Gastric cancer is the fifth most diagnosed cancer and the fourth leading cause of cancer-related mortality worldwide (1). Advances in genomic technologies have enabled a better understanding of cancers by uncovering genetic heterogeneity that enabled molecular classifications (2-5) which can affect prognosis and therapeutic outcomes in gastric cancer. The Cancer Genome Atlas (TCGA) project classified gastric cancers into four molecular subtypes: tumors positive for Epstein-Barr virus (EBV), chromosomal instable tumors, tumors with microsatellite instability (MSI), and genomically stable (GS) tumors. Of these, the GS subtype is associated with epithelial-to-mesenchymal transition (EMT) features (2). Later, the EMT-activated gastric cancer subtype was consistently found in diverse cohorts of gastric cancer (4-6). The EMT-subtype gastric cancers are also chemo-resistant due to their low proliferation rate (6). In addition, the rare druggable oncogenic mutations and low mutation burden of this subtype render targeted therapies and immunotherapies ineffective, respectively (7, 8). Consequently, there are vital unmet needs that demand the development of new therapeutic interventions targeting the EMT-subtype gastric cancer to improve the survival outcomes of these patients.
In the present study, we conducted a large-scale chemical screen using 37 gastric cancer cell lines and 48,467 synthetic small-molecule chemical compounds to identify novel drug candidates with selective cytotoxicity against EMT-subtype gastric cancer cell lines. We further investigated the molecular targets and the underlying selectivity mechanisms of one of the hit compounds, YK-135. This study highlights the altered central energy metabolism as a synthetic lethal target pathway for the selective intervention of EMT-subtype gastric cancers. Hence, the findings from this study provide a novel precision drug candidate for further drug development.
Previously, we classified a panel of 29 gastric cancer cell lines into intrinsic molecular subtypes by analyzing whole exome sequencing and RNA sequencing data (9). Here, we have expanded this panel by adding eight additional gastric cancer cell lines, SNU5, SK-GT-2, IM95, MKN7, NUGC3, NUGC4, RERF-GC-1B, and SCH (Supplementary Table 1). We performed an unsupervised hierarchical clustering analysis based on expression levels of EMT signature genes (5, 9) and found that seven of the 37 gastric cancer cell lines exhibit robust EMT features (Fig. 1A). We then performed a high-throughput drug screen using 37 gastric cancer cell lines and 48,467 small-molecule compounds to identify compounds with selective cytotoxicity against EMT-subtype gastric cancer cell lines (Fig. 1B). The readout for our screening process was the cytotoxicity of each compound using the CellTiter-Go assay, which measures ATP concentration in each well of the screening plates. The measurements from this assay were highly correlated with the live cell counts measured by combined staining of nuclei with Hoechst and propidium iodide (PI) (Supplementary Fig. 1). A series of screening identified seven hit compounds (Fig. 1B, Supplementary Data 1) that had a wider selective margin between the EMT cancer cell lines and the non-EMT lines in a cumulative distribution function (AUC KS test P ≤ 0.05 and KS distance (
Our initial observation that YK-135 inhibited mitochondrial respiration in SNU484 led us to further examine the consequence of YK-135 treatment in an expanded gastric cancer cell line panel including two EMT-(SNU484 and HGC27) and two non-EMT-subtype gastric cancer cell lines (MKN45 and NCI-N87). Measuring the cellular oxygen consumption rate (OCR) using the Seahorse XF analyzer revealed a significant inhibition of OCR by YK-135 in all cell lines at concentrations of 5-10 μM (Fig. 2A), which was more potent than the biguanides (metformin or phenformin), as they usually inhibit OCR at millimolar concentrations (10). Next, to determine which mitochondrial complex YK-135 inhibits, we examined mitochondrial respiration in membrane-permeabilized cells by supplying either complex I or complex II substrate. In the complex I-linked respiration assay supplied with adenosine-5’-diphosphate (ADP) and the complex I substrates pyruvate and malate, YK-135 (10 μM) inhibited oxygen consumption in permeabilized SNU484 cells similar to the complex I inhibitor, rotenone (Fig. 2B, Left). Meanwhile, YK-135 had a minor effect on the oxygen consumption in the cells supplemented with ADP and the complex II substrate succinate, wherein mitochondrial complex II inhibitor, 2-thenoyltrifluoroacetone (TTFA), efficiently inhibited complex II-linked respiration (Fig. 2B, Right). YK-135 also dose-dependently inhibited the mitochondrial complex I enzyme activity in isolated mitochondria of SNU484 cells
Biguanide-mediated inhibition of mitochondrial respiration activates AMP-activated protein kinase (AMPK), a major energy sensor in the cell, by increasing the intracellular AMP and ADP levels (11). YK-135 treatment activated AMPK, as demonstrated by increased phospho-AMPK (Thr172) levels, only in EMT-subtype gastric cancer cell lines (Fig. 3A), indicating that the ATP depletion in the EMT-subtype gastric cancer cell lines was more robust than that in the non-EMT cell lines. AMPK activation causes a cell cycle arrest and inhibits the synthesis of macromolecules required for the cell growth and proliferation (12), as evidenced by reduced S6K phosphorylation in EMT-subtype cell lines (Fig. 3A). Therefore, we next examined whether YK-135 selectively exerts anti-proliferative effects on EMT-subtype cancer cell lines. Our investigation revealed that YK-135 treatment induced G2/M cell cycle arrest specifically in EMT-subtype gastric cancer cell lines (HGC27 and SNU484) but not in non-EMT gastric cancer cell lines (MKN45 and NCI-N87) (Fig. 3B). It is known that the intact mitochondrial respiratory electron transport chain (ETC) generates an electrochemical proton gradient that establishes the mitochondrial membrane potential used by mitochondrial complex V to generate ATP. However, impaired mitochondrial ETC leads to a decreased mitochondrial membrane potential and increased harmful reactive oxygen species (ROS) generation, which also contributed to cell cycle arrest and cell proliferation inhibition (13). We, therefore, examined the effect of YK-135 on the mitochondrial membrane potential and mitochondrial ROS production. Upon YK-135 treatment, SNU484 and HGC27 cells, but not MKN45 and NCI-N87 cells, showed a decreased mitochondrial membrane potential (measured by the MitoTrackerTM Red CMXRos dye staining) (Fig. 3C) and significantly increased ROS production (Fig. 3D). To further understand the consequence of mitochondrial dysfunction caused by YK-135 treatment in EMT-subtype gastric cancer cells, we examined its effect on cell death. Flow cytometry analysis using PI and Annexin V staining (Annexin V-FITC Apoptosis Kit) showed significant increases in cellular apoptosis 72 h post-YK-135 treatment, as compared with the vehicle control in SNU484 and HGC27 cells but not in MKN45 and NCI-N87 cells (Fig. 3E). We further examined the expression of cleaved PARP-1 (c-PARP), a hallmark of cellular apoptosis. We observed that YK-135 treatment-induced PARP-1 cleavage only in the EMT-subtype gastric cancer cells in a time-dependent manner (Fig. 3F). These data collectively indicated that YK-135 induced robust ATP depletion and consequently led to selective apoptotic cell death in EMT-subtype gastric cancer cell lines.
Since YK-135 treatment activates AMPK only in the EMT-subtype gastric cancer cells (Fig. 3A), we hypothesized that EMT-subtype cancer cells might have a defect in compensatory mechanisms for dealing with mitochondrial ATP depletion. In this regard, GSEA analysis revealed that the glycolysis-related gene set was significantly depleted in the EMT-subtype gastric tumors in the TCGA dataset (Fig. 4A). Since glycolysis is critical for the survival of cancer cells in the presence of mitochondrial complex I inhibitor (14), we measured extracellular acidification rate (ECAR), an indicator of glycolysis, in 24 gastric cancer cell lines, and found that there was a positive correlation between glycolytic capacity and resistance to YK-135 (
The oxidative phosphorylation (OXPHOS) pathway plays a crucial role in cancer cells by providing ATP and metabolites to support rapid proliferation and migration. Thus, OXPHOS inhibition holds excellent potential for an anticancer strategy with many therapeutic opportunities. To date, the biguanide metformin which can act as a putative mitochondrial complex I inhibitor has been tried in multiple clinical trials as an anticancer agent in breast and colorectal cancers (14, 15). Beyond metformin, the more potent biguanide phenformin (for hepatocellular carcinoma (16) and mitochondrial complex I inhibitors such as IACS-010759 (for Glioblastoma and AML (17) have shown efficacy in clinical trials. In addition, other mitochondrial OXPHOS inhibitors, such as Gracillin (complex II inhibitor for non-small cell lung carcinoma (18) and gboxin (complex V inhibitor for glioblastoma (19), have shown anticancer efficacy in pre-clinical studies. Despite these successes, the use of OXPHOS inhibitors has some limitations. First, in the absence of any precise biomarker, these drugs are likely to exhibit some side effects toward non-cancerous cells with high-energy demands. Mitochondrial dysfunction or fragmentation by OXPHOS inhibitors induces a compensatory shift to glycolysis and pyruvate fermentation, resulting in secretion and accumulation of lactate leading to lactic acidosis (16, 20). Second, cancer cells exhibit flexibility in switching energy metabolism between glycolysis and OXPHOS to adapt to the tumor microenvironment, which causes resistance to OXPHOS inhibitor treatment. In the case of malignant lymphocytes, metabolic adaptation to biguanides is acquired by transcriptionally reprogramming glucose metabolism (21). Thus, identifying tumor subtypes that may benefit from OXPHOS inhibition is necessary to overcome the limitation of mitochondrial inhibitors in clinical applications.
Thus, our discovery that EMT-subtype gastric tumors and gastric cancer cell lines exhibit significantly reduced expression of the glycolysis-related genes and low glycolytic capacity, respectively, provides evidence supporting the utility of mitochondrial complex I inhibitors against EMT-subtype gastric cancers. However, of the various mitochondrial complex I inhibitors, YK-135 showed the highest selectivity against EMT-subtype gastric cancer cell lines, indicating that YK-135 has a unique pharmacological property differing from other mitochondrial complex I inhibitors. One of the possible modes of action of YK-135 explaining the higher selectivity to the EMT-subtype gastric cancer cell lines might be its partial inhibitory effect on mitochondrial complex II, as shown in Fig. 2B and Supplementary Fig. 4, since mitochondrial complex II can serve as a bypass when complex I is blocked. Despite the potent efficacy of YK-135 in gastric cancer cell lines, it has limited bioavailability and stability
In conclusion, the present study demonstrates that a novel mitochondrial complex I inhibitor, YK-135, exhibits specific cytotoxicity toward EMT-subtype gastric cancer cell lines exerted due to their impaired glycolytic capacity. This finding holds tremendous translational implications for synthetic lethal strategies against treatment-refractory EMT-subtype gastric cancers. Further studies are needed to improve the poor pharmacokinetic properties of YK-135 and to determine if YK-135 has similar effects in in other cancer types.
Materials and methods are available in the Supplemental Information.
The chemical library used in this study was kindly provided by Korea Chemical Bank (http://www.chembank.org/) of Korea Research Institute of Chemical Technology (Daejeon, Korea).
This study was supported by grants from the Korea Health Technology R&D Project through the Korea Health Industry Development Institute (HI14C1324), the National Research Foundation of Korea (NRF) (2020R1A2C3007792 and 2022 R1A2B5B03001199), and the “Team Science Award” of Yonsei University College of Medicine (6-2021-0194). SBK was supported by the Brain Pool Program funded by the Ministry of Science and ICT through the NRF (2019H1D3A2A01050712). HK was supported by the Global Ph.D. Fellowship Program funded by the NRF (2019H1A2A1075632).
HSK is a founder, chief scientific officer, and shareholder of Checkmate Therapeutics Inc. The authors have no other conflicting interests.