A GSE97766 dataset comprising data of genistein-resistant and parental PC cells was downloaded from the Gene Expression Omnibus (GEO) database. The data were adjusted and analyzed using the Limma R Package. Differentially expressed genes between drug-resistant and parental PC cells were identified using ∣Fold Change∣ ≥ 2 and p- value < 0.05 as the thresholds. A gene set variant analysis (GSVA) was conducted based on the differentially expressed genes to identify the aberrantly activated signaling pathways. The GSVA was performed with the 50 gene sets from the h.all.v7.0.symbols.gmt of the Molecular Signatures Database (MSigDB) as the background. The activity of the signaling pathways in each sample was scored. Potential target genes of genistein were predicted on the Swiss Target Prediction system.¹ The transcription factors binding to the promoter of DEPTOR were predicted on JASPAR.²

Human PC cell lines Panc-1 and PaCa were procured from Shanghai Institute of Biochemistry and Cell Biology, Chinese Academy of Sciences (Shanghai, China). Cells were cultured in Dulbecco’s Modified Eagle Medium containing 10% fetal bovine serum and 1% penicillin/streptomycin (Thermo Fisher Scientific, Inc., Waltham, MA, United States) at 37°C in humidified air enriched with 5% CO₂. Exponentially growing cells were collected for the subsequent experiments.

Overexpression vectors (oe) or short hairpin (sh) RNAs targeting DEPTOR and ELK1 were procured from GenePharma Co., Ltd (Shanghai, China). The recombinant vector and shRNA were used as negative control. All vectors and shRNAs were transfected into well growing parental or drug-resistant Panc-1 and PaCa cells using Lipofectamine 3000 Reagents (Invitrogen; Thermo Fisher Scientific, Carlsbad, CA, United States) according to the kit’s instructions.

All cells were maintained in a 37°C incubator containing humidified air enriched with 5% CO₂. Panc-1 and PaCa were continuously exposed to genistein (Sigma-Aldrich, St. Louis, MO, United States) to construct genistein-resistant cells. In brief, cells were exposed to different doses of genistein. Initially, the cells were exposed to 5 μM genistein for 48 h, once a week. When cells entered an exponential growth phase, the concentration of genistein was gradually increased till 100 μM. After 6 months of genistein treatment, the drug-resistant cells were sub-cultured and treated with 20 μM genistein every month to maintain the resistance. In the subsequent experiments, the genistein-resistant Panc-1 and PaCa cells were placed in a genistein-free condition for 14 days before use.

Cells were cultured in 96-well microtiter plates at 5 × 10³ cells/well for the measurement of half maximal inhibitory concentration (IC50). After 48 h, the Panc-1 and PaCa cells were exposed to different doses of genistein (5, 10, 20, 40, 60, and 100 μM) for another 24 h. Then, 10 μl CTG solution was added into each well (Promega Corp., Madison, WI, United States) for 2 h of incubation in the dark. Then, the optical density (OD) of each well at 450 nm was examined by a microplate reader, and the genistein-induced reduction rate in cell viability was determined.

Activity of caspase-3/7 in cells was detected utilizing a Caspase-Glo^® 3 Assay Kit (Promega) according to the kit’s protocols.

Parental or drug-resistant cells were cultured in 96-well plates at 5 × 10³ cells/well for 48 h. Then, the cells were centrifuged at 350 g at 4°C for 5 min with the supernatant discarded. Thereafter, each well was added with 150 μl lactate dehydrogenase (LDH) release reagent (Cat. No. C0016; Beyotime Biotechnology, Shanghai, China) for 1 h of incubation at 37°C. The OD value at 490 nm was examined by the microplate reader to evaluate the level of LDH release in cells.

Annexin V and DNA staining was used to measure the apoptosis rate of cells during the process of genistein treatment (20 μM for continuous 48 h). The Annexin V was stained by fluorescein isothiocyanate, while the nucleic acid was stained by propidium iodide. The fluorescence was examined by a LSRFortessa flow cytometer (BD Biosciences, San Jose, CA, United States) and analyzed using the FlowJo software (Tree Star, Ashland, OR, United States).

Cells were cultured in a condition with or without ELM treatment for 48 h and exposed to 20 μM genistein. After genistein treatment, the cells were immediately seeded in fresh medium at 200 cells per dish (60 mm) for 15 days. After that, the cells were fixed in formaldehyde and labeled with 0.1% crystal violet. The number of cell colonies (over 50 cells) was counted under a stereomicroscope (Olympus Optical, Tokyo, Japan).

Once reaching a 60% cell confluence, the cells were fixed with 4% paraformaldehyde and in cold 100% methanol. Then, the cells were incubated with the mouse monoclonal antibody anti-mTOR (phospho S2448; 1:1,000, ab109268; Abcam Inc., Cambridge, MA, United States) at room temperature, and then with rabbit monoclonal Alexa Fluor 488 (1:750; Santa Cruz Biotechnology, Inc., Santa Cruz, CA, United States) for 60 min. The nuclei were stained with 4 ′, 6-diamidino-2-phenylindole.

Total RNA from the transfected cells was extracted using the TRIzol reagent^® (Ambion, Thermo Fisher Scientific). The extracted mRNA was quantified using a ND-2000 spectrophotometer (Thermo Fisher Scientific). Then, 1 μg of RNA sample was reversely transcribed to cDNA utilizing a ReverTra Ace qPCR RT Kit (Toyobo Life Science, Osaka, Japan). Next, real-time qPCR was conducted using a SYBR^® RT-PCR kit on the an ABI7500 fast real-time PCR system (Applied Biosystems; Thermo Fisher Scientific). The primers used were acquired from Thermo Fisher Scientific. Gene expression was quantified using the 2^–ΔΔCt method with GAPDH serving as the internal loading control.

Cells were lysed in RIPA buffer containing fresh protease inhibitor (Beyotime). The protein concentration was detected using a bicinchoninic acid kit (Beyotime). Next, an equal amount of protein sample was run on 10% SDS-PAGE and transferred to PVDF membranes (Thermo Fisher Scientific). After being blocked by 5% non-fat milk for 1 h, the membranes were cultured with the antibodies against DEPTOR (1:1,000, ab171944, Abcam), ELK1 (1:1,000, ab32042, Abcam), anti-mTOR (phospho S2448; 1:1,000, ab109268, Abcam) and GAPDH (1:5,000, #5174; CST, Beverly, MA, United States) at 4°C overnight. Then, the membranes were further co-cultured with HRP-labeled goat anti-rabbit (1:5,000, ab205718, Abcam) at 37°C for 1 h. The blots were visualized utilizing an enhanced chemiluminescence kit (WP20005, Invitrogen). Relative protein expression was assessed by Image-Pro Plus 6.0.

The animal study was performed based on a previous report (Ma et al., 2019). NOD/SCID nude mice (4 weeks old, 20 ± 2 g) were acquired from the SLAC Laboratory Animal Co., Ltd. (Shanghai, China). To explore the function of genistein on the growth of xenograft tumors in vivo, a 100 μl suspension of parental and drug-resistant cells (1 × 10⁶) with stable transfection was implanted into mice through subcutaneous injection (n = 5 in each group). From the 12th day after cell implantation, the mice were intraperitoneally injected with genistein once every 3 days for a total of three times (3 mg/kg each time). ELM was administrated in to mice at a 3-day interval as well through intragastric administration, while the control mice were given an equal volume of normal saline. Then, the volume (V) of tumors was determined routinely as follows: V = 0.5 × L × W², where “L” indicates the length while “W” indicates the width of tumor. On the 35th day after cell implantation, the mice were euthanized by 150 mg/kg of pentobarbital sodium (intraperitoneal injection). Then, the tumors were taken out and weighed, and the tumor tissues were fixed and embedded in paraffin for the following histological staining. All animal studies were ratified by the Ethical Committee on Animal Research of Zhongshan Hospital. Great attempts were made to reduce the pain of animals. During the animal procedures, if the animals showed conditions such as a significant decline in body weight (>10%), piloerection, catalepsy, dyspnea, sialorrhea, tremor, spasm, or falling a dead sleep were, an appropriate amount of anesthetic or analgesic was given, such as 0.02 mg/kg of atropine or 30 mg/kg of ketamine. If the decline in body weight exceeded by over 25% or the spasm lasted for over 10 min, or the mice slept for over 1 h, the animals were euthanized using 150 mg/kg of pentobarbital sodium. After 30 min, the animal death was confirmed by the loss of neural reflex and link reflex, unresponsiveness to stimuli, and the loss of heartbeat rather than simple cardiac arrest.

The tissues were cut into sections, dewaxed, and rehydrated, and terminal deoxynucleotidyl transferase dUTP nick end labeling (TUNEL) assay was performed to examine the cell apoptosis rate in the tissues. The TUNEL assay was conducted as previously reported (Kyrylkova et al., 2012).

Immunohistochemical (IHC) staining was carried out according to a previous study (Ma et al., 2019). In brief, the paraffin-embedded sections (4 μm) were dewaxed and rehydrated. After antigen retrieval in citrate buffer, the sections were blocked in 100% normal goat serum and then incubated with anti-Ki67 (ab92742, 1:500, Abcam) at 4°C overnight, and then with HRP-conjugated goat anti-rabbit (1:1, Cat No. KIT-5905; Maixin Biotechnology, Fujian, China) at 37°C for 2 h. After that, the sections were counter-stained with hematoxylin and observed under the microscope.

The wild-type (WT) sequence containing the putative binding site with the DEPTOR promoter was obtained and amplified from the human genome DNA using PCR. The sequence was inserted into the downstream of the pGL3-promoter luciferase vector to construct recombinant pGL3-enhancer luciferase reporter vector. Then, the recombinant vector was co-transfected with oe-ELK3 into 293T cells using the Lipofectamine TM 2000 kit (Life Technologies, Carlsbad, CA, United States). After 24 h, the luciferase activity in cells was evaluated using a dual-luciferase-reporter-gene detection system (Promega).

A chromatin Immunoprecipitation (ChIP) assay kit (Cat No. #53008; Active Motif, Carlsbad, CA, United States) was used according to the kit’s instructions. The cells were cross-linked in 1% formaldehyde at 20°C for 10 min and neutralized by glycine for 5 min. Next, the cells were washed in PBS and resuspended in SDS, followed by ultrasonic treatment. After further centrifugation, the supernatant was collected and diluted in immunoprecipitation dilution buffer. Anti-ELK1 (CST) and the control anti-IgG (Cat No. #2729S, CST) was applied for immunoprecipitation reaction. Then, the precipitates were incubated with cytarabine for 1 h. Next, the precipitates were washed and de-crosslinked, and the DNA was purified for qPCR.

Graphpad prism software (version 8; GraphPad Software, San Diego, CA, United States) was applied for data analysis. Differences between two groups were compared using the Student’s t- test, while those among multiple groups were analyzed by one-way or two-way analysis of variance (ANOVA) followed by Tukey’s multiple comparison. Data were collected from three independent experiments and presented as the mean ± standard deviation (SD). p < 0.05 represents significant difference.

To explore the molecules responsible for genistein-resistance in PC cells, we first obtained a GEO GSE97766 dataset comprising data from genistein-resistant and parental PC cells. The Limma R Package analyzed that DEPTOR was significantly downregulated in the drug-resistant cells (Figure 1A). Next, a GSVA was further performed based on the identified genes, and the activity of the PI3K/AKT-mTOR signaling pathway was found to be increased significantly (Figure 1B). According to the data in The Cancer Genome Atlas (TCGA)-Genotype-Tissue Expression (GTEx) Database, DEPTOR was predicted to be highly expressed in tumor samples (Figure 1C). Next, we explored the candidate targeting genes of genistein using the Swiss Target Prediction system. It was predicted that genistein can directly suppress EGFR (Figures 1D,E; Supplementary Table 1). However, EGFR has been reported to be negatively correlated with DEPTOR expression and activate the mTOR signaling pathway (Zhou et al., 2016). We therefore speculated that the EGFR was no longer suppressed by genistein in genistein-resistant cells, which blocks DEPTOR expression, therefore activating the PI3K/AKT/mTOR axis and increasing the drug resistance of PC cells.

To validate this, the parental Panc-1 and PaCa cells were treated with an ascending doses of genistein to induce drug-resistant cells. After 6 months, the IC50 of genistein in cells was examined using a CTG kit. It was found that the genistein-resistant cells were successfully established since the drug resistance of cells was significantly increased following serial genistein treatment (Figure 2A). Next, the mRNA expression of DEPTOR was obviously decreased in the genistein-resistant cell lines (Figure 2B), and the mTOR activity, correspondingly, was increased (Figure 2C).

To validate the role of DEPTOR in the sensitivity of PC cells to genistein, downregulation of DEPTOR was introduced in parental cells through administration of sh-DEPTOR, and upregulation of DEPTOR was induced in genistein-resistant cell lines through administration of oe-DEPTOR. The transfection efficacy was validated by reverse transcription quantitative polymerase chain reaction (RT-qPCR) (Figure 2D). Thereafter, the IC50 of genistein in cells was examined. Downregulation of DEPTOR in parental Panc-1 and PaCa cells decreased the sensitivity of cells to genistein (Figure 2E), whereas upregulation of DEPTOR in genistein-resistant Panc-1 and PaCa cells reduced the IC50 of genistein (Figure 2F). The IC50 of genistein in parental PC cells was approximately 20 μM. Therefore, cells were treated with 20 μM genistein in subsequent experiments to avoid a too drastic cytotoxicity on parental cells. After that, the activity of caspase-3/7 in cells was measured. It was found that the caspase-3/7 activity in parental cells was significantly decreased upon DEPTOR knockdown, while overexpression of DEPTOR increased the caspase-3/7 activity in the genistein-resistant cells (Figure 2G). In addition, the cytotoxicity of genistein in cells was further evaluated using an LDH kit. Likewise, the LDH release in parental cells was decreased after DEPTOR silencing, but it was increased in genistein-resistant cells upon DEPTOR upregulation (Figure 2H). Moreover, the viability of cells after genistein treatment was detected using a colony formation assay. It was found that the number of cell colonies formed by parental cells was increased after DEPTOR inhibition. Again, the number of colonies formed by genistein-resistant cells was decreased after DEPTOR overexpression (Figure 2I). The phosphorylation of mTOR in cells was further detected by immunofluorescence staining. As expected, DEPTOR negatively regulated the mTOR phosphorylation in cells according to the Western blot assays (Figure 2J).

Following the findings above that mTOR activation was closely correlated with the genistein resistance in PC cells, a commercial mTOR-specific inhibitor ELM was used to treat the cells. After ELM administration, the IC50 of genistein in both parental and drug-resistant cells was significantly decreased (Figure 3A). Next, ELM was co-administrated with genistein in parental and genistein-resistant cells, after which the caspase-3 activity (Figure 3B), the LDH release (Figure 3C) and the apoptosis rate (Figure 3D) in cells were significantly increased. Moreover, the colony formation assay indicated that administration of ELM strengthened the suppressive effect of genistein on the colony formation ability of PC cells (Figure 3E).

To further evaluate the relevance of ELM to the sensitivity of genistein on the growth of PC cells, the parental and genistein-resistant PaCa and Panc-1 cells were implanted into NOD/SCID mice for in vivo experiments. The mice were given genistein through intraperitoneal injection and given ELM through intragastric administration. The volume of xenograft tumor in mice was detected. The growth rate of xenograft tumor in mice was decreased after ELM administration (Figures 4A,B). In addition, the tumor tissues were collected for histological staining. The IHC staining results showed that the staining intensity of KI67 and PCNA in the tumor tissues was significantly decreased after ELM treatment (Figures 4C,D). In addition, the TUNEL assay suggested that the cytotoxicity of genistein to PC cells was strengthened by ELM (Figure 4E).

The findings of bioinformatics analyses above suggested that genistein might target EGFR (Figures 1D,E; Supplementary Table 1). As mentioned above, we speculated that the inhibitory effect of genistein on EGFR might be blocked in drug-resistant cells. To validate this, we determined the expression of downstream transcription factors of this pathway in the parental and drug-resistant cells. The expression of c-Fos, c-Myc, and ELK1 in genistein-resistant cells was significantly elevated (Figures 5A,B). Subsequently, the data on the Human Protein Atlas website³ suggested that the staining intensity of c-Fos, c-Myc, and ELK1 in PC tissues was higher than that in normal pancreatic tissues (Figure 5C). To determine possible regulators of DEPTOR, we searched the JASPAR system to explore if there are binding relationships between these candidate transcription factors and DEPTOR. Consequently, ELK1 was predicted as the only one that owns putative binding sites with the DEPTOR promoter region (Figures 5D,E). In addition, we determined the correlation between c-Fos, c-Myc and ELK1, and DEPTOR in The Cancer Genome Atlas-Pancreatic adenocarcinoma (TCGA-PAAD) database using the GEPIA website.⁴ Also, ELK1 showed a high and negative correlation with DEPTOR expression (Figure 5F). To further validate the binding relationship between ELK1 and DEPTOR, a ChIP-qPCR assay was performed. An abundance of DEPTOR fragments was found in the precipitates pulled down by anti-ELK1 compared to anti-IgG (Figure 5G). Then, a luciferase assay was conducted where pGL3-enhancer luciferase reporter vectors containing the promoter sequence of DEPTOR were co-transfected with oe-ELK1 into 293T cells. The luciferase activity in cells was decreased with the increase of ELK1 concentration (Figure 5H). Further, Panc-1 and PaCa cells were transfected with oe-ELK1 as well, after which the expression of DEPTOR was decreased accordingly (Figure 5I). These findings suggested that ELK1 suppresses DEPTOR transcription. ELK1 is downstream the MAPK kinase pathway. Therefore, to examine whether the MAPK pathway is involved in the genistein resistance in PC cells, a MEK inhibitor trametinib (TMT) was administrated into the PC cells. After MEK inhibition, it was found that the sensitivity of both parental or drug-resistant cells to genistein was enhanced (Supplementary Figure 1).

Next, we further verified the relevance of ELK1 expression to the genistein resistance in cells. Knockdown of ELK1 was introduced in parental PC cells pre-transfected with sh-DEPTOR, and upregulation of ELK1 was administrated in drug-resistant cells pre-transfected with oe-DEPTOR. Then, it was observed that the abundance of DEPTOR in parental cells was increased and the mTOR phosphorylation was decreased after ELK1 knockdown. By contrast, further overexpression of ELK1 reduced DEPTOR expression and enhanced phosphorylation of mTOR in drug-resistant cells (Figures 6A,B). Accordingly, the IC50 of genistein in parental cell lines was enhanced after ELK1 downregulation, whereas that in genistein-resistant cells was significantly enhanced upon ELK1 overexpression (Figure 6C). The caspase-3 activity and LDH release in parental cells were increased after ELK1 overexpression, but inverse trends were observed in the drug-resistant cells overexpressing ELK1 (Figures 6D,E). Likewise, the flow cytometry results suggested that the number of apoptotic cells induced by genistein was increased by sh-ELK1 in parental cells but reduced by oe-ELK1 in drug-resistant cells (Figure 6F). The number of cell colonies, according to the colony formation assay, was reduced after ELK1 downregulation but increased after ELK1 overexpression (Figure 6G).

Since ELM was found to suppress the mTOR activity in cells, ELK1 was found to trigger mTOR phosphorylation through transcriptionally suppressing DEPTOR. To validate the molecular mechanism, the parental and drug-resistant Panc-1 and PaCa cells co-treated with ELM and genistein were further administrated with ELK1 overexpression vector. The transfection efficacy was validated by RT-qPCR again (Figure 7A), and the mTOR phosphorylation in cells was increased (Figure 7B). After ELK1 overexpression, the sensitivity of cells to genistein increased by ELM was reduced (Figure 7C). Under this condition, it was found that the cell apoptosis, caspase-3 activity, and LDH release in cells were reduced (Figures 7D–F). In addition, the colony formation ability suppressed by ELM was recovered upon ELK1 knockdown (Figure 7G). Collectively, these results indicated that downregulation of ELK1 blocks the promoting role of ELM in the genistein-sensitivity in PC cells.