Mammalian UDP-glucuronosyltransferase (UGT) isozymes facilitate detoxification of endogenous metabolites and numerous potentially injurious dietary and environmental lipid-soluble phenols. The isozymes detoxify by attaching glucuronic acid to lipophilic chemicals and generating water-soluble excretable products. Glucuronidation reactions prevent tissue accumulation of neurotoxic levels of plasma bilirubin and dietary and environmental phenols and inactivate many chemicals to avert mutagenicity and carcinogenicity of aromatic hydrocarbons, including benzo(a)pyrene, which is found in cigarette smoke and automobile emissions. Conversely, extensive glucuronidation can be disadvantageous. The premature clearance of many orally administered therapeutic drugs is a long-standing problem associated with UGT metabolism. The administration of higher doses of drugs can compensate for such metabolism but can lead to serious side effects. Thus, for decades, drug inefficiency has been the impetus for developing inhibitor(s) of UGT. The enzymatic mechanism(s) and properties that enable a limited number of UGTs to convert numerous, structurally diverse lipid-soluble phenols to innocuous glucuronides has remained unknown. We discovered and characterized the UGT1 locus, which encodes 13 genes with a shared carboxy-terminus, and cloned UGT2B family members. Accordingly, we aim to understand the properties and mechanism(s) of the various isozymes that enable UGT enzymes to detoxify a vast number of agents in order to maintain chemical homeostasis.
Expansion of phosphorylation requirement of UGT1 family members: phosphorylation of UGT1 family members by PKC with signaling
Basu, Garza, Mitra, Change, Mbas-Jonas, Banerjee; in collaboration with Rivera
Curcumin- or calphostin-C–dependent inhibition of UGT1A6 and UGT1A9 individually expressed in COS-1 cells suggested that each UGT isozyme undergoes phosphorylation. Curcumin and calphostin-C inhibited UGT1A6 activity in a time-dependent manner. Inhibition of UGT1A6 by curcumin (50 microM) showed a modest level of reversal by five hours. While mutants at T73/T74 or T202/T206 in each isozyme caused null activity, S432G/S434G mutants for UGT1A7 and 1A10 were not affected or exhibited either a marked shift in pH optimum from 8.5 to 6.4 or a minor shift in pH optimum. S434G/A/D/R/K, the equivalent mutant in UGT1A6, showed a shift from a single broad pH optimum to two pH optima while S432G/A/D/E, the equivalent mutant for UGT1A9, showed a progressive diminution of activity for mycophenolic acid. Similarly, mutants inserted in the UGT1A9-His construct expressed reduced activity designed for Western blot analysis. Curcumin and calphostin-C inhibited UGT1A9 activity in a dose- and time-dependent manner. Whereas reversal of inhibition by curcumin occurred within five hours for UGT1A9 owing to curcumin’s metabolism, reversal was much slower in the case of UGT1A6, which does not metabolize the agent. The evidence indicates that phospho-serine/threonine in UGT1A6 and UGT1A9 directly or indirectly controls activity.
The properties and enzymatic mechanism(s) that enable endoplasmic reticulum–bound UGT isozymes to convert innumerable structurally diverse lipophiles to excretable glucuronides are unknown. Inhibition of cellular UGT1A7 and UGT1A10 activities and of [33P]orthophosphate incorporation into immunoprecipitable proteins following exposure to curcumin or calphostin-C (see above) indicates that the isozymes are phosphorylated. Furthermore, inhibition of UGT phosphorylation and activity by treatment with a PKCepsilon-specific inhibitor peptide supports the notion that PKC is involved. Computer analysis revealed that each UGT isozyme has between four and six PKC sites. Co-immunoprecipitation, co-localization via immunofluorescence, and cross-linking studies of PKCepsilon and UGT1A7-His revealed that the proteins reside within 11.4 Å of each other. Mutation of three PKC sites in each UGT isozyme demonstrated that T73A/G and T202A/G caused null activity, whereas S432G-UGT1A7 caused a major shift in the enzyme’s pH 8.5 optimum to 6.4 with new substrate preferences, including 17beta-estradiol. We confirmed PKCepsilon involvement by demonstrating that PKCepsilon overexpression enhanced activity of UGT1A7 but not that of its S432 mutant and that S432G-1A7, but not unmutated 1A7, converts 17beta-[14C]estradiol. Consistent with these observations, treatment of 1A7-transfected cells with the PKCepsilon-specific inhibitor peptide or general PKC inhibitors dramatically increased 17beta-estradiol catalysis with parallel decreases in serine-432 phosphorylation. Thus, protein kinase C–mediated phosphorylation of serine/threonine in UGT regulates substrate specificity, which possibly confers survival benefit.
Basu NK, Ciotti M, Hwang MS, Labanyamoy L, Mitra PS, Cho JW, Owens IS. Differential and special properties of the major human UGT1-encoded gastrointestinal UDP-glucuronosyltransferases enhance potential to control chemical uptake. J Biol Chem 2004;279:1429-1441.
Basu NK, Kole L, Kubota S, Owens IS. Human UDP-glucuronosyltransferases show atypical metabolism of mycophenolic acid and inhibition by curcumin. Drug Metab Dispos 2004;32:768-773.
Basu NK, Korava M, Garza A, Kubota S, Saha T, Mitra PS, Banerjee R, Rivera J, Owens IS. Phosphorylation of UDP-glucuronosyltransferase regulates substrate specificity. Proc Natl Acad Sci USA 2005;102:6285-6290.
Basu NK, Kubota S, Meselhy MR, Ciotti M, Chowdhury B, Hartori M, Owens IS. Gastrointestinally distributed UDP-glucuronosyltransferase 1A10, which metabolizes estrogens and non-steroidal anti-inflammatory drugs, depends upon phosphorylation. J Biol Chem 2004;279:28320-28329.
Phosphorylation of a UGT2 family member by src tyrosine kinase
Mitra, Basu
UGT2B7, a UGT2B family member that metabolizes endogenous agents such as bile acid and the estrogenic substrates 4-hydroxy-estrone and estriol, requires occupation of tyrosine kinase phosphorylation sites at tyrosines-236 and -438. We showed that UGT2B7 requires phosphorylation by src tyrosine kinase (srcTK). While separate mutations at three PKC sites in UGT2B7 had no effect on activity, double and triple mutations either slightly decreased or increased 17-epiestriol activity between 1.5- and two-fold. By contrast, a single or double mutation at the tyrosine sites caused null activity for all substrates. Comparison of incorporation of [33P]orthophosphate into wild type, PKC-site triple mutants, tyrosine-site double mutants, and PKC- and tyrosine-site penta-mutants shows maximum labeling that progressively decreased. Herbimycin-C, genistein, lavendustin A, and geldanamycin, which are tyrosine kinase inhibitors, inhibited in cellulo UGT2B7 activity. Vitamin D, the srcTK-specific activator, caused a two-fold increase in activity and in the level of phospho-Y438 detected by anti-phosphoY438-UGT2B7 without affecting the UGT protein level. In addition, srcTK, [33gammaP]ATP, and peptide KKWDQFYSEV (position Y236) or RVINDPSYKENV (position Y438) incorporated five-fold higher label than Y236F or Y438F mutant peptide in an in vitro assay system. Cross-linking studies with and without the srcTK-specific inhibitor PP2, but not PP3, analyzed by Western blots with anti–UGT, anti-phospho-Y438-UGT2B7, and anti-srcTK showed that the kinase and UGT2B7 are within 11.4 Å of each other and that the kinase likely phosphorylates Y438 in UGT2B7. The results suggest that various combinations of phosphorylated PKC sites cause variations in the level of activity, whereas srcTK sites are obligatory for activity.
Purification of UGT1A7
Garza, Basu; in collaboration with Negishi
We adapted UGT1A7 cDNA at the 3´ end to specify the protein carboxy-terminus with serially linked Thrombin/His/Myc sites and inserted it into baculoviral-based pBlueBac vector for infecting Sf9 insect cells. We designed the pUGT1A7Th-3/His construct to allow high-level production of the enzyme, which we adapted to allow removal of the His-tag ligand site, if necessary. To date, we have optimized strategies for purification of the wild-type UGT1A7Th-3His protein by using specific detergent combinations and affinity chromatography. The system is now ready to undergo scale-up to grow quantities of Sf9 cells for high-level production of endoplasmic reticulum–bound UGT1A7 for purification of sufficient quantities to allow structural studies.
Structural analysis of UGT
Banerjee, Basu, Pennington1
Given that UDP-glucuronosyltransferases are bound to the membrane of the endoplasmic reticulum and are thus difficult to purify for crystallization, structural analysis of these critical detoxifying isozymes has been difficult. All previous attempts to find structurally determined proteins homologous to the UGT1 isozymes have failed. Matthew Pennington, a summer college student, carried out computer- and homology-based molecular modeling searches of the Protein Data Base, seeking structural matches for UT1A10. He used a Silicon Graphics O2 workstation and Insight II software (Molecular Simulation, Inc.) with the Homology, MODELER, Discover, Biopolymer, and SeqFold expansion modules. The first modules are necessary for all homology or protein work. The SeqFold module is a sequence-homology search engine that uses a threading technique to identify potentially homologous proteins based on the predicted secondary structure of the target sequence and known secondary structures within structurally solved protein databases (PDBs). All searches were performed against a nonredundant version of the Brookhaven PDB. Pennington created a protein homologue by using SeqFold and then using the structure of that sequence homologue to map the sequence of an unknown structure to a set of 3D coordinates. The application first focused on protein regions near the core and on well-conserved sections. Regions of poor conservation were then built onto the core by using a fragment database and conformation searching techniques. Pennington optimized the initial model by using a simulated-annealing technique and then selecting SeqFold, the new secondary-structure prediction–based search engine, which is proficient in identifying structural homologues with low direct sequence-sequence identity to the target protein. Although SeqFold uncovered many low-identity homologues that bound to substrates similar to UGT, the most prominent homologue with the most chemically appropriate structure was a UDP-galactoseÆUDP-glucose isomerase, known as 1XEL in the PBD. Highly homologous regions of UGT1A10 and 1XEL were located and confirmed to be those involved in binding UDP-glucose, clearly analogous to UDP-glucuronic acid (UDPGA). Pennington primarily identified two lysine residues (positions 314 and 315) and one asparagine residue (position 292) as likely to be critical in recognizing the uracil-diphosphate portion of the donor substrate UDPGA.
Our recent studies show that mutants of UGT1A10 at 314, but not 315, caused null activity. Activity for mutants at lysine 404, shown to be proximal to lysine residues 314 and 315 in the predicted UDPGA binding site, was not affected or had a sharper pH optimum. Binding of the affinity ligand 5-N3-UDPGA to UGT1A10 inhibited activity. We adapted recombinant UGT1A10 with a His-affinity ligand at its carboxy-terminus and purified it. After binding the ligand to [33P]5-N3-UDPGA to identify the binding site, we will fragment the protein by treatment with BNPS-Skatole and sequence the appropriate peptides.
Mackenzie PI, Bock KW, Burchell B, Guillemette C, Ikushiro SI, Iyanagi T, Miners JO, Owens IS, Nebert DW. Nomenclature update for the mammalian UDP-glycosyltransferase (UGT) gene superfamily. Pharmacogenet Genomics 2005;15:677-685.
Owens IS, Basu NK, Banerjee R. Gene structures of the UGT1 and UGT2 families. Methods Enzymol 2005;400 (in press).
1Matthew Pennington, former Summer Student
Collaborators
Antony McDonagh, PhD, University of California San Francisco, San Francisco, CA
Masahiko Negishi, PhD, Laboratory of Reproductive and Developmental Toxicology, NIEHS, Research Triangle Park, NC
Juan Rivera, PhD, Molecular Immunology and Inflammation Branch, NIAMS, Bethesda, MD
For further information, contact owens@helix.nih.gov.