Quantitative Imaging of Cell-Permeable Magnetic Resonance Contrast Agents Using X-Ray Fluorescence

Abstract

The inability to transduce cellular membranes is a limitation of current magnetic resonance imaging probes used in biologic and clinical settings. This constraint confines contrast agents to extracellular and vascular regions of the body, drastically reducing their viability for investigating processes and cycles in developmental biology. Conversely, a contrast agent with the ability to permeate cell membranes could be used in visualizing cell patterning, cell fate mapping, gene therapy, and, eventually, noninvasive cancer diagnosis. Therefore, we describe the synthesis and quantitative imaging of four contrast agents with the capability to cross cell membranes in sufficient quantity for detection. Each agent is based on the conjugation of a Gd(III) chelator with a cellular transduction moiety. Specifically, we coupled Gd(III)–diethylenetriaminepentaacetic acid DTPA and Gd(III)–1,4,7,10-tetraazacyclododecane-1,4,7,10-tetraacetic acid with an 8–amino acid polyarginine oligomer and an amphipathic stilbene molecule, 4-amino-4'-(N,N-dimethylamino)stilbene. The imaging modality that provided the best sensitivity and spatial resolution for direct detection of the contrast agents is synchrotron radiation x-ray fluorescence (SR-XRF). Unlike optical microscopy, SR-XRF provides two-dimensional images with resolution 10³ better than ¹⁵³Gd gamma counting, without altering the agent by organic fluorophore conjugation. The transduction efficiency of the intracellular agents was evaluated by T₁ analysis and inductively coupled plasma mass spectrometry to determine the efficacy of each chelate-transporter combination.

ADVANCES IN MAGNETIC RESONANCE IMAGING (MRI) have provided a new tool for the study of developmental biologic processes, such as cell lineage and fate mapping.^1,2 The ability to observe long-term developmental events in whole animals from descendants of individual precursors is producing a significant impact on the understanding of these complex processes. For example, an intact embryo can be labeled by microinjection of a stable, nontoxic magnetic resonance lineage tracer and images acquired. As a result, a complete time series of high-resolution three-dimensional MRIs can be analyzed forward or backward in time to reconstruct cell divisions and movements. A principal barrier to the development of new lineage tracers and contrast agents is the inherent inability of these complexes to cross cell membranes.^3–7 On cell transduction, these agents must then produce an observable effect on the MRI signal. Ideally, the agent has adequate synthetic versatility to allow for modification with a small-molecule (nonviral) delivery vehicle that does not increase toxicity and is efficient enough to deliver a large quantity of the agent.

The development of cell-permeable peptides and small molecules has led to the identification of numerous carrier molecules. A number of reports have described delivery vehicles that direct the agent to a particular cell type, such as transferrin^8–10 and folate receptor–targeted agents.^11–13 Similar to receptor-mediated delivery, cationic protein transduction domains such as polyarginine^14–17 and human immunodeficiency virus (HIV)-1 trans activating protein (TAT)^18–20 have been extensively used as translocation vehicles to facilitate delivery of all classes of modified MRI contrast agents.

The focus of this work is twofold: the development of highly efficient and passive vehicles for the intracellular transport of MRI contrast agents and the testing of new quantitative methods for evaluating uptake. We have synthesized four cell-permeable agents and characterized the efficiency of translocation by synchrotron radiation x-ray fluorescence (SR-XRF), inductively coupled plasma mass spectrometry (ICP-MS), and MRI. We compared the cellular transduction efficiencies of two different cell-permeable vehicles with entirely different translocation properties and mechanisms. This was accomplished by conjugating a cellular transduction moiety to diethylenetriaminepentaacetic acid (DTPA) and 1,4,7,10-tetraazacyclododecane-1,4,7,10-tetraacetic acid (DOTA) (Figures 1, 2, and 3).

SR-XRF uses photoelectric absorption of incident hard x-rays to cause the ejection of K shell electrons from an atom. With high atomic weight elements, this electron vacancy creates an excited state that relaxes with emission of a photon. ²¹ Owing to the monochromatic x-ray beam and zone plate optics available with SR-XRF, two-dimensional images with single-cell resolution (0.3 × 0.3 μm) can be recorded. A major advantage of SR-XRF over standard fluorescence microscopy is that images are obtainable without altering the agent by attachment of an organic fluorophore. Although gamma emitters such as ¹⁵³Gd and ¹²⁵I and high-resolution gamma counting offer the same advantage, both techniques requires handling and preparation of radioactive compounds, and the best resolution currently obtainable is 190 × 190 μm. ²²

Figure 1.

Structures of the intracellular agents: 1, 8–amino acid polyarginine oligomer coupled to Gd(III) DOTA; 2, 8–amino acid polyarginine oligomer coupled to Gd(III) DTPA; 3, 4-amino-4′-(N,N-dimethylamino)stilbene coupled to Gd(III) p-NH₂-DOTA; 4, 4-amino-4′-(N,N-dimethylamino)stilbene coupled to Gd(III) p-NH₂-DTPA.

Figure 2.

Synthesis of complex 2. A, (1) Piperdine; (2) DTPA dianhydride, DIEA; B, (1) 95% TFA, 2.5% H₂O, 2.5% TIS; (2) Gd(OH)₃, H₂O, 80°C, freeze-dry. Numbers in parentheses indicate multiple steps in a reaction sequence. Numbers in bold refer to molecules.

Studies aimed at the elucidation of in vivo cellular processes and in vivo fate mapping could improve our knowledge of key biologic processes, such as immune defense mechanisms and tumorigenesis. ²³ Therefore, modification of a contrast agent with a cellular transduction moiety and quantitative analysis of its translocation efficacy is a key step in the development of biologically compatible intracellular contrast agents.^{2,14,15,23–25}

Materials and Methods

All reagents and solvents were of the highest purity attainable from Aldrich (Milwaukee, WI) and TCI America (Portland, OR) unless otherwise noted. Modified Wang resin and amino acids were purchased from Novabiochem (San Diego, CA). NIH/3T3 cells, MDCK cells, and RAW 264.7 cells were purchased from American Type Culture Collection (ATCC, Manassas, VA). Also purchased from ATCC were Dulbecco's Modified Eagle's Medium (DMEM, with 4 mM l-glutamine modified to contain 4.5 g/L glucose and 1.5 g/L sodium carbonate) and Eagle's Minimum Essential Medium (EMEM, with Earle's balanced salt solution and 2 mM l-glutamine modified to contain 1.0 mM sodium pyruvate, 0.1 mM nonessential amino acids, and 1.5 g/L sodium bicarbonate). Fetal bovine serum (FBS), calf bovine serum (CBS), and 0.25% trypsin–ethylenediaminetetraacetic acid (EDTA) solution were purchased from ATCC. Trypan blue, vent-cap flasks, multiwell plates, cell scrapers, and Dulbecco's phosphate-buffered saline (DPBS) without calcium and magnesium were purchased from Fisher Scientific, and 200-mesh London finder gold grids coated with a polymer backing (formvar) were purchased from Electron Microscopy Sciences (Hatfield, PA). Stem coaxial inserts with a reference capacity of 60 μL and 2 mm OD were obtained from Wilmad LabGlass (Buena, NJ; catalog # WGS-5BL).

Complexes 1 ^{* 25} and 5 ²⁶ were synthesized following previously published procedures. Compound 7 was converted to compound 8 by following previously published procedures. ²⁷

¹H and ¹³C nuclear magnetic resonance (NMR) spectra were obtained on a Varian (Palo Alto, CA) Inova spectrometer at 400 and 100 MHz, respectively. Compounds were run in either CDCL₃ (7.27 and 77.23 ppm used as internal references for ¹H and ¹³C NMR spectra, respectively) or D₂O (4.80 and 49.00 ppm [MeOH] used as internal references for ¹H and ¹³C NMR spectra, respectively). Mass spectrometry samples were analyzed using electrospray ionization (ESI), single-quadrupole mass spectrometry on a Varian 1200L spectrometer. The results reported for m/z are for [M + H⁺]⁺ or [M - H⁺]⁻. Elemental analyses were performed at Desert Analytics Laboratory, Tuscon, AZ.

Analytic high-performance liquid chromatography (HPLC)–mass spectrometry (MS) was performed on a computer-controlled Varian Prostar system consisting of a 410 autosampler equipped with a 100 μL sample loop, two 210 pumps with 5 mL/min heads, a 363 fluorescence detector, a 330 photodiode array detector, and a 1200L single-quadrupole ESI-MS. All separations were executed with a 1.0 mL/min flow rate using a Waters 4.6 × 250 mm 5 μm Atlantis C18 column, with a 3.1:1 split directing one part to the MS and 3.1 parts to the series-connected light and fluorescence detectors. Mobile phases consisted of Millipore Synthesis grade water (solvent A) and HPLC-grade acetonitrile (MeCN) (solvent B). Preparative HPLC was accomplished using a Varian Prostar system. Two Prostar 210 pumps with 25 mL/min heads fed a 5 mL manual inject sample loop. Detection was performed after a 20:1 split by a two-channel Prostar 325 UV-visible detector and, on the low-flow side, an HP 1046A fluorescence detector. The mobile phases were the same as in the HPLC-MS instrument. Preparative runs were on a Waters 19 × 250 mm 10 μm Atlantis C18 column.

Figure 3.

Synthesis of 4-isothiocyanato-4'-(N,N-dimethylamino)stilbene compound (5) and Gd(III) (4,7,1-triscarboxymethyl-6-[4-(3-{4-[2-(4-dimethylaminophenyl)vinyl]phenyl}-thioureido)benzyl]-1,4,7,10-tetraazacyclododec-1-yl)-acetic acid (complex 3). A, K₂CO₃, CSCl₂, CHCl₃, 0°C; B, BH₃ THF; C, BrCH₂CO₂H, H₂O pH 10.0, 70°C 2. Pd/C, H₂; D, (1) 5, DMF, 80°C; (2) GdCl₃, H₂O, pH 7.0. Numbers in parentheses indicate multiple steps in areaction sequence. Numbers in bold refer to molecules.

MRI measurements were performed on a General Electric/Bruker (Billerica, MA) Omega 400WB 9.4 T imaging spectrometer fitted with Accustar shielded gradient coils at 20°C. Spin- lattice relaxation times (T₁) were measured using an inversion recovery pulse sequence. Images were acquired using a T₁-weighted spin-echo pulse sequence on freshly harvested cells. ICP-MS was performed on a computer-controlled Thermo Elemental (Waltham, MA) PQ ExCell Inductively Coupled Plasma Mass Spectrometer. Cells were counted using a Bright-Line hemacytometer.

Synthesis of Gd(III)-DTPA-(Arg)8 (synthesis of 2)

Polystyrene-based Wang resin containing an Fmoc-protected arginine residue (3.0 g, 0.58 mmol/g) was swelled in CH₂Cl₂ for 10 minutes (×3) and washed with peptide synthesis grade dimethylformamide (DMF) (4 × 10 minutes). The resin was treated three times with a solution of 20% piperdine in DMF (10 minutes), and the deprotected resin was washed with DMF (4 × 10 minutes). In a separate vial, Fmoc-protected Pbf-arginine (2.82 g, 4.35 mmol), o-(7-azabenzotriazol-1-yl)-N,N,N′,N′-tetramethyluronium hexafluorophosphate (HATU) (1.32 g, 3.48 mmol), and N,N′-diisopropylethylamine (DIEA) (1.12 g, 8.70 mmol) were dissolved in approximately 3 mL of DMF. The resulting solution was added to the deprotected resin, and nitrogen was bubbled through the mixture for 6 to 8 hours. The peptide solution was removed from the resin, which was subsequently rinsed with DMF (4 × 10 minutes). This procedure was repeated a total of seven times to achieve the synthesis of an 8–amino acid polyarginine oligomer bound to the Wang resin.

The resin (1.0 g, 0.58 mmol/g) was deprotected with the piperdine solution and washed with DMF as described above. In a separate vial, diethylenetriamine-N,N,N′,N′,N″-pentaacetic dianhydride (0.720 g, 2.00 mmol) and DIEA (0.369 g, 2.90 mmol) were dissolved in a minimal amount of anhydrous DMF. The resulting solution was added to the deprotected resin, and nitrogen was bubbled through the mixture for 6 to 8 hours. The peptide solution was removed, and the resin was washed with DMF, CH₂Cl₂, and MeOH (4 × 10 minutes each). Following the methanol washes, the resin was dried under a vacuum. A solution of 95% trifluoroacetic acid (TFA), 2.5% triisopropylsilane (TIS), and 2.5% H₂O (50 mL) was added to the resin, and nitrogen was bubbled through the mixture for 1 hour. The resin was filtered, and to the filtrate was added methyl tert-butyl ether (MTBE) (40 mL) to precipitate a white solid that was subsequently washed with MTBE (three times). The solid was dissolved in 30 mL of water and freeze-dried to yield white flaky crystals of the chelate-peptide conjugate DTPA-Arg₈ (0.42 g 44%) with ESI-MS (m/z): 1643.02; calculated for C₆₂H₁₁₉N₃₅O₁₈+H⁺: 1642.83.

To a solution of DTPA-Arg₈ (0.289 g, 0.176 mmol) in water was added Gd(OH)₃ (0.0410 g, 0.176 mmol). The reaction mixture was heated to 80°C and stirred for 16 hours. The mixture was allowed to cool to room temperature, and the pH was adjusted to 11.0 with concentrated NH₄OH. The resulting suspension was filtered using a 0.2 μm syringe filter and purified using preparatory HPLC with the Waters Atlantis column. An elution profile of a linearly increasing MeCN gradient from 0 to 98% over 45 minutes was used. The desired fraction (retention time 8.43 minutes by ultraviolet light at 220 nm with Gd fluorescence at Excitation λ = 274 and Emission λ = 315) was collected and freeze-dried to yield a white powder, which was found to be 2 (0.275 g, 87%) by ESI-MS (m/z): 1797.00 with Gd isotope pattern; calculated for C₆₂H₁₁₅GdN₃₅O₁₈+H⁺: 1796.84.

Synthesis of 4-Isothiocyanato-4′-(N,N-Dimethylamino)stilbene (synthesis of 5)

To a stirring solution of 4-amino-4′-(N,N-dimethylamino)stilbene (1.00 g, 8.39 mmol) in CHCl₃ (50 mL) at 0°C were added simultaneously a solution of K₂CO₃ (1.16 g, 8.39 mmol) in H₂O (30 mL) and a solution of thiophosgene (0.640 mL, 8.39 mmol) in CHCl₃ (30 mL). The reaction was allowed to warm to ambient temperature over 5 hours. The organic layer was separated, washed with H₂O, dried over MgSO₄, filtered through celite, and concentrated in vacuo to yield an orange solid (1.04 g, 93%). ¹H NMR (CDCl₃): δ = 2.99 (s, 6H), 6.70 (d, J = 8.0 Hz, 2H), 6.84 (d, J = 15.6 Hz, 1H), 7.04 (d, J = 15.6 Hz, 1H), 7.17 (d, J = 8.0 Hz, 2H), 7.41 (d, J = 8.0, 2H), 7.44 (d, J = 8.0 Hz, 2H); ¹³C NMR (CDCl3): δ = 40.59, 112.53, 122.83, 125.42, 126.22, 127.03, 127.99, 128.79, 130.49, 134.84, 137.71, 150.40; ESI-MS (m/z): 281.00; calc. for C₁₇H₁₆N₂S+H⁺: 281.10. Analytically calculated for C₁₇H₁₆N₂S: C, 72.82; H, 5.75; N, 9.99. Found: C, 73.12; H, 5.72; N, 9.80.

Synthesis of 2-(4-Nitrobenzyl)-1,4,7,10-Tetraazacyclododecane Tetrahydrochloride (synthesis of 7)

Using a modified literature procedure,^26,27 macrocycle 6 (1.61 g, 4.79 mmol) was dissolved in 140 mL of anhydrous tetrahydrofuran (THF) at 0°C and treated with dropwise addition of a solution of BH₃-THF (1.0 M, 33.5 mL, 33.5 mmol) under a nitrogen atmosphere. After the addition was complete, the reaction mixture was heated to reflux for 24 hours and concentrated in vacuo. The resulting residue was dissolved in 200 mL of MeOH and refluxed for 12 hours. This mixture was dried in vacuo, dissolved in MeOH (100 mL), and refluxed for 12 hours. This procedure was repeated with 100 mL of EtOH, and the remaining residue was dissolved in 90 mL of EtOH. HCl gas was continuously bubbled through the EtOH solution at 0°C until the temperature stopped rising. On complete addition of HCl, the resulting chalky white mixture was brought to reflux for 12 hours and concentrated in vacuo to approximately 30 mL. This reaction mixture was cooled to 0°C, and the resulting white precipitate was filtered under nitrogen, washed with cold diethyl ether (Et₂O), and dried under a vacuum. Recrystallization of compound 7 was accomplished from boiling MeOH to yield white crystals (1.06 g, 65%). ¹H NMR (D₂O): δ = 2.90 to 3.22 (m, 20H), 3.42 (m, 1H), 7.50 (d, J = 8.5 Hz, 2H), 8.20 (d, J = 8.5 Hz, 2H); ¹³C NMR (D₂O with MeOH reference): δ = 38.06, 42.20, 43.88, 44.36, 45.35, 45.58, 46.00, 49.12, 55.64, 125.44, 131.90, 146.32, 148.07; ESI-MS (m/z): 308.18; calculated for C₁₅H₂₅N₅O₂+H⁺: 308.20.

Synthesis of Gadolinium (III) (4,7,1-Triscarboxymethyl-6-[4-(3-{4-[2-(4-Dimethylaminophenyl)vinyl]phenyl)-thioureido)benzyl]-1,4,7,10-Tetraazacyclododec-1-yl}-Acetic Acid (synthesis of 3)

Compound 7 (0.497 g, 0.758 mmol) was added to a stirring solution of 4-isothiocyanato-4′-(N,N-dimethylamino)stilbene (0.233 g, 0.834 mmol) in anhydrous DMF (10 mL). This solution was brought to 80°C for 12 hours. The crude mixture was concentrated to dryness and dissolved in 10 mL of H₂O. GdCl₃ hexahydrate (0.282 g, 0.758 mmol) was added, and the pH of the solution was adjusted to 7.0 with 1 M NaOH. This solution was allowed to stir for 12 hours, at which time the pH was adjusted to 11.0 with 1 M NaOH. The resulting yellow solution was filtered with a 0.2 μm syringe filter and freeze-dried to yield a crude yellow-brown solid. This solid was purified via a Sephadex G-25 size exclusion column in H₂O. The fractions containing pure product were combined and freeze-dried to yield a light yellow solid (0.645 g, 77%). ESI-MS (m/z): 942.30 with Gd isotope pattern; calculated for C₄₀H₄₇GdN₇O₈S-H⁺: 942.24. Analytically calculated for C₄₀H₄₇GdN₇O₈S · Na · 2 H₂O: C, 47.94; H, 5.13; N, 9.78. Found: C, 47.76; H, 4.92; N, 9.61.

Synthesis of Gd(III) {[2-({2-(Biscarboxmethylamino)-3-[4-(3-{4-[2-(4-Dimethylaminophenyl)vinyl]phenyl}-Thioureido)phenyl]propyl}-Carboxymethylamino)-ethyl]-Carboxymethylamino}-Acetic Acid (synthesis of 4)

p-Aminobenzyl DTPA (0.560 g, 1.12 mmol) was prepared via a procedure in the literature ²⁶ and was added to a stirring solution of compound 5 (0.410 g, 1.46 mmol) in anhydrous DMF (10 mL). The resulting solution was brought to 80°C for 12 hours. The crude mixture was concentrated to dryness and dissolved in 10 mL of H₂O. GdCl₃ hexahydrate (0.417 g, 1.12 mmol) was added, and the pH of the solution was adjusted to 7.0 with 1 M NaOH. This suspension was allowed to stir for 12 hours, after which the pH was adjusted to 11.0 with 1 M NaOH. This solution was filtered with a 0.2 μm syringe filter and freeze-dried to yield a brown solid that was purified via a Sephadex G-25 size exclusion column in H₂O. The fractions containing pure product were combined and freeze-dried to yield a light orange solid (0.647 g, 56%). ESI-MS (m/z): 930.30 with Gd isotope pattern; calculated for C₃₈H₄₁GdN₆O₁₀S -H⁺: 930.18. Analytically calculated for C₃₈H₄₁GdN₆O₁₀S · 3 Na · 2 H₂O: C, 44.05; H, 4.38; N, 8.11. Found: C, 43.87; H, 4.50; N, 8.22.

SR-XRF Analysis

Each of the three cell lines (NIH/3T3, MDCK, and RAW 264.7) was incubated with 1 to 4 at 3.0 or 10 mM concentrations. A matrix of varying contrast agents, cell lines, and incubation concentrations produced 24 unique samples. Mouse fibroblast cells (NIH/3T3) were grown in modified DMEM containing 10% CBS. Canine kidney epithelial cells (MDCK) were grown in modified EMEM containing 10% FBS. Mouse monocyte macrophage cells (RAW 264.7) were grown in modified DMEM containing 10% FBS. The NIH/3T3, MDCK, and RAW 264.7 cells were plated at 250,000, 150,000, and 500,000 cells/mL, respectively, in tissue culture–treated Costar 12-well plates for 48 hours. Complexes 1 to 4 were solubilized in the corresponding cell media for each cell line (described previously in this section) at a concentration of 3.0 and 10 mM (corroborated by ICP-MS; data not shown), filtered through a 0.2 μm syringe filter, and added to the cells. Following the addition of media, the cells were incubated at 37°C in a 5% CO₂ atmosphere for 4 hours. The medium was removed, and the cells were rinsed in triplicate with room temperature DPBS. NIH/3T3 cells and MDCK cells were incubated with 250 μL of 0.25% trypsin-EDTA and collected. RAW 264.7 cells were harvested by cell scraping in 250 μL of RAW 264.7 media. The cells were counted with a hemacytometer and checked for viability using a trypan blue assay. ²⁸ Cell suspensions were centrifuged (at 1,000g), and the supernatant was removed, leaving ≈50 μL. Approximately 15 μL of this suspension was applied to formvar-coated gold grids with a sterile glass Pasteur pipette for 1 minute and the excess supernatant was removed. This was followed by the addition of ≈15 μL of room temperature ethanol, removal of the ethanol, and drying at ambient temperature for 15 hours. The cell coverage was approximately 15 to 30 cells/grid. The rest of the cell suspension was exposed to 500 μL of neat nitric acid at 70°C for 4 hours. The dissolved cells were diluted to 10 mL in a solution of 3% nitric acid and 5 ppb of indium as an internal standard. The samples were analyzed by ICP-MS. Electron microscopy grids with cells were mounted onto a kinematic specimen mount for both visible light and x-ray fluorescence microscopy. The samples were examined on a light microscope (Leica DMXRE), and the cells to be scanned with SR-XRF were located on the grid relative to a reference point using a high spatial resolution motorized x/y stage (Ludl Bioprecision).

Synchrotron-based scanning x-ray fluorescence microscopy was carried out at the 2-ID-E beamline of the Advanced Photon Source at Argonne National Laboratory (Argonne, IL). Hard x-rays (10 keV) from an undulator source were monochromatized using a single-bounce Si >111< monochromator. The energy was selected to allow for efficient excitation of the Gd L lines and to enable the detection of the Zn K lines. A Fresnel zone plate (320 μm diameter, focal length f = 250 mm; X-radia, Concord, CA) was used to focus the monochromatic x-ray beam to a spot size of ≈0.3 × 0.3 μm² on the specimen. The sample was raster-scanned through the beam at room temperature under a helium atmosphere. At each scan position, a full fluorescence spectrum was acquired using an energy-dispersive germanium detector (Ultra-LEGe detector, Canberra, Meriden, CT). Elemental content was determined by comparison of fitted sample spectra with NBS thin film standards 1832 and 1833 (NIST, Gaithersburg, MD) using MAPS software supplemented with fitting of fluorescence spectra at every pixel. ²⁹

MRI and T1 Analysis

NIH/3T3 cells, MDCK cells, and RAW 264.7 cells were grown in Corning brand tissue culture flasks (75 cm² with vent cap) and incubated in a 5% carbon dioxide incubator at 37°C. Cells were incubated with no contrast agent (incubated with corresponding media as controls) or 3.0 mM of complex 1, 2, 3, or 4 for 4 hours at 37°C in a 5% CO₂ atmosphere. Cells were rinsed in triplicate with room temperature DPBS. NIH/3T3 and MDCK cells were collected after exposure to 1 mL of 0.25% trypsin-EDTA and RAW 264.7 cells were harvested by exposure to 1 mL of RAW 264.7 media followed by cell scraping. Viable cells were then counted using a hemacytometer and a trypan blue stain. An average of 5,000,000 NIH/3T3 cells, 2,000,000 MDCK cells, and 12,000,000 RAW 264.7 cells were loaded into NMR tube coaxial inserts after centrifugation of the cell suspensions. Upon removal of the supernatant there was approximately 100 μL of total volume left to add to the coaxial inserts. The cells were allowed to settle for 18 hours into the 2 mm OD stem (60 μL reference capacity) of the insert. All MRI data were collected at ambient temperature in a General Electric/Bruker Omega 400WB 9.4 T magnet (83 mm bore size) fitted with Accustar shielded gradient coils. Spin-lattice relaxation times (T₁) were measured using an inversion recovery pulse sequence. Images were acquired using a T₁-weighted spin-echo pulse sequence with a repetition time of 300.8 milliseconds and an echo time of 15.0 milliseconds. After the T₁ measurements and spin-echo images were acquired, the cell suspensions (100 μL) were removed from the coaxial inserts using a 20-gauge stainless steel needle. These suspensions were diluted to 1 mL with H₂O. Cell count and viability was then determined. After cell counting, the cell suspensions were digested at 70°C for 4 hours in 1 mL of concentrated nitric acid. The dissolved cells were diluted to 10 mL in a final solution of 3% nitric acid and 5 ppb of indium as an internal standard. The samples were analyzed by ICP-MS.

Results

SR-XRF Analysis

SR-XRF analysis was performed to determine the cellular association pattern of agents 1 to 4. For each combination of cell line, contrast agent, and concentration (3.0 or 10 mM), five randomly chosen cells were raster-scanned at coarse resolution (2.0 × 2.0 μm step size). From these randomly chosen cells, two cells per group were raster-scanned at fine resolution (0.30 × 0.30 μm step size). All cells incubated with contrast agent and scanned revealed that Gd was present; however, we observed large standard deviations in the Gd content between different cells of the same population. The acquired high-resolution elemental maps provide pixel by pixel data sets that were used to globally confirm, map, and quantify the Gd distribution within each sample (Figure 4). These x-ray fluorescence maps were integrated over scanned pixels in the cell area to yield a quantitative elemental spectrum for the corresponding cell. The spectra of a cell (NIH/3T3) incubated with 4 (10 mM) and a control cell (NIH/3T3) are shown in Figure 5. Whereas fluorescence peaks corresponding to elements naturally occurring in cells are present in both of the samples (eg, Fe and Zn), the cells treated with 4 show peaks corresponding to the characteristic L_α1,2, L_β1, and L_β2 fluorescence lines for Gd, at E_α1,2 = 6,057.2, 6,025.0 eV, E_β1 = 6,713.2 eV, and E_β2 = 7,102.8 eV.

The Zn, Fe, and Gd columns of Figures 6 and 7 show the respective elemental distributions within one chosen cell of each cell type incubated with 1, 2, 3, or 4 or control contrast agents at 3.0 or 10 mM for 4 hours. The colocalization column combines the data from the three elemental maps (Zn, Fe, and Gd columns) to provide information on the areas of multielement overlap. Owing to the proximity of the Gd and Fe fluorescence energy levels (see Figure 5), it is important to deconvolute potential peak overlap between Fe and Gd fluorescence. This was accomplished using modified gaussian curves that were fitted at each scan position to the acquired x-ray fluorescence spectra. ²⁹ Comparison with NIST standards (as described in Materials and Methods) allowed the extrapolation of elemental concentrations. Finally, Gd concentration was again quantified within each sampled cell population using ICP-MS (Figure 8).

Figure 4.

Synchrotron radiation x-ray fluorescence intensity-weighted elemental maps of an MDCK cell incubated with complex 3 (10 mM for 4 hours). Each image indicates the relative distribution of the specified element. P, S, Ca, Zn, and K reveal cell boundaries within which the Gd distribution pattern is detectable. The spectrum depicting the relative concentrations in each map is scaled to differing values (maximum value given within each map).

MRI and T1 Analysis

Spin-lattice relaxation times (T₁) of unlabeled cells ranged from 2.45 to 2.87 seconds, whereas the T₁ values of cells incubated with 1 to 4 were 0.28 to 1.73 seconds (Figure 9A). As shown in Figure 9B, the trend remains that the samples with the highest total concentration of Gd show the shortest T₁, corroborating the hypothesis that the relaxation rate is a function of the cellularly associated Gd concentration and not an anomaly of cell packing. These differences were visualized using a T₁-weighted spin-echo imaging sequence of RAW 264.7 cells incubated with 1, 2, and 4. The results of the imaging experiment show that the control cell image is considerably darker than the image of the cells treated with the contrast agents (Figure 10).

Figure 5.

Comparison between the integrated x-ray fluorescence spectrums of all scanned pixels of an untreated NIH/3T3 cell (black spectrum) with that of an NIH/3T3 cell incubated with 3.0 mM of complex 4 for 4 hours (orange spectrum). Whereas peaks that correspond to elements normally present in cells are visible in both spectra (eg, Fe [K_α1 = 6.403 keV] and Zn [K_α1 = 8.636 keV]), cells treated with 4 show sharp peaks corresponding to the characteristic L_α1 (6.058 keV), L_β1 (6.713 keV), and L_β2 (7.034 keV) energy levels for Gd fluorescence.

To demonstrate the T₁ differences between cell lines and their corresponding media, a sample of each cell line incubated with 3 at 3.0 mM was allowed to gravity settle within a coaxial insert. This insert was placed within a larger tube (5 mm OD) filled by the corresponding cell media. Images show that the media phantoms provide an internal reference that has a much longer T₁ (darker image) than the cells treated with contrast agent (Figure 11).

Discussion

Our results indicate variable uptake of agents 1 to 4 between the different cell lines. This disparity can be attributed to three variables: transduction domain (polyarginine or stilbene), Gd(III) chelator (DOTA or DTPA), and cell type (NIH/3T3, RAW 264.7, or MDCK). Of these variables, the choice of transduction moiety has been studied most rigorously. Polyarginine complexes have been shown to enter cells through an endocytic pathway, whereas the uptake mechanism of stilbene complexes is not well understood.^14,16,17,30 Both Skovronsky and colleagues and Kung and colleagues demonstrated that stilbene-based compounds cross the blood-brain barrier, which may be an indication of a passive uptake mechanism.^31,32 Further, we discovered that the second variable that plays an important role in the cellular transduction efficiency of MRI contrast agents is the Gd(III) chelator. Changing the Gd(III) chelate from DOTA to DTPA leads to an overall molecular charge decrease from n to n-1.

Figure 6.

A, Synchrotron radiation x-ray fluorescence–determined, background-subtracted three-element overlay maps (Zn = blue, Fe = red, and Gd = green) of one cell for each specific cell line incubated with complex 1, 2, 3, or 4 at 3.0 mM for 4 hours. B, A background-subtracted three-element overlay map of one cell for each specific cell line incubated with an arginine- (1) or stilbene- (3) modified contrast agent at 10 mM for 4 hours. Scale bars represent 2.0 μm. Colocal. = colocalization.

These variables represent two molecular properties that were systematically varied to determine the transduction efficiencies of a cationic polyarginine peptide (complexes 1 and 2) and derivatives of 4-amino-4′-(N, N-dimethylamino)stilbene (complexes 3 and 4) to associate DOTA- and DTPA-based Gd(III) contrast agents with cells of differing morphologies and uptake mechanisms.^33–36 The data obtained from testing 1 to 4 with ICP-MS, SR-XRF, T₁ analyses, and MRI provide unique information while corroborating the same transduction efficacy of the transport moieties. For example, the SR-XRF images demonstrate association of the agents with single cells and the ICP-MS data are quantitative evidence that the T₁ values and SR-XRF maps are a consequence of the Gd(III) agents. The MRI and T₁ analyses prove that 1 to 4 are viable contrast agents for MRI at the tested concentrations.

Figure 7.

A synchrotron radiation x-ray fluorescence background-subtracted three-element overlay map (Zn = blue, Fe = red, and Gd = green) of one NIH/3T3 cell for Dotarem (Gd(III)-DOTA) or Magnevist (Gd(III)-DTPA) at 3.0 or 10 mM for 4 hours. Scale bars represent 2.0 μm. Notice that the cell boundaries are outlined in the Zn map (as in Figure 4); however, the Gd distribution shows no discernible pattern. Note that each image is scaled to its respective maximum value; therefore, the Gd distribution falsely highlights the background.

The weak fluorescence quantum yield of Gd(III) and the current inability to directly visualize its in vitro location with submicron resolution required the use of SR-XRF to validate cellular uptake of the modified agents. Currently, hard, high-brilliance SR-XRF microprobes that employ Fresnel zone plates to focus incident x-rays can achieve submicron spatial resolution on comparatively thick (eg, 10–20 μm) samples.^37,38 The calculated attenuation length of similar organic material to our cells is 18.1 mm; therefore, only 0.06% of incident x-rays are absorbed by a 10 μm–thick target cell. ³⁹ This high elemental sensitivity and spatial resolution of SR-XRF microprobes make them well suited for studying the interactions of Gd(III)-based contrast agents and single cells in vitro. By mapping individual cells, SR-XRF microprobe analyses complement bulk measurements performed by ICP-MS and MRI. Importantly, SR-XRF allows elemental mapping of agents 1 to 4 without the use of organic fluorophores that may alter the transport properties of the agents.

Figure 8.

Average Gd concentrations determined via inductively coupled plasma mass spectrometry of the NIH/3T3, RAW 264.7, and MDCK cells incubated with 1 to 4 at 3.0 mM for 4 hours (same parameters as the synchrotron radiation x-ray fluorescence imaging study). The average cell count for RAW 264.7, NIH/3T3, and MDCK cells is 2,000,000, 750,000, and 600,000 cells, respectively. All samples were run in triplicate, with error bars representing 1 SD.

As shown in Figure 6, each scanned cell incubated with complexes 1 to 4 showed substantial Gd fluorescence compared with the control cells incubated with Gd(III)-DOTA or Gd(III)-DTPA or in the absence of a contrast agent (see Figure 7) (the NIH/3T3 control images are representative of the data obtained for all three cell lines studied). As expected, when the incubation concentration was increased to 10 mM, the Gd distribution patterns associated with the cells became more pronounced (see Figure 6B). Complex 1 (arginine moiety) has a more diffuse pattern inside the cell, whereas complex 3 (stilbene moiety) exhibits a more punctate pattern. This distribution may be explained by the amphipathic nature of the stilbene agents and their affinity to aggregate in aqueous media, or it may be due to cell-specific uptake mechanisms. ²⁵ However, the mechanism of uptake cannot be determined based on the data presented in this work as we focused on the uptake efficiency of intracellular contrast agents and the quantification of the Gd associated with the cells.

As previously stated, SR-XRF can map and quantify elemental concentrations at femtogram levels. ²⁹ However, owing to the size of the cells (22 × 22 μm for MDCK cells), the resolution of the acquired elemental maps (raster-scanned at 0.3 × 0.3 μm with a dwell time of 1.0 s/pixel), sample focusing (≈1 h/cell), and total scan time per cell (≈2.5 h/cell) made rigorous quantification of associated Gd by this technique feasible for only a small number of samples (see Materials and Methods). As expected, the ability to sample only a small number of cells gave rise to large variances in Gd concentration within a population. Nonetheless, these large standard deviations are simply an indication that contrast agent uptake is not homogeneous within a given cell line even at a constant concentration and time. ⁴⁰ This finding necessitated the sampling of a larger cell population to quantify and rank the transport efficacies.

Figure 9.

A, T₁ study of NIH/3T3, RAW 264.7, and MDCK cells incubated with 3.0 mM of 1 to 4, as well as untreated cells and media. T₁ analysis was accomplished using an inversion recovery pulse sequence. Error bars represent 1 SD of the slices taken of the coaxial insert (minimum five slices with 1 mm thickness). B, Inductively coupled plasma mass spectrometry–calculated amount of Gd associated with the cells from the T₁ analysis. Total moles of Gd are shown to emphasize the inverse comparison between Gd concentration (B) and T₁ (A). This inverse trend demonstrates that the relaxation rate is a function of the cellularly associated Gd concentration and not an anomaly of cell packing.

To quantitatively determine the uptake trend of complexes 1 to 4 within the three cell lines, ICP-MS was used on larger cell populations. In Figure 8, per cell Gd content is directly correlated with cell size, with MDCK cells being the largest in volume followed by NIH/3T3 and RAW 264.7 cells. This trend may be a consequence of cellular physiologic changes that affect cell membrane transduction. However, cell viability did not decrease more than 2 (3.0 mM) or 5% (10 mM) lower than the controls. This aside, attention should be drawn to the relative transduction efficiencies of the agents within each cell line. The obtained data set shows that 1 to 4 are associated with the following cell-dependent trends: NIH/3T3, 4 > 2 > 3 > 1; RAW 264.7, 2 > 4 > 1 > 3; and MDCK, 4 > 3 > 1 > 2. As the data show, MDCK cells appear to be transport molecule dependent, allowing 3 and 4 to accumulate with increased efficiency. This dependency could be due to the differentiation of MDCK cells into columnar epithelium, which may restrict active transport and limit the charge interaction between the cell membrane and molecules with multiple charges (1 and 2).^41,42 In contrast, NIH/3T3 and RAW 264.7 transduction appears to be Gd(III) chelator dependent, preferring DTPA- (2 and 4) to DOTA-based (1 and 3) contrast agents. These results could be a function of overall molecular charge or three-dimensional chelator conformations.

Figure 10.

T₁-weighted spin-echo magnetic resonance images of RAW 264.7 cells at 9.4 T incubated with 1, 2, or 4. Images were obtained using a spin-echo pulse sequence with a repetition time of 300 milliseconds and an echo time of 15 milliseconds (field of view = 22 mm, slice thickness = 0.5 mm). All incubations were performed at 3.0 mM for 4 hours. 1, Untreated RAW 264.7 cells; 2, RAW 264.7 cells incubated with 1; 3, RAW 264.7 cells incubated with 2; 4, RAW 264.7 cells incubated with 4. Scale bar represents 1.5 mm.

Figure 11.

T₁-weighted spin-echo magnetic resonance image of NIH/3T3, RAW 264.7, and MDCK cells at 9.4 T. Images were obtained using a spin-echo pulse sequence with a repetition time of 300.8 milliseconds and an echo time of 15 milliseconds (field of view = 22 mm, slice thickness = 0.5 mm). All samples (2–4) were incubated with 3 at 3.0 mM for 4 hours. 1, Deionized water in a 5 mm OD nuclear magnetic resonance tube; 2, NIH/3T3 cells (center) with an external phantom of NIH/3T3 media (outer ring); 3, RAW 264.7 cells (center) with an external phantom of RAW 264.7 media (outer ring); 4, MDCK cells (center) with an external phantom of MDCK media (outer ring). The scale bar represents 1.5 mm. The dark spots throughout the cell images are due to air pockets created during cell packing.

To corroborate the quantitative cell studies and determine their utility as MRI contrast agents, complexes 1 to 4 (at the lowest incubated concentration, 3.0 mM) were tested for contrast enhancement via MRI. Comparison of the cells incubated with 1 to 4 in each image with either control cells (see Figure 10) or media (see Figure 11) reveals increased signal intensity. Examination of Figure 9 provides transduction efficiencies (given by T₁ values) that are identical to those outlined in the ICP-MS study done on the SR-XRF samples. The images from Figures 10 and 11 demonstrate the utility of ICP-MS and SR-XRF in prediction of relevant MRI enhancement.

In conclusion, this investigation shows the ability of a polyarginine and stilbene functionalized MRI agent set to label three cell lines effectively enough to be visualized via MRI. The data obtained from ICP-MS, T₁ analyses, and acquired MRIs covalidate the transduction efficiency of 1 to 4 within each cell line. SR-XRF was used to supplement these quantitative data by visualizing contrast agent association with single cells. The transduction efficiencies are not consistent across cells lines; therefore, selection of transduction moiety and Gd(III) chelator is an important consideration when developing intracellular MRI contrast agents The uptake mechanism of each agent is currently under research.