- Email: [email protected]
Tissue Factor Activity in Dialysis Access Grafts
Basic Science Tissue Factor Activity in Dialysis Access Grafts David Hasenstab,1 Thomas R. Kirkman,2 Alexander W. Clowes,1 and Ted R. Kohler,1,3 Seattle and Mountlake Terrace, Washington
Background: Intimal hyperplasia at the venous anastomosis of dialysis grafts causes early failure. We developed a sheep model of arteriovenous prosthetic grafts that fail rapidly due to intimal hyperplasia with histologic features nearly identical to human access grafts. A prominent feature of lesion development in this model is formation of luminal thrombus that becomes organized into stenosing lesions by macrophage and myofibroblast infiltration. To better understand this process, we examined the presence and activity of tissue factor (TF) in this system. This protein is the physiological initiator of coagulation in vivo and is known to contribute to development of intimal hyperplasia after vascular injury. Methods: Expanded polytetrafluorethylene (ePTFE) grafts were placed between the carotid artery and external jugular vein in sheep. Grafts were examined for luminal TF activity using a novel ex vivo assay. In a separate series of grafts, immunohistochemistry was used to localize smooth muscle cells, monocytes, and TF protein. Results: At 2 days, luminal TF activity already was higher in the venous and arterial end of the graft than in the adventitia. This high level of activity persisted at 8 weeks. TF activity was higher in the venous end of the grafts than in the arterial end at 2 and 8 weeks (40% and 47% increase, n ¼ 5, n ¼ 3, respectively, P < 0.05). Immunohistochemistry revealed TF protein localized in regions with or adjacent to fibrin accumulation and often in regions close to the lumen. Conclusions: This study further examines the development of intimal hyperplasia in ePTFE dialysis access grafts. In this model, TF levels on the luminal surface were increased throughout the arteriovenous grafts and the adjacent vessels as early as 2 days after engraftment and for as long as 8 weeks thereafter. The highest levels of activity were found in the venous end of the graft, where hyperplasia is most robust. Increased activity of TF is associated with luminal thrombus, which provides a scaffolding for development of intimal hyperplasia. These findings present an opportunity to develop strategies to limit TF activity within the graft. Further studies using agents delivered locally or incorporated into the graft matrix to block the luminal activity of TF are warranted.
INTRODUCTION 1
Department of Surgery, University of Washington, Seattle, WA.
2
BioDevelopment Associates LLC, Mountlake Terrace, WA.
3
Department of Surgery, Seattle VA Puget Sound Health Care System, Seattle, WA. Correspondence to: Ted R. Kohler, MS, MD, Department of Surgery, Seattle VA Puget Sound Healthcare System, 112v 1660 S. Columbian Way, Seattle, WA 98108, USA; E-mail: [email protected] Ann Vasc Surg 2016; 31: 179–185 http://dx.doi.org/10.1016/j.avsg.2015.10.008 Published by Elsevier Inc. Manuscript received: April 10, 2015; manuscript accepted: October 11, 2015; published online: November 17, 2015.
Vascular access failure is a major source of morbidity, hospitalization, and cost for patients receiving chronic hemodialysis. Prosthetic grafts, most commonly composed of expanded polytetrafluorethylene (ePTFE), are used for patients who cannot have an autologous fistula due to lack of adequate venous conduit.1 Approximately 60% of these grafts fail each year due to intimal hyperplasia, primarily at the venous end. Clinical specimens reveal intimal hyperplasia containing myofibroblasts, organized thrombus, abundant microvessels, and activated macrophages, 179
180 Hasenstab et al.
which produce cytokines that enhance smooth muscle cell migration and proliferation.2,3 We developed a sheep model of access failure in which significant narrowing occurs within weeks at the venous end.4 The histology of the narrowed segment is nearly identical to the human specimens described above. Thrombus is a prominent, early feature of these developing lesions. As in the human specimens, macrophages play an important role, invading the thrombus, which is then transformed into the mature lesion by influx of myofibroblasts and capillaries (Fig. 1). It is not known if thrombus also is a significant feature of early lesion development in clinical grafts because pathologic evaluation is generally restricted to mature, stenotic segments. However, the work of Tillman et al.5 has shown that the thrombotic response to vascular surgery in sheep is very similar to that of humans. The presence of luminal thrombus in our model suggests that tissue factor (TF) is active at the surface. This protein may be derived from activated platelets, smooth muscle cells, or macrophages.6 In addition to causing thrombosis, TF activates smooth muscle cell proliferation and migration, adding to development of hyperplastic lesions. To better understand this process in dialysis access failure, we used a novel assay to measure the surface activity of TF in our sheep model and performed immunohistochemistry to localize the protein.
MATERIALS AND METHODS Experimental Design and Operative Procedures Columbia crossbred sheep of either sex (Nebeker Ranch, Inc., Lancaster, CA), weighing between 40 and 60 kg were fasted overnight, given intramuscular ceftiofur (2 mg/kg), and premedicated with intramuscular xylazine (0.2 mg/kg), and intravenous ketamine (10e15 mg/kg). After intubation, anesthesia was maintained with inhalation isoflurane. The neck was prepared with chlorhexidine, 70% isopropyl alcohol, and a final application of iodine solution. The left carotid artery and the contralateral external jugular vein were surgically exposed. Animals were anticoagulated with intravenous heparin (150 U/ kg). Using standard vascular technique, commercially available, 6-mm internal diameter ePTFE grafts (30 m internodal distance, IMPRA, Inc., Tempe, AZ) were sutured end-to-side to blood vessels with continuous 6-0 polypropylene suture. Grafts were 11 to 15 cm in length. Fascia and skin were closed in layers with bioabsorbable suture. Buprenex was given (0.005 mg/kg) every 12 hr for 48 hr after
Annals of Vascular Surgery
Fig. 1. Microphotograph of a developing lesion in the sheep model at 2 weeks. The graft (G) has granulation tissue on its outer wall and thrombus on the lumen. The edge of the thrombus is being invaded by cells, which in our prior work were shown to be macrophages (arrow). These cells engulf red cell debris including hemoglobin, which becomes the hemosiderin seen in late lesions, and cell membranes, creating fat laden macrophages. Behind this advancing front of macrophages, myofibroblasts migrate into the fibrin scaffolding to form the neointima (original magnification 4, hematoxylin and eosin staining.).
surgery to relieve postoperative pain. Intramuscular ceftiofur (2 mg/kg) was continued twice a day on days 1 and 2 and daily on days 3 through 5 after surgery. Oral aspirin (81 mg/day) was administered one day before surgery and once per day throughout the study. Patency was checked daily for the first week and then 3 times a week by palpation for a thrill and by hand-held Doppler. Animals were maintained on a standard diet (Purina Mills Northwestern Pride DietÔ 32B4 pellets and hay cubes twice per day). Studies were approved by the Institutional Animal Care and Use Committee. Three animals were sacrificed at 2 days, 6 at 2 weeks, and 3 at 8 weeks. Immunohistochemistry and TF Detection Cross sections of representative segments of ePTFE grafts and adjacent vessels that were not used for TF assay were embedded in paraffin and cut in 5-m thickness for microscopic examination, including hematoxylin and eosin staining. Paraffin embedded, formalin-fixed tissues were stained as previously described using antibodies for the following: fibrin (1:20 dilution, antifibrin #350, American Diagnostica), smooth muscle cells (antialpha actin, Boehringer Mannheim) and macrophages (Ham56, Enzo Diagnostics, Farmingdale NY). TF was detected in histologic sections using a fluorescent-labeled
Volume 31, February 2016
ligand. The inactive form of factor VIIa (inactivated Phe-Phe-Arg chloromethylketone (FFR)-VIIai, Zymogenetics, Seattle, WA), which retains full binding capacity, was coupled to the fluorescent dye FluoReporter Oregon Green 488 (Molecular Probes, Eugene, OR) as described by the manufacturer using a dye molar ratio of 1:20 with 480 mg FFR-VIIai per reaction. The labeled factor VIIai was then used to detect TF. Sections were deparaffinized and rehydrated in phosphate-buffered saline. Fifty mL of a 1:50 dilution of fluorescent VIIai in phosphate-buffered saline with 1% bovine serum albumin was placed on the section at room temperature for 1 hr. The sections were washed 3 times in phosphate-buffered saline and mounted using Vectashield aqueous mounting medium (Vector Labs, Burlingame, CA). Sections were imaged using a fluorescent microscope (Leitz Dialux 20 EB) and digital camera (Diagnostic Instruments). TF Activity Assay We developed a novel assay that allowed us to measure TF activity on the surface of the grafts using a modified Boyden chamber. At the time of graft harvest, the animals were killed by an overdose of sodium pentobarbital. The neck was then rapidly opened, and the graft with a generous portion of adjacent artery and vein were excised en bloc. The graft and adjoining vessel were flushed with lactated Ringer’s lactate and longitudinally opened. Loose connective and fibrous tissue were removed. The graft was cut in half at the midpoint and separated from the adjoining artery or vein. Specimens were kept submerged in ice-cold Ringers’ solution after removal and during all subsequent manipulations until assayed. TF activity was measured serially along the length of the graft, on the adventitia, and on the adjacent vessels using an assay based on factor Xa cleavage of a chromogenic substrate.7 The tissue was placed in a modified Boyden chamber (Microchemotaxis chamber; Neuro Probe Inc., Gaithersburg, MD) with the silicone gasket on top (Fig. 2). Both the top and bottom chambers of all wells were filled with phosphate-buffered saline. The tissue was placed on top with the side of interest (luminal or the adventitial) facing up. For luminal measurements, cut edges, holes, and tears that fell inside a well were eliminated from further analysis because of the exposure of the adventitia. The chamber isolated regions with an area of 7 mm2. The upper chamber was gently rinsed with phosphate-buffered saline and aspirated before addition of 45 mL of factor VIIa and factor X (Proplex-T; Baxter, Deerfield, IL) diluted to a final
TF activity 181
Fig. 2. Measurement of TF activity. The vessel is opened longitudinally and placed in a modified Boyden chamber with the luminal surface exposed to the bottom of the overlying wells. Factors VIIa and X are added to the chamber and allowed to incubate for 15 min. TF complexes with factor VIIa on the luminal surface and converts factor X to factor Xa. A chromogenic substrate for factor Xa is added and quantified with spectrophotometry.
concentration of 2 units per mL fVIIa. These solutions were incubated on the luminal surface for 15 min at room temperature. The contents of the well were then transferred to a 96-well plate, and a 1:20 dilution of the chromogenic substrate for factor Xa (S-2765, Chromgenix Inc., Uppsala, Sweden) was added. The optical density at 405 nm was recorded using a Molecular Dynamics plate reader. Serial dilutions of the contents of a well incubated over the adventitia were used to establish the linear relationship between optical density and TF activity. Specificity for TF activity was demonstrated by preincubation with FFR-VIIai, which blocked the reaction (data not shown). This assay measures net potential TF activity. TF that is inactivated or unavailable to the lumen (intracellular or physically blocked from access by coagulation factors) is not detected. The small size of the sampled regions allowed for 2 to 4 independent measurements every 6 mm along the entire length of the graft, adjoining vessel, and adventitia. A typical specimen yielded 50 independent optical density measurements in the graft and 50 in the adjoining vessels. Multiple measurements from each location were averaged, and the results were mapped along the graft and adjoining vessels. The reference standard was the optical density at the most proximal (upstream) surface of the adjacent carotid artery, where TF activity
182 Hasenstab et al.
consistently was at the lowest level of all locations measured. Because this assay establishes relative values for TF activity rather than an absolute value, each specimen required its own reference value, and TF activity could not be compared between specimens. Since we established that TF activity was linearly related to optical density, activity was expressed as a percentage of the optical density at the reference arterial site. Comparison of TF activity at the arterial (proximal) and venous (distal) ends of the graft was made by averaging TF activity from all locations in the 2.5 cm region of graft adjacent to each anastomosis. Statistics All results are expressed as mean and standard deviation. Paired Wilcoxon nonparametric tests were used for all TF activity analysis. Values of P < 0.05 were considered significant.
RESULTS None of the animals died, and none of the grafts were thrombosed at the time of sacrifice. TF Activity As expected, TF activity on the adventitial surface of the native vessels was much greater than the luminal activity of the native vessels. It ranged from 1.5 to 18 fold higher than the activity on the lumen of the artery upstream from the graft, where levels always were the least. Overall, TF activity in the vein segment upstream from the graft was an average of 2.6 fold higher than in the upstream artery segment of the artery (P < 0.01). At all time points, TF activity was higher on the lumen of the graft than on that of the artery or vein and higher at the venous end than the arterial end. Figure 3 shows this pattern for the largest group: 6 grafts evaluated at 2 weeks. The jugular vein had higher TF levels than the artery, and activity levels in the graft segment adjacent to the artery were significantly greater than those in the artery (P < 0.05). Activity in the graft segment adjacent to the vein was significantly greater than in the segment adjacent to the artery (P < 0.05). At 2 days, TF activity in the segment of graft adjacent to the anastomoses was the same, or higher than that of the adventitia (n ¼ 3, Fig. 4). At 2 and 8 weeks, TF levels were significantly higher in the quarter of the graft nearest the venous anastomosis than in the quarter adjacent to the artery (P < 0.05, Fig. 4). Values in the central one-half of the graft
Annals of Vascular Surgery
were consistently lower than the region adjoining the anastomosis (data not shown). TF Localization TF protein was colocalized with cells exhibiting smooth muscle cell and macrophage markers, but these cells were not present in all regions containing TF (Fig. 5). TF often was localized in or immediately adjacent to regions of fibrin accumulation. The most common pattern was a region of TF underlying or adjacent to areas of fibrin accumulation. Less common, TF expression colocalized with fibrin. At late times, TF protein was present on the luminal surface of mature lesions where thrombus was no longer present. As expected, intimal lesions contained mostly actin-positive cells, and the interstices of the graft were filled with fibrin.
DISCUSSION Recently, most of the basic research on dialysis access failure has focused on development of intimal hyperplasia in arteriovenous fistulas. In these native vessels, stenosis results from proliferation of myofibroblasts, originating in the adventitia and migrating into the vessel, subsequent matrix deposition, and negative wall remodeling under the influence of inflammation, abnormal wall shear, and uremia.8 The vascular biology of neointimal hyperplasia in PTFE grafts is similar with the exception of more matrix in the lesions within the graft and a robust macrophage response with cytokine production and angiogenesis in the surrounding tissue.9 Contemporary studies to address access graft failure have focused on reducing proliferation, improving hemodynamics, and developing more biologically compatible prosthetic grafts.9 Our work suggests another approach: pharmacologic reduction of TF activity. We have demonstrated significant luminal TF activity in this sheep model of dialysis access graft stenosis as early as 2 days and as late as 8 weeks after engraftment. TF levels were often higher in the graft and adjacent vessels than in the adventitia, where TF is found in abundance to protect against hemorrhage from injured vessels. The activity at the venous end of the graft, where we previously demonstrated that intimal hyperplasia is the most robust, was higher than at the arterial end at 2 and 8 weeks after graft placement, suggesting a causative role for TF in this process. A similar process is likely to occur in human ePTFE access grafts since mature lesions found at the venous end of the sheep grafts are similar to those found clinically, consisting of smooth muscle cells,
Volume 31, February 2016
TF activity 183
Fig. 3. Summary data at 2 weeks. Direction of blood flow is indicated by arrows. Data presented are fold increase in TF activity as compared with the reference standard: the artery immediately upstream from the graft.
Fig. 4. TF activity in the PTFE graft, adjoining artery, and adventitia. At each time point TF activity is normalized to the value at the graft region adjacent to the arterial anastomosis. Paired comparisons between the arterial and
venous ends of the graft showed increased TF activity at 2 and 8 weeks (n ¼ 6, n ¼ 3, respectively, by paired Wilcoxon test). Differences at 2 days were not significant (n ¼ 3). N/A, not availableedata not collected.
matrix, and abundant microvessels.4 Sheep are a useful model for dialysis access graft failure studies because their coagulation system is similar to humans.5 It has been demonstrated that hemodialysis patients have even higher systemic TF and TF pathway inhibitor levels than normal controls.10 There are significant differences between the sheep model and human access graft failure: the size of recipient vein is larger; and the time course of stenosis is faster.
Mural thrombus contributes to intimal hyperplasia by providing a provisional matrix for subsequent ingrowth of mesenchymal cells. In addition, thrombus contains high levels of activated platelets and growth factors. Platelet-derived growth factors found in thrombus are potent smooth muscle cell mitogens and chemoattractants. TF is also able to generate activated coagulation factors including factor Xa and thrombin. Thrombin is an activator of platelets and is a mitogen for smooth muscle cells.11 TF itself can
184 Hasenstab et al.
Annals of Vascular Surgery
Fig. 5. Localization using an antihuman fibrin antibody and fluorescent-labeled factor VIIa in a 2-week graft. ePTFE is located in lower right. Thrombus is not present in this section, but fibrin (brown) represents prior
thrombosis. It is colocalized with TF (bright green) at the site of a suture (indicated by the arrows) and in another region near the surface of the graft. Original magnification 4.
mediate cell attachment and migration through binding to factor VIIa.12 Thus, TF is likely to be a potent initiator of intimal progression, both directly as a source of chemotactic and mitogenic factors and indirectly by providing provisional matrix. The adventitia of arteries has high levels of TF needed to rapidly initiate coagulation after vascular rupture or injury. TF activity frequently was higher in the ends of the grafts than in the adventitia and was often located close to the lumen. It is surprising that none of the grafts occluded because of thrombosis despite these high luminal TF levels, possibly due to the high flow in these devices. We have made a similar observation when we over expressed TF in the lumen of a rat carotid artery. In this model, if TF activity is blocked acutely and then allowed to be active chronically, thrombus deposition is gradual and does not result in sudden thrombosis.13 The sheep grafts may be responding in a similar manner, as suggested by the finding that most TF protein was near the lumen and adjacent to or colocalized with areas of fibrin accumulation. This finding suggests that the TF is biologically active and generates sufficient downstream coagulation factors to initiate fibrin accumulation. The time course of TF induction is rapid and sustained. At 2 days, TF levels in the graft were already at or above adventitial levels. The source of this early activity is not known. At this time, the graft surface has thrombus but very few cells. Giesen et al.14 have shown that TF is present in flowing blood and can cause thrombus to form on vascular surfaces, such as pig arterial media in the absence of cells. Our finding of TF staining in acellular regions of fibrin suggests that a similar process may occur in our sheep model. At 2 and 8 weeks, TF levels in the graft
remained high. Potential later sources of luminal TF are smooth muscle cells and macrophages, both of which are capable of producing the protein and both of which are abundant in developing lesions. Immunohistochemistry revealed that TF often colocalized with these cell types but was also found in some acellular regions. This could be due to 2 factors; these cells may have been present only briefly before the grafts were assayed, or some other cell type was responsible for TF production. Data are accumulating about the importance of TF in development of intimal hyperplasia.6 Abundant TF has also been found early in the course of vein-graft hyperplasia in rabbits,15 and intimal hyperplasia after arterial injury is reduced in knock-out mice deficient in TF.16 Our study demonstrates that TF also is active during the development of intimal hyperplasia in prosthetic dialysis grafts in our model, which closely mimics clinical lesions. Our results suggest that sustained TF activity, especially at the venous anastomosis, could contribute to graft failure in the sheep dialysis access model. Based on these findings, it is logical to explore whether interventions to inhibit TF would prevent or delay graft failure. Thrombin inhibitors can reduce intimal hyperplasia after arterial injury in animal models,17,18 and direct inhibition of TF is effective in limiting thrombosis in injured animal arteries and vein grafts.19 Application of the thrombin inhibitor factor VIIai to the luminal surface of injured rabbit arteries is effective in limiting thrombosis.20 Methods are available to deliver such blocking agents either locally or even to the luminal boundary layer of ePTFE grafts.21 Further studies with blocking agents are needed to determine causality and the possible benefits of such an approach.
Volume 31, February 2016
CONCLUSIONS In this sheep model of ePTFE dialysis access graft failure, we found that TF levels on the luminal surface are remarkably high throughout the grafts and the adjacent vessels as early as 2 days after engraftment and for as long as 8 weeks thereafter. The highest levels of activity were found in the venous end of the graft, where we know hyperplasia to be most robust. Increased activity of TF is associated with luminal thrombus, which provides a scaffolding for development of intimal hyperplasia. These findings present an opportunity to develop strategies to limit TF activity within the graft. Further studies using blocking agents are needed to establish a causative role for TF activity in development of intimal hyperplasia and to lay the groundwork for therapies to prevent it.
The authors would like to thank Alen Chen for technical assistance with immunohistochemical staining, Susan Rosell for histologic sectioning, Kari Canfield and Keri Clark for surgical assistance. Supported by Grants from The Northwest Kidney Foundation KOH001 and NRSA Cardiovascular Research Training Grant HL07828.
REFERENCES 1. Murad MH, Elamin MB, Sidawy AN, et al. Autogenous versus prosthetic vascular access for hemodialysis: a systematic review and meta-analysis. J Vasc Surg 2008;48(5 Suppl): 34Se47S. 2. Roy-Chaudhury P, Kelly BS, Miller MA, et al. Venous neointimal hyperplasia in polytetrafluoroethylene dialysis grafts. Kidney Int 2001;59:2325e34. Nature Publishing Group. 3. Swedberg SH, Brown BG, Sigley R, et al. Intimal fibromuscular hyperplasia at the venous anastomosis of PTFE grafts in hemodialysis patients. Clinical, immunocytochemical, light and electron microscopic assessment. Circulation 1989;80: 1726e36. 4. Kohler TR, Kirkman TR. Dialysis access failure: a sheep model of rapid stenosis. J Vasc Surg 1999;30:744e51. 5. Tillman P, Carson SN, Talken L. Platelet function and coagulation parameters in sheep during experimental vascular surgery. Lab Anim Sci 1981;31:263e7. 6. Williams JC, Mackman N. Tissue factor in health and disease. Front Biosci (Elite Ed) 2012;4:358e72.
TF activity 185
7. Zuckerman SH, Suprenant YM. Induction of endothelial cell/macrophage procoagulant activity: synergistic stimulation by gamma interferon and granulocyte-macrophage colony stimulating factor. Thromb Haemost 1989;61: 178e82. 8. Liang M, Wang Y, Liang A, et al. Migration of smooth muscle cells from the arterial anastomosis of arteriovenous fistulas requires Notch activation to form neointima. Kidney Int 2015;88:490e502. Nature Publishing Group. 9. Roy-Chaudhury P, Kruska L. Future directions for vascular access for hemodialysis. Semin Dial 2015;28:107e13. 10. Yorioka N, Taniguchi Y, Yamashita K, et al. Tissue factor and tissue factor pathway inhibitor in hemodialysis patients. Int J Artif Organs 1998;21:699e701. 11. McNamara CA, Sarembock IJ, Gimple LW, et al. Thrombin stimulates proliferation of cultured rat aortic smooth muscle cells by a proteolytically activated receptor. J Clin Invest 1993;91:94e8. 12. Ott I, Fischer EG, Miyagi Y, et al. A role for tissue factor in cell adhesion and migration mediated by interaction with actin-binding protein 280. J Cell Biol 1998;140:1241e53. 13. Hasenstab D, Lea H, Hart CE, et al. Tissue factor overexpression in rat arterial neointima models thrombosis and progression of advanced atherosclerosis. Circulation 2000;101: 2651e7. 14. Giesen PL, Rauch U, Bohrmann B, et al. Blood-borne tissue factor: another view of thrombosis. Proc Natl Acad Sci USA 1999;96:2311e5. 15. Channon KM, Fulton GJ, Davies MG, et al. Modulation of tissue factor protein expression in experimental venous bypass grafts. Arterioscler Thromb Vasc Biol 1997;17: 1313e9. 16. Pyo RT, Sato Y, Mackman N, et al. Mice deficient in tissue factor demonstrate attenuated intimal hyperplasia in response to vascular injury and decreased smooth muscle cell migration. Thromb Haemost 2004;92:451e8. 17. Li J-M, Singh MJ, Itani M, et al. Recombinant human thrombomodulin inhibits arterial neointimal hyperplasia after balloon injury. J Vasc Surg 2004;39:1074e83. 18. Ren G, Wang Z, Li Y, et al. Expression of antisense thrombin receptor gene inhibits intimal hyperplasia of rat carotid artery after balloon injury. Zhonghua Bing Li Xue Za Zhi 2002;31:231e5. 19. Huynh TT, Davies MG, Thompson MA, et al. Local treatment with recombinant tissue factor pathway inhibitor reduces the development of intimal hyperplasia in experimental vein grafts. J Vasc Surg 2001;33:400e7. 20. Arnljots B, Ezban M, Hedner U. Prevention of experimental arterial thrombosis by topical administration of active siteinactivated factor VIIa. J Vasc Surg 1997;25:341e6. 21. Chen C, Lumsden AB, Hanson SR. Local infusion of heparin reduces anastomotic neointimal hyperplasia in aortoiliac expanded polytetrafluoroethylene bypass grafts in baboons. J Vasc Surg 2000;31:354e63.
Background: Intimal hyperplasia at the venous anastomosis of dialysis grafts causes early failure. We developed a sheep model of arteriovenous prosthetic grafts that fail rapidly due to intimal hyperplasia with histologic features nearly identical to human access grafts. A prominent feature of lesion development in this model is formation of luminal thrombus that becomes organized into stenosing lesions by macrophage and myofibroblast infiltration. To better understand this process, we examined the presence and activity of tissue factor (TF) in this system. This protein is the physiological initiator of coagulation in vivo and is known to contribute to development of intimal hyperplasia after vascular injury. Methods: Expanded polytetrafluorethylene (ePTFE) grafts were placed between the carotid artery and external jugular vein in sheep. Grafts were examined for luminal TF activity using a novel ex vivo assay. In a separate series of grafts, immunohistochemistry was used to localize smooth muscle cells, monocytes, and TF protein. Results: At 2 days, luminal TF activity already was higher in the venous and arterial end of the graft than in the adventitia. This high level of activity persisted at 8 weeks. TF activity was higher in the venous end of the grafts than in the arterial end at 2 and 8 weeks (40% and 47% increase, n ¼ 5, n ¼ 3, respectively, P < 0.05). Immunohistochemistry revealed TF protein localized in regions with or adjacent to fibrin accumulation and often in regions close to the lumen. Conclusions: This study further examines the development of intimal hyperplasia in ePTFE dialysis access grafts. In this model, TF levels on the luminal surface were increased throughout the arteriovenous grafts and the adjacent vessels as early as 2 days after engraftment and for as long as 8 weeks thereafter. The highest levels of activity were found in the venous end of the graft, where hyperplasia is most robust. Increased activity of TF is associated with luminal thrombus, which provides a scaffolding for development of intimal hyperplasia. These findings present an opportunity to develop strategies to limit TF activity within the graft. Further studies using agents delivered locally or incorporated into the graft matrix to block the luminal activity of TF are warranted.
INTRODUCTION 1
Department of Surgery, University of Washington, Seattle, WA.
2
BioDevelopment Associates LLC, Mountlake Terrace, WA.
3
Department of Surgery, Seattle VA Puget Sound Health Care System, Seattle, WA. Correspondence to: Ted R. Kohler, MS, MD, Department of Surgery, Seattle VA Puget Sound Healthcare System, 112v 1660 S. Columbian Way, Seattle, WA 98108, USA; E-mail: [email protected] Ann Vasc Surg 2016; 31: 179–185 http://dx.doi.org/10.1016/j.avsg.2015.10.008 Published by Elsevier Inc. Manuscript received: April 10, 2015; manuscript accepted: October 11, 2015; published online: November 17, 2015.
Vascular access failure is a major source of morbidity, hospitalization, and cost for patients receiving chronic hemodialysis. Prosthetic grafts, most commonly composed of expanded polytetrafluorethylene (ePTFE), are used for patients who cannot have an autologous fistula due to lack of adequate venous conduit.1 Approximately 60% of these grafts fail each year due to intimal hyperplasia, primarily at the venous end. Clinical specimens reveal intimal hyperplasia containing myofibroblasts, organized thrombus, abundant microvessels, and activated macrophages, 179
180 Hasenstab et al.
which produce cytokines that enhance smooth muscle cell migration and proliferation.2,3 We developed a sheep model of access failure in which significant narrowing occurs within weeks at the venous end.4 The histology of the narrowed segment is nearly identical to the human specimens described above. Thrombus is a prominent, early feature of these developing lesions. As in the human specimens, macrophages play an important role, invading the thrombus, which is then transformed into the mature lesion by influx of myofibroblasts and capillaries (Fig. 1). It is not known if thrombus also is a significant feature of early lesion development in clinical grafts because pathologic evaluation is generally restricted to mature, stenotic segments. However, the work of Tillman et al.5 has shown that the thrombotic response to vascular surgery in sheep is very similar to that of humans. The presence of luminal thrombus in our model suggests that tissue factor (TF) is active at the surface. This protein may be derived from activated platelets, smooth muscle cells, or macrophages.6 In addition to causing thrombosis, TF activates smooth muscle cell proliferation and migration, adding to development of hyperplastic lesions. To better understand this process in dialysis access failure, we used a novel assay to measure the surface activity of TF in our sheep model and performed immunohistochemistry to localize the protein.
MATERIALS AND METHODS Experimental Design and Operative Procedures Columbia crossbred sheep of either sex (Nebeker Ranch, Inc., Lancaster, CA), weighing between 40 and 60 kg were fasted overnight, given intramuscular ceftiofur (2 mg/kg), and premedicated with intramuscular xylazine (0.2 mg/kg), and intravenous ketamine (10e15 mg/kg). After intubation, anesthesia was maintained with inhalation isoflurane. The neck was prepared with chlorhexidine, 70% isopropyl alcohol, and a final application of iodine solution. The left carotid artery and the contralateral external jugular vein were surgically exposed. Animals were anticoagulated with intravenous heparin (150 U/ kg). Using standard vascular technique, commercially available, 6-mm internal diameter ePTFE grafts (30 m internodal distance, IMPRA, Inc., Tempe, AZ) were sutured end-to-side to blood vessels with continuous 6-0 polypropylene suture. Grafts were 11 to 15 cm in length. Fascia and skin were closed in layers with bioabsorbable suture. Buprenex was given (0.005 mg/kg) every 12 hr for 48 hr after
Annals of Vascular Surgery
Fig. 1. Microphotograph of a developing lesion in the sheep model at 2 weeks. The graft (G) has granulation tissue on its outer wall and thrombus on the lumen. The edge of the thrombus is being invaded by cells, which in our prior work were shown to be macrophages (arrow). These cells engulf red cell debris including hemoglobin, which becomes the hemosiderin seen in late lesions, and cell membranes, creating fat laden macrophages. Behind this advancing front of macrophages, myofibroblasts migrate into the fibrin scaffolding to form the neointima (original magnification 4, hematoxylin and eosin staining.).
surgery to relieve postoperative pain. Intramuscular ceftiofur (2 mg/kg) was continued twice a day on days 1 and 2 and daily on days 3 through 5 after surgery. Oral aspirin (81 mg/day) was administered one day before surgery and once per day throughout the study. Patency was checked daily for the first week and then 3 times a week by palpation for a thrill and by hand-held Doppler. Animals were maintained on a standard diet (Purina Mills Northwestern Pride DietÔ 32B4 pellets and hay cubes twice per day). Studies were approved by the Institutional Animal Care and Use Committee. Three animals were sacrificed at 2 days, 6 at 2 weeks, and 3 at 8 weeks. Immunohistochemistry and TF Detection Cross sections of representative segments of ePTFE grafts and adjacent vessels that were not used for TF assay were embedded in paraffin and cut in 5-m thickness for microscopic examination, including hematoxylin and eosin staining. Paraffin embedded, formalin-fixed tissues were stained as previously described using antibodies for the following: fibrin (1:20 dilution, antifibrin #350, American Diagnostica), smooth muscle cells (antialpha actin, Boehringer Mannheim) and macrophages (Ham56, Enzo Diagnostics, Farmingdale NY). TF was detected in histologic sections using a fluorescent-labeled
Volume 31, February 2016
ligand. The inactive form of factor VIIa (inactivated Phe-Phe-Arg chloromethylketone (FFR)-VIIai, Zymogenetics, Seattle, WA), which retains full binding capacity, was coupled to the fluorescent dye FluoReporter Oregon Green 488 (Molecular Probes, Eugene, OR) as described by the manufacturer using a dye molar ratio of 1:20 with 480 mg FFR-VIIai per reaction. The labeled factor VIIai was then used to detect TF. Sections were deparaffinized and rehydrated in phosphate-buffered saline. Fifty mL of a 1:50 dilution of fluorescent VIIai in phosphate-buffered saline with 1% bovine serum albumin was placed on the section at room temperature for 1 hr. The sections were washed 3 times in phosphate-buffered saline and mounted using Vectashield aqueous mounting medium (Vector Labs, Burlingame, CA). Sections were imaged using a fluorescent microscope (Leitz Dialux 20 EB) and digital camera (Diagnostic Instruments). TF Activity Assay We developed a novel assay that allowed us to measure TF activity on the surface of the grafts using a modified Boyden chamber. At the time of graft harvest, the animals were killed by an overdose of sodium pentobarbital. The neck was then rapidly opened, and the graft with a generous portion of adjacent artery and vein were excised en bloc. The graft and adjoining vessel were flushed with lactated Ringer’s lactate and longitudinally opened. Loose connective and fibrous tissue were removed. The graft was cut in half at the midpoint and separated from the adjoining artery or vein. Specimens were kept submerged in ice-cold Ringers’ solution after removal and during all subsequent manipulations until assayed. TF activity was measured serially along the length of the graft, on the adventitia, and on the adjacent vessels using an assay based on factor Xa cleavage of a chromogenic substrate.7 The tissue was placed in a modified Boyden chamber (Microchemotaxis chamber; Neuro Probe Inc., Gaithersburg, MD) with the silicone gasket on top (Fig. 2). Both the top and bottom chambers of all wells were filled with phosphate-buffered saline. The tissue was placed on top with the side of interest (luminal or the adventitial) facing up. For luminal measurements, cut edges, holes, and tears that fell inside a well were eliminated from further analysis because of the exposure of the adventitia. The chamber isolated regions with an area of 7 mm2. The upper chamber was gently rinsed with phosphate-buffered saline and aspirated before addition of 45 mL of factor VIIa and factor X (Proplex-T; Baxter, Deerfield, IL) diluted to a final
TF activity 181
Fig. 2. Measurement of TF activity. The vessel is opened longitudinally and placed in a modified Boyden chamber with the luminal surface exposed to the bottom of the overlying wells. Factors VIIa and X are added to the chamber and allowed to incubate for 15 min. TF complexes with factor VIIa on the luminal surface and converts factor X to factor Xa. A chromogenic substrate for factor Xa is added and quantified with spectrophotometry.
concentration of 2 units per mL fVIIa. These solutions were incubated on the luminal surface for 15 min at room temperature. The contents of the well were then transferred to a 96-well plate, and a 1:20 dilution of the chromogenic substrate for factor Xa (S-2765, Chromgenix Inc., Uppsala, Sweden) was added. The optical density at 405 nm was recorded using a Molecular Dynamics plate reader. Serial dilutions of the contents of a well incubated over the adventitia were used to establish the linear relationship between optical density and TF activity. Specificity for TF activity was demonstrated by preincubation with FFR-VIIai, which blocked the reaction (data not shown). This assay measures net potential TF activity. TF that is inactivated or unavailable to the lumen (intracellular or physically blocked from access by coagulation factors) is not detected. The small size of the sampled regions allowed for 2 to 4 independent measurements every 6 mm along the entire length of the graft, adjoining vessel, and adventitia. A typical specimen yielded 50 independent optical density measurements in the graft and 50 in the adjoining vessels. Multiple measurements from each location were averaged, and the results were mapped along the graft and adjoining vessels. The reference standard was the optical density at the most proximal (upstream) surface of the adjacent carotid artery, where TF activity
182 Hasenstab et al.
consistently was at the lowest level of all locations measured. Because this assay establishes relative values for TF activity rather than an absolute value, each specimen required its own reference value, and TF activity could not be compared between specimens. Since we established that TF activity was linearly related to optical density, activity was expressed as a percentage of the optical density at the reference arterial site. Comparison of TF activity at the arterial (proximal) and venous (distal) ends of the graft was made by averaging TF activity from all locations in the 2.5 cm region of graft adjacent to each anastomosis. Statistics All results are expressed as mean and standard deviation. Paired Wilcoxon nonparametric tests were used for all TF activity analysis. Values of P < 0.05 were considered significant.
RESULTS None of the animals died, and none of the grafts were thrombosed at the time of sacrifice. TF Activity As expected, TF activity on the adventitial surface of the native vessels was much greater than the luminal activity of the native vessels. It ranged from 1.5 to 18 fold higher than the activity on the lumen of the artery upstream from the graft, where levels always were the least. Overall, TF activity in the vein segment upstream from the graft was an average of 2.6 fold higher than in the upstream artery segment of the artery (P < 0.01). At all time points, TF activity was higher on the lumen of the graft than on that of the artery or vein and higher at the venous end than the arterial end. Figure 3 shows this pattern for the largest group: 6 grafts evaluated at 2 weeks. The jugular vein had higher TF levels than the artery, and activity levels in the graft segment adjacent to the artery were significantly greater than those in the artery (P < 0.05). Activity in the graft segment adjacent to the vein was significantly greater than in the segment adjacent to the artery (P < 0.05). At 2 days, TF activity in the segment of graft adjacent to the anastomoses was the same, or higher than that of the adventitia (n ¼ 3, Fig. 4). At 2 and 8 weeks, TF levels were significantly higher in the quarter of the graft nearest the venous anastomosis than in the quarter adjacent to the artery (P < 0.05, Fig. 4). Values in the central one-half of the graft
Annals of Vascular Surgery
were consistently lower than the region adjoining the anastomosis (data not shown). TF Localization TF protein was colocalized with cells exhibiting smooth muscle cell and macrophage markers, but these cells were not present in all regions containing TF (Fig. 5). TF often was localized in or immediately adjacent to regions of fibrin accumulation. The most common pattern was a region of TF underlying or adjacent to areas of fibrin accumulation. Less common, TF expression colocalized with fibrin. At late times, TF protein was present on the luminal surface of mature lesions where thrombus was no longer present. As expected, intimal lesions contained mostly actin-positive cells, and the interstices of the graft were filled with fibrin.
DISCUSSION Recently, most of the basic research on dialysis access failure has focused on development of intimal hyperplasia in arteriovenous fistulas. In these native vessels, stenosis results from proliferation of myofibroblasts, originating in the adventitia and migrating into the vessel, subsequent matrix deposition, and negative wall remodeling under the influence of inflammation, abnormal wall shear, and uremia.8 The vascular biology of neointimal hyperplasia in PTFE grafts is similar with the exception of more matrix in the lesions within the graft and a robust macrophage response with cytokine production and angiogenesis in the surrounding tissue.9 Contemporary studies to address access graft failure have focused on reducing proliferation, improving hemodynamics, and developing more biologically compatible prosthetic grafts.9 Our work suggests another approach: pharmacologic reduction of TF activity. We have demonstrated significant luminal TF activity in this sheep model of dialysis access graft stenosis as early as 2 days and as late as 8 weeks after engraftment. TF levels were often higher in the graft and adjacent vessels than in the adventitia, where TF is found in abundance to protect against hemorrhage from injured vessels. The activity at the venous end of the graft, where we previously demonstrated that intimal hyperplasia is the most robust, was higher than at the arterial end at 2 and 8 weeks after graft placement, suggesting a causative role for TF in this process. A similar process is likely to occur in human ePTFE access grafts since mature lesions found at the venous end of the sheep grafts are similar to those found clinically, consisting of smooth muscle cells,
Volume 31, February 2016
TF activity 183
Fig. 3. Summary data at 2 weeks. Direction of blood flow is indicated by arrows. Data presented are fold increase in TF activity as compared with the reference standard: the artery immediately upstream from the graft.
Fig. 4. TF activity in the PTFE graft, adjoining artery, and adventitia. At each time point TF activity is normalized to the value at the graft region adjacent to the arterial anastomosis. Paired comparisons between the arterial and
venous ends of the graft showed increased TF activity at 2 and 8 weeks (n ¼ 6, n ¼ 3, respectively, by paired Wilcoxon test). Differences at 2 days were not significant (n ¼ 3). N/A, not availableedata not collected.
matrix, and abundant microvessels.4 Sheep are a useful model for dialysis access graft failure studies because their coagulation system is similar to humans.5 It has been demonstrated that hemodialysis patients have even higher systemic TF and TF pathway inhibitor levels than normal controls.10 There are significant differences between the sheep model and human access graft failure: the size of recipient vein is larger; and the time course of stenosis is faster.
Mural thrombus contributes to intimal hyperplasia by providing a provisional matrix for subsequent ingrowth of mesenchymal cells. In addition, thrombus contains high levels of activated platelets and growth factors. Platelet-derived growth factors found in thrombus are potent smooth muscle cell mitogens and chemoattractants. TF is also able to generate activated coagulation factors including factor Xa and thrombin. Thrombin is an activator of platelets and is a mitogen for smooth muscle cells.11 TF itself can
184 Hasenstab et al.
Annals of Vascular Surgery
Fig. 5. Localization using an antihuman fibrin antibody and fluorescent-labeled factor VIIa in a 2-week graft. ePTFE is located in lower right. Thrombus is not present in this section, but fibrin (brown) represents prior
thrombosis. It is colocalized with TF (bright green) at the site of a suture (indicated by the arrows) and in another region near the surface of the graft. Original magnification 4.
mediate cell attachment and migration through binding to factor VIIa.12 Thus, TF is likely to be a potent initiator of intimal progression, both directly as a source of chemotactic and mitogenic factors and indirectly by providing provisional matrix. The adventitia of arteries has high levels of TF needed to rapidly initiate coagulation after vascular rupture or injury. TF activity frequently was higher in the ends of the grafts than in the adventitia and was often located close to the lumen. It is surprising that none of the grafts occluded because of thrombosis despite these high luminal TF levels, possibly due to the high flow in these devices. We have made a similar observation when we over expressed TF in the lumen of a rat carotid artery. In this model, if TF activity is blocked acutely and then allowed to be active chronically, thrombus deposition is gradual and does not result in sudden thrombosis.13 The sheep grafts may be responding in a similar manner, as suggested by the finding that most TF protein was near the lumen and adjacent to or colocalized with areas of fibrin accumulation. This finding suggests that the TF is biologically active and generates sufficient downstream coagulation factors to initiate fibrin accumulation. The time course of TF induction is rapid and sustained. At 2 days, TF levels in the graft were already at or above adventitial levels. The source of this early activity is not known. At this time, the graft surface has thrombus but very few cells. Giesen et al.14 have shown that TF is present in flowing blood and can cause thrombus to form on vascular surfaces, such as pig arterial media in the absence of cells. Our finding of TF staining in acellular regions of fibrin suggests that a similar process may occur in our sheep model. At 2 and 8 weeks, TF levels in the graft
remained high. Potential later sources of luminal TF are smooth muscle cells and macrophages, both of which are capable of producing the protein and both of which are abundant in developing lesions. Immunohistochemistry revealed that TF often colocalized with these cell types but was also found in some acellular regions. This could be due to 2 factors; these cells may have been present only briefly before the grafts were assayed, or some other cell type was responsible for TF production. Data are accumulating about the importance of TF in development of intimal hyperplasia.6 Abundant TF has also been found early in the course of vein-graft hyperplasia in rabbits,15 and intimal hyperplasia after arterial injury is reduced in knock-out mice deficient in TF.16 Our study demonstrates that TF also is active during the development of intimal hyperplasia in prosthetic dialysis grafts in our model, which closely mimics clinical lesions. Our results suggest that sustained TF activity, especially at the venous anastomosis, could contribute to graft failure in the sheep dialysis access model. Based on these findings, it is logical to explore whether interventions to inhibit TF would prevent or delay graft failure. Thrombin inhibitors can reduce intimal hyperplasia after arterial injury in animal models,17,18 and direct inhibition of TF is effective in limiting thrombosis in injured animal arteries and vein grafts.19 Application of the thrombin inhibitor factor VIIai to the luminal surface of injured rabbit arteries is effective in limiting thrombosis.20 Methods are available to deliver such blocking agents either locally or even to the luminal boundary layer of ePTFE grafts.21 Further studies with blocking agents are needed to determine causality and the possible benefits of such an approach.
Volume 31, February 2016
CONCLUSIONS In this sheep model of ePTFE dialysis access graft failure, we found that TF levels on the luminal surface are remarkably high throughout the grafts and the adjacent vessels as early as 2 days after engraftment and for as long as 8 weeks thereafter. The highest levels of activity were found in the venous end of the graft, where we know hyperplasia to be most robust. Increased activity of TF is associated with luminal thrombus, which provides a scaffolding for development of intimal hyperplasia. These findings present an opportunity to develop strategies to limit TF activity within the graft. Further studies using blocking agents are needed to establish a causative role for TF activity in development of intimal hyperplasia and to lay the groundwork for therapies to prevent it.
The authors would like to thank Alen Chen for technical assistance with immunohistochemical staining, Susan Rosell for histologic sectioning, Kari Canfield and Keri Clark for surgical assistance. Supported by Grants from The Northwest Kidney Foundation KOH001 and NRSA Cardiovascular Research Training Grant HL07828.
REFERENCES 1. Murad MH, Elamin MB, Sidawy AN, et al. Autogenous versus prosthetic vascular access for hemodialysis: a systematic review and meta-analysis. J Vasc Surg 2008;48(5 Suppl): 34Se47S. 2. Roy-Chaudhury P, Kelly BS, Miller MA, et al. Venous neointimal hyperplasia in polytetrafluoroethylene dialysis grafts. Kidney Int 2001;59:2325e34. Nature Publishing Group. 3. Swedberg SH, Brown BG, Sigley R, et al. Intimal fibromuscular hyperplasia at the venous anastomosis of PTFE grafts in hemodialysis patients. Clinical, immunocytochemical, light and electron microscopic assessment. Circulation 1989;80: 1726e36. 4. Kohler TR, Kirkman TR. Dialysis access failure: a sheep model of rapid stenosis. J Vasc Surg 1999;30:744e51. 5. Tillman P, Carson SN, Talken L. Platelet function and coagulation parameters in sheep during experimental vascular surgery. Lab Anim Sci 1981;31:263e7. 6. Williams JC, Mackman N. Tissue factor in health and disease. Front Biosci (Elite Ed) 2012;4:358e72.
TF activity 185
7. Zuckerman SH, Suprenant YM. Induction of endothelial cell/macrophage procoagulant activity: synergistic stimulation by gamma interferon and granulocyte-macrophage colony stimulating factor. Thromb Haemost 1989;61: 178e82. 8. Liang M, Wang Y, Liang A, et al. Migration of smooth muscle cells from the arterial anastomosis of arteriovenous fistulas requires Notch activation to form neointima. Kidney Int 2015;88:490e502. Nature Publishing Group. 9. Roy-Chaudhury P, Kruska L. Future directions for vascular access for hemodialysis. Semin Dial 2015;28:107e13. 10. Yorioka N, Taniguchi Y, Yamashita K, et al. Tissue factor and tissue factor pathway inhibitor in hemodialysis patients. Int J Artif Organs 1998;21:699e701. 11. McNamara CA, Sarembock IJ, Gimple LW, et al. Thrombin stimulates proliferation of cultured rat aortic smooth muscle cells by a proteolytically activated receptor. J Clin Invest 1993;91:94e8. 12. Ott I, Fischer EG, Miyagi Y, et al. A role for tissue factor in cell adhesion and migration mediated by interaction with actin-binding protein 280. J Cell Biol 1998;140:1241e53. 13. Hasenstab D, Lea H, Hart CE, et al. Tissue factor overexpression in rat arterial neointima models thrombosis and progression of advanced atherosclerosis. Circulation 2000;101: 2651e7. 14. Giesen PL, Rauch U, Bohrmann B, et al. Blood-borne tissue factor: another view of thrombosis. Proc Natl Acad Sci USA 1999;96:2311e5. 15. Channon KM, Fulton GJ, Davies MG, et al. Modulation of tissue factor protein expression in experimental venous bypass grafts. Arterioscler Thromb Vasc Biol 1997;17: 1313e9. 16. Pyo RT, Sato Y, Mackman N, et al. Mice deficient in tissue factor demonstrate attenuated intimal hyperplasia in response to vascular injury and decreased smooth muscle cell migration. Thromb Haemost 2004;92:451e8. 17. Li J-M, Singh MJ, Itani M, et al. Recombinant human thrombomodulin inhibits arterial neointimal hyperplasia after balloon injury. J Vasc Surg 2004;39:1074e83. 18. Ren G, Wang Z, Li Y, et al. Expression of antisense thrombin receptor gene inhibits intimal hyperplasia of rat carotid artery after balloon injury. Zhonghua Bing Li Xue Za Zhi 2002;31:231e5. 19. Huynh TT, Davies MG, Thompson MA, et al. Local treatment with recombinant tissue factor pathway inhibitor reduces the development of intimal hyperplasia in experimental vein grafts. J Vasc Surg 2001;33:400e7. 20. Arnljots B, Ezban M, Hedner U. Prevention of experimental arterial thrombosis by topical administration of active siteinactivated factor VIIa. J Vasc Surg 1997;25:341e6. 21. Chen C, Lumsden AB, Hanson SR. Local infusion of heparin reduces anastomotic neointimal hyperplasia in aortoiliac expanded polytetrafluoroethylene bypass grafts in baboons. J Vasc Surg 2000;31:354e63.