Altered Micrornas in Bicuspid Aortic Valve: A Comparison between Stenotic and Insufficient Valves

Abstract

Background

Bicuspid aortic valve (BAV), the most common form of congenital heart disease, is a leading cause of aortic stenosis (AS) and aortic insufficiency (AI). AS is typically due to calcific valve disease. Recently, microRNAs (miRNAs) have been shown to modulate gene expression. This study examined miRNAs that were altered in aortic valve leaflets of patients with AS compared to patients with AI. Additionally, in vitro experiments were performed to examine if these miRNAs modulate calcification-related genes.

Materials and Methods

Aortic valve samples (fused or unfused leaflets) were collected from 9 patients (mean age 44.9±13.8 years) undergoing aortic valve replacement. PIQOR™ miRXplore Microarrays containing 1421 miRNAs were used and hybridized to fused leaflet samples labeled with Cy5; unfused samples were used as control and labeled with Cy3. Quantitative reverse transcription-polymerase chain reaction (qRT-PCR) was performed to validate the miRNA array results. Cultured human aortic valve cells (AVICs) were treated with miRNA mimics and qRT-PCR was performed to determine changes in mRNAs.

Results

Seven miRNAs were statistically different between the AS and AI patients by microarray. MiR-26a and miR-195 levels were reduced by 65% and 59% respectively with p<0.05 in the stenotic samples by qRT-PCR. MiR-30b was reduced by 62% (p<0.06) in the stenotic samples by qRT-PCR. Human AVICs treated with miR-26a or miR-30b mimics had decreased mRNA levels of calcification-related genes. MiR-26a repressed BMP2 by 36%, ALKALINE PHOSPHATASE (ALPL) by 38%, and SMAD1 by 26%. MiR-30b reduced expression of SMAD1 by 18% and SMAD3 by 12%. Whereas miR-195 treated AVICs had increased mRNA levels of calcification-related genes such as BMP2 by 68% and RUNX2 by 11%.

Conclusions

MiR-26a, miR-30b, and miR-195 were decreased in the aortic valves of patients requiring valve replacement due to AS compared to those being replaced due to AI. These miRNAs appear to modulate calcification related genes in vitro.

Introduction

Aortic valve calcification/stenosis has a significant clinical impact since it is the third leading cause of adult heart disease (1) and the most common form of acquired valvular disease in developed countries (2). The risk factor most closely linked to calcific aortic stenosis is bicuspid aortic valve (BAV), present in 1–2% of the population (2). Examination of human calcified aortic valve tissue and in vitro experiments using Aortic Valve Interstitial Cells (AVICs) have demonstrated that there are several gene pathways involved in calcification including BMP2(3, 4), NOTCH1(5–8), and the SMADs(3, 4, 9). miRNA are a class of highly conserved small noncoding RNAs that have key roles in modulation of gene expression(10). miRNAs are primarily involved in repressing mRNA transcripts by binding specific sequences in 3’ untranslated region (3’UTR) of the mRNA. There is a rapidly growing body of literature examining the role of miRNAs in cardiovascular development and disease (11–16). While miRNA profiling has been performed in hepatocellular(17) and neck carcinoma(18, 19), sepsis(20), and heart failure(21), we are not aware of any reports profiling miRNAs associated with aortic valve disease and their possible role in calcification. Since BAV predisposes to both stenosis and insufficiency (regurgitation) in our cohort, we sought to evaluate the differentially expressed miRNAs in BAV with AS as compared to AI. Additionally, we examined if these miRNAs could alter expression of calcification related genes.

Materials and Methods

Aortic Valve Collection

The study protocol was approved by the institutional ethics committee and written informed consent was obtained from all patients. Preoperatively, the functional state of the aortic valve was determined by echocardiography and/or angiocardiography according to guidelines of the American Heart Association and the American College of Cardiology(22). All of the patients examined for altered miRNA expression had type 1 bicuspid aortic valve leaflets(23), with fusion of the right and left coronary leaflets, that was confirmed at the time of aortic valve replacement. The fused and unfused leaflets were inspected at the time of surgery for morphology. The leaflets were collected separately frozen in liquid nitrogen. Patients’ demographics are shown in table 1.

Table 1.

Patient demographics and clinical characteristics.

Bicuspid aortic valve (BAV)	N = 9
Demographic
Age (years)	44.9 ± 13.8
Gender (male/female)	9 m
Body mass index	26.4±3.0
Reason of aortic valve replacement
Aortic valve insufficiency	5
Aortic valve stenosis	4
Systemic disease
Hypertension	0
Chronic obstructive pulmonary disease	0
Aneurysm of ascending aorta (Ø mm)	41.3±7.3

RNA Isolation and miRNA Array

RNA was isolated from the leaflets using Trizol (Invitrogen, Carlsbad, CA) according to the manufacturer’s protocol. The RNA from the fused leaflet was labeled with Cy5 while the unfused leaflet RNA, which served as a control, was labeled with Cy3 using the PIQOR MiRXplore protocol (Miltenyi Biotech, Bergisch Gladbach, Germany). The samples were hybridized to PIQOR miRXplore Microarrays (Miltenyi Biotech) according to the manufacturer’s recommendations.

Statistical Analysis of miRNA Array Data

Normalization

Imagene files were processed using GeneSpring 7.3.1 from Agilent. The files were normalized by dividing the Cy5 channel (fused leaflet) with its matched Cy3 channel (unfused leaflet) creating a ratio. The next step in the normalization was to scale the ratios from each matched set to the 50^th intensity from the matched set. This calculation was done for each matched set, this scales each matched set to the most representative intensity. In total there are 9 matched sets so this procedure was done 9 times. This step was performed to ensure that systematic error was removed from the data.

Assessment of group homogeneity

Principal Component Analysis and hierarchical clustering (Spearman correlation, average linkage) were performed using the normalized ratio data. As a common result, three samples (microarrays 6170041, 3320029, 6620035) were reported as outliers from the respective sample group. (Table 2). These samples were excluded from further analysis.

Table 2.

Sample quality control. Principle Component and Hierarchical Clustering analyses excluded three microarrays from the statistical analysis since these arrays were outliers.

Microarray	Reason of Aortic valve replacement	Sample QC
6170041	Aortic valve insufficiency	Bad
6620037	Aortic valve insufficiency	Good
6620029	Aortic valve insufficiency	Bad
6620030	Aortic valve insufficiency	Good
6620031	Aortic valve insufficiency	Good
6620032	Aortic valve stenosis	Good
6620033	Aortic valve stenosis	Good
6620034	Aortic valve stenosis	Good
6620035	Aortic valve stenosis	Bad

Probe Quality Control

The next step was to filter the data based on the signal error within the data, this removed miRNAs that have a large amount of signal noise from statistical analysis. This filter was applied to all the data and miRNA only passed if they achieved the specified signal/noise ratio in all samples. This ensured that the data used in the proceeding statistical analysis is sound. The total size of the miRNA array was 1421, which is a combination of human, murine, viral and other species miRNAs. 66 miRNA remained after applying this filter.

Statistical Analysis

The remaining 6 samples equated to 3 replicates in the stenosis and insufficient groups. The data was analyzed for significantly differentially expressed miRNA between the groups. The miRNA used in this analysis were the 66 miRNA remaining after the signal/noise ratio filter. The test applied to this data was a Student t-test with a p-value cutoff of 0.05. No multiple testing was applied.

QRT-PCR

miRNA qRT-PCR was performed on the fused leaflet valve samples used in the miRNA array experiment (AI group n=5 and AS group n=4) to validate the miRNA microarray results. The TaqMan MicroRNA Reverse Transcription Kit (Applied Biosystems, Foster City, CA) with miRNA specific primers was used according to manufacture’s protocol. Δ ΔΔCT values were calculated; U6 was used as an endogenous control.

Transfection of human Aortic Valve Interstitial Cells (AVICs)

Human aortic valve leaflets were harvested from recipient hearts at the time of cardiac transplant with IRB approval. Human AVICs were cultured according to standard protocols(24, 25). Fully confluent AVICs were transfected with either 100pM miRNA mimic (Dharmacon, Lafayette, CO) or Block-it using Lipofectamine 2000 (Invitrogen) according to manufacture’s protocols(8). As a transfection control and to determine transfection efficiency, the cells were transfected with non-functioning siRNA (Block-It, Invitrogen)(26). After 96 hours, RNA was isolated using Trizol. cDNA was prepared using Superscript III (Invitrogen). qRT-PCR was performed using TaqMan primers (Applied Biosystems) for a panel of calcification-related genes. ΔΔCT values were calculated; GADPH served as the control.

Statistical Analysis

The statistical significance of differences between groups was determined with the unpaired t test. P≤0.05 was considered significant.

Results

miR-26a, miR-30b, and miR-195 were reduced in AS samples

miRNA Microarray Results

To identify possible miRNAs whose expression levels were altered in bicuspid aortic valve (BAV), miRNA array analysis was performed. Statistical analysis of the array data demonstrated that seven miRNAs (miR-16, miR-26a, miR-27a, miR-30b, miR-130, miR-195, and miR-497) were altered in BAV. (Table 3). All seven miRNAs were downregulated in AS group as compared to the AI group, of which three miRNAs from the same family (miR-16, miR-195, and miR-497) are found to be significant for their difference between AS and AI phenotypes.

Table 3.

miRNA microarray results. Seven miRNAs were found to have a statistically significant decrease in AS as compared to AI valve samples.

Name	Fold change	Regulated	p-value
hsa-miR-130A	−1.48148	down in AVS	0.00259
hsa-miR-30B	−1.83486	down in AVS	0.00748
hsa-miR-497	−1.58983	down in AVS	0.0119
hsa-miR-195	−1.9802	down in AVS	0.0158
hsa-miR-26A	−1.62075	down in AVS	0.0265
hsa-miR-16	−1.9685	down in AVS	0.0353
hsa-miR-27A	−1.88679	down in AVS	0.0406

miRNA qRT-PCR

To validate the miRNA arrays results, miRNA qRT-PCR was performed on the fused leaflets from both groups. Since the fused leaflet tends to be the predominate foci of disease (data not shown), the unfused leaflet was used as an internal control. miR-26a and miR-195 levels were reduced by 65% and 59% respectively with p<0.05 in the stenotic leaflets as compared to the insufficient leaflets by qRT-PCR. miR-30b was reduced by 62% in the stenotic samples by qRT-PCR with a p-value that approached significance at p<0.06. (Figure 1) The remaining miRNAs were not altered in a statistically significant manner.

Figure 1.

miR-26a and miR-195 were expressed at lower levels in AS valves. QRT-PCR was performed on the AS (n=4) and AI (n=5) valve samples to validate the miRNA microarray results. miR-26a was reduced by 65% in the AS valves while miR-195 was decreased by 59% (*p<0.05 for both). Of note, miR-30b was reduced by 62% with p<0.06.

miR-26a, miR-30b, and miR-195 modulate calcification related genes

To examine if miR-26a, miR-30b, and miR-195 have roles in modulating calcification-related genes in AVICs, mimics for these miRNAs were transfected into human AVICs and RNA was harvested for qRT-PCR after four days. A panel of calcification and valve related genes were examined via qRT-PCR.

mir-26a repressed several of the calcification-related genes in vitro (Fig 2). ALKL mRNA levels were reduced by 38% (p<0.01), BMP2 mRNA was reduced by 36% (p<0.04), SMAD1 by 26% (p<0.01). MiR-26a increased the mRNA level of genes that may have roles in inhibiting calcification including JAG2 (31% p<0.01) and SMAD7 (15% p<0.01). Of note, there was an increase in two genes that are considered pro-calcific, RUNX2 was increased by 16% p<0.01 and SMAD5 increased by 21% p<0.01.

Figure 2.

MiR-26a predominately repressed pro-calcification genes (black bars) while increasing expression of anti-calcification genes (grey bars) in human AVICs. Specifically, miR-26a treatment resulted in decreases expression of ALPL (38%), BMP2 (36%), and SMAD1 (26%). At the same time, JAG2 and SMAD7 were increased by 31% and 15% respectively in cells treated with miR-26a. However, the increases in the pro-calcification genes RUNX2 (16%) and SMAD5 (21%) goes against the trend. (n=3; *p<0.05; **p<0.01).

miR-30b also appeared to down regulate calcification-related gene pathways (fig 3). miR-30b repressed SMAD1 by 18% p<0.02 and SMAD3 by 12% p<0.05. BMP2 was repressed by 39% with a p-value that approached significance at p<0.07. JAG2 was increased by 14% p<0.01 and SMAD7 by 40% p<0.02. In contrast, NOTCH1 expression was decreased by 19% p<0.01.

Figure 3.

MiR-30b predominately repressed pro-calcification genes (black bars) while increasing expression of anti-calcification genes (grey bars) in human AVICs. AVICs treated with miR-30b mimic had decreased SMAD1 (18%) and SMAD3 (12%) mRNA levels. JAG2 and SMAD7, two anti-calcification genes, had 14% and 40% increase in mRNA expression respectively. The 19% decreased in NOTCH1 expression goes against the trend. (n=3; *p<0.05; **p<0.01). Of note, BMP2 was reduced by 68% with a p<0.07.

miR-195 activated calcification related genes by BMP2 68% p<0.01, RUNX2 11% p<0.01, SMAD1 9% p<0.04, SMAD3 4% p<0.02, SMAD5 17% p<0.01 (fig 4). There was an increase in JAG2 and SMAD7 expression by 13% p<0.01 and 26% respectively (p<0.01) which are two genes that may repress calcification.

Figure 4.

MiR-195 predominately activates pro-calcification gene expression (black bars) while repressing anti-calcification genes (grey bars). BMP2 (68%), RUNX2 (11%), SMAD1 (9%), SMAD3 (4%), and SMAD5 (17%) were increased in miR-195 treated AVICs. The increases in JAG2 (13%) and SMAD7 (26%) go against the trend. (n=3; *p<0.05; **p<0.01).

Discussion

To our knowledge, this is the first report of altered miRNA levels in aortic valve disease and their possible role in aortic valve calcification. Given that BAV, present in 1–2% of the population, is the most the most common risk factor for aortic valve calcification, we focused our efforts on comparing miRNA levels in BAV with AS or AI. We compared the miRNA levels in the fused leaflets using miRNA arrays and qPCR. The miRNA arrays identified seven miRNAs with differentially expression in the fused leaflets of BAV compared to the unfused leaflet. It is interesting that three miRNAs (miR-16, miR-195, and miR-497) from the same family are found to be significant for their difference between AS and AI phenotypes. miRNA qPCR demonstrated that miR-26a and miR-195 were decreased in AS patients as compared to AI patients in a statistically significant manner. miR-30b was downregulated with a p-value of 0.06, which approached statistically significance. We acknowledge the likelihood that additional miRNAs may be altered but were not detected by the microarray since calcification is not a uniform process throughout the leaflet and our experimental design examined the expression patterns in the entire leaflet. Furthermore, miRNAs may have a different role in the early pathogenesis while the valve samples were obtained in the late stages of the disease process.

In order to examine the role of miR-26a, miR-195, and miR-30b on calcification related genes, we transfected cultured human AVICs with miRNA mimics and performed qPCR for calcification related genes. Interestingly, we found that miR-26a repressed pro-calcification genes such as BMP2, RUNX2, SMAD1, and SMAD5, at the same time there was an increasing of anti-calcification genes such as SMAD7, and JAG2. JAG2, a ligand for the Notch signaling pathway, may repress aortic valve calcification via NOTCH1, which has been shown to repress aortic valve calcification(5, 8). SMAD7 has inhibitory roles to BMP signaling and inhibits TGF-beta signaling (27, 28). miR-30b has a similar pattern of repression calcification related genes and activating SMAD7 and JAG2. In contrast, MiR-195 appears to activate several pro-calcification related genes. Our findings indicate that these miRNAs (miR-26a, miR-195, and miR-30b) in vitro modulated the mRNA levels of calcification related genes. It appears that miRNAs may have a key role in the evolving understanding of the genetic pathways involved in aortic valve calcification and merit study as a possible treatment modality.

In our future directions, we have embarked upon the examination of these miRNAs in a large cohort of BAVs with different clinical phenotypes. One aspect of these studies will be to characterize the role of the miRNAs in different cell types within the aortic valve.

Study limitations

A limitation of the study is the significant age difference of the patients in the AI group as compared to the AS group. This difference is typical for the patient populations because AS typically requires surgery later in life. The age difference raises the possibility that age is a contributing factor to the alterations of the miRNA. The decreases in mir-26a and -30b could contribute to the calcification of the aortic valve since they modulate genes related to calcification. Further study will be required to understand how miRNAs are being regulated in the aortic valve leaflets including role of aging on miRNA expression.

Another limitation is that it was not possible to identify specific calcification-related genes that are being targeted by the miRNA studied in this report. We did attempt to perform in silico analysis using publically available target prediction programs. However, none of the genes that had altered expression in response to treatment with miRNA mimics had predicted binding sites for the relevant miRNA. As the understanding of relationships between miRNAs and their target genes improves, it likely will be easier to identify the specific calcification-related genes targeted by the miRNAs.

We found that miR-26a, miR-30b, and miR-195 were reduced in the aortic valve leaflets of patients undergoing valve replacement of the aortic valve due to calcific stenosis as compared to patients undergoing valve replacement due to insufficiency. In vitro, increased levels of miR-26a and miR-30b repress the calcification pathways by modulating calcification-related mRNAs. Conversely, miR-195 tended to activate the calcification pathways.