Platelet-derived growth factor production by canine aortic grafts seeded with endothelial cells B r a m R. K a u f m a n , M D , Paul L. Fox, P h D , and Linda M. G r a h a m , M D ,
Cltweland, Ohio The advantages of an endothelial cell surface on a prosthetic graft may be compromised by endothelial cell production of mitogens for smooth muscle cells. To study plateletderived growth factor (PDGF) production by cells lining prosthetic grafts, 15 beagles underwent placement of 20 to 22 cm long, 8 mm inner diameter, expanded polytetrafluoroethylene thoracoabdominal aortic grafts, of which seven were seeded with autologous jugular vein endothelial cells, and eight were unseeded control grafts. Grafts and adjacent arteries were explanted after 4 weeks, divided into segments, and studied in organ culture. Platelet-derived growth factor production during a 72-hour incubation period was measured by radioreceptor assay. Midgraft segments of seeded grafts produced sig,aificantly more PDGF than control grafts, 43 -+ 8 pg/cm 2 and 22 -+ 5 pg/cm z, respectively (mean -+-+SEM), p < 0.05. Platelet-derived growth factor production correlated directly with endothelial cell coverage on graft segments as measured by scanning electron microscopy (r = 0.63, p = 0.01), and inversely with platelet deposition (r := - 0 . 4 8 , p = 0.07). For all dogs, PDGF production by the distal aorta was significantly greater than that by the proximal aorta, 89---6 and 17 + 4 pg/cm 2, respectively (p < 0.0001). These results suggest that endothelial cells on prosthetic vascular grafts are a significant source of PDGF. Furthermore, the higher PDGF production by the distal arteries may offer an explanation for the clinical finding of more severe intimal hyperplasia adjacent to the distal anastomosis. (J VASC 8URG 1992; 15:699-707.)
The development o f synthetic vascular bypass grafts has been a significant advance in the treatment o f arterial disease. Still, the patency rates for reconstruction with prosthetic material in the lower extremity are unsatisfactory. Because graft thrombosis is the most c o m m o n reason for failure, research has been directed at methods to decrease surface thrombogenicity and has led to the investigation o f From the Department of Surgery, Case Western ReserveUniversity and Department of Veterans Affairs Medical Center (Drs. Kaufman and.Graham), and the Department of Vascular Cell Biology and Atherosclerosis Research, Cleveland Cfinic Research Institute (Dr. Fox), Cleveland. Supported by grants from the National Institutes of HealthNational Heart, Lung, and Blood Institute (HL-41178 and HL-40352) and the Department of Veterans Affairs. Dr. Fox is an Established Investigator of the American Heart Association. Presented at the Forty-fifth Annual Meeting of the Society for Vascular Surgery,Research Forum II, Boston, Mass., June 4-5, 1991. Reprint requests: Linda M. Graham, MD, SurgeryService [[12W, Veterans Administration Medical Center, 10701 East Blvd., Cleveland, OH 44106. 24/1/33677
seeding endothelial cells (ECs) onto prosthetic grafts. 1,2 Endothelial cell surfacing o f grafts reduces thrombogenidty and early graft failure in animal models. 3,4 Clinical studies are currently in progress to determine the usefulness o f the technique in humans. ~,6 In addition to its ability to modulate t h e biochemical and functional properties o f the graft wall, EC seeding has emerged as a potentially useful process for the introduction o f transfected ECs and gene therapy. 7,8 One concern with the application o f EC seeding is that these cells, under certain conditions, produce mitogens for smooth musde cells (SMCs) that may contribute to the development ofintimal hyperplasia. Intimal hyperplasia is characterized by SMC proliferation and is a frequent cause o f late graft thrombosis. The idea that ECs may stimulate SMC proliferation in humans is supported by the observation that intimal hyperplasia occurs at the anastomoses, the only areas covered by endothelium. Further evidence o f such an association between SMC proliferation and ECs is found in experimental 699
700 Kaufinan, Fox, and Graham
animals where SMC proliferation on prosthetic grafts is observed only under endothellum.9 Endothelial cells cultured from the vessels of many species, including human, secrete growth factors for SMCs. 1° One of these growth factors has been identified as platelet-derived growth factor (PDGF).11 Platelet-derived growth factor stimulates SMC migration and proliferation in vitro, 1°'12 and it may be an important mediator in the EC/SMC interaction at sites of intimal hyperplasia. To test the hypothesis that ECs on prosthetic vascular grafts secrete growth factors, PDGF production by EC seeded and unseeded expanded polytetrafluoroethylene (ePTFE) vascular grafts was studied in a canine model. MATERIAL A N D M E T H O D S
Platelet-derived growth factor production by vascular grafts was studied in 15 adult female beagles, ages 58 and 318 weeks, weighing 8 to 13.5 kg. Unseeded thoracoabdominal grafts were implanted in eight dogs, whereas seven dogs received grafts seeded with autologous ECs removed from external jugular vein segments. Grafts were removed 4 weeks after implantation for assessment of PDGF production. Animal studies were performed by following a protocol approved by the institutional committee on animal use, and all animal care complied with the "Principles of Laboratory Animal Care" and the "Guide for the Care and Use of Laboratory Animals" (NIH Publication No. 80-23, revised 1985). On the day of graft implantation, anesthesia was induced with intravenous thiamylal sodium, 20 mg/kg, (Parke-Davis; Morris, N.J.) and maintained with 1% to 2% isofluorane and oxygen via an endotracheal tube with mechanical ventilation used during thoracotomy. Penicillin G benzathine (450,000 units) and penicillin G procaine (450,000 units) (Pfizer; New York, N.Y.) was administered to all animals subcutaneously. Hydration was maintained by infusion of lactated Ringer's solution at a rate of 10 ml/kg/hr for the duration of the operation. Both external jugular veins were removed and ECs harvested with use of sequential incubations in 0.05% trypsin (Sigma Chemical; St. Louis, Mo.) for 5 minutes and in 630 units/ml of collagenase A (Boehringer Mannheim; Indianapolis, Ind.) for 16 minutes. The cells were resuspended in 4 ml of Medium 199 (M-I99) (Sigma), and an aliquot of approximately 100 mm 3 was removed for cell identification with antibodies to factor VIII-related antigen (Chemicon International; E1 Segundo, Calif.)
Journal of VASCULAR SURGERY ~
and ~-actin (Enzo Diagnostics, Inc.; New York, N.Y.). All were found to be positive for factor VIII-related antigen and negative for SMC actin verifying cell identity as ECs. Expanded polytetrafluoroethylene grafts (donated by W. L. Gore & Assoc.; Elkton, Md.), 23 cm long, 8 mm inner diameter (I.D.), were filled with fibronectin (Collaborative Research; Bedford, Mass.) at a concentration of 150 pxg/ml for 60 minutes. Seeded grafts were incubated at 37 ° C for 30 minutes with medium containing ECs, whereas control grafts were treated identically but filled with cell free medium. All grafts were rotated after 15 minutes of incubation to assure uniform distribution of ECs. The grafts were trimmed to 20 to 22 cm and implanted in the thoracoabdominal position with an end-to-end proximal anastomosis to the proxi.r.[~! descending thoracic aorta and an end-to-side distal anastomosis to the infrarenal aorta as previously described. 1 Four weeks after implantation, the animals were again anesthetized and 150 units/kg of heparin sodium (Elkins-Sinn, Inc.; Cherry Hill, N.J.), and 90 ng/kg of papavarine (Eli Lilly and Co.; Indianapolis, Ind.) were administered intravenously. Bilateral common carotid arteries were harvested, rinsed with M-199, cut into approximately 4 cm 2 segments, and pinned onto disks of Sylgard 184 silicone elastomer (Dow Coming Corp.; Midland, Mich.) in a manner maintaining physiologic stretch. The aorta, graft, and iliac arteries were exposed through a thoracoabdominal incision. An inflow cannula was inserted into the stump of the right carotid artery, and outflow cannulas were placed in the distal femoral arteries. As the ascending aorta was cross-clamped and major visceral branches were ligated, the graft, aorta, iliac, and femoral arteries were flushed with 1 L of M-199. The arterial segments including the entire length of the aorta as well as the iliac and proximal femoral arteries were harvested on Sylgard disks as described previously and placed in organ culture. The graft was excised, opened, divided into segments, and pinned flat on Sylgard disks. Sixty minutes after removal, each segment was again rinsed thoroughly with M-199 and placed into tissue culture wells containing 4 ml of M-199 supplemented with 15 units/ml of heparin sodium (Elkins-Sinn Inc.), 0.05 mg/ml of gentamicin (Sigma), 0.15 mg/ml of EC growth supplement isolated from bovine hypothalami (Rockland; Gilbertsville, Pa.), 2.2 mg/ml of sodium bicarbonate (Sigma), and 10% controlled processed serum replacement-2 (Sigma). This serum substitute was used because it contains very little growth pro-
Volume 15 Number 4 April 1992
moting activity, including PDGF. Tissue was incubated at 37 ° C in 5% CO 2 for 72 hours. The medium was collected for measurement of PDGF and replaced with fresh medium containing 2 ~Ci/ml of [3H]leucine (New England Nuclear; Boston, Mass.). The tissue was incubated for an additional 24 hours to determine total secreted protein synthesis. For measurement of PDGF, the conditioned medium was centrifuged at 1000g for 5 minutes to remove ceils and debris. Platelet-derived growth factor in the supernatant was quantitated by a modification of the radioreceptor assay described by Fox and DiCorleto, 13 which measures competition with 12SI-labeled PDGF for binding to PDGF receptors on a fibroblast target cell. Platelet-derived growth factor was purified from human platelets as c"scribed by Raines and R o s s , 14 and 12SI-labeled PDGF was prepared as described by Heldin et al.~s In brief, conditioned medium obtained from organ culture specimens was concentrated sevenfold with use of Centriprep concentrators (Amicon; Cherry Hill, N.J.). Sparse, quiescent cultures of human foresldn fibroblasts (approximately 10,000 cells/cm2) were incubated for 1 hour with 0.1 ml of HEPES buffered DME/F-12 mixture (Sigma) containing 2 mg/ml bovine serum albumin and 0.2 ml of the concentrated conditioned medium. The cells were rinsed thoroughly and 0.3 ml of medium containing 12SI-labeled PDGF was added for an additional 1 hour, after which time the medium was aspirated and the cells rinsed. Cells were then solubilized with 1% Triton X-100 and cell bound radioactivity determined in a gamma radiation counter. Platelet-derived growth factor was quantitated by comparison to a standard curve prepared with use of known amounts of purified human PDGF. The amount of PDGF contributed by the medium with growth supplement and serum substitute was quantitated and subtracted from the amount secreted by the tissue samples. Since fibroblasts have receptor subunits recognizing both the A-chain and B-chain of PDGF, this assay detects all classes of PDGF dimers. ~6 The lower limit of sensitivity of the assay is approximately 20 to 20 pg. Measured PDGF was normalized by the area of the tissue in organ culture to yield pg PDGF per cm 2. Secreted protein synthesis over 24 hours was determined by [SH]leucine incorporation into trichloroacetic acid precipitable material as described by Fox and DiCorleto. ~3 The conditioned medium was centrifuged at 1000g for 10 minutes, and TCA precipitation was carried out on the supernatant four times. The final pellet was solubilized with N a O H and radioactivity measured in a liquid scintillation
PDGF production by aorticgrafts
701
counter. Secreted protein synthesis was expressed as the percentage of added [aH]leucine incorporated into trichloroacetic acid precipitable material. The arterial and graft samples were fixed in 2.5% glutaraldehyde (Sigma) for 24 hours and stored in 0.1 mol/L cacodylate. One half of each segment was processed for light microscopy by dehydration in graded alcohol solutions and embedment in Spurr Low-Viscosity Resin (Polysciences, Inc,; Warrington, Pa). After polymerization of the resin at 60 ° C for 72 hours, 0.75 ~m sections were cut with a Reichert-Jung Ultracut E Microtome and stained with toluidine blue. The other half of each segment was processed for scanning electron microscopy by dehydration in graded alcohols followed by final dehydration in hexamethyldisilazine (Polysciences). The dry specimens were coated with gold in a RMC-EIKO IB-3 ion coater and scanned with a Jeol JSM-840A (Jeol Ltd.; Tokyo, ~apan) electron microscope. Random fields were scanned at x 1000 magnification and percent endothelialization as well as platelet, erythrocyte, and leukocyte deposition were quantitated by an observer blinded with respect to graft treatment. Endothelial cell coverage was quantitated on 10 random fields by determining the presence or absence of endothelium at 10 grid intersections per field. Platelet deposition was estimated on a scale of 0 to 4, with 0 representing the absence of platelets and 4 representing heavy deposition. The numbers of erythrocytes and leukocytes were counted in the same fields. At least 10 random fields per graft segment were assessed for cell deposition. Platelet-derived growth factor production between segments in the same animal was compared by use of the paired t test. Results in seeded and control groups were compared with the unpaired t test. Endothelial cell coverage, platelet deposition, and PDGF production were compared by use of linear regression analysis. RESULTS
After 4 weeks of implantation, all seeded and control grafts were patent and without evidence of thrombus formation, infection, or technical error. Dogs in the seeded and control groups were not significantly different in age, at 121 m 30 (mean + SEM) and 154 m 31. weeks, respectively. Platelet-derived growth factor production by arterial or graft segments did not correlate with age. Platelet-derived growth factor production by segments of seeded and control grafts is shown in Table I. Although growth factor production by the
702
Journal of VASCULAR SURGERY !
K a u f m a n , Fox, a n d G r a h a m
Table I. PDGF production by canine arteries and aortic grafts
Carotid artery Proximal aorta Proximal anastomosis Midgraft Distal anastomosis Distal aorta
Seeded (n = 7)
Control (n = 8)
p value
26 -+ 6 20 -+ 6 24 ± 5
21 ___ 8 14 ___ 6 12 _+ 5
NS NS NS
43 -+ 8 15 +- 5 92 + 8*
22 _+ 5 38 _+ 18 86 --_ 8*
< 0.05 NS NS
P D G F expressed as pg/cm2/72 hours (mean ___ SEM), *p < 0.0001 compared with proximal aorta.
Table II. Protein synthesis by canine arteries and aortic grafts
Carotid artery Proximal aorta Proximal anastomosis Midgraft Distal anastomosis Distal aorta
Seeded (n = 7)
Control (n = 8)
# Value
129 + 26 93 -+ 26 80 -+ 16
277 + 167 110 -+ 46 77 _+ 26
NS NS NS
56 +- 12 88 m 23 225 _+ 54*
52 + 17" 85 -+ 36* 240 _+ 9 8 t
NS NS NS
Protein synthesis expresses as percentage o f applied [3H]leucine incorporated/cm2/72 hours (mean _+ SEM) x 1000. *p < 0.05 compared with proximal aorta. tp < 0.08 compared with proximal aorta.
anastomoses and by the arterial segments proximal to the graft were similar in dogs with seeded and control grafts, midgraft segments from the seeded group produced significantly more PDGF than the control group at 43 -+ 8 pg/cm 2 and 22 -+ 5 pg/cm 2, respectively, p < 0.05. It is interesting to note that aortic segments distal to both seeded and unseeded grafts produced significantly more PDGF than aortic segments proximal to the grafts, at 92 -+ 8 pg/cm 2 and 20 +_ 6 pg/cm 2, respectively, for dogs in the seeded group (p = 0.000'1) and 86 + 8 pg/cm 2 and 14 + 6 pg/cm 2, respectively, for dogs in the control group 60 = 0.0001).
Measurement of total secreted protein synthesis showed sustained viability Of all segments. Preliminavy studies documented near-linearity of protein synthesis by segments in organ culture for more than 96 hours. No statistical differences were observed in secreted protein synthesis by graft or arterial segments between seeded and control groups (Table II). It is of specific importance that the midgraft regions of seeded grafts and control grafts demonstrated comparable rates of protein synthesis. The secreted protein produced by the distal arterial segments was
approximately two times greater than that produced by the proximal aorta for both seeded and control dogs. Light microscopic evaluation documented significant differences between seeded and control grafts. Seeded midgraft segments showed a cellular monolayer on the surface that was characteristic of endothelium (Fig. 1, A). Subendothelial tissue was minimal and lacked cells. Control midgraft specimens, on the other hand, demonstrated a thin acellular coagulum ranging in thickness from less than 10 ~tm to approximately i00 Ixm with an occasional adherent leukocyte or erythrocyte (Fig. 1, B). No significant accumulation of tissue was noted at the anastomoses at 4 weeks, as expected at this early time after graft implantation. On scanning electron microscopy seeded g r ~ surfaces had the typical appearance of endotheliti~ with a confluent monolayer of nonoverlapping cells and minimal platelet or cellular deposition (Fig. 2, A). In contrast, control grafts were covered with a coagulum consisting of fibrin, platelets, and occasional leukocytes (Fig. 2, B). This was sufficiently thin to see the underlying nodal structure of the graft. Endothelium was identified only adjacent to the anastomoses of the control grafts except for a small area of endothelium covering approximately 3% of the midportion of one control graft. Endothelial cell coverage of the midgraft averaged 64% + 9% on seeded grafts and 0.4% _+ 0.4% of control grafts (p = 0.0001). Platelet-derived growth factor production correlated positively with the degree of endothelial coverage, r = 0.63,p = 0.01, as shown in Fig. 3. Platelet deposition, quantitated by scanning electron microscopy, was significantly less on seeded than on controlgrafts at l.3 ___0.3 and3.0 _ 0.3, respectively, p = 0.002. As expected, linear regression analysis of platelet deposition as a function of EC coverage demonstrated an inverse correlation, r = -0.78, p = 0.0006. Furthermore, platelet deposition showed a weak inverse correlation with PDGF production, r = -0.48, p = 0.07 (Fig. 4). No significant difference was observed between leukocyte deposition on control and seeded grafts, with control grafts demonstrating 8.8 --- 1.5 and seeded grafts 10.6 _+_1.7 cells per x 1000 magnification field. Also no correlation occurred between leukocyte deposition and PDGF production. DISCUSSION Despite advances in surgical approaches to arterial disease, intimal hyperplasia remains a barrier to consistent long-term patency of prosthetic vascular
Volume 15 Number 4 April 1992
PDGF production by aortic grafts 703
Fig. 1. Histologic sections of the midgraft regions 4 weeks after implantation (original magnification × 500, toluidine blue stain). A, An EC seeded graft demonstrates an EC monolayer with no subendotheliat cells. B, A control graft demonstrates an acellular coagulum on the luminal surface.
]Fig. 2. Scanning electron micrographs of the midgraft regions 4 weeks after implantation (original magnification × 1000). A, An EC seeded graft demonstrates a confluent monolayer of ECs with minimal platelet or cellular deposition. B, A control graft demonstrates an acellular coagulum of fibrin, platelets, and occasional leukocytes. grafts. :Endc~thelial cell surfaces on vascular grafts have the potential to decrease thrombogenicity, but the possibility that ECs on prosthetic grafts produce mitogens stimulating intimal hyperplasia must be considered. Endothelial cells have been linked to SMC proliferation and intimal hyperplasia in the past. Clowes et al. ~7 reported that ECs and SMCs migrate together from cut ends of adjacent artery onto ePTFE grafts in baboons, and that SMC proliferation and intimal thickness is greatest at the growing edge and at the anastomoses, sites of complete endothelial coverage. 9 In porcine aortas in organ culture: for 7 days, Koo and Gotlieb ~8observed a significant increase in the number of intimal SMCs
when endothelium was intact compared with no increase when it had been denuded. Furthermore, conditioned media from the endothelialized aortic specimens stimulated proliferation of SMCs in the denuded specimens. Supporting the role of ECs in the SMC proliferation associated with intimal hyperplasia is the observation that ECs produce several mitogens for SMCs. One of these mitogens, PDGF, is both mitogenic and chemotactic for SMCs. n,12 Smooth muscle cells coculmred with bovine ECs have an increased rate of replication when compared with SMCs cultured alone, and P D G F has been shown to be responsible for part of this mitogenic activity, n
Journal o f VASCULAR SURGERY !
704 Kaufman, Fox, and Graham
8O
y = 0 . 3 6 x + 21 r = 0.63
_f /" /
J
/
40
e~ Q.
20
/
/
.
0
,
.
,
•
20
,
,
,
,
,
40
•
,
•
,
•
60
,
•
i
•
80
100
Midgraft Endothelial Cell C o v e r a g e (%)
Fig. 3. Linear regression analysis of PDGF production and midgraf~ EC coverage (%) from seven seeded and eight control grafts demonstrates a positive correlation, r = 0.63, p = 0.0l. Broken lines represent 95% confidence intervals.
80
I",..
y = - 9 x + 51 r = -0.48
60
E 40 Q. 1.1. ¢~ e~ 0-
20
,----~. 1
2
3
Mean Midgraft Platelet S c o r e
Fig. 4. Linear regression analysis of PDGF production and midgraft platelet deposition from seven seeded and eight control grafts demonstrates a weak inverse correlation, r = -0.48, p = 0.07. Broken lines represent 95% confidence intervals. Although ECs are known to secrete significant amounts of PDGF in vitro, the amount produced in vivo has not been measured directly. A significant difference between in vivo and in vitro production is suggested by the finding that PDGF B-chain m R N A is 10 to 100 times greater in human umbilical vein ECs and bovine aortic ECs grown in culture compared with ECs freshly scraped from these vessels. 19 This suggests that ECs may be activated in the unphysiologic conditions of tissue culture and produce greater quantities of PDGF in vitro than in vivo. In a similar manner, ECs on a prosthetic graft may be activated and may behave analogously to cells in culture with increased PDGF production. Accordingly, this study was designed to measure PDGF production by endothelialized graft surfaces. An organ culture system was selected for our
studies in an attempt to mimic EC function in vivo. Organ culture preserves anatomic and intercellular relationships of the arterial wall, thus minimizing perturbation of the EC. In preliminary studies we observed that PDGF production by ECs in tissue culture was greater than that by cells on organ culture specimens, 350 pg/cm 2 and 80 pg/cm 2, respectively (unpublished observations). PDGF secretion by an intact arterial wall is the result of inhibitory and stimulatory factors. Disruption of the normal cellular milieu and intercellular communication when ECs are isolated and placed in tissue culture may lead to the overall increase in PDGF production noted. Thus an organ culture system may simulate the in vivo situation with greater accuracy than tissue culture. However, an obvious drawback of the organ culture system is the difficulty of identifying the specific cell
Volume 15 Number 4 April 1992
types responsible for PDGF production. In addition, PDGF has been shown to be secreted primarily in an abluminal t~ashion,2° so PDGF assayed in the conditioned media may be an underestimate of true PDGF production. Our findings showed that endothelialized grafts produced significantly more PDGF than control grafts. On histologic studies the only apparent difference between seeded and control grafts was the EC lining. No cellular tissue was evident beneath the ECs of seeded grafts, suggesting that at 4 weeks the ECs lining the lumen were responsible for the difference in PDGF production between seeded and control grafes. Furthermore, the similarity of protein synthesis of control and seeded graft segments suggested a specific increase in PDGF production, r--'~herthan a general increase in number or metabolic activity of cells. Still, ECs as the source of PDGF was only inferred in this study, since the specific cell producing PDGF was not identified. Many cell types have been shown to be capable of producing PDGF, including modified SMCs and macrophages in addition to platelets and ECs. However, the negative correlation between PDGF production and platelet deposition .on the graft segments suggests that platelets were not the source of the mitogen in this study. Similarly, leukocyte deposition on grafts was not significantly different between seeded and control grafts and did not correlate with PDGF production, suggesting that they did not contribute significantly to the difference in measured PDGF. Because PDGF and the percent EC coverage shared a positive correlation, ECs were the likely source, but this remains to be confirmed with additional studies. A l t h o u g h the difference in PDGF production between proxhlaal aortic segments and seeded graft segments dicl not reach statistical significance in our study, EC coverage of graft segments was incomplete. With a confluent EC surface on all graft segements, a higher level of PDGF production would be anticipated. Therefore our results seem to be consistent with those of Zacharias et al.2~ and Golden et al. 22 Zacharias ct al.2~ found increased mitogenlc activity in perfusates of endothelialized grafts compared with native artery including thoracic aorta, carotid, and femoral arteries. However, the mitogens responsible were not identified in that study. Golden et al. 22demonstrated that graft intima expressed more mRNA for PDGF A chain than did native aortic intima, but did not measure PDGF production. A significant difference in PDGF production between proximal and distal aortic segments was noted for both seeded and control dogs. Protein
PDGF production by aortic grafts 705
synthesis in the distal arterial segments was only twofold greater than proximal arterial segments, whereas the PDGF production was fourfold greater, indicating that the increase in PDGF production was specific and not solely caused by increased cellularity or a generalized increase in cellular metabolism in the distal artery. However, this increase in PDGF production was not a result of graft implantation. Study of normal animals showed that PDGF production was significantly higher in the distal aorta than the proximal aorta (unpublished observations). The reason for increased PDGF production by the distal aorta is unclear, and the role of such factors as changing flow characteristics, vessel wall composition, or circulating mitogens is undefined. The higher basal level of PDGF production distally may play a role in the clinically observed pattern of atheroscierosis and increased severity of intimal hyperplasia adjacent to distal anastomoses of bypass grafts. In summary, PDGF production by EC seeded ePTFE grafts explanted from dogs at 4 weeks is greater than unseeded controls and correlates positively with EC coverage and inversely with platelet deposition. Taken together with previous findings of increased mitogenic activity from perfusates of grafts and increased mRNA for PDGF in the intima of grafts, 21,22 this suggests that endothelialized grafts may be the source of physiologically significant amounts of PDGF that may promote smooth muscle proliferation, which may in turn contribute to late graft failure. Further study to identify the cell types and understand the stimuli responsible for mitogen production may lead to improved long-term graft patency. REFERENCES
1. Graham LM, Vinter DW, Ford JW, Kahn Rift, Burkel WE, Stanley [IC. Endothelial cell seeding of prosthetic vascular grafts. Early experimental studies with cultured autologous canine endothelium. Arch Surg 1980;115:929-33. 2. Herring MB, Dilley R, Jersild RA, Boxer L, Gardner A, Glover I. Seeding arterial prostheses with vascular endothelium. The nature of the lining. Ann Surg 1979;190:84-90. 3. Hunter TJ, Schmidt SP, Sharp WV, Malindzak GS. Controlled flow studies in 4 mm endothelialized Dacron grafts. Trans Am Soc Artif Intern Organs 1983;29:177-82. 4. Stanley IC, Burkel WE, Ford JW, et al. Enhanced patency of small-diameter externally supported Dacron iliofemoral grafts seeded with endothelial cells. Surgery 1982;92:994-1005. 5. Herring MB~ Compton RS, LeGrand DR, Gardner AL, Madison DL, Glover JL. Endothelial seeding of polytetrafluoroethylene popliteal bypasses. A preliminary report. J VAsc SURG 1987;6:I14-8. 6. Ortenwall P, Wadenvik H, Kutti J, Risberg B. Endothelial cell seeding reduces thrombogenicity of Dacron graft in humans. VAsc SURG 1990;11:403-10.
706
lournal of VASCULAR SURGERY'
Kaufman, Fox, and Graham
7. Wilson JM, Birinyi LK, Salomon RN, Libby P, Callow AD, Mulligan RC. Implantation of vascular grafts lined with genetically modified endothelial cells. Science 1989;244: 1344-6. 8. Nabel EG, Plautz G, Boyce FM, Stanley JC, Nabel GJ. Recombinant gene expression in vivo within endothelial cells of the arterial wall. Science 1989;244:1342-4. 9. Clowes AW, Kirkman TR, Clowes MM. Mechanisms of arterial graft failure. II. Chronic endothelial and smooth muscle cell proliferation in healing polytettafluoroethylene prostheses. J VASCSURG 1986;3:877-84. 10. Ross R, Raines EW, Bowen-Pope DF. The biology of platelet-derived growth factor. Cell 1986;46:155-69. 11. DiCorleto PE, Bowen-Pope DF. Cultured endothelial cells produce a platelet-derived growth factor-like protein. Proc Natl Acad Sci USA 1983;80:1919-23. 12. Grotendorst GR, Sepp~i HEJ, IGeinman HK, Martin GR. Attachment of smooth muscle cells to collagen and their migration toward platelet-derived growth factor. Proc Natl Acad Sci USA 1981;78:3669-72. 13. Fox PL, DiCorleto PE. Regulation of production of a platelet-derived growth factor-like protein by cultured bovine aortic endothelial cells. J Cell Physiol 1984;121:298308. 14. Raines EW, Ross R. Platelet-derived growth factor. I. High yield purification and evidence for multiple forms. J Biol Chem 1982;257:5154-60. 15. Heldin C-H, Westermark B, Wasteson A. Platelet-derived
16. 17. 18. I9. 20. 2I. 22.
growth factor: purification and partial characterization. Proc Natl Acad Sci USA 1979;76:3722-6. Hart CE, Forstrom IW, Kelly JD, et al. Two classes of PDGF receptor recognize different isoforms of PDGF. Science 1988;240:1529-34. Clowes AW, Gown AM, Hanson SR, Reidy MA. Mechanisms of arterial graft failure, i. Role of cellular proliferation in early healing of PTFE prostheses. Am J Pathol 1985;118:43-54. Koo EW, Gotlieb AI. Endothelial stimulation of intimal cell proliferation in a porcine aortic organ culture. Am J Pathol 1989;134:497-503. Barrett TB, Gajdusek CM, Schwartz SM, McDougall JK, Benditt EP. Expression of the sis gene by endothelial cells in culture and in vivo. Proc Nat Acad Sci USA 1984;81:6772-4. Zerwes H-G, Risan W. Polarized secretion of a plateletderived growth factor-like chemotactic factor by endothelial cells in vitro. J Cell Biol 1987;105:2037-41. Zacharias RK, Kirkrnan TR, Kenagy RD, Bowen-Pope DF, Clowes AW. Growth factor production by polytetrafluoroethylene vascular grafts. J VAsc SURG 1988;7:606-10. Golden MA, Au YPT, Kirkman TR, et al. Platelet-deriv~ growth factor activity and mRNA expression in healing vascular grafts in baboons. Association in vivo of plateletderived growth factor mRNA and protein with cellular proliferation. J Clin Invest 1991;87:406-14.
Submitted June 10, 1991; accepted Sept. 9, 1991.
DISCUSSION
Dr. James C. Stanley (Ann Arbor, Mich.). Abnormal cellular proliferative events appear to be important factors in a variety of arterial lesions, including anastomotic neointimal hyperplasia, excessive inner capsule thickness in small caliber synthetic grafts, restenosis after endarterectomy, and hyperplastic cellular plaque in accelerated atherosclerosis. The control of mitogenic activity in these lesions is an area of intense investigation. This particular study presents preliminary data related to the production of the specific mitogen PDGF in the central portion of long thoracoabdominal grafts seeded with autologous endothelium. Although the data are rather convincing, the question remains as to the physiologic or pathologic relevance ascribed to this unusual release of PDGF. One could conclude that the presence of ECs in association with synthetic grafts are dysfunctional. That conclusion is not only an oversimplification, but, perhaps, erroneous. Dr. Graham, when working in our laboratories at the University of Michigan, was a coinvestigator in a study that documented reductions in inner capsule thicknesses of grafts that had developed luminal linings of endothelium after autologous EC seeding. Production of PDGF by endothelium, as might be inferred from the current study's results, appeared to have no deleterious
effect, when compared to the thicker irmer capsule of unseeded grafts that did not develop luminal EC linings. Similarly, Dr. Bush has shown that EC seeding and resurfacing of intimectomized carotid vessels reduced subsequent fibrodysplastic responses. Nevertheless, the current observations may be relevant and, are in fact consistent with the observations that ECs on a synthetic vascular graft, even when confluent, remain perturbed and possibly dysfunctional. Perhaps the authors could comment on three aspects of their study. First, differences did not exist between the proximal and distal graft regarding PDGF release, and one might question if this does not lessen the importance of your observations regarding PDGF release from the central graft. Second, the assays were undertaken after static incubations of whole graft segments, and quite different results may have occurred if a flow-through closed-circuit system been used to assess luminal production or release of PDGF. Third, the effect of PDGF may be autocrine, such has been observed with A-chain effects on vascular smooth muscle or paracrine. Although you did not probe for the A and B chains in this particular study, perhaps you could speculate on the potential targets of the released PDGF. One finding deserves note regarding what may be a
Volume 15 Number 4 April 1992
seminal observation in this experiment. That relates to documented increases in PDGF release downstream within the aorta. This may contribute to a number of accelerated cellular-proliferative events seen in the more distal portions of the circulation and, at the very least, must be considered when assessing increased PDGF activity beyond an implanted vascular device. In the past, increased downstream PDGF activity has usually been related to platelet activation at the blood-device interface causing release phenomena, rather than different phenotypic activity within the more distal arterial tree. Dr. Brain Kaufman. The contribution of the anastomotic segments to PDGF production is of interest in view of the development of intimal hyperplasia at the anastomoses of grafts implanted in humans. In this study no difference was found in PDGF production by the proximal and distal anastomotic segments that consisted of the graft anastomosis 'with a small portion of the adjacent artery. The _ltogen production by these segments was less than that of the midportion of the graft. This may be due to injury to ECs as a result of drying and surgical instrumentation at the time of implantation or manipulation at the time of removal. It is interesting to note that preliminary results of long-term studies suggest that PDGF production at the anastomoses increases with time and eventually surpasses that of the midgraft. Organ ctflture methodology was selected for this study to allow measurement of protein production while preserving the anatomic and intercellular relationships of the normal arterial wall. One of the limitations of an organ culture system is that it is a static system and the effects of flowing blood on mitogen production is not evaluated. The effect of flow on PDGF m R N A expression by EC has been studied by others, and preliminary studies on the effect of flow on PDGF production are in progress.
PDGF production by aorticgrafts 707
The ability of ECs to produce PDGF is well documented, but other cells including leukocytes and SMCs, under certain circumstances, also produce PDGF. The actual source of PDGF was not identified in this study, although the correlation between EC coverage and PDGF production suggests that ECs are responsible for at least part of the PDGF production. Endothelial cells produce but do not respond to PDGF, whereas SMCs respond to PDGF as a chemotactic and mitogenic agent. Under certain circumstances SMCs are also capable of producing PDGF, raising the possibility that once SMCs migrate onto a prosthetic graft they might produce and respond to PDGF in an autocrine fashion. The finding of increased PDGF by the distal aorta compared with the proximal aorta was a surprising and interesting finding. The elements responsible for this difference were not identified in our study but may include differences in the cellular composition of the arterial wall as well as changes in mechanical and hemodynamic characteristics such as pulse pressure, shear stress, and wall tension. Undoubtedly, it is a multifactorial process. Dr. Michael Watldns. Did you use in your organ culture explants of the grafts and of the native vessels in the presence of growth factor? Dr. Kaufman. Yes, EC growth supplement (ECGS) was used. Dr. Watlfins. Could you speculate on what the behavior of the cells might have been in organ culture in the absence of growth factor? That is very important because the presence or absence of growth factor can effect the expression or release of PDGF or a number of things. Dr. Kaufman. Preliminary experiments with and without ECGS in tissue culture and organ culture were performed. No significant difference in PDGF production in the presence or absence of ECGS was found.
Cltweland, Ohio The advantages of an endothelial cell surface on a prosthetic graft may be compromised by endothelial cell production of mitogens for smooth muscle cells. To study plateletderived growth factor (PDGF) production by cells lining prosthetic grafts, 15 beagles underwent placement of 20 to 22 cm long, 8 mm inner diameter, expanded polytetrafluoroethylene thoracoabdominal aortic grafts, of which seven were seeded with autologous jugular vein endothelial cells, and eight were unseeded control grafts. Grafts and adjacent arteries were explanted after 4 weeks, divided into segments, and studied in organ culture. Platelet-derived growth factor production during a 72-hour incubation period was measured by radioreceptor assay. Midgraft segments of seeded grafts produced sig,aificantly more PDGF than control grafts, 43 -+ 8 pg/cm 2 and 22 -+ 5 pg/cm z, respectively (mean -+-+SEM), p < 0.05. Platelet-derived growth factor production correlated directly with endothelial cell coverage on graft segments as measured by scanning electron microscopy (r = 0.63, p = 0.01), and inversely with platelet deposition (r := - 0 . 4 8 , p = 0.07). For all dogs, PDGF production by the distal aorta was significantly greater than that by the proximal aorta, 89---6 and 17 + 4 pg/cm 2, respectively (p < 0.0001). These results suggest that endothelial cells on prosthetic vascular grafts are a significant source of PDGF. Furthermore, the higher PDGF production by the distal arteries may offer an explanation for the clinical finding of more severe intimal hyperplasia adjacent to the distal anastomosis. (J VASC 8URG 1992; 15:699-707.)
The development o f synthetic vascular bypass grafts has been a significant advance in the treatment o f arterial disease. Still, the patency rates for reconstruction with prosthetic material in the lower extremity are unsatisfactory. Because graft thrombosis is the most c o m m o n reason for failure, research has been directed at methods to decrease surface thrombogenicity and has led to the investigation o f From the Department of Surgery, Case Western ReserveUniversity and Department of Veterans Affairs Medical Center (Drs. Kaufman and.Graham), and the Department of Vascular Cell Biology and Atherosclerosis Research, Cleveland Cfinic Research Institute (Dr. Fox), Cleveland. Supported by grants from the National Institutes of HealthNational Heart, Lung, and Blood Institute (HL-41178 and HL-40352) and the Department of Veterans Affairs. Dr. Fox is an Established Investigator of the American Heart Association. Presented at the Forty-fifth Annual Meeting of the Society for Vascular Surgery,Research Forum II, Boston, Mass., June 4-5, 1991. Reprint requests: Linda M. Graham, MD, SurgeryService [[12W, Veterans Administration Medical Center, 10701 East Blvd., Cleveland, OH 44106. 24/1/33677
seeding endothelial cells (ECs) onto prosthetic grafts. 1,2 Endothelial cell surfacing o f grafts reduces thrombogenidty and early graft failure in animal models. 3,4 Clinical studies are currently in progress to determine the usefulness o f the technique in humans. ~,6 In addition to its ability to modulate t h e biochemical and functional properties o f the graft wall, EC seeding has emerged as a potentially useful process for the introduction o f transfected ECs and gene therapy. 7,8 One concern with the application o f EC seeding is that these cells, under certain conditions, produce mitogens for smooth musde cells (SMCs) that may contribute to the development ofintimal hyperplasia. Intimal hyperplasia is characterized by SMC proliferation and is a frequent cause o f late graft thrombosis. The idea that ECs may stimulate SMC proliferation in humans is supported by the observation that intimal hyperplasia occurs at the anastomoses, the only areas covered by endothelium. Further evidence o f such an association between SMC proliferation and ECs is found in experimental 699
700 Kaufinan, Fox, and Graham
animals where SMC proliferation on prosthetic grafts is observed only under endothellum.9 Endothelial cells cultured from the vessels of many species, including human, secrete growth factors for SMCs. 1° One of these growth factors has been identified as platelet-derived growth factor (PDGF).11 Platelet-derived growth factor stimulates SMC migration and proliferation in vitro, 1°'12 and it may be an important mediator in the EC/SMC interaction at sites of intimal hyperplasia. To test the hypothesis that ECs on prosthetic vascular grafts secrete growth factors, PDGF production by EC seeded and unseeded expanded polytetrafluoroethylene (ePTFE) vascular grafts was studied in a canine model. MATERIAL A N D M E T H O D S
Platelet-derived growth factor production by vascular grafts was studied in 15 adult female beagles, ages 58 and 318 weeks, weighing 8 to 13.5 kg. Unseeded thoracoabdominal grafts were implanted in eight dogs, whereas seven dogs received grafts seeded with autologous ECs removed from external jugular vein segments. Grafts were removed 4 weeks after implantation for assessment of PDGF production. Animal studies were performed by following a protocol approved by the institutional committee on animal use, and all animal care complied with the "Principles of Laboratory Animal Care" and the "Guide for the Care and Use of Laboratory Animals" (NIH Publication No. 80-23, revised 1985). On the day of graft implantation, anesthesia was induced with intravenous thiamylal sodium, 20 mg/kg, (Parke-Davis; Morris, N.J.) and maintained with 1% to 2% isofluorane and oxygen via an endotracheal tube with mechanical ventilation used during thoracotomy. Penicillin G benzathine (450,000 units) and penicillin G procaine (450,000 units) (Pfizer; New York, N.Y.) was administered to all animals subcutaneously. Hydration was maintained by infusion of lactated Ringer's solution at a rate of 10 ml/kg/hr for the duration of the operation. Both external jugular veins were removed and ECs harvested with use of sequential incubations in 0.05% trypsin (Sigma Chemical; St. Louis, Mo.) for 5 minutes and in 630 units/ml of collagenase A (Boehringer Mannheim; Indianapolis, Ind.) for 16 minutes. The cells were resuspended in 4 ml of Medium 199 (M-I99) (Sigma), and an aliquot of approximately 100 mm 3 was removed for cell identification with antibodies to factor VIII-related antigen (Chemicon International; E1 Segundo, Calif.)
Journal of VASCULAR SURGERY ~
and ~-actin (Enzo Diagnostics, Inc.; New York, N.Y.). All were found to be positive for factor VIII-related antigen and negative for SMC actin verifying cell identity as ECs. Expanded polytetrafluoroethylene grafts (donated by W. L. Gore & Assoc.; Elkton, Md.), 23 cm long, 8 mm inner diameter (I.D.), were filled with fibronectin (Collaborative Research; Bedford, Mass.) at a concentration of 150 pxg/ml for 60 minutes. Seeded grafts were incubated at 37 ° C for 30 minutes with medium containing ECs, whereas control grafts were treated identically but filled with cell free medium. All grafts were rotated after 15 minutes of incubation to assure uniform distribution of ECs. The grafts were trimmed to 20 to 22 cm and implanted in the thoracoabdominal position with an end-to-end proximal anastomosis to the proxi.r.[~! descending thoracic aorta and an end-to-side distal anastomosis to the infrarenal aorta as previously described. 1 Four weeks after implantation, the animals were again anesthetized and 150 units/kg of heparin sodium (Elkins-Sinn, Inc.; Cherry Hill, N.J.), and 90 ng/kg of papavarine (Eli Lilly and Co.; Indianapolis, Ind.) were administered intravenously. Bilateral common carotid arteries were harvested, rinsed with M-199, cut into approximately 4 cm 2 segments, and pinned onto disks of Sylgard 184 silicone elastomer (Dow Coming Corp.; Midland, Mich.) in a manner maintaining physiologic stretch. The aorta, graft, and iliac arteries were exposed through a thoracoabdominal incision. An inflow cannula was inserted into the stump of the right carotid artery, and outflow cannulas were placed in the distal femoral arteries. As the ascending aorta was cross-clamped and major visceral branches were ligated, the graft, aorta, iliac, and femoral arteries were flushed with 1 L of M-199. The arterial segments including the entire length of the aorta as well as the iliac and proximal femoral arteries were harvested on Sylgard disks as described previously and placed in organ culture. The graft was excised, opened, divided into segments, and pinned flat on Sylgard disks. Sixty minutes after removal, each segment was again rinsed thoroughly with M-199 and placed into tissue culture wells containing 4 ml of M-199 supplemented with 15 units/ml of heparin sodium (Elkins-Sinn Inc.), 0.05 mg/ml of gentamicin (Sigma), 0.15 mg/ml of EC growth supplement isolated from bovine hypothalami (Rockland; Gilbertsville, Pa.), 2.2 mg/ml of sodium bicarbonate (Sigma), and 10% controlled processed serum replacement-2 (Sigma). This serum substitute was used because it contains very little growth pro-
Volume 15 Number 4 April 1992
moting activity, including PDGF. Tissue was incubated at 37 ° C in 5% CO 2 for 72 hours. The medium was collected for measurement of PDGF and replaced with fresh medium containing 2 ~Ci/ml of [3H]leucine (New England Nuclear; Boston, Mass.). The tissue was incubated for an additional 24 hours to determine total secreted protein synthesis. For measurement of PDGF, the conditioned medium was centrifuged at 1000g for 5 minutes to remove ceils and debris. Platelet-derived growth factor in the supernatant was quantitated by a modification of the radioreceptor assay described by Fox and DiCorleto, 13 which measures competition with 12SI-labeled PDGF for binding to PDGF receptors on a fibroblast target cell. Platelet-derived growth factor was purified from human platelets as c"scribed by Raines and R o s s , 14 and 12SI-labeled PDGF was prepared as described by Heldin et al.~s In brief, conditioned medium obtained from organ culture specimens was concentrated sevenfold with use of Centriprep concentrators (Amicon; Cherry Hill, N.J.). Sparse, quiescent cultures of human foresldn fibroblasts (approximately 10,000 cells/cm2) were incubated for 1 hour with 0.1 ml of HEPES buffered DME/F-12 mixture (Sigma) containing 2 mg/ml bovine serum albumin and 0.2 ml of the concentrated conditioned medium. The cells were rinsed thoroughly and 0.3 ml of medium containing 12SI-labeled PDGF was added for an additional 1 hour, after which time the medium was aspirated and the cells rinsed. Cells were then solubilized with 1% Triton X-100 and cell bound radioactivity determined in a gamma radiation counter. Platelet-derived growth factor was quantitated by comparison to a standard curve prepared with use of known amounts of purified human PDGF. The amount of PDGF contributed by the medium with growth supplement and serum substitute was quantitated and subtracted from the amount secreted by the tissue samples. Since fibroblasts have receptor subunits recognizing both the A-chain and B-chain of PDGF, this assay detects all classes of PDGF dimers. ~6 The lower limit of sensitivity of the assay is approximately 20 to 20 pg. Measured PDGF was normalized by the area of the tissue in organ culture to yield pg PDGF per cm 2. Secreted protein synthesis over 24 hours was determined by [SH]leucine incorporation into trichloroacetic acid precipitable material as described by Fox and DiCorleto. ~3 The conditioned medium was centrifuged at 1000g for 10 minutes, and TCA precipitation was carried out on the supernatant four times. The final pellet was solubilized with N a O H and radioactivity measured in a liquid scintillation
PDGF production by aorticgrafts
701
counter. Secreted protein synthesis was expressed as the percentage of added [aH]leucine incorporated into trichloroacetic acid precipitable material. The arterial and graft samples were fixed in 2.5% glutaraldehyde (Sigma) for 24 hours and stored in 0.1 mol/L cacodylate. One half of each segment was processed for light microscopy by dehydration in graded alcohol solutions and embedment in Spurr Low-Viscosity Resin (Polysciences, Inc,; Warrington, Pa). After polymerization of the resin at 60 ° C for 72 hours, 0.75 ~m sections were cut with a Reichert-Jung Ultracut E Microtome and stained with toluidine blue. The other half of each segment was processed for scanning electron microscopy by dehydration in graded alcohols followed by final dehydration in hexamethyldisilazine (Polysciences). The dry specimens were coated with gold in a RMC-EIKO IB-3 ion coater and scanned with a Jeol JSM-840A (Jeol Ltd.; Tokyo, ~apan) electron microscope. Random fields were scanned at x 1000 magnification and percent endothelialization as well as platelet, erythrocyte, and leukocyte deposition were quantitated by an observer blinded with respect to graft treatment. Endothelial cell coverage was quantitated on 10 random fields by determining the presence or absence of endothelium at 10 grid intersections per field. Platelet deposition was estimated on a scale of 0 to 4, with 0 representing the absence of platelets and 4 representing heavy deposition. The numbers of erythrocytes and leukocytes were counted in the same fields. At least 10 random fields per graft segment were assessed for cell deposition. Platelet-derived growth factor production between segments in the same animal was compared by use of the paired t test. Results in seeded and control groups were compared with the unpaired t test. Endothelial cell coverage, platelet deposition, and PDGF production were compared by use of linear regression analysis. RESULTS
After 4 weeks of implantation, all seeded and control grafts were patent and without evidence of thrombus formation, infection, or technical error. Dogs in the seeded and control groups were not significantly different in age, at 121 m 30 (mean + SEM) and 154 m 31. weeks, respectively. Platelet-derived growth factor production by arterial or graft segments did not correlate with age. Platelet-derived growth factor production by segments of seeded and control grafts is shown in Table I. Although growth factor production by the
702
Journal of VASCULAR SURGERY !
K a u f m a n , Fox, a n d G r a h a m
Table I. PDGF production by canine arteries and aortic grafts
Carotid artery Proximal aorta Proximal anastomosis Midgraft Distal anastomosis Distal aorta
Seeded (n = 7)
Control (n = 8)
p value
26 -+ 6 20 -+ 6 24 ± 5
21 ___ 8 14 ___ 6 12 _+ 5
NS NS NS
43 -+ 8 15 +- 5 92 + 8*
22 _+ 5 38 _+ 18 86 --_ 8*
< 0.05 NS NS
P D G F expressed as pg/cm2/72 hours (mean ___ SEM), *p < 0.0001 compared with proximal aorta.
Table II. Protein synthesis by canine arteries and aortic grafts
Carotid artery Proximal aorta Proximal anastomosis Midgraft Distal anastomosis Distal aorta
Seeded (n = 7)
Control (n = 8)
# Value
129 + 26 93 -+ 26 80 -+ 16
277 + 167 110 -+ 46 77 _+ 26
NS NS NS
56 +- 12 88 m 23 225 _+ 54*
52 + 17" 85 -+ 36* 240 _+ 9 8 t
NS NS NS
Protein synthesis expresses as percentage o f applied [3H]leucine incorporated/cm2/72 hours (mean _+ SEM) x 1000. *p < 0.05 compared with proximal aorta. tp < 0.08 compared with proximal aorta.
anastomoses and by the arterial segments proximal to the graft were similar in dogs with seeded and control grafts, midgraft segments from the seeded group produced significantly more PDGF than the control group at 43 -+ 8 pg/cm 2 and 22 -+ 5 pg/cm 2, respectively, p < 0.05. It is interesting to note that aortic segments distal to both seeded and unseeded grafts produced significantly more PDGF than aortic segments proximal to the grafts, at 92 -+ 8 pg/cm 2 and 20 +_ 6 pg/cm 2, respectively, for dogs in the seeded group (p = 0.000'1) and 86 + 8 pg/cm 2 and 14 + 6 pg/cm 2, respectively, for dogs in the control group 60 = 0.0001).
Measurement of total secreted protein synthesis showed sustained viability Of all segments. Preliminavy studies documented near-linearity of protein synthesis by segments in organ culture for more than 96 hours. No statistical differences were observed in secreted protein synthesis by graft or arterial segments between seeded and control groups (Table II). It is of specific importance that the midgraft regions of seeded grafts and control grafts demonstrated comparable rates of protein synthesis. The secreted protein produced by the distal arterial segments was
approximately two times greater than that produced by the proximal aorta for both seeded and control dogs. Light microscopic evaluation documented significant differences between seeded and control grafts. Seeded midgraft segments showed a cellular monolayer on the surface that was characteristic of endothelium (Fig. 1, A). Subendothelial tissue was minimal and lacked cells. Control midgraft specimens, on the other hand, demonstrated a thin acellular coagulum ranging in thickness from less than 10 ~tm to approximately i00 Ixm with an occasional adherent leukocyte or erythrocyte (Fig. 1, B). No significant accumulation of tissue was noted at the anastomoses at 4 weeks, as expected at this early time after graft implantation. On scanning electron microscopy seeded g r ~ surfaces had the typical appearance of endotheliti~ with a confluent monolayer of nonoverlapping cells and minimal platelet or cellular deposition (Fig. 2, A). In contrast, control grafts were covered with a coagulum consisting of fibrin, platelets, and occasional leukocytes (Fig. 2, B). This was sufficiently thin to see the underlying nodal structure of the graft. Endothelium was identified only adjacent to the anastomoses of the control grafts except for a small area of endothelium covering approximately 3% of the midportion of one control graft. Endothelial cell coverage of the midgraft averaged 64% + 9% on seeded grafts and 0.4% _+ 0.4% of control grafts (p = 0.0001). Platelet-derived growth factor production correlated positively with the degree of endothelial coverage, r = 0.63,p = 0.01, as shown in Fig. 3. Platelet deposition, quantitated by scanning electron microscopy, was significantly less on seeded than on controlgrafts at l.3 ___0.3 and3.0 _ 0.3, respectively, p = 0.002. As expected, linear regression analysis of platelet deposition as a function of EC coverage demonstrated an inverse correlation, r = -0.78, p = 0.0006. Furthermore, platelet deposition showed a weak inverse correlation with PDGF production, r = -0.48, p = 0.07 (Fig. 4). No significant difference was observed between leukocyte deposition on control and seeded grafts, with control grafts demonstrating 8.8 --- 1.5 and seeded grafts 10.6 _+_1.7 cells per x 1000 magnification field. Also no correlation occurred between leukocyte deposition and PDGF production. DISCUSSION Despite advances in surgical approaches to arterial disease, intimal hyperplasia remains a barrier to consistent long-term patency of prosthetic vascular
Volume 15 Number 4 April 1992
PDGF production by aortic grafts 703
Fig. 1. Histologic sections of the midgraft regions 4 weeks after implantation (original magnification × 500, toluidine blue stain). A, An EC seeded graft demonstrates an EC monolayer with no subendotheliat cells. B, A control graft demonstrates an acellular coagulum on the luminal surface.
]Fig. 2. Scanning electron micrographs of the midgraft regions 4 weeks after implantation (original magnification × 1000). A, An EC seeded graft demonstrates a confluent monolayer of ECs with minimal platelet or cellular deposition. B, A control graft demonstrates an acellular coagulum of fibrin, platelets, and occasional leukocytes. grafts. :Endc~thelial cell surfaces on vascular grafts have the potential to decrease thrombogenicity, but the possibility that ECs on prosthetic grafts produce mitogens stimulating intimal hyperplasia must be considered. Endothelial cells have been linked to SMC proliferation and intimal hyperplasia in the past. Clowes et al. ~7 reported that ECs and SMCs migrate together from cut ends of adjacent artery onto ePTFE grafts in baboons, and that SMC proliferation and intimal thickness is greatest at the growing edge and at the anastomoses, sites of complete endothelial coverage. 9 In porcine aortas in organ culture: for 7 days, Koo and Gotlieb ~8observed a significant increase in the number of intimal SMCs
when endothelium was intact compared with no increase when it had been denuded. Furthermore, conditioned media from the endothelialized aortic specimens stimulated proliferation of SMCs in the denuded specimens. Supporting the role of ECs in the SMC proliferation associated with intimal hyperplasia is the observation that ECs produce several mitogens for SMCs. One of these mitogens, PDGF, is both mitogenic and chemotactic for SMCs. n,12 Smooth muscle cells coculmred with bovine ECs have an increased rate of replication when compared with SMCs cultured alone, and P D G F has been shown to be responsible for part of this mitogenic activity, n
Journal o f VASCULAR SURGERY !
704 Kaufman, Fox, and Graham
8O
y = 0 . 3 6 x + 21 r = 0.63
_f /" /
J
/
40
e~ Q.
20
/
/
.
0
,
.
,
•
20
,
,
,
,
,
40
•
,
•
,
•
60
,
•
i
•
80
100
Midgraft Endothelial Cell C o v e r a g e (%)
Fig. 3. Linear regression analysis of PDGF production and midgraf~ EC coverage (%) from seven seeded and eight control grafts demonstrates a positive correlation, r = 0.63, p = 0.0l. Broken lines represent 95% confidence intervals.
80
I",..
y = - 9 x + 51 r = -0.48
60
E 40 Q. 1.1. ¢~ e~ 0-
20
,----~. 1
2
3
Mean Midgraft Platelet S c o r e
Fig. 4. Linear regression analysis of PDGF production and midgraft platelet deposition from seven seeded and eight control grafts demonstrates a weak inverse correlation, r = -0.48, p = 0.07. Broken lines represent 95% confidence intervals. Although ECs are known to secrete significant amounts of PDGF in vitro, the amount produced in vivo has not been measured directly. A significant difference between in vivo and in vitro production is suggested by the finding that PDGF B-chain m R N A is 10 to 100 times greater in human umbilical vein ECs and bovine aortic ECs grown in culture compared with ECs freshly scraped from these vessels. 19 This suggests that ECs may be activated in the unphysiologic conditions of tissue culture and produce greater quantities of PDGF in vitro than in vivo. In a similar manner, ECs on a prosthetic graft may be activated and may behave analogously to cells in culture with increased PDGF production. Accordingly, this study was designed to measure PDGF production by endothelialized graft surfaces. An organ culture system was selected for our
studies in an attempt to mimic EC function in vivo. Organ culture preserves anatomic and intercellular relationships of the arterial wall, thus minimizing perturbation of the EC. In preliminary studies we observed that PDGF production by ECs in tissue culture was greater than that by cells on organ culture specimens, 350 pg/cm 2 and 80 pg/cm 2, respectively (unpublished observations). PDGF secretion by an intact arterial wall is the result of inhibitory and stimulatory factors. Disruption of the normal cellular milieu and intercellular communication when ECs are isolated and placed in tissue culture may lead to the overall increase in PDGF production noted. Thus an organ culture system may simulate the in vivo situation with greater accuracy than tissue culture. However, an obvious drawback of the organ culture system is the difficulty of identifying the specific cell
Volume 15 Number 4 April 1992
types responsible for PDGF production. In addition, PDGF has been shown to be secreted primarily in an abluminal t~ashion,2° so PDGF assayed in the conditioned media may be an underestimate of true PDGF production. Our findings showed that endothelialized grafts produced significantly more PDGF than control grafts. On histologic studies the only apparent difference between seeded and control grafts was the EC lining. No cellular tissue was evident beneath the ECs of seeded grafts, suggesting that at 4 weeks the ECs lining the lumen were responsible for the difference in PDGF production between seeded and control grafes. Furthermore, the similarity of protein synthesis of control and seeded graft segments suggested a specific increase in PDGF production, r--'~herthan a general increase in number or metabolic activity of cells. Still, ECs as the source of PDGF was only inferred in this study, since the specific cell producing PDGF was not identified. Many cell types have been shown to be capable of producing PDGF, including modified SMCs and macrophages in addition to platelets and ECs. However, the negative correlation between PDGF production and platelet deposition .on the graft segments suggests that platelets were not the source of the mitogen in this study. Similarly, leukocyte deposition on grafts was not significantly different between seeded and control grafts and did not correlate with PDGF production, suggesting that they did not contribute significantly to the difference in measured PDGF. Because PDGF and the percent EC coverage shared a positive correlation, ECs were the likely source, but this remains to be confirmed with additional studies. A l t h o u g h the difference in PDGF production between proxhlaal aortic segments and seeded graft segments dicl not reach statistical significance in our study, EC coverage of graft segments was incomplete. With a confluent EC surface on all graft segements, a higher level of PDGF production would be anticipated. Therefore our results seem to be consistent with those of Zacharias et al.2~ and Golden et al. 22 Zacharias ct al.2~ found increased mitogenlc activity in perfusates of endothelialized grafts compared with native artery including thoracic aorta, carotid, and femoral arteries. However, the mitogens responsible were not identified in that study. Golden et al. 22demonstrated that graft intima expressed more mRNA for PDGF A chain than did native aortic intima, but did not measure PDGF production. A significant difference in PDGF production between proximal and distal aortic segments was noted for both seeded and control dogs. Protein
PDGF production by aortic grafts 705
synthesis in the distal arterial segments was only twofold greater than proximal arterial segments, whereas the PDGF production was fourfold greater, indicating that the increase in PDGF production was specific and not solely caused by increased cellularity or a generalized increase in cellular metabolism in the distal artery. However, this increase in PDGF production was not a result of graft implantation. Study of normal animals showed that PDGF production was significantly higher in the distal aorta than the proximal aorta (unpublished observations). The reason for increased PDGF production by the distal aorta is unclear, and the role of such factors as changing flow characteristics, vessel wall composition, or circulating mitogens is undefined. The higher basal level of PDGF production distally may play a role in the clinically observed pattern of atheroscierosis and increased severity of intimal hyperplasia adjacent to distal anastomoses of bypass grafts. In summary, PDGF production by EC seeded ePTFE grafts explanted from dogs at 4 weeks is greater than unseeded controls and correlates positively with EC coverage and inversely with platelet deposition. Taken together with previous findings of increased mitogenic activity from perfusates of grafts and increased mRNA for PDGF in the intima of grafts, 21,22 this suggests that endothelialized grafts may be the source of physiologically significant amounts of PDGF that may promote smooth muscle proliferation, which may in turn contribute to late graft failure. Further study to identify the cell types and understand the stimuli responsible for mitogen production may lead to improved long-term graft patency. REFERENCES
1. Graham LM, Vinter DW, Ford JW, Kahn Rift, Burkel WE, Stanley [IC. Endothelial cell seeding of prosthetic vascular grafts. Early experimental studies with cultured autologous canine endothelium. Arch Surg 1980;115:929-33. 2. Herring MB, Dilley R, Jersild RA, Boxer L, Gardner A, Glover I. Seeding arterial prostheses with vascular endothelium. The nature of the lining. Ann Surg 1979;190:84-90. 3. Hunter TJ, Schmidt SP, Sharp WV, Malindzak GS. Controlled flow studies in 4 mm endothelialized Dacron grafts. Trans Am Soc Artif Intern Organs 1983;29:177-82. 4. Stanley IC, Burkel WE, Ford JW, et al. Enhanced patency of small-diameter externally supported Dacron iliofemoral grafts seeded with endothelial cells. Surgery 1982;92:994-1005. 5. Herring MB~ Compton RS, LeGrand DR, Gardner AL, Madison DL, Glover JL. Endothelial seeding of polytetrafluoroethylene popliteal bypasses. A preliminary report. J VAsc SURG 1987;6:I14-8. 6. Ortenwall P, Wadenvik H, Kutti J, Risberg B. Endothelial cell seeding reduces thrombogenicity of Dacron graft in humans. VAsc SURG 1990;11:403-10.
706
lournal of VASCULAR SURGERY'
Kaufman, Fox, and Graham
7. Wilson JM, Birinyi LK, Salomon RN, Libby P, Callow AD, Mulligan RC. Implantation of vascular grafts lined with genetically modified endothelial cells. Science 1989;244: 1344-6. 8. Nabel EG, Plautz G, Boyce FM, Stanley JC, Nabel GJ. Recombinant gene expression in vivo within endothelial cells of the arterial wall. Science 1989;244:1342-4. 9. Clowes AW, Kirkman TR, Clowes MM. Mechanisms of arterial graft failure. II. Chronic endothelial and smooth muscle cell proliferation in healing polytettafluoroethylene prostheses. J VASCSURG 1986;3:877-84. 10. Ross R, Raines EW, Bowen-Pope DF. The biology of platelet-derived growth factor. Cell 1986;46:155-69. 11. DiCorleto PE, Bowen-Pope DF. Cultured endothelial cells produce a platelet-derived growth factor-like protein. Proc Natl Acad Sci USA 1983;80:1919-23. 12. Grotendorst GR, Sepp~i HEJ, IGeinman HK, Martin GR. Attachment of smooth muscle cells to collagen and their migration toward platelet-derived growth factor. Proc Natl Acad Sci USA 1981;78:3669-72. 13. Fox PL, DiCorleto PE. Regulation of production of a platelet-derived growth factor-like protein by cultured bovine aortic endothelial cells. J Cell Physiol 1984;121:298308. 14. Raines EW, Ross R. Platelet-derived growth factor. I. High yield purification and evidence for multiple forms. J Biol Chem 1982;257:5154-60. 15. Heldin C-H, Westermark B, Wasteson A. Platelet-derived
16. 17. 18. I9. 20. 2I. 22.
growth factor: purification and partial characterization. Proc Natl Acad Sci USA 1979;76:3722-6. Hart CE, Forstrom IW, Kelly JD, et al. Two classes of PDGF receptor recognize different isoforms of PDGF. Science 1988;240:1529-34. Clowes AW, Gown AM, Hanson SR, Reidy MA. Mechanisms of arterial graft failure, i. Role of cellular proliferation in early healing of PTFE prostheses. Am J Pathol 1985;118:43-54. Koo EW, Gotlieb AI. Endothelial stimulation of intimal cell proliferation in a porcine aortic organ culture. Am J Pathol 1989;134:497-503. Barrett TB, Gajdusek CM, Schwartz SM, McDougall JK, Benditt EP. Expression of the sis gene by endothelial cells in culture and in vivo. Proc Nat Acad Sci USA 1984;81:6772-4. Zerwes H-G, Risan W. Polarized secretion of a plateletderived growth factor-like chemotactic factor by endothelial cells in vitro. J Cell Biol 1987;105:2037-41. Zacharias RK, Kirkrnan TR, Kenagy RD, Bowen-Pope DF, Clowes AW. Growth factor production by polytetrafluoroethylene vascular grafts. J VAsc SURG 1988;7:606-10. Golden MA, Au YPT, Kirkman TR, et al. Platelet-deriv~ growth factor activity and mRNA expression in healing vascular grafts in baboons. Association in vivo of plateletderived growth factor mRNA and protein with cellular proliferation. J Clin Invest 1991;87:406-14.
Submitted June 10, 1991; accepted Sept. 9, 1991.
DISCUSSION
Dr. James C. Stanley (Ann Arbor, Mich.). Abnormal cellular proliferative events appear to be important factors in a variety of arterial lesions, including anastomotic neointimal hyperplasia, excessive inner capsule thickness in small caliber synthetic grafts, restenosis after endarterectomy, and hyperplastic cellular plaque in accelerated atherosclerosis. The control of mitogenic activity in these lesions is an area of intense investigation. This particular study presents preliminary data related to the production of the specific mitogen PDGF in the central portion of long thoracoabdominal grafts seeded with autologous endothelium. Although the data are rather convincing, the question remains as to the physiologic or pathologic relevance ascribed to this unusual release of PDGF. One could conclude that the presence of ECs in association with synthetic grafts are dysfunctional. That conclusion is not only an oversimplification, but, perhaps, erroneous. Dr. Graham, when working in our laboratories at the University of Michigan, was a coinvestigator in a study that documented reductions in inner capsule thicknesses of grafts that had developed luminal linings of endothelium after autologous EC seeding. Production of PDGF by endothelium, as might be inferred from the current study's results, appeared to have no deleterious
effect, when compared to the thicker irmer capsule of unseeded grafts that did not develop luminal EC linings. Similarly, Dr. Bush has shown that EC seeding and resurfacing of intimectomized carotid vessels reduced subsequent fibrodysplastic responses. Nevertheless, the current observations may be relevant and, are in fact consistent with the observations that ECs on a synthetic vascular graft, even when confluent, remain perturbed and possibly dysfunctional. Perhaps the authors could comment on three aspects of their study. First, differences did not exist between the proximal and distal graft regarding PDGF release, and one might question if this does not lessen the importance of your observations regarding PDGF release from the central graft. Second, the assays were undertaken after static incubations of whole graft segments, and quite different results may have occurred if a flow-through closed-circuit system been used to assess luminal production or release of PDGF. Third, the effect of PDGF may be autocrine, such has been observed with A-chain effects on vascular smooth muscle or paracrine. Although you did not probe for the A and B chains in this particular study, perhaps you could speculate on the potential targets of the released PDGF. One finding deserves note regarding what may be a
Volume 15 Number 4 April 1992
seminal observation in this experiment. That relates to documented increases in PDGF release downstream within the aorta. This may contribute to a number of accelerated cellular-proliferative events seen in the more distal portions of the circulation and, at the very least, must be considered when assessing increased PDGF activity beyond an implanted vascular device. In the past, increased downstream PDGF activity has usually been related to platelet activation at the blood-device interface causing release phenomena, rather than different phenotypic activity within the more distal arterial tree. Dr. Brain Kaufman. The contribution of the anastomotic segments to PDGF production is of interest in view of the development of intimal hyperplasia at the anastomoses of grafts implanted in humans. In this study no difference was found in PDGF production by the proximal and distal anastomotic segments that consisted of the graft anastomosis 'with a small portion of the adjacent artery. The _ltogen production by these segments was less than that of the midportion of the graft. This may be due to injury to ECs as a result of drying and surgical instrumentation at the time of implantation or manipulation at the time of removal. It is interesting to note that preliminary results of long-term studies suggest that PDGF production at the anastomoses increases with time and eventually surpasses that of the midgraft. Organ ctflture methodology was selected for this study to allow measurement of protein production while preserving the anatomic and intercellular relationships of the normal arterial wall. One of the limitations of an organ culture system is that it is a static system and the effects of flowing blood on mitogen production is not evaluated. The effect of flow on PDGF m R N A expression by EC has been studied by others, and preliminary studies on the effect of flow on PDGF production are in progress.
PDGF production by aorticgrafts 707
The ability of ECs to produce PDGF is well documented, but other cells including leukocytes and SMCs, under certain circumstances, also produce PDGF. The actual source of PDGF was not identified in this study, although the correlation between EC coverage and PDGF production suggests that ECs are responsible for at least part of the PDGF production. Endothelial cells produce but do not respond to PDGF, whereas SMCs respond to PDGF as a chemotactic and mitogenic agent. Under certain circumstances SMCs are also capable of producing PDGF, raising the possibility that once SMCs migrate onto a prosthetic graft they might produce and respond to PDGF in an autocrine fashion. The finding of increased PDGF by the distal aorta compared with the proximal aorta was a surprising and interesting finding. The elements responsible for this difference were not identified in our study but may include differences in the cellular composition of the arterial wall as well as changes in mechanical and hemodynamic characteristics such as pulse pressure, shear stress, and wall tension. Undoubtedly, it is a multifactorial process. Dr. Michael Watldns. Did you use in your organ culture explants of the grafts and of the native vessels in the presence of growth factor? Dr. Kaufman. Yes, EC growth supplement (ECGS) was used. Dr. Watlfins. Could you speculate on what the behavior of the cells might have been in organ culture in the absence of growth factor? That is very important because the presence or absence of growth factor can effect the expression or release of PDGF or a number of things. Dr. Kaufman. Preliminary experiments with and without ECGS in tissue culture and organ culture were performed. No significant difference in PDGF production in the presence or absence of ECGS was found.
