Synthesis and biological evaluation of water-soluble progesterone-conjugated probes for magnetic resonance imaging of hormone related cancers

Biomedical imaging has traditionally focused on detecting abnormalities in gross anatomical structures. 1 However, there has been a recent increase in the interest of developing targeted agents that noninvasively provide information about physiological and pathological processes on a molecular level. 1 - 4 In this regard, steroid receptors have emerged as particularly attractive targets for molecular imaging due to their role in promoting the growth of breast, ovarian, uterine, and prostate cancers. 5 Previous attempts to image steroid receptors have focused on positron emission tomography (PET). 1 , 6 PET imaging of estrogen receptor (ER) in breast cancer and androgen receptor (AR) in prostate cancer has been successful. 7 - 10 However, imaging of progesterone receptor (PR) has been limited in humans due to rapid metabolism of the tracers by 20-hydroxysteroid dehydrogenase. 11 - 13 In addition, PET suffers from the disadvantages of low spatiotemporal resolution and requirements of a nearby cyclotron and radiochemistry facilities. 2 Magnetic resonance imaging (MRI), offers excellent spatiotemporal resolution without exposure to harmful radiation or the need for specialized radiochemistry equipment. 2 Although the sensitivity of PET imaging agents far surpasses that of any other modality, MRI contrast agents can be optimized to enhance the otherwise low sensitivity of these probes. 2 , 14 , 15 Therefore, MRI is an excellent alternative to PET for detecting steroid receptors, and in particular, PR. Several PR-targeted MRI contrast agents have been previously developed. 16 - 18 One of these agents referred to as ProGlo, preferentially accumulated and enhanced MR signal in tissues and tumors with high PR expression. 19 In contrast to the PR-targeted PET imaging agent, ProGlo may be less rapidly metabolized due to the steric hindrance from the Gd(III) chelate preventing binding to 20-hydroxysteroid dehydrogenase. 20 Due to the low solubility of ProGlo in aqueous media, in vivo delivery of this agent was limited to subcutaneous or intraperitoneal injection routes rather than intravenous injections. 19 Furthermore, the effect of hydrophobicity on nonspecific interactions with biomolecules and the correlation between hydrophobicity and toxicity have been well-documented in numerous drugs. 21 - 23 The association of ProGlo with toxicity when intraperitoneally injected into athymic nude mice was, therefore, not surprising, and is likely a consequence of its hydrophobic properties and nonspecific interactions, necessitating the development of water-soluble contrast agents. Minimizing lipophilicity in imaging agents has been reported to decrease background signal by reducing nonspecific interactions in tissues. 1 In addition, imaging breast cancers, as opposed to tumors that form in the peritoneal space, should be significantly improved by intravenous imaging agent delivery, which can take advantage of leaky vasculature to accumulate and remain inside those tumors specifically expressing PR. The current work describes the synthesis of a novel series of water-soluble PR-targeted MR probes. These new agents demonstrated activation of PR in vitro with lower cytotoxicity than ProGlo. Higher cellular Gd(III) accumulation was observed in comparison to a non-targeted agent, particularly in PR(+) cells. Finally, these agents preferentially enhanced signal intensity in PR(+) tumors compared to PR(−) tumors and were not associated with toxicity in preliminary in vivo studies.

All click chemistry reactions were run in 1:1 methanol:water or 1:1 DMSO:water solutions. The steroid derivative (1 eq) and Gd(III) or Eu(III) chelate (1.1 eq) were dissolved and N 2 gas was bubbled through the solution to remove oxygen. Sodium ascorbate (1 eq), CuSO 4 (0.167 eq), and [(1-benzyl-1H-1,2,3-triazol-4-yl)methyl]amine (TBTA) (0.02 eq) were added and the reaction mixture was stirred at 60°C for 24 hours. The crude residue was purified by reverse phase preparative HPLC using a ramp from 0 to 100% B over 20 minutes. Analytical HPLC traces of the purified complexes are given in the Supporting Information . HRMS (ESI) m/z [M + H] + observed: 1034.4231, calculated: 1034.4267 for C 45 H 68 GdN 8 O 10 . HRMS (ESI) m/z [M + H] + observed: 977.4066, calculated: 977.40524 for C 43 H 65 GdN 7 O 9 . HRMS (ESI) m/z [M + H] + observed:1049.4078, calculated: 1049.40775 for C 45 H 66 GdN 8 O 11 . HRMS (ESI) m/z [M + H] + observed:1031.4252, calculated: 1031.4256 for C 45 H 68 EuN 8 O 10 . HRMS (ESI) m/z [M + H] + observed: 974.4058, calculated: 974.40416 for C 43 H 65 EuN 7 O 9 . HRMS (ESI) m/z [M + H] + observed:1045.4074, calculated: 1045.40487 for C 45 H 66 EuN 8 O 11 . Solutions of 1a, 2a, 3a, 4a, 5a , and 6a were prepared in 500 μL of Millipore water for T 1 and T 2 acquisition. T 1 and T 2 relaxation times were measured on a Bruker mq60 NMR analyzer equipped with Minispec V2.51 Rev.00/NT software (Billerica, MA, USA) operating at 1.41 T (60 MHz) and 37°C. T 1 relaxation times were measured using an inversion recovery pulse sequence using the following parameters: 4 scans per point, 10 data points for fitting, monoexponential curve fitting, phase cycling, 10 ms first pulse separation, and a recycle delay and final pulse separation ≥ 5 T 1 . The Gd(III) concentration of each solution was determined using ICP-MS. The inverse of the relaxation time (1/ T 1 , s −1 ) was plotted against Gd(III) concentration (mM) and fitted to a straight line with R 2 > 0.99. The slope of the fitted line was recorded as the relaxivity, r 1 . Approximately 1 mg of each compound was dissolved in 1 mL of a 1:1 mixture of water:1-octanol. After shaking the sample tube vigorously for 30 seconds, the tube was placed on a rotator for gentle mixing over 4 hours. The tube was then removed from the rotator and allowed to sit for 10 hours to ensure complete separation of the aqueous and organic phases. 50 μL was removed from each layer and subjected to ICP-MS to determine the Gd concentration in each layer. The partition coefficient was calculated from the following equation: log 10 P = log 10 ( C o / C w ), where log 10 P is the logarithm of the partition coefficient, C o is the concentration of Gd in the 1-octanol layer, and C w is the concentration of Gd in the water layer. The Eu(III) complexes were dissolved in D 2 O and H 2 O. The emission was monitored at 614 nm with excitation at 395 nm on a Hitachi F4500 Fluorescence Spectrophotometer operating in phosphorescence lifetime mode. Twenty-five scans were averaged and fit to a monoexponential decay (R 2 > 0.98) to give the phosphorescent lifetimes which were entered into this equation (corrected for one amide oscillator): q = 1.0 ( k H2O - k D2O - 0.25 - 0.075). 24 , 25 The progesterone receptor A ligand binding domain (amino acids 675–933) fused to GST (PR-LBD-GST; 80 nM), a fluorescently tagged PR ligand (fluoromone green PL; 4 nM), and either progesterone, ProGlo, 1a , or 3a (several concentrations) were incubated in PR screening buffer with 4 mM dithiothreitol (DTT) in a total volume of 100 μL for 4 hrs at room temperature according to the manufacturer's protocol (Invitrogen, Carlsbad, CA, USA). Each sample was measured using the Beacon 2000 fluorescence polarization analyzer (Invitrogen, Carlsbad, CA) located in the Northwestern University Keck Facility. The machine was used in static mode, batch blank, no delay, with an average of 1 read per cycle, at 22°C. A sample containing only buffer and PR-LBD-GST with no fluorescent PL was used as the blank to eliminate background signal from the protein or buffer. A sample with no competitor was used to determine 100% binding capacity of the PR-LBD-GST for the PL ligand. Curve fitting and error calculation was performed using Prism ® software from GraphPad™ Software, Inc (La Jolla, CA). Dulbecco's modified phosphate buffered saline (DPBS), media, sera, and dissociation reagents were purchased from Invitrogen (Carlsbad, CA). Cell culture consumables (flasks, plates, etc.) were purchased from VWR (Radnor, PA). Charcoal dextran stripped FBS was purchased from Atlanta Biologicals (Lawrenceville, GA). MDA-MB-231 cells were cultured using phenol red free α-MEM (modified to contain 20 ng/mL insulin) supplemented with 10% FBS (characterized) or with 10% charcoal dextran stripped FBS. T47D cells were cultured using phenol red free RPMI 1640 (modified to contain 1.0 mM sodium pyruvate, 1.0 mM HEPES, and 4.5 g/L glucose) supplemented with 10% FBS or 10% charcoal dextran stripped FBS. Prior to all experiments, cells were plated in the appropriate media containing non-charcoal dextran stripped media. After plating, this media was replaced with media containing the stripped FBS and allowed to sit for 24 hours at which point the media was replaced with fresh stripped media and the cells were allowed to sit for another 24 hours prior to beginning the experiment. MDA-MB-231 and T47D cells were harvested by incubation with 0.25% TrypLE for 10 minutes at 37°C in a 5.0% CO 2 incubator. All incubations were carried out at 37 °C in a 5.0% CO 2 incubator unless otherwise specified. After cell harvesting, an aliquot (15 or 30 μL) of the cell suspensions were mixed with Guava ViaCount reagent (final sample volume of 150 μL) and allowed to stain at room temperature for at least 5.0 minutes (no longer than 20 minutes). Stained samples were then vortexed for 10 seconds, after which cells were counted and percent cell viability determined via manual analysis using a Guava EasyCyte Mini Personal Cell Analyzer (PCA) and ViaCount software module. For each sample, 1000 events were acquired with dilution factors that were determined based upon optimum machine performance (∼ 25 – 70 cells/μL). Instrument reproducibility was assessed daily using GuavaCheck Beads and following the manufacturer's suggested protocol using the Daily Check software module. Contrast agents were dissolved in the appropriate media (containing stripped FBS) for each cell line (T47D and MDA-MB-231) at concentrations of 2, 1, 0.5, 0.25, and 0.125 mM of contrast agent. For concentration dependent cellular uptake, cells were incubated with 2, 1, 0.5, and 0.25 mM of each contrast agent for four hours. For the time dependent cellular uptake, cells were incubated with 0.125 mM contrast agent for 1, 4, 10, and 24 hours. After incubation, the media was removed, and the cells were rinsed twice with PBS and trypsinized. An aliquot was used for cell counting and the remaining portion was analyzed for Gd(III) content by ICP-MS. Each condition was done in triplicate. T47D and MDA-MB-231 cells were incubated with 0.250 mM of contrast agent (in the appropriate media containing stripped FBS) for 10 hours. After incubation, the cells were rinsed twice with PBS, trypsinized, pelleted (500 × g for 5 minutes), and resuspended in PBS. A portion was removed for cell counting, and the rest was pelleted (500 × g for 5 minutes) and the PBS removed. Cytoplasmic and nuclear fractions were extracted using a NE-PER Nuclear and Cytoplasmic Extraction Kit according to manufacturer protocol (Pierce, Rockford, IL). The remaining pellet after cytoplasmic and nuclear extraction was assumed to consist mainly of membrane. Each fraction was analyzed for Gd(III) content by ICP-MS. The CellTiter 96 ® AQ ueous Non-Radioactive Cell Proliferation Assay (Promega, Madison, WI) was used to measure cell viability. Cells were plated at 5000 cells/well in 96 well plates and maintained in media containing stripped serum for 48 hours before beginning the experiment. ProGlo, progesterone, 1a, 2a, 3a , and 4a were dissolved in varying concentrations in media. After 24 hours of incubation, the assay was run according to manufacturer protocol. Absorbance was measuring using a Biotek Synergy4 microplate reader in the High Throughput Analysis Facility at Northwestern University. T47D cells were grown in Phenol-free medium, cells were trypsinized and plated in 24-well plate (50,000 cells/ well). Incubation of cells with the pPRE-luciferase plasmid (100 ng/well, construct provided by Dr. Ken Korach, NIEHS, NIH), RSV-β-galactosidase (100 ng/well, provided by Dr. William T. Beck, University of Illinois at Chicago), and Lipofectamine 2000 (1 μL per well, Invitrogen, Carlsbad, CA) in Opti-MEM was performed overnight at 37°C inside a humidified incubator. The cells were treated with 1a, 2a, 3a , and 4a for an additional 24 hours. To measure luciferase production, cells were lysed in 100 μL GME buffer (25mM glycylglycine at pH 7.8, 15 mM MgSO 4 , 4mM EGTA, 1mM dithiothreitol, and 1% Triton X-100) and lysates were added to assay buffer (GME buffer, 16.5 mM KPO 4 , 2.2 mM ATP, and 1.1 mM dithiothreitol). Luciferase substrate was injected followed by a 30 second read by a FLUOstar OPTIMA (BMG Lab Tech, Offenburg, Germany). LacZ activity (50 μL lysate) was measured from cleavage of ONPG. The sample results were normalized to β-galactosidase to account for transfection efficiency by dividing the sum of the luciferase activity by the sum of the β-galactosidase activity. Fold change was calculated by taking the luciferase value, dividing by the β-galactosidase activity, and then dividing by the solvent control, which was set equal to one. Total RNA was isolated using Qiagen RNA easy columns with on column DNAse added according to the manufacture (Qiagen, Valencia, CA). RNA samples (2 μg) were then primed with random hexamers and reverse transcribed with M-MLV Reverse Transcriptase (Promega, Madison, WI.) according to manufacturer's instructions. From the original RT reaction, 1 μL was subjected to PCR amplification in a 25 μL volume with Taqman Universal PCR SYBR Green Master Mix (Applied Biosystems, Foster City, CA) under the following conditions: 50°C hold 2 minutes, 95°C hold for 10 minutes, the 40 cycles of 95°C for 15 seconds, 65°C for 30 seconds, and 72°C for 1 minute. The primers used were designed using Integrated DNA technologies primer quest for ZBTB16 forward 5′TGTTTGAGATCCTCTTCCACCGCA3′ and Reverse 5′ TCTCCAGCATCTTCAGGCACTGTT3′ and normalized to GAPDH forward 5′ ATGGGGAAGGTGAAGGTCG3′ and reverse 5′GGGGTCATTGATGGCAACAATA3′. For octanol-water partition coefficients, relaxivity, and cell studies, ACS reagent grade nitric acid (70%) was added to solutions of the agent in water or 1-octanol, cell suspensions and media (for a 1.0:1.0 v/v sample:nitric acid) in 15-mL conical tubes and placed at 65°C for at least 4.0 hours to allow for complete sample digestion. For samples in 1-octanol, tubes were vented every 30 minutes due to buildup of pressure. Nanopure H 2 O and internal standard (either indium or multi-element) were added to produce a final solution of 3.0% nitric acid (v/v) and 5.0 ng/mL internal standard. Gd(III) standards were prepared at 0.10, 0.25, 0.50, 1.0, 5.0, 10, 25, and 50 ng/mL concentrations with 3.0% nitric acid (v/v) and internal standard final concentrations. ICP-MS was performed on a computer-controlled Thermo (Waltham, MA) X Series II Inductively Coupled Plasma Mass Spectrometer equipped with a CETAC 260 autosampler. Each sample was acquired using 1 survey run and 3 main (peak jumping) runs. The isotopes selected were 156,157 Gd, as well as 115 In and 165 Ho (as internal standards for data interpolation and machine stability). Female Balb/C athymic nude mice were acquired from Harlan (Indianapolis, IN) and housed under pathogen free conditions. All animal studies were conducted at Northwestern University in accordance with the National Institutes of Health Guide for the Care and Use of Laboratory Animals and established institutional animal use and care protocols. Due to low circulating estradiol levels in nude mice, a 17β-estradiol pellet (Innovative Research of America, Sarasota, FA, 70-day release, 0.72 mg/pellet) was implanted subcutaneously in the nape of the neck to ensure growth of estrogen-dependent T47D cells. Two to seven days later, T47D and MDA-MB-231 (1-2 × 10 6 ) cells were suspended in Matrigel (1:1 volume) and injected subcutaneously into the rear flank (T47D cells on the right side and MDA-MB-231 cells on the left). Mice were monitored for tumor growth every two to three days after injection of cells. Mice were imaged two to three weeks after xenografting when tumors were palpable with a minimum weight of 20 mg (determined after tumor removal post-imaging). Xenografted nude mice were injected intraperitoneally (n = 1 per compound) with each agent (0.15mmGd/kg) dissolved in 50 μL DMSO. During imaging, mice were maintained under anesthesia (1-3% isoflurane) but were allowed to wake up and recover between imaging time points. Tubing containing heated water was positioned under the mouse to keep a constant body temperature. All MR imaging was performed on a 89 mm bore size PharmaScan 7.05 T MR imager fitted with shielded gradient coils (Bruker BioSpin, Billerica, MA, USA) using a RF RES 300 1H 089/038 quadrature transmit receive volume coil (Bruker BioSpin, Billerica, MA, USA). Standard T 1 -weighted multislice multiecho (MSME) scans with fat suppression were used for imaging the uterus, ovaries, and xenografts: TR = 700 ms, TE = 10.635 ms, FOV = 35 × 35 mm, matrix size = 256 × 256, slice thickness = 1.0 mm, contiguous slices. Images were analyzed using ImageJ. Images were analyzed using ImageJ. Contrast-to-noise ratio (CNR) was calculated using the equation CNR = (SI Tissue - SI Muscle )/σ noise where SI Tumor is the signal intensity in the tumor, SI Muscle is the signal intensity in the muscle, and σ noise is the standard deviation of the noise. CNRs were averaged over two to three axial slices in which the tumors were clearly demarcated.

ProGlo was proven to be a successful PR-targeting MRI contrast agent in vitro and in vivo . 17 , 19 However, its insolubility in aqueous media resulted in a degree of toxicity and limited in vivo delivery. 19 In order to increase the solubility of ProGlo while attempting to retain its biological properties, the core structure of the targeting steroid and the Gd(III) chelate were maintained while the linker was modified. The structures of the new PR-targeted agents and non-targeted control agents used in this study are presented in Figure 1 . The synthesis of each water-soluble PR-targeted agent began with the coupling of 1,5-dibromopentane or 5-bromovaleric acid to 21-hydroxyprogesterone off the 21-hydroxyl group because PR is known to accommodate bulky substituents on the D ring of the progesterone. 17 , 26 , 27 Reaction with sodium azide yielded the azido derivative of these steroids ( Scheme 1 ). Using copper-catalyzed click chemistry ( Scheme 2 ), 28 , 29 these steroid derivatives were attached to alkyne-functionalized Gd(III) chelates, 5a and 6a , which were synthesized using modifications of previously reported procedures. 19 , 30 - 34 The longitudinal relaxivities of the PR-targeted agents and 4a are similar to those reported for clinically used contrast agents, about 4 – 5 mM −1 s −1 ( Table 1 ). Like most clinically used contrast agents, each of the PR-targeted agents has approximately one water molecule bound to the Gd(III), as determined by analysis of hydration number ( q ) on the Eu(III) analogues of these agents ( Table 1 ). For complex 2b , the coordination of one water molecule, rather than two, indicates that the triazole ring from the click reaction coordinates to the Gd(III), an effect that was previously reported. 33 For 3b and 1b , the q of one is due to the coordination of the amide group back to the metal. Octanol-water partition coefficients (logP) were measured to determine the hydrophobicity of these agents, which correlates to the cellular permeability and tissue distribution ( Table 1 ). 21 , 37 The negative logP value for 4a was characteristic of its high water solubility. The logP values of these water-soluble PR-targeted agents are between the values for ProGlo and 4a , indicating an intermediate hydrophobicity. The new water soluble agents were expected to have receptor binding affinities similar to ProGlo because the chemical modification occurred at the same position on the steroid. Compounds 1a and 3a were chosen for comparing the effect of the ether to an ester linker in terms of PR binding ( 2a was not tested because this molecule has the same linker as 1a ). The relative binding affinities (RBAs) of these agents and of ProGlo were measured ( Table 1 ). Both 1a and 3a have RBAs similar to ProGlo and about 100-fold less than progesterone. These results are consistent with previously published data that demonstrated approximately 100-fold lower binding affinity of ProGlo compared to unmodified progesterone. 17 Although 3a demonstrated a minimal decrease in binding affinity (likely due to the presence of the additional carbonyl), the RBAs of these agents indicate that the modifications made to the linker did not significantly affect their ability to bind PR as compared to ProGlo. To determine the cell permeability of the compounds, PR(+) (T47D) and PR(−) (MDA-MB-231) human breast cancer cells were incubated with 1a, 2a, 3a , and 4a . Time- and concentration-dependent experiments were performed to identify the optimal dose and incubation times of the agents ( Figure 2 ). Each agent demonstrated increased cellular Gd(III) accumulation with increased incubation concentrations in both cell lines. However, time-dependent accumulation was observed only for 2a in the PR(+) cells. The uptake of 1a in the PR(+) cells increased slightly after 1 hour and then stabilized, while the Gd(III) levels in PR(+) cells incubated with 3a had decreased by 24 hours in comparison to earlier time points. No time-dependence in uptake was seen in PR(−) cells as the Gd(III) levels remained constant at all time points or even decreased (for 3a ). The uptake of the steroid-conjugated agents was higher than that of 4a at all time points and cell lines, except for 3a , which showed equal or lower Gd(III) than 4a at the 24 hour time point in both cell lines. These results indicate that the presence of the steroid moiety encourages cellular uptake of the targeted contrast agents when compared to the non-functionalized Gd(III) chelate. Importantly, the Gd(III) accumulation of the PR-targeted agents was significantly higher in PR(+) than in PR(−) cells. While the uptake of 4a was higher in PR(+) cells than in PR(−) cells, the difference in uptake was lower than for the progesterone-conjugated agents (particularly 1a and 2a ). These data indicate that conjugation with progesterone increases the ability of Gd(III) contrast agents to associate with cells. In addition, the increased Gd(III) accumulation in PR(+) cells compared to PR(−) cells suggests that PR likely plays a role in the accumulation and retention of these progesterone-conjugated agents. Progesterone receptors bind to a region of DNA referred to as the progesterone response element (PRE). 17 This DNA element was ligated to the luciferase gene and used to verify the ability of the compounds to cross the cell membrane and interact with PR to form a functional transcription complex. The change in luciferase activity at increasing incubation concentrations of each PR-targeted contrast agent is shown in Figure 3a (left graph). The graph on the right in Figure 3 shows the change in luciferase activity when cells were incubated with either water or 10 nM progesterone as a positive control. As expected, the control compound 4a did not induce PRE-mediated transcription to any significant extent. However, incubation with each of the progesterone derivatives resulted in luciferase activity, indicating that these compounds entered the cells and bound to PR as agonistic ligands. Interestingly, 3a was the most effective agent despite its low cell uptake and lower relative binding affinity for PR in vitro compared to 1a and 2a . Compound 2a demonstrated the lowest efficacy of the three progesterone-conjugated agents. To further investigate the ability of the contrast agents to activate transcription, PR-regulation of the gene ZBTB16 was examined based on previous reports of its induction in T47D cells ( Figure 3b ). The increase in ZBTB16 transcription in response to 1a – 4a and progesterone was measured and normalized to GAPDH. Each progesterone-conjugated agent activated transcription after incubation at 10 nM, providing further evidence that these agents cross the cell membrane and bind to PR. Furthermore, the transcriptional activation demonstrated the same relative potency as the luciferase assays in that 3a demonstrated the highest activation of endogenous gene transcription. As expected, control agent 4a did not result in transcription of endogenous ZBTB16. To examine the cellular distribution of the water-soluble PR-targeted agents as compared to 4a and ProGlo, PR(+) and PR(−) cells incubated with each agent were subjected to a nuclear cytoplasmic fractionation kit ( Figure 4 ). The pellet remaining after isolation of the nuclear and cytoplasmic fractions was assumed to contain material that was primarily associated with cell membrane proteins. As expected, the majority of each agent was found in the cytoplasm of each cell type. However, the fraction of agent in the cytoplasm was lowest for ProGlo in both PR(+) and PR(−) cells. For 4a and the water-soluble PR-targeted agents, the fraction of agent in the cytoplasm was around 80 – 90% in both cell lines except for 3a , which showed a lower fraction in the cytoplasm of the PR(+) cells. The fraction of Gd(III) in the nucleus and membrane was similar for 4a and the water-soluble agents in both cell lines, although 3a had a higher presence in the pellet fraction in the PR(+) cells. ProGlo was found in high levels (relative to 1a – 4a ) in the pellet and in the nucleus in both cell lines. The majority of PR resides in the cytoplasm, but it was expected that incubation with the targeted contrast agents would result in translocation of a fraction of the total cytoplasmic receptor to the nucleus, leading to higher levels of Gd(III) in the nucleus after incubation with the targeted agents compared to 4a . While the luciferase assays and real-time PCR indicated that PR was activated, the difference in levels of progesterone-conjugated agents versus 4a that translocated to the nucleus was likely too low to detect using ICP-MS. However, the cell fractionation data indicate that in cells incubated with the water-soluble contrast agents, the majority of the Gd(III) associated with these cells is intracellular and not interacting within the membrane. While this was true for cells incubated with ProGlo, 20 – 30% of the total ProGlo was detected in the remaining membrane pellet fraction. The hydrophobicity of ProGlo likely increased nonspecific interaction with non-targeted biomolecules and could explain high levels of the agent in the membrane, as well as the higher levels in the nucleus compared to fractions from cells incubated with the water-soluble agents. To determine the toxicity profile of the agents, a MTS assay of cell viability was conducted after incubating T47D or MDA-MB-231 cells with varying concentrations of each agent for a 24 hour period. The toxicities of the water-soluble agents were compared to progesterone, ProGlo, and the non-targeted 4a , and the data are presented in Figure 5 . As expected, in PR(+) T47D cells, the water-soluble PR-targeted agents were less toxic than ProGlo at all concentrations except for the lowest concentration tested. In addition, progesterone was more toxic than the water-soluble agents at the highest concentration tested (2.5 mM), but reached similar levels at 1.25 mM and lower. Control agent 4a was non-toxic at all concentrations tested. In PR(−) MDA-MB-231 cells, all agents had similar toxicity profiles, including 4a . While the effect on cell viability in both cell types was similar for ProGlo and progesterone, Gd-DO3A and the water-soluble PR-targeted agents were surprisingly more toxic in the MDA-MB-231 cells than in the T47D cells. 3a was the most toxic of the water-soluble agents in both cell lines. To determine whether the water-soluble progesterone-conjugated agents would increase the contrast-to-noise ratio (CNR) of PR(+) tumors relative to PR(−) tumors in vivo, athymic nude mice with PR(+) and PR(−) xenografted tumors were injected intraperitoneally with 1a, 2a , or 3a dissolved in DMSO and imaged at 2 and 6 hours post-injection. Intraperitoneal injection was chosen over subcutaneous injection because it yielded more consistent results in previous studies, and the parent compound, ProGlo, has been more extensively studied for its biodistribution after intraperitoneal injection. 19 In addition, while these agents are water-soluble, they were dissolved in DMSO to compare back to previous in vivo imaging with the insoluble ProGlo and the control agent 4a . 19 After injection of each of the PR-targeted agents, the PR(+) tumor exhibited significantly higher CNR than the PR(−) tumor at 6 hours post-injection with 2a , 2 and 6 hours post-injection with 1a , and 2 hours post-injection with 3a ( Figure 6A , top graphs). At all other time points, the CNR in the PR(+) and PR(−) were not significantly different. This is in contrast to the CNRs after injection with non-targeted 4a , in which the CNR of the PR(+) and PR(−) tumors were not significantly different at any time point. 19 The uptake of these agents in the different tumor types can be partially attributed to PR-independent mechanisms, which would explain the presence of contrast agent in the PR(−) tumors. However, the increased CNR in the PR(+) compared to the PR(−) tumor is seen only with the progesterone-functionalized agents, indicating that the agents used here are targeting PR in vivo . Most likely, the higher accumulation in the PR(+) tumors is a result of accumulation and retention due to interaction with PR as previously reported 17 rather than PR-driven uptake of the agents. Analysis of the xenograft images revealed that the PR-targeted agents each increased the CNR in both tumors over the preinjection CNR levels, particularly after injection of 1a and 3a ( Figure 6A , bottom graphs). Injection with ProGlo was previously reported to similarly increase CNR several-fold in both tumors, while injection with Gd-DO3A resulted in only minimal increases in CNR over time. Furthermore, the fold changes in CNR (compared to the preinjection CNR) in the PR(+) tumors after injection with 1a and 3a were 16% and 34% higher, respectively, at the 2 hour time point than with ProGlo. 19 Injection with 3a resulted in the highest overall increase in CNR of the three agents, but 1a demonstrated a several fold increase in CNR along with higher CNR in the PR(+) tumor than the PR(−) tumor at both the 2 and 6 hour time points. These data provide evidence that the water-soluble PR-targeted probes function better as tumor imaging agents compared to non-functionalized 4a and they preferentially target PR(+) cancer cells over PR(−) cancer cells (in similar fashion to ProGlo). Of the three water-soluble agents tested, 1a provided an optimal balance between tumor uptake and specificity for PR(+) cells compared to PR(−) cells. Representative images of the xenografts at each time point are shown in Figure 6B . Importantly, there was no observed toxicity associated with mice that were intraperitoneally injected with these water-soluble PR-targeted agents. Intraperitoneal injection with ProGlo had previously been associated with toxicity (xenografted mice injected with ProGlo in this manner became lethargic after the 6 hour time point and did not survive to 24 hours). 19 The mice injected with the water-soluble agents, however, remained active and survived to 24 hours after injection with no changes in behavior.

Steroid receptors such as PR correlate with disease prognosis and therapeutic efficacy in breast cancer. 5 , 38 Receptor status is currently determined by immunohistochemistry assays of tumor biopsy samples, but noninvasive PR imaging agents would allow for improved molecular characterization, treatment decisions, and repeat analyses. 12 The hydrophobic PR-targeted ProGlo previously described demonstrated accumulation in, and consequent enhancement of CNRs in PR-rich organs and tumors in vivo . 19 Here, we have described the synthesis and biological testing of several new water-soluble ProGlo derivatives. These agents were associated with lower toxicity than ProGlo, but retained the ability to target PR in vitro and in preliminary in vivo experiments. ProGlo is preferentially taken up by and retained in PR(+) cells and enhances the signal intensity of PR-rich tissues in vitro and in vivo . However, this agent was associated with toxicity both in vitro and in vivo 19 due to its hydrophobic nature. 21 - 23 To decrease toxicity while maintaining PR targeting and biological activity, the structure of ProGlo was modified by using click chemistry to introduce a 1,2,3-triazole ring into the linker between progesterone and the Gd(III) chelate, which resulted in increased solubility. 1,4 disubstituted 1,2,3-triazole are known to possess a large dipole moment, and the nitrogen atoms act as hydrogen bond acceptors, which improves the solubility in water. 39 In addition, this triazole is spaced several carbons away from the steroid and did not interfere with receptor binding, as evidenced by the similar RBAs for these novel agents as compared to ProGlo. The induction of luciferase and endogenous progesterone-regulated ZBTB16 by each PR-targeted agent indicates that these agents crossed the cell membrane and bound to PR at biologically significant amounts. While 3a was not taken up well by cells, the ester may have hydrolyzed (due to the presence of esterases or via other mechanisms) to release 21-hydroxyprogesterone. This cleavage product would bind more strongly to PR and could explain the higher transcriptional activity observed after incubation with 3a . The water-soluble progesterone-conjugated agents were less toxic than ProGlo due to their hydrophilic nature. The reduced toxicity is partially explained by decreased nonspecific interactions with these water-soluble agents as compared to ProGlo, as demonstrated by the cell fractionation experiments. ProGlo was distributed almost evenly across all the compartments analyzed, while the water-soluble agents were primarily located in the cytoplasm, indicating more nonspecific interaction between Proglo and off-target biomolecules. Incubation with 3a resulted in Gd(III) accumulation in the membrane of the PR(+) cells, possibly due to ester hydrolysis and subsequent production of a Gd(III) chelate with a pentanoic acid tail that could insert in cell membranes. This ester cleavage might explain the high signal intensity surrounding the PR(+) tumor in the in vivo images taken 2 hours after injection with 3a , as well. In addition, ester hydrolysis would release 21-hydroxyprogesterone, which would enter the cells at high levels and could explain the higher cytotoxicity observed after incubation with this agent. Studies in tumor xenografts demonstrated that 1a, 2a , and 3a preferentially enhanced CNR in PR(+) tumors as compared to PR(−) tumors at 2 and/or 6 hours after injection. Injection of control agent 4a , by contrast, resulted in minimal changes in CNR at 2 and 6 hours post-injection and no difference between the PR(+) and PR(−) tumors. Of these PR-targeted agents, 1a was the most promising in that it resulted in higher CNR in the PR(+) tumor than the PR(−) tumors at both 2 and 6 hours post-injection. In addition, the fold change in CNR compared to preinjection levels was higher in the PR(+) tumor than the PR(−) tumor at both timepoints after injection. The accumulation of Gd(III) in these tumors may have been time-dependent. Injection with 1a and 3a increased CNR by 2 hours, but CNR had decreased by 6 hours. Injection with 2a , however, resulted in a slow increase in CNR over time. Interestingly, this pattern corresponded to the cellular accumulation in cells where 2a was the only agent tested that demonstrated time-dependent accumulation in PR(+) cells. Studies are ongoing to determine the optimal post-injection imaging time points for each of these agents. The changes in CNR in the PR(+) tumor after injection of the water-soluble PR-targeted agents were comparable to or greater than the changes seen after injection of ProGlo previously demonstrated. 19 The difference between CNRs of the PR(+) and PR(−) tumors was greater after injection of these water-soluble agents than after intraperitoneal injection of ProGlo, indicating that these agents possibly have more bioavailability due to their enhanced solubility. Finally, these water-soluble agents were associated with less in vivo toxicity than ProGlo. The enhanced solubility and bioavailability combined with the decreased toxicity might both be attributed to the less hydrophobic nature of these agents. Further in vivo investigations are being conducted to determine biodistribution of these agents and the optimal mode of injection. Recent reports have asserted that although there has been a focus on developing drugs that accumulate in cancer cells, this accumulation might actually prevent drugs from penetrating deep within tumors and decrease their efficacy. 40 - 42 Drugs that accumulate quickly within cells tend not to penetrate deeper areas of tumors, while less cell-permeable drugs travel further within the tumor and are more effective. 42 This indicates that decreasing lipophilicity of the agent would reduce cellular accumulation and allow for improved tumor penetration. The ability to inject water-soluble agents intravenously would also increase tumor penetration as the angiogenic vasculature of the tumor would likely leak the agent allowing for accumulation. For imaging agents, this property of tumor penetration would be essential in terms of measuring receptor levels in a tumor with heterogeneous receptor distribution. 43 The PR-targeted agents described here would likely penetrate tumors better than ProGlo and increase signal intensity of the entire tumor rather than the surface especially since it can only be administered intraperitoneally or subcutaneously. Thus, in future work, the water-soluble PR-targeted agents might prove to be excellent tumor targeted contrast agents in vivo . In conclusion, a series of water-soluble PR-targeted MRI contrast agents were designed and synthesized. These agents crossed cell membranes PR(+) T47D cells and activate PR at biologically relevant levels but exhibited lower in vitro and in vivo toxicity than the more hydrophobic ProGlo. Finally, these progesterone-conjugated agents enhanced CNR in PR(+) tumors as compared to PR(−) tumors in preliminary imaging studies in xenografts, warranting further in vivo investigations to prepare PR-targeted MR agents for noninvasive diagnosis of hormone related cancers and for elucidating the molecular pathways involved in these diseases.

Unless noted, materials and solvents were purchased from Sigma-Aldrich Chemical Co. (St. Louis, MO) and used without further purification. GdCl 3 ·6H 2 O and 1,4,7,10-tetraazacyclododecane (cyclen) were purchased from Strem Chemicals (Newburyport, MA) and used without further purification. Unless noted, all reactions were performed under a nitrogen or argon atmosphere. DMSO, N,N-dimethylformamide, and methanol were purified using a Glass Contour Solvent system. Deionized water was obtained from a Millipore Q-Guard System equipped with a quantum Ex cartridge (Billerica, MA). Thin-layer chromatography (TLC) was performed on EMD 60F 254 silica gel plates. Visualization of the developed chromatogram was performed by CAM stain and platinum stain. Standard grade 60 Å 230-400 mesh silca gel (Sorbent Technologies) was used for flash column chromatography. 1 H and 13 C NMR spectra were obtained on a Bruker 500 MHz Avance III NMR Spectrometer with deuterated solvent as noted. Electrospray ionization mass spectrometry (ESI-MS) spectra were taken on a Varian 1200 L single-quadrupole mass spectrometer. High resolution mass spectrometry data was acquired on an Agilent 6210 LC-TOF (ESI, APCI, APPI). Analytical reverse-phase HPLC-MS was performed on a Varian Prostar 500 system with a Waters Atlantis C18 column (4.6 × 250, 5 μm). This system is equipped with a Varian 380 LC ELSD system, a Varian 363 fluorescence detector, a Varian 335 UV-vis detector, and a Varian 1200L Quadrupole MS detector. Preparative runs were performed on a Waters Atlantis C18 column (19 × 250, 10 μm). Mobile phase consisted of water (solvent A) and HPLC-grade acetonitrile (solvent B). The synthesis and purification of ProGlo were performed as previously described. 17 A solution of 21-hydroxyprogesterone (300 mg, 1.36 mmol), 5-bromovaleric acid (111 mg, 0.613 mmol), diisopropyl carbodiimide (DIC) (122 μL, 0.786 mmol), and 4-(dimethylamino)pyridinium-4-toluenesulfonate (DPTS) (207 mg, 0.666 mmol) in anhydrous dichloromethane (1.1 mL) was stirred for 16 hours at room temperature. The reaction mixture was diluted in dichloromethane and washed with water three times. The organic layer was dried over sodium sulfate and concentrated by rotary evaporation. The crude residue was purified by flash chromatography with hexanes/ethyl acetate (1:1) as the eluent to afford 8 as a white solid (348 mg, 75%). 1 H NMR (500 MHz, CDCl 3 ) δ 5.65 (1H, s, 4-H), 4.68 (1H, d, J =17, COCH 2 O), 4.45 (1H, d, J =17, COCH 2 O) 3.36 (2H, t, J =6.5, CH 2 Br), 2.47 – 0.80 (complex, 26H), 1.12 (3H, s, CH 3 ), 0.63 (3H, s, CH 3 ); 13 C NMR (125 MHz, CDCl 3 ) δ 203.77, 199.80, 172.71, 171.17, 124.18, 69.34, 59.31, 56.41, 53.79, 51.04, 44.91, 38.81, 38.56, 35.94, 35.77, 34.16, 33.39, 32.98, 32.1, 32.2, 24.71, 23.62, 23.07, 21.22, 17.59, 13.43 ESI-MS m/z [M + H] + observed: 494.8, calculated: 495.2. To a solution of 8 (325 mg, 0.639 mmol) in anhydrous N,N-dimethylformamide (7 mL) was added sodium azide (415 mg, 6.39 mmol). The reaction mixture was heated to 65°C and stirred overnight. Excess sodium azide was filtered off and the solvent was removed by rotary evaporation. The crude residue was dissolved in ethyl acetate and washed with water three times. The organic layer was dried over sodium sulfate and concentrated followed by flash chromatography in hexanes/ethyl acetate (4:3) to give 10 (206 mg, 71%) 1 H NMR (500 MHz, CD 3 CN) δ 5.62 (1H, s, 4-H), 4.72 (1H, d, J =17, COCH 2 O), 4.53 (1H, d, J =17, COCH 2 O) 3.31 (2H, t, J =6.5, CH 2 N 3 ), 2.55 – 0.99 (complex, 26H), 1.17 (3H, s, CH 3 ), 0.64 (3H, s, CH 3 ); 13 C NMR (125 MHz, CD 3 CN) δ 204.49, 199.18, 173.04, 171.81, 123.86, 69.81, 59.25, 56.55, 54.21, 51.43, 44.95, 39.12, 38.60, 36.20, 35.94, 34.28, 33.36, 32.96, 32.52, 28.46, 24.79, 23.62, 23.07, 21.46, 17.35, 13.20. ESI-MS m/z [M + Na] + observed: 478.3, calculated: 478.3. A mixture of 21-hydroxyprogesterone (200 mg, 0.605 mmol), 1,5-dibromopentane (1.65 mL, 12.1 mmol), 40% KOH (300 μL), and tetrabutylammonium hydroxide (60 μL) was stirred for 16 hours at room temperature. The reaction mixture was diluted in dichloromethane and washed with water three times. The organic layer was dried over sodium sulfate and concentrated by rotary evaporation. The crude residue was purified by flash chromatography with hexanes/ethyl acetate (2:1) as the eluent to afford 7 as a colorless oil (125 mg, 43%). 1 H NMR (500 MHz, CDCl 3 ) δ 5.67 (s, 1H), 3.96 (q, J = 17.2 Hz, 2H), 3.41 (t, J = 6.4 Hz, 2H), 3.38 – 3.31 (m, 2H), 2.56 (t, J = 9.1 Hz, 1H), 2.40 – 2.09 (m, 6H), 1.96 (m, 1H), 1.90 – 1.74 (m, 5H), 1.72 – 1.16 (m, 14H), 1.12 (s, 3H), 1.05 – 0.86 (m, 2H), 0.65 – 0.58 (m, 3H). 13 C NMR (125 MHz, CDCl 3 ) δ 207.55, 198.50, 169.94, 122.92, 75.79, 70.34, 61.56, 57.43, 55.12, 52.51, 43.53, 37.54, 34.61, 32.85, 31.74, 31.47, 30.82, 28.69, 27.74, 23.77, 23.47, 21.81, 19.97, 16.34, 12.55. ESI-MS m/z [M + H] + observed: 481.1, calculated: 481.2. To a solution of 7 (125 mg, 0.261 mmol) in anhydrous N,N-dimethylformamide (7 mL) was added sodium azide (169 mg, 2.61 mmol). The reaction mixture was heated to 65°C and stirred overnight. Excess sodium azide was filtered off and the solvent was removed by rotary evaporation. The crude residue was dissolved in ethyl acetate and washed with water three times. The organic layer was dried over sodium sulfate and concentrated followed by flash chromatography in hexanes/ethyl acetate (2:1) to give 9 (78 mg, 68%). 1 H NMR (500 MHz, CD 3 OD) δ 5.61 (d, J = 4.9 Hz, 1H), 3.39 (t, J = 6.4 Hz, 2H), 3.25 – 3.17 (m, 4H), 2.60 (t, J = 9.1 Hz, 1H), 2.45 – 2.33 (m, 2H), 2.25 – 2.15 (m, 2H), 2.12 – 2.03 (m, 1H), 1.99 (m, 1H), 1.89 – 1.76 (m, 2H), 1.70 – 1.29 (m, 15H), 1.29 – 1.15 (m, 2H), 1.13 (d, J = 11.0 Hz, 4H), 1.04 – 0.87 (m, 2H), 0.60 (s, 3H). 13 C NMR (125 MHz, CD 3 OD) δ 210.81, 202.29, 174.92, 124.24, 78.00, 72.34, 62.71, 59.71, 57.39, 55.16, 52.43, 45.67, 39.99, 39.72, 36.81, 34.72, 33.90, 33.20, 30.15, 29.75, 25.52, 24.46, 24.17, 23.76, 22.15, 17.67, 13.96. ESI-MS m/z [M + Na] + observed: 464.2, calculated: 464.3.