- Email: [email protected]
Seeding of Human Endothelial Cells on Valve Containing Aortic Mini-Roots: Development of a Seeding Device and Procedure
Seeding of Human Endothelial Cells on Valve Containing Aortic Mini-Roots: Development of a Seeding Device and Procedure Helmut Gulbins, MD, Anita Pritisanac, MD, Antje Uhlig, Angelika Goldemund, Bruno M. Meiser, MD, Bruno Reichart, MD, and Sabine Daebritz, MD
Purpose. Complete covering of an artificial valvular scaffold with endothelial cells may prevent thromboembolic complications and lead to an excellent biocompatibility. For this purpose, we developed a seeding device for reproducible cell seeding on valve containing aortic roots. Description. Human endothelial cells and fibroblasts were obtained from saphenous vein pieces. Cryopreserved aortic roots (n ⴝ 25) were put into an especially developed tube, set on a rotator, and incubated with the cell suspension. The device rotated in two axes (sagittal and axial), ensuring slight movements of the leaflets. The rotation alternated with resting periods, allowing cell attachment to the surface. Different resting periods were tested (groups 1, 2, and 3 were 30, 45, and 60 min, respectively; n ⴝ 5 each). Total incubation time was 24 hours followed by further culturing for 6 days. In two further groups (groups 4 and 5; n ⴝ 5 each), a modified inlay was used to allow the cell suspension to flow around the entire graft. In group 4 the grafts were again incubated with human endothelial cells; however, in group 5 pre-seeding with autologous fibroblasts was done in addition. Immunohistochemical staining with antibodies against factor VIII, CD31, laminin, collagen IV, and CD90 were done, and scanning electron microscopy was done after initial seeding and after 6 days in culture. Evaluation. Seeding resulted in homogenous cell layers on the luminal surface of the free walls in all groups. With resting periods of 45 minutes, these results were also obtained on the leaflets, whereas the other resting times resulted in defects of the endothelial cell layer on the cusps. After 6 days under culture conditions, the endothelial cell layers were confluent and viable, with the exception of the leaflets in group 1. With the modified inlay (groups 4 and 5), confluent cell layers were also achieved on the outer surface. In group 5 pre-seeding with autologous fibroblasts resulted in enhanced synthesis of extracellular matrix proteins, as was demonstrated with immunohistochemical staining for collagen IV and laminin. Conclusions. With this newly developed seeding device, confluent cell layers on valve containing aortic roots were reproducibly achieved. The technique enables further experimental research and even clinical application. (Ann Thorac Surg 2005;79:2119 –26) © 2005 by The Society of Thoracic Surgeons
T
he lack of endothelial cell (EC) covering of mechanical or biological valve prostheses under in vivo conditions in humans is one major shortcoming of these For editorial comment, see page 1831 substitutes. A confluent EC layer could increase biocompatibility of biological valve prostheses by reducing thrombogenicity. Human ECs can be grown to confluence on various tissues, but seeding on 3-dimensional Accepted for publication May 6, 2004. Address reprint requests to Dr Gulbins, Department of Cardiac Surgery, University Hospital Ulm, Steinhövelstr 9, D-89070 Ulm, Germany; e-mail: [email protected]..
© 2005 by The Society of Thoracic Surgeons Published by Elsevier Inc
objects proved to be more difficult. The purpose of our study was to develop a device for reproducible cell seeding on valve containing aortic mini-roots. This would enable cell seeding on complex 3-dimensional scaffolds to develop a custom-made heart valve seeded with autologous cells of the patient.
Technology Cell Culture Human ECs and fibroblasts were isolated from saphenous vein pieces as previously described [1– 4]. The vein pieces were leftover from routine aorto-coronary bypass 0003-4975/05/$30.00 doi:10.1016/j.athoracsur.2004.05.085
NEW TECHNOLOGY
Department of Cardiac Surgery, University Hospital Grosshadern, Munich, Germany
2120
NEW TECHNOLOGY GULBINS ET AL DEVELOPMENT OF A SEEDING DEVICE AND PROCEDURE
Ann Thorac Surg 2005;79:2119 –26
NEW TECHNOLOGY
Fig 1. Homograft, fixed on the incubator tube and placed inside the incubator viewed through the graft in the direction of blood flow. With this tube, the cell suspension reached only the luminal surface of the graft, which was shorter than the incubator, thus allowing longitudinal movements of the graft inside during the rotation. This ensured an exchange of the cell suspension between both sides of the leaflets.
operations from which the patients had given their informed consent for these pieces to be used in the laboratory. The local ethics committee approved the anonymous use of the saphenous vein pieces for this experimental study. The vein pieces were cannulated, rinsed with buffered medium, and incubated with 0.1% collagenase (Worthington, CA) for 20 minutes at 37°C and 5% CO2. Thereafter, the collagenase reaction was stopped, the cell suspension was centrifuged at 1,000 U/min for 10 minutes and the cell pellet was resuspended in EC growth medium (Promocell, Heidelberg, Germany), and was plated on culture dishes. For fibroblast isolation, the same vein pieces were again filled with 0.1% collagenase and incubated for 30 minutes. This procedure was repeated twice. Again, the cell suspension was centrifuged at 1000 U/min for 10 minutes, the pellet was re-suspended in fibroblast growth medium (Promocell, Heidelberg, Germany), and was plated on culture dishes.
Fig 2. The incubator was fixed on the self-developed rotation device. Shown is the new tube placed inside and the closing mechanism. The incubator was filled with 120 mL cell suspension and the air was removed through the 3-way connector. The two controls at the base were used to choose the speed and time settings.
incubated with M-199 (supplemented with 5% fetal calf serum) for 1 day. For cell seeding, a special device (rotation incubator) was developed in cooperation with the technical staff. The mini-roots were fixed on a tube (Fig 1), put into a flask (incubator), and set on the rotator (Fig 2). The internal volume of this incubator was 100 mL. The device rotated the tube around the axial and sagittal axes (Fig 3).
Technique Seeding on Aortic Roots Aortic roots were obtained from explanted hearts during heart transplantations or from multiorgan donors whose hearts were not used for transplantation. After sterilization in an antibiotic cocktail (ciprofloxacin, amikacin, vancomycin, metronidazol, and amphotericin B) for 3 days, the grafts were cryopreserved in M-199 with addition of dimethyl sulfoxide, and were then stored at ⫺196°C in liquid nitrogen for 45 days (range, 15 to 102 days). For seeding, the grafts were rewarmed in a sterile water bed, flushed with M-199 for 3 hours, and then
Fig 3. Rotations and other movements of the tube inside the incubator during the rotation phase (the tube rotated axially and longitudinally, whereas the whole incubator was moved counterclockwise). These movements re-suspended those cells that had not attached to the graft during the resting phase.
NEW TECHNOLOGY GULBINS ET AL DEVELOPMENT OF A SEEDING DEVICE AND PROCEDURE
Fig 4. Homograft fixed to the new tube (groups 4 and 5). This new tube design allowed the grafts to be washed around by the cell suspension. This ensured that the outer surface was also cell coated, especially with fibroblasts.
When the incubator moved around the sagittal axis, the tube inside moved longitudinally, causing a slow motion of the valve leaflets. This also allowed an exchange of the cell suspension of both sides of the valve. The resting periods and rotation periods could be set individually. One transparent incubator was built to study the technical settings. An untreated homograft was placed inside, and the motions of the valve leaflets were examined with different rotation speed settings (1 round/min, 2.5 rounds/min, and 4 rounds/min). For EC seeding, the incubator was filled with EC medium supplemented with 5% fetal calf serum. Endothelial cells were trypsinized as previously described and were re-suspended in 10 mL phosphate buffered saline. The EC suspension (inoculation density 5.5 ⫻ 105 cells/cm2) was then transferred to the rotation incubator. The total incubation time of the aortic roots lasted for 24 hours. During incubation, resting and rotation periods were alternated. The resting phases were set at 30 minutes for group 1 (n ⫽ 5), 45 minutes for group 2 (n ⫽ 5), and 60 minutes for group 3 (n ⫽ 5), resulting in 44, 30, and 23 rotation cycles. During the rotations, the grafts were turned around 800° in all groups. The medium was then removed and nonadherent cells were identified using trypan blue staining and
2121
were counted using the Neubauer chamber. Initial cell adhesion was defined as the difference between the initial cell number of the cell suspension and these remaining, nonadherent cells. The values were given in a percentage of the initial cell number. For two more groups, the design of the tube was changed to allow cell seeding also on the outer surface. In this new tube (Fig 4), the mini-roots were fixed only with few sutures at their ends, exposing the outer as well as the luminal surface to the cells for attachment. In group 4, only ECs were seeded on the grafts as previously described. For group 5, the incubator was filled with fibroblast growth medium supplemented with 10% fetal calf serum. Human fibroblasts were trypsinized, washed with phosphate buffered saline, centrifuged, and the re-suspension volume was reduced to 10 mL. With an inoculation density of 7.7 ⫻ 105 cells/cm2, the fibroblast cell suspension was added to the incubator. Again, the same settings as previously described were used. Thereafter EC seeding was done as previously described for group 4. All seeded roots (groups 1 to 5) (see Table 1) had a further culturing phase of 6 days to develop confluent cell layers. For this, the tube was placed into a culture glass filled with 80 mL EC medium at 37°C at 5% CO2 atmosphere.
Immunohistochemical Staining A specimen of the aortic wall and valve leaflets were frozen and stored in liquid nitrogen. Eight mm sections were stained using monoclonal antibodies against factor VIII, CD31 (Dako, Hamburg, Germany), and collagen IV (Sigma, Deisnhofen, Germany). For factor VIII staining, a polyclonal antibody against antirabbit IgG (Immundiagnostik, Bensheim, Germany) was used. Collagen IV and CD31 stainings were counterstained with a polyclonal antibody against mouse IgG (Chemicon Int, Inc, CA). After incubation with the primary antibody against factor VIII, CD31, and collagen IV, an antimouse IgG antibody (Dako, Hamburg, Germany) was added and stained with 3-amino-9-ethyl carbazol (Dako, Hamburg, Germany). The results were validated with negative control stainings without the primary antibody.
Scanning Electron Microscopy Only cell layers with the typical cobblestone morphology were accepted as EC layers. For analysis, 10 visual fields of each specimen were evaluated at 1,000⫻ magnification. Aortic and ventricular surfaces of the leaflets were evaluated. The endothelial cell covering was assessed
Table 1. Definition of the Five Different Groupsa
Resting phase Fibroblast seeding Endothelial cell seeding a
Group 1
Group 2
Group 3
Group 4
Group 5
30 min ⫺ ⫹
45 min ⫺ ⫹
60 min ⫺ ⫹
45 min ⫺ ⫹
45 min ⫹ ⫹
In groups 4 and 5, a modified inlay was used to allow the cell suspension to flow around the entire graft. In groups 1 to 3 the length of the resting phases differed. Group 4 served as a control group for group 5, in which additional pre-seeding with fibroblasts was tested.
NEW TECHNOLOGY
Ann Thorac Surg 2005;79:2119 –26
2122
NEW TECHNOLOGY GULBINS ET AL DEVELOPMENT OF A SEEDING DEVICE AND PROCEDURE
NEW TECHNOLOGY
Fig 5. Scanning electron microscopy (magnification, ⫻200) of the luminal surface of the free wall of a homograft after cryopreservation before endothelial cell seeding. No endothelial cells were seen, but interstitial fibers were seen.
semiquantitatively by two investigators and was classified as 100%, 75%, 50%, or 25%. A mean value of 95% and greater was believed to represent a confluent EC layer.
Results The results of cryopreserved homografts showed a nearly complete loss of the ECs on all specimen taken before cell seeding (Fig 5). Only collagen fibers and fibrocytes were seen. Although it was possible to culture fibroblasts from all mini-roots (n ⫽ 25), viable ECs could only be isolated from three mini-root pieces in a very small number. The first tests with the transparent incubator showed nearly no movements of the leaflets at a rotation speed of 1 round/min, whereas with both faster speed settings valve motions were seen. Therefore we chose 2.5 rounds/ min for further experiments as the standard setting. In groups 1, 2, and 3, incubation with ECs resulted in an initial cell adhesion of 68%, 75%, and 72%, respec-
Fig 6. Graphical demonstration of the seeding results in the different groups.
Ann Thorac Surg 2005;79:2119 –26
Fig 7. Scanning electron microscopic examination (magnification, ⫻500) of group 5 after fibroblast seeding. The fibroblasts formed a confluent cell layer. The cells could be identified by their typical long shape.
tively (Fig 6). On the luminal surface of the free walls, a confluent EC layer was seen. Only rare defects were detected, concerning less than 1% of the specimen areas. After the final culturing phase of 6 days, confluent cell layers were documented on the free walls in all cases. Staining for CD31 and factor VIII demonstrated the cells to be viable ECs. On the outer surface of the grafts only scattered ECs were seen under scanning electron microscopy. Staining against collagen IV and laminin revealed only a thin positive reaction beneath the EC layer. All three groups showed EC attachment on the valve leaflets, but in group 1 (resting phase, 30 min) and group 3 (resting phase, 60 min), seeding did not result in a homogenous cell layer. The defect area reached 30% of the specimen area in group 1 and 20% in group 3. In group 3, many ECs had attached to the bottom of the incubator, which was not the case in the other groups. This indicated less attachment to the grafts and incomplete re-suspension during the rotation phases. After the final resting phase the EC layer on the leaflets of group 3 reached confluence, but not in group 1. In contrast, in group 2 (resting phase, 45 min), homogenous cell seeding on both leaflet surfaces was achieved initially and the cells grew to confluence after the resting period. In group 4, initial EC adhesion was 74%, and the cells also formed confluent cell layers on the leaflets. In group 5, fibroblast seeding resulted in an initial adhesion of 79%. Scanning electron microscopy revealed a confluent fibroblast layer on the luminal surface as well as on the outer surface (Fig 7). The cells already started to build collagen fibers. Endothelial cell seeding onto these preseeded grafts resulted in an initial adhesion of 78%. All taken specimens showed a homogenous EC layer on the free walls as well as on the leaflets (Fig 8A, 8B). Again, viability testing with staining against CD31 and factor VIII indicated the EC to be viable. Immunohistochemical
NEW TECHNOLOGY GULBINS ET AL DEVELOPMENT OF A SEEDING DEVICE AND PROCEDURE
2123
NEW TECHNOLOGY
Ann Thorac Surg 2005;79:2119 –26
Fig 8. (A) Scanning electron microscopic examination (magnification, ⫻500) showing a leaflet of one of the grafts of group 5 after initial endothelial cell seeding. Also shows no defects within a homogenous cell layer. (B) Group 5, immunohistochemical staining for factor VIII (magnification, ⫻100, peroxidase reaction). Positive reaction (dark band) on the luminal surface indicating viable endothelial cells.
staining of group 5 specimen with antibodies against laminin and collagen IV showed a positive band on the luminal surface (Fig 9A). The reaction was clearly stronger positive compared to group 4 (Fig 9B). This indicated an increased synthesis of these components of the basement membrane. None of the specimens showed signs of an overgrowth of the fibroblasts.
Comment Endothelial cell seeding on biological prosthesis has been described several times before [4 – 8]. Confluent cell layers on the luminal surface of vascular prostheses, veins, and arteries were reproducibly achieved [9 –17]. However, on complex 3-dimensional objects it was more difficult to achieve a confluent EC layer. In our experiments we seeded human cells on homografts. Several authors reported successful EC seeding
Fig 9. (A) Group 5, immunohistochemical staining for collagen IV (magnification, ⫻100, peroxidase reaction) revealed strong positive reaction (dark band) within the luminal surface, indicating synthesis of this component of the extracellular matrix. Pre-seeding with autologous fibroblasts allowed both cell types to synthesize their specific mixture of extracellular matrix proteins, which was therefore advantageous compared with pre-coating with singular components such as fibronectin or laminin. (B) Group 4, immunohistochemical staining for collagen IV (magnification, ⫻100, peroxidase reaction). The slight positive reaction on the luminal surface (dark band) was much weaker compared with group 5 (Fig 9A).
on cryopreserved homografts [6, 7]. With regard to their physiologic shape and function, and their excellent hemodynamics of these grafts, they can be regarded as the most physiologic valve prostheses. In addition, these allografts would represent a good scaffold, because only rare donor cells survive the sterilization and cryopreservation process. The maintainance of the donor’s endothelium would be disadvantageous with regard to the immunologic reaction of the recipient. The rejection of homografts was triggered by donor cells [18 –20] thus leading to degeneration during follow-up. Because ECs also present antigens they were the first target for any immunologic reaction and were subsequently destroyed. This was even aggravated because homograft is not commonly transplanted in accordance with the ABO blood group system.
2124
NEW TECHNOLOGY
NEW TECHNOLOGY GULBINS ET AL DEVELOPMENT OF A SEEDING DEVICE AND PROCEDURE
Ann Thorac Surg 2005;79:2119 –26
When valve containing aortic roots were subjects of cell seeding, we supposed seeding on the leaflets to be the most difficult. As rotation has been described to be effective for luminal cell seeding on different vessels [11, 15, 16], our device was developed to rotate around the axial as well as the sagittal axis. The rotation phases were defined in a way that for each resting phase the root inside stopped in another position, thus exposing different parts of the surface for cell attachment. During rotation, the tube inside moved longitudinally, causing slight movements of the leaflets, allowing cells to pass the valve to avoid entrapping of the cells in two compartments on each side of the valve, possibly resulting in an inhomogeneous cell seeding. The rotation speed influenced the success of cell seeding, too. In our experiments with a transparent incubator, the slow rotation speed of 1 round/min did not cause sufficient motions of the leaflets, whereas both faster speed settings did. For EC seeding on various surfaces, different incubation times and inoculation densities have been reported [13, 15, 17]. Seeding times among 20 and 90 minutes were used, depending on the subject of endothelialization. Kent and colleagues [15] reported improved EC adhesion after 90 minutes of seeding when compared with 20 minutes or less. We set the resting period to 45 minutes, corresponding to the seeding time of the other groups, whereas the total incubation period was 24 hours. This long time was chosen due to the fact that during each rotation nonadhesive cells were re-suspended, another part of the surface was exposed optimally, and therefore a new seeding process was initiated. The duration of the resting phases played an important role on the effectiveness of EC seeding as previously reported [14 –17]. During these periods, cells were allowed to settle down and attach to the exposed surface. A resting phase of 30 minutes did not result in a homogenous cell layer, especially not on the leaflets. Therefore we assumed that 30 minutes was not long enough for a sufficient cell number to attach before being resuspended again during the rotation phase of the incubator. This agrees with the results of Kent and colleagues [15]. The importance of the duration of the resting phases could be explained by the two factors influencing cell attachment: (1) sedimentation (passive) and (2) adhesion by pseudopodia and adhesion molecules (active). The suspended cells required enough time for sedimentation onto the graft and then to establish active adhesion. Cells that had sedimented, but had not formed a stable active adhesion were re-suspended during the next rotation phase. This was the reason for the poor results with the short resting period. On the other hand, a resting phase of 60 minutes also did not result in an initial homogenous cell layer. However, EC cells could be grown to confluence after 6 days on these grafts. In these cases the relation between resting and rotation phases was obviously not optimal. The grafts rotated only 24 times in this setting, compared with 32 times when the resting phase was set to 45 minutes. Therefore, the exposed areas did not change
often enough and thus not all parts of the surface, especially of the leaflets, were exposed for cell attachment. This was supported by the finding of ECs on the bottom of the incubator. These cells obviously had enough time to sediment to the bottom and adhere, and therefore they were lost for the following seeding periods. However, cell density was obviously high enough to allow ECs to grow to confluence after 6 days. Endothelial cell seeding and culturing under static conditions led to changes in the protein synthesis of these cells. Therefore, the construction and use of bioreactors were reported [21–23] for EC seeding under continuos flow conditions. These devices are more sophisticated than the presented one, but also much more demanding in use and probably also more susceptible to bacterial contamination. For cell seeding onto polymeric valve scaffolds, especially biodegradable polymers, seeding under flow is necessary to achieve the synthesis of structural collagenes to obtain mechanical stability. In contrast, when homografts were used as scaffold for EC seeding, this stability was already ensured by the grafts themselves, and the synthesis of structural proteins such as collagene III was not that importance. However, when ECs were seeded under static conditions, they maintain their ability to react to shear stress [24], thus keeping the EC layer stable. When cells were seeded under flow, the flow ensured that the whole surface was reached by the cells. With the presented device, this was achieved by the movements of the inlay and the fact that the graft was always stopped in a different position, thus exposing another part of the graft for cell attachment. The main difference of the presented device with other bioreactors was that continuous flow was not required for complete cell covering. The potential of ECs for cell-matrix interaction influences the regenerative process after injury as well as the result of re-endothelialization by seeding [12, 17]. Therefore, preseeding with autologous fibroblasts should not only provide better initial attachment but also improve the regeneration processes within the EC layer. This agreed with the results of previous studies [10, 11] that proved the extracellular matrix built by fibroblasts was an ideal substrate for EC seeding. This made additional pre-coating with fibronectin, laminin, or other components of the extracellular matrix unnecessary. The basement membrane of ECs consists mainly of collagen IV, entactin, a heparin sulfate proteoglycan, laminin, and fibronectin [25]. Fibroblasts isolated from different tissues synthetize different types of collagen. We used vascular fibroblasts, as they were supposed to build vascular collagen, especially type IV. The importance of the extracellular matrix for EC adhesion and the resistance of such a cell layer to shear stress has been shown by different authors [9, 10, 12, 13]. As the fibroblasts in group 5 begin to build collagen fibers, their adhesion to the surface should be sufficient. In addition, they were supposed to build the physiologic extracellular matrix for EC adhesion. This was demonstrated by the more intense stainings for collagen IV and laminin on those valves preseeded with fibroblasts.
NEW TECHNOLOGY GULBINS ET AL DEVELOPMENT OF A SEEDING DEVICE AND PROCEDURE
The explanation for the changes in the synthesis of extracellular matrix proteins (such as laminin and collagen IV) was that these proteins were synthesized not only by ECs but also by vascular fibroblasts [10 –12, 25]. Because of the limited synthesis of these components of the basement membrane in groups 1 to 4, the intention was to increase the synthesis of extracellular matrix by adding autologous fibroblasts. The results showed that pre-seeding with fibroblasts indeed increased the synthesis of collagen IV and laminin, thus improving EC adhesion. The resistance to shear stress should also be improved by a functioning basement membrane [11, 12]. One major mechanism of homograft degeneration was the immunologic reaction of the recipient to the graft [18 –20]. Covering the luminal surface with ECs may avoid direct contact of immunocompetent cells with the antigens of the recipient. This could contribute to improved durability of these grafts. Another major issue in biocompatibility is the thrombogenicity of the graft’s surface. Although this was reported to be low, some minor thromboembolic events also occur in patients with homografts. A confluent EC layer could reduce this risk further, as it was reported for vascular grafts [26, 27]. With the presented technology, seeding of ECs onto polyurethane vascular prostheses was also possible [28]. Although the feasibility of clinical application of such endothelialized homografts was already shown [29], further experimental and clinical research is necessary to evaluate and establish these technologies in clinical practice. In conclusion, the presented device allowed for reproducible seeding of human ECs and fibroblasts on cryopreserved valve containing homografts. Both the initially confluent cell layers and the sustained viability of the seeded cells after 6 days make further experimental and clinical studies with such seeded grafts possible.
5. Bader A, Schilling T, Teebken OE, et al. Tissue engineering of heart valves— human endothelial cell seeding of detergent acellularized porcine valves. Eur J Cardiothorac Surg 1998;14(3):279 – 84. 6. Fischlein T, Fasol R. In vitro endothelialization of bioprosthetic heart valves. J Heart Valve Dis 1996;5(1):58 – 65. 7. Fischlein T, Lehner G, Lante W, et al. Endothelialization of cardiac valve bioprostheses. Int J Artific Org 1994;17:345–52. 8. Bengtsson L, Radegran K, Haegerstrand A. In vitro endothelialisation of commercially available heart valve bioprostheses with cultured adult human cells. Eur J Cardiothorac Surg 1993;7:393– 8. 9. Walluschek KP, Steinhoff G, Haverich A. Endothelial cell seeding on native vascular surfaces. Eur J Vasc Endovasc Surg 1996;11(3):290 –303. 10. Lalka SG, Oelker LM, Malone JM, et al. Acellular vascular matrix: a natural endothelial cell substrate. Ann Vasc Surg 1989;3(2):108 –17. 11. Lee YS, Park DK, Kim YB, Seo JW, Lee KB, Min BG. Endothelial cell seeding onto the extracellular matrix of fibroblasts for the development of a small diameter polyurethane vessel. ASAIO J 1993;39(3):M740 –5. 12. Grant DS, Kleinmann HK, Martin GR. The role of basement membranes in vascular development. Ann NY Acad Sci 1990;588:61–72. 13. Sipehia R, Martucci G, Lipscombe J. Transplantation of human endothelial cell monolayer on artificial vascular prosthesis: the effect of growth—support surface chemistry, cell seeding density, ECM protein coating, and growth factors. Artif Cells Blood Substit Immobil Biotechnol 1996; 24(1):51– 63. 14. Bowlin GL, Rittgers SE. Electrostatic endothelial cell transplantation within small-diameter (⬍ 6mm) vascular prostheses: a prototype apparatus and procedure. Cell Transplant 1997;6(6):631–7. 15. Kent KC, Oshima A, Whittemore AD. Optimal seeding conditions for human endothelial cells. Ann Vasc Surg 1992;6(3):258 – 64. 16. Mazzucotelli JP, Roudiere JL, Bernex F, Bertrand P, Leandri J, Loisance D. A new device for endothelial cell seeding of a small-caliber vascular prosthesis. Artif Organs 1993;17(9): 787–90. 17. Thompson MM, Budd JS, Eady SL, et al. Effect of seeding time and density on endothelial cell attachment to damaged vascular surfaces. Br J Surg 1993;80(3):359 – 62. 18. Yankah AC, Wottge HU. Allograft conduit wall calcification in a model of chronic arterial graft rejection. J Card Surg 1997;12(2):86 –92. 19. Moustapha A, Ross DB, Bittira B, et al. Aortic valve grafts in the rat: evidence for rejection. J Thorac Cardiovasc Surg 1997;114(6):891–902. 20. Smith JD, Ogino H, Hunt D, Laylor RM, Rose ML, Yacoub MH. Humoral response to human aortic valve homografts. Ann Thorac Surg 1995;60:S127–30. 21. Sodian R, Sperling JS, Martin DP, et al. Fabrication of a trileaflet heart valve scaffold from a polyhydroxyalkanoate biopolyester for use in tissue engineering. Tissue Eng 2000; 6:183– 8. 22. Sodian R, Lemke T, Fritsche C, et al. Tissue-engineering bioreactors: a new combined cell-seeding and perfusion system for vascular tissue engineering. Tissue Eng 2002;8(5): 863–70. 23. Jockenhoevel S, Zund G, Hoerstrup SP, Schnell A, Turina M. Cardiovascular tissue engineering: a new laminar flow chamber for in vitro improvement of mechanical tissue properties. ASAIO J 2002;48(1):8 –11. 24. Gulbins H, Pritisanac A, Petzold R, et al. A low-flow adaption phase improves sheer-stress resistance of artificially seeded endothelial cells. Thorac Cardiovasc Surg 2004 (accepted for publication).
Disclosures and Freedom of Investigation For the studies published in this article, only the official clinical research funds of the department of cardiac surgery, University Hospital Grosshadern (Munich, Germany) were used. The homografts were taken from our in-house homograft bank (responsible, co-author AU). All authors had full control of the design of the study, methods used, outcome measurements, analysis of data, and production of the written article.
References 1. Thilo DG, Muller-Kusel S, Heinrich D, Kaufer I, Weiss E. Isolation of human venous endothelial cells by different proteases. Artery 1980;8(3):259 – 66. 2. Jarrel B, Shapiro S, Levine E. Human adult endothelial growth in culture. J Vasc Surg 1984;1:623–5. 3. Haegerstrand A, Gillis C, Bengtsson L. Serial cultivation of adult human endothelium from the great saphenous vein. J Vasc Surg 1992;16:280 –5. 4. Gulbins H, Goldemund A, Anderson I, et al. Preseeding with autologous fibroblasts improves endothelialization of glutaraldehyde-fixed porcine aortic valves. J Thorac Cardiovasc Surg 2003;125:592– 601.
NEW TECHNOLOGY
2125
Ann Thorac Surg 2005;79:2119 –26
2126
NEW TECHNOLOGY GULBINS ET AL DEVELOPMENT OF A SEEDING DEVICE AND PROCEDURE
Ann Thorac Surg 2005;79:2119 –26
25. Darnell J, Lodish H, Baltimore D. Molecular cell biology, 2nd ed. Scientific American Books, 1990: 914-32. 26. Ortenwall P, Wadenvik H, Kutti J, Risberg B. Endothelial cell seeding reduces thrombogenicity of Dacron grafts in humans. J Vasc Surg 1990;11(3):403–10. 27. Shindo S, Takagi A, Whittemore AD. Improved patency of collagen-impregnated grafts after in vitro autogenous endothelial cell seeding. J Vasc Surg 1987;6(4):325–32. 28. Gulbins H, Goldemund A, Uhlig A, Meiser B, Reichart B. Implantation of an autologously endothelialized homograft. J Thorac Cardiovasc Surg 2003;126:890 –1.
29. Gulbins H, Dauner M, Petzold R, et al. Development of an artificial vessel lined with human vascular cells. J Thorac Cardiovasc Surg 2004;128:372–7.
Disclaimer The Society of Thoracic Surgeons, the Southern Thoracic Surgical Association, and The Annals of Thoracic Surgery neither endorse nor discourage use of the new technology described in this article.
NEW TECHNOLOGY
Purpose. Complete covering of an artificial valvular scaffold with endothelial cells may prevent thromboembolic complications and lead to an excellent biocompatibility. For this purpose, we developed a seeding device for reproducible cell seeding on valve containing aortic roots. Description. Human endothelial cells and fibroblasts were obtained from saphenous vein pieces. Cryopreserved aortic roots (n ⴝ 25) were put into an especially developed tube, set on a rotator, and incubated with the cell suspension. The device rotated in two axes (sagittal and axial), ensuring slight movements of the leaflets. The rotation alternated with resting periods, allowing cell attachment to the surface. Different resting periods were tested (groups 1, 2, and 3 were 30, 45, and 60 min, respectively; n ⴝ 5 each). Total incubation time was 24 hours followed by further culturing for 6 days. In two further groups (groups 4 and 5; n ⴝ 5 each), a modified inlay was used to allow the cell suspension to flow around the entire graft. In group 4 the grafts were again incubated with human endothelial cells; however, in group 5 pre-seeding with autologous fibroblasts was done in addition. Immunohistochemical staining with antibodies against factor VIII, CD31, laminin, collagen IV, and CD90 were done, and scanning electron microscopy was done after initial seeding and after 6 days in culture. Evaluation. Seeding resulted in homogenous cell layers on the luminal surface of the free walls in all groups. With resting periods of 45 minutes, these results were also obtained on the leaflets, whereas the other resting times resulted in defects of the endothelial cell layer on the cusps. After 6 days under culture conditions, the endothelial cell layers were confluent and viable, with the exception of the leaflets in group 1. With the modified inlay (groups 4 and 5), confluent cell layers were also achieved on the outer surface. In group 5 pre-seeding with autologous fibroblasts resulted in enhanced synthesis of extracellular matrix proteins, as was demonstrated with immunohistochemical staining for collagen IV and laminin. Conclusions. With this newly developed seeding device, confluent cell layers on valve containing aortic roots were reproducibly achieved. The technique enables further experimental research and even clinical application. (Ann Thorac Surg 2005;79:2119 –26) © 2005 by The Society of Thoracic Surgeons
T
he lack of endothelial cell (EC) covering of mechanical or biological valve prostheses under in vivo conditions in humans is one major shortcoming of these For editorial comment, see page 1831 substitutes. A confluent EC layer could increase biocompatibility of biological valve prostheses by reducing thrombogenicity. Human ECs can be grown to confluence on various tissues, but seeding on 3-dimensional Accepted for publication May 6, 2004. Address reprint requests to Dr Gulbins, Department of Cardiac Surgery, University Hospital Ulm, Steinhövelstr 9, D-89070 Ulm, Germany; e-mail: [email protected]..
© 2005 by The Society of Thoracic Surgeons Published by Elsevier Inc
objects proved to be more difficult. The purpose of our study was to develop a device for reproducible cell seeding on valve containing aortic mini-roots. This would enable cell seeding on complex 3-dimensional scaffolds to develop a custom-made heart valve seeded with autologous cells of the patient.
Technology Cell Culture Human ECs and fibroblasts were isolated from saphenous vein pieces as previously described [1– 4]. The vein pieces were leftover from routine aorto-coronary bypass 0003-4975/05/$30.00 doi:10.1016/j.athoracsur.2004.05.085
NEW TECHNOLOGY
Department of Cardiac Surgery, University Hospital Grosshadern, Munich, Germany
2120
NEW TECHNOLOGY GULBINS ET AL DEVELOPMENT OF A SEEDING DEVICE AND PROCEDURE
Ann Thorac Surg 2005;79:2119 –26
NEW TECHNOLOGY
Fig 1. Homograft, fixed on the incubator tube and placed inside the incubator viewed through the graft in the direction of blood flow. With this tube, the cell suspension reached only the luminal surface of the graft, which was shorter than the incubator, thus allowing longitudinal movements of the graft inside during the rotation. This ensured an exchange of the cell suspension between both sides of the leaflets.
operations from which the patients had given their informed consent for these pieces to be used in the laboratory. The local ethics committee approved the anonymous use of the saphenous vein pieces for this experimental study. The vein pieces were cannulated, rinsed with buffered medium, and incubated with 0.1% collagenase (Worthington, CA) for 20 minutes at 37°C and 5% CO2. Thereafter, the collagenase reaction was stopped, the cell suspension was centrifuged at 1,000 U/min for 10 minutes and the cell pellet was resuspended in EC growth medium (Promocell, Heidelberg, Germany), and was plated on culture dishes. For fibroblast isolation, the same vein pieces were again filled with 0.1% collagenase and incubated for 30 minutes. This procedure was repeated twice. Again, the cell suspension was centrifuged at 1000 U/min for 10 minutes, the pellet was re-suspended in fibroblast growth medium (Promocell, Heidelberg, Germany), and was plated on culture dishes.
Fig 2. The incubator was fixed on the self-developed rotation device. Shown is the new tube placed inside and the closing mechanism. The incubator was filled with 120 mL cell suspension and the air was removed through the 3-way connector. The two controls at the base were used to choose the speed and time settings.
incubated with M-199 (supplemented with 5% fetal calf serum) for 1 day. For cell seeding, a special device (rotation incubator) was developed in cooperation with the technical staff. The mini-roots were fixed on a tube (Fig 1), put into a flask (incubator), and set on the rotator (Fig 2). The internal volume of this incubator was 100 mL. The device rotated the tube around the axial and sagittal axes (Fig 3).
Technique Seeding on Aortic Roots Aortic roots were obtained from explanted hearts during heart transplantations or from multiorgan donors whose hearts were not used for transplantation. After sterilization in an antibiotic cocktail (ciprofloxacin, amikacin, vancomycin, metronidazol, and amphotericin B) for 3 days, the grafts were cryopreserved in M-199 with addition of dimethyl sulfoxide, and were then stored at ⫺196°C in liquid nitrogen for 45 days (range, 15 to 102 days). For seeding, the grafts were rewarmed in a sterile water bed, flushed with M-199 for 3 hours, and then
Fig 3. Rotations and other movements of the tube inside the incubator during the rotation phase (the tube rotated axially and longitudinally, whereas the whole incubator was moved counterclockwise). These movements re-suspended those cells that had not attached to the graft during the resting phase.
NEW TECHNOLOGY GULBINS ET AL DEVELOPMENT OF A SEEDING DEVICE AND PROCEDURE
Fig 4. Homograft fixed to the new tube (groups 4 and 5). This new tube design allowed the grafts to be washed around by the cell suspension. This ensured that the outer surface was also cell coated, especially with fibroblasts.
When the incubator moved around the sagittal axis, the tube inside moved longitudinally, causing a slow motion of the valve leaflets. This also allowed an exchange of the cell suspension of both sides of the valve. The resting periods and rotation periods could be set individually. One transparent incubator was built to study the technical settings. An untreated homograft was placed inside, and the motions of the valve leaflets were examined with different rotation speed settings (1 round/min, 2.5 rounds/min, and 4 rounds/min). For EC seeding, the incubator was filled with EC medium supplemented with 5% fetal calf serum. Endothelial cells were trypsinized as previously described and were re-suspended in 10 mL phosphate buffered saline. The EC suspension (inoculation density 5.5 ⫻ 105 cells/cm2) was then transferred to the rotation incubator. The total incubation time of the aortic roots lasted for 24 hours. During incubation, resting and rotation periods were alternated. The resting phases were set at 30 minutes for group 1 (n ⫽ 5), 45 minutes for group 2 (n ⫽ 5), and 60 minutes for group 3 (n ⫽ 5), resulting in 44, 30, and 23 rotation cycles. During the rotations, the grafts were turned around 800° in all groups. The medium was then removed and nonadherent cells were identified using trypan blue staining and
2121
were counted using the Neubauer chamber. Initial cell adhesion was defined as the difference between the initial cell number of the cell suspension and these remaining, nonadherent cells. The values were given in a percentage of the initial cell number. For two more groups, the design of the tube was changed to allow cell seeding also on the outer surface. In this new tube (Fig 4), the mini-roots were fixed only with few sutures at their ends, exposing the outer as well as the luminal surface to the cells for attachment. In group 4, only ECs were seeded on the grafts as previously described. For group 5, the incubator was filled with fibroblast growth medium supplemented with 10% fetal calf serum. Human fibroblasts were trypsinized, washed with phosphate buffered saline, centrifuged, and the re-suspension volume was reduced to 10 mL. With an inoculation density of 7.7 ⫻ 105 cells/cm2, the fibroblast cell suspension was added to the incubator. Again, the same settings as previously described were used. Thereafter EC seeding was done as previously described for group 4. All seeded roots (groups 1 to 5) (see Table 1) had a further culturing phase of 6 days to develop confluent cell layers. For this, the tube was placed into a culture glass filled with 80 mL EC medium at 37°C at 5% CO2 atmosphere.
Immunohistochemical Staining A specimen of the aortic wall and valve leaflets were frozen and stored in liquid nitrogen. Eight mm sections were stained using monoclonal antibodies against factor VIII, CD31 (Dako, Hamburg, Germany), and collagen IV (Sigma, Deisnhofen, Germany). For factor VIII staining, a polyclonal antibody against antirabbit IgG (Immundiagnostik, Bensheim, Germany) was used. Collagen IV and CD31 stainings were counterstained with a polyclonal antibody against mouse IgG (Chemicon Int, Inc, CA). After incubation with the primary antibody against factor VIII, CD31, and collagen IV, an antimouse IgG antibody (Dako, Hamburg, Germany) was added and stained with 3-amino-9-ethyl carbazol (Dako, Hamburg, Germany). The results were validated with negative control stainings without the primary antibody.
Scanning Electron Microscopy Only cell layers with the typical cobblestone morphology were accepted as EC layers. For analysis, 10 visual fields of each specimen were evaluated at 1,000⫻ magnification. Aortic and ventricular surfaces of the leaflets were evaluated. The endothelial cell covering was assessed
Table 1. Definition of the Five Different Groupsa
Resting phase Fibroblast seeding Endothelial cell seeding a
Group 1
Group 2
Group 3
Group 4
Group 5
30 min ⫺ ⫹
45 min ⫺ ⫹
60 min ⫺ ⫹
45 min ⫺ ⫹
45 min ⫹ ⫹
In groups 4 and 5, a modified inlay was used to allow the cell suspension to flow around the entire graft. In groups 1 to 3 the length of the resting phases differed. Group 4 served as a control group for group 5, in which additional pre-seeding with fibroblasts was tested.
NEW TECHNOLOGY
Ann Thorac Surg 2005;79:2119 –26
2122
NEW TECHNOLOGY GULBINS ET AL DEVELOPMENT OF A SEEDING DEVICE AND PROCEDURE
NEW TECHNOLOGY
Fig 5. Scanning electron microscopy (magnification, ⫻200) of the luminal surface of the free wall of a homograft after cryopreservation before endothelial cell seeding. No endothelial cells were seen, but interstitial fibers were seen.
semiquantitatively by two investigators and was classified as 100%, 75%, 50%, or 25%. A mean value of 95% and greater was believed to represent a confluent EC layer.
Results The results of cryopreserved homografts showed a nearly complete loss of the ECs on all specimen taken before cell seeding (Fig 5). Only collagen fibers and fibrocytes were seen. Although it was possible to culture fibroblasts from all mini-roots (n ⫽ 25), viable ECs could only be isolated from three mini-root pieces in a very small number. The first tests with the transparent incubator showed nearly no movements of the leaflets at a rotation speed of 1 round/min, whereas with both faster speed settings valve motions were seen. Therefore we chose 2.5 rounds/ min for further experiments as the standard setting. In groups 1, 2, and 3, incubation with ECs resulted in an initial cell adhesion of 68%, 75%, and 72%, respec-
Fig 6. Graphical demonstration of the seeding results in the different groups.
Ann Thorac Surg 2005;79:2119 –26
Fig 7. Scanning electron microscopic examination (magnification, ⫻500) of group 5 after fibroblast seeding. The fibroblasts formed a confluent cell layer. The cells could be identified by their typical long shape.
tively (Fig 6). On the luminal surface of the free walls, a confluent EC layer was seen. Only rare defects were detected, concerning less than 1% of the specimen areas. After the final culturing phase of 6 days, confluent cell layers were documented on the free walls in all cases. Staining for CD31 and factor VIII demonstrated the cells to be viable ECs. On the outer surface of the grafts only scattered ECs were seen under scanning electron microscopy. Staining against collagen IV and laminin revealed only a thin positive reaction beneath the EC layer. All three groups showed EC attachment on the valve leaflets, but in group 1 (resting phase, 30 min) and group 3 (resting phase, 60 min), seeding did not result in a homogenous cell layer. The defect area reached 30% of the specimen area in group 1 and 20% in group 3. In group 3, many ECs had attached to the bottom of the incubator, which was not the case in the other groups. This indicated less attachment to the grafts and incomplete re-suspension during the rotation phases. After the final resting phase the EC layer on the leaflets of group 3 reached confluence, but not in group 1. In contrast, in group 2 (resting phase, 45 min), homogenous cell seeding on both leaflet surfaces was achieved initially and the cells grew to confluence after the resting period. In group 4, initial EC adhesion was 74%, and the cells also formed confluent cell layers on the leaflets. In group 5, fibroblast seeding resulted in an initial adhesion of 79%. Scanning electron microscopy revealed a confluent fibroblast layer on the luminal surface as well as on the outer surface (Fig 7). The cells already started to build collagen fibers. Endothelial cell seeding onto these preseeded grafts resulted in an initial adhesion of 78%. All taken specimens showed a homogenous EC layer on the free walls as well as on the leaflets (Fig 8A, 8B). Again, viability testing with staining against CD31 and factor VIII indicated the EC to be viable. Immunohistochemical
NEW TECHNOLOGY GULBINS ET AL DEVELOPMENT OF A SEEDING DEVICE AND PROCEDURE
2123
NEW TECHNOLOGY
Ann Thorac Surg 2005;79:2119 –26
Fig 8. (A) Scanning electron microscopic examination (magnification, ⫻500) showing a leaflet of one of the grafts of group 5 after initial endothelial cell seeding. Also shows no defects within a homogenous cell layer. (B) Group 5, immunohistochemical staining for factor VIII (magnification, ⫻100, peroxidase reaction). Positive reaction (dark band) on the luminal surface indicating viable endothelial cells.
staining of group 5 specimen with antibodies against laminin and collagen IV showed a positive band on the luminal surface (Fig 9A). The reaction was clearly stronger positive compared to group 4 (Fig 9B). This indicated an increased synthesis of these components of the basement membrane. None of the specimens showed signs of an overgrowth of the fibroblasts.
Comment Endothelial cell seeding on biological prosthesis has been described several times before [4 – 8]. Confluent cell layers on the luminal surface of vascular prostheses, veins, and arteries were reproducibly achieved [9 –17]. However, on complex 3-dimensional objects it was more difficult to achieve a confluent EC layer. In our experiments we seeded human cells on homografts. Several authors reported successful EC seeding
Fig 9. (A) Group 5, immunohistochemical staining for collagen IV (magnification, ⫻100, peroxidase reaction) revealed strong positive reaction (dark band) within the luminal surface, indicating synthesis of this component of the extracellular matrix. Pre-seeding with autologous fibroblasts allowed both cell types to synthesize their specific mixture of extracellular matrix proteins, which was therefore advantageous compared with pre-coating with singular components such as fibronectin or laminin. (B) Group 4, immunohistochemical staining for collagen IV (magnification, ⫻100, peroxidase reaction). The slight positive reaction on the luminal surface (dark band) was much weaker compared with group 5 (Fig 9A).
on cryopreserved homografts [6, 7]. With regard to their physiologic shape and function, and their excellent hemodynamics of these grafts, they can be regarded as the most physiologic valve prostheses. In addition, these allografts would represent a good scaffold, because only rare donor cells survive the sterilization and cryopreservation process. The maintainance of the donor’s endothelium would be disadvantageous with regard to the immunologic reaction of the recipient. The rejection of homografts was triggered by donor cells [18 –20] thus leading to degeneration during follow-up. Because ECs also present antigens they were the first target for any immunologic reaction and were subsequently destroyed. This was even aggravated because homograft is not commonly transplanted in accordance with the ABO blood group system.
2124
NEW TECHNOLOGY
NEW TECHNOLOGY GULBINS ET AL DEVELOPMENT OF A SEEDING DEVICE AND PROCEDURE
Ann Thorac Surg 2005;79:2119 –26
When valve containing aortic roots were subjects of cell seeding, we supposed seeding on the leaflets to be the most difficult. As rotation has been described to be effective for luminal cell seeding on different vessels [11, 15, 16], our device was developed to rotate around the axial as well as the sagittal axis. The rotation phases were defined in a way that for each resting phase the root inside stopped in another position, thus exposing different parts of the surface for cell attachment. During rotation, the tube inside moved longitudinally, causing slight movements of the leaflets, allowing cells to pass the valve to avoid entrapping of the cells in two compartments on each side of the valve, possibly resulting in an inhomogeneous cell seeding. The rotation speed influenced the success of cell seeding, too. In our experiments with a transparent incubator, the slow rotation speed of 1 round/min did not cause sufficient motions of the leaflets, whereas both faster speed settings did. For EC seeding on various surfaces, different incubation times and inoculation densities have been reported [13, 15, 17]. Seeding times among 20 and 90 minutes were used, depending on the subject of endothelialization. Kent and colleagues [15] reported improved EC adhesion after 90 minutes of seeding when compared with 20 minutes or less. We set the resting period to 45 minutes, corresponding to the seeding time of the other groups, whereas the total incubation period was 24 hours. This long time was chosen due to the fact that during each rotation nonadhesive cells were re-suspended, another part of the surface was exposed optimally, and therefore a new seeding process was initiated. The duration of the resting phases played an important role on the effectiveness of EC seeding as previously reported [14 –17]. During these periods, cells were allowed to settle down and attach to the exposed surface. A resting phase of 30 minutes did not result in a homogenous cell layer, especially not on the leaflets. Therefore we assumed that 30 minutes was not long enough for a sufficient cell number to attach before being resuspended again during the rotation phase of the incubator. This agrees with the results of Kent and colleagues [15]. The importance of the duration of the resting phases could be explained by the two factors influencing cell attachment: (1) sedimentation (passive) and (2) adhesion by pseudopodia and adhesion molecules (active). The suspended cells required enough time for sedimentation onto the graft and then to establish active adhesion. Cells that had sedimented, but had not formed a stable active adhesion were re-suspended during the next rotation phase. This was the reason for the poor results with the short resting period. On the other hand, a resting phase of 60 minutes also did not result in an initial homogenous cell layer. However, EC cells could be grown to confluence after 6 days on these grafts. In these cases the relation between resting and rotation phases was obviously not optimal. The grafts rotated only 24 times in this setting, compared with 32 times when the resting phase was set to 45 minutes. Therefore, the exposed areas did not change
often enough and thus not all parts of the surface, especially of the leaflets, were exposed for cell attachment. This was supported by the finding of ECs on the bottom of the incubator. These cells obviously had enough time to sediment to the bottom and adhere, and therefore they were lost for the following seeding periods. However, cell density was obviously high enough to allow ECs to grow to confluence after 6 days. Endothelial cell seeding and culturing under static conditions led to changes in the protein synthesis of these cells. Therefore, the construction and use of bioreactors were reported [21–23] for EC seeding under continuos flow conditions. These devices are more sophisticated than the presented one, but also much more demanding in use and probably also more susceptible to bacterial contamination. For cell seeding onto polymeric valve scaffolds, especially biodegradable polymers, seeding under flow is necessary to achieve the synthesis of structural collagenes to obtain mechanical stability. In contrast, when homografts were used as scaffold for EC seeding, this stability was already ensured by the grafts themselves, and the synthesis of structural proteins such as collagene III was not that importance. However, when ECs were seeded under static conditions, they maintain their ability to react to shear stress [24], thus keeping the EC layer stable. When cells were seeded under flow, the flow ensured that the whole surface was reached by the cells. With the presented device, this was achieved by the movements of the inlay and the fact that the graft was always stopped in a different position, thus exposing another part of the graft for cell attachment. The main difference of the presented device with other bioreactors was that continuous flow was not required for complete cell covering. The potential of ECs for cell-matrix interaction influences the regenerative process after injury as well as the result of re-endothelialization by seeding [12, 17]. Therefore, preseeding with autologous fibroblasts should not only provide better initial attachment but also improve the regeneration processes within the EC layer. This agreed with the results of previous studies [10, 11] that proved the extracellular matrix built by fibroblasts was an ideal substrate for EC seeding. This made additional pre-coating with fibronectin, laminin, or other components of the extracellular matrix unnecessary. The basement membrane of ECs consists mainly of collagen IV, entactin, a heparin sulfate proteoglycan, laminin, and fibronectin [25]. Fibroblasts isolated from different tissues synthetize different types of collagen. We used vascular fibroblasts, as they were supposed to build vascular collagen, especially type IV. The importance of the extracellular matrix for EC adhesion and the resistance of such a cell layer to shear stress has been shown by different authors [9, 10, 12, 13]. As the fibroblasts in group 5 begin to build collagen fibers, their adhesion to the surface should be sufficient. In addition, they were supposed to build the physiologic extracellular matrix for EC adhesion. This was demonstrated by the more intense stainings for collagen IV and laminin on those valves preseeded with fibroblasts.
NEW TECHNOLOGY GULBINS ET AL DEVELOPMENT OF A SEEDING DEVICE AND PROCEDURE
The explanation for the changes in the synthesis of extracellular matrix proteins (such as laminin and collagen IV) was that these proteins were synthesized not only by ECs but also by vascular fibroblasts [10 –12, 25]. Because of the limited synthesis of these components of the basement membrane in groups 1 to 4, the intention was to increase the synthesis of extracellular matrix by adding autologous fibroblasts. The results showed that pre-seeding with fibroblasts indeed increased the synthesis of collagen IV and laminin, thus improving EC adhesion. The resistance to shear stress should also be improved by a functioning basement membrane [11, 12]. One major mechanism of homograft degeneration was the immunologic reaction of the recipient to the graft [18 –20]. Covering the luminal surface with ECs may avoid direct contact of immunocompetent cells with the antigens of the recipient. This could contribute to improved durability of these grafts. Another major issue in biocompatibility is the thrombogenicity of the graft’s surface. Although this was reported to be low, some minor thromboembolic events also occur in patients with homografts. A confluent EC layer could reduce this risk further, as it was reported for vascular grafts [26, 27]. With the presented technology, seeding of ECs onto polyurethane vascular prostheses was also possible [28]. Although the feasibility of clinical application of such endothelialized homografts was already shown [29], further experimental and clinical research is necessary to evaluate and establish these technologies in clinical practice. In conclusion, the presented device allowed for reproducible seeding of human ECs and fibroblasts on cryopreserved valve containing homografts. Both the initially confluent cell layers and the sustained viability of the seeded cells after 6 days make further experimental and clinical studies with such seeded grafts possible.
5. Bader A, Schilling T, Teebken OE, et al. Tissue engineering of heart valves— human endothelial cell seeding of detergent acellularized porcine valves. Eur J Cardiothorac Surg 1998;14(3):279 – 84. 6. Fischlein T, Fasol R. In vitro endothelialization of bioprosthetic heart valves. J Heart Valve Dis 1996;5(1):58 – 65. 7. Fischlein T, Lehner G, Lante W, et al. Endothelialization of cardiac valve bioprostheses. Int J Artific Org 1994;17:345–52. 8. Bengtsson L, Radegran K, Haegerstrand A. In vitro endothelialisation of commercially available heart valve bioprostheses with cultured adult human cells. Eur J Cardiothorac Surg 1993;7:393– 8. 9. Walluschek KP, Steinhoff G, Haverich A. Endothelial cell seeding on native vascular surfaces. Eur J Vasc Endovasc Surg 1996;11(3):290 –303. 10. Lalka SG, Oelker LM, Malone JM, et al. Acellular vascular matrix: a natural endothelial cell substrate. Ann Vasc Surg 1989;3(2):108 –17. 11. Lee YS, Park DK, Kim YB, Seo JW, Lee KB, Min BG. Endothelial cell seeding onto the extracellular matrix of fibroblasts for the development of a small diameter polyurethane vessel. ASAIO J 1993;39(3):M740 –5. 12. Grant DS, Kleinmann HK, Martin GR. The role of basement membranes in vascular development. Ann NY Acad Sci 1990;588:61–72. 13. Sipehia R, Martucci G, Lipscombe J. Transplantation of human endothelial cell monolayer on artificial vascular prosthesis: the effect of growth—support surface chemistry, cell seeding density, ECM protein coating, and growth factors. Artif Cells Blood Substit Immobil Biotechnol 1996; 24(1):51– 63. 14. Bowlin GL, Rittgers SE. Electrostatic endothelial cell transplantation within small-diameter (⬍ 6mm) vascular prostheses: a prototype apparatus and procedure. Cell Transplant 1997;6(6):631–7. 15. Kent KC, Oshima A, Whittemore AD. Optimal seeding conditions for human endothelial cells. Ann Vasc Surg 1992;6(3):258 – 64. 16. Mazzucotelli JP, Roudiere JL, Bernex F, Bertrand P, Leandri J, Loisance D. A new device for endothelial cell seeding of a small-caliber vascular prosthesis. Artif Organs 1993;17(9): 787–90. 17. Thompson MM, Budd JS, Eady SL, et al. Effect of seeding time and density on endothelial cell attachment to damaged vascular surfaces. Br J Surg 1993;80(3):359 – 62. 18. Yankah AC, Wottge HU. Allograft conduit wall calcification in a model of chronic arterial graft rejection. J Card Surg 1997;12(2):86 –92. 19. Moustapha A, Ross DB, Bittira B, et al. Aortic valve grafts in the rat: evidence for rejection. J Thorac Cardiovasc Surg 1997;114(6):891–902. 20. Smith JD, Ogino H, Hunt D, Laylor RM, Rose ML, Yacoub MH. Humoral response to human aortic valve homografts. Ann Thorac Surg 1995;60:S127–30. 21. Sodian R, Sperling JS, Martin DP, et al. Fabrication of a trileaflet heart valve scaffold from a polyhydroxyalkanoate biopolyester for use in tissue engineering. Tissue Eng 2000; 6:183– 8. 22. Sodian R, Lemke T, Fritsche C, et al. Tissue-engineering bioreactors: a new combined cell-seeding and perfusion system for vascular tissue engineering. Tissue Eng 2002;8(5): 863–70. 23. Jockenhoevel S, Zund G, Hoerstrup SP, Schnell A, Turina M. Cardiovascular tissue engineering: a new laminar flow chamber for in vitro improvement of mechanical tissue properties. ASAIO J 2002;48(1):8 –11. 24. Gulbins H, Pritisanac A, Petzold R, et al. A low-flow adaption phase improves sheer-stress resistance of artificially seeded endothelial cells. Thorac Cardiovasc Surg 2004 (accepted for publication).
Disclosures and Freedom of Investigation For the studies published in this article, only the official clinical research funds of the department of cardiac surgery, University Hospital Grosshadern (Munich, Germany) were used. The homografts were taken from our in-house homograft bank (responsible, co-author AU). All authors had full control of the design of the study, methods used, outcome measurements, analysis of data, and production of the written article.
References 1. Thilo DG, Muller-Kusel S, Heinrich D, Kaufer I, Weiss E. Isolation of human venous endothelial cells by different proteases. Artery 1980;8(3):259 – 66. 2. Jarrel B, Shapiro S, Levine E. Human adult endothelial growth in culture. J Vasc Surg 1984;1:623–5. 3. Haegerstrand A, Gillis C, Bengtsson L. Serial cultivation of adult human endothelium from the great saphenous vein. J Vasc Surg 1992;16:280 –5. 4. Gulbins H, Goldemund A, Anderson I, et al. Preseeding with autologous fibroblasts improves endothelialization of glutaraldehyde-fixed porcine aortic valves. J Thorac Cardiovasc Surg 2003;125:592– 601.
NEW TECHNOLOGY
2125
Ann Thorac Surg 2005;79:2119 –26
2126
NEW TECHNOLOGY GULBINS ET AL DEVELOPMENT OF A SEEDING DEVICE AND PROCEDURE
Ann Thorac Surg 2005;79:2119 –26
25. Darnell J, Lodish H, Baltimore D. Molecular cell biology, 2nd ed. Scientific American Books, 1990: 914-32. 26. Ortenwall P, Wadenvik H, Kutti J, Risberg B. Endothelial cell seeding reduces thrombogenicity of Dacron grafts in humans. J Vasc Surg 1990;11(3):403–10. 27. Shindo S, Takagi A, Whittemore AD. Improved patency of collagen-impregnated grafts after in vitro autogenous endothelial cell seeding. J Vasc Surg 1987;6(4):325–32. 28. Gulbins H, Goldemund A, Uhlig A, Meiser B, Reichart B. Implantation of an autologously endothelialized homograft. J Thorac Cardiovasc Surg 2003;126:890 –1.
29. Gulbins H, Dauner M, Petzold R, et al. Development of an artificial vessel lined with human vascular cells. J Thorac Cardiovasc Surg 2004;128:372–7.
Disclaimer The Society of Thoracic Surgeons, the Southern Thoracic Surgical Association, and The Annals of Thoracic Surgery neither endorse nor discourage use of the new technology described in this article.
NEW TECHNOLOGY