STING activation by teniposide: a potential direct mechanism beyond cGAS stimulation

Abstract

Introduction: The STimulator of Interferon Genes (STING) is a key adaptor protein in the innate immune response to cytosolic DNA, making it a promising therapeutic target. Identifying novel STING ligands could provide new opportunities for immune modulation.

Methods: We employed high-throughput virtual screening to identify potential STING ligands and selected Teniposide, an anticancer drug primarily used for infant leukemia. Direct binding of Teniposide to STING's cytosolic domain was confirmed via isothermal titration calorimetry (ITC) and validated using a double mutant STING variant unable to bind Teniposide. Computational docking and molecular dynamics simulations were performed to characterize the binding mode.

Results: Teniposide activated the IFN-β signaling pathway in a STING-dependent manner, independent of dsDNA sensors cyclic GMP-AMP synthase (cGAS) and Interferon Gamma Inducible Protein 16 (IFI16). ITC confirmed direct interaction, and the STING double mutant abolished binding. Computational analyses revealed a symmetrical binding mode involving two Teniposide molecules interacting with STING.

Discussion: These findings suggest that Teniposide activates STING through a previously unrecognized, cGAS-independent mechanism, while retaining potential for canonical cGAS-STING stimulation. Our combined computational and experimental evidence supports repurposing Teniposide as a STING agonist, highlighting new therapeutic possibilities for innate immune stimulation.

Keywords: Cyclic dinucleotides (CDNs); High-throughput virtual screening (HTVS); IFN-β; Immunotherapy drug repurposing; Isothermal titration calorimetry (ITC); Molecular docking; STING; Teniposide.

Figure 1

Teniposide is a potential STING agonist that induces IFNB1 gene expression. (a) Schematic representation of high-throughput virtual screening HTVS of NIH libraries targeting the STING binding pocket. (b) Representation of the 2D structures of Teniposide and cGAMP. (c) Relative IFNB1 gene expression in THP1 cells treated with varying doses of Teniposide or cGAMP. (d) Time-course IFNB1 gene expression in THP1 cells treatment with cGAMP or Teniposide. (e)MX1 and IL6 gene expression in THP1 cells 12h post-Teniposide treatment. (f) Dose response of Teniposide in BMDMs, showing relative IFNB1 expression 24h post-treatment. (g)IFNB1 time-course expression in BMDM upon stimulation with Teniposide (1µM or 3µM). cGAMP induction at 8h was used as a reference control. (h) Relative MX1 and IL6 gene expression in BMDM 48h after Teniposide treatment. Statistical significance is indicated by *p ≤ 0.05, **p ≤ 0.01, ***p ≤0.005, **** p ≤ 0.001.

Figure 2

Teniposide-induced IFNB1 expression requires STING and may occur independently of cGAS and IFI16. (a)IFNB1 relative gene induction measured by RT-qPCR in WT or STING KO THP1 cells after Teniposide 3µM treatment for 8h. (b)IFNB1 relative expression in WT or GT STING BMDM treated with 3µM Teniposide for 8h. (c) Western blot representing pTBK1, total TBK1, pIRF3, total IRF3 and bacting in THP1cells mock treated, treated for 4h with 3µM of Teniposide or 1µM of diABZI. (d)IFNB1 relative gene expression in WT THP1 cells stimulated for 18h with 10 IU of IFN-β/ml or mock-treated followed by a treatment with DMSO, cGAMP or Teniposide (3µM). (e) Same as in (d) but in WT and cGAS KO THP1 cells. (f) Same as in (d) but in WT and IFI16 THP1 KO cells. (g) Western blot verifying STING, cGAS or IFI16 in WT but not STING, cGAS, and IFI16 KO THP1 cells. Tubulin or β-Actin is shown as a loading control. Statistical significance is indicated by *p ≤ 0.05, **p ≤ 0.01, ***p ≤0.005, **** p ≤ 0.001.

Figure 3

Isothermal titration calorimetry analysis of ligand binding to recombinant STING-LBD. (a) The top panels display heat flow rates over time for the binding of Teniposide, α-mangostin, and 2´3´-cGAMP to the STING ligand-binding domain (LBD), while the bottom panels show the total heat evolved per injection. The accompanying table summarizes the affinity constants, thermodynamic parameters, and stoichiometry values for each interaction. (b) Negative controls using a double mutant STING-LBD (R238A/Y240A), showing no detectable ligand binding. Experiments were conducted at 25°C in 20 mM HEPES, 150 mM KCl, pH 7.5. Recombinant STING-LBD (20 µM) was titrated with 300 µM of each ligand.

Figure 4

Computer model of the complex of STING with two units of Teniposide. (a) (top) Stick representation of the initial conformation, and (bottom) the representative structure of the first three most populated conformers extracted from the 100 ns MD simulation of the two Teniposide units in explicit water. The ratio between the relative energy of the conformer (E_r) and the probability of each microstate according to the Boltzmann distribution (p_i) are shown. (b) PyMOL stick and cartoon representation of the representative structure extracted from the 400 ns MD simulation of the STING homodimer (monomer A in green and monomer B in cyan) in complex with the untethered conformer c2 (Teniposide A and B represented as yellow and lime green sticks, respectively) seen both from the front (left) and from the top (right) to account for the symmetrical interaction. For the sake of clarity, in the detailed view of the ligand-receptor interactions (below), only polar hydrogens and the side chains of the amino acids that interact with the two Teniposide units are shown as sticks. The hydrogen bonds established between the ligands and the proteins are shown as black dashed lines and the amino acids involved in these interactions are labelled in bold. (c) Per residue energy decomposition, extracted from the MD simulation, of the interaction of both Teniposide units with the amino acids highlighted in b from monomers A and B (green and cyan bars, respectively).