INTRODUCTION

Top Abstract Introduction Methods Results Discussion References

RECENT DISCOVERY and characterization of heparin-binding growth factors has led to a growing appreciation of their role in normal and pathological angiogenesis (14, 15, 43). Acidic fibroblast growth factor (aFGF), basic fibroblast growth factor (bFGF), and their receptors have been isolated within the cardiovascular system and found to be upregulated in the setting of myocardial ischemia (7, 8, 12, 20). These findings have brought about interest in utilizing these agents to promote therapeutic angiogenesis.

The heparin-binding fibroblast growth factor (FGF) family is composed of 10 currently identified mitogens (20, 27, 42, 48) and 4 additional homologs of unknown function (39). aFGF (or FGF-1), one of the best characterized family members, was originally isolated from brain but is comparably abundant in heart (7, 32), where it is localized within cardiac myocytes (47). Embryonic cardiac myocytes also express FGF receptors that are required for proliferation in vivo (26) and that respond mitogenically to aFGF in vitro (31) but lose these receptors and FGF responsiveness as they terminally differentiate. In contrast, vascular endothelial cells from vessels throughout fetal and adult tissues, including the heart, retain FGF receptors and their responsiveness to aFGF in vitro. aFGF is a potent endothelial cell mitogen and chemotactic factor that is active in animal models of angiogenesis (see review, Ref. 42), dermal repair and capillary growth (25), and large vessel reendothelialization (5). Purified aFGF, however, has not been demonstrated to be efficacious in models of peripheral or coronary ischemia and, in fact, has been reported not to promote angiogenesis in ischemic dog myocardium (2). A possible explanation for this latter observation is that wild-type aFGF, which was delivered in the absence of heparin that is typically used to stabilize it, is rapidly inactivated in vivo. Therefore, to resolve whether aFGF is active in cardiac ischemia, we tested a genetically stabilized mutant of human aFGF bound to heparin in a clinically relevant porcine model. We report herein that this aFGF significantly enhanced coronary flow and improved cardiac function.

	METHODS

Top Abstract Introduction Methods Results Discussion References

Growth factor and delivery system preparation. A stabilized site-directed mutant of human aFGF, denoted Ser-117, in which Cys-117 was converted to a Ser residue by site-directed mutagenesis, was expressed in Escherichia coli and purified to apparent homogeneity by a method that eliminated detectable endotoxin (23, 29). Ethylene vinyl acetate (EVAc; DuPont, Wilmington, DE) matrices were made as specified (10, 11) by adding 1.5 g of EVAc dissolved in 0.5 g of porcine serum albumin to a lyophilized solution of 5.5 ml phosphate-buffered saline; 140 µg of heparin either with or without 140 µg of purified recombinant Ser-117 aFGF were then added to this solution. The mixture was poured into a precooled mold, allowed to harden, and then put under a 600-mTorr house vacuum at 20°C for 2 days. This procedure resulted in a porous EVAc matrix containing a diffusely distributed growth factor. Sections of 160 mg of matrix containing ~14 µg of heparin with or without 14 µg of Ser-117 aFGF were cut and activated with sterile water before perivascular placement. Each piece of EVAc matrix was sterilized under ultraviolet light for 30 s before surgical implantation.

Porcine model of chronic myocardial ischemia. Fifteen Yorkshire pigs (25-35 lb, Pine Acres Farm, Norwood, MA) were mechanically ventilated and anesthetized with halothane after administration of pentobarbital (10 mg/kg im) and ketamine (30 mg/kg iv). A left thoracotomy was performed at the fourth intercostal space, the pericardium was opened, and an Ameroid occluder (Research Instruments, Corvallis, OR) was positioned around the proximal left circumflex (LCX) artery after matching for size as described (18). Before placement of the Ameroid occluder, regional blood flow was measured by colored microsphere delivery (21).

Follow-up evaluation. Coronary angiography was performed on all animals 7-9 wk after initial surgery. After pentobarbital and halothane inhalation anesthesia were administered, animals were mechanically ventilated under constant hemodynamic monitoring. A 7-Fr JR4 diagnostic angiography catheter (Cordis, Miami, FL) was introduced over a 0.035-in. J wire, and selective coronary angiography was performed on the right and left coronary arteries in multiple left and right anterior oblique projections with ionic contrast (Renografin, Squibb Diagnostics, Princeton, NJ).

Echocardiographic analysis of regional and global myocardial function. Echocardiographic parameters were assessed from standard M-mode and two-dimensional echocardiographic images (Hewlett-Packard, Andover, MA) obtained in the open-chest state. Images were compared using apical four-chamber and midventricular short-axis planes before and after 2 min of pacing at 180 beats/min. Infarct size was determined as the percentage of akinetic endocardial circumference in the short-axis plane. Global ejection fraction was determined from the four-chamber view using a modified Simpson's algorithm (40). Regional wall thickening was determined from short-axis M-mode recording images in the mid left ventricular region oriented to include the lateral wall.

Evaluation of regional coronary flow. Colored microspheres (15-µm diameter; Triton Technology, San Diego, CA) were used to determine coronary flow at the time of final study at rest and during rapid (180 beats/min) ventricular pacing. To determine the extent of LCX territory, a set of colored microspheres was injected before the Ameroid constrictor placement with the LCX artery held transiently occluded. Thus all sections of the myocardium containing <10% of microspheres from this set at the time of final study were considered to belong to the LCX territory. For determination of coronary flow at the time of final study, a left atrial line was placed under direct vision and a set of microspheres (6 × 10⁶) was forcefully injected after verification of catheter placement. Reference blood samples were withdrawn using a syringe pump at a constant rate of 4 ml/min through the femoral artery.

1

1

Statistics. All data are expressed as means ± SD. A P value  0.05 was considered significant. Continuous variables including echocardiographic comparison of regional and global left ventricular function and regional coronary flow were compared using two-tailed t-tests. Serial data for the same group were compared using paired t-tests.

	RESULTS

Top Abstract Introduction Methods Results Discussion References

Study groups. All 15 animals survived the initial surgery. Of this total, one animal (aFGF group) did not demonstrate angiographic evidence of vessel occlusion and was excluded from data analysis. In addition, one animal in the aFGF group died after coronary angiography before coronary blood flow studies and echocardiography could be completed.

View larger version (106K): [in this window] [in a new window]   Fig. 1.   Histological analysis of coronary arteries following ethylene vinyl acetate (EVAc) implantation. Representative elastin sections (×32 magnification) of left circumflex (LCX) arteries at site of EVAc implantation were taken from control (A) and acidic fibroblast growth factor (aFGF)-treated (B) animals. Note absence of significant neointimal thickening in both specimens.

Assessment of regional myocardial flow. At the time of the final study, there was no significant difference in the LAD (nonischemic) territory blood flow between the two groups of animals (coronary blood flow: 0.79 ± 0.15 for aFGF vs. 0.80 ± 0.24 ml · min¹ · g¹ for control, P = NS). However, coronary flow in the LCX territory was significantly higher in the aFGF-treated than in control animals both at rest (coronary blood flow: 0.71 ± 0.05 for aFGF vs. 0.49 ± 0.06 ml · min¹ · g¹ for control, P = 0.001) and during rapid (180 beats/min) pacing (coronary blood flow: 0.94 ± 0.17 for aFGF vs. 0.58 ± 0.22 ml · min¹ · g¹ for control, P = 0.01). Furthermore, pacing resulted in a significant increase in coronary flow in the LCX territory in aFGF-treated but not control animals (Fig. 2).

View larger version (24K): [in this window] [in a new window]   Fig. 2.   Coronary blood flow in Ameroid-occluded LCX territory for animals implanted with EVAc polymers containing heparin (control) or Ser-117 aFGF and heparin (aFGF). Blood flow, determined using colored microspheres, is shown at rest (solid bars) and during rapid pacing (open bars). Comparisons of control vs. aFGF groups were carried out using two-tailed t-tests; comparison of coronary flow in aFGF group before and during pacing is carried out using a paired t-test.

Echocardiographic evaluation of regional and global myocardial function. Two-dimensional and M-mode echocardiography were used to measure left ventricular global (Fig. 3) and regional function (Fig. 4) in open-chest animals. Treatment with aFGF resulted in highly significant improvement of global ejection fraction measured both at rest (P = 0.003) and during pacing (P = 0.002). Likewise, aFGF treatment also resulted in significant preservation of regional LCX myocardial function during pacing (P = 0.02 vs. control).

View larger version (20K): [in this window] [in a new window]   Fig. 3.   Left ventricular wall thickening in posterior wall (LCX territory), obtained using open-chest echocardiography, for heparin control and aFGF-treated groups at rest (solid bars) and during pacing stress (open bars). Comparisons are shown as aFGF vs. control group at rest and during pacing (* P < 0.05).

View larger version (20K): [in this window] [in a new window]   Fig. 4.   Global left ventricular ejection fraction, obtained using open-chest echocardiography, for heparin control and aFGF-treated groups at rest (solid bars) and during pacing stress (open bars). Comparisons are shown as aFGF vs. control group at rest and during pacing (* P < 0.05).

	DISCUSSION

Top Abstract Introduction Methods Results Discussion References

Ischemia is characterized by inadequate blood flow and tissue oxygenation that is typically the consequence of decreased patency of atherosclerotic vessels. Clinical strategies to increase blood flow in ischemic peripheral and cardiac muscle have largely depended on vasodilation, angioplasty, and surgical revascularization. The uses of these approaches are restricted by the physiological limit of vasodilation and its side effects, the inaccessibility of angioplasty sites, and the limitations of surgery including graft failure and the absence of healthy patent vessels for autologous vascular transplants. Improvement in blood flow and muscle function can also be achieved by neovascularization, or angiogenesis, providing new collateral vessels.

The therapeutic use of angiogenic growth factors has been the subject of intensive investigation over the last several years (for review, see Ref. 46) and has demonstrated promise as a potential modality to treat both acute and chronic myocardial ischemia. Thus a number of studies have shown that bFGF, whether delivered systemically at high dose or locally at lower doses, results in improvement in collateral vessel flow, histological evidence of new vessel formation, and improvements in myocardial function in models of both acute and chronic myocardial or peripheral limb ischemia (1, 4, 18, 22, 44). Similarly, administration of vascular endothelial growth factor (VEGF) (3, 17, 19, 41) and FGF-5 (16) has been reported to produce functionally meaningful angiogenesis. However, unlike these growth factors, there are limited data regarding aFGF angiogenic efficacy in myocardial ischemia. Acidic FGF delivery via a soaked sponge placed between an internal mammary artery and the LAD myocardium (2) failed to demonstrate angiographic evidence of new vessel formation, and although aFGF delivery using fibrin glue applied between the aorta and the heart produced evidence of local extramyocardial collateral formation, it did not demonstrate any physiological significance of these collaterals (13, 33). It remains unclear, however, whether this lack of angiogenic effect was due to lack of ischemic stimulus, inadequate method of local drug delivery, rapid degradation of aFGF in vivo, or inherent failure of aFGF to promote angiogenesis in the myocardium.

In this study we found that local EVAc-based delivery of a genetically modified form of aFGF resulted in significant improvement in resting collateral blood flow that was maintained during rapid pacing with parallel improvements in global and regional left ventricular function. Improvements in flow both during rest and pacing were previously seen with VEGF (17), but only during pacing with bFGF (18). Flow during pacing reveals functional reserve vascular capacity that, when decreased, can contribute to exercise-related pain experienced by patients with angina. In this regard, improved flow at rest seen in this study could be either an intrinsic advantage of aFGF over bFGF or simply a dose-related effect. Increased blood flow may reflect a larger capillary bed volume indicative of enhanced angiogenesis (38), vasodilation (9, 35), or, perhaps, improvements in microvascular circulation. With regard to the latter, we have recently shown that the microvasculature recovered from aFGF-treated ischemic regions of porcine heart is indeed more responsive to -adrenergic and ADP endothelium-dependent vasodilation than microvessels from ischemic regions of control animals (34). Increased coronary flow induced by aFGF and, possibly, a cardioprotective effect of the growth factor (30) likely contributed to the corresponding improvement in regional and global left ventricular function during both rest and pacing.

These activities might be largely attributable to neovascularization induced by aFGF, because other angiogenic proteins have also been reported to enhance flow and function in models of ischemia. Endothelial cells within ischemic cardiac muscle can become selectively responsive to angiogenic mitogens. Hypoxic capillary endothelial cells exhibit threefold-enhanced expression of high-affinity FGF receptors in culture along with an increased mitogenic and chemotactic responsiveness to bFGF (37). aFGF functions efficiently through all seven ligand-selective FGF receptor subtypes generated by alternative splicing of the four known FGF receptor genes (37). bFGF and FGF-5 appear to act principally through subsets of five and two of these receptor subtypes, respectively (28). Expression of mRNA encoding one or both alternatively spliced versions of FGF-1 is increased 2.4-fold in ischemic porcine myocardium (36). FGF-5 functions through one of the forms of this receptor, whereas bFGF and aFGF work efficiently through both forms (28). Therefore, activities exhibited by bFGF and FGF-5 would also be expected to be exhibited by aFGF. In light of the comprehensive FGF receptor subtype binding exhibited by FGF along with its activity in several in vivo angiogenesis assays (42), its reported lack of angiogenic activity in ischemic dog myocardium as discussed above is surprising.

The instability of aFGF, especially in the absence of heparin to which it binds (29), can lead to diminished potency or even complete inactivation. In the absence of heparin or other equivalent polyanions, wild-type human aFGF unfolds at or slightly below physiological temperature (45) and inactivates at 37°C in tissue culture, with a half-life of only 15 min (29). The stabilized mutant used in this study was generated (23) by replacement of one of the three buried Cys residues (6) with an isomorphous Ser residue (Ser-117). This single sulfur-to-oxygen atom substitution in the one Cys residue unique to human aFGF does not alter either the conformational unfolding temperature (unpublished observation) or mitogenic potency (29) and, because of its lack of exposure, should not present a distinct immunologic surface epitope. However, substitution of this single Cys residue increases the activity half-life to 1.4 h (29), apparently reflecting a diminished rate of formation of inactivating disulfide bonds (unpublished observations) promoted by air oxidation (29) and, perhaps, catalyzed by trace metals (23). Wild-type aFGF binds tightly to heparin, which increases its thermal stability to ~60°C (45) and its mitogenic half-life at 37°C by nearly 100-fold to 24 h (29). However, even a 1-day half-life can correspond to nearly complete loss of activity well before the end of long-duration slow-release dosing. In the presence of heparin, the Ser-117 mutant exhibits a 10-fold greater activity half-life of 240 h (29).

To deliver aFGF, we employed EVAc matrices. This choice was dictated by the relatively lesser affinity of aFGF for heparin-alginate than that of bFGF and the lesser stability of aFGF protein. EVAc matrices have previously been used to deliver a number of biologically active materials including heparin (10) and antisense oligonucleotides (11). These studies have demonstrated that the copolymer is stable at body temperature and produces no untoward effects at the site of implantation. Indeed, in this study we have not observed any inflammatory reaction at sites of polymer placement.

The previously reported (2) lack of efficacy of aFGF in a dog model of cardiac ischemia might have been either a species-related effect, a consequence of the method of delivery, or the result of relatively rapid inactivation of wild-type aFGF, especially in the absence of heparin. The current observation of the efficacy of a stabilized form of aFGF in the presence of heparin in a clinically relevant porcine model of cardiac ischemia is consistent with the known angiogenic activity of the wild-type protein. Moreover, whatever the reason for the lack of efficacy in this previous study, treatment with stabilized aFGF clearly results in therapeutically significant improvements in blood flow, myocardial perfusion, and function in cardiac ischemia.

Several issues need to be considered in evaluating the results of the present study. Although all animals had completely occluded LCX, we cannot rule out the possibility that aFGF may have influenced the rate of Ameroid closure compared with that in control animals. This possibility is potentially relevant given known, aforementioned vasoactive properties of FGF and their ability to induce vasodilation in coronary bed as well as their potential cardioprotective activity. Thus growth factor-mediated delay in the time of Ameroid closure may have influenced results in the treatment groups. In addition, in contrast to previous studies (17, 18) employing bFGF that have shown improvement in stress-induced flow in growth factor-treated animals, we have observed improved flow in the compromised territory in aFGF-treated animals both at rest and during rapid pacing. This improvement in rest coronary flow may be secondary to superior angiogenic qualities of aFGF or may be an unintended consequence of the relatively small numbers of animals in the study.