|Home | About | Journals | Submit | Contact Us | Français|
A variety of (super)paramagnetic contrast agents are available for enhanced MR visualization of specific tissues, cells, or molecules. In order to develop alternative contrast agents without the presence of metal ions, liposomes were developed containing simple bioorganic and biodegradable compounds that produce diamagnetic Chemical Exchange Saturation Transfer (DIACEST) MR contrast. This DIACEST contrast is frequency-dependent, allowing the unique generation of “multi-color” images. The contrast can be turned on and off at will, and standard images do not show the presence of these agents. As an example, glycogen, L-arginine, and poly-L-Lysine were encapsulated inside liposomes and injected intradermally into mice to image the lymphatic uptake of these liposomes. Using a frequency-dependent acquisition scheme, it is demonstrated that multi-color MRI can differentiate between different contrast particles in vivo following their homing to draining lymph nodes. Being non-metallic and bioorganic, these DIACEST liposomes form an attractive novel platform for multi-color imaging in vivo.
Magnetic resonance imaging (MRI) is a versatile medical imaging modality that is employed primarily to generate anatomical images based on water due to its high content in soft tissues. The use of metallic MR contrast agents based on paramagnetic metal ions such as gadolinium (1) or superparamagnetic iron oxide particles (2) greatly improves the detection sensitivity and/or specificity to targeted cells or molecules, allowing MRI to provide details at the cellular or even molecular level. These metal-based agents act through modulation of the water relaxation rate, either increasing or reducing the water signal intensity. Despite their great success and demonstrated applicability (1,2), such metal-based agents still have several limitations. In addition to potential issues with biocompatibility and expected lengthy procedures before approval of use in humans, the contrast produced is based only on changes in signal intensity. Although often displayed in color, intensity changes can be considered as grayscale (one color) and multiple relaxation agents cannot be distinguished. One color MR contrast ultimately limits the application of MR in monitoring complex biological systems, as it is often desirable to examine a number of different processes simultaneously (3,4). The availability of multi-color MRI would put MR at par with other technologies, such as dual or triple label immunohistochemistry and optical imaging methods (5–11), but with the in vivo advantages of unlimited penetration depth (a major obstacle with optical imaging), excellent soft-tissue contrast, superior spatial resolution, and widespread clinical availability.
Recently, it was demonstrated that exchangeable protons can be used for generating MR contrast (12–16) by detecting the effect of their selective radiofrequency irradiation (saturation) as a reduction of the water signal used for imaging. To this end, we hypothesized that the proton exchange properties of simple biodegradable sugars, amino acids, and peptides could be exploited to produce multi-color 1H MR images because the chemical shifts (MR frequencies) of these exchangeable protons differ. We developed diamagnetic liposomes (DIACEST liposomes or DLs) containing biodegradable agents for multi-frequency (multi-color) MRI, and as a proof-of-principle demonstrate that artificial colors can be assigned to these particles for multi-color imaging of lymph nodes based on their saturation frequency-dependent CEST contrast. Lymph node mapping is an important medical application for tumor staging, and, due to the complexity of the lymphatic system, multi-color imaging is preferably used to simultaneously detect and differentiate multiple processes involved in the lymphatic composition and drainage (17).
DLs were prepared as described previously (18,19) using starting solutions of 100 mg/ml L-Arg, 25 mg/mL PLL(MW=15–30 kDa), or 100 mg/ml Glyc (MW=25–100 kDa). The lipid/sterol components of the liposomes were egg phosphatidylcholine (egg PC), cholesterol, 1,2-distearoyl-sn-glycero-3-phosphoethanolamine-N-[carboxy(polyethylene glycol)-2000] (DSPE-PEG(2000)), and 1,2-dipalmitoyl-sn-glycero-3-phosphoethanolamine-N-(lissamine rhodamine B sulfonyl) (Liss Rhod PE) with a total mole % of 46.6:46.6:5:1.8. After annealing, liposomes were extruded through 50–400 nm polycarbonate cutoff filters and dialyzed overnight using 250 kDa cutoff PVDF dialysis tubing (Spectrum Laboratories, Inc). The size and concentration were measured using dynamic light scattering (Nanosizer, 90ZS, Malvern Instruments) and fluorescence (Victor V, Perkin Elmer). The final size was controlled by the size of the poylcarbonate cutoff filter used, and was between 100–200 nm.
Liposomes loaded with 111In were prepared using an active loading protocol using oxine as a transporter/channel creator (19). Briefly, 6 μL of 11 mM oxine in ethanol was added to 194 μL acetate buffer, pH=5.5, and then 111InCl3 in 3 mM HCl was added dropwise to the solution. The resulting mixture was added dropwise to liposomes containing 2 mM DTPA(diethylenetriaminepentaacetic acid) within instead of DIACEST agent and left at room temperature for 1 hour. After loading, non-entrapped 111In was complexed by adding 10 mM EDTA, followed by liposomes purification using a Sephadex G-50 1×10 cm column. 111In loaded liposomes (40 μL, 90 400 μCi) were then injected into the animal’s footpad.
8-week old C57BL/6.SJL mice bearing the CD45.1 alloantigen were purchased from Jackson Labs (Bar Harbor, Maine) and bred in-house. For vaccination, mice were injected with a cancer cell vaccine composed of irradiated (10,000 Rad) B16 melanoma along with irradiated (5,000 Rad) B78-H1 GM-CSF-expressing bystander melanoma to produce an immunoresponsive enlargement of the popliteal lymph node (PLN) 1 week prior to liposome injection (20).
All in-vitro MRI datasets were acquired at 310 K using a 9.4 T Bruker scanner and 15 mm birdcage resonator. CEST imaging was conducted as described previously (21) through collection of two sets of saturation images: a so-called water saturation shift referencing (WASSR (22)) set for B0 mapping and a CEST set for characterizing contrast. For the WASSR images, the saturation parameters were tsat=200 ms, B1 =0.5 μT (21.3 Hz), TR=1.5 sec with saturation offset incremented from −1 to +1 ppm with respect to water in 0.1 ppm steps, while for the CEST images: tsat=4 sec, B1=4.7 μT (200 Hz), TR=6 sec, with offset incremented from −5 to +5 ppm (0.2 ppm steps). The other imaging parameters were: effective TE=40 ms, RARE factor=16, acquisition bandwidth=50 kHz, slice thickness=1 mm, acquisition matrix size=128×64, field of view (FOV)=13×13 mm, and NA=2.
In vivo images were acquired on a 9.4T Bruker scanner using a 25 mm sawtooth resonator and the same imaging scheme described above with addition of a fat suppression pulse (3.4 ms hermite pulse, offset=−3.5 ppm). The acquisition parameters were: TR=5.0 sec, effective TE=21.6 ms, RARE factor=6–8, tsat=3 sec, B1=3.6 μT (150 Hz), slice thickness=0.7 mm, acquisition matrix size=128×64, FOV=20×20 mm, and NA=2. Due to the B0 field inhomogeneity in the hind limbs, we incremented the saturation offset ± 2 ppm (0.1 ppm steps) with respect to water for B0 mapping. For the CEST contrast characterization, the offset range was ± 3 ppm for the single L-Arg or Glyc unilateral injected mice and ± 5 ppm for the single PLL unilateral injected mice and the L-Arg, PLL bilateral injected mice. Images were collected twenty-four hours after the mice were intradermally injected with 40μL of ~30nM liposome solution per footpad. As for the animals injected with mixed-type probes on two sides, pre-contrast CEST images were also acquired right before the footpad injection.
All data were processed using custom-written Matlab scripts (Supporting Information). The CEST contrast was quantified by calculating the asymmetry in the magnetization transfer ratio (MTRasym) using MTRasym=(S−Δω − S +Δω)/S−Δω. ROI masks were drawn manually based on the T2w images to cover the entire PLN and the mean intensities were used for plotting MTRasym. The analysis of mice with dual injections required a separate procedure to assign colors to each voxel (Supporting Information).
At 24 hours post DL-injection, mice were transcardially perfused with 4% paraformaldehyde and both legs were fixed in 4% paraformaldehyde overnight. Following dissection of the femur bone, the knee region containing the PLN was cryopreserved with 30% sucrose and cryosectioned at 30 μm. Tissue sections were stained with hematoxilin and eosine (H&E), or counterstained with DAPI for fluorescent microscopy of rhodamine-labeled liposomes.
To determine whether the liposomes localize within intracellular vs. extracellular compartments within the PLN, they were excised immediately after sacrificing the animal, submerged in O.C.T. mounting media and flash frozen in a bath of liquid nitrogen-cooled 2-Methylbutane. 10 μm frozen sections were fixed in ice-cold acetone, air dried, and stored at −80°C. Sections were then hydrated in PBS, blocked with 2% IgG-free BSA and stained overnight at 4°C with anti-CD45.1-allophycocyanin(APC), a pan-leukocyte marker (clone A20), or alloantigen isotype control CD45.2 (clone 104) (BD Biosciences, San Jose, CA).
Contrast probes based on exchangeable protons can be made distinguishable from each other provided these exchangeable protons possess sufficiently different NMR chemical shifts. To select compounds for integration into our particles, we compiled a list of previously reported exchangeable proton species in biomolecules (12,23–25), with the chemical shifts of these protons ranging from 0.8 ppm (i.e. OH-rich compounds) to 7 ppm (i.e. heterocycles with NH groups integrated into the ring). The chemical shifts can be viewed as a way to produce an artificial DIACEST color spectrum from the different species of exchangeable protons, as illustrated in Fig. 1a. We selected three compounds: glycogen (Glyc), L-arginine (L-Arg), and poly-L-lysine (PLL), that were encapsulated in liposomes having an approximate size of 100 nm in diameter. Fig. 1b displays a cartoon illustrating the general construction of DLs, with the incorporation of PEGylated lipids for long-term stability after injection (26), rhodamine for fluorescent histological detection, and cholesterol to increase liposome stability. In order to assess the specific MR contrast changes derived from proton transfer for the different groups (NH, NH2, and OH), we measured the magnetization transfer asymmetry parameter, MTRasym=(S−Δω − SΔω/)/S−Δω, comparing the effect of frequency dependent saturation on the water signal (S) on frequencies positive and negative with respect to the water frequency (chosen as 0 ppm) and normalized to water signal on the negative frequency (S−Δω). The in vitro MTRasym values of the three DLs are plotted as a function of saturation frequency are shown in Fig. 1c. Incorporation of Glyc, L-Arg, and PLL into liposomes gave distinguishable MTRasym contrast patterns similar to the free compounds (Supporting Information), showing that the contrast properties can be retained despite encapsulation. Using the DIACEST color spectrum in Fig. 1a, we assigned three artificial colors to the liposomes according to the chemical shift of their main exchangeable protons: red for OH in Glyc liposomes (~0.8 ppm); yellow for NH2 in L-Arg liposomes (~1.8–2.2 ppm); and green for NH in PLL liposomes (~3.6 ppm).
We also tested the stability of MR contrast for the DL preparations over a 5-day period through a dialysis experiment (Supporting Information). The results indicated that the larger CEST agents Glyc and PLL released slower (64% and 37% of initial contrast respectively by day 5) than L-Arg (11% of initial contrast by day 5). These experiments also confirmed that the contrast was sufficient 3 days following synthesis for all three DLs, with MTRasym values of 20%, 22%, and 40% for PLL, L-Arg, and Glyc liposomes, respectively. This can be generated using concentrations of ~50–60 nM of DLs containing 3.5×105, 1.8×108, and 5×105 molecules/DL. The minimum amount for detection was calculated to be 2.2–6.1nM (see Supporting Information).
We tested whether our multi-color MRI acquisition scheme (23) could be used for non-invasive lymphatic imaging using DL. We first quantified the time course of popliteal lymph node (PLN) accumulation with micro-SPECT/CT imaging after injection of 111In-labeled liposomes into the foot pad of three mice. One mouse was treated with a GM-CSF cancer vaccine to enlarge the PLN (20), while the other two were untreated. For all three mice both feet were injected to reduce the number of mice, with the liposome size varied between the left and right foot (see Fig. 2). Approximately 98% of the injected liposomes remained near the injection site, but nevertheless the (draining) lymph nodes were visible on SPECT/CT images as shown in Fig. 2a–e, reaching a plateau at 24 hours post-injection (Fig. 2f). There was no significant difference in PLN uptake between the GM-CSF-vaccinated mouse and the two unvaccinated mice (0.6 vs. 0.5% of total dose, respectively in this limited study), however there was a substantial increase in PLN size (~3–5 times larger) facilitating the MRI studies.
Next we were interested in demonstrating that we could use our DL to visualize lymphatic drainage and homing to PLNs with multi-color MRI, based on our initial studies with 111In SPECT/CT. Fluorescent CEST liposomes were injected into the right footpad of vaccinated C57BL/6.SJL mice, and an MR image from a slice through both PLNs was acquired at 24 hours after injection. The non-injected contralateral site served as control. Images were post-processed to correct for field inhomogeneities (22), and to segment the CEST DL contrast from the inherent tissue contrast using a fluid filtering algorithm (Supporting Information). Fig. 3 shows representative MTRasym images for all three types of liposomes: Glyc (Fig. 3a), L-Arg (Fig. 3b), and PLL (Fig. 3c) at the three different exchangeable proton frequencies of 0.8, 1.8, and 3.6 ppm for Glyc, L-Arg, and PLL, respectively with a white arrow to indicate the side of injection. The measured magnetization transfer ratio curves for both PLN at the injection side (DL(+)) and the contralateral side (DL(−)) displayed an offset-dependent shape due to the inherent MTRasym of the tissue based on competing magnetization transfer effects for semi-solid components. However, as can be seen in the MTRasym image, the MTRasym/T2 overlay, and the quantitative MTRasym plots, the CEST contrast in the DL (+) PLN could be clearly distinguished from the background signal and was significantly higher than that for DL(−) PLN. The shape of the difference curve between the PLNs (Fig. 3a–c, right) corroborates our findings in vitro for Glyc, L-Arg, and PLL DLs (Fig. 1c). The average PLN CEST contrast enhancement, or ΔMTRasym defined by MTRasym(DL+) − MTRasym(DL−), at the resonance frequencies of the exchangeable protons in Glyc (n=3 mice), L-Arg (n=4 mice), and PLL (n=4 mice) were 4.6±1.6% (0.8 ppm), 7.3±0.9%(1.8 ppm), and 10.9±2.4% (3.6 ppm) respectively (Fig. 3e). We calculated the in vivo DL concentration to be 78nM (see Supporting Information). When unlabeled (no DIACEST) liposomes were injected, the resulting images and plots did not show any significant differences in the MTRasym between sides.
Rhodamine labeling of DL allowed us to validate the obtained MRI results. Histology revealed that DLs were indeed localized within the PLN, primarily along the periphery of the node where the afferent lymph duct vessels first enter lymph nodes, consistent with the subcapsular and draining trabecular lymphatic sinus network (Fig. 3f,g). We did not see significant amounts of DL outside of the PLNs (Fig. 3g). Additional staining with anti-CD45.1 antibody (pan-leukocyte marker) revealed there was at least a portion of DLs were not taken up by dendritic or other white blood cells at the time of MR imaging (Fig. 3h).
Each type of liposome contains exchangeable protons with their specific frequency signature (Fig. 1c). In order to demonstrate that these signatures can be discriminated in the same animal in vivo, a second set of mice (n=3) was injected with L-Arg liposomes in the right foot and PLL liposomes in the left foot. We used the algorithm described in detail in Supporting Information to compute MTRasym on a pixel-by-pixel basis. The artificial colors, yellow or green, were assigned to reflect the CEST signal intensities at frequencies of 1.8 ppm and 3.6 ppm, which therefore reflects the species of CEST agents accumulating in one particular pixel. The resulting two-color CEST map overlay on the T2w image clearly shows that the right (R) PLN can be highlighted with an assigned color (yellow), representing the L-Arg type CEST contrast, while the left (L) node voxels display a differentially assigned (green) color, representing PLL type CEST contrast (Fig. 3d images). Furthermore, the mean MTRasym plots for ROIs containing both left and right PLN display different frequency dependent patterns, representing the accumulation of the different types of DL (Fig. 3d plots). Outside of the left PLN there are also noticeable yellow spots due to imperfect filtering of edema which displays a similar CEST frequency profile as L-Arg (Supporting Information).
We have developed three diamagnetic, non-metallic liposomal bioorganic MR contrast agents (Glyc, L-Arg, and PLL DL) that allow the creation of multi-color MR images in living mice. As a proof-of-principle, we have mapped the location of different agents within the lymphatic system. DIACEST agents represent a versatile platform for multi-color MRI because of the diversity of available exchangeable functional groups such as OH, NH, and NH2 in natural biodegradable compounds, including small molecules such as glucose (24) or L-arginine (12) and biopolymers including glycogen (24), polypeptides (23), or proteins (27). We chose three agents, Glyc, L-Arg, and PLL having distinct exchangeable proton frequency signatures. Glyc is an inexpensive, non-charged, natural sugar polymer that is part of the Cori cycle for secretion and uptake of glucose. L-Arg is an FDA-approved nutritional supplement and the most sensitive agent as determined from our previously established DIACEST library in vitro (23). PLL was used in our study because it is a well-characterized model compound which produces strong MR CEST contrast (25,28). In addition, it has been widely used for gene transfection and cell labeling (29).
Our next step was to select a suitable carrier for these compounds. Liposomes are lipid-based vessels composed entirely of natural compounds, which are widely used in nanomedicine (30), including use as contrast agents (31–35). Liposomal encapsulation of our agents was found to produce no significant impact on the characteristic saturation frequency patterns for the agents (Supporting Information). We found the DIACEST contrast to decrease over the course of 5 days due to leakage of CEST agents from the intra-liposomal space (Supporting Information), suggesting DL should only be used for short-term studies. This is expected to be an advantage for applications involving slow clearance, as low molecular weight DIACEST agents should clear more rapidly upon release from liposomes than the liposomes themselves, potentially increasing the biocompatibility of the contrast particles. The clearance of liposomes for lymphatic mapping is slow, as is demonstrated by the SPECT images shown in Fig. 2.
In order to efficiently detect DL in vivo, we developed a specialized acquisition and postprocessing scheme that integrates field mapping correction with fluid, median, and contrast-to-noise ratio (CNR) global filters. Our color-sensitive saturation imaging protocol allows the identification of DL in the acquired images (Fig. 3). In addition, we demonstrated that this imaging scheme allows the recognition of the two different saturation patterns present upon injection of two different DL in the same animal (Fig 3d). As a first demonstration, we assigned colors to voxels based on the relative CEST contrast at two frequencies; however, more elaborate pattern recognition methods through fitting of the entire frequency dependence may be feasible. Extensive studies have shown previously that particle size is critical in determining drainage rate and lymphocyte internalization during passive transport in lymphatics (36–38). Using micro-SPECT and histological analysis, we found that following subcutaneous injection, 100 nm diameter DL were quickly taken up by PLN and exhibited minimal internalization by lymphocytes (Fig. 3e), which is consistent with previous reports (37). Intratumoral injection of slightly larger particles may result in significant intracellular uptake (39). The size of these liposomes also plays a critical role in CEST contrast by affecting the intra-and extra-liposomal water exchange (40,41). According to our previous study (41), water exchanges across the lipid bilayer at a rate of ~200 s−1 for 100 nm liposomes, which is on the same order as the exchange rate between intra-liposomal water and CEST agents (28). For the DL used in this study, the above considerations were taken into account.
DIACEST peptides or liposomes are not the only MR contrast agents that can be used to perform multi-color imaging. Paramagnetic CEST agents (PARACEST) such as lanthanide chelates (42–44), polymers (45–47), and nanoparticles (48,49) could potentially also be employed. However, our DIACEST agents are composed of non-metallic, natural compounds, which may have a safety advantage over using PARACEST agents. It has also been shown that the chemical shift spectrum of 19F MRS/MRSI can be used to perform two-color cell tracking when two types of perfluorocarbon are used (i.e. PFOB and PFPE nanoparticles) (50). However, our DL-enhanced imaging generates contrast based on imaging on water protons, the largest source of MR signal in tissue. This is important because 1H imaging using water is the most widely used MRI technique and does not require special coils for detection.
In conclusion, we have created a highly sensitive, biodegradable and multicolor DIACEST MRI labeling system using bioorganic molecules such as L-Arg, PLL, and Glyc without a paramagnetic or superparamagnetic component. This enabled us to produce the first in vivo multi-color MR images. We applied our DL contrast agent platform to lymph node imaging by directly monitoring the liposomal uptake in lymph nodes. The current study opens the door to future molecular and cellular imaging studies where simultaneous two or three-color MRI detection offers new ways of studying the spatial and temporal dynamics of complex biological systems.
We thank Enzo Terreno for valuable discussions. This work was supported by NIH grants R21 EB005252, EB008769-01A1, NS065284-01, R01 EB012590 and K01 EB006394.