Advertisement
Browse Subject Areas
?

Click through the PLOS taxonomy to find articles in your field.

For more information about PLOS Subject Areas, click here.

  • Loading metrics

RNA-Sequencing Analysis of HepG2 Cells Treated with Atorvastatin

RNA-Sequencing Analysis of HepG2 Cells Treated with Atorvastatin

  • Camilla Stormo, 
  • Marianne K. Kringen, 
  • Robert Lyle, 
  • Ole Kristoffer Olstad, 
  • Daniel Sachse, 
  • Jens P. Berg, 
  • Armin P. Piehler
PLOS
x

Abstract

The cholesterol-lowering drug atorvastatin is among the most prescribed drug in the world. Alternative splicing in a number of genes has been reported to be associated with variable statin response. RNA-seq has proven to be a powerful technique for genome-wide splice variant analysis. In the present study, we sought to investigate atorvastatin responsive splice variants in HepG2 cells using RNA-seq analysis to identify novel candidate genes implicated in cholesterol homeostasis and in the statin response. HepG2 cells were treated with 10 µM atorvastatin for 24 hours. RNA-seq and exon array analyses were performed. The validation of selected genes was performed using Taqman gene expression assays. RNA-seq analysis identified 121 genes and 98 specific splice variants, of which four were minor splice variants to be differentially expressed, 11 were genes with potential changes in their splicing patterns (SYCP3, ZNF195, ZNF674, MYD88, WHSC1, KIF16B, ZNF92, AGER, FCHO1, SLC6A12 and AKAP9), and one was a gene (RAP1GAP) with differential promoter usage. The IL21R transcript was detected to be differentially expressed via RNA-seq and RT-qPCR, but not in the exon array. In conclusion, several novel candidate genes that are affected by atorvastatin treatment were identified in this study. Further studies are needed to determine the biological significance of the atorvastatin responsive splice variants that have been uniquely identified using RNA-seq.

Introduction

Atorvastatin is an efficient competitive inhibitor of 3-hydroxy-3-methylglutaryl-Coenzyme A (HMG-CoA) reductase (HMGCR), the rate-limiting enzyme in cholesterol synthesis. Atorvastatin belongs to the statin class of drugs and is widely used to reduce cholesterol levels and the risk of cardiovascular disease. The cholesterol-lowering effect of statins is well documented. Statins block HMGCR and prevent the conversion of HMG-CoA to mevalonate and thereby decrease the level of sterol and non-sterol products derived from mevalonate, including cholesterol. As a compensatory effect, sterol-regulated genes, such as HMGCR and low-density lipoprotein (LDL) receptor (LDLR), are upregulated to increase de novo cholesterol synthesis and the receptor-mediated uptake of LDL-cholesterol from the blood.

Statins are generally well tolerated in most people [1][3]; however, there is a large amount of variability in the responses across individuals, which can be partly explained by genetic factors [4]. Interestingly, the expression level of a minor splice variant of HMGCR lacking exon 13 has been shown to be associated with variations in the plasma LDL-cholesterol levels and statin response [5]. Statins may also induce a range of adverse, muscle-related events in 1–5% of patients [6]. In addition, beneficial effects of statins on endothelial function, inflammation, and plaque stability have been demonstrated, suggesting that statins have effects beyond lowering cholesterol [7]. The exact molecular mechanisms underlying these LDL-cholesterol independent effects of statins are still unclear.

Studies on global gene expression effects of statin treatment have mostly been performed by microarray analysis [8][13]. Microarrays have been used in gene expression studies for nearly two decades, whereas RNA sequencing (RNA-seq) is a relatively new approach for this purpose [14], [15]. Briefly, RNA is converted into cDNA, which is fragmented and PCR-amplified before being subjected to parallel sequencing to generate millions of reads. The gene expression and splice variant levels are determined using the RNA-seq data by counting the number of sequence reads aligned to each gene in the genome.

Human hepatoma HepG2 cells are considered a useful model for studying the effects of statin treatment on hepatocytes [16][18]. Because alternative splicing has been reported to be relevant for cholesterol homeostasis and variation in statin response in a number of genes [5], [19][23], we sought to investigate statin responsive splice variant expression changes to identify novel candidate genes as potential regulators of the statin response. In this study, we used RNA-seq in combination with microarray analysis to provide a comprehensive transcriptome profile of HepG2 cells exposed to atorvastatin.

Methods

Cell culture and treatment

The human hepatoblastoma cell line HepG2 (American Type Culture Collection, Manassas, VA, USA) was maintained in collagen I-coated tissue culture flasks (BD Biosciences, San Jose, CA, USA) and modified Eagle's minimal essential medium (MEM; ATCC), supplemented with 10% heat inactivated fetal bovine serum (FBS) and 1% penicillin-streptomycin-glutamine mixture (Sigma-Aldrich, St Louis, MO, USA). In the treatment experiments, cells were seeded at 2×105 cells/mL in a 12-well collagen I-coated plate (BD Biosciences). The following day, the medium was changed to 1 mL growth medium containing 3 mg/mL of human lipoprotein deficient serum (LPDS; Millipore, Billerica, MA, USA) instead of FBS. Because we wanted to study the effects of statin that were due to the inhibition of cholesterol biosynthesis, we used the magnitude of increased HMGCR mRNA expression as a marker for statin response. The initial experiments showed that HepG2 cells grown in medium containing 10% FBS had low HMGCR mRNA levels, which remained unchanged when treated with atorvastatin. We therefore took advantage of LPDS to activate cholesterol biosynthesis and increase the expression of HMGCR in both the control and atorvastatin-treated cells, as previously demonstrated [24]. In the dose response experiments, cells were treated with water dissolved (3S, 5S)-atorvastatin sodium salt (Toronto Research Chemicals North York, Ontario, Canada) at various concentrations (0, 2.5, 5, 10, 20 and 40 µM) for 24 hours in four independent experiments. For RNA-seq and exon array experiments, cells were treated with or without 10 µM atorvastatin for 24 hours in triplicate wells.

RNA isolation

The cells were lysed by the addition of 700 µL QIAzol Lysis Reagent. The phase separation was achieved by applying 2 mL of Phase Lock Gel Heavy (5PRIME Inc., Gaithersburg, MD, USA) to the lysates. RNA was then purified from the aqueous phase using the miRNeasy Mini Kit (Qiagen, Venlo, The Netherlands), according to the manufacturer's instructions. The RNA was eluted in 30 µL of RNase/DNase-free water and stored at −80°C until analysis. The A260/A280 ratio and RNA concentration were determined using a NanoDrop ND-1000 spectrophotometer (NanoDrop Technologies, Wilmington, DE, USA). The RNA quality was assessed by microfluidic capillary electrophoresis using an Agilent 2100 Bioanalyzer and the RNA 6000 Nano Chip kit (Agilent Technologies, Santa Clara, CA, USA). An A260/A280 ratio in the range of 1.8 to 2.0 and an RIN>7 were considered acceptable. RNA samples were denatured for 2 min at 70°C prior to cDNA synthesis.

RNA-seq

Total RNA (500 ng) from each sample was prepared using TruSeq RNA sample prep reagents (Illumina, San Diego, CA, USA) according to manufacturer's instructions, with fragmentation for 4 min at 94°C. The amplified fragmented cDNA of ∼200 bp in size were sequenced in paired-end mode using the HiSeq 2000 (Illumina) with a read length of 2×100 bp. Two FASTQ files were generated for each sample. The alignment of the reads onto the UCSC reference genome (hg19) and splice site identification were performed using Bowtie/TopHat, with mapping allowing up to two mismatches [25]. An average of 25.8±0.34 (mean ± STD) and 24.0±0.42 million 2×100-bp reads per control and atorvastatin-treated sample were obtained, respectively. The aligned reads were assembled into transcripts using the Cufflinks software [26]. Cufflinks computes normalized values termed FPKM (fragments per kilobase of exon per million fragments mapped), which reflect the mRNA expression levels [26]. The reads were mapped to a total of 23,138 Refseq genes and 39,843 Refseq transcript splice variants. Statistical analysis of differentially expressed genes and transcript splice variants, differential splicing and differential promoter usage was performed using Cuffdiff, which is integrated into Cufflinks. In the differential splicing analysis, Cuffdiff calculates the changes in the relative abundance of splice variants produced from a single primary transcript sharing a common transcription start site, such as a change in the splicing pattern. In the differential promoter usage analysis, Cuffdiff tests for differential promoter use in genes with two or more promoters that generate primary transcripts with different start sites. The status code “OK” in Cuffdiff indicates that there are a sufficient number of reads in a locus to make a reliable calculation. The default false discovery rate (FDR) of Cuffdiff is 5%. The results were visualized using the CummeRbund package [27] in the statistics environment R [28]. The CummeRbund package is available from the Bioconductor website [29]. Sequence reads have been deposited in the NCBI BioSample database (http://www.ncbi.nlm.nih.gov/biosample) with the following accessions: SAMN02808181, SAMN02808182, SAMN02808183, SAMN02808184, SAMN02808185 and SAMN02808186.

Exon Array

Total RNA (100 ng) from each sample was prepared using the Ambion WT Expression Kit (Ambion Inc., Austin, TX, USA), according to manufacturer's instructions. Fragmented and labeled sense strand DNA was hybridized to the GeneChip Human Exon 1.0 ST Array (Affymetrix, Santa Clara, CA, USA). The array contained approximately four probes per exon and 40 probes per gene. The arrays were washed and stained using an FS-450 fluidics station (Affymetrix) and were scanned using a Hewlett Packard Gene Array Scanner 3000 7G (Hewlett Packard, Palo Alto, CA, USA). The scanned images were then analyzed using the Affymetrix GeneChip Command Console. The CEL files were imported into the Partek Genomics Suite software (Partek, Inc., St. Louis, MO) for data analysis. Robust microarray analysis (RMA) was applied for normalization. The exon array data were filtered to include only those probe sets derived from the core meta-probe list, representing approximately 17,000 RefSeq genes and full-length GenBank mRNAs. The gene expression level was estimated by averaging all of the core probe sets for that gene. Paired sample t-tests were performed to analyze the differential gene expression between the control and atorvastatin-treated HepG2 samples. A Benjamini-Hochberg correction was used to correct for multiple comparisons. A 5% FDR was considered statistically significant. The data have been deposited in NCBI's Gene Expression Omnibus [30] and are accessible through the GEO Series accession number GSE57071 (http://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE57071).

Comparison between RNA-seq and exon array

The strength of the linear relationship between the RNA-seq and exon array data was measured using Pearson's correlation coefficient on the log2 values obtained for the 17,151 genes that were detected using both methods. A constant offset of 1 was added to the FPKM values to handle the log2 transformation of zero values.

Pathway analysis of differentially expressed gene lists

Statistically significant genes were imported into Ingenuity Pathway Analysis (www.ingenuity.com; Ingenuity Systems, Redwood City, CA, USA). The Core analysis was used to identify the main biological functions and canonical pathways associated with the differentially expressed genes.

Validation by reverse transcription quantitative PCR (RTqPCR)

Total RNA (200 ng) was subjected to cDNA synthesis in a 20 µL reaction using the qScript cDNA Synthesis Kit (Quanta BioSciences, Gaithersburg, MD, USA) according to the manufacturer's instructions and stored at −20°C until analysis. All qPCRs were performed on a ViiA 7 Real-Time PCR System in a standard 96-well format in a 20 µL reaction mixture containing 10 µL of TaqMan Universal PCR Master Mix, 7 µL of RNase-free water, 1 µL of Taqman gene expression assay and 2 µL of a 1∶5 dilution of each cDNA sample (Applied Biosystems, Foster City, CA, USA). The cycling steps were 10 min at 95°C followed by 40 cycles of 95°C for 15 s and 60°C for 1 min. The following genes and Taqman gene expression assays (Applied Biosystems) were selected to validate the RNA-seq results: HMGCR (Hs01102990_m1), IL21R (Hs00222310_m1), PPIA (Hs99999904_m1) and TBP (Hs99999910_m1). PPIA and TBP have previously been validated in our laboratory to be suitable for normalization in atorvastatin-treated HepG2 cells [24]. Each sample was run in duplicate. The quantification cycle (Cq) values were normalized according to the ΔΔCq method for calculating fold changes in the mRNA levels [31]: ΔΔCq = 2−(ΔCq_treated – ΔCq_non-treated)

Results

Dose-response curve

The mRNA levels of HMGCR in HepG2 cells increased in response to increasing concentrations of atorvastatin (Figure S1). The increases in the HMGCR mRNA levels after 24 h of atorvastatin treatment were 1.2±0.06-fold (mean±S.E at the concentration of 2.5 µM atorvastatin), 1.2±0.16-fold (5 µM), 1.7±0.19-fold (10 µM), 2.4±0.26-fold (20 µM) and 2.6±0.29-fold (40 µM) compared to that of the untreated HepG2 cells (1.0±0.21). The concentration of 10 µM atorvastatin was within the linear range of the dose-response curve and was therefore used in subsequent experiments.

RNA-seq

Expression level.

A graph of the atorvastatin FPKM versus control FPKM expression values is shown in Figure 1. A total number of 12,426 genes were expressed when filtering for the test status "OK" in the output file containing the differential expression results. The median FPKM was 9.1 (interquartile range IQR 3.3–23) in the control samples and 9.2 (IQR 3.4–23) in the atorvastatin-treated samples.

thumbnail
Figure 1. Bland-Altman plot for the comparison of FPKM expression values of the control and atorvastatin-treated HepG2 cells.

Significantly differentially expressed genes are highlighted in red (n = 121).

https://doi.org/10.1371/journal.pone.0105836.g001

Differential gene expression analysis.

Cuffdiff identified 121 genes to be differentially expressed, with an expected FDR of 5%. A total of 110 genes were upregulated, and 11 genes were downregulated by atorvastatin treatment. The differentially expressed genes showed at least a 1.2-fold change in expression. The statistically significant fold changes ranged from -2.4-fold to +4.1-fold. HMGCR and LDLR expression, which can be used as cellular markers of an in vitro statin response, were significantly increased by 1.8-fold and 1.5-fold, respectively. The 121 differentially expressed genes were subjected to Ingenuity Pathway Analysis. The top five biological functions (Table 1 and Table S1) and canonical pathways (Table 2 and Table S2) associated with these genes are shown. Lipid metabolism and the cholesterol biosynthesis pathway were the major affected biological processes, and this finding supports the on-target effects of the in vitro atorvastatin treatment.

thumbnail
Table 1. Top biological functions identified by Ingenuity Pathway Analysis affected by atorvastatin treatment.

https://doi.org/10.1371/journal.pone.0105836.t001

thumbnail
Table 2. Top five canonical pathways associated with the differentially expressed genes (n = 121) in atorvastatin treated HepG2 cells identified by Ingenuity Pathway Analysis.

https://doi.org/10.1371/journal.pone.0105836.t002

Differential expression analysis of transcript splice variants.

A total of 98 known transcript splice variants were differentially expressed in atorvastatin-treated samples with an FDR of 5% (Table S3). A total of 91 transcript splice variants were upregulated, and seven transcript splice variants were downregulated. The differentially expressed transcript splice variants showed at least a 1.2-fold change in expression. The statistically significant fold changes ranged from -2.3-fold to +4.1-fold. Eight of these 98 splice variants originated from the same genes, i.e., two different transcript splice variants each of ACLY, FDPS, HMGCS1 and LSS were significantly upregulated. The two lists of significantly differentially expressed genes (n = 121) and splice variants (n = 98) were then compared: 11 splice variants were not encoded by the 121 significant genes (Table 3). A barplot of one of these 11 genes, BCL2L11, is shown in Figure 2. By systematically exploring the expression level of each atorvastatin responsive splice variant, we discovered four minor splice variants encoded by the ZNF195, SP1, DTNA and FAM189B genes. FAM189B was significant in both the gene and the splice variant expression analysis.

thumbnail
Figure 2. Transcript splice variants and expression of the BCL2L11 gene.

(A) The canonical splice variant with accession number NM_138621 and the alternatively spliced variant NM_006538 of BCL2L11 are shown. Sixteen additional transcript variants of BCL2L11 are annotated in the National Center for Biotechnology Information (NCBI) Reference Sequence (RefSeq) database (not shown here). Exons are represented by blue boxes separated by intervening sequences (introns). (B) The CummeRbund expression bar plot is shown. FPKM, fragments per kilobase of transcript per million fragments mapped, reflects the mRNA expression level of NM_138621 and NM_006538 in the un-treated (Ctrl) and atorvastatin treated (Stat) HepG2 cells. The HepG2 samples showed low expression levels of the other sixteen alternatively spliced variants (FPKM<1). An asterisk indicates the statistically significant down-regulation of NM_138621 upon atorvastatin treatment after a Benjamini-Hochberg correction (5% FDR). The alternatively spliced variant NM_006538 was slightly upregulated.

https://doi.org/10.1371/journal.pone.0105836.g002

thumbnail
Table 3. The 11 genes uniquely identified by differential expression analysis of transcript splice variants, but not identified by differential gene expression analysis, are shown.

https://doi.org/10.1371/journal.pone.0105836.t003

Differential promoter usage analysis.

Statistical testing for differential promoter usage after atorvastatin treatment revealed RAP1GAP to be significant. A barplot of FPKM values with confidence intervals for RAP1GAP splice variants is shown in Figure S2.

Differential splicing analysis.

A total of 11 genes had a different splicing pattern after atorvastatin treatment. These were SYCP3, ZNF195, ZNF674, MYD88, WHSC1, KIF16B, ZNF92, AGER, FCHO1, SLC6A12 and AKAP9, ordered by increasing p-values (Figure S3-13, respectively).

Exon array

Differential gene expression analysis.

No genes were found to be differentially expressed after a Benjamini-Hochberg correction was applied to the data at a FDR of less than 5%.

RNA-seq gene expression comparison with exon array

The RNA-seq and exon array data from the same samples were compared. The gene expression levels analyzed by RNA-seq and exon array were in agreement (R = 0.81, Figure 3). The dynamic range of RNA-seq was larger than that of the exon array; the FPKM fold changes varied from −4.3 to 4.1, whereas the exon array expression value fold changes varied from −1.6 to +1.9.

thumbnail
Figure 3. RNA-seq and exon array expression values.

A comparison of RNA-seq FPKM and exon array RMA intensity values for the 17,151 genes detected using both methods in HepG2 control cells (R = 0.81) is shown. Significantly differentially expressed genes identified by RNA-seq are highlighted in red.

https://doi.org/10.1371/journal.pone.0105836.g003

RT-qPCR validation of differentially expressed genes

RNA-seq, exon array and RT-qPCR analysis confirmed the stable expression of PPIA and TBP between treatment groups. The fold changes of HMGCR mRNA levels were in good agreement among the three methods: HMGCR was upregulated 1.8-fold in the RNA-seq data, 1.5-fold in the exon array data and 2.2-fold in the RT-qPCR data. IL21R was the most upregulated gene (4.1-fold change), but its mRNA level was low (FPKM = 0.74 in atorvastatin sample). In the exon array data, IL21R expression was not altered by atorvastatin treatment (1.2-fold, p = 0.157). In the RT-qPCR data, IL21R was upregulated 7.1-fold. These results indicate that RNA-seq is able to quantify changes in low-abundance transcripts that were not discovered in the exon array.

Discussion

Because the expression level of a minor transcript splice variant of HMGCR lacking exon 13 may explain variable cholesterol response to statins, we wanted to explore statin-induced gene expression at the level of splice variants. RNA-seq detected that the expression levels of 98 transcript splice variants were altered by atorvastatin treatment. In the majority of cases, atorvastatin changed the mRNA expression level of major splice variants. Unfortunately, there were not sufficient reads to test for the differential expression of the minor splice variant of HMGCR lacking exon 13. However, we discovered four minor splice variants that apparently changed more with atorvastatin than did the major ones. None of these genes were previously known to be associated with the statin response. The proteins encoded by two of the genes, ZNF195 and SP1, are zinc finger transcription factors that are involved in cellular processes such as cell growth, apoptosis, differentiation and immune responses [32], [33]. The third gene, DTNA, encodes a protein that belongs to the dystrophin family and plays a role in synapse formation and stability, whereas the protein encoded by the fourth gene, FAM189, potentially binds to a WW domain-containing protein involved in apoptosis and tumor suppression [34], [35]. The significance of increased expression of these minor splice variants in statin therapy has yet to be determined.

We could confirm that statins induce the expression of a wide range of genes, most notably genes involved in lipid metabolism [13]. Furthermore, by focusing on the expression levels of splice variants, we identified 11 additional splice variants (produced from 11 different genes) that were not on the list of differentially expressed genes. For example, the canonical BCL2L11 splice variant, which acts as an apoptotic activator, was downregulated in response to atorvastatin treatment (Figure 2), whereas the other 16 alternatively spliced variants of the BCL2L11 gene were not affected or were only slightly upregulated. This finding shows that the regulation of single splice variants does not necessarily influence the total gene expression of a gene, suggesting that a more detailed picture is needed when investigating changes in gene expression levels.

RNA-seq technology has an advantage over microarray technology in providing information at the single-base resolution. The high resolution enables the identification of exon boundaries and thus the ability to better distinguish between different splice variants [36]. In our experiment, paired-end sequencing was performed. By using paired-end reads in the analysis, the detection of splice variants and the determination of splicing patterns are improved because cDNA fragments are sequenced from both ends [37]. It is, however, challenging that many alternative exons are expressed at rather low levels. For the majority of the genes analyzed (including BCL2L11), there were not sufficient reads in the locus to make a reliable calculation of differential splicing and promoter use. This may explain the minimal overlap between significant genes in the splice variant analysis (n = 98) and the differential splicing analysis (n = 11). The zinc finger transcription factor gene, ZNF195, was the only gene that was significant in both of the analyses.

Comparative studies of the array-based and RNA-seq platforms agree that RNA-seq is more sensitive and has a greater dynamic range than do array-based methods, with nearly no background and little technical variation [38][41]. The exon array data revealed that the magnitudes of expression changes were rather low after atorvastatin treatment. After correcting for multiple testing, there were no significantly differentially expressed genes in the exon array data, though there were 121 significantly differently expressed genes in the RNA-seq data. There was nevertheless a good accordance between the exon array and RNA-seq expression values, as previously demonstrated in other studies [41]. The larger dynamic range in RNA-seq data compared with the exon array allowed for the identification of the interleukin 21 receptor (IL21R). The differential gene expression of IL21R was not detected by exon array and has not previously been shown to be involved in the statin response. The upregulation of IL21R mRNA level was confirmed by RT-qPCR supporting the reliability of our RNA-seq data. This finding also highlights the potential of RNA-seq analysis to complement and extend microarray measurements, as noted by other researchers [15], [41]. IL21R encodes a receptor for interleukin 21, a group of cytokines that have immunoregulatory activity and are important in T cells, B cells, and natural killer cell responses. Statins appear to have anti-inflammatory properties independent of the lipid-lowering effect, although the clinical significance of this is unclear [42], [43]. It would be interesting to investigate the clinical role of increased IL21R expression in statin therapy.

We used the HMGCR expression levels as a marker of the statin response and found that a concentration of 10 µM gave nearly a half-maximal response in HepG2 cells cultured in LPDS medium. A high concentration of statins (100 µM) may, however, be employed when studying processes related to its toxic effects [8], [44]. For example, the number of differently expressed genes in our study using a concentration of 10 µM atorvastatin is substantially lower than that of another study, which used a 100 µM concentration of the drug (121 genes versus 1091 genes) [8]. However, in that study, the authors also identified lipid metabolism as the major biological function affected by the atorvastatin treatment, although they did not culture the cells in LPDS medium as we did. Their PCR results that demonstrated increased expression of genes involved in cholesterol metabolism (such as ACAT2, HMGCS, HMGCR, SQLE, LSS) were similar to our results.

In vitro effects of statins appear to be largely cell type-dependent. In epithelial cells, 10 µM atorvastatin treatments for 24 hours have been shown to induce anti-thrombotic effects due to the increased expression of genes coding for endothelial nitric oxide synthase, thrombomodulin, heat shock protein 27 and tissue plasminogen activator [45]. In the human umbilical vein endothelial cell line, EA.hy926, the increased expression of various Kruppel-like transcription factors, which have tumor-suppressing functions, and the modulation of cell cycle related genes (such as CCNA2, CCNE2, CCNB1 and CCNB2) provides evidence for an anti-cancer effect of atorvastatin [46]. In human peripheral blood, an anti-inflammatory activity of atorvastatin has been detected due to decreased mRNA levels of chemokines (CCL2, CCL7, CCL13, CCL18, CXCL1) and cytokines (IL-6, IL-8, IL-1, PAI-1, TGF-beta1, TGF-beta2) [9]. None of these genes were significantly affected in our analysis of HepG2 cells.

A major limitation of the present study is the small number of biological replicates. Only three control samples and three atorvastatin-treated samples were analyzed. The results from our study should be replicated and presented with dose-response data to confirm the significance of the results we obtained, particularly the results from transcripts that were present at a low abundance. To conclude, we have identified several novel candidate genes that are affected by atorvastatin treatment in HepG2 cells. It would be interesting to study the biological significance of the atorvastatin-responsive splice variants that have been uniquely identified with RNA-seq.

Supporting Information

Figure S1.

Dose response curve of HMGCR gene expression. HepG2 cells were incubated for 24 h in culture medium containing lipoprotein deficient serum with indicated concentration of atorvastatin. HMGCR mRNA levels were determined by RT-qPCR using Taqman gene expression assay specific for HMGCR and two validated references genes for normalization. The results are shown as the mean fold change compared with mRNA levels in un-treated cells from four independent experiments. Error bars indicate standard errors (±S.E), n = 4.

https://doi.org/10.1371/journal.pone.0105836.s001

(PNG)

Figure S2.

Differential promoter usage for RAP1GAP. A barplot of expression values with confidence intervals for RAP1GAP is shown. FPKM, fragments per kilobase of transcript per million fragments mapped, reflects the mRNA expression level of transcript splice variants in the un-treated (Ctrl) and atorvastatin treated (Stat) HepG2 cells.

https://doi.org/10.1371/journal.pone.0105836.s002

(JPEG)

Figure S3.

Differential splicing of SYCP3. A barplot of expression values with confidence intervals for SYCP3 is shown. FPKM, fragments per kilobase of transcript per million fragments mapped, reflects the mRNA expression level of transcript splice variants in the un-treated (Ctrl) and atorvastatin treated (Stat) HepG2 cells.

https://doi.org/10.1371/journal.pone.0105836.s003

(PNG)

Figure S4.

Differential splicing of ZNF195. A barplot of expression values with confidence intervals for ZNF195 is shown. FPKM, fragments per kilobase of transcript per million fragments mapped, reflects the mRNA expression level of transcript splice variants in the un-treated (Ctrl) and atorvastatin treated (Stat) HepG2 cells.

https://doi.org/10.1371/journal.pone.0105836.s004

(PNG)

Figure S5.

Differential splicing of ZNF674. A barplot of expression values with confidence intervals for ZNF674 is shown. FPKM, fragments per kilobase of transcript per million fragments mapped, reflects the mRNA expression level of transcript splice variants in the un-treated (Ctrl) and atorvastatin treated (Stat) HepG2 cells.

https://doi.org/10.1371/journal.pone.0105836.s005

(JPEG)

Figure S6.

Differential splicing of MYD88. A barplot of expression values with confidence intervals for MYD88 is shown. FPKM, fragments per kilobase of transcript per million fragments mapped, reflects the mRNA expression level of transcript splice variants in the un-treated (Ctrl) and atorvastatin treated (Stat) HepG2 cells.

https://doi.org/10.1371/journal.pone.0105836.s006

(PNG)

Figure S7.

Differential splicing of WHSC1. A barplot of expression values with confidence intervals for WHSC1 is shown. FPKM, fragments per kilobase of transcript per million fragments mapped, reflects the mRNA expression level of transcript splice variants in the un-treated (Ctrl) and atorvastatin treated (Stat) HepG2 cells.

https://doi.org/10.1371/journal.pone.0105836.s007

(PNG)

Figure S8.

Differential splicing of KIF16B. A barplot of expression values with confidence intervals for KIF16B is shown. FPKM, fragments per kilobase of transcript per million fragments mapped, reflects the mRNA expression level of transcript splice variants in the un-treated (Ctrl) and atorvastatin treated (Stat) HepG2 cells.

https://doi.org/10.1371/journal.pone.0105836.s008

(PNG)

Figure S9.

Differential splicing of ZNF92. A barplot of expression values with confidence intervals for ZNF92 is shown. FPKM, fragments per kilobase of transcript per million fragments mapped, reflects the mRNA expression level of transcript splice variants in the un-treated (Ctrl) and atorvastatin treated (Stat) HepG2 cells.

https://doi.org/10.1371/journal.pone.0105836.s009

(PNG)

Figure S10.

Differential splicing of AGER. A barplot of expression values with confidence intervals for AGER is shown. FPKM, fragments per kilobase of transcript per million fragments mapped, reflects the mRNA expression level of transcript splice variants in the un-treated (Ctrl) and atorvastatin treated (Stat) HepG2 cells.

https://doi.org/10.1371/journal.pone.0105836.s010

(PNG)

Figure S11.

Differential splicing of FCHO1. A barplot of expression values with confidence intervals for FCHO1 is shown. FPKM, fragments per kilobase of transcript per million fragments mapped, reflects the mRNA expression level of transcript splice variants in the un-treated (Ctrl) and atorvastatin treated (Stat) HepG2 cells.

https://doi.org/10.1371/journal.pone.0105836.s011

(PNG)

Figure S12.

Differential splicing of SLC6A12. A barplot of expression values with confidence intervals for SLC6A12 is shown. FPKM, fragments per kilobase of transcript per million fragments mapped, reflects the mRNA expression level of transcript splice variants in the un-treated (Ctrl) and atorvastatin treated (Stat) HepG2 cells.

https://doi.org/10.1371/journal.pone.0105836.s012

(PNG)

Figure S13.

Differential splicing of AKAP9. A barplot of expression values with confidence intervals for AKAP9 is shown. FPKM, fragments per kilobase of transcript per million fragments mapped, reflects the mRNA expression level of transcript splice variants in the un-treated (Ctrl) and atorvastatin treated (Stat) HepG2 cells.

https://doi.org/10.1371/journal.pone.0105836.s013

(PNG)

Table S1.

Biological functions and associated genes identified by Ingenuity Pathway Analysis in atorvastatin treated HepG2 cells.

https://doi.org/10.1371/journal.pone.0105836.s014

(XLS)

Table S2.

Canonical pathways and associated genes identified by Ingenuity Pathway Analysis in atorvastatin treated HepG2 cells.

https://doi.org/10.1371/journal.pone.0105836.s015

(XLS)

Table S3.

A comprehensive list of the 98 differently expressed splice variants identified by RNA-seq after atorvastatin treatment in HepG2 cells (cut off: l1.2l fold change, 5% FDR).

https://doi.org/10.1371/journal.pone.0105836.s016

(DOC)

Acknowledgments

We thank the Norwegian Sequencing Centre (www.sequencing.uio.no) a national technology platform for providing the sequencing service.

Author Contributions

Conceived and designed the experiments: CS MKK DS JPB APP. Performed the experiments: CS RL OKO. Analyzed the data: CS RL OKO DS JPB APP. Contributed reagents/materials/analysis tools: CS RL OKO. Wrote the paper: CS MKK RL OKO DS JPB APP.

References

  1. 1. Heart Protection Study Collaborative Group (2002) MRC/BHF Heart Protection Study of cholesterol lowering with simvastatin in 20,536 high-risk individuals: a randomised placebo-controlled trial. Lancet 360: 7–22 S0140-6736(02)09327-3 [pii];10.1016/S0140-6736(02)09327-3 [doi].
  2. 2. Shepherd J, Blauw GJ, Murphy MB, Bollen EL, Buckley BM, et al.. (2002) Pravastatin in elderly individuals at risk of vascular disease (PROSPER): a randomised controlled trial. Lancet 360: : 1623–1630. S014067360211600X [pii].
  3. 3. Schwartz GG, Olsson AG, Ezekowitz MD, Ganz P, Oliver MF, et al.. (2001) Effects of atorvastatin on early recurrent ischemic events in acute coronary syndromes: the MIRACL study: a randomized controlled trial. JAMA 285: : 1711–1718. joc10254 [pii].
  4. 4. Gelissen IC, Brown AJ (2012) Predicting response to statins by pharmacogenetic testing. Pharmacogenomics 13: 1223–1225 10.2217/pgs.12.107 [doi].
  5. 5. Medina MW, Gao F, Ruan W, Rotter JI, Krauss RM (2008) Alternative splicing of 3-hydroxy-3-methylglutaryl coenzyme A reductase is associated with plasma low-density lipoprotein cholesterol response to simvastatin. Circulation 118: 355–362 CIRCULATIONAHA.108.773267 [pii];10.1161/CIRCULATIONAHA.108.773267 [doi].
  6. 6. Thompson PD, Clarkson P, Karas RH (2003) Statin-associated myopathy. JAMA 289: 1681–1690 10.1001/jama.289.13.1681 [doi];289/13/1681 [pii].
  7. 7. Liao JK, Laufs U (2005) Pleiotropic effects of statins. Annu Rev Pharmacol Toxicol 45: 89–118 10.1146/annurev.pharmtox.45.120403.095748 [doi].
  8. 8. Leszczynska A, Gora M, Plochocka D, Hoser G, Szkopinska A, et al.. (2011) Different statins produce highly divergent changes in gene expression profiles of human hepatoma cells: a pilot study. Acta Biochim Pol 58: : 635–639. 20110165 [pii].
  9. 9. Wang Y, Chang H, Zou J, Jin X, Qi Z (2011) The effect of atorvastatin on mRNA levels of inflammatory genes expression in human peripheral blood lymphocytes by DNA microarray. Biomed Pharmacother 65: 118–122 S0753-3322(10)00221-0 [pii];10.1016/j.biopha.2010.12.005 [doi].
  10. 10. Yu JG, Sewright K, Hubal MJ, Liu JX, Schwartz LM, et al.. (2009) Investigation of gene expression in C(2)C(12) myotubes following simvastatin application and mechanical strain. J Atheroscler Thromb 16: : 21–29. JST.JSTAGE/jat/E551 [pii].
  11. 11. Kohro T, Yamazaki T (2009) Mechanism of statin-induced myopathy investigated using microarray technology. J Atheroscler Thromb 16: : 30–32. JST.JSTAGE/jat/E812 [pii].
  12. 12. Morikawa S, Takabe W, Mataki C, Wada Y, Izumi A, et al. (2004) Global analysis of RNA expression profile in human vascular cells treated with statins. J Atheroscler Thromb 11: 62–72.
  13. 13. Morikawa S, Murakami T, Yamazaki H, Izumi A, Saito Y, et al.. (2005) Analysis of the global RNA expression profiles of skeletal muscle cells treated with statins. J Atheroscler Thromb 12: : 121–131. JST.JSTAGE/jat/12.121 [pii].
  14. 14. Pollack JR (2009) DNA microarray technology. Introduction. Methods Mol Biol 556: 1–6 10.1007/978-1-60327-192-9_1 [doi].
  15. 15. Malone JH, Oliver B (2011) Microarrays, deep sequencing and the true measure of the transcriptome. BMC Biol 9: 34 1741–7007-9-34 [pii];10.1186/1741-7007-9-34 [doi].
  16. 16. Gerber R, Ryan JD, Clark DS (2004) Cell-based screen of HMG-CoA reductase inhibitors and expression regulators using LC-MS. Anal Biochem 329: 28–34 10.1016/j.ab.2004.03.023 [doi];S0003269704002556 [pii].
  17. 17. Maeda A, Yano T, Itoh Y, Kakumori M, Kubota T, et al. (2010) Down-regulation of RhoA is involved in the cytotoxic action of lipophilic statins in HepG2 cells. Atherosclerosis 208: 112–118 S0021-9150(09)00597-8 [pii];10.1016/j.atherosclerosis.2009.07.033 [doi].
  18. 18. Mullen PJ, Luscher B, Scharnagl H, Krahenbuhl S, Brecht K (2010) Effect of simvastatin on cholesterol metabolism in C2C12 myotubes and HepG2 cells, and consequences for statin-induced myopathy. Biochem Pharmacol 79: 1200–1209 S0006-2952(09)01065-X [pii];10.1016/j.bcp.2009.12.007 [doi].
  19. 19. Burkhardt R, Kenny EE, Lowe JK, Birkeland A, Josowitz R, et al. (2008) Common SNPs in HMGCR in micronesians and whites associated with LDL-cholesterol levels affect alternative splicing of exon13. Arterioscler Thromb Vasc Biol 28: 2078.
  20. 20. Medina MW, Gao F, Naidoo D, Rudel LL, Temel RE, et al. (2011) Coordinately regulated alternative splicing of genes involved in cholesterol biosynthesis and uptake. PLoS One 6: e19420 10.1371/journal.pone.0019420 [doi];PONE-D-11-00952 [pii].
  21. 21. Medina MW, Theusch E, Naidoo D, Bauzon F, Stevens K, et al. (2012) RHOA Is a Modulator of the Cholesterol-Lowering Effects of Statin. PLoS Genet 8: e1003058 10.1371/journal.pgen.1003058 [doi];PGENETICS-D-12-00028 [pii].
  22. 22. Zhu H, Tucker HM, Grear KE, Simpson JF, Manning AK, et al. (2007) A common polymorphism decreases low-density lipoprotein receptor exon 12 splicing efficiency and associates with increased cholesterol. Hum Mol Genet 16: 1765–1772 ddm124 [pii];10.1093/hmg/ddm124 [doi].
  23. 23. Kulseth MA, Berge KE, Bogsrud MP, Leren TP (2010) Analysis of LDLR mRNA in patients with familial hypercholesterolemia revealed a novel mutation in intron 14, which activates a cryptic splice site. J Hum Genet 55: 676–680 jhg201087 [pii];10.1038/jhg.2010.87 [doi].
  24. 24. Stormo C, Kringen MK, Grimholt RM, Berg JP, Piehler AP (2012) A novel 3-hydroxy-3-methylglutaryl-coenzyme A reductase (HMGCR) splice variant with an alternative exon 1 potentially encoding an extended N-terminus. BMC Mol Biol 13: 29 1471-2199-13-29 [pii];10.1186/1471-2199-13-29 [doi].
  25. 25. Trapnell C, Pachter L, Salzberg SL (2009) TopHat: discovering splice junctions with RNA-Seq. Bioinformatics 25: 1105–1111 btp120 [pii];10.1093/bioinformatics/btp120 [doi].
  26. 26. Trapnell C, Williams BA, Pertea G, Mortazavi A, Kwan G, et al. (2010) Transcript assembly and quantification by RNA-Seq reveals unannotated transcripts and isoform switching during cell differentiation. Nat Biotechnol 28: 511–515 nbt.1621 [pii];10.1038/nbt.1621 [doi].
  27. 27. Trapnell C, Roberts A, Goff L, Pertea G, Kim D, et al. (2012) Differential gene and transcript expression analysis of RNA-seq experiments with TopHat and Cufflinks. Nat Protoc 7: 562–578 nprot.2012.016 [pii];10.1038/nprot.2012.016 [doi].
  28. 28. R Core Team (2012) R: A Language and Environment for Statistical Computing.
  29. 29. Gentleman RC, Carey VJ, Bates DM, Bolstad B, Dettling M, et al. (2004) Bioconductor: open software development for computational biology and bioinformatics. Genome Biol 5: R80 gb-2004-5-10-r80 [pii];10.1186/gb-2004-5-10-r80 [doi].
  30. 30. Edgar R, Domrachev M, Lash AE (2002) Gene Expression Omnibus: NCBI gene expression and hybridization array data repository. Nucleic Acids Res 30: 207–210.
  31. 31. Pfaffl MW (2001) A new mathematical model for relative quantification in real-time RT-PCR. Nucleic Acids Res 29: e45.
  32. 32. Crown Human Genome Center (2014) The GeneCards Human Gene Database. http://www.genecards.org/cgi-bin/carddisp.pl?id=SP1. Accessed 31 July 2014
  33. 33. Crown Human Genome Center (2014) The GeneCards Human Gene Database. http://www.genecards.org/cgi-bin/carddisp.pl?id=ZNF195. Accessed 31 July 2014
  34. 34. Crown Human Genome Center (2014) The GeneCards Human Gene Database. http://www.genecards.org/cgi-bin/carddisp.pl?id=DTNA. Accessed 31 July 2014
  35. 35. Crown Human Genome Center (2014) The GeneCards Human Gene Database. http://www.genecards.org/cgi-bin/carddisp.pl?id=FAM189. Accessed 31 July 2014
  36. 36. Wang Z, Gerstein M, Snyder M (2009) RNA-Seq: a revolutionary tool for transcriptomics. Nat Rev Genet 10: 57–63 nrg2484 [pii];10.1038/nrg2484 [doi].
  37. 37. Trapnell C, Pachter L, Salzberg SL (2009) TopHat: discovering splice junctions with RNA-Seq. Bioinformatics 25: 1105–1111 btp120 [pii];10.1093/bioinformatics/btp120 [doi].
  38. 38. Marioni JC, Mason CE, Mane SM, Stephens M, Gilad Y (2008) RNA-seq: an assessment of technical reproducibility and comparison with gene expression arrays. Genome Res 18: 1509–1517 gr.079558.108 [pii];10.1101/gr.079558.108 [doi].
  39. 39. Liu S, Lin L, Jiang P, Wang D, Xing Y (2011) A comparison of RNA-Seq and high-density exon array for detecting differential gene expression between closely related species. Nucleic Acids Res 39: 578–588 gkq817 [pii];10.1093/nar/gkq817 [doi].
  40. 40. 't Hoen PA, Ariyurek Y, Thygesen HH, Vreugdenhil E, Vossen RH, et al. (2008) Deep sequencing-based expression analysis shows major advances in robustness, resolution and inter-lab portability over five microarray platforms. Nucleic Acids Res 36: e141 gkn705 [pii];10.1093/nar/gkn705 [doi].
  41. 41. Raghavachari N, Barb J, Yang Y, Liu P, Woodhouse K, et al. (2012) A systematic comparison and evaluation of high density exon arrays and RNA-seq technology used to unravel the peripheral blood transcriptome of sickle cell disease. BMC Med Genomics 5: 28 1755-8794-5-28 [pii];10.1186/1755-8794-5-28 [doi].
  42. 42. Ridker PM, Danielson E, Fonseca FA, Genest J, Gotto AM, et al. (2008) Rosuvastatin to prevent vascular events in men and women with elevated C-reactive protein. N Engl J Med 359: 2195–2207 NEJMoa0807646 [pii];10.1056/NEJMoa0807646 [doi].
  43. 43. Bonetti PO, Lerman LO, Napoli C, Lerman A (2003) Statin effects beyond lipid lowering—are they clinically relevant? Eur Heart J 24: : 225–248. S0195668X02004190 [pii].
  44. 44. Skottheim IB, Gedde-Dahl A, Hejazifar S, Hoel K, Asberg A (2008) Statin induced myotoxicity: the lactone forms are more potent than the acid forms in human skeletal muscle cells in vitro. Eur J Pharm Sci 33: 317–325 S0928-0987(08)00002-X [pii];10.1016/j.ejps.2007.12.009 [doi].
  45. 45. Boerma M, Fu Q, Wang J, Loose DS, Bartolozzi A, et al. (2008) Comparative gene expression profiling in three primary human cell lines after treatment with a novel inhibitor of Rho kinase or atorvastatin. Blood Coagul Fibrinolysis 19: 709–718 10.1097/MBC.0b013e32830b2891 [doi];00001721-200810000-00017 [pii].
  46. 46. Gao Y, Lu XC, Yang HY, Liu XF, Cao J, et al. (2012) The molecular mechanism of the anticancer effect of atorvastatin: DNA microarray and bioinformatic analyses. Int J Mol Med 30: 765–774 10.3892/ijmm.2012.1054 [doi].