molecular mr imaging of collagen in mouse atherosclerosis...

FULL PAPER

DOI: 10.1002/ejic.201200010

Molecular MR Imaging of Collagen in Mouse Atherosclerosis by UsingParamagnetic CNA35 Micelles

Glenda S. van Bochove,[a] Honorius M. H. F. Sanders,[a,b] Mariska de Smet,[a]

Henk M. Keizer,[c] Willem J. M. Mulder,[d] Rob Krams,[e] Gustav J. Strijkers,[a] andKlaas Nicolay*[a]

Keywords: Imaging agents / Nanostructures / Collagen / Micelles / Gadolinium

Magnetic resonance imaging (MRI) is increasingly used inbiomedicine to visualize plaques in the walls of major arter-ies in relation to atherosclerosis, the prime cause of myocar-dial infarction and ischemic stroke. The present study aimsto explore the utility of contrast-enhanced MRI for improvingthe specificity of the MRI evaluation of atherosclerotic pla-ques with the use of a Gd-based paramagnetic contrast agentthat is targeted to collagen. Collagen is a major componentof the extracellular matrix and as such plays an importantrole in the stability of atherosclerotic plaques. Micelles weremade with lipid containing 45 mol-% Gd for MRI detectionand a low mol fraction of fluorescent lipid for fluorescencemicroscopic analysis. Collagen-targeted, functional micelleswere prepared by conjugation of the CNA35 protein, whilenonfunctional control micelles were conjugated with a mu-tated version of the protein. The micelles were characterized

IntroductionAtherosclerosis is a chronic inflammation that proceeds

in the intima of the walls of arterial blood vessels and leadsto the formation of so-called atherosclerotic plaques.[1] Rup-ture of advanced atherosclerotic plaques causes contact be-tween blood and thrombogenic material inside the plaque,which results in the formation of blood clots. These clotscan either directly obstruct the flow of blood or do so at adownstream location and thereby cause infarction of the

[a] Biomedical NMR, Department of Biomedical Engineering,Eindhoven University of Technology,P.O. Box 513, 5600 MB Eindhoven, The NetherlandsFax: +31-40-2432598E-mail: [email protected]

[b] Laboratory of Materials and Interface Chemistry and SoftMatter,CryoTEM Research Unit,Department of Chemical Engineering and Chemistry,Eindhoven University of Technology,Eindhoven, The Netherlands

[c] SyMO-Chem. BV,Eindhoven, The Netherlands

[d] Translational and Molecular Imaging Institute,Mount Sinai School of Medicine,New York, USA

[e] Department of Bioengineering, Imperial College London,London, United KingdomSupporting information for this article is available on theWWW under http://dx.doi.org/10.1002/ejic.201200010.

Eur. J. Inorg. Chem. 2012, 2115–2125 © 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim 2115

with respect to their magnetic, biochemical, and biophysicalproperties. Atherosclerotic plaques were induced in the rightcarotid artery of apo-E knock-out mice by surgical placementof a tapered polymeric cast. In vivo MRI was performed at6.3 Tesla before and up to 24 h after intravenous injection ofparamagnetic micelles (50 μmol Gdkg–1). MRI revealed thestrongest signal enhancements by CNA35 micelles. At earlytime points after injection of CNA35 micelles, contrast en-hancement was higher in the collagen-richer lesions com-pared to that in the collagen-poorer lesions. Confocal laserscanning microscopy confirmed co-localization of CNA35 mi-celles and collagen in the plaques. We have demonstratedmolecular MR imaging of collagen in experimental athero-sclerosis by using a CNA35-functionalized micellar contrastagent.

heart and ischemic stroke of the brain. Current diagnosticimaging tools, as applied to symptomatic patients, primarilyfocus on the extent of the narrowing of the blood vessellumen and on the visualization of the dimensions and loca-tions of atherosclerotic plaques.[2] However, it is well knownthat these features provide a poor prognosis for the futurerisk of adverse clinical events and therefore more refinedimaging biomarkers are sought to improve the clinical man-agement of patients with advanced atherosclerosis. Suchnovel imaging procedures could also provide more relevantmeasures of the efficacy of therapeutic interventions. It iswell-established that the susceptibility of atheroscleroticplaques to rupture is dependent on their composition, theirarchitecture, as well as on the activity of key components,such as macrophages and matrix metalloproteinases(MMPs).[3]

Collagen represents a family of mostly fibrous structuralproteins that are main constituents of connective tissue (fora review see ref.[4]). Collagen fibers are a major componentof the extracellular matrix that supports most tissues. Colla-gen plays an important role in the stability of atheroscler-otic plaques. Its levels are modulated by MMPs, which de-grade the extracellular matrix (ECM), and by pro-inflam-matory cytokines, which inhibit the synthesis of new colla-

K. Nicolay et al.FULL PAPERgen.[5] Reduced collagen levels can result in decreased thick-ness as well as increased fragility of the fibrous cap thatseparates arterial blood from thrombogenic material in theplaque. This situation enhances the risk of plaque ruptureand occurrence of acute clinical events, in particular incombination with a large lipid-rich necrotic core and thepresence of many inflammatory cells. On the other hand,plaques with small lipid core and thick fibrous cap withrelatively high collagen content are usually considered to besafe and stable. The noninvasive visualization of collagen inatherosclerotic plaques could therefore be of great value fordifferentiating vulnerable and stable atherosclerotic plaques.

The present study was aimed to assess the capabilities ofcontrast-enhanced magnetic resonance imaging (MRI) forimproving the specificity of the evaluation of atheroscleroticplaques by the use of collagen-targeted paramagnetic con-trast material. MRI is one of the most important diagnosticimaging techniques and it combines high spatial resolutionwith excellent soft tissue contrast. MRI has proven to be apowerful modality for in vivo imaging of the larger vesselwalls, both in human patients[2,6] and in animal models ofatherosclerosis.[7] Traditional multiparametric MRI proto-cols (using for example T1-, T2-, proton density-, and dif-fusion-weighting) allow the assessment of different plaquecharacteristics, such as size, lipid-rich necrotic core volume,and the presence of intraplaque hemorrhage and calcifi-cations. The characterization of these important manifesta-tions of plaque phenotype has been exploited for monitor-ing plaque progression with disease severity and as surro-gate endpoints of therapeutic interventions. However, themulticontrast MRI approach provides limited insight intoplaque vulnerability, and it therefore has a modest role inpredicting plaque rupture. Consequently, the field of ath-erosclerosis imaging is expanding towards molecular im-aging, which makes use of targeted contrast agents for thedetection of specific plaque-associated markers, such as ma-crophages,[8] thrombi,[9] cell-adhesion molecules,[10] and an-giogenesis.[11]

Collagen detection in plaques has not been addressed byMRI-based molecular imaging so far. Megens et al.[12] per-formed an ex vivo fluorescence microscopy study on thedistribution of collagen in atherosclerotic plaques in the ca-rotid bifurcation of apo-E knock-out mice, with use of fluo-rescent-labeled CNA35 protein, based on original develop-ments by Krahn and colleagues.[13] CNA35 is the collagen-binding fragment derived from the transmembrane CNAprotein that is found in the bacterium Staphylococcus aureusand is involved in wound infection by this pathogen. Thesoluble CNA35 domain consists of two subdomains,termed N1 and N2, which jointly wrap around the collagenmolecule to form a complex with dissociation constantsranging from 20 nm to 30 μm depending on the specific typeof collagen.[13,14] Several binding sites have been identifiedfor the association of CNA35 to both collagen types I andII. This is relevant to the use of CNA35 as a ligand forcollagen visualization, as the fibrillar collagen types thatdominate the tensile strength and the elastic behavior of thevessel wall, both normal and diseased, are type I and type

www.eurjic.org © 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim Eur. J. Inorg. Chem. 2012, 2115–21252116

III collagen.[4] Interestingly, in atherosclerotic lesions a shifttowards a higher proportion of type I collagen has beenobserved.[15] The binding mode of CNA35 suggests that theprotein might have a higher propensity for binding to indi-vidual triple helical fibers of collagen, which are hardlypresent in normal vessel walls. Degraded collagen andnewly formed collagen fibers, highly abundant in advancedstages of atherosclerosis, are associated with high pro-portions of isolated triple helices. Consequently, the use ofCNA35 as a collagen probe might be particularly suitedto assess conditions of pathological collagen turnover, evenwhen the total collagen content shows only minor changes.Along this line, it was recently shown that in vascular speci-mens derived from abdominal aorta aneurysm (AAA) tis-sue, distinct defects in the collagen microarchitecture ratherthan changes in content or subtype distribution per seunderlie the mechanical failure of the tissue.[16] AAAs arelocalized dilations of the aortic wall that share many patho-logical features with atherosclerosis.

In the present proof-of-concept study, paramagnetic flu-orescent micelles were conjugated with the CNA35 proteinas a tool for MR and fluorescence imaging of collagen inatherosclerotic lesions. Micelles were chosen as a nanopart-iculate contrast agent for two reasons: (a) they can be incor-porated with a relatively high payload of Gd chelate forincreased detection sensitivity;[17] (b) they have the ability topermeate early-stage atherosclerotic lesions needed to reachintraplaque collagen.[18] The micelles were carefully charac-terized and their utility as an in vivo collagen probe wasstudied in the apo-E knock-out mouse model of atheroscle-rosis. More specifically, we made use of the model devel-oped by Cheng et al.,[19] in which a tapered, shear-stressmodifying polymeric cast was placed around the right ca-rotid artery. This provides a suitable model for testing thepresent collagen-targeted imaging approach: the athero-sclerotic lesions that develop on either side of the castwithin the same vascular segment differ significantly in col-lagen content (15.3�1.0 % vs. 22.2�1.0 % of the lesionarea for the upstream and downstream plaques, respec-tively[19]), and the left carotid artery can be used as a non-lesioned control vessel. Functional CNA35-conjugated mi-celles were compared to nonfunctional micelles of identicaldesign, except that they were equipped with a mutant-CNA35 protein that has a strongly reduced collagen bind-ing affinity. Both types of micelles contained Gd-containinglipid for in vivo MRI detection and fluorescent lipid for exvivo fluorescence microscopy analysis of histological slices.The fluorescent lipid also afforded a convenient optical la-bel for in vitro collagen binding assays.

Results

Characterization of CNA35 Micelles

Figure 1A schematically shows the make-up and compo-sition of the micelle, which contains a paramagnetic lipid(circa 45 mol-%) for MRI detection and PEGylated lipidsto promote micelle formation and enable coupling of the

Molecular MR Imaging of Collagen in Mouse Atherosclerosis

proteinaceous ligand. For optical detection, a low mol frac-tion of a fluorescent lipid was included. Prior to in vivo usein the mouse model of atherosclerosis, we sought to care-fully characterize the paramagnetic nanoparticles in termsof their biophysical properties, collagen binding capacity,MR relaxivity, and in vivo clearance kinetics. Successfulconjugation of CNA35 and mutant-CNA35 to preformedmicelles was confirmed by sodium dodecyl sulfate - poly-acrylamide gel electrophoresis (SDS-PAGE) analysis.[20]

Quantitative measurements of protein, gadolinium, andphosphate contents led to estimated Gd-DTPA-BSA/PEG2000-DSPE/targeting ligand ratios of circa 10:10:1(BSA: bis(stearylamide); DSPE: 1,2-distearoyl-sn-glycero-3-phosphoethanolamine). This estimation implies that onaverage between two and five protein ligands were conju-gated per micelle, when assuming that each micelle con-tained between 50 and 100 lipids.[21]

Figure 1. (A) Cartoon of CNA35 micelle. (B) Cryo-TEM image ofCNA35-conjugated micelles.

Cryogenic transmission electron microscopy (cryo-TEM)images (Figure 1B) of the micelles showed small black dots,which confirmed uniform micelle size and absence of aggre-gation of the particles. Dynamic light scattering measure-ments revealed a hydrodynamic diameter of 24.0� 3.0 nmfor both CNA35 micelles and mutant-CNA35 micelles (i.e.,control micelles). These data correspond to an increase indiameter of circa 6 nm by protein conjugation, as nonfunc-tionalized micelles were found to have hydrodynamic dia-meters of 18.0�1.0 nm. Our findings compare well to thoseof Reulen et al.,[22] who reported comparable light scat-tering data for CNA35-conjugated PEG-DSPE-based mi-celles.

As one of the measures of micelle stability, the criticalmicelle concentration (CMC), was determined with two dif-ferent approaches (Figure 2): the Wilhelmy plate techniqueand proton relaxivity measurements. The Wilhelmy platemethod measures the surface tension as a function of totallipid concentration. Above the CMC, the concentration ofisolated lipids is in essence constant and thus no changein surface tension occurs with varying lipid concentration.Below the CMC, the free lipid concentration equals the to-tal lipid concentration, and the surface tension will linearlydecrease with increasing lipid concentration. Figure 2Ashows Wilhelmy plate data for CNA35 micelles leading toan estimated CMC of circa 80� 15 μm (n = 5). MR relaxi-vity measurements can also be used to obtain estimates of

Eur. J. Inorg. Chem. 2012, 2115–2125 © 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim www.eurjic.org 2117

the CMC, as micelle formation leads to alterations in themotional properties of the Gd-containing lipid. Figure 2Bdepicts longitudinal relaxation rates R1 measured at60 MHz over a wide lipid concentration range. R1 data atlow concentrations, below the CMC, yielded a molar re-laxivity of 8 mm–1 s–1, while values of 12 mm–1 s–1 were esti-mated in the high-concentration range. The intersection oflinear fits of the low- and high-concentration range yieldedan estimated CMC of 70� 20 μm (n = 5). CNA35 and mu-tant-CNA35 micelles exhibited similar CMC values, byboth the Wilhelmy plate and the MR relaxivity approaches.

Figure 2. (A) Surface tension data collected with the Wilhelmyplate technique and plotted as a function of the lipid concentration.The lines are to guide the eye. (B) Longitudinal relaxation rate R1

measurements for different micelle concentrations. Linear fits weremade of data points below and above the apparent CMC and plot-ted as straight lines. The arrows in parts (A) and (B) indicate theapparent CMCs, as deduced from the two approaches.

NMR relaxation dispersion (NMRD) profiling, in whichrelaxometric properties are quantified as a function of mag-netic field strength, was used to confirm the macromolecu-lar behavior of the relaxivity of the micelle-embedded Gdchelate used here.[23] Figure 3 shows that the molar relaxiv-ity r1 of the Gd lipid indeed displays the typical maximumaround 20 MHz. CNA35 and mutant-CNA35 micelles be-haved similarly. NMRD profiles of the micelles in the pres-ence of blood serum were essentially identical to those ofmicelles collected in regular HEPES-buffered saline (HBS)buffer. This finding strongly suggests that micellar integritywas maintained in the presence of blood plasma proteins.At 37 °C and 60 MHz, the longitudinal relaxivity r1 was12 mm–1 s–1 for both control and targeted micelles. At 6.3 T,the field strength used for the in vivo MRI measurements,the relaxivity amounted to 3.5 mm–1 s–1. The lowered r1 isdue to the increased mismatch between the frequency pro-

Figure 3. NMRD profiles of CNA35 micelles (50 mM lipid) inHBS (black squares) and human blood serum (gray triangles) at37 °C.

K. Nicolay et al.FULL PAPERfile of the rotational and translational mobility of the mi-celle-embedded Gd chelate and the Larmor frequency.[23]

The above NMRD profiles were measured above the CMC.Attempts to obtain NMRD profiles below the CMC failedbecause of the limited sensitivity of the technique.

A fluorescent binding assay was used to verify the abilityof the micelles to bind to a collagen-coated surface (Fig-ure 4). A strong signal from micelle-associated fluorescentlipid was measured upon incubation of CNA35 micelleswith collagen-coated well plates. No significant nonspecificbinding of CNA35 micelles occurred when the plates hadnot been coated with collagen. Nonfunctionalized micellesshowed negligible collagen binding, while incubation withmutant-CNA35 micelles led to a slightly higher fluores-cence signal indicative of some background binding. Sim-ilar observations were previously reported by Reulen et al.,when using non-paramagnetic micelles equipped with mu-tant-CNA35 protein.[22] The shelf life of CNA35 micelleswas shown to be at least one month on the basis of thefluorescent binding assay.

Figure 4. Fluorescence assay of micelle binding to collagen:CNA35 micelles (CNA35), mutant-CNA35 micelles (Mutant) andnonconjugated micelles (Bare) were incubated on either a collagen-coated surface or a collagen-free surface. The level of backgroundsignal was measured in wells that were not incubated with micelles(bars labeled “No”).

The collagen binding properties of the CNA35 and mu-tant-CNA35 micelles were also assessed with the high-reso-lution one-dimensional MRI depth profiling technique,which we recently employed for studying the relaxivityproperties of collagen-associated paramagnetic CNA35liposomes.[20,24] Like the above fluorescence binding assays,these measurements make use of two-dimensional collagen-coated surfaces that are incubated with different paramag-netic micelle preparations, followed by the acquisition ofone-dimensional T1 profiles in the direction perpendicularto the collagen surface. Figure 5 shows typical examples ofsuch measurements for CNA35 and mutant-CNA35 mi-celles as well as nonfunctionalized control micelles. Only inthe case of functional CNA35 micelles did T1 shorteningoccur at the collagen surface, and this effect extended overcirca 100 μm into the bulk aqueous phase. The surface re-laxation rate (ρ),[24] which is a measure of the efficacy of T1

shortening by the collagen-associated paramagnetic mi-celles, amounted to 1.3�0.1 ms–1. This value of ρ is ap-proximately threefold lower than that of CNA35 liposomes,which we reported recently.[20,24] It should be noted thatthe latter paramagnetic liposomes, on a per particle basis,contain circa 1,000-fold more Gd lipid than the micelles


used in the present study. These observations suggest thatmicelles have a higher collagen association density thanliposomes. The proposed relatively low liposomal bindingdensity is in line with recent cryo-TEM findings thatCNA35-functionalized liposomes primarily associate topoorly assembled collagen fibrils.[25] Consequently, the ex-tent of liposome binding will strongly depend on collagenultrastructure. This likely is less of an issue in case ofmicelles that are five- to tenfold smaller than liposomes.

Figure 5. One-dimensional T1 profiles of CNA35 micelles (lightgray line), mutant-CNA35 micelles (gray line) and nonconjugatedmicelles (black line), following incubation with a collagen-coatedsurface. The x axis represents the distance from the surface, whichis located at position 0 μm.

The time dependence of the T1 shortening of blood, ascaused by the paramagnetic effect of the micelles, was usedto measure their blood circulation times following intrave-nous injection in mice (data not shown). Nonconjugatedmicelles had a blood circulation half-life of 22.5 �2.8 h.Conjugation of mutant-CNA35 protein had no effect onblood clearance kinetics; in this case, the circulation half-life amounted to 22.6 �2.5 h. CNA35 coupling lowered theblood circulation half-life to 17.1�4.3 h.

In Vivo MRI

T1-weighted MRI was used to measure changes in MRIcontrast upon injection of paramagnetic micelles in the castmodel of atherosclerosis in apo-E knock-out mice. Typicalexamples of MR images of the carotid arteries of mice in-jected with CNA35 micelles and mutant-CNA35 micellesare shown in Figure 6. Mice injected with CNA35 micelles(Figure 6A–H) showed vessel wall enhancement of the rightcarotid artery downstream from the cast as early as 15 minafter injection (Figure 6F and G). A heterogeneous en-hancement of the vessel wall was observed. After 24 h amore pronounced signal increase was noted, while theheterogeneity of the enhancement was preserved (Fig-ure 6H). In addition to the downstream segment, the vesselwall at the upstream side was also clearly enhanced 24 hafter injection with CNA35 micelles (Figure 6D). Minorsignal increases were observed in animals injected with mu-tant-CNA35 micelles (Figure 6J–L and N–P). The left ca-rotid artery, which served as a nonlesioned control vessel,showed no visible wall enhancement after both CNA35 mi-celle and mutant-CNA35 micelle injection.


Figure 6. T1-weighted MRI of a mouse injected with CNA35 micelles (A–H) and a mouse injected with mutant-CNA35 micelles (I–P),showing a slice upstream (first and third row) and downstream (second and fourth row) from the cast. Original images were cropped toa field-of-view of 0.75�0.75 cm2. The lesioned right carotid artery is shown on the left. Arrowheads indicate the right carotid artery (I),trachea (II), left carotid artery (III), spinal cord (IV), external jugular vein (V), and internal jugular vein (VI).

Figure 7 depicts the mean normalized signal enhance-ment (%NSE) of the different vascular segments over time.At all time points for both downstream and upstream pla-ques, CNA35 micelles gave a higher signal enhancementthan mutant-CNA35 micelles, in agreement with the abovevisual inspection. This difference reached significance at24 h for the downstream lesion that has higher collagencontent. For the left carotid artery, in which no plaqueswere present, the reverse trend was observed, in that themutant-CNA35 micelles led to a slightly larger average sig-nal enhancement. This may have been caused by the longercirculation half-life of the mutant-CNA35 micelles as com-pared to the CNA35-conjugated material.

In the first hour after micelle administration, a more pro-nounced difference between down- and upstream plaqueswas seen. This was due to the stronger accumulation ofCNA35 micelles in the downstream, collagen-richer plaquesthan in the upstream, collagen-poorer lesions. To a lesserextent, this effect was also noted for mutant-CNA35 mi-celles, which caused a lower signal enhancement that be-came almost equal for both plaque types after 1 h.

In all studied carotid wall segments, a signal enhance-ment indicative of micelle deposition over time was ob-


Figure 7. Percentage normalized signal enhancement(mean�SEM) of the different vascular segments versus time forCNA35 micelles (n = 7) and for mutant-CNA35 micelles (n = 8).The collagen-poorer lesion is located upstream from the cast (A),the collagen-richer lesion is located downstream from the cast (B),and the left carotid artery (C) had no plaque. *: significantly en-hanced, Bonferroni, p� 0.05; #: CNA35 micelles significantly dif-ferent from mutant-CNA35 micelles, Bonferroni, p� 0.05.

served. In case of mutant-CNA35 micelles, an equal ac-cumulation in all vessel wall segments was observed 24 hafter micelle injection. However, the average signal increaseinduced by CNA35 micelles was higher for both lesiontypes compared to that in the left carotid artery.

K. Nicolay et al.FULL PAPERMicroscopy

Various histological stainings of vascular segments are pre-sented in Figure 8. These were performed to confirm pre-

Figure 8. Histology of the upstream (A–D) and downstream (E–H) atherosclerotic plaques. Stainings from left to right: hematoxylin andeosin; picrosirius red; Masson trichrome; and anti-α-smooth muscle actin–FITC. The dashed lines in the picrosirius red staining indicatethe plaque borders. The downstream lesion (F) shows a higher collagen content as deduced both from the picrosirius red staining (F) aswell as the blue staining in the Masson trichrome images (G), when compared to the upstream lesion (B and C, respectively).

Figure 9. Confocal laser scanning microscopy of upstream and downstream lesions from sections obtained 24 h after injection of (A)CNA35 micelles, or (B) mutant-CNA35 micelles. For the images in the right column, the laser intensity was threefold higher than thatused for the images in the middle column. T1-weighed MR images of the corresponding lesions as measured before and 24 h after micelleinjection are shown in the left column. Red: NIR664-DSPE of the micelles; green: elastin autofluorescence.


vious reports on the nature of the vascular lesions that de-velop in the carotid artery cast model.[19] The picrosiriusred (Figure 8B and F) and the blue staining in the Massontrichrome images (Figure 8C and G) confirmed the higher


collagen content of the downstream lesion. The latterplaque also appeared to have a little higher smooth musclecell content (Figure 8D and H). Microscopy of left carotidartery segments that corresponded to the area used forMRI analyses revealed no plaque (data not shown). Theseobservations are in line with previous work by Cheng etal.[19]

Confocal laser scanning microscopy of sections that werecollected 24 h after micelle injection confirmed a clear ac-cumulation of CNA35 micelles in both the upstream andthe downstream plaque as exemplified in Figure 9. For theupstream lesion, intense CNA35 micelle fluorescence wasfound near the vessel lumen (Figure 9A). For the down-stream lesion, the regions positive for CNA35 micellesshowed more intense fluorescence and were also larger andextended deeper into the atherosclerotic plaques. Upon ap-plication of a threefold higher laser intensity (indicated by3 � in Figure 9), weak CNA35 micelle fluorescence was ob-served in the rest of the plaque in both cases. For mutant-CNA35 micelles, the fluorescence was significantly lessbright with highest intensity near the vessel lumen (Fig-ure 9B).

Discussion

This study was aimed to develop collagen-targeted para-magnetic micelles and to apply these for contrast-enhancedMRI of early stage atherosclerotic lesions in carotid arteriesof apo-E knock-out mice. Gd-containing lipids were incor-porated in PEGylated lipid-based micelles and conjugatedwith the CNA35 or mutant-CNA35 protein, which led tonanoparticles with similar hydrodynamic diameters of circa24 nm. The above micelles displayed similar in vitro bio-physical characteristics, while differing strongly in terms oftheir collagen-binding ability: CNA35 micelles showedhigh-affinity binding to collagen-coated surfaces, while mu-tant-CNA35 micelles experienced negligible binding undersimilar circumstances. The aggregation behavior and relaxi-vity properties of both functional and nonfunctional nano-particles were similar, and therefore the prepared materialswere examined for their ability to detect collagen in vivo insubtle vascular lesions. The functional micelles had some-what shorter blood circulation half-lives, and this needs tobe considered when interpreting the in vivo MRI data.

Collagen-targeted MRI with CNA35 micelles was shownto be capable of discriminating between vascular lesions onthe basis of moderate differences in collagen content. A sig-nificantly higher signal enhancement in the collagen-richerplaque was observed for collagen-specific micelles as com-pared to that in nonfunctional, control micelles. Ex vivofluorescence microscopy lent support to the interpretationof the in vivo findings and showed pronounced accumu-lation of CNA35 micelles in the downstream plaques withthe higher collagen content, whereas mutant-CNA35 mi-celles showed much less accumulation.

Previously, Cheng et al.[19] had demonstrated in the pres-ent mouse model that collagen in the downstream plaque


occupies 22% of the surface area, as compared to 15 % inthe upstream plaque. These histological observations werequalitatively confirmed in our study (Figure 7). In agree-ment with these patterns, the downstream, collagen-richerplaque showed the highest MRI signal enhancement shortlyafter CNA35 micelle administration. The signal enhance-ment increased to 33 % after 24 h, whereas the signal en-hancement in animals injected with mutant-CNA35 par-ticles reached a significantly lower value of 17%. The up-stream plaque with the lower collagen content showed sig-nificantly less signal enhancement shortly after injection.However, signal enhancement in the upstream plaque hadincreased to 28 % 24 h after the injection of CNA35 micelles(Figure 7A), reaching similar enhancement values as thosein the collagen-richer downstream plaque (Figure 7B).

Both the downstream and upstream signal enhancementswere higher after CNA35 micelle administration comparedto the enhancement value of 13 % in the left carotid artery.We attribute this difference to specific accumulation of thefunctional micelles. At 24 h, the mutant-CNA35 particlescaused a mean enhancement of 17% in both types of pla-ques, similar to that in the control vessel without plaque.This leads us to conclude that in all examined vessel wallsegments, both control and atherosclerotic, similar nonspe-cific micelle accumulation had occurred 24 h after injection.This nonspecific accumulation might complicate the detec-tion of specific collagen binding and mask potential differ-ences in contrast enhancement of the two plaque pheno-types 24 h after injection.

The percentage MRI signal enhancements were calcu-lated by normalizing the measured signal intensity increasesto muscle signal intensity so as to enable a more straightfor-ward comparison between the acute and 24 h time pointmeasurements. However, the signal intensity in the musclewas also affected by the injected micelles and was found tobe 6–10% higher for all post-injection time points. This ef-fect was identical for the early and 24 h measurements forboth CNA35 and mutant-CNA35 micelles (not shown).Normalization to muscle signal intensity therefore resultedin a lower fractional signal enhancement. Nevertheless, nor-malization to tissue (and nearby muscle tissue is an obviouschoice) is needed to compensate for overall signal intensitydifferences between scans.

The blood circulation half-life was approximately 17 hfor CNA35 micelles. Mutant-CNA35 micelles and noncon-jugated micelles had blood circulation half-lives of circa23 h. Ideally, the blood half-life of mutant-CNA35 micellesshould have been identical to that of CNA35 micelles. Thelonger circulation time of the control micelle preparationcan be expected to lead to more nonspecific accumulation.The fact that the mutant-CNA35 particles resulted in lesscontrast enhancement is therefore additional proof thatCNA35 micelles were at least in part specifically associatedto material inside the plaque. Another important feature ofthe control particles is their hydrodynamic diameter, whichstrongly affects blood extravasation and tissue penetrationkinetics. Mutant-CNA35 micelles (their diameter was24 nm, the same as that for CNA35 micelles) are therefore a

K. Nicolay et al.FULL PAPERbetter control than nonfunctionalized micelles (with 18 nmhydrodynamic diameter).

Differences in plaque permeability between downstreamand upstream lesions might play a considerable role as canbe inferred from the initial contrast enhancement with bothtypes of micelles. At 15 min the levels of contrast enhance-ment brought about by mutant-CNA35 particles differedbetween upstream and downstream plaques. However, thelatter micelles also accumulated in the upstream plaque at60 min, and all vessel wall segments exhibited largely iden-tical signal changes after 24 h. This suggests that long mi-celle circulation times eventually lead to equal nonspecificvessel wall accumulation. The average signal enhancementby CNA35 micelles was higher than that by mutant-CNA35micelles, which strongly supports the interpretation that weare dealing with specific accumulation in the former case.

Collagen is known to be one of the most important fac-tors in determining plaque stability,[1,5] and the noninvasivemonitoring of collagen levels might be a tool to identifyplaques at risk. One of the ultimate goals of molecular im-aging of atherosclerosis is to determine plaque vulnerability.Several markers, including macrophages, matrix metallop-roteinases, and products of reactive oxygen species havebeen proposed for this purpose. Our study took a differentapproach. By targeted imaging of collagen, low-risk stablelesions could possibly be distinguished from plaques at risk.More importantly, however, the present tool might besuited to assess the effects of plaque-stabilizing therapiesthrough the assessment of collagen levels. Combining dif-ferent targeted contrast agents might enable monitoring ofchanges in plaque composition as a consequence of thera-peutic interventions. Indeed, future molecular imaging ap-proaches likely will increasingly aim for multiple markers toimprove the specificity of the read-out.

Klink et al.[26] recently used a very similar approach withCNA35 micelles for the detection of collagen in relation tothe formation of AAAs in a mouse model. Very advancedvascular lesions that are much larger than the early-stagelesions examined in our study characterize this model. Inline with our findings, Klink et al.[26] reported MRI signalchanges that agreed well with collagen content and spatialdistribution.

Previously, low-molecular-weight, nontargeted paramag-netic contrast agents have been used to identify fibrous tis-sue in atherosclerotic plaques in patients.[27] These studiesappear to be successful; however, the relation between con-trast agent accumulation, collagen content, and plaque vul-nerability remains to be established. Specific targeting ofcollagen in atherosclerotic plaques could therefore lead tomore accurate estimation of collagen content in atheroscler-otic plaques and better assessment of treatment response.

MRI has previously been used to image collagen in myo-cardial infarction by using a collagen-specific peptideequipped with three Gd-DTPA moieties.[28] These peptideprobes, which are small and cleared via the kidneys, wereshown to highlight collagen that is formed after myocardialinfarction. Using a similar contrast agent concept, Botnarand co-workers have shown that elastin-targeted peptide


MRI contrast agents provide powerful tools for assessingcritical aspects of experimental atherosclerosis, such as ves-sel wall remodeling and plaque burden.[29]

In the current work, a larger probe with a higher payloadof gadolinium and thus a higher relaxivity per particle wasused. The particle exhibited prolonged blood circulation,which allowed ample time for diffusion into the atheroscler-otic plaques to reach collagen to overcome the disadvantageof reduced tissue penetration. In future research, attentionwill be paid to alternative targeting ligands. Collagen-spe-cific peptide constructs that have been identified by phage-display techniques might be interesting candidates for li-gands.[28a,30] In addition, we will seek to tune the circulationhalf-lives of the nanoparticulate probes to reduce the levelsof non-target-associated contrast material in the plaque. Avirtue of the nanoparticle approach is that these can readilybe loaded with plaque-stabilizing drugs and thus allow foreffective combinations of diagnostics and therapy in pre-clinical studies of atherosclerosis.

A limitation of our study is that the Gd-DTPA-BSA li-pid, which was used as MRI label, has a less favorablethermodynamic and kinetic stability towards transmetalla-tion, in which Gd ions are replaced by other di- or trivalentcations that are encountered in vivo. When using the abovelipid, we have never observed any signs of this process. VanTilborg et al.[31] showed that Gd-DTPA-BSA-loaded lipo-somes fully preserved their Gd label while circulating up to48 h in mouse blood in vivo. Nevertheless, the use of a morestable Gd chelate, such as the Gd-DOTA-based paramag-netic lipid that we recently prepared,[32] is preferred to re-duce the possibility of transmetallation. This lowers the riskof adverse effects of Gd ions that are set free from the che-late and also will reduce the probability of an erroneousMRI read-out in applications such as those explored here.

Conclusions

The data presented in this study demonstrate that colla-gen-targeted MRI with paramagnetic CNA35 micelles in anexperimental model of atherosclerosis is capable of discrimi-nating two different plaques on the basis of small differ-ences in collagen content. MRI signal enhancements weremodest, which is partly due to the small size of the vascularlesions and their early stage of development. The technol-ogy presented may provide useful tools for assessing plaqueprogression and plaque phenotyping in animal models andfor the evaluation of the plaque-stabilizing effects of thera-peutic interventions.

Experimental SectionCNA35 Expression: Vector pQE30CNA35, coding for the collagenbinding part of the A-domain of Staphylococcus aureus collagenadhesin (amino acids 30–344), fused to an N-terminal His-tag wasa kind gift from Dr. Magnus Höök (Texas A&M University, USA).The plasmid was transformed into E. coli BL21(DE3) (Novagen,Nottingham, UK). The recombinant expression and purification of


CNA35 were carried out as described before.[13] For the generationof mutant-CNA35, which was used as a control, a Y175 K mu-tation was introduced into the 6His-CNA35 gene from pQE30-CNA35 by using the QuickChange Site-Directed Mutagenesis Kit(Stratagene) having the primers 5�-CGGGAAC AAG-TAGTGTTTT, CTATAAAAAAACGGGAGATATGCTACC-3�,and 5�-GGTAGCCAT ATCTCCCGTTTTTTTATAGAAAA-CACTACTTGTTCCCG-3� to yield pQE30-CNA35(Y175 K).

Synthesis of NIR664-DSPE: Commercially available N-hydroxysuc-cinimide (NHS) activated ester NIR664 (Sigma Aldrich) was conju-gated to the primary amine of DSPE (Alexis Biochemicals) withdiisopropylethylamine (DiPEA) as a base. First, NIR664 and Di-

PEA were dissolved in anhydrous DMF, followed by the additionof the DSPE dissolved in chloroform. The reaction mixture wasstirred overnight under an inert atmosphere. After purification,NIR664-DSPE was obtained with a yield of 90% (for details seeSupporting Information).

Preparation of Paramagnetic Micelles: Paramagnetic micelles wereprepared by the lipid film hydration method.[33] The lipid film wasformed by concentration of a chloroform/methanol (4:1; v/v) solu-tion of Gd-DTPA-BSA, DSPE-PEG2000, DSPE-PEG2000-male-imide, and NIR664-DSPE (or occasionally rhodamine-PE) by ro-tary evaporation in a molar ratio of 4:5:1:0.02. After concentrationto dryness by rotary evaporation, the lipid film was dried under N2

for 30 min. Next, the lipid film was hydrated with HEPES-bufferedsaline (HBS) at pH 6.7, containing 20 mm HEPES (Sigma) and135 mm NaCl (Sigma). The solution was slowly rotated at 65 °Cfor 1 h, which resulted in a clear micelle solution. CNA35 (molarratio CNA35/lipid was 1:50) was coupled to the micelles by asulfhydryl-maleimide coupling method.[33] CNA35 was treated withsuccinimidyl S-acetylthioacetate (SATA) in a molar ratio of 1:8 for1 h in a NaHCO3 buffer at pH 8.0. Uncoupled SATA was removedby using a centrifuge concentrator with a molecular weight cut-off(MWCO) of 10 kDa (Sartorius). Subsequently, deacetylation wasconducted with a hydroxylamine solution (0.5 m hydroxylamine,1 m HEPES, 32 mm EDTA, pH 7.0) for 1 h. Modified CNA35 wasthen allowed to react with preformed micelles at 4 °C overnight inHBS (pH 6.7). Uncoupled CNA35 was separated from CNA35-functionalized micelles by using centrifuge concentrators with aMWCO of 100 kDa (Sartorius). The same procedure was used formutant-CNA35 micelles.

Lipid Determination: To determine the lipid composition of the fi-nal micelle solution, both phosphate and gadolinium content weredetermined. Gadolinium content was determined with an induc-tively coupled plasma atomic emission spectroscope (ICP-AES)from Philips Research (Eindhoven, The Netherlands). The amountof phosphate was determined by the Rouser method.[34]

Protein Determination: The total amount of CNA35 coupled to themicelles was determined by using a modified Bradford method.[35]

The modification comprised equal lipid concentrations in both thecalibration line and the actual samples, ensuring the same back-ground signal for lipids. This modification allowed an accuratemeasurement of added protein.1H-MR Relaxivity Measurements: The longitudinal relaxation time(T1) of a total lipid concentration series (0.4–2 mm) was determinedat 60 MHz and 37 °C with a Bruker Minispec MQ60 (Bruker,Ettlingen, Germany). An inversion recovery sequence (4 averages)was used with 10 different inversion times, ranging from 10 to8000 ms in an exponential fashion, and with a relaxation delay of8 s. NMRD profiles were obtained from T1 measurements at dif-ferent magnetic field strengths, by using a continuum of magneticfield strengths from 0.00024 to 0.47 T (corresponding to a proton


Larmor frequency between 0.01 and 20 MHz) with a Stelar field-cycling relaxometer (Mede, Pv, Italy). Data points from 0.47 T(20 MHz) to 1.7 T (70 MHz) were collected with a Stelar Spinmas-ter spectrometer (Mede, Pv, Italy) operating at variable fields. Mea-surements were done in HBS.

Surface Tension Measurements: The surface tension of 10 mL sam-ples of a lipid concentration series (between 0.1 μm and 0.5 mm)was determined at 37 °C by the Wilhelmy Plate method.[36] A verti-cal platinum plate was attached to a balance, and the force due towetting was measured with a Krüss digital tensiometer k10T. Theplate was moved towards the surface until the meniscus connectedwith it. The surface tension was calculated from the resulting force,by using F = pγcosθ – ρgV, in which F is the force on the platemeasured by the balance, p is the perimeter of the Wilhelmy plateat the liquid/air interface, θ is the contact angle of the liquid onthe Wilhelmy plate, ρ is the liquid density, g is the gravitationalacceleration, and V is the volume of the part of the plate immersedin the liquid. The contact angle θ of the water or micelle solutionwith the platinum plate was very small and was assumed to be zero.Perimeter (p) and volume (V) are known properties of the specificWilhelmy plate that was used. Measurements were done in HBS.

Dynamic Light Scattering: The average hydrodynamic radius of themicelles was determined by using a Malvern ZetaSizer Nano S(Malvern, UK) at a temperature of 25 °C. The lipid concentrationof the solution was 50 μm in a HBS buffer at pH 7.4.

Fluorescence-Based Collagen Binding Assay: Wells of an 8-well stripplate (Corning, Schiphol-Rijk, the Netherlands) were incubatedovernight at 4 °C with rat-tail collagen type I (50 μL , 55 μgmL–1

, C7661, Sigma–Aldrich) in HBS. Next, nonbound proteins wereaspirated from the wells by using an automated Wellwash AC platewasher (Thermo, Breda, the Netherlands), and the wells wererinsed three times with HBS (300 μL) at 20 °C. The wells wereblocked with 5% (w/v) milk powder (250 μL) in HBS for 3 h at20 °C, aspirated, and again rinsed three times with HBS (300 μL).A solution (concentration: 1 mm lipid) of micelles in HBS (50 μL)was added to each well and incubated for 3 h at 20 °C. The wellswere aspirated and washed 10 times with HBS (300 μL). The fluo-rescence was subsequently measured by using a Fluoroskan AscentFL plate reader (excitation: 578 nm; emission: 620 nm).[13,20]

High-Resolution MR Depth Profiling: One-dimensional NMRdepth profiling was performed with a microimaging setup based onthe GARField method, as reported before.[24] Briefly, a saturationrecovery sequence was used to measure the longitudinal T1 relax-ation time as a function of distance perpendicular to the surface ofthe collagen layer, prepared as before.[24] All measurements wereperformed at room temperature. All parameters were the same ex-cept for the nanoparticle preparation used, that is, a micelle suspen-sion with a concentration of 20 mm total lipid was used for allincubations. The one-dimensional T1 profiles were fitted to esti-mate the surface relaxation rates of the collagen layers.[24]

Blood Circulation Half-Lives: For the determination of the circu-lation half-lives of CNA35 micelles and mutant-CNA35 micelles,apo-E knock-out mice (age 12 weeks; n = 3 each) were put on aWestern-type diet (0.21% cholesterol) for a period of 12 weeks. Themice were anesthetized with 3% isoflurane and maintained with 1–2% isoflurane, in medical air. Before injection of the contrast agent,blood (20 μL) was withdrawn from the saphenous vein. The micellesolutions (50 μmol Gd per kg body weight) were injected via thetail vein, and blood samples (20 μL) were collected from the sa-phenous vein at 2, 15, 30, 45, and 60 min. Next, the mice wereallowed to recover from anesthesia and additional blood sampleswere taken at 4, 8, 24, and 48 h after micelle injection. Mixing the

K. Nicolay et al.FULL PAPERblood with heparinized physiological salt solution (20 μL, 500 I.U.heparin per mL) prevented blood coagulation. The longitudinal re-laxation rate of the samples was measured at room temperature byusing a 6.3 T horizontal-bore scanner (Bruker Biospin, Ettlingen,Germany), equipped with a birdcage RF coil of 3 cm diameter witha fast inversion recovery segmented FLASH sequence (TE 1.5 ms;TR 3 ms; flip angle 15°; inversion time 67 ms to 4800 ms in 80steps; overall repetition time 20 s; FOV 3� 2.81 cm2; matrix128�128; slice thickness 1 mm; NEX, 2). ΔR1 values of the bloodmixtures over time were fitted by using Origin (OriginLab Corpora-tion, Northampton, USA) with a monoexponential decay function(y = Ae–t/τ) to determine the circulation half-lives (t1/2 = τln2).

Mouse Model of Atherosclerosis: For this study, 16 apo-E knock-out mice (age 12 weeks) were put on a Western-type diet (0.21%cholesterol) for a period of 3 weeks. A tapered cast was placedaround the right carotid artery to induce plaque formation down-stream and upstream from the cast.[19] The mice were initially anes-thetized with 3% isoflurane in medical air, and maintained at ap-proximately 2% isoflurane during surgery. The right carotid arterywas prepared free from the surrounding connective tissue, and astiff tapered cast (Promolding BV, The Hague, The Netherlands),with an inner diameter ranging from 500 μm upstream to 250 μmdownstream, was placed around the right carotid artery proximalto and well before the bifurcation. Post-operative analgesic medi-cation consisted of subcutaneous injection of buprenorfine (Temg-esic, 0.1 mg per kg body weight). Plaque development prior toMRI scanning was allowed for 9 weeks with continuing diet. Fol-lowing the final MRI measurement, mice were sacrificed by exsan-guination followed by arterial phosphate-buffered saline (PBS) per-fusion. The carotid arteries, liver, lung, spleen, and kidneys weredissected, embedded in 10% gelatin, snap-frozen and stored at–80 °C until further processing for histological evaluation.

In Vivo MRI: MRI was performed with a 6.3 T horizontal-boreanimal scanner (Bruker BioSpin, Ettlingen, Germany) and a quad-rature birdcage RF coil (Rapid Biomedical, Rimpar, Germany) of3 cm diameter. The mice were initially anesthetized with 3% isoflu-rane in medical air, and maintained with 1–2% isoflurane duringscanning. An infusion line was filled with contrast agent and fixedin the tail vein. The mice were placed in a custom-made cradle,which was kept at a constant temperature of approx. 37 °C to pre-vent hypothermia. Respiration was monitored with a balloon sen-sor connected to an ECG/respiratory unit (Rapid Biomedical, Rim-par, Germany). For contrast-enhanced MRI, the mice were injectedwith either CNA35 micelles or mutant-CNA35 micelles (50 μmolGd per kg body weight, n = 8 for each group). T1-weighted spin-echo MRI (TR 800 ms; TE 10.3 ms; FOV 2.56�2.56 cm2; matrix256�256; slice thickness 0.5 mm; NEX 8; scan time 27 min) wasperformed before and at 15 and 60 min as well as 24 h after micelleinjection. Slices were carefully placed perpendicular to the flow di-rection of the right carotid artery. Because of the curvature of thecarotid artery, two separate slice packages were placed upstreamand downstream from the cast.

Histology, Immunohistochemistry, and Fluorescence Microscopy:Both upstream and downstream plaques were cut into 80 μm thicksections perpendicular to the vessel direction. To determine the dif-ferences in collagen content, 8 μm thick sections were used for dif-ferent stainings at 80 μm intervals. The following stainings wereemployed: hematoxylin & eosin, picrosirius red (collagen), Mas-son’s trichrome (cytoplasm and muscle fibers in red, collagen inblue, and cell nuclei in purple), and Anti-α-Smooth Muscle Actinconjugated with fluorescein isothiocyanate (FITC) for smoothmuscle cells [dilution 1:250, clone 1A4 (Sigma–Aldrich, St. Louis,


USA)]. The nonfluorescent sections were analyzed with a ZeissAxio Observer Z1 microscope (Carl Zeiss, Inc.). Fluorescent sec-tions were analyzed with a Zeiss Axiovert 200M microscope (CarlZeiss, Inc.). Smooth muscle cells were visualized on separate sec-tions with a 455–495 nm excitation and a 500–550 nm emission fil-ter. With these filters, elastin autofluorescence was also detected.For evaluation of micelle fluorescence, confocal laser scanning mi-croscopy was performed with a Zeiss LSM 510 META system (CarlZeiss, Inc.). Elastin autofluorescence was excited with a 488 nmargon laser. Emission was filtered through a band-pass filter of500–550 nm. NIR664 was excited with a 633 nm HeNe laser, whileemission was filtered with a 650–710 nm band-pass filter.

Image Analysis: For analysis, MRI images were zero-filled to anin-plane pixel dimension of 50 �50 μm2. Mathematica (WolframResearch, Inc.) was used to analyze the slices directly upstream anddownstream from the cast for each time point. A circular region-of-interest (ROI) was drawn to select the vessel wall directly aroundthe vessel lumen. ROIs were drawn for the left and right carotidarteries, and the average signal intensities were calculated (IW).ROIs were also drawn in the surrounding muscle tissue and outsidethe mouse to calculate the muscle tissue signal intensity (IM) andthe noise level (IN), respectively. To allow the comparison of imagescollected on different days, the vessel wall signal intensities (IW)were normalized to those in a nearby muscle tissue ROI. The nor-malized percentage signal enhancement (%NSE) was calculated ac-cording to: %NSE = {[(IW,t/IM,t)/(IW,pre/IM,pre)] – 1} � 100.

Statistics: Statistical analyses were performed to test whether: (1)contrast enhancement was observed over time; (2) the two differentmicelle preparations differed in their contrast enhancement; and(3) the upstream and downstream plaques showed a different re-sponse. For this purpose, one-way analysis of variance (ANOVA)(p� 0.05) with Bonferroni correction for multiple comparisons wasperformed with Statgraphics Centurion XV (StatPoint, Inc., Vir-ginia, USA).

Supporting Information (see footnote on the first page of this arti-cle): Synthesis of lipid NIR664-DSPE and its HPLC and MALDI-TOF-MS characterization.

Acknowledgments

The authors gratefully acknowledge the expert technical assistanceof Léonie Niesen and Jo Habets with the animal experiments andMonika Breurken and Sanne Reulen with the preparation ofCNA35. We are indebted to Dr. Enzo Terreno (University of To-rino, Italy) for measuring NMRD profiles and to Dr. Henk Huin-ink (Eindhoven University of Technology) for his assistance withone-dimensional T1 profiling. This study was funded in part by theBesluit Subsidies Investeringen Kennisinfrastructuur (BSIK) pro-gram entitled Molecular Imaging of Ischemic Heart Disease (pro-ject number BSIK03033), the European Community EC-FP6-pro-ject Diagnostic Molecular Imaging (DiMI; LSHB-CT-2005–512146), and by the Dutch Heart Foundation (project number2006 T106).

[1] A. J. Lusis, Nature 2000, 407, 233–241.[2] J. Wang, N. Balu, G. Canton, C. Yuan, J. Magn. Reson. Im-

aging 2010, 32, 502–515.[3] W. Casscells, M. Naghavi, J. T. Willerson, Circulation 2003,

107, 2072–2075.[4] G. A. Plenz, M. C. Deng, H. Robenek, W. Volker, Atherosclero-

sis 2003, 166, 1–11.


[5] P. Libby, P. M. Ridker, A. Maseri, Circulation 2002, 105, 1135–1143.

[6] R. P. Choudhury, V. Fuster, J. J. Badimon, E. A. Fisher, Z. A.Fayad, Arterioscler. Thromb. Vasc. Biol. 2002, 22, 1065–1074.

[7] a) R. P. Choudhury, J. G. Aguinaldo, J. X. Rong, J. L. Kulak,A. R. Kulak, E. D. Reis, J. T. Fallon, V. Fuster, E. A. Fisher,Z. A. Fayad, Atherosclerosis 2002, 162, 315–321; b) J. E.Schneider, M. A. McAteer, D. J. Tyler, K. Clarke, K. M. Chan-non, R. P. Choudhury, S. Neubauer, J. Magn. Reson. Imaging2004, 20, 981–989.

[8] W. J. Mulder, G. J. Strijkers, K. C. Briley-Saboe, J. C. Frias,J. G. Aguinaldo, E. Vucic, V. Amirbekian, C. Tang, P. T. Chin,K. Nicolay, Z. A. Fayad, Magn. Reson. Med. 2007, 58, 1164–1170.

[9] R. M. Botnar, A. S. Perez, S. Witte, A. J. Wiethoff, J. Laredo,J. Hamilton, W. Quist, E. C. Parsons Jr., A. Vaidya, A. Kolod-ziej, J. A. Barrett, P. B. Graham, R. M. Weisskoff, W. J. Man-ning, M. T. Johnstone, Circulation 2004, 109, 2023–2029.

[10] M. Nahrendorf, F. A. Jaffer, K. A. Kelly, D. E. Sosnovik, E.Aikawa, P. Libby, R. Weissleder, Circulation 2006, 114, 1504–1511.

[11] P. M. Winter, A. M. Neubauer, S. D. Caruthers, T. D. Harris,J. D. Robertson, T. A. Williams, A. H. Schmieder, G. Hu, J. S.Allen, E. K. Lacy, H. Zhang, S. A. Wickline, G. M. Lanza, Art-erioscler. Thromb. Vasc. Biol. 2006, 26, 2103–2109.

[12] R. T. Megens, M. G. Oude Egbrink, J. P. Cleutjens, M. J. Ku-ijpers, P. H. Schiffers, M. Merkx, D. W. Slaaf, M. A. van Zand-voort, Mol. Imaging 2007, 6, 247–260.

[13] K. N. Krahn, C. V. Bouten, S. van Tuijl, M. A. van Zandvoort,M. Merkx, Anal. Biochem. 2006, 350, 177–185.

[14] Y. Zong, Y. Xu, X. Liang, D. R. Keene, A. Hook, S. Gurusid-dappa, M. Hook, S. V. Narayana, Embo J. 2005, 24, 4224–4236.

[15] J. Rauterberg, E. Jaeger, M. Althaus, Curr. Top. Pathol. 1993,87, 163–192.

[16] J. H. Lindeman, B. A. Ashcroft, J. W. Beenakker, M. van Es,N. B. Koekkoek, F. A. Prins, J. F. Tielemans, H. Abdul-Hus-sien, R. A. Bank, T. H. Oosterkamp, Proc. Natl. Acad. Sci.USA 2010, 107, 862–865.

[17] W. J. Mulder, G. J. Strijkers, G. A. van Tilborg, D. P. Cormode,Z. A. Fayad, K. Nicolay, Acc. Chem. Res. 2009, 42, 904–914.

[18] G. S. van Bochove, L. E. Paulis, D. Segers, W. J. Mulder, R.Krams, K. Nicolay, G. J. Strijkers, Contrast Media Mol. Im-aging 2011, 6, 35–45.

[19] C. Cheng, D. Tempel, R. van Haperen, A. van der Baan, F.Grosveld, M. J. Daemen, R. Krams, R. de Crom, Circulation2006, 113, 2744–2753.

[20] H. M. Sanders, G. J. Strijkers, W. J. Mulder, H. P. Huinink, S. J.Erich, O. C. Adan, N. A. Sommerdijk, M. Merkx, K. Nicolay,Contrast Media Mol. Imaging 2009, 4, 81–88.

[21] B. Ashok, L. Arleth, R. P. Hjelm, I. Rubinstein, H. Onyuksel,J. Pharm. Sci. 2004, 93, 2476–2487.


[22] S. W. Reulen, W. W. Brusselaars, S. Langereis, W. J. Mulder, M.Breurken, M. Merkx, Bioconjugate Chem. 2007, 18, 590–596.

[23] G. J. Strijkers, W. J. M. Mulder, R. B. van Heeswijk, P. M. Fre-derik, P. Bomans, P. C. M. M. Magusin, K. Nicolay, Magn. Re-son. Mater. Phys. Biol. Med. 2005, 18, 186–192.

[24] H. P. Huinink, H. M. Sanders, S. J. Erich, K. Nicolay, G. J.Strijkers, M. Merkx, O. C. Adan, Magn. Reson. Med. 2008, 59,1282–1286.

[25] H. M. Sanders, M. Iafisco, E. M. Pouget, P. H. Bomans, F. Nu-delman, G. Falini, G. de With, M. Merkx, G. J. Strijkers, K.Nicolay, N. A. Sommerdijk, Chem. Commun. 2011, 47, 1503–1505.

[26] A. Klink, J. Heynens, B. Herranz, M. E. Lobatto, T. Arias,H. M. Sanders, G. J. Strijkers, M. Merkx, K. Nicolay, V. Fuster,A. Tedgui, Z. Mallat, W. J. Mulder, Z. A. Fayad, J. Am. Coll.Cardiol. 2011, 58, 2522–2530.

[27] C. Yuan, W. S. Kerwin, M. S. Ferguson, N. Polissar, S. Zhang,J. Cai, T. S. Hatsukami, J. Magn. Reson. Imaging 2002, 15, 62–67.

[28] a) P. Caravan, B. Das, S. Dumas, F. H. Epstein, P. A. Helm, V.Jacques, S. Koerner, A. Kolodziej, L. Shen, W. C. Sun, Z.Zhang, Angew. Chem. 2007, 119, 8319; Angew. Chem. Int. Ed.2007, 46, 8171–8173; b) P. A. Helm, P. Caravan, B. A. French,V. Jacques, L. Shen, Y. Xu, R. J. Beyers, R. J. Roy, C. M.Kramer, F. H. Epstein, Radiology 2008, 247, 788–796; c) E.Spuentrup, K. M. Ruhl, R. M. Botnar, A. J. Wiethoff, A. Buhl,V. Jacques, M. T. Greenfield, G. A. Krombach, R. W. Gunther,M. G. Vangel, P. Caravan, Circulation 2009, 119, 1768–1775.

[29] a) C. von Bary, M. Makowski, A. Preissel, A. Keithahn, A.Warley, E. Spuentrup, A. Buecker, J. Lazewatsky, R. Cesati, D.Onthank, N. Schickl, S. Schachoff, J. Hausleiter, A. Schomig,M. Schwaiger, S. Robinson, R. Botnar, Circ. Cardiovasc. Im-aging 2011, 4, 147–155; b) M. R. Makowski, A. J. Wiethoff, U.Blume, F. Cuello, A. Warley, C. H. Jansen, E. Nagel, R. Razavi,D. C. Onthank, R. R. Cesati, M. S. Marber, T. Schaeffter, A.Smith, S. P. Robinson, R. M. Botnar, Nat. Med. 2011, 17, 383–388.

[30] B. A. Helms, S. W. Reulen, S. Nijhuis, P. T. de Graaf-Heuvel-mans, M. Merkx, E. W. Meijer, J. Am. Chem. Soc. 2009, 131,11683–11685.

[31] G. A. van Tilborg, G. J. Strijkers, E. M. Pouget, C. P. Reutel-ingsperger, N. A. Sommerdijk, K. Nicolay, W. J. Mulder, Magn.Reson. Med. 2008, 60, 1444–1456.

[32] S. Hak, H. M. Sanders, P. Agrawal, S. Langereis, H. Grull,H. M. Keizer, F. Arena, E. Terreno, G. J. Strijkers, K. Nicolay,Eur. J. Pharm. Biopharm. 2009, 72, 397–404.

[33] W. J. Mulder, G. J. Strijkers, A. W. Griffioen, L. van Bloois, G.Molema, G. Storm, G. A. Koning, K. Nicolay, BioconjugateChem. 2004, 15, 799–806.

[34] G. Rouser, S. Fkeisher, A. Yamamoto, Lipids 1970, 5, 494–496.[35] M. M. Bradford, Anal. Biochem. 1976, 72, 248–254.[36] N. Wu, J. Dai, F. J. Micale, J. Colloid Interface Sci. 1999, 215,

258–269.Received: January 5, 2012

Published Online: March 28, 2012

本文献由“学霸图书馆-文献云下载”收集自网络，仅供学习交流使用。

学霸图书馆（www.xuebalib.com）是一个“整合众多图书馆数据库资源，

提供一站式文献检索和下载服务”的24 小时在线不限IP

图书馆。

图书馆致力于便利、促进学习与科研，提供最强文献下载服务。

图书馆导航：

图书馆首页文献云下载图书馆入口外文数据库大全疑难文献辅助工具

http://www.xuebalib.com/cloud/

http://www.xuebalib.com/

http://www.xuebalib.com/cloud/


http://www.xuebalib.com/vip.html

http://www.xuebalib.com/db.php

http://www.xuebalib.com/zixun/2014-08-15/44.html


molecular mr imaging of collagen in mouse atherosclerosis...

Documents