This study obtained human AAA tissue and relevant peripheral blood samples from altogether 147 consecutive cases at the Department of Vascular Surgery, The First Hospital of China Medical University (CMU) following the open operation for aneurysm according to previous description from January 2009 to December 2020 [18, 24]. In addition, each case provided the informed consent for participation. All human AAA samples were collected in line with guidelines from the World Medical Association Declaration of Helsinki. The Ethics Committees of the First Hospital of CMU approved our study protocol (ethical approval number: 2019-97-2).

We collected aneurysm tissues from 147 Chinese AAA cases at the time of emergency or elective open surgical repair. In addition, we obtained the clinical and history data from all patients, like medication history, rupture history, peripheral/coronary artery disease history, or risk factors like hypertension, smoking, hyperlipidemia and diabetes mellitus (DM). Meanwhile, patients with concurrent Marfan syndrome, Ehlers-Danlos syndrome, or additional identified connective tissue or vascular diseases were excluded from this study. Within 147 AAAs, for present study, 32 AAA tissues (28 male patients and 4 female patients) were available for further analyses. For inclusive 32 AAA patients, AAA was analyzed by computed tomography angiography (CTA). For all AAA cases, their diameter of infra-renal abdominal aorta was over 30 mm or 1.5–2 folds as high as the abdominal aorta diameter in corresponding normal segment [25] under CTA diagnosis. Furthermore, the included patients had no evidence or medical history of any cancer disease.

Over the same period, abdominal aorta samples from 12 heart-beating brain-dead organ donors (10 males and 2 females) were used as the controls. For controls, those with concurrent drug history, cancer, infection and additional immune-related disorders potentially affecting this work were excluded.

TRIzol reagent (TaKaRa Bio, Shiga, Japan) was utilized to extract total mRNA from aortic tissues according to the standard protocol as described previously. The extracted total mRNA was used to directly detect the m5c RNA methylation level by adopting the fluorometricm m5C RNA methylation ELISA kit (Epigentek, Farmingdale, NY, USA). Briefly, a pipette was utilized to add mRNA (200 ng) in the detection wells. Later, an m5C detection complex solution with a specific m5c antibody was added to the wells. After washing, the wells were added with fluorescence development solution for incubation under ambient temperature away from direct light. The fluorescence development solution will turn pink in the presence of sufficient m5c products. The fluorescence was read on a fluorescence microplate reader within 2 to 10 min at 530EX/590EM nm. Then, the mRNA m5C methylation level was assessed through the m5c-modified mRNA proportion in overall mRNA (m5C%) following the specific instructions.

For the high-quality RNAs, their A260/A280 ratio was greater than 1.8. The PrimeScript RT Master Mix (Q711-02/03, Vazyme Biotech, Nanjing, China) was used to synthesize cDNA from total mRNA, and the Universal SYBR qPCR Master Mix (Q711-02/03, Vazyme Biotech, Nanjing, China) was utilized to conduct quantitative real-time PCR (qRT-PCR) in LightCycler 480 system (Roche Molecular Systems, Indianapolis, IN, USA) in accordance with the amplification protocols after modification. The PCR conditions were as follows, 30 s of initial denaturation under 95 ${}^{\circ }$C; followed by 5 s of denaturation under 95 ${}^{\circ }$C, and 30 s of annealing and 30 s of extension under 60 ${}^{\circ }$C for 40 cycles. RT-PCR was conducted for twice or more for all samples. GAPDH served as the internal control. Each primer was provided by Sangon (Shanghai, China): NSUN1, NSUN2; NSUN5, NSUN6, Dnmt1 and Aly/REF, and their sequences are shown in Table 1. An improvement of the 2ˆ (–delta delta CT) method for quantitative real-time polymerase chain reaction data analysis.

The RIPA-based reagents were utilized to extract proteins from fresh frozen tissues in line with specific instructions; later, the concentration of each sample was measured by a Pierce BCA Protein Assay kit. The samples were isolated through SDS-PAGE, followed by transfer onto PVDF membranes. Afterwards, membranes were blocked using skimmed milk and incubated using suitable primary as well as secondary antibodies. Later, the ECL detection system was employed for membrane developing in line with specific instructions. GAPDH was used for normalization. In the present work, the following primary antibodies were utilized, anti-Aly/REF (ab202894; Abcam, Cambridge, UK; dilution 1:1000) and anti-NSUN2 (Cat. No. 20854-1-AP; Proteintech, Wuhan, China; dilution 1:1000) and anti-GAPDH (dilution 1:2000; Zsbio, Beijing, China). The densitometry was performed with Image J software (Java 1.8.0_172, National Institutes of Health, Bethesda, Rockville, MD, USA) and normalized to the signal intensity of GAPDH for equal protein loading control of each sample in each experiment.

The typical 2–3-$\mu$m aortic tissue sections were utilized to carry out histological and IHC analyses. In brief, we used hematoxylin-eosin (HE) to stain paraffin-embedded sections for assessing the inflammatory cell composition, morphology and infiltration extent in each AAA sample according to previous description [14].

In IHC assay, antigen retrieval was performed through boiling tissue sections in sodium citrate solution (pH = 6.0), after washing, suitable antibodies were used to treat the sections, then they were deparaffinized and hydrated. To analyze cells within AAA wall, endothelial cells (ECs) were analyzed by anti-CD34 (1:100; Proteintech, Wuhan, China), SMCs were measured through anti-$\alpha$-SMA (1:400; Absin, Shanghai, China), T-lymphocytes were detected through anti-CD3 (1:800; Absin Bioscience, Shanghai, China), while leukocytes were examined by anti-CD45 (1:500; Proteintech, Wuhan, China). In addition, m5c modulator expression within AAA tissues was detected by adopting anti-Aly/REF and anti-NSUN2 in line with specific protocols.

The Nikon ScanScope 90i system (Nikon, Melville, NY, USA) was used to obtain the digital images of the stained tissues (slides). To be specific, the settings for digital image capturing were 20$\times$/40$\times$ magnification and 5-slide load capacity for the 90i system.

To conduct standard and traditional staining, the extent of calcification and cellularity of vascular wall were characterized for histological classification. Thereafter, basic cell count and related cell intensity were used to evaluate the grade, such as new vessel formation, infiltrates and SMCs.

To determine the expression of biomarkers in single cells in aneurysms, we made successive slides of every sample and incubated them using appropriate antibodies. In all cases, one slide was stained with an antibody to detect a certain cell type; thereafter, the anti-biomarker antibodies were used to stain the successive slides.

In brief, typical 2–3-$\mu$m paraffin-embedded sections were subjected to deparaffinate and rehydration with gradient ethanol, followed by boiling within citrate buffer (pH = 6.0) for retrieving the antigen epitopes, rinsing by PBS, and overnight incubation using the rabbit anti-Aly/REF antibody (1:400; Abcam, Shanghai, China) under 4 ${}^{\circ }$C within the humid chamber. After rinsing by Tris-buffered saline thrice, the Alexa Fluor® 488 (green) affiniPure fab fragment goat anti-rabbit antibody (dilution 1:400; Jackson ImmunoResearch Laboratories, West Grove, NJ, USA) was used to incubate the sections for an hour. Then sections were washed 3 times with Tris-buffered saline, and incubated with rabbit anti-NSUN2 antibody (dilution 1:100; Proteintech, Wuhan, China) overnight at 4 ${}^{\circ }$C in a humidified chamber. Sections were washed 3 times with Tris-buffered saline and incubated with CY3 (red) conjugated goat anti-rabbit antibody (dilution 1:300; Servicebio, Wuhan, China) was used to further incubate the membranes for 1 h. The sections were washed 3 times again, and stained by DAPI. At last, the confocal microscope (Nikon CI plus, Tokyo, Japan) was used to obtain sections.

First of all, the Magna RIP™ RNA-binding Protein Immunoprecipitation Kit (Millipore, Billerica, MA, USA) was utilized for RIP-seq in line with specific instructions. In brief, after coating with 10 $\mu$g anti-Aly/Ref antibody (ab202894, Abcam, Cambridge, UK) or corresponding IgG antibody (Millipore, Billerica, MA, USA), cell lysates were used to incubate the coated magnetic beads overnight under 4 ${}^{\circ }$C. Later, the proteinase K digestion buffer was used to treat the RNA-protein complexes. Thereafter, the phenol: chloroform: isoamyl alcohol was used to purify the coprecipitated RNAs, and the RNA content and purity were analyzed by adopting NanoDrop 2000c [26]. Additionally, we eliminated rRNAs out of the immunoprecipitated RNAs; later, the rRNA-depleted RNAs were used to input RNA samples through the NEBNext® Ultra™ II Directional RNA Library Prep Kit (New England Biolabs, Inc., Ipswich, MA, USA) in accordance with specific protocols. In addition, the BioAnalyzer 2100 system (Agilent Technologies, Inc., Palo Alto, CA, USA) was utilized to analyze the quantity and quality of libraries. At last, we loaded the clustered libraries on the reagent cartridge to carry out sequencing using the illumina Hiseq 4000 (Illumina, San Diego, CA, USA) system by the use of 150 bp paired-end reads. Then, Q30 was utilized for quality control. Finally, the Cloud-Seq Biotech (Shanghai, China) was applied in RIP-RNA-Seq high-throughput sequencing.

With regard to mRNA and lncRNA, we used hisat2 software to align high-quality reads to human reference genome (UCSC hg19). Thereafter, we adopted Cuffdiff (version 2.1.0: http://cole-trapnell-lab.github.io/cufflinks/) for obtaining the fragments per kilobase of transcript per million mapped reads (FPKM) under the guidance of Ensembl gtf gene expression profiles, which were the mRNA and lncRNA expression profiles [27]. Further, we determined p-values and fold changes (FCs) according to FPKM, and discovered the differentially expressed mRNAs and lncRNAs. The target genes of lncRNAs were estimated according to their locations relative to adjacent genes. Additionally, the predicted target genes were subjected to GO (www.geneontology.org) and KEGG (www.genome.jp/kegg) enrichment to determine the major functions and related pathways involved in differentially expressed mRNAs and lncRNAs on the basis of differentially expressed mRNAs.

Since the expression of lncRNAs shows significant relation with the surrounding protein encoding genes, numerous lncRNAs play roles of the cis regulators [28]. LncRNA and its adjacent mRNAs were integrated and differentially expressed in the biome to explore the function of lncRNA. LncExpDB (https://ngdc.cncb.ac.cn/lncexpdb) estimates lncRNA genes’ expression reliability and capacities, used for the lncRNA-mRNA interaction network analysis.

PPI networks can offer precious data about cell functions or the signal transduction pathways. In this study, we searched the interactions of DEGs-encoded proteins through the Search Tool for the Retrieval of Interacting Genes/Proteins (http://string-db.org/) online database. Thereafter, the PPI network was built using 5 calculation algorithms (EPC, Degree, EcCentricity, MNC and MCC) and visualized using Cytoscape (http://cytoscape.org/). At last, those overlapping genes obtained by the above 5 algorithms encoded core proteins that had vital biological regulatory activities.

Statistical analysis was completed using SPSS22.0 (SPSS Inc., Chicago, IL, USA). Differences among the categorical variables were compared by chi-square test between groups. In addition, the one-sample Kolmogorov-Smirnov test was used to examine the distribution of data. Later, the non-parametric Mann–Whitney U test or parametric t-test for unpaired samples was used for analysis according to variable distribution. Thereafter, partial correlation analysis was conducted to examine the associations among continuous variables after adjusting for smoking, age and sex. Correlations between continuous variables were quantified by using Spearman’s rank correlation coefficient. A difference of p 0.05 suggested statistical significance.

Table 2 presents the clinical and demographic data of AAA cases and normal subjects. For AAA cases, their age ranged from 43 to 86 years, and the average AAA maximum diameter was found to be 69.25 $\pm$ 20.37 mm. None of the control subject (age ranged 41 to 73) showed any signs of atherosclerosis and no evidence or medical history of aneurysm and the other vascular disorders was known. Age, sex, and body mass index (BMI) were comparable between two groups. At the same time, differences in concurrent diseases, i.e., Hypertension, hyperlipidemia, and renal diseases, were not significant.

The symptoms of the AAA patients are summarized in Table 2. Altogether 22% AAA cases had aneurysmal rupture, whereas 5 out of the 32 cases had iliac aneurysms. Among AAA cases, each AAA sample was semi-quantitatively and histologically characterized for evaluating each histopathological characteristic degree within AAA wall as previously [14]. In Table 3, IHC was used to differentiate between the four main cell types in AAAs, i.e., endothelial cells, lymphocytes, macrophages, and smooth muscle cells to assess the extension of the individual histopathological features in AAA wall.

According to our analysis, relative m5c mRNA methylation level within AAA tissues showed significant correlation with an increased m5c proportion in the overall mRNA relative to controls (more than 1.5-fold; p = 0.040; Fig. 1A).

Fig. 1.

Expression of m5C RNA methylation status (A) and m5C methylation modulators (B, C, D, E, F, and G) at the mRNA level in AAA tissue samples compared with the healthy control aortas analyzed by qRT-PCR. The non-parametric Mann-Whitney U test was all applied to analyze the m5C methylation ratio (A); the m5C “writers” family, NSUN1 (B), NSUN2 (C), NSUN5 (D), NSUN6 (E), Dnmt2 (F); and “readers” family Aly/REF (G). The expression of individual m5C RNA methylation modulators related to the expression of GAPDH set as 1 (100% expression). *, p 0.05; **, p 0.01. m5C, 5-Methylcytosine RNA methylation; AAA, abdominal aortic aneurysm; qRT-PCR, quantitative real-time polymerase chain reaction; ns, not significant.

Thereafter, certain typical molecules that might be related to m5C mRNA modification were analyzed for their expression at mRNA level, including NSUN1, NSUN2, NSUN5, NSUN6, Dnmt2 and Aly/REF (Fig. 1B–G). In the current study, the mRNA expressions of NSUN2, NSUN5 and Aly/REF were observed up-regulating in the AAA tissue samples relative to matched normal tissues ($>$2 folds, p = 0.037; 2 folds, p = 0.035; 2 folds, p = 0.046). Differences in NSUN1, NSUN6 and Dnmt1 mRNA levels were not significant (p = 0.645, p = 0.136 and p = 0.448).

According to our results, the m5C% in total mRNA was significantly correlated with NSUN2, NSUN5 and Aly/REF expression levels (R = 0.316, 0.320, and 0.312 respectively, and p = 0.039, p = 0.037 and p = 0.042, respectively; Table 4). Afterwards, the associations of mRNA expression were examined. As a result, NSUN2 mRNA expression showed significant relationship with the expressions of NSUN5 and Aly/REF (R = 0.538, p 0.001 and R = 0.464, p = 0.002 respectively). Aly/REF expression significantly correlated with NSUN1, NSUN5, and NSUN6 (R = 0.313, 0.642 and 0.339; p = 0.041, p 0.001 and p = 0.026, respectively) Table 4.

Thereafter, the associations of m5C modulator expression levels with clinical features were analyzed. Similarly, the expression of NSUN1, NSUN2, NUSN5, NSUN6 and Dnmt1 positively correlated to platelet hematocrit (PCT) (R = 0.429, 0.429, 0.462, 0.494, and 0.537; p = 0.016, 0.016, 0.009, 0.005, and 0.002, respectively).

NSUN5, NSUN6, and Dnmt1 expression significantly correlated to platelets count (PLT) (R = 0.443, 0.456, and 0.382; p = 0.013, 0.010, and 0.034, respectively). The expression of NSUN1 and NSUN5 at mRNA level were negative associated with mean corpuscular volume (MCV), the correlation coefficient was –0.470 and –0.419 (p = 0.008 and 0.019 respectively) in Fig. 2.

Fig. 2.

Correlation among the mRNA expressions of m5C modulators and clinical parameters. Partial correlation analysis was adjusted with age, gender, BMI, and smoking. BMI, body mass index; LY, lymphocyte; WBC, white blood cell; MONO, monocyte; EO, eosinophils; BASO, basophil; RBC, red blood cell; HGB, hemoglobin; HCT, hematocrit; MCV, mean corpuscular volume; MCH, mean corpuscular hemoglobin; RDWCV, red blood cell volume distribution width; MCHC, mean corpuscular hemoglobin concentration; PLT, platelets; PDW, platelet distribution width; PCT, platelet hematocrit; MPV, mean platelet volume; Ca, serum calcium; K, serum kalium; Cr, creatinine; Cl, serum chlorine; Mg, serum magnesium; Na, serum sodium; GLU, fasting plasma glucose; P, serum phosphate; GGT, gamma glutamyl transpeptidase; ALB, albumin; ALT, alanine aminotransferase; LDL-C, low-density lipoprotein-cholesterol; TG, triglycerides; HDL-C, high-density lipoprotein-cholesterol; CK, creatine kinase; LDH, lactate dehydrogenase; AST, aspartate aminotransferase; TC, total cholesterol; AG, anion gap; CHE, cholinesterase; PA, pre-albumin; CysC, cystatin C; ALP, alkaline phosphatase; TBA, total bile acid; TBIL, total bilirubin; TP, total protein; PT, prothrombin time; Fg, fibrinogen; APTT, activated partial thromboplastin time; PTA, prothrombin activity; INR, international normalized ratio; DD, D-dimer; TES, testosterone; AND, androgen; DHS, dehydroepiandrosterone; F-TEST, free testosterone; SHBG, sex hormone-binding globulin; Hcy, Homocysteine.

We selectively performed western blot analysis for NSUN2 and Aly/REF, because their roles on the 5-mC modification in mRNA, their expressions at mRNA levels and their interrelatedness with mRNA m5C status in human AAA tissue samples. Fig. 3A displays the representative images of NSUN2 and Aly/REF expression levels measured through Western blotting (Supplementary Fig. 1). Aly/REF and NSUN2 protein expression markedly increased within AAA tissues relative to normal controls (more than 1.5-fold for both comparison; p = 0.039, and p = 0.003, respectively; Fig. 3B,C).

Fig. 3.

Protein level of Aly/REF and NUSN2 are highly expressed in human abdominal aortic aneurysm (AAA). The representative images of blots between AAA and healthy aortas at the protein level (A); The density of the protein signals for Aly/REF (B) and NUSN2 (C) through the non-parametric Mann-Whitney U test quantification of the band intensities relative to the expression of GAPDH. *, p 0.05; **, p 0.005. The representative photographs of immunohistochemical staining for NSUN2, Aly/REF (D) and individual cell types markers in AAA tissue. Scale bar, 50 $\mu$m. AAA, abdominal aortic aneurysm; $\alpha$-SMC, smooth muscle cell; CD34, endothelial cell and neovascularization; CD45, leukocytes; CD3, T lymphocytes. (E) Confocal immunofluorescence for human AAA sections stained with NUSN2, Aly/REF and 4’, 6-diamidino-2-phenylindole (DAPI). AAA, specimens of abdominal aortic aneurysm (N = 32). Ctrl, control healthy aorta (N = 12).

Similarly, significantly increased expressions of NSUN2 and Aly/REF were also observed in AAA tissue samples, both were observed in the healthy aorta tissue through IHC analysis, date was not shown. Additionally, we examined the colocalization of NSUN2 and Aly/REF within the successive sections by IHC. As a result, NSUN2 staining showed weak colocalization with CD34+ ECs, and strongly with CD45+ leukocytes and CD3+ T lymphocytes. Aly/REF also showed strong colocalization with CD3+ T lymphocytes and CD45+ leukocytes (Fig. 3D).

Later, the AAA wall and normal aortic wall co-localization of Aly/REF with NSUN2 was examined using IF analysis (Fig. 3E). Here, we also observed that the expressions of NSUN2 and Aly/REF were higher in AAA sections compared with the controls. The result is similar to the results of western blot analysis and IHC analysis. Furthermore, for the co-localization analysis, NSUN2 and Aly/REF were co-expressed in the one cell in AAA group (Fig. 3E).

The possible Aly/REF target genes were predicted by RIP-seq assay. lncRNAs have the length of over 200 nucleotides (nt). In addition, this study analyzed those known lncRNAs for their length distribution. As a result, there were markedly more lncRNAs with the length of 500–1000 and $>$3500 bp compared with those with additional lengths. Here, using RIP-seq, we detected a total of 36,224 lncRNAs in human AAA tissue. Afterwards, reads $\ge$1.0 or those with an anti-Aly/REF group to IgG control enrichment ratio $>$1.5 were screened; finally, 477 lncRNAs that bound to Aly/REF were discovered within AAA samples. For those upregulated lncRNAs, distributions among the human chromosomes were also illustrated in Fig. 4A. A total of 221 intergenic, 20 intron sense overlapping, 136 exon sense overlapping, 38 natural antisense, 43 intronic antisense, 19 bidirectional lncRNAs were identified (Fig. 4B), the lncRNA length of Aly/REF binding lncRNAs (Fig. 4C); and relationship of Aly/REF binding lncRNAs (Fig. 4D).

Fig. 4.

The results demonstrated that 477 differentially expressed genes Aly/REF-binding lncRNAs in AAA tissue. The locus, genomic locus, chromosome of Aly/REF binding lncRNAs (A); thebiotype of Aly/REF binding lncRNAs (B); the lncRNA length of Aly/REF binding lncRNAs (C); and relationship of Aly/REF binding lncRNAs (D) in abdominal aortic aneurysm tissue. Note, exon sense-overlapping: the LncRNA’s exon is overlapping a coding transcript exon on the same genomic strand; intron sense-overlapping: the LncRNA is overlapping the intron of a coding transcript on the same genomic strand; intronic antisense: the LncRNA is overlapping the intron of a coding transcript on the antisense strand; natural antisense: the LncRNA is transcribed from the antisense strand and overlapping with a coding transcript; bidirectional: the LncRNA is oriented head-to-head to a coding transcript within 1000 bp; intergenic: there are no overlapping or bidirectional coding transcripts nearby the LncRNA.

GO analysis indicated that the functions of Aly/REF-interacting lncRNA were involved in a variety of biological processes, including immune system process and macrophages infiltration, i.e., macrophage cytokine production (GO:0010934), immune system process (GO:0002376) for Biological Process; MHC class I protein complex (GO:0042612), platelet alpha granule (GO:0031091) for Cellular Component; and MHC class I protein binding (GO:0042288), MHC protein binding (GO:0042287) for Molecular Function (Fig. 5).

Fig. 5.

Go analysis of 477 differentially expressed genes Aly/REF-binding lncRNAs in AAA tissue. Fold enrichment in biological process (A), cellular component (B), and molecular function (C); The enrichment score in biological process (D), cellular component (E), and molecular function (F).

KEGG pathway analysis indicated the up-regulation of 26 pathways, mainly including ECM-receptor interaction (KEGG: hsa04512), PPAR (KEGG: hsa03320) and phagosome (KEGG: hsa04145) signal transduction pathways, suggested that dysregulated lncRNA pathways in Aly/REF were closely associated with signal transduction in AAA tissue (Supplementary Fig. 2). Based on the FC, the top 20 dysregulated lncRNAs are summarized in Table 5. Among these dysregulated lncRNAs, distributions among the human chromosomes were also illustrated.

After RPKM value distribution was analyzed, we analyzed the sample gene expression profiles on the whole. When IP showed significant enrichment relative to input group, the combined expression for each gene of IP group increased relative to input group. Through RIP-seq, mapping data revealed that FMRP had 3060 potential target genes, which showed extensive activities in cell physiological processes, moreover, there were 369 mRNAs showing differential expression (p 0.01) in AAA tissue (anti-Aly/REF group relative to IgG control).

To reduce the mRNAs binding to Aly/REF to better investigate and enrich mRNAs that might be related to AAA, this study screened the significantly differentially expressed mRNAs (FC $>$4, p 0.01) possibly related to the protein encoding genes annotated based on GO functional annotation and scientific literature. Based on the FC, the top 40 dysregulated mRNAs are summarized in Table 6. As expected, Dnmt1 was candidate mRNAs of Aly/REF-interacting mRNA by RIP-seq identified in AAA tissue.

GO analysis indicated that the functions of Aly/REF-interacting mRNA were related to various biological processes, such as UDP-N-acetylglucosamine metabolic process and endocytic recycling; and cellular component, including eukaryotic 48S preinitiation complex, eukaryotic 43S preinitiation complex; and platelet-derived growth factor biding, histone methyltansferases activity (Supplementary Fig. 3).

KEGG Moreover, regulated pathways were enriched in protein processing in endoplasmic reticulum (KEGG: hsa04141), lysine degradation (KEGG: hsa00310), and Focal adhesion (KEGG: hsa04510) (Supplementary Fig. 4).

Altogether 369 DEGs were analyzed against the STRING database. At the same time, Cytoscape was used to construct a PPI network through neighborhood, co-expression, settings experiments, text-mining and database. Afterwards, the separated genes were removed (those that did not interact with the remaining genes), and the DEGs regulated by Aly/REF were exhibited with 745 edges and 372 nodes (Fig. 6). Based on every gene degree, 4 hub genes that had a $>$20 degree were obtained, including ALB, ATM, TRIP12, and HIF1A.

Fig. 6.

Construction of PPI network, analysis of 369 differentially expressed, and identification of hub genes.

Firstly, the lncRNA-mRNA correlations were analyzed based on 369 differentially expressed mRNAs and 477 differentially expressed lncRNAs. For better investigating the associations among the coding genes, the NIANA approach [28] was used for regulatory network analysis, which exhibited the network objects for the 246 candidate lncRNAs.

Secondly, we screened the candidates in the lncRNA-mRNA interaction network by the correlation $>$0.7 threshold, which yielded a network that consisted of 246 lncRNAs and 369 mRNAs (Fig. 7).

Similarly, 8 hub mRNA candidates (SLC3A2, CCNL1, WDR81, MLLT10, BCL2L1, FHL1, GON4L, MYO15B) also exerted vital parts in the regulatory network, as observed from Table 7. In addition, there were more genes and signaling pathways involved in the above constructed network, suggesting the complicated mechanisms by which Aly/REF-binding mRNAs and lncRNAs regulated the AAA pathogenic mechanism.

Fig. 7.

The Venn diagram of the candidates in the lncRNA-mRNA interaction network. lncRNA and mRNAs were selected with a threshold of correlation $>$0.7, resulting in a network consisting 246 lncRNAs and 369 mRNAs.