Abstract
The enzyme UGT2B7 is one of the most active UDP-glucuronosyltransferases (UGTs) involved in drug metabolism and in maintaining homeostasis of endogenous compounds. We recently reported the existence of 22 UGT2B7 mRNAs, two with a classic 5′ region but alternative 3′ ends namely UGT2B7_v5 (containing a novel terminal exon 6b) and _v7 (exon 5 excluded) that encode enzymatically inactive isoforms 2 and 4 (i2 and i4), respectively. The v1 mRNA encoding the UGT2B7 enzyme (renamed isoform 1 or i1) is coexpressed with the splice variants v5 and v7 in human liver, kidney, and small intestine and the hepatic cell lines HepG2 and C3A. The presence of alternate v5 and v7 transcripts in isolated polysomes from these hepatic cells further supports endogenous protein translation. Cellular fractionation of clonal HEK293 cell lines overexpressing UGT2B7 isoforms demonstrates that i1, i2, and i4 proteins colocalize in the microsomal/Golgi fraction, whereas i2 and i4 can also be found in the cytosol; a finding sustained by immunofluorescence experiments using tagged proteins. By modifying splice variant abundance in overexpression in HEK293 and HepG2 cells as well as RNA interference experiments in HepG2 and C3A cells, we observe drug glucuronidation phenotypes compatible with variant-mediated repression of UGT2B7 activity without consequent alteration of the apparent enzyme affinity (Km). Finally, coimmunoprecipitation experiments support a direct protein–protein interaction of i2 and i4 proteins with the functional UGT2B7 enzyme as a potential causative mechanism. These findings point toward a novel autoregulatory mechanism of the UGT2B7 glucuronidation pathway by naturally occurring alternative i2 and i4 proteins.
Introduction
Alternative splicing has been documented for more than 95% of multi-exon human genes and is responsible for transcriptome expansion and protein diversity (Pan et al., 2008). This post-transcriptional mechanism has been reported for genes involved in drug metabolism, most notably cytochrome P450 (P450) (Turman et al., 2006), sulfotransferases (He et al., 2005), glutathione S-transferase (Ross and Board, 1993), and UDP-glucuronosyltransferases (UGTs) (Lévesque et al., 2007). UGT1A family members are derived from the use of alternative promoters and exons 1, whereas UGT2B7 is subjected to extensive alternative splicing (Innocenti et al., 2008; Guillemette et al., 2010; Ménard et al., 2011).
UGT2B7 is an enzyme responsible for detoxification of xenobiotics, including drugs such as morphine, efavirenz, zidovudine, and fenofibric acid, and is involved in the homeostasis of endogenous molecules like eicosanoids, bile acids, and sex steroids (Jin et al., 1997; Barbier et al., 2000; Stone et al., 2003; Turgeon et al., 2003; Thibaudeau et al., 2006; Mano et al., 2007; Bélanger et al., 2009; Tojcic et al., 2009; Trottier et al., 2013). Its expression is high in liver and kidney, but is also detectable in the gastrointestinal tract and peripheral tissues (Nakamura et al., 2008; Court et al., 2012). The UGT2B7 gene is located on chromosome 4q13, has six classic exons, and spans 17 kb (Mackenzie et al., 2005). As for discoveries related to alternative splicing of this gene, two additional distal 5′ exons were initially observed (referred to as exons 1a and 1b, respectively, and located 44 kb and 7 kb from the classic translation start site in exon 1) (Innocenti et al., 2008). Our group then uncovered two additional 5′ exons (1c and 1d, situated between exon 1a and 1b) as well as two novel 3′ exons (6b and 6c, located in the intron between exon 5 and the classic exon 6, renamed 6a) (Ménard et al., 2011). Alternative splicing at the UGT2B7 locus produces at least 22 full-length transcripts mainly observed in kidney and liver. Consequently, the UGT2B7 enzyme has been renamed isoform 1 (i1) and is encoded by the variant transcript 1 (v1). The variant UGT2B7 mRNAs potentially encode six shorter proteins with variable N-terminal and/or C-terminal ends that were named i2 to i7. Whether these alternative proteins are biologically active remains to be demonstrated.
Recent observations related to alternative splicing of the UGT1A family members revealed nine variant proteins derived from the use of an alternative 3′ exon (5b) with a shorter C-terminal end (named isoforms 2 or i2s) that regulate UGT1A-mediated glucuronidation by UGT1A enzymes (or i1s) (Girard et al., 2007; Lévesque et al., 2007). Indeed, UGT1A_i2s are enzymatically inactive but repress glucuronidation through protein–protein interactions with active UGT1A_i1s (Bellemare et al., 2010b). Data suggest that interactions between i1 and i2 proteins lead to an inactive complex and that i2 proteins decrease rates of glucuronide formation without affecting the kinetic model and substrate binding (Bellemare et al., 2010a). The oligomeric nature of UGT proteins has been known for some time: for UGT2B7, homodimerization is suspected to lead to atypical enzyme kinetics for certain substrates that is best described by a two-sites model (Lewis et al., 2011), whereas heterodimerization can occur with several other UGT proteins and P450 enzymes resulting in altered activity (Fremont et al., 2005; Fujiwara et al., 2010).
Two UGT2B7 splice variants have a classic 5′ region but have an alternative terminal exon 6b (variant 5 or v5) or a deletion of the exon 5 (variant 7 or v7), sharing similarities to UGT1A activity-modulating splice variants (Ménard et al., 2011). Whereas the UGT2B7 enzyme (i1) is derived from transcript v1, transcripts v5 and v7 encode C-terminally truncated i2 and i4 proteins, respectively. Here we sought to characterize their modulatory potential on cellular glucuronidation. We initially tested various commercial sources of UGT2B7 antibodies with the aim of assessing expression of UGT2B7 isoforms in cellular models and human tissues. We then evaluated the effect on glucuronidation kinetics of stably coexpressing i2 and i4 proteins along with the UGT2B7 enzyme as well as their subcellular localization by immunofluorescence and Western blotting of centrifugation fractions. We then tested the influence of partial depletion of endogenous UGT2B7 variants and their overexpression on glucuronidation activity of two human liver cell lines, HepG2 and C3A. Finally, coimmunoprecipitation experiments allowed us to test the possibility of a direct protein–protein interaction of i2 and i4 proteins with the functional UGT2B7 enzyme. Our findings support that truncated UGT2B7 i2 and i4 variant isoforms have inhibitory properties in vitro and may contribute in the control of UGT2B7-mediated glucuronidation.
Materials and Methods
Cellular Models.
HepG2, HEK293, and C3A cells (American Type Culture Collection, Rockville, MD) were cultured as recommended with 10% fetal bovine serum. HepG2 is derived from a hepatocellular carcinoma and C3A is a clonal derivative of HepG2 that exhibits a better differentiated hepatic phenotype (Filippi et al., 2004). The cloning of UGT2B7_v1, v5, and v7 cDNAs encoding UGT2B7 i1, i2, and i4 proteins and methodologies for transfection of HEK293 cells to produce stable clones have been described (Ménard et al., 2011). Briefly, prior to antibiotic selection, cells were grown for 48 hours and then transfected with 5 µg DNA by electroporation using the Neon Transfection System (Life Technologies, Carlsbad, CA). For expression in HepG2 cells, plasmid constructs containing v5 and v7 with or without a V5 epitope were transfected as well as an empty pcDNA6 acting as a mock control. For RNA interference experiments, two short interfering RNAs (siRNAs) were purchased from Thermo Scientific (Waltham, MA): one directed against the exonic end and the proximal 3′-untranslated region of the exon 6b (5′-ctgctaaactgtcttccaa-3′) and the other against the ex4–ex6a splicing event (5′-ccttctagatataaagaga-3′). Specificity of siRNAs was tested using the BLAST tool (http://blast.ncbi.nlm.nih.gov/). A negative control was also purchased from Thermo Scientific (ON-TARGETplus Non-targeting siRNA #3). After transfection with siRNA, cells were grown for 48 hours without antibiotics and then harvested. Cell lysates were prepared for glucuronidation assays and Western blotting. Total RNA was extracted with Tri Reagent (Sigma-Aldrich, St. Louis, MO) for use in quantitative real-time polymerase chain reaction (PCR).
Western Blotting and Glucuronidation Assays.
Microsomes and cell lysates were prepared as described (Lépine et al., 2004). The protein concentrations were determined using the BCA protein assay (Fisher Scientific, Ottawa, ON, Canada). Polyclonal antibodies against UGT2B7 from Millipore (Billerica, MA), BD Gentest (Franklin Lakes, NJ), and Abcam (Cambridge, UK) were tested for specificity using 20 µg of microsomes of HEK293 clones overexpressing each of the UGT2Bs, following the manufacturers’ protocols. A nonselective anti-UGT2B, called EL-93, was produced in-house (Guillemette et al., 1997) and used as a positive control of UGT overexpression. Glucuronidation assays on 3′-azido-3′-deoxythymidine (AZT; Sigma-Aldrich) were performed using microsomes (20 µg) or homogenates (50 µg), and AZT glucuronides (AZT-G) were detected by a mass spectrometry–based assay as described (Thibaudeau et al., 2006; Bélanger et al., 2009). Kinetic parameters of glucuronidation were assessed in assays containing a 100- to 5000-µM range of AZT. Kinetic parameters were estimated using Prism 6 software (GraphPad Software Inc, La Jolla, CA). Densitometric measurements of the UGT2B7 enzyme (i1) protein levels by Western blotting served to normalize glucuronidation activity. Briefly, microsomes isolated from cells used in the enzyme kinetics, siRNA, and overexpression experiments were blotted and revealed using the anti-2B7 from Millipore. Western blotting with a monoclonal antibody directed against the V5 epitope further revealed the presence of i1 proteins in HEK293 membranes upon overexpression. The relative levels of i1 proteins were assessed by integrated optical density using the Bioimage program Visage 110S (Genomic Solution Inc., Ann Arbor, MI), and activity levels were normalized by these values. Statistical significance was evaluated by a paired t test (P < 0.05) using XLSTAT for Mac (Addinsoft, New York NY).
Gene Expression Profiling by Quantitative Reverse Transcription PCR in UGT2B7-Depleted Cells.
Total RNA from adult human tissues was purchased from Life Technologies. cDNA was synthesized with 1 µg RNA and 250 ng random hexamers with Superscript III reverse transcription (RT) (Life Technologies) following the manufacturer’s protocol. RT-PCR amplification of full-length transcripts was done as described (Ménard et al., 2011) using primers described in Supplemental Table 1. Real-time quantitative PCR (qPCR) assays were performed in triplicate in a StepOne real-time PCR system with the Sybr Green PCR master mix (Life Technologies), 12 ng of cDNA, and 500 nM of primers designed to specifically amplify the ex5–ex6b and ex4–ex6a splicing events specific to variant v5 and v7, respectively. Amplification of the ex5–ex6a splicing event, a splicing event characteristic of the active enzyme, was used as a control of the specificity of the siRNAs used herein. Results were normalized according to RPLP0 (36B4) gene expression. All primers used for qRT-PCR amplifications are detailed in Supplemental Table 1. During RT-qPCR optimization, five different concentrations of vectors containing v1, v5, v7, and 36B4 were used to delineate standard amplification curves, and expression data are reported as the number of Alternative splicing events per 10,000 36B4 copies.
Immunofluorescence.
HEK293 clones stably expressing UGT2B7 fusion proteins i1-V5, i2-V5, and i4-V5 were used to visualize the V5 epitope (mouse monoclonal anti-V5, 1:100; Life Technologies), the endoplasmic reticulum (ER)–specific protein calnexin (rabbit polyclonal anti-calnexin, 1:200; Enzo Life Sciences, Farmingdale, NY) and the Golgi-specific protein giantin (rabbit polyclonal anti-giantin, 1:2000; Abcam) in immunofluorescence experiments. Two secondary fluorochrome-coupled antibodies (goat anti-rabbit conjugated to Alexa Fluor 488 [green] and goat anti-mouse conjugated to Alexa Fluor 594 [red]; Life Technologies) were diluted 1:500. TO-PRO 3 (1:5000; Life Technologies) was also used to mark the nucleus. Data were acquired with a Fluoview 300 confocal microscope (Olympus, Center Valley, PA).
Coimmunoprecipitation Assays.
HEK293 cell lines stably expressing V5-tagged recombinant i1 and “empty” HEK293 cells were plated at a density of 5 × 106 cells per dish (100 mm) and transiently cotransfected 24 hours later by electroporation with 20 μg of a pcDNA3 vector containing the untagged UGT2B7_v1, v2, or v4 cDNA sequences (encoding i1, i2, and i4 proteins, respectively). After 48 hours, cells were lysed, and immunoprecipitation was performed with monoclonal anti-V5 (2 µg/ml lysate) as described (Bellemare et al., 2010b), followed by SDS-PAGE of the samples and Western blotting using anti-2B7 to detect the UGT2B7 variants (Millipore).
Analysis of Polysomes from HepG2 Cells.
Cycloheximide at 100 µg/ml was added to the HepG2 cell medium and incubated 5 minutes at 37°C. HepG2 cells were then washed with ice-cold phosphate-buffered saline containing cycloheximide (100 µg/ml) and lysed in 200 µl lysis buffer: 50 mM Tris-HCl pH 8.3, 150 mM KCl, 3 mM MgCl2, 1% (w/v) IGEPAL, 10 mM dithiothreitol, 100 µg/ml cycloheximide, 200 µg/ml heparin, 100 U/ml Protector RNase Inhibitor (Roche Applied Science, Penzberg, Germany), and Complete EDTA-Free Protease Inhibitor (Roche Applied Science). The addition of 10 mM EDTA into the lysis buffer of negative controls disrupted polysome aggregation and released translated mRNAs. Lysed cells were harvested with the use of a cell scraper. The complete protocol for polysome isolation was described elsewhere (Scantland et al., 2011).
Results
Heterologous Expression of UGT2B7 Truncated Proteins.
We previously discovered two UGT2B7 transcripts with a classic 5′ region but alternative 3′ ends, namely UGT2B7_v5 (containing a novel terminal exon 6b) and _v7 (exon 5 excluded) (Fig. 1A). The v1, v5, and v7 transcripts are coexpressed in human liver, kidney, and small intestine as well as in two human hepatic cell lines, HepG2 and C3A. HEK293 cells lack UGT expression and were initially used to overexpressed splice variant products (Fig. 1B). Because we also wanted to study endogenous proteins, we selected two human cell lines expressing UGT2B7 splice variants v1, v5, and v7. These hepatic cell lines also exhibit glucuronidation activity for a selective substrate of UGT2B7 (AZT) (Fig. 1B).
UGT2B7 splice variants. (A) mRNA structure of two alternative transcripts (named v5 and v7) and the canonical v1 derived from human UGT2B7 by alternative splicing. (B) RT-PCR amplification of v1, v5, and v7 in total RNA from human liver, kidney, and small intestine from three individuals as well as selected cell lines (HepG2, C3A, and HEK293). (C) Peptide sequence representation of variant isoforms (i1, i2, and i4) derived from v1, v5, and v7 transcripts, respectively. Protein domains are also indicated. (D) Overexpression of UGT2B7 variants in HEK293 cells (UGT2B7-negative). Clones overexpressing each individual protein (i1, i2, or i4) or double-transfectants (i1+i2 and i1+i4) were produced, and protein expression was confirmed using an anti-V5 antibody. The UGT2B7-overexpressing cell lines displayed 1:0.35 (i1:i2) and 1:0.1 (i1:i4) ratios. aa, amino acid.
With the aim to assess protein expression in cellular models and human tissues, we initially tested three commercial UGT2B7 antibodies (from Millipore, BD Gentest, and Abcam) on microsomes of HEK293 overexpressing UGT2Bs. Although the Millipore antibody was the most efficient at detecting the UGT2B7 enzyme (i1) and its cross-reaction with other UGT2Bs was minimal (Fig. 2), no immunoreactive signal was detected in human tissues at the molecular weights predicted for i2 and i4 isoforms, as assessed by Western blotting (data not shown). The lack of a suitable antibody thus hindered detection of i2 and i4 in tissues or cell lines in which we observed v5 and v7 mRNAs. However, detection of full-length mRNAs in hepatocellular models such as HepG2 and its clonal derivative C3A (Fig. 1B) and the presence of alternate transcripts in isolated polysomes from these cells supported their endogenous translation to i2 and i4 proteins (Fig. 3). In addition, variant UGT2B7 i2 and i4 proteins tagged with a V5 epitope were produced in HEK293 cells, an overexpression that was confirmed by Western blotting using anti-V5 antibody (Fig. 1D).
Specificity of commercial antibodies against UGT2B7 and other UGT2B family members. To test the specificity of commercial antibodies against UGT2B7, we used 20 μg microsomes derived from clonal HEK293 cells stably expressing UGT2B isoforms in 10% SDS-PAGE experiments. Antibodies were used as suggested by the manufacturers (Millipore, Gentest, and Abcam), and results were compared with experiments using an anti-UGT2B antibody (EL-93 targeting the C-terminal extremity of UGT2B proteins). An arrow identifies UGT2B7 signals. Ab, antibody; NS, nonspecific marking of the antibody.
Isolation of polysomal fractions from liver cells supporting translation of UGT2B7 mRNA variants v1, v5, and v7. The upper panel shows fractionation of cytosolic extracts from HepG2 cells through a linear sucrose gradient (15%–45%). The chromatogram illustrates the separation of HepG2 cytosolic extracts at assessed by measuring absorbance (254 nm) in the absence (filled line) or presence (dashed line) of EDTA. A similar pattern was observed with C3A cells (not shown). The polysomes were found in fractions 6–11 (–EDTA), whereas monosomes were mainly found in earlier fractions. The presence of EDTA breaks polysomes into ribosomal subunits leading to the release of translated mRNAs (+EDTA). The lower panel shows detection of UGT2B7 mRNA variants in polysomal fractions by specific PCR amplification. For each fraction, we assessed the presence of three UGT2B7 mRNA variants, namely UGT2B7_v1 (positive control), v5, and v7. As expected for translated mRNA, specific PCR products were mostly observed in polysomal fractions in the absence of EDTA. In contrast, the presence of EDTA shifted specific PCR products toward monosomes (80S) and ribosomal subunits (40S and 60S).
Comparative Studies of Subcellular Localization of Variant UGT2B7 Proteins.
Sequence analysis indicated that the predicted i2 and i4 proteins have a shorter C-terminal region, a feature that could affect subcellular localization (Fig. 1C). Cellular fractionation of clonal HEK293 cell lines expressing either i1 alone or with one variant (i2 or i4) demonstrated that all three UGT2B7 proteins colocalized in the microsomal/Golgi fraction (Fig. 4A). Moreover, a fraction of i2 and i4 could be recovered in the cytosolic fraction colocalizing with transketolase. The proper localization of subcellular compartment markers confirmed that the different organelles were correctly isolated in the three fractions (Fig. 4B). The localization of UGT2B7 variants in ER and Golgi was also confirmed in living cells by immunofluorescence experiments using calnexin and giantin as controls (Fig. 5).
Cellular localization of UGT2B7 protein variants. Cellular fractions [microsomes/Golgi (M+G), whole cells/nuclei (W+N), and cytosol (C)] from HEK293 cells overexpressing i1, i1+i2, or i1+i4 were separated by 10% SDS-PAGE. (A) Proteins were detected with rabbit polyclonal anti-UGT2B7 (1:1000; Millipore) (top panel) and monoclonal anti-V5 tag (1:5000) (bottom panel). (B) As positive controls to mark specific subcellular structures, we used anti-calnexin for the ER (top panel), anti-RCAS1 for the Golgi (middle panel), and anti-transketolase for the cytoplasm (bottom panel). Arrows indicate detected proteins.
Cellular localization by immunofluorescence. The subcellular localization of UGT2B7 proteins (V5-tagged proteins) was superimposed with the staining of calnexin, an ER resident protein (A), and giantin, a maker for the Golgi (B). HEK293 cells were used as a negative control.
UGT2B7 Truncated Isoforms Suppress AZT Glucuronidation Activity Likely through Direct Protein–Protein Interaction.
We previously showed that the UGT2B7_i2 and i4 proteins are enzymatically inactive on classic UGT2B7 substrates (Ménard et al., 2011). However, the presence of i2 or i4 in UGT2B7-overexpressing HEK293 cell lines significantly decreased the extent of i1-mediated AZT-G formation by up to 40% compared with cells expressing only i1 (Fig. 6A). There was no significant difference in Km values among the clonal cell lines, and all data were consistent with Michaelis-Menten kinetics (Supplemental Fig. 1). At the Km (1550 μM), AZT-G formation was reduced by 41% in the presence of i2 and by 48% with i4. At a fixed substrate concentration (200 μM AZT) and variable concentrations of cosubstrate UDP-glucuronic acid (50–5000 μM), i2 or i4 caused an 80% decrease in Vmax with no change in Km (data not shown). Because protein–protein interactions are a potential mechanism underlying the inhibitory effect of i2 and i4 on UGT2B7 activity, we examined whether i2 and i4 interact with i1. Coimmunoprecipitation experiments were performed using HEK293 cell lines expressing i1-tagV5 + i2 (Fig. 6B) or i1-tagV5 + i4 (Fig. 6C). We observed that the untagged i2 and i4 coimmunoprecipitated with i1-tagV5. Because of an apparent weaker i1-i4 interaction (likely caused by inferior transient expression of v7 in HEK293 cells), this observation was confirmed using hemagglutinin-tagged i1 and V5-tagged i4 overexpressed proteins and a hemagglutinin column in a coimmunoprecipitation experiment (data not shown).
Kinetics of AZT glucuronidation assessed in HEK293 cells overexpressing UGT2B7 splice variants and coimmunoprecipitation assays supporting a protein–protein interaction between i1 and truncated variants i2 and i4. (A) Rates of formation of AZT-G for HEK293 cell lines. Variant proteins were tagged with a V5 epitope; similar results were observed with untagged UGT2B7 variants (data not shown). Kinetic parameters (Vmax and Km) are shown below the graph and were fitted to a Michaelis-Menten profile consistent with previous reports (Barbier et al., 2000) (Supplemental Fig. 1 displays Eadie-Hofstee plots). All data represent the mean ± S.D. of two independent experiments performed in triplicate and normalized for UGT2B7 enzyme expression (i1). *P < 0.05. (B and C) Coimmunoprecipitation experiments support the interaction between the active i1 UGT2B7 enzyme and splice variants i2 (B) and i4 (C). The i1 protein was V5-tagged, whereas UGT2B7 variants were not tagged for transient expression. In all cases, monoclonal anti-V5 was used for immunoprecipitation (IP), whereas polyclonal anti-UGT2B7 (Millipore) was used to visualize the resulting bands in 10% SDS-PAGE migration experiments. Lanes 1, 2, and 3: IP with anti-V5 (1 μg); lane 1: i1-V5 + untagged isoform (i1, i2, or i4); lane 2, untagged isoform (i1, i2, or i4), negative control; lane 3, i1-V5, positive control for i1-V5 expression; lane 4, IP with rabbit Ig-G (1 μg), i1-V5 + untagged isoform (i1, i2, or i4), control for the specificity of IP.
Overexpression or Repression of UGT2B7 Splice Variants Modifies UGT2B7-Mediated Activity in Hepatic Cells.
The effect of ectopic expression of i2 or i4 (with or without a C-terminal V5 tag) was further assessed using HepG2 cells as a model—this cell line endogenously expresses enzymatically active UGT2B7_i1 and its splice variants (Fig. 1B). After overexpression of i2 or i4, AZT-G formation significantly decreased by at least 20% regardless of the presence of the tag (Fig. 7). Notably, overexpression of splice variants was confirmed by the detection of ex5–ex6b and ex4–ex6a alternative splicing events by qPCR, representing less that 10% of the expression of v1 transcripts. By using specific siRNA-mediated knockdown, we then evaluated whether repression of splice variants could influence UGT2B7 activity. We initially tested the hepatoma-derived HepG2 cells and further confirmed the effect in the clonal cell line C3A, both of which endogenously express v1, v5, and v7 mRNAs (Fig. 1B). Figure 8 illustrates that partial repression by 70% and 62% of v5 mRNA levels in HepG2 and C3A cells increased AZT-G formation (ranging from 20% to 72%). In addition, repression of v7 mRNA (by 54%) increased UGT2B7 activity by 110% in C3A cells. All observed changes were statistically significant except for the v7 knockdown in HepG2 cells, in which the ex4–ex6a siRNA had a relatively weak effect on v7 mRNA expression and AZT glucuronidation activity. In all conditions, the levels of v1 expression were unaffected. Additional glucuronidation assays also showed no significant changes in the formation of estradiol-3G, a reaction catalyzed by few UGT1As, including UGT1A1 (results not shown).
Overexpression of i2 and i4 proteins in HepG2 cells affects the rate of AZT-G formation. Microsomal extracts (20 µg) from i2- or i4-overexpressing HepG2 cell lines were used to assess UGT2B7-mediated AZT glucuronidation (top panel) and compared with the glucuronidation activity of a pcDNA6-overexpressing HepG2 cells. Data are given as the mean ± S.D. of at least two independent experiments performed in triplicate and normalized for UGT2B7 enzyme expression (i1). *P < 0.05. Quantification of a characteristic splicing event for each transgene by qPCR is expressed as the number of copies per 10,000 copies of RPLP0 (36B4, control gene) (middle panel). The bottom panel depicts a graphical representation of the real-time PCR amplification strategies used to delineate the occurrence of ex5–ex6a, ex5–ex6b, and ex4–ex6a; those splicing events were mainly noted in v1, v5, and v7 transcripts, respectively.
Knockdown of endogenous UGT2B7 alternative transcripts by RNA interference induces glucuronidation activity of two human liver cell lines. (A) Schematic representation of the UGT2B7 region targeted by each siRNA. (B) Changes in expression were evaluated by qPCR and reported as relative expression to a nontarget siRNA control. (C) Glucuronidation assays were performed with cell homogenates (50 µg) incubated 4 hours with 200 μM AZT. Data are displayed as the mean ± S.D. of two independent experiments performed in triplicate and normalized for UGT2B7 enzyme expression (i1). *P < 0.05. UTR, untranslated region.
Discussion
We report that co-overexpression of the enzymatically active UGT2B7_i1 with i2 or i4 splice forms in HEK293 (UGT negative) cells and in HepG2 cells (which endogenously express i1, i2, and i4) led to a significant reduction of UGT2B7-mediated glucuronidation. On the other hand, siRNA-mediated partial depletion of endogenous variant UGT2B7 transcripts v5 and v7 led to increased cellular glucuronidation. We were able to detect the expression of the novel v5 and v7 transcripts in human liver, kidney, and small intestine as well as in liver cell lines, but the absence of a specific antibody for i2 and i4 prohibited detection of these proteins. However, the full-length v5 and v7 transcripts could be detected in translated mRNAs found in polyribosomal fractions of HepG2 and C3A cells, confirming the production of i2 and i4 proteins in those cell lines. In addition, overexpression of v5 and v7 in HEK293 cells gave rise to detectable i2 and i4 proteins in Golgi and ER, which further hints toward the functional translatability of these variants in cells. The enzyme kinetics for AZT, a probe substrate for UGT2B7, showed that the apparent affinity (Km) was not affected by i2 and i4,whereas the Vmax was significantly reduced compared with cells expressing only the i1 enzyme. The lowering of the Vmax of UGT2B7_i1 by i2 and i4 suggests that the inhibitory mechanism of the alternative isoforms could be through a decrease in the concentration of active UGT2B7 enzymes. In support of this idea, we further demonstrated the potential of each of the two UGT2B7 C-terminally truncated proteins to interact with the i1 enzyme. This mechanism likely results in the trapping of the active UGT2B7 enzyme in an inactive higher-order complex, thus hindering glucuronidation. These observations are in line with our previous reports of the regulatory influence of enzymatically inactive truncated isoforms 2 derived from the inclusion of an alternative exon 5 (5b) at the UGT1A locus on cellular glucuronidation (Bellemare et al., 2010b,c).
Regarding protein–protein interactions, the homodimerization of UGT2B7 has been reported (Lewis et al., 2011), a feature that is also observed herein (data not shown). Our data are in line with previous reports of the oligomerization of UGT2B7 with other UGTs and P450s, a process that appears to modulate glucuronidation rates (Fremont et al., 2005; Takeda et al., 2005, 2009; Fujiwara et al., 2010). Furthermore, UGT2B7 dimerization was identified as a possible causal factor for the atypical sigmoidal kinetics of glucuronidation by UGT2B7 of 4-methylumbelliferone and 1-naphtol, and these kinetics fit well with a two-site model in which two distinct substrate-binding sites coexist (Uchaipichat et al., 2008). The replacement of a putative “dimerization signature motif” (183HSGGFIFPPSYVPVVMS200E) in UGT2B7 by a UGT2B15 homologous peptide sequence led to the loss of atypical kinetics of glucuronidation of 4-methylumbelliferone (Lewis et al., 2007), a feature attributable to unsuccessful homodimerization of the enzyme (Lewis et al., 2011). Interestingly, this potential dimerization motif is present in UGT2B7 i2 and i4 and thus could allow interaction with the UGT2B7 enzyme, as we observed. Full-length alternative transcripts are coexpressed with the UGT2B7 enzyme in human liver, kidney, and small intestine, and all three UGT2B7 expressed proteins are found in the same subcellular compartments, thus suggesting that these interactions might occur in vivo. However, UGT2B7 splice variants were determined to have no effect on the glucuronidation activity of estradiol at position 3 that is performed by UGT1As, namely UGT1A1, suggesting that their modulatory effects could be specific to the UGT2B7 enzyme. This observation is consistent with a previous study showing that coexpression of UGT2A1_i1 or UGT2A1_i2 with other UGT1A or UGT2B enzymes caused no change in UGT1A or UGT2B glucuronidation activity (Bushey and Lazarus, 2012). Additional investigations are needed to better understand the oligomerization of UGT2B7 with its alternative splicing partners as well as with other UGT pathways and its effect on glucuronidation activity in vivo.
Because i2 and i4 lack the transmembrane domain 493VIGFLLVCVATVIFIV509 (Mackenzie, 1986; Radominska-Pandya et al., 2010), their presence in Golgi and ER membranes and not exclusively in the cytosol observed by immunofluorescence could seem unexpected; however, C-terminal variants of UGT2B4 and UGT1A also lacking this conserved domain were found to be ER membrane-bound (Lévesque et al., 2007, 2010). An explanation for membrane retention of a fraction of the total amount of i4 could reside in the presence of two positively charged residues, lysine and arginine, near its alternative C-terminal end (364DIK+R+M369L), a feature very similar to the retention of UGT1A_i2s as a consequence of its C-terminal dilysine motif (Lévesque et al., 2007). Moreover, three of the other four amino acids are hydrophobic (isoleucine, methionine, leucine), also potentially favoring membrane retention (Rösch et al., 2000). The terminal sequence of i2 comprises a stretch of seven hydrophobic residues and one positively charged lysine of 12 residues (444IAASCGNCMFMK456), a sequence that could participate in the retention of the variant polypeptide in the ER. We also report the presence of i1, i2, and i4 in the Golgi, a feature that is compatible with scarce reports of weak UGT activity in this organelle (von Bahr et al., 1972; Bossuyt and Blanckaert, 2001). However, it is also plausible that these three proteins undergo chemical modifications in the Golgi; indeed, UGT2B7 is glycosylated (Nagaoka et al., 2012), a chemical modification mainly initiated and/or terminated in the Golgi (Stanley, 2011). Western blotting of centrifugation fractions and comparison of UGT2B7 signals with subcellular compartment–specific markers further substantiate that i1, i2, and i4 are all residents of the ER and Golgi, whereas i2 and i4 can also be found in the cytoplasm, in which they may have roles distinct from that of i1 inhibition. On the other hand, this observation could also mean that a fraction of the pool of UGT2B7 truncated proteins is recognized by the ER ultraspecialized machinery as being misfolded and thus sent back to the cytosol to be degraded by the proteasome (Petris et al., 2011). The consensus is that misfolded and/or deleterious polypeptides are recognized at one of various checkpoints by the proteins of the ER-associated degradation pathway and are retrotranslocated and proteolyzed before their migration to the cis region of the Golgi (Ellgaard and Helenius, 2003). Thus, the apparent localization of i2 and i4 in both the ER and Golgi along with the UGT2B7 enzyme suggests that these isoforms fold correctly and thus generally are not subjected to degradation, which is also substantiated by the observed alteration of the glucuronidation phenotype upon their overexpression or partial depletion.
Levels of endogenous i2 and i4 relative to the i1 protein could not be measured. However, we observed that the coexpression of the UGT2B7 enzyme in HEK293 cells with inferior levels of the i2 or i4 proteins (35% and 10% of i1, respectively) yielded a significant repression of AZT-G formation. In hepatic cell lines (HepG2 and C3A), the relative expression of v5 and v7 mRNAs was less than 10% that of v1 transcripts encoding the UGT2B7 enzyme with significant changes in UGT2B7-mediated activity (up to 72% increased AZT-G formation) upon partial depletion of endogenous v5 and v7 variants. Data suggest that high levels of splice forms might not be required to have a meaningful effect on cellular glucuronidation. The mRNA levels of UGT2B7 splice forms observed in hepatic cell lines are similar to what was reported recently for v2/v3 splice variants encoding UGT1A_i2s in human liver samples (Jones et al., 2012). In addition, Court et al., 2013 suggest that UGT1A-3′UTR single nucleotide polymorphisms associated with a decreased risk of unintentional acetaminophen-induced acute liver failure might regulate exon 5a/5b splicing in the liver. One could speculate that changes in UGT splicing ratios, even when splicing forms are not abundant relative to transcripts producing UGT enzymes, may have important phenotypic effects.
We conclude that a potential novel autoregulatory mechanism of the UGT2B7 glucuronidation pathway is mediated by the alternative proteins.
Acknowledgments
The authors thank Patrick Caron for mass spectrometry–based assays for glucuronide detection as well as Olivier Eap, Mario Harvey, and Lyne Villeneuve for technical support.
Authorship Contributions
Participated in research design: Guillemette.
Conducted experiments: Ménard, Collin, Margaillan.
Performed data analysis: Ménard, Guillemette.
Wrote or contributed to the writing of the manuscript: Ménard, Collin, Margaillan, Guillemette.
Footnotes
- Received July 25, 2013.
- Accepted September 30, 2013.
This research was supported by the Canadian Institutes of Health Research (CIHR) [Grant MOP-42392]. V.M. received a CIHR Frederick Banting and Charles Best studentship award. G.M. received a studentship award from the Fonds de l’enseignement et de la recherche of the Laval University Faculty of Pharmacy. P.C. received a studentship award from the Fonds de recherches en santé du Québec. C.G. holds the Canada Research Chair in Pharmacogenomics.
↵
This article has supplemental material available at dmd.aspetjournals.org.
Abbreviations
- AZT
- zidovudine
- ER
- endoplasmic reticulum
- i
- isoform
- P450
- cytochrome P450
- PCR
- polymerase chain reaction
- qPCR
- quantitative polymerase chain reaction
- RT
- reverse transcription
- siRNA
- short interfering RNA
- UGT
- UDP-glucuronosyltransferase
- v
- variant
- Copyright © 2013 by The American Society for Pharmacology and Experimental Therapeutics
