We have designed (Fig. 1) and synthesized (Supplementary Fig. S1) imidazole-substituted fatty acids and their esters - imidazole substituted oleic acid (IOA), stearic acid (ISA) (Fig. 1a, b), imidazole substituted methyl ester of oleic acid (IEOA, Fig. 1c) and imidazole substituted dodecanoic acid (IDA) (Fig. 1d) - assuming that these compounds will interact with cyt c by providing the imidazole group as a sixth ligand to the heme. A similar strategy was used to inhibit the terminal oxidation of fatty acids by CYP4A1¹⁶. Molecular modeling and docking studies of partially unfolded cyt c showed that 18-carbon long IOA and ISA (Fig. 2a, b, c) as well as IEOA (Fig. 2d) with the imidazole moiety 7-carbon atoms away from the terminal methyl group indeed positioned the heterocycle close to heme such that the nitrogen atom of imidazole was within 2.4 Å (IOA), 2.5 Å (IEOA) and 2.7 Å (ISA) from the heme-iron. A hydrophobic channel formed by the displacement of the Met₈₀ containing loop in the partially unfolded cyt c structure stabilized the interaction (Fig. 2b, colored in yellow). Notably, IDA, with the imidazole-moiety attached to the terminal CH₂-group, this specific positioning was not achieved and the imidazole nitrogen was 6.9 Å away from the heme-iron (Fig. 2e).

We next evaluated the effect of ISA, IOA and IEOA on the peroxidase activity of cyt c complexes with tetra-oleoyl-cardiolipin (TOCL) towards H₂O₂-driven oxidation of two prototypical phenolic substrates, Amplex Red (Fig. 3a) and etoposide (Fig. 3b). We found that ISA, IOA and IEOA acted as potent inhibitors of the peroxidase activity of cyt c/TOCL complexes with both substrates (Fig. 3). Computer modeling showed that IOA (Fig. 2a), IEOA (Fig. 2c) and ISA (Fig. 2d) bind to cyt c in a similar fashion. The truncated derivative IDA, did not exert any inhibitory effect (Fig. 3a), in line with our computer modeling data. Because catalytic reactive intermediates of cyt c/CL peroxidase complexes – protein-immobilized (tyrosyl) radicals (Tyr•) – can be detected by EPR spectroscopy¹⁷, we studied the effect of IEOA and IOA on H₂O₂-dependent formation of radicals (Fig. 3c). Both IOA and IEOA (but not IDA) effectively quenched generation of Tyr• radicals. Finally, we assessed the ability of ISA and IEOA to inhibit peroxidase activity using isolated mouse liver mitochondria. To avoid decomposition of H₂O₂ by catalase, we used tert-butyl hydroperoxide (tBOOH) as a source of oxidizing equivalents. Both ISA and IEOA suppressed peroxidase activity in a concentration-dependent manner (Supplementary Fig. S2).

To address whether ISA and IEOA act as strong ligands of heme-iron in cyt c/TOCL complexes, we used low temperature EPR spectroscopy¹⁸. In the reduced form, cyt c/TOCL binds NO• and produces heme-nitrosylated complexes with characteristic EPR spectra⁹. Indeed, at 77 K, typical spectra of penta-, and hexa-coordinated cyt c were detectable upon incubation of cyt c/TOCL complexes in the presence of a source of nitroxyl (HNO), Angeli’s salt (Fig. 3d). ISA and IEOA (Fig. 3d) caused a concentration-dependent decrease of the EPR signal. To provide evidence for coordination changes of the heme-iron in cyt c we performed liquid-He EPR measurements. The X-band EPR spectrum of native ferri-cyt c at pH 7.4, recorded at 20 K, exhibited anisotropic, low spin signals, with g_z=3.09, g_y=2.24 and g_x~1 (usually unobserved) (Fig. 3e) indicative of His/Met axial coordination at the heme^19,20. Upon the addition of IOA, there was no change in the EPR spectrum indicating retention of the native His/Met axial coordination (Fig. 3e). However, when IOA was added to the cyt c/TOCL complex, the EPR spectrum revealed the presence of another low-spin species (g_z =2.97, g_y=2.27, g_x~1.5, Fig. 3f) with g-values entirely consistent with His/imidazole coordination^20,21. This signal was broadened on the low-field side, suggesting a combination of the native ferri-cyt c and a His/imidazole form. Spectral simulations (not shown) confirm an approximately 50:50 mixture of the native structure and the form in which Met₈₀ has been replaced by the imidazole moiety of IOA. These were the only signals observed; in particular there were none at g ~ 6, indicating the absence of any penta- or hexacoordinate high-spin species. Similarly, ISA was able to change heme-iron coordination in cyt c whereby Met₈₀ was substituted by the imidazole moiety (Fig. 3e). These results confirm experimentally that the imidazole moiety of imidazole fatty acids can indeed serve as a coordinating ligand for the heme substituting for Met₈₀ ligation.

We then explored the ability of imidazole substituted fatty acids to inhibit apoptosis in cells. To target imidazole substituted fatty acids into mitochondria, we esterified them with TPP salt (Fig. 4 a, b), an organic cationic alcohol with delocalized electron density – known to be effectively “electrophoresed” due to the negative potential on the inside of the organelle’s membrane²². By using computer modeling we confirmed that conjugation of ISA and IOA with TPP did not significantly affect their positioning within the immediate proximity of the cyt c’s heme-iron, i.e. the nitrogen atom of the imidazole of both TPP-IOA and TPP-ISA was within 2.9 Å from the heme-iron (Fig. 4 c, d). Moreover, TPP-ISA and TPP-IOA were as effective as non-conjugated ISA and IOA in inhibiting peroxidase activity of cyt c/TOCL complexes (Fig. 3).

We further estimated the amounts of accumulated TPP-IOA in mitochondria of mouse embryonic cells using high performance liquid chromatography (HPLC) and electrospray ionization mass spectrometry (ESI-MS) (Supplementary Fig. S3). We found that most of TPP-IOA was present in mitochondria (Supplementary Fig. S3). Assuming that the volume of mitochondria constitutes ~15–25% of the total volume of a cell, the mitochondrial enrichment factor becomes even greater such that the concentration of TPP-IOA in mitochondria may be as high as ~5 mM. It is likely that endogenous esterases can hydrolyze the ester-bond and release IOA and TPP. To test this, we performed assessments of esterase activity of mitochondria and the cytosolic fractions isolated from mouse embryonic cells based on HPLC measurements of TPP-IOA (Supplementary Fig. S3). We found that hydrolysis of TPP-IOA takes place in both mitochondria and the cytosol. The hydrolysis rate in mitochondria was comparable to that in the cytosolic fraction (Supplementary Fig. S3). Thus, both TPP-IOA and its de-esterifed form, IOA, could be present in mitochondria of mouse embryonic cells. Because TPP-conjugated imidazole substituted fatty acids effectively partition into mitochondria, we examined whether they affected bioenergetic functions, particularly ATP production. Neither TPP-IOA nor TPP-ISA had any effect on ATP levels in mouse embryonic cells. Normally, mitochondria are the major source of superoxide radicals in mouse embryonic cells^23,24. Assessments of intracellular superoxide production using dihydroethidium (DHE) showed that neither TPP-IOA nor TPP-ISA had any effect on superoxide production in cells with or without radiation treatment (Supplementary Fig. S4).

For the anti-apoptotic action, it is essential that proposed heme-iron ligation is effective in inhibiting peroxidation of polyunsaturated species of CL. Therefore, we used an oxidizable tetra-linoleoyl-cardiolipin (TLCL), and conducted experiments on its cyt c/H₂O₂-induced oxidation (Fig. 5). In the presence of H₂O₂, we detected accumulation of characteristic TLCL peroxidation products with multiple oxygenated linoleic acid residues detectable by ESI-MS (Fig. 5a). We found that both TPP-IOA and TPP-ISA inhibited cyt c/H₂O₂ induced oxidation of TLCL in a concentration dependent manner. No difference in inhibition of TLCL oxidation between TPP-IOA and TPP-ISA was detected. The oxidation of TLCL was completely blocked at a ratio of cyt c to TPP-IOA or TPP-ISA of 1:20 (Fig. 5b).

We then employed a model of intrinsic apoptosis induced in mouse embryonic cells by γ-irradiation and assessed several biomarkers of apoptosis. We found that TPP-IOA and TPP-ISA had similar radiation mitigating effects on mouse embryonic cells as evidenced by PS externalization (Fig. 6a), caspase 3/7 activation (Fig. 6b) and cyt c release (Fig. 6c). In contrast, IEOA or ISA - devoid of a mitochondria-targeting TPP-moiety - exerted no protection against apoptosis in mouse embryonic cells (Fig. 6d). To verify that the protective effects were realized during mitochondrial stages of apoptosis, TPP-ISA were added to the in vitro “caspase activation system” containing S100 fraction from mouse embryonic cells²⁵. Caspase activation caused by exogenously added cyt c was completely insensitive to TPP-ISA and TPP-IOA (Fig. 6e) thus confirming that anti-apoptotic effects of TPP-IOA and TPP-ISA were realized in mitochondria. Furthermore, the anti-apoptotic activity was neither cell- nor stimulus- specific as TPP-IOA effectively inhibited apoptosis induced by a mitochondrial complex I inhibitor, rotenone, in cultured mouse lung endothelial cells (Fig. 6f).

Mitochondria are believed to be involved in orchestration of different cell death pathways^26,27. We found that necrotic cells (PI-positive, Annexin V-positive) represented 12.9% while apoptotic cells (PI-negative, Annexin V-positive) were accountable for 28.9% of total PS positive cells detectable after irradiation. Treatment with TPP-IOA caused a 2-fold decrease in the number of apoptotic cells (to 14.8%) and 1.6-fold reduction in the number of necrotic cells (to 8.2%). Similarly, TPP-ISA protected via both anti-apoptotic and anti-necrotic mechanisms (to 13.9% of apoptotic and 7.8% necrotic cells, respectively). Overall, these results suggest that TPP-IOA and TPP-ISA afforded the radiomitigation in mouse embryonic cells acting through both anti-apoptotic and anti-necrotic pathways. In addition to apoptosis, and necrosis, mitotic cell death can be also triggered in irradiated cells. Using clonogenic assay, that includes the mitotic cell death component, we demonstrated that post-irradiation treatment of mouse embryonic cells with TPP-ISA or TPP-IOA resulted in a significant protection. Using a single-hit multi-target model, we estimated that TPP-ISA or TPP-IOA increased D₀ - the dose needed to reduce cell surviving fraction to 37% (1/e) - to 1.67 ± 0.06 and 1.71 ± 0.05, respectively, compared to 1.33 ± 0.08 in untreated cells (Fig. 6g).

Given the ability of TPP-IOA and TPP-ISA to act as radiomitigators in vitro, we assessed their potential to act as radioprotectors/radiomitigators in vivo. C57BL/6NTac female mice were exposed to 9.25 Gy total body irradiation at the dose rate of 80 cGy/min using the cesium irradiator. Three independent experiments (the total number of mice in each group was 31 to 35) yielded similar results: irradiation resulted in death of animals within 13–15 days (with survival of only 20% of animals by day 30). TPP-IOA or TPP-ISA (i.p. injection, 5 mg/kg body weight, 10 min after irradiation) showed a strong radiomitigative effect for both compounds (Fig. 7a). There was no statistically significant difference in radiomitigative potency of TPP-IOA and TPP-ISA (p=0.6389, a two-sided log-rank test). We also used a clinical linear accelerator to deliver the radiation dose. The mice (three groups with 22–23 animals in each) were irradiated to 9.25 Gy at the dose rate of 100 cGy/min using a Varian TrueBeam linear accelerator (Varian Medical Systems, Palo Alto, CA) and injected i.p. with 5 mg/kg body weight of either TPP-ISA or TPP-IOA 10 min after irradiation. Similarly to the results with cesium irradiator, the survival curves over 52 days were statistically different for TPP-IOA (5 mg/kg body weight) and TPP-ISA (5 mg/kg body weight) vs irradiated controls (Table 1); however, there was no statistically significant difference between TPP-IOA and TPP-ISA (p = 0.4567, a two-sided log-rank test). Thus, after multiple experiments, using two different irradiators, we demonstrated a high potency of both TPP-IOA and TPP-ISA in mitigating radiation damage without significant difference in radiomitigative activity between them. Therefore, all subsequent in vivo experiments were conducted with TPP-IOA and linear accelerator as the radiation source. We chose a dose of 5 mg/kg body weight of TPP-IOA because a lower dose (2.5 mg/kg body weight) of drug was not effective in mitigating the mice against irradiation. When mice (10 per group) were irradiated and administered 2.5 mg/kg body weight (10 min after irradiation), there was a trend towards a greater survival that, however, did not reach the level of significance (p=0.0525, a two-sided log-rank test) (Table 1). We further tested whether TPP-IOA was protective if given at later time points (than 10 min) after irradiation (10 mice per group). Administration of TPP-IOA at 5 hrs or 24 hrs after irradiation resulted in a significant increase in survival (Fig. 7b, Table 1). TPP-IOA was also protective if given 10 min or 1 hr before irradiation (Fig. 7b; Table 1).

Given the high radioprotective/radiomitigative activity of TPP-IOA, we determined the extent to which it would be absorbed to reach radiosensitive tissues. This required the development of new LC-MS/MS protocols to quantitate TPP-IOA in tissue samples. These involved the establishment of a selective reaction monitoring (SRM) protocol for TPP-IOA that provides high selectivity and sensitivity. Our direct assessments clearly demonstrated the presence of TPP-IOA in plasma (Supplementary Fig. S5) as well as in the two most important radiosensitive tissues, bone marrow and small intestine. TPP-IOA levels were highest in plasma (54.0 ng/ml of plasma), followed by small intestine (1.5 ng/g of tissue) and bone marrow (0.2 ng/g of tissue), at the 10 min time point after i.p. injection. The levels measured in plasma are consistent with those found for decyl-TPP, a closely related compound, as seen in the study by Porteous, et al.²⁸. The authors showed that after i.v. injection, the TPP-compounds are distributed rapidly to various tissues and less than 1% remains in plasma after 15 min. In addition, we assessed the hydrolysis of TPP-IOA in tissues (plasma, bone marrow and small intestine) yielding TPP and non-esterified IOA utilizing two different LC-MS/MS approaches: a selected reaction monitoring (SRM) method to determine the levels of IOA, a likely hydrolysis products of TPP-IOA, in tissue samples, and multiple reaction monitoring (MRM) offering extremely high selectivity and sensitivity (Supplementary Fig. S6). Using these approaches, we determined that bone marrow had the least amount of hydrolysis, exhibiting a level of 0.6 ng/g tissue/min of the TPP-propyl hydrolysis product. This was followed by a level of 4.1 ng/g tissue/min and 19.0 ng/g tissue/min of the TPP hydrolysis product in plasma and small intestine, respectively. Simple calculations show that the total amounts of non-hydrolyzed TPP-IOA plus hydrolyzed (IOA + TPP) will be significantly (approximately two orders of magnitude) higher than those of the non-hydrolyzed compound measured in the small intestine and bone marrow. The importance of this is underscored by our data that both TPP-IOA and IOA are effective inhibitors of peroxidase activity of cyt c/CL complexes (Fig. 3) and CL peroxidation (Fig. 5).

C57BL/6NTac female mice were anesthetized with Nembutal (1 mg/20 gm mouse), irradiated to a total body dose of 9.25 Gy using either a Shepherd Mark 1 Model 68 cesium irradiator at a dose rate of 80 cGy/min (31–35 mice per group) or a Varian TrueBeam Linear Accelerator (Varian Medical Systems, Palo Alto, CA) at 100 monitor units or approximately 100 cGy/min using 6 MV photons with a 40 cm × 40 cm field at 100 SSD (10–23 mice per group). Mice were injected intraperitoneally (i.p.) with 5 mg/kg body weight of TPP-IOA in a 100 μl volume of water containing 25% ethanol at 1 hr before irradiation, 10 min before irradiation, 10 min after irradiation, 5 hrs after irradiation and 24 hrs after irradiation. Other groups were injected i.p. with 2.5 mg/kg body weight of TPP-IOA or 5 mg/kg body weight of TPP-ISA 10 min after irradiation. The mice were followed for the development of hematopoietic syndrome (at which time they were sacrificed). The health of the non-irradiated mice was unexceptional and no adverse side effects were noticed over the period of study (52 days) after i.p. injection of either TPP-IOA or TPP-ISA (5 mg/kg body weight). The log-rank test was used for three analyses: the comparison of overall survival which is defined as the time from the date of radiation to the date of death for all mice under study; the comparison of short term survival over the first 20 days, i.e., the overall survival count at 20 days; and the comparison of conditional survival in mice surviving 20 or more days, i.e., the time from the date of radiation to the date of death for all mice who survived 20 days or longer after radiation. All these comparisons were made between each of the treated groups and the radiation-only control group. All procedures were pre-approved and performed according to the protocols established by the Institutional Animal Care and Use Committee of the University of Pittsburgh.

Mouse embryonic cells (courtesy of Dr. X. Wang, University of Texas, Dallas) were cultured in DMEM supplemented with 15% FBS, 25 mM Hepes, 50 mg/liter uridine, 110 mg/liter pyruvate, 2 mM glutamine, 1 × nonessential amino acids, 50 μM β-mercaptoethanol, 0.5 × 10⁶ U/liter mouse leukemia inhibitory factor, 100 U/liter penicillin, and 100 mg/liter streptomycin in a humidified atmosphere of 5% CO₂/95% air at 37°C. Mouse lung endothelial cells were obtained as previously described³⁹. Briefly, lungs were flushed with HBSS containing 10 U/mL heparin, then homogenized and digested in type I collagenase using a gentle MACsTM dissociator (Miltenyi, CA). Pulmonary endothelial cells were isolated by magnetic beads coated with antibody (rat anti-mouse) to PECAM-1 (BD Pharmingen, San Diego, CA), and seeded for subculture (passage 1). At approximately passage 2, cells were incubated with fluorescently-labeled diacetylated LDL (diI-LDL) followed by FACS sorting for further purification. The enriched PECAM and diI-LDL population was sub-cultured on a collagen/gelatin matrix in 2% O₂, 5% CO₂, 93% nitrogen in a Coy Hypoxic Glove Box/Chamber in Opti-MEM (Gibco), 10% FBS, 2 mM glutamine, 0.2% retinal derived growth factor (Vec Technologies, Rensselaer, NY), 10U/mL heparin, 0.1 mM non-essential amino acid supplement (Gibco) and 55 μM β-mercaptoethanol (Counted as passage 3). Cells at passage 4 to 6 were applied to do the experiments.

Phosphatidylserine externalization was determined by Annexin V–FITC apoptosis detection kit (Biovision, Mountain View, CA) according to the manufacturer’s instruction.

Caspase–3/7 activity was measured using a luminescence Caspase–Glo^™ 3/7 assay kit (Promega, Madison, WI) according to the manufacturer’s instruction.

Cells were harvested after 48 hrs post-irradiation incubation and resuspended in lysis buffer containing 0.05% digitonin for 4 min on ice. Supernatants were collected after centrifugation for 10 min at 10,000×g. Equal amounts of protein were subject to 15% SDS-PAGE, transferred onto a nitrocellulose membrane, and probed with antibodies against cyt c (clone 7H8.2C12, BD Pharmingen, San Diego, CA) or actin (Novus, Littleton, CO) (loading control) followed by horseradish peroxidase-coupled detection. The protein band profile was analyzed by densitometry using Labworks Image Acquisition and Analysis Software (UVP, Upland, CA).

Conditions: Native cyt c (300 μM), IEOA/cyt c (2:1), TOCL/IEOA/cyt c (20:2:1), 20 mM HEPES buffer, pH 7.4. Samples (200 μL) were placed in 3 mm o.d. suprasil quartz EPR tubes and frozen in liquid nitrogen for subsequent spectroscopic measurements. X-band (9 GHz) EPR spectra were recorded on a Bruker ESP 300 spectrometer equipped with an Oxford Instruments ESR 910 flow cryostat for ultra-low-temperature measurements. Spectra were recorded at 9.8 G modulation amplitude and 200 μW microwave power. The microwave frequency was calibrated by a frequency counter and the magnetic field was calibrated with a gaussmeter. The temperature was calibrated with carbon-glass resistors (CGR-1-1000) from LakeShore. This instrument and the software (SpinCount) to analyze the EPR spectra were graciously provided by Professor Mike Hendrich, Carnegie Mellon University.

The structures of the ligands ISA, IOA, IEOA, IDA, TPP-ISA and TPP-IOA were docked to the partially unfolded structure of horse heart cyt c using the Lamarckian genetic algorithm provided by the Autodock 4.0 software^40,41. The partially unfolded structure of cyt c, where the Met₈₀ containing loop is displaced was obtained using the molecular dynamics simulation approach as described below. The X-ray crystal structure (Protein Database accession code: 1HRC) was used as the base structure for performing the MD simulations. The docking procedure used was similar to the studies performed earlier⁴², except for the following changes. A cubic box positioned at x, y, and z 0.032, −0.399, and −0.36, respectively as the center was built around the protein with 70 × 52 × 54 points and a spacing of 0.375 Å between the grid points was used. A total of 25 genetic algorithm runs were considered in each case with an initial population of 300 and a maximum number of 5,000,000 energy evaluations. The top 25 resulting orientations that have less than or equal to 0.5 Å root mean square deviation were clustered together. The best ligand bound to the partially unfolded cyt c structure in each case was chosen based on lowest energy as well as the conformation where the imidazole group was closest to the heme iron.

Molecular Dynamics simulations were performed using the MD software package NAMD⁴³ using the VEGA ZZ 2.3.2⁴⁴ package as the user interface. The model structure of horse heart cyt c (PDB ID: 1HRC) was downloaded from the Protein Data Bank, and explicit hydrogens were added using the package VEGA ZZ 2.3.2⁴⁴. A spherical solvent cluster with a radius of 48.5 Å positioned at geometric center of the protein was built and Cl⁻ ions were added to neutralize the system. This system was energy minimized for 2000 steps using conjugate gradient method. The energy-minimized structure was slowly heated from 0 to 310 K over a period of 100ps. Following this 2ns molecular dynamics simulation was performed at 310 K to obtain the partially unfolded cyt c structure.

For the in vitro experiments, the results are presented as means ± S.D. values from at least three experiments, and statistical analyses were performed by either paired/unpaired Student’s t-test or one-way ANOVA. As an exploratory analysis, p-values were not adjusted for multiple comparisons For the survival data, median survival and its 95% confidence interval were calculated for each group, and the two-sided log-rank test was used to examine the differences between irradiated mice and irradiated mice treated with TPP-ISA or TPP-IOA. In all these tests, a p-value of less than 0.05 was regarded as significant.