WO2023006390A1

WO2023006390A1 - Mesothelin antibodies and use thereof

Info

Publication number: WO2023006390A1
Application number: PCT/EP2022/069175
Authority: WO
Inventors: Karl-Peter Hopfner; Marion SUBKLEWE; Lorenza WYDER PETERS; Anna REISCHER; Björn HILLER; Nadja Fenn; Elisabeth Kremmer; Heinrich FLASWINKEL; Heinrich Leonhardt
Original assignee: Ludwig Maximilians Universitaet Muenchen LMU
Current assignee: Ludwig Maximilians Universitaet Muenchen LMU
Priority date: 2021-07-30
Filing date: 2022-07-08
Publication date: 2023-02-02
Anticipated expiration: 2024-01-30

Abstract

The present invention relates to a protein construct comprising (i) an anti-mesothelin (MSLN) IgG1 antibody or antigen-binding fragment thereof and (ii) a polypeptide capable of binding to CD47 present on the surface of a tumor cell. Further envisaged is a nucleic acid comprising a polynucleotide encoding the protein construct, a vector comprising the nucleic acid molecule, a host cell comprising the nucleic acid molecule, a method of producing the protein construct, a product produced by the method as well as a pharmaceutical composition comprising the protein construct. The present invention further relates to a method for the treatment of cancer and the pharmaceutical composition for use in the treatment of cancer.

Description

Mesothelin antibodies and use thereof

FIELD OF THE INVENTION [0001] The present invention relates to a protein construct comprising (i) an anti-mes- othelin (MSLN) IgGl antibody or an antigen-binding fragment thereof and (ii) a polypep tide capable of binding to CD47 present on the surface of a tumor cell. Further envisaged is a nucleic acid sequence comprising a polynucleotide encoding the protein construct, a vector comprising the nucleic acid sequence, a host cell comprising the nucleic acid sequence, a method of producing the protein construct, a product produced by the method as well as a pharmaceutical composition comprising the protein construct. The present invention further relates to a method for the treatment of cancer and the phar maceutical composition for use in the treatment of cancer.

BACKGROUND OF THE INVENTION [0002] Mesothelin (MSLN) is a tumor antigen that is expressed on the surface of sev eral types of cancer cells, including pancreatic cancer, ovarian cancer, lung adenocarci nomas, malignant mesothelioma and gastric cancer (Weidemann et al., 2021, Biomedi cines, 9 (4), 397). MSLN has also been shown to be expressed in breast cancer, particu larly triple negative breast cancer and for some of these indications a correlation of MSLN expression with poor prognosis could be established (Tozbikian et al., 2014, PLos ONE 9 (12), ell4900). [0003] Moreover, it has been shown that monoclonal anti-MSLN antibodies do localize to MSLN-positive tumors in patients with pancreatic or ovarian cancer (Lamberts et al., 2016, Clin Cancer Res., 22(7), 1642).

[0004] MSLN is a glycophosphatidylinositol (GPI) -linked cell-surface glycoprotein. It is synthesized as a 71 kDa precursor protein and is then cleaved to release the secreted N- terminal region, called megakaryocyte potentiating factor (MPF). The 41 kDa mature MSLN remains attached to the membrane. In normal tissues MSLN is only expressed in mesothelial cells which line the pleura, peritoneum, and pericardium and which are dis pensable. Similarly, mice in which MSLN has been knocked out are viable and do not show any gross abnormalities. Due to its distribution MSLN was identified as a differen tiation factor for mesothelial cells, but the function of MSLN in these cells remains un clear (Hassan et al., 2004, Clin Cancer Res, 10, 3937).

[0005] MSLN has further been identified as target for anti-tumor therapies, particu larly in indications with very poor prognosis, such as ovarian cancer, triple negative breast cancer or pancreatic cancer. Currently, 5-year survival rates are: 47 % for ovarian cancer (average of all types and stages), 68% for triple negative breast cancer and only 9% for pancreatic cancer.

[0006] Several therapeutic approaches have therefore been explored targeting MSLN using monoclonal antibodies (mAbs) or coupling these to toxins. MSLN is also targeted by chimeric T cells containing antibody fragments (mainly fragment variable, Fv) that recognize MSLN (CAR-T cells) ora Listeria monocytogenes vaccine expressing MSLN (LM- mesothelin). Amatuximab (MORAb-009) is a chimeric anti-MSLN monoclonal antibody (17% mouse, 83% human sequences) (see WO 1999/028471) that entered clinical trials in 2006 for patients with mesothelioma and pancreatic cancer. So far, no effect on pro- gression free survival has been seen in any of the phase I or phase II clinical trials as single agent or in combination with chemotherapy. Although the combined therapy of amatuximab plus pemetrexed/cisplatin for mesothelioma gave encouraging results in terms of the median overall survival of 14.8 months and an objective response rate of 39%, the follow-up trial with the same regimen and patient population was terminated early. As of July 2021, only a radio-labeled version of amatuximab is in clinical trials and only as diagnostic tool to help better localization of tumors and metastases. This shows that even combined with heavy chemotherapy (pemetrexed/cisplatin) amatuximab did not show any benefits for cancer patients.

[0007] MSLN has also been used as target for antibody-drug conjugate (ADC) ap proaches. MSLN is accordingly only used to bring a toxin or cytostatic that is attached to an anti-MSLN mAb, into the tumor cell and thereby kill the tumor cell. Examples of such ADCs are SS1P (CAT 5001) from the National Cancer Institute, Bethesda, USA and Cam- bridge Antibody Technology, anetumab ravtansine from Bayer Schering, BMS-986148 from Bristol Meyers Squibb, DMOT4039A from Roche/Genentech and LMB-100 from Selecta Biosciences. The first clinical trials with these agents have started in 2011, and many have been completed by now.

[0008] However, no positive breakthrough has yet emerged from these studies, i. e. in clinical trials no single agent activity has been seen so far with anti-MSLN antibodies alone or in combination with chemotherapy. Anti-MSLN ADCs only resulted in very lim ited efficacy and did not result in breakthrough therapies in cancers where MSLN has been shown to be expressed. The main issue are presumably the side effects elicited by the toxin that result in low treatment dosage and can also lead to a discontinuation of treatment. Another problem with this approach is that each and every tumor cell has to be targeted by the anti-MSLN mAb or ADC in order to efficiently eradicate the cancer. This in turn implies, that only tumor cells expressing MSLN at their surface will be killed by anti-MSLN mAbs or ADCs and that tumor cells not expressing or downregulating MSLN can escape the treatment. [0009] While a purely anti-MSLN mAb-based treatment approach has proved to be problematic so far, the concept of activating the patient's own immune system against cancer by removing all the breaks, i. e. immune checkpoints (IC), that prevent immune cells from attacking tumor cells, has emerged as a new breakthrough therapy in 2011 with the approval of the first immune checkpoint inhibitor (ICI) Yervoy (Ipilimumab). Since then, ICIs are revolutionizing cancer treatment by achieving long-term survival and even complete cures for patients. Unfortunately, depending on the tumor type, only a fraction of patients responds to ICIs.

[0010] As ICs are part of the physiological mechanism to resolve an immune response they are also expressed on normal, healthy cells. Thus, it is not surprising that treatment with ICIs can lead to serious side effects, including autoimmunity, and patients eventu ally have to stop the treatment. Several ICIs reached market approval, while many oth ers are being developed. Remarkably, a multitude of clinical trials are assessing their efficacy in combination with other anti-tumor therapies. Inevitably, the combination of two drugs often leads to even more side effects, less well-tolerated treatment and im paired quality of life for patients.

[0011] CD47-SIRPalpha has been identified as myeloid 1C. CD47 is known as "marker of self" and is expressed on every cell in the body. Its ligand SIRPalpha is found on phag ocytes including macrophages and transmits a "don't eat me"-signal upon interaction with CD47. CD47 plays an important role on normal cells, especially in the life of red blood cells (RBCs). Fresh RBCs express high levels of CD47 on their surface, and the levels decline with the age of RBCs, so that older, less functional RBCs having low CD47 levels are recognized by macrophages, phagocytosed and thus eliminated.

[0012] The CD47-SIRPalpha signalling pathway represents a very important innate 1C, as it involves macrophages, which act as the first line of defence against pathogens. Fur thermore, by phagocytosing tumor cells, macrophages present tumor antigens on their surface which in turn induce a long-lasting T-cell response (Liu et al, 2015, Nat Med; 21(10):1209), enabling the long-term survival of patients. Importantly, this T-cell re sponse is not linked or restricted to a specific tumor antigen or CD47 itself.

[0013] Notably, most tumor cells exploit CD47 to escape from the immune system by upregulating CD47 on their surface and thereby prevent the macrophage attack. Anti- CD47 mAbs are under development for several cancer indications but have been found to elicit side effects due to the binding of antibodies to normal cells. Even if the toxicity could be managed, the wide expression of CD47 on a multitude of normal cells leads to an absorption of the majority of therapeutic antibodies by healthy, non-tumor cells (an tigen sink) which drastically reduces the amount of antibodies that can in fact bind to tumor cells. This, of course, leads to a significantly decreased efficacy.

[0014] Thus, in view of the above described problems there is a clear need for the development of novel and effective anti-cancer therapies for several cancer indications with high unmet medical need not only targeting MSLN but also inducing a MSLN-inde- pendent anti-tumor response in addition in orderto improve treatment options and sur- vival of many cancer patients. In addition, there is a need to make anti-CD47 therapies more tumor-specific.

OBJECTS AND SUMMARY OF THE INVENTION

[0015] The present invention addresses these needs and provides improved MSLN- targeting agents which combine specific MSLN-tumor-targeting with local immune checkpoint blockade in the form of a protein construct comprising (i) an anti-mesothelin (MSLN) IgGl antibody or an antigen-binding fragment thereof and (ii) a polypeptide ca pable of binding to CD47 present on the surface of a tumor cell.

[0016] The present inventors have surprisingly found that the above-mentioned pro tein construct is not only capable of specifically targeting MSLN expressing cells but also of specifically blocking the CD47-SIRPalpha interaction on cells expressing both, MSLN and CD47. This is realized by binding the tumor target MSLN with high affinity while the polypeptide binding to CD47 present on the surface of a tumor cell has a low affinity and blocks the CD47-SIRPalpha interaction only on MSLN and CD47 double positive cells. With this strategy only MSLN and CD47 double positive tumor cells are efficiently elimi- nated whereas cells expressing only CD47, such as RBCs are spared. As the protein con structs bind only to tumor cells, the antigen sink on normal cells is avoided and a high tumor specificity is achieved. This tumor-restricted CD47 blockade thus reduces CD47- related toxicity on healthy cells (e. g. anemia) while increasing efficacy of tumor cell elimination compared to anti-MSLN mAbs or ADCs that lack the 1C inhibition.

[0017] MSLN is used in this context as a target to bring immune effector cells in close proximity to MSLN-expressing tumor cells. Advantageously, due to its very limited ex- pression and function in normal tissues, MSLN is generally an attractive target for anti tumor therapies, and the protein construct of the present invention can be used partic ularly in indications with high unmet medical need and very poor prognosis such as ovar ian cancer, triple negative breast cancer or pancreatic cancer as well as mesothelioma.

[0018] A further advantage of the protein construct of the present invention is that by sparing CD47 expressing healthy cells from being targeted, a molecule with a functional effector domain such as an IgGl can be used to increase activity of immune effector cells. A further advantageous aspect of the invention is that in certain embodiments an engineered version of the constant region of the IgGl (Fc enhanced) is efficiently used to increase the effector function and thus to enhance the anti-tumor activity. [0019] Another advantageous property of said protein construct is that different im mune effector cells from both the innate and the adaptive immune system are activated, either directly by the protein construct or indirectly by phagocytes presenting tumor antigens upon phagocytosis of tumor cells. For this reason, said protein construct com bines three different modes of action in one molecule (i) The protein construct leads to a direct and specific killing of MSLN-expressing cells by NK cells through the IgGl format, a mechanism called antibody-dependent cellular cytotoxicity (ADCC). (ii) The CD47 bind ing domain of the protein construct blocks the "don't eat me" signal on MSLN and CD47 double positive cells and together with the IgGl promotes the elimination of tumor cells by phagocytosis through macrophages (iii) Last but not least, through the phagocytosis of the tumor cells, tumor antigens are presented by macrophages and other phagocytes to induce a long-lasting MSLN-independent anti-tumor T-cell response. Thus, one unique property of the protein construct of the present invention is that not every single tumor cell has to be directly hit by the protein construct to achieve complete tumor regression. Ultimately, the T cells of the patient will eliminate the tumor cells based on several tumor antigens leading to a persisting anti-tumor immune response and poten tially long-term survival and cures of patients.

[0020] The protein construct of the present invention thus combines the advantages of specific MSLN-tumor-targeting with local inhibition of the CD47 immune checkpoint in one molecule and at the same time reduces the CD47-related side effects.

[0021] In a preferred embodiment, the protein construct efficiently binds MSLN ex pressing tumor cells and blocks the interaction of CD47 with signal regulatory protein alpha (SIRPalpha). [0022] In a further preferred embodiment, the blocking of the interaction of CD47 with SIRPalpha is provided by the binding of a CD47-binding polypeptide to CD47.

[002S] In yet another preferred embodiment said binding to CD47 is enhanced and/or reinforced by the protein construct's concomitant binding of the anti MSLN IgGl anti body or antigen-binding fragment thereof to MSLN on MSLN and CD47 double positive cells.

[0024] In a particularly preferred embodiment said enhancement and/or reinforce ment is caused by an increased avidity of the protein construct caused by multivalent binding to different antigens.

[0025] In another preferred embodiment said anti MSLN IgGl antibody or antigen- binding fragment thereof induces the recruitment of immune effector cells to MSLN and CD47 double positive cells and the activation of immune effector cells, thereby provok ing the elimination of MSLN and CD47 double positive cells.

[0026] In a further preferred embodiment of the protein construct of the present in vention said anti MSLN IgGl antibody or antigen-binding fragment thereof comprises: a variable heavy chain complementarity determining region 1 (CDRH1) sequence selected from the amino acid sequences of SEQ ID NOs: 28, 86, 87, 88 and 122; a variable heavy chain complementarity determining region 2 (CDRH2) sequence selected from the amino acid sequences of SEQ ID NOs: 29, 30, 31, 89, 90 and 91; a variable heavy chain complementarity determining region 3 (CDRH3) sequence selected from the amino acid sequences of SEQ ID NOs: 32, 33, 92, 93, and 94; a variable light chain complementarity determining region 1 (CDRL1) sequence selected from the amino acid sequences of SEQ ID NOs: 34, 35, 95, 96 and 97; a variable light chain complementarity determining region 2 (CDRL2) sequence selected from the amino acid sequences of SEQ ID NOs: 36, 37, 38, 39, 98, 99 and 121; and a variable light chain complementarity determining region 3 (CDRL3) sequence selected from the amino acid sequences of SEQ ID NOs: 40, 41, 100, 101 and 102. [0027] It is particularly preferred that said anti MSLN IgGl antibody or antigen-binding fragment thereof comprises (i) a variable heavy chain complementarity determining re gion 1 (CDRH1) sequence of SEQ ID NO: 28 and a variable heavy chain complementarity determining region 2 (CDRH2) sequence of SEQ ID NO: 31 and a variable heavy chain complementarity determining region 3 (CDRH3) sequence of SEQ ID NO: 32; and a vari- able light chain complementarity determining region 1 (CDRL1) sequence of SEQ ID NO: 34 and a variable light chain complementarity determining region 2 (CDRL2) sequence of SEQ ID NO: 38 and a variable light chain complementarity determining region 3 (CDRL3) sequence of SEQ ID NO: 40; or (ii) a variable heavy chain complementarity de termining region 1 (CDRH1) sequence of SEQ ID NO: 122 and a variable heavy chain com- plementarity determining region 2 (CDRH2) sequence of SEQ ID NO: 90 and a variable heavy chain complementarity determining region 3 (CDRH3) sequence of SEQ ID NO: 93; and a variable light chain complementarity determining region 1 (CDRL1) sequence of SEQ ID NO: 96 and a variable light chain complementarity determining region 2 (CDRL2) sequence of SEQ ID NO: 99 and a variable light chain complementarity determining re- gion 3 (CDRL3) sequence of SEQ ID NO: 101 or (iii) a variable heavy chain complementa rity determining region 1 (CDRH1) sequence of SEQ ID NO: 87 and a variable heavy chain complementarity determining region 2 (CDRH2) sequence of SEQ ID NO: 90 and a varia ble heavy chain complementarity determining region 3 (CDRH3) sequence of SEQ ID NO: 93; and a variable light chain complementarity determining region 1 (CDRL1) sequence of SEQ ID NO: 96 and a variable light chain complementarity determining region 2 (CDRL2) sequence of SEQ ID NO: 99 and a variable light chain complementarity deter mining region S (CDRLS) sequence of SEQ ID NO: 101.

[0028] In a further preferred embodiment, said anti MSLN IgGl antibody or antigen- binding fragment thereof comprises a variable light chain amino acid sequence selected from the amino acid sequences of SEQ ID NO: 3 to 11 or 126 and a variable heavy chain selected from the amino acid sequences of SEQ ID NO: 14 to 20 or 116 to 120 or 123 to 125.

[0029] In yet another preferred embodiment said anti MSLN IgGl antibody or antigen- binding fragment thereof comprises: (i) a variable light chain amino acid sequence of SEQ ID NO: 3 and a variable heavy chain amino acid sequence of SEQ ID NO: 14, 15, 16, or 17; or (ii) a variable light chain amino acid sequence of SEQ ID NO: 4 and a variable heavy chain amino acid sequence of SEQ ID NO: 16; or (iii) a variable light chain amino acid sequence of SEQ ID NO: 5 and a variable heavy chain amino acid sequence of SEQ ID NO: 16; or (iv) a variable light chain amino acid sequence of SEQ ID NO: 6 and a vari able heavy chain amino acid sequence of SEQ ID NO: 14, 15, 16 or 17; or (v) a variable light chain amino acid sequence of SEQ ID NO: 7 and a variable heavy chain amino acid sequence of SEQ ID NO: 16; or (vi) a variable light chain amino acid sequence of SEQ ID NO: 8 and a variable heavy chain amino acid sequence of SEQ ID NO: 16; or (vii) a variable light chain amino acid sequence of SEQ ID NO: 9 and a variable heavy chain amino acid sequence of SEQ ID NO: 16, 116, 117, 118, 119 or 120; or (viii) a variable light chain amino acid sequence of SEQ ID NO: 10 and a variable heavy chain amino acid sequence of SEQ ID NO: 18, 19 or 20; or (ix) a variable light chain amino acid sequence of SEQ ID NO: 11 and a variable heavy chain amino acid sequence of SEQ ID NO: 18, 19 or 20; or (x) a variable light chain amino acid sequence of SEQ ID NO: 126 and a variable heavy chain amino acid sequence of SEQ ID NO: 123, 124 or 125

[0030] In yet another preferred embodiment said anti MSLN IgGl antibody or antigen binding fragment thereof comprises (i) a variable light chain amino acid sequence of SEQ ID NO: 9 and a variable heavy chain amino acid sequence of SEQ ID NO: 117 or (ii) a variable light chain amino acid sequence of SEQ ID NO: 126 and a variable heavy chain amino acid sequence of SEQ ID NO: 125 or (iii) a variable light chain amino acid sequence of SEQ ID NO: 126 and a variable heavy chain amino acid sequence of SEQ ID NO: 12S [00S1] In a further preferred embodiment of the present invention, said anti MSLN

IgGl antibody or antigen-binding fragment thereof comprises a constant light chain CL domain and constant heavy chain CHI, CH2 and CHS domains, preferably a human con stant light chain CL domain and a human constant heavy chain CHI, CH2, CH3 domain; and a hinge domain, preferably a human hinge region. [0032] It is particularly preferred that said light chain CL domain comprises an amino acid sequence which is at least 95%, 96%, 97%, 98% or 99% identical to the amino acid sequence of SEQ ID NO: 27.

[0033] It is further particularly preferred that said heavy chain CHI domain comprises an amino acid sequence which is at least 95%, 96%, 97%, 98% or 99% identical to the amino acid sequence of SEQ ID NO: 22, that said CH2 domain comprises an amino acid sequence which is at least 95%, 96%, 97%, 98% or 99% identical to the amino acid se quence of SEQ ID NO: 23 or SEQ ID NO:128 and that said CH3 domain comprises an amino acid sequence which is at least 95%, 96%, 97%, 98% or 99% identical to the amino acid sequence of SEQ ID NO: 24. [0034] In a further preferred embodiment said heavy chain CH2 domain and said heavy chain CH3 domain form an Fc domain. It is particularly preferred that said Fc do main comprises an amino acid sequence which is at least 95%, 96%, 97%, 98% or 99% identical to the amino acid sequence of SEQ ID NO: 42 or SEQ ID NO: 127.

[0035] In a furtherembodiment the hinge domain comprises the amino acid sequence of SEQ ID NO: 25. [0036] In a further preferred embodiment the Fab domain of the protein construct comprises: (i) a variable light chain (VL) amino acid sequence of SEQ ID NO: 126, a vari able heavy (VH) chain amino acid sequence of SEQ ID NO: 125, a constant light (CL) chain amino acid sequence of SEQ ID NO: 27 and a constant heavy CHI chain amino acid se- quence of SEQ ID NO: 22; (ii) a variable light chain (VL) amino acid sequence of SEQ ID NO: 126, a variable heavy (VH) chain amino acid sequence of SEQ ID NO: 123, a constant light (CL) chain amino acid sequence of SEQ ID NO: 27 and a constant heavy CHI chain amino acid sequence of SEQ ID NO: 22; (iii) a variable light chain (VL) amino acid se quence of SEQ ID NO: 9, a variable heavy (VH) chain amino acid sequence of SEQ ID NO: 117, a constant light (CL) chain amino acid sequence of SEQ ID NO: 27 and a constant heavy CHI chain amino acid sequence of SEQ ID NO: 22 ; (iv) a variable light chain (VL) amino acid sequence of SEQ ID NO: 9, a variable heavy (VH) chain amino acid sequence of SEQ ID NO: 16, a constant light (CL) chain amino acid sequence of SEQ ID NO: 27 and a constant heavy CHI chain amino acid sequence of SEQ ID NO: 22; (v) a variable light chain (VL) amino acid sequence of SEQ ID NO: 7 , a variable heavy (VH) chain amino acid sequence of SEQ ID NO: 16, a constant light (CL) chain amino acid sequence of SEQ ID NO: 27 and a constant heavy CHI chain amino acid sequence of SEQ ID NO: 22; (vi) a variable light chain (VL) amino acid sequence of SEQ ID NO: 4, a variable heavy (VH) chain amino acid sequence of SEQ ID NO: 16, a constant light (CL) chain amino acid se- quence of SEQ ID NO: 27 and a constant heavy CHI chain amino acid sequence of SEQ ID NO: 22; (vii) a variable light chain (VL) amino acid sequence of SEQ ID NO: 6, a variable heavy (VH) chain amino acid sequence of SEQ ID NO: 15, a constant light (CL) chain amino acid sequence of SEQ ID NO: 27 and a constant heavy CHI chain amino acid se quence of SEQ ID NO: 22; or (viii) a variable light chain (VL) amino acid sequence of SEQ ID NO: 9, a variable heavy (VH) chain amino acid sequence of SEQ ID NO: 120, a constant light (CL) chain amino acid sequence of SEQ ID NO: 27 and a constant heavy CHI chain amino acid sequence of SEQ ID NO: 22. [00B7] In yet another preferred embodiment said polypeptide capable of binding to CD47 present on the surface of a tumor cell comprises at least one immunoglobulin-like domain of SIRPalpha or comprises an anti-CD47 single chain fragment variable (scFv).

[0038] It is particularly preferred that said polypeptide comprises one or two copies of the immunoglobulin-like domain of SIRPalpha.

[0039] It is further particularly preferred that said immunoglobulin-like domain of SIRPalpha comprises an amino acid sequence which is at least 95%, 96%, 97%, 98% or 99% identical to the amino acid sequence of SEQ ID NO: 21.

[0040] In yet another preferred embodiment said immunoglobulin-like domain of SIRPalpha has an affinity for CD47 measured by surface plasmon resonance (SPR) in the range of 100 nM to 2 mM, preferably in a range of 300 nM to 800 nM.

[0041] It is particularly preferred that said anti-CD47 scFv has an affinity for CD47 measured by surface plasmon resonance (SPR), in the range of 100 nM to 2 pM, prefer ably in the range of 300 nM to 800 nM. [0042] In another preferred embodiment the anti MSLN IgGl antibody or antigen binding fragment thereof has an affinity to its target MSLN which in comparison to the affinity of the CD47-binding polypeptide to its target CD47 is higher by a factor of at least 10, preferably at least 25, more preferably at least 35.

[0043] In further preferred embodiments said anti-CD47 scFv comprises: a variable heavy chain complementarity determining region 1 (CDRH1) sequence selected from the amino acid sequences of SEQ ID NOs: 49 to 51; a variable heavy chain complementarity determining region 2 (CDRH2) sequence selected from the amino acid sequences of SEQ ID NOs: 52 to 62; a variable heavy chain complementarity determining region 3 (CDRH3) sequence of SEQ ID NOs: 63; a variable light chain complementarity determining region 1 (CDRL1) sequence selected from the amino acid sequences of SEQ ID NOs: 64 to 68; a variable light chain complementarity determining region 2 (CDRL2) sequence selected from the amino acid sequences of SEQ ID NOs: 69 and 70; and a variable light chain complementarity determining region 3 (CDRL3) sequence selected from the amino acid sequences of SEQ ID NOs: 71 to 74.

[0044] In a particularly preferred embodiment said anti-CD47 scFv comprises: a varia ble heavy chain complementarity determining region 1 (CDRH1) sequence of SEQ ID NOs: 49; a variable heavy chain complementarity determining region 2 (CDRH2) se quence of SEQ ID NOs: 60; a variable heavy chain complementarity determining region 3 (CDRH3) sequence of SEQ ID NOs: 63; a variable light chain complementarity deter mining region 1 (CDRL1) sequence of SEQ ID NOs: 64; a variable light chain complemen tarity determining region 2 (CDRL2) sequence of SEQ ID NOs: 69; and a variable light chain complementarity determining region 3 (CDRL3) sequence of SEQ ID NOs: 72.

[0045] In a further preferred embodiment said anti-CD47 scFv comprises a variable light chain amino acid sequence selected from the amino acid sequences of SEQ ID NO: 75 to 79, 114 and 115 and a variable heavy chain amino acid sequence selected from the amino acid sequences of SEQ ID NO: 80 to 85. [0046] It is particularly preferred that said anti-CD47 scFv comprises a variable light chain amino acid sequence of SEQ ID NO: 78 and a variable heavy chain amino acid se quence of SEQ ID NO: 83.

[0047] In another group of embodiments, the anti MSLN IgGl antibody or antigen binding fragment thereof and the polypeptide capable of binding to CD47 is connected by a polypeptide linker.

[0048] Preferably, said polypeptide linker spatially separates MSLN- and CD47-binding and thereby allows for a simultaneous binding of the protein construct to MSLN and CD47.

[0049] It is further preferred that said polypeptide linker comprises or consists of 4 to 40 amino acids. [0050] In a further preferred embodiment said polypeptide linker comprises, essen tially consist of, or consists of the amino acids glycine, alanine, proline, lysine, threonine, aspartic acid, asparagine and/or serine.

[0051] In a particularly preferred embodiment said polypeptide linker comprises, es- sentially consists of, or consists of one or more of the amino acid sequence groups of SEQID NOs: 43 to 48.

[0052] In a further set of embodiments said polypeptide linker is fused to either the N-terminus of the variable light (VL) or the N-terminus of the variable heavy (VH) chain or the C-terminus of the constant light (CL) or the C-terminus of the constant heavy (CH3) chain domain of the anti MSLN IgGl antibody or antigen-binding fragment thereof. Preferably, the polypeptide linker is fused to the N-terminus of the variable light (VL) chain of the anti MSLN IgGl antibody or antigen-binding fragment thereof.

[0053] In a further aspect the present invention relates toa nucleicacid molecule com prising a polynucleotide encoding the protein construct as defined herein above or a fragment of said protein construct.

[0054] In another aspect the present invention relates to a vector comprising the nu cleic acid molecule as defined herein.

[0055] In yet another aspect the present invention relates to a host cell comprising said nucleic acid molecule or said vector. [0056] In a further aspect the present invention relates to a host cell that expresses the protein construct according to the invention, or a fragment of said protein construct.

[0057] In a further aspect the present invention relates to a method of producing the protein construct according to the invention comprising the cultivation of a host cell as defined herein, thereby expressing said protein construct. [0058] In yet another aspect the present invention relates to a product produced by the method of producing the protein construct. [0059] Further envisaged, in an additional aspect, is a pharmaceutical composition comprising the protein construct of the invention, or the product as defined herein and a pharmaceutically acceptable carrier.

[0060] In a particularly preferred embodiment, the protein construct as defined herein, the product as defined herein or the pharmaceutical composition as defined herein is for use in the treatment of cancer.

[0061] In a further aspect the present invention relates to a method for the treatment of cancer, wherein said method comprises administering to a patient in need thereof a protein construct according to the invention, the product of the invention, or a pharma- ceutical composition as defined herein.

[0062] It is particularly preferred that said cancer is ovarian cancer, ascites, mesothe lioma, triple negative breast cancer, pancreatic cancer, pancreatic adenocarcinoma, non-small cell lung cancer (NSCLC), endometrial cancer or biliary extrahepatic cancer.

BRIEF DESCRIPTION OF THE FIGURES [0063] Figure 1 shows a schematic drawing of embodiments of the invention including

A) an IgG antibody, possibly an anti-MSLN antibody (anti-MSLN IgG protein construct),

B) a protein construct comprising an IgG and two polypeptides capable of binding to CD47 (depicted on left and right hand side of the molecule) wherein the polypeptide is a scFv domain (CD47 scFv-anti-MSLN protein construct), and C) a protein construct com- prising an IgG and two polypeptides capable of binding to CD47 (depicted on left and right hand side of the molecule) wherein the polypeptide is a single SIRPalpha domain (SIRPalpha-anti-MSLN protein construct). As the IgG antibody contains two heavy chains and two light chains, two SIRPalpha domains or two anti-CD47 scFv might be present in an IgG. VL: variable light, VH: variable heavy, CHl-3: constant heavy 1-3, CL: constant light, CDRs: complementary determining regions, flexible linkers are indicated by black lines. [0064] Figure 2 shows the structural orientation and domain/fragment arrangement of the heavy and light chains of IgG antibodies and protein constructs of embodiments of the present invention. SIRPalpha domains are depicted as single domains.

[0065] Figure 3 shows purified protein constructs including anti-MSLN IgG protein constructs, CD47 scFv-anti-MSLN protein constructs and SIRPalpha-anti-MSLN protein constructs analysed by sodium dodecylsulfate-polyacrylamind gel electrophoresis (SDS- PAGE).

[0066] Figure 4 shows different purified humanized anti-MSLN IgG protein constructs, CD47 scFv-anti-MSLN protein constructs and SIRPalpha-anti-MSLN protein constructs analysed by analytical size exclusion chromatography (aSEC).

[0067] Figure 5 shows examples of binding measurements by surface plasmon reso nance (SPR) using a Biacore X100 (Cytiva) instrument. The experiment measures the binding of protein constructs comprising humanized anti-MSLN IgG protein constructs MSL-200 (SEQ ID NOs: 6 and 15) and MSL-207 (SEQ ID NOs: 9 and 16) to recombinant C- terminally His-tagged human MSLN. KD K_on and K_0ff values were calculated for each pro tein construct and are indicated.

[0068] Figure 6 shows examples of binding measurements by surface plasmon reso nance (SPR) using a Biacore X100 (Cytiva) instrument. The experiment measures the binding of protein constructs comprising humanized SIRPalpha-anti-MSLN protein con- structs MSL-215 (SEQ ID NOs: 7, 16 and 21) and MSL-217 (SEQ ID NOs: 9, 16 and 21) to recombinant C-terminally His-tagged human MSLN. KD, K_on and K_0ff values were calcu lated for each protein construct and are indicated.

[0069] Figure 7 shows examples of binding measurements by surface plasmon reso nance (SPR) using a Biacore X100 (Cytiva) instrument. The experiment measures the binding of protein constructs comprising humanized CD47 scFv-anti-MSLN protein con- structs MSL-247 (SEQ ID NOs: 111 and 83) and MSL-248 (SEQ ID NOs: 78 and 83) to re combinant C-terminally His-tagged human MSLN. KD, K_on and K_0ff values were calculated for each protein construct and are indicated.

[0070] Figure 8 shows histograms of binding of humanized SIRPalpha-anti-MSLN pro- tein constructs to MSLN and CD47 double positive OVCAR-3 and Suit-2-MSLN cells meas ured by flow cytometry.

[0071] Figure 9 shows concentration dependent binding of humanized anti-MSLN IgG protein constructs (MSL-200, MSL-205, MSL-206, MSL-207) and the SIRPalpha-anti- MSLN protein construct MSL-217 to MSLN and CD47double positive OVCAR-3 cells measured by flow cytometry. KD values are indicated in the table below. Error bars in dicate standard error of the mean.

[0072] Figure 10 depicts blocking of the CD47-SIRPalpha interaction by humanized CD47 scFv-anti-MSLN protein constructs (MSL-248, MSL-274) and SIRPalpha-anti-MSLN protein constructs (MSL-217) by SPR. SIRPalpha is coated on the Biacore chip and human CD47 extracellular domain is used as analyte and was previously incubated with buffer or with humanized CD47 scFv-anti-MSLN protein constructs and SIRPalpha-anti-MSLN protein constructs. The line labeled with "buffer" shows thus the unblocked binding of CD47 to SIRPalpha whereas if the CD47 analyte is pre-incubated with any of the protein constructs of the present invention containing a polypeptide capable of binding to CD47, the SIRPalpha-CD47 interaction is blocked. The anti-MSLN IgG protein construct MSL- 207 does not block the SIRPalpha-CD47 interaction.

[0073] Figure 11 depicts blocking of the CD47-SIRPalpha interaction by the humanized SIRPalpha-anti-MSLN protein construct MSL-217 by flow cytometry using MSLN and CD47 double positive OVCAR-3 cells. The humanized anti-MSLN IgG protein construct MSL-207 was used as negative control.

[0074] Figure 12 shows an antibody-dependent cellular cytotoxicity (ADCC) assay us ing humanized anti-MSLN IgG protein constructs (MSL-200, MSL-207) and SIRPalpha- anti-MSLN protein constructs (MSL-210, MSL-217) using MSLN and CD47 double positive Suit-2 MSLN cells as target cells. An anti-CD47 antibody was used as control. Error bars indicate standard error of the mean.

[0075] Figure 13 shows an antibody-dependent cellular cytotoxicity (ADCC) assay us- ing humanized anti-MSLN IgG protein constructs (MSL-200, MSL-207) and SIRPalpha- anti-MSLN protein constructs (MSL-210, MSL-217) using MSLN and CD47 double positive OVCAR-3 cells as target cells. An anti-CD47 antibody was used as control. Error bars in dicate standard error of the mean.

[0076] Figure 14 shows an antibody-dependent cellular phagocytosis (ADCP) assay where phagocytosis is induced by humanized anti-MSLN IgG protein constructs (MSL- 200, MSL-207), the CD47 scFv-anti-MSLN protein construct MSL-247and the SIRPalpha- anti-MSLN protein construct MSL-217 using MSLN and CD47 double positive OVCAR-3 cells as target cells. Data represent mean +/- S.D. of 4 independent experiments.

[0077] Figure 15 shows the exemplary binding measurement of humanized anti-CD47 scFvs fused to a full IgGl antibody targeting a tumor antigen to recombinant C-termi- nally His-tagged human CD47 by surface plasmon resonance (SPR) using a Biacore X100 (Cytiva). K_on, K_0ff and KD values are indicated.

[0078] Figures 16 A and B show the exemplary binding measurement of humanized CD47 scFv-anti-MSLN protein constructs (MSL-248, MSL-253, MSL-741, MSL-742, MSL- 745) to recombinant C-terminally His-tagged human CD47 by surface plasmon reso nance (SPR) using a Biacore X100 (Cytiva). K_on, K_0ff and KD values are indicated.

[0079] Figure 17 shows the exemplary binding measurement of humanized SIRPalpha- anti-MSLN protein constructs (MSL-217, MSL-711, MSL-712, MSL-715) to recombinant C-terminally His-tagged human CD47 by SPR using a Biacore X100 (Cytiva). K_on, K_0ff and KD values are indicated. [0080] Figure 18 shows purified protein constructs including anti-MSLN IgG protein constructs (MSL-701, MSL-702, MSL-705, MSL-207), CD47 scFv-anti-MSLN protein con structs (MSL-741, MSL-742, MSL-745, MSL-25B) and SIRPalpha-anti-MSLN protein con structs (MSL-711, MSL-712, MSL-715, MSL-211) analysed by sodium dodecylsulfate-pol- yacrylamind gel electrophoresis (SDS-PAGE).

[0081] Figure 19 shows examples of binding measurements by SPR using a Biacore X100 (Cytiva) instrument. The experiment measures the binding of protein constructs comprising anti-MSLN IgG protein constructs MSL-701 (SEQ ID NOs: 126 and 123), MSL- 702 (SEQ ID NOs: 126 and 124) and MSL-705 (SEQ ID NOs: 126 and 125) to recombinant C-terminally His-tagged human MSLN. K_on, K_0ff and KD values were calculated for each protein construct and are indicated.

[0082] Figure 20 shows examples of binding measurements by SPR using a Biacore X100 (Cytiva) instrument. The experiment measures the binding of protein constructs comprising SIRPalpha-anti-MSLN protein constructs MSL-711 (SEQ ID NOs: 126, 123 and 21), MSL-712 (SEQ ID NOs: 126, 124 and 21) and MSL-715 (SEQ ID NOs: 126, 125 and 21) to recombinant C-terminally His-tagged human MSLN. K_on, K_0ff and KD values were cal culated for each protein construct and are indicated.

[0083] Figure 21 shows examples of binding measurements by SPR using a Biacore X100 (Cytiva) instrument. The experiment measures the binding of protein constructs comprising CD47 scFv-anti-MSLN protein constructs MSL-741 (SEQ ID NOs: 126, 123 and 78 and 83), MSL-742 (SEQ ID NOs: 126, 124 and 78 and 83) and MSL-745 (SEQ ID NOs: 126, 125 and 78 and 83) to recombinant C-terminally His-tagged human MSLN. K_on, K_0ff and KD values were calculated for each protein construct and are indicated.

[0084] Figure 22 shows binding of of humanized anti-MSLN IgG protein constructs (MSL-701, MSL-702, MSL-705, MSL-207), CD47 scFv-anti-MSLN protein constructs (MSL-

741, MSL-742, MSL-745, MSL-253) and SIRPalpha-anti-MSLN protein constructs (MSL- 711, MSL-712, MSL-715, MSL-211) to MSLN and CD47 double positive OVCAR-3 and Suit- 2 MSLN cells measured by flow cytometry. [0085] Figure 23 shows concentration dependent binding of humanized anti-MSLN IgG protein construct MSL-705, CD47 scFv-anti-MSLN protein construct MSL-745 and SIRPalpha-anti-MSLN protein construct MSL-715 to MSLN and CD47 double positive OVCAR-3 cells measured by flow cytometry. [0086] Figure 24 depicts blocking of the CD47-SIRPalpha interaction byCD47 scFv-anti-

MSLN protein constructs and SIRPalpha-anti-MSLN protein constructs by SPR. SIRPalpha is coated on the Biacore chip and human CD47 extracellular domain is used as analyte and was previously incubated with buffer or with humanized CD47 scFv-anti-MSLN pro tein constructs (MSL-741, MSL-742, MSL-745) and SIRPalpha-anti-MSLN protein con- structs (MSL-711, MSL-712, MSL-715). The line labeled with "buffer" shows the un blocked binding of CD47 to SIRPalpha whereas, ifthe CD47 analyte is pre-incubated with any of the protein constructs of the present invention, containing a polypeptide capable of binding CD47, the SIRPalpha-CD47 interaction is blocked. The anti-MSLN IgG protein construct MSL-701, MSL-702 and MSL-705 donot block the CD47-SIRPalpha interaction. [0087] Figure 25 depicts blocking of the CD47-SIRPalpha interaction by humanized

CD47 scFv-anti-MSLN protein constructs (MSL-741, MSL-742, MSL-745, MSL-753) and SIRPalpha-anti-MSLN protein constructs (MSL-711, MSL-712, MSL-715, MSL-217) by flow cytometry using MSLN and CD47 double positive OVCAR-3 cells. The anti-CD47 mAb serves as a positive control and the anti-MSLN-lgGl protein constructs (MSL-701, MSL- 702, MSL-705, MSL-207) do not block the CD47-SIRPalpha interaction as expected.

[0088] Figures 26 A and B show an antibody-dependent cellular cytotoxicity (ADCC) assay using humanized anti-MSLN IgG protein constructs (MSL-701, MSL-702, MSL-705), SIRPalpha-anti-MSLN protein constructs (MSL-711, MSL-712, MSL-715) and CD47 scFv- anti-MSLN protein constructs (MSL-741, MSL-742, MSL-745) using MSLN and CD47 dou- ble positive OVCAR-3 cells as target cells.

[0089] Figure 27 shows an antibody-dependent cellular cytotoxicity (ADCC) assay us ing a humanized, Fc enhanced anti-MSLN IgG protein construct (MSL-709) and a human ized Fc enhanced SIRPalpha-anti-MSLN protein construct (MSL-719) in comparison to a SIRPalpha-anti-MSLN IgGl protein construct (MSL-715) with an unmodified Fc, using MSLN and CD47 double positive OVCAR-3 cells as target cells.

[0090] Figure 28 shows an antibody-dependent cellular cytotoxicity (ADCC) assay us ing a humanized anti-MSLN IgG protein construct (MSL-207), a SIRPalpha-anti-MSLN protein constructs (MSL-217) and a CD47 scFv-anti-MSLN protein construct (MSL-253) using MSLN and CD47 double positive Suit-2 MSLN cells as target cells. An anti-CD47 antibody was used as control.

[0091] Figure 29 shows an antibody-dependent cellular phagocytosis (ADCP) assay where phagocytosis is induced by humanized anti-MSLN IgG protein constructs (MSL- 701, MSL-702, MSL-705), CD47 scFv-anti-MSLN protein constructs (MSL-741, MSL-742,

MSL-745) and SIRPalpha-anti-MSLN protein constructs (MSL-711, MSL-712, MSL-715) using MSLN and CD47 double positive OVCAR-3 cells as target cells.

[0092] Figure 30 shows the comparison of an anti-MSLN IgG protein construct (MSL- 705), a CD47 scFv-anti-MSLN protein construct (MSL-745), a SIRPalpha-anti-MSLN pro- tein construct (MSL-715) and an anti-CD47 mAb in an antibody-dependent cellular phag ocytosis (ADCP) assay using MSLN and CD47 double positive OVCAR-3 cells as target cells.

[0093] Figure 31 shows the binding to human red blood cells (RBCs) of different anti- MSLN IgG protein constructs (MSL-701, MSL-702, MSL-705, MSL-207), CD47 scFv-anti- MSLN protein constructs (MSL-741, MSL-742, MSL-745, MSL-253) and SIRPalpha-anti- MSLN protein constructs (MSL-711, MSL-712, MSL-715, MSL-217) in comparison to an anti-CD47 mAb measured by flow cytometry.

[0094] Figure 32 illustrates the ability of inducing platelet (PLT) aggregation in vitro by a PLT aggregation assay. PLT aggregation after incubation with a CD47 scFv-anti-MSLN protein construct (MSL-745) is exemplary shown as percentage of aggregation, meas ured as absorbance at 595 nm on a TECAN plate reader. Anti-CD47 lgG4 serves as nega tive and anti-CD47-lgGl as positive control. DETAILED DESCRIPTION OF THE INVENTION

[0095] Although the present invention will be described with respect to particular em bodiments, this description is not to be construed in a limiting sense.

[0096] Before describing in detail exemplary embodiments of the present invention, definitions important for understanding the present invention are given.

[0097] As used in this specification and in the appended claims, the singular forms of "a" and "an" also include the respective plurals unless the context clearly dictates oth erwise.

[0098] In the context of the present invention, the terms "about" and "approximately" denote an interval of accuracy that a person skilled in the art will understand to still ensure the technical effect of the feature in question. The term typically indicates a de viation from the indicated numerical value of ±20 %, preferably ±15 %, more preferably ±10 %, and even more preferably ±5 %.

[0099] It is to be understood that the term "comprising" is not limiting. For the pur poses of the present invention the term "consisting of" or "essentially consisting of" is considered to be a preferred embodiment of the term "comprising of". If hereinafter a group is defined to comprise at least a certain number of embodiments, this is meant to also encompass a group which preferably consists of these embodiments only.

[0100] Furthermore, the terms "(i)", "(ii)", "(iii)" or "(a)", "(b)", "(c)", "(d)", or "first", "second", "third" etc. and the like in the description or in the claims, are used for distin guishing between similar elements and not necessarily for describing a sequential or chronological order. It is to be understood that the terms so used are interchangeable under appropriate circumstances and that the embodiments of the invention described herein are capable of operation in other sequences than described or illustrated herein. In case the terms relate to steps of a method or use there is no time or time interval coherence between the steps, i. e. the steps may be carried out simultaneously or there may be time intervals of seconds, minutes, hours, days, weeks etc. between such steps, unless otherwise indicated.

[0101] It is to be understood that this invention is not limited to the particular meth odology, protocols, reagents etc. described herein as these may vary. It is also to be understood that the terminology used herein is for the purpose of describing particular embodiments only and is not intended to limit the scope of the present invention that will be limited only by the appended claims. Unless defined otherwise, all technical and scientific terms used herein have the same meanings as commonly understood by one of ordinary skill in the art. [0102] As has been set out above, the present invention concerns in one aspect a pro tein construct comprising (i) an anti-mesothelin (MSLN) IgGl antibody or an antigen binding fragment thereof and (ii) a polypeptide capable of binding to CD47 present on the surface of a tumor cell.

[0103] The term "mesothelin" or "MSLN" as used herein relates to a mesothelin poly- peptide and fragments thereof which may be present on the surface of cells. Mesothelin has been described as glycophosphatidylinositol (GPI) -linked cell-surface glycoprotein, which is typically synthesized as a 71 kDa precursor protein and is then cleaved to re lease a secreted N-terminal region, called megakaryocyte potentiating factor (MPF). (Ya- maguchi et al, 1994, J Biol Chem.; 269(2):805). The 41 kDa mature MSLN remains at- tached to the membrane (Chang et al, 1996, Proc Natl Acad Sci U S A.;93(l):136).

[0104] Human mesothelin is identified by UniProt Q13421 and is also known as CAK1 antigen or Pre-pro-megakaryocyte-potentiating factor. Alternative splicing of mRNA en coded by the human MSLN gene yields four isoforms which differ in length: isoform 1 (UniProt: Q13421-1, SEQ ID NO: 103) which has been chosen as canonical isoform in UniProt and which is used as reference isoform for the numbering of the protein (and the numbering of the other isoforms), but reflects a minor form; isoform 2 (Q13421-3, SEQ ID NO: 104), which lacks amino acids corresponding to positions 409 to 416 of SEQ ID NO: 103 and which is the major form; isoform 3 (UniProt: Q13421-2, SEQ ID NO: 105), which lacks amino acids corresponding to positions 409 to 416 of SEQ ID NO: 103 and comprises sequence changes in the stretch of amino acids corresponding to positions 601 to 630 of SEQ ID NO: 103; and isoform 4 (UniProt: Q13421-4, SEQ ID NO: 106) lacks amino acids corresponding to positions 44 and 409 to 416 of SEQ ID NO: 103. Amino acids corresponding to positions lto 36 of SEQ ID NO: 103 have been identified as signal peptide, amino acids corresponding to positions 607 to 630 of SEQ ID NO: 103 have been identified as propeptide which is removed in a mature form, amino acids corresponding to positions 37 to 606 of SEQ ID NO: 103 are typically considered to constitute the ca nonical mesothelin, amino acids corresponding to positions 37 to 286 of SEQ ID NO: 103 are typically considered to constitute the megakaryocyte-potentiating factor fragment of mesothelin and amino acids corresponding to positions 296 to 606 of SEQ ID NO: 103 constitute the processed form of mesothelin, which is presented at the surface of a cell. Both, MPF and the processed form of mesothelin are typically N-glycosylated. The gly- cosylation sites are at positions 57 (in MPF), 388, 496 and 523 (in mesothelin) of SEQ ID NO 103.

[0105] The structure and function of mesothelin is described, for example, in Hassan et al., 2004, Clin Cancer Research 10, 3737.

[0106] In normal tissues MSLN is only expressed in mesothelial cells which line the pleura, peritoneum, and pericardium (Chang et al, 1992, Int J Cancer, 50(3): 373). In con trast, mesothelin has been found to be highly expressed in several types of cancers in cluding pancreatic cancer, ovarian cancer, lung adenocarcinomas, malignant mesotheli oma, gastric cancer, squamous cell carcinomas of the cervix, head and neck carcinomas, endometrial adenocarcinomas as well as triple negative breast cancer (Weidemann et al., 2021, Biomedicines, 9 (4), 397; Tozbikian et al., 2014, PLos ONE 9 (12), ell4900).

[0107] Aberrant mesothelin expression or activity has been found in several cancers in different species, which may provide the mesothelin in truncated or modified form deviating from the above defined sequences. Accordingly, the term "mesothelin" or "MSLN" refers to mesothelin from any species, preferably from mammals such as rats, mice, and primates, more preferably from humans. It mayfurtherinclude isoforms, frag ments, variants or homologues from any species. It is particularly preferred that meso- thelin is present at the surface of a cell.

[0108] The term "CD47" as used herein relates to a transmembrane polypeptide, which belongs to the immunoglobulin superfamily. CD47 partners with membrane in- tegrins and also binds the ligands thrombospondin-1 (TSP-1) and signal-regulatory pro tein alpha (SIRPalpha) and signal-regulatory protein gamma (SIRPgamma). CD47 func tions as a marker of self and transmits a "don't eat me" signal by binding to SIRPalpha expressed by myeloid cells, macrophages, dendritic cells, monocytes and neutrophils. In this context the role of CD47 is to prevent the engulfment (phagocytosis) of healthy cells by the mentioned immune cells. CD47 is also involved in a range of additional cellular processes, including apoptosis, proliferation, adhesion, migration and angiogenesis (Sick et al, 2012, Br J Pharmacol; 167(7): 1415). CD47 has been shown to be ubiquitously ex pressed in human cells and CD47 has been found to be overexpressed in many different tumor types (Willingham et a I, 2012, PNAS, 109 (17); 6662). Human CD47 (Cluster of Differentiation 47) is identified by UniProt Q08722 and is also known as integrin associ ated protein (IAP), OA3 or MER6. Alternative splicing of mRNA encoded by the human CD47 gene yields four isoforms which differ in length: isoform 1 (UniProt: Q08722-1, or OA3-323, SEQ ID NO: 107) has been chosen as the canonical isoform in UniProt and which is used as reference isoform forthe numbering of the protein (and the numbering of the other isoforms); isoform 2 (Q08722-2, OA3-293, SEQ ID NO: 108), which lacks amino acids corresponding to positions 293 to 323 of SEQ ID NO: 107; isoform 3 (Uni Prot: Q08722-3, or OA3-305, SEQ ID NO: 109), which lacks amino acids corresponding to positions 306 to 323 of SEQ ID NO: 107 and comprises sequence changes in amino acids corresponding to positions 304 to 305 of SEQ ID NO: 107; and isoform 4 (UniProt: Q08722-4, or OA3-312, SEQ ID NO: 110) which lacks amino acids corresponding to posi tions 312 to 323 of SEQ ID NO: 107.

[0109] Amino acids corresponding to positions 1 to 18 of SEQ ID NO: 107 have been identified as signal peptide, amino acids corresponding to positions 19 to 323 of SEQ ID NO: 107 are typically considered to constitute the canonical CD47. CD47 is typically N- glycosylated at one or more of positions 23, 34, 50, 73, 111 and 206 of SEQ ID NO: 107.

[0110] The structure and function of CD47 is described, for example, in Oldenborg, 2013, ISRN Hematol, Epub Jan 21. [0111] CD47 is ubiquitously expressed on normal tissue and has been found to be highly expressed in a wide range of human cancers including acute myeloid leukemia, non-Hodgkin lymphoma, ovarian tumors, breast cancer, pancreatic adenocarcinoma (Xi Q, et al., 2020, J Immunother Cancer, 8:e000253) and non-small cell lung cancer (NSCLC; Zhao et al., 2016, Sci Rep 6: 29719)) and melanoma (Chao et al., 2012, Curr Opin Immu- nol, 24(2):225-32).

[0112] Aberrant CD47 expression has been found in several cancers in different spe cies, which may provide the CD47 in modified form deviating from the above defined sequences. Accordingly, the term "CD47" refers to a CD47 polypeptide from any species, preferably from mammals such as rats, mice, and primates, more preferably from hu- mans. It may further include isoforms, fragments, variants or homologues from any spe cies. It is particularly preferred that CD47 is present at the surface of a cell.

[0113] As used herein, a "fragment", "variant" or "homologue" of a protein relates to polypeptide which comprises or consists of an amino acid sequence which has at least 70%, preferably 75%, 80%, 85%, 90%, 91%, 92%, 93%, 94%, 95%, 96%, 97%, 98%, 99% or more amino acid sequence identity to the amino acid sequence of a reference protein (e. g. the isoform of SEQ ID NO: 103, or of SEQ ID NO: 107, or of SEQ ID NO: 112). In certain embodiments fragments, variants, isoforms and homologues of a reference pro tein may be capable of performing one, more or all function(s) performed by the refer ence protein. [0114] The term "sequence identity" as used herein means that amino acids se quences (or two polynucleotides) are identical (i. e., on a residue-by-residue (or on a nucleotide-by-nucleotide) basis) over the comparison window. A percentage of se quence identity may be calculated by comparing two optimally aligned sequences over the window of comparison, determining the number of positions at which the identical elements occurs in both sequences to yield the number of matched positions, dividing the number of matched positions by the total number of positions in the comparison window (i. e., the window size), and multiplying the result by 100 to yield the percentage of sequence identity. For amino acid sequences, sequence identity may preferably be determined by using standard techniques known in the art, including the local sequence identity algorithm of Smith and Waterman, 1981, Adv. Appl. Math. 2:482, the sequence identity alignment algorithm of Needleman and Wunsch, 1970, J. Mol. Biol. 48:443, the search for similarity method of Pearson and Lipman, 1988, Proc. Nat. Acad. Sci. U.S.A. 85:2444, or computerized implementations of these algorithms such as multiple se quence alignment tools Clustal Omega (https://www.ebi.ac.uk/Tools/msa/clustalo/) , T- coffee/M-coffee (http://tcoffee.crg.cat/), BLAST (https://blast.ncbi.nlm.nih.gov), FASTA (https://www.ebi.ac.uk/Tools/sss/fasta/) or the like, preferably using the default set tings. A further envisaged example of a useful algorithm is PILEUP. PILEUP creates a mul tiple sequence alignment from a group of related sequences using progressive, pairwise alignments. It can also plot a tree showing the clustering relationships used to create the alignment. PILEUP uses a simplification of the progressive alignment method of Feng & Doolittle, 1987, J. Mol. Evol. 35:351-360. Another example of a useful algorithm is the

BLAST algorithm, described in Altschul et al., 1990, J. Mol. Biol. 215:403-410 or the WU- BLAST-2 program. WU-BLAST-2 uses several search parameters, most of which are set to the default values. An additional useful algorithm is gapped BLAST which uses BLOSUM-62 substitution scores. [0115] An "isoform" as used herein refers to a variant of the reference protein ex pressed by the same species as the species of the reference protein (e. g. Q13421-1 to Q13421-4 or Q08722-1 to Q08722-4 as described above).

[0116] A "homologue" as used herein refers to a variant of the reference protein pro duced in different species as compared to the species of the reference protein, e. g. the human species described herein above. In certain embodiments a homologue also in cludes an orthologue.

[0117] A "fragment" as used herein refers to a portion of the reference protein. A "var iant" as used herein refers to a protein having an amino acid sequence comprising one or more amino acid substitutions, insertions, deletions or other modifications relative to the amino acid sequence of the reference protein, but retaining a considerable de gree of sequence identity (e. g. at least 70%) to the amino acid sequence of the reference protein, e. g. SEQ ID NO: 103, SEQ ID NO: 107, SEQ ID NO: 112, or in further embodi ments of the processed form of mesothelin or of CD47 or of SIRPalpha, which is pre- sented at the surface of a cell as defined above. A fragment of a reference protein, e. g. of SEQ ID NO: 103, SEQ ID NO: 107, or SEQ ID NO: 112 may be of any length (by number of amino acids). In certain embodiments it may have a length of 20%, 30%, 40%, 50%, 75%, 80%, 85%, 90%, 91%, 92% 93%, 94%, 95%, 96%, 97%, 98%, or 99% of the length of the reference protein. It is particularly preferred that the fragment is capable of per forming one, more or all function(s) performed by the reference protein. The fragment of a reference protein may, for example, have a length of 70, 100, 150, 200, 250, 300 or 350 amino acids.

[0118] The term "antibody" as used herein relates to a protein including at least one or two heavy chain (HC) variable regions (abbreviated as VH), and at least one or two light chain (LC) variable regions (abbreviated as VL). The VH and VL regions can further be subdivided into regions of hypervariability, called "complementarity determining re gions", abbreviated as "CDR", interspersed with more conserved regions termed "frame work regions", abbreviated as "FR". Antibodies generally comprise six complementarity determining regions CDRs; three in the heavy chain variable (VH) region: CDRH1, CDRH2 and CDRH3, and three in the light chain variable (VL) region: CDRL1, CDRL2, and CDRL3. The six CDRs define the paratope of the antibody which is the part of the antibody that binds to the target antigen. The VH region and VL region comprise the framework re gions (FR1, FR2, FR3 and FR4) at either side of each CDR, which provide a scaffold for the CDRs to display the CDRs on the surface of the VH and VL region. From N-terminus to C- terminus, VH regions comprise the following structure: N terminus-[HC-FRl]-[CDRHl]- [HC-FR2]-[CDRH2]-[HC-FRB]-[CDRHB]-[HC-FR4]-C terminus; and VL regions comprise the following structure: N terminus-[LC-FRl]-[CDRLl]-[LC-FR2]-[CDRL2]-[LC-FR3]-[CDRL3]- [LC-FR4]-C terminus. [0119] There are several different conventions for defining antibody CDRs and FRs such as Kabat et al., 1991, Sequences of Proteins of Immunological Interest, or Chothia et al., 1987, J. Mol. Biol. 196:901-917. Further information would be known to the skilled person or can be derived from suitable literature sources such as http://www.bio- inf.org. uk/abs/info.html#cdrid (Prof. Andrew C.R. Martin's group at UCL bioinf.org.uk) or the IMGT database (Brochet et al, 2008, Nucl. Acids Res. 36, W503-508).

[0120] The term "antibody" as used herein generally refers to intact immunoglobulins, e. g. of types IgA, IgG, IgE, IgD, IgM (as well as subtypes thereof), wherein the light chains of the immunoglobulin may be of types kappa or lambda. The antibody further com prises constant regions such as light chain constant region CL, and heavy chain constant regions CHI, CH2 and CH3. Further, the antibody may comprise a hinge domain. The antibody molecules can be full-length or can, in certain embodiments, be an antigen binding fragment or antigen-binding molecule derived form an antibody. In certain em bodiments, the present invention specifically envisages the use of antigen-binding frag ments of an antibody. An "antigen-binding fragment" of an antibody refers to a molecule which is capable of binding to a target antigen or epitope, but does not have an anti body's full length or may differ from a naturally occurring antibody structure. Antigen binding fragments or synthetic antibodies or antibody derivatives are recombinant pro teins derived from gene engineering. Examples of antigen binding fragments encompass Fv, scFv, Fab, scFab, F(ab')2, Fab2, diabody formats, triabody formats, triplebody for- mats, scFv-Fc, minibodies, single domain antibodies (e. g. VHH), different variants for truncated antibodies, bispecific antibodies as long as they display binding to the relevant target molecule(s). [0121] The antigen-binding fragment according to the present invention comprises a moiety or moieties capable of binding to a target antigen(s). In some embodiments, the moiety capable of binding to a target antigen comprises an antibody heavy chain varia ble region (VH) and an antibody light chain variable region (VL) of an antibody capable of specific binding to the target antigen, or sub-forms thereof.

[0122] It is particularly preferred that the antibody is of isotype IgGl. The term "IgGl" antibody as used herein refers to an antibody of the IgG subclass which differs from other antibodies in this subclass (e. g. lgG2, lgG3, lgG4) by differences in the constant regions, in particular the hinge region and the (upper) CH2 domain (Vidarsson et a I, 2014, Front Immunol 5(16):520). These regions are involved in binding to Fc gamma re ceptors (FcgRs) and the complement protein Clq. As a result, the different subclasses have different effector functions, both in terms of triggering FcgR-expressing cells, re sulting in phagocytosis or antibody-dependent cell-mediated cytotoxicity, and activating the complement cascade. [0123] An anti MSLN IgGl antibody according to the present invention is a glycopro tein of four polypeptide chains of two light chains (LC) and two heavy chains (HC) which are connected by disulphide bonds. The molecular weight of an IgGl antibody is ~150,000 daltons (Da). Each LC consists of two domains, the variable domain (VL) and the constant domain (CL) and has ~25,000 Da. Two different types of light chains are known, lambda and kappa, where a single IgGl according to the present invention can only contain either lambda or kappa. Each HC has a molecular weight of ~50,000 Da and comprises a variable (VH) and three constant domains (CHI, CH2 and CH3). The region between CHI and CH2 is called hinge region. The enzyme papain cleaves the IgGl mol ecule in the hinge region between the CHI and CH2 domain. This cleavage results in two identical Fab (fragment antigen binding) fragments, which retain the antigen-binding site (paratope), and one Fc (fragment crystallizable) fragment. The Fc fragment or Fc domain or Fc region is glycosylated and has many different effector functions. E. g. the Fc domain can bind and activate the complement system and can bind and activate FcgRs on macrophages, monocytes or NK cells. Thus, the hinge region connects the two Fab arms to the Fc region. The hinge region allows flexibility between the two Fab do mains and the Fc domain to accommodate binding to two antigen binding sites. IgG an tibodies can be further divided into four subclasses, also known as isotypes (for humans IgGl, lgG2, lgG3 and lgG4). Treatment of an IgGl molecule with pepsin generates the F(ab')2 fragment, which consists of the two Fab domains linked by the hinge region. Be cause the F(ab')2 molecule is bivalent, it can be capable of binding two antigens (epitopes).

[0124] The term "Fc region" or "Fc domain" refers to a dimer of heavy chain constant regions CH2 and CH3 which can be linked by disulphide bonds in hinge region of the antibody. A complete antibody can typically be separated into two Fab regions and one Fc region. The term "Fab region" as used herein comprises a CL and VL domain of the light chain as well as a CHI and VH domain of the heavy chain of the antibody. The Fc region of an antibody, i. e. the combination of two CH2 and two CH3 domains, typically interacts with cell surface receptors for Fc (Fc receptors, FcRs) as well as certain proteins of the complement system. IgG FcRs are cell surface molecules situated in the mem brane of cells and are expressed by several hematopoietic cells that recognize the Fc region of antibodies and their subclasses. FcRs for IgG are the Fc gamma receptors FcgRI or CD64, FcgRII or CD32, and FcgRIII or CD16. The neonatal FcR (FcRn) expressed on cells of the intestinal epithelium, placenta, and endothelium also binds IgG type antibodies. Engagement of FcRs expressed by immune cells initiates a number of immune modula tory functions in the immune response. Some FcRs contain activation motifs, i. e. the immunoreceptor tyrosine-based activation motif (ITAM) to induce cell signalling such as phagocytosis (antibody-dependent cellular phagocytosis, ADCP), antibody-dependent cellular cytotoxicity (ADCC), complement-dependent cytotoxicity (CDC), degranulation and/or cytokine release, all of which depend on the cell type expressing the FcR and the IgG antibody class and subclass. E. g. engagement of the type I FcRs by immune com plexes such as antibodies, induces receptor aggregation followed by immune cell acti vation. Binding of an IgGl antibody to an FcR on an immune effector cell thus often allows not only to recruit the immune cell to the tumor cell, but at the same time also to activate the immune cell by signalling through FcR and thus promoting destruction of the tumor cell.

[0125] The regions of the IgGl that interact with the FcRs are known and can be mod ified in order to increase binding of an IgGl to the FcR thereby augmenting ADCC and ADCP. In particular, it has been shown that S239D/I332E mutations in the CH2 domain can enhance FcgRIIIa binding and ADCC (Lazar et al., 2006, PNAS 103, 4005). SEQ ID 127 represents such a modified CH2/CH3 chain with an enhanced FcR binding (Fc enhanced) that can be used in the protein constructs of the present invention.

[0126] ADCC is induced when Fc gamma receptors (FcgRs) on innate immune effector cells are engaged by the Fc domain of antibodies that are bound to the surface of the target cells, e. g. to viral proteins on the surface of virus-infected cells or to specific tu mor antigens on the surface of tumor cells. This interaction induces the release of cyto toxic granules (containing perforins and granzymes) resulting in killing of infected cells. Multiple innate effector cells, including natural killer (NK) cells, neutrophils, monocytes, and macrophages, are capable of ADCC in vitro. The most important contributors to ADCC in vivo are thought to be NK cells, which express primarily FcgRIIIA. ADCC has been recognized as an important mechanism of action for monoclonal antibodies that target tumor cells.

[0127] ADCP or phagocytosis is the uptake of antibody-coated target cells by phago- cytic cells. Phagocytic cells, including monocytes, macrophages, neutrophils, eosinophils and dendritic cells (DCs), express Fc receptors such as FcgRI, FcgRIla, FcgRIIc, FcgRIIIa and Fc alpha Rlllb, which can all mediate immune complex uptake and phagocytosis.

[0128] The Fc domain can also induce complement activation, contributing to cell elimination either directly, by means of complement-dependent cytotoxicity (CDC), or indirectly, through phagocytic clearance of complement-coated targets and the induc tion of an inflammatory response. Activation of the classical complement pathway oc curs when the recognition molecule Clq, in complex with the Clr and Cls serine prote ases, binds to the Fc domain of an antibody (generally IgGl and IgM) attached to the cell surface of a target cell. Upon binding of Clq, the proteases Clr and C2r are autocatalyt- ically activated, leading to cleavage of C2 and C4. The larger fragments thereof associate to form C4bC2a on the surface of target cells, and the complex gains the ability to cleave C3 and is termed the C3 convertase. The C3 convertase in turn cleaves C3 into C3a (an- aphylatoxin) and C3b (Opsonin). C3b can covalently bind to the surface of target cells and tags them as foreign, providing the opsonic signal to phagocytes for ingestion and subsequent killing or degradation. Some of the cleaved C3b remains associated with the C4b2b forming C4b2b3b, the classical pathway C5 convertase. The C5 convertase then cleaves C5 into C5a and C5b. C5b initiates the formation of the pore-forming/membrane attack (MAC) complex, resulting in lysis of the target cell. The release of anaphylatoxins C3a and C5a stimulates a pro-inflammatory environment by inducing the recruitment of immune effector cells and the activation of leukocytes, endothelial cells, epithelial cells, thrombocytes or platelets (Bordron et al, 2020, Clin Rev Allerg Immunolog; 58:155).

[0129] The term "anti-mesothelin (MSLN) IgGl antibody" as used herein particularly refers to an IgGl antibody, which specifically binds to mesothelin or MSLN as defined above.

[0130] As used herein, an antibody, e. g. an IgGl antibody, that "specifically binds" to or is "specific" for a particular polypeptide or an epitope on a particular polypeptide is an antibody that binds to that particular polypeptide or epitope on a particular polypep tide without substantially binding to any other polypeptide or polypeptide epitope with the same domain. In other words, the IgGl antibody component of the protein construct according to the present invention is specific for MSLN. This does not exclude the possi bility that other components, e. g. functional components, of the protein construct ac cording to the present invention bind to different or other polypeptides or epitopes on other polypeptides. For example, such a different component of the protein construct is the polypeptide capable of binding to CD47, i. e. a component which per se does not bind to MSLN. [0131] Typically, the term "specific binding" or "specifically binds" refers to the ability of an antibody or antigen binding fragment thereof to bind to its target, e. g. MSLN with an affinity that is at least five-fold greater than its affinity for a non-specific antigen.

[0132] In certain embodiments the anti MSLN IgGl antibody being a part of the pro- tein construct of the present invention is a chimeric antibody. A "chimeric antibody" can be produced by recombinant DNA techniques known in the art. For example, a gene encoding the Fc constant region of a murine (or other species) monoclonal antibody molecule can be digested with restriction enzymes to remove the region encoding the murine Fc, and the equivalent portion of a gene encoding a human Fc constant region may be substituted. A chimeric antibody may also be generated by recombinant DNA techniques where DNA encoding murine variable regions can be ligated to DNA encod ing the human constant regions.

[0133] In another embodiment, the anti MSLN IgGl antibody being a part of the pro tein construct of the present invention is a humanized antibody, e. g. humanized by methods known in the art. A humanized antibody consists of non-human CDRs and a framework region and a constant region of a human antibody or derived from a human antibody. For example, a common method for humanization of non-human, i. e. murine or rat antibodies, is CDR grafting. Once murine or rat antibodies are obtained, the CDRs of the non-human antibody are grafted onto the human frameworks. A human frame- work with high homology to the non-human framework region is selected as acceptor framework for CDR grafting. In other words, humanized antibodies can be generated by replacing sequences of the murine or rat fragment variable (Fv) region that are not di rectly involved in antigen binding with equivalent sequences from human fragment var iable (Fv) regions. General methods for generating humanized antibodies are known in the art. Accordingly, the present invention envisages antibodies in which specific amino acids have been substituted, deleted, added or back-mutated to the non-human frame work. In particular, preferred antibodies may have amino acid substitutions in the framework region, such as to improve, optimize (e. g. increase affinity) or diminish (e. g. decrease affinity) binding to the antigen. For example, a selected, small number of ac ceptor framework residues of the immunoglobulin chain can be replaced by the corre sponding donor amino acids. Preferred locations of the substitutions include amino acid residues adjacent to the CDR, or which are capable of interacting with any of the CDRs. Criteria for selecting amino acids from the donor may, for example, be derived from US 5,585,089. The acceptor framework may, in preferred embodiments, be a mature hu man antibody framework sequence or a consensus sequence.

[0134] It is particularly preferred that the antibody is a monoclonal antibody. Mono clonal antibodies of defined specificity may be produced using, for instance, the hybrid- oma technology developed by Kohler and Milstein (Kohler and Milstein, 1976, Eur. J. Immunol., 6: 511-519). Typically, mice or rats are immunized with a recombinant pro tein. Once an immune response is detected, e. g., antibodies specific for the antigen are detected in the mouse or rat serum, the mouse or rat spleen is harvested and spleno- cytes isolated. The splenocytes are then fused by well-known techniques to any suitable myeloma cells, for example cells from cell line SP20 or X63AG8.653. Hybridomas are selected and cloned by limited dilution. The hybridoma clones are then assayed by methods known in the art for cells that secrete antibodies capable of binding a polypep tide of the invention.

[0135] Alternatively, antibodies according to the present invention can also be gener ated using various phage or recombinant, synthetic display methods known in the art. In phage display methods, functional antibody domains are displayed on the surface of phage particles, which carry the polynucleotide sequences encoding them. In a particu lar embodiment, such phages can be utilized to display antigen binding domains ex pressed from a repertoire or combinatorial antibody library (e. g., human or murine). Phages expressing an antigen binding domain that binds the antigen of interest can be selected or identified with antigen, e. g., using labeled antigen or antigen bound or cap tured to a solid surface or beads. Phages used in these methods are typically filamentous phages including M13. Binding domains expressed from a phage like Fab, Fv or disulfide stabilized Fv antibody domains are recombinantly fused to either the phage gene III or gene VIII protein. Examples of phage display methods that can be used to produce anti bodies according to the present invention include those disclosed in Brinkman et al., 1995, J. Immunol. Methods 182: 41-50.

[0136] The antibodies or antigen binding fragments thereof, e. g. the IgGl antibody binding to MSLN, or the antibody or antigen-binding fragment or sub-form of an anti body binding to CD47 described herein may be raised in any mammal, wild-type and/or transgenic, including, for example, mice, rats, rabbits, and goat, or may be produced synthetically, e. g. by expression from vectors, plasmids or artificial chromosomes in suitable host cells. [0137] In specific embodiments, the IgGl antibody or antigen binding fragment thereof which specifically binds to MSLN is specific for a polypeptide comprised in the amino acid sequence of SEQ ID NO: 103 to 106, preferably SEQ ID NO: 103.

[0138] In preferred embodiments of the present invention the IgGl antibody portion of the protein construct, or antigen binding fragment thereof, which specifically binds to MSLN comprises a VH region comprising the following CDRs:

(1) the CDRH1 sequence selected from the amino acid sequences of SEQ ID NO: 28, 86, 87, 88 and 122; and

(2) a CDRH2 sequence selected from the amino acid sequences of SEQ ID NOs: 29, 30, 31, 89, 90 and 91; and (3) a CDRH3 sequence selected from the amino acid sequences of SEQ ID NOs: 32, 33,

92, 93 and 94; or a variant thereof in which one or two or three amino acids in one or more of CDRH1, CDRH2, or CDRH3 are substituted with another amino acid; and comprises a VL region comprising the following CDRs: (4) a CDRL1 sequence selected from the amino acid sequences of SEQ ID NOs: 34, 35,

95, 96 and 97; and

(5) a CDRL2 sequence selected from the amino acid sequences of SEQ ID NOs: 36 to 39, 98, 99 and 121; and (6) a CDRL3 sequence selected from the amino acid sequences of SEQ ID NOs: 40, 41, 100, 101 and 102 or a variant thereof in which one or two or three amino acids in one or more of CDRL1, CDRL2, or CDRL3 are substituted with another amino acid.

[0139] In further preferred embodiments of the protein construct of the present in vention the IgGl antibody portion of the protein construct, or antigen binding fragment thereof, which specifically binds to MSLN comprises a VH region comprising the follow ing CDRs

(1) the CDRH1 sequence having the amino acid sequence of SEQ ID NO: 28; and

(2) the CDRH2 sequence having the amino acid sequence of SEQ ID NO: 31; and

(3) the CDRH3 sequence having the amino acid sequence of SEQ ID NO: 32; or a variant thereof in which one or two or three amino acids in one or more of CDRH1, CDRH2, or CDRH3 are substituted with another amino acid; and comprises a VL region comprising the following CDRs:

(4) the CDRL1 sequence having the amino acid sequence of SEQ ID NO: 34; and

(5) the CDRL2 sequence having the amino acid sequence of SEQ ID NO:38; and

(6) the CDRL3 sequence having the amino acid sequence of SEQ ID NO: 40; or a variant thereof in which one or two or three amino acids in one or more of CDRL1, CDRL2, or CDRL3 are substituted with another amino acid.

[0140] In a further preferred embodiment of the protein construct of the present in vention the IgGl antibody portion of the protein construct, or antigen binding fragment thereof, which specifically binds to MSLN comprises a VH region comprising the follow ing CDRs

(1) the CDRH1 sequence having the amino acid sequence of SEQ ID NO: 122; and

(2) the CDRH2 sequence having the amino acid sequence of SEQ ID NO: 90; and

(3) the CDRH3 sequence having the amino acid sequence of SEQ ID NO: 93 and comprises a VL region comprising the following CDRs:

(4) the CDRL1 sequence having the amino acid sequence of SEQ ID NO: 96; and

(5) the CDRL2 sequence having the amino acid sequence of SEQ ID NO:99; and (6) the CDRL3 sequence having the amino acid sequence of SEQ ID NO: 101.

[0141] In a further preferred embodiment of the protein construct of the present in vention the IgGl antibody portion of the protein construct, or antigen binding fragment thereof, which specifically binds to MSLN comprises a VH region comprising the follow- ing CDRs

(1) the CDRH1 sequence having the amino acid sequence of SEQ ID NO: 87; and

(2) the CDRH2 sequence having the amino acid sequence of SEQ ID NO: 90; and

(3) the CDRH3 sequence having the amino acid sequence of SEQ ID NO: 93 and comprises a VL region comprising the following CDRs: (4) the CDRL1 sequence having the amino acid sequence of SEQ ID NO: 96; and

(5) the CDRL2 sequence having the amino acid sequence of SEQ ID NO:99; and

(6) the CDRL3 sequence having the amino acid sequence of SEQ ID NO: 101.

[0142] In a further preferred embodiment of the protein construct of the present in vention the IgGl antibody portion of the protein construct, or antigen binding fragment thereof, which specifically binds to MSLN comprises a VH region comprising the follow ing CDR

(1) the CDRH1 sequence having the amino acid sequence of SEQ ID NO: 28; and

(2) the CDRH2 sequence having the amino acid sequence of SEQ ID NO: 29; and

(3) the CDRH3 sequence having the amino acid sequence of SEQ ID NO: 32; and comprises a VL region comprising the following CDRs:

(4) the CDRL1 sequence having the amino acid sequence of SEQ ID NO: 34; and

(5) the CDRL2 sequence having the amino acid sequence of SEQ ID NO:36; and

(6) the CDRL3 sequence having the amino acid sequence of SEQ ID NO: 40.

[0143] In a further preferred embodiment of the protein construct of the present in- vention the IgGl antibody portion of the protein construct, or antigen binding fragment thereof, which specifically binds to MSLN comprises a VH region comprising the follow ing CDRs

(1) the CDRH1 sequence having the amino acid sequence of SEQ ID NO: 86; and (2) the CDRH2 sequence having the amino acid sequence of SEQ ID NO: 89; and

(3) the CDRH3 sequence having the amino acid sequence of SEQ ID NO: 92 and comprises a VL region comprising the following CDRs:

(4) the CDRL1 sequence having the amino acid sequence of SEQ ID NO: 95; and (5) the CDRL2 sequence having the amino acid sequence of SEQ ID NO:98; and

(6) the CDRL3 sequence having the amino acid sequence of SEQ ID NO: 100.

[0144] In a further preferred embodiment of the protein construct of the present in vention the IgGl antibody portion of the protein construct, or antigen binding fragment thereof, which specifically binds to MSLN comprises a VH region comprising the follow- ing CDRs

(1) the CDRH1 sequence having the amino acid sequence of SEQ ID NO: 88; and

(2) the CDRH2 sequence having the amino acid sequence of SEQ ID NO: 91; and

(3) the CDRH3 sequence having the amino acid sequence of SEQ ID NO: 94 and comprises a VL region comprising the following CDRs: (4) the CDRL1 sequence having the amino acid sequence of SEQ ID NO: 97; and

(5) the CDRL2 sequence having the amino acid sequence of SEQ ID NO: 121; and

(6) the CDRL3 sequence having the amino acid sequence of SEQ ID NO: 102.

[0145] The sequences of the CDRH1, 2, 3 and CDRL1, 2, 3 according to the present invention are provided in Tables 1 and 2, respectively.

Table 1:

Table 2:

[0146] In a further group of preferred embodiments of the protein construct of the present invention the IgGl antibody portion of the protein construct or antigen binding fragment thereof which specifically binds to MSLN comprises a variable light chain re gion amino acid sequence at least about 60%, 65%, 70%, 75%, 80%, 85%, 86%, 87%, 88%, 89%, 90%, 91%, 92%, 93%, 94%, 95%, 96%, 97%, 98%, 99%, or 100% identical to a vari able light chain region amino acid sequence as set forth in SEQ ID NOs: 3, 4, 5, 6, 7, 8, 9, 10, 11 or 126.

[0147] In a further group of preferred embodiments of the protein construct of the present invention the IgGl antibody portion of the protein construct or antigen binding fragment thereof which specifically binds to MSLN comprises a variable heavy chain re gion amino acid sequence at least about 60%, 65%, 70%, 75%, 80%, 85%, 86%, 87%, 88%, 89%, 90%, 91%, 92%, 93%, 94%, 95%, 96%, 97%, 98%, 99%, or 100% identical to a vari able heavy chain region amino acid sequence as set forth in SEQ ID NOs: 14, 15, 16, 17, 18, 19, 20, 116, 117, 118, 119, 120, 123, 124 or 125.

[0148] In a further group of preferred embodiments of the protein construct of the present invention the IgGl antibody portion of the protein construct or antigen binding fragment thereof which specifically binds to MSLN comprises a variable light chain re gion amino acid sequence at least about 60%, 65%, 70%, 75%, 80%, 85%, 86%, 87%, 88%, 89%, 90%, 91%, 92%, 93%, 94%, 95%, 96%, 97%, 98%, 99%, or 100% identical to a vari able light chain region amino acid sequence as set forth in SEQ ID NO: 3; and comprises a variable heavy chain region amino acid sequence at least about 60%, 65%, 70%, 75%, 80%, 85%, 86%, 87%, 88%, 89%, 90%, 91%, 92%, 93%, 94%, 95%, 96%, 97%, 98%, 99%, or 100% identical to a variable heavy chain region amino acid sequence as set forth in SEQ ID NOs: 14, 15, 16, or 17.

[0149] In a further group of preferred embodiments of the protein construct of the present invention the IgGl antibody portion of the protein construct or antigen binding fragment thereof which specifically binds to MSLN comprises a variable light chain re gion amino acid sequence at least about 60%, 65%, 70%, 75%, 80%, 85%, 86%, 87%, 88%, 89%, 90%, 91%, 92%, 93%, 94%, 95%, 96%, 97%, 98%, 99%, or 100% identical to a vari able light chain region amino acid sequence as set forth in SEQ ID NO: 4; and comprises a variable heavy chain region amino acid sequence at least about 60%, 65%, 70%, 75%, 80%, 85%, 86%, 87%, 88%, 89%, 90%, 91%, 92%, 93%, 94%, 95%, 96%, 97%, 98%, 99%, or 100% identical to a variable heavy chain region amino acid sequence as set forth in SEQ ID NO: 16.

[0150] In a further group of preferred embodiments of the protein construct of the present invention the IgGl antibody portion of the protein construct or antigen binding fragment thereof which specifically binds to MSLN comprises a variable light chain re gion amino acid sequence at least about 60%, 65%, 70%, 75%, 80%, 85%, 86%, 87%, 88%, 89%, 90%, 91%, 92%, 93%, 94%, 95%, 96%, 97%, 98%, 99%, or 100% identical to a vari able light chain region amino acid sequence as set forth in SEQ ID NO: 5; and comprises a variable heavy chain region amino acid sequence at least about 60%, 65%, 70%, 75%, 80%, 85%, 86%, 87%, 88%, 89%, 90%, 91%, 92%, 93%, 94%, 95%, 96%, 97%, 98%, 99%, or 100% identical to a variable heavy chain region amino acid sequence as set forth in SEQ ID NO: 16.

[0151] In a further group of preferred embodiments of the protein construct of the present invention the IgGl antibody portion of the protein construct or antigen binding fragment thereof which specifically binds to MSLN comprises a variable light chain re gion amino acid sequence at least about 60%, 65%, 70%, 75%, 80%, 85%, 86%, 87%, 88%, 89%, 90%, 91%, 92%, 93%, 94%, 95%, 96%, 97%, 98%, 99%, or 100% identical to a vari able light chain region amino acid sequence as set forth in SEQ ID NO: 6; and comprises a variable heavy chain region amino acid sequence at least about 60%, 65%, 70%, 75%, 80%, 85%, 86%, 87%, 88%, 89%, 90%, 91%, 92%, 93%, 94%, 95%, 96%, 97%, 98%, 99%, or 100% identical to a variable heavy chain region amino acid sequence as set forth in SEQ ID NOs: 14, 15, 16 or 17.

[0152] In a further group of preferred embodiments of the protein construct of the present invention the IgGl antibody portion of the protein construct or antigen binding fragment thereof which specifically binds to MSLN comprises a variable light chain re gion amino acid sequence at least about 60%, 65%, 70%, 75%, 80%, 85%, 86%, 87%, 88%, 89%, 90%, 91%, 92%, 93%, 94%, 95%, 96%, 97%, 98%, 99%, or 100% identical to a vari able light chain region amino acid sequence as set forth in SEQ ID NO: 7; and comprises a variable heavy chain region amino acid sequence at least about 60%, 65%, 70%, 75%, 80%, 85%, 86%, 87%, 88%, 89%, 90%, 91%, 92%, 93%, 94%, 95%, 96%, 97%, 98%, 99%, or 100% identical to a variable heavy chain region amino acid sequence as set forth in SEQ ID NO: 16. [0153] In a further group of preferred embodiments of the protein construct of the present invention the IgGl antibody portion of the protein construct or antigen binding fragment thereof which specifically binds to MSLN comprises a variable light chain re gion amino acid sequence at least about 60%, 65%, 70%, 75%, 80%, 85%, 86%, 87%, 88%, 89%, 90%, 91%, 92%, 93%, 94%, 95%, 96%, 97%, 98%, 99%, or 100% identical to a vari able light chain region amino acid sequence as set forth in SEQ ID NO: 8; and comprises a variable heavy chain region amino acid sequence at least about 60%, 65%, 70%, 75%, 80%, 85%, 86%, 87%, 88%, 89%, 90%, 91%, 92%, 93%, 94%, 95%, 96%, 97%, 98%, 99%, or 100% identical to a variable heavy chain region amino acid sequence as set forth in SEQ ID NO: 16.

[0154] In a further group of preferred embodiments of the protein construct of the present invention the IgGl antibody portion of the protein construct or antigen binding fragment thereof which specifically binds to MSLN comprises a variable light chain re gion amino acid sequence at least about 60%, 65%, 70%, 75%, 80%, 85%, 86%, 87%, 88%, 89%, 90%, 91%, 92%, 93%, 94%, 95%, 96%, 97%, 98%, 99%, or 100% identical to a vari able light chain region amino acid sequence as set forth in SEQ ID NO: 9; and comprises a variable heavy chain region amino acid sequence at least about 60%, 65%, 70%, 75%, 80%, 85%, 86%, 87%, 88%, 89%, 90%, 91%, 92%, 93%, 94%, 95%, 96%, 97%, 98%, 99%, or 100% identical to a variable heavy chain region amino acid sequence as set forth in SEQ ID NOs: 16, 116, 117, 118, 119 or 120.

[0155] In a further group of preferred embodiments of the protein construct of the present invention the IgGl antibody portion of the protein construct or antigen binding fragment thereof which specifically binds to MSLN comprises a variable light chain re gion amino acid sequence at least about 60%, 65%, 70%, 75%, 80%, 85%, 86%, 87%, 88%, 89%, 90%, 91%, 92%, 93%, 94%, 95%, 96%, 97%, 98%, 99%, or 100% identical to a vari able light chain region amino acid sequence as set forth in SEQ ID NO: 10; and comprises a variable heavy chain region amino acid sequence at least about 60%, 65%, 70%, 75%, 80%, 85%, 86%, 87%, 88%, 89%, 90%, 91%, 92%, 93%, 94%, 95%, 96%, 97%, 98%, 99%, or 100% identical to a variable heavy chain region amino acid sequence as set forth in SEQ ID NOs: 18, 19 or 20.

[0156] In a further group of preferred embodiments of the protein construct of the present invention the IgGl antibody portion of the protein construct or antigen binding fragment thereof which specifically binds to MSLN comprises a variable light chain re gion amino acid sequence at least about 60%, 65%, 70%, 75%, 80%, 85%, 86%, 87%, 88%, 89%, 90%, 91%, 92%, 93%, 94%, 95%, 96%, 97%, 98%, 99%, or 100% identical to a vari able light chain region amino acid sequence as set forth in SEQ ID NO: 11; and comprises a variable heavy chain region amino acid sequence at least about 60%, 65%, 70%, 75%, 80%, 85%, 86%, 87%, 88%, 89%, 90%, 91%, 92%, 93%, 94%, 95%, 96%, 97%, 98%, 99%, or 100% identical to a variable heavy chain region amino acid sequence as set forth in SEQ ID NOs: 18, 19 or 20.

[0157] In a further group of preferred embodiments of the protein construct of the present invention the IgGl antibody portion of the protein construct or antigen binding fragment thereof which specifically binds to MSLN comprises a variable light chain re gion amino acid sequence at least about 60%, 65%, 70%, 75%, 80%, 85%, 86%, 87%, 88%, 89%, 90%, 91%, 92%, 93%, 94%, 95%, 96%, 97%, 98%, 99%, or 100% identical to a vari able light chain region amino acid sequence as set forth in SEQ ID NO: 126; and com prises a variable heavy chain region amino acid sequence at least about 60%, 65%, 70%, 75%, 80%, 85%, 86%, 87%, 88%, 89%, 90%, 91%, 92%, 93%, 94%, 95%, 96%, 97%, 98%, 99%, or 100% identical to a variable heavy chain region amino acid sequence as set forth in SEQ ID NOs: 123, 124, or 125.

[0158] In a particularly preferred embodiment of the protein construct of the present invention the IgGl antibody portion of the protein construct or antigen binding frag ment thereof which specifically binds to MSLN comprises a variable light chain region amino acid sequence at least about 60%, 65%, 70%, 75%, 80%, 85%, 86%, 87%, 88%, 89%, 90%, 91%, 92%, 93%, 94%, 95%, 96%, 97%, 98%, 99%, or 100% identical to a vari able light chain region amino acid sequence as set forth in SEQ ID NO: 9; and comprises a variable heavy chain region amino acid sequence at least about 60%, 65%, 70%, 75%, 80%, 85%, 86%, 87%, 88%, 89%, 90%, 91%, 92%, 93%, 94%, 95%, 96%, 97%, 98%, 99%, or 100% identical to a variable heavy chain region amino acid sequence as set forth in SEQID NO: 117. [0159] In a particularly preferred embodiment of the protein construct of the present invention the IgGl antibody portion of the protein construct or antigen binding frag ment thereof which specifically binds to MSLN comprises a variable light chain region amino acid sequence at least about 60%, 65%, 70%, 75%, 80%, 85%, 86%, 87%, 88%, 89%, 90%, 91%, 92%, 93%, 94%, 95%, 96%, 97%, 98%, 99%, or 100% identical to a vari- able light chain region amino acid sequence as set forth in SEQ ID NO: 126; and com prises a variable heavy chain region amino acid sequence at least about 60%, 65%, 70%, 75%, 80%, 85%, 86%, 87%, 88%, 89%, 90%, 91%, 92%, 93%, 94%, 95%, 96%, 97%, 98%, 99%, or 100% identical to a variable heavy chain region amino acid sequence as set forth in SEQID NO: 125. [0160] In a particularly preferred embodiment of the protein construct of the present invention the IgGl antibody portion of the protein construct or antigen binding frag ment thereof which specifically binds to MSLN comprises a variable light chain region amino acid sequence at least about 60%, 65%, 70%, 75%, 80%, 85%, 86%, 87%, 88%, 89%, 90%, 91%, 92%, 93%, 94%, 95%, 96%, 97%, 98%, 99%, or 100% identical to a vari- able light chain region amino acid sequence as set forth in SEQ ID NO: 126; and com prises a variable heavy chain region amino acid sequence at least about 60%, 65%, 70%, 75%, 80%, 85%, 86%, 87%, 88%, 89%, 90%, 91%, 92%, 93%, 94%, 95%, 96%, 97%, 98%, 99%, or 100% identical to a variable heavy chain region amino acid sequence as set forth in SEQID NO: 123 [0161] In a particularly preferred embodiment of the protein construct of the present invention the IgGl antibody portion of the protein construct or antigen binding frag ment thereof which specifically binds to MSLN comprises a variable light chain region amino acid sequence at least about 60%, 65%, 70%, 75%, 80%, 85%, 86%, 87%, 88%, 89%, 90%, 91%, 92%, 93%, 94%, 95%, 96%, 97%, 98%, 99%, or 100% identical to a vari able light chain region amino acid sequence as set forth in SEQ ID NO: 7; and comprises a variable heavy chain region amino acid sequence at least about 60%, 65%, 70%, 75%, 80%, 85%, 86%, 87%, 88%, 89%, 90%, 91%, 92%, 93%, 94%, 95%, 96%, 97%, 98%, 99%, or 100% identical to a variable heavy chain region amino acid sequence as set forth in SEQ ID NO: 16.

[0162] In a particularly preferred embodiment of the protein construct of the present invention the IgGl antibody portion of the protein construct or antigen binding frag ment thereof which specifically binds to MSLN comprises a variable light chain region amino acid sequence at least about 60%, 65%, 70%, 75%, 80%, 85%, 86%, 87%, 88%, 89%, 90%, 91%, 92%, 93%, 94%, 95%, 96%, 97%, 98%, 99%, or 100% identical to a vari able light chain region amino acid sequence as set forth in SEQ ID NO: 9; and comprises a variable heavy chain region amino acid sequence at least about 60%, 65%, 70%, 75%, 80%, 85%, 86%, 87%, 88%, 89%, 90%, 91%, 92%, 93%, 94%, 95%, 96%, 97%, 98%, 99%, or 100% identical to a variable heavy chain region amino acid sequence as set forth in SEQ ID NO: 16.

[0163] In a particularly preferred embodiment of the protein construct of the present invention the IgGl antibody portion of the protein construct or antigen binding frag ment thereof which specifically binds to MSLN comprises a variable light chain region amino acid sequence at least about 60%, 65%, 70%, 75%, 80%, 85%, 86%, 87%, 88%, 89%, 90%, 91%, 92%, 93%, 94%, 95%, 96%, 97%, 98%, 99%, or 100% identical to a vari able light chain region amino acid sequence as set forth in SEQ ID NO: 6; and comprises a variable heavy chain region amino acid sequence at least about 60%, 65%, 70%, 75%, 80%, 85%, 86%, 87%, 88%, 89%, 90%, 91%, 92%, 93%, 94%, 95%, 96%, 97%, 98%, 99%, or 100% identical to a variable heavy chain region amino acid sequence as set forth in SEQ ID NO: 15. [0164] In an even more preferred embodiment of the protein construct of the present invention the IgGl antibody portion of the protein construct or antigen binding frag ment thereof which specifically binds to MSLN comprises a variable light chain region amino acid sequence of SEQ ID NO: 9 and variable heavy chain region amino acid se- quence of SEQ ID NO: 117.

[0165] In an even more preferred embodiment of the protein construct of the present invention the IgGl antibody portion of the protein construct or antigen binding frag ment thereof which specifically binds to MSLN comprises a variable light chain region amino acid sequence of SEQ ID NO: 126 and variable heavy chain region amino acid se- quence of SEQ ID NO: 125.

[0166] The sequences of the variable light chain region and of the variable heavy chain region of IgGl antibody portion of the protein construct and the corresponding antibod ies are shown in thefollowingTable S and the selection of protein constructs is explained in Example 6.

Table 3:

[0167] In further preferred embodiments the IgGl antibody portion of the protein construct, or antigen binding fragment thereof, which specifically binds to MSLN com prises a kappa constant light chain (CL) domain. [0168] In further preferred embodiments the IgGl antibody portion of the protein construct, or antigen binding fragment thereof, which specifically binds to MSLN com prises a CL domain which comprises an amino acid sequence which is at least about 60%, 65%, 70%, 75%, 80%, 85%, 86%, 87%, 88%, 89%, 90%, 91%, 92%, 93%, or 94% identical, preferably about 95%, 96%, 97%, 98%, 99%, or 100% identical to the amino acid se quence of SEQ ID NO: 27.

[0169] In further preferred embodiments the IgGl antibody portion of the protein construct, or antigen binding fragment thereof, which specifically binds to MSLN com prises a constant heavy chain domain 1 (CHI) which comprises an amino acid sequence which is at least about 60%, 65%, 70%, 75%, 80%, 85%, 86%, 87%, 88%, 89%, 90%, 91%, 92%, 93%, or 94% identical, preferably about 95%, 96%, 97%, 98%, 99%, or 100% identi cal to the amino acid sequence of SEQ ID NO: 22.

[0170] In further preferred embodiments the IgGl antibody portion of the protein construct, or antigen binding fragment thereof, which specifically binds to MSLN com prises a constant heavy chain domain 2 (CH2) which comprises an amino acid sequence which is at least about 60%, 65%, 70%, 75%, 80%, 85%, 86%, 87%, 88%, 89%, 90%, 91%, 92%, 93%, or 94% identical, preferably about 95%, 96%, 97%, 98%, 99%, or 100% identi cal to the amino acid sequence of SEQ ID NO: 23 or SEQ ID NO: 128.

[0171] In further preferred embodiments the IgGl antibody portion of the protein construct, or antigen binding fragment thereof, which specifically binds to MSLN com prises a constant heavy chain domain 3 (CH3) which comprises an amino acid sequence which is at least about 60%, 65%, 70%, 75%, 80%, 85%, 86%, 87%, 88%, 89%, 90%, 91%, 92%, 93%, or 94% identical, preferably about 95%, 96%, 97%, 98%, 99%, or 100% identi cal to the amino acid sequence of SEQ ID NO: 24.

[0172] In further particularly preferred embodiments, the IgGl antibody portion of the protein construct, or antigen binding fragment thereof, which specifically binds to MSLN comprises a constant heavy chain domain 1 (CHI) comprising an amino acid se quence which is at least about 95%, 96%, 97%, 98%, 99%, or 100% identical to the amino acid sequence of SEQ ID NO: 22; and constant heavy chain domain 2 (CH2) comprising an amino acid sequence which is at least about 95%, 96%, 97%, 98%, 99%, or 100% iden tical to the amino acid sequence of SEQ ID NO: 23 or SEQ ID NO: 128; and constant heavy chain domain 3 (CH3) comprising an amino acid sequence which is at least about 95%, 96%, 97%, 98%, 99%, or 100% identical to the amino acid sequence of SEQ ID NO: 24. In further specific embodiments, the IgGl antibody portion of the protein construct, or antigen binding fragment thereof which specifically binds to MSLN comprises CDR se quences as defined herein above, or VH and/or VL sequences as defined herein above together with a CL and a CHI and a CH2 and a CH3 sequence as defined herein above.

[0173] In further preferred embodiments the IgGl antibody portion of the protein construct, or antigen binding fragment thereof, which specifically binds to MSLN com prises a hinge domain comprises an amino acid sequence which is at least about 95%, 96%, 97%, 98%, 99%, or 100% identical to the amino acid sequence of the amino acid sequence of SEQ ID NO: 25. The term "hinge domain" or alternatively "hinge region" as used herein refers to a functional domain in an antibody, which connects the CHI and the CH2 domain and thus is between the Fab and Fc domain. It typically comprises two disulphide bonds to dimerize two heavy chains and thereby contributes to the three- dimensional form and structure of an antibody. Its sequence, structure and position pro vided segmental flexibility to promote the antibody functionality.

[0174] In a further preferred embodiment the IgGl antibody portion of the protein construct, or antigen binding fragment thereof, which specifically binds to MSLN com prises a Fab domain which comprises: a variable light chain (VL) domain comprising an amino acid sequence which is at least about 95%, 96%, 97%, 98%, 99%, or 100% identical to the amino acid sequence of SEQ ID NO: 3, a variable heavy chain (VH) domain amino acid sequence which is at least about 95%, 96%, 97%, 98%, 99%, or 100% identical to the amino acid sequence of SEQID NO: 14, a constant light chain(CL) domain comprising an amino acid sequence which is at least about 95%, 96%, 97%, 98%, 99%, or 100% amino acid sequence of SEQ ID NO: 27 and a constant heavy chain (CHI) domain com prising an amino acid sequence which is at least about 95%, 96%, 97%, 98%, 99%, or 100% amino acid sequence of amino acid sequence of SEQ ID NO: 22.

[0175] In a further preferred embodiment the IgGl antibody portion of the protein construct, or antigen binding fragment thereof, which specifically binds to MSLN com prises a Fab domain which comprises: a variable light chain (VL) domain comprising an amino acid sequence which is at least about 95%, 96%, 97%, 98%, 99%, or 100% identical to the amino acid sequence of SEQ ID NO: 9, a variable heavy chain (VH) domain amino acid sequence which is at least about 95%, 96%, 97%, 98%, 99%, or 100% identical to the amino acid sequence of SEQ ID NO: 117, a constant light chain(CL) domain compris ing an amino acid sequence which is at least about 95%, 96%, 97%, 98%, 99%, or 100% amino acid sequence of SEQ ID NO: 27 and a constant heavy chain (CHI) domain com prising an amino acid sequence which is at least about 95%, 96%, 97%, 98%, 99%, or 100% amino acid sequence of amino acid sequence of SEQ ID NO: 22.

[0176] In a further preferred embodiment the IgGl antibody portion of the protein construct, or antigen binding fragment thereof, which specifically binds to MSLN com prises a Fab domain which comprises: a variable light chain (VL) domain comprising an amino acid sequence which is at least about 95%, 96%, 97%, 98%, 99%, or 100% identical to the amino acid sequence of SEQ ID NO: 126, a variable heavy chain (VH) domain amino acid sequence which is at least about 95%, 96%, 97%, 98%, 99%, or 100% identical to the amino acid sequence of SEQ ID NO: 125, a constant light chain(CL) domain com prising an amino acid sequence which is at least about 95%, 96%, 97%, 98%, 99%, or 100% amino acid sequence of SEQ ID NO: 27 and a constant heavy chain (CHI) domain comprising an amino acid sequence which is at least about 95%, 96%, 97%, 98%, 99%, or 100% amino acid sequence of amino acid sequence of SEQ ID NO: 22.

[0177] In a particularly preferred embodiment said Fab domain of the protein con struct comprises: (i) a variable light chain (VL) amino acid sequence of SEQ ID NO: 126, a variable heavy (VH) chain amino acid sequence of SEQ ID NO: 125, a constant light (CL) chain amino acid sequence of SEQ ID NO: 27 and a constant heavy CHI chain amino acid sequence of SEQ ID NO: 22; (ii) a variable light chain (VL) amino acid sequence of SEQ ID NO: 126, a variable heavy (VH) chain amino acid sequence of SEQ ID NO: 12S, a constant light (CL) chain amino acid sequence of SEQ ID NO: 27 and a constant heavy CHI chain amino acid sequence of SEQ ID NO: 22; (iii) a variable light chain (VL) amino acid se quence of SEQ ID NO: 9, a variable heavy (VH) chain amino acid sequence of SEQ ID NO: 117, a constant light (CL) chain amino acid sequence of SEQ ID NO: 27 and a constant heavy CHI chain amino acid sequence of SEQ ID NO: 22 ; (iv) a variable light chain (VL) amino acid sequence of SEQ ID NO: 9, a variable heavy (VH) chain amino acid sequence of SEQ ID NO: 16, a constant light (CL) chain amino acid sequence of SEQ ID NO: 27 and a constant heavy CHI chain amino acid sequence of SEQ ID NO: 22; (v) a variable light chain (VL) amino acid sequence of SEQ ID NO: 7 , a variable heavy (VH) chain amino acid sequence of SEQ ID NO: 16, a constant light (CL) chain amino acid sequence of SEQ ID NO: 27 and a constant heavy CHI chain amino acid sequence of SEQ ID NO: 22; (vi) a variable light chain (VL) amino acid sequence of SEQ ID NO: 4, a variable heavy (VH) chain amino acid sequence of SEQ ID NO: 16, a constant light (CL) chain amino acid se quence of SEQ ID NO: 27 and a constant heavy CHI chain amino acid sequence of SEQ ID NO: 22; (vii) a variable light chain (VL) amino acid sequence of SEQ ID NO: 6, a variable heavy (VH) chain amino acid sequence of SEQ ID NO: 15, a constant light (CL) chain amino acid sequence of SEQ ID NO: 27 and a constant heavy CHI chain amino acid se quence of SEQ ID NO: 22; or (viii) a variable light chain (VL) amino acid sequence of SEQ ID NO: 9, a variable heavy (VH) chain amino acid sequence of SEQ ID NO: 120, a constant light (CL) chain amino acid sequence of SEQ ID NO: 27 and a constant heavy CHI chain amino acid sequence of SEQ ID NO: 22.

[0178] In a further embodiment the IgGl antibody portion of the protein construct, which specifically binds to MSLN comprises a full length constant heavy chain IgGl, pref erably having the sequence of SEQ ID NO: 26. [0179] The term "polypeptide capable of binding to CD47" as used herein refers to any polypeptide which specifically binds to CD47. The polypeptide may be an immunoglobu- lin-like polypeptide or it may be an immunoglobulin or an immunoglobulin-based pro tein, e. g. an antibody or an antigen binding fragment thereof. Thus, in a group of specific embodiments of the present invention, the polypeptide capable of binding to CD47 is an immunoglobulin-like domain, preferably, the polypeptide capable of binding to CD47 is an immunoglobulin-based interactor of CD47. In a further group of specific embodi ments the polypeptide capable of binding to CD47 is an immunoglobulin, preferably an antibody or an antigen-binding fragment or sub-form of an antibody binding to CD47. [0180] In a central embodiment of the present invention, the protein construct ac cording to the present invention blocks the interaction of CD47 with signal regulatory protein alpha (SI RPalpha).

[0181] The term "SIRPalpha" as used herein relates to Signal regulatory protein alpha, a cell surface receptor expressed on myeloid cells, hematopoietic stem cells and neurons. SIRPalpha has a cytoplasmic tail that includes several immunoreceptor tyrosine-based inhibitory motifs (ITIMs). Phosphorylation of tyrosines in these motifs upon binding of CD47 to SIRPalpha leads to the recruitment and activation of tyrosine phosphatases SHP-1 and SHP-2, inhibiting the accumulation of myosin-ll at the phagocytic synapse, thereby transmitting inhibitory signals and inhibiting phagocytosis of macrophages. The binding of SIRPalpha to CD47 is mediated by the N-terminal immuno-globulin-like domain of SIRPalpha. Thus, the interaction of SIRPalpha with CD47 protects CD47 positive cells from being engulfed/phagocytosed by macrophages. In addition, SIRPalpha is activated (by phosphorylation on its ITIMs) in response to a variety of mitogenic growth factors and is also involved in adhesion and cell motility. [0182] Human SIRPalpha is also known as Tyrosine-protein phosphatase non-receptor type substrate 1 (SHPS-1), CD172a, Brain Ig-like molecule with tyrosine-based activation motifs (Bit), Macrophage fusion receptor (MFR), MyD-1 antigen or p84. [018B] In humans two variants of SI RPalpha are known: Variant 1 or VI, has the amino acid sequence set out in Genbank as accession number AAH33092 (residues 31-508 constitute the mature form) and is listed in the present invention as SEQ ID NO: 112. Variant 2 or V2 form, differs by 13 amino acids and has the amino acid sequence set out in GenBank as accession number AAH26692 (residues 31-507 constitute the mature form) and is listed in the present invention as SEQ ID NO: 113. These two forms of SIRPalpha constitute about 80% of the forms of SIRPalpha present in humans. Minor forms thereof include for example but not limited to isoforms of variant 1 such as the sequences set out in NM_001040023.1; NM_001040022.1; NM_080792.2; NM_001330728.1; XM_005260670.1; or XM_005260669.1. All mentioned variants or isoforms that are endogenous to humans and have the same property of binding to CD47 and thereby triggering a "don't eat me" signal are comprised in the term "SIRPalpha". The present invention is centered particularly on the variant 1 or VI.

[0184] Amino acids corresponding to positions 1 to 30 of SEQ ID NO: 112 have been identified as signal peptide, amino acids corresponding to positions 31 to 373 of SEQ ID NO: 112 are considered to constitute the extracellular domain of SIRPalpha, amino acids corresponding to positions 374 to 394 of SEQ ID NO: 112 are considered to constitute the transmembrane domain of SIRPalpha, amino acids corresponding to positions 395 to 508 of SEQ ID NO: 112 are considered to constitute the cytoplasmic domain of SIRPalpha. Amino acids corresponding to positions 35 to 145 of SEQ ID NO: 112, which belong to the extracellular domain, are considered to constitute an Ig-like V-type domain; amino acids corresponding to positions 148 to 254 of SEQ ID NO: 112, which belong to the extracellular domain, are considered to constitute an Ig-like Cl-type 1 domain; and amino acids corresponding to positions 266 to 339 of SEQ ID NO: 112, which belong to the extracellular domain, are considered to constitute an Ig-like Cl-type 2 domain. SIRPalpha is typically N-glycosylated at amino acids corresponding to one or more of positions 245, 270, 292, or 319 of SEQ ID NO: 112.

[0185] The term "immunoglobulin-like domain of SIRPalpha" as used herein refers to the Ig-like V-type domain, amino acids corresponding to positions 35 to 145 of SEQ ID NO: 112. In certain preferred embodiments an N- and C-terminally extended Ig-like V- type domain is used comprising amino acids corresponding to positionsBlto 149 ofSEQ ID NO: 112.

[0186] The structure and function of SIRPalpha is described, for example, in Barclay 2009, Curr Opin Immunol. Feb; 21(1): 47-52.

[0187] The "blocking of the interaction of CD47 with SIRPalpha" as used herein gener ally relates to a binding to CD47 by non-native elements or non-native receptors, pref erably by a protein construct according to the present invention, more preferably by a CD47 binding polypeptide being part of a protein construct according to the present invention, which prevents the interaction of native CD47 with native SIRPalpha, e. g. with SIRPalpha present on a macrophage, dendritic cell or neutrophil. The blocking of the interaction of CD47 with SIRPalpha according to the present invention thus relates to an occupation of CD47 by a non-native or artificial or recombinant binding element, e. g. a protein construct according to the present invention, which advantageously com- petes with the binding of native SIRPalpha present on cells and leads to the diminish- ment of the natural CD47-SIRPalpha interaction. Since native SIRPalpha on phagocytes transmits a "don't eat me"-signal upon interaction with CD47, the occupation of CD47 and the corresponding "blocking" of this interaction with a protein construct according to the present invention advantageously leads to a drastic reduction of unoccupied CD47 proteins and, in consequence, to a drastic reduction of a "don't eat me"-signalling through native SIRPalpha on macrophages or other immune cells such as dendritic cells and neutrophils. In particular, cells which, in contrast to tumor cells, do not express MSLN are not affected by the reduction of unoccupied CD47 proteins. Using the protein constructs of the present invention, cellsexpressing MSLN, e. g. tumor cells, will typically exhibit a drastic reduction of unoccupied CD47 proteins and will not be able to transmit "don^'t eat me"-signals since SIRPalpha, present on macrophages or other immune cells such as dendritic cells and neutrophils, will find no, or only a drastically reduced number of unoccupied CD47 binding partners on MSLN expressing cells, i. e. tumor cells. The occupation of CD47 occurs via the polypeptide binding to CD47, preferably an anti-CD47 scFv as defined herein or a SIRPalpha domain as defined herein comprising SEQ ID NO: 21. The occupancy typically occurs in such a way that native SIRPalpha expressed on phagocytes cannot bind to CD47 on tumor cells. However, not the complete binding site of native SIRPalpha needs to be blocked by SIRPalpha or the anti-CD47 scFv within the protein construct of the present invention as the blocking merely requires the preven tion of binding of native SIRPalpha. In certain embodiments, the number of the frag ments binding to CD47, e. g. the presentation of two or more SIRPalpha domains in the protein construct, preferably separated by a polypeptide linker as defined herein, is as sumed to have an augmenting influence on the degree of blockade, e. g. more CD47 protein on the tumor cell is assumed to be occupied by the protein construct and thus cannot transmit a "don't eat me" signal, resulting in an increased engulfment (phagocy tosis) of cells.

[0188] It is particularly preferred that the blocking of the interaction of CD47 with SIRPalpha is provided by the binding of the CD47-binding polypeptide to CD47, e. g. by an immunoglobulin-like polypeptide capable of binding to CD47 or an immunoglobulin- based interactor of CD47, e. g. an antibody or an antigen-binding fragment or sub-form of an antibody binding to CD47, or a CD47-binding immunoglobulin polypeptide.

[0189] According to a specific group of embodiments, the blocking of the interaction of CD47 with SIRPalpha is enhanced and/or reinforced by the concomitant binding of the protein constructs of the present invention via its IgGl antibody to MSLN on MSLN and CD47 double positive cells. The term "enhanced and/or reinforced" as used herein means that the binding of the protein construct according to the present invention to cells expressing and presenting on their cellular surface both MSLN and CD47 is in creased in comparison to cells which express and present on their cellular surface only CD47. The enhancement and/or reinforcement is accordingly caused by an increased avidity of the protein construct. The increased avidity is induced by a multivalent binding to different antigens, e. g. by a concomitant, non-exclusive binding of two antigens or epitopes such as MSLN and CD47, on one cell. [0190] The term "MSLN and CD47 double positive cells" as used herein relates to cells which express and present on their cellular surface the proteins MSLN and CD47, e. g. as defined herein above. The term "avidity" as used herein is known to the skilled person and generally relates to the accumulated strength of multiple monovalent or multivalent affinities of individual non-covalent binding interactions. Individually, each binding in teraction contributes with its affinity to an overall binding strength known as avidity. As such, avidity is distinct from affinity, which describes the strength of monovalent bind ing. The term avidity is further used in connection with tetravalent or multivalent bind ing of the protein construct according to the present invention. The term "monovalent" or "monovalent binding" as used herein refers to the binding strength of an isolated antibody or fragment thereof with one binding site for one single epitope or antigen. The term "multivalent" or "multivalent binding" or "multivalent affinity" as used herein means interactions that result of accumulated monovalent bindings from the same epitope or the accumulation of monovalent bindings from different epitopes, e. g. an anti-MSLN IgGl antibody binding to an epitope on mesothelin with each Fab domain, would be multivalent (in this case bivalent), as it accumulates two monovalent affinities of each Fab domain. Similarly, a protein construct of the present invention comprising an anti- MSLN IgGl antibody and a polypeptide capable of binding to CD47 present on the surface of a tumor cell, accumulates the monovalent affinities of each Fab domain in the anti-mesothelin IgGl antibody plus the monovalent affinities of each polypeptide capable of binding to CD47.

[0191] The protein constructs of the instant specification are therefore multivalent, e. g. tetravalent for their binding to MSLN and CD47, comprising, for example, two antigen binding sites for MSLN and two or more binding sites for CD47. The term "tetravalent" as used herein means the accumulation of four monovalent affinities, irrespective of their specificities, epitopes, affinities and antigens bound by the protein construct of the present specification. Individual binding events may increase the likelihood of other binding interactions to happen, e. g. binding of a MSLN IgGl antibody according to the invention to MSLN and CD47 expressing cells, is assumed to increase the likelihood for the polypeptide specifically binding to CD47 to bind to CD47 due to local increase of concentration of polypeptides binding to CD47. Further information may be derived from suitable literature sources such as Vauquelin and Charlton British, 2013, Journal of Pharmacology 1681771-1785. [0192] As used herein, "binding affinity" describes the ability of a biomolecule such as a polypeptide or a protein of the herein described protein construct, e. g., an antibody to bind a selected target or antigen and form a complex with that target or antigen. Binding affinity is measured by a number of methods known to the skilled person in the art. Methods to measure binding affinities include, but are not limited to, fluorescence titration, enzyme-linked immunosorbent assay (ELISA)-based assays, including direct and competitive ELISA, isothermal titration calorimetry (ITC), and surface plasmon res onance (SPR).

[0193] Avidity may, in certain embodiments, be measured with suitable methods on functional affinity, e. g. flow cytometry (Kiigler et al., 2010 British Journal of Haematol- ogy, 150, 574-586), enzyme-linked immunosorbent assay (ELISA) (Correa et al., 2020, BiomedicalJournal, 10.009) or SPR (Lynch et al., 2014 J Immunol Methods, 404, 1-12). Avidity or accumulation of monovalent affinities or multivalent affinities of molecules such as antibodies, bispecific antibodies or the protein construct of the instant specifi cation can be measured by any method that allows measuring of more than one affinity. Example 2 shows how avidity can be measured in a specific embodiment of the present invention.

[0194] Upon binding of the protein construct according to the present invention to a MSLN and CD47 double positive cell, the presence of the IgGl, which comprises an Fc domain as described above, typically leads to the elimination of the targeted MSLN and CD47 double positive cell(s) e. g. by one of the following processes: antibody-dependent cellular cytotoxicity (ADCC), antibody-dependent cellular phagocytosis (ADCP), and com plement-dependent cytotoxicity (CDC). Typically, the IgGl antibody of the protein con struct according to the invention binds via its Fc domain to FcgRs present on the surface of immune effector cells such as NK cells, monocytes, macrophages, dendritic cells and neutrophils and activates the mentioned immune effector cells to eliminate MSLN and CD47 double positive tumor cells which are bound by the antigen binding domain of the IgGl antibody of the protein construct. This binding provokes the elimination of MSLN and CD47 double positive cells via ADCC, ADCP or other immune cell induced processes such as complement mediated cytotoxicity. Without wishing to be bound by theory it is assumed that in order to escape immune cell recognition, tumor cells may down-regu- late respective tumor associated antigens (TAA) such as e. g. MSLN so that the protein construct cannot bind tumor cells to bring or recruit immune effector cells to tumor cells by the interaction with the Fc part of an immunoglobulin such as IgGl. Using the protein construct of the present invention, macrophages will engulf tumor cells and present tu mor specific antigens to T cells, eliciting a MSLN-independent anti-tumorT cell response, which will eliminate also tumor cells that have potentially downregulated MSLN. The protein construct described herein fulfils at least two major functions: (i) the anti-mes- othelin IgGl together with the polypeptide capable of binding to CD47 target and bind MSLN and CD47 double positive tumor cells and (ii) the Fc domain of the IgGl compo nent binds immune effector cells by the interaction with FcgRs and brings immune ef fector cells in close vicinity to tumor cells while simultaneously activating them. The binding of the Fc domain of the IgGl to FcgR typically results in different biological ac tivities depending on the nature of the immune cell or activated system (e. g NK cell, macrophage, complement, etc.). The immune system is typically activated when anti bodies recognize an antigen, e. g. MSLN, and trigger effector functions through the in teraction with Fc engaging molecules such as FcgRs. The interaction, specifically the in teraction strength and thus the potential strength of effector potential of the IgG and FcRs is dependent on the IgG subclass, allotype, and glycosylation pattern, among other factors. Mostly, FcgRs, the neonatal Fc-receptor (FcRn), TRIM21 (tripartite motif-con taining protein 21), Cl (the first component of the classical complement cascade are involved in effector functions (de Taeye et al., 2019, Antibodies 8(2), 30). The protein construct of the present invention covers different mode of actions and possible activa tion mechanisms of the immune system. Tumor cell specific ADCC is induced by NK cells, which upon activation by binding to the Fc domain of the IgGl of the protein construct bind to the MSLN and CD47 double positive tumor cell, release proteins such as perforins and granzymes which cause the lysis of the tumor cell bound by the protein construct. Tumor cell specific ADCP is induced by macrophages which upon activation by binding to the Fc domain of the IgGl of the protein construct bind to the MSLN and CD47 double positive tumor cell will engulf the nearby tumor cell, a process called phagocytosis. This mechanism is further enhanced by the protein construct of the invention as the protein construct also simultaneously blocks the CD47-SIRPalpha signalling pathway, thereby in hibiting the "don't eat me"-signal normally induced by binding of CD47 to SIRPalpha ex pressed on immune effector cells such as macrophages. The described mechanisms, e. g. ADCC, ADCP and CDC are not mutually exclusive and may occur, depending on the present immune effector cells simultaneously at different sites/locations/organs of the host organism, preferably human.

[0195] In a set of embodiments, the polypeptide which is capable of binding to CD47 present on the surface of a tumor cell and thereby also of blocking the interaction of CD47 with native SIRPalpha, e. g. SIRPalpha present on a macrophage, dendritic cell or neutrophil, is or comprises SIRPalpha, or more preferably a fragment of SIRPalpha, being a part of the protein construct of the invention. Also envisaged are fragments or sub forms of SIRPalpha which have been modified, e. g. by multiplication, preferably dupli cation of a domain or single domains, i. e. Ig-like V-type domain corresponding to amino acids S5-145 of SEQ ID NO 112 or SEQ ID NO:21, which corresponds to amino acids 31- 149 of SEQ ID NO 112 and essentially comprises the Ig-like V-type domain, preferably a duplication of the single domain, more preferably a duplication of the Ig-like V-type do main of SIRPalpha or of SEQ ID NO: 21. The term "fragment of SIRPalpha" as used in the context of a polypeptide which is capable of binding to CD47 accordingly refers to a fragment of SEQ ID NO: 112 as defined herein above. It is particularly preferred that said fragment comprises the Ig-like V-type domain of SIRPalpha, e. g. as defined herein above. It is further particularly preferred that the immunoglobulin-like domain of SIR- Palpha comprises an amino acid sequence which is at least 95%, 96%, 97%, 98%, 99% or 100% identical to the amino acid sequence of SEQ ID NO: 21. "Multiplication", or pref erably "duplication", as used herein means the fusion of at least two fragments compris ing the e. g. the Ig-like V-type domain of SIRPalpha, preferably comprising the amino acid sequence of SEQ ID NO: 21. The single fragment may individually have an amino acid sequence which is at least 95%, 96%, 97%, 98%, 99% or 100% identical to the amino acid sequence of SEQ ID NO: 21. Preferably, the term "fragment of SIRPalpha" thus re lates to a polypeptide comprising the amino acid sequence of SEQ ID NO: 21. If two or more fragments are used, each fragment may have an amino acid with a different de gree of sequence identity with respect to SEQ ID NO: 21. It is preferred that the two or more fragments are identical, i. e. have the same degree of sequence identity.

[0196] The SIRPalpha fragments comprising, for example, each the Ig-like V-type do main of SIRPalpha or the amino acid sequence of SEQ ID NO: 21, are in a typical embod iment fused to each other by a polypeptide linker as defined herein. This linker fulfils the function of connecting or fusing the fragments capable of binding to CD47 of the protein construct and also allows to spatially separate the two or more Ig-like V-type domains . The multiplication of the domain binding to CD47 advantageously allows for a better binding (targeting effect) of CD47-positive tumor cells caused by the avidity ef fect (multivalent binding) of the protein construct, a more effective CD47 blockade on tumor cells as more CD47 polypeptides can be occupied by the protein construct.

[0197] In specific embodiments, the SIRPalpha polypeptide which specifically binds to CD47 is specific for a polypeptide comprising the amino acid sequence of SEQ ID NO: 107 to 110, preferably to of SEQ ID NO: 107. In further specific embodiments a SIRPalpha polypeptide which specifically binds to CD47, is capable of binding an epitope compris ing the amino acid positions 19-124 of SEQ ID NO: 107. Specifically, amino acid positions 19, 21, 22, 24, 45, 47, 48, 53, 54, 55, 57, 64, 67, 115 and 117-124 of SEQ ID NO: 107 are directly or indirectly involved in the binding. [0198] In a further set of embodiments, the polypeptide which is capable of binding to CD47 present on the surface of a tumor cell and thereby also of blocking the interac tion of CD47 with native SIRPalpha, e. g. SIRPalpha present on a macrophage, dendritic cell or neutrophil, is or comprises a suitable anti-CD47 antibody, preferably an anti-CD47 single chain fragment variable (scFv). The term "anti-CD47 single chain fragment varia ble" or "scFv" refers to a fusion protein of the variable regions of the heavy (VH) and light chains (VL) of immunoglobulins, which are typically connected with a polypeptide linker (scFv linker) of about 10 to about 25 amino acids, preferably 15 to 20 amino acids.

[0199] The "scFv linker" may comprise the amino acids glycine and serine, which al lows for a high flexibility and increased solubility, or may additionally or alternatively comprise the amino acids threonine, proline or alanine. The "scFv" linker is a flexible linker which reduces inter alia the likelihood that the linker interferes with the folding and function of the individual domains. It is particularly preferred using an scFv linker comprising, essentially consisting of or consisting of repeats of the amino acid sequence of SEQ ID NO 43 or SEQ ID NO: 44 or SEQ ID NO: 45 or SEQ ID NO: 46 or SEQ ID NO: 47 or SEQ ID NO: 48, preferably 3-5 repeats of the amino acid sequence of SEQ ID NO: 43 and/or SEQ ID NO: 44. The anti-CD47 scFv according to the present invention, being a portion of the protein construct of the present invention, retains the specificity of the original or parent anti-CD47 immunoglobulin, despite removal of the constant regions and the introduction of the polypeptide linker.

[0200] In specific embodiments, the anti-CD47 scFv which specifically binds to CD47 is specific to a polypeptide comprising the amino acid sequence of SEQ ID NO: 107 to 110, preferably to of SEQ ID NO: 107. In further specific embodiments an anti-CD47 scFv which specifically binds to CD47, is capable of binding the extracellular domain of CD47 which refers to amino acids 19-141 of SEQ ID NO: 107. The epitope is similar to or par tially overlapping with the binding site for SIRPalpha. To block the interaction between SIRPalpha on an effector cell and CD47 on a tumor cell, the scFv that specifically binds toCD47 needs to bind the amino acid positions 19-124 of SEQ ID NO: 107 or correspond ing parts thereof. [0201] In preferred embodiments of the present invention the anti-CD47 scFv portion of the protein construct which specifically binds to CD47 comprises a VH region com prising the following CDRs:

(1) a CDRH1 sequence selected from the amino acid sequences of SEQ ID NOs: 49, 50 and 51; and

(2) a CDRH2 sequence selected from the amino acid sequences of SEQ ID NOs: 52 to 62; and

(3) the CDRH3 sequence having the amino acid sequence of SEQ ID NO: 63; or a variant thereof in which one or two or three amino acids in one or more of CDRH1, CDRH2, or CDRH3 are substituted with another amino acid; and comprises a VL region comprising the following CDRs:

(4) a CDRL1 sequence selected from the amino acid sequences of SEQ ID NOs: 64 to 68; and

(5) a CDRL2 sequence selected from the amino acid sequences of SEQ ID NOs: 69 and 70; and

(6) a CDRL3 sequence selected from the amino acid sequences of SEQ ID NOs: 71 to 74, or a variant thereof in which one or two or three amino acids in one or more of CDRL1, CDRL2, or CDRL3 are substituted with another amino acid.

The sequences of the CDRH1, 2, 3 and CDRL1, 2, 3 of the CD47 scFv are provided in Tables 4 and 5, respectively.

Table 4:

Table 5:

[0202] In further preferred embodiments of the protein construct of the present in vention the anti-CD47 scFv portion of the protein construct which specifically binds to CD47 comprises a VH region comprising the following CDRs

(1) the CDRH1 sequence having the amino acid sequence of SEQ ID NO: 49; and

(2) the CDRH2 sequence having the amino acid sequence of SEQ ID NO: 60; and (S) the CDRHS sequence having the amino acid sequence of SEQ ID NO: 63; or a variant thereof in which one or two or three amino acids in one or more of CDRH1, CDRH2, or CDRH3 are substituted with another amino acid; and comprises a VL region comprising the following CDRs:

(4) the CDRL1 sequence having the amino acid sequence of SEQ ID NO: 64; and

(5) the CDRL2 sequence having the amino acid sequence of SEQ ID NO:69; and

(6) the CDRL3 sequence having the amino acid sequence of SEQ ID NO: 72; or a variant thereof in which one or two or three amino acids in one or more of CDRL1, CDRL2, or CDRL3 are substituted with another amino acid. [0203] In a further preferred embodiment of the protein construct of the present in vention the anti-CD47 scFv portion of the protein construct which specifically binds to CD47 comprises a VH region comprising the following CDRs

(1) the CDRH1 sequence having the amino acid sequence of SEQ ID NO: 49; and

(2) the CDRH2 sequence having the amino acid sequence of SEQ ID NO: 52; and

(3) the CDRH3 sequence having the amino acid sequence of SEQ ID NO: 63; and comprises a VL region comprising the following CDRs:

(4) the CDRL1 sequence having the amino acid sequence of SEQ ID NO: 64; and

(5) the CDRL2 sequence having the amino acid sequence of SEQ ID NO: 69; and

(6) the CDRL3 sequence having the amino acid sequence of SEQ ID NO: 71.

[0204] In a further preferred embodiment of the protein construct of the present in vention the anti-CD47 scFv portion of the protein construct which specifically binds to CD47 comprises a VH region comprising the following CDRs

(1) the CDRH1 sequence having the amino acid sequence of SEQ ID NO: 49; and

(2) the CDRH2 sequence having the amino acid sequence of SEQ ID NO: 52; and

(4) the CDRL1 sequence having the amino acid sequence of SEQ ID NO: 64; and

(5) the CDRL2 sequence having the amino acid sequence of SEQ ID NO: 69; and

(6) the CDRL3 sequence having the amino acid sequence of SEQ ID NO: 72.

[0205] In a further group of preferred embodiments of the protein construct of the present invention the anti-CD47 scFv portion of the protein construct which specifically binds to CD47 comprises a variable light chain region amino acid sequence at least about 60%, 65%, 70%, 75%, 80%, 85%, 86%, 87%, 88%, 89%, 90%, 91%, 92%, 93%, 94%, 95%, 96%, 97%, 98%, 99%, or 100% identical to a variable light chain region amino acid se quence as set forth in SEQ ID NO: 75, 76, 77, 78, or 79.

[0206] In a further group of preferred embodiments of the protein construct of the present invention the anti-CD47 scFv portion of the protein construct which specifically binds to CD47 comprises a variable heavy chain region amino acid sequence at least about 60%, 65%, 70%, 75%, 80%, 85%, 86%, 87%, 88%, 89%, 90%, 91%, 92%, 93%, 94%, 95%, 96%, 97%, 98%, 99%, or 100% identical to a variable heavy chain region amino acid sequence as set forth in SEQ ID NO: 80, 81, 82, 83, 84 or 85.

[0207] In a further group of preferred embodiments of the protein construct of the present invention the anti-CD47 scFv portion of the protein construct which specifically binds to CD47 comprises a variable light chain region amino acid sequence at least about 60%, 65%, 70%, 75%, 80%, 85%, 86%, 87%, 88%, 89%, 90%, 91%, 92%, 93%, 94%, 95%, 96%, 97%, 98%, 99%, or 100% identical to a variable light chain region amino acid se quence as set forth in SEQ ID NO: 78; and comprises a variable heavy chain region amino acid sequence at least about 60%, 65%, 70%, 75%, 80%, 85%, 86%, 87%, 88%, 89%, 90%, 91%, 92%, 93%, 94%, 95%, 96%, 97%, 98%, 99%, or 100% identical to a variable heavy chain region amino acid sequence as set forth in SEQ ID NO: 83.

[0208] In a further group of preferred embodiments of the protein construct of the present invention the anti-CD47 scFv portion of the protein construct which specifically binds to CD47 comprises a variable light chain region amino acid sequence at least about 60%, 65%, 70%, 75%, 80%, 85%, 86%, 87%, 88%, 89%, 90%, 91%, 92%, 93%, 94%, 95%, 96%, 97%, 98%, 99%, or 100% identical to a variable heavy chain region amino acid se quence as set forth in SEQ ID NO: 77; and comprises a variable light chain region amino acid sequence at least about 60%, 65%, 70%, 75%, 80%, 85%, 86%, 87%, 88%, 89%, 90%, 91%, 92%, 93%, 94%, 95%, 96%, 97%, 98%, 99%, or 100% identical to a variable heavy chain region amino acid sequence as set forth in SEQ ID NO: 82.

[0209] In a further group of preferred embodiments of the protein construct of the present invention the anti-CD47 scFv portion of the protein construct which specifically binds to CD47 comprises a variable light chain region amino acid sequence at least about 60%, 65%, 70%, 75%, 80%, 85%, 86%, 87%, 88%, 89%, 90%, 91%, 92%, 93%, 94%, 95%, 96%, 97%, 98%, 99%, or 100% identical to a variable light chain region amino acid se quence as set forth in SEQ ID NO: 77; and comprises a variable heavy chain region amino acid sequence at least about 60%, 65%, 70%, 75%, 80%, 85%, 86%, 87%, 88%, 89%, 90%, 91%, 92%, 93%, 94%, 95%, 96%, 97%, 98%, 99%, or 100% identical to a variable heavy chain region amino acid sequence as set forth in SEQ ID NO: 80.

[0210] In a further group of preferred embodiments of the protein construct of the present invention the anti-CD47 scFv portion of the protein construct which specifically binds to CD47 comprises a variable light chain region amino acid sequence at least about 60%, 65%, 70%, 75%, 80%, 85%, 86%, 87%, 88%, 89%, 90%, 91%, 92%, 93%, 94%, 95%, 96%, 97%, 98%, 99%, or 100% identical to a variable light chain region amino acid se quence as set forth in SEQ ID NO: 76; and comprises a variable heavy chain region amino acid sequence at least about 60%, 65%, 70%, 75%, 80%, 85%, 86%, 87%, 88%, 89%, 90%, 91%, 92%, 93%, 94%, 95%, 96%, 97%, 98%, 99%, or 100% identical to a variable heavy chain region amino acid sequence as set forth in SEQ ID NO: 80.

[0211] The sequences of the variable light chain region and of the variable heavy chain region of the anti-CD47 scFv portion of the protein construct and the corresponding an- tibodies are shown in Table 6.

[0212] The sequences of the variable light chain region and of the variable heavy chain region of the anti-CD47 scFv according to the present invention have been selected based on their overall biochemical and biophysical properties, preferably on the affinity of the anti-CD47 scFv polypeptide to CD47. The specific VH and VL combinations have been selected as their affinity to CD47 as measured by SPR have been in the suitable and advantageous range of 100-800 nM, e. g. the 2D6-022 has an affinity of 106 nM, 2D6- 031 has an affinity of 137nM, 2D6-032 has an affinity of 141 nM, 2D6-046 has an affinity of 406 nM, 2D6-056 has an affinity of 659 nM, 2D6-059 has an affinity of 568 nM, 2D6- 088 has an affinity of 460nM. Corresponding measurements are shown in Figure 15 and described in Example 3. In addition, manufacturability, and productivity, developability, immunogenicity, stability and CD47 blocking ability were considered as relevant factors for the selection; see also Example 6.

Table 6:

[0213] In specific embodiments, the VH and VL domains binding to CD47 of the scFv are connected by an scFv linker as described herein to form the polypeptide binding to CD47, e. g. anti-CD47 scFv of the protein construct. [0214] In preferred embodiments the VL and VH domain of the anti-CD47 scFv are in the orientation N-terminus-[VH scFv]-[scFv linke r]-[VL scFv]-C-terminus or N-terminus- [VL scFv]-[scFv linke r]-[VH scFv]-C-terminus, preferably N-terminus-[VH scFv]-[scFv linke r]-[VL scFv]-C-terminus.

[0215] In further preferred embodiments said anti-CD47 scFv, being a portion of the protein construct according to the present invention, has an affinity for CD47 in the range of 100 nM to 2 mM, e. g. an affinity of 100 nM, 200 nM, 300 nM, 400 nM, 500 nM, 600 nM, 700 nM, 800 nM, 900 nM, 1000 nM, 1100 nM, 1200 nM, 1300 nM, 1400 nM, 1500 nM, 1600 nM, 1700 nM, 1800 nM, 1900 nM, or 2000 nM. It is more preferred that the anti-CD47 scFv has an affinity for CD47 in the range of 300 nM to 800 nM, e. g. in the range of 400 nM to 700 nm, 500 nM to 800 nM, 400 nM to 600 nm, 400 nM to 700 nM, or 500 mM to 600 nM.

[0216] In further preferred embodiments said fragment or modified fragment of SIRPalpha, being a portion of the protein construct according to the present invention, has an affinity for CD47 in the range of 100 nM to 2 pM, e. g. an affinity of 100 nM, 200 nM, 300 nM, 400 nM, 500 nM, 600 nM, 700 nM, 800 nM, 900 nM, 1000 nM, 1100 nM, 1200 nM, 1300 nM, 1400 nM, 1500 nM, 1600 nM, 1700 nM, 1800 nM, 1900 nM, or 2000 nM. It is more preferred that the fragment or modified fragment of SIRPalpha has an affinity for CD47 in the range of 300 nM to 800 nM, e. g. in the range of 400 nM to 700 nm, 500 nM to 800 nM, 400 nM to 600 nm, 400 nM to 700 nM, or 500 mM to 600 nM. The fragment of SIRPalpha having the affinity for CD47 as indicated above is preferably the immunoglobulin-like domain of SIRPalpha, more preferably a domain comprising an amino acid sequence which is at least 95%, 96%, 97%, 98% or 99% or 100% identical to the amino acid sequence of SEQ ID NO: 21. The modified fragment of SIRPalpha having the affinity for CD47 as indicated above is preferably a polypeptide that comprises two or more immunoglobulin-like domains of SIRPalpha, more preferably two or more do mains comprising each an amino acid sequence which is at least 95%, 96%, 97%, 98% or 99% identical to the amino acid sequence of SEQ ID NO: 21.

[0217] The affinity may be measured with any suitable methodology known to the skilled person Including fluorescence titration, enzyme-linked immunosorbent assay (ELISA)-based assays, including direct and competitive ELISA, calorimetric methods, such as ITC, and SPR. It is preferred that the affinity is measured by SPR. SPR is a technology which is based on resonant oscillation of conduction electrons at the interface between negative and positive permittivity material stimulated by incident light. SPR spectros copy as envisaged for the definition of affinity values in the context of the present in vention allows to monitor the interaction between molecules in real time. The approach typically involves the attaching of one interacting partner to the surface of a sensor chip (ligand), then passing a sample containing the other interaction partner (analyte) over the surface. Binding of molecules to the sensor surface typically generates a response which is proportional to the bound mass. Binding events can accordingly be followed in real time and interaction characteristics can be determined. "Affinity" as used herein is understood as a value of dissociation constant (KD), half maximal effective concentra tion (EC50), or half maximal inhibitory concentration (IC50) measured using such any of the above mentioned methods, preferably SPR spectroscopy. Generally, a lower KD, EC50, or IC50 value reflects better (higher) binding ability (affinity). [0218] In particular preferred embodiments, the SPR analysis is performed on Biacore devices as commercialized by GE Healthcare Ltd. or Cytiva.

[0219] In specific embodiments the affinity is measured by using standard procedures for SPR spectroscopy, preferably for Biacore devices. Corresponding paramters and technical details would be known to the skilled person or can be derived from suitable literature sources such as Hearty et al., 2012, Methods Mol Biol. ;907:411-42. An exem plary and generalizable SPR spectroscopy measurement is also derivable from Example 2, which provides further details as kits and procedures which may be used for SPR- based affinity measurements. [0220] In preferred embodiments, the anti-MSLN IgGl, i. e. the portion of the protein construct which specifically binds to MSLN, has an affinity for its target MSLN, which in comparison to the affinity of the CD47-binding polypeptide, e. g. the immunoglobulin like domain of SIRPalpha or the anti-CD47 scFv, to its target CD47 is higher by a factor of at least 10, e. g. 10, 12, 15, 17, 20, 25, 27, 30, 32, 35 or more, more preferably by a factor of at least 25, even more preferably by a factor of at least 35, e. g. 40, 50, 60, 70, 80, 90, 100 or more. The affinity for the IgGl portion and the CD47-binding polypeptide portion of the protein construct of the present invention may be measured with any suitable affinity measurement technology. It is preferred that the affinity is determined with SPR as described above. In particularly preferred embodiments, the affinity is determined with the protein construct of the present invention as whole, i. e. comprising both functional domains, i. e. the IgGl anti-mesothelin antibody and the polypeptide binding to CD47, which are contacted with their corresponding targets separately, e. g. either with mesothelin, or with CD47. Particularly, the protein construct of the present specification has been captured to a SPR sensor chip as a whole, preferably a CM5 chip using the human Fc capturing kit, and either recombinant MSLN or CD47 have been used as analyte and measured sequentially. MSLN and CD47 have been used as monomeric domains. [0221] In particularly preferred embodiments the anti-mesothelin IgGl antibody, i. e. the domain(s) of the protein construct which specifically binds to MSLN, has an affinity for its target MSLN, which in comparison to the affinity of one immunoglobulin-like domain of SIRPalpha to its target CD47 is higher by a factor of at least 10, e. g. 10, 12, 15, 17, 20, 25, 27, 30, 32, 35 or more, more preferably by a factor of at least 25, even more preferably by a factor of at least 35, e. g. 40, 50, 60, 70, 80, 90, 100 or more.

[0222] In particularly preferred embodiments the anti-mesothelin IgGl, i. e. the por tion of the protein construct which specifically binds to MSLN, has an affinity for its tar get MSLN, which in comparison to the affinity of anti-CD47 scFv to its target CD47 is higher by a factor of at least 10, e. g. 10, 12, 15, 17, 20, 25, 27, 30, 32, 35 or more, more preferably by a factor of at least 25, even more preferably by a factor of at least 35, e. g.40, 50, 60, 70, 80, 90, 100 or more.

[0223] In further preferred embodiments the CD47 scFv-anti-MSLN protein constructs and SIRPalpha-anti-MSLN protein constructs of the present invention do not bind to red blood cells (RBCs) at concentrations below 50 nM as measured by flow cytometry. The term "no binding" as mentioned herein means that the ratio of the mean fluorescence intensity (MFI) of said CD47 scFv-anti-MSLN and SIRPalpha-anti-MSLN protein con structs to the MFI of the isotype control antibody is below 1.5. Further information can be derived from the Examples and Figures, in particular Example 5 and Figure 31, or from suitable literature sources.

[0224] In a further preferred embodiment said CD47 scFv-anti-MSLN protein con structs and SIRPalpha-anti-MSLN protein constructs of the present invention do not in duce platelet aggregation. The term "not induce platelet aggregation" as mentioned herein means that the anti-CD47 scFV or SIRalha domain in the protein construct does not crosslink platelets. This property is of elevated importance since platelet aggregation can lead to thrombocytopenia or similar diseases and has been observed with some CD47 targeting agents. It is preferred that said CD47 scFv-anti-MSLN protein construct and SIRPalpha-anti-MSLN protein construct does not exceed 20% of platelet aggrega tion, using platelet rich plasma (PRP) and platelet poor plasma (PPP) and where the per centage of platelet aggregation is measured by the absorbance at a wavelength of 595 nm over 30 minutes at 37°C and calculated by the formula: platelet aggregation = [(OD PRP - OD sample)/(OD PRP - OD PPP)] x 100%. For example, a number of about 20%, 15%, 10%, 8%, 6%, 4 %, 2 %, or 1 % or any value in between the mentioned values, of the platelets in a certain volume is aggregated by the CD47 scFv-anti-MSLN protein con structs and SIRPalpha-anti-MSLN protein constructs of ther present invention.

[0225] It is further preferred that the property of platelet aggregation is measured at a concentration of about lOmM to 1000 nM of said CD47 scFv-anti-MSLN protein con structs and SIRPalpha-anti-MSLN protein constructs. It is particularly preferred that the property of platelet aggregation is measured at a concentration of about lOOmM CD47 scFv-anti-MSLN protein constructs and SIRPalpha-anti-MSLN protein constructs. Further information can be derived from the Examples and Figures, in particular Example 7 and Figure 32 or from suitable literature sources.

[0226] In a further even more preferred embodiment said CD47 scFv-anti-MSLN pro tein constructs and SIRPalpha-anti-MSLN protein constructs of the present invention leads to less than 20% platelet aggregation using PRP and PPP, measured by the absorb ance at a wavelength of 595 nm over 30 minutes at 37°C and calculated by the formula: platelet aggregation = [(OD PRP - OD sample)/(OD PRP - OD PPP)] x 100%. , preferably in a concentration range of 10 nM to 1000 nM, more preferably at 100 nM.

[0227] The present invention further envisages that the anti-MSLN IgGl antibody and the polypeptide capable of binding to CD47, e. g. the immunoglobulin-like domain of SIRPalpha or the anti-CD47 scFv, is connected by a polypeptide linker. The polypeptide linker as envisaged by the present invention connects the anti-MSLN IgGl antibody to the immunoglobulin-like domain(s) of SIRPalpha or to the anti-CD47 scFv in the form of a protein fusion, i. e. the immunoglobulin-like domain(s) of SIRPalpha or the anti-CD47 scFv is fused via the linker to an IgGl antibody domain. The exact connection sites at both ends of the linker may vary according to the linker length, the linker sequence, the number of fused domains, e. g. number of immunoglobulin-like domains of SIRPalpha, the orientation of domains, e. g. of the scFv which can be used either as N-terminus-VL- scFv linker-VH-C-terminus or as N-terminus-VH-linker-VL-C-terminus or other factors. [0228] It is preferred that the polypeptide linker is fused to either the N-terminus of the variable light (VL) or the N-terminus of the variable heavy (VH) chain or to the C- terminus of the constant light (CL) or the C-terminus of the constant heavy (CH3) chain domain of the anti-mesothelin IgGl antibody. The corresponding order may hence be for the light chain either N-terminus-scFv-polypeptide linker- VL-CL-C-terminus or N-ter- minus-VL-CL-polypeptide linker-scFv-C-terminus. It is particularly preferred that the pol ypeptide binding to CD47, e. g. the anti-CD47 scFv or SIRPalpha is fused via a polypeptide linkerto the N-terminus of the variable light (VL) chain of the anti-mesothelin IgGl an tibody.

[0229] Besides the function of connecting or fusing the different domain of the protein construct, the polypeptide linker further allows to spatially separate the MSLN- and CD47-binding domains. This advantageously allows for a simultaneous binding of MSLN and CD47, which are present at the cell surface. The density and numbers of MSLN and CD47 on the cell surface varies between different cell lines, tumor cells and cell types.

[0230] In further embodiments the lgGl-CD47-binding domain linker is flexible. The term "flexible" as used herein means that the linker polypeptide has a high degree of conformational freedom, which is assumed to prevent the formation of unwanted sec ondary structures, to reduce the likelihood that the linker interferes with the folding and function of the IgGl and CD47 binding domains and to allow for a spatially variable in teraction with two targets, and simultaneous binding of MSLN and CD47 at the cell sur- face, i. e. to bind to MSLN and CD47 being in different distances to each other.

[0231] The polypeptide linker may have any suitable length and flexibility allowing for a simultaneous binding of the protein construct to MSLN and CD47 and/or allowing for prevention of the formation of unwanted secondary structures, and/or allowing for re duction of the likelihood that the polypeptide linker interferes with the folding and func tion of the IgGl and CD47 binding domains and/or allowing for a spatially variable inter action with the targets MSLN and CD47. A "simultaneous binding" as used herein refers to the state after an interaction of the protein construct with the MSLN target and the CD47 target has taken place, i. e. it means that the protein construct according to the present invention is connected to both targets, MSLN and CD47. It may be based on different dynamics in the binding process, e. g. a temporally first binding of the protein construct to MSLN, followed by a binding to CD47, or vice versa, or a temporally syn chronous binding of MSLN and CD47 by the protein construct.

[0232] In certain embodiments of the present invention the polypeptidelinker com prises or essentially comprises 4 to 40 amino acids, e. g. 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30, 31 , 32, 33, 34, 35, 36, 37,

38, 39 or 40 amino acids. In very specific embodiments, the polypeptide linker may also be longer. In additional embodiments, the polypeptide linker consists of 4 to 40 amino acids, e. g. consists of 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30, 31, 32, 33, 34, 35, 36, 37, 38, 39 or 40 amino acids.

[0233] The polypeptide linker may be composed of any suitable amino acid which ful fils at least one or, preferably, more or all of the above mentioned functions, i. e. fusing the component to the protein construct and spatially separating the components, allow ing simultaneous binding of the protein construct to MSLN and CD47, prevention of the formation of unwanted secondary structures, reduction of the likelihood that the poly peptide linker interferes with the folding and function of the IgGl and CD47 binding domains and allowing for a spatially variable interaction with the targets MSLN and CD47, on the cell surface. According to preferred embodiments, the polypeptide linker comprises, essentially consist of, or consists of the amino acid glycine, alanine, proline, lysine, threonine, aspartic acid, asparagine and/or serine. It is particularly preferred that the polypeptide linker comprises, essentially consist of, or consists of the amino acid glycine and/or serine. For example, the polypeptide linker may comprise 0%, 10%, 20%, 30%, 40%, 50%, 60%, 70%, 80%, 90% or 100% glycine, or 0%, 10%, 20%, 30%, 40%, 50%, 60%, 70%, 80%, 90% or 100% serine. The amino acids may be located at any position in the polypeptide linker and the succession of amino acids may be suitable succession, e. g. starting with glycine, followed by serine etc., or vice versa. Further, the polypeptide linker may comprise stretches of identical amino acids, e. g. stretches of about 5, 10, 15, 20, 25, 30, 35 or more glycines or serines. Further details would be known to the skilled person or can be derived from suitable literature sources such as van Rosmalen et al., 2017, Biochemistry, 56, 6565-6574 or Chen et al, 2013, Adv Drug Deliv Rev, 65(10): 1357-1369. [0234] In further particularly preferred embodiments, the polypeptide linker may comprise, essentially consist of or consist of the amino acid sequence of SEQ ID NOs: 43 to 48 as shown in the following Table 7, or any combination or multiplication thereof:

Table 7:

[0235] For example, the amino acid sequence of Linker 1 (SEQ ID NO: 43) may be pre sent in the polypeptide Linker one time, 2 times, 3 times, 4 times, 5 times, 6 times, 7 times or 8 times. One or more copies of the amino acid sequence of Linker 1 (SEQ ID NO: 43) may further be combined with one or more copies of Linker 2 (SEQ ID NO: 44), Linker 3 (SEQ ID NO: 45), Linker 4 (SEQ ID NO: 46), Linker 5 (SEQ ID NO: 47) or Linker 6 (SEQ ID NO: 48) in a random order. [0236] In a further embodiment the amino acid sequence of Linker 2 (SEQ ID NO: 43) may be present in the polypeptide Linker one time, 2 times, 3 times, 4 times, 5 times, 6 times, 7 times, 8 times, 9 times or 10 times. One or more copies of the amino acid se quence of Linker 2 (SEQ ID NO: 44) may further be combined with one or more copies of Linker 1 (SEQ ID NOs: 43), Linker 3 (SEQ ID NO: 45), Linker 4 (SEQ ID NO: 46), Linker 5

(SEQ ID NO: 47) or Linker 6 (SEQ ID NO: 48).

[0237] In a further embodiment the amino acid sequence of Linker 3 (SEQ ID NO: 45) may be present in the polypeptide Linker one time, 2 times, 3 times, 4 times, 5 times, 6 times or 7 times. One or more copies of the amino acid sequence of Linker 3 (SEQ ID NO: 45) may further be combined with one or more copies of Linker 1 (SEQ ID NOs: 43),

Linker 2 (SEQ ID NO: 44), Linker 4 (SEQ ID NO: 46), Linker 5 (SEQ ID NO: 47) or Linker 6 (SEQ ID NO: 48).

[0238] In a further embodiment the amino acid sequence of Linker 4 (SEQ ID NO: 46) may be present in the polypeptide Linker one time, 2 times, 3 times, 4 times, 5 times, 6 times, 7 times, 8 times, 9 times or 10 times. One or more copies of the amino acid se quence of Linker 4 (SEQ ID NO: 46) may further be combined with one or more copies of Linker 1 (SEQ ID NOs: 43), Linker 2 (SEQ ID NO: 44), Linker 3 (SEQ ID NO: 45), Linker 5

(SEQ ID NO: 47) or Linker 6 (SEQ ID NO: 48).

[0239] In a further embodiment the amino acid sequence of Linker 5 (SEQ ID NO: 47) may be present in the polypeptide Linker two times, 3 times, 4 times, 5 times, 6 times,

7 times, 8 times, 9 times, 10 times, 11 times, 12 times or 13 times. One or more copies of the amino acid sequence of Linker 5 (SEQ ID NO: 47) may further be combined with one or more copies of Linker 1 (SEQ ID NOs: 43), Linker 2 (SEQ ID NO: 44), Linker 3 (SEQ ID NO: 45), Linker 4 (SEQ ID NO: 46) or Linker 6 (SEQ ID NO: 48). [0240] In a further embodiment the amino acid sequence of Linker 6 (SEQ ID NO: 48) may be present in the polypeptide linker two times, 3 times, 4 times, 5 times, 6 times, 7 times, 8 times, 9 times, 10 times, 11 times, 12 times, 13 times, 14 times, 15 times, 16 times, 17 times, 18 times, 19 times or 20 times. One or more copies of the amino acid sequence of Linker 6 (SEQ ID NO: 48) may further be combined with one or more copies of Linker 1 (SEQ ID NOs: 43), Linker 2 (SEQ ID NO: 44), Linker 3 (SEQ ID NO: 45), Linker 4 (SEQ ID NO: 46) or Linker 5 (SEQ ID NO: 47).

[0241] In particularly preferred embodiments the anti-CD47 scFv as defined herein is connected by a flexible polypeptide linker as defined above to the N-terminus of the variable light (VL) domain of the anti-MSLN antibody or the N-terminus of the variable heavy (VH) domain of the anti-MSLN antibody or the C-terminus of the CL domain of the anti-MSLN antibody, oron the C-terminus of the CH3 domain of the anti-MSLN antibody, preferably to the N-terminus of the variable light (VL) domain of the anti-MSLN anti- body.

[0242] Preferably, when the CD47 scFv is fused to the light chain of the anti-MSLN antibody, the domains are organized in the following manner: i) N-terminus-[CD47 scFv]-[polypeptide linker]-[VL^MSLN]-[CL^MSLN]-C-terminus or ii) N-terminus-[VL^MSLN]-[CL^MSLN]-[polypeptide linker]-[CD47 scFv]-C-terminus. [0243] In particularly preferred embodiments the SIRPalpha domain as defined herein, preferably being present in a multiplied format as defined herein (as indicated below by the wording "lx, 2x, or more"), more preferably two domains in tandem repeats, is con nected by the flexible polypeptide linker as defined above to the N-terminus of the var iable light (VL) domain of the anti-MSLN antibody or the N-terminus of the variable heavy (VH) domain of the anti-MSLN antibody or the C-terminus of the CL domain of the anti-MSLN antibody, oron the C-terminus of the CH3 domain ofthe anti-MSLN antibody, preferably to the N-terminus of the variable light (VL) domain of the anti-MSLN anti body. Preferably, the domains are organized in the following manner: i) N-terminus-[SIRPalpha (lx, 2x, or more)]-[polypeptide linker]-[VL^MSLN]- [CL^MSLN]-C-terminus or ii) N-terminus-[VL^MSLN]-[CL^MSLN]-[polypeptide linker]— [SIRPalpha (lx, 2x, or more)]-C-terminus. [0244] Multiplied SI RPalpha domains, preferably two SIRPalpha domains are con nected to each other by a flexible polypeptide linker, e.g. N-terminus-[SIRPalpha]-[pol- ypeptide linke r]— [SI RPalpha]— [polypeptide linker]-C-terminus.

[0245] Figure 2 shows non limiting examples of the structural orientation of the men- tioned domains as envisaged by the present invention.

[0246] In a further aspect the present invention relates toa nucleicacid molecule com prising a polynucleotide encoding the protein construct as defined herein. Also envis aged is a nucleic acid molecule comprising a polynucleotide encoding a fragment of the protein construct, preferably a functional fragment of the protein construct as defined herein, e. g. a fragment fulfilling all functions of the protein construct as defined herein. Further envisaged are nucleic acid molecules encoding fragments or components of the protein construct, e. g. a heavy chain (HC) and a light chain (LC) wherein the encoded fragments or components of the protein construct are not fused via peptide bonds, or via a polypeptide Linker, e. g. different chains of an antibody together forming an IgGl antibody or Fab fragment or Fc domain.

[0247] The term "nucleic acid" or "nucleic acid molecule" as used herein refers to any nucleic acid known to the person skilled in the art, e. g. a polynucleotide like DNA, RNA, single stranded DNA, cDNA, or derivatives thereof. The nucleic acid can further be linear or circular. Preferably, the term refers to DNA polynucleotides. [0248] The nucleic acid molecule comprising a polynucleotide encoding the protein construct may provide any sequence variant which encodes a protein construct as de fine herein, e. g. making use of one or more different codons for an amino acid. It is particularly preferred that the nucleic acid comprises a sequence which has been opti mized to an organism in which it the sequence is planned to be expressed. This "codon- optimization" may be adapted to host organisms according to information on the codon usage in the corresponding organism. Further, codons or codon combinations having an influence on the transcription and/or translation processes, e. g. constituting binding motifs etc., may preferably be avoided in the optimization process. [0249] The nucleic acid molecule comprising a polynucleotide encoding the protein construct may be obtained by any suitable method. For example, suitable protein con structs or fragments thereof, e. g. the IgGl antibody as described herein, may be isolated and sequenced, e. g. by using conventional procedures, to obtain the encoding nucleic acid sequence, or, preferably, the nucleic acid molecule may be synthesized syntheti cally, for instance by using conventional procedures, on the basis of existent nucleic acid sequence information. Furthermore, the nucleic acid molecule may be modified and changed in accordance with antibody modification procedures as described herein, e. g. its sequence may be changed by modifying domain sequences, swapping domain se- quences, combining sequences encoding CDRs and FRs, humanizing the sequences, in serting point mutations etc. Typically, recombinant DNA techniques and procedures as known to the skilled person may be used to generate, modify or optimize the nucleic acid molecule for envisaged purposes such as expression in certain cells or organism etc. Suitable references include Green and Sambrook, Molecular Cloning: A Laboratory Man- ual (4th Edition), Cold Spring Harbor Press, 2012.

[0250] In a further aspect the present invention relates to a vector comprising the nu cleic acid molecule as described above. The term "vector" as used herein refers a nucleic acid molecule that can be used as a vehicle to transfer (heterologous) genetic material into a cell. Such a vehicle may be, for example, a plasmid, a virus, a cosmid, an artificial chromosome, an episome or the like. The vector itself is generally a molecule comprising a nucleotide sequence, typically a DNA sequence that comprises an insert (e. g. a transgene) and a larger sequence that serves as the backbone of the vector. Vectors may encompass additional features besides the transgene insert and a backbone such as one or more promoters, one or more genetic markers, an antibiotic resistance, a reporter gene, a targeting sequence, a protein purification tag.

[0251] In preferred embodiments the vector is an expression vectors, i. e. a vehicle comprising a nucleic acid as defined above, which is specifically designed for the expres sion of the transgene in a target or host cell. An expression vector generally comprises a control sequence such as a promoter sequence that drives expression of the transgene. The term "control sequence" as used herein refers to a DNA sequence nec essary for the expression of an operably linked coding sequence in a particular host or ganism. The control sequences that are suitable for prokaryotes, for example, include a promoter, optionally an operator sequence, and a ribosome binding site. Eukaryotic cells are known to utilize promoters, polyadenylation signals, and enhancers. A nucleic acid is "operably linked" when it is placed into a functional relationship with another nucleic acid sequence. For example, DNA for a pre-sequence or secretory leader is op erably linked to DNA for a polypeptide if it is expressed as a pre-protein that participates in the secretion of the polypeptide; a promoter or enhancer is operably linked to a cod- ing sequence if it affects the transcription of the sequence; or a ribosome binding site is operably linked to a coding sequence if it is positioned so as to facilitate translation. Generally, "operably linked" means that the DNA sequences being linked are contigu ous, and, in the case of a secretory leader, contiguous and in reading phase. However, enhancers do not have to be contiguous. Linking is accomplished by ligation at conven- ient restriction sites. The term "expression" as used herein relates to any step known to the skilled person which is involved in the production of a protein construct according to the invention including, transcription, post- transcriptional modification, translation, post-translational modification, and secretion. It is further preferred that the vector en codes the protein construct according to the invention or a fragment thereof in any suit- able way, e. g. by comprising restriction sites so that domains or sequence fragments can be introduced or removed. Examples of vectors to be used in the context of the present invention include to be used in the context of the present invention include pFUSE-CHIg-hGl, pFUSE-CLIg-hk, pFUSE-CHIg-hG4, pSecTag, pQE70, pQE60, pQE9, pcDNAB.l, pNH8A, pNH16a, pNH18A, pNH46A, pCI-Neo, pCMV, pcDNAB.4, pKK223-3, pKK233-3, pDR540, pRIT5, pET, pGEX-2TK, pGEX-4T, pGEX-5X-l, pMAL, pWLNEO, pSV2CAT, pOG44, and pSG, pGS, pETDuet, pCDFDuet-1, or pRSFDuet-1.

[0252] In a further aspect the present invention relates to a host cell comprising the nucleic acid molecule or the vector of the present invention. The term "host cell" or "target cell" is intended to refer to any individual cell or cell culture that can be or has/have been recipients forvectorsorthe incorporation of exogenous nucleicacid mol ecules, polynucleotides and/or proteins. It also is intended to include progeny of a single cell. The cell may be prokaryotic or eukaryotic, and include bacteria, yeast cells, fungi, insect cells, animal cells, and mammalian cells, e. g., murine, rat, sheep, goat, or human. For example, the protein construct of the invention can be produced in prokaryotes such as bacteria or eukaryotes such as Chinese Hamster Ovarian (CHO) cells or specialized and adapted clones thereof. After expression, the protein construct may be isolated from the host cell and can subsequently be purified through, e. g., affinity chromatog raphy, ion-exchange chromatography and/or size exclusion chromatography. Final puri fication can be carried out similar to the process for purifying antibody expressed e. g, in CHO cells. Particularly preferred host cells are those which allow for the expression of glycosylated protein constructs. Such host cells are typically derived from multicellular organisms. Examples of invertebrate cells include plant and insect cells. For example, the host cell may be a baculoviral strain infected permissive insect cell such as Spodop- tera frugiperda, Aedes aegypti, Aedes albopictus, Drosophila melanogaster, and Bombyx mori. Particularly preferred are vertebrate cells, including mammalian host cell lines. Envisaged examples are monkey kidney CV1 cell line, human embryonic kidney line, baby hamster kidney cells (BHK); Chinese hamster ovary cells (CHO), mouse Sertoli cells, VERO-76 cells, HELA cells, canine kidney cells (MDCK), or human lung cells (W138). Par ticularly preferred are CHO cells or cells with CHO background, e. g. ExpiCHO, CHO- DG44, CHO-K1, or CHO-S. Further preferred are HEK cells or cells with HEK background, e. g. HEK293, HEK293T, Expi293. Also preferred are HighFive cells, Sf9 cells and Sf21 cells.

[0253] In a further aspect the present invention also relates to a host cell expressing the protein construct as defined above. Accordingly, the protein construct of the inven tion can be produced intracellularly, in the periplasmic space, or directly secreted into the medium. If the protein construct is produced intracellularly, as a first step, the par ticulate debris, either host cells or lysed fragments, are removed, for example, by cen trifugation or ultrafiltration. Where the construct is secreted into the medium, superna tants from the expression systems is purified directly from the supernatants from the expression systems by affinity chromatography. Subsequently, the protein construct of the invention prepared from the host cells can be purified using, for example, exchange or size exclusion chromatography, dialysis, and affinity chromatography, with affinity chromatography being the preferred purification technique. [0254] In a further aspect the invention envisages a method of producing the protein construct comprising the cultivation of a host cell, thereby expressing the protein con struct. The term "cultivation" refers to the in vitro maintenance, differentiation, growth, proliferation and/or propagation of cells under suitable conditions in a medium. The medium may, for example, comprise suitable carbon sources such as glucose, dextrose, mannitol, fructose, or mannose, which are provided in a suitable concentration, e. g. between 10 g/L to 150 g/L. The medium may further comprise antibiotics such as G418 sulfate, Zeocin, hygromycin B, puromycin, and blasticidin, neomycin in any suitable con centration. The medium may further have a specific pH and comprise certain amounts of trace elements. Cultivation conditions may further be adapted to the size and form of fermentation or growth. For example, the cultivation may be a batch fermentation process, or a continuous or perfusion growth approach which envisages the continuous addition of fresh media. Further details would be known to the skilled person or can be derived from suitable literature sources such as Rodrigues et al., 2010, Biotechnol Prog, 26(2), 332-51. [0255] The present invention further relates an aspect to the product produced by the method as described above. The product may, for example, have specific form or con formation which is due to the host cell used or activities within said host cell. For exam ple, the product may be specifically glycosylated or be not glycosylated, e. g. if expressed in a mammalian or a prokaryotic host cell, respectively or otherwise contain post-trans- lational modifications. The product may further be provided in different degrees of pu rity, e. g. the product may contain host cell protein or DNA, product degradation prod ucts or product aggregates depending on the purification method used. [0256] In another aspect the present invention encompasses a pharmaceutical com position comprising the protein construct as defined above, or the product as defined above together with a pharmaceutically acceptable carrier or optionally a pharmaceuti cal adjuvant. The term "pharmaceutical composition" as used herein relates to a com- position for administration to a patient, preferably a human patient. The preferred phar maceutical composition of this invention comprises the protein construct of the inven tion. Preferably, the pharmaceutical composition comprises suitable formulations of carriers, stabilizers and/or excipients. In a preferred embodiment, the pharmaceutical composition comprises a composition for parenteral, transdermal, intraluminal, intraar- terial, intrathecal and/or intranasal administration or by direct injection into tissue. It is envisaged that said composition is administered to a patient via infusion or injection.

[0257] The term "pharmaceutically acceptable" as used herein means approved by a regulatory agency or other generally recognized pharmacopeia for use in animals, and more particularly in humans. The term "carrier" refers to a diluent, excipient, or phar- maceutical vehicle with which the protein construct is administered. Such a carrier is pharmaceutically acceptable, i. e. is non-toxic to a recipient at the dosage and concen tration employed. It is preferably isotonic, hypotonic or weakly hypertonic and has a relatively low ionic strength, such as provided by a sucrose solution. Such pharmaceuti cal carriers can be sterile liquids, such as water and oils, including those of petroleum, animal, vegetable or synthetic origin, such as peanut oil, soybean oil, mineral oil, sesame oil and the like. Saline solutions and aqueous dextrose and glycerol solutions can also be employed as liquid carriers. Suitable pharmaceutical excipients include starch, glucose, sucrose, gelatine, malt, rice, flour, chalk, silica gel, sodium stearate, glycerol monos tearate, talc, sodium ion, dried skim milk, glycerol, propylene, glycol, water, ethanol and the like. The composition, if desired, can also contain minor amounts of wetting or emul sifying agents, or pH buffering agents. These compositions can take the form of, e. g., solutions, suspensions, emulsion and the like. Examples of suitable pharmaceutical car riers are described, for example, in "Remington's Pharmaceutical Sciences" by E.W. Mar tin. Generally, the ingredients may be supplied either separately or mixed together in unit dosage form. In a specific embodiment, the pharmaceutical composition is formu lated in accordance with routine procedures as a pharmaceutical composition adapted for intravenous administration to human beings. Typically, compositions for intravenous administration are solutions in sterile isotonic aqueous buffer. Where necessary, the composition may also include a solubilizing agent and a local anesthetic such as ligno- caine to ease pain at the site of the injection. Where the composition is to be adminis tered by infusion, it can be dispensed with an infusion bottle containing sterile pharma ceutical grade water or saline, preferably comprising 0.9% NaCI. Where the composition is administered by injection, an ampoule of sterile water for injection or saline can be provided so that the ingredients may be mixed prior to administration.

[0258] The term "pharmaceutical adjuvant" as used herein relates to additional ingre dients such as chloroquine, protic polar compounds, such as propylene glycol, polyeth ylene glycol, glycerol, EtOH, 1-methyl L-2-pyrrolidone or their derivatives, or aprotic po lar compounds such as dimethylsulfoxide (DMSO), diethylsulfoxide, di-n-propylsulfox- ide, dimethylsulfone, sulfolane, dimethylformamide, dimethylacetamide, tetra- methylurea, acetonitrile or their derivatives. The pharmaceutical adjuvant may further be one or more of a surfactant, wetting agent, dispersing agent, suspending agent, buffer, stabilizer or isotonic agent. The present invention also envisages any suitable pharmaceutical adjuvant as known to the skilled person.

[0259] The pharmaceutical composition of the present invention can also comprise a preservative. Preservatives according to certain compositions of the invention include, without limitation: butylparaben; ethylparaben; imidazolidinyl urea; methylparaben; O- phenylphenol; propylparaben; quaternium-14; quaternium-15; sodium dehydroacetate; zinc pyrithione; polysorbates such as Tween-20 and the like. The preservatives are used in amounts effective to prevent or retard microbial growth. Generally, the preservatives are used in amounts of about 0.1% to about 1% by weight of the total composition with about 0.1% to about 0.8% being preferred and about 0.1% to about 0.5% being most preferred. [0260] The composition of the present invention can be administered to a subject or patient. The term "subject" or "patient" refers to a mammal. "Mammal" as used herein is intended to have the same meaning as commonly understood by one of ordinary skill in the art. Preferred mammals are primates, cows, sheep, goats, horses, dogs, cats, rab- bits, rats, mice and the like. In particularly preferred embodiments, the subject is a hu man.

[0261] The term "administered" means administration of a therapeutically effective dose of the pharmaceutical composition by any suitable route. By "therapeutically ef fective amount" is meant a dose that produces the effects for which it is administered in a patient. The exact dose will depend on the purpose of the treatment, and will be ascertainable by one skilled in the art using known techniques. As is known in the art and described herein, adjustments for systemic versus localized delivery, age, body weight, general health, sex, diet, time of administration, drug interaction and the sever ity of the condition may be necessary and will be ascertainable with routine experimen- tation by those skilled in the art. Administration of the composition may be effected in different ways, e. g., intravenously, intraperitoneally, subcutaneously, intramuscularly, topically or intradermally. In certain embodiments, the present invention provides for an uninterrupted administration of the composition comprising the protein construct. In an example, uninterrupted, i. e. continuous administration may be realized by a pump system. A subcutaneous administration may include a needle or a cannula for penetrat ing the skin of a patient and delivering the suitable composition into the patient's body. The administration may further be transdermal by way of a patch worn on the skin and replaced at intervals.

[0262] In a further aspect the present invention relates to the protein construct as defined above, the product as defined above, or the pharmaceutical composition as mentioned above for use in the treatment of cancer. Also envisaged is a method for the treatment of cancer, wherein said method comprises administering to a patient in need thereof the protein construct as defined above, the product as defined above, or the pharmaceutical composition as mentioned above. [0263] The term "treatment", unless otherwise indicated by context, refers to thera peutic treatment and/or prophylactic measures to prevent the outbreak or relapse of a disease or pathological condition, wherein the objective is to inhibit or slow down (lessen) an undesired physiological condition. For purposes of this invention, beneficial or desired clinical results include, but are not limited to, alleviation of symptoms, dimin- ishment of extent of disease, stabilized (i. e., not worsening) state of disease, delay or slowing of disease progression, amelioration or palliation of the disease state, and re mission (whether partial or total), whether detectable or undetectable. "Treatment" can also mean prolonging survival as compared to expected survival if not receiving treat- ment. Those in need of treatment include those already having the condition or disorder as well as those prone to have the condition or disorder. The treatment may further, in specific embodiments, involve a single administration of a pharmaceutical composition, protein construct or product as defined above, or multiple administrations. A corre sponding administration scheme may be adjusted to the sex or weight of the patient, the disease, the pharmaceutical composition to be used, the general health status of the patient etc. For example, the administration scheme may contemplate an administra tion every 12 h, 24 h, 28 h, 72 h, 96 h, once a week, once every two weeks, once every 3 weeks, once a month etc. Also envisaged are pauses or breaks between administration phases. These regimens can of course be adjusted or changed by the medical practi- tioner in accordance with the patient's reaction to the treatment and/or the course of disease or of the pathological condition.

[0264] The term "cancer" as used herein relates to a pathological process that results in the formation and growth of a cancerous or malignant neoplasm, i. e., abnormal tissue that grows by cellular proliferation, often more rapidly than normal and continues to grow after the stimuli that initiated the new growth cease. Malignant neoplasms typi cally show partial or complete lack of structural organization and functional coordina tion with the normal tissue and most invade surrounding tissues, metastasize to several sites, and are likely to recur after attempted removal and to cause the death of the pa- tient unless adequately treated. The term thus also includes the existence and develop ment of metastases. As used herein, the term "neoplasia" is used to describe all cancer ous disease states and embraces or encompasses the pathological process associated with malignant hematogenous, ascitic and solid tumors. Representative cancers include, for example, stomach, colon, rectal, liver, pancreatic, lung, breast, including triple neg ative breast cancer, cervix uteri, corpus uteri, ovary, prostate, including metastatic pros tate cancer, testis, bladder, renal, head and neck, throat cancer, ascites, mesothelioma, melanoma skin cancer, non-melanoma skin cancer, and kidney cancer. Also envisaged are further cancer forms known to the skilled person or derivable from suitable litera- ture sources such as Pavlopoulou et al., 2015, Oncol Rep., 33, 1, 3-18.

[0265] It is particularly preferred that the cancer is ovarian cancer, ascites, mesotheli oma, triple negative breast cancer, pancreatic cancer, pancreatic adenocarcinoma, non small cell lung cancer (NSCLC), endometrial cancer or biliary extrahepatic cancer.

[0266] The cancer may, in certain embodiments, be a refractory cancer. A cancer may be assumed to be residually present if a subject has undergone surgery as treatment for the cancer. Also envisaged are metastasizing cancerforms, e. g. of the above mentioned cancer forms.

[0267] The following examples and figures are provided for illustrative purposes. It is thus understood that the example and figures are not to be construed as limiting. The skilled person in the art will clearly be able to envisage further modifications of the prin ciples laid out herein. EXAMPLES

Example 1

Generation and purification of protein constructs. [0268] The protein constructs were designed with the aim of targeting both, MSLN and CD47 simultaneously. Therefore, specific anti-MSLN humanized IgGs had to be gen erated first, to which in a second step a domain binding to CD47 was engineered. Protein constructs are schematically shown in Figure 1A-C and the structural orientation is indi cated in Figure 2. [0269] To obtain anti-MSLN antibodies, mice and rats were immunized with the re- combinantly expressed extracellular domain of human MSLN corresponding to AA 269- 606 of SEQID NO: 107. The immunization was followed by selection of hybridoma clones and anti-MSLN antibodies that bound the recombinant human MSLN by ELISA were se lected. The selected antibodies were furthertested for their binding to MSLN-expressing tumor cell lines OVCAR-3 and Suit-2 MSLN cells stably transduced to constitutively ex press MSLN (Suit-2MSLN; Karches et a I, 2019, Clin Cancer Res, 25(19):5890). The variable (Fv) region of the selected antibodies was then sequenced from the hybridoma cells and the CDRs and frame work amino acids were determined.

[0270] Humanized anti-hMSLN antibodies were generated by grafting the CDR se- quences from variable light (VL) and heavy chains (VH) of the anti-hMSLN antibodies into human consensus frameworks derived from a publicly available list of humanized therapeutic antibodies (template antibody sequences). Sequences were selected based on: sequence similarity between VL or VH and the template antibody sequences, CDR length and optimal pairing preferences between VL and VH human germlines. Key amino acids were determined independently for each humanization using publicly available PDB structures of the template antibodies and were retained in the humanized antibod- ies. Furthermore, some additional amino acids were mutated in order to reduce immu- nogenicity of the resulting antibodies. The publicly available Immune Epitope Database (IEBD) MHC Class II binding Prediction Tool (http://tools.iedb.org/mhcii/) was used to predict immunogenicity, more specifically to predict the potential MHC Class II epitopes in our protein constructs. These MHC Class II epitopes are presented to CD4+ T-cells and activate them to elicit an immune response. The algorithm is designed to estimate the affinity between all 9-mer peptides derived from a given protein construct to the MHC class II receptor of different HLA alleles.

[0271] Surprisingly, a single mutation in the CDRH1 aimed at reducing immunogenicity did not result in decreased activity of the corresponding protein construct (see MSL-702 with MSL-705; MSL-712 with MSL-715 and MSL-742 with MSL-745, respectively, in Fig ures 26 and 29).

[0272] The desired VL and VH sequences were generated using custom gene synthesis. The VL was subcloned into the pFUSE2-CLIg-hk vector and the VH into the pFUSE-CHIg- hGl vector. The anti-MSLN antibodies were transiently transfected and expressed in Expi293F cells and analysed for their expression and binding. After this initial character ization, a domain binding to CD47 with low affinity was engineered onto the N- or C- terminus of the anti-MSLN antibody VL or VH chain. The domain binding to CD47 used in the following examples, was comprising either one immunoglobulin-like domain of SI RPalpha (SEQ ID NO: 21) or an anti-CD47 scFv.

[0273] To generate the SIRPalpha-anti-MSLN protein construct, the N-terminal Ig-like V-type domain of SIRPalpha (SEQ ID NO: 21) was synthesized using custom gene synthe sis and subcloned into the N-terminus of the anti-MSLN LC domain together with a gly- cine/serine-rich linker. Anti-CD47 scFv-anti-MSLN protein constructs were generated by cloning an anti-CD47 scFv to the N-terminus of the anti-MSLN LC domain together with a glycine-/serine-rich linker. Anti-CD47 scFvs were generated by subcloning the VH and the VL domains of an anti-CD47 antibody connected with a (G4S)3 linker in a single pol ypeptide. The corresponding plasmids were transiently transfected into Expi293F cells or other expression cell lines for protein expression. After five to seven days, the cell culture supernatant was harvested and protein constructs were purified by protein A affinity chromatography. Size exclusion chromatography (SEC) of the purified molecules was performed using a Superdex 200 increase 10/S00 column in phosphate-buffered saline (PBS) and the protein constructs were analyzed by 4-20 % SDS-PAGE under re ducing conditions and visualized by Coomassie Brilliant Blue staining (see Figure 3 and Figure 18).

[0274] All proteins were well expressed and could be purified to at least 95% purity as shown in an exemplary analytical SEC (see Figure 4) and used for subsequent functional assays.

Example 2

The protein constructs bind to MSLN

[0275] Surface Plasmon Resonance (SPR) was used to analyse binding of the different protein constructs to recombinantly expressed MSLN and to determine the affinity (see Figures 5, 6, 7, 19, 20 and 21).

[0276] Association (K_on) and dissociation (K_0ff) rate constants are determined by SPR using a Biacore X-100 (Cytiva). Using the "Human antibody capture kit" an anti-human Fc antibody is immobilized on a CM5 sensor chip in a covalent manner using the "Amine coupling Kit" according to manufacturer's instructions.

[0277] Humanized anti-MSLN IgGs are injected as ligand to reach a surface coverage of approximately 100 response units (RU) or 400 RU for low affinity interactions. Recom binantly expressed human MSLN extracellular domain is used as analyte and injected in increasing concentrations (e. g. from S.9 to lOOOnM), with standard association time of 120s and dissociation time of 600s. Experiments are performed in multi-cycle kinetics runs, with regeneration between sample injection with the respective regeneration buffer (SM MgCI2 for the human antibody capture kit). All reagents are diluted in HBS- EP+ buffer (Cytiva). Data are fit to a simple 1:1 interaction model using the global data analysis option available within BiaEvaluation 2.0.1 software. The KD is obtained from the ratio between K_on and K_0ff-

[0278] Binding to MSLN was in an affinity range expected for monoclonal antibodies (KD below 20nM). In addition, binding of the different protein constructs to OVCAR-3 tumor cells or Suit-2MSLN cells, both double positive for human MSLN and CD47 (MSLN⁺/CD47⁺), was confirmed by flow cytometry and is shown in Figure 8, 9, 22 and 23. Importantly, in this setting the avidity-dependent binding (accumulated binding strength or affinity of multiple binding interactions) is measured, as OVCAR-3 tumor cells express both, MSLN and CD47. Indeed, the fusion of SIRPalpha or of an anti-CD47 scFv to the anti-MSLN IgGl results in increased binding to OVCAR-3 cells at high antibody concentrations compared to control molecules (see Figures 9 and 23).

Example 3

Protein constructs bind to CD47 with low affinity and are able to block the interaction of CD47 with SIRPalpha [0279] Surface Plasmon Resonance (SPR), was also used to determine the affinity of the SIRPalpha-anti-MSLN and the anti-CD47 scFv-anti-MSLN protein constructs to hu man CD47. The procedure was the same as described above for MSLN binding, but in this case the hCD47 extracellular domain was used as analyte. As expected, binding to CD47 occurs with low affinity (KD from 100 to 1000 nM) and for SIRPalpha is in accord- ance with previously measured affinities of wildtype SIRPalpha for CD47 (Hatherley et al., 2007, Biol Chem , 282(19), 14567) and is in a similar range forthe isolated low-affinity anti-CD47 scFv (SEQ ID NOs: 80 and 75; SEQ ID NOs: 85 and 114; SEQ ID NOs: 85 and 115; SEQ ID NOs: 80 and 76; SEQ ID NOs: 82 and 76; SEQ ID NOs: 81 and 77; SEQ ID NOs: 83 and 78; SEQ ID NOs: 84 and 79)). Examples of binding curves are shown in Figure 15, 16 and 17 and an overview of the KD of different protein constructs to MSLN as well as

CD47 obtained by SPR and by flow cytometry using OVCAR-3 cells is shown in Table 8.

Table 8: Overview on KD measurements by SPR and flow cytometry (N.a. not applica- bale, N.d. not determined).

[0280] In order to inhibit the CD47 immune checkpoint, the protein constructs not only need to bind CD47, but they also need to block the interaction of CD47 with SIRPal- pha. To assess this, a CD47-SIRPalpha competitive binding assay by SPR was used. A hSIRPalpha-mouse Fc fusion construct was used as ligand onto an anti-mouse Fc sensor chip and a hCD47ex (extracellular domain of human CD47)-human Fc fusion construct was used as analyte. Increasing concentrations of SIRPalpha-anti-MSLN or anti-CD47 scFv-anti-MSLN-were mixed together with a fixed concentration of hCD47ex-hFc, result ing in a dose-dependent reduction of SPR signal, consistent with an inhibition of the binding between hSIRPalpha-mFc and hCD47ex-hFc. [0281] This demonstrates that the SIRPalpha-anti-MSLN and the anti-CD47 scFv-anti- MSLN protein constructs are indeed able to block the CD47-SIRPalpha immune check point interaction (see Figure 10 and 24). This was also confirmed by flow cytometry using OVCAR-3 tumor cells that are double positive for MSLN and CD47 (see Figure 11 and 25). For this purpose, OVCAR-3 tumor cells, double positive for MSLN and CD47 were incu- bated with various concentrations of SIRPalpha-anti-MSLN or anti-CD47 scFv-anti-MSLN protein constructs, or control antibodies for 30min at room temperature (RT). Without washing, a fluorescently labelled SIRPalpha construct was subsequently incubated with the cells, and incubated for an additional 30min at RT. The signal is subsequently meas ured by flow cytometry. The fluorescently labelled SIRPalpha construct binds only to CD47 sites that are not already occupied by the protein constructs. A decrease in signal therefore corresponds to blocking of CD47 by the respective protein construct. Example 4

Protein constructs eliminate tumor cells by multiple modes of action

[028B] The purpose of the engineered protein constructs is to eliminate tumor cells and lower tumor burden in patients. To test the ability of the protein constructs to kill tumor cells, two types of in vitro assays using tumor cell lines double positive for MSLN and CD47 were performed. First an antibody-dependent cellular cytotoxicity (ADCC) as say, to measure tumor cell lysis by natural killer (NK) cells, and second an antibody-de- pendent cellular phagocytosis assay (ADCP), to measure tumor cell phagocytosis by monocyte-derived macrophages were performed. ADCC is dependent only on the IgGl domain and neither binding to nor blocking of CD47 does contribute to the ADCC assay as SI RPalpha is not expressed on NK cells.

[0284] For the ADCC assay the human tumor cell lines double positive for MSLN and CD47 Suit-2 MSLN (see Figure 12 and 28) or OVCAR-3 (see Figure 13, 26 and 27) were used as target cells and were first labelled with carboxyfluorescein succinimidyl ester (CFSE) and subsequently incubated for 3h or 4h with human NK cells (effector cells), freshly isolated from peripheral blood of healthy donors in the presence of 500 fM to 50 nM of the respective protein constructs. The effector-to-target (E:T) cell ratio was 4:1 or 5:1. NK cell-mediated cell lysis was assessed by live/dead cell staining and subsequent flow cytometric analysis. Cell killing is depicted as percent dead target cells of total tar get cells.

[0285] For the ADCP assay (see Figure 14, 29 and 30) monocytes were first isolated from the peripheral blood of healthy donors and subsequently differentiated to macro phages for 5 to 7 days in presence of 100 ng/ml recombinant hM-CSF. On the day of the experiment, macrophages (effector cells) were labelled with CellTrace Calcein red/or ange AM, and the tumor cell line OVCAR-3 (target cells), double positive for MSLN and CD47, was labelled with CFSE. Macrophages and target cells were incubated at an effec- tor-to-target (E:T) ratio of 1:1 for 2h or 3h in the presence of 500 fM to 50 nM of the respective protein constructs. Macrophages were detached using 1 mM EDTA in PBS and optionally incubated for additional lh. Phagocytosis was analysed by flow cytometry and depicted as the ratio between phagocytosing double positive macrophages (CFSE+/Cal- cein red/orange AM+) and total macrophages (Calcein red/orange AM+).

[0286] It was confirmed that all protein constructs activate NK cells and induce specific and selective tumor cell lysis via binding to MSLN with high affinity. Most importantly, the fusion of the SIRPalpha domain or the anti-CD47 scFv did not alter or impair the ability of the anti-MSLN IgGl to kill tumor cells by ADCC but lead to even increased ADCC (see Figure 12, 13, 26, 28). Furthermore, Figure 27 shows that the Fc can be engineered to increase ADCC by mutations in the CH2 domain (Fc enhanced). An anti-MSLN-lgGl Fc enhanced mAb (MSL-709) displays the same level of ADCC as the SIRPalpha-anti-MSLN protein construct (MSL-715). Enhancing the Fc of the SIRPalpha-anti-MSLN protein con- struct (MSL-719), increases the ADCC even further. Additionally, all protein constructs induced strong phagocytosis of the MSLN and CD47 double positive tumor cell line OVCAR-3, relying on both, binding of MSLN as well as blocking the interaction of CD47 with SIRPalpha (see Figure 14, 29 and 30).

[0287] Figures 26 and 29 also show that, surprisingly, the CDR mutation introduced to reduce immunogenicity does not alter the function of the protein constructs (compare MSL-702 and MSL-705, MSL-712 and MSL-715 as well as MSL-742 and MSL-745, respec tively).

Example 5

Protein constructs have reduced binding affinity to cells not expressing MSLN

[0288] Since CD47 is expressed on all cells in the body, including red blood cells (RBCs), targeting CD47 with an hlgGl antibody can be a major concern. To determine if the SIR- Palpha-anti-MSLN and the anti-CD47 scFv-anti-MSLN protein constructs bind to cells that express only CD47 but not MSLN, such as RBCs, binding to RBCs was tested by flow cytometry. RBCs were isolated from peripheral blood of healthy donors by 3 cycles of centrifugation and washing with Phosphate Buffer Saline (PBS). 150,000 RBCs were stained per well of a 96-well plate with the indicated protein constructs for 30 minutes on ice, followed by staining with a fluorescently-labelled secondary antibody detecting the hlgGl for 30 minutes on ice, and final detection by flow cytometry. Dots represent raw data, curves are non-linear fit of the specific signal with a Hill slope of 1.

[0289] Due to the low affinity to CD47, the SIRPalpha-anti-MSLN and the anti-CD47 scFv-anti-MSLN protein constructs are expected to display very little binding to RBCs from healthy donors, as compared to a high affinity anti-CD47 mAb. Figure 31 shows that indeed the protein constructs of the present invention do not bind RBCs or only minimally at the highest tested concentration of luM. In contrast and as expected a high affinity anti-CD47 mAb used as control strongly binds to CD47 expressing RBCs.

Example 6

Selection of protein constructs binding to MSLN and polypeptides binding to CD47

[0290] All herein mentioned proteins constructs undergo a selection progress in which they need to pass certain selection criteria. First, the amino acid sequence of the anti bodies and fragments thereof are humanized. After humanization, proteins constructs need to be obtained in a certain quantity, e.g. at least 2 mg/L using the Expi293 expres sion system before further analysis can occur. Some protein constructs might not pass that selection criteria due to very low protein expression (low expressers) and are aban- doned. Some "low expressers" might be further mutated in their amino acid sequence which can result in higher protein construct yields. Protein constructs, antibodies and fragments thereof with suitable expression are further analysed for their binding to their specific targets, e.g. MSLN and/or CD47 by surface plasmon resonance (SPR) and/or flow cytometry. From here binding and dissociation constant (KD) values can be obtained representing the protein construct affinity and/or avidity. If the desired affinity e.g. 300- 800 nM is not obtained the protein constructs are further mutated/engineered in their CDR or framework region. Mutation is followed by expression and binding tests. It can be an iterative process until all desired selection criteria are fulfilled.

[0291] The following criteria are important: degree of humanization, expression and manufacturability/producibility, immunogenicity, binding affinity measured by SPR, binding to cells, protein construct stability and biological activity such as blocking, ADCC or ADCP. Example 7

CD47 scFv-anti-MSLN protein and SIRPalpha-anti-MSLN protein constructs do not in duce aggregation of platelets (PLTs)

[0292] CD47 is expressed on platelets (PLTs), also known as thrombocytes. Aggrega tion of PLTs can cause thrombocytopenia, a condition where the PLT count in the blood is too low. Thrombocytopenia is especially dangerous if associated with internal bleed ing. Some drugs targeting CD47 have been reported to induce thrombocytopenia in pa- tients. We assessed PLT aggregation in vitro by isolating PLTs from healthy donors. PLTs or platelet-rich-plasma (PRP) was isolated from 20-S0 ml blood from healthy donors, drawn in a plastic syringe containing 1:10 volume CPD (citrate-phosphate-dextrose) and purified according to state-of-the-art protocols according to Abeam. Isolated PLTs were gently mixed with a concentration range of a CD47 scFv-anti-MSLN protein construct (MSL-745) and of a SIRPalpha-anti-MSLN protein construct (MSL-715) and aggregation was assessed by an absorbance measurement at 595 nm wavelength every 15 seconds over 30 minutes at 37°C under shaking condition. The percentage of aggregation was calculated with reference to the absorbance of PRP incubated with PBS and platelet- poor-plasma (PPP) according to the formula: platelet aggregation = [(OD PRP - OD sam- ple)/(OD PRP - OD PPP)] x 100%.

[0293] Figure 32 illustrates that the CD47 scFv-anti-MSLN protein construct (MSL-745) and the SIRPalpha-anti-MSLN protein construct (MSL-715) do not cause PLT aggregation in contrast to the positive controls (anti-CD47 IgGl and anti-CD47 lgG4).

Claims

1. A protein construct comprising (i) an anti-mesothelin (MSLN) IgGl antibody or an antigen-binding fragment thereof and (ii) a polypeptide capable of binding to CD47 present on the surface of a tumor cell.

2. The protein construct of claim 1 wherein the protein construct efficiently binds MSLN expressing tumor cells and blocks the interaction of CD47 with signal regulatory protein alpha (SIRPalpha).

S. The protein construct of claim 2, wherein the blocking of the interaction of CD47 with SIRPalpha is provided by the binding of the CD47-binding polypeptide to CD47.

4. The protein construct of claim 3, wherein said binding to CD47 is enhanced and/or reinforced by the protein construct's concomitant binding of the anti MSLN IgGl antibody or antigen-binding fragment thereof to MSLN on MSLN and CD47 double positive cells.

5. The protein construct of claim 4, wherein said enhancement and/or reinforcement is caused by an increased avidity of the protein construct caused by multivalent binding to different antigens.

6. The protein construct of claim 4 or 5, wherein said anti MSLN IgGl antibody or antigen-binding fragment thereof induces the recruiting of immune cells to MSLN and CD47 double positive cells, thereby provoking the elimination of MSLN and CD47 double positive cells.

7. The protein construct of any one of claims 1 to 6, wherein said anti MSLN IgGl antibody or antigen-binding fragment thereof comprises: a variable heavy chain complementarity determining region 1 (CDRH1) sequence selected from the amino acid sequences of SEQ ID NOs: 28, 86,

87, 88 and 122; a variable heavy chain complementarity determining region 2 (CDRH2) sequence selected from the amino acid sequences of SEQ ID NOs: 29, 30,

31, 89, 90 and 91; a variable heavy chain complementarity determining region 3 (CDRH3) sequence selected from the amino acid sequences of SEQ ID NOs: 32, 33,

92, 93 and 94; a variable light chain complementarity determining region 1 (CDRL1) sequence selected from the amino acid sequences of SEQ ID NOs: 34, 35,

95, 96 and 97; a variable light chain complementarity determining region 2 (CDRL2) sequence selected from the amino acid sequences of SEQ ID NOs: 36 to 39, 98, 99 and 121; and a variable light chain complementarity determining region 3 (CDRL3) sequence selected from the amino acid sequences of SEQ ID NOs: 40, 41, 100, 101 and 102.

8. The protein construct of claim 7, wherein said anti MSLN IgGl antibody or antigen-binding fragment thereof comprises (i) a variable heavy chain complementarity determining region 1 (CDRH1) sequence of SEQ ID NO: 122 and a variable heavy chain complementarity determining region 2 (CDRH2) sequence of SEQ ID NO: 90 and a variable heavy chain complementarity determining region 3 (CDRH3) sequence of SEQ ID NO: 93; and a variable light chain complementarity determining region 1 (CDRL1) sequence of SEQID NO: 96 and a variable light chain complementarity determining region 2 (CDRL2) sequence of SEQ ID NO: 99 and a variable light chain complementarity determining region 3 (CDRL3) sequence of SEQ ID NO: 101; (ii) a variable heavy chain complementarity determining region 1 (CDRH1) sequence of SEQ ID NO: 87 and a variable heavy chain complementarity determining region 2 (CDRH2) sequence of SEQ ID NO: 90 and a variable heavy chain complementarity determining region 3 (CDRH3) sequence of SEQ ID NO: 93; and a variable light chain complementarity determining region 1 (CDRL1) sequence of SEQ ID NO: 96 and a variable light chain complementarity determining region 2 (CDRL2) sequence of SEQ ID NO: 99 and a variable light chain complementarity determining region 3 (CDRL3) sequence of SEQ ID NO: 101; or (iii) a variable heavy chain complementarity determining region 1 (CDRH1) sequence of SEQ ID NO: 28 and a variable heavy chain complementarity determining region 2 (CDRH2) sequence of SEQ ID NO: 31 and a variable heavy chain complementarity determining region 3 (CDRH3) sequence of SEQ ID NO: 32; and a variable light chain complementarity determining region 1 (CDRL1) sequence of SEQ ID NO: 34 and a variable light chain complementarity determining region 2 (CDRL2) sequence of SEQ ID NO:

38 and a variable light chain complementarity determining region 3 (CDRL3) sequence of SEQ ID NO: 40

9. The protein construct of any one of claims 1 to 8, wherein said anti MSLN IgGl antibody or antigen-binding fragment thereof comprises a variable light chain amino acid sequence selected from the amino acid sequences of SEQ ID NO: 3 to 11 or 126 and a variable heavy chain_selected from the amino acid sequences of SEQ ID NO: 14 to 20 or 116 to 120 or 123 to 125.

10. The protein construct of any one of claims 1 to 9, wherein said anti MSLN

IgGl antibody or antigen-binding fragment thereof comprises:

(i) a variable light chain amino acid sequence of SEQ ID NO: 126 and a variable heavy chain amino acid sequence of SEQ ID NO: 123, 124 or 125; or

(ii) a variable light chain amino acid sequence of SEQ ID NO: 9 and a variable heavy chain amino acid sequence of SEQ ID NO: 16, 116, 117, 118, 119 and 120; or

(iii) a variable light chain amino acid sequence of SEQ ID NO: 3 and a variable heavy chain amino acid sequence of SEQ ID NO: 14, 15, 16, or 17; or

(iv) a variable light chain amino acid sequence of SEQ ID NO: 4 and a variable heavy chain amino acid sequence of SEQ ID NO: 16; or

(v) a variable light chain amino acid sequence of SEQ ID NO: 5 and a variable heavy chain amino acid sequence of SEQ ID NO: 16; or

(vi) a variable light chain amino acid sequence of SEQ ID NO: 6 and a variable heavy chain amino acid sequence of SEQ ID NO: 14, 15, 16 or 17; or

(vii) a variable light chain amino acid sequence of SEQ ID NO: 7 and a variable heavy chain amino acid sequence of SEQ ID NO: 16; or

(viii) a variable light chain amino acid sequence of SEQ ID NO: 8 and a variable heavy chain amino acid sequence of SEQ ID NO: 16; or

(ix) a variable light chain amino acid sequence of SEQ ID NO: 10 and a variable heavy chain amino acid sequence of SEQ ID NO: 18, 19 or 20; or

(x) a variable light chain amino acid sequence of SEQ ID NO: 11 and a variable heavy chain amino acid sequence of SEQ ID NO: 18, 19 or 20.

11. The protein construct of any one of claims 1 to 10, wherein said anti MSLN IgGl antibody or antigen-binding fragment thereof comprises a variable light chain amino acid sequence of SEQ ID NO: 9 and a variable heavy chain amino acid sequence of SEQ ID NO: 117.

12. The protein construct of any one of claims 1 to 11, wherein said anti MSLN IgGl antibody or antigen-binding fragment thereof comprises a variable light chain amino acid sequence of SEQ ID NO: 126 and a variable heavy chain amino acid sequence of SEQ ID NO: 125.

IB. The protein construct of any one of claims 1 to 12, wherein said anti MSLN IgGl antibody or antigen-binding fragment thereof comprises a light chain CL domain and a heavy chain CHI, CH2 and CH3 domain, preferably a human light chain CL domain and a human heavy chain CHI, CH2, CH3 domain; and a hinge domain, preferably a human hinge region.

14. The protein construct of claim 13, wherein said light chain CL domain comprises an amino acid sequence which is at least 95%, 96%, 97%, 98% or 99% identical to the amino acid sequence of SEQ ID NO: 27.

15. The protein construct of claim 13 or 14, wherein said heavy chain CHI domain comprises an amino acid sequence which is at least 95%, 96%, 97%, 98% or 99% identical to the amino acid sequence of SEQ ID NO: 22.

The protein construct of any one of claims 13 to 15, wherein said heavy chain CH2 domain comprises an amino acid sequence which is at least 95%, 96%, 97%, 98% or 99% identical to the amino acid sequence of SEQ ID NO: 23.

17. The protein construct of any one of claims 13 to 16, wherein said heavy chain CH3 domain comprises an amino acid sequence which is at least 95%, 96%, 97%, 98% or 99% identical to the amino acid sequence of SEQ ID NO: 24.

18. The protein construct of any one of claims 15 and 17, wherein said heavy chain CH2 domain and said heavy chain CH3 domain form an Fc domain.

19. The protein construct of claim 18, wherein said Fc domain comprises an amino acid sequence which is at least 95%, 96%, 97%, 98% or 99% identical to the amino acid sequence of SEQ ID NO: 42.

20. The protein construct of any one of claims 13 to 19, wherein said hinge domain comprises the amino acid sequence of SEQ ID NO: 25.

21. The protein construct of any one of claims 13 to 15, wherein a Fab domain of the protein construct comprises: (i) a variable light chain (VL) amino acid sequence of SEQID NO: 126, a variable heavy (VH) chain amino acid sequence of SEQ ID NO: 125, a constant light (CL) chain amino acid sequence of SEQ ID NO: 27 and a constant heavy CHI chain amino acid sequence of SEQ ID NO: 22; (ii) a variable light chain (VL) amino acid sequence of SEQ ID NO: 126, a variable heavy (VH) chain amino acid sequence of SEQ ID NO: 123, a constant light (CL) chain amino acid sequence of SEQ ID NO: 27 and a constant heavy CHI chain amino acid sequence of SEQ ID NO: 22; (iii) a variable light chain (VL) amino acid sequence of SEQ ID NO: 9, a variable heavy (VH) chain amino acid sequence of SEQ ID NO: 117, a constant light (CL) chain amino acid sequence of SEQ ID NO: 27 and a constant heavy CHI chain amino acid sequence of SEQ ID NO: 22 ; (iv) a variable light chain (VL) amino acid sequence of SEQ ID NO: 9, a variable heavy (VH) chain amino acid sequence of SEQ ID NO: 16, a constant light (CL) chain amino acid sequence of SEQ ID NO: 27 and a constant heavy CHI chain amino acid sequence of SEQ ID NO: 22; (v) a variable light chain (VL) amino acid sequence of SEQ ID NO: 7 , a variable heavy (VH) chain amino acid sequence of SEQ ID NO: 16, a constant light (CL) chain amino acid sequence of SEQ ID NO: 27 and a constant heavy CHI chain amino acid sequence of SEQ ID NO: 22; (vi) a variable light chain (VL) amino acid sequence of SEQ ID NO: 4, a variable heavy (VH) chain amino acid sequence of SEQ ID NO: 16, a constant light (CL) chain amino acid sequence of SEQ ID NO: 27 and a constant heavy CHI chain amino acid sequence of SEQ ID NO: 22; (vii) a variable light chain (VL) amino acid sequence of SEQ ID NO: 6, a variable heavy (VH) chain amino acid sequence of SEQ ID NO: 15, a constant light (CL) chain amino acid sequence of SEQ ID NO: 27 and a constant heavy CHI chain amino acid sequence of SEQ ID NO: 22; or (viii) a variable light chain (VL) amino acid sequence of SEQ ID NO: 9, a variable heavy (VH) chain amino acid sequence of SEQ ID NO: 120, a constant light (CL) chain amino acid sequence of SEQ ID NO: 27 and a constant heavy CHI chain amino acid sequence of SEQ ID NO: 22.

22. The protein construct of any one of claims 1 to 21, wherein said polypeptide capable of binding to CD47 present on the surface of a tumor cell comprises at least one immunoglobulin-like domain of SIRPalpha or comprises an anti- CD47 single chain fragment variable (scFv).

23. The protein construct of claim 22, wherein said polypeptide comprises one or two copies of the immunoglobulin-like domain of SIRPalpha.

24. The protein construct of claim 22 or 23, wherein said immunoglobulin-like domain of SI RPalpha comprises an amino acid sequence which is at least 95%, 96%, 97%, 98% or 99% identical to the amino acid sequence of SEQ ID NO: 21.

25. The protein construct of claim 22 to 24, wherein said immunoglobulin-like domain of SIRPalpha has an affinity for CD47 measured by surface plasmon resonance (SPR) in the range of 100 nM to 2 pM, preferably in a range of 300 nM to 800 nM.

26. The protein construct of claim 22, wherein said anti-CD47 scFv has an affinity for CD47 measured by surface plasmon resonance (SPR), in the range of 100 nM to 2 mM, preferably in the range of 300 nM to 800 nM.

27. The protein construct of any one of claims 1 to 26 wherein the anti MSLN IgGl antibody or antigen-binding fragment thereof has an affinity to its target MSLN which in comparison to the affinity of the CD47-binding polypeptide to its target CD47 is higher by a factor of at least 10, preferably at least 25, more preferably at least 35.

28. The protein construct of claim 22, 26 or 27 wherein said anti-CD47 scFv comprises: a variable heavy chain complementarity determining region 1 (CDRH1) se quence selected from the amino acid sequences of SEQ ID NOs: 49 to 51, preferably SEQ ID NO: 49; a variable heavy chain complementarity determining region 2 (CDRH2) se quence selected from the amino acid sequences of SEQ ID NOs: 52 to 62, preferably SEQ ID NO: 60; a variable heavy chain complementarity determining region 3 (CDRH3) sequence of SEQ ID NOs: 63; a variable light chain complementarity determining region 1 (CDRL1) sequence selected from the amino acid sequences of SEQ ID NOs: 64 to 68, preferably SEQ ID NO: 64; a variable light chain complementarity determining region 2 (CDRL2) sequence selected from the amino acid sequences of SEQ ID NOs: 69 and 70, preferably SEQ ID NO: 69; and a variable light chain complementarity determining region 3 (CDRL3) sequence selected from the amino acid sequences of SEQ ID NOs: 71 to 74, preferably SEQ ID NO: 72.

29. The protein construct of any one of claims 22, 26, 27 or 28, wherein said anti- CD47 scFv comprises a variable light chain amino acid sequence selected from the amino acid sequences of SEQ ID NO: 75 to 79, 114, 115 and a variable heavy chain selected from the amino acid sequences of SEQ ID NO: 80 to 85. -HI-

BO. The protein construct of claim 29, wherein said anti-CD47 scFv comprises a variable light chain amino acid sequence of SEQ ID NO: 78 and a variable heavy chain amino acid sequence of SEQ ID NO: 83.

31. The protein construct of any one of claims 1 to 30, wherein the anti-MSLN IgGl antibody or antigen-binding fragment thereof and the polypeptide capable of binding to CD47 is connected by a polypeptide linker.

32. The protein construct of claim 31 wherein said polypeptide linker spatially separates MSLN- and CD47-binding and thereby allows for a simultaneous binding of the protein construct to MSLN and CD47.

33. The protein construct of claim 31 or 32, wherein said polypeptide linker comprises or consists of 4 to 40 amino acids.

34. The protein construct of any one of claims 31 to 33, wherein said polypeptide linker comprises, essentially consist of, or consists of the amino acid glycine, alanine, proline, lysine, threonine, aspartic acid, asparagine and/or serine.

35. The protein construct of any one of claims 31 to 34, wherein said polypeptide linker comprises, essentially consists of, or consists of one or more of the amino acid sequence groups of SEQ ID Nos: 43 to 48.

36. The protein construct of any one of claims 31 to 35 wherein said polypeptide linker is fused to either the N-terminus of the variable light (VL) or the N- terminus of the variable heavy (VH) chain or the C-terminus of the constant light (CL) or the C-terminus of the constant heavy (CH3) chain domain of the anti-mesothelin IgGl antibody, preferably to the N-terminus of the variable light (VL) chain of the anti-mesothelin IgGl antibody.

37. A nucleic acid molecule comprising a polynucleotide encoding the protein construct of any one of claims 1 to 36, or a fragment of said protein construct.

38. A vector comprising the nucleic acid molecule of claim 37.

39. A host cell comprising the nucleic acid molecule of claim 37 or the vector of claim 37.

40. A host cell that expresses the protein construct of any one of claims 1 to 36, or a fragment of said protein construct.

41. A method of producing the protein construct of any one of claims 1 to 36 comprising the cultivation of a host cell of claim 39 or 40, thereby expressing said protein construct.

42. A product produced by the method of claim 41.

43. A pharmaceutical composition comprising the protein construct of any one of claims 1 to 36, or the product of claim 42 and a pharmaceutically acceptable carrier.

44. The protein construct of any one of claims 1 to 36, the product of claim 42, or the pharmaceutical composition of claim 43 for use in the treatment of cancer.

45. A method for the treatment of cancer, wherein said method comprises administering to a patient in need thereof a protein construct of any one of claims 1 to 36, the product of claim 42, or a pharmaceutical composition of claim 43.

46. The protein construct for use of claim 44, the pharmaceutical composition for use of claim 44, the product for use of claim 44, or the method of treatment of claim 45, wherein said cancer ovarian cancer, ascites, mesothelioma, triple negative breast cancer, pancreatic cancer, pancreatic adenocarcinoma, non small cell lung cancer (NSCLC), endometrial cancer or biliary extrahepatic cancer.