The consequences of atherosclerosis represent a critical public health concern in Western countries and are likely to become the major public health problem in the developing world in the near future (1). One of the key initial steps in atherogenesis is the subendothelial retention of apolipoprotein B (ApoB)-containing lipoproteins (2). Plaque formation eventually leads to luminal narrowing that can be detected by magnetic resonance imaging (MRI). Early detection of atherosclerotic plaque, however, is challenging but could aid in the prevention of acute cardiovascular events such as stroke or heart attack. Another important application could be the assessment of the efficacy of treatment strategies targeted at atherosclerotic plaque stabilization or regression. The use of MR contrast agents improves the detection of morphological and functional abnormalities within the vessel wall (3–6). Gadolinium (Gd)-based contrast agents are commonly used in MRI to improve contrast between tissues (7) and several contrast agents targeting the atherosclerotic plaque are currently under investigation (8–10). They have been designed to detect different cellular and molecular plaque features, including macrophages, neovessels, activated endothelium, or thrombus in advanced lesions. This variety is due to the highly complex progression of atherosclerosis, which results in very heterogeneous plaques with highly fibrous structures, calcifications, inflamed tissue, lipid pools, and combinations of the above (11).
In the current study, we aimed to image the lipoproteins of lipid deposits in the plaque by conceptually basing our agent on the human apolipoprotein A-I mimetic peptide D-4F (12). D4-F is an amphiphatic peptide with both a polar and a nonpolar face. This structure allows it to mimic the lipid-binding properties of apolipoprotein A-I (12–14). The lipophilic side of D-4F contains four hydrophobic phenylalanine residues (15). We chose tyrosine as a simple hydrophobic moiety. The tyrosine groups are also easily quantifiable by ultraviolet (UV) spectroscopy for chemical characterization. Therefore, we hypothesized that functionalized structures, such as nanoparticles, exposing multiple tyrosine groups at their surface could be used to identify lipid-rich areas in atherosclerotic plaques. Previously reported targeted contrast agents have used antibodies (16), antibody fragments, proteins (17), peptides, peptidomimetics (18) and small molecules (19) for targeting purposes. In comparison, using only tyrosine residues on the surface of the contrast agent is a simple approach.
The MR contrast agent reported here is based on a platform we introduced recently, i.e., polyethylene glycol (PEG)-micelles that include gadolinium diethylenetriamine pentaacetic acid (Gd-DTPA)-based lipids (20). This platform has been used to image macrophages in plaque and oxidized low-density lipoprotein (LDL) and has a greater half-life than regular micelles (21, 22) due to the protective “stealth” effect of the outer shell of PEG chains, which reduces recognition by opsonins and consequent clearance (22, 23).
The aim of the current study was to prepare, characterize, and evaluate in vivo the efficacy of tyrosine micelles using an apolipoprotein E knockout (ApoE−/−) mouse model on a high-cholesterol, Western-type diet (24). MRI of the abdominal aorta was performed on a 9.4T system, before and at several time points after injection of micelles. As controls, the tyrosine micelles were injected in wild-type (WT) mice, while nonderivatized micelles were applied to ApoE−/− mice as well.
MATERIALS AND METHODS
1,2-Distearoyl-sn-glycero-3-phosphoethanolamine-N-[carboxy(polyethylene glycol)2000] (ammonium salt) (carboxy-PEG2000-DSPE), 1,2-distearoyl-sn-glycero-3-phosphoethanolamine-N-[methoxy(polyethylene glycol)-2000] (ammonium salt) (PEG2000-DSPE), and 1,2-dipalmitoyl-sn-glycero-3-phosphoethanolamine-N-(lissamine rhodamine B sulfonyl) (ammonium salt) (Rhodamine PE) were purchased from Avanti Polar Lipids (Alabaster, AL, USA). L-tyrosine methylester hydrochloride, 1-ethyl-3-(3-dimethylaminopropyl) carbodiimide hydrochloride (EDC), N-hydroxysuccinimide (NHS), pyrene, sodium hydroxide (NaOH), and hydrochloric acid (HCl) were purchased from Sigma (St. Louis, MO, USA). Gd-DTPA-bis-stearylamide (Gd-DTPA-BSA) was from iQSynthesis (St. Louis, MO, USA).
For fluorescent microscopy, monoclonal anti-mouse CD68 conjugated AlexaFluor® 488 was purchased from Serotec (Raleigh, NC, USA). Rabbit anti-mouse collagen I polyclonal antibody was from Millipore (Billerica, MA, USA), rabbit polyclonal α-smooth muscle actin and rabbit polyclonal anti-ApoB whole serum were from Abcam (Cambridge, UK), and goat anti-mouse decorin monoclonal antibody was from R&D Systems (Minneapolis, MN, USA). Secondary antibody goat anti-rabbit AlexaFluor® 488 and donkey anti-goat AlexaFluor® 488 were both from Molecular Probes, Invitrogen (Carlsbad, CA, USA). Nuclei were stained with 4′,6′-diamino-2-phenylindole (DAPI; Invitrogen).
The Institutional Animal Care and Use Committee of Mount Sinai School of Medicine approved animal procedures. All mice were purchased from Jackson Laboratories (Bar Harbor, ME, USA). Male ApoE−/− mice (N = 22) with a mean age of 11 ± 2.6 months were used for this study. Age-matched WT were used as a negative control (N = 3). The ApoE−/− mice were fed a Western diet (WD) for at least 24 weeks (mean 36.5 ± 11 weeks) before the experiment (Diet D01022601, containing 21% fat, 0.15% cholesterol; Research Diets, Inc., New Brunswick, NJ, USA).
Micelles were prepared by mixing 0.05 mmol Gd-DTPA-BSA, 0.018 mmol PEG2000-DSPE, and 0.032 mmol carboxy-PEG-2000-DSPE in a round-bottom flask. The lipids were dissolved in chloroform/methanol 4:1 (v/v) and 0.1% Rhodamine-PE was added as a fluorescent label. A thin film was obtained by slowly removing the solvents under vacuum at 65°C. The film was hydrated by adding 4 ml warm phosphate-buffered saline (PBS) buffer (70°C) to the film, which was maintained at the same temperature. The solution was stirred at 70°C until a clear solution was formed.
Tyrosine residues were coupled to the surface of the micelles, as shown in Fig. 1. L-tyrosine methylester was coupled to the carboxyl groups on the carboxy-PEG lipids by adding 0.046 mmol EDC, 0.1 mmol NHS, and 0.03 mmol L-tyrosine methylester to the micelle solution. The micelle solution was cooled to 4°C and the pH was decreased to 4.2 by the addition of diluted HCl. The EDC was added and the pH increased to 6.8 (via addition of NaOH). The NHS and amino acid was then added and the pH raised to 10 (via addition of NaOH). The sample was mixed for 30 min until the reaction was stopped by decreasing the pH to 7 (via addition of HCl). Purification was achieved by washing the functionalized micelles with PBS three times (centrifugations were performed using Vivaspin® tubes, Sartorius Corporation, Edgewood, NY, USA; cutoff 10 kDa, at 4°C, for 20 min, at 4000 rpm).
Unmodified micelles were made as described above. They were composed of 50% Gd-DTPA-BSA, 50% PEG-2000-DSPE, and 0.1% rhodamine-PE. These percentages are based on moles of lipids.
The size of the micelles was determined by dynamic light scattering using an HPPS device (Malvern Instruments, Malvern, UK). The mean diameter at 25°C was determined based upon both volume and number weighting. Relaxivity and Gd content were measured using a Minispec mq60 NMR Analyser (Bruker, Karlsruhe, Germany), as described previously (25).
Micelle stability over time was assessed by studying the evolution of particle size distribution by dynamic light scattering over time. Critical micelle concentrations (CMC) were determined in serum or PBS as a measure of micelle stability, using pyrene as a fluorescent probe as previously described (26). Briefly, aliquots of pyrene solutions in acetone were added to vials, and acetone was allowed to evaporate. Then 1 ml of tyrosine micelles at different concentrations in PBS/serum was added to the vials, giving a final concentration of 6 × 10−7 M of pyrene. The solution was incubated overnight at 50°C to reach pyrene equilibrium. Fluorescence spectra was recorded on a F-2000 fluorescence spectrophotometer (Hitachi) using λex = 339 nm and λem = 350–450 nm. Fluorescence intensity at 375 nm was plotted as a function of tyrosine micelle concentration.
The L-tyrosine methylester content was assayed by UV spectrometry at 280 nm from the second-order derivative of the spectra (UVmc2, Safas, Monaco) and by comparison to a calibration curve.
The blood half-life of tyrosine micelle was determined in ApoE−/− mice (N = 6) as previously described for unmodified micelles (22, 27). After tyrosine micelle injection, blood was drawn into heparinized blood collection tubes via heart puncture at different time points postinjection (pi), from 5 min to 24 h pi.
The abdominal aorta was scanned using a 9.4T MR imager (Bruker, Karlsruhe, Germany) equipped with a 30-mm transmit/receive birdcage coil. Mice were anesthetized continuously with a mixture of oxygen and isoflurane (0.5–2%). The contrast agents were administered to the animals (60 μmol Gd/kg) via injection into the tail vein. Imaging was performed preinjection and 6 h, 24 h, and 48 h pi.
Mice were scanned using the following protocol. The abdominal aorta was identified in a coronal section on a localizing sequence. T1-weighted (T1W) MRI was performed using a black-blood turbo spin-echo sequence with arterial saturation bands. Sixteen contiguous 500-μm-thick slices with a microscale in-plane resolution of 101 μm were acquired using a spin-echo sequence with matrix size = 256 × 256. The repetition time (TR) and echo time (TE) for the T1W images were 800 and 8.6 ms, respectively. An inflow saturation band of 3 mm was used with a slice gap of 3 mm for additional arterial luminal flow suppression. Sixteen signal averages were used for a total imaging time of 55 min per scan. A saturation pulse was used to eliminate signal from fat tissue and better delineate the boundary of the aortic wall and minimize chemical-shift artifacts.
At each time point the slices were matched to the baseline precontrast scan by using the unique vertebral and paraspinous muscular anatomy as landmarks. Five slices per study were chosen that were of good image quality, without or with little flow or motion artifacts at the four time points. The signal intensity of the atherosclerotic plaque and the paraspinous muscle were measured using eFilm Workstation™ software (Merge Technologies, Inc., Milwaukee, WI, USA). Since the plaques of mice are particularly heterogeneous (28), only the enhancing areas of the aortic wall were used to quantify the signal. The standard deviation (SD) of noise from the background was also obtained.
For each slice, the percent enhancement (% ENH) (25) relative to the muscle, was calculated (according to the equation below) in order to evaluate the enhancement of the aortic wall.
where SI is the signal measured in the tissue, and CNR is the contrast-to-noise ratio.
A group of ApoE−/− mice was injected with tyrosine micelles (N = 8). As controls, a group of WT mice (N = 3) was injected with tyrosine micelles and imaged at the same time points. Two groups of ApoE−/− mice were injected with unmodified micelles and imaged either at 6, 24, and 48 h (N = 5) or at earlier time points of 30 min and 2 h (N = 3).
Conventional fluorescence microscopy was performed on sections of abdominal aortas of ApoE−/− mice that matched the areas of enhancement in MR images from ApoE−/− mice. Mouse aortas were removed 6 h after the micelle injection, embedded in TissueTek® (Sakura, Torrance, CA, USA), immediately placed on dry ice, and kept at −80°C until sections were obtained. Frozen sections of 7 μm were fixed in PBS buffered 4% paraformaldehyde solution for 15 min. After fixation, sections were rinsed two times with PBS 1× (15 min each). Then, sections were blocked with PBS containing 5% goat serum or BSA for 30 min at room temperature (RT) for further staining.
After blocking, sections were incubated with one primary antibody against CD68 macrophages (dilution 1/20), decorin (1/15), collagen I (1/40), ApoB (1/50), or α-smooth muscle actin (1/50) in the blocking solution for 2 h at RT. After washing, the sections were stained with a secondary antibody tagged with a fluorophore at RT for 1 h; except in the case of the macrophage staining—that primary antibody was fluorescently tagged. They were both diluted to 1/500. Sections were washed then stained for nuclei with DAPI and finally were mounted with Dako fluorescent mounting medium.
Finally, several types of control experiments were performed for fluorescence microscopy. Matched ApoE−/− mice were sacrificed and stained with secondary antibodies alone. The sections demonstrated no significant fluorescence (data not shown). Controls were also performed to confirm that the red fluorescence obtained from aortic sections after injection of tyrosine-grafted micelles is unrelated to autofluorescence.
Fluorescence microscopy was performed using a Nikon Eclipse E400 (Nikon, Japan). Pictures were analyzed with Archimed 5.2.2 software (Microvision, Evry, France). Gain and exposure time were independently determined for each set of antibodies.
Statistical analysis was performed using SPSS 14.0 for Windows. A paired t-test analysis and a one-way analysis of variance (ANOVA) with Bonferroni's multiple comparison test were performed to determine the significance of wall enhancement of the experimental groups as compared to the control group. P values <0.05 were considered significant.
Dynamic light scattering revealed that the micelles (Fig. 1) composed of 50% PEG-lipids and 50% lipids bearing Gd-DTPA had a mean hydrodynamic diameter below 20 nm. Micelles that had tyrosine conjugated to the surface did not significantly differ in size and relaxivity from the unmodified micelles (Table 1). Unmodified and tyrosine micelles showed a single peak (>99%) on number-weighted chromatograms, indicative for a narrow size distribution. The values found for the unmodified and tyrosine micelle sizes were 12 ± 2.7 nm and 14.6 ± 2.5 nm, respectively. The longitudinal relaxivity (r1) was not significantly affected by grafting of tyrosine on the PEG-chains, as r1 = 10.1 ± 1.5 mM−1s−1 for unmodified micelles, while r1 = 10.3 ± 0.9 mM−1s−1 for tyrosine micelles. Moreover, both micelles were found to be stable for over 1 month at RT via negligible changes in size and relaxivity.
Table 1. Physical Properties of Unmodified Micelles and Micelles Bearing Tyrosine Residues
[Tyrosine] (% PEG chains)
Particle size (nm) (z-average ± disp.)
Half-life in ApoE−/− mice (h)
disp = dispersity.
12 ± 2.7
10.1 ± 1.5
15% ± 4%
14.6 ± 2.5
10.3 ± 0.9
Unmodified micelles have been described (23, 29) with a very low CMC of about 10−5 M. Tyrosine micelles had a CMC both in PBS and in serum of 2.5 × 10−6 M. These in vitro experiments indicate that both types of micelles will maintain their integrity even at high dilution. In comparison, the initial concentration of micelles in the blood of mice after injection is 1.8 × 10−3 M; i.e., three orders of magnitude greater than the CMC. Mice, which have a blood volume around 2 ml, were injected with 105 μl of 34 × 10−3 M micelle. Tyrosine conjugation yield was established to be 25%, resulting in 15 ± 4% of the PEG chains conjugated with tyrosine.
After injection of micelles in ApoE−/− mice, we found that adding tyrosine to the micelles increased the blood half-life from 2 h for the unmodified micelles to 21 h for tyrosine micelles (data not shown).
Enhancement of Tyrosine-Grafted Micelles
The micelles were intravenously applied to an ApoE−/− mouse model of atherosclerosis that had been on a high-fat and high-cholesterol diet, and to WT mice on a normal diet to evaluate the efficacy of these contrast agents for early atherosclerosis imaging. Mice were injected with tyrosine micelles or unmodified micelles at a dose of 60 μmol Gd/kg via the tail vein. Before injection, all mice underwent a baseline scan for comparison, which was followed by scans at 6, 24, and 48 h pi.
A group of ApoE−/− mice (N = 8) was injected with tyrosine-grafted micelles. The administration of tyrosine micelles in the ApoE−/− mice resulted in a significant increase in the vessel wall MRI signal (Fig. 2) at 6 h and 24 h pi. At 48 h pi, very little enhancement could be observed, indicating that the agent had left the plaque at this time point and that repeat imaging would be possible with this agent in this relatively short timeframe. The signal coming from the vein wall is probably due to circulating contrast agent or to blood-flow artifact, since blood-flow saturation bands were only used on the arterial side.
The mean wall enhancement due to tyrosine-grafted micelles in the aorta wall of ApoE−/− mice and unmodified micelle in ApoE−/− (N = 5) was determined at 6, 24, and 48 h pi (Fig. 3). This wall enhancement with the tyrosine-grafted micelles was very high at 6 h (104% ± 58) and remained elevated until 24 h (61% ± 41) pi. Thereafter, the enhancement decreased and the vessel wall MR signal intensity returned to baseline after 48 h pi. Analysis of the data obtained from the injection of unmodified micelles in ApoE−/− mice did not show any significant enhancement at 6 or 24 h, with the mean values for these time points being 35 ± 13% ENH and 13 ± 17% ENH, respectively (Fig. 3). Statistical testing performed on the unmodified micelles data showed no significant enhancement at any time point.
At 6 h and 24 h, the enhancement obtained with tyrosine grafted micelles was significantly different from the enhancement from the unmodified micelles in the ApoE−/− group (P < 0.05, ANOVA).
As additional controls, a group of WT mice (N = 3) was imaged before and after the injection of tyrosine micelles at the same time point (6, 24, and 48 h pi). There was no apparent enhancement at any time point, the %ENH being not significantly different from background signal.
CNR values were also measured before the injection, 6, 24, and 48 h after the injection. Preinjection CNR values were not different among the different groups of mice (P > 0.05) for tyrosine micelles in WT mice 1.03 ± 0.33, unmodified micelles in ApoE−/− mice 0.99 ± 0.25, and for tyrosine micelles in ApoE−/− mice 0.8 ± 0.23. In ApoE−/− mice injected with tyrosine micelle, the CNR at the 6 h pi time-point (1.60 ± 0.45) was significantly different from the CNR in the preinjection images (0.80 ± 0.2), confirming that the signal increased in the aortic wall. The CNR at the 6 h pi time point for tyrosine micelles in WT mice or for unmodified micelles in ApoE−/− mice was not significantly different from pre-CNR, at 0.91 ± 0.21 and 1.26 ± 0.35, respectively.
A group of ApoE−/− mice (N = 3) was also imaged at 30 min and 2 h after the injection of unmodified micelles, time points which represent, respectively, one-quarter- and one-half-life of the unmodified micelles. This control group was included as the blood clearance kinetics of the unmodified micelles was approximately 12 times faster than that of the tyrosine micelles and, therefore, in terms of half-lives, 30 min and 2 h are the equivalent time points for the unmodified micelles to the 6 h and 24 h time points for the tyrosine micelles. This resulted in no apparent enhancement, with % ENH values for the 30 min and 2 h time points of 5 ± 20% and 1 ±19%, respectively.
To examine which components within plaque are preferentially bound by tyrosine-grafted micelles, immunofluorescence staining was performed for various plaque components on aortic sections. To this aim, we first investigated whether tyrosine micelles were accumulating because of cellular uptake or by specific binding of components found in atherosclerotic plaques. At 6 h, no uptake of the tyrosine micelles in the macrophages or macrophage-derived foam cells was revealed (Fig. 4). Staining for smooth-muscle cells with α-actin did not evidence any micelle uptake (Fig. 4). This suggests that tyrosine-grafted micelles remained restricted to the extracellular space of the atherosclerotic plaque and that there was no cellular uptake of the contrast agent. Subsequently, decorin, a matrix proteoglycan strongly expressed in atherosclerotic plaque, was investigated (30–32). Some colocalization was observed between the decorin and the distribution of tyrosine-grafted micelles in the atherosclerotic plaque (Fig. 5). Furthermore, we stained for collagen I, a major extracellular matrix component, but no colocalization was observed between collagen I and the tyrosine-grafted micelles, since collagen I–rich areas (green) did not colocalize with tyrosine micelle rich areas (Fig. 5). Lastly, LDL, the major physical cholesterol transporter and a major component of the fatty streak found in the atherosclerotic plaque, was assessed with immunofluorescence. An association between ApoB and tyrosine micelles was clearly found by comparing the fluorescence from these two components in the plaque (Fig. 5), which is a strong indication for the affinity of the tyrosine micelles for the lipoprotein-rich lipid cores in plaque.
High spatial resolution, multisequential MRI allows the depiction of various components of atherothrombotic plaque, including fibrous tissue, lipids, calcium, and thrombus formation (3, 33–35). The development of atherosclerotic plaque is initiated by the deposition of lipid-rich ApoB-containing lipoproteins in susceptible but still prelesional areas of the arterial wall. The lipid deposition in the vessel wall occurs significantly before macrophage infiltration (2). Complications of atherosclerosis, such as plaque rupture, are the most common cause of death in the Western world and are becoming a major health concern in developing countries (1). Previous reports have demonstrated a strong correlation between the lipid core size, thin fibrous cap, and plaque instability (36–38). Binding the lipid core in the atherosclerotic plaque could help the diagnosis of plaque vulnerability. Such a contrast agent could also be used to identify plaque stabilization with lipid-lowering therapy (39–42).
We report here a tyrosine-functionalized paramagnetic contrast agent that binds the lipid-rich areas in the atherosclerotic plaque of ApoE−/− mice. Lipophilic properties, small particle size, and prolonged circulation allow our contrast agent to penetrate and accumulate within the plaque after intravenous administration. As a result, tyrosine micelles efficiently highlight the atherosclerotic plaque in ApoE−/− mice, providing a high and specific enhancement in the plaque. In contrast, no or significantly less pronounced vessel wall enhancement was observed in the control groups. Simply targeting the micelles with an amino acid residue could represent a significant advance over targeting using more complex molecules such as antibodies, proteins, or peptides, which have been applied previously (16–19).
The observed strong vessel wall MRI enhancement occurred at 6 h, which is significantly earlier than that found for similar contrast agents used in the past (6, 20, 25, 43). The evaluation of the distribution of tyrosine micelles revealed no cellular uptake. Tyrosine-functionalized micelles are mainly observed within the extracellular matrix and colocalize well with ApoB, and also slightly with decorin. Proteoglycans like decorin have been shown to be critical in the retention of atherogenic lipoprotein inside the vessel wall (30–32, 44). These results strongly suggest that tyrosine micelles bind the lipid-rich areas in atherosclerotic plaques.
Tyrosine micelles have a longer half-life compared to unmodified micelles (21 h vs. 2 h in ApoE−/− mouse model), which leads to more efficient vessel wall binding. The increase in the half-life of the tyrosine micelles suggest that they are transported by interactions with serum components. This carrier might also have a role in the uptake/exchange of tyrosine micelles in the atherosclerotic plaque and would allow local binding of tyrosine micelles. This could be explained by the binding of the tyrosine micelles to plasma lipoproteins, like high-density lipoprotein (HDL) and/or LDL in the blood, which subsequently carry the paramagnetic micelles into the atherosclerotic plaque. Further experiments will be done to identify the serum component(s) responsible for this effect.
Binding the lipid-rich areas in atherosclerotic plaque could help to discriminate the unstable plaque from stable plaque, as a large lipid core in combination with a thin fibrous cap strongly correlates with plaque instability (4, 45). Gadofluorine M, a nonpegylated amphiphilic micellar paramagnetic contrast agent, has been shown in vitro and in vivo to target extracellular matrix proteins of atherosclerotic plaque without direct interaction of lipid-rich areas (5, 6). Conversely, our novel tyrosine-based nanoparticles can effectively interact with lipid-rich areas and thus could be used in conjunction with Gadofluorine M to determine the ratio between lipid-rich core and fibrous cap as a tool to detect plaque vulnerability.
Furthermore, the evaluation of the dynamic changes of the lipid content of the atherosclerotic plaque could represent a very valuable tool to assess the effect of lipid-lowering strategies, like statin therapy (39–42).
The combination of a high relaxivity, paramagnetic, and versatile contrast agent that preferentially binds to lipophilic moieties led to an agent that can effectively detect lipid-rich areas in the atherosclerotic plaque by MRI. Tyrosine micelles that bear gadolinium complexes may enable earlier detection of atherosclerotic plaques using MRI, since lipid deposition in the vessel wall occurs at very early stages of the development of atherosclerosis. In addition, the detection of lipid-rich areas of the atherosclerotic plaque could allow the evaluation in other models of plaque vulnerability and the assessment of the effects of antiatherogenic therapies.
This study was supported in part by the NIH/NHLBI (grants R01 HL71021 and R01 HL078667, both to Z.A.F.).