Investigation of the role of GARP by using cell specific knockout mice

Glycoprotein A repetitions predominant protein (GARP) is a transmembrane protein that captures latent transforming growth factor beta 1 (TGF-β1) on the surface of amongst others regulatory T cells (Treg), platelets and endothelial cells. Latent TGF-b  consists of the mature TGF-b and the latency associated peptide (LAP), which prevents mature TGF-b from binding to the TGF-b receptor. In order to be able to exert its function, latent TGF-b thus needs to be activated. Currently, the exact activation mechanism is not known, but evidence is provided that avb8 integrin is involved.

In vitro and in vivo studies have demonstrated that GARP expressed on Treg is involved in the immunosuppressive function of these cells. Neutralizing monoclonal antibodies specific for GARP/latent TGF-b1 complexes or downregulation of GARP expression prevents the release of active TGF-b1 in vitro and inhibits the immunosuppressive activity of human Treg in vivo in a xenogeneic graft-versus-host-disease model. This indicates that inhibiting GARP reduces the function of human Treg and thus indirectly promotes T cell immunity in vivo. It has been demonstrated that besides Treg, platelets are also expressing GARP. Interestingly, reduced growth of MC38 colon cancer cells in mice carrying a conditional deletion of the garp gene in platelets was recently demonstrated, presumably due to reduced TGF-b activity at the tumor site.

In hemostasis, involvement of platelet GARP in thrombus formation was suggested by full knockout of GARP in zebrafish. A microarray study that compared the expression profiles of amongst others precursor cells of platelets and red blood cells, postulated GARP as an interesting new platelet receptor possibly involved in hemostasis and thrombosis. And indeed, downregulation of GARP in zebrafish resulted in a decreased thrombus size and a delay in thrombocyte adhesion in a laser induced thrombosis model.

In this thesis, we aimed to further unravel the contribution of GARP on mouse Treg and platelets in immunosuppression in experimental tumor models. Additionally, we investigated the role of GARP in hemostasis and thrombosis in human and mice.

Mice with a cell-specific genetic deletion of garp in respectively Treg, platelets or endothelial cells were generated using the Cre-LoxP recombination system. Treg and platelet conditional garp knockout mice were challenged either orthotopically with GL261 glioma cells or subcutaneously with MC38 colon carcinoma cells. Fibrosarcoma tumors were chemically induced in Treg conditional garp knockout mice using methylcholanthrene. Furthermore, we investigated whether systemic inhibition of GARP using a neutralizing antibody dosed prophylactically would be sufficient to induce protective antitumor immunity and whether combination therapy with programmed cell death-1 (PD-1) blockade has a synergistic effect.

The function of platelets carrying a genetic deletion of garp was measured by flow cytometry, spreading analysis and aggregometry induced by protease activating peptide 4-activating peptide and collagen related peptide. Additionally, clot retraction and aggregation under flow were analyzed. Finally, in vivo tail bleeding time and occlusion time of the mesenteric and carotid artery after FeCl3-induced thrombosis were determined in platelet and endothelial conditional garp knockout mice. In parallel, in vitro aggregation and agglutination studies were performed with human platelets incubated with GARP neutralizing monoclonal antibodies.

We found that inhibition of GARP either through repeated injections with neutralizing monoclonal antibodies or through conditional genetic deletion of garp on Treg or platelets did not result in prolonged survival after GL261 challenge or delayed tumor growth of MC38 tumors. Treg conditional garp knockout mice were less prone to develop carcinogen induced fibrosarcoma tumors (borderline significance). Interestingly, prophylactic GARP inhibition by monoclonal antibody administration combined with blockade of immune-checkpoint PD-1 resulted in a synergistic delay in tumor growth of MC38 tumors in wild type mice. These results demonstrate GARP mediated immunosuppression in vivo in experimental tumor models and underscore the relevance of GARP as target in immune-oncology. To our view, this is most probably due to GARP expressed on Treg as anti-PD-1 treatment of Treg conditional garp knockout mice also resulted in tumor rejection.

Although GARP surface expression was increased upon platelet activation, GARP deficient mouse platelets displayed normal agonist induced activation, spreading on fibrinogen and aggregation responses. Similarly, none of the anti-human GARP antibodies induced aggregation or agglutination nor did they affect platelet aggregation induced by adenosine 5'-diphosphate or collagen. Furthermore, absence of GARP on mouse platelets did not influence clot retraction and had no impact on thrombus formation on collagen-coated surfaces under flow. In line with this, neither the tail bleeding time nor the occlusion time of the carotid- and mesenteric artery after FeCl3-induced thrombus formation in platelet or endothelial conditional garp knockout mice were affected. Evidence is thus provided that platelet and endothelial GARP are not important in hemostasis and thrombosis in man and mice.

In summary, we confirm that GARP expressed on Treg has immunosuppressive properties and this was further elaborated in murine tumor models. The next step will be to investigate whether therapeutic instead of prophylactic anti-GARP treatment also improves antitumor immunity and thus disease outcome. No major side effects are expected with regard to thrombus formation or bleeding as we extensively demonstrated that GARP is not involved in these processes. Finally, targeting the bioavailability of active TGF-b through GARP might open interesting new perspectives in other diseases besides cancer.