Leukocyte trafficking-associated vascular adhesion protein 1 is expressed and functionally active in atherosclerotic plaques

Given the important role of inflammation and the potential association of the leukocyte trafficking-associated adhesion molecule vascular adhesion protein 1 (VAP-1) with atherosclerosis, this study examined whether functional VAP-1 is expressed in atherosclerotic lesions and, if so, whether it could be targeted by positron emission tomography (PET). First, immunohistochemistry revealed that VAP-1 localized to endothelial cells of intra-plaque neovessels in human carotid endarterectomy samples from patients with recent ischemic symptoms. In low-density lipoprotein receptor-deficient mice expressing only apolipoprotein B100 (LDLR−/−ApoB100/100), VAP-1 was expressed on endothelial cells lining inflamed atherosclerotic lesions; normal vessel walls in aortas of C57BL/6N control mice were VAP-1-negative. Second, we discovered that the focal uptake of VAP-1 targeting sialic acid-binding immunoglobulin-like lectin 9 based PET tracer [68Ga]DOTA-Siglec-9 in atherosclerotic plaques was associated with the density of activated macrophages (r = 0.58, P = 0.022). As a final point, we found that the inhibition of VAP-1 activity with small molecule LJP1586 decreased the density of macrophages in inflamed atherosclerotic plaques in mice. Our results suggest for the first time VAP-1 as a potential imaging target for inflamed atherosclerotic plaques, and corroborate VAP-1 inhibition as a therapeutic approach in the treatment of atherosclerosis.

Functional activity of VAP-1 in atherosclerotic plaques. Treatment with small molecular VAP-1 inhibitor LJP1586 for 4 weeks reduced the amount of Mac-3-positive macrophages in atherosclerotic plaques by 46% when compared with the saline group (inhibitor, 8.2% ± 3.5% vs. saline, 15% ± 5.1%; P < 0.0001; Fig. 3B). However, the inhibitor had no effect on plaque size, as determined by the intima-media ratio (inhibitor, 1.9 ± 0.8 vs. saline, 1.5 ± 0.6; P = 0.15; Fig. 3C), body weight, or plasma lipid levels (see Supplementary material, Table S1). white arrows) lining the aortic plaques of low-density lipoprotein receptor-deficient mice expressing only apolipoprotein B100 (LDLR −/− ApoB 100/100 ) after intravenous injection of an anti-VAP-1 antibody, followed by immunohistochemical detection with a fluorescent secondary antibody. Normal vessel walls in the same sections (blue arrows) were VAP-1-negative, indicating the specificity of VAP-1 for the endothelium in atherosclerotic plaques. The autofluorescence (red color) in elastic fibers was observed even in native adjacent sections that were not stained. The adipocytes around the vessel walls showed moderate VAP-1 staining. AA = ascending aorta, A = aortic arch, LC = left common carotid artery. LS = left subclavian artery, DA = descending aorta, P = plaque, W = wall, Ad = adipocyte. (B) The endothelium in healthy C57BL/6N control mice was mainly VAP-1-negative, whereas adipocytes were VAP-1-positive. (C) Sections of human carotid artery were double-stained with a biotinylated Siglec-9 peptide and an anti-VAP-1 antibody. The endothelial cell of small capillaries inside atherosclerotic plaques in the area of intima were highly VAP-1 positive (white arrows) and co-localized with the biotinylated sialic acid-binding immunoglobulin-like lectin 9 (Siglec-9) motif containing peptide (red arrows), as demonstrated by in situ immunohistochemistry methods.

Discussion
Here, we provide evidence that VAP-1 plays a role in inflammation associated with atherosclerosis, suggesting that VAP-1 is a potential target for imaging and for anti-inflammatory therapeutics used to treat vascular inflammation. First, we showed that VAP-1 is expressed on the luminal surface of endothelial cells in atherosclerotic plaques in LDLR −/− ApoB 100/100 mice, and we confirmed a previous report showing expression of VAP-1 in human intra-plaque neovessels 19 . By contrast, endothelial cells lining non-atherosclerotic vessel walls were VAP-1-negative. Second, we discovered that focal uptake of VAP-1-targeting [ 68 Ga]DOTA-Siglec-9 in atherosclerotic plaques is associated with the density of infiltrating macrophages. Finally, we found that inhibiting VAP-1 activity with small molecule LJP1586 reduced the macrophage density in atherosclerotic plaques in mice. Several studies support the notion that adhesion molecules (e.g., P-and E-selectins, vascular cell adhesion molecule 1 (VCAM-1), and intercellular adhesion molecule 1 (ICAM-1)) play a role in the development of atherosclerosis; indeed, these molecules are expressed by endothelial cells lining atherosclerotic arteries in both humans and animals [20][21][22] . Recent studies also report VAP-1 expression in atherosclerotic lesions in humans and rabbits 6,7 . In contrast to many other leukocyte homing molecules, VAP-1 is practically absent from the surface of resting vascular endothelial cells; however, in an inflammatory environment, it is translocated rapidly from intracellular storage granules to the endothelial cell surface 1,2,17 , making it an attractive target for anti-inflammatory therapy and imaging.
The adhesive function of VAP-1 can be inhibited by certain enzymes 23,24 . Inhibiting VAP-1 reduces the migration of lymphocytes, granulocytes, and monocytes into inflamed tissues; it also reduces pathological angiogenesis in many animal models by abrogating monocyte/macrophage infiltration 25,26 . Here, we used a highly potent small molecular VAP-1 inhibitor, LJP1586 (mouse IC 50 4 nM) 27 , to demonstrate the functional importance of VAP-1 in atherosclerosis. The selectivity of LJP1586 for VAP-1 has been confirmed using a broad panel of receptors and enzymes, including monoamine oxidases A and B 27 . Marttila-Ichihara and co-workers showed that treatment of tumor-bearing mice with LJP1586 led to a significant reduction in the intratumoral accumulation of Gr-11CD11b1 myeloid cells and impaired tumor progression and neoangiogenesis 28 . Here, we found that treatment with LJP1586 led to a significant reduction in the density of macrophages in atherosclerotic plaques in mice, suggesting that VAP-1 is a potential target for therapeutics aimed at reducing inflammation associated with atherosclerosis. These results are in the line with previous mouse studies using a prototypic SSAO inhibitor, semicarbazide (also inhibiting lysyl oxidases and thus not being specific for VAP-1), as a treatment 29,30 . Mice with pre-existing, advanced atherosclerotic plaques were treated with VAP-1-targeted small molecule LJP1586 that probably explains that the short-term therapy at this stage had no effect on the size/amount of plaques. However, the finding that LJP1586 affected the extent of inflammation supports the finding of a previous study showing that inflammation is a more sensitive measure than the intima-to-media ratio 31 .
Another important finding is the high uptake of the VAP-1-targeting PET tracer, [ 68 Ga]DOTA-Siglec-9, and its association with the density of Mac-3-positive macrophages in atherosclerotic plaques. The specificity of [ 68 Ga]DOTA-Siglec-9 for VAP-1 was confirmed by a competitive binding assay in atherosclerotic mice, and further supported by the co-localization of Siglec-9 and an anti-VAP-1 antibody in human atherosclerotic plaques in situ. These observations are in the line with our previous studies, which verified that [ 68 Ga]DOTA-Siglec-9 is specific for VAP-1 using three different approaches: 1) cell-binding assays using Chinese Hamster Ovary (CHO) cells stably transfected with human VAP-1 and CHO cells transfected with vector only; 2) in vivo biodistribution analysis in human VAP-1 transgenic mice and VAP-1 knockout mice; and 3) in vivo competition with excess of unlabeled Siglec-9 peptide in a mouse model of melanoma expressing VAP-1 in the vasculature. The latter model also provided an additional specificity control because VAP-1 knockout mice were available for the studies 15 .
In vivo [ 68 Ga]DOTA-Siglec-9 PET/CT imaging showed a moderate signal in the aorta of atherosclerotic LDLR −/− ApoB 100/100 mice, particularly in the aortic root ( Fig. 2A). Humans express VAP-1 on the endothelial cells of intra-plaque neovessels; however, neovessels are rarely found in mice, which make in vivo imaging of atherosclerotic lesions in mice more challenging 32 . Thus, in vivo PET/CT imaging of human atherosclerosis-induced inflammation by targeting VAP-1 is probably achievable in humans.
Several non-invasive molecular imaging modalities have been used to detect different adhesion molecules involved in atherosclerotic inflammation [33][34][35][36] . To the best of our knowledge, only three papers report the atherosclerotic plaques in the aortic root with target-to-background ratio (SUV max, aortic arch /SUV mean, blood ) 2. imaging of atherosclerotic lesions using an adhesion molecule-targeting PET probe. Nahrendorf and co-workers report a [ 18 F]4V PET probe, which allows in vivo detection of VCAM-1 in the mouse aortic root, with an aortic root-to-blood ratio of approximately 2 at 4 h post-injection; however, this tracer has not been evaluated in clinical studies 37 . Nakamura and co-workers report a [ 64 Cu]DOTA-anti-P-selectin PET tracer, which allows in vivo detection of atherosclerotic mouse aortic root with a target-to-muscle ratio of 1.3 at 36 h post-i.v. injection 38 . However, the long physical half-life (12.7 h) of Cu 64 results in high exposure to ionizing radiation, thereby limiting its use in humans. Very recently, Bala and co-workers reported a lesion-to-blood ratio of 3.3 at 3 h post-injection (ex vivo gamma counting) of a novel 18 F-labeled anti-VCAM-1 nanobody 39 .
We previously used the same LDLR −/− ApoB 100/100 mouse model and study protocols reported herein to perform a study with [ 18 F]FDG 40 . Uptake of [ 68 Ga]DOTA-Siglec-9 by aortic plaques, normal vessel wall, and adventitia was similar to that of [ 18 F]FDG. Based on the autoradiography results, the plaque-to-normal vessel wall ratios were 2.1 ± 0.4 for [ 68 Ga]DOTA-Siglec-9 versus 2.3 ± 0.5 for [ 18 F]FDG (P = 0.44), respectively. According to the ex vivo biodistribution studies, uptake of [ 68 Ga]DOTA-Siglec-9 in the heart was significantly lower, but blood radioactivity was significantly higher, than that of [ 18 F]FDG. Based on the results presented herein, VAP-1 is a potential target for the imaging of atherosclerotic plaque inflammation. Although, in vivo detection of atherosclerotic plaques by [ 68 Ga]DOTA-Siglec-9 PET/CT was moderate in mice, the results are in line with those presented in many other reports showing autoradiographic evidence of tracer accumulation, but limited in vivo imaging signals. This raises the question of whether mouse models are appropriate for preclinical evaluation of new tracers for clinical use. Furthermore, it may be that [ 68 Ga]DOTA-Siglec-9 is not the optimal tracer for detecting VAP-1 in vivo, as a relatively high proportion of the injected radioactivity remains in the systemic circulation, at least in our LDLR −/− ApoB 100/100 mice. The 25 min accumulation time-point was chosen based on our previous study 15 , and time-activity curves in Fig. 2B confirm this. Prolonging the accumulation time-point to 90 min did not improve the aorta-to-blood ratio in our atherosclerotic mouse model; therefore, it was not investigated further. In general, the lack of a sufficiently strong imaging signal can be explained by the level of target expression, the tracer contrast between the blood and vessel wall, partial volume effects, and the affinity of the radioligand for the protein. Additionally, the positron energy of the radionuclide used for PET affects both visualization and quantification. 18  In addition to endothelial cells of intra-plaque neovessels in human carotid arteries, VAP-1 is present on the endothelial cells lining inflamed atherosclerotic plaques, but not those in normal vessel walls, in mice. The uptake of the novel VAP-1-targeting imaging tracer [ 68 Ga]DOTA-Siglec-9 by atherosclerotic plaques was associated with macrophage content. Finally, short-term treatment with LJP1586 reduced the macrophage density in atherosclerotic plaques in mice. Taken together, the results presented herein indicate that VAP-1 plays a role in inflammation associated with atherosclerosis and that it might be a potential target for imaging of inflamed atherosclerotic plaques and for anti-inflammatory therapeutics designed to treat vascular inflammation. As inhibition of VAP-1 reduces development of atherosclerotic lesions, one could envision that inhibition of inflammation may start the healing process and thus, reverse atherosclerosis. The second phase examined the role of VAP-1 in atherosclerotic plaques. Twenty-five LDLR −/− ApoB 100/100 mice were treated with a VAP-1 inhibitor (LJP1586, Z-3-fluoro-2-(4-methoxybenzyl)allylamine hydrochloride; a kind gift from M. Linnik, La Jolla Pharmaceuticals, San Diego, CA, USA), or with saline. At the age of 2 months, mice were shifted to a high-fat diet for 2 months prior to the start of 4-week treatment, during which they were on normal chow. All mice were housed in an animal facility under standard conditions (12 h light/dark cycle) and allowed access to food and water ad libitum.
All animal experiments were approved by the National Animal Experiment Board in Finland and the Regional State Administrative Agency for Southern Finland, and carried out in accordance with the relevant European Union directives.
Human carotid artery samples. Frozen human carotid endarterectomy samples (all containing atherosclerotic plaques) were obtained from patients (four female and one male; mean age 41 ± 9 years) with recent ischemic symptoms and double-stained with an anti-human VAP-1 antibody (10 μ g/mL; JG2.10 gift from E. Butcher, Stanford University, CA, USA) and a biotinylated Siglec-9 peptide (20 μ g/mL; NeoMPS, Strasbourg, France). The patient study was conducted in accordance with the Declaration of Helsinki, and the study protocol was approved by the Ethics Committee of Hospital District of Southwest Finland. All patients provided written informed consent.
Detection of VAP-1 in atherosclerotic plaques. VAP-1 expression in the aorta of high-fat fed LDLR −/− ApoB 100/100 (n = 2) and C57BL/6N control mice (n = 2), and in carotid endarterectomy samples from patients with recent ischemic symptoms (n = 5), was examined by immunohistochemistry. To detect only luminal expression of VAP-1, mice were intravenously (i.v.) injected with a monoclonal rat anti-mouse VAP-1 antibody (7-88, 1 mg/kg diluted in saline) 2 10 min before sacrifice. Aorta samples were frozen and cut into 8 μ m longitudinal sections, incubated for 30 min at room temperature in the dark with a secondary goat anti-rat antibody (working dilution, 5 μ g/mL in phosphate-buffered saline (PBS) containing 5% normal mouse or human AB serum) conjugated to a fluorescent dye (Alexa Fluor 488; Invitrogen, Eugene, OR, USA), and rinsed twice in PBS for 5 min.
Co-localization of VAP-1 and the Siglec-9 motif peptide in atherosclerotic plaques was investigated in human carotid samples. Frozen endarterectomy sections (5 μ m) were first incubated for 30 min with biotinylated Siglec-9 peptide (20 μ g/mL; NeoMPS, Strasbourg, France) in Dulbecco's PBS containing magnesium and calcium, and then detected with streptavidin-phycoerythrin. After three washes in PBS, the same sections were incubated with an anti-human VAP-1 antibody (JG2.10, 10 μ g/mL) for 30 min, followed by detection with a fluorescein isothiocyanate-conjugated secondary immunoglobulin G antibody. Sections incubated in the absence of the Siglec-9 peptide served as a negative control. Tissue autofluorescence was distinguished from specific staining using the lambda scan mode and subsequent linear unmixing based on reference spectra. All fluorescent images were captured using a Zeiss LSM780 confocal microscope (Carl Zeiss MicroImaging GmbH, Jena, Germany).
Autoradiography and support staining. The distribution of [ 68 Ga]DOTA-Siglec-9 in the aorta was examined in more detail using autoradiography, as described previously 42 . Briefly, after gamma counting, excised aortas were frozen in isopentane, and sequential longitudinal cryosections (20 and 8 μ m) were cut using a cryomicrotome at − 15 °C. Sections were then thaw-mounted onto microscope slides. Next, the cryosections were apposed to an imaging plate (Fuji Imaging Plate BAS-TR2025; Fuji Photo Film Co., Ltd., Tokyo, Japan) and scanned after an exposure time of 2.5 h (Fuji Analyzer BAS-5000; Fuji, Tokyo, Japan; internal resolution, 25 μ m).
The 20 μ m sections were stained with hematoxylin and eosin and scanned with a Pannoramic 250 Flash slide scanner (3DHISTECH Ltd.), and morphology was examined using Pannoramic Viewer 1.15 software (3DHISTECH Ltd.). After carefully superimposing the autoradiographs and hematoxylin-eosin staining images, the concentration of 68 Ga-radioactivity was measured in the following ROI: 1) plaques (excluding media); 2) normal vessel wall (no lesion formation); and 3) adventitia (mainly adipose tissue around the aorta). The results were expressed as count densities (photostimulated luminescence per square millimeter; PSL/mm 2 ) using Tina 2.1 software (Raytest Isotopemessgeräte GmbH, Straubenhardt, Germany). The count density for background radiation was subtracted from the actual ROI data, and the results for each mouse were decay-corrected for injection time and exposure time and normalized for injected radioactivity dose. In total, 1142 ROIs (417 in plaques, 478 in normal vessel walls, and 347 in adventitia) were analyzed in 27 atherosclerotic and 20 control mice.
The 8 μ m aorta cryosections were immunohistochemically stained with a rat anti-mouse Mac-3 antibody (Clone M3/84, dilution 1:1000, BD Biosciences, Franklin Lakes, NJ, USA) as described previously 43 to compare the uptake of [ 68 Ga]DOTA-Siglec-9 and the macrophage density in aortic plaques. Sections were scanned with a digital slide scanner (Pannoramic 250, 3DHISTECH Ltd.). The media was outlined and the Mac-3-positive area in each plaque was calculated (and expressed as %) using an automatic color deconvolution method and ImageJ Scientific RepoRts | 6:35089 | DOI: 10.1038/srep35089 (v. 1.46) software (Fiji, National Institutes of Health, Bethesda, MD, USA). Uptake of 68 Ga-radioactivity in the same areas was evaluated using digital autoradiography.
Assessment of functional activity of VAP-1 in atherosclerotic plaques. During the 4 week treatment period, LDLR −/− ApoB 100/100 mice received thrice-weekly intraperitoneal (i.p.) injections of a small molecular VAP-1 inhibitor, LJP1586 (Z-3-fluoro-2-(4-methoxybenzyl)allylamine hydrochloride; 0.5 mg/mL diluted in PBS; a kind gift from M. Linnik, La Jolla Pharmaceuticals, San Diego, CA, USA) at a dose of 5 mg/kg. Mice injected with the physiological saline alone (200 μ L) served as negative controls. After treatment, all mice were sacrificed and the hearts were collected, formalin-fixed, and paraffin-embedded. Tissues were then cut transversely into 5 μ m sections at the level of the coronary ostia. The areas of intima and media were outlined in sections stained with Movat's pentachrome, and the intima-to-media ratio for each mouse was determined with ImageJ as described previously 44 . Adjacent sections were stained with an anti-mouse Mac-3 antibody to detect activated macrophages in the intima as described above. Sections of aortic ostium were used to determine the area of plaque occupied by macrophages in each mouse because the ostium is easily located due to the presence of valves; it is therefore the most reliable place to compare plaque deposition in individual mice.

Statistical analyses.
All results are expressed as the mean ± SD with two significant figures. Non-paired data were compared between two groups using a t test and between multiple groups using ANOVA with Tukey's correction. A paired t test or Pearson's correlation analysis was used to compare paired data between two groups. A P value < 0.05 was considered statistically significant.