NOVEL INHIBITORS OF KALLIKREIN PROTEASES AND USES THEREOF

Abstract

The present invention relates to a cyclic inhibitor of a kallikrein protease comprising or consisting of (I) the peptide (X.sup.1 (X.sup.2)(X.sup.3)R(X.sup.4)(X.sup.5)(X.sup.6)(X.sup.7)(X.sup.8)(X.sup.9)(X.sup.10)(X.sup.11) (Formula (X)), wherein (X.sup.1) is present or absent and, is preferably present, and if present, is an amino acid, and is most preferably D-alanine or G; (X.sup.2) is an amino acid with a side chain; (X.sup.3) is an amino acid with a polar uncharged side chain, is preferably with a polar uncharged side chain comprising a hydroxyl group, is more preferably T or S and is most preferably T; (X.sup.4) is an amino acid, preferably citrulline, Q or E; (X.sup.5) is an amino acid, preferably an amino acid with a hydrophobic side chain; (X.sup.6) is present or absent, is preferably absent, and, if present, is an amino acid with a negatively charged side chain, preferably D; (X.sup.7) is present or absent, is preferably present, and, if present, is an amino acid, more preferably an amino acid with a side chain comprising a pyrrole of indole, even more preferably P, hydroxyl-proline, (R)-3-piperidine carboxylic acid or W, and most preferably P; (X.sup.8) is present or absent and, if present, is an amino acid, and is preferably absent; (X.sup.9) is present or absent and, if present, is an amino acid, and is preferably absent; (X.sup.10) is an amino acid with a side chain; and (X.sup.11) is present or absent and, if present, is an amino acid, and is preferably absent; wherein the side chains of (X.sup.2) and (X.sup.10) are connected via a connecting molecule, said connecting molecule having at least two functional groups, each functional group forming a covalent bond with one of the side chains of (X.sup.2) and (X.sup.10); and wherein the kallikrein protease is Kallikrein-related peptidase (KLK5); or (II) the peptide (Y.sup.1)(Y.sup.2)(Y.sup.3) (Y.sup.4)(Y.sup.5)(Y.sup.6)(Y.sup.7)(Y.sup.8)(Y.sup.9) (Formula (Y)), wherein (Y.sup.1) is present or absent, is preferably present, and, if present, is an amino acid, preferably P, L-beta-hydroxyl-proline, D-proline, (R)-3-piperidine carboxylic acid, Q or R, and is most preferably P; (Y.sup.2) is an amino acid with a side chain; (Y.sup.3) is I, L or L-Neopentylglycine, and is preferably L; (Y.sup.4) is Y or F, and is preferably Y; (Y.sup.5) is an amino acid, preferably an amino acid with a hydrophobic side chain, or Q or R, is more preferably L, norleucine, Q, I, R or M, and is most preferably norleucine or L; (Y.sup.6) is an amino acid, preferably an amino acid with a hydrophobic side chain and is most preferably A; (Y.sup.7) is absent or present, preferably present and, if present, is an amino acid preferably Q, homoarginine, 4-guanidino-phenlalanine or R; (Y.sup.8) is an amino acid with a side chain; and (Y.sup.9) is present or absent and, if present, is an amino acid, preferably S, and is most preferably absent; wherein the side chains of (Y.sup.2) and (Y.sup.8) are connected via a connecting molecule, said connecting molecule having at least two functional groups, each functional group forming a covalent bond with one of the side chains of (Y.sup.2) and (Y.sup.8); and wherein the kallikrein protease is Kallikrein-related peptidase 7 (KLK7).

Claims

1. A cyclic inhibitor of a kallikrein protease comprising or consisting of (I) the peptide (X.sup.1)(X.sup.2)(X.sup.3)R(X.sup.4)(X.sup.5)(X.sup.6)(X.sup.7)(X.sup.8)(X.sup.9)(X.sup.10)(X.sup.11) (Formula (X)), wherein (X.sup.1) is present or absent and, is preferably present, and if present, is an amino acid, and is most preferably D-alanine or G; (X.sup.2) is an amino acid with a side chain; (X.sup.3) is an amino acid with a polar uncharged side chain, is preferably with a polar uncharged side chain comprising a hydroxyl group, is more preferably T or S and is most preferably T; (X.sup.4) is an amino acid, preferably citrulline, Q or E; (X.sup.5) is an amino acid, preferably an amino acid with a hydrophobic side chain; (X.sup.6) is present or absent, is preferably absent, and, if present, is an amino acid with a negatively charged side chain, preferably D; (X.sup.7) is present or absent, is preferably present, and, if present, is an amino acid, more preferably an amino acid with a side chain comprising a pyrrole of indole, even more preferably P, hydroxyl-proline, (R)-3-piperidine carboxylic acid or W, and most preferably P; (X.sup.8) is present or absent and, if present, is an amino acid, and is preferably absent; (X.sup.9) is present or absent and, if present, is an amino acid, and is preferably absent; (X.sup.10) is an amino acid with a side chain; and (X.sup.11) is present or absent and, if present, is an amino acid, and is preferably absent; wherein the side chains of (X.sup.2) and (X.sup.10) are connected via a connecting molecule, said connecting molecule having at least two functional groups, each functional group forming a covalent bond with one of the side chains of (X.sup.2) and (X.sup.10); and wherein the kallikrein protease is Kallikrein-related peptidase 5 (KLK5); or (II) the peptide (Y.sup.1)(Y.sup.2)(Y.sup.3)(Y.sup.4)(Y.sup.5)(Y.sup.6)(Y.sup.7)(Y.sup.8)(Y.sup.9) (Formula (Y)), wherein (Y.sup.1) is present or absent, is preferably present, and, if present, is an amino acid, preferably P, L-beta-hydroxyl-proline, D-proline, (R)-3-piperidine carboxylic acid, Q or R, and is most preferably P; (Y.sup.2) is an amino acid with a side chain; (Y.sup.3) is I, L or L-Neopentylglycine, and is preferably L; (Y.sup.4) is Y or F, and is preferably Y; (Y.sup.5) is an amino acid, preferably an amino acid with a hydrophobic side chain, or Q or R, is more preferably L, norleucine, Q, I, R or M, and is most preferably norleucine or L; (Y.sup.6) is an amino acid, preferably an amino acid with a hydrophobic side chain and is most preferably A; (Y.sup.7) is absent or present, preferably present and, if present, is an amino acid preferably Q, homoarginine, 4-guanidino-phenlalanine or R; (Y.sup.8) is an amino acid with a side chain; and (Y.sup.9) is present or absent and, if present, is an amino acid, preferably S, and is most preferably absent; wherein the side chains of (Y.sup.2) and (Y.sup.8) are connected via a connecting molecule, said connecting molecule having at least two functional groups, each functional group forming a covalent bond with one of the side chains of (Y.sup.2) and (Y.sup.8); and wherein the kallikrein protease is Kallikrein-related peptidase 7 (KLK7).

2. The inhibitor of claim 1, wherein the side chains of (X.sup.2), (X.sup.10), (Y.sup.2) and (Y.sup.8) comprise a functional group, preferably for each of (X.sup.2), (X.sup.10), (Y.sup.2) and (Y.sup.8) independently selected from —NH.sub.2 —COOH, —OH, —SH, alkene, alkyne, azide and chloroacetamide, more preferably —NH.sub.2 and —SH, and most preferably —SH.

3. The inhibitor of claim 1, wherein (X.sup.2), (X.sup.10), (Y.sup.2) and (Y.sup.8) are each independently K, ornithine, thialysine, 2,3-diaminopropanoic acid, diaminobutyric acid, D, E, C, homocysteine, penicillamine and propargylglycine, preferably C or homocysteine and most preferably all are C.

4. The inhibitor of any one of claims 1 to 3, wherein the connecting molecule is selected from the trivalent and divalent linkers, preferably the divalent linkers shown in FIG. 29 of the application, and is most preferably 2,6-bis(chromomethyl)pyridine or 1,3-dibromoacetone.

5. The cyclic inhibitor of any one of claims 1 to 4, wherein at least one of the following applies: (X.sup.1) is D-alanine or G and is preferably G; (X.sup.3) is T or S and is preferably T; (X.sup.4) is citrulline, Q or E and is preferably Q; (X.sup.5) is V, W or Y, 2-, 3-, or 4-fluorophenyl, or 1- or -2 naphthyl-alanine, and is preferably Y; (X.sup.6) is absent; (X.sup.7) is hydroxyl-proline, (R)-3-piperidine carboxylic acid or P and is preferably P; (X.sup.8) is absent; (X.sup.9) is absent; (X.sup.11) is absent.

6. The cyclic inhibitor of any one of claims 1 to 4, wherein Formula (X) is GCTRQYPC (SEQ ID NO: 1), and wherein the side chains of the two cysteines are connected via the connecting molecule.

7. The cyclic inhibitor of any one of claims 1 to 4, wherein at least one of the following applies: (Y.sup.1) is absent, P, L-beta-hydroxyl-proline, D-proline, (R)-3-piperidine carboxylic acid, Q or R, is more preferably P, L-beta-hydroxyl-proline, D-proline, (R)-3-piperidine carboxylic acid, Q or R and is preferably P; (Y.sup.3) is I, L or L-neopentylglycine and is preferably L; (Y.sup.4) is Y or F, and is most preferably Y; (Y.sup.5) is L, norleucine, M, Q, I or R, is more preferably L, norleucine, M, R, even more preferably is norleucine or L and is most preferably norleucine; (Y.sup.6) is S, A, T and is preferably A; (Y.sup.7) is Q, homoarginine, 4-guanidino-phenlalanine or R, is preferably homoarginine or R and is most preferably homoarginine; and (Y.sup.9) is absent.

8. The cyclic inhibitor of any one of claims 1 to 4, wherein Formula (Y) is PCLYLARC (SEQ ID NO: 2) or PCLY(norleucine)A(homoarginine)C, and wherein the side chains of the two cysteines are connected via the connecting molecule.

9. The cyclic inhibitor of any one of claims 1 to 8, wherein the cyclic inhibitor is linked to a component increasing serum or blood half-life of the cyclic inhibitor.

10. The cyclic inhibitor of claim 9, wherein the component is a fatty acid being capable to bind to albumin, preferably palmitic acid.

11. An ex vivo or in vitro method of inhibiting the enzymatic activity of a kallikrein protease comprising contacting the inhibitor of any of claims 1 to 10 with the kallikrein protease, wherein the kallikrein protease is preferably present in a blood, plasma or serum sample.

12. A pharmaceutical composition comprising the inhibitor of any of claims 1 to 10.

13. The inhibitor of any of claims 1 to 10 for use in the treatment or prevention of a skin disease, preferably an inflammatory skin disease.

14. Use of the inhibitor of any of claims 1 to 10 for inhibiting the enzymatic activity of a kallikrein protease ex vivo or in vitro.

15. An active compound linked to a fatty acid being capable to bind to albumin for use in the treatment or prevention of a skin disease, preferably an inflammatory skin disease, wherein the active compound is preferably a KLK5 and/or KLK7 inhibitor.

Description

[0129] The Figures show.

[0130] FIG. 1—Phage display of double-bridged peptides. (a) Library format used for the selections. The first and the last cysteine (C.sub.1 and C.sub.4) were fixed at position 2 and 18 while C.sub.2 and C.sub.3 were designed to appear variably between position 3 and 16. (b) Structures of thiol-reactive linkers used for cyclization of the linear peptide library. Two moles of linker react with one mole of peptide. (c) Isomers that potentially form during the cyclization reaction.

[0131] FIG. 2—Titers of phage display selections. (a) First selection round. (b) Second selection round. Negative selections were performed with target-free beads to confirm phage enrichment through target binding. (c) Third selection round. Linker A: 2,6-bis(bromomethyl)pyridine; linker B: 1,3-dibromoacetone; L1: E. coli titer after library inoculation and growth until OD.sub.600=0.5; L2: E. coli titer after hyperphage infection and addition of second antibiotic; S0: phage titer in supernatant of 2×YT medium after overnight phage production; S1: phage titer after PEG precipitation from the supernatant, S2: phage titer after phage reduction with TCEP and precipitation; S3: phage titer after peptide cyclization and precipitation; S4: Titers of phages captured and eluted from the beads after target panning.

[0132] FIG. 3—Sequences of phage-selected double-bridged peptides. Amino acid sequences found after three rounds of selections with an 18 amino acid linear peptide library cyclized with either 2,6-bis(bromomethyl)pyridine or 1,3-dibromoacetone. (a) Sequences found after selections against hKLK5. (b) Sequences found after selections against hKLK7. Sequences that show a consensus are shown in groups and similar amino acids are highlighted using the Rasmol color code. Inhibitory constant (K.sub.i) values of selected, synthesized peptides are shown as the mean of three measurements next to the corresponding peptide sequences.

[0133] FIG. 4—Sequence maps of consensus monocycles. Based on the assumption that the consensus surrounding cysteines were connected, sequence maps of the smallest identified monocycles were generated.

[0134] FIG. 5—Development of monocyclic inhibitors. (a) Peptides were synthesized based on amino acids found in the consensus sequences. In contrast to the first generation, they contain only two cysteines cyclized with one linker. The K values are shown as mean of three measurements ±SD. (b) The structures of peptides 035 and peptide 096, which were chosen as leads for further development.

[0135] FIG. 6—Affinity improvement with unnatural amino acids. Macrocycles with single unnatural amino acid substitutions were synthesized and tested. The K.sub.i values relative to the reference peptide from single measurements are shown. (a) For KLK5, 035 served as the reference. (b) For KLK7, 114 (not lead peptide 096) served as the reference because the C-terminal serine of peptide 096 had to be removed later for stability reasons. (c) Structures of tested unnatural derivatives. * fold-change >16.

[0136] FIG. 7—Selectivity of KLK5 inhibitor. The selectivity of the lead KLK5 inhibitor 035 to other human trypsin-like serine proteases was tested in activity assays. For each protease, their specific substrates were used.

[0137] FIG. 8—Conjugation strategies for albumin tag. (a) Structure of the used albumin binding ligand called “albumin tag”, which is based on an acylated heptapeptide. (b) Two different conjugation strategies were compared: For N-terminal conjugation the albumin tag was attached through a PEG.sub.2 linker. C-terminal conjugation was realized through sidechain coupling to a lysine in a C-terminally attached Gly-Lys-Gly linker.

[0138] FIG. 9—Affinities of albumin tag conjugates. (a) K.sub.i values of N- and C-terminal tag conjugates in comparison to unconjugated inhibitors measured by inhibition assays in the presence of HSA. (b) K.sub.d values of conjugates for HSA assessed by fluorescence polarization. A dilution serious of human serum was used. Additionally to the peptides described above, the peptide 301-tag and 302-tag were tested, which are negative controls. All data represent the mean of three measurements ±SD.

[0139] FIG. 10—Metabolic stability of inhibitor-tag conjugates in human plasma. Inhibitors were incubated in human plasma, and the relative concentration of the intact peptide over time was assessed by LC-MS. (a) The KLK5 inhibitor without (035) and with the albumin tag (035-tag). (b) The KLK7 inhibitor without (096) and with the albumin tag (096-tag).

[0140] FIG. 11—Stability improvement of the KLK7 inhibitor. (a) Degradation fragments were identified by mass, and Ser 9 was identified as the cause of hydrolytic instability. (b) The removal of Ser 9 improved the half-life to 4 h. Noteworthy, at this point Nle 5 (identified in the affinity optimization study) was incorporated in the inhibitor, which should not impair the plasma stability. Further stability issues were cause by Arg 7. (c) Replacement of Arg 7 finally lead to stable inhibitors. Homoarginine was chosen as best derivative based on affinity.

[0141] FIG. 12—Competition of small molecules with albumin tag for HSA binding. (a) Structure, binding sites and association constant (K.sub.a) values of known albumin-binding small molecules. Values form. (b) Competitive fluorescence polarization assay with albumin tag. (c) Moieties of fluorescein that potentially provide interactions with HSA in the context of the albumin tag.

[0142] FIG. 13—Structures of the final KLK5 and KLK7 inhibitors. The molecules are composed of specific, cyclic-peptide-based kallikrein inhibitors and an acylated-heptapeptide-based albumin binding ligand for half-life extension. They have single-digit nanomolar affinity for hKLK5 and hKLK7, respectively, and nanomolar affinities for HSA affording high plasma stability with a half-life of more than 90 h in human plasma.

[0143] FIG. 14—Binding to MSA assed by FP. SSSKSSS-tag (SEQ ID NO: 44) is a tag variant with all amino acids (Tyr and Glu) replaced by Ser. A dilution series auf mouse albumin in PBS was used.

[0144] FIG. 15—Pharmacokinetics in C57BL/6J mice. (a) Mice (n=3) were injected intravenously (IV), intraperitoneally (IP) or subcutaneously (SC) with 6.2 mg/kg of KLK5 inhibitor or IV (n=3) with the same dose of KLK7 inhibitor. The expected plasma concentration after IV injection was 40 μM. Blood samples were taken at different time points, plasma proteins were precipitated, and the supernatant was analyzed by analytical HPLC equipped with a fluorescence detector. Graph depicts the concentrations of intact peptide in blood plasma at different time points after injection. (b) Raw HPLC data of all mice. (c) Mice were injected SC with 6.2 mg/kg of KLK5 inhibitor (n=2) or inhibitor coinjected with 1 equivalent (26.4 mg/mL) RSA (n=2) or HSA (n=3). Blood samples were taken at different time points, plasma proteins were precipitated, and the supernatant was analyzed by analytical HPLC equipped with a fluorescence detector. Graph depicts the concentrations of intact peptide in blood plasma at different time points after injection.

[0145] FIG. 16—Biodistribution assessed by fluorescence imaging. (a) Mice (n=3 per time point) were IV injected with KLK5 inhibitor, albumin tag (without inhibitor), 035-fluorescein (non-HSA binding control) (2 μmol/kg), and vehicle (background control, n=1 per time point), sacrificed and imaging at different time points after injection. (b) Fluorescence imaging of skin patches from mice sacrificed at different time points after IV injection of 6.2 mg/kg of KLK5 inhibitor. (c) Fluorescence imaging of skin and visceral organs of three mice 48 h after IP injection of 12.4 mg/kg of KLK5 inhibitor.

[0146] FIG. 17—Characterization of SiR conjugates. (a) SDS page gel of 0.1 nmol of SiR-conjugates imaged by excitation at 628±10 nm and capturing at 678±28 nm. (b) Coomassie blue staining of the same gel. AB: mIgG anti-Flag-SiR; MSA: MSA-SiR; IN: 035-SiR-tag; TAG: SiR-tag; SIR: 6-Carboxy-silicon rhodamine. (c) Chemical structure of 035-SiR-tag conjugate.

[0147] FIG. 18—Skin distribution of inhibitors in comparison with antibodies. Mice (n=1) were IP injected with a dose of 2 μmol/kg of SiR-labeled molecules corresponding to 300 mg/kg of antibody (mIgG anti-Flag) and MSA, 6.4 mg/kg of KLK5 inhibitor, 3.6 mg/kg of tag, and 1.0 mg/kg of SiR. Mice injected with antibody or MSA were sacrificed 8 h post injection. Mice injected with KLK5 inhibitor, tag, or SiR were sacrificed 8 h post injection. Skin was isolated and cryofixed. The DAPI signal is shown in blue and the SiR signal in red. (a) Section overview. (b) View of epidermis. (c) Intensity profile plots of DAPI and SiR signal along the arrows shown in (b). Antibody: mIgG anti-Flag-SiR; MSA: MSA-SiR; Inhibitor: 035-SiR-tag; Tag: SiR-tag; SiR: 6-Carboxy-silicon rhodamine.

[0148] FIG. 19—Inhibition of mouse KLKs. Activity of KLK5 inhibitor 035-tag and 096-tag against mKLK5 and mKLK7. The same assay conditions as for the hKLK activity assays were used. Note that IC.sub.50 values are shown, as the K.sub.m values have not been determined.

[0149] FIG. 20—Characterization of inactive control peptides. 301 is the negative control for KLK5, 302 for KLK7. Inhibitors were rendered inactive by replacing the P2 amino acids with alanine and swapping the P1 and P2 positions (underlined).

[0150] FIG. 21—Peptide inhibitors used for in vivo studies. (a) 20 mg of each of the four peptides synthesized for efficacy studies is shown. In total around 90 mg were synthesized. (b) Structure, purity and mass of the four peptides synthesized for efficacy studies.

[0151] FIG. 22—H.sub.2O solubility of inhibitors. 2 mg peptide inhibitor (M≈3000 Da) were dissolved in increasing volumes of ddH.sub.2O from 20 μl to 2 mL. After thorough vortexing, the solution was centrifuged for 5 min. Visible precipitation is indicated with black arrow heads. * No precipitation was visible.

[0152] FIG. 23—Solubility of inhibitors different buffers. 10 mM stock solutions in ddH2O with 15% DMSO were diluted to 0.4 mM in different buffers, all at a pH of 7.4. The solution was centrifuged, and the peptide concentration in the supernatant was spectrophotometrically determined.

[0153] FIG. 24—Preparation of injection solutions for in vivo studies. * due to residual TFA and H.sub.2O in the powder, in practice 400 mg/ml were necessary to obtain a 100 mM solution.

[0154] FIG. 25—Subcutaneous injection of KLK5 bicycle peptide inhibitor improves KLK5-mediated NS clinical features in TghKLK5 mice. (A) Schematic representation of the protocol for subcutaneous injection with KLK5 bicycle peptide inhibitor (KLK5IN) or with the KLK5 bicycle inactive peptide inhibitor (KLK5IP). Mice were injected every 2 days during 20 days. (B) Clinical features of subcutaneously injected TghKLK5 mice with KLK5IN or KLK5IP at the age 5 days, 10 days, 15 days and 20 days. (C) Transepidermal water loss (TEWL) bi-weekly measurements for a period of 20 days following birth to the end of KLK5IN injection in TghKLK5 mice in comparison to the untreated or injected TghKLK5 mice with KLK5IP and to WT mice. Data represent mean values ±SEM from at least 6 mice per group. P-values are calculated using two-way ANOVA test. *, P<0.05; ****, P<0.0001. (D) Fluorescence microscopy image of in situ zymography with casein-BODIPY-FL substrate on skin cryosections from injected TghKLK5 mice with KLK5IN in comparison to the untreated or injected TghKLK5 mice with KLK5IP and to WT mice (scale bar, 50 μm).

[0155] FIG. 26—Abnormal epidermal proliferation was decreased in skin section from injected TghKLK5 mice with KLK5 inhibitor. (A) Hematoxylin/eosin/safranin (HES) staining of skin sections from injected TghKLK5 mice with KLK5 bicycle peptide inhibitor (KLK5IN) in comparison to those untreated or injected with KLK5 inactive bicycle peptide inhibitor (KLK5IP) and to WT mice. Scale bar, 100 μm. (B) Epidermal thickness measurements from HES staining of skin sections injected TghKLK5 with KLK5In in comparison to the untreated or injected TghKLK5 mice with KLK5IP and to WT mice. Data are mean values ±SEM from 20 measurements per skin sections of at least 4 mice. Data are mean values ±SEM from 3 dependent experiments with at least 6 mice per group. P-values are calculated using the ANOVA test ****, P<0.0001. (C) Immunofluorescence staining of Ki67, keratin 14 (Krt14) and filaggrin (Flg) on skin sections from injected TghKLK5 with KLK5IN in comparison to the untreated or injected TghKLK5 mice with KLK5IP and to WT mice. Pictures are representative of 3 independent experiments performed on skin sections from at least 6 mice per group. Scale bar, 50 μm; blue, Dapi; red, targets; dotted white lane delineates the dermal-epidermal junction.

[0156] FIG. 27—KLK5 bicycle peptide inhibitor impaired Mat cells and neutrophils infiltration in skin from TghKLK5 mice. (A) Detection of mast cells by toluidine blue staining in skin section from injected TghKLK5 mice with KLK5 bicycle peptide inhibitor (KLK5IN) in comparison to those untreated or injected with KLK5 inactive bicycle peptide inhibitor (KLK5IP) and to WT mice. Scale bar, 100 μm. (B) Enumeration of the number of Mast cells stained with toluidine blue. Data are mean values ±SEM from at least 6 mice per group. P-values are calculated using two-way ANOVA test: *, P<0.05; **, P<0.01. (C) Neutrophil sainting in skin section from treated TghKLK5 mice Scale bar, 100 μm; Blue, Dapi; Red, Neutrophils; dotted white lane delineates the dermal-epidermal junction. Pictures in this figure are representative at least 6 mice per group.

[0157] FIG. 28—KLK5 inhibitor trigger decreased expression of inflammatory cytokines in skin from TghKLK5 mice. (A) mRNA expression of II-36a, II-36b, II-36g, Ccl20 and S100a8 normalized to Hprt in skin extracts from injected TghKLK5 mice with KLK5 bicycle peptide inhibitor (KLK5IN) in comparison to those untreated or injected with KLK5 inactive bicycle peptide inhibitor (KLK5IP) and to WT mice. Data are mean values ±SEM from skin extracts of at least 6 mice. P-values are calculated using Two-way ANOVA test. *, P<0.05; **, P<0.01; ****, P<0.0001. Immunodetection of II-36a protein in (B) skin extracts and (C) skin section from injected TghKLK5 with KLK5IN mice in comparison to those untreated or injected with KLK5IP and to WT mice. Protein level measurement of II-36a was normalized to Hsc70. Scale bar, 100 μm; blue, Dapi; red, II-36a. Pictures are representative of 3 independent experiments from skin sections and skin extracts of at least 6 mice per group.

[0158] FIG. 29 Chemical structure of connecting molecules.

[0159] The Examples illustrate the invention.

EXAMPLE 1—PHAGE DISPLAY OF BICYCLIC PEPTIDE INHIBITORS OF KLK5 AND KLK7

[0160] Phage display was used to identify cyclic-peptide inhibitors of KLK5 and KLK7. A phage encoded library of linear peptides of 18 amino acids of the format X C X.sub.m C X.sub.n C X.sub.oC X (m+n+o=12) developed by Carle et al. in the lab of the inventors was used for the selections (FIG. 1a). All peptides in this library were designed to contain four cysteines for cyclization with two thiol-reactive linkers based on the novel, double-bridged peptide format recently published by our lab. This library format offers the possibility to identify double-bridged peptides, but also monocyclic peptides, in case that only one ring of the peptide contributes to target binding.

[0161] The first and last cysteines (C.sub.1 and C.sub.4) in this library are fixed at position 2 and 17 while the second and third cysteines (C.sub.2 and C.sub.3) can appear at any position between 3 and 16. This gives a theoretical combinatorial diversity of 1.5×10.sup.20. However, the realized diversity after cloning is limited by the transformation efficiency and was determined to be 3×10.sup.10 for the library that was used. This library was encoded in a phagemid as linear peptide N-terminally fused to the PIII protein of the M13 phage. The library was expressed with the help of a hyperphage, leading to multivalent display. On the surface of the phage, the expressed, linear peptides were cyclized with either of the two thiol-reactive linkers 2,6-bis(bromomethyl)pyridine or 1,3-dibromoacetone (FIG. 1b). As two linker molecules can react with one linear peptide, three different regioisomers form during the cyclization reaction, depending on which of the two cysteines is connected (FIG. 1c). This expands the genetically encoded diversity by another factor of 3 based on the structural diversity.

[0162] Three rounds of phage-display selections against immobilized hKLK5 and hKLK7 (10 μg, 5 μg, 2.5 μg—first, second, third round) were performed with this library. Samples to determine the phage titers were taken at different steps of the phage purification, peptide modification, and phage selection procedure during each of the three selection rounds (FIG. 2). These titers give valuable information about the success of the phage selection procedure. The 11′ and 12′ titers indicate the number of E. coli after library inoculation and hyperphage infection and thus give information of how much of the library diversity is covered in the culture. The ‘S0’ titer shows the amount of phages in the supernatant of the culture medium after overnight phage production and indicates how often each clone of the library is represented. The following phage titers ‘S1’ to ‘S3’ follow up the phage titers through the purification and modification to ensure that the amount of infective phages is not severally reduced by any of the steps: ‘S1’: phage titer after PEG precipitation from the supernatant, ‘S2’: phage titer after phage reduction with TCEP and precipitation; ‘S3’: phage titer after peptide cyclization and precipitation. The ‘S4’ titers finally indicate the number of phages that were captured and eluted from the beads after target panning. To ensure the majority of phages were captured by the target protein and not by the beads or the plastic tubes, negative control selections using beads without bound target were performed during the second round.

[0163] The titer ‘L2’ of the first round indicates that 2.5×10.sup.10 phage producing cells were inoculated and successfully infected with hyperphage, meaning that 2.5×10.sup.10 peptides of the library were represented in the selections (FIG. 2a). Negative control selections with beads not bound to any target were performed during the second selection round. An enrichment between 385-fold (KLK5 with 2,6-bis[bromomethyl]pyridine) and 5000-fold (KLK7 with 1,3-dibromoacetone) could be observed, indicating successful, target specific selections (FIG. 2b). Positive capture titers were target- and linker-specific, which confirms target specificity and indicates successful selections. As a target specific enrichment could already be confirmed during the second selection round, no negative control selections were performed during the third round. Furthermore, the titers confirmed that the loss of infective phages during the preparation procedure did not exceed one order of magnitude in any of the three selection rounds.

[0164] As the titers indicated the successful selection of target specific binders, the sequences of 96 clones for each target and linker were identified by Sanger sequencing (FIG. 3). For KLK5 a single consensus group per linker could be identified, while for KLK7 two or three consensus groups per linker were found. For KLK5, threonine followed by arginine was common in all sequences, while for KLK7 leucine followed by tyrosine appeared in most of the sequences. Both motifs match with the reported substrate specificities for the S2 and S1 subsides of KLK5 and KLK7, respectively, which implies active site binding with a standard mechanism of inhibition. Furthermore, it can be observed that a consensus in peptides selected against KLK5 can only be found N-terminally between the first and second cysteine.

[0165] Three bicyclic peptides per target and linker were synthesized on 25 μmol scale using 9-fluorenylmethyloxycarbonyl (Fmoc) solid-phase peptide synthesis (SPPS). The peptides were not synthesized as pure regioisomers, but if possible, the three isomers were separated during the second RP-HPLC purification. The yields of the pure compounds were between 1-5 mg per isomer, corresponding to yields between 12-60%. However, not for all peptides all three isomers could be separated, meaning they either did not form during the cyclization reaction due to steric reasons or they did not elute separately from the column. For this reason, the purity of some peptides was lower than 95%. Thus, one must consider that the measured activity of the peptide might be not precise, for examples if the isomers could not be separated at all. For this reason, the measured activity values were only taken as indicative and differences of less than one order of magnitude were not considered.

[0166] The inhibitory activity of the peptides was tested in substrate based-activity assays. The inhibitory constant (K) values of the best isomer or fraction that could be separated is shown next to the corresponding peptide sequences (FIG. 3). With K.sub.i values as low as 2 nM, inhibitors in the low nanomolar range could be successfully identified for KLK5 (peptide 001, 002, 004). With K values as low as 9 nM, inhibitors in the low-nanomolar range were also found for KLK7 (peptide 011). The high potencies of these bicyclic inhibitors confirm the successful phage display selections, and the identified inhibitors provided a good starting point for the development of KLK5- and KLK7-targeted therapeutics.

[0167] As the consensus sequence covers less than half of the 18 amino acids of the peptides, great large of each peptide probably do not contribute to target binding. It was previously shown that the peptides of one consensus group are selected as any of three possible isomers and that this selected isomer shows the strongest inhibitory activity. However, as the cysteine connectivity is not genetically encoded, it has to be unraveled by synthesizing all three isomers using orthogonal cysteine protecting groups and testing their activity. For all obtained consensus groups, it is apparent that the consensus sequence was located between two cysteines, of which one is a fixed cysteine and the other a variable cysteine. This suggests that the two, consensus surrounding cysteines are connected, and peptides were selected as isomer one (C.sub.1-C.sub.2 and C.sub.3-C.sub.4 connected). Isomer one represents the least constrained scaffold, with two, separate cycles. As a consensus sequence can only be found in one cycle, it seems that only one cycle provides the major binding interaction while the rest of the peptide does not contribute considerably to target binding. Another indicator for this was that the second variable cysteine in each sequence appeared in random positions. This suggests that monocyclic peptides might suffice for target binding and provide the same inhibitory activities as the synthesized bicyclic peptides.

[0168] Based on the hypothesis that all binders were selected as isomer one, the ring sizes of the consensus containing monocycles were analyzed. Even though various ring sizes appeared in the sequences, the minimal ring size that could be found was linker specific. It was apparent from the consensus for KLK5 as well as for KLK7 that the smallest cycles selected with 2,6-bis(bromomethyl)pyridine contained six amino acids in the ring, while when selected with 1,3-dibromoacetone they contained at least seven amino acids. It has been described previously that macrocycles with similar binding motifs often have comparable ring sizes, which suggests a correlation between a certain binding motif and its cycle topology. This would explain that with 2,6-bis(bromomethyl)pyridine as the linker no cycles with less than six amino acids were selected, while with 1,3-dibromoacetone, which is two carbons shorter, the smallest ring size was seven amino acids.

[0169] Based on this assumption, the consensus sequences of the smallest monocyclic peptides, which most likely are sufficient to provide the major binding interactions, were derived from the initially discovered sequences. To understand which amino acids are tolerated in the different positions in this smallest found macrocycles depending on target and linker, they were visualized as sequence maps (FIG. 4). The smallest possible cycles were of particular interest, as in contrast to the bigger cycles, since most of the positions in these cycles showed conserved amino acids. Additionally, their smaller size makes them more attractive for drug development than their bigger, initially selected, double-bridged progenitors, as their synthesis is much easier. Thus, they can also be easier conjugated and combined with other moieties, which was important for the further development.

EXAMPLE 2—STRUCTURE ACTIVITY RELATIONSHIP AND AFFINITY IMPROVEMENT

[0170] Development of Monocyclic Peptide Inhibitors

[0171] To validate the hypothesis that only one macrocycle of the peptide mediates target binding, several monocyclic inhibitors of KLK5 and KLK7 were synthesized (FIG. 5a). If different amino acids appeared at one position in the consensus, based on the consensus maps (FIG. 4) peptides with these variants were synthesized to assess which amino acids provided the best activity. Additionally, it was tried to analyze the impact of the two extra-cyclic positions, by synthesizing peptides lacking these amino acids (peptide 108, 109 and 114). For the KLK5 inhibitors, where the cycle size was not as clearly conserved as for the KLK7 inhibitors, several peptides with increasing cycle size were synthesized to assess the impact of cycle size on affinity (peptide 031, 051, 034 and 054).

[0172] Peptides could be successfully synthesized on 25 μmol scale with high purities of more than 95% for all peptides except of peptide 081, where an N-terminal glutamine formed pyroglutamate for around 50% of the product.

[0173] With K.sub.i values of 2.2 nM for KLK5 (peptide 035) and 16 nM for KLK7 (peptide 096), monocyclic peptides inhibited their targets with a similar efficacy as their bicyclic progenitors, while having a much smaller molecular weight of only 1 kDa versus 3 kDa for the bicyclic peptides.

[0174] For both targets, peptides based on 2,6-bis(bromomethyl)pyridine showed inhibitory activities in the same range as peptides based on 1,3-dibromoacetone. It was decided to only continue the development of 1,3-dibromoacetone based peptides. Nevertheless, the development of inhibitors based on two different linkers was valuable until this stage as it provided entirely independent controls, and as the discovered peptides are even similar, the information provided seems to be transferable between the inhibitors based on the two different linkers—in particular for position one to five.

[0175] The structure-activity relationship (SAR) analysis provided information that allowed us to narrow in on the best inhibitor based on the consensus sequences. In the fifth position of 1,3-dibromoacetone-based KLK5 inhibitors, glutamine had better K.sub.i values than glutamic acid. Additionally, glutamine was favored due to its neutral charge. In the sixth position, tyrosine outperformed tryptophan. Inhibitors with a C-terminal extra-cyclic amino acid or a bigger ring size did not greatly improve the inhibitory activity (1.15-fold and 1.47-fold, respectively) and thus peptide 035, including a glutamine, a tyrosine, and the smaller ring size for molecular weight purposes, was chosen as the lead for the further development of KLK5 inhibitors (FIG. 5b).

[0176] For the KLK7, several amino acids were compared, especially in the first, fifth, and sixth position. In the first, N-terminal, extra-cyclic position, glutamine and arginine had better K.sub.i values than proline. Nevertheless, proline was chosen for the lead peptide, as arginine could lead to stability issues, while a free, N-terminal glutamine can cyclize with the free —NH.sub.2 to form pyroglutamate and thus would cause purity problems during synthesis. In position five, leucine displayed the best activity after methionine, which was avoided due to oxidation reasons. In position six, which was not very well conserved in the sequences, serine, threonine and alanine were compared. With both linkers, alanine provided a great improvement in activity compared to the polar amino acids. Peptide 096, which has as an extra-cyclic, C-terminal serine showed a 2-fold better activity than peptide 114 without serine, and thus peptide 096 was chosen as lead for the further development of KLK7 inhibitors (FIG. 5b).

[0177] Affinity Improvement with Unnatural Amino Acids

[0178] To exploit the so far untouched chemical space of non-natural amino acids for affinity improvement, peptides with non-natural side-chain derivatives in one position were synthesized and tested on their impact on inhibitory activity. Unnatural derivatives were identified using SciFinder (https://scifinder.cas.org), searching for commercially available compounds with up to 90% similarity to the N-Fmoc-protected, natural derivative. Based the output of this search, derivatives were manually chosen based on availability and price. The structures of the chosen and tested unnatural amino acids are shown in FIG. 6c. Additionally to the unnatural derivatives, for the KLK5 inhibitor Ser and Lys were tested in position 3 and 4, respectively. Peptide 035 served as the reference for the KLK5 inhibitor and the side chain of one residue was changed at a time. For the KLK7 inhibitor, peptide 114 was used and not lead peptide 096 as reference because the C-terminal serine of peptide 096 had to be removed later for stability reasons.

[0179] As for the natural monocyclic inhibitors, the peptides were synthesized on 25 μmol scale. High purities of more than 95% could be obtained for almost all peptides with unnatural side-chain derivatives.

[0180] For KLK5, no derivatives providing any further improvement in inhibitory activity could be identified (FIG. 6a). As it was already partially apparent from the consensus, changes in positions three and four, which provide the non-primed side residues, led to big losses in the inhibitory activity, while positions five and six, which provide primed side residues, are more tolerant to structural changes. Position seven is particularly sensitive to mutations to D-amino acids and therefore probably has an important impact on the topology of the macrocycle.

[0181] Also for the KLK7 inhibitor, positions three and four were highly sensitive to structural changes, while positions five and six, were more tolerant. In position five, one derivative that improves the inhibitory activity was found: Norleucine exploits the chemical space between the previously used leucine and methionine, which also appeared in the consensus at position five but was avoided due to oxidation concerns and improves the K.sub.i value to 7±0.7 nM (FIG. 6b).

[0182] This part of the SAR study confirmed that our developed KLK5 and KLK7 inhibitors both exploit a standard mechanism of inhibition, binding the active site like a substrate. The results confirmed the assumption made base on the phage display results, that Arg 4 and Tyr 4 provide the P1 residues in the KLK5 and KLK7 inhibitor, respectively, as they were the residues which were most conserved in the phage display consensus, and also the residues most sensitive to structural changes in the affinity maturation screen. Furthermore, this means, that also the other respective residues of the two inhibitors most likely bind to homologues subsites in the two proteases. The identified residues match well with previously in literature described subsite preferences of KLK5 and KLK7 and thus allow a clear assignment of most of the side chains to the subsites of their targets.

[0183] Selectivity

[0184] Even though phage-selected peptides are in general specific for their targets and usually provide selective inhibitors, it was desired to confirm this for the molecule described herein. The inhibitory activity of the lead KLK5 inhibitor was tested in substrate-based activity assays against a group of various trypsin-like serine proteases that were available in the lab.

[0185] None of the seven tested proteases inhibited KLK5 at nanomolar concentrations, demonstrating a high selectivity of the KLK5 inhibitor (FIG. 7). FXIIa showed a weak off-target inhibition with a K.sub.i value of 40 μM, which is around 4300-fold weaker than the K.sub.i value for KLK5. Furthermore, trypsin is inhibited at almost the same concentration, which shows that this is not a specific inhibition, but already in the concentration range where many trypsin-like serine proteases might be unspecifically inhibited by arginine containing peptides.

[0186] This shows that our lead KLK5 inhibitor is selective over distantly related proteases.

EXAMPLE 3—HALF-LIFE EXTENSION AND PHARMACOKINETICS

[0187] Conjugation to Albumin Tag

[0188] Even though peptides furnish potent binders, they typically suffer in vivo from short half-lives due to fast renal clearance. To still profit from their enormous potential for therapeutic applications, several strategies to improve their half-live have been applied to peptide-based molecules.

[0189] Non-covalent albumin binding was chosen as half-life extension strategy for our tissue kallikrein inhibitors, the rationale being that the inhibitors remain small in size and promise good diffusion into tissue such as skin. To equip our inhibitors with an albumin binding moiety, it was decided to use an albumin-binding ligand recently developed in our lab (called “albumin tag”), which is based on a fatty acid peptide chimera, has a molecular weight of 2 kDa, and binds human serum albumin (HSA) with low nanomolar affinity of 50 nM (FIG. 8a).

[0190] Conjugation of inhibitors to an albumin ligand might affect the binding constants of the inhibitors. In the past was observed, for example with a FXII inhibitor, that conjugation or binding to high molecular weight proteins like albumin can dramatically reduce their activity. One reason might be that an increase in molecular weight reduces the diffusion rate and thus affects the k.sub.on value of an inhibitor. Furthermore, the increase in size might have steric impacts or intermolecular interactions with albumin can also impair the binding constants. While an increase in molecular weight is difficult to circumvent, steric impacts can be minimized by choosing the best site for conjugation.

[0191] It was decided to test two different conjugation strategies and to compare which results in molecules with better properties. The KLK5 inhibitor 035 and the KLK7 inhibitor 096 were conjugated N-terminally and C-terminally through the sidechain of a lysine to the albumin tag (FIG. 8b).

[0192] To synthesize the conjugates the previously in the lab of the inventors established synthesis protocol based on Fmoc SPPS was used.

[0193] N-terminal conjugation was realized by first synthesizing the linear inhibitor and continuing the synthesis on its free N-terminus. First, a PEG.sub.2 unit was introduced as spacer to the tag. Then the heptapeptide sequence of the tag was attached. Dde (N-(1-(4,4-dimethyl-2,6-dioxocyclohexylidene)ethyl)) was used as protection group for the side chain of the Lys within the heptapeptide as it can be selectively removed using 1% hydrazine. However, Fmoc is not stable under these conditions, thus the N-terminus must be protected with protection groups stable in 1% hydrazine such as Boc (tert-butyloxycarbonyl). For the synthesis of the N-terminal conjugates this was not necessary as 5(6)-carboxyfluorescein (5(6)-FAM) was coupled to the N-terminus. Next, the Dde on the side chain of the Lys was selectively removed and palmitic acid coupled on this primary amine. Then the molecule was cleaved off form the solid phase and cyclization of the two free Cys with 1,3-dibromoacetone was performed on the cured product in solution. Then the peptides were purified by RP-HPLC.

[0194] For synthesizing C-terminal conjugates, the inhibitors were synthesized with a C-terminal Gly-Lys-Gly linker to attach the tag on the side chain of the Lys. Thus, Lys with Dde-protected side chain was used. As last amino acid at the N-terminus, N-Boc protected amino acids were introduced. Then the side chain of the Lys within the Gly-Lys-Gly linker was selectively deprotected with hydrazine and the synthesis of the tag continued until 5(6)-FAM. Again, Lys(Dde) was integrate in the middle of the heptapeptide. In the last step, the side chain of this Lys was selectively deprotected and coupled with palmitic acid. After cleavage, cyclization with 1,3-dibromoacetone was performed in solution and the peptides were purified by RP-HPLC.

[0195] The inhibitors were synthesized on 50 μmol scale and around 15 mg pure peptide could be obtained, giving a satisfactory yield of around 10% regarding the length of the synthesis.

[0196] After successful synthesis of the molecules, the K.sub.i values for KLK5 or KLK7 in the presence of 25 μM HSA were determined (FIG. 9a). At this albumin concentration more than 99% of the inhibitor is bound to albumin (at the highest inhibitor concentration), so that the K value realistically represent the inhibitory activity of the molecule in the presence of albumin. Even though the actual concentration of albumin in human blood is at around 600 μM and thus much higher, a higher albumin concentration in the assay was not possible, as it interfered with the protease activity.

[0197] For the N-terminal conjugates a K.sub.i value of 23±5 nM was measured for KLK5 and 106±12 nM for KLK7. In comparison to the cyclic peptide alone, this means losses in the activity of both inhibitors with a particularly strong, 10-fold loss observable for the KLK5 inhibitor and a 6.6-fold loss for the KLK7 inhibitor. In contrast, C-terminal conjugation preserved the activity of both inhibitors at 1.2±0.0 nM for KLK5 and 32±6 nM for KLK7. Regarding the SAR, this confirms the contribution of the free —NH.sub.2 of the N-terminus to target binding, particularly for the KLK5 inhibitor. This assumption was first made based on the consensus sequences and could already be confirmed by the affinity optimization study. What seems rather surprising is that instead of the expected losses in inhibitory activity, a 1.8-fold improvement in the K.sub.i value could be noted for the KLK5 inhibitor. A viable explanation for this improvement is that the peptide backbone of the C-terminally attached Gly-Lys-Gly linker or the tag contributes to KLK5 binding.

[0198] Previous work showed, that conjugation also can lead to losses in the affinity for albumin. Thus, the dissociation constants (K.sub.d) of the molecules for HSA were tested with a fluorescence polarization (FP) assay using the fluorescein moiety in the albumin tag. The fluorescence anisotropy of the molecules in a dilution series of human serum was measured and Kd values were calculated. The K.sub.d value for HSA of all conjugates could be mostly preserved, independent of the conjugation strategy (FIG. 9b). With acceptable 2.1 and 2.9-fold losses, the C-terminally conjugated KLK5 and KLK7 inhibitors, reached final K.sub.d values of 119±9 nM and 164±12 nM for HSA, respectively. The losses in affinity seem to be more dependent on the conjugated inhibitor than on the terminus that was used for conjugation. This might be due to intramolecular interactions of amino acids in the inhibitor with the tag. Hydrophobic amino acids like in the KLK7 inhibitor might interact with the fatty acid moiety, which could slightly impair albumin binding.

[0199] Based on these results for both inhibitors the C-terminus was identified as suitable for conjugation without losses in potency and thus the C-terminally conjugated inhibitors were chosen for further development.

[0200] Plasma Stability of Inhibitors

[0201] The metabolic stability of the inhibitor-tag conjugates in human blood plasma was tested and compared to the inhibitors alone (FIG. 10). The peptides were added to human blood plasma at a final concentration of 80 μM and were incubated at 37° C. The relative concentration of intact peptide was assessed over time by LC-MS. Both inhibitors without the albumin tag showed a very short plasma half-lives in the range of minutes. In contrast, the half-live of both inhibitors with albumin tag was greatly prolonged. The KLK5 inhibitor 035-tag had a half-life of 90 h, meaning that the attachment of the tag extended it more than 1,00-fold (FIG. 10a). The KLK7 inhibitor 096-tag showed a half-life of 30 min. In this case the conjugation to the tag extended the half-life only 6-fold, but still significantly (FIG. 10b). While the half-life of the KLK5 inhibitor was already satisfying and suited for in vivo use, the KLK7 inhibitor had to be further engineer to achieve comparable stability. As both molecules are similar except for the exact amino acid sequences inhibitors, the result exemplifies the major impact of the amino acid sequence on the proteolytic stability of molecules without secondary structure.

[0202] Stability Improvement of KLK7 Inhibitor

[0203] To further improve the proteolytic stability of the KLK7 inhibitor 096-tag, its degradation pathway was analyzed by identifying degradation products in the LC-MS data-set of the stability experiment. Subsequently, vulnerable peptide bonds were identified and amino acids in this region modified (FIG. 11).

[0204] A clearly identifiable proteolytic fragment of the KLK7 inhibitor 096-tag was the cyclic peptide alone, with hydrolysis occurring after the C-terminal serine that connected it to the GKG linker and albumin tag (FIG. 11a). To improve this link, serine 9 was removed from the molecule as it only contributes a factor of 2 to the target binding when comparing peptide 096 and peptide 114 in the monocycle SAR screening (FIG. 5a). A new KLK7 inhibitor without serine 9 (214-tag) was synthesized and tested for its stability in human plasma. It had a 10-fold improved half-life of 4 hours. As affinity and stability improvement studies were conducted in parallel, it is noteworthy that in this peptide also leucine 5 was changed to norleucine, which was found to improve the affinity but was not expected to change the stability, but a contribution to the stability improvement cannot be entirely excluded.

[0205] As plasma half-life of 4 hours was still not satisfactory for therapeutic application the plasma stability was further improved. Analysis of degradation fragments of peptide 214-tag revealed that cleavage of the amide bonds on both sides of arginine 7 causes proteolytic instability (FIG. 11b). In the affinity optimization study with unnatural amino acids, derivatives for arginine 7 that do not reduce the inhibitory activity were already identified: 4-guanidino-phenylalanine and homoarginine. As homoarginine provides the economically favorable alternative from a manufacturing point of view, this derivative was chosen. Furthermore, it was compared to glutamine and its derivative citrulline, which also appeared in the consensus but had not yet been tested. When testing inhibitors with alternative amino acids in position 7, all three alternatives to arginine provided a significantly better stability with half-lives of more than 96 hours. However, from these three alternatives, only homoarginine (peptide 278-tag) did not negatively impair the inhibitory activity (FIG. 11c), so peptide 278-tag with homoarginine instead of arginine in position 7 was chosen as the final KLK7 inhibitor.

[0206] After this stability optimization, both inhibitors had a metabolic half-life of more than 90 hours in human plasma and were considered suited for in vivo studies regarding their inhibitory activity and plasma stability.

[0207] Replacement of Fluorescein in Albumin Tag

[0208] Because the future application of these molecules will be therapeutics, they should have suitable pharmaceutical properties. Regarding the structures of both inhibitors, however, the fluorescein moiety in the albumin tag is not ideal for this application. Even though fluorescein is FDA approved, it has been reported to potentially cause skin discolorations in humans after only a single injection for angiography. Therefore, it is probably not ideal to have a fluorescein moiety in molecules which are intended for chronic therapy in regards of patient compliance, as the chronic application of the molecule might cause discolorations.

[0209] It has previously been shown in Zoro eat al. (2017), DOI: 10.1038/ncomms16092 that the fluorescein moiety—even though initially installed for detection reasons only—strongly contributes to HSA binding. Removal or displacement of fluorescein from the albumin tag is associated with an around 30-fold loss in affinity for HSA. Even slight changes in its position in the albumin tag lead to losses in affinity for HSA. It was searched for a solution to eliminate fluorescein from the molecules without losing affinity for HAS, while the tag should still be easily synthesizable using SPPS and commercially available building blocks. To minimize the number of molecules that had to be synthesized and tested, this issue was addressed with an initial SAR experiment and subsequent rational design.

[0210] To get an idea of which moieties of fluorescein contribute to albumin binding, several known albumin-binding small molecules were tested in a competitive FP assay to identify those that compete with the tag for albumin binding. It is known, that albumin has seven binding sites for fatty acids with 10 to 18 carbons and two small molecule binding sites, called ‘drug site one’ and ‘drug site two’. ‘Drug site one’ was characterize to accommodate bulky heterocycles with negative charge, while for ‘drug site two’ aromatic carboxylic acids fit best. A set of molecules was chose that bind albumin in ‘drugs site one’ or ‘drug site two’ with reported affinities (FIG. 12a). The results showed that the drug site two binders, diflunisal and diclofenac outcompeted the albumin tag at lower concentrations than all drug site one binders (FIG. 12b). As diflunisal is one of the rather weaker albumin binders tested, but the best competitor, it could be excluded that this is due to a higher affinity for albumin. However, as diflunisal and diclofenac have multiple binding sites in HSA, it is not clear from this assay whether they compete with the fluorescein or the acylated heptapeptide part of the tag. Thus, the binding site of fluorescein in HSA in context of the albumin tag cannot be clearly identified. However, as the only common moieties in both molecules are two aromatic rings and a carboxylic acid, which are moieties that can also be found in fluorescein, it was hypothesized, that these are the three moieties in fluorescein that form the main interaction with albumin in context of the tag (FIG. 12c).

[0211] To design molecules that could provide similar target interactions as fluorescein, it was attempted to replace it with a free carboxylic acid and two aromatic rings. The strategy was to reintroduce these three groups one by one, first optimizing their position and then introducing the next group. Thus, three generations of molecules were synthesized, starting with the carboxylic acid, followed by a first, and then a second benzoic ring.

[0212] To be able to measure the affinity of these tag variants for HSA via FP assays, fluorescein was still required in the molecules. Thus, the synthesis of the tag had to be slightly changed and fluorescein was attached to the side chain of an additional lysine at the C-terminus of the tag. First, Fmoc-Lys(Dde) was coupled on the solid support. Then the heptapeptide sequence with standard Lys(Boc) in heptapeptide sequence was attached. After that, the N-terminal variants were coupled to the amine of the PEG.sub.2 unit. Then the side chain of the C-terminal Lys was selectively deprotected and 5(6)-FAM was coupled on solid phase. After cleavage, palmitic acid was coupled to the side chain of the Lys within the heptapeptide sequence using palmitic acid NHS-ester.

[0213] The peptides could be successfully synthesized on 25 μmol scale and satisfactory purities could be obtained. The two control peptides F-tag and tag-F were previously synthesized and their identity and purity were confirmed.

[0214] In order to reintroduce only the carboxylic acid, butanedioic acid and pentanedioic acid were appended to the N-terminus of the tag separated by a PEG.sub.2 unit as previously the 5(6)-FAM (peptide 262, 263). Pentanedioic acid performed slightly better than butanedioic acid and already recovered a factor of two in the HSA affinity compared to tag-F. To introduce the first aromatic ring while keeping the carboxylic acid in a comparable position, two strategies each required two coupling steps were compared: either benzenedicarboxylic acid or benzoic acid was coupled to the amine of a gamma-coupled glutamic acid (peptide 264 and 266). The molecule assembled from the benzoic acid (peptide 266) performed better, and the first aromatic ring again recovered another factor of two in affinity. Based on peptide 266, four variants introducing a second aromatic ring were compared (peptide 282, 283, 284, 285). Out of these, biphenyl-3-carboxylic (peptide 282) and 2-naphthalenecarboxylic acid (peptide 284) performed the best, and both improved the affinity by another factor of four. With K.sub.d values of 132±18 nM and 147±8 nM, respectively, the two tag variants 282 and 284 afford an affinity for HSA comparable to the fluorescein-tag (F-tag). Peptide 289 confirms the binding of the fluorescein competitor diflunisal in context of the albumin tag.

[0215] Both dye-free tag alternatives are SPPS compatible, require only cheap, commercially available building blocks and one extra coupling step. Thus, they provide dye-free albumin tag variants suitable for therapeutic application. Additionally, the whole set of tag variants allows for a tunable affinity for HSA in the range of 100-1000 nM.

[0216] Final Inhibitors

[0217] The structures of the final KLK5 inhibitor 035-tag and the final KLK7 inhibitor 278-tag after in vitro optimization, which were used in the in vivo studies, are shown in FIG. 13. The conjugates contain the cyclic peptides found in the phage display selections. The KLK7 inhibitor was further improved with two unnatural amino acid derivatives: norleucine was incorporated in position 5 for affinity reasons and homoarginine in position 7 for stability reasons. Both cyclic peptide inhibitors are C-terminally prolonged by a Gly-Lys-Gly tripeptide serving as linker for the installation of a previously developed albumin tag through the side chain of the lysine. The albumin tag consists of a PEG.sub.2 unit, followed by the heptapeptide Glu-Tyr-Glu-Lys-Glu-Tyr-Glu (SEQ ID NO: 43) with the side chain of lysine acylated with palmitic acid, and a N-terminal fluorescein moiety linked through another PEG.sub.2 unit. Derivates of the N-terminal fluorescein moiety were developed but have not been incorporated yet due to the need for sensitive detection in the various in vivo studies. The inhibitors have respective K.sub.i values of 1.2±0.0 nM for KLK5 and 7.4±1.3 nM for KLK7. They bind HSA with K.sub.d values of 119±8 nM (KLK5 inhibitor) and 164±12 nM (KLK7 inhibitor). The albumin binding moiety equips the inhibitors with ex vivo plasma half-lives of more than 90 hours in human plasma. As during the pharmacokinetic and efficacy studies of this work sensitive detection of the molecules was still essential and the fluorescein-containing albumin tag was used.

[0218] Pharmacokinetics in Mice

[0219] Before the efficacy of the molecules could be tested in a disease model, their pharmacokinetic profiles had to be analyzed in vivo in wild-type mice. Mice were chosen for the first pharmacokinetic studies as they represent the model organism that will be used for the efficacy studies. The results of the pharmacokinetic studies determine the dosing and administration route for the efficacy studies.

[0220] The albumin binding property of the peptides is expected to prevent fast renal clearance of the inhibitors in vivo. However, the albumin tag was developed to have a high affinity for human albumin and thus might have a weaker affinity for mouse serum albumin (MSA). Before performing mouse pharmacokinetic experiments, the affinity of the albumin tag for MSA was determined. The affinity for rat serum albumin (RSA) was previously tested, and a K.sub.d value of 220 nM, around 5-times weaker than for HSA, was measured.sup.116. The K.sub.d value of the F-tag for MSA was 4.2 μM and thus two orders of magnitude higher than for HSA, indicating a much weaker affinity of the albumin tag for MSA than for HSA (FIG. 14).

[0221] In order to understand which parts of the tag not forming interactions with MSA, additional control peptides were tested, such as a tag composed of Ser residues (SSSKSSS-tag, SEQ ID NO: 44) and the tag with fluorescein at the C-terminus (tag-F). Additionally, the KLK5 inhibitor 035-tag was tested. With K.sub.d values between 7 and 10 μM, the control peptides as well as the inhibitor did not perform much worse than the F-tag. The conclusion was that only the fatty acid moiety of the albumin tag efficiently binds MSA, while it was shown previously, that the heptapeptide as well as the fluorescein moiety of the tag contribute to HSA binding.

[0222] For this reason, the albumin tag is not ideal for mouse studies and the inhibitors were expected to strongly underperform in mice compared to other species in terms of pharmacokinetics. These results made a pharmacokinetic pre-study before efficacy testing very important, to understand whether efficacy testing in mice was feasible at all, and if yes, how frequently injections would need to be performed to obtain an effect in vivo.

[0223] The pharmacokinetics of the KLK5 inhibitor 035-tag and KLK7 inhibitor 278-tag were investigated in wild-type C57BL/6J mice. Mice (n=3) were either injected intravenously (IV), intraperitoneal (IP) and subcutaneously (SC) with 100 μL KLK5 inhibitor 035-tag at a concentration of 400 μM and IV (n=3) with the same dose of KLK7 inhibitor 278-tag. For both inhibitors, this corresponds to a dose of 6.2 mg/kg for a mouse of 20 g. For the IV injection this dose was expected to result in an inhibitor blood concentration of 20 μM, assuming 2 mL of blood, and a plasma concentration of 40 μM, assuming 1 mL of plasma. After injection, the concentration of the inhibitor in plasma was analyzed at different time points over a period of 24 hours. Blood samples were taken from the tail vain into EDTA coated tubes, plasma proteins were precipitated, and the supernatant was analyzed by HPLC equipped with a fluorescence detector. After IV injection, the KLK5 inhibitor 035-tag and the KLK7 inhibitor 278-tag inhibitor showed an elimination half-life of 3.9 and 5.9 h, respectively (FIG. 15a). The residual plasma concentrations 24 hours post injection were around 0.2 μM for the KLK5 inhibitor and around 1 μM for the KLK7 inhibitor, which seems to have a slightly longer half-live in vivo. For the KLK5 inhibitor 035-tag, SC and IP injection routes showed lower peak concentrations and higher residual concentrations 24 hours post injection with around 0.6 μM.

[0224] Two degradation fragments could be observed in the HPLC tracks (FIG. 15b): While the intact peptide elutes at 18 min, one fragment elutes at 21 min and the other one at 22 min. Both fragments were also overserved when testing the stability of peptide 035-tag in mouse plasma in vitro. Both fragments must still contain the fluorescein moiety as they were detected with the fluorescence detector. Furthermore, they must still contain the fatty acid as they elute later from the column than the intact inhibitor and thus are more hydrophobic. Further analysis revealed, that the main fragment, which elutes at 21 min is probably a fragment of the tag, cleaved after the second Tyr within the heptapeptide sequence. By replacing this Tyr with Glu (280-tag) the degradation fragment disappeared. However, the stability of the peptide in mouse plasma could not be improved through this replacement. That the peptide part of the tag is metabolically less stable in mouse plasma than in human plasma is most likely also due to the weaker affinity of the tag for MSA than for HSA.

[0225] To test, whether the inhibitor would have a longer half-life, if it had a better affinity for MSA, a subcutaneous co-injection with one equivalent HSA was performed. The results showed that co-injection with HSA prolongs the absorption phase and the elimination half-life to 7 hours. A plasma concentration of 0.1 μM could still be measured 48 hours post injection. RSA, used as a control, only influenced the absorption phase, but the elimination half-life was not prolonged (FIG. 15c).

[0226] These results demonstrate the strong relationship between HSA affinity and half-life. It suggests that the half-life of the inhibitors in other species might be better than would be expected by allometric scaling and will strongly depend on the affinity of the tag for the albumin of the respective species. Furthermore, it was demonstrated that the absorption phase can be prolonged through administration with adjuvants that bind the inhibitors and retard its distribution from the injection site.

EXAMPLE 4—SKIN DISTRIBUTION

[0227] Biodistribution to Skin and Organs

[0228] In addition to the plasma concentration, it was desired to assess the biodistribution of the inhibitors. The fluorescein moiety in the albumin tag theoretically allows to monitor the biodistribution with a sensitive CCD camera. However the emission signal of fluorescein (Em: 565 nm) is confined to superficial structures with <3 mm of tissue depth due to absorption and light scattering. To get an idea of the concentration of the inhibitor in skin in comparison to other organs, skin samples and all visceral organs were imaged.

[0229] Mice (n=3 per timepoint) were intravenously injected with the KLK5 inhibitor 035-tag, F-tag (without inhibitor), 035-fluorescein (non-HSA binding control), and PBS (background control) with a dose of 2 μmol/kg (corresponding to 6.2 mg/kg for the KLK5 inhibitor). Mice were imaged at various time points up to 48 h post injection (FIG. 16a). The signal of 035-fluorescein reached to almost background level only 2 hours after injection, indicating the rapid clearance of non-albumin-bound molecules. In contrast, the KLK5 inhibitor and albumin tag showed an increase in signal up to 8 hours post injection, and a significant signal over background was still detectable 48 hours post injection. First, the increase in the signal intensity up to 8 hours after IV injection confirms that superficial tissue and not the vascular system is being imaged, as IV injected molecules reach their peak concentration in the vascular system seconds after injection. In the first 4-8 hours post IV injection, a strong drop in concentration was observed in the blood. This imaging experiment confirmed that the molecules are not eliminated but instead distribute to the tissue in this time. An increase in signal originating from extravasation that was only observed for the two albumin binding molecules and not for the peptide with fluorescein only shows that extravasation to the tissue is strongly driven by albumin binding. This underlines the importance of the albumin tag for both the half-life as well as for the distribution of the molecules to the tissue.

[0230] To confirm the distribution of the inhibitor to the skin, mice were sacrificed at different time points (1 h, 4 h, 12 h) after IV injection and imaged the isolated skin (FIG. 16b). The strongest signal in the skin was detected in the 4 hour sample, matching with the observed peak concentration in the tissue between 4 and 8 hours post injection.

[0231] To compare the distribution of the molecule between skin and visceral organs, mice were euthanized 48 hours after injection of 12.4 mg/kg of inhibitor, and the organs were collected and imaged. The strongest fluorescence signal was observed in the skin, while only a low signal intensity was measured in the liver, spleen, kidneys, heart and lung. The gastro-intestinal tract showed moderate fluorescence compared to the skin and other organs (FIG. 16c). These results indicate that the molecule is secreted through the digestive tract rather than the kidneys, which can also be attributed to albumin binding.

[0232] Taken together, these data strongly indicate that the inhibitor not only reaches, but primarily distributes, to the skin. This favorable biodistribution can be attributed to the non-covalent albumin-binding strategy. Based on the pharmacokinetic and biodistribution studies, the optimal dose, administration route, and frequency of injection were determined to be 3×/week subcutaneous injections of 6.2 mg/kg of inhibitor for the first efficacy study.

[0233] Skin Distribution of Peptide Versus Antibody

[0234] The in vivo fluorescence imaging experiment confirmed that the inhibitors reach the skin, though this is not sufficient evidence that they also reach KLK5 and KLK7, which are located in the outer layers of the epidermis. The skin tissue, which can be separated from the mouse and has been imaged in the previous experiment, comprises not only the epidermis but also the vascularized dermis, and subcutaneous muscular and adipose tissue, which is easier to reach through systemic administration. Even reaching the epidermis is not sufficient, as the two kallikreins are located in the extracellular space of the outermost layer of the epidermis, the stratum corneum, which consists of tightly connected corneocytes embedded in a lipid matrix. The barrier function of the stratum corneum that protects the body from external pathogens and chemicals and internally against dehydration makes it challenging to reach KLK5 and KLK7 from the vascular system.

[0235] To get an idea of how deep our inhibitors can penetrate the skin after systemic administration fluorescence microscopy of skin sections was performed. A second aim of this experiment was to compare whether the small, peptide-based drug modality has an advantage over bigger drug modalities, such as antibodies, in penetrating the epidermal tissue. As KLK5- or KLK7-specific antibodies could not be afforded in the required amounts, a monoclonal antibody was used that does not bind KLK5 or KLK7 specifically. mIgG anti-Flag was used, which binds the flag epitope (DYKDDDDK; SEQ ID NO: 45), an artificial epitope that is not present in mice. Given that KLK5 and KLK7 are present at concentrations of 2-4 and 7-14 ng/mg stratum corneum dry weight, respectively and it was aimed to inject 0.128 mg inhibitor or 6 mg antibody, it was assumed that the amount of injected molecule should significantly exceed the amount of available target in the organism. Therefore, the biodistribution should not be primarily driven by target binding, but rather by the biophysical properties of the drug modality. To exclude potential impact of target binding on the biodistribution of the peptide, additionally the distribution of the albumin tag without the inhibitor was studied. Additionally, the biodistribution of MSA was tested. This experiment was to see whether albumin can reach the epidermis, as the biodistribution of our molecules is influenced by albumin binding. Taken together, mIgG anti-Flag-SiR, MSA-SiR, 035-SiR-tag, and SiR-tag were compared on the skin distribution by fluorescence microscopy.

[0236] Previously fluorescein was used as a dye for in vivo imaging because fluorescein-containing molecules were already available, and it was only aimed to image superficial tissue layers. However, fluorescein is not well suited for microscopy due to the strong autofluorescence of tissue at its emission maximum of around 520 nm. Much better signal-to-background ratios can be obtained with red-shifted dyes, like the by Spirochrome developed silicon rhodamine (SiR) with an emission maximum at 674 nm. For this reason, SiR conjugates were used to assess the skin distribution of our molecules by microscopy. As only milligrams of the dye were available, the synthesis of the inhibitor and tag had to be slightly changed: The N-terminus of the inhibitor was acetylated and Boc-PEG.sub.2-OH was coupled as last building block on the tag. After Lys(Dde) deprotection and palmitic acid coupling, peptides were cleaved from the resin and RP-HPLC purified. Pure inhibitor or tag was coupled at the free N-terminus of the albumin tag using the SiR—NHS ester in DMSO with DIPEA, and it was purified again by RP-HPLC. The chemical structure of the final inhibitor-SiR-tag conjugate is shown in FIG. 17c. Antibody and albumin were randomly labeled with an excess of 5 equivalents SiR—NHS ester in sodium bicarbonate buffer, aiming for a degree of labeling (DOL) of one (FIG. 16a). However, for MSA, even with 20 equivalents, only a DOL of 0.4 could be obtained. The conjugates were characterized by SDS-PAGE, confirming their correct size and labeling (FIG. 17). The peptide-conjugates were additionally checked by LC-MS and analytical HPLC.

[0237] Equimolar doses of 2 μmol/kg for all molecules were injected. Additionally, dye only was used as background control. As the mice had a weight of around 20 g, 200 μl at a concentration of 200 μM were injected. These doses corresponds to 300 mg/kg of antibody or MSA, 6.4 mg/kg of 035-SiR-tag, 3.6 mg/kg of SiR-tag, and 1.0 mg/kg of SiR. The concentrations of the injection solutions were calculated based on labeled protein as determined by absorbance at 652 nm, so that the lower DOL of MSA was considered.

[0238] Mice (n=1) were injected IP with prepared solution. It was aimed at analyzing the skin distribution at the time when a high concentration of molecule in the skin is expected. Thus mice injected with antibody or MSA were sacrificed 24 hours after injection, while mice injected with inhibitor, tag, or SiR were sacrificed 8 hours after injection. 8 hours correspond to the previously overserved peak concentrations of the inhibitors in the tissue. The injection with dye only served as a background control, as the dye was not expected to distribute to the tissue and should get eliminated within 2 hours after injection, as it was previously observed for fluorescein. Ventral skin tissue was isolated, fixed for 30 min in a PFA solution to prevent diffusion, cryopreserved, and analyzed by fluorescence microscopy.

[0239] Regarding the section overviews of all five samples, the SiR signal from the injected molecules was only visible for the antibody, the inhibitor, and the tag, but not for MSA (FIG. 18a). As the MSA-SiR conjugate was characterized for size and labeling, either a problem with the injection or an unexpected weak distribution to the tissue could explain the unexpectedly weak signal. As only one mouse was injected per molecule, issues with the injection cannot be excluded, and more replicates of the experiment would give further information. As the antibody, inhibitor, tag, and SiR were visible, their distribution in the skin was further analyzed by having a closer look at the dermis and epidermis. The inhibitor and the tag both showed a strong signal intensity particularly localized to the epidermis. In contrast, the distribution of the antibody seemed to be limited to the dermis and the signal intensity decreased towards the outer layers (FIG. 18b). To confirm the stronger epidermal localization of the inhibitor as opposed to the antibody, intensity profile plots were generated along arrows through the epidermis and dermis (FIG. 18c). In these plots, the beginning of the epidermis is identifiable by the strong DAPI signal originating from the compact, mono-layered stratum basale. While the signal of the antibody was very weak in the epidermis and did not overlay with the DAPI signal, the signal originating from the inhibitor was strong and completely overlaid with the DAPI signal originating from the epidermis/stratum basale.

[0240] These initial skin distribution data suggest a clear advantage of our molecular format as opposed to a standard IgG in terms of epidermal targeting. Two apparent differences in the molecules might contribute to their very different skin distribution: First, they have very different biophysical properties. While the 3-kDa inhibitor has only 2% of the size of an IgG, it is also much more hydrophobic due to its fatty acid moiety. The small size might be advantageous for diffusing into the extracellular space of the epidermis, as all keratinocytes are tightly connected through desmosomes, which might limit the accessibility of large molecules. Further the hydrophobicity might be an advantage as the keratinocytes are embedded into a matrix of various lipids, which might limit the accessibility of hydrophilic molecules. The second difference in the molecules that might influence their skin distribution is their mode of biodistribution: While the inhibitor is distributed with albumin, the distribution of IgG is mostly driven by its Fc region. Both, albumin and IgG can bind the FcRn receptor, but they bind to different sites with different affinities, which, for example leads to different levels of transplacental transport and might also lead to different levels of transport to the epidermis. It is noteworthy that this experiment should be repeated with more replicates to confirm these results.

EXAMPLE 5—EFFICACY STUDIES

[0241] Inhibition of Mouse KLKs

[0242] To be able to test the efficacy of the inhibitors in NS mouse models other than Tg-hKLK5, which only express mKLKs, it was important that mKLK5 and mKLK7 were inhibited. For this reason, the IC.sub.50 values of the inhibitors for mKLK5 and mKLK7 had to be determined.

[0243] The IC.sub.50 value of peptide 035-tag for mKLK5 is 2 nM and the IC.sub.50 value of peptide 096-tag for mKLK7 is 20 nM (FIG. 19). For both inhibitors, the IC.sub.50 values for their respective mouse KLKs lie in the same range as for the human KLKs This confirms the structural comparison-based assumption that both orthologues are comparably inhabitable with the same peptide. Thus, the inhibitors are expected to show activity also in mouse models expressing mKLK5 and mKLK7.

[0244] Negative Control Inhibitors

[0245] Two inactive control peptides were synthesized for the in vivo efficacy studies. The inhibitors were rendered inactive by replacing the P2 amino acids with alanine and swapping the P1 and P2 positions.

[0246] Even though swapping the positions would be sufficient to render the inhibitors inactive, the additional amino acid change gives active and inactive peptides of different masses that allows the clear identification of the molecules by mass to be able detect contaminations or possible confusions also later in samples.

[0247] The inactive peptides (301 and 302) were compared to their corresponding active forms (035 and 278) on target inhibition. For both control peptides IC.sub.50 values >100 μM could be confirmed for their respective targets (FIG. 20). After confirming their inactivity, the peptides were synthesized with C-terminal albumin tag for the in vivo studies (301-tag and 302-tag).

[0248] Upscaling of Synthesis

[0249] To obtain sufficient material for in vivo efficacy studies, the synthesis was scaled up without changing the synthesis protocol (FIG. 9b). The whole synthesis involved 23 coupling steps, of which 22 were performed on solid phase and one in solution. 20 out of the 22 solid phase couplings are based on standard Fmoc-chemistry, while two are based on Dde-deprotection with hydrazine to branch the peptide chain.

[0250] The synthesis was performed in the scale of 6×50 μmol=300 μmol. This corresponded to 900 mg (M=3000 g/mol) and around 90 mg pure peptide could be obtained, giving a satisfactory yield of around 10% (FIG. 21a). The identity of the compound was validated by ESI-MS and the purity was analyzed with RP-HPLC. All peptides used for in vivo studies had a purity of >95% (FIG. 21b).

[0251] Formulation

[0252] For the in vivo studies, a suitable formulation had to be found. First, the solubility of both molecules in ddH.sub.2O was tested. 2 mg of peptide were dissolved in an increasing volume of ddH.sub.20, while visually observing whether the peptide dissolved or precipitated. While peptide 035-tag was entirely soluble already in 20 μl ddH.sub.2O, which gives a good water solubility of >100 mg/mL, peptide 278-tag was only poorly water soluble and did not entirely dissolve in 2 mL, giving a water solubility of <1 mg/mL (FIG. 22). As expected, the amino acid sequence in the inhibitor strongly impairs the water solubility, with more than half of the amino acid composition of the KLK7 inhibitor coming from hydrophobic amino acids.

[0253] For parenteral administration in animals, an aqueous solution with pH 7.4 and salts for tonicity is required. Various buffers were tested. A buffer published for the peptide drug semaglutide, which is based on disodium phosphate with propylene glycol as a tonicity agent and a pH of 7.4 was taken as starting point.sup.144. Then it was tried to be optimized further the buffer for both peptide by addition of SDS, PEG400, DMSO or HSA.

[0254] For the experiment, 10 mM peptide stock solutions in ddH.sub.2O with 15% DMSO were diluted into different buffers to a final concentration of 0.4 mM, which corresponds to the desired concentration of the injection solution. The dilution of the stock solution with buffer reduces the final DMSO concentration to 0.6%. The solution was centrifuged to identify and remove precipitated peptide, and the concentration of peptide in the supernatant was spectrophotometrically determined to quantify how much of the expected concentration of 0.4 mM could be reached (FIG. 23).

[0255] First, the reference buffer 1, was compared to PBS (buffer 2). Buffer 1 did provide a slightly better solubility for peptide 035-tag than PBS and the solubility to the required 0.4 mM could already be reached. Peptide 278-tag was comparably soluble in buffer 1 and PBS, but the aimed concentration of 0.4 mM could not be reached. Then it was tried to improve the solubility of peptide 278-tag by addition of SDS and PEG400 to PBS (buffer 3 and 4), but both rather reduced the solubility. Finally, the aimed 0.4 mM concentration of peptide 278-tag could be achieved by either increasing the DMSO concentration to final 1.6% (buffer 5 and 6) or adding 0.1 mM HSA (buffer 8).

[0256] For peptide 035-tag, buffer 1 with 0.6% final DMSO concentration was used for all studies. For peptide 278-tag, buffer 6 with a final DMSO concentration of 1.6% was used for the pharmacokinetic experiments, for which only single injections were performed and thus no DMSO toxicity was expected. However, for the efficacy studies wherein repetitive injections were planned to be performed, it was desired to avoid possible DMSO toxicity and used buffer 1 with additional 0.4 mM HSA and a final DMSO concentration of 0.6%. The preparation of the injection solution for in vivo experiments is shown in FIG. 24. The final DMSO concentration could be slightly further reduced to 1.4% (without HSA) or 0.4% (with 0.4 mM HSA) by preparing 10 mM stock solution in ddH.sub.2O with only 10% DMSO.

[0257] Efficacy Studies

[0258] To conduct efficacy studies, 0.4 mM injection solution of 30 mg KLK5 inhibitor 035-tag (KLK5IN) and 30 mg corresponding negative control 301-tag (KLK5IP) were prepared.

[0259] KLK5 Bicyclic Peptide Inhibitor Reduced Major Clinical NS Clinical Features in TghKLK5 Mice

[0260] TghKLK5 were subcutaneously injected with the KLK5 bicycle peptide inhibitor (KLK5IN) every 2 days for a period of 20 days following birth (FIG. 25A). Mice were euthanized at the age of 20 days according the Ethical European guidelines. A KLK5 inactive bicycle peptide inhibitor (KLK5IP) was injected in TghKLK5 as negative control for KLK5IN. In this manuscript, injected TghKLK5 mice were compared with untreated TghKLK5 and with WT mice. At 5 days of birth, injected TghKLK5 mice with KLK5IN showed a decreased number of scales and their skin appear to be more hydrated than those injected with KLK5IP. From 10 days t 20 days of aged the skin of injected TghKLK5 mice with KLK5IP displayed skin redness, numerous scales on their entire body and large crusts on their ears, while the injected TghKLK5 with KLK5IN showed a significant reduction NS clinical features (FIG. 25B). Injected TghKLK5 mice with KLK5IN displayed a transepidermal water loss (TEWL) close to that of WT mice in comparison to the untreated and injected TghKLK5 mice with KLK5IP, which showed a high hydric loss (FIG. 25C). In situ zymography assay using skin cryosections from TghKLK5 mice and casein-BODIPY-FL substrate showed that injected TghKLK5 mice with KLK5IN was associated with decreased caseinolytic activity in comparison to the untreated and injected TghKLK5 mice with KLK5IP (FIG. 25D).

[0261] NS Histological Features and Abnormal Proliferative Markers were Reduced by the Injection of KLK5 Bicyclic Peptide Inhibitor in TghKLK5 Mice

[0262] Histological analysis of skin sections from injected TghKLK5 mice with KLK5IN showed a strong reduction of NS histological features including acanthosis, parakeratosis and hyperkeratosis in comparison to those untreated or injected with KLK5IP. Histological features of injected TghKLK5 mice with KLK5IN were close to the WT (FIG. 26A). In addition a significant decreased epidermal thickness was observed in comparison to untreated TghKLK5 mice or injected TghKLK5 with the inactive peptide. The epidermal thickness of injected mice with KLK5IN was close to WT but remained still higher (FIG. 26B). Immunostaining of proliferative markers including Ki67 and keratin 14 (Krt14) as well as differentiation markers such as Filaggrin were performed to validate the improvement of NS histological features in Injected TghKLK5 mice with KLK5IN (FIG. 26C). Many keratinocytes on the basal layer of the epidermis were stained with Ki67 in skin section from untreated TghKLK5 mice or injected TghKLK5 mice with the inactive peptide, while the skin sections from injected TghKLK5 mice with the KLK5 inhibitor look similar to the WT with few numbers of stained keratinocytes. Increased number of stained keratinocytes with Krt14 was observed in untreated and injected TghKLK5 mice with KLK51P, while the number of stained cells with Krt14 was close to WT mice (FIG. 26C). No variation of Filaggrin expression was observed in skin sections from untreated or injected TghKLK5 mice with KLK5IN or KLK5IP in comparison to WT (FIG. 26C).

[0263] Immune Cells Infiltration was Impaired in Section from Injected TghKLK5 Mice with KLK5 Bicycle Peptide

[0264] Mast cell coloration with toluidine blue showed a significant increased of Mast cells in the dermis of skin section from untreated or injected TghKLK5 with the KLK5IP in comparison to WT. Increased number of Mast cells was strongly reduced in skin section from injected TghKLK5 mice with KLK51N but remained still higher than WT (FIGS. 27A and B). A significant increased number of neutrophils was detected in skin dermis of untreated or injected TghKLK5 mice with KLK51P in comparison to WT. In skin section from injected TghKLK5 mice with KLK5IN, no stained neutrophils or a few numbers of stained neutrophils was detected in skin section. Stained neutrophils did not localize in dermis unlike the untreated TghKLK5 but remained stocked in blood vessels with no infiltration in the dermis (FIG. 27C).

[0265] KLK5 Bicycle Peptide Strongly Decrease Inflammatory Gene Signature in Skin from TghKLK5 Mice.

[0266] To assess the impact of KLK5 inhibition by KLK5IN on skin from TghKLK5, RT-qPCR of several inflammatory genes including 11-36 cytokines, Ccl20 and S100a8 was performed. Increased expression of II-36a, II-36b, II-36g, Ccl20 and S100a8 observed in skin lysates from untreated or injected TghKLK5 with KLK5IP were strongly reduced in skin lysates from injected TghKLK5 with KLK5IN (FIG. 28A). Decreased expression of II-36a previously seen at mRNA level in skin lystates from injected TghKLK5 with KLK5IN was confirmed by SDS-PAGE/immunoblotting (FIG. 4B) and by immunostaining assays (FIG. 28C). In addition, only one low intensity band was detected in skin lysate from WT corresponding to the pro-II-36 form, while the untreated and injected TghKLK5 with KLK5IN or KLK5IP showed a second band at low molecular weight, which corresponds to the cleaved II-36a active form (FIG. 28B).

[0267] Materials and Methods Used in the Efficacy Studies

[0268] Mutant Mouse Line and Treatment

[0269] The IVL-TghKLK5 mouse lines were previously established and characterized in AH lab (Furio et al., 2014). The TghKLK5 new-borns were injected intraperitoneally with 6.2 mg/kg of the KLK5 inhibitor or the inactive peptide every 2 days during 21 days. A clinical follow-up was performed every 2 days to monitor clinical NS features including TEWL, mice behaviours and body weight. Experiments were carried out according to Institution guidelines and EU legislations.

[0270] TEWL Measurement.

[0271] TEWL was Measured Using Tewameter TM300 Apparatus (Courage+Khazaka, Cologne, Germany).

[0272] Immunofluorescence Staining

[0273] Formalin-fixed paraffin-embedded skin sections (7-μm thickness) from TghKLK5 and WT mice were performed for immunofluorescence staining, hematoxylin/eosin/safranin coloration and toluidine blue coloration. Dsg3 and Dsg4 stainings were performed on skin cryosections. Antibodies used in this study are listed in Table S2. Nuclei were stained using DAPI-Fluoromount-G® (0100-20, Cliniscience). Image acquisition was performed using Leica TCS SP8 SMD confocal microscope and pictures were further processed using ImageJ software.

[0274] RNA Isolation and Quantitative Polymerase Chain Reaction

[0275] Total mRNA was extracted using RNeasy® Fibrous Tissue Mini Kit (Qiagen, Redwood, Calif., USA). Reverse transcription was carried out with 100 ng of total RNA using Super Script IV Reverse transcription kit (Life Technologies, Carlsbad, Calif., USA). cDNA was amplified using Mesa Green qPCR kits for SYBR® Assay (Eurogentec, Liège, Belgium) and Applied Biosystems 7300 Real Time PCR System (Life Technologies, Carlsbad, Calif., USA). Primers for each target gene are listed in Table S1. Relative gene expression was normalized to Hprt and calculated using the 2.sup.{circumflex over ( )}−ΔΔct method.

[0276] Immunoblotting

[0277] Tissue samples were lysed in RIPA buffer (Life Technologies, Carlsbad, Calif., USA) containing Protease Inhibitor Cocktail Tablets (Roche, Basel, Switzerland). Lysates were centrifuged at 13,000×g (4° C.) during 20 min to remove insoluble materials and total protein was measured by Bradford assay (Bio-Rad, Hercules, Calif., USA). Proteins were mixed with laemmli buffer and loaded on a 4-12% precast gel (Invitrogen, Carlsbad, Calif., USA) for SDS-PAGE separation. Proteins were transferred to nitrocellulose membranes using Trans-Blot Turbo system (Bio-Rad, Hercules, Calif., USA). The antibodies used are listed in Table S2.

[0278] In Situ Zymography

[0279] Skin cryosections (5 μm thickness) were washed with 2% Tween 20 in PBS and incubated overnight at 37° C. with 100 μg/ml BODIPY FL casein (Life Technologies, Carlsbad, Calif., USA). Caseinolytic activity was visualized and analyzed using Leica SP8 SMD confocal microscope and ImageJ software, respectively.

[0280] Statistical Analysis

[0281] Experiments were performed at least three times using at least 4 mice. Significance level was assessed using a t test, Mann-Whitney t test or Two-way ANOVA test: * p<0.05, ** p<0.005, *** p<0.0005, **** p<0.0001.