Deletion of p63 exon 13 in mice reveals C-terminal isoform–specific functions in epithelial development

Significance

p63 is essential for a proper development of epithelia. Mutations within the C-terminal domains of p63 are associated with severe skin disorders like AEC syndrome. Yet, the in vivo function of the p63 C-terminus is poorly understood. Here, we use a mouse model harboring a deletion of p63 C-terminus in keratin-14-expressing tissues. The resulting shorter p63α-depleted isoforms bind more frequently proteins associated with transcription and can be detected more readily at promoters, which leads to excessive expression of adhesion-related genes. Mice lacking p63α isoforms show abnormal adhesion of keratinocytes, systemic inflammation and die prematurely. Altogether, our findings highlight a critical role for p63α C-terminus domains in correctly orchestrating the transcriptional program to ensure proper formation of epithelia.

Abstract

The transcription factor p63 is an essential regulator of epithelial development. Yet, the complexity at the 3′UTR, which gives rise to the three distinct C-terminal protein isoforms (α, β, and γ), remains unresolved and opens an investigation on the in vivo role of the C-terminus. This region, codified by exon 13, harbors genetic mutations leading to AEC syndrome. Here, we generated a mouse with a deletion of p63 exon 13 in keratin-14-expressing tissues and employed transcriptome, genome-wide occupancy, and interactome studies to characterize the role of the p63 C-terminus in vivo. In this model mouse, the p63 protein is expressed at the correct level in time and space but predominantly as the β isoform instead of the α isoform, thereby providing insights into the function of the C-terminus. We show that p63β interacts more readily with the core promoter transcription machinery and p63α-depleted isoforms bind more frequently the promoter region of target genes, resulting in inappropriate overexpression of extracellular matrix organization genes in the skin. This leads to the aberrant adhesion of epidermal keratinocytes to the basal lamina and triggers systemic inflammation, growth abnormalities, and premature death. We found a significant role of the full-length ΔNp63a isoform which cannot be substituted by the other isoforms (β or γ). Our studies highlight a crucial role for p63α in correctly orchestrating the gene expression program to ensure proper formation of epithelia.

The transcription factor p63 is a master regulator of epithelial gene expression during embryonic development of the craniofacial tissues, limbs, thymus, and some simple and stratified epithelia. With regard to the latter, it guarantees epithelial stemness and homeostasis (1–4). TP63 gene transcription can be initiated at two different promoters, generating N-terminal variants, full-length TAp63 and N-deleted ΔNp63, whereas alternative splicing events at the 3′ end give rise to additional variants α, β, and γ (5). All isoforms share a DNA binding domain (DBD) and an oligomerization domain. The full-length TAp63 protein has an N-terminal transactivation domain (TAD) that is absent from the ΔNp63 isoform, yet both isoforms are able to regulate transcription (6, 7). The longer α isoform harbors sterile alpha-motif (SAM) and transactivation inhibitory (TID) domains at its C-terminus, whereas the β and γ isoforms lack both of these domains. The function of the SAM domain is unclear, but it seems to mediate protein–protein interactions, whereas the TID is responsible for transcriptional inhibition (8, 9). In addition, structural and biochemical studies indicate that the p63α C-terminus contains multiple regulatory sequences that undergo posttranslational modifications (8, 10, 11).

All the TAp63 isoforms show high affinity for canonical p53-responsive element-activating genes involved in cell cycle arrest and apoptosis (12, 13). The ΔNp63 isoforms instead activate a specific transcriptional program to allow proliferation, differentiation, and adhesion (14–17). At the genomic level, p63 binds to response elements located mainly at enhancer regions (18–22), where it interacts and collaborates with chromatin-remodeling complexes, including the DNA methyltransferase 3 alpha, the histone methyltransferase KMT2D, the histone deacetylases HDAC1 and HDAC2 (23–26), the nucleosome modifier SWItch/Sucrose Non-Fermentable (BRG1- or BRM-associated factors) complex, and the insulator protein CCCTC-binding factor CTCF (27, 28), to organize the three-dimensional chromatin structure and regulate chromatin accessibility, leading to a specific transcriptional signature.

The role of the N-terminal variants, that is, the difference between the TAp63 and ΔNp63 isoforms, has been extensively studied in vivo through the generation of ΔNp63- and TAp63-specific knockout mice (29–32). The developmental abnormalities observed in ΔNp63^−/− mice (33) clearly demonstrate that ΔNp63 plays an important role in epithelial biology. Conversely, TAp63^−/− mice develop normally but present abnormalities in cellular senescence that cause premature tissue aging and defects in lipid and glucose metabolism (32). TAp63 is also expressed in the primary oocytes that are arrested in meiotic prophase I within primary follicles (34, 35), and TAp63a activation is essential for eliminating oocytes with DNA damage via classic programmed cell death pathways. Therefore, TAp63 maintains the integrity of the female germline (36, 37). The TAp63α activation mechanism was genetically demonstrated in our previous work (9), which revealed that heterozygous female mice with one exon 13-deleted (Δ13p63^+/−) allele are sterile because of the rapid loss of the primary oocyte reserve. These findings indicate that the p63 C-terminus plays an essential role in the activity of the TAp63 isoform and that inhibitory regulation by SAM and TID are required in TAp63α-expressing tissues to control cell death (9). However, because these female mice are completely infertile, it was not possible to evaluate the role of the C-terminus in the ΔNp63 isoform. To gain insight into the contribution of the C-terminus in the ΔNp63α isoform (SI Appendix, Fig. S1A), we generated a conditional deletion of exon 13 under the control of the keratin-14 promoter (KRT14-Cre-mediated excision), leading to genetically modified mice in which exon 13 was deleted (KRT14;p63^Δ13/Δ13) to replace p63α with p63β in K14-expressing tissues, including the epidermis.

Here, we report that the resulting homozygous mice lacking the C-terminal domain of ΔNp63α exhibit an impaired growth phenotype, skin abnormalities, and systemic inflammation. Detailed transcriptome profiling and genome occupancy analyses revealed preferential binding of ΔΝp63β to specific promoters, leading to the increased expression of extracellular matrix (ECM)–related genes. The protein interactome generated via BioID followed by mass spectrometry analysis revealed a marked preference of ΔNp63β for cooperation with the basal transcriptional machinery, consistent with the observed enrichment of this isoform at proximal promoters. Taken together, our findings provide deep molecular insights into the distinct functions of the C-terminal domains of p63 in epithelial cells in vivo.

Results

K14 Conditional p63^Δ13/Δ13 Mice Predominantly Express the ΔΝp63β Isoform in the Epidermis.

To study the role of the ΔNp63 C-terminus in vivo (SI Appendix, Fig. S1A), we generated model mice with a floxed p63 exon 13 allele derived from our previous total knockout (9). To achieve homozygous conditional deletion of exon 13 of Trp63 (p63^Δ13/Δ13) in keratin-14 (K14)-positive cells, we mated heterozygous floxed mice (Trp63-Ex13^fl/+) with Tg(KRT14-cre) mice (Fig. 1A). The KRT14-cre;Trp63-Ex13^fl/fl homozygous pups (hereinafter referred to as p63^Δ13/Δ13) were obtained by crossing two TgKRT14-cre;Trp63-Ex13^fl/+ mice. We were able to detect the intact Trp63 exon 13 floxed allele in gDNA from K14/Cre-negative tissues such as the liver and brain and the deletion allele in the K14/Cre-positive epidermis of p63^Δ13/Δ13 mice (SI Appendix, Fig. S1B). Similarly, at the mRNA level, we observed Trp63 exon 13 expression specifically in the epidermis of p63^+/+ mice. As expected, p63^{Δ13/ Δ13} primary mouse keratinocytes expressed Trp63 mRNA lacking exon 13 (Fig. 1B). At the protein level, only a shorter p63β isoform was detected in p63^Δ13/Δ13 primary mouse keratinocytes, and there were no significant differences in the amount of p63 between the two genotypes (Fig. 1C). The loss of the C-terminal domains in p63^Δ13/Δ13 newborn mice in p63^Δ13/Δ13 back skin was also evident in comparisons of immunofluorescence staining with a p63α-specific antibody and a pan-p63 antibody (Fig. 1D). RNA sequencing analysis of epidermis from p63^+/+ and p63^Δ13/Δ13 newborn mice demonstrated that the remaining exons were expressed at the same levels (Fig. 1E). The ΔΝp63α isoform was detected predominantly in the p63^+/+ epidermis, and p63^Δ13/Δ13 mice predominantly expressed the ΔΝp63β isoform in the epidermis tissues, whereas the TAp63 isoform was undetectable in both genotypes (Fig. 1F).

Fig. 1.

K14-conditional p63^Δ13/Δ13 mice exhibit an impaired growth phenotype. (A) Schematic of the p63^Δ13/Δ13 allele and crossing with Tg(KRT14-cre) to obtain homozygous mice. The primers used for RT–PCR and antibodies used to determine the specific expression of the deleted isoform are shown in the boxes at the bottom. (B) Semiquantitative RT–PCR analysis of complementary DNA and western blotting of total protein extracts from proliferating cultured mouse keratinocytes of the indicated genotypes were performed. β-actin was used as a normalization control. The image shown is representative of three samples. (C, Left) Western blot of p63 levels (pan-p63 antibody) in total protein extracts from proliferating cultured mouse keratinocytes of the indicated genotypes. β-actin was used as a loading control. (Right) Densitometry analysis of the intensities of the upper (p63α) and lower (p63β) bands normalized to the intensities of the β-actin and p63α bands in the p63^+/+ sample. The data are shown as the means ± SDs. n = 4 independent keratinocyte preparations from n = 2 to 4 mice each. P was determined by two-way ANOVA. (D) IF staining of p63 in the back skin of newborn mice. (Scale bar, 25 μm.) (E) University of California, Santa Cruz (UCSC) Genome Browser screenshots showing epidermis RNA-seq peaks within the Trp63 gene. (F) Expression of the six p63 isoforms shown in (E), as analyzed by RNA-seq. The values are shown as the proportion of each isoform compared with the sum expression of all isoforms (TPM). n = 4 mice for each group. P was determined by two-way ANOVA.

p63^Δ13/Δ13 Mice Exhibit an Impaired Growth Phenotype.

p63^Δ13/Δ13 mice appeared normal at birth (SI Appendix, Fig. S2A) but presented impaired growth (Fig. 2A) and a 40% reduction in survival over 3 mo. Among the surviving p63^Δ13/Δ13 animals, a significant reduction in body weight was observed over 8 mo, starting from the first month (Fig. 2B). The gross phenotypes of the mice included features of premature aging, with altered skin pigmentation (tail and ears), hair loss, kyphosis, and signs of toe necrosis (SI Appendix, Fig. S2A). To better understand these phenotypes, we performed a detailed histological analysis of the organs of 2-mo-old mice (Fig. 2C and SI Appendix, Fig S2B). We observed thickening of the keratin layer in K14-positive tissues such as the skin and forestomach. We observed increased thickness of the squamous epithelium and increased loricrin expression at the squamous–glandular junction of the stomach, whereas no alterations were observed in the corpus or esophagus (SI Appendix, Fig. S2 C–E). Several visible alterations in K14-negative tissues were observed in 40 to 70% of the analyzed mice; for example, smaller lipid droplets in brown adipose tissue and smaller skeletal muscle fibers (Fig. 2D). Compared with those of wild-type mice, the food and liquid intake, activity, and respiratory exchange ratio of p63^Δ13/Δ13 mice were not significantly different (SI Appendix, Fig. S2F). However, the serum leptin concentration was decreased (Fig. 2E), and energy expenditure was increased (Fig. 2F), suggesting indirect physiological consequences of the changes in K14-positive tissues across the whole organism. Taken together, these findings show that the expression of p63β in K14-positive tissues leads to organ abnormalities, impaired growth, and premature death.

Fig. 2.

p63Δ13/Δ13 mice exhibit an impaired growth phenotype. (A) Gross phenotype and % survival of p63^+/+ and p63^Δ13/Δ13 mice up to 3 mo of age. (B) Weights of p63^+/+ and p63^Δ13/Δ13 mice up to 8 mo of age. n = 3 to 5 mice. P was determined by two-way ANOVA. (C) Sankey plot summarizing the results of the histological analysis and abnormalities observed in the tissues of p63^+/+ and ^p63Δ13/Δ13 mice. (D) Hematoxylin and eosin staining of brown adipose tissue and gastrocnemius skeletal muscle from p63+/+ and p63Δ13/Δ13 mice. (Scale bar, 200 mm.) (E) Violin plot of the serum leptin concentration (ng/mL) in 1-mo-old p63^+/+ and p63^Δ13/Δ13 mice. n = 14 (p63^+/+) and n = 10 (p63^Δ13/Δ13) mice. P was determined via unpaired Student’s t test. (F) Violin plot of energy expenditure (kcal × h⁻¹ × kg⁻¹) of 2-mo-old p63^+/+ and p63^Δ13/Δ13 mice calculated by metabolic cage parameter value analysis. n = 14 mice. P was determined by Student’s t test.

The Epidermal Structure and Transcriptome Are Compromised in p63^Δ13/Δ13 Mice.

A detailed characterization of the epidermis at 1 mo of age in p63^Δ13/Δ13 mice revealed a significant increase in epidermal thickness (Fig. 3A). Accordingly, the distributions of the epidermal markers K14, K10, and loricrin were altered in p63^Δ13/Δ13 mice. For example, the epidermis of p63^Δ13/Δ13 mice expressed p63 and K14 in the suprabasal layer and contained K14/K10 and K14/loricrin double-positive keratinocytes (Fig. 3B). The same analysis of skin thickness and marker evaluation at the newborn stage revealed normal development of epidermis and keratinocyte differentiation (SI Appendix, Fig. S3 A and B) in both genotypes. Importantly, the number of Ki67-positive basal keratinocytes was greater in both newborn and 1-mo-old p63^Δ13/Δ13 mice than in their wild-type littermates (Fig. 3B and SI Appendix, Fig. S3B). To investigate alterations in signaling pathways related to proliferation and differentiation, we performed total RNA sequencing of p63^+/+ and p63^Δ13/Δ13 epidermis tissues (Fig. 1E). We identified 34 and 109 genes significantly enriched in p63^+/+ or p63^Δ13/Δ13 epidermis tissues, respectively (abs(log2FC)>0.5 and P < 0.05) (Fig. 3C), with the top 60 significantly modulated genes shown as a heatmap in SI Appendix, Fig. S3C. We validated these data by analyzing the expression of 10 selected genes via RT–qPCR in epidermis tissues from different animals (SI Appendix, Fig. S3D). Among the top genes enriched in p63^Δ13/Δ13 tissues, genes related to the prostaglandin and G-protein coupled receptors (GPCR) pathways were identified (Fig. 3D), and the most significantly modulated genes included nidogen 2 (Nid2). Since local activation of the prostaglandin and GPCR pathways in p63^Δ13/Δ13 mouse skin can trigger systemic inflammation, we measured IL-1β and IL-6 serum levels and observed a significant increase in cytokine levels in p63^Δ13/Δ13 mice compared with those in p63^+/+ mice (Fig. 3E). In summary, we observed dysregulation of differentiation, upregulation of prostaglandin/GPCR pathways in the epidermis at the transcriptional level and activation of inflammation-related genes in p63^Δ13/Δ13 mice.

Fig. 3.

Epidermal structure and gene expression profiles are compromised in p63^Δ13/Δ13 mice. (A, Left) Hematoxylin and eosin staining of the skin of 1-mo-old p63^+/+ and p63^Δ13/Δ13 mice; magnified areas are shown on the right. (Scale bar, 50 mm.) (Right) Violin plot of epidermal thickness. n = 7 mice. P was calculated with Student’s t test. (B, Left) IF staining of epidermal markers (keratin-14, p63, keratin 10, and loricrin) and the proliferation marker Ki67 in the back skin of p63^+/+ and p63^Δ13/Δ13 mice; magnified areas are shown at the Bottom Right. (Scale bar, 25 μm.) (Right) Violin plot of the percentage of Ki67-positive basal cells. n = 7 mice. P was determined by Student’s t test. (C) Volcano plot of modulated genes in p63^+/+ and p63^Δ13/Δ13 mouse epidermis as determined by RNA sequencing. (D) Heatmap of z scores of the expression of genes from the GPCR and prostaglandin signaling pathways. (E) IL-1β and IL-6 serum concentrations (pg/mL) in 1-mo-old p63^+/+ and p63^Δ13/Δ13 mice. n = 14 (p63^+/+) and n = 10 (p63^Δ13/Δ13) mice. P was determined by unpaired Student’s t test.

ECM-Related Genes Are Upregulated in p63^Δ13/Δ13 Keratinocytes.

Given the observed alterations in gene expression at the tissue level, we questioned whether these changes were characteristic for all epidermal cells or intrinsic to only p63+/K14+ basal proliferating keratinocytes. Therefore, we isolated, cultured, and differentiated primary mouse keratinocytes in vitro. The expression of differentiation markers (Trp63, Krt14, Cdkn1a, Tgm1, and Tgm5, Lor, and Inv) was determined via RT–qPCR at 0, 1, or 3 d of CaCl₂-induced differentiation and found to be largely unchanged, with a significant increase in only Ivl expression confirmed at the protein level (SI Appendix, Fig. S4 A and B). Furthermore, the proliferation rate of cultured p63^Δ13/Δ13 primary keratinocytes was significantly increased, consistent with our observations in the epidermis (SI Appendix, Fig. S4C). Notably, these changes were not linked to differences in the expression or stability of the ΔNp63α and ΔNp63β proteins, as evident during the in vitro differentiation process (SI Appendix, Fig. S4B), or to the presence of a CHX/MG132 pulse (SI Appendix, Fig. S4D) in culture.

We then used cultured keratinocytes as a model to carry out more detailed gene expression profiling to study the impact of p63 different isoforms on epidermal proliferating basal cells and, consequently, on the entire skin. We performed total RNA sequencing of primary keratinocytes from both homozygous genotypes and identified 48 and 114 genes with significant enrichment in p63^+/+ and p63^Δ13/Δ13 keratinocytes, respectively (abs(log2FC) > 0.5 and P < 0.05) (Fig. 4A), with the top 60 significantly modulated genes shown in a heatmap in SI Appendix, Fig. S4E. We validated these observations by RT–qPCR of nine selected genes via in additional pools of keratinocytes from different animals (SI Appendix, Fig. S4F). Gene set enrichment analysis via Mouse Reactome pathway collection (GSEA mReactome) revealed that the most significantly altered pathways in p63^+/+ cells were associated with translation, whereas genes enriched in p63^Δ13/Δ13 keratinocytes were associated with ECM organization and adhesion (Fig. 4 B and C). Overall, the genes upregulated in p63^Δ13/Δ13 keratinocytes were also upregulated in p63^Δ13/Δ13 epidermal tissues and vice versa (Fig. 4D); however, only 26 genes were significantly upregulated both in p63^Δ13/Δ13 epidermises and keratinocytes, while there were no common p63^+/+ enriched genes (SI Appendix, Fig. S4G). The shared 26 genes enriched in p63^Δ13/Δ13 animals included basement membrane nidogen 2 (Nid2), metallocarboxypeptidase (Cpz), calcium/calmodulin-dependent protein kinase IV (Camk4), beaded filament protein gene (Bfsp2), and type I transmembrane receptor embigin (Emb) involved in ECM pathway.

Fig. 4.

ECM-related genes are upregulated in p63^Δ13/Δ13 keratinocytes. (A) Volcano plot of modulated genes in cultured p63^+/+ and p63^Δ13/Δ13 mouse keratinocytes as determined by RNA-seq. (B) Gene set enrichment analysis (GSEA) pathway enrichment analysis of the differentially expressed genes. ECM-related pathways are highlighted in beige. (C) Heatmap of z scores of the expression of genes from the ECM organization pathway. (D) Boxplot showing z scores of the expression of the top overexpressed genes in cultured keratinocytes or epidermis tissues of p63^+/+ and p63^Δ13/Δ13 mice. n = 3 mouse keratinocyte preparations or n = 4 epidermis tissues for each group. P was calculated with the Wilcoxon matched-pairs test. (E) Electron microscopy of skin samples from 3-mo-old p63^+/+ and p63^Δ13/Δ13 mice. Schematics depict the presence of gaps and a reduced number of hemidesmosomes. (Scale bar, 1 mm.)

Given the amplitude of increased expression and the large number of genes enriched in p63^Δ13/Δ13 keratinocytes, we focused our attention on the “ECM organization” pathway. Multiple genes associated with collagen production, hemidesmosome organization, and adhesion to the basal lamina were upregulated in p63^Δ13/Δ13 keratinocytes (Fig. 4C). We detected increased expression of additional genes (integrins and laminins) from related pathways via RT–qPCR in six pools of keratinocytes collected from different animals (SI Appendix, Fig. S4H). The overexpression of several ECM proteins was previously reported in the basal cells of human psoriatic tissues (38). We therefore investigated the same genes in our in vivo model and observed a trend toward upregulation of their expression in keratinocytes of p63^Δ13/Δ13 mice (SI Appendix, Fig. S4I).

To assess the impact of aberrant expression of ECM genes on skin organization, we used electron microscopy to analyze the skin of 2-mo-old mice (Fig. 4E). Compared with that of their wild-type littermates, p63^Δ13/Δ13 skin displayed gaps (lacunae) in the basal lamina and presented a reduced number of hemidesmosomes. In summary, the loss of the p63 C-terminus in keratinocytes leads to the overexpression of ECM-related genes and the alteration of keratinocyte adhesion to the basal lamina.

p63α-Depleted Isoform Occupancy Is Enriched at Promoters.

To understand whether the observed changes in gene expression in p63^Δ13/Δ13 keratinocytes were due to altered p63 DNA binding, we performed ChIP-seq in p63^+/+ and p63^Δ13/Δ13 keratinocytes using a pan-anti-p63 antibody. We identified 28395 and 35024 p63 peaks in p63^+/+ and p63^Δ13/Δ13 keratinocytes, respectively (hereafter, p63α and p63Δα peaks) (SI Appendix, Fig. S5A). The number of peaks, number of genes associated with the peaks and peak distribution within the genome in wild-type keratinocytes were comparable to those from a previously published ChIP-seq analysis carried out in a similar model (SI Appendix, Fig. S5B) (39). As expected, the p63α and p63Δα peaks had a canonical p63 motif (SI Appendix, Fig. S5C). While most of the peaks were shared between p63α and p63Δα isoforms, we identified 3,805 p63α-specific peaks and 10,668 p63Δα-specific peaks, i.e., peaks that were significantly enriched in only one of the two samples (Fig. 5A). Notably, the p63Δα-specific peaks accounted for almost one-third of all p63Δα peaks, and those peaks were less significantly enriched for the p63 motif than the p63α-specific or p63-shared peaks were (Fig. 5B). The shared and p63α-specific peaks were mainly intronic or distal intergenic, with only ~7 to 10% of the peaks found within promoters (0 to 5 kb from the transcription start site (TSS)), which is consistent with the p63 binding landscape in human keratinocytes (21). Unexpectedly, approximately 30% of the p63Δα-specific peaks were located within promoters with higher level of significance compared with p63α-specific peaks (Fig. 5 C, Top). Analysis of p63α and p63Δα isoform occupancy at ENCODE-predicted cis-regulatory elements (CREs) revealed a notable increase in p63Δα isoforms binding at promoter-like regions, whereas p63Δα isoform occupancy at distal intergenic-like elements was comparable between the two genotypes (Fig. 5 C, Bottom). Notably, the distribution, i.e. % of peaks located within promoters or distal regions, of p63Δα-specific peaks was similar to the landscape of human p53 in different cell lines (SI Appendix, Fig. S5D).

Fig. 5.

p63α-depleted isoform occupancy is enriched at promoters. (A) Heatmap showing the intensity of p63 occupancy across all p63 peaks in the input or p63 ChIP-seq samples from p63^+/+ and p63^Δ13/Δ13 keratinocytes. p63 peaks that reached statistical thresholds in either p63^+/+ or p63^Δ13/Δ13-only samples are denoted as “specific,” whereas peaks present in both samples are denoted as “shared.” (B, Top) Venn diagram showing overlap between identified p63 peaks from (A). (Bottom) MEME motif enrichment analysis of shared or specific peaks. (C, Top) Distribution of peak locations on DNA. The P value and odd ratio of specific peak enrichment within promoter regions by the Fisher exact test are shown below. (Bottom) Heatmap showing the intensity of p63 occupancy across mouse ENCODE-predicted CREs (promoter-like and distal enhancer-like). (D, Top) Venn diagram showing the number of genes associated with shared or specific peaks. (Bottom) % of genes with p63 peaks within proximal elements, distal elements, or both. Schematic illustrating the more frequent p63β binding at proximal regions of p63 target genes. (E) GSEA pathway enrichment analysis of modulated genes bound by p63Δα. ECM-related pathways are highlighted in beige. The full graph is shown in SI Appendix, Fig. S5G. (F) UCSC Genome Browser screenshots showing the distribution of p63 signals, peaks, and p63 motifs at the Col18a1 and Itga2 loci. (G) Boxplots showing z scores of the expression of p63Δα-bound genes in the “ECM organization” Reactome pathway. The values are shown for p63^+/+ and p63^Δ13/Δ13 keratinocytes. n = 3 mouse keratinocyte preparations. P was calculated with the Wilcoxon matched-pairs test. (H) Heatmap showing the intensity of the H3K27ac ChIP-seq signal across all H327ac peaks in p63^+/+ and p63^Δ13/Δ13 keratinocytes. (I) Venn diagram showing the number of shared and unique superenhancers (SEs) in p63^+/+ and p63^Δ13/Δ13 keratinocytes. (J) UCSC Genome Browser screenshots showing the distribution of H3K27ac signals, peaks, and p63Δα-specific peaks and identified SEs at the Col18a1 and Itga2 loci.

Overall, the p63α- and p63Δα-bound regions were associated with approximately the same set of genes, with 79% overlap (SI Appendix, Fig. S5E), and most of the specific peaks were related to the shared p63α/p63Δα genes (Fig. 5 D, Top) rather than to distinct genes. Indeed, a greater proportion of p63Δα-bound genes than p63α-bound genes had peaks within both the promoter (0 to 5 kb from the TSS) and distal regions (5 to 100 kb from the TSS) (Fig. 5 D, Bottom). Furthermore, p63Δα bound target genes at greater numbers of regions within promoters than did p63α (SI Appendix, Fig. S5F).

Genes bound specifically by p63Δα were strongly associated with ECM-related pathways (Fig. 5E and SI Appendix, Fig. S5G); for instance, the Col18a1 and Itga2 loci were more frequently bound by p63Δα isoforms (Fig. 5F). The overall expression of p63Δα-specifically bound ECM-related genes was significantly upregulated in keratinocytes of p63^Δ13/Δ13 mice (Fig. 5G). In contrast, no pathways were significantly enriched among the genes with p63α-specific binding (SI Appendix, Fig. S5G). Of note, most of modulated ECM-related genes were already bound by p63α in p63^+/+ keratinocytes and these regions had a canonical p63 motif (SI Appendix, Fig. S5H). Collectively, our findings indicate that p63Δα isoforms bind promoters of p63 targets more readily and thus with higher density, thereby resulting in the increased expression of ECM-related genes.

p63 is recognized as a pioneer transcription factor that is able to bind closed chromatin and recruit chromatin modelers to increase chromatin accessibility (18). Therefore, we wondered whether increased genome occupancy of p63Δα isoforms was associated with changes in the chromatin state. To address this issue, we performed ChIP-seq for acetylated lysine 27 of histone 3 (H3K27ac) (Fig. 5H and SI Appendix, Fig. S5I). Overall, there were no notable changes in H3K27ac distribution across common H3K27ac peaks (Fig. 5H). No changes were detected at p63α/p63Δα -specific bound loci (SI Appendix, Fig. S5J) or at promoter-like or distal element-like predicted CREs in p63^Δ13/Δ13 versus p63^+/+ keratinocytes (SI Appendix, Fig. S5K). The only exceptions were a few loci with significantly enriched p63 and H3K27ac levels near genes that were dramatically upregulated in p63^Δ13/Δ13 keratinocytes and epidermis tissues, such as Nid2, Myot, Fut9, Grm4, and Dpp6 (SI Appendix, Fig. S5L).

Moreover, moderate changes in histone acetylation at specific loci resulted in a superenhancer (SE) loss or gain of ~40% (Fig. 5I). Compared with their wild-type counterparts, genes related to stable SEs presented only minor differences in gene expression in p63^Δ13/Δ13 keratinocytes, whereas genes associated with lost SEs tended to be expressed at lower levels in p63^Δ13/Δ13 keratinocytes (SI Appendix, Fig. S5M). The genes associated with SE gain, including Col18a1 and Itga2, exhibited strong increases in gene expression (Fig. 5J) and were enriched in the ECM and focal adhesion pathways (SI Appendix, Fig. S5N). Taken together, these findings indicate that p63Δα isoforms lead to the formation of new SEs associated with ECM/focal adhesion genes and thereby promotes their gene expression.

ΔNp63α Preferentially Interacts with the General Transcription Machinery.

Our data suggest that the loss of the C-terminus of p63 results in its relocation to new chromatin regions, but that those regions are also enriched for p53 family motifs. Thus, changes in p63 occupancy are unlikely to be due to altered DNA sequence recognition. Therefore, we hypothesized that the absence of the carboxyterminal domains might alter the p63 interactome. To address this possibility, we performed BioID mass spectrometry analysis of the p63 interactome in human 293 T cells. This technique utilizes the enzyme BirA* to biotinylate proteins that are in physical proximity to one another in the presence of extracellular biotin.

For BioID, we generated cell lines with doxycycline-inducible expression of human ΔΝp63α or ΔNp63β, a major p63Δα isoform expressed in our animal model, fused to the FLAG-tagged BirA* protein or BirA*-FLAG only (empty vector) (Fig. 6A). We confirmed the successful induction of the expression of the FLAG-tagged fusion proteins by doxycycline treatment (SI Appendix, Fig. S6 A, Left). Notably, both isoforms of p63 in our model localized to the nucleus (SI Appendix, Fig. S6 A, Right) and were functional, as assessed by measuring the luciferase activity of the BPAG1 reporter upon induction with doxycycline (SI Appendix, Fig. S6B). We then obtained lysates of the induced cells for pull-down with streptavidin beads (SI Appendix, Fig. S6C), followed by mass spectrometry. A comparison of the peptides identified in the ΔΝp63α or ΔNp63β pulldown samples from three biological replicates with those identified in the empty vector lysates revealed 96 and 115 high-confidence binding partners of ΔΝp63α and ΔNp63β, respectively. These proteins included several known cofactors of p63 published in the BioGRID database (SI Appendix, Fig. S6D). We then assessed the differential abundance of identified binding partners of p63. Fifty proteins were bound to both isoforms of p63, whereas 29 and 51 cofactors showed stronger interactions with ΔΝp63α and ΔNp63β, respectively (Fig. 6B and Dataset S1). Reactome pathway analysis of ΔΝp63α-enriched binding partners revealed only a handful of pathways with low significance. The cofactors shared by ΔΝp63α/ΔNp63β were related mainly to gene transcription and chromatin organization, consistent with previous reports (40). In contrast, the ΔNp63β-enriched binding partners were more strongly associated with general transcription machinery, for example, GTF2i, GTF2A1, TEAD1, and MED27 (Fig. 6C). To confirm these findings in vivo, we isolated primary keratinocytes from p63^+/+ and p63^Δ13/Δ13 mice and carried out a proximity ligation assay (PLA) between p63 and either chromatin remodeled SmarcC2 or the general transcription factor Gft2i on pools of keratinocytes (Fig. 6D). p63-SmarcC2 PLA yielded a high number of foci per nucleus, and the numbers of spots were comparable between the two groups. In contrast, the number of p63-Gtf2i PLA foci per nucleus was larger in p63^Δ13/Δ13 keratinocytes. Taken together, our findings strongly suggest that p63β interacts preferentially with general transcription machinery proteins such as Gtf2i.

Fig. 6.

ΔNp63α preferentially interacts with the general transcription machinery. (A) Schematic showing the BioID/MS strategy. (B) Volcano plot showing common and differentially enriched binding partners of ΔNp63α or ΔNp63β from BioID/MS analysis. Significantly enriched proteins with abs ([ΔNp63β/ΔNp63α] MS signal differences) > 0.5 and P < 0.05 are highlighted in green (ΔNp63α-enriched) or magenta (ΔNp63β). Shared binding partners are in gray. (C) Reactome pathway enrichment of p63 binding partners from (A). (D) PLA analysis of the interaction between selected binding partners and p63α or p63β in cultured p63^+/+ and p63^Δ13/Δ13 mouse keratinocytes, respectively. Representative images and the number of PLA spots per nucleus are shown. Analysis of a pool of keratinocytes from n = 4 mice was performed. Five fields from each sample were quantified. (Scale bar, 10 μm.)

Discussion

p63 is a master regulator of epithelial homeostasis and oocyte integrity. The functional properties of different p63 domains have been comprehensively investigated in vitro, and the roles of the central DBD and N terminus have been established via in vivo knockout models. However, the role of C-terminal domains in vivo remains elusive (SI Appendix, Fig. S1A), especially considering that this region harbors genetic mutations that leads to AEC syndrome. In particular, the functions of the SAM and TID domains in the ΔNp63 isoform remain unclear. Here, we generated a KRT14 conditional p63^Δ13/Δ13 mouse expressing p63β (a natural variant lacking C-terminal SAM/TID domains due to the absence of exon 13) in place of p63α (Fig. 7) in only keratin-14-positive tissues such as the skin, esophagus, and forestomach. Our results indicate a distinct functional role for p63α-depleted isoforms compared to p63α highlighting the essential role of the C-terminus: The full-length ΔNp63α isoform is essential for skin development and cannot be substituted by the other isoforms (β, γ, or others).

Fig. 7.

Loss of the C-terminus drives p63 to promoters, leading to dysregulation of epithelial homeostasis. Loss of the C-terminal domain of p63 enhances its interaction with the basal transcriptional machinery, which results in increased p63 occupancy at promoters and increased expression of ECM-related genes in mouse keratinocytes. Abnormal gene expression leads to aberrant adhesion of basal keratinocytes to the basal lamina and may trigger systemic inflammation, growth arrest, and premature death.

We recently demonstrated the importance of the C-terminal domains of the TAp63 isoform in oocyte genome quality check and its inhibitory function in vivo (9). Structurally, the C-terminus impaired the formation of a transcriptionally active tetramer, masking the TAD of this isoform. The SAM and TID domains within the p63 C-terminus are described as protein–protein interaction and regulatory domains, respectively (8, 9). Accordingly, p63 binding partners include the main chromatin remodeling factors and complexes (23–28), while posttranslational modifications are reported to regulate p63 activity and half-life (11, 41, 42). Our BioID interactome profiling followed by mass spectrometry confirmed that both ΔNp63α and ΔΝp63β share multiple binding partners, including epigenetic modulators such as SMARCC2, KDM6A, KMT2D, and NCOR1, indicating that the C-terminal domains are not strictly required for p63 to join the remodeling complexes. Interestingly, we found an enrichment of general transcription factors, such as GTF2I, among the ΔΝp63β binding partners (Fig. 7), contributing to the phenotype observed.

Consistent with the notion that ΔΝp63β cooperates with the general transcription machinery, we observed more frequent occupancy of p63α-depleted isoforms at promoters, although at most of the same genes associated with p63α. These data suggest that, overall, the loss of the C-terminus does not lead to rewiring of the p63 landscape; rather, it enhances p63 binding to its specific target loci (Fig. 7). It is, therefore, plausible that p63, as an evolutionary pioneer factor, first recognizes canonical binding sites within distal elements, recruiting its cofactors as supported by H3K27 acetylation, which is not affected in the presence of p63α-depleted isoforms. Subsequently, p63 binds to low-affinity motifs at promoters, probably establishing chromatin loops to increase transcription activation. The observation that target genes are mostly conserved between p63α- and p63Δα-expressing cells and that only their expression level is affected indicates that the p63 C-terminus is important for selectively guiding p63 toward high-affinity binding sites, predominantly at distal regulatory elements. However, more effort is needed to clarify how and whether p63 binding shifts from high-affinity sites to low-affinity sites. Notably, our ChIP-seq carried out using a pan-p63 antibody cannot distinguish between TA and ΔN isoform occupancy (43); therefore, additional experiments must be performed to address this issue.

Transcriptomic analysis revealed a strong increase in the expression of ECM organization genes, including collagens, laminins, and integrins, among the p63 target genes modulated in the presence of the p63α-depleted isoforms variant in mouse p63^Δ13/Δ13 keratinocytes. ECM- and adhesion-related genes were identified among the top modulated ΔNp63 targets in RNAi and overexpression experiments (14, 44). Our data show that ECM genes whose expression was increased in p63^Δ13/Δ13 keratinocytes were already bound by p63α and the occupied regions had canonical p63 motif. Therefore, it is plausible that ECM genes are among the strongest p63α targets and hence immediately affected by p63α-depleted isoforms which further interact with the general transcription machinery, leading to increased expression of ECM genes and triggering an imbalance in the levels of these proteins (Fig. 7) (45). Of note, we were able to see a trend for upregulation of these genes in epidermises of p63^Δ13/Δ13 mice as well; however, the upregulation only of few genes was statistically significant. This is possibly due to the presence of multiple cell types in the epidermis other than basal proliferating keratinocytes, constituting only a small proportion of the epidermis, leading to an averaging of bulk gene expression.

As a consequence of this aberrant expression of ECM proteins, ultrastructural defects in p63^Δ13/Δ13 mouse skin were observed. In particular, electron microscopy revealed gaps between basal keratinocytes and the basal lamina, as well as a reduced number of desmosomes (Fig. 7). Importantly, the upregulation of integrin expression has been reported to inhibit epidermal differentiation (46), suggesting that both cultured keratinocytes and p63^Δ13/Δ13 mouse epidermal cells exhibit an increased proliferation rate.

The deletion of the p63 C-terminus does not quite recapitulate the loss of the full-length p63 (2), in fact p63^Δ13/Δ13 mice do not show any evident changes at birth. These observations suggest that the C-terminus is not essential for tissue development, differentiation, and stratification. However, adult p63^Δ13/Δ13 mice exhibit severe skin and forestomach abnormalities. Deregulation of ΔNp63 activity has been shown to promote proinflammatory, psoriasis-like phenotype (44). Interestingly, our transcriptomic analysis of the p63^Δ13/Δ13 epidermis revealed a notable increase in the expression of prostaglandin and GPCR signaling-related genes, which have been shown to be associated with skin inflammation and hyperplasia (47–51) as well as chronic inflammation. An excess of collagen and fibronectin fragments can also stimulate chemokine expression by tissue-specific immune cells (52), while alterations in ECM protein expression can be observed in inflammatory skin diseases (38). Unfortunately, we used bulk RNA-seq, which cannot distinguish between different cell types; therefore, we could not establish exactly which type is responsible for the overall increase in prostaglandin/GPCR genes. Notably, an increase in the serum levels of Il1β and Il6 in p63^Δ13/Δ13 mice is often associated with systemic inflammation, which may lead to the growth abnormalities and premature death observed in p63^Δ13/Δ13 mice (53, 54) (Fig. 7). These data suggest that α-domain of p63 prevents chronic inflammation by maintaining correct expression levels of structural proteins in basal epithelial cells and overall epithelial homeostasis.

In summary, our findings demonstrate an “expected” crucial physiological role of the SAM/TID domains in the p53 family, as the C-terminus is essential for the proper activity of TAp63 (9) and some functions of p73 (55, 56). However, our data also indicate an “unexpected” inhibitory role for the C-terminus of ΔNp63 in the underlying molecular mechanism. The presence of SAM/TID domains in ΔNp63α seems to reduce its interaction with the general transcription machinery while maintaining strong cooperation with chromatin remodeling complexes. This guides p63 toward selective occupancy at distal regulatory elements with stronger binding motifs to allow balanced expression of its target genes. In contrast, the loss of the C-terminus leads to enhanced p63 binding to proximal promoters of target genes, similar to the pattern observed for its sibling p53 (57). These data clearly indicate that p63β or other p63α-depleted isoforms cannot replace p63α in vivo.

In conjunction with the findings of our previous report (9), these data demonstrate that the C-terminus inhibits the transcriptional activity of both TAp63 and ΔΝp63 via different mechanisms. In the ovary, loss of the C-terminus generates a “p53-like” constitutively active ΤΑp63β, which induces uncontrolled expression of apoptosis-related genes. Similarly, in the skin, ΔNp63β behaves as an overactive “p53-like” isoform, leading to disproportionately high ECM gene expression. While the exact role of the C-terminus in inhibiting protein–protein interactions is still under investigation, we can affirm that the carboxy-terminal domains of the ΔNp63α isoform play an essential role in ensuring the proper epithelial organization and homeostasis required for organism growth and survival in vivo.

Limitation of Our Study.

It is worth noting that while ΔNp63 is the major isoform expressed in Keratin 14 positive epithelia, we cannot exclude an impact of TAp63β isoform to the observed phenotype, as TAp63 is expressed in a subset of epithelial cells or during specific stages of tissue development (30, 43, 58, 59). Therefore, additional work must be carried out to clarify this point.

Methods

Generation of KRT14-cre;Trp63-Ex13^fl/fl Mice.

Heterozygous Trp63 exon 13 floxed allele mice (Trp63-Ex13^fl/+) were generated by Ozgene (Ozgene Pty Ltd., Bentley, Australia), and the colony was amplified by crossing with wild-type C57Bl/6J mice. Homozygous floxed/floxed mice (Trp63-Ex13^fl/fl) were then crossed with Tg(KRT14-cre)1Amc/J mice (60) to obtain male and female Tg(KRT14-cre) hemizygous floxed/+ mice. These mice were crossed to obtain homozygous Tg(KRT14-cre) floxed/floxed mice with Cre mediated p63 exon 13 deletion and, as a consequence, p63 β isoform expression in KRT14 expressing tissues (KRT14-cre;Trp63-Ex13^fl/fl). All experiments were approved by the Institutional Animal Care and Use Committee (IACUC) and were carried out according to the Italian and European rules (D.L.116/92; C.E. 609/86; European Directive 2010/63/EU). Licence for mice generation and experiments was approved by the Italian Ministry of Health (protocol number 703/2020PR).

See SI Appendix for additional Methods.

Supplementary Material

Appendix 01 (PDF)

Dataset S01 (XLSX)

Data, Materials, and Software Availability

All study data are included in the manuscript and/or supporting information.