Metabolic Hallmarks for Purine Nucleotide Biosynthesis in Small Cell Lung Carcinoma

This article is featured in Selected Articles from This Issue, p. 5

Small cell lung cancer (SCLC) is a neuroendocrine tumor subtype that accounts for approximately 13% of all lung cancers (1, 2). SCLC involves a high frequency of mutations in the TP53 and RB1 genes; however, the proportion of gene mutations that represent effective drug targets is not as high as that in lung adenocarcinoma (3–8). Although recent evidence has shown the efficacy of immunotherapy for patients with SCLC (9, 10), SCLC remains a disease with a poor prognosis and limited treatment options are available. SCLC tumors are composed of small-sized cells with a round-to-fusiform shape, scant cytoplasm, and finely granular nuclear chromatin, and the pathology includes high rates of mitosis, apoptosis, and necrosis (11, 12).

Purine nucleotides and their derivatives are utilized for a variety of critical functions in cancer cells (13–15), and they are either biosynthesized through the de novo pathway or recycled from their free bases through the salvage pathway (16–18). One-carbon units supplied in the folate metabolic pathway, which are required for the purine de novo pathway, enable cancer cells to maintain high proliferation rates (19–21). Therefore, antifolates were among the earliest classes of drugs developed for clinical use in cancer chemotherapy (22), and they inhibit dihydrofolate reductase (DHFR), glycinamide ribonucleotide formyltransferase (GART), or thymidylate synthetase (TYMS; refs. 20, 23–25). The regulatory mechanisms of biosynthesis through the de novo pathway have recently been elucidated in detail (16, 26–29), and studies have described differences in purine salvage between normal and tumor cells and revealed clinical applications associated with such differences (30, 31). Previously, we have found that intermediate levels in the purine nucleotide metabolic pathway are elevated in human SCLC tumors and are involved in resistance to the anticancer drug PI3K inhibitors (32). Recent findings also show that phosphoribosyl pyrophosphate amidotransferase (PPAT) and inosine monophosphate dehydrogenase (IMPDH) in the purine de novo pathway are key enzymes for SCLC tumor growth that have attracted attention as potential therapeutic targets (33–35). However, the biological importance of the purine salvage pathway in SCLC progression has not yet been evaluated.

In this study, we used a SCLC model to demonstrate that deficiency of the purine salvage pathway inhibits tumor growth and increases sensitivity to antifolate, which inhibits the purine de novo pathway. Moreover, it was suggested that different types of SCLC cells exist that depend on either the de novo or salvage pathway for purine nucleotide biosynthesis. These results are important for understanding the regulatory mechanisms of purine metabolism in SCLC and for developing therapeutic strategies.

SCLC cell lines were purchased from the ATCC, European Collection of Authenticated Cell Culture (ECACC), and RIKEN Bio Resource Center. All cells were grown in RPMI1640 medium (Sigma-Aldrich) containing 10% (v/v) FBS (Sigma-Aldrich) at 37°C in a humidified atmosphere with 5% CO₂. All cells were annually tested for mycoplasma using the BioMycoX Mycoplasma PCR Detection Kit (CellSafe).

This study was approved by the Institutional Review Board (IRB) of the National Cancer Center (IRB no. 2011–201) and adhered to the principles of the Declaration of Helsinki. Human plasma from patients with SCLC was provided by the National Cancer Center Biobank after obtaining written informed consent for the use of their biological material. Plasma from healthy volunteers was purchased from Kohjin Bio Co., Ltd.

We used the C-SCOPE package for metabolome analysis, which was conducted by Human Metabolome Technologies, Inc. using capillary electrophoresis (CE) time-of-flight mass spectrometry for cation analysis and CE-tandem mass spectrometry (CE/MS-MS) for anion analysis, based on previously described methods (36, 37).

Formalin-fixed, paraffin-embedded sections of 59 SCLC tumor samples were subjected to hematoxylin and eosin (H&E) and IHC staining. For IHC, antigen retrieval was performed by boiling the slides at 95°C for 20 minutes in 10 mmol/L citrate buffer (pH 6.0). Primary antibodies were diluted in Antibody Diluent (Agilent) at a 1:100 dilution and incubated overnight at 4°C with a rabbit mAb against HPRT1 (ab109021; Abcam). The slides were then incubated with horseradish peroxidase (HRP)-conjugated secondary antibody (Agilent) and developed with 3,3-diaminobenzidine (Fujifilm) in 50 mmol/L Tris-buffer (pH 7.6) containing 0.3% hydrogen. A Nanozoomer system (Hamamatsu Photonics) was used for imaging. The positive of negative results for assessment of IHC of each case were listed in Supplementary Table S1.

Total RNA from the cell lines was isolated using TRizol Reagent (Thermo Fisher Scientific). cDNA was synthesized using the SuperScript VILO cDNA Synthesis Kit (Thermo Fisher Scientific). Real-time RT-PCR was performed using specific primers and a QuantStudio 3 system (Thermo Fisher Scientific). Real-time fluorescence monitoring of the PCR products was performed using TB Green Premix Ex Taq (TaKaRa Bio). The expression levels of specific genes are reported relative to the level of expression of GAPDH in the same master reaction. Synthesized primers were purchased from TaKaRa Bio with primer set ID HPRT1 (HA067805). GAPDH was used as the control, and the relative quantitation value compared with the calibrator for that target was expressed as 2^−(Ct-Cc).

Cell lysates were prepared using RIPA buffer (Cell Signaling Technologies) supplemented with the protease inhibitor cOmplete Mini Protease Inhibitor Cocktail (Roche). The protein concentrations were measured using the BCA assay (Thermo Fisher Scientific), and equal amounts of protein were separated on 4% to 20% Tris-glycine SDS-PAGE gels (Bio-Rad) and blotted onto polyvinylidene difluoride membranes (Bio-Rad). The membranes were blocked for 1 hour at room temperature with 5% milk in TBS supplemented with 0.1% Tween-20 (TBS-T; Cell Signaling Technologies), and the blocked membrane was then incubated with primary antibodies overnight at 4°C. After washing with TBS-T, the membranes were incubated with horseradish peroxidase–conjugated secondary antibody (GE Healthcare), and signals were developed using the ECL Prime Western Blotting Detection System (GE Healthcare). The blots were stripped using Restore stripping buffer (Thermo Fisher Scientific) and re-stained with GAPDH antibody (#2118; Cell Signaling Technologies). Chemiluminescence signals were acquired and analyzed using a FUSION Chemiluminescence Imaging System (VILBER). Because images showing the full-length blots were not obtained, the original images, including the positions of molecular weight markers above and below the bands of interest, were used.

The cytotoxic effects of the antifolates on human SCLC cells were analyzed using a WST-8 Cell Counting Kit (Dojindo) after 72 hours of exposure. Cells (1 × 10⁴) were plated in 96-well plates (Corning), with six to eight wells allotted per experimental condition (n = 6–8), and the compounds were added at eight decreasing concentrations. Aminopterin (AMT; Selleck), methotrexate (MTX; Towa Pharmaceutical Co.), pemetrexed (PMX; Selleck), and lemetrexol (LMX; Merck) were used in this study, and the end-product reversal of antifolates was analyzed in SCLC cells. MTX was kindly gifted by Towa Pharmaceutical Co. Cell proliferation was measured using the WST-8 assay in the presence of nucleotide precursor alone [thymidine (THY) or hypoxanthine (HXN)], antifolates alone, or antifolates + nucleotide precursor.

HPRT CRISPR/Cas9 knockout (KO) plasmids were obtained from Santa Cruz Biotechnology (sc-417332). The sgRNA sequences targeting the HPRT1 exon were obtained from three predesigned gRNAs: gRNA#A, 5′-GTTATGGCGACCCGCAGCCC-3′; gRNA#B, 5′-CTGTCCATAATTAGTCCATG-3′; and gRNA#C, 5′-TCTTGCTCGAGATGTGATGA-3′. SCLC cells were transfected with plasmids and incubated for 2 days under 5% CO₂ at 37°C. Cells were selected and cloned using puromycin. Up to 40 clones were used for the ELISA evaluation of HPRT1 protein expression using an anti-HPRT1 antibody (ab109021; Abcam).

The retrovirus packaging cell line Platinum-E (Plat-E) and retroviral vector pMXs (Funakoshi) were used. This retroviral system was developed by Dr. Kitamura (38). Plasmid construction was performed using a retrovirus system according to the manufacturer's instructions, and the human HPRT1 gene was inserted. Transfection was performed using Lipofectamine 2000 reagent (Thermo Fisher Scientific) according to the manufacturer's instructions.

All experimental nude mice were handled in accordance with the institutional guidelines established by the Animal Care Committee of the National Cancer Center. SCLC cells and HPRT1-knockout cells were injected into the subcutaneous tissues of 8-week-old female nude mice (CLEA). Tumor volume was calculated as the product of a scaling factor of 0.5, and tumor length and width were measured. After the tumor size reached >100 mm³, treatment with LMX (25 mg/kg) was initiated. LMX or vehicle was administered intravenously every 2 days (39). For IHC analysis, organs were obtained from the mice and fixed in 10% formalin. Xenograft SCLC tumors resected from mice after size measurements were used for metabolite measurements.

The excised tissues were cut into pieces, immediately frozen in liquid nitrogen, and stored at −80°C until metabolite extraction. The concentrations of intracellular anionic and cationic metabolites were measured using CE-TOF-MS, as previously described in detail (40, 41). To interpret the data, the metabolites levels significantly altered by HPRT1 KO (FDR < 0.05) were analyzed for pathway enrichment using the web-based MetaboAnalyst 5.0 software (http://www.metaboanalyst.ca) and SMPD library (https://www.smpdb.ca/).

The experimental design was described in a previous paper (42). The cells were cultured in RPMI1640 medium at a density of 5 × 10⁵ cells/mL. After adding [¹³C₂,¹⁵N]-glycine (Cambridge Isotope Laboratories, Catalog No. CNLM-1673-H) or [¹³C₅,¹⁵N₄]-hypoxanthine (Cambridge Isotope Laboratories, Catalog No. CNLM-7894-PK), the cells were cultured for an additional 2 and 6 hour, harvested, and pelleted. The reaction was quenched with 80% cold methanol. The cells were centrifuged at 9,100 × g for 10 minutes, and the metabolites in the supernatant were analyzed using CE-TOFMS. Synthesis (or flux) of AMP, ADP, and ATP through the de novo purine synthesis pathway was measured by [¹³C₂,¹⁵N] incorporation into the cells. Synthesis (flux) of AMP, ADP, and ATP via the purine salvage pathway was measured by [¹³C₅,¹⁵N₄] incorporation.

The GSE60052 dataset (43) that included SCLC tumors and nontumor lung tissues was obtained from the Gene Expression Omnibus (GEO) database. The normalized expression values of HPRT1 were analyzed using the GEO2R tool available in GEO datasets. Genes were plotted on a log2(TPM + 1) scale, with TPM representing transcripts per million.

Graphical data are presented as mean values, and error bars indicate standard deviation. The P values for pairwise comparisons were determined using two-tailed Student t test. Statistical differences among the groups were assessed via one-way analysis of variance, and adjusted P values were calculated using Dunnett multiple comparisons test. Statistical details on the experiments are provided in the associated figure legends.

Metabolome data are included in Supplementary Table S2. All other data will be available upon reasonable request.

To determine whether therapies targeting the purine de novo pathway effectively inhibit the growth of SCLC cells, we first examined the effects of antifolates on cell survival in the SCLC cell lines DMZ273, H1048, and SBC-5. The antifolate agents used in this study were AMT, MTX, PMX, and LMX, and their target enzymes and associated metabolic pathways are shown in Fig. 1A (44, 45). We found that in the PMX, MTX, and LMX treatments, the IC₅₀ of SBC-5 cells was obviously smaller than that of H1048 and DMS273 cells (Fig. 1B–E), suggesting that SBC-5 cells were highly sensitive to the inhibition of de novo purine nucleotide synthesis and that the sensitivity varied across SCLC cell lines. Compared with the other inhibitors, AMT showed a similar suppressive effect on cell growth in all cell lines (Fig. 1B).

In addition to purine de novo synthesis through DHFR and GART, several antifolates inhibit TYMS in pyrimidine metabolism (refs. 20, 23–25; Fig. 1A). Thus, we examined whether metabolites downstream of the enzymatic reaction inhibited by antifolates, such as thymidine and hypoxanthine (Fig. 1A), restored cell survival under antifolate treatment conditions. Although the combination of thymidine and hypoxanthine did not affect the growth of SCLC cells (Fig. 2A), it prevented the inhibitory effects of AMT, MTX, PMX, and LMX (Fig. 2B–E). LMX prevented purine de novo biosynthesis by blocking GART only (Fig. 1A), and its inhibitory effect on cell proliferation was reversed by treatment with hypoxanthine alone (Fig. 2), suggesting that LMX specifically inhibits the purine de novo pathway.

Next, we examined whether the purine salvage pathway is important for nucleic acid synthesis and cell proliferation in SCLC cells. HPRT1 is a key enzyme in the purine salvage pathway (46, 47). HPRT1 catalyzes the conversion of hypoxanthine to inosine monophosphate (IMP) and guanine to guanosine monophosphate (GMP) via transfer of the 5-phosphoribosyl group from 5-phosphoribosyl 1-pyrophosphate (refs. 16, 17, 47; Fig. 1A). To determine whether human SCLC tumor tissues express HPRT1, we performed IHC evaluations of tumors in 59 SCLC cases. Representative images of HPRT1 positivity (HPRT1+) and HPRT1 negativity (HPRT1−) with anti-HPRT1 antibody in SCLC tissues are presented in Fig. 3A and B. HPRT1 was positive in 86.4% of cases (51 samples) and negative in 13.6% of cases (eight samples; Supplementary Table S1). HPRT1 expression was also observed in all SCLC cell lines, and its levels varied among the lines (Fig. 3C; Supplementary Fig. S1). In addition, HPRT1 mRNA levels in healthy lung and SCLC tissues were evaluated using a public dataset (GSE60052; ref. 43), and although HPRT1 expression tended to be higher in SCLC tumors, no significant differences were observed (Supplementary Fig. S2).

In SCLC cells, variations in the expression of HPRT1 in the purine nucleotide salvage pathway and sensitivity to inhibition of the de novo purine nucleotide synthesis pathway indicate that the dependency of purine nucleotide synthesis on these pathways is cell-specific. Therefore, we investigated whether HPRT1 expression is involved in sensitivity to inhibition of the de novo purine nucleotide synthesis pathway using antifolates. We deleted the HPRT1 gene through CRISPR/Cas9 gene editing and screened HPRT1-KO clones (deleted HPRT1: DH) via ELISA (Supplementary Fig. S3). We confirmed the loss of HPRT1 protein in DMS273 (Fig. 4A), H1048 (Fig. 4B), and SBC-5 cells (Fig. 4C) by western blotting. Interestingly, cells lacking HPRT1 displayed more severe survival defects in the presence of LMX, and this effect was more apparent in DMS273 and H1048 cells than in SBC-5 cells (Fig. 4D–F). To confirm whether this phenotype was specific to HPRT1 function, we transduced wild-type (WT) HPRT1 into HPRT1-KO cells (Supplementary Fig. S4). Ectopic expression of HPRT1 completely rescued the sensitivity to LMX treatment (Fig. 4G–I). Furthermore, hypoxanthine enhanced the LMX resistance induced by HPRT1 expression (Fig. 4J). These results indicate that HPRT1 expression negatively regulates sensitivity to inhibition of de novo purine nucleotide synthesis.

To examine whether the purine nucleotide salvage pathway is required for tumor growth in vivo, we used an immunodeficient murine xenograft model subcutaneously transplanted with HPRT1-KO cells. HPRT1-KO or parent cells as controls (1 × 10⁷ cells) were subcutaneously transplanted into the flanks of SCID mice, and tumor size was measured over time (Fig. 5A–D). HPRT1-KO significantly suppressed the tumor size in DM273 and H1048 cells (Fig. 5A and B). However, loss of HPRT1 had no effect on the growth of tumors derived from SBC-5 cells (Fig. 5C). To confirm the loss of HPRT1 in subcutaneous tumors, we performed IHC staining for the HPRT1 protein and observed a lack of staining in HPRT1-KO tumors but not in WT tumors (Supplementary Fig. S5). Furthermore, expression of HPRT1 in DMS273/HPRT1-KO cells restored tumor growth (Fig. 5D). These results suggest that HPRT1 expression promotes tumor growth in SCLC, which is dependent on the salvage pathway of purine nucleotide biosynthesis.

Next, we performed a metabolome analysis of xenografted tumors from HPRT1-KO cells (Supplementary Table S2). The results showed that 21 and 9 metabolites were significantly upregulated and downregulated by HPRT1 deletion, respectively (Fig. 6A). Interestingly, levels of 5-aminoimidazole-4-carboxamide-1-β-d-ribofuranoside (AICAR), which is intermediate in the purine de novo pathway, were drastically increased in the HPRT1-KO tumors (Fig. 6A and C). Pathway enrichment analysis of the metabolites altered by HPRT1-KO identified the pentose phosphate pathway (PPP) and nucleotide sugar metabolism pathway (Fig. 6B). PPP is involved in the purine de novo pathway (Fig. 1A), and the levels of PPP intermediates fructose 1,6-bisphosphate (F1,6P), 3PG, and 6-phosphogluconate (6PG) were significantly upregulated by HPRT1-KO in vivo (Fig. 6C). Furthermore, ectopic HPRT1 induction in HPRT1-KO tumors reduced the levels of AICAR and PPP intermediates (Fig. 6D). These results suggest that purine salvage pathway activity affects the de novo synthesis of purine nucleotides.

To further examine the relationship between the purine de novo and salvage pathways, we conducted a stable isotope tracer analysis using [¹³C₂, ¹⁵N]-glycine and [¹³C₅, ¹⁵N₄]-hypoxanthine to assess the activity of the two pathways in purine metabolism, respectively (Fig. 7A). HPRT1-KO cells transfected with the control vector (KO+V) or HPRT1 cDNA (KO+H) were used in this analysis. The production of [¹³C₂, ¹⁵N]-glycine-derived nucleotides AMP (M+3), ADP (M+3), and ATP (M+3) in HPRT1-KO (KO+V) cells was higher than that in HPRT1-expressing KO+H cells, suggesting that the purine de novo pathway was activated by HPRT1 deletion (Fig. 7B; Supplementary Fig. S6). In contrast, [¹³C₅, ¹⁵N₄]-hypoxanthine-derived AMP(M+9), ADP (M+9), and ATP (M+9) levels were only increased in the KO+H cells (Fig. 7C; Supplementary Fig. S6). These results indicate that HPRT1 deletion, which prevents the purine salvage pathway, promotes de novo purine synthesis and increases the dependence on the de novo pathway in purine nucleotide biosynthesis and cell survival.

We then examined the effect of purine de novo pathway inhibition on tumor growth in HPRT1-KO cells in vivo. The size of tumors from KO+V cells was significantly decreased by the LMX treatment (Fig. 7D, left), whereas ectopic HPRT1 expression (KO+H) abolished this effect (Fig. 7D, right). Under these experimental conditions, mice displayed no weight loss or side effects (Supplementary Fig. S7). These result suggest that HPRT1 expression reduces the dependency of the purine de novo pathway in purine nucleotide biosynthesis and the suppressive effect of LMX on tumor progression.

In this study, we revealed that the purine salvage pathway regulates tumor growth and the effects of drugs targeting the purine de novo pathway on cell survival in SCLC. Our results also suggest that different SCLC cell lines exhibit a preference for either the de novo or salvage pathway for purine nucleotide biosynthesis. Moreover, SCLC cells exhibit plasticity between the purine de novo and salvage pathways.

We performed a metabolomic analysis of plasma from patients with SCLC and surprisingly found that purine nucleotide-related metabolites, including hypoxanthine (HXN), IMP, and UDP-glucose (UDP-Glc), were significantly higher in the plasma of patients with SCLC than in healthy volunteers (HV; Supplementary Fig. S8). Many apoptotic and necrotic cells are observed in SCLC tumors (11, 12), which may indicate that abundant nucleic acid-derived metabolites are leaked from dead cells into the extracellular matrix and blood. This would be advantageous for the survival and proliferation of SCLC cells that rely on the purine salvage pathway for nucleotide biosynthesis. Nucleic acids in the microenvironment derived from dead cells have also been reported to function as “find me” and “goodbye” signals for phagocytosis (48–51), holding unique significance in tumor biology. During the process of malignant growth (52, 53), tumor tissues often exhibit nutritional heterogeneity, such as glucose deprivation; thus, the mechanism of recycling dead cell-derived nucleic acids could represent a strategy for SCLC cells to adapt to areas of nutrient deprivation.

Meanwhile, differences in sensitivity to LMX, PMX, and MTX were observed among SCLC cell lines, although only AMX showed no such differences, suggesting that AMX has low specificity for the inhibition of folate metabolism. Given that LMX inhibits the purine biosynthesis pathway by blocking GART only and that its cell growth inhibitory effect was inhibited by hypoxanthine alone (Fig. 2), LMX may be the most specific inhibitor of the purine de novo pathway compared with other antifolates. SBC-5 cells were highly sensitive to PMX, MTX, and LMX, whereas the cells exhibited a low effect of HPRT1 deletion on tumor suppression in vivo (Fig. 5) and promotion of LMX sensitivity (Fig. 4). These suggest that cell proliferation in SBC-5 is more purine de novo pathway-dependent than that in H1048 and DMS273 cells, which may indicate that the preference for purine de novo and salvage pathways differs from cell to cell.

In this study, we found that HPRT1 is a promising therapeutic target for SCLC treatment. If we could select purine salvage pathway-dependent cancers and inhibit HPRT1, then the antitumor effect could be expected to be high. A new class of HPRT1 inhibitors has been developed for the treatment of malaria (54, 55), and a human HPRT1 inhibitor with high inhibitory activity has also been reported, which may hold promise as an anticancer agent against SCLC. Conversely, if purine de novo pathway-dependent SCLC is selected and treated with antifolate, which is already in clinical application, then it may represent a highly effective treatment. An important step in this process is the development of biomarkers to determine purine de novo and salvage preferences; however, in this study, a simple correlation between HPRT1 expression and either in vitro sensitivity to LMX or in vivo tumor growth was not observed (Fig. 3; Supplementary Fig. S1), suggesting that HPRT1 expression alone was not sufficient to determine such preferences. Cancer therapy strategies that inhibit both de novo and salvage pathways are also attractive, although many problems may need to be overcome in terms of side effects on normal tissues. On the clinical side, antifolate agents such as PMX have been used as a treatment for non–small cell lung cancer (NSCLC) rather than SCLC; however, further research is required on the purine de novo and salvage preferences in NSCLC.

In addition, intracellular purine nucleotide levels positively regulate mTORC1 signaling (56), and HPRT1-KO might reduce the mTORC1 activity in SCLC. mTOR is important for SCLC growth and is defined as a promising therapeutic target (57). It is possible that the antitumor effects of HPRT1 inhibition are partially mediated through the suppression of mTORC1.

Metabolic flux analysis using stable isotopes revealed compensatory plasticity in the purine de novo and salvage pathways. The biological significance and molecular mechanism of plasticity and whether it differs between individual cells remain to be further examined. Although we focused on HPRT1 in this study, the purine salvage pathway is also mediated by adenine phosphoribosyltransferase and methylthioadenosine phosphorylase (16, 31), and the contribution of HPRT1 to the salvage pathway in SCLC requires further investigation.

In this study, we show that the HPRT1-mediated salvage pathway is important for the proliferation of a particular type of SCLC, and further suggest that there is a preference for purine de novo and salvage pathways among SCLC cells; there is also plasticity in these pathways. These findings provide a deeper understanding of purine metabolism and open new avenues for the development of therapeutic interventions for SCLC.

S. Umemura reports personal fees from Chugai pharmaceutical and MSD outside the submitted work. H. Udagawa reports grants from Boehringer Ingelheim and Takeda outside the submitted work. M. Tsuboi reports personal fees from Eli Lilly Japan, Chugai Pharmaceutical Co., Ltd., Taiho Pharma, Medtronic Japan, Ono Pharmaceutical Co., Ltd., MSD, Bristol-Myers Squibb KK, Teijin Pharma, Novartis; grants from MSD, AstraZeneca KK, Ono Pharmaceutical Co., Ltd., Bristol-Myers Squibb KK, Novartis, and MiRXES outside the submitted work. K. Goto reports grants and personal fees from Amgen Inc., Amgen K.K., AstraZeneca K.K., Bayer Yakuhin, Ltd., Boehringer Ingelheim Japan, Inc., Blueprint Medicines Corporation, DAIICHI SANKYO Co., Ltd., Eisai Co., Ltd., Eli Lilly Japan K.K., Haihe Biopharma Co., Ltd., Janssen Pharmaceutical K.K., Merck Biopharma Co., Ltd., Novartis Pharma K.K., Ono Pharmaceutical Co., Ltd., Riken Genesis Co., Ltd., Taiho Pharmaceutical Co., Ltd., Takeda Pharmaceutical Co., Ltd.; grants from Craif Inc., Amgen Astellas BioPharma K.K., Bristol-Myers Squibb K.K., Ignyta, Inc., Kissei Pharmaceutical Co., Ltd., Kyowa Kirin Co., Ltd., Life Technologies Japan Ltd., Loxo Oncology, Inc., LSI Medience Corporation, Medical& Biological Laboratories Co., Ltd., Merus N.V., MSD K.K., Pfizer Japan Inc., Pfizer R&D Japan G.K., Precision Medicine Asia Co., Ltd., Sumitomo Pharma Co., Ltd., Spectrum Pharmaceuticals, Inc., Sysmex Corporation, Turning Point Therapeutics, Inc., and Xcoo, Inc.; personal fees from Bayer HealthCare Pharmaceuticals Inc., Guardant Health Inc., Medpace Japan K.K., Nippon Kayaku Co., Ltd., Otsuka Pharmaceutical Co., Ltd. outside the submitted work. S.S. Kobayashi reports grants from Boehringer Ingelheim, MiRXES, Johnson&Johnson, and Taiho; personal fees from AstraZeneca, Boehringer Ingelheim, Bristol Meyers Squibb, Chugai, and Takeda outside the submitted work. No disclosures were reported by the other authors.

S. Tabata: Conceptualization, data curation, investigation, visualization, methodology, writing–original draft, writing–review and editing. S. Umemura: Conceptualization, resources, data curation, supervision, investigation, visualization, methodology, writing–original draft, writing–review and editing. M. Narita: Data curation, validation, investigation, methodology. H. Udagawa: Investigation, methodology, writing–review and editing. T. Ishikawa: Investigation, methodology. M. Tsuboi: Investigation, methodology, writing–review and editing. K. Goto: Resources, supervision, writing–original draft. G. Ishii: Resources, supervision, investigation, methodology, writing–review and editing. K. Tsuchihara: Resources, supervision, writing–review and editing. A. Ochiai: Conceptualization, resources, supervision, methodology, writing–review and editing. S.S. Kobayashi: Resources, supervision, methodology, writing–review and editing. T. Soga: Conceptualization, resources, supervision, methodology, writing–review and editing. H. Makinoshima: Conceptualization, resources, data curation, supervision, funding acquisition, investigation, visualization, methodology, writing–original draft, project administration, writing–review and editing.

We thank Ami Maruyama, Yuzo Sato, Dr. Joji Nakayama, and Yuka Nakamura for their technical assistance, and Dr. Phillip Wong for carefully reviewing the manuscript and providing critical comments. We also thank all members of the Shonai Regional Industry Promotion Center for their assistance. We would like to thank Editage (www.editage.com) for the English language editing. This study was supported by the National Cancer Center Research and Development Fund (31-A-6) and JSPS KAKENHI, with grants 17K07189 and 20K07627 awarded to H. Makinoshima. This work was supported in part by research funds from the Yamagata Prefectural Government and City of Tsuruoka.

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.