All procedures in the present study were approved by the Ethics Committee of the Liaocheng People's Hospital (Liaocheng, China). All subjects provided written informed consent prior to enrollment into the study.

The 8 control patients who admitted for routine testing and 8 patients with DCM underwent internal medical treatment were enrolled in the present study at Liaocheng People's Hospital between July 2015 and May 2016 (Table I). The inclusion criteria for patients with DCM were as follows: i) Left ventricular end-diastolic diameter (LVED) of >50 mm (females) or >55 mm (males); and ii) left ventricular ejection fraction (LVEF) of <45% and/or left ventricular fraction shortening of <25%. Patients with clear causes of idiopathic DCM, including hypertension, coronary artery diseases, valvular disease, congenital defect, alcoholic cardiomyopathy, tachycardia-induced cardiomyopathy and peripartum cardiomyopathy, were excluded from the present study. Patients with a positive family history of DCM were also excluded. The following data were gathered for the DCM and control subjects: Sex, age, blood pressure on admission, heart rate, New York Heart Association classes (18), LVED, LVEF, N-terminal pro b-type natriuretic peptide, medical history and diagnostic testing (Table I).

Blood samples for RNA sequencing were collected from 8 typical patients with DCM and 8 healthy controls. In addition, blood samples from another 20 patients with DCM (16 males and 4 females, 43.2±8.7 years old) and 20 healthy controls (16 males and 4 females, 45.3±7.9 years old) were collected to identify the expression levels of the top 10 lncRNAs with the highest degree via Reverse transcription-polymerase chain reaction (RT-PCR), which was conducted as described below. All the patients were diagnosed with DCM by two experienced clinicians and underwent internal medical treatment at Liaocheng People's Hospital based on the American Heart Association guidelines (19,20). Peripheral venous blood (5 ml) samples were collected from all participants in sterile tubes with ethylenediaminetetraacetic acid.

Total RNA was isolated and enriched using TRIzol^® reagent (cat. no. 15596; Invitrogen; Thermo Fisher Scientific, Inc., Waltham, MA, USA). Next, ribosomal RNA was removed using the Ribo-Zero rRNA Removal kit (cat. no. MRZB12424; Illumina, Inc., San Diego, CA, USA). Ribosomal RNA-depleted RNAs were used to construct RNA libraries with the TruSeq Stranded Total RNA Library Prep kit (cat. no. RS-122-2301; Illumina, Inc.). Libraries were controlled for quality and quantified using the Agilent BioAnalyzer 2100 system (cat. no. 5067-1505; Agilent Technologies Inc., Santa Clara, CA, USA). Libraries (10 pM) were denatured as single-stranded DNA molecules, captured on Illumina flow cells, amplified in situ as clusters and finally sequenced for 150 cycles on the Illumina HiSeq sequencer (HiSeq 4000; Illumina, Inc.). All these procedures were conducted according to the manufacturer's protocol.

Paired-end reads were harvested from Illumina HiSeq sequencer and quality controlled by Q30 (P<0.001). Subsequent to removing the 3′ adaptor-trimming and low-quality reads with cutadapt software (version 1.9.3, http://cutadapt.readthedocs.io/en/stable/index.html), the high-quality trimmed reads were aligned to the reference genome (UCSC hg19) guided by the Ensembl Gff gene annotation file with HISAT2 software (version 2.0.4, http://ccb.jhu.edu/software/hisat2/index.shtml). The cuffdiff software (version 2.2.1; part of Cufflinks, http://cole-trapnell-lab.github.io/cufflinks/cuffdiff/index.html) was then used to obtain the expression profiles of lncRNAs and mRNAs in terms of the fragments per kilobase of transcript per million fragments mapped (FPKM). Next, the fold-change (FC) and P-values were calculated based on the FPKM, and differentially expressed lncRNAs and mRNAs were identified between the DCM and control groups. The criteria were set as an FC of ≥2 or ≤-2, and a P-value of <0.05 between two groups. P-values were corrected with the false discovery rate method. Hierarchical clustering with average linkage was used to calculate the distinguishable expression patterns. RNA sequencing data were uploaded to the Gene Expression Omnibus database (accession no. GSE101585; http://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?&acc=GSE101585).

Total RNA from the blood samples was isolated and enriched with TRIzol^® reagent (Thermo Fisher Scientific, Inc.), and then the concentration was measured with a NanoDrop ND-1000 spectrophotometer (Nanodrop Technologies; Thermo Fisher Scientific, Inc., Wilmington, MA, USA). Total RNA was reverse transcribed into cDNA with the SuperScript III reverse transcriptase (cat. no. 18080044; Invitrogen; Thermo Fisher Scientific, Inc.). Subsequently, 2X PCR master mix (CloudSeq Biotech, Inc., Shanghai, China) was applied to perform the PCR with the target RNAs and internal reference, β-actin (ACTB) of the samples. The primers were as follows: TCONS_00028585, forward: 5′-CGCAGAAGCACAGGGTTAAT-3′, reverse, 5′-GCTGTAGCTTGCTGGCTCTT-3′; ENST00000439796, forward, 5′-TGAAAAACACCGCAGTCAAA-3′, reverse, 5′-AGGCCCTAAAGAAGGCCATA-3′; TCONS_l2_00009351, forward, 5′-ATGAAGCTGGGCAAATCAGT-3′, reverse, 5′-AGATGGCACAAATGCTCTCC-3′; ENST00000448134, forward 5′-TTGGTGGACGGAACAGTGTA-3′, reverse, 5′-GAGTTAGCCAGTGGGGTGAA-3′; ENST00000607876, forward, 5′-CCCTCACACCTCTGTGTCCT-3′, reverse, 5′-GTGACACAGCCACAGTCCAC-3′; and ACTB, forward, 5′-GTGGCCGAGGACTTTGATTG-3′ and reverse, 5′-CCTGTAACAACGCATCTCATATT-3′.

The thermal cycling conditions were set to 40 cycles at 95°C for 10 sec and 60°C for 60 sec, following predegeneration for 10 min at 95°C. Data were analyzed using the 2^−∆∆Cq method (21,22).

The cis target genes were analyzed through the gene annotations provided by the University of California (Oakland, CA, USA) (23), and trans target genes were predicted using BLAT (24). GO analysis, which contained three domains, namely the ‘biological process’, ‘cellular component’ and ‘molecular function’, was performed to explore the functions of differentially expressed coding genes with the Database for Annotation, Visualization and Integrated Discovery (http://david.abcc.ncifcrf.gov) (25). Pathway analysis was also adopted to explore potential pathways of differentially expressed coding genes, according to KEGG pathways (https://www.genome.jp/kegg/). The criteria were set as follows: Fold-change of ≥2 or ≤-2, and P-value of <0.05 between the DCM and control groups.

Pearson's correlation coefficient (PCC) was calculated for each of the lncRNA-mRNA pairs across the DCM and control groups. Co-expressed lncRNA-mRNA pairs with PCC>0.99 and P<0.01 were selected. In total, two lncRNA-mRNA co-expression networks were initially constructed based on co-expressed lncRNA-mRNA pairs in the DCM and control groups. The ‘lost’ network presents the lncRNA-mRNA co-expression pairs, which only appeared in the control group, and the ‘obtained’ network presents the co-expression pairs, which only appeared in the DCM group. Finally, the dynamic lncRNA-mRNA co-expression network was constructed by combining the ‘lost’ and ‘obtained’ networks. Subsequently, key lncRNA nodes were identified in the combined network. According to the mRNAs paired with key lncRNAs, GO analysis was performed to separately identify the significantly enriched biological processes in the DCM and the normal co-expression networks. P<0.05 was considered to indicate a statistically significant difference.

RNA sequencing detected a total of 88,768 lncRNAs in the serum of DCM and control subjects. In order to assess the differentially expressed lncRNAs, the criteria were set as fold-change of ≥2 or ≤-2, and P-value of <0.05 between the DCM and control groups. A hierarchical clustering approach was adopted to verify the consistency in the specimens, according to dysregulated lncRNAs (Fig. 1A). A heat map of partially dysregulated lncRNAs is presented in Fig. 1B. Compared with the control group, 998 dysregulated lncRNAs were identified in patients with DCM, including 661 unregulated lncRNAs and 327 downregulated lncRNAs (Fig. 1C). In addition, a scatter plot was implemented to identify differentially expressed lncRNAs between DCM and control subjects (Fig. 1D).

The dysregulated lncRNAs were subsequently classified in five categories based on their genomic locations, which included intergenic (n=654), bidirectional (n=220), exonic (n=25), antisense (n=6) and intronic (n=5) lncRNAs (Fig. 2A). In terms of length distribution, dysregulated lncRNAs ranged mainly between 500 and 10,000 base pairs (bp; Fig. 2B). These differentially expressed lncRNAs were scattered throughout the human chromosomes (Fig. 2C).

The top 10 upregulated lncRNAs presenting the highest fold-change were TCONS_00028585, ENST00000439796, TCONS_l2_00009351, ENST00000448134, ENST00000607876, TCONS_00015639, ENST00000489016, TCONS_00011165, ENST00000322227 and ENST00000458361. The top 10 downregulated lncRNAs presenting the highest fold-change were TCONS_00003609, TCONS_00018640, ENST00000606664, uc003uey.1, ENST00000494231, ENST00000419628, ENST00000435049, ENST00000609755, TCONS_00026839 and ENST00 000500180 (Fig. 3A). Next, five dysregulated lncRNAs were selected to confirm the lncRNA expression, including TCONS_00028585 (Fig. 3B), ENST00000439796 (Fig. 3C), TCONS_l2_00009351 (Fig. 3D), ENST00000448134 (Fig. 3E) and ENST00000607876 (Fig. 3F). The results verified that the expression levels of these lncRNAs were significantly upregulated in DCM patients as compared with those in control subjects.

In order to assess the potential functions of these dysregulated lncRNAs, GO and KEGG pathway analyses were performed with the target coding genes of downregulated lncRNAs and upregulated lncRNAs, separately. GO analysis was comprised of three networks, including the ‘biological process’, ‘cellular component’ and ‘molecular function’. Through GO analysis of neighbor coding genes, the downregulated lncRNAs were found to be mostly enriched as follows: System development, anatomical structure development and cellular component organization in the ‘biological process’ analysis (Fig. 4A); organelle, intracellular organelle and intracellular part in the ‘cellular component’ analysis (Fig. 4B); and protein binding, cytoskeletal protein binding and enzyme binding in the ‘molecular function’ analysis (Fig. 4C). KEGG pathway analysis revealed that downregulated lncRNAs were mainly associated with gap junction, hypertrophic cardiomyopathy, phagosomes and DCM pathways (Fig. 4D). In addition, upregulated lncRNAs were mainly associated with the regulation of metabolic process, organ morphogenesis and regulation of nucleobase-containing compound metabolic process in the biological process analysis (Fig. 5A), with the cell projection, cell junction and intracellular part in the ‘cellular component’ analysis (Fig. 5B), and with protein binding, enzyme binding and calcium ion transmembrane transporter activity in the ‘molecular function’ analysis (Fig. 5C). The upregulated lncRNAs exhibited a strong association with hypertrophic cardiomyopathy and various cancer pathways, according to KEGG pathway analysis (Fig. 5D).

By RNA sequencing, a total of 1,418 differentially expressed mRNAs (800 upregulated mRNAs and 618 downregulated mRNAs) were identified among the 20,315 detected mRNAs between the DCM and control groups (Fig. 6A), according to the criteria of a fold-change of ≥2 or ≤-2, and P<0.05. A scatter plot was implemented to identify the differentially expressed mRNAs (Fig. 6B). The top 10 upregulated mRNAs were DCAF15, DEPDC4, RASA4, ZNF227, ZNF559-ZNF177, POLR1C, TES, KCNIP2, RAD52 and CD300LF. The top 10 downregulated mRNAs were HLA-A, PHB, SNRPN, DPEP2, OR6S1, ZNF233, DUT, ACD, KIAA1324 and C11orf48 (Fig. 6C).

A dynamic lncRNA-mRNA co-expression network was constructed based on the PCC to identify the correlation between differentially expressed lncRNAs and mRNAs. Initially, the DCM and control lncRNA-mRNA co-expression networks were constructed individually. A total of 1,331 lncRNA-mRNA pairs were detected in the DCM co-expression network (‘obtained’ network), while 1,348 lncRNA-mRNA pairs were identified in the control co-expression network (‘lost’ network). Next, the two networks were combined into a total co-expression network containing 676 nodes (69 lncRNA nodes and 607 mRNA nodes) and 2,661 edges (Fig. 7). The top 10 lncRNA nodes (hub nodes) with the highest degree were XLOC_004581, XLOC_l2_010969, ENSG00000202198, ENSG00000251408, XLOC_008043, BC015774, ENSG00000267172, AK055324, AK094644 and CEP83-AS1 (Fig. 8A).

Furthermore, the blood samples from another 20 patients with DCM and 20 healthy controls were collected to identify the expression levels of the top 10 lncRNA nodes with the highest degree using RT-PCR assay (Fig. 8B). According to the GO analysis, the top 10 lncRNA-mRNA co-expression pairs with the highest degree were mainly associated with positive regulation of cyclin-dependent protein serine/threonine kinase activity, translational initiation, DNA-templated transcription and regulation of muscle cell differentiation in the DCM group (‘obtained’ network; Fig. 8C), and with negative regulation of apoptotic process, nuclear-transcribed mRNA catabolic process and response to stress in the control group (‘lost’ network; Fig. 8D). Finally, an lncRNA-mRNA co-expression network was constructed in order to identify the target mRNAs of the top 10 lncRNA nodes with the highest degree (Fig. 9).