Abnormal mechanical stress on bicuspid aortic valve induces valvular calcification and inhibits Notch1/NICD/Runx2 signal

Patient recruitment and sample collection

Two bicuspid aortic valve patients and two tricuspid aortic valve volunteers were enrolled in the construction of finite element model. BAV was diagnosed based on the cardiac ultrasonic examination (Michelena et al., 2014). The clinical characteristics of all participants were similar, including the height, weight, BMI, and the valves were functioning properly without regurgitation or stenosis based on cardiac ultrasonography (Table 1).

The BAV samples were collected from 22 BAV patients who underwent aortic valve replacement and the TAV samples were collected from 17 end-stage dilated cardiomyopathy patients who received donor heart from June 2016 to December 2016 at the Department of Cardiovascular Surgery of Wuhan Union Hospital. The BAV was diagnosed by direct observation during surgery to exclude the pseudobivalvular deformity, combined with echo report. The clinical demographic data were shown in Table 2.

Consent details

In accordance with the Declaration of Helsinki, the study was approved by review boards of Union Hospital and Tongji Medical College (IORG0003571). Written informed consents were obtained from all participants at their inclusion in the study.

Antibodies and primers

Antibodies for immunoblot in this study included the following: anti-Notch1 (ab280898; Abcam, Cambridge, UK), anti-Runx2 (ab114133; Abcam), anti-NICD (ab52301; Abcam), anti-GAPDH (ab8245; Abcam). Primers for qPCR in this study included the following: Notch1 forward 5′-GCGACAACGCCTACCTCTG-3′, Notch1 reverse 5′-AAGCCATTGATGCCGTCC-3′, Runx2 forward 5′-GGACGAGGCAAGAGTTTCAC-3′, Runx2 reverse 5′-GAGGCGGTCAGAGAACAAAC- 3′, NICD forward 5′-CTATAGACGACATTGACGAGTGTGAC-3′, NICD reverse 5′-TGTAGAATTCAGAGGACAGTTGCACTTG-3′, GAPDH forward 5′-AATCCCATCACCATCTTCCAG- 3′, GAPDH reverse 5′-GAGCCCCAGCCTTCTCCAT-3′.

Establishment of the aortic valve geometric model

The aortic valve geometric model was established by a previously described method with modification (Jernigan et al., 2007; Nappi et al., 2020). Briefly, the ascending aorta CTA images in the DICOM format were imported into Mimics 10.01 software (Materialise, Leuven, Belgium). Then, based on the tissue density of cardiac valve, set the range of threshold between 329 to 560 Hounsfield units, and 3D reconstruction was performed using ‘Calculate 3D’ to obtain the geometric model of the aortic valve and ascending aortic arch. Next, the 3D solid model of aortic leaflets was built through meshes using the Remesh module in Magics.

Parameters of the aortic leaflet model

Fibrous reinforcement material properties were utilized to model aortic valve in FEM module of the Mimics software based on the grayscale of CTA images. The ascending aorta was set as a large vessel with a uniform thickness of 2.3 mm and described by a linear hexahedral element. The aortic valve was set as thin tissue and described by a 4-node shell unit (Kim et al., 2007). The remaining tissues such as myocardium were set as isotropic linear elastic materials and the material properties were set as following: Young modulus was 2MPa, Poisson ratio was 0.3, tissue density was 1.1 g/cm³. The response of leaflet to mechanical stimuli was considered to be highly elastic, isotropic and incompressible.

Finite element analysis

The files of the aortic valve solid model were imported into ANSYS software (Materialise, Leuven, Belgium) for finite element analysis. The left ventricular end-diastolic pressure was set as 80mmHg, and the friction coefficient of aortic valve was set as 0.05. The finite element parameters of aortic valve were set according to isotropic, uniform continuous linear elastic material. The cardiac cycle was set as 0.8 s and we analyzed the equivalent stress on aortic valve and the deformation of leaflet during one cardiac cycle.

Alizarin red stain

The aortic valve excised at the time of surgery was placed in a container filled with cold sodium chloride physiological solution, then fixed in 4% paraformaldehyde, dehydrated, and subsequently embedded in paraffin. Alizarin red stain was performed in paraffin sections of aortic valve tissues. After incubation with 0.2% alizarin red solution for 30 min, excess dye was removed by washing with distilled water. The images were captured using optical microscope (Olympus, Tokyo, Japan), and the calcification nodules appeared orange/red.

Immunoblot analysis

Specimen from aortic valves were cut into pieces and smashed with liquid nitrogen. Protein extracted from aortic valve was then lysed in RIPA Lysis (Beyotime, Jiangsu, China) and Extraction Buffer (Beyotime) containing phenylmethanesulfonyl fluoride (PMSF, Beyotime) for 30 mins on ice. After 12,000 g centrifugation for 10 mins at 4 °C, the supernatants were then transferred to new Eppendorf tubes. The concentration of protein was measured using the BCA method and whole cell lysates were then subjected to western blot using standard procedures.

RNA isolation, cDNA synthesis and RT-PCR

Specimens from aortic valves were cut into pieces and smashed with liquid nitrogen. Total RNAs from leaflets were prepared using Trizol reagent (Thermo Fisher, Waltham, MA, USA). The supernatants containing total RNAs were then purified using Direct-zol RNA MicroPrep Kit (Zymo Research). The concentration of total RNAs was measured using NanoDrop (Thermo Fisher, Waltham, MA, USA) and 0.1–1 µg of total RNAs were reverse transcribed with cDNA Synthesis Kit (Thermo Fisher, Waltham, MA, USA) in accordance with the manufacturers’ instruction. The cDNAs were further sixfold diluted and subjected into RT-qPCR experiments using QuantStudio 6 Real-Time PCR Detection System (Thermo Fisher; Waltham, MA, USA) and KAPA SYBR FAST qPCR Master Mix (2X) Kit (Kapa Biosystems, Wilmington, DE, USA). Primers for RT-qPCR were designed using Primer 5.0 software and synthesized by Invitrogen. Expression was normalized to GAPDH. The data were analyzed using the 2-ΔΔCT method.

Statistical analysis

All results were analyzed by SPSS 22.0 software (SPSS lnc, Chicago, IL, USA). The Chi-square test was used to compare the count data between two groups. One-way ANOVA was used for comparation between multiple groups. Tukey test for multiple comparisons was used to generate the P value between two groups in one-way ANOVA. Data were shown as mean ± SD and displayed using GraphPad Prism version 8 (GraphPad Software Inc., San Diego, CA, USA). The P value < 0.05 was considered statistically significant.

The construction of FEM of aortic valve and ascending aorta

The image data of four participants, including two BAV patients and two healthy volunteers were used to generate 3D geometric model of aortic valve. The shape of aortic valve and the cardiac function of four participants were evaluated using cardiac ultrasonic test (Fig. 1A). These two patients were diagnosed with left–right (RL) leaflet fusion BAV and the function of aortic valve was normal. The data of ascending aorta CTA examination were used to construct 3D geometric surface model (Fig. 1B).

The 3D aorta geometric surface model was generated based on a series of CT angiography images, and the final geometry included the intact aortic valve as well as ascending aortic arch. The structure of heart valve is similar to the thin-walled leaflet, and bending stress is the main cause of valve damage during cardiac cycle (Weinberg et al., 2010). Shell element was used in our study to construct the FEM of aortic valve since bending stress was taken into consideration in it (Kim et al., 2007), and linear hexahedral element was used to generate the FEM of ascending aorta (Fig. 2A). Further, based on the tissue density of heart valve, the 3D geometric model of BAV (Fig. 2B) and TAV (Fig. 2C) were constructed and used for equivalent stress analysis.

Dynamic simulation of the aortic valve

For BAV patients, the abnormal hemodynamics at the aortic valve can cause the valvular calcification. More evidence indicated that mechanical stimulation is the main cause of cardiac valve calcification  (Bogdanova et al., 2018). Based on the FEM, we investigated the equivalent stress on different part of heart valves during one heart cycle. Dynamic FE analyses of the aortic valve model during one heart cycle under different pressure were performed.

At the across valve pressure difference of 25mmHg, von Mises stress distribution on the valve leaflets during the opening or closing phase were shown in Fig. 3A. At the fully open position, the equivalent stress at the belly of the BAV large leaflet was higher than that of TAV. At the closing phase, there was a local high stress zone formed at the belly of BAV large leaflet. In order to investigate the mechanical stimulation at different parts of leaflet, we draw the time-dependent aortic pressure curve during one cardiac cycle (Fig. 3B). There was a stress peak at the systolic phase and the equivalent stress of different parts of the heart valve had no significant difference. At diastole phase, the peak values of BAV large leaflet belly, BAV large leaflet free edge, BAV small leaflet belly, BAV small leaflet free edge, TAV leaflet belly and TAV leaflet free edge were 102 KPa, 54 KPa, 52 KPa, 50 KPa, 45 KPa and 60 KPa respectively.

A similar phenomenon was observed at the across valve pressure of 50mmHg. Overall, the stress on aortic leaflet increased with the elevation of across valve pressure. At the diastole phase, the BAV large leaflet belly suffered higher stress compared with other parts of BAV and TAV (Fig. 4A). The time-dependent stress curve showed that the peak values of BAV large leaflet belly, BAV large leaflet free edge, BAV small leaflet belly, BAV small leaflet free edge, TAV leaflet belly and TAV leaflet free edge were 195 KPa, 100 KPa, 73 KPa, 98 KPa, 105 KPa, and 115 KPa, respectively, at the diastole phase (Fig. 4B).

In general, with the same valve pressure, the difference between BAV and TAV leaflets was not significant at the systolic phase. The equivalent stress on BAV was higher on TAV leaflet at the diastole phase, especially on BAV large valve belly.

Notch1/NICD/Runx2 signaling pathway participate in the calcification of BAV

At diastole phase, the oscillatory shear stress on fibrosa layer of leaflet can promote the valvular calcification (Back et al., 2013). The Notch1 signaling pathway plays an important role in the pathological process of CAVD. Based on the findings of FEA, the higher stress on BAV large valve belly might cause disturbance of Notch1 pathway, leading to the calcification of BAV. We collected the aortic valves from BAV and DCM patients with TAV and their clinical characteristics were shown in Table 1. There was no significant difference of age, gender and CAVD risk factors between these two groups. ECG parameters differed between BAV patients who had valvular calcification and DCM patients with heart failure.

We further explored the expression level of molecules in the Notch1 pathway at different part of BAV and TAV leaflets (Fig. 5A). Alizarin red staining showed that large BAV had significantly more calcification nodules compared with small BAV and TAV (Fig. 5B). The expression of Notch1 on large BAV leaflet belly was significantly lower than that on small BAV leaflet and TAV, which is negatively associate with equivalent stress on leaflet. The activation of Notch1 mediates the release of NICD, and the activation of NICD can inhibit the expression of Runx2, which mediates the calcification of aortic valve (Toshima et al., 2020). The expression pattern of NICD on different parts of aortic valve corresponded with Notch1, and the level of Runx2 on large BAV leaflet was significantly elevated as insufficient of NICD (Figs. 5C–5D). We also extracted the mRNA from different parts of aortic valve, and we found the mRNA of Notch1 was significantly lower at large BAV leaflet belly, on the contrary, the level of pro-calcification gene Runx2 was significantly higher at the root of large BAV (Fig. 5E).