All cell lines used in this study were obtained from Type Culture Collection of the Chinese Academy of Sciences (Beijing, China). Flag, DPYSL3, MEK, ERK, pERK and GAPDH primary antibodies were purchased from SAB Biotech (College Park, MD, USA). Deferoxamin and 5-FU were from Sigma-Aldrich (Billerica, MA, USA). DPYSL3 siRNA (5’-GGAUAAUACUCUACCACCAT-3’) was synthesized from Genepharma, Shanghai, China. DPYSL3 cDNA was cloned into pIRES2-EGFP-SF plasmid (Vigene Biosciences, Jinan, China). The vectors of truncating DPYSL3 were prepared based on wild type DPYSL3 vector using mutagenesis PCR. The following PCR primer sequences were listed in Suppl. Table 1. The stably knock-down or knock-in DPYSL3 COLO-684, HEC-108, Ishikawa and HEC-59 cell lines were prepared via plasmid transfection using Lipofectamine 3000 (Thermo Scientific Fisher, Waltham, MA, USA), drug screening with 400 μg/mL G418 (Sigma-Aldrich) and maintained in DMEM (Gibco, Carlsbad, CA, USA) containing 10% (v/v) FBS (Gibco) supplemented with 250 μg/mL of G418.

The ChIP assay was performed as previously described (Peng et al., 2020). Briefly, 5 × 10⁶ cells were fixed with 1% formaldehyde, quenched with 0.125 M glycine at room temperature and then lysed in 500 μL of lysis buffer (10 mM Tris-HCl (pH 8.0), 10 mM NaCl, and 0.2% IGEPAL CA-630 (Thermo Fisher Scientific)) on ice for 30 min. The genomic DNA was sonicated into 200–500 bp. Ten percent of each whole-cell lysate was stored as input, and the rest of the lysate was incubated with 1 μg appropriate antibodies at 4°C overnight. DNA fragments bound with protein were pull down by protein-A beads (Thermo Fisher Scientific) at 4°C for 2 h. Primers were designed to encompass about 100 bp up- and downstream of potential p53 DNA binding sites in the DPYSL3 gene. Their sequences were listed in Suppl. Table 1.

In brief, six genomic DNA fragments containing p53 binding sites of DPYSL3 (CBS1-4) were cloned into pGL3-promoter-reporter (Promega), respectively, and mixed with pcDNA3-wtp53 or mtp53 (R248G) as well as pRL-TK (Renilla luciferase-expressing construct; Promega), then co-transfected into Calu-6 cells. 48 h later, the Renilla-luciferase substrate coelenterazine (2.5 μg/mL) was added into the plates to detect the luciferase activity as a measure of transfection efficiency. The firefly luciferase substrate D-luciferin (100 μg/mL) was added for detection of p53-driven reporter activation.

TRIzol (Thermo Fisher Scientific) was used to extract total RNA according to the manufacturer’s instructions. cDNA synthesis was performed using the QuantiTect Reverse Transcription Kit (Qiagen, Hilden, Germany). All qPCR was performed using SYBR Premix ExTaq Kit (Roche, Basel, Switzerland) to analysis mRNA. The primers were listed in Suppl. Table 1.

1 × 10⁵ cells after 48-h transfection were plated in 96 well plate, followed by adding 10 μL CCK-8 solution (Solarbio, Beijing, China) to incubate additional 1 h. Absorption values of 450 nm were examined using Multiskan FC microplate reader (Thermo Fisher Scientific) to construct the regression equation and calculate IC50. The IC50 values were evaluated the cell proliferation.

1 × 10⁵ cells were cultured within 200 μL suspension in upper and 800 μL fresh medium in lower transwell chamber (Corning, New York, NY, USA) on 24-well plates for 24 h. The cells at the lower chamber were cross-linked by 1% paraformaldehyde for 10 min and stained by 0.5% crystal violet (Sangon Biotech, Shanghai, China) for 5 min. The stained cells were counted under Olympus BX-51 microscope to evaluate the cell invasion.

1 × 10⁵ cells were sub-cultured in 6 well plate with 80% density, and a straight line was gently created using a p200 pipette tip. Cells were then washed with PBS and cultured in medium for the period of the assay. After 8-h and 24-h cell culture, the distance between the two sides of the scratch was captured, measured the dynamic change of width and analyzed the cell migration by Image J. Amount of scratch closure was calculated as previously described (Ma et al., 2020).

Transfected with appropriate plasmids and empty vectors used as the experimental controls accordingly, the whole cell lysates or nuclear extracts were mixed 1 μg Flag, MEK or Rabbit IgG antibody, and 40 μL flurry IgA beads (Invitrogen) for rotating overnight at 4°C. Immunoprecipitates were washed by IP buffer (20 mM HEPES [pH 7.9], 350 mM NaCl, 0.1% NP-40, 1 mM DTT, 0.2 mM PMSF, 2 mg/mL leupeptin and 2 mg/mL aprotinin) and western blotted for appropriate antibodies. For immunoblotting, the transfer-ready membranes were blocked overnight in TBS (10 mM Tris-HCl [pH 7.5], 150 mM NaCl) containing 5% nonfat milk and 0.1% Tween-20 at 4°C, followed by incubation with appropriate primary antibodies. The secondary antibodies of horseradish peroxidase-conjugated anti-mouse, -rabbit, and -goat antibodies were used at a 1:5000 dilution.

Animal experiments in the current study were approved by our Institutional Laboratory Animals Committee and conducted following The Guidelines for Laboratory Animal Care and Use by the Affiliated Yantai Yuhuangding Hospital of Qingdao University. A total 16 BALB/c nude mice (male, weight 20–25 g) were supplied by Shanghai SLAC Laboratory Animal (Shanghai, China). Xenograft experiments were performed as described previously (Brown and Fan, 2016). In brief, mice were anesthetized using sodium pentobarbital (20 mg/kg). The stably DPYSL3 silencing COLO-684 cells and the stable DPYSL3 over-expressed Ishikawa cells (5 × 106) were suspended with 0.2 mL serum-free EGM-2 media and Matrigel matrix (v/v 1:1) and then injected onto the back of nude mice under sterile conditions. Tumor volume was calculated according to the formula: V = width 2 × length × 0.5. Mice were sacrificed on the 28th day.

The full length of amino acid of DPYSL3 and MEK were achieved and input into I-TASSER server (Zhang, 2008) (https://zhanglab.ccmb.med.umich.edu/) for homologous modeling by the comparison between candidate protein sequences and the known protein database, and given the protein structure. The models with highest score are selected. Both the files of DPYSL3 and MEK protein structure were put into Z-DOCK server (Pierce et al., 2014) (http://zdock.umassmed.edu/) to generate the protein docking. The given interaction complexes were visualized and analyzed by pyMOL to pick up confident model.

Data are presented as means ± standard deviations for three independent experiments. The difference of values was analyzed using Student’s t-test. P-values less than 0.05 was considered as statistical significance.

Commercialized kits, reagents, instruments, software, antibodies, etc., used in the research, shall be provided with their full name, along with the information of the Manufacturers/suppliers/software details (Name, City, Province/State, Country).

Accession numbers of RNA, DNA and protein sequences used in the manuscript should be provided.

To validate the presence of DPYSL3 up-regulation by p53 over-expression in our previous RNA sequencing data, DPYSL3 mRNA and protein level were examined in EC cells of COLO-684 and HEC-108 (wild type p53, wtp53) as well as Ishikawa and HEC-59 (mutated p53, mtp53) treated with a hypoxia-mimic drug deferoxamine (DFO) or 5-fluorouracil (5-FU) known to damage DNA. The higher expression of DPYSL3 was observed after DFO or 5-FU exposure compared to non-treatment both in COLO-684 but not in Ishikawa cells (Figs. 1A–1D), which indicated that DPYSL3 was induced in a wild type p53-regulated manner under hypoxic stress and DNA damage condition.

As a transcription factor, p53 has been well revealed its DNA binding domain sequence (El-Deiry et al., 1992). Interestingly, DPYSL3 promoter region was also found containing four putative p53 consensus binding sequences (CBS). To verify the contribution of CBS1-4 of DPYSL3 to p53-dependent transcriptional regulation, all four CBSs were cloned individually and inserted into upstream of the minimal SV40 promoter in the pGL-3 luciferase reporter plasmid. CBS1, 2 and 4 reporters were significantly activated in COLO-684 and HEC-108 with wtp53. Notably, the activity of CBS1 and 4 were obviously compromised in Ishikawa (M246V) and HEC-59 (R273H) (Fig. 2A). To further determine the interaction between p53 and these candidate CBSs of DPYSL3, we conducted ChIP assay, and observed that CBS1, 2 and 4 were the preferential binding domains of wtp53 in COLO-684 and HEC-108 but showed lower binding ability by mtp53 in Ishikawa and HEC-59, while S3 could not be recognized by wtp53 or mtp53 (Fig. 2B). Taken together, our data demonstrated that p53 targeted DPYSL3 promoter region and facilitated its transcriptional level to respond the hypoxia stress.

Now that DPYSL3 has been demonstrated to be promoted by wtp53 under hypoxia condition in EC cells, therefore, the potential effect of DPYSL3 upon tumor cells were further investigated. Generating the stable DPYSL3 RNA interference EC cell lines COLO-684 and the stable DPYSL3 overexpression of Ishikawa (mtp53), we examined CCK-8 assay (Fig. 3A), transwell assay (Fig. 3B) and wound healing assay (Fig. 3C) and observed that DPYSL3 indeed inhibited cell growth and invasion. Furthermore, xenograft mouse models were also used to verify that tumor cells with overexpressed DPYSL3 grew slowly than control group, while silencing DPYSL3 enhanced the malignancy in vivo (Fig. 3D).

To study the anti-proliferative mechanism of DPYSL3 in EC cells, we constructed Stag-Flag-DPYSL3 stably expressing COLO-684 cells, conducted Flag-IP and explored the probable binding protein of DPYSL3 using blue staining. Unlike in empty vector, one interested band near 45 kD was observed and identified as MEK1 using mass spectrometry assay. Moreover, to define the regions of DPYSL3 which are responsible for the interaction with MEK1, we performed a series of deletion analysis on DPYSL3 and conducted reverse IP to validate the interaction between DPYSL3 and MEK. We observed that the N-terminal (2–345) of DPYSL3 was essential for the interaction with MEK (Fig. 4A). And we further narrowed down the N-terminal of DPYSL3 and concluded that amino acids 2–16 and 143–159 (DPYSL3 tetramer interface domain) might be responsible for the interaction with MEK (Fig. 4B). Nevertheless, how does DPYSL3 affect MEK in tumor cells and whether MEK is involved in the inhibition of tumor cell proliferation mediated by DPYSL3 are both unknown. Therefore, we detected the phosphorylation level of ERK1/2 which is the downstream substrate of MEK1. Consistent with IP-assay, Thr202 phosphorylation level of ERK1/2 regulated by the DPYSL3 with A.A. 143–159 domain deletion group was higher compared to wild type and other DPYSL3 deletion groups (Fig. 4C), which indicated that the tetramer interface domain of DPYSL3 played an important role to suppress the MEK1 activity. Finally, we performed protein docking and speculated that the DPYSL3 might interact with MEK and suppress its kinase activity mainly through blocking MEK phosphorylation sites of 218 and 222 serine residues (Figs. 4D and 4E).