Pancreatic ductal adenocarcinoma (PDAC) is one of the most life-threatening malignancies. Although the deoxycytidine analog gemcitabine has been used as the first-line treatment for PDAC, the primary clinical challenge arises because of an eventual acquisition of resistance. Therefore, it is crucial to elucidate the mechanisms underlying gemcitabine resistance to improve treatment efficacy. To investigate potential genes whose inactivation confers gemcitabine resistance, we performed CRISPR knockout (KO) library screening. We found that deoxycytidine kinase (DCK) deficiency is the primary mechanism of gemcitabine resistance, and the inactivation of CRYBA2, DMBX1, CROT, and CD36 slightly conferred gemcitabine resistance. In particular, gene expression analysis revealed that DCK KO cells displayed a significant enrichment of genes associated with MYC targets, folate/one-carbon metabolism and glutamine metabolism pathways. Evidently, chemically targeting each of these pathways significantly reduced the survival of DCK KO cells. Moreover, the pathways enriched in DCK KO cells represented a trend similar to those in PDAC cell lines and samples of patients with PDAC with low DCK expression. We further observed that short-term treatment of parental CFPAC-1 cells with gemcitabine induces the expression of several genes, which promote synthesis and transport of glutamine in a dose-dependent manner, which suggests glutamine availability as a potential mechanism of escaping drug toxicity in an initial response for survival. Thus, our findings provide insights into novel therapeutic approaches for gemcitabine-resistant PDAC and emphasize the involvement of glutamine metabolism in drug-tolerant persister cells.

Implications:

Our study revealed the key pathways involved in gemcitabine resistance in PDAC, thus providing potential therapeutic strategies.

This article is featured in Highlights of This Issue, p. 387

Pancreatic ductal adenocarcinoma (PDAC) is a lethal form of cancer. The incidence of PDAC has been increasing dramatically over the past few decades, and it is expected that this trend will continue and PDAC will become the major cause of cancer-related mortality (1). There has been substantial improvement in the screening and treatment of solid cancer, which has helped many patients to achieve an early cure. However, the 5-year survival rate of patients with PDAC remains approximately 9% (2). Lack of corresponding clinical symptoms is a common scenario in patients with primary PDAC. Biomarkers such as CA19–9 exhibit minimum sensitivity and specificity during diagnosis (3). As most patients are diagnosed at the metastatic stage of the disease and are not eligible for surgical resection, they are managed with broad-spectrum chemotherapy (4). Chemotherapeutic drug resistance is a common consequence that causes treatment failure in patients with PDAC. Therefore, it is imperative to identify the factors responsible for chemoresistance (5, 6).

Gemcitabine [2′,2′-difluoro-2′-deoxycytidine (dFdC)] is a deoxycytidine analog that has been used as a first-line treatment for PDAC as well as against a wide range of solid tumors (7). Although patients with PDAC generally show considerable initial response to gemcitabine-based treatment, gemcitabine resistance develops gradually, which ultimately prevents the achievement of maximum clinical outcomes (7). Furthermore, combinatorial treatments with gemcitabine, which include cytotoxic agents (5-FU, cisplatin, oxaliplatin, and capecitabine) and biological agents (erlotinib, cetuximab, and bevacizumab) have been reported and conducted in patients. Although patient survival has comparatively increased due to combinatorial treatments, their efficacy is limited as a first-line treatment compared with gemcitabine monotherapy (5). Hence, the acquisition of resistance to gemcitabine by tumor cells is a major cause of treatment failure during the course of chemotherapy. Studies have reported several mechanisms of gemcitabine resistance, including reduced cellular uptake and high efflux rate of gemcitabine, intracellular transformation, and upregulation of antiapoptotic pathways (8–10). However, to date, the manipulation of these pathways has shown no significant success in overcoming gemcitabine resistance.

In mammalian cells, deoxyribonucleotide triphosphates (dNTPs) are synthesized through the de novo and nucleotide salvage pathways (11). Deoxycytidine kinase (DCK) catalyzes the rate-limiting step of the salvage pathway, which provides cells with DNA precursors. Further, DCK catalyzes the phosphorylation of deoxycytidine, deoxyguanosine, and deoxyadenosine, with ATP and UTP as phosphoryl donors (12). Furthermore, DCK phosphorylates and converts gemcitabine to metabolically active gemcitabine monophosphate (F2dCMP), which is further phosphorylated to di- (F2dCDP) and triphosphates (F2dCTP). F2dCDP blocks DNA de novo synthesis by inhibiting ribonucleotide reductases, which are required for the synthesis of deoxyribonucleotide precursors for DNA synthesis. Two major subunits of ribonucleotide reductase, RRM1 and RRM2, are strongly inhibited by gemcitabine metabolites; thus, cells ultimately stop DNA synthesis and die (13, 14).

Nucleotide synthesis is essential for cell proliferation and is upregulated in cancer cells to meet their increasing demand. Hence, multiple layers of regulations and interconnections exist with other metabolism pathways to ensure sufficient nucleotide synthesis. In fact, studies have clearly demonstrated the interplay between glucose and amino acid metabolism pathways (15–17). However, the mechanism by which DCK inactivation alters the diverse regulatory pathways of nucleotide synthesis remains obscure. Previous studies have reported that DCK was frequently inactivated in cancer cells with acquired gemcitabine resistance (18, 19). In addition, patients with PDAC with lower expression of DCK showed worse overall survival with gemcitabine treatment (20). Therefore, it is important to determine how cells overcome the negative impacts of reduced DCK activity on their proliferation and survival, which can lead to the identification of the novel vulnerabilities of cells and can be used for cancer treatment.

In this study, we applied genome-wide CRISPR/Cas9 library screening to identify the genes responsible for gemcitabine sensitivity in a human PDAC cell line. We found that DCK deficiency is the primary mechanism of gemcitabine resistance. Moreover, the introduction of single-guide RNAs (sgRNA) for CRYBA2 (crystallin beta A2), DMBX1 (diencephalon/mesencephalon homeobox 1), CROT (carnitine O-octanoyltransferase), and CD36 (scavenger receptor) led to gemcitabine resistance at low doses. RNA sequencing (RNA-seq) analysis of DCK-deficient cells revealed the upregulation of the MYC pathway, one-carbon metabolism pathway, and arginine/proline/glutamine metabolism pathways. In support of these findings, DCK-deficient cells became sensitive to the inhibitors of these pathways, such as c-Myc inhibitor, methotrexate, and 6-diazo-5-oxo-L-norleucine (DON). Furthermore, these pathways were upregulated in PDAC cell lines and samples of patients with PDAC with low DCK expression. Overall, our study suggests new treatment options for PDAC cells with acquired gemcitabine resistance induced by DCK inactivation.

Cell lines

Human PDAC cell lines, including Capan-2 (RRID:CVCL_0026), Panc10.05 (RRID:CVCL_1639), CFPAC-1 (RRID:CVCL_1119), HPAF-II (RRID:CVCL_0313), SW1990 (RRID:CVCL_1723), BxPC-3 (RRID:CVCL_0186), and AsPC-1(RRID:CVCL_0152), were directly obtained from the ATCC. All cell lines were authenticated using the Promega GenePrint 10 system. Mycoplasma testing was performed using Mycoblue mycoplasma detector kit (Vazyme). Capan-2 cells were maintained in McCoy's 5A medium (GE Healthcare Life Science); CFPAC-1 cells were maintained in Iscove's modified Dulbecco's medium [IMDM; FUJIFILM Wako Pure Chemical Corporation (FUJIFILM-Wako)]; HPAF-II cells were maintained in Eagle Minimum Essential Medium (FUJIFILM-Wako); and Panc10.05, SW1990, BxPC-3, and AsPC-1 cells were maintained in RPMI1640 medium (FUJIFILM-Wako), which were cultured at 37°C in a humidified atmosphere containing 5% CO2. All cell culture media were supplemented with 10% heat-inactivated FBS (Gibco #10270–106) and 1% penicillin–streptomycin (10,000 U/mL; Thermo Fisher Scientific, #15140–122). Cells were maintained in culture for less than 15 passages.

Cell viability assay

Cellular viability was evaluated using Cell Count Reagent SF (WST-8; Nacalai Tesque, #07553–15). Cells were seeded into 96-well plates at 5,000 cells per well 24 hours before treatment with gemcitabine, methotrexate, 5-fluorouracil, c-Myc inhibitor 10058-F4, Osalmid, and DON at the indicated doses. At 72 hours after treatment, 10 μL of WST-8 [2-(2-methoxy-4-nitrophenyl)-3-(4-nitrophenyl)-5-(2,4-disulfophenyl)-2H-tetrazolium] was added to each well and incubated for 2 hours at 37°C, and then the absorbance was measured immediately at 450 nm using a microplate reader (iMark; Bio-Rad). The background readings were subtracted from each original reading. The cellular viability assay was performed in triplicates and repeated at least 3 times. The IC50 values were calculated using the curves constructed by plotting cellular viability versus drug concentration (21).

Lentivirus production

For lentivirus production, 293JD packaging cells (a kind gift from Dr. James Ellis, University of Toronto) were seeded and cultured onto 6-cm dishes at 2 × 106 cells per dish in DMEM (Gibco, #11995–065) supplemented with 10% heat-inactivated FBS (Gibco, #10270–106) and 1% penicillin–streptomycin (10,000 U/mL; Thermo Fisher Scientific, #15140–122) 24 hours before performing lentiviral vector transfection. For packaged lentiviral production, 6 μL of Fugene 6 (Promega, #E2691), 94 μL of Opti-MEM (Gibco, #31985–070), 0.75 μg of psPAX2 plasmid (Addgene, #12260, RRID:Addgene_12260), 0.5 μg of pMD2.G (Addgene, #12259, RRID:Addgene_12259), and 1 μg of a lentiviral vector plasmid per 6-cm dish were used. To obtain the optimized multiplicity of infection (transfection efficiency > 40%), the supernatant containing viral particles was collected from the dishes at 48 hours post transfection and then concentrated using a LentiX concentrator (Takara, #631231).

Genome-wide CRISPR/cas9 knockout library screening

The Human Activity-Optimized CRISPR knockout (KO) library (two sublibraries in lentiCRISPRv1; Addgene pooled library, #1000000067) was used to identify the genes responsible for gemcitabine resistance in the CFPAC-1 cell line (22). The workflow for the forward genetic screening is illustrated in Fig. 1B. The lentiCRISPRv1 library comprises two sublibraries, each of them containing approximately 90,000 sgRNAs. In other words, the library contains 178,896 sgRNAs in total and targets 18,166 genes. It also contains 1,000 control nontarget sgRNAs. First, 5.4 × 106 CFPAC-1 cells were seeded onto 6-well plates (3 × 105 cells per well, total 18 wells) 24 hours before transfection to cover 60-fold of each sublibrary. An aliquot of polybrene (Nacalai Tesque, #12996–81, stock 8 mg/mL) was added to each well at 0.625 μL/ml, mixed with viral particles, and transfected into CFPAC-1 cells. After 24 hours of transfection, the cells were cultured in regular IMDM for another 24 hours, followed by selection using 2 μg/mL of puromycin (Gibco, #A11138–03) for 72 hours. The selection process was confirmed using parental CFPAC-1 cells without viral infection as the reference. The established cell line containing the library, pre-gemcitabine–treated (pre-gem) cells, was treated with 300 nmol/L gemcitabine for 3 days followed by the culture in normal medium (for 10 days for the 1st, and 4 days for the 2nd and the 3rd round of selections) to expand the surviving cells. This process was repeated 3 times. After the treatment, 3 × 107 cells were harvested as post-gemcitabine–treated (post-gem) cells. Genomic DNA was extracted from pre-gem and post-gem cells. The inserted sgRNA region was confirmed using a universal primer and sgRNA barcode PCR primers (Supplementary Table S1). PCR was performed using 2× phusion PCR master mix (NEB, #M0531S). The sequences of amplified PCR products were analyzed using next-generation sequencing to identify the enriched sgRNAs in post-gem cells compared with pre-gem cells. The screening process with each sublibrary was conducted independently.

Figure 1.

Identification of determinants of gemcitabine resistance via CRISPR-Cas9 library screening in a gemcitabine-sensitive CFPAC-1 PDAC cell line. A, Gemcitabine dose–response curve in seven PDAC cell lines. Cells were treated with different concentrations of gemcitabine for 3 days. Data are expressed as mean ± SD from three independent triplicate experiments. B, Schematic diagram illustrating the workflow of genome-wide CRISPR-Cas9 KO library screening. C, Gemcitabine dose–response curve in pre-gem and post-gem CFPAC-1 cells. Cells were treated with gemcitabine at the indicated concentrations for 3 days. Data are expressed as mean ± SD from three independent triplicate experiments. D and E, Distribution of fold change in the top 100 sgRNA clones for the screening using pool A (D) and pool B (E) libraries. The sgDCK#2, sgDCK#3, sgRPS6KC1#1, sgSALL1#5, and sgSTBXP1#1 from pool A library and the sgDCK#10, sgCRYBA2#10, sgDMBX1#10, sgCROT#9, sgCD36#8, and sgFAM186A#6 from pool B library were identified as hits with a cutoff of log2-transformed fold change of >10. F, Venn diagram presenting genes targeted by sgRNA clones, which are shown as hits in D and E. G, Relative mRNA expression levels of DCK in seven PDAC cell lines analyzed using RT-qPCR. Bars indicate the average. Error bars, SD.

Figure 1.

Identification of determinants of gemcitabine resistance via CRISPR-Cas9 library screening in a gemcitabine-sensitive CFPAC-1 PDAC cell line. A, Gemcitabine dose–response curve in seven PDAC cell lines. Cells were treated with different concentrations of gemcitabine for 3 days. Data are expressed as mean ± SD from three independent triplicate experiments. B, Schematic diagram illustrating the workflow of genome-wide CRISPR-Cas9 KO library screening. C, Gemcitabine dose–response curve in pre-gem and post-gem CFPAC-1 cells. Cells were treated with gemcitabine at the indicated concentrations for 3 days. Data are expressed as mean ± SD from three independent triplicate experiments. D and E, Distribution of fold change in the top 100 sgRNA clones for the screening using pool A (D) and pool B (E) libraries. The sgDCK#2, sgDCK#3, sgRPS6KC1#1, sgSALL1#5, and sgSTBXP1#1 from pool A library and the sgDCK#10, sgCRYBA2#10, sgDMBX1#10, sgCROT#9, sgCD36#8, and sgFAM186A#6 from pool B library were identified as hits with a cutoff of log2-transformed fold change of >10. F, Venn diagram presenting genes targeted by sgRNA clones, which are shown as hits in D and E. G, Relative mRNA expression levels of DCK in seven PDAC cell lines analyzed using RT-qPCR. Bars indicate the average. Error bars, SD.

Close modal

Establishment of individual KO cell lines

To confirm the hits obtained via screening with each sublibrary, the oligos corresponding to each sgRNA sequence of the hits, sgDCK#2, sgDCK#3, sgDCK#10, sgRPS6KC1#1, sgSALL1#5, sgSTBXP1#1, sgCRYBA2#10, sgDMBX1#10, sgCROT#9, sgCD36#8, and sgFAM186A#6 and controls, sgNT1 and sgGFP (Supplementary Table S2), were synthesized, annealed, and subcloned into lentiCRISPRv2 (Addgene, #52961, RRID:Addgene_52961; ref. 23). The insertion of the desired sgRNA sequences in lentiCRISPRv2 was confirmed by sequencing. The CFPAC-1 cells were seeded onto 6-well plates at 3 × 105 cells per well and cultured in regular IMDM for 24 hours. On the following day, the cells were transduced with lentiviral particles prepared from each sgRNA lentiviral vector. After 24 hours of infection, the cells were cultured for another 24 hours in regular IMDM, and then the sgRNA-introduced cells were selected using 2 μg/mL of puromycin for 72 hours.

Focus formation assay and crystal violet staining

Cells were seeded onto 6-well plates at 6 × 102 cells per well for DCK#2, DCK#3, and DCK#10, and 8 × 102 cells for the other clones 24 hours before treatment with gemcitabine. Then, they were cultured in the presence of gemcitabine for 72 hours, washed twice, and cultured for an additional 10 days in gemcitabine-free regular culture medium. Next, the cells were washed twice with PBS and stained with 600 μL of 0.2% crystal violet [Sigma, # C6158; dissolved in 80% methanol (FUJIFILM-Wako #131–01826)]. After 30 minutes of incubation at room temperature, the plates were washed twice with PBS and dried at room temperature for 1 hour before capturing the images. The area of the foci was quantified by ImageJ (RRID:SCR_003070).

RT-qPCR

Total RNA was extracted from cells using the Monarch Total RNA Miniprep Kit (NEB, #T2010). First-strand cDNA was synthesized using the iScript cDNA Synthesis Kit (Bio-Rad, #1708840). RT-qPCR was performed using SYBR Green PCR Master Mix (NEB, #M3003L), with primers specific for target genes of interest. HPRT1 and GUSB genes were used as internal controls. The primer sequences are listed in Supplementary Table S3. PCR was performed as follows: an initial incubation step at 95°C for 30 seconds, followed by 40 cycles at 95°C for 5 seconds and 60°C for 30 seconds. Data were collected using the StepOnePlus Real-time PCR system (Applied Biosystems). The specificity of the detected signals was confirmed by the presence of a dissociation curve containing a single peak and via electrophoresis of the PCR products.

Western blotting

To prepare whole cell protein lysates, 3.2 × 106 cells were collected via trypsinization with 0.25% trypsin–EDTA (Gibco, #25200–072), followed by washing twice with PBS and centrifugation at 200 g for 5 minutes at 4°C. Next, the cells were gently mixed with 300 μL of 1× SDS-PAGE sample buffer (50 mmol/L Tris-HCl, pH 6.8, 5% glycerol, 2% SDS, 0.02% bromophenol blue, and 0.66% β-mercaptoethanol) and heated at 95°C for 15 minutes. The cell lysates were loaded on 12% polyacrylamide gel for conducting SDS-PAGE. Precision plus protein dye (Bio-Rad, #1610374) was used as a size marker. DCK protein was detected using anti-DCK antibody produced in rabbits (GeneTex, #GTX102800, RRID:AB_11163443, 1:1,000). HA protein was detected with anti-HA tag antibody (Proteintech, #51064–2-AP, RRID:AB_11042321, 1:1,000). β-actin antibody (Santa Cruz, #sc-47778, RRID:AB_626632 1:1,000) was used as a loading control.

Sequencing of DCK sgRNA target regions

To confirm CRISPR/Cas9-mediated genome editing for DCK, the DNA fragments containing the target regions of each sgRNA were sequenced. Genomic DNA was extracted from CFPAC-1 cells containing sgDCK#2, sgDCK#3, and sgDCK#10 using the QIAamp DNA Mini Kit (Qiagen, #51304). DNA extracted from CFPAC-1 cells was used as the wild-type control. The DNA regions covering the sgRNA target sites were amplified by PCR using the primers listed in Supplementary Table S4. The DNA fragments with the expected sizes were purified using agarose gel extraction, and the blunt ends of the DNA fragments were phosphorylated by T4PNK (NEB, #M0201S). The resulting DNA fragments were cloned into EcoRV-digested pBluescript SK (+) plasmid. The insertion of the desired DNA fragments was confirmed by sequencing.

Production of CRISPR/cas9-resistant cDNA of wild-type DCK and the kinase-dead mutant

To rescue DCK expression in DCK-deficient CFPAC-1 cells (DCK#10 cells), wild-type DCK cDNA was amplified by RT-PCR. The template cDNA was prepared using mRNA isolated from CFPAC-1 cells using the iScript cDNA synthesis kit (Bio-Rad, #1708891). The PCR primers were designed to introduce EcoRI site at the 5′ end and XhoI site at the 3′ end. The EcoRI/XhoI fragment purified from the agarose gel was subcloned into the pcDNA3 vector containing the N-terminus HA-tag sequence (HA-pcDNA3, a kind gift from Dr. K. Miyazono, University of Tokyo), with the methionine start codon in-frame with the EcoRI restriction site (HA-DCK). After confirming the entire sequence of the cloned wild-type DCK cDNA in both directions, four translationally silent point mutations were introduced adjacent to the PAM region targeted by sgDCK#10 for DCK cDNA using PCR, which presumably avoids Cas9-mediated degradation of DCK cDNA in cells that constitutively express Cas9 and sgDCK#10. The insertion of the desired mutations was confirmed by sequencing. In addition, to generate a kinase-dead mutant of DCK, the nucleotides were replaced to convert three amino acids within the kinase domain of DCK, ERS (positions 127–129) to AAA. The resulting HA-DCK with or without the kinase domain mutation (HA-DCK-KD and HA-DCK, respectively) was subcloned into the lentiviral expression vector CSII-CMV-MCS-IRES2-Bsd (RIKEN BRC #RDB04385) at the NheI/XhoI site. The lentivirus-transduced CFPAC-1 cells were selected in IMDM containing 8 μg/mL of blasticidin S for 5 days. The primer sequences used in this step are listed in Supplementary Table S3.

RNA-seq

Total RNA was extracted using the Monarch Total RNA Miniprep Kit (NEB, #T2010). Library preparation, sequencing, and data processing were performed using DNAFORM (Yokohama, Kanagawa, Japan). The qualities of total RNA were evaluated using Bioanalyzer (Agilent) to ensure that the RNA integrity number is > 7.0. After poly (A) + RNA enrichment by NEBNext poly(A) mRNA Magnetic Isolation Module (NEB E7490), double-stranded cDNA libraries (RNA-seq libraries) were prepared using the SMARTer Stranded RNA-seq Kit (Clontech, #634836). The RNA-seq libraries were sequenced using paired-end reads (50 nt of read 1 and 25 nt of read 2) on a NextSeq 500 instrument (Illumina). The obtained raw reads were trimmed and quality-filtered using the Trim Galore! (version 0.4.4, RRID:SCR_011847), Trimmomatic (version 0.36, RRID:SCR_011848), and cutadapt (version 1.16) software. The trimmed reads were then mapped to the human GRCh38 genome using STAR (version 2.7.2b, RRID:SCR_004463). The reads mapped to annotated genes were counted using featureCounts (version 1.6.1). The FPKM values were calculated from the mapped reads by normalizing the total transcript count. To plot the heatmap, DEseq2 (version 1.20.0, RRID:SCR_015687) analyses were performed to compare relative transcript levels between CFPAC-1 and DCK#10 cell lines with default parameters, and differentially expressed genes were determined based on a cutoff of the Padj value of < 0.05. The heatmap cluster analysis was performed using R version 4.1.2 with library (pheatmap).

Gene set enrichment analyses

Normalized gene expression data were obtained from the Cancer Cell Line Encyclopedia (CCLE) project (24, 25) for AsPC-1, BxPC-3, Capan-2, CFPAC-1, HPAF-II, Panc10.05, and SW1990 cell lines and the Pan-Cancer Atlas (The Cancer Genome Atlas [TCGA]) (26) for patients with PDAC. Gene set enrichment analysis (GSEA; RRID:SCR_003199) analyses (27) were conducted to compare the following groups: CFPAC-1 versus DCK-deficient CFPAC-1 (DCK#10); patients with PDAC with high DCK expression (n = 11, Z-score > 1.25) versus those with low DCK expression (n = 12, Z-score < −1.25); and PDAC cell lines with high DCK expression (RPKM > 10; Capan-2 and CFPAC-1) versus those with low DCK expression (RPKM < 10; AsPC-1, BxPC-3, HPAF-II, Panc10.05, and SW1990). The analyses were performed using the following gene sets: HALLMARK and KEGG gene sets and GOBP_ONE_CARBON_METABOLIC_PROCESS, WP_TRANSSULFURATION_AND_ONE_CARBON_METBOLISM, REACTOME_GLUTAMATE_AND_GLUTAMINE_METABOLISM, and KEGG_ARGININE_PROLINE_METABOLISM_PATHWAY from MSigDB.

The normalized enrichment score was calculated using the GSEA software. A FDR q-value of < 0.25 was considered to be significant.

Glutamine/glutamate assay

CFPAC-1 cells were seeded on 96-well plates 24 hours before treatment with gemcitabine. At the same time, replica plates were prepared for cell counting. The intracellular glutamine /glutamate levels were measured using Glutamine/Glutamate-Glo Assay (Promega) in triplicate and repeated twice. The luminescence signals were measured with Wallac 1420 ARVO MX Multilabel Counter (PerkinElmer). The values from the treated and untreated cells were normalized by the cell number from the replica plates.

Statistical analysis

Results are expressed as mean ± SD. Differences between two groups were analyzed using an unpaired two-tailed Student t test. Multiple groups were compared using one-way analysis of variance. P < 0.05 was considered to be significant.

Data availability

RNA-seq data obtained in this study have been deposited under the following accession numbers: Experiment DRX370437-DRX370440 in the DDBJ (DNA Data Bank of Japan).

CRISPR/cas9 library screening identified that DCK is responsible for gemcitabine resistance

To evaluate the gemcitabine sensitivity of PDAC cell lines, seven PDAC cell lines were treated with the indicated concentrations of gemcitabine. CFPAC-1 and HPAF-II cells were most sensitive and most resistant, respectively, among the seven cell lines (Fig. 1A). To identify the critical genes involved in gemcitabine resistance, genome-wide CRISPR/Cas9 KO library screening was performed using the most gemcitabine-sensitive PDAC cell line, CFPAC-1, to minimize the background effect of preexisting gemcitabine resistance on the screening (Fig. 1B). The Human Activity-Optimized CRISPR KO library, lentiCRISPRv1, contains two sublibraries (pool A and pool B). For each pool, 5.4 × 106 cells were screened, which represented approximately 60-fold coverage of each pool. The parental CFPAC-1 and the library transfected pre-gem cells showed similar gemcitabine sensitivity (Supplementary Fig. S1A). After three rounds of gemcitabine treatment, the treated (post-gem) cells became markedly resistant to gemcitabine (Fig. 1C; Supplementary Fig. S1A) with higher growth rate (Supplementary Fig. S1B). Then, genomic DNA was extracted from the pre-gem and post-gem cells and subjected to next-generation sequencing. Five sgRNAs were identified from the screening of positive hits of pool A and six were identified from that of pool B, as they showed a log2 fold change of >10 in the post-gem cell pools A and B compared with the corresponding pre-gem pools (Fig. 1D and E). Among these top 11 enriched sgRNAs in the post-gem cell pools, three were against the DCK gene (Fig. 1F), indicating that DCK inactivation is highly responsible for gemcitabine resistance. In agreement with these findings, DCK levels were lower in the post-gem cells (Supplementary Fig. S1C). Next, to correlate gemcitabine sensitivity with DCK expression, the mRNA expression of DCK was assessed in seven PDAC cell lines using RT-qPCR. Results showed that DCK expression levels positively correlated with gemcitabine sensitivity, except for Capan-2 (Fig. 1G).

Confirmation of individual gemcitabine-resistant genes

To validate the results of CRISPR/Cas9 screening, 13 CFPAC-1 cell lines expressing the following individual sgRNAs were established: sgDCK#2, sgDCK#3, sgDCK#10, sgRPS6KC1#1, sgSALL1#5, sgSTBXP1#1, sgCRYBA2#10, sgDMBX1#10, sgCROT#9, sgCD36#8, sgFAM186A#6, sgNT1, and sgGFP (sgNT1 and sgGFP as controls). To evaluate gemcitabine resistance, these cells were treated with the indicated doses of gemcitabine (Fig. 2A and B). DCK#2, DCK#3, and DCK#10 cells exhibited marked resistance to gemcitabine, with DCK#10 cells being the most resistant among them. Other cells, such as CRYBA2#10, DMBX1#10, CROT#9, and CD36#8, exhibited slight resistance only at low concentrations of gemcitabine. The remaining cells, including RPS6KC1#1, SALL1#5, sgSTBXP1#1, and sgFAM186A#6, showed no resistance.

Figure 2.

Validation of selected hits for gemcitabine resistance. A and B, Gemcitabine dose–response curve of CFPAC-1 cell lines transduced with the indicated sgRNAs from pool A (A) and pool B (B) libraries. The cells were treated with gemcitabine for 3 days. Data are expressed as mean ± SD from three independent triplicate experiments. NT1 (nontargeting sgRNA) was used as a control. C and D, Representative images of focus formation assay. The DCK#2-, DCK#3-, or NT1-transduced CFPAC-1 cells (C) and the DCK#10-, DMBX1#10-, CD36#8-, CRYBA2#10-, or CROT#9-transduced CFPAC-1 cells (D) were treated with the indicated concentrations of gemcitabine for 3 days. The cells were further cultured in regular media for 10 days. E, Bar graph showing the quantification of dye-stained areas of the foci derived from DCK#2, DCK#3 and DCK#10 treated with gemcitabine at the indicated concentrations. The values were normalized to the area of untreated cells (set at 100%). F, Immunoblot showing protein expression of DCK in DCK#2, DCK#3, and DCK#10 cells. GFP indicates the control sgGFP-transduced cells. β-actin was used as a loading control. G, Immunoblot showing the DCK protein rescue by transduction of exogenous HA-tagged DCK. β-actin used as a loading control. H, Gemcitabine dose–response curve of DCK#10 cells expressing wild-type or kinase-dead mutant of DCK. Cells were incubated with gemcitabine for 3 days. Exogenous wild-type DCK but not kinase-dead mutant sensitizes DCK#10 cells to gemcitabine. Data are expressed as mean ± SD from three independent triplicate experiments. *, P < 0.05; **, P < 0.01; ***, P < 0.001; ns, not significant. CFPAC-1, parental CFAPC-1 cell line; EV, empty vector; HA–DCK, HA-tagged wild-type DCK; and HA-DCK–KD, HA-tagged kinase-dead mutant of DCK.

Figure 2.

Validation of selected hits for gemcitabine resistance. A and B, Gemcitabine dose–response curve of CFPAC-1 cell lines transduced with the indicated sgRNAs from pool A (A) and pool B (B) libraries. The cells were treated with gemcitabine for 3 days. Data are expressed as mean ± SD from three independent triplicate experiments. NT1 (nontargeting sgRNA) was used as a control. C and D, Representative images of focus formation assay. The DCK#2-, DCK#3-, or NT1-transduced CFPAC-1 cells (C) and the DCK#10-, DMBX1#10-, CD36#8-, CRYBA2#10-, or CROT#9-transduced CFPAC-1 cells (D) were treated with the indicated concentrations of gemcitabine for 3 days. The cells were further cultured in regular media for 10 days. E, Bar graph showing the quantification of dye-stained areas of the foci derived from DCK#2, DCK#3 and DCK#10 treated with gemcitabine at the indicated concentrations. The values were normalized to the area of untreated cells (set at 100%). F, Immunoblot showing protein expression of DCK in DCK#2, DCK#3, and DCK#10 cells. GFP indicates the control sgGFP-transduced cells. β-actin was used as a loading control. G, Immunoblot showing the DCK protein rescue by transduction of exogenous HA-tagged DCK. β-actin used as a loading control. H, Gemcitabine dose–response curve of DCK#10 cells expressing wild-type or kinase-dead mutant of DCK. Cells were incubated with gemcitabine for 3 days. Exogenous wild-type DCK but not kinase-dead mutant sensitizes DCK#10 cells to gemcitabine. Data are expressed as mean ± SD from three independent triplicate experiments. *, P < 0.05; **, P < 0.01; ***, P < 0.001; ns, not significant. CFPAC-1, parental CFAPC-1 cell line; EV, empty vector; HA–DCK, HA-tagged wild-type DCK; and HA-DCK–KD, HA-tagged kinase-dead mutant of DCK.

Close modal

To validate long-term gemcitabine sensitivity, focus formation assay was performed for the cell lines with resistance. Consistent with the abovementioned WST assay results, DCK#2, DCK#3, and DCK#10 formed higher number of foci (Fig. 2CE; Supplementary Fig. S1D) than other cells. DCK#10 lacked DCK protein expression (Fig. 2F), whereas DCK protein expression was detectable in DCK#2 and DCK#3. Next, the deletion of DCK was analyzed in DCK#2, DCK#3, and DCK#10, which revealed that DCK#10 contained longer deletion and frameshift than other cells, supporting the loss of DCK protein expression. In contrast, in DCK#2 and DCK#3, approximately 50% of the clones had an intact DCK region, which might be the reason for the presence of DCK protein in the western blotting results (Fig. 2F; Supplementary Fig. S2A–S2C). These results suggest that DCK#2 and DCK#3 mutants act as dominant negatives because DCK functions as a dimer, although further validation is required for confirmation in future studies. Because DCK protein was undetectable in DCK#10, these cells were used for subsequent analyses. To confirm that the loss of DCK caused gemcitabine resistance in DCK#10, it was reintroduced with HA tag (HA-DCK), which contained four nucleotide base pair changes adjacent to the PAM site of sgDCK#10 and was resistant to sgDCK#10/Cas9-induced degradation (Fig. 2G; Supplementary Fig. S3A). The introduction of HA-DCK made DCK#10 cells more sensitive to gemcitabine than the control presumably because of a higher DCK expression level (Fig. 2H). HA-DCK was also generated using the kinase-dead mutant with three amino acid changes in the kinase domain (HA-DCK-KD: 127–129 ERS > AAA), and HA-DCK-KD was introduced into DCK#10 (Fig. 2G; Supplementary Fig. S3B). However, it could not restore the gemcitabine sensitivity of DCK#10 (Fig. 2H). These results imply that gemcitabine resistance is solely dependent on DCK loss in DCK#10.

DCK inactivation enhanced the expression of genes involved in MYC, one-carbon metabolism, and glutamine metabolism pathways

To further characterize the DCK KO cell line, DCK#10, we analyzed the gene expression using RNA-seq. A total of 1,825 genes were found to be differentially expressed in DCK#10 with a P value (Padj) of < 0.05 (Fig. 3A). GSEA was performed to identify the differentially expressed gene profiles between DCK#10 and the parental CFPAC-1 cells, which revealed that genes associated with MYC targets, E2F targets, and oxidative phosphorylation (OXPHOS) pathways were significantly upregulated in DCK#10 (Fig. 3B). Furthermore, the one-carbon metabolism pathway and arginine/proline/glutamine metabolism pathways were upregulated in DCK#10 (Fig. 3CE). To confirm the GSEA results, the mRNA expression of the following genes crucial to the one-carbon metabolism pathway was investigated using RT-qPCR: DHFR, PHGDH, SHMT1, and MTHFD1L. The results showed that the expression of PHGDH, SHMT1, and MTHFD1L was significantly upregulated in DCK#10 cells (Fig. 3F). Next, the gene expression of the enzymes RRM1, RRM2, and RRM2B involved in the de novo nucleotide synthesis was analyzed. The results showed that RRM2B was significantly upregulated in DCK#10 cells (Fig. 3G). For the arginine/proline/glutamine metabolism pathways, we selected several crucial interconnected genes involved in the metabolism of these amino acids, including CAD, GLS2, GLUD1, OAT, SLAC1A5, P5CS, and PYCR1. These genes are essential for maintaining the arginine/proline/glutamine shuttle in cells (28–30). We observed that all genes involved in noncanonical glutamine-mediated energy production were significantly upregulated in DCK#10 cells (Fig. 3H). These findings corroborate that DCK KO is responsible for the upregulation of the one-carbon metabolism pathway; DNA de novo synthesis pathway; and arginine/proline/glutamine metabolism pathways.

Figure 3.

DCK inactivation induced upregulation of MYC, and one-carbon metabolism, and glutamine metabolism pathways. A, Heatmap showing a clustered gene expression pattern of 1,825 differentially expressed genes, with a Padj value of < 0.05, between parental CFPAC-1 and DCK#10 groups. Each group contains two biological replicates. B, GSEA plot showing a significant enrichment of MYC targets (HALLMARK_MYC_TARGETS_V1 and HALLMARK_MYC_TARGETS_V2), E2F targets (HALLMARK_E2F_TARGETS), and genes associated with OXPHOS pathway (HALLMARK_OXIDATIVE_PHOSPHORYLATION) in DCK#10 cells. C–E, GSEA plot showing a significant enrichment of genes associated with folate/one-carbon metabolism pathway (WP_TRANSSULFURARTION_AND_ONE_CARBON_METABOLISM and GOBP_ONE_CARBON_METABOLIC_PROCESS; C), arginine/proline metabolism pathway (KEGG_ARGININE_AND_PROLINE_METABOLISM; D) and glutamine metabolism pathway (REACTOME_GLUTAMATE_AND_GLUTAMINE_METABOLISM) in DCK#10 cells. F–H, RT-qPCR showing expression of genes involved in the one-carbon metabolism pathway (F), DNA de novo synthesis (G), and arginine/proline/glutamine metabolism pathways (H). Data are expressed as mean ± SD from three independent triplicate experiments.*, P < 0.05; **, P < 0.01; ***, P < 0.001; ns, not significant. NES, normalized enrichment score.

Figure 3.

DCK inactivation induced upregulation of MYC, and one-carbon metabolism, and glutamine metabolism pathways. A, Heatmap showing a clustered gene expression pattern of 1,825 differentially expressed genes, with a Padj value of < 0.05, between parental CFPAC-1 and DCK#10 groups. Each group contains two biological replicates. B, GSEA plot showing a significant enrichment of MYC targets (HALLMARK_MYC_TARGETS_V1 and HALLMARK_MYC_TARGETS_V2), E2F targets (HALLMARK_E2F_TARGETS), and genes associated with OXPHOS pathway (HALLMARK_OXIDATIVE_PHOSPHORYLATION) in DCK#10 cells. C–E, GSEA plot showing a significant enrichment of genes associated with folate/one-carbon metabolism pathway (WP_TRANSSULFURARTION_AND_ONE_CARBON_METABOLISM and GOBP_ONE_CARBON_METABOLIC_PROCESS; C), arginine/proline metabolism pathway (KEGG_ARGININE_AND_PROLINE_METABOLISM; D) and glutamine metabolism pathway (REACTOME_GLUTAMATE_AND_GLUTAMINE_METABOLISM) in DCK#10 cells. F–H, RT-qPCR showing expression of genes involved in the one-carbon metabolism pathway (F), DNA de novo synthesis (G), and arginine/proline/glutamine metabolism pathways (H). Data are expressed as mean ± SD from three independent triplicate experiments.*, P < 0.05; **, P < 0.01; ***, P < 0.001; ns, not significant. NES, normalized enrichment score.

Close modal

DCK inactivation increased dependency on MYC, one-carbon metabolism, and glutamine metabolism pathways

The results of GSEA indicated the upregulation of MYC target genes, OXPHOS, and one-carbon and glutamine metabolism pathways in DCK-deficient DCK#10 cells. Hence, we tested whether the inactivation of these pathways affects DCK#10 cell proliferation. DCK#10 and NT1 (a negative control) cells were treated with the inhibitors of the one-carbon metabolism [methotrexate (MTX) and 5-fluorouracil (5-FU)], MYC (10058-F4), DNA de novo synthesis (Osalmid), and glutamine metabolism (DON) pathways. We observed that DCK#10 cells became highly sensitive to MTX with an IC50 value of 70 ± 5 nmol/L, compared with NT1 cells whose IC50 value was 550 ± 50 nmol/L (Fig. 4A). DCK#10 cells were also sensitive to another inhibitor of the one-carbon metabolism pathway, 5-FU, with an IC50 value of 3.60 ± 0.2 μmol/L, and the IC50 value of NT1 cells was 6.3 ± 0.2 μmol/L (Fig. 4B). Next, the cells were treated with a potent c-Myc inhibitor 10058-F4, which inhibits the Myc–Max interaction and prevents the transactivation of Myc target genes (31). The IC50 value of DCK#10 cells was 35.5 ± 0.5 μmol/L, whereas that of NT1 cells was 78 ± 1 μmol/L (Fig. 4C), suggesting that DCK deletion increased the dependency on the MYC pathway. The gene expression of RRM2B was also upregulated in DCK#10 cells, presumably because of the increased demand of cells for de novo nucleotide synthesis. Upon treatment with Osalmid, the IC50 value of DCK#10 cells was 42 ± 2 μmol/L, whereas that of NT1 cells was 142 ± 1 μmol/L (Fig. 4D). In addition, genes associated with the arginine/proline/glutamine metabolism pathways were upregulated in DCK#10 cells. Previous studies have also reported that the upregulation of these amino acids plays a crucial role in cancer cell growth by upregulating nucleotide synthesis and aerobic glycolysis (16, 17, 32). As arginine and proline can be synthesized from glutamine, we treated DCK#10 cells with the glutamine metabolism inhibitor DON. We observed that DCK#10 cell proliferation was markedly reduced by DON treatment, with an IC50 value of 2.8 ± 0.05 μmol/L, whereas the NT1 cells showed an IC50 value of 31 ± 0.5 μmol/L (Fig. 4E).

Figure 4.

DCK KO cells are sensitive to methotrexate, 5-fluorouracil, Osalmid, MYC inhibitor, and the glutamine metabolism pathway inhibitor DON. A–E, Dose–response curve of methotrexate (0–10 μmol/L; A), 5-fluorouracil (0–100 μmol/L; B), MYC inhibitor 10058-F4 (0–200 μmol/L; C), Osalmid (0–200 μmol/L; D), and the glutamine analog DON (0–80 μmol/L; E) in DCK KO (DCK#10) and control NT1 cells. The data represent the mean ± SD of triplicate samples from three independent experiments. *, P < 0.05; **, P < 0.01, ***, P < 0.001. Cells were seeded at a density of 5,000 cells/well in 96-well plates and then incubated with the indicated drugs for 3 days. DCK#10 cells showed a higher sensitivity to these drugs than NT1 cells.

Figure 4.

DCK KO cells are sensitive to methotrexate, 5-fluorouracil, Osalmid, MYC inhibitor, and the glutamine metabolism pathway inhibitor DON. A–E, Dose–response curve of methotrexate (0–10 μmol/L; A), 5-fluorouracil (0–100 μmol/L; B), MYC inhibitor 10058-F4 (0–200 μmol/L; C), Osalmid (0–200 μmol/L; D), and the glutamine analog DON (0–80 μmol/L; E) in DCK KO (DCK#10) and control NT1 cells. The data represent the mean ± SD of triplicate samples from three independent experiments. *, P < 0.05; **, P < 0.01, ***, P < 0.001. Cells were seeded at a density of 5,000 cells/well in 96-well plates and then incubated with the indicated drugs for 3 days. DCK#10 cells showed a higher sensitivity to these drugs than NT1 cells.

Close modal

We further confirmed the importance of MYC and glutamine in the gemcitabine resistant post-gem cells. The post-gem cells showed higher MYC expression (Supplementary Fig. S4A) and increased sensitivity to 10058-F4 (Supplementary Fig. S4B). In addition, co-treatment with gemcitabine and 10058-F4 showed synergistic effects (Supplementary Fig. S4C). In terms of glutamine, the post-gem cells were more sensitive to DON (Supplementary Fig. S4D). The sequential treatment of the post-gem cells with DON and gemcitabine showed synergistic effects (Supplementary Fig. S4E) as previously reported (33). These findings indicate that DCK inactivation in PDAC cells increased their dependency on the one-carbon metabolism, MYC, and glutamine metabolism pathways.

MYC, OXPHOS, and glutamine metabolism pathways are activated in patients with PDAC with low DCK expression

To examine whether our findings in DCK#10 cells have any clinical relevance, we analyzed the PDAC patient datasets in TCGA (paad_tcga_pan_can_atlas_2018). We categorized the patients into two groups of high and low DCK mRNA expression [high DCK, n = 11 (Z-score > 1.25); low DCK, n = 12 (Z-score < −1.25)]. Patients with low DCK expression levels showed significantly lower progression-free survival (P = 0.0386; Fig. 5A and B). In the GSEA, hallmark MYC targets (V1 and V2), OXPHOS pathway, and KEGG arginine and proline metabolism pathways were significantly upregulated in patients with low DCK expression compared with those with high DCK expression (Fig. 5C and D), which corroborate the same scenario as observed in DCK-deficient cells. In further support of these findings in patients with PDAC, a principal component analysis of the genes belonging to MYC TARGETS V2 segregated the low and high DCK expression patient clusters (Fig. 5E). We also analyzed the gene expression datasets of the seven PDAC cell lines in CCLE broad 2019. The cell lines were classified into two groups on the basis of DCK expression, i.e., high DCK (RPKM > 10; CFPAC-1 and Capan-2) and low DCK (RPKM < 10; HPAF-2, AsPC-1, BxPC-3, Panc10.05, and SW1990) expression cell lines. GSEA revealed the upregulation of MYC targets V1 and V2, the OXPHOS pathway, and arginine and proline metabolism pathways in low DCK expression cell lines (Fig. 5F and G). These findings further support that the upregulation of MYC target genes, OXPHOS pathway, and KEGG arginine and proline metabolism pathways is a common feature observed in patients with PDAC and cell lines with low DCK expression, and this upregulation is required to maintain the proliferation of PDAC cells.

Figure 5.

MYC, OXPHOS, and glutamine metabolism pathways are activated in patients with PDAC with low DCK expression. A, Kaplan–Meier plot for progression-free survival of TCGA pancreatic cancer cases with low DCK (n = 12, Z-score <−1.25) or high DCK (n = 11, Z-score > 1.25) expression (P = 0.0386, Gehan-Breslow-Wilcoxon test). B, Box plot showing the relative mRNA expression levels (RSEM counts) of DCK in cases shown in A. The boxes extend the 25–75 percentiles; the line in the middle of the box represents the median value; the whiskers represent the 10 to 90 percentiles; and the patients are shown as dots. Gene expression data were obtained from TCGA PDAC dataset. C and D, GSEA revealed a significant enrichment of MYC targets and OXPHOS pathway (C) and arginine and proline metabolism pathways (D) in patients with low DCK expression compared with those with high DCK expression, which is shown in A and B. E, Principal component analysis of genes belonging to HALLMARK_MYC_TARGETS_V2 for 23 pancreatic cancer cases is shown in A and B. Gene expression levels were analyzed as logarithmic values. Sample Name: high, cases with high DCK expression; low, cases with low DCK expression. The cases with low and high DCK expression are partially segregated by expression profiles of the two groups. F and G, GSEA revealed a significant enrichment of MYC targets and OXPHOS pathway (F) and arginine and proline metabolism pathways (G) in PDAC cell lines with low DCK expression. Gene expression data were obtained from CCLE for the low DCK (AsPC-1, BxPC-3, HPAF-II, Panc10.05, and SW1990 cell lines) (mRNA expression level, RPKM < 10) and high DCK (Capan2 and CFPAC-1, RPKM > 10) expression groups. TCGA, The Cancer Genome Atlas.

Figure 5.

MYC, OXPHOS, and glutamine metabolism pathways are activated in patients with PDAC with low DCK expression. A, Kaplan–Meier plot for progression-free survival of TCGA pancreatic cancer cases with low DCK (n = 12, Z-score <−1.25) or high DCK (n = 11, Z-score > 1.25) expression (P = 0.0386, Gehan-Breslow-Wilcoxon test). B, Box plot showing the relative mRNA expression levels (RSEM counts) of DCK in cases shown in A. The boxes extend the 25–75 percentiles; the line in the middle of the box represents the median value; the whiskers represent the 10 to 90 percentiles; and the patients are shown as dots. Gene expression data were obtained from TCGA PDAC dataset. C and D, GSEA revealed a significant enrichment of MYC targets and OXPHOS pathway (C) and arginine and proline metabolism pathways (D) in patients with low DCK expression compared with those with high DCK expression, which is shown in A and B. E, Principal component analysis of genes belonging to HALLMARK_MYC_TARGETS_V2 for 23 pancreatic cancer cases is shown in A and B. Gene expression levels were analyzed as logarithmic values. Sample Name: high, cases with high DCK expression; low, cases with low DCK expression. The cases with low and high DCK expression are partially segregated by expression profiles of the two groups. F and G, GSEA revealed a significant enrichment of MYC targets and OXPHOS pathway (F) and arginine and proline metabolism pathways (G) in PDAC cell lines with low DCK expression. Gene expression data were obtained from CCLE for the low DCK (AsPC-1, BxPC-3, HPAF-II, Panc10.05, and SW1990 cell lines) (mRNA expression level, RPKM < 10) and high DCK (Capan2 and CFPAC-1, RPKM > 10) expression groups. TCGA, The Cancer Genome Atlas.

Close modal

Upregulation of glutamine metabolism was an early event observed among gemcitabine-resistant PDAC cells in persisters

Our analyses identified the importance of glutamine metabolism in gemcitabine sensitivity after long-term exposure to gemcitabine. A previous study demonstrated glutamine-dependent mTOR activation and chemoresistance in pancreatic cancer (34). Furthermore, disruption of the glutamine pathway could restore gemcitabine sensitivity in a PDAC cell line with acquired gemcitabine resistance (33). Therefore, we examined the importance of glutamine metabolism at an early phase of gemcitabine treatment. CFPAC-1 cells were treated with the doses of 13 nmol/L (IC50) and 10 μmol/L (IC90) of gemcitabine for 72 hours. The expression levels of the glutamine transporter SLC1A5 were upregulated in the two treatment groups in a dose-dependent manner (Fig. 6A). We also observed a significant increase in the expression of the genes involved in glutamine metabolism, including carbamoyl phosphate synthetase 2 (CAD), ornithine aminotransferase (OAT), and pyrroline-5-carboxylate reductase 1 (PYCR1), with gemcitabine treatment (Fig. 6A), whereas the expression of the mitochondrial glutamine pathway enzymes glutaminase 2 (GLS2) and glutamate dehydrogenase 1 (GLUD1) was not altered. To confirm the increase of glutamine uptake in the gemcitabine IC90-treated CFPAC-1 cells, we measured intracellular glutamine/glutamate levels. The gemcitabine treatment significantly increased intracellular glutamine/glutamate (Fig. 6B). These results suggest that the increase of glutamine uptake and glutamine metabolism play a crucial role in maintaining the persisters in the early stage of gemcitabine-treated PDAC cells. It has been previously reported that enhanced glutaminolysis in persisters induced more ferroptosis-mediated cell death (35). In our study, the genes responsible for ferroptosis (GLS2 and GLUD1) were not upregulated in this context. In addition, MYC was significantly increased in the gemcitabine treated CFPAC-1 cells (Fig. 6C). Altogether, our study findings suggest that MYC and glutamine metabolism pathways are novel targets to overcome gemcitabine resistance.

Figure 6.

Short-term treatment of gemcitabine induced the expression of glutamine-related genes in drug-naïve parental CFPAC-1 cells. A, RT-qPCR revealed an increased expression of genes involved in glutamine transport and metabolism in residual adherent cells under gemcitabine treatment at the concentrations of IC50 and IC90. B, Increased glutamine uptake in CFPAC cells treated with gemcitabine at IC90. C, RT-qPCR showing increased MYC expression in CFPAC cells treated with gemcitabine at IC50 and IC90. *, P < 0.05; **, P < 0.01; ***, P < 0.001; ns, not significant.

Figure 6.

Short-term treatment of gemcitabine induced the expression of glutamine-related genes in drug-naïve parental CFPAC-1 cells. A, RT-qPCR revealed an increased expression of genes involved in glutamine transport and metabolism in residual adherent cells under gemcitabine treatment at the concentrations of IC50 and IC90. B, Increased glutamine uptake in CFPAC cells treated with gemcitabine at IC90. C, RT-qPCR showing increased MYC expression in CFPAC cells treated with gemcitabine at IC50 and IC90. *, P < 0.05; **, P < 0.01; ***, P < 0.001; ns, not significant.

Close modal

Most patients with pancreatic cancer have poor prognosis. As a standard treatment for unresectable pancreatic cancer, gemcitabine plays a central role in the first-line chemotherapy; however, its maximum benefit is often hampered by chemoresistance during PDAC treatment (9). Although several combination therapies have been reported to sensitize gemcitabine-resistant PDAC cells (36–38), there is a need for further investigation of the mechanisms underlying chemoresistance to expand the therapeutic options. In this study, we conducted a CRISPR KO screening using a gemcitabine-sensitive CFPAC-1 cell line and demonstrated that DCK depletion is vital for the chemoresistance. RNA interference–mediated inhibition of DCK was previously reported to confer gemcitabine resistance in PDAC cell lines (19). The present result from our unbiased CRISPR/Cas9 KO screening indicates that DCK is the dominant effector of gemcitabine cytotoxicity in a gemcitabine-sensitive PDAC cell line. We further identified the four genes CRYBA2, DMBX1, CD36, and CROT9 as gene hits, and the respective inactivation of these genes was found to induce a slight resistance at low concentrations of gemcitabine (Fig. 2). In fact, all single KOs of these genes variably reduced the expression of DCK compared with the control, as demonstrated using RT-qPCR (Supplementary Fig. S4F). Although whether this is a direct or indirect effect has not yet been elucidated, this finding may partially support the mechanistic involvement of these gene hits in nucleotide metabolism.

Most previous studies conducted to date have investigated the resistance mechanism of gemcitabine cytotoxicity by focusing on the drug metabolism, including its transport, and DCK-mediated phosphorylation processes (39). Limited research has been reported on gene expression alterations induced by DCK deficiency, although it has been reported that lower DCK protein levels correlated with poor survival following gemcitabine treatment (20). When GSEA was applied to our gene expression data, it revealed a significant enrichment of several pathways, including MYC targets, folate/one-carbon pathway metabolism as well as glutamine/arginine/proline metabolism pathways, in DCK-deficient cells (Fig. 3). Increased demand for DNA synthesis is essential to maintain a highly proliferative phenotype of cancer cells (40). As DCK is a rate-limiting enzyme of the nucleotide salvage pathway, we speculate that our observation can be attributed to compensatory mechanisms for producing nucleotides. MYC acts as a transcription factor that directly activates genes responsible for nucleotide synthesis (41, 42). MYC also reprograms cellular metabolism in response to different cellular stresses and plays a crucial role in the folate/one-carbon metabolism pathway as well as glutaminolysis (43, 44). Along with this, the upregulated folate/one-carbon metabolism pathway presumably enhances the de novo nucleotide synthesis to retain a sufficient nucleotide pool in growing cancer cells (45). Furthermore, glutamine is a major source of carbon and nitrogen for ATP production as well as nucleotide synthesis (16), thereby sustaining cell proliferation. Because the arginine/proline metabolism pathways are directly coupled with glutamine metabolism pathway, these upregulated pathways are likely to be tightly interconnected (Fig. 7). Remarkably, the inhibitors of MYC, folate metabolism, and glutamine metabolism (10058-F4, MTX, 5-FU, DON) successfully reduced the survival of gemcitabine-resistant DCK KO cells compared with the inhibitor of de novo DNA synthesis, Osalmid (Fig. 4). Therefore, this result indicates that these pathways are functionally important for the survival of gemcitabine-resistant CFPAC-1 cells and proposes a druggable mechanistic connection involving amino acid metabolism and nucleotide synthesis (Fig. 7). Importantly, the upregulation of these pathways coupled with low DCK expression was similarly observed in cell lines as well as in patients with PDAC (Fig. 5). Of note, Capan-2 was an exception with higher levels of DCK but was resistant to gemcitabine among the cell lines tested (Fig. 1A and G). While DCK is a major determinant for the chemoresistance, additional genes could contribute to the chemoresistance in Capan-2. Indeed, our screening identified more than 20 hits (Fig. 1D and E). Meanwhile, we further observed a significant upregulation of the OXPHOS pathway in DCK KO cells (Fig. 4B). The mitochondrion acts as a major source of energy supplier as well as reactive oxygen species as a byproduct (46). Although the OXPHOS pathway inhibitor was not tested in this study, the possibility of using such an inhibitor in cancer therapy has been previously reported (47). Further investigation is required to determine the role of the OXPHOS pathway in the absence of DCK.

Figure 7.

Schematic model showing the compensatory mechanism to restore DNA pool in DCK KO cells. DCK KO contributes to gemcitabine resistance. MYC pathway was reported to be a critical regulator of the glutamine and one carbon pathways in various types of cancers (43–45). In the present study, we have shown that DCK KO upregulates MYC-mediated one-carbon pathway and glutamine metabolism, which leads to dependency on the de novo nucleotide synthesis pathway. Thus, MYC/glutamine dependency is a therapeutic vulnerability for gemcitabine-resistant pancreatic cancer caused by DCK inactivation.

Figure 7.

Schematic model showing the compensatory mechanism to restore DNA pool in DCK KO cells. DCK KO contributes to gemcitabine resistance. MYC pathway was reported to be a critical regulator of the glutamine and one carbon pathways in various types of cancers (43–45). In the present study, we have shown that DCK KO upregulates MYC-mediated one-carbon pathway and glutamine metabolism, which leads to dependency on the de novo nucleotide synthesis pathway. Thus, MYC/glutamine dependency is a therapeutic vulnerability for gemcitabine-resistant pancreatic cancer caused by DCK inactivation.

Close modal

Recent studies have clearly demonstrated that drug-tolerant persister cancer cells exhibit vulnerability to ferroptosis, a nonapoptotic form of iron-dependent cell death resulting from the accumulation of peroxidized phospholipids (48, 49). Therefore, the persister cell populations are preferentially sensitive to the inhibition of GSH/GPX4-dependent ROS scavenging (50). Consistently, the gemcitabine-resistant DCK KO CFPAC-1 cells showed a significant increase in the gene expression of glutamine/arginine/proline metabolism and folate/one-carbon metabolism pathways (Fig. 3), which could provide the source for GSH biosynthesis. Interestingly, our short-term (3 days) treatment of the drug-naïve parental CFPAC-1 cells with gemcitabine enhanced the expression of genes, which promote synthesis and transport of glutamine in the residual surviving cells in a dose-dependent manner (Fig. 6). This observation is probably an early reaction in drug-tolerant persisters, suggesting glutamine availability as a potential mechanism of escaping drug toxicity in an initial response for survival. Combining gemcitabine with glutamine inhibitors at the start of chemotherapy may help delay recurrence by reducing the pool of drug-tolerant persister cells.

To summarize, the unbiased genetic screening performed in our study revealed that DCK deficiency is the major mechanism underlying gemcitabine resistance in a gemcitabine-sensitive PDAC cell line. We further identified amino acid metabolism and nucleotide synthesis involving MYC as a therapeutic link for gemcitabine-resistant DCK-deficient PDAC cells. Furthermore, our results emphasized glutamine metabolism as a survival mechanism in drug-tolerant persisters. These druggable pathways were upregulated in patients with PDAC with low DCK expression. Therefore, our findings provide insights into expanding the therapeutic approaches for this refractory disease.

S. Dash reports other support from Japanese Government (MEXT) scholarship during the conduct of the study. M. Kawazu reports grants from Kobayashi Foundation for Cancer Research; and personal fees from Pacific Biosciences during the conduct of the study. H. Okada reports grants from Grant-in-Aid for Scientific Research on Innovative Areas; and grants from Vehicle Racing Commemorative Foundation during the conduct of the study. No disclosures were reported by the other authors.

S. Dash: Data curation, formal analysis, validation, investigation, visualization, writing–original draft. T. Ueda: Conceptualization, formal analysis, supervision, validation, methodology, writing–review and editing. A. Komuro: Investigation, methodology, writing–review and editing. H. Amano: Methodology, writing–review and editing. M. Honda: Methodology, writing–review and editing. M. Kawazu: Resources, data curation, writing–review and editing. H. Okada: Conceptualization, supervision, funding acquisition, methodology, project administration, writing–review and editing.

S. Dash is supported by Japanese Government (MEXT) Scholarship program. This work was supported by a grant from Grant-in-Aid for Scientific Research on Innovative Areas and the Vehicle Racing Commemorative Foundation to H. Okada.

We thank Y. Mine, S. Kurashimo, and the members of the core research facilities of Kindai University Faculty of Medicine for their support and excellent technical assistance. S. Dash is supported by Japanese Government (MEXT) Scholarship program. This work was supported by a grant from Grant-in-Aid for Scientific Research on Innovative Areas and the Vehicle Racing Commemorative Foundation to H. Okada.

The publication costs of this article were defrayed in part by the payment of publication fees. Therefore, and solely to indicate this fact, this article is hereby marked “advertisement” in accordance with 18 USC section 1734.

Note: Supplementary data for this article are available at Molecular Cancer Research Online (http://mcr.aacrjournals.org/).

1.
Pourshams
A
,
Sepanlou
SG
,
Ikuta
KS
,
Bisignano
C
,
Safiri
S
,
Roshandel
G
, et al
.
The global, regional, and national burden of pancreatic cancer and its attributable risk factors in 195 countries and territories, 1990–2017: a systematic analysis for the global burden of disease study 2017
.
Lancet Gastroenterol Hepatol
2019
;
4
:
934
47
.
2.
Rawla
P
,
Sunkara
T
,
Gaduputi
V
.
Epidemiology of pancreatic cancer: global trends, etiology, and risk factors
.
World J Oncol
2019
;
10
:
10
27
.
3.
Kaur
S
,
Smith
LM
,
Patel
A
,
Menning
M
,
Watley
DC
,
Malik
SS
, et al
.
A combination of MUC5AC and CA19–9 improves the diagnosis of pancreatic cancer: a multicenter study
.
Am J Gastroenterol
2017
;
112
:
172
83
.
4.
McGuigan
A
,
Kelly
P
,
Turkington
RC
,
Jones
C
,
Coleman
HG
,
McCain
RS
.
Pancreatic cancer: a review of clinical diagnosis, epidemiology, treatment and outcomes
.
World J Gastroenterol
2018
;
24
:
4846
61
.
5.
Long
J
,
Zhang
Y
,
Yu
X
,
Yang
J
,
LeBrun
DG
,
Chen
C
, et al
.
Overcoming drug resistance in pancreatic cancer
.
Expert Opin Ther Targets
2011
;
15
:
817
28
.
6.
Oberstein
PE
,
Olive
KP
.
Pancreatic cancer: why is it so hard to treat?
Therap Adv Gastroenterol
2013
;
6
:
321
37
.
7.
Jia
Y
,
Xie
J
.
Promising molecular mechanisms responsible for gemcitabine resistance in cancer
.
Genes Dis
2015
;
2
:
299
306
.
8.
Bergman
AM
,
Pinedo
HM
,
Peters
GJ
.
Determinants of resistance to 2',2'-difluorodeoxycytidine (gemcitabine)
.
Drug Resist Updat
2002
;
5
:
19
33
.
9.
Shah
AN
,
Summy
JM
,
Zhang
J
,
Park
SI
,
Parikh
NU
,
Gallick
GE
.
Development and characterization of gemcitabine-resistant pancreatic tumor cells
.
Ann Surg Oncol
2007
;
14
:
3629
37
.
10.
Bafna
S
,
Kaur
S
,
Momi
N
,
Batra
SK
.
Pancreatic cancer cells resistance to gemcitabine: the role of MUC4 mucin
.
Br J Cancer
2009
;
101
:
1155
61
.
11.
Reichard
P
.
Interactions between deoxyribonucleotide and DNA synthesis
.
Annu Rev Biochem
1988
;
57
:
349
74
.
12.
Arner
ES
,
Eriksson
S
.
Mammalian deoxyribonucleoside kinases
.
Pharmacol Ther
1995
;
67
:
155
86
.
13.
Jung
CP
,
Motwani
MV
,
Schwartz
GK
.
Flavopiridol increases sensitization to gemcitabine in human gastrointestinal cancer cell lines and correlates with downregulation of ribonucleotide reductase M2 subunit
.
Clin Cancer Res
2001
;
7
:
2527
36
.
14.
Cao
X
,
Mitra
AK
,
Pounds
S
,
Crews
KR
,
Gandhi
V
,
Plunkett
W
, et al
.
RRM1 and RRM2 pharmacogenetics: association with phenotypes in HapMap cell lines and acute myeloid leukemia patients
.
Pharmacogenomics
2013
;
14
:
1449
66
.
15.
Choi
BH
,
Coloff
JL
.
The diverse functions of nonessential amino acids in cancer
.
Cancers
2019
;
11
:
675
.
16.
Wise
DR
,
Thompson
CB
.
Glutamine addiction: a new therapeutic target in cancer
.
Trends Biochem Sci
2010
;
35
:
427
33
.
17.
Liu
W
,
Hancock
CN
,
Fischer
JW
,
Harman
M
,
Phang
JM
.
Proline biosynthesis augments tumor cell growth and aerobic glycolysis: involvement of pyridine nucleotides
.
Sci Rep
2015
;
5
:
17206
.
18.
Ohhashi
S
,
Ohuchida
K
,
Mizumoto
K
,
Fujita
H
,
Egami
T
,
Yu
J
, et al
.
Downregulation of deoxycytidine kinase enhances acquired resistance to gemcitabine in pancreatic cancer
.
Anticancer Res
2008
;
28
:
2205
12
.
19.
Saiki
Y
,
Yoshino
Y
,
Fujimura
H
,
Manabe
T
,
Kudo
Y
,
Shimada
M
, et al
.
DCK is frequently inactivated in acquired gemcitabine-resistant human cancer cells
.
Biochem Biophys Res Commun
2012
;
421
:
98
104
.
20.
Sebastiani
V
,
Ricci
F
,
Rubio-Viqueira
B
,
Kulesza
P
,
Yeo
CJ
,
Hidalgo
M
, et al
.
Immunohistochemical and genetic evaluation of deoxycytidine kinase in pancreatic cancer: relationship to molecular mechanisms of gemcitabine resistance and survival
.
Clin Cancer Res
2006
;
12
:
2492
7
.
21.
Oka
N
,
Komuro
A
,
Amano
H
,
Dash
S
,
Honda
M
,
Ota
K
, et al
.
Ascorbate sensitizes human osteosarcoma cells to the cytostatic effects of cisplatin
.
Pharmacol Res Perspect
2020
;
8
:
e00632
.
22.
Wang
T
,
Birsoy
K
,
Hughes
NW
,
Krupczak
KM
,
Post
Y
,
Wei
JJ
, et al
.
Identification and characterization of essential genes in the human genome
.
Science
2015
;
350
:
1096
101
.
23.
Shalem
O
,
Sanjana
NE
,
Hartenian
E
,
Shi
X
,
Scott
DA
,
Mikkelson
T
, et al
.
Genome-scale CRISPR-Cas9 knockout screening in human cells
.
Science
2014
;
343
:
84
7
.
24.
Ghandi
M
,
Huang
FW
,
Jane-Valbuena
J
,
Kryukov
GV
,
Lo
CC
,
McDonald
ER
3rd
, et al
.
Next-generation characterization of the cancer cell line encyclopedia
.
Nature
2019
;
569
:
503
8
.
25.
Nusinow
DP
,
Szpyt
J
,
Ghandi
M
,
Rose
CM
,
McDonald
ER
3rd
,
Kalocsay
M
, et al
.
Quantitative proteomics of the cancer cell line encyclopedia
.
Cell
2020
;
180
:
387
402
.
26.
Hoadley
KA
,
Yau
C
,
Hinoue
T
,
Wolf
DM
,
Lazar
AJ
,
Drill
E
, et al
.
Cell-of-origin patterns dominate the molecular classification of 10,000 tumors from 33 types of cancer
.
Cell
2018
;
173
:
291
304
.
27.
Subramanian
A
,
Tamayo
P
,
Mootha
VK
,
Mukherjee
S
,
Ebert
BL
,
Gillette
MA
, et al
.
Gene set enrichment analysis: a knowledge-based approach for interpreting genome-wide expression profiles
.
Proc Natl Acad Sci USA
2005
;
102
:
15545
50
.
28.
Kuo
MT
,
Chen
HHW
,
Feun
LG
,
Savaraj
N
.
Targeting the proline-glutamine-asparagine-arginine metabolic axis in amino acid starvation cancer therapy
.
Pharmaceuticals
2021
;
14
:
72
.
29.
D'Aniello
C
,
Patriarca
EJ
,
Phang
JM
,
Minchiotti
G
.
Proline metabolism in tumor growth and metastatic progression
.
Front Oncol
2020
;
10
:
776
.
30.
Phang
JM
,
Liu
W
,
Hancock
CN
,
Fischer
JW
.
Proline metabolism and cancer: emerging links to glutamine and collagen
.
Curr Opin Clin Nutr Metab Care
2015
;
18
:
71
7
.
31.
Yin
X
,
Giap
C
,
Lazo
JS
,
Prochownik
EV
.
Low molecular weight inhibitors of Myc-Max interaction and function
.
Oncogene
2003
;
22
:
6151
9
.
32.
Qiu
F
,
Huang
J
,
Sui
M
.
Targeting arginine metabolism pathway to treat arginine-dependent cancers
.
Cancer Lett
2015
;
364
:
1
7
.
33.
Chen
R
,
Lai
LA
,
Sullivan
Y
,
Wong
M
,
Wang
L
,
Riddell
J
, et al
.
Disrupting glutamine metabolic pathways to sensitize gemcitabine-resistant pancreatic cancer
.
Sci Rep
2017
;
7
:
7950
.
34.
Feng
M
,
Xiong
G
,
Cao
Z
,
Yang
G
,
Zheng
S
,
Qiu
J
, et al
.
LAT2 regulates glutamine-dependent mTOR activation to promote glycolysis and chemoresistance in pancreatic cancer
.
J Exp Clin Cancer Res
2018
;
37
:
274
.
35.
Gao
M
,
Yi
J
,
Zhu
J
,
Minikes
AM
,
Monian
P
,
Thompson
CB
, et al
.
Role of mitochondria in ferroptosis
.
Mol Cell
2019
;
73
:
354
63
.
36.
Namba
T
,
Kodama
R
,
Moritomo
S
,
Hoshino
T
,
Mizushima
T
.
Zidovudine, an anti-viral drug, resensitizes gemcitabine-resistant pancreatic cancer cells to gemcitabine by inhibition of the Akt-GSK3beta-Snail pathway
.
Cell Death Dis
2015
;
6
:
e1795
.
37.
Liu
ZJ
,
Xu
W
,
Han
J
,
Liu
QY
,
Gao
LF
,
Wang
XH
, et al
.
Quercetin induces apoptosis and enhances gemcitabine therapeutic efficacy against gemcitabine-resistant cancer cells
.
Anticancer Drugs
2020
;
31
:
684
92
.
38.
Yoshida
K
,
Toden
S
,
Ravindranathan
P
,
Han
H
,
Goel
A
.
Curcumin sensitizes pancreatic cancer cells to gemcitabine by attenuating PRC2 subunit EZH2, and the lncRNA PVT1 expression
.
Carcinogenesis
2017
;
38
:
1036
46
.
39.
Marechal
R
,
Bachet
JB
,
Mackey
JR
,
Dalban
C
,
Demetter
P
,
Graham
K
, et al
.
Levels of gemcitabine transport and metabolism proteins predict survival times of patients treated with gemcitabine for pancreatic adenocarcinoma
.
Gastroenterology
2012
;
143
:
664
74
.
40.
Pavlova
NN
,
Thompson
CB
.
The emerging hallmarks of cancer metabolism
.
Cell Metab
2016
;
23
:
27
47
.
41.
Liu
YC
,
Li
F
,
Handler
J
,
Huang
CR
,
Xiang
Y
,
Neretti
N
, et al
.
Global regulation of nucleotide biosynthetic genes by c-Myc
.
PLoS One
2008
;
3
:
e2722
.
42.
Mannava
S
,
Grachtchouk
V
,
Wheeler
LJ
,
Im
M
,
Zhuang
D
,
Slavina
EG
, et al
.
Direct role of nucleotide metabolism in C-MYC-dependent proliferation of melanoma cells
.
Cell Cycle
2008
;
7
:
2392
400
.
43.
Wise
DR
,
DeBerardinis
RJ
,
Mancuso
A
,
Sayed
N
,
Zhang
XY
,
Pfeiffer
HK
, et al
.
Myc regulates a transcriptional program that stimulates mitochondrial glutaminolysis and leads to glutamine addiction
.
Proc Natl Acad Sci USA
2008
;
105
:
18782
7
.
44.
Sun
L
,
Song
L
,
Wan
Q
,
Wu
G
,
Li
X
,
Wang
Y
, et al
.
cMyc-mediated activation of serine biosynthesis pathway is critical for cancer progression under nutrient deprivation conditions
.
Cell Res
2015
;
25
:
429
44
.
45.
Newman
AC
,
Maddocks
ODK
.
One-carbon metabolism in cancer
.
Br J Cancer
2017
;
116
:
1499
504
.
46.
Solaini
G
,
Harris
DA
.
Biochemical dysfunction in heart mitochondria exposed to ischemia and reperfusion
.
Biochem J
2005
;
390
:
377
94
.
47.
Ashton
TM
,
McKenna
WG
,
Kunz-Schughart
LA
,
Higgins
GS
.
Oxidative phosphorylation as an emerging target in cancer therapy
.
Clin Cancer Res
2018
;
24
:
2482
90
.
48.
Viswanathan
VS
,
Ryan
MJ
,
Dhruv
HD
,
Gill
S
,
Eichhoff
OM
,
Seashore-Ludlow
B
, et al
.
Dependency of a therapy-resistant state of cancer cells on a lipid peroxidase pathway
.
Nature
2017
;
547
:
453
7
.
49.
Rodriguez
R
,
Schreiber
SL
,
Conrad
M
.
Persister cancer cells: iron addiction and vulnerability to ferroptosis
.
Mol Cell
2022
;
82
:
728
40
.
50.
Hangauer
MJ
,
Viswanathan
VS
,
Ryan
MJ
,
Bole
D
,
Eaton
JK
,
Matov
A
, et al
.
Drug-tolerant persister cancer cells are vulnerable to GPX4 inhibition
.
Nature
2017
;
551
:
247
50
.