Abstract
The UDP glucuronosyltransferase (UGT) superfamily comprises glycoproteins that reside in the endoplasmic reticulum membranes and that undergo post-translational modifications (PTMs). UGT2B7 is of particular interest because of its action on a wide variety of drugs. Most studies currently survey common variants and examine only a small fraction of the genetic diversity; however, rare variants (frequency <1%) might have a significant effect because they are predicted to greatly outnumber common variants in the human genome. We discovered a rare single nucleotide UGT2B7 variant of potential pharmacogenetic relevance that encodes a nonconservative amino acid substitution at codon 121. This low-frequency variation, found in two individuals of a population of 305 healthy volunteers, leads to the translation of an asparagine instead of an aspartic acid (UGT2B7 p.D121N). This amino acid change was predicted to create a putative N-glycosylation motif NX(S/T) subsequently validated upon endoglycosidase H treatment of microsomal fractions and inhibition of N-glycosylation of endogenously produced UGT2B7 with tunicamycin in human embryonic kidney (HEK293) cells. The presence of an additional N-linked glycan on the UGT2B7 enzyme, likely affecting proper protein folding, resulted in a significant decrease of 49% and 40% in the formation of zidovudine and mycophenolic acid glucuronides, respectively. A systematic survey of the Short Genetic Variations database uncovered 32 rare, naturally occurring missense variations predicted to create or disrupt N-glycosylation sequence motifs in the other UGT2B enzymes. Collectively, these variants have the potential to increase the proportion of variance explained in the UGT pathway resulting from changes in PTMs, such as N-linked glycosylation with consequences on drug metabolism.
Introduction
Metabolic enzymes of the UDP-glucuronosyltransferase (UGT) superfamily catalyze glucuronidation reactions involved in the disposition of endogenous molecules, drugs, and other xenobiotics. UGTs are glycoprotein residents of the endoplasmic reticulum (ER) membranes that undergo post-translational modifications (PTMs) such as glycosylation and phosphorylation (Mackenzie 1990; Chakraborty et al., 2012; Riches and Collier 2015). This family of proteins is involved in drug metabolism, accounting for approximately 55% of the 200 most prescribed drugs (Guillemette et al., 2014). UGT2B7 is of particular interest because it is the most prevalent member of this enzyme family, and it conjugates almost one-fifth of all drugs known to be conjugated to glucuronic acid (Stingl et al., 2014). For instance, UGT2B7 is involved in the inactivation and elimination of carboxylic acid-containing drugs, including the nonsteroidal anti-inflammatory drugs, and of a large variety of other drugs, including opioids, anticonvulsant valproic acid, morphine, codeine, efavirenz, fenozfibric acid, mycophenolic acid (MPA), and zidovudine (AZT) (Stingl et al., 2014).
Glucuronide formation varies greatly between individuals and may be explained in part by the presence of single nucleotide polymorphisms (SNPs) (Guillemette et al., 2014). In the case of UGT2B7, there is a well-established connection to variations in drug efficacy and toxicity, especially for the most common polymorphism UGT2B7*2 allele, which encodes a nonconservative amino acid substitution His268Tyr in the substrate-binding domain linked to altered pharmacokinetics of several drugs but not all substrates examined (Bhasker et al., 2000; Stingl et al., 2014). Most pharmacogenomics studies currently survey common variants and thus are observing only a small fraction of the genetic diversity in any gene. Different forms of genetic variations within a UGT locus, including common and rare coding and regulatory variants, can exist and have separate, yet cumulative effects. Rare variants (allele frequency <1%) may also cause interindividual differences in therapeutic effects and adverse reactions to drugs but have been much less studied (Nelson et al., 2012). This is highly relevant since rare variants are predicted to greatly outnumber common variants in the human genome (Marth et al., 2011) and may be very important in the genomic contribution to treatment response and toxicity (Ramsey et al., 2012; Gillis et al., 2014). For example, it was recently established that as much as 92% of known variants in the cytochrome P450 superfamily have an allelic frequency below 1% and 83% below 0.1% (Fujikura et al., 2015). Similar distribution of variants by allelic frequency might be observed in UGT2B7 and other UGTs, underscoring the importance of studying rare variants.
Here we report the identification of a novel rare missense UGT2B7 variant (NM_001074.2:c.361G>A) in a cohort of 305 healthy white volunteers. This rare variation leads to the translation of an asparagine (Asn) instead of an aspartic acid (Asp) at position 121 of the UGT2B7 protein (NP_001065.2:p.Asp121Asn; D121N). Investigation of its functional impact using bioinformatics tools predicted that this variation creates a putative N-glycosylation site, which was experimentally validated. We then studied whether this additional PTM site affects glucuronidation of AZT and MPA, demonstrating a drastic reduction in UGT2B7 catalytic activity. Lastly, we established a comprehensive dataset of 32 rare variants potentially affecting the gain or loss of N-glycosylation sites in other family members, with some located in regions highly conserved across UGT genes, by integrating data from the dbSNP data base along with their respective allelic frequency. Collectively, these low-frequency variants have the potential to increase the proportion of variance explained in the glucuronidation pathway.
Materials and Methods
Genotyping of Healthy White Volunteers.
The variation at position 361 in the UGT2B7 gene causing amino acid change Asp121Asn was initially observed in one of 52 participants of a previous pharmacokinetic study (Levesque et al., 2007, 2008). These healthy volunteers were selected in a population of 305 white subjects who were subsequently genotyped for the variation at codon 121 using specific primers, as described (Levesque et al., 2007, 2008), and a second subject was found to carry this variation.
Heterologous Expression of the Variant UGT2B7 Enzyme and Enzymatic Assays.
A HEK293 cell line expressing the variant UGT2B7 protein at codon 121 (UGT2B7Asn121) was established by mutagenesis using primers 5′-CCCAACAACTCATCCTCTCTTAAAATTGAAA-3′ (forward) and 5′-TTTCAATTTTAAGAGAGGATGAGTTGTTGGG-3′ (reverse) based on the reference UGT2B7 cDNA (Menard et al., 2011). Relative quantification of UGT protein content in microsomal proteins was performed by immunoblot analysis using an anti-UGT2B antibody EL-93 (dilution 1:2000) (Lepine et al., 2004) and with an anti-calnexin antibody (dilution 1:5000; Stressgen Biotechnologies, Victoria, BC) to normalize for sample loading. Kinetic parameters were assessed for both cell lines in the presence of increasing concentrations of substrates, MPA (25–1500 µM; MP Biomedicals, Santa Ana, CA), or AZT (100–5000 µM, fixed UDPGA (5 mM); Sigma-Aldrich, St. Louis, MO) or cosubstrate UDPGA [50–5000 µM, fixed AZT (500 µM or 1 mM); Sigma Aldrich] for 1-hour incubation at 37°C. Enzymatic assays and liquid chromatography-mass spectrometry were performed to assess glucuronide (G) formation as described (Benoit-Biancamano et al., 2007). Absolute velocities (Vmax: pmol/min per millligram of protein) were normalized for UGT protein content assessed by Western blotting and expressed as relative Vmax (pmol/min per milligram of protein/UGT content). Kinetic parameters, according to the Michaelis-Menten model, were calculated with Sigma Plot 11 using the Enzyme Kinetics 1.3 module (SYSTAT Software Inc., San Jose, CA). Data are derived from at least two independent experiments performed in triplicates. P-value calculations using Student’s t test were performed with Excel 2016 (Microsoft Corporation, Redmond, WA).
Experimental Confirmation of N-Glycosylation Using Endoglycosidase Digestion.
Enzymes were obtained from New England BioLabs Inc. (Ipswich, MA). Microsomes (20 µg) from HEK293-UGT2B7 cell lines were incubated in Glycoprotein Denaturing Buffer (0.5% SDS, 40 mM dithiothreitol) for 10 minutes at 100°C to ensure denaturation of the protein content. For endoglycosidase H (Endo H) treatment, GlycoBuffer 3 (50 mM sodium acetate (pH 6 at 25°C) and Endo H (500 U) were added to microsomes, whereas for O-glycosidase treatment, microsomes were supplemented with GlycoBuffer 2 [50 mM sodium phosphate (pH 7.5 at 25°C)], NP-40 (1%), neuraminidase (100 U) and O-glycosidase (40 000 U), in a 20-µl final volume. Samples were incubated for 1 hour at 37°C and subsequently resolved by 10% SDS-PAGE using standard procedure. Immunodetection of UGT2B7 was conducted with a polyclonal UGT2B7 antibody (1:5000; 16661-1-APl ProteinTech Group, Rosemont, IL).
Inhibition of endogenous glycosylation of UGT2B7 was conducted as described with some modifications (Nakajuma et al., 2010). Briefly, HEK293 cells plated in 10-cm dishes were transiently transfected with 4 μg of pcDNA3 constructs driving the expression of reference UGT2B7 or the c121 variant (Menard et al., 2011) using LipofectAMINE 2000 (Invitrogen, Carlsbad, CA). Inhibition of glycosylation was achieved with tunicamycin (Sigma-Aldrich) added at the time of transfection (final concentration, 0.1 and 1 µg/ml). Controls consisted of vehicle only (0.1% ethanol). Cells were harvested 40 hours post-transfection by washing twice in phosphate-buffered saline, and then homogenates were prepared by harvesting cells in phosphate-buffered saline containing 0.5 mM dithiothreitol. To control for ER stress, cells were transfected as above, thapsigargin (0.5 µM; Sigma-Aldrich) was added at the time of transfection, and cells were harvested 24 hours post-treatment as described already. Protein concentration in homogenates was determined by a bicinchoninic acid assay. Immunodetection of UGT2B7 was performed as described.
Analysis of UGT2B Genetic Variations.
Single nucleotide variations (SNPs) for UGT2B coding sequences were retrieved from the dbSNP database using the National Center for Biotechnology Information browser (U.S. National Library of Medicine, Bethesda, MD). Alignments of UGT2B sequences were obtained using the Clustal O (1.2.1) multiple-sequence alignment tool. Coding variations were analyzed to establish whether they affect the sequence context Asn-X-Ser/Thr (create or disrupt NX(S/T) motif), where X is any amino acid except proline.
Results and Discussion
Our analysis of UGT2B7 gene sequences in a population of 305 healthy volunteers uncovered a novel missense variation observed in two subjects, corresponding to a nonsynonymous coding variation at codon 121 (Asp121Asn). A Sanger sequencing chromatogram is presented in Fig. 1A designating a double peak of an adenine (A) and a guanine (G) at position 361 (allelic frequency of 0.328% for the variant A allele). This population was previously used to identify candidates carrying specific UGT1A and UGT2B7 genetic variations for a pharmacokinetic study of mycophenolate mofetyl (Levesque et al., 2007, 2008). An in silico analysis for prediction of PTM sites using the NetNGlyc 1.0 tool (http://www.cbs.dtu.dk/services/NetNGlyc/), also confirmed with the GlycoEP software (Chauhan et al., 2013), revealed that this amino acid change creates a putative asparagine-linked N-glycosylation motif NX(S/T) consisting of an asparagine, followed two positions downstream by a threonine as pictured in Fig. 1D. Three such motifs are typically found in the UGT2B7 protein, of which glycosylation at positions 68 and 315 was experimentally validated (Nagaoka et al., 2012). Western blot analysis of HEK293 microsomal preparations showed that the variant UGT2B7Asn121 protein has a higher molecular weight than the UGT2B7Asp121 enzyme, which is consistent with the addition of an oligosaccharide by N-glycosylation. In contrast to O-glycosidase treatment, when treated with Endo H, oligosaccharides are removed, leaving both UGT2B7 proteins at the same lower molecular weight, thus confirming the presence of an additional N-glycosylation on the variant UGT2B7Asn121 protein (Fig. 1B). As further evidence of N-glycosylation causing the enhanced mobility shift of the UGT2B7Asn121 protein, N-glycosylation of endogenously produced UGT2B7 was inhibited with tunicamycin (Nakajima et al., 2010). Both the reference and variant UGT2B7 were detected as multiple protein bands with low drug concentration, suggesting a partially perturbed glycosylation, whereas the high drug concentration prevented the formation of slower migrating proteins, thus confirming that the mobility shift is caused by N-glycosylation (Fig. 1C). Inhibition was not due to a general ER stress impairing glycosylation, given that thapsigargin did not perturb the mobility of either UGT2B7 (Fig. 1C). We also used mass spectrometry analysis with the goal of detecting the glycosylated peptide containing the sequence at codon 121. This approach, using cellular fractions from HEK293-UGT2B7 cell models enriched for UGT2B7 by affinity purification and treated or not with endoglycosidase PNGase F, was inconclusive. It permitted detection of multiple UGT2B7 tryptic peptides (protein coverage up to 29%; data not shown) but not the codon 121-bearing peptide.
(A) Sanger sequencing chromatogram revealing an individual heterozygote for the novel nonsense variation at position 361 ((NC_000004.12:g.69096881G>A, NM_001074.2:c.361G>A, NP_001065.2:p.Asp121Asn)) of the UGT2B7 gene. The arrow indicates the position of the nucleotide variation. (B) N-glycosylation profile of the UGT2B7 enzyme (Asp121) and variant protein (Asn121) assessed by Western blot analysis. Microsomal preparations (20 μg) were untreated or treated with endoglycosidases to cleave attached oligosaccharides. (C) Inhibition of endogenous N-glycosylation. HEK293 cells transiently transfected to express UGT2B7 (Asp121) or variant UGT2B7 (Asn121) were treated with either vehicle (CTR), tunicamycin (0.1 or 1.0 µg/ml), or thapsigargin (0.5 µM) at the time of transfection. UGT2B7 were immunodetected in cell homogenates (10 µg). (D) Graphical representation of the UGT2B7 protein sequence with putative domains, N-glycosylation sites (blue), those experimentally validated (marked with a star (Nagaoka et al., 2012)), and the new site created in the UGT2B7Asn121 variant protein (red) as predicted with the NetNGlyc 1.0 tool. (E) Glucuronide formation by the UGT2B7Asp121 reference enzyme (blue) and the variant UGT2B7Asn121 (red) using varying concentrations of MPA (25–1500 µM), AZT (100–5000 µM) or UDPGA (50–5000 µM in the presence of 500 µM of AZT). (F) Partial sequence alignment of the 19 UGT1 and UGT2 enzymes in which putative N-glycosylation motifs (gray) and positions of rare variants (listed in Table 1) either creating (green) or disrupting (red) such motifs are indicated. Amino acid positions given above sequence alignments correspond to the UGT2B7 protein; letters represent clusters of putative N-glycosylation sites among UGTs (see Table 2).
Previous data support that N-glycosylation plays a significant role in the enzymatic activity of UGT2B7 (Barbier et al., 2000; Nagaoka et al., 2012), suggesting that the novel UGT2B7Asn121 variant may affect enzyme activity. A significant alteration in the conjugation of AZT and MPA was observed, with a decreased activity by 49% and 40%, respectively, associated with the variant Asn121 protein compared with the reference Asp121 enzyme, suggesting an altered protein folding. No significant differences were noted in the affinity (Km) of the enzyme (Table 1; Fig. 1E). Nagaoka and colleagues showed that disruption of N-glycosylation sites at position 68 and 315 by mutagenesis leads to significant changes in the activity of the UGT2B7 enzyme (Nagaoka et al., 2012). Similarly, abolition of the N-glycosylation of UGT2B15 resulted in decreased enzyme activity without changing Km (Barbier et al., 2000). In line, a rare UGT1A1 variant creating a glycosylation site (K402T) was reported to cause a drastic decrease in UGT activity associated with severe hyperbilirubinemia (Crigler-Najjar type 1) (Sneitz et al., 2010). These observations underscore the need for further research on the impact of the N-glyco variants on UGT2B7 protein function as well as on other UGTs.
Kinetic parameters for UGT2B7 substrates (AZT and MPA) using microsomal protein preparations isolated from HEK293 cells stably expressing the UGT2B7 enzyme (Asp121) and the novel variant UGT2B7 protein (Asn121)a
Our goal was next to build a framework for better understanding the effects of nonsynonymous variations on the N-glycosylation of UGT1 and UGT2 enzymes. Through a systematic survey of publicly available data (dbSNP database), we revealed numerous naturally occurring missense variations, most with low allelic frequency (below 0.01%), predicted to affect N-glycosylation of UGT enzymes (Fig. 1F; Table 2). Very little information is still available regarding the glycosylation profile of UGT enzymes, precluding us from establishing the potential functional relevance of these rare variants. Clustering of the variants according to their position in the UGT sequences helped show that many of the variants localize in the same regions of UGT1, UGT2A, or UGT2B enzymes. These potentially correspond to conserved and functional NX(S/T) motifs across several UGT proteins (Fig. 1F), hence making them interesting candidates for further in vitro validation. For example, the first cluster is noteworthy because N-glycosylation in this region of UGT2B7 and UGT2B15, on residues 68 and 65, respectively, was experimentally validated in liver tissue by a method combining multiple digestion and hydrazide chemistry (Chen et al., 2009). Likewise, codon 69 of UGT2B4, UGT2B11, and UGT2B28 is affected by naturally occurring rare variants creating at this position the final serine or threonine of an NX(S/T) glycosylation motif, whereas a similar motif is present in this region for UGT1A and UGT2A enzymes. A comprehensive survey of the impact of these additional rare variants on the N-glycosylation of UGT enzymes will be required to evaluate whether variations that lead to changes in the glycosylation pattern of a UGT protein can be damaging.
Naturally occurring rare variants in UGT2B family members predicted to create (gain) or disrupt (loss) N-glycosylation sites (NX(S/T) motif
Spacing groups variants according to their positions.
In conclusion, we discovered a rare UGT2B7 variant resulting in a NX(S/T) glycosylation gain that significantly affects the rates of drug glucuronidation. A more thorough understanding of the significance of this variant in the context of drug treatment would require genotyping of a larger population, initially to acquire a better appreciation of the allelic frequency across populations and to perform pharmacokinetic analyses in individuals carrying this variation. Likewise, our exhaustive analysis of data on variations in additional drug-conjugating UGT enzymes exposes numerous missense variations with low allelic frequency, potentially creating or disrupting N-glycosylation sites, with some in regions conserved across all UGT enzymes. Accordingly, a large proportion of variability in the UGT pathway may be due to rare variants of significant effect size with a profound impact on their biologic function resulting from changes in PTM, such as N-linked glycosylation.
Acknowledgments
The authors thank Patrick Caron and Véronique Turcotte for technical assistance and Michèle Rouleau for helpful discussion.
Authorship Contributions
Participated in research design: Benoit-Biancamano, Guillemette.
Conducted experiments: Girard-Bock, Benoit-Biancamano, Villeneuve, Desjardins.
Performed data analysis: Girard-Bock, Benoit-Biancamano, Villeneuve, Desjardins, Guillemette.
Wrote or contributed to the writing of the manuscript: Girard-Bock, Benoit-Biancamano, Guillemette.
Footnotes
- Received May 31, 2016.
- Accepted September 8, 2016.
This work is supported by the Canadian Institutes of Health (Grant CIHR MOP-42392). CGB received studentships from the Fonds d’enseignement et de recherche (FER) of Laval University’s Faculty of Pharmacy and from the Fondation Desjardins and Fondation du Centre Hospitalier Universitaire de Québec. C.G. holds a Canada Research Chair in Pharmacogenomics (Canadian Research Chair Program).
Abbreviations
- AZT
- zidovudine (3′-azido-3′-deoxythymidine)
- Asn
- asparagine
- Asp
- aspartic acid
- Endo H
- endoglycosidase H
- ER
- endoplasmic reticulum
- HEK293
- human embryo kidney 293 cells
- MPA
- mycophenolic acid
- PTM
- post-transational modification
- SNP
- single nucleotide polymorphism
- UDPGA
- UDP-glucuronic acid
- UGT
- UDP-glucuronosyltransferase
- Copyright © 2016 by The American Society for Pharmacology and Experimental Therapeutics
References
In this issue
A Rare Functional UGT2B7 D121N Variant
