Defining the epitope of a blood-brain barrier crossing single domain antibody specific for the type 1 insulin-like growth factor receptor

Abstract

Ligand-activated signaling through the type 1 insulin-like growth factor receptor (IGF1R) is implicated in many physiological processes ranging from normal human growth to cancer proliferation and metastasis. IGF1R has also emerged as a target for receptor-mediated transcytosis, a transport phenomenon that can be exploited to shuttle biotherapeutics across the blood-brain barrier (BBB). We employed differential hydrogen-deuterium exchange mass spectrometry (HDX-MS) and nuclear magnetic resonance (NMR) to characterize the interactions of the IGF1R ectodomain with a recently discovered BBB-crossing single-domain antibody (sdAb), VHH-IR5, in comparison with IGF-1 binding. HDX-MS confirmed that IGF-1 induced global conformational shifts in the L1/FnIII-1/-2 domains and α-CT helix of IGF1R. In contrast, the VHH-IR5 sdAb-mediated changes in conformational dynamics were limited to the α-CT helix and its immediate vicinity (L1 domain). High-resolution NMR spectroscopy titration data and linear peptide scanning demonstrated that VHH-IR5 has high-affinity binding interactions with a peptide sequence around the C-terminal region of the α-CT helix. Taken together, these results define a core linear epitope for VHH-IR5 within the α-CT helix, overlapping the IGF-1 binding site, and suggest a potential role for the α-CT helix in sdAb-mediated transcytosis.

Figure 1

sdAb-IGF1R binding and effects on IGF1R signalling. (a) SPR sensorgrams showing the pre-loading of the sdAb VHH-IR5 to surface-immobilized human eIGF1R followed by the binding of human IGF-1. Experiments were carried out in duplicate with SEC-purified eIGF1R, IGF-1 and VHH-IR5 (see Fig. S1 of the Supplementary Materials). (b) SPR sensorgrams showing the pre-loading of human IGF-1 to surfaced immobilized eIGF1R followed by the binding of VHH-IR5. (c) Cell-based assays showing IGF1R activation by IGF-1 (green) and and lack of activation by VHH-IR5 (black) alone. (d) Cell-based assays showing effects of VHH-IR5 on IGF1R activation in the presence of 3 nM IGF-1 (blue).

Figure 2

A global survey of HDX-MS profiles of eIGF1R in response to IGF-1 and VHH-IR5 binding. (a) IGF-1 response profile projected on a structural model of the 1:1 eIGF1R:IGF-1 complex (PDB: 6PYH). The (αβ) ′ monomer, except for α-CT′, is rendered as a transparent surface, while the (αβ) monomer and α-CT′ are shown as a ribbon cartoon. The schematic in the inset shows the domain organization of IGF1R where the (αβ) monomer is colored in green, and (αβ)′ in black. (b) VHH-IR5 response profile projected on the same structural model as in (a) (PDB: 6PYH), except that the structure is rotated counterclockwise by 90°. Significant structural destabilizations are shown in red, stabilizations in blue, lack of significant changes in grey, and missing sequence coverage in black. Residues are colored based on differences in deuteration at a single time point (± 2 SD, p = 0.02). In both (a) and (b), IGF-1 molecules are shown as magenta spheres. Of note, residues of the rhesus eIGF1R used for HDX-MS data collection were mapped by BLAST onto the mouse eIGF1R before rendering the HDX-MS results onto the 3D structure of mouse eIGF1R (PDB: 6PYH).

Figure 3

Detailed analysis of the HDX-MS profiles in the IGF1R α-CT region. (a) HDX profile of the α-CT’ bound to IGF-1 (magenta) (based on PDB: 6PYH). Residues 684-706 of mouse IGF1R α-CT’ (residues 683-705 in rhesus/human eIGF1R) are shown as a ribbon cartoon. Significant structure destabilization is shown in red and stabilization in blue, while lack of significant changes in grey and missing coverage in black. (b–e) HDX-MS kinetics of peptides covering the α-CT helix. Data collected in quadruplicate, and error bars represent 1× SD. Deuteration was normalized to theoretical maximum uptake of 45%. Free eIGF1R is shown as black circles, eIGF1R/IGF-1 as blue diamonds, and eIGF1R/VHH-IR5 as red triangles.

Figure 4

NMR signal perturbations of the ¹⁵N-labelled Ubi-ID protein by VHH-IR5. Weighted deviations are calculated as the square root of the weighted frequency shifts along both the ¹H and ¹⁵N dimensions of the HSQC spectra of the free Ubi-ID as compared to a 1:1 complex of Ubi-ID with unlabelled VHH-IR5 (Fig. S3a) and plotted according to residue-specific assignments achieved for Ubi-ID (Fig. S4). The ubiquitin moiety from 1 to 76 exhibits relatively small differences between the free protein and its complex with VHH-IR5. Some residues of the IGF1R ID 694-742 moiety show pronounced perturbations, especially for T675-E687 (or residues 84-96 of Ubi-ID). Bars with an arrow indicate those residues whose HSQC signals disappeared in the complex of ¹⁵N-labelled Ubi-ID with VHH-IR5. Hatched boxes indicate that no HSQC signals were found for this region of Ubi-ID (R689-R709), except for R704. HSQC signals of the ID segment re-emerge from residue 119 to 151 (R710-E742), with essentially no responses to VHH-IR5 binding.

Figure 5

Comparative responses of VHH-IR5 to IGF1R ID fragments. HSQC spectra of ¹⁵N-labelled VHH-IR5 identify effects of interactions with three unlabelled IGF1R ID fragments. In black are the HSQC spectra of free VHH-IR5 collected immediately before additions of Ubi-ID, Ubi-s-ID and Pep5 while superimposed in red are the HSQC spectra of the sdAb complexes at ~ 1:1 molar ratio. (a) spectral comparisons showing the effects of Ubi-ID binding. (b) spectra showing the effects of Ubi-s-ID binding. Note that Ubi-ID (a) and Ubi-s-ID (b) binding induce almost the same perturbations, i.e. disappearances of many HSQC signals of ¹⁵N-labelled VHH-IR5, especially those of residues T51 and I52 at the beginning of the CDR2 loop and many other residues in the framework region. All perturbed residues were labelled using the resonance assignments of ¹⁵N-labelled VHH-IR5 (Fig. S5). (c) Widespread spectral displacements of ¹⁵N-labelled VHH-IR5 were produced by binding of the peptide fragment Pep5.

Figure 6

VHH-IR5-binding profile of 15-residue peptide fragments covering IGF1R 674-742. The horizontal axis corresponds to the first residue in each peptide and the vertical axis is the ELISA readout (Table S2) in the PepScan assay. A high level of binding of peptide 692-FENFLHNSIVPRPE-706 (containing R704) should be noted in light of the large NMR signal perturbations of the same Arg residue.