|Home | About | Journals | Submit | Contact Us | Français|
CES1 encodes carboxylesterase-1, an important drug metabolizing enzyme with high expression in the liver. Previous studies have demonstrated a genomic translocation of the 5′ region from the poorly expressed pseudogene CES1P1, to CES1, yielding structural variant CES1VAR. The purpose of this study was to characterize this translocation and its effect on CES1 expression in human liver.
Experiments were performed in human liver tissues and cell culture (HepG2). The promoter and exon-1 of CES1 were sequenced by Sanger and Ion Torrent sequencing to identify gene translocations. The effect of CES1 5′UTRs on mRNA and protein expression were assessed by quantitative real-time-PCR, allelic ratio mRNA analysis by primer extension (SNaPshot), quantitative targeted proteomics, and luciferase reporter gene assays.
Sequencing of CES1 identified two translocations. First, CES1VAR (17% minor allele frequency, MAF) comprising the 5′UTR, exon-1, and part of intron-1. A second shorter translocation, CES1SVAR, was observed excluding exon-1 and intron-1 regions (<0.01% MAF). CES1VAR associates with 2.6-fold decreased CES1 mRNA and ~1.35-fold lower allelic mRNA. Luciferase reporter constructs demonstrate CES1VAR decreases luciferase activity 1.5-fold, while CES1SVAR slightly increases activity. CES1VAR was not associated with CES1 protein expression or metabolism of the CES1 substrates enalapril, clopidogrel or methylphenidate in liver.
The frequent translocation variant CES1VAR reduces mRNA expression of CES1 in liver by ~30%, but protein expression and metabolizing activity in liver were not detectably altered – possibly due to variable CES1 expression masking small allelic effects. Whether drug therapies are affected by CES1VAR will require further in vivo studies.
A member of the carboxylesterase family, carboxylesterase-1 (CES1) plays a significant role in the metabolism of drugs with ester or amide bonds in the human liver. Common drugs metabolized by CES1 include the anti-platelet prodrug clopidogrel , antivirals (oseltamivir) [2, 3], ADHD medications (methylphenidate) , chemotherapeutic agents (irinotecan) , ACE inhibitors (imidapril, enalapril, trandolapril, ramipril) [6–8] and others. CES1 is closely related to the pseudogene CES1P1, which constitutes less than 2% of total CES1 hepatic gene expression . Recent studies have shown that polymorphisms within CES1 can have significant effects on enzyme function and drug efficacy [10, 11]. In the case of clopidogrel, CES1 inactivates 85% of the drug, acting at three levels: the prodrug itself, an intermediate metabolite, and the active metabolite . The relatively rare coding variant rs71647871 (G143E, minor allele frequency (MAF) = 0.01–0.04) has been shown to ablate CES1 metabolic activity  and has been proposed as a biomarker to predict clopidogrel efficacy  as decreased CES1 metabolic activity leads to increased clopidogrel efficacy. Numerous additional variants have been associated with drug responses or changes in gene expression indicating CES1 has potential to impact a wide array of substrates (Table 1) [8, 11–14]. While regulatory variants affecting transcription and RNA biology have been identified for a number of drug metabolizing enzymes, no systematic study has been carried out with CES1 [15, 16].
Previous studies have characterized the structure of the CES1 gene family locus which consists of CES1 and CES1P1 in a tail-to-tail configuration (Figure 1) [9, 17]. Showing high sequence identity with CES1, the pseudogene CES1P1 expresses at low levels either a non-functional truncated mRNA or the protein coding CES1P1VAR mRNA. Sequence differences between CES1 and CES1P1 exist primarily in the 5′UTR, exon 1, and intron 1 regions, as well as at a premature stop codon within the exon 3 of CES1P1 (Figure 2, Table 2). CES1 consists of at least two alleles [9, 18], the wild-type and a variant allele carrying a translocation from CES1P1 (5′UTR, exon 1 and intron 1 regions) which replaces CES1 sequence. Referred to as CES1 variant (CES1VAR) (Figure 1, Figure 2, Table 2), this isoform consists of 11 highly linked singe nucleotide polymorphisms (SNPs) whose sequence matches the CES1P1VAR gene locus. Since the non-synonymous variants in the protein coding portion of exon 1 included in CES1VAR do not occur at CES1 catalytic sites, we hypothesized that CES1VAR expression could be affected by the acquired CES1P1VAR sequence, now driven by the expression of the functional CES1 promoter.
Molecular genetic studies testing variants within CES1 for function or association with changes in gene and protein expression are limited [8, 18]. Studies testing major CES1 variants (such as CES1VAR), have not typically found significant associations with drug response or toxicity [19, 20]. In one case CES1P1VAR expression has been related to the activation of irinotecan to CPT-11 . Further, an additional sequence variant similar to CES1VAR has also been identified but is limited to CES1P1 SNPs in the protein coding region of CES1 exon 1 . Adding complexity to this region is the inclusion of multiple transcription start sites along the annotated 5′UTR . It is unclear whether and to what extent CES1 translocation variation affects mRNA isoforms or expression.
The goal of this study is to better understand variation at the 5′ region of the CES1 gene locus. Further, we aim to define how this 5′ variation affects mRNA expression, mRNA isoforms, and protein expression, to determine whether CES1 5′UTR sequences such as CES1VAR can serve as biomarkers when CES1 is a primary metabolic enzyme determining drug response.
A total of 227 human liver samples were obtained from the Cooperative Human Tissue Network (Midwest and Western Division) and XenoTech LLC (Lenexa, KS), under approval of the Ohio State University (n=125, primarily Caucasians and 15% with African ancestry, 45% male and 55% female) and the University of Michigan (n=102, 92 Caucasians, 6 African-Americans, 2 Hispanics, and 2 classified as ‘others’, 45% male and 55% female) Institutional Review Boards. An additional fifty liver samples were obtained from other sources that included the Medical College of Virginia (Richmond, VA, USA), Medical College of Wisconsin (Milwaukee, WI, USA), Indiana University School of Medicine (Indianapolis, IN, USA), and University of Pittsburgh (Pittsburgh, PA, USA). All livers were used under protocols approved by committees for the conduct of human research at the Ohio State University and the University of Michigan.
Genomic DNA and RNA were prepared from liver tissue samples using either nuclei lysis buffer or Trizol respectively (Life Technologies, Carlsbad, CA) as previously described . Briefly, following tissue lysis genomic DNA samples were phenol chloroform purified, ethanol washed, and resuspended in TE. RNA isolation used SpinSmart Columns (Denville, Metuchen, NJ) with DNaseI treatment to prevent contamination of genomic DNA in cDNA synthesis. cDNA was prepared using RNA and Reverse Transcriptase SSIII (Invitrogen, San Francisco, CA, USA), with controls lacking reverse transcriptase or template to test for residual gDNA contamination. SNPs in CES1 were genotyped in liver samples using a primer extension assay (SNaPshot, Life Technologies), GC-clamp real time-PCR assay or fluorescently labeled PCR-restriction fragment length polymorphism (RFLP) analysis as previously described [21, 22]. PCR conditions and primers for all PCR assays (including genotyping) are provided in Supplemental Digital Content, Table 1.
Initially Sanger sequencing was used to screen the immediate 5′UTR, exon 1 and intron 1 of CES1 to confirm the CES1VAR genetic sequence. Following association of CES1VAR with changes in mRNA and protein expression, the CES1 promoter and upstream region (4.387 kb) were sequenced to scan for regulatory variants. This region included the annotated 5′UTR of CES1, through intron 1. PCR products were separated on agarose gel and the products mixed and fragmented on a Covaris S220 (Covaris, Woburn, MA). NEB Next Fast DNA Library Prep Set for Ion Torrent (New England Biosciences, Ipswich, MA) was used to generate barcoded libraries that were run on an Ion Torrent PGM (Life Technologies). Sequencing results were analyzed in CLC Bio’s Genomics Workbench (CLC Bio, Katrinebjerg, Denmark), and SNP calling required, at minimum, 20 reads at a given base with 20% reads per allele.
The length of CES1 5′UTR (s) was measured using 5′RACE product synthesized using the First Choice RLM-RACE RNA Ligase Mediated RACE kit (Life Technologies) from a CES1/CES1 individual. CES1 5′UTR length was visualized on an ABI3730 using product amplified with a CES1 specific fluorescent tagged primer (Table 1, Supplemental Digital Content). Length was determined using the internal size standard Gene Scan 500 ROX (Life Technologies) on the ABI3730.
Structure (2D) and free energy was assessed using mfold , with sequence based on the intermediate length 5′ UTR (derived from 5’RACE analysis), CES1 exon, and the first 18 bases of pGL3-Basic downstream of the insertion site. Folding structures and free energy were recorded and compared between the various constructs by One-way ANOVA (GraphPad Prism). To best visualize the overall difference in the predicted structures, three dimensional models were generated using RNAFold and RNAComposer (Figure 3) .
Luciferase assay vectors were prepared using pGL3-Basic (Promega, Madison WI) and a CES1 promoter region-5′UTR-exon 1 region fragment (1327 bases, Figure 4A). The Infusion Cloning System (Clontech, Mountain View, CA) was used to insert the fragment at the Hind III multiple cloning site of pGL3-Basic. Completed reactions were transformed into XL-2 (Agilent, Santa Clara, CA) cells and plated on LB agar with carbenicillin. Individual clones were screened and plasmids isolated by the Zyppy Mini Prep kit (Zymo Research, Irvine, CA). Plasmids were screened for inserts by Hind III digestion and sequenced for confirmation of the insert present in frame with the luciferase gene. Positive clones with haplotypes of interest were re-subcloned into DH5α cells to avoid colonized effects. Three positive subclones per haplotype were then combined and cultured, and plasmid DNA prepared with a Qiagen MidiPrep Kit (Qiagen, Valencia, CA). In total four constructs were generated, including, CES1, CES1VAR, CES1SVAR and CES1 with CES1 promoter SNP rs3815583. Luciferase assays were performed in HepG2 cells with high endogenous expression of CES1 (www.biogps.org) . Prior to transfection, cells were plated in 12-well plates and grown to ~60% confluence. Transfection plasmid mixes of CES1-pGL3B and pRL (Promega), a Renilla fluorescence vector (to serve as control to normalize transfection efficiency), were prepared in a ratio of 1 µg to 0.2 µg in OptiMem Media (Life Technologies). Plasmid mix was added to an equivalent OptiMem mix of transfection reagent FuGene HD (Promega) at 1µg DNA/3 µl reagent. Six hours post transfection, media containing antibiotics were added to transfected wells (Penicillin/Streptomycin, Life Technologies). Following incubation for 24 hours, luciferase activity was measured by Dual-Glo Luciferase Assay System (Promega) on a Packard Fusion plate reader (PerkinElmer Life and Analytical Sciences, Shelton, CT). Luciferase activity was determined as a ratio of luciferase fluorescence over renilla fluorescence. Luciferase was measured in three experiments, each with three replicates and two measurements per replicate.
CES1 mRNA expression was measured using quantitative real time PCR (qRT-PCR) on a 7500 Fast Real Time PCR System (Applied Biosystems, Carlsbad, CA, USA). Amplification of CES1 mRNA captured CES1, CES1VAR, CES1SVAR, CES1P1 and CES1P1VAR (the latter two accounting for ~2% of total CES1 mRNA). Relative quantitation used GAPDH as a housekeeping gene where expression is representative of overall RNA quality in human liver and has low interindividual variability .
The SNaPshot protocol has been previously described [15, 22]. In short, PCR amplification surrounding a marker SNP in an exonic region, from a heterozygous sample, is visualized using an extension primer and fluorescent ddNTPs. Allelic mRNA ratios are determined on an ABI3730 sequencer (Life Technologies). The average of the allelic ratios of gDNA is used to normalize allelic mRNA ratios in each tissue. The primary marker SNP for CES1 was rs12149370, derived from the translocation affected portion of CES1VAR and CES1SVAR. Because of CES1’s high identity with the CES1P1/CES1P1VAR gene locus, the genomic DNA control PCR product was selected to be specific to the CES1 locus regardless of translocation, providing accurate allelic gDNA ratios. Allelic mRNA ratios were measured at rs12149370 using PCR conditions detecting both CES1 (including CES1VAR) and CES1P1/CES1P1VAR, the latter expressed at low levels in the liver. qRT-PCR indicated that CES1P1/CES1P1VAR transcripts account for only ~2% of all CES1 mRNA, which was considered a negligible contribution to both allelic mRNA ratios and total mRNA. SNaPshot measurements were completed in duplicate for gDNA in all 28 livers heterozygous for rs12149370 (variant included in CES1VAR and CES1SVAR). Allelic mRNA ratios were measured in triplicate for 26 of 28 samples, while L36 and L120 were measured in duplicate. Primers for this assay are provided in Supplemental Digital Content, Table 1. Significant allelic mRNA ratios deviating from unity were considered greater than 1.2-fold and was termed allelic expression imbalance (AEI), an indicator of cis-acting regulatory variation.
Absolute CES1 protein expressions were determined in 102 individual human liver S9 fraction (HLS9) samples using a novel LC-MS/MS-based targeted absolute quantitative proteomics method (TAQSI) recently established in our laboratory . Briefly, an aliquot of 20 µg protein of HLS9 samples was mixed with the internal standard SILAC HepG2 cell S9 fractions (40 µg protein). After reduction and alkylation, proteins were digested with trypsin (Worthington Biochemical Corporation, Freehold, NJ) at an enzyme/protein ratio of 1:500 in an incubation shaker at 200 rpm at 37°C for 16 h. Digested peptides were then extracted (Oasis HLB columns, Waters, Milford, MA) and vacuum dried (Speed Vac SPD1010, Thermo Scientific, Hudson, NH). Finally, samples were reconstituted in 50 µl 50% acetonitrile and subjected to LC-MS/MS analysis. Standard calibration curves established from unlabeled CES1 protein calibrators (R&D system, Minneapolis, MN) and the HepG2 SILAC internal standards were employed to quantify absolute CES1 protein expressions in individual human livers. The CES1 calibrators ranged from 0.59 to 11.8 pmol, which the covered normal range of CES1 expressions in 20 µg proteins of HLS9 preparations. Three quality control (QC) samples (1.18, 2.36 and 4.73 pmol recombinant CES1) were utilized to evaluate the accuracy and precision of the assay.
To determine CES1 catalytic activity in the 102 individual human livers, the CES1 substrates clopidogrel (300 µM), enalapril (500 µM), or methylphenidate (50 µM) were incubated with individual HLS9 samples at 37°C for 20, 10, and 60 minutes, respectively. The protein concentrations of HLS9 for the three substrates were 1.5, 0.5, 0.5 mg/ml, respectively. Preliminary experiments were performed to ensure the formation of hydrolytic metabolites was linear with CES1 concentrations in the HLS9 samples. Following incubation, the concentrations of the metabolites clopidogrel carboxylate, enalaprilat, and ritalinic acid were measured using the LC-MS/MS assays previously established in our laboratory [10, 28, 29]. All incubations were carried out in triplicate.
Association of CES1VAR with gene expression and protein quantity was tested in SPSS (IBM, Armonk, NY). Stepwise linear regression including covariates, gender, race, rs3815583 carrier status, rs2244613 carrier status, and CES1VAR carrier status were included in these analyses. T-tests, correlation statistics for RNA-Protein relationship and one-way ANOVA for luciferase activity were performed in GraphPad Prism. P-values < 0.05 were considered significant for all associations. Allele frequencies for CES1VAR and other variants of interest were assessed using Helix Tree SNP and Variation Suite (Golden Helix, Bozeman, MT, USA) (Table 2B).
To understand the architecture of CES1 and its relationship to CES1P1, Sanger sequencing of the CES1 gene locus was performed. Previous studies have characterized the gene loci of CES1 and CES1P1 and identified CES1VAR (where a 5′ portion of CES1P1 translocates to CES1, MAF = 0.17, Table 2B) . Sequencing gDNA from CES1/CES1, CES1/CES1VAR and CES1VAR/CES1VAR homozygotes defined the major regions affected by CES1VAR. The CES1VAR 5′ boundary begins between 5′UTR SNP rs3815583 (outside of the region affected by translocation) and rs12149373 (first SNP of CES1VAR), and ends in intron 1. In total, this includes 11 variants in complete linkage disequilibrium (LD) attributable to CES1VAR in place of the original CES1 sequence (Figure 2). When screening for variants using CES1 specific primers, 3 samples were identified with 5 CES1P1VAR SNPs in the 5′UTR of CES1, while the exon 1 coding regions and intron 1 were unaltered. This shorter translocation variant is referred to here as CES1SVAR, representing a rare 5′ sequence variation seen at the CES1 gene locus (MAF < 0.01 Table 2B).
To select the most abundant 5′ mRNA isoforms for allelic mRNA analysis, CES1 5′UTR length was analyzed by 5′ RACE. In a CES1 homozygote, fluorescently labeled CES1-RACE PCR products indicate that CES1 mRNA in human liver is largely ~ 50 bases shorter than the mRNA annotated by RefSeq and the UCSC Genome Browser (Figure 1, Supplemental Digital Content), in relative concordance with previously reported transcription start sites . qRT-PCR confirms this medium-length 5′UTR length mRNA as the primary species, while the annotated long 5′UTR accounts for ~2% of total CES1 expression (including CES1, CES1, CES1P1 and CES1P1VAR data not shown). A shorter mRNA isoform beginning at the exon 1 coding region was also detectable with the RACE analysis, but only in small quantities (Figure 1, Supplemental Digital Content). Based on these observations, the medium 5′UTR is an abundant mRNA isoform beginning just downstream from CES1VAR and CES1SVAR associated sequence. Notably, this medium 5′UTR causes the CES1VAR variant rs12149373 to exist in the immediate promoter region, potentially affecting gene expression.
Considering the variants associated with CES1, CES1VAR and CES1SVAR, and the abundance of the medium–length 5′UTR, representative 5′UTR/exon1 sequences were analyzed for differences in RNA folding using mfold. Sequence variation can be viewed in Figure 2, starting at the beginning of the medium 5′UTR, through exon 1. The CES1 sequence showed 2 folding patterns, with an average final ΔG = −27.74. The CES1VAR sequence yielded 4 folding patterns with an average ΔG = −40.58. The CES1SVAR sequence yielded 3 structures with an average ΔG = −40.68. When compared to CES1, the CES1VAR and CES1SVAR sequences had significantly lower free energy (P <0.001, One-way ANOVA). Though there is some overlap in the predicted structures, the difference in ΔG indicates that CES1VAR and CES1SVAR are more stable in their predicted structures when compared to CES1. RNAFold outputs confirmed mfold free energy predictions. The RNAFold 2-D structure was visualized in 3-D using RNA Composer, with drastically different 3-D structures between the CES1, CES1VAR and CES1SVAR sequences (Figure 3). These indicate overall structural differences that may affect both transcription and translation.
To determine the function of the various CES1 5′ variant alleles, luciferase assays were performed in HepG2 cells. Constructs contained complete 5′UTR and exon 1 regions corresponding to CES1, CES1VAR, CES1SVAR and CES1+ rs3815583 (Figure 4A and Table 2, Supplemental Digital Content), with an additional 1172 bases of promoter region. CES1VAR showed a significant 35% decrease in luciferase activity when compared to CES1 (P <0.01, at 24 hrs) (Figure 4B). In contrast, CES1SVAR showed an average 36% increase in luciferase activity when compared to CES1 (P <0.01), though there was considerable variation in activity between replicates. CES1SVAR was not further studied because of its low MAF. Presence of rs3815583 (a SNP specific to CES1) showed no significant difference in luciferase activity. The difference between CES1 and CES1VAR luciferase activity was confirmed by measuring activity after 48 hours. Here, CES1VAR luciferase activity is significantly decreased by 22% (P = 0.012, Figure 2, Supplemental Digital Content).
Previous studies have shown that CES1 gene expression is highly variable, while the contributions of CES1 variants were unknown. Total CES1 mRNA was measured in 60 human liver samples (CES1 = 29, CES1/CES1VAR = 24, CES1VAR/CES1VAR =7). Analysis by stepwise linear regression including gender, race, and the previously reported rs3815583 and rs224816, indicated that CES1VAR carrier status was the only significant contributor to gene expression (p = 0.003; CES1SVAR was too infrequent and not included). Even with substantial variability in mRNA levels, mean total CES1 expression was 2.6-fold lower in CES1VAR carriers (heterozygotes and homozygotes) than in non-carriers (p =0.015, two-tailed t-test, WT; n = 29; VAR; n = 31) (Figure 5).
Allelic mRNA ratios were measured to determine whether CES1 is under the influence of cis-acting regulatory polymorphism(s). Because of the minor contribution of CES1P1/CES1P1VAR to total gene expression, allelic mRNA ratios were measured using common CES1 primers, amplifying almost exclusively CES1 mRNA. Allelic mRNA ratios were screened in 28 human livers heterozygous for the 5′ UTR rs12149370 located in the abundant medium-length mRNA transcript of both CES1VAR (minor allele frequency = 0.17) and CES1SVAR (MAF < 0.01). Different PCR primers were used for amplifying gDNA (to distinguish between CES1 and CES1P1 gene loci) and mRNA. To account for variability of allelic gDNA and mRNA ratios, and differences in amplification efficiency between reactions, the threshold for observing AEI was set at >1.20-fold. Of the 28 livers, 18 display AEI ratios >1.2-fold (1.23 to 1.4-fold, Figure 6), indicating that the CES1VAR allele was expressed at a lower level compared to the CES1 allele, and that cis-acting regulatory polymorphisms influence CES1 expression. CES1SVAR carriers, L50, L51, and LL10 show some degree of AEI (2 of 3 >1.2 fold), with similar interindividual variability observed in CES1VAR carriers. In 10 livers heterozygous for either translocation, the allelic mRNA ratios did not reach significance or were close to unity. This result could have been caused by variability in the allelic ratio analysis (4 samples close to the cutoff), variability in the magnitude of the allele effect caused by trans-acting factors, presence of additional regulatory variants compensating for the CES1VAR effect, or an alternative truly functional regulatory variant in partial LD with CES1VAR –though no variants in high LD and with the requisite frequency are present in the 1000 genomes database. In some cases, allelic mRNA ratios greater than 2-fold were observed and may be the result of an uncommon variant (representative sample L18, Fig. 6).
To test whether AEI was associated with a cis-acting functional polymorphism other than CES1VAR or CES1SVAR, 4.387 kb of the promoter region was sequenced in 10 human livers (5 AEI positive and 3 AEI negative CES1/CES1VAR individuals, and 1 AEI positive and 1 AEI negative CES1/CES1SVAR individuals). No single SNP was significantly associated with the allelic mRNA ratios. Promoter SNP rs38135583 was present in all AEI-negative samples (n = 4), and present in only 1 of 6 AEI-positive samples. This SNP is located in the long low-abundance 5′UTR isoform of CES1 and is not in LD with CES1VAR though it has been associated with appetite reduction during methylphenidate therapy . Analysis in additional liver cohorts indicate rs38135583 was not associated with AEI status (n=28) or mRNA expression (n=60).
The association of CES1VAR with decreased CES1 mRNA expression prompted us to determine allelic effects at the protein level. The novel targeted proteomics assay TAQSI was found to be highly accurate and precise for absolute quantification of CES1 expression in liver. The inter- and intra-day precision measured as relative standard deviations in the three quality controls were equal or less than 9.0% and 5.5%, respectively. The inter- and intra-day accuracy were within the ranges of 96.1%–102.2% and 94.6%–106.1%, respectively. The distributions of CES1 genotypes among those samples are: CES1SVAR heterozygote: 1; CES1VAR homozygotes: 6; CES1VAR heterozygotes: 32, CES1VAR non-carriers: 63. The CES1 protein expression levels in CES1VAR carriers did not significantly differ from those of non-carriers (161 ± 63 vs 180 ± 76 pmole/mg protein, p = 0.24) (Figure 7A). In addition, we measured CES1 mRNA expression in 50 livers randomly selected from the whole set of 102 samples, and found that CES1 protein expressions did not significantly correlate to CES1 mRNA levels (Figure 7B, p = 0.67, r = 0.06, n = 50).
The effect of CES1VAR on CES1 enzymatic activities was further interrogated using a total of 102 HLS9 samples. The hydrolyzing velocity of the CES1 substrates, enalapril, clopidogrel and methylphenidate, was measured and tested for association with CES1VAR carrier status by linear regression including gender, race, as well as the previously reported variants, including rs3815583 and rs2244613. No predictor was significantly associated with enzyme activity. Hydrolysis of enalapril in CES1VAR carriers was not significantly decreased when compared to the non-carriers (p = 0.93, two-tailed t-test, WT: 53 ± 38 (pmole/min/mg protein, n = 63) vs. VAR: 54 ± 37 (pmole/min/mg protein, n = 38)) (Figure 8A). Similar results were obtained when comparing CES1 activities between the CES1VAR carriers and non-carriers on the metabolisms of clopidogrel (p = 0.82, two-tailed t-test, WT: 130 ± 58 (pmole/min/mg protein, n =63) vs. VAR: 137 ± 64 (pmole/min/mg protein, n = 38)) (Figure 8B) and methylphenidate (p = 0.95, two-tailed t-test, WT: 205 ± 98 (pmole/hr/mg protein, n =63) vs. VAR: 208 ± 95 (pmole/hr/mg protein, n = 38)) (Figure 8C). Notably, G143E, a well defined loss of function allele was significantly associated with CES1 activity on hydrolysis for all three CES1 substrates (data not pictured).
This study confirms previous reports of the CES1P1-CES1 translocation generating the CES1VAR allele (17% MAF) with 11 SNPs in the 5′UTR, exon1 and intron 1 derived from the CES1P1 sequence. In addition we identified a shorter translocation involving only the 5′UTR, termed CES1SVAR, with MAF < 0.01 (Figure 2). As CES1P1 is poorly expressed and accounts for only ~2% of the CES1 mRNA in the liver, we tested whether the translocation affects expression of CES1 mRNA. Though CES1 mRNA was variable in human liver, results of this study indicate that CES1VAR decreases mRNA expression in human liver to a moderate extent which is supported by results from a luciferase reporter gene assay. Similarly, CES1SVAR displayed only a modest effect in a luciferase reporter gene assay, but was not further studied because of its low frequency. However, the studies of CES1 protein expression and metabolic activity on hydrolyzing enalapril, clopidogrel, and methylphenidate in human livers failed to reveal significant differences between the CES1VAR genotypes. Overall, our results suggest a modest reduction in CES1 mRNA expression caused by the translocation in the CES1VAR allele. Such small effects could be buffered by large interindividual variability in CES1 expression and activity, as well as possible post-transcriptional regulatory processes. As a result, any effect of CES1VAR on CES1 protein expression and activity might not be readily detected. The results are consistent with a recent study by Wang and colleagues, suggesting that CES1VAR does not significantly affect hepatic CES1 protein expression and enzymatic activity on the activation of ACE inhibitor prodrugs .
Reduced expression due to CES1VAR is supported by total and allelic mRNA expression data, and luciferase reporter gene analysis. Our allelic mRNA expression results are consistent with this finding but leave open the possibility of additional regulatory variants. A previous study had suggested that less common exon 1 CES1 regulatory variants exist in addition to CES1VAR . Although we identified CES1SVAR by sequencing, allelic mRNA ratios failed to reveal any new regulatory variants in the screened 5′ region of CES1. Our study also provides clarification on CES1 5′UTR length. Measurement of the average 5′UTR length by 5′RACE indicates that the most frequent species is shorter than the RefSeq annotation. This medium-length 5′UTR is in concordance with transcription start sites reported previously for CES1 .
Luciferase assays support 5′UTR variation as a functional variant, indicating that CES1VAR decreases protein expression one-third in HepG2 cells, while CES1SVAR affected luciferase activity with a one-third increase. Construct CES1 + rs3815583 (5′ UTR SNP) includes a variant only present in the long 5′ UTR transcript. While this SNP has been associated with methylphenidate side effects the construct showed no effect on luciferase activity.
The sequences of CES1VAR and CES1SVAR introduce multiple factors that may affect expression. First, the medium 5′UTR begins just downstream from the first CES1VAR and CES1SVAR associated variant, potentially affecting transcription. In the exonic coding region, CES1VAR contains multiple non-synonymous coding variants and includes several intron 1 SNPs, all with unknown effects. Second, in silico analysis of CES1VAR and CES1SVAR reveals changes in computed folding structure of the mRNA, with CES1 displaying higher free energy when compared to both variant alleles. Here, significant decreases in free energy may cause the variant allele’s mRNA to be less receptive to translation. These factors are all potential mechanisms by which CES1VAR and CES1SVAR could impact transcription, mRNA processing, and translation at the CES1 gene locus – some effects potentially cancelling each other out. Interestingly, the poorly expressed CES1P1 specific sequence which creates CES1VAR might also be expected to drastically decrease CES1VAR expression in a similar manner. However, we see only a small CES1VAR specific effect at the genetic level, with no effect at the protein level. Despite the possible regulatory elements contained in the translocated region, this indicates that more pervasive regulatory elements are present outside the translocation.
Here we have shown that the previously described CES1VAR decreases CES1 mRNA expression in the human liver by ~30%, possibly reflecting complex regulation of the 5′ region. Multiple UTR lengths, alternative sequences, and potential changes in mRNA folding suggest this region is driving variation in CES1 expression. However, the associations of CES1VAR genotypes with CES1 protein expression and activity in human livers were insignificant in the present study; any small effect could have been confounded by the marked interindividual variability, while we also cannot exclude possible involvement of post-translation regulation in CES1 protein expression. Future studies should assess the effect of CES1 regulatory variants on the metabolism of CES1 substrate drugs in clinical settings to evaluate the utility of those variants as clinical biomarkers to improve therapeutic outcomes of many drugs metabolized by CES1.
Funding: This study was supported by a grant from the NIH General Medical Sciences U01 GM092655, and partially by a grant from the National Center for Research Resources (UL1RR025755). The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Center for Research Resources or the National Institutes of Health.
Conflicts of Interest: None declared