Atherosclerosis, the thickening and stiffening of arteries, is causative of myocardial infarction and stroke when arteries become blocked after hemorrhage or thrombus formation (1). The accumulation of harmful lipoproteins in the vessel wall and the resulting local responses are responsible for the development of atherosclerotic plaques (2). Lipoproteins that have accumulated in the vessel wall of these arteries are phagocytosed by macrophages. Eventually the macrophages are converted into foam cells, and an early atherosclerotic plaque is formed. The plaque is stabilized by the migration of smooth muscle cells, resulting in the formation of a fibrous cap (1). The thickness of the cap, in relation to the lipid core and the composition of the plaque, is believed to be indicative of the risk of plaque rupture, which is the main cause of the above acute clinical syndromes (3, 4). In advanced stages of atherosclerosis, plaque formation results in remodeling of the artery (e.g., by narrowing of the lumen). For this reason, conventional diagnostic methods (mainly angiography) have been designed to visualize the lumen of the arteries (5, 6). The identification of atherosclerosis with this method has two major drawbacks. First, if plaque formation causes the outward remodeling of arteries, narrowing of the lumen will not be observed (7). More importantly, using this technique, plaque composition and thus plaque stability cannot be characterized. Since plaque composition is a much better indicator for plaque stability and the risk of a clinical event, the development of diagnostic imaging methods in the past two decades has shifted toward visualizing plaque composition (1, 8). Plaque composition can be characterized noninvasively with magnetic resonance imaging (MRI) (9, 10) and may lead to improved diagnosis and prognosis. Furthermore, MRI allows the study of the progression or regression of the disease over time, e.g., to monitor the effect of lipid-lowering therapies (11, 12).
A relatively new and emerging diagnostic method for imaging atherosclerosis is MRI based on targeted contrast agents (CAs) for molecular imaging. The aim of the technique is to visualize pathological processes related to and associated with atherosclerosis at the cellular and molecular levels for improved characterization of disease progression and characterization of plaque phenotype, particularly stability (13, 14). Molecular imaging of macrophages may be of great use in predicting the severity of the disease, since macrophages play a predominant role in plaque formation and progression (3).
MRI, as a molecular imaging modality, has two important limitations that have to be taken into consideration. The first limitation is related to the relatively low sensitivity of MRI as compared to nuclear methods (15). Second, although MRI outclasses the spatial resolution of most noninvasive imaging modalities (e.g., positron emission tomography [PET] and single-photon emission computed tomography [SPECT]), the resolution is limited to approximately 0.5 mm on clinical systems and to 0.05 mm on a high-field experimental system. In order to deal with these limitations, sophisticated contrast materials need to be developed for molecular MRI (15).
In this study we developed and applied a macrophage-specific CA for MRI based on paramagnetic and pegylated immunomicelles equipped with an optical label for parallel detection with fluorescence imaging and microscopy. The micelles were prepared from a gadolinium diethylene triamine pentaacetic acid (Gd-DTPA)-based amphiphile and a pegylated lipid (16). Specificity for macrophages was introduced by covalently conjugating antibodies specific for the macrophage scavenger receptor (MSR, CD204) to the paramagnetic micelles (17). For fluorescence imaging the micelles were either equipped with rhodamine or prepared such that a fraction carried a quantum dot (QD) in the micelle core. QDs are semiconductor nanocrystals in the size range of 2–6 nm that upon excitation emit photons of a defined wavelength depending on the size and composition of the QD. QDs have several important advantages over organic fluorophores, including a broad excitation window, a narrow and tunable emission profile, and resistance to photobleaching (18, 19). Furthermore, QDs exhibit a high quantum yield. These properties make QDs excellent tools for molecular imaging (20), especially when combined with a label for a complementary imaging modality, such as MRI (16).
High-resolution MRI of the abdominal aorta was performed prior to and 24 h after the macrophage-targeted CA was applied in an apolipoprotein E knockout (apoE-KO) mouse model of atherosclerosis. Nontargeted micelles served as a control agent. The uptake of the CA was also visualized with optical imaging in excised and intact aortas of the animals. Last, 10-μm-thick sections of the aortas were assessed by histopathology and fluorescence microscopy.
MATERIALS AND METHODS
Paramagnetic micelles were prepared by lipid film hydration. In short, PEG-DSPE (1,2-distearoyl-sn-glycero-3-phosphoethanolamine-N-[methoxy(polyethylene glycol)-2000]) and Gd-DTPA-BSA (Gd-DTPA-bis(stearylamide) were dissolved in chloroform/methanol (10/1) in a 1:1 ratio and evaporated to dryness by rotary evaporation at 40°C. The obtained film was further dried under nitrogen for 15 min and heated to 65°C. Next, the film was hydrated with a 4-2-hydroxyethyl-1-piperazineethanesulfonic acid (HEPES) buffer (20 mM HEPES, 135 mM NaCl, pH 6.7) at 65°C. This suspension was vigorously vortexed and kept at 65°C until a clear solution was obtained, indicating a small size of the formed aggregates. For micelles that were used to attach antibodies to the distal end of the PEG-chains, 10% PEG-DSPE was replaced by Mal-PEG-DSPE (1,2-distearoyl-sn-glycero-3-phosphoethanolamine-N-[maleimide(poly(ethylene glycol))2000]). For fluorescence microscopy, 0.1 mol percent of rhodamine-PE was added to these preparations. The longitudinal and transverse relaxivity r1 and r2 in HEPES buffer at 60 MHz and 40°C were 10 and 15 mM–1 s–1, respectively.
MRI-detectable QD micelles were prepared as described previously (16). In short, high-quality core shell tri-n-octylphosphine oxide/hexadecylamine (TOPO/HDA) capped QDs and PEG-DSPE, Mal-PEG-DSPE, and Gd-DTPA-BSA in a molar ratio of 0.4/0.1/0.5 were dissolved in chloroform/methanol (10/1). The solvents were evaporated gently until a dry film of lipids and QDs was obtained. Thereafter, the lipid film was heated to 70°C and hydrated with a HEPES buffer (20 mM HEPES, 135 mM NaCl, pH 6.7) of the same temperature. This suspension was heated and vigorously stirred until a clear suspension was obtained. The resulting suspension consisted of a mixture of small micelles(QD) containing a single QD per micelle, and larger aggregates containing multiple QDs per micelle. The aggregates were removed from the mixture by centrifugation. Size determinations with Dynamic Light Scattering of a suspension of micelles and a suspension of micelles(QD) are depicted in Fig. 1a. The mean size of both nanoparticles was 15 nm, with a narrow size distribution. The micelle(QD) suspension was then evaluated for its optical properties, as shown in Fig. 1b and c. Figure 1b is a representative image of the micelles(QD) under daylight illumination, and Fig. 1c shows the effect of illumination with ultraviolet (UV) light at 254 nm. The green luminescence from the suspension is clearly visible.
The longitudinal and transverse relaxivity (r1 and r2) in HEPES buffer at 60 MHz and 40°C were 13 and 19 mM–1 s–1, respectively.
The monoclonal rat anti mouse CD204 (Serotec, Raleigh, NC, USA) antibody against the MSR was coupled to micelles and QD-micelles containing Mal-PEG2000-DSPE by a sulfhydryl-maleimide coupling method as described previously (21, 22). In short, CD204 monoclonal antibody (5 mg/ml) was modified with N-succinimidyl S-acetylthioacetate (8:1 SATA:antibody mole:mole ratio) by incubation for 45 min on a roller-bench at room temperature. Free SATA was separated from the antibody/SATA solution by centrifugation on a Vivaspin concentrator with a 30-kD MW cutoff filter, followed by thorough washing four times with HBS. The SATA-derivatized antibody was deacetylated by incubation with a hydroxylamine solution for 1 h at room temperature. The activated antibody was added to the Mal-PEG2000-DSPE containing micelles and QD-micelles. This preparation was stored at 4°C under N2 overnight. Uncoupled antibody was separated from immunomicelles by washing twice on a Vivaspin concentrator.
Hereafter, we refer to the bare micelles as “micelles” and the antibody-conjugated micelles as “Ab-micelles.” When rhodamine was used as a fluorescent label the micelles are specified as “Ab-micelles(Rh),” while QD-labeled Ab-micelles are specified as “Ab-micelles(QD).”
Ten 13-month-old apoE-KO mice (Jackson Laboratory, Bar Harbor, ME, USA) were used for this study. The apoE-KO animals were fed a Western diet (WD, containing 21% fat, 0.15% cholesterol; Diet D01022601, Research Diets, Inc., New Brunswick, NJ, USA) for 20 weeks. The animals were housed in the Center for Laboratory Animal Science, and the protocols were approved by the Animal Use and Care Committee of the Mount Sinai School of Medicine. All micelle formulations were administered at a dose of 0.075 mmol Gd/kg via intravenous injection into the tail vein. Four mice were administered control micelles, and six mice were administered Ab-micelles. In the latter experiment two mice received Ab-micelles(Rh), and four mice received Ab-micelles(QD).
In vivo MRI was performed with a Bruker 9.4T, 89-mm-bore system operating at a proton frequency of 400 MHz (Bruker Instruments, Billerica, MA, USA). A gradient insert (inner diameter = 75 mm) was used that was capable of generating a maximum of 50 Gauss/cm. MRI parameters were optimized to visualize the aortic wall. The mice were anesthetized with continuously inhaled isoflurane (1.5–2%) and placed head-up in a vertical 30-mm birdcage coil. Constant body temperature of 37°C was maintained using a thermocouple/heater system. Cardiac and respiratory motion of the aorta in the thorax mandated a study of the infrarenal abdominal aorta. The abdominal aorta was first identified in an approximately coronal section using a localizing sequence. Twenty-one contiguous, 0.5-mm-thick axial slices were obtained, starting immediately caudal to the aortic bifurcation and progressing in the cranial direction. A 256 × 256 matrix spin-echo sequence was used for 2D axial imaging. The resultant voxel size was 0.109 × 0.109 × 0.5 mm3. The repetition time (TR) and echo time (TE) for T1-weighted imaging were 800 and 9 ms, respectively. Fourteen signal averages were used. A saturation pulse was used to eliminate signals from fat tissue to better delineate the outer boundary of the aortic wall and minimize chemical shift artifacts. The total imaging time per scan was 45 min. T1-weighted images were acquired before and 45 and 90 min postinjection, followed by recovery of the animals from anesthesia. At 24 h postinjection the mice were anesthetized again and high-resolution T1-weighted images were acquired according to the same protocol.
Images were analyzed using Mathematica (Wolfram Research, Inc., USA). For each time point, typically six to 12 slices perpendicular to the aorta were analyzed. Coregistration of the pre- and postcontrast images was performed by accurately positioning the animal and determining the exact same slice position after verifying the shape and position of the spinal cord. Atherosclerotic plaques were then identified on the precontrast images. Next, circular regions of interest (ROIs) were drawn on the pixels in the thickened wall directly surrounding the lumen of the aorta (Iwall). Second and third ROIs were drawn in a portion of the surrounding muscle tissue (Imuscle) and in the aorta lumen. Furthermore, an ROI was placed outside the animal to measure the noise level (Inoise). The contrast-to-noise ratio (CNR) was defined as CNR = (Iwall – Imuscle)/Inoise, which is a measure for how well the vessel wall can be discriminated from the surrounding tissue. Signal enhancement (SE) was defined as the relative difference between the signal intensity in the T1-weighted images before and after CA injection: SE = (Ipost – Ipre)/Ipre. Post-CA-injection images were color-coded as follows: First an ellipsoid-shaped ROI with a thickness of 6 pixels was drawn in the precontrast image in the vessel wall surrounding the lumen of the abdominal aorta. This ROI was then warped and translated to match the vessel wall in the corresponding post-CA-injection image.
Data are presented as the mean ± SD. For differences between the two groups, Student's t-test was used. A value of P < 0.05 was considered statistically significant.
Confocal Fluorescence Microscopy and Histopathology
Abdominal aortas matching the MRI area were removed from apoE-KO mice and 8-μm-thick frozen sections were fixed in PBS-buffered 4% paraformaldehyde. The sections were then incubated with Alexa Fluor 647® conjugated rat anti-mouse CD 68 (clone FA-11; Serotec). After two washes with hypertonic PBS, the sections were rinsed with PBS and immediately mounted with DAPI containing VectaShield®, covered with coverslips, and kept at 4°C until confocal imaging was performed. Confocal imaging was performed using a Zeiss LSM 510 META microscope (Carl Zeiss AG, Oberkochen, Germany) in an inverted configuration. Pinhole settings were adjusted for equal “optical sections.” When antibodies with far-red emission were used, the far-red emission channel was “falsely” colored green to allow for overlay with rhodamine-tagged nanoparticles, and the original green emission channel was turned off. Data were captured and analyzed using Zeiss LSM 510 Meta and Image Browser software (Carl Zeiss AG). PhotoShop 7 (Adobe Systems Inc., San Jose, CA, USA) was used for postprocessing of images.
In Vivo MRI
Bare untargeted paramagnetic micelles served as a nonspecific control CA and were injected into apoE-KO mice (N = 4). To that end, an infusion line was placed in the tail vein of the animals to allow administration of the CA when the animals were in the MRI scanner. First, high-resolution T1-weighted images were acquired prior to CA administration (Fig. 2a). The administration of the CA was monitored by dynamic contrast-enhanced (DCE)-MRI of the abdominal aorta. At 45 and 90 min postinjection, high-resolution MR images were acquired with the same imaging parameters as before CA injection (Fig. 2b and c). At 24 h after the CA injection, the abdominal aorta region of the animals was scanned again (Fig. 2d). SE of 30–60% between the thickened wall and the surrounding muscle tissue was observed at the time points after the CA injection.
Mice that were injected with MSR-specific micelles showed much more pronounced enhancement of the vessel wall. Representative MR images of atherosclerotic plaques of apoE-KO mice before and 24 h after injection with Ab-micelles (N = 6) are depicted in Fig. 3a–f. A prominent SE of atherosclerotic regions in the aortic wall was observed. For improved visualization, the SE was color-coded (insets) according to the pseudo color code on the right.
Analysis of the data obtained from mice that were injected with the nonspecific CA did not show significant differences in the signal-to-noise ratio (SNR) of the vessel wall (Fig. 4a) or the CNR between the vessel wall and surrounding muscle tissue (Fig. 4b) at any of the time points studied. Mice that were injected with the MSR-targeted CAs (Ab-micelles) showed a significant enhancement (up to 200%) of the voxels located at the ROIs 24 h postinjection (Fig. 4a). The CNR of the atherosclerotic vessel wall increased from approximately –1 to approximately 3 at 24 h postinjection (Fig. 4b), while there was no significant contrast between muscle and vessel wall in regions of the vessel that did not contain plaque, as confirmed by histopathology (data not shown). The SNR in muscle showed no significant changes for both the nontargeted and targeted CAs when pre- and postcontrast images were compared (Fig. 4c).
Lastly, the CNR between the lumen and the vessel wall was determined (Fig. 4d). The nontargeted CA did not induce differences in the CNR between the lumen and the vessel wall, while upon injection of the targeted micelles, the CNR between the vessel wall and lumen doubled 24 h postinjection.
Optical Imaging and Histopathology of Excised Aortas
After the MRI experiments the animals were killed and the aortas were excised for optical imaging and histological analysis. Histological sections and MR images were registered as published previously (23). The aortas of mice injected with Ab-micelles(QD) were illuminated with UV light in order to identify regions with a high CA uptake. These regions could be identified by green fluorescence originating from the QDs (Fig. 5a). From these regions sections were taken and evaluated with histopathology (Fig. 5b). Fluorescence microscopy from embedded sections was not possible, since QDs are highly sensitive to oxidation. The embedding and fixation procedure solubilizes the micellar coating, which causes the QDs to come in contact with water and oxygen, resulting in their oxidation and the loss of their optical properties. Therefore, two mice were injected with Ab-micelles(Rh). In Fig. 5c–f a DAPI-colored plaque section (Fig. 5d) counterstained for macrophages (Fig. 5e) of such an animal is shown that reveals the association of the Ab-micelles(Rh) (Fig. 5c) with macrophages in the plaque.
Atherosclerotic plaques with a macrophage-rich phenotype are more likely to be unstable than plaques with a relatively low macrophage abundance and relatively high smooth muscle cell content (3). Novel vascular imaging methods to discriminate unstable from stable plaques would be very valuable and improve diagnosis. Over the past decade MRI has developed as a promising modality for atherosclerosis imaging because of its ability to depict soft tissue with high spatial resolution. Contrast can be generated on the basis of intrinsic differences in tissue T1, T2, and diffusion. Different plaque components have been identified using multicontrast MRI (9). Improved visualization of plaques has been realized by using nonspecific macromolecular CAs such as Gadofluorine M (24, 25), which accumulates in plaques due to its lipophilic properties. Furthermore, dextran-coated iron-oxide particles (USPIO) have shown their potential as a negative CA for the visualization of macrophage-rich plaques in both animal models (26) and humans (27). In contrast to the positive contrast generated with Gd-based agents, the accumulation of iron-oxide particles causes dark spots in T2- and T2*-weighted images, which is referred to as negative contrast. This may complicate the interpretation of the images, since one has to detect a decrease of signal in T2*-weighted images that usually already have a poor SNR (27).
For improved plaque visualization, the so-called “black-blood” spin-echo MRI sequences are used. These MRI sequences depict the arterial lumen as black regions on the image to facilitate discrimination of the surrounding vessel wall, muscle, or plaque from the vessel lumen (8). Since the iron-oxide nanoparticles cause darkening of the tissue, the identification of plaque containing such a CA is hampered (28). Recently, several techniques have been developed that allow for the generation of positive signal in the presence of iron-oxide nanoparticles (29). However, this type of macrophage imaging is based on passive targeting, which may limit its prognostic value.
In this study we aimed to improve the visualization of macrophage activity in plaques by actively targeting the CA with an antibody specific for MSR. The CA was based on paramagnetic and pegylated micelles, which contain a QD in their core to allow parallel fluorescence imaging. In contrast to iron-oxide-based CAs, the nanoparticles developed for this study contain Gd-DTPA-like amphiphiles, which cause positive contrast on T1-weighted images. This may enable improved plaque visualization as compared to the identification of hypointense regions caused by iron-oxide nanoparticles on T2*-weighted images.
The Ab-micelles actively targeted to the MSR caused a pronounced SE (locally up to 200%) of atherosclerotic plaques on T1-weighted images. As discussed above, this SE was much more pronounced when the CAs were conjugated to MSR-specific antibodies. The vessel wall of regions in the aorta that did not contain plaque showed no contrast enhancement. This pronounced SE facilitates comparison between pre- and postcontrast situations as compared to previous studies that employed actively targeted CAs for improved plaque detection with MRI (28, 30). The quantification of macrophages and the analysis of macrophages, matrix metalloproteinases, etc. is critical, especially in light of a recent report by Verhoeven et al. (31). However, that was beyond the scope of this study and is the subject of ongoing future work. Furthermore, since we did not observe vessels in the plaques, we cannot ascribe the nonspecific enhancement to plaques rich in neovessels. We therefore do not attribute the SE of the targeted micelles to neovascularization.
Although we were not able to fully exploit the unique properties of QDs (18–20) in this study, we are the first to demonstrate the feasibility of using QDs for target-specific plaque detection. For ex vivo histological preparations the QD-micelles used in this study were not suitable, since the fixation and embedding procedure solubilizes the micellar coating. Nevertheless, we have shown the value of this CA for identifying regions with high CA content by illuminating intact aortas with UV light. With regard to toxicity, the QDs used in this study have specific features that make them suitable for in vivo experiments. They consist of a CdSe core, but are covered with a ZnS shell to improve their stability and decrease toxicity.
The hydrophilic PEG coating of the CA has several benefits. First, it serves as a protective shield that reduces the interaction of the micelles with blood components, cells of the immune system, and the reticuloendothelial system (RES) (32). This results in an increased circulation time and bioavailability of the CA. Preliminary studies using the control micelles show the blood half-life in mice to be longer than 2 h. Second, due to their small size (approximately 15 nm), the micelles can easily penetrate into tissue with increased permeability, such as plaques (33). Most Gd-based nanoparticulate carriers have sizes that exceed 150 nm (21, 22, 30, 34, 35), and therefore mainly serve as CAs to be directed to vascular targets. The macrophages targeted in this study are an extravascular target, since they are present in the vessel wall. The CA, therefore, must first extravasate from the circulation into the plaque and then target the MSR. This extravasation process may be facilitated by the small size of the micelles and their flexible nature. Finally, the present micelle system has an advantage over other systems (e.g., Gadofluorine M (24, 25)) in that it is hydrophilic and not lipophilic, and thus may be expected to be less prone to nonspecifically accumulate in the plaque. In this study the nontargeted micelles indeed did generate little background signal as compared to the MSR-targeted micelles. All the above properties of this micellar CA create the opportunity to use it also to specifically visualize other molecular markers overexpressed in atherosclerotic plaques, e.g., phosphatidyl serine exposed on the surface of apoptotic cells. In addition to applying this CA for improved plaque imaging, it may also be employed to target other surface receptors on cells present in the extravascular space (e.g., HER2/NEU (36)) that are overexpressed by multiple types of cancer cells.
In conclusion, a novel CA based on hydrophilic and pegylated micelles for improved plaque detection and characterization by molecularly targeting macrophages was presented. The size, specificity, and bimodal character of the micellar CA allow MR and optical molecular imaging of extravascular targets and may be employed to detect other markers of atherosclerosis, as well as extravascular targets in other disorders.
This study was supported in part by grants from the NIH/NHLBI (R01 HL71021 and R01 HL78667; to Z.A.F.).