Modulation of the PTPRS proteoglycan switch by antibodies binding to the membrane-proximal fibronectin-type III domain

Abstract

Protein tyrosine phosphatases (PTPs) receptor type II A (R2A) are negatively regulated through oligomerization upon binding of their extracellular domains to glycosaminoglycans (GAGs) on heparan sulfate proteoglycans (HSPGs). Inactivation of receptor PTP sigma (PTPRS) by HSPGs promotes the aggressive behavior of fibroblast-like synoviocytes (FLS) in rheumatoid arthritis (RA). Blocking the binding of its N-terminal, membrane-distal immunoglobulin-like 1 and 2 (Ig1&2) domains to its GAG ligands on the HSPG syndecan-4 (SDC4) promotes PTPRS activity and reverses the pathogenic phenotype of FLS. The potential for therapeutically leveraging other PTPRS ectodomain regions is, however, unknown. We show targeting the membrane-proximal fibronectin type III-like 9 (Fn9) domain offers a novel avenue to activate PTPRS. We mapped PTPRS Fn9 as the binding site of three antibodies (Abs) (13G5, 22H8, 49F2) and characterized their effects on cells. Despite sharing similar epitopes, we found large differences in the ability of these Abs to regulate PTPRS activity. One of these, 13G5, reduced PTPRS-dependent cell migration, PTPRS co-localization with SDC4, and PTPRS oligomerization. Single-chain variable fragment Abs of 13G5 and 22H8 were similarly effective at activating cellular PTPRS as 13G5. Replacing the entire 13G5 constant region enhanced its binding and cellular activity, indicating the Ab's potency can be optimized via isotype engineering. Treatment of cells with recombinant Fn9 protein acted as a decoy, disrupting PTPRS colocalization with SDC4 and oligomerization, and inhibiting FLS migration. Finally, significant disease mitigation in mice using 13G5-derived Abs suggests a viable strategy for the generation of novel drugs for RA therapy.

Keywords: PTPRS; antibody; cell migration; proteoglycan; receptor regulation; rheumatoid arthritis; synoviocyte; tyrosine phosphatase.

Figure 1

Anti-PTPRS monoclonal antibody (Ab) 13G5 decreases FLS migration by blocking PTPRS clustering and association with SDC4.A, schematic representation of PTPRS and SDC4 interaction with Ig1&2-Fc decoy protein. B, 13G5 and 22H8 association/dissociation curves at 500 nM, 400 nM, 300 nM, 200 nM, 100 nM, 50 nM, 30 nM, and 10 nM concentration and 49F2 association/dissociation curves at 100 nM, 80 nM, 60 nM, 40 nM, 20 nM, 10 nM, 5 nM, and 1 nM concentration measured by biolayer interferometry. K_D, K_on and K_off values against PTPRS ectodomain were calculated (n = 3). C, relative transwell migration of mouse FLS in the presence of 13G5, 22H8, and 49F2 (200 nM). PTPRS Ig1&2-hFc fusion (Ig1&2) is used for comparison, human Fc and mouse IgG are negative controls. Each dot represents a biological replicate (n = 4). D, relative transwell migration of RA FLS in the presence of 13G5 at the indicated concentrations. Mouse IgG is used as a negative control. Each dot represents a biological replicate (n = 3). E, relative transwell migration of Ptprs and Sdc4 KO mouse FLS in the presence of 13G5 (200 nM). Each dot represents a biological replicate (n = 3). F, relative transwell migration of mouse FLS in the presence of 13G5, Ig1&2, or both 13G5 and Ig1&2 (13G5: 200 nM; Ig1&2: 20 nM). Each dot represents a biological replicate (n = 5). G, relative hPTPRS/mPTPRS co-localization by proximity ligation assay (PLA) in the presence of 13G5, 22H8, or 49F2 (200 nM) (left) and representative assay images (right). Scale bar 10 μm. Ig1&2 is used for comparison. Each dot represents a biological replicate (n = 3). H, relative hPTPRS/mSDC4 co-localization by PLA in the presence of 13G5, 22H8, or 49F2 (200 nM) (left) and representative assay images (right). Scale bar 10 μm. Each dot represents a biological replicate (n = 3). I, relative co-localization between hPTPRS WT or QuadK mutant and mSDC4 by PLA in the presence of 13G5 (200 nM). Each dot represents a biological replicate (IgG: n = 4; 13G5: n = 3). Mean ± SEM are shown. ∗∗∗p ≤ 0.001, ∗∗p ≤ 0.01, and ∗p ≤ 0.05 by ordinary one-way ANOVA (C–F, H, and I) and Kruskal-Wallis (G).

Figure 2

Anti-PTPRS Abs share an epitope in the PTPRS juxtamembrane region.A, hydrogen-deuterium exchange mass spectrometry (HDX-MS) of mouse PTPRS ectodomain in the presence or absence of 13G5, 22H8, and 49F2. Ribbon diagram of the difference between PTPRS alone and PTPRS+13G5 (top, blue indicates peptides that exchange faster in the absence of the Ab) and graphs showing the number of hydrogen atoms exchanged at 10, 100, 1000, and 10,000 s for selected peptides (bottom). Numbers refer to the sequence of the species used in the experiment, starting at the mature polypeptide. B, immunoprecipitation of full-length (FL) or N-terminally truncated PTPRS with 13G5, 22H8, and 49F2 blotted for HA tag. Gel image (right) and quantification normalized to PTPRS^Δ326 expression (left) are shown. Each dot represents a biological replicate (n = 3). C, schematic view of the Fn9 mutation within hPTPRS (top). Representative flow cytometry histogram showing 13G5 binding to WT PTPRS and PTPRS Fn9mut (bottom). Overton percentage represents the ratio of antibody binding to overexpressed PTPRS. D, relative WT or Fn9mut hPTPRS/mSDC4 co-localization in the presence of 13G5 (200 nM). Each dot represents an experimental replicate (WT IgG: n = 4; others: n = 3). Mean ± SEM are shown. ∗∗∗∗p ≤ 0.0001, ∗∗∗p ≤ 0.001, ∗∗p ≤ 0.01, and ∗p ≤ 0.05 by ordinary one-way ANOVA (B) and Kruskal-Wallis (D).

Figure 3

Ab isotype affects its potency in vitro.A, schematic representation of 13G5, 22H8, 13G5^IgG1 and 22H8^IgG2b Abs. B and C, association/dissociation kinetics of 13G5^IgG1 (B) and 22H8^IgG2b (C) at 500 nM, 400 nM, 300 nM, 200 nM, 100 nM, 50 nM, 30 nM, and 10 nM concentrations measured by bio-layer interferometry (n = 3). D, relative transwell migration of mouse FLS in the presence of 13G5 or 13G5^IgG1 at the indicated concentrations. Each dot represents a biological replicate (n = 3). E, relative transwell migration of mouse FLS in the presence of 13G5, 13G5^IgG1, 22H8 or 22H8^IgG2b (200 nM). Each dot represents a biological replicate (13G5 & 22H8: n = 3; others: n = 4). F, relative hPTPRS/mSDC4 co-localization by proximity ligation assay (PLA) in the presence of 13G5 (200 nM) or 13G5^IgG1 (100 nM). Each dot represents a biological replicate (n = 3). G, relative hPTPRS/mPTPRS co-localization by PLA in the presence of 13G5 (200 nM) or 13G5^IgG1 (100 nM). Each dot represents a biological replicate (n = 3). Mean ± SEM are shown. ∗∗∗p ≤ 0.001, ∗∗p ≤ 0.01, and ∗p ≤ 0.05 by ordinary one-way ANOVA (D–G).

Figure 4

scFv-Fc fusions reduce FLS migration, PTPRS/SDC4 proximity, and PTPRS clustering.A, schematic representation of scFv-Fc fusions. B, ELISA showing binding of the scFv-Fcs at different concentrations to the mPTPRS ectodomain. C, binding kinetics of scFv-Fc fusion proteins 13G5^HL-scFv-Fc (denoted 13G5-HL in the figure) (left) and 22H8^HL-scFv-Fc (22H8-HL) (right) at 50 nM, 40 nM, 30 nM, 20 nM, 10 nM, 5 nM, 2.5 nM, and 1 nM concentrations to immobilized hPTPRS Fn9His by BLI. D, binding kinetics of 13G5 and 22H8 at 500 nM, 400 nM, 300 nM, 200 nM, 100 nM, 50 nM, 30 nM, and 10 nM concentration to immobilized hPTPRS Fn9His by BLI. E, relative transwell migration of mouse FLS in the presence of scFv-Fc fusion proteins 13G5^HL-scFv-Fc (13G5-Fc HL) and 22H8^HL-scFv-Fc (22H8-Fc HL) or Fc control protein (200 nM) (n = 4). F and G, relative hPTPRS/mSDC4 co-localization (F) or hPTPRS/mPTPRS clustering (G) by proximity ligation assay (PLA) in the presence of scFv-Fc fusion proteins 13G5-Fc HL and 22H8-Fc HL or Fc control protein (200 nM) (F: n = 6; G: n = 3). Each dot represents a biological replicate. Mean ± SEM are shown. ∗∗∗p ≤ 0.001 and ∗p ≤ 0.05 by ordinary one-way ANOVA. † indicates that calculated parameters might be inaccurate due to extremely slow dissociation.

Figure 5

hPTPRS Fn9-Fc fusions reduce FLS migration, PTPRS/SDC4 proximity and PTPRS clustering.A, schematic representation (left) and sequence (right) of the human PTPTRS Fn9-Fc fusion protein. B and C, relative transwell migration of WT (B) or Ptprs KO (C) mouse FLS in the presence of Fn9-Fc (1 μM). Each dot represents a biological replicate (n = 3). D and E, relative hPTPRS/mSDC4 co-localization (D) and hPTPRS/mPTPRS clustering (E) by proximity ligation assay (PLA) in the presence of Fn9-Fc (1 μM). Each dot represents a biological replicate (n = 3). Mean ± SEM are shown. ∗∗∗p ≤ 0.001, ∗∗p ≤ 0.01, and ∗p ≤ 0.05 by ordinary one-way ANOVA.

Figure 6

13G5^IgG1 decreases arthritis severity in mice.A, scheme representing the K/BxN STIA model and the dosing schedule. B, clinical score of STIA mice treated with 1.0 mg of mIgG1 control or 13G5^IgG1; (n = 5). C, representative images of safranin-O staining. Scale bar 100 μm. D, histopathological evaluation of synovial inflammation, proteoglycan loss and bone erosion of hind ankles from mice in (B). Bone erosion, proteoglycan loss, and synovial inflammation were assessed separately, while histological score represents the sum of all parameters (n = 5). Mean ± SEM are shown. ∗p ≤ 0.05 by two-way ANOVA (B) and Mann–Whitney (D).