Homeobox gene Dlx3 is regulated by p63 during ectoderm development: relevance in the pathogenesis of ectodermal dysplasias

During embryonic development and organ formation, a series of signals between epithelial cells and the underlying mesenchymal cells are the basis for the formation of a variety of appendages and/or organs(Pispa and Thesleff, 2003). Anomalies in epithelialmesenchymal-derived organs are characteristics of human pathological conditions defined as ectodermal dysplasias (EDs)(Priolo and Lagana, 2001).

Mutations in DLX3 and p63, among other genes, have been directly linked with EDs. The Dlx and p63 families of transcriptional effectors are essential for the development of the epidermis and/or embryonic appendages (Panganiban and Rubenstein,2002; Merlo et al.,2003; Morasso and Radoja,2005; Koster and Roop,2004). Dlx3 expression has been detected in the hair follicle, tooth, limb bud, branchial arches, labyrinthe layer of the placenta,osteoblasts and epidermis (Morasso et al.,1995; Morasso et al.,1999; Hassan et al.,2004). Here, we present evidence that Dlx3 is regulated by p63 as part of a transcriptional regulatory pathway relevant to specific EDs.

p63 regulates multiple signaling pathways, such as the bone morphogenetic protein (BMP) and fibroblast growth factor (FGF) pathways(Laurikkala et al., 2006; Barbieri and Pietenpol, 2006). Transgenic and knockout (KO) mouse models indicate that p63 has essential roles in the development and maintenance of the stratified epidermis(Yang et al., 1999; Mills et al., 1999; Koster et al., 2004; Koster and Roop, 2004). The p63 gene is transcribed from two distinct promoters, giving rise to proteins that either contain (TAp63) or lack (ΔNp63) the amino terminal transactivating domain. The TA and ΔN isoforms both possess the DNA-binding and oligomerization domains, and, by alternative splicing at the 3′ end, produce isoforms with different C-termini, termed alpha(α), beta (β) and gamma (γ)(Yang et al., 1998). Theα isoforms contain a sterile α motif (SAM) - a domain with reputed importance in protein-protein interactions(Qiao and Bowie, 2005). p63 isoforms act as transcriptional activators and/or repressors(Ghioni et al., 2002; King et al., 2003; Wu et al., 2005), and bind to two or more tandem repeats of RRRCWWGYYY, but preferentially activate the RRRCGTGYYY sequence (Osada et al.,2005).

Mutations in the p63 gene have been associated with EDs that include ectrodactyly-ectodermal dysplasia-cleft lip/palate (EEC), limb-mammary syndrome (LMS), split hand-foot malformation (SHFM) and ankyloblepharon-ectodermal dysplasia-clefting (AEC) syndrome. There is a correlation between the position of the mutation and the observed abnormal phenotype (van Bokhoven et al.,2001; McGrath et al.,2001; van Bokhoven and Brunner, 2002). Mutations in the DLX3 gene are linked to tricho-dento-osseus (TDO) syndrome, which, like AEC, is characterized by defects in the development of hair, teeth and bone, and by absence of overt limb malformations (Price et al.,1998).

Partial-overlapping mRNA expression and phenotypes of specific human malformations caused by molecular lesions in either p63 or DLX3 suggest that these genes are components of common signaling pathways during embryonic development. Here, we show that p63 is able to bind and transactivate Dlx3 both in vitro and in vivo. Mutant p63 proteins derived from AEC patients exhibit an impaired ability to transactivate Dlx3, indicating that the misregulation of the DLX3 gene is involved in the pathogenesis of human syndromes associated to AEC.

The -117 to +60 DNA fragment of the Dlx3 promoter(Park and Morasso, 1999) was inserted into the pGL3-Basic and pCAT-Basic vectors (Promega). Mutations in the p63-binding sites of the Dlx3 promoter were obtained using the ExSite Mutagenesis kit (Stratagene). The coding sequences forΔ Np63α, ΔNp63β, andΔ Np63γ were cloned into pBK-CMV (Stratagene). The TAp63α, TAp63β and TAp63γconstructs were a gift from E. Candi (University of Rome `Tor Vergata', Rome,Italy) and G. Melino (University of Rome `Tor Vergata', Rome, Italy). The p63 mutants L518F, L518V and Q540L (AEC); E639X (SHFM); FS525 (EEC); and G76W and ΔAA (LMS) were kindly provided by H. van Bokhoven (Radboud University, Nijmegan, The Netherlands).

Primary mouse keratinocytes were grown according to Park and Morasso(Park and Morasso, 1999). Normal human epidermal keratinocytes [NHEK; a gift of M. Simon (SUNY, Stony Brook, NY, USA)] were derived from newborn foreskin and cultured in K-SFM(Invitrogen). The human osteosarcoma Saos-2 cell line was maintained in RPMI 1640 and 10% FCS. The immortalized human keratinocyte HaCaT and H1299 non-small-lungcarcinoma cell lines were grown in DMEM and 10% FBS.

Transient transfections of keratinocytes were performed with FuGENE6(Roche) in a 1:3 ratio. PRL-SV40 vector was used as an internal control. Luciferase activity was measured 24-36 hours after transfection using the Dual-Luciferase Reporter Assay System (Promega). Transient transfections of Saos-2 and H1299 cells were performed according to Calabrò et al.(Calabrò et al., 2002). CAT reaction was performed 48 hours after transfection using 90 μg of cell extract. β-Gal was used to normalize for transfection efficiency.

Nuclear extracts and EMSA analysis were carried out according to Park and Morasso (Park and Morasso,1999) using the Dlx3 p63site1+2 and non-specific competitor AP-2 binding site.

Chromatin immunoprecipitation (ChIP) was performed with chromatin from mouse keratinocytes and H1299 cells transfected with TAp63α according to Caretti et al. (Caretti et al.,2004) using no antibody, IgG or p63 antibody (Santa Cruz, H137 and 4A4). Real-time PCR was performed using the Mx3000P System (Stratagene) and SyberGreen MasterMix (Applied Biosystems) with independent DNA samples and the following oligonucleotides: for mouse, Dlx3(F)5′-GAGAAAGCGCGAGCGTGTTTTGCC-3′ and Dlx3(R)5′-CCGGCTGTCGGTCAGTCGCTGCGT-3′; for human, DLX3(F)5′-AGAGAGGCGGAAGAGACGAG-3′ and DLX3(R)5′-GAGGAGGGAGGAGAGAAGGA-3′; and for JAG2(F)5′-CAAGTGGTGAACAAGGGAGACT-3′ and JAG2(R)5′-ACTGCTGCCTTCTGGAAACTC-3′. Data are presented as fold differences relative to input and values are obtained by IgG with the formula 2^{[(CtIgG-CtInput)-(CtAb-CtInput)]}, where Ct is threshold cycles,IgG is normal rabbit IgG, Ab is specific antibody and Input is input genomic DNA. ACHR amplification was performed as a control for H1299 transfected with TAp63α using ACHR(F) 5′-TGCCTCGGGTGAACTAAGATG-3′ and ACHR(R)5′-GCCTCATTCGTCTTGGGAACT-3′.

For analysis of the expression of p63 in mouse keratinocytes and embryonic tissues, the following oligonucleotides were used: TAp63(F)5′-AGACAAGCGAGTTCCTCAGC-3′, TAp63(R)5′-TGCGGATACAATCCATGCTA-3′, ΔNp63(F)5′-ATGTTGTACCTGGAAAACAATG-3′, ΔNp63(R)5′-GATGGAGAGAGGGCATCAAA-3′. These oligonucleotides are designed in regions of the mRNA common to all isoforms (α, β and γ), and do not distinguish between these variants.

For Dlx3 expression in mouse keratinocytes, the following oligonucleotides were used: Dlx3(F) 5′-ATTACAGCGCTCCTCAGCAT-3′ and Dlx3(R) 5′-GCCTATAGGATCCCCCGTAG-3′. In embryonic tissues: Dlx3(F)5′-CGTTTCCAGAAAGCCCGTA-3′ and Dlx3(R)5′-CGTGGAATGGGAAGATGTGT-3′. For normalization: GAPDH(F)5′-TGTCAGCAATGCATCCTGCA-3′ and GADPH(R)5′-TGTATGCAGGGATGATGTTC-3′.

For the analysis of p63 and Dlx3 mRNA levels in embryonic tissues, E10.5, E11.5 and E12.5 anterior or posterior limb buds from wild-type embryos were dissected, pooled in TRIZOL (Roche) and extracted, as indicated by the manufacturer. Real-time PCR was performed with a LightCycler (Roche)using FastStart DNA MasterPLUS SYBR-Green I (Roche). Standard curves were performed using wild-type cDNA with four calibration points: TQ; 1:3; 1:9 and 1:27. All samples were done in duplicates and the analysis was repeated twice. GAPDH was used for normalization, calculated using LightCycler Software 3.5.3. The results are expressed with the value relative to E10.5 (set at 1) for each mRNA.

Immunohistochemistry was performed on 11 μm cryostatic sections of E10.5 embryonic forelimbs. Sections were blocked with 10% goat serum in PBS for 40 minutes at room temperature. Antibodies used were: mouse monoclonal anti-p63(4A4, 1:100, Santa Cruz) and rabbit anti-distal-less [pan-anti-Dlx, 1:100;kindly provided by G. Boekhoff-Falk (University of Wisconsin Medical School,WI, USA)]. As secondary antibody, anti-mouse-Cy2, anti-rabbit-Cy3 (1:100;Jackson Immuno-Research) and Envision antirabbit HRP (Dako) were used. Fluorescence micrographs were taken by confocal microscopy.

Whole-mount in situ hybridization was performed according to Acampora et al. (Acampora et al., 1999) on E10.5 p63 KO embryos [Brdm2 line of p63 KO kindly provided by D. Roop (Baylor College of Medicine, Houston, USA)] using a Dlx3probe (Morasso et al.,1995).

Total RNA from human Saos-2 and HaCaT cells was prepared with TRIZOL(Roche). For reverse transcription (RT)-PCR, 3-4 μg of total RNA were reverse-transcribed using SuperScript II (Invitrogen). The following oligonucleotides were used: Dlx3(F) 5′-ACCTACGGAGCCTCCTACCG-3′,Dlx3(R) 5′-ACTCAGGTTCTGTGCGTGAT-3′, p63α(F)5′-GTCTCCATCTTCATATGGTAAC-3′, p63α(R)5′-CACACTGACTGTAGAGGCA-3′, p63β(F)5′-GTCTCCATCTTCATATGGTAAC-3′, p63β(R)5′-CTTGCCAAATCCTGACAATGCTGC-3′, p63γ(F)5′-GAGGATAGCATCAGAAAACAGCAAG-3′ and p63γ(R)5′-CTCCACAAGCTCATTCCTGAAGC-3′. For normalization: Cyclophillin(F)5′-ATCACCATTGCTGACTGTGG-3′, Cyclophillin(R)5′-ACTCTGCAATCCAGCTAGGC-3′, GAPDH(F)5′-GTCTCCATCTTCATATGGTAA-3′ and GAPDH(R)5′-CCACAGTCCATGCCATCACT-3′ were used.

The existence of malformations caused by either p63 or DLX3 gene mutations that translate to partially overlapping phenotypes suggests that these genes are transcriptional effectors in common signaling cascades regulating epidermal development. The severity of the phenotype in p63-null mice suggests that it is a crucial upstream regulator of these signaling pathways (Mills et al.,1999; Yang et al.,1999). Detailed analysis of the Dlx3 proximal promoter region revealed a sequence with two p63-like overlapping binding sites immediately upstream of the CCAAT box, located from -89 to -80 bp (site 1) and from -84 to -75 bp (site 2) of the transcriptional start site(Park and Morasso, 1999). Because the expression patterns of p63 and Dlx3 overlap throughout embryonic development (Morasso and Radoja, 2005), we proceeded to test the ability of different p63 isoforms to transactivate the Dlx3 promoter. The Dlx3_-117/+60 construct, which contains the two overlapping sites,was transiently transfected into primary mouse keratinocytes in either the absence or presence of expression vectors encoding TAp63α, TAp63β,TAp63γ, ΔNp63α, ΔNp63β or ΔNp63γ. The TA isoforms activated the Dlx3_-117/+60 promoter at a magnitude of twelve-, three- and seven-fold, respectively, compared to normal activation,whereas the exogenous expression of the ΔN isoforms resulted in a two-to four-fold greater transactivation compared with wild type(Fig. 1A). These effects are specific for p63 isoforms, because p53 did not transactivate the Dlx3promoter (data not shown).

In order to test each overlapping site (p63-binding-site 1 and p63-binding-site 2, Fig. 1B),we compared the activities of the Dlx3_-117/+60,Dlx3_-117/+60 p63M1 (mutated p63 site 1) and Dlx3_-117/+60p63M2 (mutated p63 site 2) constructs (Fig. 1B) by cotransfection performed in the presence of the TA andΔNp63 isoforms (Fig. 1C,D). Our results indicate that TAp63α induction of Dlx3 is mediated through either of the two p63 sites, whereas TAp63γ required an intact regulatory region. These results are not cell-type specific, because a similar profile was obtained using Saos-2 cells(Fig. 1E).

Since Dlx3 is induced in keratinocytes cultured in 0.12 mM Ca²⁺ (Park and Morasso,1999), we compared the endogenous expression of the p63isoforms with Dlx3 after 12- and 24-hours of 0.12 mM Ca²⁺treatment (Fig. 2A-C). The real-time PCR results showed a correlation between the upregulation of Dlx3 and TAp63 and the downregulation ofΔ Np63 mRNAs associated by Ca²⁺-induced differentiation, and are consistent with the recent report by King et al.(King et al., 2006). The specificity of the PCR products was corroborated by sequencing (data not shown). TAp63 proteins are found in normal adult tissues(Nylander et al., 2002) and during mouse embryonic development (Koster et al., 2004) (also our own data). Findings of Koster et al.(Koster et al., 2004) support a role for TAp63 as a molecular switch for the initiation of epithelial stratification. Our findings support a working model in which, once transactivated by TAp63α, Dlx3 will in turn regulate the expression of terminal differentiation markers(Morasso et al., 1996).

In order to demonstrate direct binding to the p63 region in the Dlx3 promoter, we performed EMSA with a fragment comprising -89 to-75 bp (Dlx3 p63 sites 1 and 2) using nuclear extracts from primary keratinocytes (NE). A shift was detected(Fig. 2D, lane NE), and the complexes were competed with a specific competitor(Fig. 2D, lane SC), but not with a nonspecific DNA competitor (Fig. 2D, lane NC). We next evaluated whether p63 bound this region of the Dlx3 promoter in vivo. In mouse keratinocytes, ChIP experiments were performed with a p63-specific antibody and analyzed by regular PCR(Fig. 2E) and real-time PCR(Fig. 2F). The data shows that p63 specifically binds to the Dlx3 promoter in vivo (eightfold higher than with IgG control). Moreover, ChIP experiments on TAp63α-transfected H1299 cells, which are devoid of p63, demonstrated the direct binding of TAp63α to the Dlx3 promoter (Fig. 2G).

To further explore the relationship of Dlx3 and p63 in vivo, we analyzed their colocalization by immunofluorescence on E10.5 embryonic forelimb sections with anti-p63 and anti-distal-less antibodies. The latter reagent recognizes Dlx3 in the limb ectoderm. p63 and Dlx3 immunoreactivity were found to colocalize in the same nuclei (Fig. 3A). Comparison of the expression of Dlx3, TAp63 andΔ Np63 in the limbs at embryonic stages E10.5, E11.5 and E12.5 was performed by real-time PCR. Between E11.5 and E12.5, the relative abundance of both TAp63 and Dlx3 mRNA increased from three-to eight-fold relative to their expression at E10.5 in the anterior (AL) and posterior limbs (PL) (Fig. 3B),whereas expression of ΔNp63 was only moderately increased. These results for p63 in the limb ectoderm of embryos at E10.5-E12.5 are in agreement with reported data (Koster et al., 2004) and show for the first time that a good correlation is observed between the expressions of Dlx3 and TAp63.

To provide further evidence that p63 is an upstream regulator of Dlx3, we studied the effect of p63 ablation on Dlx3 mRNA and protein expression in the Brdm2 p63 KO mice(Mills et al., 1999)(Fig. 3C-D). As assessed by immunohistochemistry, the abundance of Dlx3 protein was significantly reduced in p63-KO limb ectoderm (Fig. 3C). Analysis by whole-mount in situ hybridization with a Dlx3 antisense probe demonstrated that the absence of p63 led to a downregulation of Dlx3 (Fig. 3D).

We next studied the clinical relevance of p63-mediated regulation of Dlx3 expression. We examined the transcriptional activity of the Dlx3 promoter in the presence of p63 mutants causative of human AEC,EEC, SHFM or LMS (Fig. 4A) in the H1299 cell line. The p63 AEC mutants - L518V, L518F and Q540L -are all point substitutions within the SAM domain, present only in TAp63α and ΔNp63α. The p63αFS mutant contains a mutation found in EEC that generates a frameshift at amino acid 525, which leads to a premature stop. The E639X mutation, in exon 14, was isolated in a SHFM patient, whereas the 2-bp deletion, in exon 14 (ΔAA), was present in one family with LMS. The G76W mutation, in exon 3, was isolated from a LMS patient and affects all p63 isoforms. Co-transfection experiments were performed with expression vectors encoding TAp63 and ΔNp63 mutants. As shown in Fig. 4B and 4C, the AEC mutants failed to yield a significant level of reporter-gene expression despite having intact amino terminal and DNA-binding domains. These results point to a crucial role of the C-terminus of p63α on the regulation of Dlx3 transcription. The LMS-derived mutants G76W and ΔAA; as well as the FSEEC mutant, SHFM-derived E639X TA and ΔN proteins; showed a similar mode of regulation of Dlx3-promoter activity compared with the corresponding wild-type proteins(Fig. 4B,C). A similar profile,albeit with a lower amount of induction, was obtained upon transfection in HaCaT cells (data not shown). The level of expression of the mutant proteins was corroborated by immunoblot analysis with anti-p63 antibody(Fig. 4D).

The hampered ability of the AEC mutants, as well as the partial overlapping phenotypes of specific malformations caused by to p63 (i.e. AEC) or Dlx3 (i.e. TDO) gene mutations, suggest that Dlx3misregulation is involved in aspects of the pathogenesis of AEC. AEC is characterized by ectodermal dysplasia, ankyloblepharon and cleft lip with cleft palate, and by the lack of limb involvement. The absence of limb defects in AEC may reflect the possibility that a putative role of Dlx3 in the limb is compensated for by other Dlx proteins(Panganiban and Rubenstein,2002; Morasso and Radoja,2005).

To determine the modulation of endogenous Dlx3 by p63, we used Saos-2 cells to express wild-type and mutant p63(Fig. 4E). Interestingly,TAp63α increased Dlx3 expression(Fig. 4E, lane 2), and this effect was shared, although to different extents, with the EEC, SHFM and LMS mutants (Fig. 4E; lanes 6, 7 and 8). TAp63β did not alter Dlx3 levels(Fig. 4E, lane 3). Remarkably,an AEC mutant (Fig. 4E, lane 5)and the TAp63γ isoform abolished Dlx3 expression. Transfections of ΔNp63 isoforms, both wild-type and mutant, had no significant effect on Dlx3 transcription (Fig. 4E, bottom panel). TAp63α and TAp63γ are potent transactivators (Barbieri and Pietenpol,2006) and, in our in vitro studies, both isoforms were able to transactivate Dlx3 at similar levels. However, whereas TAp63αinduction was mediated through either of the two p63 sites in the Dlx3 promoter, TAp63γ required an intact regulatory region. Surprisingly, a slightly different outcome was obtained for the endogenous Dlx3 regulation, where upregulation of Dlx3 was detected with TAp63α and a complete downregulation was found when expressing TAp63γ in Soas-2 cells. To determine if these results could be attributed to cell context, the experiments were also performed in HaCaT cells. In these cells, overexpression of TAp63α showed upregulation of endogenous Dlx3, whereas TAp63γ once again caused a complete downregulation of Dlx3 expression (Fig. 4F). These differences might be attributed to the specific p63RE-CCAAT-box chromatin architecture. An important feature of the Dlx3 promoter is that the overlapping p63-binding sites are in close proximity to CCAAT box that binds NF-Y in keratinocytes(Park and Morasso, 1999). NF-Y is a general promoter organizer that presets chromatin structure locally. A recent report shows that p63α regulates the transcription of the hsp70 gene through interactions with NF-Y(Wu et al., 2005). Although we have not determined the significance of NF-Y and p63 interactions on Dlx3 transcriptional regulation, it might be proposed that there is a dual role for the overlapping p63-binding sites, and that Dlx3 will be transcriptionally active or repressed depending on the specific p63 isoform bound to the promoter, on which of the p63 sites is occupied and on interactions with NF-Y CCAAT binding factor.

Dlx3 and p63 both function as part of a complex series of cascades that ultimately lead to the formation of ectoderm-derived organs. Unraveling the function of each protein at specific times of embryonic development will prove to be complex because of the differential expression of the p63 isoforms in distinct tissues (Nylander et al.,2002) and the cross-regulation with other developmentally relevant signaling pathways [i.e. FGF, BMP, and Notch(Laurikkala et al., 2006; Nguyen et al., 2006)]. The characterization of p63 target genes promises to improve our knowledge of the signaling cascades that are directly involved in normal ectodermal development. In summary, our study proves a functional relationship between p63 and Dlx3, with Dlx3 demonstrated to be a direct target of p63. The findings also provide evidence that the misregulation of Dlx3 is involved in the pathogenesis of p63 molecular lesions in AEC.

We thank S. J. Stimpson, Y. Rivera, S. Bertuzzi, T. Lozito, A. Pollice, K. King and G. Caretti for helpful comments; H. van Bokhoven, E. Candi and G. Melino for their gifts of plasmids; M. Simon for NHEK; G. Boekhoff-Falk for the anti-distalless antibody; and D. Roop for providing the Bmdr2 p63-KO mouse line. The research was supported by the Intramural Research Program of the NIAMS of the National Institutes of Health; Telethon (GGP05056) to L.G.;Telethon (GGP030326), AIRC and MIUR to G.L.M.; CIB to V.C.; and G.R.M. is supported by a career award from Fondazione Telethon (TCP99003) and Fondazione San Paolo, Italy.