A large haplotype on chromosome 19p13.11 tagged by rs10401969 in intron 8 of SURP and G patch domain containing 1 (SUGP1) is associated with coronary artery disease (CAD), plasma LDL cholesterol levels, and other energy metabolism phenotypes. Recent studies have suggested that TM6SF2 is the causal gene within the locus, but we postulated that this locus could harbor additional CAD risk genes, including the putative splicing factor SUGP1. Indeed, we found that rs10401969 regulates SUGP1 exon 8 skipping, causing non-sense-mediated mRNA decay. Hepatic Sugp1 overexpression in CD1 male mice increased plasma cholesterol levels 20–50%. In human hepatoma cell lines, SUGP1 knockdown stimulated 3-hydroxy-3-methylglutaryl-CoA reductase (HMGCR) alternative splicing and decreased HMGCR transcript stability, thus reducing cholesterol synthesis and increasing LDL uptake. Our results strongly support a role for SUGP1 as a novel regulator of cholesterol metabolism and suggest that it contributes to the relationship between rs10401969 and plasma cholesterol.
In the largest genome-wide association study meta-analysis of lipids published to date, only ten loci were associated with both LDL cholesterol (LDLC) and triglycerides (TG) levels significantly (1). Of these, a locus on Chr19p13.11 has also been associated with incidence of coronary artery disease (CAD), as shown in both case–control and prospective cohort studies of Caucasians and Asians (1–3). The haplotype in this locus tagged by the C minor allele of rs10401969 has been associated with reduced plasma LDLC and/or TG in African American, Hispanic and Native American populations (3–10), as well as in several populations of European ancestry (1,4,11–14). This locus has also been associated with Type 2 diabetes (11,15,16), hepatic steatosis, non-alcoholic fatty liver disease (17–20) and body mass index (21), suggesting that genetic variation within this region has a broad impact on human metabolism.
Recently, three studies have implicated TM6SF2 as the causative gene within the Chr19p13.11 locus. Exome sequencing studies identified an association between rs58542926, a non-synonymous (Glu167Lys) variant in TM6SF2, and risk of myocardial infarction and non-alcoholic fatty liver disease (22,23). Expression of the recombinant TM6SF2 variant encoding p.Glu167Lys in HepG2 cells resulted in 50% less protein compared to wildtype. Tm6sf2 knockdown in an AAV murine model (23) increased hepatic TG and decreased very low-density lipoprotein (VLDL) secretion, consistent with the effects reported in vitro (24).
Notably, rs58542926 is in very strong linkage disequilibrium (LD) with rs10401969 in Europeans (r2 = 0.95 and D′ = 1 in 503 EUR, 1000 Genomes Project, Phase 3). In fact, in individuals of European descent, rs10401969 (EUR MAF = 7.1%) resides in a large LD block that extends ∼360 kb across at least 14 genes and contains 32 other genetic variants that are all in strong LD (r2 > 0.8) with rs10410969. Although it may be simplest to attribute all of the phenotypic associations with the 19p13.11 haplotype described above to one causal gene, such as TM6SF2, the complex genetic architecture of this region suggests other possibilities. For instance, any number of the 33 variants in strong LD (and perhaps rarer SNPs, via synthetic association) (25) could have functional effects that contribute to the associated phenotypes, perhaps by acting on different genes. In addition, even a single variant could impact multiple genes that contribute to the same phenotype, especially because functionally related genes are often clustered in the genome. Consistent with these possibilities, Blattmann et al. (26) reported that knockdown of at least six genes in the Chr19p13.11 locus in Hela-Kyoto cells impacted DiI-LDL uptake, intracellular cholesterol levels or both. rs10401969 is located within an intron of SURP and G patch domain containing 1 (SUGP1), previously known as splicing factor 4 (SF4). Because splicing is a mechanism that regulates cholesterol homeostasis (27) and genetic variants associated with variation in plasma LDLC levels can influence splicing of the key cholesterol metabolism genes 3-hydroxy-3-methylglutaryl-CoA reductase (HMGCR) and low-density lipoprotein receptor (LDLR) (28,29), we sought to investigate whether genetic variation within a splicing factor itself (30) can affect cellular cholesterol homeostasis. Thus, here we investigate whether SUGP1 may contribute to the relationship between rs10401969 and plasma lipids through the regulation of alternative splicing.
SUGP1 impacts cholesterol metabolism in vitro
To determine whether SUGP1 impacts cellular cholesterol metabolism, we transfected human hepatoma HepG2 and Huh7 cell lines with siRNAs targeting SUGP1, which reduced SUGP1 transcript levels by 45–70% (n = 12) and protein levels by 72–91% (n = 2) (Fig. 1A and B). SUGP1 knockdown increased rates of DiI-LDL uptake 10–15% (Fig. 1C) and reduced APOB in the cell culture media (Fig. 1D), but did not alter APOE concentrations (Fig. 1E) and HMGCR enzyme activity ∼20% (Fig. 1F). Consistent with this observation, we also found that SUGP1 transcript levels were positively correlated with variation in HMGCR enzyme activity in 118 human lymphoblastoid cell lines (P = 0.0215, r2 = 0.04; Fig. 1G). To demonstrate the specificity of these effects, we performed a rescue experiment in which SUGP1 was overexpressed in HepG2 cells after SUGP1 knockdown (Fig. 1H) and APOB in the media (Fig. 1I). Similarly, SUGP1 overexpression alone also reduced rates of DiI-LDL uptake (31), did not produce consistent and statistically significant changes in measures of cellular cholesterol metabolism (
SUGP1 alters cholesterol metabolism in vivo
To test the physiological relevance of SUGP1, we delivered murine Sugp1 to the liver by hydrodynamic tail vein injection in CD-1 mice. After 7 days, hepatic overexpression of Sugp1 was assessed. Sugp1 transcript and protein levels were increased (Fig. 2A and B andFig. 2C). Similar effects were also observed after 28 days of Sugp1 overexpression (Fig. 2D). Consistent with in vitro results, Sugp1 overexpression resulted in greater hepatic Hmgcr enzyme activity (Fig. 2E), whereas we observed no change in hepatic Hmgcr transcript levels (
SUGP1 knockdown increases expression of alternatively spliced HMGCR transcripts
Given the effects of SUGP1 knockdown on LDL uptake and HMGCR enzyme activity, and its putative role as a splicing factor, we postulated that SUGP1 might directly affect LDLR or HMGCR transcript structure. Because alternative splicing of LDLR is extensive and highly varied, we tested the four major LDLR splice variants, specifically, those lacking exons 3, 4, 12 or 14 (28). SUGP1 knockdown did not produce consistent changes in expression levels of any of these LDLR transcripts in vitro (
To assess effects of SUGP1 on HMGCR structure, we first identified HMGCR splice variants expressed in primary liver tissue derived from human donors (n = 3), human hepatoma cell lines (HepG2, Hep3B and Huh7) and lymphoblastoid cell lines (LCLs) (n = 8) by RNAseq. In addition to the known HMGCR splice variants, which lack exon 13, HMGCR13(−), or exon 18, HMGCR18(−), we also detected transcripts with shortened versions of exons 11 (ex11b) and 18 (ex18b). Ex11b is missing 93 bp at its 5′ end, whereas ex18b is missing 45 bp at its 3′ end (Fig. 3A–C). Although HMGCR11b and HMGCR18b have a similar level of abundance, RT-PCR with primers specific for ex11b to ex18b failed to amplify a product (Fig. 3D). In contrast, there were no consistent changes in HMGCR13(−), the most abundant HMGCR splice variant, and a trend of increased HMGCR13(+). We found detectable expression levels of these splice variants in the mouse, and consistent with these effects, increased Hmgcr alternative splicing was observed after Sugp1 overexpression in vivo (Fig. 2F).
To determine whether these novel HMGCR splice variants are subject to non-sense-mediated decay (NMD), we treated HepG2 cells with cycloheximide, an inhibitor of NMD, and quantified HMGCR splice variants over 3 hours. HMGCR18(−) levels dramatically increased, whereas HMGCR11b and HMGCR18b transcripts levels remained unchanged, indicating that HMGCR18(−) abundance is likely regulated by NMD (
SUGP1 regulates HMGCR alternative splicing and transcript stability
SUGP1 is predicted to function as a splicing factor (30). However, to date, the function of SUGP1 has not yet been validated. To determine whether SUGP1 regulates HMGCR alternative splicing, HepG2 and Huh7 cell lines were transfected with a mini-gene containing the HMGCR genomic sequence from intron 17 to 19 (mHMGCR17-19) following SUGP1 knockdown (Fig. 4A). HMGCR transcripts derived from the mini-gene were quantified by quantitative PCR. SUGP1 knockdown increased the relative expression levels of the mini-gene-derived HMGCR18b and HMGCR18(−) transcripts, compared to HMGCR18(+) transcripts in vitro (Fig. 4B).
We previously reported that HNRNPA1, another RNA binding protein, increases the relative abundance of HMGCR13(−) transcript compared to HMGCR13(+) transcript by promoting alternative splicing as well as preferentially stabilizing the HMGCR13(−) transcript (32). To test whether SUGP1 also impacts HMGCR transcript stability, we treated HepG2 and Huh7 cell lines with actinomycin D upon SUGP1 knockdown and measured endogenous HMGCR transcripts. SUGP1 knockdown dramatically reduced HMGCR transcript half-lives (Fig. 4C andFig. 4D). No decay of HMGCR18(−) was detected after actinomycin D treatment, and thus half-life was not determined (
rs10401969 modulates SUGP1 alternative splicing and protein levels
rs10401969, despite being located in SUGP1 intron 8, has not been associated with SUGP1 transcript levels in any publically available eQTL dataset, including liver (33). Thus, we postulated that rs10401969 may influence SUGP1 transcript structure. Using RT-PCR and Sanger sequencing of cDNA from human hepatoma RNA, we detected a novel SUGP1 splice variant lacking exon 8, SUGP1 exon 8(−) (Fig. 5A), whereas no difference in overall SUGP1 expression levels was detected. Similar trends were detected in human livers derived from the St. Jude’s Liver Repository (Fig. 5B). In addition, we found that the rs10401969 C allele carriers also had reduced levels of SUGP1 protein compared to TT homozygotes (P = 0.01, n = 60, Fig. 5C and
GWAS-identified SNPs are often assumed to be in LD with functionally relevant SNPs and not the causal variant themselves. Although rs10401969 is 80 base pairs downstream from the exon 8 splice donor, there are no other SNPs in high LD (r2 > 0.9) with rs10401969 that are predicted to modulate exon 8 skipping. We tested whether rs10401969 directly alters exon 8 alternative splicing using an SUGP1 mini-gene (mSUGP1) construct (Fig. 5D). Transcripts derived from the mini-gene with the rs10401969 C allele had a higher frequency of exon 8 skipping than those expressed from the T allele (Fig. 5E), consistent with our observations in the CAP LCLs. Using Human Splicing Finder (34), we noted that rs10401969 lies immediately adjacent to a predicted binding site for HNRNPA1, a splicing factor that we have previously implicated in cholesterol metabolism (32). Using an RNA electrophoretic mobility shift assay with probes containing SUGP1 intron 8 with either the rs10401969 C or T allele, we found that the C allele had 27% stronger binding affinity to HNRNPA1 protein compared to the T allele as indicated by reduced intensity of the free probe (Fig. 5F andFig. 5F and
Inter-individual variation in SUGP1 transcript levels is correlated with variation in HMGCR alternative splicing in an rs10401969 allele-dependent fashion
To determine whether endogenous variation in SUGP1 expression was associated with variation in HMGCR alternative splicing, we quantified SUGP1 and HMGCR splice variants in 57 LCLs. We found evidence of a positive correlation between SUGP1 transcript levels with variation in HMGCR11b/Total HMGCR (P = 0.0045, r2 = 0.14) and HMGCR18b/Total HMGCR ratios (P = 0.0018, r2 = 0.16), whereas there was no association with either HMGCR13(−)/Total HMGCR or HMGCR18(−)/Total HMGCR ratios. Interestingly, we found that this relationship was modified by rs10401969. As shown in Figure 6, correlations between SUGP1 and HMGCR11b/Total (P = 6.3e −5, r2 = 0.51) or HMGCR18b/Total (P = 0.0002, r2 = 0.46) were observed in rs10401969 minor allele carriers only (n = 25), with statistically significant interaction terms.
Here, we sought to determine whether SUGP1 contributes to the relationship between rs10401969 and plasma lipids through regulation of alternative splicing. Through cellular and animal studies, we found that SUGP1 is a novel regulator of cholesterol metabolism that post-transcriptionally modifies HMGCR, impacting cholesterol synthesis. In addition, we show that rs10401969 is a functional variant, regulating SUGP1 alternative splicing. Together, these results implicate SUGP1 as a contributor to inter-individual variation in plasma cholesterol.
SUGP1 was first identified as a putative splicing factor based on domain composition analysis (30). The SUGP1 protein is predicted to contain two SURP motifs, domains of alternative splicing regulators thought to be involved in RNA binding (35), as well as a G-patch domain, a region of six highly conserved glycine residues commonly found in eukaryotic RNA-processing proteins (36). Although SUGP1 has been shown to co-purify with components of the spliceosome in multiple studies (37–39), prior to this report, no functional studies of SUGP1 have been published. Here, we confirm that manipulation of SUGP1 does lead to changes in the structure and stability of some transcripts, consistent with its annotation as an RNA regulatory protein.
Importantly, we found that one of SUGP1’s targets is HMGCR, a key cholesterol regulatory gene. SUGP1 knockdown increased expression levels of several alternatively spliced HMGCR transcripts, HMGCR11b, HMGCR18b and HMGCR18(−) by both promoting exon skipping and reducing transcript stability in a splice variant-dependent fashion. Although the open reading frame is retained in these splice variants, they omit a portion of the catalytic domain, and if translated we predict these variants to encode HMGCR isoforms with reduced or abolished enzyme activity. We found that in vitro SUGP1 knockdown reduced HMGCR enzyme activity, rate of cholesterol synthesis, and APOB in the culture media, as well as increased LDL uptake. Consistent with these effects, we observed elevated plasma cholesterol and hepatic Hmgcr expression, which could be attributed to the increased hepatic triglyceride secretion, elevated Hmgcr enzyme activity and reduced expression of Hmgcr splice variants after Sugp1 overexpression in vivo. Together, these data strongly support the likelihood that SUGP1 is a novel cholesterol regulatory gene.
We also demonstrate that rs10401969, a GWAS identified SNP associated with plasma lipids, CAD and hepatic steatosis (4,22,23,40,41), regulates SUGP1 alternative splicing. Located 80 bp upstream of the exon 8 splice donor, we found that the rs10401969 C allele promotes binding of HNRNPA1, a well known factor that stimulates alternative splicing, and induces skipping of SUGP1 exon 8. Loss of exon 8 disrupts the open reading frame, stimulating non-sense-mediated decay of the alternatively spliced transcript. SUGP1 knockdown both increases HMGCR alternative splicing and reduces overall HMGCR transcript stability, consistent with our observation that the correlation between inter-individual variation in SUGP1 expression quantified in the LCLs with variation in HMGCR alternative splicing differs by rs10401969 genotype. Together, these observations support the likelihood that rs10401969 itself is a functional variant that affects SUGP1 regulation of HMGCR. Because HMGCR encodes the rate-limiting step of the cholesterol biosynthesis pathway, knockdown also reduced cholesterol synthesis in vitro, whereas overexpression increased plasma cholesterol in vivo. These relationships are consistent with many GWAS studies showing an association between the rs10401969 C allele and reduced plasma cholesterol (
These results suggest that SUGP1-induced changes in cholesterol metabolism are mediated by the regulation of HMGCR; however, RNA-binding proteins are not usually specific for a single target gene. For example, although it is tempting to conclude that the reduced cholesterol synthesis stimulates LDL uptake by promoting SREBF2 activation, there is no evidence to suggest that these are cause and effect. We failed to observe increased expression levels of SREBF2 target genes including MVK, LDLR and PCSK9 (data not shown) after SUGP1 knockdown in vitro, indicating a lack of SREBF2 activation. Although greater hepatic Hmgcr transcript levels (another SREBF2 target gene) were detected after Sugp1 overexpression in vivo, this up-regulation could be attributed to direct effects of Sugp1 on Hmgcr transcript stability. Further study is required to determine whether SUGP1 targets other key cholesterol regulators beyond HMGCR and contributes to the overall cholesterol phenotype observed.
As previously described, rs10401969 is in a large LD block containing many genetic variants in individuals of European ancestry, and one or more of these variants could have a functional impact. Additional fine mapping and conditional association analyses in ethnic populations with smaller haplotypic blocks in this region could help reveal the number of functional variants that contribute to the genetic association signals in individuals of European ancestry. In particular, rs10401969 is in strong LD with a functional variant in TM6SF2, p.Glu167Lys. A trio of recent studies utilizing human genetics and functional studies in cellular and animal models all implicate TM6SF2 as a causative functional gene within the Chr19.13p11 locus. Using an AAV model, Holzmen et al.(2014) found that overexpression of human TM6SF2 in C57BL/6J mice increased plasma TC, LDLC and triglycerides and reduced HDL cholesterol after 5 days. Consistent with these findings, Kozlitina et al.(23) reported that Tm6sf2 knockdown reduced plasma cholesterol and triglycerides after 8 weeks, and the effect was attributed to a defect in VLDL secretion. TM6SF2 knockdown was also shown to reduce secretion of triglyceride rich lipoproteins (24), whereas TM6SF2 overexpression was reported to increase cholesterol synthesis (42) in human hepatoma cell lines. Notably, these studies suggest that TM6SF2 is the only causative gene within the locus. Conditional analyses found that the TM6SF2 p.Glu167Lys variant remained significantly associated with plasma LDLC and triglycerides when controlled for other SNPs within the locus, whereas the signal of these other SNPs (including rs10401969) was abolished after controlling for the TM6SF2 variant (22,23). In addition, using a panel of >200 human livers, variation in the expression of TM6SF2, but not any of the other 18 genes tested within the locus, was correlated with both rs10401969 genotype and variation in hepatic triglyceride content (24).
Given these findings, it is challenging to interpret our results and the relevance of SUGP1 and rs10401969. Because the function of TM6SF2 has not been fully established, it is possible that the effect of TM6SF2 on lipid metabolism may be mediated (at least in part) by SUGP1 or vice versa. However, in preliminary studies, we have failed to identify any direct effect of TM6SF2 knockdown on SUGP1 expression or vice versa (data not shown). In addition, LCLs do not express TM6SF2. Thus, the relationships we observed between SUGP1 and HMGCR enzyme activity and alternative splicing using these cells cannot be attributed to a potential effect of either the TM6SF2 SNP or gene. In addition, although our results do not dispute the importance of the TM6SF2 gene or variant as a causative gene within the 19p13.11 locus, the consistency in our findings across cellular, animal and genetic analyses supports the conclusion that SUGP1 is a novel cholesterol regulator. Finally, although the conditional genetic analyses suggest that the effect of rs10401969 alone is not sufficient to impact plasma lipids in vivo, they do not discount the possibility that rs10401969 may act in an additive or synergistic manner with the TM6SF2 SNP, especially given the high LD of the two SNPs. The minor allele of each SNP reduces SUGP1 or TM6SF2 protein levels, and these two proteins have concordant effects on plasma cholesterol as overexpression of each gene alone increases plasma cholesterol. Thus, the relationships observed between genetic variation within this locus and plasma cholesterol may be due to the combined effect of SUGP1 and TM6SF2.
In summary, our results have shown SUGP1 to be a multi-functional RNA regulatory protein that modulates expression levels of several HMGCR splice variants, altering HMGCR enzyme activity and impacting both cellular and in vivo cholesterol metabolism. Furthermore, the finding that rs10401969, a GWAS-identified SNP associated with plasma LDLC and CAD, regulates SUGP1 implicates SUGP1 as a potential contributor to this genetic association. Overall, the present results reinforce the role of genetic regulation of transcript structure in pathways affecting cholesterol metabolism and cardiovascular disease risk.
Materials and Methods
HepG2, Hep3B and Huh7 cells were obtained from the American Type Culture Collection and grown under standard culture conditions. LDL was isolated as previously described (43). Cells were exposed in replicate to conditions of sterol depletion (10% lipoprotein deficient serum and 2 µm activated simvastatin), and after 24 hours, 50 µg/ml LDLC was added for an additional 24 hours. Immortalized LCLs (n = 54) were derived from donors of the CAP clinical trial and grown as previously described (44).
siRNA transfections and cellular measurements
SUGP1, ATP13A1 and MAU2 knockdown were achieved using the Ambion Silence Select siRNAs (SUGP1-1: s33721, SUGP1-3: s33722, ATP13A1: s32750, MAU2: s225955) or non-targeting control (NTC: AM4611) according to the manufacturer’s protocol. Forty-eight hours after knockdown, RNA was isolated. cDNA was synthesized and transcripts were quantified by quantitative PCR with assays shown in44). For rescue experiments, cells were transfected in replicate with SUGP1-1 siRNA or NTC as described above. Forty-eight hours post transfection, cells SUGP1-1 siRNA treated cells were transfected with either the SUGP1 overexpression plasmid (pCMV-SUGP1) or an empty vector control (pCMV) and incubated for an additional 48 hours. APOB and APOE were quantified in triplicate by sandwich-style ELISA in culture media. HMGCR enzyme activity and DiI-LDL uptake were quantified as previously described (44,45). Cholesterol synthesis was measured by the conversion of [1-13C] sodium acetate to 13C-labeled cholesterol as described (46). Analysis of variance with post hoc two-tailed paired t-tests were used to determine statistically significant differences between treatments.
In vivo study
An Sugp1 overexpression construct was created by subcloning the murine Sugp1 gene (purchased from Origene) into the pLIVE-MCS vector (Mirus Bio). Sugp1-pLIVE was injected (25 μg) into the tail vein of 6-week-old CD-1 male mice (n = 10; Charles River Laboratories) using the TransIT-EE hydrodynamic gene delivery system (Mirus Bio), as previously described (47). In a second set of CD-1 male mice (n = 10), the empty vector was injected to serve as the negative control. Five mice for each construct were euthanized at 7 and 28 days after injection by cardiac puncture. Animal studies were approved by the Children’s Hospital Oakland Research Institute Animal Care and Use Committee and conform to all regulatory standards. Whole blood, liver and plasma were collected. Plasma cholesterol was quantified using a LIASYS 330 chemistry analyzer (AMS Diagnostics). Hepatic gene expression levels were quantified as described above, with values normalized to the geomean of Clptm1 and Rplp0. All RNAs were DNaseI treated during isolation. Hepatic Sugp1 protein levels were quantified by Western blot analysis.
RNA electrophoretic mobility gel shift assay (RNA EMSA)
3′ end-biotinylated RNA oligonucleotides were synthesized from Sigma-Aldrich. Purified His-tagged human recombinant HNRNPA1 and SRSF1 proteins were purchased from Prospec and ProteinOne. RNA gel shift assay was performed using LightShift Chemiluminescent RNA EMSA kit (Thermo Scientific). RNA oligos with either the SUGP1 rs10401969 ‘T’ or ‘C’ allele were incubated for 30 minutes in the presence or in the absence of either His-HNRNPA1 or His-SRSF1 protein in 1x binding buffer, 5% glycerol, 2 µg tRNA, and 6.25 nm biotin-labeled RNA oligos. 5x loading dye was added to each reaction, loaded onto a 4–20% Novex® TBE Gels (Life Technologies), and run at 100 V for 2 hours in 0.5x Novex® TBE Running Buffer (Invitrogen) at 4 °C. The RNA–protein complexes were transferred from the gel to a nylon membrane (Thermo Scientific) at 35V for 40 minutes in a 4 °C cold room, and crosslinked using the Stratalinker UV crosslinker (Stratagene). Biotin-labeled RNA oligos were detected by the Chemiluminescent Nucleic Acid detection Module (Thermo Scientific), and quantified using the GelQuant.NET software.
SUGP1 and HMGCR mini-gene construct
To create the SUGP1 and HMGCR mini-genes, SUGP1 introns 7–8 and HMGCR introns 16–19 were cloned into pDEST exon-trap vector provided by Dr. Stephan Stamm. The plasmids were DNA sequence verified. The SUGP1 mini-gene was co-transfected into HepG2 and Huh7 cells with the HNRNPA1 overexpression plasmid. The HMGCR mini-gene was co-transfected one of the two SUGP1 targeting siRNA or NTC.
Identification of novel HMGCR splice variants
Whole transcriptome sequencing (RNA-seq) was performed on LCLs from eight African American individuals (seven female and one male) as previously described (48). Libraries were sequenced on an Illumina GAII machine and aligned to hg19 using TopHat v1.2.0 and Bowtie v0.12.7 (49,50). Splice junctions detected by TopHat were compared to Ensembl v61 annotations to identify potentially novel junctions in HMGCR. From this analysis, two novel splice variants were identified with truncated versions of exons 11 and 18 using both direct evidence of junction spanning reads, and indirect evidence based the distance of paired end reads. Novel junctions were validated using RT-PCR with Sanger sequencing. The effect of SUGP1 on the stability of these HMGCR transcripts was assessed in HepG2 and Huh7 cells incubated with actinomycin D, with transcript half-life calculated as previously described (48).
The CAP LCLs used in these studies were provided by Sheila Pressman and Dr. Jerome I. Rotter. Human liver RNA-seq data was generated through the PGRN Network-wide RNA-Seq Project. We thank Dr. Ronald M. Krauss for his scientific discussion and critical review of this manuscript.
Conflict of Interest statement. None declared.
This work was supported by the National Institutes of Health [R01 HL104133], [U19 HL069757], [P50 GM115318], and the American Heart Association [12POST10430005]. Funding to pay the Open Access publication charges for this article was provided by NIH R01 HL104133.
- coronary arteriosclerosis
- ldl cholesterol lipoproteins
- hepatic tumor
- alternative splicing
- carcinoma, hepatocellular
- cell line
- coenzyme a
- rna splicing
- rna, messenger
- cholesterol measurement test
- malignant neoplasm of liver
- protein overexpression
- cholesterol biosynthesis
- cholesterol metabolism