Introduction

Top Abstract Introduction Materials and Methods Results Discussion References

Since their discovery over 15 years ago, several angiogenic growth factors have been isolated and purified (Folkman and Klagsbrun, 1987). The concept that such agents could be used therapeutically to induce both small and large vessel growth (angiogenesis and arteriogenesis, respectively) in states of chronic ischemia due to atherosclerosis has driven both experimental and clinical studies. In both ischemic limbs and heart muscle, the majority of studies have focused on the delivery of a single growth factor to stimulate vascular growth. The most extensively studied proteins are members of the VEGF and FGF families (Isner et al., 1996; Mack et al., 1998; Laham et al., 1999; Unger et al., 2000). There is evidence that delivery of these compounds, in the form of protein or gene therapy, can enhance blood flow to ischemic tissue in various experimental models (Unger et al., 1994; Mack et al., 1998). However, for a number of acknowledged reasons, results obtained in experimental models often do not translate directly into clinical practice.

One unanswered question about therapeutic angiogenesis is whether delivery of a single growth factor to an ischemic organ will be sufficient to induce growth of relatively large conduit vessels (arteriogenesis). Since ischemic syndromes resulting from arteriosclerosis affect the conduit vessels and not the small arterioles or capillaries, the growth of large vessels can be considered requisite for successful therapy. Many factors participate in the process of arteriogenesis (Beck and D'Amore, 1997). Recent studies have identified synergistic effects of angiogenic agents in the induction of vascular growth (Gajdusek et al., 1993; Ramoshebi and Ripamonti, 2000). Moreover, while prior published animal studies using single-factor strategies noted above have shown improvements, none has shown normalization of blood flow to treated ischemic myocardium. Identification of the factors involved, clarification of the role each factor plays, and an understanding of the events involved in the process of arteriogenesis currently constitute areas of active investigation (Carmeliet, 2000).

Recently, a naturally occurring growth factor mixture (GF_m) isolated from bovine long bones was shown to enhance bone formation in several models, including a standard rabbit model of lumbar spinal fusion (Boden et al., 1997). Early in the course of investigating the bone formation stimulation properties of GF_m, it became evident that this mixture also promoted vascular growth. This was particularly evident in a rat ectopic bone development assay in which a collagen disk saturated with GF_m was placed subcutaneously over the chest wall. By 3 weeks after implant, the collagen was replaced with a bony ossicle, and an extensive vascular network (including large vessels) was observed on the surface of the ossicles. In the present study, we follow up on these preliminary findings by formally studying the angiogenic properties of GF_m. We present results indicating that indeed this multiple growth factor strategy effectively induces large vessel growth in chronically ischemic myocardium.

	Materials and Methods

Top Abstract Introduction Materials and Methods Results Discussion References

GF_m. The proteins found in GF_m are purified as one mixture from a noncollagenous protein extract of bovine femurs as detailed previously (Poser and Benedict, 1994). Briefly, mid-diaphyseal segments of bovine femurs are cleaned, pulverized, and demineralized. Proteins of apparent molecular weights between 10,000 and 100,000 are extracted by ultrafiltration, lyophilized, and purified by reverse-phase high-pressure liquid chromatography (HPLC). The protein mixtures containing active ingredients for bone formation used in this study have been shown to elute between 35.0 and 37.1% acetonitrile (v/v) (Poser and Benedict, 1994). Fractions eluting at slightly higher volume percentages have been shown to be inactive for bone formation (inactive bone protein, IBP) and were used in the present study as negative controls.

GF_m Content Analysis. One- and two-dimensional polyacrylamide gel electrophoresis (PAGE), HPLC, mass spectrometry, and Western blot analyses were used to investigate the contents of GF_m. PAGE analysis was conducted according to standard techniques. Tropomyosin (33 kDa, pI 5.2) and lysozyme (14 kDa, pI 10.5-11) were added to the samples as internal markers. Gels were stained with Coomassie Blue, and protein spots were excised from dried gels for mass spectrometry and sequence analysis.

Studies in Chorioallantoic Membranes (CAM). The angiogenic activity of GF_m was compared with that of a range of concentrations of recombinant human bFGF (10 µg/ml) and VEGF (10 µg/ml) or their combination (5 µg/ml each) (proteins obtained from R & D Systems Inc., Minneapolis, MN), as well as their carrier (povidone), in quail CAMs (n = 6 per group) (Parsons-Wingerter et al., 2000). Previous studies have shown that vascular growth response in CAMs to either bFGF or VEGF at concentrations of 10 µg/ml are on the plateau of the respective dose-response curves (Parsons-Wingerter et al., 1998, 2000). Fertilized Japanese quail eggs (Coturnix coturnix japonica) were opened into Petri dishes on day 3 postfertilization. After 4 days in culture at 37°C (i.e., day 7), the growth factors were solubilized in prewarmed carrier solution and added in a total volume of 0.5 ml to each embryo. The test material was evenly distributed on the surface of the CAM, which was cultured for 24 h at 37°C. Embryos were fixed in 4% paraformaldehyde/2% glutaraldehyde solution in phosphate-buffered saline. The CAMs were dissected from the embryos and were mounted on glass slides. Digital images were acquired at 10× magnification with a computer-supported digital camera attached to a microscope. The fractal dimension (Df) of each image was determined with previously validated software (Parsons-Wingerter et al., 1998). The Df (baseline) for a day 7 CAM is 1.372 (Parsons-Wingerter et al., 1998). The amount of vascular growth in CAMs treated for 24 h was reported as the percentage of change in Df relative to control (povidone-treated) CAMs.

Pilot Study in Canine Myocardium. Six adult mongrel dogs (20-25 kg) were anesthetized with thiopental sodium (15 mg/kg, i.v.) and maintained with 0.5 to 2.0% inhaled isoflurane. Via a left lateral thoracotomy, the proximal left anterior descending artery (LAD) was isolated and an ameroid constrictor was placed to induce ischemia over time. The dogs received intramyocardial growth factor injections of GF_m diluted in povidone (1 mg/ml, n = 4) or IBP (1 mg/ml; n = 2). All injections were 0.15 ml and were spaced ~1 injection/cm². Injections were made in both the anterior (ischemic) and posterior (normal) walls with five to nine injections performed in each region. Animals survived for either 2 or 6 weeks. All animals received postoperative bromodeoxyuridine (BrdU) injections (schedule provided below). Dogs were euthanized with pentobarbital (100 mg/kg), hearts were removed, and transmural tissue blocks were submitted for histologic and immunohistochemical evaluation (Masson's Trichrome stain; BrdU; smooth muscle actin; von Willebrand factor).

Efficacy Study in Chronically Ischemic Canine Myocardium. A randomized, blinded, placebo-controlled study was performed. An ameroid constrictor was placed in 21 adult mongrel dogs (20-30 kg) to create chronic ischemia. To minimize collateral flow in the acute setting, all visible epicardial obtuse marginal or posterior branches seen to connect with the LAD or LAD diagonal vessels were ligated (4-0 polypropylene sutures). One silicon tube (Tygon, Cardiovascular Instrument Corp., Wakefield, MA) was chronically implanted into the left atrium and another into the descending aorta.

Statistics. Data were expressed as mean ± S.E.M. Between-group comparisons of ordinal data were performed with a Kruskal-Wallis test followed by repeated Mann-Whitney U tests to determine individual differences. Within-group differences were tested by the Friedman test; if significant, this was followed by repeated Wilcoxon tests. Between-group comparisons of continuous variable were done by one-way analysis of variance followed by Scheffe post hoc test. Within-group comparisons were done by paired samples t test. p < 0.05 was considered statistically significant.

	Results

Top Abstract Introduction Materials and Methods Results Discussion References

GF_m Composition. PAGE analysis of GF_m included 13 major bands, representing 65% of the protein by weight. These major bands have been identified by mass spectrometry. Two major components are BMP-3 and TGF--2. Other identified proteins that are probably not contributors to angiogenic activity include three that are related to histone H1.1 and three that matched with the ribosomal proteins S20, L6, and L32. Also identified were cathepsin L and proteins related to -2-macroglobulin receptor-associated protein, retinoic acid receptor responder protein 2, secreted phosphoprotein 24, and lysyl oxidase-related protein.

Quail CAM Assay. Compared with povidone alone, CAMs exposed to GF_m for 24 h showed a greater vascular density and more vessel branchings (Fig. 1, A and B). The rate of vascular growth in CAMs subjected to povidone alone was similar to that of control CAMs (data not shown) and was set at 100% (Fig. 1C). The rate of vascular growth was approximately doubled with bFGF, VEGF, or their combination. GF_m at concentrations of 1 or 10 mg/ml elicited a slightly greater (although not statistically significant) vascular growth relative to that stimulated by bFGF, VEGF, or their combination.

View larger version (76K): [in this window] [in a new window]   Fig. 1.   Effects of growth factors on vascular growth on CAMs of fertilized Japanese quail eggs. Representative digitized pictures of CAMs exposed to either placebo (A) or GFm 10 mg/ml (B). C, summary of quantitative image analysis of percentage of change in fractal dimension in response to different growth factors (Df, mean ± S.D., n = 6 in each group).

Pilot Study. In GF_m (1.0 mg/ml)-treated animals allowed to survive for 2 weeks, large, BrdU-positive, conduit-sized vessels with diameters up to 300 µm could be detected in areas surrounding the injection sites, both in ischemic and nonischemic areas of the heart (Fig. 2, A-D). Six weeks after treatment with GF_m, numerous well organized large vessels as well as smaller arterioles and capillaries in the surrounding myocardium exhibited abundant BrdU incorporation (Fig. 2, E-H). Compared with nontreated ischemic tissue, GF_m-treated ischemic myocardium exhibited a greater than 5-fold increase in BrdU-stained, newly formed arterioles, with diameters 50 µm containing 2 BrdU-positive cells (6.6 ± 4.0 versus 1.3 ± 1.8 vessels/cm², p < 0.05). In contrast, IBP (1.0 mg/ml, an inactive protein fraction derived from bovine bones) failed to induce any significant growth of vessels larger than capillaries either close to the injection sites or in remote areas after 2 weeks (Fig. 3, a and b), although the inflammatory response at the injection site appeared similar with all the factors. These observations suggest that GF_m can induce large vessel growth and the major component is not due to inflammation.

View larger version (93K): [in this window] [in a new window]   Fig. 2.   Masson's Trichrome-stained section (A, 40× original magnification) showing large vessel with several side branches 2 weeks after GFm injection into ischemic myocardium. As shown in adjacent sections, immunostained with antibodies against von Willebrand factor (B) and smooth muscle actin (C), the vessel is lined by endothelium surrounded by poorly organized layers of smooth muscle cells. That von Willebrand factor staining was not limited to the endothelial layer provides another indication that this vessel was not fully organized. Nuclei staining for proliferating cell nuclear antigen (200×, D) is indicative of neovascularization. Similar vessels were identified in both the ischemic and nonischemic areas of these hearts. Six weeks after ameroid placement and GFm injection, relatively large but apparently mature vessels were seen (E-H). Evidence of vascular growth is based on the abundant BrdU incorporation seen in the endothelial and smooth muscle cells.

View larger version (116K): [in this window] [in a new window]   Fig. 3.   Reactive inflammatory response after treatment with inactive bone protein (IBP, 1 mg/ml) shows scarring with no significant growth of vessels larger than capillaries (a and b, original magnification 40×).

Effects in Chronic Myocardial Ischemia. Histologic samples were examined and graded for vascular growth outside the scar observed at the injection site according to a semiquantitative overall vascular growth index (0, no angiogenic response; 1+, mild-to-moderate angiogenic response; 2+, significant angiogenic response). According to this analysis, as summarized in Table 1, GF_m induced a robust concentration-dependent neovascular response in ischemic canine myocardium. The size of the scar was also concentration-dependent.

View larger version (110K): [in this window] [in a new window]   Fig. 4.   Sequence of in vivo angiograms from a dog's heart at baseline (3 weeks after ameroid constrictor placement; A and B) and 6 weeks after treatment (C and D) with GFm (10 mg/ml). At baseline, the ameroid was completely occluded and the LAD demonstrates minimal filling, both early (A) and late (B) during the contrast injection. Six weeks after treatment, the ameroid was again seen to completely occlude the LAD (C), but the LAD showed complete reconstitution late in the injection (D). A newly formed collateral (C1) was detected surrounding the experimentally occluded segment. LADD, diagonal branch; LCx, circumflex artery; RVb, right ventricular branch.

View larger version (96K): [in this window] [in a new window]   Fig. 5.   Sequence of representative ex vivo angiograms taken at the terminal experiment 6 weeks after treatment with GFm (10 mg/ml; A and B) or povidone (placebo; C and D). The contrast agent was injected via a catheter engaged in the left main coronary artery, revealing the left circumflex coronary artery (LCx) and a right ventricular branch (RVb). Ameroid constrictors completely occluded the LAD in both cases. After GFm treatment, the LAD opacified early during the injection (A) via two collateral vessels (C1 and C2). Later during the injection (B), the entire LAD was seen. With placebo, the distal LAD is not seen early (C) and is barely evident late during the injection (D).

View larger version (42K): [in this window] [in a new window]   Fig. 6.   Hearts were cut into 5 layers (A-E, base to apex) to produce 32 transmural blocks (A). Blood flow at each time point at rest and during vasodilation (adenosine) of each segment was plotted as a function of segment number to reveal ischemic and normally perfused segments (B). This result, obtained from a dog treated with GFm 10 mg/ml, shows no significant difference between baseline and 6 weeks after treatment. Results from endo- and epicardial layers (analyzed separately) were subsequently pooled because they were highly similar.

	Discussion

Top Abstract Introduction Materials and Methods Results Discussion References

GF_m, a protein mixture derived from bovine long bones, is an effective angiogenic agent in quail CAMs and canine myocardium. Several of the identified ingredients of GF_m are among the list of known angiogenic factors, and synergism between some GF_m components (e.g., BMP-7, bFGF, and TGF--1) has already been demonstrated in the CAM assay (Ramoshebi and Ripamonti, 2000). In ischemic myocardium, GF_m treatment was associated with growth of vessels with diameters as large as 300 µm with abundant BrdU incorporation. Analysis of the composition of GF_m confirmed the presence of several known angiogenic growth factors, such as bone morphogenic proteins BMP-2 through 7, TGF--1 through 3, as well as FGF-1. However, approximately 30% of the protein mass is unidentified. Thus, the observed effects could be related to any one or combination of the known growth factors, or they could reflect activity of as yet unidentified and possibly unknown factors. However, an inflammatory response is not the predominant mechanism underlying the effects of GF_m since similar vascular growth responses were not observed with IBP, which did incite an inflammatory response similar (histologically) to GF_m. However, although not sufficient, this does not exclude the possibility that an inflammatory response is necessary for the angiogenic response of GF_m to occur.

Angiographic findings in the chronically ischemic dog hearts provided additional evidence of large vessel growth. All ameroid constrictors were completely occluded so that at the baseline ischemic evaluation, there was only faint visualization of the distal LAD during contrast agent injections. A GF_m concentration-dependent increase in distal LAD opacification (blinded analysis) was observed, suggesting that collateralization to the distal LAD was improved. This was associated with the appearance of new collateral vessels either bridging around the ameroid constrictor or connecting marginal branches of the circumflex artery to diagonal vessels of the LAD. These vessels, which were particularly evident on ex vivo angiograms, took circuitous courses that are typical of new vessels.

As identified in prior studies of dogs (Unger et al., 1993) and pigs (Giordano et al., 1996), resting blood flow 3 weeks after ameroid constrictor placement was not significantly decreased compared with the normally perfused region, the result of natural collateralization. Since distal LAD opacification was poor at this time point, it is evident that normalization of resting flow is due to natural angiogenesis that occurs in the setting of chronic ischemia (Ware and Simons, 1997). However, despite the histologic and angiographic findings indicating the presence of large vessel growth, blood flow to the anterior wall was not improved during vasodilatory stress induced by adenosine. Distal LAD opacification on resting angiography does not in any way indicate the maximal blood flow capacity into the vascular bed; it merely indicates that new anatomic connections exist. This is completely analogous to the common clinical experience where, although a totally occluded epicardial vessel can frequently be visualized angiographically by flow of contrast agent from collateral vessels, a myocardial perfusion defect is typically observed during stress on a nuclear perfusion imaging study.

Thus, while GF_m effectively induces large vessel growth, it could be that the delivery strategy (direct mid-myocardial injections at ~1-cm interinjection spacing) may not be optimal for improving maximal blood flow to ischemic myocardium. Use of a delivery method to ensure that new vessels grow from well perfused normal epicardial vessels neighboring the ischemic territory [e.g., an intra-arterial injection strategy (Giordano et al., 1996) or preferential subepicardial delivery] might be more effective.

The application of single exogenous growth factors either as protein or gene therapy has thus far been the standard in clinical trials of therapeutic angiogenesis. However, angiogenesis is not a process induced, sustained, or completed by a single molecule (Folkman and Klagsbrun, 1987; Beck and D'Amore, 1997; Coussens et al., 1999; Carmeliet, 2000; Carmeliet and Jain, 2000). Published preclinical studies of single growth factor strategies, whether with proteins or gene therapies, whether with FGF or VEGF, whether injected into the muscle or into an artery, have never demonstrated complete normalization of blood flow to the ischemic territory; such results are summarized in recent reviews (Ware and Simons, 1997; Simons et al., 2000). Clinically, although results of a few small (some unblinded) studies of single growth factor strategies have provided encouraging results (Schumacher et al., 1998; Vale et al., 2001; Laham et al., 1999), results from two relatively large-scale multicenter, double-blind studies now available have been negative; specifically, the VIVA trial of intracoronary and intravenous VEGF (Henry et al., 1999) and the FIRST trial of intracoronary bFGF (M. Simons, unpublished observations). Even in the unblinded clinical study, the extent of revascularization was incomplete, as has been observed in the animal studies. There are multiple possible explanations for variable results between animal studies and for the negative results reported in the earlier noted clinical trials. The fact that single growth factor strategies were used is among the possibilities (mode of delivery being another important factor), but it would be premature to conclude at this time that this is the decisive factor.

Conclusions. The choice of growth factor(s) and delivery strategies for therapeutic angiogenesis are currently the topic of intensive research. Some clinical trials based upon positive preclinical studies have provided negative results (see for example, Henry et al., 1999; Simons et al., 2000) warning about the potentially limited power of preclinical studies to predict clinical utility. Nevertheless, the present results indicate that GF_m, a mixture of growth factors, induces the formation of relatively large vessel. Lack of improved blood flow in treated animals does not diminish the importance of the histologic images of large new vessels seen in response to GF_m treatment. This does highlight the fact, however, that there are limitations to the overall strategy used in the present study. This could relate to the use of an intramyocardial route of administration, the density of injections, the dose of GF_m, or other unidentified factors that may also have impacted the results. As such, the present results serve to emphasize the point that, although necessary, it is not sufficient for an angiogenic therapy to induce vascular growth. The overall strategy, which includes the substance and its mode of delivery, must ensure that the new vessels become part of an effective vascular bed.