CDX2 regulation by the RNA-binding protein MEX3A: impact on intestinal differentiation and stemness

Abstract

The homeobox transcription factor CDX2 plays a crucial role in intestinal cell fate specification, both during normal development and in tumorigenic processes involving intestinal reprogramming. The CDX2 regulatory network is intricate, but it has not yet been fully uncovered. Through genome-wide screening of a 3D culture system, the RNA-binding protein MEX3A was identified as putatively involved in CDX2 regulation; therefore, its biological relevance was addressed by setting up cell-based assays together with expression studies in murine intestine. We demonstrate here that MEX3A has a repressive function by controlling CDX2 levels in gastric and colorectal cellular models. This is dependent on the interaction with a specific binding determinant present in CDX2 mRNA 3'untranslated region. We have further determined that MEX3A impairs intestinal differentiation and cellular polarization, affects cell cycle progression and promotes increased expression of intestinal stem cell markers, namely LGR5, BMI1 and MSI1. Finally, we show that MEX3A is expressed in mouse intestine, supporting an in vivo context for interaction with CDX2 and modulation of stem cell properties. Therefore, we describe a novel CDX2 post-transcriptional regulatory mechanism, through the RNA-binding protein MEX3A, with a major impact in intestinal differentiation, polarity and stemness, likely contributing to intestinal homeostasis and carcinogenesis.

Figure 1.

Establishment of AGS 3D cultures and characterization of CDX2 expression. (A) Schematic representation of the culture systems and cellular morphology at culture day 4 in bright field microscopy (original magnification, ×100). (B) Western blot and qPCR of CDX2 expression in 2D and 3D after 2 weeks culture. Cardinal numbers represent biological replicates. Values for CDX2 mRNA expression in 2D culture were referred to as 1. (C) Western blot of CDX2 protein expression on treatment with MG132. (D) qPCR of CDX2 mRNA expression on treatment with actinomycin D. Expression levels in the absence of treatment were set at 100%. Depicted half lives were calculated using exponential regression (**P = 0.003).

Figure 2.

Transcriptome profiling of 2D and 3D AGS cultures. (A) Microarray heat map displaying differentially expressed transcripts (P < 0.01, fold change > 1.5). Values obtained for MEX3A are highlighted. (B) Functional annotation analysis depicting the preponderance of differentially expressed genes in terms of signaling networks (P < 0.05, fold change > 1.5). (C) qPCR validation of MEX3A expression. Values for MEX3A mRNA expression in 2D culture were referred to as 1.

Figure 3.

CDX2 regulation by MEX3A in AGS cells. (A) Western blot of transient transfections with a myc-tagged MEX3A expression vector. (B) qPCR of CDX2 and MEX3A mRNA expression in the previous samples. The value for CDX2 mRNA expression in the empty vector transfected cells at 24 h was referred to as 1. (C) Immunofluorescence for MEX3A and CDX2 in MEX3A stably transfected cells at 48 h culture (original magnification,×400; scale bar 20 µm). (D) Western blot of MEX3A and CDX2 expression in stably transfected cells at different time points.

Figure 4.

Mechanistics of MEX3A interaction with CDX2 mRNA in AGS cells. (A) Terminal sequence of CDX2 mRNA, with the predicted MRE in the 3′UTR highlighted in bold and underlined. (B) qPCR showing CDX2 and GAPDH mRNAs immunoprecipitated with myc-tag antibody in MEX3A transfected cells (*P = 0.018). The values for CDX2 and GAPDH mRNA levels in the IgG sample were referred to as 1. (C) Schematic representation of pRL constructs. (D) qPCR showing Rluc mRNA immunoprecipitated with myc-tag antibody for pRLCDX2 transfection (****P < 0.0001). Values for Rluc mRNA expression in IgG samples were referred to as 1. (E) Luciferase activity assay for the different pRL constructs (**P = 0.006 for pRLControl/pRLCDX2 and *P = 0.036 for pRLCDX2/pRLΔCDX2). The values obtained for luciferase expression in the pRLControl co-transfected cells were referred to as 1.

Figure 5.

MEX3A modulation of intestinal phenotype in Caco-2 cells. (A) Western blot of MEX3A inhibition with transfection of siRNAs performed at day −3 and day −1. (B) qPCR of CDX2 mRNA expression in the same samples. The value for CDX2 at day −2 for the siControl sample was referred to as 1. (C) Western blot of MEX3A stably transfected cells at different confluences for differentiation and proliferation markers. (D) Flow cytometry analysis of MEX3A transiently transfected cells. Dot plots depicting negative control (secondary antibody only) and FITC-conjugated MEX3A expression levels are shown on top. DNA content histograms showing population percentages for the different cell cycle phases of the empty vector transfected cells and MEX3A-transfected cells (gate MEX3A exogenous) are shown below (*P = 0.03 for G0/G1 phase and *P = 0.05 for S phase).

Figure 6.

Expression profiles of intestinal stem cell markers in Caco-2 cells. (A) qPCR of LGR5 mRNA expression in mock and MEX3A overexpressing cells. The value for LGR5 at day −3 for the mock sample was referred to as 1. (B) qPCR of BMI1 mRNA expression in the same samples. The value for BMI1 at day −3 for the mock sample was referred to as 1. (C) Western blot analysis of MSI1 expression in the same samples and in Caco-2 parental cell line with CDX2 inhibition by siRNAs at the confluence time point.

Figure 7.

Polarity alterations induced by MEX3A in Caco-2 cells. (A) Immunofluorescence showing ZO-1 expression in Caco-2 mock and MEX3A stably transfected cell lines at day −2 of culture (original magnification,×400). (B) Morphology of Caco-2 mock and MEX3A cysts in bright field microscopy during 3D culture (original magnification, ×100) and CDX2/MEX3A expression (original magnification, ×630). (C) Quantification of cysts with lumen or no lumen at culture day 8. (D) Expression of E-cadherin and Phalloidin staining in Caco-2 cysts (original magnification, ×630; all scale bars 20 µm).

Figure 8.

Expression patterns of MEX3A and CDX2 in mouse normal intestine. Representative immunofluorescence data for MEX3A and CDX2 in small intestine and colon are shown (original magnification, ×200; scale bars 50 µm; insert original magnification, ×400).