Hepatitis C Virus Mediated Changes in miRNA-449a Modulates Inflammatory Biomarker YKL40 through Components of the NOTCH Signaling Pathway

Liver disease due to hepatitis C virus (HCV) infection is an important health problem worldwide. HCV induced changes in microRNAs (miRNA) are shown to mediate inflammation leading to liver fibrosis. Gene expression analyses identified dysregulation of miRNA-449a in HCV patients but not in alcoholic and non-alcoholic liver diseases. By sequence analysis of the promoter for YKL40, an inflammatory marker upregulated in patients with chronic liver diseases with fibrosis, adjacent binding sites for nuclear factor of Kappa B/P65 and CCAAT/enhancer-binding protein alpha (CEBPα) were identified. P65 interacted with CEBPα to co-operatively activate YKL40 expression through sequence specific DNA binding. In vitro analysis demonstrated that tumor necrosis factor alpha (TNFα) mediated YKL40 expression is regulated by miRNA-449a and its target NOTCH1 in human hepatocytes.NOTCH1 facilitated nuclear localization of P65 in response to TNFα. Further, HCV patients demonstrated upregulation of NOTCH1 along with downregulation of miRNA-449a. Taken together it is demonstrated that miRNA-449a plays an important role in modulating expression of YKL40 through targeting the components of the NOTCH signaling pathway following HCV infection. Therefore, defining transcriptional regulatory mechanisms which control inflammatory responses and fibrosis will be important towards developing strategies to prevent hepatic fibrosis especially following HCV recurrence in liver transplant recipients.


Introduction
Liver diseases resulting from hepatitis C virus (HCV) infection is a major health issue worldwide as well as the United States [1,2]. It is estimated that about 4 million people are infected with HCV in the United States and about 300 million worldwide [1]. The natural history of HCV infection in the liver is characterized by slow progression to fibrosis and cirrhosis, end-stage liver diseases, and high risk of developing hepatocellular carcinoma (HCC) [3]. YKL40 (CHI3L1) is a member of the ''mammalian chitinaselike proteins,'' secreted by activated macrophages and neutrophils during inflammation in various tissues including liver, smooth muscle and cancer cells [4]. YKL40 is elevated in patients with chronic liver diseases that are characterized by inflammation and increased extra-cellular remodeling [5,6]. Although increased levels of YKL40 have been shown to be induced by tumor necrosis factor alpha (TNFa), the molecular mechanisms are not clearly identified [7]. TNFa, an inflammatory cytokine regulates gene expression in the nuclear factor of Kappa B (NFKB) signaling pathway [8]. Components of the mammalian NFKB family of transcription factors includes NFKB1 (P105/P50), NFKB2 (P100/ P52), RelA (P65), RelB and c-Rel [9]. The NFKB component P65 is a multimeric DNA binding transcription factor involved in inflammatory and immune disorders especially autoimmune diseases and cancer [10]. NOTCH1 is one of the upstream regulator of NFKB complex and downregulation of NOTCH1 impairs its function [11,12]. It has been shown that NOTCH1 and TNFa regulate nuclear retention of NFKB [13,14]. CCAAT/ enhancer-binding protein alpha (CEBPa) is a homodimeric DNA binding bZIP transcription factor that controls cell proliferation and differentiation [15]. CEBPa is differentially regulated in cases of HCC and targets expression of a wide range of genes and microRNAs (miRNA) involved in liver diseases [16,17].
miRNAs have been shown to play an important role in immune evasion, regulation of cell cycle and in cancer progression [18,19,20]. HCV infection results in modulation of miRNA particularly those that control viral particle entry and propagation, thus playing an important role in host immune evasion [21]. In this study we defined the molecular mechanisms of YKL40 expression that involves HCV induced miRNA modulation and regulation by novel pathways including NOTCH1, NFKB and CEBPa.

Patients
Liver biopsies were obtained from 10 chronic HCV patients, 10 alcoholic hepatitis patients, 10 non-alcoholic steatohepatitis patients and 10 normal donor livers (control) at the time of transplantation at Washington University Medical Center/Barnes-Jewish Hospital (Table 1). Patients with hepatitis B virus and/or HIV were excluded from the study. All of the human studies were approved by the human research protection committee at Washington University (protocol 201104075) and patients were enrolled after written informed consent was obtained.

miRNA and mRNA Expression Analysis
Total RNA was isolated from the liver biopsies or hepatocytes using the RNAaqueous kit (Ambion, NY). Expression level of miRNA-449a was determined using the TaqManH MicroRNA assays and TaqManH Universal Master Mix II (Life technologies, NY) using predesigned primers. Quantitative PCR (qPCR) to analyze YKL40 and NOTCH1 was performed using a BioRad Real-Time PCR System with cycling conditions of 95uC for 10 min followed by 95uC for 15 sec and 60uC for 60 sec for a total of 40 cycles. Each TaqMan assay was run in triplicate. The DDCt value was calculated by normalizing the threshold (CT) values with GAPDH expression and respective gene expression in controls.

Primary Hepatocytes and HEPG2 Cell Line Transfection
Primary human hepatocytes were purchased from Life Technologies, New York and grown in 24 well plates in Williams Medium E supplemented with 5% FCS, 100 units/mL penicillin, 100 units/ml Amphotericin, 0.1% Albumin, 300 nM insulin, 2 mM L-glutamine and 0.1 nM Hydrocortisone. HEPG2 cells were grown in RPMI with 10% FBS, 1 mM sodium-pyruvate, 10 mM HEPES, 2 mM L-glutamine, and 100 units/mL penicillin/streptomycin. 20 ng/ml of TNFa (Sigma, St. Louis, MO) was added for 6 hours wherever indicated. The optimal concentration of TNFa was determined by dose-dependent analyses.
For transfection of hepatocytes and HEPG2 cells (ATCC), 0.2610 5 cells were seeded into each well of a 24 well plate and grown for 24-48 hours. On the day of transfection, medium was changed and 500 ng of DNA was transfected using Lipofectamine TM LTX and Plus Reagent (Invitrogen, NY). For siRNA delivery 0.1610 5 cells were grown in each well of a 24 well plate for 24-48 hours in antibiotic free medium and 80 pico moles of siRNA were transfected using Lipofectamine TM RNAiMAX (Invitrogen). Cells were harvested 48 hours post transfection and efficiency was measured by qPCR, immunostaining and western blot.

Luciferase Assay
Human primary hepatocytes (1610 5 ) were transfected in 24 well plates as mentioned earlier with 1 mg pGL4.11 luciferase reporter vector or pGL4.11 driven by the YKL40 promoter or deletion constructs. For miRNA regulation studies the reporter construct was transfected in combination with 1 mg of either control vector or vector expressing miRNA-449 precursor. For NOTCH1 regulation studies the reporter construct was transfected in combination with 80 picomoles of either non specific siRNA or siRNA specific for NOTCH1. For transcription factor studies the reporter constructs were transfected in combination with 2 mg of empty pcDNA3 vector, or pcDNA3 expressing P65 or CEBPa or HCV Viral load (X10 6 /mL) 1.460.3 n/a n/a n/a HCV genotype 1 4 n/a n/a n/a % 21.5% n/a n/a n/a 1a 4 n/a n/a n/a % 30.8% n/a n/a n/a 1b 2 n/a n/a n/a % 15.4% n/a n/a n/a

Chromatin Immunoprecipitation
Chromatin Immunoprecipitation was carried out with ChIP-IT TM Express (Active Motif, Carlsbad, CA). Briefly, hepatocytes (1610 6 ) were crosslinked with 1% formaldehyde and quenched with 0.375 M glycine. Nuclei were isolated and sonicated in 350 ml of shearing buffer to prepare chromatin extracts. 1 mg of antibodies for control IgG, P65 or CEBPa were added to 60 ml of sheared chromatin along with Protein G Magnetic Beads. Antibody-lysate mix was washed and DNA was eluted according to the instructions. YKL40 promoter regions were amplified by PCR.

miRNA-449a is Downregulated in HCV Patients
A genomewide miRNA analysis in liver biopsies obtained from chronic HCV infected patients demonstrated a distinct expression profile when compared to the normal liver. Particularly, a significant downregulation of microRNA-449a was observed in the HCV infected livers.
To analyze the specific role of miRNA-449a following HCV infection, biopsies were obtained from 10 chronic HCV patients, 10 alcoholic hepatitis patients, 10 non-alcoholic Steatohepatitis (NASH) patients and 10 control normal donor livers at the time of liver transplant. Total RNA was isolated from the liver biopsies and expression level of miRNA-449a was determined by qPCR using specific primers and the results were normalized to GAPDH expression. Expression analysis demonstrated that miRNA-449a is downregulated more than two fold in livers obtained from HCV patients whereas no significant differences in the expression was observed in alcoholic hepatitis patients, NASH patients and Hepatocytes were transfected with an 23000 bp YKL40-promoter driven reporter construct with (+) or without (2) TNFa. Firefly luciferase activity was measured 48 hours after transfection and normalized to a Renilla luciferase internal control. The numbers represent fold-change over control (average of three independent experiments); error bars represent SD. The '*' represents p value,0.05 obtained by a two-tailed t test. (Lower panel): hepatocytes from 2A were immunoblotted with anti-YKL40 with (+) or without (2) TNFa. ACTIN was used as the loading control. B. HEPG2 cells were immunostained with anti-YKL40 antibody without (2) or with (+) TNFa. C. Quantification of the YKL40 immunostaining signal in HEPG2 cells (1B). The numbers represent the average fluorescence intensity of YKL40 (n = 100). D. Essential regions in the YKL40 promoter required for TNFa mediated expression. Hepatocytes were transfected with luciferase reporters driven by deletion constructs of YKL40 promoter (23000 bp, 22000 bp, 21000 bp, 2500 bp, filled black bars on left) construct with (+) or without (2) TNFa. Firefly luciferase activity was measured 48 hours after transfection and normalized to a Renilla luciferase internal control. The luciferase activity was normalized to the control empty luciferase vector and the numbers represent fold-change over control (average of three independent experiments); error bars represent SD calculated from three independent experiments. doi:10.1371/journal.pone.0050826.g002  Figure 1A). This suggests that miRNA-449a is specifically downreguated in patients with liver diseases following HCV infection.

YKL40 is Upregulated in HCV Patients with Fibrosis
To determine the expression level of YKL40 in liver diseases, biopsies were obtained from 10 chronic HCV patients, 10 alcoholic hepatitis patients, 10 NASH patients and 10 control normal donor livers at the time of liver transplant. Total RNA was isolated from the liver biopsies and expression level of YKL40 was determined by qPCR using specific primers and the results were normalized to GAPDH expression. Expression analysis showed that YKL40 to be upregulated in alcoholic hepatitis patients (2.4 fold), NASH patients (2.9 fold) and HCV patients (2.6 fold) compared to normals ( Figure 1B). These results demonstrate that YKL40 is elevated in patients with chronic liver diseases which are accompanied by inflammation.

TNFa Regulates the Expression of YKL40 at the Transcriptional Level
We and others have shown that YKL40 expression is elevated in patients with chronic liver diseases accompanied by inflammation [5,6]. In vitro studies have shown that inflammatory cytokines such as TNFa have the ability to induce expression of YKL40 [7]. To study the molecular mechanisms of TNFa mediated regulation of YKL40, 23000 base pairs (bp) of the human YKL40 promoter was cloned upstream of a luciferase reporter gene and the construct was introduced into human hepatocytes. The cells were treated either with or without 20 ng/ml TNFa for 6 hours and luciferase activity was measured. More than four fold increase in the YKL40 promoter activity was observed in cells treated with TNFa compared to untreated (Figure 2A, upper panel). Further, western blot analysis showed increased expression of YKL40 in TNFa treated cells (Figure 2A, lower panel). This reporter gene analysis suggests that TNFa regulates the expression of YKL40 through modulation of upstream transcriptional complexes that interact with the YKL40 promoter.
To further demonstrate that TNFa induces expression of YKL40, human HEPG2 cells were cultured and treated with TNFa. Immunostaining of the HEPG2 cells with anti-YKL40 showed a more than 2 fold increase in expression of YKL40 in the cells treated with TNFa compared to untreated cells ( Figures 2B  and 2C).

Reporter Analysis Identified Essential Regions in the YKL40 Promoter Required for TNFa Regulated Expression of YKL40
Since TNFa regulated the expression of YKL40 at the transcriptional level we hypothesized the possible interaction of TNFa with upstream regulatory complexes of the YKL40 gene. To identify the essential regions for TNFa mediated expression deletion mutants of the YKL40 promoter regions were cloned upstream of a luciferase reporter gene. Sequential deletion mutants of the YKL40 promoter region ( Figure 2D, black filled bars on the left) were introduced into hepatocytes. The cells were treated with or without TNFa and the luciferase activity was measured. Significant increases (.10 fold) in the expression from the YKL40 promoter deletion constructs were observed in the cells treated with TNFa compared to untreated ( Figure 2D). However, deletion of the 21000 to 2500 bp region of the YKL40 promoter impaired TNFa mediated transcriptional induction. This indicates that this region (21000 to 2500 bp) contains binding sites for transcriptional regulatory elements on the YKL40 promoter.

Computational Prediction of Transcription Factors Regulating YKL40 Expression
In order to identify transcription factors that regulate the expression of YKL40 a computational analysis of 24000 bp upstream of the open reading frame was done using the transcription element search system. The program predicted consensus binding sites in the DNA for several transcription factors that included NFKB subunit P65 and CEBPa ( Figure 3A). The software predicted a consensus binding site for p65 (GGAATTTCC) at 2578 bp position at the promoter. Similarly several DNA binding sites for CEBPa (CCAAT) was also predicted throughout the YKL40 promoter. Most of the CEBPa binding sites were concentrated in the 23000 to 22000 bp region of the promoter. Interestingly, two CEBPa binding sites were identified in close proximity to the P65 DNA binding site on the promoter ( Figure 3A) suggesting that these two transcription factors may bind to the YKL40 promoter by forming a transcriptional regulatory complex.

Mutation of NFKB (P65) Binding Site Inhibits TNFa Mediated YKL40 Induction
Computational analysis predicted a putative binding site for NFKB subunit P65 at 2578 bp position of the human YKL40 promoter. Further the YKL40 promoter analysis using the deletion constructs identified 21000 to 2500 bp region to be essential for TNFa mediated induction of YKL40 in hepatocytes. To test whether P65 binding plays a role in TNFa mediated upregulation of YKL40 promoter a 2578 bp wildtype luciferase reporter construct, a 2578 bp construct with point mutations in the P65 binding site (wildtype: GGAATTTCC, point mutation: AGCT-TATCA) and a 2500 bp construct with P65 binding site deleted were prepared. Hepatocytes were transfected with these three reporter constructs and cells were treated either with or without TNFa and Luciferase activity was measured. As expected the wildtype 2578 bp promoter construct with an intact P65 binding site showed two fold increase in transcriptional activity in the presence of TNFa ( Figure 3B). However, point mutation in the P65 binding site or deletion of the P65 binding site (2500 bp) completely abolished TNFa mediated induction of YKL40 ( Figure 3B). Further, we demonstrated that siRNA mediated knockdown of P65 resulted in impairment of transcriptional activation by the YKL40 promoter in TNFa treated cells.

CEBPa in an Upstream Transcription Factor which Activate YKL40 Expression
Computational analysis identified CEBPa as a putative DNA binding factor for transcriptional regulation of YKL40. To test whether CEBPa regulates YKL40 expression an empty vector or numbers represent fold-change over the 2578 wildtype construct without TNFa treatment (average of three independent experiments); error bars represent SD. The P65 binding site mutation is shown in the lower panel. The '*' represents p value ,0.05 obtained by a two-tailed t test. C. CEBPa in an upstream transcription factor to activate YKL40 expression. Hepatocytes were transfected with luciferase reporters driven by deletion constructs of YKL40 promoter (23000 bp, 22000 bp, 21000 bp, 2500 bp, filled black bars on left) along with an empty vector or vector expressing CEBPa and treated with TNFa. Firefly luciferase activity was measured 48 hours after transfection and normalized to a Renilla luciferase internal control. The numbers represent fold-change over the control empty luciferase vector (average of three independent experiments); error bars represent SD. doi:10.1371/journal.pone.0050826.g003 a vector expressing CEBPa was overexpressed in hepatocytes along with deletion mutants of the YKL40 promoter cloned upstream of a luciferase reporter gene. The deletion mutants included 23000 bp, 22000 bp, 21000 bp and 2500 bp of the YKL40 promoter region ( Figure 3C, black filled bars on the left) and the luciferase activity was measured after the cells were treated with TNFa for 6 hours. Significant increases (.9 fold) in the YKL40 expression from the 23000 bp promoter construct was noted when CEBPa was overexpressed compared to control empty vector ( Figure 3C). However, deletion of the 23000 bp to 22000 bp region of the YKL40 promoter impaired CEBPa mediated transcriptional induction of YKL40 as that region contained several putative CEBPa binding sites as shown on our computational analysis ( Figure 3A).

CEBPa Interacts with NFKB/P65 to Bind YKL40 Promoter and Cooperates to Activate Transcription in Hepatocytes
Reporter analysis demonstrated that both P65 and CEBPa regulate expression from the YKL40 promoter. To test whether P65 and CEBPa interact and bind to adjacent consensus sites present on the YKL40 promoter a chromatin immunoprecipitation analysis was done. TNFa treated hepatocytes were fixed with formaldehyde and chromatin extracts were prepared. The DNAprotein complexes were immunoprecipitated with anti-P65 or anti-CEBPa or isotype control IgGs. Chromatin fragments were isolated from the immunoprecipitated DNA-protein complexes and subjected to PCR amplification using primers specific for YKL40 promoter regions as shown in Figure 4A. The PCR products were resolved on agarose gel. Chromatin immunoprecipitation analysis showed that both P65 and CEBPa bind to the YKL40 promoter ( Figure 4B). P65 binds to the 2300 bp to 2700 bp region of the YKL40 whereas no binding was observed with other regions of the promoter. This region encompasses the P65 binding site at 2578 bp position. CEPBa showed binding to several regions of the YKL40 promoter where maximum band intensity was seen at 21900 bp to 22300 bp region that encompasses most of the CEBPa binding sites. Immunoprecipitation with isotype IgGs did not enrich any of these YKL40 promoter regions demonstrating the specificity for P65 or CEBPa. No amplification of DNA was observed in PCR reactions performed with primers specific for ACTIN promoter.
Binding of both CEBPa and P65 to the YKL40 promoter in adjacent DNA binding sites suggests the possibility that they interact with each other. To test this, crude lysates from TNFa treated hepatocytes were prepared and subjected to immunoprecipitation with either isotype control IgG or anti-P65 followed by immunoblotting with anti-CEBPa. CEBPa was co-immunoprecipitated with endogenous P65 whereas no CEBPa was observed with the control IgG ( Figure 4C). To confirm the interaction a reverse co-immunoprecipitaion was done by immunoprecipitation with either isotype control IgG or anti-CEBPa followed by immunoblotting with anti-P65 ( Figure S1). To further demonstrate the interaction between P65 and CEBPa, HEPG2 cells were treated with or without TNFa and co-immunostained with anti-CEBPa and anti-P65. CEBPa primarily localized to the nucleus in both TNFa treated and untreated cells. P65 remained exclusively cytoplasmic in the untreated cells with little to no nuclear localization. In the TNFa treated cells, a significant amount of P65 translocated into the nucleus and co-localized with CEBPa ( Figure 4D).
Immunoprecipitation and co-localization analyses indicated that both P65 and CEBPa interact with each other and bind to their consensus sites on the YKL40 promoter. To determine their role in transcriptional regulation from the YKL40 promoter, hepatocytes were co-transfected with a luciferase reporter construct driven by the YKL40 promoter (23000 bp) with P65 or CEBPa or both. Overexpression of both P65 and CEBPa resulted in significant increase in transcriptional activation of the reporter construct compared to either alone ( Figure 4E, upper panel). Immunoblotting of the cell lysates with anti-P65 and anti-CEBPa confirmed the elevated expression levels of these factors compared to endogenous levels ( Figure 4E, lower panel). This demonstrates that CEBPa cooperates with NFKB to regulate expression from the YKL40 promoter.

NOTCH1 Regulates Nuclear Retention of NFKB/P65 in Response to TNFa
Results presented clearly demonstrates that TNFa mediated regulation of YKL40 is dependent on the NFKB subunit P65. To demonstrate nuclear translocation of P65 in response to TNFa in hepatocytes, cells were treated with or without TNFa and cytoplasmic and nuclear extracts were prepared. Whole cell lysates, the cytoplasmic fraction and the nuclear fraction were subjected to immunoblotting with anti-P65. The expression of P65 was not affected by TNFa as no difference was observed in the whole cell lysate ( Figure 5A, left panel). However, significant cytoplasmic exclusion and nuclear enrichment of P65 was observed in TNFa treated cells compared to untreated cells. One of the essential upstream regulators of NFKB complex is NOTCH1 [11,12]. To analyze the role of NOTCH1, an upstream factor required for TNFa mediated nuclear localization of P65, and its functionality in facilitating downstream gene regulation, hepatocytes were transfected either with scrambled siRNA or siRNA specific for human NOTCH1 and subjected to TNFa treatment. The siRNA mediated knockdown of NOTCH1 was confirmed by immunoblotting with anti-Notch1 ( Figure S2A). Immunoblot using anti-P65 demonstrated that knockdown of NOTCH1 resulted in impairment of TNFa mediated cytoplasmic exclusion and nuclear translocation of P65 ( Figure 5A, right  panel).
To further demonstrate that NOTCH1 is required for TNFa mediated translocation of P65, HEPG2 cells were treated with or . CEBPa interacts with NFKB/P65 to bind YKL40 promoter and cooperates to activate transcription in hepatocytes. A. YKL40 promoter showing binding sites for P65 and CEBPa. The black horizontal bars represent regions amplified by the PCR primers. B. Chromatin was immunoprecipitated with anti-P65 or anti-CEBPa or isotype control IgG from hepatocytes. Segments of the YKL40 promoter (indicated in 4A) were amplified by PCR. The first three lanes show immunoprecipitated chromatin (IP) and the fourth lane show input chromatin (Input). ACTIN promoter amplification is shown as the negative control. C. Co-immunoprecipitation of P65 with CEBPa in hepatocytes. Whole cell lysates were subjected to immunoprecipitation with either rabbit IgG or anti-P65. CEBPa in the cell lysates (Input) and immunoprecipitated complexes (IP) was detected by immunoblotting with anti-CEBPa. P65 was detected by immunoblotting with anti-P65. D. HEPG2 cells were treated with (+) or without (2) TNFa and co-immunostained with anti-CEBPa and anti-P65. E. Hepatocytes were transfected with a luciferase construct driven by the 23000 bp YKL40 promoter in addition to the control vector or vector expressing P65 or CEBPa or both. Firefly luciferase activity was measured 48 hours after transfection and normalized to a Renilla luciferase internal control. The numbers represent fold-change over the control empty vector (average of three independent experiments); error bars represent SD. Bottom panel, expression of P65 and CEBPa was verified by immunoblotting with anti-P65 and anti-CEBPa respectively. doi:10.1371/journal.pone.0050826.g004 without TNFa and probed for P65 localization by immunostaining with anti-P65 ( Figure 5B). In the untreated cells P65 remained exclusively cytoplasmic with little to no nuclear localization observed. In the TNFa treated cells, a significant amount of P65 Figure 5. Notch1 regulates nuclear retention of NFKB/P65 in response to TNFa. A. (Right panel) Hepatocytes were treated with (+) or without (2) TNFa and cytoplasmic and nuclear fractions were extracted. (Left panel) hepatocytes were transfected with either non-specific siRNA or siRNA specific for NOTCH1 and treated with TNFa and cytoplasmic and nuclear fractions were extracted. Whole cell lysates, cytoplasmic and nuclear extracts were subjected to immunoblotting with anti-P65. B. HEPG2 cells were treated with (+) or without (2) TNFa and immunostained with anti-P65. The arrows indicate nuclear localization of P65. C. Quantification of the P65 immunostaining signal in HEPG2 cells (5B). The numbers represent the average fluorescence intensity of P65 (n = 100). D. HEPG2 cells were transfected with either non-specific siRNA or siRNA specific for NOTCH1 and treated with TNFa. The arrows indicate nuclear localization of P65. E. Quantification of the P65 immunostaining signal in HEPG2 cells (5D). The numbers represent the average fluorescence intensity of P65 (n = 100). The '*' represents p value,0.05 obtained by a two-tailed t test. doi:10.1371/journal.pone.0050826.g005 Figure 6. microRNA 449a regulates YKL40 expression by targeting NOTCH1 for silencing. A & B. Hepatocytes were transfected with an empty vector (2) or vector expressing miRNA-449a (+) and expression of NOTCH1 (6A) and miRNA-449a (6B) were determined by Q-PCR. The DDCT value was calculated by normalizing the threshold (CT) values with GAPDH and expression of NOTCH1 and miRNA-449a respectively in controls. The '*' represents p value,0.05 obtained by a two-tailed t test. C. (Upper panel) hepatocytes were transfected with a luciferase construct driven by the YKL40 promoter in addition to the control vector or vector expressing miRNA-449a or non-specific siRNA or siRNA specific for NOTCH1 or siRNA specific for P65 in the presence of TNFa. Firefly luciferase activity was measured 48 hours after transfection and normalized to a Renilla luciferase internal control. The numbers represent fold-change over the control vector (average of three independent experiments); error bars represent SD. (Lower panel) Downregulation of YKL40, NOTCH1 and P65 is verified by immunoblotting with anti-YKL40 or anti-NOTCH1 or anti-P65 respectively. ACTIN is shown as the loading control. D. Hepatocytes were transfected with a luciferase construct driven by the YKL40 promoter in addition to the control vector or vector expressing miRNA-449a construct with (+) or without (2) TNFa. Firefly luciferase activity was measured 48 hours after transfection and normalized to a Renilla luciferase internal control. The numbers represent fold-change over the control vector without TNFa (average of three independent experiments); error bars represent SD. The '*' represents p value,0.01 obtained by a two-tailed t test. doi:10.1371/journal.pone.0050826.g006 translocated into the nucleus and quantification showed more than three fold nuclear abundance of P65 compared to the untreated cells ( Figure 5C). Next, the HEPG2 cells were transfected with either scrambled siRNA or siRNA specific for human NOTCH1 and subjected to TNFa treatment. Immunostaining with anti-Notch1 showed more than four fold knockdown of NOTCH1 by siRNA ( Figure S2B, C). Knockdown of NOTCH1 resulted in impairment of TNFa mediated nuclear translocation of P65 by two fold (Figures 5D and E). TNFa mediated nuclear localization of P65 in cells transfected with scrambled siRNA was not affected. This finding indicates that NOTCH1 acts as an upstream regulatory factor and controls TNFa mediated nuclear translocation of P65.

miRNA-449a Regulates YKL40 Expression by Modulating NOTCH1 Expression
Genomewide microarray analysis in our laboratory followed by miRNA gene expression analysis showed that miRNA-449a is downregulated more than two fold in HCV patients compared to non-HCV liver diseases and normals ( Figure 1A). Computational target prediction of miRNA-449a using Targetscan (Targetscan.org) identified NOTCH1 to be a putative target for translational silencing. To test this, hepatocytes were transfected with either empty vector of vector expressing miRNA-449a and expression of both miRNA-449a and NOTCH1 were determined by qPCR. The results were normalized to GAPDH expression. Increased expression of miRNA-449a resulted in more than two fold downregulation of NOTCH1 ( Figure 6A and B).
Earlier we have shown that NFKB component P65, a protein regulated by NOTCH1, activates YKL40 expression through sequence specific promoter interaction (Figures 3 and 4). It is likely that downregulation of miRNA-449a in HCV infected patients ( Figure 1A) results in activation of NOTCH1/NFKB signaling that leads to upregulation of YKL40 expression. To test whether miRNA-449a regulates YKL40 expression, hepatocytes were transfected with either a control vector or vector expressing miRNA-449a along with an YKL40 promoter-driven luciferase reporter construct. In TNFa treated cells expression of YKL40 is reduced by more than two fold in the presence of miRNA-449a ( Figure 6C, upper panel). qPCR analysis also showed downregulation of YKL40 by increased expression of miRNA-449a ( Figure S3).
Since, computational target analysis did not identify YKL40 to be a direct target for miRNA-449a; results obtained from promoter based reporter analysis suggest that miRNA-449a regulates the expression of YKL40 by silencing components of upstream transcriptional regulatory complexes such as NOTCH1/NFKB. To demonstrate that downregulation of YKL40 expression by miRNA-449a is mediated by silencing NOTCH1, hepatocytes were transfected with either scrambled siRNA or siRNA specific for NOTCH1 along with an YKL40driven luciferase reporter construct and cells were treated with TNFa. siRNA mediated knockdown of NOTCH1 impaired expression from the YKL40 promoter ( Figure 6C, upper panel). We have demonstrated that nuclear P65, regulated by NOTCH1, activates YKL40 expression in response to TNFa. siRNA mediated knockdown of P65 also impaired expression from the YKL40  promoter ( Figure 6C, upper panel). Western blot analysis using anti-YKL40 showed downregulation of YKL40 by expression of miRNA-449a or siRNA mediated knockdown of NOTCH1 or P65 in hepatocytes ( Figure 6C, lower panels). qPCR analysis also confirmed downregulation of YKL40 by miRNA-449a ( Figure S3). Immunoblot analysis using anti-NOTCH1 demonstrated that expression of miRNA-449a resulted in downregulation of NOTCH1. Since, P65 is a downstream factor for NOTCH1, knockdown of P65 did not affect its expression ( Figure 6C, lower panels). This suggests that miRNA-449a regulates expression of YKL40 by modulating the NOTCH1 signaling pathway.
To further determine if TNFa mediated activation of YKL40 is regulated by miRNA-449a, hepatocytes were transfected with either a control vector or vector expressing miRNA-449a along with an YKL40-driven luciferase reporter construct (23000 bp) and cells were treated with or without TNFa. In the cells expressing the empty vector, TNFa induced expression from the YKL40 promoter ( Figure 6D). However, in presence of miRNA-449a this TNFa mediated upregulation of YKL40 was impaired.

NOTCH1 Expression is Upregulated in HCV Patients
To determine whether downregulation of miRNA-449a in HCV infection is accompanied by upregulation of its target NOTCH1, biopsies were obtained from 10 chronic HCV patients, 10 alcoholic hepatitis patients, 10 NASH patients and 10 control normal donor livers at the time of liver transplant. Total RNA isolation followed by qPCR demonstrated NOTCH1 to be significantly upregulated in livers obtained from HCV patients. However, no significant difference in the expression of NOTCH1 was observed in alcoholic hepatitis patients, NASH patients and normal livers (Figure 7). Based on our in-vitro results obtained with hepatocytes, upregulation of NOTCH1 and YKL40 (Figure 1) in HCV patients can be attributed to downregulation of miRNA-449a.

Discussion
YKL40, a member of the mammalian chitinase-like protein, has been shown to be elevated in patients with chronic liver diseases with fibrosis and cirrhosis ( Figure 1) [5,6]. In chronic liver disease patients, YKL40 expression has been shown to have a strong correlation with degree of fibrosis progression, extracellular matrix (ECM) synthesis, and serves an early indicator of liver fibrosis [23,24]. In HCC patients, YKL40 expression is highly elevated in both serum and liver tissue [25]. Here we demonstrate that YKL40 expression in human livers is regulated by co-operative action of several promoter-bound transcription factors. By computational analysis and subsequent in vitro studies we have defined putative binding sites for NFKB subunit P65 and CEBPa in the YKL40 promoter ( Figure 3A). The NFKB pathway plays an important role in liver fibrosis, as its activation in hepatocytes can lead to activation of surrounding tissue macrophages and thus leading to fibrosis [26]. We demonstrated that TNFa mediated YKL40 expression is regulated by P65, a component of the NOTCH/ NFKB signaling pathway (Figures 2 and 3). Additionally our study demonstrates that NOTCH1 is essential for nuclear retention of P65 in human liver cells ( Figure 5). Studies in animal models have shown that knockdown of NOTCH1 resulted in impairment of DNA binding and transcriptional activation ability of P65 and impacted dendritic cell differentiation [12]. We have shown for the first time that the NFKB subunit P65 cooperates with CEBPa to regulate expression of YKL40 through direct DNA binding in hepatocytes (Figures 3 and 4). Several studies have shown that CEBPa regulates activation of hepatic stellate cells which play key roles in hepatic fibrosis [27,28]. Differential modulation of CEBPa has been shown in HCC patients [17,29]. Our analysis on the regulation of the inflammatory biomarker YKL40 expression at the transcriptional level provides new insight into role of components of the NOTCH and NFKB signaling pathways in HCV induced hepatic fibrosis and HCC. It is of interest that HCV core protein NS3 can activate the NOTCH signaling pathway resulting in development of HCV-induced HCC [30]. Activation of NOTCH signaling also promotes TGFb1 induced epithelial-mesenchymal transition, an initial step in the development of fibrosis, by directly interacting with the transcriptional machinery [31,32,33]. Modulation of both NOTCH1 and NFKB pathways have also been implicated in several cancers including HCC [34,35,36,37].
In addition to demonstrating the interaction of NOTCH, CEBPa and NFKB pathways in YKL40 expression, we determined an important role for modulation of miRNA by HCV. The understanding of the complex role of miRNAs in the various physiological and pathological processes is still emerging. miRNAs have been implicated in regulation of pro-inflammatory cytokines, anti-inflammatory cytokines, and interferons [38]. miRNA-21 has been shown to regulate chronic rejection and has been implicated in the development of fibrosis following liver transplantation [39,40]. miRNA-21 modulates resident fibroblasts, epithelial cells and lymphocytes to produce pro-fibrotic cytokines resulting in deposition of ECM components [39,41]. It has been shown that several liver specific miRNAs including miRNA-122, miRNA-148, miRNA-194 are sensitive biomarkers for hepatocyte injury and rejection after liver transplantation [42]. In this study we identified a novel miRNA (miRNA-449a) that is modulated in HCV infection. Further, we have shown that miRNA-449a regulates HCV induced inflammatory responses (YKL40) implicated in allograft liver fibrosis.
Previous studies in our laboratory have demonstrated an upregulation of autoimmune Th17 inflammatory cascade leading to liver fibrosis in HCV infection, particularly in recurrent HCV following orthotopic liver transplantation [43]. Viral modulations of miRNA have been well known to influence transcriptional regulation in T cell responses, inflammation and fibrosis [44]. Although suggested in literature, a direct effect of HCV mediated modulation of cellular inflammatory responses and fibrosis is yet to be determined. In our current study using promoter analyses techniques we provide direct evidence for the role of miRNA-449a, which is down regulated in HCV infection (Figure 1), in the upregulation pro-inflammatory YKL40 fibrotic cascade. miRNA-449a has been implicated transcriptional dysregulation affecting cell proliferation in several human diseases including cancers [45,46]. In vitro studies have also shown that miRNA-449a can arrest cell proliferation and induce apoptosis [46,47]. Thus, HCV induced down regulation of miRNA-449a in human livers can upregulate transcriptional factors leading to increased inflammatory response; promoting cell proliferation that can result in HCC.
We have also demonstrated by in vitro analysis that miRNA-449a regulates TNFa mediated induction of YKL40 by targeting components of the NOTCH signaling pathway (NOTCH1) ( Figure 6). We have shown for the first time in human hepatocytes that miRNA-449a targets NOTCH1 for translational silencing. Studies have shown that miRNA-34a is downregulated in patients with chronic hypoxia kidney diseases and promotes epithelialmesenchymal transition by targeting components of the NOTCH signaling pathway [48]. In HCV infected patients expression of miRNA-449a was significantly downregulated ( Figure 1A). In consistence with our in vitro results, in the same HCV patients downregulation of miRNA-449a was accompanied by significant upregulation of NOTCH1 (Figure 7). Thus, results obtained from patient samples and our in vitro analysis using hepatocytes indicate that upregulation of NOTCH1 resulting from downregulation of miRNA-449a stabilizes nuclear P65 to activate YKL40 expression in patients with HCV mediated hepatic fibrosis (Figure 1). The increased expression of YKL40 in patients with HCV mediated liver fibrosis [5,6] can be attributed to this novel pathway ( Figure 8). Since, YKL40 is elevated in patients with multiple liver diseases ( Figure 1B); it is likely that other parallel pathways for its transcriptional regulation may exist in non-HCV mediated liver fibrosis.
Taken together our results provide new insight into the mechanisms by which miRNAs mediate changes in the inflammatory process by modulating components of the transcriptional machinery. The results from this study should assist in the development of novel strategies for identifying non-invasive biomarkers that can prognosticate patients and monitor those at increased risk for development of cirrhosis and HCC following HCV infection. Figure S1 CEBPa interacts with P65. Co-immunoprecipitation of CEBPa with P65 in hepatocytes. Whole cell lysates were subjected to immunoprecipitation with either mouse IgG or anti-CEBPa. P65 in the cell lysates (Input) and immunoprecipitated complexes (IP) was detected by immunoblotting with anti-P65. CEBPa was detected by immunoblotting with anti-CEBPa. (TIF) Figure S2 siRNA mediated knockdown of NOTCH1. A. Hepatocytes were transfected with either non-specific siRNA or siRNA specific for NOTCH1 and treated with TNFa. Lysates were subjected to immunoblotting with anti-NOTCH1. B. HEPG2 cells were transfected with either non-specific siRNA or siRNA specific for NOTCH1, treated with TNFa and immunostained with anti-NOTCH1. C. Quantification of the NOTCH1 immunostaining signal in HEPG2 cells (S2B). The numbers represent the average fluorescence intensity of P65 (n = 100). (TIF) Figure S3 miRNA-449a regulates YKL40 expression. Hepatocytes were transfected with an empty vector (-) or vector expressing miRNA-449a (+) and expression of YKL40 was determined by Q-PCR. The DDCT value was calculated by normalizing the threshold (CT) values with GAPDH and expression of YKL40 in controls. The '*' represents p value,0.05 obtained by a two-tailed t-test. (TIF)