Identification of Hit Compounds Using Artificial Intelligence for the Management of Allergic Diseases

Abstract

This study aimed to identify and evaluate drug candidates targeting the kinase inhibitory region of suppressor of cytokine signaling (SOCS) 3 for the treatment of allergic rhinitis (AR). Utilizing an artificial intelligence (AI)-based new drug development platform, virtual screening was conducted to identify compounds inhibiting the SH2 domain binding of SOCS3. Luminescence assays assessed the ability of these compounds to restore JAK-2 activity diminished by SOCS3. Jurkat T and BEAS-2B cells were utilized to investigate changes in SOCS3 and STAT3 expression, along with STAT3 phosphorylation in response to the identified compounds. In an OVA-induced allergic rhinitis mouse model, we measured serum levels of total IgE and OVA-specific IgE, performed real-time PCR on nasal mucosa samples to quantify Th2 cytokines and IFN-γ expression, and conducted immunohistochemistry to analyze eosinophil levels. Screening identified 20 hit compounds with robust binding affinities. As the concentration of SOCS3 increased, a corresponding decrease in JAK2 activity was observed. Compounds 5 and 8 exhibited significant efficacy in restoring JAK2 activity without toxicity. Treatment with these compounds resulted in reduced SOCS3 expression and the reinstatement of STAT3 phosphorylation in Jurkat T and BEAS-2B cells. In the OVA-induced allergic rhinitis mouse model, compounds 5 and 8 effectively alleviated nasal symptoms and demonstrated lower levels of immune markers compared to the allergy group. This study underscores the promising nonclinical efficacy of compounds identified through the AI-based drug development platform. These findings introduce innovative strategies for the treatment of AR and highlight the potential therapeutic value of targeting SOCS3 in managing AR.

Keywords: SOCS3; allergic rhinitis; artificial intelligence; new drug development platform.

Figure 1

Virtual screening by AI using a virtual ligand database for the target combination site and 20 hit compounds with outstanding binding ability. (A) DB converted into a 3D virtual ligand library was used to conduct virtual screening of >8 million stereoisomers and to calculate mm/GBSA ΔG binding values, afterwards, the compounds that inhibit the SH2 domain binding of SOCS3 were screened. (B) After molecular dynamics calculations, compounds with little fluctuation in the binding mode were ordered. After screening their drug inhibitory effects on the targeted sites based on AI, 20 hit compounds with excellent binding ability were identified.

Figure 2

The hit compounds exhibited non-toxicity in both BEAS-2B and Jurkat cells, while restoring the suppressed JAK2 activity induced by SOCS3. (A) At concentrations up to 100 µM, compound 5 demonstrated no significant decrease in cell viability in BEAS-2B, but at 800 µM, the viability dropped below 50%. (B) At concentrations up to 100 µM, compound 8 demonstrated no significant decrease in cell viability in BEAS-2B, but at 800 µM, the viability dropped below 50%. (C) At concentrations up to 100 µM, compound 5 demonstrated no significant decrease in cell viability in Jurkat cell, but at 800 µM, the viability dropped below 10%. (D) At concentrations up to 100 µM, compound 8 demonstrated no significant decrease in cell viability in Jurkat cell, but at 800 µM, the viability dropped below 10%. (E) When the concentration of SOCS3 reached 5.2 µM, the residual activity of JAK2 was 10%. (F) When 100 µM of hit compounds 5 or 8 was added, the activity of JAK2, which was reduced from SOCS3 to approximately 25%, recovered to >50%.

Figure 3

In Jurkat T and BEAS-2B cells, SOCS3 expression was reduced after treatment with hit compounds, and STAT3 phosphorylation was recovered. (A) In the Jurkat T cell line, hit compound 8 could inhibit SOCS3, from 0.5 h to 1 h, thus recovering phosphorylation of STAT3. (B) In the BEAS-2B cell line, hit compounds 5 and 8 could inhibit SOCS3 at 1 h, thus recovering phosphorylation of STAT3. Results presented in the graph are from three independent experiments.

Figure 4

Effects of novel drug treatment on mouse symptom scores and immunoglobulin E levels in comparison to allergy group. (A) Nasal rubbing frequency of mice was reduced in hit compounds 5 and 8 treatment groups. (B) Nasal sneezing frequency of mice was reduced in hit compounds 5 and 8 treatment groups. (C) Expression of total IgE was reduced in hit compounds 5 and 8 treatment groups. (D) OVA-specific IgE expression was reduced in compound 8 group. ***: p < 0.001.

Figure 5

Reduction in eosinophils in mouse nasal mucosa and lungs in drug treatment groups compared to allergy group. (A) Expression of IL-13 was reduced in treatment groups. (B) Although the expression of IL-5 was also reduced in treatment groups, the difference was not statistically significant. (C) Although the expression of IL-4 was reduced in treatment groups, the difference was not statistically significant. (D) Eosinophils in mouse nasal mucosa decreased in treated groups. (E) Eosinophil levels in mouse lungs decreased in treated groups. ***: p < 0.001.