Critical Role of Coaptive Strain in Aortic Valve Leaflet Homeostasis: Use of a Novel Flow Culture Bioreactor to Explore Heart Valve Mechanobiology.

BACKGROUND: Aortic valve (AV) disease presents critical situations requiring surgery in over 2% of the US population and is increasingly the reason for cardiac surgery. Throughout the AV cycle, mechanical forces of multiple types and varying intensities are exerted on valve leaflets. The mechanisms whereby forces regulate leaflet homeostasis are incompletely understood. We used a novel flow bioreactor culture to investigate alteration of AV opening or closure on leaflet genes. METHODS AND
RESULTS: Culture of rat AV was conducted in a flow bioreactor for 7 days at 37°C under conditions approximating the normal stroke volume. Three force condition groups were compared: Cycling (n=8); always open (Open; n=3); or always closed (Closed; n=5). From each culture, AV leaflets were pooled by force condition and RNA expression evaluated using microarrays. Hierarchical clustering of 16 transcriptome data sets from the 3 groups revealed only 2 patterns of gene expression: Cycling and Closed groups clustered together, whereas Open AV were different (P<0.05). Sustained AV opening induced marked changes in expression (202 transcripts >2-fold; P<0.05), whereas Closed AV exhibited similar expression pattern as Cycling (no transcripts >2-fold; P<0.05). Comparison with human sclerotic and calcific AV transcriptomes demonstrated high concordance of >40 Open group genes with progression toward disease.
CONCLUSIONS: Failure of AV to close initiates an extensive response characterized by expression changes common to progression to calcific aortic valve disease. AV coaptation, whether phasic or chronic, preserved phenotypic gene expression. These results demonstrate, for the first time, that coaptation of valve leaflets is a fundamentally important biomechanical cue driving homeostasis.

Introduction

The aortic valve (AV) transiently seals the outflow tract, a function critical to life. In the United States, valve diseases presenting situations requiring surgery account for nearly 20% of all cardiac surgeries.1, 2 Understanding of the mechanisms of homeostasis or deterioration of AV leaflet is critically important for management of valve disease in the clinical setting. Undergoing more than 40 million cycles per human year without wearing out, heart valves are normally quite durable tissues. This indicates a highly robust ability to self‐repair, yet the mechanistic basis of this homeostatic maintenance and repair process is still poorly understood.

The question we pursued in this study is how the mechanical forces experienced by cycling AV affect normal leaflet homeostasis as well as the responses to its disruption by disease or mechanical intervention. To provide greater insight into the mechanisms actively mediating leaflet tissue homeostasis, we specifically asked how disruption of the valve cycle affected the normal expression program of AV leaflet genes. We developed a novel ex vivo AV flow‐culture system with which we could systematically alter the forces to which leaflets are exposed in a controlled manner, and then study how disrupting cyclic valve coaptation modified leaflet gene expression.

In clinical settings, where valve closure is disrupted by aortic valve regurgitation, leaflets undergo negative remodeling of their extracellular matrix to a thickened and fibrotic state of reduced flexibility and elasticity.3, 4 This tissue‐level response to impaired coaptive stretch and disrupted flow conditions indicates a major role of blood‐flow–induced mechanical forces (stretch, flow shear) in the maintenance of valve leaflet architecture.

In this report, we describe the effects of interrupting the AV cycle on leaflet gene expression. We found that alteration of the forces acting on AV by keeping valves open markedly increases the expression of multiple types of genes, whereas allowing them to undergo the coaptive stretch of closure and remain closed has essentially the same effect as allowing them to cycle.

Methods

Animals

Female Sprague‐Dawley rats (250 g; Charles River Laboratories, Hollister, CA) were used for these studies under a Stanford University institutional review board–approved protocol. Under aseptic conditions, aortic valve grafts comprised of minimal myocardial cuff and aorta were harvested and mounted on sterile custom‐fabricated cannula and stent fittings for anterograde flow. Both coronary arteries were ligated proximally to enable cyclic valve closure.

Novel Valve Culture Bioreactor System

We designed a novel bioreactor system specifically for use in these studies to replicate the normal valve cycle sequence: opening, closing, and pressurization by the arterial filling volume. It consisted of four main components: 500‐mL medium reservoir (Corning, Edison, NJ); peristaltic roller pump (Masterflex; Cole‐Parmer, Vernon Hills, IL); a custom‐fabricated glass chamber serving as the ventricular surrogate; and a custom‐fabricated acrylic afterload compliance chamber (Figure 1). Medium (EGM‐2 MV; Lonza, Walkersville, MD) was continuously pumped in a pulsatile manner (Easy‐load II pump head; Cole‐Parmer) from the reservoir into the ventricular chamber in which 8 rat valve grafts underwent anterograde flow through individual outflow channels connected to a compliance chamber that provided an afterload filling volume for cyclic valve closure (“Cycling” condition) at 80 beats per minute (bpm) and returned medium to the reservoir. A separate outflow connection from the “ventricular” chamber shunted excess flow directly to the reservoir. Two force‐designated groups of 4 valves each were studied—either Cycling and Open, or Cycling and Closed conditions—within a single culture experiment, keeping all other variables common. The rate of fluid flow through each Cycling valve graft was adjusted to achieve a complete valve cycle function at a moderate cycle rate (80 bpm; compared with ≈350‐bpm normal rat heart rate) under a flow rate that approximated the known right ventricular stroke volume (160 μL) of the rat heart5, 6 and a pressure cycle (5–28 mm Hg). Flow rate was monitored with inline ultrasonic flow probes (Transonic Systems Inc., Ithaca, NY). Bioreactor media was maintained at 37°C with a surface heater (Omega Engineering, Stamford, CT) beneath the reservoir and aerated with a gas mixture (5% CO2, 21% O2, and 74% N2) to maintain a pH of 7.4.

Figure 1

Bioreactor flow culture system for aortic valves. The principle components of the flow culture system are shown. Culture media in a 0.5‐L volume glass side‐arm flask reservoir (1) is heated to maintain 37°C with a surface heater. A peristaltic pump (2) moves media into a glass ventricular surrogate (3) reactor vessel (250‐mL volume). A total of 8 intact rat aortic valves are mounted in a circular array inside the reactor (3) to provide physiologic anterograde flow. The outflows from the reactor vessel are connected to the afterload chamber (4). Outflow from the afterload chamber returns media to the reservoir (1).

Three different conditions of mechanical force were studied: Cycling, as described above, mimics normal physiological valve opening, closure, and full coaptation, indicating that the physiomimetic stretch loading of leaflets and pressurization of sinus by filling volume induces the normal transient coaptive stretch of leaflets. Second, an “Open” condition was used to sustain valve leaflets in a neutral‐buoyancy, open state parallel to the outflow tract wall with minimal leaflet motion. The constantly open position state, in which AV experience neither stretch nor laminar flow shear, is a baseline zero point for forces acting on valves in which the only major force acting on the leaflets is the “ventricular” hydrostatic pressure. The third force condition we studied is that of the chronically closed valve (Closed) achieved by blocking forward flow distal to the valves. Full coaptation indicated physiomimetic stretch loading of leaflets.

Expression Microarray Analysis of Leaflet RNA

After 7 days of bioreactor culture, valve grafts were explanted to ice‐cold PBS, then leaflets from each force condition group were harvested, pooled, and flash frozen in liquid nitrogen. Total RNA was isolated from frozen leaflets with TRIzol reagent (Invitrogen, San Diego, CA) according to manufacturer's protocols, followed by an additional RNA purification with an RNAeasy Mini Kit (Qiagen, Redwood City, CA). Quality of RNA was assessed by UV absorption spectrophotometry (NanoDrop Technologies, ThermoFisher, Santa Clara, CA), and on‐chip electrophoretic separation with quantitation (Bioanalyzer 2100; Agilent Technologies, Santa Clara, CA).

Intact total RNA from each sample pool was used for amplification, labeling, and hybridization on microarray slides for expression analysis. A total of 16 array data sets were thereby derived from 64 rats in 8 experiments using 8 aortic valves cultured under 2 force conditions in groups of 4 valves. Given that the Cycling condition is common to each experiment, there were 8 Cycling samples, 5 Closed, and 3 Open subjected to microarray analysis. Samples were hybridized on Rat Genome 230 2.0 Array slides (Affymetrix, Santa Clara, CA) using standard protocols for cRNA labeling, hybridization, and detection procedures. GeneSpring GX 12.6 (Agilent Technologies) software and the Partek Genomic Suite (version 6.6; St Louis, MO) were used for statistical analysis and to visualize the microarray data.

Statistical Analysis

The resulting raw array data CEL format files were uploaded to GeneSpring GX (version 12.6). Probe‐level data were then compiled, normalized, and transformed using Robust Multichip Analysis. The following criteria were applied to filter the differentially expressed transcript list: a fold change of >2 and a Wilcoxon rank‐sum test where P<0.05 with Benjamini‐Hochberg estimation of false discovery rate (FDR). The resulting list of differentially expressed transcripts was uploaded into a BioBase ExPlain 3.1 data system (BIOBASE Biological database; Qiagen) and Ingenuity Pathway Analysis (IPA; Ingenuity Systems, Redwood City, CA) as the starting points for Gene Ontogeny (GO) analysis and generation of biological networks. In both systems, a P value is calculated to estimate the probability that each biological function and/or disease assigned to the data set of interest is not attributed to chance alone. Additionally, differentially expressed transcripts and genes are classified either by proprietary ontology's or by MESH terms according to the categories of canonical pathways, therapeutic target, biomarker, and molecular mechanism. The statistical significance of the association between the data set and the categories is estimated from the ratio of the number of proteins from the data set that map to the category divided by the total number of proteins that map to the canonical pathway. To define and identify patterns of gene expression profiles among the 3 valve culture groups (total n=16), unsupervised Hierarchical Clustering analysis was performed based on the values of the differentially expressed genes to determine the relatedness of expression patterns among the entire 16‐sample group.

Real‐time polymerase chain reaction (PCR) was performed as previously reported7 to confirm expression of selected genes identified by microarray analysis. Briefly, total RNA from the 16 RNA pools used for array and additional experiments was reverse‐transcribed (RT) to cDNA using the Superscript II RNase H Reverse Transcriptase kit (Invitrogen). An initial RT reaction minus RNA template confirmed that all RNA samples were free of contaminating genomic DNA. Forward and reverse gene‐specific primers (Table 1) were used for quantitative real‐time PCR analysis. Reactions were run on a ViiA 7 PCR instrument (Life Technologies, South San Francisco, CA), and mRNA levels were determined using the standard curve method. Expression values were normalized to the GAPDH expression level in a parallel reaction with the same amount of cDNA template. The fold change in expression was computed as the ratio of expression in Open condition divided by the expression in Cycling or Closed condition for paired tissue sample pools from each experiment in a total of 3 experiments per condition.

Table 1

Primers Used for PCR Analysis

Gene	Primer for Human Gene	Amplicon Length (bp)	Gene Bank ID
Apln	TGCCTCCAGATGGGAAAGG CCCTGGTCCAGTCCTCGAA	80	AF179679
Aplnr	CCCTTCCTCTATGCCTTCTTTG TGGTCACAGCAGAGCATGGA	68	NM_031349
Cldn5	GCTGCCAGAGGAATGCGTTA GGGCAAGTCCTTTGGTTCAG	76	NM_031701
RT1‐Da	CATCATCCAGGCGGAGTTCT CGTCACCGTCAAAGTCAAACA	71	NM_001008847
RT1‐Ba	AGCCCCTGTGGAGGTCAAG CATATTTATACCATAGGCGGCTACGT	63	NM_001008831
RT1‐Bb	AGAGTGTTTTGGTTGTGTTTGAAGAG GCAGGATTTGATGCGGAAA	67	NM_001004088
RT1‐Db1	GACGCAGCCCCTGAAACA ACGCTGCCAGGGTAGAAGTC	66	NM_001008884
Pecam1	GGAAACCAACAGCCATTACGA AGGGAGGACACTTCCACTTCTG	69	NM_031591

bp indicates base pairs; PCR, polymerase chain reaction.

Results

AV Flow Culture Bioreactor System

The primary design strategy for the bioreactor system used in these studies (Figure 1) was to induce continuous cyclic opening and closing of the AV (Cycling condition) in a physiomimetic manner, which was verified visually and monitored by video during culture. Eight valves were cultured simultaneously (Figure 2A). Imaging under Cycling condition demonstrated complete closing (Figure 2B) and opening (Figure 2C) of valves with full coaptation of all leaflets, demonstrating valve competence. Our bioreactor system therefore replicates a normal valve functional cycle. A video recording of the valve dynamics during a typical culture experiment is provided in Video S1. In contrast to Cycling condition, we also cultured valves in a “Closed” condition to examine a state similar to the AV of nonpulsatile ventricular assist device (VAD) patients. The Closed condition induces full coaptation and thereby the physiomimetic stretch loading of leaflets that normally occurs transiently in Cycling valves, although considered a nonphysiological condition in general. A third, “Open” condition was also studied to model a state in which minimal forces are exerted on the AV, during which forward transvalvular flow was blocked distally and the leaflets remained in a constantly open position without laminar shear or strain forces. The Open condition was designed to create a high level of disruption of leaflet homeostasis and a baseline for the assessing effects of flow and coaptation on the self‐maintenance process.

Figure 2

Flow‐induced cyclic closure of rat aortic valves in flow culture. Eight valve grafts were cultured within the same experiment (A). Physiomimetic valve cycling was demonstrated by complete cyclic closing (B) and opening (C) of valves under flow condition. Ruler in (B and C), 5 mm. A video clip demonstrating cyclic closure is provided in Video S1.

Force Alters Gene Expression in AV

Hierarchical clustering analysis of genes regulated in the Open group compared to the Cycling and Closed groups was performed to objectively assess the statistical heterogeneity in expression pattern among the 3 groups. The expression “heat” map array (Figure 3) depicts the hierarchical clustering dendrogram from this analysis. Two notable outcomes were apparent: First, based solely on their gene expression magnitude patterns, the 3 different culture conditions formed only 2 major clusters, the Open condition samples, and another cluster containing both Cycling and Closed condition samples. Second, expression changes observed in the Open group cluster were highly distinct from the other cluster, exhibiting a preponderance of upregulated genes (red bars in the heat map). Upregulation of gene expression in valves that stay open indicates that mechanical forces (laminar flow shear, stress, and strain) sustain a normally repressive effect on their expression. The clustering dendrogram at the top of Figure 3 shows that the Cycling and Closed groups are very highly related, even though the samples are from independent experiments in which the forces attendant to coaptation were experienced either phasic and cyclical or chronically static (Closed) for the culture duration.

Figure 3

Hierarchical clustering analysis results arrayed as heat map. Transcriptome data for open, closed, and cycling AV were analyzed by hierarchical clustering and the results arrayed as heat map. The treatment condition is indicated below the data columns: O, Open (n=3); C, Closed (n=5), unlabeled, Cycling (n=8). The results of clustering analysis are represented by the dendrogram above the data columns. AV indicates aortic valve.

A summary of the up‐ and downregulated genes and their fold‐change cutoff are listed in Figure 4A. We identified 202 genes in the Open valve group (fold >2.0; P<0.05, t test, followed by Benjamini and Hochberg FDR correction) that were differentially regulated compared with Cycling valves, and 211 genes in Open group versus Closed group comparison (Figure 4A). However, no significant differential expression was observed between the Cycling and Closed groups at the P<0.05 cutoff for >2‐fold difference, indicating significant homogeneity of the 2 groups. Furthermore, the data of Figure 4A confirm the very high degree of overlap in expression values between Cycling and Closed valves. Table S1 lists the gene transcripts expressed >2‐fold in the Open condition compared with the other 2 culture conditions. As shown in the Venn diagrams (Figure 4B and 4C), 190 differentially expressed transcripts were common between these 2 force condition groups (Figure 4B), and the majority of genes were upregulated. When Open valves were compared with either the Closed or Cycling valves, more than 90% of the >2‐fold regulated transcripts were common to both groups. A similar high level of concordance was observed in the subsets of genes expressed at >4‐fold (Figure 4A and 4C) and >10‐fold changes. To further assess the homogeneity of the Closed and Cycling group transcriptomes, we compared the 28 transcripts regulated in the >1.2‐ to <2‐fold change level and confirmed that the few differences between the 2 groups were of no apparent biological significance.

Figure 4

Integrated analysis of gene expression patterns. The data table (A) indicates the number of differentially regulated transcripts as a function of the fold‐change cut‐off value for the 3 comparison groups. Venn diagrams depict the transcripts with overlapping expression among the 3 culture conditions at a 2‐fold change level (B) and a 4‐fold change level (C) in the study group (1‐way ANOVA corrected P≤0.05). The overlap in the Venn diagrams indicates the transcripts present in both comparisons.

Genes Markedly Regulated in AV That Stay Open

In valves that stay open, we observed that over 30 genes were highly changed (>4‐fold) in expression level. In Table 2, we list the top 10 upregulated and top 6 downregulated genes. The P‐value scores for the significance of their fold change are also provided. The magnitude of change was far greater for upregulated genes, which comprised 4 genes classically associated with antigen presentation to immune cells and 3 genes for receptor molecules. The most downregulated genes code for molecules with diverse functions. Voltage‐sensitive ion channel genes were also highly regulated: potassium channel (Kcne3) up‐ and chloride channel (Clcn4) downregulated.

Table 2

List of the Top 10 Genes Up‐ and Down‐regulated in Open Versus Cycling and Open Versus Closed Groups, Based on the Statistical Significance P Value

Gene Symbol	Gene Annotation	Open vs Cycling		Open vs Closed
		Fold	P Value	Fold	P Value
Cd74	Cd74 molecule, major histocompatibility complex	17.64	2.09E‐05	26.26	1.20E‐05
Tifab	TRAF‐interacting protein with forkhead‐associated domain	17.57	4.13E‐04	16.69	8.93E‐04
RT1‐Da	Major histocompatibility complex, class II, Da	16.17	3.30E‐06	23.38	1.94E‐06
RT1‐Bb	Major histocompatibility complex, class II, Bb	15.51	1.90E‐09	20.16	2.77E‐08
RT1‐Ba	Major histocompatibility complex, class II, Ba	13.60	3.33E‐08	17.61	1.51E‐09
Kcne3	Potassium voltage‐gated channel, Isk‐related family 3	12.73	1.02E‐03	11.04	2.75E‐03
RT1‐Db1	Major histocompatibility complex, class II, Db1	10.79	1.31E‐07	12.34	1.67E‐07
Aplnr	Apelin receptor	10.69	2.67E‐05	11.17	4.71E‐05
Kdr	Kinase insert domain receptor	10.39	1.07E‐03	11.86	1.26E‐03
Cldn5	Claudin 5	9.15	7.58E‐07	8.54	2.50E‐06
Scrn1	Secernin 1	−2.41	2.16E‐04	−2.52	2.70E‐04
Clcn4	Chloride channel, voltage‐sensitive 4	−2.30	1.15E‐03	−2.42	2.59E‐03
Pdlim3	PDZ and LIM domain 3	−2.74	3.10E‐05	−2.68	8.16E‐05
Slc6a15	Solute carrier family 6, member 15	−3.34	8.29E‐04	−3.59	9.56E‐04
Scx	Scleraxis	−3.66	1.37E‐05	−2.77	2.74E‐04
Mlana	Melan‐A	−3.89	7.91E‐05	−3.50	3.30E‐04

P value adjusted by Benjamini and Hochberg false discovery rate correction.

Mechanical Force Regulates Multiple Gene Classes

We used GO analysis to classify the gene transcripts with altered expression between the Open and Cycling/Closed AV and quantitate their representation among 3 functional types: Molecular Function, Biological Process, and Cellular Components. Biological Process and Cellular Components were the predominantly represented classes at 45% and 40%, respectively. Molecular Function represented the smallest class at 15%.

Analysis using ExPlain 3.1 was used to explore statistically over‐represented groups and identify significantly represented GO categories. In the regulated gene sets of both Open versus Cycling/Closed groups at a cutoff of P<0.01, over 120 GO classes were enriched. The most highly enriched GO classes include Cardiovascular System Development, Angiogenesis, and Cell Proliferation. Figure 5 shows the major GO listings of in these 2 analyses. Given the very high degree of overlap, we considered the Cycling and Closed groups as a single group and henceforth compared the Open and Cycling groups in the GO analysis reported below.

Figure 5

Gene ontology in open valve upregulated genes. Enriched Gene Ontology (GO) categories are identified along with number of genes in each category. Blue bars represent actual enriched number of genes in Open versus Cycling AV, whereas red bars represent actual enriched number of genes in Open versus Closed AV. The x‐axis indicates the number of genes in each category (“Hits”). AV indicates aortic valve; MHC, major histocompatibility complex.

Functional Gene Pathway Analysis Identifies Key Canonical Pathways and Transcriptional Regulators

Functional analysis of genes with over 2‐fold altered expression in the 3 different force conditions was conducted using IPA and several distinct relationships were identified. The major gene network functions identified were Cardiovascular system development and function, Organismal development, and Cardiovascular diseases. The statistically significant canonical pathways in each of the 3 different conditions are shown in Tables S2 and S3. Pathway activity analysis indicates the major canonical pathways that are activated with z‐score >2.0. IPA upstream regulator analysis identified 58 activated and 17 inhibited regulators in Open versus Cycling. The top regulator genes are vascular endothelial growth factor (VEGF), fibroblast growth factor 2 (FGF2), tumor necrosis factor (TNF), hypoxia‐inducible factor (HIF) 1alpha, and nuclear factor kappa B (NF‐κB) complex (Table 3). In Table 4, we compare the genes common to the 5 most represented pathways. The detailed list of the genes targeted by these 5 regulator genes is provided in Table S4.

Table 3

Listing of Top Upstream Analysis‐Predicted Activation State in Differentially Regulated Genes

Upstream Regulator	Molecule Type	Predicted State	Activation z‐Score	P Value of Overlap
TNF	Cytokine	Activated	3.327	4.84E‐11
VEGF	Group	Activated	3.516	4.21E‐09
HIF1A	Transcription regulator	Activated	2.657	6.87E‐09
FGF2	Growth factor	Activated	3.50	1.87E‐07
NF‐κB	Complex	Activated	3.227	3.01E‐06

FGF2 indicates fibroblast growth factor 2; HIF1A, hypoxia‐inducible factor 1A; NF‐κB, nuclear factor kappa B; TNF, tumor necrosis factor; VEGF, vascular endothelial growth factor.

Table 4

Common Target Genes Activated in the Top Upstream Regulators

TNF Regulator	Vegf Regulator	HIF1A Regulator	FGF2 Regulator	NF‐κB Regulator	Fold Change
ADORA2A	ADORA2A				2.06
ANGPT2	ANGPT2		ANGPT2		5.33
APLN	APLN	APLN			3.48
BDNF		BDNF	BDNF	BDNF	2.42
CCR5		CCR5			2.52
CXCR4	CXCR4	CXCR4	CXCR4	CXCR4	2.22
	DLL4	DLL4	DLL4		2.13
EFNA1	EFNA1			EFNA1	2.16
IGFBP3	IGFBP3	IGFBP3	IGFBP3		4.03
IL1B		IL1B	IL1B	IL1B	2.11
KDR	KDR		KDR	KDR	10.39
LYVE1	LYVE1		LYVE1		3.37
Notch4	Notch4		Notch4		2.63
NR5A2	NR5A2				2.07
PECAM1	PECAM1			PECAM1	2.16
PLK2				PLK2	2.67
SLC2A1		SLC2A1	SLC2A1	SLC2A1	2.24
VEGFA	VEGFA	VEGFA	VEGFA	VEGFA	2.69

FGF2 indicates fibroblast growth factor 2; HIF1A, hypoxia‐inducible factor 1A; NF‐κB, nuclear factor kappa B; TNF, tumor necrosis factor; VEGF, vascular endothelial growth factor.

Force‐Regulated Rat AV Genes Are Also Active in Progression of Human AV Disease

We compared the transcripts regulated in Open AV with published transcriptome data for human diseased AV. The National Center for Biotechnology Information Gene Expression Omnibus (GEO) Datasets collection was used to access transcriptome data (GSE51472) for 3 groups in a total of 15 leaflet specimens of human AV: presclerotic (normal), sclerotic, and calcific AV generated by Ohukainen et al.8 and available at GEO (http://www.ncbi.nlm.nih.gov/gds/?term=GSE51472). Multiple genes that we identified as significantly upregulated in Open rat AV are also reported to be upregulated progressively with advancing human AV disease (Table 5). Overall, more than 40 genes identified as upregulated in Open rat AV are also upregulated in Human AV exhibiting precalcific sclerosis and in calcific AV disease. For example, CXCR4, CD74, ADAMTS9, HLA‐DA1, ANGPT2, and KDR were upregulated in both human and rat AV cultured under Open condition.

Table 5

Regulated Rat AV Genes also Expressed in Human Diseased AV

Rat AV This Study		Description	Human AV
Genes	Fold		Gene
Acp5	5.04	Acid phosphatase 5, tartrate resistant	ACP5
Adamts9	3.20	ADAM metallopeptidase with thrombospondin type 1 motif	ADAMTS9
Angpt2	5.33	Angiopoietin 2	ANGPT2
Ccr5	2.52	Chemokine (C‐C motif) receptor 5	CCR5
Cd74	17.64	Cd74 molecule, major histocompatibility complex, class II	CD74
Chst1	2.85	Carbohydrate (keratan sulfate Gal‐6) sulfotransferase 1	CHST1
Cldn5	9.15	Claudin 5	CLDN5
CXCR4	2.22	Chemokine (C‐X‐C motif) receptor 4	CXCR4
RT1‐Ba	13.60	Major histocompatibility complex, class II	HLA‐DQA1
RT1‐Bb	15.51	Major histocompatibility complex, class II	HLA‐DQB1
RT1‐Da	16.17	Major histocompatibility complex, class II	HLA‐DRA
RT1‐Db	10.79	Major histocompatibility complex, class II	HLA‐DRB1
Igfbp3	4.03	Insulin‐like growth factor binding protein 3	IGFBP3
Kcne3	12.73	Potassium voltage‐gated channel, Isk‐related family 3	KCNE3
Kdr	10.39	Kinase insert domain receptor	KDR
Lyve1	3.37	Lymphatic vessel endothelial hyaluronan receptor 1	LYVE1
Pdk1	2.24	Pyruvate dehydrogenase kinase, isozyme 1	PDK1
Ptp4a3	2.15	Protein tyrosine phosphatase type IVA, member 3	PTP4A3
Stc1	2.44	Stanniocalcin 1	STC1
Tgm2	3.29	Transglutaminase 2, C polypeptide	TGM2

AV indicates aortic valve.

PCR Validation of Differentially Expressed Genes

To confirm the differential expression observed in microarray analyses, we performed quantitative real‐time RT‐PCR analysis (Table 6) of 8 representative genes: Apelin (Apln); apelin receptor (Aplnr); claudin5 (Cldn5); endothelial cell adhesion molecule 1 (Pecam1); and a group of the highest regulated gene family: Rattus norvegicus RT1 class II antigen (Ba, Bb, Da, and Db1), orthologs of human MHC classes II B and D. Upregulation in Open condition was confirmed in all 8 genes. For these confirmed genes, RT‐PCR results represent a 90% concordance with the microarray data (ie, highest expression level of those genes with high signal intensity on the microarray).

Table 6

Comparison of PCR and Microarray Expression Values

Column ID	Gene	Open vs Cycling		Closed vs Cycling		Gene Bank ID
		PCR	Array	PCR	Array
1389651_at	Apln	3.68±1.86	3.48	1.56±0.70	1.23	AF179679
1379772_at	Aplnr	5.80±4.74	10.69	4.47±1.00	−1.04	NM_031349
1374104_at	Cldn5	9.19±2.57	9.15	2.32±1.89	1.07	NM_031701
1371514_at	CD31	ND	2.16	1.25±0.15	−1.07	NM_031591
1370883_at	RT1‐Da	14.38±10.71	16.17	1.77±0.80	−1.45	NM_001008847
1377334_at	RT1‐Ba	9.12±11.18	13.60	1.48±0.71	−1.29	NM_001008831
1371065_at	RT1‐Bb	ND	15.51	2.10±1.16	−1.30	NM_001004088
1370383_a	RT1‐Db1	12.52±2.36	10.79	1.62±0.63	−1.14	NM_001008884

Values are mean±SD of fold change in expression for n=3 independent determinations. ND indicates not determined; PCR, polymerase chain reaction.

Discussion

Heart valves continuously undergo changes in mechanical forces. The cyclical patterns of force change are sensed by valve leaflet endothelial and interstitial cells (VEC and VIC), and encoded as homeostatic signals that are still poorly understood. The focus of this study is the role of coaptive stretch forces on AV leaflet phenotype. We used a novel flow culture bioreactor system to isolate the separate effects of stretch and flow on rat AV grafts cultured either with normal cycling and forward flow, or without cycling or flow and constantly open (Open condition). In addition, we studied leaflet coaptive stretch in isolation by culturing with the AV constantly closed (Closed condition). The primary question we posed is how these 3 force conditions differentially affected leaflet gene expression patterns. We found that in valves remaining open during culture, multiple genes are expressed at much higher levels than in valves that close, whether closure is tonic or cyclic. This result demonstrated that valves that stay open (with leaflets unstretched) undergo activation of multiple gene expression pathways associated with valve remodeling or repair.

Analysis of our microarray expression data demonstrated that 2 signaling pathways were prominently activated when coaptive stretch was prevented: angiogenesis and Hif1alpha‐directed hypoxia (Hif) signaling. The increased expression of angiogenic genes is consistent with a response to repair endothelial cell damage through proliferative replacement. VECs are believed to be the primary source of new VICs as well, through the process of endothelial to mesenchymal transdifferentiation that generates valve leaflet precursor cushion structures during valve development.9 Activation of the Hif signaling pathway in Open valves raises the possibility that in the absence of coaptive stretch (and forward flow), these leaflets become oxygen deficient. Given that Hif activation was much less in Closed or Cycling valves, our data indicate that it is the chronic open, unstretched state, rather than an ambient nutrient deficiency, that drives Hif expression.

The Hif pathway is also known to regulate angiogenesis, for example, in tumor growth and also drives fibrosis by Vegf, Angpt2, and Igf. Therefore, given that the 2 most highly targeted pathways identified in Open valves are also highly interactive, Hif may be the primary transcriptional regulator. It is conceivable that the absence of leaflet stretch impairs the delivery of oxygen as well as nutrients to the core of leaflets. It has been shown that leaflets exist on the verge of core hypoxia because of their lack of vascularization,10 and that leaflet motion including coaptation helps to convectively deliver nutrients to the deeper regions of valves.11, 12 Indeed, human valves are in the 300‐ to 700‐μm thickness range,13 beyond the theoretical passive oxygen diffusion range of 100 μm14 when unloaded, yet remain fully viable. Of note, AV area increases at least 20% to 30% at coaptation.15 We speculate that this increase in leaflet area attributed to stretching may be accompanied by a sufficient reduction in leaflet thickness to significantly reduce the diffusion distance for oxygen. Consistent with an effect of stretch to facilitate leaflet oxygen delivery, we note that Closed AVs have the same low level of hypoxia‐regulated gene expression as do Cycling valves, suggesting that sustained physiological stretch of closure may be sufficient to reduce the diffusion barrier for oxygen, even without the convective assistance of leaflet cycling between stretched and relaxed states.

The findings of this study identify coaptive leaflet strain as a driver of leaflet homeostatic maintenance: Closed and Cycling AVs have a similar level of gene expression, whereas Open AVs uniquely exhibit marked upregulation of genes common to all 3 groups. Surprisingly, a role of forward flow‐mediated laminar shear force in homeostatic maintenance is not demonstrated by our results, given that forward flow was essentially absent from both Open and Closed conditions, yet gene expression was markedly different between these 2 conditions. The role of forward flow in regulating this gene expression response was further demonstrated to be minimal, given that closed valves that had no forward flow (Closed) had the same expression pattern as cycling valves with forward flow (Cycling). Therefore, although shear stress from forward flow and stretch from coaptation are expectedly interactive during normal valve cycling, our results indicate that flow has less influence on gene expression than does stretch. Importantly, flow shear is sensed by valve endothelium, which then indirectly regulates neighboring cells, whereas stretch is sensed directly by leaflet cells. The response of a larger cell population (VEC plus VIC) to direct stimulation by stretch, pressure, and torsion may partially explain the observed predominant effect of stretch on gene expression. Much of the literature on heart valve mechanobiology focuses on the role of laminar flow shear as a dominant driver of valve homeostasis.3, 11 This well‐accepted role of flow is a conceptual extension from vascular endothelial cells (ECs) to VEC: Flow shear forces affect the biology of both VEC and vascular ECs with many similarities and some differences.16, 17, 18 However, other forces acting on leaflets also clearly influence valve cellular biology.19 Elevated transvalvular pressure (eg, in the setting of hypertension) is a recognized risk factor for commissural fusion during VAD therapy as well as valvular diseases presumably attributed to the increased static pressure on valve cells and the increased strain loading on the closed AV root.13, 20 Connelly et al Leaflet stretch is recognized as an integral part of the valve cycle in studies of valve mechanics,19 though apparently considered a force the leaflet must endure to prevent pathology rather than a driver of valve homeostasis. Importantly, in the conditions of our study, chronically Closed valves had no laminar shear flow exposure, and were chronically stretched, demonstrating that chronic closure was not a strong inducer of altered transcription compared with AV that stay open and unstretched. Our results therefore provide strong evidence that the physiological straining of leaflets may be a major regulator of leaflet homeostasis. Expanded analysis of the AV tissue in our study and of the biological roles of the differentially expressed genes will help to confirm these intriguing initial findings.

Many of the genes regulated in Open AV are also regulated in human sclerotic and calcific valves based upon a comparison of published transcriptome data sets.8 Overall, more than 40 genes identified as upregulated in Open condition rat AV are also reported to be upregulated in human AV exhibiting precalcific sclerosis and in calcific AV disease (Table 5). This congruence in disease‐related gene expression changes across species demonstrates the potential clinical relevance of our ex vivo AV culture system as a platform for exploring the mechanisms through which mechanical forces regulate leaflet homeostasis. Furthermore, our study results therefore support the concept that reduced leaflet tissue nutrient and/or oxygen delivery may be a causative factor in human valve disease progression in a feed‐forward manner as leaflets thicken because of disease, or even simply aging.

Limitations

Changes in mRNA are inferred as changes in the biology relevant to this study. Although alterations in gene expression indicate definite cell‐level responses typically orchestrated at the nuclear level, they do not necessarily inform about potentially important changes at the levels of stored‐protein release, protein catalytic activity, and post‐translational modifications, such as phosphorylation and other subcellular signaling, which must be directly measured. RNA changes are informative, but we acknowledge that they are just a starting point to understanding how systems are modified by the perturbations under study. The observed changes necessarily constitute averages over the entire valve leaflets, so focal changes in a specific region and/or cell type may not be detected, and small changes may be obscured. Appreciating this reality, we have emphasized the most significant changes pending follow‐up analyses in ongoing studies.

The conditions of heart rate and pressure gradients present in this ex vivo culture system are both significantly lower than normal or peak in vivo levels of rats (ca 400 bpm and 90 mm Hg). This was deliberate and necessitated by limitations of the bioreactor system, which does nonetheless replicate what our data demonstrate to be critically important aspects of valve physiology. We note that preserving flow rate and valve sinus pressurization was sufficient to reveal significant differences in gene expression that we demonstrate to be consistent with and parallel to human calcific aortic valve disease, demonstrating that model systems in other species need not fully replicate in vivo physiology in order to reveal important biological mechanisms relevant to human disease. Importantly, it has been demonstrated that a pressure of 20 mm Hg is sufficient to completely uncrimp (relax and extend) the collagen fibrils of the aortic valve leaflet.21 Beyond this pressure, the AV leaflets resist stretch and act more like tendons. Because our focus was the role of leaflet stretch on homeostasis, the pressure range we studied is therefore more than adequate to achieve full stretch. Above 20 mm Hg to the peak systolic range of 90 mm Hg, compression of leaflets occurs and this force is expectedly sensed by all leaflet cells; an up to 30% further increase in VIC nuclear deformation is reported to occur between 20 and 90 mm Hg.22 Therefore, though our studies likely capture the full effect of leaflet stretch, they do not fully capture in vivo–level compressive forces and this limitation is acknowledged. The design and implementation of a bioreactor system that replicates additional aspects rat physiology is underway for the further investigation of our results. The additional increment in peak pressure to 90 mm Hg could conceivably alter the magnitude of the changes we observed; whether different genes become regulated will be determined in future studies.

Summary and Conclusions

In summary, we used gene expression profiling to assess the effects of disrupting normal aortic valve cyclic closure in a flow culture bioreactor model system. Valves that stay open and remain without the strain of coaptation exhibit high‐magnitude changes in the expression of many genes. Valves that remain closed with a physiological level of coaptive stretch exhibit a gene profile that essentially the same as that of cycling valves. Of note, comparing our results with published transcriptome data of human AV disease specimens revealed that many of the genes we observed as upregulated in rat AVs are also increased in human AVs exhibiting stenosis‐induced sclerosis and also valves with calcific disease. We therefore conclude that disruption of the normal mechanical forces upon valves, coaptive strain in particular, is sufficient to markedly alter their gene expression in a manner similar to that associated with human valvular disease. Our results also demonstrate that the absence of laminar flow shear (Open and Closed AV) may be of less critical importance relative to coaptive stretch, and this is a surprising finding, the implication/relevance of which needs to be further explored to be better understood.

Sources of Funding

This work was supported by the Lucile Packard Foundation for Children's Health, Alex Vibber Endowment (to Hanley), and the Oak Foundation (Grant No.: OUSA‐07‐096; to Hanley).

Disclosures

None.

Table S1. List of Differentially Regulated Genes in 1 Culture Condition Compared With the Other Culture Condition (fold >2.0; P<0.05). Values are the numeric ratio of their expression level (calculated as the fold change; negative values indicate downregulation).

Table S2. Identified Canonical Pathways and Their Significantly Regulated Genes in Open Versus Closed Comparison

Table S3. Identified Canonical Pathways and Their Significantly Regulated Genes in Open Versus Cycling Comparison

Table S4. Target Genes and Their Fold Changes for the Major Identified Regulators

Click here for additional data file.

Video S1. Grafts are imaged from the ventricular side of the outflow tract to demonstrate valve cycling with coaptation. The first 4 valves are in the Cycling condition. The next 4 valves are in the Closed position with sustained coaptation. The video ends with a return to the first Cycling valve shown at the start of the sequence.

Click here for additional data file.