Inactivation of Prostaglandin E2 as a Mechanism for UGT2B17-Mediated Adverse Effects in Chronic Lymphocytic Leukemia

High expression of the metabolic enzyme UDP-glucuronosyltransferase UGT2B17 in chronic lymphocytic leukemia (CLL) cells was associated with poor prognosis in two independent studies. However, the underlying mechanism remains unknown. We hypothesized that UGT2B17 impacts intracellular levels of hormone-like signaling molecules involved in the regulation of gene expression in leukemic cells. We initially confirmed in a third cohort of 291 CLL patients that those with high UGT2B17 displayed poor prognosis (hazard ratio of 2.31, P = 0.015). Consistent with the unfavorable prognostic significance of elevated UGT2B17 expression in CLL patients, high UGT2B17 expression was associated with enhanced proliferation of MEC1 and JVM2 malignant B-cell models. Transcriptomic analyses revealed that high UGT2B17 was linked to a significant alteration of genes related to prostaglandin E2 (PGE2) and to its precursor arachidonic acid, both in cell models and a cohort of 448 CLL patients. In functional assays, PGE2 emerged as a negative regulator of apoptosis in CLL patients and proliferation in cells models, whereas its effect was partially abrogated by high UGT2B17 expression in MEC1 and JVM2 cells. Enzymatic assays and mass-spectrometry analyses established that the UGT2B17 enzyme inactivates PGE2 by its conjugation to glucuronic acid (GlcA) leading to the formation of two glucuronide (G) derivatives. High UGT2B17 expression was further associated with a proficient inactivation of PGE2 to PGE2-G in CLL patient cells and cell models. We conclude that UGT2B17-dependent PGE2 glucuronidation impairs anti-oncogenic PGE2 effects in leukemic cells, thereby partially contributing to disease progression in high UGT2B17 CLL patients.

High expression of the metabolic enzyme UDP-glucuronosyltransferase UGT2B17 in chronic lymphocytic leukemia (CLL) cells was associated with poor prognosis in two independent studies. However, the underlying mechanism remains unknown. We hypothesized that UGT2B17 impacts intracellular levels of hormone-like signaling molecules involved in the regulation of gene expression in leukemic cells. We initially confirmed in a third cohort of 291 CLL patients that those with high UGT2B17 displayed poor prognosis (hazard ratio of 2.31, P = 0.015). Consistent with the unfavorable prognostic significance of elevated UGT2B17 expression in CLL patients, high UGT2B17 expression was associated with enhanced proliferation of MEC1 and JVM2 malignant B-cell models. Transcriptomic analyses revealed that high UGT2B17 was linked to a significant alteration of genes related to prostaglandin E2 (PGE 2 ) and to its precursor arachidonic acid, both in cell models and a cohort of 448 CLL patients. In functional assays, PGE 2 emerged as a negative regulator of apoptosis in CLL patients and proliferation in cells models, whereas its effect was partially abrogated by high UGT2B17 expression in MEC1 and JVM2 cells. Enzymatic assays and mass-spectrometry analyses established that the UGT2B17 enzyme inactivates PGE 2 by its conjugation to glucuronic acid (GlcA) leading to the formation of two glucuronide (G) derivatives. High UGT2B17 expression was further associated with a proficient inactivation of PGE 2 to PGE 2 -G in CLL patient cells and cell models. We conclude that UGT2B17-dependent PGE 2 glucuronidation impairs anti-oncogenic PGE 2 effects in leukemic cells, thereby partially contributing to disease progression in high UGT2B17 CLL patients.

INTRODUCTION
Chronic lymphocytic leukemia (CLL) is the most common form of adult leukemia in the western world (1). It is characterized by an accumulation of mature B-cells in peripheral blood, bone marrow, and other lymphoid organs. CLL has a highly variable course and generally develops at a slower pace than other leukemia subtypes. The accumulation of CLL cells in patients can be explained, in part, by defects in the regulation of apoptosis (2), but other studies have also shown evidence of clonal evolution (3,4) and the importance of active proliferation in progressive disease (5,6).
A significant increase in therapeutic options for CLL patients has been associated with improved survival. However, CLL remains incurable (7). Recent investigations have yielded insights into molecular indicators used to better predict clinical outcome of this disease. Of those, uridine diphosphoglucuronosyltransferase 2B17 (UGT2B17) expression was identified as a novel molecular marker for CLL progression in two independent CLL cohorts (8,9). The initial report associated high UGT2B17 mRNA expression with poor prognosis, unmutated-IGHV and other expression-based markers, such as LPL and CD38 (9). The second study reported high UGT2B17 expression as a prognostic marker particularly for mutated-IGHV individuals, a subgroup of patients for which few indicators of progression currently exist (8). The UGT2B17 gene encodes a member of a superfamily of metabolic enzymes responsible for the conjugation of small lipophilic substrates to glucuronic acid (GlcA) derived from its co-substrate UDP-GlcA (10). Glucuronidation inactivates the enzyme's substrates, increases their polarity and facilitates their elimination through bile or urine. The function of this metabolic route is to maintain homeostasis of endogenous molecules while protecting cells from potentially harmful chemicals arising from exogenous sources, including pharmacological compounds (11).
The underlying mechanism(s) by which the UGT2B17 protein may affect CLL malignancy and disease progression in patients remains unknown. Amongst the 19 human UGT isoforms, UGT2B17 is the only significantly expressed UGT in CLL cells. The UGT2B17 protein is enzymatically functional, as it was shown to conjugate UGT2B17 substrates such as androgens in cells isolated from CLL patients (9). It is plausible that UGT2B17 influences intracellular levels of hormone-like signaling molecules involved in the regulation of gene expression, with subsequent impacts on cancer cell growth and survival. We hypothesized that overexpression of UGT2B17 perturbs the bioavailability and response to endogenous B-cell modulators with consequences on the transcriptome of leukemic cells and neoplastic behavior. We report that high UGT2B17 modifies expression of genes related to prostanoids. It also impairs prostaglandin-mediated growth inhibition of malignant cells through direct inactivation of these hormone-like molecules by glucuronidation.

Patients, Cell Lines, and Culture
The B-cell neoplastic cell lines MEC1 and JVM2 were purchased from DSMZ (Braunschweig, Germany) in 2010 and ATCC (Manassas, VA) in 2015, respectively. The MEC1-2B17 and JVM2-2B17 cell models were generated by electroporation of parental cells with a pcDNA6 vector containing the coding sequence of UGT2B17 for stable overexpression. Electroporation was achieved with the Neon Transfection System (Thermo Fisher scientific, Waltham, MA). The JVM2-CTRL cells were produced by electroporation with the pcDNA6 vector and the MEC1-CTRL cells were prepared as described (9). All cell lines were cultured in RPMI medium supplemented with 10% fetal bovine serum (FBS), 1% penicillin/streptomycin, 1% sodium pyruvate and 1% L-glutamine. Selection of MEC1-2B17 cells was achieved by supplementing culture medium with 20 µg/mL blasticidin for 3 weeks, after which selection was maintained by growing cells with 10 µg/mL of blasticidin. MEC1-CTRL cells were supplemented with 3 µg/mL of puromycin. For JVM2-CTRL and JVM2-2B17 cells, selection for 3 weeks and subsequent growth was achieved by supplementing medium with 5 µg/mL blasticidin. All cell culture components were purchased from Wisent Bioproducts (St-Bruno, QC). Cells were regularly tested for mycoplasma, with the most recent carried out on February 9th, 2019.
Cryopreserved peripheral blood mononuclear cells (PBMCs) from 15 CLL patients diagnosed between 1987 and 2011 at Vienna General Hospital were used (Table S1). Cell purity was assessed by measuring CD5 and CD19 surface expression by cytometry prior to experimentation, with CLL cells representing 67% of total PBMCs on average. Primary cells from CLL patients were cultured in RPMI supplemented with 10% FBS, 1% sodium pyruvate, and 1% L-glutamine without antibiotics. All subjects gave written informed consent in accordance with the Helsinki Declaration and the study was evaluated and approved by local Ethical Research Committees of the Medical University of Vienna (Ethics vote 1499/2015) and the Center Hospitalier Universitaire (CHU) de Québec (A14-10-1205).

Immunoblotting
For western blot, microsomal fractions were prepared from MEC1 and JVM2 by resuspending 10 8 cells in 1 mL microsome buffer (4 mM potassium phosphate, 20% glycerol, pH 7.0). Cells were sonicated three times for 30 s, alternating with 30-s pauses on ice. Extracts were centrifuged twice at 12,000 × g for 22 min at 4 • C. Supernatants were centrifuged at 105,000 × g for 60 min at 4 • C to isolate the microsomal fraction that was resuspended in microsome buffer containing 0.5 mM dithiothreitol (DTT). Microsomal fractions (20 µg) were mixed with Laemmli sample buffer (Bio-Rad, Mississauga, ON), heated at 95 • C for 5 min prior to SDS-PAGE and transfered to nitrocellulose membranes. Protein detection was adapted from a previously described immunoblotting strategy (12) using the polyclonal anti-UGT2B antibody EL-93 (1:2,000) for detection of UGT2B17 (13) or an anti-calnexin antibody (1:2,000, Enzo Life Science, Farmingdale, NY) as a loading control.

Gene Expression Analysis
Publicly available data sets were used for analysis of overall survival (OS) of CLL patients and CLL expression profiles in relation to UGT2B17 levels using the affy (v1.48) and limma (v3.26.9) packages for R (http://www.bioconductor.org). The first dataset was obtained from the International Cancer Genome Consortium (ICGC-CLLE-ES, n = 291) (15,16). A second dataset was from the Gene Expression Omnibus (GEO-GSE13159, n = 448 untreated patients) (17). For MEC1 and JVM2 cell models, total RNA from three biological replicates was extracted using RNeasy mini kit, as per manufacturer's instructions (Qiagen, Toronto, ON). Samples were subjected to ribosomal RNA depletion before Illumina HiSeq2000 paired-end sequencing at Genome Québec McGill University and at the CHU de Québec Research center -Université Laval. Raw data was processed using the MUGQIC pipeline version 1.3. Briefly, reads were quality trimmed and aligned to the hg38 human genome. Differential gene expression analysis was performed using the edgeR and DESeq2 tools for R v3.2.2 (18,19). Differences in gene expression were considered significant if Benjamini-Hochberg adjusted P-values for both tools were below 0.05. Analysis of enriched biological pathways was carried out using g:profiler (20) with the Reactome (21), KEGG (22)(23)(24), and GO biological process databases. Reactome FI plugin for Cytoscape v3.2.1 was also used for clustering of genes into modules and visualization of enriched pathways (25). Altered expression of selected genes was validated by quantitative real time PCR. Briefly, total RNA was DNase I-treated and purified using the RNeasy MinElute Cleanup kit (Qiagen, Hilden, DE) following the manufacturer's instructions. First-strand cDNA synthesis was accomplished using Superscript IV RNase H-RT (Invitrogen Life Technologies, Burlington, ON, CA), PCR purification kit (Qiagen, Hilden, DE) was used to purify cDNA. Oligoprimer pairs were designed by GeneTool 2.0 software (Biotools Inc, Edmonton, AB, CA), their specificity was verified by BLAST alignment to human RefSeq sequences and were synthesized by IDT (Integrated DNA Technology, Coralville, IA, USA) ( Table S2). RT-qPCR quantification was carried out using the LightCycler 480 (Roche Diagnostics, Mannheim, DE). Reagent LightCycler 480 SYBRGreen I Master (Roche Diagnostics, Indianapolis, IN, USA) was used as described by the manufacturer with 2% DMSO. Relative quantity was calculated using the second derivative method and by applying the delta Ct method (26). B2M, HPRT1, and UBC were used as reference genes for normalization (27). Quantitative qPCR measurements were performed in compliance with MIQE guidelines by the Gene Expression Platform of our institution (28,29).

Cell Proliferation and Viability Assays
Cells were plated at 1 × 10 4 cells/well (MEC1) or 5 × 10 4 cells/well (JVM2) in 96-well U-bottom tissue culture plates (BD Bioscience, Mississauga, ON). For treatment assays, growth media was supplemented with PGs or vehicle (ethanol) at time of plating and renewed every 48 h. PGE 2 , Butaprost and PGE 1 -OH, used at concentrations indicated in the text, were purchased from Cayman chemicals (Ann Arbor, MI). Every 24 h, an aliquot of cells was stained with trypan blue (50%) and counted with a TC-10 automated cell counter (Bio-Rad). Assays were replicated at least three times in duplicate. For MTS assays, 20 µL of CellTiter aqueous one solution cell proliferation reagent (MTS Promega, Madison, WI, USA) was added to 100 µL of cells. Absorbance at 490 nm was read after a 4-h incubation at 37 • C. To further assay cell proliferation, cells were labeled with 5 µM of CFSE (Thermo Fisher Scientific) and incubated at 37 • C for 10 min before rinsing once with DPBS and plating at 4 × 10 6 cells/mL in 12-well culture plates with whole medium 72 h prior to analysis.

Migration Assays
CLL PBMCs were plated at 3 × 10 6 cells/mL in 24-well culture plates 24 h prior to migration assays. Cells were then treated for 24 h with PGE 2 (1-5 µM) before initiation of migration experiments. Transwell 5 µm-pore inserts (Corning, New-York, NY) were then placed in a 24-well plate with 600 µL of medium in the lower well and 100 µL in the upper well, containing 5 × 10 5 CLL cells. Cells were treated with either vehicle (PBS) or CXCL12 (200 ng/mL) before incubation for 4 h at 37 • C. Cells in the upper and lower compartments were recovered in separate tubes by incubation with a non-enzymatic cell dissociation solution (Sigma-Aldrich, Oakville, ON) for 10 min at 37 • C. Each tube was then spiked with 50 µL of 123eBeads (Thermo Fisher Scientific) and counted using a FACS Canto II flow cytometer (BD Bioscience). Migration was calculated with the following formula: % Migration = Cells in lower chamber Cells in lower chamber + Cells in upper chamber × 100

Flow Cytometry Analyses
Each patient PBMC sample was thawed and diluted to 3 × 10 6 cells/ml prior to identification of CLL cells by analysis of cell surface markers using PerCPCy5.5-conjugated anti-human CD5 (BD Pharmingen), PE-Cy7-conjugated anti-human CD19 (BD Bioscience) and APC-Cy7-conjugated anti-human CD45 (BD Bioscience). All analyses were conducted with a FACS Canto II flow cytometer (BD Bioscience).

Cell Death
Aliquots of 5 × 10 5 cells were centrifuged at 520 × g and rinsed twice with Dulbecco's Phosphate Buffered Saline (DPBS) before resuspension in Annexin V binding buffer (50 mM HEPES, 700 mM NaCl, 12.5 mM CaCl 2 , pH 7.4). Cells were labeled with Alexa Fluor 647-conjugated Annexin V as per manufacturer's instructions (Thermo Fisher scientific) and propidium iodide (PI; 4 ng/mL) in the dark, on ice, for 30 min prior to analysis.

Statistical Analysis
The half maximal inhibitory concentrations (IC 50 ) were calculated by fitting variable slope non-linear curves to normalized response data from PGE 2 treatments. Statistical analysis was carried out using a two-tailed Student's t-test, unless otherwise indicated in the legends. Analysis of overall survival was done using the Kaplan-Meier method and the Log-Rank test. Statistics were performed using GraphPad Prism v5 (GraphPad Software Inc., La Jolla, CA) and R v3.2.2. Statistical significance is defined as * P < 0.05, * * P < 0.01. Each experiment was performed with three biological replicates, unless otherwise indicated in figure legends.

UGT2B17 Confers a Proliferative Advantage to Malignant B-Cell Models
A significant reduction in overall survival (OS) was established in 291 CLL patients from the International Cancer Genome Consortium (ICGC) expressing high UGT2B17 levels ( Figure 1A). Neoplastic B-cell models MEC1 and JVM2 stably overexpressing UGT2B17 were created to examine its role in CLL progression. These two cell lines were chosen on the basis of relatedness to CLL and diversity of cytogenetic aberrations. Overexpression was confirmed at the mRNA (2.6fold) and protein (2.6-fold) levels in the MEC1-2B17 cells, and the functionality of the enzyme was supported by an enhanced glucuronidation activity for characteristic UGT2B17 substrates vorinostat (SAHA), testosterone (Testo), dihydrotestosterone (DHT), and estradiol (E 2 ) by 1.6-1.7-fold (P < 0.05), compared to control cells levels (Figures 1B-D). In the JVM2-2B17 model, a 2.5-fold enhanced mRNA expression and 2.2-fold higher protein levels were observed with a corresponding 2.1-2.5fold enhanced UGT2B17 activity (P < 0.05) (Figures 1B-D). Consistent with the unfavorable prognostic significance of UGT2B17 expression in CLL patients (above) (8,9), high UGT2B17 expression was associated with enhanced proliferation by 1.7 (P < 0.05) and 2.0-fold (P < 0.01) for MEC1 and JVM2, respectively (Figures 1E,F). A significantly shorter doubling time for high UGT2B17 expressing cells was evidenced in cell lines by cell counts, MTS assays, and CFSE labeling (Figures 1E-I).
No significant difference in cell viability was noted by Annexin V/PI and trypan blue exclusion assays when cells were grown in either basal or serum starvation conditions (data not shown). Robustness of cellular phenotypes was measured periodically for several months to confirm stability of UGT2B17 cell models.

UGT2B17 Modulates Expression of Genes Related to Prostanoids in Cell Models and Leukemic Cells From Patients
RNA sequencing revealed a strong effect of elevated UGT2B17 on global gene expression. A total of 5474 and 2880 genes were differentially expressed in MEC1-2B17 and JVM2-2B17 cells, respectively, when compared to control cells (FDR < 0.05) (Figure 2A). Of those, 683 modulated genes were concordant between models, among which 272 had an absolute fold-change (FC) > 1.3 (FDR < 0.05) in UGT2B17 overexpressing cells (Table S3). Pathway analyses using the Reactome and KEGG databases revealed biological pathways overrepresented among common modulated genes such as cell adhesion, chemokine signaling, antigen processing, translation, apoptosis, antigenreceptor signaling as well as steroid and lipid signaling pathways ( Figure 2B). The latter was well-represented with 54 genes connected to the metabolism and biosynthesis of lipids or eicosanoids (prostanoids and leukotrienes) significantly modified by UGT2B17 (Table S4). This is plausible given that several bioactive lipids are glucuronidated by the UGT pathway (30). This set of genes was selected for further investigation. Genes involved in arachidonic acid metabolism such as FADS1, FADS2, and ACSL4 were significantly down-regulated in both cell models (Figures 2C,D; Table S4). The leukotriene biosynthesis genes ALOX5 and LTA4H as well as genes from prostaglandin-related pathways were also differentially expressed. This included lower expression of the genes encoding the prostaglandin (PG) E receptors PTGER2 and PTGER4, higher expression of the PGinactivating enzyme PTGR2, as well as decreased expression of the prostanoid biosynthesis enzymes PTGES2 and CBR1. Quantitative PCR confirmed these observations (Figure 2E).
Their clinical relevance was supported by the altered expression of these genes in CLL patients expressing high levels of UGT2B17 in a cohort of 448 cases (GSE13159) dichotomized on the basis of median UGT2B17 expression ( Figure 2E). Notably, genes of the PG biosynthesis pathway and PG receptors were significantly down-regulated in CLL patients with high UGT2B17 expression ( Figure 2E). Also, the expression of genes coding for membrane transporters known to mediate PG influx such as SLCO3A1 and SLCO4A1 was reduced in CLL patients with high UGT2B17, while the PG efflux transporter gene ABCC4 was enhanced (not shown). Pathway enrichment analysis of genes commonly altered in both MEC1 and JVM2 further revealed that response to PGE 2 was a significantly enriched pathway (FDR = 0.032). According to these observations coherent for CLL patients and cell models, the effect of PGE 2 on B-cell phenotypes and the influence of UGT2B17 expression were investigated further.

PGE 2 Increases Cell Death and Inhibits Migration of Primary CLL Patient Cells
PBMCs from 15 CLL patients were analyzed following treatment with PGE 2 for potential differences in cell death, activation and migration by flow cytometry. Sample purity was assessed by analysis of CD5, CD19, and CD45 expression. A trend toward increased cell death was observed for primary CLL cells treated for 24 h with 5 µM PGE 2 (P = 0.067). By 48 h of treatment, cell death was significantly enhanced by PGE 2 (P = 0.0007), suggesting that PGE 2 promotes cell death of CLL cells  ( Figure 3A). CLL cells showed a trend for inhibition of CXCL12directed migration by 5 µM PGE 2 when compared to cells treated only with CXCL12 ( Figure 3B; P = 0.102). PGE 2 did not alter the activation of PBMCs triggered by IgM stimulation, which was assessed with CD80 and pERK markers (Figures 3C,D). Based on these data, potential anti-oncogenic effects of PGE 2 and possible interactions with UGT2B17 were investigated further using in vitro cell models of B-cell malignancies.

High UGT2B17 Expression Hinders PGE 2 -Mediated Growth Inhibition in Cell Models
MEC1 cells treated with PGE 2 or with synthetic agonists of the PGE 2 receptors EP 2 (butaprost) or EP 4 (PGE 1 -OH) showed considerable growth inhibition ( Figure 4A). MEC1 cells exposed to increasing physiologically relevant concentrations of PGE 2 (1-20 µM) further supported a repressive effect of PGE 2 on B-cell growth, with an inhibition profile significantly different for MEC1-2B17 compared to control cells (Figure 4B). High UGT2B17 significantly decreased the responsiveness of B-cells to PGE 2 , evidenced by a 1.5-fold higher half maximal inhibitory concentration (IC 50 ) of 12.4 vs. 8.2 µM for high vs. low expressing cells (P = 0.0006). We then tested the effects of other PGs. In these experiments, the level of UGT2B17 expression affected the inhibitory effect of PGE 2 by up to 33% (P ≤ 0.022) but not of the other PGs (Figures 4C,D).

UGT2B17 Inactivates PGE 2 and Other Related PGs
PGs share a lipid backbone and comprise several functional groups (hydroxyl and carboxyl groups) that may be susceptible to conjugation with GlcA by the UGT pathway ( Figure 5A). Amongst all 19 human UGT enzymes, UGT2B17 largely predominates in human leukemic B-cells, although with significant variability (CV UGT2B17 = 242.9%; n = 291 CLL patients) (Figure 5B). We investigated the possibility that PGE 2 may be a substrate of UGT2B17. The initial series of experiments revealed the formation of two glucuronide (G) derivatives (PGE 2 -G1 and PGE 2 -G2) detected by MS in MRM mode, upon incubations of PGE 2 with MEC1-2B17 microsomes ( Figure 5C). The patterns of fragmentation of individual glucuronides were consistent with the loss of the GlcA moiety (molecular mass of 176) ( Figure 5C) and support UGT2B17 targeting two of the functional groups on the PGE 2 molecule. Enzymatic assays with PGE 2 further established that primary CLL samples and cell lines expressing high UGT2B17 levels produced significant amounts of PGE 2 -G1 and PGE 2 -G2 derivatives whereas they  were not detected in those expressing low UGT2B17 levels in CLL patients ( Figure 5D).

DISCUSSION
CLL is a chronic disease characterized by a high degree of heterogeneity, with complex clonal dynamics resulting in an imbalance between B-cell proliferation and death that favors an accumulation of leukemic cells during progressive disease (5). We demonstrated that high UGT2B17 expression is associated with adverse outcome in a cohort of 291 CLL patients, consistent with two independent previous reports (8,9). UGT2B17 is the only appreciably expressed UGT in CLL cells, although heterogeneously expressed. Our work revealed that high UGT2B17 expression confers a growth advantage to malignant B-cells that is mediated, at least in part, through significant changes in gene expression in PGE 2 -related pathways and its inactivation by the UGT2B17 enzyme. This enhanced proliferation rate results in a significantly larger B-cell population only after several days in culture, consistent with the slow rate of B-cell accumulation in CLL. PGE 2 emerges as a potential regulator of B-cell growth and apoptosis, a notion supported by previous reports in mouse models and in human B cells (31)(32)(33)(34)(35)(36). Similarly, our analysis of primary cells from CLL patients exposed to physiological concentrations of PGE 2 revealed an enhanced cell death after 48 h and a slight inhibition of CXCL12-mediated migration. Furthermore, in CLL patients' samples expressing high UGT2B17 levels, we also observed PGE 2 inactivation but not in those with low UGT2B17 levels. One mechanism by which UGT2B17 may modify cancer cell behavior may be through inactivation of PGE 2 , thereby altering homeostatic levels of PGE 2 and gene expression related to PGE 2 synthesis, transport and action. PGE 2 is a key lipid hormone-like signaling molecule of the eicosanoid family, and plays important roles in disease-associated inflammation and normal physiological functions (37). PGE 2 is produced from a cascade of enzymatic modifications to arachidonic acid (Figure 5), a polyunsaturated fatty acid esterified in cell membrane phospholipids and a previously reported substrate of UGT2B7 (not expressed in B-cells) leading to its inactivation in the liver (38). Two PGE 2 glucuronide derivatives were produced by UGT2B17 in leukemic cells, consistent with the conjugation of hydroxyl groups that corresponds to the preferred substrate moieties of the enzyme (39), whereas the hepatic UGT2B7 enzyme may preferentially target the carboxyl group (40). Analysis of PGE 2 -G by nuclear magnetic resonance will be required to confirm this reasoning. PGE 2 displays tumor suppressor functions in B-cells and other immune cells by impairing BCR activation and cell proliferation, as well as by triggering cell death (31,35). PGE 2 signals through different types of E-series PG receptors (EPs), namely EP 1 , EP 2 , EP 3 , and EP 4 (41). Past studies in murine cells have demonstrated inhibition of B-cell activation by PGE 2 through binding and activation of EP 2 and EP 4 (31,32,36,42). EP 4 was shown to be a negative feedback regulator of B-cell activation via the BCR signaling cascade, which is an integral pathway in CLL (31). Our findings that PGE 2 receptor agonists mimicked the growth-inhibition induced by PGE 2 support a receptormediated mechanism of action in human neoplastic B-cells. The expression of genes related to PGE 2 signaling, including the EP receptors, was reduced in CLL patient cells and B-cell models with high UGT2B17 expression, likely further disrupting PGE 2 homeostasis. The mechanism by which UGT2B17 modulates PGE 2 -related gene expression remains to be addressed, but likely involves regulatory PGE 2 feedback loops. Together, this altered expression profile combined with the direct PGE 2 inactivation by UGT2B17 support the notion that CLL cells with high UGT2B17 are less responsive to PGE 2 , thus conferring a proliferative advantage that could influence CLL disease course.
Our results provide a first potential mechanism for understanding the role UGT2B17 in CLL by circumventing the actions of endogenous signaling factors such as PGE 2 . This study does have limitations, namely the use of two cell models imperfectly representing CLL, the variability in CLL sample purity and limited number of available patients to investigate the correlation between UGT2B17 expression levels and PGE 2 response in detail. UGT2B17-dependent glucuronidation of additional endogenous effectors remain to be identified and may contribute to divergent expression profiles and phenotypes reported here for low and high UGT2B17 expressing B-cells.
In conclusion, our findings imply a relevant anti-oncogenic function of PGE 2 in CLL cells that is blocked by high UGT2B17 expression, through a direct metabolic inactivation of PGE 2 and altered PGE 2 -related gene expression leading to PGE 2 unresponsiveness.

DATA AVAILABILITY
The datasets generated during the current study are available in the Gene Expression Omnibus repository with the accession number GSE121626.

ETHICS STATEMENT
All subjects gave written informed consent in accordance with the Helsinki Declaration and the study was evaluated and approved by local Ethical Research Committees of the Medical University of Vienna (Ethics vote 1499/2015) and the Center Hospitalier Universitaire (CHU) de Québec (A14-10-1205).

AUTHOR CONTRIBUTIONS
EA, LV, PC, and VT performed the experiments. EA, MR, LV, EL, and CG analyzed the data. EA performed bioinformatics analyses. EA, MR, EL, and CG designed the research. TL and KV provided patient samples, collected clinical data, and revised the manuscript. EA, MR, and CG wrote the manuscript.

FUNDING
The Cancer Research Society, the Leukemia and Lymphoma Society of Canada, the Canadian Institutes of Health Research (FRN-152986), and the Canada Research Chair Program financially supported this study.