- Email: [email protected]
The effect of thrombin on low-density lipoprotein permeability and uptake by an arterial endothelial smooth muscle cell bilayer
The effect of thrombin on low-density lipoprotein permeability and uptake by an arterial endothelial smooth muscle cell bilayer J. Jeffrey Alexander, M D , Remedios Miguel, MS, D e b r a G r a h a m , M D , and Joseph J. Piotrowski, M D , Cleveland, Ohio Thrombin, a mediator of thrombosis, has been shown to directly alter the fimction of vascular cells. We studied the effect of thrombin on low-density lipoprotein permeability and uptake by an arterial endothelial cell-smooth muscle cell bilayer to determine its potential role in atherogenesis. Confluent cell bilayers were incubated in media containing thrombin (10 or 50 units/ml) for a period of 24 hours to 9 days. Iodine 125 02sI)-LDL (10 ~g protein/ml) was then added to the media, and after a 3-hour incubation, ~2SI-LDL transit through the endothelial cell layer as well as membrane binding and uptake by each cell type were measured. The lower concentration of thrombin caused a delayed increase in both the permeability (p < 0.0001) and uptake (p < 0.05) of LDL, but had no effect on membrane binding of the lipoprotein. The higher thrombin concentration led to an immediate increase in endothelial cell permeability to LDL (p < 10 -7) and a significant reduction in both cellular uptake (p < 10 -7) and membrane binding (p < 0.0005). In contrast, smooth muscle cell binding and uptake were unaffected at the lower concentration of thrombin. At the higher concentration, smooth muscle cell uptake of LDL was increased (p < 10 -7) disproportionately to a more limited increase in membrane binding (i0 < 0.05). Endothelial DNA content, reflecting cell number, was increased at 10 units/ml thrombin ~ < 0.001) but markedly reduced at 50 units/ml thrombin (p < 0.0005), whereas smooth muscle cell DNA content remained unchanged, all-adenine and chromium 51 release assays indicated a greater susceptibility of endothelial cells to thrombin-induced injury. These results suggest that endothelial cells and smooth muscle cells are affected differently by thrombin. Thrombin-induced endothelial toxicity, in combination with enhanced LDL uptake by underlying smooth muscle cells, may promote the formation of atherosclerotic plaque. (J VASC SURG 1992;15:718-25.)
Histologic evidence o f fibrotic transformation and lipid incorporation by organizing mural thrombus has implicated intravascular thrombosis in the pathogenesis o f atherosclerosis) -3 Although this mechanism may not be applicable to fatty streak formation, it does provide a plausible explanation for the nature o f established or rapidly progressing atherosclerotic plaque, where the presence o f fibrin and platelets would indicate a thrombotic event. 4 This theory does not necessarily exclude the more commonly cited response to injury hypothesis, which From the Department of Surgery, Case Western ReserveUniversity, ClevelandMetropolitan GeneralHospital. Presented at the Forty-fifth Annual Meeting of the Society for Vascular Surgery,Research Forum II, Boston, Mass., June 4-5, 1991. Reprint requests: I. leffrey Alexander, MD, Departmen~~'of Surgery,ClevelandMc~tropolitanGeneralHospital, 3395 Scranton Rd., Cleveland;~OH 44109, 24/1]34220 718
proposes that a break in either the functional or structural integrity o f the endothelial barrier results in platelet adherence, lipid infiltration, and a mitogenic stimulation o f underlying smooth muscle cells (SMCs). s Under such circumstances, thrombus formation may occur, providing a site for a localized and enhanced interaction between the vessel wall and circulating platelets, coagulation factors, and plasma lipids. These interactions may lead to changes in the arterial wall that are characteristic o f atherosclerotic plaque. Thrombin is an essential element in thrombus formation, providing a stabilizing network o f fibrin through its conversion offibrinogen. It has also been found to promote platelet aggregation and release. 6 In addition to its role in coagulation, thrombin interacts directly with the endothelial cell (EC), to which it binds in a specific and saturating fashion. 7 Previous studies have indicated that it may also affect
Voltune 15 Number 4 ~.pril 1992
Thrombin alters LDL uptake by an EC-SB4C bilayer 719
Table I. ~2e;I-LDLcontent of the lower chamber media (media counts/available counts ___ SD × 10 3) Time 24 hours 48 hours 96 hours 9 days
Thrombin 10 units/ml
Control 366.4 380.8 354.2 360.8
-± ± ±
40.9 41.5 32.1 29.3
391.6 393.0 415.0 414.0
± ± ± ±
34.8 46.0 39.4 9.7
p~
Thrombin 50 units/ml
Control
NS NS <0,0001 <0.005
447.2 446.6 416.5 470,5
± 27.9 _+ 17.1 ± 32.2 ± 5.1
532,8 557,2 523,4 542,3
± ± ± ±
23.3 43,3 40.9 30.0
p~ < 10 -7 < 10 -s <0.0005 <0.001
tp unpaired two-tailed Student's t test.
Table II. Endothelial cell binding and uptake of ~2SI-LDL (counts/available counts + SD × 10 a) Time
Control
Thrombin 10 units/ml
p~
Control
Thrombin 50 units/ml
p~
Binding 24 hours 48 hours 96 hours 9 days
152.9 108.0 98.8 118.1
-+ 38.8 ± 15.3 -+ 13,4 _+ 16.2
149,8 167.9 105.4 107.8
± 50.2 +- 59.7 ± 11.2 ± 11.8
NS <0.005 NS NS
227.2 235.3 185.6 152.9
± 35.6 ± 31.5 -+ 18.1 ± 25,2
186.1 183.1 150.3 76.3
± 38.9 ± 36.4 ± 13.4 _+ 5.5
<0.01 <0.0005 <0.005 <0.001
:24 hours ,48 hours 96 hours 9 days
485.7 645.7 474.6 252.7
-+ ---±
505.4 756.6 648,2 437.8
-+ 112.6 -- 150.0 ± 126,3 _+ 141.7
NS NS <0,05 <0.01
567.0 471.8 402.8 273.0
± ± ± ±
357.5 329.2 288.6 97.7
± 44.5 ± 39,9 -+ 61.6 ± 12.4
< 10 -7 < 10 -s <0.05 <0.001
Uptake 145.6 197.1 198.5 111.1
68.3 77,3 92,2 60.2
*p unpaired two-tailed Student's t test.
cellular proliferation, 8 prostanoid secretion,9 and matrix production by arterial cells. 1° In addition, thrombin increases EC contractility, resulting in cellular retraction, n,~2 This has been associated with an increased permeability of endothelial monolayers to macromolecules,13 which may represent a significant alteration of barrier function. It is therefore possible that thrombin may promote not only platelet adhesion and activation, but also lipid uptake through widened intercellular junctions within the endothelium. The purpose of these experiments was to study the effect of' thrombin on low-density lipoprotein (LDL) permeability and uptake by an EC-SMC bilayer and the potential influence of thrombininduced cell :injury on these processes. MATERIAl, A N D M E T H O D S Cell harvesthlg and culture Endothelial cells and SMCs were harvested from bovine aortas as previously described) ~ Cells in passages 6 through 8 were used for study. Endothelial cells were identified by cellular morphology and immunofluoresccnt staining of factor-VIII related antigen. Smooth muscle cells were identified by their characteristic morphologic appearance. Both cell types were plated on transwell culture plates (Costax Corp., Cambridge, Mass.) consisting
of two separate culture chambers separated by a porous polycarbonate filter. 14 Smooth muscle cells were initially placed in the lower well in 750 mm 3 of culture medium with a plating density of 25,000 cells/cm2. The ECs were then grown on the polycarbonate filters in the upper well in 200 mm 3 of culture medium with the same plating density. Confluent growth, which was achieved in approximately 5 days, was verified by phase contrast microscopy. Once cell confluence has been reached, bovine thrombin (95 units/mg protein; Sigma Diagnostics, St. Louis, Mo.) was added to the upper and lower well media of experimental wells to achieve concentrations of either 10 or 50 units/ml. The bilayers were then incubated for periods ranging from 24 hours to 9 days. The cells were inspected daily, and the media were changed every 3 days. After completion of the incubation period, the media were removed and replaced with Dulbecco's modified Eagle's medium (DMEM) containing 2% fetal bovine serum (FBS) and bovine thrombin. Iodine 125 (12sI)-LDL (10 ~g protein/ml; specific activity 172 to 879 cpm/~g protein) was added to the upper well media. Iodine 125-LDL was prepared weekly by the method of Markwell, 15 using human LDL (Sigma; 6.4 mg protein/ml), and achieving a 12SI-LDL binding efficiency of 91% to 94%. Low-density lipoprotein was maintained for not longer than 3 weeks in refriger-
720
Journal of VASCULAR SURGER~.~
A l e x a n d e r et al.
Table III. Smooth muscle cell binding and uptake of ~25I-LDL (counts/available counts + SD × 103) Time
Thrombin 10 units/ml
Control
p~
Control
Thrombin 50 units/ml
p~
Binding 24 hours 48 hours 96 hours 9 days
295.6 285.9 323.2 438.4
± 78.9 -+ 21.8 -+ 72.3 ± 44.2
255.6 320.2 321.2 363.9
-+ 23.1 ± 58.2 ± 41.0 ± 71.2
NS NS NS NS
260.7 284.7 206.7 233.7
± ± ± ±
24 hours 48 hours 96 hours 9 days Uptake/binding ratio
149.9 151.0 176,9 174,2
± ± ± ±
125.8 149.3 192.4 155.1
_+ 11.5 ± 10.8 ± 34.3 ± 19.9
< .005 NS NS NS
128.4 189.9 91.3 110.2
34.5 45.5 34.1
17.7
269.8 315.6 249.9 287.6
± ± ± ±
41.1 56.9 22.4 49.0
± 17.6 ± 19.1 _+ 5.7 ± 6.7
248.9 505.3 351.4 393.6
± 22.5 ± 71.9 ± 46.4 _+ 36.8
NS NS <0.05 <0,05
Uptake
24 hours 48 hours 96 hours 9 days
14.8 22.8 25.4 39.1
< 10 -7 < 10 -7 < 10 7 <0.001
pt 0.51 0.53 0.55 0.40
0,49 0,46 0.60 0.42
NS < 0.01 < 0.05 NS
pt 0.49 0.67 0.44 0.47
0.92 1.60 1.40 1.37
< < < <
10 -7 10 -7 10 -7 10~f
~p unpaired two-tailed Student's t test; p~-chi-square analysis.
Table IV. Thrombin-induced 3H-adenine and 5~Cr release from endothelial and smooth muscle ceils ~ Thrombin 3H-Adenine Release Control 10 units/ml 50 units/ml 5ICr release Control 10 units/ml 50 units/ml ~24-hour exposure; % release =
Endothelial
p~
Smooth muscle
p~
11.83 ± 0.47 24.20 ± 1.05 31.63 ± 1.37
< i0 -s < 10 -5
20.57 ± 1.08 20.56 ± 1.17 30.39 -+ 0.52
NS <10 s
28.43 ± 0.62 35.65 ± 0.93 39.77 ± 1.28
< 0.00005 < 0.000005
30.65 ± 0.74 32.53 +- 0.50 36.10 ± 0.73
<0.001 <0.00005
counts media counts (ceils + media)
± SD x 100
ated storage (4 ° C) to prevent oxidation. The cells were incubated for 3 hours, after which the medium from each chamber was removed and its 125I-LDL content determined by a gamma counter (Packard 5000; Packard, Downers Grove, Ill.). The 125I-LDL media content of the lower well was used as a measure of EC permeability. The ECs on the polycarbonate filter and the SMCs on the lower culture chamber were washed with phosphate-buffered saline solution (PBS; Wit-taker Bioproducts, Walkersville, Md.) to remove free radiolabel. Both cell types were released from their culture surface with trypsinethylenediamine tetraacetic acid (EDTA) (Sigma). The cell suspensions were centrifuged ( × 800g) for 5 minutes. The supernatant fraction containing membrane bound a25I-LDL was removed and counted. The remaining cell pellet was washed with PBS, and its content of intracellular 125I-LDL was determined.
D N A assay The D N A content of both ECs and SMCs was measured by the method of Cesarone et al. 16After the 125I-LDL assay of the cell pellets was completed, 100 m m 3 of 0.19% Triton (Aldrich Chemical Company, Milwaukee, Wis.) was added to each. Aliquots (50 mm 3) of the resultant solution were diluted in 2 ml PBS to which 1 ml of Hoechst dye no. 33258 (0.15 mmol/L; Polysciences, Warrington, Pa.) was also added. The samples were stirred and allowed to stand in the dark for 10 minutes. The fluorescence of the solution was measured with use of a spectrofluorometer (Perkins Elmer, Norwalk, Conn.). This measurement was compared with a standard that was generated with calf thymus D N A (Sigma) with a range of 0.25 to 10 txg. In this manner, the D N A content of the cells was determined as an index of cell number.
Volume 15 Number 4 ..pril 1992
all-adenine and chromium 51 release assay To study the effect of thrombin on cellular metabolic function and viability, both all-adenine and chromim-n 51 (~Cr) release assays were performedY Separate monolayers of both ECs and SMCs were established in 24-well culture plates (Costar) by use of the same culture techniques. When confluent growth had been achieved, the media were changed to DMEM with 2% FBS containing either 3H-adenine (12.60 Ci/mmol/L) or S~Cr (374 mCi/mg; New England Nuclear, Boston, Mass.) at a concentration of i I*Ci/ml. The ceils were incubated for 24 hours. The media were then removed and the ceils washed with PBS. Fresh media containing 0, 10, or 50 units/ml bovine thrombin (Sigma) were added and the cells incubated for an additional 24 hours. T' ": media were rcmoved and counted with a liquid scintillation counter (Tricarb 2000; Packard, Meriden, Conn.) or a gamma counter (Packard). The ceils were washed with PBS released from the culture plates with trypsin-EDTA, and counted separately. The perceniEage release was calculated as media counts/(media + cell) counts × 100. Data analysis Endothelial cell permeability to ~2SI-LDL was expressed as'. counts in the lower well media and adjusted for specific activity of the radiolabel. Endothelial cell and SMC binding and uptake data were expressed as counts of radiolabel adjusted for specific activity and counts available on the media. Seven experiments were performed with quadruplicate values. Statistical analysis was performed with a microcomputer and the program True Epistat. Permeability, binding, and uptake data were compared with use of an unpaired, two-tailed Student's t test, with a p value of less than 0.05 being considered significant. Low-density lipoprotein uptake to binding ratios were compared by chi-square analysis. Cellular D N A content measurements were derived from two representative bilayer experiments and compared by means of the same statistical analysis. Two chromium and adenine release assays were performed with triplicate measurements at each concentration of thrombin. Calculated percentage release values were compared with an unpaired two-tailed Student's t test. RESULTS Cell morphology Examination of ceils exposed to thrombin under phase contrast microscopy demonstrated no gross morphologic change of the SMCs at either concen-
Thrombin alters LDL uptake by an EC-SMC bilayer
721
tration. Endothelial cells showed retraction at 24 hours that was more pronounced and prolonged at the higher concentration of thrombin (50 units/ml). Despite cellular retraction, loss of cellular adherence was not noted. Ceils exposed to thrombin 10 units/ml regained their normal architecture within 48 to 72 hours, whereas changes at the higher concentration were persistent. Endothelial permeability to 125I-LDL In bilayer culture, ECs exposed to thrombin at a concentration of 10 units/ml showed a minimal increase in permeability to 12SI-LDL within the first 48 hours of exposure. This increase became significant by 96 hours and persisted to 9 days (Table I). In contrast, at a concentration of 50 units/ml, thrombin resulted in an immediate and significant increase in EC permeability to 12SI-LDL, which persisted, although to a lesser degree, to 9 days. Endothelial cell binding and uptake of 12~I-LDL Exposure of ECs to the lower concentration of thrombin caused no significant alteration in 12SI-LDL binding to the cell membrane (Table II). An increase in cellular uptake of lipid was seen, although this was significant only at 96 hours and 9 days. The higher concentration of thrombin had the opposite effect, reducing both binding and uptake to a significant degree within 24 hours. Smooth muscle binding and uptake o f 125I-LDL Smooth muscle ceils in bilayer culture showed no significant response to the lower concentration of thrombin in either the binding or uptake of~2~I-LDL (Table III). At the higher concentration, ~25I-LDL binding was increased only at the later time points, but the cellular uptake of ~25I-LDL showed a marked and sustained increase beginning at 24 hours and persisting to 9 days. The uptake-to,binding ratio, reflecting the relative activity of receptor related LDL uptake, was minimally- changed at the lower concentrations of thrombin but significantly increased at the higher concentration. Cell number and viability Assay of the cellular DNA content, as an indicator of cell number, showed an increase in EC DNA with exposure to the lower concentration of thrombin (17.3 + .02 ~g DNA/well vs 1.48 + 0.08 ~g DNA/well;p < 0.001), but a marked decrease at the higher concentration of thrombin (1.28 + .09 t*g DNA/well) suggesting cell toxicity and loss at this level. Smooth muscle cells showed no significant
722 Alexander et al.
difference in D N A content with exposure to either the lower (4.59 _+ 0.62 /~g DNA/well vs 4.50 _ 0.47 /~g DNA/well) or the higher concentration of thrombin (4.65 _+ 0.68 t~g DNA/well). 51Cr release was increased in both cell types after exposure to thrombin, but to a greater extent in ECs (Table IV). This release was close related. ~H-adenine release, as a measure of cell injury, was increased in a dose related fashion by ECs. It was not increased by SMCs at the lower concentration of thrombin, but was significantly increased at the concentration of 50 units/ml. DISCUSSION Atherosclerosis is commonly thought to involve a break in the integrity of the arterial endothelium, resulting both in an incorporation of circulating plasma lipids and in a mitogenic stimulation of underlying smooth muscle cells by platelet-derived growth factors.18 Although evidence exists to support the belief that these events are a consequence of direct endothelial injury, s histologi c examination of atherosclerotic plaque has revealed elements of organizing thrombus, including platelets and fibrin. 1-s This finding has led Rokitansky s and others to postulate a role of thrombosis in the origin of atherogenesis. More recently, Jorgensen et al.19 have suggested that hemodynamic factors may promote the interaction of platelets and plasma proteins with focal areas of the arterial wall, which may potentiate the formation of thrombus within these regions, and that these events can occur in the absence of flank endothelial desquamarion. It is likely, therefore, that functional as well as structural abnormalities of the endothelium may be responsible for plaque formation.18 Activated platelets have been shown to contribute to endothelial dysfunction and to changes in SMC proliferation and metabolism} 9 Platelets could represent a primary factor linking thrombus and atherosclerotic plaque. However, platelet adherence to the endothelium is not uniformly found, nor are platelet effects on lipid uptake always evident. 2° This inability to attribute the development of atheroscleroric plaque entirely to platelet activity has led to a search for other factors involved in thrombus formation that could cause an alteration of arterial wall function or metabolism. Thrombin is a serine protease that stabilizes thrombus by promoting the formation of fibrin. It can also induce the aggregation and degranulation of platelets. 6 Thrombin has been found to interact directly with ECs by binding to membrane receptors in a specific, saturating fashion. 7 In addition, we have
Journal of VASCULAB~ SURGERt
previously reported that thrombin may influence cellular proliferation and matrix production by arterial cells and possibly influence both arterial healing and atherosclerotic plaque formation. 1° Small amounts ofthrombin have a marked effect on EC morphology, creating rapid but reversible changes in cellular architecture. 21 These changes appear to involve an energy-dependent contractile process that causes cellular retraction but does not appear to affect cell viability. ~1 Such thrombininduced alterations on cell conformation have been associated with an increase in the permeability of endothelial monolayers to macromolecules} 3 This effect can be inhibited by agents that specifically block the enzymatic site of the molecule. 22 These findings raise the possibility that thrombin's interaction with the vascular endothelium may increase its permea~ity to macromolecules and, as a result of cellular retraction, provide exposure of the subendothelium to circulating platelets and plasma proteins. ~2 Previous studies in our laboratory have indicated that small concentrations of thrombin can increase the permeability of an endothelial monolayer to albumin and stimulate SMC proliferation despite the presence of the endothelial barrier. 2s Others have shown that thrombin binding to ECs is greater in regions of decreased cell density. 24 It is possible, therefore, that the thrombin effect can occur without cell desquamarion and may be enhanced in zones of increased cell turnover, which can result from such factors as hemodynamic stress. 2s These zones of rapid cell turnover have been associated with abnormal levels of lipid uptake and deposirion. 26 In the present study, attempts were made to determine whether thrombin by itself could affect LDL permeability and uptake by an EC-SMC bilayer. To incorporate aspects of an injury model while allowing for the study of LDL permeability, both ECs and SMCs in bilayer culture were exposed to the same concentrations of thrombin. This system, however, may not represent an ideal physiologic model, because it makes the assumption that in areas of injury with thrombus formation both underlying SMCs and adjacent ECs are exposed to similar concentrations of thrombin. Exposure of the bilayer to a moderate concentration of thrombin (10 units/ml) did result in a limited and reversible retraction of ECs without loss of cellular adherence. These gross, morphologic changes were accompanied by an early but mild increase in LDL permeability, which became statistically significant at 96 hours, and persisted to 9 days despite apparent
Volume 15 Number 4 April 1992
normalization of monolayer architecture. It is not clear whether cell retraction alone was responsible for this permeability change. However, normalization of EC morphology during the period of increased permeabilitT would suggest that the latter was due to an alteration in cellular function. This might be supported by the paralleled increase in cellular uptake ofLDL. ~m absence of change in membrane binding would imply that this increase in cell uptake was not receptor mediated. A higher concentration of thrombin (50 units/ml) resulted in an early and severe increase in EC permeability to LDL, and a marked decrease in both the binding and uptake of LDL by the EC. These changes corresponded to severe and prolonged cellular reta:action, a marked reduction in cellular tuNA content, and a significant increase in both SlCr and all-adenine release. Such findings suggest that high concentrations of thrombin are directly toxic to ECs, resulting in cell loss and a release of adenine nucleotides. As suggested by Pearson and Gordon 27 this effect could potentially expose the subendothelium to platelets and plasma lipids, and provide free adenine as a stimulant for platelet aggregation and release. In contrast, a smaller concentration of thrombin, although causing a more limited release of 3H-adenine and 51Cr, increased the cellular DNA content. T]hese results might suggest a mitogenic response of ECs to lower concentrations of thrombin, which in conjunction with an increase in permeability, might lead to an increase in lipid infiltration without overt cellular loss. Such results would support the theory that factors involved in thrombus formation could affect lipid handling by the arterial endothelium without endothelial desquamation. Unlike the ECs, the SMCs revealed no gross morphologic change at either concentration of thrombin. The lower concentration of thrombin had no effect on either the cellular binding or uptake of ~25I-LDL. However, the higher concentration of thrombin markedly increased cellular uptake of LDL throughout the course of study, while causing a more limited and late increase in cellular binding. The LDL uptake-to-binding ratio, reflecting the level of nonreceptor mediated uptake, was unchanged at the lower concentration of thrombin but significantly increased at the higher level. These data, again, suggest that SMC lipid uptake is enhanced by thrombin and must occur through either a receptor independent process (bulk phase uptake) or an increase in receptor turnover. Smooth muscle cells demonstrated an increase in
Thrombin alters LDL uptake by an EC-SMC bilayer 723
SlCr release in response to thrombin which was dose related but less pronounced than for ECs. This release was not associated with a change in DNA content, possibly indicating an increase in cell turnover without net cell loss. 3H-adenine release was not stimulated by the lower concentration of thrombin but was significantly increased by the higher concentration, pointing to a thrombin-induced alteration in cellular metabolism. From these studies, it is apparent that thrombin can directly affect LDL permeability and uptake by arterial ECs and SMCs in coculture. Endothelial cells appear to be exquisitely more sensitive to thrombin than do SMCs, showing evidence of structural and functional abnormality at moderate doses of thrombin and of overt toxicity at higher concentrations. By comparison, SMCs are relatively insensitive to the effects of moderate concentrations of thrombin, but display a marked increase in LDL uptake in response to higher concentrations. These findings raise an important question as to whether thrombin can mediate or significantly influence the proliferation and utilization of lipid by cells of the arterial wall. Although serum thrombin levels have been found to be lower than the concentrations used in this study, 28 the levels achieved near sites of active thrombus formation have been calculated to be 9 units/ml. 29 It may be postulated that thrombin levels within thrombus itself might be considerably greater, although direct measurement of these levels has not been reported. In such areas of expected high thrombin concentration, endothelial toxicity combined with an enhanced uptake of LDL by underlying SMCs may contribute to the formation of atherosclerotic plaque. REFERENCES 1. Duguid }'B. Thrombosis as a factor in the pathogenesis of aortic atherosderosis, l Pathol BacterioI 1948;60:57-61. 2. More RM, Movat HZ, Hanst DM. Role of mura! fibrin thrombi of the aorta in genesis of arteriosclerotic plaques. Arch Pathol 1957;63:612-20. 3. Schwartz CJ, Valente AJ, Kelly JL, Sprague EA, Edwards EH. Thrombosis and the development of atherosclerosis: Rokitansky revisited. Semin Thromb Hemost 1988;14:189-95. 4. Woolf N, Carstairs KC. Infiltration and thrombosis in atherogenesis. A study using immunofluorescent techniques. Am ~ Pathol 1967;51:373-86. 5. Ip JH, Fuster V, Badiman L, Badiman I, Taubman MD, Chesbro JH. Syndromes of accelerated atherosclerosis; role of vascular injury and smooth muscle cell proliferation. ~Am Coil Cardiol 1990;15:1667-87. 6. Walsh PN, Schmaier AH. Platelet-coagulant protein interactions. In: Colman RW, Hirsh J, Marder VI, Salzman EW, eds. Hemostasis and thrombosis. Philadelphia: JB Lippincott, I987:689-709.
724
Journal of VASCULAR SURGERY
Alexander et al.
7. Awbry BJ, Hoak JC, Owen WG. Binding of human thrombin to cultured human endothelial cells. J Biol Chem 1979;254: 4092-5. 8. Gospodarowicz D, Brown KD, Birdwell CR, Zetter BR. Control of proliferation of human vascular endothelial cells. J Cell Biol 1978;77:774-88. 9. Weksler BB, Ley CW, Jaffee EA. Stimulation of endothelial cell prostacyclin production by thrombin, trypsin, and the ionophore A23187. J Clin Invest 1978;62:923-30. 10. Graham DJ, Alexander JJ. The effects ofthrombin on bovine aortic endothelial and smooth muscle cells. I VAse SURG i990;11:307-13. 11. Galdal KS, Evensen SA, Nilsen E. Thrombin-induced shape changes of cultured endothelialcells: metabolic and functional observations. Thromb Res 1983;32:57-66. 12. Laposata M, Dovnarsky DK, Shin HS. Thrombin-induced gap formation in confluent endothelial cell monolayers in vitro. Blood 1983;62:549-56. 13. Garcia JG, Siflinger-Birnboim A, Bizios R, DelVecchio P, Fenton JW II, Malik AB. Thrombin-induced increases in albumin transport across cultured endothelial monolayers. J Cell Physiol i986;128:96-104. 14. Alexander Jl, Miguel R, Graham D. Low-density lipoprotein uptake by an endothelial-smooth muscle cell bilayer, l VASC SURG 1991;13:444-5i. 15. Markwell MA. A new solid state reagent to iodinate proteins. Anal Biochem 1982; 125:427-32. 16. Cesarone CF, Bolognesi C, Santi L. Improved microfluorometric DNA determination in biological material using 33258 Hoechst. Anal Biochem 1979;100:188-97. 17. Andreoli SP, Baehner ILL, Bergstein JM. In vitro detection of endothelial cell damage using 2-deoxy-D-3H-glucose: comparison with chromium 51, 3H-leucine, 3H-adenine and lactate dehydrogenase, l Lab Clin Med 1985;106:25361. 18. Morrel EM, Holland JA, Pritchard ICA, Cotton CK, Stemerman MB. Endothelial perturbation and low density lipoprotein. Ann NY Acad Sci I987;516:412-7.
i9. Jorgensen L, Peckham MA, Rowsell HC, Mustard IF. Deposition of formed elements of blood on the intima and signs ofintimal injury in the aorta of rabbit, pig and man. Lab Invest I972;27:341-50. 20. Jorgensen L, Rowsell HC, Hovig T, Mustard IF. Resolution and organization of platelet-rich mural thrombi in carotid arteries of swine. Am J Pathol 1967;51:681-93. 21. Galdal KS, Evensen SA. Effects of divalent cations and various vasoactive and haemostatistically active agents on the integrity of monolayers of human endothelial cells. Thromb Res 1981;21:273-84. 22. DeMichele /VIA, Moon DG, Fenton JW II, Minnear FL. Thrombin's enzymatic activity increases permeability of endothelial cell monolayers. J Appl Physiol 1990;69:1599-606. 23. Graham DJ, Alexander JJ, Miguel R. Thrombin alters permeability and proliferation of co-cultured endothelial and smooth muscle cells. Surg Forum 1990;46:322-4. 24. Isaacs J, Savion N, Gospodarowicz D, Shuman MA. Effect of cell density on thrombin binding to a specific site on bovine vascular endothelial cells, l Cell Biol 1981;90:670-4. 25. Schwartz SM, Benditt EP. Aortic endothelial cell replications.. I. Effects of age and hypertension in the rat. Circ Res 1977;41:248-55. 26. Lin S-J, Jan K-M, Weinbaum S, Chien S. Transendothelial transport of low density lipoprotein in association with cell mitosis in rat aorta. Arteriosclerosis 1989;9:230-6. 27. Pearson ID, Gordon JL. Vascular endothelial and smooth muscle cells in culture selectively release adenine nucleotides. Nature 1979;281:384-6. 28. Zimmerman TS, Frereo I, Rothberger H. Blood coagulation and inflammatory response. Semin Haematol 1977;14:391404. 29. Hubbell IA, McIntire LV. Platelet active concentration profiles near growing thrombi: a mathematical consideration. Biophys I 1986;50:937-45. Submitted June 10, 1991; accepted Oct. 7, 1991.
DISCUSSION
Dr. John Sharefldn (Bethesda, Md.). This is an elegantly conducted study showing that alpha thrombin can, at suffidently high doses o f up to 50 units/ml, induce an increase in the uptake of L D L by SMCs even when these are beneath an initially intact endothelial layer. The fnding is of interest because the hypothesis has been around awhile. Kaddish and Folkman actually had suggested in 1979 that a fibrin clot from which active thrombin can be released by fibrinolysis could cause loss o f cell-to-ceU contact in the endothelial layer for several hours, and they proposed that persistence of fibrin clots could induce the early stages of atherosclerosis by disrupting the endothelial layer to allow lipid uptake. I have three questions for the authors about this study. What do you think the actual mechanism o f thrombin-
induced increases in SMC L D L uptake is? Thrombin is known to be able to both release platelet-derived growth factor from ECs, even in doses o f one tenth of a unit per ml sometimes, and certainly at 1 unit/ml it can induce the increase in gene expression for platelet-derived growth factor in endothelium. Platelet-derived growth factor in turn can increase the amount of 12~I labeled human L D L bound to and internalized by SMCs within approximately 24 hours, in part by increasing the number of LDL receptors by up to fourfold. D o you think this mechanism could be tested by either blocking thrombin-iuduced P D G F released with phosphodiesterase inhibitors such as amrinone or by adding dibuterol cyclic adenosine monophosphate or by adding platelet-derived growth factor antagonists or using monospecific antiplatelet-derived
Volume 15 Number 4 April 1992
growth factor activities? It seems this might be an intriguing way to see if that was one of the mechanisms for the increased uptake you observed. My second question is really what the authors think is the comparative importance of the roles of this sort of action of thrombin inducing loss of integrity in an EC layer versus other effects, because thrombin can also induce the expression of endothelial leukocyte and circulating monocyte adhesior~ molecules as inducers of atherogenic processes. The fact is that an atherosclerotic lesion's earliest form, such as the fatty streak, detectable for instance at sites of flow disturbance in hypercholesterolemic rabitt, is there even in the total absence of gross thrombus formation; and do the authors think a model system where they added a monocyte line such as U937 or normal peripheral blood monocytes would be able to test for this other hypothesis about the initiation of atherogenesis simply because the endothelial layers in the fatty streaks are intact and show no c. ,dence of retraction? These lesions begin to occur with an intact endothelial layer, even when monocytes can be seen beneath them. I think my final and maybe most important question is the choice ,of the dose of 50 u n i t s - N I H units I p r e s u m e - o f alpha thrombin per milliliter. Do you think that such a dose, which had some lethality to ECs, is even compatible with the finding of a totally intact endothelial layer over early lesions such as fatty streaks? I have been unable in looking at the literature of thrombin effects on vascular wall cells to find use of a dose outside the range of 0.1 to 10 units a milliliter, and the reason for this choice is based on a number of studies. Both Schuman and Medieros in 1976, and Aronson and Finlas in 1977 measured thrombin activity in the highest possible conditions, which is the totally static clotting of fresh whole blood, and they did not get levels over 6 to 7 or 10 units a milliliter, and those only peaked at 11 minutes and rapidly went down thereafter. That was in the absence of dilution of it with regard to flow. Much lower doses than this can produce the effects like retraction of human cells as observed at 0.1 unit of thrombin. We can persistently induce PDGF R N A transcript increase in human ECs in the range 1 to 10 units and all sorts of other effects on smooth muscle like c-phos gene activation and phosphatidyl inositol 3,4,5triphosphate phosphate activation can occur at this lower range. In a similar way all these changes can occur at this much lower dose range. My question is do the authors have
Thrombin alters LDL uptake by an EC-SMC bilayer 725
evidence for the occurrence of this higher levei of thrombin activity being attained in bovine coagulation systems at all or do they think there is a large difference perhaps in sensitivity to thrombin between bovine and human vascular cells that required use of this higher dose of 50 units? Dr. J. Jeffrey Alexander. In regard to the first question concerning the production of growth factors by the endothelial layer, it is my impression that the endothelial layer has multiple roles in this bilayer system. We have previously looked at other injury models using this same culture system, and have found that the ECs do affect underlying SMCs, altering their rate of replication as well as their utilization of LDL. I refer to several studies primarily by Somer and Minick in which in vivo models demonstrate that lipid uptake by the arterial wall is greater in areas where endothelial regeneration has occurred, as compared to areas of denuded endothelium. This finding suggests that the EC regulates lipid uptake by the underlying SMCs and that metabolic dysfunction of the endothelium influences SMC function. Further study of this interaction is needed, including the role of growth factors in intercellular dynamics. Regarding the second question, the study presented today really has little to do with early fatty streak formation. This model would be more applicable to mature or fibrous plaque. Fatty streak formation would not be expected to involve thrombin deposition or interaction with the arterial endothelium. A need certainly exists to expand this bilayer culture model using monocytes. Lipid accumulation is a net result of both deposition and clearance. In addition, it is possible that monocytes may affect arterial cell lipid metabolism either directly or indirectly. The dose of 50 units of thrombin is a high one. Previous studies have shown significant effects on the EC at doses ranging from 1 to 10 units per milliliter. The concentration of thrombin occurring in thrombus is not well defined, and it was supposed that this could be higher than the 10 units/ml that has been measured in serum. We, therefore, arbitrarily chose a level of 50 units/m[ as our higher concentration, and compared these results with the lower concentration of 10 units/ml as suggested by our sponsor, Dr. Linda Graham. I suspect that results obtained using smaller concentrations ofthrombin would be similar, although perhaps not as pronounced as those that have been presented today.
Histologic evidence o f fibrotic transformation and lipid incorporation by organizing mural thrombus has implicated intravascular thrombosis in the pathogenesis o f atherosclerosis) -3 Although this mechanism may not be applicable to fatty streak formation, it does provide a plausible explanation for the nature o f established or rapidly progressing atherosclerotic plaque, where the presence o f fibrin and platelets would indicate a thrombotic event. 4 This theory does not necessarily exclude the more commonly cited response to injury hypothesis, which From the Department of Surgery, Case Western ReserveUniversity, ClevelandMetropolitan GeneralHospital. Presented at the Forty-fifth Annual Meeting of the Society for Vascular Surgery,Research Forum II, Boston, Mass., June 4-5, 1991. Reprint requests: I. leffrey Alexander, MD, Departmen~~'of Surgery,ClevelandMc~tropolitanGeneralHospital, 3395 Scranton Rd., Cleveland;~OH 44109, 24/1]34220 718
proposes that a break in either the functional or structural integrity o f the endothelial barrier results in platelet adherence, lipid infiltration, and a mitogenic stimulation o f underlying smooth muscle cells (SMCs). s Under such circumstances, thrombus formation may occur, providing a site for a localized and enhanced interaction between the vessel wall and circulating platelets, coagulation factors, and plasma lipids. These interactions may lead to changes in the arterial wall that are characteristic o f atherosclerotic plaque. Thrombin is an essential element in thrombus formation, providing a stabilizing network o f fibrin through its conversion offibrinogen. It has also been found to promote platelet aggregation and release. 6 In addition to its role in coagulation, thrombin interacts directly with the endothelial cell (EC), to which it binds in a specific and saturating fashion. 7 Previous studies have indicated that it may also affect
Voltune 15 Number 4 ~.pril 1992
Thrombin alters LDL uptake by an EC-SB4C bilayer 719
Table I. ~2e;I-LDLcontent of the lower chamber media (media counts/available counts ___ SD × 10 3) Time 24 hours 48 hours 96 hours 9 days
Thrombin 10 units/ml
Control 366.4 380.8 354.2 360.8
-± ± ±
40.9 41.5 32.1 29.3
391.6 393.0 415.0 414.0
± ± ± ±
34.8 46.0 39.4 9.7
p~
Thrombin 50 units/ml
Control
NS NS <0,0001 <0.005
447.2 446.6 416.5 470,5
± 27.9 _+ 17.1 ± 32.2 ± 5.1
532,8 557,2 523,4 542,3
± ± ± ±
23.3 43,3 40.9 30.0
p~ < 10 -7 < 10 -s <0.0005 <0.001
tp unpaired two-tailed Student's t test.
Table II. Endothelial cell binding and uptake of ~2SI-LDL (counts/available counts + SD × 10 a) Time
Control
Thrombin 10 units/ml
p~
Control
Thrombin 50 units/ml
p~
Binding 24 hours 48 hours 96 hours 9 days
152.9 108.0 98.8 118.1
-+ 38.8 ± 15.3 -+ 13,4 _+ 16.2
149,8 167.9 105.4 107.8
± 50.2 +- 59.7 ± 11.2 ± 11.8
NS <0.005 NS NS
227.2 235.3 185.6 152.9
± 35.6 ± 31.5 -+ 18.1 ± 25,2
186.1 183.1 150.3 76.3
± 38.9 ± 36.4 ± 13.4 _+ 5.5
<0.01 <0.0005 <0.005 <0.001
:24 hours ,48 hours 96 hours 9 days
485.7 645.7 474.6 252.7
-+ ---±
505.4 756.6 648,2 437.8
-+ 112.6 -- 150.0 ± 126,3 _+ 141.7
NS NS <0,05 <0.01
567.0 471.8 402.8 273.0
± ± ± ±
357.5 329.2 288.6 97.7
± 44.5 ± 39,9 -+ 61.6 ± 12.4
< 10 -7 < 10 -s <0.05 <0.001
Uptake 145.6 197.1 198.5 111.1
68.3 77,3 92,2 60.2
*p unpaired two-tailed Student's t test.
cellular proliferation, 8 prostanoid secretion,9 and matrix production by arterial cells. 1° In addition, thrombin increases EC contractility, resulting in cellular retraction, n,~2 This has been associated with an increased permeability of endothelial monolayers to macromolecules,13 which may represent a significant alteration of barrier function. It is therefore possible that thrombin may promote not only platelet adhesion and activation, but also lipid uptake through widened intercellular junctions within the endothelium. The purpose of these experiments was to study the effect of' thrombin on low-density lipoprotein (LDL) permeability and uptake by an EC-SMC bilayer and the potential influence of thrombininduced cell :injury on these processes. MATERIAl, A N D M E T H O D S Cell harvesthlg and culture Endothelial cells and SMCs were harvested from bovine aortas as previously described) ~ Cells in passages 6 through 8 were used for study. Endothelial cells were identified by cellular morphology and immunofluoresccnt staining of factor-VIII related antigen. Smooth muscle cells were identified by their characteristic morphologic appearance. Both cell types were plated on transwell culture plates (Costax Corp., Cambridge, Mass.) consisting
of two separate culture chambers separated by a porous polycarbonate filter. 14 Smooth muscle cells were initially placed in the lower well in 750 mm 3 of culture medium with a plating density of 25,000 cells/cm2. The ECs were then grown on the polycarbonate filters in the upper well in 200 mm 3 of culture medium with the same plating density. Confluent growth, which was achieved in approximately 5 days, was verified by phase contrast microscopy. Once cell confluence has been reached, bovine thrombin (95 units/mg protein; Sigma Diagnostics, St. Louis, Mo.) was added to the upper and lower well media of experimental wells to achieve concentrations of either 10 or 50 units/ml. The bilayers were then incubated for periods ranging from 24 hours to 9 days. The cells were inspected daily, and the media were changed every 3 days. After completion of the incubation period, the media were removed and replaced with Dulbecco's modified Eagle's medium (DMEM) containing 2% fetal bovine serum (FBS) and bovine thrombin. Iodine 125 (12sI)-LDL (10 ~g protein/ml; specific activity 172 to 879 cpm/~g protein) was added to the upper well media. Iodine 125-LDL was prepared weekly by the method of Markwell, 15 using human LDL (Sigma; 6.4 mg protein/ml), and achieving a 12SI-LDL binding efficiency of 91% to 94%. Low-density lipoprotein was maintained for not longer than 3 weeks in refriger-
720
Journal of VASCULAR SURGER~.~
A l e x a n d e r et al.
Table III. Smooth muscle cell binding and uptake of ~25I-LDL (counts/available counts + SD × 103) Time
Thrombin 10 units/ml
Control
p~
Control
Thrombin 50 units/ml
p~
Binding 24 hours 48 hours 96 hours 9 days
295.6 285.9 323.2 438.4
± 78.9 -+ 21.8 -+ 72.3 ± 44.2
255.6 320.2 321.2 363.9
-+ 23.1 ± 58.2 ± 41.0 ± 71.2
NS NS NS NS
260.7 284.7 206.7 233.7
± ± ± ±
24 hours 48 hours 96 hours 9 days Uptake/binding ratio
149.9 151.0 176,9 174,2
± ± ± ±
125.8 149.3 192.4 155.1
_+ 11.5 ± 10.8 ± 34.3 ± 19.9
< .005 NS NS NS
128.4 189.9 91.3 110.2
34.5 45.5 34.1
17.7
269.8 315.6 249.9 287.6
± ± ± ±
41.1 56.9 22.4 49.0
± 17.6 ± 19.1 _+ 5.7 ± 6.7
248.9 505.3 351.4 393.6
± 22.5 ± 71.9 ± 46.4 _+ 36.8
NS NS <0.05 <0,05
Uptake
24 hours 48 hours 96 hours 9 days
14.8 22.8 25.4 39.1
< 10 -7 < 10 -7 < 10 7 <0.001
pt 0.51 0.53 0.55 0.40
0,49 0,46 0.60 0.42
NS < 0.01 < 0.05 NS
pt 0.49 0.67 0.44 0.47
0.92 1.60 1.40 1.37
< < < <
10 -7 10 -7 10 -7 10~f
~p unpaired two-tailed Student's t test; p~-chi-square analysis.
Table IV. Thrombin-induced 3H-adenine and 5~Cr release from endothelial and smooth muscle ceils ~ Thrombin 3H-Adenine Release Control 10 units/ml 50 units/ml 5ICr release Control 10 units/ml 50 units/ml ~24-hour exposure; % release =
Endothelial
p~
Smooth muscle
p~
11.83 ± 0.47 24.20 ± 1.05 31.63 ± 1.37
< i0 -s < 10 -5
20.57 ± 1.08 20.56 ± 1.17 30.39 -+ 0.52
NS <10 s
28.43 ± 0.62 35.65 ± 0.93 39.77 ± 1.28
< 0.00005 < 0.000005
30.65 ± 0.74 32.53 +- 0.50 36.10 ± 0.73
<0.001 <0.00005
counts media counts (ceils + media)
± SD x 100
ated storage (4 ° C) to prevent oxidation. The cells were incubated for 3 hours, after which the medium from each chamber was removed and its 125I-LDL content determined by a gamma counter (Packard 5000; Packard, Downers Grove, Ill.). The 125I-LDL media content of the lower well was used as a measure of EC permeability. The ECs on the polycarbonate filter and the SMCs on the lower culture chamber were washed with phosphate-buffered saline solution (PBS; Wit-taker Bioproducts, Walkersville, Md.) to remove free radiolabel. Both cell types were released from their culture surface with trypsinethylenediamine tetraacetic acid (EDTA) (Sigma). The cell suspensions were centrifuged ( × 800g) for 5 minutes. The supernatant fraction containing membrane bound a25I-LDL was removed and counted. The remaining cell pellet was washed with PBS, and its content of intracellular 125I-LDL was determined.
D N A assay The D N A content of both ECs and SMCs was measured by the method of Cesarone et al. 16After the 125I-LDL assay of the cell pellets was completed, 100 m m 3 of 0.19% Triton (Aldrich Chemical Company, Milwaukee, Wis.) was added to each. Aliquots (50 mm 3) of the resultant solution were diluted in 2 ml PBS to which 1 ml of Hoechst dye no. 33258 (0.15 mmol/L; Polysciences, Warrington, Pa.) was also added. The samples were stirred and allowed to stand in the dark for 10 minutes. The fluorescence of the solution was measured with use of a spectrofluorometer (Perkins Elmer, Norwalk, Conn.). This measurement was compared with a standard that was generated with calf thymus D N A (Sigma) with a range of 0.25 to 10 txg. In this manner, the D N A content of the cells was determined as an index of cell number.
Volume 15 Number 4 ..pril 1992
all-adenine and chromium 51 release assay To study the effect of thrombin on cellular metabolic function and viability, both all-adenine and chromim-n 51 (~Cr) release assays were performedY Separate monolayers of both ECs and SMCs were established in 24-well culture plates (Costar) by use of the same culture techniques. When confluent growth had been achieved, the media were changed to DMEM with 2% FBS containing either 3H-adenine (12.60 Ci/mmol/L) or S~Cr (374 mCi/mg; New England Nuclear, Boston, Mass.) at a concentration of i I*Ci/ml. The ceils were incubated for 24 hours. The media were then removed and the ceils washed with PBS. Fresh media containing 0, 10, or 50 units/ml bovine thrombin (Sigma) were added and the cells incubated for an additional 24 hours. T' ": media were rcmoved and counted with a liquid scintillation counter (Tricarb 2000; Packard, Meriden, Conn.) or a gamma counter (Packard). The ceils were washed with PBS released from the culture plates with trypsin-EDTA, and counted separately. The perceniEage release was calculated as media counts/(media + cell) counts × 100. Data analysis Endothelial cell permeability to ~2SI-LDL was expressed as'. counts in the lower well media and adjusted for specific activity of the radiolabel. Endothelial cell and SMC binding and uptake data were expressed as counts of radiolabel adjusted for specific activity and counts available on the media. Seven experiments were performed with quadruplicate values. Statistical analysis was performed with a microcomputer and the program True Epistat. Permeability, binding, and uptake data were compared with use of an unpaired, two-tailed Student's t test, with a p value of less than 0.05 being considered significant. Low-density lipoprotein uptake to binding ratios were compared by chi-square analysis. Cellular D N A content measurements were derived from two representative bilayer experiments and compared by means of the same statistical analysis. Two chromium and adenine release assays were performed with triplicate measurements at each concentration of thrombin. Calculated percentage release values were compared with an unpaired two-tailed Student's t test. RESULTS Cell morphology Examination of ceils exposed to thrombin under phase contrast microscopy demonstrated no gross morphologic change of the SMCs at either concen-
Thrombin alters LDL uptake by an EC-SMC bilayer
721
tration. Endothelial cells showed retraction at 24 hours that was more pronounced and prolonged at the higher concentration of thrombin (50 units/ml). Despite cellular retraction, loss of cellular adherence was not noted. Ceils exposed to thrombin 10 units/ml regained their normal architecture within 48 to 72 hours, whereas changes at the higher concentration were persistent. Endothelial permeability to 125I-LDL In bilayer culture, ECs exposed to thrombin at a concentration of 10 units/ml showed a minimal increase in permeability to 12SI-LDL within the first 48 hours of exposure. This increase became significant by 96 hours and persisted to 9 days (Table I). In contrast, at a concentration of 50 units/ml, thrombin resulted in an immediate and significant increase in EC permeability to 12SI-LDL, which persisted, although to a lesser degree, to 9 days. Endothelial cell binding and uptake of 12~I-LDL Exposure of ECs to the lower concentration of thrombin caused no significant alteration in 12SI-LDL binding to the cell membrane (Table II). An increase in cellular uptake of lipid was seen, although this was significant only at 96 hours and 9 days. The higher concentration of thrombin had the opposite effect, reducing both binding and uptake to a significant degree within 24 hours. Smooth muscle binding and uptake o f 125I-LDL Smooth muscle ceils in bilayer culture showed no significant response to the lower concentration of thrombin in either the binding or uptake of~2~I-LDL (Table III). At the higher concentration, ~25I-LDL binding was increased only at the later time points, but the cellular uptake of ~25I-LDL showed a marked and sustained increase beginning at 24 hours and persisting to 9 days. The uptake-to,binding ratio, reflecting the relative activity of receptor related LDL uptake, was minimally- changed at the lower concentrations of thrombin but significantly increased at the higher concentration. Cell number and viability Assay of the cellular DNA content, as an indicator of cell number, showed an increase in EC DNA with exposure to the lower concentration of thrombin (17.3 + .02 ~g DNA/well vs 1.48 + 0.08 ~g DNA/well;p < 0.001), but a marked decrease at the higher concentration of thrombin (1.28 + .09 t*g DNA/well) suggesting cell toxicity and loss at this level. Smooth muscle cells showed no significant
722 Alexander et al.
difference in D N A content with exposure to either the lower (4.59 _+ 0.62 /~g DNA/well vs 4.50 _ 0.47 /~g DNA/well) or the higher concentration of thrombin (4.65 _+ 0.68 t~g DNA/well). 51Cr release was increased in both cell types after exposure to thrombin, but to a greater extent in ECs (Table IV). This release was close related. ~H-adenine release, as a measure of cell injury, was increased in a dose related fashion by ECs. It was not increased by SMCs at the lower concentration of thrombin, but was significantly increased at the concentration of 50 units/ml. DISCUSSION Atherosclerosis is commonly thought to involve a break in the integrity of the arterial endothelium, resulting both in an incorporation of circulating plasma lipids and in a mitogenic stimulation of underlying smooth muscle cells by platelet-derived growth factors.18 Although evidence exists to support the belief that these events are a consequence of direct endothelial injury, s histologi c examination of atherosclerotic plaque has revealed elements of organizing thrombus, including platelets and fibrin. 1-s This finding has led Rokitansky s and others to postulate a role of thrombosis in the origin of atherogenesis. More recently, Jorgensen et al.19 have suggested that hemodynamic factors may promote the interaction of platelets and plasma proteins with focal areas of the arterial wall, which may potentiate the formation of thrombus within these regions, and that these events can occur in the absence of flank endothelial desquamarion. It is likely, therefore, that functional as well as structural abnormalities of the endothelium may be responsible for plaque formation.18 Activated platelets have been shown to contribute to endothelial dysfunction and to changes in SMC proliferation and metabolism} 9 Platelets could represent a primary factor linking thrombus and atherosclerotic plaque. However, platelet adherence to the endothelium is not uniformly found, nor are platelet effects on lipid uptake always evident. 2° This inability to attribute the development of atheroscleroric plaque entirely to platelet activity has led to a search for other factors involved in thrombus formation that could cause an alteration of arterial wall function or metabolism. Thrombin is a serine protease that stabilizes thrombus by promoting the formation of fibrin. It can also induce the aggregation and degranulation of platelets. 6 Thrombin has been found to interact directly with ECs by binding to membrane receptors in a specific, saturating fashion. 7 In addition, we have
Journal of VASCULAB~ SURGERt
previously reported that thrombin may influence cellular proliferation and matrix production by arterial cells and possibly influence both arterial healing and atherosclerotic plaque formation. 1° Small amounts ofthrombin have a marked effect on EC morphology, creating rapid but reversible changes in cellular architecture. 21 These changes appear to involve an energy-dependent contractile process that causes cellular retraction but does not appear to affect cell viability. ~1 Such thrombininduced alterations on cell conformation have been associated with an increase in the permeability of endothelial monolayers to macromolecules} 3 This effect can be inhibited by agents that specifically block the enzymatic site of the molecule. 22 These findings raise the possibility that thrombin's interaction with the vascular endothelium may increase its permea~ity to macromolecules and, as a result of cellular retraction, provide exposure of the subendothelium to circulating platelets and plasma proteins. ~2 Previous studies in our laboratory have indicated that small concentrations of thrombin can increase the permeability of an endothelial monolayer to albumin and stimulate SMC proliferation despite the presence of the endothelial barrier. 2s Others have shown that thrombin binding to ECs is greater in regions of decreased cell density. 24 It is possible, therefore, that the thrombin effect can occur without cell desquamarion and may be enhanced in zones of increased cell turnover, which can result from such factors as hemodynamic stress. 2s These zones of rapid cell turnover have been associated with abnormal levels of lipid uptake and deposirion. 26 In the present study, attempts were made to determine whether thrombin by itself could affect LDL permeability and uptake by an EC-SMC bilayer. To incorporate aspects of an injury model while allowing for the study of LDL permeability, both ECs and SMCs in bilayer culture were exposed to the same concentrations of thrombin. This system, however, may not represent an ideal physiologic model, because it makes the assumption that in areas of injury with thrombus formation both underlying SMCs and adjacent ECs are exposed to similar concentrations of thrombin. Exposure of the bilayer to a moderate concentration of thrombin (10 units/ml) did result in a limited and reversible retraction of ECs without loss of cellular adherence. These gross, morphologic changes were accompanied by an early but mild increase in LDL permeability, which became statistically significant at 96 hours, and persisted to 9 days despite apparent
Volume 15 Number 4 April 1992
normalization of monolayer architecture. It is not clear whether cell retraction alone was responsible for this permeability change. However, normalization of EC morphology during the period of increased permeabilitT would suggest that the latter was due to an alteration in cellular function. This might be supported by the paralleled increase in cellular uptake ofLDL. ~m absence of change in membrane binding would imply that this increase in cell uptake was not receptor mediated. A higher concentration of thrombin (50 units/ml) resulted in an early and severe increase in EC permeability to LDL, and a marked decrease in both the binding and uptake of LDL by the EC. These changes corresponded to severe and prolonged cellular reta:action, a marked reduction in cellular tuNA content, and a significant increase in both SlCr and all-adenine release. Such findings suggest that high concentrations of thrombin are directly toxic to ECs, resulting in cell loss and a release of adenine nucleotides. As suggested by Pearson and Gordon 27 this effect could potentially expose the subendothelium to platelets and plasma lipids, and provide free adenine as a stimulant for platelet aggregation and release. In contrast, a smaller concentration of thrombin, although causing a more limited release of 3H-adenine and 51Cr, increased the cellular DNA content. T]hese results might suggest a mitogenic response of ECs to lower concentrations of thrombin, which in conjunction with an increase in permeability, might lead to an increase in lipid infiltration without overt cellular loss. Such results would support the theory that factors involved in thrombus formation could affect lipid handling by the arterial endothelium without endothelial desquamation. Unlike the ECs, the SMCs revealed no gross morphologic change at either concentration of thrombin. The lower concentration of thrombin had no effect on either the cellular binding or uptake of ~25I-LDL. However, the higher concentration of thrombin markedly increased cellular uptake of LDL throughout the course of study, while causing a more limited and late increase in cellular binding. The LDL uptake-to-binding ratio, reflecting the level of nonreceptor mediated uptake, was unchanged at the lower concentration of thrombin but significantly increased at the higher level. These data, again, suggest that SMC lipid uptake is enhanced by thrombin and must occur through either a receptor independent process (bulk phase uptake) or an increase in receptor turnover. Smooth muscle cells demonstrated an increase in
Thrombin alters LDL uptake by an EC-SMC bilayer 723
SlCr release in response to thrombin which was dose related but less pronounced than for ECs. This release was not associated with a change in DNA content, possibly indicating an increase in cell turnover without net cell loss. 3H-adenine release was not stimulated by the lower concentration of thrombin but was significantly increased by the higher concentration, pointing to a thrombin-induced alteration in cellular metabolism. From these studies, it is apparent that thrombin can directly affect LDL permeability and uptake by arterial ECs and SMCs in coculture. Endothelial cells appear to be exquisitely more sensitive to thrombin than do SMCs, showing evidence of structural and functional abnormality at moderate doses of thrombin and of overt toxicity at higher concentrations. By comparison, SMCs are relatively insensitive to the effects of moderate concentrations of thrombin, but display a marked increase in LDL uptake in response to higher concentrations. These findings raise an important question as to whether thrombin can mediate or significantly influence the proliferation and utilization of lipid by cells of the arterial wall. Although serum thrombin levels have been found to be lower than the concentrations used in this study, 28 the levels achieved near sites of active thrombus formation have been calculated to be 9 units/ml. 29 It may be postulated that thrombin levels within thrombus itself might be considerably greater, although direct measurement of these levels has not been reported. In such areas of expected high thrombin concentration, endothelial toxicity combined with an enhanced uptake of LDL by underlying SMCs may contribute to the formation of atherosclerotic plaque. REFERENCES 1. Duguid }'B. Thrombosis as a factor in the pathogenesis of aortic atherosderosis, l Pathol BacterioI 1948;60:57-61. 2. More RM, Movat HZ, Hanst DM. Role of mura! fibrin thrombi of the aorta in genesis of arteriosclerotic plaques. Arch Pathol 1957;63:612-20. 3. Schwartz CJ, Valente AJ, Kelly JL, Sprague EA, Edwards EH. Thrombosis and the development of atherosclerosis: Rokitansky revisited. Semin Thromb Hemost 1988;14:189-95. 4. Woolf N, Carstairs KC. Infiltration and thrombosis in atherogenesis. A study using immunofluorescent techniques. Am ~ Pathol 1967;51:373-86. 5. Ip JH, Fuster V, Badiman L, Badiman I, Taubman MD, Chesbro JH. Syndromes of accelerated atherosclerosis; role of vascular injury and smooth muscle cell proliferation. ~Am Coil Cardiol 1990;15:1667-87. 6. Walsh PN, Schmaier AH. Platelet-coagulant protein interactions. In: Colman RW, Hirsh J, Marder VI, Salzman EW, eds. Hemostasis and thrombosis. Philadelphia: JB Lippincott, I987:689-709.
724
Journal of VASCULAR SURGERY
Alexander et al.
7. Awbry BJ, Hoak JC, Owen WG. Binding of human thrombin to cultured human endothelial cells. J Biol Chem 1979;254: 4092-5. 8. Gospodarowicz D, Brown KD, Birdwell CR, Zetter BR. Control of proliferation of human vascular endothelial cells. J Cell Biol 1978;77:774-88. 9. Weksler BB, Ley CW, Jaffee EA. Stimulation of endothelial cell prostacyclin production by thrombin, trypsin, and the ionophore A23187. J Clin Invest 1978;62:923-30. 10. Graham DJ, Alexander JJ. The effects ofthrombin on bovine aortic endothelial and smooth muscle cells. I VAse SURG i990;11:307-13. 11. Galdal KS, Evensen SA, Nilsen E. Thrombin-induced shape changes of cultured endothelialcells: metabolic and functional observations. Thromb Res 1983;32:57-66. 12. Laposata M, Dovnarsky DK, Shin HS. Thrombin-induced gap formation in confluent endothelial cell monolayers in vitro. Blood 1983;62:549-56. 13. Garcia JG, Siflinger-Birnboim A, Bizios R, DelVecchio P, Fenton JW II, Malik AB. Thrombin-induced increases in albumin transport across cultured endothelial monolayers. J Cell Physiol i986;128:96-104. 14. Alexander Jl, Miguel R, Graham D. Low-density lipoprotein uptake by an endothelial-smooth muscle cell bilayer, l VASC SURG 1991;13:444-5i. 15. Markwell MA. A new solid state reagent to iodinate proteins. Anal Biochem 1982; 125:427-32. 16. Cesarone CF, Bolognesi C, Santi L. Improved microfluorometric DNA determination in biological material using 33258 Hoechst. Anal Biochem 1979;100:188-97. 17. Andreoli SP, Baehner ILL, Bergstein JM. In vitro detection of endothelial cell damage using 2-deoxy-D-3H-glucose: comparison with chromium 51, 3H-leucine, 3H-adenine and lactate dehydrogenase, l Lab Clin Med 1985;106:25361. 18. Morrel EM, Holland JA, Pritchard ICA, Cotton CK, Stemerman MB. Endothelial perturbation and low density lipoprotein. Ann NY Acad Sci I987;516:412-7.
i9. Jorgensen L, Peckham MA, Rowsell HC, Mustard IF. Deposition of formed elements of blood on the intima and signs ofintimal injury in the aorta of rabbit, pig and man. Lab Invest I972;27:341-50. 20. Jorgensen L, Rowsell HC, Hovig T, Mustard IF. Resolution and organization of platelet-rich mural thrombi in carotid arteries of swine. Am J Pathol 1967;51:681-93. 21. Galdal KS, Evensen SA. Effects of divalent cations and various vasoactive and haemostatistically active agents on the integrity of monolayers of human endothelial cells. Thromb Res 1981;21:273-84. 22. DeMichele /VIA, Moon DG, Fenton JW II, Minnear FL. Thrombin's enzymatic activity increases permeability of endothelial cell monolayers. J Appl Physiol 1990;69:1599-606. 23. Graham DJ, Alexander JJ, Miguel R. Thrombin alters permeability and proliferation of co-cultured endothelial and smooth muscle cells. Surg Forum 1990;46:322-4. 24. Isaacs J, Savion N, Gospodarowicz D, Shuman MA. Effect of cell density on thrombin binding to a specific site on bovine vascular endothelial cells, l Cell Biol 1981;90:670-4. 25. Schwartz SM, Benditt EP. Aortic endothelial cell replications.. I. Effects of age and hypertension in the rat. Circ Res 1977;41:248-55. 26. Lin S-J, Jan K-M, Weinbaum S, Chien S. Transendothelial transport of low density lipoprotein in association with cell mitosis in rat aorta. Arteriosclerosis 1989;9:230-6. 27. Pearson ID, Gordon JL. Vascular endothelial and smooth muscle cells in culture selectively release adenine nucleotides. Nature 1979;281:384-6. 28. Zimmerman TS, Frereo I, Rothberger H. Blood coagulation and inflammatory response. Semin Haematol 1977;14:391404. 29. Hubbell IA, McIntire LV. Platelet active concentration profiles near growing thrombi: a mathematical consideration. Biophys I 1986;50:937-45. Submitted June 10, 1991; accepted Oct. 7, 1991.
DISCUSSION
Dr. John Sharefldn (Bethesda, Md.). This is an elegantly conducted study showing that alpha thrombin can, at suffidently high doses o f up to 50 units/ml, induce an increase in the uptake of L D L by SMCs even when these are beneath an initially intact endothelial layer. The fnding is of interest because the hypothesis has been around awhile. Kaddish and Folkman actually had suggested in 1979 that a fibrin clot from which active thrombin can be released by fibrinolysis could cause loss o f cell-to-ceU contact in the endothelial layer for several hours, and they proposed that persistence of fibrin clots could induce the early stages of atherosclerosis by disrupting the endothelial layer to allow lipid uptake. I have three questions for the authors about this study. What do you think the actual mechanism o f thrombin-
induced increases in SMC L D L uptake is? Thrombin is known to be able to both release platelet-derived growth factor from ECs, even in doses o f one tenth of a unit per ml sometimes, and certainly at 1 unit/ml it can induce the increase in gene expression for platelet-derived growth factor in endothelium. Platelet-derived growth factor in turn can increase the amount of 12~I labeled human L D L bound to and internalized by SMCs within approximately 24 hours, in part by increasing the number of LDL receptors by up to fourfold. D o you think this mechanism could be tested by either blocking thrombin-iuduced P D G F released with phosphodiesterase inhibitors such as amrinone or by adding dibuterol cyclic adenosine monophosphate or by adding platelet-derived growth factor antagonists or using monospecific antiplatelet-derived
Volume 15 Number 4 April 1992
growth factor activities? It seems this might be an intriguing way to see if that was one of the mechanisms for the increased uptake you observed. My second question is really what the authors think is the comparative importance of the roles of this sort of action of thrombin inducing loss of integrity in an EC layer versus other effects, because thrombin can also induce the expression of endothelial leukocyte and circulating monocyte adhesior~ molecules as inducers of atherogenic processes. The fact is that an atherosclerotic lesion's earliest form, such as the fatty streak, detectable for instance at sites of flow disturbance in hypercholesterolemic rabitt, is there even in the total absence of gross thrombus formation; and do the authors think a model system where they added a monocyte line such as U937 or normal peripheral blood monocytes would be able to test for this other hypothesis about the initiation of atherogenesis simply because the endothelial layers in the fatty streaks are intact and show no c. ,dence of retraction? These lesions begin to occur with an intact endothelial layer, even when monocytes can be seen beneath them. I think my final and maybe most important question is the choice ,of the dose of 50 u n i t s - N I H units I p r e s u m e - o f alpha thrombin per milliliter. Do you think that such a dose, which had some lethality to ECs, is even compatible with the finding of a totally intact endothelial layer over early lesions such as fatty streaks? I have been unable in looking at the literature of thrombin effects on vascular wall cells to find use of a dose outside the range of 0.1 to 10 units a milliliter, and the reason for this choice is based on a number of studies. Both Schuman and Medieros in 1976, and Aronson and Finlas in 1977 measured thrombin activity in the highest possible conditions, which is the totally static clotting of fresh whole blood, and they did not get levels over 6 to 7 or 10 units a milliliter, and those only peaked at 11 minutes and rapidly went down thereafter. That was in the absence of dilution of it with regard to flow. Much lower doses than this can produce the effects like retraction of human cells as observed at 0.1 unit of thrombin. We can persistently induce PDGF R N A transcript increase in human ECs in the range 1 to 10 units and all sorts of other effects on smooth muscle like c-phos gene activation and phosphatidyl inositol 3,4,5triphosphate phosphate activation can occur at this lower range. In a similar way all these changes can occur at this much lower dose range. My question is do the authors have
Thrombin alters LDL uptake by an EC-SMC bilayer 725
evidence for the occurrence of this higher levei of thrombin activity being attained in bovine coagulation systems at all or do they think there is a large difference perhaps in sensitivity to thrombin between bovine and human vascular cells that required use of this higher dose of 50 units? Dr. J. Jeffrey Alexander. In regard to the first question concerning the production of growth factors by the endothelial layer, it is my impression that the endothelial layer has multiple roles in this bilayer system. We have previously looked at other injury models using this same culture system, and have found that the ECs do affect underlying SMCs, altering their rate of replication as well as their utilization of LDL. I refer to several studies primarily by Somer and Minick in which in vivo models demonstrate that lipid uptake by the arterial wall is greater in areas where endothelial regeneration has occurred, as compared to areas of denuded endothelium. This finding suggests that the EC regulates lipid uptake by the underlying SMCs and that metabolic dysfunction of the endothelium influences SMC function. Further study of this interaction is needed, including the role of growth factors in intercellular dynamics. Regarding the second question, the study presented today really has little to do with early fatty streak formation. This model would be more applicable to mature or fibrous plaque. Fatty streak formation would not be expected to involve thrombin deposition or interaction with the arterial endothelium. A need certainly exists to expand this bilayer culture model using monocytes. Lipid accumulation is a net result of both deposition and clearance. In addition, it is possible that monocytes may affect arterial cell lipid metabolism either directly or indirectly. The dose of 50 units of thrombin is a high one. Previous studies have shown significant effects on the EC at doses ranging from 1 to 10 units per milliliter. The concentration of thrombin occurring in thrombus is not well defined, and it was supposed that this could be higher than the 10 units/ml that has been measured in serum. We, therefore, arbitrarily chose a level of 50 units/m[ as our higher concentration, and compared these results with the lower concentration of 10 units/ml as suggested by our sponsor, Dr. Linda Graham. I suspect that results obtained using smaller concentrations ofthrombin would be similar, although perhaps not as pronounced as those that have been presented today.