Skip to main content
NCBI home page
NIHPA Author Manuscripts logoLink to NIHPA Author Manuscripts
. Author manuscript; available in PMC: 2023 Jul 1.
Published in final edited form as: Cancer Lett. 2022 Oct 27;552:215981. doi: 10.1016/j.canlet.2022.215981

ENT1 blockade by CNX-774 overcomes resistance to DHODH inhibition in pancreatic cancer

Nicholas J Mullen 1, Ravi Thakur 1, Surendra K Shukla 2, Nina V Chaika 1, Sai Sundeep Kollala 1, Dezhen Wang 1, Chunbo He 2, Yuki Fujii 2, Shikhar Sharma 2, Scott E Mulder 1, David B Sykes 3, Pankaj K Singh 1,2,4,5,*
PMCID: PMC10305837  NIHMSID: NIHMS1907807  PMID: 36341997

Abstract

Inhibitors of dihydroorotate dehydrogenase (DHODH), a key enzyme for de novo synthesis of pyrimidine nucleotides, have failed in clinical trials for various cancers despite robust efficacy in preclinical animal models. To probe for druggable mediators of DHODH inhibitor resistance, we performed a combination screen with a small molecule library against pancreatic cancer cell lines that are highly resistant to the DHODH inhibitor brequinar (BQ). The screen revealed that CNX-774, a preclinical Bruton tyrosine kinase (BTK) inhibitor, sensitizes resistant cell lines to BQ. Mechanistic studies showed that this effect is independent of BTK and instead results from inhibition of equilibrative nucleoside transporter 1 (ENT1) by CNX-774. We show that ENT1 mediates BQ resistance by taking up extracellular uridine, which is salvaged to generate pyrimidine nucleotides in a DHODH-independent manner. In BQ-resistant cell lines, BQ monotherapy slowed proliferation and caused modest pyrimidine nucleotide depletion, whereas combination treatment with BQ and CNX-774 led to profound cell viability loss and pyrimidine starvation. We also identify N-acetylneuraminic acid accumulation as a potential marker of the therapeutic efficacy of DHODH inhibitors. In an aggressive, immunocompetent pancreatic cancer mouse model, combined targeting of DHODH and ENT1 dramatically suppressed tumor growth and prolonged mouse survival. Overall, our study defines CNX-774 as a previously uncharacterized ENT1 inhibitor and provides strong proof of concept support for dual targeting of DHODH and ENT1 in pancreatic cancer.

Keywords: Pancreatic Cancer, Nucleotide Metabolism, DHODH inhibitor, Therapy resistance, Nucleoside transporter

INTRODUCTION

Pancreatic ductal adenocarcinoma (PDAC) is a devastating disease that afflicts more than 56,000 Americans each year, resulting in over 45,000 deaths [1, 2]. The shocking lethality of PDAC is largely due to the lack of any effective therapy. Indeed, standard of care treatments, which include FOLFIRINOX and gemcitabine in combination with paclitaxel, provide only a marginal survival benefit and frequently cause severe toxicity that necessitates treatment discontinuation [3]. Moreover, immunotherapies that have demonstrated efficacy against other malignancies have repeatedly failed in clinical trials for PDAC [4]. Thus, new and more effective treatment strategies are needed to improve patient survival and quality of life.

Metabolic reprogramming, first recognized in malignant cells nearly a century ago [5], is now an established cancer hallmark [6]. Pyrimidine nucleotide biosynthesis has been implicated in numerous studies as a critical metabolic dependency of many different cancer types, including neurological malignancies, leukemia, lymphoma, and various carcinomas, including PDAC [7–27]. Dihydroorotate dehydrogenase (DHODH), an essential enzyme in the de novo pyrimidine biosynthesis pathway, is the target of several clinically approved and experimental drug compounds. However, despite remarkable preclinical activity, the potent and selective DHODH inhibitor brequinar (BQ) failed to show efficacy across several human trials in which it was given as monotherapy against various solid cancers, including PDAC [28–32]. The striking discordance between the preclinical and clinical effectiveness of BQ led us to investigate whether PDAC cells have intrinsic resistance to DHODH inhibition and if so, could this be overcome by combination drug treatment.

We characterized the BQ response across a panel of PDAC cell lines and identified several highly resistant lines. We then screened >350 kinase inhibitors to identify compounds that restored BQ sensitivity in BQ-resistant PDAC cells. The small molecule screen and hit validation experiments revealed that the combination of BQ and CNX-774 causes synergistic loss of cell viability and pyrimidine nucleotide depletion in BQ-resistant PDAC cell lines.

CNX-774 was developed as an inhibitor of Bruton tyrosine kinase (BTK), though our mechanistic studies determined that the cooperativity of CNX-774 with BQ is independent of BTK inhibition and is instead due to inhibition of nucleoside salvage. Specifically, engagement by CNX-774 of equilibrative nucleoside transporter 1 (ENT1, also called SLC29A1) inhibits the uptake of extracellular uridine and thereby blocks replenishment of nucleotide pools via the nucleoside salvage pathway. Deletion of ENT1 impaired uridine uptake and profoundly sensitized PDAC cells to BQ in vitro, and ENT1 knockout improved tumor response to BQ, resulting in dramatically enhanced survival in an orthotopic, immunocompetent PDAC mouse model. Our study identifies a novel ENT1 inhibitor, highlights the essential role of nucleoside salvage in DHODH inhibitor resistance, and provides proof of concept evidence for combined targeting of de novo and salvage pyrimidine synthesis as a therapeutic strategy for PDAC and other cancers.

MATERIALS AND METHODS

Cell culture and reagents

C57BL/6J-congenic LSL-KrasG12D/+;LSL-Trp53R172H/+;Pdx-1-Cre (KPC) tumor-derived pancreatic cancer cell lines KPC 1245 and KPC 1199 were a kind gift from Dr. David Tuveson’s laboratory (Cold Spring Harbor, NY, USA). Cell lines were validated by PCR to detect mutant Kras and p53. The S2–013 cell line is a clonal derivative of the Suit2 cell line and was a kind gift from the Dr. Tony Hollingsworth’s laboratory at the University of Nebraska Medical Center (Omaha, NE, USA). All other cell lines in this study were obtained from American Type Culture Collection (Manassas, VA, USA). All human cell lines were authenticated by STR profiling by the Genetics Core at University of Arizona (Tucson, AZ, USA). Cells were routinely (at time of initial revival from liquid nitrogen storage and at least every 6 months) determined to be free of mycoplasma contamination by PCR-based methods. Cells were cultured in Dulbecco’s modified Eagle medium (Sigma-Aldrich, St Louis, MO, USA) supplemented with 50 IU/mL penicillin, 50 μg/mL streptomycin, and incubated at 37 °C in a humidified incubator with 5% CO2. Cells were maintained at 10% fetal bovine serum (FBS). Upon reaching 70–80% confluency, cells were passaged by washing with phosphate-buffered saline (PBS) before adding 0.25% trypsin (Caisson Labs, Smithfield, UT, USA) and plating at 25% confluency. Dialyzed FBS used in several experiments was obtained from Cytiva (Marlborough, MA, USA; catalog #SH300079.03) and confirmed to be depleted of nucleosides by LC-MS/MS analysis. Drug compounds were purchased from the following vendors: brequinar sodium from Sigma-Aldrich (catalog number SML0113), CNX-774 from Selleck Chemicals (Houston, TX, USA; catalog number S7257), gemcitabine from Sagent Pharmaceuticals (Schaumberg, IL, USA; NDC 25021-243-10), fludarabine from Selleck Chemicals (catalog number S1491), 5-fluoro-5’-deoxyuridine from Sigma Aldrich (catalog number F8791), 6-thioguanine from Sigma-Aldrich (catalog number A4882), 5-fluorouracil from Intas Pharmaceuticals (Ahmedabad, Gujarat, India; NDC 16729-276-03), 3-deazauridine from Toronto Research Chemicals (North York, ON, Canada; catalog number D203240), cytarabine from MedChem Express (Monmouth Junction, NJ, USA; catalog number HY-13605/CS-2177), and NBMPR (full name S-(nitrobenzyl)-6-thioinosine) from Cayman Chemical Company (Ann Arbor, MI, USA; catalog number 16403). Compounds were dissolved in water or dimethyl sulfoxide (DMSO) according to manufacturer’s instructions and stored at −80° C (except gemcitabine and 5-fluourouracil, which are stored at room temperature).

Kinase inhibitor small molecule screen

2×103 S2–013 cells per well were seeded in opaque white 96 well plates and allowed to equilibrate overnight. The following day, plates were treated with compounds from the Selleck kinase inhibitor library at a final concentration of 100nM, and either 25μM BQ or 0.02% DMSO as a vehicle control. After 72 hours, cell viability was assessed by CellTiter-Glo from Promega (Madison, WI, USA) according to manufacturer’s instructions. Each compound was assigned a score by (viability under BQ + kinase inhibitor)/(viability under vehicle + kinase inhibitor), and compounds with the lowest numerical scores were prioritized for hit validation by dose-response experiments. The screen was performed in triplicate and the average for each condition was used to generate each compound score shown in Fig 2A.

Figure 2: CNX-774 restores BQ-sensitivity in S2–013 cells in a BTK-independent manner.

Figure 2:

Open in a new tab

A) Representation of relative cell viability under (BQ + kinase inhibitor) treatment divided by (vehicle + kinase inhibitor) treatment for each kinase inhibitor in the screen. Screen was conducted with biological triplicates. B) Structure of CNX-774, with Michael acceptor group outlined in red. C-D) Dose-response curve for BQ in the presence (red line) or absence (black line) of CNX-774 in BQ-resistant S2–013, KPC 1245, KPC 1199, B16F10, HCT116 (C) or BQ-sensitive CFPAC-1, A549, and A375 (D) cell lines. Data represent mean +/− SEM of 3 biological replicates. E) Area under curve (AUC) quantification for viability experiments in C and D, error bars represent 95% confidence interval. F) Relative BTK protein expression from various cancer cell lines, derived from cancer cell line encyclopedia (CCLE) Q06817 proteomics dataset. Blood- and lymphoma-derived cancer cell lines are highlighted in blue, pancreatic cancer cell lines highlighted in red. G) Dose-response curve for CNX-774 (left) or the clinically approved BTK inhibitors ibrutinib (middle) or acalabrutinib (right) in the presence (red line) or absence (black line) of BQ (25μM). Data represent mean +/− SEM of three biological replicates. H) Immunoblot for BTK in various cancer cell lines. Human chronic lymphocytic leukemia (CLL) cell lines were used as a positive control. Actin was used as a loading control.

Cell viability assays

Cells were seeded in 96 well plates (1×103 to 5×103 cells per well, depending on the cell line) and allowed to equilibrate overnight. Seeding density was optimized for each cell line based on its growth kinetics and ability to grow out from a small number of cells. Cells were then treated with indicated compounds for 72 hours and viability was assessed by CellTiter-Glo cell viability assay. Luminescence values for each condition were normalized to the average luminescence of the vehicle-treated control replicates. Each experiment was conducted at least three times independently. For each assay, three independent biological replicates were utilized for each condition. Dose-response curves were fit to nonlinear regression models using Prism9 software ([Inhibitor] vs response (variable slope, four parameters)), which was also used to quantify area under the curve (AUC).

Cell proliferation assays and live cell imaging

Cells were seeded and treated the same way as for cell viability assays in transparent 96-well plates. Cell number was serially quantified by brightfield imaging using Nexcelom Celigo Imaging Cytometer, and representative images are shown in Fig 1D. Instrument focus procedure, intensity threshold, minimum cell area, cell diameter, and other analysis parameters were optimized for each cell line.

Figure 1: Brequinar-resistant PDAC cell lines maintain pyrimidine pools and continue proliferation under DHODH inhibition.

Figure 1:

Open in a new tab

A) Schematic of de novo pyrimidine biosynthesis pathway, with DHODH step blocked by BQ. B) BQ dose-response cell viability experiments were performed on a panel of human and murine PDAC cell lines. Cells were treated with indicated doses of BQ for 72 hours and viability was determined by CellTiter Glo assay. Data represent mean +/− SEM of at least 3 biological replicates. C) CFPAC-1 (left) or S2–013 (right) cells were cultured with indicated BQ concentrations for four passages. Cells were counted by image cytometry at each passage. D) Brightfield images of BQ-sensitive CFPAC-1 cells (top) or BQ-resistant S2–013 cells (bottom) after 72 hour exposure to the indicated dose of BQ, captured by image cytometry. E-I) CFPAC-1 or S2–013 cells were treated with indicated concentrations of BQ or vehicle for 8 hours and subjected to LC-MS/MS metabolomics analysis of polar metabolites. E) Principal component analysis (PCA) plot for CFPAC-1 (left) or S2–013 (right) cells. Each data point represents an individual biological replicate. F) Heatmap for top 50 differentially altered metabolites in BQ-treated vs control groups. Pyrimidine pathway metabolites are highlighted in violet. G) Pathway enrichment analysis for KEGG pathways significantly altered upon BQ-treatment. H) Quantification of relative abundance of indicated metabolites, normalized to respective vehicle-treated control. Data represent mean +/− SEM of 4 biological replicates. * p < 0.05, **** p < 0.0001 by two-way ANOVA with Bonferroni’s post-hoc test.

CRISPR/Cas9 gene editing

For the generation of stable knockout cell lines, lentiviral vectors encoding Cas9 and sgRNA sequences against DHDOH or ENT1 (or a nontargeting sgRNA for control cell lines) were obtained from Cellecta (Mountain View, CA USA) and packaged into lentiviral particles using HEK-293FT cells. Target cell lines were transduced and then selected by blasticidin resistance. Single cell clones were then isolated, and genetic knockout was validated by immunoblotting (for DHODH) or by PCR amplification and Sanger sequencing of genomic region targeted by sgRNA (for ENT1).

Liquid chromatography – tandem mass spectrometry metabolomics analysis

For metabolomics experiments, 5 × 105 cells were seeded in 6-well plates and allowed to equilibrate overnight. At the start of each assay, the cell culture media was changed and fresh media with desired conditions was added (to eliminate metabolite depletion from overnight equilibration as a confounding variable). Following 8 hour treatment with indicated compounds of cancer cell lines (or in the case of Fig 4F–I, 8 hour exposure to nucleoside-depleted media), polar metabolites were extracted and quantified as previously described [33]. For analysis of media (Fig 4J, 6F), 200μl of cell culture media was harvested at indicated time points and polar metabolite extraction with 80% v/v methanol was performed as for cellular metabolites. Datasets were processed using Skyline (MacCoss Lab Software) and Metaboanalyst5.0. Relative metabolite abundances were normalized to the average peak area of the experimental control group and were compared using two-way ANOVA with Bonferonni’s post-test correction for multiple comparisons. P < 0.05 was considered significant.

Figure 4: DHODH knockout phenocopies BQ treatment and is synthetic lethal with CNX-774.

Figure 4:

Open in a new tab

A) Immunoblot for DHODH in S2–013 clonally-derived Cas9-expressing cell lines with sgRNA targeting DHODH (sgDHODH) or a nontargeting sgRNA control (sgNT) vector. B) BQ dose-response cell viability experiment in sgDHODH or sgNT cells. Data represent mean +/− SEM of three biological replicates. C-D) Cell proliferation assay of DHODH knockout or sgNT control cell lines under various concentrations of exogenous uridine (C) or cytidine (D). Cell numbers were quantified by image cytometry after 5-day culture. Data represent mean +/− SEM of 3 biological replicates. E). CNX-774 dose-response cell viability experiment with DHODH knockout or sgNT control cell lines. Data represent mean +/− SEM of 3 biological replicates. F-H) sgDHODH or sgNT control S2–013 cells were cultured in nucleoside-depleted media (10% dialyzed FBS) for 8 hours and subjected to LC-MS/MS-based metabolomic analysis for polar metabolites. Principal component analysis plot (F), heatmap visualization of top 50 altered metabolites (G), and volcano plot of sgDHODH-A12 versus sgNT cells (H). Metabolites related to pyrimidine metabolism are listed in violet color in (G). I) Quantification of relative abundance of indicated metabolites normalized to sgNT control. Data represent mean +/− SEM of six biological replicates. * p < 0.05, ** p < 0.01, *** p < 0.001, and **** p < 0.0001 by two-way ANOVA with Bonferonni’s post-hoc test. J) LC-MS/MS-based quantification of exogenous uridine in cell culture media collected at indicated time points under control or CNX-774-treated conditions. Data represent mean +/− SD of 3 biological replicates, # indicates p < 0.001 by two-way ANOVA with Bonferroni’s post-hoc test.

Figure 6: ENT1 knockout phenocopies CNX-774, ablates cooperativity between BQ and CNX-774, and enhances in vivo response to BQ in syngeneic orthotopic mouse model A-C).

Figure 6:

Open in a new tab

Dose-response cell viability experiments for BQ (A), gemcitabine (B), and 5-fluoro-5’-deoxyuridine (doxifluridine) (C) in ENT1-knockout or sgNT control KPC 1245 cells. Data represent mean +/− SEM of three biological replicates. D) Cell proliferation assay with sgNT control or ENT1-knockout KPC 1245 cells in the presence (dotted lines) or absence (solid lines) of BQ (10μM), cell number serially quantified by image cytometry, data indicate mean +/− SD of three biological replicates. E) Uridine rescue experiment in ENT1-knockout or sgNT cells treated with BQ (10μM). Experiment done in nucleoside-depleted media with 10% dialyzed FBS. Data represent mean +/− SEM of three biological replicates. F) LC-MS/MS-based quantification of exogenous uridine in cell culture media at indicated time points, data represent mean +/− SD of 3 biological replicates. G) BQ dose-response cell viability experiments in the presence or absence of CNX-774 in ENT1 knockout and sgNT control cells. Data represent mean +/− SEM of three biological replicates. H-I) ENT1 knockout or sgNT control KPC 1245 cells were implanted orthotopically in syngeneic C57BL/6J mice, which were then treated with BQ (10mg/kg IP daily) or vehicle (0.9% NaCl IP daily) during the indicated time period (right of dotted line in H and shaded area of I). H) Tumor volume measurements. Data represent mean +/− SEM, statistical comparison of day 30 tumor volume is shown, # indicates p < 0.001 by one-way ANOVA with Tukey’s multiple comparison test. I) Kaplan-Meier survival plots. Median survival (days post implantation) is shown in parentheses in graph legend for each cohort. # Indicates p ≤ 0.0001 by Log-rank (Mantel-Cox) test. N = 10 for sgNT BQ and sgENT1-F5 vehicle, N = 8 for sgENT1-F13 vehicle, and N = 9 for all other groups.

Immunoblot

Immunoblotting procedure was described previously [33]. Antibodies for BTK (clone D3H5, catalog number: 8547) and DHODH (catalog number: 80981) were obtained from Cell Signaling Technology (Danvers, MA USA). Human chronic lymphocytic leukemia cell line lysates, used as a positive control for BTK expression in Fig 2H, were a kind gift from the Dr. Dalia ElGamal laboratory at the University of Nebraska Medical Center.

Real time quantitative polymerase chain reaction

1×105 S2–013 cells were seeded in 6-well plate format, allowed to equilibrate overnight, and then treated with BQ (10μM) or vehicle for 72 hours. RNA was isolated using Trizol reagent per manufacturer’s instructions (Thermo Fisher Scientific, Waltham, MA, USA). cDNA synthesis was performed using iScript cDNA synthesis kit (catalog number 1708891) from Bio-Rad (Hercules, CA, USA) according to manufacturer’s instructions. Reactions were run using QuantStudio5 instrument from Applied Biosystems (Waltham, MA, USA) with previously specified thermocycle settings [34]. Data was normalized to 18S ribosomal RNA and is displayed in Fig 5E as 2^-(gene of interest Ct – 18S Ct). Beta actin mRNA was used as an additional loading control. Data in Figure 5E represent mean +/− SD of four replicates. An unpaired student’s t -test was used to compare vehicle and BQ-treated samples, p < 0.05 was considered significant. RT-qPCR primer sequences are provided in supplementary Table 1.

Figure 5: CNX-774 shows drug interactions across a panel of cytotoxic nucleoside analog drugs identical to that of ENT1 inhibitor NBMPR.

Figure 5:

Open in a new tab

A) Schematic illustrating nucleoside salvage pathway with candidate CNX-774 targets highlighted. B-H) Dose-response of indicated cytotoxic nucleoside analog drugs in the presence (red line) or absence (black line) of CNX-774 (B-D) or ENT1 inhibitor NBMPR (F-H) along with cartoon showing activation pathway for each nucleoside analog drug. Data represent mean +/− SEM of three biological replicates. E) Relative mRNA expression for indicated genes in S2–013 cells assessed by real time quantitative PCR after 72 hour treatment with BQ (10μM) or vehicle. Data represent mean +/− SD of four replicates. N.D. stands for not detected. I) Table of drug interactions of CNX-774 and NBMPR across a panel of nucleoside analog drugs. Red indicates antagonism, blue indicates cooperativity, black indicates no interaction. Dose response experiments for other drugs in table shown in Figure S5.

Mice studies

The procedure for orthotopic implantation of murine PDAC cells into syngeneic immunocompetent hosts has been previously described [35]. Briefly, KPC 1245 cells were trypsinized, counted, washed three times with PBS and then resuspended in PBS, and kept on ice until implantation. 10-week-old female C57BL6/J mice (Jackson Laboratory; Bar Harbor, ME, USA) were used for orthotopic implantation of PDAC cells into the pancreas. For survival analysis, mice were monitored daily for signs of euthanasia criteria or actual demise. Tumor growth was monitored by serial caliper measurement made by an independent investigator blinded to the experimental conditions. Brequinar for in vivo studies was provided by Clear Creek Bio (Cambridge, MA, USA; batch number PS02256–26-E-P1) and dissolved in physiologic saline (0.9% NaCl) at pH 8.0. This solvent was used as vehicle for injection into control cohorts. All procedures were approved by the Institutional Animal Care and Use Committee (IACUC) at the University of Nebraska Medical Center.

Statistical analysis

For animal studies, post-implantation day 30 tumor volume was compared between groups using an ordinary one-way ANOVA with Tukey’s multiple comparison test; p < 0.05 was considered significant. Mouse survival was compared using the log-rank (Mantel-Cox) test; p < 0.05 was considered significant. For metabolomics experiments, relative metabolite abundances were compared between groups (at 8 hour time point for media uridine depletion assays) using a two-way ANOVA with Bonferonni’s multiple comparison test; p < 0.05 was considered significant. Principal component analysis, partial least squares discriminant analysis, volcano plot analysis, heatmap visualization, and pathway enrichment analysis were performed using Metaboanalyst5.0 web-based platform (https://www.metaboanalyst.ca). Data was input into Metaboanalyst as follows: no filtering was applied (as recommended for datasets with less than 5000 features), data was normalized using log (base 10) transformation and Pareto scaling. For real time quantitative polymerase chain reaction (RT-qPCR) experiments, normalized mRNA relative abundances for each gene were compared using an unpaired two-tailed student’s t-test; p < 0.05 was considered significant.

RESULTS

Brequinar-resistant PDAC cell lines maintain pyrimidine pools despite DHODH inhibition

DHODH is required for de novo synthesis of pyrimidine nucleotides, and its inhibition is expected to result in depletion of intracellular pyrimidines (Fig 1A). To evaluate the effect of DHODH inhibition on PDAC proliferation and viability, dose-response experiments with the DHODH inhibitor brequinar (BQ) were performed across a panel of human and murine PDAC cell lines (Fig 1B). Some cell lines undergo nearly complete loss of viability under DHODH inhibition (BQ-sensitive), while others continue to proliferate, albeit at a slower rate (BQ-resistant). This observation was confirmed by cell proliferation assays (Fig 1C) and live cell imaging (Fig 1D) with BQ-sensitive CFPAC-1 and BQ-resistant S2–013 cells.

To investigate the metabolic phenotypes associated with sensitivity/resistance to DHODH inhibition, we performed metabolomics analysis on CFPAC-1 and S2–013 cells following BQ treatment. Treatment with BQ (8 hours) caused profound metabolic perturbations in both cell lines, as evidenced by separation of control and BQ groups upon principal component analysis (Fig 1E) and by volcano plot analysis (Fig S1A). Heatmap visualization of the top 50 differentially altered metabolites for each cell line (Fig 1F) demonstrated perturbation of the steady-state levels of diverse metabolites, most prominently pyrimidine nucleosides/nucleotides (as expected), but also intermediates in several other metabolic pathways. Pathway enrichment analysis (Fig 1G) suggested that commonly perturbed metabolic pathways across both cell lines include pyrimidine metabolism and alanine/aspartate/glutamate metabolism. However, the most striking difference between the two cell lines was the degree of pyrimidine nucleotide depletion following BQ treatment. Across the detected pyrimidine nucleotide species (Fig S1B), and especially the pyrimidine NTPs (Fig 1H, right), CFPAC-1 cells underwent a much greater depletion than S2–013 cells (>90% versus <50%), despite a comparable accumulation of upstream de novo pyrimidine pathway metabolites N-carbamoyl-aspartate and dihydroorotate (DHO) (Fig 1H, left) and despite exposure to 50-fold higher BQ concentration in S2–013 compared to CFPAC-1 cells (25μM vs 500nM). This suggests that S2–013 cells can maintain adequate pyrimidine nucleotide pools to support proliferation even upon DHODH inhibition, explaining their BQ-resistance.

Interestingly, we observed BQ-induced accumulation of N-acetylneuraminic acid (Neu5Ac, aka sialic acid) in both cell lines (Fig S1A–B), due to depletion of CTP, which is required for the conversion of Neu5Ac to Neu5Ac-CMP [36] (Fig S1C). Concordantly, Neu5Ac accumulation was greater in CFPAC-1 than S2–013 cells (Fig S1B, right). Overall, these results suggest that BQ-resistance reflects a cell’s capacity to maintain pyrimidine pools sufficient to support proliferation despite DHODH inhibition.

A chemical library screen identifies CNX-774 as a sensitizer for BQ in resistant PDAC cells

To probe the mechanism of BQ-resistance, we screened a library of ~350 small molecule protein kinase inhibitors for the ability to induce BQ-sensitivity in S2–013 cells (Fig 2A). We hypothesized that since the activity of many metabolic enzymes can be modulated by cellular kinases, this screen might reveal signaling and/or metabolic pathways that are essential for BQ-resistance. Top scoring hits (calculated as viability upon treatment with BQ and inhibitor combination divided by viability with inhibitor treatment alone) were prioritized for validation in dose-response experiments. The most robust hit was CNX-774 (Fig 2B), under investigation as a Bruton tyrosine kinase (BTK) inhibitor. CNX-774 enhanced BQ efficacy (i.e. maximal loss of viability) across a panel of BQ-resistant cancer cell lines representing diverse cancer types (Fig 2C), including cell lines derived from autochthonous LSL-KRASG12D/+; LSL-TRP53R172H/+; PDX-1-Cre (KPC) murine PDAC tumors [35], which were previously reported to be resistant to DHODH inhibition [8]. However, CNX-774 had no effect on BQ efficacy in various BQ-sensitive cell lines (Fig 2D, 2E).

Because BTK is not expressed in human PDAC, nor in most human carcinoma samples (Human Protein Atlas, Fig S2A), nor in carcinoma-derived cell lines (CCLE, Fig 2F), we sought to determine if CNX-774 exerts its BQ-sensitizing activity in a BTK-independent manner. We compared the activity of CNX-774 and the clinically approved BTK inhibitors ibrutinib and acalabrutinib in the presence or absence of BQ (Fig 2G). Neither ibrutinib nor acalabrutinib replicated the BQ-sensitizing activity of CNX-774. Furthermore, neither S2–013 nor KPC 1245 PDAC cell lines expressed BTK as assessed by immunoblot (Fig 2H). Thus, we concluded that CNX-774 sensitizes BQ-resistant cancer cell lines to BQ in a BTK-independent manner.

CNX-774 inhibits pyrimidine nucleoside salvage and combines with BQ to cause pyrimidine nucleotide depletion

To explore the mechanism of CNX-774 cooperativity with BQ, we quantified metabolites in S2–013 cells treated with vehicle, BQ, CNX-774, or the combination. Principal component analysis (PCA) distinguished BQ treatment and combination treatment from each other and from the vehicle control, but not the control from CNX-774 monotherapy (Fig 3A), suggesting that CNX-774 causes significant metabolic perturbations only in the presence of BQ. To identify which metabolites accounted for the independent clustering of BQ and combination groups upon PCA, we quantified the top 15 metabolites most responsible for this separation as reflected by their variable importance in projection (VIP) scores determined by partial least squares discriminant analysis (PLSDA, Fig 3B). Depletion of pyrimidine nucleotides, UDP-glycoconjugates (dependent on UTP for their synthesis), and CDP-choline (dependent on CTP for its synthesis), as well as accumulation of Neu5Ac (reflective of CTP depletion) were largely responsible for separation of BQ and combination treatment groups, and this was confirmed by volcano plot analysis comparing the combination versus BQ alone (Fig 3C).

Figure 3: CNX-774 inhibits nucleoside salvage and causes profound pyrimidine nucleotide depletion in combination with BQ.

Figure 3:

Open in a new tab

A-D) S2–013 cells were treated with vehicle, BQ (5μM), CNX-774 (2μM), or the combination thereof for 8 hours and subjected to LC-MS/MS-based metabolomic analysis for polar metabolites. A) Principal component analysis (PCA) plot of the 4 experimental groups. Each data point represents an individual biological replicate. B) List of top 15 differentially altered metabolites between BQ and combination groups ranked by VIP score using partial least squares discriminant analysis (PLS-DA). Pyrimidine nucleotide species are listed in violet color. C) Volcano plot of BQ + CNX-774 vs BQ only treatment groups. Fold change > 2 and p < 0.05 were considered significant. D) Heatmap of top 50 differentially altered metabolites. Metabolites related to pyrimidine metabolism are listed in violet color. E) Quantification of relative abundance of indicated metabolites normalized to vehicle control. Data represent mean +/− SEM of six biological replicates. * p < 0.05, ** p < 0.01, *** p < 0.001, **** p < 0.0001 by two-way ANOVA with Bonferonni’s post-hoc test. F) Uridine and cytidine nucleotides quantified as described in E, in an independent experiment also including ibrutinib and ibrutinib + BQ treatment groups. G) Uridine addback cell viability experiments in cells treated with either BQ (black line) or BQ + CNX-774 (red line), experiment performed with nucleoside-depleted media with 10% dialyzed FBS. Data represent mean +/− SEM of three biological replicates. H) Schematic of working hypothesis suggesting CNX-774 inhibits nucleoside salvage.

Heatmap visualization (Fig 3D) and pyrimidine nucleotide relative abundances (Fig 3E, middle) confirmed that the combination of BQ and CNX-774 caused a more profound depletion of pyrimidine metabolites than BQ alone. Notably, CNX-774 alone had little impact on pyrimidine levels and did not lead to an accumulation of DHO or N-carbamoyl-aspartate (Fig 3E, left), indicating that CNX-774 does not inhibit de novo pyrimidine biosynthesis nor does it enhance the BQ inhibitory effect on DHODH. Concordantly, Neu5Ac accumulation was observed with combination treatment but not with either single agent (Fig 3E, right). Furthermore, ibrutinib did not mimic the effects of CNX-774 on nucleotide abundance when combined with BQ (Fig 3F) in an independent experiment, arguing against BTK inhibition as the mechanism of BQ sensitization. One parsimonious explanation for these results is that CNX-774 blocks the nucleoside salvage pathway (NSP), by which uridine (present in fetal bovine serum and thus in cell culture media) is converted to UMP and incorporated into pyrimidines independently of DHODH (Fig 3H).

To test this hypothesis, we confirmed that supplementing supraphysiologic concentration of exogenous uridine (100μM) can completely rescue BQ-induced loss of viability (in otherwise nucleoside-free media with 10% dialyzed FBS) (Fig S3A). We then titrated exogenous uridine (Fig 3G) in the media of cells treated with BQ in the presence or absence of CNX-774 and observed that rescue of cell viability by uridine was blocked by CNX-774, although this could be overcome at very high uridine concentrations. Thus, we conclude that CNX-774 inhibits the salvage of extracellular uridine and thereby cooperates with DHODH inhibition to cause pyrimidine nucleotide depletion (Fig 3H), explaining its efficacy when combined with BQ.

Genetic knockout of DHODH phenocopies BQ and exhibits synthetic lethality with CNX-774

To ensure that the CNX-774 cooperativity with BQ was on-target with respect to DHODH, we generated clonal DHODH-null S2–013 cell lines (Fig 4A). DHODH knockout (DHODH-KO) cells were completely resistant to BQ-mediated loss of viability (Fig 4B), confirming the on-target effect of BQ. Additionally, DHODH-KO cells required a minimum extracellular concentration of 10μM uridine or cytidine to support their proliferation (Fig 4C–D). As expected, DHODH-KO cells were profoundly sensitized to CNX-774 compared to control cells (Fig 4E). Thus, CNX-774 exhibits synthetic lethality DHODH deletion, and its cooperativity with BQ is strictly dependent on DHODH inhibition.

To investigate the metabolic effect of DHODH ablation, we assayed metabolites from DHODH-KO and control S2–013 cell lines under salvage-limiting conditions by using nucleoside-depleted (10% dialyzed FBS) media. The loss of DHODH resulted in profound metabolic perturbations, as evidenced by PCA (Fig 4F), heatmap visualization (Fig 4G), and volcano plot analysis (Fig 4H, S4A). DHODH-KO cells, after 8 hours in nucleoside-depleted media, displayed a striking depletion of pyrimidine nucleotides (Fig 4I, middle), accumulation of N-carbamoyl-aspartate and DHO (left), and buildup of Neu5Ac (right), consistent with the metabolic changes observed following BQ + CNX-774 treatment. In a separate experiment, we supplemented exogenous uridine and monitored its depletion from the media, due to cellular uptake. Uridine was more rapidly depleted from the media by DHODH-KO cells (consistent with their complete reliance on nucleoside salvage to generate pyrimidine nucleotides), and this depletion was blocked by CNX-774 (Fig 4J), confirming that CNX-774 inhibits the uptake of extracellular uridine.

A complementary chemical biology approach confirms ENT1 as the target of CNX-774

Having established that CNX-774 inhibits uridine salvage, sensitizing cells to BQ, we sought to pinpoint the target of CNX-774. The flux of extracellular nucleosides through the NSP is a two-step process entailing 1) the uptake of nucleosides through plasma membrane resident nucleoside transporters and 2) the enzymatic phosphorylation of nucleosides to yield nucleotide monophosphates (Fig 5A). The lack of uridine uptake from the media following treatment with CNX-774 (Fig 4J) could theoretically result from inhibition of either of these two steps.

Uridine-cytidine kinases (UCK1 and UCK2) phosphorylate uridine and cytidine to form UMP and CMP. Deoxycytidine kinase (DCK) phosphorylates various deoxynucleosides, including deoxycytidine, to form their cognate dNMPs (Fig 5A). To interrogate whether CNX-774 inhibits UCK1/2 or DCK, we leveraged the fact that certain cytotoxic nucleoside analogs are only activated following phosphorylation.

The uridine analog 3-deazauridine is activated by UCK1/2 [37]. CNX-774 had no effect on the activity of 3-deazauridine (Fig 5B) and thus, UCK enzymes were ruled out as the target of CNX-774. Gemcitabine is a deoxycytidine analog and obligate DCK substrate [38, 39]. CNX-774 decreased gemcitabine cytotoxicity (Fig 5C) suggesting that it inhibited either the uptake of gemcitabine or its phosphorylation by DCK. Fludarabine is a deoxyadenosine analog and obligate DCK substrate [40, 41]. Treatment with CNX-774 had no effect on fludarabine efficacy (Fig 5D), arguing against DCK as its target. Notably, a DCK inhibitor would not be expected to block rescue of BQ-mediated toxicity by uridine (Fig 3G), as it is a ribonucleoside and therefore not a DCK substrate. Together, these results suggest that CNX-774 inhibits the NSP at the level of uridine uptake rather than uridine phosphorylation.

Of the seven known nucleoside transporters, only ENT1, ENT2, ENT3, ENT4, and CNT1 could be detected at the mRNA level by RT-qPCR in S2–013 cells, and none of these transporters were transcriptionally induced by BQ treatment (72 hours) (Fig 5E). ENT3 localizes to the mitochondrial and lysosomal membranes, but not to the plasma membrane, and is not known to mediate import of extracellular nucleosides [42]; ENT3 inhibition therefore cannot account for impaired nucleoside uptake. Likewise, since ENT4 transports only adenosine (but not pyrimidine nucleosides) [43, 44], ENT4 inhibition cannot account for impaired uridine uptake. Finally, while CNT1 can mediate uridine uptake, it is thought to be a low-capacity transporter that is responsible for sensing nucleoside abundance rather than maintaining homeostatic cellular nucleoside levels [45], and so it too was considered an unlikely candidate.

To query which of the two remaining ENT proteins (ENT1 and ENT2) is the target of CNX-774, we compared CNX-774 activity to that of the known ENT1 inhibitor nitrobenzylmercaptopurine riboside (NBMPR), which preferentially inhibits ENT1 >> ENT2 at nanomolar concentrations [46, 47]. Like CNX-774, NBMPR (100nM) displayed cooperativity with BQ, antagonism with gemcitabine, and no interaction with fludarabine or 3-deazauridine (Fig 5F–H). Furthermore, we compared CNX-774 and NBMPR in combination with a panel of other nucleoside analogs, including cytarabine (Fig S5A), 5-fluoro-5’-deoxyuridine (also known as doxifluridine) (Fig S5B), 6-thioguanine (Fig S5C), and 5-fluorouracil (Fig S5D), and found that their effect on nucleoside analog activity was identical in all cases (summarized in Fig 5I). Together, these results strongly suggest that CNX-774 inhibits ENT1-mediated nucleoside uptake.

ENT1 loss phenocopies the effect of CNX-774 treatment and negates CNX-744 cooperativity with BQ

To confirm that CNX-774 inhibits ENT1, we generated clonal ENT1-knockout (ENT1-KO) cells. Like CNX-774 and NBMPR treatment, ENT1 knockout conferred sensitivity to BQ (Fig 6A) and resistance to gemcitabine (Fig 6B). Additionally, all three perturbations conferred resistance to 5-fluoro-5’-deoxyuridine (doxifluridine) (Fig 6C and S5B). Doxifluridine is a 5-FU prodrug and is metabolized to 5-FU by uridine phosphorylase; because its 5’ ribose carbon has no hydroxyl group, it cannot be phosphorylated. Thus, resistance to doxifluridine must be caused by impaired uptake via ENT1. Importantly, ENT1 knockout did not significantly inhibit cell proliferation under control conditions, while proliferation of ENT1-KO cells was completely abrogated by BQ-treatment (Fig 6D).

Concordantly, like CNX-774 treatment (Fig 3G), ENT1 knockout dramatically impaired rescue of BQ toxicity by uridine (Fig 6E). When we monitored uptake of exogenous uridine from the cell culture media, we found that BQ treatment accelerates uridine uptake in ENT1-competent control cells and that ENT1-KO cells take up uridine much less rapidly, regardless of BQ presence (Fig 6F). This is consistent with our previous observation (Fig 4J) that DHODH knockout accelerates uridine uptake, which is inhibited by CNX-774. Finally, while control KPC 1245 cells retain BQ-resistance and are sensitized by CNX-774, ENT1-KO cells are sensitive to BQ, and their sensitivity is not enhanced by addition of CNX-774 (Fig 6G). Taken together, these results confirm that ENT1 knockout is synthetic lethal with BQ treatment and that CNX-774 efficacy in combination with BQ is due to inhibition of ENT1. CNX-774 therefore represents a previously unknown ENT1 inhibitor that leads to nucleotide starvation in combination with DHODH inhibition.

Combined targeting of ENT1 and DHODH is effective against PDAC in a murine model

Having demonstrated that either CNX-774 treatment or ENT1 knockout can inhibit uridine uptake and restore BQ sensitivity in various in vitro experiments, we sought to determine if ENT1 ablation could cooperate with BQ to control tumor growth in vivo. Unlike in cell culture systems, cancer cells in vivo receive a constantly replenishing supply of extracellular nucleosides from the systemic circulation and from neighboring stromal cells in the tumor microenvironment. Therefore, it is possible that BQ-induced ENT1 dependence could be mitigated or even eliminated in vivo.

To test this possibility, we implanted control or ENT1-KO KPC 1245 cells into the pancreata of syngeneic (C57BL6/J) mice and assessed tumor growth and mouse survival following BQ (10 mg/kg IP daily) or vehicle treatment. In vitro, ENT1 knockout marginally accelerated cell proliferation, while BQ slowed proliferation of ENT1-competent cells and completely abrogated proliferation of ENT1-KO cells (Fig 6D). Concordantly, while BQ delayed the growth of ENT1-competent tumors (Fig 6H) and conferred a marginal improvement in median survival (Fig 6I), these effects were not significant. Conversely, BQ sharply inhibited growth of ENT1-KO tumors (Fig 6H), resulting in a dramatic and highly significant survival benefit compared to their respective vehicle controls and versus BQ-treated ENT1-competent tumors (Fig 6I). Overall, these results provide powerful proof of concept evidence for the anti-tumoral efficacy of combined pyrimidine de novo and salvage pathway inhibition, which warrants further study as a therapeutic approach for PDAC and other malignancies.

DISCUSSION

Despite robust and reproducible preclinical activity, DHODH inhibitors have failed to demonstrate efficacy across multiple clinical trials in the treatment of patients with PDAC and other solid tumors. Cancer cells may be resistant to the inhibition of metabolic enzymes by utilizing redundant pathways to bypass the inhibited reaction and avoid depletion of critical metabolic intermediates. Our study implicates uridine salvage as a mechanism of resistance to DHODH inhibition (Fig 5A). We demonstrate 1) that this resistance mechanism is active in 2D culture of PDAC cells (as well as BQ-resistant cell lines derived from other cancers) and can be ablated by genetic knockout of ENT1 or by its inhibition with CNX-774, and 2) that combined targeting of ENT1 and DHODH is highly efficacious in an aggressive murine model of PDAC.

Our specific experimental setup for in vivo studies was chosen for several reasons. We chose an immunocompetent mouse model because of the well-known immunomodulatory effects of BQ and the clinically approved DHODH inhibitor teriflunomide (and its prodrug leflunomide) [48], which is used to treat autoreactive T cell driven autoimmune syndromes such as rheumatoid arthritis [49] and multiple sclerosis [50]. Since this could theoretically favor tumor progression (by inhibiting anticancer adaptive immunity), we did not want to exclude this factor by using an immunocompromised mouse model. We found no evidence that BQ impaired anticancer immunity (as BQ did confer modest benefit even against ENT1-competent tumors), but nor can we rule out that possibility, and so the effects of DHODH inhibition on anticancer immunity warrants further investigation.

Implanting murine (as opposed to human) cancer cells also avoided a species-specific DHODH enzyme bias, since BQ has greater affinity for human DHODH than its murine counterpart [11, 51], which could introduce an artifactual therapeutic window. We chose to test genetic deletion of ENT1 rather than CNX-774 administration because BTK inhibition with ibrutinib has been previously shown to inhibit PDAC tumor growth by promoting anticancer immunity [52], and so we sought to exclude BTK inhibition as an explanation for enhanced therapeutic efficacy. We chose to use an orthotopic model (with cancer cells injected into the mouse pancreas) to more closely imitate the tumor microenvironment of human PDAC. Our studies indicate that while combined ablation of ENT1 and DHODH can effectively suppress tumor growth, this approach was insufficient to cause complete tumor regression in our model system. Previous studies have demonstrated that targeting de novo pyrimidine synthesis can sensitize PDAC tumors to chemotherapy [18, 53], and it is possible that more profound pyrimidine starvation by combined de novo and salvage pathway inhibition will further accentuate chemotherapy efficacy. Thus, combination with other therapies may be required for disease eradication, and this will be the subject of future studies. However, given the antagonistic effect of CNX-774 and NBMPR with gemcitabine (Fig 5C, 5G), but not 5-fluorouracil (Fig S5D), the effect of combined inhibition of DHODH/ENT1 with various chemotherapies will need to be carefully delineated.

Inhibition of ENT1/2 by the FDA-approved drug dipyridamole (DPM) synergizes with BQ and other DHODH inhibitors in cell culture models, and BQ + DPM showed significant albeit modest benefit (versus BQ monotherapy) in a mouse model of MYCN-amplified neuroblastoma [54]. However, BQ + DPM was no better than BQ monotherapy in a colon cancer xenograft model [55]. One possible explanation for this is that DPM is rapidly cleared [56, 57], and therefore administration at least every few hours is necessary to sustain ENT inhibition (ongoing clinical trials of DPM for COVID-19 use three (NCT05166876) or four (NCT04391179) doses per day). Additionally, while DPM inhibits both ENT1 and ENT2 at therapeutic concentrations, our animal studies with ENT1-KO tumors (Fig 6I–J) show that ENT2 inhibition is not required to confer BQ-sensitivity in our model, and so selective ENT1 inhibition may be desirable to avoid unnecessary disruption of host nucleotide homeostasis downstream of ENT2 blockade. Potent, specific, and pharmacokinetically stable (i.e. long-acting) ENT1 inhibitors, which are currently not clinically available, could therefore be useful. However, our data cannot rule out tumor type-specific efficacy differences. The present study defines CNX-774 as a previously unrecognized ENT1 inhibitor and provides insights that will help address this unmet need.

In similar fashion to our experiments, Abt et al. reported that CNX-774 scored in a screen for NSP inhibitors [58], although they did not investigate its mechanism. They demonstrated that a preclinical JNK inhibitor (JNK-IN-8) engages ENT1, and they point out that both CNX-774 and JNK-IN-8 contain an electrophilic Michael acceptor group (Fig 2B, red box) predicted to form covalent bonds with nucleophilic cysteine residues. This is the mechanism by which CNX-774 and ibrutinib inhibit BTK, and the crystal structure of the ibrutinib Michael acceptor group covalently bound to BTK active site cysteine 481 (C481) has been solved [59]. Interestingly, structural studies on human ENT1 have suggested that it contains several cysteine residues that could serve as the target site for Michael acceptors [60]. Our work supports this concept, and further studies may determine the precise mechanism of ENT1 inhibition by CNX-774.

The activity of CNX-774 as an inhibitor of BTK was a potential confounding factor that was rigorously exonerated as the mechanism of BQ sensitization (Fig 2F–H, S2A, 3F). However, there are certain clinical contexts in which combined inhibition of BTK and ENT1 may be desirable. For example, BTK has been proposed as a therapeutic target for patients with acute myeloid leukemia (AML) [61], and clinical trials of DHODH inhibitors for AML and other hematologic malignancies are ongoing (NCT04609826, NCT02509052 NCT05246384), based on strong preclinical data [10, 11]. In selected AML or other leukemia patients with dysregulated BTK activity, CNX-774 and BQ could therefore be a useful therapeutic combination, as CNX-774 could enhance BQ-mediated pyrimidine depletion (by inhibiting ENT1) and inhibit oncogenic BTK signaling. Also, as previously mentioned, BTK inhibitors have been suggested as a component of combination therapy for PDAC based on preclinical evidence that inhibition of stromal BTK signaling favorably alters the immune landscape of the tumor microenvironment to promote anticancer immunity [52].

Our work provides novel insights into PDAC cell metabolism under DHODH inhibition and nominates increased N-acetylneuraminic acid (Neu5Ac) as a metabolic marker of effective CTP depletion (Fig S1B–C, 3B–E, 4G–I). As CMP-Neu5Ac is required for protein sialylation (Fig S1C), nucleotide depletion would be predicted to decrease global protein sialylation. While hyperactive sialylation has been shown to promote tumor growth [62, 63] and immune evasion [64] in PDAC, loss of sialylation was recently reported to accelerate progression of AML [65]. Thus, the effects of decreased protein sialylation downstream of pyrimidine starvation may be tumor type dependent, and future studies will be aimed at dissecting this phenomenon in various contexts. Furthermore, upon DHODH blockade, we observed consistent perturbation of aspartate metabolism, including an accumulation of aspartate (Fig 1F, 3D, 4G–H, S4A) and products of aspartate metabolism such as asparagine (Fig 1F, right), fumarate (Fig 1F right, 4G–H, S4A), and arginosuccinate (Fig 1F right, 3D, 4G), likely because DHODH inhibition prevents aspartate flux into de novo pyrimidine synthesis (Fig 5A).

Interestingly, BQ treatment (or DHODH knockout) led to depletion of UDP-glycoconjugates such as UDP-glucose and UDP-N-acetylglucosamine (UDP-GlcNAc) (Fig 1F, S1A-B, 4G–I, S4A), which was strongly accentuated by CNX-774 (Fig 3B–E). Global upregulation of protein O-GlcNAcylation (which requires UDP-GlcNAc) is frequently observed in human cancer and has been implicated in driving several malignant behaviors [66], including hyperactive glucose metabolism [67], immune evasion [68], and metastasis [69, 70]. Inhibition of protein O-GlcNAcylation (by UDP-GlcNAc depletion) has also been proposed as a potential mechanism of BQ-induced differentiation of AML leukemic blasts [10]. Thus, the consequences of impaired protein O-GlcNAcylation downstream of pyrimidine depletion remain to be fully characterized and warrant further study.

Overall, our work demonstrates the potential utility of combined DHODH and ENT1 inhibition as a component of PDAC treatment that has not yet been clinically explored. We demonstrate CNX-774 is a novel ENT1 inhibitor and support previous studies suggesting cysteine-targeting drugs as possible ENT1 inhibitors. Further development of ENT1 inhibitors is also of interest for the treatment of viral infection (including COVID-19, for which BQ + DPM is undergoing clinical trials [NCT05166876, NCT04391179] based on promising preclinical activity [71]), cardiovascular disease (as DPM is an approved anti-angina medicine), and psychological/neurological disorders such as alcoholism [72] and anxiety disorder [73], in which ENT1 has been heavily implicated. Additional studies are thus urgently needed to unlock the full potential of therapies targeting nucleotide metabolism for cancer and other diseases.

Supplementary Material

Supplementary Table 1
Supplementary Figures and Legends

Figure S1: BQ-sensitive CFPAC-1 cells undergo greater pyrimidine depletion than BQ-resistant S2–013 cells. A) Volcano plot for significantly altered metabolites in BQ-treated vs control groups for CFPAC-1 (top) or S2–013 (bottom) cells, p < 0.05 and fold change > 2 was considered significant. B) Quantification of relative abundance of indicated metabolites, normalized to the respective vehicle-treated control. Data represent mean +/− SEM of 4 biological replicates. * p < 0.05, **** p < 0.0001 by two-way ANOVA with Bonferroni’s post-hoc test. C) Schematic explaining how CTP depletion leads to accumulation of Neu5Ac.

Figure S2: Bruton tyrosine kinase (BTK), the putative target of CNX-774, is not expressed in human PDAC samples nor human PDAC cell lines A) Percent of human samples with positive BTK expression as assessed by immunohistochemical staining with antibody HPA001198, from Human Protein Atlas (https://www.proteinatlas.org/ENSG00000010671-BTK/pathology).

Figure S3: Exogenous nucleosides can rescue BQ-induced toxicity A) BQ dose-response viability experiment in the presence or absence of 100μM uridine. Data represent mean +/− SEM of three biological replicates.

Figure S4: A) Volcano plot of significantly altered metabolites in sgDHODH-A28 vs sgNT.

Figure S5: Exploration of CNX-774 and NBMPR interactions with additional nucleoside analogs

A-D) Dose response cell viability experiments with indicated nucleoside analogs in the presence or absence of CNX-774 (left two panels) or NBMPR (right two panels), with cartoon showing nucleoside analog activation (middle). Data represent mean +/− SEM of three biological replicates.

ACKNOWLEDGMENTS

This work was supported in part by funding from the National Institutes of Health (R01CA163649, R01CA210439, R01CA216853, R01CA270234, U54CA274329, NCI) to PKS and to NJM (F30CA265277, NCI).

Footnotes

Conflict of Interest Statement: DBS is a co-founder and holds equity in Clear Creek Bio. The other authors declare no competing interests.

REFERENCES

  • 1.da Costa WL Jr, Oluyomi AO, and Thrift AP, Trends in the Incidence of Pancreatic Adenocarcinoma in All 50 United States Examined Through an Age-Period-Cohort Analysis. JNCI Cancer Spectrum, 2020. 4(4). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 2.Cai J, et al. , Advances in the epidemiology of pancreatic cancer: Trends, risk factors, screening, and prognosis. Cancer Letters, 2021. 520: p. 1–11. [DOI] [PubMed] [Google Scholar]
  • 3.Conroy T, et al. , FOLFIRINOX versus Gemcitabine for Metastatic Pancreatic Cancer. New England Journal of Medicine, 2011. 364(19): p. 1817–1825. [DOI] [PubMed] [Google Scholar]
  • 4.Principe DR, et al. , Trials and tribulations of pancreatic cancer immunotherapy. Cancer Letters, 2021. 504: p. 1–14. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 5.Warburg O, On the origin of cancer cells. Science, 1956. 123(3191): p. 309–14. [DOI] [PubMed] [Google Scholar]
  • 6.Hanahan D and Weinberg Robert A. Hallmarks of Cancer: The Next Generation. Cell, 2011. 144(5): p. 646–674. [DOI] [PubMed] [Google Scholar]
  • 7.Koundinya M, et al. , Dependence on the Pyrimidine Biosynthetic Enzyme DHODH Is a Synthetic Lethal Vulnerability in Mutant KRAS-Driven Cancers. Cell Chemical Biology, 2018. 25(6): p. 705–717.e11. [DOI] [PubMed] [Google Scholar]
  • 8.Li L, et al. , Identification of DHODH as a therapeutic target in small cell lung cancer. Science Translational Medicine, 2019. 11(517): p. eaaw7852. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 9.Boukalova S, et al. , Dihydroorotate dehydrogenase in oxidative phosphorylation and cancer. Biochimica et Biophysica Acta (BBA) - Molecular Basis of Disease, 2020. 1866(6): p. 165759. [DOI] [PubMed] [Google Scholar]
  • 10.Christian S, et al. , The novel dihydroorotate dehydrogenase (DHODH) inhibitor BAY 2402234 triggers differentiation and is effective in the treatment of myeloid malignancies. Leukemia, 2019. 33(10): p. 2403–2415. [DOI] [PubMed] [Google Scholar]
  • 11.Sykes DB, et al. , Inhibition of Dihydroorotate Dehydrogenase Overcomes Differentiation Blockade in Acute Myeloid Leukemia. Cell, 2016. 167(1): p. 171–186.e15. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 12.Wang X, et al. , Targeting pyrimidine synthesis accentuates molecular therapy response in glioblastoma stem cells. Science Translational Medicine, 2019. 11(504): p. eaau4972. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 13.Biancur DE, et al. , Functional Genomics Identifies Metabolic Vulnerabilities in Pancreatic Cancer. Cell Metabolism, 2020. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 14.Zhu XG, et al. , Functional Genomics In Vivo Reveal Metabolic Dependencies of Pancreatic Cancer Cells. Cell Metabolism, 2020. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 15.Brown KK, et al. , Adaptive Reprogramming of De Novo Pyrimidine Synthesis Is a Metabolic Vulnerability in Triple-Negative Breast Cancer. Cancer Discovery, 2017. 7(4): p. 391–399. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 16.Santana-Codina N, et al. , Oncogenic KRAS supports pancreatic cancer through regulation of nucleotide synthesis. Nat Commun, 2018. 9(1): p. 4945. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 17.Bajzikova M, et al. , Reactivation of Dihydroorotate Dehydrogenase-Driven Pyrimidine Biosynthesis Restores Tumor Growth of Respiration-Deficient Cancer Cells. Cell Metab, 2019. 29(2): p. 399–416.e10. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 18.Shukla SK, et al. , MUC1 and HIF-1alpha Signaling Crosstalk Induces Anabolic Glucose Metabolism to Impart Gemcitabine Resistance to Pancreatic Cancer. Cancer Cell, 2017. 32(1): p. 71–87.e7. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 19.Zhang P, et al. , Targeting OXPHOS de novo purine synthesis as the nexus of FLT3 inhibitor–mediated synergistic antileukemic actions. Science Advances, 2022. 8(37): p. eabp9005. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 20.Mao C, et al. , DHODH-mediated ferroptosis defence is a targetable vulnerability in cancer. Nature, 2021. 593(7860): p. 586–590. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 21.Eriksen-Gjerstad M, et al. , Dihydroorotate dehydrogenase inhibition acts synergistically with tyrosine kinase inhibitors to induce apoptosis of mantle cell lymphoma cells. EJHaem, 2022. 3(3): p. 913–918. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 22.Shi DD, et al. , De novo pyrimidine synthesis is a targetable vulnerability in IDH mutant glioma. Cancer Cell, 2022. 40(9): p. 939–956.e16. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 23.Pal S, et al. , A druggable addiction to de novo pyrimidine biosynthesis in diffuse midline glioma. Cancer Cell, 2022. 40(9): p. 957–972.e10. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 24.Olsen TK, et al. , DHODH is an independent prognostic marker and potent therapeutic target in neuroblastoma. JCI Insight, 2022. 7(17). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 25.He D, et al. , De novo pyrimidine synthesis fuels glycolysis and confers chemoresistance in gastric cancer. Cancer Lett, 2022: p. 215837. [DOI] [PubMed] [Google Scholar]
  • 26.So J, et al. , Inhibition of pyrimidine biosynthesis targets protein translation in acute myeloid leukemia. EMBO Mol Med, 2022. 14(7): p. e15203. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 27.Dai M, et al. , Nuclear-translocation of ACLY induced by obesity-related factors enhances pyrimidine metabolism through regulating histone acetylation in endometrial cancer. Cancer Letters, 2021. 513: p. 36–49. [DOI] [PubMed] [Google Scholar]
  • 28.Maroun J, et al. , Multicenter phase II study of brequinar sodium in patients with advanced lung cancer. Cancer Chemotherapy and Pharmacology, 1993. 32(1): p. 64–66. [DOI] [PubMed] [Google Scholar]
  • 29.Cody R, et al. , Multicenter phase II study of brequinar sodium in patients with advanced breast cancer. Am J Clin Oncol, 1993. 16(6): p. 526–8. [DOI] [PubMed] [Google Scholar]
  • 30.Natale R, et al. , Multicenter phase II trial of brequinar sodium in patients with advanced melanoma. Ann Oncol, 1992. 3(8): p. 659–60. [DOI] [PubMed] [Google Scholar]
  • 31.Moore M, et al. , Multicenter phase II study of brequinar sodium in patients with advanced gastrointestinal cancer. Invest New Drugs, 1993. 11(1): p. 61–5. [DOI] [PubMed] [Google Scholar]
  • 32.Urba S, et al. , Multicenter phase II trial of brequinar sodium in patients with advanced squamous-cell carcinoma of the head and neck. Cancer Chemother Pharmacol, 1992. 31(2): p. 167–9. [DOI] [PubMed] [Google Scholar]
  • 33.Olou AA, et al. , MUC1 oncoprotein mitigates ER stress via CDA-mediated reprogramming of pyrimidine metabolism. Oncogene, 2020. 39(16): p. 3381–3395. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 34.Shukla SK, et al. , Silibinin-mediated metabolic reprogramming attenuates pancreatic cancer-induced cachexia and tumor growth. Oncotarget, 2015. 6(38): p. 41146–61. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 35.King RJ, et al. , CD73 induces GM-CSF/MDSC-mediated suppression of T cells to accelerate pancreatic cancer pathogenesis. Oncogene, 2022. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 36.Münster A-K, et al. , Nuclear Localization Signal of Murine CMP-Neu5Ac Synthetase Includes Residues Required for Both Nuclear Targeting and Enzymatic Activity *. Journal of Biological Chemistry, 2002. 277(22): p. 19688–19696. [DOI] [PubMed] [Google Scholar]
  • 37.van Kuilenburg AB and Meinsma R, The pivotal role of uridine-cytidine kinases in pyrimidine metabolism and activation of cytotoxic nucleoside analogues in neuroblastoma. Biochim Biophys Acta, 2016. 1862(9): p. 1504–12. [DOI] [PubMed] [Google Scholar]
  • 38.Huang P, et al. , Action of 2′,2′-Difluorodeoxycytidine on DNA Synthesis. Cancer Research, 1991. 51(22): p. 6110. [PubMed] [Google Scholar]
  • 39.Saiki Y, et al. , DCK is frequently inactivated in acquired gemcitabine-resistant human cancer cells. Biochemical and Biophysical Research Communications, 2012. 421(1): p. 98–104. [DOI] [PubMed] [Google Scholar]
  • 40.Månsson E, et al. , Down-regulation of deoxycytidine kinase in human leukemic cell lines resistant to cladribine and clofarabine and increased ribonucleotide reductase activity contributes to fludarabine resistance. Biochemical Pharmacology, 2003. 65(2): p. 237–247. [DOI] [PubMed] [Google Scholar]
  • 41.Lachmann N, et al. , Deoxycytidine-kinase knockdown as a novel myeloprotective strategy in the context of fludarabine, cytarabine or cladribine therapy. Leukemia, 2015. 29(11): p. 2266–2269. [DOI] [PubMed] [Google Scholar]
  • 42.Kang N, et al. , Human equilibrative nucleoside transporter-3 (hENT3) spectrum disorder mutations impair nucleoside transport, protein localization, and stability. J Biol Chem, 2010. 285(36): p. 28343–52. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 43.Xia L, et al. , Membrane localization and pH-dependent transport of a newly cloned organic cation transporter (PMAT) in kidney cells. Am J Physiol Renal Physiol, 2007. 292(2): p. F682–90. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 44.Barnes K, et al. , Distribution and functional characterization of equilibrative nucleoside transporter-4, a novel cardiac adenosine transporter activated at acidic pH. Circ Res, 2006. 99(5): p. 510–9. [DOI] [PubMed] [Google Scholar]
  • 45.Pastor-Anglada M and Pérez-Torras S, Emerging Roles of Nucleoside Transporters. Front Pharmacol, 2018. 9: p. 606. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 46.Miller SR, et al. , Nucleoside Reverse Transcriptase Inhibitor Interaction with Human Equilibrative Nucleoside Transporters 1 and 2. Drug Metabolism and Disposition, 2020. 48(7): p. 603. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 47.Ward JL and Tse CM, Nucleoside transport in human colonic epithelial cell lines: evidence for two Na+-independent transport systems in T84 and Caco-2 cells1Portions of this work were published as an abstract in Gastroenterology 14 (1998) A431.1. Biochimica et Biophysica Acta (BBA) - Biomembranes, 1999. 1419(1): p. 15–22. [DOI] [PubMed] [Google Scholar]
  • 48.Klotz L, et al. , Teriflunomide treatment for multiple sclerosis modulates T cell mitochondrial respiration with affinity-dependent effects. Science Translational Medicine, 2019. 11(490): p. eaao5563. [DOI] [PubMed] [Google Scholar]
  • 49.Breedveld F and Dayer J, Leflunomide: mode of action in the treatment of rheumatoid arthritis. Annals of the rheumatic diseases, 2000. 59(11): p. 841–849. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 50.Klotz L, et al. , Teriflunomide treatment for multiple sclerosis modulates T cell mitochondrial respiration with affinity-dependent effects. Science Translational Medicine, 2019. 11(490): p. eaao5563. [DOI] [PubMed] [Google Scholar]
  • 51.Ullrich A, et al. , Recombinant expression of N-terminal truncated mutants of the membrane bound mouse, rat and human flavoenzyme dihydroorotate dehydrogenase. A versatile tool to rate inhibitor effects? Eur J Biochem, 2001. 268(6): p. 1861–8. [PubMed] [Google Scholar]
  • 52.Gunderson AJ, et al. , Bruton Tyrosine Kinase–Dependent Immune Cell Cross-talk Drives Pancreas Cancer. Cancer Discovery, 2016. 6(3): p. 270–285. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 53.Gu Z, et al. , Tumor microenvironment and metabolic remodeling in gemcitabine‐based chemoresistance of pancreatic cancer. Cancer Letters, 2021. 521: p. 98–108. [DOI] [PubMed] [Google Scholar]
  • 54.Yu Y, et al. , Therapeutic targeting of both dihydroorotate dehydrogenase and nucleoside transport in MYCN-amplified neuroblastoma. Cell Death & Disease, 2021. 12(9): p. 821. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 55.Cuthbertson CR, et al. , The Dihydroorotate Dehydrogenase Inhibitor Brequinar Is Synergistic with ENT1/2 Inhibitors. ACS Pharmacology & Translational Science, 2020. 3(6): p. 1242–1252. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 56.Krishan A, et al. , Synergistic Effect of Prochlorperazine and Dipyridamole on the Cellular Retention and Cytotoxicity of Doxorubicin. Clinical Cancer Research, 2000. 6(4): p. 1508. [PubMed] [Google Scholar]
  • 57.Beiki D, et al. , Comparative Bioavailability of Two Tablet Formulations of Dipyridamole in Healthy Volunteers. Iranian Journal of Pharmaceutical Research, 2003. 2. [Google Scholar]
  • 58.Abt ER, et al. , Metabolic Modifier Screen Reveals Secondary Targets of Protein Kinase Inhibitors within Nucleotide Metabolism. Cell Chem Biol, 2020. 27(2): p. 197–205.e6. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 59.Gu D, et al. , Targeting Bruton tyrosine kinase using non-covalent inhibitors in B cell malignancies. Journal of Hematology & Oncology, 2021. 14(1): p. 40. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 60.Yao SYM, et al. , Role of cysteine 416 in N-ethylmaleimide sensitivity of human equilibrative nucleoside transporter 1 (hENT1). Biochem J, 2018. 475(20): p. 3293–3309. [DOI] [PubMed] [Google Scholar]
  • 61.Rushworth SA, et al. , Identification of Bruton’s tyrosine kinase as a therapeutic target in acute myeloid leukemia. Blood, 2014. 123(8): p. 1229–1238. [DOI] [PubMed] [Google Scholar]
  • 62.Sagar S, et al. , MUC4 enhances gemcitabine resistance and malignant behaviour in pancreatic cancer cells expressing cancer-associated short O-glycans. Cancer Letters, 2021. 503: p. 91–102. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 63.Radhakrishnan P, et al. , Immature truncated O-glycophenotype of cancer directly induces oncogenic features. Proceedings of the National Academy of Sciences, 2014. 111(39): p. E4066–E4075. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 64.Rodriguez E, et al. , Sialic acids in pancreatic cancer cells drive tumour-associated macrophage differentiation via the Siglec receptors Siglec-7 and Siglec-9. Nature Communications, 2021. 12(1): p. 1270. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 65.Lee D. h., et al. , Genome wide CRISPR screening reveals a role for sialylation in the tumorigenesis and chemoresistance of acute myeloid leukemia cells. Cancer Letters, 2021. 510: p. 37–47. [DOI] [PubMed] [Google Scholar]
  • 66.Lam C, et al. , The hexosamine biosynthetic pathway and cancer: Current knowledge and future therapeutic strategies. Cancer Letters, 2021. 503: p. 11–18. [DOI] [PubMed] [Google Scholar]
  • 67.Chen Y, et al. , Sublethal heat stress-induced O-GlcNAcylation coordinates the Warburg effect to promote hepatocellular carcinoma recurrence and metastasis after thermal ablation. Cancer Letters, 2021. 518: p. 23–34. [DOI] [PubMed] [Google Scholar]
  • 68.Yuan Y, et al. , Exosomal O-GlcNAc transferase from esophageal carcinoma stem cell promotes cancer immunosuppression through up-regulation of PD-1 in CD8+ T cells. Cancer Letters, 2021. 500: p. 98–106. [DOI] [PubMed] [Google Scholar]
  • 69.Ferrer CM, Sodi VL, and Reginato MJ, O-GlcNAcylation in Cancer Biology: Linking Metabolism and Signaling. Journal of molecular biology, 2016. 428(16): p. 3282–3294. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 70.Hanover JA, Chen W, and Bond MR, O-GlcNAc in cancer: An Oncometabolism-fueled vicious cycle. Journal of Bioenergetics and Biomembranes, 2018. 50(3): p. 155–173. [DOI] [PubMed] [Google Scholar]
  • 71.Demarest JF, et al. , Brequinar and dipyridamole in combination exhibits synergistic antiviral activity against SARS-CoV-2 in vitro: Rationale for a host-acting antiviral treatment strategy for COVID-19. Antiviral Res, 2022. 206: p. 105403. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 72.Choi DS, et al. , The type 1 equilibrative nucleoside transporter regulates ethanol intoxication and preference. Nat Neurosci, 2004. 7(8): p. 855–61. [DOI] [PubMed] [Google Scholar]
  • 73.Chen J, et al. , The type 1 equilibrative nucleoside transporter regulates anxiety-like behavior in mice. Genes Brain Behav, 2007. 6(8): p. 776–83. [DOI] [PMC free article] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

Supplementary Table 1
NIHMS1907807-supplement-Supplementary_Table_1.xlsx (9.3KB, xlsx)
Supplementary Figures and Legends

Figure S1: BQ-sensitive CFPAC-1 cells undergo greater pyrimidine depletion than BQ-resistant S2–013 cells. A) Volcano plot for significantly altered metabolites in BQ-treated vs control groups for CFPAC-1 (top) or S2–013 (bottom) cells, p < 0.05 and fold change > 2 was considered significant. B) Quantification of relative abundance of indicated metabolites, normalized to the respective vehicle-treated control. Data represent mean +/− SEM of 4 biological replicates. * p < 0.05, **** p < 0.0001 by two-way ANOVA with Bonferroni’s post-hoc test. C) Schematic explaining how CTP depletion leads to accumulation of Neu5Ac.

Figure S2: Bruton tyrosine kinase (BTK), the putative target of CNX-774, is not expressed in human PDAC samples nor human PDAC cell lines A) Percent of human samples with positive BTK expression as assessed by immunohistochemical staining with antibody HPA001198, from Human Protein Atlas (https://www.proteinatlas.org/ENSG00000010671-BTK/pathology).

Figure S3: Exogenous nucleosides can rescue BQ-induced toxicity A) BQ dose-response viability experiment in the presence or absence of 100μM uridine. Data represent mean +/− SEM of three biological replicates.

Figure S4: A) Volcano plot of significantly altered metabolites in sgDHODH-A28 vs sgNT.

Figure S5: Exploration of CNX-774 and NBMPR interactions with additional nucleoside analogs

A-D) Dose response cell viability experiments with indicated nucleoside analogs in the presence or absence of CNX-774 (left two panels) or NBMPR (right two panels), with cartoon showing nucleoside analog activation (middle). Data represent mean +/− SEM of three biological replicates.

NIHMS1907807-supplement-Supplementary_Figures_and_Legends.docx (8.1MB, docx)

RESOURCES