Objective: Animal models of hindlimb ischemia are critical to our understanding of peripheral vascular disease and allow us to evaluate therapeutic strategies aimed to improve peripheral collateral circulation. To further elucidate the processes involved in revascularization following ischemia, we evaluated the temporal association between tissue ischemia, vascular endothelial cell growth factor (VEGF) release, angiogenesis (capillary sprouting), arteriogenesis (growth of the larger muscular arteries), and reserve blood flow (functional collateral flow). Methods: New Zealand White rabbits (male 3–4 kg) were evaluated at specific days (0, 5, 10, 20 or 40) following femoral artery removal for measurement of hindlimb blood flow, skeletal muscle lactate production and VEGF content, capillary density (a marker of angiogenesis), and angiographic score (a marker of arteriogenesis). Results: Maximal capillary sprouting occurred within 5 days of femoral artery removal and was temporally associated with reduced resting hindlimb blood flow, increased lactate release and detectable levels of skeletal muscle VEGF. The growth of larger angiographically visible collateral vessels occurred after 10 days and was not temporally associated with ischemia or skeletal muscle VEGF content, but did coincide with a large functional improvement in the reserve blood flow capacity of the limb. Conclusions: Following femoral artery removal in the rabbit, the time course of angiogenesis and arteriogenesis were clearly distinct. Tissue ischemia and/or VEGF may stimulate capillary sprouting, but this response does not translate to a significant improvement in collateral flow. The growth and development of the larger collateral vessels was correlated with a large functional improvement in collateral flow, and occurred at a time when VEGF levels were undetectable.
Time for primary review 21 days.
Peripheral vascular disease (PVD) has been reported to affect 7.5% of the population aged 60–64 in the United States . In any given year, PVD is responsible for 200 lower limb amputations per million in the non-diabetic population and 3900 per million in diabetic patients . When surgical treatment is not an option, the clinical manifestations of the disease are dependent upon the balance between the rapidity and extent of ‘natural’ collateral vessel growth versus the progression of the occlusive arterial disease. Therefore, therapies aimed to improve peripheral blood flow by stimulating the growth and development of collateral vessels could translate to substantial improvements in mobility and exercise tolerance for patients with intermittent claudication or critical limb ischemia.
The terms ‘spontaneous neovascularization’ and ‘autocollateralization’ have been used by others to describe the natural growth and development of collateral vessels that occurs in the ischemic tissue of animals subjected to permanent arterial ligation or removal [3,4]. The sprouting of new capillaries, referred to as angiogenesis, is clearly different from the development of preexisting arterial connections into true collateral vessels, a process now termed arteriogenesis [5–8].
Mounting evidence has suggested that hypoxia-induced vascular endothelial growth factor (VEGF) plays an important regulatory role in angiogenesis in both physiological and pathological conditions. Upregulation of VEGF in vivo has been demonstrated in various animal models of ischemia including the heart , brain , and lung . It remains unclear however, whether VEGF is a stimulus for both angiogenesis and arteriogenesis, and furthermore, which of these processes contribute to, or are necessary to improve collateral blood flow following an arterial obstruction or occlusion.
The purpose of this study was to gain a better understanding of the processes that contribute to improve collateral blood flow in response to an arterial obstruction. This was accomplished by examining the temporal association of ischemia, VEGF production, angiogenesis and arteriogenesis within the hindlimb following femoral artery removal. Although there has been widespread interest in therapeutic angiogenesis for the treatment of peripheral vascular disease, an understanding of the mechanisms involved in revascularization is needed before we can gain insight into how the growth and development of collateral vessels could be enhanced for therapeutic benefit.
2.1 Animal model
Sixty-three male New Zealand White rabbits (3–4 kg, Covance)were used for the completion of this study. All procedures related to the use of animals were approved by the Institutional Animal Care and Use Committee at Merck Research Laboratories, West Point, PA and conform with the Guide for the Care and Use of Laboratory Animals published by the US National Institutes of Health (NIH Publication No. 85-23, revised 1996). The initial surgery to induce hindlimb ischemia has been described by others . Briefly, the rabbits were pre-medicated with xylazine (10 mg/kg, i.m.) and anesthetized with ketamine (70 mg/kg, i.m.). Under sterile surgical conditions, a longitudinal incision was made on the medial thigh of one hindlimb extending from the inguinal ligament to a point just proximal to the patella. Through this incision, using surgical loops, the entire length of the femoral artery was dissected free, as were all major branches of the femoral artery including the inferior epigastric, deep femoral, lateral circumflex, and superficial epigastric arteries. The external iliac artery, as well the arteries mentioned above, were ligated. The femoral artery was completely excised from its proximal origin as a branch of the external iliac to the point distally where it bifurcates into the saphenous and popliteal arteries. The excision of the femoral artery results in retrograde thrombus propagation to the origin of the external iliac artery. Therefore, blood flow to the hindlimb after excision of the femoral artery is dependent upon flow through the internal iliac artery. The incision was closed in three layers with 3.0 silk. All rabbits were closely monitored by veterinary staff and received an analgesic (buprenorphine, 0.04 mg/kg s.c., b.i.d.) and antibiotic (enrofloxacin, 6.0 mg/kg, s.c.) for 5 days following surgery.
2.2 Experimental groups
Rabbits were evaluated either naïve (non-ischemic, n = 9), acutely after the surgical induction of ischemia, (day 0, n = 10), or at specific days post-surgery (day 5, n = 11; day 10, n = 10; day 20, n = 12 or day 40, n = 11).
2.3 Measurement of resting and reserve hindlimb blood flow
The resting and reserve hindlimb blood flows were determined for each experimental group. Anesthetized rabbits were placed in the supine position on a temperature-controlled heating pad and a tracheotomy was performed that allowed the animals to be mechanically ventilated with room air (30–35 breaths/min, tidal volume of 15 ml). A 5-cm midline incision was made in the lower abdomen through the skin and underlying abdominal muscles. The descending aorta was isolated and cleared at the point of bifurcation into the common iliac arteries. The common iliac arteries were then cleared and both epigastric arteries were ligated to eliminate blood flow to the lateral compartments of the abdomen which would complicate the measurement of hindlimb blood flow. Perivascular flow probes (1.5 RB, Transonic Systems, Inc.) were placed on the right and left iliac arteries to continuously measure blood flow to both limbs simultaneously.
To obtain a functional measurement of collateral blood flow, the hyperemic response following a series of acute arterial occlusions was measured. It is known that the increase in blood flow following abrupt arterial occlusion (i.e. reactive hyperemia) represents the reserve blood flow capacity and is related to both the diameter and number of microvessels and collateral arteries within the tissue(s) whose blood supply is dependent upon the artery that was occluded [13,14]. For this test, hindlimb blood flow was interrupted by temporarily applying an arterial clip on the descending aorta at a position just proximal to the common iliac bifurcation. Upon removal of the clip, a rapid increase in blood flow was observed that gradually returned to resting values within 20–30 s. Three separate tests of increasing duration of arterial occlusion (i.e. 10, 20, and 40 s) were performed in duplicate with 10 min allowed between tests for the equilibration of blood flow. Using a data acquisition program (MacLab Instruments), the total area under the hyperemic response curve during the first 30 s after release of the arterial clip was calculated and the baseline flow during the time of integration was subtracted out; the data is presented as bulk flow (ml).
2.4 Measurement of hindlimb lactate release
Following the measurements of reserve flow, a blood sample from the external iliac vein was obtained by direct withdrawal using a 22G, 0.5 in. needle attached to a 1-cc syringe. An arterial sample was also collected from the cannula inserted into the carotid artery. The concentration of lactate in the plasma was determined by an enzymatic assay (Sigma Diagnostics) read on a spectrophotometer at 540 nm. The data was standardized for blood flow (ml/min) and lactate release from the hindlimb is presented as mg lactate/min.
2.5 Measurement of VEGF protein
Following femoral artery ligation in the rabbit, the adductor muscle has previously been shown to be one of the most severely affected muscles in the hindlimb . A tissue sample of the adductor muscle was removed and homogenized at each time point and the amount of VEGF was measured with a chemiluminescent sandwich ELISA (R&D Systems). In this assay, tissue homogenates were assayed on plates coated with a murine monoclonal antibody against VEGF. After washing, the secondary polyclonal antibody (conjugated to horseradish peroxidase) was added, followed by the addition of a hydrogen peroxide/luminol mixture. The intensity of the color was measured at 450 nm. VEGF protein levels were normalized to the total amount of protein, determined in duplicate with BSA as the standard (Bio-Rad Protein Assay Kit, Bio-Rad Laboratories) and expressed as picograms per milligrams of total protein.
2.6 Internal iliac angiography
Angiography of the hindlimb was performed to obtain an anatomical measurement of the growth and development of the larger conduit vessels in the medial portion of the upper limb (i.e. the original location of the femoral artery) following the induction of ischemia. After femoral venous blood sampling, a polyethylene catheter (PE-100) was inserted into the common iliac artery of the ischemic limb, with the tip of the catheter positioned just proximal to the origin of the internal iliac artery. It has been reported by others and confirmed in our laboratory that the ‘corkscrew’ collateral arteries in the medial thigh region of the ischemic limb originate from the internal iliac artery [5,16]. The hindlimb of the rabbit was positioned 20 cm below the output beam of a fluoroscope (General Electric, Stenoscop). Immediately following an intraarterial bolus injection of the vasodilator sodium nitroprusside (300 μg in 1 ml saline), iodinated contrast media (Isovue-370) was infused intraarterially at a constant rate of 60 ml/min for 5 s. Perfusion of the hindlimb was observed on a monitor in real time and an angiographic image was taken exactly 4 s after the start of contrast media infusion. The angiographic images were analyzed using the methods described by others to quantitatively determine the amount of collateral vessels [4,16]. Briefly, a grid overlay that comprised 2.5 mm diameter circles arranged in rows 5 mm apart was placed over the 4-s angiographic image. The number of contrast-opacified vessels intersecting a circle was counted by two independent investigators blinded to the experimental groups (inter-observer variability was <3%). The angiographic score therefore represents the number of circles containing a vessel, divided by the total number of circles in the medial thigh region.
2.7 Histological determination of capillary density
Immediately after the animal was sacrificed, the hindlimbs were dissected and samples of the adductor muscle were removed for histological evaluation. The tissues were placed in a plastic cassette and covered with O.C.T. compound before freezing for 30 s in a bath of liquid nitrogen. Multiple frozen sections were cut (10 μm thickness) on a cryostat and placed on glass slides. Tissue sections were stained for alkaline phosphatase using the indoxyl-terazolium method to detect capillary endothelial cells as previously described [17,18] and counterstained with eosin. The number of capillaries were counted under 20× objective using an image analysis system (Phase 3 Imaging). The system automatically identified and counted the number of capillaries in a chosen field (a total of 60 fields from six different sections were analyzed per muscle) and also determined the total area of muscle. Hence, capillary density was calculated as the number of capillaries/mm2 of muscle.
2.8 Statistical analysis
All data are presented as mean±S.E.M. Comparisons within the same animal were performed using a paired Student's t-test. For intergroup comparisons, a one-way ANOVA followed by an unpaired Student's t-test was used. Probability values of less than 0.05 were required for assumption of statistical significance.
3.1 Resting blood flow, lactate release and VEGF response
Resting blood flow in the right and left hindlimbs of an anesthetized naive rabbit were found to be 15.2±0.7 and 16.7±0.8 ml/min, respectively. As can be seen in Fig. 1, following femoral artery removal (day 0), there was an immediate reduction (∼60%) in the resting blood flow to the operated limb (6.1±0.8 ml/min), which remained significantly lower than the pre-surgery value for 5 days. However, when measured 10 days after surgery, resting blood flow to the operated limb had recovered to 15.8±1.7 ml/min and was no longer significantly different from the blood flow to the contralateral normal limb (17.4±1.4 ml/min) or the resting blood flow to the operated limb before femoral artery removal. The resting blood flow of the operated limb was sustained at normal levels when measured 20 (17.8±2.0 ml/min) and 40 (17.5±1.4 ml/min) days post-surgery.
Venous lactate concentration is commonly used as a metabolic marker of tissue hypoxia . Following femoral artery removal, hindlimb lactate release increased significantly from 1.50±0.18 to 2.48±0.11 mg/min (Fig. 2). There was a moderate reduction in lactate release when measured on day 5 (1.97±0.21 mg/min), however, this value was still significantly elevated compared to the control group. When measured at 10, 20 and 40 days, there was no significant difference in hindlimb lactate release compared to pre-surgery.
Using the ELISA described, VEGF protein was undetectable in normal non-ischemic skeletal muscle (0/11=number of rabbits with a detectable level of VEGF/total number of rabbits in the group). On day 5 after surgery, when resting blood flow was still reduced and lactate release was increased, there were detectable levels of VEGF in the homogenates from the ischemic muscle (11/11, mean=0.68±0.09 ng/mg protein, P<0.01 versus all other groups, Fisher's exact test). The time course and peak levels (0.48 ng/mg)  of VEGF expression are similar to what others have observed in rat models of hypoxic pulmonary hypertension  and brain infarction . Aside from the day 5 group in which all 11 rabbits, had a measurable level of VEGF, there was no detectable levels of VEGF at any of the other post-surgery time points: day 10, 0/10; day 20, 0/12; and day 40, 0/11.
3.2 Angiogenic response following hindlimb ischemia
The changes in skeletal muscle capillary density measured in response to femoral artery removal are graphed in Fig. 3. There was no acute change in the capillary density following the induction of hindlimb ischemia (non-ischemic, 216±24 caps/mm2 vs. acutely ischemic, 232±18 caps/mm2); however, there was a significant increase measured 5 days post-ischemia (351±36 caps/mm2). Interestingly, capillary density then decreased back towards pre-ischemia values by 10 days (260±22 caps/mm2) and was not significantly different from non-ischemic muscle (i.e. control rabbits) when measured on days 20 (228±17 caps/mm2) and 40 (222±37 caps/mm2) post-ischemia.
3.3 Arteriogenesis following hindlimb ischemia
Angiographic images of the rabbit hindlimbs and the corresponding angiographic scores are shown in Figs. 4A–F and 5, respectively. The hindlimbs of non-ischemic rabbits (i.e. with the femoral artery intact) revealed very few visible collateral vessels in the medial thigh (Fig. 4A). Within hours after femoral artery removal, there was a small and insignificant increase in the angiographic score (day 0=16.1±2.5), which may reflect an endogenous collateral network in the upper limb. There was no significant improvement in angiographic score during the first 10 days after the induction of hindlimb ischemia, however there was a dramatic increase observed between day 10 (Fig. 4D) and 20 (Fig. 4E) post-ischemia, with no significant change thereafter.
3.4 Measurement of functional collateral blood flow
The reserve blood flow capacity of a tissue is dependent upon both the number and diameter of microvessels and collateral arteries; therefore, this measurement was used to obtain a functional assessment of revascularization. In normal non-ischemic rabbits, there was a direct relationship between the duration of the aortic occlusion (AO) and the bulk flow as shown in Fig. 6 (10 s AO, 4.3±0.6 ml; 20 s AO, 6.1±0.7 ml; 40 s AO, 7.5±1.0 ml). Following femoral artery removal (i.e. day 0), there was a reduction in the reserve flow capacity of the hindlimb; bulk flow decreased 80% to 0.8±0.8 ml, 85% to 0.9±0.2 ml, and 88% to 0.9±0.2 ml in response to the 10, 20 and 40 s AO, respectively. In contrast to resting flow which was reestablished by 5 days (Fig. 1), there was no significant improvement in reserve flow on day 5. There was, however, a large increase (∼100%) in reserve blood flow observed on day 20 (10 s AO 2.1±0.3; 20 s AO 3.0±0.4 ml; 40 s AO 4.1.±0.5 ml). Although not as great as the improvement between days 5 and 20, reserve flow continued to increase when measured on day 40. The responses measured on day 40 (10 s AO 2.5±0.3; 20 s AO 3.6±0.3 ml; 40 s AO 4.5±0.4 ml) also revealed that there was still a significant reduction compared to the non-ischemic group; thus reserve flow was not restored during the 40 day time course of this study.
The results from this study enhance our understanding of the relationships between ischemia, angiogenesis, arteriogenesis and functional improvements in collateral blood flow. We have shown that both angiogenesis and arteriogenesis occur in the rabbit hindlimb following femoral artery removal. However, the time course of each process, their temporal association with tissue ischemia, VEGF production, and their contribution to improved collateral blood flow were clearly distinct. The angiogenic response was found to occur rapidly, within the first 5 days, and was associated with ischemia and increased levels of skeletal muscle VEGF. The growth of larger collateral arteries was observed in the absence of resting ischemia, occurred at a time when VEGF levels had decreased, and was correlated with a dramatic increase in the functional reserve flow capacity of the limb.
Angiogenesis is defined as the growth of new capillaries and should be used to describe the process whereby preexisting capillaries proliferate and sprout to form new capillary networks. This complex phenomenon consists of several distinct processes that include endothelial cell proliferation, migration, and differentiation, as well as the degradation and subsequent reestablishment of the basement membrane. Importantly, these newly formed capillary tubes lack vascular smooth muscle cells have little or no effect on total vascular resistance. Arteriogenesis, however, involves the growth, development and remodeling of preexisting vessels, mainly arterioles, into larger collateral arteries. These vessels serve as conduit arteries and thus have the ability to influence vascular resistance.
Our results strongly support the efforts of Schaper's group to clarify the terms angiogenesis and arteriogenesis to describe the two different types of vascular growth processes that occur in response to ischemia [5,21]. In contrast to the limited post-ischemia time points and end-point measurements reported by that group, our study employed a more extensive time course and a combination of several biochemical, metabolic, anatomic and hemodynamic measurements to allow for a more complete representation of the interplay and temporal association of these variables in response to hindlimb ischemia. In addition, this study also examined the potential role of VEGF as a potential mediator of angiogenesis and/or arteriogenesis.
The temporal association observed between capillary sprouting and ischemia during the first 5 days of this study suggests that angiogenesis is probably mediated by an endogenous factor(s) released from ischemic tissue. Although many different cell types, including keratinocytes, corneal fibroblasts and monocytes are capable of producing a variety of angiogenic factors in response to low oxygen, VEGF has received the most attention due to its mitogenic specificity for endothelial cells [22,23], and the upregulation of its receptors VEGFR-1 (Flt-1) and VEGFR-2 (Flk-1 or KDR) following the induction of hypoxia [11,24]. Furthermore, additional studies have shown that VEGFR-1, and VEGFR-2 are required for normal embryological development of the vascular system [25,26]. Recent data has suggested, however, that VEGF may not exclusively be an endothelial specific mitogen and angiogenic factor, but rather a non-specific stress-induced gene that affects endothelial cell survival [27,28].
In contrast to the stimulus for angiogenesis, our results suggest that the stimulus for arteriogenesis is probably not VEGF, or any other biochemical factor released in response to ischemia. In this study, the largest increase in angiographic score occurred at a time (between days 20 and 40) in which resting blood flow was normalized and therefore suggests that the stimulus for arteriogenesis must be present beyond the time of ischemia or have a delayed onset of action. This finding of a dissociation of ischemia and collateral growth is complemented by the study of Paskins-Hurlburt and Hollenberg  who observed a temporal, but not spatial, dissociation of ischemia and collateral growth in the rat hindlimb. Following the induction of hindlimb ischemia, and using angiographic methods similar to ours, this group observed continued vascular growth despite normalization of perfusion .
Rather than an ischemia-related mediator, there is evidence the stimulus for arteriogenesis may be mediated by alterations in mechanical pressure or shear force and the subsequent invasion and activation of circulating monocytes [30,31]. Stenosis or occlusion of a major artery invokes the redirection of blood flow through existing vessels that otherwise were not exposed to the sustained hemodynamic conditions of increased flow velocity and shear forces. Increased shear stress and cyclic strain are mechanical events known to activate the endothelium leading to the attraction, adhesion and activation of circulating cells, mainly monocytes [32,33]. The involvement of monocytes is supported by data that showed local infusion of monocytes chemoattractant protein-1 (MCP-1) caused an increase in collateral conductance following femoral artery ligation in the rabbit . Activation of monocytes and secretion of angiogenic cytokines by this process could thereby contribute to the proliferation, growth, and remodeling of vessels into mature conduits for collateral flow in the absence of a resting blood flow deficiency (i.e. ischemia).
Our results showed that the early angiogenic response did not translate to an improvement in collateral blood flow (i.e. reserve blood flow). On the other hand, the arteriogenic response observed between days 20 and 40 post-ischemia did correlate with a large functional improvement in reserve blood flow. These results therefore provide evidence that angiogenesis, in its strict definition, cannot replace or functionally compensate for an occluded artery. Compensation to an occluded artery can only come from low-resistance connections between a donor artery in a non-ischemic region and the post-occlusion arterial system of the recipient ischemic region (i.e. arteriogenesis).
The idea that angiogenic growth factors such as VEGF and bFGF, or their genes, could have therapeutic value for the treatment of ischemia related disorders such as peripheral vascular disease and coronary artery disease remains to be proven. Several clinical phase I trials suggest the feasibility and short-term safety with these growth factors. However, the VIVA trial, which is the only phase II trial published in this field, delivered VEGF protein by intracoronary infusion followed by several intravenous infusions and found no VEGF-related increases in exercise time or reductions in angina versus placebo . It has been suggested by others and now supported by our findings, that VEGF probably does not stimulate the growth and development of the larger collateral vessels, those vessels that would most likely impart a functional improvement.
In summary, our results suggest that compared to angiogenesis, arteriogenesis is an ischemia-independent process and a much more efficient mechanism to compensate for the gradual or intermittent occlusion of a major artery. These findings directly support the suggestion by Ito et al.  that although capillary sprouting may deliver some relief to the underperfused territory, only true collateral arteries are principally capable of providing large enough amounts of blood flow to the ischemic areas at risk for necrosis or loss of function.
The authors wish to express their gratitude to John Gehret, Tamara Montgomery, Kelly Sandavov and Ashliegh Bone for providing expert veterinary care.