Gilbert syndrome redefined: A complex genetic haplotype influences the regulation of glucuronidation§


  • Potential conflict of interest: Nothing to report.

  • Supported by the Deutsche Forschungsgemeinschaft SFB621 project C3 (to C.P.S.).


Gilbert syndrome (GS) is characterized by intermittent unconjugated hyperbilirubinemia without structural liver damage, affecting about 10% of the white population. In GS the UGT1A1*28 variant reduces bilirubin conjugation by 70% and is associated with irinotecan and protease inhibitor side effects. The aim of this study was to characterize potential in vivo consequences of UGT1A gene variability in GS. Three hundred GS patients (UGT1A1*28 homozygous) and 249 healthy blood donors (HBD) were genotyped for UGT1A (UGT1A1*28, UGT1A3-66 T>C, UGT1A6*3a, UGT1A7*3) and transporter single nucleotide polymorphisms (SNPs) (SCLO1B1 p.V174A, SCLO1B1 p.N130D, ABCC2 p.I1324I, ABCC2-24 UTR) using TaqMan-5′-nuclease-assays. A humanized transgenic UGT1A-SNP and corresponding wildtype mouse model were established carrying the GS-associated UGT1A variant haplotype. UGT1A transcript and protein expression, and transcriptional activation were studied in vivo. Homozygous UGT1A1*28 GS individuals were simultaneously homozygous for UGT1A3-66 T>C (91%), UGT1A6*2a (77%), and UGT1A7*3 (77%). Seventy-six percent of GS and only 9% of HBD were homozygous for the variant haplotype spanning four UGT1A genes. SCLO1B1 and ABCC2 SNPs showed no differences. In transgenic humanized UGT1A SNP and wildtype mice this UGT1A haplotype led to lower UGT1A messenger RNA (mRNA) expression and UGT1A protein synthesis. UGT1A transcriptional activation by dioxin, phenobarbital, and endotoxin was significantly reduced in SNP mice. Conclusion: Our data redefine the genetic basis behind GS. In vivo data studying the genotype present in 76% of GS individuals suggest that transcription and transcriptional activation of glucuronidation genes responsible for conjugation and detoxification is directly affected, leading to lower responsiveness. This study suggests that GS should be considered a potential risk factor for drug toxicity. (HEPATOLOGY 2012;55:1912–1921)

Gilbert syndrome (GS) is one of the most common genetically defined variants of bilirubin metabolism affecting about 10% of the white population.1 GS is viewed as a benign condition because it does not lead to liver inflammation, cellular destruction, fibrosis, or cirrhosis but is merely clinically characterized by intermittent episodes of uncomplicated unconjugated hyperbilirubinemia. GS is associated with the UDP-glucuronosyltransferase (UGT) 1A1*28 polymorphism, an insertion of an additional TA-repeat into the promoter region of the UGT1A1 gene resulting in a A(TA)7TAA promoter sequence that differs from the more prevalent A(TA)6TAA, which reduces bilirubin glucuronidation by 70%.2, 3 UGT1A1*28 has an allelic frequency of 40% in white individuals. Homozygous UGT1A1*28 is found in 16% of European, 12% of Indian, 8% of Egyptian, and 23% of African-American individuals.1

In GS, the phenotype of fluctuating levels of unconjugated hyperbilirubinemia is therefore the result of impaired glucuronidation and subsequent elimination of bilirubin. In humans UGT1A1 is the only physiological enzyme capable of forming water-soluble bilirubin glucuronides.4 Glucuronidation utilizes UDP-glucuronic acid as cosubstrate for the formation of hydrophilic glucuronides from nonmembrane-associated substrates. It represents an important specialized metabolic and detoxifying function in higher organisms. The responsible gene for bilirubin glucuronidation is encoded on chromosome 2 as part of the UGT1A gene complex encompassing (3′ to 5′) UGT1A1, UGT1A3, UGT1A4, UGT1A6, UGT1A7, UGT1A9, UGT1A10, and UGT1A8.1 The multitude of genes at this locus leads to a large potential spectrum of endo- and xenobiotic substrates that can be targeted for detoxification and elimination by glucuronidation.

However, this at first glance benign variant affecting glucuronidation activity has been associated with metabolic risks. UGT1A1*28 has been identified as a risk marker for irinotecan toxicity because UGT1A1 contributes to the detoxification of the active irinotecan metabolite SN-38, which is over 100-fold more active than irinotecan and can induce intestinal and bone marrow toxicity.5-7 Apart from bilirubin and SN-38 glucuronidation, UGT1A1 is capable of detoxifying many other endogenous and exogenous substrates.8 These include 2-hydroxy-estrone and estradiol, and a number of therapeutic drugs such as ethinylestradiol, gemfibrozil, simvastatin, and buprenorphine. Also, mutagenic xenobiotics such as N-hydroxy-PhIP and benzo(a)pyrenes undergo conjugation and detoxification by UGT1A1.9 Moreover, not only UGT1A1*28 alone but its combination with genetic variants of the UGT1A3 and UGT1A7 genes, which by themselves lead to altered gene expression, transcriptional regulation, and protein function10-13 have been studied in vitro. Combined variants of UGT1A1, UGT1A3, and UGT1A7 have been associated with hyperbilirubinemia in atazanavir and indinavir therapy.14, 15 GS may therefore exert a complex effect on detoxification.

The aim of this study was therefore to provide evidence for the complexity of variations affecting glucuronidation as well as transport genes (ABCC2, SCLO1B1) that are associated with GS. This study shows that 76% of 300 GS individuals homozygous for UGT1A1*28 simultaneously carried homozygous single nucleotide polymorphisms (SNPs) in four UGT1A genes (UGT1A1, UGT1A3, UGT1A6, UGT1A7). To examine the functional relevance of this finding in vivo, a transgenic humanized mouse model was created carrying SNPs of these UGT1A genes. The GS-associated variant haplotype reduced UGT1A transcript and protein expression in addition to reducing transcriptional activation of UGT1A gene expression by dioxin, phenobarbital, and lipopolysaccharide (LPS), suggesting that GS represents a complex condition relevant for glucuronidation and detoxification capability.


AhR, arylhydrocarbon receptor; ARE, antioxidant response element; HBDs, healthy blood donors; LPS, lipopolysaccharide; Nrf2, nuclear factor erythroid related factor 2; PCR, polymerase chain reaction; SNP, single nucleotide polymorphism; TCDD, 2,3,7,8-tetrachlordibenzo-p-dioxin; UGT, UDP-glucuronosyltransferase; WT, wildtype; XRE, xenobiotic response element.

Materials and Methods


Genomic DNA was isolated from whole blood of 300 homozygous UGT1A1*28 GS individuals and 249 healthy blood donors,16 all of white northern European descent. The Ethics Committee of Hannover Medical School approved the genotyping.


Genomic DNA (10 ng) was used as a template in TaqMan 5′-nuclease assays. Primers and probes specific for each SNP were designed with Primer Express software (Applied Biosystems, Darmstadt, Germany) and labeled with either 6-FAM or VIC as reporter dyes and MGB-NFQ (Applied Biosystems) as a quencher as described13 (Table 1). TaqMan assays were performed using qPCR MasterMix Plus (Eurogentec, Cologne, Germany) and probes from Applied Biosystems. Briefly, a run involved a hot start Taq at 95°C for 10 minutes, and 35 cycles of 94°C for 15 seconds, and 61°C for 1 minute performed in 25 μL reactions (96-well trays) using an ABI 7000 instrument (Applied Biosystems).

Table 1. Sequences of Primers and Probes Used for UGT1A Genotyping
GenePrimer and Probes

Variants of the transporter genes ABCC2 and SCLO1B1 were determined using specific TaqMan Drug Metabolism Genotyping Assays purchased from Applied Biosystems (ABCC2-24UTR: C_2814642_10; ABCC2 I1324I: C_11214910_20; SLCO1B1 N130D: C_1901697_20 and SLCO1B1 V174A: C_30633906_10) following the manufacturer's instructions.

Development of a Humanized Transgenic UGT1A Mouse Model

Bacterial Artificial Chromosome (BAC) Clones.

Sequence comparisons were performed in commercially available BAC clones (imaGenes, Berlin, Germany) with the Ensembl Genome Browser ( DNA was amplified by polymerase chain reaction (PCR) using BioTherm Taq DNA polymerase (Genecraft, Cologne, Germany) and gene specific primers (Eurofins MWG, Ebersberg, Germany) prior to sequencing (ABI Prism 300 automated sequencer, Applied Biosystems). A 13-kb 5′ fragment was removed from clone RP11-943B10 by Red/ET recombination using the Counter Selection BAC Modification Kit (Gene Bridges, Dresden, Germany) according to the manufacturer's instructions, resulting in a linear DNA fragment. This contained a kanamycin resistance and streptomycin sensitivity cassette flanked by 50 nucleotides homologous to DNA sequences flanking the region of the UGT1A gene locus to be deleted. Primers used for amplification of the selection cassette were: 5′-CT TTT ATA TGA AAC ACG GAC ACG AAA GGT AAC TCG TAC CCG CTT TTG GTA CTT AAG GGC CTA CAC GGT ATC TTT GTA TTG TAA ATT GTT ACC ACC TTT GGC CCA TGT CCC TTC-3′ (forward) and 5′-GA AGG GAC ATG GGC CAA AGG TGG TAA CAA TTT ACA ATA CAA AGA TAC CGT GTA GGC CCT TAA GTA CCA AAA GCG GGT ACG AGT TAC CTT TCG TGT CCG TGT TTC ATA TAA AAG-3′ (reverse). The BAC clones were then propagated in E. coli bacteria and recombination was verified by PCR and direct sequencing. Highly purified DNA for pronuclear injection was isolated using the NucleoBond BAC 100 kit (Macherey-Nagel, Düren, Germany). Fifty μg of DNA were linearized by digestion with the restriction enzyme NotI (New England Biolabs, Frankfurt am Main, Germany). DNA fragments were purified on a sepharose CL4b (Sigma-Aldrich, Taufkirchen, Germany) column. Transgenic founder animals were generated by pronuclear injection of linearized DNA fragments into fertilized mouse oocytes (strain C57BL/6).

Humanized Transgenic UGT1A Mice (htgUGT1A wildtype [WT] and htgUGT1A SNP).

DNA was isolated from mouse tails using the DirectPCR Lysis Reagent (Viagen, Los Angeles, CA) and Proteinase K (Qiagen, Hilden, Germany) according to the manufacturers' instructions. Presence of the human transgene was verified by PCR. A 3′, middle and a 5′ portion were chosen as templates to ensure that only animals with integration of the entire construct were selected for breeding. Gene copy numbers were analyzed using DNA from mouse tails (DNeasy Blood & Tissue Kit, Qiagen, according the manufacturer's instructions) by real-time quantitative PCR (RT-qPCR). All reactions were carried out as duplex reactions in an ABI Prism 7000 Sequence Detection System (Applied Biosystems) using qPCR MasterMix (Eurogentec). Probes and primers were specific for human UGT1A7 and mouse beta-actin. Ct-values were normalized against genomic DNA of nontransgenic C57BL/6 mice mixed with known amounts of BAC DNA.

Chromosomes were isolated from cycling bone marrow cells of transgenic mice cultured overnight at 37°C in RPMI-1640 medium supplemented with 10% fetal calf serum and phytohemagglutinin (PHA). One hour prior to isolation, cells were arrested in metaphase by 0.5 μg/mL colcemid (Gibco, Eggenstein, Germany). Cells were pelleted by centrifugation and incubated for 30 minutes in hypotonic buffer (75 mM KCl). After fixation in a 3:1 mixture of methanol and acetic acid, cells were concentrated by centrifugation and chromosomal preparations were spread on tissue slides. A specific probe for fluorescence in situ hybridization (FISH) was created by amplification and subcloning of a 2,050 basepair (bp) fragment of the human UGT1A gene locus. Human UGT1A4 exon1 and a 1,150 bp intronic region were amplified from BAC clones with gene-specific primers carrying restriction sites for XhoI (5′-CA GGA GAC CGG TAG AGT CGT CTT CTG AGC TCG TCG-3′) or HindIII (5′-G GTA ATT CCC TCG GTA GTT CGA AGG ACC-3′), respectively. The DNA fragment was then cloned into pcDNA3.1(-). BAC clone DNA was directly labeled with spectrum orange (Vysis, Downer's Grove, IL) by nick translation according to standard procedures. FISH was performed on metaphase chromosomes according to standard procedures. Metaphases were analyzed on an epifluorescence microscope (Zeiss, Oberkochen, Germany) with suitable filter combination and the ISIS image analysis system (MetaSystems, Altlussheim, Germany).

Transgenic UGT1A mice were kept at the Central Animal Facility (Ztm), Hannover Medical School, and bred to nontransgenic C57BL/6 mice (Charles River, Sulzfeld, Germany). All experiments were approved by the local Institutional Animal Care and Research Advisory Committee and authorized by the local state government of Lower Saxony, Germany. For all experiments, animals were sacrificed at 8-12 weeks of age.

Tissue Isolation from Transgenic Mice.

Tissue from different organs was isolated at 4°C, snap-frozen in liquid nitrogen, and stored at −80°C. For tissue slides, tissue was incubated in 4% paraformaldehyde (PFA) at 4°C overnight and transferred to phosphate-buffered saline (PBS) with 0.1% sodium azide for short-term storage. Tissue was embedded in paraffin and stored at 4°C. Tissue was cut at 2 μm and stained with hematoxylin and eosin (H&E).

Gene Expression Analysis.

RNA was isolated from tissue of sacrificed transgenic and nontransgenic (C57BL/6) mice using Trizol (Invitrogen, Karlsruhe, Germany) according to the manufacturer's instructions. For further analyses, equal amounts of RNA from four animals per group were pooled. Five μg of RNA was incubated with DNase I (Invitrogen) at room temperature for 15 minutes, followed by inactivation at 65°C for 10 minutes. DNase I-treated RNA was then used as template in oligo (dT)-primed complementary DNA (cDNA) synthesis reaction using the SuperScript III First-Strand Synthesis System (Invitrogen). For gene expression analysis, cDNA concentrations were determined by qPCR relative to mouse beta-actin. qPCR reactions with gene-specific primers and probes were performed in an ABI Prism 7000 Sequence Detection System (Applied Biosystems) using qPCR MasterMix (Eurogentec). Primer and probe sequences are available upon request. Calibration curves were determined in triplicate in three independent runs with correlation coefficients greater than 0.995. For gene expression analysis, all reactions were performed at least in duplicate. Expression relative to mouse beta-actin was calculated using Q-Gene.

Western Blots.

For western blot analyses, microsomal protein was isolated from 100 mg of liver tissue as described.17 Western blots were performed as described18 using different antibodies (UGT1Aall:18; UGT1A3: H00054659-M02, Abnova, Walnut, CA; UGT1A4: AV46829, Sigma-Aldrich; UGT1A6: 458416, BD Biosciences, San Jose, CA).

In Vivo Induction Experiments.

HtgUGT1A SNP and WT mice carrying five and six copies of the transgene, and nontransgenic C57BL/6 mice were treated with phenobarbital (Desitin, Hamburg, Germany), 2,3,7,8-tetrachlordibenzo-p-dioxin (TCDD) (Wellington Laboratories, Berlin, Germany), or solvent. Each substance was administered by intraperitoneal injection for 3 days. Phenobarbital and LPS were dissolved in a mixture of 50% PBS and 50% polyethylene glycol (PEG), whereas TCDD was dissolved in 10% dimethyl sulfoxide (DMSO) and 90% corn oil. Injection was at a volume of 10 mL/kg. Animals were sacrificed at day 4 and organs were collected as described above. For treatment with LPS (purified from E. coli 055:B5) (Sigma-Aldrich), animals received a single dose intraperitoneally and were sacrificed at different timepoints after intraperitoneal injection (24, 48, and 72 hours). Per substance, four male mice with htgUGT1A WT or htgUGT1A SNP genotype were treated. For further analyses, RNA or protein of all four animals was pooled.

Statistical Analysis.

Statistical analyses were performed using two-tailed Student's t test in groups of animals where more than one pool was available. For comparison of a single pool of treated animals to untreated mice, the 95% confidence interval of all pools of the reference group was calculated. Differences in gene expression levels were considered statistically significant if gene expression levels in treated mice were lower or higher than the 95% confidence interval.


UGT1A Variants in Homozygous GS Individuals and Healthy Blood Donors.

In a cohort of 300 individuals homozygous for UGT1A1*28, genotyping was performed for UGT1A3-66T>C, UGT1A6 p.S7A, UGT1A6 p.T181A, UGT1A6 p.R184S, UGT1A7 p.N129K/p.R131K, UGT1A7-57T>G/p.W208R. The frequency of homozygous UGT1A3, UGT1A6, and UGT1A7 variants was 79%-91% (Fig. 1A). For the UGT1A6*2a genotype 78% were homozygous (combining UGT1A6 p.S7A, p.T181A, and p.R184S), and 77% for UGT1A7*3 (combining UGT1A7 p.N129K/p.R131K and p.W208R). When analyzed for the presence of all variants, 76% of GS individuals were homozygous for all of these SNPs (Fig. 1B). Genotyping of 249 controls (healthy blood donors; HBD) expectedly showed lower SNP frequencies; 12% were homozygous for UGT1A1*28. In HBD the homozygous UGT1A1-UGT1A3-UGT1A6-UGT1A7 SNP haplotype was present in 9% versus 76% in GS individuals (Fig. 1B). To control for an influence of selection the 300 homozygous UGT1A1*28 GS individuals were compared with the homozygous UGT1A1*28 individuals identified from the HBD, showing no differences (Fig. 1C,D). In 127 GS individuals total serum bilirubin measurements at the time of clinical presentation were available (mean 83,66 μmol/L). In individuals carrying the UGT1A SNP haplotype (n = 106), mean total serum bilirubin (84.51 μmol/L) was higher than in those 21 individuals with UGT1A1*28 alone (38.5 μmol/L). This was not statistically significant because of high standard deviations (Fig. 1E).

Figure 1.

(A) Genotyping results of 300 homozygous (UGT1A1*28) Gilbert syndrome (GS) individuals and 249 healthy blood donors (HBD) showing the percentage of homozygous individuals for single nucleotide polymorphisms (SNPs) of four UGT1A genes in addition to the ABCC2-24UTR and p.I1324I, and the SLCO1B1 p.N130D and p.V174A SNPs. In GS individuals homozygous UGT1A SNPs are significantly (***P < 0,001) more frequent than in HBD, whereas there is no difference (n.s.; not significant) regarding the tested ABCC2 and SLCO1B1 variants. In HBD, 12% are homozygous for UGT1A1*28. In GS individuals 79%-91% exhibit other homozygous UGT1A variants. (B) Presentation of the genotyping data according to the defined genotypes UGT1A6*2a and UGT1A7*3. When combined, 9% of HBD and 76% of GS individuals exhibit all homozygous UGT1A SNPs tested involving the UGT1A1, UGT1A3, UGT1A6, and UGT1A7 genes. (C) Those individuals among the HBD found to be homozygous for UGT1A1*28 in (A) were compared with the GS cohort, which revealed no differences in frequency of homozygous carriers of the UGT1A3, UGT1A6, and UGT1A7 as well as the ABCC2 and OATP1B1 variants, indicating that prior diagnosis of GS leading to the GS cohort does not bias the distribution of genotypes. (D) Presentation of the comparison in (C) by predefined genotype shows no differences between prediagnosed GS individuals and those identified within the HBD group. (E) Total serum bilirubin levels in 127 GS individuals were found to be higher in those 106 with the identified UGT1A haplotype compared with those 21 with UGT1A1*28 alone (comparison not significant; bars indicate standard deviation of the mean).

SCLO1B1 and ABCC2 Variants in Homozygous GS Individuals and HBDs.

To expand the genotyping analysis, variants of hepatic uptake (SLCO1B1) or conjugate export (ABCC2) were determined. In contrast to the UGT1A variants ABCC2-24UTR and p.I1324I, as well as SLCO1B1 p.N130D and p.V174A were not found to be more prevalent in GS but were detected at expected frequencies both in HBD and GS (Fig. 1A,C).

Development of a Humanized (h) Transgenic (tg)UGT1A WT and htgUGT1A SNP Mouse Model.

A gene fragment from human chromosome 2 spanning UGT1A1 up to a portion of UGT1A10, which encodes the most prevalent (wildtype, WT) sequence and a fragment of equal length containing all aforementioned SNPs were inserted into mice (Fig. 2A). The UGT1A transgene was found to be integrated into a single site demonstrated by FISH using a human-specific probe, which was absent in WT C57BL/6 mice (Fig. 2B). Histologies of the liver and other organs (not shown) of these mice showed no abnormalities (Fig. 2B).

Figure 2.

(A) BAC clones were adjusted in length by deletion of a 13-kb fragment from the “wild type” sequence. The cartoon displays the region of the human UGT1A gene locus in both BAC clones. UGT1A polymorphisms in the UGT1A SNP construct are indicated at the bottom (not drawn to scale). (B) FISH demonstrated binding of a human UGT1A specific probe in htgUGT1A WT and htgUGT1A SNP mice but not in nontransgenic C57BL/6 mice. Binding was observed in a single region in metaphase chromosomes. No cells with two or multiple probe signals were detected. H&E-stained liver tissue slides from male htgUGT1A WT and htgUGT1A SNP mice showed regular liver architecture and normal cell morphology. All other organs analyzed were of normal appearance upon histological examination (data not shown). (C) Human UGT1A mRNA expression in male htgUGT1A WT mice is demonstrated. Primers and probe designated “UGT1A all” bind in the common exon 2-5 region shared by all UGT1A isoforms and therefore demonstrate overall UGT1A expression. All other results are UGT1A isoform specific. Human UGT1A expression is demonstrated in different tissues of the established mouse model. (D) Expression of human UGT1A mRNA relative to mouse beta-actin in male tgUGT1A WT mice using three independent pools of four animals each are demonstrated. Expression of total human UGT1A (UGT1Aall) relative to mouse beta-actin in the liver, small, and large intestine of tgUGT1A WT (□) and tgUGT1A SNP mice (▪) show reduced expression in the htgUGT1A SNP mouse model. The specific analysis of UGT1A1 transcription in liver shows a drastic reduction in htgUGT1A SNP mice. Bars represent mean mRNA levels of six pools for each genotype. *P < 0.05; **P < 0.01.

UGT1A Transcript Expression in htgUGT1A WT and htgUGT1A SNP Mice.

First, organ-specific UGT1A expression was analyzed in male htgUGT1A WT mice. RNA expression analyses were performed by real-time PCR and confirmed the tissue-specific expression of the human UGT1A1, UGT1A3, UGT1A4, UGT1A5, UGT1A6, UGT1A7, and UGT1A9 genes as well as overall human UGT1A expression using a common exon 2-5 binding primer in htgUGT1A WT mouse liver and in other organs (Fig. 2C). This pattern is similar to that observed in humans.19

Second, the influence of SNPs was analyzed by determining an effect of SNPs on overall UGT1A (using a primer binding to the common UGT1A exon 2-5 region) messenger RNA (mRNA) expression (Fig. 2D). UGT1A transcription was reduced in SNP versus WT mice in liver, small intestine, and large intestine. UGT1A1 expression in liver was drastically reduced in htgUGT1A SNP animals compared with their WT counterparts. The reduction of UGT1A expression in htgUGT1A SNP mice was isoform and tissue-dependent. Comparable mRNA expression levels between WT and SNP mice for some isoforms in other tissues (data not shown) indicate that differential UGT1A gene expression in both genotypes is likely a result of tissue-specific transcription factors and not related to the integration site of the transgene.

UGT1A Protein Expression in htgUGT1A WT and htgUGT1A SNP Mice.

Overall UGT1A protein expression (antibody directed against exon 2 common sequence of all UGT1A18) and UGT1A3, UGT1A4, and UGT1A6 protein was studied in mice by western blot. Although some crossreactivity of the anti-UGT1A and anti-UGT1A6 antibodies with mouse protein was observed, UGT1A protein expression in the liver of htgUGT1A SNP animals was clearly reduced in comparison with their WT counterparts (Supporting Fig. 1). In combination (Fig. 2), these data establish a humanized transgenic mouse model that allows for a comparison of UGT1A WT gene and SNP variant effects.

Transcriptional Activation in Response to Inducers.

HtgUGT1A SNP and WT animals were treated intraperitoneally with TCDD, phenobarbital, and LPS to examine potential differential responses of SNP haplotype carriers and WT mice. UGT1A mRNA expression was assessed by TaqMan PCR using isoform-specific probes.


TCDD is a potent inducer of UGT1A transcription, which is seen exemplified for UGT1A1, UGT1A3, and UGT1A6 in the liver, the jejunum (exception UGT1A6), and the colon of htgUGT1A WT mice. In htgUGT1A SNP mice TCDD-mediated activation was considerably lower. The UGT1A SNP haplotype reduced TCDD-mediated transcriptional activation of UGT1A genes (Fig. 3A).

Figure 3.

(A) Differences of UGT1A expression detected by TaqMan PCR in htgUGT1A WT (WT) and htgUGT1A SNP (SNP) mice in response to treatment with 2,3,7,8-tetrachlordibenzo-p-dioxin (TCDD). Shown are UGT1A1, UGT1A3, and UGT1A3 in liver, small, and large intestine. With the exception of UGT1A6 mRNA in jejunum, activation by TCDD is lower in SNP mice. (B) The response of UGT1A1, UGT1A3, and UGT1A6 expression to phenobarbital in liver is considerably reduced in the liver of htgUGT1A SNP mice. White and gray bars represent untreated or vehicle-injected mice. (C) Differences in hepatic UGT1A expression between htgUGT1A WT (WT) and htgUGT1A SNP (SNP) mice 24 hours after endotoxin injection. Expression of hepatic isoforms UGT1A3, UGT1A6 was reduced and lower in SNP mice. The activation observed for UGT1A1 was not observed in the SNP mice. White bars, untreated animals (three pools each group); dotted bars, mice injected with vehicle (PEG/PBS); gray and black bars, LPS-treated animals, three different timepoints. UGT1A expression levels were determined by analysis of pooled RNA of four animals and are shown relative to mouse beta-actin. The response to LPS is altered in the presence of UGT1A SNPs. In (A-C) the confidence intervals were calculated for the untreated mouse pool data. When data obtained after induction was outside this range significant induction was present (*P < 0.05).


Similar to the effects observed with TCDD, hepatic activation of overall UGT1A as well as UGT1A1, UGT1A3, and UGT1A6 expression in response to phenobarbital was inhibited in htgUGT1A SNP mice (Fig. 3B).


Induction by endotoxin (LPS) was studied at three timepoints showing an early activation of UGT1A transcription after 24 hours in the liver, which was also observed for the induction of UGT1A1 in liver. In htgUGT1A SNP mice this effect of LPS was not observed, indicating that the UGT1A haplotype altered the UGT1A transcriptional activation in response to endotoxin (Fig. 3C).

Serum bilirubin levels (mean ± standard deviation of the mean) were measured in the treated mice and were higher in htgUGT1A SNP than in WT animals but remained within normal limits: TCDD, 0.35 ± 0.07 μM/L (WT) and 2.45 ± 0.07 μM/L (SNP); phenobarbital, 1.65 ± 0.07 μM/L (WT) and 2.5 ± 0.28 μM/L (SNP); LPS, 1.4 ± 0.42 μM/L (WT) and 2.85 ± 0.5 μM/L (SNP).

In combination, these three examples indicate that in the established in vivo model the UGT1A haplotype observed in 76% of GS individuals led to profound alterations not only of human UGT1A expression but also transcriptional regulation and activation by inducers of transcription.


In this study we performed a detailed analysis of UGT1A gene locus variability in a cohort of 300 homozygous UGT1A1*28 individuals in order to elucidate the potential impact of this variant affecting approximately 10% of the white population. Based on earlier studies that suggest the existence of more extensive haplotypes of UGT1A gene variants7, 15, 20 genotyping included the UGT1A3, UGT1A6, and UGT1A7 genes apart from UGT1A1*28. These UGT1A isoforms are important proteins for detoxification. Substrates of UGT1A3 include xenobiotics such as polyaromatic hydrocarbons and hydroxylated benzo(a)pyrenes, amines, nonsteroidal antiinflammatory drugs such as ibuprofen, flurbiprofen, and ketoprofen, statins, flavonoids, bile acids, estrogens, vitamin D derivatives, and ezetimibe (reviewed19). Substrates of UGT1A6 and UGT1A7 include xenobiotics, acetaminophen, coumarin derivatives, and irinotecan metabolites (reviewed19). Moreover, the in vitro characterization of SNPs of many UGT1A genes has shown that, similar to UGT1A1*28,2 SNPs located within the promoter region of UGT1A3 (UGT1A3-66T>C)10 and UGT1A7 (UGT1A7-57T>G)13 alter transcriptional regulation and activation of these genes. Variants such as UGT1A7 p.N129K/p.R131K and p.W208R lead to catalytically less active UGT1A proteins.16, 21 In view of the individual catalytic activities, combinations of UGT1A SNPs carry potential relevance for detoxification, a feature that may be closely associated with the commonly observed GS.

A cohort of 300 homozygous UGT1A1*28 individuals that would roughly represent a population of 3,000 individuals was analyzed for the UGT1A3-66T>C, UGT1A6 p.S7A, UGT1A6 p.T181A, UGT1A6 p.R184S, UGT1A7 p.N129K/p.R131K, UGT1A7-57T>G/p.W208R variants. Genotyping of the UGT1A4, UGT1A9 and UGT1A10 genes was not undertaken because of low frequencies of these SNPs in the population.22 The frequency of homozygous UGT1A3, UGT1A6, and UGT1A7 variants was observed to be 79%-91%, indicating that the vast majority of GS individuals also carry variants of these three UGT1A genes. In all, 76% of GS individuals were homozygous for all SNPs, a haplotype found in 9% of HBD. This finding reveals a surprisingly high degree of genetic variability encountered in GS individuals. To expand on this analysis, transporter genes with relevance for bilirubin and conjugate trafficking were also analyzed. Previous publications have hypothesized GS to be associated with variations of transport (reviewed1). We therefore studied variants of uptake (SLCO1B1) or conjugate export (ABCC2) in GS.23 SCLO1B1 variants have been associated with statin-induced myopathy,24, 25 and statins are also substrates for glucuronidation by UGT1A1 and UGT1A3.26 However, neither ABCC2-24UTR and p.I1324I, nor SLCO1B1 p.N130D and p.V174A25 were more prevalent in GS but were observed in expected frequencies both in HBD and GS. Based on these findings a complex array of variants of at least four UGT1A genes is present in a vast majority of GS individuals. When total serum bilirubin levels of 127 of the genotyped 300 GS individuals were correlated with genotype there was a trend toward higher levels in those individuals carrying the identified UGT1A haplotype compared with those with UGT1A1*28 alone (Fig. 1E). This may indicate that genotyping of more than UGT1A1*28 can identify individuals at risk for higher levels of total serum bilirubin. In addition, further studies are needed to show whether the determination of additional UGT1A SNPs may be of relevance for the prediction of toxicity associated with UGT1A1*28 such as in irinotecan therapy,7 which would potentially change the perception of GS and the required clinical, chemical and genetic tests employed for its diagnosis.

Because the functional aspects of UGT1A variants have been studied in vitro, we then sought to study the potential impact of the GS-associated SNP haplotype in vivo. We created a humanized transgenic UGT1A mouse model containing a gene fragment from human chromosome 2 spanning UGT1A1 up to a portion of UGT1A10, which encodes the most prevalent (WT) sequence and a fragment of equal length containing all aforementioned SNPs. The UGT1A transgene was integrated into a single site, and RNA expression analyses confirmed the tissue-specific expression of the human UGT1A genes in liver and in other organs similar to the human pattern.19 In htgUGT1A SNP mice, overall expression of UGT1A transcripts was significantly reduced in liver, small and large intestine in comparison to the htgUGT1A WT mice. In the liver, UGT1A1 was significantly reduced in htgUGT1A SNP mice. This was also confirmed at the protein level. The UGT1A haplotype found in 76% of GS individuals therefore reduced transcription and protein expression of different UGT1A genes in different tissues in vivo.

It is important to also consider the impact this may have on the response to environmental stimuli acting to influence the transcriptional regulation of detoxification by glucuronidation. We therefore studied whether the observed differences between htgUGT1A WT and SNP mice would also affect UGT1A regulation in response to typical activators of transcription. In a first series of experiments mice were treated with TCDD. TCDD was chosen because TCDD-inducibility has been characterized for several UGT1A genes including UGT1A1, UGT1A3, and UGT1A7.11, 27, 28 TCDD activation of UGT1A genes has been demonstrated to proceed by way of cis-acting DNA xenobiotic response elements (XRE) involving the arylhydrocarbon receptor (AhR)/AhR nuclear translocator (ARNT) pathway. This represents a classical mechanism for xenobiotic activation of drug metabolism genes. Previous in vitro studies have shown that individual SNPs can affect XRE-mediated signal transduction and transcriptional activation.11, 28 In htgUGT1A SNP mice transcriptional activation of UGT1A1, UGT1A3, and UGT1A6 by TCDD in the liver and the gastrointestinal tract was indeed considerably reduced compared with the WT counterparts. The presence of the GS UGT1A SNP haplotype therefore not only down-regulated basal UGT1A expression but also significantly reduced responsiveness to xenobiotic (TCDD) exposure. Inducibility was also assessed with phenobarbital, used to induce UGT activity, i.e., in patients with Crigler Najjar type 2 syndrome.29, 30 Similarly, activation of hepatic UGT1A transcription by phenobarbital was inhibited in htgUGT1A SNP versus htgUGT1A WT mice. This may indicate that signaling by way of the constitutive androsterone receptor (CAR) is likely to be affected by genetic variation at the UGT1A locus.31 Finally, the response to Gram-negative sepsis was studied (LPS). UGT1A genes were shown to be regulated by nuclear factor erythroid related factor 2 (Nrf2) as a result of oxidative stress.11 LPS treatment was chosen to test the activation of UGT1A transcription as an indirect antioxidant response in a situation of inflammation and potentially associated oxidative stress. LPS-treated mice showed a time-dependent response with an ultimate decrease of transcription. In the liver, UGT1A1 induction was observed at 24 hours, which was not seen in the htgUGT1A SNP mice. All three examples show an effect on UGT1A gene activation by the presence of SNPs in vivo.

In combination, our in vivo data suggest that a haplotype of multiple genetic variants at the UGT1A gene locus, which is present in a vast majority of GS individuals, profoundly alters the expression and the responsiveness of the human UGT1A genes. Our data therefore redefine the molecular and functional basis of GS as a condition with a complex genotype that may be capable of significantly altering metabolism, and may thus represent a risk factor for drug therapy and potentially in infection.