The naming of the UGT1A1 isoform 2 (UGT1A1_i2 for protein and UGT1A1_v2 for gene product) was done according to the Human Gene Nomenclature Guidelines and was approved by the UGT nomenclature committee.
Potential conflict of interest: Nothing to report.
UDP-glucuronosyltransferase 1A1 (UGT1A1) is involved in a wide range of biological and pharmacological processes because of its critical role in the conjugation of a diverse array of endogenous and exogenous compounds. We now describe a new UGT1A1 isoform, referred to as isoform 2 (UGT1A1_i2), encoded by a 1495-bp complementary DNA isolated from human liver and generated by an alternative splicing event involving an additional exon found at the 3′ end of the UGT1A locus. The N-terminal portion of the 45-kd UGT1A1_i2 protein is identical to UGT1A1 (55 kd, UGT1A1_i1); however, UGT1A1_i2 contains a unique 10-residue sequence instead of the 99–amino acid C-terminal domain of UGT1A1_i1. RT-PCR and Western blot analyses with a specific antibody against UGT1A1 indicate that isoform 2 is differentially expressed in liver, kidney, colon, and small intestine at levels that reach or exceed, for some tissues, those of isoform 1. Western blots of different cell fractions and immunofluorescence experiments indicate that UGT1A1_i1 and UGT1A1_i2 colocalize in microsomes. Functional enzymatic data indicate that UGT1A1_i2, which lacks transferase activity when stably expressed alone in HEK293 cells, acts as a negative modulator of UGT1A1_i1, decreasing its activity by up to 78%. Coimmunoprecipitation of UGT1A1_i1 and UGT1A1_i2 suggests that this repression may occur via direct protein–protein interactions. Conclusion: Our results indicate that this newly discovered alternative splicing mechanism at the UGT1A locus amplifies the structural diversity of human UGT proteins and describes the identification of an additional posttranscriptional regulatory mechanism of the glucuronidation pathway. (HEPATOLOGY 2007;45:128–138.)
Glucuronidation, catalyzed by UDP-glucuronosyltransferase (UGT) enzymes, represents a major phase II conjugation pathway in humans and is involved in the metabolism and excretion of several endogenous and exogenous compounds. UGTs are membrane glycoproteins located in the endoplasmic reticulum (ER) that transfer the glucuronic acid moiety from UDP–glucuronic acid to aglycone substrates, resulting in an increased polarity of the substrate that facilitates its excretion from the body through bile and urine. This metabolic process is involved in the elimination of bilirubin, steroids, bile acids, toxic dietary components, and several drugs, including morphine, irinotecan, and mycophenolate mofetil, to name a few.1–3
The UGT gene superfamily includes four mammalian UGT families, namely UGT1, UGT2, UGT3, and UGT8.4 Over the last few years, the UGT1 and UGT2 gene structures have been extensively characterized in humans; together, these genes encode 16 functional proteins.5, 6UGT1A comprises 17 exons, spans more than 198 kb, is located on chromosome 2q37, and encodes 9 functional UGT1A proteins. This gene is characterized by 13 first exons 1 shared to common exons 2 to 5, the latter encodes the C-terminal region of UGT proteins with the UDP–glucuronic acid binding and transmembrane domains.6–8 Each unique exon 1, which is transcribed by unique promoter for tissue-specific expression, encodes the amino-terminal half of the protein that imparts aglycone specificity.
UGT1A1 is one of the most studied UGT enzymes due to its major role in the biliary excretion of bilirubin, a toxic breakdown product of heme metabolism. Its physiological role is also exemplified by its involvement in the conjugation of steroid and thyroid hormones.9, 10 Genetic polymorphisms in the promoter region of UGT1A1, for instance, are associated with reduced transcriptional activity and result in Gilbert's syndrome (mild unconjugated hyperbilirubinemia). In turn, more significant lesions in UGT1A1 can lead to severe forms of hyperbilirubinemia known as Crigler-Najjar type I and II.11–13 Additional evidence supports a critical role for UGT1A1 in the metabolism of many therapeutic drugs. UGT1A1 is involved in the metabolism of the topoisomerase I inhibitor irinotecan, the topoisomerase II inhibitor etoposide, and the oral contraceptive steroid 17α-ethinyl estradiol.14–16 Therefore, any genetic and/or environmental influences that alter the glucuronidation activity of UGT1A1 may have significant physiological and pharmacological consequences.
Recent preliminary data from our laboratory using a specific antibody against human UGT1A1 highlight the presence of immunoreactive proteins of lower molecular weight in several human tissues. These data suggest that additional forms of this enzyme may exist. Furthermore, a search of public databases revealed a shorter UGT1A complementary DNA (cDNA), which supports the existence of a new class of UGT proteins. We describe the isolation and the characterization of a new UGT1A1 isoform 2 (UGT1A1_i2) that is generated by alternative splicing of an additional exon in the common 3′ region of UGT1A. We show that UGT1A1_i2 negatively regulates the function of the conjugating UGT1A1 enzyme (UGT1A1_i1, isoform 1).
cDNA, complementary DNA; ER, endoplasmic reticulum; UGT, UDP-glucuronosyltransferase.
Materials and Methods
UDP-glucuronic acid was obtained from Sigma (St. Louis, MO), geneticin (G418) was obtained from Wisent Inc. (St.-Bruno, Canada), blasticidin was obtained from Invitrogen (Carlsbad, CA), and Lipofectin Reagent was obtained from Stratagene (La Jolla, CA). Protein assay reagents were obtained from Bio-Rad (Richmond, CA). HEK293 cells were obtained from the American Type Culture Collection (Manassas, VA). Total RNA from human tissues was purchased from Ambion (Austin, TX). Four of the liver tissue samples used in this study were kindly provided by Dr. Ted T. Inaba from the University of Toronto.17 All other tissues and microsomes were purchased from Tissue Transformation Technology (Edison, NJ). Superscript II reverse transcriptase was obtained from Invitrogen (Carlsbad, CA). Estradiol (E2) and its metabolites were purchased from Steraloids (Newport, RI).
Isolation of the Human UGT1A1_v2 cDNA.
UGT1A1 variant 2 (UGT1A1_v2) was amplified by RT-PCR from total RNA obtained from human liver. Total RNA (1 μg) was denatured in the presence of 100 pmol of oligo(dT) primer at 65°C for 15 minutes. The reaction was performed at 42°C in 20 μl containing Tris-HCl (pH 8.3), 2.5 mM MgCl2, 10 mM dithiothreitol, and 0.5 mM of each dNTP. The RNA was incubated at 42°C for 5 minutes before the addition of 200 U of Superscript II (Gibco BRL). The reaction was then incubated at 42°C for 50 minutes and at 70°C for 15 minutes and chilled on ice. Specific 3′ oligonucleotides were designed for intron 4 of the common region of UGT1A upstream of the putative polyadenylation signals. The UGT1A1_v2 cDNA was cloned via PCR amplification as follows. The PCR reaction (50 μl) contained 2 mM MgCl2, 0.2 mM of each dNTP, and 0.4 μM of each primer (sense #1483 5′-GAGAGAAAGCTTCGAACCTCTGGCAGGAGCAAA-3′ and antisense #1484 5′-GAGAGACTCGAGTATCCAGTGCCACCACACACACATTAGCACCTCAAA-3′). The sense primer was designed to incorporate a HindIII restriction endonuclease cleavage site 5′ of the initiation ATG codon, whereas the antisense primer contained a Xho1 site 3′ of the stop codon. Finally, 2 U of Taq polymerase were added, and the reaction was incubated at 94°C for 30 seconds followed by 40 cycles of 94°C for 30 seconds, 55°C for 30 seconds, and 72°C for 1 minute, with a final incubation at 72°C for 5 minutes. The amplified product of 1,495 bp was separated on a 1% agarose gel and digested with specific restriction endonucleases as mentioned above, subcloned into vectors pcDNA3 and pcDNA6, and sequenced.
Tissue Distribution of the UGT1A1_v1 and UGT1A1_v2 mRNA.
Reverse transcription was performed in 20 μl with 1 μg of total human RNA and with pdN6 random hexamer primers according to the manufacturer's protocol (Invitrogen). First, using a common oligonucleotide located in exon 2 in UGT1A (#1470 5′-GAATTTGAAGCCTACATTAATGCTTCTGGAGAACAT-3′) along with a 3′ oligonucleotide located in exon 5b for UGT1A_v2 (#1528 5′-TCACATCTGTCTTCCTGACTGC-3′) or exon 5a for UGT1A_v1 (#1529 5′'-TCAATGGGTCTTGGATTTGTGG-3′), we performed PCR reactions using cDNA samples from tissues. PCR conditions were 95°C for 1 minute for denaturation, followed by 40 cycles at 95°C for 30 seconds, 55°C for 30 seconds, and 72°C for 1 minute, with a final extension at 72°C for 7 minutes. PCR products were purified on Qiagen quick columns (Qiagen Inc., Mississauga, Ontario, Canada) and sequenced. UGT1A1_v2 mRNA was amplified with 1 μl of the reverse transcription reaction with primers #615 5′-GAGAGAGGTGACTGTCCAGGAC-3′ and #1528, and UGT1A_v1 with primers #615 and #1529 using the same PCR conditions above.
Protein Expression Analysis.
The presence of UGT1A1_i1 and UGT1A1_i2 proteins was determined with a specific polyclonal antibody against UGT1A1 (Ab1A1#518) raised against the N-terminal portion (residues 63-144) of UGT1A118 and with a specific commercial UGT1A1 antibody (BD Gentest, San Jose, CA). Immunoreactive bands were visualized using a chemiluminescence kit (ECL; Perkin Elmer, Woodbridge, Ontario, Canada) with exposure to Kodak XB film. The relative levels of UGT proteins were determined by integrated optical density using the Bioimage program Visage 110S (Genomic Solution Inc., Ann Arbor, MI).
Stable Expression of UGT1A1 Isoforms 1 (UGT1A1_i1) and 2 (UGT1A1_i2).
HEK293 cells were transfected with pcDNA3.1/UGT1A1_i1-HisMyc, pcDNA6/V5-HisA-UGT1A1_i2 and pcDNA3/UGT1A1_i2 expression plasmids as described previously, using geneticin (1 mg/ml) or blasticidin (10 μg/ml).19 Several clones expressing both UGT1A proteins (clones #2 and #10) were isolated from the initial pool to obtain different expression levels of these two proteins in the same cell and to ensure stable expression level over time.
Preparation of Microsomal Fractions and Enzymatic Assays.
Membrane fractions were prepared from 800 × 106 cells of each population as described9 with cells disrupted using 3 × 10 seconds of sonication. Enzymatic assays were performed using 20 μg of proteins as previously reported.20 Reactions were initiated by addition of E2 (5, 25, 75, and 200 μM) or bilirubin (200 μM). All bilirubin assays were performed under minimal light conditions for 1 hour, or 3 hours for estradiol, at 37°C. Bilirubin and estradiol assays were stopped with 100 μl ice-cold methanol (0.02% butylated hydroxytoluene) and 100 μl ice-cold methanol, respectively. Estradiol glucuronide measurements were performed as described previously9 and activities are expressed as pmol glucuronides/min/mg protein. Bilirubin-glucuronide analysis was performed via high-performance liquid chromatography (Alliance 2690, Waters, Milford, MA) using a Luna C8 100 × 4.6-mm 3-μm column (Phenomenex, Torrance, CA) and eluted at a flow rate of 0.9 ml/min. The initial conditions were 30% A (H2O, 1 mM ammonium formate) and 70% B (methanol, mM ammonium formate) followed by a linear gradient up to 90% B in 2.5 minutes. The effluent from the high-performance liquid chromatography system was connected directly to an API 3000 triple quadrupole mass spectrometer (Sciex, Toronto, Canada) equipped with a turbo-ion spray source with a split of 1:4 in positive mode. The mass spectrometer was operated in the multiple reaction monitoring modes using the following conditions: spray probe temperature at 500°C, ionization voltage at 5,000 V, and the orifice and the ring at 45 V and 200 V, respectively. Data were acquired with a dwell time of 400 milliseconds, a pause time of 5 milliseconds, and a scan time of 1.2 seconds. Bilirubin glucuronidation activities were calculated as area/min/mg protein and are expressed as a percentage of activity versus UGT1A1_i1 activity. All enzyme activities were subsequently divided by the UGT1A1_i1 content.
Stable HEK-293/UGT1A1_i1/UGT1A1_i2 and control HEK-293 cells (75 × 103) were used for these experiments as described.9 For UGT1A1_i1, rabbit anti-c-myc primary antibody (Sigma-Aldrich) was used, and for UGT1A1_i2, mouse anti-V5 primary antibody (Invitrogen, Burlington, Canada) was used; both antibodies were diluted 1:200. For UGT1A1_i1 and UGT1A1_i2 a goat anti-rabbit secondary antibody 1:500 (Alexa Fluor 488, green and Alexa Fluor 594, red, respectively) were used. The expression of the ER resident protein calnexin was also assessed using a rabbit anti-calnexin primary antibody (Stressgen Biotechnologies, Victoria, Canada) diluted 1:200. Visualization and image acquisition was achieved using a fluorescent microscope with a ×100 oil objective coupled with a digital camera.
Protein G Sepharose 4 fast flow (GE Healthcare, Piscataway, NJ) was centrifuged at 12,000g for 20 seconds and washed three times with 1 ml of high-salt buffer (500 mM NaCl, 1% [vol/vol] IGEPAL, 50 mM Tris [pH 7.5]). At the final wash, the beads were incubated for 1 hour at 4°C with rocking in 1 ml of buffer. Finally, the solution was centrifuged at 12,000g for 20 seconds, and a 50% slurry mix was prepared by adding an equivalent volume of high-salt buffer. Both solubilized and sonicated microsomes were used for coimmunoprecipitation experiments. Solubilized microsomes were prepared as described previously.21 Briefly, microsomes from HEK-293–transfected cells were incubated for 30 minutes at 4°C in 25 mM Tris-HCl (pH 7.4) containing 0.8% (w/v) sodium cholate, 0.1 mM dithiothreitol, and 20% glycerol. The solution was centrifuged at 105,000g for 60 minutes, and the supernatant was collected for co-immunoprecipitation experimentation. Microsomes (50 μg) were mixed with 1 μg of specific monoclonal antibody (Invitrogen, Burlington, Ontario, Canada) in 1 ml of high-salt buffer and incubated at 4°C with 50 μl of protein G Sepharose 4 fast flow (50% slurry) for 15 hours. The beads were washed 3 times with 1 ml high-salt buffer and finally with 1 ml 50 mM Tris (pH 7.5). Beads containing the immunoprecipitated proteins were resuspended with 30 μl of ×1 SDS-PAGE solution, heated at 100°C for 2 minutes, and centrifuged at 12,000g for 20 seconds. The supernatant was analyzed via SDS-PAGE. The membrane blots were probed with a specific monoclonal antibody linked with horseradish peroxidase (Invitrogen, Burlington, Ontario, Canada).
Identification of a New cDNA Encoding UGT1A1 Isoform 2 (UGT1A1_i2).
A cDNA that lacks the portion encoded by the terminal exon 5 at the UGT1A locus (GenBank AF297093) was recently isolated from human kidney (GenBank BC053576). The deduced amino acid sequence of the protein predicts a shorter C-terminal portion compared with other UGT1As. Based on this observation, we screened the common region of UGT1A for additional putative exons and identified an intronic acceptor site followed by a 28-bp open reading frame associated with several downstream polyadenylation signals (GenBank AF297093). We subsequently designed specific oligonucleotides that recognize these regions to isolate potential UGT1A1 cDNAs via RT-PCR. In human liver, a 1,495-bp UGT1A1 cDNA was isolated with an open reading frame of 1,335 bp and a 3′-untranslated region of 133 bp (GenBank DQ364247). The sequence of a new shorter UGT1A1 cDNA, UGT1A1 variant 2 (UGT1A1_v2), corresponded to exons 1-4 flanked with a previously unidentified 28-bp 3′ sequence followed by a stop codon (TGA) (31 bp total) (Fig. 1B, right panel). The sequence from exon 2 to the 3′ UTR was identical to the cDNA sequence BC053576. Comparison of the sequence with the full-length UGT1A1 cDNA (UGT1A1_v1 herein) revealed that UGT1A1_v2 lacked 298 bp (encoding 99 amino acids) of the common exon 5. This portion is replaced by a shorter sequence of 31 bp encoded by a new exon located in intron 4 (GenBank AF297093) (Fig. 1A). Thus, the exon 5 originally described by Ritter et al.7 was renamed exon 5a, and the new exon located in intron 4 was referred to as exon 5b (Fig. 1A).
The UGT1A1_v2 cDNA is predicted to encode a 45-kd protein. Its deduced primary structure (Fig. 1B) reveals an ER-targeting signal peptide and complete binding sites for UDP-GlcA and aglycone substrates; however, it also reveals a lack of a characteristic hydrophobic transmembrane domain (residues 445-530) encoded by exon 5a. The 10–amino acid sequence (RKKQQSGRQM) encoded by the newly discovered exon 5b contains a typical dilysine motif KKXX, presumably for ER retention, as well as four positively charged residues that may promote the interaction of the protein with negatively charged ER membranes.
Wide Expression Profile of UGT1A1_v2 in Human Tissues.
Tissue distribution of UGT1A1 isoform 2 was studied via RT-PCR and Western blotting. Using a common oligonucleotide located in exon 2 of UGT1A (#1470; Fig. 2A) associated with a 3′ oligonucleotide located in the new exon 5b (#1528) or an oligonucleotide in exon 5a (#1529) (Fig. 2A), it was demonstrated that both UGT1A amplicons are expressed in all the human tissues tested (Fig. 2B). We then used a specific UGT1A1 primer for exon 1 (#615; Fig. 2A) combined with each of the 2 primers for exons 5b and 5a, which yielded a 886-bp UGT1A1_v1 amplicon in the liver, colon, and small intestine, in agreement with previous reports.22 On the other hand, the 619-bp amplicon specific to the UGT1A1_v2 transcript was expressed in the same tissues as UGT1A1_v1 with the exception of the kidney, which showed only the short form for UGT1A1_v2. These data suggest that both UGT1A1 spliced mRNAs are not always coexpressed in the same tissues.
To correlate gene expression data with protein expression, we used a UGT1A1-specific antibody18 in addition to the commercially available anti-UGT1A1. We detected at least 2 forms of UGT1A1 of different molecular weights (Fig. 3). Human liver microsomes expressed the 55-kD UGT1A1_i1 and a protein of approximately 45 kD, consistent with the predicted molecular weight of UGT1A1_i2. In agreement with RT-PCR results, the kidney expressed only the 45-kD UGT1A1_i2 protein.
To evaluate the relative expression of both proteins in individual liver samples and explore their subcellular localization, purified microsomes and tissue homogenates from 4 liver samples were studied via semiquantitative Western blots in parallel with a commercially available pool of liver microsomes (Fig. 4A). UGT1A1_i2 was detected in all tissue samples tested, and the intensity of the bands ranged from 5%-57% (homogenates) and 2%-7% (microsomes) relative to the UGT1A1_i1 bands. In all liver samples, the relative abundance of UGT1A1_i2 was higher in crude tissue homogenates compared with microsomes, suggesting that the UGT1A1_i2 form is not exclusively a resident of the ER (Fig. 4A, lower panel). Furthermore, in jejunum microsomes samples, UGT1A1_i2 was 50% lower compared with UGT1A1_i1, whereas ileum microsomes had equivalent amounts of both proteins. These observations suggest differential regulation of the expression of these proteins along the gastrointestinal tract with stronger expression of UGT1A1_i2 in the distal gastrointestinal tract and stronger UGT1A1_i1 expression in the proximal intestine.
Preparation of UGT1A1_i1/UGT1A1_i2 HEK293 Cell Lines and Assessment of Their Enzymatic Functions.
For functional experiments, the UGT1A1_v2 cDNA was subcloned into pcDNA3 and pcDNA6/HisA-V5 vectors and stably transfected into HEK293 cells and a HEK293/UGT1A1_i1 stable cell line, respectively. These coexpression experiments were critical based on 2 aspects: 1) UGT1A proteins form homodimers/tetramers23, 24 and 2) UGT1A1_i1 and UGT1A1_i2 are often expressed in the same tissue (Fig. 4B).
Western blot analysis confirmed the presence of a 45-kD protein in the HEK293-UGT1A1_i2 line and a 60-kD protein in the HEK293-UGT1A1_i1-His-Myc line (Fig. 5). Cells expressing both proteins showed a 48-kD UGT1A1_i2 because a His-V5 tag sequence (≈3 kD) was introduced in frame with the UGT1A1_i2 sequence for further characterization experiments. UGT1A1_i1 is expressed significantly higher in microsomes compared with homogenates (Fig. 5B). In contrast, UGT1A1_i2 is expressed more highly in homogenates compared with microsomes, suggesting possible extramicrosomal localization of the shorter UGT1A1_i2 protein as inferred with human tissues. In UGT1A1_i1/_i2 cell microsomes, UGT1A1_i2 expression was 2-fold higher than UGT1A1_i1.
We next assessed the enzymatic function of UGT1A1_i2 expressed alone or coexpressed with UGT1A1_i1 using the UGT1A1 substrates bilirubin and estradiol. UGT1A1_i2 had no detectable transferase activity for either substrate (Fig. 6 A, B). Next, we used the stably cotransfected HEK293-UGT1A1 _i1+_i2 cells to test if UGT1A1_i2 alters UGT1A1_i1 activity. In the presence of UGT1A1_i2, UGT1A1_i2 transferase activity for bilirubin and estradiol was significantly lower compared with UGT1A1_i1 microsomes from both stable cell lines (Fig. 6). The inhibition of UGT1A1_i1 activity by UGT1A1_i2 for both bilirubin and estradiol varied from 21%-78% and was significant for all substrate concentrations tested. These differences in velocity cannot be explained by differences in microsomal UGT1A1_i1 concentration because the data were normalized to the amount of UGT1A1_i1 (Fig. 6C).
Subcellular Localization of UGT1A1_i1 and UGT1A1_i2.
To further explore the subcellular localization of UGT1A1_i2, a series of immunofluorescence analyses were performed with the 2 antibodies against UGT1A1 (Ab#518) and anti-UGT1A (Ab#RC-71) using HEK-293 cells stably expressing UGT1A1_i1 or UGT1A1_i2. Both proteins were expressed in the ER and perinuclear structure (data not shown). Colocalization analyses were then performed with anti-Myc and anti-V5 for UGT1A1_i1 and UGT1A1_i2 visualization, respectively, in cells expressing both proteins. Subsequent colocalization analyses confirmed the colocalization of UGT1A1_i1 and UGT1A1_i2 in the ER as well as in the perinuclear structure (Fig. 7).
Interaction Between UGT1A1_i1 and UGT1A1_i2.
Based on the inhibitory effect of UGT1A1_i2 on UGT1A1_i1-mediated glucuronidating activity and their partial subcellular localization, we tested if UGT1A1_i1 and _i2 interact by using coimmunoprecipitation assays with microsomes from stable cell lines expressing both isoforms. Western blotting showed a 48-kD protein corresponding to UGT1A1_i2-His/V5, indicating that UGT1A1_i2 coprecipitates with UGT1A1_i1 (Fig. 8, lane 5). Coimmunoprecipitation experiments with UGT1A1_i1 or UGT1A1_i2 revealed that anti-Myc did not cross-react with UGT1A1_i2 (Fig. 8, lanes 3 and 4). These observations suggest that UGT1A1_i1 and UGT1A1_i2 may interact directly.
We report the isolation and characterization of a new UGT1A1 protein, UGT1A1_i2, generated by alternative splicing of an additional exon in the common 3′ region of the gene UGT1A. UGT1A1_i2 comprises 444 residues, with the first 434 residues identical to UGT1A1_i1; the conserved C-terminal domain encoded by exon 5a of UGT1A1_i1, however, is replaced by a 10-residue sequence encoded by an additional exon (5b) located within intron 4. Despite these structural changes, UGT1A1_i2 retains the consensus substrate and cosubstrate binding domains—encoded by exons 1-4—that are characteristic of UGT proteins. UGT1A1_i2 resides mainly in the ER, and in human tissues it is widely expressed, although the expression level varies between individuals. Functional enzymatic data strongly imply that UGT1A1_i2, which lacks transferase activity when expressed alone, acts as a negative modulator of the conjugating activity of UGT1A1_i1 potentially through direct protein–protein interactions.
We have shown that the UGT1A locus encodes one additional exon and that the common region of the UGT1A locus comprises 5 common exons (exons 2-5a and 5b) instead of 4 (exons 2-5), as described.6, 7 We further demonstrated that exons 5a and 5b are alternatively spliced in human tissues, generating 2 different UGT1A1 mRNAs, and that the levels of these transcripts vary significantly in different tissues and among individuals. These data suggest that other transcriptional regulatory mechanisms are involved in the expression of a shorter form of UGT1A1—namely UGT1A1_i2—in various human tissues, in addition to those mechanisms that dictate tissue-specific expression for UGT1A family members.25 In support of these results, we observed that UGT1A1_i1 and UGT1A1_i2 mRNAs are differentially expressed in human hepatic and extrahepatic tissues, namely the liver, colon, and small intestine (Fig. 2).
At physiological levels, UGT1A1 represents a critical enzyme involved in the inactivation of bilirubin as well as several steroid hormone molecules and therapeutic drugs. The constitutive expression of UGT1A1 is under the influence of a common polymorphic repeat sequence (TA5-8) located in its regulatory region. The presence of the TA repeat is now recognized as one of the key factors contributing to interindividual and racial heterogeneity in UGT1A1 protein levels and bilirubin-conjugating capacity in humans.13, 26 The most frequent allele, UGT1A1*28 (TA7), is associated with a 30%-50% reduction in transcriptional activity and is linked to mild hyperbilirubinemia, susceptibility to cancer, and drug-induced toxicity.26–29 The reduction in UGT1A1_i1 conjugating activity, apparently induced by the presence of the UGT1A1_i2 in vitro, suggests that the expression of the isoform 2 is an additional determinant of variable glucuronidation capacity. We propose that the extent of the role of this posttranscriptional regulatory mechanism is dictated by the UGT1A1_i2 expression level. Alternative splicing-induced variant proteins that exert repressive activity have been reported for many proteins.30, 31
The tissue expression pattern for UGT1A1_i2 suggests that the functional consequences of its putative negative regulatory function predominate in extrahepatic tissues. In the liver, the UGT1A1_i2 level in microsomes appears to be minimal (<10%) compared with UGT1A1_i1; this proportionality probably is needed to preserve the high level of glucuronidation activity that maintain homeostasis of key factors such as bilirubin. In contrast, UGT1A1_v2 seems to be enriched in the kidney and in the proximal and distal segments of the small intestine (jejunum and ileum). The presence of high levels of the UGT1A1_v2 form in the kidney remains undefined and deserves further investigation. Interestingly, the UGT1A1_v1 content decreases along the distal gastrointestinal tract, whereas that of UGT1A1_v2 increases; this trend would be expected to effect a progressive decline in transferase activity. Therefore, the inhibitory effect of UGT1A1_i2 on UGT1A1_i1 represents an additional posttranscriptional mechanism involved in the fine tuning of endogenous and exogenous metabolism in humans and is predicted to be greater in extrahepatic tissues. It is proposed that specific regulatory mechanisms may exist for the processing of both the 5′ and 3′ end regions of UGT1A.
We found that UGT1A1_i2 has no detectable UGT activity toward common UGT1A1_i1 substrates, suggesting that the amino acid sequence encoded by exon 5a (residues 435-530) is required for UGT1A1 transferase activity. This requirement could reflect either a functional or structural role. Several studies demonstrated that truncation of specific UGT domains, including the C-terminal domain, impairs UGT enzymatic activity but does not prevent binding to the ER.32, 33 These studies also support our observed ER localization of UGT1A1_i2 despite the lack of the transmembrane domain; we note, however, that a portion of the protein did not localize within the ER (Fig. 7). The subcellular localization to the ER may be conferred, at least in part, by the dilysine motif KKXX, encoded by the new exon 5b, that can serve as an ER retention/retrieval signal.34, 35 Another study also suggested that Asp446, encoded by exon 5 of rat UGT1A6, is essential for proper secondary structure, overall folding, and activity of the enzyme.36 Because the entire exon 5a is missing in UGT1A1_i2, this isoform may fold differently than UGT1A1_i1, with a consequent loss of transferase activity.
Based on our findings, we hypothesize that UGT dimerization is one of the mechanisms involved in the reduced UGT1A1 glucuronidation activity caused by UGT1A1_i2. In support of this notion, several reports have described the homodimerization of UGT1A1 enzymes.23, 24, 37 Futhermore, Ghosh et al.20 reported that the interactions between UGT1A1 proteins are not abolished by partial deletion of the C-terminal domain and suggested that homodimerization of UGT1A1 may explain some of the dominant-negative effects of mutated UGT1A1 proteins. Homodimerization of UGT1A1 proteins is also supported by the clinical description of a Crigler-Najjar patient with only one nonsense mutation in the coding region of the gene, inferring that the dysfunctional protein had a dominant negative effect with severe clinical consequences.38 Our results are in agreement with the concept of a dominant-negative effect through direct protein–protein interactions between UGT1A1_i1 and UGT1A1_i2, as previously reported by Ghosh et al.24 This hypothetical mechanism is plausible because UGT1A1_i1 and UGT1A1_i2 both localize in microsomes (Fig. 4) where they potentially could interact. However, other mechanisms might also be responsible for the decrease in UGT1A1_i1 transferase activity induced by the presence of UGT1A1_i2. These two proteins could form homodimers, which could prohibit proper folding of the catalytic domain of UGT1A1_i1. This, in turn, could accelerate the degradation of the protein, thereby altering the net transferase activity of the complex toward various substrates. In addition, UGT1A1_i2 may bind specific substrates and/or cofactors that are required for UGT1A1_i1 activity.
In conclusion, the evolution of alternative splicing mechanisms at the UGT1A locus has served to amplify the structural diversity of these biologically and pharmacologically important proteins beyond the basic pattern provided by the distinct gene classes. It is expected that the alternative splicing mechanism occurring in the common region of the gene generates several other novel UGT1A_i2 proteins expressed in various tissues. Our preliminary findings support this statement (unpublished data). Thus, UGT1A expression may involve highly regulated and complex transcriptional and splicing mechanisms at the extremities of this locus, which may explain the observed diversity among UGT proteins expressed from this gene.
We thank Dr. Ted T. Inaba from the University of Toronto for kindly providing liver tissue from 4 individuals. We also thank Drs. Alain Bélanger and Olivier Barbier for critical reading of the manuscript and Olivier Bernard for helping with DNA preparations.