Effects of linear energy transfer and hypoxia on radiation-induced immunogenicity through STING

Purpose: Preclinical studies have demonstrated that cancer cells may produce innate immune signals such as type-I interferons following radiation damage, which derives from activation of the cGAS-STING pathway following detection of cytosolic dsDNA. Limited studies have explored how these mechanisms vary from the conditions of the radiation exposure. High- linear energy transfer (LET) radiation induces more DNA double-strand breaks (DSB) per dose than low-LET radiation, thus is expected to be more immunogenic. However, DNA damage in hypoxic cells is more probable to undergo chemical repair due to limitations in oxygen fixation, thus is expected to be more immunosuppressive. Our goal is to study and model the dose response characteristics of IFNβ and Trex1 in vitro following exposure of radiations with varying LET and to develop techniques for further study in vivo.

Methods: Reference data from Vanpouille-Box (2017) on STING dose response was applied to develop empirical models of cytosolic dsDNA and Trex1 regulation as a function of dose and quantity of DNA DSB, the latter of which is dependent on particle LET and oxygenation and is calculated using Monte Carlo Damage Simulation (MCDS) software. These models were used as preliminary data to guide in vitro experiments using Merkel cell carcinoma cells. The dose response of pro-inflammatory IFNβ and exonuclease Trex1, an anti-inflammatory suppressor of cGAS-STING, was measured post-irradiation. MCDS was again used to model fast neutron relative biological effectiveness for DSB induction (RBEDSB) and compared to laboratory measurements of the RBE for IFNβ production (RBEIFNβ). RBEIFNβ models were applied to radiation transport simulations to quantify the potential secretion of IFNβ in representative clinical beams. To enable intra-tumor radiation targeting of tumor hypoxia, mice were seeded with syngeneic tumors and imaged longitudinally with PCT- spectroscopy to determine local variations hemoglobin concentration (Hb) and oxygen saturation (SaO2) over time. Hypoxia classification was based on SaO2 levels in voxels containing hemoglobin relative to a “hypoxia threshold” of SaO2 < 0.2.

Results: Based on analysis of published data, our preliminary models of cytosolic DNA and Trex1 dose responses demonstrate dose enhancements from high-LET radiation, such as that at the distal edge of a Bragg peak, and suppression from cellular hypoxia. This manifests as an RBE-dependent ‘shift’ in STING response. Laboratory measurements in MCC13 cells show peak IFNβ production at 6.1 Gy following fast neutron irradiation and 14.5 Gy following x-rays (RBEIFNβ = 2.4). However, IFNβ signal amplitudes were not significantly different between these radiation types. Trex1 signal increased linearly with dose, with fourfold higher upregulation per dose for fast neutrons. Modeling of RBE in clinical beams suggests that ion sources may induce spatially localized IFNβ near their end of range, which is potentially advantageous for initiation of tumor-specific immune activity. Uncharged sources stimulate IFNβ more uniformly with depth. Longitudinal PCT-S scanning is able to localize and distinguish chronic and acute hypoxia in vivo. Changes in the hypoxic classification from tumor growth and following anti-angiogenic therapy are distinguishable.