Positron emission tomography imaging of vessel wall matrix metalloproteinase activity in abdominal aortic aneurysm

Abstract

Background.

Matrix metalloproteinases (MMPs) play a key role in the pathogenesis of abdominal aortic aneurysm (AAA). Imaging aortic MMP activity, especially using positron emission tomography (PET) to access high sensitivity, quantitative data, could potentially improve AAA risk stratification. Here, we describe the design, synthesis, characterization, and evaluation in murine AAA and human aortic tissue of a first in class MMP-targeted PET radioligand, ⁶⁴Cu-RYM2.

Methods.

The broad spectrum MMP inhibitor, RYM2 was synthetized, and its potency as an MMP inhibitor was evaluated by a competitive inhibition assay. Toxicology studies were performed. Tracer biodistribution was evaluated in a murine model of AAA induced by angiotensin II (Ang II) infusion in Apolipoprotein E-deficient (Apoe^−/−) mice. ⁶⁴Cu-RYM2 binding to normal and aneurysmal human aortic tissues was assessed by autoradiography.

Results.

RYM2 functioned as an MMP inhibitor with nanomolar affinities. Toxicology studies showed no adverse reaction in mice. Upon radiolabeling with Cu-64, the resulting tracer was stable in murine and human blood in vitro. Biodistribution and metabolite analysis in mice showed rapid renal clearance and acceptable in vivo stability. In vivo PET/CT in a murine model of AAA showed a specific aortic signal, which correlated with ex vivo measured MMP activity and Cd68 gene expression. ⁶⁴Cu-RYM2 specifically bound to normal and aneurysmal human aortic tissues in correlation with MMP activity.

Conclusion.

⁶⁴Cu-RYM2 is a first in class MMP-targeted PET tracer with favorable stability, biodistribution, performance in preclinical AAA, and importantly, specific binding to human tissues. These data set the stage for ⁶⁴Cu-RYM2-based translational imaging studies of vessel wall MMP activity, and indirectly, inflammation, in AAA.

Introduction

Abdominal aortic aneurysm (AAA), a potentially life-threatening disease due to the risk of rupture, primarily affects males aged 65 and older with a history of smoking. In the absence of any effective medical therapy,¹ active surveillance with serial imaging and eventual aneurysm repair is currently the standard management strategy for patients with asymptomatic AAA.² The risk of AAA rupture increases with aneurysm size and growth rate, and patients who become symptomatic or those with an AAA exceeding a specific size or growth rate threshold are directed to surgical or endovascular repair.² While this approach has reduced AAA-associated mortality in the general population, current guidelines and imaging techniques fall short of identifying patients with smaller AAA size that may have an increased risk of rupture. In fact, 10–15% of all ruptures of AAA occur at a size below the threshold for elective repair.³ For these reasons, improvement in rupture risk stratification remains an unmet clinical need, which if addressed, would decrease AAA mortality and morbidity by guiding patients with high-risk AAA toward AAA repair, while avoiding intervention in patients with low-risk aneurysms, irrespective of size.

Matrix metalloproteinases (MMPs) are a family of zinc-dependent endopeptidases with the ability to degrade different components of extracellular matrix.⁴ Most MMPs are produced and secreted as inactive zymogens, which can undergo proteolytic activation to remove an N-terminal pro-domain that exposes their active site responsible for MMP activity.⁵ MMP activity is tightly regulated, and its dysregulation contributes to a number of pathophysiological conditions^6–8 including AAA development.^{6, 9, 10} While several intrinsic vascular cells (e.g., smooth muscle cells) can produce MMPs, macrophages are a major source of MMPs and enhanced MMP activity may be a mechanism through which inflammation promotes aneurysm expansion and rupture.^{10, 11} Accordingly, non-invasive molecular imaging of MMP activity within the aneurysm wall could potentially serve as a robust diagnostic tool to evaluate AAA rupture risk.

Among different molecular imaging modalities, positron emission tomography (PET) stands out for its high sensitivity and quantitative nature that facilitates the detection of biological processes in small targets such as the aortic wall.¹² Moreover, compared to its sister modality, single photon computed tomography (SPECT), PET has a higher sensitivity and a better spatial resolution that is especially important for reducing signal quantification errors in small targets.¹² Keeping in mind the potential for clinical translation, we designed a macrocyclic hydroxamate MMP inhibitor with a zwitterionic side chain and a hexadentate chelator for metal complexation, RYM2, as a first in class MMP-targeted PET tracer precursor. Here we describe the synthesis and characterization of the copper-64 (⁶⁴Cu)-labeled MMP inhibitor, ⁶⁴Cu-RYM2, as a PET radiotracer for molecular imaging of activated MMPs, and its evaluation in murine AAA and human aortic tissue. Our data demonstrate favorable characteristics of ⁶⁴Cu-RYM2, which enabled in vivo imaging and quantification of MMP activity, and indirectly inflammation, in a murine AAA model, and ex vivo human aortic tissues, as a prelude to clinical translation.

Methods

All the data presented in this article are available upon reasonable request for purposes of reproducing the results or replicating the procedure.

Synthetic chemistry

The synthetic pathway and key intermediates to access the radiotracer precursor, RYM2 and their chemical characterization are presented in the Supplement.

MMP inhibition assay

Evaluation of RYM2 inhibition constant (K_i) was performed using activated recombinants MMPs (R&D Systems) and the fluorogenic MMP substrate peptide Mca-KPLGL-Dpa-AR-NH2 (R&D Systems), following previously described methods.¹³ Briefly, recombinant human and murine MMPs were activated using 4-aminophenylmercuric acetate (APMA) following manufacturer’s instructions and mixed with different amounts of the fluorogenic substrate (0.1 – 6.3 μM) and RYM2, Cu-RYM2 or GM6001 (6.3 – 100 nM). Fluorescence was recorded over 30 min on a microplate reader (Synergy HT, BioTek) to measure the initial rate of reaction (V), and subsequently fitted to the competitive inhibition model to determine the K_i.

Radiochemistry

Cyclotron generated ⁶⁴Cu-CuCl₂ (~ 296 MBq, ~ 8 μl) produced at the Washington University School of Medicine Mallinckrodt Institute of Radiology Cyclotron Facility, was converted to ⁶⁴Cu-Cu(OAc)₂ by adding 50 μL of 0.5 M NH₄OAc (pH 3.5) and mixing well. RYM2 (5 μg in 5 μL DMSO) was mixed with 60 μL of 0.1 M NH₄OAc (pH 5.5), added to the radioisotope, and heated at 45 °C for 30 min. The radiolabeled solution was cooled down to room temperature before subjecting it to quality control by radio high performance liquid chromatography (HPLC). Conjugation of non-radioactive copper (^63/65Cu) to RYM2 was performed following a similar procedure using Cu(OAc)₂ (see Supplementary Materials and Methods).

HPLC analysis was performed using a Phenomenex, Jupiter 4μm, Proteo 90Å, 100 × 4.6mm reverse-phase analytical column with C18 as stationary phase and water containing 0.16% ammonium formate (A) and a solution of 9:1 acetonitrile/solution A (B) as mobile phase for gradient elution. The gradient was initiated at a flow rate of 1 mL/min, starting from 90% A for 4 min, followed by a linear gradient to 10% A over 16 min, and an isocratic run for another 10 min, and was linearly brought back to the initial gradient by the end of the run. UV chromatograms were monitored at both 254 nm and 214 nm with the ⁶⁴Cu-labeled product monitored by a radioactivity detector connected in parallel with the UV detector.

Blood partition and in vitro stability analysis

⁶⁴Cu-RYM2 was incubated at 37 °C in murine and human blood, collected in heparinized tubes. Samples were collected at 5, 30, 60 and 120 min of incubation and centrifuged to separate blood cells from plasma. Plasma proteins were precipitated using methanol. Radioactivity in blood cells, plasma protein and plasma free fraction was quantified by gamma-counting (Wizard², PerkinElmer). The plasma free fraction was further analyzed by HPLC.

Animals

To generate abdominal aortic aneurysm, male Apolipoprotein E-deficient (Apoe^−/−) mice fed a chow diet were infused with recombinant human angiotensin II (Ang II) at 1000 ng/kg/min for up to 4 weeks, using osmotic pumps (model 2004; Alzet) implanted subcutaneously (n = 48). Male Apoe^−/− mice fed a diet containing 1.25% cholesterol and 0.5% sodium cholate (TD.88051, Envigo) were used to generate carotid artery aneurysm via a 20 min-long peri-adventitial application of CaCl₂ 10% (n = 6), as described.¹⁴ In accordance with prior work and to achieve a higher prevalence of aneurysm, only male animals were used.¹³ Additional age-matched untreated male Apoe^−/− mice (n = 7) and wild-type C57Bl/6 mice (n = 47) fed a chow diet were used to evaluate tracer stability, biodistribution, toxicology and metabolites. A flow chart summarizing all animal experiments is presented in the supplemental materials (Figure S1). The sample sizes are based on prior studies.¹³ All animals were originally purchased from Jackson Laboratories or bred locally, and housed in standard conditions (12 h light–dark cycle, corn cob bedding, ad libitum access to food and water). Surgical procedures were performed on animals anesthetized with isoflurane 1–3% or ketamine 100 mg/kg and xylazine 10 mg/kg, intraperitoneally. Postoperative analgesia (carprofen 5 mg/kg/day, subcutaneous) was maintained for 2 days. The experiments were performed under protocols approved by Institutional Animal Care and Use Committees of Yale University (10455), VA Connecticut Health Care System (MS0003), and Washington University (19–0940).

Metabolite analysis

In vivo radiochemical stability was assessed in blood and urine of C57BL/6J mice (n = 3). At 1 h post tracer injection (i.v., tail vein), blood was collected from the heart into a lithium heparin tube (BD Vacutainer^™ Capillary Blood Tube). After centrifugation at 7600 g for 10 min, 100 μL of the supernatant was diluted with 200 μL of water, passed through a 0.22 μm syringe filter, and the resulting filtrate injected into a reverse-phase HPLC (Agilent 1200 Series). HPLC fractions were collected every 0.5 min, and radioactivity was measured using a gamma counter (WIZARD, PerkinElmer). Urine was collected from the bladder and diluted with water (2-fold). After filtration using a 0.22 μm syringe filter, serial HPLC fractions were collected, and their radioactivity was measured using a gamma counter.

Toxicology

Preliminary toxicology of RYM2 was evaluated in C57Bl/6J mice (n = 20) at the Washington University Medical Center, Division of Comparative Medicine Research Animal Diagnostic Lab, following standard procedures, using 100-fold murine equivalent of a human dose of 50 μg (1.025 mg/kg).

Contrast-enhanced CT, PET/CT, and image analysis

To enrich the experimental group in animals with AAA, a group of Apoe^−/− mice at 19 or 20 days of Ang II infusion were injected with Exitron nano 12000 (Viscover Imaging) 15 min prior to a CT acquisition on a small-animal dedicated SPECT/CT scanner (U-SPECT⁴CT, MILabs). For this and all other imaging experiments, the animals were maintained under anesthesia with isoflurane 1–3%, and their body temperature maintained using a bed heating system. CT images were reconstructed using manufacturer’s software (reconstruction software, version 11, MILabs) at an isotropic voxel size of 0.1 mm with a filtered back projection algorithm. Contrast-enhanced CT images were qualitatively analyzed to identify the presence of AAA, based on 1) the presence of a false-lumen, 2) irregular shape of the abdominal aortic lumen. Those without evidence of vascular remodeling were not used for PET imaging.

For PET/CT imaging, Ang II-infused Apoe^−/− (n = 23) and control Apoe^−/− (n = 6) under isoflurane anesthesia were injected with 30.2 ± 11.8 MBq of ⁶⁴Cu-RYM2 and Exitron nano 12000 (Viscover Imaging). After contrast agent injection, a CT acquisition was performed followed by a 10 min-long PET acquisition starting at 50 min after tracer injection, or a 30 min-long PET acquisition starting 30 min p.i. (in which case, only the last 10 min-long frame was included in the final analysis). A subset of animals (n = 6) underwent a second imaging protocol on the following day, including animals (n = 4) preinjected with 500-fold excess of cold Cu-RYM2 (blocking). CT images were reconstructed at a down-sampled resolution (0.815 × 0.815 × 0.796 mm) using Feldkamp cone beam algorithm for attenuation correction. An additional non-down-sampled reconstruction was performed using a 3D ordered subset expectation maximization method (OSEM3D) algorithm, at an isotropic voxel size of 0.111 mm for registration with PET images. Emission data were reconstructed using the OSEM3D method (2 iterations, 16 subsets) and maximum a posteriori probability algorithm (MAP; 25 iterations) to a voxel size of 0.388 × 0.388 × 0.796 mm with decay, attenuation, scatter, normalization and randoms corrections. All PET/CT data were acquired using a small animal dedicated PET/CT scanner (Inveon PET/CT, Siemens Medical Solutions) and reconstructed using the manufacturers’ software (Inveon Workstation, version 2.0, Siemens Medical Solutions). PET images were corrected by the injected dose and body weight. Tracer uptake is presented in standardized uptake values (SUV). Quantification of blood and aortic activity was performed using standard-size volumes of interest drawn in the left ventricle cavity and at the site of highest uptake along the suprarenal abdominal aorta, avoiding spill over from adjacent structures with high activity, using coregistered contrast-enhanced CT images for anatomical reference. The analysis was performed independently by two investigators (JT and MG), with an intraclass correlation coefficient for combined measurements of 0.97, indicating an excellent interobserver agreement. The maximal SUV value (SUV_max) in the AAA/abdominal aorta served as the primary endpoint for PET studies.

Biodistribution studies

C57Bl/6J mice (n = 21) were injected with ⁶⁴Cu-RYM2. After 1 h, 4 h and 24 h, animals were euthanized and samples of tissues of interest were collected, weighed and their activity recorded on a gamma-counter (Wizard², PerkinElmer). Tracer activity was expressed as percent of injected dose per gram of tissue (%ID/g).

Murine tissue analysis

After PET/CT, Ang II-infused Apoe^−/− and control Apoe^−/− mice were euthanized and samples of tissues of interest were collected, weighed and their activity recorded on a gamma-counter. Aortic samples were collected using a binocular microscope, embedded in OCT and deep frozen. Visual inspection of the aorta of Ang II-infused Apoe^−/− mice during tissue collection, focused on the identification of focal bulging, was used to classify animals as low remodeling or AAA. Frozen tissue was processed (CM1850, Leica) to obtain 5 μm-thick tissue sections at 0.8 mm interval for morphometry analysis. Tissue between the sections was used for protein zymography and gene expression analysis (secondary endpoints).

Carotid aneurysm studies

⁶⁴Cu-RYM2 uptake in carotid aneurysm in basal and blocking condition, as well as the tracer’s biodistribution, were studied in Apoe^−/− mice at 4 weeks after CaCl₂-induced carotid aneurysm. The animals were injected with ⁶⁴Cu-RYM2 alone (basal conditions) or the tracer following the pre-injection of 500-fold molar excess of the MMP inhibitor RYM (blocking condition). Blood samples were collected up to 2 h post-injection (p.i.). At 2 h p.i., the animals were euthanized, and samples of different tissues were collected to evaluate the tracer’s biodistribution. The aorta and major arterial branches were carefully collected under a binocular microscope for autoradiography.

Human tissues and blood

All experiments were performed under protocols approved by Yale University, VA Connecticut Healthcare System, and Washington University Institutional Review Boards and the New England Organ Bank. Blood was collected from anonymous, healthy blood donors for in vitro stability assays. AAA and non-diseased (normal) thoraco-abdominal aorta tissues were collected from anonymized patients undergoing surgical aneurysm repair and deceased transplant donors, respectively. Tissues were deep-frozen, embedded in OCT and processed to obtain 5 μm-thick sections, and used for autoradiography and Movat’s pentachrome staining. Adjacent tissue sections were used for in vitro binding assays, protein zymography, and gene expression analysis.

Tissue homogenate binding assay

Human aortic tissue was homogenized in phosphate buffered saline (PBS) and incubated with ⁶⁴Cu-RYM2 for 2 h at 37 °C. A 500-fold molar excess of GM6001 was added in blocking conditions. After incubation, the tissue was centrifuged at 8000 rounds/min for 1 min, supernatant was removed, and the tissue pellet was washed with tris buffered saline containing 0.1% Tween 20 (TBST) and PBS successively. Proteins were extracted using an ethane-1,2-diyldinitrilotetraacetic acid (EDTA)-free lysis buffer. After determining protein concentration, the tissue lysate radioactivity was measured by gamma counting. To analyze tracer binding, decay-corrected radioactivity was normalized to total protein content.

Protein zymography

Proteins from human and murine aortic tissues were extracted in EDTA-free lysis buffer (NaCl 0.3 M, Tris 50 mM, Triton X-100 1%, cOmplete^™ Protease Inhibitor Cocktail, Sigma-Aldrich). MMP activity of 1 μg of protein lysate was measured using a fluorometric zymography assay (SensoLyte 520 Generic MMP Activity Kit, AnaSpec), as described.¹³

Gene expression analysis

Human and murine aortic tissues were used for RNA isolation (GenElute Mammalian Total RNA Miniprep Kit, Sigma-Aldrich) and reverse transcription (QuantiTect Reverse Transcription Kit, Qiagen). The resulting cDNA was used for real-time quantitative PCR analysis (7500 Real-Time PCR System, Applied Biosystems) to assess gene expression using TaqMan probes (Cd68: Mm03047343_m1, Mmp2: Mm00439498_m1, Mmp3: Mm00440295_m1, Mmp9: Mm00442991_m1, Mmp12: Mm0050054_m1, Mmp13: Mm00439491_m1 and Actb: Mn00607939_s1, CD68: Hs02836816_g1, ACTB: Hs01060665_g1, TaqMan Gene Expression Assays, Thermo Fisher Scientific).

Statistics

Continuous variables are presented as mean ± SD. Normality of data distribution was tested using Shapiro-Wilk test. The difference between the means of multiple groups, two independent groups and two related groups, was compared using one-way ANOVA with Tukey’s test for multiple comparisons, unpaired t-test or paired t-test for normally distributed variables, respectively, and Kruskal-Wallis test with Dunn’s test for multiple comparisons, Mann-Whitney test and Wilcoxon test for not normally distributed variables, respectively. The association between variables was assessed using Spearman’s rank correlation coefficient. Prism 9 (GraphPad) was used for statistical analysis and to create charts. Interobserver agreement was evaluated with the intraclass correlation coefficient. Statistical outcomes were considered significant when P < 0.05.

Results

Rationally designed PET radioligand precursor, RYM2, and Cu-RYM2 bind with moderate to high affinity to MMPs

We designed a novel MMP PET tracer precursor, RYM2, by integrating a hydroxamate group acting as a strong zinc binding moiety, a stereo-selective side chain at P1′-segment that interacts within the hydrophobic S1′ pocket of MMPs, a zwitterionic arginine side chain for improving hydrophilicity and the hexadentate chelator, 2,2’,2”-(1,4,7-triazacyclononane-1,4,7-triyl)triacetic acid (NOTA) for Cu-64 conjugation, into a single molecule (Figure 1A). RYM2 was synthetized following the scheme presented in Figure S2 and methods described in supplements. Briefly, intermediate C-5 with NOTA was prepared in a 5-step reaction scheme starting from Fmoc-methoxy-triethylbenzenesulfonyl-L-arginine, and coupled with a macrocyclic hydroxamic acid intermediate prepared as described previously.¹⁵ The characterization by nuclear magnetic resonance (NMR) spectroscopy of the final compound and unpublished intermediates is presented in supplements (Figure S3–S10). Conjugation of cold copper to NOTA chelator to generate Cu-RYM2 was performed using copper acetate in a single step rapid reaction (Figure S11A). Characterization of RYM2 and Cu-RYM2 by liquid chromatography–mass spectrometry (LC-MS) is presented in supplements (Figure S11B, C, S12 and S13).

Figure 1.

Design, radiochemistry, radiochemical stability and blood partition of ⁶⁴Cu-RYM2. (A) Design of RYM2. (B) Schematic of radiolabeling reaction of RYM2 with ⁶⁴Cu-CuCl₂. (C) Radiochromatogram spectra of ⁶⁴Cu-CuCl₂ and ⁶⁴Cu-RYM2 maintained for up to 24 h in the radiolabeling media. (D) Radiochromatograms of the plasma free fraction from human blood incubated with ⁶⁴Cu-RYM2 for up to 2 h at 37 °C. (E) ⁶⁴Cu-RYM2 binding to blood cells (BC) and plasma protein (PP), and activity in plasma free fraction (PFF).

Evaluation of RYM2 and Cu-RYM2 inhibition constants (K_i) towards several activated recombinant human MMPs confirmed their moderate to high affinity for MMPs (Table 1, see Figure S14 for an example of competitive inhibition data in comparison with the broad-spectrum MMP inhibitor, GM6001). In this regard, Cu-RYM2 consistently showed 1.2 to 3.2-fold higher affinities compared to its unconjugated precursor, RYM2, with K_i values ranging from 9 ± 2 to 98 ± 37 nM (for human MMP-12 and MMP-9, respectively). Notably, while inhibition was observed for all tested MMPs, both molecules exhibited modest selectivity towards MMP-12.

Table 1.

Inhibition constants (K_i) of RYM2 and Cu-RYM2.

	Ki (nM)*
	RYM2	Cu-RYM2
rhMMP-2	48 ± 23	17 ± 7
rhMMP-7	51 ± 5	30 ± 12
rhMMP-9	124 ± 14	98 ± 37
rhMMP-12	11 ± 3	9 ± 2
rhMMP-13	180 ± 98	56 ± 47

^*K_i values represent the mean ± standard deviation of 3 – 5 experiments.

⁶⁴Cu-RYM2 is radiochemically stable in vitro and displays favorable pharmacokinetics, metabolic stability, and safety profile for in vivo vascular imaging

Efficient conjugation of ⁶⁴Cu, as copper acetate to the NOTA moiety of RYM2 was achieved using a single step, rapid radiolabeling procedure (Figure 1B), to access ⁶⁴Cu-RYM2 with a median specific activity of 61 MBq/nmol [interquartile range (IQR): 47 – 89 MBq/nmol, n = 8]. When analyzed by high-performance liquid chromatography (HPLC), the labeling product showed a single peak at the expected retention time, which remained stable in the radiolabeling media for up to 24 h (Figure 1C). Incubation of ⁶⁴Cu-RYM2 in human blood for up to 2 h in vitro led to no observable degradation (Figure 1D). Blood partition analysis showed that 71 –74 % of radioactivity was in the free fraction of the plasma with no evolution over time (Figure 1E). A similar analysis of ⁶⁴Cu-RYM2 stability in mouse blood did not reveal any radiochemical instability (Figure S15).

Evaluation of ⁶⁴Cu-RYM2 biodistribution in wild-type mice showed rapid renal clearance of the tracer with low residual blood activity [0.7 ± 0.1 %injected dose (ID)/g] at 1 h post-injection (p.i.) (Table 2). HPLC analysis of mouse blood and urine collected at 1 h p.i. showed the presence of radiometabolites at a retention time corresponding to free ⁶⁴Cu, with the parent compound representing 84.4 ± 8.1 % and 94.2 ± 5.6 % of blood and urine radioactivity, respectively (Figure 2). Preliminary toxicology analysis of RYM2 (1.025 mg/kg, i.v.) in wild type mice, including blood chemistry, hematology and urinalysis parameters showed no adverse effect compared to vehicle-treated animals at 24 hr p.i. (Table S1).

Table 2.

Biodistribution of ⁶⁴Cu-RYM2 in C57Bl/6 mice, as percent injected dose per gram of tissue (%ID/g), at 1 h, 4 h and 24 h post-injection.

	%ID/g
	1 h	4 h	24 h
Blood	0.7 ± 0.1	0.4 ± 0.1	0.2 ± 0.1
Heart	1.3 ± 0.3	0.8 ± 0.1	0.5 ± 0.1
Lungs	2.7 ± 1.0	2.5 ± 0.3	1.5 ± 0.4
Liver	3.4 ± 1.2	4.1 ± 0.6	2.4 ± 0.3
Spleen	1.9 ± 0.4	1.1 ± 0.3	0.4 ± 0.1
Fat	0.6 ± 0.2	0.2 ± 0.1	0.1 ± 0.1
Muscle	0.9 ± 0.4	0.4 ± 0.1	0.1 ± 0.0
Brain	0.1 ± 0.0	0.1 ± 0.0	0.1 ± 0.0
Bone	1.4 ± 0.7	0.8 ± 0.3	0.2 ± 0.1
Bone Marrow	0.6 ± 0.5	1.0 ± 0.8	1.0 ± 0.2
Pancreas	1.4 ± 0.5	0.8 ± 0.2	0.3 ± 0.1
Stomach	1.6 ± 0.5	1.5 ± 0.5	0.6 ± 0.2
Small intestine	3.3 ± 1.0	3.8 ± 1.1	0.9 ± 0.2
Kidney	114 ± 39	38 ± 14	1.8 ± 0.4

Values represent mean ± standard deviation from 7 animals.

Figure 2.

⁶⁴Cu-RYM2 stability in vivo. High-performance liquid chromatography analysis of ⁶⁴Cu-RYM2 radiometabolites in mouse blood (A) and urine (B) collected at 1 h after injection. a.u.: arbitrary units. Data from three animals are shown in different colors on each graph.

⁶⁴Cu-RYM2 PET/CT can differentiate AAA from non-aneurysmal aorta

Apoe^−/− mice infused with angiotensin II (Ang II) for up to 4 weeks to induce AAA and untreated, control Apoe^−/− mice were used to evaluate ⁶⁴Cu-RYM2 for PET imaging of aortic MMP activity. The animals underwent PET/CT at 50 to 60 min after intravenous injection of ⁶⁴Cu-RYM2, followed by animal euthanasia and tissue collection. After a first set of studies, to enrich the Ang II-infused group of animals for the presence of AAA, those without evidence of vascular remodeling on contrast-enhanced CT images were excluded from PET (Figure S1). This enrichment was confirmed on visual examination of the aorta at the time of tissue harvesting, which showed an AAA in 43% (10/23) of the Ang II-infused mice who underwent PET/CT, with the remaining 57% (13/23) showing low remodeling of the suprarenal abdominal aorta. None of the control, age-matched Apoe^−/− mice (n = 6) exhibited any degree of aortic remodeling (Figure 3A–C). A posteriori morphometry analysis was consistent with visual inspection as the maximal external diameter of the suprarenal abdominal aorta was significantly larger in animals with AAA on visual inspection compared to control animals and Ang II-infused animals with low remodeling (P < 0.0001 vs. both), with a cut-off value of 1.3 mm separating the AAA group from the other two groups (Figure 3D).

Figure 3.

Morphometry of suprarenal abdominal aorta. (A) Examples of Verhoeff–van Gieson staining of the suprarenal abdominal aorta from control, Apoe^−/− mice (Con, A), angiotensin II-infused Apoe^−/− mice with low remodeling (LR, B) and mice with abdominal aortic aneurysm (AAA, C). (D) Quantification of the maximal external diameter of suprarenal abdominal aorta. Scale bar: 1 mm. **** P < 0.0001, One-way analysis of variance.

On PET/CT images of animals with AAA, focal uptake of ⁶⁴Cu-RYM2 could be detected in the aneurysmal region of the aorta identified on contrast-enhanced CT images, whereas no clear area of high activity was identifiable on PET images of control animals and animals with low remodeling (Figure 4A–C). Quantification of the maximal signal along the suprarenal abdominal aorta showed 2.2- and 3.1-fold higher uptake of ⁶⁴Cu-RYM2 in the AAA group (SUV_max: 1.82 ± 0.51, n = 10) compared to Ang II-infused Apoe^−/− mice with low remodeling (SUV_max: 0.83 ± 0.17, n = 13) and control animals (SUV_max: 0.59 ± 0.05, n = 6, P = 0.004 and P < 0.0001 respectively, Fig, 4D). No difference in blood activity, whether quantified in the left ventricle on PET/CT images or by gamma counting of blood samples collected at the time of euthanasia, was observed between the 3 groups of animals. Similarly, the biodistribution of the tracer in other tissues, as assessed by ex vivo gamma counting, was similar between the 3 groups of animals (Table S2).

Figure 4.

In vivo ⁶⁴Cu-RYM2 PET/CT imaging. (A-C) Examples of PET/CT images of control Apoe^−/− mice (Con, A), and angiotensin II-infused Apoe^−/− mice with low remodeling of the abdominal aorta (LR, B) or abdominal aortic aneurysm (AAA, C) in transverse, coronal, and sagittal views (from left to right) at 50–60 min after injection of ⁶⁴Cu-RYM2. White arrows point to ⁶⁴Cu-RYM2 uptake in AAA; white arrowheads point to the aorta; red arrows indicate the location of the transverse section. (D) Quantification of ⁶⁴Cu-RYM2 uptake in suprarenal abdominal aorta ** P < 0.01, **** P < 0.0001, Kruskal–Wallis test.

To establish the specificity of ⁶⁴Cu-RYM2 uptake in vivo, a group of Ang II-infused animals underwent PET/CT on two consecutive days. The second day imaging in a subgroup (n = 4) of animals was performed following administration of 500-fold excess of cold Cu-RYM2. Evaluation of the suprarenal abdominal aortic PET signal showed a significant 59% reduction in ⁶⁴Cu-RYM2 signal with in vivo blocking (SUV_max: 1.87 ± 0.56 vs 0.77 ± 0.09, n = 4, P = 0.03), to levels comparable to control animals (Figure 5). The remaining two animals underwent a PET/CT acquisition on two consecutive days in basal conditions and showed comparable PET signal in AAA on both days (SUV_max: 2.52 and 2.58, and 2.13 and 2.06 for the two animals on consecutive days).

Figure 5.

Specificity of the ⁶⁴Cu-RYM2 signal in vivo. (A-B) Examples of fused PET/CT images in transverse, coronal and sagittal views (from left to right) of an angiotensin II-infused Apoe^−/− mouse at 50–60 min after the injection of ⁶⁴Cu-RYM2 alone (A) or with pre-injection of 500-fold excess non-radioactive Cu-RYM2 (B). White arrows point to ⁶⁴Cu-RYM2 uptake in abdominal aortic aneurysm; grey arrows indicate the corresponding location in the blocking conditions; red arrows indicate the location of the transverse section. (C) Quantification of ⁶⁴Cu-RYM2 signal in suprarenal abdominal aorta. * P < 0.05, Paired t test.

A second animal model, where carotid aneurysm is induced through local CaCl₂ application, was used to confirm ⁶⁴Cu-RYM2 specificity. In this model, higher ⁶⁴Cu-RYM2 uptake was detected by autoradiography in the aneurysmal left carotid artery than the control, saline-treated right carotid artery. Along with decreasing ⁶⁴Cu-RYM2 uptake in several other organs (e.g., skin, stomach, and duodenum), pre-injection of an excess of RYM inhibitor led to 89% reduction in the tracer’s uptake in the aneurysmal tissue at 2 h post-injection (P < 0.0001, Figure S16).

Aortic ⁶⁴Cu-RYM2 signal on in vivo PET images correlates with ex vivo-determined tissue MMP activity and inflammation

Zymographic evaluation of the suprarenal abdominal aortic tissue showed significantly higher MMP activity in the AAA group compared to low remodeling and control groups (P = 0.01 and P < 0.0001, respectively), and higher MMP activity in the low remodeling group compared to control animals (P = 0.02, Figure 6A). This was associated with higher aortic expression of Mmp2, Mmp12 and Mmp13, as detected by RT-PCR in AAA compared to control animals (P < 0.0001, P = 0.002 and P = 0.005, respectively), a higher expression of Mmp2 and Mmp13 in low remodeling compared to control animals (P = 0.049 and P = 0.008, respectively), and a higher expression of Mmp2 and Mmp12 in AAA compared to the low remodeling group (P = 0.04 and P = 0.03, respectively, Figure 6B), while no difference in Mmp3 and Mmp9 gene expression was observed between different experimental groups (Figure S17). Importantly, the aortic ⁶⁴Cu-RYM2 signal on in vivo PET/CT images correlated with ex vivo measured aortic MMP activity quantified by zymography (P = 0.0007, Figure 6C), as well as Mmp2, Mmp12 and Mmp13 gene expression (P < 0.0001, P < 0.0001, P = 0.004, respectively, Figure 6D), but not Mmp3 and Mmp9 gene expression (Figure S18).

Figure 6.

Tissue characterization and correlates of ⁶⁴Cu-RYM2 signal in suprarenal abdominal aorta. (A) Aortic matrix metalloproteinase (MMP) activity. (B) Aortic Mmp2, Mmp12 and Mmp13 gene expression. (C) Correlation between aortic ⁶⁴Cu-RYM2 signal and ex vivo measured aortic MMP activity (D), and Mmp2, Mmp12 and Mmp13 gene expression. (E) Aortic Cd68 gene expression, (F) Correlation between aortic ⁶⁴Cu-RYM2 signal and Cd68 gene expression. Gene expression data are expressed as 2^−ΔCT normalized to Acta2. Sample size: Con n = 5–6, LR n = 12, AAA n = 5–6. * P < 0.05, ** P < 0.01, **** P < 0.0001, One-way analysis of variance (A, E) or Kruskal–Wallis test (B).

Finally, evaluation of Cd68 as a marker of vessel wall inflammation showed significantly higher expression in the AAA group compared to low remodeling and control animals (P < 0.0001 for both, Figure 6E). Consistent with the role of inflammatory cells as a major source of MMP activity, there was a significant correlation between the aortic ⁶⁴Cu-RYM2 signal on in vivo PET/CT images and Cd68 gene expression (P < 0.0001, Figure 6F and Table S3).

⁶⁴Cu-RYM2 specifically binds to human aortic tissue in correlation with MMP activity and inflammation

As a prelude to in vivo human studies, binding of ⁶⁴Cu-RYM2 to human AAA and non-aneurysmal, control aortic tissue homogenates was evaluated by gamma counting. After normalization to protein content, ⁶⁴Cu-RYM2 exhibited significantly higher binding to AAA compared to control aortic tissue homogenates (P = 0.002, Figure 7A), which was significantly reduced in the presence of an MMP inhibitor, GM6001(P = 0.005, Figure 7B). Using adjacent tissue to evaluate MMP activity by zymography and gene expression analysis, a significant correlation was demonstrated between aortic ⁶⁴Cu-RYM2 tissue binding and MMP activity (r = 0.63, P = 0.02, Figure 7C) and CD68 gene expression (r = 0.82, P = 0.0006, Figure 7D).

Figure 7.

Ex vivo ⁶⁴Cu-RYM2 binding to human aortic tissue homogenates. (A) ⁶⁴Cu-RYM2 binding to control aorta and abdominal aortic aneurysm (AAA). (B) ⁶⁴Cu-RYM2 binding to AAA in the absence or presence of 500-fold molar excess of GM6001. (C-D) Correlation between ⁶⁴Cu-RYM2 binding to aortic tissue homogenates and matrix metalloproteinase (MMP) activity (C) or CD68 gene expression (D). ** P < 0.01, unpaired t test (A) or paired t test (B).

Specific binding of ⁶⁴Cu-RYM2 to human AAA was confirmed by incubating AAA and normal aortic tissue sections to the tracer in the absence or presence of GM6001, followed by autoradiography (Figure 8A and B). Quantification of autoradiography images showed significantly higher ⁶⁴Cu-RYM2 binding to human AAA than to normal aorta tissue sections (P = 0.001, Figure 8C), which was significantly reduced in the presence of GM6001 (P = 0.02, Figure 8D). ⁶⁴Cu-RYM2 binding to aortic tissue sections correlated with MMP activity (r = 0.66, P = 0.02, Figure 8E) and CD68 gene expression (r = 0.83, P = 0.002, Figure 8F).

Figure 8.

Autoradiographic analysis of ⁶⁴Cu-RYM2 binding to human aortic tissue. (A) Examples of Movat pentachrome staining of a normal aorta (left) and an abdominal aortic aneurysm (AAA, right). (B) Examples of autoradiography obtained after incubation of adjacent aortic tissue sections with ⁶⁴Cu-RYM2 alone (upper row), and ⁶⁴Cu-RYM2 in presence of 500-fold molar excess of GM6001 (bottom row). (C) Autoradiographic quantification of ⁶⁴Cu-RYM2 binding to normal aortae and AAA. (D) Autoradiographic quantification of ⁶⁴Cu-RYM2 binding to AAA in the absence or presence of 500-fold molar excess of GM6001. (E-F) Correlation between ⁶⁴Cu-RYM2 binding quantified by autoradiography and aortic tissue MMP activity (E) and CD68 gene expression (F). * P < 0.05, ** P < 0.01, Mann-Whitney test (A) or Wilcoxon test (B). L: luminal side. Scale bar: 5 mm.

Discussion

Given the central role of proteases in the pathogenesis of aortic aneurysms, the unmet clinical need to improve AAA risk stratification can potentially be addressed by in vivo molecular imaging of MMP activity. As a key step toward this objective, here we report the design, synthesis, in vitro characterization, and in vivo evaluation in murine AAA of a broad-spectrum MMP inhibitor, RYM2, and the corresponding radiotracer, ⁶⁴Cu-RYM2. Specific binding of ⁶⁴Cu-RYM2, which correlated with MMP activity in murine aorta in vivo and human aortic tissue including AAA specimens ex vivo, signifies the potential of ⁶⁴Cu-RYM2 for MMP imaging of AAA in humans.

Among different imaging modalities, nuclear techniques have the unique advantage of high sensitivity for in vivo, non-invasive molecular cardiovascular imaging. This sensitivity is especially important for vessel wall imaging, where the target, even in the case of AAA, is relatively small. For human imaging, PET has a better spatial resolution and sensitivity compared to SPECT and is more quantitative in nature. This is in contrast with small animal imaging, where the spatial resolution of microSPECT (which can reach 0.25 mm) is more favorable compared to microPET, where the best spatial resolutions that can be achieved for ¹⁸F and ⁶⁴Cu are in the 1–2 mm range. With the goal of clinical translation in mind, we designed RYM2 as a novel MMP inhibitor and radiotracer precursor for ⁶⁴Cu labeling. The selection of ⁶⁴Cu was motivated by the longer half-life of ⁶⁴Cu (12.7 h) compared to ¹⁸F (1.8 h), which would eventually facilitate off-site production and overnight shipping of the tracer to imaging centers, at a cost of potentially higher exposure for the patients. Dosimetry studies with ⁶⁴Cu-RYM2 remain to be carried out. Additionally, metal chelation chemistry enables efficient radiolabeling using a limited amount of precursors. The positron energy of ⁶⁴Cu is comparable to ¹⁸F and therefore their resolution loss due to positron free path are comparable; in this regard, ⁶⁴Cu is much more favorable compared to gallium-68, another radiometal commonly used for PET radiotracers. A limitation of ⁶⁴Cu is a suboptimal branching fraction for positron emission, requiring a tradeoff between image quality and radiation exposure.

The RYM2 design as a hydroxamate with an anti-succinate-based macrocycle scaffold, was based on the information gained from similarly structured ¹¹¹In- and ^99mTc-labeled macrocyclic hydroxamate MMP tracers for SPECT in preclinical imaging cardiovascular disease models,^16–20 including aneurysms.^{13, 21} Previous work identified the P3′ site as a potential location for introducing a linker and chelator for radiolabeling.¹⁵ NOTA was selected as the chelator, as chelation by NOTA is an efficient approach to radiolabeling with ⁶⁴Cu,^{22, 23} and NOTA has been used in several copper-based radiopharmaceuticals evaluated with a good safety profile in humans.^{24, 25} Copper chelation on the NOTA group of RYM2 was efficient using a simple single-step method and the resulting tracer, ⁶⁴Cu-RYM2, showed excellent stability in vitro, including in murine and human blood. Notably, in vivo evaluation of tracer stability revealed a modest amount of radiometabolite with the retention time of free copper in mouse blood and urine. The presence of radiometabolites is not uncommon for PET tracers, and this relatively low level, if confirmed in humans, should not affect the performance of ⁶⁴Cu-RYM2 in humans. Incorporation of arginine, which is positively charged at physiologic pH, on the P3’ substituent, was previously shown to improve blood clearance on a related ^99mTc-labeled tracer,¹³ and resulted in a similarly rapid blood clearance of ⁶⁴Cu-RYM2 in mice, allowing in vivo imaging under 1 hour post-tracer administration. As expected from this design, the affinities of Cu-RYM2 toward a select panel of human recombinant MMPs are favorable for in vivo imaging and comparable with the values previously reported for analog compounds used successfully for cardiovascular imaging in animals.^{13, 16} In this regard, Cu complexation slightly improved RYM2 affinity for different MMPs. While the reason behind this phenomenon is beyond the scope of this report, it may be explored in future studies to improve tracer affinity, if needed.

Hydroxamate MMP inhibitors were initially introduced as therapeutic agents. However, following initial evaluations in clinical trials, their use as therapeutic agents was hampered by musculoskeletal toxicity following chronic utilization.²⁶ The use of this class of molecules at trace amounts in imaging studies is not expected to be associated with any adverse reaction. Accordingly, preliminary toxicology of RYM2, showed no adverse effect in mice after the administration of the equivalent of 100-fold expected human imaging dose converted based on body surface area.

⁶⁴Cu-RYM2 PET/CT in vivo showed a focal signal overlaying the AAA in Ang II-infused mice, significantly higher than the signal observed in suprarenal abdominal aorta in Ang II-infused animals without a distinct AAA or control mice. Pre-administration of 500-fold non-radioactive ^63/65Cu-RYM2 led to a significant decrease in the AAA PET signal, to levels comparable to the signal in animals without AAA, establishing the specificity of the ⁶⁴Cu-RYM2 signal in AAA. Blocking experiments were reproduced in a second murine model of aneurysm. This was associated with reduced ⁶⁴Cu-RYM2 uptake in several other organs such as the skin, stomach, and duodenum, which reflects a component of baseline MMP activity and thus, specific ⁶⁴Cu-RYM2 uptake. In the case of the kidneys, altered tracer clearance may have also contributed to the reduction in ⁶⁴Cu-RYM2 uptake in this organ. Consistent with the low levels of MMP activity in normal arteries, the baseline aorta signal in the absence of AAA is non-specific, and the multiple folds difference in the ⁶⁴Cu-RYM2 signal is sufficient for distinguishing AAA from normal aorta. Importantly, our data showed a significant correlation between the ⁶⁴Cu-RYM2 signal and MMP activity of the aorta. In addition to inherent limitations of microPET image quantification stemming from partial volume effects due to the mismatch between spatial resolution and the size of the aorta, the location of AAA in the vicinity of the kidneys, where there is a very high signal due to the tracer’s route of clearance (and to a lesser extend in the liver) may affect the accuracy of signal quantification. To overcome potential interference from the kidney signal, we focused on the signal which could be clearly distinguished from adjacent structures. We cannot exclude the possibility that some segments of aorta with high activity were omitted during image quantification because of their proximity to kidneys. This might have led to underestimation of the correlation between MMP activity, determined ex vivo, and tracer uptake quantified on PET images.

Ang II-induced aortopathy in Apoe^−/− mice was first described as an experimental model of aortic aneurysm, subsequently as dissection, and most recently as pseudoaneurysm.²⁷ Regardless of the pathological lesion, inflammation, macrophage recruitment, and protease production occur within days of initiating Ang II and are necessary for remodeling of the vessel wall. In line with previous reports,²⁸ we found several members of the MMP family to be upregulated in AAA. While MMP activity is regulated at several levels, including MMP gene expression and the presence of tissue inhibitors, a strong correlation was found between aortic ⁶⁴Cu-RYM2 uptake and Mmp2 and Mmp12 gene expression. Given the role of MMP-12 (macrophage elastase) in the degradation of elastic laminae, it is possible that MMP-12 expression may relate to AAA evolution. In this regard, the relation between MMP activity detected by molecular imaging in vivo and AAA outcome has been demonstrated in Ang II-infused mice using an analog SPECT tracer.²¹ Given its relation to vessel wall MMP activity, the ⁶⁴Cu-RYM2 signal is expected to similarly be predictive of AAA evolution in this preclinical model. Ultimately, the performance of ⁶⁴Cu-RYM2 PET in predicting AAA expansion rate or rupture in patients with AAA is the key factor for its clinical application and will need to be addressed in future clinical trials. As a prelude to such human studies, it was confirmed that ⁶⁴Cu-RYM2 binds specifically to human aortic tissue in correlation with tissue MMP activity.

To date, several molecular imaging methods, including ultrasmall superparamagnetic iron oxide-enhanced magnetic resonance imaging (to detect inflammation based on phagocytosis of the particles by macrophages),²⁹ PET imaging with ¹⁸F-fluorodeoxyglucose (to detect inflammation based on enhanced glucose uptake in activated macrophages)³⁰ and ¹⁸F-sodium fluoride (to detect microcalcification)³¹ have been tested in clinical trials. Among these, PET imaging with ¹⁸F-sodium fluoride appears to be the most promising for prediction of AAA outcome independent of other clinical predictors such as AAA diameter, but this observation should be fully validated in additional studies and remains mostly unexplained based on the current knowledge on the role of calcification in AAA. Given the direct role of MMPs in the pathogenesis of AAA,^{10 64}Cu-RYM2 PET imaging might be a superior approach to assessing the risk of AAA rupture. Beside the direct role of MMPs in extracellular matrix remodeling, it is worth noting that the ⁶⁴Cu-RYM2 signal, which reflects MMP activity, correlated well with vessel wall inflammation, both in vivo in murine AAA and ex vivo using human aorta tissues.

Future clinical studies should establish whether ⁶⁴Cu-RYM2 PET will provide substantial benefit compared to the current AAA risk assessment paradigm and be cost-effective. The information reported here regarding the tracer’s stability, biodistribution, performance in preclinical AAA, and importantly, specific binding to human tissues justifies the testing of this agent in clinical studies. If validated in human AAA, ⁶⁴Cu-RYM2 PET could also be of potential value in many other clinical settings, where MMP activity and inflammation play a role.

Clinical Perspective.

Current guidelines recommend screening for abdominal aortic aneurysm (AAA), especially in men aged 65 or more with a history of smoking. Those with AAA are followed with regular imaging studies to track the aneurysm size and are referred for elective endovascular or open repair based on a threshold AAA size, AAA growth rate or symptoms. However, 10 to 15% of all ruptures occurs in patients that do not meet the criteria for repair. The mortality and morbidity from aneurysm repair are increased in the presence of co-morbidities. As such, a non-invasive method for better assessment of the risk of rupture can improve the management of patients with AAA, as those with high-risk AAA, even if they do not meet the current criteria, could be referred for elective repair, while intervention might be avoided in patients with large, low risk AAA. Our current understanding of the AAA pathophysiology suggests that matrix metalloproteinase (MMP) activity may be a determinant of AAA susceptibility to rupture. Therefore, the detection and quantification of MMP activity by positron emission tomography (PET) imaging might be a powerful tool for AAA risk stratification. To this end, here we have introduced a novel PET tracer, ⁶⁴Cu-RYM2, and validated it for PET imaging of MMP activity in murine models of aneurysm. If validated in humans, ⁶⁴Cu-RYM2 PET may help improve the outcome of patients with AAA.

Non-standard Abbreviations and Acronyms:

AAA

abdominal aortic aneurysm

Ang II

angiotensin II

(CE)CT

(contrast-enhanced) [X-ray] computed tomography

HPLC

high performance liquid chromatography

LC-MS

liquid chromatography–mass spectrometry

MMPs

matrix metalloproteinases

NMR

nuclear magnetic resonance

NOTA

2,2’,2”-(1,4,7-triazacyclononane-1,4,7-triyl)triacetic acid

PCR

polymerase chain reaction

SPECT

single photon computed tomography

SUV

standardized uptake value