European Journal of Nuclear Medicine and Molecular Imaging

, Volume 34, Supplement 1, pp 1–8

Targeting of matrix metalloproteinase activation for noninvasive detection of vulnerable atherosclerotic lesions


  • Dagmar Hartung
    • School of MedicineUniversity of California
    • Department of RadiologySchool of Medicine
  • Michael Schäfers
    • Department of Nuclear MedicineUniversity of Münster
  • Shinichiro Fujimoto
    • School of MedicineUniversity of California
  • Bodo Levkau
    • Institute of PathophysiologyUniversity of Duisburg-Essen
  • Navneet Narula
    • School of MedicineUniversity of California
  • Klaus Kopka
    • Department of Nuclear MedicineUniversity of Münster
  • Renu Virmani
    • Cardiovascular Pathology
  • Chris Reutelingsperger
    • Cardiovascular Research Institute
  • Leo Hofstra
    • Cardiovascular Research Institute
  • Frank D. Kolodgie
    • Cardiovascular Pathology
    • School of MedicineUniversity of California
  • Jagat Narula
    • School of MedicineUniversity of California

DOI: 10.1007/s00259-007-0435-0

Cite this article as:
Hartung, D., Schäfers, M., Fujimoto, S. et al. Eur J Nucl Med Mol Imaging (2007) 34: 1. doi:10.1007/s00259-007-0435-0



Inflammation plays an important role in vulnerability of atherosclerotic plaques to rupture and hence acute coronary events. The monocyte–macrophage infiltration in plaques leads to upregulation of cytokines and metalloproteinase enzymes.

Matrix metalloproteinases result in matrix dissolution and consequently expansive remodeling of the vessel. They also contribute to attenuation of fibrous cap and hence susceptibility to rupture. Assessment of metalloproteinase expression and activity should provide information about plaque instability.


Vascular remodelingAtherosclerosisThin cap fibroatheromaMatrix metalloproteinaseRadionuclide imaging
Disruption of an atherosclerotic plaque accounts for more than two-thirds of acute coronary events [1, 2]. The plaques vulnerable to rupture demonstrate large necrotic cores and positive remodeling of the vessel. The overlying fibrous cap is markedly attenuated and often harbors intense inflammation (Fig. 1). It has been proposed that the inflammatory process plays a critical role in destabilization of atherosclerotic plaque. Macrophage infiltration, lipid insudation, and macrophage apoptosis are associated with matrix metalloproteinase (MMP) expression and activation in the atherosclerotic plaque [3, 4]. It is believed that MMP expression contributes to smooth muscle and collagen dissolution, which in turn result in plaque cap thinning and expansive vascular remodeling. Therefore, the ability to detect MMP expression should allow identification of atherosclerotic plaques prone to rupture [5].
Fig. 1

Morphologic characteristics of atherosclerotic plaque vulnerable to rupture. a, b Low-power views (×20) of a coronary artery representing a stable plaque (a; FA fibroatheroma) and a vulnerable plaque (b; TFCA thin fibrous cap atheroma). Fibroatheroma consists predominantly of smooth muscle and collagen tissue with small lipid inclusions; these plaques have thick fibrous caps. TFCA comprises a large necrotic core, attenuated fibrous cap, and macrophage inflammation in fibrous caps. c Immunostaining of a TCFA demonstrates intense infiltration of CD68-positive macrophages (Macs) in fibrous cap area (×200). d Significant MMP-9 localization can be seen in the macrophage-rich region of TCFA (×200)

Various noninvasive imaging modalities have been proposed for the detection of the inflammatory component of atherosclerotic plaques [5]. 18F-labeled fluorodeoxyglucose is known to accumulate in the cells with high respiratory burst and hence identifies macrophage-rich plaques [69]. 99mTc-labeled annexin A5 has been employed to identify foam cells, macrophage apoptosis, and intraplaque hemorrhage in patients with carotid vascular disease with recent cerebrovascular accidents [10]. Anti-oxidized LDL antibody fragments target the intra-plaque macrophages which have developed SRA-I and II receptors [11, 12], and radiolabeled nonspecific immunoglobulins target macrophage Fcγ receptors that contribute to lipid ingestion [13]. Expression of receptors for adhesion molecules and chemokines is upregulated on the infiltrating plaque monocytes, and iodinated MCP-1 [14, 15] and VCAM-1 [16] have been shown to target cells of monocyte-macrophage lineage in atherosclerotic lesions. Although these strategies may offer an indirect index of MMP expression, it has become possible to identify MMP activity directly in atherosclerotic lesions.

Role of matrix metalloproteinases in vascular remodeling and fibrous cap disruption

MMPs are Zn2+-requiring enzymes that are secreted by or expressed on the cell surface of various vascular cell types and contribute to extracellular matrix degradation [17]. The MMP family includes more than 25 proteases, which are stratified, based on the substrate specificity and structural homology, into five subgroups: interstitial collagenases (MMP-1, -8, -13), gelatinases or basement membrane degrading MMPs (MMP-2, -9), stromelysins or matrilysins (MMP-3, -7), membrane-type MMPs (MMP-14 to -17), and others (MMP-12) [17]. MMPs play an important role in diverse physiologic processes such as organ development, angiogenesis, and tissue repair, and contribute to pathology including inflammation and cancer [18]. In vascular pathology, MMPs play a role in vascular remodeling, aneurysm formation, post-angioplastic restenosis, progression of atherosclerosis, and plaque destabilization [19]. MMPs are expressed predominantly in macrophages, and also in vascular smooth muscle cells (SMCs), lymphocytes, and endothelial cells [20]. Most MMPs, except MMP-11 and membrane-type MMP, are secreted as inactive pro-forms and enabled extracellularly [17]. The activity of MMPs is regulated by the level of gene induction, the extent of vesicle trafficking and secretion, proteolytic activation of the exocytosed inactive zymogens, and annulment of activity by endogenous tissue inhibitors of MMPs (TIMP) or nonselective proteinase inhibitors [17, 21].

An overexpression of the interstitial collagenases MMP-1, -8, and -13 [2224] and of gelatinases MMP-2 and MMP-9 is observed in human atherosclerotic plaques [24]. Elevated MMP mRNA, protein, and activity level have been found in the cap and core area of the atherosclerotic plaque [23]. MMP-1 overexpression occurs in areas of the fibrous cap with increased circumferential tensile stress [25]. MMP-7 and MMP-12, however, are predominantly localized within the fibrous areas surrounding the necrotic core [26]. Collagen breakdown products are abundantly observed in the lipid-rich atherosclerotic plaques as evidence for enhanced extracellular matrix degradation [27]. On the other hand, elevated circulating MMP levels are observed in patients with acute myocardial infarction and unstable angina [28], and increased MMP-9 serum levels are directly related with cardiovascular mortality [29].

Human monocytes and macrophages produce MMP-9 gelatinase most abundantly [17]. It is upregulated early after contact with the endothelial cells and facilitates macrophage entry through the basement membrane [30]. Once in the arterial subintima, inflammatory cytokines such as IL-1 and TNF-α and oxidized LDL accentuate MMP expression [30]. MMPs have been demonstrated to increase in the process of expansive remodeling and plaque vulnerability [24, 31]. Additionally, there is attenuation of SMC content and thinning of fibrous cap, which renders them susceptible to disruption.

Polymorphism of MMP-3 promoter region is associated with progression of coronary atherosclerosis [32], and epidemiologically a functional gene polymorphism is related to the exaggerated risk of myocardial infarction [33]. Ironically, the dietary administration of broad-spectrum MMP inhibitor (RS-130830) in apolipoprotein E knockout mice with already established plaques did not change the incidence of plaque instability [34]. Comparable results were obtained in the pilot clinical trial using the broad spectrum MMP inhibitor doxycycline (MIDAS) [35], as the treatment did not significantly reduce the incidence of acute coronary events compared with the placebo group.

Radiolabeled inhibitors of active MMP for the detection of MMP activity in atherosclerosis and vascular remodeling

MMP imaging in experimentally induced atherosclerosis in rabbits

For direct imaging of MMPs in atherosclerotic plaques [36] and post-infarct healing myocardium [37], a radiolabeled broad-based inhibitor of MMPs (MPI) has been employed. In the former study, 111In-labeled MPI (RP780; MMP1–3, 7–9, 13; Ki 1–15 nM) (Bristol-Myers Squibb Imaging, North Billerica, MA) was used for noninvasive imaging of experimentally induced atherosclerotic lesions in NZW rabbits; the effect of dietary modification on MMP expression in atherosclerotic plaque was also evaluated [36]. Of 24 NZW male rabbits, atherosclerosis was produced in 16 rabbits by balloon de-endothelialization of the abdominal aorta using a 4F Fogarty embolectomy catheter (12-040-4F; Edwards Lifesciences LLC, Irvine, CA) followed by a high-cholesterol diet (HC) containing 0.5% cholesterol and 6% peanut oil for 4 months. HC was administered continuously or interrupted with normal chow (NC). Among 16 HC animals, HC diet was administered for 4 months (HC4) in six animals, withdrawn after 2 months and replaced by normal chow (HC2+NC2) in four animals, and interrupted in the third and restarted in the fourth month (HC2+NC1+HC1) in six animals. Five unmanipulated rabbits were used as controls and received NC for 4 months (NC4). MPI was labeled with 111In for noninvasive imaging. In addition, three NC4 animals were used for radiotracer biodistribution and blood clearance studies.

For radiolabeling, MPI (150 μg) was added to a shielded and crimped 5-ml vial, dissolved in 2.0 ml sodium ascorbate (0.5 M, pH 6.0), followed by the addition of 185 MBq (5.0 mCi; 10–20 μl) 111InCl3 in 0.05 N HCl [specific activity 0.81 μg/MBq (30 μg/mCi)]. The reaction mixture was allowed to stand at room temperature for 20 min prior to being analyzed by high-performance liquid chromatography. A final concentration of 92.5 MBq/ml (2.5 mCi/ml) was obtained.

111In-labeled MPI (70 ± 11 MBq) was injected intravenously. Planar images were acquired 3 h after radiotracer administration using a symmetrical 15% window centered on 173 keV and 247 keV 111In photopeaks in a 128 × 128 word matrix by a gamma camera (Vertex Plus ADAC). After imaging, animals were sacrificed (sodium pentobarbital 120 mg/kg), and the aorta was explanted and imaged ex vivo. The aorta was then segmented, weighed, and gamma counted (PerkinElmerWallac Inc., Gaithersburg, MD) for calculation of the percent total injected dose per gram tissue (%ID/g).

For further histologic and immunohistochemical investigations, one-half of every aortic segment was snap frozen and the other half was fixed with HEPES-buffered 4% formalin. The tissue was then dehydrated in a graded series of ethanol. Each segment was subdivided into three equidistant sections and embedded on-edge in paraffin. Serial 4-μm-thick sections were cut, mounted on charged slides, and stained with hematoxylin and eosin and Movat Pentachrome elastin stain. Representative tissue samples were obtained which demonstrated high and low quantitative MPI uptake for pathologic characterization of MMP expression. Atherosclerotic lesions were characterized according to criteria established by the American Heart Association (AHA) [3840]; only type II–IV lesions were observed. For identification of SMCs a primary antibody against actin isotypes α and β (HHF-35 Enzo, Farmingdale, NY, dilution 1:40, 1-h incubation) was used. The marker RAM-11 (DAKO, dilution 1:200 overnight incubation) was used for localization of macrophages. A biotinylated link antibody directed against mouse using a peroxidase-based kit (LSAB, Dako) was used to label primary antibody. Immunostains were visualized (red reaction product) by an AEC substrate–chromogen system (Dako) and counterstained with Gill’s hematoxylin. For MMP characterization, mouse monoclonal antibodies against human MMP-1 (collagenase-1, IM35L, Oncogene Research Products), rabbit MMP-3 (stromelysin-1, IM45L, Oncogene Research Products), and human MMP-9 (gelatinase-B, IM10L, Oncogene Research Products) were used [41].

The atherosclerotic lesions in the abdominal aorta were visualized best at 3 h after radiolabeled MPI injection in HC4 animals (Fig. 2a–c). No uptake was seen in control animals (Fig. 2d–f) and only minimal tracer uptake was detected in the diet withdrawal (HC2+NC2) and diet interrupted (HC2+NC1+HC1) groups. The HC4 rabbits showed more advanced lesions (AHA type II–IV lesions) with a higher content of neointimal macrophages and low SMC prevalence associated with a significantly higher %ID/g uptake of MPI (lesion-to-nonlesion ratio 11:1) in the atherosclerotic lesions compared with the control animals (0.033 ± 0.021 versus 0.003 ± 0.001; p < 0.0015). The quantitative MPI uptake in the diet withdrawal (HC2+NC2) group was not significantly different from that in the control group (0.011 ± 0.003 versus 0.003 ± 0.001; p=ns), but significantly lower than the uptake in the HC4 group (0.033 ± 0.001; p < 0.007) (Fig. 2b). In the animals which were restarted on diet (HC2+NC1+HC1), the uptake increased to almost 50% (0.015 ± 0.006) of that in HC4 rabbits. Histology showed a significant decrease in macrophage population and increase in SMC content in animals with dietary modification. Histologic stability of plaque paralleled the resolution of MPI uptake in atherosclerotic lesions. Significantly higher MMP-1, -3, and -9 expression was observed in higher MPI uptake segments as compared to lower uptake segments (Fig. 3). In addition, image threshold analysis of histologic sections after immunostaining showed a higher MMP expression in plaque segments demonstrating high MPI uptake (MMP-1 = 12.5 ± 2.1; *MMP-3 = 14.3 ± 1.8; *MMP-9 = 1.3 ± 0.5 mm2) versus those with low MPI uptake (MMP-1 = 7.7 ± 0.2; MMP-3 = 9.1 ± 1.5; MMP-9 = 0.2 ± 0.05 mm2; *p < 0.03). Macrophage expression of MMP-1, -3, and -9 decreased with a lipid-lowering diet such as HC interruption and complete HC diet withdrawal.
Fig. 2

111In-labeled MPI uptake in experimental atherosclerotic lesions. A Planar left decubitus images of an atherosclerotic HC4 rabbit obtained immediately after intravenous MPI administration outline the aortic blood pool activity (a). MPI localizes in the atherosclerotic lesions gradually and can be visualized best at 3 h later (b). Ex vivo image of the explanted aorta confirms in vivo evidence of tracer uptake (c). df Corresponding images of a control rabbit. The aortic blood pool is seen at the time of intravenous MPI injection (d, arrow). In contrast, radiotracer uptake was absent in the abdominal aorta of the control rabbit at 3 h after injection (e, arrows). Ex vivo aortic image of the control animal demonstrates the absence of MPI uptake (f). B Bar graphs showing 111In-labeled MPI uptake as the mean percent injected dose per gram (%ID/g) within the abdominal aorta relative to the study group. The uptake was maximum in HC4 abdominal aorta (p values are shown). The uptake in the diet withdrawal group (HC2+NC4) was not significantly different from that in the NC4 group but lower than that in the diet interruption group (HC2+NC1+HC1) (p=ns)
Fig. 3

Comparison of the scintigraphic and histopathologic results. The upper panel demonstrates Movat’s pentachrome stain (left) and immunohistochemical staining for MMP-1 (middle) and MMP-3 (right) expression (MMP-9 staining is not shown). The quantitative correlation revealed that the higher the MPI uptake, the higher was the macrophage infiltration (RAM-11) and the active MMP-1, -3, and -9 expression but the lower the SMC content (actin). Diet withdrawal decreased MMP expression and macrophage content, but increased the SMC population (data not shown)

It is expected that a precise assessment of the extent of MMP expression would allow noninvasive detection of the plaque vulnerability. Our study demonstrated the feasibility of noninvasive targeting of MMP expression in atherosclerotic lesions. The radiolabeled MPI uptake was associated with histologically verified higher expression of MMP-1, -3, and -9 and correlated with higher macrophage and lower SMC content of the lesion. These results confirm that in vivo quantification of MMPs in atherosclerotic plaques is feasible and correlates with the pathologic distribution of MMPs in plaque. It has been previously demonstrated that lipid lowering by either dietary modification or HMG-CoA reductase inhibitors decreases the macrophage deposition in atheroma [4143]. Our data confirm that a lipid-lowering diet such as HC diet interruption and complete HC diet withdrawal reduces macrophage accumulation and leads to abrogation of MMPs in atherosclerotic lesion.

Radionuclide imaging of vascular lesions in a mouse model of atherosclerosis

A similar synthetic broad-spectrum MPI (CGS 27023A) for characterization of MMP activity has been used in a model of vascular lesion formation and remodeling in mice deficient for apolipoprotein E (apoE−/−). The chosen MPI binds only to the active catalytic domain of MMPs [44]. After modification of the original compound, the MPI (HO-CGS 27023A) was radioiodinated (123I-MPI for SPECT, 124I-MPI for PET, or 125I-MPI for autoradiography) and its imaging properties studied in apoE−/− mice. In these, a localized vascular lesion containing high levels of MMPs was generated by ligation of the left common carotid artery followed by an HC diet for 4 weeks [45]. Sham-operated apoE−/− mice and carotid-ligated wild-type mice on NC were used as controls. For histology and immunohistochemistry, serial cryostat cross-sections (10 μm) of the common carotid arteries were air dried and fixed for 10 min in 3.75% paraformaldehyde, and MMP-9 and macrophage staining was performed using a rabbit anti-mouse MMP-9 antibody (Chemicon) and a rat anti-mouse Mac-3 antibody, respectively (Pharmingen).

Ex vivo microautoradiography was performed in four ligated carotid arteries of apoE−/− mice 2 h after intravenous injection of 20 MBq of 125I-MPI. For processing, the excised carotid artery was frozen and cross-sections were examined. Dynamic in vivo gamma imaging was performed after intravenous injection of 5–10 MBq of 123I-MPI in 11 apoE−/− mice with carotid ligation and HC, four wild-type mice with carotid ligation and NC, and two sham-operated apoE−/− mice with NC (Siemens Multispect 3, UHR collimator, matrix 256 × 256, zoom 2.0). To assess the specificity of the radioligand, predosing experiments were performed by injecting 200 μl of 6 mmol/l unlabeled MPI 2 h before radiotracer administration in six of the 11 apoE−/− mice with carotid ligation. Regions of interest (ROI) were drawn over the areas of the carotid arteries with maximum tracer uptake as well as other body organs for evaluation of the biodistribution in 10-min summed images. After decay correction, time-activity curves were obtained and the uptake of the radioligand was quantified as %ID/g tissue. Ex vivo imaging of the carotid vessels was also performed. The efficacy of 123I-MPI binding to and inhibition of MMP was confirmed with fluorogenic in vitro enzymatic activity assays for MMP-2 and MMP-9.

After intravenous injection, radioligands rapidly cleared from the circulation, allowing plaque visualization by scintigraphic imaging. In vivo imaging demonstrated a steadily increasing specific radioligand uptake in the carotid lesions during the 2-h period after radioligand administration. Uptake of radioligand in the carotid lesion was almost completely abolished upon pretreatment with unlabeled MPI. No significant uptake of the radiolabeled MPI was detected in the contralateral carotid arteries, in the carotid arteries of sham-operated apoE−/− mice, or in the ROI of carotid-ligated wild-type animals. Figure 4 shows a coronal 0.4-mm-thick PET slice through the left carotid lesion with intense uptake of the 124I-MPI into the lesion 30 min after intravenous administration. The specific uptake in the carotid arteries was significantly higher in mice given the radioligand alone than in mice pretreated with an excess unlabeled MPI at time points ≥80 min. These results were confirmed by ex vivo tissue gamma counting of tracer uptake in carotid arteries as well as by microautoradiography studies. The radioactivity in the unblocked lesional carotid areas of apoE-deficient mice was significantly increased compared with the contralateral artery (299 ± 59 versus 110 ± 34 cpm/mg; p = 0.027) and compared with the carotid arteries predosed with unlabeled MPI (299 ± 59 versus 39 ± 11 cpm/mg; p = 0.006). There was no significant difference between the uptake of radioligand in unblocked and blocked contralateral carotid arteries (110 ± 34 versus 53 ± 28 cpm/mg; p = 0.194). The intense autoradiographic signal in the carotid lesions corresponded to MMP-9-rich lesion areas identified by immunostaining (Fig. 5). Studies in animal models of atherosclerosis and vascular remodeling suggest that MMP targeting in vascular lesions and noninvasive in vivo imaging are feasible and could provide a powerful approach for evaluation of the level of inflammation in atherosclerotic plaques.
Fig. 4

Site of ligated left common carotid artery (left panel) and a corresponding whole-body coronal slice (0.4 mm thick) through a left carotid lesion (right panel) 4 weeks after ligation and a HC diet in an apoE−/− mouse. Intense uptake of the radiolabelled broad spectrum MMP inhibitor 124I-HO-MPI in the left carotid lesion (arrow) 30 min after intravenous injection is visible using high-resolution small animal PET
Fig. 5

Ex vivo microautoradiography and MMP immunostaining. Adjacent sections of the left common carotid artery of an apoE−/− mouse 4 weeks after carotid ligation were explanted 2 h following injection of [125I]I-HO-MPI. Top: immunostaining for MMP-9. Bottom: microautoradiography. Arrows indicate corresponding regions of MMP-9 expression and autoradiographic signal

Fluorescent activatable MMP substrates for the detection of MMP expression in atherosclerosis

Metalloproteinase expression in an atherosclerotic lesion has also been targeted by the use of MMP substrates [46]. In such an approach, an activatable probe is used that becomes detectable only after it is cleaved by the active proteinase [47]. For this purpose, a near-infrared fluorescence (NIRF) probe was used to detect enzymatic cleavage by gelatinases, MMP-2, and MMP-9 in atherosclerotic aortas of hypercholesterolemic apoE−/− mice [46]. ApoE−/− mice were fed 1.25% HC diet for 12 weeks; age-matched wild-type mice, fed NC, were used as control. The probe consists of a gelatinase substrate sequence, GGPRQITAG, that is amenable to similar cleavage by MMP-2 or MMP-9. After gelatinase-based cleavage, the fluorescence of the probe is increased by 200-fold at excitation 675 nm and emission 694 nm. Twenty-four hours after intravenous administration of the gelatinase-imaging probe, ex vivo NIRF reflectance imaging of excised mice aortas (n = 19) was performed, in visible light and NIRF spectra with a 2-min acquisition captured by a 12-bit monochrome CCD camera (Kodak, New Haven, CT) equipped with an f/1.2 12.5- to 75-mm zoom lens and an emission long-pass filter at 700 nm (Omega Optical, Brattleboro, VT). The specificity of the probe was established with a gelatinase inhibitor (MMP-2/9 Inhibitor III, Calbiochem, EMD Biosciences, Inc, Darmstadt, Germany).

Visible light images showed atherosclerotic lesions in aortas of apoE−/− mice. No lesions were observed in wild-type mice. NIRF images of gelatinase substrate probe demonstrated signals in atherosclerotic aortas of apoE−/− mice but not in wild-type mouse aorta. Pretreatment with the gelatinase inhibitor for 8 h before probe administration abrogated NIRF signal in atherosclerotic aorta. The NIRF reflectance areas colocalized with macrophage accumulation and with MMP-2 and MMP-9 expression in macrophage-rich lesions, but not with intimal smooth muscle cells or endothelial cells (Fig. 6). It is expected that this technique will become useful for imaging of MMP expression/activity in superficial peripheral arteries noninvasively or in coronary or deep-seated vessels by catheter-based detectors.
Fig. 6

Correlation of NIRF signals ex vivo and MMP-9 expression and macrophage accumulation in situ. Aortas from wild-type mice yielded negligible fluorescence signal (left sample in NIRF reflectance image, top left, i) and had undetectable levels of immunoreactive MMP-9 (panel i). However, atherosclerotic aorta of HC-fed apoE−/− mice with fluorescence signals (right sample in NIRF reflectance image, top left, ii and iii) colocalized with MMP-9 accumulation in the intima sampled at the indicated sites (panels ii and iii). Macrophages constitute the major source of NIRF signals for gelatinase action in mouse atheroma. NIRF signals (700 nm, red) (right, top) localized with immunoreactive macrophages (Mac-3) (right middle), but not smooth muscle cells (SMC) (right, bottom) in the atherosclerotic plaque of apoE−/− HC mice. Autofluorescence signals (480 nm, green) were merged for better orientation of the tissue


Although standard risk factors of atherosclerosis and systemic biomarkers may predict patients at risk for an acute coronary event (vulnerable patients), imaging modalities need to be developed to identify the culprit lesions. Imaging strategies that can identify and quantify inflammation in atherosclerotic lesions are needed as tools for selective intervention on culprit lesions and for monitoring therapeutic interventions. Macrophages in atherosclerotic plaques can be localized by specific targeting of neo-expression of surface receptors, their characteristic alterations during apoptosis, and metalloproteinases secreted in the plaque.


This study was supported by National Institute of Health grant HL-78681 (to Dr. Narula) and Bristol-Myers Squibb grant (to Dr. Narula) as well as by GE Healthcare, Amersham, UK (to Dr. Schäfers and Dr. Levkau) and the German Research Foundation DFG, SFB 656, projects A1 & A2 (to Dr. Schäfers).

Copyright information

© Springer-Verlag 2007