EYA2 suppresses the progression of hepatocellular carcinoma via SOCS3-mediated blockade of JAK/STAT signaling

Abstract

Background: Somatic mutations are involved in hepatocellular carcinoma (HCC) progression, but the genetic mechanism associated to hepatocarcinogenesis remains poorly understood. We report that Eyes absent homolog 2 (EYA2) suppresses the HCC progression, while EYA2(A510E) mutation identified by exome sequencing attenuates the tumor-inhibiting effect of EYA2.

Methods: Whole-exome sequencing was performed on six pairs of human HCC primary tumors and matched adjacent tissues. Focusing on EYA2, expression level of EYA2 in human HCC samples was evaluated by quantitative real-time PCR, western blot and immunohistochemistry. Loss- and gain-of-function studies, hepatocyte-specific deletion of EYA2 (Eya2^-/-) in mice and RNA sequencing analysis were used to explore the functional effect and mechanism of EYA2 on HCC cell growth and metastasis. EYA2 methylation status was evaluated using Sequenom MassARRAY and publicly available data analysis.

Results: A new somatic mutation p.Ala510Glu of EYA2 was identified in HCC tissues. The expression of EYA2 was down-regulated in HCC and associated with tumor size (P = 0.001), Barcelona Clinic Liver Cancer stage (P = 0.016) and tumor differentiation (P = 0.048). High level of EYA2 was correlated with a favorable prognosis in HCC patients (P = 0.003). Results from loss-of-function and gain-of-function experiments suggested that knockdown of EYA2 enhanced, while overexpression of EYA2 attenuated, the proliferation, clone formation, invasion, and migration of HCC cells in vitro. Delivery of EYA2 gene had a therapeutic effect on inhibition of orthotopic liver tumor in nude mice. However, EYA2(A510E) mutation led to protein degradation by unfolded protein response, thus weakening the inhibitory function of EYA2. Hepatocyte-specific deletion of EYA2 in mice dramatically promoted diethylnitrosamine-induced HCC development. EYA2 was also down-regulated in HCC by aberrant CpG methylation. Mechanically, EYA2 combined with DACH1 to transcriptionally regulate SOCS3 expression, thus suppressing the progression of HCC via SOCS3-mediated blockade of the JAK/STAT signaling pathway.

Conclusions: In our study, we identified and validated EYA2 as a tumor suppressor gene in HCC, providing a new insight into HCC pathogenesis.

Keywords: Eyes absent homolog 2; JAK/STAT signaling pathway; Somatic mutation; Tumor suppressor gene; Unfolded protein response; Whole-exome sequencing.

Fig. 1

EYA2 mutations in HCC. (A) Somatic mutations identified in EYA2 gene, which were confirmed by IGV visualization and Sanger sequencing of DNA samples from HCC and case-matched adjacent tissues. The black arrows indicate the somatic mutation sites. (B) Prediction of the functional effects of R255K and A510E substitutions (R, Arg; K, Lys; A, Ala; E, Glu). (C) Multiple sequence alignments of EYA2 paralogs from different species. The mutation sites in EYA2 are indicated by black arrows. (D) Genomic organization of EYA2 exon regions and protein domain structure. The EYA2 somatic mutations in HCC are indicated by red and black arrows which represent our research findings and COSMIC database, respectively. The asterisk represents nonsense mutation. (E) Somatic mutation frequency of EYA2 in liver cancer in the population of different countries obtained from the ICGC database

Fig. 2

Expression and prognostic significance of EYA2 in HCC tissues. (A) EYA2 expression in HCC and adjacent tissues according to the three datasets GES22058, GSE14520 and ICGC_LIRI. (B) Western blot analysis of EYA2 expression in HCC and adjacent tissues. T: HCC tissues, N: Adjacent tissues. (C) A representative immunohistochemical staining of EYA2 in HCC and case-matched adjacent tissue. (D) Immunohistochemistry scores associated to the EYA2 expression in HCC and case-matched adjacent tissues (n = 94 pairs). Kaplan-Meier analysis of overall survival (E) and recurrence-free survival (F) of HCC patients with high or low EYA2 expression (n = 94 cases). **P < 0.01, ***P < 0.001. (A, B, D) mean ± SEM, Student’s t-tests; (E, F) Kaplan-Meier analysis

Fig. 3

EYA2 suppresses the malignant phenotypes of HCC cells in vitro and in vivo. (A) Western blot analysis of the expression of EYA2 in HCC cells transiently transfected with siRNA or with the overexpression and mutant vectors. The effect of siRNA-mediated knockdown of EYA2 and pcDNA3.1-mediated overexpression of constructs encoding EYA2 wild-type, EYA2(R255K) and EYA2(A510E) on cell proliferation (B), cell invasion (C), clone formation (D) and cell migration (E) of HCC cells in vitro. (F) The effect of EYA2 knockdown, EYA2 wild-type and EYA2(A510E) mutant on HCC tumor growth in nude mice. (G) The scheme of treatment of orthotopic HCC tumors in nude mice by lentivirus-EYA2. (H) Images, volume and weight of the orthotopic liver tumors of the mice. (I) Expression of EYA2 in tumor tissues detected by immunohistochemistry analysis. ns, not significant, *P < 0.05, **P < 0.01, ***P < 0.001. (B–F, H–I) mean ± SEM, Student’s t-tests

Fig. 4

Aberrant CpG methylation of the first intron region of EYA2 down-regulates the expression of mRNA in HCC. (A) mRNA expression of EYA2 in four HCC cell lines by qRT-PCR after treatment with 5-aza-2′-deoxycytidine for 12 h compared with DMSO-treated cells. (B) The methylation level of EYA2 CpG islands in HCC and adjacent tissue (left), and negative correlation between methylation level of EYA2 CpG islands and EYA2 mRNA expression based on MethHC database (right). (C) The location of CpG in EYA2 intron 1 and PCR primers used for the methylation analysis. (D) Profiling of the methylation levels of CpG sites in EYA2 intron 1 presented as a dot chart. (E) Average methylation level of each CpG site in HCC and adjacent tissues (n = 30 pairs). (F) mRNA expression of EYA2 (left) and 10 significantly different CpG methylation sites (right) in 30 paired HCC tissues and adjacent tissues. (G) Correlation analysis between mean methylation of 10 significantly different CpGs and corresponding mRNA expression of EYA2. *P < 0.05, **P < 0.01, ***P < 0.001. (A, B, E, F) mean ± SEM, Student’s t-tests; (B, G) Pearson’s correlation test

Fig. 5

EYA2(A510E) mutation leads to lysosomal or proteasomal degradation by triggering the unfolded protein response. (A) EYA2(A510E) shortened the half-life of EYA2 protein. EYA2 and EYA2(A510E) stably transfected HEK293T cells were treated with 100 μg/ml cycloheximide (CHX) and protein lysates were collected at the indicated times for western blot analysis. (B) Confocal immunofluorescence analysis of the expression of Flag-tagged EYA2(A510E) and Flag-tagged EYA2. (C) Structural prediction of EYA2(A510E) mutation. (D) Western blot analysis of the expression of UPR-related proteins in EYA2(A510E) mutant-expressed Huh-7 and MHCC-97H cells. Western blot analysis of EYA2 expression in EYA2(A510E) stably transfected HEK293T cells treated with chloroquine (E) or MG132 (F). (G) Immunoprecipitation analysis of the ubiquitination of EYA2(A510E) mutation. Western blot densitometry was performed from three independent experiments. EYA2 was used as the normalization control. (H) Colocalization of Flag-tagged EYA2(A510E) and lysosomes in the presence of chloroquine detected by confocal immunofluorescence. The arrows point the colocalization of Flag-tagged EYA2(A510E) and Lysosomes-RFP. (I) Colocalization of Flag-tagged EYA2(A510E) and proteasome-20S in the presence of MG132 detected by confocal immunofluorescence. The arrows point the colocalization of Flag-tagged EYA2(A510E) and proteasome-20S. ns: not significant. (G) mean ± SEM, Student’s t-tests

Fig. 6

EYA2 interacts with DACH1 to suppress HCC progression via SOCS3-mediated blockade of JAK/STAT signaling. (A) Scatter plot showing the differentially expressed genes regulated by the overexpression of EYA2. (B) Heat map showing the representative differentially expressed genes modulated by the overexpression of EYA2. (C) KEGG pathway analysis showing the most enriched pathways of the differentially expressed genes. (D) GSEA showing the enrichment of JAK/STAT-related gene signatures in the EYA2 overexpression cells. (E) Co-immunoprecipitation of endogenous DACH1 with anti-EYA2 antibody (upper) and endogenous EYA2 with anti-DACH1 antibody (lower) in MHCC-97H and Huh-7 cells. (F) SOCS3 promoter constructs (− 2000/− 1) co-transfected with pcDNA3.1-EYA2 or/and GV141-DACH1 and the relative luciferase activity measured in HEK293T cells. (G) Expression of SOCS3, p-STAT3, p-JAK2, STAT3 and JAK2 in Hep3B and Huh-7 cells co-transfected with GV141-DACH1/sh-EYA2 or si-DACH1/pcDNA3.1-EYA2 detected by western blot. (H) Western blot analysis of expression of STAT3, p-STAT3, JAK2 and p-JAK2 in Hep3B and Huh-7 cells co-transfected with GV492-SOCS3/sh1-EYA2 or si1-DACH1/GV492-SOCS3. (I) Scheme of Eya2^−/− mice induced with DEN. (J) Images of the liver of Eya2^−/− and Eya2^+/+ mice treated with DEN at 7, 9 and 11 months. (K) H&E staining and immunohistochemistry analysis of the expression of EYA2, SOCS3 and PCNA in liver tissues from mice treated with DEN at 7, 9 and 11 months. T indicates tumor nodule. Number of tumors (L) and tumor sizes (M) from mice treated with DEN at 9 and 11 months. qRT-PCR analysis of EYA2 (N) and SOCS3 (O) expression in liver tissues from mice treated with DEN at 7, 9 and 11 months (n = 3 for each group). *P < 0.05, **P < 0.01, ***P < 0.001. (F, L, M, N, O) mean ± SEM, Student’s t-tests

Fig. 7

A proposed model for the tumor-suppressive activity of EYA2