Functional Study of miR-27a in Human Hepatic Stellate Cells by Proteomic Analysis: Comprehensive View and a Role in Myogenic Tans-Differentiation

We previous reported that miR-27a regulates lipid metabolism and cell proliferation during hepatic stellate cells (HSCs) activation. To further explore the biological function and underlying mechanisms of miR-27a in HSCs, global protein expression affected by overexpression of miR-27a in HSCs was analyzed by a cleavable isotope-coded affinity tags (cICAT) based comparative proteomic approach. In the present study, 1267 non-redundant proteins were identified with unique accession numbers (score ≥1.3, i.e. confidence ≥95%), among which 1171 were quantified and 149 proteins (12.72%) were differentially expressed with a differential expression ratio of 1.5. We found that up-regulated proteins by miR-27a mainly participate in cell proliferation and myogenesis, while down-regulated proteins were the key enzymes involved in de novo lipid synthesis. The expression of a group of six miR-27a regulated proteins was validated and the function of one miR-27a regulated protein was further validated. The results not only delineated the underlying mechanism of miR-27a in modulating fat metabolism and cell proliferation, but also revealed a novel role of miR-27a in promoting myogenic tans-differentiation during HSCs activation. This study also exemplified proteomics strategy as a powerful tool for the functional study of miRNA.

Introduction microRNAs (miRNAs) regulate gene expression post-transcriptionally by binding primarily to the 39untranslated region (39UTR) of their target mRNAs, resulting in mRNA destabilization or translational repression [1]. Genes encoding 2042 mature human miRNAs have so far been identified (miRBase v.19) [2] and miRNAs are predicted to regulate the expression of up to 60% of human protein-encoding genes [3]. The best way to understand the biological function of a miRNA is to identify the genes that it regulates. Several bioinformatics methods have been developed for miRNA target prediction, including TargetScan (www.targetscan. org), miRanda (www.microrna.org), TarBase (diana.cslab.ece.ntua.gr), PicTar (pictar.mdcberlin. de) et al. However since the mechanism of miRNA target recognition is still not fully understood, target gene prediction is not accurate and sometimes over predict [4]. In addition, a single miRNA can target hundreds of proteins and a single protein can be influenced by multiple miRNAs [5]. Thus comprehensive understanding of the phenotypic effects of miRNAs at the cellular level is currently difficult.
The use of quantitative proteomic strategies to characterize targets of miRNAs has opened new avenues to miRNA biology study [6]. The method of cleavable isotope-coded affinity tags (cICAT) coupling with nano LC-MS/MS is a quantitative proteomic approach that enables rapid, comprehensive and reliable analysis of the proteomes of two comparable samples [7]. More importantly, compared with other quantitative proteomic strategies, cICAT based approach could greatly reduce the sample complexity, therefore those low abundance proteins could be readily identified.
We have previously reported that miR-27a,b suppresses fat accumulation and promotes cell proliferation during hepatic stellate cells (HSCs) activation [8]. Thereafter, miR-27 has been evidenced to act as negative regulator of adipocyte differentiation [9] or lipid metabolism [10], and positive regulator of cell proliferation [11] by several groups. It has also been regarded as an oncogene in some malignant tumor [12,13]. To further explore the possible functions and underlying mechanism of miR-27a during HSCs activation, human stellate cell line LX2/miR-27a stable transfectants was established and validated. Global protein expression profiles were compared between LX2/miR-27a and LX2/miR-neg control by cICAT-based proteomic approach. We found that out of 1267 identified proteins, 149 proteins were differentially expressed, and 75 were repressed by miR-27a overexpression among which, 15 proteins were predicted miR-27a targets. The bio-significance of miR-27a was analyzed based on the functional annotation of miR-27a regulated proteins. Individual siRNA mediated knock-down of one miR-27a regulated protein was performed to demonstrate the phenotypic effects.

Establishment of stable transfectants
Human hepatic stellate cell line LX2 cells [14] were maintained in DMEM (Invitrogen), supplemented with 10% FBS (Invitrogen), and were incubated in a humidified atmosphere of 5% CO 2 and 95% air at 37uC. The medium was changed every 48 hours. Stable transfectants were constructed using LX2 cells that had been plated at approximately 1610 5 per 60-mm diameter culture dish and cultured overnight. The cells were transfected with 5 mg pcDNA6.2-GW/EmGFP-mir-27a or mir-neg control plasmids by Lipofectamine 2000 (Invitrogen). Transfection efficiencies were checked by EmGFP expression under fluorescent microscope. Clones were selected and maintained in DMEM supplemented with 10 mg/ml Blasticidin (Invitrogen). Two stably transfected cell lines, LX2/miR-27a and LX2/miR-neg were isolated after 28 days' selection.

Real-time reverse transcription PCR (RT-PCR)
Total RNA from LX2 cells was extracted using Trizol reagent (Invitrogen). cDNAs and the first-strand cDNAs of miRNA were produced according to the manufacturer's instructions for Thermoscript RT-PCR system (Invitrogen) or NCode miRNA First-Strand cDNA Synthesis kits (Invitrogen). For the quantitative detection of miR-27a and mRNAs of interested genes, the templates and primer sets (Table S1) were mixed with SYBR qPCR master mix (TaKaRa, Dalian, China), and real-time PCR was performed using Rotor-Gene 3000 (Corbett Research, Sydney). The cycling parameters were: initial denaturing at 94uC for 15 sec, followed by 40 cycles of 94uC denaturing for 10 sec, primer annealing and extension at 60uC for 40 sec. To ensure the specificity of the amplification reaction, melting curve analysis was performed. The expression of miR-27a was normalized to U6snRNA, and mRNAs were normalized to GAPDH. Relative gene expression was presented by comparative CT method.

Quantitative proteomic analysis
Global protein expression profile changes of LX2/miR-27a transfectants were analyzed by a cleavable isotope-coded affinity tags (cICAT) labeling coupled with online 2D nanoLC-MS/MS based quantitative proteomic approach. cICAT reagents were from Applied Biosystems (Foster City, CA).
(A) cICAT labeling. Proteins from LX2/miR-27a and LX2/ miR-neg control were labeled with isotopically heavy (H) and light (L) cICAT reagents respectively following the manufacture's protocol. Briefly 100 mg total protein collected from LX2/miR-27a and negative control LX2/miR-neg were labeled, respectively, with isotopically light ( 12 C for LX2/miR-neg) and heavy ( 13 C for LX2/miR-27a ) cICAT reagents at 37uC for 2 hours. The labeled preparations were combined and digested with trypsin (Promega, madison, WI) overnight at 37uC using an enzyme-toprotein ratio of 1:50 w/w. The resulting peptides were subsequently purified by cation exchange chromatography and avidin affinity chromatography (Applied Biosystems). The biotin group on the tag was removed by acid cleavage and the peptides were dried by vacuum-evaporation using a Speedvac TM system (Thermo Scientific).
(B) 2D nanoLC-MS/MS analysis. The dried peptides were resuspended in 80 ul of an aqueous solution containing 0.1% FA and 5% acetonitrile, the resulting solution was loaded onto a 30*0.5 mm strong cation exchange column (Agilent Technologies) and separated into 17 fractions with a step gradient of 0 mM, 10 mM, 20 mM, 30 mM, 40 mM, 50 mM, 60 mM, 70 mM, 80 mM, 90 mM, 100 mM, 125 mM 150 mM, 200 mM, 300 mM, 400 mM, 500 mM and 900 mM, 0.1% FA, 5% acetonitrile. The elutions from SCX column were further separated on a 150*0.075 mm Vydac C18 reverse phase column (Grace, inc) in line after a nanotrap column (Agilent Technologies) using a nanoHPLC 1100 system (Agilent Technologies). Separation of the peptides was performed at 400 nl/min and was coupled to online analysis by tandem mass spectrometry using a QstarXL MS/MS system (Applied Biosystems) equipped with a nanospray ion source (Applied Biosystems). Elution of the peptides into the mass spectrometer was performed with a linear gradient from 95% mobile phase A (0.1% FA, 99.9% water) to 35% mobile phase B (0.1% FA, 99.9% acetonitrile) over 120 minutes followed by 80% mobile phase B for 10 min. The peptides were detected in positive ion mode using an IDA (information dependent acquisition) method in which three most abundant ions detected in a MS scan were selected for MS/MS analysis. Two independent analyses were performed.
(C) Data Analysis. For protein identification and quantification, all MS/MS spectra were searched against the IPI human protein database (V3.83) using ProteinpilotTM 3.0.1 (Applied Biosystem). The software compares relative intensity of proteins present in samples based on the intensity of reporter ions released from each labeled peptide and automatically calculates protein ratios and p-values for each protein. For protein identification, 95% confidence was used and the corresponding FDR ,1%.

GOfact
(http://61.50.138.118/gofact/cgi/gofact2009.cgi) strategy [15,16] which based on the structured and controlled vocabularies -Gene Ontology (GO), and the GO annotation from related databases was used to identify the functional distribution and the enriched functional categories of miR-27a regulated proteins in LX2 cells. The subcellular locations and bio-functions of proteins were also annotated by Protein Knowledgebase (UniprotKB) (http://www.uniprot.org/).

Transfection of siRNA
Transfection of siRNA was performed according to the manufacturer's protocol (Sigma, Saint Louis, MO). LX2 and Function of miR-27a in Human Hepatic Stellate Cells PLOS ONE | www.plosone.org LX2/miR-27a transfectants cultured in 24-well plates or 6-cm dishes were transfected at 50-70% confluence with siRNA targeting human four and a half LIM domains 1 (FHL1) by means of the siRNA transfection reagent RNAiMAX (Invitrogen). NTC (Non-targeting control) siRNA was transfected simultaneously as negative control. After 48 h transfection, the efficiency of siRNA-mediated mRNA degradation was assessed by real-time RT-PCR.

Proliferation and migration assays
The effects of siRNA transfection on LX2/miR-27a transfectants migration were measured by using a modified Boyden chamber assay. Two days after transfection, 2610 4 cells in serumfree DMEM were plated on the upper chamber of each Transwell (Costar, Corning Inc., NY) with 8 mm pores, while the lower chamber contained 800 ml completed medium. Transfected cells were incubated for 16 h at 37uC in 5% CO2. Non-migrating cells were carefully removed from the upper surface of the membrane with cotton swabs. Membranes were stained with crystal violet and mounted onto glass slides. Migration was quantified by counting cells in eight 200x microscopic fields.
Forty-eight hours after siRNAs transfection, the cell proliferation of LX2 cells was detected by incorporation of 5-ethynyl-29deoxyuridine (EdU) with the Cell-Light EdU Apollo 567 Cell Proliferation Kit (Ruibo Biotech, Guangzhou, China). According to the kit's protocol, cells were incubated with 10 mM EdU for 16 h before fixation, permeabilization, and EdU staining. EdU was detected by Apollo fluorescent dye at 567 nm wave length and nuclei were counterstained with 5 mg/ml Hoechst 33342. For each well, eight fields were counted at a 200x magnification. The results were expressed as the labeling index according to the following formula: number of EdU-positive nuclei6100/number of total nuclei.

Statistics assay
Data are expressed as the mean 6 SD. Comparison between groups were made by Student's t test (two tailed) or one-way ANOVA followed by Tukey's multiple comparison test. The relationship between two data sets was analyzed by linear regression. Differences were considered significant if P,0.05. Unless otherwise specified, all assays were performed in triplicate.

Biological characterization of LX2/miR-27a stable transfectants
To explore the biological effects of miR-27a overexpression on HSCs, we established a LX2/miR-27a stable transfectants ( Figure 1A). The expression of mature miR-27a increased significantly in LX2/miR-27a stable transfectants ( Figure 1B). As it was expected, LX2/miR-27a stable transfectants showed increased cell proliferation and migration compared to LX2/ miR-neg stable transfectants ( Figure 1C and D). The influence of Function of miR-27a in Human Hepatic Stellate Cells PLOS ONE | www.plosone.org miR-27a over expression on lipid metabolism was not measurable due to the already activated HSC phenotype of LX2 cell line.

Identification of miR-27a regulated proteins by cICAT-based proteomic analyses
Global protein expression profiles were compared between LX2/miR-27a and LX2/miR-neg stable transfectants by a cICAT-based quantitative proteomic approach (Figure 2A-C). Two biological replications were analyzed (Table S2). To estimate the analytical reproducibility of our proteomics study, linear regression analyses were performed on H/L ratios of duplicate analyses of samples 1 and 2 ( Figure 2D). Pearson correlation coefficient for sample 1 and 2 was 0.8039 (P,0.01). Thus, the ratios of the two duplicate analyses were significantly positively correlated, indicating the good analytical reproducibility of the online 2D LC/MS/MS system. Thereby, spectral data from two duplicate analyses were merged and searched again to enhance the coverage of protein identification and to ''average'' the expression ratios of proteins identified in samples 1 and 2 (Table S3).
In the present study, 1267 non-redundant proteins were identified with unique accession numbers (score $1.3, i.e. confidence $95%), among which 1171 were quantified (Table  S3). In the present study, based on the expression ratio of housekeeping proteins such as b-actin (ACTB, H/L = 1.0637) and tubulin b chain (TUBB, H/L = 1.0274), a differential protein expression ratio of 1.5 was selected as significant threshold [17], thus 149 (12.72%) proteins were differentially expressed. Of these 149 proteins, 74 were up-regulated (i.e. H/L $1.5) and 75 were down-regulated (i.e. H/L #0.6667), the number of up-regulated proteins was almost equal to that of down-regulated (Table S4). Compared with our previous study on HSCs activation [18], the extent of protein expression changes is relatively small in miR-27a overexpressed LX2, only 6 proteins increase up to 3-fold (i.e. H/L $3.0) and 2 proteins reduced below 3-fold (i.e. H/L #0.3333).
The results also corroborated a recent finding that a single miRNA  could regulate the production of hundreds of proteins, but the regulation was typically relatively mild [5].

Correlation between miR-27a target prediction and down-regulated proteins in LX2/miR-27a identified by cICAT
Next, we tried to figure out how miR-27a target prediction correlated with miR-27a down-regulated proteins in HSCs identified by cICAT-based proteomics analyses. TargetScan is one of the widely recognized databases for biological targets prediction of miRNAs [19]. By searching TargetScan Human Release 6.2 (http://www.targetscan.org/vert_61/), we found that only 2 out of the 75 down-regulated proteins were predicted targets of miR-27a, namely SMAD5 (mothers against decapentaplegic homolog 5) and ACLY (ATP-citrate synthase). SMAD5, a key component of TGF-beta signaling pathway, is an experimentally confirmed target of miR-27 [20]. ACLY is the primary enzyme responsible for the synthesis of cytosolic acetyl-CoA in many tissues and has a central role in de novo lipid synthesis. We further searched the predicted consequential pairing of miR-27a target region in the 39 UTR of the remaining 73 down-regulated proteins in TargetScan Human Release 6.2. As shown in Table 1, 15 (20%) out of 75 down-regulated proteins could be potential targets of miR-27a, while the other 60 (80%) down-regulated proteins did not have consequential pairing of miR-27a target region in the 39 UTR. Moreover, 74 proteins were even upregulated in LX2/miR-27a stable transfectants. These findings suggested that the miRNA responsive proteins were not necessarily the predicted endogenous targets, they also reflected indirect effects. The underlying mechanisms deserve further investigation, as it has also been reported that miRNAs can even stimulate gene expression post transcriptionally by direct and indirect mechanisms [21].

Validation of proteomic findings by real-time RT-PCR
Six of the differentially expressed proteins identified in two replicate cICAT assays, ATP-citrate synthase (ACLY), leukotriene A4 hydrolase (LTA4H), cathepsin L1 (CTSL1), thrombospondin-1 precursor (THBS1), four and a half LIM domains 1 (FHL1) and high-mobility group box 1(HMGB1), were validated by real-time RT-PCR. The relationship between fold changes of protein detected by cICAT and fold changes of protein encoding gene detected by PCR was assessed by linear regression analysis. Pearson correlation coefficient for cICAT and real-time RT-PCR expression data was 0.9745 (P = 0.001). The PCR results confirmed the expression pattern observed in cICAT quantitative proteomics analysis (Figure 3).

Overall distribution of miR-27a regulated proteins in LX2 cells
The subcellular location and bio-function of miR-27a regulated proteins in LX2 cells were categorized by using Protein Knowledgebase (UniprotKB) ( Table S4). The subcellular localization of miR-27a regulated proteins is wide, including cytoplasm, nucleus, plasma membrane and extracellular space ( Figure 4A). Enzymes, kinase, peptidase and phosphatase constituted the largest part of miR-27a regulated proteins in LX2 cells (49 out of 134 annotated differentially expressed proteins, 37%), followed by transcription regulator (11 out of 134, 8%). Therefore, by preferentially influencing the expression of enzymes and transcription regulators, miR-27a could perform its bio-function with high efficiency ( Figure 4B). GOfact was used to identify the enriched functional categories. The data of functional categorizing was inspiring, according to their molecular functions, most of the altered proteins could be well assigned into the categories involved in de novo lipid synthesis, cell proliferation, apoptosis, cell adhesion and migration, which were closely associated with the mechanisms participating in HSCs activation (Table 2, 3).
A large number of the down-regulated proteins were involved in de novo lipid synthesis ( Figure 5), among which three groups were most concerned: (1) aconitase (ACO2), malate dehydrogenase (MDH2), and ATP-citrate synthase (ACLY), which are important enzymes participating in tricarboxylic acid cycle and favor the production of acetyl-CoA; (2) glucose 1-dehydrogenase/6-phosphogluconolactonase (H6PD), the rate-limiting enzyme for pentose phosphate pathway that supplies NADPH; (3) 6-phosphofructokinase type C (PFKP) and fructose-bisphosphate aldolase C (ALDOC), are involved in glycolytic pathway that provides glycerol-3-phosphate, and the former is a rate-limiting enzyme ( Table 2). Acetyl-CoA, NADPH and glycerol-3-phosphate are all required in de novo lipid synthesis. On the other hand, one negative regulator of lipid synthesis called 59-AMP-activated protein kinase catalytic subunit alpha-1 (PRKAA1) was significantly up-regulated (Table 3). By phosphorylation, PRKAA1 can inactivate acetyl-CoA carboxylase that catalyzes the rate-limiting reaction in the biosynthesis of long-chain fatty acids [22,23]. So miR-27a may affect HSCs fat accumulation by directly regulating a group of genes that are involved in the biosynthesis of triglyceride.
The present proteomic study not only provided the possible mechanism underlying the previously reported miR-27 function in     HSCs, but also casted new light on a novel role of miR-27a in myogenesis, which was consistent with the myofibroblast transdifferentiation during HSCs activation. In 9 up-regulated cytoskeleton related proteins, 4 are structural constituents of muscle, including tropomyosin alpha-1 chain (TPM1), tropomyosin beta chain (TPM2), myosin-IXb (MYO9B) and myosin regulatory light chain 2 (MYL9); 4 are in regulation of actomyosin structure and function, including protein phosphatase 1 regulatory subunit 12A (PPP1R12A) [29]; calponin 2 (CNN2) [30]; transforming protein RhoA (RHOA) [31] and FHL1 [32]. The up-regulation of TPM1, MYO9B and MYL9 by miR-27a in LX2 cells was further validated by RT-PCR ( Figure S1). In a previous study, it has also been evidenced that miR-27a can up-regulate cardiac myosin heavy chain (MHC) gene (b-MHC) expression via thyroid hormone signaling [33]. And miR-27a has also been reported to be able to influence muscle stem cell behavior [34]. It is the first time for us to recognize a novel role of miR-27a in promoting myogenic tans-differentiation in HSCs. The finding also suggested similar bio-functions of the same miRNA in different types of tissues or cells. However, further effort is needed to determine the role of miR-27a in myogenic trans-differentiation of activated HSCs.

The biological significance of miR-27a regulated protein in HSCs
In order to validate the biological significance of miR-27a regulated proteins identified by cICAT proteomic strategy, the function of FHL1, one of the highest increased proteins which not only related to cell growth [28] but also played a crucial role in embryonic skeletal muscle myogenesis [32], was evaluated in miR-27a transfectants. Three different siRNA targeting FHL1 were compared. The one possessed the highest knockdown efficiency ( Figure S2) was used in the following experiment. Our data showed that FHL1 involved in miR-27a related HSCs proliferation and migration, knockdown of FHL1 significantly inhibited the proliferation and migration of LX2/miR-27a transfectants (Figure 6). Interestingly, in a recent study based on 2-dimensional Proteins from LX2/miR-27a were labeled with heavy isotope (H) tagging and those from LX2/miR-neg were labeled with light isotope (L) tagging. Data were from two independent cICAT-based quantitative analyses. doi:10.1371/journal.pone.0108351.t003 Figure 5. Altered proteins that are involved in metabolism processes related to de novo lipid synthesis: aconitase 2 (ACO2) and malate dehydrogenase (MDH2), which participate in tricarboxylic acid cycle (TAC) (i) decreased; ATP-citrate synthase (ACLY), the primary enzyme responsible for the synthesis of cytosolic acetyl-CoA (ii) decreased; 59-AMP-activated protein kinase catalytic subunit alpha-1 (PRKAA1) that repress the synthesis of malonyl-CoA (iii) by phosphorylation of acetyl-CoA carboxylase increased; glucose 1-dehydrogenase/6-phosphogluconolactonase (H6PD), the rate-limiting enzyme in pentose phosphate pathway (PPP) (iv) decreased; 6-phosphofructokinase type C (PFKP) that acts as the rate-limiting enzyme, fructose-bisphosphate aldolase C (ALDOC), which are involved in glycolytic pathway(v) decreased. doi:10.1371/journal.pone.0108351.g005 polyacrylamide gel electrophoresis (2D-PAGE) proteomic approach, FHL-1 was identified as one of the most prominently upregulated proteins in pulmonary hypertension mouse model, and a similar effects of FHL-1 on promoting pulmonary arterial smooth muscle cell migration and proliferation has also been evidenced [35].

Conclusions
The data of present study indicated that miR-27a influenced the activation of HSCs by affecting several groups of proteins. These results not only explained our previous finding that overexpression of miR-27a promoted HSC activation with reduced cytoplasmic lipid drops and increased cell proliferation [8], but also revealed a novel role of miR-27a in promoting the myogenic trans-differentiation of activated HSC into myofibroblast. The pattern of miR-27a regulation on protein expression might well reflect the emerging picture of miRNA regulation in animals is far richer and more complex than the crisp linear pathways [1]. Our study also validated proteomic strategy as a promising tool for functional study of miRNA. In the future, it will be interesting to uncover the mechanisms underlying the regulation of miR-27a on these functionally related genes. Figure S1 Validation of myogenesis related genes found by cICAT proteomic analyses. The expression of TPM1, MYO9B and MYL9 encoding mRNA was evaluated by RT-PCR in LX2/miR-27a stable transfectants. *P,0.05, compared with LX2/miR-neg. (TIF) Figure S2 Knockdown efficiency of FHL1 siRNA, LX2 cells were transfected with FHL1 specific siRNA or with NTC siRNA, after 48 hours, their mRNA levels were determined by quantitative polymerase chain reaction. GAPDH was used as housekeeping gene. NTC, non-targeting control siRNA transfected cells. **P,0.01 compared with NTC. (TIF)

Author Contributions
Conceived and designed the experiments: JLJ YHJ JSZ. Performed the experiments: JLJ YHJ WWW. Analyzed the data: JLJ YHJ. Contributed reagents/materials/analysis tools: JLJ YHJ JSZ. Contributed to the writing of the manuscript: JLJ YHJ.