Scintigraphic Imaging of Matrix Metalloproteinase Activity in the Arterial Wall In Vivo
Background— Matrix metalloproteinases (MMPs) are enzymes involved in the proteolytic degradation of extracellular matrix. They play an important role in several disease processes, such as inflammation, cancer, and atherosclerosis.
Methods and Results— In this study, we have used the broad-spectrum MMP inhibitor CGS 27023A to develop the radioligand [123I]I-HO-CGS 27023A for in vivo imaging of MMP activity. Using this radioligand, we were able to specifically image MMP activity by scintigraphy in vivo in the MMP-rich vascular lesions that develop after carotid artery ligation and cholesterol-rich diet in apolipoprotein E-deficient mice. These results were confirmed by gamma counting of lesional tissue (counts per minute per milligram).
Conclusions— Imaging of MMP activity in vivo is feasible using radiolabeled MMP inhibitors. Additional studies are needed to test the potential of this approach as a novel noninvasive clinical diagnostic tool for the management of human MMP-related diseases.
Received November 26, 2003; de novo received March 11, 2004; accepted March 11, 2004.
Matrix metalloproteinases (MMPs) are a family of endopeptidases that degrade extracellular matrix components. They play an important role in different physiological (eg, organ development) and pathological (eg, wound healing, inflammation, and cancer) processes.1 MMP activity has been implicated both in the development and clinical complications of atherosclerosis attributable to plaque rupture, such as myocardial infarction and stroke. Dysregulation or misexpression of MMPs in the atherosclerotic lesion may result in mechanical destabilization and rupture of the atherosclerotic plaque, potentially leading to thrombosis and vessel occlusion with life-threatening clinical sequelae.2 Despite primary and secondary prevention, such events still account for one third of all deaths worldwide and constitute a major source of disability and healthcare costs.3 Therefore, identifying individual patients with a high risk of plaque rupture is an important challenge in clinical medicine. Several noninvasive techniques, such as MRI, multislice computed tomography, and electron beam tomography, are currently being tested for their applicability to identify such patients by assessing morphological criteria associated with high risk of atherosclerotic plaque rupture. In contrast, noninvasive scintigraphic techniques such as SPECT and PET make use of radiolabeled molecules (radioligands) designed to specifically target individual metabolic or enzymatic activities involved in a particular molecular process. This allows the imaging of disease processes in vivo with high intrinsic sensitivity (nanomolar/picomolar range) and specificity.4 Several subtype-selective or broad-spectrum MMP inhibitors are currently being developed for treatment and assessed in clinical trials for several applications.5 We have investigated the feasibility of scintigraphic imaging of MMPs in vivo using a radiolabeled broad-spectrum MMP inhibitor in an established animal model of arterial remodeling and lesion development where MMPs are induced and activated.
Synthesis of [123I/125I]I-HO-CGS 27023A
The broad-spectrum MMP inhibitor CGS 27023A (Figure 1a) was synthesized de novo, as recently described,9 and modified for labeling with [123I]iodine or [125I]iodine; the methoxy moiety of the lead compound CGS 27023A was substituted with the hydroxy group as previously published.6 This compound was then radioiodinated by a slightly modified standard procedure,7 resulting in [123I]I-HO-CGS 27023A (Figure 1b; specific activity, 16.2 GBq/μg) or [125I]I-HO-CGS 27023A (specific activity, 0.2 GBq/μg), respectively. The radioligands were identified by analytical reversed-phase high-pressure liquid chromatography with ultraviolet and γ detection using I-HO-CGS 27023A as a reference compound, which was prepared by the iodination of the hydroxy derivative with iodine monochloride.8 Chemical and radiochemical purities were >95%. Details of the synthesis of the radioligands as well as biodistribution studies in mice have been recently published.9
Apolipoprotein E-deficient mice and wild-type mice (Charles River, Belgium) (CL 57/BL6, 8 weeks old, 20 to 28 g) were anesthetized by intraperitoneal injection of xylazine/ketamine, and the left common carotid artery was ligated near the bifurcation using 5-0 silk (Ethicon).10 In sham-operated controls, the thread was passed under the left common carotid artery without tightening. The animals were given a high-cholesterol diet (15% cocofat, 1.0% cholesterol, 0.5% sodium-cholate) for 4 weeks. The protocol was approved by the local university committee on Institutional Animal Care. For histological and autoradiographic analysis, anesthetized animals were perfused with Langendorff buffer for 3 minutes and the left carotid artery was removed and snap-frozen in liquid nitrogen. Serial groups of 5 sections (1×60 μm for microautoradiography, 4×10 μm for histopathology) were collected.
Histology and Immunohistochemistry
Serial cryostat cross-sections (10 μm) of the ligated left common carotid artery were air dried and fixed for 10 minutes in 3.75% paraformaldehyde and acetone (4°C) for MMP-9 (rabbit anti-mouse MMP-9, Chemicon) and macrophage (rat anti-mouse Mac-3, BD Pharmingen) immunostaining, respectively. The control antibodies used were rabbit IgG (DAKO) and rat IgG1 (BD Pharmingen). Detection was performed with biotinylated goat anti-rabbit IgG (H+L) (Chemicon) and mouse anti-rat IgG1/2A (BD Pharmingen), respectively, followed by streptavidin-horseradish peroxidase (DAKO) and amino-ethyl-carbazole (K0696, DAKO) each for 25 minutes and a hematoxylin counterstain.
The synthetic broad-spectrum fluorogenic MMP substrate (7- methoxycoumarin-4-yl) acetyl-pro-Leu-Gly-Leu-(3-(2,4-dinitrophenyl)-l-2,3-diamino-propionyl)-Ala-Arg-NH2 (R&D Systems) was used to measure the activity of active human recombinant MMP-2 and MMP-9 (Calbiochem) in the presence and absence of CGS 27023A and I-HO-CGS 27023A.11 MMP-2 (1 nmol/L) and MMP-9 (2 nmol/L), respectively, were preincubated with 10 pmol/L to 1 mmol/L of each inhibitor in 50 mmol/L Tris-HCl (pH 7.5), 0.2 mol/L NaCl, 5 mmol/L CaCl2, 20 μmol/L ZnSO4, and 0.05% Brij 35 at 37°C for 30 minutes. An aliquot of substrate (5 μmol/L, 10 μL) was added to 90 μL of the MMP/inhibitor mixture, and MMP activity was determined by measuring fluorescence at 37°C over time using a Fusion Universal Microplate Analyzer (Packard Bioscience) with excitation and emission wavelengths of 330 and 390 nm, respectively. Inhibition rates were determined from the initial 10 minutes of the reaction, where product release was linear with time. Nonlinear regression analysis was performed using dedicated software (Grace 5.1.8).
In Vivo Microautoradiography
Four carotid-ligated apolipoprotein E-deficient mice were injected retroorbitally with 20 MBq of [125I]I-HO-CGS 27023A in 200 μL 0.9% NaCl and euthanized after 2 hours. The left common carotid artery was quickly removed and frozen, and cross-sections (60 μm) were processed for microautoradiography as described.12 Briefly, slide-mounted carotid artery sections were post-fixed in paraformaldehyde at 80°C for 2 hours before dipping into Hypercoat emulsion LM-1 (Amersham Bioscience Ltd) at 42°C under darkroom conditions. Carotid slices were exposed in light-proof boxes for 8 days at 4°C. Subsequently, the emulsion was developed (D19 developer, Kodak Ltd) for 5 minutes and fixed in IF 23 (Ilford Ltd) for 10 minutes. Finally, slides were rinsed in running tap water for 20 minutes and air dried.12
In Vivo Imaging
A total of 11 apolipoprotein E-deficient mice with carotid ligation, 4 wild-type mice with carotid ligation, and 2 sham-operated apolipoprotein E-deficient mice were anesthetized by inhalation of isoflurane (2%)/oxygen for the entire length of the imaging studies. [123I]I-HO-CGS 27023A (5 to 10 MBq in 200 μL 0.9% NaCl) was injected retroorbitally and imaged dynamically on a planar gamma camera (Siemens Multispect 3, UHR collimator, matrix 256×256, zoom 2.0) for 120 minutes post injectionem (p.i.) To additionally test the specificity of radioligand uptake, we repeated the in vivo imaging in a subgroup of 6 apolipoprotein E-deficient mice with carotid ligation, which were given an additional injection of 200 μL of 6 mmol/L unlabeled CGS 27023A 2 hours before the injection of [123I]I-HO-CGS 27023A. Dynamic imaging data were summed to 10-minute frames, circular regions of interest (ROIs) were drawn over the maximum activity in the ligation area for all studies, and additional ROIs were drawn over the liver, kidneys, bladder, brain, and thorax. For each ROI, decay-corrected time-activity curves were calculated and the uptake of the radioligand was quantified as percentage injected dose (% ID).
Ex Vivo Tissue Counting
Three carotid-ligated apolipoprotein E-deficient mice were injected intravenously with 10 MBq of [123I]I-HO-CGS 27023A in 200 μL 0.9% NaCl and euthanized after 2 hours. In these, the left common carotid artery, which holds the lesion, was microsurgically precisely excised, dabbed dry, and weighed on a microbalance (Mettler-Toledo M3, Mettler-Toledo GmbH). We then counted the radioactivity of the sample in a gamma counter (counts per minute [cpm]; Wallac 1480 Wizard). The same procedure was performed with the right carotid as a control. We then calculated cpm per milligram of tissue for all samples. In a second series of experiments, we predosed another 3 carotid-ligated apolipoprotein E-deficient mice by an additional injection of 200 μL of 6 mmol/L unlabeled CGS 27023A 2 hour before the injection of [123I]I-HO-CGS 27023A and repeated the above procedure.
Values are expressed as mean±SEM. Differences of the tracer uptake in the lesion between the different experiments were tested using one-way ANOVA followed by post-hoc analysis (Bonferroni). Differences between groups in the ex vivo tissue-counting experiments were tested by unpaired Student’s t tests. P<0.05 was considered significant.
Synthesis and In Vitro Characterization of the MMP Radioligand [123I]I-HO-CGS 27023A
We modified and radioiodinated the broad-spectrum MMP inhibitor CGS 27023A (a hydroxamate derivative), which binds only to the active catalytic domain of MMPs,13 to produce [123I]I-HO-CGS 27023A (Figures 1a and 1b). To test the efficacy of the radioligand for MMP binding and inhibition, we performed fluorogenic in vitro enzymatic activity assays for MMP-2 and MMP-9 with unlabeled I-HO-CGS 27023A in comparison to the lead compound CGS 27023A (Figures 1C and 1D). Nonlinear regression analysis of the concentration-response curves yielded nanomolar IC50 values for MMP inhibition (I-HO-CGS 27023A, 298 and 153 nmol/L for MMP-2 and MMP-9, respectively; CGS 27023A, 11 and 59 nmol/L for MMP-2 and MMP-9, respectively). These results suggested that [123I]I-HO-CGS 27023A may be a suitable radioligand for in vivo imaging of MMP activity.
In Vivo Imaging of MMP Activity in a Murine Model of Arterial Remodeling and Vascular Lesion Development
The ligation of the carotid artery in the apolipoprotein E-deficient mouse is an established model of arterial remodeling because of the development of macrophage- and MMP-rich vascular lesions.10 We have chosen this animal model to produce a single precisely localized vascular lesion that expresses10 and depends on active MMPs14 and that, therefore, may be suitable for imaging MMP activity in the vasculature in vivo. To test this, we injected 5 to 10 MBq of [123I]I-HO-CGS 27023A 4 weeks after carotid ligation and cholesterol-rich diet in apolipoprotein E-deficient mice and measured the uptake of the compound in the carotid lesions by planar scintigraphy (n=11). We detected a steadily increasing specific uptake of the radioligand in the carotid lesion during the first 2 hours after injection (Figure 2a), which was abolished by predosing with an excess of unlabeled CGS 27023A (6 mmol/L; n=6) 2 hours before the experiment (Figures 2a and 2b). Neither sham-operated apolipoprotein E-deficient mice nor carotid-ligated wild-type mice showed any specific uptake in the region of interest (Figures 2c and 2d). These results suggested a specific uptake of the radiolabeled MMP inhibitor in the intimal lesion developing after carotid ligation and cholesterol-rich diet in the apolipoprotein E-deficient mouse. The specific radioactivity signal in the lesion was quantified as percentage injected dose (% ID) (Figure 2e). One-way ANOVA analysis confirmed a significantly higher lesional uptake in unblocked mice compared with predosed mice at imaging time points ≥80 minutes p.i. (Figure 2e). Biodistribution studies revealed rapid clearance of the radioligand from the blood via renal and hepatic excretion and no appreciable signal in the thoracic cavity or brain during the same time (Figure 2f).
Because our imaging approach (planar scintigraphy) cannot differentiate between vascular (carotid lesion) and paravascular accumulation of our radioactive compound, we performed gamma counting of microsurgically excised left (holding the lesion) and right (control) common carotid arteries 2 hours after intravenous injection of 10 MBq of [123I]I-HO-CGS 27023A in 6 apolipoprotein E-deficient mice 4 weeks after carotid ligation and cholesterol-rich diet. Half of these mice were predosed with an excess of unlabeled CGS 27023A (6 mmol/L) 2 hours before the experiment. Weight-corrected radioactivity was significantly higher (2.72-fold) in the left compared with the right common carotid artery in unblocked mice (299±59 versus 110±34 cpm/mg; P=0.027; Figure 3). Furthermore, uptake in the left common carotid artery was significantly blocked by predosing with cold CGS 27023A, whereas there was no difference in uptake between unblocked and predosed mice in the right carotid (left carotid, 299±59 cpm/mg versus 39±11 cpm/mg; P=0.006; right carotid, 110±34 cpm/mg versus 53±28 cpm/mg; P=0.194) (Figure 3). Thus, it seems that the lesional uptake observed in the in vivo imaging experiments is predominantly localized within the lesion and is specific.
To confirm binding of the radioligand to regions of the lesion with high MMP activity, we performed microautoradiography studies of carotid lesions explanted 2 hours after injection of [125I]iodine-labeled I-HO-CGS 27023A. There was an intense autoradiographic signal in the carotid lesion corresponding to areas rich in MMP-9, as identified by immunostaining of adjacent sections (Figure 4). These results additionally suggested that the scintigraphically measured uptake of [123I]I-HO-CGS 27023A in vivo was attributable to its specific accumulation in MMP-rich areas of the carotid lesion.
In this study, we report the establishment of a novel noninvasive scintigraphic imaging technology for the measurement of MMP activity in vivo using the radiolabeled MMP inhibitor [123I]I-HO-CGS 27023A. We have tested the radioligand in an in vivo model of increased MMP activity, which is the arterial lesion developing in cholesterol-fed apolipoprotein E-deficient mice after ligation of the carotid artery. The lesions are macrophage- and MMP-rich, with MMP-9 as the most prominently expressed enzyme,10 which also causally contributes to lesion development in this model.14 Because I-HO-CGS 27023A is a broad-spectrum MMP inhibitor, we cannot differentiate between different MMP activities in the lesion. However, it may be even preferable to have a broad-specificity MMP radioligand in this context, because it allows imaging of all MMPs activated in the lesion. Furthermore, a broad-specificity radioligand such as [123I]I-HO-CGS 27023A may have potential applications in other experimental settings where different MMPs are expressed simultaneously or where the individual MMPs have not been characterized.
Although radiolabeling of the MMP inhibitor CGS 27023A with 18F and 11C has been recently reported,15 no imaging studies were performed in vivo. In our study, the structural modification of the original CGS compound did not have a major effect on the affinity of its binding to and inhibition of MMP-2 and MMP-9. This is a critical step in the development of modified compounds, because a structurally different (peptide-based) radioiodinated MMP inhibitor that we synthesized had IC50 values in the millimolar range, making it unsuitable for in vivo applications (data not shown). The in vivo specificity of [123I]I-HO-CGS 27023A for MMPs in the developing lesion was demonstrated by the successful blocking of its uptake in the lesion with the original CGS 27023A compound and by the lack of signal in sham-ligated apolipoprotein E-deficient mice and ligated or nonligated wild-type animals, respectively. The latter additionally excludes accumulation of the radiopharmaceutical or free radioactive iodide in the thyroid gland.
In the carotid lesion, 0.38% ID of [123I]I-HO-CGS 27023A is accumulated 2 hours p.i. compared with 0.24% ID after blockade, which corresponds to an approximately 1.5-fold increase in accumulation attributable to specific uptake (in vivo imaging). To additionally investigate this, we microsurgically excised the left carotid (holding the lesion) and right carotid 2 hours after injection of the radioisotope and counted these in a gamma counter. Ex vivo, the left carotid lesional uptake was approximately 3-fold that of the right carotid. Furthermore, this lesional uptake was specific, because predosing with the unlabeled MMP inhibitor greatly suppressed the uptake in the left and had no effect on the right carotid. These experiments additionally prove the specific lesional uptake of the radioisotope as the main source of the radioactive signal measured by scintigraphy in vivo.
In contrast to the ex vivo experiment, we observe a background signal in the in vivo imaging experiment that may be attributable to unspecific uptake in the scar/inflammatory area surrounding the ligated artery after surgery. Therefore, we may be underestimating the specific lesional uptake by in vivo imaging. In native atherosclerosis, where no such surgical artifacts are present, the background activity in the thorax (which would be the expected background area for coronary artery imaging in humans) would be equivalent to the one we observe in nonoperated control mice (0.02% ID of [123I]I-HO-CGS 27023A after 2 hours). This could theoretically increase the target-to-background ratio to approximately 20-fold and enable imaging with a much higher image contrast in native atherosclerosis.
Of major interest is the rapid clearance of the radiotracer from the blood via liver, kidneys, and bladder in biodistribution studies, favorably increasing the target-to-background ratio as well as the low background signal in the brain and the thorax. The latter could be of practical significance because the lack of signal interference from the myocardium or the lungs may potentially allow imaging of MMP activity in lesions of atherosclerosis both in coronary arteries and aorta. An interesting observation is that the radiotracer [123I]I-HO-CGS 27023A still accumulates in the carotid lesions at late times after injection (1 to 2 hours p.i.) despite low blood levels. However, we have no experimental evidence of the pharmacokinetics of the tracer in the microenvironment of the lesion. Both the lesional microcirculation and the lipid accumulation in the cells (and extracellularly) attributable to the high-cholesterol diet may have profound impact on the rate of penetration and local retention of the tracer.
The scintigraphic technique we use is suitable for the quantitative detection of MMP activity in vivo, because the radioactivity signal we detect in the lesion corresponds to the total amount of labeled MMP inhibitor bound to active MMPs. When a dedicated compartmental mathematical model based on the in vivo biodistribution results is used, scintigraphic techniques are able to absolutely quantify levels of target molecules per tissue mass down to the nanomolar/picomolar range, because the injected amount of carrier-free radiopharmaceutical is pharmacologically nonsignificant. Thus, the binding of [123I]I-HO-CGS 27023A to active MMPs is not saturated and correlates linearly to the radioactivity signal.
Recently, an optical imaging approach has been described for noninvasive imaging of MMP activity.16 In contrast to the nonpeptidic MMP inhibitors in our study, the authors have used MMP-2 peptide substrates with quenched near-infrared fluorochromes coupled to a polymer and demonstrated successful detection of MMP activity in subcutaneous tumors in mice. Although optical imaging techniques are an exciting new approach, their application remains restricted to pathological processes close to the body surface (eg, skin tumors), because the near-infrared light applied undergoes serious scattering by even thin overlying tissue layers. In contrast, scintigraphic molecular imaging methods such as SPECT and PET have been used for many years and have received broad applications in several clinical disciplines. Gamma photons with high energy (511 keV for PET) easily penetrate long distances within the human body, and the absolute quantification of the dynamic distribution of the radioligand in vivo has been established and validated.4 Because the appropriate imaging devices are available for both animal and human studies, new radioligands validated in animal models can be consequently tested for diagnostic applications in humans. Furthermore, scintigraphic radioligands are preferably nonpeptidyl small molecules, such as the MMP inhibitor [123I]I-HO-CGS 27023A, which are either authentically radiolabeled (eg, by exchange of 12C with 11C) or only slightly modified (analogously radiolabeled). Their advantage compared with peptide-based compounds is that, in the nanomolar concentrations they are routinely applied, they do not influence the biological systems and are able to trace even rapid metabolic pathways.
In our study, we have shown that imaging of MMP activity is feasible in vivo using radiolabeled MMP inhibitors. Clearly, additional studies and developments are needed to test the potential of this approach as a novel noninvasive clinical diagnostic tool for the management of human MMP-related diseases.1 MMPs have been implicated in the progression of atherosclerotic lesions toward a vulnerable phenotype that is prone to rupture and accounts for the acute clinical complications of the disease. Activated macrophages secrete several MMPs in human atherosclerotic plaques with the result of increased collagenolysis and extracellular matrix degradation.17 One of the hallmarks of vulnerable lesions is the weakening of the fibrous cap,18 and this has been correlated with the increased MMP-9 activity in the vulnerable shoulders of the plaque.19 However, the role of MMPs is much more complex and still not entirely understood, and it is important to emphasize that a conclusive demonstration of a causal role of MMPs in plaque rupture remains to be provided. Additional studies using genetically engineered MMP-deficient mice on an atherosclerosis-susceptible background or plaque-specific overexpression of MMPs are clearly the way to address this. Although no animal model of atherosclerosis allows extrapolation of findings on MMP activity directly to human atherosclerotic disease and its complications, an exciting study would be to test the activity of MMPs using the inhibitor [123I]I-HO-CGS 27023A in a recently-described first animal model of atherosclerotic plaque rupture in the brachiocephalic artery of apolipoprotein E-deficient mice.20 Serial monitoring of MMP activity at rupture-prone lesion sites with [124I]I-HO-CGS 27023A-PET may help to understand the role of MMPs in this process.
In combination with the high-resolution morphological imaging techniques, such as MRI and computed tomography, that are currently being developed for both human and animal studies, the molecular imaging of individual disease parameters such as MMP activity in lesions of atherosclerosis may help design approaches for the prediction and prevention of coronary events attributable to plaque rupture of an individual lesion in an individual patient.
This work was supported in part by a research grant from Amersham Health plc, Amersham, UK. The authors gratefully acknowledge the technical assistance of Christine Bätza, Irmgard Hoppe, Kornelia Parusel, Silke Schröer, Sandra Schröer, and Monika Trub.
↵*These authors contributed equally to this study.
This article originally appeared Online on May 3, 2004 (Circulation. 2004;109:r107–r112).
Galis ZS, Khatri JJ. Matrix metalloproteinases in vascular remodeling and atherogenesis: the good, the bad, and the ugly. Circ Res. 2002; 90: 251–262.
World Health Report 2002. Reducing risks, promoting healthy life. World Health Organization; 2002.
MacPherson LJ, Bayburt EK, Capparelli MP, et al. Discovery of CGS 27023A, a nonpeptidic, potent, and orally active stromelysin inhibitor that blocks cartilage degradation in rabbits. J Med Chem. 2002; 40: 2525–2532.
Schnur RC, Corman ML. Preparation of 17-amino-22-(4′-azido-3′-[125I]iodophenacyl)-17-demethoxygeldanamycin: an ansamycin for photoaffinity labelling. J Labelled Cpd Radiopharm. 1994; 34: 529–535.
Edgar KJ, Falling SN. An efficient and selective method for the preparation of iodophenols. J Org Chem. 1990; 55: 5287–5291.
Ivan E, Khatri JJ, Johnson C, et al. Expansive arterial remodeling is associated with increased neointimal macrophage foam cell content: the murine model of macrophage-rich carotid artery lesions. Circulation. 2002; 105: 2686–2691.
Huang W, Meng Q, Suzuki K, et al. Mutational study of the amino-terminal domain of human tissue inhibitor of metalloproteinases 1 (TIMP-1) locates an inhibitory region for matrix metalloproteinases. J Biol Chem. 1997; 272: 22086–22091.
Hidalgo M, Eckhardt SG. Development of matrix metalloproteinase inhibitors in cancer therapy. J Natl Cancer Inst. 2001; 93: 178–193.
Galis ZS, Johnson C, Godin D, et al. Targeted disruption of the matrix metalloproteinase-9 gene impairs smooth muscle cell migration and geometrical arterial remodeling. Circ Res. 2002; 91: 852–859.