Chelates were generated from clinical products Gd-EOB-DTPA (Eovist) or Gd-BOPTA (Multihance). Metal sources were: GdCl3.6H2O, EuCl3.6H2O, TbCl3.6H2O, HoCl3.6H2O, DyCl3.6H2O, ScCl3.6H2O, and YCl3.6H2O, all from Sigma Aldrich. PrCl3.5H2O was from Strem Chemicals, Newburyport, MA.

Following a procedure reported by Patrick, et al., [10] the first step involved the oxalic acid assisted de-complexation of Gd-EOB-DTPA or Gd-BOPTA to separate out the EOB-DTPA or BOPTA ligand and Gd-oxalate (Fig. 1A). In a typical procedure, 52 mg of oxalic acid dihydrate (0.4 mmol) was dissolved in 0.6 mL DI water to obtain a clear solution. Next, 0.4 mL Gd-EOB-DTPA (0.1 mmol) was added to this solution resulting in immediate precipitation of white Gd-oxalate. The reaction mixture was centrifuged and the supernatant that contained the free EOB-DTPA ligand was pipetted out and lyophilized to obtain a crystalline solid. This solid was dissolved in 1 mL water and purified on an MPLC (medium-pressure liquid chromatography) column (C18 reverse phase column, 30 g; solvent system: A = 25 mM ammonium formate; B = methanol; eluent run composition: 0% methanol to 100% methanol, compound was eluted at 25–35% methanol in ~ 4 min, flow rate: 40 mL/min) (Fig. 1B). The product was recovered as a white solid via solvent removal and lyophilization. Mass spectrometry was used to verify the product. The same procedure was used to generate free BOPTA from Gd-BOPTA, though Gd-BOPTA is 2X concentrated versus Gd-EOB-DTPA.

Synthesis of metal chelates. A) Chemical schematic of the synthesis of Y-EOB-DTPA from Gd-EOB-DTPA. This generalized procedure was carried out for all chelates, for both EOB-DTPA and BOPTA. B, C) MPLC traces for the empty EOB-DTPA chelate (B) and Y-EOB-DTPA (C). D, E) Mass spec for Y-EOB-DTPA (D) and Y-BOPTA (E) showing chelated product and numerous sodium adducts.

To generate the metal EOB-DTPA and BOPTA complexes, the free ligands were first dissolved in 5 mL DI water, resulting in pH ~ 2–3. Metal chlorides were then added in a 1:1 proportion, further reducing the pH. Using 0.1 M NaOH, the pH of the solution was slowly brought to pH 6–6.5, and the reaction was stirred overnight at room temperature to ensure complete complexation. The products were then recovered via filtration and lyophilization. Final products were purified by MPLC under the same conditions as above and lyophilized to dryness (Fig. 1C). Chelate formation was validated by mass spectrometry and metal content was measured using ICP-QQQ-MS. Quantification was performed using multi-element calibration standards, and measurements were validated across the concentration range used in this study.

Unmodified HEK293 cells and cells stably expressing rat OATP1B2 or human OATP1B3 [11,12,13] were maintained in DMEM supplemented with 10% fetal bovine serum and 1% penicillin/streptomycin at 37 °C, 5% CO₂. Cocktails containing all eight EOB-DTPA based metal chelates or BOPTA based chelates (each individual chelate at 0.1 mM) was prepared in culture medium. This enabled multiplexed, simultaneous quantification of each chelate in a single sample. Keeping the total chelate concentration at 0.8 mM ensured the concentration was below the reported Km for hepatic transporters [14]. Cells were incubated (n = 3 replicates) with each cocktail for 10 min, washed three times with cold PBS, harvested, dried, and digested in 70% HNO₃. Metal content was quantified by ICP-QQQ-MS, with calibration against multi-element ICP standards containing all of the metals under investigation. ICP-QQQ-MS validated that each chelate in the multi-metal cocktail was present at the target concentration of 0.1 mM ± 5%, confirming dosing accuracy for both in vitro and in vivo studies.

All animal procedures were in performed according to an approved MSU IACUC protocol in accordance with AALAS guidelines for animal experiments. Animal experiments were performed in wild-type mice and in mice expressing human OATP1B1 and OATP1B3 on a rodent OATP knockout background (OATP1B1/1B3 mice) [15]. Wild-type mice (n = 3 per group) were injected via tail vein with 250 µL of either the 8-metal EOB-DTPA-based cocktail or BOPTA-based cocktail (each individual chelate at 0.1 mM). OATP1B1/1B3 mice (n = 3) were injected via tail vein with 250 µL of only the 8-metal EOB-DTPA-based cocktail. After 10 min, the time previously determined as peak liver uptake [16], mice were perfused with saline, and liver, kidney, and quadriceps tissues were harvested. Samples were microwave-digested in 70% HNO₃ for 2 h and analyzed by ICP-QQQ-MS.

Clearance studies were performed using a combined formulation of Gd-EOB-DTPA and Y-EOB-DTPA. Wild-type and human OATP1B1/1B3 knock-in mice (n = 3 per group) were injected intravenously with 250 µL of agent (0.1 mM). Mice were housed individually in metabolic cages for 24 h following injection. Urine and feces were collected separately, dried, and digested in 70% nitric acid. Renal and hepatobiliary clearance were quantified by ICP-QQQ-MS as the fraction of recovered metal excreted in urine versus feces.

For each experiment, uptake was measured in three independent cell experiments or three animals, with all eight metals assessed in every case. Because each sample was measured across all metals, the data were analyzed using a repeated-measures framework, treating metal as a within-subject factor.

To account for differences in overall uptake between samples, we used within-subject statistical approaches. The primary analysis consisted of repeated-measures ANOVA with metal as the within-subject factor. When assumptions were uncertain or confirmation was needed, we additionally applied the Friedman test as a nonparametric alternative. When a significant overall effect of metal was observed, post-hoc comparisons were performed using paired Wilcoxon signed-rank tests to compare metals within the same sample. Holm correction was applied to adjust for multiple testing, and metal chelates were considered distinct only if they remained significant after correction.

Variability between samples was further explored by visually examining uptake profiles across metals. For visualization purposes only, some plots used within-sample normalization (demeaning); all statistical analyses were conducted on the raw data. Analyses were performed in Python (v3.11) using pandas, scipy, and statsmodels, with statistical significance defined as p < 0.05.