Abstract

   Tangeretin (Tan), a citrus flavonoid, possesses a strong anti-tumor
   efficacy in various human cancers. However, the precise role of Tan in
   the development of esophageal squamous cell carcinoma (ESCC) remains
   unclear. RNA sequencing (RNA-seq) analysis was performed to observe the
   Tan-related genes in Tan-treated TE-1 cells. The direct relationship
   between GLI family zinc finger 2 (GLI2) and the promoter of
   glycoprotein non-metastatic melanoma protein B (GPNMB) was predicted by
   bioinformatics analysis and validated by luciferase reporter and
   chromatin immunoprecipitation (ChIP) assays. Cell survival after Tan
   treatment was assessed by CCK8 assay. Gene expression levels were
   evaluated by a qRT-PCR, western blot, or immunofluorescence method.
   Cell migration and invasion were detected by wound-healing and
   transwell assays. The function of Tan in vivo was examined using
   xenograft studies. Our data indicated anti-migration and anti-invasion
   functions of Tan in ESCC cells in vitro. Tan also diminished tumor
   growth in vivo. Mechanistically, Tan diminished the expression and
   transcriptional activity of GLI2 in ESCC cells. Silencing of GLI2
   resulted in decreased expression of GPNMB by inhibiting GPNMB
   transcription via the binding site at the GPNMB promoter at position
   +(1539–1550). Moreover, Tan down-regulated GPNMB expression in ESCC
   cells, and re-expression of GPNMB reversed anti-migration and
   anti-invasion functions of Tan in ESCC cells. Our findings uncover
   anti-migration and anti-invasion effects of Tan in ESCC cells by
   down-regulating GPNMB by suppressing GLI2-mediated GPNMB transcription,
   providing new evidence that Tan can function as a therapeutic agent
   against ESCC.

Introduction

   As the most frequent type of esophageal cancer, esophageal squamous
   cell carcinoma (ESCC) is a deadly and high-mortality malignancy
   worldwide [[36]1, [37]2]. Although clinically therapeutic regimens,
   including surgery and radiochemotherapy, can treat many primary ESCC
   tumors effectively, these therapies remain restricted in controlling
   the metastasis of ESCC cells and curbing metastatic or advanced ESCC
   [[38]3]. Hence, new anti-metastasis and anti-cancer therapeutic
   approaches are needed in ESCC.

   Tangeretin (Tan), a type of bioactive polymethoxylated flavones found
   in citrus peel, has been unveiled to possess many pharmacological
   functions, such as anti-inflammatory, anti-oxidant, anti-viral,
   cardioprotective, neuroprotective, and hepatoprotective properties
   [[39]4]. Of note, the anti-tumor efficacy of Tan has recently become
   increasingly clear due to its growth repression, apoptosis induction,
   anti-angiogenesis, and anti-metastasis effects [[40]5]. Furthermore,
   Tan enhances the sensitivity of the chemotherapy drugs and reduces
   chemotherapy-induced toxicity [[41]6, [42]7]. Functional studies in
   various cancer cell lines have highlighted the tumor-inhibitory
   potential of Tan [[43]8–[44]10], but its precise action in ESCC
   pathogenesis remains unclear.

   Transcription factors (TFs) might prove crucial in promoting human
   carcinogenesis by controlling the transcription of the key genes by
   binding to their promoters [[45]11]. GLI family zinc finger 2 (GLI2), a
   key TF that mediates the Sonic Hedgehog signaling, actively
   participates in human carcinogenesis [[46]12, [47]13]. The activation
   and aberrant expression of GLI2 have been reported to induce the
   malignant phenotypes of multiple cancers, such as castration-resistant
   prostate cancer, renal cell carcinoma and gallbladder cancer
   [[48]14–[49]16]. In human esophagus adenocarcinoma, GLI2 is highly
   expressed and contributes to carcinoma formation [[50]17]. Importantly,
   long noncoding RNA NR2F1-AS1 contributes to ESCC development by
   activating the Sonic Hedgehog pathway by elevating GLI2 expression
   [[51]18], suggesting the oncogenic activity of GLI2 in ESCC. Although
   several other TFs, such as nuclear liver X receptor α (LXRα) and signal
   transducer and activator of transcription 3 (STAT3), have been
   illuminated as the downstream effectors of Tan [[52]19, [53]20], no
   reports showed the involvement of GLI2 in the anti-cancer property of
   Tan in human tumors.

   Here, we sought to elucidate the activity and molecular determinant of
   Tan in ESCC, with the hope that these findings might provide new
   evidence for developing Tan as a promising anti-tumor agent.

Materials and methods

Cell culture and treatment

   Human ESCC cell line TE-1 (Procell, Wuhan, China) was propagated in 10%
   FBS RPMI-1640 (Servicebio, Wuhan, China), and the KYSE150 cell line
   (Procell) was cultured in 10% FBS RPMI-1640 containing Ham’s F-12
   (Servicebio). The 293T cell line (Procell) used for luciferase reporter
   assay was grown in complete medium provided by Procell. Human normal
   esophageal HET-1A cells (Wanwusw, Hefei, China) were cultured in BEGM
   Kit (Wanwusw). All cells were maintained in a fully humidified
   incubator in the presence of 5% CO[2] at a temperature of 37°C.

   Tan (≥98% purity) was procured from Macklin (Shanghai, China) and
   dissolved in DMSO for all experiments. For treatment of Tan, TE-1 and
   KYSE150 cells were incubated with or without Tan for 24 h at a
   concentration of 20 μg/ml.

Determination of cell survival after Tan treatment

   The survival rate of Tan-treated cells was evaluated by CCK8 assay
   based on the standard protocols [[54]21]. Cells (5.0 × 10^3 cells/well)
   were seeded in regular growth media into 96-well culture dishes. 12 h
   post-seeding, the cells were challenged with different concentrations
   of Tan. After incubation of 48 h, 10 μl CCK8 reagent (Beyotime
   Biotechnology, Shanghai, China) was added to each well. After addition,
   one-hour incubation was allowed at 37°C. Using a microplate reader
   (Thermo Fisher Scientific, MA, USA), the absorption was gauged at a
   wavelength of 450 nm. Surviving fraction (%) in Tan treatment group was
   estimated relative to that of the untreated counterpart. The IC50 value
   for Tan was determined from a plot of the percentage of surviving cells
   (50%) versus Tan concentration.

RNA sequencing

   For RNA sequencing (RNA-seq) as described [[55]22], TE-1 cells were
   treated with 20 μg/ml Tan or DMSO for 24 h. RNA was prepared from the
   cells using magnetic beads with Oligo (dT) and fragmented. The quality
   and quantity of extracted RNA were analyzed using Qubit 2.0 fluorimeter
   (Thermo Fisher Scientific). RNA-seq analysis of high-quantity RNA for
   the construction of sequencing libraries was conducted by BGI
   Technology (Shenzhen, China) using the BGISEQ-500 sequencing platform
   (BGI Technology) following standard protocols. Read alignment mapped to
   the human genome grch37 was performed using Hisat2 software [[56]23],
   and the read counts per group were calculated using the Python package
   HTseq [[57]24]. The tab-delimited text files included FPKM values for
   each sample. Differentially expressed genes were defined under the
   conditions: log2|fold-change| > 1 and P < 0.05. Gene Ontology (GO) and
   Kyoto Encyclopedia of Genes and Genomes (KEGG) enrichment analyses were
   conducted using the R package “Cluster Profiler”. A loop graph was
   drawn for the visualization of the enrichment results.

Bioinformatics analysis

   The differentially expressed genes (log2|fold-change| > 1 and P < 0.05)
   in ESCC tissues (n = 77) and normal esophageal tissues (n = 11) were
   obtained from The Cancer Genome Atlas (TCGA) database
   ([58]https://www.tcga.org/). All human transcriptional factors (hTFs)
   were downloaded from AnimalTFDB database
   ([59]http://bioinfo.life.hust.edu.cn/AnimalTFDB/). The TF binding sites
   for human were predicted using the Gene Transcription Regulation
   Database (GTRD) database ([60]http://gtrd20-06.biouml.org/). To obtain
   the downstream targets of GLI2 TF, we used the database hTF-target at
   [61]http://bioinfo.life.hust.edu.cn/hTFtarget#!/. To search the binding
   sites between GLI2 and the promoter of glycoprotein non-metastatic
   melanoma protein B (GPNMB), we interrogated the JASPAR database
   ([62]https://jaspar.genereg.net/).

siRNAs, plasmids and cell transfection

   Human GLI2-specific siRNA (si-GLI2) and a non-silencing siRNA mock
   (si-NC) were procured from Wuzhoukangjian Biological Technology Co.,
   LTD (Tianjin, China). The coding sequence of human GPNMB was obtained
   by PCR amplification and inserted into pECMV-MCS-FLAG expressing
   plasmid (Miaoling Biotechnology, Wuhan, China). As control, the
   scrambled sequence was processed in the same way to generate a plasmid
   control (vec).

   TE-1 and KYSE150 cells (5.0 × 10^5 cells/well) were seeded in 24-well
   culture dishes 24 h prior to transfection using Lipofectamine 3000
   (Thermo Fisher Scientific). The following day, 50 nM of siRNA or/and
   500 ng of plasmid was transfected into cells as per the manufacturing
   recommendations for transfection. Cells were collected 24 h
   post-transfection and subjected to Tan treatment.

qRT-PCR of mRNA in ESCC cell lines

   RNA preparation from TE-1 and KYSE150 cells after Tan treatment or/and
   transfection was done using RNAeasy™ Animal RNA Isolation Kit with Spin
   Column based on the recommendations of the manufacturer (Beyotime
   Biotechnology). For mRNA analysis, cDNA was oligo(dT) primed using
   SweScript RT II First Strand cDNA Synthesis Kit (Servicebio). Generated
   cDNA was diluted 10 folds, and qRT-PCR was performed on ABI
   QuantStudio6 Felx Real-Time System (Thermo Fisher Scientific) using the
   primers listed in [63]Table 1. mRNA expression data were obtained using
   the 2^-ΔΔCt method [[64]25] after normalization with reference to
   expression of β-actin.

Table 1. Primers sequences used for qRT-PCR.

   Name                              Primers for qRT-PCR (5’-3’)
   GLI2                      Forward TTGACATGCGACACCAGGAA
                             Reverse AGAACGGAGGTAGTGCTCCA
   BARX2                     Forward CCAAGGAGACCTGCGATTACT
                             Reverse TGTTCCGTCTCTGACTCGCT
   EN1                       Forward GCAACCCGGCTATCCTACTT
                             Reverse TCGTTGAGGCTGAGTTCCTG
   BMP7                      Forward TGGTCCACTTCATCAACCCG
                             Reverse CCGGACCACCATGTTTCTGT
   GPNMB                     Forward GCGAGATCACCCAGAACACA
                             Reverse ACACCAAGAGGGAGATCACAG
   β-actin                   Forward CTCGCCTTTGCCGATCC
                             Reverse GGGGTACTTCAGGGTGAGGA
   GPNMB promoter +(343–354) Forward CAAGCAATCACGAGCACAGG
                             Reverse TGTGGTGCCTCCCTCTCTAT
   [65]Open in a new tab

Western blot

   Lysates of TE-1 and KYSE150 cells after Tan treatment or siRNA
   transfection were conducted using RIPA lysis buffer (Beyotime
   Biotechnology) with protease and phosphatase inhibitor cocktail
   (Beyotime Biotechnology) for 20–30 min on ice. For western blot
   [[66]16], approximately 35 μg of protein was loaded per lane and
   electrophoresed on 5–12% SDS-PAGE gels. After the resulting gels were
   blotted onto PVDF membranes (Millipore, Bedford, MA, USA), probing was
   conducted with primary antibodies (rabbit polyclonal) including GLI2
   (18989-1-AP, 1:1000, Proteintech, Wuhan, China), GPNMB ([67]GB111475,
   1:1000, Servicebio), Vimentin ([68]GB111308, 1:1000, Servicebio),
   E-Cadherin (E-Cad, GB11082, 1:1000, Servicebio), N-Cadherin (N-Cad,
   [69]GB111273, 1:1000, Servicebio), β-Tubulin (GB11017, 1:1000,
   Servicebio), Slug (12129-1-AP, 1:2000, Proteintech), Snail (13099-1-AP,
   1:1000, Proteintech), VEGF (19003-1-AP, 1:1500, Proteintech), and
   Cyclin D1 (26939-1-AP, 1:10000, Proteintech), and β-actin (GB11001,
   1:1000, Servicebio) antibodies. Following incubation with secondary
   HRP-tagged anti-rabbit IgG antibody (GB23303, 1:3000, Servicebio),
   signals were developed with Ultra sensitive ECL Chemiluminescence Kit
   as described by the manufacturer (Servicebio).

Luciferase reporter assay

   To evaluate the impact of Tan treatment on GLI2 transcriptional
   activity, 5.0 × 10^5 TE-1 cells grown in 24-well culture dishes were
   co-transfected using Lipofectamine 3000 with 100 ng pRL-TK control
   plasmid, constitutively expressing Renilla luciferase (for
   normalization, Promega), and 400 ng GLI2-luc luciferase reporter
   plasmid (Yeasen, Shanghai, China). Cells were subjected to Tan
   treatment (20 μg/ml, 24 h) at 24 h after transfection and assayed for
   luciferase activities. Firefly luciferase activity was normalized to
   that of Renilla luciferase.

   To determine the binding site between GLI2 and the GPNMB promoter, the
   fragments (about 100 bp) of the GPNMB promoter containing the
   +(343–354) (named BS-1) position sequence or +(1539–1550) (named BS-2)
   position sequence, synthesized by Qingke Biotechnology (Beijing,
   China), were subcloned into the 3’UTR of the firefly luciferase coding
   region into pGL3-basic luciferase plasmid (Miaoling Biotechnology) as
   described previously [[70]26]. For luciferase assays, 5.0 × 10^5 293T
   cells growth in 24-well culture dishes were co-transfected using
   Lipofectamine 8000 (Beyotime Biotechnology) with reporter construct,
   siRNA and pRL-TK control plasmid based on the manufacturer’s
   suggestion. 48 h post-transfection, cells were lysed for luciferase
   activities and firefly luciferase activity was normalized to that of
   Renilla luciferase.

Chromatin immunoprecipitation (ChIP) assay

   For ChIP assay [[71]26], the ChIP Kit was applied as recommended by the
   manufacturer (Beyotime Biotechnology). Briefly, 1.0 × 10^7 TE-1 cells
   were fixed with 1% formaldehyde (Macklin) for 10 min, lysed in SDS cell
   lysis buffer (Beyotime), and sonicated to fragment DNA length into
   200–800 bp using the Ultrasonic Cell Disruptor (Lichen, Shanghai,
   China). IP complexes were immunoprecipitated with anti-GLI2 antibody
   (ab226390, Abcam, Cambridge, UK) or rabbit IgG control (10284-1-AP,
   Proteintech) overnight at 4°C. DNA was purified with DNA Product
   Purification Kit (Solarbio, Beijing, China) and amplified by qRT-PCR
   using specific primers ([72]Table 1).

Measuring cell migration by wound-healing assay

   TE-1 and KYSE150 cells after Tan treatment or/and plasmid transfection
   were seeded in 6-well culture dishes in 2 ml total volume to yield a
   monolayer of 80–90% confluence following overnight incubation. A
   homogeneous wound track was created using the 200 μl of sterile pipette
   tips. After 24 h culture in regular media at 37°C, images were acquired
   and analyzed using Image J software (National Institutes of Health,
   Bethesda, USA). We scored cell migration ability by gauging cell
   motility distance [[73]27].

Transwell migration and invasion assays

   TE-1 and KYSE150 cells after Tan treatment or/and plasmid transfection
   were resuspended in non-serum medium and placed in 24-transwell inserts
   (Corning, New York, USA) precoated without (for migration analysis, 4.0
   × 10^5 cells/well) or with (for invasion analysis, 2.0 × 10^6
   cells/well) 1 mg/ml Matrigel (BD Biosciences). Then, the inserts were
   placed in 24-wells containing 10% FBS growth medium. Following 24-h
   incubation at 37°C, the cells that passed the membranes to the
   undersurface were photographed and counted in 10 random microscopic
   fields as described elsewhere [[74]27].

Xenograft studies

   Six-week-old female BALB/c athymic nude mice (n = 10, Beijing Vital
   River Laboratory Animal Technology Co., Ltd., Beijing, China) were used
   for evaluation of the tumor-inhibitory effect of Tan. For formation of
   xenograft tumors [[75]28], KYSE150 cells (5 × 10^6 cells/mouse) were
   implanted into the right flanks of nude mice via subcutaneous
   injection. Five days later, the mice were administrated with Tan (20
   mg/kg) or PBS at the same volume by intraperitoneal injection every
   three days. Each group included five mice. All mice were euthanized at
   day 35 by a CO[2] overdose method by gradually elevating the
   concentration of CO[2] in the anesthesia chamber. After the mice
   stopped breathing and heartbeat completely, the tumors were removed,
   weighed, and paraffin-embedded. All animal procedures were conducted in
   accordance with a protocol approved by Animal Care and Use Committee of
   Henan Provincial People’s Hospital.

Immunofluorescence

   For cell immunofluorescence staining [[76]29], TE-1 and KYSE150 cells
   after Tan treatment or/and plasmid transfection were fixed for 15 min
   in 4% paraformaldehyde, permeabilized by incubation for 20 min in 0.5%
   Triton X-100, and blocked for 30 min in 3% BSA at room temperature. For
   tumor immunofluorescence staining, paraffin-embedded tumors were
   sectioned at 4 μm thickness, dehydrated, boiled in 10 nM sodium citrate
   for 30 min, and blocked in 3% BSA for 30 min. PCNA or MMP9 was probed
   with a rabbit polyclonal antibody (Servicebio, Wuhan, China) to PCNA
   (GB11010, 1:500) or MMP9 (GB11132, 1:1000), respectively. The Alexa
   Fluor 488 conjugated goat anti-rabbit IgG (for cell staining, GB25303,
   1:500, Servicebio) or Cy3 conjugated goat anti-rabbit IgG (for tumor
   staining, GB21303, 1:300, Servicebio) was used for visualization.
   Nuclei were subsequently stained by DAPI (Servicebio). Fluorescent
   images were obtained under an Olympus fluorescence microscope (BX53,
   Olympus, Tokyo, Japan) and analyzed using ImageJ software (NIH,
   Bethesda, MA, USA).

Statistical analysis

   Results were compared using ANOVA with Tukey’s post hoc test (for
   multiple comparisons) or Student’s t-test (two-tailed, for two-group
   comparison). Data were shown as mean ± SEM of three independent
   experiments (n ≥ 3) that were done in triplicate. Difference between a
   single or combination treatment and control was considered significant
   at a p value less than 0.05.

Results

Tan exerts anti-migration and anti-invasion functions in ESCC cells in vitro

   We first evaluated the anti-tumor effects of Tan in TE-1 and KYSE150
   ESCC cells in vitro. The cells were exposed to various concentrations
   of Tan, and cell viability was analyzed at 48 h. We found that exposure
   of Tan caused a remarkable reduction in cell survival rate, indicating
   that Tan hindered cell growth with the IC50 being 85.92 μg/ml in TE-1
   cells and 94.34 μg/ml in KYSE150 cells ([77]Fig 1A). We then treated
   cells with 20 μg/ml of Tan and evaluated the effect on cell migration
   and invasiveness after 24 h. Through wound-healing and transwell
   assays, we observed that treatment of Tan led to a striking reduction
   in the migratory abilities of the two ESCC cell lines ([78]Fig 1B and
   1C). Furthermore, treatment of Tan diminished cell invasiveness in
   vitro ([79]Fig 1D). Immunofluorescence results showed that TE-1 and
   KYSE150 cells treated by Tan exhibited lower levels of proliferating
   marker PCNA and metastasis-related protein MMP9 compared with controls
   ([80]Fig 2A and 2B). These results demonstrate that Tan can diminish
   ESCC cell growth, migration, and invasiveness, thereby exerting
   anti-ESCC activity.

Fig 1. Anti-migration and anti-invasion effects of Tan in TE-1 and KYSE150
ESCC cells.

   [81]Fig 1
   [82]Open in a new tab

   (A) TE-1 and KYSE150 cells were exposed to incremental concentrations
   of Tan for 48 h and checked for cell viability by CCK8 assay. Also
   shown were the IC50 values in the two cell lines. (B) Cells were
   treated with 20 μg/ml of Tan for 24 h, and cell migration was
   quantified as the wound-healed distance. Also shown were the
   representative pictures. Scale bar: 100 μm. (C and D) Representative
   images depicting transwell migration and invasion assays performed with
   control (Con) or Tan-treated cells. Scale bar: 100 μm. *P < 0.05, **P <
   0.01, ***P < 0.001.

Fig 2. Regulation of Tan in PCNA and MMP9 expression in ESCC cells.

   [83]Fig 2
   [84]Open in a new tab

   (A and B) Immunofluorescence assay showing the fluorescence intensity
   of PCNA and MMP9 in cells treated with control (Con) or Tan. Scale
   bars: 100 μm (A) and 50 μm (B). **P < 0.01.

Tan diminishes tumor growth in vivo

   To elucidate whether Tan possesses a tumor-inhibitory function, we
   implanted KYSE150 cells subcutaneously into the right flanks of nude
   mice, with administration of Tan every three days. As shown in [85]Fig
   3A and 3B, images showed a significant reduction in tumor volume as a
   result of Tan treatment. Moreover, Tan treatment reduced average weight
   of the xenograft tumors ([86]Fig 3C). This result was also verified by
   immunofluorescence for cell proliferation using PCNA. Tan treatment
   suppressed the expression of PCNA in the xenograft tumors ([87]Fig 3D).
   Additionally, immunofluorescence results revealed that Tan treatment
   led to a clear down-regulation of MMP9 expression in the tumors
   ([88]Fig 3E). All these data indicate the tumor-inhibitory effect of
   Tan in vivo.

Fig 3. The tumor-inhibitory effect of Tan in vivo.

   [89]Fig 3
   [90]Open in a new tab

   KYSE150 cells (5 × 10^6) were implanted subcutaneously into right
   flanks of nude mice, with administration of Tan every three days. 35
   days later, the tumors were harvested from the mice. (A) Images of mice
   with the xenograft tumors. (B) Images of the xenograft tumors. (C) Mean
   weight of the xenograft tumors. (D and E) Immunofluorescence for PCNA
   and MMP9 levels in the harvested tumors. Scale bar: 100 μm. *P < 0.05.

Transcriptional profiling of Tan-treated ESCC cells and selection of the
regulatory TFs of Tan in ESCC

   To find important genes in relation to the anti-ESCC activity of Tan,
   we performed RNA sequencing (RNA-seq) on Tan-treated TE-1 ESCC cells
   and control cells. We identified 587 genes with a significant variation
   following Tan treatment in TE-1 cells, in which 361 genes were
   down-regulated and 226 genes were up-regulated ([91]S1 Table), and the
   cluster heat map of differentially expressed genes was shown in [92]Fig
   4A. We then performed GO and KEGG enrichment analyses to elucidate the
   mechanisms of the 587 genes on biological process and KEGG pathway.
   Results showed that these genes were closely associated with
   extracellular structure organization (GO: 0043062), collagen-containing
   extracellular matrix (GO: 0062023), extracellular matrix organization
   (GO: 0030198), ossification (GO: 0001503), heparin binding (GO:
   008201), glycosaminoglycan binding (GO: 00055393), sulfur compound
   binding (GO: 1901681), extracellular matrix part (0044420), and
   interstitial matrix (0005614). KEGG pathway enrichment analysis pointed
   to the hippo signaling pathway (hsa04390), Complement and coagulation
   cascades (hsa04610), Basal cell carcinoma (hsa05217), wnt signaling
   pathway, and cholesterol metabolism ([93]Fig 4B). To obtain the TFs
   that are not only aberrantly expressed in ESCC but also are controlled
   by Tan treatment in ESCC cells, we combined these genes including the
   587 genes altered by Tan treatment in TE-1 ESCC cells ([94]S1 Table),
   the all 1665 hTFs from AnimalTFDB database ([95]S2 Table), and the
   differentially expressed genes in ESCC tissues and normal esophageal
   tissues using the TCGA database ([96]S3 Table). As shown in [97]Fig 4C,
   Venn diagram revealed a total of 18 TFs that could be regulated by Tan
   in TE-1 ESCC cells and were deregulated in ESCC. By combining the 18
   TFs with the 2684 up-regulated genes in ESCC and the 361 genes
   down-regulated by Tan in ESCC TE-1 cells, we found a total of 6 TFs
   (BNC1, GLI2, BARX2, FOXN1, EN1 and NHLH2) ([98]Fig 4D). Using the GTRD
   database, among the 6 TFs, we found that GLI2, BARX2 and EN1 had been
   shown to contain binding sites for human, which were identified by
   ChIP-seq experiments (-1000, +100).

Fig 4. Selection of TFs that were up-regulated in ESCC and were inhibited by
Tan treatment in TE-1 ESCC cells.

   [99]Fig 4
   [100]Open in a new tab

   (A) Heat map obtained from high-throughput RNA-seq profiling of
   Tan-treated TE-1 ESCC cells (20 μg/ml, 24 h) and control cells (Con).
   (B) The loop graph showing the markedly enriched biological processes,
   molecular function, cellular components, and KEGG pathways. The red and
   blue dots represent up-regulated and down-regulated genes,
   respectively. (C) Venn diagram revealing the 18 TFs that were not only
   aberrantly expressed in ESCC but also were controlled by Tan treatment
   in TE-1 ESCC cells. (D) Differential ordering diagram of 18 TFs in ESCC
   vs normal tissues, as well as in Tan treatment and con group.

Tan reduces the expression and transcriptional activity of GLI2 in ESCC cells

   Next, we wanted to evaluate whether the three TFs (GLI2, BARX2 and EN1)
   are involved in the anti-ESCC activity of Tan. We analyzed their
   expression in Tan-treated TE-1 and KYSE150 ESCC cells by qRT-PCR.
   Treatment of Tan led to a significant down-regulation in GLI2 mRNA
   expression, and BARX2 and EN1 mRNA levels did not reduce following Tan
   treatment in the cells ([101]Fig 5A and 5B). We thus focused on the
   GLI2 TF as a potential target gene of Tan. Our western blot results
   further confirmed the reduction of GLI2 expression at the protein level
   following Tan treatment in TE-1 and KYSE150 ESCC cells ([102]Fig 5C).
   To determine whether Tan could influence the transcriptional activity
   of GLI2, we adopted luciferase assays in TE-1 cells. Treatment of Tan
   caused a clear down-regulation in the luciferase activity of the
   GLI2-luc luciferase reporter plasmid ([103]Fig 5D), supporting our
   hypothesis that Tan could inhibit the transcriptional activity of GLI2
   in ESCC cells.

Fig 5. Tan inhibits the expression and transcriptional activity of GLI2 in
ESCC cells.

   [104]Fig 5
   [105]Open in a new tab

   (A and B) qRT-PCR of the mRNA levels of GLI2, BARX2 and EN1 in
   Tan-treated TE-1 and KYSE150 ESCC cells (20 μg/ml, 24 h) and control
   cells (Con). (C) Western blot analysis of GLI2 protein expression in
   Tan-treated TE-1 and KYSE150 ESCC cells (20 μg/ml, 24 h) and control
   cells (Con). (D) TE-1 cells were introduced with GLI2-luc luciferase
   reporter plasmid and pRL-TK Renilla control vector and then subjected
   to Tan treatment (20 μg/ml, 24 h) or control (Con), followed by the
   evaluation of the luciferase activity. *P < 0.05, ***P < 0.001.

Tan down-regulates GPNMB expression by suppressing GLI2-mediated
transcription of GPNMB in ESCC cells

   To identify GLI2 downstream genes responsible for the anti-ESCC
   activity of Tan, we integrated these genes including the 587 genes with
   a significant variation following Tan treatment in TE-1 cells ([106]S1
   Table), the differentially expressed genes in ESCC tissues and normal
   esophageal tissues using the TCGA database ([107]S3 Table), and the
   downstream target genes of GLI2 using hTF-target database ([108]S4
   Table). Of the 24 genes that overlapped among the 3 lists ([109]Fig
   5A), we found that 8 genes (ABCC3, BMP7, TM4SF1, GPNMB, COBLL1, FRY,
   BCAT1, ENHO) were positively correlated with GLI2 expression in ESCC
   tissues ([110]Fig 6A). Furthermore, only GPNMB, BMP7, BCAT1 levels were
   markedly elevated in ESCC ([111]Fig 6B, [112]S3 Table) and were
   down-regulated following Tan treatment in TE-1 ESCC cells ([113]Fig 6B,
   [114]S1 Table). In consideration of the markedly correlation of GPNMB
   and BMP7 with P < 0.01 ([115]Fig 6A), we further confirmed the
   expression of GPNMB and BMP7 in GLI2-lossed TE-1 and KYSE150 cells. As
   expected, silencing of GLI2 resulted in reduced mRNA levels of GPNMB in
   the two cell lines ([116]Fig 6C), thereby, we selected GPNMB for
   further exploration. Western blot results also confirmed the
   down-regulation of GPNMB protein level in GLI2-silenced TE-1 and
   KYSE150 cells ([117]Fig 6D), indicating that GLI2 positively regulated
   GPNMB expression in ESCC cells.

Fig 6. GPNMB is a potential target gene of GLI2 in ESCC cells.

   [118]Fig 6
   [119]Open in a new tab

   (A and B) Venn diagram revealing the 24 genes that overlapped among the
   3 lists. Correlation clustering heat map of GLI2 and 24 target genes.
   Red “*” indicates positive correlation, and blue “*” indicates
   negative. (B) The volcano map of differentially expressed genes in ESCC
   vs normal tissue and Tan treatment vs Con. (C) qRT-PCR of the mRNA
   levels of GPNMB and BMP7 in TE-1 and KYSE150 ESCC cells after
   transfection by si-GLI2 or si-NC. (D) Western blot of GPNMB protein
   level in cells transfected as indicated. *P < 0.05, **P < 0.01, ***P <
   0.001.

   To search the binding region between GLI2 and the GPNMB promoter, we
   interrogated the JASPAR database and found a putative binding motif
   ([120]Fig 7A) and two putative binding sites at positions +(343–354)
   (named BS-1) and +(1539–1550) (named BS-2) in the GPNMB promoter
   ([121]Fig 7B). To verify this, we cloned the fragments (about 100 bp)
   of the GPNMB promoter containing the +(343–354) (named BS-1) position
   sequence or +(1539–1550) (named BS-2) position sequence into the
   pGL3-basic luciferase plasmid and transfected them into 293T cells
   together with si-GLI2. With the BS-2 reporter construct and GLI2
   silencing led to a reduction in luciferase activity ([122]Fig 7B). When
   the BS-1 reporter plasmid was introduced, no reduction in luciferase
   was observed with GLI2 silencing ([123]Fig 7B), demonstrating that GLI2
   could bind to the GPNMB promoter through the site at position
   +(1539–1550). Furthermore, ChIP experiments showed that the GPNMB
   promoter at position +(1539–1550) was preferentially enriched in
   GLI2-associating immunoprecipitates compared with IgG controls
   ([124]Fig 7C), reinforcing the binding relationship between GLI2 and
   the GPNMB promoter. All these findings indicate that GLI2 positively
   regulates GPNMB transcription by binding to the GPNMB promoter.

Fig 7. Tan inhibits GPNMB expression via GLI2-mediated transcription of GPNMB
in ESCC cells.

   [125]Fig 7
   [126]Open in a new tab

   (A) The putative binding motif between GLI2 and the GPNMB promoter
   predicted by the JASPAR database. (B) Luciferase reporter activity from
   293T cells cotransfected with the BS-2 reporter construct or the BS-1
   reporter plasmid and si-GLI2 or si-NC. (C) ChIP assay with the lysates
   of TE-1 cells using an antibody against GLI2 or IgG control. qPCR with
   specific primers was utilized to gauge the enrichment level of the
   GPNMB promoter at position +(1539–1550). (D and E) GPNMB mRNA level by
   qRT-PCR and its protein expression by western blot in Tan-treated TE-1
   and KYSE150 ESCC cells (20 μg/ml, 24 h) and control cells (Con). **P <
   0.01, ***P < 0.001.

   Having established that Tan inhibits GLI2 transcriptional activity, we
   next determined whether Tan could affect GPNMB expression in ESCC
   cells. Notably, treatment of Tan down-regulated GPNMB expression at
   both mRNA and protein in TE-1 and KYSE150 ESCC cells ([127]Fig 7D and
   7E). In summary, these observations suggest that Tan inhibits GPNMB
   expression at least in part by suppressing GLI2-mediated transcription
   of GPNMB in ESCC cells.

Re-expression of GPNMB reverses anti-migration and anti-invasion functions of
Tan in ESCC cells

   To determine whether the anti-tumor activity of Tan in ESCC is due to
   the reduction of GPNMB, a GPNMB cDNA plasmid was introduced before Tan
   treatment and assayed for cell migration and invasiveness. The efficacy
   of the cDNA plasmid in increasing GPNMB expression was validated by
   qRT-PCR in ESCC cells ([128]Fig 8A). Indeed, re-expression of GPNMB
   strongly abolished Tan-driven suppression of viability, migration, and
   invasiveness of TE-1 and KYSE150 cells ([129]Fig 8B–8E). Furthermore,
   GPNMB re-expression abated Tan-mediated reduction of PCNA and MMP9
   levels in the two cell lines ([130]Fig 9A and 9B). Collectively, these
   findings point to the notion that Tan diminishes ESCC cell migration
   and invasion partially via the suppression of GPNMB.

Fig 8. Anti-migration and anti-invasion activities of Tan in ESCC cells are
due to the down-regulation of GPNMB.

   [131]Fig 8
   [132]Open in a new tab

   (A) qRT-PCR of GPNMB mRNA in cells transfected with vec or GPNMB. (B-E)
   TE-1 and KYSE150 cells were introduced with or without the GPNMB cDNA
   plasmid or control plasmid (vec) before Tan treatment (20 μg/ml, 24 h)
   and control treatment (Con). (B) Cell viability was determined by CCK8
   assay. (C) Representative images showing a cell migration assay with
   cells treated as indicated and cell migration by wound-healing assay.
   Scale bars: 100 μm. (D and E) Representative images depicting transwell
   migration and invasion assays performed with cells treated as
   indicated. Scale bar: 100 μm. *P < 0.05, **P < 0.01, ***P < 0.001.

Fig 9. Regulation of Tan in the expression of PCNA and MMP9 in ESCC cells
through GPNMB.

   [133]Fig 9
   [134]Open in a new tab

   (A and B) TE-1 and KYSE150 cells were introduced with or without the
   GPNMB cDNA plasmid or control plasmid (vec) before Tan treatment (20
   μg/ml, 24 h) and control treatment (Con), followed by the detection of
   PCNA (A) and MMP9 (B) expression by immunofluorescence. Scale bars: 100
   μm (A) and 50 μm (B). *P < 0.05, **P < 0.01, ***P < 0.001.

Discussion

   Emerging evidence supports the notion that Tan exhibits potent
   anti-tumor efficacy against numerous human cancers [[135]5, [136]6].
   Identifying the molecular determinants underlying the activity of Tan
   has been challenging. In the current work, we first demonstrate the
   anti-cancer activity of Tan in TE-1 and KYSE150 ESCC cells.
   Furthermore, we uncover the relevance of GLI2-mediated GPNMB
   transcription to the anti-ESCC activity of Tan, providing a new
   understanding of the inhibitory mechanism of Tan in ESCC development
   and a rationale for developing Tan as a potential agent against ESCC.

   Functional studies in cancer cell lines have unveiled the
   tumor-inhibitory potential of Tan. For instance, in hepatocellular
   carcinoma HepG2 cells, Tan possesses the suppressive effects on cell
   growth and migration by elevating autophagy via the JNK/Bcl-2/beclin1
   pathway [[137]30]. In colorectal carcinoma COLO205 cells, Tan strongly
   enhances cell cycle arrest at G1 phase at least in part by diminishing
   cyclin-dependent kinases 2 (Cdk2) and 4 (Cdk4) activities [[138]31].
   Treatment of Tan in prostate cancer PC-3 cells leads to suppressed EMT
   by attenuating Vimentin expression and increasing E-cad level through
   the PI3K/Akt/mTOR pathway [[139]32]. Moreover, Tan-zinc oxide quantum
   dots perform a strong anti-metastasis property in metastatic lung
   cancer NCI-H358 cells [[140]33]. In the current report, we identify,
   for the first time, the anti-migration, anti-invasion, and anti-tumor
   functions of Tan in ESCC cells. Intriguingly, Tan (1–100 μg/ml) did not
   significantly affect the growth of normal esophageal HET-1A cells, and
   the IC50 value of Tan was 520.6 μg/ml against HET-1A cells ([141]S1
   Fig), suggesting the safety of Tan at low doses as an anti-tumor drug.

   By analyzing the transcriptional profiling of Tan-stimulated TE-1 ESCC
   cells, the GLI2 TF seems to be a candidate for the Tan target genes.
   The Sonic Hedgehog signaling has established a vital role in cancer
   biology [[142]34]. As the downstream gene of the pathway, numerous
   reports have implicated GLI2 in the regulation of human carcinogenesis
   by inducing cancer cell malignant phenotypes, suggesting the potential
   of GLI2 as a new target for cancer treatment [[143]13]. For example,
   GLI2 participates in the enhanced invasive ability of gallbladder
   cancer cells by increasing cell EMT [[144]16]. Knocking down GLI2 in
   cervical cancer cells can cause suppressed cell growth and migration
   [[145]35]. Moreover, GLI2 has been identified as a potent oncogene in
   esophageal cancer, including ESCC, and it is present at high levels in
   these diseases [[146]17, [147]18]. Our results first discovered that in
   TE-1 and KYSE150 ESCC cells, Tan down-regulates GLI2 expression by
   diminishing the transcriptional activity of GLI2. We thus speculate
   that GLI2 deregulation might be responsible for the anti-ESCC activity
   of Tan. In addition to the Hedgehog signaling, GLI2 has identified as a
   direct target of the TGF-β/SMAD pathway [[148]13]. Previous work also
   uncovered the inhibition of Tan in TGF-β1 expression in human retinal
   pigment epithelial cells [[149]36]. These conclusions suggest that Tan
   may reduce GLI2 expression by suppressing the TGF-β/SMAD pathway.
   Related field research is warranted in future work.

   TFs can modulate the transcription of the key cancer-related genes and
   thus actively participate in human carcinogenesis [[150]11]. According
   to a series of bioinformatics analyses and our siRNA experiments, GPNMB
   is identified as a target gene of GLI2. Furthermore, we first define
   that GLI2 positively regulates GPNMB transcription by binding to the
   GPNMB promoter at position +(1539–1550). GPNMB, an endogenous
   transmembrane protein, has emerged as a critical positive contributor
   to human carcinogenesis and is present at high levels in various types
   of cancers, thereby targeting GPNMB as an attractive strategy against
   cancer [[151]37, [152]38]. GPNMB functions as a pro-tumorigenic factor
   predominantly by promoting tumor migration and invasiveness and
   elevating the levels of tumorigenic factors [[153]37]. Bioinformatics
   analysis revealed the overexpression of GPNMB in ESCC. Nonetheless, no
   reports proved whether GPNMB overexpression is causally involved in
   ESCC pathogenesis. In the current work, we first demonstrate the
   inhibitory effect of Tan on GPNMB expression in TE-1 and KYSE150 ESCC
   cells. More intriguingly, our study defines the causal association
   between the reduction of GPNMB and the anti-migration and anti-invasion
   functions of Tan in TE-1 and KYSE150 ESCC cells. Some important
   metastasis-related markers, such as EMT specific molecule Snail,
   angiogenesis-related molecule VEGF, and cell cycle regulator Cyclin D1,
   have been identified as target genes of GLI2 TF [[154]39, [155]40]. Our
   data also uncovered the repression of Tan in the expression of Slug,
   Snail, VEGF, and Cyclin D1 in ESCC cells ([156]S2 Fig), suggesting that
   Tan may downregulate their expression via GLI2 and thus influences ESCC
   metastasis. Future work will build on the findings by elucidating
   whether Tan exerts an anti-metastasis function in ESCC in vivo and
   whether the novel mechanism is involved in the anti-metastasis function
   of Tan in vivo. Moreover, more studies are required to determine the
   long-term efficacy and safety of Tan in various experimental models.
   Additionally, MMP9 is a crucial regulator implicated in ESCC outcome
   and prognosis [[157]41], and its expression can be regulated by GLI2 in
   melanoma cells [[158]42]. A previous document proved that GPNMB can
   influence MMP9 expression in chronic obstructive pulmonary disease
   [[159]43]. Our data also showed that Tan can down-regulate MMP9
   expression by diminishing GPNMB level in ESCC cells. With these
   findings, we speculate that Tan in ESCC cells down-regulates MMP9
   expression by suppressing GLI2-mediated GPNMB transcription.

   Owing to the diverse pharmacological activities and minimal side
   effects, the flavonoids, including Tan, have gained much attention in
   the prevention and treatment of cancer [[160]5]. However, certain
   limitations, such as the difficult purification process and poor
   solubility, impede their clinical use [[161]5]. Therefore, it is very
   important to study how these limitations are resolved for their
   clinical use.

Conclusions

   Here, we conclude that Tan exerts anti-migration and anti-invasion
   effects in ESCC cells by down-regulating GPNMB by suppressing
   GLI2-mediated GPNMB transcription, providing a new understanding of the
   tumor-inhibitory mechanism of Tan and the basis for the development of
   Tan as a therapeutic agent against ESCC.

Supporting information

   S1 Fig. The IC50 value of Tan in normal esophageal HET-1A cells.

   (TIF)
   [162]pone.0291531.s001.tif^ (113.2KB, tif)
   S2 Fig. Effect of Tan on the expression of VEGF, Slug, Snail and Cyclin
   D1.

   (TIF)
   [163]pone.0291531.s002.tif^ (457.8KB, tif)
   S1 Table. The 587 genes with a significant variation following Tan
   treatment in TE-1 cells.

   (XLSX)
   [164]pone.0291531.s003.xlsx^ (134.6KB, xlsx)
   S2 Table. The all 1665 human transcriptional factors.

   (XLSX)
   [165]pone.0291531.s004.xlsx^ (109.9KB, xlsx)
   S3 Table. The differentially expressed genes in ESCC tissues and normal
   esophageal tissues using the TCGA database.

   (XLSX)
   [166]pone.0291531.s005.xlsx^ (595.8KB, xlsx)
   S4 Table. The genes that could be regulated by GLI2 from hTF-target
   database.

   (XLSX)
   [167]pone.0291531.s006.xlsx^ (117.4KB, xlsx)
   S1 Raw images

   (PDF)
   [168]pone.0291531.s007.pdf^ (1.5MB, pdf)

Data Availability

   All relevant data are within the paper and its [169]Supporting
   Information files.

Funding Statement

   This research was supported by Key Science and Technology Projects in
   Henan Province (No. 212102310669, No. 222102310045), and Henan Province
   Medical Science and Technology Research Plan Joint Construction Project
   (No. SBGJ202102022) and “23456 Talent Project” of Henan Provincial
   People’s Hospital.

References