Noninvasive Assessment of Angiogenesis by Ultrasound and Microbubbles Targeted to αv-Integrins
Background— Noninvasive methods for characterizing neovessel formation during angiogenesis are currently lacking. We hypothesized that angiogenesis could be imaged with the use of contrast-enhanced ultrasound (CEU) with microbubbles targeted to αv-integrins.
Methods and Results— Microbubbles targeted to αv-integrins were prepared by conjugating echistatin (MBE) or monoclonal antibody against murine αv (MBα) to their surface. Control microbubbles (MBc) were also prepared. The microvascular behavior of these microbubbles was assessed by intravital microscopy of the cremaster muscle in mice treated for 4 days with sustained-release FGF-2. Microvascular retention was much greater (P<0.01) for MBE (11±6 mm−3) and MBα (10±7 mm−3) than that for MBc (1±1 mm−3). Retained MBE and MBα attached directly to the microvascular endothelial surface. Microbubble retention in 4 control mice was minimal. Subcutaneous matrigel plugs enriched with FGF-2 were created in 12 mice and studied 10 days later. Neovessels within the matrigel stained positive for αv-integrins. CEU demonstrated greater (P<0.01) acoustic intensity for MBE (16.0±5.9 U) and MBα (17.0±5.5 U) compared with MBc (5.8±2.6 U). The signal from targeted microbubbles (MBE and MBα) correlated well (r=0.90) with the matrigel blood volume determined by CEU perfusion imaging.
Conclusions— CEU with microbubbles targeted for αv-integrins may provide a noninvasive method for assessing therapeutic angiogenesis.
Received August 15, 2002; revision received September 27, 2002; accepted October 1, 2002.
There is currently great interest in either promoting angiogenesis in chronically ischemic tissue or arresting angiogenesis in neoplasms or diabetic retinopathy. A noninvasive technique capable of temporally assessing neovessel formation with high resolution would be desirable for assessing these processes. In this study, we examined whether microbubbles targeted to endothelial cell receptors could be used to assess angiogenesis. This strategy is based on our previous experience with contrast-enhanced ultrasound (CEU) and site-targeted microbubble agents that are retained in regions of disease by virtue of their shell properties or by conjugation of specific ligands to their surface. The ability to noninvasively assess inflammatory responses in vivo by using CEU has been demonstrated with the use of microbubbles targeted activated leukocytes1,2 or to endothelial cell adhesion molecules.3
We hypothesized that CEU with microbubbles targeted to αv-integrins expressed on neovascular endothelium4,5 could be used to assess early angiogenesis noninvasively. Lipid microbubbles were targeted by surface conjugation of monoclonal antibody (mAb) or an RGD-containing peptide with a high affinity for αv-integrins. Retention of these agents in growth factor–stimulated vessels was assessed by intravital microscopy in the mouse cremaster muscle. The ability of CEU to image microbubbles retained in angiogenic vessels was also assessed in growth factor–enriched subcutaneous matrigel plugs in mice.
Two microbubble agents targeted to αv integrins were constructed bearing either mAbs against the integrin αv chain (MBα) or the RGD-containing disintegrin echistatin (MBE). Control microbubbles bearing isotype control mAb (MBc) were also constructed. First, biotinylated lipid-shelled microbubbles containing decafluorobutane gas were prepared.6 Either 2.5 μg of biotinylated echistatin (Sigma), 75 μg of biotinylated rat anti-mouse monoclonal IgG1 against αv-integrins (RMV-7), or isotype control antibody (R3-34, Pharmingen Inc) were conjugated to their surface.3 For intravital microscopy studies, microbubbles were fluorescently labeled with dioctadecyl-oxacarbocyanine (Molecular Probes). For perfusion imaging, nontargeted lipid-shelled microbubbles (MP1950) were also prepared.1 Microbubble concentration and size distribution were determined by electrozone sensing with a Coulter Multisizer IIe (Beckman-Coulter).
In Vitro Integrin Expression of Murine Endothelial Cells
Study protocols were approved by the Animal Research Committee at the University of Virginia. Expression of αv-integrins in response to FGF-2 was tested by immunofluorescent microscopy of cultured murine endothelial cells prepared from the thoracic aorta of mice, as previously described.7 Confluent cells were stimulated by 50 ng/mL of FGF-2 for 48 hours. Plates containing stimulated or control cells were incubated with the mAb RMV-7 for 10 minutes, washed, then incubated with phycoerythrin-labeled goat anti-rat secondary antibody (Pharmingen Inc) for 10 minutes. Separate cells were also incubated with the fluorescent secondary antibody alone to exclude nonspecific cellular interactions. Fluorescent microscopy (excitation filter 530 to 560 nm) of washed cells was performed with the use of a microscope (Axioskop2-FS, Carl Zeiss, Inc) and a high-sensitivity silicone-intensified tube camera (SIT68, Dage-MTI).
FGF-2 Stimulation of Cremaster Microvessels
Heparin-agarose beads (10 mg/mL) were mixed with calcium alginate (1.2% wt/vol), and the slurry was dropped through a 25-gauge needle into a 1.5% CaCl2 hardening solution to form 400- to 600-μm microcapsules. Fifteen microcapsules were combined with 1 μg FGF-2 for 24 hours to create a controlled release platform.8,9 Intrascrotal implantation of FGF-2–containing (n=6) or control (n=4) pellets was performed by aseptic technique in 10 wild-type C57Bl/6 mice (Jackson Laboratory, Bar Harbor, Maine). Pellets were “reloaded” once daily for 4 days with supplemental intrascrotal injections of either FGF-2 (125 ng) or saline for control mice.
Microvascular Behavior of Targeted Microbubbles
Intravital microscopy of the cremaster muscle was performed 4 days after pellet implantation to assess the microvascular behavior of targeted and control microbubbles. Animals were anesthetized with an intraperitoneal injection (12.5 μL/g) of a solution containing ketamine hydrochloride (10 mg/mL), xylazine (1 mg/mL), and atropine (0.02 mg/mL). A jugular vein was cannulated for administration of microbubbles and the cremaster muscle was exteriorized and fixed to a microscopy stage. Observations were made with the use of an Axioskop2-FS microscope (Carl Zeiss, Inc) with a saline-immersion objective (SW 40/0.8 numerical aperture). Video recordings were made with a high-resolution CCD camera (C2400, Hamamatsu Photonics) connected to an S-VHS recorder (S9500, JVC).
Microbubble behavior was assessed by fluorescent epi-illumination with a 460- to 500-nm excitation filter. Twenty optical fields (total tissue volume 4.2×10−3 cm3)10 were recorded 5 minutes after injection of 1×107 fluorescently labeled MBc, MBα, or MBE microbubbles in random order. Vascular density was assessed by fluorescent epi-illumination after intravenous injection of 5 μg FITC-dextran-70 in 4 mice in each group. Cumulative linear density was measured by calibrated video calipers in 4 optical fields (×20), and data were averaged for vessels with diameters of ≤10 μm, 10 to 20 μm, and >20 μm.
Matrigel Model of Angiogenesis
Subcutaneous matrigel plugs were created in 12 wild-type C57BL6 mice to promote neovessel formation.11 Matrigel (Collaborative Biomedical), a tumor basement membrane matrix extract, was thawed at 4°C and enriched with heparin (64 U/mL) and FGF-2 (500 ng/mL). Mice were anesthetized and 1 mL of enriched Matrigel (4°C) was injected subcutaneously in the left ventral region. Matrigel hardened to form discrete ellipsoid plugs.
Fluorescent confocal microscopy was performed in 3 mice to characterize microbubble retention in matrigel plugs. Ten days after matrigel injection, either MBc, MBα, or MBE microbubbles (1×107) labeled with dioctadecyl-tetramethylindocarbocyanine (Molecular Probes) were injected intravenously. After 15 minutes, vascular staining was performed with intra-arterial injection of 1 mg fluorescein Lycopersicon esculentum lectin.12 Perfusion fixation was performed with 4% paraformaldehyde in PBS, and thick sections of matrigel were examined with the use of a confocal laser scanning microscope (LSM510, Carl Zeiss Inc), using dual fluorescence overlays.
CEU of matrigel plugs was performed in 9 anesthetized mice 10 days after matrigel injection. Imaging was performed with a Sonos 5500 system (Philips Ultrasound), with harmonic power Doppler imaging at a transmit frequency of 1.6 MHz and a mechanical index of 1.0, with a small water bath serving as an acoustic interface. The acoustic focus was placed at the mid-level of the matrigel. Gain settings, compression, and pulse repetition frequency settings were optimized and held constant. Data were recorded on a magnetic optical disk and transferred to a computer for off-line analysis.
CEU imaging was performed after intravenous injection of 5×106 MBc, MBα, or MBE in random order at 20-minute intervals. Imaging was initiated 15 minutes after each injection to allow microbubble retention and clearance of most circulating microbubbles from the blood pool.1,3 The signal from the initial frame represented both retained and freely circulating microbubbles.1,3 After destruction of these microbubbles by 2 to 3 seconds of continuous imaging, intermittent imaging at a pulsing interval (PI) of 10 seconds was performed to derive the signal only from freely circulating microbubbles.13 These frames were averaged and digitally subtracted from the initial frame to create a single image representing only retained microbubbles. Background-subtracted acoustic intensity (AI) was measured from this image.14
Perfusion imaging of the matrigel plug was also performed with an infusion of nontargeted MP1950 microbubbles (5×106 per minute). Background images were acquired during continuous imaging at 30 Hz, and intermittent imaging was performed with progressive prolongation of the PI from 200 ms to 20 seconds. Averaged background frames were digitally subtracted from averaged frames at each PI, and the acoustic intensity was measured within specified regions of interest. PI versus acoustic intensity data were fit to the function, y=A (1−e−βt), where y is AI at the PI t, A is plateau video intensity (an index of blood volume), and β is the rate constant that provides a measure of microvascular red blood cell (RBC) velocity.13 Color-coding for visual analysis was performed on a pixel basis by assigning a color from a scale expanded to 256 levels, based on background-subtracted AI.14
Histopathology and Immunohistochemistry
Paraffin-embedded matrigel sections were stained with hematoxylin and eosin. Immunostaining for αv-integrins was performed on frozen sections of matrigel. Affinity-purified rat mAb against murine αv-integrins (RMV-7, Pharmingen Inc) was used as a primary antibody followed by a biotinylated goat anti-rat secondary antibody (Vector Laboratories). Staining was performed with a peroxidase kit (ABC Vectastain Elite, Vector Laboratories) and 3,3′-diaminobenzidine chromagen (DAKO). Slides were counterstained with hematoxylin.
Data are expressed as mean±SD. Comparisons between groups was performed with ANOVA. When differences were found, comparisons between two groups were performed by means of a nonpaired Student’s t test with Bonferroni correction. Correlations were performed by means of least-squares-fit linear regression analysis. Differences were considered significant at P<0.05 (2-sided).
Microbubble Retention in the Cremaster Microcirculation
Expression of αv-integrins by murine endothelial cells in vitro after exposure to FGF-2 for 48 hours was detected by immunofluorescent microscopy (Figure 1A). Intravital microscopy images of the microcirculation of the cremaster muscle of mice receiving control or FGF-2-eluting alginate pellets are shown in Figure 1B. In FGF-2–treated mice, focal regions of increased vascular density were observed, characterized by tortuous, poorly defined microvessels. Abnormal vascular permeability was seen in these areas, manifest by extravascular blush of FITC-dextran. The linear density of vessels was greater for microvessels ≤10 μm in diameter for FGF-2–treated compared with control mice (Table).
In mice receiving control pellets, microbubble retention in the cremaster microcirculation was uncommon for all 3 agents (Figure 2A). Most of this infrequent retention was from microbubble attachment to activated leukocytes adherent to the venular endothelium at sites of surgical trauma. In FGF-2–treated animals, retention of MBα and MBE was substantially greater than in control animals, whereas retention of MBc tended to decrease with FGF-2 treatment (Figure 2A). Enhanced retention of MBα and MBE microbubbles in FGF-2–treated mice was almost entirely due to their attachment directly to the microvascular endothelial surface, with a slight preference toward small (<30 μm) arterioles (Figure 2B), although attachment was also seen in small venules and capillaries (<10 μm). Flux of RBCs around microbubbles retained in capillaries was frequently observed, indicating endothelial attachment rather than size-dependent entrapment. Direct endothelial attachment was not observed for MBc.
CEU Assessment of Matrigel Angiogenesis
Neovessel formation in the matrigel plugs was observed on histology in all 9 mice (Figure 3A), with a central region devoid of angiogenic vessels present in most. Intravascular RBCs were seen, indicating the presence of flow in these vessels. Vessel diameter ranged from 5 to 150 μm, and both their size and density tended to increase with distance from the center. Immunohistochemistry demonstrated αv-integrin expression on the endothelial surface of these vessels, the density of which greatest at the outer one-third margin of the plug (Figure 3B). Leukocytes were occasionally found in the matrigel and also stained for the αv-integrin subunit.
Illustrated in Figure 4 are representative CEU images and the corresponding PI versus AI curve from a matrigel plug during continuous infusion of nontargeted microbubbles. Progressive prolongation of the PI resulted in an incremental increase in AI until plateau was reached. Microvascular blood volume, represented by the image at the longest PI or quantitatively as the plateau AI,13 tended to be greatest at the peripheral margins of the matrigel and was absent at the center of plugs that were determined by histology to have an avascular core. The size of the avascular region varied between 0% and 25% of the total area.
Confocal microscopy revealed a poorly organized vascular network within the matrigel plug (Figure 5). Retention of fluorescently labeled targeted but not control microbubbles was observed on dual fluorescence imaging (Figure 5). On CEU imaging, the signal from retained microbubbles retained in matrigel 15 minutes after their intravenous administration was significantly greater for MBα and MBE compared with MBc (Figure 6). Regions with the highest MBα and MBE signal tended to correlate spatially with regions with the greatest microvascular blood volume seen on perfusion imaging. The background-subtracted signal obtained during intermittent imaging at a PI of 10 seconds was low for all microbubble agents (MBc=2.6±1.4, MBα=2.9±2.0, MBE=3.0±1.5 U), indicating the presence of few freely circulating microbubbles at 15 minutes.
The signal intensity for retained MBα and MBE correlated well with the matrigel microvascular blood volume (A value of the PI versus AI curves) irrespective of whether analysis was performed for the entire matrigel plug or excluding nonperfused areas (regions where microbubble signal was absent on perfusion imaging) from analysis (Figure 7). The signal from retained microbubbles did not correlate with microvascular blood velocity (β value of the PI versus AI curves).
In the present study, we have demonstrated for the first time that angiogenesis can be characterized in vivo by using CEU and microbubble contrast agents targeted to endothelial cell surface receptors expressed in neovessels. Targeted microbubbles were created by conjugation of either mAbs or peptide ligands for αv-integrins to their surface. Both αvβ3 and αvβ5 are differentially expressed in tumor and developmental models of angiogenesis, and αvβ3 in particular appears to play a functional role in neovessel formation.6,7,15 The ability of αvβ3 to bind to many of the components of the extracellular matrix contributes to cell migration, adhesion, and differentiation during vasculogenesis or angiogenesis. This binding relies largely on integrin recognition of the RGD peptide sequence, which was the basis for microbubble targeting with the use of echistatin, a peptide that contains an RGD sequence in a mobile loop.16 This disintegrin was chosen because of its relative affinity to αvβ3 over nonactivated αIIbβ3 or the fibronectin receptor α5β1.16,17 In growth factor–stimulated cremaster muscles and subcutaneous matrigel plugs, targeting efficiency was equivalent for echistatin and αv-specific mAb strategies. Advantages of using short peptides for targeting in the clinical setting include less immunogenicity, lower cost, and problems with antibody availability.
Microbubble targeting was tested in models designed to mimic vascular responses to FGF-2, one of several proangiogenic factors under clinical investigation for treatment of ischemia. Although controlled-release FGF-2 treatment of the cremaster was performed only to stimulate endothelial αvβ3 expression, focal regions consistent with early angiogenesis were observed. Longer duration of treatment was not possible because of the hyperplastic and matrix responses that preclude videomicroscopy. Similar to previous descriptions, subcutaneous injection of FGF-2–enriched matrigel, composed of basement membrane proteins from murine Engelbreth-Holm-Swarm tumor, resulted in de novo vessel formation and penetration of neovessels. Expression of αvβ3 was confirmed by immunohistochemistry and has been shown to be a requirement for matrigel angiogenesis.18
Enhanced retention of targeted microbubbles in these models was demonstrated by both microscopy and CEU. Retention of MBE and MBα varied considerably, in part because of lower signal for the second targeted agent administered, implying competitive inhibition from residual free ligand. In the matrigel model there was also substantial variability in the degree of angiogenesis. Intravital microscopy confirmed direct attachment of targeted microbubbles to the endothelium rather than other potential mechanisms such as bubble accumulation in regions of abnormal perfusion or shear. Microbubble attachment occurred in small microvessels, with a slight preference toward small arterioles, where αvβ3 expression is likely to be high as the result of its putative role in extracellular matrix remodeling and endothelial cell invasion.19 Occasional attachment of microbubbles to activated leukocytes was also observed. This process is mediated primarily by serum complement for microbubbles with shells composed of lipid.20 Retention of nontargeted microbubbles was reduced by FGF-2, in accord with previous reports that FGF-2 can inhibit leukocyte recruitment in venules.21 Yet the presence of leukocytes on matrigel histology indicated that microbubble-leukocyte interaction was still most likely responsible for the low-intensity ultrasound signal for control microbubbles.
The ultrasound imaging protocol used to detect retention of targeted microbubbles in matrigel neovessels has been used previously to image tissue inflammation with microbubbles targeted to activated leukocytes or endothelial P-selectin.1–3 By imaging late after a single intravenous injection of contrast, signal from retained microbubbles relative to that of freely circulating microbubbles is optimized because of gradual removal of the latter from the blood pool. In the present study, the signal from freely circulating microbubbles 15 minutes after injection was very low, accounting for <15% of the contrast signal intensity on initial frames with the targeted agents. Ultrasound was performed with a high acoustic power, or mechanical index, to generate sufficient signal from retained microbubbles, which are susceptible to damping from cellular attachment.22 Harmonic power Doppler processing was chosen because of its enhanced sensitivity for detecting microbubbles in tissue. In the future, quantification of angiogenesis in physiological models with targeted CEU probably will require other image processing methods that have a greater dynamic range and are more reliable in terms of the relation between microbubble concentration and acoustic intensity. Nonetheless, we found a good relation between the signal from αv-targeted microbubbles and neovessel blood volume in matrigel.
There are several limitations of this study. First, it cannot be assumed that the models we used to test the feasibility of angiogenesis-targeted microbubbles are identical to disease-related angiogenesis in the clinical setting. On a more practical level, precise spatial correlation between CEU and immunohistology was not possible. This limitation was due to the difficulty spatially matching the ultrasound plane with matrigel sections that distorted after explantation and because 2-dimensional ultrasound images are formed from signals received over a beam elevation of several millimeters. Studies in physiological models of ischemic disease will be needed to address the question of whether this technique can image αv-integrin expression that occurs in large vessels during arteriogenesis or in small microvessels. Finally, although prior studies have indicated that very few retained microbubbles are required to produce an easily detectable CEU signal,22 the overall sensitivity for detecting angiogenesis with αv-targeted microbubbles remains unknown.
Our results suggest that imaging microvascular phenotype with CEU can provide additional information by evaluating early response to growth factor stimulation in terms of spatial extent and amount of angiogenesis. Such information is likely to be very useful for guiding focal delivery of proangiogenic proteins or genes that require direct administration into ischemic tissue. In this setting, CEU signal from targeted microbubbles is likely to precede neovascularization since αv-integrin expression may occur in preexisting vessels. Imaging angiogenic phenotype may also be useful for tumor detection and for providing information on prognosis, metastatic potential, or susceptibility to therapy. This has prompted early investigation into methods for detecting angiogenic phenotype with novel contrast agents used with MRI and positron emission tomography.23–25 Since acoustic destruction of “payload-bearing” microbubbles can be used to focally deliver therapeutic drugs or augment gene transfection,26 we believe that the role of angiogenesis-targeted microbubbles may uniquely extend beyond imaging and into the realm of site-specific therapy for ischemic tissues or tumors. Further characterization of αv-targeted microbubbles in physiological models of ischemic or tumor angiogenesis will be necessary to explore these potential applications.
This study was supported by grants (R01-HL-48890 and R01-HL-65704 to Dr Kaul; K08-HL-03810 to Dr Lindner) from the National Institutes of Health, Bethesda, Md, and from the Cardiovascular Research Initiative, Entertainment Industry Fund. Dr Leong-Poi is supported by a Research Fellowship Award from the Canadian Institutes of Health Research and the Heart and Stroke Foundation of Canada (Ottawa, Canada). Dr Christiansen is supported by a Postdoctoral Fellowship Award from the American Heart Association, Mid-Atlantic Affiliate, Baltimore, Md.
Lindner JR, Song J, Xu F, et al. Noninvasive ultrasound imaging of inflammation using microbubbles targeted to activated leukocytes. Circulation. 2000; 102: 2745–2750.
Christiansen JP, Leong-Poi H, Klibanov AL, et al. Noninvasive imaging of myocardial reperfusion injury using leukocyte-targeted contrast echocardiography. Circulation. 2002; 105: 1764–1767.
Lindner JR, Song J, Christiansen JP, et al. Ultrasound assessment of inflammation and renal tissue injury with microbubbles targeted to P-selectin. Circulation. 2001; 104: 2107–2112.
Brooks PC, Clark RAF, Cheresh DA. Requirement of vascular integrin αvβ3 for angiogenesis. Science. 1994; 264: 569–571.
Friedlander M, Thesefeld C, Sugita L, et al. Involvement of integrin αvβ3 and αvβ5 in ocular neovascular diseases. Proc Natl Acad Sci U S A. 1996; 93: 9764–9769.
Klibanov AL, Gu H, Wojdyla JK, et al. Attachment of ligands to gas-filled microbubbles via PEG spacer and lipid residues anchored at the interface. Proceedings of the 26th International Symposium on Controlled Release of Bioactive Materials. 1999:124–125.
Shi W, Wang NJ, Shih DM, et al. Determinants of atherosclerosis susceptibility in the C3H and C57BL/6 mouse model: evidence for involvement of endothelial cells but not blood cells or cholesterol metabolism. Circ Res. 2000; 86: 1078–1084.
Edelman ER, Nugent MA, Karnovsky MJ. Perivascular and intravenous administration of basic fibroblast growth factor: vascular and solid organ deposition. Proc Natl Acad Sci U S A. 1993; 90: 1513–1517.
Wei K, Jayaweera AR, Firoozan S, et al. Quantification of myocardial blood flow with ultrasound-induced destruction of microbubbles administered as a constant venous infusion. Circulation. 1998; 97: 473–483.
Friedlander M, Brooks PC, Shaffer R, et al. Definition of two angiogenic pathways by distinct αv integrins. Science. 1995; 270: 1500–1503.
Wierzbicka-Patynowski I, Niewiarowski S, Marcinkiewicz, et al. Structural requirements of echistatin for the recognition of αvβ3 and α5β1 integrins. J Biol Chem. 1999; 274: 37809–37814.
Thibault G, Tardif P, LaPalme G. Comparative specificity of platelet αIIbβ3 integrin antagonists. J Pharmacol Exp Ther. 2001; 296: 690–696.
Lindner JR, Coggins MP, Kaul S, et al. Microbubble persistence in the microcirculation during ischemia-reperfusion and inflammation: integrin- and complement-mediated adherence to activated leukocytes. Circulation. 2000; 101: 668–675.
Tromp SC, oude Egbrink MGA, Dings RPM, et al. Tumor angiogenesis factors reduce leukocyte adhesion in vivo. Int Immunol. 2000; 12: 671–676.
Lindner JR, Dayton PA, Coggins MP, et al. Non-invasive imaging of inflammation by ultrasound detection of phagocytosed microbubbles. Circulation. 2000; 102: 531–538.
Anderson SA, Rader RK, Westlin WF, et al. Magnetic resonance contrast enhancement of neovasculatures with αvβ3-targeted nanoparticles. Magn Reson Med. 200; 44: 433–439.
Haubner R, Wester HJ, Weber WA, et al. Non-invasive imaging of αvβ3 integrin expression using 18F-labeled RGD-containing glycopeptide and positron emission tomography. Cancer Res. 2001; 61: 1781–1785.
Shohet RV, Chen S, Zhou Y, et al. Echocardiographic destruction of albumin microbubbles directs gene delivery to the myocardium. Circulation. 2000; 101: 2554–2556.