lable at ScienceDirect

ARTICLE IN PRESS

Biomaterials xxx (2009) 1–9

Contents lists avai

Biomaterials

journal homepage: www.elsevier .com/locate/biomateria ls

The remodeling of cardiovascular bioprostheses under influence of stem cellhoming signal pathways

Geofrey De Visscher a,*, An Lebacq a, Lindsay Mesure a, Helga Blockx a, Ilse Vranken a, Ruth Plusquin a,Bart Meuris a, Marie-Christine Herregods b, Hans Van Oosterwyck c, Willem Flameng a

a Laboratory for Experimental Cardiac Surgery, Department of Cardiovascular Diseases, Katholieke Universiteit Leuven, Leuven, Belgiumb Division of Cardiology, Katholieke Universiteit Leuven, Leuven, Belgiumc Division of Biomechanics and Engineering Design, Katholieke Universiteit Leuven, Leuven, Belgium

a r t i c l e i n f o

Article history:Received 7 August 2009Accepted 4 September 2009Available online xxx

Keywords:BioprosthesisCell adhesionFibronectinHeart valve & Vascular grafts

* Corresponding author. Laboratory for Experimenment of Cardiovascular Diseases, Katholieke Uerbroedersstraat 17, 3000 Leuven, Belgium. Tel.: þ337122.

E-mail address: geofrey.devisscher@med.kuleuven

0142-9612/$ – see front matter � 2009 Elsevier Ltd.doi:10.1016/j.biomaterials.2009.09.016

Please cite this article in press as: De Visschsignal pathways, Biomaterials (2009), doi:10

a b s t r a c t

Optimizing current heart valve replacement strategies by creating living prostheses is a necessity toalleviate complications with current bioprosthetic devices such as calcification and degeneration.Regenerative medicine, mostly in vitro tissue engineering, is the forerunner of this optimization search,yet here we show the functionality of an in vivo alternative making use of 2 homing axes for stem cells. Inrats we studied the signaling pathways of stem cells on implanted bioprosthetic tissue (photooxidizedbovine pericardium (POP)), by gene and protein expression analysis. We found that SDF-1a/CXCR4 andFN/VLA4 homing axes play a role. When we implanted vascular grafts impregnated with SDF-1a and/orFN as carotid artery interpositions, primitive cells were attracted from the circulation. Next, bioprostheticheart valves, constructed from POP impregnated with SDF-1a and/or FN, were implanted in pulmonaryposition. As shown by CD90, CD34 and CD117 immunofluorescent staining they became completelyrecellularized after 5 months, had a normal function and biomechanical properties and specifically thecombination of SDF-1a and FN had an optimal valve-cell phenotype.

� 2009 Elsevier Ltd. All rights reserved.

1. Introduction

The preferred diseased heart valve treatment is repair, with themajority being replaced by artificial ones and even if valve repair isfeasible it is not necessarily durable [1]. Synthetic prostheses aredurable but anticoagulation related bleeding complications arelife-threatening. These are significantly reduced by the use ofbiological tissue prostheses. However, their non-living structureremains responsible for its reduced long term durability. Tissueengineering was suggested to overcome these complications, butissues like the lack of immediate availability, potential problemswith autologous cell culturing and uncertain durability hamperthis technique.

Therefore we investigate the combination of ‘in vivo’, insteadof ‘in vitro’, tissue engineering using the existing durable

tal Cardiac Surgery, Depart-niversiteit Leuven, Mind-

32 16 337120; fax: þ32 16

.be (G. De Visscher).

All rights reserved.

er G, et al., The remodeling o.1016/j.biomaterials.2009.09

bioprosthetic as a core matrix. The basic idea was to attract adultcirculating progenitors and stem cells to the implanted functioningbioprosthetic valve, acting as its own bioreactor. We previouslyshowed that collagen containing biomaterial attracts stem cells invivo when implanted intraperitoneally [2]. Microarray baseddifferential gene expression generated insights into the differentdevelopmental pathways of this tissue [3], from which the usedhoming proteins are derived. As shown heart valves from intra-peritoneally preseeded POP become living tissue derived frommacrophages and peritoneal cells and in its mature dispositioncomprises EC matrix and (myo)fibroblasts similar to native valves[4,5]. The first part of this study unravels the homing signal path-ways of stem cells related to intraperitoneal implantation andshows a major role of Sdf-1a/Cxcr4 and a4-integrin-Vcam1/fibro-nectin (Fn) homing axes. This is not surprising knowing thatSDF-1a/CXCR4 together with the compensatory a4-integrin-VCAM1/FN axis [6], are crucial in homing and engraftment of adultprogenitor and stem cell therapies [6,7]. Additionally the SDF-1a/CXCR4 axis is necessary in embryogenesis [8], more particularly fordevelopment of heart and vascular tissue throughout the body[9,10] and plays a role in the homing of smooth muscle progenitors[11]. SDF-1a has been well preserved during evolution with only 1

f cardiovascular bioprostheses under influence of stem cell homing.016

G. De Visscher et al. / Biomaterials xxx (2009) 1–92

ARTICLE IN PRESS

amino acid difference between humans and mice [8]. The sameholds for fibronectin and a-integrins being present in evolutionarydistant species [12,13].

Providing homing capacity by pre-activating a target tissue [7]has been proposed in myocardial infarction by direct delivery orimplanting SDF-1a producing cells [14], which resulted in CD117þ

cell homing and recellularisation [14]. Noteworthy is that SDF-1a,released from platelets, mediates homing of bone marrow derivedprogenitors was observed in the arterial bed [15].

Consequently, we implemented these pathways into our heartvalve cellularization paradigm by pre-activating, impregnationwith the mentioned trigger peptides, bioprostheses for stem cellhoming and assessed the primitive cell attraction after implan-tation in the circulation. Finally, recellularisation and cell pheno-types were studied on pulmonary bioprosthetic heart valvesimpregnated with FN, SDF-1a or both and implanted for 5 monthsinto sheep.

2. Materials and methods

2.1. SDF-1a/CXCR4 and FN/VLA4 gene and protein expression

SDF-1a – CXCR4 and alpha4 integrin – fibronectin homing pathways wereexamined in previously described intraperitoneal primitive cell homing [2]. Here,POP patches (Cardiofix�, Sulzer Carbomedics) were harvested after 1.5 or 3 days.Neotissue-derived and intraperitoneal lavage (control) cells were collected for RNAextraction and PCR. The neotissue-derived supernatant obtained after centrifugation(2200 G) of the explants was analyzed by ELISA. IP fluid served as control, plasmaand bone marrow as technical control.

Total RNA was extracted from similar 1.5 and 3 day neotissue and control cellsusing Trizol (Invitrogen) and RNeasy Mini RNA isolation columns (Qiagen) accordingto the University of Nebraska Medical Center Microarray Core Facility protocol(http://www.unmc.edu/microarray). RNA integrity and purity/concentration werechecked using the RNA 6000 Nano Kit with the Agilent 2100 Bioanalyzer (AgilentTechnologies). 100 ng total RNA was reversely transcribed with the MJ Mini GradientThermal Cycler (Bio-Rad) using 100 mM random hexamer primers (Roche AppliedScience) and 300 units revertAid H Minus M-Mulv Reverse Transcriptase Enzyme(MBI Fermentas).

The PCR primers were designed, accounting for intron-exon structures toexclude genomic DNA contamination, using Primer3 software (http://frodo.wi.mit.edu/cgi-bin/primer3/primer3_www.cgi):

b-actin: L AGATTTGGCACCACACTTTCTACR AACACAGCCTGGATGGCTAC

Sdf-1a: L AGGTCGTCGCTGTGCTGR GATGTTTGACGTTGGCTCTG

Cxcr4: L GTGCAGCAGGTAGCAGTGAR ATAGATGGTGGGCAGGAAGA

Fn: L AGACCCCAGGCACCTATCACR TCGGTCACTTCCACAAACTG

Vla-4: L ATAGCGGGGACCTTGACCR AACTTTTGGGTGTGGCTCTG

The polymerase chain reaction (PCR) was performed using 2.5 units recombi-nant Taq DNA polymerase (MBI Fermentas) on an MJ Mini Gradient Thermal Cycler(denaturation: 2 min at 94 �C; 30 cycli: 30 sec at 94 �C, 1 min at 60 �C and 45 sec at72 �C; final elongation: 10 min at 72 �C). 5 ml of the Sdf-1a, Cxcr4, Fn or Vla-4 PCR-product was mixed with 5 ml of the b-actin PCR-product (positive control andnormalization). 1 ml of each mix was analyzed using the DNA 1000 Kit with the 2100Bioanalyzer (Agilent Technologies).

SDF-1a-protein in 1.5 and 3d intraperitoneal implants was assessed by ELISA,using normal intraperitoneal fluid and blood plasma as control. 96 well plates(Microlon 200, Greiner Bio-One) were incubated (overnight at room temperature(RT)) with 100 ml/well of 2 mg/ml (in PBS) SDF-1a capture antibody (79018, R&DSystems) and blocked with 200 ml 1% BSA and 5% sucrose in PBS at RT for 1 h.100 ml of the prepared samples as well as a standard (mouse SDF-1a, 200–6400 pg/ml) was applied to the plate in duplicate and incubated 2 h at RT. Thewells were incubated 90 min (RT) with 100 ml of 200 ng/ml biotinylated SDF-1adetection antibody (polyclonal, Serotec) and developed with streptavidin-HRP(20 min, RT) and TMB (Bio-Rad; 30 min, RT). The plates were read (Multiskan EXmicroplate reader, Thermo Electron Corp.) at 450 and 540 nm for backgroundcorrection.

Please cite this article in press as: De Visscher G, et al., The remodelingsignal pathways, Biomaterials (2009), doi:10.1016/j.biomaterials.2009.09

Total protein content was determined by the Coomassie protein assay (PerbioScience), referenced to standard protein dilutions (Bovine Serum Albumin). Absor-bance was measured at 595 nm in the microplate reader.

Explant samples, snap-frozen in Tissue Freezing medium (Leica) with liquidnitrogen, were cryosectioned (7 mm) on a Microm HM500 OM cryostat (Prosan).Immunohistochemistry was performed with the antibodies to CXCR4 (polyclonal,Abcam), FN (clone IST-9, Abcam), SDF-1a (polyclonal, Abcam) and a4 integrin (cloneTA-2, Serotec). FITC-conjugated secondary antibodies were used for detection. Theslides were mounted using Vectashield with DAPI (Vector Laboratories). For nega-tive controls primary antibodies were omitted.

2.2. Matrix coating by impregnation

Two pairs of 25 mm2 POP patches were impregnated with either 3.2 mg/cm2

SDF-1a or 32 mg/cm2 FN in 400 ml PBS for 0, 6 and 24 h. Thereafter matrices werefrozen, sectioned and stained for FN or SDF-1a as described above.

Binding efficiency was assessed by incrementally impregnating POP (tripli-cates) with SDF-1a (0.16,0.8,1.6 or 3.2 mg/cm2), FN (1.6,8,16 or 32 mg/cm2) or thecombination thereof. Veritas collagen matrices (Synovis) were used as controls.SDF-1a binding was assessed in SDF-1a and FN-SDF-1a coated samples witha modified ELISA assay. FN was assessed in the FN coated samples. Fixed (0.4%paraformaldehyde) matrices were incubated with 100 ml of polyclonal bio-tinylated anti-human SDF-1a (Serotec) at 200 ng/ml in PBS (1% BSA, pH 7.4) or100 ml of HRP conjugated anti-FN antibody (1/1000 in PBS (1% BSA, pH 7.4)) for1.5 h or 2 h at RT, respectively. For SDF-1a detection 100 ml of streptavidin-HRP (1/1000 in PBS (1% BSA, pH 7.4)) was added for 30 min. The assay was developedwith 100 ml TMB solution (Bio-Rad) for 30 min and stopped by adding 100 ml 1 NH2SO4 solution. Absorption was determined at 450 nm and at 540 nm to correctfor optical imperfections. Matrices were collected and dried to determine watercontent.

The SDF-1a dosages were chosen around the theoretical maximum of0.57 mg/cm2 for double-sided coating, calculated as a monolayer using thefollowing parameters: M: 8 kDa, theoretical density [16]: 1.49 g/cm3 and a 3D/2Dsphere packing coefficient: resp. 0.7405 and 0.9069. An experimental maximumwas established at 0.99 mg/cm2, yet 3.2 mg/cm2 was preferred to ensuresaturation.

2.3. Carotid artery implants

24 Male Wistar rats (380–400 g) were selected and cared for according to the‘‘Guide for the Care and Use of Laboratory Animals’’ (NIH publication 85-23, revised1985). Local Ethics Committee approval was obtained.

POP (�0.3 cm2) was impregnated (24 h, 4 �C) with murine SDF-1a (3.2 mg/cm2,Peprotech EC), rat FN (32 mg/cm2, Biomedical Technologies) or a combinationthereof and unimpregnated controls were used. Carotid artery grafts were madeusing a 26 G Vialon IV catheter (BD Biosciences) for sizing. Grafts were implantedunder 2% isoflurane anesthesia (100% O2) using the 7 o’clock stitch technique [17].Neck muscles were closed with a running suture (Ticron 3-0) and the skin wassutured intradermally (Ticron 3-0) to avoid opening of the wound by grooming.After recovery the animals were placed in individual cages.

Explanted 24 h later, the grafts were filled with and embedded in TissueFreezing medium (Leica), snap-frozen in liquid nitrogen and cryosectioned asdescribed above as was the staining and quantification for the following markers [2]:CD90 (OX-7, Serotec, Oxford, UK), c-kit (H-300, Santa Cruz, Heidelberg, Germany)and CD34 (QBEnd 10, DAKO, Heverlee, Belgium).

2.4. Pulmonary valve implants

Selected young pure bred female ‘‘Lovenaar’’ sheep (n¼ 24, 264 [133, 395] days,51[40, 63] kg) were cared for by a veterinarian, according to the NIH publication 85-23 (revised 1985). Local Ethics Committee approval was obtained. They wererandomly allocated to 3 groups of 8 animals, each receiving pulmonary implants ofPOP valves, as described before [4], impregnated (24 h, 4 �C) with human SDF-1a(80 mg, Peprotech EC), human FN (800 mg, Biomedical Technologies) or a combina-tion thereof. Native valve controls were harvested from non-conflictingexperiments.

Before sacrifice all animals were echocardiographically examined (Vivid Five &2.5-MHz GE Ultrasound probe, GE Medical Systems). The mean gradient andregurgitation score were assessed with continuous-wave or colour flow Dopplerechocardiography, respectively.

Longitudinal valve cryo-sections (7 mm) were immunohistochemically stainedand quantified, as described before [5], for the following markers: ASMA (1A4,DAKO), desmin (D33, Serotec), SMMS-1 (DAKO), smoothelin (polyclonal, SantaCruz), vimentin (V9, DAKO), and ecNOS (3, BD Pharmingen).

2.5. Microscopy

An Axioplan 2 imaging microscope with an Axiocam MRc5 camera (Zeiss) andthe following objective lenses; Plan-NEOFLUAR 1x/0.025, Plan-APOCHROMAT 10x/

of cardiovascular bioprostheses under influence of stem cell homing.016

G. De Visscher et al. / Biomaterials xxx (2009) 1–9 3

ARTICLE IN PRESS

0.45 and 20x/0.75, was used. Image analysis and morphometry were performedwith Axiovision Rel. 4.4 software.

2.6. Statistical analysis

Complication rates were compared with an exact Pearson’s c2 test (StatXact4.0.1, Cytel Inc). Continuous data from the different groups (rat or sheep study) werefirst compared with a Kruskal-Wallis test. If the obtained p-value was lower than0.05, between group comparisons were performed with a Wilcoxon-Mann-Whitneytest (SPSS 15.0 for Windows, SPSS Inc.).

3. Results

3.1. Sdf-1a/Cxcr4 and Fn/Vla4 homing axes expression

Neotissue derived cells from 1.5 to 3 days intraperitoneallyimplanted POP patches in rats were compared to steady stateperitoneal cells and not unimplanted material because this wasacellular. Simultaneous loading of the gene of interest and the b-actin control (Fig. 1a) allowed assessment of the genes in these 3conditions. The Cxcr4-gene, the Sdf-1a receptor, and those of theFn-Vla-4 homing axes were present and equally expressed. Sdf-1a

was not present in steady state peritoneal cells, in contrast to bothimplant groups (Fig. 1b). The soluble SDF-1a fraction detected byELISA (Fig. 1c) showed an increased level in the 3 day implants only.No soluble SDF-1a could be detected in the controls and the 1.5 dayexplants. As a control of our ELISA, SDF-1a was absent in bloodplasma from the same rats but high levels were observed in bonemarrow supernatant, the original source [18]. Immunofluorescentdetection of the 4 proteins confirmed the finding for the implants

Fig. 1. Gene and protein expression of intraperitoneal implant adhered cells. (a) Electrophcontrols). (b) Sdf-1a gene expression (BA corrected) of 1.5 and 3 day intraperitoneal implexpression (total protein corrected) measured by ELISA assay (n¼ 3/group). (d) From leftintraperitoneal implants (bar¼ 100 mm). POP designates the implanted acellular photooxid

Please cite this article in press as: De Visscher G, et al., The remodeling osignal pathways, Biomaterials (2009), doi:10.1016/j.biomaterials.2009.09

after 3 days of implantation (Fig. 1d). The SDF-1a panel clearlyshows the presence of the cell bound form of this protein and thecells expressing the protein appear in clusters. CXCR4þ cells wereclearly detected, although to a lesser extent than those expressingSDF-1a. FN was ubiquitously present in the matrix materialdeposited by the cells and VLA-4 expressed by approximately 8% ofthe cells adhering to the implants.

Because both homing axes were present and SDF-1a was evendifferentially expressed both at gene and protein level, we assessedwhether the homing proteins, SDF-1a and FN, could be coated byimpregnation on POP to be used as an experimental cardiovascularprosthesis.

3.2. Impregnation coating of POP with FN and/or SDF-1a

Impregnation of POP (12.6 [9.1, 17.8] mg dry weight/cm2) withFN (32 mg/cm2) and/or SDF-1a (3.2 mg/cm2) could be achieved asillustrated in (Fig. 2). Modified ELISA showed dose-dependentincreases in the amounts of FN (Fig. 2a) or SDF-1a (Fig. 2b). For FNthe increments were measurable in all concentrations and FNadhesion was higher in POP as compared to collagen patches (SYN).SDF-1a was present in comparable amounts in the SDF-1a and FN-SDF-1a co-impregnation group but only from 1.6 mg/cm2 onwards.For SDF-1a the collagen patch showed similar impregnationproperties as POP. Lacking absolute quantification these data wereconfirmed by immunofluorescent stained sections of the high doseimpregnation of FN or SDF-1a (Fig. 2c). Co-impregnation was notevaluated because if FN could be coated then FN would be able to

erogram of an Sdf-1a and b-actin (BA) containing sample (M15 and M1500: loadingants compared to steady state intraperitoneal cells (n¼ 3/group). (c) SDF-1a proteinto right, SDF-1a, CXCR4, FN and VLA-4 immunofluorescent stained sections of 3 dayized pericardium.

f cardiovascular bioprostheses under influence of stem cell homing.016

Fig. 2. Impregnation coating of POP. (a) Bound FN by impregnation with incremental doses of FN (n¼ 3/dose). 32 SYN represents an unfixed collagen patch impregnated with 32 mg/cm2 FN. (b) Bound SDF-1a on samples impregnated with the molecule and on samples precoated with FN (FN-SDF). (c) Immunofluorescent stained sections of FN or SDF-1aimpregnated POP (bar¼ 100 mm).

G. De Visscher et al. / Biomaterials xxx (2009) 1–94

ARTICLE IN PRESS

bind SDF-1a and present it to cells in a natural way [19]. In contrastto control (T0), a layer of either protein was present at respectively6 and 24 h. This shows that the molecules, present in our model ofin vivo stem cell attraction, can be coated to bioprosthetic materialby impregnation.

Please cite this article in press as: De Visscher G, et al., The remodelingsignal pathways, Biomaterials (2009), doi:10.1016/j.biomaterials.2009.09

3.3. Primitive cellattraction to FN and/or SDF-1a coated vascular grafts

The influence on the cells that home or bind FN and/or SDF-1acoated POP was assessed on blood vessel implanted into thecirculation (Fig. 3a). A longitudinal vascular graft section, implanted

of cardiovascular bioprostheses under influence of stem cell homing.016

G. De Visscher et al. / Biomaterials xxx (2009) 1–9 5

ARTICLE IN PRESS

in the carotid artery for 24 h, is shown (Fig. 3b). The coating did notchange the amount of cells residing on the luminal surface of thesegrafts (Fig. 3c; p¼ .297). However, focusing on stem cells markers,already used to identify these cells in intraperitoneal implants [2],and comparing controls to impregnation coated (FN, SDF-1a or thecombination) groups, we observed significant differences in thefraction of CD90þ, CD34þ and CD117þ cells (Fig. 3d–f, p¼ .017, .006and<.001 respectively). CD90, a marker of stromal cells, a subset ofhematopoietic and mesenchymal stem cells [20] and very fewperipheral blood cells [21], was significantly increased on bloodvessel grafts impregnated with FN (p¼ .002) and FN in combinationwith SDF-1a (p¼ .041). For CD34, expressed by hematopoieticstem/progenitor and endothelial precursor cells [22], both the FN(p¼ .002) and FN (p¼ .002) with SDF-1a impregnated blood

Fig. 3. Cell homing to a vascular implant. (a) From left to right an overview of the surg(b) Representative longitudinal section of a vascular implant 24 h after implantation (flow: fadventitial side of the implant (n¼ 6/group). (d–f) CD90þ, CD34þ and CD117þ cells, respectivsignificant difference from control or FN & FN-SDF respectively).

Please cite this article in press as: De Visscher G, et al., The remodeling osignal pathways, Biomaterials (2009), doi:10.1016/j.biomaterials.2009.09

vessels showed significantly higher fractions. Finally CD117 or c-kit,expressed by hematopoietic stem cells, committed progenitors,circulating immature cells and mesenchymal stem cells [20,23],was significantly increased in the FN (p¼ .002), SDF-1a (p¼ .002)and the combination (p¼ .002) thereof compared to controls. Wealso observed a somewhat lower fraction in the group impregnatedsolely with SDF-1a compared to the FN (p¼ .015) and FN with SDF-1a (p¼ .002) impregnated groups.

Because the used proteins mediate in immune responses andwound healing we assessed the presence of leucocytes (CD45), T-cells (CD3), B-cells (CD72), macrophages (CD11b, CD68), gran-ulocytes (His48) and mast cells (TBO) (Supplementary Table 1). Nodifferences between the impregnation groups mutually andcontrols could be observed.

ical technique used to implant the handmade POP vascular implant (scale in mm).rom left to right). (c) Cell counts of the luminal area, excluding the original carotid andely, on FN and/or SDF-1a coated grafts compared to uncoated controls (* and # indicates

f cardiovascular bioprostheses under influence of stem cell homing.016

G. De Visscher et al. / Biomaterials xxx (2009) 1–96

ARTICLE IN PRESS

3.4. Self-seeding matrices as functional heart valves

Next we investigated how bioprosthetic valves impregnated inthe same way as the vascular grafts recellularize when implantedfor 5 months into pulmonary position. How they recellularizespontaneously has already been published [4,5] and the controlrange (untreated 5 month POP implants) is added to the respectivegraphs. Here we compared the 3 impregnations to native ovinepulmonary heart valves.

20/24 sheep were endocarditis free and the incidence didn’tdiffer between groups (p¼ .575) at 5 months after implantation.The mean pressure gradients, across the prosthetic valve, were 3.8[1.0; 8.8], 2.7 [1.8; 4.5] and 5.7 [3.8; 6.8] mmHg for the FN, SDF-1aand combination group respectively. The group coated with SDF-1ahad a significantly lower mean gradient than the group impreg-nated with the combination of FN and SDF-1a. No valve regurgi-tation (score� 1/4), echocardiographic signs of valve thrombosis ordysfunction were observed in any of the groups.

At 5 months we found a complete recellularisation of the leaf-lets, the cell phenotype resembling those of native valves. Fig. 4aand b show the recellularisation and the differentiation betweenimplanted (pseudo-blue) and newly deposited (red) matrix ofa valve leaflet coated with both FN and SDF-1a. The recellularisa-tion was further visualized by plotting the position of the nuclei,thus plotting the position of every individual cell detected by theimage analysis system. The differentiation between the implantedand newly deposited matrix was obtained from Sirius red stainedsections imaged with normal and polarized light. The whole leafletwas covered and grown in by cells, interstitial as well as endothelialcells (Fig. 4c). As a first quantitative assessment of recellularisation,the number of cells of one leaflet per valve (3 section average) wascounted (Fig. 4d) and compared to the cell counts of 10 individualnative ovine pulmonary valves. The observed cell number in theleaflets after 5 months as a functioning valve didn’t differ betweenthe groups but more importantly it didn’t differ from the countobserved in native ovine pulmonary valve leaflets (p¼ .192). Thephenotype of these cells and their population fraction in the leafletsis an important issue, especially in the light the possible correlationof these features with valve abnormality and pathology [4,5].Vimentin, a general mesenchymal cell marker [24], present infibroblasts, myofibroblasts and smooth muscle cells, was notdifferentially expressed between the groups and didn’t deviate fromthe values observed in native valves (Fig. 4e, p¼ .358). With respectto the alpha smooth muscle actin expressing cell fraction (Fig. 4f) wefound significant differences (p< .001) between the groups. All ofour impregnated valves had fractions of ASMAþ cells exceeding thatof native valves, but interestingly the co-impregnation group (5.8[4.3; 14.8]%) showed a very low level in these cells compared to FN(24.1 [10.8; 40.4]%) and SDF-1a (55.8[40.1;63.3]) individually. Bothmarkers for smooth muscle cells, desmin (Fig. 4g, p< .001) andsmoothelin (Fig. 4h, p< .001), showed to be moderately increased inthe impregnated groups compared to native valves where thesecells are virtually absent. However, significant shortening of theleaflets was not observed (Supplementary Table 2).

Another important issue with respect to the coating andimplantation of the biological prostheses is the impact on theimmune response and its potential role in the degeneration ofbioprosthetic material. To asses this, the impregnation groups werecompared to each other, but not to the native valves because noinflammatory cells are to be expected in normal valve tissue(Supplementary Table 2). Leucocytes (CD45þ) showed a limited andequal presence in all groups. Further phenotyping revealed verysmall fractions of T-cells (CD8 and CD25), B-cells (B-B1) andcomplete absence of mast cells (TBO). Macrophages (CD11b, CD14and iNOS), important in implant related immune response, showed

Please cite this article in press as: De Visscher G, et al., The remodelingsignal pathways, Biomaterials (2009), doi:10.1016/j.biomaterials.2009.09

to be present in variable yet limited amounts in all impregnatedvalves. Obviously this limited inflammatory reaction was associatedwith normal tissue properties (leaflet length, thickness, depositedmatrix, collagen density and organization), absence of calcification,biochemical properties (collagen, elastin and free amine content) orbiomechanical properties (shrinkage temperature, E-modulus,Fmax and strength) and no differences were detected between thegroups.

4. Discussion

Building on experiences with primitive cells homing in a modelof intraperitoneal implantation of collagen based material [2], westudied and confirmed the presence of 2 major and compensatoryhoming axes for primitive cells [6] in this specific situation. SDF-1agene and protein presence in our model even showed the typicaltime lag between gene and protein expression. Whereas SDF-1awas confined to a subset of cells, FN seemed ubiquitously present inthe matrix deposited by cells adhering to the implanted material.Within the same primitive cell homing model we confirmed thepresence of cells expressing the relevant receptors, CXCR4 and VLA-4. Interestingly, the SDF-1a mediated axis was activated in a verysimilar approach where host stem/progenitor cells were recruitedby an activated collagen graft resulting in increased vascularizationof the hind limb [25].

Implementing and enhancing this natural process by impreg-nating or coating biologic heart valves, we provide a novel way tocreate autologous cell adhesion and tissue development in situ.Before addressing this ultimate goal we needed to address 2 majorissues. Firstly, was the matrix material coatable with FN, SDF-1a orthe combination of both. FN contains a collagen binding domain[26] and is known to present SDF-1a to cells, thereby enhancingtheir migration [19]. Furthermore, SDF-1a binds other constituentsof the extracellular matrix [27] and collagen IV [28]. Thus FNfacilitates SDF-1a binding to matrices, here POP (90% collagen,predominantly type I [29]), chosen because it is completely acel-lular and used as a cardiovascular repair patch. Binding of FN and ofSDF-1a with or without FN precoating was shown by both ELISAand immunofluorescent staining. This proved that POP could becoated with the proteins, laying at the origin of both primitive cellhoming axes. The second issue was critical because we needed toprove that primitive cells home differentially or rather preferen-tially to FN, SDF-1a or FN-SDF-1a coated matrices. The coated graftsdidn’t enhance cell adhesion to vascular implants; however theyaltered the profile of adhered cell phenotypes. The increased cellfractions expressing primitive cell markers, especially CD34 andCD117, showed differential homing in all coated groups except forCD34 in the matrices coated with SDF-1a only. Even for CD117þ

cells SDF-1a coating had a smaller effect. Overall the observationsshowed coating induced preferential homing of primitive cells,something already shown for CD34þ progenitors in vitro [30].Alternatively increased primitive cell presence could be explainedby enhanced proliferation (up to 5-fold within in 24 h), but this isunlikely for cells having a slow proliferation rate [31] and thetendency to differentiate after cell division [32]. Also, it is knownthat SDF-1a has anti-apoptotic effects [8], however the differencesbetween the coated grafts and controls would require apoptotic cellfrequencies several times those reported (<2% within this time-frame) in arterial remodeling [33].

Most important is the application of the coatings to heart valveprostheses and the results over a longer period of time. Previously,we constructed POP valves and implanted them as pulmonary valvesubstitutes in sheep [4,5], where they functioned well for 5 monthsbut spontaneous recellularisation was insufficient and contractilecells were overrepresented [5]. In comparison, coated valves were

of cardiovascular bioprostheses under influence of stem cell homing.016

Fig. 4. Pulmonary valve recellularisation induced by coating of FN (n¼ 7), SDF-1a (n¼ 6) or both (FNSDF; n¼ 7). (a) X-Y plot of the cell positions (image analysis) of a total valveleaflet. (b) Red and blue (pseudocolor) representation of a Sirius red stained section of the same leaflet. Red represents the matrix that was stained but not detected by polarizedlight microscopy, hence the unorganized collagen. Blue represents the collagen that could be detected with polarized light microscopy and represents organized collagen fibers,largely coinciding with the original implanted matrix material. (c) Endothelial layer (ecNOS immunofluorescence) on the valve leaflet (bar¼ 100 mm). (d) Cells present in the leafletsof each coated valve as compared to the counts of native pulmonary valves. (e) The vimentinþ cells, (f) ASMAþ contractile cells, (g) desminþ cells and (h) smoothelinþ cells asquantified with immunofluorescent staining. (a, b and c indicated different from all other groups, native or SDF-1a group respectively; CR indicates the control range [5]).

G. De Visscher et al. / Biomaterials xxx (2009) 1–9 7

ARTICLE IN PRESS

far superior in terms of recellularization. However, the final goal ofbio-engineering is the development of tissue comparable to nativevalves. Therefore we have related the results to those obtained ina control group of native valves. Based on the impregnation studies,the valves were coated with 800 mg FN and/or 80 mg SDF-1a, rep-resenting 32 and/or 3.2 mg/cm2 respectively. In vitro CD34þ

progenitor attraction seemed optimal at 0.3 mg/ml [30], a lowerconcentration. However, recent in vivo data on SDF-1a inducedangiogenesis in mice infarcted hearts showed improvement in

Please cite this article in press as: De Visscher G, et al., The remodeling osignal pathways, Biomaterials (2009), doi:10.1016/j.biomaterials.2009.09

heart function at a 1 mg topical injection [34]. Despite unreportedabsolute infarct sizes, the percentages and actual mice heart sizessuggest that the reported dose is close to the one used here.

With respect to the recellularisation potential, all coatingprotocols proved successful because in none of them the observedamount of cells, after serving as a well functioning non calcifyingvalve substitute, differed from the native pulmonary valve’s cellcount. The recellularization comprised ingrown cells producingnew matrix (data not shown) and cells covering the implant with

f cardiovascular bioprostheses under influence of stem cell homing.016

G. De Visscher et al. / Biomaterials xxx (2009) 1–98

ARTICLE IN PRESS

new matrix and endothelium. Despite comparable amounts ofmesenchymal cells (fibroblasts in native valves [35]), a clearbeneficial effect with respect to the amount of contractile cells wasobserved in the group were FN was used to present SDF-1a. Incontrast to the single protein coatings, the co-impregnated valveshad a marked decrease in contractile cells, attributed to a decreasein myofibroblasts (ASMA [36]) rather than a decrease in smoothmuscle cells (desmin [37] and smoothelin [38]). This is a markedimprovement on our previous work [5] where intraperitonealseeding resulted in leaflet shortening and valve dysfunction bya combination of increased cell counts and exaggerated amounts ofcontractile cells. Noteworthy, is that for the FN/SDF-1a coating theamount of ASMAþ cells is only slightly above the 2–5% reported fornormal valves [39].

The obtained valves contained some immune cells remnants,which was most probably unavoidable based on our previousobservations with the fixed biological matrix [4]. However, theobtained recellularisation didn’t induce leaflet shortening andregarding leaflet thickening, apparent in intraperitoneally seededvalves [5], only a limited increase was found combined with matrixdeposition. Besides the lack of valve calcification, apparent withother valves in this model [40], the valves clearly maintained theirstrength which was well above the 2.4 N of native pulmonaryvalves reported previously [5].

5. Conclusions

With the FN/SDF-1a coating we have shown that it is possible toprovide a prosthesis that will attract specific primitive cells fromthe blood stream. Moreover these cells will, under influence of thecoating and the milieu they are exposed to, develop into nativevalve cells depositing their own matrix and therefore be able tomaintain the biohybrid valve prostheses.

Source of funding

The project was partially funded by Fonds voor WetenschappelijkOnderzoek (G.0549.06).

Disclosures

Prof. Dr. W. Flameng and Dr. G. De Visscher are inventors ona pending patent application.

Appendix

Figures with essential colour discrimination. Certain figures inthis article, in particular Figs. 1–4, are difficult to interpret in blackand white. The full colour images can be found in the on-lineversion, at doi:10.1016/j.biomaterials.2009.09.016.

Appendix. Supplementary data

The supplementary data associated with this article can befound in the on-line version, at doi:10.1016/j.biomaterials.2009.09.016.

References

[1] Flameng W, Herijgers P, Bogaerts K. Recurrence of mitral valve regurgitationafter mitral valve repair in degenerative valve disease. Circulation2003;107(12):1609–13.

[2] Vranken I, De Visscher G, Lebacq A, Verbeken E, Flameng W. The recruitmentof primitive Lin- Sca-1þ, CD34þ, C-kitþ and CD271þ cells during the earlyintraperitoneal foreign body reaction. Biomaterials 2007;29(7):797–808.

Please cite this article in press as: De Visscher G, et al., The remodelingsignal pathways, Biomaterials (2009), doi:10.1016/j.biomaterials.2009.09

[3] Lebacq A, De Visscher G, Vranken I, Flameng W. Homing, signaling, andstructural molecule gene expression profiling in the early-phase foreign bodyreaction. Int J Low Extrem Wounds 2007;6(3):198.

[4] De Visscher G, Vranken I, Lebacq A, Van Kerrebroeck C, Ganame J, Verbeken E,et al. In vivo cellularisation of a cross-linked matrix by intraperitonealimplantation: a new tool in heart valve tissue engineering. Eur Heart J2007;28(11):1389–96.

[5] De Visscher G, Blockx H, Meuris B, Van Oosterwyck H, Verbeken E,Herregods MC, et al. Functional and biomechanical evaluation of a completelyrecellularized stentless pulmonary bioprosthesis in sheep. J Thorac CardiovascSurg 2008;135(2):395–404.

[6] Pelus LM, Fukuda S. Chemokine-mobilized adult stem cells; defining a betterhematopoietic graft. Leukemia 2008;22(3):466–73.

[7] Chavakis E, Urbich C, Dimmeler S. Homing and engraftment of progenitorcells: a prerequisite for cell therapy. J Mol Cell Cardiol 2008;45(4):514–22.

[8] Claps CM, Corcoran KE, Cho KJ, Rameshwar P. Stromal derived growth factor-1alpha as a beacon for stem cell homing in development and injury. CurrNeurovasc Res 2005;2(4):319–29.

[9] Smith TK, Bader DM. Signals from both sides: control of cardiac developmentby the endocardium and epicardium. Semin Cell Dev Biol 2007;18(1):84–9.

[10] Lu W, Gersting JA, Maheshwari A, Christensen RD, Calhoun DA. Developmentalexpression of chemokine receptor genes in the human fetus. Early Hum Dev2005;81(6):489–96.

[11] Schober A, Karshovska E, Zernecke A, Weber C. SDF-1alpha-mediated tissuerepair by stem cells: a promising tool in cardiovascular medicine? TrendsCardiovasc Med 2006;16(4):103–8.

[12] Har-el R, Tanzer ML. Extracellular matrix. 3: evolution of the extracellularmatrix in invertebrates. FASEB J 1993;7(12):1115–23.

[13] Srivastava M, Begovic E, Chapman J, Putnam NH, Hellsten U, Kawashima T,et al. The Trichoplax genome and the nature of placozoans. Nature2008;454(7207):955–60.

[14] Askari AT, Unzek S, Popovic ZB, Goldman CK, Forudi F, Kiedrowski M, et al.Effect of stromal-cell-derived factor 1 on stem-cell homing and tissueregeneration in ischaemic cardiomyopathy. Lancet 2003;362(9385):697–703.

[15] Massberg S, Konrad I, Schurzinger K, Lorenz M, Schneider S, Zohlnhoefer D,et al. Platelets secrete stromal cell-derived factor 1alpha and recruit bonemarrow-derived progenitor cells to arterial thrombi in vivo. J Exp Med2006;203(5):1221–33.

[16] Fischer H, Polikarpov I, Craievich AF. Average protein density is a molecular-weight-dependent function. Protein Sci 2004;13(10):2825–8.

[17] Kirsch WM, Zhu YH, Hardesty RA, Chapolini R. A new method for microvas-cular anastomosis: report of experimental and clinical research. Am Surg1992;58(12):722–7.

[18] Tashiro K, Tada H, Heilker R, Shirozu M, Nakano T, Honjo T. Signal sequencetrap: a cloning strategy for secreted proteins and type I membrane proteins.Science 1993;261(5121):600–3.

[19] Pelletier AJ, van der Laan LJ, Hildbrand P, Siani MA, Thompson DA, Dawson PE,et al. Presentation of chemokine SDF-1 alpha by fibronectin mediates directedmigration of T cells. Blood 2000;96(8):2682–90.

[20] Baddoo M, Hill K, Wilkinson R, Gaupp D, Hughes C, Kopen GC, et al. Charac-terization of mesenchymal stem cells isolated from murine bone marrow bynegative selection. J Cell Biochem 2003;89(6):1235–49.

[21] Zucchini A, Del Zotto G, Brando B, Canonico B. CD90. J Biol Regul HomeostAgents 2001;15(1):82–5.

[22] Krause DS, Fackler MJ, Civin CI, May WS. CD34: structure, biology, and clinicalutility. Blood 1996;87(1):1–13.

[23] Okada S, Nakauchi H, Nagayoshi K, Nishikawa S, Miura Y, Suda T. In vivo and invitro stem cell function of c-kit- and Sca-1-positive murine hematopoieticcells. Blood 1992;80(12):3044–50.

[24] Guarita-Souza LC, Carvalho KA, Woitowicz V, Rebelatto C, Senegaglia A,Hansen P, et al. Simultaneous autologous transplantation of coculturedmesenchymal stem cells and skeletal myoblasts improves ventricular functionin a murine model of Chagas disease. Circulation 2006;114(Suppl. 1):I120–4.

[25] Suuronen EJ, Zhang P, Kuraitis D, Cao X, Melhuish A, McKee D, et al. Anacellular matrix-bound ligand enhances the mobilization, recruitment andtherapeutic effects of circulating progenitor cells in a hindlimb ischemiamodel. FASEB J 2009;23(5):1447–58.

[26] Wierzbicka-Patynowski I, Schwarzbauer JE. The ins and outs of fibronectinmatrix assembly. J Cell Sci 2003;116(Pt 16):3269–76.

[27] Amara A, Lorthioir O, Valenzuela A, Magerus A, Thelen M, Montes M, et al.Stromal cell-derived factor-1alpha associates with heparan sulfates throughthe first beta-strand of the chemokine. J Biol Chem 1999;274(34):23916–25.

[28] Yang BG, Tanaka T, Jang MH, Bai Z, Hayasaka H, Miyasaka M. Binding oflymphoid chemokines to collagen IV that accumulates in the basal lamina ofhigh endothelial venules: its implications in lymphocyte trafficking. J Immu-nol 2007;179(7):4376–82.

[29] Schoen FJ, Tsao JW, Levy RJ. Calcification of bovine pericardium used in cardiacvalve bioprostheses. Implications for the mechanisms of bioprosthetic tissuemineralization. Am J Pathol 1986;123(1):134–45.

[30] Aiuti A, Tavian M, Cipponi A, Ficara F, Zappone E, Hoxie J, et al. Expression ofCXCR4, the receptor for stromal cell-derived factor-1 on fetal and adult humanlympho-hematopoietic progenitors. Eur J Immunol 1999;29(6):1823–31.

[31] Mahmud N, Devine SM, Weller KP, Parmar S, Sturgeon C, Nelson MC, et al. Therelative quiescence of hematopoietic stem cells in nonhuman primates. Blood2001;97(10):3061–8.

of cardiovascular bioprostheses under influence of stem cell homing.016

G. De Visscher et al. / Biomaterials xxx (2009) 1–9 9

ARTICLE IN PRESS

[32] Wodarz D. Effect of stem cell turnover rates on protection against cancer andaging. J Theor Biol 2007;245(3):449–58.

[33] Durand E, Mallat Z, Addad F, Vilde F, Desnos M, Guerot C, et al. Time courses ofapoptosis and cell proliferation and their relationship to arterial remodelingand restenosis after angioplasty in an atherosclerotic rabbit model. J Am CollCardiol 2002;39(10):1680–5.

[34] Sasaki T, Fukazawa R, Ogawa S, Kanno S, Nitta T, Ochi M, et al. Stromal cell-derived factor-1alpha improves infarcted heart function through angiogenesisin mice. Pediatr Int 2007;49(6):966–71.

[35] Della Rocca F, Sartore S, Guidolin D, Bertiplaglia B, Gerosa G, Casarotto D, et al.Cell composition of the human pulmonary valve: a comparative study withthe aortic valve–the VESALIO Project. Vitalitate Exornatum SuccedaneumAorticum labore Ingegnoso Obtinebitur. Ann Thorac Surg 2000;70(5):1594–600.

Please cite this article in press as: De Visscher G, et al., The remodeling osignal pathways, Biomaterials (2009), doi:10.1016/j.biomaterials.2009.09

[36] Ninomiya K, Takahashi A, Fujioka Y, Ishikawa Y, Yokoyama M. Trans-forming growth factor-beta signaling enhances transdifferentiation ofmacrophages into smooth muscle-like cells. Hypertens Res 2006;29(4):269–76.

[37] Paulin D, Li Z. Desmin: a major intermediate filament protein essential for thestructural integrity and function of muscle. Exp Cell Res 2004;301(1):1–7.

[38] van der Loop FT, Schaart G, Timmer ED, Ramaekers FC, van Eys GJ. Smoothelin,a novel cytoskeletal protein specific for smooth muscle cells. J Cell Biol1996;134(2):401–11.

[39] Mendelson K, Schoen FJ. Heart valve tissue engineering: concepts, approaches,progress, and challenges. Ann Biomed Eng 2006;34(12):1799–819.

[40] Flameng W, Meuris B, Yperman J, De Visscher G, Herijgers P, Verbeken E.Factors influencing calcification of cardiac bioprostheses in adolescent sheep. JThorac Cardiovasc Surg 2006;132(1):89–98.

f cardiovascular bioprostheses under influence of stem cell homing.016