Oligomerization of the UDP-glucuronosyltransferase 1A Proteins

UDP-glucuronosyltransferases (UGTs) are membrane-bound proteins localized to the endoplasmic reticulum and catalyze the formation of β-d-glucopyranosiduronic acids (glucuronides) using UDP-glucuronic acid and acceptor substrates such as drugs, steroids, bile acids, xenobiotics, and dietary nutrients. Recent biochemical evidence indicates that the UGT proteins may oligomerize in the membrane, but conclusive evidence is still lacking. In the present study, we have used fluorescence resonance energy transfer (FRET) to study UGT1A oligomerization in live cells. This technique demonstrated that UGT1A1, UGT1A3, UGT1A4, UGT1A6, UGT1A7, UGT1A8, UGT1A9, and UGT1A10 self-oligomerize (homodimerize). Heterodimer interactions were also explored, and it was determined that UGT1A1 was capable of binding with UGT1A3, UGT1A4, UGT1A6, UGT1A7, UGT1A8, UGT1A9, and UGT1A10. In addition to the in vivo FRET analysis, UGT1A protein-protein interactions were demonstrated through co-immunoprecipitation experiments. Co-expression of hemagglutinin-tagged and cyan fluorescent protein-tagged UGT1A proteins, followed by immunoprecipitation with anti-hemagglutinin beads, illustrated the potential of each UGT1A protein to homodimerize. Co-immunoprecipitation results also confirmed that UGT1A1 was capable of forming heterodimer complexes with all of the UGT1A proteins, corroborating the FRET results in live cells. These preliminary studies suggest that the UGT1A family of proteins form oligomerized complexes in the membrane, a property that may influence function and substrate selectivity.

UDP-glucuronosyltransferases (UGTs) 2 are type I membrane-bound proteins localized to the endoplasmic reticulum (ER) and play an essential role in the elimination and detoxifi-cation of drugs, xenobiotics, and a host of endogenous compounds such as steroids, bile acids, bilirubin, and fatty acids (1). This process proceeds by the transfer of glucuronic acid from UDP-glucuronic acid to targeted substrates. The addition of the polar glucuronic acid to the lipophilic substrates increases the metabolite's water solubility, thereby facilitating excretion into water compartments of the cell for excretion into either the urine or bile.
The UGTs are classified into two gene families, UGT1 and UGT2 (2). The UGT2 subfamily of proteins is encoded by individual genes located in a gene cluster on chromosome 4 at 4q13-q21 (3)(4)(5). These are composed of the UGT2A1 and UGT2A2, along with the UGT2B members, which include UGT2B4, 2B7, 2B10, 2B11, 2B15, 2B17, and 2B28. The UGT2A proteins are localized to the olfactory system and are felt to be important in eliminating olfactory stimuli (6). The UGT2B proteins are involved in the metabolism of endogenous compounds such as bile acids (7), steroids (8), hormones (9), and fatty acids but are also important for the elimination of therapeutic drugs (10). In contrast, the UGT1 locus is located on chromosome 2q37 (11), and its gene products are linked to the metabolism of exogenous compounds, although UGT1A1 is the only UGT capable of conjugating the heme byproduct bilirubin (12). The UGT1 locus encodes nine functional UGT1A proteins (UGT1A1 and UGT1A3 through UGT1A10) (2,13), each derived through a unique process called exon sharing (14). The UGT1 locus is organized with a tandem array of 9 open reading frames encoding each of the exon 1 regions of the respective UGT1A proteins, flanked by open reading frames encoding exons 2-5. Based upon the start of transcription, each respective first exon sequence is processed with common exons 2-5, resulting in the synthesis of the 9 UGT1A gene products. Thus, each UGT1A protein is composed of a variable amino region (exon 1) and an identical carboxyl region (exons 2-5) (13,15). The carboxyl region of the UGT1A proteins displays high sequence identity to the same region on the UGT2 proteins, implicating the conserved function of this region for recognition and binding of UDP-glucuronic acid. Within the UGT2 family of proteins, both the amino-terminal and the carboxyl regions have been shown to play an important role in substrate specificity (16,17).
Preliminary findings suggest that the family of UGTs may be highly organized within the ER, existing as monomeric proteins as well as oligomers in homodimeric and heterodimeric states.
In rat liver microsomes, UGT2B1 was shown to exist as a functional dimer, with evidence that the amino-terminal region was important in the protein-protein interaction of the UGT (18). In addition, UGT2B1 was also shown to directly heterodimerize with the family of UGT1A proteins (19), although identification of the UGT1A proteins involved in the oligomerization was not characterized. Cross-linking studies using expressed human UGT1A1, in addition to a two-hybrid analysis, demonstrated that UGT1A1 formed a homodimer complex. It was also observed that the reduction of disulfide bond formation was shown to have little impact on UGT1A1 dimerization and mutagenesis of cysteine 223 to a tyrosine completely abolished UGT1A1 oligomerization, as did the deletion of a membraneembedded region spanning amino acids 152-180. Consistent with the UGT2B1 findings, the amino-terminal region was found to be important for UGT1A1 dimerization (20).
The focus of the current study is to explore the intermolecular interactions of the UGTs in living cells using fluorescence resonance energy transfer (FRET). FRET is a non-invasive spectroscopic method that can be used in live cells to monitor spatial and temporal intermolecular interactions of chimeric proteins that have acceptor and donor fluorophores attached (21). FRET involves the transfer of energy from a donor fluorophore in its excited state to an acceptor fluorophore by a nonradiative dipole-dipole interaction (22) and is only able to occur if the proteins reside within angstroms from each other. To further complement the in vivo analysis of UGT protein-protein interactions by FRET, co-immunoprecipitation experiments provided additional biochemical evidence to support the existence of the UGT1A proteins forming homodimeric or heterodimeric aggregates in the membrane.

EXPERIMENTAL PROCEDURES
Generation of the pUGT1A-CFP, pUGT1A-YFP, and pUGT1A-HA Plasmids-For the following study monomeric cyan fluorescent protein (mCFP) and monomeric yellow fluorescent protein (mYFP) variants were chosen as the donor and acceptor fluorophores, respectively. Because of the intrinsic property of the wild type CFP and YFP proteins to spontaneously aggregate at high concentrations, we elected to use the monomeric forms of the fluorophores, which would prevent spontaneous aggregation of the fluorophores in the membrane (23). mCFP and mYFP plasmids were kindly provided by Dr. Roger Tsien (University of California, San Diego). Both mCFP and mYFP were amplified by PCR with the sense primer (5Ј-ggatccaatggtgagcaagggcgaggag-3Ј) containing a BamHI site (underlined) and the antisense primer (5Ј-gcggccgctttacttgtacagctcgtccatgcc-3Ј) containing a NotI site (underlined). The mCFP and mYFP PCR products were subcloned into pCR 2.1-TOPO vector (Invitrogen). The amplified TOPO clones containing mCFP and mYFP were digested with BamHI and NotI restriction enzymes and each insert purified. The pEGFP-N1 expression vector (Clontech), which contains the green fluorescent protein gene flanked by a multicloning site, was digested with BamHI and NotI to remove the green fluorescent protein portion. The mCFP and mYFP inserts were ligated into the linearized pEGFP-N1 expression vector and the resulting plasmids identified as mCFP-N1 and mYFP-N1. To make the chimeric UGT1A1-CFP and UGT1A1-YFP proteins, UGT1A1 was amplified by PCR with a sense primer (5Ј-ctcgagATGgctgtggagtcccag-3Ј) containing an XhoI site (underlined) followed by an initiation ATG codon and an antisense primer (5Ј-ggatccccaatgggtcttggatttg-3Ј) with a BamHI site (underlined) followed by a mutated stop codon shown in bold. After the cloning of UGT1A1 PCR product into the pCR 2.1-TOPO vector, the insert was removed, purified by digestion with XhoI and BamHI, and subcloned into the XhoI and BamHI sites in the mCFP-N1 and mYFP-N1 plasmids, creating the pUGT1A1-CFP and pUGT1A1-YFP plasmids. To generate the hemagglutinin (HA) recognition sequence fused to the UGT1A1 protein, a UGT1A1-HA fragment was generated by PCR with a sense primer (5Ј-ctcgagatggctgtggagtcccag-3Ј) containing an XhoI site (underlined) and antisense primer (5Ј-ggatcctcaggcataatctggcacatcataagggtattccatccaatgggtcttgggattt-3Ј) containing a BamHI site (underlined) and the HA sequence in bold. The UGT1A1-HA PCR product was subcloned into pCR 2.1-TOPO vector (Invitrogen) and digested with XhoI and BamHI restriction enzymes to remove the insert. The UGT1A1-HA fragment was subcloned into the XhoI and BamHI multiple cloning site of the pcDNA 3.1(-) expression vector to produce the pUGT1A1-HA plasmid.
Construction of UGT1A-CFP-, UGT1A-YFP-, and UGT1A-HA-tagged Expression Plasmids-To generate a library of the UGT1A-CFP-, UGT1A-YFP-, and UGT1A-HA-tagged constructs, we took advantage of the feature that all of the UGT1A proteins have a unique exon 1 region and a conserved 2-5 exon region. The pUGT1A1-CFP, pUGT1A1-YFP, and pUGT1A1-HA expression vectors were used to generate all subsequent constructs. This was done by removing the exon 1, exon 2, exon 3, and a portion of the exon 4 region of UGT1A1 from the CFP, YFP, and HA expression vectors and inserting in their place the accompanying regions of the other UGT1A cDNAs. The following procedures were performed for each UGT1A isoform. pUGT1A1-CFP, pUGT1A1-YFP, and pUGT1A1-HA plasmids were each digested with the restriction enzymes NheI (located in the multiple cloning site of each expression vector) and BstEII (located in exon 4 of each UGT1A isoform). Having constructed previously in our laboratory pcDNA 3.1 plasmids expressing UGT1A3, UGT1A4, UGT1A6, UGT1A7, UGT1A8, UGT1A9, and UGT1A10, each of the UGT1A inserts encoding these regions was isolated by digestion with NheI and BstEII. Each of the UGT1A fragments was subcloned into the linearized pUGT1A1-CFP, pUGT1A1-YFP, and pUGT1A1-HA expression vectors that had the NheI and BstEII fragment removed. In all, 24 UGT1A constructs were made with each expressing a fusion protein with either CFP, YFP, or HA carboxyl-terminal tags. A list of the plasmid constructs is shown in Table 1.
Microscopy and FRET Analysis-COS cells were plated at a density of 3.5 ϫ 10 5 cells/well in 6-well plates and co-transfected with 5 g of pUGT1A-CFP and 5 g of pUGT1A-YFP plasmids using Lipofectamine 2000 (Invitrogen) according to the manufacturer's protocol. After 24 h the live cells were imaged using a Zeiss Axiovert 200 M inverted microscope. Donor (CFP) was excited at 455 nm and fluorescence emission detected in a bandwidth of 440 -480 nm (CFP channel). Accep-tor (YFP) was excited at 515 nm and fluorescence emission detected in a bandwidth of 530 -535 nm (YFP channel). FRET excitation was conducted at 455 nm and fluorescence emission detected in a bandwidth of 530 -535 nm (FRET channel). To correct for fluorescence bleed through into the FRET channel, cells transfected with only pUGT1A-CFP were used to determine the Donor correction factor and cells transfected with only pUGT1A-YFP were used to determine the Acceptor correction factor. Images of UGT1A CFP and UGT1A YFP expression in co-transfected cells were sequentially acquired with the Donor (CFP) channel, Acceptor (YFP) channel, and FRET channel under identical conditions. The image obtained with the FRET channel was evaluated using Carl Zeiss AxioVision FRET software and FRET calculated using the Acceptor ratio correction method (24). The Acceptor ratio method corrects for the cross-talk from the Donor alone, the Acceptor alone, and background intensity using the following equation: Ac ϭ Fc/(accgv-bgacc), where Ac ϭ FRET concentration as Acceptor ratio and Fc ϭ FRET concentration according to Youvan et al. (24) calculated by the equation Fc ϭ (fretgv-bgfret) Ϫ cfdon*(dongv-bgdon) Ϫ cfacc* (accgc-bgacc), with gv ϭ intensity as gray value, bg ϭ background intensity, cf ϭ correction factor, fret ϭ FRET image, don ϭ Donor image, acc ϭ Acceptor image.
FRET efficiency was also analyzed by photobleaching the Acceptor (YFP) and measuring the increase in Donor signal (CFP). Cells co-transfected with pUGT1A1-CFP and pUGT1A1-YFP were initially imaged using the CFP and YFP channels. The same cell was then illuminated using the photobleaching cube (YFP excitation 515 nm) with the laser power set to 100% for 1-min intervals, followed by CFP and YFP image acquisitions. This was repeated every minute over a 10-min time course. To quantify changes in the CFP and YFP fluorescence intensity over the photobleaching time course, selected regions of interest were measured using the Carl Zeiss Axiovision FRET software and quantified using the Siegel evaluation method (25). The evaluation method uses the following equation: BL ϭ [1-(dongvbef-bgdon-bef)/(dongvaft-bgdon-aft)*100, where BL ϭ FRET concentration by Acceptor bleaching, Don ϭ Donor image, gv ϭ intensity as gray value, bef ϭ before bleaching, aft ϭ after bleaching.
Co-immunoprecipitation of Expressed UGT1A Proteins-COS cells were plated at density of 3.5 ϫ 10 5 cells/well in 6-well plates and co-transfected with 5 g of pUGT1A-HA and 5 g of pUGT1A-CFP plasmids using Lipofectamine 2000 (Invitrogen). After 24 h, cells were washed twice with phosphate-buff-ered saline and lysed for 1 h on ice with 1 ml of 1ϫ radioimmunoprecipitation buffer (0.05 M Tris-HCl, pH 7.4, 0.15 M NaCl, 0.25% deoxycholic acid, 1% Nonidet P-40, 1 mM EDTA) supplemented with 1% protease mixture inhibitor (Sigma). The lysed cells were centrifuged for 10 min at 13,000 ϫ g and the supernatant collected. The entire supernatant from one transfected well was mixed with 30 l of anti-HA beads (Roche Applied Science) and incubated at 4°C on a rocker platform overnight. The anti-HA beads were centrifuged for 1 min at 1000 ϫ g and the supernatant discarded. The anti-HA beads were washed three times with 1 ml of cold lysis buffer. The anti-HA beads were then resuspended in 60 l of electrophoresis sample buffer and heated to 95°C for 10 min. The beads were centrifuged for 1 min at 1,000 ϫ g and the supernatant collected and used for Western blot analysis.
Western Blot Analysis-All Western blots were performed using 4 -12% NuPAGE Bis Tris polyacrylamide gels as outlined by the supplier (Invitrogen) and resolved under denaturing conditions (50 mM MOPS, 50 mM Tris base, pH 7.7, 0.1% SDS, 1 mM EDTA). The resolved protein was transferred onto nitrocellulose membrane using a semi-dry transfer system (Norvex). The membrane was blocked with 5% nonfat dry milk in 10 mM Tris-HCl, pH 8, 0.15 M NaCl, and 0.05% Tween 20 (Tris-buffered saline) for 1 h at room temperature. Each membrane was washed in Tris-buffered saline solution and incubated with an anti-UGT1A antibody, kindly provided by Dr. Alain Belanger (Laval University, Quebec, Canada), in Tris-buffered saline containing 0.02% sodium azide at 4°C overnight. Membranes were then washed five times with Tris-buffered saline solution and incubated with horseradish peroxidase-conjugated rabbit secondary antibody at a 1:5000 dilution in Tris-buffered saline solution containing 2% nonfat milk for 1 h of shaking at room temperature. Membranes were washed five times with Trisbuffered saline solution and visualized using Western Lightning Chemiluminescence reagents according to the manufacturer's instructions (PerkinElmer Life Sciences) followed by exposure to x-ray film (Kodak).
Functional Analysis of UDP-glucuronosyltransferase Activity-COS cells were plated at a density of 3.5 ϫ 10 5 cells/well in 6-well plates and transfected with either 2 g of each plasmid (co-transfection) or 5 g of plasmid (single transfection) using Lipofectamine 2000 (Invitrogen). Whole cell lysates were prepared by resuspending cells in 50 mM Tris-HCl, pH 7.6, 10 mM MgCl 2 followed by homogenization and sonication. Each catalytic assay was conducted in a 100-l reaction that contained 350 g of cellular protein, 50 mM Tris-HCl, pH 7.6, 10 mM MgCl 2 , 0.04 Ci of [ 14 C]UDPGA (255.6 pmol/reaction), 8.5 mM saccharolactone, 100 g of phosphatidylcholine, 0.5 mM UDPglucuronic acid, and a final concentration of 500 M for each substrate. The substrate octylgallate was used to determine UGT activity for UGT1A1 and UGT1A3 isoforms, whereas 2-napthol was used for UGT1A7. Each reaction was incubated in a 37°C water bath for 60 min and then terminated with the addition of 100 l of cold methanol. The protein was spun down by centrifugation at 13,000 ϫ g for 20 min and 100 l of the supernatant spotted onto a Whatman glass-backed thin layer chromatography (TLC) plate. Chromatography was performed in a mixture (35:35:10:20 v/v) of n-butanol:acetone:ace-  FEBRUARY 16, 2007 • VOLUME 282 • NUMBER 7

UGT1A Proteins Oligomerize in Living Cells
tic acid:water. Once the solvent front reached 2 cm from the top of the TLC plate, the plate was removed from the tank, dried, and exposed to a phosphor screen (GE Healthcare) overnight.
The location of the glucuronides on the plate was identified, and that region of the TLC plate was scraped and placed in scintillation fluid for quantification with a liquid scintillation counter (Beckman Coulter LS6500, Fullerton, CA).

RESULTS
Determination of FRET Analysis with Expressed UGT1A CFP and UGT1A YFP -To explore intermolecular protein-protein interactions between UGT1A1 proteins in vivo, UGT1A1-CFPtagged and UGT1A1-YFP-tagged fusion constructs were made. Expression patterns of the UGT1A1 CFP and UGT1A1 YFP proteins were evaluated by transfecting pUGT1A1-CFP and pUGT1A1-YFP plasmids into COS cells, respectively, followed by fluorescence microscopy analysis. In Fig. 1, the top panel (1A1-CFP) and lower panel (1A1-YFP) of cells illustrate that both UGT1A1 CFP and UGT1A1 YFP fusion proteins exhibit fluorescence. Protein expression of each construct was excluded from the nucleus and resembles an ER localization expression pattern.
The application of FRET has been used to monitor a vast array of protein interactions (26,27) such as oligomerization and multimeric complex formation (28 -33). To determine homodimerization of UGT1A1 in vivo, COS cells were co-transfected with pUGT1A1-CFP and pUGT1A1-YFP plasmids and the live cells evaluated for FRET. In Fig. 2A, co-expression of both UGT1A1 CFP and UGT1A1 YFP is indicated in the CFP and YFP channels, respectively. Analysis with the FRET channel after correction for Donor and Acceptor bleed through illustrates increased fluorescence, indicating that the two proteins reside in close proximity to each other within the ER membrane. Evaluation of FRET using Acceptor Ratio analysis depicted distinct color-coded regions, illustrating FRET efficiency ranging from 40% (green) up to 80% (red) throughout the cell.
The acceptor photobleaching method was also implemented as an alternative method to quantify the absolute efficiency of FRET between the Donor (CFP) and Acceptor (YFP) fluorophores. For this experiment COS cells were co-transfected with pUGT1A1-CFP and pUGT1A1-YFP and the cells illuminated at 515 nm wavelength for 60 s, followed by CFP and YFP image acquisitions. This was repeated every minute over a 10-min time course. Using the acceptor photobleaching method, FRET is manifested by the dequenching of the Donor CFP signal by selective photobleaching of the Acceptor YFP over time, resulting in increased CFP signal. In Fig. 2B the Acceptor graph showed a sharp decrease in YFP fluorescence after each photobleaching exposure, whereas the Donor graph depicted a concomitant increase in CFP fluorescence throughout the photobleaching time course, indicating FRET between UGT1A1 CFP and UGT1A1 YFP . Moreover, the decrease in fluorescence gray value of the Acceptor (YFP) was directly proportional to the fluorescence gray value gained by the Donor (CFP), further confirming the high FRET efficiency observed in the Acceptor Ratio analysis in Fig. 2A. Because FRET can elucidate proteinprotein interactions within angstrom distances, these results infer that the UGT1A1 CFP and UGT1A1 YFP fusion proteins are in physical contact within the membrane.
FRET analysis suggests that UGT1A1 forms a homodimer complex. The UGT1A proteins share 100% amino acid identity in 244 amino acids at the carboxyl end with varying degrees of sequence divergence in the amino-terminal region. Hydropathy profiles indicate that each of the UGT1A proteins shows a conserved pattern of hydrophobicity with UGT1A1, allowing us to speculate that the other UGT1A proteins may behave in a similar pattern as UGT1A1 in the membrane. To examine this possibility, homodimerization among the other UGT1A proteins was explored. COS cells were co-transfected with the corresponding pUGT1A-CFP and pUGT1A-YFP plasmids to examine homodimerization potential as measured by FRET analysis (Fig. 3). As a negative control, COS cells were co-transfected with the pCFP and pYFP vectors. FRET analysis using Acceptor Ratio evaluation from pCFP/pYFP co-transfected cells showed no FRET signal as indicated by the dark blue colorcoded region throughout the cell. However, Acceptor Ratio evaluations of cells co-transfected with the pUGT1A3-, pUGT1A4-, pUGT1A6-, pUGT1A7-, pUGT1A8-, pUGT1A9-, or pUGT1A10-CFP and -YFP pairs showed high FRET signals ranging from 40% (green) to 90% (red-white) FRET efficiency. It can be seen in each of the FRET channels that all of the UGT1A-CFP/YFP combinations produced a FRET signal. No FRET was identified in select regions of the cell, as indicated by the black and dark blue regions, particularly the nucleus.
To examine the potential of UGT1A1 to dimerize with the other UGT1A proteins, co-transfection of COS cells with pUGT1A1-CFP and either pUGT1A3-YFP, pUGT1A4-YFP, pUGT1A6-YFP, pUGT1A7-YFP, pUGT1A8-YFP, pUGT1A9-YFP, or pUGT1A10-YFP was performed and evaluated for FRET. In Fig. 4, it is shown that Acceptor Ratio evaluation for each UGT1A1 CFP /UGT1A YFP pair resulted in a high FRET signal ranging from 40% (green) to 100% (white) FRET efficiency among the cells. This evidence suggests that UGT1A1 is able to heterodimerize with either UGT1A3, UGT1A4, UGT1A6, UGT1A7, UGT1A8, UGT1A9, or UGT1A10. Combined, these results suggest that each of the UGT1A proteins is capable of forming homodimers and heterodimers in the ER.
Co-immunoprecipitation Confirms the Existence of Homoand Heterodimerization of the UGT1A Proteins-In addition to the in vivo analysis of protein-protein interactions using FRET, co-immunoprecipitation experiments were conducted to examine the potential of the UGT1A proteins to form heterodimer complexes. To carry out these experiments, we first constructed expression plasmids of each of the UGT1A cDNAs that were aligned in-frame with a HA sequence. The HA-tagged expressed proteins can be recognized with mono-specific HA antibodies, and when these plasmids are expressed in COS cells their expressed UGT1A HA fusion products can be immunoprecipitated with anti-HA beads. The addition of the HA sequence tag (11 amino acids) at the carboxyl end of each of the UGT1A proteins does not add significantly to the molecular mass of the protein, and they can be resolved on SDS-polyacrylamide gels as ϳ56-kDa proteins. We rationalized that if the UGT1A HA proteins could be expressed in conjunction with UGT1A CFP proteins, dimerization of the proteins following immunoprecipitation could be identified by the resolution of the 86-kDa UGT1A CFP proteins and the 56-kDa UGT1A HA proteins.
To confirm homodimerization of UGT1A1, both pUGT1A1-HA and pUGT1A1-CFP were expressed individually (Fig. 5, lanes 1 and 2) or co-expressed (lane 3) and whole cell lysates immunoprecipitated with the anti-HA beads. Lane 1 illustrates that UGT1A1 HA was immunoprecipitated with the anti-HA beads, whereas UGT1A1 CFP was not immunoprecipitated (lane 2), demonstrating the specificity of the anti-HA beads for the HA-tagged protein. In lane 3, COS cells were co-transfected with pUGT1A1-HA and pUGT1A1-CFP followed by immunoprecipitation with anti-HA beads. Western blot analysis with the anti-UGT1A antibody illustrated the presence of two bands, one at ϳ56 kDa (UGT1A1 HA ) and the other at ϳ86 kDa (UGT1A1 CFP ), indicating that UGT1A1 CFP was co-immunoprecipitated with the anti-HA beads only in the presence of expressed UGT1A1 HA protein. In lane 4, individual transfections with pUGT1A1-HA and pUGT1A1-CFP constructs, followed by mixing the whole cell lysates prior to co-immunoprecipitation with anti-HA beads, did not result in the co-immunoprecipitation of UGT1A1 CFP . This result suggests that protein-protein interactions leading to dimerization may require coordinated synthesis of the UGTs in the cell.
Homodimerization of all the UGT1A proteins and heterodimerization of the UGT1A proteins with UGT1A1 were also evaluated by the co-immunoprecipitation technique. When each of the pUGT1A-HA and UGT1A-CFP plasmids was transfected individually, expressed UGT1A HA proteins FIGURE 2. Detection of FRET in COS cells co-transfected with pUGT1A1-CFP and pUGT1A1-YFP. A, COS cells were co-transfected with 5 g of pUGT1A1-CFP and pUGT1A1-YFP for 24 h. Co-expression of both UGT1A1 CFP and UGT1A1 YFP is indicated by fluorescence in both the CFP channel and YFP channel. Correction for Donor and Acceptor bleed through was performed, and FRET signal was analyzed with the FRET channel. FRET efficiency was evaluated with the Carl Zeiss AxioVision FRET software using the Acceptor Ratio analysis and depicted as a color-coded scale ranging from 0 to 100% FRET efficiency. B, COS cells were co-transfected with 5 g of pUGT1A1-CFP and pUGT1A1-YFP and initially imaged with the CFP and YFP channels. The same cell was then photobleached by excitation at 515 nm wavelength for 60-s intervals, followed by CFP and YFP image acquisitions. This was repeated every minute over a 10-min time course. Changes in the CFP and YFP fluorescence intensity over the photobleaching time course were quantified using the Siegel evaluation method in the Carl Zeiss AxioVision FRET software. FEBRUARY 16, 2007 • VOLUME 282 • NUMBER 7

UGT1A Proteins Oligomerize in Living Cells
were identified from whole cell lysates following immunoprecipitation with anti-HA beads, whereas the anti-HA beads did not immunoprecipitate any of the expressed UGT1A CFP proteins (Fig. 6). This result confirms that the anti-HA beads will identify only UGT1A HA -expressed proteins. Homodimerization of the UGT1A proteins was tested by co-transfecting pUGT1A-CFP with the corresponding pUGT1A-HA constructs followed by co-immunoprecipitation with anti-HA beads. As shown in Fig. 7A, Western blot analysis revealed two distinct bands indicating the presence of the individual UGT1A HA fusion proteins (lower band) as well as the UGT1A CFP fusion proteins (upper band). These data indicate that each of the UGT1A proteins is able to homodimerize. Heterodimerization analysis between UGT1A1 and each of the UGT1A proteins was evaluated by co-transfecting COS cells with pUGT1A1-HA and each pUGT1A-CFP plasmid, followed by co-immunoprecipitation with anti-HA beads (Fig. 7B). Western blot analysis with the anti-UGT1A antibody revealed two distinct bands, the lower band representing UGT1A1 HA and the upper band in each lane representing expression of each UGT1A CFP protein. These data confirm FRET analysis and lead us to conclude that UGT1A1 is able to heterodimerize with each of the UGT1A proteins.
Functional Analysis of Oligomeric UGT Isoforms-Functional analysis of several chimeric UGT isoforms was performed to ensure that the large fluorescent tags at the carboxyl terminus did not alter UGT activity. Three chimeras were chosen as representatives for functional analysis. Expressed  proteins by FRET analysis. COS cells were co-transfected with 5 g of pUGT1A1-CFP and 5 g of each pUGT1A-YFP plasmid for 24 h. Co-expression of both UGT1A1 CFP and UGT1A YFP is indicated by fluorescence in both the CFP channel and YFP channel. Correction for Donor and Acceptor bleed through was performed, and FRET signal was analyzed. FRET efficiency was evaluated with the Carl Zeiss AxioVision FRET software using the Acceptor Ratio analysis and depicted as a color-coded scale ranging from 0 to 100% FRET efficiency. UGT1A1 CFP and UGT1A3 CFP fusion proteins displayed similar levels of UGT activity for the substrate octylgallate as compared with the untagged UGT1A1 and UGT1A3 proteins, illustrated by similar intensities of bands representing the glucuronides in TLC analysis (Fig. 8A). Expressed UGT1A7 CFP also had equal levels of UGT activity for the substrate 2-napthol as compared with untagged UGT1A7 protein (Fig. 8A). These data indicate that the addition of the carboxyl-terminal tags did not affect the functionality of the UGTs.
The functional importance of UGT protein-protein interactions was also explored by co-transfecting two UGT isoforms and measuring the effect on specific activity. In Fig. 8B the cotransfection of pUGT1A1 and pUGT1A7 yielded approximately equivalent protein concentrations of each UGT isoform as compared with the single transfections of pUGT1A1 and pUGT1A7. Specific activity for 2-napthol was measured, and it was observed that UGT1A7 had 3-fold higher specific activity as compared with UGT1A1 (Fig. 8C). Co-transfection of pUGT1A1/pUGT1A7 resulted in a slightly increased specific activity for 2-napthol as compared with UGT1A7, potentially indicating an additive effect of specific activities (Fig. 8C).   (Table 1). After 24 h, the cells were lysed in 1 ml of radioimmunoprecipitation buffer supplemented with 1% protease inhibitor mixture and immunoprecipitated with the anti-HA beads overnight. Western blot analysis was performed using an anti-UGT1A antibody. Each lane shows the location of expressed UGT1A CFP and UGT1A HA proteins. B, in this experiment, 5 g of pUGT1A1-HA was co-transfected with 5 g of either pUGT1A3-CFP, pUGT1A4-CFP, pUGT1A6-CFP, pUGT1A7-CFP, pUGT1A8-CFP, pUGT1A9-CFP, or pUGT1A10-CFP. After 24 h, the cells were lysed and the immunoprecipitated proteins detected by Western blot analysis with the anti-UGT1A antibody. FIGURE 8. Functional analysis of chimeric and oligomeric UGT1A proteins. A, COS cells were transfected with 5 g of untagged pUGT1A1, pUGT1A3, or pUGT1A7 or 5 g of pUGT1A1-CFP, pUGT1A3-CFP, or pUGT1A7-CFP plasmids. Whole cell lysates were prepared, and each catalytic assay was conducted in a 100-l UGT reaction mix that contained 350 g of cellular protein and a final concentration of 500 M substrate (octylgallate for UGT1A1 and UGT1A3 isoforms and 2-napthol for UGT1A7). 100 l of the supernatant was spotted onto a TLC plate, and chromatography was performed in a mixture (35:35:10:20 v/v) of n-butanol:acetone:acetic acid:water. The dried TLC plate was then exposed to a phosphor screen overnight. B, COS cells were transfected with either 2 g of pUGT1A1 or 2 g of pUGT1A7 or co-transfected with 2 g of each plasmid. Whole cell lysates were prepared and Western blot analysis was performed using an anti-UGT1A antibody. C, COS cells were transfected with either 2 g of pUGT1A1 or 2 g of pUGT1A7 or co-transfected with 2 g of each plasmid. UGT assays were performed with 500 M 2-napthol as the substrate. The location of the glucuronides on the plate was identified, and that region of the TLC plate was scraped and quantified.

DISCUSSION
In the present study, we have utilized FRET to determine quaternary structure and intermolecular interactions among UGT1A proteins in live cells. In addition, we have provided complementary biochemical evidence for homo-and heterodimerization of these proteins by co-immunoprecipitation analysis. In this study, we have designed a collection of plasmids that express fluorescently tagged UGT1A proteins, allowing us to explore the intermolecular interactions among the UGT1A proteins in vivo. This technique is highly sensitive and allows for the detection of molecular interactions occurring within 1-10-nm distance and does not require the proteins to be cross-linked to detect this interaction.
Analysis of FRET between membrane-bound associated UGTs illustrates that expressed UGT1A1 CFP was able to homodimerize with UGT1A1 YFP . Association of UGT1A1 as a homodimer supports previous findings where nearest neighbor cross-linking studies followed by gel filtration have suggested the close proximity of the UGT1A1 proteins within the ER (20). Homodimerization analysis of other UGT1A isoforms by FRET also revealed that UGT1A3, UGT1A4, UGT1A6, UGT1A7, UGT1A8, UGT1A9, and UGT1A10 all have the capacity to homodimerize. Homodimerization of UGT1A9 by FRET supports previously published biochemical data that UGT1A9 forms a stable homodimer (34). The evaluation of UGT1A1 protein-protein interactions with other UGT1A proteins by FRET analysis also revealed the promiscuous nature of UGT1A1 to heterodimerize with UGT1A3, UGT1A4, UGT1A6, UGT1A7, UGT1A8, UGT1A9, and UGT1A10. Although it has been suggested from two-hybrid mammalian studies that UGT1A1 and UGT1A6 are unable to oligomerize (20), FRET analysis in conjunction with co-immunoprecipitation studies indicates that these two proteins are able to oligomerize in vivo. Two-hybrid systems require the bait and prey constructs to translocate to the nucleus and initiate transcriptional activation of the reporter genes. It is possible that the membrane-bound localization of the UGT proteins within the ER may have prevented translocation of the UGT bait and prey constructs from entering the nucleus. Our studies support other findings where interactions between UGT1A and UGT2B1 proteins have been suggested by co-immunoprecipitation experiments (19) and heterodimerization between UGT1A4 and UGT1A9 has also been observed (35).
In addition to the in vivo FRET technique, the homodimerization and heterodimerization of the UGT1A proteins were also evaluated by co-immunoprecipitation analysis. For this experiment each UGT1A protein was fused with a HA epitope at the carboxyl terminus. These co-immunoprecipitation experiments, conducted with anti-HA beads, provided an alternative and complementary technique to the in vivo FRET analysis. The co-immunoprecipitation data indicated homodimerization of UGT1A proteins with their respective isoform, clearly illustrated by the presence of UGT1A HA (ϳ56 kDa) and UGT1A CFP (ϳ86 kDa) proteins. This technique also confirmed the FRET results that UGT1A1 heterodimerized with all of the UGT1A plasmids. The generation of the library of UGT1A plasmids containing the CFP/YFP/HA epitopes provides a use-ful molecular tool to evaluate any combination of UGT interactions.
The functional implications of homo-or hetero-oligomerization of UGT1A proteins is of interest because several studies have suggested the importance of oligomerization for function. UGT2B1 has been suggested to form oligomers, and the functional relevance of this interaction became apparent when two different inactive mutant forms of UGT2B1 were co-expressed in COS cells and activity was restored (18). Another study suggests that dimerization may facilitate a dominant negative effect. Crigler-Najjar syndrome type II shows dominant-negative inheritance for bilirubin glucuronidation. When the mutated form of UGT1A1 found in Crigler-Najjar type II patients was co-expressed in COS cells with wild type UGT1A1, activity was found to be 6% that of the wild type (36). This result suggests that the oligomerization of mutant UGT1A1 with wild type UGT1A1 leads to an inactive aggregate, whereas expression of the wild type UGT1A1 is completely functional. The effects of dimerization on function were explored in this study by co-expression of UGT1A1 with UGT1A7 in COS cells. The data indicated that there was no significant difference in specific activities between cells co-expressing UGT1A1/UGT1A7 as compared with cells expressing only UGT1A7, whereas cells expressing only UGT1A1 had a much lower specific activity for 2-napthol. The FRET data and co-immunoprecipitation data suggest that UGT1A1 and UGT1A7 are capable of forming heterodimers, and therefore one can speculate that potentially 50% of the proteins in the co-expressing cells would be UGT1A1/ UGT1A7 heterodimers, while only 25% would be UGT1A7 homodimers. Because there was no significant difference in specific activities between cells expressing only UGT1A7 and cells co-expressing UGT1A1/1A7, this would suggest that the UGT1A1/UGT1A7 heterodimer has equal specific activity for 2-napthol as UGT1A7 homodimer. Although UGT1A1 was found to have a much lower specific activity for 2-napthol than UGT1A7, the presence of UGT1A1 in the heterodimer complex did not seem to inhibit UGT1A7 activity. Unlike the oligomerization studies performed with UGT2B1 mutants (18) and UGT1A1 mutants (36), the UGTs used in these experiments were wild type and fully functional and revealed that the heterodimer complex had functional activity similar to the homodimer.
Recent findings have also implicated protein-protein interactions between drug-metabolizing enzymes (37). Phase I enzymes, such as cytochrome P450 (CYP), and phase II enzymes, such as the UGTs, play important roles in detoxification of endogenous and exogenous compounds. During drug metabolism it is not uncommon for a metabolite produced by a cytochrome P450 to be sequentially metabolized by other enzymes to further facilitate elimination. The efficiency of concerted metabolism of compounds may be enhanced if enzymes physically interact with one another (38), implicating the significance of observed interactions between phase I and phase II proteins. CYP1A1 has been found to associate with microsomal epoxide hydrolase, UGTs, and NADPH cytochrome P450 reductase (39). Immunoprecipitation of CYP3A4 from human liver microsomes resulted in co-immunoprecipitation of UGT2B7, UGT1A1, and UGT1A6 (40). The functional rele-vance of CYP-UGT interaction was tested by co-expressing CYP3A4 and UGT2B7 in COS cells, and it was found that UGT2B7-catalyzed glucuronidation of morphine was altered by the presence of CYP3A4 (41). Studies have also indicated that baculovirus-expressed cytochrome P450 catalytic activity may be altered by the co-expression of different cytochrome P450s (42)(43)(44)(45). In summary, these studies indicate protein-protein interactions between drug-metabolizing enzymes in the ER and suggest the importance of such cooperative interactions in potentially modulating function.
Evidence provided in this work, as well as in the current literature, supports the hypothesis that UGTs form homo/heterodimer complexes. Under the stringent conditions of the whole cell lysis buffer, which contained 1% Nonidet P-40 and 0.25% deoxycholic acid as solubilizing agents, it might be anticipated that the protein-protein interactions are highly specific. Because all of the UGT1A proteins were found to homodimerize with their respective isoforms and heterodimerize with UGT1A1, we hypothesize that the dimerization domain may exist in the conserved carboxyl region (exons 2-5), although it has been suggested that the amino terminus of UGT2B1 mediates dimerization (18). The simultaneous insertion of the UGTs into the membrane also appears to be a requirement for oligomerization because dimerization was not observed when UGT1A1 HA and UGT1A1 CFP whole cell lysates were mixed and immunoprecipitated with anti-HA beads, which supports a previous observation by Kurkela et al. (34). This study provides evidence for biochemical and in vivo data for UGT1A proteinprotein interactions, but it is unclear whether the protein-protein interactions observed are important for function. Further studies are required to isolate the region(s) responsible for the protein interactions. Therefore, the focus of our future work will be to determine the structural domains involved in facilitating oligomerization of the UGT1A proteins and investigate the functional importance of the protein interactions among the UGTs.