Structural basis of IL-23 antagonism by an Alphabody protein scaffold

Abstract

Protein scaffolds can provide a promising alternative to antibodies for various biomedical and biotechnological applications, including therapeutics. Here we describe the design and development of the Alphabody, a protein scaffold featuring a single-chain antiparallel triple-helix coiled-coil fold. We report affinity-matured Alphabodies with favourable physicochemical properties that can specifically neutralize human interleukin (IL)-23, a pivotal therapeutic target in autoimmune inflammatory diseases such as psoriasis and multiple sclerosis. The crystal structure of human IL-23 in complex with an affinity-matured Alphabody reveals how the variable interhelical groove of the scaffold uniquely targets a large epitope on the p19 subunit of IL-23 to harness fully the hydrophobic and hydrogen-bonding potential of tryptophan and tyrosine residues contributed by p19 and the Alphabody, respectively. Thus, Alphabodies are suitable for targeting protein-protein interfaces of therapeutic importance and can be tailored to interrogate desired design and binding-mode principles via efficient selection and affinity-maturation strategies.

Figure 1. Schematic overview of the Alphabody platform.

(a) Alphabodies are encoded by a single polypeptide featuring three α-helices (A–C) connected by linkers (L1, L2). The sequences of the three generic libraries (scLib_AC11, scLib_C9 and scLib_AC7) as well as the maturation library (matLib) used for biopanning are depicted. Amino-acid positions denoted with ‘x’ mark positions that were fully or partially randomized. The N terminus (N) of the Alphabody consists of the first 21 amino acids from the pelB leader sequence followed by an Asp in scLib_AC7 or Ser in all other cases. The linkers connecting the helices (L1 and L2) are identical within each library (La=T (GGSG)₄MS, Lb=T(GGSG)₄MD). The C terminus (C) differed between the generic libraries (TPGGSGGAAAHHHHHHGRAQ) and the maturation library (AAAHHHHHHQ) where in the latter the glutamine residue was encoded by an amber stop codon allowing for the translation of PIII fusion proteins. The sequence numbering corresponds to the maturation library, matLib. (b) Schematic rendering of the Alphabody scaffold as a single-chain three-helix antiparallel coiled-coil. (c) Schematic representation of the interactions between the core residues in an antiparallel coiled-coil. Hydrophobic interactions between isoleucines at positions a and d in each heptad repeat stabilize the Alphabody fold. Additional stabilizing interactions can arise from amino acids at positions e and g in helices A and C, positions g and g in helices B and C and positions e and e in helices A and B.

Figure 2. Stability and folding of the Alphabody scaffold.

(a) Stability of the scRef_L16 reference Alphabody revealed by circular dichroism thermal denaturation experiments performed in 3–6 M guanidine hydrochloride (GuHCl). (b) Melting temperatures from thermal denaturation experiments as a function of GuHCl concentration. Alphabody scRef_L16 was tested together with scRef_L8, and compared with Alphabody scShort having four different linker lengths (L6–L18). (c) Circular dichroism wavelength scan on purified scRef_L16 confirming the correct folding and alpha-helical nature of the scaffold in its purified form.

Figure 3. Binding properties of affinity-matured Alphabodies against human IL-23.

(a) Comparison of ELISA binding profiles for matured Alphabodies from each maturation campaign against non-matured Cl59 Alphabody. (b) Comparison of K_I values obtained via the splenocyte functional assay and K_D values determined by equilibrium ELISA. Cl59, MA12 and MB23 are indicated. The thick diagonal straight line corresponds to K_I=K_D, thin straight lines correspond to K_I deviating by a factor 3 from binding K_D. Dashed lines represent deviations by a factor 10. (c) Specificity of MA12 and MB23 towards human IL-23 versus human IL-12 determined by ELISA binding assays.

Figure 4. An affinity-matured Alphabody is able to prevent topical inflammation driven by IL-23 in mice.

The ear thickness of 20 mice, subdivided into four groups (n=5), was measured as a function of time. Control, control group injected with PBS on days 1 to 15 every other day; Groups A–C, mice injected with human IL-23 according to the same scheme; groups B and C, mice that received injections with MB23-PEG on days −1, 0, 3, 6, 9, 12 and 15. Group B mice received intradermal injections of 10 μg of MB23-PEG and group C received intraperitoneal injections of MB23-PEG at 40 mg kg⁻¹. Error bars represent the s.e.m.

Figure 5. Biochemical and biophysical characterization of the IL-23:MA12 interaction.

(a) Size-exclusion chromatography profile for the isolation of IL-23:MA12 complex in the presence of an excess of MA12 Alphabody. Elution volumes of protein standards are indicated at the top. The inset shows a Coomassie-stained SDS–polyacrylamide gel electrophoresis gel corresponding to the IL-23:MA12 complex elution peak. Molecular weights of protein standards are indicated. (b) ITC thermogram and analysis of the titration of IL-23 (4.9 μM in the microcalorimeter cell) with the MA12 Alphabody (53.2 μM in the titration syringe). Data were fitted to a ‘single-site binding model’, giving the apparent molar reaction enthalpy (ΔH°), entropy (ΔS°), Gibbs free energy (ΔG°), dissociation constant (K_D) and stoichiometry of binding (N) of complex formation. (c) SPR sensorgrams for the association of recombinant human IL-23 (0.5–40 nM) with immobilized MA12 Alphabody (43 RU) were fitted to a 1:1 Langmuir binding model, giving the apparent dissociation constant (K_D), association rate constant (k_on) and dissociation rate constant (k_off) of complex formation.

Figure 6. Structural analysis of the IL-23:MA12 complex.

(a) Assembly of the IL-23–MA12 complex. (Left) the IL-23:MA12 complex with IL-23 in surface mode and the MA12 Alphabody in cartoon mode. (right) The complex is rotated by 180° and IL-23 is shown in cartoon mode and MA12 in surface mode. MA12 is coloured in pink, p19 in green and p40 in grey. The MA12-interacting surface of p19 (800 Ų) is indicated in dark green and the p19-interacting surface of MA12 (840 Ų) in magenta. The linker regions between helices A and B (L1) and B and C (L2) in MA12 are represented as dashed lines. (b) Top view of the IL-23–MA12 complex. MA12 interacts with helix D and the AB and BC loops of p19. (c) View of the isoleucine core of the Alphabody core in MA12 and engagement of the p19 subunit via the interhelical groove presented by helices A and C of MA12. (d) Detailed view of the IL-23:MA12 interactions around residue W156 in the p19 subunit of IL-23. (e) Detailed view of the interactions of MA12 helix C with p19. Residue numbering in the structure of human IL-23 reported herein reflects the sequence numbering of the protein in Uniprot. Thus, residue numbers in the p19 subunit of human IL-23 differ by 19 with respect to equivalent residues in PDB entries 3DUH, 3D85, 3D87, 3QWR and 4GRW (for example, W156 in the p19 subunit of human IL-23 is equivalent to W137 in previously reported structures). (f) Comparisons of sequences corresponding to helices A and C in the two best matured Alphabodies MA12 and MB23, against the reference sequences in matLib. Positions labelled with ‘x’ corresponding to variable amino-acid position. Amino acids in MA12 that are involved in binding to human IL-23 are shown in bold.

Figure 7. Comparison of binding modes of human IL-23 with antagonists.

Comparison of structures for different IL-23:antagonist complexes: Alphabody:IL-23 complex, adnectin:IL-23 complex (pdb code 3QWR) and 7G10 antibody:IL-23 complex (PDB code 3D85). The complexes are oriented based on structural superpositions against the p19 subunit of human IL-23.