De Novo Design of Integrin α5β1 Modulating Proteins to Enhance Biomaterial Properties

Abstract

Integrin α5β1 is crucial for cell attachment and migration in development and tissue regeneration, and α5β1 binding proteins can have considerable utility in regenerative medicine and next-generation therapeutics. We use computational protein design to create de novo α5β1-specific modulating miniprotein binders, called NeoNectins, that bind to and stabilize the open state of α5β1. When immobilized onto titanium surfaces and throughout 3D hydrogels, the NeoNectins outperform native fibronectin (FN) and RGD peptides in enhancing cell attachment and spreading, and NeoNectin-grafted titanium implants outperformed FN- and RGD-grafted implants in animal models in promoting tissue integration and bone growth. NeoNectins should be broadly applicable for tissue engineering and biomedicine.

Keywords: RGD; biomaterial; de novo protein design; hydrogel; integrin α5β1; regenerative medicine; titanium.

Figure 1

α5β1 binder design strategy. A) Schematic representation of the structure of the inactive, apo‐integrin α5β1, Fibronectin‐bound, active integrin α5β1(top panel), and NeoNectin‐bound (blue circle) integrin α5β1 (bottom panel). Integrin α5 subunit is shown as slate blue and β1 subunit is shown as orange. B) Schematic of designed NeoNectin in biomaterial applications. C,D) Specificity design challenge highlighted by the similar electrostatic potential of integrins α5β1, αvβ3 (structures are from complexes with their cognate ligand peptides; PDB:4WK2 and 1L5G, respectively). Main differences are highlighted with arrows. Glycan molecules are shown as yellow sticks. Zoomed‐in views of the RGD binding interfaces of α5β1 and αvβ3 are shown below. E) Design strategy for α5β1 specific NeoNectin. F) Computational model of a designed α5β1 binder colored by site saturation mutagenesis results. The NeoNectin parent design 1 was colored by positional Shannon entropy, using a gradient from blue to red, with blue indicating positions of low entropy (conserved) and red those of high entropy (not conserved). G) Site Saturation Mutagenesis analysis of NeoNectin parent designs was sorted by FACS in the presence of α5β1 at different concentrations and the affinity of each variant was calculated. The affinity of each variant of NN parent design 1 (NN P1) and parent design 2 (NN P2) were highlighted as green and orange circles, respectively. The upper left corner are variants specific to α5β1.

Figure 2

NeoNectin binds α5β1 with high affinity and specificity. A–D) BLI binding affinity traces for NeoNectin candidates 1 (NN‐C1) or 2 (NN‐C2) against the α5β1 ectodomain in integrin resting buffer (20 mm Tris, pH 7.4, 1 mm Ca²⁺, 1 mm Mg²⁺) or otherwise noted. Global kinetic fits, assuming a 1:1 binding model, are shown as black lines. E) Flow cytometry measurements of biotinylated NeoNectin candidates on K562 cells gave K_D​ values of 1.9 nm for NN‐C1 and 124 nm for NN‐C2. F) Competing the binding of CF647 labeled NN‐C1 by NN‐C1, FN, and RGD peptide on K562 cells gave K_D values of 0.9 ± 0.2, 612 ± 105, and 150 000 ± 20,000 nm respectively (see Methods for calculating K_D values). G–K) BLI binding affinity traces for NN‐C1 against αvβ1, αvβ3, αvβ6, αvβ8, and α8β1 ectodomain in integrin resting buffer. L) BLI binding affinity traces for NN‐C2 against integrin αvβ3 in the resting buffer.

Figure 3

Structural characterization of integrin α5β1:NeoNectin complexes. A–C) Representative 2D negative stain class averages of α5β1 alone, in presence of NN‐C1 or NN‐C2 in activating (1 mm Mn²⁺) and non‐activating (5 mm Ca²⁺) buffer conditions. Integrins are categorized into three canonical conformations: extended open (EO), extended closed (EC), and bent closed (BC). The number of classes shown is representative of the number of total particles in that conformation. D) Cryo‐EM map of α5β1 bound to NeoNectin. The sharpened, locally refined map is shown in color, superimposed with the unsharpened map in semi‐transparent white. The color code is as follows: α5 (lavender), β1 (light orange), Neonectin (turquoise), coordinated cations (plum), and glycans (yellow). E) Two views of the ribbon model of α5β1 bound to NeoNectin displayed within the unsharpened density shown in A). F) An overlay of the NN‐C1 designed model (gray) and the experimentally determined model (turquoise). G) Close‐up of NN‐C1 Loop1 (L1, ⁶HRGDFP)^{[ 12 ]} and α5β1. R8 and D10 directly interact with α5β1 and other residues contribute to stabilizing the loop. H) Close‐up of NN‐C1 Loop3 (L3, ³³DHK)^{[ 35 ]} and α5β1 interface. I) Close‐up of NN‐C1 Loop5 (L5, ⁵⁷RGLW)^{[ 60 ]} and α5β1 interface.

Figure 4

Soluble NeoNectin inhibits α5β1‐mediated cell spreading and migration. A) Schematic of the experimental design monitoring MCF10A cell attachment in presence of soluble NeoNectin on collagen I or FN‐coated surface. B) Confocal imaging of MCF10A cells plated on collagen I or FN coated surface in presence of soluble NeoNectin for 30 min. The scale bar is 20 µm. C) Quantification of cell area in B). D) RNA‐seq results from MCF10A cells plated on fibronectin surface in presence or absence of 500 nm NeoNectin for four h, with genes related to the focal adhesion pathway highlighted in blue, genes significantly affected highlighted in red. (E) Heat map representation of genes (as Z‐score of logCPM) involved in focal adhesion pathway. ‐NN: Cells spread on FN‐coated surface. +NN: Cells spread on FN‐coated surface in the presence of 500 nm NeoNectin. F) Schematic of the experimental design monitoring MCF10A cell migration with/without soluble NeoNectin. G) Trajectories of individual cells tracked over an 18‐hour imaging period in presence of 0 or 500 nm NeoNectin. H) Quantification of cell velocity in µm/min of individual cells from G) and Figure S6C (Supporting Information). I) Quantification of accumulated traveled distance of individual cells from G) and Figure S6C (Supporting Information). J) Schematic of the tube formation assay. K) Representative decrease in vascular stability by 10 nm soluble NeoNectin. Soluble NeoNectin was added to HUVEC cells at 0, 1, 10, and 100 nm. Vascular stability was analyzed after 12 h. L) The number of nodes, meshes, and tubes was quantified using an angiogenesis analyzer plug‐in in ImageJ. The scale bar is 100 µm. Statistical significance was analyzed using One‐way Anova Bonferroni's multiple comparison test. All experiments have at least three biological replicates.

Figure 5

Immobilized NeoNectin stimulates FN‐like cells spreading in 3D and 2D cultures. A) Representative immunofluorescence images of MSCs after 5 days of 3D culture within the different functionalized hydrogels. Adhesion modifications were included at 1 mm concentration. The scale bar denotes 50 µm. RGD: CRGDS. Nuclei are stained in blue, and Actin cytoskeleton in red. B) Quantification of cell spread area in A). One Way ANOVA, Tukey's Post‐hoc Test. **p < 0.01, ***p < 0.001, ****p < 0.0001. C) Representative immunofluorescence images of MSCs (top) and FFs (bottom) after 4 h of adhesion on the different functionalized surfaces. FN and NeoNectin were covalently linked through free amines. The RGD peptide (Cys‐(Ahx)3‐GRGDS) was covalently attached through the Cys. The scale bar denotes 100 µm. Nuclei are stained in blue, and Actin cytoskeleton in red. D) Representative immunofluorescence images of MSCs to detect phosphorylation of FAK. Nuclei are stained in blue, Actin cytoskeleton is stained in red, and pFAK (Tyr397) is in green. The scale bar denotes 20 µm. E) Representative immunofluorescence images of MSCs to detect phosphorylation of FAK after treatment with blebbistatin. Nuclei are stained in blue, Actin cytoskeleton is stained in red, and pFAK (Tyr397) in green. The scale bar denotes 20 µm. F,G) Scatterplots comparing relative gene expression (fold change of FN‐, NN‐, or RGD‐grafted Ti over bare Ti) for MSCs F) and FFs G). FN‐grafted samples are on the y‐axis; NN‐ (left) or RGD‐treated (right) samples on the x‐axis. Similarities of whole transcriptome (black) and TGF‐β pathway genes (red) with FN‐grafted samples were assessed by Pearson correlation. H,I) Heat map representation of top differentially expressed genes compared to cells spread on bare titanium surface (Log of Fold Change (LogFC) > 1.5 or < −1.5 in MSCs). FN: Cells spread on FN‐grafted titanium surface. NN: Cells spread on NeoNectin‐grafted titanium surface. RGD: Cells spread on RGD peptide‐grafted titanium surface. ECM: Extracellular Matrix. TF: Transcription Factors.

Figure 6

NeoNectin‐grafted titanium implants outperform FN‐grafted, RGD‐grafted, and bare titanium (Ti) implants in stimulating implant integration and bone growth. A) Schematic of the in vivo experimental procedure with rabbits. Implants were randomly inserted into the tibia of rabbits, and samples were collected for histomorphometric analyses 3 and 6 weeks after the surgical intervention. N = 7 for the 3‐week and 6‐week groups. B) Representative micro‐CT 3D reconstruction images showing bone (yellow) around the grafted or bare titanium implants (gray) 3 weeks post‐surgery. C) Calculated percentage of bone volume versus total volume (BV/TV) from micro‐CT images collected from animals at 3 weeks B) and 6 weeks groups (Figure S11A, Supporting Information) post‐surgery. Non‐parametric Mann Whitney's test (*p < 0.05). Data are presented as mean ± standard error of the mean. Direct comparisons between FN‐ and NeoNectin‐grafted titanium implants were highlighted in red for both the post‐3‐week B) and 6‐week groups (Figure S11A, Supporting Information). D–G) Representative histological staining (left) and SEM (right) images of longitudinal sections 3 weeks post‐implantation showing the implants conjugated with indicated molecules inserted into the tibia of rabbits. Bones are stained in green, and muscle in red. The scale bar denotes 200 µm. H) Calculated bone‐implant contact (BIC) percentage from SEM images D–G), and Figures S11C–F (Supporting Information). Non‐parametric Mann Whitney's test (*p < 0.05). Data are presented as mean ± standard error of the mean. Direct comparisons between FN‐ and NeoNectin‐grafted titanium implants were highlighted in red for both the post‐3‐week D–G) and 6‐week groups (Figures S11C‐F, Supporting Information). I–L) Zoomed‐in view of the boxed area in D–G). The scale bar denotes 50 µm. M) Calculated percentage of new bone from the SEM images D–G), and Figures S11C–F, Supporting Information). Non‐parametric Mann Whitney's test (*p < 0.05). Data are presented as mean ± standard error of the mean. Direct comparisons between FN‐ and NeoNectin‐grafted titanium implants were highlighted in red for both the post‐3‐week and 6‐week groups.