Use of Cyclic Backbone NGR-Based SPECT to Increase Efficacy of Postmyocardial Infarction Angiogenesis Imaging

As CD13 is selectively expressed in angiogenesis, it can serve as a target for molecular imaging tracers to noninvasively visualize angiogenic processes in vivo. The CD13-targeting moiety NGR was synthesized and cyclized by native chemical ligation (NCL) instead of disulfide bridging, leading to a cyclic peptide backbone: cyclo(Cys-Asn-Gly-Arg-Gly) (coNGR). Beside this new monomeric coNGR, a tetrameric NGR peptide co(NGR)4 was designed and synthesized. After radiolabeling, their in vitro and in vivo characteristics were determined. Both coNGR-based imaging agents displayed considerably higher standardized uptake values (SUVs) at infarcted areas compared to the previously reported disulfide-cyclized cNGR imaging agent. Uptake patterns of 111In-coNGR and 111In-co(NGR)4 coincided with CD13 immunohistochemistry on excised hearts. Blood stability tests indicated better stability for both novel imaging agents after 50 min blood incubation compared to the disulfide-cyclized cNGR imaging agent. In mice, both coNGR peptides cleared rapidly from the blood mainly via the kidneys. In addition, co(NGR)4 showed a significantly higher specific uptake in infarcted myocardium compared to coNGR and thus is a promising sensitive imaging agent for detection of angiogenesis in infarcted myocardium.


Introduction
Angiogenesis is an endogenous healing process which serves to restore tissue blood supply in response to ischemic injury [1]. The extent of angiogenic activity is correlated with infarct healing and postmyocardial infarction (MI) remodeling [2]. Augmentation of experimental angiogenesis has several beneficial effects on post-MI remodeling, including reduced apoptosis of hypertrophic cardiomyocytes in the border zone, attenuated collagen deposition and scar formation in the noninfarcted zone, and improved long-term ventricular function [2,3]. However, while positive results were shown in animal models of MI, the benefit for MI patients has yet to be shown. So far, results from double-blind placebocontroled trials were disappointing [4][5][6][7]. These negative outcomes were most likely attributable to a combination of factors, including patient selection, choice of delivery platforms for therapeutic agents, and, importantly, lack of sensitive noninvasive detection methods for angiogenesis [8].
Frequently employed imaging agents for noninvasive nuclear imaging of cardiovascular angiogenesis in animal models are based either on the v 3 integrin-targeting  Arg-Gly-Asp (RGD) amino acid sequence or on vascular endothelial growth factor (VEGF) [9]. Several single-photon emission computed tomography (SPECT) and positron emission tomography (PET) based studies reported positive results with RGD targeting agents (reviewed in [10,11]). However, in competition studies, the RGD motif was found to have a lower target homing ratio (target to control tissue) compared to the Asn-Gly-Arg (NGR) tripeptide sequence which binds to selectively upregulated CD13 on angiogenically active endothelial cells [12]. Given the high target homing ratio of cyclic NGR peptides, employing a cyclic NGRbased imaging agent might result in better image quality. The CD13-targeting NGR motif has been explored at our institute as a molecular angiogenesis imaging agent for fluorescence microscopy [13], magnetic resonance imaging (MRI) [14], and more recently single-photon emission computed tomography (SPECT) imaging [15]. These cyclic NGRbased molecular imaging agents showed enhanced uptake in the infarct area and infarct border zone 7 days after MI in a mouse model [13][14][15]. Although each study established specific binding of their respective NGR-based ligand in the infarct area and infarct border zone, none of these molecular imaging agents had the potential for clinical translation at this stage. While localizing the fluorescently labeled NGR-based molecular imaging probe would require an invasive procedure, the cadmium-selenium core of cyclic NGR-conjugated paramagnetic quantum dots (pQD) in combination with their accumulation in liver and spleen hampers translation of the MRI angiogenesis imaging agent into clinical trials. For the NGR-based SPECT tracer to be considered as a potential clinical angiogenesis imaging tool, specific target uptake has to increase.

Stability Tests. 111
In-cNGR, 111 In-coNGR, or 111 Inco(NGR) 4 was added to 1 × 10 mL human blood in heparin (BD Biosciences, Vianen, The Netherlands). Blood stability was tested at 0, 10, 30, and 50 min. For each time point imaging agents were separated from blood cells and proteins by adding 0.5 mL MeCN (VWR International BV, Breda, The Netherlands) to 0.5 mL blood sample followed by centrifugation (2,000 ×g, 3 min). Supernatant samples were analyzed by HPLC using the above-mentioned method.

Octanol/Water Partition Coefficient.
The log value of each compound was determined in three separate experiments as described previously [22].

Animal Studies.
In 10-12-week-old male Swiss mice, we induced MI by ligation of the left anterior descending coronary artery (LAD) as described before [23] to test coNGR ( = 5) or co(NGR) 4 ( = 8). A group of sham-operated Swiss mice ( = 5) was used as control. SPECT imaging was performed 7 days after MI or sham surgery. An overview of the number of animals used per tracer is given in Table S1. All animals were held under the guidelines of the animal care facility (Maastricht University). All animal experiments were approved by the Committee for Animal Welfare of Maastricht University.

Micro-SPECT.
Mice were anesthetized with isoflurane (induction 2.5%; maintenance 1.5%), a catheter was placed in a tail vein, and animals were positioned in the SPECT camera (MILabs, Utrecht, The Netherlands). Prior to image acquisition, a bolus injection of 99m Tc-sestamibi and 111 In-coNGR or 111 In-co(NGR) 4 was given intravenously (i.v.) in a maximum volume of 200 L. Table S1 in Supplementary Material available online at https://doi.org/10.1155/2017/8638549 provides an overview of the average injected dose. Immediately after injection, 4 consecutive time frames of 15 min each were acquired. For image quantification we used the last of these frames.

SPECT Image
Reconstruction. Acquired list mode data was reconstructed using MILabs reconstruction software (version 2.51) employing the POS-EM algorithm (6 iterations and 16 subsets, reconstructed at a voxel size of 0.4 mm). 99m Tc and 111 In images were reconstructed by selecting photo peak and background energy levels as described before [15].

Image Quantification.
To allow quantification of imaging agent uptake, in vivo isotope-specific conversion factors (CF) were determined for the 0.6 mm collimator in a representative phantom with a known activity. Using the previously described method [15] the following conversion factors (CF), CF 99mTc 612 MBq/mL and CF 111In 643 MBq/mL, were found. PMOD 3.7 cardiac tool PCARD (PMOD technologies, Zürich, Switzerland) was used to segment the heart in the 17-segment model. Uptake per segment was subsequently expressed as a mean standardized uptake value (SUV mean ), also using the previously described method [15]. For body weight, we assumed that 1 g equalled 1 mL.

Biodistribution.
After imaging, the vital organs were harvested and kept for gamma-counting (Wallac Wizard, Turku, Finland). Acquired data were expressed as percentage injected dose per gram tissue (% ID/g).

CD13 Immunohistological Staining.
Hearts were dissected and fixated in HEPES-buffered formaldehyde containing 150 mM saline for 24 h at 4 ∘ C. Hereafter, hearts Contrast Media & Molecular Imaging 5 were placed in 70% ethanol for maximal one month before embedding in paraffin. Paraffin-embedded hearts were cut at 4 m thickness. Sections were deparaffinized and rehydrated after which endogenous peroxidase was blocked by 0.3% H 2 O 2 in methanol. After antigen retrieval (DAKO, target retrieval solution) sections were blocked in 5% goat serum for one hour. Next, sections were incubated overnight with primary monoclonal antibody against CD13 (1 : 500, Sigma). After washing, sections were incubated with BrightVision poly-horseradish peroxidase (HRP) Goat anti-Rabbit antibody (Immunologic, Duiven, The Netherlands). HRP was visualized by NOVAred substrate kit (VECTOR Laboratories Inc., Burlingame, CA, USA) and sections were counterstained with hematoxyline. Finally, sections were covered by cover glass with entellan and visualized using Leica Application Suite X (Leica Microsystems, Wetzlar, Germany).

Statistics.
All data were expressed as mean ± SEM. To test for significant differences we performed an unpaired student's -test with < 0.05 being considered statistically significant. Data were analyzed using Microsoft Excel (version 2010).

Results and Discussion
In this study, an NGR peptide cyclized via native chemical ligation (NCL) and its tetravalent analog were designed and synthesized. The feasibility of these two new NGR peptidebased ligands for radionuclide imaging of CD13 expression in a mouse MI model with SPECT was explored. The peptides were radiolabeled with 111 In and used for dualisotope SPECT with 99m Tc-sestamibi as myocardial perfusion agent. Uptake of 99m Tc-sestamibi and 111 In-coNGR and 111 In-co(NGR) 4 was quantified in the 17-segment model.  Figure S1. Radiochemical purity generally exceeded 95% for coNGR, co(NGR) 4 , and cNGR and did not require further purification. From stability studies it appeared that disulfide bond-cyclized cNGR was less stable after 50 minutes of incubation in blood (7.5% intact product, Table 2) compared to both coNGR-based tracers. The percentage of intact tetrameric coNGR (25.3%) is slightly higher compared to coNGR (20.2%).
Octanol-water partition coefficients indicated that coNGR was more hydrophilic than co(NGR) 4 (Table 3). Remarkably, the reference compound cNGR appeared initially to be the most lipophilic compound (−3.30 ± 0.14). However, over time its log value decreased soon to −4.56 ± 0.05 (<60 min), most likely explained by instability of this compound. Apparently, backbone cyclization via NCL resulted in a more stable ring structure than cyclization via disulfide bond formation. 4 Uptake. The 17-segment model was used to determine standardized   4 25.31 cNGR 7.55 Table 3: log values of 111 In-labeled coNGR, co(NGR) 4 , and cNGR [15]. log values were determined in three separate experiments, hence indicated as log 1, log 2, and 3. = 3 for each measurement. uptake values (SUVs) of the tracers [24]. 99m Tc-sestamibi allowed visualization of infarcted myocardium 7 days after LAD ligation. Infarcted areas were defined as areas with significantly reduced uptake of 99m Tc-sestamibi in either coNGR or co(NGR) 4 injected mice with LAD ligation compared to sham-operated mice (Table S2). Highest uptake of 111 In-coNGR and 111 In-co(NGR) 4 was observed in the area in and around the apex, which corresponds to the infarct area (Figures 3 and 4 and Table S3). Uptake patterns of coNGR and co(NGR) 4 can be seen in Figures 5 and 6, respectively. To examine whether multimerization indeed resulted in higher target uptake, SUVs of co(NGR) 4 and coNGR in infarcted and noninfarcted areas were compared for each tracer separately. Uptake of co(NGR) 4 in segments that were affected by MI was significantly higher (0.57 ± 0.03) than in segments that were not affected by MI (0.43 ± 0.02), whereas uptake of coNGR in MI-affected segments (0.83 ± 0.05) was not significantly higher than in unaffected segments (0.76 ±  Figure 5: Representative in vivo fusion images of 99m Tc-sestamibi and coNGR uptake in MI mice. In "grey" is the uptake of 99m Tc-sestamibi in myocardium, while uptake of coNGR is in "hot metal" color. In (a) and (b) slices of an infarcted heart are shown. The infarct area is visualized by the decreased uptake of 99m Tc-sestamibi (in grey) which is clearly visible in the anterolateral region of the heart (arrowheads). Enhanced uptake of coNGR is clearly visible in the infarct area (arrows). Note: the uptake is color coded and relative to the injected dose per animal. SA: short axis; VLA: vertical long axis. 0.05). This suggests that co(NGR) 4 has a more specific uptake in infarcted areas than coNGR.

SPECT and Analysis of coNGR and co(NGR)
Although coNGR and co(NGR) 4 were not directly compared with cNGR in an in vivo study, both coNGR-based ligands showed a higher uptake than cNGR in the infarcted area. However, only co(NGR) 4 indicated a more targetspecific uptake which makes co(NGR) 4 a more optimal agent than cNGR or coNGR for imaging of angiogenesis after MI. 4 . One hour postinjection (p.i.), urinary excretion in MI mice was 72.5 ± 5.2 for coNGR and 54.9 ± 9.1 for co(NGR) 4 (% ID/g ± SEM, = 0.23). Retention of coNGR appeared to be significantly higher in blood, muscle, lungs, and intestines compared to retention of co(NGR) 4 . It could be that co(NGR) 4 clears a bit more via the hepotobiliary route than coNGR, though liver uptake of co(NGR) 4 and coNGR was not statiscally different.  Figure 6: Representative in vivo fusion images of 99m Tc-sestamibi and co(NGR) 4 uptake. In "grey" is the uptake of 99m Tc-sestamibi in myocardium, while uptake of co(NGR) 4 imaging is in "hot metal" color. (a) and (c): infarcted heart. Infarct area is signified by decreased uptake of 99m Tc-sestamibi which is clearly visible in the anterolateral region of the heart (arrowheads). Enhanced uptake of co(NGR) 4 is clearly visibly in the infarct area (arrows). (b) and (d): sham-operated heart. Uniform uptake of 99m Tc-sestamibi can be seen in combination with low overall uptake of co(NGR) 4 . Note: uptake is color coded and relative to injected dose per animal. SA: short axis; VLA: vertical long axis.

Biodistribution of coNGR and co(NGR)
Other organs displayed a low uptake of coNGR that did not differ from the uptake of co(NGR) 4 (Figure 7).
Unlike with cyclic RGD, cyclic NGR multimerization did not result in higher target uptake as coNGR exceeded co(NGR) 4 uptake. Changing the currently used short SMCC spacer, which only allows statistical rebinding, for a longer and less rigid spacer that might enable binding to multiple CD13 receptors simultaneously, could improve the avidity and thereby the affinity of the tracer. For example, a flexible PEG spacer with a length that could bridge the width of the CD13 receptor of 131Å [25] might be suitable.

CD13 Immunohistochemistry.
To validate uptake patterns of both coNGR-based tracers, cardiac CD13 expression was evaluated through immunohistological staining. A low level of CD13 expression was observed within the myocardium of sham-operated control animals (Figures 8(a) and 8(d)). * * *  4 in MI mice and co(NGR) 4 in sham-operated mice. Substantial kidney uptake of coNGR and co(NGR) 4 was observed. coNGR had significantly higher uptake than co(NGR) 4 in blood, lungs, and intestines whereas uptake in other organs was similar. * < 0.05 is considered statistically significant. Note that expression of CD13 is not restricted to blood vessels but that endothelium of blood vessels is positive in the infarcted heart as opposed to the sham (arrows).
While the level of CD13 expression in the noninfarcted myocardium of MI animals appeared marginally higher compared to the expression level in control animals (Figures 8(c) and 8(f)), the level of CD13 expression was markedly higher in the infarcted myocardium of MI animals ( Figures  8(b) and 8(e)). The enhanced uptake of coNGR and co(NGR) 4 in the healthy myocardium of infarcted hearts points towards an overall angiogenic response of the heart in MI animals. Additionally, for co(NGR) 4 the highest level of uptake was found in and around the infarcted areas of MI animals which also correlated to the histological findings as the highest level of CD13 expression was found in and around the infarcted areas. The same trend was observed for coNGR. It is highly likely that the uptake in the infarcted area as well as in the healthy myocardium is specific and related to increased CD13 expression.

Conclusions
Two CD13-targeting SPECT tracers for angiogenesis, coNGR and co(NGR) 4 , were designed and synthesized. However, target uptake of cyclic NGR-based imaging agents does not seem to increase with increasing valency. Instead, the key to enhance cyclic NGR-based imaging agent uptake was to stabilize the ring structure through NCL. Additional studies with different linkers conjugated to the lysine scaffold are warranted to investigate a possible increased avidity effect with coNGR-based imaging agents.