BINDERS FOR INHIBITING FORMATION OF MULTIMERIC PROTEINS

Abstract

The invention relates to variants of OB-fold proteins, in particular of the Sac7d family that are able to bind a subunit of a multimeric protein and inhibit the formation of the multimer.

Claims

1-25. (canceled)

26. A polypeptide comprising a variant of a Sac7d protein, wherein the variant comprises 4 to 22 mutated amino acids in a binding site of the protein, and wherein the variant binds to a subunit of a multimeric protein but does not bind to the subunit of the multimeric protein when the subunit is assembled as a fully formed multimeric protein in a native state.

27. The polypeptide of claim 26, wherein the mutated amino acids of the variant are selected from the group consisting of V2, K3, K5, K7, Y8, K9, G10, E11, K13, E14, T17, K21, K22, W24, V26, G27, K28, M29, S31, T33, Y34, D35, D36, N37, G38, K39, T40, G41, R42, A44, S46, E47, K48, D49, A50 and P51 of Sac7d.

28. The polypeptide of claim 26, wherein the variant comprises 4 to 17 mutated amino acids selected from the group consisting of K7, Y8, K9, E11, K21, K22, W24, V26, M29, S31, T33, D35, T40, G41, R42, A44, and S46 of Sac7d.

29. The polypeptide of claim 26, wherein the multimeric protein belongs to a Tumor necrosis factor superfamily, preferably selected from the group consisting of TNF-alpha, RANKL and, TRAIL.

30. The polypeptide of claim 26 comprising SEQ ID NO: 17 or SEQ ID NO: 18.

31. The polypeptide of 26 comprising SEQ ID NO: 16.

32. The polypeptide of 26 comprising SEQ ID NO: 16, in which from 1 to 8 amino acids selected from the group consisting of V7, M8, F9, K11, V21, H22, M24, Q26, L29, E35, D41, F44 and P46 have been replaced by another amino acid.

33. The polypeptide of 26 comprising SEQ ID NO: 19 or SEQ ID NO: 20.

34. The polypeptide of claim 26, wherein the variant of the Sac7d protein is linked or fused to another protein or polypeptide.

35. The polypeptide of claim 34, wherein the other protein or polypeptide comprises a variant of a Sac7d protein.

36. The polypeptide of claim 34, wherein the other protein or polypeptide is an antibody, preferably binding to IL17, TNF-alpha, or Her2/neu.

37. The polypeptide of claim 26, wherein the polypeptide is conjugated to an organic molecule.

38. A method for producing the polypeptide of claim 26 comprising: (a) culturing a cell culture comprising cells that have been transformed by a genetic construct comprising a DNA sequence coding for the polypeptide; and (b) recovering the polypeptide.

39. A method for obtaining a protein that binds to a monomer of a multimeric protein comprising: (a) providing a combinatorial library of variants of an OB-fold protein, in which 4 to 22 residues of a binding interface of the OB-fold protein have been randomized, wherein the OB-fold protein is Sac7d or Sac7e from Sulfolobus acidocaldarius, Sso7d from Sulfolobus solfataricus, DBP 7 from Sulfolobus tokodaii, Ssh7b from Sulfolobus shibatae, Ssh7a from Sulfolobus shibatae, or p7ss from Sulfolobus solfataricus, wherein the randomized residues are selected from the group consisting of V2, K3, K5, K7, Y8, K9, G10, E14, T17, K21, K22, W24, V26, G27, K28, M29, S31, T33, D35, D36, N37, G38, K39, T40, G41, R42, A44, S46, E47, K48, D49, A50 and P51 of Sac7d; (b) expressing the variants; (c) presenting the variants to a multimeric protein that has been conjugated to a solid support so that multimerization has been abolished and present the face involved in multimerization to the variants; (d) selecting variants that bind to the face; (e) obtaining a sub-library of the selected variants that bind to the interface; (f) optionally, repeating (b) through (e) from one to three times; (g) selecting a variant from the sub-library comprising 5 to 20 residues mutated in the binding interface that bind to the face of the multimeric protein involved in multimerization and inhibits multimerization of the multimeric protein.

40. The method of claim 39, wherein the conjugation of the multimeric protein to a solid ligand comprises: (a) grafting/conjugating the ligand to the multimeric protein; (b) immobilizing the grafted/conjugated protein on a solid surface via the ligand; and (c) washing the surface in order to eliminate unbound multimeric protein.

41. A method for treating a patient in need thereof comprising administering a therapeutic amount of the polypeptide according to claim 26.

42. The method of claim 41, wherein the polypeptide binds to a monomer of TNF-alpha but not to a trimerized TNF-alpha, and wherein the patient is in need of treatment for rheumatoid arthritis, psoriatic arthritis, ankylosing spondylitis, Crohn's disease, ulcerative colitis, chronic psoriasis, hidradenitis suppurativa, juvenile idiopathic arthritis, Behcet's disease, or plaque psoriasis.

43. The method of claim 41, wherein the polypeptide binds to a monomer of TNF-alpha but not to a trimerized TNF-alpha, wherein the patient is in need of treatment for a cancer, and wherein the polypeptide is administered in combination with a chemotherapy of treatment with CAR-T cells.

44. The method of claim 41, wherein the polypeptide binds to a monomer of RANKL but not to a trimerized RANKL, and wherein the patient is in need of treatment for preventing or treating bone loss or in need of treatment for cancer.

Description

DESCRIPTION OF THE FIGURES

[0209] FIG. 1: alignment of proteins of the Sac7d family.

[0210] FIG. 2: representation of the binding curves of N11 to TNF-alpha (TNFα), in presence of TNF Receptor I (TNFRI), TNF Receptor II (TNFRII) Adalimumab (ADA) and Infliximab (IFX).

[0211] FIG. 3: OD450 nm representing binding of N11 in presence of human TNFa (hTNFa), murine TNFa (mTNFa) or no target.

[0212] FIG. 4: OD450 nm representing binding of variants of N11 in presence of human TNFa

[0213] FIG. 5: schematic representation of TNFa mutants. The E146A and Y115A are part of the ADA epitope.

[0214] FIG. 6: OD450 nm representing binding of N11 in presence of mutants of human TNFa (Y151A, Y115A, E146A and R32A).

[0215] FIG. 7: Analysis of the functionality of the Y151A mutant of the TNFalpha as compared to the wild type of TNFa (TNF WT).

[0216] FIG. 8: SDS-PAGE showing the presence of a complex of N11 with a dimer or a monomer of TNFa (arrows).

[0217] FIG. 9: schematic representation of the transition between monomeric, dimeric and trimeric forms of TNFα, and of the type of mutants generated. The S95C/G148C mutant is locked in its trimeric form, whereas the Y119A and Y119R mutants are impaired for their ability to self-assemble.

[0218] FIG. 10: illustration of the number of species of TNFa (monomer, dimer, trimer) for wild-type (WT) and mutant TNFα.

[0219] FIG. 11: biological functionality of WT and mutants TNFa in the L929 assay.

[0220] FIG. 12: OD450 nm representing binding of N11 on the Y119 mutants, wild-type (WT) or no target (A) or of the other TNFa mutants, wild-type (WT) or no target (B)

[0221] FIG. 13: blood measurement of IL6 in mice, 6h after administration of TNFa co-incubated with various molecules: N11, N9 variant of Sac7d binding TNFα, N9mut42A (the N9 variant with a mutation to abolish binding to TNF) and a irrelevant variant of Sac7d Orr NF).

EXAMPLES

Example 1. Production of a NF Against TNF or RANK Ligand

[0222] The screening was performed using a library comprising random mutations in 17 positions of the sequence of Sac7d (SEQ ID NO: 1).

[0223] The mutated positions were K7, Y8, K9, E11, K21, K22, W24, V26, M29, S31, T33, D35, T40, G41, R42, A44, and S46.

[0224] The targets were TNF-alpha and RANKL.

[0225] The targets, in the trimeric soluble form, were biotinylated and attached via biotin/avidin liaison on solid beads. They were then thoroughly washed in order to dissociate the trimers.

[0226] Screening essentially as disclosed in WO 2008/068637 was performed by ribosome display expression of the variants of the library and exposition to the bound targets.

[0227] For the selection and the identification of the clones, biotinylated hTNFa (R&D systems) and mTNFa (R&D systems) were used. The biotinylation was performed by incubation of a 50 μM solution of the target protein with a 5-fold molar excess of sulfosuccinimidyl-6-(biotinamido) hexanoate (Sulfo-NHS-LC-LC-Biotin, Pierce) in PBS (Sigma-Aldrich) on ice for 1 h. The biotinylated proteins were buffer-exchanged using protein desalting spin columns from Pierce equilibrated in 20 mM Tris-HCl, 150 mM NaCl pH 7.4 (TBS). The degree of biotinylation was determined, using the HABA assay (Sigma), as being about 0.5 to 5 molecules of biotin per protein molecule. The generation of the library corresponding to the random mutagenesis of positions as indicated above in Sac7d protein has been previously described (in particular in Mouratou et al, Methods Mol Biol. 2012; 805:315-31; Mouratou et al, Proc Natl Acad Sci USA. 2007 Nov. 13; 104(46):17983-8; WO 2008/068637).

[0228] The PCR-amplified library was transcribed and the selection was done at 4° C. essentially as previously described (Mouratou et al, 2007 and 2012, op.cit., Binz et al, Nat Biotechnol 2004, 22:575-582). For the first two rounds of selection, streptavidin (66 nM, 100 μl/well; Sigma-Aldrich) or neutravidin (66 nM, 100 μl/well; Thermo Scientific) in TBS150 (50 mM Tris HCl, pH 7.4, 150 mM NaCl) was alternately immobilized on a Maxisorp plate (Nunc) by overnight incubation at 4° C. The wells were then blocked with 300 μl 0.5% BSA (Sigma-Aldrich) in TBS150 for 1 h at 4° C. Biotinylated hTNFa (100 μl, 0.1 μM) in TBS150 with 0.5% BSA was allowed to bind for 1 h at 4° C. Before the ribosome-display round, the wells were extensively washed with washing buffer WBT (50 mM Tris acetic acid, pH 7.5, 150 mM NaCl, 50 mM Mg(CH.sub.3COO-)2, 0.05% Tween 20). Round 3 and 4 were performed similarly except that the target was presented alternately on streptavidin or neutravidin coated magnetic beads.

[0229] A ribosome-display round consisted of a 1 h pre-panning steps on streptavidin (or neutravidin) coated material and a 1 h binding step on the target protein. After washing, RNA purification and reverse transcription, the DNA pool was amplified by RT-PCR

[0230] Multiple variants were obtained for each target and further characterized to show that they bind to the target in other assays.

[0231] One assay to verify the binding to the target is performed using a variant produced with a His-Tag.

[0232] Biotinylated antigens are immobilized on streptavidin-coated plates as described above, except that the blocking step is performed for 1h at room temperature. For the identification of the positive clones, 100 μl of E. coli crude extracts are applied to wells with or without immobilized antigen for 1 h at RT. The binding is visualized using anti-RGSHis antibody HRP conjugate (Qiagen) which detects RGSHis6 tag from the variant and a solution at 1 mg/ml of o-phenylenediamine substrate (OPD) in revelation buffer (0.05 M citric acid, 0.05% hydrogen peroxide). Absorbance at 450 nm is measured using an ELISA plate reader. Quantitative ELISAs is done in the same manner, except purified protein is used.

[0233] Characterization of a specific variant binding to TNF-alpha, called N11, is disclosed in more details in the following examples.

[0234] Characterization of a variant binding to RANKL, called F10, is also disclosed below.

Example 2. Further Characterization of a Specific Variant (N11)

Example 2.1 Competition with TNF Receptors and Known Binders

[0235] The ability of other binders to TNF-alpha to compete with the binding of N11 was assessed.

[0236] Briefly, the other binders (TNF Receptor I (TNFRI), TNF Receptor II (TNFRII) or two monoclonal antibodies Adalimumab (ADA) and Infliximab (IFX)) were co-incubated with N11 and the soluble TNF.

[0237] As shown in FIG. 2, binding of N11 is neutralized by TNFRI, TNFRII, and Adalimumab whereas binding of N11 is not neutralized by Infliximab

Example 2.2. Cross Binding Between Human and Mouse TNF

[0238] EC100 for N11-human TNFa (hTNFa) interaction was determined by ELISA using plate functionalized with hTNFa and a concentration range of N11.

[0239] Cross reactivity of N11 on hTNFa and mTNFa (murine TNFa) was also monitored by ELISA at a single concentration (EC100). A plate was functionalized with hTNFa and mTNFa. Wells not functionalized with TNFa have been used as negative control to highlight the specifity of the signal observed.

[0240] It was shown that N11 does cross-react between human and mouse TNF (FIG. 3).

Example 2.3. Scan for Important Residues

[0241] In order to determine which residues are responsible for the observed binding, each residue of N11 (SEQ ID NO: 16) differing from the Sac7d (SEQ ID NO: 1) sequence was replaced by an Alanine (Alascan).

[0242] Exposition of the obtained proteins to TNFa was then performed and binding was measured by ELISA (measure of EC100 using plate functionalized with hTNFa and a concentration range of the variants of N11).

[0243] If there is a loss of binding, this indicates that the residue is very important for the binding to the subunit.

[0244] As seen in FIG. 4, it was shown that residues 31, 33, 40 and 42 are very important.

[0245] Residues 22 and 24 are also important, although mutations thereof would essentially reduce the affinity.

[0246] Modification of the other mutated residues doesn't essentially impact the binding.

[0247] It thus appears that sequences SEQ ID NO: 17 (including the very important amino acids 31, 33, 40, 42 of N11) and SEQ NO: 18 (including the very important amino acids 31, 33, 40, 42 and residues 22 and 24 of N11) represent minimal sequences that maintain the binding to TNFa.

[0248] Conclusion

[0249] The above information [0250] Binding human and mouse TNFa [0251] Competition with ADA, TNFRI, TNFRII but not IFX [0252] nature of the important amino acids

[0253] was used for in silico modeling, that showed that N11 would likely bind to the trimerisation core of TNF (two models were obtained, both binding to the core).

[0254] This was further demonstrated experimentally by engineering mutants of TNFalpha by directed mutagenesis (example 3).

[0255] It was further shown that N11 doesn't bind to other members of the TNF superfamily.

Example 3. Design of TNF Mutants and Further Characterization

[0256] 4 single mutants of TNFalpha were generated (FIG. 5) [0257] TNFvariant R32A [0258] TNFvariant E146A [0259] TNFvariant Y115A [0260] TNFvariant Y151A

[0261] The binding of N11 on plates coated with TNFalpha variants was checked (FIG. 6). It was shown that binding was essentially abolished for mutant Y151A.

[0262] It was however shown that Y151 mutant was still functional (FIG. 7).

[0263] This was performed by the L929 assay.

[0264] This indicates that the mutant Y151A is most likely well folded and that the residue is involved in the epitope recognized by N11.

[0265] Conclusion

[0266] N11 binding is significantly impacted by Y151A mutation.

[0267] Y151A mutant was demonstrated as being fully functional.

[0268] Residue Y151 is involved in the epitope of N11

[0269] It appears that residue Y151 is burried in the core of the TNF alpha trimer

[0270] This is an additional argument pointing on a mode of interaction relying on the blockade of the TNFalpha trimerization

Example 4. Further Binding Analysis—Inhibition of Formation of Trimer

[0271] Cross Linking Experiments were Performed

[0272] FIG. 8 shows that bands could be attributed to a complex of N11 with a monomer and a dimer of TNFa were observed on SDS-PAGE gel after cross-linking, whereas there was no evidence of an interaction between N11 and the trimer of TNFa. It was further observed complexes of lower molecular weight when TNFa is incubated with N11 for 24h at RT, that could be attributed to a monomer of TNFa in complex with one Nanofitin and a dimer of TNFa in complex with one Nanofitin. In a similar cross linking experiment where TNF was co-incubated with N11 overtime, it was observed that rates of TNFa trimer and dimer decreased as the time of incubation increased, whereas rates of TNFa monomer and dimer in complex with N11 increased as the time of incubation increased.

[0273] Similarly to what was observed in the cross linking experiments, species of lower molecular weight appeared after incubation of TNFa with N11.

Example 5. Inhibition of Trimer Formation

[0274] The above data strongly supports the hypothesis that N11 bind both the monomer and dimer of TNFa but not the trimer.

[0275] Mutants of TNFa were engineered to strengthen this finding (FIG. 9).

[0276] a) Mutants that Generate a Stabilized Trimer of TNFa

[0277] The stabilization of the trimer is achieved by the formation of disulfide bridges via the double mutation S95C/G148C

[0278] b) Mutants that are Impaired for Trimerization

[0279] Position 119 is critical for the trimerization of TNFalpha. Mutations at the position 119 impair the trimer formation

[0280] A low proportion of trimer observed in Y119 mutants compared to WT TNFalpha (FIG. 10).

[0281] The functionality of the TN Fa mutants was verified (FIG. 11).

[0282] The stabilized trimer remained functional (Mutant S95C/G148C) in L929 assay

[0283] The mutants Y119 were not functional in L929 assay. This observation was expected as the mutation at this position alters the trimerization and the active form of TNFalpha is the trimer. The L929 assay being not relevant to confirm that the mutations at Y119 do not impair with a correct folding of the respective TNFalpha mutants, other tests were performed to verify the functionality of Y119 mutants. A binding assay with Infliximab showed that a similar binding of infliximab (IFX) was observed on WT and Y119 TNFalpha variants. Since the Y119 mutants are still functional for binding to IFX, this indicated that the proteins are well folded.

[0284] It was shown that N11 does not bind the trimer but remains fully capable of binding the TNFalpha in its lower oligomeric states (FIG. 12). The absence of binding on the stabilized TNFa trimer correlates well with the observation made from the cross linking experiments

[0285] Binding of N11 for TNFa could also be monitored on Octet RED96 with Y119 mutants but not with WT TNFa (not shown)

[0286] Conclusions

[0287] Cross linking experiments have shown that N11 binds both monomer and dimer of TNFalpha but not the trimer. Epitope of N11 involves a residue that is burned in the core of TNFalpha trimer. N11 cannot bind the stabilized trimer of TNFalpha. Binding of N11 is facilitated when TNFalpha is present in its lower oligomeric states

Example 6. In Vivo Experiments

[0288] TNFa neutralization by N11 was demonstrated in an in vivo set up

[0289] A mixture of TNFa and N11, pre-incubated for 24h was administered i.v. at mice and the measure of IL6 (cytokine that is produced in response to TNFa) in the blood was performed after 6 h.

[0290] Further controls were used: N9, another (less efficient) variant isolated in Example 1, N9mut42A (the N9 variant with a mutation to abolish binding to TNF) and a irrelevant variant of Sac7d Orr NF).

[0291] FIG. 13 shows that both N11 and N9 are able to abolish production of IL6 in vivo thereby inhibiting the activity of TNFa in vivo.

[0292] Other in vivo experiments showed efficacy of various variants used orally in a murine model of acute colitis.

Example 7. Characterization of a Specific Variant (F10)

Example 7.1. Binding of the Sac7d Variant

[0293] A variant binding to RANKL, called F10, was also characterized. Such variant's sequence is depicted by SEQ ID NO: 19. Another variant, called F3 (sequence not shown) was also further characterized.

[0294] It was shown that both variants F3 and F10 bind to RANKL as well as to other members of the TNF superfamily (TNF-alpha, CD40 ligand and TRAIL).

[0295] In view of the fact that these Sac7d variants bind to various members of the TNF superfamily, its epitope must be localized in a region that is homologous for all proteins, namely the trimerization interface. Consequently, the variant is able to bind to the monomer when non-involved in the trimer, when such interface is available and not to the protein once trimerization is achieved.

[0296] Using bio-layer interferometry, it was shown that the epitopes of F3 and F10 are different.

Example 7.2. Cross Binding Between Human and Mouse TNF

[0297] Cross binding between human and mouse RANKL was determined, with protocols similar as the one disclosed in example 2.2.

[0298] It was shown that both F3 and F10 cross-react between human and mouse RANKL.

Example 7.3. In Vitro Experiments

[0299] CD14+ monocyte were cultured in the presence of RANKL in a medium complete medium also containing MCSF (macrophage colony-stimulating factor).

[0300] Such conditions trigger differentiation of the cells into osteoclasts.

[0301] Presence of the monoclonal antibody to RANKL IK 22.5 (1 μg/ml) abolished the differentiation of the cells.

[0302] Various concentration of Sac7d variants F3 and F10 were tested and it was shown that abolishment of differentiation was obtained from concentration as low as 25 nM (about 50% abolishment).

Example 7.4. In Vivo Experiments

[0303] Ovariectomized mice (Female c57b16 mice, 8 weeks old) were used as a model of bone loss. After recovery from surgery and acclimation for 7 days, the mice were randomly divided into a treatment and a control group (n=8 per group). A polypeptide comprising the F10 variant, associated with a peptide extending half-life in vivo was administered daily by intraperitoneal injection at 10 mg/kg. Non-ovariectomized mice were included as healthy controls. During the experimental period, the body weight of animals was monitored. After treatment for 6 weeks, mice were anesthetized with isoflurane and euthanized by cervical disruption. Hind limbs and vertebrae were collected and stored at 4° C. in 4% formol until further analysis. Internal organs were harvested, fixed in 4% formol, and paraffin-embedded for toxicity screening.

[0304] Analysis of the tissues show that administration of such F10 variant preserves the trabecular bone (ratio Bone volume/Total volume and bone surface) as compared to non-treated mice.

[0305] The anti-RANKL variant thus shows activity in vivo to protect from bone loss.

[0306] The data herein disclosed show that the screening method herein provided makes it possible to obtain variants of Sac7d protein that can bind to a subunit of proteins of the TNF superfamily, at the trimerization site and prevent multimerization. The variants don't bind to the fully multimerized native proteins.

[0307] In vivo examples show that the variants are able to inhibit the action of the multimerized proteins (presumably, but without being bound by this theory) by displacing the equilibrium between the subunits in the free for or in the multimerized protein. Although the above examples only exemplify specific proteins, other variants can be obtained with different libraries or further screening steps. Although the above examples only exemplify specific clinical conditions, the reported results show that the variants interact with the proteins that are involved in various clinical conditions and that these variants are thus useful in such diseases where such proteins are involved. Although the examples only exemplify proteins of the TNF super family, the screening test can be performed with subunits involved in other multimeric proteins and the results herein reported for the TNF proteins (binding the subunit in the free form and not the multimerized form) can be achieved by repeating the teachings to these other multimeric proteins.