Cepharanthine Blocks Presentation of Thyroid and Islet Peptides in a Novel Humanized Autoimmune Diabetes and Thyroiditis Mouse Model

Abstract

Autoimmune polyglandular syndrome type 3 variant (APS3v) refers to an autoimmune condition in which both type 1 diabetes (T1D) and autoimmune thyroiditis (AITD) develop in the same individual. HLA-DR3 confers the strongest susceptibility to APS3v. Previously we reported a unique amino acid signature pocket that predisposes to APS3v. We found that this pocket is flexible and can trigger APS3v by presenting both thyroid (Tg.1571, TPO.758) and islet (GAD.492) peptides to induce autoimmune response. We hypothesized that blocking the specific APS3v-HLA-DR3 pocket from presenting thyroid/islet antigens can block the autoimmune response in APS3v. To test this hypothesis we performed a virtual screen of small molecules blocking APS3v-HLA-DR3, and identified 11 small molecules hits that were predicted to block APS3v-HLA-DR3. Using the baculovirus-produced recombinant APS3v-HLA-DR3 protein we tested the 11 small molecules in an in vitro binding assay. We validated 4 small molecule hits, S9, S5, S53 and S15, that could block the APS3v-HLA-DR3 pocket in vitro. We then developed a novel humanized APS3v mouse model induced by co-immunizing a peptide mix of Tg.1571, TPO.758 and GAD.492. The immunized mice developed strong T-cell and antibody responses to the thyroid/islet peptides, as well as mouse thyroglobulin. In addition, the mice showed significantly lower free T4 levels compared to controls. Using the APS3v mouse model, we showed that one of the 4 small molecules, Cepharanthine (S53), blocked T-cell activation by thyroid/islet peptides ex vivo and in vivo. These findings suggested Cepharanthine may have a therapeutic potential in APS3v patients carrying the specific APS3v-HLA-DR3 pocket.

Keywords: autoimmune polyglandular syndrome; autoimmune thyroiditis; cepharanthine; immune therapy; type 1 diabetes.

Figure 1

A flow chart showing the approach we took to identify Cepharanthine (S53) as a compound that can block HLA-DR3 in APS3v. First we performed a virtual screen which identified 11 compounds. This was followed by ELISA testing which we validated 4 top hit compounds out of the 11 predicted by the virtual screen (S9, S5, S53 and S15). We then generated a humanized mouse model of APS3v and tested whether S53 can block T-cell activation in the APS3v mouse model both ex vivo and in vivo. S53 was selected for further testing because it is the only compound out of the 4 validated small molecules that has been used in humans before.

Figure 2

A representative depiction of the binding of Cepharanthine to the binding pocket of the APS3v-associated HLA-DR3. The surface is colored by the electrostatic potential, where blue indicates a positive and red a negative electrostatic potential. Note the anchoring of the methylenedioxy group next to the Arg-β74 (blue) and the positively charged methyl-piperidine group next to Glu-α36.

Figure 3

11 top scoring small molecules were tested for blocking APO peptide binding to APS3v-HLA-DR3 in ELISA. Figure 3 shows the in vitro screening results of the 11 top scoring small molecules identified in the virtual screen. 4 small molecules (S9, S15, S53 and S5) inhibited APO peptide binding to APS3v-HLA-DR3 by more than 50%.

Figure 4

The top 4 hit small molecules (S9, S5, S15, S53) were tested for blocking biotinylated TPO.758, GAD.492 and Tg.1571 binding to APS3v-HLA-DR3 in ELISA. Figure 4 shows the ELISA results of blocking TPO.758, GAD.492 and Tg.1571 binding APS3v-HLA-DR3 by S9 (A), S5 (B), S15 (C) and S53 (D).

Figure 5

S53 at decreasing doses (0.4 mM, 0.2 mM, 0.1 mM, 0.05 mM) was tested for blocking the binding of biotinylated APO, TPO.758, GAD.492 and Tg.1571 to APS3v-HLA-DR3 by ELISA. The ELISA results showed that of S53 blocked APO (A), TPO.758 (B), GAD.492 (C) and Tg.1571 (D) in a dose-dependent manner.

Figure 6

Twenty-two humanized NOD-DR3 mice were co-immunized with Tg.1571, GAD.492 and TPO.758. Supernatant from stimulated splenocytes were collected and assayed for cytokine production. Sera were also collected for autoantibody measurements. Figure 6 shows interferon gamma production (A) and autoantibody response to Tg.1571 (B), GAD.492 (C) and TPO.758 (D) of humanized NOD-DR3 mice co-immunized with the three thyroid and islet peptides. NC, negative control peptide. Anti-CD3/CD28 beads, positive control. Control mice were immunized with PBS and adjuvant. *p < 0.05; **p < 0.01; ***p < 0.001.

Figure 7

Twenty-two humanized NOD-DR3 mice were co-immunized with Tg.1571, GAD.492 and TPO.758. Figure 7 shows the humoral response and thyroid functions in humanized NOD-DR3 mice co-immunized with Tg.1571, GAD.492 and TPO.758. Autoantibody response to mouse Tg (A) and free T4 levels (B) of humanized NOD-DR3 mice co-immunized with the three thyroid and islet peptides, compared to control mice immunized with PBS and adjuvant. *p < 0.05; ***p < 0.001.

Figure 8

Eight humanized NOD-DR3 mice were co-immunized with Tg.1571, GAD.492 and TPO.758. Splenocytes were co-incubated with peptides and small molecules ex vivo. Stimulated splenocytes were collected and assayed for cytokine production. Figure 8 shows interferon gamma and IL-2 production when GAD.492 (A), TPO.758 (B) and Tg.1571 (C) were co-incubated with the top four small molecules (S9, S5, S53 and S15) in splenocytes of the immunized mice. *p < 0.05; **p < 0.01; ***p < 0.001.

Figure 9

Four humanized NOD-DR3 mice were co-immunized with Tg.1571, GAD.492 and TPO.758 and treated with Cepharanthine (S53) or vehicle. Stimulated splenocytes were collected and T-cell proliferation assay was performed. Figure 9 shows the stimulation index in response to GAD.492 (A), TPO.758 (B) and Tg.1571 (C). *p < 0.05; **p < 0.01.