miR-214 is Stretch-Sensitive in Aortic Valve and Inhibits Aortic Valve Calcification

Abstract

miR-214 has been recently found to be significantly downregulated in calcified human aortic valves (AVs). ER stress, especially the ATF4-mediated pathway, has also been shown to be significantly upregulated in calcific AV disease. Since elevated cyclic stretch is one of the major mechanical stimuli for AV calcification and ATF4 is a validated target of miR-214, we investigated the effect of cyclic stretch on miR-214 expression as well as those of ATF4 and two downstream genes (CHOP and BCL2L1). Porcine aortic valve (PAV) leaflets were cyclically stretched at 15% for 48 hours in regular medium and for one week in osteogenic medium to simulate the early remodeling and late calcification stages of stretch-induced AV disease, respectively. For both stages, 10% cyclic stretch served as the physiological counterpart. RT-qPCR revealed that miR-214 expression was significantly downregulated during the late calcification stage, whereas the mRNA expression of ATF4 and BCL2L1 was upregulated and downregulated, respectively, during both early remodeling and late calcification stages. When PAV leaflets were statically transfected with miR-214 mimic in osteogenic medium for 2 weeks, calcification was significantly reduced compared to the control mimic case. This implies that miR-214 may have a protective role in stretchinduced calcific AV disease.

Introduction

Aortic stenosis (AS) is one of the most prevalent valvular diseases in the world, affecting more than 12% of the elderly population⁴¹. The primary reasons of AS are AV fibrosis¹⁰ and calcification⁴⁵. Currently, there are no therapeutic drugs available to treat this disease²⁹ and the only treatment options are surgical or transcatheter AV replacement⁵⁵. However, these procedures entail several complications, such as bleeding²¹, thrombosis⁹, etc.

It has been well established that mechanical forces (such as stretch and shear) play a key role in AV pathogenesis^{1, 4, 22}. For example, both elevated mechanical stretch (15%) and low, oscillatory shear stress (i. e. disturbed flow; ± 5 dynes/cm²) have been shown to promote AV calcification^{3, 48}. However, high mechanical stretch induces 5-fold higher AV calcification compared to low, oscillatory shear stress⁴⁶, implying that elevated mechanical stretch has a more pronounced effect on AV calcification than disturbed flow. Interestingly, 15% stretch was also found to correspond to a hypertensive condition in AV⁷³, with hypertension being one of the major risk factors for calcific aortic valve disease (CAVD)⁴⁴. Hence, 15% stretch is considered to represent the pathological level of mechanical stretch, whereas 10% stretch is representative of the physiological level^{58, 73}.

In the pursuit of effective therapeutics to treat CAVD, manipulation of microRNA (miRNA) expression levels by miRNA mimic or inhibitor has recently emerged as an exciting candidate⁶². miRNAs are small (~ 22 nucleotides in length) non-coding RNAs that modulate gene expression by silencing their targets²⁴. Due to their role in modulating gene expression, miRNAs have been widely considered as a potential therapeutic option for different diseases, such as cancer⁵¹. Interestingly, calcified human AVs have distinct miRNA expression profiles compared to healthy AVs^{13, 56, 65}. However, considering the hundreds of potential miRNAs for therapeutic application in CAVD, identification of the most effective ones is a critical first step. To this end, mechanosensitive miRNAs are promising, given that perturbations to mechanical stimuli promote AV pathogenesis^{1, 4, 22}. In recent years, several studies have focused on the identification of shear-sensitive miRNAs and their functional relations to CAVD^{25, 28, 47}. However, stretch-sensitive miRNAs relevant to CAVD remain to be identified or functionally described⁴², even though elevated mechanical stretch induces increased AV calcification compared to disturbed flow⁴⁶. In a recent study⁴², it was found that 14% stretch downregulated the expression of miR-148a-3p in human aortic valve interstitial cells (HAVICs) compared to static condition. This downregulation resulted in increased expression of IKBKB (a target of miR-148a-3p) and activation of downstream inflammatory signaling. However, to the best of our knowledge, no studies have investigated the effects of physiological and pathological stretch on miRNA expression in AV to date.

Among the plethora of discovered miRNAs so far, miRNA 214 (or miR-214) is one of the most studied miRNAs, especially in cancer^{59, 70}. miR-214 (along with miR-199a) is encoded by the opposite strand of DNM3 (DNM3os)^{15, 66} and well known to inhibit osteogenesis^{34, 53, 69}. In line with the anti-osteogenic effect, miR-214 expression was found to be downregulated in calcified human AVs compared to healthy AVs^{13, 56, 65}.Recently, miR-214 was found to be shear-sensitive in AV⁴⁷. However, the functional role of miR-214 in stretch-induced AV calcification is not known. To this end, Activating Transcription Factor 4 or ATF4 may play a key role in mediating the effect of miR-214 in AV calcification. ATF4, which is one of the major targets of miR-214⁶⁴, is a component of the PERK-eIF2α signaling pathway, which is one of the three endoplasmic reticulum (ER) stress pathways²⁷. These pathways significantly regulate cellular processes like inflammation¹⁹, apoptosis⁴³, etc. Interestingly, ER stress has been shown to promote AV calcification, in which the PERK-eIF2α-ATF4 signaling pathway plays a major role^{8, 63}. Mechanical stretch has been found to promote inflammation and apoptosis in smooth muscle cells via induction of ER stress³⁰. Under ER stress, both transcription and translation of ATF4 are increased, resulting in cell death situations via upregulation of C/EBP-Homologous Protein or CHOP expression¹⁶.

Interestingly, ATF4 has been shown to promote mineralization in vascular smooth muscle cells³⁶, osteoblasts⁶⁸, mesenchymal stem cells⁷⁴, etc. On the other hand, CHOP has been shown to have a potentially repressive effect on the expression of pro-survival gene Bcl-2-Like Protein 1 or BCL2L1 under apoptotic conditions²⁰. Hence, considering the anti-osteogenic effect of miR-214 and pro-osteogenic effect of ATF4, this study investigates the role of miR-214 in stretch-induced AV calcification, especially focusing on the ATF4/CHOP/BCL2L1 pathway.

Materials and Methods

Experimental Simulation of Different Stages of Stretch-Induced Calcific Aortic Valve Disease (CAVD)

The cyclic stretch bioreactor [Supplemental Figure 1] used in this study was previously validated in our laboratory^{2, 3}. An ex vivo experimental approach using porcine aortic valves (PAVs)^{2, 3, 47} was adopted as it incorporates the contributions from both valvular endothelial cells (VECs) and valvular interstitial cells (VICs). Cyclic stretch experiments were carried out to simulate two different stages of stretch-induced calcific aortic valve disease (CAVD), namely, early remodeling and late calcification. Stretch-induced early remodeling stage was experimentally simulated by cyclic stretching at 15% for 48 hours in regular medium^{2, 3}, whereas 10% stretch (with all other conditions remaining the same) represented the physiological counterpart. It was previously found that cyclic stretching at 15% for 48 hours in regular medium induced significant changes in the expression levels of key extracellular matrix (ECM) remodeling proteins (such as MMP2, Cathepsin S and Cathepsin K) in PAV leaflets compared to 10% stretch².

On the other hand, stretch-induced late calcification stage was experimentally simulated by cyclic stretching at 15% for one week in osteogenic medium (regular medium supplemented with 3.8 mM NaH₂PO₄.H₂O, 1 mM β-Glycerophosphate, 10 µM Dexamethasone and 1 ng/mL TGF-β1)³, whereas 10% stretch (with all other conditions remaining the same) again represented the physiological counterpart. When PAV leaflets were cyclically stretched at 15% for one week in osteogenic medium, it was found that the degree of calcification (as quantified by Calcium Arsenazo assay) was at least two-fold higher than the one at 10% stretch [Supplemental Figure 2 (a)]. Similar qualitative results were obtained from Alizarin Red and Von Kossa staining [Supplemental Figure 2 (b)]. Balachandran et al.³ observed similar increase in PAV calcification under 15% stretch with osteogenic medium over a culture duration of two weeks.

miR-214 Overexpression in PAV Leaflets

PAV leaflets were transfected with either negative control mimic (QIAGEN) or miR-214 mimic (QIAGEN) in static culture for 48 hours and under 15% stretch for 72 hours. It should be noted here that for the 1-week long 15% stretch and 2-week long static miR214 overexpression experiments, the mimics were added only at Day 0 and (Day 0 and Day 7), respectively. In all transfection experiments, osteogenic medium was used to promote a calcification environment.

RNA Isolation and Real-time Quantitative Polymerase Chain Reaction (RT-qPCR)

Total RNA was isolated using a RNA isolation kit (ZYMO RESEARCH). cDNA was synthesized using a Reverse Transcription kit (QIAGEN). RT-qPCR was carried out in triplicates using a 96-well Real-Time PCR system. U6 and 18S were used as the housekeeping genes for miRNA and mRNA RT-qPCR, respectively. The primer assays for U6 and miR-214 were directly obtained from QIAGEN, whereas the forward and reverse primers for ATF4, CHOP and BCL2L1 mRNAs were custom designed [Table 1]. The relative expression levels of miR-214 and mRNAs were calculated using the ΔCT method⁵².

Table 1.

Primer sequences for 18S, ATF4, CHOP and BCL2L1

mRNA	Forward Primer (5’→3’)	Reverse Primer (5’→3’)
18S	AGGAATTGACGGAAGGGCACCA	GTGCAGCCCCGGACATCTAAG
ATF4	AGTCCTTTTCTGCGAGTGGG	GAGAAGCGCCATGGCCTAAG
CHOP	CCCTGGAAATGAGGAGGAGTC	TGACTGGAATCAGGCGAGTG
BCL2L1	TGACCACCTAGAGCCTTGGA	CGTCAGGAACCATCGGTTGA

Assessment of PAV Calcification

PAV calcification was assessed by Calcium Arsenazo assay, Alizarin Red and Von Kossa stains³.

Statistical Analysis

Data are presented as mean ± standard error of mean. During statistical analysis, the first step was to discard outliers from each independent group of data. Outliers were detected based on the interquartile range. When comparing two independent data sets, independent samples t-test was used if both data sets were normally distributed. In case where either one or both data sets were not normally distributed, non-parametric Mann-Whitney U test was used to compare these two groups. A p-value of 0.05 or less was considered to indicate statistical significance. All statistical tests were carried out in the IBM SPSS Statistics software. In addition to assessing statistical significance, effect size (Hedges’ g-value) was calculated for each case of comparison²⁶. Small, medium and large effect sizes were represented by Hedges’ g-values of 0.2, 0.5 and 0.8, respectively.

Results

miR-214 Expression During Stretch-Induced Early Remodeling of PAV Leaflets

When PAV leaflets were cyclically stretched at 15% for 48 hours in regular medium, there was no significant change in miR-214 expression (9.9% increase with Hedges’ g- value of 0.14) compared to 10% stretch [Figure 1 (a)].

Figure 1:

(a) miR-214 [n = 8] and (c) ATF4, CHOP and BCL2L1 mRNA [n = 7 – 8] expression during the early remodeling stage, and (b) miR-214 [n = 9 – 10] and (d) ATF4, CHOP and BCL2L1 mRNA [n = 7 – 8] expression during the late calcification stage of stretch-induced PAV disease.

ATF4, CHOP and BCL2L1 mRNA Expression During Stretch-Induced Early Remodeling of PAV Leaflets

When PAV leaflets were cyclically stretched at 15% for 48 hours in regular medium,ATF4 mRNA expression was significantly upregulated (37.6% increase with Hedges’ g-value of 1.19) and BCL2L1 mRNA expression was significantly downregulated (47.3% decrease with Hedges’ g-value of 2.25) compared to the ones at 10% stretch [Figure 1 (c)]. However, the mRNA expression of CHOP remained the same under both stretch conditions (0.0% change with Hedges’ g-value of 0.000356) [Figure 1 (c)].

miR-214 Expression During Stretch-Induced Late Calcification of PAV Leaflets

When PAV leaflets were cyclically stretched at 15% for 1 week in osteogenic medium, miR-214 expression was significantly downregulated (71.6% decrease with Hedges’ g- value of 1.52) compared to 10% stretch [Figure 1 (b)]. Similar downregulation of miR-214 expression was also observed in calcified human AVs compared to healthy control^{13, 56, 65}.

ATF4, CHOP and BCL2L1 mRNA Expression During Stretch-Induced Late Calcification of PAV Leaflets

When PAV leaflets were cyclically stretched at 15% for 1 week in osteogenic medium, there was considerable upregulation of ATF4 mRNA expression (114.5% increase with Hedges’ g-value of 0.91) and downregulation of BCL2L1 mRNA expression (53.1% decrease with Hedges’ g-value of 0.91) compared to the ones at 10% stretch [Figure 1 (d)]. These trends were similar to the ones observed for the early remodeling stage of stretch-induced aortic valve (AV) disease [Figure 1 (c)]. Additionally, there was slight increase in CHOP mRNA expression (27.1% increase with Hedges’ g-value of 0.28) at 15% stretch compared to 10% [Figure 1 (d)]. Interestingly, calcified human AVs have been shown to express increased levels of ATF4 and CHOP compared to healthy control^{8, 63}.

Effect of miR-214 Overexpression on ATF4, CHOP and BCL2L1 mRNA Expression in PAV Leaflets

When PAV leaflets were transfected with miR-214 mimic (50 nM) in static culture with osteogenic medium for 48 hours, miR-214 expression was significantly upregulated (15868121.8% increase with Hedges’ g-value of 0.82) compared to the case with control mimic (50 nM) [Figure 2 (a)]. Consequently, miR-214 overexpression resulted in significant downregulation of ATF4 and BCL2L1 mRNA expression (47.4% and 41.9% decrease with Hedges’ g-values of 1.19 and 1.12, respectively) compared to the control mimic case [Figure 2 (b)]. Additionally, there was considerable downregulation of CHOP mRNA expression (42.6% decrease with Hedges’ g-value of 0.83) with miR-214 mimic compared to the control mimic case [Figure 2 (b)].

Figure 2:

(a) miR-214 overexpression [n = 4] and (b) effect of miR-214 overexpression on ATF4, CHOP and BCL2L1 mRNA expression [n = 7 – 8] after 48-hour static PAV culture in osteogenic medium.

Effect of miR-214 Overexpression on PAV Calcification

When PAV leaflets were transfected with miR-214 mimic (50 nM) in static culture with osteogenic medium for 2 weeks and under 15% stretch with osteogenic medium for 1 week, miR-214 expression was significantly upregulated (19306.3% increase with Hedges’ g-value of 1.47 and 771.0% increase with Hedges’ g-value of 3.45, respectively) compared to the case with control mimic (50 nM) [Figures 3 (a) and 6 (a)]. Consequently, miR-214 overexpression resulted in significant reduction of PAV calcification (37.4% decrease with Hedges’ g-value of 1.14 in static culture and 20.6% decrease with Hedges’ g-value of 0.98 under 15% stretch) compared to the control mimic case, as determined by Calcium Arsenazo assay [Figures 3 (b) and 6 (b)]. Alizarin Red [Figure 4 (a) and (b)] and Von Kossa [Figure 5 (a) and (b)] staining showed similar reduction in PAV calcification (59.9% and 28.2% decrease with Hedges’ g- values of 1.39 and 1.81, respectively) upon transfection with miR-214 mimic.

Figure 3:

(a) miR-214 overexpression [n = 7 – 8] and (b) effect of miR-214 overexpression on PAV calcification [n = 7 – 8] as determined by Calcium Arsenazo assay after 2-week static culture in osteogenic medium.

Figure 6:

(a) miR-214 overexpression [n = 6 – 7] and (b) effect of miR-214 overexpression on PAV calcification [n = 8] as determined by Calcium Arsenazo assay after 1-week 15% stretch in osteogenic medium.

Figure 4:

(a) Representative Alizarin Red staining images (top: lower magnification, bottom: higher magnification) and (b) quantification of whole Alizarin Red stained PAV sections after 2-week static culture in osteogenic medium [n = 6; arrows indicate calcification].

Figure 5:

(a) Representative Von Kossa staining images (top: lower magnification, bottom: higher magnification) and (b) quantification of whole Von Kossa stained PAV sections after 2-week static culture in osteogenic medium [n = 4 – 6; arrows indicate calcification].

Discussion

In this study, we investigated the role of miR-214 in stretch-induced calcific aortic valve disease (CAVD), especially focusing on the ATF4/CHOP/BCL2L1 pathway. We found that elevated cyclic stretch (15%) downregulates miR-214 expression in PAVs during the late calcification stage of stretch-induced aortic valve (AV) disease. We also found that ATF4 and BCL2L1 expression was upregulated and downregulated, respectively, by 15% stretch during both early remodeling and late calcification stages of stretch- induced AV disease. When miR-214 was overexpressed in PAVs under static culture with osteogenic medium for 48 hours, there was significant reduction in ATF4 and BCL2L1 expression. Additionally, miR-214 overexpression resulted in significant reduction of PAV calcification after 2 weeks. Similar regulation of ATF4 expression by miR-214 was reported for chondrogenesis⁴⁹ and osteogenesis^{34, 64, 72}. In addition to that, inhibition of ATF4 expression by siRNA was found to downregulate the expression of osteogenic markers in valvular interstitial cells (VICs)⁸. However, to the best of our knowledge, the stretch-sensitive nature of this miR-214/ATF4 interaction has not been reported for AV, especially in relation to AV calcification.

It was intriguing to observe no significant change in miR-214 expression during the early remodeling stage of stretch-induced AV disease. This can be explained by the effect of TWIST1 on miR-214 expression. TWIST1, a basic helix-loop-helix (bHLH) transcription factor, is a positive regulator of miR-214 expression³³. TWIST1 expression was previously shown to be induced by TIMP1¹⁴. Interestingly, Balachandran et al.² showed that there was no significant change in TIMP1 expression between 10% and 15% stretch after 48 hours of cyclic stretching in regular medium. Consequently, similar TIMP1 expression under both stretch levels implies that there may be no significant change in TWIST1 expression, explaining the observed non-significant difference in miR-214 expression during the early remodeling stage of stretch-induced AV disease. On the other hand, TWIST1 expression was found to be significantly downregulated in calcified human AVs compared to healthy AVs⁷⁵. In addition to that, TWIST1 has an inhibiting effect on osteogenesis^{6, 38} and cyclic stretching at 15% for a longer duration (>> 48 hours) may downregulate its expression via Scleraxis (SCX), another bHLH transcription factor^{5, 39, 50}. This may explain the significantly downregulated miR-214 expression during the late calcification stage of stretch-induced AV disease.

It has been well established that ATF4 is a pro-osteogenic transcription factor^{8, 36}. Valentine et al.⁶⁰ showed that exposure of mouse type II alveolar epithelial cells (ATII) to 15% stretch resulted in significant increase of ATF4 expression compared to static control. Yang et al.⁷¹ reported that 10% stretch significantly upregulated ATF4 expression in human periodontal ligament cells (hPDLCs) compared to static control. Since elevated cyclic stretch (15%) induces higher AV calcification compared to the physiological level (10%)³, the observed upregulation of ATF4 expression (15% vs. 10% stretch) in PAV leaflets makes sense from both biomechanical and functional point of view. ATF4 can upregulate the expression of MMP2 and MMP9 in cancer cells^{17, 76}. Similar upregulation of MMP2 and MMP9 expression was observed in PAV leaflets under pathological cyclic stretch (15%) compared to the physiological level (10%)².

It was surprising to observe that CHOP expression remained effectively the same during the early remodeling and late calcification stages of stretch-induced AV disease. Jia et al.³⁰ reported that 18% stretch significantly upregulated CHOP expression in mouse aortic smooth muscle cells (SMCs) compared to static control. However, Cheng et al.¹² showed that 20% stretch induced a brief (during the culture period from 12 to 18 hours) increase in CHOP expression of vascular smooth muscle cells (VSMCs) compared to 10% stretch, which was attenuated afterwards (culture period > 18 hours). Therefore, it is possible that pathological cyclic stretch (15%) induces similar transient increase of CHOP expression in PAV leaflets compared to the physiological level (10%), which is later attenuated by some adaptive response of the tissue itself. Tribbles Homolog 3 (TRB3) is one of the transcriptional targets of ATF4 that is upregulated when there is an increase in ATF4 expression⁴⁰. It should be noted here that CHOP is another transcriptional target of ATF4¹⁶. Interestingly, increased TRB3 expression works in a negative feedback loop to inhibit the transcriptional induction of CHOP³¹. Therefore, it can be hypothesized that the insensitivity of CHOP expression to pathological cyclic stretch (15%) during the early remodeling and late calcification stages of stretch- induced AV disease may be caused by TRB3-mediated inhibition of its transiently increased expression at 15% stretch compared to 10%. On the other hand, the observed non-significant effect of miR-214 overexpression on CHOP mRNA expression in PAV leaflets makes sense since miR-214 has not been predicted to target CHOP at the 3’UTR region¹⁸.

BCL2L1 is mainly a pro-survival gene due to its predominant anti-apoptotic isoform, BCL-XL⁵⁴. Contrary to our findings, 15% stretch was found to upregulate BCL-XL expression in aortic smooth muscle cells (SMCs) compared to both static condition and 5% stretch³⁷. One of the positive regulators of BCL2L1 is B-Cell Lymphoma/Leukemia 11B (BCL11B)^{23, 32}. Interestingly, BCL11B expression was found to be downregulated with increased arterial stiffness^{61, 67}. Therefore, it can be hypothesized that decrease in BCL2L1 expression during the early remodeling and late calcification stages of stretch- induced AV disease may be mediated by downregulation of BCL11B expression. Surprisingly, both BCL11B and BCL2L1 expression was found to be upregulated in calcified human AVs compared to healthy control^{7, 35}. This indicates that disturbed flow may have a more pronounced effect in upregulating BCL2L1 expression compared to the downregulation induced by elevated mechanical stretch⁵⁷. In addition to that, miR-214 overexpression resulted in significant reduction of BCL2L1 expression in PAV leaflets, as observed in our study. This confirms the already known anti-proliferative effect of miR-214, especially in cancer cells¹¹.