Introduction

The liver is a central hub for vital functions ensuring maintenance of body fuel, detoxification, hormonal regulation and immune functions. It regulates body homeostasis by filtering and detoxifying multiple small metabolites carried from distal organs including stomach, pancreas, and intestines through the portal vein as well as from exogenous sources. It is also a highly metabolic and secretory organ, regulating blood levels of multiple proteins, sugars and other small metabolites, sex steroid hormone metabolism, lipid and protein synthesis and signaling. Less known but equally important, the liver has a significant role in immune functions.

The metabolic enzyme UDP-glycosyltransferase UGT2B17 is well expressed in the liver, and predominantly in hepatocytes.1,2 It is one of the 22 human UGT enzymes that regulate drug response and the bioactivity of endogenous metabolites by catalyzing the conjugation of small molecules to glucuronic acid. Known UGT2B17 functions include the detoxification of drugs such as the histone deacetylase inhibitor vorinostat, the prostaglandin D2 receptor inhibitor MK-7246, and several anti-leukemics, and the modulation of endobiotics bioactivity such as androgens and signaling lipids.3,4,5,6,7,8,9

Expression of UGT2B17 has attracted a significant level of attention because of its association with the risk and progression of multiple cancers such as leukemia, prostate and colon cancers, and of other clinical conditions including bone mineral density, puberty and autoimmune diseases.10,11,12,13,14,15,16,17 It also predicted resistance to anti-leukemic drug treatments in chronic lymphocytic leukemia.9,15 The mechanisms by which UGT2B17 contributes to the etiology, progression and drug response of these diseases remain elusive, given that its endogenous functions are mostly unknown, and may involve both metabolic and non-enzymatic functions.3,15,18 This knowledge gap may have a critical impact on the individuals that are UGT2B17 gene-deficient, owing to a complete natural germline deletion of the gene, a genetic variation found in 9% of Caucasians and over 70% Asian individuals.1,19,20,21 We have recently exposed the severe consequence of a complete UGT2B17 gene deletion on the circulating metabolome of prostate cancer patients, affecting most classes of metabolites including androgens.1 Notably, circulating bile acids, triacylglycerols, sphingolipids and kynurenine were particularly altered in deficient vs homozygous gene-carrier prostate cancer individuals.

In this study, we investigated the impact of a complete UGT2B17 gene deletion on the human liver proteome to gain further mechanistic insights about the endogenous functions of UGT2B17 in the liver and its role in liver homeostasis. The distinctive hepatic protein profile of deficient individuals exposed in this study lends support to a significant influence of UGT2B17 on numerous metabolic pathways, redox control and endoplasmic reticulum (ER)-related functions such as secreted protein synthesis. The UGT2B17 deficiency was further associated with a rewired metabolism that might trigger several stress responses, enhanced inflammation and immune responses.

Methods

Liver specimens

Normal liver tissue samples (n = 15) were from males (5) and females (10) of European descent. As described previously, samples were taken from patients undergoing liver surgery, mostly due to metastasis resection, at the Department of General, Visceral, and Transplantation Surgery at the Charité (Campus Virchow, University Medical Center Charité, Humboldt University Berlin, Germany).22 Only tumor-free material was part of the liver bank and the diagnosis of undiseased liver specimen was confirmed by an independent pathologist with expertise in liver diseases, before storage at − 80 °C. The local ethical committees of the Charité, Humboldt University (Berlin, Germany) as well as the University Clinic Tuebingen, Germany and CHU de Québec Université Laval (2012–262), approved the use of anonymized human liver tissues for proteomics analyses. All liver donors or their legal guardian gave their written informed consent, and the study was conducted in accordance with the Declaration of Helsinki. Descriptive characteristics of each donor are provided in Table 1. The germline UGT2B17 genetic status (at least one gene copy (proficient, n = 8) or complete deletion of both gene copies (deficient, n = 7)) of each donor was also established by PCR analysis of genomic DNA extracted from liver tissues as previously described.19 The deletion status was further confirmed by targeted proteomics using a UGT2B17 signature peptide and by Western blotting with the mAb UGT2B17 monoclonal antibody.19,23 The gene-proficient group was composed of individuals with one or two gene copies given the documented minor effect of copy number variation on UGT2B17 expression or activity.8,19,24.

Table 1 Characteristics of liver donors.
Full size table

Proteomic analysis

Sample preparation

Each liver tissue (10–60 mg) was dissolved in a protein extraction buffer containing 10 mM HEPES–KOH pH 7.5, 1 M ammonium bicarbonate and 8 M urea. The samples were sonicated for 2 min (pulse on: 10 s, pulse off: 5 s) with an amplitude of 25% (Sonic Dismembrator Model 120, Thermo Fisher Scientific, Waltham, MA, USA). The samples were spun at 21,100 × g for 10 min at 4 °C to remove insoluble debris. The protein concentration was determined by bicinchoninic acid protein assay and adjusted to 1.5 μg/μl with the protein extraction buffer. Proteins (75 μg) were reduced by adding dithiothreitol (DTT) to a final concentration of 5 mM and by heating at 95◦C for 2 min, followed by a 30-min incubation at room temperature. The alkylation of proteins was carried out by adding chloroacetamide (Sigma-Aldrich, St-Louis, MO, USA) to a final concentration of 7.5 mM followed by a 20-min incubation at room temperature protected from light. Samples were diluted with three volumes of 1 M ammonium bicarbonate to bring the urea concentration to 2 M. The proteins were digested with 1 μg Pierce mass-spectrometry (MS)-grade trypsin (Thermo Fisher Scientific) and incubated overnight at 30 °C with shaking. Trifluoroacetic acid (TFA) (Sigma-Aldrich) was added to a final concentration of 0.2% to stop the digestion. The peptides were purified with ZipTip 100 μl micropipette tips containing a C18 column (EMD Millipore, Burlington, VT, USA). ZipTips were equilibrated in 100% acetonitrile (ACN) then in 0.1% TFA buffer. Each peptide sample was passed on a ZipTip twice, washed in 0.1% TFA and peptides were eluted three times with 100 μl of 50% ACN/1% formic acid (FA) buffer. The peptides concentrated by centrifugal evaporation at 60 °C until complete dryness, resuspended in 50 μl of 1% FA buffer, transferred to a glass vial (Thermo Fisher Scientific) and stored at − 20 °C until MS analysis.

DIA LC–MS analysis

For each sample, peptides (250 ng) were injected into an HPLC (nanoElute, Bruker Daltonics, Billerica, MA, USA) and loaded onto a trap column with a constant flow of 4 µL/min (Acclaim PepMap100 C18 column, 0.3 mm id × 5 mm, Dionex Corporation, Sunnyvale, CA, USA) then eluted onto an analytical C18 Column (1.9 µm beads size, 75 µm × 25 cm, PepSep, city) heated at 50 °C. Peptides were eluted over a 2-h gradient of ACN (5–37%) in 0.1% FA at 400 nL/min while being injected into a TimsTOF Pro ion mobility mass spectrometer equipped with a Captive Spray nano electrospray source (Bruker Daltonics). Data was acquired using diaPASEF mode. Briefly, for each single TIMS (100 ms) in diaPASEF mode, we used 1 mobility window consisting of 27 mass steps (m/z between 114 to 1414 with a mass width of 50 Da) per cycle (1.27 s duty cycle). These steps cover the diagonal scan line for + 2 and + 3 charged peptides in the m/z-ion mobility plane. Two replicate LC–MS/MS analysis was conducted per sample.

Protein identification

The raw files were analyzed using MaxQuant (version 2.0.1.0) and the Uniprot human proteome database (21/03/2020, 75,776 entries), with uploading in silico generated human library files (available at: http://annotations.perseus-framework.org/) to run Maxquant in DIA discovery mode. The settings used for the analysis (with TIMS-MaxDIA type in group-specific parameters) were: 1 miscleavage allowed; fixed modification was carbamidomethylation on cysteine; enzymes were Trypsin (K/R not before P); variable modifications included in the analysis were methionine oxidation and protein N-terminal. A mass tolerance of 20 ppm and 40 ppm was used for precursor and fragment ions, respectively. Identification values “PSM FDR”, “Protein FDR” and "Site decoy fraction" were set to 0.05. The minimum peptide count was set to 1. Both the “Second peptides” and "Match between runs" options were allowed. MaxQuant was run with a transfer q value of 0.3. Proteins positive for at least either one of the “Reverse” or “Potential contaminant” categories were eliminated.

Statistical analysis of protein quantification was conducted with the Perseus software (version 2.0.6.0).25 Proteins only identified by site and identified with only one peptide were not considered. LFQ intensities (sum of replicate analysis) were Log2-transformed, proteins identified in at least 70% samples in at least one group (proficient or deficient) were retained for subsequent analyses, and missing values were imputed according to normal distribution. Proteins differentially expressed between UGT2B17-deficient and proficient individuals were identified by a two-sided Student t-test. None of the differentially expressed proteins passed the significance threshold when adjusted for multiple hypothesis testing (Benjamini–Hochberg false discovery rate). Principle component analysis and Volcano plots were generated with the Perseus software. Due to an insufficient number of male liver samples, this study could not assess the sex-specific protein profiles of UGT2B17-deficient livers. However, given the documented sex-dependent expression of UGT2B17,23 the sex of liver donors was indicated in relevant graphs for sake of comparison. The lists of measured and differentially expressed proteins are provided in Supplementary Table 1. Because of the high protein sequence identity between UGT2B15 and UGT2B17, the relative abundance of UGT2B15 and UGT2B17 proteins reported in Supplementary Table 1 was manually calculated based on the intensity of their unique peptides with a posterior error probability < 0.05 without imputation. Unique peptides for the quantification of UGT2B15 were IHHDQPMKPLDR, SVINDPVYKENVMK, NYLEDSLLK, ILDRWIYGVSK, SVINDPVYK and for UGT2B17 were FSVGYTVEK, LCEDAVLNKK, SVINDPIYKENIMK, NDLEDFFMK, LCEDAVLNK. The complete list of quantified peptides in this study is provided in Supplementary Table 3. Functional annotation clustering of proteins significantly changed (P ≤ 0.05) were conducted with the Database for Annotation, Visualization and Integrated Discovery (DAVID26) with a Kappa similarity term overlap of 5 and default values for all other options, and with ShinyGO,27 using the REACTOME database. ADME and related protein expression included 360 proteins, as listed in PharmaADME.org and our previous ADME panel28,29 (Supplementary Table 2).

Metabolomics analysis

The quantification of water-soluble metabolites in human liver tissue by untargeted metabolomics and targeted bile acid metabolomics has been previously described.30,31,32 In brief, for untargeted analysis, liver tissue was homogenized and subsequently metabolites extracted in a two-step procedure. Dried aqueous metabolite extracts were reconstituted in water/acetonitrile (5:95, v/v) with 100 µl/mg tissue and HILIC separated on a UPLC System coupled to a QTOF-MS. Data was preprocessed by targeted extraction of annotated features and peak areas were normalized by median-normalization following LOESS correction based on QC samples. In summary, 98 metabolites were annotated and included in the statistical analyses of differential abundant in UGT2B17 deficient samples. For the targeted quantification of bile acids by LC–MS, liver cytosol was diluted with water and subjected to methanol precipitation in presence of deuterium-labeled internal standards. Calibration samples for the generation of calibration curves were processed in parallel. Dried metabolites were re-suspended in methanol:water (1:9, v/v) prior to LC–MS analysis. The precision and accuracy of the method were evaluated by analyzing quality control (QC) samples, prepared like the calibration samples. Bile acid quantitative data was normalized to protein amount. Statistical analysis of metabolite quantification was performed by group comparisons using the Wilcoxon rank sum test. P-value < 0.05 was defined as significant. The differences were no longer significant when adjusted for multiple hypothesis testing (Benjamini–Hochberg false discovery rate).

Results

Approximately 3000 proteins were quantified with a minimum of two peptides in normal liver sample (n = 15) by data-independent acquisition (DIA) mass spectrometry (Fig. 1A; Supplementary Table 1). The UGT2B17-deficient status was confirmed through multiple methods: genetic analysis of germline DNA from the same donors, immunoblotting with a specific monoclonal antibody, peptide-specific mass spectrometry quantification, and the detection of unique UGT2B17 peptides identified in this study (Table 1).19,23 Given the documented minor effect of copy number variation on UGT2B17 expression and activity,8,19,24 the gene-proficient group was composed of individuals with one or two gene copies (Table 1). The Spearman correlation between quantification by the two MS approaches was elevated (r = 0.953, P = 0.0001). As expected, UGT2B17 was the most differentially negatively expressed in deficient livers (Fig. 1B). UGT2B17-deficient tissues were characterized by 233 differentially expressed (DE) proteins relative to livers of gene-carriers, with a majority (82%) more abundant in UGT2B17-deficient livers (Fig. 1B, Supplementary Table 1). The names of DE proteins, along with their corresponding gene names in bold, are given in the text and in Supplementary Table 1. The distinctive hepatic proteome of UGT2B17-deficient livers was illustrated by a principal component analysis (Fig. 1C).

Fig. 1
figure 1

Protein profiling in UGT2B17-deficient livers. (A). Study pipeline. Normal liver samples were genotyped to identify UGT2B17 gene carriers or homozygote germline deletions then quantitatively profiled by label-free data-independent proteomics to identify differentially expressed proteins. The UGT2B17 gene status was genotyped by PCR and further confirmed by multiple reaction monitoring mass spectrometry and immunoblotting. Created with BioRender.com. (B) Differentially higher (red dots) and lower (blue dots) expressed proteins. Top differentially expressed proteins are labeled. Unchanged proteins are shown in grey. (C) Principal component analysis identified discriminating protein profiles in a majority of samples. (D) Enriched subcellular localization of the differentially expressed proteins. (E) Pathway enrichment analysis of the 233 significantly changed proteins identified two main clusters of affected pathways.

Full size image

Proteins displaying most enriched abundance included the copper metabolism MURR1 domain (COMMD)-containing protein 3 (COMMD3), a multifunctional protein involved in copper homeostasis and endosomal sorting, the cytochrome b-245 heavy chain (CYBB) encoding the monocytic NADPH oxidase (NOX2) enzyme complex, and the ATP-binding cassette sub-family F3 transporter (ABCF3), as the most upregulated. On the opposite, the heavy metal scavenger metallothionein-1 (MT1M), transcription initiation factor TFIID subunit 2 (TAF2) as well as the detoxifying glutathione S-transferase theta-1 (GSTT1) enzyme were the most downregulated (Fig. 1B; Table 2).

Table 2 Hepatic proteins most differentially expressed in livers of UGT2B17-deficient individuals.
Full size table

DE proteins were enriched in endoplasmic reticulum (ER) and lysosomal lumen proteins (Fig. 1D) and clustered in two main functional pathways related to biological insults, namely (i) detoxification and cellular stress and (ii) the immune system (Fig. 1E), suggesting a constitutive stress response ongoing in the liver proteome deficient in UGT2B17 protein expression.

As anticipated, given the function of UGT2B17 in chemical defense processes, proteins associated with cellular stress and drug response pathways were considerably divergent in UGT2B17-deficient livers, including several "absorption, distribution, metabolism, and excretion" (ADME) proteins, suggesting potential compensatory mechanisms but not by other UGT enzymes (Fig. 2A,B; Supplementary Table 2).

Fig. 2
figure 2

Expression of absorption, distribution, metabolism, excretion (ADME) proteins in UGT2B17-deficient livers. (A) Based on a list of 360 ADME proteins and related proteins,38,39 152 ADME were quantitatively profiled in liver tissues. (B) Differentially expressed (DE) ADME proteins in deficient livers. Log2FC < 0: Lower abundance; Log2FC > 0: Higher abundance in UGT2B17-deficient livers. (C–E) Divergent protein abundance in UGT2B17 proficient ( +) and deficient (−) individuals: (C) Redox control DE enzymes: biliverdin reductase (BVRA), glutathione peroxidase 2 (GPX2) and superoxide dismutase 2 (SOD2). (D) Phase II DE enzymes: glutathione S-transferase M3 (GSTM3), glutathione S-transferase T1 (GSTT1), nicotinamide methyl transferase (NNMT). (E) Other significantly DE ADME proteins dehydrogenase/reductase SDR family X (DHRSX), aldehyde dehydrogenase member X (ALDH1B1) and urocanate hydratase (UROC1) in UGT2B17 proficient and deficient individuals. In boxplots, sex of liver donors is indicated (blue dots, males; red dots, females), the line represents median expression, “+ ” represents mean expression, and whiskers the range of values. *P < 0.05, **P < 0.01. Expression of ADME enzymes is detailed in Supplementary Table 2.

Full size image

ADME. The oxidative stress-related proteins superoxide dismutase (SOD2), glutathione peroxidase 2 (GPX2), and biliverdin reductase A (BVRA), were significantly elevated, whereas peroxiredoxin 5 (PRDX5) was lower in UGT2B17-deficient livers (Fig. 2C, Supplementary Table 1). Three phase II conjugating enzymes were differentially expressed, namely the glutathione-S-transferases GSTM3 and GSTT1, respectively higher and lower in UGT2B17-deficient livers, as well as a higher abundance of the nicotinamide N-methyltransferase NNMT (Fig. 2D). In all, eight ADME proteins were significantly changed, including an increased abundance of the dehydrogenase/reductase SDR family member X (DHRSX), and lower abundance of the aldehyde dehydrogenase X (ALDH1B1) and the urocanate hydratase UROC1 (Fig. 2B-E) in UGT2B17-deficient livers.

Additional ADME proteins were differentially expressed but narrowly missed significance, namely the ABC transporter ABCA2, alcohol dehydrogenases ADH4 and ADH6, glutathione S-transferases GSTA2 and MGST3, solute carrier organic anion transporter SLCO1B3, and the superoxide dismutase SOD3 that tended to be reduced, whereas GSTM1 and the solute carrier SLC22A18 tended to be considerably higher (Fig. 2B; Supplementary Table 1). By contrast, none of the 10 other measured UGT proteins, nor the 25 cytochrome P450 (CYP) enzymes measured were affected by the absence of UGT2B17 hepatic expression.20.

Endobiotic homeostasis. There is a significant functional overlap for proteins involved in drug-related pathways that also largely contribute to endobiotic or metabolite homeostasis. For instance, the lower abundance of the ADME ALDH1B1 enzyme involved in the oxidative detoxification of alcohol (Fig. 2E), also participates in the oxidation of retinaldehyde and in the metabolism of corticosteroid. Two hormone-related enzymes, the 17β-hydroxysteroid dehydrogenases 17β-HSD7 (HSD17B7) and 17β-HSD12 (HSD17B12), were significantly more abundant in UGT2B17-deficient livers (Fig. 3A). Both 17β-HSDs contribute to the biosynthesis of estradiol, potentially leading to significant changes in estrogen biotransformation in favor of increased estradiol exposure in individuals lacking UGT2B17.

Fig. 3
figure 3

Divergent expression of steroid- and NAD+ pathway enzymes in UGT2B17 proficient (+) and deficient (−) livers. (A) 17β-hydroxysteroid reductases (HSD17B) enzymes significantly differentially expressed. (B) Altered enzymes of the NAD+ metabolism pathway. A simplified scheme of the NAD+ synthesis pathway is shown with the differentially expressed proteins. In boxplots, sex of liver donors is indicated (blue dots, males; red dots, females), the line represents median expression, “ + ” represents mean expression, and whiskers the range of values. *, P < 0.05, t, P < 0.1.

Full size image

Similarly, the phase II ADME enzyme nicotinamide N-methyltransferase (NNMT) is a critical regulator of nicotinamide adenine dinucleotide (NAD) availability. It displayed a higher abundance, along with nicotinamide phosphoribosyltransferase (NAMPT), whereas nicotinate phosphoribosyltransferase (NAPRT) was less abundant in deficient livers (Fig. 3B). This rewiring suggests that cellular NAD+ and NADP+ pools may be altered in livers of UGT2B17-deficient individuals. Previous metabolomics profiling of these liver tissues reveals disrupted endobiotic homeostasis in UGT2B17-deficient livers,30,31,32 characterized by elevated taurocholic acid and reduced levels of both adenine and hydroxyisovalerylcarnitine, a product of leucine metabolism (Supplementary Fig. 1).

Carbohydrate homeostasis. A disrupted endogenous carbohydrate homeostasis was evidenced by the significantly different abundance of key enzymes involved in glucose-related pathways between deficient and proficient livers. Glucokinase (GCK), a master regulator of glucose homeostasis that catalyzes the first and rate-limiting step of glycolysis, was one enzyme with the most reduced abundance, by nearly eightfold in UGT2B17-deficient livers (Figs. 1B and 4). GCK generates glucose-6-phosphate, the key glucose metabolite feeding glycolysis, the pentose phosphate pathway (PPP), and the uronic acid pathway (Fig. 4). An increased abundance of the PPP enzymes hexose-6-phosphate dehydrogenase (H6PD) and transketolase (TKT) suggested a more active PPP pathway, by diverting glucose-6-phophate for the synthesis of nucleotide and the redox NADPH metabolite. The phosphoglycerate dehydrogenase (PHGDH), which diverts another glycolytic intermediate for serine and glycine biosynthesis and redox NADH metabolite was rather less abundant. Collectively, these observations support disturbed glycolysis and redox capacity in UGT2B17-deficient livers.

Fig. 4
figure 4

Enzymes of energy metabolic pathways divergent in UGT2B17-deficient livers. Glucose-6-phosphate is a central metabolite produced by the phosphorylation of glucose by glucokinase (GCK). Its fate is diverse, feeding glycolysis, the pentose phosphate pathway (PPP) and the uronic acid pathway. Lower GCK and higher hexose 6-phosphate dehydrogenase (H6PD) and transketolase (TKT) expression could favor the pentose phosphate pathway and production of NADPH. Serine, glycine and NADH production catalyzed by 3-phosphoglycerate dehydrogenase (PHGDH) is reduced. GCK: glucokinase; H6PD: hexose 6-phosphate dehydrogenase; UGGT1: UDP-glucose glycoprotein transferase 1; TKT: transketolase. PC: pyruvate carboxylase; TCA: tricarboxylic acid cycle. Created with BioRender.com.

Full size image

ER homeostasis. The synthesis of secreted and membrane-bound proteins is characterized by the co-translational folding of nascent proteins in the ER lumen and is driven by a tightly regulated set of N-linked glycosylation steps and chaperone proteins, and by their vesicular transport to the Golgi (Fig. 5). Proteins encoded by ALG2, STT3A, STT3B, PRKCSH, MLEC, CALR, and UGGT1 of the lectin-dependent folding process, PDIA3, PDIA6, ERO1L, TXNDC5, regulating disulfide-redox protein folding, and by PSMC2, PSMD2, PSMD4, PSMD6, the proteasome subunits associated with protein degradation of improperly folded proteins, were all more abundant in UGT2B17-deficient livers (Supplementary Table 1). An increased abundance of proteins in the secreted/membrane protein synthesis pathway might indicate an ER-stress response to unfolded proteins.

Fig. 5
figure 5

The ER quality control of membrane and secreted protein synthesis is globally impacted in the liver of UGT2B17-deficient individuals. Synthesis of secreted and membrane proteins is characterized by their co-translational folding in the ER lumen and is tightly regulated by chaperone proteins and N-linked glycosylation. N-linked glycosylation is initiated by the loading of a preassembled dolichol glycostructure (ALG2) onto context-specific asparagines of the nascent protein by the oligosaccharyltransferase OST (STT3A and STT3B). The trimming of initial glycostructure by glucosidases I and II (PRKCSH) enables the binding of folding chaperone lectins such as malectin (MLEC) and calreticulin (CALR). Folding of nascent proteins also requires protein disulfide isomerases such as PDIA3, PDIA6, ERO1L and TXNDC5 for proper disulfide bond formation. Properly folded proteins may then traffic to the Golgi by vesicle-mediated transport. The folding sensor UGGT1 regulates misfolded proteins so that they may interact again with chaperone lectins to engage another round of folding. Proteins that fail to fold properly undergo further trimming of the N-linked glycan to target them for protein degradation by the proteasome and the ER-associated lysosomes (ER-phagosomes). Multiple proteins upregulated in UGT2B17-deficient livers (red labels) are involved in this biological process. Created with BioRender.com.

Full size image

Immune functions. A considerable set of proteins with a role in immune-related pathways were significantly differentially expressed in UGT2B17-deficient livers, and included several of the most elevated proteins, by 8 to 19 -fold, namely COMMD3, NOX2, legumain (LGMN), α-actinin-2 (ACTN2). Proteins released by neutrophil and platelet degranulation, a process activated by inflammation, constituted an important group of proteins with increased abundance in deficient livers, including matrix metalloproteinase 9 (MMP9), neutrophil defensing 1 (DEFA1), vesicle-associated membrane protein 8 (VAMP8), galectin-3-binding protein (LGALS3BP), α-1-antichymotrypsin (SERPINA3) and the leukocyte elastase inhibitor (SERPINB1) (Fig. 6, Supplementary Table 1). Several of these proteins, as well as the chaperone calreticulin (CALR), and cathepsin S (CTSS), are also key components of the antigen processing-cross presentation pathway, all significantly more abundant in UGT2B17-deficient livers. The significant number and increased abundance of these proteins collectively provide evidence for an activated immune response and a state of inflammation in livers of UGT2B17 deficient individuals.

Fig. 6
figure 6

Activation of immune-related pathways in livers of UGT2B17-deficient individuals and summary of findings. (A) Neutrophil activation is characterized by the release of granule proteins at sites of cell damage or infection. Several proteins key to neutrophil activation and ER-associated degradation (ERAD) of misfolded damaged proteins were more abundant in UGT2B17-deficient livers. Note that this schematic representation is simplified from a highly complex process, to solely highlight the role of differentially expressed proteins in the context of immune pathways. Created with BioRender.com. (B) Differential abundance of NADPH oxidase 2 (NOX2) and neutrophil granule proteins. MMP9: matrix metalloproteinase 9; LYZ: lysozyme; DEFA1: defensin 1; PSAP: prosaposin; CTSS: cathepsin S in UGT2B17 proficient (+) and deficient (−) individuals. In boxplots, sex of liver donors is indicated (blue dots, males; red dots, females), the line represents median expression, “ + ” represents mean expression, and whiskers the range of values. *P < 0.05, **P < 0.01. (C) Graphical summary of findings.

Full size image

Discussion

The liver is crucial for maintaining metabolic homeostasis in the body and ensuring the proper functioning of various physiological processes. Disruption in these processes may lead to metabolic imbalances and health problems. Through this exploration of the hepatic protein profiles associated with a deficiency of the UGT2B17 metabolic pathway, we expose a significantly divergent liver proteome implying a critical role of UGT2B17 in maintaining liver metabolic functions (Fig. 6C). Remarkably, the significant number and increased abundance of many non-metabolic proteins involved in fundamental cellular pathways, such as the synthesis of secretory proteins and immune-related functions, highlight the pleiotropic aspects of a UGT2B17 deficiency that extend far beyond its function as a glycosyltransferase enzyme.

Using data-independent acquisition, this study offers a remarkably detailed quantitative profiling of proteins in unfractionated liver specimens of UGT2B17-deficient individuals, revealing the metabolic rewiring associated with this deficiency, compared to UGT2B17-proficient individuals. The majority of 233 impacted proteins were more abundant in UGT2B17-deficient livers, many of which are biochemical enzymes suggesting an enhanced metabolic activity. These proteins are involved in diverse pathways, including drug, steroid hormones, redox control, carbohydrate, and energy metabolism. These findings are consistent with recent untargeted metabolomics studies revealing reduced glucuronide-conjugated steroids and bile acids and increased levels of several circulating metabolites in healthy individuals carrying a complete UGT2B17 gene deletion and in UGT2B17-deficient cancer patients.1,33 As well, untargeted and targeted metabolomics profiling of the same liver tissues identified divergent metabolites, such as taurocholic acid.30,31,32.

As a detoxification organ, the liver expresses a wide array of ADME enzymes, including UGT2B17, which work together to regulate liver and overall body homeostasis. UGT2B17 deficiency had only a minor impact on the abundance of most other ADME enzymes, including CYPs, transporters and SULTs. Furthermore, UGT2B17 deficiency was not compensated by an increased abundance of any of the ten other UGT enzymes detected in the liver. In turns, a drastic change was noted for glutathione and methyl transferases, as well as for antioxidant enzymes such as those encoded by SOD2 and GPX2, arguing for an adaptation of redox control, a function that is currently undocumented for the UGT2B17 protein but has already been highlighted for other UGT proteins, such as the bilirubin-conjugating UGT1A1.34 A complementary analysis of RNA-seq data of a larger set of liver specimens including those of the current proteomics analysis suggests that a subset of UGT2B17-deficient individuals may also carry a GSTT1 gene deletion and/or GSTM1 gene retention, two other frequently deleted human genes.20 Given that these genes are not genetically linked (UGT2B17: chr4; GSTT1: chr22, GSTM1: chr1), the potential co-occurrence of gene deletion or retention in certain individuals warrants further investigation. The limited sample size of our study did not permit to clarify this possibility.

We found that the UGT2B17 deficiency may have an impact on key cellular signaling pathways. Individuals with complete UGT2B17 gene deletion have elevated levels of circulating androgens, reflecting the UGT2B17 enzymatic activity towards major androgens like testosterone that serves as a precursor to estradiol.1,4,33 Here, we observed a translational adaptation of the hepatic sex steroid metabolic pathway favoring the biosynthesis of the biologically active estradiol, sustained by the enhanced expression of 17β-HSD7 and 17β-HSD12 enzymes, two ER-resident enzymes that both catalyze the NAD+-dependent reduction of estrone into estradiol.35,36 In contrast, none of the enzymes involved in the primary metabolism of bile acids, which are also derived from cholesterol, were affected. Metabolic perturbations of androgens, progestins and lipids may also occur given that 17β-HSD7 is involved in the biotransformation of hormone receptor ligands, the conversion of androgenic DHT and progesterone into inactive metabolites (5α-androstane-3β-17β-diol and 4-pregen-3β-ol-20-one respectively) and participates in cholesterol biosynthesis,37 whereas 17β-HSD12 is a key enzyme in the elongation of omega-6 polyunsaturated fatty acid for the synthesis of arachidonic acid in the ER (reviewed in35). The sphingolipid pathway could also be rewired as we observed an increased abundance of the hepatic acid ceramidase ASAH1 and β-galactosidase GLB1 enzymes in UGT2B17-deficient livers. These observations are fully in keeping with the divergent systemic steroidome and sphingolipid profiles recently observed in healthy UGT2B17-deficient individuals and those with cancer.1,33,38.

Unexpectedly, our observations expose a critical role of the ER-resident UGT2B17 enzyme in the maintenance of ER functions beyond detoxification discussed above, and include carbohydrate and lipid metabolism, protein synthesis and transport, and glycosylation. We raise the concept that a UGT2B17 deficiency affects multiple biochemical pathways that sufficiently impair metabolic homeostasis to trigger ER-stress responses. The ER is a vital metabolic hub involved in anabolic and catabolic processes, including protein, hormone, drug metabolism as well as lipid and glucose-related metabolism and storage.39 ER thus acts as a metabolic sensor in which stress responses are activated upon metabolic dysfunctions. Several pieces of evidence support the critical role of UGT2B17 in ER-stress responses.

First, an energy deficiency is anticipated given the divergent expression of multiple enzymes regulating glucose metabolism. Notably, the reduced abundance of glucokinase, coupled with higher hexose-6-phosphate dehydrogenase levels suggest a rewiring of glucose into the pentose phosphate pathway, crucial for nucleotide synthesis and NADPH production. This might reflect an adaptation to favor the maintenance of NADPH pools, an essential co-factor in hundreds of biochemical reactions, at the expense of other critical pathways including energy production and nucleotide-activated sugar synthesis. Consequently, metabolic transactions dependent on glycolytic derivatives are likely affected by the UGT2B17 deficiency, including fatty acid and nucleotide synthesis, but also NADPH pools and glycoconjugates.

Second, a dysregulated secreted protein synthesis in UGT2B17-deficient livers is hinted from the globally enhanced expression of many proteins orchestrating secretory/membrane protein synthesis, folding and degradation. The synthesis of secretory/membrane proteins occurring in the ER, highly relies on glycoconjugates and N-linked glycosylation. As a critical metabolic sensor, this may indicate an ER-stress response to impaired cellular energy and redox control, prevalent in UGT2B17-deficient livers, due to the increased abundance of ER-quality control proteins, potentially triggering the unfolded protein response.

Third, enhanced expression of the protein folding chaperones calreticulin and protein disulfide isomerase 3 not only mediates the protein folding process in the ER, they also constitute two key damage-associated molecular pattern (DAMP) ER proteins. DAMPs are endogenous peptides released by stressed, damaged or dying cells, triggering immunogenic cell death and immune responses to favor clearance of cell debris.40 There is an important interplay between the clearance of defective proteins and damaged cells and immune pathways, both of which appear activated in UGT2B17-deficient livers. The more abundant expression of ER-quality control and neutrophil/platelet granule proteins lends a strong support to an activation of immune responses for the clearance of damaged hepatocyte cells in UGT2B17-deficient livers by recruiting neutrophils, crucial for the resolution of immunogenic cell death. Because neutrophil activation also generates a high level of oxidative stress, it may contribute to the upregulation of antioxidant enzymes discussed above.

The increased abundance of proteins involved in chronic liver inflammation, including (i) antioxidant defense such as glutathione peroxidase 2, biliverdin reductase, and superoxide dismutase 2, (ii) proteins associated with liver fibrosis such as matrix metalloproteinases, as well as (iii) key components of neutrophil degranulation, support the notion that an inflammatory condition of the liver may prevail in UGT2B17-deficient individuals. This hypothesis is further supported by elevated taurocholic acid levels in UGT2B17-deficient hepatic tissues that are linked with liver toxicities, and the higher circulating levels of specific metabolites including lysophospholipids and plasmalogens in UGT2B17-deficient healthy individuals, evidenced by untargeted metabolomics.33 Consistent with our findings, a recent study reported that low protein expression of UGT1A1 may be associated with the progression of liver damage in patients with acute chronic liver failure.41.

It is still unknown whether these observations may be a direct result of a UGT2B17 gene deficiency. Indeed, our study was limited by a small sample size (n = 15) although it enabled the quantification of over 3,000 proteins, with 233 showing differential expression. A recent exploratory study with an even smaller sample size (n = 6), identified 16 differentially expressed hepatic proteins by data-dependent acquisition mass spectrometry.42 This study, conducted in S9 subcellular fractions, enriched for enzyme systems, from UGT2B17 proficient and deficient individuals, identified that 12 out of the 16 altered proteins (75%) were more abundant, which aligns with our observations that extend to numerous other proteins. All samples were of normal histological appearance as confirmed by the pathology, and included similar numbers of adult male and female donors in each group. The limited number of samples constitute a limitation of the study, given that levels of hepatic enzymes may be affected by the overall liver health, or by sex and age, as documented for UGT2B17.24 It is also well-documented that a complex array of genetic and environmental factors influence liver homeostasis, such as other genetic variants and metabolic gene deficiencies, diet and nutrition, medication and lifestyle habits. Therefore, the findings of this exploratory study warrant investigations in larger cohorts to better define the direct impact of the frequent complete UGT2B17 deficiency in populations.

In conclusion, this study exposes the multifaceted effect of UGT2B17 deficiency on liver functions and is fully consistent with an altered circulating metabolome and associated health conditions.33 The characteristic hepatic proteome of UGT2B17-deficient individuals appears to reflect an adaptive response to important metabolic changes and a chronic ER-stress (Fig. 6C). The far-reaching effects of a complete UGT2B17 deletion on cellular metabolism, protein expression and immune responses have likely implications for disease mechanisms, biomarker discovery, and therapeutic intervention. The molecular mechanisms by which a UGT2B17 deficiency might trigger this collection of adaptive responses remain to be addressed.