Hypercholesterolemia Exacerbates Transplant Arteriosclerosis via Increased Neointimal Smooth Muscle Cell Accumulation
Studies in Apolipoprotein E Knockout Mice
Background Hypercholesterolemia is thought to be a significant risk factor for coronary vasculopathy in cardiac transplant recipients.
Methods and Results We examined the development of arteriosclerosis in mouse carotid artery loops allografted from B.10A(2R) (H-2h2) donors to normocholesterolemic C57BL/6J (H-2b) recipients and hypercholesterolemic C57BL/6J recipients in which the apolipoprotein (apo) E gene had been knocked out. Luminal occlusion and cross-sectional neointimal area were greater in arteries allografted into hypercholesterolemic recipients at 15 and 30 days after transplantation. We also measured cellular and extracellular matrix components of the neointima by computerized planimetry of the fractional areas subtended by smooth muscle cells (anti–α-actin stain), collagen (Masson’s trichrome), lipid (oil red O), and leukocytes (anti-CD45). The neointimal area stained for smooth muscle cells was significantly greater in hypercholesterolemic recipients than in normocholesterolemic recipients at 15 and 30 days after allografting. Lipid contributed to neointimal area to a lesser degree, and there was no significant increase in the contribution of collagen or leukocytes.
Conclusions Smooth muscle cell accumulation appears to be the principal contributor to the increase in neointimal area observed in arteries allografted into hypercholesterolemic mice.
Although graft rejection and infection remain the major predictors of mortality during the first year after heart transplantation, accelerated coronary artery disease becomes the chief impediment to survival thereafter.1 Hypercholesterolemia has been shown by some investigators to be a significant risk factor for the development of coronary vasculopathy in transplant recipients.2 3 4 5 Kobashigawa and colleagues6 showed that treatment with a 3-hydroxy-3-methylglutaryl coenzyme A (HMG-CoA) reductase inhibitor (pravastatin) reduces the incidence of coronary vasculopathy, yet they found no correlation between cholesterol levels and development of coronary vasculopathy during a 12-month period of observation. This paradoxical finding suggests that pravastatin may ameliorate vasculopathy through a mechanism other than the lowering of serum cholesterol. Other studies in rabbits, rats, and mice have yielded variable results.7 8 9 10 11 Thus, both the role and the mechanism of hypercholesterolemia in transplant vasculopathy remain open questions.
We have used a mouse carotid artery transplantation model12 to identify immunologic components of transplant arteriosclerosis.13 In the work reported here, we used the same model to study the role of hypercholesterolemia in transplant arteriosclerosis. These experiments allowed us to compare the tissue components contributing to arteriosclerotic lesions and the magnitude of lesion development in carotid artery loops allografted into hypercholesterolemic mice (in which the apolipoprotein E [apo E] gene had been deleted14 15 ) with that in loops allografted into normocholesterolemic control mice of the same genetic background. We found that a neointima developed faster and that luminal occlusion increased in arteries allografted into hypercholesterolemic mice. Smooth muscle cell accumulation appeared to be the principal cause of this increase in neointimal size.
Carotid artery transplantation was performed on 8-week-old male mice weighing 28 to 32 g (Jackson Laboratory) in the following donor-recipient combinations: B.10A(2R) (H-2h2) into C57BL/6J (H-2b) (allograft, normocholesterolemic recipient); B.10A into C57BL/6J in which the apo E gene had been deleted14 15 (allograft, hypercholesterolemic recipient); and C57BL/6J apo E knockout into C57BL/6J apo E knockout (isograft, hypercholesterolemic recipient). The apo E knockout mice had been backcrossed over six generations to C57BL/6J (Jackson Laboratory). The animals were maintained on a normal mouse chow diet (No. 5010, Purina Mills) before and after surgery and were not treated with an immunosuppres-sive agent.
A donor carotid artery segment was anastomosed onto the carotid artery12 of a histoincompatible (or compatible) recipient in which the apo E gene had been deleted (or not, control). The operation was performed on anesthetized mice under a dissecting microscope (Wild, model M3Z). A midline incision was made on the ventral side of the neck from the suprasternal notch to the chin. The left carotid artery was dissected from the bifurcation in the distal end toward the proximal end as far as was technically possible. The artery was then occluded with two microvascular clamps (8 mm long, ROBOZ Surgical Instruments), one at each end, and two longitudinal arteriotomies (0.5 to 0.6 mm) were made with a fine needle (30 gauge) and scissors. In the donor mouse, both the left and the right carotid arteries were fully dissected from the arch to the bifurcation. The graft was then transplanted paratopically into the recipient in an end-to-side anastomosis with an 11/0 continuous nylon suture under ×16 or ×25 magnification, and the skin incision was closed with a 4/0 interrupted suture. The time during which the graft was ischemic did not exceed 60 minutes. Grafts were harvested 15 or 30 days after transplantation.
Histology and Morphometry
The proximal half of the transplanted loop (≈2.5 mm) was fixed for 3 to 4 hours in methyl Carnoy’s solution at 4°C and embedded in paraffin. The other half was fixed in 4% paraformaldehyde for 3 hours, dehydrated in 30% sucrose for 48 hours, embedded in medium (OCT Compound, Miles), frozen in powdered dry ice, and kept at −80°C. Consecutive serial sections were obtained from the center of the donor loop to the site of the anastomosis with the recipient carotid artery. Tissue samples fixed in methyl Carnoy’s solution were stained with Verhoeff’s elastin,12 anti–α-actin,12 anti-CD45,12 Masson’s trichrome,13 or anti-macrophage antibody F4/80 (Caltag Laboratories). Frozen samples were stained with oil red O. Histological, morphometric, and immunocytochemical analyses were performed as described.12
The areas of the neointima, the media, and the lumen were measured by computerized planimetry of sections obtained 150 and 300 μm from the junction of the donor carotid loop and the recipient carotid artery. Color video images were recorded from a Nikon Labophot-2 light microscope equipped with a Sony DXC-760MD video camera and control unit. The images were captured on a Power Macintosh 7100/80 computer (Apple Computer) with the Scion LG3 frame grabber (Scion, Inc) program. Image analysis routines were created within a modified version (Organize-It Software) of the NIH Image 1.1 program. After tracing the area to be measured with the cursor, the operator sampled five pixels of the color that defined a given stain. The area encompassed by pixels (not always contiguous) in the color range for anti–α-actin, Masson’s trichrome, oil red O, or anti-CD45 was then computed automatically by the software. The neointima was defined as the area bounded by the internal elastic lamina and the lumen. The media was defined as the region between the internal and external elastic laminae. The lumen was defined as the clear part of the vessel in the center of the section.
The results of Wilks-Shapiro tests of normality were consistent with gaussian distributions for all measurements of area subtended by the four assay reagents (Fig 4⇓; anti–α-actin for smooth muscle, Masson’s trichrome for collagen, oil red O for lipid, and anti-CD45 for leukocytes) when the areas were normalized against each corresponding total intimal area. These normalized measurements were then entered into a MANOVA with two grouping factors: apo E status (with the levels normocholesterolemic control and hypercholesterolemic apo E knockout) and day (with the levels day 15 and day 30).16 The four variates in the MANOVA design were the normalized areas subtended by each of the four stains. Because the apo E status/day data cells represented 3 mice in the control/day 15 combination, 4 mice in the knockout/day 15 combination, 4 mice in the control/day 30 combination, and 5 mice in the knockout/day 30 combination, the design was slightly unbalanced. Only the oil red O variate (lipid) demonstrated a significant apo E status × day interaction (P=.0012). Therefore, the oil red O data were analyzed separately in two univariate ANOVAs. Because eight distinct null hypotheses were tested (linear additive parameters for apo E status=0 and for day=0 for each of the four variates), the Bonferroni method was used to adjust the nominal α-value for statistical significance from .05 to .05÷8=.00625 to correct for spurious significance due to multiple comparisons.17 Therefore, comparisons were regarded as significant only at P<.00625. All estimates of variability are stated as mean±SD. The data presented in Fig 4⇓ were also analyzed with a stepwise multiple regression program18 to explore interdependencies among the four stain variates and overall intimal area.
Carotid artery loops from B.10A donors were allografted into C57BL/6J recipients (normocholesterolemic mice) and C57BL/6J recipients in which the apo E gene had been deleted (hypercholesterolemic apo E knockouts). As a control, loops from apo E knockout donors were isografted into apo E knockout recipients.
Serum Cholesterol Levels
Mean serum cholesterol levels measured by enzymatic assay (Boehringer-Mannheim) were 102±17 (SD) mg/dL (n=6) in B.10A mice and 90±8 mg/dL (n=4) in normocholesterolemic C57BL/6J mice before transplantation. Mean serum cholesterol levels were 652±187 mg/dL (n=9) in hypercholesterolemic C57BL/6J apo E knockout mice after transplantation.
Region of Donor Loop Sampled
In hypercholesterolemic apo E knockout mice not subjected to surgery, atherosclerotic lesions develop preferentially at arterial bifurcations,19 an observation consistent with the classic appreciation of atheromatous lesion distribution in humans. Therefore, in the present study we examined cross-sectional neointimal area in sections from the center of the transplanted carotid artery loop, as we had in our previous studies,12 13 as well as neointimal area in sections close to the junction between the transplanted arterial loop and the recipient carotid artery. This examination allowed us to determine the possible effects of turbulent blood flow at bifurcations, which exacerbates arteriosclerotic lesion formation in the setting of hypercholesterolemia.
We found no significant difference at 15 days after transplantation between the intimal area [expressed as (intimal area/intimal+medial area)×100] in the junctional region and that in the center in arterial loops allografted into normocholesterolemic mice [40.6±15.1 (SD) versus 38.5±17.8, P=.70]. In loops allografted into hypercholesterolemic animals, however, the neointimal cross-sectional ratio at 15 days was substantially greater in the junctional region (not including the suture line) than in the center (69.9±9.1 versus 25.7±8.9, P<.0001). This observation suggests that turbulent blood flow has a substantial effect on lesion development in hypercholesterolemic but not normocholesterolemic mice. We subsequently measured differences between normocholesterolemic and hypercholesterolemic mice in sectionsof donor loop obtained 150 and 300 μm from the anastomosis.
Size of Neointima and Degree of Luminal Occlusion
Carotid artery allotransplantation into hypercholesterolemic mice resulted in a higher degree of luminal occlusion and formation of neointimas with a larger cross-sectional area in the region near the anastomosis than did allotransplantation into normocholesterolemic mice at both 15 (Figs 1 through 3⇓⇓⇓) and 30 (Figs 2⇓ and 3⇓) days after transplantation. As expected, intimal area increased and residual luminal area decreased between days 15 and 30 (Figs 2⇓ and 3⇓) in loops allografted into normocholesterolemic as well as hypercholesterolemic recipients. An alternative analysis of these comparisons that normalized for vessel diameter [(intimal area/intimal+medial area)×100] gave similar results: loops allografted into normocholesterolemic recipients versus hypercholesterolemic recipients at 15 days, 40.6±15 versus 50.5±14, P=.04; at 30 days, 56.8±6.8 versus 70.7±4.8, P<.0001.
We found previously that isotransplantation of carotid artery loops into normocholesterolemic mice results in minimal formation of neointima.12 Isotransplantation of nonatherosclerotic arteries from 8-week-old apo E knockout donors (an age at which the vessels are still clear in these hypercholesterolemic mice) into 8-week-old apo E knockout recipients (Fig 2⇑) did result in formation of a modest, eccentric neointima [11.8±16×103 (SD) μm2] at 30 days after transplantation that was much smaller than the neointimas formed in arterial loops allografted into apo E knockouts at both 15 and 30 days after transplantation (63.09±13.62×103 and 112.51±27.68×103 μm2, respectively). Although a neointima was evident on histological examination in five of the eight apo E knockout arteries isografted into apo E knockout recipients, there was no neointima whatsoever in the remaining three. The 8-week-old apo E knockout mice used as donors did not manifest atherosclerosis in the carotid artery (data not shown). It is likely that the lesions observed in the apo E isografts proceeded independently of the transplantation operation.
There was little variability in cholesterol values in apo E knockout mice (as indicated under “Serum Cholesterol Levels,” the SD was small: 652±187 mg/dL). There was also no significant correlation between cholesterol values in individual animals and intimal lesion size (r=.12), probably because the range of variability in both cholesterol values and cross-sectional lesion areas was small.
Number of Nuclei in the Neointima
To determine whether the increase in the size of the neointima in arteries allografted into hypercholesterolemic mice was due to an increase in cellularity or to an increase in cell size or deposition of extracellular matrix, we compared the number of nuclei in sections of B.10A arterial loops that had been allografted into normocholesterolemic C57BL/6J recipients with that in sections that had been allografted into hypercholesterolemic C57BL/6J apo E knockout recipients. Although there was no significant difference in the number of nuclei at 15 days, at 30 days the number of nuclei in the intimas of arteries allografted into apo E knockouts was significantly higher [814±158 (SD) versus 565±221 nuclei/1000 μm2, P=.037]. These findings suggest that cellular proliferation or migration in hypercholesterolemic mice contributes at least in part to the increase in neointimal size.
Contributions to Neointimal Areas of Component Cells
The cellular and extracellular matrix components contributing to the increase in neointimal area were measured by computerized planimetry of the areas subtended by colors representing anti–α-actin (which stains for smooth muscle cells), Masson’s trichrome (which stains for collagen), oil red O (which stains for lipid), and anti-CD45 (which stains for leukocytes). Each area measurement was made independently in separate and consecutive vessel sections. These measurements permitted quantification of cellular components and matrix substances (smooth muscle cells, lipid, and collagen) that are not amenable to conventional cell counting.
On day 15 after transplantation, smooth muscle cells made the major contribution to the neointimal area in arterial loops allografted into hypercholesterolemic mice (Figs 1⇑, right, and 4). The contributions of collagen (Fig 4⇓), lipid (Figs 4⇓ and 5⇓, center), and leukocytes (Fig 4⇓) were but trivial. By day 30, smooth muscle cells remained the major contributor to intimal area (Figs 4⇓ and 6⇓, right), but the contribution of lipid became more significant (Figs 4⇓ and 5⇓, right). On histological examination, lipid was deposited both intracellularly and extracellularly. Double staining with macrophage-specific antibody F4/80 and anti–α-actin, as well as single staining with the two antibodies on adjacent sections, showed that foam cells in the neointima arose from both the smooth muscle cell and the macrophage lineages. Approximately 90% of the foam cells were stained with anti–α-actin and 10% were stained with F4/80 (data not shown).
There was no significant difference between collagen deposition in loops allografted into normocholesterolemic and hypercholesterolemic mice at day 15 or day 30 (Fig 4⇑). However, a comparison between day-15 and day-30 neointimas in loops allografted into hypercholesterolemic as well as normocholesterolemic mice indicated a significant increase in collagen deposition (Fig 4⇑). In hypercholesterolemic mice at day 30 after transplantation, collagen was the third-most-prevalent component studied (behind smooth muscle and lipid) of the neointima in the allografted artery (Fig 4⇑). Leukocytes were significantly more prevalent in arterial loops allografted into normocholesterolemic recipients than in those allografted into hypercholesterolemic recipients, at both 15 and 30 days after transplantation (Fig 4⇑). Because the major contribution to the enlarged neointimas in arteries allografted into hypercholesterolemic mice came from smooth muscle cells, smooth muscle cell accumulation appears to be a consequence of hypercholesterolemia, lipid deposition in the vessel wall, or chemotactic (or growth) factors secreted by foam cells.
The sum of the areas subtended by the four stains studied did not equal the total area of the neointima. For example, the areas stained by anti–α-actin (smooth muscle), Masson’s trichrome (collagen), oil red O (lipid), and anti-CD45 (leukocytes) accounted for only 75% of the measured neointimal area in arterial loops allografted into apo E knockout recipients at 30 days after transplantation. The presence of immature smooth muscle cells, which do not contain α-actin,20 21 and the fact that extracellular matrix contains many substances other than collagen are two possible explanations for this difference. It should also be noted that the antibody to CD45 stains the leukocyte cell membrane, in contrast to the antibody to α-actin, which stains the smooth muscle cell cytoplasm. This difference in staining would result in an underestimation of the area occupied by leukocytes. These disparities notwithstanding, however, smooth muscle cell accumulation appears to be the principal reason for the larger neointimas and greater degree of luminal occlusion in arteries allografted into hypercholesterolemic as opposed to normocholesterolemic recipients. Lipid-filled foam cells or extracellular lipid (revealed by oil red O) made a smaller contribution to neointimal size, and extracellular matrix (Masson’s trichrome) and inflammatory cells (anti-CD45) made negligible contributions.
A stepwise regression analysis showed that changes in the intimal area correlated best with changes in the area stained by anti–α-actin (P<.001 compared with the areas stained by oil red O, Masson’s trichrome, and anti-CD45) and that changes in the area stained by anti–α-actin in turn correlated best with changes in the area stained by oil red O (P<.001 compared with the areas stained by Masson’s trichrome and anti-CD45). By indicating that smooth muscle cells accounted for most of the increase in intimal area and that lipid accounted for most of the increase in smooth muscle cells, this analysis confirms the comparative analysis presented in Fig 4⇑.
Characteristics of the Media
There was significant deposition of lipid in the media [10±4% (SD) of the medial area] by 15 days after transplantation in B.10A arterial loops allografted into C57BL/6J apo E knockout (hypercholesterolemic) recipients (Fig 5⇑, center). Little lipid deposition was visible in B.10A arterial loops allografted into normocholesterolemic C57BL/6J recipients (not shown). By 30 days after transplantation, medial lipid deposition had increased to 24±13% in loops allografted into hypercholesterolemic animals (Fig 5⇑, right) compared with 3.4±2.5% in loops allografted into normocholesterolemic controls (Fig 5⇑, left). Lipid particles deposited in the medias of B.10A arteries allografted into normocholesterolemic C57BL/6J recipients (Fig 5⇑, left) were smaller than lipid particles deposited in those allografted into C57BL/6J apo E knockout recipients (Fig 5⇑, center and right). This difference may reflect the presence of fragmented extracellular membrane rather than foam cells. The larger globules observed in the B.10A to C57BL/6J apo E knockout combination are more consistent with foam cells.
If the intima were to expand and simply impinge on the lumen without affecting the cross-sectional area of the vessel overall, then the sum of the intimal and residual luminal areas should not differ between arteries allografted into normocholesterolemic and hypercholesterolemic mice, despite any difference in intimal area. If there was arterial remodeling, however, the overall area inside the internal elastic lamina should increase. We studied this question by comparing the cross-sectional vessel area (intimal+luminal area). At 30 days after transplantation (Fig 6⇑), there was a significant increase in cross-sectional vessel area in arterial loops allografted into hypercholesterolemic in comparison with normocholesterolemic recipients [110.0±21.9×103 (SD) versus 76.7±21.7×103 μm2, P=.0006]. At 15 days (not shown), there was no significant difference (P=.13). These findings suggest a degree of arterial remodeling caused by massive formation of neointima in hypercholesterolemic recipients that does not occur in normocholesterolemic recipients.
Although clinical and experimental observations indicate that hypercholesterolemia accelerates transplant arteriosclerosis,2 3 4 7 8 the mechanism of this acceleration has not been documented. We developed a mouse model of carotid artery allotransplantation that permits quantification of the cellular composition and size of arteriosclerotic lesions.12 We have used the model to define components of the immune system responsible for transplant arteriosclerosis13 and apply it here to an examination of accelerated arteriosclerosis in the presence of hypercholesterolemia. At both 15 and 30 days after transplantation, the area of neointima formed in arterial loops allografted into hypercholesterolemic apo E knockout recipients was significantly greater than that in loops allografted into normocholesterolemic recipients (Fig 2⇑), as was the degree of luminal occlusion (Fig 3⇑).
We noted that the neointimas formed in hypercholesterolemic mice contained a larger number of nuclei than those formed in normocholesterolemic controls, which suggested that cellular proliferation was in part responsible for the increase in area. To determine the cellular and extracellular matrix components contributing to this increase in neointimal size, we stained sections of allografted arteries individually with anti–α-actin (to show smooth muscle cells), Masson’s trichrome (to show collagen, an extracellular matrix component), oil red O (to show extracellular lipid as well as lipid within foam cells), and anti-CD45 (to show inflammatory cells). Automated computerized planimetry specific to each stain permitted an objective evaluation of the fractional area of neointima subtended by that stain. We found that the area occupied by smooth muscle cells made a major contribution to the increase in neointimal size overall (Fig 4⇑), although intracellular lipid deposition contributed as well. Changes in collagen and inflammatory cells made but minor contributions to overall neointimal area, and they did not account for the increased neointimal size in arteries allografted into hypercholesterolemic recipients. Rather than just impinging on the lumen, the very large neointima that formed 30 days after transplantation in arteries allografted into hypercholesterolemic recipients (Fig 6⇑) caused an overall expansion of the vessel’s diameter, in essence an example of arterial remodeling.22
It is not surprising that foam cells should occur in arteriosclerotic lesions developing in hypercholesterolemic animals. Indeed, lipid deposition frequently accompanies transplant arteriosclerosis in human specimens.23 What is surprising is the prominent role played by smooth muscle cells in the augmented lesion in hypercholesterolemic animals. Smooth muscle cell accumulation is a characteristic of human transplant arteriosclerosis,23 and we have also observed it in carotid arteries allografted into normocholesterolemic mice.12 Smooth muscle cell accumulation is probably mediated by an immunologic process in which helper T cells, humoral antibody, and macrophages participate.13 The stimulus for the substantial additional smooth muscle cell accumulation seen in arteries allografted into hypercholesterolemic mice is not apparent. One possibility is that the additional accumulation derives from the action of a growth factor secreted in greater amounts by macrophage-derived foam cells than by non–lipid-containing macrophages. Wang et al24 showed that interleukin (IL)-8 mRNA and protein levels are three to four times higher in human foam cells than in non–lipid-containing macrophages. IL-8 is a potent stimulus to DNA synthesis and cell migration in vascular smooth muscle.25 Although IL-8 has not been identified yet in the mouse, there are two forms of the IL-8 receptor in this species.26 Because these receptors are present in the mouse, ligands to them, such as N51, may play the role of IL-8.27 28 Our studies do not address the issue of whether HMG-CoA reductase inhibitors manifest their beneficial effects in humans by lowering plasma cholesterol or by inhibiting smooth muscle cell proliferation directly.6 To differentiate between the two possibilities, it will be necessary to repeat the studies in LDL receptor–deficient mice, a model in which HMG-CoA reductase inhibitors have no effect.
Another unusual finding, generally not observed in transplant arteriosclerosis in humans, is that lipid deposition was significant in the medias of arterial loops allografted into hypercholesterolemic animals, particularly at 30 days after transplantation (Fig 5⇑). The very high levels of lipid in apo E knockout mice may have accounted for this difference, because oxidation of the abundant lipid29 could have contributed additionally to medial smooth muscle cell proliferation and migration to the neointima.
In conclusion, hypercholesterolemia appears to increase the rate of neointima formation and vascular occlusion in experimental transplant vasculopathy by a mechanism that depends on smooth muscle cell accumulation. This observation should lead to a reevaluation of the importance of controlling cholesterol and lipid levels after clinical cardiac transplantation.
This work was supported by a grant from the Bristol-Myers Squibb Pharmaceutical Research Institute. We are grateful to Mark A. Perrella and Choon-Joo Kho for their advice about the design of our studies.
- Received April 21, 1997.
- Revision received June 4, 1997.
- Accepted June 6, 1997.
- Copyright © 1997 by American Heart Association
Kobashigawa JA, Katznelson S, Laks H, Johnson JA, Yeatman L, Wang XM, Chia D, Terasaki PI, Sabad A, Cogert GA, Trosian K, Hamilton MA, Moriguchi JD, Kawata N, Hage A, Drinkwater DC, Stevenson LW. Effect of pravastatin on outcomes after cardiac transplantation. N Engl J Med. 1995;333:621-627.
Tanaka H, Sukhova GK, Libby P. Interaction of the allogeneic state and hypercholesterolemia in arterial lesion formation in experimental cardiac allografts. Arterioscler Thromb. 1994;14:734-745.
Shi C, Russell ME, Bianchi C, Newell JB, Haber E. Murine model of accelerated transplant arteriosclerosis. Circ Res. 1994;75:199-207.
Shi C, Lee WS, He Q, Zhang D, Fletcher DL Jr, Newell JB, Haber E. Immunologic basis of transplant-associated arteriosclerosis. Proc Natl Acad Sci U S A. 1996;93:4051-4056.
Piedrahita JA, Zhang SH, Hagaman JR, Oliver PM, Maeda N. Generation of mice carrying a mutant apolipoprotein E gene inactivated by gene targeting in embryonic stem cells. Proc Natl Acad Sci U S A. 1992;89:4471-4475.
Dixon WJ, ed. BMDP Program 4V, Release 7. Berkeley, Calif: University of California Press; 1992:1259-1310.
Morrison DF. Multivariate Statistical Methods. New York, NY: McGraw-Hill Book Co; 1976:32-34.
Dixon WJ, ed. BMDP Program 2R, Release 7. Berkeley, Calif: University of California Press; 1992:387-425.
Nakashima Y, Plump AS, Raines EW, Breslow JL, Ross R. ApoE-deficient mice develop lesions of all phases of atherosclerosis throughout the arterial tree. Arterioscler Thromb. 1994;14:133-140.
Owens GK, Loeb A, Gordon D, Thompson MM. Expression of smooth muscle-specific α-isoactin in cultured vascular smooth muscle cells: relationship between growth and cytodifferentiation. J Cell Biol. 1986;102:343-352.
Langille BL. Remodeling of developing and mature arteries: endothelium, smooth muscle, and matrix. J Cardiovasc Pharmacol. 1993;21(suppl 1):S11-S17.
Wang N, Tabas I, Winchester R, Ravalli S, Rabbani LE, Tall A. Interleukin 8 is induced by cholesterol loading of macrophages and expressed by macrophage foam cells in human atheroma. J Biol Chem. 1996;271:8837-8842.
Yue TL, Wang X, Sung CP, Olson B, McKenna PJ, Gu JL, Feuerstein GZ. Interleukin-8: a mitogen and chemoattractant for vascular smooth muscle cells. Circ Res. 1994;75:1-7.
Bozic CR, Gerard NP, von Uexkull-Guldenband C, Kolakowski LF Jr, Conklyn MJ, Breslow R, Showell HJ, Gerard C. The murine interleukin 8 type B receptor homologue and its ligands: expression and biological characterization. J Biol Chem. 1994;269:29355-29358.
Heinrich JN, Bravo R. The orphan mouse receptor interleukin (IL)-8Rβ binds N51: structure-function analysis using N51/IL-8 chimeric molecules. J Biol Chem. 1995;270:4987-4989.
Heinrich JN, Bravo R. N51 competes 125I-interleukin (IL)-8 binding to IL-8Rβ but not IL-8Rα: structure-function analysis using N51/IL-8 chimeric molecules. J Biol Chem. 1995;270:28014-28017.