A 3,2-Hydroxypyridinone-based Decorporation Agent that Removes Uranium from Bones In Vivo

Searching for actinide decorporation agents with advantages of high decorporation efficiency, minimal biological toxicity, and high oral efficiency is crucial for nuclear safety and the sustainable development of nuclear energy. Removing actinides deposited in bones after intake is one of the most significant challenges remaining in this field because of the instantaneous formation of highly stable actinide phosphate complexes upon contact with hydroxyapatite. Here we report a hydroxypyridinone-based ligand (5LIO-1-Cm-3,2-HOPO) exhibiting stronger affinity for U(VI) compared with the reported tetradentate hydroxypyridinone ligands. This is further revealed by the first principles calculation analysis on bonding between the ligand and uranium. Both in vitro uranium removal assay and in vivo decorporation experiments with mice show that 5LIO-1-Cm-3,2-HOPO can remove uranium from kidneys and bones with high efficiencies, while the decorporation efficiency is nearly independent of the treatment time. Moreover, this ligand shows a high oral decorporation efficiency, making it attractive for practical applications.

C ontinued development of in vivo decorporation agents that target key radionuclides from nuclear fuel cycle, especially actinides, is critical for public safety control during emergency responses to nuclear accidents and could serve as a preemptive measure to mitigate the damages from nuclear incidents linked to terrorism. Uranium is the most important naturally-occurring actinide and a crucial resource for nuclear power, which is also recognized as a global environmental contaminant due to its combined radio-and chemo-toxicities 1,2 . Although different routes that introduce uranium into the human body exist, e.g., through ingestion, inhalation, and wound related absorption, approximately two-thirds of uranium is subsequently eliminated from plasma and excreted through the kidney. However, the portion of uranium that remains is retained in deposits found in kidneys and bone tissues in the form of the hexavalent uranyl ion (UO 2 2+ ) 2 , leading to both acute and chronic renal damage, as well as a heightened risk of osteosarcoma and osteogenesis [3][4][5][6][7][8] . The latter hazard suffers from long-term and unsolved therapy challenges because of the instantaneous formation of highly insoluble uranyl phosphate complexes when uranium contacts calcium hydroxyapatite (HAP) in bones. A decorporation agent which can effectively remove uranium from these parts in humans is highly desirable.
Chelation therapy is the most effective treatment for uranium decorporation. A suitable multidentate ligand (i.e., chelator) is administered with the intent of selectively coordinating toxic metals and accelerating their excretion from the body, thereby reducing the toxicity via competitive formation of thermodynamically favorable and water-soluble molecular complexes 9 . A feasible chelation agent is chosen based on the criteria that the ligand displays a strong affinity towards uranyl ion and the coordination environment facilitates their removal 9 , while also exhibiting biocompatibility and low toxicity as a bare complexant. Diethylenetriamine pentaacetate calcium sodium salt (CaNa 3 -DTPA) and its zinc sodium salt (ZnNa 3 -DTPA) are currently the only Food and Drug Administration (FDA)-approved drugs for this purpose 10 . Additionally, NaHCO 3 is also recommended by FDA in case of internal contamination of uranium. Although DTPA has been reported for insufficient removal of intracellular actinide deposits owing to its hydrophilic nature that limits its ability to cross the cell membranes 10-12 , Grémy et al. developed a liposomal DTPA with enhanced removal efficiency of Pu/Am in liver with either prophylactic or delayed treatment 13,14 . Moreover, although DTPA agents are effective with low-valent actinides such as Pu(IV) and Am(III), they show minimal effects on U(VI) [15][16][17][18] . Macrocycles have shown strong affinity for U(VI), however, decorporation assays aimed at their interaction with U (VI) in either animal or cellular assays are limited with relatively high toxicity 19 . Functional groups, such as catechol (CAM), terephthalamide (TAM), and hydroxypyridinone (HOPO) were evaluated for their decorporation efficiency with actinides 9,20 . Most of these ligands were found to be capable of removing significantly more U(VI) from the entire body and kidneys compared to CaNa 3 -DTPA under the same conditions. Despite that CAM(S), MeTAM, and 3,4,3-LI-CAM(C) ligands can significantly reduce U(VI) deposited in bone using a high ligand to U(VI) molar ratio, CAM(S) ligands were found to be severely toxic by introducing kidney damage, and both 4-LI-MeTAM and 5-LI-MeTAM caused serious damages to the kidneys, liver, and/ or spleen 9 . Recently, a 3,2-hydroxypyridinone-grafted chitosan oligosaccharide nanoparticle (COS-HOPO) with low toxicity was shown to be effective for reactive oxygen species removal and can prevent uranium from depositing on kidney and bones by prophylactic administration 21 . In addition, a bidentate ligand 3-hydroxy-2-pyrrolidinone (HPD) was recently reported with a low cytotoxicity 22 . However, in these two works, the decorporation efficiency on removing uranium from kidney was limited, and more importantly no obvious removal effect from bones was observed by prompt intraperitoneal injection. Among all candidates reported, 5LIO-(Me-3,2-HOPO) and 3,4,3-LI(1,2-HOPO) are the two most promising chelators because of their high body removal efficiency for U(VI), low toxicities, and high oral activities 15,23,24 . The most important and probably the only drawback of these two ligands is their poor performances in removing uranium from bones at low ligand to metal molar ratio.
From previous studies, these HOPO ligands were intentionally designed to form intramolecular hydrogen bonds between the amide and hydroxide groups in order to be pre-deprotonated in the physiological pH range and could therefore more successfully coordinate U(VI) in a sense of kinetic control 23 . However, the formation of the intramolecular hydrogen bonding would partially offset the negative potential of the oxygen donors in the HOPO units, possibly leading to the decrease of the ligand's affinity for U(VI). Therefore, in this work, we introduce a chelating agent (5LIO-1-Cm-3,2-HOPO) with an additional methyl group inserted between the amide and HOPO groups in order to significantly weaken these intramolecular hydrogen bonds. This ligand exhibits significantly elevated coordination capability towards U(VI), as demonstrated by potentiometric titration experiment, synchrotron radiation-based extended X-ray adsorption fine structure (EXAFS) analysis, and density functional theory (DFT) simulation. Ultimately, results of both in vitro and in vivo assays suggest that 5LIO-1-Cm-3,2-HOPO is a potential solution to the long-term challenge of efficiently removing uranium from bones.

Results
Ligand design and synthesis. In order to weaken the intramolecular hydrogen bond formed between the amide and hydroxide groups found in many tetradentate HOPO ligands, a ligand {N,N′-[oxybis(ethane-2,1-diyl)]bis[2-(3-hydroxy-2-oxopyridin-1(2H)-yl) acetamide], denoted as 5LIO-1-Cm-3,2-HOPO, was designed by introducing the carboxylic group on the N site of pyridine ring. The distance between the amide and carbonyl group is increased by the addition of a methyl group. 5LIO-1-Cm-3,2-HOPO was obtained from a 4-step synthesis as illustrated in Fig. 1 [25][26][27] . Commercially available 1,2-dihydro-2,3-pyridinediol was first alkylated at the nitrogen position of the pyridinone ring by reacting it with excess amounts of ethyl bromoacetate at 150°C to yield the ester (product A). Next, the hydroxyl group in A was protected with a benzyl moiety, and the ester group was activated via hydrolysis to afford product B. 5LIO-1-Cm-3,2-HOPOBn was then obtained via amidation of the carboxylate group with the amine backbone. Benzyl protection was finally removed by 5% Pd/ C to obtain 5LIO-1-Cm-3,2-HOPO with a total yield of 43%. The chemical shifts observed in the spectra of 1 H NMR and 13 C NMR all corroborate with the chemical structure of the molecules, and the results of elemental analysis and LC-MS are consistent with the formula of 5LIO-1-Cm-3,2-HOPO (Supplementary Figure 1). In the FTIR spectra, the pattern of 5LIO-1-Cm-3,2-HOPO contains peaks at 2947 cm −1 and 1080 cm −1 , corresponding to -CH 2 group and the symmetric stretch of C-O-C, respectively (Supplementary Figure 2a).
Solution thermodynamic studies. The solution thermodynamic data of 5LIO-1-Cm-3,2-HOPO was first measured to evaluate its complexation behavior with U(VI). The protonation constants of the free ligand, 5LIO-1-Cm-3,2-HOPO (denoted as LH 2 in this part), were determined by potentiometric titrations. The formation constants could be calculated with equation (1) in the experimental section (Table 1). The two protonation constants, pK a1 and pK a2, were measured to be 8.3(5) and 9.3(4), respectively.
Characterizations of UO 2 −5LIO-1-Cm-3,2-HOPO Complex. The 1 H NMR and 13 C NMR spectra of the complex in DMSO-d 6 were collected (Supplementary Figure 1f). A slight difference between the chemical shifts of the carbon atoms from UO 2 −5LIO-1-Cm-3,2-HOPO complex and from the ligand can be observed in the 13 C NMR spectra collected in DMSO-d 6 . More notably, the feature at 8.97 ppm in 1 H NMR spectrum, which can be observed for the raw ligand and assigned to the hydroxyl group, disappears for the UO 2 -5LIO-1-Cm-3,2-HOPO complex, initially suggesting the complexation between the ligand and U (VI). In comparison with the FTIR spectrum of 5LIO-1-Cm-3,2-HOPO, the spectrum of UO 2 -5LIO-1-Cm-3,2-HOPO exhibits an additional peak at 899 cm −1 attributed to the uranyl group (Supplementary Figure 2b). A significant intensity reduction of the peak at 3282 cm −1 assigned to the hydroxyl group was observed for the spectrum of UO 2 −5LIO-1-Cm-3,2-HOPO, when compared with that of 5LIO-1-Cm-3,2-HOPO. Nevertheless, much more powerful evidence comes from the elemental analysis and LC-MS analysis, confirming the formation of UO 2 -5LIO-1-Cm-3,2-HOPO complex with a metal to ligand molar ratio of 1:1 (Supplementary Figure 1f).
Density functional theory (DFT). We performed DFT calculations to reveal the effect of intramolecular hydrogen bonds on the interaction between the ligand and U(VI). In 5LIO-(Me-3,2-HOPO), the strong intramolecular hydrogen bond (approximately 1.96 Å) between the amide group and the hydroxyl group of the pyridinone ring promotes the formation of a planar local structure between the two bonding components (Fig. 2a). This strong hydrogen bond and local planar structure remain intact when chelating a UO 2 2+ cation (Fig. 2b). On theother hand, for 5LIO-1-Cm-3,2-HOPO, we obtained two different stable structures, one containing-NH···N (pyridine) intramolecular hydrogen bonds and the other containing -NH···O (pyridinone) intramolecular hydrogen bonds [ Fig. 2c State I and State II]. Stable structure I is more energetically favorable than II by a relatively small energy difference of approximately 0.96 kcal mol −1 . The calculated transition state (Fig. 2c) between the two structures shows a 1.37 kcal mol −1 energy barrier for this structural transformation. Such a low energy barrier and small energy difference imply that the C-N and C-C bonds on either side of the methylene group can freely rotate and that the two types of intramolecular hydrogen bonds in 5LIO-1-Cm-3,2-HOPO can spontaneously transform at room temperature. Compared to 5LIO-(Me-3,2-HOPO), the increase in the local degrees of freedom of 5LIO-1-Cm-3,2-HOPO can be attributed to the cooperation of the inversion of the pyridinone ring and the addition of the methylene group, leading to the weakening of the intramolecular hydrogen bonds (Fig. 2c). We then calculated the Gibbs free energy change for the  Table 2). The small difference (~0.02 eV) in the values of their binding energies between the UO 2 2+ cation and the chelator indicates that alteration of the hydrogen-bonding scheme would not drastically affect the complexing ability of the ligand. The intramolecular hydrogen bonds transformation and minor change of the chain folding form are not expected to significantly affect the binding energy (−9.23/−9.21 eV). In contrast, the binding energy difference between UO 2 -5LIO-(Me-3,2-HOPO) and UO 2 -5LIO-1-Cm-3,2-HOPO (−8.81 eV vs −9.23/−9.21 eV) is obvious. Therefore, this calculation result reveals the enhancement of the binding ability of the 5LIO-1-Cm-3,2-HOPO ligand, especially the oxygen denticity.
We propose that the enhanced uranyl binding in 5LIO-1-Cm-3,2-HOPO originates from its intrinsic structural and electronic features. From a classical point of view, the negatively charged oxygen of the deprotonated hydroxyl group is strongly attracted to the hydrogen atom of the nearby amide in 5LIO-(Me-3,2-HOPO), thereby generating a strong -NH···O (pyridine) hydrogen bond (Fig. 3a). In contrast, when the pyridinone ring is reversed, the negatively charged oxygen of the deprotonated hydroxyl group in 5LIO-1-Cm-3,2-HOPO can be completely exposed to the environment without any intramolecular interactions (Fig. 3b). This intrinsic structural feature is expected to endow a spatial advantage in coordinating the positively charged UO 2 2+ cation. From a quantum perspective, DFT calculations show that, for both ligands, the global electrostatic potential (ESP) minima over the electron density surface are located between the two oxygen atoms of the pyridinone ring (Fig. 3c, d). Consequently, these two oxygen atoms act as the targeted chelating sites during complexation reactions. The calculated ESP minimum of 5LIO-1-Cm-3,2-HOPO (−189.21 kcal mol −1 ) is lower than that of 5LIO-(Me-3,2-HOPO) (−178.25 kcal mol −1 ), indicating that 5LIO-1-Cm-3,2-HOPO could provide more effective long-range electrostatic attractions for UO 2 2+ . Moreover, the negative ESP area originating from the two oxygen atoms is further broadened over the electron density surface in 5LIO-1-Cm-3,2-HOPO (46.07 Å 2 ) than in 5LIO-(Me-3,2-HOPO) (38.67 Å 2 ). This observation indicates that 5LIO-1-Cm-3,2-HOPO could provide a wider effective landing surface region for UO 2 2+ complexation. These calculation results indicate that the exposure of the deprotonated hydroxyl group and the weak and transformable intramolecular hydrogen bond synergistically contributes to the unique distribution of the negative ESP between the two oxygen atoms. To help directly visualize and compare the ESP features, we depicted the two ESP isosurfaces for the two ligands, at a same value of +/− 188.25 kcal mol −1 , in Fig. 3e, f. Clearly, the negative ESP distribution space is much larger in 5LIO-1-Cm-3,2-HOPO than in 5LIO-(Me-3,2-HOPO). In addition, Morokuma scheme energy decomposition analyses (EDA) results show that the E(elstat) term in UO 2 -5LIO-1-Cm-3,2-HOPO is ca. 1.16 eV larger than that in UO 2 -5LIO-(Me-3,2-HOPO). This result confirms our perspective that the electrostatic effect of 5LIO-1-Cm-3,2-HOPO can be fully released.
Most surprisingly, DFT calculations further reveal an important interaction between the axial oxygen of uranyl and the side  Fig. 4c. Formation of the two intermolecular hydrogen bonds are mainly attributed to the longer and more flexible side chain of the 5LIO-1-Cm-3,2-HOPO chelator. In contrast, for 5LIO-(Me-3,2-HOPO), the side chain of this chelator is relative shorter and more rigid, thus only intramolecular -NH···O (pyridine) hydrogen bonds can be formed (Fig. 2b).
In order to quantitatively evaluate the bonding contributions of the intermolecular hydrogen bond, we also performed topology analyses for the UO 2 -5LIO-1-Cm-3,2-HOPO complex (Fig. 2d, right). The UO 2 -5LIO-(Me-3,2-HOPO) complex (Fig. 2b) was also subjected to this analysis for comparison. Based on the quantum theory of atoms in molecules (QTAIM), bond critical points (BCP) between relevant H and O atoms were found in the ground state electron densities, and the potential energy density, V(r), of these BCPs were calculated. According to the relationship between V(r) and the hydrogen bond energy, E HB34 , the E HB can be estimated as E HB = V(r BCP ) / 2. It can be seen that the E HB value of the intramolecular -NH···O (pyridinone) hydrogen bond in UO 2 -5LIO-(Me-3,2-HOPO) is about -0.36 eV, signifying a relatively stronger hydrogen bond interaction. This intramolecular -NH···O (pyridinone) hydrogen bond was significantly weakened to −0.27 eV in UO 2 -5LIO-1-Cm-3,2-HOPO. The E HB values of the intermolecular -NH···O (uranyl) hydrogen bonds is −0.11 eV, implying relatively weaker intermolecular hydrogen bond interactions (Supplementary Table 3). Although the binding contributions (−0.16 eV in total) of the intermolecular hydrogen bond is relatively small compared to the total binding energy (−9.21 eV), it can be expected that the intermolecular hydrogen bond could provide an additional driving force for the coordination of uranyl and therefore enhance the chelating ability of 5LIO-1-Cm-3,2-HOPO.
Cytotoxicity of chelating agents. Given the high affinity of 5LIO-1-Cm-3,2-HOPO for uranyl, the U(VI) sequestration performance and toxicity of 5LIO-1-Cm-3,2-HOPO were evaluated and compared with those of the clinically-used ZnNa 3 -DTPA and the previously reported most optimal tetradentate ligand 5LIO-(Me-3,2-HOPO) in vitro and in vivo. Renal injury is one of the major concerns in the case of uranium contamination. Therefore, U(VI) uptake and release assays were first conducted using renal proximal tubular epithelial cells from rat (NRK-52E cells). The toxicity assay of U(VI) with chelation therapy agent has been performed, and 12.4 μM was considered as the acceptable concentration for the following cellular assays 35    show that the comprehensive toxicity of UO 2 -5LIO-1-Cm-3,2-HOPO is slightly lower than that of UO 2 -ZnNa 3 -DTPA at low dosage, and notably lower than that of UO 2 -5LIO-(Me-3,2-HOPO) (Supplementary Table 4, Fig. 5a).
U(VI) uptake and release. In vitro U(VI) uptake and release assays were conducted to investigate the uranyl removal efficiency of 5LIO-1-Cm-3,2-HOPO at the cellular level. In vitro assay for U (VI) uptake and release from NRK-52E cells were performed by adding 12.4 μM U(VI) solution to cells, followed by 320.0 μM solution of the chelators. Fig. 5b shows that the addition of 5LIO-1-Cm-3,2-HOPO can remove 99.5% of uranium from NRK-52E cells, whereas ZnNa 3 -DTPA can remove only 12.5% of uranium under the same condition. Clearly, the uranium removal efficiency of 5LIO-1-Cm-3,2-HOPO is much higher than that of ZnNa 3 -DTPA at an equal molar dosage (Supplementary Table 5). However, since the NRK-52E cells were cultured in medium that contain uranium throughout the experiment, question remains as whether the ligand mainly play the role of preventing the uranium uptake of the cells, or enhancing the release of the intracellular uranium. Therefore, another in vitro assay for U(VI) release from NRK-52E cells was designed by adding 12.4 μM U (VI) solution to cells first, then removing the U(VI) culture medium, followed by the addition of 320.0 μM solution containing the chelators. As shown in Fig. 5b, the treatment of 5LIO-1-Cm-3,2-HOPO resulted in a uranium removal efficiency of 75.8%, illustrating that 5LIO-1-Cm-3,2-HOPO plays a major role for enhancing the U(VI) release from cells (Supplementary Table 6).
In the multiple-dose group, 5LIO-1-Cm-3,2-HOPO and ZnNa 3 -DTPA (97 μmol kg −1 ; molar ratio to uranium of 46:1) were ip injected at 3 min, 6 h, 24 h, and 48 h after the initial iv injection of U(VI). At 72 h after the iv injection of 0.5 mg U(VI) kg −1 , the kidneys and femurs from the control group contained 5.22 ± 1.13 and 3.50 ± 0.39 μg U(VI) per gram of tissue, respectively (Fig. 6f). Compared with the control group, 5LIO-1-Cm-3,2-HOPO significantly reduced U(VI) levels in the femur by 47.4%. Another multiple-dose assay was conducted by ip injection of 5LIO-1-Cm-3,2-HOPO and ZnNa 3 -DTPA (97 μmol kg −1 ; molar ratio to uranium is 46:1) at 1, 7, 25, and 49 h after the initial iv injection of U(VI). Similarly, kidneys and femur samples were obtained 72 h after the initial U(VI) injection. Notably, 70.9% of U(VI) deposited in the kidneys and 50.0% in femurs were removed in the 5LIO-1-Cm-3,2-HOPO group compared with the untreated U(VI)-injected controls (Fig. 6f). These values are almost identical with those of the multiple-dose group treated 3 min after the U(VI) injection. This achievement overcomes a large hurdle in developing actinide chelation agents and opens up the possibility of complete uranium removal from bones by increasing the dosage further (Supplementary Table 13).
In most nuclear accidents, an immediate response to radionuclide introduction into human bodies is not applicable. Therefore, decorporation agents that can still function despite delayed administration are preferred for chelation therapy with elevated significance. To fully study the relationship between the decorporation efficiency and the time of administration, 5LIO-1-Cm-3,2-HOPO (193 μmol kg −1 ; molar ratio to uranium is 92:1) was given via ip injection at the following three different time intervals: (1) 6, 12, 30, and 54 h; (2) 12, 18, 36, and 60 h; (3) 24, 30, 48, and 72 h, after the initial iv injection of U(VI). The three groups with 6, 12, and 24 h delayed multiple-dose administration display a similar removal level of U(VI) from kidneys of 61.4%, 62.4%, and 65.0%, respectively. For the decorporation efficiency in femurs, all three groups significantly reduced U(VI) accumulation in femurs by 39.6, 24.2, and 30.8%, respectively (Fig. 6g and Supplementary Table 14), suggesting that the decorporation efficiency is nearly independent of the time of administered treatment.
Hydroxyapatite, or HAP, is the main inorganic mineral component in bones and can form highly stable and insoluble complexes with the uranyl cation. Therefore, an in vitro desorption experiment was designed to demonstrate the thermodynamic possibility of 5LIO-1-Cm-3,2-HOPO desorbing the complexed uranyl from HAP to further support the experimental performance of the ligand removing uranium from bones. For comparison, the desorption experiments using 5LIO-(Me-3,2-HOPO) and ZnNa 3 -DTPA were also conducted using the same method. As shown in Fig. 6h, after that the sorption equilibrium between U(VI) and HAP had been reached after 180 min, the addition of 5LIO-1-Cm-3,2-HOPO resulted in rapid uranyl desorption from HAP with a high desorption rate of 79.6%. This signifies that the process of decorporating uranium from bone with 5LIO-1-Cm-3,2-HOPO is indeed thermodynamically favorable. In sharp contrast, 5LIO-(Me-3,2-HOPO) and ZnNa 3 -DTPA only resulted in a desorption rate of 18.6 and 3.4%, respectively.

Discussion
The results described above demonstrate that weakening the intramolecular hydrogen bonds between the oxygen donor sites within the HOPO ligand can significantly increase their uranium coordination capabilities. According to this simple idea, we have successfully synthesized a unique chelating ligand, 5LIO-1-Cm-3,2-HOPO. The combination of solution titration studies, EXAFS analysis, FTIR analysis, NMR analysis, LC-MS analysis, elemental analysis, and first principles theoretical analysis demonstrates the coordination mode and the thermodynamic feasibility for decorporating uranium from kidneys and bones. DFT calculations indicate that with the additional acyl derivative on the ring nitrogen, the formation of strong intramolecular hydrogen bond is greatly prohibited, giving rise to strengthened uranium-HOPO interaction consequently. In vitro U(VI) removal assays results show that 5LIO-1-Cm-3,2-HOPO can significantly enhance the release of intracellular U(VI) release from cells. Further, in vivo uranium decorporation assays demonstrate that this ligand not only improve the reduction of U(VI) levels in kidneys, but also show a record high removal efficiency of uranium from bones even for the oral and delayed treatments, while maintaining a low toxicity at the similar level with ZnNa 3 -DTPA. Overall, these results signify that 5LIO-1-Cm-3,2-HOPO represents one of the most promising U(VI) decorporation agents and its practical application is expected to be realized in the near future.
1,2-dihydro-2,3-pyridinediol (22.6 g, 0.2 mol) and ethyl bromoacetate (133.6 g, 0.8 mol) were stirred together, then flushed with nitrogen for 1 h and refluxed under nitrogen for 24 h. The solution was allowed to cool down to room temperature, yielding a beige precipitate. The solution was filtered and the product was washed with acetone for at least three times. Recrystallization from 95% ethanol yielded colorless needle-shape crystals, Ethyl 2- 5% Pd/C was added slowly to a solution of 5LIO-1-Cm-3,2-HOPOBn (4.0 g, 6.8 mmol) dissolved in 100 mL methanol. The mixture was then stirred under H 2 at room temperature for 4 h. After filtration, the product was re-dissolved in DMF and filtered. The filtrate was subjected to rotary evaporation to give a pale gray solid. The product was dried at 50°C under vacuum to yield the final product 5LIO  (2.5-11). Data were also refined by nonlinear least-squares and the results are included in Table 1 and Supplementary Table 1.
The species distribution of 5LIO-1-Cm-3,2-HOPO and metal ions were calculated with Hyss at the condition of 10 −3 M L and 10 −4 M metal ions from pH 3 to 11 (Supplementary Figure 3), taking the formation of uranyl-hydroxide and uranyl-carbonate complexes into consideration.
Supplementary Figure 5 shows the refinement details for the cumulative constants of 5LIO-1-Cm-3,2-HOPO and metal elements with nonlinear least-squares program (hyperquad 2013). For the UO 2 -L system, although the formation constants measurement of UO 2 -L was carried out by competition titration with EDTA (denoted L e H 4 ), only a small quantity of UO 2 L e H + and UO 2 L e exist at low pH. The results suggest that EDTA is not able to compete with 5LIO-1-Cm-3,2-HOPO (Supplementary Figure 5a). Extended X-ray Absorption Fine Structure (EXAFS). X-ray absorption spectroscopy measurements were performed at beamline 14 W1 of the Shanghai Synchrotron Radiation Facility with a Si (111) double crystal monochromator in transmission mode for the uranium L 3 -edge spectra. The electron beam energy of the storage ring was 3.5 GeV, and the maximum stored current was approximately 210 mA. Energy calibration was performed using a zirconium foil (~17,998 eV). The sample was measured thrice, and the spectra were averaged. The uranium L 3 -edge EXAFS data were analyzed using the standard procedures in the Demeter program 38 . Double-electron excitations affect the EXAFS signal and can influence the results of the data analysis. Thus, in the uranium L 3 -edge EXAFS experimental spectra, the double-electron excitations were subtracted as a reflection of the data translated to the position in energy of the excitation using the standard procedures in Demeter. Supplementary Figure 4 shows the uranium L 3 -edge EXAFS data before and after subtracting the double-electron excitation in k-and R-space, in which the features at very low distances (R ≈ 1 Å) clearly improved. The coordination number of the uranyl center is expected to be influenced, while the bond length is subtly influenced by double-electron excitations. Theoretical EXAFS data were calculated using FEFF 9.0 39 . Fitting procedure was performed on the k 3 -weighted FTEXAFS from 2-14 Å −1 . An R window of 1-4 Å was used for the fitting. The amplitude reduction factor, S 0 2 was fixed at 0.83 in the EXAFS fits, and the shifts in the threshold energy, ΔE 0 , were constrained to be the same value for all fitted shells.
Computational method. First principle calculations based on density functional theory (DFT) were performed by using the Gaussian 09 program 40 to investigate the structural, binding properties, and the quantitative analyses of molecular surfaces for two types of ligands and two types of ligand-uranyl chelates. For the 5LIO-(Me-3,2-HOPO) and the 5LIO-1-Cm-3,2-HOPO ligands, two monomer fragments containing their foremost structural features were used as the computational models, as shown in Fig. 3a and b. For the ligand-uranyl chelate models, the complete double-deprotonated chelating agent molecules were used. Since it is widely recognized that the penta-coordinate UO 2 2+ is the dominant species in aqueous solutions, an additional water molecule was added to supplement the vacant coordination site of the uranyl center, as seen in Fig. 3c and d. Full geometric optimizations for all models were carried out in the liquid phase at the B3LYP-D3/SDD~6-31G(d) level. The hybrid-B3LYP exchange-correlation functional 41,42 combined with the D3 version of Grimme's dispersion 43 were employed. The Stuttgart/Dresden relativistic effective core potential 44 and associated valence basis set (SDD, 60 core electrons and 32 valence electrons) 45 were applied for uranyl, while the standard Gaussian-type 6-31G(d) basis sets 46 were used for other light atoms (C, H, O, and N). The solvation effects were included by using SMD implicit solvent model 47 with water as the solvent. Harmonic vibrational frequencies were calculated afterwards to confirm that the obtained structures possessed stationary points (stable structures) or saddle points (transition states) on their potential energy surfaces. The binding energies (E b ) between the UO 2 2+ cation and the remaining ligands (containing the chelating agent and one water molecule, with two negative charges) in ligand-uranyl chelates were calculated by E b = E(complex) − E(UO 2 2+ ) − E(remaining), where the three terms of the right side of this formula represent the total energies of the chelates, the UO 2 2+ cations, and the remaining ligands. All geometries used for calculating these three energy terms maintained the same atomic arrangements as the complex structures. Based on the ground state electron densities of the two single-deprotonated chelator structures (Fig. 4a), the global electrostatic potential (ESP) minima on the van der Waals (VDW) surfaces (isodensity = 0.001 a.u.) and the local negative ESP areas, contributed from the two oxygen atoms, were then derived to evaluate the electrostatic properties of the two ligands. The quantitative analyses of molecular surfaces, RGD, and topology analyses were performed using the MultiWFN program 48 .
In order to support the electrostatic amplification effect more strongly, we performed Morokuma scheme energy decomposition analyses (EDA) based on two simplest but most typical models, as seen in Supplementary Figure 6. In these two models, the side chains of two ligands were removed, allowing us to only focus on the O-U interactions. The structures of Fig. 2a, b were optimized at B3LYP-D3/ SDD~6-31G* level in gas phase. Then, the EDA calculations were performed at ZORA-PBE/TZP level in gas phase. According to the Morokuma EDA scheme, the total binding energy (E b ) between the uranyl cation and the ligand fragments can be decomposed as: where, the E(elstat), E(Pauli) and E(orb) represent the electrostatic interaction, the Pauli repulsion and the orbital interaction, respectively. The E(elstat) and E(orb) are positive interactions (with negative values) for the binding and the E(Pauli) is the negative interaction (with positive value) for the binding. The EDA results are summarized in Supplementary Figure 6c.
The ΔG value was calculated by the equation (3) and (4), and the G H +, G LHn−1 , G LH values were calculated by theoretical calculation.
Cytotoxicity assays of uranyl and chelating agents. To compare the cytotoxicity among the ligands and determine the optimized concentrations of U(VI) and chelating agents for removal assays, cytotoxicity assays were performed first. Cells in log phase were cultured in a 96-well plate for 24 h, and then 0.1 mL culture medium contained 12.4 μM U(VI) and different concentrations of chelating agents: 20.0, 40.0, 80.0, 160.0, and 320.0 μM. After 48 h, 10.0 μL of CCK-8 was added to every well and the cells were cultured for 1-2 h. The absorbance spectrum of every well was obtained by ELIASA (FilterMax F5). The cell survival rate could be calculated using the formula: OD (Experimental group) / OD (control group) × 100%. Each group of above assays was performed with six parallel samples, and the results are shown in Supplementary Table 4. All the independent experiments were statistically tested with SAS. 320.0 μM was considered to be an acceptable concentration of the chelating agents for further assays. In both assays, only culture medium was added to the wells of the control group.
In vitro uranium removal assays. For the U(VI) uptake and release assay, exponentially growing NRK-52E cells were cultured for 24 h, and then 12.4 μM U (VI) and 320.0 μM of chelating agents were added to the experimental group, while only 12.4 μM U(VI) was added to the control group. After 48 h cultured with U(VI) and chelators, the cell cultures were washed with PBS, trypsinized, counted, and lysed. For the U(VI) release assay, the NRK-52E cells were cultured for 24 h, followed by the addition of 2.0 mL culture medium contained 12.4 μM U(VI) into the experimental and control groups, and then the culture medium contains uranium was discarded 24 h later. Then 320.0 μM of chelating agent was added to the experimental group, while only culture medium was added to the control group. After 24 h the addition of chelating agents, the cell cultures were washed with PBS, trypsinized, counted, and lysed. Each group was performed with three parallel samples. The U(VI) content of the cells were analyzed by ICP-MS (Thermo Scientific) and converted to ng per 10 6 cells (Supplementary Table 5 and 6).
In vivo uranium decorporation assays. For the single dosage group with intraperitoneal (ip) injection, U(VI) solutions (UO 2 (NO 3 ) 2 ·6H 2 O, 3.0 mg) were prepared as that the standard dosage (0.5 mg 238 U(VI) kg −1 ) was contained in 20.0 mL of 0.14 M NaCl at pH 4-5; all the chelating agent sample solutions were adjusted to pH 7-8 with 0.1 M NaOH, and molar ratio of all the ligands to U(VI) is 92:1, except for the NaHCO 3 solution, whose molar ratio to U(VI) is 184:1. The 5LIO-1-Cm-3,2-HOPO (29.0 mg) or 5LIO-(Me-3,2-HOPO) (29.3 mg) was dissolved in 1.0 mL DMSO, 2.0 mL 0.14 M NaCl and a certain volume of 0.1 M NaOH solution, then diluted to 6.0 mL by adding 0.14 M NaCl solution; ZnNa 3 -DTPA (37.4 mg) was firstly dissolved in 3.0 mL 0.14 M NaCl, and then adjusted the pH to 7-8 with 0.1 M NaOH, finally diluted to 6.0 mL by adding 0.14 M NaCl solution; 3 mL of HEDP (7.4 mg) or NaHCO 3 (6.0 mg) solution was prepared following the procedure of ZnNa 3 -DTPA. For the single dosage group with oral administration, all the chelating agent sample solutions were adjust to pH 7-8 with 0.1 M NaOH, and the molar ratio of all the ligands to U(VI) is 307:1. 5LIO-1-Cm-3,2-HOPO (48.5 mg) or 5LIO-(Me-3,2-HOPO) (48.1 mg) was dissolved in 1.0 mL DMSO, 1.0 mL 0.14 M NaCl and a certain volume of 0.1 M NaOH solution, then diluted to 3.0 mL by adding 0.14 M NaCl solution; ZnNa 3 -DTPA (62.3 mg) was firstly dissolved in 1.0 mL 0.14 M NaCl, and then adjusted the pH to 7-8 with 0.1 M NaOH, finally diluted to 3.0 mL by adding 0.14 M NaCl solution. For multiple dosage groups, 5LIO-1-Cm-3,2-HOPO (58.0 mg, molar ratio to U (VI) is 46:1) was dissolved in 2.0 mL DMSO, 10.0 mL 0.14 M NaCl and a certain volume of 0.1 M NaOH solution, then diluted to 24.0 mL by adding 0.14 M NaCl solution; ZnNa 3 -DTPA (74.8 mg, molar ratio to U(VI) is 46:1) was firstly dissolved in 10.0 mL 0.14 M NaCl, and then adjust the pH to 7-8 with 0.1 M NaOH, finally diluted to 24.0 mL by adding 0.14 M NaCl solution. 5LIO-1-Cm-3,2-HOPO (145.0 mg, molar ratio to U (VI) is 92:1) for the 6 h, 12 h, and 24 h delayed multiple dosage groups was dissolved in 5.0 mL DMSO and diluted to 30.0 mL followed the above method. Female Kunming mice (84 to 86 days old, 30 ± 4 g body weight) were obtained for the in vivo uranium decorporation assay. These animal assays were approved by the Animal Care and Use Committee of Soochow University and were in accordance with the National Institutes of Health guidelines in its guide for the care and use of laboratory animals.
The single dosage group with ip injection: The first assay was carried out to compare the removal efficiency between 5LIO-1-Cm-3,2-HOPO and ZnNa 3 -DTPA, and the second assay was carried out to compare the removal efficiency between 5LIO-1-Cm-3,2-HOPO and 5LIO-(Me-3,2-HOPO), the third assay was carried out to compare the removal efficiency between HEDP and NaHCO 3 . Experimental and control groups of five mice each were intravenously (iv) injected with 0.2 mL 238 U(VI) solution and 0.5 mL ligand solution was administered by intraperitoneal (ip) injection promptly after. 238 U-injected control group were only given 0.5 mL saline. All mice were fed with water and food 4 h after the 238 U injection. Mice were dissected to obtain the kidneys, femur, liver, spleen and muscle samples. The samples of tissues were digested in aqua regia, and the contents of 238 U in each were determined by ICP-MS (Thermo Scientific). The results of these assays are listed in Supplementary Table 7-10. The single dosage group with oral administration: Similarly, two assays were carried out to compare the removal efficiency among 5LIO-1-Cm-3,2-HOPO, 5LIO-(Me-3,2-HOPO), and ZnNa 3 -DTPA. The mice had been fasted for 16 h before administering uranium and chelating agents. Experimental and control groups of five mice each were intravenously (iv) injected with 0.2 mL 238 U(VI) solution, and 0.5 mL ligand solution was administered by gastric tube promptly after. 238 U-injected control group were only given 0.5 mL saline. Urine and feces were collected and mice were killed 24 h after the 238 U injection. Mice were dissected to obtain the kidneys, femur, liver, spleen and muscle samples. The samples of tissues and excreta were digested in aqua regia, and the contents of 238 U in each were determined by ICP-MS (Thermo Scientific). The results of these assays are listed in Supplementary Table 11 and 12. The multiple dosage and delayed multiple dosage groups: two assays were carried out to compare the removal efficiency between 5LIO-1-Cm-3,2-HOPO and ZnNa 3 -DTPA with different treated time and dosage, respectively. For the multiple dosage and 1 h multiple dosage groups, the experimental and control groups of five mice were i.v.-injected with 0.2 mL 238 U(VI) solution. For the multiple dosage group, 0.5 mL ligand solution was given by ip injection at 3 min, 6, 24, 48 h, while the 1 h multiple dosage group was given 0.5 mL ligand solution by ip injection at 1, 7, 25, and 49 h. 238 U-injected control group were only given 0.5 mL saline. All mice were fed with water and food 4 h after the initial 238 U injection. Urine and feces were collected and the mice were killed 72 h after the 238 U injection. Mice were dissected to obtain the kidney and femur samples. For the 6 or 12 or 24 h delayed multiple dosage groups, 0.5 mL ligand solution was given by ip injection at 6, 12, 30, and 54 h after the initial iv injection of U(VI) for the 6 h delayed multiple dosage group, the 12 and 24 h delayed multiple dosage group was given 0.5 mL ligand solution by ip injection at 12, 18, 36, 60 h and 24, 30, 48, and 72 h after the initial iv injection of U(VI), respectively. 238 U-injected control group were only given 0.5 mL saline. All mice were fed with water and food 4 h after the initial 238 U injection. Mice were dissected to obtain the kidney, femur, liver, spleen and muscle samples. The samples of tissues were digested in aqua regia, and the contents of 238 U in each were determined by ICP-MS (Thermo Scientific). These results are listed in Supplementary Table 13 and 14.