Zebrafish husbandry and ventricular cryoinjury

Zebrafish were obtained from a breeding line maintained in the zebrafish core facility at the University of Helsinki. They were housed at 28 °C with a 14:10 h light:dark cycle. Ventricular cryoinjury was conducted as described previously [36]. Briefly, zebrafish were anesthetized with 0.03% tricaine and a small incision was made on the ventral side of the fish for direct access to the heart. A metal probe precooled in liquid nitrogen was placed on the ventricular surface for 15 seconds. The sham-operated zebrafish underwent the same procedures using a metal probe kept at room temperature. After two weeks post-injury, the hearts were dissected, snap-frozen in liquid nitrogen, and stored at −80°C. Given that the surgical impact appears to be the greatest at one dpi, but negligible at later time points [27], we thus employed healthy hearts as controls.

To assess the role of Col5A1 in fibrotic scar remodelling, thirty-two adult zebrafish were randomly assigned to two groups matched by body size: a negative control (NC) group and a COL5A1 GapmeR G2 (COL5A1 G2) group. After anaesthesia, fish received a single intraperitoneal injection of either antisense LNA™ COL5A1 HS DR2 GapmeR (ID 339523 LG00851814-FDA, A*G*T*A*T*T*C*A*C*C*A*T*C*T*G*G, in vivo ready, Qiagen) or antisense LNA™ Negative Control A (ID 339515 LG00000002-FDA, A*A*C*A*C*G*T*C*T*A*T*A*C*G*C, in vivo ready, Qiagen) at a dose of 2.5 mg/kg body weight, followed by ventricular cryoinjury. The GapmeRs were designed to target col5a1 with 100% sequence homology to the human COL5A1. Eleven fish survived the procedure in the NC group and twelve in COL5A1 G2 group. Cardiac function was monitored at 0 (baseline), 7, and 14 days post-injury (dpi). Hearts, livers, and kidneys were then collected. Six hearts were fixed in 10% formalin and embedded in paraffin. Organs from two fish were pooled, snap frozen on dry ice, and stored at −80°C, except for one NC sample, which was processed individually.

Human cardiomyopathy patients

The left ventricular (LV) samples of ischemic cardiomyopathy (ICM) patients were obtained from patients undergoing cardiac transplantation in Helsinki University Hospital between 2014 and 2019. Control samples were derived from the LVs of organ donors without cardiac disease, whose hearts could not be used as whole organ grafts due to tissue type or size mismatch. The myocardium samples were taken from the non-infarcted regions and immediately frozen in isopentane solution (2-methylbutane) cooled in liquid nitrogen and stored at −80°C. The age, gender, and medical and medication history of the patients have been previously described [50, 51].

Echocardiography of adult zebrafish

Echocardiography of adult zebrafish was performed with Vevo 2100® Image System and Vevo Imaging Station (VisualSonics, Amsterdam, Netherlands) equipped with a high frequency transducer (MS700, 30–70MHz) as described previously [53]. In brief, fish were anaesthetized with 0.02% tricaine in system water. Echocardiographic images were acquired in longitudinal axis (LAX) view to record B-mode videos. After echocardiography, fish were placed into a tank filled with fresh system water and usually recovered within 20 seconds.

Image analysis was conducted off-line with the Vevo Lab™ analysis software (VisualSonics) by experienced personnel. Briefly, speckle-tracking analysis of ventricular wall motion was performed to assess ejection fraction (EF) and global longitudinal strain (GLS) from five consecutive cardiac cycles. Ventricular inner border of compact myocardium was defined as the internal chamber. Subsequently, the software calculated EF, strain, and heart rate (HR). EF is defined as the difference between end-diastolic volume (EDV) and end-systolic volume (ESV) divided by EDV, EF = (EDV–ESV)/EDV. Strain (ε) is defined as the change in the length (L) of a deformed object divided by the original length (L0), ε = L–L0/L0. GLS represents the average of six segmental strain values of myocardial wall.

RNA sequencing

RNA from snap-frozen zebrafish hearts (MI n = 4, Ctrl n = 4) was extracted with TRIsure™ Kit (Bioline Ltd, United Kingdom) following the manufacturer’s instructions. Briefly, heart was homogenized with a Tissue-Tearor (BioSpec, Bartlesville, OK, USA) in 0.8 mL of TRIsure. After incubating for 5 minutes at room temperature (RT), 250 μL of chloroform was added and mixed vigorously for 15 seconds. Aqueous phase was separated by centrifuge at 12,000 × g for 15 minutes at 4 °C and transferred to a fresh tube for RNA extraction. To precipitate RNA, 600 μL of cold isopropyl alcohol was added. Followed incubation for 10 minutes at RT, RNA was precipitated with centrifuge at 12,000 × g for 10 minutes at 4°C. Pellet was washed with 800 μL of 75% ethanol, air-dried, and dissolved in DEPC-treated water.

High-throughput RNA-sequencing was performed with Illumina HiSeq™ 2000 (Illumina, San Diego, CA, USA) at BGI (Shenzhen, China) as pair-end sequencing for read of 100 bp. The data quality was analysed with FastQC (http://www.bioinformatics.babraham.ac.uk/projects/fastqc/) and summarized with MultiQC [13]. A light quality trimming was applied to the data with the Trimmomatic software [6]. The sample reads were aligned against Danio rerio genome GRCz11(GCA_000002035.4) (https://www.ncbi.nlm.nih.gov/assembly/GCF_000002035.6/) or miRBase release 22 [28] with STAR-aligner [9]. Alignment statistics were collected with the Qualimap tool [37]. Read quantifications were calculated with FeatureCounts software [30]. Read qualities were accessed automatically in the normalization and statistical processes inside DESeq2 [32].

Analysis of differentially expressed genes and miRNAs

The differential expression analyses were performed with the DESeq2 software in R environment [32]. Negative binomial linear model and Wald test were used to produce p-values. All p values were adjusted for multiple comparisons using Benjamini-Hochberg correction. Probe sets with an adj. p-value < 0.05 were considered significantly differentially expressed. The basic gene annotations came from Ensembl Release 92 [56] and the required data quires were done in R environment with BioMart tool [11] (Supplementary Table 1). In terms of miRNA analysis, infarct 3 appeared to be an outlier and was thus excluded (Supplementary Table 2). Omitting this one sample did not change the most top differentially expressed miRNAs.

Gene set enrichment analysis

Gene set enrichment analysis (GSEA) was performed with the GSEA program (https://www.gsea-msigdb.org/gsea/index.jsp) using all genes to calculate enrichment scores. The enrichment scores were normalized to account for the size of the translational gene set. Genes were ranked according to the correlation between their expression values and the phenotype class distinction using the signal-to-noise ratio. The statistical significance of the normalized enrichment score was estimated using gene set-based permutation test [45].

Similar enrichment analysis was conducted with the general enrichment tool in R (GAGE) using fold changes. The result was subjected to pathway analysis against Kyoto Encyclopedia of Genes and Genomes (KEGG) (https://www.genome.jp/kegg/pathway.html) and Gene Ontology (GO). The enriched pathways were visualized with Pathview tool (https://bioconductor.org/packages/release/bioc/html/pathview.html).

Collagen network analysis

The protein-protein interaction network of all COL genes detected in the RNA-seq dataset from hearts of ICM patients (Supplementary Table 3) was constructed using the STRING database (https://string-db.org/) with a confidence score threshold of 0.7. The resulting COL network was partitioned into five clusters using K-means clustering. Relative expression across cardiac cell types was extracted from summarized single-cell RNA-sequencing profiles provided by the Human Protein Atlas (https://www.proteinatlas.org/). Expression values were Z-score standardized prior to heatmap generation, which was performed in R (version 4.3.1).

Prediction of miRNA-mRNA interactions

Predicted targets for the differentially expressed miRNAs (adj. p-value < 0.05) between healthy and cryoinjured hearts were obtained from TargetScan 6.2 (https://www.targetscan.org/vert_61/). Pearson correlations and corresponding p-values were calculated between mRNA and miRNA normalized expression using all healthy and cryoinjured heart samples. Correlated mRNA-miRNA pairs were extracted using a Pearson correlation p-value <0.05. The predicted targets that were assigned a total context + score in the top 25 percentile and significantly differentially expressed (adj. p-value <0.05) between healthy and cryoinjured hearts were selected. The interactions between miRNAs and their target genes were drawn with RAWGraphs (https://www.rawgraphs.io/).

RNA extraction and real-time quantitative RT-PCR

Owing to the conservation of miRNA gene families among species, we employed human miRNA primers to analyse homologous miRNA genes in zebrafish. Sequence search against miRBase database (http://www.mirbase.org) using the primers as queries identified the respective homologous miRNAs in zebrafish. To analyse miRNA and gene expressions, 30 zebrafish were randomly divided into three groups: control, sham, and cryoinjury. RNA was extracted from 3–4 pooled zebrafish hearts with miRNeasy Mini Kit (Qiagen, Hilden, Germany) according to the manufacturer’s instructions. Reverse transcription was performed with a miScript II RT kit (Qiagen). Quantitative PCR was performed with a miScript primer assay (Qiagen) and miScript SYBR Green PCR kit (Qiagen) using Light Cycler 480 II instrument (Roche Applied Science, Penzberg, Germany). Small Nucleolar RNA, C/D Box 61 (SNORD 61, Catalog #MS00033705, Qiagen) served as an internal control for normalization. The expression levels of investigated miRNAs in sham-operated hearts are similar to those in healthy hearts; therefore, healthy hearts were used to as control for cryoinjured hearts. To confirm the downregulation of col5a1 upon the administration of COL5A1 G2, RNA was extracted from 1–2 pooled hearts, livers, and kidneys of cryoinjured zebrafish.

To detect gene expression, reverse transcription was carried out using a SuperScript® VILO™ cDNA synthesis kit (Invitrogen, Carlsbad, CA, USA). Quantitative PCR was performed with gene specific primers and the LightCycler® 480 SYBR Green I Master (Roche). Glyceraldehyde 3-phosphate dehydrogenase (gapdh) and elongation factor 1α 1a (ef1α1a) served as reference genes for normalization. RT-qPCR was performed at least three times with three technical replicates for each sample. Primers are listed in Table 1.

Table 1. List of primers used in the study

To analyse miRNA expressions in human ventricular samples, RNAs were extracted from snap-frozen LV samples as previously described [51]. Briefly, LV samples were homogenized with OMNI Bead Ruptor for 30 s (6 m/s) in QIAzol reagent (Qiagen), followed by RNA extraction with miRNeasy Mini Kit (Qiagen). Reverse transcription was performed with a miScript II RT kit (Qiagen). Quantitative PCR was performed with a miScript primer assay (Qiagen) and miScript SYBR Green PCR kit (Qiagen). Small Nucleolar RNA, U5G (RNU5G, Catalog #339350, Qiagen) and miR-103a (Qiagen) served as internal controls for normalization. Primers are listed in Table 1.

Cell culture

Human primary cardiac fibroblasts (HCF, Catalog #6300, ScienCell™ Research Laboratories, Carlsbad, CA, USA) were cultured in poly-L-lysine (Sigma Aldrich, St. Louis, MO, USA) coated flasks. Cells were grown in FGM™−3 Cardiac Fibroblast Growth Medium-3 BulletKit™ medium (Lonza, Walkersville, MD, USA) at 37 °C in a 5% CO2 atmosphere until 70–90% confluent. For overexpression or silencing of miR-145-5p, HCF cells were transfected with 10 nM hsa-miR-145-5p miRCURY LNA miRNA Mimic (ID YM00470014; 5'GUCCAGUUUUCCCAGGAAUCCCU, Qiagen) or 50 nM hsa-miR-145-5p miRCURY LNA miRNA Inhibitor (ID YI04102423; GGATTCCTGGGAAAACTGGA, Qiagen) using HiPerFect Transfection Reagent (ID 301704, Qiagen) according to the manufacturer’s protocol. Briefly, HCFs were seeded in 6-well plates, followed by addition of mimic/inhibitor–transfection reagent complexes in serum-free medium, and cultured for 48 hours. AllStars Negative Control siRNA (ID 1027280, Qiagen) and miRCURY LNA miRNA Inhibitor Negative Control (ID YI00199006; TAACACGTCTATACGCCCA, Qiagen) served as negative control of mimic and inhibitor, respectively. For knockdown of COL5A1, HCF were transfected with 10 nM Antisense LNA™ COL5A1 HS DR1 GapmeR (ID 339511 LG00851813-DDA, A*G*C*T*T*C*A*G*A*G*A*G*T*T*G*A, Qiagen) or Antisense LNA™ COL5A1 HS DR2 GapmeR (ID 339511 LG00851814-DDA, A*G*T*A*T*T*C*A*C*C*A*T*C*T*G*G, Qiagen) using HiPerFect Transfection Reagent for 72 hours. Antisense LNA™ Negative Control A (ID 339515 LG00000002-DDA, A*A*C*A*C*G*T*C*T*A*T*A*C*G*C, Qiagen) and Antisense LNA™ GAPDH (human) Positive Control (ID 339515 LG00000005-DDA, A*G*A*T*T*C*A*G*T*G*T*G*G*T*G*G, Qiagen) were included as negative and positive control, respectively. For co-transfection of miR-145-5p inhibitor and Antisense LNA™ COL5A1 GapmeR, 25 nM miR-145-5p inhibitor together with 5 nM Antisense LNA™ COL5A1 HS DR2 GapmeR were added to the cells. HCFs were exposed to cilengitide at concentrations of 1, 5 or 10 µM for 48 hours to examine the optimal dosage of cilengitide. For co-treatment experiments, cells were incubated with 5 µM cilengitide in combination with 50 nM miR-145-5p inhibitor for 48 hours. All experiments were carried out in triplicates, and the experiments were repeated three times, unless otherwise indicated.

Luciferase reporter assay

Two potential binding regions for miR-145-5p in the COL5A1 3′UTR, CCAAGAACGTGCAATAAATTGGAA (741–764), and ATTTTTTCCTCTCAATATATATAATTGGAC (2264-2293) (miRTarBase-#MIRT053211; https://mirtarbase.cuhk.edu.cn/~miRTarBase/miRTarBase_2019/php/detail.php?mirtid=MIRT053211) were cloned into the dual Luc vector pEZX-MT06 (GeneCopoeia, Rockville, MD). The correct insertion was confirmed by sequencing. The primer sets used for cloning are listed in Table 1. HEK293 cells (a kind gift from Dr. Tiina Rasila, University of Helsinki) were grown in DMEM (Gibco) supplemented with 10% FCS and 100 U/ml penicillin and 100 µg/ml streptomycin (Sigma-Aldrich, St. Louis, MO, USA). The cells were transfected with Luc-COL5A13´UTR constructs together with 20 nM miR-145-5p mimic or mimic control in 96-well plate for 24 hours using HiPerFect Transfection Reagent (Qiagen). Cell lysates were subjected to measurements of firefly and Renilla Luc activities using the Dual-luciferase reporter assay kit (Promega, Madison, WI). The firefly signals were normalized using the Renilla signals.

Western blotting

Western blotting was performed as described previously [51]. Briefly, cells were lysed in RIPA buffer (50 mM Tris-HCl, pH 7.4, 150 mM NaCl, 0.25% sodium deoxycholates, 1mM EDTA, 1% NP–40) supplemented with EDTA-free protease inhibitor and phosphatase inhibitor cocktails (Roche, Basel, Switzerland). Protein extracts were resolved on 7.5–10% SDS-PAGE and transferred onto PVDF membranes (Bio-Rad Laboratories, Hercules, CA, USA). Membranes were blocked with EveryBlot Blocking buffer (BioRad) and incubated with the primary antibodies followed by horseradish peroxidase-conjugated or fluorescence-labelled secondary antibodies. The blots were detected using Pierce™ Enhanced Chemiluminescence (ECL) substrate (Pierce Biotechnology, Rockford, IL, USA) of horseradish peroxidase. Quantification was performed using a Gel Doc Image Analyzer (Bio-Rad). Western blotting was performed at least three times. Antibodies are listed in Table 2.

Table 2 List of antibodies used in this studyImmunocytochemistry

HCF cells were cultured on coverslips pre-coated with poly-L-lysine and transfected with miR-145-5p mimic, miR-145-5p inhibitor, or corresponding controls for 48 hours. Cells were fixed in 4% paraformaldehyde in PBS for 10 min at RT and blocked with blocking buffer (10% normal donkey serum, 1% BSA, 0.3% Triton X-100 in PBS), followed by incubation with primary antibodies against αSMA and vimentin overnight at 4°C. The immunoreactivity was detected with AlexaFluor 488 and AlexaFluor 594 labelled secondary antibodies. Nuclei were labelled with DAPI (4`,6-Diamidino-2-Phenylindole) (Molecular Probes, Eugene, OR, USA). Coverslips were mounted in ProLong Diamond antifade mountant (Molecular Probes) and imaged with a Zeiss LSM780 confocal microscope (Carl Zeiss Microscopy GmbH, Jena, Germany). ImageJ 1.53f51 software was used to process images and to quantify the αSMA-positive cells.

Wound healing assay

HCF cells were seeded in a 48-well plate and transfected for 48 hours. A vertical scratch was made with a 200 μL pipette tip followed by extensive washes with PBS to remove the detached cells. Subsequently, a complete growth medium was added. At 0, 4, and 24 hours post-treatment, images were captured using an Invitrogen EVOS™ microscope (Thermo Scientific) and analysed with Image J. The wound closure was calculated at 4 hours post-treatment.

Real‑time cell proliferation and migration assay

Proliferation of HCF cells transfected with miR-145-5p mimic, miR-145-5p inhibitor, or corresponding controls was measured using an xCELLigence Real-Time Cell Analysis (RTCA) instrument with E-16 plates (Agilent Technologies, San Diego, CA, USA). Transfected cells for 48 hours were seeded at a density of 1 × 104 cells/well on poly-L-lysine coated microtiter plates with gold microelectrode biosensors (E-plate 16, Agilent Technologies). Cell density was detected by monitoring changes in electrical impedance at 15-min intervals for 12 hours followed by 30-min intervals for 12 hours with RTCA Software 1.2 (Agilent Technologies). Cell Index (CI) that is related to cellular density was normalized to that at 2 hours. Calculated CI values were plotted on a line graph. Each experimental condition was tested in triplicate. The effect of COL5A1 downregulation on HCF proliferation was performed as described above except that HCF cells were transfected with Antisense LNA™ COL5A1 HS DR2 GapmeR or Negative Control for 48 hours prior to the xCELLigence assays. To assess the effect of cilengitide alone or in combination with the miR-145-5p inhibitor on HCF proliferation, HCFs were treated with 5 µM cilengitide alone or together with 50 nM miR-145-5p inhibitor for 48 hours prior to seeding onto fibronectin-coated (Sigma-Aldrich) E-16 plates. The cell density was monitored for 24 hours.

Migration of HCF cells transfected with Antisense LNA™ COL5A1 HS DR2 GapmeR or Negative Control was measured using an xCELLigence RTCA DP instrument with CIM-plate 16 according to instruction (Agilent Technologies). Transfected cells for 48 hours were suspended and seeded at a density of 2 × 104 cells/well into upper chamber of CIM-plate 16 containing low FBS (1%) FGM™−3 media. The wells of the lower chamber were loaded with FGM™−3 complete media. As the cells migrated towards the chemoattractant across microelectronics sensors, the impedance was measured for 24 hours at 15-min intervals. All experiments were performed in triplicate and repeated three times. However, the proliferation assays of HCF treated with cilengitide alone or in combination with miR-145-4p inhibitor were repeated twice.

Histology

Acid fuchsin orange G (AFOG, Sigma-Aldrich, Saint Louis, MO, USA) staining was performed to detect fibrotic tissue, as previously described [52]. Briefly, paraffin sections were dewaxed and post-fixed in Bouin´s solution for 60 minutes at 60 °C in a water bath, then stained with Weigert´s iron haematoxylin and AFOG solution. The infarct region and ventricle size were measured using ImageJ. Fibrin and collagen stains in the infarct region were analysed using QuPath 0.6.0 software.

Statistical analysis

The data were analysed using Prism software (version 10.3.0; GraphPad, San Diego, CA, USA) and are presented as mean ± SD. One-way ANOVA with Tukey's adjustment or two-way ANOVA with Dunnett´s adjustment was used to calculate differences between more than two groups with normally distributed data. A two-sample, unpaired, two-tailed test was performed to compare two groups with normally distributed data. Data that did not pass the normality and lognormality tests were analysed with the Mann–Whitney test to compare the difference between two groups, or the Kruskal–Wallis followed by Dunn's test for three groups. Statistical significance was set at p<0.05.