SPECT/CT Imaging of High-Risk Atherosclerotic Plaques using Integrin-Binding RGD Dimer Peptides

Vulnerable atherosclerotic plaques with unique biological signatures are responsible for most major cardiovascular events including acute myocardial infarction and stroke. However, current clinical diagnostic approaches for atherosclerosis focus on anatomical measurements such as the degree of luminal stenosis and wall thickness. An abundance of neovessels with elevated expression of integrin αvβ3 is closely associated with an increased risk of plaque rupture. Herein we evaluated the potential of an αvβ3 integrin-targeting radiotracer, 99mTc-IDA-D-[c(RGDfK)]2, for SPECT/CT imaging of high-risk plaque in murine atherosclerosis models. In vivo uptake of 99mTc-IDA-D-[c(RGDfK)]2 was significantly higher in atherosclerotic aortas than in relatively normal aortas. Comparison with the negative-control peptide, 99mTc-IDA-D-[c(RADfK)]2, proved specific binding of 99mTc-IDA-D-[c(RGDfK)]2 for plaque lesions in in vivo SPECT/CT and ex vivo autoradiographic imaging. Histopathological characterization revealed that a prominent SPECT signal of 99mTc-IDA-D-[c(RGDfK)]2 corresponded to the presence of high-risk plaques with a large necrotic core, a thin fibrous cap, and vibrant neoangiogenic events. Notably, the RGD dimer based 99mTc-IDA-D-[c(RGDfK)]2 showed better imaging performance in comparison with the common monomeric RGD peptide probe 123I-c(RGDyV) and fluorescence tissue assay corroborated this. Our preclinical data demonstrated that 99mTc-IDA-D-[c(RGDfK)]2 SPECT/CT is a sensitive tool to noninvasively gauge atherosclerosis beyond vascular anatomy by assessing culprit plaque neovascularization.

to predict the location of possible adverse events 3,4 . Current diagnostic strategies predominantly focus on anatomical issues such as myocardial ischemia 5 , hemodynamic luminal narrowing, or morphological abnormalities of atheromas, but not on the biological aspects of atherosclerotic lesions. This traditional strategy has proven disappointing in preventing myocardial infarction or prolonging life, except in limited patient groups [6][7][8] .
Atherosclerotic plaques comprise a heterogeneous mixture of cellular and acellular elements 9 . In the past decade, considerable efforts have been devoted to determine the specific compositional features of unstable vulnerable plaques [9][10][11][12] . Atherosclerotic plaques at the highest risk of rupture clearly exhibit a large lipid-rich necrotic core, thin fibrous cap, neovascularization, spotty calcium, and abundant inflammatory cells; these features are distinctive from those of stable lesions [9][10][11] . Therefore, the assessment of plaque composition is potentially more important than the traditional detection of simple intraluminal stenosis for predicting devastating arterial events 7,8,13,14 . As a result, there is a compelling need to develop diagnostic imaging techniques to gauge the biological details of plaques that trigger the conversion of asymptomatic atheromas to rupture-prone lesions and subsequent fatal thrombosis. Molecular imaging strategies now provide an approach other than assessing vessel stenosis and wall thickness, and shed light on the in vivo pathology of atherosclerotic plaques [13][14][15][16] .
In particular, neovascularization is a key process of advanced atherosclerotic plaques and an independent predictor of future adverse clinical outcomes [17][18][19] . Newly formed vasculature within a plaque is closely related to the inflammatory process, which is another important determinant of plaque rupture because it facilitates monocyte recruitment and transmigration. Intraplaque hemorrhage, a critical event that provokes lesion destabilization by providing erythrocyte-derived phospholipids and free cholesterol, is caused by damage to neovessels because of their immature, fragile characteristics resulting from the lack of smooth muscle cells 20 . Aspects of neoangiogenesis have therefore surfaced as emerging major targets for molecular imaging of atherosclerosis. Angiogenesis has been extensively studied for cancer diagnosis 21 ; however, imaging neovasculature to identify patients at risk for major clinical manifestations of atherosclerosis is relatively new, compared to the long history of imaging plaque inflammation [22][23][24] . Previous investigations of imaging neovascular proliferation in plaques are limited to trials using nanoparticle-enhanced molecular magnetic resonance imaging (MRI) 25,26 and to a few recent studies using positron emission tomography (PET) with simple arginyl-glycyl-aspartic acid (RGD) monomer peptides 27,28 .
The RGD, an excellent targeting moiety for integrin α v β 3 of activated endothelial cells, has been successfully validated for imaging tumor angiogenesis in numerous preclinical and clinical studies 21,29,30 . We recently developed a new radiotracer, 99m Tc-labeled RGD peptide ( 99m Tc-IDA-D-[c(RGDfK)] 2 ) and performed single photon emission computed tomography (SPECT) imaging for targeting glioblastoma 31 . Compared to other reported RGD monomer-based agents 30,32,33 [e.g., 99m Tc-(CO) 3 -pyrazoyl conjugate of c(RGDyK) 34 and 123 I-c(RGDyV)], the developed RGD dimer agent 99m Tc-IDA-D-[c(RGDfK)] 2 showed specific integrin-binding affinity, high tumor accumulation and desirable pharmacokinetic properties for tumor xenograft imaging 31 . Similar to tumor angiogenesis imaging, this radiotracer is expected to be effective for imaging neovessel-rich atherosclerotic plaques and show better imaging performance compared with previously reported monomeric RGD probe based approaches 27,28 .
Here we describe a molecular imaging strategy that uses α v β 3 integrin-targeted probe 99m Tc-IDA-D-[c(RGDfK)] 2 with SPECT to achieve improved atherosclerosis staging through assessment of neovascularization. To illustrate the utility of this approach, we demonstrate SPECT/CT imaging of atherosclerotic mouse models and analyze correlation between in vivo uptake of the radiotracer and ex vivo autoradiography signal and corresponding histopathological signatures. Comparative measurements with conventional monomeric RGD derivatives reveal superior sensitivity of the designed dimer RGD probe for noninvasive nuclear imaging as well as tissue fluorescence imaging. Finally we discuss the potential of angiogenesis targeted approach for ideal noninvasive imaging to pinpoint high-risk atherosclerotic plaques before they lead to fatal clinical events.

Development of High-risk Atherosclerosis and Validation Studies.
To explore the ability of α v β 3 integrin-targeted probes to detect atherosclerotic lesions with a high risk of rupture, we established mouse models of atherosclerosis by feeding a high-cholesterol diet to apoE-deficient (ApoE-/-) mice. After 40 weeks of this special diet, their hearts and aortas were excised and carefully analyzed. As Fig. 1A shows, atherosclerotic lesions (white areas) were present in the aorta-predominantly in the aortic arch; in the origins of the brachiocephalic, left subclavian, and left common carotid arteries; and throughout the thoracic and abdominal aorta. After the gross anatomical examination, we performed Oil-Red-O (ORO) staining of the excised aortic arch tissues and descending aortas in the apoE transgenic mice and the wild-type C57BL/6J mice (Figs 1B,C). Lipid pool zones with a strong ORO-positive response were identified in the apoE-deficient mice, compared to the control mice, which demonstrated significant development of lipid-rich high-risk lesions; this was observed in a previous study 35 .
The x-ray CT scans with a vascular contrast agent revealed detailed anatomy of heart and aorta regions, which included the aortic arch and descending aorta structures (Figs 1D-I). As a complement to SPECT imaging, x-ray CT provides structural high-resolution visualization of specific locations for developed atherosclerotic lesions. radiolabeling protocol as in a previous report 31 , we prepared 99m Tc-IDA-D-[c(RGDfK)] 2 , a diagnostic imaging agent for angiogenesis with chemical and radiochemical purities greater than 99% and specific activity greater than 55 GBq/μ mol. This agent was designed to have enhanced hydrophilicity of the integrin-binding RGD dimer peptide. Its superior pharmacokinetic properties and high metabolic stability have been proven in a previous study 31 . Herein we evaluated the feasibility of SPECT/CT imaging using 99m Tc-IDA-D-[c(RGDfK)] 2 to noninvasively detect high-risk plaques in established mouse models of atherosclerosis by feeding a high-cholesterol diet to ApoE-/-mice ( Fig. 2). High local signals of 99m Tc-IDA-D-[c(RGDfK)] 2 in the aortic arch lesions (i.e., well-known prevalent plaque regions, which is identified by contrast-enhanced CT scans in Fig. 2A 35 ) was detected at 30 min post-injection (Fig. 2B). We next compared the imaging performance by using the same mouse with the 99m Tc-labeled negative-control peptide 99m Tc-IDA-D-[c(RADfK)] 2 (Fig. 2C) on the next day. In the aortic arch region containing atherosclerotic plaques, we observed marked uptake by 99m Tc-IDA-D-[c(RGDfK)] 2 but only scant uptake by 99m Tc-IDA-D-[c(RADfK)] 2 . The injection of pure radioisotope 99m Tc-pertechnetate in a transgenic mouse showed no specific accumulation in the same territories and only nonspecific uptake in the salivary glands (Fig. 2D). An unmanipulated wild-type C57BL/6J control mouse that received 99m Tc-IDA-D-[c(RGDfK)] 2 also showed no significant radioactivity, except noncleared radiotracer signal in the gall bladder (Fig. 2E). Quantification of 99m Tc-IDA-D-[c(RGDfK)] 2 accumulation revealed significantly higher uptake in the atherosclerotic aortas than in the relatively normal thoracic aortas [n = 4, percentage injected dose per gram of tissue (%ID/g) was 2.98 ± 0.64 versus 0.41 ± 0.10, respectively; P < 0.001, Fig. 2F]. Preferential in vivo accumulation in aortic plaque suggested that 99m Tc-IDA-D-[c(RGDfK)] 2 has good specificity for use in staging high-risk atherosclerosis.

QD605-D-[c(RGDfK)] 2 and QD605-c(RGDyK) Uptake by Tissue-based Assay. To verify in vivo
SPECT/CT imaging data (Fig. 4), we next investigated the use of a fluorescently labeled RGD-dimer peptide, QD605-D-[c(RGDfK)] 2 , for more sensitive targeting of atherosclerotic plaques compared to a fluorophore-conjugated RGD-monomer peptide (i.e., QD605-c(RGDyK)) in the tissue-based assay. To compare the RGD dimer and monomer peptides' uptake in aortic tissue sections, we conjugated fluorescent quantum dots (QD605, emission approximately 605 nm) to each peptide and used them to stain the plaque cryosections. Using confocal fluorescence microscopy, we readily identified the binding signal in consecutive sections. This study showed a substantially elevated uptake of QD605-D-[c(RGDfK)] 2

Discussion
Most acute vascular events result from sudden luminal thrombosis due to rupture of an atherosclerotic plaque. Preventing such complications of atherosclerosis is the most urgent need to improve the survival of patients with cardiovascular disease. Contrast-enhanced x-ray angiography, which is the gold standard imaging tool used in clinics, only identifies luminal anatomy and rarely captures arterial wall characteristics, although the association between plaque composition and lesion instability has become obvious 5 . Thus, accurate discrimination between stable and vulnerable plaques remains a clinical challenge 5,7,8,13,14 .
In the present study, we demonstrated the feasibility of SPECT/CT imaging with α v β 3 integrin-targeted 99m Tc-IDA-D-[c(RGDfK)] 2 for specific detection of rupture-prone high-risk atherosclerotic plaques in a mouse model of atherosclerosis. The use of 99m Tc-IDA-D-[c(RGDfK)] 2 was based on the premise that neovascularization is deeply associated with atheroma disruption or erosion, and thus the expression of integrin α v β 3 by angiogenic endothelial cells can provide an important target for atherosclerosis staging. Imaging by SPECT/CT showed focal increases in the 99m Tc-IDA-D-[c(RGDfK)] 2 signal in advanced lipid-rich plaques inside a mouse aorta, compared to the significantly low signal in normal areas of the same aorta or in the aortas of wild-type control mice (2.98 ± 0.64%ID/g versus 0.41 ± 0.10%ID/g for atherosclerotic and normal aortas, respectively; P < 0.001). Autoradiography and histopathology corroborated the in vivo data by revealing specific vulnerable plaque characteristics such as a large necrotic core and a thin fibrous cap.
The biological insights and experimental knowledge in understanding key processes of atherosclerosis that contribute to a lesion's initiation, progression and complication have advanced markedly 9-12 . This has spurred many efforts to develop molecular imaging strategies to identify destabilized atherosclerotic plaques that are likely provoke the onset of acute thrombotic events 13,14,16 . The cyclic peptide RGD is perhaps the best known ligand for targeting angiogenesis through its specific binding affinity for integrin α v β 3 and it has been widely used as a cancer diagnostic agent 21,29,30 . An increasing number of reports have recently indicated that angiogenesis is a very pertinent hallmark that can be used for staging atherosclerosis [17][18][19] . Shown in this study demonstrates successful application of RGD peptides to sensitively detect neoangiogenesis in high-risk lesions for clinical SPECT/CT imaging.
Focus of this study was particularly given to comparing monomeric and dimeric RGD-based tracers for gauging atherosclerosis. A direct comparison of 99m Tc-IDA-D-[c(RGDfK)] 2 with RGD monomer based 123 I-c(RGDyV) indicate that 99m Tc-IDA-D-[c(RGDfK)] 2 has better in vivo targeting with a 3.7-fold higher affinity for unstable atherosclerotic lesion. In vitro tissue assay using QD605-D-[c(RGDfK)] 2 and QD605-c(RGDyK) also showed superior binding property of the RGD dimer-based probe. Different fluorophore labeling using carboxyfluorescein and intensity quantification showed identical results (Supplementary Figure 4), proving such sensitivity difference arise from intrinsic property of RGD derivatives not fluorophore conjugation. Enhanced specific targeting may be due to improved avidity to integrin α v β 3 of dimeric tracer over monomeric form as the interaction between integrins and their physiologic binding partners in the extracellular matrix involves multivalent binding sites with clustering of integrins.
Of note, the approach showcased in this study can be readily translated into the clinic, where its ultimate utility can be assessed. The noninvasive nuclear imaging technique SPECT has high sensitivity and quantification ability. The radionuclide 99m Tc can be obtained by daily elution from the 99 Mo/ 99m Tc-generator, and is thus convenient and suitable for routine clinical use. Furthermore, in vivo imaging showed sufficient signal intensity for delineating aortic lesions and superior imaging performance compared with  other monomer RGD based strategies. As expected, based on our previous study 31 , the mean uptake of 99m Tc-IDA-D-[c(RGDfK)] 2 in atherosclerotic plaques (2.98 ± 0.34%ID/g) was several folds lower (because of volume difference) than the uptake reported in α v β 3 integrin-expressing tumors (12.4 ± 3.89%ID/g) 31 .
Despite the small dimension of atherosclerotic lesions, our results suggest that visualization of high-risk plaques in human artery may be possible with α v β 3 integrin-specific SPECT/CT imaging. It must be highlighted that 99m Tc-IDA-D-[c(RGDfK)] 2 depicts minimum background signal in chest SPECT/CT, which is especially beneficial for coronary artery imaging. By contrast, atherosclerosis imaging with 18 F-FDG PET has been suffered with great background myocardial uptake because of glucose consumption by the heart muscle itself 36 . In addition, 99m Tc-IDA-D-[c(RGDfK)] 2 SPECT/CT may have an advantage over previous preclinical MRI studies for complete targeting of intraplaque microvessels because of the small probe size compared to MRI contrast agents, α v β 3 integrin-specific nanoparticles 25,26 .
To reach beyond the tools available in laboratory research, generalized, large, prospective clinical trials are needed to confirm the illustrated results of preclinical small animal imaging with 99m Tc-IDA-D-[c(RGDfK)] 2 SPECT/CT. Typical atherosclerotic plaque regions that can be imaged in mouse models only include the larger vessels such as the abdominal aorta, the carotid arteries, the aortic arch, and the aortic root, as displayed in this report. It is challenging, but there is great interest in directly imaging thrombosis-prone plaques in small coronary arteries, which commonly cause acute myocardial infarction. Therefore Animal SPECT/CT Imaging and Analysis. We anesthetized mice with 2% isoflurane gas anesthesia. The ApoE-/-mice were then administered intravenously 99m Tc-IDA-D-[c(RGDfK)] 2 (n = 4), 99m Tc-IDA-D-[c(RADfK)] 2 (n = 3), 123 I-c(RGDyV) (n = 1), or 99m Tc-pertachnate (n = 2) (each 37 MBq in 0.3 mL of saline) and C57BL/6J was administered 99m Tc-IDA-D-[c(RGDfK)] 2 (n = 3). The mice were placed supine on the bed of an animal SPECT/CT scanner (NanoSPECT/CT, Bioscan Inc., Washington DC, USA). At 30 min post-injection, a high-resolution static scan of chest region was acquired in helical scanning mode in 24 projections during a 30 min period using a four-head scanner with 4 × 9 (1.4 mm) pinhole collimators. The energy window was set at 140 keV ± 15%. The SPECT imaging was followed by CT image acquisition with the animal in the same position. The CT images were obtained with the x-ray source set at 45 kVp and 177 μ A after the injection of vascular contrast agent Fenestra VC (MediLumine Inc., Montreal, QC, Canada) to demonstrate the ability to visualize the vasculature with the CT scanner for correlation to a SPECT studies (Figs 1D-I and 2A-C). Following intravenous injection of 10 μ l/g of Fenestra VC, CT was performed at a mean time of 10 minutes post-injection with acquisition time 270 s per CT scan. The analysis software HiSPECT (Version 1.0, Bioscan Inc., Poway, CA, USA) and InVivoScope (Version 1.43, Bioscan Inc., Poway, CA, USA) were used for image reconstruction and quantification, respectively. The SPECT images were reconstructed to produce an image size of 176 × 176 × 136 voxels with a voxel size of 0.2 × 0.2 × 0.2 mm. The CT images were 48 μ m resolution acquisition with a voxel-pixel size of 0.20 : 0.192 mm. Details for image reconstruction and processing have been reported elsewhere 31 . Manually drawn two-dimensional regions of interest (ROIs) or three-dimensional volumes of interest (VOIs) were used to determine the accumulated radioactivity in units of %ID/g (with decay corrected to the time of injection) using 37-55.5 MBq radioactivity of 99m Tc as the reference source.
Statistical Analysis. Comparisons between regions were performed using SPSS Statistics 19 (IBM, Armonk, NY, USA). All data were analyzed using SigmaStat software, version 3.5 (Systat Software, San Jose, CA, USA). Differences with a P value less than 0.001 were considered significant.
Autoradiography. After performing animal SPECT/CT imaging with 99m Tc-IDA-D-[c(RGDfK)] 2 and the negative-control peptide 99m Tc-IDA-D-[c(RADfK)] 2 , we dissected the aortic tissues of the atherosclerotic animals and laid these specimens flat on a phosphor imager (Fuji BAS-5000; Fujifilm Life Sciences, Stamford, USA) for 24 h. The generated autoradiographs were analyzed using a computer-based image analysis system (Multi Gauge software, Fujifilm Life Sciences). The specific uptake was expressed as photostimulated luminescence per millimeter squared (PSL/mm 2 ).

Histological Evaluation.
To confirm the development of typical lipid-rich atherosclerotic plaques, the excised aortas were imaged intact. They then underwent Oil-Red-O staining. All photographic images were obtained by a digital camera.
To evaluate noninvasive imaging of atherosclerotic lesions, histopathology was performed on in vivo imaged aortas with an intense 99m Tc-IDA-D-[c(RGDfK)] 2 signal. After undergoing SPECT/CT and autoradiography imaging, the excised aortas were fixed with 10% formalin, embedded in paraffin, cut into 5-μ m sections and deparaffinized. The sections were subsequently stained with hematoxylin and eosin or Masson's trichrome stain to characterize the morphology and composition of the recorded peak signal of the aorta. Bright field color micrographs were obtained on a BX51 microscope equipped with DP71 camera (Olympus Optical Co., Ltd., Tokyo, Japan).
In Vitro Fluorescent Staining Assay in Plaque Tissue. Aorta tissues were dissected from apoE transgenic (ApoE-/-) mice (which were fed a high-cholesterol diet for 40 weeks) and wild-type C57BL/6J mice. The resected aortas were embedded in a tissue-freezing medium (Triangle Biomedical Sciences, Durham, NC, USA), frozen, and consecutively cryosectioned in 10-μ m segments using a Cryocut Microtome (CM3050S, Leica, Solms, Germany). The tissue sections were thaw-mounted onto silane-coated microscope slides (Muto Pure Chemicals co., Tokyo, Japan), dried in an aeration room, and stored at -80 °C until use. To confirm in vivo data and to compare plaque tissue binding ability, we performed fluorescence staining with QD605-D-[c(RGDfK)] 2 and QD605-c(RGDyK) on consecutive sections by the following steps: the 10-μ m sections were washed with phosphate-buffered saline (PBS) and incubated with either QD605-D-[c(RGDfK)] 2 or QD605-c(RGDyK) (10 pmol of peptide, 4 μ g of QD605) for 30 min. The slides were then washed with PBS several times, counterstained with Hoechst33342 and mounted with Prolong Gold Antifade Reagent (Life Technologies, Carlsbad, CA, USA). The fluorescence images were captured with a confocal microscope (TCS NT4D, Leica, Solms, Germany) to identify the binding difference between the two agents.