MOLECULAR BIOLOGY, REGULATION, AND BIOCHEMISTRY OF UDP-GLUCURONOSYLTRANSFERASE ISOZYMES
Nikhil K. Basu, PhD, Staff Scientist
Rajat Banerjee, PhD, Visiting Fellow
Kushal Chakraborty, PhD, Visiting Fellow
Sunit K. Chakraborty, PhD, Visiting Fellow
Mousumi Basu, BS, Technical Training Fellow
![Photo of Ida Owens, Ph.D.](graphics/owens.jpg)
UDP-glucuronosyltransferase (UGT) isozymes carry out the essential role of converting innumerable, frequently encountered, structurally diverse lipophilic chemicals, including toxic metabolites, dietary constituents, environmental carcinogens, and therapeutics to glucuronides. Transformation of lipophilic chemicals to glucuronides hastens excretion, preventing tissue accumulation and toxic effects in the body. Neurotoxic bilirubin is the most important endogenous substrate. Given that the UGT isozymes prevent bilirubin neurotoxicities in children and inactivate common mutagens and carcinogens and prematurely clear therapeutic chemicals, it is important to understand the mechanism of glucuronidation in order to control chemical dispositions and thereby accelerate the removal of toxic chemicals while retaining medications to achieve maximal therapeutic benefits. Moreover, the enzymatic mechanism(s) and properties that enable a limited number of endoplasmic reticulum–bound UGTs to convert numerous structurally diverse lipophiles to innocuous glucuronides are unknown. We discovered and characterized the novel complex UGT1A locus encoding 13 UGT genes, organized to share a common carboxyl terminus, and cloned UGT2B7 and UGT2B15 and characterized catalysis. Thus, an important research aim is to determine the properties, mechanism(s), and molecular events controlling these isozymes in order to understand how exogenous as well as endogenous chemicals lead to the disease process.
Control of the UGT1 family of isozymes by PKC phosphorylation
UGTs, distributed primarily in the liver, kidney, and gastrointestinal tract, inactivate aromatic-like metabolites and a vast number of dietary and environmental chemicals, thus reducing the risks of toxicities, mutagenesis, and carcinogenesis. Even though UGT isozymes conjugate to glucuronic acid several chemical toxins present in our daily diet and environment, the properties and enzymatic mechanism(s) that enable endoplasmic reticulum (ER)–bound UGT isozymes to convert innumerable structurally diverse lipophiles to excretable glucuronides remain unknown. Inhibition of UGT activity and of immunoprecipitable [33P]orthophosphate in UGT1A7 and UGT1A10 following exposure to curcumin or calphostin-C indicates that the isozymes are phosphorylated (Basu et al., J Biol Chem 2004;279:28320). Furthermore, inhibition of UGT phosphorylation and activity by treatment with PKCe-specific inhibitor peptide supports PKC involvement. Computer analysis revealed that each UGT isozyme has four to six PKC sites. Co-immunoprecipitation, colocalization via immunofluorescence, and cross-linking studies of PKCe and UGT1A7His show 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, for S432G-UGT1A7, we observed a major shift of the 8.5 pH optimum to 6.4 with new substrate specificities, including 17beta-estradiol. We confirmed PKCe involvement by demonstrating (1) that PKCe overexpression enhances UGT1A7 activity, but not that of its S432 mutant, and (2) that conversion of 17beta-[14C]estradiol is catalyzed by S432G-UGT1A7, but not by UGT1A7. Consistent with our observations, treatment of UGT1A7-transfected cells with PKCe-specific inhibitor peptide or general PKC inhibitors dramatically increased 17beta-estradiol catalysis with parallel decreases in phosphoserine-432. Thus, protein kinase C–mediated phosphorylation of serine/threonine in UGT1A7 altered and expanded the enzymes’ substrate range, which necessarily confers survival benefit.
Inhibition by curcumin- or calphostin-C of recombinant UGT1A6 and UGT1A9 expressed in COS-1 provides further evidence that each UGT isozyme requires phosphorylation. Time-dependent inhibition of UGT1A6 by curcumin (50 µM) showed a modest level of reversal by 5 hours. The equivalent mutant K in UGT1A6 (S434G/A/D/R/) showed a shift from a single broad pH optimum to two pH optima. For UGT1A9, the equivalent mutant S432G/A/D/E showed a progressive diminution of activity for mycophenolic acid, its substrate. Similarly, mutants inserted in the UGT1A9-His construct expressed reduced activity, as assessed by Western blot. Curcumin and calphostin-C inhibited UGT1A9 activity in a dose-dependent manner. For UGT1A6, reversal of 95 percent inhibition by curcumin was much slower than for UGT1A9. The evidence shows that phospho-serine/threonine in UGT1A6 and UGT1A9 directly or indirectly controls activity. Thus, our cumulative evidence indicates that each UGT requires phosphorylation.
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;102:6285-90.
Owens IS, Basu NK, Banerjee R. UDP-glucurosyltransferases: gene structures of UGT1 and UGT2 families. Methods Enzymol 2005;400:1-22.
Dependence of UGT2B7 activity on phosphorylation by c-Src kinase
Earlier, we cloned UDP-glucuronosyltransferase-2B7 (UGT2B7) and demonstrated that it metabolizes genotoxic catechol estrogens produced in mammary gland and other estrogen target tissues. UGT2B7 contains three PKC and two tyrosine phosphorylation sites and incorporates immunoprecipitable [33P]orthophosphate. Mutagenesis indicated that Y236 and Y438 phosphorylation is required, but not that of T123, S132, or S437. Inhibition of UGT2B7 activity by the Src-specific inhibitor PP2, but not by PP3, in parallel with loss of phosphotyrosine-438-UGT2B7 content, indicated that Src phosphorylates UGT2B7. We accumulated additional strong evidence that Src-tyrosine kinase (TK) phosphorylates UGT2B7. We observed (1) a two-fold enhancement of both UGT2B7 activity and Y-438-phosphorylation by cotransfection of COS-1 cells with wild-type or activated Src-TK, but not by dominant negative Src-TK; (2) that mutation of Src-TK binding sites in UGT2B7 lowered anti–UGT2B7 pull-down of Src-TK in COS-1 cells transfected with solubilized UGT2B7; (3) colocalization of active Src-TK and UGT2B7, which was disrupted by PP2 pretreatment; (4) crosslinking of UGT2B7 and Src-Tk, which was disrupted by PP2; (5) trivial UGT2B7 activity following its expression in SYF−/− fibroblast cells compared with 14-fold–higher activity following expression in SYF+/− cells; and (6) increased in vitro phosphorylation of partially affinity-purified UGT2B7 by SrcTK and increased in vitro activity with ATP. Finally, dramatic losses of Src/active Src content and Src kinase activity, which parallel loss of glucuronidation by UGT2B7 and phosphotyrosine-438-UGT2B7 content in microsomes isolated from breast carcinomas compared with matched controls, are consistent with the notion that Src phosphorylation of UGT2B7 is needed to support UGT2B7’s activity. As Src is not only required for normal breast development but is also associated with breast carcinoma, we have provided strong evidence that Src phosphorylates UGT2B7, leading to inactivation of catechol estrogens, prevention of superoxide-mediated DNA depurination with possible tumor initiation, and, in effect, thus enables 2B7 to serve as a Src-dependent tumor suppressor.
Dependence of phosphorylation and activity of UGT2B15 on protein kinase C-epsilon and tyrosine kinase
Earlier, we cloned prostate- and testis-distributed UGT2B15 and demonstrated that it metabolizes several androgen intermediates as well as dihydrotestosterone, which is associated with benign prostate hyperplasia and prostate cancer. Treatment of UGT2B15-transfected COS-1 cells with PKC-kinase or tyrosine kinase inhibitors downregulated activity. Appropriate concentrations of the PKC-specific inhibitor calphostin-C inhibited activity by 97 percent, whereas the tyrosine kinase inhibitors genistein and herbimycin inhibited activity by between 10 and 25 percent. Computer analysis demonstrated that UGT2B15 contains three PKC and two tyrosine phosphorylation sites. Mutation analysis of UGT2B15 demonstrated that phosphorylation of a single PKC and of a single tyrosine site is required for activity. In contrast to trivial levels of UGT2B7 activity in transfected SYF−/− cells, UGT2B15 expressed high levels of activity in SYF−/−, which was inhibited by 64 and 88 percent when expressed in SYF+/− and SYF+/+ cells, respectively. The evidence suggests that UGT2B15 is inhibited by c-Src.
Regulation of UGT phosphorylation via signaling
The discovery that UGTs require phosphorylation derives from the finding that rapid and reversible downregulation of human UDP-UGT in LS180 colon cells follows curcumin treatment. Our aim is to identify the relevant kinases and mechanism(s) regulating phosphorylation of constitutive UGTs in LS180 cells and of 10 UGTcDNAs representing family A isozymes individually transfected into COS-1 cells. Time- and concentration-dependent inhibition of immunodetectable [33P]orthophosphate in UGTs and PKCe, following treatment of LS180 cells with curcumin or the PKC inhibitor calphostin-C, suggested that UGT phosphorylation is supported by active PKC(s). Immunofluorescence and co-immunoprecipitation studies with UGT-transfected COS-1 cells showed colocalization of UGT1A7His with PKCe and of UGT1A10His with PKCa or PKCd; inhibitors disrupted colocalization, providing evidence that certain PKCs and UGTs are proximately located. A PKCe-specific translocation antagonist peptide inhibited UGT activity, and typical PKC agonists dramatically stimulated curcumin-downregulated UGTs in LS180 cells, verifying a central role for PKC. Finally, catalase or herbimycin-A inhibition of constitutive or hydrogen peroxide–activated UGTs demonstrated that reactive oxygen species– (ROS) related oxidants behave as second messengers in maintaining constitutive PKC-dependent signaling, evidently sustaining UGT phosphorylation and activity. Given that cells use the signal transduction process to detect and respond appropriately to a changing environment, our results, combined with our earlier demonstration of phospho-dependent substrate selections by UGT, suggest that regulated phosphorylation functions efficiently by allowing adaptations to UGTs’ differential phosphate use.
UGT complex
To detail the required phosphorylation of UGTs, we have attempted to purify a catalytically active UGT protein for structural analysis. UGT1A7cDNA, adapted at the 3¢ end to specify Thrombin/His/Myc sites at the enzyme’s C-terminus and expressed by pBlueBac-based baculoviral and pSVL-based vectors in Sf9 insect cells and COS-1 cells, respectively, allows removal of the His-tag ligand site, if necessary. Because membrane-bound UGT requires an effective but non-inactivating detergent system, we developed an effective two-detergent system that permitted partial affinity-purification of UGT1A7His and UGT1A10His. UGT1A7 His, eluted from the affinity chromatography column, reproducibly contains 15 to 20 proteins. In addition, Western blot analysis showed a 225-kDa protein complex of one molecule of UGT1A7 (55-58 kDa), one of beta-COP (RACK1) (110 kDa), and two of 14-3-3 phosphoserine/threonine-binding protein (29 kDa). We confirmed involvement of the 14-3-3 chaperone protein by its colocalization with UGT1A1 His, UGT1A6His, UGT1A7His, UGT1A10, or UGT2B7His. Computer-based searches predicted that each UGT would contain at least two 14-3-3 binding sites: those in family A isozymes are S162 and T403/406, as well as S445 in UGT1A1, and those in family B isozyme UGT2B7 are T245 and T346. Mutants S162A- and S403A-UGT1A7 had null and 33 percent activity, respectively, whereas the S162A/S403A double mutant exhibited 25 percent activity. Mutational studies thus indicated that S162 is the critical 14-3-3 binding site in UGT1A7. Co-immunoprecipitation and colocalization studies, combined with mutational analysis of 14-3-3 binding sites, suggested that UGT1A7 is a cellular complex necessary for sustaining activity via protein kinase(s) signaling. Our findings are consistent with the greater stability of solubilized UGT with retention of the complex formation but loss of activity with disruption of the complex.
Structural analysis and identification of the common donor-substrate binding site in UGT1A10
Given that UGTs are bound to the membrane of the endoplasmic reticulum and are thus difficult to purify for crystallization, researchers have encountered difficulties in performing structural analyses of these critical detoxifying isozymes. Whereas all previous attempts to find structurally determined proteins homologous to the UGT1 isozymes have failed, Matthew Pennington, a former summer college student, carried out computer- and homology-based molecular modeling searches of the Protein Data Bank (PDB). He sought structural matches for UT1A10 by using the Silicon Graphics O2 workstation in Insight II with the software (Molecular Simulation, Inc.) Homology, Modeller, Discover, Bioploymer, and SeqFold expansion modules. The sequence homology search engine SeqFold uses a threading technique to identify potentially homologous proteins based on a predicted secondary structure of the target sequence and known secondary structures. It searches data banks of structurally solved proteins restricted to the non-redundant version of the Brookhaven PDB. With SeqFold, Pennington created a protein homologue and then used the structure of that sequence homologue to map the sequence of an unknown structure to a set of three-dimensional coordinates. He first applied the technique to protein regions near the core and on well-conserved sections and then built regions of poor conservation onto the core by using a fragment database and conformation-searching techniques. He optimized the initial model by using a simulated annealing technique.
Given that any sequence can be made into a model based on any protein, Pennington thoroughly evaluated the model for feasibility. After establishing a suitability model, he analyzed the model similar to that for crystallographic structure. He selected the new secondary-structure prediction–based search engine that is proficient in identifying structural homologues low on direct sequence-sequence identity to the target protein. Although SeqFold uncovered many low-identity homologues that bound similar substrates to UGT, the most prominent one with the most chemically proper structure was a UDP-galactose®UDP-glucose isomerase known as 1XEL in the PDB. Pennington located highly homologous regions of UGT1A10 and UDP-glucose isomerase and confirmed that they were those involved in binding to UDP-glucose, clearly analogous to UDP-glucuronic acid (UDPGA). He identified primarily three lysines (positions 314, 315, and 404) and one asparagine (position 292) predicted to be critical in identifying the uridine-diphosphate portion of the common donor substrate UDPGA. Site-directed mutagenesis data showed that all mutations at K-314 caused null activity and that substitutions at N-292, K-315, and K-404 had differential activities depending on the residue. Kinetics-of-inhibition studies with UDP-glucose and UDP-hexanolamine showed biphasic curves. Scatchard analysis with a radiolabeled photoactive chemical analogue of UDP-glucuronic acid revealed a high- and a low-affinity specific site. With the same affinity ligand, peptide-gel analysis for wild-type and mutant UGT1A10 also revealed that K-314 is the critical binding site.
Marked improvement in drug efficacy by targeted inhibition of glucuronidation in mice
The finding that UGTs require protein kinase C-mediated phosphorylation is important information because it allows manipulation of this critical system. With UGTs inactivated by downregulating PKCs with reversibly acting dietary curcumin, we used the immunosuppressant mycophenolic acid (MPA) in mice to determine the impact of gastro-intestinal glucuronidation on free drug uptake and efficacy. When expressed in COS-1 cells, mouse Ugt1a1, which typically is expressed in the animal's GI, glucuronidated both curcumin and MPA. As with human UGTs, the mouse isozyme underwent irreversible and reversible dephosphorylation by the PKC-specific inhibitor calphostin-C and general-kinase inhibitor curcumin, respectively, with parallel effects on activity. Moreover, oral administration of curcumin to mice reversibly inhibited glucuronidation in GI tissues. Finally, successive oral administration of curcumin and MPA to antigen-treated mice increased serum-free MPA and immunosuppression up to nine-fold. With MPA as a model, the results indicated that targeted inhibition of GI-glucuronidation in mice markedly improved free chemical uptake and efficacy.
Basu NK, Kole L, Basu M, McDonagh AF, Owens IS. Targeted inhibition of glucuronidation markedly improves drug efficacy in mice—a model. Biochem Biophys Res Commun 2007;360:7-13.
1 Amanda Garza, BS, former Predoctoral Fellow
2 Chimere Mbas-Jones, former Summer Student
3 Elaine Chang, former Summer Student
4 Partha S. Mitra, PhD, former Visiting Fellow
5 Labanyamoy Kole, PhD, former Postdoctoral Fellow
6 Matthew Pennington, former Summer Student
COLLABORATOR
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, contactowens@helix.nih.gov.