Preclinical research

In vivo cellularization of a cross-linked matrix byintraperitoneal implantation: a new tool in heart valvetissue engineering

Geofrey De Visscher*, Ilse Vranken, An Lebacq, Christiaan Van Kerrebroeck, Javier Ganame,Erik Verbeken, and Willem Flameng

Laboratory for Experimental Cardiac Surgery, Department of Cardiovascular Diseases, Katholieke Universiteit Leuven,Minderbroedersstraat 17, 3000 Leuven, Belgium

Received 21 April 2006; revised 26 October 2006; accepted 10 November 2006; online publish-ahead-of-print 23 January 2007

This paper was guest edited by Prof. Julie H. Campbell, University of Queensland, Australia

Aim To use in vivo instead of in vitro cell seeding in heart valve tissue engineering.Methods and results Intraperitoneally preseeded, photo-oxidized bovine pericardial pulmonaryvalve constructs (group 1) were compared with non-preseeded constructs (group 2) implanted insheep. All valves functioned normally and were macroscopically intact at explantation [1 week(n ¼ 6) and 1 month (n ¼ 6) in each group], except for one thrombosed leaflet in a group-2 valve at1 month. Almost 10-fold higher neomatrix deposition and doubling of the leaflet thickness werefound in group 1 vs. 2 (P , 0.05). A concomitant significant decrease in leaflet length (15%) wasfound at 1 month in group 1. The total cross-sectional surface and total amount of collagen of theoriginal matrix remained unchanged in both groups at all times. Immunohistochemistry showed a lowimmune response, stem/progenitor cell infiltration, appropriated differentiation, and spontaneousendothelialization of the valves. Significantly, increased re-cellularization was found after IP preseedingcompared with spontaneous seeding: cell coverage of the leaflet was 71–100 vs. 8–26% (P , 0.05),respectively.Conclusion Complete re-cellularization can be obtained by IP preseeding of an acellularized cross-linked matrix. Well-functioning valve constructs show cellularization and differentiation into myofibro-blast phenotype and concomitant neomatrix deposition.

KEYWORDSTissue engineering;

Endogenous seeding;

Heart valve

Introduction

The disadvantages of actual mechanical and bioprostheticheart valves are related to thromboembolic and bleedingcomplications as well as to calcification or degenerativevalve dysfunction.1 Recent attempts to overcome these pro-blems by tissue engineering of heart valves focus on thetechnology of in vitro seeding of host cells on biodegradablesynthetic scaffolds or non-cross-linked acellularizedmatrices.2 Although some experimental success wasclaimed, durability of these constructs in terms of theabsence of calcification and failure of the prosthetic valveremains to be proved. Furthermore, permanent and com-plete re-cellularization of the valve structures protectingthe valve against degeneration and thrombosis was neverdemonstrated.

The reasons for these flaws is basically the inherentpotential of the scaffold to degenerate and the need for apermanent and complete repopulation of the construct byhighly differentiated valve cells capable of producing anew matrix material having natural biophysical properties.Although it must be theoretically possible to generate a

functional heart valve using the concept of in vitro cellseeding on a degradable matrix, we preferred a completelydifferent approach. At first, we used a stable, cross-linkedmatrix that would not be degraded over time, to guaranteea durable heart valve, having adequate biophysical proper-ties and lasting for a long time without degradation.Second, we applied the concept of endogenous cellseeding using the natural ‘milieu interior’ for attraction,homing, and differentiation of progenitor cells, in contrastto the artificial and limited in vitro seeding conditions. Weused the organism as its own bioreactor, based on thework of Campbell et al.,3 who generated a functional vascu-lar graft from an intraperitoneally (IP) deposited capsulearound an implanted silastic tube. The capsule comprises

& The European Society of Cardiology 2007. All rights reserved. For Permissions, please e-mail: journals.permissions@oxfordjournals.org

* Corresponding author. Tel: þ32 16 33 71 20; fax: þ32 16 33 71 22.E-mail address: geofrey.devisscher@med.kuleuven.be

European Heart Journal (2007) 28, 1389–1396doi:10.1093/eurheartj/ehl422

an internal layer of macrophages, some layers of (myo)fibro-blasts and an external layer of mesothelial cells and is gen-erally described as ‘foreign body reaction’.4 By studying thecell morphology and co-localization of alpha smooth muscleactin (ASMA) and CD172a, it has been proven that the entirelayer of (myo)fibroblasts is of haematopoietic origin5 andthat the myofibroblasts in those conditions were derivedfrom macrophages.6 On the other hand, the presence ofan undifferentiated cell population in the explants wasalso reported.6 ASMA is a generally used marker for myofi-broblasts,7 but CD172a is not an exclusive marker for macro-phages and is also expressed in a subpopulation ofhaematopoietic stem cells.8,9 Therefore, the possible pre-sence of stem cells in granulation tissue renders the studyof the ‘early’ foreign body reaction as a means to repopulatea biological but stabilized matrix attractive. Based on thepresence of approximately 4.5% lineage negative cells on 3day IP implants in rats (unpublished results), we performed3 day endogenous IP seeding by the valve recipient itselffor this study. As a cross-linked scaffold we have used a non-cytotoxic photo-oxidized bovine pericardium,10 which isshown to tolerate in vitro cell seeding.11

The aim of this study was primarily to prove completere-cellularization of the scaffolds and to demonstrateappropriate cell differentiation and phenotype. On theother hand, we also wanted to show that the valvesremain haemodynamically stable up to 1 month afterimplantation in the pulmonary position, free of thrombosisin the absence of anti-coagulation therapy and free ofcalcification.At present, it was not our intention to compare this

valve to any type of existing bioprostheses in terms ofdurability because the actual design of the valve is experi-mental and meant only as a tool to prove re-cellularizationpotential.

Methods

Animals and study design

Lovenaar sheep [n ¼ 24, ewes 60 kg (45–75)] were selected andcared for in accordance with the ‘Guide for the Care and Use ofLaboratory Animals’ (NIH publication 85-23, revised 1985). Thestudy was approved by the local Ethics Committee.Photo-oxidized (CardiofixTM, Sulzer Carbomedics, TX, USA)

pericardium was used.The study comprised four groups to which the sheep were

randomly assigned and the groups contained six animals each:group 1, spontaneous seeding and 1 week in pulmonary position;group 2, spontaneous seeding and 1 month in pulmonary position;group 3, IP preseeding and 1 week in pulmonary position; andgroup 4, IP preseeding and 1 month in pulmonary position. Thesheep of groups 3 and 4 were implanted in the peritoneal cavitywith a pericardial patch suspended in a stainless steel cage. Threedays later, the implant was aseptically retrieved and a valve(Figure 1) was made, similar to the one described by Whiteet al.12 In short, a 6.8 � 3.8 cm pericardium patch was foldedalong the long axis leaving an extra 2 mm at on end. The fold wasfixed with one 5-0 prolene suture at each side, the whole flap wasfolded along the short axis, and the short axes were sewn togetherwith a running suture. The double-walled tube was suspended on alarge forceps with the suture on one of the tips. At two-thirds of thelength, distal from the sutured edge, one suture was used to createa leaflet and the same was done on the other side of the double-walled tube. A left thoracotomy was performed, cardio-pulmonarybypass installed, the native pulmonary valve excised, and the

in vivo seeded valvular construct implanted as an interposition inthe pulmonary artery. The valves were retrieved after 1 week or 1month in pulmonary position for groups 3 and 4, respectively. Thesheep of groups 1 and 2 only received the valve without priorIP seeding. Similarly, the valves were retrieved after 1 week or1 month for groups 1 and 2, respectively.

Procedure for ovine implant echocardiography

At sacrifice, 7 or 31 days postoperatively, the animals were first pre-medicated with ketamine and anaesthesia was induced with isoflur-ane in oxygen. The animal was then placed on the operation table inright lateral recumbent position and attached to the anaesthesiamachine. The echocardiography was performed by an experiencedmedical doctor using a Vivid Five echo system (GE medicalsystems) and a 2.5 MHz GE Ultrasound probe. Prior to excision ofthe valve, the animal was heparinized to prevent blood clottingand sacrificed with an overdose of pentobarbital and KCl.

Histology and immunohistochemistry

After overnight fixation in 4% paraformaldehyde, the samples wereincubated overnight in 20% sucrose in PBS. The samples were snap-frozen in Neg-50 medium (Prosan, Belgium) and stored at 2808C.Seven micrometre cryosections of the samples were placed onpoly-L-lysine-coated slides and stored at 2208C until staining.

Cross-linked matrix stability assessment was performed by meansof picrosirius red staining13 allowing measurement of the total col-lagen as well as the organized collagen (Figure 2). We measuredthe total area of the pericardium (At, between green lines) andthe area of the red stained material within that area (Ar). Anadditional picture was taken using polarized light to measure theorganization of the collagen. To obtain this picture, the sectionwas placed in between two perpendicularly orientated polarizationfilters, the collagen bundles within the section themselves also actas polarization filters thus causing these bundles to be visible again.On this image, the area of the red illuminated bundles (Ab) ismeasured. The collagen density and organized collagen are calcu-lated by the following formulas.

(1) Collagen density: (Ar/At) � 100(2) Organized collagen: (Ab/Ar) � 100

Both are expressed as a percentage: collagen density as a percen-tage of the total area and organized collagen as a percentage of thecollagen area.

Figure 1 Finalised 21 mm diameter valve construct ready for implantation.

1390 G. De Visscher et al.

This method allows studying the matrix integrity both on amicroscopic and molecular organization level. On the microscopiclevel, it allows assessment of gross structural damagessuch as lacerations and increase or decrease in spongiosity. On themolecular organization level, it allows the study of bundle break-down, an important feature that might occur during cellularinfiltration.

Before immunohistochemistry (IHC) was performed, the presenceof cells was verified using standard haematoxylin and eosin(H&E) staining and microscopic screening. Additional MOVATpentachrome staining was performed to evaluate the compo-sition of connective tissue elements in the newly depositedmatrix.13

The samples were immunohistochemically stained for the follow-ing markers: CD45 (clone: 1.11.32, Serotec, Oxford, UK), CD44(clone: BAT31A, A&E Scientific SPRL, Marcq, Belgium); major histo-compatibility complex (MHC)-I (clone: H58A, A&E Scientific SPRL,Marcq, Belgium); MHC-II (clone: TH14B, A&E Scientific SPRL,Marcq, Belgium); phosphohistone H3 [rabbit polyclonal (06-570Upstate), Biognost, Heule Belgium]; vimentin (clone: V9, DAKO,Heverlee, Belgium); ASMA (clone: 1A4, DAKO, Heverlee, Belgium);smooth muscle heavy chain (SMMS-1, clone: SMMS-1, DAKO,Heverlee, Belgium); CD172a (clone: DH59B, A&E Scientific SPRL,Marcq, Belgium); CD34 (clone QBEnd10, DAKO, Heverlee,Belgium); CD117 [rabbit polyclonal (ab5633), Abcam, Cambridge,UK]; and endothelial nitric oxide synthase (ecNOS, clone: 3, BDPharmingen, Erembodegem, Belgium). The cells from one completeleaflet section were counted for each valve, the length of theleaflet was measured as well as the thickness and surface of thedeposited new material. Length, thickness, and surface area ofthe neomatrix were measured on calibrated micrographs byimage analysis, using an Axioplan 2 imaging microscope (Zeiss,Belgium) and the Axiovision 4.3 software package (Zeiss,Belgium). For neomatrix surface calculation, the surface of theoriginal matrix was subtracted from the total leaflet surface,both measured on the same picrosirius red stained section. Forcellular phenotyping, a total of 500 cells were assessed for eachleaflet and the results were expressed as a percentage. To avoidbias, several pictures from each leaflet were taken, divided intoquarters, and counted according to a randomization list.

Statistical analysis

The sample size of the groups was based on previous experiencesand in accordance, non-parametric statistics were used. A firstanalysis was performed by means of a two-tailed Kruskal–Wallistest for n independent samples (SPSS 11, SPSS Inc., Chicago, IL,USA) for continuous data or Pearson’s x2 test for the ordinal data.P-values less than or equal to 0.10 were considered significantand additional testing was only performed for these groups. Forcontinuous data, exact non-paired two-tailed Wilcoxon–Mann–Whitney tests were used to identify the differences between theindividual groups and P-values less than 0.05 were regarded as sig-nificant. For the ordinal data, Pearson’s x2 test was used andP-values less than 0.05 were regarded as significant. Data wereonly compared when the groups were either within the sameseeding protocol or within the same time frame. Since the IP testsamples were only primed with cells, the observed fractions ofdifferent cells were only compared with the valves made out ofthis material.

Results

Functionality of the ovine valve implants

As shown in Table 1, all animals were within the same rangeof age and weight, a requirement for implantation of a21 mm diameter valve construct. As shown in the sametable, all except one valve construct were adequatelyimplanted. In one valve, one leaflet was caught by thedistal running suture of the end-to-end anastomosis withthe native pulmonary artery, resulting in one small andtwo large leaflets. Assessment of the haemodynamic func-tion was performed but showed no differences betweenany of the groups. Equal gradients and pulmonary insuffi-ciency scores were observed. Regarding thrombus depo-sition, none of the six non-preseeded valves, explanted at1 week, showed valve thrombosis. At 1 month, however,one out of six valves presented an obstructive thrombus.All preseeded valves were free of thrombosis at 1 weekand at 1 month.

Re-cellularization and remodelling of the ovinevalve implants

A four- to seven-fold higher cell count was found in theIP-seeded group as compared with the spontaneouslyseeded group, at 1 week and 1 month, respectively(Table 2 and Figure 3; H&E stain). In and on spontaneouslyseeded valve leaflets, cellularization remained constant.Although a further numerical increase in cell count wasapparent in the IP-preseeded valves at 1 month, thisincrease did not reach a statistical significance.Cell density and the amount of leaflet covered by the new

material were estimated, as shown in Figure 4. A signifi-cantly higher cell density was found in the IP-preseededgroups as compared with the spontaneously seeded groups(upper panel). A similar result was found regarding the cov-erage of the implanted material with newly depositedmaterial (lower panel). Figure 3 shows H&E or MOVAT penta-chrome stained representative sections of both control andIP-preseeded valves after 1 week or 1 month in the pulmon-ary position.The IP-preseeded leaflets showed a significant 15% short-

ening after being 1 month in the pulmonary position.Leaflet thickness of IP-preseeded valves is twice that ofthe spontaneously seeded valves. De novo deposited or

Figure 2 Microscopy images of picrosirius red stained photo-oxidized bovinepericardium with clear field illumination (upper left) and polarized illumina-tion (upper right). The lower panels are representative micrographs of eachgroup. Spontaneous, non-IP-seeded valve group. Each individual micrographis showing the picrosirius red stained collagen and in blue pseudo-color theorganized collagen bundles photographed with polarized light.

In vivo seeding 1391

neomatrix surface is approximately 10-fold larger whencomparing IP-preseeded with spontaneously seeded valves(Table 2 and Figure 3; MOVAT stain). Due to one outlier inthe spontaneously seeded valves at 1 week after implan-tation, statistical significance between both 1 week groups

could not be attained. With respect to the new matrix depo-sition, we found that collagens were present (picrosiriusred, data not shown), but that organization into bundleswas not yet achieved. The blue-green staining of the newmatrix material also revealed the presence of ground sub-stance and proteoglycans.

Matrix stability

Table 2 also summarizes the matrix stability data. No differ-ences in collagen density could be observed. All these datawere within the range of collagen density observed in thetest samples, which were only implanted in the peritonealcavity for 3 days [84.1 (73.2, 90.1)%]. Regarding thepercentage of organized collagen, we found a limited butsignificant decrease in collagen organization in thespontaneously seeded implants in the 1 month group ascompared with the 1 week group. This decrease wasalready apparent in the 1 week IP-preseeded group but didnot change with time. Additionally, we also found that thisdecreased collagen organization was already present after3 days of IP implantation [90.3 (77.6, 94.8)%].

Cell phenotyping in valve constructs

Table 3 summarizes the immunohistochemical cellular phe-notyping of spontaneously seeded and IP-preseeded valves,and IP test sample implanted in the sheep’s peritonealcavity for 3 days. Since these are percentages, only the frac-tion of positive cells is shown, keeping in mind the large

Table 2 Image analysis and matrix stability data

Spontaneous IP-preseeded KWP-value

1 week 1 month 1 week 1 month

Cell count 3753 [1450, 7531] 3345 [1928, 3737] 12126 [7416, 25373]a 20404 [4894, 27898] ,0.01Leaflet length (mm) 16.5 [14.6, 18.2] 16.2 [15.6, 18.2] 17.2 [15.8, 18.7] 14.3 [13.6, 15.4]b,c 0.02Leaflet thickness (mm) 0.60 [0.43, 1.18] 0.60 [0.50, 0.69] 1.23 [0.63, 1.31] 1.32 [0.91, 1.68] 0.04Neomatrix surface (mm2) 0.94 [0.44, 7.74] 1.03 [0.37, 2.04] 10.99 [2.47, 12.58] 9.38 [4.11, 15.57]b 0.02Collagen density (%) 87.7 [82.1, 94.9] 86.6 [80.9, 88.3] 83.3 [78.3, 88.5] 82.3 [74.0, 88.3] 0.50Organized collagen (%) 96.7 [94.9, 97.6] 92.7 [89.8, 94.8]c 93.3 [88.1, 95.4]a 92.3 [85.1, 95.4] 0.10

Data are presented as median [interquartile range]. Spontaneous, non-IP-seeded valve group. KW indicates Kruskal–Wallis test.aSignificant difference between 1 week groups.bSignificant difference between both 1 month groups.cSignificant difference between either both control groups or between both IP-seeded groups.

Table 1 Age, weight, and valve function

Spontaneous IP-preseeded P-value

1 week 1 month 1 week 1 month

Age (days) 737 [384, 753] 705 [368, 713] 575 [458, 756] 653 [581, 708] 0.52 (KW)Weight (kg) 67.5 [66.2, 68.2] 66.0 [46.8, 70.5] 65.0 [59.8, 69.5] 68.0 [66.8, 73.2] 0.16 (KW)Normal function 6/6 6/6 6/6 5/6 0.37 (P)Peak gradient (mmHg) 9.0 [8.3, 22.0] 11.4 [8.0, 18.0] 11.4 [10.3, 23.0] 9.5 [8.4, 26.0] 0.80 (KW)Pulmonary insufficiency 1/4 [0/4, 1/4] 1/4 [1/4, 1/4] 0/4 [0/4, 2/4] 1/4 [1/4, 2/4] 0.19 (P)Thrombus 0/6 1/6 0/6 0/6 0.37 (P)

Data are presented as median [interquartile range]. Spontaneous, non-IP-seeded valve group. KW indicates Kruskal–Wallis test and P indicates Pearson’sx2 test.

Figure 3 Representative H&E and MOVAT stained sections of valves fromeach group. Spontaneous, non-intraperitoneally seeded valve group.

1392 G. De Visscher et al.

differences

intotal

cellsob

served,large

differences

inthe

absolute

celltyp

ecount

areap

parent.

CD44þcells,

absent

inIP

testsam

ples,

showed

asignifi

cantincrease

from1

week

to1

month

inspontaneously

seeded

valvesonly.

Expression

ofMHC-I

and-II

was

low.At1

week,

MHC-I

expression

significantly

differed

betw

eenspontaneously

seeded

andIP-preseed

edvalves.

Phosphohistone

H3

showed

alow

but

consistentpresence

ofdivid

ingcells.

Inthe

spontaneously

seeded

constructs,ASM

Aþ

orvim

entinþ

cellsincreased

significantly

from1

week

to1

month

tolevels

alreadyattained

byIP-preseed

edvalves

at1week.

Atboth

timepoints,

therewas

aninsignifi

cantpresence

ofSM

MS-1þcells.

CD34þcells

were

present

inlow

percentages

inall

groupsbut

notin

allind

ividual

valves.CD117þcells,

Table 3 Immunohistochemical data

In situ period Spontaneous IP test samples IP-preseeded KWP-value

1 week 1 month 3 days 1 week 1 month

CD45 (%) 4.23 [2.75, 6.75] 5.32 [3.06, 9.44] 7.56 [2.93, 13.49] 11.10 [4.54, 14.72] 6.05 [4.53, 10.80] 0.17CD44 (%) 4.81 [1.75, 9.16] 17.03 [8.81, 32.40]a 0.00 [0.00, 0.00] 4.65 [2.02, 8.05]d 8.40 [1.67, 13.19]d 0.02MHC-I (%) 0.69 [0.00, 2.36] 1.47 [0.11, 2.27] 0.17 [0.09, 3.97] 2.55 [2.03, 4.15]b 1.42 [0.93, 3.34] 0.08MHC-II (%) 0.00 [0.00, 0.00] 0.00 [0.00, 0.83] 0.00 [0.00, 1.10] 0.10 [0.00, 1.11] 0.00 [0.00, 0.32] 0.34Phosphohistone H3 (%) 1.29 [0.41, 1.57] 0.74 [0.42, 2.28] 1.30 [0.27, 3.64]c 1.13 [0.51, 1.47] 0.85 [0.53, 1.19] 0.94Vimentin (%) 18.73 [12.48, 27.96] 86.78 [60.61, 91.93]a 13.25 [8.32, 53.74] 59.49 [29.64, 85.09]b 64.85 [37.11, 75.65]d 0.01ASMA (%) 4.83 [1.74, 7.58] 33.07 [21.89, 38.92]a 0.78 [0.00, 5.22] 34.12 [13.56, 51.91]b,d 42.68 [16.27, 56.68]d 0.02SMMS-1 (%) 0.00 [0.00, 0.45] 0.06 [0.00, 0.68] 0.00 [0.00, 0.00] 0.00 [0.00, 0.00] 1.77 [0.00, 5.33]d 0.03CD172a (%) 36.62 [21.67, 58.96] 31.20 [6.55, 61.67] 52.03 [27.67, 71.94] 16.42 [11.68, 22.49] 13.90 [7.69, 39.45] 0.21CD34 (%) 0.00 [0.00, 0.44] 1.69 [0.00, 4.71] 1.49 [0.28, 2.87] 1.09 [0.00, 2.79] 1.92 [0.48, 3.31] 0.21CD117 (%) 0.00 [0.00, 0.00] 0.93 [0.30, 3.01]a 0.37 [0.21, 1.69]c 0.86 [0.52, 1.61]b 1.44 [0.74, 3.88] ,0.01ecNOS 1/6 3/6 0/5 5/6b,d 6/6d 0.01

Data are presented as medians [interquartile range]. Spontaneous, non-IP-seeded valve group; IP test samples, 3 day IP-implanted control patches. KW indicates Kruskal–Wallis test and P indicates Pearson’s x2 test.aSignificant difference between either both control groups or between both IP-seeded groups.bSignificant difference between 1 week groups.cNo statistical analysis could be performed because n , 6.dSignificantly different from IP test samples.

Figu

re4

(A)Cell

density

data

[med

ian(interq

uartilerange)]

fromthe

spontan

eouslyseed

edand

IP-preseed

edphoto-oxid

izedvalve

constructsim

planted

inthe

pulm

onaryposition

foreither

1week

or1

month.

(B)

Leaflet

coverage[(length

ofnew

lydep

ositedmaterial/leafl

et

length)�

100]from

thespontan

eouslyseed

edand

IP-preseed

edphoto-

oxidized

valveconstructs

implanted

inthe

pulm

onaryartery

foreither

1week

or1month.

Invivo

seeding

1393

absent in 1 week spontaneously seeded valves, increasedsignificantly at 1 month, to the level already attained byIP-preseeded valves at 1 week of pulmonary implantation.We realize that the valve tissue treatment as we per-

formed is not optimal for maintaining the endothelium:ideally the valves should have been fixed in situ with para-formaldehyde. However, the initial H&E stains revealedthat we could have an endothelial cell layer, which westained for endothelial NO synthase (ecNOS). Figure 5shows different aspects of the re-cellularization. PanelsA–C show the stainings for vimentin, ASMA, and SMMS-1,respectively. These micrographs illustrate the presence ofblast-like and spindle-shaped vimentinþ cells (A), ASMAþ

spindle-shaped cells, and only limited presence of SMMS-1þ

cells. Panels D and E illustrate the presence of CD117þ orCD34þ cells, respectively. The DAPI-stained section in panelF shows completely re-cellularized photo-oxidized bovinepericardium. Panel G shows an incompletely organized endo-thelial layer on an IP-seeded valve construct that wasimplanted in the pulmonary artery for 1 week. However, at 1month we found organized endothelial layers on the wall andfibrosa side of the leaflet (H) and on both sides of the leaflet(I). Presence of endothelium (ecNOS) was scored as present(1) or absent (0) and summarized in Table 3. At 1 week,IP-preseeded valves had a significantly higher proportion ofendothelialized valves than the spontaneously seeded valves.There was also a significant increase in endothelialization

when compared with the IP test samples, which were comple-tely devoid of any endothelial cell.

Discussion

To provide evidence that the combination of photo-oxidizedbovine pericardium and cells seeded by the short IP implan-tation is suitable as heart valve prosthesis, we implantedthis material in the pulmonary valve position and comparedthem to valves made from the same material but without IPcell seeding. Despite the fact that the valves were hand-crafted and therefore prone to variations, no abnormalvalve function could be detected. This was independent ofthe treatment of the valves, that is, IP-preseeded or not.Only the absence of valve thrombosis suggests the superior-ity of the IP-preseeded valves.

Re-cellularization was clearly dependent upon prior IPseeding of the matrix for three days. IP-preseedingcovered the valves almost completely with new tissue con-taining more cells and new extracellular matrix than couldbe achieved by spontaneous seeding of similar constructs.On the other hand, an average 15% shrinkage of the leafletswas observed in IP-seeded valves after being 1 month in thecirculation. The actual valves were designed to assure ade-quate leaflet coaptation in spite of this 15% shrinkage. Sincecells have infiltrated the original matrix, it was important tolook at the structural integrity of the matrix. No decrease in

Figure 5 (A) Vimentinþ cells (red) with blast-like morphology (outer layer) and spindle shape (interior layer) in IP-preseeded valves after 1 month in thepulmonary position. (B) ASMAþ cells, mainly spindle-shaped cells, in the interior layer in IP-preseeded valves after 1 month in the pulmonary position.(C) Smooth muscle heavy chain myosin positive cells. (D) CD117þ cells on an IP-preseeded valve after 1 week in the pulmonary position. (E) CD34þ cells onan IP-preseeded valve after 1 week in the pulmonary position (CD34, green; nuclei, blue). (F) Completely re-cellularized photo-oxidized matrix (nuclei,blue). (G) One-week-old unorganized endothelial layer (ecNOS, green; nuclei, blue). (H) One-month-old organized endothelial layer on the wall (left) andthe fibrosa side (right) of the leaflet (ecNOS, green; nuclei, blue). (I) One-month-old organized endothelial layer on the fibrosa (left) and ventricularis side(right) of the leaflet (ecNOS, green; nuclei, blue).

1394 G. De Visscher et al.

the amount of collagen was observed, only a small decreasein organization. However, this was coinciding with the repo-pulation of the material and can be caused by the cellsgrowing and migrating in between the collagen fibres, indu-cing a limited dissociation of the collagen structure.Cells having a role in inflammation and the immune

response are present in low percentages. It is interestingthat in spontaneously seeded valves cells positive forCD44, a receptor for hyaluronic acid14 involved in homing,of among others, primitive cells,15 increase three timesfrom 1 week to 1 month after implantation in the pulmonaryartery and are three times higher than the CD45þ cells, apanleukocytic marker.16 The IP test samples were comple-tely devoid of CD44þ cells, whereas the IP-preseededvalves had a significantly larger proportion of these cells.This suggests either a differentiation of these cells or anattraction of cells from the blood. In this respect it is alsonoteworthy to mention the existence of CD452/CD44þ

cells, of bone marrow stromal origin, also expressingmarkers like Stro-1.17

The low presence or even absence of antigen presentingmolecules MHC-I and -II also shows that the material isonly minimally immunogenic.18 To define whether we havea proliferative tissue, we assessed the proliferation rate ofthe cells by staining for phosphohistone H3.19 In general,we found a low mitotic index (around 1%) in all valves,which was not different from the mitotic index of the IPtest samples. It is known that in foreign body reaction thecells are rather attracted to the material than proliferatingin situ.4,20,21 Furthermore, mitotic indices of 10–20% arenecessary for cardiac muscle regeneration,22 whereas insuf-ficient myocardial repair in normal mice relies on cellattraction.23

Since the method of IP preseeding is based on the attrac-tion of immature cells, we have also studied their differen-tiation into (myo)fibroblasts and smooth muscle cells, theoriginal heart valve interstitial cells.7 Three markers wereused:1 vimentin, specific for mesenchymal cells includingfibroblasts;2 ASMA, found in myofibroblasts and smoothmuscle cells; and3 heavy chain myosin, specifically foundin smooth and skeletal muscle cells.7,24 From this phenotyp-ing, we can conclude that a dramatic differentiation of thecells occurs when we compare the 1 week to the 1 monthspontaneously seeded group: there is a six-fold increase inthe amount of myofibroblasts. The IP test samples werealready primed with cells of mesenchymal origin and thisresulted in a differentiation of the cells after beingimplanted in the pulmonary artery for 1 week, which wascomparable to what could only be achieved after leavingan unseeded valve in the same site for 1 month. A small pro-portion of the cells further differentiates into a smoothmuscle cell phenotype, as shown by the approximately 2%cells staining positive for heavy chain myosin in theIP-preseeded valves remaining in the pulmonary positionfor 1 month.Stem cells were attracted by the matrix after IP implan-

tation as well as after implantation in the circulation ofboth unseeded and IP-preseeded valves in the sheep. Inthe first approach, we used a CD172a antibody labellingmacrophages6 and stem cells.9,25 We saw a large presenceof these cells in the all samples, confirming the findings ofCampbell et al.6 Comparing the fractions of CD172aþ andMHC IIþ cells, we see that there is a large difference that

is remarkable because macrophages should be MHC IIþ.26

Two possible explanations are apparent: (i) since thematerial is non-immunogenic and coated with host proteins,the macrophages are not exerting their function by anMHC II-mediated pathway; or (ii) the CD172aþ fraction ismade up of a large proportion of stem/progenitor cells.Therefore, we also studied two markers expressed bystem/progenitor cells. Regarding CD34þ cells, representing0.5% of the cells in peripheral blood,27 we found alreadyan enrichment in the IP test samples. Despite the fact thatsome numerical variation was observed, these cells werenot differently present in any of the other samples. Moreinteresting were the findings with CD117þ cells.28 Duringspontaneous seeding, these cells could only be detectedafter the valves remained in the pulmonary position for 1month. On the contrary, the IP seeding primed the matrixwith a small fraction of these cells, which remainedpresent in the IP-preseeded valves at both time points.For the identification of endothelial cells, we chose an

antibody directed against the ecNOS.29 ecNOS providesadditional information regarding the endothelial function,30

but more importantly the commonly used markers, vonWillebrand Factor (vWF) or CD31, might yield false positivesin this particular setting. Being part of the blood clottingcascade, vWF can insulate the implant surface31 resultingin aspecific staining of the external cell layer. CD31 canalso be expressed by T-cells32 and macrophages,33 one ofthe key cell types in the early foreign body reaction.Scoring the presence of endothelial cells revealed thatIP-preseeded valves had a better potential towardsre-endothelialization than the nude matrix material.Because of the morphology and patchy development of theendothelial layer in the preseeded valves, we think thatthe cells are originating from circulating progenitors thathome to the newly deposited matrix or are attracted bythe deposited cells. In controls, on the other hand, theobserved layer of endothelial cells is the external layer ofpannus, originating from native vessel cell overgrowth.This is an important finding since it is known that innormal aortic and pulmonary valves the endothelium regu-lates the anti-thrombogenicity34 and causes anNO-dependent regulation of the intrinsic contractility ofthe leaflets. Furthermore, myofibroblasts enable this intrin-sic contractility by acting on NO, endothelin-1, and prosta-cyclin released by the endothelium.30

These findings clearly illustrate the potential of thesecells to revitalize the stabilized biological matrix with myo-fibroblasts. Furthermore, the 1 month implants alreadyshowed signs of spontaneous re-endothelialization. Theprocess of re-cellularization seems to be self-limiting,since the amount of newly deposited material is not continu-ously increasing and is getting covered by endothelium. Thisshould prevent new cells from adhering and contributing tothe re-cellularization process.Although we have used a stabilized biological matrix, the

cells obtained by peritoneal seeding are able to modify thismatrix as well as to deposit their own. An important issuethat remains to be studied in this regard is whether or notrepopulation is improving the compliance of the material.This biohybrid valve constructed out of xenogeneic matrix

material and autologous cells combines the reliability of thematrix with the viability of the cells. At present, the originof these cells and the mediators, such as interleukins,

In vivo seeding 1395

integrins, etc. are largely unknown. We know that the cellsare of mesenchymal origin and can differentiate into theappropriate phenotype, i.e. myofibroblast, for cellrepopulation. Considering the mediators, we also knowthat the cells and possibly the differentiation are not sus-tained by a standard cell-culturing medium. At present,the group of Campbell5,6 has shown that the cells are poss-ibly derived from macrophages attracted to the material.The present study provides evidence for our alternativehypothesis that the repopulation, although mediated orinitiated by macrophages, is (haematopoietic) stem cellderived.

Acknowledgements

The project was partially funded by Fonds voor WetenschappelijkOnderzoek (G.0549.06). The authors wish to thank lab assistantsRuth Plusquin, Veerle Leunens, Monique Vercalsteren, and KristofReyniers.

Conflict of interest: none declared.

References

1. Butany J, Fayet C, Ahluwalia MS, Blit P, Ahn C, Munroe C, Israel N,Cusimano RJ, Leask RL. Biological replacement heart valves.Identification and evaluation. Cardiovasc Pathol 2003;12:119–139.

2. Rabkin E, Schoen FJ. Cardiovascular tissue engineering. CardiovascPathol 2002;11:305–317.

3. Campbell JH, Efendy JL, Campbell GR. Novel vascular graft grown withinrecipient’s own peritoneal cavity. Circ Res 1999;85:1173–1178.

4. Butler KR, Benghuzzi HA, Puckett A. Morphometric evaluation oftissue-implant reaction associated with ALCAP and TCP bioceramicsin vivo. J Invest Surg 2001;14:139–152.

5. Campbell JH, Efendy JL, Han C, Girjes AA, Campbell GR. Haemopoieticorigin of myofibroblasts formed in the peritoneal cavity in response toa foreign body. J Vasc Res 2000;37:364–371.

6. Bayes-Genis A, Campbell JH, Carlson PJ, Holmes DR Jr, Schwartz RS.Macrophages, myofibroblasts and neointimal hyperplasia after coronaryartery injury and repair. Atherosclerosis 2002;163:89–98.

7. Della Rocca F, Sartore S, Guidolin D, Bertiplaglia B, Gerosa G,Casarotto D, Pauletto P. Cell composition of the human pulmonaryvalve: a comparative study with the aortic valve—the VESALIO project.Vitalitate Exornatum Succedaneum Aorticum labore IngegnosoObtinebitur. Ann Thorac Surg 2000;70:1594–1600.

8. Kuci S, Wessels JT, Buhring HJ, Schilbach K, Schumm M, Seitz G, Loffler J,Bader P, Schlegel PG, Niethammer D, Handgretinger R. Identification of anovel class of human adherent CD34-stem cells that give rise toSCID-repopulating cells. Blood 2003;101:869–876.

9. Vogel W, Grunebach F, Messam CA, Kanz L, Brugger W, Buhring HJ.Heterogeneity among human bone marrow-derived mesenchymal stemcells and neural progenitor cells. Haematologica 2003;88:126–133.

10. Moore MA. PhotoFix: unraveling the mystery. J Long Term Eff MedImplants 2001;11:185–197.

11. Jansson K, Bengtsson L, Swedenborg J, Haegerstrand A. In vitro endothe-lialization of bioprosthetic heart valves provides a cell monolayer withproliferative capacities and resistance to pulsatile flow. J ThoracCardiovasc Surg 2001;121:108–115.

12. White JK, Agnihotri AK, Latremouille C, Messas E, Carpentier A, TorchianaDF. A method of using the pulmonary trunk to form a trileaflet valve.J Thorac Cardiovasc Surg 2005;129:677–679.

13. Rabkin E, Hoerstrup SP, Aikawa M, Mayer JE Jr, Schoen FJ. Evolution ofcell phenotype and extracellular matrix in tissue-engineered heart

valves during in-vitro maturation and in-vivo remodeling. J Heart ValveDis 2002;11:308–314.

14. Jackson DG. Human leucocyte heparan sulphate proteoglycans and theirroles in inflammation. Biochem Soc Trans 1997;25:220–224.

15. Deguchi T, Komada Y. Homing-associated cell adhesion molecule (H-CAM/CD44) on human CD34þ hematopoietic progenitor cells. Leuk Lymphoma2000;40:25–37.

16. Sunderland CA, McMaster WR, Williams AF. Purification with monoclonalantibody of a predominant leukocyte-common antigen and glycoproteinfrom rat thymocytes. Eur J Immunol 1979;9:155–159.

17. Clark BR, Keating A. Biology of bone marrow stroma. Ann N Y Acad Sci1995;770:70–78.

18. Ketchedjian A, Jones AL, Krueger P, Robinson E, Crouch K, Wolfinbarger LJr, Hopkins R. Recellularization of decellularized allograft scaffolds inovine great vessel reconstructions. Ann Thorac Surg 2005;79:888–896.

19. Brenner RM, Slayden OD, Rodgers WH, Critchley HO, Carroll R, Nie XJ,Mah K. Immunocytochemical assessment of mitotic activity with anantibody to phosphorylated histone H3 in the macaque and humanendometrium. Hum Reprod 2003;18:1185–1193.

20. Tang L, Jennings TA, Eaton JW. Mast cells mediate acute inflammatoryresponses to implanted biomaterials. Proc Natl Acad Sci USA 1998;95:8841–8846.

21. Tang L, Eaton JW. Inflammatory responses to biomaterials. Am J ClinPathol 1995;103:466–471.

22. Leferovich JM, Bedelbaeva K, Samulewicz S, Zhang XM, Zwas D, LankfordEB, Heber-Katz E. Heart regeneration in adult MRL mice. Proc Natl AcadSci USA 2001;98:9830–9835.

23. Askari AT, Unzek S, Popovic ZB, Goldman CK, Forudi F, Kiedrowski M,Rovner A, Ellis SG, Thomas JD, DiCorleto PE, Topol EJ, Penn MS. Effectof stromal-cell-derived factor 1 on stem-cell homing and tissue regener-ation in ischaemic cardiomyopathy. Lancet 2003;362:697–703.

24. Yperman J, De Visscher G, Holvoet P, Flameng W. Molecular andfunctional characterization of ovine cardiac valve-derived interstitialcells in primary isolates and cultures. Tissue Eng 2004;10:1368–1375.

25. Handgretinger R, Gordon PR, Leimig T, Chen X, Buhring HJ, NiethammerD, Kuci S. Biology and plasticity of CD133þ hematopoietic stem cells. AnnN Y Acad Sci 2003;996:141–151.

26. Ahn JS, Konno A, Gebe JA, Aruffo A, Hamilton MJ, Park YH, Davis WC.Scavenger receptor cysteine-rich domains 9 and 11 of WC1 are receptorsfor the WC1 counter receptor. J Leukoc Biol 2002;72:382–390.

27. Krause DS, Fackler MJ, Civin CI, May WS. CD34: structure, biology, andclinical utility. Blood 1996;87:1–13.

28. Rombouts WJ, Ploemacher RE. Primary murine MSC show highly efficienthoming to the bone marrow but lose homing ability following culture.Leukemia 2003;17:160–170.

29. Laufs U, Endres M, Stagliano N, Amin-Hanjani S, Chui DS, Yang SX,Simoncini T, Yamada M, Rabkin E, Allen PG, Huang PL, Bohm M, SchoenFJ, Moskowitz MA, Liao JK. Neuroprotection mediated by changesin the endothelial actin cytoskeleton. J Clin Invest 2000;106: 15–24.

30. Pompilio G, Rossoni G, Sala A, Polvani GL, Berti F, Dainese L,Porqueddu M, Biglioli P. Endothelial-dependent dynamic and antithrom-botic properties of porcine aortic and pulmonary valves. Ann ThoracSurg 1998;65:986–992.

31. Pankowsky DA, Ziats NP, Topham NS, Ratnoff OD, Anderson JM.Morphologic characteristics of adsorbed human plasma proteins onvascular grafts and biomaterials. J Vasc Surg 1990;11:599–606.

32. Caligiuri G, Rossignol P, Julia P, Groyer E, Mouradian D, Urbain D, Misra N,Ollivier V, Sapoval M, Boutouyrie P, Kaveri SV, Nicoletti A, Lafont A.Reduced immunoregulatory CD31þ T cells in patients with atherosclero-tic abdominal aortic aneurysm. Arterioscler Thromb Vasc Biol 2006;26:618–623.

33. Curat CA, Miranville A, Sengenes C, Diehl M, Tonus C, Busse R,Bouloumie A. From blood monocytes to adipose tissue-resident macro-phages: induction of diapedesis by human mature adipocytes. Diabetes2004;53:1285–1292.

34. Ku DD, Nelson JM, Caulfield JB, Winn MJ. Release of endothelium-derivedrelaxing factors from canine cardiac valves. J Cardiovasc Pharmacol1990;16:212–218.

1396 G. De Visscher et al.