Effects of artemisinin on ventricular arrhythmias in response to left ventricular afterload increase and microRNA expression profiles in Wistar rats

Animal groups

Overall, 70 male Wistar rats weighing between 225 and 275 g (average, 247.73 g) were obtained from Beijing Vital River Laboratory Animal Technology Co., Ltd (Beijing, China). The care of the animals complied with the principles of laboratory animal care formulated by the animal ethics committee of Peking University People’s Hospital {approval number: [2013] (65)}. Animals received water and food ad libitum until the research began, and they were maintained at a temperature of 24 °C in a 12 h light–dark cycle. A total of 70 Wistar rats were randomly assigned as follows (10 rats/group):

Group 1: (control group) sham surgery, group 2: (model group) transverse aortic constriction (TAC) surgery, group 3: ART 75 mg/kg+TAC surgery, group 4: ART 150 mg/kg+TAC surgery, group 5: ART 300 mg/kg+TAC surgery, group 6: ART 600 mg/kg+TAC surgery, group 7: VER 1 mg/kg+TAC surgery. The ART-pretreated rats were administered by oral gavage once daily for 3 days. The rats of the VER-pretreated groups were slowly administered the medication via tail vein injections over a 10 min period before TAC surgery. The dosage of VER was calculated in accordance with methods reported previously (Bhatti & Foster, 1997; Deng et al., 2017).

Drugs and chemicals

Artemisinin was purchased from Tokyo Chemical Industry Co., Ltd (Tokyo, Japan) and VER was purchased from Shanghai Harvest Pharmaceutical Co., Ltd (Shanghai, China). The VER was dissolved in 0.9% saline solution (Beijing Leagene Biotechnology Co., Ltd, Beijing, China) and ART in Dimethyl Sulfoxide, separately. Heparin and chloral hydrate were obtained respectively from Beijing Solarbio Science & Technology Co., Ltd (Beijing, China) and Sinopharm Chemical Reagent Co., Ltd (Shanghai, China). TRIzol reagent and RNAlater were purchased from Thermo Fisher Scientific Inc (Waltham, MA, USA). Fresh drug solutions were prepared at the beginning of each experiment during the entire experimental period.

TAC surgical procedure

The rats were anesthetized with chloral hydrate (10%), which was administered intraperitoneally at a dose of 3 ml/kg. By clamping the subcutaneous needles with electrodes, a standard electrocardiograph (ECG) lead (Lead II) between the left hind leg and the right foreleg was continuously recorded throughout the surgery. ECG was used to assess the susceptibility to VA related to mechanoelectrical feedback and the side effects of ART and VER. We placed a polyethylene-50 (PE-50) catheter into the right common carotid artery and inserted it into the LV cavity through the aortic valve to measure LV pressure. The PE-50 catheter was filled with 0.3% heparin saline (3 ml/kg), connected to a transducer and, in turn, to a data acquisition and preservation system (Powerlab; ADInstruments Pty Ltd, Bella Vista, NSW, Australia) via a three-way tube. Subsequently, we elicited a rapid pressure overload-induced ventricular arrhythmia using an aortic banding procedure. A mid-sternal thoracotomy was executed to expose the ascending aorta for TAC surgery. The aortic arch between the origin of the left common carotid and right brachiocephalic artery was freed by vascular forceps and threaded by a silk suture (Deknatel 2-0). A 0.7 mm-diameter wire was placed parallel to the aortic arch. A silk suture was tied against the wire and the aortic arch; the wire was then promptly pulled out to constrict the aortic arch. The silk suture was loosened after 5 min of constriction. ECG parameters were recorded, digitalized and analyzed by Labchart data analysis software version 7.2 (ADInstruments Pty Ltd, Bella Vista, NSW, Australia). The sham-surgery group underwent the same surgical process except tightening of the suture was performed first (Fischer et al., 1997). The surgical incisions were closed by suturing, and they were wiped with a povidone-iodine solution. Benzylpenicillin (100,000 U/kg) was administered intramuscularly for 3 days to the rats in all groups. Two days later, all rats were anesthetized, and the hearts were isolated from the body. The LV apex tissue was dissected and placed into centrifuge tubes with RNAlater solutions for RNA stabilization and stored at 4 °C overnight. Then tissues were stored at −80 °C for long-term storage.

Definitions and measurements and of the ECG characteristics

PR intervals and heart rate corrected QT (QTc) intervals were calculated in all groups before TAC surgery. PR interval is the period that extends from the beginning of the P wave until the QRS complex initiation. The QT interval is a time scale between the start of the Q wave and the end of the T wave. The QTc interval was calculated by applying Bazett’s formula, namely, QTc = QT/RR^1/2 (Bealer & Little, 2013).

Walker et al. (Curtis et al., 2013) published the Lambeth conventions to define VA in animal models. PVC is defined as a QRS wave that is different in shape from the normal ventricular complex and is premature in relation to the preceding ventricular complex. VT is defined as a run of four or more consecutive PVCs in animals. A VA scoring system modified by Yang et al. (1995) was used to execute a comparative quantitative analysis of VASs and was defined as follows: 0, no arrhythmias; 1, occasional VPC; 2, frequent VPCs (three or more VPCs within 1 min); 3, VT (one or two episodes); 4, VT (three to five episodes); 5, VT (more than five episodes). The most severe type of ventricular arrhythmia observed in each rat was assigned as one corresponding score during the entire TAC surgery period (5 min). The VASs and the number of rats that exhibited different types of atrioventricular blocks (AVBs) were determined in all groups.

RNA extraction and miRNA library sequencing

Three LV tissues were randomly selected in groups 2 and 4 (indicated by the results in “Effects of ART and VER on VASs,” mean VAS in group 4 was the lowest in ART pretreated groups). Total RNA was isolated from the LV apex using TRIzol Reagent according to the manufacturer’s protocol. RNA purity was estimated by ND-1000 Nanodrop. RNA concentration and integrity were evaluated by Agilent 2200 TapeStation (Agilent Technologies, Santa Clara, CA, USA). Qualified RNAs were connected to a 3′-adapter and followed by a 5′-adapter closely. Subsequently, the adapter-ligated RNAs were reverse-transcribed to cDNAs and cDNAs were amplified with a low-cycle using a reverse transcription-polymerase chain reaction. Then the cDNA products with a length of 18–40 nt were selected by polyacrylamide gel electrophoresis followed by the introduction of a NEBNext® Multiplex Small RNA Library Prep Set for Illumina® (Illumina, San Diego, CA, USA). We constructed and evaluated a cDNA library using the Agilent 2200 TapeStation. The cDNA library was diluted to 10 pM for sequencing (1 × 50 bp) on a HiSeq 2500 platform.

Analysis of sequencing data and DEMs

All the raw reads were obtained from the sequencing machine. After removing adaptors at both ends of the sequences, eliminating low quality reads and cleaning up the pollutant, we obtained clean reads. We aligned the clean reads to the miRBase version 21.0 database to annotate the known miRNAs and calculate the expressions of miRNAs. A bioconductor tool referred to as empirical analysis of digital gene expression in R (Robinson, McCarthy & Smyth, 2010) was performed on normalized expression profiles to identify DEMs between the samples of group 4 and group 2. Significant differences in miRNAs expression were carefully filtered using a normalized fold change value which was named as |[log2(fold change)]| = |log 2 (mean expression ratio (group 4/group 2))| > 1 and p < 0.05.

Target genes prediction, functional and pathway enrichment analysis

We used the TargetScan, miRDB, miRTarBase and miRWalk algorithms to predict target mRNAs of DEMs. The intersections of target genes predicted by the above four algorithms were selected to be the candidate mRNAs. To further understand the signaling pathways and biological functions enriched from the candidate mRNAs, GO and Kyoto Encyclopedia annotation of Genes and Genomes (KEGG) enrichment analyses were carried out using the DAVID gene annotation tool.

Construction of miRNA-mRNA integrated network

A MiRNA-mRNA interaction network was generated. The network consisted of 16 miRNAs manifested by their names, 1,846 pink nodes for mRNAs and each line represented the relationship between each miRNA and mRNA.

DEMs validated by RT-qPCR

Real time quantitative polymerase chain reaction (RT-qPCR) was used to quantify the expression of 16 DEMs in three tissues of group 2 and group 4. Following this, one μg of total RNA extracted from each sample was reverse-transcribed using a miDETECT A Track™Uni-RT Primer. The reactions were incubated at 42 °C for 60 min and at 72 °C for 10 min. Real-time PCR was performed in triplicate wells using CFX96 (Bio-Rad Laboratories, Hercules, CA, USA) according to the protocol. In a 10 μl reaction mixture, 1.0 μl of cDNA was used as a template, with five μl of SYBR green Supermix, 0.4 μl of specific forward primer, 0.4 μl of reverse primer and treated utilizing the following procedure: 95 °C for 10 min, followed by 40 cycles of 95 °C for 10 s, 60 °C for 20 s and 70 °C for 10 s. The 2^−ΔΔCt method was carried out to calculate the relative expressions of DEMs (vs U6 snRNA).

Statistical methods

Electrocardiograph characteristics and LV pressure in all groups were analyzed by one-way ANOVA analyses. We used a two-sided Fisher’s exact test to select the significant pathways and classify functional categories. DEMs expression of groups 2 and four validated by RT-qPCR were analyzed by a Student’s t-test. All results were presented as means and standard deviations. A value of p < 0.05 was considered significantly different.

Characteristics of ECG changes during TAC surgery

The ECG graphs were used to identify where a sinus rhythm appeared in all rats before TAC surgery. LV pressure increased rapidly during TAC surgery (Fig. 1). VA occurred in response to increased afterload as indicated by PVCs (premature QRS complex noted by arrow in Fig. 2C) and VTs (Figs. 2B and 2D); however, AVBs were not generated. Type I second degree AVBs (absent QRS complex noted by arrow in Fig. 2E) and two to one second degree AVBs (absent QRS complex noted by arrow in Fig. 2F) were both observed in the ART and VER pretreated groups (Fig. 2).

Effects of ART and VER on PR and QTc intervals

The average PR intervals in groups 3, 4, 5 and 6 were unchanged compared with the control group (50.37 ± 4.54, 50.30 ± 4.13, 50.27 ± 3.47, 51.89 ± 2.80 vs 50.00 ± 3.66, p > 0.05), while the PR interval was significantly prolonged in group 7 compared with the control group (56.31 ± 3.28 vs 50.00 ± 3.66, p < 0.01) as shown in Fig. 3A. QTc intervals in groups 3, 4, 5, 6 and 7 were unchanged compared with the control group (149.34 ± 6.61, 144.88 ± 6.07, 144.28 ± 5.57, 145.95 ± 7.30 and 144.17 ± 9.79 vs 147.39 ± 5.62, p > 0.05) as shown in Fig. 3B.

Preventive effects of ART and VER on ventricular arrhythmias

As shown in Fig. 4A, the mean VAS was significantly higher in the model group (3.70 ± 0.67) compared with the control group (0.30 ± 0.48). VASs in group 3 (2.70 ± 0.48 vs 3.70 ± 0.67, p < 0.01), group 4 (1.70 ± 0.95 vs 3.70 ± 0.67, p < 0.01), group 5 (2.80 ± 0.79 vs 3.70 ± 0.67, p < 0.01), group 6 (2.60 ± 0.97 vs 3.70 ± 0.67, p < 0.01) and group 7 (1.40 ± 0.52 vs 3.70 ± 0.67, p < 0.01) were significantly decreased compared with the model group (3.70 ± 0.67).

Number of rats that presented with AVBs

Third-degree AVBs were not observed in ART and VER pretreated groups. 1/10 rats in group 3, 2/10 rats in group 4, 2/10 rats in group 5, 5/10 rats in group 6 and 5/10 rats in group 7 presented with type I second degree AVB. In addition, 1/10 rats in group 3, 3/10 rats in group 4, 3/10 rats in group 5, 6/10 rats in group 6, 8/10 rats in group 7 presented with two to one second degree AVB (Figs. 3E and 3F). ART pretreatment with a dosage of 150 mg/kg (group 4) appeared to have the best preventive effect as the average numbers of rats that presented with type I second degree AVB and two to one second degree AVB were the lowest (Fig. 4B).

Deep sequencing results of miRNAs

The expression levels of the 16 miRNAs [|log2(foldchange)| > 1 and p-value < 0.05] in six samples were shown by the heat map (Fig. 5). We identified 16 DEMs in 765 known miRNAs, which were changed significantly [|log2(foldchange)| > 1, p < 0.05)] between group 2 and group 4. Expression levels of rno-miR-200a-3p, rno-miR-341, rno-miR-21-3p, rno-miR-202-5p, rno-miR-370-3p, rno-miR-337-5p, rno-miR-127-5p, rno-miR-495, rno-miR-16-3p, rno-miR-204-5p, rno-miR-743b-5p, rno-miR-134-5p, rno-miR-434-5p, rno-miR-743a-5p and rno-miR-375-3p were significantly upregulated and rno-miR-6319 was downregulated in group 4 compared with group 2 (Table 1).

Prediction of target genes

These results showed that 16 DEMs had 1,846 target genes.

Functional bioinformatics analysis

Figure 5 illustrates the results of the GO annotation analysis; 1,846 predicted target genes were significantly enriched in terms of 361 biological processes, 61 cellular components and 62 molecular functions. Notably, organ morphogenesis, dephosphorylation, neuron projection, axon, protein domain specific binding and transcriptional activator activity and RNA polymerase II transcription regulatory region sequence-specific binding seemed to be the most significantly enriched with GO terms (Fig. 6A). Among the 1,846 predicted target genes, according to the KEGG enrichment results, the main enriched pathways were proteoglycans in cancer, MAPK signaling pathway, the Wnt signaling pathway, lysosome and the Hippo signaling pathway (Fig. 6B; Table 2).

As shown in the miRNA-mRNA regulatory network, rno-miR-495 regulates most of the 431 target genes (Fig. 7). Among candidate target genes in the present study, 25/1,846 genes have four or more miRNA predictive binding sites; Rere in particular can be one of the target genes in up to 5/16 miRNAs, and 24 genes have 4/16 miRNA binding sites (Fig. 8; Table 3).

Results of DEMs expression collected by RT-qPCR

Real time quantitative polymerase chain reaction results supported the expression pattern of miRNAs obtained by high throughput sequencing, with the exception of rno-miR-204-5p, rno-miR-21-3p and rno-miR-6319. In all DEMs, rno-miR-370-3p was not detected by RT-qPCR as its expression was very low (Table 4).