spatial heterogeneity of endothelial phenotypes correlates...

Spatial Heterogeneity of Endothelial Phenotypes CorrelatesWith Side-Specific Vulnerability to Calcification in Normal

Porcine Aortic ValvesCraig A. Simmons, Gregory R. Grant, Elisabetta Manduchi, Peter F. Davies

Abstract—Calcific aortic valve sclerosis involves inflammatory processes and occurs preferentially on the aortic side ofendothelialized valve leaflets. Although the endothelium is recognized to play critical roles in focal vascular sclerosis,the contributions of valvular endothelial phenotypes to aortic valve sclerosis and side-specific susceptibility tocalcification are poorly understood. Using RNA amplification and cDNA microarrays, we identified 584 genes asdifferentially expressed in situ by the endothelium on the aortic side versus ventricular side of normal adult pig aorticvalves. These differential transcriptional profiles, representative of the steady state in vivo, identify globally distinctendothelial phenotypes on opposite sides of the aortic valve. Several over-represented biological classifications withputative relevance to endothelial regulation of valvular homeostasis and aortic-side vulnerability to calcification wereidentified among the differentially expressed genes. Of note, multiple inhibitors of cardiovascular calcification weresignificantly less expressed by endothelium on the disease-prone aortic side of the valve, suggesting side-specificpermissiveness to calcification. However, coexisting putative protective mechanisms were also expressed. Specifically,enhanced antioxidative gene expression and the lack of differential expression of proinflammatory molecules on theaortic side may protect against inflammation and lesion initiation in the normal valve. These data implicate theendothelium in regulating valvular calcification and suggest that spatial heterogeneity of valvular endothelialphenotypes may contribute to the focal susceptibility for lesion development. (Circ Res. 2005;96:792-799.)

Key Words: transcriptional profiling � microarray analysis � calcific aortic sclerosis � hemodynamics

Calcific aortic valve sclerosis, characterized by thickeningand calcification of the valve leaflets, is a common

disease associated with significant morbidity.1 Until recently,calcific aortic sclerosis was considered a passive degenerativeprocess secondary to aging and the accumulation of mechan-ical damage to the valve matrix. However, recent studiesdemonstrate an association between clinical risk factors foratherosclerosis and the development of valvular disease,2

suggesting a more complex etiology than appreciated previ-ously. Early degenerative lesions in human valves are char-acterized by increased cellularity, increased extracellularmatrix deposition, and the accumulation of oxidized lipopro-teins, nonfoam cell and foam cell macrophages, and occa-sional T cells within the valve interstitium.3,4 These histolog-ical findings resemble early sclerotic lesions of thevasculature, and together with the shared risk factors, suggestthat, as in atherosclerosis, the initiation of aortic valvularsclerosis involves chronic inflammatory processes potenti-ated by systemic factors. In advanced calcific valvular le-sions, prominent features include mineralized deposits com-posed of hydroxyapatite, several bone matrix proteins, and

mature osteoblasts and osteoclasts.5–7 Thus, valvular calcifi-cation, rather than being attributable to passive, unregulatedprecipitation of calcium phosphate, appears to be a highlyregulated, active ossification process.

Several theories have been proposed to explain themechanisms by which cardiovascular calcification occurs.8

Observations from mouse mutants9,10 suggest that ectopicmineralization can occur when there is a loss of activeinhibition. In healthy cardiovascular tissue, many inhibi-tors are constitutively expressed, thereby preventing “de-fault” mineralization.11 The identification of subpopula-tions of vascular smooth muscle cells,12,13 adventitialmyofibroblasts,14 and valvular interstitial myofibroblasts15

that can differentiate to osteoblast/chondroblast-like cellssuggests calcification can also occur through induction ofbone formation. Among the most potent stimulants ofmyofibroblast osteogenic differentiation are macrophage-derived inflammatory cytokines, reactive oxygen species(ROS), and lipid oxidation products,15–18 suggesting amechanism by which the inflammatory response in earlylesions contributes indirectly to mineral formation. How-

Original received January 11, 2005; revision received February 11, 2005; accepted February 25, 2005.From the Institute for Medicine and Engineering (C.A.S., P.F.D.), the Departments of Pathology & Laboratory Medicine (P.F.D.) and Bioengineering

(C.A.S., P.F.D.), and the Center for Bioinformatics (G.R.G., E.M.), University of Pennsylvania, Philadelphia.Correspondence to Peter F. Davies, Institute for Medicine and Engineering, University of Pennsylvania, 1010 Vagelos Laboratories, 3340 Smith Walk,

Philadelphia PA 19104. E-mail [email protected]; or Craig A. Simmons, Institute of Biomaterials and Biomedical Engineering, University ofToronto, 5 King’s College Road, Toronto, ON, Canada M5S 3G8. E-mail [email protected]

© 2005 American Heart Association, Inc.

Circulation Research is available at http://www.circresaha.org DOI: 10.1161/01.RES.0000161998.92009.64

792

by guest on May 20, 2018

http://circres.ahajournals.org/D

ownloaded from

by guest on M

ay 20, 2018http://circres.ahajournals.org/

Dow

nloaded from



ownloaded from

by guest on M


Dow

nloaded from



ownloaded from

by guest on M


Dow

nloaded from



ownloaded from

by guest on M


Dow

nloaded from



ownloaded from

by guest on M


Dow

nloaded from



ownloaded from

by guest on M


Dow

nloaded from



ownloaded from

by guest on M


Dow

nloaded from



ownloaded from

by guest on M


Dow

nloaded from



ownloaded from

by guest on M


Dow

nloaded from



ownloaded from

by guest on M


Dow

nloaded from



ownloaded from

by guest on M


Dow

nloaded from



ownloaded from

by guest on M


Dow

nloaded from



ownloaded from

by guest on M


Dow

nloaded from



ownloaded from

http://circres.ahajournals.org/



























ever, particularly in the context of aortic valve sclerosis,these mechanisms are poorly understood.

In the vascular system, the endothelium is an importantregulator of physiology and pathology, including atherogen-esis. Similarly, the endothelium lining the surface of valveleaflets is presumed to be involved in valve homeostasis andpathology, although the contribution of endothelial pheno-types and dysfunction to valve pathologies at the cellular andmolecular level has received little attention. Notably, inhumans3,19 and pigs (supplemental Figure I, available onlineat http://circres.ahajournals.org), lipid deposits and calcificlesions in the aortic valve occur preferentially in the fibrosa,the layer of the valve immediately beneath the endotheliumon the aortic side of the valve. The preferential susceptibilityto lesion formation on the aortic rather than ventricularsurface of the aortic valve may result from coordinatedregulation of gene expression by the respective endothelia,resulting in side-specific endothelial phenotypes that favor orinhibit calcification. Furthermore, during the cardiac cycle,the aortic valve endothelium is subjected to complex fluiddynamics that are distinctly different on either side of thevalve.20 Thus, there is a spatial correlation between the focalnature of calcific lesions and the local hemodynamic envi-ronment, similar to that observed for atherosusceptible re-gions in the large arteries.21 Local environmental factors,including biomechanical forces and molecular transport, maytherefore also contribute to differential endothelial pheno-types that define the sidedness of the valve and the focalsusceptibility to calcification.

To investigate phenotypic heterogeneity in valve endothe-lium and its implications for the focal genesis of valvulardisease, differential endothelial gene expression was analyzedon the aortic side versus ventricular side of normal adult pigaortic valves. The resulting expression profiles, representa-tive of the steady state in vivo, reveal that although the aorticside of normal nondiseased valves is permissive to calcifica-tion, it is generally protected from inflammatory processesassociated with the initiation of sclerosis in the absence ofadditional risk factors.

Materials and MethodsAortic valve leaflets were harvested from eight adult male pigs froma local abattoir (Hatfield Industries, Hatfield, PA) immediately after

death. Endothelial cells were isolated separately from the aortic andventricular surfaces of each valve leaflet using a modified Hautchenmethod.22 Total RNA from the isolated cells (100 ng�10 000cells/side) was linearly amplified by one round of T7-mediated invitro transcription.23 cDNA probes were synthesized from 2 �g ampli-fied antisense RNA using indirect labeling (described in the online datasupplement, available at http://circres.ahajournals.org). Paired fluores-cently labeled cDNA probes from the aortic and ventricular surfaces ofthe same valve were competitively hybridized to an Agilent Human 1cDNA microarray containing 12 922 clones (Agilent Technologies).

Arrays (n�8) were scanned at 10-�m resolution, and images wereanalyzed with Agilent Feature Extraction Software (version A.7.1.1).The intensity ratios on each array were log transformed andnormalized via global lowess with the functions provided in theComprehensive R Archive Network (CRAN) Statistical MicroarrayAnalysis package (sma version 0.5.14; http://cran.r-project.org/;online data supplement). Differential expression analysis was per-formed with PaGE (see http://www.cbil.upenn.edu/PaGE and theonline data supplement). Patterns from Gene Expression (PaGE) is afalse discovery rate (FDR)–based method24 that uses a permutationapproach to estimate the FDR. The normalized log ratios were testedwith a one-sample T statistic and the FDR was 5%. The relativeexpression of selected genes was validated by quantitative real-timepolymerase chain reaction (QRT-PCR) using paired aortic andventricular endothelial samples isolated from 10 additional adultmale pigs. Statistical significance was tested using the one-sample Tstatistic and one-tailed permutation P values. The list of differen-tially expressed genes was analyzed for statistically over-representedbiological themes using Expression Analysis Systematic Explorer(EASE).25 Data were additionally mined with Genespring (SiliconGenetics) for annotation and with literature searches. The completeannotated study is publicly available in a MIAME-compliant frame-

Figure 1. A, Histological section of a normal porcine aortic valvestained with hematoxylin and eosin. B and C, Immunostainingfor vWF demonstrated intact endothelium on the aortic (B) andventricular (C) surfaces of the valves (arrowheads).Bars�200 �m in A and 40 �m in B and C.

TABLE 1. Comparison of Relative Expression Levels Measuredby Microarray Analysis and by QRT-PCR for Selected Genes

A/V Fold Change†

Gene* Microarray QRT-PCRP

Value‡

CX43 �1.72 �1.33 0.0050

FGFR2 2.14 �3.24� 0.00083

GPX3 �1.35 �1.92 0.0020

GSTO1 1.40 1.57 0.18

LGALS1 �2.08 �2.17 0.0017

NOS3 1.44 1.29 0.0033

OSTF1 1.24 1.43 0.0025

PECAM1 1.44 1.26 0.037

PTGS1 �1.54 �2.50 0.00030

SELP 1.66 2.05 0.00083

VCAM1 1.37 �1.02 1.00

VWF 1.60 2.61 0.0025

ICAM1 Not differentially expressed 1.00 1.00

SELE Not differentially expressed Not detected �

* CX43 indicates connexin 43; FGFR2, fibroblast growth factor receptor 2;GPX3, glutathione peroxidase 3; GSTO1, glutathione S-transferase-�1;LGALS1, galectin 1; OSTF, osteoclast-stimulating factor 1; PECAM1, platelet/endothelial cell adhesion molecule; PTGS1, prostaglandin–endoperoxide syn-thase 1; SELP, selectin P; VCAM1, vascular cell adhesion molecule 1; ICAM,intracellular cell adhesion molecule 1; SELE, selectin E.

†A/V fold change indicates aortic-side to ventricular-side fold change;‡permutation-based one sample T statistic, one-tailed in direction predicted bymicroarray; �ratio based on 6 of 10 samples because FGFR2 was detected onthe aortic side but not on the ventricular side of four samples, resulting ininfinite A/V fold changes.

Simmons et al Heterogeneity of Valvular Endothelial Phenotypes 793



ownloaded from


work through the RNA Abundance Database (http://www.cbil.upenn.edu/RAD/php/displayStudy.php?study_id�1270).26 Addi-tionally, all supplemental material can be accessed at http://www.cbil.upenn.edu/RAD/PigValveStudy/.

For histological and immunohistochemical staining, fresh porcineaortic valve leaflets were fixed in 10% neutral-buffered formalin,paraffin-embedded, and serial sectioned. Sections were stained withhematoxylin and eosin or for various antigens using standardperoxidase-based detection methods (online data supplement).

ResultsSample CharacterizationThe gross appearances of the porcine valves used for RNAisolation were normal and consistent with histological exam-ination, which showed no pathology and intact endothelium(Figure 1). Immunostaining of cells isolated by the modifiedHautchen technique confirmed their purity22 (supplementalFigure II). From the freshly isolated endothelial cells, �100ng of intact (28S:18S ribosomal RNA ratio �2) total RNA,representative of the steady state in vivo, was obtained fromeach side of the valve. We have shown previously that thisamount of RNA is sufficient for amplification and identifi-cation of differential expression with fidelity and enhancedsensitivity.27

Side-Dependent Differential Gene ExpressionDifferential gene expression by endothelial cells from theaortic side versus ventricular side of normal aortic valves wasassessed using PaGE. At 5% FDR, 584 genes were identifiedas differentially expressed, with higher expression of 285genes and lower expression of 299 genes on the aortic side ofthe valve relative to the ventricular side. (A fully annotatedlist of the differentially expressed genes is available at

http://www.cbil.upenn.edu/RAD/PigValveStudy/.) The arraypredictions were validated by QRT-PCR for 83% of a subsetof genes relevant to valvular disease (Table 1), consistentwith our previous reports in arterial cells.27,28

Analysis of Biological Themes and PathwaysUsing EASE, the list of differentially expressed genes wasanalyzed globally to identify biological themes that weresignificantly over-represented in the data set (Table 2). Thenonrandom coordinated expression of multiple genes in anover-represented biological theme is suggestive of the func-tional involvement of that pathway or class of genes inside-dependent susceptibility to disease. Included among themost prominent classifications were groups of genes relatedto the cell cycle and apoptosis, nuclear and metabolic activity,intracellular signaling, cytoskeletal organization, and skeletaldevelopment. The gene groupings identified by EASE aredetailed at http://www.cbil.upenn.edu/RAD/PigValveStudy/and, together with the annotated gene list, provide a richpublic database. Although an exhaustive analysis of thecomplete data are beyond the scope of this article, differentialexpression patterns incorporating several prominent classifi-cations with putative significance to valve pathology wereidentified.

Aortic-Side Endothelial Gene Expression IsPermissive to Valvular CalcificationOf particular relevance to valvular pathology was the signif-icant over-representation of differentially expressed genesrelated to skeletal development and vascular calcification(Table 2). Most remarkable was the lower expression on theaortic side of several transcripts for inhibitors of vascular and

TABLE 2. Prominent Biological Classifications Identified by EASE*

BiologicalClassification

No. of GenesDifferentiallyExpressed†

#GenesRepresented

on ArrayP

Value‡

Peripheral nervous system development 4 8 0.00059

Rho protein signal transduction 5 16 0.0015

Eicosanoid biosynthesis 4 10 0.0016

mRNA polyadenylation 3 5 0.0017

Apoptosis 22 202 0.0022

Aspartate family amino acid metabolism 3 6 0.0032

Cell motility 20 201 0.0094

Glutamate channel activity 3 9 0.011

Tumor suppressor 4 17 0.013

Gluconeogenesis 3 10 0.016

Steroid hormone nuclear receptors 5 25 0.022

Actin cytoskeletal organization 6 41 0.026

Transforming growth factor-� signaling 7 47 0.038

Cell adhesion 28 353 0.042

Skeletal development 9 83 0.044

*Shown are the top 10 over-represented classifications containing three or more genes and fiveother highly represented categories potentially relevant to valvular pathology. (The complete list isavailable at www.cbil.upenn.edu/RAD/PigValveStudy.)

†Aortic side vs ventricular side; ‡Fisher exact probability.

794 Circulation Research April 15, 2005



ownloaded from


valvular calcification (Table 3). These included osteoprote-gerin (OPG; tumor necrosis factor receptor superfamily;member 11b), C-type natriuretic peptide (CNP), and parathy-roid hormone (PTH), each of which has been shown to inhibitcardiovascular calcification (Discussion). Also underex-pressed on the aortic side was the transcript for chordin, asecreted protein that inhibits the osteoinductive effects ofbone morphogenetic proteins (BMPs) by sequestering themin latent complexes.29 We immunostained aortic valve sec-tions with an anti-OPG antibody and, consistent with themicroarray predictions, noted striking differential expressionof the OPG protein, with strong staining in and around theventricular-side endothelium but little expression on theaortic side (Figure 2A through 2C).

Coupled with the relative lack of expression of inhibitorson the aortic side was increased expression of severaltranscripts associated with bone formation. Most notable wasthe higher expression of BMP4 by the endothelium on theaortic side of the valve. Also more highly expressed on theaortic side were transcripts for several proteins associatedwith the extracellular matrix of bone or cartilage (Table 3),suggesting that the aortic side endothelium may contribute toa matrix environment that is permissive to calcification.Notably, transcripts for several molecules with demonstratedroles in vascular and valvular calcification (eg, matrix glaprotein, and transforming growth factor-�1) were not identi-fied as differentially expressed in the normal valve (supple-mental Table 2).

Aortic-Side Endothelial Gene Expression IsAntioxidative and Not InflammatoryThe histological similarities between early atheroscleroticvascular lesions and calcific valvular lesions suggest valvularcalcification may initiate through inflammation, leukocyteadhesion and invasion, and oxidative processes. Many ofthese processes are regulated by the endothelium. Notably,relevant transcripts were differentially expressed on oppositesides of normal aortic valves (Tables 2 and 4). However, inthe normal valves, the aortic-side expression profile wasdecidedly noninflammatory. The proinflammatory molecules

CCL13 and Duffy blood group were less expressed on theaortic side of the valve. None of the prototypical inflamma-tory cytokines/chemokines (eg, the interleukins) were differ-entially expressed. Consistent with the lack of inflammation

TABLE 3. Differentially Expressed Genes Related to Skeletal Development and Vascular Calcification

Lower Expression on Aortic Side Higher Expression on Aortic Side

Gene*Accession

No.A/V FoldChange†

PutativeEffect‡ Gene*

AccessionNo.

A/V FoldChange†

Putativeeffect‡

TNFRSF11B U94332 �3.53 � BMP4 NM_001202 1.57 �

NPPC D90337 �3.12 � PTN AU120808 1.53 �

CHRD AF209928 �1.37 � HAPLN1 U43328 1.49 �

PTH V00597 �1.31 � FBN1 X63556 1.39 �

COL11A1 J04177 �1.44 – CHAD AF371328 1.37 �

BMP1 NM_006129 �1.52 ? OSTF1 BC007459 1.24 ?

BMP6 AA426586 �1.29 ?

*CHAD indicates chondroadherin; CHRD, chordin; COL3A1, collagen type III �1; COL11A1, collagen type XI �1; FBN1, fibrillin 1;HAPLN1, hyaluronan and proteoglycan link protein 1; NPPC, natriuretic peptide precursor C (CNP); OSTF1, osteoclast-stimulating factor1; PTN, pleiotrophin; TNFRSF11B, tumor necrosis factor receptor superfamily member 11b (osteoprotegerin).

†Aortic side to ventricular side fold change by microarray analysis; ‡�procalcific on aortic side;–anticalcific on aortic side; ?unknown;refer to supplemental Table 3.

Figure 2. A through C, Immunostaining for OPG revealed strik-ing side-dependent expression (A), with little OPG expression onthe aortic side (B) but strong staining on the ventricular side ofthe valve (C; arrowheads). D through F, 12-lipoxygenase (12-LOX) was also clearly differentially expressed (D), with less en-dothelial expression on the aortic side (E) of the normal valvethan on the ventricular side (F; arrowheads). Strong, heteroge-nous expression of 12-lipoxygenase was also observed in the in-terstitial myofibroblasts. Differential endothelial protein expressionsillustrated in this figure are consistent with the microarray analyses.Bars�200 �m in A and D and 40 �m in B, C, E, and F.




ownloaded from


on the aortic side, the transcripts for galectin-1 and CNP, bothof which are upregulated by proinflammatory mediators,30,31

were underexpressed. Of the adhesion molecules associatedwith early inflammatory responses and leukocyte adhesion,only P-selectin and von Willebrand factor (vWF) weredifferentially expressed. E-selectin, vascular cell adhesionmolecule 1, and intracellular adhesion molecule 1 were notdifferentially expressed, as confirmed by QRT-PCR (Table1). Although P-selectin and vWF mRNA were more highlyexpressed on the aortic side (confirmed by QRT-PCR),differential vWF protein expression was not evident byimmunostaining (Figure 1B and 1C).

The transcription of genes for many proinflammatorycytokines, chemokines, and adhesion molecules is regulatedin vascular endothelial cells by the nuclear factor �B (NF-�B)pathway.32 Among the most potent activators of the NF-�Bpathway are ROS generated by oxidative stress and impli-cated in the initiation and progression of atherosclerosis.33

Data mining revealed a strong antioxidative profile in theaortic-side endothelium that may protect against ROS- andNF-�B–mediated inflammation (Table 4). The transcripts forseveral intracellular antioxidative enzymes, including micro-somal glutathione S-transferase 2 and peroxiredoxin 2, weremore highly expressed on the aortic side of the valve.Glutathione S-transferase-�1 was also more highly expressedon the aortic side by microarray analysis and by QRT-PCRbut was not statistically significant in the QRT-PCR valida-tion because of large intra-animal variability (Table 1).Glutathione peroxidase 3 was underexpressed on the aorticside of the valve, but this antioxidative enzyme acts extracel-lularly, and therefore its role in maintaining the intracellularendothelial redox state is unclear. Further contributing to theantioxidant profile on the aortic side was higher expression ofendothelial NO synthase (eNOS; associated with severalbeneficial effects including resistance to oxidative stress34),and lower expression of the oxidized low-density lipoproteinreceptor 1, which binds oxidized lipoproteins to promoteROS-induced activation of NF-�B35 and the endothelialexpression of calcification-related genes.36 Additionally, ara-

chidonate 12-lipoxygenase expression was significantlylower on the aortic side of the valve. Lipoxygenase enzymesoxidize polyunsaturated fatty acids to synthesize hydroper-oxyacids, which are potent pro-oxidant mediators. In endo-thelial cells, 12/15 lipoxygenase and its products are impli-cated in mediating atherosclerosis.37 Consistent with theaortic-side antioxidative transcriptional profile, immuno-staining clearly revealed lower expression of the 12-lipox-ygenase protein by endothelial cells on the aortic side ofthe valve relative to the ventricular side (Figure 2Dthrough 2F). Thus, the balance of oxidative transcripts wasdecisively shifted toward an antioxidative state on theaortic side of the valve, and this protective profile may beresponsible for suppressing the expression of proinflam-matory genes in the normal valve.

DiscussionAlthough the endothelium is recognized to play critical rolesin the initiation and progression of sclerotic diseases of thevasculature, the contribution of valvular endothelial pheno-types to calcific aortic sclerosis and its focal genesis arepoorly understood. Using RNA amplification and transcrip-tional profiling of aortic-side versus ventricular-side endothe-lium from normal porcine valves, we identified spatiallydistinct endothelial phenotypes on opposite sides of the aorticvalve in vivo. The differential gene expression profilessuggest that the endothelium on the disease-prone aortic sideof the valve is permissive to calcification but is protected inthe normal valve against inflammation and lesion initiationby antioxidative mechanisms. These data demonstrate that theendothelium may play critical roles in regulating valvularcalcification through inhibitory and stimulatory mechanisms,and the spatial heterogeneity of valvular endothelial pheno-types may contribute to the focal susceptibility for lesiondevelopment.

A striking observation from the gene expression data wasthe relative absence on the aortic side of the valve of OPG,CNP, PTH, and chordin. OPG plays a role in the local andsystemic regulation of bone resorption38 but also appears to

TABLE 4. Differentially Expressed Genes Related to Inflammation and Oxidation

Lower Expression on Aortic Side Higher Expression on Aortic Side

Gene*Accession

No.A/V FoldChange†

PutativeEffect‡ Gene*

AccessionNo.

A/V FoldChange†

PutativeEffect‡

ALOX12 M62982 �4.53 – MGST2 U77604 1.82 –

NPPC D90337 �3.12 – NOS3 BG741096 1.44 –

LGALS1 BC001693 �2.22 – PRDX2 BC000452 1.28 –

FY AF030521 �1.97 – SELP NM_003005 1.66 �

CCL13 U59808 �1.53 – vWF X04385 1.60 �

OLR1 AB017444 �1.32 –

GPX3 D00632 �1.36 �

*ALOX12 indicates arachidonate 12-lipoxygenase; CCL13, chemokine (C-C motif) ligand 13; FY, Duffy blood group; GPX3,glutathione peroxidase 3; LGALS1, galectin 1; MGST2, microsomal glutathione S-transferase 2; NPPC, natriuretic peptide precursorC; OLR1, oxidized low-density lipoprotein receptor 1; PRDX2, peroxiredoxin 2; SELP, selectin P.

†Aortic side to ventricular side fold change by microarray analysis; ‡�proinflammatory/oxidative on the aortic side; –anti-inflammatory/oxidative on the aortic side; refer to supplemental Table 3.




ownloaded from


suppress local cardiovascular calcification because mice lack-ing OPG have calcified arteries,9 and aortic valve myofibro-blasts treated with receptor activator of nuclear factor kappaB ligand (RANKL) in the absence of OPG differentiate to anosteogenic phenotype in vitro.39 Similarly, CNP- and PTH-related polypeptide inhibit vascular myofibroblast calcifica-tion in vitro.40,41 In vivo, PTH administered intermittently toLDLR�/� mice promotes skeletal bone formation whilesimultaneously suppressing vascular osteogenesis and calci-fication, in part through direct action of PTH on the osteo-genic differentiation of vascular myofibroblasts.42 Chordin isan antagonist of BMP2 and BMP4 and therefore may regulateosteogenic differentiation indirectly by mitigating the os-teoinductive effects of endogenous BMPs. Local release ofeach of these inhibitors from the ventricular-side endotheliummay act in a paracrine manner to inhibit ectopic mineraliza-tion or myofibroblast osteogenic differentiation. Conversely,the absence of these inhibitors on the aortic side may, in thepresence of a procalcific challenge, permit the phenotypictransition of myofibroblasts to osteoblast-like cells and ec-topic mineralization. Previous studies have identified OPG inthe interstitium of normal but not sclerotic aortic valves.39

However, the striking side-dependent expression we observedat the mRNA and protein levels in situ argues strongly for apreviously unrecognized role for the endothelium in regulat-ing local calcification processes through paracrine signalingmechanisms. Additionally, OPG is an autocrine antiapoptoticfactor for endothelial cells,43 and therefore, a deficiency ofOPG on the aortic side may contribute to endothelial injury asan initiating event in lesion formation.

Also notable was the higher aortic-side expression ofBMP4, which, together with the closely related proteinBMP2, has been observed in advanced atherosclerotic le-sions,13,44 calcific valvular lesions,45 and in vascular endothe-lial cells in human coronary arteries, where BMP4 appears tomediate inflammatory responses.46 Endothelial-derivedBMP2 has been shown to regulate orthotopic bone forma-tion47 and to regulate vascular myofibroblast osteogenesis invitro,48 suggesting endothelial-derived BMPs may have thepotential to stimulate valvular calcification. Although BMP4mRNA was more highly expressed on the aortic side, we didnot observe histological evidence of inflammation or calcifi-cation, suggesting compensatory mechanisms limit the effectsof BMP4 gene expression in the normal valve. Similarly,Sorescu et al46 only observed BMP4 protein expression inendothelial cells when they were overlying foam cells ininflamed vascular regions.

Side-dependent endothelial phenotypes are likely to be acombination of intrinsic phenotype determined during devel-opment and maintained postnatally, and spatially sensitivephenotype determined by local environmental factors, such ashemodynamic characteristics. Chi et al49 demonstrated intrin-sic regional differences in endothelial phenotypes by profil-ing endothelial cells isolated from various vascular beds andgrown in culture for several passages. The different endothe-lia had distinct and characteristic gene expression profiles,indicating that they displayed intrinsic phenotypic differenceseven in the absence of tissue-specific microenvironmentalstimuli. A similar comparison of cultured vascular and

valvular endothelial cells also demonstrated distinct tran-scriptional profiles.50 At a higher level of spatial resolution,we have shown in the current and previous studies28 site-specific heterogeneity of endothelial phenotypes within asingle tissue in vivo. In the case of the valvular endothelium,the side-specific phenotypic differences appear to be in partintrinsic because we have observed morphological and func-tional differences between endothelial cells derived fromopposite sides of aortic valves and grown in culture underidentical conditions for multiple population doublings(Simmons et al, unpublished data, 2004). Phenotypic differ-ences in adult aortic and ventricular-side valvular endothelialcells may reflect functional differences determined early indevelopment.

However, it is probable that as in the vasculature, localenvironmental cues regulate the biology of the valvularendothelium. As with atherogenesis, the correlation betweenthe differential blood flow characteristics and the focal natureof valvular calcification suggests that hemodynamic forcesmay play a critical regulatory role. Although the relationshipbetween flow and vascular endothelial cell transcriptionalprofiles has been well studied in vitro51 and recently in vivo,28

little is known about the response of valvular endothelial cellsto hemodynamic forces. It was reported recently that aorticvalve endothelial cells are sensitive to laminar shear stress invitro and, at least in their flow-mediated alignment, responddistinctly from vascular endothelium.52 We observed differ-ential expression by the valvular endothelium of severalgenes that are mechanoregulated in vascular endothelial cells(eg, BMP4, CNP, eNOS, and SMAD6, etc). Although thedirection of regulation of some of these genes was similar inlesion-prone regions of the valve and the vasculature,28 thiswas not always the case. Contributing to the distinct re-sponses may be the unique and complex biomechanicalstimuli experienced by the valvular endothelium in vivo, thephenotypic differences between valvular and vascular endo-thelial cells, or other nonbiomechanical factors. Therefore,although there appears to be a correlative link betweenhemodynamics, valvular endothelial phenotypes, and suscep-tibility to calcification, further investigations are required tounderstand these links and to demonstrate causality.

AcknowledgmentsThis work was supported by an American Heart Association post-doctoral fellowship 0325666U (C.A.S.) and National Institutes ofHealth grants K25-HG-02296 (G.G.), K25-HG-00052 (E.M.), andHL36049 MERIT, HL62250, and HL64388 (P.F.D.). We thank DrsAnthony Passerini, Emile Mohler III, Robert Levy, and ChristianStoeckert of the University of Pennsylvania for thoughtful discus-sions and critical reading of this manuscript, and Dr Peter White(Center for Molecular Studies in Digestive and Liver Disease) foruse of the microarray scanner. We gratefully acknowledge the expertassistance of Rebecca Riley, Nadeene Francesco, and Drs ArmenKaramanian, Congzhu Shi, and Jenny Zilberberg.

References1. Otto CM, Lind BK, Kitzman DW, Gersh BJ, Siscovick DS. Association

of aortic-valve sclerosis with cardiovascular mortality and morbidity inthe elderly. N Engl J Med. 1999;341:142–147.

2. Mohler ER III. Are atherosclerotic processes involved in aortic-valvecalcification? Lancet. 2000;356:524–525.




ownloaded from


3. O’Brien KD, Reichenbach DD, Marcovina SM, Kuusisto J, Alpers CE,Otto CM. Apolipoproteins B, (a), and E accumulate in the morpholog-ically early lesion of ‘degenerative’ valvular aortic stenosis. ArteriosclerThromb Vasc Biol. 1996;16:523–532.

4. Olsson M, Thyberg J, Nilsson J. Presence of oxidized low densitylipoprotein in nonrheumatic stenotic aortic valves. Arterioscler ThrombVasc Biol. 1999;19:1218–1222.

5. Rajamannan NM, Subramaniam M, Rickard D, Stock SR, Donovan J,Springett M, Orszulak T, Fullerton DA, Tajik AJ, Bonow RO, SpelsbergT. Human aortic valve calcification is associated with an osteoblastphenotype. Circulation. 2003;107:2181–2184.

6. O’Brien KD, Kuusisto J, Reichenbach DD, Ferguson M, Giachelli C,Alpers CE, Otto CM. Osteopontin is expressed in human aortic valvularlesions. Circulation. 1995;92:2163–2168.

7. Mohler ER III, Gannon F, Reynolds C, Zimmerman R, Keane MG,Kaplan FS. Bone formation and inflammation in cardiac valves. Circu-lation. 2001;103:1522–1528.

8. Speer MY, Giachelli CM. Regulation of cardiovascular calcification.Cardiovasc Pathol. 2004;13:63–70.

9. Bucay N, Sarosi I, Dunstan CR, Morony S, Tarpley J, Capparelli C,Scully S, Tan HL, Xu W, Lacey DL, Boyle WJ, Simonet WS.Osteoprotegerin-deficient mice develop early onset osteoporosis and ar-terial calcification. Genes Dev. 1998;12:1260–1268.

10. Luo G, Ducy P, McKee MD, Pinero GJ, Loyer E, Behringer RR, KarsentyG. Spontaneous calcification of arteries and cartilage in mice lackingmatrix GLA protein. Nature. 1997;386:78–81.

11. Schinke T, McKee MD, Karsenty G. Extracellular matrix calcification:where is the action? Nat Genet. 1999;21:150–151.

12. Wada T, McKee MD, Steitz S, Giachelli CM. Calcification of vascularsmooth muscle cell cultures: inhibition by osteopontin. Circ Res. 1999;84:166–178.

13. Bostrom K, Watson KE, Horn S, Wortham C, Herman IM, Demer LL.Bone morphogenetic protein expression in human atherosclerotic lesions.J Clin Invest. 1993;91:1800–1809.

14. Vattikuti R, Towler DA. Osteogenic regulation of vascular calcification:an early perspective. Am J Physiol Endocrinol Metab. 2004;286:E686–E696.

15. Mohler ER III, Chawla MK, Chang AW, Vyavahare N, Levy RJ, GrahamL, Gannon FH. Identification and characterization of calcifying valvecells from human and canine aortic valves. J Heart Valve Dis. 1999;8:254–260.

16. Tintut Y, Patel J, Territo M, Saini T, Parhami F, Demer LL. Monocyte/macrophage regulation of vascular calcification in vitro. Circulation.2002;105:650–655.

17. Parhami F, Morrow AD, Balucan J, Leitinger N, Watson AD, Tintut Y,Berliner JA, Demer LL. Lipid oxidation products have opposite effects oncalcifying vascular cell and bone cell differentiation. A possible expla-nation for the paradox of arterial calcification in osteoporotic patients.Arterioscler Thromb Vasc Biol. 1997;17:680–687.

18. Mody N, Parhami F, Sarafian TA, Demer LL. Oxidative stress modulatesosteoblastic differentiation of vascular and bone cells. Free Radic BiolMed. 2001;31:509–519.

19. Otto CM, Kuusisto J, Reichenbach DD, Gown AM, O’Brien KD. Char-acterization of the early lesion of ‘degenerative’ valvular aortic stenosis.Histological and immunohistochemical studies. Circulation. 1994;90:844–853.

20. Nicosia MA, Cochran RP, Einstein DR, Rutland CJ, Kunzelman KS. Acoupled fluid-structure finite element model of the aortic valve and root.J Heart Valve Dis. 2003;12:781–789.

21. Caro CG, Fitz-Gerald JM, Schroter RC. Arterial wall shear and distri-bution of early atheroma in man. Nature. 1969;223:1159–1160.

22. Simmons CA, Zilberberg J, Davies PF. A rapid, reliable method to isolatehigh quality endothelial RNA from small spatially defined locations. AnnBiomed Eng. 2004;32:1453–1459.

23. Van Gelder RN, von Zastrow ME, Yool A, Dement WC, Barchas JD,Eberwine JH. Amplified RNA synthesized from limited quantities ofheterogeneous cDNA. Proc Natl Acad Sci U S A. 1990;87:1663–1667.

24. Benjamini Y, Hochberg Y. Controlling the false discovery rate—apractical and powerful approach to multiple testing. J R Stat Soc B.1995;57:289–300.

25. Hosack DA, Dennis G Jr, Sherman BT, Lane HC, Lempicki RA. Iden-tifying biological themes within lists of genes with EASE. Genome Biol.2003;4:R70.

26. Manduchi E, Grant GR, He H, Liu J, Mailman MD, Pizarro AD, WhetzelPL, Stoeckert CJ Jr. RAD and the RAD Study-Annotator: an approach to

collection, organization and exchange of all relevant information forhigh-throughput gene expression studies. Bioinformatics. 2004;20:452–459.

27. Polacek DC, Passerini AG, Shi C, Francesco NM, Manduchi E, GrantGR, Powell S, Bischof H, Winkler H, Stoeckert CJ Jr, Davies PF. Fidelityand enhanced sensitivity of differential transcription profiles followinglinear amplification of nanogram amounts of endothelial mRNA. PhysiolGenomics. 2003;13:147–156.

28. Passerini AG, Polacek DC, Shi C, Francesco NM, Manduchi E, GrantGR, Pritchard WF, Powell S, Chang GY, Stoeckert CJ Jr, Davies PF.Coexisting proinflammatory and antioxidative endothelial transcriptionprofiles in a disturbed flow region of the adult porcine aorta. Proc NatlAcad Sci U S A. 2004;101:2482–2487.

29. Piccolo S, Sasai Y, Lu B, De Robertis EM. Dorsoventral patterning inXenopus: inhibition of ventral signals by direct binding of chordin toBMP-4. Cell. 1996;86:589–598.

30. La M, Cao TV, Cerchiaro G, Chilton K, Hirabayashi J, Kasai K, OlianiSM, Chernajovsky Y, Perretti M. A novel biological activity for galec-tin-1: inhibition of leukocyte-endothelial cell interactions in experimentalinflammation. Am J Pathol. 2003;163:1505–1515.

31. Suga S, Itoh H, Komatsu Y, Ogawa Y, Hama N, Yoshimasa T, Nakao K.Cytokine-induced C-type natriuretic peptide (CNP) secretion fromvascular endothelial cells– evidence for CNP as a novel autocrine/paracrine regulator from endothelial cells. Endocrinology. 1993;133:3038–3041.

32. Collins T, Cybulsky MI. NF-�B: pivotal mediator or innocent bystanderin atherogenesis? J Clin Invest. 2001;107:255–264.

33. Harrison D, Griendling KK, Landmesser U, Hornig B, Drexler H. Role ofoxidative stress in atherosclerosis. Am J Cardiol. 2003;91:7A–11A.

34. Wung BS, Cheng JJ, Shyue SK, Wang DL. NO modulates monocytechemotactic protein-1 expression in endothelial cells under cyclic strain.Arterioscler Thromb Vasc Biol. 2001;21:1941–1947.

35. Matsunaga T, Hokari S, Koyama I, Harada T, Komoda T. NF-�B acti-vation in endothelial cells treated with oxidized high-density lipoprotein.Biochem Biophys Res Commun. 2003;303:313–319.

36. Cola C, Almeida M, Li D, Romeo F, Mehta JL. Regulatory role ofendothelium in the expression of genes affecting arterial calcification.Biochem Biophys Res Commun. 2004;320:424–427.

37. Reilly KB, Srinivasan S, Hatley ME, Patricia MK, Lannigan J, BolickDT, Vandenhoff G, Pei H, Natarajan R, Nadler JL, Hedrick CC. 12/15-Lipoxygenase activity mediates inflammatory monocyte/endothelialinteractions and atherosclerosis in vivo. J Biol Chem. 2004;279:9440–9450.

38. Simonet WS, Lacey DL, Dunstan CR, Kelley M, Chang MS, Luthy R,Nguyen HQ, Wooden S, Bennett L, Boone T, Shimamoto G, DeRose M,Elliott R, Colombero A, Tan HL, Trail G, Sullivan J, Davy E, Bucay N,Renshaw-Gegg L, Hughes TM, Hill D, Pattison W, Campbell P, BoyleWJ, et al. Osteoprotegerin: a novel secreted protein involved in theregulation of bone density. Cell. 1997;89:309–319.

39. Kaden JJ, Bickelhaupt S, Grobholz R, Haase KK, Sarikoc A, Kilic R,Brueckmann M, Lang S, Zahn I, Vahl C, Hagl S, Dempfle CE,Borggrefe M. Receptor activator of nuclear factor �B ligand andosteoprotegerin regulate aortic valve calcification. J Mol Cell Cardiol.2004;36:57– 66.

40. Jono S, Nishizawa Y, Shioi A, Morii H. 1,25-Dihydroxyvitamin D3increases in vitro vascular calcification by modulating secretion of en-dogenous parathyroid hormone-related peptide. Circulation. 1998;98:1302–1306.

41. Huang Z, Li J, Jiang Z, Qi Y, Tang C, Du J. Effects of adrenomedullin,C-type natriuretic peptide, and parathyroid hormone-related peptide oncalcification in cultured rat vascular smooth muscle cells. J CardiovascPharmacol. 2003;42:89–97.

42. Shao JS, Cheng SL, Charlton-Kachigian N, Loewy AP, Towler DA.Teriparatide (human parathyroid hormone (1–34)) inhibits osteogenicvascular calcification in diabetic low density lipoprotein receptor-deficient mice. J Biol Chem. 2003;278:50195–50202.

43. Malyankar UM, Scatena M, Suchland KL, Yun TJ, Clark EA, GiachelliCM. Osteoprotegerin is an alpha vbeta 3-induced, NF-�B-dependentsurvival factor for endothelial cells. J Biol Chem. 2000;275:20959–20962.

44. Dhore CR, Cleutjens JP, Lutgens E, Cleutjens KB, Geusens PP, KitslaarPJ, Tordoir JH, Spronk HM, Vermeer C, Daemen MJ. Differentialexpression of bone matrix regulatory proteins in human atheroscleroticplaques. Arterioscler Thromb Vasc Biol. 2001;21:1998–2003.




ownloaded from


45. Kaden JJ, Bickelhaupt S, Grobholz R, Vahl CF, Hagl S, Brueckmann M,Haase KK, Dempfle CE, Borggrefe M. Expression of bone sialoproteinand bone morphogenetic protein-2 in calcific aortic stenosis. J HeartValve Dis. 2004;13:560–566.

46. Sorescu GP, Sykes M, Weiss D, Platt MO, Saha A, Hwang J, Boyd N,Boo YC, Vega JD, Taylor WR, Jo H. Bone morphogenic protein 4produced in endothelial cells by oscillatory shear stress stimulates aninflammatory response. J Biol Chem. 2003;278:31128–31135.

47. Bouletreau PJ, Warren SM, Spector JA, Peled ZM, Gerrets RP,Greenwald JA, Longaker MT. Hypoxia and VEGF up-regulate BMP-2mRNA and protein expression in microvascular endothelial cells:implications for fracture healing. Plast Reconstr Surg. 2002;109:2384 –2397.

48. Shin V, Zebboudj AF, Bostrom K. Endothelial cells modulate osteo-genesis in calcifying vascular cells. J Vasc Res. 2004;41:193–201.

49. Chi JT, Chang HY, Haraldsen G, Jahnsen FL, Troyanskaya OG, ChangDS, Wang Z, Rockson SG, van de Rijn M, Botstein D, Brown PO.Endothelial cell diversity revealed by global expression profiling. ProcNatl Acad Sci U S A. 2003;100:10623–10628.

50. Farivar RS, Cohn LH, Soltesz EG, Mihaljevic T, Rawn JD, Byrne JG.Transcriptional profiling and growth kinetics of endothelium revealsdifferences between cells derived from porcine aorta versus aortic valve.Eur J Cardiothorac Surg. 2003;24:527–534.

51. Garcia-Cardena G, Comander J, Anderson KR, Blackman BR, GimbroneMA Jr. Biomechanical activation of vascular endothelium as a deter-minant of its functional phenotype. Proc Natl Acad Sci U S A. 2001;98:4478–4485.

52. Butcher JT, Penrod AM, Garcia AJ, Nerem RM. Unique morphology andfocal adhesion development of valvular endothelial cells in static andfluid flow environments. Arterioscler Thromb Vasc Biol. 2004;24:1429–1434.




ownloaded from


Craig A. Simmons, Gregory R. Grant, Elisabetta Manduchi and Peter F. DaviesVulnerability to Calcification in Normal Porcine Aortic Valves

Spatial Heterogeneity of Endothelial Phenotypes Correlates With Side-Specific

Print ISSN: 0009-7330. Online ISSN: 1524-4571 Copyright © 2005 American Heart Association, Inc. All rights reserved.is published by the American Heart Association, 7272 Greenville Avenue, Dallas, TX 75231Circulation Research

doi: 10.1161/01.RES.0000161998.92009.642005;96:792-799; originally published online March 10, 2005;Circ Res.

http://circres.ahajournals.org/content/96/7/792World Wide Web at:

The online version of this article, along with updated information and services, is located on the

http://circres.ahajournals.org/content/suppl/2005/06/06/01.RES.0000161998.92009.64.DC1 http://circres.ahajournals.org/content/suppl/2005/03/20/01.RES.0000161998.92009.64v1.DC1

Data Supplement (unedited) at:

http://circres.ahajournals.org//subscriptions/

is online at: Circulation Research Information about subscribing to Subscriptions:

http://www.lww.com/reprints Information about reprints can be found online at: Reprints:

document. Permissions and Rights Question and Answer about this process is available in the

located, click Request Permissions in the middle column of the Web page under Services. Further informationEditorial Office. Once the online version of the published article for which permission is being requested is

can be obtained via RightsLink, a service of the Copyright Clearance Center, not theCirculation Researchin Requests for permissions to reproduce figures, tables, or portions of articles originally publishedPermissions:



ownloaded from

http://circres.ahajournals.org/content/96/7/792

http://circres.ahajournals.org/content/suppl/2005/03/20/01.RES.0000161998.92009.64v1.DC1

http://circres.ahajournals.org/content/suppl/2005/06/06/01.RES.0000161998.92009.64.DC1

http://www.ahajournals.org/site/rights/

http://www.lww.com/reprints

http://circres.ahajournals.org//subscriptions/


of 10

Online Supplemental Methods and Tables for Simmons et al. “Spatial heterogeneity of endothelial phenotypes correlates with side-specific vulnerability to calcification in normal porcine aortic valves" Supplemental Methods Supplemental material is available directly at http://www.cbil.upenn.edu/RAD/PigValveStudy/ Valve endothelial cell collection. Aortic valve leaflets were harvested from eight adult male pigs (Landrace X Yorkshire, ~250 pounds, castrated; Hatfield Industries, Hatfield, PA) and rinsed thoroughly with cold diethyl pyrocarbonate (DEPC)-treated phosphate buffered saline (PBS). Immediately after harvest, endothelial cells were isolated simultaneously from the aortic and ventricular surfaces of each valve leaflet by freezing the endothelium to glass coverslips, as described previously 1. This modified Hautchen method has been shown to yield high quality side-specific pure endothelial RNA (1 and Online Figure 2). The cells were rapidly transferred to a guanidine thiocyanate lysis buffer (Absolutely RNA Nanoprep Kit, Stratagene, La Jolla, CA) and vortexed. Lysates from the aortic and ventricular surfaces were collected separately and the same aliquot of lysis buffer was used for all three leaflets from the same valve. Lysates were stored on dry ice for transportation back to the lab. RNA extraction, characterization, and amplification. Total RNA was extracted using the Absolutely RNA Nanoprep Kit (Stratagene) according to the manufacturer’s protocol. Purified total RNA was eluted in ~18 µL nuclease-free water and concentrated to 10 µL by vacuum centrifugation. Total RNA quantity and integrity were evaluated using an Agilent Bioanalyzer 2100 and RNA 6000 Nano Labchips (Agilent Technologies, Palo Alto, CA). The RNA was frozen at -80°C until amplification. Valve endothelial mRNA from 100 ng total RNA (~ 10,000 cells) was linearly amplified using the MessageAmp aRNA Kit (Ambion, Austin, TX) following the manufacturer’s protocol. This T7-mediated in vitro transcription 2 typically produced >2 μg amplified antisense RNA (aRNA) from one round of amplification (equivalent to >2,000 fold amplification). aRNA quantity was measured using a NanoDrop ND-1000 spectrophotometer (NanoDrop Technologies, Inc., Rockland, DE) and integrity was assessed using the Agilent Bioanalyzer. cDNA probe synthesis and indirect labeling. cDNA probes were synthesized from aRNA using an indirect labeling protocol. Two micrograms aRNA were combined with 2 μg random hexamer primer (Amersham Biosciences, Piscataway, NJ) and 20 U RNasin Plus (Promega, Madison, WI) in an 11.5 μL reaction volume, denatured for 10 min at 70°C, and then cooled to 4°C for 5 min. The RNA/primer solution was combined with 8.5 μL RT mixture consisting of RT buffer (20 mM Tris-HCl, pH=8.4; 50 mM KCl; 5 mM MgCl2), 300 U Superscript™ III RT, 10 mM DTT (all Invitrogen), 0.5 mM dATP, 0.5 mM dGTP, 0.5 mM dCTP, 0.25 mM dTTP (Amersham Biosciences), and 0.25 mM amino-allyl dUTP (Ambion) and then incubated at 50°C for 2 h. The reaction was terminated and the RNA was hydrolyzed with 4 μL 1 M NaOH and incubation for 15 min at 65°C. The solution was neutralized with 10 μL 1 M HEPES, pH 7.0. The cDNA was purified with MinElute PCR Purification kits (Qiagen Inc., Valencia, CA), with

of 10

3 x 10 μL elutions (the first in EB buffer, the other two in water) and then concentrated to 5 μL by vacuum centrifugation. Mono-functional NHS ester Cy3 or Cy5 dye (40 nmol; RPN5661, Amersham Biosciences) was reconstituted in 5 μL 0.1 M sodium bicarbonate and incubated with the aminoallyl-cDNA at room temperature for 1 h in the dark. Paired cDNA samples (n = 8) from the aortic and ventricular surfaces of the same valve were labeled with Cy3 and Cy5 in a randomized manner. The labeling reaction was quenched by incubating with 6 μL 4 M hydroxylamine hydrochloride at room temperature for 15 min in the dark. After addition of 14 μL 50 mM sodium acetate (pH 5.2) to each reaction volume, paired labeled cDNA samples from the aortic and ventricular sides of the same valve were combined and purified with MinElute PCR Purification kits (Qiagen). The probes were concentrated to 7 μL by vacuum centrifugation, and stored on ice in the dark prior to microarray hybridization. Microarray hybridization. A hybridization mixture (25 μL) was prepared by combining the labeled cDNA with 3 μg human Cot-1 DNA (Invitrogen), 2.5 μL deposition targets (Agilent), and hybridization buffer (Agilent). The hybridization solution was incubated for 2 min at 98°C to denature the cDNA probe and then cooled to room temperature. The hybridization mixture was applied to an Agilent Human 1 cDNA microarray. In previous studies, we obtained similar validation rates by QRT-PCR when using human aRNA with human cDNA arrays or pig aRNA with human cDNA arrays, suggesting pig-to-human cross-hybridization is efficient 3, 4. After incubation for 17 h at 65°C, arrays were submerged briefly in 0.5X SSC, 0.01% SDS to remove the coverslip, and then washed once in 0.5X SSC, 0.01% SDS for 6 min with agitation and once in 0.06X SSC for 3 min with agitation. Arrays were dried by centrifugation at room temperature for 2 min at 400g. Microarray scanning and image analysis. Arrays were scanned with an Agilent DNA Microarray Scanner at 10 μm resolution with 100% laser power and variable photomultiplier tube (PMT) voltage settings to obtain maximal signal intensities without saturation. Images were analyzed with Agilent Feature Extraction Software (version A.7.1.1) with raw fluorescence intensity values determined using the “CookieCutter” method of spot analysis. Data preprocessing and analysis. Control features, array and print defect features, and features with non-uniform intensities (e.g., due to scratches) were discarded from the analyses. The (few) features flagged as “saturated” and the features flagged as “not found” were retained in the analyses, but the flags were noted. Preprocessing was performed using R (CRAN: http://cran.r-project.org/) version 1.8.1. For each channel, the mean signal measure from the Agilent software was used as input signal intensity (no background subtraction was performed). The intensity ratios on each array were log transformed and normalized via global lowess with the functions provided in the Statistical Microarray Analysis (sma) package (version 0.5.14) available at the CRAN website. Differential expression analysis was performed with PaGE (http://www.cbil.upenn.edu/PaGE). PaGE is a False Discovery Rate (FDR) based method that uses a permutation approach to estimate the FDR. For any specified constant, permutations of the data matrix are used to estimate the rate of false positives in any set of genes having a T-statistic greater than the

of 10

constant. An appropriate constant is chosen to guarantee the desired FDR, in our case 0.05. Confidences are then transferred to all genes in the set. In this way the gene-by-gene confidences reported can be considered as corrected for multiple testing. For example, if 10 genes with confidence 0.8 are examined, the expected number of false positives is approximately equal to two. PaGE also produces "levels" of differential expression, based on the confidence parameter. The array data analyzed consisted of normalized log ratios and the one-sample T-statistic option was used. If more than 4 (of 8) measurements were missing for any gene then that gene was discarded. The permutations for each row of data consist of taking the reciprocals of a subset of the ratios. This is done in all possible (28 = 256) ways. For detailed information on the PaGE algorithm, see http://www.cbil.upenn.edu/PaGE/doc/PaGE_documentation_technical_manual.pdf. The list of differentially expressed genes was analyzed for over-represented biological themes using Expression Analysis Systematic Explorer (EASE) 5. Genes were classified by Gene Ontology (GO), Kyoto Encyclopedia of Genes and Genomes (KEGG) Pathway, and Genmapp Pathway annotations, and statistically over-represented categories were determined using the Fisher Exact test. The data were additionally annotated using Genespring (Silicon Genetics, Redwood City, CA) based on the KEGG, Genmapp, and GO classifications, and mined with literature searches to identify genes and gene products potentially relevant to aspects of valvular calcification. In accordance with proposed standards of the Microarray Gene Expression Data Society (http://www.mged.org), the complete annotated study is publicly available in a MIAME compliant framework through the RNA Abundance Database (http://www.cbil.upenn.edu/RAD/php/displayStudy.php?study_id=1270) 6. Gene expression validation studies. Relative expression of selected genes was validated by quantitative real-time PCR (QRT-PCR) using side-specific endothelial RNA isolated from ten additional adult male pigs. Total RNA (40-100 ng) was reverse transcribed to cDNA using Superscript II™ Reverse Transcription reagents (Invitrogen) in a 20 μL reaction volume. This cDNA was used in a 10 μL QRT-PCR reaction using the FastStart DNA Master SYBR Green I Kit and the LightCycler® System (Roche Applied Science, Indianapolis, IN). Primer sets specific to the selected genes were designed using Oligo Primer Analysis Software (Molecular Biology Insights, Inc., Cascade, CO) or Primer3 7 (Online Table 1), and Mg2+ concentration, annealing temperature, and primer concentration were optimized for each gene. Ubiquitin (GenBank Accession Number U72496) was used for normalization of cDNA quantity, as it was unchanged between aortic and ventricular side endothelium. QRT-PCR for each gene was performed in duplicate on paired aortic and ventricular side samples from ten animals, and a ratio was calculated for each animal. Statistical significance was tested by applying the same permutation strategy as with the microarray data, again with the one-sample T-statistic, but this time to produce permutation P-values. Histology and immunostaining. The following primary antibodies were used for immunohistochemical staining: monoclonal mouse anti-PECAM-1 Lot #62 1:10 (kindly provided by Dr. S. Albelda, University of Pennsylvania, Philadelphia, PA); polyclonal rabbit anti-mouse leukocyte 12-lipoxygenase 1:50 (Cayman Chemical, Ann Arbor, MI); polyclonal goat anti-human osteoprotegerin 1:6 (R&D Systems, Minneapolis, MN); and polyclonal rabbit anti-human von Willebrand factor 1:250 (Dako Corporation, Carpinteria, CA).

of 10

To assess purity of the cells recovered by the modified Hautchen method, the frozen cells on the coverslips were fixed in 4% paraformaldehyde and then permeabilized with 0.1% Triton-X. Non-specific primary antibody binding was blocked with 3% bovine serum albumin (BSA) and the coverslips were incubated overnight at 4°C with rabbit anti-human von Willebrand Factor and mouse anti-human PECAM-1. The following day, secondary antibody staining was done at room temperature (1 hour incubation time) with goat anti-mouse IgG FITC-labeled (1:100 dilution in blocking buffer; Organon Teknika Corp. West Chester, PA) and goat anti-rabbit IgG TRITC-labeled (1:100; Zymed Laboratories Inc., San Francisco, CA). Nuclei were stained with Hoechst 33258 (Sigma), and the coverslips were mounted with SlowFade Light (Molecular Probes, Inc., Eugene, OR) and sealed. The coverslips were observed on an Axioplan microscope (Carl Zeiss, Inc., Thornwood, NY) and images were acquired using a Zeiss AxioCam MR5 digital camera. For whole valve leaflet histology, porcine aortic valve leaflets were fixed in 10% neutral-buffered formalin, paraffin-embedded, and serial sectioned. Some sections were stained with hematoxylin and eosin using standard protocols. For immunohistochemical staining, sections were de-paraffinized and re-hydrated, and in some cases, antigens were retrieved by incubating the sections for 5 minutes at 37°C in a proteinase K solution (0.01 mg/mL Proteinase K; 50 mM Tris-CL, pH 8.0; 5 mM EDTA, pH 8.0). Immunohistochemical staining was performed using the DAKO Envision System Peroxidase (AEC) Kit (for rabbit or mouse primary antibodies) or the R&D Systems Cell and Tissue Staining (HRP-AEC) Kit (for goat primary antibodies) according to the manufacturer’s instructions. In all cases, sections were incubated with primary antibodies overnight at 4°C. For both kits, primary antibodies were detected using horseradish peroxidase-based techniques and 3-amino-9-ehtylcarbazole as the chromogenic substrate. Sections were counterstained with Mayer’s hematoxylin (Sigma-Aldrich, Inc., St. Louis, MO) and mounted using aqueous mounting medium (R&D Systems). The histological sections were observed on an Axioplan microscope (Carl Zeiss, Inc., Thornwood, NY) and images were acquired using a Zeiss AxioCam MRc5 digital camera.

of 10

Online Table 1 – Primer sets used for QRT-PCR validation

Gene Genbank Accession Number

Primer pair Amplicon size

CX43 X86023 Forward 5’ ATA CGC TTA TTT CAA TGG CTG CTC 3’ Reverse 5’ CAT GGG AGT TAG AGA TGG TGC TT 3’ 203

FGFR2 AF044944 Forward 5’ AAC GAT TAC GGG TCC ATC AA 3’ Reverse 5’ CCG TTC TTT TCC ACG TGT TT 3’ 191

GPX3 AY368622 Forward 5’ GAA CTG AAT GCA CTG CAG GA 3’ Reverse 5’ GTT CAC GTC CCC TTT CTC AA 3’ 183

GSTO1 NM_214050 Forward 5’ GCC AGA TGA CCC CTA TGA GA 3’ Reverse 5’ GGG CCA GAT GAG GTA ATC AA 3’ 229

ICAM AF156712 Forward 5’ GAT CAA TGG AAC CGA GAA GGA GCA 3’ Reverse 5’ CAG GGC AGC AGA GCA GGA GAA 3’ 301

LGALS1 AY604429 Forward 5’ GGC AAA GAC AGC AAC AAC CT 3’ Reverse 5’ GCA GCT TGA TGG TGA GGT CT 3’ 199

NOS3 AY266137 Forward 5’ TGC ATG ACA TTG AGA GCA AAG G 3’ Reverse 5’ GAT GGT CGA GTT GGG AGC AT 3’ 81

OSTF NM_214005 Forward 5’ CTG AAC AGG CAG AAT CCA TTG ACA 3’ Reverse 5’ CCC TTC CAG GCA GCA GCA T 3’ 253

PECAM1 X98505 Forward 5’ CCT CGC CCA TTT CCT ACC AAC TTT 3’ Reverse 5’ CAG ACT CCA CCT CCT CGC TCA G 3’ 237

PTGS1 AF207823 Forward 5’ CCA AAG GGA AGA AGC AGT TG 3’ Reverse 5’ CAT AAA TGT GGC CGA GGT CT 3’ 210

SELE L39076 Forward 5’ TCT CTG CTC TCC CTT TGG T 3’ Reverse 5’ ATT GCT TTG CTT ATT ATT TGG TTC 3’ 308

SELP L39075 Forward 5’ AGC TCT TAT TAC TGG ATT GGG ATG 3’ Reverse 5’ GCT TTC GTT ATT GGG CTC ATT 3’ 120

UBA52 U72496 Forward 5’ TGA CCA GCA GCG TCT GAT T 3’ Reverse 5’ TCT TGT CGC AGT TGT ATT TCT GAG 3’ 167

VCAM1 L43124 Forward 5’ CAG CCC TCA GTA AAG ACA ACA CCA 3’ Reverse 5’ GTC ATC ATC ACG GAG TCA CCT TCT 3’ 116

VWF S78431 Forward 5’ GGT CAG CGG TGT GGA CGA G 3’ Reverse 5’ ACA AAC TCC GTG CTC CTG TTG AAG 3’ 243

of 10

Online Table 2 – Confidence levels of differential expression of calcification-related genes

Gene Genbank Accession Number PaGE Confidence Level*

MGP AW99947 0

AHSG (fetuin) M16961 0

ENPP1 NM_006208 0

OPN D14813 0.089

SPARC (osteonectin) J03040 0

SMAD6 AF037469 0.99

CA2 J03037 0

BGN BG111541 0

DCN BC005322 0

IBSP NM_004967 0

ALPI NM_001631 0

TGFB1 X02812 0.735

TNC X78565 0

EPHX2 AI301066 0.327

ESR1 X03635 0

TIMP1 BC000866 0.786

MMP1 AK024818 0

MMP2 J03210 0.216

MMP9 BG684746 0

MMP3 Not on array –

ANKH Not on array –

* Confidence level predicted by PaGE = (1 - False Discovery Rate). Therefore, the higher the confidence level, the more likely the gene is to be differentially expressed.

of 10

Online Table 3 – Supporting references for putative effects assigned in Tables 3 and 4

Calcification-related

BMP4 8, 9

CHAD 10, 11

CHRD 12

COLL11A1 13

FBN1 14, 15

HAPLN1 16

NPPC 17

PTH 17-19

PTN 20-22

TNFRSF11B 23, 24

Inflammation/Oxidation-related

ALOX12 25, 26

CCL13 27

FY 28

GPX3 29, 30

LGALS1 31

MGST2 32, 33

NOS3 34

NPPC 35

OLR1 36, 37

PRDX2 38

SELP 39

VWF 40-42

of 10

References for Supplemental Material 1. Simmons CA, Zilberberg J, Davies PF. A rapid, reliable method to isolate high quality

endothelial RNA from small spatially-defined locations. Ann Biomed Eng. 2004;32:1453-9.

2. Van Gelder RN, von Zastrow ME, Yool A, Dement WC, Barchas JD, Eberwine JH. Amplified RNA synthesized from limited quantities of heterogeneous cDNA. Proc Natl Acad Sci U S A. 1990;87:1663-7.

3. Passerini AG, Polacek DC, Shi C, Francesco NM, Manduchi E, Grant GR, Pritchard WF, Powell S, Chang GY, Stoeckert CJ, Jr., Davies PF. Coexisting proinflammatory and antioxidative endothelial transcription profiles in a disturbed flow region of the adult porcine aorta. Proc Natl Acad Sci U S A. 2004;101:2482-7.

4. Polacek DC, Passerini AG, Shi C, Francesco NM, Manduchi E, Grant GR, Powell S, Bischof H, Winkler H, Stoeckert CJ, Jr., Davies PF. Fidelity and enhanced sensitivity of differential transcription profiles following linear amplification of nanogram amounts of endothelial mRNA. Physiol Genomics. 2003;13:147-56.

5. Hosack DA, Dennis G, Jr., Sherman BT, Lane HC, Lempicki RA. Identifying biological themes within lists of genes with EASE. Genome Biol. 2003;4:R70.

6. Manduchi E, Grant GR, He H, Liu J, Mailman MD, Pizarro AD, Whetzel PL, Stoeckert CJ, Jr. RAD and the RAD Study-Annotator: an approach to collection, organization and exchange of all relevant information for high-throughput gene expression studies. Bioinformatics. 2004;20:452-9.

7. Rozen S, Skaletsky HJ. Primer3 on the WWW for general users and for biologist programmers. In: S K, S M, eds. Bioinformatics Methods and Protocols: Methods in Molecular Biology. Totowa, NJ: Humana Press; 2000:365-86.

8. Shin V, Zebboudj AF, Bostrom K. Endothelial cells modulate osteogenesis in calcifying vascular cells. J Vasc Res. 2004;41:193-201.

9. Kaden JJ, Bickelhaupt S, Grobholz R, Vahl CF, Hagl S, Brueckmann M, Haase KK, Dempfle CE, Borggrefe M. Expression of bone sialoprotein and bone morphogenetic protein-2 in calcific aortic stenosis. J Heart Valve Dis. 2004;13:560-6.

10. Larsson T, Sommarin Y, Paulsson M, Antonsson P, Hedbom E, Wendel M, Heinegard D. Cartilage matrix proteins. A basic 36-kDa protein with a restricted distribution to cartilage and bone. J Biol Chem. 1991;266:20428-33.

11. Neame PJ, Sommarin Y, Boynton RE, Heinegard D. The structure of a 38-kDa leucine-rich protein (chondroadherin) isolated from bovine cartilage. J Biol Chem. 1994;269:21547-54.

12. Zhang D, Ferguson CM, O'Keefe RJ, Puzas JE, Rosier RN, Reynolds PR. A role for the BMP antagonist chordin in endochondral ossification. J Bone Miner Res. 2002;17:293-300.

13. Eyre DR, Eyre D, Schurman DJ. Collagens and cartilage matrix homeostasis. Clin Orthop. 2004:S118-22.

14. Kitahama S, Gibson MA, Hatzinikolas G, Hay S, Kuliwaba JL, Evdokiou A, Atkins GJ, Findlay DM. Expression of fibrillins and other microfibril-associated proteins in human bone and osteoblast-like cells. Bone. 2000;27:61-7.

of 10

15. Pereira L, Lee SY, Gayraud B, Andrikopoulos K, Shapiro SD, Bunton T, Biery NJ, Dietz HC, Sakai LY, Ramirez F. Pathogenetic sequence for aneurysm revealed in mice underexpressing fibrillin-1. Proc Natl Acad Sci U S A. 1999;96:3819-23.

16. Neame PJ, Barry FP. The link proteins. Experientia. 1993;49:393-402. 17. Huang Z, Li J, Jiang Z, Qi Y, Tang C, Du J. Effects of adrenomedullin, C-type natriuretic

peptide, and parathyroid hormone-related peptide on calcification in cultured rat vascular smooth muscle cells. J Cardiovasc Pharmacol. 2003;42:89-97.

18. Jono S, Nishizawa Y, Shioi A, Morii H. 1,25-Dihydroxyvitamin D3 increases in vitro vascular calcification by modulating secretion of endogenous parathyroid hormone-related peptide. Circulation. 1998;98:1302-6.

19. Shao JS, Cheng SL, Charlton-Kachigian N, Loewy AP, Towler DA. Teriparatide (human parathyroid hormone (1-34)) inhibits osteogenic vascular calcification in diabetic low density lipoprotein receptor-deficient mice. J Biol Chem. 2003;278:50195-202.

20. Neame PJ, Young CN, Brock CW, Treep JT, Ganey TM, Sasse J, Rosenberg LC. Pleiotrophin is an abundant protein in dissociative extracts of bovine fetal epiphyseal cartilage and nasal cartilage from newborns. J Orthop Res. 1993;11:479-91.

21. Sato Y, Takita H, Ohata N, Tamura M, Kuboki Y. Pleiotrophin regulates bone morphogenetic protein (BMP)-induced ectopic osteogenesis. J Biochem (Tokyo). 2002;131:877-86.

22. Tare RS, Oreffo RO, Clarke NM, Roach HI. Pleiotrophin/Osteoblast-stimulating factor 1: dissecting its diverse functions in bone formation. J Bone Miner Res. 2002;17:2009-20.

23. Bucay N, Sarosi I, Dunstan CR, Morony S, Tarpley J, Capparelli C, Scully S, Tan HL, Xu W, Lacey DL, Boyle WJ, Simonet WS. Osteoprotegerin-deficient mice develop early onset osteoporosis and arterial calcification. Genes Dev. 1998;12:1260-8.

24. Kaden JJ, Bickelhaupt S, Grobholz R, Haase KK, Sarikoc A, Kilic R, Brueckmann M, Lang S, Zahn I, Vahl C, Hagl S, Dempfle CE, Borggrefe M. Receptor activator of nuclear factor kappaB ligand and osteoprotegerin regulate aortic valve calcification. J Mol Cell Cardiol. 2004;36:57-66.

25. Harats D, Shaish A, George J, Mulkins M, Kurihara H, Levkovitz H, Sigal E. Overexpression of 15-lipoxygenase in vascular endothelium accelerates early atherosclerosis in LDL receptor-deficient mice. Arterioscler Thromb Vasc Biol. 2000;20:2100-5.

26. Reilly KB, Srinivasan S, Hatley ME, Patricia MK, Lannigan J, Bolick DT, Vandenhoff G, Pei H, Natarajan R, Nadler JL, Hedrick CC. 12/15-Lipoxygenase activity mediates inflammatory monocyte/endothelial interactions and atherosclerosis in vivo. J Biol Chem. 2004;279:9440-50.

27. Berkhout TA, Sarau HM, Moores K, White JR, Elshourbagy N, Appelbaum E, Reape RJ, Brawner M, Makwana J, Foley JJ, Schmidt DB, Imburgia C, McNulty D, Matthews J, O'Donnell K, O'Shannessy D, Scott M, Groot PH, Macphee C. Cloning, in vitro expression, and functional characterization of a novel human CC chemokine of the monocyte chemotactic protein (MCP) family (MCP-4) that binds and signals through the CC chemokine receptor 2B. J Biol Chem. 1997;272:16404-13.

28. Chaudhuri A, Rodriguez M, Zbrzezna V, Luo H, Pogo AO, Banerjee D. Induction of Duffy gene (FY) in human endothelial cells and in mouse. Cytokine. 2003;21:137-48.

of 10

29. Maddipati KR, Marnett LJ. Characterization of the major hydroperoxide-reducing activity of human plasma. Purification and properties of a selenium-dependent glutathione peroxidase. J Biol Chem. 1987;262:17398-403.

30. Bierl C, Voetsch B, Jin RC, Handy DE, Loscalzo J. Determinants of human plasma glutathione peroxidase (GPx-3) expression. J Biol Chem. 2004;279:26839-45.

31. La M, Cao TV, Cerchiaro G, Chilton K, Hirabayashi J, Kasai K, Oliani SM, Chernajovsky Y, Perretti M. A novel biological activity for galectin-1: inhibition of leukocyte-endothelial cell interactions in experimental inflammation. Am J Pathol. 2003;163:1505-15.

32. Hayes JD, Flanagan JU, Jowsey IR. Glutathione Transferases. Annu Rev Pharmacol Toxicol. 2004;45:51-88.

33. Sjostrom M, Jakobsson PJ, Heimburger M, Palmblad J, Haeggstrom JZ. Human umbilical vein endothelial cells generate leukotriene C4 via microsomal glutathione S-transferase type 2 and express the CysLT(1) receptor. Eur J Biochem. 2001;268:2578-86.

34. Wung BS, Cheng JJ, Shyue SK, Wang DL. NO modulates monocyte chemotactic protein-1 expression in endothelial cells under cyclic strain. Arterioscler Thromb Vasc Biol. 2001;21:1941-7.

35. Suga S, Itoh H, Komatsu Y, Ogawa Y, Hama N, Yoshimasa T, Nakao K. Cytokine-induced C-type natriuretic peptide (CNP) secretion from vascular endothelial cells--evidence for CNP as a novel autocrine/paracrine regulator from endothelial cells. Endocrinology. 1993;133:3038-41.

36. Cola C, Almeida M, Li D, Romeo F, Mehta JL. Regulatory role of endothelium in the expression of genes affecting arterial calcification. Biochem Biophys Res Commun. 2004;320:424-7.

37. Matsunaga T, Hokari S, Koyama I, Harada T, Komoda T. NF-kappa B activation in endothelial cells treated with oxidized high-density lipoprotein. Biochem Biophys Res Commun. 2003;303:313-9.

38. Fujii J, Ikeda Y. Advances in our understanding of peroxiredoxin, a multifunctional, mammalian redox protein. Redox Rep. 2002;7:123-30.

39. Andre P. P-selectin in haemostasis. Br J Haematol. 2004;126:298-306. 40. Ruggeri ZM. Von Willebrand factor, platelets and endothelial cell interactions. J Thromb

Haemost. 2003;1:1335-42. 41. Theilmeier G, Michiels C, Spaepen E, Vreys I, Collen D, Vermylen J, Hoylaerts MF.

Endothelial von Willebrand factor recruits platelets to atherosclerosis-prone sites in response to hypercholesterolemia. Blood. 2002;99:4486-93.

42. Methia N, Andre P, Denis CV, Economopoulos M, Wagner DD. Localized reduction of atherosclerosis in von Willebrand factor-deficient mice. Blood. 2001;98:1424-8.

A B

C D

E F

Online Fig. 1. Photomicrographs and histological sections of an aortic valve leaflet from a mature diabetic pig fed a high cholesterol diet for six months. As in humans, lesions formed preferentially on the (A) aortic surface (red arrows) as opposed to the (B)ventricular surface of the leaflet. Lipid deposition within the lesions was confirmed by positive oil red O staining in the subendothelial region (C, D). Calcified nodules, detected by alizarin red S in a near adjacent section, were also observed in the lesions (E, F). In panels C-F, the aortic surface of the leaflet is toward the top. Scale bar is 200 μm (C,E) or 50 μm (D,F).

A B

Online Fig. 2. Cells isolated in patches by the modified Hautchen method from the entire (A) aortic and (B) ventricular surfaces of the same valve leaflet stained positively for CD31 (green) and von Willebrand factor (red), consistent with endothelial phenotype. Positive staining for α-smooth muscle actin and CD45 (characteristic of valve interstitial myofibroblasts and leukocytes, respectively) was not observed (data not shown). Scale bar is 20 μm.