Measuring Dynamic Changes in the Labile Iron Pool in Vivo with a Reactivity-Based Probe for Positron Emission Tomography.

Redox cycling of iron powers various enzyme functions crucial for life, making the study of iron acquisition, storage, and disposition in the whole organism a worthy topic of inquiry. However, despite its important role in biology and disease, imaging iron in animals with oxidation-state specificity remains an outstanding problem in biology and medicine. Here we report a first-generation reactivity-based probe of labile ferrous iron suitable for positron emission tomography studies in live animals. The responses of this reagent to systemic changes in labile iron disposition were revealed using iron supplementation and sequestration treatments in mice, while the potential of this approach for in vivo imaging of cancer was demonstrated using genetically and pathologically diverse mouse models, including spontaneous tumors arising in a genetically engineered model of prostate cancer driven by loss of PTEN.

Introduction

Iron is a transition metal nutrient whose redox cycling is utilized by iron-dependent enzymes to perform various biological functions essential to life.[1,2] The storage, transport, and utilization of iron is highly regulated in normal cells and tissues due to the potential of free ferrous iron to promote Fenton chemistry and the formation of reactive oxygen species. Accordingly, iron in biology is stored and transported in an inert ferric state, bound with high affinity by proteins including ferritin (storage) and transferrin (transport). Unloading of iron from transport and storage proteins involves reduction to the soluble ferrous state and entry into the cytosolic labile iron pool (LIP),[3,4] from which iron is utilized to produce essential enzyme cofactors including iron heme[1] and iron–sulfur clusters.[5,6] Dysregulation of iron homeostasis occurs in diverse disease states significantly impacting public health, including neurodegenerative disease,[7] cardiovascular disease,[8] inflammation,[9] and cancer.[10,11]

Given its important role in biology and disease, the development of chemical reagents to detect iron with metal-ion and oxidation-state specificity is an important problem in chemical biology.[12,13] One of the earliest and most widely used probes, calcein AM, is a chelation-based reagent that unfortunately exhibits poor selectivity for Fe(II) over Fe(III) and other metal ions.[14,15] Improved metal and oxidation-state selectivity was later achieved in the first generation of reactivity-based iron probes, such as Rho-Nox1[16] and IP-1.[17] Subsequently, Renslo and co-workers introduced the reactivity-based probe TRX-PURO,[18] an iron-caged form of puromycin that is activated by Fe(II)-promoted fragmentation of a 1,2,4-trioxolane (TRX) moiety. Immunofluorescence imaging of TRX-PURO treated cells affords improved sensitivity[18] compared to Rho-Nox1 and IP-1, but its use in mice requires the laborious ex vivo analysis of mouse tissues. Caged fluorescent probes are also unsuitable for in vivo studies due to the limited tissue penetration of short wavelength probes.[19,20] Caging of d-aminoluciferin with the TRX moiety used in TRX-PURO produced the iron probe ICL-1,[21] which enables imaging of labile iron in live, luciferase-expressing mice.

Despite the advances described above, reactivity-based imaging of labile ferrous iron in nontransgenic animals and humans remains an outstanding problem in biology and medicine. We considered that this challenge might be met in Fe(II)-sensitive, TRX-based radiotracers for positron emission tomography (PET). Incorporation of positron-emitting radionuclides such as fluorine-18 or carbon-11 in biologically active small molecules generally has minimal pharmacological impact.[22] Moreover, PET is used widely in both basic science and clinical settings, affording 3-D noninvasive views of an experimental animal model or human patient.[23]18F radionuclide is also widely available for routine preclinical and clinical use, and has a convenient 110 min half-life that allows sufficient time for radiotracer preparation and imaging, while ensuring a relatively low radiation dose in patients.[24]

The degradation mechanism of 1,2,4-trioxolane antimalarials like artefenomel[25] (Figure A) has been widely studied, with Fe(II)-promoted cleavage of the peroxide bond and a subsequent β-scission reaction producing carbon-centered radicals that can be intercepted with stable radical reagents like TEMPO,[26] or by nearby biological macromolecules when formed in cells.[27] This radical trapping chemistry thus provides a mechanism by which a putative PET reagent would become sequestered in cells and tissues in an Fe(II)-dependent fashion. Accordingly, we designed the trioxolane-based reagent 18F-TRX in which a short linker bearing 18F is introduced on the adamantane moiety of artefenomel (Figure A). The 4-fluorobenzamide function in 18F-TRX would be introduced in a final reaction step, using well-established N-succinimidyl-4-[18F]fluorobenzoate radiochemistry. We anticipated that 18F-TRX, like artefenomel, would be stable and well-distributed in vivo, while reacting and becoming sequestered in those cells/tissues where aberrant concentrations of labile ferrous iron are encountered (Figure B).

Figure 1

(A) Structure of artefenomel and 18F-TRX. (B) Reaction with Fe2+ yields 18F-TRX-linked biomolecule adducts that are sequestered in cells at site(s) of reaction.

Herein, we describe the synthesis of 18F-TRX, a first-generation reactivity-based PET probe of labile ferrous iron. We show that this new scaffold retains Fe(II)-dependent reactivity in vitro and accumulates in cells in an Fe(II)-dependent fashion. We describe mouse pharmacokinetic and biodistribution studies in healthy mice with macro- and microdosing, where we observed relatively rapid elimination of 18F-TRX from the blood pool and significant radiotracer uptake in small intestine at 60 min postinjection. The iron-sensitivity of 18F-TRX biodistribution in vivo was confirmed using iron supplementation and withholding strategies. Finally, we show enhanced accumulation of 18F-TRX in human and mouse tumors (up to ∼5% ID/g) compared to normal reference tissue compartments like blood and muscle. Overall, this study marks a first step toward a general method for sensitive and selective imaging of labile iron in living animals and suggests a new approach for functional imaging of cancer.

Results and Discussion

Radiosynthesis of 18F-TRX

The synthesis of 18F-TRX required the novel 1,2,4-trioxolane precursor reagent TRX-amine in which an amine-bearing side chain has been introduced at a bridgehead position on the adamantane ring (Scheme ). This reagent was synthesized in eight steps, based on the general approach described for the synthesis of artefenomel[25] and related antimalarial trioxolanes. Full details regarding the synthesis of TRX-amine and 19F-TRX analytical standard are provided in the Supporting Information.

Scheme 1

: Radiosynthesis of 18F-TRX

The radiosynthesis of 18F-TRX began with the automated preparation of 18F-N-succinimidyl 4-fluorobenzoate. Using an ELIXYS automated radiosynthesizer, 18F-SFB was prepared in 75 min to a decay-corrected radiochemical yield of approximately 70%.[28,29] For the coupling reaction, 20 mCi of 18F-SFB was added to TRX-amine (5 mg of a formate salt) and 10% (v/v) DIPEA in anhydrous DMF (1 mL). The reaction was stirred at 40 °C for 30 min. The reaction produced only one major radioactive peak, which comigrated with the 19F-TRX standard (Figure A). The crude reaction was purified using semipreparative HPLC (1:10 CH3CN:H2O to 19:1 CH3CN:H2O over 20 min) to obtain the radiotracer 18F-TRX to a decay-corrected radiochemical yield of 67 ± 7.2%. The purity of the compound was verified by reinjection on semiprep HPLC (Figure B). 18F-TRX was concentrated, and immediately reconstituted for additional in vitro or animal studies (see the Experimental Section). The specific activity of 18F-TRX was calculated to be 0.052 ± 0.02 Ci/μmol (Figure S1).

Figure 2

Synthesis and isolation of 18F-TRX. (A) Representative semipreparative rad-HPLC traces showing, from top, the UV spectrum at 254 nm for TRX-amine starting material, the UV spectrum at 254 nm for the pure 19F-TRX analytical standard, the UV spectrum at 254 nm for complex radiochemistry reaction mixture at 40 min, and the rad spectrum of the complex radiochemistry reaction mixture at 40 min. Peaks of interest are labeled with the retention time in minutes. (B) Semipreparative rad-HPLC traces of the reinjected fraction isolated after purification. The retention time in minutes of the peak of interest representing 18F-TRX is labeled. (C) A rad spectrum collected 30 min after incubation of 18F-TRX in aqueous ammonium Fe(II) sulfate (30 mg/mL) showing disappearance of the peak corresponding to 18F-TRX and appearance of a new peak with a different retention time.

In Vitro Studies Show That TRX Analogues React with Fe(II) and Cross-Link Cellular Proteins

To confirm that 18F-TRX retains reactivity with Fe(II), the purified compound was exposed to aqueous ammonium Fe(II) sulfate (FAS, 30 mg/mL). After 30 min, 18F-TRX was entirely consumed and cleanly converted to a single radioactive byproduct as observed on reverse phase rad-HPLC (Figure C). This result indicates that the reactivity of 18F-TRX with Fe(II) is highly regioselective, as was expected based on earlier mechanistic studies[26] of the antimalarial compounds on which 18F-TRX is based. To confirm that the 18F-bearing side chain in 18F-TRX becomes sequestered in cells following reaction with Fe(II), we prepared the nonradioactive “clickable” probe HC2-TRX (Figure A) in which an alkyne function replaces the fluorine atom in 18F-TRX. Fluorescence cell imaging of PC3 cells treated with HC2-TRX followed by fixing and copper(I)-catalyzed alkyne–azide cycloaddition (CuAAC) reaction with Alexa488-azide were consistent with irreversible labeling of cytoplasmic targets (Figure B). An in-gel TAMRA fluorescence analysis confirmed the promiscuous labeling of proteins in PC3 cells treated with HC2-TRX (Figure C). Moreover, pretreating PC3 cells with ferric ammonium citrate (FAC) or the iron chelator desferrioxamine (DFO) prior to HC2-TRX revealed more and less labeling, respectively, consistent with iron-dependent activation. These findings with HC2-TRX indicate that the analogously positioned 18F atom in 18F-TRX would similarly become sequestered in cells in an Fe(II)-dependent fashion, as designed.

Figure 3

Cellular probe HC2-TRX promiscuously labels proteins in PC3 cells. (A) Structure of cellular “clickable” probe HC2-TRX. (B) Representative fluorescence images of PC3 cells treated with HC2-TRX for 6 h and visualized by CuACC reaction with Alexa488-azide (green, FITC channel); cell nuclei are visible in the DAPI channel (blue). Scale bars denote 50 μm. (C) In-gel TAMRA fluorescence analysis reveals promiscuous protein labeling following incubation of PC3 cells with HC2-TRX. Enhanced labeling is observed when PC3 cells were pretreated (2 h) with the Fe(III) source FAC (500 μM) before treatment with HC2-TRX. Conversely, PC3 cells pretreated with the iron chelator DFO (300 μM) showed reduced labeling intensity. Coomassie (Coo) staining is used as a loading control.

18F-TRX Biodistribution Is Fe(II)-Dependent in Normal Mouse Tissues

To determine the biodistribution of 18F-TRX in clinically relevant mouse models, the radiotracer was injected into immunocompetent C57Bl/6J mice and studied over time. Region of interest analysis of a dynamic PET/CT acquisition from 0 to 60 min postinjection revealed several enlightening trends. First, 18F-TRX rapidly cleared from the mediastinal blood pool at a rate ∼60 times faster than what we observed from a classic PK assessment of 19F-TRX (Figure A and Figure S2A). Moreover, 18F-TRX rapidly accumulated in liver from 0 to 60 s, suggesting that the radiotracer may be metabolized and cleared through this organ (Figure B,C) as is the case for artefenomel in humans.[30] Kidney accumulation of the radiotracer was also observed, but the uptake was significantly lower than liver. Moreover, kidney uptake plateaued within 30 s postinjection, suggesting that (as with artefenomel) renal clearance is not the dominant mechanism of clearance of radiotracer. Focal uptake in a region of the small intestine was also observed early after radiotracer injection, and the accumulation steadily increased from 400 to 3600 s (Figure S2B). This observation is also consistent with a model of hepatobiliary excretion. To study radiotracer distribution over a broader window of time, ex vivo biodistribution studies were conducted at 30, 60, and 90 min postinjection of 18F-TRX in a separate cohort of mice (Figure D). These studies generally corroborated the PET findings, showing that 18F-TRX uptake was dominant in the liver, components of the small intestine, and the kidneys. Moreover, radiotracer uptake generally increased from 30 to 90 min in these tissues. Radiotracer uptake was also observed in the spleen, pancreas, stomach, large intestine, and lungs. Uptake in the brain was very low, consistent with previous observations that other TRX conjugates do not cross the blood–brain barrier.[31]

Figure 4

An in vivo assessment of the biodistribution of 18F-TRX in tumor naïve immunocompetent mice. (A) Plasma concentration–time curve for 19F-TRX administered via tail vein injection in C57Bl/6J mice revealing a plasma half-life of ∼30 min. (B) Time activity curves derived from region of interest analysis of a 1 h dynamic PET scan reveal radiotracer biodistribution in vivo. The radiotracer rapidly clears from the blood (red) with a calculated serum half-life of about 25 s. The radiotracer is also sequestered within 60 s by the liver (tan) and kidneys (blue), with no additional accumulation from 60 to 3600 s. Liver uptake is significantly higher than what is observed in kidneys. Also, 18F-TRX accumulation in the small intestine (gray) steadily increases from 0 to 3600 s, consistent with a model of hepatobiliary clearance for the radiotracer. Little uptake is observed in the muscle (green). Figure S2C shows the location of the manually drawn regions of interest. (C) Maximum intensity projections sampled serially over short time frames from the dynamic acquisition show the biodistribution of the radiotracer in normal tissues over time. A diffuse signal is observed at early time points, which gradually consolidates into the liver, kidney, and small intestine. (D) Biodistribution data acquired at 30, 60, and 90 min postinjection of 18F-TRX show continuous accumulation of the radiotracer in many abdominal organs. The highest uptake was observed in the liver, kidneys, and small intestine. Radiotracer accumulation was low in the blood pool and muscle, as expected from the MIPs. (E) A PET/CT shows the biodistribution of 18F-TRX in components of the gastrointestinal tract after dissection from a mouse. Prominent uptake was observed in the duodenum and jejunum, while comparatively lower radiotracer uptake was noted in the stomach, ileum, cecum, and large intestine. Liver, kidneys, and muscle are included for perspective on the upper and lower bound of radiotracer uptake in tissues.

To better understand the localization of the radiotracer within regions of the small intestine, components of the GI tract were excised from a representative mouse 60 min postinjection, and radiotracer biodistribution was assessed with PET/CT (Figure E). This study suggested that radiotracer uptake was predominant in the duodenum and jejunum. Comparatively lower uptake was observed in the ileum, cecum, and a truncated segment of the large intestine (purged manually to remove fecal matter).

We next tested whether 18F-TRX biodistribution was impacted by exogenous treatments designed to alter tissue concentrations of ferrous iron. 20 minutes prior to i.v. injection of 18F-TRX, immunocompetent mice were treated with an i.p. bolus of PBS, ferric ammonium citrate (FAC, 20 mg/kg), desferrioxamine (50 mg/kg, DFO), deferiprone (50 mg/kg, DFP), or a mixture of FAC (20 mg/kg) precomplexed with desferrioxamine (50 mg/kg, FAC + DFO). The use of a ferric iron source ensures that bioconversion to ferrous iron by normal cellular processes is a necessary prelude to reaction with 18F-TRX. The biodistribution of the radiotracer at 60 min postinjection was strikingly different between treatment arms on PET/CT (Figure A). For instance, while 18F-TRX most visibly accumulated in the liver of mice treated with PBS, FAC treatment elevated uptake in the liver, small intestine, and gall bladder. Iron depleting strategies (e.g., DFO, FAC + DFO, DFP) reduced radiotracer uptake in the liver, while clearly redistributing 18F-TRX to components of the small intestine. Quantitative biodistribution studies showed a statistically significant increase in 18F-TRX uptake in the liver, spleen, pancreas, and duodenum of mice treated with FAC versus those treated with PBS (Figure B and Figure S3). Moreover, treatment with iron chelating agents significantly reduced 18F-TRX uptake in virtually all tissues, with the notable exceptions of the duodenum, ileum, cecum, and large intestine in the FAC + DFO treatment arm. In summary, these data demonstrate that 18F-TRX biodistribution is substantially influenced by exogenous treatments designed to modulate tissue concentrations of ferrous iron.

Figure 5

18F-TRX biodistribution is substantially altered by exogenous treatments that change intracellular iron homeostasis. (A) (top) A timeline outlining the sequence of injections in immunocompetent C57Bl/6J mice prior to PET/CT and biodistribution studies. (bottom) Representative maximum intensity projections (MIPs) showing the biodistribution of 18F-TRX in mice from each treatment arm from 50 to 60 min postinjection. Radiotracer uptake was predominantly in the liver of mice pretreated with PBS, while pretreatment with ferric ammonium citrate (FAC) augmented radiotracer uptake in nearly all organs, including the liver and components of the gastrointestinal tract. Pretreatment with various iron depleting agents, including desferrioxamine (DFO), FAC complexed with DFO, and deferiprone (DFP), generally reduced radiotracer uptake in organs and accelerated clearance. The MIPs, while inherently semiquantitative, were derived from decay-corrected PET data with scale bars adjusted to the range 0–50% ID/g to enable gross comparison. (B) Biodistribution data collected at 60 min postinjection show the percent changes in radiotracer uptake for selected organs in each treatment arm compared to mice receiving PBS. Relative increases in radiotracer uptake due to FAC treatment were observed in nearly all organs, while iron depleting treatments generally reduced organ uptake of the radiotracer. Treatment with FAC + DFO enhanced radiotracer uptake in components of the gastrointestinal tract, which may reflect accelerated clearance of the radiotracer.

18F-TRX Detects Tumor Xenografts Sensitive to Trioxolane Prodrugs

We next evaluated if 18F-TRX can detect human tumors derived from cell lines previously shown to harbor sensitivity to TRX-based iron(II)-activatable prodrugs.[32] Biodistribution studies were first conducted in intact male nu/nu mice bearing subcutaneous PC3 xenografts at 30, 60, and 90 min postinjection of 18F-TRX. The uptake of the radiotracer steadily increased from 30 to 90 min postinjection. Radiotracer levels in the tumor significantly exceeded blood and muscle at 90 min postinjection with a tumor to blood ratio of 1.97 ± 0.4 and a tumor to muscle ratio of 1.90 ± 0.3 (Figure A and Figure S4A,B).

Figure 6

18F-TRX detects tumor tissue in vivo in genetically and pathologically diverse cancer models. (A) Biodistribution data acquired at 30, 60, and 90 min postinjection of 18F-TRX in male nu/nu mice with subcutaneous PC3 xenografts. The radiotracer uptake in the tumor continually increases from 30 to 90 min, consistent with a reactivity-based mechanism of action. Moreover, radiotracer uptake in the tumor exceeds the level observed in blood and muscle at 90 min postinjection, two standard reference compartments for background radiotracer accumulation. The human prostate cancer model PC3 was prioritized as it was previously shown to be highly sensitive to an Fe(II)-sensitive TRX prodrug bearing a chemotherapeutic payload.[32] *P < 0.01 compared to blood and muscle. Figure S4 shows the biodistribution values for the entire repertoire of tissues from this animal cohort as well as the tumor to normal tissue ratios. (B) Biodistribution data acquired 90 min postinjection of 18F-TRX shows radiotracer uptake in tumor exceeding background for PC3 tumors implanted in the renal capsule (rcPC3), and subcutaneous EKVX and U251 tumors (scEKVX, scU251). Figure S5 shows a MR image highlighting the tumor burden in renal capsule. Figure S6 shows the complete biodistribution data sets, and the tumor to normal tissue ratios. (C) PET/CT imaging data showing uptake of 18F-TRX in tumor and normal tissues for mice bearing subcutaneous U251 tumors. The data were acquired at 90 min postinjection. (D) H&E (left) and digital autoradiography (right) showing 18F-TRX distribution within a representative section of U251 tumors. 18F-TRX appears to be present in all regions of the slice, with the highest relative uptake appearing to colocalize with the area of densest cellularity on H&E. (E) (left) A photograph of the surgically excised whole prostate (Pr.), a piece of muscle from the hindlimb (Mu.), and the seminal vesicles (SV) of a 10 month old Pb-Cre:Ptenfl/fl mouse with fully invasive adenocarcinoma. Cysts extending from the anterior prostate are evident by eye. (middle) A volume rendered CT image of the tissues acquired on a small animal PET/CT. (right) A volume rendered PET/CT image of 18F-TRX uptake in the tissues acquired 90 min postinjection of 18F-TRX. The image clearly shows relatively higher accumulation of radiotracer in the diseased prostate compared to muscle or seminal vesicles. 18F-TRX was excluded from the cysts, as expected.

We further tested if 18F-TRX can detect PC3 tumors implanted in the renal capsule, which provides a better vascularized environment than subcutaneous implants. Biodistribution studies conducted 90 min postinjection of 18F-TRX showed equivalent radiotracer uptake in a PC3 tumor embedded in the renal capsule compared to the extent of uptake in subcutaneous PC3 tumors (Figure B and Figure S5). 18F-TRX uptake was also significantly higher than background (blood and muscle) in subcutaneous EKVX and U251 tumors, two models of human lung adenocarcinoma and glioblastoma, respectively (Figure B and Figure S6). The U251 tumor, the model with the highest 18F-TRX uptake, was visually obvious on small animal PET/CT (Figure C). Ex vivo analysis of the spatial distribution of the radiotracer in U251 tumors showed that 18F-TRX was well-distributed through the tumor xenograft, with the regions of highest uptake appearing to have the densest cellularity on H&E (Figure D). Collectively, these data show that genetically and pathologically diverse models of human cancer harbor high avidity for 18F-TRX in vivo.

We next conducted a pilot imaging study in a genetically engineered mouse model of prostate cancer.[33] A 10 month old Pb-Cre:Ptenfl/fl mouse with invasive adenocarcinoma was treated with 18F-TRX, and after 90 min, the whole prostate was resected post mortem and imaged with PET/CT. Radiotracer uptake was visually higher in the prostate tissue compared to seminal vesicles and muscle (Figure E). Moreover, 18F-TRX uptake was predominant in the prostate tissue, and not observed in the cysts that routinely develop in this disease model.

Conclusion

Here we reported the design, synthesis, and characterization of 18F-TRX, a reactivity-based PET radiotracer that enables quantitative imaging of labile Fe2+ in living animals, including its application to several clinically relevant human and mouse cancer models. Based on an antimalarial with Fe(II)-dependent pharmacology, 18F-TRX can react with Fe2+ and become sequestered in tissues at the site(s) of its Fe(II)-promoted activation in vivo. Biodistribution studies and ex vivo imaging revealed high levels of radiotracer uptake in liver and small intestine that could be altered by pretreatment with bioavailable iron sources (FAC) or iron chelators (DFO). Furthermore, 18F-TRX was capable of detecting elevated Fe2+ levels in tumor compared to blood pool and muscle in PC3, EKVX, and U251 xenograft models and in a genetic prostate cancer model. These results suggest that labile Fe2+ represents an actionable analyte for cancer imaging with PET in animals.

Experimental Section

Radiosynthesis of 18F-TRX

18F-SFB was prepared in 75 min to a decay-corrected radiochemical yield of approximately 70%. 18F-SFB was transferred from the ELIXYS on a C18 Sep-Pak cartridge, and the 18F-SFB was eluted from the cartridge using neat CH3CN. The acetonitrile was removed under vacuum and a gentle stream of N2(g), and to the 18F-SFB (20 mCi) was added TRX-amine (5 mg, as formate salt) and 10% (v/v) DIPEA in anhydrous DMF (1 mL). The reaction was stirred at 40 °C. Reaction progress was monitored by rad-HPLC, and the reaction was terminated at 30 min, as the coupling of 19F-SFB to TRX-amine under analogous conditions was complete at 30 min. The crude reaction was purified using semipreparative HPLC (1:10 CH3CN:H2O to 19:1 CH3CN:H2O over 20 min) to obtain the radiotracer 18F-TRX to a decay-corrected radiochemical yield of 67 ± 7.2%. The purity of the compound was verified by reinjection on semiprep HPLC.

Cell Culture and Cell/Protein Labeling Experiments

The PC3, EKVC, and U251 cell lines were obtained from ATCC (Manassas, VA) and cultured in Dulbecco’s modified Eagle’s medium (VWR) supplemented with 10% fetal bovine serum (Gemini Bio) and 1× penicillin/streptomycin (Life Technologies). Cells were cultured at 37 °C supplemented with 5% CO2.

For the fluorescence cell imaging experiment, PC3 cells were cultured in a 96-well Greiner μClear plate until reaching 75% confluence. Media was removed and replaced with media containing trioxolane HC2-TRX (20 μM, 0.1% DMSO) or vehicle (0.1% DMSO). After 6 h, cells were washed with PBS twice and then fixed with 4% paraformaldehyde for 10 min at room temperature (rt). Cells were washed with PBS twice and incubated with PBS supplemented with 0.1% Triton X-100 for 5 min. Cell were then washed with PBS three times, then incubated for 1 h with click master mix (5 mM sodium ascorbate, 2 mM THPTA, 500 uM CuSO4, 10 uM Alexa488-azide), and then washed with PBS three times. Cells were then treated with PBS containing Hochest nuclear stain for 10 min, washed with PBS twice, and imaged with an IN Cell 6500 automated cell imager at 40× magnification.

For the protein labeling and in-gel fluorescence experiment, PC3 cells were cultured in 6-well plates until reaching 75% confluence. Media was then removed and replaced with fresh media or fresh media containing FAC (500 μM) or DFO (300 μM). After 2 h, media was removed and replaced with media containing HC2-TRX (20 μM, 0.1% DMSO) or vehicle (0.1% DMSO). After 6 h, cells were washed with PBS twice then incubated in cold PBS for 5 min. Cells were collected using a plastic cell scraper and pelleted. The pellet was resuspended in 100 μL of cold 0.1% NP40 lysis buffer (100 mM Hepes, 150 mM NaCl, 0.1% NP40, pH 7.5) with 5 mg/mL EDTA free protease inhibitor (Roche) for 30 min on ice. The following lysate was centrifuged at 13 000 rpm for 10 min. Protein concentration was determined using Pierce BCA protein assay kit (Thermo Fisher) and subsequently normalized to 1 mg/mL. To 50 μL of this lysate was added the click reaction cocktail (final concentrations: 25 μM TAMRA-Azide, 1 mM CuSO4, 0.1 mM TBTA, 1 mM TCEP), and it was incubated at room temperature for 1 h in the dark.

Protein was precipitated by addition of 1 mL of cold methanol and cooled to −80 °C. Protein was pelleted by centrifugation at 14 000 rpm for 5 min at 4 °C. Methanol was decanted, and the protein pellet was washed with 1 mL of cold methanol and pelleted again as described. Methanol was decanted, and the pellet was resuspended in 50 μL of 1× laemmli buffer (Bio-Rad) supplemented with β-mercaptoethanol. Sample was boiled for 5 min, and 25 μg of protein was loaded onto Bolt 4–12% Bis-Tris Plus Gel (Thermo Fisher). Fluorescence was visualized by a ChemiDoc system and displayed in a grayscale. Following imaging, total protein was visualized by Coomassie stain.

Animal Experiments

To reconstitute 18F-TRX for animal studies, the probe was trapped on a C18 Sep-Pak cartridge, and eluted with a small volume of ethanol. Ethanol was removed at 50 °C under vacuum and a gentle stream of N2(g) to afford neat 18F-TRX. Formulation of 18F-TRX for in vivo studies proved challenging due to the very poor aqueous solubility of this material. After some experimentation, we adopted a formulation comprising 10% DMSO in a 20 mM aqueous sodium phosphate solution at pH 3 used previously by Charman and co-workers[25] for IV pharmacokinetic studies of artefenomel. A 5 mg/kg dose of 19F-TRX in 100 μL of this formulation was well-tolerated in mice when administered by tail vein injection.

All animal experiments were conducted under the approval from Institutional Animal Care and Use Committee (IACUC) at UCSF. Male nu/nu or C57BL6/J mice (4–6 weeks) were purchased from Charles River. All the mice were well-housed in the USCF with free access to food and water. Nu/nu mice were inoculated with 2–5 × 106 PC3 cells in a mixture of media (RPMI) and Matrigel (Corning) (1:1 v/v) subcutaneously into one flank. Tumors in mice were palpable within 3–4 weeks after the implanting. Male Rag2 RAGN12 mice (Taconic) were used for the renal capsule tumor implants. The mice were anesthetized with isofluorane (2–3%) and performed with a dorsal midline incision (0.5 cm). Via pressuring on the muscle wall, one kidney was pulled gently and carefully through the small incision. PC3 tumor cells (5 × 106 in 50 mL of PBS) were then injected into the pocket under the kidney capsule, which was lifted from the kidney parenchyma. The kidney was placed back to the body of mice, and then the skin incision was closed using 3 surgical sutures. Carprofen (5–10 mg/kg) was used to treat the mice for easing recovery. Mice were observed carefully over 24 h for the signs of postoperative bleeding, pain, and (or) other complications. After surgery, a 14 T Agilent small animal MRI was used to monitor the tumor progression for the following 7–14 days.

Small Animal PET/CT

18F-TRX (∼300 μCi) was injected via tail vein in 100–150 μL of 10% DMSO in a 20 mM aqueous sodium phosphate solution. For treatment studies, vehicle (100 μL PBS), FAC (20 mg/kg in PBS, 100 μL), both FAC (20 mg/kg in PBS, 100 μL) and DFO (50 mg/kg in PBS, 100 μL), DFP (30 mg/kg in PBS, 100 μL), or DFO (30 mg/kg in PBS, 100 μL) were injected intraperitoneally 20 min prior to the intravenous injection of 18F-TRX. The mice were anesthetized with 2–3% isoflurane, and imaged with a Siemens Inveon microPET/CT. For dynamic acquisitions, the mice were anesthetized prior to injection, and injected while positioned on the scanner bed. All imaging data were decay-corrected, reconstructed, and analyzed with AMIDE software. Maximum intensity projections (MIPs) were generated by AMIDE software. Regions of interest (ROI) were manually placed to calculate SUV data from the dynamic acquisitions.

Biodistribution Studies

Mice were euthanized with CO2(g) asphyxiation and dissected at dedicated time points postinjection. The blood and tissues were removed, washed, dried, and weighed. The activity of each tissue was measured with a γ counter. All data were decay-corrected. PRISM software was used to express a percentage (% ID/g) of the injected dose per gram of tissue.

Digital Autoradiography

Post mortem, tumors were flash frozen in OCT on liquid nitrogen. The tissue was sectioned with a microtome into 20 μm thickness slices and mounted on glass slides. The slides were exposed in a GE phosphor storage screen for 10 radionuclide half-lives. The phosphor screen was developed on an Amersham Typhoon 9400 phosphorimager. The images were processed using ImageJ software. H&E staining was performed by the Pathology core facility at UCSF.

Statistics

All statistical analysis was performed using PRISM v6.0 or ORIGIN software. An unpaired, two-tailed Student’s t test was used to determine statistically significant differences in the data. Changes at the 95% confidence level (P < 0.05) were reported as statistically significant.