Materials & Methods
Synthesis of Eu-DOTA-Gly4
DOTAM-Gly, the tetraglycineamide derivative of DOTA, was synthesized using a previously reported method in. Unless otherwise noted, all reagents were purchased from Acros Organics (Geel, Belgium), or Sigma-Aldrich Inc. (MO, USA), and used without further purification. DN-(2-Bromoethanoyl) ethyl glycinate was synthesized from bromoethanoyl bromide (10.5 ml, 120 mmol) and glycine ethyl ester (14 g, 100 mmol). DOTAM-Gly-OEt was synthesized through exhaustive alkylation of cyclen (1,3,5,7-tetraazacyclododecane; 1.72 g, 10 mmol) with N-(2-bromoethanoyl) ethyl glycinate (9.19 g, 41 mmol) in acetonitrile with K2CO3 (11.06 g, 80 mmol) at 70°C for 6 h under N2 gas. After the reaction, undissolved materials were removed by filtration and the final product was obtained by evaporating the solvent (7.40 g, quantitative yield). The hydrolysis of DOTAM-Gly-OEt (7.45 g, 10 mmol) was conducted in ethanol:water (1:1) at 60°C by controlling pH at approximately 11 with 1 N-NaOH. The 1-h reaction was traced with thin-layer chromatography. The reaction mixture was cooled, acidified to pH 3 with 1 N-HCl, lyophilized, redissolved in water, and purified with liquid chromatography (solid-phase: Amberlite® XAD-1600; eluent: water, 5.04 g, 80% yield).
Synthesis of [Eu-DOTA-Gly4]42–G5PAMAM
A G5PAMAM dendrimer (Dentritech Inc.; MI, USA) was labeled with Eu-DOTA-Gly4 using the method previously published in. Briefly, NHS (0.21 g, 1.83 mmol) and 1-ethyl-3-(3-dimethylaminopropyl) carbodiimide.HCl (0.88 g, 4.6 mmol) was added to a stirred solution of Eu-DOTA-Gly4 (0.81 g, 1.03 mmol) in 2-(N-morpholino)ethanesulfonic acid buffer (20 ml, pH 6.5) at 0°C for 1 h. The resulting active ester, Eu-DOTA-Gly4 NHS, was added in aliquots to a stirred solution of G5PAMAM dendrimer (0.08 g, 72 µmol) in 3.5 ml of phosphate-buffered saline (PBS) at pH 7.4 and then allowed to stir at room temperature for 24 h. The solution was filtered using a centrifugal filter unit with a 10,000 molecular weight cut-off (Millipore Inc., MA, USA). Finally, the solution was lyophilized to obtain 420 mg of a white solid, which represented a 17% yield. Matrix-assisted laser desorption/ionization mass spectrometry was used to characterize the average number of DOTA-Gly4 ligands conjugated to G5PAMAM.
Synthesis of [DyLight® 680]1–[Eu-DOTA-Gly4]42–G5PAMAM
A solution of DyLight 680 NHS ester (50 mg, ~0.071 mmol; Thermo Fisher Scientific, IL, USA)) in dimethyl sulfoxide was added to a stirred solution of Eu-G5PAMAM (200 mg, 0.071 mmol) in 2 ml of PBS, and the reaction was stirred at room temperature for 24 h. The reaction mixture was diafiltrated using Amicon Ultra centrifugal filter unit with a 10,000 molecular weight cut-off. The solution was lyophilized to obtain 210 mg of solid (~0.66 mmol, 93% yield). The emission profile of the final product was determined and compared with that of DyLight 680.
Synthesis of [DyLight 680]1-G5PAMAM
The same procedure used to label Eu-G5PAMAM with a fluorophore was used to label a G5PAMAM dendrimer. A solution of DyLight 680 NHS ester (150 mg, 0.16 mmol) in dimethyl sulfoxide was added to a stirred solution of G5PAMAM dendrimer (0.040 g, 1.3 mmol) in 2 ml of PBS, and the reaction was stirred at room temperature for 24 h. The reaction mixture was diafiltrated using Amicon Ultra centrifugal filter unit with a 10,000 molecular weight cut-off. The solution was lyophilized to obtain 0.041 g of solid (~1.4 mmol, 87% yield). The emission profile of the final product was determined and compared with that of DyLight 680.
The animal study was conducted according to approved procedures of Institutional Animal Care and Use Committee of Henry Ford Hospital (MI, USA). U87 human glioma cells were provided by Tom Mikkelsen from the Henry Ford Hospital. The cells were cultured in 75 cm2 tissue culture flasks with DMEM supplemented with 10% fetal bovine serum, penicillin (100 IU/ml) and streptomycin (100 µg/ml) until they were 80–90% confluent. The cells were collected by trypsinization, washed and centrifuged to make a cell suspension of 4 × 105 cells/5 µl.
An athymic nude rat that was 6–8 weeks old and weighed 160 g (Charles River Laboratory Inc., MD, USA) was anesthetized by intraperitoneal injection, using a mixture of 100 mg/kg ketamine and 15 mg/kg xylazine. Orthotopic human glioma was created in nude rats by implanting 4 × 105 human U87 glioma cells according to a recently published method in. The orthotopic tumor was allowed to grow for 4–5 weeks to a diameter of approximately 5–7 mm, before conducting the MRI study.
In Vitro CEST MRI Studies
Solutions of DyLight 680-Eu-G5PAMAM that ranged in concentration from 46.9–2750 µM (on a per dendrimer basis) were prepared in PBS at pH 7. The concentrations of these solutions were validated using inductively coupled plasma mass spectrometry. Images were acquired using a 7T Varian MR scanner that was equipped with a 12-cm bore magnet and a 38-mm diameter homemade transmit/receive quadrature birdcage coil. The CEST effects of these samples were measured at 37°C using a rapid acquisition with refocused echoes pulse sequence (echo time: 5 s; repetition time: 10 ms; rapid acquisition with refocused echoes factor: 8; one image slice at 2-mm slice thickness; 128 × 128 matrix, 250 × 250 µm in-plane resolution; 24 × 24 mm field of view; one average). This experiment was prepended with a selective saturation pulse applied at 17 µT for 1.6 s at MR frequencies ranging from +80 to -60 ppm in 1 ppm increments (where the water resonance is referenced to 0 ppm). This saturation time and saturation power was considered to generate a maximum CEST effect, because a saturation time of 2 s and a saturation power of 20 µT produced the same CEST effects. The percent CEST was calculated by comparing the images acquired with selective saturation at +55 ppm (MS) with images acquired with selective saturation at -55 ppm (M0) (Equation 1). Image contrast was measured using ImageJ (NIH, MD, USA), and the calibration of the CEST effect with respect to concentrations was evaluated using a Hanes-like linear analysis method, which was evaluated with MS Excel (Microsoft Inc., MI, USA).
In vivo CEST MRI Studies
The same Varian MRI scanner used for in vitro studies was also applied for the in vivo study. To prepare for the MRI scans, the rat was initially anesthetized with 3.0% isoflurane in a 2:1 N2O:O2 mixture administered via a facemask, followed by 1.2–2.0% isoflurane during the MRI scans. The respiration rate and rectal temperature were continuously monitored using an automated feedback system (SA Instruments Inc., NY, USA). During MRI studies, the rectal temperature was automatically maintained at 37.0 ± 0.2°C using warmed air. The rat was secured in a customized MRI-compatible cradle to prevent injury during the MRI exam and to stabilize the head with ear bars in order to eliminate motion artifacts during MRI scans.
An isotropic 3D fast imaging employing steady state acquistion MR image was acquired to locate the U87 glioma in the rat model using 3T GE Excite clinical MRI system fitted with dedicated small animal coil (Litzcage small animal imaging system, Doty Scientific Inc., SC, USA). The fast imaging employing steady state acquistion images were obtained with the following parameters – echo time: 11.4 ms; repetition time: 5.61 ms; 0.3-mm slice thickness; 200 × 200 matrix; 300 × 300 µm in-plane resolution; 60 × 60 mm field of view and; two averages. The tumor rim was visualized by thresholding the image to obtain a contiguous loop of pixels that had the lowest signal amplitude in the region of the glioma. This tumor rim was used to define the region of interest for subsequent analyses.
The same CEST–rapid acquisition with refocused echoes MRI protocol used for in vitro studies was also applied for the in vivo study, except that a 32 × 32 mm field of view was used. This protocol provided a 80-s temporal resolution that did not cause sample heating. A series of axial images with saturation frequencies applied at +55 ppm were acquired for 48 min. After the first six images were acquired, a solution of 0.045 mmol/kg DyLight 680-Eu-G5PAMAM was injected through the tail vein catheter in 600 µl volume, which equated to a concentration of 12 mM that was prepared using the average molecular weight of DyLight 680-Eu-G5PAMAM, determined from matrix-assisted laser desorption/ionization mass–mass spectrometry analysis. This injection was prepended by injection of 100 µl heparinized saline and followed by 100 µl of saline to flush the catheter. The total volume of 800 µl was manually injected within approximately 1 min. Following MRI, the rat was moved to another facility for optical imaging.
The % CEST was calculated by comparing the average water signal at each post-injection time-point (MS) with the water signal of the tumor determined from the average of the images acquired before injection (M0) (Equation 1). The average water signal was obtained from the region of interest that was determined from the 3D FIESTA MRI study. The standard deviations of the CEST measurements were determined from the pixel-wise distributions of water signals in the region of interest. The contrast:noise ratio, or the CEST:noise ratio in this type of experiment, must exceed 2√2 in order to ensure that the contrast has a 95% probability of being real. Although this statistical threshold for the CEST:noise ratio is based on Rician noise, this threshold is also valid for Gaussian noise present in this study due to the high signal:noise ratio of 133:1 for the MR image of the tumor before injection. To determine the noise level, the standard deviation of the MR signal of an image region that represented air was multiplied by √2 to account for the subtraction of two images before and after injection. Only pixels that reached this CEST:noise threshold were used for subsequent analyses.
Near-infrared fluorescence imaging of the rat was performed after acquisition of the MR images using a Kodak Carestream Multispectral Imaging System (Carestream, NY, USA). To obtain an optimal emission image devoid of nonspecific fluorescent emission, fluorescence images were acquired using multiple excitation wavelength filters, ranging from 540 to 690 nm and a single emission wavelength filter of 750 nm. Spectral profiles were created using the spectral analysis software (Carestream), from which subtracted optimal images were obtained. For each optical image set, an x-ray image was also obtained to validate the anatomic location of the tumor using the same animal position and field of view (the near-infrared and x-ray images were coregistered).
After in vivo imaging, the rat was euthanized using CO2 asphyxiation and perfused by intracardiac injection of 100 ml PBS followed by 3% paraformaldehyde, and then kept in a solution of 3% sucrose and 3% paraformaldehyde during storage. An ex vivo fluorescence image of the whole brain was acquired, which was registered to the in vivo fast imaging employing steady state acquistion MR image by comparing the outline of the autofluorescence and MR signal of the normal brain tissue. The brain was snap-frozen and cut into 15–20µm thick sections for further analysis with fluorescence microscopy. For fluorescent microscopic detection of DyLight 680 conjugated to Eu-G5PAMAM, a proper excitation and emission filter was used. Sections were stained with fluorescein isothiocyanate-labeled tomato lectin to delineate endothelial lining of functional and nonfunctional blood vessels (neovascularization). Nuclei were visualized with 4',6-diamidino-2-phenylindole (Sigma-Aldrich), using standard histological staining procedures as recommended by the suppliers of the reagent.
Nanomedicine. 2012;7(12):1827-1837. © 2012 Future Medicine Ltd.