Characterization of genetically modified human embryonic kidney 293 cells lacking equilibrative nucleoside transporter subtype 2, or both subtypes 1 and 2, and the impact of their loss on sensitivity to chemotherapeutic purine/pyrimidine analogs

Abstract

Equilibrative nucleoside transporters (ENTs) 1 and 2 are considered critical to the cellular uptake of purine and pyrimidine analogs used to treat cancer and viral infections. However, a detailed understanding of the discrete and overlapping roles of these ENT subtypes in drug activity remains limited. A significant barrier to progress has been the absence of model systems that enable functional characterization of individual nucleoside transporters in the context of their native environment. To address this, we developed and characterized a panel of CRISPR/cas9-engineered human embryonic kidney 293 cell lines with selective deletion of ENT subtypes: ENT1 knockout, ENT2 knockout, and dual knockout. These models were used to dissect subtype-specific roles of ENT1 and ENT2 in nucleoside/nucleobase analog uptake and cytotoxicity. Our data show that ENT1 and ENT2 in their endogenous environment have a similar affinity for a range of both endogenous and chemotherapeutic nucleoside and nucleobase analogs. Deletion of ENT1 generally enhanced the sensitivity of cells to these drugs, particularly the nucleobase analogs, likely due to reduced nucleoside salvage by the cells via ENT1. Deletion of ENT2, on the other hand, dramatically reduced the ability of a number of the tested drugs to impact cell viability, by mechanisms beyond those related to reduced cellular uptake of the drugs. This study highlights distinctive roles of ENT1 and ENT2 in the actions of nucleoside/nucleobase analog drugs.

Significance Statement

A panel of genetically modified human embryonic kidney 293 cells has been created as a model to screen novel nucleoside transporter inhibitors and substrates. Using these cell lines, it was revealed that ENT2 may play a more functionally significant role in nucleoside analog chemotherapeutic drug activity than previously appreciated.

1. Introduction

Nucleoside and nucleobase analogs remain foundational chemotherapeutic agents for the management of diverse malignancies and viral infections due to their effects on nucleic acid metabolism and synthesis.¹ The clinical efficacy of these drugs is influenced by their ability to enter cells, a process reliant on specialized membrane transporters such as the equilibrative nucleoside transporters (ENTs, specifically ENT1 and ENT2) and the concentrative nucleoside transporters (CNTs).1, 2, 3, 4, 5 Among these, ENT1 has typically been highlighted as the primary nucleoside transporter involved in the uptake of chemotherapeutic nucleoside analogs.^6,7 However, reduced ENT2 expression at the plasma membrane has also been linked to chemoresistance.⁸ Notably, ENT2 may be important for the transport of nucleobase analog drugs such as 5-fluorouracil, for which ENT1 status does not reliably predict cellular response.^9,10 ENT2 upregulation in various malignancies also suggests a potential involvement in tumor-specific drug responses and resistance mechanisms.^4,11 Despite the established importance of ENTs, a detailed understanding of their discrete and overlapping roles in drug uptake and the regulation of their expression remains limited. A significant barrier to progress has been the absence of model systems that enable functional characterization of individual nucleoside transporters in the context of their native environment. To address this, we have established CRISPR/cas9-engineered human embryonic kidney 293 (HEK293) cell models deficient in ENT1 (ENT1 knockout [ENT1KO]), ENT2 (ENT2 knockout [ENT2KO]), and both ENT1 and ENT2 (ENT1 and ENT2 knockout [E1E2KO]). The characterization of the ENT1KO cell model has been reported by us previously.¹² The development and characterization of the ENT2KO and E1E2KO cell lines are reported herein. The current study illustrates how this platform of genetically engineered cell lines may be used to dissect transporter-specific contributions to drug uptake and effects on cellular function.

2. Materials and methods

2.1. Materials

[8-³H]2-Chloroadenosine (1 mCi/mL), [³H]nitrobenzylmercaptopurine riboside (nitrobenzylthioinosine) ([³H]NBMPR) (1 mCi/mL), and [³H]water (1 mCi/g) were obtained from Moravek Biochemicals. HEK293 cells were purchased from American Type Culture Collection. Dipyridamole (DY), geneticin, Dulbecco’s modified Eagle’s medium, FBS, penicillin-streptomycin solution, 2-mercaptoethanol, light mineral oil, 2-chloroadenosine, guanosine, adenosine, thymidine, uridine, cytosine, adenine, hypoxanthine, clofarabine, cytarabine, and gemcitabine were purchased from Sigma-Aldrich. NBMPR, dilazep, azidothymidine, 5-fluorouracil, and cladribine were purchased from Tocris Bioscience. All primers, Alt-R crRNA (20 nmol scale), Alt-R ATTO-550 tracrRNA (20 nmol scale), and nuclease-free Duplex Buffer, were ordered through Integrated DNA Technologies. Agarose, Oligo (dT)12–18 primer, HEPES, sodium pyruvate, thiazolyl blue tetrazolium bromide (MTT), bicinchoninic acid protein assay kit, PowerUp SYBR Green, Halt Protease Inhibitor Cocktail, TRIzol Reagent, SuperScript III Reverse Transcriptase and Taq DNA Polymerase, and DreamTaq PCR Master Mixes were obtained from Thermo Fisher Scientific. The 100 bp DNA Ladder was supplied by Truin Science. EnGen SpyCas9 (M0646T) was from New England Biolabs. Lipofectamine CRISPRMAX (CMAX00-001) and Opti-MEM were from Thermo Fisher Scientific. The antibodies used for immunoblotting were mouse monoclonal IgG1 anti-Myc antibody (Clone 4A6, 05-724, Lot #3013479; EMD Millipore), monoclonal mouse anti-ENT1 (F-12: sc-377283; Santa Cruz), polyclonal rabbit anti-ENT2 (Invitrogen PA572317; Fisher Scientific), and mouse monoclonal anti–β-actin (C4, sc-47778, Lot #B0719; Santa Cruz Biotechnology Inc). Secondary antibodies, mouse anti-rabbit IgG–horseradish peroxidase (HRP) (sc-2357, Lot #2517) and m-IgGk binding protein-HRP (sc-516102, Lot #F1016), were purchased from Santa Cruz Biotechnology. The ECL Prime western blotting system was purchased from Cytiva Life Sciences. The Ecolite liquid scintillation cocktail was purchased from MP Biomedical. The Dow DOWSIL 550 Silicone Fluid (silicone oil) was from Ellsworth Adhesives Canada Corporation.

2.2. Generation of knockout cell models

CRISPR/cas9 guide RNA (5′-CGGGGCGTCTCCTCGCGCC-3′) targeting human SLC29A2 was prepared as previously described,^12,13 using Alt-R CRISPR/cas9 tracrRNA, ATTO-550. The genomic region targeted corresponded to bases 66371571–66371589 on exon 1 of the SLC29A2 gene on Homo sapiens chromosome 11 (NC_000011.10; GRCh38.p14 Primary Assembly), which is immediately upstream of the translation initiation codon of the transcribed mRNA (Fig. 1A). mRNA-EnGen Spy Cas9 NLS (New England Biolabs) was assembled into ribonucleoprotein complexes and transfected into HEK293 cells (to create the ENT2KO cell line) and the previously generated¹² ENT1KO cells (to create the E1E2KO cell line) using Lipofectamine CRISPRMAX as per the manufacturer’s instructions. Twenty-four hours following transfection, cells positive for ATTO-550 were sorted into the wells of a 96-well plate (one cell per well in growth media) on a BD FACSAria III instrument by the Flow Cytometry Core at the University of Alberta. Clonal growth was monitored for the next 2–3 weeks until approximately 70% confluency was reached; then approximately 1 × 10⁵ cells were lysed in DirectPCR Lysis Reagent (Viagen Biotech, Inc) containing 1 mg/mL Proteinase K (Thermo Fisher) for 16 hours at 55 °C. Remaining cells were transferred to a 24-well plate for continued growth. Following lysis, Proteinase K was heat inactivated at 85 °C for 45 minutes, and 1 μL of lysate was used as a template for polymerase chain reaction (PCR) amplification of the corresponding target site. Genomic DNA was isolated from the monoclonal populations and analyzed via Sanger sequencing, and compared with HEK293 cells, to confirm cas9-mediated gene disruption (Fig. 1, B and C).

Fig. 1.

CRISPR/cas9-mediated disruption of the SLC29A2 gene. (A) Intron/exon arrangement of SLC29A2 on Homo sapiens chromosome 11 (NC_000011.10; GRCh38.p14 primary assembly). The base sequence of the 3′ end of exon 1 is shown along with the predicted cas9-target site based on the guide RNA (gRNA) indicated. The start of the translation initiation codon is also indicated by a blue-shaded box. (B) Chromatogram obtained from Sanger sequencing of the wild-type (WT) HEK293 cells, and (C) the CRISPR/cas9 modified exon 1 of SLC29A2 showing the genomic sequence disruption at the predicted target region in ENT2KO cells.

2.3. Cell culture and proliferation assessment

The cells were cultured in Dulbecco’s modified Eagle’s medium containing penicillin (100 U/mL), streptomycin (100 μg/mL), sodium pyruvate (1 mM), and 10% FBS for ENT1KO and parental HEK293 cells and 20% FBS for ENT2KO and E1E2KO cell lines (which both exhibited reduced proliferation capacity at early passage). The cells were washed with PBS and then removed from flasks by exposure to 0.05% trypsin/0.18 mM EDTA for 10 minutes at 37 °C, and the suspended cells were then washed in PBS buffer solution immediately prior to use in subsequent assays. To assess the rate of cell proliferation, each cell line was seeded at 1.2 × 10⁵ cells per well in a 6-well plate. After every 24 hours for 4 consecutive days, the cells were harvested as mentioned above. An aliquot (10 μL) was then assessed for cell density using a Bio-Rad TC10 Automated Cell Counter (Bio-Rad Laboratories).

2.4. Flow cytometry/cell cycle analysis

Cells were harvested 24–48 hours after seeding, washed twice in PBS, and adjusted to a concentration of 1 × 10⁶ cells per tube. After centrifugation at 700 × g for 10 minutes, the supernatant solution was removed and the cell pellet fixed in ice-cold 70% ethanol (adding dropwise to the pellet with gentle mixing to ensure fixation of all cells and to minimize clumping) for 30 minutes on ice. After fixation, cells were again centrifuged and washed twice with cold PBS, and then resuspended in staining buffer (PBS containing 100 μg/mL RNase A and 50 μg/mL propidium iodide). The final solution (1 × 10⁶ cells in 1 mL of PBS) was filtered through a 100-μm filter/cell strainer to remove any large clumps and then transferred to 5 mL tubes, wrapped in foil to protect from light, and incubated overnight at 4 °C. The following day, the samples were analyzed for cell cycle stage using an Attune NxT flow cytometer (Thermo Fisher Scientific) in the University of Alberta Flow Cytometry Core Facility.

2.5. Immunoblotting

HEK293, ENT1KO, ENT2KO, and E1E2KO cell lysates were extracted in radioimmunoprecipitation buffer (150 mM NaCl, 50 mM Tris, 1% NP-40, 0.5% sodium deoxycholate, 1% SDS) containing Halt Protease Inhibitor Cocktail + EDTA. The concentrations of the protein extract were assessed using a bicinchoninic acid–based assay, and then diluted to 1.5 μg/μL protein and adjusted to 2% β-mercaptoethanol for reducing conditions. Proteins (30 μg) were resolved using sodium dodecyl sulfate-polyacrylamide gel electrophoresis on 12.5% acrylamide gels at 80 V for 15 minutes, followed by 150 V for 45 minutes. Proteins were then electrotransferred from gel onto Immobilon-P polyvinylidene difluoride membranes (Millipore Corporation) for 1.5 hours at a constant current of 280 mA. Following transfer, membranes previously rinsed in Tris-buffered saline (TBS-T: 150 mM NaCl, 50 mM Tris, pH 7.5 containing 0.1% v/v Tween-20) were incubated in 5% w/v skim milk (TBS-TM1: 2.5 g of milk powder in 50 mL of TBS-T) for 1 hour at room temperature with gentle rocking to block nonspecific binding of the polyvinylidene difluoride membrane. Membranes were then incubated overnight (approximately 16 hours) at 4 °C with gentle rocking in the presence of primary antibodies such as mouse anti-ENT1, rabbit anti-ENT2, or mouse anti–β-actin at 1:1000, 1:500, 1:1000, and 1:500 dilution, respectively, in TBS-TM2 (containing 1% skim milk powder). The following day, the membranes were washed 4–5 times with TBS-TM2 and then incubated for 1–2 hours at room temperature with anti-rabbit IgG-HRP or anti-mouse IgG-HRP at 1:5000 or 1:3000 dilution, respectively, in TBS-TM2. Following this incubation, membranes were further washed (4–5 times) with TBS-T, and protein was detected using ECL Prime solution and visualized via chemiluminescence using an Amersham Imager 680 (GE Healthcare). ImageJ software¹⁴ was used to conduct densitometry analyses.

2.6. Real-time semiquantitative PCR

For the extraction of RNA, cells from 10-cm plates were suspended in 1 mL of TRIzol reagent and homogenized according to the manufacturer’s protocol (Thermo Fisher Scientific). Total RNA concentration and purity were determined using a NanoDrop 2000 spectrophotometer (Life Technologies Inc). Two micrograms of total RNA were reverse transcribed to cDNA using Oligo (dT)12–18 primer and SuperScript III Reverse Transcriptase (Thermo Fisher Scientific). For semiquantitative reverse transcriptase PCR, cDNA (approximately 200 ng/well) was prepared with the primer sets specific for the genes, as shown in Supplemental Table 1, using PowerUp SYBR Green fluorescence on a Roche Light Cycler 480 System (Cardiovascular Research Centre, University of Alberta, Edmonton, Canada). Primer efficiency and melt curves were assessed to validate the primers prior to their use for gene expression analyses. Semiquantitative reverse transcriptase PCR conditions were 2 minutes at 50 °C (uracil-DNA glycosylase activation), 2 minutes at 95 °C (denaturation), followed by 50 cycles of 15 seconds at 95 °C, and 60 seconds at 60 °C for amplification, with a final melt curve analysis. Gene expression was normalized to either glyceraldehyde-3-phosphate dehydrogenase (GAPDH) alone, or the geometric mean of 3 separate reference genes, GAPDH, ribosomal RNA 18S (nuclear variant 5, RNA18SN5), and β-actin (ACTB), and the relative expression of the genes was determined using the ΔΔCt method with HEK293 cells as the comparator.

2.7. [³H]NBMPR binding

The high-affinity binding of the selective ENT1-probe [³H]NBMPR was assessed to quantify the number of ENT1 proteins in each cell line. Approximately 1 × 10⁶ cells suspended in PBS were incubated at room temperature with a range of concentrations of [³H]NBMPR in the absence (total binding) and presence (nonspecific binding) of 10 μM DY. When steady-state binding was reached (approximately 45 minutes), cells were filtered through Whatman GF/B filters (Fisher Scientific) using a 24-port Brandel Cell Harvester (Brandel Biomedical Research). [³H] content on each filter was then determined using liquid scintillation techniques. Specific binding was defined as the difference between total binding and nonspecific binding. K_d and B_max values were determined by fitting a hyperbolic curve to the site-specific [³H]NBMPR binding versus the equilibrium free concentration of [³H]NBMPR.

2.8. [³H]2-Chloroadenosine transport assay

To assess transporter function, cell lines were examined for their ability to accumulate [³H]2-chloroadenosine, an established ENT substrate that is poorly metabolized by adenosine kinase and adenosine deaminase.^15,16 To eliminate the potential contribution of sodium-dependent CNTs, cells were suspended in sodium-free buffer (140 mM N-methylglucamine, 5 mM KCl, 4.2 mM KHCO₃, 0.36 mM K₂HPO₄, 0.44 mM KH₂PO₄, 10 mM HEPES, 0.5 mM MgCl₂, 1.3 mM CaCl₂, pH 7.4) containing 0.1% dimethyl sulfoxide (solvent control; total uptake), 100 nM NBMPR (to selectively block ENT1), 5 mM adenosine (complete ENT1 and ENT2 block; nonmediated uptake), or 3 μM NBMPR/DY (alternative nonmediated uptake determination). These normally adherent cells were removed from the plates and used in suspension to allow for the rapid kinetic analysis of substrate flux by ENTs, which requires time points in the 3–15-second range. Following a 15-minute incubation at room temperature, 250 μL of cell suspension was added to 250 μL of [³H]2-chloroadenosine layered over 21:4 (v:v) silicone:mineral oil (200 μL) in microcentrifuge tubes. The uptake reaction was terminated after 10 seconds by centrifugation of the cells through the oil layer at 10,000 × g. This incubation time was selected from preliminary time courses of 10 μM [³H]2-chloroadenosine uptake in each cell model as the shortest time point technically achievable that approximated the initial near-linear portion of the hyperbolic relationship (initial rate estimate) while allowing a sufficient signal-to-noise ratio for analysis. After the reactions were terminated by centrifugation, the aqueous layer was aspirated, and the tube above the oil was washed with approximately 1 mL of deionized water. The water and the oil were then sequentially removed, leaving only the cell pellet. This pellet was digested in 1 M sodium hydroxide for approximately 24 hours after which the radioactive content of each sample was measured using liquid scintillation spectroscopy. Mediated uptake was defined as the difference between total and nonmediated uptake. The nonmediated component reflects passive diffusion of [³H]2-chloroadenosine into the cells as well as extracellular [³H]2-chloroadenosine remaining in the cell pellet after processing the samples. Data were analyzed as initial rate of uptake (picomole per microliter per second) against the concentration of 2-chloroadenosine used. Cell volume (microliters) was measured by incubating cells with [³H]-water for 2 minutes, centrifuging through 21:4 silicone:mineral oil, and then determining the [³H] content of the cell pellet and the supernatant. Total cellular volume was defined as the ratio of the disintegrations per minute of the cell pellet to the disintegrations per minute per microliter of the supernatant and was used to define the V_max values as picomole per microliter per second. Michaelis-Menten curves were fit to the concentration-dependent uptake data for determination of K_m and V_max values. Statistical differences between K_m and V_max values were determined using the extra sum-of-squares F test. Significant differences between groups were assessed using a one-way or two-way ANOVA, as appropriate, corrected for multiple comparisons with the Dunnett method. For inhibition studies, 10 μM [³H]2-chloroadenosine and a range of concentrations of the inhibitors/substrates were used with a 15-second incubation time. Cells were preincubated with established ENT inhibitors (ie, NBMPR, dilazep, and DY) for 15 minutes prior to the assay. However, no preincubation was performed with the test substrates (which are readily metabolized by these cells), which were added to the assay concomitant with the [³H]2-chloroadenosine. All compounds were initially tested at the highest concentration possible as limited by aqueous solubility. Those that showed greater than 50% inhibition at these concentrations were subsequently assessed over a broad concentration range to determine their IC₅₀ values for inhibition of [³H]2-chloroadenosine uptake in the various cell lines. K_i values were then calculated using the Cheng-Prusoff equation¹⁷ based on the IC₅₀ values derived from the concentration-response curves and the K_m of [³H]2-chloroadenosine for ENT1 and ENT2 in the ENT2KO and ENT1KO cell models.

2.9. Cell viability analysis

The impact of SLC29A1 and SLC29A2 deletion on the sensitivity of ENT1KO, ENT2KO, and E1E2KO cells to chemotherapeutic drugs was assessed using the MTT-based cell viability assay. Cells were seeded into 24-well plates at a density of 1.1 × 10⁵ cells per well in complete culture medium. One hour following plate seeding, complete medium containing a range of concentrations of the test compound was added and incubated for 48 hours at 37 °C in a humidified 5% CO₂ atmosphere (compound-mediated response). To determine the baseline cell viability (100% cell viability), cells were treated with medium containing DMSO only (at the concentration used to dissolve the test compounds). Medium containing the test compound in the absence of cells was used to determine the 0% cell viability (background). After 48-hour treatment, the media was removed and replaced with 250 μL of Dulbecco’s PBS (137 mM NaCl, 2.7 mM KCl, 6.3 mM Na₂HPO₄, 1.5 mM KH₂PO₄, 0.5 mM MgCl₂, 0.9 mM CaCl₂, pH 7.4) containing MTT (1 mg/mL) for 90 minutes at 37 °C in a humidified cell culture incubator. The resultant formazan crystals were solubilized in 450 μL of DMSO and transferred to a 96-well plate in 195 μL duplicates. Absorbance was measured at 570 nm in a SpectraMax i3x Multi-Mode Detection Platform (Molecular Devices). Percent cell viability was calculated by dividing compound-mediated response by 100% cell viability after subtraction of the background and plotted as % cell viability versus log drug concentration. These concentration-response curves were analyzed for their shape (monophasic or biphasic) to assess the presence of distinct sensitive and resistant cell populations. For biphasic curves, the data were fitted to a 2-site competition model to quantify the sensitive and resistant cell populations, calculated as the percentage of cells showing reduced viability in the first phase (sensitive) and second phase (resistant), respectively. To determine the best fit, an extra sum-of-squares F test was performed (GraphPad Prism 10.6, GraphPad) to compare the biphasic and monophasic models, with a significant difference (P < .05) indicating a better fit to the biphasic curve; if no significant difference was found, the data were fitted to a monophasic curve. LogEC₅₀ values were determined for each phase of the biphasic curve (n = 5, mean ± SEM) (∗P < .05 vs HEK293 cells).

2.10. Data analysis and statistics

A sample size of 5 independent experiments (average of 2–3 technical replicates from 5 different cell lots) was used for most (including the substrate uptake and cell viability analyses) of the experimental sets reported herein. This is the minimum sample size needed for statistical analysis based on the historical variability inherent in these types of studies. Nonlinear curves were fitted to data, and statistical analyses were conducted using GraphPad Prism 10.6 software. In all cases, if the P value determined from a statistical test was less than .05, the difference was considered significant and the null hypothesis (no difference between data sets) was rejected.

3. Results

3.1. Characterization of the ENT2KO and E1E2KO cell models

The deletion of SLC29A2 (ENT2) in the E1E2KO and ENT2KO cell models was confirmed through multiple approaches: (1) absence or presence of ENT1 or ENT2 immunoreactivity, (2) ENT1- versus ENT2-mediated transport activity based on the rate of cellular uptake of [³H]2-chloroadenosine, and (3) assessment of the high-affinity binding of the ENT1-specific probe [³H]NBMPR.

3.2. Protein expression

There was no significant ENT2 protein detected in the ENT2KO or E1E2KO cells, nor was ENT1 protein detected in the E1E2KO cells that were derived from the ENT1KO cell line reported on previously (Fig. 2).¹² Notably, the ENT2KO cells had similar levels of ENT1 protein as the original HEK293 cells. This, combined with the fact that loss of SLC29A1 in the ENT1KO cells did not affect ENT2 expression, suggests a lack of cross-regulation of the expression of these 2 major ENT subtypes in these cells.

Fig. 2.

ENT1 and ENT2 protein expression in HEK293, ENT1KO, ENT2KO, and E1E2KO. (A) Protein was extracted from each cell type, resolved on 12.5% acrylamide gels, and then probed with antibodies specific for β-actin (loading control), ENT1, or ENT2. Data are shown as the expression of ENT1 and ENT2 relative to β-actin. The bars represent the mean ± SEM from 3 independent cell culture plates. (B) Representative immunoblots are shown along with the molecular sizing ladder (L) provided for comparison. ∗Signifies a significant difference from HEK293 cells (P < .05; one-way ANOVA with Dunnett’s multiple comparison post-test).

3.3. [³H]NBMPR binding

The presence or absence of ENT1 was determined using the specific ENT1 probe, [³H]NBMPR. Figure 3A shows the binding of [³H]NBMPR to the base HEK293 cells in the absence (total binding) and presence (nonspecific binding) of DY. Based on these data, the specific binding of [³H]NBMPR (which represents binding to ENT1 protein) had a B_max of 53,400 ± 900 sites/cell and a K_d of 0.063 ± 0.004 nM. The ENT2KO cells had a similar number of [³H]NBMPR binding sites (B_max = 54,700 ± 2900 sites/cell; K_d = 0.062 ± 0.013 nM) (Fig. 3B), but no site-specific (DY-sensitive) [³H]NBMPR binding was detected in the E1E2KO cells (Fig. 3C).

Fig. 3.

Binding of [³H]NBMPR to the ENT1 protein of (A) HEK293, (B) ENT2KO, and (C) E1E2KO cells. Cells were incubated with the indicated concentrations of [³H]NBMPR (abscissa) for 45 minutes in the absence (total binding) or presence (nonspecific binding) of 10 μM DY and then filtering the cell suspension through Whatman GF/B filters. Specific binding in the HEK293 and ENT2KO cells was determined as the difference between total and nonspecific binding. Note that there was no specific binding detected in the E1E2KO cells. The data are shown as the number of molecules of NBMPR bound per cell (ordinate), and each point represents the mean ± SEM from 5 independent experiments conducted in duplicate.

3.4. Impact of ENT1 and ENT2 loss on cell cycle and proliferation of HEK293 knockout cell models

Loss of SLC29A1 in the ENT1KO and E1E2KO cells led to a significant decrease in plating efficiency in the first 24 hours after seeding, whereas the ENT2KO cells were similar to the HEK293 cells in this regard (Fig. 4A). The number of ENT1KO and ENT2KO cells was similar to the HEK293 cell counts on days 2 and 3 of culture, but exceed HEK293 cells in number by 96 hours after plating (Fig. 4A). Removing SLC29A1 in the ENT1KO cells had no significant effect on cell cycle (Fig. 4B). However, deletion of SLC29A2 in the ENT2KO cells led to a slight, but significant, increase in the percentage of cells in the G0/G1 cell cycle stage and a corresponding decrease in S phase (Fig. 4B). In contrast, E1E2KO cells had a significantly reduced proliferation rate across the full culture period (Fig. 4A) and a significant 60% increase in the percentage of the cell population in the G2 cell cycle stage and a corresponding decrease of cells in both the G0/G1 and S stages (Fig. 4B).

Fig. 4.

Effect of SLC29A1 and/or SLC29A2 deletion on cell proliferation and cell cycle. (A) Cell proliferation: cells were plated at equal densities and then counted each day for 4 consecutive days. Data are presented as cell count relative to the original seeding density. Each bar represents the mean ± SEM, n = 10. Significant differences were assessed using two-way ANOVA with Tukey’s multiple comparison post-test. ∗Denotes a significant difference from HEK293 cells (P < .05). (B) Cell cycle analysis: the relative number of cells (harvested 24–36 hours after seeding) in each stage of the cell cycle was assessed by flow cytometry. Each bar represents the mean ± SEM from 3 independent cell lots. Significant differences from HEK293 cells were assessed using two-way ANOVA with Dunnett’s multiple comparison post-test.

3.5. Assessment of compensatory changes in gene expression in ENT1KO, ENT2KO, and E1E2KO cells

To gain an understanding of the cellular response SLC29A2 deletion in HEK293 cells, an exploratory screen of the gene expression levels of other purine and pyrimidine transporters and enzymes involved in nucleoside/nucleobase metabolism was conducted (Fig. 5). Because of limited sample sizes, differences were considered significant only if there was greater than a 2-fold change (t test, P < .05) in expression relative to that found in HEK293 cells.

Fig. 5.

Semiquantitative PCR screen of various nucleoside/nucleobase transporters and enzymes involved in nucleoside metabolism in HEK293 cells. mRNA was extracted from HEK293, (A) ENT2KO, and (B) E1E2KO cells and analyzed by semiquantitative PCR as described in the text. Data were analyzed with the ΔΔCt method using the geometric mean of 3 reference genes (GAPDH, ACTB, and ribosomal RNA 18S) and expression in HEK293 cells (dashed horizontal line) as the comparator. (A, B) Each bar is the mean ± SEM from the number of independent experiments shown above each bar. ∗Indicates a significant difference from the expression in HEK293 cells (1-sample t test, P < .05); shaded blue for an increase and orange for a decrease in expression. (C) For dCK, APOBE3C, and APOBE3D, mRNA was extracted from HEK293, ENT1KO, ENT2KO, and E1E2KO cells and assessed for the expression of these transcripts, in parallel. Each bar represents the mean ± SEM from 5 experiments. ∗Indicates a significant difference in expression from that found in HEK293 cells (one-way ANOVA with Tukey’s post-test for multiple samples, P < .05).

Deletion of SLC29A2 in the ENT2KO cells had a broad impact on the expression of numerous genes associated with nucleoside and nucleobase metabolism (Fig. 5A). Specifically, with respect to other nucleoside and nucleobase transporters, the loss of SLC29A2 resulted in a large 350-fold increase in the expression of SLC28A1 and a 3-fold increase in SLC29A4. However, there was no significant change in the expression of transcript for the other major ENT subtype, SLC29A1. Concerning the enzymes involved in purine metabolism, there was more than a 100-fold increase in ADSS transcript, a 6-fold increase in ADA expression, and a 2.5-fold increase in HPRT with the loss of SLC29A2 in ENT2KO cells.

Loss of both SLC29A1 and SLC29A2 in the E1E2KO cell line (Fig. 5B) resulted in an increase in SLC28A1 similar to that seen for the ENT2KO cells, but also resulted in a significant decrease in SLC28A3 and ABCC4 expression. With respect to the metabolic enzymes, there was an increase in ADSS expression (as was seen for the ENT2KO cells), as well as a significant increase in the expression of IMPDH2 (which also had a trend toward an increase in the ENT2KO cells), and more than a 5-fold decrease in both NT5C and NT5E.

Other genes encoding enzymes that have been reported to contribute to resistance to cytosine/cytidine analog chemotherapeutic drugs such as gemcitabine and cytarabine were also screened in the 4 cell models (Fig. 5C). There were no changes in the expression of DCK (deoxycytidine kinase) or APOBE3C (member of the cytosine deaminase family) transcripts with deletion of either SLC29A1, SLC29A2, or both. However, APOBE3D (encoding a cytosine deaminase) was significantly (one-way ANOVA with Tukey’s post-test for multiple comparisons, P < .05) decreased in the ENT2KO cells.

3.6. Functional assessment of ENT subtype expression based on [³H]2-chloroadenosine uptake

Kinetic analysis of ENT1- and ENT2-mediated uptake of [³H]2-chloroadenosine was conducted in the ENT2KO and E1E2KO cell lines. The data obtained in the base HEK293 cells are shown for comparison (Fig. 6A). Initial rates of uptake of [³H]2-chloroadenosine were measured in the presence or absence of 100 nM NBMPR (to inhibit ENT1) or 5 mM adenosine (to inhibit both ENT1 and ENT2). These experiments confirmed robust ENT1-mediated uptake in the ENT2KO cells with a K_m of 30 ± 13 μM and a V_max of 4.8 ± 0.6 pmol/μL per second (Fig. 6B). This V_max of ENT1-mediated uptake in the ENT2KO cells was found to be significantly higher (unpaired t test, P < .05) than that determined for the HEK293 cells (V_max = 3.2 ± 0.3 pmol/μL per second), leading to a similar total mediated uptake of [³H]2-chloradenosine in the 2 cell lines. ENT2KO cells had no detectable ENT2-mediated uptake (Fig. 6B), functionally confirming the knockout. E1E2KO cells displayed no ENT1-mediated or ENT2-mediated uptake of [³H]2-chloroadenosine (Fig. 6C), verifying the functional absence of both transporters.

Fig. 6.

Kinetics of [³H]2-chloroadenosine transport by (A) HEK293, (B) ENT2KO, and (C) E1E2KO cells. Cells were incubated at room temperature with the indicated concentrations of [³H]2-chloroadenosine (abscissa) for 10 seconds in the absence (total uptake) and presence of 100 nM NBMPR or 5 mM adenosine. The rate of accumulation of the substrate is shown as pmol/μL cell water/s (ordinate). Total mediated uptake (ENT1 + ENT2 mediated) = total uptake minus that determined in the presence of 5 mM adenosine; ENT2-mediated = uptake in the presence of 100 nM NBMPR minus that determined in the presence of 5 mM adenosine; ENT1-mediated uptake is defined as the difference between the total mediated uptake and ENT2-mediated uptake. Each point is the mean ± SEM of 5 independent experiments conducted in duplicate.

3.7. Inhibition of ENT1- and ENT2-mediated 2-chloroadenosine uptake by subtype-selective agents

To further assess the relative contributions of ENT1 and ENT2 to the cellular influx of [³H]2-chloroadenosine in these cell models, full concentration-inhibition profiles for compounds known to distinguish between ENT subtypes were constructed (Fig. 7). It is well established that NBMPR is a specific inhibitor for ENT1 at nanomolar concentrations (eg, 0.1–1 nM), while it requires micromolar concentrations to inhibit ENT2.^18,19 HEK293 cells showed a biphasic NBMPR inhibition profile with high-affinity (logIC₅₀ = −9.2 ± 0.2 M) and low-affinity (logIC₅₀ = −5.8 ± 0.2 M) components consistent with the presence of both ENT1 and ENT2 activity (Fig. 7A). Substrate uptake by the ENT2KO cells was inhibited by NBMPR in a monophasic manner with a logIC₅₀ of −8.9 ± 0.1 M, typical of ENT1, whereas in the ENT1KO cells, NBMPR concentrations in excess of 1 μM were required to obtain any inhibition of [³H]2-chloroadenosine uptake by the cells and had an estimated logIC₅₀ of −5.5 ± 0.1 M consistent with ENT2. Dilazep was similar to NBMPR in that it showed a biphasic inhibition profile in the HEK293 cells (Fig. 7B) with logIC₅₀ values of −7.8 ± 0.3 M and −5.1 ± 0.2 M for the high and low affinity components, respectively. In contrast, only the high-affinity (logIC₅₀ = −8.1 ± 0.2 M) inhibition profile was apparent in the ENT2KO cells, and only the low affinity (logIC₅₀ = −5.2 ± 0.2 M) inhibition component was seen in the ENT1KO cells. DY had monophasic inhibition profiles for inhibiting [³H]2-chloroadenosine uptake in HEK293, ENT2KO, and ENT1KO cells, with logIC₅₀ values of −6.6 ±.01 M, −7.2 ± 0.2 M, and −6.3 ± 0.1 M, respectively (Fig. 7C).

Fig. 7.

Comparative inhibition of [³H]2-chloroadenosine uptake by (A) NBMPR, (B) dilazep, and (C) DY in HEK293, ENT1KO, and ENT2KO cells. Cells were incubated for 15 minutes with the specified inhibitor concentrations; then the 15-second uptake of 10 μM [³H]2-chloroadenosine was assessed. Data are shown as percent of the uptake seen in the absence of inhibitor (control). Each point is the mean ± SEM from 5 independent experiments conducted in duplicate.

3.8. Pharmacological analysis of endogenous ENT1- and ENT2-mediated 2-chloroadenosine uptake

The abilities of a range of nucleosides and nucleobases (both endogenous and chemotherapeutic analogs) were assessed for the ability to inhibit [³H]2-chloroadenosine uptake mediated by ENT1 in ENT2KO cells and compared with their previously assessed affinity for ENT2 determined using the ENT1KO cells¹² (Table 1). Clofarabine and vidarabine, not assessed in the previous study, were also tested in parallel for their affinity for ENT1 and ENT2. These assays were performed using suspended cells to allow for analysis of the effect of the transport process itself on the cellular accumulation of drugs. This can be assessed only at short time points (15 seconds or less) to minimize the confounding effects of intracellular trapping of the radiolabeled substrate by intracellular metabolism, which may also be impacted by the test compounds. Nucleobases were poor inhibitors of both ENT1- and ENT2-mediated [³H]2-chloroadenosine uptake, producing less than 50% inhibition at the highest concentrations tested with no significant differences between their effects on ENT1 versus ENT2 (Fig. 8A). The nucleoside analogs azidothymidine and cytarabine were also relatively weak inhibitors of ENT1 and ENT2. The other compounds tested (Fig. 8, B and C) were found to be potent inhibitors of ENT1 and ENT2 in these cell models allowing the construction of full concentration-response relationships to examine their relative ENT1:ENT2 affinity. All of the endogenous nucleoside ENT substrates tested showed a similar affinity for ENT1 and ENT2, with the exception of guanosine, which had a significantly higher affinity for ENT1 over ENT2 (Fig. 8B, Table 1); adenosine had the highest and uridine the lowest affinity for ENT1. None of the nucleoside analog drugs showed a significant difference in their affinity for ENT1 versus ENT2 (Fig. 8C, Table 1). Furthermore, they all shared a similar affinity (K_i approximately 50 μM), with the exception of gemcitabine, which had about a 10-fold lower affinity (K_i approximately 800 μM) for inhibiting [³H]2-chloroadenosine uptake by both ENT1 and ENT2 (Table 1).

Table 1.

Inhibition of [³H]2-chloroadenosine uptake by ENT1 and ENT2 by endogenous and chemotherapeutic nucleoside analogs

	ENT1 Kia μM	ENT2 Ki μM
Endogenous nucleosides
Adenosine	4.9 ± 1.9	2.6 ± 0.9b
Inosine	25 ± 10	65 ± 24b
Guanosine	29 ± 13c	160 ± 52b
Thymidine	36 ± 14	42 ± 8b
Uridine	89 ± 31	180 ± 30b
Chemotherapeutic analogs
Clofarabine	40 ± 16	75 ± 49
Cladribine	62 ± 24	23 ± 10b
Vidarabine	74 ± 26	60 ± 24
Gemcitabine	839 ± 308	950 ± 150b

^aFrom Shahid et al,¹² shown for comparison.

^bK_i values (mean ± SEM, n = 5) were calculated via the Cheng-Prusoff relationship using the IC₅₀ values obtained from the concentration-inhibition curves shown in Fig. 8, A and B, and the K_m of ENT1 for [³H]2-chloroadenosine provided in the text.

^cSignificant difference from the ENT2 K_i (unpaired t test, P < .05).

Fig. 8.

Inhibition of [³H]2-chloroadenosine uptake in ENT2KO cells by endogenous nucleosides/nucleobases and nucleoside/nucleobase analogs. Cells were exposed to the test compound and [³H]2-chloroadenosine (10 μM) simultaneously and the uptake reaction terminated after 15 seconds. (A) Compounds were initially tested at the highest concentration achievable in aqueous solution and compared with what was determined previously using ENT1KO cells.¹² Each bar represents the mean ± SEM from 5 experiments. ∗Indicates a significant decrease from control uptake levels (multiple unpaired t test, P < .05). (B) Endogenous nucleosides and (C) nucleoside/nucleobase analogs that produced greater than 50% inhibition in the initial screen were then tested over a range of concentrations to obtain IC₅₀ values for inhibition of ENT1 in the ENT2KO cell line. Each point represents the mean ± SEM from 5 independent experiments. The K_i values derived from these analyses are shown in Table 1.

3.9. Effect of ENT1 and/or ENT2 loss on the capacity of chemotherapeutic drugs to decrease cell viability

With the exception of 5-flurouracil and cladribine, all of the drugs tested produced a biphasic concentration-response profile for decreasing the viability of the unmodified HEK293 cells, but the relative proportion of resistant cells in the population varied with the drug used (Fig. 9). These data were analyzed to compare the logEC₅₀ values for both the high and low affinity components of the biphasic curve profiles and the percentage of relatively resistant cells making up the low affinity component. Values derived from the profiles shown in Fig. 9 are summarized in Table 2. Cladribine (Fig. 9G) showed a clear monophasic concentration-response profile, with almost 100% of the cells being nonviable after 48-hour treatment with the highest concentration of cladribine tested (100 μM). Likewise, the 5-flurouracil data also fit best to a monophasic profile (Fig. 9A). For the other drugs, approximately 50% of the cell population was relatively resistant to the effects of the cytidine analogs, gemcitabine and cytarabine (Fig. 9, D and E), and the nucleobase analog, 6-mercaptopurine (Fig. 9B), whereas only approximately 20% of the cells were resistant to 6-thioguanine and clofarabine (Fig. 9, C and F). Overall, the order of potency of these drugs in the HEK293 cells was gemcitabine > clofarabine > cytarabine > 6-thioguanine > cladribine = 6-mercaptopurine > 5-fluorouracil. In general, the loss of SLC29A1 in the ENT1KO cells had a minimal effect on the ability of the drugs tested to decrease cell viability; where there was a significant effect, loss of SLC29A1 increased the sensitivity of the cells to the drugs either by shifting the curve profile to the left (increased potency; eg, 5-fluorouracil and cladribine) or by decreasing the relative percentage of drug-resistant cells (eg, gemcitabine, cytarabine, and clofarabine). The loss of SLC29A2 (ENT2KO) cells had no significant effect on the nucleobase analog drugs (Fig. 9, A–C) but did significantly decrease the effects of the nucleoside analogs (Fig. 9, D–G). In particular, the ENT2KO cells were almost completely resistant to the effect of pyrimidine analogs, gemcitabine and cytarabine, and the purine analog, clofarabine, whereas cladribine was significantly less potent in this regard (HEK293 logEC₅₀, −4.98 ± 0.11; ENT2KO logEC₅₀, −4.47 ± 0.08). Loss of both ENT1 and ENT2 in the E1E2KO cell model also led to a significantly reduced sensitivity to 6-mercaptopurine, 6-thioguanine, gemcitabine, cytarabine, and clofarabine, but had less impact on the effects of 5-fluorouracil and cladribine.

Fig. 9.

Effect of nucleoside/nucleobase analogs on the viability of HEK293, ENT1KO, ENT2KO, and E1E2KO cell lines. Cells were plated in complete media and treated with the compounds indicated (nucleobase analogs, panels A–C; nucleoside analogs, panels D–G) for 48 hours at 37 °C. Cells were then assessed for viability using the MTT assay. Data are shown as mean ± SEM from 5 independent experiments. Biphasic or monophasic curves were fit to these data (best fit determined using the F test) and the EC₅₀ values and the percentage of cells showing relative resistance to the drugs calculated from these curve fits are shown in Table 2.

Table 2.

Effect of chemotherapeutic nucleoside/nucleobase analogs on the viability of HEK293, ENT1KO, ENT2KO, and E1E2KO cells

Cell Line	HEK293	ENT1KO	ENT2KO	E1E2KO
5-fluorouracil
% resistanta	Monophasic	Monophasic	Monophasic	48 ± 10
LogEC50-1a	−4.03 ± 0.07	−5.07 ± 0.07b	−4.25 ± 0.20	−5.11 ± 0.25b
LogEC50-2a	–	–	–	−2.97 ± 0.34
6-mercaptopurine
% resistant	43 ± 3	66 ± 4b	41 ± 3	70 ± 4b
LogEC50-1	−5.11 ± 0.10	−5.53 ± 0.21	−4.79 ± 0.09	−4.58 ± 0.24b
LogEC50-2	ND	ND	ND	ND
6-thioguanine
% resistant	20 ± 4	10 ± 3	25 ± 5	Monophasic
LogEC50-1	−5.50 ± 0.09	−5.77 ± 0.07	−5.28 ± 0.16	−4.11 ± 0.05b
LogEC50-2	−2.85 ± 0.49	ND	ND	–
Gemcitabine
% resistant	33 ± 4	14 ± 2b	76 ± 3b	Monophasic
LogEC50-1	−7.38 ± 0.21	−7.37 ± 0.08	−6.36 ± 0.27b	–
LogEC50-2	−3.42 ± 0.32	−3.33 ± 0.44	−2.54 ± 0.19	−2.74 ± 0.22
Cytarabine
% resistant	51 ± 2	37 ± 3b	Monophasic	Monophasic
LogEC50-1	−6.17 ± 0.10	−5.63 ± 0.09b	–	–
LogEC50-2	−2.22 ± 0.33	−2.34 ± 0.43	−2.48 ± 0.10	−1.95 ± 0.27
Clofarabine
% resistant	21 ± 2	6 ± 2b	70 ± 3b	Monophasic
LogEC50-1	−6.91 ± 0.06	−6.76 ± 0.09	−6.38 ± 0.18b	−3.86 ± 0.13b
LogEC50-2	ND	ND	−2.91 ± 0.49	–
Cladribine
% resistant	Monophasic	Monophasic	Monophasic	Monophasic
LogEC50-1	−5.11 ± 0.07	−5.40 ± 0.08b	−4.51 ± 0.07b	−4.78 ± 0.06b
LogEC50-2	–	–	–	–

ND, not determined.

^aDerived from the curve fits shown in Fig. 9. % resistant refers to the percentage of the cell population that was relatively resistant to the drug (plateau of a biphasic curve fit). LogEC₅₀-1 and LogEC₅₀-2 were derived from the high and low affinity phases, respectively, of the curve fits. Each value is the mean ± SEM, n = 5.

^bSignificant difference from value obtained in unmodified HEK293 cells (one-way ANOVA with Dunnett’s post-test for multiple comparisons, P < .05).

4. Discussion

The primary focus of this study was the creation and validation of genetically modified HEK293 cells lacking ENT2 (while retaining ENT1) (ENT2KO) or lacking both ENT1 and ENT2 (E1E2KO). HEK293 cells were selected for this work owing to their low expression of other nucleoside transporters, such as ENT3, ENT4, CNT1, CNT2, and CNT3,^12,20 making ENT1 and ENT2 the primary mechanisms by which nucleosides cross the plasma membrane in these cells. The HEK293 cell line, while not the normal target of chemotherapeutic nucleoside/nucleobase analog drugs, is still sensitive to these drugs that act by inhibiting nucleic acid synthesis in any proliferating cells. However, cancer cells typically show increased sensitivity to this class of chemotherapeutic drugs relative to noncancerous cells like HEK293 cells. This is thought to be due to the upregulation of ENT proteins by the cancer cells to salvage nucleosides/nucleobases from their environment to support their increased proliferative rate. This difference must be considered when assessing the translatability of the findings of the current study to the clinical situation.

The CRISPR/cas9-mediated disruption of the SLC29A2 gene was confirmed by Sanger sequencing, and the loss of the encoded ENT2 protein was confirmed via immunoblotting, and functionally by [³H]2-chloroadenosine uptake assays and selective inhibitor affinities. ENT1 and ENT2 in these cell models had affinities for their endogenous nucleoside substrates that were comparable to those reported previously,²¹ including the ENT1 selectivity of guanosine.²² Classical non-nucleoside ENT inhibitors (DY, NBMPR, and dilazep) demonstrated greater affinity for ENT1 than ENT2, consistent with previous reports.^18,23,24 The ENT2KO cell line, in combination with the ENT1KO cell line that we reported previously,¹² provides a screening panel for assessing the affinities of inhibitors and substrates for endogenously expressed ENT1 and ENT2. In addition, the E1E2KO cell line provides a nucleoside/nucleobase transporter null, human context, background for future studies on transporter oligomerization and the functional characterization of ENT mutants/polymorphisms. A similar CRISPR/cas9 approach has been used previously to selectively knock out SLC29A1 or SLC29A2 in HeLa cells and assess the effect of transporter deletion on substrate and drug uptake by the cells.²⁵ The results described in that study were qualitatively different, in some respects, from what was found in the current study, suggesting that cellular background may impact ENT affinities. However, the differences between the studies more likely reflect the fact that the study in HeLa cells used a 2-minute uptake time to assess cellular accumulation of [³H]uridine; this would measure the impact of inhibitors on both the transport process and intracellular metabolic trapping of uridine by phosphorylation. We used a rapid 15-second uptake time in the current study and used a more poorly metabolized substrate ([³H]2-chloroadenosine), which would provide results more indicative of inhibitor interactions with the transport process itself. A parallel evaluation of the HEK293 and HeLa cell lines using the same methodology and substrate is necessary to assess the potential impact of cellular background.

This panel of genetically modified HEK293 cells was used in the current study to examine the impact of the loss of ENT1, ENT2, or both subtypes, on the ability of a number of cancer chemotherapeutic drugs to reduce cell viability. The compounds tested have all been shown to enter cells via one or more subtypes of nucleoside transporters, and/or that their activity is sensitive to changes in nucleoside transporter expression.1, 2, 3, 4, 5 Antimetabolite chemotherapeutic drugs are typically more effective against actively proliferating cells (S phase), which is why they were applied in this study 2 hours after the initial plating when the cells are well below confluence and would be expected to be actively proliferating. Cell cycle analysis of the unmodified HEK293 cells indicated that approximately 50% of the cell population were in the S phase of the cell cycle and would thus be expected to be sensitive to the drugs tested. This was indeed the case for 6-mercaptopurine, 6-thioguanine, gemcitabine, cytarabine, and clofarabine, which displayed biphasic profiles for reducing cell viability with approximately 50%–70% of the cells showing high sensitivity to the drug and the remainder requiring concentrations in excess of 300 μM to be impacted. 5-Fluorouracil and cladribine were exceptions in that regard, showing relatively monophasic cell viability profiles over the range of concentrations tested, implying significant activity against nonproliferating cells. In the case of 5-fluorouracil, this may reflect its reported direct effects on mitochondrial function via free radical formation,²⁶ which would be detected by the MTT assay. Cladribine, although most active against cells in S phase, has previously been reported to have cell cycle independent cytotoxic effects on nonproliferating cells.^27,28 The order of potency of these drugs to reduce cell viability (gemcitabine > clofarabine > cytarabine > 6-thioguanine > cladribine = 6-mercaptopurine > 5-fluorouracil) had no relationship with their order of affinity for the ENT1 or ENT2 transporters (as measured by inhibition of [³H]2-chloroadenosine uptake). Contrary to previous reports,¹ this implies that their ability to be taken up by the cells via ENTs is not the major factor in their ability to decrease cell viability. This may reflect a difference between their effects on HEK293 cells versus cancer cells (although a similar low affinity of gemcitabine and cytarabine for inhibiting ENTs was reported for the analogous HeLa cell model²⁵), or it may reflect the fact that the transporter inhibition studies were performed in suspension after removal of the cells from the plates, whereas the cell viability studies were performed on adherent cell monolayers where additional environmental factors may impact drug-cell interactions. Nevertheless, the observed effects of genetic deletion of SLC29A1 and/or SLC29A2 from these cells on the ability of drugs to decrease cell viability do suggest that these transporters are impacting drug sensitivity independently of their ability to transport the drugs into the cells.

4.1. Effect of ENT1 loss

Deletion of SLC29A1 either had no effect (cytarabine, 6-mercaptopurine, and 6-thioguanine) or enhanced the effects of drugs on cell viability. This most likely reflects a decrease in endogenous nucleoside salvage by the cells due to the absence of the major nucleoside transporter ENT1. With respect to 5-fluorouracil, the loss of ENT1-mediated thymidine salvage makes cells more dependent on de novo thymidylate synthesis. Inhibition of thymidylate synthase by 5-fluorouracil under these conditions would lead to a more pronounced cytotoxic effect. Enhanced endogenous nucleoside salvage mechanisms are known to be involved in resistance to many of these drugs,²⁹ and blocking nucleoside salvage has been explored as a way to enhance the effectiveness of chemotherapy.³⁰ Our results would further support this approach.

4.2. Effect of ENT2 loss

Interpretation of the effects of the deletion of SLC29A2 on chemotherapeutic drug cytotoxicity is complicated by the significant compensatory changes in the expression of other genes associated with purine/pyrimidine metabolism. Although deletion of SLC29A1 had minimal effects on the expression of other genes involved in nucleoside metabolism,¹² the loss of SLC29A2 led to significant upregulation of SLC29A4 transcript (encoding ENT4), as well as an increase in the expression of SLC28A1, encoding CNT1. There was also an upregulation of HPRT (encoding hypoxanthine-guanine phosphoribosyltransferase) and ADSS (encoding adenylosuccinate synthase), both of which would enhance the production of adenine nucleotides from inosine monophosphate in these cells. Although the encoded protein levels were not assessed in the present study, the upregulation of these transporters and enzymes might be expected to enhance the salvage of endogenous nucleosides by the cells, as well as their subsequent conversion to nucleotides, thereby diminishing the cytotoxic potency of the nucleoside analog drugs. In addition, a larger proportion of the ENT2KO cell population was relatively insensitive to nucleoside analog drugs compared with the parent HEK293 cells. This does not appear to be due to changes in cell cycle because the ENT2KO cells had a similar spectrum of cell cycle stages as the HEK293 cells, and the nucleobase analog drugs were not similarly affected by the loss of ENT2. Nor was it due to changes in the CDK/NT5C ratio, an established measure of cell sensitivity to these nucleoside analogs.³¹ A possible explanation may involve ENT2 splice variants that have been identified in HEK293 cells localized to the nuclear envelope. Because exon 1 is maintained in these splice variants, CRISPR/cas9-mediated disruption in exon 1 of the ENT2KO cells would delete these variants along with the full-length ENT2. Regulation of alternative splicing of ENT2 has been correlated with proliferative status of the cell.³² A recent study has also hypothesized that chronic lymphocytic leukemia cells expressing high levels of nuclear ENT2 protein variants may be less affected by anticancer nucleoside analogs due to lower drug uptake,³³ but direct experimental evidence is limited at this time.

4.3. Effect of combined ENT1 and ENT2 loss

Deletion of both SLC29A1 and SLC29A2 from the HEK293 cells (E1E2KO cells) had the greatest impact on the effects of the drugs on cell viability. For the nucleoside analogs gemcitabine, cytarabine, and clofarabine, the E1E2KO cells were similar to the ENT2KO cells in their response to the drugs, suggesting that it was primarily the loss of SLC29A2 that was responsible. However, nucleobase analogs 6-mercaptopurine and 6-thioguanine were also much less effective in the E1E2KO cells, whereas loss of SLC29A2 alone had no effect. The E1E2KO cells had a much lower initial seeding efficiency and delayed proliferation compared with the other cell lines, with a greater percentage of the cells being in G2 phase. This means that the cells are relatively nonproliferative at the time of initial exposure to the drugs and may contribute to their insensitivity to the nucleobase analogs. Interestingly, the 2 drugs that were not affected by the dual loss of SLC29A1 and SLC29A2, 5-fluorouracil and cladribine, were also those drugs that showed relatively monophasic cell viability profiles reflecting their broader spectrum of activity against both proliferating and nonproliferating cells.

5. Conclusions

We have validated a panel of cell lines that have been genetically modified to express only ENT1, ENT2, or neither. These cell lines have been used in this study to screen the impact of selective ENT loss on the ability of clinically used nucleoside and nucleobase analogs to affect HEK293 cell viability. This study highlights distinct roles for ENT1 and ENT2 in the effects of these drugs, either through differential contributions to their cellular uptake or via compensatory changes in cellular function subsequent to the loss of the transporter.

Conflict of interest

The authors declare no conflicts of interest.