2.1.

Near-Infrared Fluorescent Magnetonanoparticles

Briefly, a dextran-coated monodisperse iron oxide nanoparticle colloid (MION) was prepared and cross-linked with epichlorohydrin (CLIO). The mixture was then reacted with ammonia to yield ${\mathrm{NH}}_2$ -functionalized cross-linked iron oxide $\left(\mathrm{CLIO}\text{‐}{\mathrm{NH}}_2\right)$ , as described previously.^{4, 5} A NHS-ester near-infrared fluorochrome (cyanine 5.5, excitation/emission $674∕694\mathrm{nm}$ , Amersham Biosciences, United Kingdom) was subsequently conjugated to the nanoparticle by standard chemistry to produce the MFNP. The final product contained an average of 1.5 fluorochromes per nanoparticle. The mean particle size was $32\mathrm{nm}$ as determined by laser light scattering.⁴

2.2.

Experimental Model of Atherosclerosis for Intravital Imaging

All animal studies were approved by the Subcommittee on Research Animal Care at Massachusetts General Hospital. Apolipoprotein E deficient (apoE -/-) mice were employed as a well-characterized experimental model of atherosclerosis.⁹ Wild-type C57/B6 mice without atherosclerosis were included as additional controls. Female apoE -/- mice (Jackson Laboratory, Bar Harbor, Maine) were placed on an atherogenic diet (21% fat, 0.15% cholesterol, Harlan Teklad, Madison, Wisconsin) ad libitum beginning at 10 weeks until 32 to 36 weeks of age. Wild-type C57/B6 mice (Jackson Laboratory, Bar Harbor, Maine) were maintained on a regular diet until the time of the experiment.

For intravital imaging, MFNP was dissolved in phosphate-buffered saline and then injected via tail vein at a dose of 0.2-nmol fluorochrome/gram weight in the active (apoE -/- MFNP) group $(\mathrm{n}=7)$ and in the wild-type MFNP control group $(\mathrm{n}=4)$ . Saline (0.9%) was injected in the apoE -/- saline control group $(\mathrm{n}=3)$ . The total volume of the injection ranged between 100 and $150\mathrm{\mu }\mathrm{L}$ . After $24\mathrm{h}$ , mice were anesthetized by inhalation anesthesia (2% isoflurane, 1-L/min ${\mathrm{O}}_2$ ) using an isoflurane vaporizer (Braintree Scientific, Braintree, Massachusetts). We modified a surgical model that provides access to atherosclerotic plaques in the apoE -/- mouse.^{10, 11} Briefly, the distal right common carotid artery was carefully exposed from the periadventitial tissues (Fig. 1 ).

Fig. 1

Direct surgical exposure of the right carotid artery in an apoE -/- mouse in preparation for in vivo laser scanning fluorescence microscopy. A typical yellowish-white atherosclerotic plaque was present at the distal common carotid artery bifurcation. A fluorescent phantom was used as reference during imaging (noted adjacent to the plaque). The dashed box represents the region displayed in laser scanning fluorescence microscopy imaged in Fig. 2b.

Fig. 2

(a) Representative intravital laser scanning fluorescence microscopy Cy5.5 channel images from an apoE -/- MFNP-injected animal, an apoE -/- saline control, and a wild-type MFNP control. Strong NIRF signal in plaques was evident in apoE -/- MFNP-injected animals but not in controls. The Cy5.5 channel NIRF images were windowed identically. (b) Additional intravital laser scanning fluorescence microscopy images from a different apoE -/- MFNP-injected mouse also revealed bright, focal Cy5.5 signal in plaque. Signal in the ICG channel prior to indocyanine green injection was minimal at baseline. Subsequent ICG injection provided a definition of the vascular space and revealed an intravascular filling defect, further confirming the location of plaques detected in the MFNP channel.

Atherosclerotic plaques were visually identified and a fluorescent phantom was placed next to it as a reference during imaging. Animals were placed on a warmed glass plate and maintained on inhalation anesthesia during the one hour imaging session.

2.3.

Laser Scanning Fluorescence Microscopy of Atherosclerotic Plaques

After isolating carotid atheroma, multichannel fluorescence imaging was performed with a prototypical laser scanning fluorescence microscope (Olympus Corporation, Japan) specifically developed for intravital near-infrared imaging of mouse organs.¹² Three laser lines at 488, 633, and $748\mathrm{nm}$ were used. Image acquisition was $1\mathrm{s}$ . The FluoView 300 software program (Olympus) was used to control the microscope and collect images of $512\times 512$ pixels with a pixel size of about $5.4\mathrm{\mu }\mathrm{m}∕\mathrm{pixel}$ and a total image size of about $2.75\times 2.8\mathrm{mm}$ . Images were stored as multilayer 16-bit tagged image file format (TIFF) files. Images in the FITC channel (bandpass 505- to 525-nm filter), Cy5.5 channel (bandpass 660- to 730-nm filter), and Cy7 channel (long-pass 770-nm filter) were collected concomitantly using custom-made filters and dichroic mirrors (Olympus, Japan). A dry objective ( $4\times \mathrm{UplanApo}$ N.A. 0.16, Olympus, Japan) with a field of view of $3.25\mathrm{mm}$ and a theoretical lateral resolution of about $2.6\mathrm{\mu }\mathrm{m}$ at $680\mathrm{nm}$ was used. Wide spectral response photomultiplier tubes (model R928P, Hamamatsu, Japan) were used as detectors for both visible light and near-infrared signals. After initial images were obtained, a NIR vascular agent, indocyanine green (ICG), (excitation/emission $800∕830\mathrm{nm}$ , Akorn, Buffalo Grove, Illinois) was administered at a dose of $5\mathrm{mg}∕\mathrm{kg}$ , and postinjection images were taken in the LSFM Cy7 channel.

The NIRF signal was determined as mean signal intensities (SI) from manually drawn regions of interest (ROI) on areas of plaque using ImageJ software (National Institutes of Health, Bethesda, Maryland). The plaque target-to-background ratio (TBR) was calculated as follows: $\mathrm{TBR}=\left[\mathrm{SI}\right(\mathrm{plaque})∕\mathrm{SI}(\mathrm{blood}\left)\right]$ .

2.4.

Fluorescence Reflectance Imaging

After LSFM imaging, mice were euthanized and perfused with saline and 4% paraformaldehyde. Excised carotid arteries were imaged with a custom-made NIR fluorescence reflectance imaging (FRI) system¹³ equipped with a 150-W halogen light source and multichannel filter sets, including green fluorescent protein (GFP)/FITC (bandpass excitation 406 to $450\mathrm{nm}$ , bandpass emission 495 to $525\mathrm{nm}$ ), Cy5.5 (bandpass excitation 615 to $645\mathrm{nm}$ , bandpass emission 680 to $720\mathrm{nm}$ ), and ICG (bandpass excitation 716 to $756\mathrm{nm}$ , bandpass emission 780 to $820\mathrm{nm}$ ) (Omega Optical, Brattleboro, Vermont). Light, Cy5.5, and GFP/FITC images were obtained (0.075, 240, and 120-s acquisitions, respectively) with a 12-bit charge-coupled device camera (Kodak, Rochester, New York) equipped with a special C-mount lens. The NIRF signal was determined as mean signal intensities from manually drawn ROI. The plaque TBR was calculated as follows: $\mathrm{TBR}=\left[\mathrm{SI}\right(\mathrm{plaque})∕\mathrm{SI}(\text{normal}$ $\text{vessel}\left)\right]$ . Epifluorescence microscopic images of resected carotid plaques were obtained with an upright fluorescence microscope (Eclipse 80i, Nikon Instruments, Melville, New York) with a cooled charge-coupled device camera (Cascade 512B, Photometries, Tucson, Arizona).

2.5.

Immunohistochemistry and Fluorescence Microscopy

Excised tissues were embedded in optimal cutting temperature (OCT) compound (Tissue Tek, Sakura Finetek, Japan) and frozen in chilled isopentane. Five micrometer sections were cut from the carotid artery specimens, and fluorescence microscopy was performed in multiple channels using an upright Nikon Eclipse 80i mentioned before (FITC filter $480\pm 20\mathrm{nm}$ ex/ $535\pm 25\mathrm{nm}$ em/Q505LP bs; Cy5.5 filter $650\pm 22.5\mathrm{nm}$ ex/ $710\pm 25\mathrm{nm}$ em/Q680LP bs; ICG filter $775\pm 25\mathrm{nm}$ ex / $845\pm 27.5\mathrm{nm}$ em/ Q810LP bs). Exposure times ranged between $50\mathrm{ms}$ to $2\mathrm{s}$ . Images were analyzed with IPLab Spectrum software (Version 3.9.3, Scanalytics, Virginia). The exact same sections were then immunostained to identify macrophages within the plaque using a Mac-3 primary antibody (BD Biosciences, San Jose, California) and biotinylated secondary antibody and Vectastain ABC reagent (Vector Labs, Burlingame, California). The reaction was visualized with 3-amino-9-ethyl carbazol as substrate (AEC, Sigma Chemical, Saint Louis, Missouri), yielding red reaction product in the cytoplasm. Nuclei were counterstained with Mayer’s hematoxylin solution (Sigma Chemical). Adjacent sections were stained with hematoxylin and eosin for general morphology.

2.6.

Image Analysis

Digitized immunohistochemistry and fluorescence microscopy images were analyzed using IPLab Spectrum software. For quantitative image analysis, ten plaque sections were imaged with fluorescence microscopy. All images were then identically segmented by thresholding grayscale intensity at a fixed level (1750 arbitrary fluorescence units, chosen based on the average signal intensity across all images). Pixels with signal intensity above this threshold were counted as positive and summed for each section. The percent NIRF positive surface area per vessel section was calculated as follows: ${\mathrm{NIRF}}_{\mathrm{pos}}$ $\mathrm{SA}(\%)=100\%*\Sigma$ (pixels above threshold in the section)/ $\Sigma$ (total pixels in the section). The same sections were subsequently immunostained for macrophages (as described before) and then digitized and identically segmented by thresholding red intensity at a fixed level (88 arbitrary units, chosen based on the average signal intensity across all images) for all images. Pixels with signal intensity above this threshold were counted as positive. The percent Mac-3 positive surface area per vessel section $\left[\mathrm{Mac}\text{‐}{3}_{\mathrm{pos}}\mathrm{SA}\right(\%\left)\right]$ was calculated in a similar fashion as ${\mathrm{NIRF}}_{\mathrm{pos}}$ $\mathrm{SA}(\%)$ . For further analysis, 12 representative regions of interest (1000 pixel uniform area) in four plaque sections of varying macrophage content were randomly selected, and the average NIRF signal intensity in each ROI (in arbitrary fluorescence units) was measured using IPLab Spectrum software. The number of cells expressing a Mac-3 antigen (defined as a red reaction product associated with a blue hematoxylin nuclear counterstain) was then counted by an experienced pathologist (EA) in ten high-power fields for each corresponding ROI region using an eyepiece reticle ( $10\mathrm{mm}$ , 1-mm division; Fisher Scientific) and correlated with the average NIRF signal intensity of the same ROI.

2.7.

Statistical Methods

All results are reported as $\mathrm{mean}\pm \mathrm{SD}$ . For differences between multiple groups, a one-way ANOVA followed by a post-hoc Tukey’s test for multiple comparisons was used. All statistics including linear regressions were performed using GraphPad Prism version 4.03 for Windows (GraphPad Software, San Diego, California). A p-value of $<0.05$ was considered significant.

3.1.

Laser Scanning Fluorescence Microscopy Identifies Magnetofluorescent Nanoparticles in Carotid Atheroma

Discrete right common carotid atherosclerotic lesions were visually identified in situ (Fig. 1) and imaged in vivo in all apoE -/- animals ( $\mathrm{n}=10$ , Fig. 1); lesions were not present in any of the four wild-type animals. In apoE -/- animals injected with MFNP, the LSFM images showed focal areas of NIRF signal in carotid atheroma, consistent with MFNP uptake [Fig. 2a ]. In comparison, there was minimal NIRF signal in the apoE -/- saline control group [Fig. 2a] consistent with low autofluorescence in the NIR window.⁸ A relatively low level of intravascular NIRF signal was also seen, likely from residual circulating MFNP. In addition, no significant vascular wall NIRF signal was seen in the wild-type MFNP group, consistent with the absence of atherosclerosis. Subsequent injection of a vascular agent, indocyanine green (ICG), defined the arterial lumen and visualized plaques as angiographic filling defects [Fig. 2b] The target-to-background ratio between the plaque and the adjacent blood (TBR) in the Cy5.5 channel in the apoE -/- MFNP group was $3.9\pm 1.8$ .

3.2.

Ex-vivo Near-Infrared Fluorescence Imaging Confirms Magnetofluorescent Nanoparticle Localization in Plaque

Fluorescence reflectance images and epifluorescence microscopic images of excised carotid atheromata were obtained to corroborate the LSFM findings. The images showed focal NIRF signal in plaque, with minimal NIRF signal in the normal arterial wall [Fig. 3a ]. The plaque target-to-background ratio was $4.8\pm 1.8$ in the MFNP-injected group, $1.3\pm 0.2$ in the saline-injected group, and $1.1\pm 0.1$ in the wild-type group ( $\mathrm{p}<0.01$ ANOVA):[Fig. 3b]

Fig. 3

Fluorescence reflectance imaging of an excised carotid atheroma from an apoE -/- MFNP-injected animal showed bright enhancement in the Cy5.5 channel, consistent with MFNP localization in plaque and distinct from autofluorescence seen in the FITC channel. (b) The plaque TBR from apoE -/- MFNP-injected animals was significantly greater than the control groups. $*\mathrm{p}<0.01$ versus each of the two control groups.

3.3.

Atherosclerotic Plaque Near-Infrared Fluorescence Signal Colocalizes with Macrophages

Multiwavelength fluorescence microscopy was performed on frozen sections of carotid arteries. Fluorescence microscopy in the Cy5.5 channel showed strong signal in regions of atherosclerotic plaques [Fig. 4a ]. The plaque NIRF signal colocalized with the cellular-rich zones of plaques, typically superficial in the employed atherosclerotic model. Images in multiple fluorescence channels were obtained and confirmed that Cy5.5 signal was distinct from autofluorescence [Fig. 4a]. A low level of nonspecific ICG fluorescent signal was seen on the endoluminal surface of the vessels and in the periadventitial areas. Correlative immunohistochemistry confirmed that the NIRF signal colocalized with areas rich in macrophages [Fig. 4b], consistent with earlier studies.^{4, 5} The percent of NIRF-positive surface area per vessel section correlated with the percent of Mac-3-immunostaining-positive surface area per vessel section $(\mathrm{r}=0.77)$ [Fig. 5a ]. Furthermore, when macrophage staining was quantified as the number of cells per region of interest (ROI), the NIRF-macrophage correlation became stronger $(\mathrm{r}=0.86)$ [Fig. 5b].

Fig. 4

(a) Fluorescence microscopy showed focal NIR signal in cellular-rich areas of plaques from apoE -/- MFNP-injected animals and negligible NIR signal in control plaques. The Cy5.5 channel NIRF images were windowed identically. (b) Correlative immunohistochemistry confirmed colocalization of the NIRF signal with macrophages as detected by Mac-3 immunostaining.

Fig. 5

(a) Quantitative histological analysis showed a strong correlation between the percent NIRF-positive plaque signal $[{\mathrm{NIRF}}_{\mathrm{pos}}$ $\mathrm{SA}(\%)]$ and the macrophage Mac-3 immunostained surface area $\left[\mathrm{Mac}\text{‐}{3}_{\mathrm{pos}}\mathrm{SA}\right(\%\left)\right]$ $(\mathrm{r}=0.77)$ . (b) Furthermore, the average NIRF signal intensity per ROI also correlated strongly with the number of macrophages per ROI $(\mathrm{r}=0.86)$ . $\mathrm{SI}=\mathrm{signal}\mathrm{intensity}$ , $\mathrm{AFU}=$ arbitrary fluorescence units.