Decellularized tissue-engineered blood vessel as anarterial conduitClay Quinta, Yuka Kondob, Roberto J. Mansonc, Jeffrey H. Lawsonc, Alan Dardikb, and Laura E. Niklasona,d,1

Departments of aBiomedical Engineering, bSurgery, and dAnesthesiology, Yale University, New Haven, CT 06520; and cDepartment of Surgery, DukeUniversity Medical Center, Durham, NC 27708

Edited* by Robert Langer, Massachusetts Institute of Technology, Cambridge, MA, and approved April 19, 2011 (received for review January 1, 2011)

Arterial tissue-engineering techniques that have been reportedpreviously typically involve long waiting times of several monthswhile cells from the recipient are cultured to create the engineeredvessel. In this study, we developed a different approach to arterialtissue engineering that can substantially reduce the waiting time fora graft. Tissue-engineered vessels (TEVs) were grown from bankedporcine smooth muscle cells that were allogeneic to the intendedrecipient, using a biomimetic perfusion system. The engineeredvessels were then decellularized, leaving behind the mechanicallyrobust extracellular matrix of the graft wall. The acellular graftswere then seeded with cells that were derived from the intendedrecipient—either endothelial progenitor cells (EPC) or endothelialcell (EC)—on the graft lumen. TEV were then implanted as end-to-side grafts in the porcine carotid artery, which is a rigorous testbeddue to its tendency for graft occlusion. The EPC- and EC-seeded TEVall remained patent for 30 d in this study, whereas the contralateralcontrol vein grafts were patent in only 3/8 implants. Going alongwith the improved patency, the cell-seeded TEV demonstrated lessneointimal hyperplasia and fewer proliferating cells than did thevein grafts. Proteins in the mammalian target of rapamycin signal-ing pathway tended to be decreased in TEV compared with veingrafts, implicating this pathway in the TEV’s resistance to occlusionfrom intimal hyperplasia. These results indicate that a readily avail-able, decellularized tissue-engineered vessel can be seeded withautologous endothelial progenitor cells to provide a biological vas-cular graft that resists both clotting and intimal hyperplasia. In ad-dition, these results show that engineered connective tissues can begrown from banked cells, rendered acellular, and then used fortissue regeneration in vivo.

bypass graft | collagen | mechanical conditioning

Arterial bypass graft implantation remains the primary ther-apy for patients with advanced cardiovascular disease. The

ongoing need for arterial conduits is due to the poor clinicalefficacy of existing synthetic grafts in small diameter arteryapplications (<6 mm) (1). In addition, many patients with arte-rial disease lack adequate saphenous vein or other conduit forbypass procedures (2, 3). Tissue-engineering approaches can beused to generate biologically based conduits to address the needfor vascular grafts (4, 5). In one approach, native vessels can bedecellularized and/or cross-linked, and can serve as conduits forsmall diameter arterial grafting (6, 7). However, procuring nativeallogeneic vessels for decellularization is a significant practicalhurdle, which has heretofore limited the extensive application ofsuch an approach. We report here the results of a large animalstudy of a tissue-engineered graft that is engineered from allo-geneic cells, and is decellularized to provide a suitable matrixfor endothelialization.A viable endothelium is generally required to prevent

thrombosis of small caliber vascular grafts in animal models,especially for decellularized vessels that are mostly comprised ofcollagen (8, 9). The source of endothelium to coat the vasculargraft could be obtained from a minimally invasive peripheralblood draw to harvest endothelial progenitor cells, as opposedto isolating endothelial cells from vein by a surgical procedure

(10–12). The approach of integrating a tissue-engineered vessel,decellularization, and an endothelial progenitor cell populationhas not been reported previously to the best of our knowledge.Decellularization of a completely tissue-engineered vessel couldprovide a designer vessel that could be stored like synthetic (e.g.,Teflon or Dacron) materials and would be available whenneeded as a bypass graft (13). The decellularization of a tissue-engineered vessel is more efficient than engineering a vessel foran individual patient and eliminates the extended lead time forautologous vessel culture that may limit the widespread usecompletely autologous grafts (14). In addition, demonstratingutility of an acellular, engineered connective tissue should pointthe way to producing other acellular tissues that may involve noculture, or only a short culture period for treating patients with avariety of connective tissue defects.

ResultsTissue-Engineered Vessel and Mechanical Properties. A biomimeticperfusion system was used to culture small diameter vessels, asreported (Fig. 1A) (15, 16). Aortic smooth muscle cells werecultured on a degradable polyglycolic acid mesh scaffold inbioreactors for 10 wk. At the conclusion of culture, the freshtissue-engineered vessel had a robust, dense layer of smoothmuscle cells with minimal residual scaffold polyglycolic acid(PGA) fibers (Fig. 1 B and C). The vessel wall thickness of theengineered vessels (442 ± 13 μm; n = 8) was not significantlydifferent after decellularization (459 ± 17 μm; n = 8). Beforedecellularization, engineered tissues had 15–20 layers of smoothmuscle cells interposed with layers of collagenous extracellularmatrix. Few PGA polymer residuals were observed near the lu-minal aspects of the engineered vessels. After decellularization,nuclei were removed from the engineered vessels (Fig. 1D),whereas collagen was found throughout the engineered anddecellularized vessels (Fig. 1 E and F). The histological findingswere confirmed by assays for DNA and collagen content. TheDNA of the fresh vessel (5.7 ± 0.4% of dry weight; n = 8) wassignificantly reduced after decellularization (0.8 ± 0.05% dryweight; n = 8; P < 0.001). In contrast, the collagen content in-creased as a percentage of the dry weight after decellularization(69 ± 2.0; n = 8) compared with the fresh vessels (51 ± 2.4; n =8; P < 0.001). Because the collagen matrix remains after decel-lularization and other cellular proteins are removed, the amountof collagen tends to increase as a percentage of dry weight afterdecellularization. The loss of cellular proteins was also evaluatedby immunohistochemistry and Western blot analysis (Fig. S1).

Author contributions: C.Q., J.H.L., A.D., and L.E.N. designed research; C.Q., Y.K., R.J.M.,J.H.L., and A.D. performed research; C.Q. analyzed data; and C.Q. and L.E.N. wrote thepaper.

Conflict of interest statement: L.E.N. has a financial interest in Humacyte, Inc., a regener-ative medicine company. Humacyte did not fund these studies, and Humacyte did notaffect the design, interpretation, or reporting of any of the experiments herein.

*This Direct Submission article had a prearranged editor.1To whom correspondence should be addressed. E-mail: laura.niklason@yale.edu.

This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10.1073/pnas.1019506108/-/DCSupplemental.

9214–9219 | PNAS | May 31, 2011 | vol. 108 | no. 22 www.pnas.org/cgi/doi/10.1073/pnas.1019506108

Dow

nloa

ded

by g

uest

on

Aug

ust 5

, 202

0

On immunohistochemistry, major histocompatibility complex I(MHC-I) was removed after decellularization (Fig. S1 A and B).By Western blot analysis, the smooth muscle protein calponin, aswell as β-actin and GAPDH, were removed after decellulariza-tion (Fig. S1C). Hence, the decellularization method resulted ina loss of cellular components but retention of the collagenousextracellular matrix.Porcine tissue-engineered vessels had mechanical properties

that were comparable to human saphenous vein. The burstpressure of the fresh (1,337 ± 103 mm Hg; n = 4) and decel-lularized tissue-engineered vessels (1,300 ± 59; n = 6) weresimilar, and near the burst pressure of a human saphenous veinat 1,680 ± 307 (17). The mechanical properties of the fresh anddecellularized vessels were also compared by obtaining stress–strain curves (Fig. S2). Stress–strain curves for engineered anddecellularized vessels were similar, although decellularized ves-sels had a somewhat lower ultimate tensile strength [fresh TEV,1.44 ± 0.068 MPa (n = 2) and decellularized TEV, 1.03 ± 0.208MPa (n = 4)]. Therefore, the mechanical properties, collagencontent, and wall thickness of the decellularized engineered ves-sels were similar to engineered vessels and human saphenous vein.

Endothelial Progenitor Cells (EPCs) Isolated from Peripheral Blood.Peripheral blood EPCs were isolated from pigs that were des-tined to received engineered grafts. EPCs exhibited character-istic endothelial properties by morphology, phenotype, andfunction. As shown by phase contrast microscopy, EPCs hada typical cobblestone morphology (Fig. 2A) that was similar tothat of differentiated porcine endothelium (Fig. S3A), which wasalso obtained from pigs destined to receive grafts. The EPCsuniformly expressed CD31 (platelet endothelial cell adhesion

molecule-1; Fig. 2B) and CD144 (vascular endothelial cadherin;Fig. 2C). EPCs were also positive for von Willibrand factor(vWF) that was present as granules in the cytoplasm (Fig. 2D).The functional characteristics of the EPCs were shown bythe uptake of acetylated LDL (Fig. 2E) and the formation ofcapillary-like tubes when cultured on Matrigel (Fig. 2F). Flowcytometry verified the immunofluorescent staining with expres-sion of CD31 (Fig. 2G), and absence of hematopoietic markerCD45 (Fig. 2H). After exposure to laminar shear stress of 15dyne/cm2 for 24 h, EPCs and differentiated endothelial cells(ECs) both showed an increase in eNOS protein (Fig. 2I), whichis expected for functional EC (18, 19). Hence, the characteristicsof porcine peripheral blood-derived EPC were similar in all waysassessed to differentiated porcine EC (Fig. S3) (20, 21).

Porcine Implantation. A porcine implantation model was used toassess the efficacy of the EC- or EPC-seeded decellularizedengineered vessels as small diameter arterial grafts. As shown inthe study outline (Fig. S4), the preparation of the TEV for im-plantation included forming an engineered vessel, then decellu-

Fig. 1. Tissue-engineered vessel from porcine SMC. (A) Porcine tissue-engi-neered vessel after 10 wk in culture. (B) Histology of the tissue-engineeredvessel by H&E staining. H&E of the tissue-engineered vessel before (C) andafter (D) decellularization with a loss of nuclei. Masson’s Trichrome stain be-fore (E) and after (F) decellularization demonstrating preservation of collagen(blue) throughout the matrix. (Scale bars: B, 500 μm; C–F, 100 μm.)

Fig. 2. Characterization of porcine EPCs from peripheral blood. (A) EPCsexhibit typical cobblestone endothelial cell morphology. Immunofluorescentdetection of CD31 (B), CD144 (C), vWF (D) demonstrate endothelial cellphenotype. Functional properties of the EPCs included incorporation of aLDL(E) and formation of capillary-like tubes on Matrigel (F). (Scale bars: 50 μm.)Flow cytometry is positive for CD31 (G) and negative for CD45 (H) (blackpeak is antibody stained). (I) EPC-derived EC and differentiated aortic ECresponse to shear stress on a decellularized porcine artery by immunoblot-ting for eNOS protein: low shear = 1 dyne/cm2; high shear = 15 dynes/cm2.

Quint et al. PNAS | May 31, 2011 | vol. 108 | no. 22 | 9215

MED

ICALSC

IENCE

S

Dow

nloa

ded

by g

uest

on

Aug

ust 5

, 202

0

larizing the tissue, harvesting ECs or EPCs from graft recipients,retroviral cell labeling with GFP, seeding the lumen of thedecellularized tissue with labeled EC (or EPC), and thenimplanting the cell-seeded graft into the recipient carotid. Thestability of the GFP-labeled EPC was evaluated in vitro overa 45-d period and 3–4 passages using flow cytometry, whichshowed that >99% of EPC were GFP+ after 45 d (Fig. S5). EC(or EPC) seeding onto the graft lumen resulted in an averagepercent coverage of 64 ± 9% (n = 5). A total of five cell-seededvessels were implanted, including n = 3 for EPC-seeded engi-neered grafts and n = 2 for EC-seeded grafts. The controlgrafts for the study were nonseeded, decellularized matrices(n = 3) and internal jugular vein (n = 8). Aspirin and clopi-dogrel were given 1 d before surgery and continued for theduration of the study.All grafts were implanted in the common carotid as end-to-

side grafts, to mimic clinical vascular bypass, for a 30-day period(Fig. 3A). In vivo imaging by angiography or MRI was performedat 30 d, which was combined with histological analysis to assesspatency (Fig. 3B). The patency rate of the cell-seeded TEVvessels was 5/5 (EC- and EPC-seeded; Table 1), whereas thenonseeded TEV had a patency rate of 0/3, with histology re-vealing luminal clot likely due to exposure of host platelets to thecollagenous TEV wall (Fig. S6B). Explants of cell-seeded TEVdemonstrated migration of host cells into the residual engi-neered matrix and minimal inflammatory response (Fig. 3C). Incontrast to cell-seeded TEV, vein grafts demonstrated sub-stantial intimal hyperplasia in the 3/8 patent grafts (Fig. 3D andFig. S6 C and D) and in most of the 5/8 occluded grafts. In 3/5occluded vein grafts, there was significant intimal hyperplasiawith a thrombosis in the remnant lumen (Fig. 5E). The othertwo occluded vein grafts appeared to have failed by a throm-botic event that occurred before substantial intimal hyperplasiacould develop.An endothelial layer was found on all of the explanted cell-

seeded TEV, as indicated by vWF staining (Fig. 3F). All of thecell-seeded TEV had α-actin–positive staining on the outside ofthe residual engineered vessel (Fig. 3G), and two of the cell-seeded engineered vessels also had α-actin positive staining inthe neointima. In the two samples with α-actin–positive stainingin the neointima, the inner and outer α-actin layers were sepa-rated by an α-actin negative region that was the residual TEV(Fig. S6A). However, a vimentin stain for fibroblasts was positivein the residual TEV at 30 d (Fig. 3H), indicating that althoughα-actin cells did not extensively repopulate the matrix by 30 d,there were notable numbers of fibroblasts in the matrix at thattime. There was not a significant inflammatory response in theresidual engineered vessels, as shown by sparse CD45 positivecells (Fig. 3I).The lumens of the cell-seeded engineered vessels stained posi-

tively for GFP-labeled EC or EPC before implantation (Fig. 3J).At the 30-d explant point, GFP+ cells were found in 4/5 cell-seeded grafts and the luminal coverage ranged from 5% to 60%,with an average of 35% (Fig. 3K). The non-GFP+ endothelial cellson the TEVs (average 65%) were likely host-derived, because invitro studies showed GFP retention by EC for at least 45 d. Inaddition, the GFP+ cells were found only at the luminal surfaceand did not appear to migrate into the neointima or residual vessel

matrix. Thus, the GFP+ ECs and EPCs did not appear to con-tribute directly to any neointima in cell-seeded TEV.Morphometric analysis quantified the graft differences of the

explanted cell-seeded engineered grafts compared with veingrafts. The luminal area (Fig. 4A) of the cell-seeded tissue-engineered vessels (2.31 ± 0.49 mm2; n = 5) tended to be largerthan those of the vein grafts (1.22 ± 0.51 mm2; n = 6), and theintimal area (Fig. 4B) of the cell-seeded TEV (1.32 ± 0.84 mm2;n = 5) tended to be lower than that of the vein grafts (3.63 ±0.59 mm2; n = 6) (Fig. S7). Although intima-to-media ratiosare frequently reported when analyzing vein grafts, this ap-proach does not directly apply to tissue-engineered vesselsbecause there is no defined media. In the cell-seeded tissue-engineered vessels, the neointima-to-residual engineered vessel

Fig. 3. Implantation of vessels. (A) TEV anastomosed as an end-to-side by-pass in the porcine carotid artery. (B) Angiogram of a vein graft (left, arrowsat anastomoses) and an EPC-seeded TEV (right, arrowheads at anastomoses)30 d after implantation. (C) Low magnification cross-section of an explantedEPC-seeded engineered vessel with cellular repopulation of the matrix.Patent vein graft (D) and an occluded vein graft (E), both demonstratingintimal hyperplasia. Immunohistochemical staining of explanted cell-seededgrafts for vWF on the luminal surface (arrow) (F), SMC α-actin expressingcells (arrow) on the periphery of the residual engineered matrix (G), andvimentin-positive cells (arrow) populating the residual graft matrix (H). (I)CD45 (arrow) stain for leukocytes. (J) The preimplant engineered matrixseeded with GFP+-labeled EPCs. After 30 d of implantation, GFP+-labeledEPCs were detected on the lumen of the engineered vessel (K). (Scale bars:C–E, 500 μm; F–K, 100 μm.)

Table 1. Graft patency rates

Graft Patency

Cell-seeded tissue-engineered vessel 5/5EPC-seeded TEV 3/3EC-seeded TEV 2/2

Non-seeded tissue engineered Vessel 0/3Vein graft (internal jugular vein) 3/8

9216 | www.pnas.org/cgi/doi/10.1073/pnas.1019506108 Quint et al.

Dow

nloa

ded

by g

uest

on

Aug

ust 5

, 202

0

ratio was 0.55 ± 0.34 (n = 5) compared with vein grafts with anintima-to-media ratio of 1.34 ± 0.34 (n = 6). All of the veingrafts with intimal hyperplasia (6/8) were included to quantifyintimal hyperplasia and luminal (or residual luminal) area.Cellular proliferation in the neointima and the graft wall wereevaluated by PCNA staining, which showed that vein grafts(25.4 ± 9.1; n = 6) appeared to have a greater number ofproliferating cells per high power field in comparison with thecell-seeded TEV (13.3 ± 6.1) (Fig. 4C and Fig. S8). Althoughthe proliferation rates (i.e., PCNA+ cells as a fraction of totalcells) in the vein grafts (19.1 ± 5.3; n = 6) were similar to thecell-seeded tissue-engineered vessels (18.6 ± 7.6; n = 6), therewere overall greater number of cells in the vein grafts. Hence,the cell-seeded tissue-engineered vessels tended to have greaterluminal area, less neointimal area, and fewer proliferating cellsthan did the vein grafts, all of which likely contributed to im-proved patency of the cell-seeded TEV. Although these obser-vations did not reach statistical significance, the trends wereentirely consistent.The mammalian target of rapamycin (mTOR) signaling

pathway has been implicated in cell proliferation of smoothmuscle cells in vein grafts, and inhibitors of the mTORC1complex have been shown to ameliorate intimal hyperplasia invascular stents (22, 23). To evaluate elements of the mTORpathway in explanted specimens, midgraft segments of the cell-seeded TEV and vein grafts were analyzed for the expression ofthe downstream effector phosphorylated p70S6K and the up-stream activator phosphorylated Akt. By immunoblotting, theprotein expression of p-p70S6K tended to be higher in the veingrafts than in the cell-seeded TEV, although this difference didnot achieve significance {vein graft 3.6-fold greater than TEV[TEV 3,920 ± 1,096 optical density (OD), n = 5 and vein graft14,124 ± 4,795 OD, n = 6]; Fig. 4D}. The expression of phos-phorylated Akt was also somewhat higher in vein grafts than in

engineered grafts [vein graft 1.8 fold greater than TEV (TEV6,361 ± 468 OD, n= 5 and vein graft 11,602 ± 5,122 OD, n= 6);Fig. S9]. An increased expression of p-Akt tends to occur earlierthan 4 wk in the development of hyperplasia (24), and so thedifference in p-Akt may have been higher at 2 wk than at our4-wk time point. Hence, decreased activation of the mTORpathway in the TEVs compared with the vein grafts may havecontributed to decreased neointimal hyperplasia and improvedpatency of the cell-seeded TEV.

DiscussionIn the pig carotid model, the EPC- and EC-seeded TEV out-performed the vein graft by demonstrating a higher patency rate(100% vs. 38%) and a trend toward a lower neointimal response.The end-to-side graft implantation in the pig model is an ac-celerated intimal hyperplasia model, which typically results inlower patency rates than porcine interpositional vein grafts (25).Although both the cell-seeded TEV and the vein grafts had anendothelial layer, there was a substantial difference in graft pa-tency that correlated to the degree of intimal hyperplasia. Ob-viously, the tissue-engineered matrix is acellular, whereas thevein graft has a living smooth muscle layer that can contribute toneointimal thickening and luminal occlusion. In contrast, hostcells appear to migrate slowly into the decellularized engineeredmatrix, perhaps retarding the formation of neointima. Activationof mTOR pathway proteins was less in engineered grafts com-pared with veins, and because the immunoblot data were nor-malized to β-actin expression, this observation means that p-Aktand p-p70S6k were lower for cells in engineered grafts comparedwith veins. The explant morphometric data, the cellular pro-liferation results, and the protein expression levels of the mTORpathway all provide supportive data regarding the resistance ofthe cell-seeded tissue-engineered vessel to both intimal hyper-plasia and to thrombotic occlusion.Endothelial progenitor cells have been used to coat vascular

grafts to produce an antithrombotic surface and preventthrombosis (7, 8). In vitro studies have shown that EPCs, evenwhen harvested from patients with cardiovascular disease, sup-port a similar antithrombotic phenotype to differentiated ECswhen exposed to laminar shear stress (26). The GFP+ ECs andEPCs that survived over the 30-d implant period could haveplayed a direct role in maintaining an antithrombotic surface,possibly through the local release of nitric oxide because thepreconditioning to shear was shown to increase eNOS (27, 28).The non-GFP+ endothelial cells on the explanted cell-seededTEV likely originated from adjacent ECs or circulating EPCs inthe host. The EC labeling here demonstrated that the EC orEPC can be less than 100% confluent on the lumen at the time ofimplant and still maintain patency. However, the decellularizedTEV occluded in this animal model if ECs or EPCs were notseeded onto the lumen.The allogenic decellularized tissue-engineered matrix did not

induce a substantial inflammatory response, which is importantbecause an inflammatory response can contribute to stenosis (29,30). Although smooth muscle cells were not found in the residualmatrix, there was a smooth muscle cell layer outside all of theengineered vessels at explant. Only two of the five cell-seededengineered vessels had any neointima that was positive forsmooth muscle cells, but even in those tissues, there was a sep-aration of the neointima from the outer layer of smooth musclecells. Hence, it is unlikely that smooth muscle cells in the neo-intima migrated from outside the graft, but instead may havebeen derived from adjacent host tissue at the anastomosis. Inaddition, a circulating bone marrow progenitor may have alsocontributed to the development of the neointima (31, 32).Intimal hyperplasia as a result of cellular proliferation is as-

sociated with the mTORC1 pathway and its downstream effec-tor, p70S6K (33, 34). The pharmacological inhibitor rapamycin

Fig. 4. Characterization of explanted grafts. The luminal area (A) and in-timal area (B) were measured for the cell-seeded tissue-engineered vessels(TEV) (n = 5) and the vein grafts (VG) [n = 6, patent vein grafts (n = 3) andoccluded by intimal hyperplasia (n = 3)]. (C) Number of PNCA+ nuclei perhigh power field of the engineered graft neointima and vessel wall com-pared with vein graft intima and media. Immunoblots of engineered cell-seeded vessels (n = 5) or vein grafts (n = 6) for phosphorylated p70S6K andquantified and normalized by β-actin expression (D and E). TEV, 3,920 ±1,096 OD; VG, 14,124 ± 4,795 OD. Data represents mean ± SEM.

Quint et al. PNAS | May 31, 2011 | vol. 108 | no. 22 | 9217

MED

ICALSC

IENCE

S

Dow

nloa

ded

by g

uest

on

Aug

ust 5

, 202

0

has been shown to have an antiproliferative effect on vascularsmooth muscle cells in vitro and in vivo and has been used togreat clinical advantage with drug-eluting stents (23, 35). In thisstudy, phosphorylated p70S6K and Akt tended to have higherexpression levels in the vein grafts than in the cell-seeded TEV.The higher expression level of p-p70S6K correlated with anincreased number of proliferative cells in the vein graft comparedwith the cell-seeded TEV. This observation implies that differ-ences in mTORC1 pathway activation may underlie the improve-ments in patency of the engineered grafts.In summary, a tissue-engineered vessel can be decellularized

to create a readily available engineered connective tissue. Be-cause the engineered connective tissues (vessels) can be pro-duced from allogeneic cells, the months-long process required toculture a collagen-rich and mechanically robust tissue is moved“off-line,” not requiring cells from the actual recipient. EPC thatare obtained by using a minimally invasive blood draw can beseeded onto such a graft, to result in a functional arterial pros-thesis within only a few weeks. This decellularized tissue func-tions well as an arterial graft and is gradually remodeled in vivoby host cells. The paradigm of using allogeneic cells to engineeracellular connective tissues may find applicability in other areasof tissue engineering, thereby shortening the wait time forengineered replacements of other connective tissues.

MethodsCulture and Decellularization of Engineered Vessel. The bioreactor systems, cellseeding, and medium replenishment proceeded as described (15, 16, 36).Briefly, the scaffold for the tissue-engineered vessel consisted of a PGA mesh(Concordia) that was sewn by using Dexon suture (Syneture) around a sili-cone tube (Saint Gobain) with an outer diameter of 4 mm and a length of12 cm. A suspension of passage three aortic smooth muscle cells at a densityof 12 × 106 cells were used to seed the PGA mesh. After 10 wk under pul-satile culture conditions in the bioreactor, the vessels were decellularized byusing a two solution detergent process as described (13, 37). The tissue-engineered vessel was placed into the first decellularization solution con-taining 8 mM CHAPS (Sigma), 1 M NaCl (Sigma), and 25 mM EDTA (BostonBioproducts) in PBS. The vessel was stirred for 1 h on a stirplate in an in-cubator at 37 °C. The vessel was rinsed in PBS three times for 10 min each.The vessel was placed in a second decellularization solution containing 1.8mM SDS (Sigma), 1 M NaCl (Sigma), and 25 mM EDTA (Boston Bioproducts)in PBS. Again, it was stirred for 1 h in an incubator at 37 °C and rinsed in PBSthree times for 10 min each. The vessel was rinsed in EBM-2 medium (Lonza,without hydrocortisone) with 10% FBS (HyClone) for 24 h in an incubator at37 °C and rinsed in PBS three times for 10 min each.

Isolation of ECs from Artery and EPCs from Peripheral Blood. ECs were isolatedfrom the saphenous artery of 6- to 8-wk-old Yorkshire pigs by scraping thecells off the artery and plating on fibronectin-coated plates (BD Science), andculturing in EGM2 (Lonza, without hydrocortisone) (8, 11). The EPCs werecollected from the blood of 6- to 8-wk-old Yorkshire pigs. After the initial5 mL of blood was discarded, 30 mL was collected and 1,000 U heparin wasadded. The mononuclear fragment was obtained by a Histopaque densitygradient (Sigma) centrifugation at 400 × g for 30 min with the brake turnedoff. The cell pellet was washed three times in PBS and resuspended in EGM2supplemented with 20% FBS (HyClone). The cell seeding density was 1 to 3 ×107 cells per well, and the wells were precoated with fibronectin. The me-dium was changed at 48 h, and was changed every 2 d thereafter until cellcolonies appeared.

EC and EPC Labeling. ECs and EPCs were labeled with retroviral vectorsexpressing GFP (38). Cells were infected at 20–30% confluence and again at70–80% confluence. The infection was repeated for a total of four rounds.The cells were then sorted by flow cytometry for GFP expression, and thenseeded onto the TEV.

Preconditioning of ECs or EPCs on the TEV. Vessels were placed in a flowchamber and seeded at a density of 2 × 105 EC or EPC cells per cm2 by rotatingthe vessel 90° every 30 min. After 360° rotation (2 h), medium was replacedwith fresh medium and remained static for another 2 h. Shear stress wascalculated by using Poiseuille’s equation: τ = 4 μQ/πr3, where τ is shear stress,μ is fluid viscosity, Q is medium flow rate, and r is radius of the vessel. Shearstress was started at 1 dyne/cm2 for 2 d and gradually increased to 15 dynes/cm2 over 36 h. The vessels were maintained at 15 dynes/cm2 for 24 h. Theedge of each graft was analyzed for confluency by examining the luminalsurface by staining for DAPI (Vector Labs) and F-actin (Invitrogen) and usingImageJ to quantify the coverage of the luminal area.

Porcine Implant Study. The implanation protocol was approved by the In-stitutional Animal Care and Use Committee of Yale and Duke universities incompliance with animal care and handling with the Guide for the Care andUse of Laboratory Animals published by National Institute of Health. One daybefore surgery, aspirin (5 mg/kg) and clopidogrel (1 mg/kg) were given.Aspirin and clopidogrel were continued daily over the 30-d study. At the timeof implantation, heparin (100 IU/kg) was administered i.v. before arterialclamping. TEV (with or without cell seeding) were implanted by using 6-0 pro-lene as an end-to-side anastomosis to the common carotid artery. Vein graftswere autologous internal jugular vein and were implanted in the contra-lateral common carotid artery.

Explant Characterization. Standard histological stains of hematoxylin & eosin(H&E), and Masson’s trichrome stain for collagen, were used. Immunohis-tochemical staining for SMC α-actin (Dako), vimentin (Abcam), and vWF(Dako) used a streptavidin-biotinylated peroxidase kit (Vector 6101 or6102). Immunofluorescent staining with anti-GFP antibody (Abcam) useda FITC secondary (Abcam). Morphometric analysis was performed by usingImageJ software to obtain luminal area and intimal area using elastic-VanGieson stain.

Western Blot Analysis. Cell lysates were prepared as described and separatedby SDS/PAGE and transferred to polyvinylidene (PVDF) membrane (39). Thefollowing antibodies were used for Western blot analysis: eNOS (BD trans-duction laboratories), calponin (DAKO), GAPDH (Millipore), β-actin (Sigma),p-p70S6K (Cell Signaling), and p-AKT (Cell Signaling).

Vessel Characterization. Mechanical testing included burst pressure and su-ture retention and was described (15). Stress–strain curves were obtained byplacing hooks through a 4-mm ringlet and using an Instron (5800) with thecross-head speed set at 1 mm/sec. Collagen and DNA quantification wasdetermined as described (37).

Statistics. The properties (vessel thickness, collagen content, DNA quantifi-cation, and burst pressure) of the fresh and decellularized engineered vesselsare expressed as mean ± SEM. Statistical significance was evaluated by usingthe unpaired Student t test. Statistical significance was evaluated by usingthe nonparametric Mann–Whitney test for the explanted vessel propertiesluminal area, intimal area, and protein expression).

ACKNOWLEDGMENTS. This work was supported by National Institutes ofHealth Grants HL-083895, EB-008366 (to L.E.N.), and 1F32 HL083662 (to C.Q.).

1. Yeager RA, et al. (1982) Differential patency and limb salvage for polytetrafluoro-ethylene and autogenous saphenous vein in severe lower extremity ischemia. Surgery91:99–103.

2. Veith FJ, Moss CM, Sprayregen S, Montefusco C (1979) Preoperative saphenousvenography in arterial reconstructive surgery of the lower extremity. Surgery 85:253–256.

3. Giannoukas AD, et al. (1997) Pre-bypass quality assessment of the long saphenousvein wall with ultrasound and histology. Eur J Vasc Endovasc Surg 14:37–40.

4. Weinberg CB, Bell E (1986) A blood vessel model constructed from collagen andcultured vascular cells. Science 231:397–400.

5. Kakisis JD, Liapis CD, Breuer C, Sumpio BE (2005) Artificial blood vessel: The Holy Grailof peripheral vascular surgery. J Vasc Surg 41:349–354.

6. Schmidt CE, Baier JM (2000) Acellular vascular tissues: Natural biomaterials for tissuerepair and tissue engineering. Biomaterials 21:2215–2231.

7. Amiel GE, et al. (2006) Engineering of blood vessels from acellular collagen matricescoated with human endothelial cells. Tissue Eng 12:2355–2365.

8. Kaushal S, et al. (2001) Functional small-diameter neovessels created using endothelialprogenitor cells expanded ex vivo. Nat Med 7:1035–1040.

9. Cho SW, et al. (2005) Small-diameter blood vessels engineered with bone marrow-derived cells. Ann Surg 241:506–515.

10. Asahara T, et al. (1997) Isolation of putative progenitor endothelial cells forangiogenesis. Science 275:964–967.

11. Mead LE, Prater D, Yoder MC, Ingram DA (2008) Isolation and characterization ofendothelial progenitor cells from human blood. Curr Protoc Stem Cell Biol, Chapter 2,Unit 2C 1.

12. Kalka C, et al. (2000) Transplantation of ex vivo expanded endothelial progenitor cellsfor therapeutic neovascularization. Proc Natl Acad Sci USA 97:3422–3427.

9218 | www.pnas.org/cgi/doi/10.1073/pnas.1019506108 Quint et al.

Dow

nloa

ded

by g

uest

on

Aug

ust 5

, 202

0

13. Dahl SL, Koh J, Prabhakar V, Niklason LE (2003) Decellularized native and engineeredarterial scaffolds for transplantation. Cell Transplant 12:659–666.

14. McAllister TN, et al. (2009) Effectiveness of haemodialysis access with an autologoustissue-engineered vascular graft: A multicentre cohort study. Lancet 373:1440–1446.

15. Niklason LE, et al. (1999) Functional arteries grown in vitro. Science 284:489–493.16. Niklason LE, et al. (2001) Morphologic and mechanical characteristics of engineered

bovine arteries. J Vasc Surg 33:628–638.17. L’Heureux N, Pâquet S, Labbé R, Germain L, Auger FA (1998) A completely biological

tissue-engineered human blood vessel. FASEB J 12:47–56.18. Topper JN, Cai J, Falb D, Gimbrone MA, Jr. (1996) Identification of vascular endothelial

genes differentially responsive to fluidmechanical stimuli: Cyclooxygenase-2, manganesesuperoxide dismutase, and endothelial cell nitric oxide synthase are selectively up-regulated by steady laminar shear stress. Proc Natl Acad Sci USA 93:10417–10422.

19. Brown MA, Wallace CS, Angelos M, Truskey GA (2009) Characterization of umbilicalcord blood-derived late outgrowth endothelial progenitor cells exposed to laminarshear stress. Tissue Eng Part A 15:3575–3587.

20. Case J, et al. (2007) Human CD34+AC133+VEGFR-2+ cells are not endothelialprogenitor cells but distinct, primitive hematopoietic progenitors. Exp Hematol 35:1109–1118.

21. Timmermans F, et al. (2009) Endothelial progenitor cells: Identity defined? J Cell MolMed 13:87–102.

22. Braun-Dullaeus RC, et al. (2001) Cell cycle protein expression in vascular smoothmuscle cells in vitro and in vivo is regulated through phosphatidylinositol 3-kinaseand mammalian target of rapamycin. Arterioscler Thromb Vasc Biol 21:1152–1158.

23. Morice MC, et al.; RAVEL Study Group. Randomized Study with the Sirolimus-CoatedBx Velocity Balloon-Expandable Stent in the Treatment of Patients with de NovoNative Coronary Artery Lesions (2002) A randomized comparison of a sirolimus-eluting stent with a standard stent for coronary revascularization. N Engl J Med 346:1773–1780.

24. Stabile E, et al. (2003) Akt controls vascular smooth muscle cell proliferation in vitroand in vivo by delaying G1/S exit. Circ Res 93:1059–1065.

25. Tsutsumi H, Miyawaki F, Arakawa H, Tsuji T, Tanigawa M (2001) Experience of veingrafting in Göttingen minipigs. Exp Anim 50:191–195.

26. Stroncek JD, et al. (2009) Comparison of endothelial cell phenotypic markers of late-outgrowth endothelial progenitor cells isolated from patients with coronary arterydisease and healthy volunteers. Tissue Eng Part A 15:3473–3486.

27. Kibbe MR, et al. (2001) Adenovirus-mediated gene transfer of human inducible nitricoxide synthase in porcine vein grafts inhibits intimal hyperplasia. J Vasc Surg 34:156–165.

28. Ahanchi SS, Tsihlis ND, Kibbe MR (2007) The role of nitric oxide in the pathophysiologyof intimal hyperplasia. J Vasc Surg 45(Suppl A):A64–A73.

29. Davis C, Fischer J, Ley K, Sarembock IJ (2003) The role of inflammation in vascularinjury and repair. J Thromb Haemost 1:1699–1709.

30. Owens CD (2010) Adaptive changes in autogenous vein grafts for arterialreconstruction: Clinical implications. J Vasc Surg 51:736–746.

31. Shimizu K, et al. (2001) Host bone-marrow cells are a source of donor intimal smooth-muscle-like cells in murine aortic transplant arteriopathy. Nat Med 7:738–741.

32. Diao Y, et al. (2008) Long-term engraftment of bone marrow-derived cells in theintimal hyperplasia lesion of autologous vein grafts. Am J Pathol 172:839–848.

33. Martin KA, et al. (2004) The mTOR/p70 S6K1 pathway regulates vascular smoothmuscle cell differentiation. Am J Physiol Cell Physiol 286:C507–C517.

34. Wang Y, et al. (2007) Interferon-gamma induces human vascular smooth muscle cellproliferation and intimal expansion by phosphatidylinositol 3-kinase dependentmammalian target of rapamycin raptor complex 1 activation. Circ Res 101:560–569.

35. Sousa JE, et al. (2005) Four-year angiographic and intravascular ultrasound follow-upof patients treated with sirolimus-eluting stents. Circulation 111:2326–2329.

36. Dahl SL, et al. (2011) Readily available tissue-engineered vascular grafts. Sci TranslMed 3:68ra9.

37. Gui L, Chan SA, Breuer CK, Niklason LE (2010) Novel utilization of serum in tissuedecellularization. Tissue Eng Part C Methods 16:173–184.

38. Zheng L, Ben LH, Pober JS, Bothwell AL (2002) Porcine endothelial cells, unlike humanendothelial cells, can be killed by human CTL via Fas ligand and cannot be protectedby Bcl-2. J Immunol 169:6850–6855.

39. Gong Z, Niklason LE (2008) Small-diameter human vessel wall engineered from bonemarrow-derived mesenchymal stem cells (hMSCs). FASEB J 22:1635–1648.

Quint et al. PNAS | May 31, 2011 | vol. 108 | no. 22 | 9219

MED

ICALSC

IENCE

S

Dow

nloa

ded

by g

uest

on

Aug

ust 5

, 202

0