Identification of Whole Blood C Ircular RNA Expression Profiles and Potential Regulatory Roles in Children with Human A denovirus Pneumonia


 Human adenoviruses (HAdVs) causes infections predominantly in infants and young children, and severe HAdVs pneumonia results in high lethality rate in children. CircRNAs are novel identified long non-coding RNAs and play important roles in gene regulation and pathogenesis of disease. To elucidate the roles of circRNAs in HAdVs pneumonia, we analyzed the circRNAs profiles of healthy children and children with HAdVs pneumonia including mild and severe cases, and identified 139 significant upregulated circRNAs in children with HAdVs pneumonia vs healthy controls, and 18 significant upregulated circRNAs in children with severe HAdVs pneumonia vs mild HAdVs pneumonia. In particular, hsa_circ_0002171 was found differently expressed in both groups, which may be taken as diagnostic biomarkers for HAdVs pneumonia and severe HAdVs pneumonia. To identify the underlying mechanisms of circRNAs in HAdVs pneumonia, we analyzed the transcriptome of children with HAdVs pneumonia and established circRNA-mRNA regulatory network. Enrichment of the differentially expressed (DE) target mRNAs demonstrated that DE genes between healthy controls vs HAdVs pneumonia were mainly involved in RNA splicing, while DE genes between children with mild and severe HAdVs pneumonia were mainly from the regulation of lymphocyte activation. Receiver operating characteristic (ROC) curve analysis suggested that hsa_circ_0002171 had significant value in HAdVs pneumonia and severe cases diagnosis. Taken together, profile of circRNA expression was altered in children with HAdVs pneumonia and severe cases. These results demonstrate that hsa_circ_0002171 as a potential diagnostic biomarker of HAdVs pneumonia and severe cases and may be a novel therapeutic target.


Introduction
Human adenoviruses (HAdVs) belong to Adenoviridae family and are nonenveloped double-stranded DNA viruses. As important causes of acute respiratory tract illness in children, HAdVs account for 3.8-11% of pneumonias [1][2][3]. HAdVs had the highest correlation with severe pneumonia in children, accounting for 20-33.3% of severe pneumonia [1,[4][5][6]. The fatality rates for untreated severe HAdVs induced pneumonia may exceed to 50% [1,4,5].It is a signi cant burden to society due to associated medical expenses [3]. Unfortunately, evaluation of HAdVs pneumonia remains to base on clinical manifestations, is short of sensitive and speci c biomarkers. Furthermore, no adenovirus-speci c approved antivira drugs are available for severe cases currently [7]. Early diagnosis and treatment of severe HAdVs pneumonia remain to be major challenge. Therefore, in order to decrease the morbidity and improve the prognosis, it is particularly important to explore novel early indenti cation biomarkers and potential antiviral mechanisms in severe cases.
The HAdVs infection process has been considered to involve RNA splicing and interaction with the innate immunity of the host [8]. Notably, HAdVs RNA splice sites were detected in previous study and some of them are possibly non-coding RNAs [9]. Nonetheless, the role of non-coding RNA in HAdVs pneumonia still needs to be investigated further.
Circular RNAs (circRNAs) have been identi ed as a class of noncoding RNAs that formed as a closed loop from the covalent linkage of the 3' and 5' ends. Thus, unlike linear RNA, a special circular covalently bonded structure of circRNAs enables high stability and resistance against RNA exonuclease [10]. As a novel type of endogenous noncoding RNA, circRNA is expressed widely in different species and has been demonstrated to be a class of RNA with tissue speci city [11,12]. With stability, large abundance and evolutionary conservation, circRNAs have key functions in regulating various physiological and pathological processes [13,14]. It has been demonstrated that circRNAs could serve as competing endogenous RNAs to inhibit the activity of microRNAs (miRNAs). Increasing evidence reveals that circRNAs have regulatory roles in many human diseases, such as cardiovascular disease, diabetes and cancer, which make circRNA potential diagnostic and prognostic biomarkers [14][15][16].
Studies have demonstrated that the virus generated circRNAs or differentially expressed host circRNAs can be used as candidate biomarkers for viral infection [17][18][19]. Meanwhile, circRNAs are involved in regulating antiviral immuneresponse [17], viral replication [20], and pathogenesis of infectious diseases [19]. Moreover studies have shown that several circRNAs are involved in respiratory diseases. For example, Circular RNA GATAD2A promotes H1N1 replication by inhibiting autophagy [20].
However, expression pro ling and the potential role of circular RNAs in HAdVs pneumonia is largely unknown. In this regard, the present study is designed to identify the whole blood circular RNAs expression pro les and potential regulatory roles in children with HAdVs Pneumonia, which may provide several novel candidate clinical diagnostic and therapeutic biomarkers for HAdVs pneumonia and severe cases especially.

Materials And Methods
Patients and ethics statement Children were recruited from Guangzhou Women and Children's Medical Center. Peripheral whole blood samples was collected from 24 healthy children (denoted as healthy control), 20 children with mild HAdVs pneumonia (denoted as mild), and 18 children with severe HAdVs pneumonia (denoted as severe).
The characteristic of all participants selected in the study are presented in Table 1. High-resolution computed tomography of the chest in children with mild and severe HAdVs pneumonia was shown in Figure 1. Among them, a cohort of 6 patients (4 mild and 2 severe) and 3 healthy controls was used for highthroughput sequencing, and a cohort of 53 children (21 healthy controls, 16 mild, and 16 severe) was used for validation. The ow path of the present study cohort was shown in Figure 2. Pneumonia was de ned as the presence of fever, acute respiratory symptoms (cough, tachypnoea, and di cult breathing), or both, along with the presence of new in ltrate detected by chest radiography and/or consolidation. The evidence of adenovirus infection was identi ed by positive multiplex polymerase chain reaction (PCR) for HAdVs from nasopharyngeal swab and/or bronchial alveolar lavage uid. Exclusion criteria of this study included: (1) there was evidence of infection with other organisms; (2) corticosteroids were used as part of therapy before the study; (3) there were signi cant underlying comorbidities, including malnutrition, chronic cardiac or chronic pulmonary disease, asthma, tumor, congenital malformation, mmunode ciency and immunosuppressive medications before admission,etc. Pneumonia severity classi cation was performed according to the British Thoracic Society guidelines [21].
This study protocol was conducted in accordance with the declaration of Helsinki and approved from the Ethics Committee of Guangzhou Women and Children's Medical Center, Guangzhou Medical University. All participants were provided written informed consent form by legal guardian for using their clinical and laboratory data from their medical reports.

RNA isolation
The total RNA of each whole blood sample was isolated using TRIzol reagent (New England Biolabs) according to the manufacturer's protocol. The extracted RNAs were used for circRNA and mRNA high-throughput sequencing, or qRT-PCR experiment.

High-throughput sequencing
The circRNA and mRNA sequencing were performed using the NEB Next Ultra Directional RNA LibraryPrep Kit for Illumina (NEB, Ispawich, USA). We selected the differentially expressed circRNAs and mRNAs with criterion of p-value ≤ 0.05 and log2FoldChange ≥ 2, while for differentially expressed miRNA with criterion of p-value < 0.05 and effect size > 1.0.
Quantitative real-time PCR assay cDNA for circRNA and mRNA was generated using PrimeScript RT reagent Kit (Takara) according to the manufacturer's protocol. The expression of circRNA and mRNA were analyzed with SYBR Premix Ex Taq (TaKaRa) kit. The circRNA and mRNA were normalized with β-actin and evaluated by the 2 -ΔΔCt method.

Data analysis
The R package DESeq2 (v1.26.0) was used to perform the differential expression analysis on each group from the circRNA and mRNA counts 33 . Differential expressed circRNAs and mRNAs were de ned as those with a p-value ≤ 0.05 and absolute log2FoldChange ≥ 2. These differential expressed circRNAs were then used to predict the corresponding sponger miRNAs by miRanda (3.3a). The R package multiMiR (v.1.8.0) was used to screen the validated mRNA targets of these miRNAs from databases like mirecords, mirtarbase and tarbase [22][23][24]. Enrichment analysis of Gene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway was performed by clusterPro ler (v.3.14.0) to investigate functions of differential expressed circRNAs and mRNAs. Besides, the overlaps between validated mRNAs and differential expressed mRNAs were also used as candidates for enrichment analysis. GO terms and KEGG pathways with p-value ≤ 0.05 were considered signi cantly enriched.

Results
Identi cation of differentially expressed circRNA pro les in children with HAdVs pneumonia To explore the different expressed circRNAs in HAdVs pneumonia patients, we collected whole blood samples from three healthy children, four children with mild HAdVs pneumonia, and two children with severe HAdVs pneumonia for circRNA pro ling. Of note, by comparing the samples from children suffered HAdVs pneumonia with healthy controls, we were able to identify the roles of circRNAs in HAdVs pneumonia, while comparing the samples from mild group with those from severe group, we were able to analyze the role of circRNAs in pediatric severe HAdVs pneumonia. We identi ed 208 circRNAs were differentially expressed (DE) in samples from children with HAdVs pneumonia v.s. samples from healthy controls ( Figure 3A), while 92 circRNAs were differentially expressed in samples from severe group v.s. samples from mild group ( Figure 3B). The hierarchical clustering analysis showed that 69 upreguated circRNAs plus 139 downregulated circRNAs in the samples from HAdVs infected children ( Figure 3A and 3C), and 74 upreguated circRNAs plus 18 downregulated circRNAs in the samples from severe group passed our fold-change lter (log2 fold change > 2.0, and P value < 0.05) ( Figure 3B and 3D). The 10 top up-and down-regulated circRNAs between children with HAdVs pneumonia and healthy controls were listed in Table 2. The 10 top up-and downregulated circRNAs between mild group and severe group were shown in Table 3. These data indicated that the expression of circRNAs in children with HAdVs pneumonia were different from that in healthy controls, and the expression of circRNAs in children with severe HAdVs pneumonia were also different from that in children with mild HAdVs pneumonia.
Validation of the differentially expressed circRNAs using qRT-PCR To con rm the high-throughput sequence data, we selected the DE circRNAs in Table 2 and Table 3 for further validation with qRT-PCR with another 6 groups of samples. We found that hsa_circ_0002171, hsa_circ_0022787, hsa_circ_0005013, hsa_circ_0027651, hsa_circ_0025965, and hsa_circ_0027659 were upregulated in samples from children with HAdVs pneumonia v.s. healthy children ( Figure 4A), while hsa_circ_0014349, hsa_circ_0002171, hsa_circ_0003646, and hsa_circ_0009627 were upregulated in samples from severe group v.s. mild group ( Figure 4B), which was consistent with our sequencing data. In particular, we found that hsa_circ_0002171 was differently expressed among both children with HAdVs pneumonia v.s. healthy control children group and severe group v.s. mild group. Based on the result of 10 samples from severe group, 10 samples from mild group and 15 healthy controls, we found that hsa_circ_0002171 was signi cantly increased in samples from children with HAdVs pneumonia children while compared with healthy controls ( Figure 4C), and signi cantly increased in samples from severe group while compared with mild group ( Figure 4D). Taken together, these results indicated that hsa_circ_0002171, hsa_circ_0022787, hsa_circ_0005013, hsa_circ_0027651, hsa_circ_0025965, and hsa_circ_0027659 may re ect HAdVs pneumonia in children, while hsa_circ_0014349, hsa_circ_0002171, hsa_circ_0003646, and hsa_circ_0009627 may re ect severe HAdVs pneumonia. Thus, these circRNAs can likely be taken as candidate biomarkers for HAdVs pneumonia and severe HAdVs pneumonia in children respectively. Furthermore, hsa_circ_0002171 was con rmed as a candidate biomarker for severe adenovirus pneumonia.

Transcriptional regulation of circRNAs in children with HAdVs pneumonia
To further explore the possible mechanisms of the differentially expressed circRNAs in children with HAdVs pneumonia, we performed RNA-sequencing and constructed the circRNA-mRNA regulatory network. We combined the DE mRNAs and circRNAs with circRNA target predictions to obtain genuine circRNA targets in children with HAdVs pneumonia. According to the results of the DE mRNAs and circRNA expression analysis, the regulatory relationships between circRNAs and mRNAs were predicted. Based on the results, we found that 283 upreguated target mRNAs plus 993 downregulated target mRNAs in the samples from children with HAdVs pneumonia compared with healthy control children( Figure 5A), and 465 upreguated target mRNAs plus 51 downregulated target mRNAs in the samples from severe group compared with mild group passed our fold-change lter (log2 fold change > 2.0, and P value < 0.05) ( Figure  5B). We further performed the functional enrichment analysis including Kyoto Encyclopedia Genomes Genes (KEGG) pathway analysis ( Figure 6A and 6B) and Gene Ontology (GO) analysis ( Figure 6C and 6D) of DE mRNAs. The top 10 enriched GO terms and pathways of DE mRNAs were ranked by enrichment score.
The KEGG pathway of DE mRNAs between healthy controls and children with HAdVs pneumonia showed that the DE mRNAs were involved in viral infection ( Figure 6A), and the KEGG pathway of DE mRNAs between mild and severe group showed that the DE mRNAs were also involved in viral infection ( Figure 6B). However, the GO analysis of the DE mRNAs between healthy controls and children with HAdVs pneumonia indicated that RNA processing pathway ( Figure 6C), while the GO analysis of the DE mRNAs between mild and severe group indicated that immune pathway ( Figure 6D). The DE mRNAs of top 1 enriched GO pathway were showed in Table 4. We further established the circRNA-miRNA-mRNA gene regulatory networks from the above circRNA-mRNA gene pairs among HAdVs pneumonia group v.s. healthy control group ( Figure 7A) and severe group v.s. mild group ( Figure 7B) using Cytoscape software. These results indicated that genes involved in mild HAdVs pneumonia were different from genes invloved in severe HAdVs pneumonia. These genes may be the anti-HAdVs infection targets or pneumonia immunotherapy targets, and re ect the mechanisms of HAdVs infection and HAdVs-induced pneumonia.
ROC Curve Analysis of Con rmed circRNAs ROC curve analysis was conducted to evaluate the potential diagnostic value of signi cantly and differentially expressed circRNAs. ROC curves of con rmed circRNAs showed that levels of hsa_circ_0002171 could separate the patients with HAdVs pneumonia from the healty controls, as well as severe cases from mild cases. Results of ROC curve analysis showed that the diagnostic accuracy of hsa_circ_0002171 in predicting HAdVs pneumonia and severe cases was 77.33% and 79%, respectively( Figures 8A, B). If the cut-off value of hsa_circ_0002171 was set at 0.8401, the sensitivity and speci city of predicting HAdVs pneumonia was 70% and 80%, respectively. While a sensitivity of 80% and a speci city of 70% to predict severe HAdVs Pneumonia casess at the cut-off value of 1.633 ( Figure 8C). Thus, it appeared that hsa_circ_0002171 may be valuable as a biomarker for HAdVs pneumonia diagnosis.

Discussion
Human adenoviruses (HAdVs play a signi cant role in pediatric pneumonia, and it is frequently associated with severe pediatric pneumonia, causing a signi cant morbidity and mortality in infants and children [25]. However, though high mortality rates are often reported in cases of severe HAdVs pneumonia, there is no no speci c invention for severe HAdVs pneumonia and the underlying mechanisms of HAdVs pneumonia remain to be discovered [6]. Severe HAdVs may cause serious long-term sequelae or even demise. Thus, investigation of the molecular mechanism network of interactions between the HAdVs and host factors, and identi cation of the biomarkers for severe HAdVs pneumonia are crucial to the early diagnosis and innovative therapeutic targets.
The severity of pediatric HAdVs pneumonia is associated with the age and immune state of host, subtype of HAdVs. Current recognition of severe HAdVs pneumonia precailingly rely on the clinical characteristics manifesting in the late stage[6]. At present, there is still a lack of indicators for early speci c recognition of HAdVs pneumonia, especially for severe cases.
CircRNAs are ubiquitously biological functions in various tissues and cells.They are recently reported as unique class of long non-coding RNAs that affect the expression of the target miRNA, and thus regulate the mRNA expression and in uence many biological processes [26,27]. CircRNAs are discovered as novel clinical diagnostic and therapeutic biomarkers for several diseases [14,28]. The Non-coding RNAs have been considered to be involve in RNA splicing of HAdVs infection [9]. It was reported that highly upregulated microRNAs may have crucial roles in HAdVs pneumonia pathogenesis and are potential biomarkers [29]. Moreover, studies have shown that several circRNAs are involved in respiratory diseases or disorders. For example, Circular RNA GATAD2A promotes H1N1 replication by inhibiting autophagy in cell model [20]. However, there is no studies have been reported the expression pro les and role of circRNAs in whole blood from children with HAdVs pneumonia. Hence, it would be meaningful to pro le circRNA expression and search for new biomarkers, which may contribute to providing new directions and strategies for disease diagnosis and treatment.
In our present study, circRNA expression pro les in whole blood from children with HAdVs pneumonia were rstly analyzed and validated. From the difference analysis, our study identi ed 208 differentially expressed (DE) circRNAs, with 69 upreguated and 139 downregulated, in whole blood of children with HAdVs pneumonia, as compared to healthy controls. And 92 circRNAs were differentially expressed in whole blood from severe cases, including 74 upreguated circRNAs and 18 downregulated circRNAs, as compared to mild group. Furthermore, we increased the number of samples and carried out real-time qPCR to validate the sequencing results. We chose ten circRNAs from the upregulated ones according to their expression distribution in each specimen. As a consequence, hsa_circ_0002171, hsa_circ_0022787, hsa_circ_0005013, hsa_circ_0027651, hsa_circ_0025965 and hsa_circ_0027659 were veri ed to be markedly upregulated in the HAdVs pneumonia group; hsa_circ_0014349, hsa_circ_0002171, hsa_circ_0003646 and hsa_circ_0009627 were veri ed to be markedly upregulated in the severe HAdVs pneumonia group. These ndings suggested that HAdVs pneumonia can change circRNAs expression patterns in children, and this change may be related to the severity of the disease. These ndings also suggested that differentially expressed circRNAs may be involved in regulating the pediatric HAdVs pneumonia process. Our results may enrich the study of the pathogenesis of HAdVs pneumonia and provide a theoretical basis for the indepth exploration of the function of circRNAs in HAdVs pneumonia,especially in severe cases.These differentially expressed circRNAs might be implicated in HAdVs pneumonia, while the exact mechanism requires further investigation. In addition, these circRNAs may be used as biomarkers for diagnosis and prediction of HAdVs pneumonia, and may also be therapeutic targets for future innovative treatments.
In this study, we found that a set of circRNA could serve as potential biomarkers for pediatric HAdVs pneumonia. Previous study has shown that non-coding RNA in whole blood can be used as a potential biomarker, such as hsa-miR-127-3p, hsa-miR-493-5p, and hsa-miR-409-3p [29]. However, no association between these markers and disease severity has been established, and sensitivity and speci city are unclear. In our study differential expression of hsa_circ_0002171 was detected between the HAdVs pneumonia group and the healthy control group, as well as between the severe and mild cases. As expected, the average expression levels of hsa_circ_0002171 in the whole blood of patients with HAdVs pneumonia was signi cantly higher than those of the healthy controls.
Furthermore, the expression of hsa_circ_0002171 in the whole blood of severe cases was signi cantly higher than that of the mild cases. Therefore, in this study we found that the expression levels of hsa_circ_0002171 were positively correlated with the severity of HAdVs pneumonia in children. By ROC curve analysis, hsa_circ_0002171 showed signi cant value of diagnosis of HAdVs pneumonia and severe cases, with AUC 0.7733 (95% CI 0.6183-0.9284, p = 0.0063) and AUC 0.7900(95% CI 0.5861-0.9939, p = 0.0284), respectively. The value of diagnosis of severe cases was higher than procalcitonin, interleukin 6, erythrocyte sedimentation rate, C-reactive protein and prealbumin (AUC were all less than 0.656) reported in previous research [30]. The above results of our study indicated that hsa_circ_0002171 may have potential as a novel biomarkers in diagnosis of pediatric HAdVs pneumonia and p rediction of severe cases, also proving that hsa_circ_0002171 is of great importance for practical applications. Unfortunately, to our knowledge, the biological role of hsa_circ_0002171 have not been reported by previous study.
After integrating transcriptome analysis, we have identi ed the target mRNAs of DE circRNAs. With GO analysis of the DE target mRNAs, we found that DE genes between healthy controls vs HAdVs pneumonia were mainly involved in RNA splicing. This was consistent with the presence of RNA splicing during adenovirus infection[8].Among these genes, several genes have been reported involved in viral replication including FUS, PPP2CA, EIF4A3, PCBP2, and TARDBP, which implicated the underlying mechanism for HAdVs pneumonia, and will provide therapeutic targets for HAdVs pneumonia. Interestingly, when we analyzed the GO enrichment of DE genes between children with mild vs severe HAdVs pneumonia, we found that the genes were mainly involved in lymphocyte activation, which is different from pathway of HAdVs pneumonia as compared with healthy controls. These results indicated that severe HAdVs pneumonia is not only associated with virus infection, but also involved in triggered by host excessive immune response during infection. Thus, the treatment for severe HAdVs pneumonia should be combined with immunotherapy. This is consistent with the greater need for immunoregulatory therapy (such as glucocorticoids and intravenous immunoglobulin) in the treatment of severe HAdVs pneumonia [6].The DE genes between children with mild vs severe HAdVinduced pneumonia may be good targets for the immunotherapy.
For the rst time, to our knowledge, our results provide a pro le of circRNA expression and circRNA-miRNA-mRNA regulatory network in pediatric HAdVs pneumonia and severe. The association between hsa_circ_0002171 and disease severity suggests that hsa_circ_0002171 plays a critical role in the development of pediatric HAdVs pneumonia. In summary, hsa_circ_0002171 is of potentially signi cant value in pediatric HAdVs pneumonia diagnosis, disease severity prediction and therapeutic target, but there remains a need to further study the mechanisms of hsa_circ_0002171 in pediatric HAdVs pneumonia.

Declarations
Authors' contributions DYY and GL created the concept and design of this study. KS were responsible for the statistical analysis. HFF and TTS participated in sample diagnosis and data collection. FH and XXC performed the experiments. DYY and GL drafted, revised and edited the manuscript. Availability of sequencing data and materials The datasets generated and analysed during the current study are available from the corresponding author on reasonable request.

Compliance with ethical standards
Con ict of interest The authors declare no con ct of interest.
Ethical approval This study was approved from the Ethics Committee of Guangzhou Women and Children's Medical Center, Guangzhou Medical University.   Flowchart of the study. HC: Healthy control group; Mild: Mild HAdVs pneumonia group; Severe: Severe HAdVs pneumonia group.

Figure 3
The circRNAs expression pro le of whole blood from HAdVs pneumonia children.Volcano plot (a) and hierarchical cluster analysis (c) of differently expressed circRNAs between HAdVs pneumonia group and healthy control group.Volcano plot (b) and hierarchical cluster analysis (d) of differently expressed circRNAs between mild HAdVs pneumonia group and severe HAdVs pneumonia group.

Figure 7
The circRNA-miRNA-mRNA regulatory network in HAdVs pneumonia group.The circRNA-miRNA-mRNA regulatory networks from circRNA-mRNA gene pairs among HAdVs pneumonia group v.s. healthy control group(a) and severe group v.s. mild group(b) was analyzed using Cytoscape software. Squares, triangles and circles represent circRNA, miRNA and target genes, respectively. Orange nodes represent genes that are upregulated in severe HAdVs pneumonia, and blue nodes represent downregulated genes.

Figure 8
Hsa_circ_0002171 may be potential indicator of HAdVs pneumonia and severe cases. Receiver operating characteristic (ROC) analysis of the ability of hsa_circ_0002171 to predict HAdVs pneumonia (a) and severe cases (b). The cut-off point of hsa_circ_0002171 as well as their sensitivity and speci city for prediction HAdVs pneumonia and severe cases(c).