Role of chromatin and transcriptional co-regulators in mediating p63-genome interactions in keratinocytes
© Sethi et al.; licensee BioMed Central Ltd. 2014
Received: 12 June 2014
Accepted: 18 November 2014
Published: 29 November 2014
The Transcription Factor (TF) p63 is a master regulator of epidermal development and differentiation as evident from the remarkable skin phenotype of p63 mouse knockouts. Furthermore, ectopic expression of p63 alone is sufficient to convert simple epithelium into stratified epithelial tissues in vivo and p63 is required for efficient transdifferentiation of fibroblasts into keratinocytes. However, little is known about the molecular mechanisms of p63 function, in particular how it selects its target sites in the genome. p63, which acts both as an activator and repressor of transcription, recognizes a canonical binding motif that occurs over 1 million times in the human genome. But, in human keratinocytes less than 12,000 of these sites are bound in vivo suggesting that underlying chromatin architecture and cooperating TFs mediate p63-genome interactions.
We find that the chromatin architecture at p63-bound targets possess distinctive features and can be used to categorize p63 targets into proximal promoters (1%), enhancers (59%) and repressed or inactive (40%) regulatory elements. Our analysis shows that the chromatin modifications H3K4me1, H3K27me3, along with overall chromatin accessibility status can accurately predict bonafide p63-bound sites without a priori DNA sequence information. Interestingly, however there exists a qualitative correlation between the p63 binding motif and accessibility and H3K4me1 levels. Furthermore, we use a comprehensive in silico approach that leverages ENCODE data to identify several known TFs such as AP1, AP2 and novel TFs (RFX5 for e.g.) that can potentially cooperate with p63 to modulate its myriad biological functions in keratinocytes.
Our analysis shows that p63 bound genomic locations in keratinocytes are accessible, marked by active histone modifications, and co-targeted by other developmentally important transcriptional regulators. Collectively, our results suggest that p63 might actively remodel and/or influence chromatin dynamics at its target sites and in the process dictate its own DNA binding and possibly that of adjacent TFs.
Tp63 is an important transcription factor of the p53/p63/p73 family that dictates a wide range of cellular properties including but not limited to stem cell renewal, lineage choices and maintaining the balance between proliferation and differentiation [1, 2]. This diverse function of p63 is critical for morphogenesis during development, particularly for epithelial-enriched tissues such as the skin and its appendages such as the hair follicles and mammary glands. Indeed, p63-null mice die after birth and exhibit a dramatic agenesis of epithelial-rich structures and widespread developmental defects of the limb, orofacial region, and external genitalia [3, 4]. These p63-deficient structural defects are thought to be the result of a failed program of epithelial stratification and/or diminished capacity for stem cell renewal, both of which can jeopardize normal epithelial-stromal interactions needed during embryonic organ development [5, 6]. In agreement with the mouse phenotype, p63 mutations in humans lead to congenital abnormalities such as abnormal limb development and ectodermal dysplasia, which are associated with a spectrum of developmental disorders including AEC or EEC syndrome [7, 8].
The biological function of p63 is mediated by several isoforms derived from distinct transcripts . These include the longer TAp63 isoforms and N-terminal deleted ΔNp63 isoforms generated from an internal promoter located within intron 3. Furthermore alternative splicing can result in α, β, and γ isoforms, which differ in the C-terminus. All p63 isoforms share the DNA-binding and oligomerization domains, which are analogous to that of p53. It is now well-established that ∆Np63, especially ∆Np63α is the predominant isoform that is present in most epithelial cells such as the keratinocytes of the skin . Importantly both gene complementation studies and isoform specific knockouts have conclusively affirmed that ∆Np63 harbors most of the function and biological activity of p63, particularly as it pertains to the epithelial tissues [10–14].
The role of p63 in regulating transcription during development has been extensively studied in skin where ∆Np63 is highly expressed and regulates the transition from simple ectodermal cells to stratified epithelium [5, 15]. Given the master regulatory function of p63, it is not surprising that the repertoire of p63-targets is vast and represents practically every crucial gene regulatory and signaling pathway. This is evident from the ~11,000 binding sites for p63 in human keratinocytes as determined by chromatin immunoprecipitation with next-generation sequencing (ChIP-seq) studies . p63 controls expression of basal keratin genes K5 and K14 and regulates MYC levels thereby controlling keratinocyte proliferation via the Wnt/β-catenin and Notch signaling pathways [10, 17, 18]. The keratinocyte differentiation program is also regulated by p63, in part via its effect on the ZNF750-KLF4 regulatory axis .
While the identification of p63 bound cis-regulatory elements in keratinocytes has received much attention, the mechanics of p63-DNA interaction is still relatively unknown. p63 binds a canonical motif, defined as closely spaced 2 decamers (RRRCRWGYYY, RRRCWYGYYY), although there is growing evidence that p63 can target sites that do not completely conform to this consensus sequence, including half sites . Given the degenerate nature of the p63 binding motif, it is not surprising that by conservative estimates, there are more than 1 million such potential sites in the human genome. However, as is the case with most other Transcription Factors (TFs), only a small subset of these sites are bound by p63 in vivo [16, 21]. It is likely that the local chromatin architecture, among other factors plays an important deterministic role in dictating how and why p63 selects its target DNA. Hence, this is an important area of future investigation; especially given the increasing evidence that p63 can play an important role in modulating the chromatin structure. Indeed, recent studies have demonstrated that p63 can functionally interact with several epigenetic factors in keratinocytes, which can in turn profoundly influence p63-dependent transcriptional activation and repression. Examples of such interactions include the reinforcement of p63 mediated repression of p16 by Lsh, a member of the SNF2 family of chromatin remodeling ATPases , direct recruitment of histone deacetylases, HDAC1 and HDAC2 by ∆Np63 during repression of target genes in the embryonic epidermis  and the crosstalk between p63 and chromatin organizer Satb1 in regulating keratinocyte differentiation genes . p63 can also control higher-order chromatin structure in epidermal progenitor cells during skin development by regulating Brg1, a ATP-dependent chromatin remodeler . Given these emerging links between p63 and chromatin, it is important that any comprehensive studies on the mechanism of p63-genome interactions takes into account the underlying state of epigenetic modifications.
Here we have utilized the p63 ChIP-Seq dataset and available chromatin modification datasets for Normal Human Epidermal Keratinocytes (NHEK) to investigate the rules that govern binding of p63 to its target DNA. We find that p63 binds to a canonical motif (2 decamers with zero spacer in-between) at the majority (73.3%) of its sites, whereas non-canonical motifs containing 1–15 spacer between decamers are present in only 16.4% of the sites. The chromatin at p63 binding sites is largely marked by active histone modifications (H3K4me1 or H3K4me3 and H3K27ac). Moreover, chromatin accessibility with H3K4me1 can accurately predict bona-fide bound p63 sites without the need for any additional DNA sequence information. Finally, using a comprehensive in silico approach, we identify several cooperating TFs that appear to define specific classes of p63 regulated genes.
Underlying sequence patterns and chromatin architecture of p63 targets
Several groups have determined global p63 binding locations in various primary and immortalized keratinocytes using ChIP-chip or ChIP-Seq techniques [16, 21, 26–28]. For our studies, we focused on the most comprehensive p63 ChIP-Seq data  available to date. It had an added benefit of being generated from primary keratinocytes (NHEK) and more importantly conforming to ENCODE guidelines . To facilitate uniform comparisons across other ENCODE datasets, we re-aligned the p63 ChIP-Seq to the latest human genome build (hg19) with Bowtie . In strong agreement with Kouwenhoven et al., by using high stringency conditions (p-value: 1e-10), we identified a reliable and robust dataset of 11632 p63 binding sites that were common among the three biological replicates. On examining the underlying DNA sequence of these p63-ChIPed elements, we found that 73.3% of these sites have at least one p63 canonical motif. Among these, 32% show a close match to the p63 consensus (strong motif) while the remaining 41.3% are a weaker match (Additional file 1: Figure S1). Both the strong and weak canonical motifs are significantly enriched at the p63 ChIPed regions compared to random genomic regions (P value <1×10-200). An additional 16.4% of the binding sites show a close match to the non-canonical p63 motif, which has 1–15 base spacers separating the two half-sites (Additional file 2: Figure S2). Interestingly, 10% of the p63 bound genomic sites do not have a recognizable p63 motif raising the possibility that p63 can perhaps also be recruited to target regulatory elements through indirect mechanisms such as protein-protein interactions. Our analysis also revealed that only a few of the p63 binding sites contained just a half site (Additional file 2: Figure S2).
p63 has been shown to target different types of regulatory regions (such as promoters and enhancers) and involved in both activation and repression of gene expression [33–36]. Recent studies have demonstrated that specific states of chromatin modifications at the regulatory regions are strongly associated with the level of gene expression for the corresponding genes . To determine whether such differences exists for distinct classes (active vs. repressed for e.g.) of p63 targets, we clustered the p63 bound regulatory regions by their underlying chromatin architecture. We first divided p63 targets into two groups (Cluster A and B) based on the magnitude of signal (chromatin state) and then performed unsupervised clustering using the spatial arrangement of the histone modifications (chromatin architecture) .
Annotating p63 targets clustered by chromatin profiles
Size of cluster
Chromatin based segmentation
Median distance to nearest TSS
Median expression of nearest gene (RPKM)
Top GO terms (Biological Process) (Binomial FDR Q-val)
Top PANTHER pathways (Binomial FDR Q-val)
Proximal (898 bp)
1. Regulation of programmed cell-death (2.98e-2)
1. P53 pathway (1.27e-3)
Distal (35 kb)
1. Anti-apoptosis (8.28e-14)
1. Apoptosis signaling pathway (1.94e-9)
2. Hemi-desmosome assembly (1.166e-10)
2. Integrin signaling pathway (1.67e-7)
3. Epidermis development (1.85e-9)
3. Interleukin signaling pathway (1.1299e-6)
Distal (37 kb)
1. Epidermis development (3.29e-9)
1. T cell activation (2.06e-4)
2. Hair follicle development (3.25e-6)
3. Hair cycle (3.88e-6)
Distal (49.3 kb)
1. Response to mechanical stimulus (9.75e-10)
1. T cell activation (1.9e-8)
2. Negative regulation of intracellular protein kinase cascade (1.54e-6)
2. VEGF signaling pathway (3.77e-5)
3. Response to reactive oxygen species (2.6e-6)
Distal (57.4 kb)
Hair cycle (1.19e-9)
No term enriched
Hair follicle development (1.56e-9)
Cellular response to gonadotropin stimulus (4.16e-4)
We next used Genomic Regions Enrichment of Annotations Tool (GREAT)  to determine whether the five clustered groups of p63 targets could be segregated into distinct classes of genes involved in specific biological pathways. Our analysis revealed that each cluster indeed was overrepresented by genes that were involved in closely related, yet disparate biological activity. For example, cluster B1, which primarily encodes for strong enhancers is enriched for Gene-Ontology Biological Process (GO-BP) of epidermis development, while cluster B3 representing repressed/inactive sites is enriched for GO-BP of hair cycle. This raises the intriguing possibility that p63 might play an important role in actively driving the epidermal developmental processes while keeping the hair cell fate repressed – a notion that is supported by data from prior transgenic mouse studies . The complete annotation for each group of p63 targets is provided in Table 1. Thus, collectively the data obtained by chromatin architecture-dependent clustering of p63 targets allowed us to determine both the active and poised targets of p63 and annotate these to specific biological processes (Additional file 3: Table S1).
p63 binding can be accurately predicted from chromatin modifications
Identification of p63 Cooperating TFs
It is well established that many TFs often act in a combinatorial fashion to govern tissue-specific gene expression. Hence, we wanted to examine the repertoire of p63 associated TFs that might play such role in modulating p63 binding and possibly influencing p63-dependent gene expression in keratinocytes. We therefore used a combination of in-silico analysis of the p63 binding sites and careful data mining of large-scale genomic datasets such as RNA-Seq and ChIP-Seq from the ENCODE project. We posit that the TF that are likely to directly cooperate with p63 will have the following features. First, their DNA binding motifs will likely be enriched at p63 bound elements, keeping in mind that co-occurrence of such motifs adjacent to p63 sites is not a prerequisite for such interactions. Second, their in-vivo binding profiles will overlap with p63 binding profile. Here we reasoned that TF binding to some extent could be extrapolated from ChIP-Seq in other cell types, as it has been shown that TFs share a large number of common binding sites across different cell lines . Finally, we contend that the relevant TF should be expressed in keratinocytes.
The potential role of p63 in modulating different biological functions in coordination with various cooperating TFs
We also discovered possible new target genes under the control of p63-bound distal regulatory regions that otherwise would have been missed by the conventional strategy of assigning a p63-regulated element to its closest gene. As shown in Figure 8B, one such interesting candidate is RFX5, a novel cooperating TF of p63 as discussed above. p63 binds a regulatory element 469.4 kb upstream of the RFX5 gene and motif analysis suggests that AP1, E-box binding factors and BACH1 are part of the transcriptional complex at this site (Figure 8B). Hence, there might exist a transcriptional regulatory loop whereby p63 in cooperation with additional TFs activates RFX5, which then in turn, modulates p63 binding to its target sites. Similar to the case with the KRT14 gene, this distal regulatory element for RFX5 is predicted to be also linked to the TUFT1 and RORC genes. In the future in-depth studies such as 3C experiments will help to confirm these distal enhancer-promoter interactions and to firmly establish the true identity of the p63-driven gene network in keratinocytes.
This study aimed to decipher the mechanics of p63 binding by determining the minimal in-vivo motif required for binding, distinguishing between chromatin architecture of bound and unbound motifs and identifying cooperating TFs that modulate p63 biological activity.
p63 requires a full site for binding to target sites
The transcription factor p63 binds as a homotetramer to two decamers RRRCRWGYYY, RRRCWYGYYY separated by a 0–15 base pair spacer region. We found that the p63 sites containing the decamer pair with intervening spacers was much less prevalent (16.4%) than those where the two half sites were juxtaposed to each other (73.3%). A very small subset (only 8 genomic locations) of p63 binding sites consisted of only a half site (1 decamer) (Additional file 2: Figure S2). This is in contrast to a similar study that found 3-4% of p53 binding sites having a half site . This observation can be explained by a slight difference in the consensus motifs of the two factors, which also results in 3 fold lower binding affinity of p63, in comparison to p53 . One possibility is that the dimer-dimer interactions are important for p63 DNA binding specificity and therefore p63 requires a full site to bind DNA efficiently. Indeed, such differences in the DNA-protein interactions among p53 family members are quite evident from recent structural studies with p73 . Interestingly, not only the distance between two p73 half-sites influences the p73 quaternary structure, but tellingly transcriptional activity is also more affected by spacer length in p73 response element than in p53. Finally, it is worth noting that ~10% of p63 ChIPed sites in keratinocytes do not have a recognizable p63 binding sequence suggesting that the p63 binding at these sites is driven by indirect mechanisms that might involve other DNA-binding TFs and/or non-canonical p63 motifs.
The p63 consensus motif is not required for predicting binding events in keratinocytes
We utilized two computational modeling approaches to uncover key characteristics defining p63 binding sites. First, we utilized discriminant modeling that allowed us to predict p63 binding as a binary score (presence or absence). Second, in a parallel strategy, we used regression modeling that predicted the degree of p63 occupancy. Our approach was distinct from other published methods in that we trained our statistical models on random genomic sites which might or might not have p63 binding sequences . This allowed our models to include the 10% binding sites that do not have p63 DNA-binding motif. Surprisingly, we found that our final models constituting only chromatin marks (H3K4me1, H3K27me3 and accessibility data) predicted p63 binding with high accuracy (Discriminant Model - 89.8%(±0.3) sensitivity, Regression model – 0.083(±0.001) MSE). Adding sequence information to the models, did not lead to any significant improvement even though p63 motif is a statistically significant predictor (Additional file 5: Table S2, Additional file 6: Table S3). One possible explanation is that in keratinocytes, regions of the genome that have a functional p63 motif are on average more accessible and marked with active chromatin marks. This result is not surprising if p63 is a key component of the regulatory complex that is involved in remodeling the chromatin at its binding sites. Indeed p63 target sequences dictate higher nucleosome occupancy than random genomic sequences according to the nucleosome-DNA interaction model (Figure 4). It can be then hypothesized that p63 binding shifts the nucleosomes creating an accessible active chromatin structure at its targets. Support for such a role for p63 comes from a recent study, which examined the selective loss versus gain of DHSs targeted by lineage regulating TFs during lineage differentiation from ESCs . While the recognition landscape for p63 remains largely unchanged during development of all other lineages, there is a significant and selective gain of p63 binding elements in the DHSs of the human skin keratinocytes, which represent the ectodermal lineage. Such deterministic function of p63 is further evident by data showing that p63 in combination with KLF4 can efficiently convert human fibroblast into keratinocytes . These interesting correlative findings, together with our results presented here strongly suggest that p63 functions as part of a pioneering complex which can target and remodel chromatin at many of its sites.
Binding of p63 in coordination with cooperating TFs
It is likely that the p63-depedent regulation of target genes in keratinocytes requires co-operation of other TFs. We have used a multi-pronged approach to identify such p63-associated cooperating TFs by processing data from NHEK RNA-seq, available ENCODE ChIP-seq and computational prediction methods based on TF motifs. Our analysis led to a few surprising observations about the identity of candidate TFs that were likely to be involved in p63-genomic interactions. One striking result from our study is that many of the p63-associated factors belong to broadly expressed family of TFs such as AP1, AP2, MYC and STAT rather than highly tissue-specific factors. Although at first glance, this result may seem disappointing, we think that given the master regulatory role of p63, such a finding makes biological sense. Indeed, given the fact that p63 is highly expressed in a lineage-restricted fashion and plays a crucial role in dictating keratinocyte cell fate, it is conceivable that some of the p63-associated cooperating TFs might just provide ancillary role in regulating gene expression. Another interesting possibility is that the keratinocyte-specific gene expression is mediated by a combinatorial interaction of multiple TFs as suggested by prior studies . However it is important to stress that many of the broadly expressed TFs such as AP1 and AP2 do have keratinocyte-specific roles that are often masked due to functional redundancy from expression of multiple family members [47, 49]. Future functional studies on these TFs that are part of the p63-driven transcriptional network, including ones that are relatively under studied such as RFX5 will shed important insights into gene regulatory mechanisms in keratinocytes.
Despite the wealth of information obtained from our data-mining studies, long term follow-up experimental studies are needed to better elucidate the p63 TF network and the role of chromatin in regulating myriad biological functions of p63. Unraveling the complex nature of the distal regulatory elements such as enhancers, which are by far the most common sites of p63 binding is a formidable challenge. The new insight into the dynamic interplay between p63, its many cooperating TFs and the local chromatin environment, as reported here is the first step towards tackling such challenges.
Determining p63 binding profile in the genome
Global p63 binding locations in keratinocytes was determined from ChIP-Seq datasets generated by Kouwenhoven et al. . The Illumina FASTQ sequencing files from three independent replicates were aligned to hg19 with bowtie  with the following parameters: m = 1 (i.e. removes all those alignments with more than one match). P63 binding locations were then identified in each experiment under stringent conditions with MACS (cutoff p-value = 1e-10) . The 11632 locations that were common in all three replicates were used in this study.
Finding p63 motif in p63 bound locations
Patser  was used to search for the occurrence of p63 canonical motif (defined as 2 decamers (RRRCRWGYYY, RRRCWYGYYY) with zero spacer in between)  in a 500 bp window around the 11632 p63 bound locations. To determine a cutoff score, above which the motif would be termed as a significant match, we created a background model with 100000 random genomic sites. The motif score for which the probability of any stronger motif occurring by random chance would be less than 0.01, was selected as the cutoff for strong motifs (For p63 - cutoff score of 7.14). A relaxed cutoff of 2.24, corresponding to a weaker motif (0.1 probability of occurring by random chance) was also determined. We then repeated the above procedure for possible non-canonical p63 motifs. For this we modified the position weight matrix (PWM) used earlier by inserting spacers (1–15 with 0 weight for each base, i.e. each base assumed to be equally likely) and again used the background model to calculate cutoffs and determine significant matches. We also did this for a half-site (only 1 decamer).
Identifying the chromatin profile at p63 targets
Histone modification ChIP-Seq data for 5 active histone modifications (H3K4me1, H3K4me2, H3K4me3, H3K9ac, H3K27ac) and 2 repressive histone modifications (H3K9me3, H3K27me3) in NHEK (Normal Human Epidermal Keratinocytes) cell-line were obtained from ENCODE . The coordinates for the 11632 p63 bound locations and 30,000 negative genomic sites (any genomic site not within 5 kb of a p63 bound location was termed as a negative site) that had strong p63 canonical motif were obtained. The histone marks were plotted for a 1 kb window at 10 bp resolution in standardized tag count space.
Clustering p63 targets by histone signature
An average signal across a 1 kb window centered at the p63 binding site was plotted for the 5 active (H3K4me1, H3K4me2, H3K4me3, H3K9ac, H3K27ac) and 2 repressive (H3K9me3, H3K27me3) histone modifications. Using k-means clustering algorithm, implemented in Cluster 3.0 software package , with k = 2, the p63 targets were divided into Group A which contained high overall signal for the different histone modifications and Group B, which contained sites with lower overall signal. For each of the two groups, the seven histone modifications were standardized and again plotted in count space, this time at 10 bp resolution. This was done to take into account the spatial arrangement of the histone modifications. Both the groups were individually clustered using k-means algorithm, with k = 4. Pearson correlation was calculated between each pair of clusters, in both directions, for both the groups. Clusters with Pearson correlation higher than 0.9 were grouped together.
Training and test dataset for computational models of p63 binding
The 100,000 random genomic locations were filtered to generate 94760 negative sites (Sites not within 5 kb of a p63 bound location were termed as negative sites). These, along with 11,632 p63 bound locations were used to train and test our Fisher’s discriminant and regression models. Five active (H3K4me1, H3K4me2, H3K4me3, H3K9ac, H3K27ac) and two repressive (H3K9me3, H3K27me3) histone modifications as measured by ChIP-Seq, along with accessibility as measured by DNase-Seq and FAIRE-Seq, in NHEK cell-line were plotted as an average signal across a 1 kb window for each of the 106,392 genomic coordinates, in sqroot space. All the datasets were standardized to 30 million tag count so as to be comparable to each other. These 9 datasets along with interaction terms (calculated as product of two datasets for all possible combinations – e.g. DF (DNase*FAIRE)) were used as predictor variables along with the p63 canonical motif score as generated by Patser, for the computational models. p63 tag count in standardized sqroot space was the response variable for the regression model and a categorical variable with two possible values (0 (bound) and 1 (unbound)) was the response variable for the discriminant model. The 94,760 negative genomic coordinates and 11,632 positive genomic coordinates were then randomly divided into training and test datasets, such that the training dataset had 45,000 negative genomic sites and 5000 positive genomic sites. For the discriminant model, the training dataset was further filtered so as to keep only those negative genomic locations that had p63 tag count (standardized sqroot space) less than 0.25, resulting in 30000 negative genomic sites and 5000 positive genomic sites. This random division into training and test datasets was then repeated 10 times to obtain the mean sensitivity (true positive rate) and mean specificity (true negative rate) for the discriminant model and mean MSE (mean square error) and mean R2 (fraction of variance in p63 occupancy explained by the model) for the regression model.
Fisher’s discriminant model
DISCRIM procedure in SAS statistical software package was used to construct a Fisher’s Discriminant Model. Using all the variables in the discriminant training dataset (9 chromatin features (H3K4me1, H3K4me2, H3K4me3, H3K9ac, H3K27ac, H3K9me3, H3K27me3, DNase, FAIRE) and one interaction term (DNase*FAIRE)), we created a full discriminant model. To simplify the model, we used the STEPDISC procedure to determine 8 variables with significant predictive power. They were used to make the significant chromatin marks model. This was further simplified into the Best 3 variable model, by using only the top 3 variables (H3K4me1, DNase, H3K27me3), with the highest predictive power. The Best 3 variable model was then tested on the test dataset.
REG procedure in SAS statistical software package was used to create a regression model for p63 occupancy based on the chromatin features (no sequence information). We started with a model with no predictors and used stepwise selection to add significant predictors to the model, starting with the one that had the smallest P value (till P value for entry was less than 0.01). At each step the P value for exit was also calculated and the predictor was retained only if its P value was less than 0.01. The 9 chromatin features (H3K4me1, H3K4me2, H3K4me3, H3K9ac, H3K27ac, H3K9me3, H3K27me3, DNase, FAIRE) and one interaction term (DNase*FAIRE) were all found to be statistically significant and formed the full regression model. To simplify this model and find the features with the most predictive power we then used MAXR (Maximum R2 Improvement) selection method, with parameter STOP = 3, to determine the best 3 variable model for predicting p63 occupancy.
Identifying p63 cooperating TFs
We utilized the motifs database by Genomatix software and used Patser to search for the occurrence of 900 TF motifs in a 1 kb window centered at the 11632 p63 bound locations (test set) and 94760 negative genomic sites (background set). Chi-Square test was done to statistically determine which of the TFs motifs were enriched at the p63 binding sites versus the background. % overlaps of the motifs with p63 binding profile were also determined. We then examined for co-occurrence of the TFs motifs with p63 motif. For this our test set contained 8,375 p63 bound locations that had at least a weak p63 motif (score greater than 2.24) and the background set contained 515,933 unbound genomic sites (containing a weak motif). Again chi-square test was done to find the statistically enriched motifs within 100 bp of p63.
The second step to our approach was to use the in vivo binding profiles of TFs, to find the potential cooperating TFs of p63. 264 ChIP-Seq experiments as carried out by ENCODE, capturing an in-vivo binding profile of TFs across 30 cell-lines were used for this analysis. Overlaps were determined between these experiments and p63 binding profile. The same was repeated for the 94760 negative genomic sites. Chi-square test determined the TFs showing enriched binding at p63 targets. We also used NHEK RNA-Seq data to eliminate TFs with RPKM <2 in keratinocytes.
Correlation matrix of cooperating TFs
ChIP-Seq correlation matrix: 13 TFs (CEBPB, CFOS, FOSL2, JUND, BACH1, TFAP2C, STAT1, STAT3, MAX, c-MYC, USF2, RFX5, ELK1) were identified as the most probable cooperating TFs of p63. ChIP-Seq alignment files in Hela-S3 cell-line were obtained from ENCODE via the UCSC genome browser for 11 of the 13 TFs. For BACH1 and FOSL2 we used the alignment files from K562 and A549 cell-lines respectively. An average signal across a 1 kb window was plotted for each of the 13 factors across the 11632 p63 bound locations. Then Pearson correlation coefficient (r) was calculated for each pair. Motif correlation matrix: The 13 TFs corresponded to 10 Position weight matrices (PWMs) (CEBPB, AP1 (CFOS, FOSL2, JUND), BACH1, AP2 (TFAP2C), STAT1, STAT3, MYCMAX (MAX, c-MYC), USF(USF2), RFX(RFX5), ELK1). For each PWM, a Patser generated motif score was obtained for the 11,632 locations. Again, Pearson correlation coefficient (r) was calculated for each pair.
Clustering p63 targets by cooperating TFs motifs
The 10 PWMs (CEBPB, AP1, BACH1, AP2, STAT1, STAT3, MYCMAX, USF, RFX, ELK1) were used to search for the occurrence of REs in a 1 kb window centered at each of the 11632 p63 binding sites. The default cutoffs determined by Patser based on the information content of each of the weight matrices were used to assign a binary score of 0 (Motif absent) and 1 (Motif present) across the 11632 genomic locations. This binary matrix was then clustered using k-means algorithm, implemented in Cluster 3.0 software package , with k = 5.
This work is partially supported by grants R21DE021137 and R03HD073891 to SS, and NY State Department of Health C026714 to MJB.
- Yang A, Kaghad M, Wang Y, Gillett E, Fleming MD, Dotsch V, Andrews NC, Caput D, McKeon F: p63, a p53 homolog at 3q27-29, encodes multiple products with transactivating, death-inducing, and dominant-negative activities. Mol Cell. 1998, 2 (3): 305-316. 10.1016/S1097-2765(00)80275-0.PubMedView ArticleGoogle Scholar
- Crum CP, McKeon FD: p63 in epithelial survival, germ cell surveillance, and neoplasia. Annu Rev Pathol. 2010, 5: 349-371. 10.1146/annurev-pathol-121808-102117.PubMedView ArticleGoogle Scholar
- Yang A, Schweitzer R, Sun D, Kaghad M, Walker N, Bronson RT, Tabin C, Sharpe A, Caput D, Crum C, Mckeon F: p63 is essential for regenerative proliferation in limb, craniofacial and epithelial development. Nature. 1999, 398 (6729): 714-718. 10.1038/19539.PubMedView ArticleGoogle Scholar
- Mills AA, Zheng B, Wang XJ, Vogel H, Roop DR, Bradley A: p63 is a p53 homologue required for limb and epidermal morphogenesis. Nature. 1999, 398 (6729): 708-713. 10.1038/19531.PubMedView ArticleGoogle Scholar
- Koster MI, Kim S, Mills AA, DeMayo FJ, Roop DR: p63 is the molecular switch for initiation of an epithelial stratification program. Genes Dev. 2004, 18 (2): 126-131. 10.1101/gad.1165104.PubMed CentralPubMedView ArticleGoogle Scholar
- Senoo M, Pinto F, Crum CP, McKeon F: p63 is essential for the proliferative potential of stem cells in stratified epithelia. Cell. 2007, 129 (3): 523-536. 10.1016/j.cell.2007.02.045.PubMedView ArticleGoogle Scholar
- Celli J, Duijf P, Hamel BC, Bamshad M, Kramer B, Smits AP, Newbury-Ecob R, Hennekam RC, Van Buggenhout G, van Haeringen A, Woods CG, van Essen AJ, de Waal R, Vriend G, Haber DA, Yang A, McKeon F, Brunner HG, van Bokhoven H: Heterozygous germline mutations in the p53 homolog p63 are the cause of EEC syndrome. Cell. 1999, 99 (2): 143-153. 10.1016/S0092-8674(00)81646-3.PubMedView ArticleGoogle Scholar
- McGrath JA, Duijf PH, Doetsch V, Irvine AD, de Waal R, Vanmolkot KR, Wessagowit V, Kelly A, Atherton DJ, Griffiths WA, Orlow SJ, van Haeringen A, Ausems MG, Yang A, McKeon F, Bamshad MA, Brunner HG, Hamel BC, van Bokhoven H: Hay-Wells syndrome is caused by heterozygous missense mutations in the SAM domain of p63. Hum Mol Genet. 2001, 10 (3): 221-229. 10.1093/hmg/10.3.221.PubMedView ArticleGoogle Scholar
- Rizzo JM, Romano RA, Bard J, Sinha S: RNA-seq studies reveal new insights into p63 and the transcriptomic landscape of the mouse skin. J Investig Dermatol Adv Online Publication. 2014Google Scholar
- Romano RA, Ortt K, Birkaya B, Smalley K, Sinha S: An active role of the DeltaN isoform of p63 in regulating basal keratin genes K5 and K14 and directing epidermal cell fate. PLoS ONE. 2009, 4 (5): e5623-10.1371/journal.pone.0005623.PubMed CentralPubMedView ArticleGoogle Scholar
- Romano RA, Smalley K, Magraw C, Serna VA, Kurita T, Raghavan S, Sinha S: DeltaNp63 knockout mice reveal its indispensable role as a master regulator of epithelial development and differentiation. Development. 2012, 139 (4): 772-782. 10.1242/dev.071191.PubMed CentralPubMedView ArticleGoogle Scholar
- Chakravarti D, Su X, Cho MS, Bui NH, Coarfa C, Venkatanarayan A, Benham AL, Flores Gonzalez RE, Alana J, Xiao W, Leung ML, Vin H, Chan IL, Aquino A, Muller N, Wang H, Cooney AJ, Parker-Thornburg J, Tsai KY, Gunaratne PH, Flores ER: Induced multipotency in adult keratinocytes through down-regulation of DeltaNp63 or DGCR8. Proc Natl Acad Sci U S A. 2014, 111 (5): E572-E581. 10.1073/pnas.1319743111.PubMed CentralPubMedView ArticleGoogle Scholar
- Pignon JC, Grisanzio C, Geng Y, Song J, Shivdasani RA, Signoretti S: p63-expressing cells are the stem cells of developing prostate, bladder, and colorectal epithelia. Proc Natl Acad Sci U S A. 2013, 110 (20): 8105-8110. 10.1073/pnas.1221216110.PubMed CentralPubMedView ArticleGoogle Scholar
- Candi E, Rufini A, Terrinoni A, Dinsdale D, Ranalli M, Paradisi A, De Laurenzi V, Spagnoli LG, Catani MV, Ramadan S, Knight RA, Melino G: Differential roles of p63 isoforms in epidermal development: selective genetic complementation in p63 null mice. Cell Death Differ. 2006, 13 (6): 1037-1047. 10.1038/sj.cdd.4401926.PubMedView ArticleGoogle Scholar
- Botchkarev VA, Flores ER: p53/p63/p73 in the epidermis in health and disease. Cold Spring Harb Perspect Med. 2014, 4 (8): a015248-10.1101/cshperspect.a015248.PubMedView ArticleGoogle Scholar
- Kouwenhoven EN, van Heeringen SJ, Tena JJ, Oti M, Dutilh BE, Alonso ME, de la Calle-Mustienes E, Smeenk L, Rinne T, Parsaulian L, Bolat E, Jurgelenaite R, Huynen MA, Hoischen A, Veltman JA, Brunner HG, Roscioli T, Oates E, Wilson M, Manzanares M, Gomez-Skarmeta JL, Stunnenberg HG, Lohrum M, van Bokhoven H, Zhou H: Genome-wide profiling of p63 DNA-binding sites identifies an element that regulates gene expression during limb development in the 7q21 SHFM1 locus. PLoS Genet. 2010, 6 (8): e1001065-10.1371/journal.pgen.1001065.PubMed CentralPubMedView ArticleGoogle Scholar
- Romano RA, Birkaya B, Sinha S: A functional enhancer of keratin14 is a direct transcriptional target of deltaNp63. J Investig Dermatol. 2007, 127 (5): 1175-1186. 10.1038/sj.jid.5700652.PubMedView ArticleGoogle Scholar
- Wu N, Rollin J, Masse I, Lamartine J, Gidrol X: p63 regulates human keratinocyte proliferation via MYC-regulated gene network and differentiation commitment through cell adhesion-related gene network. J Biol Chem. 2012, 287 (8): 5627-5638. 10.1074/jbc.M111.328120.PubMed CentralPubMedView ArticleGoogle Scholar
- Sen GL, Boxer LD, Webster DE, Bussat RT, Qu K, Zarnegar BJ, Johnston D, Siprashvili Z, Khavari PA: ZNF750 is a p63 target gene that induces KLF4 to drive terminal epidermal differentiation. Dev Cell. 2012, 22 (3): 669-677. 10.1016/j.devcel.2011.12.001.PubMed CentralPubMedView ArticleGoogle Scholar
- Perez CA, Ott J, Mays DJ, Pietenpol JA: p63 consensus DNA-binding site: identification, analysis and application into a p63MH algorithm. Oncogene. 2007, 26 (52): 7363-7370. 10.1038/sj.onc.1210561.PubMedView ArticleGoogle Scholar
- McDade SS, Henry AE, Pivato GP, Kozarewa I, Mitsopoulos C, Fenwick K, Assiotis I, Hakas J, Zvelebil M, Orr N, Lord CJ, Patel D, Ashworth A, McCance DJ: Genome-wide analysis of p63 binding sites identifies AP-2 factors as co-regulators of epidermal differentiation. Nucleic Acids Res. 2012, 40 (15): 7190-7206. 10.1093/nar/gks389.PubMed CentralPubMedView ArticleGoogle Scholar
- Keyes WM, Pecoraro M, Aranda V, Vernersson-Lindahl E, Li W, Vogel H, Guo X, Garcia EL, Michurina TV, Enikolopov G, Muthuswamy SK, Mills AA: DeltaNp63alpha is an oncogene that targets chromatin remodeler Lsh to drive skin stem cell proliferation and tumorigenesis. Cell Stem Cell. 2011, 8 (2): 164-176. 10.1016/j.stem.2010.12.009.PubMed CentralPubMedView ArticleGoogle Scholar
- LeBoeuf M, Terrell A, Trivedi S, Sinha S, Epstein JA, Olson EN, Morrisey EE, Millar SE: Hdac1 and Hdac2 act redundantly to control p63 and p53 functions in epidermal progenitor cells. Dev Cell. 2010, 19 (6): 807-818. 10.1016/j.devcel.2010.10.015.PubMed CentralPubMedView ArticleGoogle Scholar
- Fessing MY, Mardaryev AN, Gdula MR, Sharov AA, Sharova TY, Rapisarda V, Gordon KB, Smorodchenko AD, Poterlowicz K, Ferone G, Kohwi Y, Missero C, Kohwi-Shigematsu T, Botchkarev VA: p63 regulates Satb1 to control tissue-specific chromatin remodeling during development of the epidermis. J Cell Biol. 2011, 194 (6): 825-839. 10.1083/jcb.201101148.PubMed CentralPubMedView ArticleGoogle Scholar
- Mardaryev AN, Gdula MR, Yarker JL, Emelianov VU, Poterlowicz K, Sharov AA, Sharova TY, Scarpa JA, Joffe B, Solovei I, Chambon P, Botchkarev VA, Fessing MY: p63 and Brg1 control developmentally regulated higher-order chromatin remodelling at the epidermal differentiation complex locus in epidermal progenitor cells. Development. 2014, 141 (1): 101-111. 10.1242/dev.103200.PubMed CentralPubMedView ArticleGoogle Scholar
- Pozzi S, Zambelli F, Merico D, Pavesi G, Robert A, Maltere P, Gidrol X, Mantovani R, Vigano MA: Transcriptional network of p63 in human keratinocytes. PLoS ONE. 2009, 4 (3): e5008-10.1371/journal.pone.0005008.PubMed CentralPubMedView ArticleGoogle Scholar
- Vigano MA, Lamartine J, Testoni B, Merico D, Alotto D, Castagnoli C, Robert A, Candi E, Melino G, Gidrol X, Mantovani R: New p63 targets in keratinocytes identified by a genome-wide approach. EMBO J. 2006, 25 (21): 5105-5116. 10.1038/sj.emboj.7601375.PubMed CentralPubMedView ArticleGoogle Scholar
- Yang A, Zhu Z, Kapranov P, McKeon F, Church GM, Gingeras TR, Struhl K: Relationships between p63 binding, DNA sequence, transcription activity, and biological function in human cells. Mol Cell. 2006, 24 (4): 593-602. 10.1016/j.molcel.2006.10.018.PubMedView ArticleGoogle Scholar
- Landt SG, Marinov GK, Kundaje A, Kheradpour P, Pauli F, Batzoglou S, Bernstein BE, Bickel P, Brown JB, Cayting P, Chen Y, DeSalvo G, Epstein C, Fisher-Aylor KI, Euskirchen G, Gerstein M, Gertz J, Hartemink AJ, Hoffman MM, Iyer VR, Jung YL, Karmakar S, Kellis M, Kharchenko PV, Li Q, Liu T, Liu XS, Ma L, Milosavljevic A, Myers RM, et al: ChIP-seq guidelines and practices of the ENCODE and modENCODE consortia. Genome Res. 2012, 22 (9): 1813-1831. 10.1101/gr.136184.111.PubMed CentralPubMedView ArticleGoogle Scholar
- Langmead B, Trapnell C, Pop M, Salzberg SL: Ultrafast and memory-efficient alignment of short DNA sequences to the human genome. Genome Biol. 2009, 10 (3): R25-10.1186/gb-2009-10-3-r25.PubMed CentralPubMedView ArticleGoogle Scholar
- Consortium EP: A user’s guide to the encyclopedia of DNA elements (ENCODE). PLoS Biol. 2011, 9 (4): e1001046-10.1371/journal.pbio.1001046.View ArticleGoogle Scholar
- Lai WK, Bard JE, Buck MJ: ArchTEx: accurate extraction and visualization of next-generation sequence data. Bioinformatics. 2012, 28 (7): 1021-1023. 10.1093/bioinformatics/bts063.PubMedView ArticleGoogle Scholar
- Helton ES, Zhu J, Chen X: The unique NH2-terminally deleted (DeltaN) residues, the PXXP motif, and the PPXY motif are required for the transcriptional activity of the DeltaN variant of p63. J Biol Chem. 2006, 281 (5): 2533-2542.PubMedView ArticleGoogle Scholar
- Birkaya B, Ortt K, Sinha S: Novel in vivo targets of DeltaNp63 in keratinocytes identified by a modified chromatin immunoprecipitation approach. BMC Mol Biol. 2007, 8: 43-10.1186/1471-2199-8-43.PubMed CentralPubMedView ArticleGoogle Scholar
- Ortt K, Raveh E, Gat U, Sinha S: A chromatin immunoprecipitation screen in mouse keratinocytes reveals Runx1 as a direct transcriptional target of DeltaNp63. J Cell Biochem. 2008, 104 (4): 1204-1219. 10.1002/jcb.21700.PubMedView ArticleGoogle Scholar
- Barton CE, Johnson KN, Mays DM, Boehnke K, Shyr Y, Boukamp P, Pietenpol JA: Novel p63 target genes involved in paracrine signaling and keratinocyte differentiation. Cell Death Dis. 2010, 1: e74-10.1038/cddis.2010.49.PubMed CentralPubMedView ArticleGoogle Scholar
- Dong XJ, Greven MC, Kundaje A, Djebali S, Brown JB, Cheng C, Gingeras TR, Gerstein M, Guigo R, Birney E, Weng ZP: Modeling gene expression using chromatin features in various cellular contexts. Genome Biol. 2012, 13 (9): R53-10.1186/gb-2012-13-9-r53.PubMed CentralPubMedView ArticleGoogle Scholar
- Lai WK, Buck MJ: ArchAlign: coordinate-free chromatin alignment reveals novel architectures. Genome Biol. 2010, 11 (12): R126-10.1186/gb-2010-11-12-r126.PubMed CentralPubMedView ArticleGoogle Scholar
- Ernst J, Kheradpour P, Mikkelsen TS, Shoresh N, Ward LD, Epstein CB, Zhang X, Wang L, Issner R, Coyne M, Ku M, Durham T, Kellis M, Bernstein BE: Mapping and analysis of chromatin state dynamics in nine human cell types. Nature. 2011, 473 (7345): 43-49. 10.1038/nature09906.PubMed CentralPubMedView ArticleGoogle Scholar
- McLean CY, Bristor D, Hiller M, Clarke SL, Schaar BT, Lowe CB, Wenger AM, Bejerano G: GREAT improves functional interpretation of cis-regulatory regions. Nat Biotechnol. 2010, 28 (5): 495-501. 10.1038/nbt.1630.PubMedView ArticleGoogle Scholar
- Romano RA, Smalley K, Liu S, Sinha S: Abnormal hair follicle development and altered cell fate of follicular keratinocytes in transgenic mice expressing DeltaNp63alpha. Development. 2010, 137 (9): 1431-1439. 10.1242/dev.045427.PubMed CentralPubMedView ArticleGoogle Scholar
- Song L, Zhang Z, Grasfeder LL, Boyle AP, Giresi PG, Lee BK, Sheffield NC, Graf S, Huss M, Keefe D, Liu Z, London D, McDaniell R, Shibata Y, Showers KA, Simon JM, Vales T, Wang T, Winter D, Zhang Z, Clarke ND, Birney E, Iyer VR, Crawford GE, Lieb JD, Furey TS: Open chromatin defined by DNaseI and FAIRE identifies regulatory elements that shape cell-type identity. Genome Res. 2011, 21 (10): 1757-1767. 10.1101/gr.121541.111.PubMed CentralPubMedView ArticleGoogle Scholar
- Lai WK, Buck MJ: An integrative approach to understanding the combinatorial histone code at functional elements. Bioinformatics. 2013, 29 (18): 2231-2237. 10.1093/bioinformatics/btt382.PubMedView ArticleGoogle Scholar
- Kaplan N, Moore IK, Fondufe-Mittendorf Y, Gossett AJ, Tillo D, Field Y, LeProust EM, Hughes TR, Lieb JD, Widom J, Segal E: The DNA-encoded nucleosome organization of a eukaryotic genome. Nature. 2009, 458 (7236): 362-366. 10.1038/nature07667.PubMed CentralPubMedView ArticleGoogle Scholar
- Lidor Nili E, Field Y, Lubling Y, Widom J, Oren M, Segal E: p53 binds preferentially to genomic regions with high DNA-encoded nucleosome occupancy. Genome Res. 2010, 20 (10): 1361-1368. 10.1101/gr.103945.109.PubMed CentralPubMedView ArticleGoogle Scholar
- Wang J, Zhuang J, Iyer S, Lin X, Whitfield TW, Greven MC, Pierce BG, Dong X, Kundaje A, Cheng Y, Rando OJ, Birney E, Myers RM, Noble WS, Snyder M, Weng Z: Sequence features and chromatin structure around the genomic regions bound by 119 human transcription factors. Genome Res. 2012, 22 (9): 1798-1812. 10.1101/gr.139105.112.PubMed CentralPubMedView ArticleGoogle Scholar
- Eckert RL, Adhikary G, Young CA, Jans R, Crish JF, Xu W, Rorke EA: AP1 transcription factors in epidermal differentiation and skin cancer. J Skin Cancer. 2013, 2013: 537028-PubMed CentralPubMedView ArticleGoogle Scholar
- Lopez RG, Garcia-Silva S, Moore SJ, Bereshchenko O, Martinez-Cruz AB, Ermakova O, Kurz E, Paramio JM, Nerlov C: C/EBPalpha and beta couple interfollicular keratinocyte proliferation arrest to commitment and terminal differentiation. Nat Cell Biol. 2009, 11 (10): 1181-1190. 10.1038/ncb1960.PubMedView ArticleGoogle Scholar
- Wang X, Pasolli HA, Williams T, Fuchs E: AP-2 factors act in concert with Notch to orchestrate terminal differentiation in skin epidermis. J Cell Biol. 2008, 183 (1): 37-48. 10.1083/jcb.200804030.PubMed CentralPubMedView ArticleGoogle Scholar
- Kira M, Sano S, Takagi S, Yoshikawa K, Takeda J, Itami S: STAT3 deficiency in keratinocytes leads to compromised cell migration through hyperphosphorylation of p130(cas). J Biol Chem. 2002, 277 (15): 12931-12936. 10.1074/jbc.M110795200.PubMedView ArticleGoogle Scholar
- Gulati N, Krueger JG, Suarez-Farinas M, Mitsui H: Creation of differentiation-specific genomic maps of human epidermis through laser capture microdissection. J Investig Dermatol. 2013, 133 (11): 2640-2642. 10.1038/jid.2013.190.PubMed CentralPubMedView ArticleGoogle Scholar
- Sanyal A, Lajoie BR, Jain G, Dekker J: The long-range interaction landscape of gene promoters. Nature. 2012, 489 (7414): 109-113. 10.1038/nature11279.PubMed CentralPubMedView ArticleGoogle Scholar
- Thurman RE, Rynes E, Humbert R, Vierstra J, Maurano MT, Haugen E, Sheffield NC, Stergachis AB, Wang H, Vernot B, Garg K, John S, Sandstrom R, Bates D, Boatman L, Canfield TK, Diegel M, Dunn D, Ebersol AK, Frum T, Giste E, Johnson AK, Johnson EM, Kutyavin T, Lajoie B, Lee BK, Lee K, London D, Lotakis D, Neph S: The accessible chromatin landscape of the human genome. Nature. 2012, 489 (7414): 75-82. 10.1038/nature11232.PubMed CentralPubMedView ArticleGoogle Scholar
- Menendez D, Nguyen TA, Freudenberg JM, Mathew VJ, Anderson CW, Jothi R, Resnick MA: Diverse stresses dramatically alter genome-wide p53 binding and transactivation landscape in human cancer cells. Nucleic Acids Res. 2013, 41 (15): 7286-7301. 10.1093/nar/gkt504.PubMed CentralPubMedView ArticleGoogle Scholar
- Chen C, Gorlatova N, Kelman Z, Herzberg O: Structures of p63 DNA binding domain in complexes with half-site and with spacer-containing full response elements. Proc Natl Acad Sci U S A. 2011, 108 (16): 6456-6461. 10.1073/pnas.1013657108.PubMed CentralPubMedView ArticleGoogle Scholar
- Ethayathulla AS, Tse PW, Monti P, Nguyen S, Inga A, Fronza G, Viadiu H: Structure of p73 DNA-binding domain tetramer modulates p73 transactivation. Proc Natl Acad Sci U S A. 2012, 109 (16): 6066-6071. 10.1073/pnas.1115463109.PubMed CentralPubMedView ArticleGoogle Scholar
- Pique-Regi R, Degner JF, Pai AA, Gaffney DJ, Gilad Y, Pritchard JK: Accurate inference of transcription factor binding from DNA sequence and chromatin accessibility data. Genome Res. 2011, 21 (3): 447-455. 10.1101/gr.112623.110.PubMed CentralPubMedView ArticleGoogle Scholar
- Stergachis AB, Neph S, Reynolds A, Humbert R, Miller B, Paige SL, Vernot B, Cheng JB, Thurman RE, Sandstrom R, Haugen E, Heimfeld S, Murry CE, Akey JM, Stamatoyannopoulos JA: Developmental fate and cellular maturity encoded in human regulatory DNA landscapes. Cell. 2013, 154 (4): 888-903. 10.1016/j.cell.2013.07.020.PubMed CentralPubMedView ArticleGoogle Scholar
- Chen Y, Mistry DS, Sen GL: Highly rapid and efficient conversion of human fibroblasts to keratinocyte-like cells. J Investig Dermatol. 2014, 134 (2): 335-344. 10.1038/jid.2013.327.PubMed CentralPubMedView ArticleGoogle Scholar
- Sinha S, Degenstein L, Copenhaver C, Fuchs E: Defining the regulatory factors required for epidermal gene expression. Mol Cell Biol. 2000, 20 (7): 2543-2555. 10.1128/MCB.20.7.2543-2555.2000.PubMed CentralPubMedView ArticleGoogle Scholar
- Feng J, Liu T, Zhang Y: Using MACS to identify peaks from ChIP-Seq data. Curr Protoc Bioinformatics. 2011, Chapter 2: Unit 2 14-PubMedGoogle Scholar
- Hertz GZ, Stormo GD: Identifying DNA and protein patterns with statistically significant alignments of multiple sequences. Bioinformatics. 1999, 15 (7–8): 563-577.PubMedView ArticleGoogle Scholar
- de Hoon MJ, Imoto S, Nolan J, Miyano S: Open source clustering software. Bioinformatics. 2004, 20 (9): 1453-1454. 10.1093/bioinformatics/bth078.PubMedView ArticleGoogle Scholar
This article is published under license to BioMed Central Ltd. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/4.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly credited. The Creative Commons Public Domain Dedication waiver (http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated.