Endogenous Protein Interactome of Human UDP-Glucuronosyltransferases Exposed by Untargeted Proteomics

Abstract

The conjugative metabolism mediated by UDP-glucuronosyltransferase enzymes (UGTs) significantly influences the bioavailability and biological responses of endogenous molecule substrates and xenobiotics including drugs. UGTs participate in the regulation of cellular homeostasis by limiting stress induced by toxic molecules, and by controlling hormonal signaling networks. Glucuronidation is highly regulated at genomic, transcriptional, post-transcriptional and post-translational levels. However, the UGT protein interaction network, which is likely to influence glucuronidation, has received little attention. We investigated the endogenous protein interactome of human UGT1A enzymes in main drug metabolizing non-malignant tissues where UGT expression is most prevalent, using an unbiased proteomics approach. Mass spectrometry analysis of affinity-purified UGT1A enzymes and associated protein complexes in liver, kidney and intestine tissues revealed an intricate interactome linking UGT1A enzymes to multiple metabolic pathways. Several proteins of pharmacological importance such as transferases (including UGT2 enzymes), transporters and dehydrogenases were identified, upholding a potential coordinated cellular response to small lipophilic molecules and drugs. Furthermore, a significant cluster of functionally related enzymes involved in fatty acid β-oxidation, as well as in the glycolysis and glycogenolysis pathways were enriched in UGT1A enzymes complexes. Several partnerships were confirmed by co-immunoprecipitations and co-localization by confocal microscopy. An enhanced accumulation of lipid droplets in a kidney cell model overexpressing the UGT1A9 enzyme supported the presence of a functional interplay. Our work provides unprecedented evidence for a functional interaction between glucuronidation and bioenergetic metabolism.

Keywords: UGT; affinity purification; human tissues; mass spectrometry; metabolism; protein-protein interaction; proteomics.

Figure 1

UGT1A interaction network investigated by untargeted proteomics. (A) The nine UGT1A enzymes are distinguished by the amino acid sequence of their substrate binding domain (unique peptides) whereas they share identical C-terminal co-substrate and transmembrane domains (common peptides). The anti-UGT1A antibody used in this study was raised against a C-terminal peptide common to all nine UGT1A enzymes but does not recognize the main spliced alternative isoforms 2 or UGT1A_i2s. (B) Experimental approach to establish endogenous UGT1A protein interactomes in drug metabolizing tissues and in the colon cancer cell model HT-29. Immunoprecipitation of UGT1A enzymes was conducted with the anti-UGT1A antibody. The numbers of common and unique UGT1A protein partners identified by mass spectrometry and above confidence threshold are represented in the Venn diagrams. Datasets were established on a minimum of two biological replicates. A Venn diagram for the 4 matrices is presented in Supplementary Figure 2. A list of proteins in each group is provided in Supplementary Table 3.

Figure 2

Quantitative overview of UGT1A enzymes immunoprecipitated from each tissue. Identification of immunoprecipitated UGT1A enzymes was based on the detection of peptides unique to specified UGT1A enzymes. (A) The quantitative assessment of each immunoprecipitated UGT1A is given by the total number of spectral counts for peptides unique to each UGT1A identified by mass spectrometry. Total spectral counts for peptides common to all UGT1A enzymes (Liver: 465; Kidney: 67; Intestine: 1561) were not considered in the quantitative assessment of specific UGT1As. (B) For each tissue, the number of peptides unique to each UGT1A identified by MS/MS analysis is represented in ring charts. Detailed quantification and unique/common UGT1A peptides identified are presented in Supplementary Table 1.

Figure 3

UGT1A interaction network in drug metabolizing tissues. UGT1A interacting proteins were classified according to KEGG pathways with ClueGO/CluePedia (Bindea et al., 2009, 2013). Node size is representative of pathway enrichment significance. Interactome was enhanced with significant interaction partners not part of KEGG pathways that are functionally related based on Uniprot and literature. These proteins are not linked to nodes but are grouped according to global functions. Structural proteins, ribosomal protein subunits and other RNA-binding proteins involved in mRNA splicing are not shown but were significantly enriched in UGT1A IPs. Full protein names are provided in Table 2. Complete lists of UGT1A interacting proteins are provided in Supplementary Table 2.

Figure 4

Validation of selected protein interactions by immunoprecipitation and immunofluorescence in a UGT negative kidney cell model. (A) Immunoprecipitation (IP) of UGT1A9, with purified anti-UGT1A antibodies, was conducted in HEK293-UGT1A9_myc/his transiently transfected with the indicated protein partner. UGT1A9 was immunodetected with anti-myc, whereas protein partners were detected with anti-tag antibodies as specified below immunoblots. Control IPs were conducted with normal rabbit immunoglobulins (IgG). Lysates (IP input) are shown as references. Protein bands denoted by the asterisk are the rabbit IgGs used in IPs. (B) Co-localization of UGT1A9 and the protein partners ACOT8 and SH3KBP1/CIN85 assessed by immunofluorescence in HEK293-UGT1A9_myc/his transiently expressing specified partners. Confocal microscope images are representative of three independent experiments. Partial co-localization is detected by yellow labeling in merged images. Insets present enlargements of boxed regions in merged images. Bar = 20 μm.

Figure 5

Accumulation of cellular lipid droplets in UGT1A9 expressing HEK293 cells. (A) Representative images of lipid droplets (green fluorescence) stained with Nile Red in HEK293-UGT1A9_myc-his or control HEK cells (stably transfected with the empty pcDNA3.1 vector—UGT negative cells). Bar = 20 μm. (B) Average number of lipid droplets per cell stably expressing UGT1A9 or control HEK cells. Lipid droplets per cell were counted in at least 140 cells per condition and averaged (n = 3 independent experiments).