Isorhamnetin Suppresses Human Gastric Cancer Cell Proliferation through Mitochondria-Dependent Apoptosis

Derivates of natural products have been wildly utilized in the treatment of malignant tumors. Isorhamnetin (ISO), a most important active ingredient derived from flavonoids, shows great potential in tumor therapy. However, the therapeutic effects of ISO on gastric cancer (GC) remain unclear. Here, we demonstrate that ISO treatment dramatically inhibited the proliferation of two types of GC cells (AGS-1 and HGC-27) both in vitro and in vivo in time- and dose-dependent manners. These results are consistent with the transcriptomic analysis of ISO-treated GC cells, which yielded hundreds of differentially expressed genes that were enriched with cell growth and apoptosis. Mechanically, ISO treatment initiated the activation of caspase-3 cascade and elevated the expression of mitochondria-associated Bax/Bcl-2, cytosolic cytochrome c, followed by the activation of the cleavage of caspase-3 as well as poly ADP-ribose polymerase (PARP), resulting in the severe reduction of the mitochondrial potential and the accumulation of reactive oxygen species (ROS), while pre-treatment of the caspase-3 inhibitor could block the anti-tumor effect. Therefore, these results indicate that ISO treatment induces the apoptosis of GC cells through the mitochondria-dependent apoptotic pathway, providing a potential strategy for clinical GC therapy.


Introduction
Cancer is a primary cause of death worldwide. In 2020, statistics showed that there were about 19 million new cases and about 10 million deaths caused by cancer worldwide. Among different cancer types, gastric cancer (GC) is a fatal malignancy that causes over 1,000,000 new cases and about 770,000 deaths globally in 2020, exhibiting a high incidence (fifth) and mortality rate (fourth) [1]. Factors such as Helicobacter pylori infection, genetic alteration, diet, obesity, and tobacco use are the main causes of GC [2]. Although the mortality of GC has been reduced by treatment options such as surgery, chemotherapy, radiotherapy, targeted therapy, and immunotherapy, the treatment of GC still has serious problems, including drug resistance, poor prognosis, and a high recurrence rate [3]. Thus, novel strategies are needed to alleviate the progression and mortality of GC. Due to the advantages of multi-target, multi-link, and small side effects, natural medicine has recently become an emerging option for clinical anticancer drug investigation [4]. For example, natural product-derived doxorubicin and camptothecin have been widely used as an anticancer drug [5]. Nevertheless, GC is not sensitive to these compounds. Hence, new effective derivates from natural products still need to be developed and tested in GC therapy.
Flavonoids are abundant in daily intakes of food, such as fruits, vegetables, tea, and red wine. Recently, a variety of flavonoid agents have been applied to different human disease therapies, such as cardiovascular disease, breast cancer, and ovarian cancers [6][7][8]. Among these agents, isorhamnetin is a natural product isolated from various plants, including Hippophae rhamnoides L., Ginkgo biloba L., and Tamarix ramosissima, and a kind of immediate metabolite of quercetin in mammals [9,10]. Diverse studies have demonstrated that ISO exhibits remarkable effects on immunomodulatory, anti-inflammatory, and cardiovascular as well as cerebrovascular protection [11][12][13]. Recently, ISO has received attention due to its tumor-suppressing role in different human cancers, including colorectal, skin, lung, and breast cancers [14][15][16][17][18]. In these cancers, ISO exhibits comprehensive anti-tumor activities by repressing cell proliferation and migration and activating apoptosis [19][20][21]. Even though the cytostatic and pro-apoptotic effects of ISO have been extensively studied, the potential effects of ISO in GC therapy and the molecular mechanisms by which ISO induces apoptosis are still unknown.
Apoptosis plays critical functions in controlling cell homeostasis and proliferation. Hence, the aberrant activation of apoptosis has been observed in various diseases and contributes to numerous human cancers [22], emphasizing the activation of apoptosis in cancer therapy [23,24]. Ramachandran et al. reported that ISO could modulate PPAR-γ cascade activity and further inhibit proliferation and invasion, while the use of PPAR-γ inhibitor mildly blocked the effects of ISO [25]. In addition, Lee et al. previously uncovered that ISO activated the apoptosis of LLC cells in part through mitochondria-cytochrome C-caspase-9 cascade [26], and Hu et al. demonstrated that ISO has an important role in ROSmediating CaMKII/Drp1 signaling by regulating mitochondrial fission and apoptosis [12]. These studies have suggested that the anti-tumor function of ISO is mainly based on the apoptosis-dependent pathway. Our group previously found that ISO treatments could enhance the radiosensitivity and apoptosis of A549 cells through the NF-κB and IL-13 signaling pathways [15]. Collectively, ISO has exhibited pro-apoptotic effects in various tumor therapies.
Due to the profound pro-apoptotic effects of ISO, in this study, we further investigated the anti-tumor activities of ISO in GC cells and revealed the antineoplastic properties of ISO both in vitro and in vivo. In addition to the PPAR-γ activation induced by ISO treatment, we found that ISO could induce high cytotoxicity of GC cells and significantly inhibit cell proliferation and migration. We also revealed that the mechanisms underlying this effect were involved in mitochondria-dependent caspase-3 cascade. Overall, these findings implicate the potential of ISO in gastric cancer treatment and further reveal the primary pathways of ISO-induced apoptosis.

ISO Induces Apoptosis of GC Cells
In order to evaluate the potential effects of ISO on GC cells, we first tracked the morphology of ISO-treated GC cells. Tubulin-based analyses of cell morphology exhibited significant changes in the two types of GC cells (AGS-1 and HGC-27) during ISO treatment; rounded cells with progressive nuclear shrinkage and vesicle formation were observed ( Figure 1A). Severe disruption of the cell morphology indicated that the ISO treatment may have induced cell apoptosis. DNA fragmentation is the hallmark of apoptosis [21]; therefore, we conducted a TUNEL assay to evaluate the effects of ISO on this event. The results show that the ISO treatment increased the proportion of TUNEL-positive cells in a time-dependent manner ( Figure 1B,C). The TUNEL fluorescence intensity of AGS-1 cells ( Figure 1D) and HGC-27 cells ( Figure 1E) had increases of 300-500-fold and 300-800-fold, respectively. Furthermore, we performed annexin V/PI staining and flow cytometric analysis to detect the apoptosis levels of ISO-treated cells. The staining results show that the ISO treatment drastically increased the proportion of cells in the early apoptotic states at 24 h, while prolonged treatment (48 h) further increased the proportion of cells in the late apoptotic stage ( Figure 1F,G). The analysis of the apoptosis rate (assessed by the proportion of all apoptotic cells) further revealed that the ISO treatment induced apoptosis in a timedependent manner ( Figure 1H,I). Based on these results, we tested the anti-tumor effects of Furthermore, we performed annexin V/PI staining and flow cytometric analysis to detect the apoptosis levels of ISO-treated cells. The staining results show that the ISO treatment drastically increased the proportion of cells in the early apoptotic states at 24 h, while prolonged treatment (48 h) further increased the proportion of cells in the late apoptotic stage ( Figure 1F,G). The analysis of the apoptosis rate (assessed by the proportion of all apoptotic cells) further revealed that the ISO treatment induced apoptosis in a timedependent manner ( Figure 1H,I). Based on these results, we tested the anti-tumor effects of ISO in vivo. The results show that the ISO treatment drastically suppressed tumor growth in mice with AGS-1 tumor xenografts ( Figure 1J,K).

ISO Induces the Caspase-3 Cascade in GC Cells
It is well established that the cleavage of caspase-3 is the key regulatory event in the activation of apoptotic signaling [27]. Thus, we performed co-staining of caspase-3 and annexin V to detect caspase-3 activity and apoptosis levels. Consistently, caspase-3 and annexin V were both strongly activated after ISO treatment in AGS-1 and HGC-27 cells (Figure 2A,B). Cleaved caspase-3 functions as an executioner caspase and cleaves most of its substrates to induce apoptosis [28,29]. To further confirm the ISO-induced activation of caspase-3 in GC cells, we examined the cleavage level of caspase-3 and PARP. After

ISO Induces the Caspase-3 Cascade in GC Cells
It is well established that the cleavage of caspase-3 is the key regulatory event in the activation of apoptotic signaling [27]. Thus, we performed co-staining of caspase-3 and annexin V to detect caspase-3 activity and apoptosis levels. Consistently, caspase-3 and annexin V were both strongly activated after ISO treatment in AGS-1 and HGC-27 cells (Figure 2A,B). Cleaved caspase-3 functions as an executioner caspase and cleaves most of its substrates to induce apoptosis [28,29]. To further confirm the ISO-induced activation of caspase-3 in GC cells, we examined the cleavage level of caspase-3 and PARP. After ISO treatment at 20 μM for 6 h, higher levels of cleaved caspase-3 and cleaved PARP were observed at 48 h ( Figure 2C-H). We then questioned whether the ISO-induced apoptosis was caspase-3 activationdependent. Thus, we pre-incubated AGS-1 and HGC-27 cells with the caspase-3 inhibitor (Ac-DEVD-CHO) for 30 min, and then treated these cells with ISO for 72 h. The viability of AGS-1 and HGC-27 cells, as measured by the CCK-8 assay, decreased after treatment with ISO ( Figure 2I,J), while treatment with the 50 μM Ac-DEVD-CHO alone had little effect on the viability of the GC cells. However, pre-treatment with Ac-DEVD-CHO drastically abolished the ISO-induced suppression in cell viability in an ISO dose-dependent manner; further, 50 μM Ac-DEVD-CHO completely blocked ISO-induced cell apoptosis ( Figure 2I,J). Similarly, the anti-tumor effects of ISO were also blocked in vivo by Ac-DEVD-CHO ( Figure 1J,K), indicating that the ISO-induced apoptosis activation was caspase-3 dependent.
A previous study demonstrated that ISO modulated PPAR-γ cascade activity and that PPAR-γ inhibitor (GW9662) could partially reverse ISO-induced apoptosis in 12 h We then questioned whether the ISO-induced apoptosis was caspase-3 activationdependent. Thus, we pre-incubated AGS-1 and HGC-27 cells with the caspase-3 inhibitor (Ac-DEVD-CHO) for 30 min, and then treated these cells with ISO for 72 h. The viability of AGS-1 and HGC-27 cells, as measured by the CCK-8 assay, decreased after treatment with ISO ( Figure 2I,J), while treatment with the 50 µM Ac-DEVD-CHO alone had little effect on the viability of the GC cells. However, pre-treatment with Ac-DEVD-CHO drastically abolished the ISO-induced suppression in cell viability in an ISO dose-dependent manner; further, 50 µM Ac-DEVD-CHO completely blocked ISO-induced cell apoptosis ( Figure 2I,J). Similarly, the anti-tumor effects of ISO were also blocked in vivo by Ac-DEVD-CHO ( Figure 1J,K), indicating that the ISO-induced apoptosis activation was caspase-3 dependent.
A previous study demonstrated that ISO modulated PPAR-γ cascade activity and that PPAR-γ inhibitor (GW9662) could partially reverse ISO-induced apoptosis in 12 h [25]. Here, we also involved the treatment of GW9662 in both AGS-1 and HGC-27 cells and compared the block effect with Ac-DEVD-CHO. After 72 h of ISO treatment, we found that the pre-treatment of GW9662 rescued 20% of ISO-induced suppression, which was much weaker than that of the Ac-DEVD-CHO (~70%) ( Figure 2I,J). This result shows that the ISO-induced apoptosis activation primarily depended on caspase-3 activation.

ISO Induces the Mitochondrion-Dependent Apoptosis of GC Cells
The mitochondria membrane potential (MMP) has been identified as an important indicator of mitochondrial homeostasis, and mitochondrial depolarization is generally associated with apoptosis [30,31]. To determine whether ISO treatment induced mitochondriondependent apoptosis, we first detected the MMP of ISO-treated cells using MitoTracker Red CMXRos. The results show that the MMP gradually decreased with the extension of the ISO treatment duration ( Figure 3A,B). The fluorescence intensity of MitoTracker Red CMXRos in AGS-1 cells ( Figure 3C) and HGC-27 cells ( Figure 3D) was also examined. Similarly, we performed JC-1 staining to detect MMP loss in GC cells under ISO treatment. The mitochondrial depolarization analysis showed that ISO treatment markedly enhanced the mitochondrial depolarization in both AGS-1 ( Figure 3E) and HGC-27 cells ( Figure 3F). Moreover, we found that ISO treatment upregulated reactive oxygen species (ROS) production in AGS-1 ( Figure 3G) and HGC-27 ( Figure 3H) cells; this indicates that the ISO treatment impaired mitochondrial homeostasis. [25]. Here, we also involved the treatment of GW9662 in both AGS-1 and HGC-27 cells and compared the block effect with Ac-DEVD-CHO. After 72 h of ISO treatment, we found that the pre-treatment of GW9662 rescued 20% of ISO-induced suppression, which was much weaker than that of the Ac-DEVD-CHO (~70%) (Figure 2I,J). This result shows that the ISO-induced apoptosis activation primarily depended on caspase-3 activation.

ISO Induces the Mitochondrion-Dependent Apoptosis of GC Cells
The mitochondria membrane potential (MMP) has been identified as an important indicator of mitochondrial homeostasis, and mitochondrial depolarization is generally associated with apoptosis [30,31]. To determine whether ISO treatment induced mitochondrion-dependent apoptosis, we first detected the MMP of ISO-treated cells using Mito-Tracker Red CMXRos. The results show that the MMP gradually decreased with the extension of the ISO treatment duration ( Figure 3A,B). The fluorescence intensity of Mito-Tracker Red CMXRos in AGS-1 cells ( Figure 3C) and HGC-27 cells ( Figure 3D) was also examined. Similarly, we performed JC-1 staining to detect MMP loss in GC cells under ISO treatment. The mitochondrial depolarization analysis showed that ISO treatment markedly enhanced the mitochondrial depolarization in both AGS-1 ( Figure 3E) and HGC-27 cells ( Figure 3F). Moreover, we found that ISO treatment upregulated reactive oxygen species (ROS) production in AGS-1 ( Figure 3G) and HGC-27 ( Figure 3H) cells; this indicates that the ISO treatment impaired mitochondrial homeostasis. Bcl-2 family proteins, including anti-apoptotic protein Bcl-2 and the pro-apoptotic protein Bax, are major regulators of mitochondrial integrity and apoptosis [32]. Thus, we investigated if these proteins participated in mitochondrial permeability changes. The ISO treatment did not affect Bax or Bcl-2 expression (data not shown). Next, we isolated the mitochondria and examined the expression levels of these proteins. As shown in Figure 3I,J, Bax expression increased markedly during the ISO treatment, which concomitantly repressed the expression of mitochondrion-associated Bcl-2. Consequently, the ratio of Bax/Bcl-2 gradually increased with the ISO treatment in both the AGS-1 and HGC-27 cells ( Figure 3K). In drug-induced intrinsic apoptosis signaling, Bax activation contributes to cytochrome c release from the mitochondria and subsequent caspase activation [33]. As expected, with the increase in the duration of ISO administration, we also detected the gradual upregulation of cytochrome c in the total cell lysates ( Figure 3I,J,L), which was a likely consequence of the loss of MMP observed in Figure 3A-F. To further clarify that ISO-induced apoptosis relies on Bax activation, we used short hairpin RNA (shRNA) to knock down Bax in both AGS-1 and HGC-27 cells (Figure 3M), and found that Bax knockdown could also block cell ISO-induced apoptosis ( Figure 3N).
Collectively, these results demonstrate that ISO-induced apoptosis in GC cells is mitochondrion-dependent and mediated by caspase activation.

Transcriptome Analysis of ISO-Treated GC Cells
To determine the regulatory mechanism underlying the ISO-induced anti-tumor activity in GC cells, we performed RNA-seq of the total RNA from AGS-1 and HGC-27 cells that were treated with 20 µM ISO for 24 h or not. Correlation analysis of RNA-seq revealed that the duplicates were highly correlated ( Figure S1A), while PCA analysis showed that the ISO treatment resulted in differences in the transcriptomes ( Figure S1B). In the case of the consistent transcriptome levels ( Figure 4A), we identified 465 and 339 differentially expressed genes in the AGS-1 cells (ISO vs. Control) ( Figure 4B) and the HGC-27 (ISO vs. Control) cells ( Figure 4C), respectively. Heat map analysis was used to depict these differentially expressed genes ( Figure 4D and Figure S1C). Using gene ontology (GO) analysis, we further found that these differentially expressed genes were enriched in cell growth, apoptosis, migration, and cytoskeleton pathways ( Figure 4E,F). Meanwhile, these differentially expressed genes could be classified into three major categories: biological processes, cellular components, and molecular functions ( Figure S1D,E).  Taken together, these results indicate that ISO treatment regulates various processes associated with cell homeostasis, including cell growth, apoptosis, and other related biological processes. Taken together, these results indicate that ISO treatment regulates various processes associated with cell homeostasis, including cell growth, apoptosis, and other related biological processes.

ISO Inhibits GC Cell Proliferation
To further assess the ISO-mediated inhibition of human GC cell proliferation, we conducted CCK-8 and colony formation assays using the two types of GC cells. Upon treatment with increasing doses of ISO (0, 10, 20, 50, and 100 µM) and an increasing time course (24,48, and 72 h), AGS-1 and HGC-27 cells exhibited enhanced growth inhibition ( Figure 5A-D). Accordingly, pre-treatment with ISO repressed the colony-formation ability of AGS-1 and HGC-27 cells in a dose-dependent manner ( Figure 5E). The colony-formation ability of both AGS-1 and HGC-27 cells pre-treated with 20 µM ISO was inhibited by >50% after 14 days of culture ( Figure 5F). Collectively, these results demonstrate that the ISO treatment suppressed cell proliferation via apoptosis activation.

ISO Suppresses GC Cell Migration
To further evaluate the effect of ISO on the cell migration of GC cells, we performed Transwell and wound-healing analyses of these cells after ISO treatment (20 μM, for 24 h or 48 h). As shown in Figure 6A,B, the ISO treatment distinctly suppressed the migration ability of AGS-1 and HGC-27 cells. The efficiency of this suppression was further calculated by assessing the OD570, which indicated a suppression efficiency of >50%. Next, we performed a wound-healing assay via 24-h continuous real-time imaging and analyzed the relative wound gap area at 6, 12, and 24 h. Compared to the case at 0 h, the low-dosage (10 μM) ISO treatment showed a negligible effect on cell migration, whereas a higher dosage (20-100 μM) of ISO strongly inhibited cell migration ( Figure 6E-H). Collectively, these results demonstrate that the ISO treatment suppressed cell proliferation via apoptosis activation.

ISO Suppresses GC Cell Migration
To further evaluate the effect of ISO on the cell migration of GC cells, we performed Transwell and wound-healing analyses of these cells after ISO treatment (20 µM, for 24 h or 48 h). As shown in Figure 6A,B, the ISO treatment distinctly suppressed the migration ability of AGS-1 and HGC-27 cells. The efficiency of this suppression was further calculated by assessing the OD 570 , which indicated a suppression efficiency of >50%. Next, we performed a wound-healing assay via 24-h continuous real-time imaging and analyzed the relative wound gap area at 6, 12, and 24 h. Compared to the case at 0 h, the low-dosage (10 µM) ISO treatment showed a negligible effect on cell migration, whereas a higher dosage (20-100 µM) of ISO strongly inhibited cell migration ( Figure 6E-H). Thus, our results show that ISO treatment notably suppressed the migration of AGS-1 and HGC-27 cells in a dose-and time-dependent manner.

Discussion
Gastric cancer (GC) is a fatal malignant tumor worldwide. In spite of the advancement in the early detection and chemotherapy of GC, there has been little effect on increasing survival among GC patients. In recent decades, due to the promising efficacy and fewer side effects, derivatives of medicinal natural products exhibit great potential and have attracted extensive attention in cancer therapy. ISO (a kind of immediate 3'-O-methylated metabolite of quercetin) has recently been reported to possess remarkable anti-tumor activity against several human cancer types [19,34,35]. Specifically, ISO repressed TNF-α-induced inflammation and cell proliferation in BEAS-2B cells (human bronchial epithelial cell lines), via the MAPK and NF-κB pathways [36]. Further, our group previously discovered that ISO treatment could enhance the radiosensitivity of lung cancer Thus, our results show that ISO treatment notably suppressed the migration of AGS-1 and HGC-27 cells in a dose-and time-dependent manner.

Discussion
Gastric cancer (GC) is a fatal malignant tumor worldwide. In spite of the advancement in the early detection and chemotherapy of GC, there has been little effect on increasing survival among GC patients. In recent decades, due to the promising efficacy and fewer side effects, derivatives of medicinal natural products exhibit great potential and have attracted extensive attention in cancer therapy. ISO (a kind of immediate 3'-O-methylated metabolite of quercetin) has recently been reported to possess remarkable anti-tumor activity against several human cancer types [19,34,35]. Specifically, ISO repressed TNF-α-induced inflam-mation and cell proliferation in BEAS-2B cells (human bronchial epithelial cell lines), via the MAPK and NF-κB pathways [36]. Further, our group previously discovered that ISO treatment could enhance the radiosensitivity of lung cancer cells through the IL-13 and NF-κB signaling pathways [14]. However, the anti-tumor effects of ISO on GC cells and how ISO suppresses cell proliferation remain poorly understood.
Cancer cells exhibit common characteristics of infinite proliferation. To this end, we conducted CCK-8 assays to explore the anti-cancer activity of ISO in the two types of GC cells. Interestingly, the ISO was selectively cytotoxic to both types of GC cells in a time-and concentration-dependent manner but exerted no toxic effects on normal human epithelial GES-1 cells under the same conditions. A decrease in the colony formation of GC cells (under ISO treatment) and obvious morphological changes in these cells (condensed nuclear chromatin, vesicle formation, and increased numbers of apoptotic bodies) further confirm the potential tumor-suppressor ability of ISO against GC, which is consistent with the effects of ISO observed in vivo. In cancer therapy, therapeutic efficacy and prognostic survival are mainly relied on in tumor cell movement. In advancing the impact of ISO on cell proliferation inhibition, we also observed that the ISO treatment dramatically mediated the suppression of GC cells' migration ability in this study, which occurred in an ISO dose-dependent manner.
The RNA-seq data from ISO-treated GC cells yielded genes with differential expressions that were mainly related to cell growth, migration, cytoskeleton, and the mitochondrial membrane, whereas the GO and KEGG analyses of these genes revealed that they were enriched in signaling pathways associated with apoptosis. In addition to apoptosis, the GO analysis also showed that ISO treatment plays an important role in several signaling cascades, including Hippo signaling, tight junctions, MAPK signaling, and ion hemostasis. Nevertheless, the regulatory network for ISO appears to be complicated, and multiple surveillance mechanisms are required to clarify the regulatory effects of ISO on various signaling pathways.
Apoptosis is extremely important in the progression of benign tumors into malignant neoplasms; thus, it is considered a potential target for the treatment of different cancers [37][38][39]. Nucleosome-sized fragments, membrane blebs, and condensed cellular compartments are involved in this complex, sequential process. As expected, ISO-treated GC cells revealed many TUNEL-positive cells, annexin V-FITC, and PI co-staining, further substantiating that the cytotoxicity of the ISO treatment was mediated by the induction of apoptosis. Meanwhile, we observed decreased MMP and JC-1 staining intensities in ISO-treated GC cells; this indicates that the ISO treatment leads to severe mitochondrial dysfunction and suggests that ISO facilitates the apoptosis destiny of GC cells.
The imbalance between the antiapoptotic and proapoptotic Bcl-2 family members facilitates the cytochrome c release from mitochondria and subsequent activation of caspase-9 and -3 and further apoptosis. To further elucidate the role of ISO in inducing apoptosis, the levels of apoptotic signaling proteins (e.g., Bax, Bcl-2, caspase-3) were measured under ISO treatment in AGS-1 and HGC-27 cells. The immunoblotting demonstrated that ISO increased the expression of the mitochondrion-associated Bax, cytoplasmic cytochrome c, cleaved caspase-3, and cleaved PARP proteins, while the expression of mitochondriaassociated Bcl-2 protein decreased. These striking changes in the expression levels of these proteins indicate that ISO might contribute to apoptosis activation, thereby giving rise to the anti-tumor effects. A previous study revealed that ISO modulated PPAR-γ cascade activity and induced apoptosis [25]. Here, we compared the effects of PPAR-γ inhibitors and caspase-3 inhibitors in ISO-induced GC apoptosis; the results show that caspase-3 inhibitor pre-treatment completely blocked long-term ISO treatment-induced apoptosis, indicating that ISO treatment initiates a mitochondrion-dependent intrinsic apoptotic pathway. However, apoptosis activation is regulated via several important pathways and mitochondrion-dependent apoptosis is composed of various steps. The precise role of ISO in mitochondrion-related apoptotic activation remains to be determined. In this regard, which specific signaling pathway in ISO-mediated apoptosis activation is necessary should be studied in the future.
In conclusion, this study shows that ISO treatment in GC cells initiated the activation of caspase-3 cascade, the upregulation of cytochrome c, Bax/Bcl-2, and cytosolic cytochrome c, and the cleavage of caspase-3 as well as PARP, resulting in mitochondrial homeostasis imbalance and apoptosis. This effect of ISO is primarily mitochondrial-dependent and contributes to the selective anti-tumor activity in the suppression of the migration and invasion of human GC cells. Collectively, ISO targeting mitochondrion-dependent apoptosis indicates the potential of using ISO as a therapeutic agent for gastric cancer treatment.

Subcutaneous Xenograft Tumor Models
Male nude recipient mice, 4-6-weeks old, were purchased from the SLAC Laboratory Animal Company and raised under pathogen-free conditions. Cells for xenograft (5 × 10 5 ) were implanted subcutaneously within a Matrigel matrix (total 0.1 mL) into the mice's right flanks. At 5 days after the xenograft, ISO (5 mg/kg) treatment was conducted through intraperitoneal injection every 3 days. For apoptosis inhibition, caspase-3 inhibitor Ac-DEVD-CHO (3 mg/kg) was administered every day through intraperitoneal injection after ISO treatment until sacrifice. A vernier caliper was used to measure the tumor volume at the indicated time and the tumor was allowed to grow to reach an average size of 150 mm 3 .

shRNA Plasmid Construction, Lentivirus Production, and Cell Infection
For shRNA constructs, Bax target sequences and a scrambled sequence were individually cloned into a pLKO.1-TRC vector (AgeI and EcoRI sites).
For lentiviral particle production, 5 µg of shRNA construct, 3.75 µg of psPAX2, and 1.5 µg of pMD2.G were co-transfected into HEK293FT cells (70-80% confluence) in a 6 cm dish. The supernatant containing lentivirus was harvested twice at 48 and 72 h after transfection and further filtered through a 0.45 µm filter before use. To infect the GC cells with lentivirus, they were cultured in a medium containing a lentivirus supplement with 1 mg/mL polybrene (Beyotime, C0351 Sigma).

Cell Morphology Analysis
AGS-1 and HGC-27 cells were seeded on 35-mm glass bottom dishes and incubated overnight. After ISO treatment, the cells were fixed with 4% formaldehyde (15 min) and permeabilized with 0.05% Triton X-100 (5 min). The cells were then blocked with 5% BSA (30 min) and incubated with anti-tubulin (1:2000) for 1 h at room temperature. After washing 3 times with PBS, the cells were incubated with fluorescent secondary antibody (1:1000) for another hour. The nuclei were stained with DAPI (5 min). Fluorescent images were collected using an Olympus FV3000 microscope and processed by ImageJ.

Colony Formation Assay
A colony formation assay was performed as described with a few modifications [40]. Cells were seeded on a 60-mm dish (1000 cells per group). After 24 h of incubation, ISO (0, 20, 50 µM) was used to treat the cells for another 24 h, then they were rinsed with PBS, and replaced with fresh medium. After 14 days of incubation, the colonies were fixed in 4% formaldehyde (15 min) and stained with 0.5% crystal violet (30 min), followed by manually counting.

TdT-Mediated dUTP Nick-End Labeling (TUNEL)
AGS-1 and HGC-27 cells were seeded on 35-mm glass bottom dishes and incubated for 24 h. The cells were then fixed in 4% formaldehyde (30 min) and permeabilization with 0.05% Triton X-100 (5 min). After rinsing, the cells were incubated with a One-Step TUNEL Apoptosis Assay Kit (Beyotime, C1086). Images were captured by an Olympus FV3000 microscope and analyzed using Fiji ImageJ.

Apoptosis Assays
After treatment with 20 µM ISO for 24 or 48 h, the cells were digested with trypsin. The cells were then washed with PBS and detected with an Annexin V-FITC Apoptosis Detection Kit (Beyotime, C1062L), followed by measurement with a BD FACS Celesta TM (BD Biosciences, Franklin Lakes, NJ, USA). Caspase-3 and annexin V staining was conducted with (Beyotime, C1077M) according to the protocol. Images were captured by an Olympus FV3000 microscope and further analyzed by Fiji ImageJ.

Mitochondrial Staining of Cells
After treatment with 20 µM ISO, the cells were stained with Mito-Tracker Red CMXRos working fluid (37 • C, 30 min), followed by staining with Hoechst for 10 min. After staining, the cells were washed with PBS and pre-warmed. FluoroBrite DMEM (GIBCO) supplemented with 10% FBS was used for the lice-cell imaging. Images were captured by an Olympus FV3000 confocal microscope.

Mitochondrial Membrane Potential Measurement (MMP)
The MMP was detected as described with a few modifications [41]. After treatment with 20 µM ISO for 24 h, the MMP of the cells was detected using a JC-1 detection kit (BD Biosciences, 551302). The cells were incubated with JC-1 working solution (37 • C, 20 min) and further washed with 1× assay buffer twice, followed by flow cytometry analysis. The statistics of the MMP were measured by the ratio of JC-1 monomers to JC-1 aggregates.

Reactive Oxygen Species (ROS) Assay
After treatment with 20 µM ISO for 24 or 48 h, the cells were digested with trypsin. Then, the cells were stained with DCFH-DA (Beyotime, S0033M) at 37 • C for 20 min, followed by washing with PBS 3 times and flow cytometry analysis.

Cell Mitochondrial Isolation
The mitochondria fraction was collected using a Cell Mitochondria Isolation Kit (Beyotime, C3601) with a few modifications. After treatment with ISO, 5 × 10 7 cells were harvested with trypsin and washed with PBS. The cells were then homogenized, and the lysates were centrifuged at 1000× g (5 min) for the removal of unbroken cells. The supernatant was centrifuged at 12,000× g (10 min), and the mitochondria fractions were obtained in the pellets. The pellets were further lysed with 1 × sodium dodecyl sulfate (SDS) and analyzed using Western blotting.

Cell Migration Assay
Cell migration assays were performed using Transwell Permeable Supports (Costar Corning, NY, USA) with a pore size of 8.0 um, according to the Boyden chamber method. AGS-1 and HGC-27 cells seeded on the upper chamber supplemented with RPMI-1640 (serum-free) were pre-treated with 20 µM ISO. The lower chamber was supplemented with normal RPMI-1640 medium. After 24 h of incubation, the cells that migrated into the lower chamber were stained with 0.5% crystal violet (30 min) and observed using an Olympus IX73. The cells were further eluted by 50% ethanol and 0.1% acetic acid and detected by BioTek Synergy NEO at 550 nm.

Wound-Healing Assay
A wound-healing assay was performed as described with a few modifications [42]. AGS-1 and HGC-27 cells were seeded in a 96-well plate overnight before ISO treatment. A sterile pipette tip was used to generate the wounded cell line, and the floating cells were removed using a PBS wash. The wounded cells were cultured for up to 24 h and images were captured by EnSight™ Multimode Microplate Reader (PerkinElmer, Waltham, MA, USA).

Preparation of RNA-seq Libraries and Sequencing
RNA samples were extracted using a TRIzol and QIAGEN RNeasy Mini Kit. The cDNA libraries were prepared by the VAHTSTM mRNA-seq V3 Library Prep Kit (Vazyme, Nanjing, China, NR611-01) and subjected to an Illumina NextSeq sequencer.

Western Blotting
After the indicated ISO treatment, AGS-1 and HGC-27 cells were lysed in 1 × SDS. The protein samples were resolved in 10% SDS-PAGE gel and transferred to PVDF membranes. After blocking with 5% non-fat dry milk, each protein on separate membranes was detected by incubation with the indicated primary antibodies at 4 • C overnight, followed by the corresponding HRP-conjugated secondary antibodies, and then detected using an ECL detection kit and Tanon 5200.

Statistical Analysis
The data are presented as the means ± S.D. in this study. Statistical analyses (two-tailed Student's t test, Mann-Whitney Test) were calculated using GraphPad Prism 7. For correlation, the Spearman rank correlation was used. p < 0.05 is considered statistically significant.