Tropoelastin: A novel marker for plaque progression and instability

Abstract

Background

Elastolysis and ineffective elastogenesis favor the accumulation of tropoelastin, rather than cross-linked elastin, in atherosclerotic plaques. We developed gadolinium-labeled tropoelastin-specific magnetic resonance contrast agents (Gd-TESMAs) for tropoelastin imaging in animal models.

Methods and Results

Two peptides, VVGSPSAQDEASPLS and YPDHVQYTHY were selected to target tropoelastin. In vitro binding, relaxivity, and biodistribution experiments enabled characterization of the probes and selecting the best candidate for in vivo MRI. MRI was performed in atherosclerotic apolipoprotein E-deficient (ApoE^-/-) mice and New Zealand white rabbits with stable and rupture-prone plaques using Gd-TESMA. Additionally, human carotid endarterectomy specimens were imaged ex vivo. The VVGSPSAQDEASPLS-based probe discriminated between tropoelastin and cross-linked elastin (64±7% vs 1±2%, P=0.001), had high in vitro relaxivity in solution (r_1-free=11.7±0.6mM⁻¹s⁻¹, r_{1-bound to tropoelastin} = 44±1mM^-1s^-1) and favorable pharmacokinetics. In vivo mice vascular enhancement (4wks=0.13±0.007mm², 8wks=0.22±0.01mm², 12wks=0.33±0.01mm², P<0.001) and R₁ relaxation rate (4wks=0.90±0.01 s^-1, 8wks=1.40±0.03 s^-1, 12wks=1.87±0.04s^-1, P<0.001) increased with atherosclerosis progression after Gd-TESMA injection. Conversely, statin-treated (0.13±0.01mm², R₁ =1.37±0.03s^-1) and control (0.10±0.005mm², R₁ =0.87±0.05s^-1) mice showed less enhancement. Rupture-prone rabbit plaques had higher R₁ relaxation rate compared with stale plaques (R₁=2.26±0.1s^-1 vs R₁=1.43±0.02s^-1, P=0.001), after administration of Gd-TESMA that allowed detection of rupture-prone plaques with high sensitivity (84.4%) and specificity (92.3%). Increased vascular R₁ relaxation rate was observed in carotid endarterectomy plaques after soaking (R_1pre= 1.1±0.26 s^-1 vs R_1post= 3.0±0.1s^-1, P=0.01). Ex vivo analyses confirmed the MRI findings and showed uptake of the contrast agent to be specific for tropoelastin.

Conclusions

MRI of tropoelastin provides a novel biomarker for atherosclerotic plaque progression and instability.

Introduction

Cardiovascular disease (CVD) accounts for nearly one-third of sudden and premature deaths worldwide¹. Despite improvements in the management of patients, CVD related deaths are projected to affect 12 million people by 2030². Atherosclerosis is the underlying cause of the majority of cardiovascular events. Atherosclerotic plaque progression involves the dynamic turnover of the extracellular matrix (ECM) protein, elastin, through elastolysis^3,4 (degradation) and elastogenesis (de novo synthesis)^5–8. Elastolysis^3,4 induced by inflammatory processes that upregulate the expression of elastases⁹ has been considered the major pathway for elastin remodeling. Conversely, dysfunctional elastogenesis and its contribution to plaque progression and instability, is less understood^5–8.

Elastin is a highly abundant ECM protein found in the arterial wall that contributes to 50% of its dry weight¹⁰. Mature elastin is an insoluble and hydrophobic polymer, with a very low turnover rate. Mature elastin is formed by cross-linking of tropoelastin, a 60-72 kDa soluble monomer ¹¹, by lysyl oxidase (LOX). Mature elastin contains several tropoelastin molecules. Tropoelastin has two alternating domains: (1) hydrophilic, cross-linked, domains rich in Lys and Ala and (2) hydrophobic domains (responsible for elasticity) rich in Val, Pro, and Gly, which often occur in repeats of VPGVG or VGGVG¹¹. Endothelial cells, vascular smooth muscle cells (VSMCs) and adventitial fibroblasts produce tropoelastin during the late fetal and early neonatal periods of life, after which elastogenesis ceases¹¹. Secreted tropoelastin is “chaperoned” to the extracellular space by the elastin-binding protein (EBP)^12,13 where tropoelastin is stabilized and aligned along microfibrils, that contain glycoproteins (e.g. fibrillins) and microfibril-associated glycoproteins (e.g., MAGP-1) prior to enzymatic cross-linking^14,15. Thus, normal arteries contain cross-linked elastin but negligible amounts of tropoelastin.

Dysfunctional ECM synthesis and degradation contribute to the initiation, progression and complication of arterial diseases including atherosclerosis^6,16,17, abdominal aortic aneurysm (AAA)¹⁸, supravalvular aortic stenosis¹⁹, in-stent restenosis^20,21 and Marfan’s syndrome²². In atherosclerosis, VSMCs^23,24 and macrophages^6,8 resume production of tropoelastin that often fails to cross-link into elastic fibers as a result of the reduced expression or absence of LOX^25–30, or any of the components of the microfibrillar scaffold required for fiber assembly^31–33. Simultaneously, elastin degradation by elastases⁹ generates elastin-derived peptides, including the VGVAPG, that are chemotactic for inflammatory cells³⁴, participate in lipoprotein retention, and upregulate matrix metalloproteinases¹⁰, all of which are involved in the progression and instability of a plaque.

Over the past 10 years there have been significant developments in imaging of molecular components of the vessel wall in vivo. Imaging of vascular elastin has been achieved using the elastin-specific MRI contrast agent (ESMA)³⁵ in preclinical models of atherosclerosis^36–38, AAA³⁹, coronary artery injury⁴⁰ and myocardial remodeling after infarction^41,42. Collagen-binding paramagnetic nanoparticles have also successfully been used to image collagen remodeling in murine atherosclerosis⁴³; while in vivo optical analysis has been used to quantify changes in collagen and elastin after wall injury⁴⁴.

As elastogenesis and elastolysis favor the accumulation of tropoelastin, rather than fully processed cross-linked elastin, we hypothesized that tropoelastin may serve as a new imaging biomarker to detect atherosclerosis progression and lesion instability. Herein, we have developed, characterized and validated novel tropoelastin-specific MR contrast agents to image tropoelastin turnover in models of atherosclerosis and plaque instability and shown potential to bind to symptomatic plaques taken from patients undergoing carotid endarterectomy.

Methods

The data, analytic methods, and study materials will not be made available to other researchers for purposes of reproducing the results or replicating the procedure.

Elastin and tropoelastin-binding probes

The tropoelastin peptides were selected as described in the online Supplementary Methods. The probes, (DOTA)-VVGSPSAQDEASPLS and K(DOTA)YPDHVQYTHY, their corresponding scrambled versions DOTA-GAESAPLVSSVQSPD (scrambled-V) and K(DOTA)-HQVYTYPHDY (scrambled-Y), as well as the rhodamine-labeled VVGSPSAQDEASPLS peptide derivative were custom synthesized by Peptide Synthetics Ltd. (Hampshire, UK). The DOTA-peptides were complexed in house with lanthanide ions: europium (Eu³⁺), lanthanum (La³⁺) or gadolinium (Gd³⁺) for in vitro binding assays, relaxivity and MRI experiments; or ⁶⁴Cu²⁺ for biodistribution studies. The complexed (DOTA)-VVGSPSAQDEASPLS and K(DOTA)YPDHVQYTHY probes were denominated as tropoelastin-binding MR contrast-agents (TESMAs). The diethylenetriaminepentaacetic acid (DTPA)-based elastin-specific contrast agent was provided by Lantheus Medical Imaging (North Billerica, USA)³⁵, either complexed to gadolinium, (Gd-ESMA; 856 Da) for relaxivity and MRI experiments or in its free form (ESMA), complexed in house with Eu³⁺ for in vitro binding assays. Eu³⁺ labeled DTPA solutions (Eu-DTPA) complexed in house were used as controls.

Synthesis of the lanthanide(III) complexes, in vitro binding and relaxivity studies were performed as described in the online Supplementary Methods.

Murine model of atherosclerosis progression

Male apolipoprotein E knockout (ApoE^-/-) mice and wild-type (WT) C57BL/6J were purchased from Charles Rivers Laboratories (Edinburgh, UK). 8 week-old ApoE^-/- mice were switched to a high-fat diet (HFD) containing 21% fat from lard, and 0.15% (wt/wt) cholesterol (Special Diets Services, Witham, UK). ApoE^-/- mice were imaged at 4, 8, and 12 weeks after HFD-feeding (n=10 per group). In the treatment group, ApoE^-/- mice (n=10) received pravastatin (40mg/kg/day) (Kemprotec Ltd, Middlesborough, UK) administered in the drinking water simultaneously with the HFD for 12 weeks. 8 week-old WT mice (n=10) were fed a normal chow diet for 12 weeks and scanned at 20-weeks.

Rabbit model of plaque instability

Six 3 month-old male New Zealand White rabbits (Harlan, Wyton, England; mean weight, 2.5 kg) were fed a 1% cholesterol diet (Special Diet Services, Witham, England) for 2 weeks before and 6 weeks after balloon injury of the abdominal aorta. This was followed by 4 weeks of normal chow diet as previously described^37,45,46. Diseased rabbits received intraperitoneal administration of Russell’s viper venom (0.15 mg/kg; Enzyme Research Laboratories, Swansea, UK), a procoagulant factor, followed 30 minutes later by intravenous administration of histamine dihydrochloride (0.02 mg/kg; Sigma-Aldrich, Dorset, England), a vasoconstrictor in rabbits, to induce plaque disruption and thrombosis. This procedure was repeated twice within 4 hours. All procedures were approved by the United Kingdom Home Office and were in accordance with institutional guidelines.

Biodistribution of ⁶⁴Cu-labeled tropoelastin probes

Biodistribution experiments are described in the Online Supplementary Methods.

In vivo molecular imaging of tropoelastin at 3T

All imaging experiments were performed using a 3T Philips Achieva MR scanner (Philips Healthcare, Best, The Netherlands) equipped with a clinical gradient system (30mT m^–1, 200mT/m/ms).

Mice

Mice (n=10 per group) were imaged using a single–loop surface coil (diameter=23mm). The animals were placed in prone position and the brachiocephalic artery was imaged 30-40min post injection of the tropoelastin-binding contrast agent Gd-TESMA [(Gd-DOTA)-VVGS] (0.2mmol/kg). A subgroup of animals (n=5 per group) was also scanned 1h (optimal time point) after injection of Gd-ESMA (0.2mmol/kg) on the previous day for comparison (Online Supplementary Data Figure 1A). Additionally, 12 weeks HFD-fed animals (n=3) were imaged 30-40min after injection of the scrambled-VVGS probe (0.2mmol/kg), and another 2 mice were imaged with a 1:1 (Gd-TESMA + La-TESMA) cocktail for in vivo competition experiments. A detailed MRI acquisition protocol is described in the online Supplementary Methods.

Rabbits

The abdominal aorta of diseased rabbits was scanned using a 32-channel cardiac coil. Rabbits received general anaesthesia and were imaged in a supine position twice before (pre) and one time after (post) pharmacological triggering (Online Supplementary Data 1B). The pre-trigger MR imaging included two separate scanning sessions; one before and after administration of Gd-ESMA and one before and after administration of Gd-TESMA. The second MRI session was performed two days after the first to allow vessel wall clearance of Gd-ESMA. During these sessions, native zoom T1-weighted black-blood (BB) and delayed-enhanced (DE) images and T1 maps were acquired 2h after administration of Gd-ESMA (0.2 mmol/kg) and 40min after administration of Gd-TESMA (0.2 mmol/kg). After the second scan, the rabbits were triggered twice for plaque rupture and the final post-triggering MR imaging session was performed 8h later. The post-trigger MRI included acquisition of native zoom T1-weighted BB images to visualize thrombus. A detailed MRI acquisition protocol is described in the online Supplementary Methods.

Ex vivo T1 mapping experiments using symptomatic plaques taken from patients undergoing carotid endarterectomy at 3T

Fresh symptomatic carotid endarterectomy specimens were imaged before and after soaking the in Gd-TESMA (Online Supplementary Data Figure 1C). A detailed tissue handling and MRI acquisition protocol is described in the online Supplementary Methods. Collection and use of human samples was approved by an institutional review committee and subjects gave informed consent.

MRI image analysis

In mice, vessel wall area was calculated by manually segmenting the visually enhanced region of the vessel wall as seen on the DE-MRI images using OsiriX (OsiriX Foundation, Geneva, Switzerland). To ensure that the segmented area encompassed the vessel wall the DE-MRI images were co-registered and fused with the magnetic resonance angiography images. T₁ values were computed on a pixel-by-pixel basis using a 3 parameter fit implemented in Matlab³⁶. In rabbits, pre-trigger native T1w-BB images and contrast enhanced IR and T1 maps were used to assess the vascular remodelling. Post-trigger native T1w-BB images were only used to identify the presence or absence of thrombus as an endpoint for the classification of plaques as rupture-prone and stable, respectively. The T1 maps were automatically generated after the data acquisition using a 3-parameter fit model. The T1 maps were imported into Osirix and the vessel wall was manually segmented to calculate the R₁. When segmenting the T1 maps the anatomical T1w-BB images were used to aid the identification of the vessel wall contours. The analysis was performed on anonymized datasets in all experiments. T1 maps of ex vivo carotid endarterectomy specimens were analyzed as described above for the rabbit images.

Histology

Mice

The aortic root, aortic arch, brachiocephalic and carotid arteries were removed en bloc, pinned down to maintain tissue morphology and fixed in 10% formaldehyde for 48h (n=4 per group). Tissues were embedded in paraffin and sectioned transversely (5-μm thick). Verhoeff-Van Gieson elastin staining (HT25A-1KT, Sigma, Dorset, UK) and Trichrome Stain (Masson) Kit (HT15-1KT, Sigma, Dorset, UK) were used to investigate vessel wall morphology and elastin and collagen fibres, respectively. Immunohistochemistry for tropoelastin was performed using an anti-mouse rabbit polyclonal antibody (1:100, Abcam, ab21600, Cambridge, MA, USA). Vessel wall area was calculated using the Verhoeff-Van Gieson images as [adventitia area−the luminal area (mm²)] using ImageJ (NIH). The immunopositive areas were segmented and expressed as normalized tropoelastin area (%tropoelastin= tropoelastin immunopositive area/vessel wall area) × 100. Fluorescent microscopy was performed using a custom synthesized rhodamine-labeled VVGS peptide derivative (rhod-VVGS). Sections were incubated with a 200nM solution for 24h at 4°C followed by nuclear counterstain using Hoechst (ThermoFischer 33342, 1:3000, for 15 min at room temperature). Slides were shielded from light at all times and mounted with a Mowiol containing 2.5% 1,4-diazobicyclo-[2.2.2]-octane (DABCO, Sigma, D2522) medium.

Rabbits

Transverse cryosections (10μm thick) were collected throughout the length of each segment at 500μm intervals and stained with Verhoeff-Van Gieson to visualize the elastin fibres and calculate the total elastin content. Immunohistochemistry for tropoelastin was performed using an anti-mouse rabbit monoclonal antibody (1:100, NB100-2076, Novus Biologicals, Abingdon, UK). Computer-assisted colour image analysis (ImageJ) was used to quantify the percentage total elastin and tropoelastin content, using the Verhoeff-Van Gieson elastin and tropoelastin immunohistochemistry stainings, respectively. For registration of the in vivo MR images and histologic sections, the distance of the proximal end of each segment from the renal branches and iliac bifurcation, the gross morphology, and internal plaque and/or thrombus landmarks visible on both the MR and histologic images were used as references.

Human carotid endarterectomy (CEA) specimens

After the imaging experiments, the plaques from endarterectomy were fixed in formalin, decalcified, embedded in paraffin, and sectioned transversely at 5μm. Sections were stained for collagen using Masson’s trichrome, elastin with Verhoeff-Van Gieson and tropoelastin by immunohistochemistry (1:100, Abcam, ab21600, Cambridge, MA, USA).

Western Blot experiments for quantification of vessel wall tropoelastin concentration are described in the online Supplementary Methods.

Inductively coupled plasma mass spectrometry (ICP–MS) was used to quantify vessel wall gadolinium concentrations as described in the online Supplementary Methods.

Statistical analyses

The detailed statistical methods are described in the online Supplemental Methods. P-values <0.05 were considered statistically significant.

Results

Chemical structures and in vitro binding assays

The chemical structures and molecular weight of the gadolinium-labelled tropoelastin-binding probes are shown in Figures 1A-B. In vitro binding experiments showed that both K(Eu-DOTA)-YPDH and (Eu-DOTA)-VVGS probes have high selectivity towards tropoelastin compared with other proteins including collagen I, fibronectin, and human serum albumin (HSA) (Figure 1C). Importantly, the (Eu-DOTA)-VVGS probe showed superior discrimination between tropoelastin and mature elastin (64±7% vs 1±2%, P=0.001) compared with K(Eu-DOTA)-YPDH (84± 4% vs 53±5%, P=0.009). Conversely, Eu-ESMA showed equal binding to tropoelastin and mature elastin (40±1% vs 41±2%, P=NS). Additionally, both (Eu-DOTA)-VVGS and K(Eu-DOTA)-YPDH did not show binding to HSA, whereas Eu-ESMA has been reported to bind 15% to HSA [value included in Figure 1C⁴⁰]. Finally, Eu-DTPA and complexes of the scrambled VVGS and YPDH probes showed no binding to all proteins studied. The saturation binding plots (Figure 1D-E) showed that both probes bind to tropoelastin whereas no binding to HSA was observed. Finally, no binding of the scrambled probes to tropoelastin was observed. The K_D values showed high affinity of the probes towards tropoelastin (Figure 1F). Further competition studies showed the specificity of the two probes (Online Data Supplement Figure 2).

Figure 1. Chemical structures and in vitro binding experiments of the tropoelastin-binding probes.

A, B, Chemical structures of the probes. C, In vitro binding show superior discrimination between tropoelastin and mature elastin for the (Eu-DOTA)-VVGS compared with K(Eu-DOTA)-YPDH whereas Eu-ESMA binds equally to tropoelastin and mature elastin. The new probes show less binding towards other proteins compared with tropoelastin whereas the scrambled peptides do not bind to tropoelastin. Eu-DTPA shows little binding to tropoelastin, collagen I and human serum albumin. D-E, The saturation binding plots show that both probes bind to tropoelastin whereas no binding to HSA was observed. Finally, no binding of the scrambled probes to tropoelastin was observed. F, The K_D values showed that the probes have high affinity for tropoelastin. Statistical differences between Kd values are shown.

No binding signifies that the data could not be fit, and lines indicate that the assay was not performed (n=2 for elastin, n=3 for all other proteins). Eu: europium(III), ESMA: elastin-specific MR contrast agent.

Relaxivity studies using Gd-labeled tropoelastin and elastin-binding contrast agents

¹H nuclear magnetic relaxation dispersion (NMRD) profiles obtained in PBS were characteristic of small-molecular weight complexes (Figure 2A-C). As expected, the relaxivity values of the new probes were higher (r₁ =8.9±0.4 and 11.7±0.6mM⁻¹s⁻¹ for K(Gd-DOTA)-YPDH and (Gd-DOTA)-VVGS, respectively, at 20MHz and 37°C) compared with non-peptide based contrast agents (e.g., Gd-DTPA=4.02mM^-1s^-1 at 20MHz, 37°C)⁴⁷ (Figure 2D) due to the increased molecular weight (reducing the rotational correlation time), which also contributes to second- and outer-sphere proton relaxation effects⁴⁷. The relaxivity of Gd-ESMA was measured to be 4.9±0.3mM^-1s^-1 (20MHz, 37°C).

Figure 2. Relaxivity studies and nuclear magnetic relaxation dispersion (NMRD) profiles of Gd-labeled tropoelastin and elastin-binding contrast agents at 20MHz and 37°C.

A-C, ¹H-NMRD in the absence (open circles) and presence (black circles) of tropoelastin. The profiles obtained in PBS (open circles) are characteristic of small-molecular weight complexes. In the presence of tropoelastin (black circles), the relaxivity increases suggesting binding of the probes to the protein. The relaxivity increase was more evident for both (Gd-DOTA)-VVGS and K(Gd-DOTA)-YPDH, compared with Gd-ESMA. D, Summary of the relaxivity values, (n=3 for all studies).

In the presence of tropoelastin (TE), the relaxivity of the bound fraction was calculated taking into account the K_D and a stoichiometry of 1:2 tropoelastin:probe as determined by the fluorescence studies (Figure 1F) and was plotted as a function of the magnetic field strength (Figure 2A-C). These NMRD profiles correspond to slow tumbling systems with a typical high field peak around 20-40MHz showing overall higher relaxivity for the tropoelastin-bound fraction: 36±1, 44±1 and 25±1mM^-1s^-1 for TE-[K(Gd-DOTA)-YPDH)], TE-[(Gd-DOTA)-VVGS)], and TE-[(Gd-ESMA)], respectively.

To investigate the metabolic stability of all Gd-probes we measured the R₁ relaxation rates of 1mM solution of the probes incubated with murine and human blood and plasma every 30min, up to 4h at 3T. These data showed that the probes were stable in solution for up to 4h as the changes in the R₁ relaxation rate were not statistically significant (data not shown).

Biodistribution of ⁶⁴Cu-labeled tropoelastin probes

Biodistribution studies showed favorable pharmacokinetics with high renal clearance for both probes and some liver uptake for the K(⁶⁴Cu-DOTA)-YPDH (Online Data Supplement Figure 3).

In vivo comparison of vessel wall enhancement by MRI using the elastin (ESMA) and tropoelastin binding (TESMA) contrast agent

Considering that (Gd-DOTA)-VVGS was superior in differentiating tropoelastin from cross-linked elastin and had less liver uptake compared to the K(DOTA)-YPDH imaging probe, the (Gd-DOTA)-VVGS, denominated as Gd-TESMA, was selected as the best candidate for in vivo molecular imaging of tropoelastin by MRI. For initial in vivo MRI experiments, control and atherosclerotic ApoE^-/- mice were scanned, after the administration of Gd-ESMA, and 24h later, after the administration of Gd-TESMA (Figure 3). Pilot studies showed clearance of Gd-ESMA from both the blood and vessel wall after 24h (data not shown). Fused maximum intensity projection (MIP) MRA and delayed-enhanced (DE) images after injection of Gd-TESMA showed focal uptake in the brachiocephalic artery (BCA) of an atherosclerotic mouse (Figure 3A). DE-MRI images of the BCA acquired from a control animal after injection of Gd-ESMA showed vessel wall enhancement (Figure 3B-C), as Gd-ESMA binds to endogenously found cross-linked elastin, present in non-diseased arteries. Conversely, there was no enhancement of the control vessel wall after injection of Gd-TESMA due to the lack of tropoelastin in the absence of disease (Figure 3D-E). However, two different atherosclerotic mice, showed enhancement of the vessel wall after administration of both Gd-ESMA and Gd-TESMA (Figure 3F-M). DE-MRI showed a larger area of enhancement after administration of Gd-ESMA compared with Gd-TESMA as the diseased vessel wall contains a mixture of cross-linked elastin and tropoelastin and Gd-ESMA binds to both fractions whereas Gd-TESMA only binds to tropoelastin (also see Figure 1C & online Data Supplement Figure 4A-B).

Figure 3. In vivo MRI comparison of vessel wall enhancement using the elastin (ESMA) and tropoelastin (TESMA) binding contrast agents in mice.

A, Fused MIP reconstructed MRA and DE-MRI images after administration of Gd-TESMA show focal uptake of Gd-TESMA in the BCA of an atherosclerotic ApoE^-/- mouse. B-E, MRI images of the BCA acquired from a control animal, scanned 24h apart, showed vessel wall uptake of Gd-ESMA (B, C) but no uptake of Gd-TESMA (D, E) due to the lack of tropoelastin in the absence of disease. F-M, MRI images of the BCA acquired from two different diseased animals showed enhancement of the vessel wall after administration of both agents due to the presence of both cross-linked elastin and tropoelastin in the atherosclerotic lesion. MIP: maximum intensity projection, BCA: brachiocephalic artery. L.: left, Ao: aortic, (n=5 per group).

In vivo molecular MRI of tropoelastin shows accumulation of tropoelastin during atherosclerosis progression in mice

DE-MRI (Figure 4A₁-D₁ & 4A₂-D₂) and R₁ maps (Figure 4A₃-D₃) acquired after administration of Gd-TESMA showed progressive enhancement and increased vessel wall R₁ relaxation rate of the BCA in ApoE^-/- mice exposed to a HFD compared with wild-type mice. Statin-treated ApoE^-/- mice showed significantly less enhancement and lower relaxation rate compared with untreated ApoE^-/- mice (Figure 4E_1-3) The uptake of Gd-TESMA by DE-MRI and R₁ mapping paralleled the growth of atherosclerotic lesions (Figure 4A₄-E₄) as seen by Van Gieson elastin staining and the deposition of tropoelastin fibers as seen by immunostaining (Figure 4A₅-E_5, Figure 4A₆-E₆ and Figure 4A₇-E₇). The higher magnification images (Figure 4A₆-E₆) of tropoelastin immunopositive areas (black signal) show that the antibody does not bind to endogenously and cross-linked elastin found in the media of control arteries and even in tissues from animals fed with a HFD for 4 weeks. However, as the disease progresses and new tropoelastin fibers are deposited within the growing intima, and also as media elastin becomes fragmented by elastases, the immunopositive areas increase in both of these layers of the vessel wall. ApoE^-/- mice fed a HFD for 12 weeks showed less vessel wall enhancement after administration of the scrambled peptide probe (DOTA-GAESAPLVSSVQSPD) compared with those injected with the non-scrambled probe (Figure 4F₁-F₃). In vivo competition experiments showed displacement of the (Gd-DOTA)-VVGS probe by the (La-DOTA)-VVGS probe in the brachiocephalic artery of atherosclerotic mice (on line Data Supplement Figure 5). Ex vivo fluorescent microscopy experiments showed fluorescent signal originating from the rhodamine-labeled VVGS peptide derivative within tropoelastin-rich areas of diseased vessels as identified by corresponding immunohistochemistry. Conversely, fluorescent signal originating from the probe was found to be nearly non-existent in the regions containing mature/crosslink elastin-rich and collagen in diseased vessels (on line Data Supplement Figure 6).

Figure 4. Molecular in vivo MRI of tropoelastin in atherosclerotic mice.

A₁-E₁ & A₂-E₂ & A₃-E₃, DE-MRI images and R₁ relaxation rate maps after injection of Gd-TESMA show progressive enhancement of the brachiocephalic artery and vessel wall relaxation rate with disease progression compared with control and statin-treated mice. F₁-F₃, DE-MRI and R₁ maps after injecting the scrambled peptide (DOTA-GAESAPLVSSVQSPD) shows less vessel wall enhancement and relaxation rate. A₄-E₄, Van Gieson elastin staining shows the development of atherosclerotic lesions. A₅-E₅, A₆-E₆ and A₇-E₇, Immunohistochemistry shows accumulation of tropoelastin molecules (dark purple staining) in the lesion (n=10 per group). DE-MRI: delayed-enhanced MRI, TESMA: tropoelastin-specific MR contrast agent, HFD: high-fat diet.

Quantitative MRI, histology and western blot measurements of murine vessel wall tropoelastin

Quantification of the MR data showed that the DE-MRI area (Figure 5A, control=0.10±0.005, 4wks=0.13±0.01, 8wks=0.22±0.01, 12wks=0.33±0.01mm², P<0.001) and the vessel wall R₁ (Figure 5B, control=0.87±0.05, 4wks=0.90±0.01, 8wks=1.40±0.03, 12wks=1.87±0.04s^-1, P<0.001) significantly increased with disease progression and decreased with statin-treatment (DE-MRI treated=0.13±0.01mm², R₁ treated=1.37±0.03s^-1). Less uptake was found in 12wk-old HFD-fed ApoE^-/- mice injected with a scrambled peptide of VVGS (0.13±0.01mm² vs 0.33±0.01, P=0.001) and lower R₁ (0.91±0.06 vs 1.87±0.04, P=0.01) compared with animals injected with the non-scrambled Gd-TESMA peptide. Segmentation of the tropoelastin immunopositive areas showed a significant accumulation of tropoelastin fibers during atherosclerosis progression and reduction of tropoelastin in the statin-treated ApoE^-/- mice (% tropoelastin, control mice=3.8±0.50, 4wks HFD =3.74±0.60, 8wks HFD =12.10±0.52, 12wks HFD =18.54±0.30, statin-treated at 12wks with HFD =11.41±0.35, P=0.002) (Figure 5C). Quantification of the vessel wall gadolinium concentration by ICP-MS showed similar trends (control=176±50, 4wks=165±38, 8wks=244±52, 12wks=303±38, statin-treated at 12wks=207±23 nmol/gr tissue, P=0.007) (Figure 5D). There was a significant correlation between the vessel wall R₁ and gadolinium concentration [r=0.97, P=0.004, 95% CI (0.69-0.99)]. A linear regression analysis of average R₁ and average [Gd] per time point of disease progression revealed an in vivo bound relaxivity of r₁=7.15 mM^-1s^-1 (Figure 5E).

Figure 5. Quantitative MRI, histology and western blotting of murine vessel wall tropoelastin.

A-B, DE-MRI area and vessel wall R₁ measured after administration of Gd-TESMA increase with disease progression and decrease with statin-treatment (n=10 per group). C, Tropoelastin accumulates during atherosclerosis progression and reduces in statin-treated mice (n=4 per group). D, Vessel wall gadolinium concentration increases with disease progression (n=4 per group). E, The measured in vivo bound relaxivity of TESMA is r₁=7.15 mM^-1s^-1. F, DE-MRI area correlates with vessel wall R₁ measured after administration of Gd-TESMA (n=10 per group). G-H, The tropoelastin content measured histologically correlates with both the DE-MRI (G) and vessel wall R₁ relaxation rate (H) measured by MRI after administration of Gd-TESMA (n=4 per group). I, Dilutions of recombinant human tropoelastin were used to generate a standard curve for the quantification of the tropoelastin concentration from tissue lysates using western blotting. J, Western blotting of purified elastin showed that cross-linked elastin is an insoluble polymer with a high molecular band (>250kDa) whereas soluble tropoelastin has a molecular weight at 70kDa. Western blotting for tropoelastin extracted from the vessel wall (n=3 per group) showed negligible amounts in control arteries and increased deposition of the protein during atherosclerosis progression in ApoE^−/− mice. Tropoelastin content was decreased in the statin-treated ApoE^−/− mice. K, Quantification of tropoelastin concentration in control and diseased arteries as measured by western blotting.

We found a significant correlation between the DE-MRI plaque area and vessel wall R₁ measured after administration of Gd-TESMA (r=0.58, P<0.001, 95% CI=0.46-0.69) (Figure 5F). There were also strong positive correlations between plaque % tropoelastin (measure immunohistochemically) and both the in vivo DE-MRI area (r=0.76, P<0.001, 95% CI=0.47-0.89) (Figure 5G) and vessel wall R₁ (r=0.93, P<0.001, 95% CI=0.76-0.96) (Figure 5H) suggesting that the increased MRI signal observed in this model reflects the histological accumulation of tropoelastin fibers within the growing lesion.

In addition to the quantification of tropoelastin by immunohistochemistry (Figure 5C), the modulation of the protein levels during atherosclerosis progression and in response to treatment were also measured by western blotting. Dilutions of recombinant tropoelastin were used to generate a standard curve for the quantification of the tropoelastin concentration from tissue lysates (Figure 5I). Western blotting of purified cross-linked elastin and soluble tropoelastin showed that the cross-linked polymer gives a high molecular weight band (>250kDa) whereas the tropoelastin monomer has a molecular weight of 70kDa (Figure 5J). Western blotting of tropoelastin extracted from the vessel wall showed negligible amounts in control arteries and increased deposition during atherosclerosis progression in ApoE^−/− mice. Tropoelastin content was decreased in statin-treated ApoE^−/− mice (Figure 5J). Quantification of tropoelastin concentration in control and diseased arteries is shown in Figure 5K (control=0.62±0.04, 4wks=0.86±0.05, 8wks=1.52±0.06, 12wks=1.99±0.15, statin-treated at 12wks=1.17±0.09 mg/gr tissue, P<0.001).

In vivo molecular MRI of tropoelastin allows detection of rupture-prone rabbit plaque

To test the ability of molecular MRI of tropoelastin in discriminating between stable and rupture-prone plaques we used a rabbit model of atherosclerosis and controlled plaque rupture. Atherosclerosis was observed at 12-weeks in all 6 injured and cholesterol-fed rabbits, and thrombosis occurred in 5 (83%) animals at rates similar to those previously reported^37,45,46,48. Figure 6 shows corresponding native T1wBB, delayed-enhanced images and T₁ maps after administration of Gd-ESMA and Gd-TESMA acquired at 12 weeks and before triggering for plaque rupture and native T1wBB images acquired after triggering for plaque rupture. Representative pre-trigger T1wBB images of a stable (Figure 6A-E) and rupture-prone plaques (Figure 6G-K) show the anatomy of the vessel wall and the location of the plaque. The delayed-enhanced images show an enhancement of the vessel wall and shortening of the T1 relaxation time both after the administration of Gd-ESMA and also after the administration of Gd-TESMA. The corresponding T1wBB images acquired after triggering show the lack of thrombus at the level of the stable plaque and the presence of thrombus at the location of the rupture-prone lesion (Figure 6F and 6L). Although both Gd-ESMA and Gd-TESMA lead to vessel wall enhancement, quantitative assessment of the R₁ relaxation rate showed that the uptake of contrast agent was significantly different between stable and rupture-prone lesions only after administration of Gd-TESMA (Figure 6M). Rupture-prone plaques uptake significantly more Gd-TESMA compared with stable plaques (R₁=2.26±0.1 vs R₁=1.43±0.02, P=0.001) whereas the uptake of Gd-ESMA was similar between stable and rupture-prone plaques (R₁=2.14±0.05 vs R₁=2.44±0.07, P=NS). The sensitivity, specificity, positive predictive value and negative predictive value of the quantitative assessment of vascular elastin and tropoelastin (R₁) after administration of Gd-ESMA and Gd-TESMA, respectively in detecting the rupture-prone plaque are tabulated in Figure 6N. These measurements suggest that quantitative assessment of vessel wall contrast uptake using T₁ mapping after administration of Gd-TESMA allows better in vivo detection of rupture-prone plaques.

Figure 6. In vivo molecular MRI of tropoelastin allows detection of rupture-prone rabbit plaque.

Pre-trigger and post-trigger MRI images of rabbit stable (A-F) and rupture-prone (G-L) plaques. A,G, Pre-trigger native zoom T1wBB images show the plaque. B, H, C, I, Delayed-enhanced inversion recovery images and T₁ maps after administration of Gd-ESMA show a strong enhancement of the aortic wall and shortening of the T₁ relaxation time in both stable and rupture-prone plaque. D, J, E, K, Corresponding delayed-enhanced inversion recovery images after administration of Gd-TESMA show vessel wall enhancement and lower T1 values in ruptured-prone compared with stable plaque. F, L, Corresponding native zoom T1wBB post-trigger images show the presence of thrombus only at the side of the ruptured plaque. M, Quantification of vessel wall R₁ relaxation rate show a significantly higher R₁ value in rupture-prone compared with stable plaque only after administration of Gd-TESMA.

Histological analyses show accumulation of tropoelastin in rabbit rupture-prone compared with stable plaque

Histological analyses of rabbit sections using Van Gieson elastin showed multiple elastin lamellae consisting of mature, cross-linked elastin in control arteries (Figure 7A; black stain) but no corresponding immunopositive tropoelastin fibers suggesting the antibody is specific for monomeric tropoelastin (Figure 7B and 7C).

Figure 7. Histological analyses show accumulation of tropoelastin in rabbit rupture-prone compared with stable plaque.

A, D, G, Van Gieson elastin staining (dark purple indicates elastin fibers) shows the presence of organized elastin fibers in the media of control tissue and a net accumulation of elastin fibers in the intima of stable and rupture-prone plaque. B-C, E-F, H-I, Corresponding immunohistochemistry for tropoelastin (brown staining) shows lack of tropoelastin positive fibers in control aortas, upregulation in stable and deposition of a dense network of tropoelastin fibers in rupture-prone plaque. J-K, Quantification of total elastin and tropoelastin shows significantly higher tropoelastin in ruptured compared with stable plaque.

In both stable and ruptured plaques there was deposition of elastin in the intima (Figure 7D, 7G) and in ruptured plaques there was also additional disintegration and fragmentation of the media elastin as seen by Van Gieson elastin. Immunohistochemistry for tropoelastin (brown stain) showed deposition of tropoelastin in the intima of stable plaque (Figure 7E-F) and a more extensive network of tropoelastin in the intima of ruptured plaque (7H-I). In addition, there were immunopositive tropoelastin areas in the media of ruptured plaques that could be due to either deposition of new tropoelastin fibers or exposure of the tropoelastin epitope when mature elastin fibers become fragmented by elastases. Quantification of total elastin and tropoelastin (Figure 7J-7K) showed that measuring the tropoelastin accumulation in the vessel wall allowed a better discrimination between stable and ruptured plaques (16.6±3.14 vs 42.2±4.0, P=0.001) compared with measuring the net increase of elastin (28.8±3.0 vs 44.35±4.83 %, P=0.03) that includes both the cross-linked and non cross-linked elastin fibers. Only negligible amounts of tropoelastin (7.3±1.45%) were found in control vessel wall segments compared with elastin (26.3±0.9%).

Ex vivo molecular MRI of tropoelastin in human carotid endarterectomy samples

To test the translational value of the tropoelastin contrast agent we performed ex vivo T₁ mapping experiments after soaking the tissues in Gd-TESMA. The MRI results showed retention of Gd-TESMA and significant increase in the R₁ relaxation rate of carotid plaques (R_{1 pre}= 1.1±0.26s^-1 vs R_{1 post}= 3.0±0.1s^-1, P=0.01) (Figure 8A-E) and the histological analyses showed upregulation of tropoelastin in the lesion (Figure 8F-H).

Figure 8. Ex vivo T1 mapping experiments using human carotid endarterectomy specimens at 3T.

A, Ex vivo photograph of an excised left carotid artery. B-C, A T2w image shows the plaque in the internal carotid artery (B) and a corresponding T1 map shows the baseline relaxation values (C). D, A repeated T1 mapping experiment after soaking the specimen in Gd-TESMA shows a reduction of the T1 values. E, Quantification of the changes shows a significant uptake of Gd-TESMA in human carotid plaques. F-G, Corresponding histology shows the lesion (F-G) and the deposition of tropoelastin (dark purple staining) (H).

Discussion

Elastolysis and ineffective elastogenesis favor the accumulation of tropoelastin, rather than mature cross-linked elastin fibers, in atherosclerosis and may trigger plaque progression and instability (6). We have identified tropoelastin-binding peptides and developed a novel gadolinium-based tropoelastin-binding contrast agent that can be used for MR imaging of tropoelastin turnover in atherosclerosis. We demonstrate that the newly developed Gd-TESMA probe (Gd-DOTA)-VVGS: (i) has high selectivity for tropoelastin, (ii) discriminates between tropoelastin and cross-linked elastin, (iii) has high in vitro relaxivity (r₁) that increases upon binding to tropoelastin in solution, and (iv) has favorable pharmacokinetics with blood clearance within 1h via renal excretion. Using (Gd-DOTA)-VVGS, we demonstrate, for the first time, the feasibility of in vivo imaging of tropoelastin turnover in a murine model of atherosclerotic plaque progression and response to treatment and in a rabbit model of plaque instability. In addition, we show the potential for this novel contrast agent to bind to symptomatic plaques taken from patients undergoing carotid endarterectomy. Molecular imaging of tropoelastin in mice showed increased vessel wall enhancement and R₁ relaxation rates during plaque progression compared with statin-treated and control animals. Histology and western blotting confirmed the accumulation of tropoelastin within the growing lesion and ICP-MS confirmed the accumulation of gadolinium in the vessel wall. Molecular imaging of tropoelastin in rabbits showed significantly higher R₁ relaxation rates in rupture-prone compared with stable lesions showing potential for the detection of high-risk plaques with high sensitivity and specificity. Finally, molecular imaging of human carotid endarterectomy specimens showed detection of intra-plaque tropoelastin after soaking the tissues in Gd-TESMA suggesting that this contrast agent has potential in man. Non-invasive quantification of dysfunctional tropoelastin turnover, that fails to cross-link into mature elastin fibres, using a newly developed tropoelastin-binding MR contrast agent may therefore have diagnostic value in detecting not only plaque progression and response to treatment but also detection of unstable atherosclerotic lesions prior to plaque rupture.

The main targeted Gd-based low molecular weight contrast agents that have been developed bind to proteins including fibrin (EP-1873 and EP-2104R)⁴⁹, collagen (EP-3533 and EP-3600)^50,51, and elastin (BMS 753951 and LMI 1174)^35,41,47 that are highly abundant and thus amenable for MR imaging. Most of these agents were initially developed as DTPA chelates and then modified to DOTA chelates, to increase their thermodynamic and kinetic stability⁴⁹, and all are clear from blood within 1-2h after injection. Compared to these agents, our new tropoelastin agent with a single Gd-DOTA conjugated to the N-terminal of the 15 amino-acid peptide (2kDa), showed comparable affinity (K_D≈µM), relaxivities and pharmacokinetics to these agents^{35,36,40,49–51}.

Previous studies have shown that the VVGS^12,13,52 probe is expected to bind to all sequences that follow the GXXPG and XGXPG motif (where X is a hydrophobic amino acid) that repeats multiple times in the tropoelastin molecule. In murine tropoelastin, these commonly repeated sequences represent 21% of the entire molecule⁵³ that may explain our observation of higher vessel wall ICP-MS gadolinium measurements compared with the tropoelastin concentration as measured by western blotting. The in vivo vessel wall bound relativity of (Gd-DOTA)-VVGS was 7.15 mM^-1s^-1 and the image acquisition window lasted up to 1.5h post-injection. As some of the binding sites of tropoelastin might be masked and the amount of water is lower within the atherosclerotic lesion this could explain why the in vivo bound relaxivity was lower compared with the relaxivity in solution. Future improvements of the chemical properties of the agent, e.g., by chelating more Gd³⁺ per peptide, or conjugating the peptide to a different chelate unit that enables two water molecules in the inner coordination sphere of Gd³⁺ could increase the relaxivity of the free, but more importantly of the bound fraction of the agent, and also prolong the imaging acquisition window to allow better detection of vascular tropoelastin and imaging of larger vascular segments.

Our finding that tropoelastin accumulates within the growing lesion and reduces with statin-treatment (using immunohistochemistry and western blotting) in ApoE^-/- mice is in agreement with a previous study³⁶. Using western blotting we found that tropoelastin accumulation is up to 4-fold increased in the murine plaque, compared with control tissue, and reaches a concentration of about 2mg/gr of tissue. Although statins directly reduce plaque burden through a lipid-lower mechanism, they were also shown to abolish the down-regulation of LOX produced by tumour necrosis factor-α (TNFα) and LDL particles through a RhoA/Rho kinase-dependent mechanism²⁷. Normalization of LOX could promote cross-linking of the newly synthesized tropoelastin leading to an overall reduction of tropoelastin, as seen in our study.

When compared with the Gd-ESMA, the new tropoelastin-agent resulted in a smaller area of contrast uptake at the same time points by MRI in atherosclerotic ApoE^-/- mice. A previous study also reported a higher uptake of (¹⁵³Gd-DTPA)-ESMA compared with (⁶⁴Cu –DOTA)-VVGS (13.2 vs. 7.2 %ID/g)⁴¹. As shown in this and other studies^36–38,40 Gd-ESMA binds equally to cross-linked elastin and tropoelastin. For this reason, enhancement of endogenous cross-linked elastin in non-diseased arteries was also observed in both mice and rabbits. To this end, Gd-ESMA allows measurement of total elastin content and the percentage increase that occurs from the control to the diseased condition by MRI.

In our current study, we found that quantitative assessment of vessel wall contrast uptake using T₁ mapping after administration of Gd-TESMA improved the detection of rupture-prone plaques compared with Gd-ESMA. Rabbit rupture-prone plaques uptake significantly more Gd-TESMA (higher vessel wall R₁) compared with stable plaques whereas the uptake of Gd-ESMA was similar between the two groups (similar R₁ values). The elimination of signal from endogenously present cross-linked elastin, and detection of only immature tropoelastin, using the Gd-TESMA, agent allowed a stronger discrimination between rupture-prone and stable rabbit plaques and high prediction of unstable lesions compared with Gd-ESMA. In our previous study using the same rabbit model³⁷, we showed that administration of Gd-ESMA allowed measurement of plaque burden, that was significantly increased in diseased vessel walls (higher R₁) compared with control vessels and also improved the assessment of positive vascular remodeling that predicted rupture-prone plaques. Finally, to determine whether tropoelastin could be imaged in human tissue we carried out ex vivo T₁ mapping experiments in carotid plaque samples obtained from endarterectomy and soaked in Gd-TESMA. The MRI results showed retention of the agent in carotid plaques and the histological analyses revealed increased levels of tropoelastin in these plaques. Our findings are in agreement with studies demonstrating that synthetic VSCMs^24,54 and macrophages^6,8 secrete tropoelastin in rabbit atherosclerotic plaque and human carotid atheroma.

In conclusion, this study shows that the accumulation of tropoelastin is associated with plaque progression and instability in atherosclerotic models. In addition, the novel contrast agent is able to demonstrate the presence in tropoelastin in symptomatic plaques taken from patients supporting a potential role for this novel contrast agent in man.

Study Limitations

Our findings on the value of tropoelastin as a marker of plaque burden and instability relies on the use of animal models of atherosclerosis that reflect, only partly, the complex composition of human atherosclerotic plaques and their evolution towards rupture. Although a small intermediate step of using ex vivo human endarterectomy specimens was used to show the translational potential of our work, future studies including larger numbers of human samples will be needed. In addition, the ability of molecular imaging of tropoelastin to detect rupture-prone plaques was based on a relatively small a number of rabbits, larger animal studies will be need to confirm these findings.

A potential limitation of our study is that an independent black-blood (e.g., double inversion recovery) sequence for the entire vessel wall anatomy was not acquired for the murine experiments. Instead, we implemented a flow-independent delayed-enhancement MR imaging protocol that was clinically established for myocardial infarction and coronary vessel wall imaging in humans instead of a flow–dependent black-blood (BB) MR imaging. Another limitation our study was the lack of an automated/semi-automated software for region-of-interest (ROI) analysis, which was performed manually for all experiments. Future implementation of threshold-based segmentation algorithms may improve vessel wall segmentation and reduce observer bias. Despite the initial potential of using the tropoelastin agent for imaging human plaques further optimisation and extensive safety checks will be needed prior to using this agent for in vivo imaging of tropoelastin in man.

Conclusions

We demonstrate the development of a new tropoelastin-binding contrast agent and the first in vivo implementation of molecular imaging of tropoelastin in animal models of plaque progression and instability, and show uptake of the agent human carotid plaques ex vivo. Tropoelastin may represent a new imaging biomarker for plaque progression and instability and for the monitoring of the effectiveness of therapeutic interventions.

Clinical Perspectives.

A tropoelastin-binding peptide labeled with gadolinium is a novel probe for molecular imaging of dysfunctional elastogenesis by in vivo MRI. Our new probe combined with quantitative T₁ mapping may be used as a new imaging biomarker to quantify ineffective elastogenesis related to plaque progression and instability and identify plaques that may be susceptible to rupture.

Molecular imaging of tropoelastin using our presented imaging approach may provide information on plaque progression and instability and for monitoring the effectiveness of established and novel interventions in different animal models and human tissue. We showed that statins, a widely used intervention, reduced tropoelastin accumulation in atherosclerotic plaques but the long term effects on plaque stability remains to be elucidated. Similarly, the effectiveness of novel interventions, that specifically aim elastin remodeling, can be and remain to be investigated using this contrast agent. Importantly, in our work we implemented MRI protocols that have already been used in the clinical setting enabling translation of our current findings in bigger animal models of coronary atherosclerosis to identify plaques that may be susceptible to rupture.