University of Toronto; the Divisions of Pathology and Cardiovascular Research, Research Institute, The Hospital for Sick Children; and The Toronto Hospital Research Institute,¶ Toronto, Ontario, Canada
In vitro studies have demonstrated that smooth muscle cells in culture respond to cyclic changes in strain of the substrate on which they are grown. Pulsatile strain up-regulates mitogenesis and release of fibroblast growth factor (FGF)-2, and it alters actin, myosin, and h-caldesmon expression; 10-16 however, it has been impossible to reproduce in cell culture the three-dimensional strain and complex matrix interactions that smooth muscle cells experience in vivo. Thus, the role that cyclic mechanical strain plays in arterial remodeling in physiological and pathological conditions has not been fully elucidated.
To study the role of wall tension and strain on arterial remodeling, we have developed a model in which an external cuff was used to reduce vessel diameter and eliminate cyclic strain produced by pulsatile pressure in the rabbit abdominal aorta. A marked medial atrophy, smooth muscle cell loss, and DNA fragmentation was observed under the cuff, implying that a reduction of physiological strain levels can be transduced into signals that induce apoptotic cell death. Thus, mechanical signals play an important role in determining cell number and structural remodeling.
Surprisingly, reducing circumferential strain after balloon denudation did not effect the development of a neointima even in the presence of severe medial atrophy. Thus, intimal proliferative responses appear to be independent of tensile loads and independent of extensive medial remodeling and atrophy.
A total of 56 New Zealand White rabbits were premedicated with intramuscular injections of Ketamine (100 mg) and Xylazine (16 mg). Anesthesia was then maintained with intravenous injections of Ketamine and Xylazine as required. A laparotomy was performed, and a portion of the abdominal aorta immediately proximal to the caudal mesenteric artery was isolated. Cuffs manufactured from sterile, medical-grade polyethylene tubing (3.9-mm inner diameter) were passed around the aorta and tied closed with 3–0 silk. Cuffs produced an approximately 25% reduction in arterial diameter at systolic pressure, as measured in four rabbits in which the arterial system was cast by infusion of Batson’s casting compound at systolic pressure from 1 to 6 weeks postoperatively. Cuffed aortic segments were compared with segments of aorta immediately upstream from the cuff.
We performed a separate series of experiments to control for tissue reactions to the foreign material comprising the cuff. For these experiments, 30% of the circumference was removed from the upstream half of the cuff (region A in Figure 1 ). Aorta in this region of the cuff was free to distend under pressure but was in contact with polyethylene for most of its circumference. The downstream half of the cuff was identical to the standard cuffs described above.
After closure of the incisions, all rabbits were given the analgesic buprenorphine (0.03 mg).
For histology, 11 standard cuffs and 6 control cuffs were implanted in 17 rabbits that were sacrificed by anesthetic overdose at 6 weeks postoperatively. The aorta was perfusion fixed at 100 mm Hg with phosphate-buffered 3% paraformaldehyde, and transverse aortic sections from the downstream limit of the cuff to 1 cm upstream from the cuff were prepared.
DNA was extracted from arterial tissues and purified using a standard extraction method. The proximal and cuff portions of abdominal aorta were minced and incubated overnight in DNA lysis buffer (20 mmol/L Tris/HCl, pH 7.4, 1% SDS, 5 mmol/L EDTA, 100 μg/ml proteinase K) at 50°C. DNA was extracted with salt-saturated phenol and chloroform and precipitated with 100% ethanol. Optical measurements were used to estimate the purity, and total DNA was extracted.
Fragmentation of DNA was measured with a modified end-labeling procedure. 17 Five micrograms of arterial tissue DNA was incubated with [32P]dCTP (10 μCi; ICN Biomedical Canada, St. Laurent, Quebec, Canada) and Klenow polymerase (10 U; Pharmacia Biotech, Baie d’Urfe, Quebec, Canada) for 15 minutes at 30°C. The end-labeling reaction was terminated by addition of 10 mmol/L EDTA. Unincorporated nucleotides were removed using a Magic DNA Clean-up system (Promega Corp., Madison, WI). Radiolabeled DNA was electrophoresed on a 1.8% agarose gel and blotted onto Hybond nylon membrane, and the membrane was exposed to Kodak X-Omat x-ray film. In a separate lane, we loaded 1 μg of DNA size markers consisting of multiples of 100-bp fragments (Gibco BRL Life Technologies, Gaithersburg, MD) that was similarly end-labeled with [32P]dCTP. The appearance of bands of labeled DNA from arterial tissue at multiples of approximately 200 bp indicated apoptotic fragmentation DNA into oligonucleosomes.
In a separate group of four rabbits, the aorta was cuffed for 1 or 2 weeks at which time propidium iodide (5 μmol/L/kg) was injected intravenously 15 minutes before sacrifice and perfusion fixation. An en face preparation of the abdominal aorta was examined from each rabbit with confocal microscopy to detect nonviable cells, as propidium iodide is excluded from viable cells. 18
All rabbits were killed by anesthetic overdose and perfusion fixed at 100 mm Hg with phosphate-buffered 3% paraformaldehyde. Consecutive sections of abdominal aorta were taken from the area underneath the cuff and extending to 1 cm proximal of the cuff. For animals with no cuff, three sections were taken immediately proximal to the caudal mesenteric artery. BrdU immunohistochemistry and quantitative morphometry were performed to calculate cell numbers, dividing cells and medial and intimal areas.
Aortic sections from under standard cuffs were compared with regions of aorta proximal to the cuff. At 6 weeks postoperatively, all regions under the cuff displayed decreased medial area, decreased medial cell number, and increased cell density (Table 1) . No intimal thickening was observed at any location when the cuff was placed on uninjured aorta. Histological sections revealed numerous patent small distributing vessels between the cuff and the adventitia of the aorta, indicating that the cuffs did not block blood flow to the adventitia.
DNA extracted from regions of vessel under the cuff displayed a distinctive laddering pattern at 1, 2, 3, and 6 weeks postoperatively (Figure 2) . Extensive DNA fragmentation was not observed in DNA extracted from regions of the aorta proximal to the cuff. Transmission electron microscopy revealed irreversibly damaged cells in the portion of the vessel beneath the cuff, nuclear and cytoplasmic condensation, swelling of intracellular organelles, large cytoplasmic inclusions and membrane blebbing in scattered cells throughout the media (Figure 3) .
Propidium iodide is excluded from viable cells and when injected intravenously it labels nuclei in cells that are not viable. In rabbits treated with propidium iodide, en face preparations of the aorta from cuffed regions displayed scattered cells in which the dye had been internalized (Figure 4) . These cells also displayed a distinct nuclear fragmentation typical of apoptotic cell death.
In control cuffs (30% of circumference removed from upstream half) all regions under the closed portion of the cuff produced a marked thinning of the media when compared with regions in the open cuff or proximal vessel (Figure 1) . No decrease in medial thickness was observed under the open cuff. There was no significant effect of the open region on medial area or total cells per cross section; medial cells per section were 3230 ± 280 in the open region versus 3260 ± 200 in the region of the vessel outside the cuff. Although the open and closed cuffs produced some perivascular fibrosis, this did not extend into the vessel wall. Inflammatory infiltrates were not observed in the adventitia under the cuff.
At 3 weeks postoperatively, the cuffs produced a marked, statistically significant reduction in medial cross-sectional area compared with regions upstream from the cuff in the same animals (Table 3) ; however, the cuff produced no statistically significant effect on the size of the intima at 3 weeks. (A power analysis of the data indicated that the experimental design was robust enough to determine a 20% or greater difference in intimal area.) Also, no statistically significant difference was found between cuffed and upstream vessels for the total intimal cells per section, or medial and intimal cell density.
There was no statistically significant difference, between cuffed and upstream segments, in the percentage of intimal or medial cells proliferating over the 24 hours before sacrifice at 3 weeks. There was, however, considerable variability in the labeling indices.
A perivascular cuff placed around the rabbit abdominal aorta to reduce circumferential strain caused a marked thinning and loss of cells in the media over 6 weeks. The smooth muscle cell loss showed classical hallmarks of apoptotic cell death, including DNA fragmentation into multiples of 200 bp, incorporation of propidium iodide in vivo, and a distinctive cellular morphology that included condensation and fragmentation into apoptotic bodies. 19-22 Taken together, these results indicate that a perivascular cuff induces extensive smooth muscle cell apoptosis. This response was not seen under the open regions of control cuffs, which maintained close perivascular contact yet did not reduce strain, indicating that remodeling in the media was not a response to a foreign body or to tissue irritation due to the cuff. Extensive medial cell loss under a cuff has not been reported previously. The most likely explanation for this observation is that smooth muscle cells of the media are responding to changes in tensile load. By cuffing the vessel, the average strain in the wall is reduced and the pulsatile strain eliminated. It was not possible to distinguish the independent roles of reduced mean strain level versus the loss of pulsatile strain in arterial remodeling. As mentioned earlier, in vitro models suggest that vascular smooth muscle cells display a wide range of responses to cyclic strain; however, it is also possible that a step reduction in the mean strain level will produce profound cellular responses and vessel remodeling.
In the media of uncuffed arteries, hydrostatic pressure decreases from arterial blood pressure at the intimal side of the vessel wall to near atmospheric pressure at the adventitial side of the media. Our standard cuffs will cause hydrostatic pressures throughout the media to approach arterial pressure, as the constricting cuff will bear most of the wall tension, but tissue pressure cannot exceed arterial blood pressure. As innermost smooth muscle cells in unmanipulated arteries are untraumatized by this level of pressure, it is unlikely that any injury associated with hydrostatic forces can account for medial atrophy in our model.
Cuffs may induce other perturbations to the vessel wall, including possible disruption of adventitial innervation, vascularization, and transport of fluid across the media. However, such perturbations occur after application of loose-fitting perivascular cuffs with sealed ends. The latter cuffs cause intimal proliferation without substantial changes, eg, atrophy, to the media. 23,24 Consequently, this previous work argues against a role for such perturbations in the atrophy seen in the current work. A decrease in nutrient delivery produced by the cuff would be most pronounced on the adventitial side of the media, and smooth muscle cells immediately adjacent to the lumen would receive adequate nutrient supply. In our experiments the morphology of cell loss and the propidium staining in cuffed vessels clearly showed that smooth muscle cell apoptosis was equally distributed throughout the vessel media. These findings argue against a primary role for reduced nutrient delivery as the primary cause for the apoptotic cell loss. We have not, however, proven conclusively that reduced nutrient delivery may be involved in the remodeling process in cuffed vessels.
DNA fragmentation data indicated that cell death through apoptosis in cuffed regions occurs over an extensive postoperative period, with evidence of elevated cell death present even at 6 weeks postoperatively. This implies a chronic turnover of the cell population in the media. The cuffed regions of vessel also displayed marked remodeling of the extracellular matrix, evidenced histologically by a compacting of the elastic lamellar units and an extensive loss of non-cell-associated area. Loss of attachment to the extracellular matrix has been associated with apoptosis, 25,26 and matrix-integrin interactions have been implicated in mediating cell survival. 27 It is possible that loss or restructuring of extracellular matrix in response to reduced wall tension is a stimulus for apoptosis in our model.
Although changes in tensile strain are the most likely stimulus for remodeling under the cuff, the reduction in diameter produced by the cuff would also elevate the luminal shear stress under the cuffed portion of the vessel. Increased shear stress can induce release of endothelial-derived relaxing factor (nitric oxide). Cuffing may have led to a frustrated vasodilatory response in which nitric oxide was chronically released. Nitric oxide is an important mediator of cell death in a number of cell culture systems, including vascular smooth muscle. 28-31 Consequently, it is possible that chronically elevated levels of nitric oxide released from the endothelium may lead to medial smooth muscle cell loss; however, our balloon denudation experiments in the rabbit do not support this hypothesis. Loss of medial smooth muscle cells was evident even after extensive balloon denudation in which the endothelium was stripped from the vessel. At 3 weeks postoperatively, medial cell loss was evident despite the absence of endothelium. Also, an intact endothelial cell monolayer is a primary determinant of fluid flux through the wall, 32 and therefore, disruption of the endothelium is likely to have a marked impact on fluid flux. As similar medial cell loss was observed with and without balloon denudation, it is unlikely that changes in fluid flux are a primary stimulus.
The application of a perivascular cuff delayed the onset of intimal thickening after balloon denudation; however, the intimal area under the cuffed vessel was similar to the unconfined regions by 3 weeks postoperatively. Delayed onset of intimal thickening under the cuff may have been the result of increased shear stress on the luminal surface. A 20% to 25% decrease in diameter in the cuffed region would have doubled the peak luminal shear stress (assuming Poiseuille flow, for which shear stress varies as the inverse cube of vessel diameter), and elevated shear stress slows neointimal thickening. Intimal proliferation induced by balloon injury in rat carotid arteries was inhibited by shear stress transiently (2 to 4 weeks) but not at steady state 8 weeks after injury. 33 Thus, elevated shear stress may have contributed to the delayed smooth muscle cell migration into the intima of the cuffed vessels.
Medial trauma after balloon injury has been linked to neointimal formation; however, our data clearly indicate that chronic medial cell apoptosis in response to vascular cuffing does not influence intimal thickening after balloon denudation. Unlike the necrotic cell death produced by ballooning, apoptosis is much less likely to produce an inflammatory reaction even if present in large numbers, 21,22 and consequently the release of cytokines that could contribute to intimal proliferation is probably modest. However, recent evidence has suggested that apoptotic cell death may not be completely benign. In cell co-cultures of endothelium with fibroblasts, induction of apoptosis enhanced tissue factor procoagulant activity, possibly due to increased exposure of phosphatidylserine. 34 In our model, extensive medial cell loss under the cuff did not promote any observable thrombosis, inflammation, or increased intimal growth. These data show that extensive vascular smooth muscle cell death can take place within the vessel wall without promoting or initiating these pathological processes.
When the cuff was applied after balloon denudation, it promoted medial atrophy yet did not affect long-term neointimal formation. Possibly, the very organized structure of the extracellular matrix in the media is important in the transduction of mechanical stimuli. Alternatively, it may be that phenotypic differences between the smooth muscle cell populations in media versus within the neointima 35,36 account for differences in responses to cuffing. Finally, it should be noted that inferences concerning circumferential tensions imposed on neointima should be made with caution, regardless of whether a cuff is present or not. Neointima in uncuffed vessels forms on a prestressed substrate (the media), but it is not clear that intimal tissues acquire wall tension (become stretched) as they are elaborated. Indeed, if medial stretch and structure are unchanged during neointimal formation, then medial tension may remain in equilibrium with intraluminal pressure.
Finally, our findings may be relevant to remodeling of atherosclerotic vessels in humans, in which lesion tissues may reduce tensile loads on adjacent media. They may also be relevant to many surgical interventions that alter blood vessel mechanics and promote intimal hyperplasia. The cuff procedure we present here represents a novel model in which to study, in vivo, the role of medial smooth muscle cell apoptosis in blood vessel response to injury.