The KYSE-series cell lines (ACC 351, ACC 375, ACC 379, ACC 363, ACC 374 ACC 380, ACC 387; DSMZ) were cultured in RPMI-1640 medium (cat. no. 23400) supplemented with 10% fetal bovine serum (FBS; cat. no. 10270) and 1% penicillin-streptomycin (cat. no. 15070). NE-1 cells were cultured in Defined Keratinocyte SFM medium with Defined Keratinocyte SFM Growth Supplement (cat. no. 10744019). 293T, HKESC-2 and SLMT cells were cultured in DMEM (cat. no. 12100) supplemented with 5% FBS and 5% Newborn Calf Serum (NCS; cat. no. 16010), and 1% penicillin-streptomycin. 293T cells served as host cells for lentiviral production. Cells were seeded and incubated until full confluence and passaged every 2–3 days using Trypsin/EDTA (cat. no. 25200) for continuous culture at 37°C with 5% CO₂. All reagents used for cell culture were purchased from Invitrogen; Thermo Fisher Scientific, Inc. The cell lines were genotyped, authenticated and tested for mycoplasma prior to the experiments.

Mouse esophageal tissue isolation, organoid generation and culture were performed as previously described with modifications (15). In brief, 8–10-week old BALB/c/nu/nu athymic mice (n=4) were sacrificed through anesthetic overdose by intraperitoneal injection of ketamine/xylazine (300 mg/kg ketamine and 30 mg/kg xylazine) followed by carbon dioxide asphyxiation by placing mice in a carbon dioxide-filled container. The esophagus was surgically removed, and the mucosa was pulled away from the submucosa layer using forceps. The mucosa layer was cut into small pieces with a scalpel, dissociated in Trypsin-EDTA at 37°C for 30 min and subsequently neutralized with RPMI-1640 supplemented with 10% FBS. To generate organoids, the dissociated cells were suspended in Matrigel^® (cat. no. 356234; Corning, Inc.) and placed in 24-well low-bind culture plates (cat. no. 174930; Thermo Fisher Scientific, Inc.) at 37°C to solidify. The mNEEO was cultured in growth medium comprising advanced DMEM/F12 (cat. no. 12634), 1X N2 (cat. no. 17502) and 1X B27 Supplements (cat. no. 17504), 1X Glutamax (cat. no. 35050), 1X HEPES (cat. no. 15630), 1X penicillin/streptomycin, 1 mM N-acetyl-L-cysteine (cat. no. A9165, Sigma-Aldrich; Merck KGaA), 100 µm gastrin (cat. no. G9020, Sigma-Aldrich; Merck KGaA), 10 mM nicotinamide (cat. no. N0636, Sigma-Aldrich; Merck KGaA), 10 µm SB202190 (cat. no. S7067, Sigma-Aldrich; Merck KGaA), 50 ng/ml epidermal growth factor (cat. no. PHG0314), 100 ng/ml Noggin (cat. no. 120-10C; PeproTech, Inc.), 100 ng/ml Wnt3A (cat. no. 315-20; PeproTech, Inc.), 100 ng/ml R-Spondin 2 (cat. no. 120-43; PeproTech, Inc.) and 500 nM A8301 (cat. no. 2939; Tocris Bioscience) as previously described (15) and passaged every 7 days for continuous culture. All reagents were supplied by Thermo Fisher Scientific, Inc. unless otherwise specified. Ethics approval for animal studies was obtained from the Animal Ethics Committee of The University of Hong Kong (approval no. CULATR 3377-14).

For all-trans retinoic acid (ATRA; cat. no. R2625; Sigma-Aldrich; Merck KGaA) and Compound E (cat. no. sc-221433; Santa Cruz Biotechnology, Inc.) treatment of mNEEO, 2,000 dissociated esophageal epithelial cells were embedded in 4 wells/group in a low-bind 96-well culture plate (cat. no. 15237905; Thermo Fisher Scientific, Inc.). The esophageal epithelial cells were cultured in growth medium for 24 h prior to ATRA and Compound E treatment. Growth medium was replaced with growth medium containing 100 nM ATRA, 10 µM Compound E or DMSO (cat. no. D418; Sigma-Aldrich; Merck KGaA) vehicle after 24 h, and the cells were cultured for 9 days to generate organoids.

ZNF750 (NM_024702.3) was amplified using Q5 High-Fidelity DNA Polymerase (cat. no. M0491; New England Biolabs, Inc.) from KYSE30 cell line cDNA by PCR. The PCR procedure was performed using a Veriti™ 96-Well Thermal Cycler (cat. no. 4375786; Thermo Fisher Scientific, Inc.) under the following cycling conditions: Initial denaturation at 98°C for 30 sec, followed by 30 cycles of annealing at 56°C for 30 sec and extension at 72°C for 2 min. The oligonucleotide pairs used for amplification were ZNF750 forward, 5′-GGAATTCGCCACCATGAGTCTCCTCAAAGAGCGG-3′ and reverse, 5′-GCTCTAGATTACTTGTCGTCATCGTCTTTGTAGTCGGACACCCGGGCCCTC-3′. The amplified DNA was subsequently cloned into a pLVX-EF1α-puro vector. The ZNF750 sequence (pLVX-ZNF750) was validated by Sanger sequencing (16) at the Centre for PanorOmic Sciences (CPOS) at the University of Hong Kong (Hong Kong, SAR, China). pLVX-puro (cat. no. 632164; Clontech Laboratories, Inc.) was modified by removing the CMV promoter and replacing it with an EF1α promoter to construct a pLVX-EF1α-puro vector (17). The pLVX-EF1α-puro vector was used to insert a GFP gene to create the pLVX-GFP control plasmid (17).

293T cells were seeded in preparation for viral production in T25 culture flasks. When 293T cells reached 60% confluence, the transfection mixture was prepared using 8 µl X-tremeGENE™ HP DNA Transfection Reagent (cat. no. 6366546001; Roche Molecular Systems, Inc.), 2 µg plasmid DNA (pLVX-GFP or pLVX-ZNF750), 1.5 µg viral packaging vector psPAX2 (cat. no. 12260; Addgene, Inc.) and 0.5 µg viral envelope vector pMD2.G (cat. no. 12259; Addgene, Inc.) in 400 µl serum-free DMEM and incubated for 20 min at room temperature. The transfection mixture was added dropwise to the culture flask of HEK293T cells and incubated for 72 h at 37°C. Subsequently, the viral supernatant was passed through a 0.45-µm filter and aliquoted. The virus was stored at −80°C until further experiments. For lentiviral infection, KYSE30 and KYSE180 cells were seeded to achieve a confluency of ~30% at 24 h post-seeding. A 1 ml aliquot of pre-prepared virus expressing GFP or ZNF750 with 0.006 µl 5 µg/ml polybrene (cat. no. H9268; Sigma-Aldrich; Merck KGaA) was added to cells in tissue culture flasks to facilitate lentiviral infection. The medium was changed at 24 h post-infection to allow the cells to recover. The cells were passaged at day 2 post-infection for expansion and prepared for assays after the second passage.

RNA was extracted from cells and organoids using the TRI Reagent™ solution (cat. no. AM9738; Invitrogen; Thermo Fisher Scientific, Inc.), according to the manufacturer's instructions and reverse-transcribed to complementary DNA using PrimeScript RT Master Mix (Perfect Real Time) (cat. no. RR036; Takara Bio, Inc.) according to the manufacturer's instructions. FastStart Universal SYBR^®-Green Master (Rox) (cat. no. 4913914001; Roche Life Science) was used to determine the mRNA expression levels of target genes. Each reaction comprised 1X FastStart Universal SYBR^® Green Master mix, 30 ng/µl cDNA, and 0.4 mM each of forward and reverse primer. The reactions were performed in the LightCycler^® 480 Instrument II (cat. no. 05015243001; Roche Life Science) using the default SYBR^®-Green program. mRNA expression levels were quantified using the 2^−∆∆Cq method (18). Human GAPDH forward, 5′-AAGGTGAAGGTCGGAGTC-3′ and reverse, 5′-GAAGATGGTGATGGGATTTC-3′, was used as the internal control. The primer sequences were as follows: Human: ZNF750 forward, 5′-AAAGCACAGAATGCCTACCTG-3′ and reverse, 5′-GGGTCTCCGTTCACAACATT-3′; keratin 4 (KRT4) forward, 5′-GATGCTCTTCTGGGGGATT-3′ and reverse, 5′-GCTCTGGTTGATGGTGACCT-3′; SPRR1A forward, 5′-CCGTATACCAGCTTTCTGTCTC-3′ and reverse, 5′-GCTGGCAAGGTTGTTTCAC-3′; SPRR2E forward, 5′-TGGTACTTGAGCACTGATCTGC-3′ and reverse, 5′-TGCACTGCTGCTGTTGATAA-3′; TGM3 forward, 5′-TGGAAGGACTCTGCCACAAT-3′ and reverse, 5′-TGAGCGTACGAGATCTTTATGG-3′; and involucrin (IVL) forward, 5′-GCAGGAGGGACAGCTGAA-3′ and reverse, 5′-ATCTGCTCCTCTGGGACCTC-3′; mouse: Zfp750 forward, 5′-GAGCCAGCGTGAGAACAGA-3′ and reverse, 5′-CAGGAGAGTTCCTTCCGTCA-3′; Ivl forward, 5′-CCCTAGTTGCTGCTCAGCTC-3′ and reverse, 5′-TGGAAGCTTCTGTAGACCAAAA-3′; Krt4 forward, 5′-AACAAATTCGCGTCCTTCAT-3′ and reverse, 5′-CTGCTGGAGGAGGTTCCAT-3′; and Krt14 forward, 5′-ATCGAGGACCTGAAGAGCAA-3′ and reverse, 5′-TCGATCTGCAGGAGGACATT-3′.

Protein lysates were obtained by lysing KYSE30 and KYSE180 cells for 10 min at 4°C in RIPA buffer (cat. no. 89900; Thermo Fisher Scientific, Inc.) supplemented with 1X Halt™ Protease and Phosphatase Inhibitor Cocktail (cat. no. 78442; Thermo Fisher Scientific, Inc.), and 1 unit of Pierce Universal Nuclease (cat. no. 88700; Thermo Fisher Scientific, Inc.). Protein lysates were collected by centrifugation at 12,000 × g at 4°C for 10 min. The proteins were denatured at 95°C for 10 min prior to electrophoresis. A total of 50 µg protein per sample were electrophoresed using 12% SDS-PAGE at 80V for 30 min and 120V for 1.5 h using the Mini Gel Tank (cat. no. A25977; Thermo Fisher Scientific, Inc.). Following electrophoresis, the proteins within the gel were transferred to methanol-activated 0.45-µm PVDF membranes (cat. no. IPVH00010; MilliporeSigma) using a Mini Blot system (cat. no. B1000; Thermo Fisher Scientific, Inc.) at 30V for 1 h. Subsequently, the membranes were blocked with 3% bovine serum albumin (v/v) in Tris-buffered saline with 0.1% Tween (TBST) for 45 min at room temperature. The membranes were hybridized with the primary antibodies at 4°C overnight with constant shaking. The membranes were briefly washed with TBST and probed with a horseradish peroxidase (HRP)-conjugated secondary antibody diluted 1:5,000 for 1 h at room temperature with gentle agitation, followed by extensive washing with TBST prior to signal detection. Signals were detected by WesternBright ECL HRP substrate (cat. no. K-12045; Advansta, Inc.) and Super RX X-ray films (Fujifilm Corporation). Band intensities were measured using ImageJ software version 1.53a (National Institutes of Health) (19) and quantified by normalization to the GFP control. The primary antibodies used were polyclonal rabbit anti-human ZNF750 (1:1,000; cat. no. 21752-1-AP; Proteintech Group, Inc.), polyclonal rabbit anti-human SPRR3 (1:1,000; cat. no. A12041; ABclonal, Inc.), monoclonal mouse anti-human IVL (1:1,000; cat. no. MA5-11803; Thermo Fisher Scientific, Inc.), monoclonal mouse anti-human p84 (1:1,000; cat. no. GTX70220; GeneTex, Inc.), monoclonal rabbit anti-human GAPDH (1:5,000; cat. no. 5174; Cell Signaling Technology, Inc.) and monoclonal rabbit anti-GFP (1:1,000; cat. no. 2956; Cell Signaling Technology, Inc.). The secondary antibodies used were polyclonal anti-mouse IgG (1:5,000; cat. no. GTX213111-01) and polyclonal anti-rabbit IgG (1:5,000; cat. no. GTX213110-01) (both from GeneTex, Inc.).

Total RNAs of pLVX-GFP- or pLVX-ZNF750-transduced KYSE180 cells were extracted with the AllPrep DNA/RNA Micro kit (cat. no. 80284; Qiagen, Inc.) according to the manufacturer's instructions for RNA sequencing. Library preparation and Illumina sequencing (paired-end sequencing of 151 bp) were performed at CPOS. Bioanalyzer RNA Nano 6000 (Agilent Technologies) was used to determine the quality of the RNA. Ribosomal RNA (rRNA) depletion was performed using NEBNext^® rRNA Depletion Kit (Human/Mouse/Rat) (cat. no. E6310; New England BioLabs, Inc.). cDNA libraries were prepared using the NEBNext^® Ultra II Directional RNA Library Prep Kit (cat. no. E7760; New England BioLabs, Inc.). A total of 500 ng RNA was used as starting material for ribosomal RNA depletion. Cytoplasmic and mitochondrial rRNA were firstly hybridized to ssDNA capture probes and then digested by Ribonuclease H. Subsequently, the residual ssDNA probes were digested by DNase I. The rRNA-depleted RNA was purified by Agencourt^® RNAClean™ XP beads (cat. no. A63987; Beckman Coulter, Inc.). The purified rRNA-depleted RNA was fragmented to 200–500 bp by incubating at 94°C for 10 min in the presence of magnesium ions (NEBNext^® Magnesium RNA Fragmentation Module; cat no. E6150; New England BioLabs, Inc.). The fragmented rRNA-depleted RNA was then applied as a template to synthesize the first-strand cDNA using random hexamer-primer and a reverse transcriptase. In the second strand cDNA synthesis, the mRNA template was removed, and a replacement strand was generated to form the blunt-end double-stranded (ds) cDNA. The ds cDNA underwent 3′adenylation and indexed adaptor ligation, followed by treatment with User Enzyme. The adaptor-ligated libraries were enriched and indexed using NEBNext^® Multiplex Oligos for Illumina^® (96 Unique Dual Index Primer Pairs) (cat. no. E6440; New England BioLabs, Inc.) by 8 cycles of PCR according to the manufacturer's instructions. The concentration was determined using qPCR following library construction, and the libraries were denatured and diluted to 1.2 nM. Illumina NovaSeq 6000 Reagent Kit (cat. no. 20028312; Illumina, Inc.) was used for paired-end 151bp sequencing. Raw RNA-Seq data were cleaned and aligned to the hg19 reference genome using Tophat (version 2.0.14; bowtie version 2.2.4) (20) with library-type fr-firststrand parameter. Gene expression profile and differentially expressed genes were calculated by Cufflinks (version 2.2.1) with the Ensemble gene annotation file and Cuffdiff (21), respectively. Differential gene expression was calculated by log₂ fold-change (FC) expression. Genes with log₂ (FC) ≥2 were used for further analysis.

Gene Set Enrichment Analysis (GSEA) (22) and ToppFun (https://toppgene.cchmc.org/enrichment.jsp) (23) were used for the analysis of enriched pathways in ZNF750-overexrpressing cells. Using GSEA version 4.1.0, transcriptomic data were analyzed to identify the enriched gene sets in the ZNF750-overexpressing group compared with the GFP-expressing group to identify the genes driving the enrichment. In ToppFun, differentially upregulated genes in the ZNF750-overexpressing group compared with the control group were used for analysis. The gene sets from the Molecular Signatures Database (MsigDB; http://www.broad.mit.edu/gsea/msigdb/index.jsp) were used for both analyses. The consensus gene sets between the two analyses with a false discovery rate (FDR) ≤0.25 were considered to be enriched.

TCGA data were downloaded from the National Cancer Institute Genomic Data Commons (https://gdc.cancer.gov/) (24). Normalized gene expression profiles of SCC samples with RNA-seq data from the TCGA-ESCA project (dbGaP study accession no. phs000178) (25) were extracted using R version 4.0.2 (26) and the TCGAbiolinks R/Bioconductor package (27) (http://bioconductor.org/packages/release/bioc/html/TCGAbiolinks.html). ESCC data were divided into high-ZNF750 expressing tumors (ZNF750-high; n=40) and low-ZNF750 expressing tumors (ZNF750-low; n=41) using median normalized expression levels as the cut-off value. GSEA was used to determine the enriched gene sets in the ZNF750-high compared with the ZNF750-low group using the same cut-off values as aforementioned.

To assess cell viability, 2.5×10³ KYSE30 and 5×10³ KYSE180 cells per well were seeded in a 96-well plate with 100 µl medium. For visualization, 30 µl/well 5 mg/ml MTT dye (cat. no. M6494; Thermo Fisher Scientific, Inc.) was added to the cells and incubated at 37°C for 2 h. Subsequently, the medium was aspirated, and 100 µl DMSO was added to dissolve the formazan crystals. Absorbance was measured at 540 nm using an Ao Absorbance Microplate Reader (Azure Biosystems) daily for 3 days post-seeding.

For the ATRA treatment, parental KYSE30 and KYSE180 cells were seeded as aforementioned and cultured for 24 h. Subsequently, the culture medium in each well was replaced with medium containing 50 or 100 nM ATRA or DMSO and cultured for 3 days. Concurrently, cells from the same batch were seeded in T25 culture flasks and treated with ATRA. RNA was extracted as aforementioned from the ATRA-treated cells at 48 h post-treatment and subjected to RT-qPCR analysis. Experiments were performed on KYSE30 and KYSE180 cell lines using previously reported ATRA concentrations as a reference to determine the optimal ATRA concentration for each cell line (28).

For the colony formation assay, 1×10³ KYSE30 or KYSE180 cells were seeded into a single well of a 6-well culture plate to form colonies. The colonies were fixed with 37% formaldehyde for 30 min at room temperature and stained with 1:20 Giemsa solution in water overnight at room temperature. The solution was washed away with 1X PBS until the buffer was clear. Colonies were counted using the Gel Doc XR+ system (cat. no. 170-8195; Bio-Rad Laboratories, Inc.).

ESCC tumorigenicity was evaluated by subcutaneous inoculation of ESCC cells into the flanks of 8–10-week-old BALB/c/nu/nu athymic nude mice. A total of 1×10⁶ KYSE30 cells were resuspended in 150 µl of serum-free medium and injected in both flanks of each mouse (n=4 mice per group). Tumor growth was monitored weekly for ≥4 weeks and measured by calipers to calculate the tumor volume as follows: Volume=(length × width × height). At the experimental endpoint, mice were sacrificed by anesthetic overdose through intraperitoneal injection of ketamine/xylazine (300 mg/kg ketamine and 30 mg/kg xylazine) followed by carbon dioxide asphyxiation by placing mice in a carbon dioxide-filled container. Cessation of cardiac function was used as the method for verification of death. All experimental procedures were approved by the Committee on the Use of Live Animals in Teaching and Research at The University of Hong Kong (approval no. CULATR 3377-14).

Data are presented as the mean ± SEM. Each in vitro experiment was repeated at least three times in duplicate, unless otherwise stated. Unpaired Student's t-test was performed for all statistical analyses using GraphPad Prism version 8.4.3 (GraphPad Software, Inc.). P<0.05 was considered to indicate a statistically significant difference.

mNEEOs were used to determine the role of Zfp750 in the esophageal keratinocyte differentiation program. Following ATRA treatment, the organoids demonstrated increased expression levels of Ivl and Krt4 and decreased levels of Krt14 compared with those in the vehicle-treated mNEEOs (Fig. 1A), suggesting an induction of the keratinocyte differentiation program and subsequent decrease in the basal cell layers within the organoid. By contrast, following Compound E treatment, the expression levels of Ivl and Krt4 were markedly decreased, whereas those of Krt14 were increased compared with those in the vehicle-treated group (Fig. 1B), suggesting the inhibition of the differentiation program and an increase in the basal cell layers in these organoids. Changes in Zfp750 expression corresponded with the expression levels of the markers of differentiation, Ivl and Krt4, after both treatments (Fig. 1C). These results suggested that Zfp750 expression may be proportionally affected by the stimulation and inhibition of the esophageal keratinocyte differentiation program, and that Zfp750 may be involved in this cellular process in esophageal keratinocytes.

To determine whether ZNF750 expression may affect the keratinocyte differentiation program in ESCC tumors, RNA-seq data of ESCC tumor samples (n=81) from the TCGA-ESCA dataset were analyzed using GSEA. The samples were stratified into ZNF750-high [median, 20.07; interquartile range (IQR), 19.03–21.11] and ZNF750-low (median, 17.86; IQR, 16.45–19.27) groups (Fig. 2A). The results of GSEA confirmed that epithelial and epidermal differentiation-related cellular processes including ‘cornification’, ‘keratinization’, ‘keratinocyte differentiation’ and ‘epidermal cell differentiation’ (Table SI) were enriched in the ZNF750-high group compared with the ZNF750-low group. ‘Keratinocyte differentiation’ and ‘epidermal cell differentiation’ gene sets were among the most enriched sets (Fig. 2B), whereas SPRR, TGM and IVL were among the most enriched genes (Table SII). These results supported the involvement of ZNF750 in keratinocyte differentiation in ESCC.

ZNF750 expression levels were downregulated in the ESCC cell line models compared with those in the immortalized human esophageal epithelial cell line NE1 (Fig. 3A); therefore, to study the suppressive function of ZNF750 in ESCC, ZNF750 was overexpressed in KYSE30 and KYSE180 cells by lentiviral transduction of pLVX-GFP and pLVX-ZNF750. These cell lines were selected as the models for further experiments as they exhibited decreased expression of ZNF750 at the mRNA and protein level (Fig. 3A and B). ZNF750 expression levels were upregulated in both cell lines in the pLVX-ZNF750 groups compared with those in the corresponding pLVX-GFP groups, as demonstrated by RT-qPCR and western blot analyses (Fig. 3B). In vitro MTT cell viability and colony formation assays and in vivo subcutaneous tumorigenicity model were used to assess the tumor-suppressive ability of ZNF750. In vitro, ZNF750 overexpression suppressed ESCC cell viability compared with that in the control group (Fig. 3C). Colony formation ability was also reduced in the ZNF750-overexpressing groups compared with that in the controls (Fig. 3D). In the in vivo subcutaneous tumor model, mice in the ZNF750 overexpression group developed smaller tumors (n=8; mean ± SEM, 184±26.5 mm³) in mice compared to the GFP control (n=8; mean ± SEM, 311.17±47.76 mm³) (P<0.05; Fig. 3E). These results demonstrate the ability of ZNF750 to suppress cell viability and growth in vitro and in vivo.

To determine the pathways affected by ZNF750 overexpression in the in vitro model, ZNF750 and GFP lentivirus-transduced KYSE180 cells were prepared for RNA-seq and subjected to subsequent bioinformatics analyses. The results of the analyses using the MSigDB revealed enrichment of gene sets involved in epithelial differentiation including ‘epithelial cell differentiation’, ‘epidermal cell differentiation’ and ‘keratinocyte differentiation’ (Table SIII). Consistent with the GSEA analysis of ZNF750-high ESCC tumors, these results confirmed the enrichment of keratinocyte differentiation and epidermal cell differentiation gene sets in ZNF750-overexpressing ESCC cells compared with the controls (Fig. 4A). The upregulation of differentiation-related genes KRT4, SPRR1A, SPRR2E, TGM3, IVL and SPRR3 were validated by RT-qPCR or western blotting (Fig. 4B). These results suggested that ZNF750 overexpression induced the expression of differentiation-related genes in the ESCC cell line model.

To determine the effects of the induction of the keratinocyte differentiation program on ESCC cell viability, parental KYSE30 and KYSE180 cells were treated with ATRA and subsequently tested for changes in cell viability using the MTT assay. The results demonstrated a significant suppression in the viability of the ATRA-treated parental ESCC cells compared with that in the vehicle-treated group at 48 h post-treatment (Fig. 5A). RT-qPCR analysis revealed that ATRA treatment upregulated the expression levels of IVL compared with those in the vehicle group (Fig. 5B), indicating the induction of terminal differentiation. These results suggested that induction of the keratinocyte differentiation program through to terminal differentiation may contribute to the suppression of ESCC cell viability following ATRA-induced stimulation of differentiation.