Therapeutic Potential of Shark Anti-ICOSL VNAR Domains is Exemplified in a Murine Model of Autoimmune Non-Infectious Uveitis

Abstract

Induced costimulatory ligand (ICOSL) plays an important role in the activation of T cells through its interaction with the inducible costimulator, ICOS. Suppression of full T cell activation can be achieved by blocking this interaction and has been shown to be an effective means of ameliorating disease in models of autoimmunity and inflammation. In this study, we demonstrated the ability of a novel class of anti-ICOSL antigen-binding single domains derived from sharks (VNARs) to effectively reduce inflammation in a murine model of non-infectious uveitis. In initial selections, specific VNARs that recognized human ICOSL were isolated from an immunized nurse shark phage display library and lead domains were identified following their performance in a series of antigen selectivity and in vitro bioassay screens. High potency in cell-based blocking assays suggested their potential as novel binders suitable for further therapeutic development. To test this hypothesis, surrogate anti-mouse ICOSL VNAR domains were isolated from the same phage display library and the lead VNAR clone selected via screening in binding and ICOS/ICOSL blocking experiments. The VNAR domain with the highest potency in cell-based blocking of ICOS/ICOSL interaction was fused to the Fc portion of human IgG1 and was tested in vivo in a mouse model of interphotoreceptor retinoid-binding protein-induced uveitis. The anti-mICOSL VNAR Fc, injected systemically, resulted in a marked reduction of inflammation in treated mice when compared with untreated control animals. This approach inhibited disease progression to an equivalent extent to that seen for the positive corticosteroid control, cyclosporin A, reducing both clinical and histopathological scores. These results represent the first demonstration of efficacy of a VNAR binding domain in a relevant clinical model of disease and highlight the potential of VNARs for the treatment of auto-inflammatory conditions.

Keywords: autoimmunity; biologic therapeutics; phage display; shark; single chain binding domain; uveitis; variable domain of shark new antigen receptor.

Figure 1

Selection of anti-ICOSL VNARs. (A) Screening of outputs from a selection campaign with human ICOSL. The scatter plot represents screening data of phage monoclonals from each round of selection for specific binding to huICOSL (Y axis) vs non-specific binding to human serum albumin (X axis). Each circle denotes a single clone. (B) Screening of outputs from a selection campaign with mouse ICOSL. The scatter plot represents screening data of phage monoclonals from each round of selection for specific binding to mICOSL (Y axis) vs non-specific binding to human serum albumin (X axis). (C) Cell-based binding and huICOS-huICOSL blocking assay. Monoclonal VNAR outputs from third round of selection with huICOSL were expressed in periplasm, and periplasmic fractions were tested in cell-based binding and ICOS-ICOSL blocking assays. The X axis indicates CHO-huICOSL binding with higher signals corresponding to stronger binders and the Y axis identifies clones with decreased signals that are capable of blocking the interaction of ICOSL with CHO-huICOS. The circled area captures all clones which are both strong huICOSL binders and can block ICOS/ICOSL interaction. The human ICOSL positive control is the mouse monoclonal anti-huB7-H2 antibody (MAB165, R&D).

Figure 2

Characterization of anti-huICOSL lead domains binding. (A) Lead anti-huICOSL VNAR domains were tested for binding to human and mouse ICOSL in ELISA. (B) Binding of lead clones to the IgC and/or IgV domain of the ICOS ligand.

Figure 3

Binding specificity of anti-mICOSL lead domains and blocking of ligand/receptor interaction. (A) Binding to mouse ICOSL. Titration curves of four lead anti-mICOSL domains binding to recombinant mouse ICOSL in ELISA and calculated EC50 values. (B) Blocking of ligand/receptor binding. Concentration-dependent inhibition of mICOSL-Fc binding to cell surface expressed hICOS by the addition of serial dilutions of anti-mICOSL VNAR domains (from 30 to 500 nM). 2V is the VNAR isotype control used in this experiment. (C) Binding to the IgC and IgV domains of the mouse ICOSL. (D) Binding to IgC and IgV part of the human ICOSL.

Figure 4

Characterization of anti-mICOSL lead domains after Fc reformatting. (A) SDS-PAGE of HEK293 expressed VNAR Fc. (B) Binding to mouse ICOSL. Titration curves of anti-mICOSL-Fc domains binding to recombinant mouse ICOSL. (C) Blocking of ligand/receptor binding. Concentration-dependent inhibition of recombinant mICOSL-Fc binding to cell surface expressed hICOS by the addition of serial dilutions of anti-mICOSL-Fc domains. 2V-Fc is the isotype control. (D) Species cross-reactivity of anti-mICOSL-Fc domains.

Figure 5

Clinical scores and histopathology sections from the interphotoreceptor retinoid-binding protein-induced uveitis study. (A) Bodyweights. All animals were weighed three times a week. Data are presented as mean ± SEM of percentage initial (Day 0) bodyweights. No bodyweight loss was observed. (B) Clinical scores. Retinal imaging by topical endoscopic fundal imaging (TEFI) was analyzed by one-way ANOVA followed by Dunn’s test for multiple comparisons between experimental days. A significant increase in TEFI scores was observed on Day 21 when compared to Day 7 in the untreated group, as expected for this model of uveitis. Cyclosporin A, administered from Day 1 until the end of the experiment, induced a significant reduction in the clinical scores when compared to the untreated group at Day 21. A5-Fc, administered from Day 1 until Day 14 of the experiment, delivered a comparable result to the Cyclosporin A group treated for 28 days. (C) Histopathology scores. Histopathology scores were analyzed by the Kruskal–Wallis test followed by Dunn’s post-test for multiple comparisons between experimental groups. The pathological changes observed were consistent with those reported for the model. Greater inflammation was observed in Group 1 (untreated). Cyclosporin A administered from Day 1 until Day 28 and A5-Fc administered from Day 1 until Day 14 caused an equivalent reduction in the histopathology scores. (D) Histopathology sections. Dissected eyes were embedded in paraffin wax, sectioned, and stained with hematoxylin and eosin for detailed histopathology analysis at the cellular level. (1) Healthy entire eye glob section and normal retina. (2) Untreated (=uveitis): inflammatory cells in the vitreous (upper panel), a cuff of inflammatory cells surrounding the vessel (middle panel) and neutrophilic inflammation in the drainage angle, anterior chamber, and ciliary body (lower panel). (3) Cyclosporin A treated: mild inflammation of the vitreous (upper and middle panels) and mild vasculitis (lower panel). (4) A5-Fc treated: a low number of inflammatory cells (mild inflammation) in the vitreous (above and middle panels) and mild vasculitis/cuffing (lower panel).

Figure 6

Corneal penetration of VNAR, VNAR Fc, and mAb. A mouse eye with a scratched cornea (to remove epithelium) was treated for 20 min with VNAR (two animals), VNAR-Fc (two animals), or mAb (two animals) by applying drops (4 × 3 µl) directly onto the eye. Anterior fluids were collected and analyzed in ELISA for the presence of VNAR, VNAR Fc, or mAb. VNAR, but not mAb or VNAR-Fc, was detected in anterior fluid of both mice.