WO2014076179A1 - New polypeptide - Google Patents

Abstract

The present disclosure relates to a class of engineered polypeptides having a binding affinity for Human Epidermal Growth Factor Receptor 2 (HER2), and provides a HER2 binding polypeptide comprising the sequence LAX₃AKX₆TAX₉Y HLX₁₃X₁₄X₁₅GVSDX₂₀ YKX₂₃LIDKX₂₈KT VEGVX₃₅ALYX₃₉X₄₀ ILX₄₃ALP. The present disclosure also relates to said HER2 binding polypeptide also having an affinity for albumin. Furthermore, the present disclosure relates to the use of such a HER2 binding polypeptide as a diagnostic agent and/or a medicament.

Description

NEW POLYPEPTIDE

Field of the invention

The present disclosure relates to a class of engineered polypeptides having a binding affinity for Human Epidermal Growth Factor Receptor 2 (in the following referred to as HER2). The present disclosure also relates to said HER2 binding polypeptide having affinity for albumin.

Furthermore, the present disclosure relates to use of such a HER2 binding polypeptide as a diagnostic agent and/or as a medicament.

Background

HER2 and its role in cancer diseases

The HER2 proto-oncogene encodes the production of a 185 kDa cell surface receptor protein known as the HER2 protein or receptor (Hynes NE et al (1994) Biochim Biophys Acta 1 198:165-184). This gene is also sometimes referred to as neu, HER2/neu or c-erbB-2, and the protein is often denoted "ErbB2" instead of "HER2" (in the present disclosure, these terms are sometimes used interchangeably). Neu was first discovered in rats that had been treated with ethyl nitrosourea, and exhibited mutation of this gene (Shih C et al (1981 ) Nature 290:261 -264). The mutated version of neu results in the production of a constitutively active form of the receptor, and constitutes a potent oncogene that can transform cells at low copy number (Hynes NE et al, supra).

Normal cells express a small amount of HER2 protein on their plasma membranes in a tissue-specific pattern. No known ligand to HER2 has been elucidated; however, HER2 has been shown to form heterodimers with HER1 (the epidermal growth factor receptor, EGFR), HER3 and HER4 in complex with the ligands for these receptors. Such heterodimer formation leads to the activated HER2 receptor transmitting growth signals from outside the cell to the nucleus, thus controlling aspects of normal cell growth and division (Sundaresan S et al (1999) Curr Oncol Rep 1 :16-22).

In tumor cells, errors in the DNA replication system may result in the existence of multiple copies of a gene on a single chromosome, which is a phenomenon known as gene amplification. Amplification of the HER2 gene leads to an increased transcription of this gene. This elevates HER2 mRNA levels and increases the concomitant synthesis of HER2 protein, which results in HER2 protein over-expression on the surface of these tumor cells. This over-expression can result in HER2 protein levels that are 10- to 100-fold greater than those found in the adjacent normal cells. This, in turn, results in increased cell division and a concomitantly higher rate of cell growth.

Amplification of the HER2 gene is implicated in transformation of normal cells to the cancer phenotype (Hynes NE et al, supra; Sundaresan S et al, supra).

Over-expression of HER2 protein is thought to result in the formation of homodimers of HER2, which in turn results in a constitutively active receptor (Sliwkowski MX ef a/ (1999) Semin Oncol 26(4 Suppl 12):60-70). Under these conditions, growth-promoting signals may be continuously transmitted into the cells in the absence of ligands. Consequently, multiple intracellular signal transduction pathways become activated, resulting in unregulated cell growth and, in some instances, oncogenic transformation (Hynes NE et al, supra). Thus, the signal transduction mechanisms mediated by growth factor receptors are important targets for inhibiting cell replication and tumor growth.

Breast cancer is the most common malignancy among women in the

United States, with 226870 new cases projected to occur in 2012 (Siegel R et al (2012) CA Cancer J Clin 62:10-29). In approximately 25 % of all breast cancer patients, there is an over-expression of the HER2 gene due to amplification thereof (Slamon DJ ef a/ (1989) Science 244:707-712). This over-expression of HER2 protein correlates with several negative prognostic variables, including estrogen receptor-negative status, high S-phase fraction, positive nodal status, mutated p53, and high nuclear grade (Sjogren S et al (1998) J Clin Oncol 16(2):462-469). According to Slamon et al {supra), the amplification of the HER2 gene was found to correlate strongly with shortened disease-free survival and shortened overall survival of node-positive patients.

For these reasons, it has been, and is still, an important goal to further pursue investigations into the role of HER2 in the pathogenesis and treatment of breast cancer. The identification of molecules that interact with HER2 forms one part of this effort.

Preclinical in vitro studies have examined whether inhibition of HER2 activity could affect tumor cell growth. Treatment of SK-BR-3 breast cancer cells over-expressing HER2 protein with 4D5, one of several murine anti- HER2 monoclonal antibodies, did indeed inhibit tumor cell proliferation, compared with treatment with a control monoclonal antibody. Administration of 4D5 to mice bearing human breast and ovarian cancers (xenografts) that over-express HER2 prolonged their tumor-free survival time. Similar studies demonstrated the growth inhibition by anti-HER2 monoclonal antibodies in human gastric cancer xenografts in mice (Pietras RJ et al (1994) Oncogene 9:1829-1838).

Among the approaches to inhibiting the HER2 protein abundantly present on tumor cell surfaces with an antibody, one therapy has become commercially available during recent years. Thus, the humanized variant of monoclonal antibody 4D5, or trastuzumab, is marketed for this purpose by F Hoffman-La Roche and Genentech under the trade name of Herceptin®.

Additionally, small-molecule inhibitors of HER tyrosine kinases, including the HER1 inhibitors gefitinib (Iressa) and erlotinib (Tarceva®) and the dual HER1/HER2 inhibitor lapatinib (Tykerb®), have also been developed.

Over-expression of HER2 has thus been described for breast cancer. It has also been connected to i.a. ovarian cancer, stomach cancer, bladder cancer, salivary cancer, lung cancer (Holbro et al, Annu. Rev. Pharmacol. Toxicol. 2004. 44:195-217) and cancer in the esophagus (Ekman et al, Oncologist 2007; 12;1 165-1 177, see in particular pages 1 170-1 171 ).

Notwithstanding the obvious advantages shown by antibody therapy against cancers characterized by over-expression of HER2 protein, the fact remains that a variety of factors have the potential of reducing antibody efficacy (see e.g. Reilly RM ef a/ (1995) Clin Pharmacokinet 28:126-142). These include the following: (1 ) limited penetration of the antibody into a large solid tumor or into vital regions such as the brain; (2) reduced extravasation of antibodies into target sites owing to decreased vascular permeability; (3) cross-reactivity and nonspecific binding of antibody to normal tissues, reducing the targeting effect; (4) heterogeneous tumor uptake resulting in untreated zones; (5) increased metabolism of injected antibodies, reducing therapeutic effects; and (6) rapid formation of HAMA and human antihuman antibodies, inactivating the therapeutic antibody.

In addition, toxic effects have been a major obstacle in the

development of therapeutic antibodies for cancer (Carter P (2001 ) Nat Rev Cancer 1 :1 18-129; Goldenberg DM (2002) J Nucl Med 43:693-713; Reichert JM (2002) Curr Opin Mol Ther 4:1 10-1 18). Cross-reactivity with healthy tissues can cause substantial side effects for unconjugated (naked)

antibodies, which side effects may be enhanced upon conjugation of the antibodies with toxins or radioisotopes. Immune-mediated complications include dyspnoea from pulmonary toxic effects, occasional central and peripheral nervous system complications, and decreased liver and renal function. On occasion, unexpected toxic complications can be seen, such as the cardiotoxic effects associated with the HER2 targeting antibody

trastuzumab (Schneider JW et al (2002) Semin Oncol 29(3 suppl 1 1 ):22-28). Radioimmunotherapy with isotope-conjugated antibodies also can cause bone marrow suppression.

Despite the recent clinical and commercial success of the currently used anticancer antibodies, a substantial number of important questions thus remain concerning the future of the use of antibodies. As a consequence, the continued provision of agents with a comparable affinity for HER2 remains a matter of substantial interest within the field, as well as the provision of uses of such molecules in the diagnosis and treatment of disease.

HER2 binding Z variant molecules and diagnostic use thereof

Molecules related to protein Z, derived from domain B of

staphylococcal protein A (SPA) (Nilsson B ef a/ (1987) Prot Eng 1 :107-133), have been selected from a library of randomized such molecules using different interaction targets (see e.g. WO95/19374 and Nord K et al (1997) Nat Biotech 15:772-777). In WO2005/003156, a substantial number of Z variants with an ability to interact with HER2 are disclosed. Baum et al (2010), J Nucl Med 51 (6):892-897, describes the use of one of these variants, denoted Z_HER2:342 or ABY-002, for molecular imaging in human patients.

WO2009/080810 discloses HER2 binding polypeptides with a re- engineered scaffold compared to the Z variants of WO2005/003156, as well as use of such re-engineered polypeptides for the diagnosis in general of cancer diseases in mammalian subjects characterized by the over-expression of HER2. In the experiments of WO2009/080810, the new polypeptides are used for molecular imaging studies in mice with a view to visualize HER2 bearing tumors. These experiments are also disclosed in Ahlgren et al (2010), J Nucl Med 51 (7):1 131 -1 138. Serum albumin

Serum albumin is the most abundant protein in mammalian sera (40 g/l; approximately 0.7 mM in humans), and one of its functions is to bind molecules such as lipids and bilirubin (Peters T, Advances in Protein

Chemistry 37:161 , 1985). The half-life of serum albumin is directly

proportional to the size of the animal, where for example human serum albumin (HSA) has a half-life of 19 days and rabbit serum albumin has a half- life of about 5 days (McCurdy TR et al, J Lab Clin Med 143:1 15, 2004). Human serum albumin is widely distributed throughout the body, in particular in the intestinal and blood compartments, where it is mainly involved in the maintenance of osmolarity. Structurally, albumins are single-chain proteins comprising three homologous domains and totaling 584 or 585 amino acids (Dugaiczyk L et al, Proc Natl Acad Sci USA 79:71 , 1982). Albumins contain 17 disulfide bridges and a single reactive thiol, C34, but lack N-linked and O- linked carbohydrate moieties (Peters, 1985, supra; Nicholson JP et al, Br J Anaesth 85:599, 2000). The lack of glycosylation simplifies recombinant expression of albumin. This property of albumin, together with the fact that its three-dimensional structure is known (He XM and Carter DC, Nature 358:209 1992), has made it an attractive candidate for use in recombinant fusion proteins. Such fusion proteins generally combine a therapeutic protein (which would be rapidly cleared from the body upon administration of the protein per se) and a plasma protein (which exhibits a natural slow clearance) in a single polypeptide chain (Sheffield WP, Curr Drug Targets Cardiovacs Haematol Disord 1 :1 , 2001 ). Such fusion proteins may provide clinical benefits in requiring less frequent injection and higher levels of therapeutic protein in vivo. Fusion or association with HSA results in increased in vivo half-life of proteins Serum albumin is devoid of any enzymatic or immunological function and, thus, should not exhibit undesired side effects upon coupling to a bioactive polypeptide. Furthermore, HSA is a natural carrier involved in the endogenous transport and delivery of numerous natural as well as therapeutic molecules (Sellers EM and Koch-Weser MD, "Albumin Structure, Function and Uses", eds Rosenoer VM et al, Pergamon, Oxford, p 159, 1977). Several strategies have been reported to either covalently couple proteins directly to serum albumins or to a peptide or protein that will allow in vivo association to serum albumins. Examples of the latter approach have been described e.g. in WO91/01743. This document describes inter alia the use of albumin binding peptides or proteins derived from streptococcal protein G for increasing the half-life of other proteins. The idea is to fuse the bacterially derived, albumin binding peptide/protein to a therapeutically interesting peptide/protein, which has been shown to have a rapid clearance in blood. The thus generated fusion protein binds to serum albumin in vivo, and benefits from its longer half-life, which increases the net half-life of the fused therapeutically interesting peptide/protein. Association with HSA results in decreased immunogenicity

In addition to the effect on the in vivo half-life of a biologically active protein, it has been proposed that the non-covalent association with albumin of a fusion between a biologically active protein and an albumin binding protein acts to reduce the immune response to the biologically active protein. Thus, in WO2005/097202, there is described the use of this principle to reduce or eliminate the immune response to a biologically active protein. Albumin binding domains of bacterial receptor proteins

Streptococcal protein G is a bi-functional receptor present on the surface of certain strains of streptococci and capable of binding to both IgG and serum albumin (Bjorck et al, Mol Immunol 24:1 1 13, 1987). The structure is highly repetitive with several structurally and functionally different domains (Guss et al, EMBO J 5:1567, 1986), more precisely three Ig-binding motifs and three serum albumin binding domains (Olsson et al, Eur J Biochem 168:319, 1987). The structure of one of the three serum albumin binding domains has been determined, showing a three-helix bundle domain (Kraulis et al, FEBS Lett 378:190, 1996). This motif was named ABD (albumin binding c/omain) and is 46 amino acid residues in size. In the literature, it has subsequently also been designated G148-GA3.

Other bacterial albumin binding proteins than protein G from

Streptococcus have also been identified, which contain domains similar to the albumin binding three-helix domains of protein G. Examples of such proteins are the PAB, PPL, MAG and ZAG proteins. Studies of structure and function of such albumin binding proteins have been carried out and reported e.g. by Johansson and co-workers (Johansson et al, J Mol Biol 266:859-865, 1997; Johansson et al, J Biol Chem 277:81 14-8120, 2002), who introduced the designation "GA module" (protein G-related albumin binding module) for the three-helix protein domain responsible for albumin binding. Furthermore, Rozak et al have reported on the creation of artificial variants of the GA module, which were selected and studied with regard to different species specificity and stability (Rozak et al, Biochemistry 45:3263-3271 , 2006; He et al, Protein Science 16:1490-1494, 2007).

Variants of the G148-GA3 domain have been developed, with various optimized characteristics. Such variants are for example disclosed in PCT publications WO00/23580 (alkali-stabilized variant denoted "ABDmut"; sequence disclosed in Figure 3 of WO00/23580), WO2009/016043 and WO2012/004384.

In short, drawbacks of current antibody based cancer therapies, especially in solid tumor targeting applications, include the large size of antibodies which impairs their vascular permeability and tumor penetration capability. On the other hand, known smaller binding molecules suffer the drawback of rapid renal clearance and therefore short in vivo half-life. Thus therapies based on smaller binding molecules, have to be administered more frequently and/or in higher dose leading to stressful treatment regimes and even undesirable side effects, such as pulmonary and cardiotoxicity.

As is evident from the different sections of this background description, the provision of polypeptide molecules of small size and with a high affinity for HER2 as well as increased in vivo half-life could be desirable in the

development of efficient therapies targeting various forms of cancer, in particular breast cancer.

Disclosure of the invention

It is an object of the present disclosure to provide a molecule allowing for efficient therapy targeting various forms of cancer while alleviating the abovementioned and other drawbacks of current therapies.

It is another object of the present disclosure to provide a HER2 binding polypeptide, which may allow better tissue penetration than existing molecules, while simultaneously exhibiting an increased in vivo half-life.

These and other objects which are evident to the skilled person from the present disclosure are met by different aspects of the invention as claimed in the appended claims and as generally disclosed herein.

Thus, in the first aspect of the disclosure, there is provided a HER2 binding polypeptide, comprising an amino acid sequence selected from i) LAX₃AKX₆TAX₉Y HLXi₃Xi Xi5GVSDX₂₀ YKX23LIDKX28KT

VEGVX35ALYX39X40 ILX43ALP wherein, independently of each other,

X3 is selected from A, G, P, S and V; Χβ is selected from D and E;

Xg is selected from L and N;

Xl3 is selected from D and T;

Xl4 is selected from K and R;

Xl5 is selected from I, L, M, T and V;

X20 is selected from F and Y;

X23 is selected from D and R;

X28 is selected from A and V;

X35 is selected from K, M and R;

X39 is selected from A, F and L;

X40 is selected from A and E; and

X43 is selected from A, H, K, P, R, T, Q and Y and ii) an amino acid sequence which has at least 83 % identity to the

sequence defined in i).

As the skilled person will realize, the function of any polypeptide, such as the HER2 binding capacity of the polypeptide of the present disclosure, is dependent on the tertiary structure of the polypeptide. It is therefore possible to make minor changes to the sequence of amino acids in a polypeptide without affecting the function thereof. Thus, the invention encompasses modified variants of the HER2 binding polypeptide, which are such that the HER2 binding characteristics are retained. In another embodiment the invention encompasses modified variants of the HER2 binding polypeptide which retain HER2 binding characteristics and albumin binding

characteristics.

In this way, also encompassed by the present disclosure is a HER2 binding polypeptide comprising an amino acid sequence with 86 % or greater identity to a polypeptide as defined in i). In some embodiments, the inventive polypeptide may comprise a sequence which is at least 89 %, such as at least 91 %, such as at least 93 %, such as at least 95 %, such as at least 97 % identical to the polypeptides as defined in i).

In some embodiments, such changes may be made in all positions of the sequences of the HER2 binding polypeptide as disclosed herein. In other embodiments, such changes may be made only in the non-variable positions, also denoted as scaffold amino acid residues. In such cases, changes are not allowed in the variable positions, i.e. positions denoted with an "X" in sequence i). For example, it is possible that an amino acid residue belonging to a certain functional grouping of amino acid residues (e.g. hydrophobic, hydrophilic, polar etc) could be exchanged for another amino acid residue from the same functional group.

The term "% identity", as used throughout the specification, may for example be calculated as follows. The query sequence is aligned to the target sequence using the CLUSTAL W algorithm (Thompson et al, Nucleic Acids Research, 22: 4673-4680 (1994)). A comparison is made over the window corresponding to the shortest of the aligned sequences. The shortest of the aligned sequences may in some instances be the target sequence. In other instances, the query sequence may constitute the shortest of the aligned sequences. The amino acid residues at each position are compared, and the percentage of positions in the query sequence that have identical

correspondences in the target sequence is reported as % identity. In the alignment, deletions and additions of one or a few amino acid residues are suitably taken into account. When present, a deletion or addition of one amino acid residue in one of the aligned sequences is counted as one difference in the calculation of identity between the sequences.

In one embodiment of the first aspect, there is provided a HER2 binding polypeptide, which is also capable of binding albumin. Albumin binding ability, when present in HER2 binding polypeptides of the present disclosure, is thought to arise as a consequence of retaining the original albumin-binding capacity of the G148-GA3 domain (or "ABDwt") and the stabilized version thereof ("ABD" in the Examples and Figures herein;

"ABDmut" in WO00/23580) discussed in the Background section. It is advantageous for the HER2 binding polypeptide to also bind to albumin, because such albumin binding is expected to prolong the in vivo half life of the polypeptide, by for example avoiding rapid renal clearance. The relatively small HER2 binding polypeptide of the present disclosure, which is capable of binding to albumin in addition to HER2, is expected, due to its small size, not to be impaired in terms of vascular permeability and tumor penetration capability. This is in contrast to many significantly larger molecules.

In one embodiment of a polypeptide according to this aspect, X₃ in sequence i) is selected from A, G and P. In one embodiment, X₃ in sequence i) is selected from A and P. In another embodiment, X₃ in sequence i) is selected from A and G. In yet another embodiment X₃ in sequence i) is A.

In one embodiment, Χβ in sequence i) is E.

In one embodiment, Xg in sequence i) is L.

In one embodiment, Xl3 n sequence i) is D.

In one embodiment, Xl4 n sequence i) is R.

In one embodiment, Xl5 n sequence i) is selected from L and V.

In one embodiment, Xl5 n sequence i) is L.

In one embodiment, Xl5 n sequence i) is V.

In one embodiment, Xi₄Xi₅ in sequence i) is selected from RL and RV.

In one embodiment, X20 n sequence i) is F.

In one embodiment, X20 n sequence i) is Y.

In one embodiment, X23 n sequence i) is D.

In one embodiment, X23 n sequence i) is R.

In one embodiment, X28 n sequence i) is A.

In one embodiment, X35 n sequence i) is R.

In one embodiment, X39 n sequence i) is selected from F and L.

In one embodiment, X3g in sequence i) is F.

In one embodiment, X3g in sequence i) is L.

In one embodiment, X40 in sequence i) is E.

In one embodiment, X43 in sequence i) is selected from H, P and R.

In yet further embodiments of the HER2 binding polypeptide according to this aspect, sequence i) is selected from the group consisting of SEQ ID NO:1 -24. In one embodiment, sequence i) is selected from the group consisting of SEQ ID NO:1 -7, SEQ ID NO:9-19 and SEQ ID NO:21 -24. In another embodiment, sequence i) is selected from the group consisting of SEQ ID NO:1 -12. In yet another embodiment, sequence i) is selected from SEQ ID NO:1 -7 and SEQ ID NO:9-12.

In another embodiment, sequence i) is selected from the group consisting of SEQ ID NO:1 -24 and 30-177. In another embodiment, sequence i) is selected from SEQ ID NO:1 -12 and 30-103. In another embodiment, sequence i) is selected from the group consisting of SEQ ID NO:32, 33, 37, 38, 39, 43, 44, 87, 90, 91 , 93, 106, 107, 1 1 1 , 1 12, 1 13, 1 17, 1 18, 161 , 164, 165 and 167. In another embodiment, sequence i) is selected from the group consisting of SEQ ID NO:32, 33, 37, 38, 39, 43, 44, 87, 90, 91 and 93. The terms "HER2 binding" and "binding affinity for HER2" as used in this specification refer to a property of a polypeptide which may be tested for example by the use of surface plasmon resonance (SPR) technology. For example as described in the examples below, HER2 binding affinity may be tested in an experiment in which HER2, or a fragment thereof, is immobilized on a sensor chip of the instrument, and the sample containing the polypeptide to be tested is passed over the chip. Alternatively, the polypeptide to be tested is immobilized on a sensor chip of the instrument, and a sample containing HER2, or a fragment thereof, is passed over the chip. The skilled person may then interpret the results obtained by such experiments to establish at least a qualitative measure of the binding affinity of the

polypeptide for HER2. If a quantitative measure is desired, for example to determine a K_D value for the interaction, surface plasmon resonance methods may also be used. Binding values may for example be defined in a Biacore (GE Healthcare) or ProteOn XPR 36 (Bio-Rad) instrument. HER2 is suitably immobilized on a sensor chip of the instrument, and samples of the polypeptide whose affinity is to be determined are prepared by serial dilution and injected in random order. K_D values may then be calculated from the results using for example the 1 :1 Langmuir binding model of the

BIAevaluation 4.1 software, or other suitable software, provided by the instrument manufacturer.

The terms "albumin binding" and "binding affinity for albumin" as used in this specifications refer to a property of a polypeptide which may also be tested for example by the use of SPR technology, such as in a Biacore or ProteOn XPR36 instrument, in an analogous way to the example described above for HER2.

In one embodiment, the HER2 binding polypeptide is capable of binding to HER2 such that the K_D value of the interaction is at most 1 x 10^"8 M, such as at most 1 x 10^"9 M, such as at most 1 x 10^"10 M, such as at most 1 x 10^"11 M.

In one embodiment of those polypeptides disclosed herein that have a dual affinity for both HER2 and albumin, the HER2 binding polypeptide is capable of binding to albumin such that the K_D value of the interaction is at least 1 x 10^"8 M, such as at least 1 x 10^"7 M, such as at least 1 x 10^"6 M, such as at least 1 x 10^"5 M. In one embodiment, the K_D value of the interaction with albumin is within the range from 1 x 10^{"1 1} M to 1 x 10^"6 M, such as within the range from 1 x 10^"9 M to 1 x 10^"6 M. In one embodiment of those polypeptides disclosed herein that have a dual affinity for both HER2 and albumin, the affinity of the HER2 binding polypeptide for HER2 is higher than its affinity for albumin, such that the HER2 binding polypeptide is preferentially bound to HER2 in the presence of both HER2 and albumin, but binds to albumin in the absence of HER2.

In one embodiment, said albumin is human serum albumin.

The skilled person will understand that various modifications and/or additions can be made to a HER2 binding polypeptide according to any aspect disclosed herein in order to tailor the polypeptide to a specific application without departing from the scope of the present disclosure.

For example, any HER2 binding polypeptide disclosed herein may comprise further C terminal and/or N terminal amino acids. Such a

polypeptide should be understood as a polypeptide having one or more additional amino acid residues at the very first and/or the very last position in the polypeptide chain, i.e. at the N- and/or C-terminus of sequence i) or ii). Thus, a HER2 binding polypeptide may comprise any suitable number of additional amino acid residues, for example at least one additional amino acid residue. Each additional amino acid residue may individually or collectively be added in order to, for example, improve production, purification, stabilization in vivo or in vitro, coupling, or detection of the polypeptide. Such additional amino acid residues may comprise one or more amino acid residues added for the purpose of chemical coupling. One example of this is the addition of a cysteine residue. Such additional amino acid residues may also provide a "tag" for purification or detection of the polypeptide, such as a His6 tag or a "myc" (c-myc) tag or a "FLAG" tag for interaction with antibodies specific to the tag or immobilized metal affinity chromatography (IMAC) in the case of

The further amino acids as discussed above may be coupled to the HER2 binding polypeptide by means of chemical conjugation (using known organic chemistry methods) or by any other means, such as expression of the HER2 binding polypeptide as a fusion protein or joined in any other fashion, either directly or via a linker, for example an amino acid linker.

The further amino acids as discussed above may for example comprise one or more polypeptide domain(s). A further polypeptide domain may provide the HER2 binding polypeptide with another function, such as for example yet another binding function, or an enzymatic function, or a toxic function (e.g. an immunotoxin), or a fluorescent signaling function, or combinations thereof.

A further polypeptide domain may moreover provide another HER2 binding moiety with the same HER2 (and optionally albumin) binding function. Thus, in a further embodiment, there is provided a HER2 binding polypeptide as a multimer, such as dimer. Said multimer is understood to comprise at least two HER2 binding polypeptides as disclosed herein as monomer units, the amino acid sequences of which may be the same or different. Multimeric forms of the polypeptides may comprise a suitable number of domains, each having a HER2 binding motif, and each forming a monomer within the multimer. These domains may have the same amino acid sequence, but alternatively, they may have different amino acid sequences. In other words, the HER2 binding polypeptide of the invention may form homo- or

heteromultimers, for example homo- or heterodimers.

Additionally, "heterogenic" fusion polypeptides or proteins, or conjugates, in which a HER2 binding polypeptide according to the invention, or multimer thereof, constitutes a first domain, or first moiety, and the second and further moieties have other functions than binding HER2, are also contemplated and fall within the ambit of the present disclosure. The second and further moiety/moieties of the fusion polypeptide or conjugate in such a protein suitably have a desired biological activity.

Thus, in a second aspect of the present disclosure, there is provided a fusion protein or a conjugate, comprising a first moiety consisting of a HER2 binding polypeptide according to the first aspect, and a second moiety consisting of a polypeptide having a desired biological activity. In another embodiment, said fusion protein or conjugate may additionally comprise further moieties, comprising desired biological activities that can be either the same or different from the biological activity of the second moiety.

Non-limiting examples of such a desired biological activity comprise a therapeutic activity, a binding activity, and an enzymatic activity. In one embodiment, the second moiety having a desired biological activity is a therapeutically active polypeptide.

Non-limiting examples of therapeutically active polypeptides are biomolecules, such as molecules selected from the group consisting of human endogenous enzymes, hormones, growth factors, chemokines, cytokines and lymphokines. In one embodiment of this aspect of the present disclosure, there is provided a HER2 binding polypeptide, fusion protein or conjugate which further comprises a cytotoxic agent. Non-limiting examples of cytotoxic agents are agents selected from the group consisting of auristatin, anthracycline, calicheamycin, combretastatin, doxorubicin, duocarmycin, the CC-1065 anti- tumorantibiotic, ecteinsascidin, geldanamycin, maytansinoid, methotrexate, mycotoxin, ricin and its analogues, taxol and derivates thereof and

combinations thereof.

As the skilled person understands, the HER2 binding polypeptide according to the first aspect may be useful in a fusion protein or as a conjugate partner to any other moiety. Therefore, the above lists of therapeutically active polypeptides and cytotoxic agents should not be construed as limiting in any way.

Other possibilities for the creation of fusion polypeptides or conjugates are also contemplated. Thus, an HER2 binding polypeptide according to the first aspect of the invention may be covalently coupled to a second or further moiety or moieties, which in addition to or instead of target binding exhibit other functions. One example is a fusion between one or more HER2 binding polypeptide(s) and an enzymatically active polypeptide serving as a reporter or effector moiety.

With regard to the description above of fusion proteins or conjugates incorporating an HER2 binding polypeptide according to the invention, it is to be noted that the designation of first, second and further moieties is made for clarity reasons to distinguish between HER2 binding polypeptide or polypeptides according to the invention on the one hand, and moieties exhibiting other functions on the other hand. These designations are not intended to refer to the actual order of the different domains in the polypeptide chain of the fusion protein or conjugate. Thus, for example, said first moiety may without restriction appear at the N-terminal end, in the middle, or at the C-terminal end of the fusion protein or conjugate.

The above aspects furthermore encompass polypeptides in which the HER2 binding polypeptide according to the first aspect, or the HER2 binding polypeptide as comprised in a fusion protein or conjugate according to the second aspect, further comprises a label, such as a label selected from the group consisting of fluorescent dyes and metals, chromophoric dyes, chemiluminescent compounds and bioluminescent proteins, enzymes, radionuclides and particles. Such labels may for example be used for detection of the polypeptide.

For example, in embodiments where the labeled HER2 binding polypeptide comprises a HER2 binding polypeptide according to the first aspect of the disclosure and a label, the labeled polypeptide may for example be used for indirect labeling of HER2 expressing tumors cells as well as metastatic cells.

In other embodiments, the labeled HER2 binding polypeptide is present as a moiety in a fusion protein or conjugate also comprising a second moiety having a desired biological activity. The label may in some instances be coupled only to the HER2 binding polypeptide, and in some instances both to the HER2 binding polypeptide and to the second moiety of the conjugate or fusion protein. Furthermore, it is also possible that the label may be coupled to a second moiety only and not the HER2 binding moiety. Hence, in yet another embodiment, there is provided an HER2 binding polypeptide comprising a second moiety, wherein said label is coupled to the second moiety only.

When reference is made to a labeled polypeptide, this should be understood as a reference to all aspects of polypeptides as described herein, including fusion proteins and conjugates comprising a HER2 binding polypeptide and a second and optionally further moieties. Thus, a labeled polypeptide may contain only the HER2 binding polypeptide and e.g. a therapeutic radionuclide, which may be chelated or covalently coupled to the HER2 binding polypeptide, or contain the HER2 binding polypeptide, a therapeutic radionuclide and a second moiety such as a small molecule having a desired biological activity, for example a therapeutic efficacy.

In embodiments where the HER2 binding polypeptide, fusion protein or conjugate is radiolabeled, such a radiolabeled polypeptide may comprise a radionuclide. A majority of radionuclides have a metallic nature and metals are typically incapable of forming stable covalent bonds with elements presented in proteins and peptides. For this reason, labeling of proteins and peptides with radioactive metals is performed with the use of chelators, i.e. multidentate ligands, which form non-covalent compounds, called chelates, with the metal ions. In an embodiment of the HER2 binding polypeptide, fusion protein or conjugate, the incorporation of a radionuclide is enabled through the provision of a chelating environment, through which the

radionuclide may be coordinated, chelated or complexed to the polypeptide. One example of a chelator is the polyaminopolycarboxylate type of chelator. Two classes of such polyaminopolycarboxylate chelators can be distinguished: macrocyclic and acyclic chelators.

In one embodiment, the HER2 binding polypeptide, fusion protein or conjugate comprises a chelating environment provided by a

polyaminopolycarboxylate chelator conjugated to the HER2 binding polypeptide via a thiol group of a cysteine residue or an epsilon amine group of a lysine residue.

The most commonly used macrocyclic chelators for radioisotopes of indium, gallium, yttrium, bismuth, radioactinides and radiolanthanides are different derivatives of DOTA (1 ,4,7,10-tetraazacyclododecane-1 ,4,7,10- tetraacetic acid). In one embodiment, a chelating environment of the HER2 binding polypeptide, fusion protein or conjugate is provided by DOTA or a derivative thereof. More specifically, in one embodiment, the chelating polypeptides encompassed by the present disclosure are obtained by reacting the DOTA derivative 1 ,4,7,10-tetraazacyclododecane-1 ,4,7-tris- acetic acid-10-maleimidoethylacetamide (maleimidomonoamide-DOTA) with said polypeptide.

Additionally, 1 ,4,7-triazacyclononane-1 ,4,7-triacetic acid (NOTA) and derivatives thereof may be used as chelators. Hence, in one embodiment, there is provided a HER2 binding polypeptide, fusion protein or conjugate, wherein the the polyaminopolycarboxylate chelator is 1 ,4,7- triazacyclononane-1 ,4,7-triacetic acid or a derivative thereof.

The most commonly used acyclic polyaminopolycarboxylate chelators are different derivatives of DTPA (diethylenetriamine-pentaacetic acid). Hence, polypeptides having a chelating environment provided by

diethylenetriaminepentaacetic acid or derivatives thereof are also

encompassed by the present disclosure. In a third aspect of the present disclosure, there is provided a polynucleotide encoding a HER2 binding polypeptide or a fusion protein as described herein.

Also encompassed by this disclosure is a method of producing a polypeptide or fusion protein as described above comprising expressing a polynucleotide; an expression vector comprising the polynucleotide; and a host cell comprising the expression vector. Also encompassed is a method of producing a polypeptide, comprising culturing said host cell under conditions permissive of expression of said polypeptide from its expression vector, and isolating the polypeptide.

The HER2 binding polypeptide of the present disclosure may

alternatively be produced by non-biological peptide synthesis using amino acids and/or amino acid derivatives having protected reactive side-chains, the non-biological peptide synthesis comprising

- step-wise coupling of the amino acids and/or the amino acid derivatives to form a polypeptide according to the first aspect having protected reactive side-chains,

- removal of the protecting groups from the reactive side-chains of the polypeptide, and

- folding of the polypeptide in aqueous solution. It should be understood that the HER2 binding polypeptide according to the present disclosure may be useful as a therapeutic or diagnostic agent in its own right or as a means for targeting other therapeutic or diagnostic agents, with e.g. direct or indirect effects on HER2. A direct therapeutic effect may for example be accomplished by inhibiting HER2 signaling.

Hence, in another aspect of the present disclosure, there is provided a

HER2 binding polypeptide, fusion protein or conjugate as described herein for use as a medicament. Also, in another embodiment there is provided a HER2 binding polypeptide, fusion protein or conjugate as described herein for use in diagnosis. In one embodiment, there is provided a HER2 binding polypeptide, fusion protein or conjugate for use in the treatment or diagnosis of a HER2 related condition, such as cancer. Non-limiting examples of HER2 related conditions are cancers selected from the group consisting of breast cancer, ovarian cancer, gastric cancer, uterine cancer, testicular cancer, esophageal cancer, bladder cancer, salivary cancer and lung cancer.

In yet another aspect, there is provided a composition comprising a

HER2 binding polypeptide, fusion protein or conjugate as described herein and at least one pharmaceutically acceptable excipient or carrier. In one embodiment thereof, the composition further comprises at least one additional active agent, such as at least two additional active agents, such as at least three additional active agents. Non-limiting examples of additional active agents that may prove useful in such a combination are immunostimulatory agents, radionuclides, toxic agents, enzymes, factors recruiting effector cells (e.g. T or NK cells) and photosensitizers.

In a related aspect, there is provided a method of treatment of a HER2 related condition, comprising administering to a subject in need thereof an effective amount of a HER2 binding polypeptide, fusion protein or conjugate HER2 binding polypeptide as described herein. Consequently, in the method of treatment, the subject is treated with a HER2 binding polypeptide or a HER2 binding combination according to the invention. In a more specific embodiment of said method, the HER2 binding polypeptide, fusion protein or conjugate inhibits HER2 mediated signaling by binding to HER2 expressed on a cell surface.

While the invention has been described with reference to various exemplary aspects and embodiments, it will be understood by those skilled in the art that various changes may be made and equivalents may be

substituted for elements thereof without departing from the scope of the invention. In addition, many modifications may be made to adapt a particular situation or molecule to the teachings of the invention without departing from the essential scope thereof. Therefore, it is intended that the invention not be limited to any particular embodiment contemplated, but that the invention will include all embodiments falling within the scope of the appended claims.

Brief description of the figures

Throughout the Examples and Figures, the synonym term "ErbB2" is used interchangeably with the term "HER2".

Figure 1 is a listing of amino acid sequences of examples of HER2 binding polypeptides according to the disclosure (SEQ ID NO:1 -24 and 30- 177), as well as of HER2 binding Z variant polypeptides Z02891 (SEQ ID NO:25) and Z477 (SEQ ID NO:26) as well as alkali stabilized G148-GA3 (SEQ ID NO:27; in the Examples and Figures denoted "ABD"; see

Background section above and PCT publication WO00/23580, in which the same polypeptide is denoted "ABDmut") and the wildtype G148-GA3 domain (SEQ ID NO:28; see Background section above and PCT publication

WO00/23580, in which the same polypeptide is denoted "ABDwt"). Figure 2A shows the strategy for the phage display selection of the first generation binding molecule ABD_HER2-I (SEQ ID NO:12). Three rounds (1 -3) of selection with decreasing target concentrations and increasing number of washes were performed in four (A-D) parallel tracks.

Figure 2B shows an alignment of the sequences of the selected first generation HER2 binding ABD-variant ABD_HER2-I (SEQ ID NO:12) and the starting sequence ABD (SEQ ID NO:27). The eleven positions that were diversified in the combinatorial library are marked by dots, whereas boxes indicate the three alpha helices of the ABD polypeptide.

Figure 2C shows sensorgrams for ABD_HER2-I (SEQ ID NO:12) binding to immobilized human HER2 (left) and to human serum albumin (HSA) (right). From the top, curves represent 100 nM (right panel only), 33 nM, 1 1 nM, 3.7 nM and 1 .2 nM ABD_HER2-I - Solid lines indicate a Langmuir 1 :1 kinetic fit.

Figure 3A shows the library design for affinity maturation. A

conservative library and a semi-conservative library were designed as indicated. Positions marked with X were randomized by NNK. ABD (SEQ ID NO:27) and ABD_HER2-I (SEQ ID NO:12) are included for comparison.

Figure 3B shows the selection strategy for affinity maturation.

Figure 3C shows a sequence logotype derived from a group of sequenced polypeptides obtained in selections without presence of HSA (271 sequences). The eleven randomized positions are shown from left (N- terminus) to right (C-terminus) and their location in the 46 amino acid sequence are indicated by numbers.

Figure 3D shows the corresponding logotype for sequences from selections with presence of HSA (167 sequences).

Figure 4A shows the sequences of polypeptides ABD_HER2-mati-n (SEQ ID NO:1 -1 1 ) obtained after affinity maturation and selection as described in Example 4, as compared to the starting ABD sequence (SEQ ID NO:27). Also indicated is the number of times each clone was observed (among 438 sequenced clones) and the selection strategy tracks in which the

corresponding sequence was found.

Figure 4B shows sensorgrams for ABD_HER2-mati (SEQ ID NO:1 ) binding to immobilized HER2. From the top, curves represent 100 nM, 33 nM, 1 1 nM, 3.7 nM and 1 .2 nM ABD_HER2-mati - Solid lines indicate a Langmuir 1 :1 kinetic fit.

Figure 4C shows the overlapping CD spectra (average of five individual wavelength scans) of ABD_HER2-mati (SEQ ID NO:1 ) before and after heating to 90 °C. Figure 5 shows flow cytometry data of batch output from selection tracks without HSA (A, C, E and G) and with HSA (B, D, F and H), incubated with 50 nM HER2 and with or without 1 μΜ HSA. Each diagram presents an overlay of contour plots of the expressing populations of cells from two separate samples from the same batch incubated with HER2 only or with HER2 and HSA.

Figure 6 shows comparisons of HSA-binding for batches of ABD molecules from selection tracks A-H displayed on cells. HSA binding is shown on the y-axis, and expression level on the x-axis. As controls, cells expressing ABDHER2-I (SEQ ID NO:12) or starting ABD (SEQ ID NO:27) are included.

Figure 7 shows a competition assay of HER2 binding ABD variants from track A and HER2 binding Z variant molecules Z02891 and Z477.

Figure 8 shows the results from two repeated cell-based affinity screenings as described in Example 8, using variants enriched after four rounds of sorting by staphylococcal display as described in Example 7.

Z02891 , ABDHER2-I and ABD where included as positive or negative controls. The variant SEQ ID NO:7 was assayed in duplicates as an internal assay control. A) Screening of HER2 binding. The mean fluorescence intensity (MFI), normalized against the ABD_HER2-I signal, is shown on the Y-axis. B) Screening of HSA binding. MFI, normalized against the ABD signal, is shown on the Y-axis. C) Screening of HER2 binding in the presence of HSA. MFI, normalized against the ABD_HER2-I signal, is shown on the Y-axis.

Figure 9 shows the result of a competition assay of HER2 binding ABD variants, Herceptin®-scFv and Z variant molecules Z02891 and Z477, as described in Example 10. Staphylococci expressing (A) the HER2 binding ABD variant SEQ ID NO:43, (B) the HER2 binding ABD variant SEQ ID NO:90, (C) Herceptin®-scFv or (D) the Z variant polypeptide Z02891 were each incubated with the following different sample compositions l-V. I: HER2; II: HER2 + SEQ ID NO:43; III: HER2 + SEQ ID NO:90; IV: HER2 + Z477 and V: HER2 + Herceptin-scFV. Fluorescence normalized against the signal for HER2 alone (I) is plotted on the y-axes, and data from two replicate assays are shown.

The invention will now be illustrated further through the non-limiting description of experiments conducted in accordance therewith. Unless otherwise specified, conventional chemistry and molecular biology methods were used throughout. Examples

As used throughout the example section of this specification, HER2 binding polypeptides according to the invention are referred to according to the ABDHER2-N and ABD_HER2-matN nomenclature, wherein N is an integer and "mat" indicates affinity matured molecules.

Example 1

Phage display selections, DNA sequencing and sub-cloning

Summary

In this example, two separate phage display selections were performed using a recombinant human HER2 Fc fusion protein as target. The first selection was carried out in three successive rounds with decreased target concentration (Figure 2A) and the second was performed in four cycles with somewhat altered selection parameters in an attempt to increase the number of positive clones. DNA sequencing of clones originating from the second and third rounds from both selections and from the fourth round of the second selection revealed a clear dominance of one resulting sequence, which was denoted ABD_HER₂-I (SEQ ID NO:12) (Figure 2B).

Materials and methods

Phage display was used to select binders to the extracellular domain of human HER2 from a combinatorial library based on an albumin-binding domain derived from streptococcal protein G and denoted in these Examples as "ABD" (Aim T et al (2010) Biotechnol J 5(6):605-17). Phagemid- transformed RR1 AM15 Escherichia coli {E. coli) cells (Ruther U (1982) Nucleic Acids Res 10(19):5765-72), in excess amount compared to the library size, were grown in 500 ml tryptic soy broth (TSB; Merck, Darmstadt,

Germany) supplemented with 5 % (w/v) yeast extract (Merck), 2 % (w/v) glucose and 100 pgml^"1 ampicillin. An aliquot of cells (8 ml in selection rounds 1 -2, 4 ml in round 3) in the log phase were incubated with a 15-fold excess of helper phage (M13K07; New England Biolabs) during a still 2 h incubation at 37 °C. The cells were harvested by centrifugation and inoculated to 500 ml (100 ml in round 3) of TSB supplemented with 5 % (w/v) yeast extract, 100 gml^"1 ampicillin, 50 pgrnl^"1 kanamycin and 1 mM isopropyl^-D-1 - thiogalactopyranoside (IPTG; Apollo Scientific, Derbyshire, UK) and incubated at 30 °C over night (ON). Phages were precipitated with

polyethylene glycol/NaCI from the culture supernatants using standard protocols and re-suspended in phosphate buffered saline (PBS) pH 7.4 supplemented with 3 % (w/v) bovine serum albumin (BSA) and 0.1 % (v/v) Tween 20 (3 % PBSTB).

Three rounds of selection were performed using a recombinant human HER2-Fc fusion (Cat. No. 1 129-EP, R&D Systems, Minneapolis, MN, USA) as target. In rounds 2-3, phages were incubated with 100 nM polyclonal human Fc (Cat. No. P80-104, Bethyl Laboratories, Montgomery, TX, USA) for 30 min at room temperature (RT). 0.6 mg of washed Dynabeads® Protein A magnetic beads (Invitrogen, Carlsbad, CA, USA), which had been washed twice in PBS with 0.1 % (v/v) Tween 20 (PBST) and blocked for 20 min with PBS supplemented with 5 % (w/v) BSA and 0.1 % (v/v) Tween 20 (5 % PBSTB), were added to capture Fc and non-specifically bound phages. The pre-selected phages (supernatant) were transferred to a new tube and incubated with HER2-Fc for 2 h at RT according to Figure 2A. 1 .6 mg of Dynabeads® Protein A (0.6 mg in round 3) were added for capture of target- bound phages, washed with PBST and eluted by addition of 500 μΙ of 50 mM glycine-HCI pH 2.7 for 10 min. The beads used for pre-selection were treated with the same washing and elution procedure as during selection. The eluates were neutralized by dilution in 450 μΙ PBS supplemented with 50 μΙ 1 M Tris- HCI pH 8.0 and used to infect fresh cultures of RR1 AM15 for 30 min at 37 °C followed by plating on tryptone yeast extract (TYE) plates with 100 pgml^"1 ampicillin and 2 % (w/v) glucose and incubation ON. Colonies were collected and used to produce phages for the following round of selection. Phage titers were determined for samples from phage stocks after precipitation, the last wash and eluate from both pre-selection and selection. The concentration of HER2 was decreased for each cycle; the concentration of Tween 20 in the washing buffer was increased as well as the number of washes (Figure 2A). At least a 10000-fold excess of phages compared to the eluate in the previous round were used in following cycles. All tubes were blocked with 5 % PBSTB before use and always incubated end-over-end.

In a separate experiment aimed to isolate additional HER2 binders from the library, the full selection procedure was repeated from a new phage stock prepared from a separate RR1 AM15 freeze stock of the original phagemid library. The selection parameters were modified in this selection compared to the one described above; 100 nM of target was used in the first round and the concentrations were decreased to 25 or 0.5 nM in a fourth round. The number of washes was increased from three in round one to seven in the last round. Dynabeads® Protein A or Dynabeads® Protein G (Invitrogen) were used in alternating cycles for capture and elution was performed with 50 mM glycine-HCI pH 2.2. Fc was used for pre-selection in all four rounds (100 nM in 1 -3 and 50 nM in 4).

DNA encoding selected ABD variants from individual colonies isolated after selection were amplified by polymerase chain reaction (PCR),

sequenced with specific primers and Big Dye terminators (GE Healthcare, Uppsala, Sweden) and analyzed on an ABI Prism® 3700 DNA sequencer (Applied Biosystems, Foster City, CA, USA). The same procedure was used for sequence verification of fragments sub-cloned in the expression- and cell- display vectors, respectively.

For sub-cloning of phage selected ABD-variants into an expression vector, phagemids were purified from small-scale cultivations (QIAquick plasmid miniprep kit, Qiagen, Solna, Sweden). PCR-fragments with flanking EcoRI and Xhol restriction sites were produced, purified (QIAquick PCR purification kit, Qiagen), restricted (EcoRI-HF and Xhol-HF; New England Biolabs (NEB), Ipswich, MA, USA), purified and ligated in an expression vector containing a T7 promoter, a kanamycin resistance gene and an N- terminal His6-tag that had been cleaved with the same enzymes,

dephosphorylated with alkaline phosphatase (NEB) and purified from a 1 % agarose gel (DNA gel extraction kit, Qiagen). The same procedure was used to clone the non-randomized ABD sequence into the same vector, using the plasmid pTrpABDstable (Gulich S et al (2000) J Biotechnol 80(2):169-78; Kraulis PJ ef a/ (1996) FEBS lett 378(2):190-4) as a template for PCR and for sub-cloning of affinity matured ABD molecules.

Results

Selections revealed a clear dominance of one sequence, referred to as ABD_HER2-i (SEQ ID NO:12) (Figure 2B). This sequence was found in 81 % (942) of a total of 1 169 colonies sequenced after the third (both replicate selections) and fourth (second selection) rounds of selection. ABD_HER2-I was always represented by the same nucleotide sequence and found in all parallel selection tracks of both selections. Sequencing of 180 colonies after the second round resulted in a more diverse data set, although 20 % of the colonies were ABD_HER2-I - This indicated that convergence on this sequence occurred early. A few additional sequences showed up more than once, however clustering and alignment of all unique sequences did not reveal any apparent similarities to the dominating candidate.

Example 2

Expression, purification and characterization of first generation candidates

Materials and methods

Mass spectrometry (MS) analysis, circular dichroism (CD) analysis and surface plasmon resonance (SPR) spectroscopy was performed as described in Example 5.

Results

ABDHER2-I and five additional candidates (data not shown), identified as described in Example 1 , were sub-cloned into an expression vector, expressed and purified to homogeneity. Four of the additional candidates that were selected beside ABD_HER2-I were represented more than once in the sequence data and none contained any cysteines. Molecular weights were verified by MS (data not shown). CD-analysis demonstrated that all variants except one shared a similar spectrum and the spectra were similar to what has been measured previously for G148-GA3 (SEQ ID NO:28) and the stabilized ABD used in this library (SEQ ID NO:27) (Gulich S et al, supra). For all proteins except the deviating variant, spectra before and after thermal denaturation overlapped, and all melting temperatures were above 60 °C. ABDHER2-I denatured at 68 °C (Table 1 ) and the starting polypeptide ABD had a T_m above 80 °C, which is comparable to what has been reported before using a different expression vector (Gulich S et al, supra).

SPR studies confirmed that all variants had retained the affinity of the ABD polypeptide to HSA, but revealed that only the most commonly observed variant, ABD_HER2-I, was able to bind HER2 (Figure 2C). For ABD_HER2-I, an equilibrium dissociation constant (K_D) of 75 nM was calculated from the SPR analysis, both from surfaces with HER2-Fc and HER2-His6. Thus, data from those surfaces was treated together. Kinetic parameters are shown in Table 1 . The starting ABD domain bound to albumin with an affinity of 8 nM (Table 1 ), which is similar to what has been measured for this molecule as well as for the wild-type G148-GA3 before (Nilvebrant J et al (201 1 ) PLoS One

6(10):e25791 , Jonsson A et al (2008) Protein Eng Des Sel 21 (8):515-27).

Example 3

Alanine scan mutagenesis and library construction for affinity maturation Summary

To evaluate the dominating first generation candidate ABD_HER2-I , identified and studied in Examples 1 and 2, respectively, alanine point mutants of this protein were generated. Data obtained from binding analysis of these mutants of ABD_HER2-I were used to design two affinity maturation libraries (Figure 3A). First, a conservative library maintaining T7, Y10, H1 1 and Y38 was designed based on data from point mutants of ABD_HER2-I - Positions marked with X were randomized by NNK. Second, a semi- conservative library was constructed. The randomization in this library was similar to that in the conservative library, but here, T7, Y10, H1 1 , Y38 and A39 were conserved to 50 % and randomized by NNK in the remaining 50 % of the library.

Materials and methods

In ABDHER2-I, each of the eleven positions targeted for randomization in the initial library were mutated to alanine. Because the two last randomized positions in the third helix were alanines in ABD_HER2-I, they were replaced by valine or phenylalanine. DNA sequences for the 13 mutants were assembled from pairs of mutagenic oligonucleotides (MWG Eurofins, Ebersberg,

Germany) that spanned the first 65 bases (eight forward primers for unmodified ABD_HER2-I and the S3A, D6A, T7A, Y10A, H1 1A, R14A and V15A mutants) or last 93 bases (seven reverse primers for unmodified ABD_HER2-I and the R35A, Y38A, A39V, A39F, A43V and A43F mutants) of the molecule with a 20 bp overlapping section over the region encoding the second helix. For each mutant, two oligonucleotides (100 pmol each) were assembled and extended by six cycles of PCR. The extended products were amplified in 35 cycles using external primers that also introduced restriction sites for EcoRI and Xhol. For mutants S3A, A43V and A43F, clone-specific external primers were used because mutations were located close to the ends. Purified cleaved fragments were ligated into the expression vector.

Data from binding analysis of mutants of ABD_HER2-I were used to design two affinity maturation libraries (Figure 3A). In a conservative library design, the four residues T7, Y10, H1 1 and Y38, which were necessary for retained HER2 binding, were maintained, whereas the remaining seven positions were randomized using NNK degeneracy. The library insert was assembled from two degenerate oligonucleotides (Integrated DNA

Technologies, Coralville, IA, USA) using a similar methodology as when constructing the point mutants. A forward oligonucleotide with T7, Y10 and H1 1 conserved and NNK in the remaining four positions was assembled with a reverse oligonucleotide that retained Y38 and introduced NNK at the remaining three positions. A semi-conservative library was also designed where eight forward and four reverse oligonucleotides were mixed to partly preserve the four positions mentioned above but still allow some variation. A39 was also maintained to the same degree due to a lack of data on the A39V/F-mutants. The forward oligonucleotides for the semi-conservative design either conserved all three of T7, Y10 and H1 1 , all possible pairs of two of them, one of them or used NNK for all three positions (eight primers in total). NNK was used in the remaining 4-7 positions. The reverse

oligonucleotides for the semi-conservative library contained NNK in all four randomized positions, NNK in three with Y38 or A39 conserved or NNK in two positions with both Y39 and A39 conserved (four primers in total). 200 pmol of forward primers (mixed at an equal ratio) were assembled with 200 pmol of reverse primers (mixed at an equal ratio), assembled and extended by six cycles of PCR and amplified in 15 cycles using Phusion DNA polymerase (NEB) and external primers that introduced restriction sites for EcoRI and Xhol for cloning in the phagemid vector pMLII (Aim T et al, supra). 150 pmol of each primer was used for the conservative library. The inserts were purified using a PCR purification kit (Qiagen), restricted with EcoRI-HF and Xhol-HF (NEB) and purified again. Prior to ligation, pMLII was restricted with the same enzymes, dephosphorylated and purified from a 1 % agarose gel as described above for the expression vector. Ligation products were purified using QIAquick gel extraction kit columns and transformed to

electrocompetent SS320 E. coli cells (Lucigen, Middleton, Wl, USA).

To examine whether the theoretical library design had been correctly represented in the libraries actually expressed, 174 individual clones from the conservative library and of 160 clones from the semi-conservative library were picked at random, and their DNA sequences were analyzed.

Results

To generate more information on the binding properties of ABD_HER2-I as input for affinity maturation, mutants in which the eleven positions targeted for mutagenesis in the initial library were individually mutated to alanine were produced. For A39 and A43, the alanines were mutated to either valine or phenylalanine. Valine was chosen as a small, non-polar residue that is only a little larger than alanine. Phenylalanine, on the other hand, represents a large aromatic residue with properties different from alanine. Twelve of the mutants could be cloned into the expression vector, and eleven were efficiently expressed and purified. The A39V and A39F mutants did not pass all steps; A39F did not yield any transformants after cloning and A39V could not be expressed, despite several attempts at different temperatures and expression times (not shown).

Characterization of the mutants by CD showed that all had spectra similar to ABD_HER2-I and melting temperatures between 55 and 71 °C (Table 1 ). SPR analysis clearly showed that four residues were critical for binding, because mutants T7A, Y10A, H1 1A and Y38A totally lost their HER2 binding even at a 15-fold increase (1500 nM instead of 100 nM) of the concentration being injected. The remaining mutants all bound HER2 with similar kinetics. As expected, only negligible effects on the original albumin binding were observed (Table 1 ). CD and SPR analysis was essentially performed as described in Example 5.

In Table 1 , kinetic parameters are presented as mean values with standard deviation. K_D was calculated from k_d/k_a and the number of replicates is indicated within brackets. No binding to HER2 could be observed even at elevated concentrations for the negative control ABD or the four mutants T7A, Y10A, H1 1A and Y38A. Melting temperatures (T_m) from CD measurements are also included. A)

ABD No binding at highest concentrations used

ABDHER2-1

SEQ I D NO:12 2.4 (±0.4)-10⁵ 1 .8 (±0.3)-10^"2 75 [13]

ABDHER2-1 (S3A) 3.4 (±0.8)-10⁵ 1 .4 (±0.3)-10^"2 41 [7]

ABDHER2-1 (D6A) 2.9 (±1 .0)-10⁵ 4.3 (±1 .4)-10^"2 148 [8]

ABDHER2-1 (T7A) Ruins binding

ABDHER2-1 (Y10A) Ruins binding

ABDHER2-1 (H11A) Ruins binding

ABDHER2-1 (R14A) 2.5 (±0.3)-10⁵ 3.5 (±0.7)-10^"2 140 [5]

ABDHER2-1 (V15A) 2.3 (±0.3)-10⁵ 1 .7 (±0.3)-10^"2 74 [6]

ABDHER2-1 (R35A) 3.3 (±1 .2)-10⁵ 4.5 (±1 .3)-10^"2 136 [5]

ABDHER2-1 (Y38A) Ruins binding

ABDHER2-1 (A43V) 2.4 (±0.6)-10⁵ 1 .1 (±0.2)-10^"2 46 [3]

ABDHER2-1 (A43F) 3.7 (±1 .0)-10⁵ 1 .7(±0.5)-10^"2 46 [5]

B)

HSA

k_a (NTs -¹) k_d (s-¹) K_D (nM) T_m (°C)

ABD 4

5.4 (±2.1 ) ^■10 4.4 (±0.4)-10 8.1 [6] >80

ABDHER2-1 5

SEQ I D NO:12 2.3 (±1 .5) ^■10 2.4 (±0.9)-10^"4 1 .0 [8] 68

5

ABDHER2-1 (S3A) 2.1 (±0.8) ^■10 2.7 (±0.6)-10^"4 1 .3 [6] 71

5

ABDHER2-1 (D6A) 1 .9 (±0.9) ^■10 3.3 (±0.6)-10^"4 1 .7 [6] 64

5

ABDHER2-1 (T7A) 2.0 (±0.2) ^■10 3.4 (±0.7)-10^"4 1 .7 [6] 65

5

ABDHER2-1 (Y10A) 2.8 (±0.1 ) ^■10 3.7 (±0.4)-10^"4 1 .3 [6] 60

5

ABDHER2-1 (H11A) 1 .1 (±0.2) ^■10 3.5 (±0.6)-10^"4 3.2 [6] 63

5

ABDHER2-1 (R14A) 2.0 (±0.8) ^■10 3.3 (±0.8)-10^"4 1 .7 [6] 61

5

ABDHER2-1 (V15A) 3.1 (±1 .9) ^■10 3.0 (±0.3)-10^"4 1 .0 [6] 62

5

ABDHER2-1 (R35A) 1 .5 (±0.2) ^■10 3.4 (±0.4)-10^"4 2.3 [6] 59

5

ABDHER2-1 (Y38A) 1 .6 (±0.5) ^■10 3.7 (±0.7)-10^"4 2.3 [6] 55

5

ABDHER2-1 (A43V) 1 .4 (±0.7) ^■10 1 .9 (±1 .0)-10 ⁴ 1 .4 [6] 56

5

ABDHER2-1 (A43F) 1 .4 (±0.5) ^■10 2.5 (±0.7)-10^"4 1 .8 [5] 65 Table 1. Affinities for HER2 (A) and HSA and melting temperatures (T_m) (B) of non-randomized ABD, ABD_HER2-I and point mutants of ABD_HER2-I. Binding data from the different mutants of ABD_HER2-I suggested that there was flexibility in the sequence to design new libraries for affinity maturation. Two parallel approaches were taken, a conservative approach that maintained all four residues, the replacement of which by alanine had led to disruption of the HER2 binding, and a semi-conservative approach that also allowed some variation in those positions (Figure 3A). In the semi- conservative library, T7, Y10, H1 1 and Y38 were all conserved to 50 % in the primer mix used for library assembly. Due to the lack of data on A39 mutants, this position was hypothesized to be important for expression and was therefore preserved to 50 % in the semi-conservative approach. The remaining 50 % and six additional positions were randomized by NNK.

Replacements of the four residues that were fully retained in the conservative library to other residues than alanine may allow selection of further improved binding molecules from the semi-conservative library.

After cloning, the libraries were transformed to E. coli SS320 to yield library sizes of 3.2 x 10⁷ (theoretical size 1 .3 x 10⁹) for the conservative library and 2.2 x 10⁷ (theoretical size 2.0 x 10¹⁴) for the semi-conservative library. Library sizes were calculated from transformation frequencies (plating) and number of electroporations performed. Sequence evaluation of 174 clones picked at random from the conservative library demonstrated a strong agreement with the design. As desired, T7, Y10, H1 1 and Y38 were present in all sequences and all 20 amino acids and the amber stop codon encoded by NNK were observed at least once in all of the remaining seven positions. Data from 160 clones picked at random from the semi-conservative library identified all 20 amino acids and the amber stop in all eleven positions.

Moreover, the degree of conservation of T7, Y10, H1 1 , Y38 and A39 were 48 % on average with a standard deviation of 2.9 % (as compared to the theoretical values 52 % for Y and H that are encoded by 1/32 codons in NNK or 53 % for A and T that are encoded by 2/32). 22 sequences contained deletions of 1 -9 bases, three sequences with single base insertions and one with a single base substitution were observed in addition to the 160 correct inserts mentioned above. Example 4

Phage display selection of second generation HER2 binding ABD variants Summary

To enrich higher affinity HER2 binding variants from the two affinity maturation libraries designed in Example 3, they were subjected to four rounds of phage display selection in four parallel tracks that branched out into eight tracks (A-H) from cycle two, as shown in Figure 3B. Materials and methods

For each library, half of the selection tracks from cycle two were performed in the presence of 1 μΜ unlabeled HSA and half were performed without HSA. HER2-Fc chimera (Cat. No. 1 129-ER, R&D Systems) was used as target in half of the tracks and biotinylated (using Biotin-XX-succinimidyl ester B-1606, Invitrogen) HER2-His6 (Sino Biological) was used in the remaining tracks (Figure 3B). To increase the stringency and limit avidity effects, only the monovalent HER2-His6 was used in the last cycle of all tracks. Dynabeads® M280 streptavidin (Invitrogen) were used for capture in tracks with biotinylated HER2-His6 and Dynabeads® Protein A were used for HER2-FC. In tracks using HER2-Fc, pre-selection against 50 nM human polyclonal Fc (Bethyl Laboratories) was performed in all cycles. Target concentrations down to 5 nM were used, and up to 15 washes were performed (Figure 3B). The selection was performed essentially as described in Example 1 with one main difference, bacteria infected with the selection eluates were simultaneously (without a plating step in-between) infected with helper phage (M13K07, NEB), harvested by centrifugation and inoculated to fresh medium. Following ON incubation new phage stocks were prepared. 3 % PBSTB was used for washing and 50 mM glycine-HCI pH 2.2 was used as elution buffer.

Figure 3B illustrates the selection strategy for affinity maturation. In brief, four rounds (1 -4) of phage display selection were performed in eight parallel tracks (A-H). Tracks A-D were selected from the conservative library and E-H from the semi-conservative library. Tracks A, B, E and F used HER2- His6 as target, tracks C, D, G and H used HER2-Fc. HER2-His6 was used in all tracks in the fourth selection round. Selections A, C, E and G were performed in the absence of HSA whereas tracks B, D, F and H contained 1 μΜ of HSA in round 2-4. The number of washes was 3, 5, 10 and 15 in rounds 1 -4.

ABDHER2-I and the starting sequence ABD (SEQ ID NO:27) were cloned into the staphylococcal display vector pSCABDI (Nilvebrant J et al, supra) using Nhel and Xhol and the leader sequence DEAVDANS. The output from the eight (A-H) phage display selection tracks for affinity maturation were cloned into pSCABDI using the same procedure. At least 60000 phagemid-containing RR1 AM15 cells were grown ON for plasmid preparation. ABD variant inserts were amplified by PCR, restricted, purified and ligated in pSCABDI . Twelve individual colonies of S. carnosus per track (A-H) were PCR-screened and sequenced to enable a comparison with sequence data obtained after phage display selection.

Example 5

Expression, purification and characterization of selected ABD variants

Materials and methods

All ABD variants, cloned in the expression vector as described in Example 4, were transformed to RR1 AM15 and sequence verified plasmids were transformed to Rosetta (DE3) E. coli (Novagen, Madison, Wl, USA). Single colonies were inoculated in 10 ml TSB supplemented with 50 pgml^"1 kanamycin and 20 pgml^"1 chloramphenicol and grown ON. 100 ml of TSB supplemented with 5 % (w/v) yeast extract, and the same antibiotics were inoculated with 1 ml of ON culture, grown to log phase at 37 °C and 150 rpm, induced with IPTG to 1 mM final concentration and incubated at 25 °C ON. Cells were harvested by centrifugation, re-suspended in Tris-buffered saline (TST; 25 mM Tris-HCI, 200 mM NaCI, 1 mM EDTA, 0.05 % (v/v) Tween 20, pH 8.0) and lysed by sonication (Vibra-Cell, Sonics & Materials Inc.,

Newtown, CT, USA). Clarified lysates were filtered 0.45 μιτι and loaded onto 7.5 ml column volume (CV) HSA-sepharose gravity flow columns equilibrated with 10 CV of TST. The columns were washed with 10 CV of TST, 7 CV of 5 mM NH₄Ac pH 5.5 and eluted in 1 ml fractions with 0.5 M HAc pH 2.8. Pure fractions, as determined by sodium dodecyl sulphate polyacrylamide gel electrophoresis (SDS-PAGE), were dried by evaporation in a Savant

AES2010 SpeedVac system (Thermo Scientific, Rockford, IL, USA), re- suspended in PBS and pooled. Protein molecular weights were determined by mass spectrometry (MS) on a 6520 Accurate Q-TOF LC/MS (Agilent, Santa Clara, CA, USA). Secondary structure content and melting temperatures (T_m) were evaluated by circular dichroism (CD) at 25 °C on a Jasco J-810 Spectropolarimeter (Jasco, Essex, UK). Proteins were diluted in PBS to 0.4 mg/ml, scanned in a 1 mm quartz crystal cuvette at wavelengths from 250-195 nm, heated to 90 °C (at 5 °C s^"1) with simultaneous recording of the CD signal at 221 nm followed by a new wavelength scan once the sample temperatures returned to 25 °C. Melting temperatures were determined from thermal denaturation curves and refolding ability was assessed by comparing CD spectra before and after heating.

The concentration of ABD_HER2-I, the most common variant observed after the selection, was determined by amino acid analysis

(Aminosyraanalyscentralen, Uppsala, Sweden). All other protein

concentrations were determined by the bicinchoninic acid (BCA) assay, using ABDHER2-I as a calibrant, according to the manufacturer's protocol (Thermo Scientific). Proteins and standards were diluted in PBS, BCA reagents A (containing bicinchoninic acid) and B (containing 4 % cupric sulfate) were mixed 50 parts to 1 and 200 μΙ was added to 25 μΙ of diluted protein samples and standards in a microplate. After 30 min incubation at 37 °C, absorbance at 562 nm was measured on a plate reader. Protein concentration in mgml^"1 was calculated from a standard curve based on a serial dilution of amino acid analyzed ABD_HER2-I -

Surface plasmon resonance (SPR) spectroscopy was used to analyze the target-binding ability of the selected candidates and, when appropriate, determine binding affinities for HER2 and HSA. HER2-Fc chimera (R&D Systems), HER2-His6 (Sino Biological, Beijing, China), Fc (Bethyl

Laboratories) and HSA were immobilized on General Layer Medium (GLM) sensor chips using a ProteOn XPR36 Protein interaction array system (Bio- Rad, Hercules, CA, USA). Proteins were diluted to 5 pgrnl^"1 in 10 mM NaAc pH 4.5 (2 pgml^"1 for HSA) and immobilized by amine coupling according to the supplier's recommendations. Immobilization levels of 2100-3500 response units (RU) for HER2-Fc, 2000-4300 RU HER2-His₆, 600 RU Fc and 1 100- 1600 RU HSA were used. Surfaces were deactivated by 1 M ethanolamine and one channel was used as a blank for referencing. An association time of 252 s and dissociation for 1500 s was used in all injections. All analyses were performed at 25 °C with a flow rate of 50 μΐιτιίη^"1 using PBST pH 7.4 as running buffer and 20 mM HCI for regeneration. For kinetic determination five concentrations (different dilution series (1 :3) starting from 30 or 100 nM) and a blank were injected. All kinetic parameters were determined at least in duplicates using at least two different immobilization levels, dilution series or sensor chips. Data were fitted to a 1 :1 Langmuir binding isotherm using the ProteOn Manager software 3.1 .0.6. For alanine mutants of ABD_HER2-I , concentrations of up to 1500 nM were used in the SPR analysis in cases when 100 nM did not result in detectable binding. Affinity matured ABD molecules were only evaluated on surfaces immobilized with HER2-His6 (Sino Biological) and HSA.

SPR was used to examine the polypeptides selected from the affinity matured libraries in more detail. Simultaneous binding to HSA was assessed using the co-inject command with injection of 100 nM ABD variant followed by either 50 nM HSA or 50 nM HER2-His₆ over surfaces immobilized with HER2- His6 or HSA. As a complementary assay, 100 nM ABD was pre-incubated with a 5-fold molar excess of HSA for 1 h prior to injection over immobilized HER2-His6 and compared to a simultaneous injection of ABD alone. In an attempt to map the binding site on HER2, binding to six recombinant sections spanning 97-167 amino acid regions of the extracellular domain of human HER2 was evaluated. The recombinant HER2 derived sequences were expressed as fusions to an albumin-binding protein and an N-terminal His6- tag and purified by immobilized metal ion affinity chromatography (IMAC). The six HER2-derived proteins spanned amino acids 42-186 (including most of domain 1 ), 236-363 (domain 2), 274-400 (most of domain 2), 347-492 (most of domain 3), 364-530 (domain 3) and 531 -626 (domain 4) and were injected at 100 nM over immobilized ABD_HER2-mati (SEQ ID NO:1 ). Three of the HER2-derived proteins (the ones corresponding to full domain sequences, see above) were also immobilized by amine coupling and tested for binding to all ABD variants injected at 100 nM. In the same experiment, immobilized ABD_HER2-mati was evaluated for binding to HER2 alone (20 nM) or 20 nM

HER2 pre-incubated (1 h) with 100 nM of ABD_HER2-mati-ABD_H ER2-mati i to verify that all candidates recognized the same epitope on HER2.

Results

Following four rounds of phage display selection aimed at affinity maturation and selection of molecules with the ability to bind HER2 in the presence or absence of HSA, individual colonies derived from all eight parallel selection tracks (A-H, Figure 3B) were sequenced. A total of 438 full length ABD-sequences were analyzed; 268 from selections from the conservative library and 170 from the semi-conservative library. 92 unique sequences were found in the data from the conservative library, of which 34 were observed more than once and the most common sequence appeared 34 times. Data from the semi-conservative library indicated a similar degree of convergence, because 63 unique sequences were found, 19 were found repeatedly and the most common sequence was seen in 29 colonies. A comparison between selections with (167 sequences) and without (271 ) presence of HSA resulted in 50 unique sequences with HSA (19 repeated and the most common 32 times) and 1 18 unique without HSA (35 repeated and one occurring 19 times). This may indicate a larger degree of

convergence in the presence of HSA, which is also expected when more selection pressure is applied.

Interestingly, the four residues that were retained in the conservative library (T7, Y10, H1 1 and Y38) were observed in every one of the 438 clones from both libraries, in selections both with and without HSA. In the semi- conservative library, only a fraction (-1/16) of the initial, non-selected library contained this combination of residues. This clearly demonstrates the importance of those residues and verifies the results from the alanine scanning of the binding residues of ABD_HER2-I - NO differences were observed between the two forms of HER2 used as targets. Clustering of all sequences showed that many only differed in one or a few positions and, interestingly, sequences from both libraries and selections with/without HSA did not fall into separate clusters. Only three small differences were observed between the groups of sequences from selections with and without HSA when sequence logotypes were compared (Figures 3C and D). Figures 3C and D show sequence logotypes from sequenced colonies from selections without HSA present (271 sequences) and with HSA present (167 sequences),

respectively. The eleven randomized positions are shown from left (N- terminus) to right (C-terminus) and their location in the full-length sequence of 46 amino acids is indicated. Logotypes were generated using Weblogo 3.3 (Crooks et ai, Genome Research 14:1 188-1 190 (2004)) The overall height of each residue corresponds to its degree of conservation and the height within each stack relates to the relative frequency. The maximum sequence conservation per site is described by Iog₂(20) for 20 possible amino acids (^« 4.3 bits). Proline was the most common residue in position 43 with HSA present (35 %) and was seldom (1 %) observed in data where HSA was not included. M35 was often found with HSA present (17 %) and never seen in the opposite data set. Lysine and arginine were however the most common residues in this position in both data sets. All sequences from selections with HSA present contained E6 as compared to almost all clones (92 %) from selections without HSA. As a measure of enrichment, the proportion of the initial semi- conservative library that contained E6 in this NNK-position and had T7, Y10, H1 1 and Y38 preserved is expected to be around 0.2 % (-1/512). The corresponding proportion in the conservative library is about 3 % (1/32).

Sequence analysis also revealed that eight mutations in the scaffold that were not intentionally included in the libraries were present between one and twelve times among the sequences of enriched clones. Four mutations were observed in tracks without HSA (D13E (2 times), D13T (4), V34A (1 ) and A36T (2)) and eight in tracks with HSA (A2G (1 ), D13T (7), Y20C (2), A28V (12), E32K (1 ), G33A (4), L37Q (1 ), and A44V (1 )). Several mutations occurred in clones with non-identical sequences, which means that the mutation was enriched in more than one individual clone, and many that were found after selections with HSA present affect residues in close proximity to the previously mapped HSA-binding site (Linhult M et al (2002) Protein Sci 1 1 (2):206-13).

Eleven clones that were frequently observed in the sequence data (between 1 1 -34 times, Figure 4A), and also represented different regions of the global sequence cluster, different selection tracks as well as different selection pressures, were chosen for further characterization. These 1 1 selected clones were denoted ABD_HER2-mati - ABD_HER2-matn , and are listed in Figure 1 and in the accompanying sequence listing as SEQ ID NO:1 -1 1 . By way of deduction, the mutations conferring HER2 binding to the starting ABD sequence in ABD_HER2-I and ABD_HER2-mati - ABD_HER2-matn were also virtually introduced into a previously described ABD variant sequence disclosed and studied in WO2012/004384, and the resulting, deduced, sequences are listed in Figure 1 and the sequence listing as SEQ ID NO:13-24.

Track D was underrepresented in the sequence data set. Two sequences contained mutations in the scaffold residues that were not intentionally diversified in the libraries. All except one (ABD_HER2-mat8; SEQ ID NO:8) could be purified by HSA affinity chromatography (Figure 4B) and subjected to kinetic evaluation by SPR. Interestingly, ABD_HER2-mat8 contained the A28V scaffold mutation that is in close proximity to albumin binding residues (Linhult M et al, supra) (Figure 4A). All candidates bound HER2 with high affinity; 9- to 83-fold improvements in HER2 binding affinity (K_D) compared to ABD_HER2-I were observed (Table 2, Figure 4B). Albumin binding affinities were in the low nanomolar range. CD demonstrated spectra similar to the starting polypeptide ABD (SEQ ID NO:27) and melting temperatures of from 60 °C to above 80 °C. Most variants were able to refold completely after heating (Figure 4C).

All clones were analyzed for their ability to bind HER2 and HSA simultaneously, using several SPR-based experimental setups with co- injections or pre-incubations. None of the tested candidates was able to bind both targets simultaneously and binding of one protein blocked binding of the second. Binding to six different fragments that spanned regions of the extracellular domain of human HER2 was also evaluated, but none of the ABD variants bound any of the protein fragments included in the analysis. As expected from sequence similarities and their common origin, all ABD variants were shown to bind to the same epitope on HER2 in a competition assay between immobilized ABD_HER2-mati and injected HER2 pre-incubated with an excess of one ABD variant molecule at a time.

In Table 2, kinetic parameters are presented as mean values with standard deviation. K_D was calculated from k_d/k_a and the number of replicates is indicated within brackets.

A)

HER2

k_a (M^_1s k_d (s ¹) K_D (nM)

6.4 (±1 .1 )-10^"4 2.3 [7]

4.4 (±0.5)-10^-4 2.0 [7]

ABDHER2-mat3 5

SEQ ID N0:3 3.2 (±0.4) 10 3.3 (±0.5)-10^"4 1 .1 [7]

ABDHER2-mat4 5

SEQ ID N0:4 5.8 (±1 .0) 10 6.0 (±0.8)-10^"4 1 .0 [7]

ABDHER2-mat5 5

SEQ ID N0:5 3.5 (±0.7) 10 1 .1 (±0.1 )-10 ³ 3.1 [7]

3.6 (±0.4)-10 ⁴ 0.9 [7]

ABDHER2-mat7 5

SEQ ID N0:7 3.9 (±0.5) 10 1 .1 (±0.2)-10^"3 2.9 [7]

ABDHER2-mat9 5

SEQ ID N0:9 2.9 (±0.4) 10 2.7 (±0.4)-10^"3 9.4 [7]

ABDHER2-mat10 5

SEQ ID NO:10 2.6 (±0.4) 10 2.1 (±0.2)-10^"3 8.1 [7]

ABDHER2-mat11 5

SEQ ID N0:11 2.0 (±0.3) 10 7.8 (±0.8)-10^"4 3.9 [7]

B)

HSA

k_a (M^_1s ^]) k_d (s-¹) K_D (nM)

2.3 (±0.8)-10^"3 20 [2]

2.7 (±0.4)-10^"4 2.9 [2]

ABDHER2-mat3 5

SEQ ID N0:3 2.0 (±0.1 ) 10 2.2 (±0.3)-10^"4 1 .1 [2]

ABDHER2-mat4 5

SEQ ID N0:4 2.3 (±0.5) 10 3.0 (±0.5)-10⁴ 1 .3 [2]

ABDHER2-mat5 5

SEQ ID N0:5 1 .3 (±0.3) 10 4.6 (±0.6)-10^"4 3.7 [2]

ABDHER2-mat6 5

SEQ ID N0:6 1 .9 (±0.1 ) 10 2.4 (±0.4)-10^"4 1 .3 [2]

4.9 (±0.4)-10^"4 3.3 [2]

ABDHER2-mat9 5

SEQ ID N0:9 1 .2 (±0.3) 10 2.0 (±0.5)-10 ³ 17 [2]

ABDHER2-mat10 4

SEQ ID NO:10 8.3 (±1 .1 ) 10 7.7 (±0.1 )-10^~4 9.3 [2]

ABDHER2-mat11 4

SEQ ID N0:11 6.8 (±0.7) 10 3.7 (±0.2)-10^"4 5.5 [2]

Table 2. Affinities of matured ABD variants for HER2 (A) and HSA (B) Example 6

Cell-display evaluation of HER2 binding ABD variants Summary

In this example, the ability of the ABD_HER2 polypeptides to bind HER2 in the presence of HSA using a staphylococcal cell surface display assay is studied. Further, epitope binding specificity was addressed using said staphylococcal cell surface display assay to compare output from selection track A with HER2 binding Z variant molecules Z02819 (SEQ ID NO:25) and Z477 (SEQ ID NO:26).

Materials and methods

To facilitate better visualization of the enriched ABD variant

populations and the selection of top candidates, staphylococcal cell surface display was employed to display the HER2 binding ABD variants on cells. Ligated and sequence verified plasmid pSCABDI encoding ABD_HER2-I was transformed to electrocompetent Staphylococcus carnosus TM300 using a previously described protocol (Lofblom J et al (2007) Appl Environ Microbiol 73(21 ):6714-21 ) and analyzed on a Gallios™ flow cytometer (Beckman Coulter, Brea, CA, USA). Briefly, a single colony was grown in TSB

supplemented 5 % (w/v) yeast extract with 20 pgml^"1 chloramphenicol at 37 °C ON. Approximately 10⁶ cells were washed with PBS supplemented with 0.1 % (v/v) Pluronic ® F108 NF Surfactant (PBSP; BASF Corporation, Mount Olive, NJ, USA), pelleted by centrifugation, re-suspended in PBSP with 100 nM HSA-Alexa Fluor 488 conjugate (Invitrogen, labeled according to the supplier's recommendations) or 50 nM biotinylated human HER2-His6 (Sino Biological) and incubated end-over-end for 1 h at RT. After washing with ice cold PBSP, cells were incubated with polyclonal human IgG-Alexa Fluor 647 conjugate (Invitrogen) to measure expression level and, for detection of HER2, streptavidin-phycoerythrin (Invitrogen). After a 30 min incubation on ice, cells were washed and re-suspended in ice cold PBSP and analyzed on the flow cytometer.

Output from the eight parallel affinity maturation selections (A-H) were transformed to electrocompetent SS320 E. coli and plasmids were prepared from ON SS320 cultures using a Jetstar 2.0 plasmid maxiprep kit (Genomed, Bad Oeynhausen, Germany) and transformed to S. carnosus. Cells from all tracks were analyzed on the flow cytometer with 50 nM HER2, 50 nM HER2 and 1 μΜ HSA (unlabeled) or 50 nM HSA-Alexa Fluor 488, as described above. Staphylococci expressing ABD (starting sequence) were used as a negative control, whereas cells expressing a HER2 binding Z variant (Z02891 (SEQ ID NO:25); Feldwisch J ef a/ (2010) J Mol Biol 398(2):232-47) as a positive control.

A competition assay was performed to evaluate if HER2 binding ABD variants displayed on cells or in solution competed with the HER2 binding Z variant Z02891 for binding to soluble HER2. Staphylococci expressing Z02891 or clones from selection track A were incubated with 50 nM

biotinylated HER2 (Sino Biological) or 50 nM biotinylated HER2 pre-incubated (1 h, RT) with 0.5 μΜ Z477 (a HER2 binding Z variant that is closely related to Z02891 ; Orlova A (2006) Cancer Res 66(8):4339-48) or 50 nM biotinylated HER2 pre-incubated with 0.5 μΜ ABD_HER2-mati■ Cells were washed and incubated with secondary reagents as described above and analyzed by flow cytometry.

Results

Because of the large sequence diversity observed after four rounds of phage display selection, staphylococcal cell surface display was applied to allow better analysis of the selection output and facilitate future fine sorting of top performing clones using a previously described method (Kronqvist N et al (2008) Protein Eng Des Sel 21 (4):247-55), Nilvebrant J et al, supra).

ABDHER2-I and the starting sequence ABD were successfully displayed on staphylococci and bound to both HER2 and HSA (ABD_HER2-I) or to only HSA (ABD) in a flow cytometry based binding assay. The output from selections A- H was cloned in batch to pSCABDI using the same strategy with a leader sequence that is also present in the expression and phagemid vectors.

Electroporation of batch-cloned material from tracks A-H resulted in 1 .4 x 10⁵- 2.6 x 10⁶ S. carnosus transformants, which is in excess compared to the minimum number of phagemid-carrying E. coli used for plasmid preparation and sub-cloning and also a large number compared to the sizes of the libraries used for affinity maturation. Sequencing of a total of 96 individual S. carnosus colonies (12 per track) revealed 56 unique sequences and 10 sequences that occurred more than once. All distinct sequence characteristics that were observed in the larger E. coli data set (438

sequences described above) were seen.

Flow-cytometric analysis of HER2 binding demonstrated strong binding signals for all tracks (Figure 5). HER2 binding was detected through streptavidin-phycoerythrin (y-axis) and expression levels were detected by IgG Alexa Flour 647 conjugate (x-axis) in Figure 5. The contours in Figure 5 represent cell density in the respective regions corresponding to 20, 40, 60 or 80 % of the maximum density observed. Each sample is represented by 25- 30 000 events.

Interestingly, the decrease in binding signals upon addition of a 20-fold molar excess of unlabeled HSA was more apparent in tracks selected without HSA present (A, C, E and G) as compared to tracks where HSA was present during selection (B, D, F and H), see Figure 5. To investigate if this loss of HER2 binding was a result of lower albumin binding affinities of clones in batches selected in the presence of HSA, cells from all tracks were incubated with labeled HSA (Figure 6). HSA binding is shown on the y-axis and expression levels were detected by IgG Alexa Flour 647 conjugate (x-axis) in Figure 6. This analysis revealed that clones selected only with HER2 present were more homogenous and had binding patterns to HSA similar to

ABDHER2-I and the starting sequence ABD. However, clones from selection tracks B, D, F and H, selected with an excess of HSA present, were more heterogeneous and formed two distinct populations with high and low binding signals to HSA, respectively.

Because all ABD variants are derived from the same first generation HER2 binding molecule (ABD_HER2-I), they are expected to recognize the same epitope on HER2. To investigate if this epitope is the same as, or overlapping with, the epitope of the HER2 binding Z variant Z02891 (Feldwisch J et al, supra), a flow-cytometry based blocking assay was used to compare the HER2 binding of ABD variants from track A to the Z variant. Staphylococcal cells expressing either Z02891 or clones from track A were incubated with only HER2 or HER2 pre-incubated with an excess of ABD_HER2-mati or of the Z variant Z477 (Figure 7). This assay showed that both molecules, as expected, are able to block the binding of HER2 to cells expressing the same molecule as used for blocking. On the other hand, binding of HER2 to Z02891 on cells cannot be blocked by ABD_HER2-mati , and HER2 binding to cell-displayed ABD variants from track A cannot be blocked by a soluble Z variant. Therefore, it seems as though the HER2 binding ABD variants and the HER2 binding Z variant recognize two distinct epitopes on HER2.

Example 7

Enrichment of HER2 binding ABD variants using staphylococcal display

Summary

Staphylococcal display was applied to further enrich higher affinity HER2 binding variants. The output from the eight tracks following four rounds of phage display selection as described in Example 4 were subcloned to a staphylococcal display vector. Selections were performed in four rounds in the presence or absence of HSA. DNA sequencing of isolated clones after two and four selection rounds, respectively, identified both new unique variants and variants identical in sequence to ABD_HER2 binders previously identified by phage display.

Materials and methods

Construction of sub-libraries: PCR-products derived from plasmid prepped phagemids from the eight parallel tracks (A-H) after four rounds of phage display selection as described in Example 4 were used as templates for cloning to the staphylococcal display vector pSCABDI . The constructs were digested over night at 37 °C using restriction enzymes Xho\ and Nhe\ (NEB) after gel extraction purification from 2 % Seakem GTG agarose (Lonza, Basel, Switzerland) with a gel extraction kit (Qiagen, Venlo,

Netherlands). The digested fragments were then ligated into the

staphylococcal surface display vector pSCABDI , digested with the same enzymes and purified by gel extraction.

The eight assembled sub-libraries (one per phage display selection track A-H) were purified and transformed by electroporation into

electrocompetent E. coli SS320 (Lucigen). The transformed cells were grown over night in TSB supplemented with 100 g/ml ampicillin and the plasmid DNA was extracted and purified using a Jetstar maxiprep kit (Genomed, Lohne, Germany). The purified plasmids were further purified by

chlorophorm:phenol extraction and electroporated into electrocompetent S. carnosus cells. Fluorescence activated cell sorting: The sub-libraries were inoculated in TSB+YE supplemented with chloramphenicol (20 pg/ml) for overnight growth at 37 °C and 150 rpm. The following day, cells were harvested by centrifugation and washed in PBSP before addition of biotinylated HER2-His6 (Sino Biological). Cells were incubated with gentle mixing for 1 h at RT prior to washing with ice-cold PBSP and labeling with Streptavidin, R- Phycoerythrin conjugate (SAPE; Invitrogen). Binding signals were normalized to the expression level using IgG labeled with Alexa-647. Cells were washed and re-suspended in ice-cold PBSP.

Fluorescence activated cell-sorting was performed using a MoFlo

Astrios flow cytometer (Beckman Coulter, Indianapolis, IN). Four rounds of sorting were performed using decreasing concentrations of HER2; 50 nM in round one, 10 nM in round 2 and 1 nM in round 3, and in the absence or presence of 1 μΜ of unlabeled HSA. An off-rate selection strategy was applied in the fourth round using 50 nM HER2 for 1 h followed by a wash and a subsequent incubation with 100 nM unlabeled HER2 for 3 h prior to sorting. Prior to round 1 , the sub-libraries were pooled to 4 tracks as follows: A+C (originating from conservative design), B+D (originating from conservative design + HSA), E+G (originating from semi-conservative design) and F+H (originating from semi-conservative design + HSA). Prior to round 3, the sub- libraries were pooled to 2 tracks based on whether the selection had been performed in the presence or absence of albumin. For each round of sorting, approximately ten times the library size was analyzed in the flow cytometer. Subsequently, sorted cells were inoculated in TSB+YE with chloramphenicol (20 pg/ml) for overnight amplification prior to the next sorting round. Finally, isolated cells were spread on agar plates containing chloramphenicol and individual colonies were picked for BigDye Thermo Cycle Sequencing reactions using an ABI Prism 3700 instrument (Applied Biosystems, Foster City, CA).

Results

The isolation of high-affinity binders was performed by phage display (described in Example 4), followed by four rounds of flow-cytometric sorting. DNA sequencing was carried out on a set of clones isolated in the second and fourth round of sorting, respectively. In total, 74 new and unique polypeptide variants were identified, and these are listed in Figure 1 and in the accompanying sequence listing as SEQ ID NO:30-103. By way of deduction, the mutations conferring HER2 binding to the starting ABD sequence in these variants were also virtually introduced into a previously described ABD variant sequence disclosed and studied in WO2012/004384, and the resulting, deduced, sequences are listed in Figure 1 and the sequence listing as SEQ ID NO:104-177. Of the 22 unique variants enriched and identified after four rounds of sorting, three of the variants had been identified by phage selection (SEQ ID NO:3, SEQ IQ NO:7, SEQ ID NO:12; see Examples 4-5) and 19 were new variants (SEQ ID NO:30-44 and 90-93).

Example 8

Cell display evaluation of additional HER2 binding ABD variants

Summary

In this example, the ABD_HER2 polypeptides enriched and identified after four rounds of flow cytometric sorting were subjected to affinity ranking in terms of 1 ) HER2 binding, 2) HSA binding, and 3) HER2 binding in the presence of HSA, using staphylococcal surface display and flow cytometric screening.

Materials and methods

The cell display evaluation was performed essentially as described in Example 6. In brief, each of the selected unique staphylococcal clones from the fourth round of sorting were inoculated in TSB+YE with chloramphenicol (20 Mg/ml) and grown overnight at 37 °C and 150 rpm. Cells were pelleted by centrifugation and re-suspended in PBSP with 5 nM of biotinylated HER2, 5 nM of biotinylated HER2 and 1 μΜ HSA (unlabeled), or 20 nM HSA-Alexa Fluor 488. After 1 h incubation at RT with gentle mixing, cells were washed with ice-cold PBSP and labeled with SAPE and IgG-Alexa Fluor 647 for 30 minutes on ice. Finally, cells were washed and re-suspended in ice-cold PBSP. All samples were ranked based on the ratio between mean

fluorescence intensities (MFI) from binding to HER2 or HSA and cell surface expression signals in a Gallios flow cytometer (Beckman Coulter). In addition, the HER2 binder ABD_HER2-I (SEQ ID NO:12; see Examples 4-5), the alkali stabilized ABD (SEQ ID NO:27) and a HER2-binding Z variant molecule Z02891 (SEQ ID NO:25) were analyzed for comparison. Results

The 22 unique clones identified from the fourth round of flow cytometric sorting as described in Example 7 were screened for HER2 and HSA binding. Figure 8A shows the screening of HER2 binding, Figure 8B shows the screening of HSA binding, and Figure 8C shows the HER2 binding in the presence of HSA. All variants bound to HER2 (Figure 8A). Seven variants were shown to bind HER2 in the presence of unlabeled HSA (Figure 8C), but the same variants were shown not to not bind HSA (Figure 8B).

Example 9

Expression, purification and characterization of selected HER2 binding ABD variants Materials and methods

Eleven of the HER2 binding ABD variants identified in Example 7 (SEQ ID NO:32, 33, 37, 38, 39, 43, 44, 87, 90, 91 and 93) were selected for further characterization. Variants conferring albumin binding were cloned, expressed and purified essentially as described in Example 4 and 5. Variants unable to bind albumin were expressed with a histidine tag and purified by IMAC.

Subsequent analyses by mass spectrometry (MS), circular dichroism (CD) spectroscopy and surface plasmon resonance (SPR) spectroscopy were performed essentially as described in Example 5. In contrast to Example 5, however, murine epidermal growth factor receptor 2 (mHER2) and mouse serum albumin (MSA) were included in the SPR analysis.

Results

All eleven subcloned candidates were successfully expressed and purified in high yields. The correct molecular weight for each polypeptide variant was confirmed by HPLC-MS. Analysis by circular dichroism

spectroscopy showed that all variants refolded after heat denaturation, and the melting points, Tm, were estimated to be in the range of 60-75 °C.

The kinetic parameters for binding of the ABD_HER2 variants to the human HER2, HSA and MSA are shown in Table 3. The kinetic parameters are presented as mean values with standard deviations. K_D was calculated from kd ka, and the number of replicates is indicated within brackets. No binding to mHER2 was detected for any of the variants. A)

B)

HSA

k_a (M-V¹) k_d (s¹) K_D (nM)

SEQ ID NO:32 1.1 (±0.4)-10 1.4 (±0.9)10 1.3 [2]

SEQ ID NO:33 0.9 (±0.5)-10⁵ 8.9 (±5.2)-10^"5 1.0 [2]

SEQ ID NO:37 No binding

SEQ ID NO:38 1.6 (±0.7)-10⁵ 1.6 (±0.9)-10^"4 1.0 [2]

SEQ ID NO:39 1.1 (±0.7)-10⁵ 1.1 (±0.5)- 10^"4 1.0 [2]

SEQ ID NO:43 1.1 (±0.4)-10⁵ 7.0 (±3.5)10^" 0.6 [3]

SEQ ID NO:44 1.6 (±0.2)-10⁵ 1.3 (±0.1)10^" 0.8 [6]

SEQ ID NO:87 9.8 (±0.2)-10⁴ 1.4 (±0.2)10^" 1.4 [4]

SEQ ID NO:90 No binding

SEQ ID N0:91 Weak binding

SEQ ID NO:93 No binding C)

MSA

k_a (MV¹) k_d (s ¹) K_D (nM)

SEQ ID NO:32 5.6 -10 4.6 -10 14 [1 ]

SEQ ID NO:33 3.2 -10⁴ 7.9 -10^"4 14 [1 ]

SEQ ID NO:37 No binding

SEQ ID NO:38 6.0 -10⁴ 8.3 -10^"4 14 [1 ]

SEQ ID NO:39 2.4 -10⁴ 5.4 -10^"4 14 [1 ]

SEQ ID NO:43 5.5 (±2.0)-10⁴ 2.8 (±0.2)-10^"4 5.1 [2]

SEQ ID NO:44 ND ND ND

SEQ ID NO:87 ND ND ND

SEQ ID NO:90 No binding

SEQ ID NO:91 Weak binding

SEQ ID NO:93 No binding

Table 3: Affinities ofABD variants for HER2 (A), HSA (B) and MSA (C)

Example 10

Competition binding assay using trastuzumab

Summary

A competition binding assay was performed, in order to assess whether the HER2 binding ABD variants interact with HER2 at the same epitope as the clinically approved therapeutic antibody trastuzumab

(Herceptin®), which interferes with the HER2/neu receptor signaling. Materials and methods

Staphylococci expressing either the HER2 binding ABD variants SEQ ID NO:43 or SEQ ID NO:90, the HER2 binding Z variant Z02891 (SEQ ID NO:25) or a scFv derived from the humanized mAb trastuzumab, Herceptin®- scFv, were each incubated with 5 nM biotinylated HER2 (Sino Biological) or 5 nM biotinylated HER2 pre-incubated (1 h, RT) with 50 nM of SEQ ID NO:43, SEQ ID NO:90, Z477 (SEQ ID NO:26) or Herceptin-scFV, respectively. Cells were washed and incubated with secondary reagents as described above and analyzed by flow cytometry. Results

Because all ABD_HER2 variants are derived from the same first generation HER2 binding molecule (ABD_HER2-I), they are expected to recognize the same epitope on HER2. In Example 6, it was shown that the HER2 binding ABD variants disclosed herein do not bind to the same epitope as the HER2 binding Z variant Z02891 . In this Example, it was investigated whether the ABD_HER2 variants bind to same epitope as the clinically approved therapeutic antibody trastuzumab. Z02891 was included again for

comparison. The result of the blocking assay is shown schematically in Table 4, and data from the flow-cytometric analysis is shown in Figure 9. The assay showed that, as expected, all four molecules are able to block the binding of HER2 to cells expressing the same molecule as used for blocking. However, binding of HER2 to Herceptin®-scFv on cells could be blocked by the

ABD_HER2 variants SEQ ID NO:43 or SEQ ID NO:90, and HER2 binding to said cell-displayed ABD_HER2 variants could be blocked by Herceptin-scFv.

Therefore, it is concluded that the HER2 binding ABD variants disclosed herein recognize the same epitope on HER2 as does trastuzumab, which is an epitope distinct from that recognized by the HER2 specific Z variant Z02891 .

Table 4. Schematic overview of the results from the blocking assay ITEMIZED LISTING OF EMBODIMENTS

1 . HER2 binding polypeptide, comprising an amino acid sequence selected from i) LAX₃AKX₆TAX₉Y HLXi₃Xi Xi5GVSDX₂₀ YKX23LIDKX28KT

VEGVX35ALYX39X40 ILX43ALP wherein, independently of each other,

X3 is selected from A, G, P, S and V;

Xe is selected from D and E;

XQ is selected from L and N;

Xl3 is selected from D and T;

Xl4 is selected from K and R;

Xl5 is selected from I, L, M, T and V;

X20 is selected from F and Y;

X23 is selected from D and R;

X28 is selected from A and V;

X35 is selected from K, M and R;

X39 is selected from A, F and L;

X40 is selected from A and E; and

X43 is selected from A, H, K, P, R, T, Q and Y and ii) an amino acid sequence which has at least 86 % identity to the

sequence defined in i). 2. HER2 binding polypeptide according to item 1 , wherein X₃ in sequence i) is selected from A, G and P.

3. HER2 binding polypeptide according to any preceding item, wherein X3 in sequence i) is selected from A and G. 4. HER2 binding polypeptide according to item 2, wherein X₃ in sequence i) is selected from A and P.

5. HER2 binding polypeptide according to any one of items 3-4, wherein X₃ in sequence i) is A.

6. HER2 binding polypeptide according to any preceding item, wherein Χβ in sequence i) is E. 7. HER2 binding polypeptide according to any preceding item, wherein

Xg in sequence i) is L.

8. HER2 binding polypeptide according to any preceding item, wherein Xi3 in sequence i) is D.

9. HER2 binding polypeptide according to any preceding item, wherein Xi₄ in sequence i) is R.

10. HER2 binding polypeptide according to any preceding item, wherein Xi₅ in sequence i) is selected from L and V.

11. HER2 binding polypeptide according to item 10, wherein Xi₅ in sequence i) is L. 12. HER2 binding polypeptide according to item 10, wherein Xi₅ in sequence i) is V.

13. HER2 binding polypeptide according to any one of items 1 -10, wherein Xi Xi₅ in sequence i) is selected from RL and RV.

14. HER2 binding polypeptide according to any preceding item, wherein X₂o in sequence i) is F.

15. HER2 binding polypeptide according to any one of items 1 -13, wherein X₂o in sequence i) is Y. 16. HER2 binding polypeptide according to any preceding item, wherein X₂₃ in sequence i) is D.

17. HER2 binding polypeptide according to any one of items 1 -15, wherein X₂₃ in sequence i) is R.

18. HER2 binding polypeptide according to any preceding item, wherein X₂₈ in sequence i) is A. 19. HER2 binding polypeptide according to any preceding item, wherein X₃₅ in sequence i) is R.

20. HER2 binding polypeptide according to any preceding item, wherein X₃₉ in sequence i) is selected from F and L.

21 . HER2 binding polypeptide according to item 20, wherein X₃₉ in sequence i) is F.

22. HER2 binding polypeptide according to item 20, wherein X₃g in sequence i) is L.

23. HER2 binding polypeptide according to any preceding item, wherein X40 in sequence i) is E. 24. HER2 binding polypeptide according to any preceding item, wherein X^ in sequence i) is selected from H, P and R.

25. HER2 binding polypeptide according to any preceding item, wherein sequence i) is selected from SEQ ID NO:1 -24 and 30-177, such as selected from SEQ ID NO:1 -24.

26. HER2 binding polypeptide according to item 25, wherein sequence i) is selected from SEQ ID NO:1 -7, SEQ ID NO:9-19 and SEQ ID NO:21 -24. 27. HER2 binding polypeptide according to item 25, wherein sequence i) is selected from SEQ ID NO: 1 -12 and 30-103, such as selected from SEQ ID NO:1 -12. 28. HER2 binding polypeptide according to item 27, wherein sequence i) is selected from SEQ ID NO:1 -7 and SEQ ID NO:9-12. 29. HER2 binding polypeptide according to item 25, wherein sequence i) is selected from the group consisting of SEQ ID NO:32, 33, 37, 38, 39, 43, 44, 87, 90, 91 , 93, 106, 107, 1 1 1 , 1 12, 1 13, 1 17, 1 18, 161 , 164, 165 and 167.

30. HER2 binding polypeptide according to item 29, wherein sequence i) is selected from the group consisting of SEQ ID NO:32, 33, 37, 38, 39, 43,

44, 87, 90, 91 and 93.

31 . HER2 binding polypeptide according to any preceding item, which is capable of binding to HER2 such that the K_D value of the interaction is at most 1 x 10^"8 M, such as at most 1 x 10^"9 M, such as at most 1 x 10^"10 M, such as at most 1 x 10^{"1 1} M.

32. HER2 binding polypeptide according to any preceding item, wherein said polypeptide is capable of binding albumin.

33. HER2 binding polypeptide according to item 32, which is capable of binding to albumin such that the K_D value of the interaction is at least 1 x 10^"8 M, such as at least 1 x 10^"7 M, such as at least 1 x 10^"6 M, such as at least

1 x 10^"5 M.

34. HER2 binding polypeptide according to any one of items 32-33, wherein said albumin is human serum albumin.

35. HER2 binding polypeptide according to any preceding item as a multimer, such as a dimer.

36. Fusion protein or conjugate comprising

- a first moiety consisting of HER2 binding polypeptide according to any preceding item; and

- a second moiety consisting of a polypeptide having a desired biological activity. 37. Fusion protein or conjugate according to item 36, wherein said desired biological activity is a therapeutic activity.

38. Fusion protein or conjugate according to item 36, wherein said desired biological activity is a binding activity.

39. Fusion protein or conjugate according to item 36, wherein said desired biological activity is an enzymatic activity. 40. Fusion protein or conjugate according to item 37, wherein the second moiety having a desired biological activity is a therapeutically active polypeptide.

41 . Fusion protein or conjugate according to any one of items 36, 37 and 40, wherein the second moiety having a desired biological activity is selected from the group consisting of human endogenous enzymes, hormones, growth factors, chemokines, cytokines and lymphokines.

42. HER2 binding polypeptide, fusion protein or conjugate according to any preceding item, further comprising a cytotoxic agent.

43. HER2 binding polypeptide, fusion protein or conjugate according to item 42, wherein said cytotoxic agent is selected from the group consisting of auristatin, anthracycline, calicheamycin, combretastatin, doxorubicin, duocarmycin, the CC-1065 anti-tumorantibiotic, ecteinsascidin, geldanamycin, maytansinoid, methotrexate, mycotoxin, ricin and its analogues, taxol and derivates thereof and combinations thereof.

44. HER2 binding polypeptide, fusion protein or conjugate according to any preceding item, further comprising a label.

45. HER2 binding polypeptide, fusion protein or conjugate according to item 44, wherein said label is selected from the group consisting of

fluorescent dyes and metals, chromophoric dyes, chemiluminescent compounds and bioluminescent proteins, enzymes, radionuclides and particles. 46. HER2 binding polypeptide, fusion protein or conjugate according to any preceding item, comprising a chelating environment provided by a polyaminopolycarboxylate chelator conjugated to the HER2 binding polypeptide via a thiol group of a cysteine residue or an amine group of a lysine residue.

47. HER2 binding polypeptide, fusion protein or conjugate according to item 46, wherein the polyaminopolycarboxylate chelator is 1 ,4,7,10- tetraazacyclododecane-1 , 4, 7,10-tetraacetic acid or a derivative thereof.

48. HER2 binding polypeptide, fusion protein or conjugate according to item 47, wherein the 1 ,4,7,10-tetraazacyclododecane-1 ,4,7,10-tetraacetic acid derivative is 1 ,4,7,10-tetraazacyclododecane-1 ,4,7-tris-acetic acid-10- maleimidoethylacetamide.

49. HER2 binding polypeptide, fusion protein or conjugate according to item 46, wherein the polyaminopolycarboxylate chelator is 1 ,4,7- triazacyclononane-1 ,4,7-triacetic acid or a derivative thereof. 50. HER2 binding polypeptide, fusion protein or conjugate according to item 46, wherein the polyaminopolycarboxylate chelator is

diethylenetriaminepentaacetic acid or derivatives thereof.

51 . Polynucleotide encoding a HER2 binding polypeptide or a fusion protein according to any one of items 1 -41 .

52. Method of producing a polypeptide or a fusion protein according to any one of items 1 -41 , comprising expressing a polynucleotide according to item 51 .

53. Expression vector comprising a polynucleotide according to item

51 .

54. Host cell comprising an expression vector according to item 53. 55. Method of producing a polypeptide according to any one of items 1 - 41 , comprising

- culturing a host cell according to item 54 under conditions permissive of expression of said polypeptide from said expression vector, and

- isolating said polypeptide.

56. Composition comprising a HER2 binding polypeptide, fusion protein or conjugate according to any one of items 1 -50 and at least one pharmaceutically acceptable excipient or carrier.

57. Composition according to item 56, further comprising at least one additional active agent.

58. HER2 binding polypeptide, fusion protein or conjugate according to any one of items 1 -50 or a composition according to any one of items 56-57 for use as a medicament.

59. HER2 binding polypeptide, fusion protein or conjugate according to any one of items 1 -50 or a composition according to any one of items 56-57 for use in diagnosis.

60. HER2 binding polypeptide, fusion protein or conjugate according to any one of items 1 -50 or a composition according to any one of items 56-57 for use in the treatment or diagnosis of a HER2 related condition.

61 . HER2 binding polypeptide, fusion protein, conjugate or composition for use according to item 60, wherein said HER2 related condition is cancer.

62. HER2 binding polypeptide, fusion protein, conjugate or composition for use according to item 61 , wherein said cancer is selected from the group consisting of breast cancer, ovarian cancer, gastric cancer, uterine cancer, testicular cancer, esophageal cancer, bladder cancer, salivary cancer and lung cancer. 63. Method of treatment of a HER2 related condition, comprising administering to a subject in need thereof an effective amount of a HER2 binding polypeptide, fusion protein or conjugate according to any one of items 1 -50 or a composition according to any one of items 56-57.

64. Method of treatment according to item 63, wherein said HER2 binding polypeptide, fusion protein, conjugate or composition inhibits HER2 mediated signaling by binding to HER2 expressed on a cell surface.

65. Method according to any one of items 63-64, wherein said HER2 related condition is cancer.

66. Method according to item 65, wherein said cancer is selected from the group consisting of breast cancer, ovarian cancer, gastric cancer, uterine cancer, testicular cancer, esophageal cancer, bladder cancer, salivary cancer and lung cancer.

Claims

1 . HER2 binding polypeptide, comprising an amino acid sequence selected from i) LAX₃AKX₆TAX₉Y HLXi₃Xi Xi5GVSDX₂₀ YKX₂₃LIDKX₂₈KT

VEGVX35ALYX39X40 ILX43ALP wherein, independently of each other,

X₃ is selected from A, G, P, S and V;

Χβ is selected from D and E;

Xg is selected from L and N;

Xi3 is selected from D and T;

Xi₄ is selected from K and R;

Xi₅ is selected from I, L, M, T and V;

X₂o is selected from F and Y;

X₂3 is selected from D and R;

X₂8 is selected from A and V;

X35 is selected from K, M and R;

X39 is selected from A, F and L;

X₄o is selected from A and E; and

X₄₃ is selected from A, H, K, P, R, T, Q and Y; and ii) an amino acid sequence which has at least 86 % identity to the

sequence defined in i).

2. HER2 binding polypeptide according to claim 1 , wherein sequence i) is selected from SEQ ID NO:1 -24.

3. HER2 binding polypeptide according to claim 2, wherein sequence i) is selected from SEQ ID NO:1 -7 and SEQ ID NO:9-12.

4. HER2 binding polypeptide according to any preceding claim, which is capable of binding to HER2 such that the K_D value of the interaction is at most 1 x 10^"8 M, such as at most 1 x 10^"9 M, such as at most 1 x 10^"10 M, such as at most 1 x 10^{"1 1} M.

5. HER2 binding polypeptide according to any preceding claim, wherein said polypeptide is capable of binding albumin.

6. HER2 binding polypeptide according to claim 5, which is capable of binding to albumin such that the K_D value of the interaction is at least 1 x 10^"8

M, such as at least 1 x 10^"7 M, such as at least 1 x 10^"6 M, such as at least 1 x 10^"5 M.

7. Fusion protein or conjugate comprising

- a first moiety consisting of HER2 binding polypeptide according to any preceding claim; and

- a second moiety consisting of a polypeptide having a desired biological activity.

8. Polynucleotide encoding a HER2 binding polypeptide or fusion protein according to any one of claims 1 -7.

9. Expression vector comprising a polynucleotide according to claim 8.

10. Host cell comprising an expression vector according to claim 10.

1 1 . Method of producing a polypeptide according to any one of claims 1 -7, comprising

- culturing a host cell according to claim 1 1 under conditions

permissive of expression of said polypeptide from said expression vector, and

- isolating said polypeptide.

12. Composition comprising a HER2 binding polypeptide, fusion protein or conjugate according to any one of claims 1 -7 and at least one pharmaceutically acceptable excipient or carrier.

13. HER2 binding polypeptide, fusion protein or conjugate according to any one of claims 1 -7 or composition according to claim 13 for use as a medicament.

14. HER2 binding polypeptide, fusion protein or conjugate according to any one of claims 1 -7 or composition according to claim 13 for use in diagnosis.

15. HER2 binding polypeptide, fusion protein or conjugate according to any one of claims 1 -7 or composition according to claim 13 for use in the treatment or diagnosis of a HER2 related condition, for example cancer, for example cancer selected from the group consisting of breast cancer, ovarian cancer, gastric cancer, uterine cancer, testicular cancer, esophageal cancer, bladder cancer, salivary cancer and lung cancer.