Multi-omic analysis of nonhuman primate heart after partial-body radiation with minimal bone marrow sparing

Abstract

High-dose radiation exposure results in hematopoietic and gastrointestinal acute radiation syndromes followed by delayed effects of acute radiation exposure, which encompasses multiple organs, including heart, kidney, and lung. Here we sought to further characterize the natural history of radiation-induced heart injury via determination of differential protein and metabolite expression in the heart. We quantitatively profiled the proteome and metabolome of left and right ventricle from non-human primates following 12 Gy partial body irradiation with 2.5% bone marrow sparing over a time period of 3 weeks. Global proteome profiling identified more than 2200 unique proteins, with 220 and 286 in the left and right ventricles, respectively, showing significant responses across at least three time points compared to baseline levels. High-throughput targeted metabolomics analyzed a total of 229 metabolites and metabolite combinations, with 18 and 22 in the left and right ventricles, respectively, showing significant responses compared to baseline levels. Bioinformatic analysis performed on metabolomic and proteomic data revealed pathways related to inflammation, energy metabolism, and myocardial remodeling were dysregulated. Additionally, we observed dysregulation of the retinoid homeostasis pathway, including significant post-radiation decreases in retinoic acid, an active metabolite of vitamin A. Significant differences between left and right ventricles in the pathology of radiation-induced injury were identified. This multi-omic study characterizes the natural history and molecular mechanisms of radiation-induced heart injury in NHP exposed to PBI with minimal bone marrow sparing.

INTRODUCTION

High-dose radiation exposure results in hematopoietic acute radiation syndromes (H-ARS) and gastrointestinal acute radiation syndrome (G-ARS), followed by delayed effects of acute radiation exposure (DEARE) (MacVittie et al. 2019, Fish et al. 2020). DEARE encompasses multiple organs, including heart, kidney, and lung (MacVittie et al. 2012). The partial body irradiation (PBI) with minimal, 2.5% or 5%, bone marrow (BM) sparing non-human primate (NHP) model was developed to mimic intentional or accidental radiation exposures in humans. Such exposures are likely to include some bone marrow sparing and to permit the concurrent analysis of coincident short- and long- term damage to organ systems in a time- and dose-dependent manner (MacVittie et al. 2012). Previous publications from our team have described this PBI model, the analysis of resulting tissue injury, as well as its use for medical countermeasure (MCM) development (MacVittie et al. 2012, Jones et al. 2015, MacVittie et al. 2015, Shea-Donohue et al. 2016, Cohen et al. 2017, Prado et al. 2017, Carter et al. 2019, Cohen et al. 2019, Farese et al. 2019, MacVittie et al. 2019, MacVittie et al. 2019, Parker et al. 2019, Parker et al. 2019, Parker et al. 2019). Here we sought to further characterize the natural history of radiation-induced injury in NHP via determination of differential protein and metabolite expression in the heart. A more in-depth characterization of the natural history of radiation-induced damage in NHPs exposed to PBI with minimal BM sparing increases the usefulness of the model for MCM development, as understanding of the natural history of an injury is a general expectation of drug development under the FDA Animal Rule (FDA 2015).

Characterizing radiation exposure to the heart has many beneficial clinical effects. Forty years after full-body radiation exposure, Japanese atomic bomb survivors have increased mortality from cardiovascular disease (Preston et al. 2003, Yamada et al. 2004, McGale and Darby 2005). Radiation is an extremely common treatment in cancer, and radiation exposure to the heart is possible for many cancers, including breast and lung. More than half of all breast cancer survivors whom received radiotherapy have cardiac damage (Hardenbergh et al. 2001) and have a greatly increased risk of cardiac-related death (Cuzick et al. 1994, Paszat et al. 1998, Yeh et al. 2004, Hooning et al. 2007, Darby et al. 2013, Bodai and Tuso 2015). The increased risk of cardiovascular injury and disease is also seen in Hodgkin’s disease patients who received radiotherapy (Hancock et al. 1993, Lee et al. 2000, Swerdlow et al. 2007). Additionally, studies have shown childhood leukemia patients treated with radiation developed cardiac abnormalities later in life (Pihkala et al. 1994). There is a strong link between radiation exposure and delayed cardiac disease, however, the mechanism of injury is not fully established. This study encompasses the first 3 weeks following radiation and may elucidate potential initiating events in radiation-induced heart injury.

Previous reports detailing the effect of radiation in rats and NHP often employed echocardiography and histology staining to assess heart injury (Baker et al. 2009, Lenarczyk et al. 2013, Medhora et al. 2015, DeBo et al. 2016, Jacobs et al. 2019, Parker et al. 2019, Zhang et al. 2020). While these techniques provide detailed information about the resulting heart injury post-irradiation, they do not provide data on the mechanism of injury. This study employs a comprehensive, comparative proteomic and metabolomic approach to identify differences between irradiated and non-irradiated NHP hearts. Proteomics is capable of measuring hundreds of thousands of proteins at a time and can identify a subset of proteins that changes significantly between groups. Our targeted metabolomics assay provided absolute quantitation of 188 metabolites across 8 classes. Together, these methods provide a powerful tool for the identification of the mechanism of injury of radiation-induced heart damage.

Additionally, this study characterized the effect of radiation exposure on retinoic acid (RA) metabolism. RA is the active metabolite of Vitamin A and a master regulator of gene expression through ligand-activated control of transcription mediated by the nuclear receptors retinoic acid receptor (RAR) and retinoid X receptor (RXR) (Germain et al. 2006, Germain et al. 2006). In the heart, RA levels are essential for proper development and function (Pan and Baker 2007, Xavier-Neto et al. 2015, Wang et al. 2018, Wang et al. 2018). Dysregulated RA levels in the heart have been associated with heart disease (Minicucci et al. 2010, Liu et al. 2016, Park et al. 2018, Park et al. 2019). In addition to the heart, radiation exposure has also been shown to cause a significant reduction in jejunum and lung RA levels (Jones et al. 2014, Huang et al. 2019, Huang et al. 2020).

To help inform on the natural history of the NHP PBI/BM2.5 model, we quantitatively profiled the proteome and metabolome of NHP left (LV) and right ventricles (RV) over a time period of 3 wk. Bioinformatic analysis revealed significant pathways that were perturbed post irradiation with the intent of identifying mechanism of radiation-induced heart injury. Our study also sought to compare the radiation response in LV and RV.

MATERIALS AND METHODS

Radiation animal model.

All animal procedures were conducted in accordance with the NIH guidelines for the care and use of laboratory animals and experiments were performed with prior approval from the University of Maryland Institutional Animal Care and Use Committee (IACUC). Tissue was collected from left and right ventricles of male rhesus macaques (Macaca mulatta); tissues were snap frozen in liquid nitrogen and stored at −80°C until assay. Samples were provided by the laboratory of Thomas J. MacVittie, University of Maryland School of Medicine, Department of Radiation Oncology (Baltimore, MD). Description of the animal models, including radiation exposure and dosimetry, medical management (supportive care and health monitoring), as well as collection of plasma and tissue have been previously described (MacVittie et al. 2012). NHP were exposed to 12 Gy partial body irradiation with 2.5% bone marrow sparing (PBI/BM2.5) with a peak 6 MV linear accelerator (LINAC)-derived photons with an average energy of 2 MV at 0.80 Gy min⁻¹. Bone marrow sparing was accomplished with tibiae outside the beam field. Left and right ventricle heart tissue was obtained from a natural history study where timed euthanasia was planned at day 4, 8, 15, and 21 after radiation. Additional heart samples were obtained from animals that were euthanized for cause according to criteria specified in the IACUC protocol. We grouped animals at day 8 and 9, day 11 and 12, and day 21 and 22 to increase the numbers of n at the timepoints across our study. Heart tissue from two non-irradiated animals were used as baseline controls. Thirty-two left ventricle samples and thirty-two right ventricle samples were analyzed in total with the numbers of NHP at each time point (n) as follows: day 0 (baseline), n=2; day 4, n=4; day 8-day9, n=11; day 11-day 12, n=7; day 15, n=5; day 21–22, n=3. The mean age of the irradiated animals was 3.7 years (median 3.7 years, range 2.9–4.7 years). The body weight of the irradiated animals was 5.2 kg (median 5.0 kg, range 3.0–7.0 kg). The normal NHP were 2.5 and 2.6 years old and were 4.3 and 5.0 kg. A full list of animal samples is provided in Supp. Table 1.

Proteomic sample preparation.

Tissue samples were stored in −80°C until processed. Left and right ventricle tissues were homogenized in phosphate buffered saline using Precellys CK14 lysing kit (Bertin Corp., Rockville, MD). Proteins were extracted, purified from tissue lysates, trypsinolyzed and desalted as described previously (Huang et al. 2020).

Liquid chromatography-tandem mass spectrometry data acquisition.

Tryptic peptides were separated on a nanoACQUITY UPLC analytical column (BEH130 C18, 1.7 μm, 75 μm x 200 mm, Waters) over a 165-minute linear acetonitrile gradient (3 – 40%) with 0.1 % formic acid on a Waters nano-ACQUITY UPLC system and analyzed on a coupled Thermo Scientific Orbitrap Fusion Lumos Tribrid mass spectrometer as previously described (Williamson et al. 2016) and previously used by us (Defnet et al. 2019, Kim et al. 2019). Full scans were acquired at a resolution of 240,000, and precursors were selected for fragmentation by collision induced dissociation (normalized collision energy at 35 %) for a maximum 3-second cycle.

Liquid chromatography-tandem mass spectrometry data analysis.

Tandem mass spectra were searched against a UniProt Macaca mulatta reference proteome using Sequest HT algorithm described previously (Eng et al. 2008) and MS Amanda algorithm developed by Dorfer et al. (Dorfer et al. 2014) with a maximum precursor mass error tolerance of 10 ppm. Carbamidomethylation of cysteine and deamidation of asparagine and glutamine were treated as static and dynamic modifications, respectively. Resulting hits were validated at a maximum false discovery rate of 0.01 using a semi-supervised machine learning algorithm Percolator developed by Käll et al. (Käll et al. 2007). Label-free quantifications were performed using Minora, an aligned AMRT (Accurate Mass and Retention Time) cluster quantification algorithm (Thermo Scientific, 2017). Protein abundance ratios between groups were measured by comparing the MS1 peak volumes of peptide ions, whose identities were confirmed by MS2 sequencing as described above. A full list of protein name, gene names, and accession numbers is provided in Supp. Table 2.

Proteomic bioinformatic and statistical analysis.

Pathway and gene ontology analysis were performed with Qiagen Ingenuity and DAVID databases, as described by Kramer et al. and Huang et al., respectively (Huang da et al. 2009, Kramer et al. 2014, Mi et al. 2017). Proteins showing at least a 2-fold change (FC) with an FDR adjusted ANOVA p-value < 0.05 were considered significantly changed and used for further analysis. Fisher’s exact p-values of < 0.05 were used in the gene ontology analyses to identify biological processes associated with observed protein changes. Ingenuity Pathway Analysis (IPA) analysis was used to predict canonical pathways according to the proteins that were significantly different using a Benjamini-Hochberg corrected Fisher’s exact test p-value < 0.05 with a non-zero absolute activation z-score.

Retinoid analysis.

Tissue samples were stored in −80°C until processed. Only glass containers, pipettes, and syringes were used to handle retinoids. Extraction of retinoids was performed under yellow lights using a two-step liquid-liquid extraction that has been described in detail previously using 4,4-dimethyl-RA as an internal standard for RA and retinyl acetate as an internal standard for retinol and total retinyl ester (Kane et al. 2005, Kane et al. 2008, Kane and Napoli 2010, Jones et al. 2015). Briefly, for the extraction of retinoids, 85.7 ± 17.7 mg of heart tissue was homogenized in 2 mL of 0.9% NaCl (normal saline) and two 1,000-μL aliquots were extracted as technical duplicates. To each homogenate aliquot, 2 mL of 0.025 M KOH in ethanol was added to the homogenate followed by addition of 10 mL hexane to the aqueous ethanol phase. The samples were vortexed and centrifuged for 1 to 3 min at 1,000 rpm in a Dynac centrifuge (Becton Dickinson) to facilitate phase separation and pellet precipitated protein. The hexane (top) phase containing nonpolar retinoids [retinol and total retinyl esters (RE)] was removed. Then 4 M HCl (65 μL) was added to the remaining aqueous ethanol phase, samples were vortexed, and then polar retinoids (RA) were removed by extraction with a second 10 mL aliquot of hexane as described above. Organic hexane phases were evaporated under nitrogen while heating at approximately 30°C in a water bath (model N-EVAP 112, Organomation Associates, Berlin, MA, USA). All samples were resuspended in 60 μL acetonitrile.

Levels of RA were determined by liquid chromatography-multistage tandem mass spectrometry (LC-MRM³) which is an LC-MS/MS method utilizing two distinct fragmentation events for enhanced selectivity (Jones et al. 2015). RA was measured using a Shimadzu Prominence UFLC XR liquid chromatography system (Shimadzu, Columbia, MD) coupled to an AB Sciex 6500+ QTRAP hybrid triple quadrupole mass spectrometer (AB Sciex, Framingham, MA) using atmospheric pressure chemical ionization (APCI) operated in positive ion mode as previously described (Jones et al. 2015). For the LC separation, the column temperature was controlled at 25°C, the autosampler was maintained at 10°C and the injection volume was typically 20 μL. All separations were performed using an Ascentis Express RP-Amide guard cartridge column (Supelco, 50 × 2.1 mm, 2.7 μm) coupled to an Ascentis Express RP-Amide analytical column (Supelco, 100 × 2.1 mm, 2.7 μm). Mobile phase A consisted of 0.1% formic acid in water, and mobile phase B consisted of 0.1% formic acid in acetonitrile. Endogenously occurring retinoid isomers including all-trans-retinoic acid (RA), 9-cis retinoic acid, 13-cis retinoic acid, and 9,13-di-cis retinoic acid are resolved using a gradient separation at a flow rate of 0.4 mL min⁻¹ with gradient conditions described previously (Jones et al. 2015). The APCI source conditions and MRM³ detection parameters were as previously described where the MRM³ transition for RA was m/z 301.1 → m/z 205.1 → m/z 159.1 and for 4,4-dimethyl RA was m/z 329.2 → m/z 151.2 → m/z 100.0 (Jones et al. 2015).

Levels of retinol and total retinyl esters (RE) were quantified via HPLC-UV according to previously published methodology (Kane et al. 2008, Kane and Napoli 2010). Retinol and RE were resolved by reverse-phase chromatography (Zorbax SB-C18, 4.6 × 100 mm, 3.5 μm) on a Waters Acquity UPLC H-class system and were quantified by UV absorbance at 325 nm. Analytes were separated at 1 mL min⁻¹ with a gradient separation as described previously with a typical injection volume of 30 μL.

The amount of retinoic acid (RA), retinol (ROL) and total retinyl ester (RE) was normalized per gram of tissue. Data are expressed as mean ± SEM with statistical significance assessed with a two-sided, unpaired student’s t-test as compared to the baseline value using Graphpad Prism V7.03.

High-throughput targeted metabolomics.

Targeted, quantitative metabolomics was performed using Biocrates AbsoluteIDQ p180 kit (Biocrates, Life Science AG, Innsbruck, Austria). The AbsoluteIDQ p180 kit was prepared as described by the manufacturer. The assay quantifies up to 188 metabolites from five metabolite classes: acylcarnitines, amino acids, biogenic amines, glycerophospholipids, sphingolipids, and hexose. Internal standards, analyte derivatization and metabolite extraction are integrated into a 96-well plate kit. Metabolite detection is done via pre-selected selected reaction monitoring (SRM) transitions.

Tissue samples were stored in −80°C until processed. Heart tissue was homogenized in ethanol/phosphate-buffered saline (85:15, v/v) (10 mg of tissues in 30 μl solvent) using Precellys CK14 lysing kit (Bertin Corp., Rockville, Maryland, US). The mixture was centrifuged at 10,000 g for 5 min at 4 °C. After centrifugation, 20 μL of supernatant was loaded onto the 96 well kit. 10 μL of internal standard cocktail was added followed by drying with nitrogen. Next, a 5% solution of phenylisothiocyanate in ethanol:water:pyridine (1:1:1, v/v/v) was added for derivatization of biogenic amines and amino acids. Metabolite extraction was achieved with 5 mM ammonium acetate in methanol. Analyses were preformed according to the manufacturer’s instructions on a tandem mass spectrometry platform that consisted of an ACQUITY I-Class UPLC (Waters Corporation, Milford, MA, USA) coupled to a Xevo TQ-S tandem quadrupole mass spectrometer (Waters corporation, Milford, MA, USA). MetIQ software (Biocrates) controlled the assay workflow including sample registration, calculation of metabolite concentrations, and assay validation. A total of 229 analytes passed QC.

Metabolomic nomenclature.

Lipid notations listed as “Class CXX:Y” notate particular lipids indicating the XX total carbon number in the fatty acid chains and the Y number of double bonds. For example, PCaa C36:3 is phosphatidylcholine (PC) 36:3; i.e., a phosphatidylcholine with 36 total carbons among the two alkyl chains with 3 double bonds present. PC lipids listed as “PCaa” have both moieties at the sn-1 and sn-2 position as fatty acids bound to the glycerol backbone via ester bonds. PC lipids listed as “PCae” have one of the moieties, either the sn-1 or sn-2 position, as a fatty alcohol bound via an ether bond. For sphingomyelins (SM), the number of double bonds or the presence of a hydroxyl group (OH) are indicated for the fatty acid in the amide bond with the assumption that the backbone is sphingosine (d18:1). SFA refers to a sum of PC lipids containing fully saturated fatty acids (SFA), MUFA refers to a sum of PC lipids containing monounsaturated fatty acids (MUFA), and PUFA refers to a sum of PC lipids containing polyunsaturated fatty acids (PUFA). Lipid species measurements include potential isobaric and isomeric species (Biocrates 2017).

Metabolomic bioinformatic and statistical analysis.

Statistical analyses were performed using the MetaboAnalyst 5.0 web-based statistical package and GraphPad Prism (v 7.03, La Jolla, CA) (Chong et al. 2018, Chong et al. 2019) The data generated from the AbsoluteIDQ p180 kit which included analyte name and calculated concentration were imported into MetaboAnalyst for multivariate analysis. The metabolite data were normalized via autoscaling (mean-centered and divided by the standard deviation of each variable). MetaboAnalyst requires a minimum of three replicates for statistical analysis, which required the duplication of our two Day 0 samples to establish baseline levels. For metabolomic pathway analysis, human KEGG library was chosen under the assumption that human pathway library is similar to rhesus macaque.

RESULTS

Study design

The left and right ventricles proteome and metabolome were assayed after radiation exposure as part of a natural history study in NHP after 12 Gy PBI/BM2.5. Heart samples were interrogated at various times after radiation up to 21 d after exposure, with sampling occurring at timed euthanasia as planned at day 4, 8, 15, and 21 after radiation (MacVittie et al. 2012).

Expression of proteins most changed after radiation

The heart proteome was assessed via LC-MS/MS to identify proteins that were altered in LV and RV NHP after radiation exposure. In the LV samples, a total of 2255 unique proteins were quantified, and protein abundances were compared between LV NHP samples at baseline and d 4, 8–9, 11–12, 15, and 21–22. Of those proteins, 220 proteins showed significantly altered abundances at three or more time points compared to baseline (Fig. 1, Supp. Table 3). In the RV, a total of 2454 unique proteins were quantified and 286 showed significantly altered abundances at three or more time points compared to baseline (Fig 2, Supp. Table 4). Significant changes in protein expression was determined with FDR < 0.01, protein expression FC > 2-fold, and FDR corrected ANOVA p-value cut-off < 0.05. Most of the proteins showed strong and consistent changes over the times interrogated.

Figure 1. Expression of proteins most changed in left ventricle (LV) after radiation.

LC-MS/MS proteomic analysis of NHP LV after 12 Gy PBI/BM2.5. Minimum 2-fold change of expression for at least 3 time points and FDR adjusted ANOVA p-value < 0.05 were criteria for inclusion.

Figure 2. Expression of proteins most changed in right ventricle (RV) after radiation.

LC-MS/MS proteomic analysis of NHP RV after 12 Gy PBI/BM2.5. Minimum 2-fold change of expression for at least 3 time points and FDR adjusted ANOVA p-value < 0.05 were criteria for inclusion.

Canonical pathways and upstream regulators altered by radiation

Ingenuity Pathway Analysis (IPA) was used to predict canonical pathways and upstream regulators that were significantly altered in response to radiation (Tables 1 and 2, respectively) (Kramer et al. 2014). Canonical pathway analysis provides information about key biological events post-irradiation. The calculated z-score is a statistical measure of the match between expected relationship direction from published literature and observed gene expression from the experimental dataset. It was used to infer likely activation states of canonical pathways based on comparison with a model that assigns random regulation direction. A non-zero absolute activation z-score and a Benjamini-Hochberg corrected p-value < 0.05 was used for inclusion in Table 1. Six canonical pathways were significantly altered in LV, including LXR/RXR activation and coagulation system. Three pathways were significantly altered in RV, including LXR/RXR activation, tRNA charging, and acute phase response signaling. A Benjamini-Hochberg corrected Fisher exact test p < 0.01 and a non-zero z-score were inclusion criteria for inferred upstream regulators in Table 2. Eight inferred upstream regulators were significantly activated (2) or inhibited (6) in LV. Nineteen inferred upstream regulators were significantly activated (14) or inhibited (5) in RV. PPARA is inhibited in LV and activated in RV.

Table 1. Canonical pathways altered by radiation .

Criteria for pathway changes were non-zero absolute activation z-score and Benjamini-Hochberg corrected Fisher’s exact test p-value < 0.05, as compared to baseline values. Derived from LC-MS/MS proteomic analysis of NHP left and right ventricles after 12 Gy PBI/BM2.5.

Canonical Pathways	B-H p-value	Ratio	Activation z-score	Proteins
Left Ventricle
LXR/RXR Activation	<0.000005	0.083	−1.0	KNG1, LPA, APOA4, SAA1, FASN, S100A8, PCYOX1, SERPINF2, APOD, CETP
Coagulation System	<0.00005	0.171	0.8	KNG1, F12, SERPINA5, SERPINF2, F2, SERPIND1
Cdc42 Signaling	<0.01	0.042	−1.3	MYL10, PPP1R12A, MYL4, HLA-F, CLIP1, HLA-DPA1, MYL7
Actin Cytoskeleton Signaling	<0.01	0.034	−2.4	KNG1, MYH4, MYL10, PPP1R12A, FLNA, MYL4, F2, MYL7
ILK Signaling	<0.05	0.034	−0.8	MYH4, PARVB, PARVA, PPP1R12A, FLNA, MYL4, MYL7
Protein Kinase A Signaling	<0.05	0.023	0.4	MYH4, PRKAR2B, MYL10, PPP1R12A, FLNA, ADD1, MYL4, TNNI1, MYL7
Right Ventricle
LXR/RXR Activation	<0.0005	0.083	1.3	HPX, LPA, ORM1, SAA1, C9, CD14, S100A8, PLTP, AGT, APOC3
tRNA Charging	<0.0005	0.154	2.4	NARS, YARS, CARS2, TARSL2, MARS, QARS
Acute Phase Response Signaling	<0.005	0.050	1.6	RAP2B, HP, HPX, ORM1, SAA1, C9, FGB, SERPINA3, AGT

Table 2. Upstream regulators altered by radiation in right and left ventricles.

Criteria for upstream regulator interference were a nonzero absolute activation z-score and a Benjamini-Hochberg corrected Fisher’s exact test p-value < 0.01, as compared to baseline values. Derived from LC-MS/MS proteomic analysis of NHP left and right ventricles after 12 Gy PBI/BM2.5.

Upstream regulator	Activation z-score	B-H corrected p-value	Target proteins in dataset
Left- activated
INSIG1	2.4	<0.01	ACACB, ACLY, FASN, G6PD, S100A8, VNN1
CFTR	1.0	<0.005	CFTR, CPA3, HSD17B11, MYL7, S100A8, SAA1
Left- Inhibited
OSM	−1.3	<0.01	ANXA1, CDA, CLIP1, COL3A1, CRP, FASN, HLA-F, MARCKS, PDAP1, S100A8, SAA1, SYNE1, TPM1
Nr1h	−1.3	<0.005	ACACB, ACLY, ALDH1A2, APOD, CETP, FASN, PLIN1, THY1
Ins1	−1.4	<0.01	ADH1C, CETP, CYB5A, FASN, G6PD, NPPA, PLIN1, PTGFRN, SLC2A1, THY1
CD38	−1.9	<0.01	GLIPR2, NCAM1, PRDX4, S100A4, SLC2A1, SPCS2, THY1
AGTR2	−2.0	<0.01	COL3A1, FASN, KNG1, NPPA
PPARA	−2.1	<0.01	APOA4, BCKDK, C8A, FASN, G6PD, HSD17B11, NPPA, PC, PLIN1, S100A8, SAA1, VNN1
Right- Activated
MRTFA	2.6	<0.005	CAMP, CMA1, CPA3, PRTN3, S100A8, S100A9, SLC1A2, TUBB1
PRL	2.6	<0.001	CNN3, COL1A2, COL3A1, EIF2AK2, GPNMB, IGFBP5, KRT5, MARS, OAS1, PCNA, PSME1, SERPINA3, TRIM25
MRTFB	2.4	<0.005	CAMP, CMA1, CPA3, PRTN3, S100A8, S100A9, TUBB1
SRF	2.3	<0.005	BPGM, CALB1, CAMP, CMA1, CPA3, ITGA2B, MYH2, PMFBP1, PRTN3, S100A8, S100A9, SLC2A1, TCAP, TMOD3
MYOD1	2.3	<0.01	ACACB, AGT, COL4A1, Hrc, IGFBP5, POSTN, PRKAR2B, SLC2A1, TJP1
MYB	2.1	<0.001	BAX, CNN3, COL1A2, COL4A1, COPA, IGFBP5, ITGA2B, PCNA, PRTN3
CA4	2.0	<0.01	ANK1, CA1, CA2, SPTB
IL6	1.6	<0.001	AGT, BAX, CD14, COL3A1, DPP3, FGB, GAP43, HP, HPX, IGFBP5, ITLN1, LPA, MPO, ORM1, PCNA, PRPH, PRTN3, PSME1, S100A9, SAA1, SERPINA3, SERPINA7, STOM
CEBPB	1.5	<0.0005	ADH7, AGT, ALDH1A2, APOC3, CD14, COL1A2, CYB5A, GLIPR2, HBB, HP, HPX, INMT, ORM1, PCNA, PRKAR2B, PRTN3, SAA1
PPARA	1.4	<0.0001	ACADL, APOC3, BAX, BCKDK, C9, CAT, CYB5B, DYNLL1, ECI1, FGB, HPX, IGFBP5, PCNA, PKLR, PLTP, S100A8, S100A9, SAA1, TJP1
MTOR	1.4	<0.005	ACACB, ACADL, CNP, FGB, GAP43, GPHN, NDUFA6, OAS1, PCNA, PRKAR2B, RAP1GDS1, RPL32, SLC2A1, SQSTM1
NFE2L2	1.3	<0.005	ADH7, CALB1, CAT, COL3A1, DAD1, DYNLL1, HMBS, INMT, NARS, PHGDH, S100P, SAA1, SERPINA3, SLC2A1, SQSTM1
TGFB1	1.2	<0.01	ALOX15, BAX, CAMK2G, CAMP, CAT, CD14, CNN3, COL16A1, COL1A2, COL3A1, COL4A1, DAD1, DMTN, DSC2, DYNLL1, FGB, IGFBP5, ITGA5, MFAP2, MRC1, NRBP1, PCNA, PHGDH, POSTN, PPP2R5A, PRPS1, SERPINA3, SERPINE2, SLC1A2, SLC2A1, SRI, TJP1, TMOD3, TPM1, WFS1
SREBF1	1.1	<0.01	ACACB, APOC3, BAX, CD14, CYB5A, GLB1, GPNMB, PKLR, PLTP, SERPINA3
Right- Inhibited
AHR	−1.3	<0.005	BAX, CD14, COL16A1, COL1A2, COL3A1, COL4A1, HP, HSPH1, ITGA5, PKLR, SAA1, SERPINA7, SLC2A1
HNF4A	−1.5	<0.005	AGT, APEH, APOC3, BLVRB, BPGM, BTN2A1, C8G, CAT, CHCHD2, DAD1, DNAJC14, DSC2, EHD3, ERO1A, FGB, FN3KRP, GPR137, HLA-C, HPX, HSPH1, LPA, LYRM4, MAIP1, MPP1, NEK7, ORM1, PCNA, PKLR, PPFIBP1, PROZ, PRPS1, RAP2B, RPRD1B, S100A9, SAA1, SERPINA3, SLC43A1, STOM, SULT1A1, TMOD2, TTC38
IRF2	−1.7	<0.01	CD14, EIF2AK2, FGB, OAS1, PSME1, S100P, TPM1
NFE2	−2.2	<0.00005	CAT, HBB, HMBS, ITGA2B, SPTA1, TUBB1
GATA1	−2.4	<0.00001	AHSP, ANK1, BLVRB, CA1, CA2, CMA1, CNN3, CPA3, DMTN, EPB42, HBB, ITGA2B, PCNA, PLEK, SPTA1, SPTB, TUBB1

Proteins showing consistently increased or decreased expression after radiation

Proteins showing consistent responses throughout the natural history study may provide diagnostic utility in identifying populations exposed to radiation. The majority of proteins (213/220 in LV and 284/286 in RV) had consistent elevation or depression in expression after radiation (Figs,.1, 2). The majority of the RV differentially expressed proteins showed upregulation, while the majority of the differentially expressed LV proteins showed downregulation compared to baseline.

Biological processes that were enriched in all the proteins that were consistently activated or inhibited post-irradiated are shown in Table 3. The inclusion criteria were protein FC > 2, FDR corrected ANOVA p-value < 0.05, and significant biological process by Fisher’s exact test p < 0.05. In the left ventricle, a total of 15 biological processes were consistently enriched, including actin cytoskeleton reorganization and cholesterol biosynthetic process. In the right ventricle, a total of eight biological processes were consistently enriched, including extracellular fibril organization, negative regulation of endoplasmic reticulum calcium ion concentration, and ER to Golgi vesicle-mediated transport. Different biological processes were enriched in each ventricle.

Table 3. Biological processes altered by radiation.

Criteria for biological process enrichment was Fisher’s exact test p-value < 0.05, as compared to baseline values. Derived from LC-MS/MS proteomic analysis of NHP left and right ventricles after 12 Gy PBI/BM2.5.

Biological process	P-Value	Fold Enrichment	Genes
Left Ventricle
actin cytoskeleton reorganization	<0.005	13	FLNA, PARVA, PARB, ANXA1
cholesterol biosynthetic process	<0.005	31	G6PD, CFTR, APOA4
vesicle docking involved in exocytosis	<0.01	23	RAB8B, CFTR, SCFD1
positive regulation of release of sequestered calcium ion into cytosol	<0.01	20	A0N064, P61143, F7EW76
actomyosin structure organization	<0.01	20	F2, CNN1, CNN2
regulation of gene expression	<0.05	9	F2, DDX39B, PHGDH, HRG
peripheral nervous system axon regeneration	<0.05	96	APOD, NEFL
negative regulation of exocytosis	<0.05	96	SNCA, ANXA1
activation of cysteine-type endopeptidase activity involved in apoptotic process	<0.05	11	SNCA, BID, S110A8
negative regulation of fibrinolysis	<0.05	57	SERPINF2, HRG
positive regulation of triglyceride catabolic process	<0.05	57	ABHD5, APOA4
very-low-density lipoprotein particle remodeling	<0.05	48	APOA4, CETP
positive regulation of lipoprotein lipase activity	<0.05	48	ABHD5, APOA4
negative regulation of platelet-derived growth factor receptor signaling pathway	<0.05	41	SNCA, APOD
regulation of blood vessel size	<0.05	41	NPPA, GCLM
Right Ventricle
extracellular fibril organization	<0.005	53	MFAP5, MFAP4, COL3A1
negative regulation of endoplasmic reticulum calcium ion concentration	<0.005	44	BCAP31, RAP1GDS1, BAX
ER to Golgi vesicle-mediated transport	<0.005	8	BCAP31, COPA, SEC24C, COPE, NRBP1
cytosolic transport	<0.05	88	SRSF10
gamma-aminobutyric acid metabolic process	<0.05	88	PHGDH, ABAT
muscle contraction	<0.05	11	MYH2, TMOD3, TMOD2
long-chain fatty acid catabolic process	<0.05	59	LIPE, ACADL
retinoic acid metabolic process	<0.05	44	ALDH1A2, ADH7

Proteins related to retinoic acid activity

Multistage mass spectrometry analysis was used to quantify retinoids in left and right ventricles (Fig. 3, Supp. Fig. 1 and 2). In both ventricles, RA was reduced after radiation. In the LV, RA had an initial 44% decrease at d 4, and continued to decline until d 21–22 where it was reduced 60% (Fig. 3a). In the RV, RA had an initial 30% decrease at d 4, and continued to decline until d 21–22 where it was reduced 57% (Fig. 3b). Retinol and total retinyl esters (RE) levels were also quantified via HPLC-UV and neither exhibited consistent changes in response to radiation (Supp. Fig. 1 and 2).

Figure 3. Retinoic acid (RA) is reduced after radiation in LV (a) and RV (b).

RA quantification in LV (a) and RV (b) after 12 Gy PBI/BM2.5 was performed by LC with multiple reaction monitoring. Radiation days after radiation dose are noted. Data are mean ± SEM. *p-value < 0.05, **p-value < 0.01, and ***p-value < 0.001 using student’s t-test between groups as compared to Day 0 (baseline). Pmol g⁻¹: picomole/gram.

IPA upstream regulator analysis also identified proteins related to RA activity affected by radiation exposure (Figs. 4 and 5), where a Benjamini-Hochberg corrected p-value < 0.05 was the criteria for inclusion. In the LV, sixteen proteins associated with RA activity were found to be significantly perturbed by radiation exposure, with seven upregulated and nine downregulated (Fig. 4). In the RV, 23 proteins associated with RA activity were found to be significantly perturbed by radiation exposure, with eighteen upregulated and five downregulated (Fig. 5). ADA was upregulated in both LV and RV. ACACB and COL3A1 were downregulated in LV and upregulated in RV.

Figure 4. Proteins in LV regulated by RA and significantly perturbed by radiation exposure.

Benjamini-Hochberg adjusted Fisher’s exact test p-value < 0.05 was criteria for inclusion. Red color indicates significant upregulation and green color indicates downregulation, with intensity of color corresponding to magnitude of regulation.

Figure 5. Proteins in RV regulated by RA and significantly perturbed by radiation exposure.

Benjamini-Hochberg adjusted Fisher’s exact test p-value < 0.05 was criteria for inclusion. Red color indicates significant upregulation and green color indicates downregulation, with intensity of color corresponding to magnitude of regulation.

Metabolites most changed after radiation

The heart metabolome was assessed via targeted, quantitative LC-MS/MS with the Biocrates AbsoluteIDQ p180 kit to identify metabolites that were altered in NHP LV and RV after radiation exposure. Metabolites in the following classes were quantified: acylcarnitines (AC), amino acids (AA), biogenic amines (BA), glycerophospholipids (PC), spingolipids (SM), and hexose. A total of 229 metabolite and metabolite combinations were analyzed in both LV and RV (Supp. Tables 5 and 6). MetaboAnalyst was used for multivariate analysis to identify metabolites that show significantly altered abundances. Heatmaps (Figs. 6, 7) and PLS-DA plots (Supp. Fig. 3) were able to distinguish between the baseline/non-irradiated and irradiated NHPs in both the LV and RV.

Figure 6. Hierarchical clustering displays statistical metabolite differences between irradiated and non-irradiated NHP LV.

Heatmap displaying the top 25 metabolites based on t-test/ANOVA, Euclidean distancing, and Ward clustering, comparing metabolite profiles of LV at d0 (baseline) (red), d4 (green), d8–9 (blue), d11–12 (light blue), d15 (magenta), and d21–22 (yellow).

Figure 7. Hierarchical clustering displays statistical metabolite differences between irradiated and non-irradiated NHP RV.

Heatmap displaying the top 25 metabolites based on t-test/ANOVA, Euclidean distancing, and Ward clustering, comparing metabolite profiles of RV at d0 (baseline) (red), d4 (green), d8–9 (blue), d11–12 (light blue), d15 (magenta), and d21–22 (yellow).

Mulitivariate analysis with one-way ANOVA identified metabolites significantly altered expression as a result of radiation exposure, with a p-value < 0.05 and FDR < 5% as criteria for inclusion in Tables 4 and 5. In the LV, 18 metabolites and metabolite combinations were identified, including 1 acylcarnitine, 3 amino acids, 3 biogenic amines, and 6 glycerophospholipids (Table 4). Most of these 18 metabolites decreased in response to radiation, except Val and lysoPC a C16:0 which reached peak levels at d 4 and d 8–9, respectively. Additionally, C9, t4-OH-Pro and SDMA/Arg levels initially decreased and began to return to baseline levels at d 15. In the RV, 22 metabolites and metabolite combinations were identified, including 3 acylcarnitines, 4 amino acids, 3 biogenic amines, and 7 glycerophospholipids (Table 5). Most of these 22 metabolites decreased in response to radiation, except C5 which reached peak levels at d 8–9 and C5:1-DC and Orn/Arg which both reached peak levels at d 21–22, respectively. Also, PC aa C36:2 and Total PC aa both exhibited initial slight increases at d 4 but had an overall trend of decreased levels compared to baseline. Additionally, C9, His, and Ser levels initially decreased and began to return to baseline levels at d 15.

Table 4. Metabolites significantly changed by radiation in left ventricle.

NHP were exposed to 12 Gy PBI/BM2.5. Metabolites assayed from NHP left ventricles at each time point (n) as follows: Day 0 (baseline), n=2; Day 4, n=4; Days 8–9, n=11; Days 11–12, n=7; Day 15, n=5; Days 21–22, n=3. AC: Acylcarnitine. AA: Amino acids. BA: Biogenic amines. PC: Glycerophospholipids.

			Day 0		Day 4		Days 8–9		Days 11–12		Day 15		Days 21–22
Metabolite	Class	p-value	Mean	SD	Mean	SD	Mean	SD	Mean	SD	Mean	SD	Mean	SD
C9	AC	<0.005	0.027	0.006	0.020	0.005	0.021	0.005	0.017	0.003	0.028	0.006	0.022	0.007
Cit	AA	<1×10−9	59.40	4.9	21.15	10.2	5.16	3.0	5.70	3.8	25.81	19.5	21.00	11.1
His	AA	<5×10−5	65.85	20.0	93.08	17.8	39.72	11.8	33.76	15.3	57.68	14.3	65.30	30.4
Val	AA	<0.001	35.30	10.2	67.93	26.5	105.46	20.8	87.21	40.8	70.90	17.4	49.50	26.0
SDMA	BA	<0.001	0.31	0.03	0.33	0.09	0.16	0.1	0.11	0.06	0.21	0.09	0.23	0.09
Serotonin	BA	<5×10−5	0.059	0.02	0.063	0.03	0.029	0.01	0.019	0.004	0.019	0.003	0.025	0.01
t4-OH-Pro	BA	<0.005	23.10	6.1	16.98	2.2	13.18	5.4	11.27	5.2	21.26	6.1	21.87	6.1
lysoPC a C16:0	PC	<0.005	4.68	0.3	9.08	2.1	6.93	1.3	4.97	2.4	5.05	1.5	7.42	0.7
PC aa C34:3	PC	<5×10−5	4.80	0.4	4.25	1.3	2.75	0.8	2.06	0.9	2.39	0.3	3.05	0.6
PC aa C36:2	PC	<0.0005	131.50	9.2	141.75	27.1	96.60	21.0	73.37	35.9	68.60	22.1	84.23	22.6
PC aa C36:3	PC	<1×10−4	65.80	22.9	54.65	11.5	35.72	8.1	27.21	12.4	31.28	12.4	36.40	4.4
PC ae C36:2	PC	<1×10−4	8.68	2.0	6.60	1.0	4.77	1.5	3.58	1.6	5.02	1.2	4.47	0.5
PC ae C38:2	PC	<1×10−4	2.21	0.1	1.75	0.5	1.21	0.2	1.14	0.5	1.19	0.2	1.26	0.4
ADMA / Arg	BA / AA	<0.005	0.019	0.005	0.023	0.005	0.0075	0.008	0.0054	0.007	0.0076	0.007	0..011	0.005
Cit / Arg	AA	<1×10−8	2.72	0.2	0.77	0.5	0.20	0.06	0.34	0.2	1.26	1.1	0.91	0.4
Cit / Orn	AA	<5×10−9	15.70	3.1	5.16	4.5	1.63	1.0	1.57	0.9	4.87	4.3	3.03	1.6
SDMA / Arg	BA / AA	<5×10−7	0.014	0.001	0.012	0.001	0.0061	0.001	0.0061	0.002	0.0092	0.001	0.010	0.004
Total DMA / Arg	BA / AA	<1×10−4	0.032	0.006	0.034	0.007	0.013	0.009	0.011	0.008	0.017	0.007	0.021	0.009

Table 5. Metabolites significantly changed by radiation in right ventricle.

NHP were exposed to 12 Gy PBI/BM2.5. Metabolites assayed from NHP right ventricles at each time point (n) as follows: Day 0 (baseline), n=2; Day 4, n=4; Days 8–9, n=11; Days 11–12, n=7; Day 15, n=5; Days 21–22, n=3. AC: Acylcarnitine. AA: Amino acids. BA: Biogenic amines. PC: Glycerophospholipids.

			Day 0		Day 4		Days 8–9		Days 11–12		Day 15		Days 21–22
Metabolite	Class	p-value	Mean	SD	Mean	SD	Mean	SD	Mean	SD	Mean	SD	Mean	SD
C5	AC	<0.005	0.52	0.006	0.32	0.1	2.45	1.3	1.14	0.8	1.04	0.5	1.05	0.6
C5:1-DC	AC	<0.005	0.043	0.004	0.040	0.01	0.041	0.01	0.025	0.008	0.032	0.007	0.061	0.02
C9	AC	<0.005	0.032	0.004	0.025	0.003	0.022	0.007	0.027	0.006	0.019	0.005	0.032	0.008
Cit	AA	< 5 × 10−10	54.30	15.4	14.86	9.9	4.52	1.7	4.24	3.1	18.58	13.3	19.33	10.4
His	AA	<0.005	65.30	7.4	51.68	15.4	36.88	12.1	29.2	19.9	44.72	12.4	57.67	22.7
Pro	AA	<0.005	79.30	8.1	51.50	8.8	55.41	14.7	37.43	25.7	36.36	7.6	52.17	13.2
Ser	AA	<0.0001	154.00	15.6	116.00	29.7	81.04	19.5	79.33	60.7	89.08	24.6	174.33	45.3
ADMA	BA	<0.0001	0.52	0.2	0.26	0.2	0.10	0.2	0.054	0.09	0.14	0..1	0.092	0.2
SDMA	BA	<0.005	0.37	0.1	0.23	0.1	0.14	0.06	0.13	0.1	0.19	0.1	0.22	0.06
Serotonin	BA	< 1 × 10−9	0.12	0.02	0.068	0.03	0.029	0.02	0.021	0.005	0.019	0.005	0.017	0.003
lysoPC a C18:2	PC	<0.005	31.40	11.2	12.11	8.8	15.32	5.8	11.02	9.1	11.11	5.1	17.10	0.8
PC aa C34:2	PC	<0.005	209.50	88.4	195.00	54.6	136.64	29.5	93.65	59.8	94.50	30.1	153.33	28.4
PC aa C34:3	PC	< 5 × 10−5	5.83	0.2	3.79	1.2	2.98	0.7	1.96	1.3	1.97	0.4	3.37	0.2
PC aa C36:2	PC	<0.00005	104.40	`6.4	124.45	21.8	75.92	11.9	50.63	32.6	50.92	19.3	62.83	8.9
PC aa C36:3	PC	< 1 × 10−7	72.90	5.6	51.35	10.9	33.50	8.1	21.72	14.1	25.80	11.2	40.77	2.9
PC ae C36:2	PC	<0.00005	8.30	1.1	5.62	1.6	4.95	1.3	2.87	1.8	3.88	1.1	4.73	0.6
PC ae C38:2	PC	<0.0001	1.77	0.2	1.32	0.4	1.13	0.2	0.72	0.5	0.92	0.1	1.03	0.03
Cit / Arg	AA	< 1 ×10−6	1.66	0.01	0.58	0.2	0.21	0.06	0.29	0.2	0.96	0.8	0.81	0.4
Cit / Orn	AA	< 1 × 10−5	14.70	3.8	4.16	2.9	3.16	4.7	1.38	0.7	3.33	2.5	1.80	0.6
Orn / Arg	AA	<0.00005	0.12	0.03	0.16	0.05	0.14	0.09	0.21	0.04	0.28	0.04	0.44	0.09
Serotonin / Trp	BA / AA	<0.005	0.017	0.004	0.0093	0.004	0.0040	0.002	0.0084	0.009	0.0028	0.0008	0.0027	0.0006
Total PC aa	PC	<0.005	680.00	101.8	851.25	113.0	590.00	66.2	414.71	262.1	493.40	132.2	577.67	62.6

Metabolomic pathways altered by radiation

MetaboAnalyst Pathway Analysis was used to identify the pathways most impacted by radiation exposure in NHP LV and RV (Table 6). Metabolite data from d 8–9 and d 11–12 were grouped together to represent the irradiated NHP group and compared to baseline for pathway analysis in each ventricle. In the LV, 15 pathways were significantly affected by radiation, where 9 pathways were significantly affected by radiation in the RV. Of these pathways, five were altered in both LV and RV, including tryptophan metabolism, arginine biosynthesis, sphingolipid metabolism, histidine metabolism, and cysteine and methionine metabolism.

Table 6. Pathway analysis of metabolites affected by radiation in left and right ventricles.

Criteria for inclusion were p-value <0.05. Hits refers to the number of matched metabolites from the uploaded data, over the total number of compounds in the pathways. Raw p is the original p value calculated from the enrichment analysis. Holm p is the p value adjusted by Holm-Bonferroni method. The FDR p is the p value adjusted using False Discovery Rate. Impact is the pathway impact value calculated from pathway topology analysis. Results were normalized via Autoscaling (mean-centered and divided by the standard deviation of each variable). Pathway Enrichment Analysis was Global Test. Pathway Topology Analysis was Relative-betweenness Centrality. Pathway library was Homo sapiens (KEGG).

Pathway	Hits	Raw p	Holm adjust	FDR	Impact
Left ventricle
Tryptophan metabolism	3 / 14	<0.005	0.0074	0.0074	0.3422
Pantothenate and CoA biosynthesis	2 / 19	<0.005	0.0592	0.0263	0
Arginine biosynthesis	7 / 14	<0.005	0.0773	0.0263	0.4822
Valine, leucine and isoleucine degradation	3 / 40	<0.005	0.0970	0.0263	0
Valine, leucine and isoleucine biosynthesis	4 / 8	<0.005	0.1459	0.0325	0
Fatty acid degradation	1 / 39	<0.05	0.4380	0.0744	0
Sphingolipid metabolism	2 / 14	<0.05	0.5363	0.0744	0
Phenylalanine, tyrosine and tryptophan biosynthesis	2 / 4	<0.05	0.6877	0.0744	1
Phenylalanine metabolism	2 / 14	<0.05	0.6877	0.0744	0.3571
Tyrosine metabolism	3 / 14	<0.05	0.6877	0.0744	0.3803
Aminoacyl-tRNA biosynthesis	19 / 14	<0.05	0.6877	0.0744	0.1667
Ubiquinone and other terpenoid-quinone biosynthesis	1 / 9	<0.05	0.6877	0.0744	0
Histidine metabolism	5 / 14	<0.05	0.6877	0.0750	0.5000
beta-Alanine metabolism	5 / 14	<0.05	0.7615	0.0816	0.1119
Cysteine and methionine metabolism	2 / 33	<0.05	1.0000	0.1265	0.1263
Right ventricle
Tryptophan metabolism	3 / 14	<5×10−7	1.15×10−5	1.15×10−5	0.3422
Arginine biosynthesis	7 /14	<0.005	0.1004	0.0515	0.4822
Glycine, serine and threonine metabolism	4 /33	<0.05	0.5842	0.1070	0.5558
Sphingolipid metabolism	2 /21	<0.05	0.6021	0.1070	0
Histidine metabolism	5 /16	<0.05	0.6246	0.1070	0.5000
Cysteine and methionine metabolism	2 /33	<0.05	0.6380	0.1070	0.1263
beta-Alanine metabolism	5 /21	<0.05	0.6380	0.1070	0.1119
Arginine and proline metabolism	7 /38	<0.05	0.9409	0.1433	0.4273
Glyoxylate and dicarboxylate metabolism	4 /32	<0.05	1	0.20352	0.14815

DISCUSSION

In this study, the left (LV) and right ventricle (RV) NHP proteome and metabolome were quantitatively profiled following PBI/BM2.5 at 12 Gy over three weeks. Sampling included NHP tissue at timed euthanasia as planned at d 4, 8, 15, and 21 after radiation, and from NHP that were euthanized for cause according to pre-defined euthanasia criteria. Previous publications have described this PBI model and the analysis of resulting tissue injury (MacVittie et al. 2012, Jones et al. 2015, Shea-Donohue et al. 2016, Cohen et al. 2017, Prado et al. 2017, Carter et al. 2019, Cohen et al. 2019, MacVittie et al. 2019, MacVittie et al. 2019, Parker et al. 2019, Parker et al. 2019, Parker et al. 2019, Huang et al. 2020, Huang et al. 2020). We sought to further characterize the natural history of radiation-induced injury in NHP via determination of differential protein and metabolite expression in heart. Additionally, we considered the proteins and metabolites yielded by our data toward supporting efforts to identify exposed populations and may be useful during drug development under the FDA Animal Rule or could inform on medical management of radiation-induced injuries post-exposure. Through our multi-omic analysis, we identified the pathways most affected by radiation exposure in both the left and right ventricles. The proteomic and metabolomic data from irradiated NHP LV and RV reported here show some similarities in the canonical pathways dysregulated in irradiated NHP plasma and jejunum, as well as murine models of lung and ileum irradiation, as previously reported by us (Huang et al. 2019, Huang et al. 2019, Huang et al. 2020, Huang et al. 2020).

Liver X receptor/retinoid X receptor (LXR/RXR) signaling is important for regulating lipid and glucose metabolism, inflammation, and cholesterol homeostasis (Cannon et al. 2016). Activation of LXR/RXR is important for myocardial protection in various diseases, such as atherosclerosis, hypertension and diabetes, and dysregulated LXR/RXR activation is associated with heart disease (Cannon et al. 2016, DeLeon-Pennell et al. 2018, Lin et al. 2019, Sun et al. 2019). In this study, LXR/RXR was upregulated in RV, but downregulated in LV. Of the proteins involved in LXR/RXR signaling, LPA, SAA1, and S100A8 significantly altered in both LV and RV. LPA, a low-density lipoprotein variant, contributes to atherogenesis and atherosclerosis (Pan et al. 2004, Tsimikas et al. 2005). Greater LPA levels are directly associated with greater risk of coronary heart disease (Burgess et al. 2018). SAA1 is an apolipoprotein of the HDL complex, which increases in response to infection, trauma, inflammatory and immune diseases, and neoplasia (De Buck et al. 2016). SAA1 levels are elevated in patients who suffered myocardial infarction and is a prognostic factor for unstable angina and myocardial damage (Kisilevsky and Tam 2002). The increase in SAA1 levels are consistent with our previous studies of NHP irradiation in plasma and jejunum (Huang et al. 2020, Huang et al. 2020) and has been identified as a potential biomarker of radiation (Ossetrova et al. 2014, Ossetrova et al. 2014). S100A8 regulates inflammatory and immune responses through heterodimerizing with S100A9 (Averill et al. 2012). While the specific role of S100A8/A9 in the heart is unknown, this complex is a biomarker of cardiovascular disease (Averill et al. 2012, Li et al. 2019). APOD and FASN are both RA-related genes (Fig. 4) involved in LXR/RXR activation that were also shown to be downregulated in LV (Table 1). APOD is a widely produced glycoprotein with cardioprotective activity, likely through the prevention of cardiomyocyte injury (Muffat and Walker 2010, Tsukamoto et al. 2013). FASN catalyzes palmitate synthesis during fatty acid synthesis and is shown to be increased in heart failure (Abdalla et al. 2011, Razani et al. 2011, Karlstaedt et al. 2018).

Acute phase response signaling rapidly responds to a variety of inflammatory stimuli including infection, tissue injury, trauma, and immunological disorders, and was upregulated in RV following irradiation (Moshage 1997). Of all the proteins identified as members of the acute phase response signaling pathway, only RAP2B was not seen in irradiated NHP ileum and plasma (Huang et al. 2020, Huang et al. 2020). RAP2B stimulates vascular smooth muscle cell migration as part of arteriogenesis (Pöling et al. 2011). Activation of both the LXR/RXR and acute phase response pathways in the RV is consistent with previously reported data from both NHP and murine models of irradiation (Huang et al. 2019, Huang et al. 2019, Huang et al. 2020, Huang et al. 2020). Coagulation system pathway was activated in LV, as seen in previous reports of NHP plasma irradiation (Huang et al. 2020).

RA is the active metabolite of Vitamin A and a master regulator of gene expression through ligand-activated control of transcription mediated through the nuclear receptors retinoic acid receptor (RAR) and retinoid X receptor (RXR) (Germain et al. 2006, Germain et al. 2006). In the heart, RA kevels are essential for proper development and function (Pan and Baker 2007, Xavier-Neto et al. 2015, Wang et al. 2018, Wang et al. 2018). Dysregulated RA levels in the heart have been associated with heart disease (Minicucci et al. 2010, Liu et al. 2016, Park et al. 2018, Park et al. 2019). In this study, RA levels were decreased by 60% and 57% in the left and right ventricles, respectively, three weeks after radiation exposure (Fig. 3). This decrease in heart RA levels is similar to the decrease in RA that was observed in Rbp1 global knockout mice, an animal model of dysregulated RA biosynthesis, compared to wild-type mice (Pierzchalski et al. 2013, Pierzchalski et al. 2014, Zalesak-Kravec 2021).

RA levels are tightly regulated to maintain proper cell function, so the decreased levels of RA following radiation exposure represent significant dysregulation in RA homeostasis and, therefore, disrupted essential nutrient homeostasis (Napoli 2012, Napoli 2020). Retinoic acid metabolic process showed 44-fold enrichment in RV (Table 3). RA is biosynthesized by a two-step enzymatic conversion of vitamin A (retinol). Retinol dehydrogenases catalyze the first rate-limiting oxidation of diet-derived retinol to retinal (Napoli 2012). Of the retinol dehydrogenases, only ADH7 was significantly affected by irradiation and was only seen in irradiated RV samples. There were no significant changes in retinol levels after irradiation in either LV or RV (Supp. Fig. 1). Retinal reductases counterbalance retinol dehydrogenases by catalyzing the conversion of retinal to retinol and none were found to be significantly altered after irradiation. RDH14, a retinal reductase previously known as PAN2, was significantly upregulated in the RV (Belyaeva and Kedishvili 2002). The ALDH1A family of retinal dehydrogenases catalyzes the second step of RA biosynthesis, the irreversible oxidation of retinal to RA (Napoli 2012). ALDH1A1 was significantly downregulated in the LV, while ALDH1A2 was significantly upregulated in both the LV and RV. Retinol-binding proteins protect retinoids from non-specific oxidation and facilitate their transport and metabolism (Napoli 2012, Napoli 2017). RBP1, the main intracellular chaperone for retinol and retinol, showed decreased expression in the LV and increased expression in the RV. RBP4, which facilitates vitamin A uptake, had significantly decreased expression only in LV. RA can also be derived by the cleavage of beta-carotene, which is cleaved symmetrically by BCO1 to two molecules of retinal (Harrison and Kopec 2020). However, BCO1 is not expressed in adult hearts, unlike BCO2, which mediates the asymmetric cleavage of beta-carotene, resulting in apo-carotenals (Lee et al. 2014). BCO2 was significantly altered by radiation, downregulated in the LV and upregulated in the RV. Asymmetric cleavage by BCO2 protects mitochondria from the toxic effects of oxidative stress caused by excess carotenoids, which is seen in various diseases including metabolic disorders and cancer (Ip et al. 2014, Guo et al. 2017, Tan et al. 2017, Wu et al. 2017, Harrison and Quadro 2018).

IPA upstream regulator analysis identified 16 proteins in the LV and 23 proteins in the RV that were associated with retinoic acid activity and were significantly perturbed after irradiation (Fig. 4 and 5, respectively). Adenosine deaminase (ADA) was upregulated in both LV and RV. Adenosine serves as a cardioprotective agent through coronary vasodilation and inhibition of platelet and neutrophil activity, and ADA activity may be useful as a biomarker of chronic heart failure severity (Kowalczyk et al. 2008, Khodadadi et al. 2014). ACACB and COL3A1 were downregulated in LV and upregulated in RV. ACACB, also known as ACC2, is involved in fatty acid oxidation through production of malonyl CoA and a cardiac-specific deletion of ACAB was shown to prevent metabolic remodeling that occurs during the development of pressure-overload hypertrophy (Kolwicz et al. 2012). In the injured heart, such as after a myocardial infarction, ventricular remodeling occurs. As remodeling continues in the LV, collagen type III is replaced with collagen type I, which leads to increased stiffness and LV systolic dysfunction (Chello et al. 1996, Uchinaka et al. 2018).

Myocardial remodeling pathways were disrupted in both LV and RV. LV showed 13-fold and 20-fold enrichment of actin cytoskeleton reorganization and actomyosin structure organization, respectively (Table 3) and significantly downregulation of actin cytoskeleton signaling (Table 1). RV showed 53-fold enrichment of extracellular fibril organization (Table 3). Myocardial remodeling and fibrosis are common side effects of radiation exposure, as seen in multiple murine and NHP radiation models (Baker et al. 2009, Yusuf et al. 2011, Medhora et al. 2015, DeBo et al. 2016, Jacobs et al. 2019, Parker et al. 2019, Zhang et al. 2020). Myocardial remodeling occurs after injury and proper cardiac remodeling depends on RA signaling affecting the RhoA and renin-angiotensin systems (RAS) (Wang et al. 2002, Choudhary et al. 2008, Wang et al. 2018, Wang et al. 2020). Altered RA levels can impact heart remodeling after injury in several ways (Minicucci et al. 2010, Wang et al. 2020). For example, activation of epicardial gene expression and RA signaling in the heart are required for heart repair in zebrafish (Kikuchi et al. 2011). However, if left unchecked, cardiac remodeling can lead to hypertrophy and subsequent heart failure (Choudhary et al. 2008). Multiple proteins from the RhoA and RAS systems were significantly altered in response to radiation exposure. In LV, COL3A1, COL14A1, MYL4 and MYL7 were downregulated (Fig. 1). In RV, COL1A1, COL3A1, COL 4A1, COL16A1, MYH2, MYH14, and MYH15 were all upregulated (Fig. 2).

The heart requires large amounts of energy to function properly, and obtains most of its ATP from fatty acid β-oxidation (Lopaschuk et al. 2010). Excess fatty acid availability causes metabolic stress and leads to lipotoxic cardiomyopathy (Suzuki et al. 2009). Fatty acid metabolism was affected by radiation in both LV and RV. FASN was downregulated (Table 1) and the fatty acid degradation pathway was significantly impacted (Table 6) post-irradiation in LV. In the RV, the long-chain fatty-acid catabolic process had 59-fold enrichment post-irradiation, attributed to upregulation of LIPE and ACADL (Table 3). LIPE, hormone-sensitive lipase, contributes to cardiac lipid metabolism and disrupted LIPE levels correlate with cardiac lipotoxicity (Suzuki et al. 2009). ACADL is a member of the acyl-CoA dehydrogenase family that catalyzes the first step of mitochondrial fatty acid β-oxidation (Andresen et al. 1999). Studies in mice and humans with deficient ACADL show increased cardiomyopathy (Andresen et al. 1999, Cox et al. 2009). The transport of acylcarnitine esters into the mitochondria for subsequent fatty acid β-oxidation was also disrupted as evident by decreased levels of various acylcarnitines; C9 in LV and C5, C5:1-DC, and C9 in RV (Tables 4 and 5). No proteins related to carnitine transport or biosynthesis were significantly affected due to radiation exposure in either LV or RV.

Of the amino acids quantified, citrulline and metabolite combinations containing citrulline (Cit / Arg and Cit / Orn) were some of the most significantly impacted by radiation exposure in both LV and RV. Citrulline is the natural precursor of arginine, which is necessary for the production of nitric oxide (NO) via nitric oxide synthase (Romero et al. 2006). Improved NO production improves cardiovascular function in various cardiovascular diseases, including hypertension and heart failure (Romero et al. 2006). The downregulation of citrulline at all timepoints was also seen in gastrointestinal models of murine and NHP irradiation (Jones et al. 2015, Kumar et al. 2020).

Pathway analysis performed by MetaboAnalyst software identified 15 metabolite pathways in the LV and nine metabolite pathways in the RV that were significantly altered post-irradiation (Table 6). Most identified pathways involved amino acid metabolism, biosynthesis, and degradation; with five metabolite pathways common among both LV and RV: tryptophan metabolism, arginine biosynthesis, sphingolipid metabolism, histidine metabolism, and cysteine and methionine metabolism (Table 6). Tryptophan metabolism was the most significantly affected pathway in both LV and RV, based on p-value. Tryptophan, along with other aromatic amino acids, is not metabolized by a healthy heart and provides a marker of protein flux (Drake et al. 2012). Histidine was one of the few amino acids that was significantly downregulated in both LV and RV, and histidine metabolism was affected post-irradiation in both LV and RV. Histidine, along with arginine, alanine, aspartate, glutamate, and glutamine, are utilized by the oxygen-limited heart for energy (Drake et al. 2012). Studies have shown that branched-chain amino acid metabolism is impaired in heart disease (Huang et al. 2011, Yang et al. 2020). Of the BCAA, only valine was significantly decreased in the LV and pathways regulating valine, leucine, and isoleucine degradation and biosynthesis were only dysregulated in the LV.

While no individual SM were identified as significantly changed (Tables 4 and 5), the sphingolipid metabolism pathway was significantly impacted in response to irradiation (Table 6). Dysregulated sphingolipid levels are associated with various cardiovascular diseases, including coronary artery disease and atherosclerosis (Chiu et al. 2001, Knapp et al. 2013, Kasumov et al. 2015, Kovilakath and Cowart 2020, Kovilakath et al. 2020). Some sphingolipids are also being studied as potential biomarkers of cardiovascular disease (Bellis et al. 2014, Zordoky et al. 2015, Tabassum et al. 2019).

This study has highlighted a clear difference in early radiation injury between the LV and RV. The majority of significantly altered proteins in LV were downregulated compared to baseline (Table 1), whereas the majority of significantly altered proteins in RV were upregulated compared to baseline (Table 2), including many of the top affected proteins, including LPA, SAA1, S100A8, BCO2, COL3A1, and COL4A1. The upregulation of significantly altered proteins seen in RV is consistent with previous reports of NHP irradiation in jejunum and plasma from our group (Huang et al. 2020, Huang et al. 2020). These differences indicate a difference in the pathology of radiation-induced injury in the heart ventricles. Differences in the LV and RV are also seen in a variety of cardiovascular diseases. LV hypertrophy (LVH) is the heart’s adaptive response to hypertension consisting of myocardial remodeling and fibrosis (Gradman and Alfayoumi 2006). LVH presents clinically as ischemia, ventricular arrhythmias, and diastolic dysfunction, and leads to increased risk of coronary heart disease, sudden death, heart failure, and stroke (Gradman and Alfayoumi 2006). Previous studies have shown that LV fibrosis and LV end-diastolic pressure are impacted by radiation exposure (Ghobadi et al. 2012, Medhora et al. 2015, van der Veen et al. 2015, DeBo et al. 2016). Interestingly, transient right ventricular hypertrophy has been seen in rat models of thoracic radiation (Ghosh et al. 2009, Medhora et al. 2015). Further studies exploring changes in ventricle weight over a variety of time-points are necessary to determine the immediate and delayed effects of radiation-induced heart injury.

CONCLUSION

In summary, we used a mass spectrometry-based proteomics and metabolomics approach to interrogate the left and right ventricle in order to characterize the natural history of the NHP PBI/BM2.5 model. Radiation exposure is linked with delayed cardiac disease, and these data are useful in identifying potential initiating events of radiation-induced heart injury. Proteins and metabolites related to inflammation, energy metabolism, and myocardial remodeling were dysregulated. The most significantly altered pathways included LXR/RXR activation, acute phase response, RA metabolism and signaling, cytoskeletal and extracellular reorganization, fatty acid β-oxidation, and amino acid metabolism. In response to radiation exposure, the majority of significantly altered LV proteins were downregulated, but the majority of significantly altered RV proteins were upregulated compared to baseline levels. This observation indicates clear differences between the pathology of radiation-induced injury in the LV and RV. This multi-omic study characterizes the natural history and molecular mechanisms of radiation-induced heart injury in NHP exposed to PBI with minimal bone marrow sparing.