Macroscopically, the TEBV appeared as a homogeneous tubular tissue strikingly resembling a human artery (Fig. 1A). Histological analysis revealed well-defined tissues: intima, media, and adventitia (Fig. 1B). The adventitia exhibited very dense collagenous layers as well as abundant fibroblasts. In the media, SMC appeared as elongated cells with circumferential or longitudinal orientations. SMC density, although very high for an in vitro model, was still lower than in normal vascular media. SMC did not penetrate the dense IM even though they were in contact with it for more than 8 wk and exposed to a gradient of nutriments leading to the lumen. Metabolic labeling and immunostaining of the endothelium (Fig. 1C) revealed confluent and active ECs as demonstrated by ac-LDL uptake and by von Willebrand factor, two specific EC functions (21, 26–28). SMC stained positively for muscular markers -smooth muscle actin (not shown) and desmin (Fig. 1D) (23, 29). Adventitial cells were negative for desmin or -smooth muscle actin, but fibroblasts expressed vimentin and synthesized high amounts of elastin assembled in small fibers, which were organized in large circular arrays (Fig. 1E). Immunostaining also indicated that the ECM contained type I, III, and IV collagens as well as laminin, fibronectin, and chondroitin sulfates (not shown).
Mechanical strength of the TEBV
Because mechanical strength has been the weak point of previous biological TEBV, we first evaluated this property in our vessel. To maximize its burst strength, we decided to optimize the strength of the engineered adventitia since this layer was clearly stronger than the media. The extent of both culture and maturation periods influenced the mechanical properties of this tissue. The culture period before detachment and rolling of the fibroblast sheet determined the thickness of the resulting adventitia (Fig. 2A). As expected, the burst pressure of the adventitia was similarly dependent on culture time (Fig. 2B). From these results, a culture period of 5 wk was chosen for all fibroblast sheets (IM and adventitias) since it gave the optimal strength to culture time ratio. During the maturation period, the burst strength of the adventitia steadily increased from the 1st to the 7th wk, when it reached a plateau at 2232 ± 251 mmHg (Fig. 2C). This plateau was maintained until at least the 12th wk, and some adventitias were matured for 24 wk without significant change in strength (2382±249 mmHg; n=5). Consequently, TEBV were kept in culture for at least 8 wk after the addition of the adventitia. The burst strength of the complete TEBV (2594±501 mmHg) was comparable to that of the adventitia alone, although the IM itself had a burst strength of over 1000 mmHg (Fig. 2D). Most important, TEBV burst strength was significantly higher than that of human saphenous veins (1680±307 mmHg), which are currently considered the optimal grafts for lower limb vascular reconstruction (3).
We attribute the mechanical strength of this model to the particularly well-organized collagenous matrix of the adventitia (Fig. 3A). Ultrastructural analysis revealed long collagen fibrils, featuring the characteristic 67 nm cross-striation periodicity, organized in closely packed bundles of parallel fibrils as observed in vivo (30). Furthermore, fibril bundles were oriented in perpendicular directions to one another, and a network of 10–12 nm elastin-associated microfibers was observed parallel to the collagen fibrils. Such a level of ultrastructural organization is characteristic of a mature collagen scaffold (30, 31).
Because increased matrix metalloproteinase (MMP) synthesis was proposed as a critical factor in long-term instability of collagen-based engineered organs in vitro (17), we followed collagenase synthesis during our vessel production and maturation. Normal fibroblasts cultured in vitro produce large quantities of MMP-2 (32), also called 72 kDa type IV collagenase/gelatinase or gelatinase A. Gelatin zymography of conditioned culture medium (Fig. 3B) revealed that fibroblast sheets prior to rolling produced mainly 70 kDa and 60 kDa gelatinases (lane 1), likely the pro-enzyme form of MMP-2 and its enzymatically active 62 kDa form (32). SMC sheets produced mostly the 72 kDa form of MMP-2 and also produced a gelatinolytic doublet of about 50 kDa (lane 3). When cell sheets were rolled, a rapid and marked overall decrease in collagenase production was observed for both fibroblasts (lane 1 vs. 2) and SMC (lane 3 vs. 4). However, the rolling step specifically induced expression of the 50 kDa gelatinolytic doublet by fibroblast sheets when rolled either alone (lane 1 vs. 2) or over SMC (lanes 1 + 4 vs. 5). This induction subsided within 2 wk (lane 6). During long-term culture, down-regulation of gelatinase expression was maintained and the remaining activity was observed mainly in the 72 kDa band (lane 7 and 8).
The ECs seeded on the IM covered 99.2 ± 0.7 % (n=6) of the luminal surface at a cell density of 5.4 ± 1.1 x 104 cells/cm2 (n=5), as measured by morphometric analysis. These ECs expressed a functional thrombin receptor since they responded to thrombin stimulation by an 2.9 ± 0.7 (n=5) fold increase in PGI2 synthesis, a potent inhibitor of platelet aggregation. Heparinized human whole blood was used in a functional assay to evaluate the blood compatibility of this endothelium. When ECs were absent, platelets adhered to the IM in small clusters, whereas conflu~ent ECs inhibited platelet adhesion by 93.2 ± 3.6 % and limited the phenomenon to single platelet adhesion in intercellular gaps (Fig. 4).
Grafting of the TEBV
Based on our promising in vitro data, we extended the model to a preliminary in vivo assessment. In addition to the ability to withstand hydrostatic pressure, TEBVs must also resist the various mechanical insults involved in vascular grafting procedures: manipulation with metal instruments and suture-related strains. To assess in vivo graftability, 5 cm-long unendothelialized human TEBV (inner diameter =3 mm) were implanted as interposition femoral grafts in mongrel dogs. TEBV were not endothelialized in order to avoid the immediate thrombosis associated with the acute rejection of endothelialized vascular xenograft (33, 34). Suturability and handling characteristics of the graft were evaluated as `tissue-like' by an experienced vascular surgeon. When blood flow was reestablished, no transmural blood loss was observed and immediate patency was confirmed by a palpable pulse and a visible downstream pulsatile flow. After 7 days, graft patency was assessed by angiography and explantation (Fig. 5). Even with the severe problems associated with xenografts, we were able to demonstrate a 50% patency rate (three out of six grafts). Patent grafts showed a smooth thrombus-free luminal surface and did not show signs of degradation, tearing, or dilatation. In all grafts, intramural blood infiltrations were observed between tissue layers, but these did not correlate with decreased patency; some patent grafts had significant infiltrations (Fig. 5B) whereas some thrombosed grafts had minimal infiltrations. Aside from blood infiltration, histological analysis revealed that the graft's architecture was retained. Graft failures occurred during the first days, as evaluated by Doppler signaling, and were the result of occlusive thrombus formation. This is in keeping with the thrombogenic nature of the collagen matrix observed in vitro in the absence of an endothelium (Fig. 4A).