FoxE1 is a thyroid-specific forkhead transcription factor essential for thyroid gland development, as well as for the maintenance of the thyroid differentiated state in adults. FoxE1 recognizes and binds to a short DNA sequence present in thyroglobulin (Tg) and thyroperoxidase (Tpo) promoters, but FoxE1 binding to regulatory regions other than Tg and Tpo promoters remains almost unexplored. Improving knowledge of the regulatory functions of FoxE1 is necessary to clarify its role in endocrine syndromes and cancer susceptibility.
In order to further investigate downstream FoxE1 targets, we performed a genome-wide expression screening after knocking-down FoxE1 and obtained new insights into FoxE1 transcriptional networks in thyroid follicular cells. After validation, we confirmed Adamts9, Cdh1, Duox2 and S100a4 as upregulated genes and Casp4, Creld2, Dusp5, Etv5, Hsp5a, Nr4a2 and Tm4sf1 as downregulated genes when FoxE1 was silenced. In promoter regions of putative FoxE1-regulated genes and also in the promoters of the classical thyroid genes Nis, Pax8 and Titf1, we performed an in silico search of the FoxE1 binding motif that was in close proximity to the NF1/CTF binding sequence, as previously described for other forkhead factors. Using chromatin immunoprecipitation we detected specific in vivo FoxE1 binding to novel regulatory regions in two relevant thyroid genes, Nis and Duox2. Moreover, we demonstrated simultaneous binding of FoxE1 and NF1/CTF to the Nis upstream enhancer region, as well as a clear functional activation of the Nis promoter by both transcription factors.
In search for potential downstream mediators of FoxE1 function in thyroid cells, we identified two novel direct FoxE1 target genes. To our knowledge, this is the first evidence regarding the implication of Nis and Duox2 in executing the transcriptional program triggered by FoxE1. Furthermore, this study points out the important role of FoxE1 in the regulation of a large number of genes in thyroid cells.
Citation: Fernández LP, López-Márquez A, Martínez ÁM, Gómez-López G, Santisteban P (2013) New Insights into FoxE1 Functions: Identification of Direct FoxE1 Targets in Thyroid Cells. PLoS ONE 8(5): e62849. https://doi.org/10.1371/journal.pone.0062849
Editor: Alfredo Fusco, Consiglio Nazionale delle Ricerche (CNR), Italy
Received: December 20, 2012; Accepted: March 26, 2013; Published: May 13, 2013
Copyright: © 2013 Fernández et al. This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.
Funding: This work was supported by Grants BFU-2010-16025 from the Dirección General de Proyectos de Investigación; RD06/0020/0060 and RD12/0036/0030 from FIS, Instituto de Salud Carlos III; and S2011/BMD-2328 TIRONET project from the Comunidad de Madrid (Spain). LP Fernández holds a postdoctoral grant of the Juan de la Cierva programme of the Spanish Government. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.
Competing interests: The authors have declared that no competing interests exist.
Coordinated expression of thyroid transcription factors Pax8, FoxE1/Ttf2 and Ttf1/Nkx2-1 is essential for maintaining the differentiated thyroid function, which involves synthesis and secretion of thyroid hormones. These factors are encoded by genes with paired box, forkhead box and homeobox domains, respectively. Thyroid hormones are iodinated, and therefore thyroid cells actively concentrate iodide through a sodium dependent co-transporter, Nis, a glycoprotein located in the basal membrane. The iodide is transported to the apical membrane, where thyroperoxidase (Tpo) iodinates the tyrosine residues of the main thyroid protein thyroglobulin (Tg) that serves as a storage for thyroid hormones , .
FoxE1, formerly known as thyroid transcription factor 2 or Ttf2, is a thyroid-specific transcription factor that belongs to the forkhead/winged-helix family . Fox proteins are a superfamily of evolutionarily conserved transcriptional regulators, which share a highly conserved forkhead box or winged helix DNA binding domain. Forkhead factors control a wide range of biological processes, and some of them are key regulators of embryogenesis and play important roles in cell differentiation and development, hormone responsiveness and aging , .
FoxE1, as a member of the Fox family, is able to interact with nucleosomes through its winged-helix DNA binding domain and to alter chromatin structure, creating a locally exposed domain necessary for the action of other transcription factors . This intrinsic property defines FoxE1 as a pioneer transcription factor , essential during thyroid development and differentiation, as well as for the maintenance of the thyroid differentiated state in adults .
Foxe1-null mice exhibit an ectopic (50%) or completely absent (50%) thyroid gland and severe cleft palate. Consequently, FoxE1 seems to be crucial either for the migration of the thyroid cell precursors or for repressing differentiation until migration has completed .
Human FOXE1 mutations cause the Bamforth–Lazarus syndrome (OMIM 241850), which is associated with congenital hypothyroidism, cleft palate and spiky hair, with or without choanal atresia, bifid epiglottis and ocular hypertelorism , . Moreover, FOXE1 variations have been associated with susceptibility to several types of cancer , , , including papillary thyroid cancer , , .
FoxE1 was initially identified as a nuclear protein  that recognizes and binds to DNA sequences present in the promoters of two thyroid-specific genes: thyroglobulin (Tg)  and thyroperoxidase (Tpo) . FoxE1 usually promotes transcriptional activation of the Tg and Tpo genes; however, it can also act as a promoter-specific transcriptional repressor of both genes .
Putative FoxE1-binding sites previously identified in the Tg and Tpo promoters share the core sequence AAACA . Moreover, in the Tpo promoter FoxE1 forms part of an interaction-complex together with the transcription factor NF1/CTF, whose final result is to turn on the expression of the Tpo gene in response to external hormonal stimuli .
Nevertheless, FoxE1 binding to DNA sequences other than the Tg and Tpo promoters remains almost unexplored. Only two studies have reported other FoxE1 targets, but both were conducted in heterologous expression systems , .
In order to further investigate FoxE1 downstream targets in thyroid epithelial cells, we performed a genome-wide screening using expression arrays in FoxE1 knock-down cells followed by a search of direct target genes containing in their promoters both FoxE1 and NF1/CTF binding sites. The results obtained in this study provide new insights into FoxE1 transcriptional networks in differentiated thyroid cells and predict involvement of FoxE1 in relevant biological processes and pathways. These data may lead to a better understanding of thyroid biology.
Materials and Methods
PCCl3 cells, a continuous line of rat thyroid follicular cells , were cultured in Coon’s modified Ham’s F-12 medium supplemented with 5% donor calf serum and a six-hormone mixture . These cells express thyroid-specific genes (Tg, Tpo, and Nis) as well as thyroid-specific transcription factors (Ttf1, FoxE1, and Pax8)  and therefore constitute a good model system to study differentiation and growth regulation in a thyroid epithelial cell setting.
HeLa cells were grown in Dulbecco’s modified Eagle’s medium (DMEM) supplemented with 5% fetal bovine serum.
PCCl3 cells were transfected with 25 nM of FoxE1 siRNA (Rat FoxE1 ON-TARGETplus SMARTpool) or with scrambled siRNA (ON-TARGETplus Non-targeting Pool) using DharmaFECT 1 Transfection Reagent and following the manufacturer’s protocol (Dharmacon, Denver, USA).
In order to establish optimal conditions for FoxE1 silencing, duplicate samples were harvested at different time points (24, 48 and 72 h) after transfection, and total protein was extracted. FoxE1 silencing was tested by western blotting using a polyclonal FoxE1 antibody (Biopat, Milan, Italy) (data not shown). Once 48 hours was defined as the best time point for FoxE1 silencing, we performed additional transfections, four biological replicates for each condition, to isolate total protein and total RNA using TRIzol reagent (Invitrogen, Carlsbad, CA) following the manufacturer’s recommended protocol.
FoxE1-dependent gene expression was tested using expression arrays (Agilent SurePrint Rat 60 K). We established two main comparisons: FoxE1-silenced PCCl3 cells (siFoxE1 PCCl3) vs. scrambled siRNA-treated PCCl3 cells (siScramble PCCl3) and FoxE1-silenced PCCl3 cells vs. wild type PCCL3 cells (wt PCCl3). This last condition was included to consistently analyse expression array signals in basal cellular conditions. Each comparison was performed using quadruplicates and dye swaps (Experimental design shown in Fig. 1A).
Panel A: Experimental design of expression array experiments. Two comparisons were done: FoxE1-silenced PCCl3 cells (siFoxE1 PCCl3) vs. scrambled siRNA-treated PCCl3 cells (siScramble PCCl3), and FoxE1-silenced PCCl3 cells vs. wild type PCCL3 cells (wt PCCl3). Each comparison was performed in quadruplicate and using dye swaps. Panel B: Western Blot of extracts from control, siScramble and siFoxE1-treated cells from quadruplicate samples used for microarray analysis. Hybridizations were done with anti-FoxE1 antibodies; anti-tubulin antibodies were used as loading controls.
Ten µg of total RNA for each condition were sent to the Genomics Core Unit of the Spanish National Cancer Research Centre (CNIO, Madrid) for RNA quality evaluation, amplification, labelling and hybridization to Agilent SurePrint Rat 60 K arrays according to the manufacturer’s protocols.
Signal quantification was carried out with Agilent Feature Extraction Software 10.7 (Agilent Technologies, Palo Alto, CA), using default analysis parameters for Agilent’s whole rat genome 60 K gene expression arrays. To normalize the data set, we performed loess within-array normalization and quantiles between-array normalization. Differential expression analysis was done using Bioconductor’s limma package (http://www.bioconductor.org). At a later stage, we used the file “SurePrint G3 Rat GE 8×60 K Microarray” to obtain the annotations of the rat genome from Agilent. Genes that showed adjusted p-values <0.005 both in siFoxE1 vs. wild type and in siFoxE1 vs. siScramble PCCl3 cells were considered differentially expressed. Functional analysis of Gene Ontology (GO) terms was carried out using the FatiGO tool and gene set enrichment analysis was performed using FatiScan , . All microarray data can be downloaded from the Gene Expression Omnibus (GEO; http://www.ncbi.nlm.nih.gov/geo/) database under accession number GSE42497.
Experimental Validation of Expression Array Data
Technical validations of expression data were performed by real-time quantitative PCR (qPCR) and western blots after performing four independent experiments of FoxE1 silencing.
Total RNA was isolated from PCCl3 cells after silencing following the standard TRIzol Reagent protocol (Invitrogen, Carlsbad, CA). RNA was reverse transcribed with random primers and quantitative PCR reactions were assembled in triplicate using the SYBR Green PCR Kit (Kapa Biosystems, Woburn, MA) following the manufacturer’s protocol. Expression level changes were studied for 6 thyroid-specific genes (FoxE1, Tpo, Tg, Nis, Pax8 and Titf1) as well as for 12 putative FoxE1-regulated genes, using beta-glucuronidase (Gus) for gene expression normalization (oligos are shown in Table S1).
Total protein extracts were obtained by scraping the cells in RIPA buffer (1% PBS, 0.5% (wt/vol) Nonidet, 0.1% (wt/vol) sodium deoxycholate, 0.1% SDS). Protein extracts (30 µg) were separated on 10% SDS-PAGE gels and immunodetected after western blotting. FoxE1 antibody was from Biopat (Milan, Italy), Actin and Cdh1 antibodies were from Santa Cruz Biotechnology, Inc. (Santa Cruz, CA), and Nis and Duox2 antibodies were generous gifts from Dr. N. Carrasco (Department of Cellular and Molecular Physiology, Yale School of Medicine, New Haven, CT, USA)  and Dr. C. Dupuy (Institut Gustave Roussy,Villejuif, France) , respectively.
We searched for FoxE1 binding sites in promoter regions (+/−1000 bp relative to the transcription start site) of genes statistically significantly regulated by FoxE1 (p-values <0.005). We performed an automated search of genes containing both the FoxE1 core binding site [5′-ACAAA-3′] and the NF1/CTF binding motif [5′-TTGG -3′] localised between 5 to 30 bp from the FoxE1 site, as previously described for FoxE1 binding within the Tpo promoter [5′-TTGG-(5–30)bp-ACAAA-3′] . FASTA promoter sequences of FoxE1-regulated genes, as well as promoter sequences of their orthologous genes, were extracted from the Ensembl database (http://www.ensembl.org).
ChIP and ReChIP Assays
Chromatin immunoprecipitation (ChIP) and sequential ChIP (ReChIP) were performed using the Diagenode HighCell ChIP kit following the manufacturer’s standard protocol. Cross-linked PCCl3 chromatin was immunoprecipitated using polyclonal antibodies against FoxE1 (Biopat, Milan, Italy) and NF1 (Abcam, Cambridge, UK). Two independent ChIP experiments were performed using two different batches of Biopat FoxE1 antibodies.
ReChIP was used to analyse simultaneous binding of FoxE1 and NF1/CTF proteins to a common DNA region. After a first immunoprecipitation with FoxE1 antibody using the Diagenode HighCell ChIP kit, samples were rinsed with RIPA buffer, and the first antibody was stripped from the beads by incubating in 1% SDS at 65°C for 15 minutes. Then a second round of immunoprecipitation with NF1 antibody was performed using the same Diagenode HighCell ChIP kit and following the manufacturer’s standard protocol.
FoxE, NF1 and FoxE1/NF1 -immunoprecipitated and input samples were analysed by real-time PCR using specific primers for the analysed regions (see Table S2). The known FoxE1 targets Tpo and Tg were used as positive IP controls whereas intron and promoter regions of Gad1 (glutamate decarboxylase 1) and Afm (afamin or alpha-albumin) were used as negative controls. qPCR reactions were assembled in triplicate using the SYBR Green PCR Kit (Kapa Biosystems, Woburn, MA), following the manufacturer’s protocol. The enrichment of target sequences in ChIP material was calculated relative to the Afm and Gad1 negative controls, and normalized to their relative amplification in the input sample .
Transfection and Luciferase Assays
HeLa cells were plated at a density of 2×105 cells/well in 6-well tissue culture plates, 24 h before transfection. Transfections were performed by calcium phosphate co-precipitation as described previously . The rat promoter pNIS 2.8-Luc, which contains the NIS Upstream Enhancer (NUE)  was transiently transfected alone (1.5 µg) or in combination with 1 µg of the expression vectors CMV-FoxE1 (formerly CMV-TTF2)  or RSV-CTF/NF1-C . One hundred nanograms of the CMV-Renilla vector were cotransfected to assess transfection efficiency. After 48 h, cells were harvested, lysed, and analysed for Luciferase and Renilla activities by the Dual-Luciferase reporter assay system (Promega, Madison WI). The promoter activity in cells transfected with expression vectors was determined as the ratio between luciferase and Renilla activities, relative to the ratio obtained in cells transfected with the corresponding control vectors. The results shown are the average ± SD of six different experiments performed in triplicate. Statistical significance was determined by t test analysis (two-tailed), and differences were considered significant at p<0.05. Western blots were done where indicated to determine the levels of expression of the transfected expression vectors (data not shown).
Gene Expression Profile of FoxE1-silenced Cells
In order to further investigate FoxE1 function and to identify putative FoxE1 target genes in thyroid cells, we performed whole genome microarray analysis in PCCl3 thyroid cells after knocking down FoxE1. Silencing was performed for 48 h after transfection. Data from four independent silencing experiments were analysed. We obtained a FoxE1 silencing grade of more than 85% (Fig. 1B). Statistically significant probes (p<0.005) from the siFoxE1 PCCl3 vs siScrambl PCCl3 comparison showed 74 differentially deregulated probes, including 24 upregulated and 50 downregulated probes. Comparison of siFoxE1 PCCl3 vs wt PCCl3 generated 211 differentially regulated probes, consisting of 99 upregulated and 112 downregulated probes (Table 1). Array results of both comparisons are shown in datasets S1 and S2, respectively. Statistically significant probes (p<0.005) from both comparisons are shown in Table S3. Combination of significant results from both comparisons showed 55 differentially deregulated genes, 17 of which were upregulated and 38 were downregulated (Tables 1 and 2). As expected, expression of previously known FoxE1 targets (Tpo and Tg) was downregulated, in the absence of FoxE1, in both comparisons.
Gene ontology (GO) analysis using FatiGO identified common GO terms, including endoplasmic reticulum (ER) overload response (GO:0006983), response to ER stress (GO:0034976), cellular response to glucose starvation (GO:0042149), protein folding (GO:0006457), heat shock protein binding, and oxidoreductase activity acting on sulfur group of donors, disulfide as acceptor (GO:0016671).
Independent Validation of Genes Regulated by FoxE1 Silencing
We performed validations of microarray data using four independent FoxE1 silencing experiments, 48 h after siRNA transfection. First of all, we analysed expression levels of FoxE1, in order to check and confirm its inhibition at the mRNA level. We also checked the mRNA expression status of the FoxE1 target genes Tpo and Tg, as well as mRNA levels of the classical, most representative, thyroid genes: the thyroid iodide symporter gene, Nis, and thyroid transcription factors Pax8 and Ttf1 (Fig. 2A).
Relative expression assessed by means of qRT-PCR of 6 thyroid-specific genes (panel A) and 12 additional genes differentially down- and upregulated in FoxE1-silenced PCCl3 cells (panel B). As FoxE1-dependent positive controls, we evaluated Tpo and Tg mRNA expression levels. Relative gene expression in siFoxE1 samples was calculated using the corresponding siScrambl samples as a reference ( = 1). Results are mean ± SEM of four independent experiments. Total protein extracts were prepared and submitted to western blot analysis to assess the protein levels of FoxE1 and Nis (panel C), and of Cdh1 and Duox2 (panel D). Actin was used as loading control. Representative western blot assays of four independent experiments are shown.
Moreover, we validated 12 genes randomly chosen from the list of putative FoxE1 targets, 5 of which were upregulated and 7 of which were downregulated after FoxE1 silencing (Table 2). Genes selected for this validation included: metallopeptidase Adamts9; cadherin 1 (Cdh1), a cell-cell adhesion glycoprotein; Crip2, a repressor of NF-kappaB; dual oxidase 2 (Duox2), a glycoprotein located at the apical membrane of thyroid follicular cells; S100a4, a calcium-binding protein; caspase 4 (Casp4), an apoptosis-related cysteine peptidase; the cysteine-rich with GF-like domains protein 2, Creld2; dual specificity phosphatase 5 (Dusp5); transcription factor Etv5; 70 kDa heat shock protein 5 (Hsp5a); steroid-thyroid hormone-retinoid receptor Nr4a2; and transmembrane 4L six family member 1, Tm4sf1.
After FoxE1 silencing, qRT-PCR analysis confirmed downregulation of FoxE1 as well as its previously described target genes Tpo and Tg , . Interestingly, we also observed inhibition of Nis, and its mRNA levels were decreased even more than Tg mRNA levels (Fig. 2A). We confirmed the microarray data for selected genes (Fig. 2B) with the only exception of the Crip2 gene, whose qRT-PCR expression levels did not show differences after FoxE1 silencing. All validation experiments were performed four times, attesting to the robustness of the results.
We confirmed that mRNA levels of Adamts9, Cdh1, Duox2 and S100a4 were upregulated, and those of Casp4, Creld2, Dusp5, Etv5, Hsp5a, Nr4a2 and Tm4sf1 were downregulated when FoxE1 was silenced in thyroid cells.
The protein expression of FoxE1 was analysed in order to check and confirm its inhibition at the protein level. We then checked protein levels of three FoxE1 putative target genes, Nis, Duox2 and Cdh1 (Figs 2C and 2D). After FoxE1 silencing, Duox2 protein levels did not change, but we confirmed inhibition of Nis expression and overexpression of Cdh1, at protein level.
Identification and Functional Analysis of Direct FoxE1 Targets
With the aim of investigating new direct FoxE1 targets, first we searched for the specific FoxE1 core binding sequence, AAACA, in promoter regions of putative FoxE1 regulated genes (+/−1000 pb). However, AAACA is a short sequence that can be easily found at random in the genome. We therefore performed an automatic search in the promoter sequences for FoxE1 binding sites in close proximity (5–30 bp) to the NF1/CTF binding motif, [5′-TTGG-(5–30)bp-ACAAA-3′], as previously described for FoxE1 binding to the Tpo promoter  and for other forkhead family members .
FoxE1-NF1/CTF binding sequences were present in promoter regions of 26 out of 55 differentially FoxE1-regulated genes. Results are summarized in Table 3. Considering their previously described function in thyroid biology, we selected three of these genes: Duox2, Cdh1 and Nr4a2 for performing ChIP in vivo analysis of FoxE1 binding to their promoters.
We also wanted to study in more detail the role of FoxE1 in Nis regulation (Fig. 2A). As we detected that Nis mRNA levels were even more strongly reduced than Tg mRNA levels, we tried to find FoxE1-NF1/CTF binding sequences in both the Nis proximal promoter (+/−1000 pb) and in the Nis upstream enhancer element (NUE ) localised between –2495 and –2264 bp . A FoxE1-NF1/CTF binding sequence was detected in the NUE, close to a previously described Pax8 binding sequence  (Fig. 3A), but not in the proximal Nis promoter.
Chromosomal location of putative FoxE1 and NF1/CTF binding motifs in the Nis upstream enhancer (panel A) and the Duox2 gene (panel B). Oligos used and exons are represented in italics.
In order to analyse in vivo FoxE1 binding to Duox2, Cdh1, Nr4a2 and Nis promoter sequences, we performed chromatin immunoprecipitation (ChIP) using a polyclonal antibody against FoxE1. We analysed immunoprecipitated DNA of two independent experiments using qPCR. Results of IP ratios normalized to the Afm negative control are shown in Fig. 4. The same results were obtained when we used Gad1 as a negative control (data not shown). Previously described FoxE1 binding sequences in the Tg and Tpo promoters were used as positive controls.
qPCR analysis of chromatin immunoprecipitation performed on PCCl3 cells with FoxE1 antibody. The enrichment of target sequences was calculated as the IP ratio (arbitrary units) relative to the negative control Afm, and normalized to their relative amplification in the input sample. Sequences from the Tg and Tpo promoters were used as positive controls. Two regulatory regions were analysed in the Duox2 promoter (called Duox2-1 and Duox2-2). Results are mean ± SEM of two independent experiments, each performed in triplicate.
The FoxE1 antibody precipitated FoxE1 target sequences in the Tg and Tpo promoters with relative immunoprecipitation (IP) ratios of 22.8 and 3.6, respectively. We found binding of FoxE1 to chromatin regions containing the Nis upstream enhancer element (relative IP ratio: 12.6). We also observed positive FoxE1 binding to two DNA sites located in the Duox2 gene; one in the promoter sequence at −351 bp (relative IP ratio: 28.1) and the other in the third exon at +696 bp (relative IP ratio: 9.3) (Figs 3 and 4). We did not detect direct FoxE1 interactions with either Cdh1 or Nr4a2.
In order to better understand whether FoxE1 and NF1/CTF are bound together on the Nis and Duox2 promoter regions, we performed ChIP assays using a polyclonal antibody against NF1. Next, we also performed ReChIP experiments with FoxE and NF1 antibodies. We quantified immunoprecipitated DNA regions using qPCR. The previously described FoxE1 binding sequence in the Tg promoter was used as positive control.
In our experimental conditions, we did not observe in vivo NF1/CTF binding to either of the two putative FoxE1 binding sequences in Duox2 (data not shown). Nevertheless, we detected positive binding of NF1/CTF to the NUE. Moreover, we observed simultaneous FoxE1 and NF1/CTF binding to the NUE region. Results of IP ratios normalised to the Afm negative control are shown in Fig. 5A. Similar results were obtained when Gad1 was used as a negative control (data not shown).
A ReChIP assay was used to analyse simultaneous binding of FoxE1 and NF1/CTF proteins to the NUE (panel A). qPCR was done to analyse chromatin immunoprecipitates of PCCl3 cells using FoxE1 and NF1 antibodies. The enrichment of target sequences was calculated as the IP ratio (arbitrary units) relative to the negative control Afm, and normalized to their relative amplification in the input sample. A sequence from the Tg promoter was used as positive control. HeLa cells were transfected with 1 µg of a FoxE1 or NF1/CTF expression vector or the empty vector, and 1.5 µg and 0.1 µg of pNIS 2.8-Luc and CMV-Renilla constructs respectively (panel B). Forty-eight hours after transfection, cells were collected for the measurement of luciferase and renilla levels. Results are shown as the mean±SD of the luciferase levels relative to the non-regulated renilla levels of six independent experiments. (*): p<0.05; (**): p<0.01; (***): p<0.001; two tailed t-test.
As we wanted to study in depth the role of FoxE1 and NF1/CTF in Nis regulation, we cotransfected HeLa cells with FoxE1 and NF1/CTF expression vectors, together with a luciferase expression vector containing the NUE. Forty-eight hours later, cells were lysed and luciferase levels were measured. The results showed a significant 2-fold increase in the activity of NUE when FoxE1 or NF1/CTF were expressed individually in HeLa cells; thus, the effect was additive when both transcription factors were transfected together (Fig. 5B).
In summary, we present the first description of specific in vivo FoxE1 binding to regulatory regions of two relevant thyroid genes, Nis and Duox2. These FoxE1 core binding sequences are located close to a NF1/CTF transcription factor binding site. Moreover, we demonstrate that FoxE1 and NF1/CTF are simultaneously bound to the Nis upstream enhancer and that they cooperate in the regulation of NIS-promoter activity.
FoxE1 is a forkhead transcription factor essential for thyroid differentiation and function. In humans, FOXE1 mutations cause the Bamforth–Lazarus syndrome (OMIM 241850) , . In spite of this, FoxE1 expression levels seem to be unaltered in human tumours. As reflected in the Gene Expression Omnibus database, FOXE1 variations have been associated with susceptibility to several types of cancer , , , including papillary thyroid cancer , , . FoxE1 recognizes and binds to the Tg and Tpo promoters. However, little is known about other downstream targets of FoxE1 that could explain the phenotypes observed in humans carrying variations in FoxE1. Only two studies reporting novel genes regulated by FoxE1 have been published until now , , but both were done in a heterologous system overexpressing FoxE1. Our goal was to find new potential downstream mediators of FoxE1 function in vivo, but importantly, instead of a heterologous system we used a thyroid cell line. Results obtained by performing a genome-wide approach followed by chromatin immunoprecipitation analysis, led us to the identification of direct FoxE1 target genes in thyroid cells.
In this study, we found 54 putative novel FoxE1 target genes. We obtained a very restrictive list of significantly regulated genes, based on overlapping expression array data of two comparisons. This restrictive list did not allow us to obtain relevant information about main pathways affected by FoxE1 action. We identified and validated eleven genes regulated by FoxE1; four of them (Adamts9, Cdh1, Duox2 and S100a4) showed an increased expression, whereas the remaining seven genes (Casp4, Creld2, Dusp5, Etv5, Hsp5a, Nr4a2 and Tm4sf1) showed decreased mRNA levels when FoxE1 is silenced. Moreover, we have not ruled out that the other 43 statistically significant genes are true FoxE1 targets, and in future studies we will take advantage of these additional array results.
In order to identify genes directly regulated by FoxE1, we performed an in silico search of the FoxE1 binding motif in the promoter regions of 54 statistically significant genes. As the FoxE1 core binding sequence is very short, only five nucleotides, we were forced to perform a more restrictive search including the NF1/CTF binding sequence, as previously described for the Tpo promoter  and for other forkhead factors in the serum albumin gene promoter . It has been proposed that the conserved interaction between NF1/CTF and forkhead proteins could reflect a general mechanism of action of both families of transcription factors. The NF1/CTF-binding site is masked inside the nucleosomal structure; binding of forkhead proteins to their cognate sites opens this structure and makes the NF1/CTF site accessible to exert its transactivation effect. The ability of the forkhead domain to induce DNA bending  would promote contact with NF1/CTF factors. It has been hypothesised that this interaction may play an important role in cell type-specific transcription and might be a widespread phenomenon .
Duox2 codes for the enzyme that catalyzes H2O2 production in the thyroid gland , , . Duox2 expression is found in the thyrocyte more abundantly than in several other cell types . In humans, mutations in the corresponding gene are associated with congenital hypothyroidism , . Duox2 regulation by the classical thyroid transcription factors Pax8 and Ttf1 has been studied in depth, however, two main studies reported controversial results , . Nevertheless, until now there were no indications in the literature of a role of FoxE1 in Duox2 regulation.
In the present work we demonstrate that FoxE1 interacts in vivo with two DNA regions located in Duox2. We observed FoxE1 binding to the promoter region at −351 bp and to a second regulatory region located in exon 3 at +696 bp (Figs 3 and 4). However Duox2 did not seem to be regulated by FoxE1 at the protein level, indicating the involvement of strong post-transcriptional regulatory mechanisms. Since Duox2 mRNA expression increases when FoxE1 is silenced and since FoxE1 binds directly to two Duox2 regulatory regions, we can hypothesise that FoxE1 is acting as a transcriptional repressor of Duox2.
The repressor function of FoxE1 has been previously reported. FoxE1 represses the transcriptional activity of Ttf1 and Pax8 during thyroid development through its C-terminal region, which contains an alanine-rich domain , . FoxE1 also inhibits estrogen receptor α transactivation through this domain . In line with our hypothesis, a direct repression function that may require both target DNA binding and dimerization domains  has been demonstrated for other forkhead family members such as FoxP.
FoxE1 binding domains in Duox2 regulatory regions are localised at 26 and 27 bp from the NF1/CTF binding motif (Table 3). Nevertheless, in our experimental conditions, we were not able to detect direct in vivo binding of NF1/CTF to either of the two FoxE1 core sequences in Duox2. Thus, the exact FoxE1-mediated repression mechanism of Duox2 expression remains to be elucidated.
Cdh1 belongs to a superfamily of glycoproteins that mediate calcium-dependent cell-cell adhesion. Cdh1 is highly expressed in the normal thyroid gland and its expression leads to aggregation of thyrocytes . Moreover, thyroid-stimulating hormone (TSH) in dog and human thyroid cell cultures regulates Cdh1 transcription , suggesting a relationship between Cdh1 expression and thyroid differentiation. FoxE1 expression itself is under TSH control . Therefore it seems plausible that TSH, via FoxE1 and other players, might regulate expression levels of Cdh1. In fact, we observed an increased Cdh1 mRNA and protein expression when FoxE1 was silenced; nevertheless, we were not able to detect a direct interaction between FoxE1 and Cdh1.
Nr4a2 is a transcription factor that belongs to the steroid-thyroid hormone-retinoid receptor superfamily. In addition, Nr4a2 is a member of the NR4A subfamily of nuclear orphan receptors that function as ligand-independent early response genes involved in proliferation, apoptosis, and inflammation . Nr4a2 cooperates with another member of the forkhead protein family, FoxA2, during dopamine neuron differentiation  and Nr4a2 overexpression enhances migration of mesenchymal stromal cells . Our data reveals a downregulation of Nr4a2 expression when FoxE1 is absent, but, in our experiments, FoxE1 did not directly bind to its promoter.
Further studies will be done to analyse in depth the apparently indirect regulation by FoxE1 of Nr4a2 and Cdh1 and its implication in cellular migration and/or thyroid differentiation.
Surprisingly, in the absence of FoxE, we observed both reduced Nis mRNA levels and reduced Nis protein levels (Fig. 2A and C). We found one FoxE1 core binding sequence and three NF1/CTF motifs in the NUE (Fig. 3A). NUE stimulates transcription in a thyroid-specific, cAMP-dependent manner and involves the most relevant aspect of Nis regulation. NUE contains Ttf1 binding sites that have no known effect on Nis transcription, Pax8 binding sites, and a degenerate cAMP responsive element sequence . In their original study, Ohno et al. performed DNase I footprinting analysis of the NUE, but no footprints were detected when FoxE1 was used . The first and foremost possibility that might explain the apparent discrepancy with our results is that Ohno et al. did in vitro experiments. In a later study, Li et al. suggested that FoxE1 participates in a human NIS repressor complex, together with PARP-1 . Our study clearly demonstrates a direct in vivo interaction of FoxE1 with the NUE and a modification of Nis transcriptional activity in the absence of FoxE1. In addition, we observed direct NF1/CTF binding to the NUE, and we also demonstrated that FoxE1 and NF1/CTF are bound together on the NUE region. Binding of both transcription factors together resulted in an increased transcriptional activity of the NUE, confirming their cooperation in Nis regulation (Fig. 5).
Our work emphasizes the idea of a cross-talk of thyroid transcription factors during the process of transcriptional regulation, controlling key cellular processes for thyrocyte biology, with different roles of several players of this complex regulatory system that are still emerging.
On the one hand, a master thyroid gene, Nis, contains in its NUE binding sequences for Pax8, Ttf-1 and FoxE1-NF1/CTF that appear to be linked in a complex network of reciprocal regulatory interactions. On the other hand, another of the best candidate genes in our genome-wide study is the thyroid gene Duox2; thus, evidence is accumulating regarding the existence of a regulatory network in the thyroid involving FoxE1.
Finally, our study points out the importance of cooperating transcription factors and the usefulness of previously used tools, especially the in silico search of transcription factor binding sequences. Core sequences are necessary but not sufficient, and flanking sequences determine transcription factor binding specificity , . The use of the FoxE1-NF1/CTF binding motif  led us to obtain a workable restrictive list of potential direct FoxE1 targets. Moreover, this opens a window to further investigating the cooperative function of FoxE1 and NF1/CTF in transcriptional regulation of our candidate genes, as we demonstrated to occur for Nis expression.
In search of potential downstream mediators of FoxE1 function, we identified Nis and Duox2 as novel direct FoxE1 target genes. To our knowledge, this is the first evidence regarding the implication of Nis and Duox2 in executing the transcriptional programme triggered by FoxE1. Both are pivotal proteins required for thyroid gland differentiation and function.
Although mainly acting as a regulator of the expression of the thyroid genes Tg, Tpo, Nis and Duox2, FoxE1 may also act in thyroid cells as a transcriptional regulator of many other candidate genes, some of which are suggested in the present study.
Oligonucleotides used for experimental validation of expression array data.
Oligonucleotides used for ChIP analysis.
Statistically significant probes (p<0.005) common to both comparisons (siFoxE1 PCCl3 vs siScrambl PCCl3, and siFoxE1 PCCl3 vs wt PCCl3).
Microarray results for siFoxE1 PCCl3 vs siScramble PCCl3 comparison.
We thank David G. Pisano and Enrique Carrillo de Santa Pau from the Bioinformatics Unit of the Spanish National Cancer Research Centre (CNIO) for their contributions to microarray data analysis and Dr. Ronald Hartong for critical reading of the manuscript. We also thank Dr. Nancy Carrasco (Department of Cellular and Molecular Physiology, Yale School of Medicine, New Haven, CT, USA) and Dr. Corinne Dupuy (Institut Gustave Roussy,Villejuif, France) for the generous gifts of Nis and Duox2 antibodies respectively.
Conceived and designed the experiments: LPF PS. Performed the experiments: LPF AL-M. Analyzed the data: LPF AMM GG-L. Contributed reagents/materials/analysis tools: PS. Wrote the paper: LPF PS.
- 1. Damante G, Tell G, Di Lauro R (2001) A unique combination of transcription factors controls differentiation of thyroid cells. Prog Nucleic Acid Res Mol Biol 66: 307–356.
- 2. De Felice M, Di Lauro R (2004) Thyroid development and its disorders: genetics and molecular mechanisms. Endocr Rev 25: 722–746.
- 3. Zannini M, Avantaggiato V, Biffali E, Arnone MI, Sato K, et al. (1997) TTF-2, a new forkhead protein, shows a temporal expression in the developing thyroid which is consistent with a role in controlling the onset of differentiation. EMBO J 16: 3185–3197.
- 4. Myatt SS, Lam EW (2007) The emerging roles of forkhead box (Fox) proteins in cancer. Nat Rev Cancer 7: 847–859.
- 5. Carlsson P, Mahlapuu M (2002) Forkhead transcription factors: key players in development and metabolism. Dev Biol 250: 1–23.
- 6. Cuesta I, Zaret KS, Santisteban P (2007) The forkhead factor FoxE1 binds to the thyroperoxidase promoter during thyroid cell differentiation and modifies compacted chromatin structure. Mol Cell Biol 27: 7302–7314.
- 7. Zaret KS, Carroll JS (2011) Pioneer transcription factors: establishing competence for gene expression. Genes Dev 25: 2227–2241.
- 8. De Felice M, Ovitt C, Biffali E, Rodriguez-Mallon A, Arra C, et al. (1998) A mouse model for hereditary thyroid dysgenesis and cleft palate. Nat Genet 19: 395–398.
- 9. Bamforth JS, Hughes IA, Lazarus JH, Weaver CM, Harper PS (1989) Congenital hypothyroidism, spiky hair, and cleft palate. J Med Genet 26: 49–51.
- 10. Clifton-Bligh RJ, Wentworth JM, Heinz P, Crisp MS, John R, et al. (1998) Mutation of the gene encoding human TTF-2 associated with thyroid agenesis, cleft palate and choanal atresia. Nat Genet 19: 399–401.
- 11. Venza I, Visalli M, Tripodo B, Lentini M, Teti D, et al. (2009) Investigation into FOXE1 genetic variations in cutaneous squamous cell carcinoma. Br J Dermatol 162: 681–683.
- 12. Venza I, Visalli M, Tripodo B, De Grazia G, Loddo S, et al. (2009) FOXE1 is a target for aberrant methylation in cutaneous squamous cell carcinoma. Br J Dermatol 162: 1093–1097.
- 13. Brune K, Hong SM, Li A, Yachida S, Abe T, et al. (2008) Genetic and epigenetic alterations of familial pancreatic cancers. Cancer Epidemiol Biomarkers Prev 17: 3536–3542.
- 14. Gudmundsson J, Sulem P, Gudbjartsson DF, Jonasson JG, Sigurdsson A, et al. (2009) Common variants on 9q22.33 and 14q13.3 predispose to thyroid cancer in European populations. Nat Genet 41: 460–464.
- 15. Landa I, Ruiz-Llorente S, Montero-Conde C, Inglada-Perez L, Schiavi F, et al. (2009) The variant rs1867277 in FOXE1 gene confers thyroid cancer susceptibility through the recruitment of USF1/USF2 transcription factors. PLoS Genet 5: e1000637.
- 16. Bullock M, Duncan EL, O’Neill C, Tacon L, Sywak M, et al. (2012) Association of FOXE1 Polyalanine Repeat Region with Papillary Thyroid Cancer. J Clin Endocrinol Metab 97: E1814–1819.
- 17. Santisteban P, Acebron A, Polycarpou-Schwarz M, Di Lauro R (1992) Insulin and insulin-like growth factor I regulate a thyroid-specific nuclear protein that binds to the thyroglobulin promoter. Mol Endocrinol 6: 1310–1317.
- 18. Aza-Blanc P, Di Lauro R, Santisteban P (1993) Identification of a cis-regulatory element and a thyroid-specific nuclear factor mediating the hormonal regulation of rat thyroid peroxidase promoter activity. Mol Endocrinol 7: 1297–1306.
- 19. Perrone L, Pasca di Magliano M, Zannini M, Di Lauro R (2000) The thyroid transcription factor 2 (TTF-2) is a promoter-specific DNA-binding independent transcriptional repressor. Biochem Biophys Res Commun 275: 203–208.
- 20. Francis-Lang H, Price M, Polycarpou-Schwarz M, Di Lauro R (1992) Cell-type-specific expression of the rat thyroperoxidase promoter indicates common mechanisms for thyroid-specific gene expression. Mol Cell Biol 12: 576–588.
- 21. Ortiz L, Aza-Blanc P, Zannini M, Cato AC, Santisteban P (1999) The interaction between the forkhead thyroid transcription factor TTF-2 and the constitutive factor CTF/NF-1 is required for efficient hormonal regulation of the thyroperoxidase gene transcription. J Biol Chem 274: 15213–15221.
- 22. Hishinuma A, Ohmika N, Namatame T, Ieiri T (2004) TTF-2 stimulates expression of 17 genes, including one novel thyroid-specific gene which might be involved in thyroid development. Mol Cell Endocrinol 221: 33–46.
- 23. Venza I, Visalli M, Parrillo L, De Felice M, Teti D, et al. (2010) MSX1 and TGF-beta3 are novel target genes functionally regulated by FOXE1. Hum Mol Genet 20: 1016–1025.
- 24. Fusco A, Berlingieri MT, Di Fiore PP, Portella G, Grieco M, et al. (1987) One- and two-step transformations of rat thyroid epithelial cells by retroviral oncogenes. Mol Cell Biol 7: 3365–3370.
- 25. Costamagna E, Garcia B, Santisteban P (2004) The functional interaction between the paired domain transcription factor Pax8 and Smad3 is involved in transforming growth factor-beta repression of the sodium/iodide symporter gene. J Biol Chem 279: 3439–3446.
- 26. Medina DL, Santisteban P (2000) Thyrotropin-dependent proliferation of in vitro rat thyroid cell systems. Eur J Endocrinol 143: 161–178.
- 27. Al-Shahrour F, Minguez P, Tarraga J, Medina I, Alloza E, et al. (2007) FatiGO +: a functional profiling tool for genomic data. Integration of functional annotation, regulatory motifs and interaction data with microarray experiments. Nucleic Acids Res 35: W91–96.
- 28. Medina I, Carbonell J, Pulido L, Madeira SC, Goetz S, et al. (2010) Babelomics: an integrative platform for the analysis of transcriptomics, proteomics and genomic data with advanced functional profiling. Nucleic Acids Res 38: W210–213.
- 29. De la Vieja A, Reed MD, Ginter CS, Carrasco N (2007) Amino acid residues in transmembrane segment IX of the Na+/I- symporter play a role in its Na+ dependence and are critical for transport activity. J Biol Chem 282: 25290–25298.
- 30. Dupuy C, Pomerance M, Ohayon R, Noel-Hudson MS, Deme D, et al. (2000) Thyroid oxidase (THOX2) gene expression in the rat thyroid cell line FRTL-5. Biochem Biophys Res Commun 277: 287–292.
- 31. Ruiz-Llorente S, Carrillo Santa de Pau E, Sastre-Perona A, Montero-Conde C, Gomez-Lopez G, et al. (2012) Genome-wide analysis of Pax8 binding provides new insights into thyroid functions. BMC Genomics 13: 147.
- 32. Garcia B, Santisteban P (2002) PI3K is involved in the IGF-I inhibition of TSH-induced sodium/iodide symporter gene expression. Mol Endocrinol 16: 342–352.
- 33. Nebl G, Cato AC (1995) NFI/X proteins: a class of NFI family of transcription factors with positive and negative regulatory domains. Cell Mol Biol Res 41: 85–95.
- 34. Jackson DA, Rowader KE, Stevens K, Jiang C, Milos P, et al. (1993) Modulation of liver-specific transcription by interactions between hepatocyte nuclear factor 3 and nuclear factor 1 binding DNA in close apposition. Mol Cell Biol 13: 2401–2410.
- 35. Ohno M, Zannini M, Levy O, Carrasco N, di Lauro R (1999) The paired-domain transcription factor Pax8 binds to the upstream enhancer of the rat sodium/iodide symporter gene and participates in both thyroid-specific and cyclic-AMP-dependent transcription. Mol Cell Biol 19: 2051–2060.
- 36. Pierrou S, Hellqvist M, Samuelsson L, Enerback S, Carlsson P (1994) Cloning and characterization of seven human forkhead proteins: binding site specificity and DNA bending. EMBO J 13: 5002–5012.
- 37. Bjorkman U, Ekholm R, Denef JF (1981) Cytochemical localization of hydrogen peroxide in isolated thyroid follicles. J Ultrastruct Res 74: 105–115.
- 38. Bjorkman U, Ekholm R (1984) Generation of H2O2 in isolated porcine thyroid follicles. Endocrinology 115: 392–398.
- 39. Dupuy C, Ohayon R, Valent A, Noel-Hudson MS, Deme D, et al. (1999) Purification of a novel flavoprotein involved in the thyroid NADPH oxidase. Cloning of the porcine and human cdnas. J Biol Chem 274: 37265–37269.
- 40. Christophe-Hobertus C, Christophe D (2007) Human Thyroid Oxidases genes promoter activity in thyrocytes does not appear to be functionally dependent on Thyroid Transcription Factor-1 or Pax8. Mol Cell Endocrinol 264: 157–163.
- 41. Moreno JC, Bikker H, Kempers MJ, van Trotsenburg AS, Baas F, et al. (2002) Inactivating mutations in the gene for thyroid oxidase 2 (THOX2) and congenital hypothyroidism. N Engl J Med 347: 95–102.
- 42. Vigone MC, Fugazzola L, Zamproni I, Passoni A, Di Candia S, et al. (2005) Persistent mild hypothyroidism associated with novel sequence variants of the DUOX2 gene in two siblings. Hum Mutat 26: 395.
- 43. D’Andrea B, Iacone R, Di Palma T, Nitsch R, Baratta MG, et al. (2006) Functional inactivation of the transcription factor Pax8 through oligomerization chain reaction. Mol Endocrinol 20: 1810–1824.
- 44. Park E, Gong EY, Romanelli MG, Lee K (2012) Suppression of estrogen receptor-alpha transactivation by thyroid transcription factor-2 in breast cancer cells. Biochem Biophys Res Commun 421: 532–537.
- 45. Wang B, Lin D, Li C, Tucker P (2003) Multiple domains define the expression and regulatory properties of Foxp1 forkhead transcriptional repressors. J Biol Chem 278: 24259–24268.
- 46. Yap AS, Stevenson BR, Keast JR, Manley SW (1995) Cadherin-mediated adhesion and apical membrane assembly define distinct steps during thyroid epithelial polarization and lumen formation. Endocrinology 136: 4672–4680.
- 47. Brabant G, Hoang-Vu C, Behrends J, Cetin Y, Potter E, et al. (1995) Regulation of the cell-cell adhesion protein, E-cadherin, in dog and human thyrocytes in vitro. Endocrinology 136: 3113–3119.
- 48. Ortiz L, Zannini M, Di Lauro R, Santisteban P (1997) Transcriptional control of the forkhead thyroid transcription factor TTF-2 by thyrotropin, insulin, and insulin-like growth factor I. J Biol Chem. 272: 23334–23339.
- 49. Maijenburg MW, Gilissen C, Melief SM, Kleijer M, Weijer K, et al. (2011) Nuclear receptors Nur77 and Nurr1 modulate mesenchymal stromal cell migration. Stem Cells Dev 21: 228–238.
- 50. Lee HS, Bae EJ, Yi SH, Shim JW, Jo AY, et al. (2010) Foxa2 and Nurr1 synergistically yield A9 nigral dopamine neurons exhibiting improved differentiation, function, and cell survival. Stem Cells 28: 501–512.
- 51. Li W, Ain KB (2010) Human sodium-iodide symporter (hNIS) gene expression is inhibited by a trans-active transcriptional repressor, NIS-repressor, containing PARP-1 in thyroid cancer cells. Endocr Relat Cancer 17: 383–398.