Novel insights into the role of 5-Methylcytosine RNA methylation in human abdominal aortic aneurysm

1. Abstract 2. Introduction 3. Materials and methods 3.1 China Medical University Aneurysm Biobank and tissue collection 3.2 RNA extraction and total mRNA m5C level determination 3.3 Quantitative real-time PCR analysis 3.4 Western blotting analysis 3.5 Histological and Immunohistochemical (IHC) analyses 3.6 Immunofluorescence (IF) staining of NSUN2 and Aly/REF 3.7 RNA Immunoprecipitation sequencing (RIP-seq) 3.8 LncRNA and mRNA interaction network analysis 3.9 PPI network construction and visualization 3.10 Statistical analysis 4. Results 4.1 Characterization of AAA subjects and histological analysis 4.2 Increased m5c mRNA methylation occurred in AAA tissue samples 4.3 The mRNA expression of m5c modulators in AAA tissue samples and association with clinical data 4.4 The correlations among the mRNA m5C status and mRNA expression levels of m5Cc modulators in AAA tissue samples 4.5 The protein expressions of mRNA m5C modulators and their cellular colocalization in AAA tissues 4.6 RIP-seq Identifies Aly/REF-interacting lncRNAs in AAA 4.7 RIP-seq Identifies Aly/REF-interacting mRNAs in AAA 4.8 Regulatory network (lncRNA-mRNA) analysis 5. Discussion 6. Conclusions 7. Author contributions 8. Ethics approval and consent to participate 9. Acknowledgment 10. Funding 11. Conflict of interest 12. Availability of data and materials 13. References


Abstract
Background: It remains largely unclear about the function of 5-methylcytosine (m5C) RNA modification in the context of abdominal aortic aneurysm (AAA). In this regard, the present work focused on investigating m5C RNA methylation and related modulator expression levels in AAA. Materials and methods: To this end, we quantified the m5C methylation levels in AAA tissues (n = 32) and normal aortic tissues (n = 12) to examine the mRNA m5C status and m5C modulator expression at mRNA and protein levels. Meanwhile, mod-ulator localization within AAA tissue samples was detected by immunohistochemistry (IHC). Moreover, RNA immunoprecipitation-sequencing (RIP-seq) was also used to analyze the lncRNAs and mRNA binding to Aly/REF, as an m5C reader. Results: m5C expression markedly elevated in AAA in comparison with normal aortic samples in the AAA cases. The major 5-methylcytosine modulators, including NSUN2, NSUN5, and Aly/REF, which represented the major parameters related to the abnormal m5C modification level, were observed up-regulating in AAA tissues at both protein and mRNA levels. In addition, NSUN2 mRNA level remarkably related to Aly/REF expression, and they were co-expressed in the same cells in AAA group. Regarding the cellular location, Aly/REF was associated with inflammatory (CD45+, CD3+) infiltrates. Simultaneously, after screening for reads in AAA tissue compare with anti-Aly/REF group relative to IgG as control, we obtained totally 477 differentially expressed Aly/REF-binding lncRNAs and 369 differentially expressed Aly/REF-binding mRNAs in AAA tissue. The functions of Aly/REF-interacting lncRNA were involved in immune system process and macrophages infiltration. Through regulatory network (lncRNA-mRNA) analysis, our findings predicted the potential mechanism of Aly/REF-induced lncBCL2L1 and Aly/REF-lncFHL1 axis in AAA and inspire the understanding of m5C and lncRNA in AAA. Conclusions: This study is the first to examine m5A modification within human AAA samples. Our results indicate that m5C modulators, namely, Aly/REF and NUSN2, play vital parts in the human AAA pathogenic mechanism, which shed new lights on the function of m5C modification within AAA. Taken together, findings in this work offer a possible RNA methylation modification mechanism within clinical AAA.

Introduction
Abdominal aortic aneurysm (AAA) accounts for a primary cause leading to cardiovascular event among the old male population [1,2]. AAA is featured by the local while persistent abdominal aortic weakening and expansion [3], and it may be symptomatic, asymptomatic, or may present as rupture. The open retroperitoneal or transperitoneal selection operation has been the frequently adopted repair method [4]. Nonetheless, it is now suggested that the placement of the endoluminal stent graft in aneurysm may be applied in replacement of open intervention [5,6]. There is no effective pharmacological treatment capable of limiting AAA progression or avoiding AAAs rupture. So far, pharmacological intervention is unavailable, while monitoring aneurysm size prior to operation is the only choice [7].
The pathogenic mechanism of AAA has been suggested to be complicated and multifactorial. In previous studies, AAA is suggested to be related to the deficient ad-ventitial/medial arterial layers, like fibroblasts and smooth muscle cells (SMCs). Recent research with either human tissue or animal models has led to a shift regarding AAA, and AAA development is now considered to be part of a significant and dynamic remodeling process in the vessel [8]. Moreover, the critical pathological features are vascular inflammation [9], oxidative stress (OS) [10], aortic extracellular matrix (ECM) destruction [11], and aortic wall thinning due to vascular smooth muscle cell (VSMC) losses [12].
Epigenetic alterations, such as DNA methylation [13], histone modification [14] or RNA modification [15], has been found to exert vital parts in AAA due to their strong impacts on regulating gene levels. N6methyladenosine (m6A) has been well investigated and most frequently observed in mRNA, which is found to be related to pre-mRNA translation, processing, mRNA decay and miRNA biogenesis. m6A modifications are reversible and dynamic among mammalian cells, and they are suggested to be the other epigenetic regulating layer associated with histone and DNA modifications [16,17]. Our recent study has reported that aberrant RNA epigenetic modifications, such as m6A RNA methylation, are present in AAA [18]. The above results can shed novel lights on the AAA pathogenic mechanism.
Hence in this study, we focused on the function of m5C mRNA modification involved in human AAA tissues, and detected m5C mRNA methylation level as well as related modulator expression. In addition, this study examined the relationship of m5C modification with the clinical data of patients. Moreover, RNA immunoprecipitationsequencing (RIP-seq) was also used to analyze the lncRNAs and mRNA binding to Aly/REF, and to investigate the role of m5C RNA methylation in progression of AAA.

China Medical University Aneurysm Biobank and tissue collection
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 descrip- 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.

RNA extraction and total mRNA m5C level determination
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 flu-orescence 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.

Quantitative real-time PCR analysis
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  Table 1. An improvement of the 2ˆ(-delta delta CT) method for quantitative real-time polymerase chain reaction data analysis.

Western blotting analysis
The RIPA-based reagents were utilized to extract proteins from fresh frozen tissues in line with specific in-structions; 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

Histological and Immunohistochemical (IHC) analyses
The typical 2-3-µm 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].
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×/40× 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, ba-sic 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.

Immunofluorescence (IF) staining of NSUN2 and Aly/REF
In brief, typical 2-3-µm 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 • 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 • 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.

RNA Immunoprecipitation sequencing (RIP-seq)
First of all, the Magna RIP™ RNA-binding Protein Immunoprecipitation Kit (Millipore, Billerica, MA, USA) was utilized for RIP-seq in line with specific in-structions. In brief, after coating with 10 µg 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 • 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-trapnelllab.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.

LncRNA and mRNA interaction network analysis
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 network construction and visualization
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
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 ± 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.

Characterization of AAA subjects and histological analysis
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.

Increased m5c mRNA methylation occurred in AAA tissue samples
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).

The mRNA expression of m5c modulators in AAA tissue samples and association with clinical data
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).

The correlations among the mRNA m5C status and mRNA expression levels of m5Cc modulators in AAA tissue samples
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, respec-tively; Table 4). Afterwards, the associations of mRNA expression were examined. As a result, NSUN2 mRNA ex-

The protein expressions of mRNA m5C modulators and their cellular colocalization in AAA tissues
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).
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
Later, the AAA wall and normal aortic wall colocalization 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).

RIP-seq Identifies Aly/REF-interacting lncRNAs in AAA
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 ≥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 over- tion 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.

RIP-seq Identifies Aly/REF-interacting mRNAs in AAA
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 ex-pressed 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/REFinteracting 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 plateletderived 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, coexpression, 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.

Regulatory network (lncRNA-mRNA) analysis
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 NI-ANA 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.

Discussion
AAA is one of the most severe vascular diseases in the vascular surgery [29,30]. Because of the complex pathogenesis of AAA, efficient medical treatment for preventing the occurrence and rupture of AAA is lacking so far [31]. Among the researches on the AAA pathogenesis, epigenetic regulation occupies an increasing decisive position [32]. 5-methylcytosine modification on mRNA is a novel mRNA epigenetic modification, which have been found to play an essential role in other diseases [33,34]. However, the evidence on the relationship between AAA and mRNA m5C modification is still lack. Therefore, this work aimed to examine the relationship of mRNA m5C expression with AAA. And for the first time, we observed that increased mRNA m5C modification occurred in AAA tissues, compared with the healthy controls.
Here, we observed up-regulating in the AAA tissue samples relative to matched normal tissues (>1.5-fold) at mRNA and protein level. Clinically, the NSUN2 expression was positively correlated to platelet hematocrit in AAA patients. NSUN2, firstly recognized as a tRNA m5C-methyltransferases [35], was also identified to methylate mRNA. Several researches have reported previously that NSUN2 regulated the stability, the translation of mR-NAs and further affected gene expressions in cell proliferation [36], oxidative stress [37], inflammation reactions [38] and other pathophysiological processes [39]. These processes might participate in the progression of AAA [40] and other cardiovascular diseases [41]. Recently, Miao et al. [40] suggested that Nsun2 regulated hyperhomocysteinemia (HHcy)-deteriorated AAA progression mostly through   elevating the expression and production of endothelial autotaxin, along with the migration of T cells, and this accounts for a new mechanism for the HHcy-deteriorated pathogenic mechanism and vascular inflammation in AAA. In the current study, we found that NSUN2 was higher within human AAA tissues relative to normal tissues, significantly correlated with m5C status of mRNA, and was localized in the inflammatory cells through IHC analysis. Our results indicated that NSUN2 may play an important role in the m5C methylation of mRNA in the AAA progression. Further research should focus on its potential function on regulating expressions of target genes in AAA, which may provide a new insight on the etiological study of abdominal aortic aneurysm.
Aly/REF is a specific mRNA m5C reader, which can bind to m5C sites in mRNA [33,42], and contributes to the regulation of mRNA export [43]. Aly/REF plays a critical role on promoting the nuclear-cytoplasmic shuttling mRNA export in conjunction with NSUN2 [33]. It has been reported previously that either knock-down of NSUN2 or knock-down of Aly/REF affected the cytoplasmic to nuclear ratios of mRNAs [36]. In the current study, we found a higher level of Aly/REF in AAA group. Additionally, the expression of Aly/REF was strongly correlated with m5C status of mRNAs and the expression of NSUN2. Furthermore, according to the IHC analysis and IF analysis, NSUN2 and Aly/REF were co-expressed in the inflammatory cells in AAA tissues, which illustrated that NSUN2 and Aly/REF might mediate mRNA m5C modification in inflammatory cells. Yang et al. [33] have suggested a similar point about m5C formation in mRNAs is mainly catalyzed by the RNA methyltransferase NSUN2, and m5C is specifically recognized by the mRNA export adaptor Aly/REF as shown by in vitro and in vivo studies. NSUN2 modulates Aly/REF's nuclear-cytoplasmic shuttling, RNAbinding affinity and associated mRNA export. The modification may be related with inflammatory infiltration in AAA. It has been reported that NSUN2 regulated AAA formation by promoting T cell recruitment. However, the research on the association of inflammation with m5C modification and Aly/REF is still lack. Phenotypic transformation of immune cells is an important mechanism of AAA progression. Further researches should focus on the potential role of m5C modification in the phenotypic transformation of immune cells.
Considering the higher level of Aly/REF in AAA and its specific binding to mRNA m5C sites as identified by other researches, Aly/REF may play a critical role in the mRNA m5C status in AAA. Therefore, we tried to perform RIP-Sequence of Aly/REF in AAA tissues to find out the downstream target lncRNA and mRNA of Aly/REF. According to our result, in AAA tissue samples, the downstream mRNAs were involved in several pathophysiological processes. As this study explored, Dnmt1 was candidate mRNAs of Aly/REF-interacting mRNA by RIP-seq identified in AAA tissue. Previous study demonstrated that high Aly/REF expression and low DNMT1 expression were both associated with poor head and neck squamous cell carcinoma prognosis [44]. Simultaneously, the target differential expression mRNA regulated by Aly/REF were exhibited by PPI network, four hub genes were obtained, including ALB, ATM, TRIP12, and HIF1A. As well as Aly/REF-HIF1A interaction, a bioprocesses of bladder cancer cells were demonstrated by a series of experiments in vitro, they found that hypoxia-inducible factor-1alpha (HIF-1A) indirectly up-regulated the expression of PKM2 by activating Aly/REF in addition to activating its transcription directly [45]. Future cellular and molecular studies are required to further validate our findings and to better understand the genes targeted for m5C modifications during AAA progression.
For example, Aly/REF-interacting lncRNAs were associated with various biological processes, including immune system process and macrophages infiltration, i.e., macrophage cytokine production, MHC class I protein complex, MHC class I protein binding, and MHC protein binding. Meanwhile, lncRNAs also participated in Phagosome pathway, PPAR signaling pathway and ECMreceptor interaction pathway, which have been identified to be modulated in the current study. Long non-coding RNAs (lncRNAs), which are the main non-coding RNAs, are transcripts longer than 200 nt. They are known to play a key role in chromatin remodeling, transcription, and posttranscriptional regulation [46]. At present, few studies exist on m5C related lncRNAs. Studies have performed quantitative mapping of the m5C sites in Arabidopsis thaliana on a transcriptome range, and found more than 1000 m5C sites in mRNA, long non-coding RNA [47]. NSUN2 may play a similar role in lncRNAs. Sun et al. [48] showed that NSUN2 deficiency significantly decreased the half-life of H19 RNA (lncRNA), which might be regulated by NSUN2mediated m5C modification in hepatocellular carcinoma.   Nevertheless, the mechanism of m5C methylation in lncR-NAs promoting AAA progress is unclear, and deeper exploration would be helpful for understanding the pathogenesis of AAA. Through bioinformation method, we also predicted eight hub mRNA candidates (SLC3A2, CCNL1, WDR81, MLLT10, BCL2L1, FHL1, GON4L, MYO15B) exerted vital parts in the lncRNAs-mRNAs regulation network. The sequence profile provided reliable basis for the mechanism of m5C methylation regulating AAA progression. However, further researches are needed to focus on the relationship of m5C target RNA and AAA, which may uncover a new cause of AAA occurrence.

Conclusions
We were first to observe m5A modification in human abdominal aortic aneurysm tissues. The results also reveal the important roles of m5C modulators, including NUSN2 and Aly/REF, in the pathogenesis of human AAA and provide a new view on m5C modification in AAA. Our findings suggest a potential mechanism of RNA methylation modification in clinical AAA. Understanding how these m5C RNA modifications occur, and the correlation between lncRNA changes in structure and function, may open up new therapeutic possibilities in AAA. Future molecular and cellular studies are required to further identify our findings and to better understand the genes targeted for m5C RNA methylation modifications during AAA progression.

Author contributions
Conceptualization-JZ and YSH; methodology-YCH, FXY and YHW; software-YSH and YCH; formal analysis-SYW, YSH and YCH; investigation-JZ and YSH; data curation-HZ and YCH; writing-original draft preparation-PPG, YCH and YSH; writing-review and editing-SJX and JZ; supervision-JZ and YSH; funding acquisition-YSH and JZ. All authors have read and agreed to the published version of the manuscript.

Ethics approval and consent to participate
Human AAA samples collection conducted according to the Guidelines of the World Medical Association Declaration of Helsinki, and was approved by the Ethics Committees of First Hospital of China Medical University (ethical approval number: 2019-97-2).

Acknowledgment
We thank Cloud-Seq Biotech Ltd. Co. (Shanghai, China) for the RIP-transcriptome sequencing service and the subsequent bioinformatics analysis.