Assessment of Endogenous and Therapeutic Arteriogenesis by Contrast Ultrasound Molecular Imaging of Integrin Expression
Background— We hypothesized that molecular imaging with contrast-enhanced ultrasound (CEU) and microbubbles targeted to endothelial integrins could be used to noninvasively assess early angiogenic responses to ischemia and growth factor therapy.
Methods and Results— Hindlimb ischemia was produced in 48 rats by ligation of an iliac artery. Half of the animals received intramuscular sustained-release fibroblast growth factor-2 (FGF-2). Immediately after ligation and at subsequent intervals from 4 to 28 days, blood flow and oxygen tension in the proximal adductor muscles were measured by CEU perfusion imaging and phosphor quenching, respectively. Targeted CEU imaging of αv- and α5β1-integrin expression was performed with microbubbles bearing the disintegrin echistatin. Iliac artery ligation produced a 65% to 70% reduction in blood flow and oxygen tension. In untreated ischemic muscle, muscle flow and oxygen tension partially recovered by days 14 to 28. In these animals, signal from integrin-targeted microbubbles was intense and peaked before flow increase (days 4 to 7). In comparison to untreated animals, FGF-2–treated muscle had a greater rate and extent of blood flow recovery and greater signal intensity from integrin-targeted microbubbles, which peaked before maximal recovery of flow. On immunohistology, arteriolar but not capillary density increased in the ischemic limb after ligation, the rate and degree of which were greater in FGF-2–treated rats. Immunofluorescence demonstrated intense staining for αv in arterioles, the temporal course of which correlated with targeted imaging.
Conclusions— Targeted CEU can be used to assess endogenous and therapeutic arteriogenesis before recovery of tissue perfusion. These results suggest that molecular imaging of integrin expression may be useful for evaluating proangiogenic therapies.
Received June 4, 2004; revision received December 29, 2004; accepted February 7, 2005.
Despite advances in percutaneous and surgical revascularization techniques, the number of patients with severe ischemic heart or peripheral vascular disease who are not eligible for revascularization is growing. To address this problem, new strategies for improving blood flow by promoting new vessel growth in ischemic tissue are being developed. Promising methods include treatment with proangiogenic factors or their respective genes.1,2 Because of the urgent need for palliative therapy, clinical trials have been initiated before optimal treatment regimens have been definitively established. A noninvasive, high-resolution technique capable of assessing neovessel formation would be valuable for characterizing the cellular processes responsible for vascular remodeling and for evaluating new proangiogenic therapies or dosing regimens.
See p 3188
Contrast-enhanced ultrasound (CEU) imaging of molecular events in vivo has recently been achieved by targeting microbubble contrast agents to events that occur within the vascular compartment. Microbubbles have been targeted to angiogenesis by surface conjugation of echistatin that binds to αv- and α5β1-integrins expressed by the neovascular endothelium.3,4 These targeted microbubbles adhere in vivo to microvessels in tissue treated with fibroblast growth factor-2 (FGF-2) and to neovessels in a matrigel model of angiogenesis.3 CEU with these microbubbles has recently been used to assess tumor angiogenesis in glioblastomas.4
In the present study, we hypothesized that CEU with angiogenesis-targeted microbubbles can spatially and temporally assess endogenous angiogenic responses to chronic ischemia and proangiogenic responses to growth factor administration. We tested our hypothesis in a rat model of ischemic hindlimb skeletal muscle with and without treatment with sustained-release FGF-2.
Lipid-shelled decafluorobutane microbubbles were targeted to angiogenesis by shell surface conjugation of the disintegrin echistatin at the end of a polyethylene glycol tether (MBE), as previously described.3 This conjugation strategy results in ≈2500 ligands conjugated per square micrometer of surface area.5 Control microbubbles (MBC) without echistatin were also prepared. For perfusion imaging, nontargeted lipid-shelled decafluorobutane microbubbles (MP1950) were prepared.6 Microbubble concentration and size distribution were determined by electrozone sensing (Multisizer IIe, Beckman-Coulter).
Flow-Chamber Cell Attachment Studies
Primary human umbilical vein endothelial cells (HUVECs) were harvested, grown to confluence in M199 medium with 20% FBS and 5 mg · mL−1 endothelial cell growth supplement (Sigma), and studied at passages 2 to 3. Microbubble attachment to HUVECs was tested with the use of an inverted parallel-plate flow chamber (Glycotech). MBE or MBC (6×105) were drawn through the chamber at a shear stress of 0.7 dyne · cm−2 over 5 minutes, followed by bubble-free medium for 5 minutes. The number attached in 20 optical fields (0.3 mm−2) was determined by microscopy. Studies were performed in triplicate. Studies were also performed 1 hour after antibody blockade (30 μg · mL−1) of αvβ3- (LM609, Chemicon) and α5-integrins (IIA1, BD PharMingen). Integrin expression was confirmed by FITC-conjugated secondary antibody staining for the blocking antibodies.
The study protocol was approved by the Animal Research Committee at the University of Virginia. Proximal hindlimb ischemia was produced in 48 Sprague-Dawley rats. Rats were anesthetized with intraperitoneal injection of ketamine hydrochloride (10 mg · kg−1), xylazine (8 mg · kg−1), and atropine (0.02 mg · kg−1). With the use of an aseptic technique, the left common iliac artery and small proximal branches were ligated. Half of the animals (n=24) were treated with implantation of 50 to 60 alginate microcapsules (400 to 600 μm) containing heparin-agarose beads for controlled release of FGF-2.3,7 Microcapsules were combined with 5 μg FGF-2 for 24 hours before their implantation into the adductor muscles at 3 sites distal to arterial ligation. The incision was closed in layers, and animals were recovered.
CEU of the proximal hindlimb adductor muscles (adductor magnus and semimembranosus) was performed with ultraharmonic imaging (Sonos 5500, Philips Ultrasound). To ensure image registration, the hindlimbs were immobilized on the imaging stage, and the transducer was fixed in a swinging arm clamp that maintained an elevational plane to image both limbs. Imaging was performed at a mechanical index of 1.0 and a transmission frequency of 1.3 MHz. Gain settings and compression were optimized and held constant. Data were recorded on magnetic-optical disk and transferred to a computer for offline analysis.
Perfusion was assessed during continuous infusion of MP1950 (1×107 min−1). Background images were acquired at 30 Hz for subtraction of both tissue and large intramuscular vessels.8 Intermittent imaging was performed by progressive prolongation of the pulsing interval (PI) from 0.2 to 20 seconds. Several averaged background frames were digitally subtracted from averaged contrast-enhanced frames at each PI. PI versus video intensity (VI) data were fit to the function y=A(1−e−βt), where y is VI at the pulsing interval t, A is plateau VI, which is an index of blood volume, and β is the rate constant that provides a measure of microvascular blood velocity.8,9 Microvascular blood flow was calculated by the product of A and β. Parametric images of microvascular blood flow were created in which the value of A×β is displayed on a pixel-by-pixel basis.10
Targeted Imaging of Integrin Expression
Targeted imaging was performed as previously described.6 Imaging was paused for 15 minutes after an intravenous injection of 3×107 MBC or MBE to allow retention of microbubbles and clearance of freely circulating microbubbles from the blood pool. A single image reflecting only retained microbubbles was created by capturing the initial frame and subtracting averaged frames subsequently obtained at a PI of 10 seconds to eliminate signal from the few freely circulating microbubbles. Because the number of microbubbles retained is influenced by microbubble influx, the image representing retained microbubbles was normalized to microvascular blood flow from parametric CEU perfusion images, thereby producing an image reflecting microbubble retention fraction.4
Tissue Oxygen Tension
Tissue oxygen tension was measured by oxygen-mediated phosphor quenching.11 Muscle phosphorescence at an excitation wavelength of 635 nm was measured (PMOD 2000, Oxygen Enterprises, Ltd) 5 minutes after intravenous injection of 5 mg/kg Oxyphor-R2 (Oxygen Enterprises, Ltd). The returning light was optically filtered and relayed to a digital processor for determination of phosphorescence decay. Oxygen pressure was calculated according to the Stern-Volmer equation with the use of known constants for temperature and pH.11
Staining was performed on fixed, paraffin-embedded sections of muscle in the imaging plane. Polyclonal antibodies for αv- (AB1930, Chemicon International Inc) and α5-integrin (SC6593, Santa Cruz) were used for primary labeling. Species-appropriate FITC-labeled or biotinylated secondary antibodies for Vector red staining were used (Vector Laboratories). For endothelial cell staining, primary mouse anti-rat CD31 (MAB1393, Chemicon) was used with an ALEXA-555–labeled secondary antibody (Molecular Probes). For smooth muscle α-actin staining, a FITC-labeled primary monoclonal antibody (1A4, Sigma) was used. Fluorescent microscopy was performed with excitation filters of 460 to 500 (for FITC) and 530 to 560 nm (for Vector red and ALEXA-555). A minimum of 15 random nonoverlapping optical fields from ≥4 animals for each time and treatment cohort were selected for analysis performed blinded to animal identity. The density and number of capillaries per myocyte were determined by vessels that stained for CD31 but not α-actin on sections in the transaxial plane of the muscle fibers. Noncapillary microvascular density was determined by the number α-actin–positive vessels and expressed per tissue area. On the basis of reanalysis of 20 sections, intraobserver variability was 7% for capillary and 5% for noncapillary microvascular density.
Imaging of the ischemic hindlimb adductor muscles and the control contralateral muscles was performed 1 hour after iliac artery ligation and at either 4, 7, 14, or 28 days after ligation (n=6 at each interval for FGF-2–treated and untreated rats). For each study, targeted CEU with MBC and MBE was performed in random order 20 minutes apart. Perfusion of the ischemic and contralateral muscles and tissue oxygen tension were assessed. At the completion of each study, skeletal muscle tissue from the ischemic adductor muscle and the contralateral control muscle was obtained for histology.
Data are expressed as mean±SD. Comparisons between multiple stages were made with 1-way ANOVA. When differences were found, interstage comparisons were performed with the use of nonpaired Student t test with Bonferroni correction. Data for ischemic and control hindlimbs were compared by paired Student t test. Differences were considered significant at P<0.05 (2-sided).
Muscle Perfusion and Vascular Density
Blood flow in the adductor muscles was reduced to approximately one third of normal immediately after iliac artery ligation (Figure 1A). In the first 7 days after ligation, blood flow in the ischemic muscle increased only in FGF-2–treated animals. In untreated animals, flow increased later after ligation (days 14 to 28) but remained significantly lower than that in FGF-2–treated animals. The Table shows the source blood flow data from the ischemic and control skeletal muscle. Tissue oxygen tension measurements (Figure 1B) demonstrated a nearly identical pattern of recovery as blood flow for both treatment groups.
On visual inspection, surface vessel proliferation (Figure 2A) was observed consistently for ischemic muscles late (14 to 28 days) after artery ligation for both treatment groups. On histology, an increase in noncapillary microvascular (arteriolar and venular) density occurred in ischemic muscle over the first 2 weeks, the rate and extent of which were greater in FGF-2–treated muscles (Figure 2B and 2C). Although noncapillary microvascular density in FGF-2–treated vessels decreased slightly after day 7, the diameter of these vessels tended to increase, consistent with a process of selection and maturation. The capillary density and ratio of capillaries to muscle fiber did not change significantly over time in either treatment group.
Endothelial αv-integrin expression was observed in all ischemic muscle at days 4 and 7 after ligation and was most prominent in large penetrating vessels and transverse arterioles (Figure 3). Staining was also intense in nonendothelial vascular layers and perivascular tissues. Low-intensity staining for αv-integrin was seen in capillaries. Expression of αv-integrin decreased markedly by day 28 and became undetectable in untreated animals. In FGF-2–treated animals, positive αv-integrin staining was also detected for extravascular leukocytes infiltrating the muscle (Figure 3), most consistent with monocyte recruitment observed previously.12 In nonischemic muscle, endothelial αv-integrin expression was either absent or present at very low levels at all study intervals.
Endothelial expression of α5-integrin was also observed in intramuscular noncapillary microvessels in ischemic muscle (Figure 4) but was less prominent and much more heterogeneous among vessels than αv. Vascular α5-integrin staining in untreated ischemic muscles was seen primarily at days 4 to 7 and was uncommon at later time intervals. Much of this signal appeared to be localized to regions other than the endothelial surface. In FGF-2–treated animals, vascular staining at days 4 to 7 was more intense than in untreated animals and was detected in some vessels even at day 28, although again much of this signal appeared to be nonendothelial. Leukocyte staining for α5 was also observed.
Targeted CEU Imaging
Flow chamber studies were performed to evaluate the selective targeting of microbubbles to endothelial integrins as intended. HUVECs stained positive for both αvβ3 and α5. In flow chamber experiments, attachment to HUVECs was greater for targeted (MBE) versus control (MBC) microbubbles (67±42 versus 12±9 mm−2; P<0.05). MBE attachment was incrementally reduced by blocking αvβ3 (24±15 mm−2) and both αvβ3 and α5 (12±9 mm−2).
In ischemic muscle, signal from retained control microbubbles (MBc) increased marginally 1 week after ligation, whereas signal from targeted microbubbles (MBE) increased markedly in the first week after ligation, the degree of which was greater for FGF-2–treated animals (Figure 5A). By day 28, MBE signal had decreased in both groups but remained slightly elevated compared with baseline for FGF-2–treated animals. Comparison with the temporal pattern of blood flow (Figure 1) revealed that the signal from MBE peaked before significant changes in perfusion had occurred for untreated animals. For FGF-2–treated animals, signal from MBE peaked earlier than maximal flow changes. In control muscle (Figure 5B), the signal for MBE and MBC remained low at all time points.
Examples of targeted imaging 4 days after iliac artery ligation are illustrated in Figure 6. Signal enhancement from MBE was greater and more diffuse in the FGF-2–treated muscle than in the untreated ischemic muscle, where enhancement was located primarily around the region of the major vascular bundles.
Methods for imaging disease-related processes at the cellular or molecular level are now being developed to better characterize pathophysiology, diagnose disease early, and evaluate response to therapy. In the present study, we demonstrated that targeted CEU imaging of integrin expression can be used to assess the endogenous response to occlusive peripheral arterial disease, as well as the therapeutic response to growth factor administration. Molecular imaging detected the expression of integrins associated with vascular remodeling before maximal changes in limb skeletal muscle perfusion had occurred for both endogenous responses to ischemia and therapeutic response to proangiogenic therapy with FGF-2.
Assessment of Arteriogenesis by Perfusion Imaging
The model of rat hindlimb ischemia used in this study was designed to study vascular adaptation in the proximal hindlimb muscles. The presence of ischemia immediately after ligation was indicated by an ≈70% reduction in resting blood flow by CEU and by a dusky appearance of the muscle. In animals not treated with FGF-2, endogenous angiogenic responses were manifest by an increase in tissue perfusion 2 to 4 weeks after ligation. These findings are in accord with those observed with radiolabeled microsphere and surface laser-Doppler measurements in models of more distal lower extremity occlusion in rats.13,14
The underlying changes in vascular architecture responsible for restitution in flow in this model are complex and not yet fully characterized. Angiography has demonstrated that arteriogenesis, the growth of new collateral arteries and arterioles or the recruitment and maturation of preexisting channels, plays a dominant role in late recovery of proximal hindlimb blood flow and flow reserve.15,16 Proliferative responses at the capillary level tend to be more prominent in distal limb tissue.15 In accord with this notion, we found a time-dependent increase in small penetrating arteries and medium-sized arterioles and venules in the ischemic proximal hindlimb adductor muscle, whereas the capillary-to-myocyte ratio did not change.
FGF-2 has been shown to have direct effects on endothelial and smooth muscle cell growth and migration and to augment neovascularization in models of limb ischemia, in part via arteriogenesis.14,17 In this study treatment of the ischemic muscle with sustained-release FGF-2 resulted in a greater and more rapid recovery of blood flow and greater density of noncapillary microvessels. Although the microvascular density peaked early (at 7 days) and then declined slightly, the vessel diameter tended to progressively increase, consistent with a process of vessel selection and maturation that has been previously observed in larger vessels.18 It should be noted, however, that morphological changes do not necessarily equate to functional changes.
Molecular Imaging of Vascular Remodeling
We sought to determine whether molecular imaging of angiogenesis-related integrins could herald later changes in microvascular morphology and flow. For targeting integrins, the disintegrin echistatin was conjugated to microbubbles. This ligand contains an RGD motif and has enhanced selectivity for αv-integrins, particularly for αvβ3.19 These integrins are selectively expressed in angiogenic vessels, although their functional role has recently been debated.20,21 Echistatin can also bind α5β1,19 which also participates in the angiogenic response, either by direct signaling effects or by regulating αvβ3.22,23 We observed endothelial expression of both αv- and α5-integrins in ischemic hindlimbs early after arterial ligation. Although this suggests that signal from MBE was not specific for any one integrin, we believe that the majority of targeted signal was from αv-integrins because its expression was more extensive on immunohistology. Moreover, our flow-chamber studies using HUVECs that express α5β1- and αv-integrins at early passage22,23 indicated that selective attachment of MBE was mediated primarily by the latter. Unlike many other targeted contrast agents, microbubbles are pure intravascular tracers and therefore are unlikely to be influenced by the αv-integrin signal seen in perivascular tissues.
Targeted radionuclide imaging of αv-integrin expression has recently been used to detect microvascular proliferation in infarcted myocardium.24 Our study is novel in that it examined the temporal relationship between integrin expression and flow recovery during chronic hypoperfusion, with and without proangiogenic therapy. In animals not undergoing FGF-2 treatment, signal from targeted microbubbles peaked at day 7 (Figure 5), around the time when integrin expression on immunohistology was greatest. The signal from targeted microbubbles cannot necessarily distinguish capillary from noncapillary microvascular enhancement. However, immunohistology revealed that integrin expression was found primarily in arterioles and venules. Together with the histological data on noncapillary microvascular proliferation, these findings indicate that molecular imaging of integrin expression with CEU can detect vascular remodeling associated with endogenous arteriogenic responses to limb ischemia. For untreated ischemic muscles, signal enhancement from molecular imaging occurred before any measurable increase in skeletal muscle perfusion or tissue oxygen content (Figure 1).
The proangiogenic effects of FGF-2 are thought to involve αv- and α5β1-dependant signaling pathways.20,22 Treatment of the ischemic muscle with FGF-2 in this study produced an earlier and greater peak signal intensity with integrin-targeted microbubbles. The peak intensity from targeted imaging occurred before the time of maximal noncapillary microvascular density, suggesting that signal intensity was dependent more on the degree of expression than on the number of vessels. These data also suggest that molecular imaging of integrin expression can provide information on the very early stages of vascular remodeling associated with growth factor therapy.
A major limitation of this study was that we could not image each individual animal at all of the specified study intervals because of the need to procure tissue at each stage and to avoid any influence of the integrin-blocking effect of echistatin. Another major limitation of this and many other angiogenesis-targeting imaging studies is the broad and overlapping integrin expression profile for neovascular endothelium and monocytes that participate in vascular remodeling. The signal from targeted microbubbles may have been influenced by stimulation of the inflammatory response by FGF-212 and secondary attachment of targeted microbubbles to α5β1 or αv on activated monocytes within the vascular space. We are somewhat reassured, however, that intravital microscopy of tissue treated with FGF-2 has demonstrated primarily endothelial attachment for echistatin-bearing microbubbles.3 We also used only 1 imaging plane to ensure image registration between targeted and perfusion imaging. Differences in vascular responses according to the distance from the site of ligation could have produced very different results if distal tissues were examined. It must also be acknowledged that the integrin expression profile will likely differ according to the proangiogenic therapy that is used. Finally, blood flow measurements were relative to the normal limb rather than in absolute flow per muscle mass.
In summary, our results indicate that imaging endothelial integrin expression with targeted CEU can be used to assess endogenous and therapeutic arteriogenesis. We also demonstrated for the first time that CEU perfusion imaging alone may provide a method for noninvasively evaluating the effect of proangiogenic growth factor therapy. In the context of the mixed results of clinical studies investigating proangiogenic protein and gene therapy to date, we believe that imaging techniques such as targeted CEU can potentially enhance our ability to define optimal treatment strategies in the preclinical testing stage and ultimately in clinical trials.
This study was supported by grants (R01-EB-002069, R01-DK-063508, R01-HL-074443, and R01-HL-078610) from the National Institutes of Health, Bethesda, Md; a Grant-in-Aid from the American Heart Association, Mid-Atlantic Affiliate, Baltimore, Md; and a Research Fellowship Award from the Canadian Institutes of Health Research and the Heart and Stroke Foundation of Canada, Ottawa, Canada.
Dr Klibanov has received a research grant from Philips Research and delivered an invited presentation at Bracco Research. He has an ownership interest in Targeson LLC and stock in Tyco Mallinckrodt. Dr Lindner has served as a consultant to VisualSonics and has ownership interest in Targeson LLC.
Leong-Poi H, Christiansen JP, Klibanov AL, et al. Noninvasive assessment of angiogenesis by ultrasound and microbubbles targeted to αv-integrins. Circulation. 2003; 107: 455–460.
Ellegala DB, Leong-Poi H, Carpenter JE, et al. Tumor angiogenesis imaging with contrast ultrasound and microbubbles targeted to αv-integrins. Circulation. 2003; 108: 336–341.
Lindner JR, Song J, Xu F, et al. Noninvasive ultrasound imaging of inflammation using microbubbles targeted to activated leukocytes. Circulation. 2000; 102: 2745–2750.
Dawson D, Vincent MA, Clarke A, et al. Assessment of capillary recruitment in skeletal muscle using contrast-enhanced ultrasound. Am J Physiol. 2002; 282: E714–E720.
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.
Linka AZ, Sklenar J, Wei K, et al. Assessment of transmural distribution of myocardial perfusion with contrast echocardiography. Circulation. 1998; 98: 1912–1920.
Yang HT, Deschenes MR, Ogilvie RW, et al. Basic fibroblast growth factor increases collateral blood flow in rats with femoral arterial ligation. Circ Res. 1996; 79: 62–69.
Hershey JC, Baskin EP, Glass JD, et al. Revascularization in the rabbit hindlimb: dissociation between capillary sprouting and arteriogenesis. Cardiovasc Res. 2001; 49: 618–625.
Baffour R, Berman J, Garb JL, et al. Enhanced angiogenesis and growth of collaterals by in vivo administration of recombinant basic fibroblast growth factor in a rabbit model of acute lower limb ischemia: dose-response effect of fibroblast growth factor. J Vasc Surg. 1992; 16: 181–191.
Hoefer IE, van Royen N, Buschmann IR, et al. Time course of arteriogenesis following femoral artery occlusion in the rabbit. Cardiovasc Res. 2001; 49: 609–617.
McLane MA, Marinkiewicz C, Vijay-Kumar S, et al. Viper venom disintegrins and related molecules. Proc Soc Exp Biol Med. 1998; 219: 109–119.
Brooks PC, Clark RAF, Cheresh DA. Requirement of vascular integrin αVβ3 for angiogenesis. Science. 1994; 264: 569–571.
Kim S, Harris M, Varner JA. Regulation of integrin αvβ3-mediated endothelial cell migration and angiogenesis by integrin αvβ3 and protein kinase A. J Biol Chem. 2000; 275: 33920–33928.
Klein S, de Fougerolles AR, Blaikie P, et al. α5β1 integrin activates an NF-κB–dependent program of gene expression important for angiogenesis and inflammation. Mol Cell Biol. 2002; 22: 5912–5922.