Compositions and methods for diagnosis and treatment of cancer

12516104 · 2026-01-06

Assignee

Inventors

Cpc classification

International classification

Abstract

The present invention relates to the diagnosis and treatment of diseases expressing Fibronectin Extra Domain B (EDB) such as diseases characterized by tissue remodeling and/or angiogenesis, in particular cancerous diseases, such as head and neck, brain, colorectal, lung, prostate and breast cancer. More particularly, the invention concerns peptides targeting Fibronectin Extra Domain B.

Claims

1. A recombinant nucleic acid encoding a Fibronectin Extra Domain B (EDB) binding peptide, wherein said peptide comprises the amino acid sequence: TABLE-US-00005 (SEQIDNO:4) (Xaa)n1Cys(Xaa)n2ArgIle/ValArgLeu(Xaa)n3 Cys(Xaa)n4Cys(Xaa)n5Cys(Xaa)n6Cys(Xaa)n7 Cys(Xaa)n8, wherein the Cys residues form a cystine knot structure, each Xaa is independently any amino acid, and n1, n2, n3, n4, n5, n6, n7, and n8 are the respective numbers of amino acids, wherein n1 is 0 to 4, n2 is 3 to 10, n3 is 0 to 3, n4 is 3 to 7, n5 is 2 to 6, n6 is 1 to 3, n7 is 3 to 7, and n8 is 0 to 4, and each amino acid Xaa and the number of amino acids n1, n2, n3, n4, n5, n6, n7 and n8 are selected such that a cystine knot structure can form between the Cys residues; and wherein the recombinant nucleic acid is in the form of a vector or RNA.

2. The recombinant nucleic acid of claim 1, wherein n1 is 1 or 2, n2 is 4, 5, 6 or 7, n3 is 0 or 1, n4 is 4, 5 or 6, n5 is 2, 3 or 4, n6 is 1 or 2, n7 is 4, 5 or 6, and n8 is 1 or 2.

3. The recombinant nucleic acid of claim 1, which comprises the amino acid sequence: TABLE-US-00006 (SEQIDNO:5) (Xaa)n1Cys(Xaa)n2ArgIle/ValArgLeu(Xaa)n3Cys ArgArgAspSerAspCys(Xaa)n5CysIleCysArg GlyAsnGlyTyrCys(Xaa)n8; or (SEQIDNO:6) (Xaa)n1Cys(Xaa)n2ArgIle/ValArgLeu(Xaa)n3Cys ArgArgAspSerAspCys(Xaa)n5CysIleCysArg GlyAsnGlyTyrCysGly.

4. The recombinant nucleic acid of claim 1, which comprises an amino acid sequence selected from: TABLE-US-00007 (i) (SEQIDNO:8) SerValCysLysAsnValSerIleMetArgIleArgLeuCysArgArg AspSerAspCysProGlyAlaCysIleCysArgGlyAsnGlyTyrCys Gly, (ii) (SEQIDNO:9) SerValCysAlaHisTyrAsnThrIleArgValArgLeuCysArgArg AspSerAspCysProGlyAlaCysIleCysArgGlyAsnGlyTyrCys Gly, (iii) (SEQIDNO:11) ProMetCysThrGlnArgLysAsnArgIleArgLeuCysArgArgAsp SerAspCysThrGlyAlaCysIleCysArgGlyAsnGlyTyrCysGly, (iv) (SEQIDNO:12) SerValCysLysGlnAlaAsnPheValArgIleArgLeuCysArgArg AspSerAspCysProGlyAlaCysIleCysArgGlyAsnGlyTyrCys Gly, (v) (SEQIDNO:13) AlaMetCysThrGlnArgLysAsnArgIleArgLeuCysArgArgAsp SerAspCysThrGlyAlaCysIleCysArgGlyAsnGlyTyrCysGly, (vi) (SEQIDNO:14) ProAlaCysThrGlnArgLysAsnArgIleArgLeuCysArgArgAsp SerAspCysThrGlyAlaCysIleCysArgGlyAsnGlyTyrCysGly, (vii) (SEQIDNO:15) ProMetCysAlaGlnArgLysAsnArgIleArgLeuCysArgArgAsp SerAspCysThrGlyAlaCysIleCysArgGlyAsnGlyTyrCysGly, (viii) (SEQIDNO:16) ProMetCysThrAlaArgLysAsnArgIleArgLeuCysArgArgAsp SerAspCysThrGlyAlaCysIleCysArgGlyAsnGlyTyrCysGly, (ix) (SEQIDNO:17) ProMetCysThrGlnAlaLysAsnArgIleArgLeuCysArgArgAsp SerAspCysThrGlyAlaCysIleCysArgGlyAsnGlyTyrCysGly, (x) (SEQIDNO:18) ProMetCysThrGlnArgAlaAsnArgIleArgLeuCysArgArgAsp SerAspCysThrGlyAlaCysIleCysArgGlyAsnGlyTyrCysGly, (xi) (SEQIDNO:19) ProMetCysThrGlnArgLysAlaArgIleArgLeuCysArgArgAsp SerAspCysThrGlyAlaCysIleCysArgGlyAsnGlyTyrCysGly, (xii) (SEQIDNO:24) ProMetCysThrGlnArgLysAsnArgIleArgLeuCysAlaArgAsp SerAspCysThrGlyAlaCysIleCysArgGlyAsnGlyTyrCysGly, and (xiii) (SEQIDNO:25) ProMetCysThrGlnArgLysAsnArgIleArgLeuCysArgAlaAsp SerAspCysThrGlyAlaCysIleCysArgGlyAsnGlyTyrCysGly.

5. The recombinant nucleic acid of claim 1, wherein the peptide forms or is part of a scaffold.

6. The recombinant nucleic acid of claim 1, wherein the peptide is stabilized by a covalent modification.

7. The recombinant nucleic acid of claim 6, wherein said covalent modification is cyclization via one or more disulfide bridges.

8. The recombinant nucleic acid of claim 1, wherein the peptide forms and/or is part of an inhibitor cystine knot structure.

9. A Fibronectin Extra Domain B (EDB) binding agent comprising the EDB binding peptide encoded by the recombinant nucleic acid of claim 1; and wherein n2 is 5 or 6, n4 is 5, n5 is 3, n6 is 1, and n7 is 5.

10. An EDB binding agent, which comprises at least two subunits which are covalently and/or non-covalently associated, each of said subunits comprising an EDB binding peptide encoded by the recombinant nucleic acid of claim 1, wherein the EDB binding peptides are identical or different; and wherein n2 is 5 or 6, n4 is 5, n5 is 3, n6 is 1, and n7 is 5.

11. The EDB binding agent of claim 10, covalently and/or non-covalently associated with at least one detectable label or reporter and/or at least one therapeutic effector moiety.

12. A method of treating a patient comprising administering to the patient the recombinant nucleic acid of claim 1, wherein the patient has cancer or is at risk of developing cancer.

13. The EDB binding peptide of claim 11, wherein n2 is 5 or 6, n4 is 5, n5 is 3, n6 is 1, and n7 is 5.

Description

FIGURES

(1) FIG. 1: A: Analytical SDS-PAGE of FN-proteins after IMAC and SEC purification. A total of 10 g protein was applied to SDS-PAGE under reducing (+-mercaptoethanol) and non-reducing (-mercaptoethanol) conditions. Protein batch purity was furthermore determined via densitometric analysis using ImageQuant software from unmodified SDS-PAGE image. Contrast and brightness have been altered in this image for better visualization. B: Enzyme linked immunosorbent assay (ELISA)-based binding analysis of FN-6789, FN-67B89 and FN-B with anti-His- and BC-1 antibody. Error bars represent standard deviations resulting from duplicate measurements, with 1 g proteins coated per well respectively.

(2) FIG. 2: Enriched cystine-knot miniprotein sequences after three phage screening rounds of MCopt 1.0 and MCopt 2.0 libraries. Clone name, cystine-knot miniprotein sequence and proportion of total screen clones are depicted for each candidate. Variable amino acids are shown in bold letters. The identified common R-I/V-R-(L) motif is highlighted in grey (from top to bottom: SEQ ID NOs: 7, 8, 9, 10, 11 and 12).

(3) FIG. 3: EDB binding cystine-knot miniproteins resulting from phage display screening of MCopt 1.0 library and following hit identification process. Ranking values were calculated on the basis of the signal (FN-B) to noise (BSA) ratios determined via ELISA and were further normalized to the expression rate of Trx-cystine-knot miniproteins. Variable amino acids are shown in bold letters (from top to bottom: SEQ ID NOs: 10, 7 and 11).

(4) FIG. 4: Specificity analysis of Trx-cystine-knot miniprotein clones MCopt 1.0-1/-2/-3 and MCopt 2.0-1/-2/-3. 200 nM of each variant were applied to immobilized FN-B and FN-67B89 target proteins as well as to control proteins FN-6789, milk powder and bovine serum albumin (coated at 1 g/well). An HRP-conjugated anti-s-tag antibody was used to analyze binding. ELISA was performed in duplicates with Trx-MC-Myc-010 serving as negative control.

(5) FIG. 5: Extended specificity analysis of Trx-cystine-knot miniprotein clones MCopt 1.0-2/-3 and MCopt 2.0-1/-2/3. In total 200 nM of each variant were applied to FN-B, T7-TEV-B, FN-67B89 and FN-B(8-14) target proteins as well as negative controls FN-6789, milk powder, bovine serum albumin, lysozyme, ovalbumin and aldolase (coated at 1 g/well). ELISA was performed in duplicates, error bars represent standard deviation and Trx-MC-Myc-010 served as negative cystine-knot miniprotein control.

(6) FIG. 6: Saturation-binding curves of MCopt 1.0-2/-3 and MCopt 2.0-1/-2/-3 EDB binding. Trx-cystine-knot miniprotein binding was assayed against FN-B, FN-67B89 and FN-B(8-14) targets as well as against FN-6789 control protein (coated at 1 g/well) in ten different concentrations. Binding of Trx-cystine-knot miniproteins was detected with an HRP-conjugated anti-s-tag-antibody. ELISAs were performed in duplicates (single values for FN-B(8-14)). A: from top to bottom SEQ ID NOs: 7, 8, and 12; and B: from top to bottom SEQ ID NOs: 11, 8, and 29.

(7) FIG. 7: A: Respective amino acids in the first, second or fifth loop of parental MC-FN-010 sequence were exchanged with alanine. Alanine substitutions are highlighted in grey (from top to bottom: SEQ ID NOs: 11, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, and 26). B: Parental MC-FN-010 and alanine scan variants as Trx-fusion proteins (50 nM to 1.563 nM) were incubated with pre-coated human FN-B. Binding was detected with 10 ng of HRP-conjugated anti-s-tag-antibody. ELISA was performed in duplicates and in three independent assays. Relative binding of each variant was calculated by determination of apparent binding constants and comparison to parental MC-FN-010. Error bars represent the standard deviation of three duplicate measurements. C: Sequence of parental MC-FN-010 (SEQ ID NO: 11). Bold letters and brackets indicate the amino acid cysteine and the disulfide bond connectivity. Residues highlighted in light grey are relevant for target-binding, those that are marked in grey do not contribute in binding interaction or only to a minor extend.

(8) FIG. 8: Specific binding of MC-FN-010 to glioblastoma xenograft tumor sections. Representative results of immunofluorescence staining of U-87 MG tumor tissues with EDB ligand MC-FN-010 and negative control MC-FN-0115. The tissue sections (5 m) were stained with tetramerized cystine-knot miniprotein-biotin strepatividin-Cy3 complex (red) and with an anti-CD31 antibody to visualize vasculature (green). Scale bars indicate 100 m.

(9) FIG. 9: A: Specificity analysis of Trx-MC-FN-010 to human FN-67B89 and FN-6789 using ELISA. B: Specificity analysis of Trx-MC-FN-016 to human FN-67B89 and FN-6789 using ELISA. Binding of Trx-MC-FN-010 and Trx-MC-FN-016 to FN-67B89 with concentrations ranging from 1.56-50 nM in comparison to FN-6789 was detected with 50 ng of HRP-conjugated anti-S-tag-antibody. ELISA was performed in duplicates using 1 g coated FN-67B89 or FN-6789 per well.

(10) FIG. 10: A: Kinetic parameters of MC-FN-010 and MC-FN-016 resulting from surface plasmon resonance analysis. Biotinylated human FN-67B89 was immobilized to streptavidin sensor to perform affinity determination of MC-FN-010 and MC-FN-016 using two-fold serial dilutions starting from 4000 nM. Kinetic parameters were calculated using a 1:1 Langmuir fitting model applied to generated sensorgrams. B: Kinetic parameters of AF680-(MC-FN-010).sub.3 and AF680-(MC-FN-016).sub.3 resulting from surface plasmon resonance with single cycle kinetic analysis. Biotinylated human FN-67B89 was immobilized to streptavidin chip to perform binding determination of AF680-(MC-FN-010).sub.3 and AF680-(MC-FN-016).sub.3 using two-fold serial dilution starting from 10 nM. Kinetic parameters were calculated using a 1:1 Langmuir fitting model applied to generated sensorgrams.

(11) FIG. 11: A: In vivo and ex vivo imaging of U-87 MG bearing mice. Mice with tumors arising from s.c. injected human U-87 MG cells were imaged after i.v. application of 3.36 nmol AF680-(MC-FN-010).sub.3, AF680-(MC-FN-016).sub.3 (EDB binder) and control AF680-(MC-FN-0115).sub.3. Groups were stratified with three mice per group, each carrying a different tumor size. Imaging was performed 1 h, 2 h and 6 h post injection. After in vivo imaging organs and tumors were excised, weighed and used for ex vivo fluorescence signal analysis. B: Development of fluorescence signals over time in tumor, kidney, liver and lung. Fluorescence signals of organs were quantified using Living Image 2.5 imaging analysis software and normalized to the respective organ/tumor weight. Mean of data sets are shown resulting from triplicates SE. Statistical significance was calculated with two-way ANOVA (*P <0.0342; **P<0.0055; ***P=0.0001; ****P<0.0001; n. s.=not significant).

(12) FIG. 12: Representative immunofluorescence staining of normal mouse brain with EDB ligand MC-FN-010 and negative control MC-FN-0115. Tissue sections (6 m) were stained with tetramerized cystine-knot miniprotein-biotin/strepatividin-Cy3 complex and an anti-CD31 antibody to visualize vasculature. Scale bars, 100 m.

(13) FIG. 13: Specific binding of MC-FN-010 to tissue sections derived from the human U-87 MG glioblastoma cell line grown as mouse xenograft tumor. Representative immunofluorescence staining of U-87 MG tumor tissues (A) and normal mouse brain (B) with trimeric MC-FN-010 and negative control MC-FN-0115. Tissue sections (6 m) were stained with Alexa Fluor 680 conjugated trimeric cystine-knot miniproteins and an anti-CD31 antibody detected with a secondary antibody to visualize vasculature. Scale bars, 20 m.

(14) FIG. 14: A: In vivo imaging of U-87 MG bearing mice. Tumors derived from human U-87 MG cells injected s.c. in flanks of Fox n1/nu mice were subjected to imaging after i.v. application of 3.34 nmol AF680-(MC-FN-016).sub.3 alone, or in combination with a 3- and 5-fold molar excess of DOTA-(MC-FN-016).sub.3 injected concurrently (co-injection) or 30 min prior to it (pre-injection). Triple alanine-mutant peptide AF680-(MC-FN-0115).sub.3 served as a negative control. B: Imaging of organs 6 h after i.v. injection of the probes. C: Fluorescence signals in tumors were quantified and normalized to the respective tumor weight.

(15) FIG. 15: SPR binding analysis of cystine-knot miniprotein variants to EDB. Kinetic parameters of EDB-specific cystine-knot miniprotein variants as measured by surface plasmon resonance analysis and calculated using a 1:1 Langmuir fitting model. Biotinylated human FN-67B89 was immobilized to a streptavidin sensor to determine affinity of DOTA-(MC-FN-016).sub.3 using two-fold serial dilutions.

EXAMPLES

(16) The techniques and methods used herein are described herein or carried out in a manner known per se and as described, for example, in Sambrook et al., Molecular Cloning: A Laboratory Manual, 2nd Edition (1989) Cold Spring Harbor Laboratory Press, Cold Spring Harbor, N.Y. All methods including the use of kits and reagents are carried out according to the manufacturers' information unless specifically indicated.

Example 1: Material and Methods

(17) Target Expression, Purification and Biotinylation

(18) Recombinant human fibronectin EDB either as a single domain (FN-B, Uniprot ID P02751, isoform 7, amino acid E1265-T1355) or flanked by its surrounding type III domains (FN-67B89, amino acid G1080-E1455) served as target protein in this study while domains 6-9 without EDB (FN-6789) was used as control. All variants were expressed in E. coli with a c-terminal hexahistidine (H6) tag and purified via immobilized metal ion affinity chromatography (IMAC) and size exclusion chromatography (SEC). For this, codon-optimized DNA sequences were synthesized by Thermo Fisher Scientific, cloned into pET-21a expression vector (Novagen) and introduced into E. coli BL21 (DE3) cells (Agilent). Proteins were expressed in a 750-mL scale at 30 C., 120 rpm until an OD.sub.600 of approximately 0.7 was reached. For induction of protein production, 750 L 1 M IPTG were added to the main-culture and incubated at 25 C., 120 rpm overnight. Cells were harvested, re-suspended in 10 mL equilibration buffer (20 mM Tris-HCl pH 8.0, % glycerol, 500 mM NaCl, 10 mM imidazole) and lysed by sonification (Branson Digital Sonifier 250). The supernatant was purified by IMAC with a 1 mL HisTrap column (GE Healthcare) using an KTAprime plus system (GE Healthcare) and a linear gradient from 10-500 mM imidazole in 20 min. Subsequently, proteins were dialyzed against PBS (14 mM NaCl, 2.7 mM KCl, 10 mM Na.sub.2HPO.sub.4, 1.8 mM KH.sub.2PO.sub.4, pH 7.5) at 4 C. overnight and further purified by size exclusion chromatography using a HiLoad 26/600 Superdex 200 g (for FN-67B89 and FN-6789) or 75 g (for FN-B) column (GE Healthcare). Final purified proteins were analyzed via SDS-PAGE, analytical SEC and via ELISA using the EDB specific BC-1 antibody (ab154210, Abcam). Proteins were stored in aliquots in PBS supplemented with 5% mannitol and 5% trehalose at 20 C. FN-B, FN-67B89 and FN-6789 proteins were biotinylated via primary NH.sub.2-groups using EZ-Link Sulfo-NHS-LC-Biotin (Thermo Fisher Scientific) according to manufacturer's introductions.

(19) Quality Control of FN-67B89 Target Protein

(20) To determine the quality of expressed and purified FN-67B89 protein, ELISA was used as technique to analyze protein-protein interactions in a 96-well format. In a first step, 10 g/mL target proteins were coated on a Maxisorb 96-well plate in a volume of 100 L via passive absorption under alkaline conditions, using 50 mM Na.sub.2CO.sub.3 pH 9.4 coating buffer. Coating of ELISA plates was performed overnight at 4 C. After removal of coating buffer the plates were washed three times with 300 L PBS-T and 300 L blocking buffer (3% BSA in PBS) was added. Blocking was performed for 2 h at RT. For primary incubation, plates were washed three times with 300 L PBS-T and incubated for 1 h with 100 L of BC-1 antibody diluted 1:1000 in PBS-T, at 4 C. A HRP-conjugated anti-mouse antibody (554002, BD Pharmingen) diluted 1:5000 in PBS was added after washing three times with PBS-T in 100 L/well and incubated for 1 h at 4 C.

(21) Plates were then washed three times with PBS-T and three times with PBS prior to detection of antigen-antibody complexes via HRP mediated conversion of TMB substrate. 100 L TMB substrate was added to the wells and a blue color developed in proportion to the amount of analyte present in the sample. Color development was stopped by adding of 50 L 0.2 M HCl and the specific absorbance was measured at 450 nm.

(22) Selection of EDB-Specific Ligands via Phage Display

(23) EDB-specific cystine knot miniproteins were selected via phage display using two different combinatorial libraries which are based on an open chain variant of the trypsin inhibitor II from Momordica cochinchinensis (oMCoTI-II). Both libraries are based on an M13 phagemid system but differ with respect to the applied randomization scheme. While in the MCopt1.0 library loop one (including a length variation of 6, 9 and 12 amino acids) and three as well as the two amino acids at the N-terminus have been randomized, the MCopt2.0 library was built via randomization of loop 1 with 10 amino acids only. Additionally, they are also different in the type of display as MCopt1.0 is presented via major coat protein pVIII and MCopt2.0 via pIII.

(24) In total three screening rounds based on streptavidin-coated (SA) magnetic beads were performed with each library. For a screening round 250 L Dynabeads M-280 Streptavidin (Life Technologies) were transferred to a 2 mL tube each and washed with 1 mL TBS-T (50 mM Tris, 150 mM NaCl, 0.1% Tween-20, pH 7.4). 100 g biotinylated FN-B in 200 L TBS (50 mM Tris, 150 mM NaCl, pH 7.4) was added to the first tube, 200 L TBS without target to the second tube (negative selection of phages) and beads were incubated on the rolling mixer for 20 min at 30 rpm. Tubes were placed back into the magnet, remaining solution was discarded and the beads washed twice with TBS-T. Beads were blocked with 2% milk powder (Carl Roth) in TBS for 1 h at 4 C. and 30 rpm. While biotinylated FN-B coated SA-beads were blocked for another 30 min, 710.sup.13 (1st round) or 710.sup.12 (2nd and 3rd round) phages were added to the uncoated SA beads in 1 mL 2% milk powder in TBS and incubated for 30 min at RT, 30 rpm (negative selection). The blocking solution of target-coated beads was discarded; beads washed twice with TBS-T and phage supernatant from negative selection was added. Target-coated beads were incubated with the phage suspension for 1 h at RT and 30 rpm. Subsequently, unbound phages were washed off, washing the beads six times with TBS-T and twice with TBS. To elute bound phages, a pH-shift elution was performed, adding 50 L 100 mM trimethylamine (TEA) to the washed beads. The TEA bead suspension was incubated for 6 min at 3000 rpm, placed back into the magnet and supernatant transferred to a fresh tube containing 100 L 1M Tris/HCl pH 7 for neutralization. A second elution step was performed by adding 50 L 100 mM glycine (pH 2) to the target-coated beads and incubating the mixture for 10 min in a thermomixer at max rpm. Tubes were placed back into the magnet and the supernatant was transferred to the Tris/TEA mixture from the first elution. Phage eluate was used to infect exponentially growing E. coli XL1-Blue (Agilent) for phage pool amplification. 1800 L E. coli XL1-Blue cells at an OD.sub.600 of 0.5 were added to the eluate and incubated for 30 min at 37 C. without and for 30 min at 37 C. with agitation at 150 rpm. The infected XL1-Blue cells were then dispended on two large agar plates containing 100 g/mL carbenicillin and 0.4% glucose and incubated overnight at 37 C., before phage rescue was performed on the following day.

(25) For phage rescue 4 mL LB media were distributed per plate and cells scraped off with a cell scraper. A 50-mL culture was inoculated to a final OD600 of 0.2. Cells were grown until an OD.sub.600 of 0.5 and infected with 0.510.sup.12 VCSM13 helper phages (Agilent) for phage production from phagemid vector. E. coli XL1-Blue cells were incubated for 30 min, 37 C., without agitation followed by 30 min, 37 C., with agitation at 150 rpm. The bacterial suspension was centrifuged at 4500 g for 10 min at RT and the supernatant discarded. Bacterial pellets were taken up in 50 mL LB media (supplemented with 100 g/mL carbenicillin, 25 g/mL kanamycin, 1 mM IPTG) for induction of protein production and E. coli phage production cultures grown overnight at 30 C., 250 rpm. E. coli cells were harvested by centrifugation at 4500 g for 15 min, 4 C. 10 mL PEG/NaCl solution (25% (w/v) polyethylene glycol, 15% (w/v) NaCl) was added to 40 mL of the phage containing supernatant, inverted and placed on ice for 30 min. Phage particles were precipitated via centrifugation at 15.000 g, 20 min at 4 C. Supernatants were discarded and phage pellets were taken up in 1600 L Tris/HCl (pH 8.0) for subsequent centrifugation at 15.000 g, 10 min at 4 C. The phage containing supernatant was added to 400 L PEG/NaCl again and incubated on ice for 20 min. After the second PEGNaCl precipitation, tubes were centrifuged at 15.000 g, 15 min at 4 C. Phage pellets were re-suspended in 800 L Tris/HCl and heated to 65 C., 15 min in a final purification step. Phage suspensions were centrifuged at RT, 15.000 g for 10 min, supernatants were taken and phage particle concentration was determined photometrically using the dual wavelength modus OD.sub.269-OD.sub.320. Phage concentrations were calculated according to the nucleotide content and molar extinction coefficient of M13 phages (Barbas et al. 2004).

(26) Hit Identification of Selected Cystine-Knot Miniproteins

(27) In the identification process, enriched screening pool phagemids derived from the 3.sup.rd screening round were prepared in order to amplify cystine-knot miniprotein sequences by PCR. PCR-inserts were cloned into the expression vector pET-32-LibEx, a derivative of pET-32a (Novagen), to enable an expression of cystine-knot miniproteins as thioredoxin-A fusion variants. This vector carries DNA sequences in successive order encoding for E. coli thioredoxin-A to allow for efficient disulfide bond formation in the cytoplasm, a H6-tag for rapid purification, an s-tag for detection by an antibody and a thrombin cleavage site to remove the fusion tag. DNA fragments encoding for cystine-knot miniprotein sequences were introduced downstream of the thrombin cleavage site into pET-32-LibEx vector via unique Barn HI and Kpn I restriction sites. Vectors were introduced into E. coli SHuffle T7 Express cells (New England BioLabs) via heat-shock and the cells were plated on selective agar plates. Single colonies were picked, sequenced and transferred to a 96-well plate for small scale expression in 1 mL autoinduction media (MagicMedia, Thermo Fisher Scientific). Protein production was conducted at 30 C. and 220 rpm overnight. Cells were harvested by centrifugation at 3000 g for 15 min, lysed via incubation in buffer (20 mM Tris, 2 mM MgCl.sub.2, 20 mM NaCl, pH 8) containing 0.1 mg/mL lysozyme (Merck Millipore) and 5 U/mL benzonase (Merck Millipore) combined with a freeze-thaw cycle and heated for min at 80 C. After a final centrifugation (3000 g, 15 min, 4 C.) to remove cell debris, the supernatant was collected for E-PAGE and binding analysis. E-PAGE (Life Technologies) was used as a high throughput gel system to simultaneously analyze 96-probes in parallel for a quantification of produced and heat purified proteins. E-PAGE analysis was performed according to manufacturer's instructions.

(28) ELISA was used as technique to analyze protein-protein interactions. Cavities of 96-well microtiter plates (Nunc MaxiSorp, Thermo Fisher Scientific) were coated with 1 g FN-B, bovine serum albumin (BSA, Eurobio), milk powder, streptavidin (Sigma Aldrich), T7-His-TEV-B (LD BioPharma), FN-B(8-14) (R&D Systems), lysozyme, ovalbumin (GE Healthcare Life Science), aldolase (GE Healthcare Life Science) or 0.6 g anti-c-myc-antibody (M4439, Sigma) overnight at 4 C. The wells were washed three times with PBS-T (1PBS with 0.1% (w/v) Tween-20), blocked with 1 Casein buffer (Sigma Aldrich) diluted in PBS for 2 h at RT and washed again as described. 20 L of supernatant containing the respective heat-purified fusion protein was added to 80 L PBS-T or 200 nM of MC-Myc-010 fusion protein (wells with anti-c-myc-antibody) diluted in PBS-T, applied to the cavities and incubated for 1 h at 4 C. After three times of washing with PBS-T, binding of the respective variant was detected with a horseradish peroxidase (HRP)-conjugated anti-S-tag antibody (ab18589 or ab19324, Abcam). Enzymatic reaction was measured with TMB as a chromogenic substrate and stopped with 0.2 M HCl after approximately 5 min. The measurement of absorbance at 450 nm was performed using an Infinite M200 PRO Microplate Reader (Tecan).

(29) In order to compare binding signals among different plates, ELISA signals were normalized to the internal plate control (c-myc binding). Normalized FN-B signals were then referenced to normalized BSA signals to evaluate binding ability of selected cystine-knot miniproteins. In addition, the target binding signals were correlated to the protein expression rate, resulting in a ranking value for the identification of hits.

(30) Recombinant Cystine-Knot Miniprotein Production

(31) Recombinant protein production was carried out using E. coli SHuffle T7 Express cells carrying pET-32-LibEx vector encoding for the respective cystine-knot miniprotein sequence in a 750-mL scale at 30 C., 120 rpm. After the culture reached an OD.sub.600 of approximately 0.7, induction of production was achieved by adding 750 L 1 M IPTG and incubation at 25 C., 120 rpm overnight. E. coli cells were harvested, re-suspended in 10 mL equilibration buffer, lysed by sonification and heated to 80 C. for 10 min. After centrifugation of cell debris (15.000 g for 30 min, 4 C.), the supernatant was purified by IMAC with a 1 mL HisTrap column using an KTAprime plus system and a linear gradient from 10-500 mM imidazole in 20 min. The cystine-knot miniprotein fusion protein containing fractions were collected and dialyzed against thrombin cleavage buffer (20 mM Tris, 150 mM NaCl, 1.5 mM CaCl.sub.2 and 5% (w/v) glycerol, pH 8.45) at 4 C. overnight.

(32) Fusion proteins were either directly used for ELISA-based assays or processed further in case that the untagged miniprotein was needed, e.g. for SPR analysis. For this, fusion proteins were cleaved with 0.5 U of thrombin (Sigma-Aldrich) per 1 mg protein and incubated at 37 C. overnight.

(33) Separation of protein fragments was performed by reverse phase chromatography with Agilent 1260 Infinity Quaternary LC system (Agilent) and a 3 mL RESOURCE RPC column (GE Healthcare) using a linear gradient from 2-80% acetonitrile in H.sub.2O supplemented with 0.05% trifluoroacetic acid (TFA). Respective fractions containing cystine-knot miniprotein were lyophilized in a RVC 2-18-CD Plus SpeedVac (Christ) as a final step. Amount of cystine-knot miniprotein was determined by weighing and the peptides were stored in lyophilized form at 20 C. Identity was verified by mass spectrometry with a LCMS Single Quad G6130B System (Agilent Technologies) using a standard electrospray ionization protocol.

(34) Alanine Scanning Mutagenesis of Selected MC-FN-010

(35) In order to identify the residues within MC-FN-010 that contribute to EDB-binding an alanine scanning mutagenesis was performed. This method includes a systemic substitution of amino acids against alanine at defined sequence positions and subsequent binding analysis of the generated mutants. For generation of alanine scanning MC-FN-010 derivatives mutations were either introduced via PCR or the whole coding sequence was assembled via direct synthesis of GeneArt Strings fragments (Thermo Fisher Scientific). Respective DNA fragments were cloned into pET-32-LibEx expression vector using unique BamHI and KpnI restriction sites and introduced into E. coli SHuffle T7 Express competent cells (New England BioLabs). All mutations were verified by DNA sequencing. The alanine scan mutagenesis variants were expressed in 24-well format using 5 mL of selective autoinduction media. Production and fusion protein purification was performed as described above for the 96-well format, but included a further purification step of the supernatant using HisPur Ni-NTA spin columns (Thermo Fisher Scientific). Binding ability and specificity of cystine-knot fusion proteins to target and off-target protein was carried out with an antibody-based ELISA assay as described above.

(36) Surface Plasmon Resonance Spectroscopy

(37) Binding kinetics of monomeric and trimeric cystine-knot miniprotein ligands to its target protein was determined using a Biacore T-100 device (GE Healthcare Life Science) with PBS-T as running buffer. For this, the biotinylated FN-67B89 protein (200-300 g/mL) was captured by binding to a flow cell of a SA sensor chip (GE Healthcare Life Science). To analyze monomeric ligands an immobilized target density of maximum 750 response units (RU) was applied and for trimeric variants a RU of maximum 400 was aimed for. Binding analysis of monomeric ligands was performed using a multi cycle kinetic method with concentrations ranging from 50 to 4000 nM. A cycle started with an association period of 90 sec, followed by a dissociation period of 420 sec and a final regeneration step. Kinetic measurement was conducted applying a flow rate of 20 L/min. Trimeric variants were analyzed under the same association and dissociation conditions, but using the single cycle kinetic measurement mode in a constant flow of 30 L/mL. In this case the analyte concentration was between 1.25 nM to 10 nM. Binding kinetics and steady state analysis were calculated using a global kinetic fit model (1:1 Langmuir, Biacore T-100 Evaluation Software, GE Healthcare Life Science).

(38) Immunofluorescence Staining

(39) For immunofluorescence staining cryopreserved tumor or brain pieces were cut in five micron thick sections, fixed in ice cold acetone for 5 min and air-dried. Slides were then blocked in PBS with 3% BSA at RT for 5 min. For staining of EDB, 1 g of the respective biotinylated cystine-knot miniprotein was incubated with 2.9 g streptavidin-Cy3 conjugate (Rockland Immunochemicals) at RT for 30 min. The pre-formed complex was then added to the tumor sections and incubated for min at 37 C. Afterwards, slides were washed three times with PBS containing 1% BSA. CD31 staining was performed with a rat anti-mouse CD31 IgG antibody (clone 390, eBioscience) diluted 1:100 in PBS with 1% BSA for 30 min at 37 C. After three washing steps in PBS, cell nuclei were stained with Hchst 33342 (Thermo Fisher Scientific) diluted 1:5000 in PBS for 30 min at RT. Slides were washed again as described above and covered with coverslips in a thin layer of mounting medium (Dako). Images were captured with a Zeiss Apotome microscope (Carl Zeiss) and analyzed with ZEN software (Carl Zeiss).

(40) Peptide Synthesis

(41) Trimeric Alexa Fluor 680 (AF680) conjugated ligands as well as N-terminally biotinylated miniproteins were purchased from Pepscan. All obtained peptidic constructs were stored as 100 g aliquots at 20 C. For experiments all peptides were dissolved in 100 L DPBS (Gibco) resulting in a concentration of 1 g/L. For all constructs identity was verified by ESI mass spectrometry and purity was analyzed by analytical reverse phase chromatography (Pepscan). Additionally, trimers were analyzed via SDS-PAGE and SPR in order to characterize target binding properties (binding to FN-67B89) and specificity (binding to FN-6789).

(42) U-87 MG Xenograft Mouse Model

(43) Human glioblastoma U-87 MG (ATCC) cell line was cultured in EMEM medium (ATCC) supplemented with 10% FCS under aseptic conditions at 37 C. with 5% CO.sub.2 and 95% humidity.

(44) Mice were housed in the animal facility at BioNTech AG and all animal protocols were approved by Tierschutzkommision des Landesuntersuchungsamts Rheinland-Pfalz. Four weeks old Fox n1/ nu mice ranging in weights between approximately 25 and 28 g were obtained from Janvier. For xenograft mouse studies 710.sup.6 human U-87 MG cells were subcutaneously injected into the right flank of Fox n1/ nu mice and tumors were allowed to grow for approximately five weeks. Subcutaneous tumor size was determined using ellipsoid formula

(45) ( width length 2 2 ) .
All animals with tumor volume between 100-1200 mm.sup.3 were included in the studies and mice were randomly assigned to experimental cohorts.
In Vivo and Ex Vivo Imaging

(46) Mice carrying a desired tumor size were included for analysis of biodistribution and tumor targeting of trimeric constructs. All trimeric constructs were injected intravenously via retrobulbar venous plexus in a final volume of 100 l PBS buffer (3.34 nmol/mice). Mice (n=3 for each construct) were imaged in an IVIS Spectrum System (Perkin Elmer) using excitation range of 615-665 nm and monitoring emission signals at 695-770 nm. Imaging process was performed 1 h, 2 h or 6 h post-injection and after euthanization the tumor and specific organs were excised, imaged, weighed and cryo-conservated for further analysis. Fluorescence intensity of regions of interest was quantified using Living Image software (PerkinElmer). Statistically significance was calculated based on triplicate data sets using two-way ANOVA analysis in GraphPad Prism.

Example 2: Screening and Selection of EDB-Specific Ligands

(47) Since cystine-knot miniproteins have been shown to be ideally suited as agents for tumor imaging (Kimura et al. 2009; Moore et al. 2013; Miao et al. 2009; Soroceanu et al. 1998; Veiseh et al. 2007, Nielsen et al. 2010; Hackel et al. 2013; Zhu et al. 2014), we used the open chain sequence of Momordica cochinchinensis trypsin inhibitor-II (oMCoTI-II) as basis for a combinatorial phage library construction to select target binding ligands (Hernandez et al. 2000). Our first library (MCopt 1.0) comprises sequences with randomized amino acids in the first loop, scattered positions in the third loop and two variable residues in front of the first cysteine. Cystine-knot miniprotein sequences were genetically fused to the major coat protein (pVIII) of M13 phages. In addition, a second library (MCopt 2.0) was developed with randomization in the first loop of the sequence and a presentation of proteins via the minor coat protein (pIII) of M13 phages. The libraries thus distinguish in the randomized loop positions and sequence length as well as the protein presentation valency, which may lead to variation in ligand selection outcome. Both libraries were applied in parallel in order to identify cystine-knot miniproteins directed against fibronectin extra domain B (EDB) which is known to be highly expressed in different tumor entities while absent from most normal tissues expect interstitium of the ovary or synovial cells (Carnemolla et al. 1989; Castellani et al. 1994).

(48) To generate suitable target and control proteins for the subsequent screening and hit identification process we recombinantly produced the single EDB domain (FN-B), EDB flanked by the surrounding type III domains (FN-67B89) and type III domains 6-9 without EDB (FN-6789). Correct protein sizes of all FN variants could be confirmed and yielded purity were above 93% as shown in FIG. 1A. By using a monoclonal antibody (BC-1), which distinguishes between fibronectin containing EDB and fibronectin without EDB (Carnemolla et al. 1992), we evaluated a native folding of the FN-67B89 protein. Furthermore, the C-terminally H6-tag was detected in all FN-fusion proteins (FIG. 1 B). Both phage libraries were screened in three consecutive rounds against biotinylated FN-B and after completion forty-six single clones were sequenced. In the MCopt 1.0 screening one cystine-knot miniprotein was strongly enriched, dominating the pool with 40%. Additionally, two other cystine-knot miniprotein clones were enriched with 4% and 2% (FIG. 2). In the case of the MCopt 2.0 screening three different cystine-knot miniprotein clones were enriched to a proportion of 13%, 10% and 2% of the total sequences. Interestingly, five out of six amplified sequences comprise a common R-IN-R-(L) motif at the c-terminal end of loop 1 (FIG. 2). Encouraged by these findings, we assessed FN-B binding ability of enriched sequences resulting from the screening of the MCopt 1.0 library. To this end, cystine-knot miniproteins were expressed, C-terminally fused to thioredoxin, his-tag and s-tag (Trx-cystine-knot miniprotein), in a 96-well mini scale format. Binding of proteins to FN-B and BSA was assayed in an ELISA and additionally the expression rate of each clone was determined via EPAGE analysis. Based on the derived signal-to-noise ratio and the expression value, we calculated a ranking score for each candidate as a measure for FN-B interaction. Three different Trx-cystine-knot miniprotein variants showed an increased interaction to FN-B compared to the BSA control (FIG. 3) and the sequences correspond to the enriched clones from the screening pool as expected. These three candidates as well as the R-IN-R-(L) motif containing clones from the MCopt2.0 screening were included for subsequent deeper binding analysis.

Example 3: Specificity Analysis of Enriched Cystine-Knot Miniprotein Candidates

(49) We next focused on target binding specificity of the six remaining cystine-knot miniproteins using in-house produced EDB target proteins (FN-B and FN-67B89) as well as off-target protein (FN-6789) and different control proteins (milk powder, streptavidin and bovine serum albumin). FN-6789 represents a perfect corresponding off-target protein, because fibronectin lacking extra domain B is expressed by many different cell types (Mao and Schwarzbauer 2005). All candidates display a reasonable EDB target binding, being equally high for recombinant FN-B and FN-67B89 target proteins (FIG. 4). Variants chosen on the basis of the common R-IN-R-(L) motif (MCopt 1.0-2/-3 and MCopt 2.0-1/-2/-3) showed medium to low off-target and control protein signals. While MCopt 1.0-2 and MCopt 1.0-3 were already identified in MCopt 1.0 hit identification, MCopt 2.0-1, -2 and -3 were solely identified from the MCopt 2.0 pool on the basis of their common motif and were thus not assayed for FN-B target binding before. MCopt 1.0-1 however which does not contain the R-I/V-R-(L)-motif failed in specificity evaluation, because we observed a high interaction to the off-target FN-6789 as well as to all control proteins expect milk powder. These data strongly indicate that the observed amino acid motif is relevant for EDB binding.

(50) We then further assessed EDB specificity of the five promising candidates by including also commercially available T7-TEV-B (LD BioPharma), a T7-TEV N-terminally flanked EDB domain, and FN-B(8-14) (R&D Systems), EDB with C-terminal domains 8-13 and 1/2 of domain 14. Again, binding signals were equally high for all tested EDB containing target proteins with relatively low off-target signals (FIG. 5). Since all used target proteins present EDB in different formats, EDB alone (FN-B), flanked by an artificial N-terminal construct (T7-His-TEV), flanked by natural C- and N-terminal type III domains (FN-67B89) as well as only flanked by its neighboring natural C-terminal type III domains (FN-B(8-14)), it can be concluded that is the observed binding activities are indeed specific for the EDB center piece.

(51) Finally, we generated dose-dependent binding data of four R-IN-R-(L)-motif containing candidates (MCopt 1.0-2/-3 and MCopt2.0-1/-2) towards FN-B, FN-67B89 and FN-B(8-14) with different receptor saturation concentrations, as presented in FIG. 6. Background signals observed towards FN-6789 off-target are generally much lower than for all EDB containing target proteins, which indicate a clear discrimination between fibronectin type III domains in FN-6789 and EDB. As it has already been seen in specificity ELISAs before, clone MCopt 2.0-3 illustrates a high degree of unspecific binding towards the FN-6789 off-target. Even though clone MCopt 2.0-3 shares the common R-I/V-R-(L) -motif which seems to have a crucial function in EDB target binding, other residues in the randomized cystine-knot miniprotein loop one might facilitate unspecific binding e.g. due to hydrophobic interactions.

(52) For intended application of cystine-knot miniprotein as imaging agent, we then studied binding ability of tag-free proteins by surface plasmon resonance (SPR) analysis. Surprisingly, only MCopt 1.0-3 revealed strong binding towards FN-67B89 target protein across all five candidates in the tested concentration range from 50-1000 nM (data not shown). Therefore, MCopt 1.0-3, in the following called MC-FN-010, was chosen for further analysis and optical imaging probe development.

Example 4: Mapping of the MC-FN-010 Binding Site

(53) Even though our previous findings of the amino acid motif R-IN-R-(L) in different selected cystine-knot miniproteins already suggested a high sequence contribution to EDB binding, we then experimentally evaluated its relevance in more detail. Single alanine substitutions in the MC-FN-010 target sequence was addressed and lead in total to fourteen derivate constructs (consecutively numbered from MC-FN-011 till MC-FN-0114 as shown in FIG. 7A). All constructs were tested against single domain FN-B. As expected, seven constructs with alanine exchanges in the beginning of the sequence still showed strong target interaction suggesting that those positions are not crucial for EDB binding. In contrast, four constructs with exchange in the common motif positions revealed loss of binding. A further alanine substitution in the fifth loop also led to a reduced target interaction, indicating its relevance as well (FIG. 7 B). These results confirm that four amino acid residues in the first loop (RIRL) and also the arginine residue in the fifth loop have a direct effect on the binding interaction to FN-67B89 or an indirect influence on the miniprotein conformation as summarized in FIG. 7 C.

Example 5: Specificity Analysis of Parental MC-FN-010 Miniprotein on Tumor Tissues

(54) We furthermore studied the specificity of parental MC-FN-010 in a cellular context using an U-87 MG tumor xenograft section comprising the natural EDB protein in the microenvironment. Human glioblastoma tumors are known to harbour the fibronectin EDB isoform in vascular structures (Mariani et al. 1997). Based on alanine scanning mutagenesis, we generated a negative control construct (MC-FN-0115) with alanine substitutions in three positions (PMCTQRANRIAACRRDSDCTGACICRGNGYCG (SEQ ID NO: 27). For immunofluorescence imaging experiments MC-FN-010 and MC-FN-0115 as biotinylated formats were tetramerized with Cy3-labeled streptavidin. Tetramerized MC-FN-010-bio almost solely decorated areas around vessels as confirmed with an Alexa Fluor 647-conjugated antibody against CD31, a ubiquitously expressed surface protein of endothelial cells reputed as vascular marker (FIG. 8). Merge image of Cy3 and Alexa Fluor 647 demonstrates the co-localization of both fluorescence signals associated to blood vessels. In addition, a localization of tetramerized MC-FN-010 in surrounding perivascular areas could be detected. In contrast, U-87 MG tumor sections stained with the negative control construct MC-FN-0115 showed no fluorescence signal at all (FIG. 8). No fluorescence signal for tetramerized MC-FN-010 and MC-FN-0115 was observed on normal mouse brain section (FIG. 12).

Example 6: Binding and Affinity of Cystine-Knot Miniproteins to FN-67B89

(55) Prior to a usage of targeting proteins as diagnostic tool, they need to be specifically conjugated with imaging agents (Spicer und Davis 2014). Our lead candidate MC-FN-010 contains a lysine in loop one, which is not favorable for selective agent linkage to primary amines. Based on our previous analysis this amino acid does not actively contributes to EDB binding, so we choose a derivate construct MC-FN-016 as a second candidate. We observed a binding of Trx-MC-FN-016 towards FN-67B89 in a dose-dependent manner with signals that were comparable with parental Trx-MC-FN-010 (FIG. 9). In contrast, the overall background signal to FN-6789 was continuously relatively low. Both cystine-knot miniprotein candidates exclusively targets FN-67B89 with no interaction to FN-6789, which is important since fibronectin is widely expressed in multiple cell types (Pankov und Yamada 2002). The affinities of tag-free MC-FN-010 and MC-FN-016 towards biotinylated FN-67B89 were assayed by SPR analysis. The binding kinetics of both cystine-knot miniproteins revealed low binding affinity in the one digit micromolar range with fast off-rates (FIG. 10 A).

Example 7: Generation and Evaluation of an EDB-Specific Optical Imaging Probe

(56) To achieve a stronger binding strength, the ligands were chemically trimerized via oxime ligation to take advantage of a potential avidity effect. Additionally, the molecule was tagged with a near-infrared fluorescence dye, Alexa Fluor 680, to enable an observation of distribution and localization after administration into mice. After chemical synthesis of all three trimeric constructs, different assays were performed in order to control the correct size and purity. SDS-PAGE and reverse phase chromatography analysis revealed no critical conspicuities in all constructs (data not shown). Notably, our oligomerization strategy led to enormously improved affinity of both EDB binding cystine-knot miniproteins (AF680-(MC-FN-010).sub.3 and AF680-(MC-FN-016).sub.3) resulting in a three digit picomolar affinity constant and remarkably slower off-rates compared to the monomeric variants (FIG. 10 B).

(57) Previously, it was shown that other EDB targeting molecules were applied as diagnostic reagent to image glioblastomas (Albrecht et al. 2016; Mohammadgholi et al. 2017). To this end, we focused on the feasibility of our EDB binding cystine-knot miniproteins to target human glioblastoma-bearing Fox n1/ nu mice. Whole-body and ex vivo imaging of organs were performed after intravenous (r. o.) injection of 3.34 nmol AF680-(MC-FN-010).sub.3, AF680-(MC-FN-016).sub.3 and the negative control AF680-(MC-FN-0115).sub.3. FIG. 11A presents fluorescence images with strong tumor signals resulting from AF680-(MC-FN-010).sub.3 and AF680-(MC-FN-016).sub.3 compared to the negative control AF680-(MC-FN-0115).sub.3. All trimeric constructs could also be detected in liver, gallbladder and kidney in the early timeframe. After 6 h the different organ signals decreased expect for the gallbladder, but importantly the tumor signal generated from AF680-(MC-FN-010).sub.3 and AF680-(MC-FN-016).sub.3 remained. Furthermore, the fluorescence signals of the organs were correlated with their respective weights as depicted in FIG. 11 B. The parental AF680-(MC-FN-010).sub.3 had significantly stronger tumor signals in comparison to the negative control at all time points. However, AF680-(MC-FN-016).sub.3 revealed lower signals than AF680-(MC-FN-010).sub.3, but nevertheless higher as the negative control AF680-(MC-FN-0115).sub.3.

Example 8: Specific Binding of MC-FN-010 to Tissue Sections Derived from the Human U-87 MG Glioblastoma Cell Line Grown as Mouse Xenograft Tumor

(58) Cryopreserved tumor or brain pieces were cut in six micron thick sections, fixed in l ice cold acetone for 5 min and air-dried. Slides were then blocked in PBS with 3% BSA at RT for 5 min. For staining of EDB, 0.1 vt.g AF680-(MC-FN-010).sub.3 and anti-mouse CD31 antibody (RB-10333-P1, Thermo Fisher) diluted 1:100 in PBS with 1% BSA was then added to the tumor sections and incubated for 30 min at 37 C. Afterwards, slides were washed three times with PBS containing 1% BSA. CD31 staining of primary anti-mouse CD31 antibody was detected with a secondary anti-rabbit IgG-Cy3 antibody (111-165-003, Jackson ImmunoResearch) diluted 1:400 in PBS with 1% BSA for 30 min at 37 C. After three washing steps in PBS, cell nuclei were stained with Hchst 33342 (Thermo Fisher Scientific) diluted 1:5000 in PBS for 30 min at RT. Slides were washed again as described above and covered with coverslips in a thin layer of mounting medium (Dako). Images were captured with a Zeiss Apotome microscope (Carl Zeiss) and analyzed with ZEN software (Carl Zeiss). FIG. 13 presents an additional immunofluorescence staining with the trimeric constructs (AF680-(MC-FN-010).sub.3 and control AF680-(MC-FN-0115).sub.3) on human glioblastoma xenograft tumor and normal brain tissue samples. AF680-(MC-FN-010).sub.3 stained the areas around the tumor vessels on U-87 MG sections that was localized with the vascular marker CD31, while the control AF680-(MC-FN-0115).sub.3 showed no staining at all. No staining was observed in normal mouse brain section with AF680-(MC-FN-010).sub.3, indicating tumor vasculature specificity.

Example 9: Specific Tumor Targeting with Selected Cystine-Knot Miniprotein

(59) U-87 MG xenograft mouse model was arised as described in Example 1 (U-87 MG xenograft mouse model).

(60) For in vivo competition experiment, mice carrying a desired tumor size (200 mm.sup.3) were injected intravenously via retrobulbar venous plexus with the unlabeled trimeric probe (DOTA-(MC-FN-016).sub.3) in 3- or 5-fold molar excess as competitor along with AF680-labeled trimer (3.34 nmol). Mice were imaged in an IVIS Spectrum System (Perkin Elmer) using excitation range of 615-665 nm and monitoring emission signals at 695-770 nm. Whole mice imaging process was performed 1 h, 2 h or 6 h post-injection. After 6 h mice were euthanized, tumor and specific organs were excised, imaged, weighed and cryo-conservated for further analysis. Fluorescence intensity of regions of interest was quantified using Living Image software (PerkinElmer). The kinetics of tumor fluorescence intensity in these mice was compared by in vivo and ex vivo imaging to that in mice, which were treated with the labeled AF680-(MC-FN-016).sub.3 without competitor. DOTA-(MC-FN-016).sub.3 was well suited for the competition experiment as the measured apparent binding constant to FN-67B89 was comparable to AF680-(MC-FN-016).sub.3 (FIG. 15). The tumor signals measured in vivo in mice treated with the competitor were substantially reduced at each time point (FIG. 14 A). As in previous experiments (FIG. 11), the negative control peptide AF680-(MC-FN-0115).sub.3 showed no enrichment in the tumor. After 6 h the mice were euthanized, the tumor and organs were excised, and ex vivo fluorescence imaging analysis (FIG. 14 B) was performed. Measured fluorescence intensities were normalized to the tumor weight. Competition by the unlabeled trimer was confirmed and dependency of signal reduction on the competitor dose was observed (FIG. 14 C). Injection of the competitor 30 min prior to the labeled trimer was found to be more effective as compared to its concurrent injection.

(61) Our study describes the selection of a cystine-knot miniprotein (MC-FN-010) out of a phage library against recombinant EDB. MC-FN-010 and its derivate MC-FN-016 were engineered as molecular scaffolds for tumor imaging approaches. Both EDB-binding molecules showed strong accumulation in U87-MG xenograft tumor and low background signals except for the kidneys. These results demonstrate the high potential of MC-FN-010 and MC-FN-016 as agents for tumor diagnostic technology.

REFERENCES

(62) Albrecht, Valerie; Richter, Antonia; Pfeiffer, Sarah; Gebauer, Michaela; Lindner, Simon; Gieser, Eugenie et al. (2016): Anticalins directed against the fibronectin extra domain B as diagnostic tracers for glioblastomas. In: International journal of cancer 138 (5), S. 1269-1280. DOI: 10.1002/ ijc.29874. Barbas, Carlos F.; Burton, D. R.; Scott, J. K.; Silverman, G. J. (2004): Phage Display: Cold Spring Harbor Laboratory Pr. Carnemolla, B.; Balza, E.; Siri, A.; Zardi, L.; Nicotra, M. R.; Bigotti, A.; Natali, P. G. (1989): A tumor-associated fibronectin isoform generated by alternative splicing of messenger RNA precursors. In: The Journal of cell biology 108 (3), S. 1139-1148. Carnemolla, Barbara; Leprini, Alessandra; Allemanni, Giorgio; Saginati, Marc; Zardi, Luciano (1992): The inclusion of the type III repeat ED-B in the fibronectin molecule generates conformational modifications that unmask a cryptic sequence. In: Journal of Biological Chemistry 267 (34), S. 24689-24692. Castellani, Patrizia; Viale, Giuseppe; Dorcaratto, Alessandra; Nicolo, Guido; Kaczmarek, Janusz; Querze, Germano; Zardi, Luciano (1994): The fibronectin isoform containing the ed-b oncofetal domain. A marker of angiogenesis. In: Int. J. Cancer 59 (5), S. 612-618. DOI: 10.1002/ ijc.2910590507. Hackel, Benjamin J.; Kimura, Richard H.; Miao, Zheng; Liu, Hongguang; Sathirachinda, Ataya; Cheng, Zhen et al. (2013): 18F-fluorobenzoate-labeled cystine knot peptides for PET imaging of integrin vr6. In: Journal of nuclear medicine: official publication, Society of Nuclear Medicine 54 (7), S. 1101-1105. DOI: 10.2967/ jnumed.112.110759. Hernandez, J. F.; Gagnon, J.; Chiche, L.; Nguyen, T. M.; Andrieu, J. P.; Heitz, A. et al. (2000): Squash trypsin inhibitors from Momordica cochinchinensis exhibit an atypical macrocyclic structure. In: Biochemistry 39 (19), S. 5722-5730. Kimura, Richard H.; Cheng, Zhen; Gambhir, Sanjiv Sam; Cochran, Jennifer R. (2009): Engineered knottin peptides: a new class of agents for imaging integrin expression in living subjects. In: Cancer Research 69 (6), S. 2435-2442. DOI: 10.1158/0008-5472. CAN-08-2495. Mao, Yong; Schwarzbauer, Jean E. (2005): Fibronectin fibrillogenesis, a cell-mediated matrix assembly process. In: Matrix biology: journal of the International Society for Matrix Biology 24 (6), S. 389-399. DOI: 101.1016/ j.matbio.2005.06.008. Mariani, G.; Lasku, A.; Balza, E.; Gaggero, B.; Motta, C.; Di Luca, L. et al. (1997): Tumor targeting potential of the monoclonal antibody BC-1 against oncofetal fibronectin in nude mice bearing human tumor implants. In: Cancer 80 (12 Sunni), S. 2378-2384. Miao, Zheng; Ren, Gang; Liu, Hongguang; Kimura, Richard H.; Jiang, Lei; Cochran, Jennifer R. et al. (2009): An engineered knottin peptide labeled with 18F for PET imaging of integrin expression. In: Bioconjugate chemistry 20 (12), S. 2342-2347. DOI: 10.1021/ bc900361g. Mohammadgholi, Mohsen; Sadeghzadeh, Nourollah; Erfani, Mostafa; Abediankenari, Saeid; Abedi, Seyed Mohammad; Emrarian, Iman et al. (2017): Human Fibronectin Extra-Domain B (EDB)-Specific Aptide (APTEDB) Radiolabelling with Technetium-99m as a Potent Targeted Tumour-Imaging Agent. In: Anti-cancer agents in medicinal chemistry. DOI: 10.2174/1871520617666170918125020. Moore, Sarah J.; Hayden Gephart, Melanie G.; Bergen, Jamie M.; Su, YouRong S.; Rayburn, Helen; Scott, Matthew P.; Cochran, Jennifer R. (2013): Engineered knottin peptide enables noninvasive optical imaging of intracranial medulloblastoma. In: Proceedings of the National Academy of Sciences of the United States of America 110 (36), S. 14598-14603. DOI: 10.1073/ pnas.1311333110. Nielsen, Carsten H.; Kimura, Richard H.; Withofs, Nadia; Tran, Phuoc T.; Miao, Zheng; Cochran, Jennifer R. et al. (2010): PET imaging of tumor neovascularization in a transgenic mouse model with a novel 64Cu-DOTA-knottin peptide. In: Cancer Research 70 (22), S. 9022-9030. DOI: 10.1158/0008-5472. CAN-10-1338. Pankov, Roumen; Yamada, Kenneth M. (2002): Fibronectin at a glance. In: Journal of cell science 115 (20), S. 3861-3863. Soroceanu, L.; Gillespie, Y.; Khazaeli, M. B.; Sontheimer, H. (1998): Use of chlorotoxin for targeting of primary brain tumors. In: Cancer Research 58 (21), S. 4871-4879. Spicer, Christopher D.; Davis, Benjamin G. (2014): Selective chemical protein modification. In: Nature communications 5, S. 4740. DOI: 10.1038/ ncomms5740. Veiseh, Mandana; Gabikian, Patrik; Bahrami, S-Bahram; Veiseh, Omid; Zhang, Miqin; Hackman, Robert C. et al. (2007): Tumor paint: a chlorotoxin:Cy5.5 bioconjugate for intraoperative visualization of cancer foci. In: Cancer Research 67 (14), S. 6882-6888. DOI: 10.1158/0008-5472. CAN-06-3948. Zhu, Xiaohua; Li, Jinbo; Hong, Yeongjin; Kimura, Richard H.; Ma, Xiaowei; Liu, Hongguang et al. (2014): 99mTc-labeled cystine knot peptide targeting integrin vr6 for tumor SPECT imaging. In: Molecular Pharmaceutics 11 (4), S. 1208-1217. DOI: 10.1021/ mp400683q.