H1299 non-small cell lung cancer cells (CRL-5803, American Type Cell Culture (ATCC) repository) were cultured in RPMI 1640 basal cell culture medium (Corning, NY, USA) supplemented with 10% fetal bovine serum (FBS; Atlas Biologicals, Fort Collins, CO, USA). Oncogenically non-transformed, immortalized human bronchial epithelial (HBEC) HBEC-3KT cells were a kind gift from Dr. John Minna (University of Texas Southwestern Medical Center, Dallas, TX, USA) and cultured as described previously in Keratinocyte Serum Free Medium (Thermo Fisher, Waltham, Massachusetts, USA) (20–22). WTK1 lymphoblastoid cells were maintained in suspension culture using RPMI 1640 basal medium supplemented with 10% FBS (23). Eight-week-old female C57BL/6 and C3H mice were purchased from Charles River Laboratories (Charleston, SC, USA), and husbandry was conducted according to protocols approved by the Institutional Animal Care and Use Committee (IACUC) at the University of Texas Southwestern Medical Center (UTSW).

In vitro cellular irradiations were conducted utilizing a Mark 1 ¹³⁷Cs sealed source irradiator (J.L. Shepherd and Associates, San Fernando, CA, USA) at a dose rate of 3–4 Gy/min. Animal irradiations were conducted utilizing an X-rad 320 irradiator (Precision X-Ray, Madison, CT, USA) running at 250 kVp at 15 mA (24). Irradiations to mimic clinical SAbR protocols were conducted utilizing a Cx225 (Precision X-Ray, Madison, CT, USA) irradiator capable of image-guided treatment planning using an on-board CT. For a detailed breakdown of each irradiator’s usage, dosimetry, and field geometry, please refer to Supplementary Table 2.

During irradiation, animals were immobilized using isoflurane inhalation at a 1%–2.5% concentration in oxygen. Animals were euthanized using CO₂ inhalation at a flow rate sufficient to displace the volume of the container at a rate of 30%–70% per minute. After 5 minutes of asphyxiation, cervical dislocation was utilized as a confirmation of euthanasia.

HBEC3 KT cells were irradiated with 10 Gy of ¹³⁷Cs γ-rays. At 24 hours post-irradiation, cells were fixed and stained for the presence of γH2AX and 53BP1 foci, representing DNA double-strand breaks (DSBs). Samples were stained and analyzed in biological triplicate (irradiated separately) and pooled for analysis. Foci in over 100 cells were counted for each replicate. In addition, more than 200 were evaluated for the presence of micronuclei, and the number of micronuclei per cell was reported for each treatment cohort (25).

To determine whether AVA altered the mutational burden of surviving cells, the mutation status of the thymidine kinase (TK) gene was tested using the WTK1 lymphoblastoid cell line that harbors a heterozygous mutation in the TK locus that makes it an effective model to test for deleterious mutation production by radiation (23, 26). The assay was performed in biological triplicate (cell cultures plated 48 hours prior to irradiation), and samples were irradiated separately and plated separately for mutational analysis. Error bars represent the SEM of three biological replicates. Additional information is included in the Supplementary Methods.

To determine whether AVA altered the induction of chromosome and chromatid aberrations 24 hours post-IR, AVA was added to the cultures of WTK1 cells 1 hour prior to a single acute dose of 4 Gy. Cells were allowed to incubate for 24 hours post-exposure, treated with Colcemid solution for 2.5 hours to enrich the percentage of cells in metaphase, and fixed to examine chromosome aberrations. The preparation of metaphase chromosome spreads and cytogenetic analysis were performed as previously reported (27, 28). Three independent cultures were irradiated per treatment group, and 50 metaphases were examined per biological replicate. Error bars represent the SEM of three independent biological replicates. Additional information is included in the Supplementary Methods.

Ten-week-old female C57BL/6J mice were focally irradiated to the left lung using a 3-mm collimated X-ray beam to total doses of 54, 60, and 70 Gy in a single fraction. Targeted irradiation was performed using fluoroscopy. AVA was delivered once as a single 24 mg/kg i.p. injection delivered 30–60 minutes prior to irradiation. Lungs were perfused, and tissues were collected at time points of 4, 8, 12, and 24 weeks post-irradiation. Doses of 60 and 70 Gy were selected to determine if a dose threshold exists that limits the efficacy of AVA as a radioprotector, and tissues were collected at a time point of 12 weeks post-irradiation for histological analysis for these doses. Error bars represent the SEM of the mean, and the statistical unit of analysis was the individual animal.

Due to the limited amount of lung volume irradiated, a modified Ashcroft scale was developed and used by a board-certified veterinary pathologist in a double-blind manner. A fibrosis score was generated according to the following scale and assessed using both Masson’s trichrome (MTC) and H&E-stained consecutive sections. For MTC-stained sections, 0 = no significant findings, 1 = <10% of the lung area affected, 2 = 10%–20% of the lung area affected with a single focal area, and 3 = 10%–20% of the lung affected with multiple focal lesions. For H&E-stained sections, 0 = no fibrosis, 1 = mild fibrosis, 2 = moderate fibrosis, and 3 = severe fibrosis.

Semi-quantitative image analysis was utilized to provide an alternative independent method for establishing the extent and severity of fibrosis. The standard ImageJ package was utilized without the use of additional scripts, plugins, or macros to analyze ×40 Masson’s trichrome-stained images of murine lung as described previously (29). Image acquisition and image analysis were conducted by a board-certified veterinary pathologist, and images were blinded for independent analysis by a second individual. Briefly, full color images were gated to minimize all but blue (collagen)-stained regions, and a mask was applied to create a binary image. Gating threshold values were maintained throughout image analysis. Seven 100 × 100 µm squares were randomly placed throughout the field of three separate fibrotic frames from each animal, as well as three frames from the unirradiated portion of the lung of the same animal. Those numbers were averaged to measure an individual animal’s fibrotic area value and normalized against the density of the unirradiated lung. Individual animal fibrotic area measurements were averaged within each treatment group to generate the overall fibrosis score.

Animals were focally irradiated using a single fraction of 54 Gy delivered with a 10-mm sphere of radiation dose using the X-Rad 320 irradiator (Precision X-Ray). AVA was delivered as a 24 mg/kg i.p. injection once prior to irradiation to replicate the protection studies. Additional treatment arms with AVA included delivery starting 24 hours post-irradiation and continuing as daily injections (Monday through Friday) for 1, 4, 8, 12, or 20 weeks post-irradiation or starting at 1, 4, 8, or 12 weeks post-irradiation and continuing until euthanasia at week 20. Error bars represent the SEM of the mean, and the statistical unit of analysis was the individual animal.

Based on previous studies, the snouts of 10-week-old female C3H mice were focally irradiated using a 10-mm collimated beam with a single acute dose of 17 Gy to cause radiation-induced OM in 100% of animals at 11–13 days post-irradiation (30). AVA was delivered i.p. as a 24 mg/kg injection i) once 30–60 minutes prior to irradiation, ii) daily starting 24 hours post-irradiation until euthanasia, or iii) once prior to irradiation and then daily until euthanasia. On day 11 post-irradiation, animals were euthanized; tongues were excised and stained with toluidine blue to determine where epithelial ulceration had occurred and then fixed in 10% neutral buffered formalin for 48 hours. Following fixation, tongues were embedded in paraffin and stained with H&E. Tongue sections were imaged using a Keyence BZ-X710 (Keyence, Minneapolis, MN, USA), and epithelial layer thickness was measured as described previously (31, 32). Error bars represent the SEM of the mean, and the statistical unit of analysis was the individual animal.

Animals were focally irradiated to the snout with a single dose of 17 Gy and given a 21-day window to resolve (5 days in addition to the 16-day recovery period determined by measuring the epithelial layer thickness of the tongue following irradiation). On day 21 post-irradiation, animals were then focally re-irradiated to the snout with a single dose of either 12 Gy (the dose necessary to induce radiation-induced OM in ~50% of naive animals) or 17 Gy as before, with euthanasia and tissue harvest conducted on day 11 after the second irradiation. AVA was delivered i.p. as a 24 mg/kg injection 30–60 minutes prior to re-irradiation and then daily until euthanasia. Error bars represent the SEM of the mean, and the statistical unit of analysis was the individual animal.

All statistical analyses were conducted utilizing a two-tailed Student’s t-test with a probability of a type 1 error set to 0.05 as the threshold for significance and a power of 95%. For a detailed breakdown of the number of animals in each group and linked to a specific figure, please refer to Supplementary Table 1.

Previously, it was demonstrated that pre-treatment with AVA potentiated high-dose-per-fraction radiation therapy in NSCLC tumors through a mechanism that was dependent upon limited catalase activity in human tumor xenograft models (10). Presumably, this is due to increased DNA damage caused by elevated levels of hydrogen peroxide in the tumor that cannot be detoxified quickly enough by the activity of catalase. To test this hypothesis, H1299 NSCLC cells were irradiated with a single acute dose of 10 Gy with or without a 24-µM dose of AVA present. Staining of γH2AX and 53BP1 DNA damage foci was performed 24 hours post-exposure to detect the presence of residual DNA double-strand breaks. Representative immunofluorescence-stained images are found in Figure 1A. γH2AX and 53BP1 foci were significantly increased in the AVA-treated cells determined 24 hours post-IR exposure (Figures 1B, C). As further confirmation of an increase in DNA damage, the number of cells with micronuclei in AVA-treated cultures was also higher (Figure 1D), confirming an increase in DNA damage, which supports the results from clonogenic survival assays published previously. Representative micronucleus images can be found in Supplementary Figure 2.

To determine if the normal tissue radioprotective effect (Supplementary Figure 1) is moderated through a reduction in DNA damage (in contrast to the increase observed in H1299 NSCLC cells), immunofluorescence staining to determine the levels of γH2AX and 53BP1 foci was also performed 24 hours post-exposure to IR in cells irradiated with 10 Gy with or without AVA treatment. Representative immunofluorescence-stained images are found in Figure 1E. AVA, in contrast to the results observed in the H1299 tumor line, demonstrated significantly reduced numbers of DNA damage foci (Figures 1F, G, bottom panel) and numbers of cells with micronuclei (Figure 1H, bottom panel), with representative images found in Supplementary Figure 2. Therefore, these results indicate a differential response to the combination of IR and AVA treatment in tumor vs. normal cells.

To determine the impact of AVA on the radiation-induced mutation frequency in non-oncogenic cells surviving radiation exposure, WTK1 lymphoblastoid cells, commonly used for mutational analysis, were irradiated with doses of either 1 or 4 Gy, which resulted in a dose-dependent increase in TK^−/− mutant frequency (Figure 2A). However, when AVA was added to the cell cultures 1 hour prior to irradiation and allowed to incubate for 24 hours post-exposure, the observed mutational burden induced by radiation was reduced. Furthermore, cytogenetic analysis of WTK1 cells 24 hours after a single dose of 4 Gy revealed a significantly reduced number of total chromosome (combined chromosome and chromatid) aberrations in irradiated cells treated with AVA (Figure 2B). Representative images are also shown (Figure 2C).

The reduced DNA and chromosomal damage, along with the reduced mutational burden in cells treated with AVA, suggested that long-term radiation effects may also be reduced. Therefore, the efficacy of AVA in reducing late effects was tested in preclinical animal models of RILF at doses and treatment schedules approximating the treatment of NSCLC with SAbR.

C57BL/6 mice were irradiated with a single acute dose of 54, 60, or 70 Gy delivered as a vertical 5-mm collimated cone to the left lung, as determined utilizing fluoroscopic imaging. AVA was delivered as a single 24 mg/kg dose 30–60 minutes prior to irradiation. Fibrosis (as detected using Masson’s trichrome staining in histological sections) was examined 24 weeks post-irradiation at 54 Gy and 12 weeks post-exposure in cohorts receiving a higher dose (60, 70, and 80 Gy). Representative images at both ×2 and ×40 magnifications are represented in Figure 3A. Pre-treatment with AVA greatly reduced lung fibrosis at 25 weeks in the cohorts treated with 54 Gy and at 12 weeks in the cohorts irradiated with 60 and 70 Gy, as determined using two independent metrics: the pathological score and the modified Ashcroft scale (Figure 3B). The fibrotic area, as determined via semi-quantitative image analysis of the irradiated region of the lung, was reduced in the 54- and 60-Gy cohorts (Figure 3C). No protective effect of AVA for fibrosis was observed in the 80-Gy cohorts (data not shown). This suggests that fibrosis can be reduced with the application of AVA prior to at least as much as 60 Gy of radiation by either metric, while results for doses above 60 Gy are inconclusive and depend on the assay used to quantify fibrosis.

Since AVA reduces superoxide and daughter ROS regardless of source and is effective as a radioprotector in the context of radiation-induced lung fibrosis, whether AVA is also an effective radiomitigator of fibrosis generated by chronically upregulated oxidative stress and inflammatory signaling was tested. Animals were irradiated with a single dose of 54 Gy delivered with a 10-mm collimator to the peripheral lung. AVA dosing schedules started at 24 hours post-irradiation for increasing durations of time or started at later times post-irradiation until 20 weeks post-irradiation. Fibrosis was then determined at 24 weeks post-irradiation for all cohorts. The various dosing schedules of AVA administration are depicted in Figure 4, as well as the quantification of the fibrotic area and the relevant p-values to describe statistical significance. These results support the hypothesis that AVA is a radiomitigator when delivered starting 24 hours post-radiation exposure, whose mitigative properties increase as the time of use increases. However, if AVA delivery is delayed, the reduction in fibrosis is reduced and is eliminated if delivery begins 4 weeks or more post-irradiation.

AVA demonstrated efficacy as a radioprotector reducing the extent and intensity of radiation-induced OM in clinical trials of patients receiving conventionally fractionated intensity-modulated radiotherapy (IMRT) with cisplatin to treat HNSCC (13, 15) and esophagitis (17). However, because AVA enhanced the response of human HNSCC tumor xenografts to high-dose-per-fraction radiation exposure, whether AVA would be effective in protecting normal oral mucosa of the tongue from ablative radiation exposure (>8 Gy), akin to SAbR regimens sometimes used for HNSCC radiotherapy, was tested in animal models (10). Figure 5A describes the schematic of experiments to answer this question, while Figure 5B depicts three representative images of the tongue at ×10 magnification from each treatment arm. The quantification of epithelial layer thickness can be observed in Figure 5C. Radiation alone significantly reduced the thickness of the epithelial cell layer by an average of ~40% by day 11 post-irradiation (group 1 vs. group 2), with the epithelial layer thickness returning to that of the unirradiated epithelial layer by day 16 post-exposure (group 6). AVA was effective as a radioprotector (single dose 30–60 minutes prior to irradiation), significantly ameliorating thinning of the epithelial layer by day 11 post-irradiation (group 2 vs. group 3). When AVA was delivered as a radiomitigator (daily AVA dose starting 24 hours post-exposure until euthanasia), it was also effective at ameliorating the reduction in the epithelial layer thickness, albeit much less so than the protection arm (group 2 vs. group 4, and group 3 vs. group 4). Finally, when AVA was delivered as both a radioprotector and a radiomitigator, a significant reduction in the epithelial layer thickness due to radiation exposure was not observed (group 1 vs. group 5). Furthermore, epithelial layer thickness was greater than that of the protection-only arm, indicating that the most effective treatment protocol with AVA for preventing radiation-induced OM in the context of SAbR is as both a protector and a mitigator (group 3 vs. group 5).

Re-irradiation can elicit a phenomenon referred to as radiation recall, whereby normal tissues “remember” the original radiation dose and experience severe toxicities. Whether AVA could reduce radiation recall toxicity was also tested in the mouse tongue model with re-irradiation at day 21 after primary irradiation and utilizing the optimal protection + mitigation protocol where AVA was delivered both once prior to the initial irradiation and daily until euthanasia (Figure 6A).

Re-irradiation with a second dose of 17 Gy resulted in five of six animals not surviving to 11 days after the second dose, presumably because the severity of OM prevented food intake (data not shown). A re-irradiation dose of 12 Gy, however, was nearly as effective as the single primary dose of 17 Gy in reducing the thickness of the epithelial layer, i.e., recalling the first irradiation, indicating that a second, lower dose of radiation is effective at inducing radiation-induced OM in all animals (group 2 vs. group 7). AVA treatment beginning immediately before 12 Gy re-irradiation result significantly attenuated the effects of radiation recall (group 7 vs. group 8) (Figure 6B). Representative images on ×10 tongue sections can be found in Figure 6C.