Gene 527 (2013) 529–536

Contents lists available at SciVerse ScienceDirect

Gene

j ourna l homepage: www.e lsev ie r .com/ locate /gene

Methods paper

TFII-I regulates target genes in the PI-3K and TGF-β signaling pathways through anovel DNA binding motif

Maria Segura-Puimedon a,b, Cristina Borralleras a,b,c, Luis A. Pérez-Jurado a,b,c, Victoria Campuzano a,b,c,⁎a Unitat de Genètica, Departament de Ciències Experimentals i de la Salut, Universitat Pompeu Fabra, 08003 Barcelona, Spainb Centro de Investigación Biomédica en Red de Enfermedades Raras (CIBERER), 08003 Barcelona, Spainc IMIM—Institut Hospital del Mar d'Investigacions Mèdiques, 08003 Barcelona, Spain

Abbreviations:WBS,Williams–Beuren syndrome; TFIII-I; ES, embryonic stem cells; MEFs, mouse embryoniccomparative genome hybridization; qRT-PCR, quantchain reaction; ChIP, chromatine immunoprecipitation;DEG, differentially expressed genes; NGF, nerve growth⁎ Corresponding author at: Genetics Unit, Universitat P

Spain. Tel.: +34 93 3160855; fax: +34 93 3160901.E-mail address: victoria.campuzano@upf.edu (V. Cam

0378-1119/$ – see front matter © 2013 Elsevier B.V. Allhttp://dx.doi.org/10.1016/j.gene.2013.06.050

a b s t r a c t

a r t i c l e i n f o

Article history:Accepted 21 June 2013Available online 3 July 2013

Keywords:TFII-IWilliams–Beuren syndromeTranscriptional regulationDNA binding domainMouse modelEmbryonic stem cells

General transcription factor (TFII-I) is a multi-functional protein involved in the transcriptional regulation ofcritical developmental genes, encoded by the GTF2I gene located on chromosome 7q11.23. Haploinsufficiencyat GTF2I has been shown to play a major role in the neurodevelopmental features of Williams–Beuren syn-drome (WBS). Identification of genes regulated by TFII-I is thus critical to detect molecular determinants ofWBS as well as to identify potential new targets for specific pharmacological interventions, which are cur-rently absent. We performed a microarray screening for transcriptional targets of TFII-I in cortex and embry-onic cells from Gtf2imutant and wild-type mice. Candidate genes with altered expression were verified usingreal-time PCR. A novel motif shared by deregulated genes was found and chromatin immunoprecipitation as-says in embryonic fibroblasts were used to document in vitro TFII-I binding to this motif in the promoter re-gions of deregulated genes. Interestingly, the PI3K and TGFβ signaling pathways were over-representedamong TFII-I-modulated genes. In this study we have found a highly conserved DNA element, common toa set of genes regulated by TFII-I, and identified and validated novel in vivo neuronal targets of this proteinaffecting the PI3K and TGFβ signaling pathways. Overall, our data further contribute to unravel the complex-ity and variability of the different genetic programs orchestrated by TFII-I.

© 2013 Elsevier B.V. All rights reserved.

1. Introduction

General transcription factor (TFII-I) is member of an ubiquitouslyexpressed, multifunctional transcription factor family which operate asa molecular switch to convey signals frommultiple pathways andmedi-ate cellular response (Roy, 2001). TFII-I was originally identified as a pro-tein that binds to the initiator (Inr) core promoter element andwas latershown to bind to various upstream elements that include the E box (Royet al., 1997), the downstream immunoglobulin control element (DICE)sequence (Tantin et al., 2004) and the consensus BRGATTRBR sequence(Chimge et al., 2008). Structurally, TFII-I consists of multiple I-repeats,each of which contains a putative helix–loop–helix motif that is poten-tially important for protein–protein interactions (Roy, 2007). Growthfactor signaling leads to rapid tyrosine phosphorylation, followed by nu-clear translocation of TFII-I, and subsequent activation of target genes(Misra et al., 2009; Parker et al., 2001; Roy, 2007). In addition to its

I-I, general transcription factorfibroblast; aCGH, microarray

itative real-time polymeraseCNVs, copy number variants;factor.ompeu Fabra, 08003 Barcelona,

puzano).

rights reserved.

function as a transcription factor, cytosolic TFII-I regulates calcium ho-meostasis by modulating agonist-induced extracellular Ca2+ entry(Caraveo et al., 2006; Patterson et al., 2002). In the other hand, recentdata suggest that other member of the family GTF2IRD2 has evolved asan inhibitor of TFII-I (Palmer et al., 2012).

The gene encoding TFII-I, GTF2I (Entrez Gene ID 2969), has beenmapped to an interval of the human chromosome 7q11.23, a regioncommonly deleted in Williams–Beuren syndrome (WBS; OMIM no.194050). WBS is a rare developmental disorder with sporadic occur-rence (1/7000-1/20000), characterized by mild intellectual disabil-ities with a cognitive profile including relatively preserved verbalskills, very deficient visuospatial abilities and a characteristic person-ality showing high sociability with strangers and increased anxiety(Morris, 2010). Additional features include craniofacial dysmorphism,growth retardation, odynoacusis, as well as cardiovascular, endocrineand connective tissue abnormalities (Morris et al., 1988; Tassabehji,2003). The implication of GTF2I haploinsufficiency in the origin ofthis disorder is extensively documented but the concrete pathwaysand pathogenic mechanisms remain unclear (Antonell et al., 2010a;Ferrero et al., 2010; Sakurai et al., 2011). In mouse, the Gtf2i gene(Entrez Gene ID 14886, MGI 1202722) is located on chromosome 5G2.Gtf2i expression is important for embryonic development, as heterozy-gous knock-out mice featured anomalies similar to those observed inWBS: retarded growth, microcephaly, craniofacial and skeletal defects,

530 M. Segura-Puimedon et al. / Gene 527 (2013) 529–536

odynacusis and hypersociability (Enkhmandakh et al., 2009; Lucena etal., 2010; Sakurai et al., 2011).

The mechanisms whereby decreased TFII-I can promote anomaliessimilar to those observed inWBS have been subject to exhaustive cel-lular and biochemical studies (Antonell et al., 2010a; Dai et al., 2009;Roy, 2007; Stasyk et al., 2005). Transcriptomic profiles resulting fromchanges in TFII-I dosage have been already defined in cell lines andtissues (Chimge et al., 2007b; Enkhmandakh et al., 2009). So far, in-volvement of TFII-I in gene regulation has been validated for 20 directtarget genes involved in transcriptional regulation, chromatin remod-eling, cell cycle, muscle development and neurogenesis (Chimge etal., 2008; Lazebnik et al., 2009; Makeyev et al., 2010, 2013). Herein,we have extended these observations by comparative transcriptomeanalysis of brain tissues and ES cells from mutants of TFII-I. Our dataunveil the existence of a new consensus binding domain that in-creases the number of distinct transcriptional networks that dependon the TFII-I signals further endorsing the concept of the function do-mains as key regulator of the biochemical, genetic and biological out-comes of TFII-I.

2. Materials and methods

2.1. Animals and cell culture maintenance

Animals were maintained under standard animal housing conditionsin a 12-h dark-light cycle with free access to food and water. Animalcare was in accordance with ethical guidelines (European CommunitiesCouncil Directive 86/609/EEC) and approved by the Local Ethical Com-mittee (IMAS-IMIM/UPF). Animals used were already described (Lucenaet al., 2010). After behavioral phenotyping, cortex from Gtf2i+/Δexon2 andWT animals were immediately dissected from fresh brains and quicklyfrozen by liquid nitrogen until RNA extractions. The age of mice vary be-tween 20 and 25 weeks.

Clone XS0353 was obtained from the Sanger Institute Gene Trap Re-source (SIGTR) mouse ES cell line collection (http://www.sanger.ac.uk/PostGenomics/genetrap/). aCGH analysis between XS0353 and wild-type cell lines did not shown neither major genomic alterations (dele-tions or duplications) in chromosome 5 nor in the rest of chromosomes.We found a total of 11 CNVs containing 24 genes (Table S1, supplemen-tary information). None of them were found deregulated in the expres-sion array analysis, in concordance with previous findings (Henrichsenet al., 2009). XS0353 cell linewas cultured in standard conditions, Knock-out DMEM medium (Invitrogen), supplemented with 20% KnockoutSerum Replacement for ES cells (Invitrogen), LIF, β-mercaptoethanol,no essential aminoacids and penicillin/streptomycin. Cells were alwayscultured in a monolayer of feeder cells and maintained at 37 °C in a hu-midified 5% CO2 chamber. We performed RT-PCR, qRT-PCR and westernanalysis to ensure the loss-of-function of the gene-trap allele (Fig. S1,supplementary information).

2.2. RNA isolation and quantification

Total RNA was extracted, from XS0353 cell line and cortex, byusing TRIZOL reagent (Invitrogen, Carlsbad, CA, USA), followed by asecond extraction using RNeasy (Qiagen) in both cases according tothe manufacturer's instructions. Quality of all RNA samples waschecked using an Agilent 2100 Bioanalyzer (Agilent Technologies).Only those samples with an RNA Integrity Number (RIN) N7 wereused for hybridization.

2.3. Microarray and data analysis

Cortex tissues were hybridized to the Agilent 4x44K v1 MouseWhole Genome chips. Fluorescent images were obtained with theAgilent G2565BA Microarray Scanner System (Agilent Technologies)and TIFF images were quantified with the use of the Spot program

(http://experimental.act.cmis.csiro.au/Spot/index.php) under the Renvironment (http://www.r-project.org). The resulting raw valueswere filtered and an intensity cut-off was applied, selecting thosepoints with a foreground median/background median N3 in at leastone channel. A series of programs, collectively packaged as ArrayFile Maker 4.0 (AFM), were use to manipulate and manage DNA mi-croarray data (Breitkreutz et al., 2001). AFM 4.0, Quantarray DataHandler 3.0, and Array Database 1.0 can be downloaded at the TyersLab Home Page [http://www.mshri.on.ca/tyers/] and are copyrightedagainst commercial gain.

ES cells were hybridized using duplicates to the GeneChip MouseGenome 430 2.0. Array (Affymetrix)microarray imageswere processedwith Microarray Analysis Suite 5.0 (Affymetrix).

All samples demonstrated characteristics of high-quality cRNA (3′/5′ratio of probe sets for glyceraldehyde-3-phosphate dehydrogenase andβ-actin of b1.5) andwere subjected to subsequent analysis. Rawexpres-sion values obtained directly from CEL files were preprocessed using theRMA method (Irizarry et al., 2003), a three-step process which inte-grates background correction, normalization and summarization ofprobe values. These normalized valueswere the basis for all the analysis.Previous to any analysis, data were submitted to non-specific filtering toremove low signal genes (those genes whosemean signal in each groupdid not exceed a minimum threshold) and low variability genes (thosegenes whose standard deviation between all samples did not exceed aminimum threshold).

2.4. Statistical analysis

The selection of differentially expressed genes was performedusing a linear model approach following the methodology (“Empiri-cal Bayes and Linear Models for Microarrays”) (Smyth, 2004) andimplemented in the limma Bioconductor package.

Briefly, the method consists of fitting a linear model (similar tostandard ANOVA) for each gene followed by variance regularizationbased on empirical Bayes modelling. The empirical Bayes analysisimplemented yields several performance measures which help todecide which genes appear to be differentially express. We called dif-ferentially expressed those genes having a B statistic (it roughly indi-cates the logarithm of the odds that a gene is differentially expressedversus that it is not) greater than 0.

2.5. cDNA obtention, qRT-PCR experiments and data analysis

To validate the expression results obtained in the array analysis, 1 μgof the same mRNA used for the array hybridization was used forfirst-strand cDNA synthesis with Superscript II (Invitrogen). Primersand probes were designed to span an intron in all cases using the Prim-er3 software Version 0.4.0 (Rozen et al., 2000) (Table S2, supplementaryinformation). Real-time PCR was performed using the SYBR GreenReady Master Mix according to the manufacturer's instructions on theABI PRISM 7900HT Sequence Detection System (Applied Biosystems).The standard curve method was used for the analysis. The resultswere normalized respect to a housekeeping gene selected for its stableexpression among the different tissues and cell lines. A reagent-only(no DNA) negative control sample was always included in each run. Ex-periments were performed a minimum of 3 times in 384-well plateswith three replicates per sample. Raw data were obtained using SDS2.4 software (Applied Biosystems).

2.6. Chromatin immunoprecipitation

ChIP experiments were performed using home-made MEF cell lineswith endogenous TFII-I expression for mouse genes and Hela cell linewas used in human analysis. We selected three wild-type cells lines andtwo homozygous (Gtf2iΔex2/Δex2) and two heterozygous (Gtf2i+/Δex2)cell lines derived from embryos of Gtf2i+/Δex2 intercrosses. MEFs were

531M. Segura-Puimedon et al. / Gene 527 (2013) 529–536

crosslinked in 1% formaldehyde for 10 minutes, lysed and sonicated tosize smaller than 600 bp. To prevent unequal shearing of DNA sampleswe keep the tip end of the sonicator near the bottom of the sampletubes. Immunoprecipitations were performed overnight at 4 °C withTFII-I antibody (Lucena et al., 2010) and goat anti-mouse IgG (Sc-2028,Santa Cruz Biotechnology) used as negative control. Fifty microlitersof the chromatin supernatant was saved as input control. Primerswere designed to span the motif region and were located in its closevicinity. All primer sequences are shown in Table S2 (Supplementaryinformation).

2.7. Mutagenesis analysis

The sequence of interest of the promoter region of Birc1f contain-ing the newly described motif was inserted in a pGem®-T Easy Vector(Promega) and One Shot® TOP10 cells (Life Technologies) weretransformed with the vector. Nine clones were analyzed for insertionusing the EcoRI restriction enzyme. The recombinant clones were se-quenced and correct ones were selected for posterior experiments.Mutagenesis experiment was performed using the QuikChange Kitprotocol (Stratagene—an Agilent Technologies company) followingmanufacturer's instructions. The conserved guanine a 3′ of the motifwas substituted by an adenine. Five clones were selected and se-quenced for correct insertion, obtaining one correct clone. Mutagene-sis studies were made in COS7 (a monkey derived cell line) to abolishinterference with mouse specific primers. Transfection experimentswere performed using the jetPEI® reagent (Poliplus Transfection) fol-lowing the manufactures recommendations.

2.8. Pathway definition

In order to identify which were the affected pathways in the dif-ferent groups we used Consensus Path database release 24 (CPDB;http://cpdb.molgen.mpg.de/) (Kamburov et al., 2009), which containfunctional molecular interactions obtained from 13 publicly availableresources. We converted the RefSeq ID mouse genes in the humanorthologues using DAVID (DAVID Bioinformatics Resources 6.7)(Huang da et al., 2009a, 2009b).

Gene information was obtained from Entrez Gene (http://www.ncbi.nlm.nih.gov/gene) and Ensembl databases (www.ensembl.org).

2.9. Accession number

Accession numbers for supporting microarray data for review areas follows: http://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?token=zzcdzwaswqsaqjk&acc=GSE41251

http://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE48790

3. Results

3.1. Identification of deregulated genes in Gtf2i mutant mouse cortex andES cell line

In our effort to elucidate the role of TFII-I in the cognitive pheno-type of WBS also observed in different mouse models (Li et al.,2009; Lucena et al., 2010), we searched for differentially expressedgenes (DEG) by gene expression microarray analysis of cortex fromheterozygous Gtf2i mutant mice (Gtf2i+/Δexon2).

Previous studies of dissected brain regions had noticed considerablerandom variation among individual samples. To reduce the false posi-tives due to this random noise, cDNA was prepared from pools (fivemice per pool), extracted from Gtf2i+/Δexon2 and WT mice. After a re-strictive analysis, a total of 16026 probes could be analyzed in cortex.DEG were selected for the statistic B N 0. Around 10% of probes wereidentified as deregulated in cortex, and the magnitudes of thesechanges was generally small, ranking between −3.27 and 2.5-fold. In

order to validate themicroarray results we carried out qRT-PCR analysisin a subset of selectedDEG. The results confirmed themicroarray data inmagnitude for 19 of the 20 genes tested (Fig. 1A).

On the other hand, it is well established that undifferentiated stemcells are a very interesting tool for gene expression studies as they ex-press most genes in the organism, including many important for devel-opment. To identify genes already deregulated in early developmentalstages, we also studied the embryonic stem (ES) cell line XS0353 withthe Gtf2i+/Δexon2 genotype. After different corrections and quality con-trols of the results from two replicates, we discriminated 12402 cDNAclones. DEG were selected in each comparison for B N 0 and p b 0.01.Verification of array data was performed by qRT-PCR using 20 chosengenes: all 20 genes were deregulated in the same direction and magni-tude (Fig. 1B). Taken together, we found a very high correlation of mi-croarray data with qRT-PCR results.

Only 19% of the DEG found in the cortex was also deregulated inthe undifferentiated XS0353 cell line, five of them are validated inboth (Figs. 1A and B). To define the degree of concordance with pre-vious reports we also compared our data with the set of DEG found inWBS human skin fibroblasts (Henrichsen et al., 2011) and with theset of DEG found in WBS human lymphoblastoid cell lines comparingWBS patients with different size deletions versus controls (Antonellet al., 2010b). In the first case, we obtained 6 and 20 genes when com-paring respectively with the XS0353 or cortex DEG. In the secondcase, comparison gave us 5 genes in common with XS0353 and 10with cortex. We also compared our results with previous publicationin mouse embryonic fibroblasts (Chimge et al., 2007b). A total of 44genes were found in common between the 158 published DEG andthe XS0353; but only 6 with cortex DEG.

3.2. Recruitment of TFII-I to the proximal promoter of target genes

To identify potential direct target genes of TFII-I among the candidateDEG, we first searched the consensus TFII-I binding site BRGATTRBR onpositive strands within 1 kb (−1000 to +1) from the transcriptionstart site of each locus in the mouse genome (Chimge et al., 2008). Theconsensus binding motif was found in the promoter of ≈17% of DEG(both cell line and cortex DEG), which is not significantly enriched withrespect to the entire data set (DBTSS) (Wakaguri et al., 2008). Of thegenes with the motif, 50% are down-regulated while ≈28% are up-regulated. These results agree with the fact that the presence of themotif does not correlate with a specific direction of expression changes(Ouyang et al., 2009).

A set of 27 genes deregulated in cortex with no BRGATTRBR bind-ing motif was subjected to an in silico analysis with the MEME pro-gram (Bailey et al., 1994). After manual curation, a single novelmotif with a high degree of significance in multiple DEG (p =1.49 × 1012 − 1.72 × 10−7) was identified and selected. Phylogenet-ic footprints were generated by aligning the motifs in the context offlanking upstream sequences of selected genes (Fig. 2). A core con-served site containing the sequence 5′-CAGCCWG-3′ was identifiedwith strong evidence of further conservation in flanking sequences.Similar to the BRGATTRBR motif, the consensus CAGCCWG motifwas found in ≈17% of annotated promoters of the DEG (≈48%down-regulated and ≈27% up-regulated), which is not significantlyenriched with respect to the entire data set.

Six candidate target genes with the BRGATTRBR consensus bind-ing domain and 10 target genes with the regulatory CAGCCWG ele-ment were selected for further analyses. To assess whether TFII-Ican be recruited to promoters of selected target genes we performedchromatin immunoprecipitation (ChIP) analysis in MEFs that endog-enously expressed TFII-I proteins. As expected, our ChIP resultsshowed specific enrichment of TFII-I in promoters of 5 of the 6 select-ed genes containing the BRGATTRBR binding motif (Acta1, Cyp51,Pik3r1, Rpusd3 and Slc25a31) (Fig. 3A) and also for 9 of the 10 target

Fig. 1. Validation of deregulated genes by qRT-PCR. Selected genes found deregulated in microarray transcriptome data were validated by qRT-PCR comparing their relative expres-sion in mRNA. A. Cortex of Gtf2i+/Δexon2 versus wild-type mice. B. XS0353 versus AB2.2 cell lines. Fold changes are represented in dark grey for the array and in light grey for theqRT-PCR results.

532 M. Segura-Puimedon et al. / Gene 527 (2013) 529–536

genes with the CAGCCWG binding motif (Birc1f, Cnp1, Htr3a Jmjd5,Pik3r3, Shc1, Snw1, Utrn and Zfp280c) (Fig. 3B).

However, no binding was detected for Keap1 and Ptpn1 to theBRGATTRBR and CAGCCWG motifs respectively (Figs. 3A and B) inour conditions, despite the presence of the complete consensus bind-ing sequence.

Furthermore, ChIP assays in the human HeLa cell line showed anenrichment of TFII-I binding in the human promoter of NAIP6, thehuman orthologous of Birc1f (Fig. 3C) confirming the in vivo function-al relevance of this motif under our basal cell culture conditions.

3.3. The CAGCBVG motif is necessary for correct binding by TFII-I

To better define the relevant sites of CAGCCWG for TFII-I binding,we performed additional ChIP analyses in five genes containing simi-lar domains in their promoters with small variations with respect tothe consensus. We could observe positive binding in genes that con-tain a variation of the consensus sequences (CAGCBVG) in their pro-moters, such as Ascc2, Nnat, Slc24a3 and Igfbp3 (Fig. 4A). In addition,we substituted the 3′ conserved guanine for adenine by targeted mu-tagenesis. This mutation was sufficient to abrogate TFII-I binding inthe Birc1f gene (Fig. 4B), demonstrating the importance of this nucle-otide into the consensus domain in TFII-I binding to DNA.

3.4. Deletion of functional TFII-I domains abolishes its binding to the nov-el motif

The Gtf2i+/Δexon2mutant mice generated a truncated protein lackingthe first 140 amino acids (Δ140TFII-I), then missing the leucine zippermotif (residues 23 to 44) and disrupting the first I-repeat (residues104 to 176). In vitro studies have shown that the deletion of theN-terminal 90 amino acids of a TFII-I-Δ isoform not only affects itsDNA binding ability but also the proper homomerization leading to a

fail to activate transcription on TFII-I-dependent reporters (Cheriyathet al., 2001). Consequently, we decided to examinewhether the expres-sion ofΔ140TFII-I could also alter transcriptional regulation through thedescribed motifs.

We performed ChIP experiments into the promoters of positivegenes in two genetically modified mouse embryonic fibroblast celllines (MEFs) with either homozygous (Gtf2iΔexon2/Δexon2) or heterozy-gous (Gtf2i+/Δexon2) mutations expressing Δ140TFII-I. No clear bind-ing pattern was found, suggesting that the binding is gene promoterdependent. In some cases (Rpusd3 for BRGATTRBR, and Snw1 andZfp280c, for CAGCCWG), the presence of wild-type TFII-I in heterozy-gous animals form was enough for binding (Fig. 5A). In the case ofCyp51 and Slc25a31 for BRGATTRBR and Jmjd5 for CAGCCWG, thecomplete lack of binding in both homo and heterozygous cells sug-gested that proper TFII-I dimerization could be required for binding(Fig. 5B). In the case of Pik3r1 for BRGATTRBR, no binding wasfound in the heterozygous but normal binding in homozygous cells,which would suggest that abnormal stability of the dimers might in-terfere with binding (Fig. 5C). For genes like Birc1f and Utrn, the nor-mal binding in both, homo and heterozygous cell lines, indicated thatthe first 140 amino acids are not required for TFII-I binding (Fig. 5D).

By qRT-PCR we analyzed the expression levels of some of thesegenes in both genotypes. We observed that Rpusd3, Jmjd5 and Birc1f,with different promoter binding pattern in the mutant cells (Figs. 5A,B and D) showed the same pattern of expression; they are all genesup-regulated in the heterozygous genotype and even more in the ho-mozygous genotype. Genes validated in XS0353 like Cyp51 with notbinding in any mutant cells (Fig. 5B) showed limited down-regulationin heterozygous genotype (−1,29) and a more clear down-regulationin homozygous genotype (−1,45). In the other hand, we observednot deregulated expression for Pik3r1with not binding in heterozygousbut binding in homozygous mutant cells (Fig. 5C). These observationsare in agreementwith previous reports that pointed to a lowcorrelation

Fig. 2. Motif description and conservation. The evolutionary conservation of a putative DNA-binding motif in different genes and organisms allowed the definition of a conserved5′-CAGCCWG-3′ sequence. Identical nucleotides are starred. The height of the letters in the Logo picture correlates with degree of conservation.

533M. Segura-Puimedon et al. / Gene 527 (2013) 529–536

between promoter binding and expression of the corresponding genes(Bayarsaihan et al., 2012).

3.5. Functional annotation of TFII-I modulated genes

Finally, we analyzed whether deregulated genes in cortex and/orES cell line, harboring the different TFII-I binding motifs shared func-tional pathways. Overall, 347 TFII-I modulated genes were shownto bear the two binding motifs, 542 DEG bear only the consensusBRGATTRBR and 500 DEG bear only the consensus CAGCCWG, intheir proximal promoters (Fig. 6). CPDB analysis of the first set ofgenes revealed the cell cycle, gene expression and transcription, asthe most significant pathways regulated by TFII-I. We could also ap-preciated significant enrichment (P b 0.03) for genes involved inEGFR1 signaling, PI3K/AKT activation and TGFβ receptor signaling,among others. These pathways were also significantly enriched in

the set of TFII-I modulated genes bearing only one of the motifs. Aspecific enrichment in the signaling pathway by NGF was significantonly in the set of 500 TFII-I modulated genes bearing the CAGCCWGmotif (P = 0.0014) (Table 1).

4. Discussion

Decreased levels of TFII-I protein during development appear tocontribute to several relevant features of the WBS phenotype, includ-ing craniofacial shape, abnormal sound sensitivity, sensorimotor gat-ing phenotype and altered cognitive profile, as observed in differentmouse models (Li et al., 2009; Lucena et al., 2010). Containing ahelix–loop–helix, a DNA binding domain, a leucine zipper (dimeriza-tion motif) at the N-terminus and six unique I-repeats with DNAbinding properties, TFII-I is capable of partnering with a vast arrayof both cytoplasmic and nuclear factors, thus affecting diverse signal

Fig. 3. In vivo recruitment of TFII-I to proximal promoters of target genes. A. ChIP anal-ysis of the selected genes containing the BRGATTRBR binding domain in their promoterregion. B. ChIP analysis of the selected genes containing the CAGCCWG binding domainin their promoter region. C. ChIP analysis of the NAIP6 conserved CAGCCWG motif inthe HeLA human cell line. Reln was the negative gene control for TFII-I immunoprecip-itation in all cases.

Fig. 5. ChIP analyses of several gene promoters in Gtf2i+/Δexon2 and Gtf2iΔexon2/Δexon2 celllines. A. The presence of wild-type TFII-I in Gtf2i+/Δexon2 heterozygous cells was enoughfor binding at these three gene promoters (Rpusd3, Snw1 and Zfp280c). B. No bindingwas obtained in any of the mutant cell lines at these three gene promoters (Cyp51,Slc25a31 and Jmjd5). C. At the Pik3r1 promoter, binding was found only in theGtf2iΔexon2/Δexon2 homozygous cell lines. D. At the Birc1f and Utrn promoters, bindingwas present in both mutant cell lines.

534 M. Segura-Puimedon et al. / Gene 527 (2013) 529–536

transduction cascades and modulating the expression of variousgenes (Ashworth et al., 2007; Caraveo et al., 2006; Ogura et al., 2006).

We have studied gene expression profile differences induced byaltered TFII-I function in ES cells and adult brain tissues in order toidentify deregulated genes and pathways. Unexpectedly, we onlyfound 19% of common DEG between adult cortex tissue and ES cells.A major degree of concordance (30%) can be found between DEG re-ported in mouse embryonic fibroblast (Chimge et al., 2007a) and ourES cells, but expected as consequence of different expression patternsamong cells with adult brain tissues (4%). Similarly, little concordanceis found among our analysis and the ones carried out in patients sam-ples (lymphoblast and fibroblast) (Antonell et al., 2010b; Henrichsenet al., 2011). This low concordance is reasonable taking into accountthat in the human studies all the genes in the WBS are deleted com-pared to only Gtf2i in the present study.

Deregulated genes were used as tools to look for novel highly con-served DNA binding site for TFII-I. We identify a new conserved

Fig. 4. Variability of the CAGCCWG conserved sequence. A. ChIP analysis of target genescontaining variations in the conserved motifs, showing that the presence of CAGCBVGwas enough for binding. B. Following mutagenesis of the conserved 3′ guanine to anadenine, binding of TFII-I to the Birc1f promoter was no longer present.

binding motif and experimental validation of this binding site forsome deregulated genes was documented. Conservation of upstreamelement containing the putative TFII-I-binding site goes from mam-mals to fish. For humans, we could observe, an enrichment of TFII-I

Fig. 6. Four-way Venn diagram of identified target genes. Target genes were obtainedfrom the DBTSS for the two binding motifs and from array expression analyses forthe cortex and the ES cell line.

Table 1Pathway involvement of TFII-I modulated genes.

Genes Pathway Count P value

TFII-I modulated geneswith both motifs in theirproximal promoters(n = 347)

Cell cycle 21 0.0001663Gene expression 39 0.0006854Insulin signaling 10 0.0014353Glucose transport 5 0.0022387EGFR1 21 0.0023685CDO in myogenesis 4 0.0049139Actin organization and cellmigration by PI3K

3 0.0066944

PI3K/AKT activation 4 0.0104319Ubiquitin mediated proteolysis 8 0.0148252Signaling by VEGFR1 andVEGFR2

5 0.0234935

TGFβ receptor signaling 4 0.0287998TFII-I modulated geneswith BRGATTRB motif intheir proximal promoters(n = 542)

DNA replication 18 0.0000078Glycolysis and gluconeogenesis 10 0.0000081TNFα 20 0.0000369EGFR1 29 0.0012183VEGFR1 specific signals 5 0.0033832Purine metabolism 5 0.0051766Cell cycle 25 0.0005387TGFβ receptor signaling 12 0.0193252Actin organization and cellmigration by PI3K

3 0.0195390

Insulin signaling 5 0.0225993PI3K cascade 6 0.0381539

TFII-I modulated geneswith CAGCWG motif intheir proximal promoters(n = 500)

EGFR1 35 0.0000017Cell cycle 28 0.0000279DNA replication 14 0.0005531Signalling by NGF 7 0.0014121TNFR1 signaling 4 0.0018261TNFα 15 0.0031993NCAM signaling for neuriteout-growth

3 0.0045506

TNFR22 signaling 3 0.0059235Pyrimidine metabolism 10 0.0120670Purine metabolism 11 0.0180500TGFβ signaling 3 0.0253695

535M. Segura-Puimedon et al. / Gene 527 (2013) 529–536

binding in the promoter of NAIP suggesting a functional role for thissequence. In agreement with our previous results about the in vivorelevance of the N-terminal part of TFII-I (Lucena et al., 2010), weprovide new experimental data demonstrating the importance ofthis region to maintain binding capacity of TFII-I to different promotersequences of various genes and thus, altering its transcriptional activ-ity. No correlation exists between the presence of any of the motivesanalyzed and that the expression is being affected in a certain direc-tion in the array results. In both cases we find that from the totalbinding motif genes, 50% are down-regulated and about 30% up-regulated following the same pattern as in the overall analysis inwhich 68% of the genes appear down-regulated. Similar conclusioncould be extracted after analysis between binding motif and levelsof expression in the different mutant genotypes, in agreement withprevious reports (Bayarsaihan et al., 2012).

A fact demonstrated by ChIP validations performed in experimentswith the same cell type but with different genotypes, is that the bindingproperties of the motif are not identical in all genes. Assuming that themutant TFII-I studied is affected in their ability to form heterodimers(Lucena et al., 2010), and that analysis here is performed with endoge-nous expression of all genes, we propose that the susceptibility toforming and binding of these complexes is different among genes, per-haps due to the genomic context in which they are found.

In linewith previous reports, pathway classification of TFII-I putativetargets showed significant enrichment in genes involved in glycolysisand gluconeogenesis (with experimental validation for Cyp51) and in-sulin pathway (Antonell et al., 2010b), suggesting a possible contribu-tion of GTF2I to the glucose intolerance or diabetes present in 75% ofadult WBS patients (Pober, 2010).

TFII-I has been previously reported to have a role in the PI3K/AKTpathway (Chimge et al., 2012). We further demonstrated this rele-vant function with the experimental validation of direct regulationof the Pik3r1 gene encoding the 85kD regulatory subunit of phos-phatidylinositol 3-kinase (PI3K). Other putative targets of TFII-I in-clude the Cab39l, Pik3ca, Stk11, Irs1, Gab1, Eif4ebp1, Trib3, Eif4b, Ptenand Rps6kb2 genes, implicating a significant enrichment of PI3K cas-cade and Nup88, Pik3c2a, Nup85, Atp8b1, and Plcd4 showed a signifi-cant enrichment of the phosphatidylinositol phosphate metabolism.The PI3K/AKT signaling pathway plays a key role in diverse physio-logic processes, including dendritic spine formation during develop-ment and structural synaptic plasticity (Diez et al., 2012).

It has been previously shown that TFII-I proteins play an impor-tant role in the regulation of the TGFβ signaling pathway (Ku et al.,2005; Stasyk et al., 2005). Here we provide experimental evidencethat TFII-I is recruited to the promoter of Shc1 and Snw1, membersof the TGFβ signaling pathways.

The 5-hydroxytryptamine (serotonin) receptor 3A gene (Htr3a)appears to be another new target of TFII-I. Interactions between theprefrontal cortex and the amygdala are of great interest in the inves-tigation of neural mechanisms underlying part of the WBS phenotypebecause of their role in anxiety and social cognition. In mouse modelswith mutations of TFII-I family members, alterations in serotonin5-HT1A currents have been demonstrated in the prefrontal cortex(Proulx et al., 2010). Our data suggest that TFII-I could participate inthese alterations through its direct regulation of Htr3a, a gene in-volved in serotonergic synapse.

5. Conclusions

TFII-I is an interesting and unusual transcription factor, particular-ly in light of its role in the pathogenesis of the neurodevelopmentaldisorder, WBS. We have expanded the knowledge about TFII-I func-tion by defining a novel regulatory element, CAGCCWG, identifyingadditional target genes and enriched pathways under TFII-I regula-tion. The final elucidation of the biological role of TFII-I may facilitatea deeper understanding of relevant molecular mechanisms underly-ing human development and cognition.

Supplementary data to this article can be found online at http://dx.doi.org/10.1016/j.gene.2013.06.050.

Author contributions

V.C. and L.A.P.J. conceived and designed the experiments. M.S.P.,V.C. and L.A.P.J. analyzed the results and wrote the article. M.S.P.,C.B. and V.C. performed the molecular and cellular biology experi-ments. All authors read and approved the final article.

Conflict of interest

The authors have no conflict of interest.

Acknowledgements

We thank Verena Terrado and q-Genomics for technical assistance.This work was supported by grants from the SpanishMinistry of Health(FIS 04/0433, to V.C.), AGAUR (2009 SGR1274) and the VI FrameworkProgramme of the European Union (LSHG-CT-2006-037627, to L.P.J.).M.S.-P. and C.B. were supported by a CIBERER and UPF Fellowships.V.C. is a FIS Investigator.

References

Antonell, A., et al., 2010a. Partial 7q11.23 deletions further implicate GTF2I andGTF2IRD1 as the main genes responsible for the Williams–Beuren syndromeneurocognitive profile. J. Med. Genet. 47, 312–320.

536 M. Segura-Puimedon et al. / Gene 527 (2013) 529–536

Antonell, A., et al., 2010b. Transcriptome profile in Williams–Beuren syndrome lym-phoblast cells reveals gene pathways implicated in glucose intolerance and visuo-spatial construction deficits. Hum. Genet. 128, 27–37.

Ashworth, T., et al., 2007. Cutting edge: TFII-I controls B cell proliferation via regulatingNF-kappaB. J. Immunol. 178, 2631–2635.

Bailey, T.L., et al., 1994. Fitting a mixture model by expectation maximization to discov-er motifs in biopolymers. Proc. Int. Conf. Intell. Syst. Mol. Biol. 2, 28–36.

Bayarsaihan, D., et al., 2012. Epigenetic modulation by TFII-I during embryonic stemcell differentiation. J. Cell. Biochem. 113, 3056–3060.

Breitkreutz, B.J., et al., 2001. AFM 4.0: a toolbox for DNA microarray analysis. GenomeBiol. 2 (SOFTWARE0001).

Caraveo, G., et al., 2006. Action of TFII-I outside the nucleus as an inhibitor of agonist-induced calcium entry. Science 314, 122–125.

Cheriyath, V., et al., 2001. Structure-function analysis of TFII-I. Roles of the N-terminalend, basic region, and I-repeats. J. Biol. Chem. 276, 8377–8383.

Chimge, N.O., et al., 2007a. Expression profiling of BEN regulated genes in mouseembryonic fibroblasts. Mol. Dev. Evol. 308, 209–224.

Chimge, N.O., et al., 2007b. Gene expression analysis of TFII-I modulated genes inmouse embryonic fibroblasts. Mol. Dev. Evol. 308, 225–235.

Chimge, N.O., et al., 2008. Identification of the TFII-I family target genes in the verte-brate genome. Proc. Natl. Acad. Sci. U. S. A. 105, 9006–9010.

Chimge, N.O., et al., 2012. PI3K/Akt-dependent functions of TFII-I transcription factorsin mouse embryonic stem cells. J. Cell. Biochem. 113, 1122–1131.

Dai, L., et al., 2009. Is it Williams syndrome? GTF2IRD1 implicated in visual-spatial con-struction and GTF2I in sociability revealed by high resolution arrays. Am. J. Med.Genet. A 149A, 302–314.

Diez, H., et al., 2012. Specific roles of Akt iso forms in apoptosis and axon growth reg-ulation in neurons. PLoS One 7, e32715.

Enkhmandakh, B., et al., 2009. Essential functions of the Williams–Beuren syndrome-associated TFII-I genes in embryonic development. Proc. Natl. Acad. Sci. U. S. A.106, 181–186.

Ferrero, G.B., et al., 2010. An atypical 7q11.23 deletion in a normal IQ Williams–Beurensyndrome patient. Eur. J. Hum. Genet. 18, 33–38.

Henrichsen, C.N., et al., 2009. Copy number variants, diseases and gene expression.Hum. Mol. Genet. 18, R1–R8.

Henrichsen, C.N., et al., 2011. Using transcription modules to identify expression clus-ters perturbed in Williams–Beuren syndrome. PLoS Comput. Biol. 7, e1001054.

Huang da, W., et al., 2009a. Bioinformatics enrichment tools: paths toward the compre-hensive functional analysis of large gene lists. Nucleic Acids Res. 37, 1–13.

Huang da, W., et al., 2009b. Systematic and integrative analysis of large gene lists usingDAVID bioinformatics resources. Nat. Protoc. 4, 44–57.

Irizarry, R.A., et al., 2003. Exploration, normalization, and summaries of high densityoligonucleotide array probe level data. Biostatistics 4, 249–264.

Kamburov, A., et al., 2009. ConsensusPathDB—a database for integrating human func-tional interaction networks. Nucleic Acids Res. 37, D623–D628.

Ku, M., et al., 2005. Positive and negative regulation of the transforming growth factorbeta/activin target gene goosecoid by the TFII-I family of transcription factors. Mol.Cell. Biol. 25, 7144–7157.

Lazebnik, M.B., et al., 2009. Williams–Beuren syndrome-associated transcription factorTFII-I regulates osteogenic marker genes. J. Biol. Chem. 284, 36234–36239.

Li, H.H., et al., 2009. Induced chromosome deletions cause hypersociability and otherfeatures of Williams–Beuren syndrome in mice. EMBO Mol. Med. 1, 50–65.

Lucena, J., et al., 2010. Essential role of the N-terminal region of TFII-I in viability andbehavior. BMC Med. Genet. 11, 61.

Makeyev, A.V., et al., 2010.Molecular basis ofWilliams–Beuren syndrome: TFII-I regulatedtargets involved in craniofacial development. Cleft Palate Craniofac. J. 48, 109–116.

Makeyev, A.V., et al., 2013. ChIP-Chip identifies SEC23A, CFDP1 and NSD1 as TFII-Itarget genes in human neural crest progenitor cells. Cleft Palate Craniofac. J. 50,347–350.

Misra, U.K., et al., 2009. Transcription factor TFII-I causes transcriptional upregulationof GRP78 synthesis in prostate cancer cells. J. Cell. Biochem. 106, 381–389.

Morris, C.A., 2010. The behavioral phenotype of Williams syndrome: a recognizablepattern of neurodevelopment. Semin. Med. Genet. 154C, 427–431.

Morris, C.A., et al., 1988. Natural history of Williams syndrome: physical characteristics.J. Pediatr. 113, 318–326.

Ogura, Y., et al., 2006. TFII-I down-regulates a subset of estrogen-responsive genesthrough its interaction with an initiator element and estrogen receptor alpha.Genes Cells 11, 373–381.

Ouyang, Z., et al., 2009. ChIP-Seq of transcription factors predicts absolute and differen-tial gene expression in embryonic stem cells. Proc. Natl. Acad. Sci. U. S. A. 106,21521–21526.

Palmer, S.J., et al., 2012. GTF2IRD2 from the Williams–Beuren critical region encodes amobile element-derived fusion protein that antagonizes the action of its relatedfamily members. J. Cell Sci. 125, 5040–5050.

Parker, R., et al., 2001. Identification of TFII-I as the endoplasmic reticulum stress re-sponse element binding factor ERSF: its autoregulation by stress and interactionwith ATF6. Mol. Cell. Biol. 21, 3220–3233.

Patterson, R.L., et al., 2002. Phospholipase C-gamma is required for agonist-inducedCa2+ entry. Cell 111, 529–541.

Pober, B.R., 2010. Williams–Beuren syndrome. N. Engl. J. Med. 362, 239–252.Proulx, E., et al., 2010. Enhanced prefrontal serotonin 5-HT(1A) currents in amousemodel of

Williams–Beuren syndrome with low innate anxiety. J. Neurodev. Disord. 2, 99–108.Roy, A.L., 2001. Biochemistry and biology of the inducible multifunctional transcription

factor TFII-I. Gene 274, 1–13.Roy, A.L., 2007. Signal-induced functions of the transcription factor TFII-I. Biochim.

Biophys. Acta 1769, 613–621.Roy, A.L., et al., 1997. Cloning of an inr- and E-box-binding protein, TFII-I, that interacts

physically and functionally with USF1. EMBO J. 16, 7091–7104.Rozen, S., et al., 2000. Primer3 on the WWW for general users and for biologist pro-

grammers. Methods Mol. Biol. 132, 365–386.Sakurai, T., et al., 2011. Haploinsufficiency of Gtf2i, a gene deleted in Williams Syn-

drome, leads to increases in social interactions. Autism Res. 4, 28–39.Smyth, G.K., 2004. Linear models and empirical bayes methods for assessing differen-

tial expression in microarray experiments. Stat. Appl. Genet. Mol. Biol. 3 (Article3).Stasyk, T., et al., 2005. Phosphoproteome profiling of transforming growth factor

(TGF)-beta signaling: abrogation of TGFbeta1-dependent phosphorylation of tran-scription factor-II-I (TFII-I) enhances cooperation of TFII-I and Smad3 in transcrip-tion. Mol. Biol. Cell 16, 4765–4780.

Tantin, D., et al., 2004. Regulation of immunoglobulin promoter activity by TFII-I classtranscription factors. J. Biol. Chem. 279, 5460–5469.

Tassabehji, M., 2003. Williams–Beuren syndrome: a challenge for genotype-phenotypecorrelations. Hum. Mol. Genet. 12, R229–R237 (Spec No 2).

Wakaguri, H., et al., 2008. DBTSS: database of transcription start sites, progress report2008. Nucleic Acids Res. 36, D97–D101.