WO2010010051A1

WO2010010051A1 - Tnfsf single chain molecules

Info

Publication number: WO2010010051A1
Application number: PCT/EP2009/059269
Authority: WO
Inventors: Oliver Hill; Christian Gieffers; Meinolf Thiemann
Original assignee: Apogenix Gmbh
Priority date: 2008-07-21
Filing date: 2009-07-18
Publication date: 2010-01-28
Also published as: JP2014230546A; EP2604693B1; JP2014230545A; EP2604693A3; HRP20150425T1; EP2806029B1; US10519208B2; JP2017029158A; US20110162095A1; US11214605B2; EP2604693A2; EP3103875A1; US8450460B2; PL2310509T3; US8921519B2; CA3092223C; JP2011528562A; US20170081381A1; CY1116288T1; EP2310509B1

Abstract

The present invention refers to single-chain fusion proteins comprising three soluble TNF superfamily (TNFSF) cytokine domains and nucleic acid molecules encoding these fusion proteins. The fusion proteins are substantially non-aggregating and suitable for therapeutic, diagnostic and/or research applications.

Description

TNFSF Single Chain Molecules

Description

The present invention refers to single-chain fusion proteins comprising three soluble TNF superfamily (TNFSF) cytokine domains and nucleic acid molecules encoding the fusion proteins. The fusion proteins are substantially non-aggregating and suitable for therapeutic, diagnostic and/or research applications.

State of the Art

It is known that trimerisation of TNFSF cytokines, e.g., the CD95 ligand (CD95L), is required for efficient receptor binding and activation. Trimeric complexes of TNF superfamily cytokines, however, are difficult to prepare from recombinant monomeric units. WO 01/49866 and WO 02/09055 disclose recombinant fusion proteins comprising a TNF cytokine and a multimerisation component, particularly a protein from the C1q protein family or a collectin. A disadvantage of these fusion proteins is, however, that the trimerisation domain usually has a large molecular weight and/or that the trimerisation is rather inefficient.

Schneider et al. (J Exp Med 187 (1989), 1205-1213) describe that trimers of TNF cytokines are stabilised by N-terminally positioned stabilisation motifs. In CD95L, the stabilisation of the receptor binding domain trimer is presumably caused by N-terminal amino acid domains which are located near the cytoplasmic membrane.

Shiraishi et al. (Biochem Biophys Res Commun 322 (2004), 197-202) describe that the receptor binding domain of CD95L may be stabilised by N-terminally positioned artificial α-helical coiled-coil (leucine zipper) motifs. It was found, however, that the orientation of the polypeptide chains to each other, e.g. parallel or antiparallel orientation, can hardly be predicted. Further, the optimal number of heptad- repeats in the coiled-coil zipper motif are difficult to determine. In addition, coiled-coil structures have the tendency to form macromolecular aggregates after alteration of pH and/or ionic strength. WO 01/25277 relates to single-chain oligomeric polypeptides which bind to an extracellular ligand binding domain of a cellular receptor, wherein the polypeptide comprises at least three receptor binding sites of which at least one is capable of binding to a ligand binding domain of the cellular receptor and at least one is incapable of effectively binding to a ligand binding domain of the cellular receptor, whereby the single-chain oligomeric polypeptides are capable of binding to the receptor, but incapable of activating the receptor. For example, the monomers are derived from cytokine ligands of the TNF family, particularly from TNF-α. WO 2005/103077 discloses single-chain fusion polypeptides comprising at least three monomers of a

TNF family ligand member and at least two peptide linkers that link the monomers of the TNF ligand family members to one another. Recent experiments, however, have shown that these single-chain fusion polypeptides show undesired aggregation. It was an object of the present invention to provide single-chain fusion proteins comprising at least three TNF cytokine domains which allow efficient recombinant manufacturing combined with good stability concerning aggregation.

Summary of the Invention The present invention relates to a single-chain fusion polypeptide comprising:

(i) a first soluble TNF-family cytokine domain, (ii) a first peptide linker,

(iii) a second soluble TNF-family cytokine domain, (iv) a second peptide linker, and (v) a third soluble TNF-family cytokine domain, which is substantially non-aggregating.

The invention further relates to a nucleic acid molecule encoding a fusion protein as described herein and to a cell or a non-human organism transformed or transfected with a nucleic acid molecule as described herein. The invention also relates to a pharmaceutical or diagnostic composition comprising as an active agent a fusion protein, a nucleic acid molecule, or a cell as described herein.

The invention also relates to a fusion protein, a nucleic acid molecule, or a cell as described herein for use in therapy, e.g., the use of a fusion protein, a nucleic acid molecule, or a cell as described herein for the preparation of a pharmaceutical composition in the prophylaxis and/or treatment of disorders caused by, associated with and/or accompanied by dysfunction of TNFSF cytokines, particularly proliferative disorders, such as tumours, e.g. solid or lymphatic tumours; infectious diseases; inflammatory diseases; metabolic diseases; autoimmune disorders, e.g. rheumatoid and/or arthritic diseases; degenerative diseases, e.g. neurodegenerative diseases such as multiple sclerosis; apoptosis-associated diseases or transplant rejections.

Description of the Figures

Figure 1 Domain structure of the inventive single-chain fusion polypeptide. I., II., III. soluble TNF- family cytokine domains.

Figure 2 Schematic picture representing the general structure of TNF-SF proteins. ■ ■ ■ cell membrane, N-terminus located within the cell, 1. anti-parallel β-fold of receptor-binding domain (RBD), 2. interface of RBD and cell membrane, 3. protease cleavage site.

Figure 3 Schematic picture representing the structure of the native TNF-SF trimer. Cylindric structures represent RBDs, N-termini connect RBD with the cell membrane.

Figure 4 Schematic picture representing the structure of three soluble domains comprising the receptor-binding domain of a TNF cytokine. I., II., III. soluble TNF-family cytokine domains.

Figure 5 Trimerisation of the soluble domains comprising the RBD of a TNF cytokine, characterised in that the N- and C-termini of the three soluble domains form a surface.

Figure 6 Schematic picture representing the structure of the single-chain TNF-SF comprising all or a part of the stalk-region illustrating the requirement of longer linkers to compensate for the distance to the N-terminus of the next soluble domain.

Figure 7 scFv-TNF-SF fusion protein known from the art.

Figure 8 Fc-TNF-SF fusion protein known from the art.

Figure 9 9A Single-chain fusion polypeptide comprising an additional Fab antibody fragment. 9B Single-chain fusion polypeptide comprisjng an additional scFv antibody fragment.

Figure 10 Dimerisation of two N-terminally fused scFc fusion polypeptides via disulfide bridges.

Figure 11 Dimerisation of two C-terminally fused scFc fusion polypeptides via disulfide bridges.

Figure 12 Dimerisation of single-chain fusion polypeptides via a linker. Figure 13 Single-chain fusion polypeptide comprising an additional Fab antibody fragment further fused to a second fusion polypeptide or to a scFv fusion polypeptide.

Figure 14 Dimerisation of two scFab fusion polypeptides via disulfide bridges.

Figure 15 N-terminally fused scFc fusion polypeptides further comprising a Fv and/or Fab antibody fragment.

Figure 16 C-terminally fused scFc fusion polypeptides further comprising a Fv and/or Fab antibody fragment.

Figure 17 SEC analysis of recombinantly expressed, purified TNF-SF members under native conditions. Exemplarily shown are two SEC analyses of purified TNF-SF members on a Superdex200 column under native condition (e.g.: PBS, pH 7.4). The diagrams show the absorption at 280nm (imAU) plotted against the elution volume (ml). The filled arrow indicates the elution peak for the fraction containing defined, soluble trimeric TNF-SF protein. The triangle indicates the elution peak for the oligomerised TNF-SF . The open arrow indicates the void volume of the SEC-column that contains protein-aggregates, which are too big to be separated (>800kDa).

Figure 17A: TNF-SF protein Aggregation Diagram A exemplarily shows an analysis of a TNF-SF protein preparation that contains a high amount of oligomerised/aggregated protein( indicated by the high amount of protein eluting in the void volume and the high amount of oligomeric protein). Figure 17 B: TNF-SF protein defined soluble protein Diagram B exemplarily shows an analysis for a TNF-SF protein preparation that contains almost exclusively defined soluble protein (indicated by the absence of protein eluting in the void volume and by the very limited amount of protein eluting as oligomer).

Figure 18 SEC analysis of recombinantly expressed, affinity purified Fab-scTRAII_R2-SSSS.

SEC analysis of Fab-scTRAILR2-SSSS on a Superdex200 column using PBS, pH 7.4. The diagram shows the absorption at 280nm (imAU) plotted against the elution volume (ml). The protein elutes as a distinct peak with an elution volume of 14.56ml, corresponding to an apparent MW of 68 kDa. No additional protein peaks with lower retention volume, indicating oligomerised/aggregated protein, could be observed.

Figure 19 SEC analysis of recombinants expressed, affinity purified Fab-scTRAII_R2-SNSN. SEC analysis of Fab-scTRAILR2-SNSN on a Superdex200 column using PBS, pH 7.4.

The diagram shows the absorption at 280nm (imAU) plotted against the elution volume (ml). The protein elutes as a distinct peak with an elution volume of 14.12ml, corresponding to an apparent MW of 87 kDa. No additional protein peaks with lower retention volume, indicating oligomerised/aggregated protein, could be observed.

Figure 20 SEC analysis of recombinantly expressed, affinity purified Fab-scTRAILwt-SNSN.

SEC analysis of Fab-scTRAILwt-SNSN on a Superdex200 column using PBS, pH 7.4. The diagram shows the absorption at 280nm (imAU) plotted against the elution volume (ml). The protein elutes as a distinct peak with an elution volume of 13.99ml, corresponding to an apparent MW of 94 kDa. A small additional protein peak at 12.00ml could be observed. The apparent Mw of this peak corresponds to about 270 kDa, indicating a defined trimerisation of Fab-scTRAILwt-SNSN. The total protein amount of the peak at 12.00ml accounts for <3% of the total protein. More than 97% of the analysed Fab-scTRAILwt-SNSN has a defined soluble state (correct assembly of the three receptor binding modules). The peak at 16.12ml corresponding to a MW of 28 kDa contains Fab-light-chain polypeptide and was not included for the analysis of peak areas.

Figure 21 Human scTRAIL Linker glycosylate

Figure 21 A Amino acid sequence of the linker(s) used to combine the receptor binding modules of single chain TRAIL constructs. Gly281 encodes the last amino acid of a respective receptor binding module, the sequence GSGN/SGN/SGS encodes the linker sequence, Arg121 encodes the first amino acid of the following TRAIL receptor binding domain. The designed linker sequences contains two putative N-linked glycosylate sites at position 1 or 2 as indicated. These positions were permutated as indicated (version I, II, III). Figure 21 B Combination of linker positions: The scTRAIL molecules contain three homologue modules (grey barrels) that are connected with linker 1 and linker 2 as indicated. Each of the two linkers, can be designed for N-linked glycosylate as described in "A". A complete set of 9 different proteins containing all possible combinations of linkers can be designed based on the sequences shown in B for linker 1 and 2. (Six of these proteins were expressed - see "C"). Figure 21 C Nomenclature of scTRAIL constructs expressed to test the influence of different linker sequences on glycosylation

Figure 22 Western Blot analysis of recombinant scTRAIL constructs

Single chain TRAIL proteins with different linker sequences were recombinantly expressed, separated by SDS-PAGE and transferred to a PVDF-membrane. Bound proteins were detected with a mouse monoclonal antibody recognising the Strep -Tag followed by a Peroxidase-conjugated secondary anti-mouse antibody. Different TRAIL variants were loaded as indicated. Note the MW-shift indicating differential glycosylation of scTRAIL-linker variants.

Figure 23 Cell culture supernatant of HEK293 cells, transiently expressing scCD95L (SEQ-ID NO:27) was collected and used to stimulate Jurkat cells at varying concentrations. The supernatant was used either directly without further modifications or an anti-Streptag antibody (2 microgram/ml) was added to cross-link the scCD95L protein. Jurkat cells were incubated with HEK293 cell culture supernatant for three hours at 37°, lysed and analysed for caspase activity. Only cell supernatant that contained cross-linked scCD95L-St increased caspase activity in Jurkat cells, indicating that scCD95L alone does not form higher order aggregates able to be pro-apoptotic.

Figure 24 The protein scCD95L (SEQ ID NO:27) can be produced by transient transfection of HEK293 cells, stable transfection of other eukaryotic cells or by expression using prokaryotic cells. The recombinant protein can be affinity purified by using StrepTactin

Sepharose matrix. Bound protein can be eluted with a buffer containing desthio-biotin. Figure 2 shows a silver stained SDS-PAGE of the elution fractions (lanes 1 to 5; fraction 2 is positive) of the affinity purification. The elution fraction containing scCD95L could be applied to size exclusion chromatography (SEC). It is expected, that the protein shows only a low aggregate content.

Figure 25 Cell culture supematants of HEK293 cells, transiently expressing single chain TRAIL proteins with different linkers (derived from SEQ ID 28) were collected and used to stimulate Jurkat cells at varying dilutions (exemplarily, a dilution of 1:8 is shown in this figure). The supematants were used either directly without further modifications or an anti-Streptag antibody (2 microgram/ml Strep MAB Immo) was added to cross-link the scTRAIL proteins. Jurkat cells were incubated with HEK293 cell culture supernatant for three hours at 37°, lysed and analysed for caspase activity. Cell culture supernatant that contained cross-linked scTRAILwt proteins induced an increased caspase activity in Jurkat cells, indicating that scTRAILwt proteins alone do form only a low amount of higher order aggregates able to be pro-apoptotic.

Figure 26 Influence of the module succession of scTRAIL-construct components on their expression rate of Fab-scTRAIL fusion proteins. Western blot of HEK293T cell culture supematants from transient expression experiments. The polypeptide chains necessary for the formation of the Fab-scTRAIL proteins were either expressed separately (lanes 1 to 10) or alternatively co-expression experiments were performed (lanes 11 -13). After reducing SDS-PAGE, proteins were transferred to a nitrocellulose membrane and proteins containing a Streptag werde detected, using an anti-Streptag specific imAB as primary AB . The light-chain-scTRAIL(R2-specific) proteins were secreted even in the absence of the accessory heavy chain (lanes 1-4). In contrast, the heavy-chain- scTRAIL(R2-specific) fusion proteins were not secreted in the absence of the acessory light chain (lanes 5-8). As exemplified in lane 13. , the heavy -chain-scTRAIL(R2-specific) fusion proteins were only secreted in the presence of the light chain.

Figure 27 Cell culture supematants of HEK293T cells, transiently expressing scTRAILwt-Fc fusion proteins with different linkers were collected and used to stimulate Jurkat cells at varying dilutions. The supematants were used directly without further modifications (Figure XX-

A). Jurkat cells were incubated with HEK293T cell culture supernatant for three hours at 37°, lysed and analysed for caspase activity. There was already a pronounced proapoptotic capacity present in the scTRAILwt-Fc containing supematants, indicating that scTRAILwt-Fc fusion proteins alone do form dimeric assemblies able to be pro- apoptotic.

Figure 28 It is well known that the use of artificially cross-linked or a membrane-bound ligand of the TNF superfamily has superior bioactivity as compared to soluble, homotrimeric ligand. Thus the local enrichment of single chain TRAIL (scTRAIL) constructs on cells that express the antigen Her2 via the Her2-selective Fab-fragment ("Pertuzumab") fused to these scTRAIL proteins should increase their cytotoxic bioactivity. Likewise, the blocking of the Her2 binding sites on cells by pre-incubation with the Her2-specific Fab-fragment (Pertuzumab-Fab) only should decrease the cytotoxic bioactivity of Fab-scTRAIL fusion proteins. As shown in Figure 28A, scTRAIL constructs induce the death of HT1080 cells, as the viability decreases with incresing protein concentration. In accordance, the preincubation of HT1080 cells with the Fab-fragment (Pertuzumab-Fab), followed by co- incubation with the Fab-scTRAIL constructs (Fab-scTRAILR2-SNSN or Fab-scTRAILwt- SNSN) over night, reduced the cytotoxic activity of the Fab-scTRAIL constructs (figure

28B), whereas the Fab only induced no cell death (Pertuzumab-Fab). This means that the Fab-scTRAIL constructs bind to HT1080 cells via the Fab fragment thus increasing the cytotoxic bioactivity of scTRAIL.

Detailed Description of the Invention

According to the present invention a substantially non-aggregating fusion polypeptide comprising at least three soluble TNF family ligand domains connected by two peptide linkers is provided.

The term "non-aggregating" refers to a monomer content of the preparation of > 50%, preferably > 70% and more preferably > 90%. The ratio of monomer content to aggregate content may be determined by examining the amount of aggregate formation using size-exclusion chromatography (SEC). The stability concerning aggregation may be determined by SEC after defined time periods, e.g. from a few to several days, to weeks and months under different storage conditions, e.g. at 4°C or 25°C. For the fusion protein, in order to be classified as substantially non -aggregating, it is preferred that the monomer content is as defined above after a time period of several days, e.g. 10 days, more preferably after several weeks, e.g. 2, 3 or 4 weeks, and most preferably after several months, e.g. 2 or 3 months of storage at4⁰C, or25°C.

As an increase of e.g. the apoptosis inducing potential in the case of scCD95L on human Jurkat cells correlates with its aggregation state, the stability of the fusion polypeptide concerning aggregation may also be determined by examining the biological activity of the fusion polypeptide.

The single-chain fusion polypeptide may comprise additional domains which may be located at the N- and/or C-termini thereof. Examples for additional fusion domains are e.g. single-chain antibodies or antibody fragments or other targeting molecules or a further cytokine domain, e.g. an interleukin. The single-chain fusion protein comprises three soluble domains derived from a cytokine of the TNF superfamily. Preferably, those soluble domains are derived from a mammalian, particularly human cytokine including allelic variants and/or derivatives thereof. The soluble domains comprise the extracellular portion of a TNFSF cytokine including the receptor binding domain without membrane located domains. Proteins of the TNF superfamiliy are anchored to the membrane via an N -terminal portion of 15-30 amino acids, the so-called stalk-region. The stalk region contributes to trimerisation and provides a certain distance to the cell membrane. However, the stalk region is not part of the receptor binding domain (RBD). Importantly, the RBD is characterised by a particular localisation of its N- and C-terminal amino acids. Said amino acids are immediately adjacent and are located centrally to the axis of the trimer. The first N-terminal amino acids of the RBD form an anti-parallel beta-strand with the C-terminal amino acids of the RBD (Fig. 2 and 3).

Thus, the anti-parallel beta-strand of the RBD forms an interface with the cell membrane, which is connected to and anchored within the cell membrane via the amino acids of the stalk region. It is highly preferred that the soluble domains of the single-chain fusion protein comprises a receptor binding domain of the TNF-SF cytokine lacking any amino acids from the stalk region (Figs. 4 and 5). Otherwise, a long linker connecting the C-terminus of one of the soluble domains with the N -terminus of the next soluble domain would be required to compensate for the N-terminal stalk-region of the next soluble domain (Figure 6), which might result in instability and/or formation of aggregates. A further advantage of such soluble domains is that the N- and C-terminal amino acids of the RBD are not accessible for any anti-drug antibodies.

Preferably, the single-chain fusion polypeptide is capable of forming an ordered trimeric structure comprising at least one functional binding site for the respective cytokine receptor. The fusion polypeptide may comprise one, two or three functional cytokine receptor binding sites, i.e. amino acid sequences capable of forming a complex with a cytokine receptor. Thus, at least one of the soluble domains is capable of binding to the corresponding cytokine receptor. In one embodiment, at least one of the soluble domains is capable of receptor activation, whereby apoptotic and/or proliferative activity may be effected. In a further embodiment, one or more of the soluble domains are selected as not being capable of receptor activation. The soluble domain may be derived from TNF superfamily members, e.g. human TNFSF-1 to -18 and EDA-A1 to -A2 as indicated in Table 1 , preferably from LTA (SEQ ID N0:1), TNFα (SEQ ID N0:2), LTB (SEQ ID N0:3), OX40L (SEQ ID N0:4), CD40L (SEQ ID N0:5), CD95L (SEQ ID N0:6), CD27L (SEQ ID N0:7), CD30L (SEQ ID N0:8), CD137L (SEQ ID N0:9), TRAIL (SEQ ID NO:10), RANKL (SEQ ID N0:11), TWEAK (SEQ ID N0:12), APRIL 1 (SEQ ID N0:13), APRIL 2 (SEQ ID N0:14), BAFF (SEQ ID

N0:15), LIGHT (SEQ ID N0:16), TL1A (SEQ ID N0:17), GITRL (SEQ ID N0:18), EDA-A1 (SEQ ID N0:19) and EDA-A2 (SEQ ID NO:20). Preferred soluble domains of the respective proteins are indicated in Table 1 (NH₂-aa to COOH-aa) and, e.g., comprise amino acids 59-205, 60-205 or 64-205 of LTA (SEQ ID NO:1), 86-233 of TNFα (SEQ ID NO:2), 82-244 or 86-244 of LTB (SEQ ID NO:3), 52-183 or 55-183 of OX40L (SEQ ID NO:4), 112-261, 117-261 or 121-261 of CD40L (SEQ ID NO:5), 51-193 or 56-193 of CD27L (SEQ ID NO:7), 97-234, 98-234 or 102-234 of CD30L (SEQ ID NO:8), 86-254 of CD137L (SEQ ID NO:9), 161 -317 of RANKL (SEQ ID NO:11), 103-249, 104-249, 105-249 or 106-249 of TWEAK (SEQ ID NO:12), 112-247 of APRIL 1 (SEQ ID NO:13), 112-250 of APRIL 2 (SEQ ID NO:14), 140-285 of BAFF (SEQ ID NO:15), 91-251, 93-251 or 97-251 of TL1A (SEQ ID NO:17), 52-177 of GITRL (SEQ ID NO:18), 245-391 of EDA-A1 (SEQ ID NO:19), 245-389 of EDA-A2 (SEQ ID NO:20). More preferably, the soluble domains are derived from CD95L, TRAIL or LIGHT. In an especially preferred embodiment, the soluble domains are selected from human CD95L, particularly starting from amino acids 144, 145 or 146 and comprise particularly amino acids 144-281 or 145-281 or 146-281 of SEQ ID NO:6 or human TRAIL, particularly starting from amino acids 120-122 and comprise particularly amino acids 120-281, 121-281 or 122-281 of SEQ ID NO:10. Optionally, amino acid Lys145 of SEQ ID NO:6 may be replaced by a non-charged amino acid, e.g. Ser or GIy. Optionally, amino acid Arg121 of SEQ ID NO:10 may be replaced by a non-charged amino acid, e.g. Ser or GIy. In a further preferred embodiment, the soluble domains are selected from human LIGHT, particularly starting from amino acids 93, 94 or 95 of SEQ ID NO:16 and particularly comprise amino acids 93-240, 94-240 or 95-240 of SEQ ID NO:16.

As indicated above, the soluble domains may comprise the wild-type sequences as indicated in SEQ ID NO: 1-20. It should be noted, however, that it is possible to introduce mutations in one or more of these soluble domains, e.g. mutations which alter (e.g. increase or decrease) the binding properties of the soluble domains. In one embodiment, soluble domains may be selected which cannnot bind to the corresponding cytokine receptor. An example of such a mutation is a replacement of amino acid Y218 in human CD95L (SEQ ID NO:6) by another amino acid, e.g. R, K, S or D. Further, a mutation may be introduced which alters the binding to other cellular and/or extracellular components, e.g. the extracellular matrix. An example of such a mutation is a replacement of amino acid K177 in CD95L (SEQ ID NO: 6) by another amino acid, e.g. E, D or S.

In a further preferred embodiment of the invention, the soluble cytokine domain (i) comprises a mutant of the cytokine of the TNF superfamily or a receptor binding domain thereof which binds and/or activates TRAIL-receptor 1 (TRAILR1) and/or TRAIL-receptor 2 (TRAILR2). The binding and/or activity of the mutant may be, e.g., determined by the assays as described in van der Sloot et al. (PNAS, 2006,

103:8634-8639), Kelley et al. (J. Biol. Chem., 2005, 280:2205-2215), or MacFarlane et al. (Cancer Res., 2005, 65: 11265-11270).

The mutant may be generated by any technique and is known by the skilled person, e.g., the techniques described in van der Sloot et al. (PNAS, 2006, 103:8634-8639), Kelley et al. (J. Biol. Chem., 2005, 280:2205-2215), or MacFarlane et al. (Cancer Res., 2005, 65: 11265-11270) and may comprise any type of structural mutations, e.g., substitution, deletion, duplication and/or insertion of an amino acid. A preferred embodiment is the generation of substitutions. The substitution may affect at least one amino acid of the cytokine of the TNF superfamily or a receptor binding domain thereof as described herein. In a preferred embodiment, the substitution may affect at least one of the amino acids of TRAIL, e.g., human TRAIL (e.g., SEQ ID NO: 10). Preferred substitutions in this regard affect at least one of the following amino acids of human TRAIL of SEQ ID NO:10: R130, G160, Y189, R191 , Q193, E195, N199, K201 , Y213, T214, S215, H264, 1266, D267, D269. Preferred amino acid substitutions of human TRAIL of SEQ ID NO:10 are at least one of the following substitutions: R130E, G160M, Y189A, Y189Q, R191K, Q193S, Q193R, E195R, N199V, N199R, K201R, Y213W, T214R, S215D, H264R, I266L, D267Q, D269H, D269R, or D269K.

The amino acid substitution(s) may affect the binding and/or activity of TRAIL, e.g., human TRAIL, to or on either the TRAILR1 or the TRAILR2. Alternatively, the amino acid substitution(s) may affect the binding and/or activity of TRAIL, e.g., human TRAIL, to or on both, the TRAILR1 and the TRAILR2. The binding and/or activity of the TRAILR1 and/or TRAILR2 may be affected positively, i.e., stronger, more selective or more specific binding and/or more activation of the receptor. Alternatively, the binding and/or activity of the TRAILR1 and/or TRAILR2 may be affected negatively, i.e., weaker, less selective or less specific binding and/or less or no activation of the receptor. Examples of mutants of TRAIL with amino acid substitution(s) of the invention that affect binding and/or activation of both TRAILR1 and TRAILR2 may be found, e.g., in Table 1 of MacFarlane et al. (cf. above) and may comprise a human TRAIL mutant with the following two amino acid substitutions of SEQ ID NO: 10 Y213W and S215D or with the following single amino acid substitution : Y189A. Examples of mutants of TRAIL with amino acid substitution(s) of the invention that affect binding and/or activation of TRAILR1 may be found, e.g., in Table 1 of MacFarlane et al. (cf. above) and may comprise a human TRAIL mutant with the following four amino acid substitutions of SEQ ID NO: 10 N 199V, K201R, Y213W and S215D or with the following five amino acid substitutions: Q193S, N199V, K201R, Y213W and S215D, or may be found in Table 2 of Kelley et al. (cf. above) and may comprise a human TRAIL mutant with the following six amino acid substitutions: Y213W, S215D, Y189A, Q193S, N199V, and K201R, or with Y213W, S215D, Y189A, Q193S, N199R, and K201R.

Examples of mutants of TRAIL with amino acid substitution(s) of the invention that affect binding and/or activation of TRAILR2 may be found, e.g., in Table 1 of MacFarlane et al. (cf. above) or in Table 2 of

Kelley et al. (cf. above) and may comprise a human TRAIL mutant with the following six amino acid substitutions of SEQ ID NO: 10: Y189Q, R191 K, Q193R, H264R, I266L, and D267Q , or may be found in

Table 2 of van der Sloot et al. (cf. above) and may comprise a human TRAIL mutant with the following single amino acid substitution: D269H, or with the following two amino acid substitutions: D269H and

E195R or D269H and T214R.

Thus one preferred embodiment is a fusion protein as described herein wherein at least one of the soluble domains comprises a mutant of TRAIL or of a receptor binding domain thereof which binds and/or activates TRAILR1 and/or TRAILR2.

Further examples of mutants of TRAIL, which show reduced TRAIL induced receptor aggregation are

H168 (S, T, Q), R170 (E, S, T, Q) and H177 (S, T).

One preferred embodiment of a fusion protein comprising a mutant of TRAIL or of a receptor binding domain as described herein is a fusion protein wherein component (i) comprises at least one amino acid substitution, particularly as indicated below.

Such an amino acid substitution affects at least one of the following amino acid positions of human

TRAIL (SEQ ID NO: 10): R130, G160, H168, R170, H177, Y189, R191 , Q193, E195, N199, K201 , Y213,

T214, S215, H264, 1266, D267, D269. Such an amino acid substitution is at least one of the following: R13OE, G16OM, H168 (S, T, Q), R170

(E, S, T, Q), H177 (SJ)₁ Y189A, Y189Q, R191K, Q193S, Q193R, E195R, N199V, N199R, K201R,

Y213W, T214R, S215D, H264R, I266L, D267Q, D269H, D269R, or D269K.

A preferred TRAIL-R2 selective domain comprises amino acid substitutions Y189Q, R191K, Q193R,

H264R, I266L and D267Q. A preferred TRAIL-R1 selective domain comprises amino acid substitutions Y189A, Q193S, N199V,

K201R, Y213W and S215D.

The single-chain fusion molecule of the present invention comprises additionally three soluble cytokine domains, namely components (i), (iii) and (v). According to the present invention, it was surprisingly found that the stability of a single-chain TNF family cytokine fusion polypeptide against aggregation is enhanced, if the second and/or third soluble TNF family cytokine domain is an N-terminally shortened domain which optionally comprises amino acid sequence mutations. Thus, preferably, both the second and the third soluble TNF family cytokine domain are N-terminally shortened domains which optionally comprise amino acid sequence mutations in the N-terminal regions, preferably within the first five amino acids of the N-terminus of the soluble cytokine domain. These mutations may comprise replacement of charged, e.g. acidic or basic amino acids, by neutral amino acids, particularly serine or glycine. In contrast thereto, the selection of the first soluble TNF family cytokine domain is not as critical. Here, a soluble domain having a full-length N-terminal sequence may be used. It should be noted, however, that also the first soluble cytokine domain may have an N -terminally shortened and optionally mutated sequence.

In a preferred embodiment of the present invention, the soluble TNF family cytokine domains (i), (iii) and (v) are soluble CD95L domains, particularly soluble human CD95L domains. The first soluble CD95L domain (i) may be selected from native, shortened and/or mutated sequences. The N-terminal sequence of the first domain (i) may e.g. start between amino acid Glu142 and Val146 of human CD95L, wherein Arg144 and/or Lys145 may be replaced by a neutral amino acid, e.g. by Ser or GIy. The second and third soluble CD95L domains (iii) and (v), however, are selected from shortened and/or mutated sequences. Preferably, at least one of the soluble CD95L domains, (iii) and (v), has an N- terminal sequence which starts between amino acid Arg144 and Val146 of human CD95L, and wherein Arg144 and/or Lys145 may be replaced by a neutral amino acid, e.g. by Ser and/or GIy. In an especially preferred embodiment, the second and third soluble CD95L domain start with an N-terminal sequence selected from:

(a) Arg144 - (Gly/Ser) 145 - VaI (146)

(b) (Gly/Ser) 144 - Lys145 - VaI (146) and (C) (Gly/Ser) 144 - (Gly/Ser) 145 -VaI (146).

Further, it is preferred that the CD95L domain ends with amino acid Leu 281 of human CD95L.

The soluble CD95L domain may comprise a mammalian, e.g. a human wild-type sequence. In certain embodiments, however, the CD95L sequence may comprise a mutation which results in a reduction or complete inhibition of the binding to the extracellular matrix, e.g. a mutation at position Lys177, e.g. Lys177→ GIu, Asp or Ser and/or a mutation which reduces and/or inhibits binding to the CD95L receptor, e.g. a mutation at position Tyr218, e.g. Tyr218 → Arg, Lys, Ser, Asp. In certain embodiments of the present invention, one of the three soluble CD95L modules is a sequence variant with a reduced receptor binding. In other embodiments, two of the modules contain mutations resulting in reduced receptor binding.

In a further preferred embodiment of the present invention, the soluble TNF family cytokine domains (i), (iii) and (v) are soluble TRAIL domains, particularly soluble human TRAIL domains. The first soluble TRAIL domain (i) may be selected from native, shortened and/or mutated sequences. Thus, the first soluble TRAIL domain (i) has an N-terminal sequence which may start between amino acid GIu116 and

VaM 22 of human TRAIL, and wherein Arg121 may be replaced by a neutral amino acid, e.g. by Ser or GIy. The second and third soluble TRAIL domains (iii) and (v) have a shortened N-terminal sequence which preferably starts between amino acid Gly120 and VaM 22 of human TRAIL and wherein Arg121 may be replaced by another amino acid, e.g . Ser or GIy.

Preferably, the N-terminal sequence of the soluble TRAIL domains (iii) and (v) is selected from:

(a) Arg121 - Val122 - Ala123 and

(b) (Gly/Ser)121.

The soluble TRAIL domain preferably ends with amino acid Gly281 of human TRAIL. In certain embodiments, the TRAIL domain may comprise internal mutations as described above.

In a further preferred embodiment of the present invention, the soluble TNF family cytokine domains (i), (iii) and (v) are soluble LIGHT domains, particularly soluble human LIGHT domains. The first soluble LIGHT domain (i) may be selected from native, shortened and/or mutated sequences. Thus, the first soluble LIGHT domain (i) has an N-terminal sequence which may start between amino acid Glu91 and Ala95 of human LIGHT. The second and third soluble LIGHT domains (iii) and (v) have a shortened N- terminal sequence which preferably starts between amino acid Pro94 and Ala95 of human LIGHT. The soluble LIGHT domain preferably ends with amino acid Val240.

Components (ii) and (iv) of the single-chain fusion polypeptide are peptide linker elements located between components (i) and (iii) or (iii) and (v), respectively. The flexible linker elements have a length of 3-8 amino acids, particularly a length of 3, 4, 5, 6, 7, or 8 amino acids. The linker elements are preferably glycine/serine linkers, i.e. peptide linkers substantially consisting of the amino acids glycine and serine. In cases in which the soluble cytokine domain terminates with S or G (C-terminus), e.g. human TRAIL, the linker starts after S or G. In cases in which the soluble cytokine domain starts with S or G (N-terminus), the linker ends before this S or G. It should be noted that linker (ii) and linker (iv) do not need to be of the same length. In order to decrease potential immunogenicity, it may be preferred to use shorter linkers. In addition it turned out that shorter linkers lead to single chain molecules with reduced tendency to form aggregates. Whereas linkers that are substantially longer than the ones disclosed here may exhibit unfavourable aggregations properties. If desired, the linker may comprise an asparagine residue which may form a glycosylate site Asn-Xaa- Ser. In certain embodiments, one of the linkers, e.g. linker (ii) or linker (iv) comprises a glycosylate on site. In other embodiments, both linkers (iv) comprise glycosylate sites. In order to increase the solubility of the scTNF-SF proteins and/or in order to reduce the potential immunogenicity, it may be preferred that linker (ii) or linker (iv) or both comprise a glycosylate site.

Preferred linker sequences are selected from GSGSGSGS (SEQ ID NO:52), GSGSGNGS (SEQ ID NO:53), GGSGSGSG (SEQ ID N0:21), GGSGSG (SEQ ID NO:22), GGSG (SEQ ID NO:23), GGSGNGSG (SEQ ID NO:24), GGNGSGSG (SEQ ID NO:25) and GGNGSG (SEQ ID NO:26) The fusion protein may additionally comprise an N-terminal signal peptide domain, which allows processing, e.g. extracellular secretion, in a suitable host cell. Preferably, the N-terminal signal peptide domain comprises a protease cleavage site, e.g. a signal peptidase cleavage site and thus may be removed after or during expression to obtain the mature protein. Further, the fusion protein may additionally comprise a C-terminal element, having a length of e.g. 1-50, preferably 10-30 amino acids which may include or connect to a recognition/purification domain, e.g. a FLAG domain, a Strep-tag or Strep-tag Il domain and/or a poly-His domain.

Further, the fusion polypeptide may additionally comprise N -terminally and/or C-terminally a further domain, e.g. a targeting domain such as a single-chain antibody or an antibody fragment domain. Specific examples of suitable antibodies are anti-tumour antibodies, such as antibodies against EGFR- familiy members. Suitable examples of other targeting molecules are cytokines, such as interleukins. Examples of specific fusion proteins of the invention are SEQ ID NOs: 27, 28, 29, 43, 45, 47, 49 and 51. A further aspect of the present invention relates to a nucleic acid molecule encoding a fusion protein as described herein. The nucleic acid molecule may be a DNA molecule, e.g. a double-stranded or single- stranded DNA molecule, or an RNA molecule. The nucleic acid molecule may encode the fusion protein or a precursor thereof, e.g. a pro- or pre-proform of the fusion protein which may comprise a signal sequence or other heterologous amino acid portions for secretion or purification which are preferably located at the N- and/or C-terminus of the fusion protein. The heterologous amino acid portions may be linked to the first and/or second domain via a protease cleavage site, e.g. a Factor X₃, thrombin or IgA protease cleavage site. Examples of specific nucleic acid sequences of the invention are SEQ ID NOs: 30, 31 32, 44, 46, 48 and 50.

The nucleic acid molecule may be operatively linked to an expression control sequence, e.g. an expression control sequence which allows expression of the nucleic acid molecule in a desired host cell. The nucleic acid molecule may be located on a vector, e.g. a plasmid, a bacteriophage, a viral vector, a chromosal integration vector, etc. Examples of suitable expression control sequences and vectors are described for example by Sambrook et al. (1989) Molecular Cloning, A Laboratory Manual, Cold Spring Harbor Press, and Ausubel et al. (1989), Current Protocols in Molecular Biology, John Wiley & Sons or more recent editions thereof. Various expression vector/host cell systems may be used to express the nucleic acid sequences encoding the fusion proteins of the present invention. Suitable host cells include, but are not limited to, prokaryotic cells such as bacteria, e.g. E.coli, eukaryotic host cells such as yeast cells, insect cells, plant cells or animal cells, preferably mammalian cells and, more preferably, human cells. Further, the invention relates to a non-human organism transformed or transfected with a nucleic acid molecule as described above. Such transgenic organisms may be generated by known methods of genetic transfer including homologous recombination.

A further aspect of the present invention relates to a pharmaceutical or diagnostic composition comprising as the active agent at least one fusion protein, a respective nucleic acid encoding therefore, or a transformed or transfected cell, all as described herein.

At least one fusion protein, respective nucleic acid encoding therefore, or transformed or transfected cell, all as described herein may be used in therapy, e.g., in the prophylaxis and/or treatment of disorders caused by, associated with and/or accompanied by dysfunction of TNF-SF cytokines, particularly proliferative disorders, such as tumours, e.g. solid or lymphatic tumours; infectious diseases; inflammatory diseases; metabolic diseases; autoimmune disorders, e.g. rheumatoid and/or arthritic diseases; degenerative diseases, e.g. neurodegenerative diseases such as multiple sclerosis; apoptosis-associated diseases or transplant rejections.

The term "dysfunction of TNF-SF cytokines" as used herein is to be understood as any function or expression of a TNF-SF cytokine that deviates from the normal function or expression of a TNF-SF cytokine, e.g., overexpression of the TNF-SF gene or protein, reduced or abolished expression of the TNF-SF cytokine gene or protein compared to the normal physiological expression level of said TNF-SF cytokine, increased activity of the TNF-SF cytokine, reduced or abolished activity of the TNF-SF cytokine, increased binding of the TNF-SF cytokine to any binding partners, e.g., to a receptor, particularly a CD95 or TRAIL receptor or another cytokine molecule, reduced or abolished binding to any binding partner, e.g. to a receptor, particularly a CD95 or TRAIL receptor or another cytokine molecule, compared to the normal physiological activity or binding of said TNF-SF cytokine. The composition may be administered as monotherapy or as combination therapy with further medications, e.g. cytostatic or chemotherapeutic agents, corticosteroids and/or antibiotics. The fusion protein is administered to a subject in need thereof, particularly a human patient, in a sufficient dose for the treatment of the specific conditions by suitable means. For example, the fusion protein may be formulated as a pharmaceutical composition together with pharmaceutically acceptable carriers, diluents and/or adjuvants. Therapeutic efficacy and toxicity may be determined according to standard protocols. The pharmaceutical composition may be administered systemically, e.g. intraperitoneally, intramuscularly or intravenously or locally, e.g. intranasally, subcutaneously or intrathecally. Preferred is intravenous administration.

The dose of the fusion protein administered will of course be dependent on the subject to be treated, on the subject's weight, the type and severity of the disease, the manner of administration and the judgement of the prescribing physician. For the administration of fusion proteins, a daily dose of 0.001 to 100 mg/kg is suitable.

Examples

1. Manufacture of a single-chain CD95L fusion protein (scCD95L)

In the following, the general structure of the recombinant proteins of the invention (Figure 1) is shown exemplified for the receptor binding domain of the human CD95 ligand.

1.1 Polypeptide structure

A) Amino acids Met1 -Ser21

IgKappa-signal peptide, assumed signal peptidase cleavage site after amino acid Gly20

B) Amino acids Glu22-Leu161

First soluble cytokine domain of the human CD95 ligand (CD95L; amino acids 142-281 of SEQ ID NO: 6 including a K145S mutation).

C) Amino acids Gly162-Gly169 First peptide linker element.

D) Amino acids Arg170-Leu307

Second soluble cytokine domain of the human CD95 ligand (CD95L; amino acids 144-182 of SEQ ID NO: 6 including a K145S mutation).

E) Amino acids Gly308-315 Second peptide linker element. F) Amino acids Arg316-Leu453

Third soluble cytokine domain of the human CD95 ligand (CD95L; amino acids 144-281 of SEQ ID NO: 6 including a K145S mutation).

G) Amino acid Gly457-Lys472

Peptide linker with a Strep-tag Il motif.

The amino acid sequence of sc CD95L is shown in SEQ ID NO. 27. The fusion polypeptide comprises first and second peptide linkers having the sequence GGSGSGSG (SEQ ID NO: 21). Further preferred linker sequences are SEQ ID NOs: 22-26 as described above. It should be noted that the first and second peptide linker sequences need not to be identical.

The signal peptide sequence (A) may be replaced by any other suitable, e.g. mammalian signal peptide sequence. The Strep-tag Il motif (G) may be replaced by other motifs, if desired, or deleted. As shown in Figure 23, cell culture supernatant of HEK293 cells, transiently expressing scCD95L (SEQ ID NO:27) was collected and used to stimulate Jurkat cells at varying concentrations. The supernatant was used either directly without further modifications or an anti-Streptag antibody (2 microgram/ml) was added to cross-link the scCD95L protein. Only cell supernatant that contained cross-linked scCD95L-St increased caspase activity in Jurkat cells, indicating that scCD95L alone does not form higher order aggregates able to be pro-apoptotic.

1.2 Gene cassette encoding the polypeptide

The synthetic gene may be optimised in view of its codon-usage for the expression in suitable host cells, e.g. insect cells or mammalian cells. A preferred nucleic acid sequence is shown in SEQ ID NO: 30.

1.3 Cloning strategy

The synthetic gene may be cloned, e.g. by means of a restriction enzyme hydrolysis into a suitable expression vector.

2. Manufacture of a single-chain TRAIL fusion protein (sc TRAIL wt) 2.1 Polypeptide structure

A) Amino acids Met1 -Gly20

Ig -Kappa-signal peptide, assumed signal peptidase cleavage site after amino acid GIy 20.

B) Amino acids Gln21 - Gly182

First soluble cytokine domain of the human TRAIL ligand (TRAIL, amino acid 120 - 281 of SEQ ID NO: 10)

C) Amino acids GIy 183 - Ser 190

First peptide linker element, wherein the two amino acids designated X are both S or one is S and the other one is N.

D) Amino acids Arg191 - Gly351

Second soluble cytokine domain of the human TRAIL ligand (TRAIL, amino acids 121 - 281 of SEQ ID NO:10)

E) Amino acids GIy 352 - Ser359 Second peptide linker element wherein the two amino acids designated X are both S or one is S and the other one is N.

F) Amino acids Arg360 - Gly520 Third soluble cytokine domain of the human TRAIL ligand (TRAIL, amino acids 121 - Gly281 of SEQ ID NO:10).

G) Amino acids Gly521 - Lys538

Peptide linker element with a Streptag Il motif. The amino acid sequence of sc TRAIL wt is shown in SEQ ID NO: 28.

The indicated linkers may be replaced by other preferred linkers, e.g. as shown in SEQ ID NOs: 21.26.

It should be noted that the first and second peptide linkers do not need to be identical.

The signal peptide sequence (A) may be replaced by any other suitable, e.g. mammalian signal peptide sequence. The Strep-tag Il motif (G) may be replaced by other motifs, if desired, or deleted. Cell culture supematants of HEK293 cells, transiently expressing single chain TRAIL proteins with different linkers (derived from SEQ ID 28, in total nine different linker combinations) were collected and used to stimulate Jurkat cells at varying dilutions (exemplarily, a dilution of 1 :8 is shown in Figure 25). The supematants were used either directly without further modifications or an anti-Streptag antibody (2 microgram/ml Strep MAB Immo) was added to cross-link the scTRAILwt proteins. Jurkat cells were incubated with HEK293 cell culture supernatant for three hours at 37°, lysed and analysed for caspase activity. Cell culture supernatant that contained cross-linked scTRAILwt proteins induced an increased caspase activity in Jurkat cells (results shown on the right hand side of the graph), indicating that scTRAILwt proteins alone do form only a low amount of higher order aggregates able to be pro- apoptotic.

2.2 Gene cassette encoding the polypeptide

The synthetic gene may be optimised in view of its codon usage for the expression in suitable host cells, e.g. insect cells or mammalian cells. A preferred nucleic acid sequence is shown in SEQ ID NO: 31.

3. Manufacture of a single-chain mutated TRAIL fusion protein (scTRAIL (R2-specific))

In the following, the structure of a single-chain TRAIL polypeptide comprising a mutation for selective binding to TRAIL receptor R2 is shown.

3.1 Polypeptide structure

A) Amino acids Met1 - Ser29

Ig -Kappa signal peptide, assumed signal peptidase cleavage site after amino acid Gly20 and peptide linker

B) Amino acids Arg29 - GIyI 90

First soluble cytokine domain of the human TRAIL ligand (TRAIL, amino acids 121-281 of SEQ ID NO: 10 including the mutations Y189Q, R191 K, Q193R, H264R, I266L and D267Q ) C) Amino acid Gly191 - Ser198

First peptide linker element, wherein the amino acids designated X are as indicated in Example 2

D) Amino acids Arg199 - Gly359

Second soluble cytokine domain of the human TRAIL ligand (TRAIL amino acids 121-281 of SEQ ID NO: 10 including the mutations as indicated in B)

E) Amino acids Gly360 - Ser367

Second peptide linker element, wherein the amino acids X are as indicated in Example 2

F) Amino acids Arg368 - Gly528

Third soluble cytokine domain of the human TRAIL ligand (TRAIL, amino acids 121-281 of SEQ ID NO: 10 including the mutations as indicated in B)

G) Amino acids Gly529 - Lys546

Peptide linker with a Strep-tag Il motif

The amino acid sequence of scTRAIL(R2-specific) is shown in SEQ ID NO: 29.

The indicated linkers may be replaced by other preferred linkers, e.g. as shown in SEQ ID NOs: 21-26. It should be noted that the first and second peptide linkers do not need to be identical.

The signal peptide sequence (A) may be replaced by any other suitable, e.g. mammalian signal peptide sequence. The Streptag Il motif (G) may be replaced by other motifs, if desired, or deleted.

3.2 Gene cassette encoding the polypeptide

The synthetic gene may be optimised in view of its codon usage for the expression in suitable host cells, e.g. insect cells or mammalian cells. A preferred nucleic acid sequence is shown in SEQ ID NO: 32.

4. Expression and Purification a) Cloning, expression and purification of fusion polypeptides

Hek293T cells grown in DMEM + GlutaMAX (GibCo) supplemented with 10% FBS, 100 units/ml Penicillin and 100 μg/ml Streptomycin were transiently transfected with a plasmid containing an expression cassette for a fusion polypeptide. In those cases, where a plurality of polypeptide chains is necessary to achieve the final product, e.g. for the Fab-scTNF-SF fusion proteins (Figure 9A), the expression cassettes were either combined on one plasmid or positioned on different plasmids during the transfection. Cell culture supernatant containing recombinant fusion polypeptide was harvested three days post transfection and clarified by centrifugation at 300 x g followed by filtration through a 0.22 μm sterile filter. For affinity purification Streptactin Sepharose was packed to a column (gel bed 1 ml), equilibrated with 15 ml buffer W (100 mM Tris-HCI, 150 mM NaCI, pH 8.0) or PBS pH 7.4 and the cell culture supernatant was applied to the column with a flow rate of 4 ml/min. Subsequently, the column was washed with 15 ml buffer W and bound polypeptide was eluted stepwise by addition of 7 x 1 ml buffer E (100 mM Tris HCI, 150 mM NaCI, 2.5 mM Desthiobiotin , pH 8.0). Alternately, PBS pH 7.4 containing 2.5 mM Desthiobiotin can be used for this step. The protein amount of the eluate fractions was quantitated and peak fractions were concentrated by ultrafiltration and further purified by size exclusion chromatography (SEC).

SEC was performed on a Superdex 200 column using an Akta chromatography system (GE- Healthcare). The column was equilibrated with phosphate buffered saline and the concentrated, Streptactin-purified polypeptide was loaded onto the SEC column at a flow rate of 0.5 ml/min. The elution profile of the polypeptide was monitored by absorbance at 280 nm.

For determination of the apparent molecular weight of purified fusion polypeptide under native conditions a Superdex 200 column was loaded with standard proteins of known molecular weight. Based on the elution volume of the standard proteins a calibration curve was plotted and the apparent molecular weight of purified fusion polypeptide was determined.

5. Apoptosis Assay

A cellular assay with a Jurkat A3 permanent T-cell line was used to determine the apoptosis inducing activity of different CD95-ligand (CD95L) and TRAIL fusion polypeptide constructs. Jurkat cells were grown in flasks with RPMI 1640-medium + GlutaMAX (GibCo) supplemented with 10% FBS, 100 units/ml Penicillin and 100 μg/ml Streptomycin. Prior to the assay, 100,000 cells were seeded per well into a 96-well microtiterplate. The addition of different concentrations of fusion peptides to the wells was followed by a 3 hour incubation at 37°C. Cells were lysed by adding lysis buffer (250 mM HEPES, 50 mM MgCI₂, 10 mM EGTA, 5% Triton-X-100, 100 mM DTT, 10 mM AEBSF, pH 7.5) and plates were put on ice for 30 minutes to 2 hours. Apoptosis is paralleled by an increased activity of caspases, e.g. Caspase-3. Hence, cleavage of the specific caspase substrate Ac-DEVD-AFC (Biomol) was used to determine the extent of apoptosis. In fact, Caspase activity correlates with the percentage of apoptotic cells determined morphologically after staining the cells with propidium iodide and Hoechst-33342. For the caspase activity assay, 20 μl cell lysate was transferred to a black 96-well microtiterplate. After the addition of 80 μl buffer containing 50 mM HEPES, 1% Sucrose, 0.1% CHAPS, 50 μM Ac-DEVD-AFC, and 25 mM DTT, pH 7.5, the plate was transferred to a Tecan Infinite 500 microtiterplate reader and the increase in fluorescence intensity was monitored (excitation wavelength 400 nm, emission wavelength 505 nm).

5.1 Cell death assay

For the determination of cell death in HT1080 fibrosarcoma cells 15,000 cells were plated in 96-well plates over night in RPM1 1640-medium + GlutaMAX (GibCo) supplemented with 10 % FBS (Biochrom). Cells were coincubated with cycloheximide (Sigma) at a final concentration of 2.5 g/ml. Cell death was quantified by staining with buffer KV (0.5% crystal violet, 20% methanol). After staining, the wells were washed with water and air-dried. The dye was eluted with methanol and optical density at 595 nm was measured with an ELISA reader.

6. Stability/Aggregation Test

6.1. Principle of the aggregation analysis (Definition for soluble protein)

The content of monomers (defined trimeric assembly of TNF-SF receptor binding modules) and aggregates is determined by analytical SEC as described in Example 4. For this particular purpose the analysis is performed in buffers containing physiological salt concentrations at physiological pH (e.g. 0.9% NaCI, pH 7.4; PBS pH 7.4). A typical aggregation analysis is done on a Superdex200 column (GE Healthcare). This column separates proteins in the range between 10 to 800 kDa. For determination of the apparent molecular weight of purified fusion polypeptide under native conditions a Superdex 200 column is loaded with standard proteins of known molecular weight. Based on the elution volume of the standard proteins a calibration curve is plotted and the apparent molecular weight of purified fusion polypeptide is calculated based on the elution volume. SEC analysis of soluble, non aggregated proteins, - e.g. trimeric TNF-SF, typically shows a distinct single protein peak at a defined elution volume. This elution volume corresponds to the apparent native molecular weight of the particular protein and approximately complies to the theoretical molecular weight calculated on the basis of the primary amino acid sequence.

If protein aggregation occurs the SEC analysis shows additional protein peaks with lower retention volumes. For TNF-SF family members the aggregation of soluble proteins occurs in a characteristic manner. The proteins tend to form oligomers of the "trimers", forming nonamers (3 x 3) and 27mers (3 x 9). These oligomers serve as aggregation seeds and a high content of oligomers potentially leads to aggregation of the protein. Oligomers of large molecular weight and aggregates elute in the void volume of the Superdex200 column and cannot be analysed by SEC with respect to their native molecular weight. Examples for SEC analysis of a defined soluble trimeric and a oligomerised/aggregated preparation of TNF-SF proteins are shown in Figure 17.

Due to the induction of (complete) aggregation, purified preparations of TNF-SF fusion proteins should preferably contain only defined trimeric proteins and only a very low amount of oligomerised protein. The degree of aggregation/oligomerisation of a particular TNF-SF protein preparation is determined on basis of the SEC analysis by calculating the peak areas of the OD280 diagram for the defined trimeric and the oligomer/aggregate fraction, respectively. Based on the total peak area the percentage of defined trimeric protein is calculated as follows:

(% Trimer content = [Peak area trimer] I [Total peak area] x 100)

The definition for soluble protein as used in this text, describes a protein preparation of purified TNF-SF protein in a buffer of physiological salt concentrations at physiological pH that contains a defined soluble protein (trimeric assembly of TNF-SF domains) content of >90% within a typical protein concentration range from 0.2 to 10.0 mg/ml.

6.2 SEC aggregation analysis for purified sc-TRAIL variants

Three different sc-TRAIL variants were transfected and affinity purified as described. The purified proteins were subsequently analysed for their content of defined soluble protein using SEC analysis as described in 6.1. In the particular case of single chain fusion proteins a trimer describes a trimeric assembly of three encoded TNF-SF domains encoded by a single polypeptide chain. (Formally single chain TNF-SF proteins are monomers, since single chain assemblies do only form intramolecular interactions [all protein domains are encoded by a single polypeptide chain] and do not form intermolecular interactions between distinct individual polypeptide chains.)

The proteins analysed by SEC were:

1.) Fab-sc-TRAIL(R2-specific)-SNSN (Figure 19):

Fusion protein comprising an Fab domain fused N-terminal to a single chain fusion protein of TRAIL specific for TRAIL-receptor 2 interaction, glycosylated

2.) Fab-sc-TRAIL(R2-specific)-SSSS (Figure 18)

Fusion protein comprising an Fab domain fused N-terminal to a single chain fusion protein of TRAIL specific for TRAIL-receptor 2 interaction, non glycosylated

3.) Fab-sc-TRAIL-wt-SNSN (Figure 20): Fusion protein comprising an Fab domain fused N-terminal to a single chain TRAIL, glycosylated

The SEC analysis for the three purified Fab-sc-constructs of TRAIL revealed a single protein peak for all proteins indicating defined soluble protein fractions (>95% trimer). The calculated apparent MW for the proteins (based on calibration of the column) strongly indicate a trimeric association of the TNF-SF- domains for the purified proteins. None of the analysed proteins showed indications for aggregation (Figures 18, 19, 20).

Comparing the potentially glycosylated "Fab-sc-TRAIL-R2-SNSN" with the non glycosylated "Fab-sc- TRAIL-R2-SSSS" indicates a significant difference of the apparent native MW that is due to glycosylate of Fab-sc-TRAIL(R2-specific)-SNSN. Expression of sc-TNF-SF members as fusion protein with an antibody fv-fragment is known to facilitate aggregation of the protein. The construction principle of the Fab-sc-TRAIL variants revealed no aggregation of the expressed TRAIL variants and is therefore beneficial with respect to solubility of the protein.

6.3 Differential glycolsylation of sc-TRAIL-linker variants

Glycosylate of proteins can be beneficial for recombinant sc-TNF-SF constructs with regard to potential immunogenicity and stability. In order to get glycosylate of the sc-TRAIL construct, specific linker sequences were designed that contained putative N-linked glycosylate sites at defined positions (see Figure 21 -A). Recombinant expression and subsequent Western-Blot analysis revealed that the respective position of the Asparagine (N) within the linker sequence is important for the subsequent glycosylation of the protein. Surprisingly, the preferential linker position of the glycosylated asparagine was identified to be at position "2" as described in Figure 21 -A, (G S G S G N G S). If the asparagine is localised at other positions (e.g. position "1" [G S G N G S G S] see Figure 21 -A), glycosylation of the respective asparagines(s) is abolished. This aspect could be confirmed by Western-Blot analysis of different sc-TRAIL variants. If both asparagines of linker 1 and linker 2 were localised at position"2" a significant glycosylation dependant MW-shift could be observed for the respective sc-TRAIL variant (Figure 22). A MW-shift of the glycosylated sc-TRAIL linker variant could also be confirmed by calculating the apparent MW after SEC analysis (Figure 18, 19). The non glycosylated Fab-sc- TRAIL(R2-specific)SSSS has a clearly lower MW (68 k Da) compared to glycosylated Fab-sc- TRAIL(R2-specific)SNSN (87 kDa).

Based on this analysis we claim differential glycosylation of the sc-TRAIL constructs by modifying the position of the asparagines within the linker sequence(s). Glycosylation protects the linker sequence towards proteolytic degradation and might stabilise the protein. In addition glycosylation of the linker sequence potentially prevents recognition of the linker sequence by the immune system and potentially reduces the immunogenicity of the protein. Therefore glycosylation of the linker sequence is beneficial with regard to immunogenicity and proteolytic stability of the sc-TRAIL constructs and has potential influence on the half life of the protein. The linker specific differential glycosylation can be used to modify the immunogenicity and stability of recombinant TNF-SF members.

6.3. Expression and analysis of a sc-TRAIL with prolonged linker sequence and N-terminal stalk residues (sc-TRAIL-(95-281)-long)

In WO/2005/103077 a single chain TRAIL-fusion polypeptide, herein named sc-TRAIL-(95-281)-long, is described, wherein each TRAIL module comprise residues 95 to 281 of SEQ ID NO:10. The TRAIL modules are linked by Glycin Serin linker comprising of at least 12 amino acids (GGGSGGGSGGGS). Compared to the TRAIL modules of the present invention (comprising residues 121 -281 of SEQ ID NO:10), additional 25 amino acids including the stalk region are present in each of the adjacant TRAIL modules. In order to analyse the influence of the linker sequence on sc-TRIAL constructs, sc-TRAIL-(95-281)- long is analysed. Expression, purification and subsequent SEC analysis reveals that sc-TRAIL-(95- 281)-long with the 12 aa linker and the additional stalk sequence is expressed and secreted to the cell culture supernatant of HEK293T cells. However, SEC analysis of the purified protein indicates that sc- TRAIL-(95-281)-long shows multiple peaks comprising a large amount of protein in an oligomerised or aggregated from. Aggregation of sc-TRAIL-(95-281)-long is a direct effect of the prolonged linker sequences in combination with the additional residues of the N-terminal stalk. The results indicate that the longer linker used in this construct leads to increased aggregation properties of the construct.

7. Construction of single-chain fusion polypeptides comprising one or more additional domains

7.1. Assembly of soluble TNF-SF and antibody fragments known from the art

It is known from the art that soluble TNF-SF cytokine domains may be fused to antibody fragments in order to obtain trimerisation and/or dimerisation of trimers. Single-chain scFv-TNF-SF fusion proteins have been constructed consisting of a single-chain antibody and a soluble domain comprising a TNF-

RBD and the stalk-region. The corresponding trimers consist of three single-chain antibodies and three soluble domains (Fig. 7).

In addition, Fc-TNF-SF fusion proteins, wherein each fusion protein comprises an N-terminal intramolecular Fc-domain and a C-terminal soluble domain have been constructed (Figure 8). The dimerisation of soluble domains is accomplished by assembly of two Fc-domains via disulfide bridges.

Trimers are subsequently obtained by a combination of two soluble domains from one Fc-TNF-SF fusion protein and one soluble domain from another Fc-TNF-SF fusion protein. As can be deduced from

Fig.4, dimerisation of trimers is also mediated by the N-terminal Fc-TNF-SF fusion. In conclusion, three Fc-antibody fragments are present per dimer of the trimer. However, such fusion proteins are likely to form higher molecular weight aggregates, which represents a major disadvantage.

7.2 Fusion proteins of the invention comprising one or more additional domains

The inventive fusion proteins comprising one or more additional domains can be constructed in several ways. In the following, the construction of fusion proteins with additional domains is exemplified with the antibody pertuzumab directed against the cell surface antigen ErbB2. The amino acid sequence of the heavy chain is shown in SEQ ID NO: 33:

1 EVQLVESGGG LVQPGGSLRL SCAASGFTFT DYTMDWVRQA PGKGLEWVAD VNPNSGGSIY

61 NQRFKGRFTL SVDRSKNTLY LQMNSLRAED TAVYYCARNL

GPSFYFDYWG QGTLVTVSSA 121 STKGPSVFPL APSSKSTSGG TAALGCLVKD YFPEPVTVSW

NSGALTSGVH TFPAVLQSSG 181 LYSLSSWTV PSSSLGTQTY ICNVNHKPSN TKVDKKVEPK SC The amino acid sequence of the light chain is shown in SEQ ID NO: 34

1 DIQMTQSPSS LSASVGDRVT ITCKASQDVS IGVAWYQQKP GKAPKLLIYS ASYRYTGVPS 61 RFSGSGSGTD FTLTISSLQP EDFATYYCQQ YYIYPYTFGQ

GTKVEIKRTVAAPSVFIFPP

121 SDEQLKSGTA SWCLLNNFY PREAKVQWKV DNALQSGNSQ ESVTEQDSKD STYSLSSTLT

181 LSKADYEKHK VYACEVTHQG LSSPVTKSFN RGEC

7.2.1

In one embodiment, the fusion polypeptide of the invention further comprises an N- or C-terminal Fab- antibody fragment (Fig.9A).

The fusion of an antibody Fab-fragment to the N-terminus of scTNF-SF fusion polypeptide may be accomplished by the following two strategies:

(i) The heavy chain sequence is extended by further amino acids from the IgGI hinge region and fused to the single-chain TNF-SF fusion protein.

The IgGI hinge region comprises the amino acid sequence SEQ ID NO: 35:

....KSC₁DKTHTCIPPCSPAPE...

In a preferred embodiment, the Fab-domain is chosen such that the C-terminal cysteine of the heavy chain (C1 of the hinge region) terminates the CH1 domain. This cysteine is required for forming a disulfide linkage to the light chain.

The subsequent linker comprises portions of the IgG hinge region (e.g. DKTHT or DKT), however without further cysteines of the hinge region. Alternatively, a glycine/serine inker is used. Due to the absence of further cysteines, a monomeric fusion protein comprising two polypeptide chains is obtained. The linker preferably has a length of 3-15 amino acids. More preferably, the linker is selected from the linker 1-7 as shown below.

1. DKTHTG(S)a(G)b; (a=0-5; b=0 Oder 1) 2. DKTHTGS(S)a(GS)bG(S)c (a, b =0,1-6; c=0 Oder 1)

3. DKTG(S)a(G)b; (a=0-5; b=0 Oder 1) 4. DKTG(S)a(GS)bG(S)c (a, b =0,1-6; c=0 Oder 1)

5. SSG(S)a(GS)bG(S)c (a, b =0,1 -6; c=0 Oder 1)

6. SS(GGGS)aG(S)b (a=0, 1-4; b=0 Oder 1)

7. GSPGSSSSSS(G)a (a=0 Oder Preferred amino acid sequences with the heavy chain module positioned N-terminal to the scTNF-SF module are shown in SEQ ID NO: 45, SEQ ID NO: 47 and SEQ ID NO: 49. For production purposes, these polypeptide chains are coexpressed with the Fab light chain polypeptide (SEQ ID NO: 40) to finally achieve the Fab-scTRAIL fusion polypeptides, (ii) The light-chain sequence is fused to the single chain TNF-SF fusion protein. The constant region of the light chain (e.g. SEQ ID NO: 34) ends with a C-terminal cysteine residue. This residue may be covalently bridged with the C1 hinge cysteine of the heavy chain. Preferably, the linkers 1-7 as shown below are used for the connection between the light chain sequence and the TNF- SF fusion protein. Linkers 5-7 are preferred (see above). Preferably, the last amino acid in the linker adjacant to the cytokine module is either GIy or Ser. In the following, preferred linker sequences are shown:

Further, the linker may comprise N-glycosylation motifs (NXS/T, wherein X may be any amino acid). One embodiment of the amino acid sequences with the light chain module positioned N-terminal to the scTNF-SF module is shown in SEQ ID NO: 51. In the case of the Fab-scTNF-SF fusion proteins, the co-expression of two polypeptide chains is necessary to achieve the correct assembly of the Fab module in addition to the scTNF-SF module (see figure 9A). The Pertuzumab heavy and light chain modules (SEQ ID NO: 33 and SEQ ID NO: 34) were equipped with a signal peptide, backtranslated and the resulting synthetic genes (SEQ ID NO: 41 and SEQ ID NO: 42) genetically fused upstream of the scTRAILwt- or scTRAILR2-specific gene modules (SEQ ID NO: 31 and SEQ ID NO: 32). Examples for the resulting gene cassettes are shown in SEQ ID NO: 46, 48 and 50. After subcloning into appropriate expression vectors, a selection of the resulting plasmids was used for transient protein expression in HEK293T cells. The heavy chain TRAIL or light chain TRAIL expression plasmids were transfected either alone or in combination with the necessary light or heavy chain encoding vectors of the Fab-Fragment (F igure 26). Surprisingly, the module combination within the fusion proteins influenced the relative stability of the scTRAIL-protein during secretory based expression. If the light-chain module of the Fab-domain is fused N-terminal to the scTRAIL-domain (exemplified in SEQ ID NO: 51), the expression product is stable itsself and secreted, when expressed separatly (Lanes 1-4, figure 26). It can be therefore expected, when such a fusion polypeptid is coexpressed with a heavy-chain module, that two major protein species will be formed during a potential production process: (1) the Fab-scTRAIL fusion protein consisting of two polypeptide chains and (2) as contamination a light-chain-scTRAIL fusion protein without a functional Fab domain. Therefore, fusing the heavy-chain module N-terminal to the scTNF-SF-module for the expression is preferred to avoid this technical disadvantage. A functional analysis of recombinant inventive Fab comprising-scTRAIL fusion proteins with the heavy- chain module fused N-terminal to the scTRAIL-module (Fab-scTRAILR2-SNSN or Fab-scTRAILwt- SNSN) is shown in figure 28. As final purification step, size exclusion chromatography was employed as examplified in figures 19 and 20. Superior bioactivity compared to soluble, homotrimeric ligands can easily be achieved by the use of artificially cross-linked or a membrane-bound ligand of the TNF superfamily. Thus the local enrichment of single chain TRAIL (scTRAIL) constructs on cells that express the antigen Her2 via the Her2- selective Fab-fragment ("Pertuzumab") fused to these scTRAIL proteins should increase their cytotoxic bioactivity. Likewise, the blocking of the Her2 binding sites on cells by pre-incubation with the Her2- specific Fab-fragment (Pertuzumab-Fab) only should decrease the cytotoxic bioactivity of Fab-scTRAIL fusion proteins. As shown in Figure 28A, scTRAIL constructs induce the death of HT1080 cells, as the viability decreases with increasing protein concentration. In accordance, the pre-incubation of HT1080 cells with the Fab-fragment (Pertuzumab-Fab), followed by co-incubation with the Fab-scTRAIL constructs (Fab-scTRAILR2-SNSN or Fab-scTRAILwt-SNSN) over night, reduced the cytotoxic activity of the Fab-scTRAIL constructs (figure 28B), whereas the Fab only induced no cell death. An increased technical effect may be achieved by use of artificially cross-linked or a membrane-bound ligands of the TNF superfamily resulting especially in superior bioactivity as compared to soluble, homotrimeric ligand. Thus the local enrichment of ligands or single chain ligands such as examplifed by single chain TRAIL (scTRAIL) on cells or on neighbouring cells should increase the bioactivity of these fusion proteins. The local enrichment (or targeting) of these single chain ligands can be specifically induced for instance by fusing the single chain ligands with amino acid sequences that bind to any antigen present on cells such as for instance tumor cells. Examples for antigen binding sequences may be derived from antibodies such as scFv or Fab fragments. Examples for antigens expressed on target cells may be receptors such as from the EGFR family or any other antigen to which a binding antibody can be generated. Of special interest in this context are cell surface antigens specific for tumor or cancer cells.

7.2.2

In another embodiment, the fusion polypeptide of the invention further comprises an additional N - or C- terminal scFv-antibody fragment (Fig.9B).

In this embodiment linkers 5-7 as described above may be used. Further, the linkers may comprise N- glycosylation motifs. A preferred single chain Fv-pertuzumab fragment for fusing to the single-chain cytokine fusion protein may comprise amino acids GIuI-SeM 19 of SEQ ID NO: 33 and Asp-Lys107 or Thr109 of SEQ ID NO: 34. The VH and VL fragments may be connected by a linker.

One embodiment of a scFv-domain of pertuzumab is shown in the following SEQ ID NO: 36:

1 METDTLLLWV LLLWVPAGNG EVQLVESGGG LVQPGGSLRL

SCAASGFTFT DYTMDWVRQA 61 PGKGLEWVAD VNPNSGGSIY NQRFKGRFTL SVDRSKNTLY

LQMNSLRAED TAVYYCARNL 121 GPSFYFDYWG QGTLVTVSSG GGGSGGGGSG GGGSDIQMTQ

SPSSLSASVG DRVTITCKAS 181 QDVSIGVAWY QQKPGKAPKL LIYSASYRYT GVPSRFSGSG

SGTDFTLTIS SLQPEDFATY 241 YCQQYYIYPY TFGQGTKVEI KRT

Amino acids 1 -20 (underlined) constitute an N -terminal secretory signal peptide.

7.2.3

In a further embodiment, the fusion polypeptide of the invention comprises an additional N- or C- terminal Fc-antibody fragment (Figs. 10 and 11 ).

Preferably, the Fc-antibody fragment domain is derived from a human immunoglobulin G heavy chain, particularly from a human immunoglobulin IgGI heavy chain. In an especially preferred embodiment, the amino acid sequence of the Fc-domain is shown in SEQ ID NO: 37.

1 KSCDKTHTCP PCPAPELLGG PSVFLFPPKP KDTLMISRTP

EVTCVWDVS HEDPEVKFNW 61 YVDGVEVHNA KTKPREEQYN STYRWSVLT VLHQDWLNGK

EYKCKVSNKA LPAPIEKTIS

121 KAKGQPREPQ VYTLPPSREE MTKNQVSLTC LVKGFYPSDI AVEWESNGQP ENNYKTTPPV

181 LDSDGSFFLY SKLTVDKSRW QQGNVFSCSV MHEALHNHYT

QKSLSLSPGK

Amino acids Lys1-Glu16 define the hinge region. For a C-terminal fusion (Fig. 11) the Fc-domain preferably comprises the complete constant domain (amino acids 17-230 of SEQ ID NO: 37) and a part or the complete hinge region, e.g. the complete hinge region or the hinge region starting from amino acid Asp4.

Preferred linkers for connecting a C-terminal Fc-antibody fragment (e.g. Fig. 11) are shown in the following:

Linker 8 scCD95L/scTRAIL....GG(P/S)a(GS)_b(G/S)c KSCDKTHTCPPCPAPE... (a=0 oder 1; b=0-8; c=0-8)

Linker 9 scCD95L/scTRAIL....GG(P/S)a(GSSGS)_bGS(G/S)c DKTHTCPPCPAPE...

(a=0 oder 1; b=0-8; c=0-8) Linker 10 scCD95L/scTRAIL....GG(P/S)a(S)b(GS)c(G/S)_d DKTHTCPPCPAPE... (a=0 oder 1; b=0-8; c=0-8; d=0-8)

All linkers start with GIyGIy taking in account, however, that the C-terminal amino acid of TRAIL is a GIy. At position 3 of the linker, alternatively Pro or Ser are present. Linker 8 comprises the Cys1 cysteine of the heavy chain.

It should be noted that linkers 8-10 are also suitable for the C-terminal fusion of other polypeptides, e.g. a further scTNF-SF fusion protein.

In detail, the scTRAILwt module (SEQ ID NO: 28), the scTRAIL(R2-specific)-module (SEQ ID NO: 29) and the scCD95L-module (SEQ ID NO: 27) were fused N-terminally to the Fc-domain of human IgGI , starting with Asp4 of SEQ ID NO: 37 employing four linker elements as shown in table 2.

Table 2: Sequences linking the Fc-domain C-terminally to scTNF-SF module. The N-terminal amino- acid of the IgGI CH2-domain is underlined. The N-terminal Glycine of the linking sequence is shown in brackets. For TNF-SF proteins with a glycine as the C-terminal amino acid (e.g. TRAIL), the N-terminal glycine of the linking sequence formally belongs to the scTNF-SF module.

For purification and characterisation, a Strep-tag Il (amino acid sequence WSHPQFEK) was placed C- terminally to the Fc-domain. This affinity tag was linked to the CH3-domain by a flexible linker element (amino acid sequence SSSSSSA), replacing the C-terminal lysine residue of the CH3-sequence. The amino acid sequences of the scTNF-SF fusion proteins as well as for the described protein modules were backtranslated and their codon usage was optimised for mammalian cell-based expression. Gene synthesis was done by ENTELECHON GmbH (Regensburg, Germany). The expression cassettes for larger fusion proteins were assembled by common cloning procedures starting with DNA-modules of suitable size and suitable restriction enzyme pattern. Exemplarily, the resulting gene cassette for the single chain TRAILwt FC01 fusion protein (scTRAILwt-FC01) is shown in SEQ ID NO: 44 and the encoded protein sequence is shown in SEQ ID NO: 43. The gene cassettes encoding the shortened linker variants (table 1) were generated by PCR based subcloning strategies, starting from SEQ ID NO: 44. The final expression cassettes were released from intermediate cloning vectors and subcloned into to pCDNA4-HisMax-backbone, using unique Hind-Ill-, Not-I- or Xba-I sites of the plasmid. For the assembly of the Fab- and Fc-fusions proteins, a unique SgS-I site was introduced into the vector backbone, replacing the Not-l-site. All expression cassettes were routinely verified by DNA sequencing. The proteins were transiently expressed in HEK293T cells and the cell culture supematants were monitored regarding their pro-apoptotic activity. As shown in figure 27, the scTRAIL-Fc fusion proteins of the invention, were able to induce a pronounced increase in caspase activity, confirming the potency of the Fc-based dimerisation of two scTRAILwt-modules. Similar results were obtained for scTRAIL(R2- specific)-Fc fusion proteins (data not shown).

If an Fc-antibody fragment is fused to the N-terminus of an scTNF-SF fusion protein (cf. Fig. 10), the amino acid squence of the Fc-module is preferably as shown in SEQ ID NO: 38: 1 METDTLLLWV LLLWVPAGNG DKTHTCPPCP APELLGGPSV

FLFPPKPKDT LMISRTPEVT 61 CVWDVSHED PEVKFNWYVD GVEVHNAKTK PREEQYNSTY

RWSVLTVLH QDWLNGKEYK 121 CKVSNKALPA PIEKTISKAK GQPREPQVYT LPPSREEMTK

NQVSLTCLVK GFYPSDIAVE 181 WESNGQPENN YKTTPPVLDS DGSFFLYSKL TVDKSRWQQG

NVFSCSVMHE ALHNHYTQKS 241 LSLSPG

Amino acids 1-20 (underlined) constitute an N -terminal secretory signal peptide. For connecting the Fc-module to the ScTNF-SF fusion protein, preferably Gly/Ser linkers are used. All linkers preferably start with a serine and preferably end with glycine or serine. Preferred linker sequences 11-12 are shown in the following:

11. (S)a(GS)_bG(S)c (a, b =0,1-6; c=0 Oder 1)

12. S(GGGS)aG_b(S)c (a, b =0,1-6; c=0 Oder 1)

7.3 Dimerisation of the single-chain fusion proteins of the invention

7.3.1 Single-chain fusion polypeptides comprising one additional domain

The trimeric fusion proteins of the invention can further be dimerised.

In one embodiment, dimerisation will be obtained if the C-terminus of a first fusion protein is directly connected to the N-terminus of a second fusion protein via a linker structure as defined herein (Fig. 12).

In another embodiment, a fusion protein of the invention comprising an Fab -antibody fragment as an additional domain, may be connected via a linker as defined herein directly with a further fusion protein of the invention or indirectly via an scFv-antibody fragment fused to a further fusion protein of the invention (Fig. 13). Thereby, dimerisation of the trimeric fusion proteins of the invention is accomplished.

In another embodiment, dimerisation of trimers may be obtained via the assembly of two fusion proteins of the invention comprising a Fab-antibody fragment as an additional domain (Fig. 14). In this case, intermolecular disulfide bridges are formed.

For the construction of dimerising Fab fragments N-terminal to the scTNF-SF domain (e.g. Fig. 14), preferably the natural cysteine residues of the IgG hinge region (SEQ ID NO: 35) are used. Preferably the C-terminal cysteine of the Fab-sequence corresponds to the C1 -residue of the hinge region, which forms a disulfide bond with the light chain. The second cysteine C2 may be used for the covalent linkage of two Fab-modules. A third cysteine residue C3 may be open or linked with the C3 of the neighbouring chain. Preferred linkers between the Fab heavy chain sequence and the N-terminus of the scTNF-SF domain are linkers 13-22 as shown below.

13. DKTHTCPGSS(GS)_aG(S)b 14. DKTHTCPGSSaG(S)_b 15. DKTHTC(GSSGS)_aGSG(S)b 16. DKTHTCGSS(GS)_aG(S)b 17. DKTHTCGSS_aG(S)b 18. DKTHTC(GSSGS)_aGS(G)b

19. DKTHTCPPCPGSSGSGSGS(G)_b

20. DKTHTCPPCP(GSSGS)_aGS(G)b 21. DKTHTCPPCPGSS(GS)_aGS(G)b

22. DKTHTCPPCPGSS_aGS(G)b

Further, the linkers may be modified by incorporation of N-glycosylation motifs as described above. In a further embodiment, dimerisation of the fusion proteins of the invention comprising an Fc -antibody fragment as an additional N- and/or C-terminal domain, may be obtained by the formation of intermolecular disulfide bridges between two of said fusion proteins. In that case, only one Fc -antibody fragment is present per dimer of a trimeric fusion protein. Thereby, in contrast to the Fc -antibody fragment fusion proteins of the art, formation of higher molecular weight aggregates is not very likely.

7.3.2 Single-chain fusion polypeptides comprising a plurality of additional domains

The single-chain fusion polypeptide may comprise one or more additional domains, e.g. a further antibody fragment and/or a further targeting domain and/or a further cytokine domain. A fusion protein of the invention comprising an Fc-antibody fragment as one additional domain may be connected to a further Fab- or scFv-antibody fragment via the N-terminus of an N-terminal fused Fc- antibody fragment (Fig. 15) or directly via its N-terminus through a further linker structure (Fig. 16), if the Fc-antibody fragment is connected to the fusion protein of the invention via its C-terminus. In addition to a further antibody fragment or instead of the further antibody fragment, a further cytokine, preferably an interleukin, may be connected to the fusion protein. Thereby, it is possible to obtain a combination of an agonistic scCD95L and an antagonistic scCD95L molecule or alternatively combinations of scTRAIL (R1 -specific) and scTRAIL (R2-specific).

Said fusion proteins are especially useful for the induction of apoptosis.

Claims

1. A single-chain fusion polypeptide comprising:

(i) a first soluble TNF superfamily cytokine domain, (ii) a first peptide linker,

(iii) a second soluble TNF superfamily cytokine domain, (iv) a second peptide linker, and (v) a third soluble TNF superfamily cytokine domain, wherein the fusion polypeptide is substantially non-aggregating.

2. The polypeptide of claim 1 , wherein the soluble TNF superfamily cytokine domain is selected from soluble CD95L or TRAIL domains.

3. The polypeptide of claim 1 or 2, wherein the second and/or third soluble TNF superfamily cytokine domain is an N-terminally shortened domain which optionally comprises amino acid sequence mutations.

4. The polypeptide of any one of claims 1-3, wherein at least one of the soluble TNF superfamily cytokine domains, particularly at least one of the soluble TNF superfamily cytokine domains (iii) and (v) is a soluble CD95L domain with an N-terminal sequence which starts between amino acid Arg

144 to Val146 of human CD95L, and wherein Arg144 and/or Lys145 may be replaced by a neutral amino acid, e.g. by Ser or GIy.

5. The polypeptide of claim 4, wherein at least one of the soluble TNF superfamily cytokine domains, particularly at least one of the soluble TNF superfamily cytokine domains (iii) and (v) is a soluble

CD95L domain with an N-terminal sequence selected from

(a) Arg 144 - (Gly/Ser) 145 - VaI (146)

(b) (Gly/Ser) 144 - Lys 145 - VaI (146) and (C) (Gly/Ser) 144 - (Gly/Ser) 145 - VaI (146).

6. The polypeptide of any one of claims 2-5 wherein the soluble CD95L domain ends with Leu281 of human CD95L and/or optionally comprises a mutation at position Lys177, e.g. Lys177 → GIu, Asp or Ser and/or at position Tyr218, e.g. Tyr218 → Arg, Lys, Ser, Asp.

7. The polypeptide of any one of claims 1-3 wherein at least one of the soluble TNF superfamily cytokine domains, particularly at least one of the soluble TNF superfamily cytokine domains (iii) and (v) is a soluble TRAIL domain with an N-terminal sequence which starts between amino acid Gln120 and VaM 22 of human TRAIL and wherein Arg121 may be replaced by a neutral amino acid, e.g. Ser or GIy.

8. The polypeptide of claim 7, wherein at least one of the soluble TNF superfamily cytokine domains, particularly at least one of the soluble TNF superfamily cytokine domains (iii) and (v) is a soluble TRAIL domain with an N-terminal sequence selected from (a) Arg121 - Val122 - Ala123 and

(b) (Gly/Ser) 121-Val122-Ala123.

9. The polypeptide of claims 7 or 8 wherein the soluble TRAIL domain ends with amino acid Gly281 of human TRAIL and/or optionally comprises a mutation at positions R130, G160, H168, R170, H177, Y189, R191 , Q193, E195, N199, K201 , Y213, T214, S215, H264, I266, D267 or D269 or at two or more of said positions.

10. The polypeptide of any one of claims 1 -9 wherein the first and second peptide linkers independently have a length of 3-8 amino acids, particularly a length of 3, 4, 5, 6, 7 or 8 amino acids, and preferably are glycine/serine linkers, optionally comprising an asparigine residue which may be glycosylated.

11. The polypeptide of any one of claims 1 -10 which additionally comprises an N-terminal signal peptide domain, which may comprise a protease cleavage site, and/or which additionally comprises a C- terminal element which may comprise and/or connect to a recognition/purification domain.

12. The polypeptide of any one of claims 1-11 which additionally comprises N-terminally and/or C- terminally a further domain, e.g. a single-chain antibody or antibody fragment, such as a Fab or Fc fragment domain.

13. A nucleic acid molecule encoding a fusion polypeptide of any one of claims 1-11 , preferably in operative linkage with an expression control sequence.

14. A cell or a non-human organism transformed or transfected with a nucleic acid molecule of claim 13, wherein the cell is e.g. a prokaryotic cell or a eukaryotic cell, preferably a mammalian cell or more preferably a human cell.

15. A pharmaceutical or diagnostic composition comprising as an active agent a fusion protein of any one of claims 1-12 or a nucleic acid molecule of claim 13, particuarly for use in therapy, more particularly in the prophylaxis and/or treatment of disorders caused by, associated with and/or accompanied by dysfunction of TNF cytokines, particularly proliferative disorders, such as tumours, e.g. solid or lymphatic tumours; infectious diseases; inflammatory diseases; metabolic diseases; autoimmune disorders, e.g. rheumatoid and/or arthritic diseases; degenerative diseases, e.g. neurodegenerative diseases such as multiple sclerosis; apoptosis-associated diseases or transplant rejections.