Integrin-targeted Imaging of Inflammation in Vascular Remodeling

Abstract

Objective

Inflammation plays a key role in the development of vascular diseases. Monocytes and macrophages express αvβ3 integrin. We used an αv integrin-specific tracer, ^99mTc-NC100692, to investigate integrin-targeted imaging for detection vessel wall inflammation.

Methods and Results

The binding of a fluorescent homologue of NC100692 to αvβ3 on human monocytes and macrophages was shown by flow cytometry. Vessel wall inflammation and remodeling was induced in murine carotid arteries through adventitial exposure to CaCl₂. NC100692 microSPECT-CT imaging was performed after 2 and 4 weeks and showed significantly higher uptake of the tracer in CaCl₂-exposed left carotids compared to sham-operated contra-lateral arteries. Histological analysis at 4 weeks demonstrated significant remodeling of left carotid arteries and considerable macrophage infiltration which was confirmed by real-time polymerase chain reaction. There was no significant difference in normalized αv, β3 or β5 mRNA expression between right and left carotid arteries. Finally, NC100692 uptake strongly correlated with macrophage marker expression in carotid arteries.

Conclusions

NC100692 imaging can detect vessel wall inflammation in vivo. If further validated, αv-targeted imaging may provide a non-invasive approach for identifying patients who are at high risk for vascular events and tracking the effect of anti-inflammatory treatments.

Inflammation is a common feature of many vascular diseases and plays a central role in their pathogenesis. Typical examples include atherosclerosis and aneurysm formation where an inflammatory process is critical to the development of the disease and its complications. It is therefore not surprising that many therapeutic interventions aim at modulating vessel wall inflammation. One of the limitations of the modern approach to managing vascular diseases is the lack of reliable approaches to detecting and tracking the effect of interventions on vessel wall biology. This may be addressed by targeting molecular signatures of relevant processes by molecular imaging.

Endothelial activation, leukocyte recruitment and activation, and matrix remodeling are integral parts of inflammation, which is closely intertwined with vessel wall angiogenesis. αv integrin-targeted imaging has been introduced for detection of angiogenesis associated with myocardial infarction ¹, peripheral arterial disease ², atherosclerosis ³ and neoplasm ⁴. In studies of αvβ3-targeted imaging of ischemia-induced angiogenesis part of the tracer localized in what appeared to be the inflammatory infiltrate associated with angiogenesis ². This led us to investigate whether αv-targeted imaging may be used for detection of vessel wall inflammation in vivo. Here, we demonstrate that peripheral blood monocytes and monocyte-derived macrophages express αvβ3 integrin and bind to NC100692, a cyclic RGD peptide with specificity for activated αv integrins ^{2, 5, 6}, to levels comparable to that of endothelial cells (ECs). In a mouse model of vessel wall inflammation, NC100692 uptake was clearly detectable by microSPECT/CT imaging in chemically injured carotid arteries and the uptake correlated well with the presence of macrophages.

Material and Methods

Materials

Materials were obtained from Sigma (St. Louis, MO), unless indicated otherwise. NC100692 precursor and its fluorescent-labeled homologue were provided by GE healthcare (Buckinghamshire, UK). NC100692 radiolabeling with ^99mTc was performed using kits provided by GE according to the manufacturer’s instructions ⁶. Each kit contained ~44 nmol NC100692 (molecular weight 1697) and was labeled with 1.1 GBq sodium pertechnetate (^99mTc).

Cell Culture

Human umbilical vein ECs were isolated and cultured as described ⁷. Peripheral blood mononuclear cells (PBMCs) were isolated under protocols approved by Yale Human Investigation Committee from normal anonymous donors’ leukapheresis product by gradient density centrifugation following standard procedures. Monocytes were isolated to high purity from PBMCs by magnetic cell sorting using anti-CD14-coated beads according to manufacturer’s instructions(Stemcell Technologies, Vancouver, BC). Monocyte purity was verified by flow cytometry and was found to be >85%. Purified monocytes were cultured for 10 days in RPMI plus 10% fetal bovine serum (Lonza, Walkersville, MD), 2 mM L-glutamine, 100 U/ml penicillin and 100 ug/ml streptomycin in the presence of recombinant GM-CSF (50ng/ml, PeproTech, Rocky Hill, NJ) to generate type 1 macrophages ⁸.

Flow cytometry

Expression of surface proteins was analyzed by staining live cells with conjugated anti-CD14 (BD pharmigen), αvβ3 integrin (LM609, Millipore corporation, Temecula, CA) antibody, the corresponding isotype control antibodies or a fluorescent RGD peptide homologue of NC100692 (GE Healthcare). To investigate the effect of integrin activation, cells were exposed to MnCl₂ (0.2 mM) in calcium and magnesium free phosphate buffered saline for 10 minutes before staining. MnCl₂ was kept in all buffers during staining and flow cytometry. At least 2,500 cells that satisfied a gate on forward and side scatter were acquired using a FACS Calibur flow cytometer (Becton Dickinson, Mountain View, CA). Data analysis was performed using CellQuest software (San Jose, CA).

Animal model

Seventeen animals underwent surgery to induced carotid artery inflammation and remodeling as described ⁹ (Supplemental Fig I). Briefly, in 8- to 10-week old female apoE^−/− mice (Jackson Laboratory, Bar Harbor, ME) fed a high-cholesterol chow (1.25% cholesterol, Harlan Teklad, Madison, WI) for 1 week the carotid arteries were surgically exposed under anesthesia (ketamine 100 mg/kg and xylazine 10 mg/kg, ip). The left common carotid artery just below carotid bifurcation was adventitially painted with a 10% solution of CaCl₂ for 20 minutes. The opposite carotid artery was exposed to normal saline and served as control for imaging studies. Ibuprofen (0.11 mg/kg/day, po) was used for postoperative analgesia. Experiments were performed according to regulations of Yale University’s Animal Care and Use Committee.

Imaging

MicroSPECT/CT imaging was performed as described ^{9, 10} with minor modifications on 9 animals at two weeks after surgery. Of these, 5 underwent repeat imaging followed by tissue analysis at 4 weeks. Images could not be obtained from two of this latter group of animals. An additional group consisting of 5 animals underwent imaging followed by tissue analysis at 4 weeks. Images obtained at either 2 (n=9) or 4 (n=8) weeks were combined for analysis of tracer uptake at each time point. Briefly, 41 ± 1.1 MBq NC100692 (^99mTc-labeled) was administered through a right jugular vein intravenous catheter placed under anesthesia (isofluorane 1–3%). Animals were imaged after 2 hours on a high-resolution small animal imaging system (X-SPECT, Gamma Medica-Ideas, Northridge, CA) with 1-mm low-energy pinhole collimators. The following acquisition parameters were used for microSPECT imaging: 360 degree, 128 projections, 30 seconds/projection (~80 minute image acquisition), with 140 keV photopeaks ±10% window. After completion of microSPECT imaging, animals were injected with a continuous infusion of iodinated CT contrast (iohexol 100 μL/min) over 2 minutes or Fenestra (200μl, ART Advanced Research Technologies, Montreal, QC, Canada), and CT imaging was performed (energy 75 kVp/280 μA, matrix 512×512) to identify anatomic structure. To avoid tissue damage we did not perform any additional ex vivo imaging and preserved the tissue immediately for mRNA and immune-histological analysis. To establish imaging specificity, three animals were injected with 50-fold excess unlabeled precursor prior to NC100692 imaging at two weeks after surgery. For quantitative analysis of tracer uptake, cylindrical regions of interest (ROIs) were drawn at the level of carotid artery bifurcation (2×2×2 mm). A ROI immediately posterior to both carotids was used to calculate the background activity. Data were expressed as background-corrected counts per voxel (cpv)/MBq injected.

Morphometric analysis and immunostaining

After imaging, carotid arteries were harvested, embedded in OCT compound, snap-frozen, and stored at −80°C. Hematoxylin and eosin immunostaining were performed according to standard protocols on 5 μm-thick cryostat sections. Morphometric analysis was performed on cryostat sections with NIH ImageJ software (National Institutes of Health, Bethesda, MD), as previously described ¹¹. The area within the external elastic lamina representing total vessel area was calculated by averaging measurements from serial sections at 200 μm intervals from 200 μm to 2000 μm below carotid bifurcation. For immunostaining, primary antibodies were anti-mouse αv integrin (Millipore), anti-smooth muscle α-actin (Sigma), anti-CD31 (BD Pharmingen, San Jose, CA), and F4/80 (Invitrogen, Carlsbad, CA). Isotype-matched antibodies were used as controls. Nuclei were detected with DAPI.

Quantitative reverse transcription polymerase chain reaction (RT-PCR)

Suitable tissue was available from 8 animals (including 6 who had undergone successful imaging) for analysis. Total RNA was isolated, reverse transcribed, and real time RT-PCR performed as described ⁹ using the following Taqman® primer sets (Applied Biosystems, Foster City, CA). CD68 (Mm00839636_g1), EMR1 (Mm00802529_m1), smooth muscle α-actin (Mm01546133-m1), CD31 (Mm00476702-m1) and αv (Mm00434506-m1), β3 (Mm00443980-m1), β5 (Mn00439825-m1), GAPDH (Mm99999915_g1). The results were normalized to Glyceraldehyde 3-phosphate dehydrogenase (GAPDH).

Statistical analysis

Statistical analysis was performed using GraphPad Prism (La Jolla, CA). Data are presented as mean ± standard error (SE). Differences between two groups were tested using two-tailed paired or unpaired Student’s t test, as appropriate. Association between any 2 variables was addressed using Pearson correlation. Significance was set at the 0.05 level.

Results

αvβ3 expression and activation in monocytes and macrophages

Monocyte-derived macrophages constitute a major component of inflammatory cells in the vessel wall. As a prelude to imaging studies, we assessed αvβ3 integrin expression on human monocytes, macrophages, and ECs. Integrin expression was readily detectable by flow cytometry on peripheral blood monocytes (Fig 1). NC100692 is a ^99mTc-labeled cyclic RGD peptide with specificity for αv integrins ^{2, 5}. To investigate NC100692 binding properties, we assessed the binding of a fluorescent homologue of NC100692 to ECs by flow cytometry. Low level of NC100692 homologue binding was detected in resting ECs. However, integrin activation with MnCl₂ (0.2 mM) considerably enhanced peptide binding to ECs (Supplemental Fig II). Similarly, integrin activation with MnCl₂ enhanced RGD peptide binding to monocytes (Fig 1a) without changing cell membrane αvβ3 expression, indicating that similar to resting ECs ¹², αv integrins on resting monocytes are in a non-fully activated state. Double color staining of monocytes using an anti-αvβ3 antibody and the fluorescent RGD peptide indicated that they both stain the same cells (Fig 1b). Interestingly, there were considerable differences in the extent of integrin activation in monocytes from different donors. To address the effect of monocyte differentiation into macrophages on αv expression and NC100692 binding, purified monocytes were differentiated into type I macrophages. Macrophage differentiation was confirmed by their distinct morphology. Similar to resting monocytes, macrophages expressed high levels of αvβ3 integrin and bound to NC100692. MnCl₂ enhanced fluorescent NC100692 binding to macrophages (Fig 1a) without altering the integrin expression level. Due to the time required for macrophage differentiation, monocytes and macrophages from the same donor were stained on different days. The changes in integrin expression and RGD peptide binding observed in the course of monocyte to macrophage differentiation were not consistent and varied from experiment to experiment.

Fig 1.

Flow cytometric assessment of αvβ3 expression and activation in monocytes and macrophages. a) Representative histograms of αvβ3 (top row) and RGD (bottom row) immunostaining of monocytes and monocyte-derived macrophages b) Representative contour plots demonstrating co-staining of monocytes with anti- αvβ3 antibody and RGD peptide. ab: antibody, ctrl: control, RGD: fluorescent homologue of NC100691.

Imaging αv integrin activation in vessel wall inflammation

To investigate αv-targeted imaging for detection of vessel wall inflammation in vivo, we used an established model of vascular inflammation. In this model, adventitial application of CaCl₂ to common carotid arteries of high fat fed apoE^−/− mice triggers an inflammatory response that leads to aneurismal dilatation of the artery over a period of 4 weeks ⁹. ApoE^−/− mice underwent ^99mTc-NC100692 microSPECT imaging at 2 or 4 weeks after surgery. NC100692 signal was readily visible on inflamed left, but not sham-operated right, carotid arteries identified by CT angiography at either time point after surgery (Fig 2). Quantitative analysis of NC100692 uptake from in vivo images showed significantly higher tracer uptake in the left, as compared to control right, carotid arteries (0.52 ± 0.09 vs 0.08 ± 0.02 cpv/MBq injected, n= 9, p<0.001 at 2 weeks, and 0.42 ± 0.04 vs 0.13 ± 0.02 cpv/MBq injected, n= 8, p<0.001 at 4 weeks, Fig 2b). As expected, considerable NC100692 uptake was also present in the surgical wound. Tracer uptake specificity in inflamed arteries was investigated in a group of animals at two weeks after surgery who were pretreated with 50-fold excess non-labeled precursor prior to ^99mTc-NC100692 administration. Left carotid (as well as surgical wound) uptake was significantly reduced following administration of excess unlabeled precursor (0.16 ± 0.03 cpv/MBq, n=3 vs 0.52 ± 0.09 cpv/MBq without blocking, n=9, p=0.004), establishing specificity of NC100692 uptake (Fig 3).

Fig 2.

MicroSPECT/CT imaging of αvβ3 activation in vascular inflammation. a) Examples of contrast-enhanced CT and NC100692 microSPECT-CT fused images of an apoE^−/− mouse 4 weeks after surgery to induce left common carotid artery vascular inflammation. Arrows point to common carotid arteries. Tracer uptake is also detected in the surgical wound (arrowheads). R: right, L: left, T: transverse, C: coronal, S: sagittal, cpv: counts per voxel. b) MicroSPECT-derived quantification of NC100692 uptake in remodeling left and sham-operated right common carotid artery at 2 (n=9) and 4 (n=8) weeks after surgery. *: p<0.001,

Fig 3.

NC100692 uptake specificity in vascular inflammation. MicroSPECT derived quantification of NC100692 signal in remodeling carotid artery in animals without (n=9) or with (n=3) injection with 50-fold excess unlabeled precursor prior to tracer administration. *: p=0.004, cpv: counts per voxel.

Ex vivo analysis of Integrin expression and inflammation

As expected, total vessel area at 4 weeks after surgery was ~2-folds higher in CaCl₂-treated left, compared to NaCl-treated right, carotid arteries (respectively, 0.192 ± 0.010 and 0.093 ± 0.007 mm², n=8, p<0.0001, Fig 4). Immunostaining with a macrophage-specific marker, F4/80, showed the presence of a large number of macrophages in the vessel wall in injured arteries (Fig 5). CD31 (EC) and smooth muscle α-actin (VSMC) staining demonstrated the presence of small blood vessels in the vessel wall (Supplemental Fig III). αv integrin immunostaining was detected in the intima and media of injured arteries, and its distribution resembled macrophage staining (Fig 5). Macrophage content of the vessel wall was quantified by real time RT-PCR which showed significantly higher levels of GAPDH-normalized CD68 and EMR1 mRNA expression in injured, as compared to sham-operated arteries (n=8, p=0.03 for CD68 and 0.003 for EMR1, Fig 6a). Interestingly, while there was no difference in CD31 mRNA expression between right and left carotid arteries, smooth muscle α-actin mRNA expression was significantly reduced in aneurismal arteries, indicating loss of VSMCs (n=8, p=0.045). There was no significant difference in GAPDH-normalized αv, β3, or β5 expression between control right and aneurismal left carotid arteries (Fig 6b).

Fig 4.

Inflammation-induced vascular remodeling in carotid arteries. a) examples of hematoxylin and eosin staining of CaCl₂-exposed left (L) and NaCl-exposed right (R) carotid arteries at 4 weeks after surgery demonstrating considerable remodeling of the left carotid artery. Scale bar: 100 μm, b) Morphometric analysis of total vessel area of common carotid arteries at 4 weeks after surgery, n=8, *: p<0.0001.

Fig 5.

Representative examples of macrophage (F4/80) and αv immunostaining (in red) of control right and remodeling left carotid arteries at 4 weeks after surgery. Nuclei are stained with DAPI in blue and elastic membrane autofluorescence is seen in green. L: lumen, Scale bar: 20 μm.

Fig 6.

Gene expression in carotid arteries at 4 weeks after surgery. a) GAPDH-normalized smooth muscle α-actin, CD31, CD68, and EMR-1 mRNA expression in control right and remodeling left carotid arteries detected by real time RT-PCR. n=8, *:p<0.05, **:p<0.01. b) GAPDH-normalized integrin mRNA expression in control right and remodeling left carotid arteries detected by real time RT-PCR demonstrating no significant difference. n=8.

Biological correlate of NC100692 uptake in carotid arteries

A number of cells in the vessel wall, including ECs, VSMCs and monocyte-derived macrophages, express αv integrins and may bind to NC100692 in vivo. The expression levels of cell-specific markers assessed by real time RT-PCR were used to define the biological correlates of NC100692 uptake in carotid arteries. NC100692 uptake significantly correlated with CD68 (r=0.67, p=0.02, Fig 7) expression, while there was no correlation between NC100692 uptake and GAPDH-normalized CD31, or smooth muscle α-actin expression (not shown).

Fig 7.

Vascular inflammation and integrin αvβ3 tracer uptake in carotid arteries. There is a significant correlation between CD68 expression and NC100692 uptake in the same animal. Pearson’s r=0.67, p=0.02.

Discussion

Integrins are a large family of heterodimeric adhesion molecules which mediate cell-cell and cell-matrix interactions ¹³. In vertebrates, 8 β subunits associate with 18 α subunits to generate 24 distinct integrins. Most integrins bind to ligands which contain an RGD tripeptide sequence. An important aspect of integrin biology is the role of conformational changes which modulate integrin function. Many integrins are expressed in a low affinity (off) state and upon activation, whether through outside-in or inside-out signals, convert to a high affinity (on) state which can bind specific ligands and trigger signaling ¹³. Integrin activation state is cell-dependent. For example, αvβ3 integrin is mostly in a low affinity state in JY lymphoblastoid cells, while in melanoma cell lines it is present in an active conformation ¹⁴. In resting ECs, αvβ3 is mostly in a low affinity state and EC activation, e.g., with shear stress, increases high affinity integrin ¹².

Integrin αvβ3 is expressed at high density on proliferating ECs and αvβ3-targeted imaging appears as a promising approach for detection of angiogenesis associated with tumors ⁴ and myocardial or hindlimb ischemia ^{1, 2}. Expression of αvβ3 integrins by other cells raises the possibility that this integrin may be targeted for imaging other processes where cell proliferation and integrin-mediated cell-cell and cell-matrix interactions are critically involved. Indeed, RP748, an ¹¹¹In-labeled αvβ3-targeted tracer localizes in murine or human arteries following mechanical or immune injury in parallel with changes in cell proliferation ^{7, 11}. In a previous study of targeted imaging of angiogenesis NC100692 appeared to localize in part in the inflammatory infiltrate in addition to ECs ². This, in conjunction with the reports on integrin expression in monocyte-macrophages ^{15, 16}, led us to investigate NC100692 for imaging vascular inflammation in vivo.

Several tracers have been developed for imaging αvβ3 expression, predominantly in angiogenesis ^17–19. In general, the binding motif in these probes is structured based on RGD tripeptide and they show broader specificity for αv integrins. We have previously shown that RP748, a peptidomimetic quinolone, preferentially interacts with the active conformation of the integrin ¹¹. Here, we showed a similar preferential binding to integrin active conformation for NC100692, a cyclic RGD peptide. In monocytes and macrophages Mn-induced integrin activation enhanced RGD peptide binding, indicating that similar to ECs, αv integrins in resting monocytes and macrophages are not in a fully activated state.

Inflammation plays a key role in the pathogenesis of several vasculopathies, including atherosclerosis and aneurysm. In atherosclerosis, vessel wall inflammation has been linked to plaque vulnerability and imaging vessel wall inflammation may help identify patients at high risk for acute coronary syndromes and stroke ²⁰. Similarly, vessel wall inflammation is linked to aortic aneurysm expansion and rupture and detection of vessel wall inflammation in vivo may help stratify patients based on their risk of rupture ²¹. As such, imaging vessel wall inflammation can potentially transform the clinical care of patients with atherosclerosis, aneurysm and other vascular pathologies. This is especially true for non-invasive imaging modalities, such as SPECT and PET, which are routinely used in humans. Potential applications of such imaging approaches include identification of patients at high risk for morbid events who may benefit from early treatment and tracking and optimizing therapeutic interventions. A number of tracers, predominantly those targeting endothelial adhesion molecules (vascular cell adhesion molecule-1 ²²), cellular metabolism (with ¹⁸F-fluorodeoxy glucose ^{23, 24}) and protease activity (matrix metalloproteinases ^{9, 25–28}, cathepsins ^{29, 30}) have been studied for their ability to track vessel wall inflammation in vivo. Recently, ex vivo studies have raised the possibility of αvβ3-targeted imaging of vessel wall inflammation. Using autoradiography, ¹⁸F-Galacto-RGD was shown to localize in atherosclerotic lesions and RGD uptake correlated with the density of nuclei and ³H-fluorodeoxy glucose uptake ³¹. Similarly, RGD-Cy5.5 localized in the arterial wall following carotid ligation in the mouse and the uptake was detectable by ex vivo near-infrared fluorescence reflectance imaging ³². Here, we demonstrated that NC100692 microSPECT/CT imaging can detect remodeling carotid arteries in a prototypic model of vascular inflammation in apoE^−/− mice in vivo. Blocking with excess unlabeled precursor confirmed the specificity of NC100692 signal. A similar protocol was used to image matrix metalloproteinase activation in vascular remodeling, where the approach to in vivo quantification of carotid signal was validated with ex vivo measures of tracer uptake ^{9, 10, 28}.

αvβ3 targeted paramagnetic nanoparticles have been used to image atherosclerotic plaque angiogenesis by magnetic resonance imaging ³. Unlike these nanoparticles, smaller probes such as NC100692 are not confined to intravascular space and any αvβ3 integrin expressing vascular cell may be target for NC100692 binding in vivo. In our experiments, the intense autofluorescence of elastic laminae did not permit direct co-localization of fluorescent RGD peptide with specific vascular cells. However, αvβ3 expressing proliferating ECs and inflammatory cells are both components of the inflammatory response in the vessel wall. Therefore, by targeting multiple cellular events linked to inflammation, our imaging approach may prove to be highly effective for detection of inflammation. Cellular content and target expression in the vessel wall is often measured by immunostaining. Because immunostaining is at best a semi-quantitative technique and only a limited number of histological sections (5–7 μm) are evaluated, such data may not reliably relate to imaging data obtained from much larger segments (~2mm) of the artery. Because the small size of murine carotid arteries prohibits the use of a more quantitative approach (e.g., Western blotting) for protein measurement, we relied on mRNA analysis to quantify macrophage content and integrin expression in carotid arteries. Using this approach, we found a strong correlation between NC100692 uptake in vivo and GAPDH-normalized CD68 (macrophage marker) expression in the vessel wall, validating αv-targeted imaging for detection of vessel wall inflammation in vivo. Importantly, despite the marked difference in NC100692 uptake, there was no significant difference in normalized αv, β3, or β5 integrin expression between control and remodeling carotid arteries. This, in conjunction with the preferential binding of NC100692 to active conformation of integrins, may indicate that αv integrins are in an active state in remodeling arteries.

In conclusion, we demonstrate that NC100692, a tracer with preferential binding to active conformation of αv integrins, specifically localizes in inflamed carotid arteries of apoE^−/− mice and provides a signal that is detectable by microSPECT/CT imaging in vivo. NC100692 uptake in the artery correlates well with macrophage content of the vessel wall, indicating that this RGD peptide may be used to image vascular inflammation in vivo. Further validation of our observations in other models of vessel wall inflammation may lead to the development of a novel imaging approach for identifying patients who are at high risk for vascular events and tracking the effect of anti-inflammatory treatments.