Short-term metabolic disruptions in urine of mouse models following exposure to low doses of oxygen ion radiation

Abstract

Molecular alterations as a result of exposure to low doses of high linear energy transfer (LET) radiation can have deleterious short- and long-term consequences on crew members embarking on long distance space missions. Oxygen ions (¹⁶O) are among the high LET charged particles that make up the radiation environment inside a vehicle in deep space. We used mass spectrometry-based metabolomics to characterize urinary metabolic profiles of male C57BL/6J mice exposed to a single dose of 0.1, 0.25 and 1.0 Gy of ¹⁶O (600 MeV/n) at 10 and 30 days post-exposure to delineate radiation-induced metabolic alterations. We recognized a significant down regulation of several classes of metabolites including cresols and tryptophan metabolites, ketoacids and their derivatives upon exposure to 0.1 and 0.25 Gy after 10 days. While some of these changes reverted to near normal by 30 days, some metabolites including p-Cresol sulfate, oxalosuccinic acid, and indoxylsul-fate remained dysregulated at 30 days, suggesting long term prognosis on metabolism. Pathway analysis revealed a long-term dysregulation in multiple pathways including tryptophan and porphyrin metabolism. These results suggest that low doses of high-LET charged particle irradiation may have long-term implications on metabolic imbalance.

1. Introduction

The primary sources of ionizing radiation in deep space are solar particles and Galactic Cosmic Rays (GCR), comprised of a multitude of high-linear energy transfer (LET) charged particles of various sizes, charges, and energies.^1–6 Exposure to low doses of high-LET charged particles is associated with various adverse biological effects in animal models, which raises the concern about the health of future travelers in deep space.^7–10 Although mechanisms by which low doses of high-LET radiation alter cellular and tissue physiology are not fully known, the molecular effects may include cellular oxidative stress and DNA damage, which may trigger many cellular responses including inflammation and cell death.^11–16

Although the material that makes up the hull of the spaceship can shield astronauts from some smaller charged particles, a portion of the heavier ions such as ⁵⁶Fe will be fragmented by the space craft hull, leading to a radiation environment of smaller ions like ⁴He, ¹²C and ¹⁶O in the interior of the spaceship.^17,18 These fission products can result in a significant fraction of the total mission dose to the crew members and can result in normal tissue injury.¹⁹

To help identify health risks associated with long-distance space missions, investigations into molecular alterations due to exposure to low doses of high LET radiation may provide insight in short- and long-term physiological consequences. We have previously reported on significant changes in the composition and function of gut microbiota as part of a multimodal response to exposure to single low doses of ¹⁶O (600 MeV/n) in male C57BL/6J mice.⁶ In this study, we identified changes in metabolic profiles of urine samples obtained from these same mice 10 and 30 days after exposure to ¹⁶O.

Exposure to single doses of ¹⁶O (600 MeV/n, 0.05 – 1.0 Gy) as a representative of space type radiation has been shown to cause metabolic perturbations that may contribute to tissue and organ damage in various animal models.^20,21 Going into this study there were multiple outstanding questions including: What metabolic pathways are altered after acute exposure to a low dose of ¹⁶O? Can we find evidence of metabolic perturbations in body fluids like urine that can be developed as a minimally invasive assay? Can we design a prediction model to confirm exposure?

We utilized liquid chromatography mass spectrometry – time of flight (LCMS-TOF) technology to characterize metabolic profiles, followed by tandem mass spectrometry (MS/MS) to confirm the putative identity of metabolites that were significantly different between irradiated mice and the un-irradiated control group. Pathway analysis was used to delineate significantly altered metabolic pathways. The overall study design and workflow is detailed in Figure 1. Finally, dysregulated metabolites observed at the early time point (10 days) augmented the development of a biodosimetry model that identified radiation exposure and may also predict metabolism related long term consequences. Because of the large number of available samples, an approach of training, testing and validation could be used.

Figure 1.

The experimental design and workflow. In this study, male C57BL/6J mice (6 months of age) were divided into 4 groups depending on the dose of ¹⁶O irradiation. Urine samples were collected at 10 days and after 30 days after exposure. Samples were extracted, screened, and analyzed using LCMS-QTOF. Detected peaks for each feature obtained from irradiated mice were compared to sham (0 Gy). Statistically significant features were further annotated using tandem MS/MS analysis and statistical analysis was performed.

2. Material and methods

2.1. Mice and irradiation procedure

The animal model and radiation exposures have been described before.^6,22 In short, 4-week old male C57BL/6J mice (Jackson Laboratory) were kept at the Division of Laboratory Animal Medicine, University of Arkansas for Medical Sciences (UAMS) with unlimited access to food (soy-free rodent diet 2020X, Harlan Teklad) and water. Mice were kept under alternating cycles of 12 hours light and 12 hours dark and allowed to age to 6 months (comparable to an age of about 40 years in humans). At the age of 6 months, animals were transferred to Brookhaven National Laboratory (BNL) and left to acclimate for one week with the same cage mates and the same chow and water supply as at UAMS.⁶ Each mouse was placed in a well-ventilated transparent Lucite cubes (3 × 1½ × 1½ in.) and exposed to whole body ¹⁶O irradiation (600 MeV/n; 0.1, 0.25, or 1.0 Gy, 0.21–0.28 Gy/min), with 34–35 mice per dose group, at the NASA Space Radiation Laboratory (NSRL) at BNL. Sham-irradiated (or 0 Gy) mice were transported to NSRL and put in the same holders but without exposure to ¹⁶O. One day after (sham-) irradiation, all mice were brought back to UAMS by air transportation. Upon return to UAMS, all mice were placed on 2020X diet including 0.68 g/kg fenbendazole (Harlan Teklad) as part of the standard UAMS rodent quarantine protocol. At 10 and 30 days after irradiation, urine samples were collected for mass spectrometry based global metabolomics. Urine collection was performed between 8 and 10 am. Each mouse was restrained and held up briefly, and a urine sample was collected. Urine samples were spun down for 5 minutes, and supernatants collected and snap frozen before storage at −80 °C.

2.2. Sample preparation for mass spectrometry analysis

In this study, 376 urine samples were collected from a total of 138 mice exposed to ¹⁶O or sham irradiation. Metabolites were extracted by combining 20 μL of urine with 80 μL of 50:50 acetonitrile/water containing the internal standards (10 μL debrisoquine (1 mg/mL in ddH₂O), 50 μL of 4-nitrobenzoic acid (1 mg/mL in methanol) per 10 mL acetonitrile/water). The samples were kept on ice for 20 minutes, then centrifuged at 15,493 × g at 4 °C for 20. The supernatant was transferred to mass spectrometry vials for LC-MS analysis.

2.3. Urine analysis using UPLC-QTOF mass spectrometry

For metabolomics analysis, 1.5 μL of each sample was injected onto a 130 Å, 1.7 μm, 2.1 mm × 100 mm Acquity UPLC BEH C18 reverse-phase column (Waters Corporation, Milford, MA) kept at 40 °C using an Acquity UPLC system (Waters) with a gradient mobile phase consisting of 100% water containing 0.1% formic acid (Solvent A) and 100% acetonitrile containing 0.1% formic acid (Solvent B) and resolved for 11 min at a flow rate of 0.5 ml/min. The gradient began at 95% A and was held for 0.5 minutes. It then changed to 80% A by 4 minutes and then slowly changed to 5% A by 8.0 minutes. The gradient was then held for 1 minute before returning to the starting gradient at 11 minutes. The column eluent was introduced directly into the mass spectrometer by electrospray ionization. Mass spectrometry was performed on a Synapt G2-Si QToF MS (Waters Corporation), operating in either electrospray negative-ion (ESI−) or positive-ion (ESI+) - ionization mode with a capillary voltage of 1.52 kV for positive mode and 1.00 kV for negative mode respectively. The sample cone voltage was set to 25 V in positive mode and 30 V in the negative mode and the source offset was set to 80 °C. The desolation gas flow was set to 1000 liters/hour and the temperature was set to 450 °C. The cone gas flow was 25 liters/hour, and the source temperature was 120 °C. Accurate mass was maintained by infusing Leucine Enkephalin (556.2771 [M + H]⁺ or 554.2615 [M−H]⁻) at a concentration of 1 ng/μL in 50% aqueous ACN and a rate of 7 μL/min via the Lockspray interface (Waters). Data were acquired in centroid mode from 50 to 1200 m/z in MS scanning. Pooled QC (quality control samples) consisting of an aliquot of each of the samples were run throughout the batch to ensure minimal shifts in retention times and intensities (Supplementary Figure 1 A).

2.4. LC-Ms data processing and statistical analysis

The UPLC-QTOF-MS data were converted to NetCDF using MassLynx (Waters) and preprocessed in R using an in-house implementation of XCMS (Scripps Institute, La Jolla CA) to generate a retention time corrected peak intensity table filtered and detected by the matched filter algorithm that bins the data into predefined widths and masses and then compares them to known peaks of similar distributions. Retention time correction was performed using the Ordered Bijective Interpolated Warping (OBI-Warp) algorithm.²³ The parameters for both algorithms were optimized using the Isotopologue Parameter Optimization (IPO) R package.²⁴ Preprocessing resulted in 5956 features in positive mode and 4482 features in negative mode. Intensities of features (mass-to-charge ratio with associated retention time) were normalized to the internal standards added during sample preparation and by Quality control–based robust LOESS (locally estimated scatterplot smoothing) signal correction (QC-RLSC) normalization (Supplementary Figure 1B).²⁵ The data were log transformed and Pareto scaled, and binary comparisons were performed between the sham and irradiated samples at each dose (0.1 Gy, 0.25 Gy, 1.0 Gy) and time point (10 days, 30 days) to identify significantly dysregulated features using unpaired t-tests (false discovery rate (FDR) < 0.05). Significantly dysregulated features were subjected to MS/MS and annotated against spectral databases. Additionally, pathway analysis using the LC-MS features was conducted using Mummichog v2.06.²⁶ Principle component analysis (PCA) was performed to visualize group differences based on overall metabolite profiles.

Mass search was performed using the “cmmr” R package (CEU Mass Mediator RESTful API) for metabolite annotation. The databases searched by the “cmmr” R package included KEGG, HMDB, LipidMaps, METLIN, and PubChem. Next, in order to confirm the identity of significantly dysregulated metabolites, we performed tandem mass spectrometry wherein the MS/MS spectra were matched against the METLIN database. For biomarker panel selection we used a 100-fold cross-validation approach to determine hyper parameters and calibrate the prediction model in the discovery set and testing set. Then, the optimal value of lambda, obtained by the cross-validation procedure, was used to fit the model. Finally, all the features with non-zero coefficients were retained as the candidate biomarker panel. Receiver operator characteristic curves were used to determine the efficiency of biomarker panels using training and validation sets. The ROC curve can be understood as a plot of the probability of classifying the positive samples correctly against the rate of incorrectly classifying true negative samples. Therefore, the AUC measure of an ROC plot is a measure of predictive accuracy.

3. Results

3.1. Exposure to ¹⁶O caused dose- and time-dependent perturbations in urine metabolites

Differences in urine metabolite profiles between experimental groups were visualized using three-dimensional PCA plots that were based on all detected features in all radiation doses after 10 days (Figure 2A) and 30 days (Figure 2B). At day 10 after doses of 0.1, 0.25, and 1.0 Gy, a total of 3781, 3011, and 49 features were dysregulated, respectively and at day 30, the number of statistically significant dysregulated features for 0.1, 0.25, and 1.0 Gy were: 4273, 4008, and 3994, respectively (Supplementary Table 1 A). These results indicate that the lower doses of ¹⁶O (0.1 and 0.25 Gy) have more prominent effects shortly after irradiation, while the relatively highest dose examined (1.0 Gy) was more influential after a period of time. Moreover, group separation decreased with increasing radiation dose (Figure 2A). Meanwhile, there was a significant difference between each individually tested dose and sham group at 30 days (Figure 2B). The total number of statistically significant altered features following each dose administration at both time points and the extent of metabolic overlapping amongst different cohorts was visualized as a Venn diagram (Figure 3 and Supplementary Table 1B), which showed that the vast majority of altered features were detected upon administration of the lower doses (0.1 Gy and 0.25 Gy), especially at 10 days post exposure. Interestingly, the highest radiation dose (1.0 Gy) did not cause major metabolic alterations at this time point. On the other hand, 30 days after exposure, all three doses shared similar metabolic perturbations.

Figure 2.

A three-dimensional principal component analysis (PCA) plot (top 3 principal components) reveals a robust metabolic response at low doses of ¹⁶O radiation at day 10 (panel A) and day 30 (panel B). The 0.1 Gy dose resulted in the most significant disruptions in the metabolic profiles compared to the other doses at day 10. At day 30, on the other hand, there was a significant difference in metabolomic profiles between all tested doses and the sham group.

Figure 3.

A Venn diagram illustrating the degree of overlap between detected metabolites for the three tested doses of ¹⁶O radiation at each time point. At day 10, there was a high degree of overlap between the altered metabolites in the lower doses compared to the higher dose. At day 30, on the other hand, the majority of detected metabolites were altered in all 3 radiation doses, revealing a delayed effect of the higher dose.

To further investigate the identity of the dysregulated metabolites, MS/MS putative annotation analysis was performed. Multiple metabolites were found to be downregulated following radiation exposure. Supplementary Table 2 lists all annotated metabolites, their m/z values, and their corresponding collision-induced dissociation (CID) key fragments. Furthermore, Supplementary Table 3 discloses all annotated metabolites, the p-value, FDR-adjusted p-value and the corresponding fold change compared to sham at the same time point. Figure 4 comprises heatmap that illustrates the overall behavior of each individual metabolite along with the corresponding FDR and fold change values. For the most part, these metabolites were downregulated at 10 days after 0.1 and 0.25 Gy. Meanwhile, the 1.0 Gy group showed most differences from sham at the 30-day time point, emphasizing a delayed metabolic perturbation after the highest dose.

Figure 4.

A heatmap showing annotated metabolites that were significantly altered accompanied by p-values, False Discovery Rate adjusted (FDR-adjusted) p-values and fold change. The group averages demonstrate a decline in most annotated metabolites as a result of exposure to low doses (0.1 Gy and 0.25 Gy) of ¹⁶O, while this was not the case at the higher dose (1.0 Gy).

3.2. Pathway analysis discloses various commonalities in altered pathways that may contribute to the early response to the lower radiation doses

Next, we performed pathway analysis with all detected features that showed significant alterations in both modes utilizing Mummichog v2.06. The outcome of Mummichog encompasses putative metabolite annotations and proposes incorporated pathways based on a statistical inference using built-in pathway information.²⁶ Ten days after 0.1 and 0.25 Gy, several metabolic pathways were found to be dysregulated, including porphyrin metabolism, purine metabolism, sialic acid metabolism, androgen/estrogen metabolism, phosphatidylinositol phosphate metabolism, and biopterin metabolism. Meanwhile, tryptophan metabolism, tyrosine metabolism and porphyrin metabolism were the main perturbed metabolic pathways at 30 days after both low doses. The highest dose of ¹⁶O radiation (1.0 Gy) resulted in alterations in metabolic pathways including linoleate metabolism, androgen/estrogen biosynthesis and metabolism at both time points, while porphyrin metabolism and tryptophan metabolism appear to play a key role at the 30-day time point (Figure 5). Supplementary Table 4 includes all pathways as obtained from Mummichog 2.06 analysis.

Figure 5.

Pathway analysis created by Mummichog v2.06 based on all detected metabolites in both ESI + and ESI− Modes.

To use the information obtained from this study for biodosimetry, we constructed receiver operating characteristic (ROC) curves. After MS/MS putative annotation of significantly altered metabolites, normalized intensities of the annotated metabolites were used for metabolite selection using an ROC regulated learning technique, which uses the least absolute shrink-age and selection operator (LASSO) penalty. Using a training, testing, and validation study design, we randomly divided half of the sample set as discovery and the remainder ¼ as the internal testing cohort and ¼ as the external validation set. The classification performance of the biomarker panel is assessed using the area under the ROC curve (AUC). The ROC curve is created by plotting the true positive rate (TPR) against the false positive rate (FPR) at various threshold settings. We used 5 selected metabolites with an AUC closest to unity to develop a prediction model that can help in future diagnosis of radiation exposure (Figure 6).

Figure 6.

A prediction model for low dose ¹⁶O radiation effect at 10 days based on selected 5 metabolites panel. Panel A: Receiver Operating Characteristic (ROC) curve of the biomarker panel in differentiating radiation from the sham group in training/testing. Panel B: The ROC curves of the radiation effect prediction model in discrimination of irradiated group from Sham in the validation set. AUC = Area Under Curve.

4. Discussion

Male C57BL/6J mice were exposed to a single dose of ¹⁶O (600 MeV/n, 0.1 – 1.0 Gy), and urine samples collected at 10 and 30 days after irradiation were analyzed. At 10 days post exposure, the lower doses (0.1 Gy and 0.25 Gy) of ¹⁶O caused a dysregulation in the levels of p-Cresol and its metabolites as well as indole and its metabolite, leading to an imminent decrease in their urine levels (FDR adjusted p value < 0.05). This phenomenon could be due to the activation of cellular inflammatory or repair mechanisms which can impart a broad array of metabolic perturbations.^27,28 Meanwhile, these lower radiation doses may not have been sufficient to trigger cell death. The higher dose (1.0 Gy) caused more lasting changes in urine metabolites at 30 days, which may be explained by a more robust cellular damage or even some cell death.

We also found evidence to the occurrence of perturbations in the folate metabolism pathway, which is responsible for the single carbon transfer mechanisms and tyrosine metabolism. The compound p-Cresol is one of the metabolites of the amino acids tyrosine and phenylalanine, which are converted to p-hydroxyphenylacetic acid by intestinal bacteria, before being decarboxylated to p-Cresol. Sulfatation and glucuronidation by anaerobic intestinal bacteria generate p-Cresol sulfate and p-Cresol glucuronide. The dwindling levels of p-Cresol in urine of 16O exposed animals may reflect a distortion in intestinal microbiota. A reduction in p-Cresol production in the intestine consequently leads to a decrease in the levels of the secondary metabolites p-Cresol sulfate and p-Cresol glucuronide.^29–31 p-Hydroxyphenylacetic acid is a member of the 1-hydroxy-2-unsubstituted benzenoids and is another secondary metabolite of p-Cresol, however, it is also biosynthesized in mammalian tissues. Cinnamoylglycine is the primary urinary metabolite of cinnamic acid in mammals.³² The decrease in the urinary level of cinnamoylglycine is suggestive of a disruption in the microbiota environment due to high LET radiation exposure. Furthermore, 2-Methylhippuric acid is biosynthesized by microbiota through the coupling of glycine to benzoic acid or its derivative. A reduction in the level of this metabolite may be due to the disturbance in the domestic bacterial environment in the mammalian gut.^33,34 Lastly, the compound 3-methylindole is a urinary metabolite of L-tryptophan and is produced by the intestinal microbiota, and indole-3-carboxylic acid is a urinary indolic tryptophan metabolite usually produced by domestic bacteria in the mammalian gut. The radiation-induced decreases in these indole-based metabolites may indicate dysregulation in gut microbiota and alterations in the tryptophan metabolism pathway. Altogether, many of the metabolic perturbations in the urine of the irradiated mice suggest changes in the intestinal microbiome. These results corroborate our prior studies that highlighted long-term changes in the microbiome of the same ¹⁶O exposed mice that were studied here.⁶

Moreover, we observed downregulation of N- Acetyl glucosamine-6-sulfate in mice exposed to lower doses of oxygen radiation; this metabolite is a physiological intermediate generated during the degradation of keratan sulfate and usually hydrolyzed intralysosomally by N-acetylglucosamine-6-sulfate sulfatase. A decrease in the levels of this metabolite may indicate a disorder in the lysosomal enzyme system or an induction of the hydrolysis of this metabolite via an alternative route.³⁵ The compound L-3-Amino-isobutanoic acid is produced from S-methylmalonate semialdehyde by the enzyme 4-aminobutyrate aminotransferase and can be indicative of pyrimi-dine metabolism.^36–38

The pathway analysis indicates that the low doses of ¹⁶O (0.1 and 0.25 Gy) triggered changes in metabolic pathways at the early stages after exposure such as phosphatidylinositol phosphate metabolites, androgen/estrogen biosynthesis and metabolism, porphyrin metabolism and bioptyrin metabolism, with sialic acid metabolism being the single unique pathway that was manifested exclusively in the low doses after 10 days. Sialic acid is an integral component of glycosylated protein and it is usually part of cellular walls of microbiota. Dysregulation in urinary secretion of sialic acid may be due to the imbalance in bacteria flora. Also, it is widely present in a broad array of tissues and may be an evidence of excessive tissue injury.^39,40

The later effect of the small doses involves tryptophan metabolism, tyrosine metabolism, and porphyrin metabolism. Furthermore, pathway analysis of the higher dose (1.0 Gy) reveals that a delayed metabolic imbalance (30 days after irradiation) could be due to the dysregulation in porphyrin metabolism, not seen the earlier time point. Unlike tryptophan metabolism, which is modulated by stress signaling, porphyrin metabolism was reported to decrease as a result of exposure to ionizing radiation.^41,42 Based on these findings, it seems that the 0.1 and 0.25 Gy doses of ¹⁶O and the 1.0 Gy dose cause physiological changes through distinct mechanisms.

Finally, we used the data obtained at 10 days after irradiation to develop a prediction model based on the five urinary metabolites showing significant dysregulation, to diagnose prior exposure to ¹⁶O ionizing radiation and to anticipate future prognoses. Due to the importance of early detection and diagnosis of radiation exposure, we constructed the model based on the data extracted from the 10-day time point. Compounds S-(Formylmethyl) glutathione, p-Cresol glucuronide, Phenylacetylglycine, 1-Methyladenosine, and N-acetylglucosamine-6-sulfate were chosen to develop this model. As indicated by the value of AUC (approaching unity), the prediction model successfully distinguished between the irradiated group and the sham group for the low doses (0.1 Gy and 0.25 Gy). However, the model failed to discriminate between the higher dose (1 Gy) irradiated group and the sham as demonstrated by the low AUC value in the validation set (AUC = 0.543). This might be related to the differential cellular response to low doses of high LET radiation as compared to the higher dose. We have previously reported similar findings with microbiome and fecal metabolite profiles wherein low doses (<0.5 Gy) seem to trigger a hypersensitivity response such that cell continues to divide while accumulating damage whereas 1 Gy seems to be a threshold dose that triggers repair processes in response to radiation exposure.⁶

In summary, to the best of our knowledge, this is one of the first reports on short-term urine metabolic changes in irradiated mouse models using LC-MS. In this discovery sample set, we noticed dramatic changes in the level of urine metabolites upon exposure to low doses of ionizing ¹⁶O radiation (0.1 Gy and 0.25 Gy) compared to the higher dose of 1.0 Gy. While most of the detected metabolites were downregulated 10 days after 0.1 and 0.25 Gy, these metabolites showed no change after 1.0 Gy. These early changes in metabolite levels after low doses are possibly due to perturbations in a myriad of pathways including tryptophan metabolism, porphyrin metabolism, purine metabolism and sialic acid metabolism. While the impact of the low doses seemed consistent from 10 up to 30 days after exposure, interestingly, the higher dose created a delayed metabolic imbalance which may include the porphyrin and tryptophan metabolisms as essential key metabolic pathways. The putatively identified metabolites were consistent with the literature and affirmed the occurrence of changes in gut microorganisms. While due to practical limitations, we were only able to study single exposures to ¹⁶O at high dose rates, to improve the modeling of radiation exposures as encountered in deep space, future studies need to be designed to identify metabolic perturbations of chronic exposures to low-dose rate high-LET radiation. In addition, the role of gender difference and aging in the response to heavy ion irradiation may need to be thoroughly investigated in future studies.

Data availability statement

The data that support the findings of this study are openly available in Dryad Digital Repository at https://doi.org/10.5061/dryad.gb5mkkwn9.