Feedforward loop profile among transcription factor,miRNA and mRNA in influenza

Yanyan Bao ,Yujing Shi ,Yingjie Gao ,Lirun Zhou ,Xiaolan Cui*

1Biosafety Laboratory,Institute of Chinese Materia Medica,China Academy of Chinese Medical Sciences,Beijing,100700,China.

Abstract Purpose:In the present study,we focused on the 46 microRNAs and 719 genes in the microRNA-gene network,reported by us,and aimed to build a research blueprint of feedforward loops and reveal the key TFs in H1N1-infected mouse lung.Method:Based on microRNAs and genes in the microRNA-gene network previously reported by us,we used Jemboss software to find relationships between TFs and microRNAs (or genes),and then built a TF-microRNA-gene network exploiting the interactions between TFs and microRNAs (or genes).Next,we searched the sequences of above genes or microRNAs near the transcription start site (TSS)area,and then used the MatchTM algorithm to predict relevant TFs,and built the TF-Gene-Network.Result:We built a TF-microRNAgene network and exploreed eight key TFs,namely NF-AT1,GKLF,SRY,SOX10,AML1,MZF1,CRX and myogenin,in the network,and then constructed subgraphs of these eight TFs.Simultaneously,we predicted the possible target genes of microRNAs and identified the feedforward regulation relationship of possible TFs,microRNAs and mRNAs.The results showed that all eight factors with a score greater than 100 were TFs,namely NF-AT1,GKLF,SRY,SOX10,AML1,CRX,myogenin and MZF1.We then constructed subtables of the above eight TFs.Conclusion:In this study,TFs including NF-AT1,GKLF,SRY,SOX10,AML1,MZF1,CRX and myogenin showed the highest score (>100)not only in the TF-microRNA-gene network but also in feedforward loops,indicating that these eight TFs play the most important roles in mouse H1N1 influenza virus infection biology.

Keywords:H1N1;Influenza A virus;Transcription factor;MicroRNA;Gene

Introduction

Influenza A viruses can cause outbreaks or epidemics of local infections worldwide.Influenza A viruses infection have high mortality.Even in years without mass outbreaks,influenza A viruses could lead to more than half a million deaths each year [1].Based on the antigenicity of the surface structural proteins,influenza A viruses are divided into different subtypes,such as H1N1,H5N1 and H3N2.H1N1 influenza viruses infect not only humans but also livestock,and could cause respiratory disease including mild infection and severe pneumonia.In 2009,human infection with a novel H1N1 influenza virus of swine origin was reported and rapidly spread worldwide,which WHO declared be the first influenza pandemic in more than 40 years [2].

MicroRNA is involved in many biological processes and is an important factor for maintaining biological balance [3-7].MicroRNAs inhibit the expression of genes by complementing the sequence of the target gene or causing translational blockade [8].Not only microRNA but also its target genes can be regulated by transcription factors (TFs).A feedforward loop refers to a network motif in which TFs and microRNAs coordinately regulate target genes.Feedforward loops play increasingly important roles in living organisms[9-10]and understanding the feedforward loop profile caused by H1N1 virus is important to finding out H1N1 virus infection biology which helps predict and control future outbreaks.

In this study,based on the microRNAs and genes in the microRNA-gene network we have previously reported [11-12],we built a TF-microRNA-gene network and predicted 1072 feedforward loops of TFs,microRNAs and genes through the TRANSFAC database.TFs including NF-AT1,GKLF,SRY,SOX10,AML1,MZF1,CRX and myogenin showed the highest score (>100)not only in the TF-microRNA-gene network but also in feedforward loops,indicating that these eight TFs play the most important roles in mouse H1N1 influenza virus infection biology.

1 Materials and methods

1.1 H1N1 influenza virus

H1N1 influenza virus (A/Puerto Rico/8/34,PR8)(ATCC,Manassas,VA)was used in this study.PR8 virus was passaged with 10-day-old egg embryos.The 50% tissue culture infectious dose (TCID 50)was detected in MDCK cells and calculated by the method developed by Reed and Muench [13].All experiments were performed in an ABSL-2 biosafety laboratory.

1.2 Mouse pneumonia model induced by PR8 virus

Male and female mice (ICR,13-15g)were used for the infection experiments,mice were purchased from Beijing Vital River Laboratory Animal Technology Co.,Ltd (Beijing,China).Mice was infected with PR8 virus as previously [14].Briefly,the mice of 2-day and 5-day groups (n=12/group)were intranasally inoculated with 104 TCID50 PR8 viruses after anesthetized with ether,while normal control group (n=12)was intranasally inoculated with physiological saline.Mice were executed and lungs were harvested after weighing.The lung index was used to evaluate the lung damage,lung index (%)=(lung weight/body weight)× 100%.

1.3 RNA of mouse lung isolation

Mouse lungs were homogenized in liquid nitrogen.Total RNA was extracted from mouse lungs using the Qiagen miRNeasy Mini Kit (Qiagen,Hilden,Germany).

1.4 Histopathology

Lungs were fixed in 10% formalin fixative,dehydrated in ascending ethanol concentrations,embedded in paraffin,sectioned into 4μm slices,and stained with hematoxylin and eosin.Histopathology photos were taken using a phase inverted microscope (Olympus,Japan).Histopathological result was used to evaluate the lungs damage.

1.5 Microarray analysis of microRNA and mRNA

Total RNA was processed with microRNA and mRNA microarray analysis,using Affymetrix miRNA 4.0 and Affymetrix GeneChip Mouse 1.0 (Affymetrix,USA).There were three pools per group;each pool included four mice.The random-variance model (RVM)F-test was applied to select differentially expressed genes for the control and infected groups,because the RVM F-test is an efficient analysis method in small sample cases.After significance and false discovery rate analyses,differentially expressed genes were filtered according to their p-value threshold [15-17].

1.6 TF-microRNA-gene network

TF analysis illustrates how TFs regulate genes.Based on microRNAs and genes in the microRNA-gene network previously reported by us [11-12],we used Jemboss software to find relationships between TFs and microRNAs (or genes),which were determined by counting correlations between gene or microRNA sequences and TF sequences.Next,we built a TFmicroRNA-gene network using the interactions between TFs and microRNAs (or genes).TF was the network's core and had the biggest score [18-19].

1.7 Feedforward loop analysis

We first searched the sequences of above genes or microRNAs near the transcription start site (TSS)area,and then used the MatchTMalgorithm to predict relevant TFs,which used a library of position weight matrices collected in the TRANSFAC® database and therefore enabled us to search for a wide variety of different TF binding sites [20].The search algorithm used two score values:the matrix similarity score (mSS)and the core similarity score (cSS)to evaluate the results,which ranged from 0.0 to 1.0,with 1.0 meaning an exact match.MSS and CSS,are calculated using the same formula (see below).Whereas MSS is calculated using all positions of the matrix,CSS is calculated using the core positions only.In this stuty,we select uniform MSS and CSS cut-offs >0.7 that applied to all matrices.Next,we built the TF-Gene-Network using the interactions among genes and TFs,and set cut-offthresholds for the mSS and cSS.The network's core TF was the most important factor and had the biggest score[18].

The mSS (as well as the cSS)for a subsequence x of lengthLwas calculated in the following way:

fi,B,frequency of nucleotide B to occur at position i of the matrix (B{A,T,G,C})

fimin,frequency of the nucleotide which is rarest in position i in the matrix

fimax,highest frequency in position i.

The information vector describes the conservation of position i in a matrix [21].

1.8 Statistical analysis

The lung index was expressed as the mean ± standard deviation.Statistical analysis was performed using ANOVA for multiple comparisons andp＜ 0.05 was considered statistically significant.

2 Results

2.1 Mouse pneumonia model induced by PR8 virus

Mice were anesthetized and infected as previously described.The lung index and histopathology result both showed when mice were infected with 104 TCID50 of PR8 viruses,the mortality rate was zero,but weight loss and lung damage increased with time[11-12].

2.2 TF-microRNA-gene network

IBased on the microRNAs and genes in the microRNAgene network,which we have previously reported[11-12],we predicted related TFs and built a TFmicroRNA-gene network (Figure 1).At the same time,the key TFs with core regulation could be isolated by quantitative calculation,which provides a possible explanation for the mechanism of discovering gene co-expression regulation.The name and score of each gene,TF or microRNA in the TF-microRNA-gene network are displayed in Table S1,and the results showed that all eight factors with a score greater than 100 were TFs,namely NF-AT1,GKLF,SRY,SOX10,AML1,MZF1,CRX and myogenin,in order of score (Supplementary Table 1).We then constructed subgraphs of these eight TFs (Figure 2A-H).

2.3 Feedforward loop relationship

MicroRNAs can affect gene expression by modulating target genes,and most of the cell-specific expression patterns of microRNAs are mediated by TFs at the transcriptional level.TFs can not only directly regulate target genes,but also indirectly regulate target genes through microRNAs,thereby forming a feedforward loop among TFs,microRNAs and genes.In this study,we used the TRANSFAC database in Biobase to predict TFs that regulate microRNA precursors and genes.Using the MatchTM algorithm,we predicted possible TF binding sites within a region of the TSS adjacent to the gene or microRNA precursor.At the same time,we predicted the possible target genes of microRNAs and identified the feedforward regulation relationship of possible TFs,microRNAs and mRNAs.We filtered 1072 feedforward loop relationships of TFs,microRNAs and genes (Supplementary Table S2),and recorded the names and scores of genes,TFs or microRNAs involved in these feedforward relationships(Supplementary Table S3).The results showed that all eight factors with a score greater than 100 were TFs,namely NF-AT1,GKLF,SRY,SOX10,AML1,CRX,myogenin and MZF1,ordered by score (Table 2).We then constructed subtables of the above eight TFs(Table 3-10)to provide prioritized feedforward loops(Figure 3A-H).

3 Discussion

Due to the high spread and wide host range of influenza A viruses,they can cause local outbreaks and global pandemics.As such,they are among the major causative agents of human disease and death and it is therefore very important to study the biological mechanism of influenza virus infection.We have been engaged in related research in this area,especially that involving up-and downstream regulation of microRNA during influenza virus infection.H1N1 virus is a major pathogenic subtype of influenza A virus and causes respiratory infections in humans and livestock that range from mild infection to severe pneumonia associated with acute respiratory distress syndrome.We have previously reported microRNA expression profiles and networks,and also dynamic gene expression in H1N1-infected mouse lungs [11-12].

We had reported that 46 microRNAs and 719 related target genes were differentially expressed.Focusing on them,we built a microRNA-gene network [11-12].In the present study,we focused on the 46 microRNAs and 719 genes in the microRNA-gene network and aimed to build a research blueprint of feedforward loops and reveal the key TFs in H1N1-infected mouse lung.Jemboss software was used to identify the relationship between TFs and microRNAs (or genes),which was determined by counting the correlations between gene or microRNA sequences and TF sequences.Following this,we built a TF-microRNA-gene network with the interactions between TFs and microRNAs (or genes).Eight key factors (score >100)were filtered,and the eight factors meeting this criterion were TFs,namely

NF-AT1,GKLF,SRY,SOX10,AML1,MZF1,CRX and myogenin.We then constructed subgraphs of the above eight TFs.The TRANSFAC database was used to predict TFs that regulate microRNA precursors and genes.Using the MatchTM algorithm,we predicted possible TF binding sites within a specified region of the TSS adjacent to the gene or microRNA precursor.At the same time,we predicted possible target genes of microRNAs and found the feedforward regulation relationship of possible TFs,microRNAs and genes.One thousand and seventy-two feedforward loops were built and eight key factors with a score greater than 100 were filtered,namely NF-AT1,GKLF,SRY,SOX10,AML1,CRX,myogenin and MZF1.

?

?

?

?

?

?

?

?

?

?

Figure 1 TF-microRNA-gene network

Figure 2 TF-microRNA-gene subgraphs

In summary,TFs including NF-AT1,GKLF,SRY,SOX10,AML1,MZF1,CRX and myogenin showed the highest score (>100)not only in the TFmicroRNA-gene network but also in feedforward loops,indicating that these eight TFs play the most important roles in mouse H1N1 influenza virus infection biology.These results provide a research blueprint for microRNA transcriptional regulation research in mouse H1N1 influenza virus infection and a prelude to advancements in mouse H1N1 influenza virus infection biology.In future research,we will focus on these eight transcription factors to study the relevant mechanisms of influenza virus infection.