Identification of a Novel GRHL3/HOPX/Wnt/β-Catenin Proto-oncogenic Axis in Squamous Cell Carcinoma of the Esophagus

Abstract

Background & aims: Esophageal squamous cell carcinoma (ESCC) is an aggressive malignancy with a poor long-term prognosis. The molecular mechanisms underlying the initiation and progression of this tumor are largely unknown. The transcription factor GRHL3 functions as a potent tumor suppressor in SCC of skin, head, and neck. This study aims to determine whether GRHL3 also plays a role in the homeostasis of the esophageal epithelium and in the development of ESCC.

Methods: The effects of Grhl3 deletion on squamous epithelial homeostasis in embryos and adult mice were examined using immunohistochemistry, transmission electron microscopy, and real-time polymerase chain reaction. The conditionally deleted mice were subsequently used to determine susceptibility to ESCC. Whole-transcriptome sequencing (RNA-seq) was performed on ESCC in wild-type and Grhl3 deleted animals. To decipher the signaling pathways, real-time polymerase chain reaction, immunohistochemistry, analysis of chromatin immunoprecipitation sequencing, chromatin immunoprecipitation-polymerase chain reaction, and RNA seq datasets were used. Primary human samples were used to validate the findings in the mouse model.

Results: Loss of Grhl3 perturbs the proliferation-differentiation balance in the esophageal epithelium, thereby increasing the susceptibility to esophageal carcinogenesis in adult mice. Grhl3 imparts its tumor suppressor function by regulating the expression of HOPX. We have identified the Wnt/β-catenin pathway as the downstream effectors of GRHL3 and HOPX through our integrated approach using patient-derived ESCC samples and mouse models.

Conclusions: GRHL3 conveys its tumor suppressor function in ESCC through regulating its target gene HOPX, which limits Wnt/β-catenin signaling. Targeted therapies to inhibit this pathway could be a potential treatment strategy for ESCC patients with reduced GRHL3 expression.

Keywords: Cancer Signaling Pathways; RNA Sequencing; Transcriptional Regulation.

Graphical abstract

Figure 1

Grhl3 deletion during embryogenesis causes hyperproliferation of the esophageal epithelium. (A and B) Representative images of transverse sections of the middle esophagus of E18.5 wild-type (Grhl3^+/+) and Grhl3-knockout (Grhl3^-/-) mice (n = 3 each) immunohistochemically analyzed with β-galactosidase and Grhl3. Scale bar = 50 μm. (C) Representative H&E images of Grhl3^+/+ and Grhl3^-/- mice middle esophagus showing increased thickness of esophageal epithelium in Grhl3^-/- mice. Scale bar = 50 μm. (D) Esophageal epithelial thickness was quantified from H&E images of E18.5 Grhl3^+/+ and Grhl3^-/- mice (n = 3 each) using ImageJ software. Three sections representing cervical, thoracic, and abdominal areas were examined, and 4 points per section were measured to calculate the average thickness of the esophagus. Statistics were calculated using unpaired, two-tailed Student t test. (E) Representative transmission electron micrograph of E18.5 Grhl3^+/+ and Grhl3^-/- mice middle esophagus (n = 3 each). Nucleated suprabasal squamous epithelium shown in Grhl3^-/- embryo esophagus. Scale bar = 50 μm. (F) Graph shows number of Ki67 positive cells in the esophagus in E18.5 Grhl3^+/+ and Grhl3^-/- mice (n = 3 each), enumerated from micrographs immunohistochemically analyzed with Ki67 (representative image shown, G). Statistics were calculated using unpaired, two-tailed Student t test (∗P < .05). (H) Representative immunohistochemical images of E18.5 Grhl3^+/+ and Grhl3^-/- mice middle esophagus (n = 3 each) stained with caspase antibody. Scale bar = 50 μm.

Figure 2

Grhl3 deletion during embryogenesis causes perturbation of proliferation-differentiation equilibrium of esophageal epithelium. Representative immunohistochemical images of E18.5 Grhl3^+/+ and Grhl3^-/- mice middle esophagus (n = 3 each) stained with K6 (A), K8 (B), K5 (C), K14 (D), K13 (E), and K4 (F) antibodies. Scale bar = 50 μm.

Figure 3

Grhl3 deletion is a driver of esophageal squamous cell carcinoma. (A) Transverse sections of middle esophagus of Grhl3 heterozygous mice (Grhl3^+/-) showing β-galactosidase expression driving from LacZ reporter gene using X-gal staining and antibodies against β-galactosidase. Scale bar = 50 μm. (B) The mRNA expression of Grhl3 was detected by quantitative real-time PCR from esophageal epithelium collected from WT (n = 3) and Grhl3-cKO mice (n = 6). Relative gene expression was calculated using HPRT as the housekeeping gene. Data were represented as mean ± standard deviation of replicates. Statistics were calculated using unpaired, two-tailed Student t test. (Inset) Genomic DNA from the esophageal epithelium of 12-week-old WT and Grhl3-cKO mice were analyzed using PCR to detect deletion of floxed Grhl3 allele. The undeleted (flox) band of 425 base pairs and the deleted (Δ) band of 282 base pairs are indicated. (C) Representative H&E images showing moderate hyperplasia of esophageal mucosa in Grhl3-cKO mice compared with WT mice. Scale bar = 50 μm. (D) Graph shows number of Ki67 positive cells in the esophagus in WT (n = 5) and Grhl3-cKO mice (n = 3), enumerated from micrographs immunohistochemically analyzed with Ki67 antibody (representative image shown). Statistics were calculated using unpaired, two-tailed Student t test. (E) Representative immunohistochemical images of WT and Grhl3-cKO mice esophagus (n = 3 each) stained with K6 antibody. Scale bar = 50 μm. (F) Representative immunohistochemical images of WT and Grhl3-cKO mice esophagus (n = 3 each) stained with K14 antibody. Scale bar = 50 μm. (G) Kaplan-Meier survival curve of esophageal tumor-free survival in WT (n = 10) and Grhl3-cKO mice (n = 10). Mice were euthanized upon losing 20% of their body weight. P value from the log-rank (Mantel-Cox) test is .04. (H) Representative H&E images of invasive SCC of the esophagus and occluding squamous papilloma in mid esophageal area of Grhl3-cKO mice. Scale bar = 50 μm. (I) Graph representing classification of esophageal lesions in WT (n = 10) and Grhl3-cKO mice (n = 10). (J) The mRNA expression of Grhl3 was detected by quantitative real-time PCR from esophageal tumors collected from WT (n = 10) and Grhl3-cKO mice (n = 8). Relative gene expression was calculated using HPRT as the housekeeping gene. Data were represented as mean ± standard deviation of replicates. Statistics were calculated using unpaired, two-tailed Student t test (∗∗P < .01).

Figure 4

Loss of GRHL3 induces transcriptome changes in esophageal SCC. (A) Heat map of RNA-seq transcriptome analysis showing differential gene expression in ESCC from WT and Grhl3-cKO mice. Sixty-one genes were up-regulated, and 75 genes were down-regulated (false discovery rate [FDR] <0.05) in Grhl3-cKO ESCC compared with its WT counterparts. (B) List of genes that are differentially expressed more than 2-fold in Grhl3-cKO mice compared with WT mice. (C) Graph shows enriched ontology clusters in differentially expressed genes in Grhl3-cKO mice using Metascape analysis.

Figure 5

Loss of GRHL3 induces transcriptome changes in esophageal SCC. (A) The mRNA expression of selected genes by Q-RT-PCR from esophageal tumor samples collected from WT and Grhl3-cKO mice. Relative gene expression was calculated using HPRT as the housekeeping gene. Data were represented as mean ± standard deviation of 3–5 replicates. Statistics were calculated using multiple unpaired t test using GraphPad prism (∗P < .05). (C) The mRNA expression of selected genes by Q-RT-PCR from esophageal epithelium collected from WT and Grhl3-cKO mice. Relative gene expression was calculated using HPRT as the housekeeping gene. Data were represented as mean ± standard deviation of 4–8 replicates. Statistics were calculated using multiple unpaired t test using GraphPad prism (∗P < .05). (C) Table showing putative Grhl3 target genes that are differentially expressed in Grhl3-cKO mice.

Figure 6

HOPX is a direct target of GRHL3. (A) Location of GRHL3 binding site in the human HOPX genome. Blue arrowhead indicates start codon. (B) Alignment of GRHL3 binding site (red) in the indicated species showing conservation of the site in multiple species. (C) ChIP-seq peak in the promoter region of human HOPX genome from GRHL3 ChIP performed on primary human keratinocytes (NHEK, GSE95199), aligned in the hg19 UCSC genome assembly. (D) ChIP-seq peak in the promoter region of mouse Hopx genome from GRHL3 ChIP performed on E16.5 mouse back skin (GSE52648), aligned in the UCSC mm9 genome assembly. (E) ChIP-Q-PCR from EPC cells demonstrating more than 2-fold enrichment over immunoglobulin G (IgG) in the binding site of HOPX promoter (n = 4 replicates). Statistics were calculated using unpaired, one-tailed Student t test.

Figure 7

Reduced expression of HOPX activates Wnt signaling pathway in esophageal epithelium. (A) Representative immunohistochemical images of middle esophagus from WT E18.5 and Grhl3KO embryo esophagus (n = 3 each) stained with HOPX antibody. Scale bar = 50 μm. (B) Representative immunohistochemical images of middle esophagus from WT and Grhl3-cKO adult esophagus (n = 3 each) stained with HOPX antibody. Scale bar = 50 μm. (C) The mRNA expression of Hopx was detected by Q-RT-PCR from esophagus collected from WT (n = 3) and Grhl3KO mice (n = 6). Relative gene expression was calculated using HPRT as the housekeeping gene. Data were represented as mean ± standard deviation. Statistics were calculated using unpaired, two-tailed Student t test. (D) The mRNA expression of indicated genes by Q-RT-PCR from esophagus collected from WT (n = 4) and Grhl3KO mice (n = 4). Relative gene expression was calculated using HPRT as the housekeeping gene. Data were represented as mean ± standard deviation. Statistics were calculated using multiple unpaired t test using GraphPad prism. (E) The mRNA expression of indicated genes by Q-RT-PCR from esophageal epithelium collected from WT (n = 6) and Grhl3-cKO mice (n = 6). Relative gene expression was calculated using HPRT as the housekeeping gene. Data were represented as mean ± standard deviation. Statistics were calculated using multiple unpaired t test using GraphPad Prism. (F) The mRNA expression of indicated genes by Q-RT-PCR from esophageal tumors collected from WT (n = 5–6) and Grhl3-cKO mice (n = 5–6). Relative gene expression was calculated using HPRT as the housekeeping gene. Data were represented as mean ± standard deviation. Statistics were calculated using multiple unpaired t test using GraphPad Prism. (G) ChIP-Q-PCR from EPC cells demonstrating more than 4-fold enrichment over immunoglobulin G (IgG) in WNT2 and WNT9a promoter (n =3 replicates). Statistics were calculated using unpaired, one-tailed Student t test.

Figure 8

Wnt pathway is dysregulated esophageal SCC samples. (A) Representative immunohistochemical images of WT and Grhl3-cKO mice esophageal tumors (n = 3 each) stained with CTNNB antibody. Scale bar = 50 μm. (B) The mRNA expression of indicated genes by Q-RT-PCR from esophageal tumors collected from WT (n = 8–9) and Grhl3-cKO mice (n = 8–9). Relative gene expression was calculated using HPRT as the housekeeping gene. Data were represented as mean ± standard deviation. Statistics were calculated using multiple unpaired t test using GraphPad Prism.

Figure 9

GRHL3 and HOPX expression is reduced in patient-derived esophageal SCC samples leading to activation of Wnt signaling pathway. (A) GRHL3 mRNA expression in 51 oral squamous cell carcinoma (OSCC) samples compared with its own paired normal tissue, calculated from GSE23400. Statistics were calculated using unpaired, two-tailed Student t test, and P values are indicated on the graph. (B) HOPX mRNA expression in 51 OSCC samples compared with its own paired normal tissue, calculated from GSE23400. Statistics were calculated using unpaired, two-tailed Student t test, and P values are indicated on the graph. (C) Correlation between GRHL3 and HOPX mRNA expression in patient samples calculated from GSE23400. Pearson r method using GraphPad prism is used to calculate the correlation between the mRNA expression. (D) WNT2, WNT9A, and MET mRNA expression in 51 OSCC samples compared with its own paired normal tissue, calculated from GSE23400. Statistics were calculated using unpaired, two-tailed Student t test, and P values are indicated on the graph. (E) Representative immunohistochemical images of 3 human patient esophageal samples immunohistochemically stained with CTNNB. Scale bar = 50 μm. (F) Representative immunohistochemical images of human patient esophageal samples immunohistochemically stained with GRHL3, HOPX, and WNT9A. Scale bar = 50 μm.