Circulating MicroRNAs in Plasma of Hepatitis B e Antigen Positive Children Reveal Liver-Specific Target Genes

Background and Aim. Hepatitis B e antigen positive (HBeAg-positive) children are at high risk of severe complications such as hepatocellular carcinoma and cirrhosis. Liver damage is caused by the host immune response to infected hepatocytes, and we hypothesise that specific microRNAs play a role in this complex interaction between virus and host. The study aimed to identify microRNAs with aberrant plasma expressions in HBeAg-positive children and with liver-specific target genes. Methods. By revisiting our previous screen of microRNA plasma levels in HBeAg-positive and HBeAg-negative children with chronic hepatitis B (CHB) and in healthy controls, candidate microRNAs with aberrant plasma expressions in HBeAg-positive children were identified. MicroRNAs targeting liver-specific genes were selected based on bioinformatics analysis and validated by qRT-PCR using plasma samples from 34 HBeAg-positive, 26 HBeAg-negative, and 60 healthy control children. Results. Thirteen microRNAs showed aberrant plasma expressions in HBeAg-positive children and targeted liver-specific genes. In particular, three microRNAs were upregulated and one was downregulated in HBeAg-positive children compared to HBeAg-negative and healthy control children, which showed equal levels. Conclusion. The identified microRNAs might impact the progression of CHB in children. Functional studies are warranted, however, to elucidate the microRNAs' role in the immunopathogenesis of childhood CHB.


Introduction
Children with chronic hepatitis B (CHB) have a lifetime risk of developing hepatocellular carcinoma (HCC) up to 25% and an incidence of cirrhosis of 2-3% per year [1,2]. It is widely accepted that the natural course of CHB is determined by the host-virus interaction; however, the exact mechanisms responsible for disease progression in children are not fully understood.
Evidence suggests that microRNAs play a role in the complex interaction between the hepatitis B virus and host [3]. Our group recently identified a panel of 16 microRNAs aberrantly expressed in plasma of children with CHB and suggested a potential role of these microRNAs in the pathogenesis of childhood CHB [4].
Risk of progressive liver disease primarily applies to hepatitis B e antigen positive (HBeAg-positive) children and seroclearance of HBeAg is a key event in the natural course of disease [5]. Most children who undergo HBeAg seroconversion are defined inactive carriers, with absent or low viral replication, and usually inactive liver histology [5]. Inactive carriers with no signs of cirrhosis at seroconversion do not show disease progression over long-term follow-up (24-29 years) [6][7][8]. In the present study we hypothesise that microRNAs aberrantly expressed in plasma of HBeAgpositive children might be involved in the development of progressive liver disease.
It is well known that specific host factors in the liver tissue are tightly regulated in patients with CHB [9]. MicroRNAs were free from known medical conditions. Blood samples were obtained prior to anaesthesia. Parents of all participants provided informed written consent prior to any study procedure.

Blood
Samples. Blood samples were processed as previously described [4]. Briefly, blood samples were collected in EDTA tubes, centrifuged at 2,500 g for 10 minutes at room temperature, separated, aliquoted, and stored at −80 ∘ C until further use.
International Journal of Hepatology 3

Screen of Plasma MicroRNA Levels in Children with CHB and in Healthy Controls.
Recently, our group published an initial screen of plasma microRNA levels in HBeAgpositive, HBeAg-negative, and healthy children [4]. Briefly, microRNA polymerase-chain-reaction (PCR) panels (human panel I and II V2.M/R), miRCURY LNA Universal RT PCR system (Exiqon, Vedbaek, Denmark) were employed to measure plasma levels of 739 human microRNAs in a total of three samples: one sample contained plasma from 10 HBeAg-positive children, one sample contained plasma from 10 HBeAg-negative children, and one sample contained plasma from 10 healthy controls. Analyses were performed per Exiqon's instructions. According to the manufacturer, the microRNAs covered in the microRNA PCR panels are generally higher expressed and more likely differentially expressed in disease, or more often cited in the literature.

Selection of Candidate MicroRNAs.
For selection of candidate microRNAs we followed stringent criteria and included only data from HBeAg-positive and HBeAgnegative children. Firstly, based on raw data, microRNAs with values above 35 in HBeAg-positive and/or HBeAgnegative children were excluded. Secondly, raw data were normalised using three differentapproaches: global mean, U6, and geometric mean of miR-22-5p, -26a-5p, and 221-3p as previously described [4]. By using the comparative method, the relative expression of each microRNA between the groups was calculated [16]. We focused on microRNAs aberrantly expressed in HBeAg-positive children and identified the top up-and downregulated microRNAs for further analyses.

Bioinformatics Analysis.
Bioinformatics analysis was performed on all candidate microRNAs identified by revisiting our previous screen. Our analysis included retrieval of microRNA target genes with CLIP-Seq (cross-linking immunoprecipitation-high-throughput sequencing) overlap from starBase (sRNA target Base, release 2.1) and subsequently liver-specificity filters were applied on the retrieved microRNA target genes with CLIP-Seq overlap.

Retrieval of MicroRNA Target Genes with CLIP-Seq
Overlap. MicroRNA target genes were retrieved using the starBase, a database that allows a comprehensive exploration of microRNA-target gene interaction maps from CLIP-Seq and Degradome-Seq data [17]. The predicted microRNAtarget gene interactions in starBase are processed from five target prediction tools (TargetScan, PicTar, PITA, miRanda, and RNA22) overlapping with the CLIP-Seq data. Only those target genes that intersected the CLIP-Seq data sets with a biological complexity ≥2 (a measure of reproducibility between biological replicates or experiments to further reduce false positives) were retrieved.

Liver-Specificity Filter for MicroRNA Target Genes.
Human liver-specific genes were retrieved from the TiGER (Tissue Specific-Gene-Expression and Regulation, version 1.0) [18] and TiSGeD (Tissue-Specific Genes Database) [19] databases. The TiGER database encompasses human tissuespecific gene expression profiles or expressed sequence tag (EST) data, cis-regulatory module (CRM) data, and combinatorial gene regulation data for interacting transcription factor (TF) pairs.
TiGER data is based on analyses of 30 human tissues identifying tissue-specific genes (specificity is determined by expression enrichment scores and associated −log10 ( value)), TFs and CRMs. A gene is defined as tissue-specific if it satisfies the following two conditions: enrichment score >5 and value < 10 −3.5 . In case of the TiSGeD, tissuespecific genes are based on biomedical literature and data mining of gene expression profiles for over 100 human tissues. In TiSGeD, relative tissue-specificity of a gene is based on a statistical parameter, SPM, to quantitatively measure the specificity of a gene over tissues. An SPM parameter is a sensitive indicator in quantitative estimation of gene expression patterns. SPM ranges from 0 to 1.0. A value close to 1.0 indicates high tissue specificity of a gene.
From the TiGER database, liver-specific genes were retrieved: 383 genes based on ESTs (309 nonredundant genes); 300 CRM detections (105 nonredundant genes); and 160 TF pairs coregulating in liver (96 nonredundant genes). Two-hundred-and-fifty liver-specific genes were retrieved from the TiSGeD database that included human liver, fetal liver, hepatoma, and HepG2 specific genes. In total, 542 liverspecific genes were retrieved from these resources.
Liver-specific target genes were identified for the microR-NAs of interest by comparing all the target genes with four liver-specific gene lists based on TiGER (EST, CRM, TF) and TiSGeD. A target gene was deemed to be liver-specific if it was present in any of the four liver-specific gene lists.

Validation of Candidate MicroRNAs.
Candidate micro-RNAs targeting liver-specific genes were selected for further validation by quantitative real-time PCR (qRT-PCR). Candidate microRNAs were individually quantified by standard qRT-PCR using total RNA extracted from plasma of 34 HBeAg-positive, 26 HBeAg-negative, and 60 healthy controls.
2.11. RNA Extraction. Total RNA was extracted from plasma using the miRNeasy mini kit (Qiagen, Hilden, Germany) as previously described [4]. RNA was stored at −80 ∘ C until further use.
2.12. cDNA Synthesis. cDNA synthesis was performed using the Universal cDNA Synthesis kit (Exiqon, Vedbaek, Denmark) as previously described [4] and cDNA was stored at −20 ∘ C.
Resultant cDNA was diluted ×50 and assayed in 10 L PCR reactions according to the protocol for miRCURY LNA Universal RT microRNA PCR (Exiqon, Vedbaek, Denmark). Negative controls with no template from the reverse transcription reaction were included and profiled like the samples. Thermal cycling was performed on a CFX384 Real-Time thermal cycler (Biorad, Hercules, California, USA) in 384 well plates as per Exiqon's instruction.
(max) was set to 40 amplification cycles. Analyses were run in triplicate.
2.14. Analysis of qRT-PCR Data. Quality control of the qRT-PCR was performed as previously described [4,20]. Raw data were normalised using the geometric mean of miR-22-3p, -26a-5p, and -221-3p [4,21,22] and the comparative method was used to analyse the data [16]. Statistical analyses were performed as previously described [4], by using SAS software, version 9.2 (SAS Institute, Cary, NC, USA). Briefly, statistical significances were determined using Mann-Whitney test and correlation analyses were performed in two steps, both of which used analysis of variance on ranks ( values from Chi-Squared tests). Due to multiple testing only < 0.0028 was regarded as significant (Bonferroni correction).
Furthermore, the remaining ten microRNAs had no liverspecific target genes and did not pass the liver-specificity filter. For these microRNAs, the total numbers of non-liverspecific target genes with CLIP-seq overlap retrieved from the starBase were 1378 (833 nonredundant). Only those target genes predicted by two or more target prediction tools are presented: in total 522 target genes of which 299 are nonredundant (Table S1 in
To further investigate the microRNA expression patterns of children with CHB we included data from 57 healthy controls in the statistical analyses. Of the six microRNAs with significantly higher plasma levels in HBeAg-positive children than in HBeAg-negative children, two microRNAs (miR-28-5p and miR-30a-5p) showed similar expression levels in plasma from HBeAg-negative and healthy children, > 0.004, whereas four microRNAs (miR-30e-3p, 378a-3p, 574-3p, and let-7c) showed lowest expression in plasma from healthy controls, < 0.001 (Figure 1). miR-654-3p that showed lower expression in HBeAgpositive children compared to HBeAg-negative children was expressed at same levels in plasma from HBeAg-negative and healthy control children, = 0.52 (Figure 1). Levels (fold changes) of the eight microRNAs in plasma from HBeAg-positive and HBeAg-negative children were calculated relative to the plasma levels in healthy controls (Table 3).

MicroRNA Plasma Levels and Virological Parameters.
We examined the correlations between plasma microRNA levels and the virological parameters: viral load and HBsAg quantity. Only children with CHB were included in these analyses. Firstly, focusing on univariate analyses, we observed strong positive correlations between plasma levels of six microRNAs (miR-28-5p, -30a-5p, -30e-3p, -378a-3p, -574-3p, and -let-7c) and with both viral load and HBsAg quantity, < 0.001. All six of these microRNAs showed significantly higher plasma levels in HBeAg-positive children when compared to HBeAg-negative children.
Secondly, multivariate analyses were performed taking into account the effect on plasma microRNA levels of the different virological parameters: HBeAg-status, viral load, and HBsAg quantity. Interestingly, the strongest correlations were observed between plasma microRNA levels and HBsAg quantity. The correlations were significant for four microR-NAs (miR-28-5p, -30e-3p, -574-3p, and -let-7c), < 0.001, all of which showed significantly higher levels in HBeAgpositive children when compared to HBeAg-negative children ( Figure 2).

Discussion
HBeAg-positive children are especially vulnerable to developing HCC and cirrhosis, and we hypothesise that specific microRNAs impact the progression of CHB. This study is the first to identify microRNAs with aberrant plasma expressions specifically in HBeAg-positive children and with liver-specific target genes.
For miR-122, known to be highly expressed in liver tissue [35], we could not identify liver-specific target genes due to nonavailability of CLIP-seq data for a number of microR-NAs, including mir-122-3p. Additionally, the application of a liver-specificity filter excluded candidate microRNAs with substantial functions in tissues other than the liver.
The other interesting microRNA expression profile observed in the present study was for miR-654-3p. miR-654-3p was downregulated in HBeAg-positive children when compared to HBeAg-negative and healthy control children, which showed equal levels. In line with our finding, miR-654-3p has an inhibitory role in the H1N1 Influenza A virus [52] and acts as a tumor suppressor in prostate cancer [53]. Two liver-specific target genes, CPOX and SF1, are targeted by miR-654-3p. CPOX encodes the enzyme Coproporphyrinogen-III oxidase that has a role in the haem and chlorophyll biosynthetic pathways [54] whereas SF1 encodes Splicing Factor 1 that is involved in the formation of the spliceosome complex [55].
We propose that these microRNAs identified with aberrant plasma expressions specifically in HBeAg-positive children and with liver-specific target genes are biomarkers for disease progression and might impact the development of HCC, and perhaps also cirrhosis, in children with CHB. Functional studies are warranted, however, to further elucidate these microRNAs' role in the immunopathogenesis of childhood CHB.
interventions has shown great promise [11], and the first artificial microRNA antagonist is already in clinical trial [56]. MicroRNAs with a biological function in the immunopathogenesis of childhood CHB might serve as targets for the development of new antiviral treatment strategies. Whether the microRNAs identified in the present study would have the potential to serve as targets in future therapeutics needs to be investigated.
In conclusion, we identified 13 microRNAs with aberrant plasma expressions of HBeAg-positive children and with liver-specific target genes. In particular, we observed two distinct microRNA expression patterns: three microRNAs were upregulated and one was downregulated in HBeAgpositive children compared to HBeAg-negative and healthy control children, which showed equal levels. Five liverspecific target genes were identified for the four microRNAs.
The microRNAs might be biomarkers for disease progression in children with CHB. Further studies are needed to elucidate the microRNAs' role in childhood CHB, hopefully leading to the identification of future therapeutic targets and to enhanced management of childhood CHB.