a chromatin immunoprecipitation screen in mouse keratinocytes reveals runx1 as a direct...
TRANSCRIPT
Journal of Cellular Biochemistry 104:1204–1219 (2008)
A Chromatin Immunoprecipitation Screen in MouseKeratinocytes Reveals Runx1 as a DirectTranscriptional Target Of DNp63
Kori Ortt,1 Eli Raveh,2 Uri Gat,2 and Satrajit Sinha1*1Department of Biochemistry, State University of New York at Buffalo, Buffalo, New York 142142Department of Cell and Animal Biology, Silberman Life Sciences Institute, Edmond Safra Campus atGivat-Ram, The Hebrew University, Jerusalem 91904, Israel
Abstract Development of the skin epidermis and appendages such as hair follicles involves coordinated processesof keratinocyte proliferation and differentiation. The transcription factor p63 plays a critical role in these steps as evidentby a complete lack of these structures in p63 null mice. The p63 gene encodes for two proteins TAp63 and DNp63, thelatter being the more prevalent and dominant isoform expressed in keratinocytes. Although numerous p63 target geneshave been identified, these studies have been limited to transformed human keratinocyte cell lines. Here, we haveemployed a genomic screening approach of chromatin immunoprecipitation (ChIP) coupled with an enrichment strategyto identify DNp63 response elements in primary mouse keratinocytes. Analysis of p63-ChIP-derived DNA segmentshas revealed more than 100 potential target genes including novel as well as mouse counterparts of established humanp63 targets. Among these is Runx1, a transcription factor important for hair follicle development. We demonstrate thatDNp63 binds to a p63-response element located within a well-conserved enhancer of the Runx1 gene. Furthermore,siRNA mediated reduction ofDNp63 in mouse keratinocytes reduces Runx1 expression. Consistent with this, endogenousRunx1 levels are lower in the skin of p63þ/� animals as compared to wild type animals. Lastly, we demonstrate thatDNp63 and Runx1 are co-expressed in specific compartments of the hair follicle in a dynamic fashion. Taken togetherour data demonstrate that p63 directly regulates Runx1 gene expression through a novel enhancer element and suggestsa role for these two transcription factors in dictating skin keratinocyte and appendage development. J. Cell. Biochem. 104:1204–1219, 2008. � 2008 Wiley-Liss, Inc.
Key words: transcription; Runx1; p63; keratinocyte; chromatin immunoprecipitation; hair follicle
The p53 family of transcription factorsincludes p73 and p63 each of which share somedegree of structural and functional similarities
to the founding member p53 [Yang et al., 2002].There are, however, unique physiological rolesattributed to each factor that do not involvecanonical p53-linked pathways such as growtharrest and apoptosis, but rather developmentaldecisions. This is particularly evident in thecase of p63, which is highly and selectivelyexpressed in epithelial cells, where it is thoughtto play important roles in stem cell self-renewal,morphogenesis and differentiation programs[Koster and Roop, 2004; Barbieri and Pietenpol,2006; Blanpain and Fuchs, 2007; King andWeinberg, 2007]. The critical role of p63 inregulating various facets of epithelial cellbiology is prominently displayed in the strikingphenotype of p63 null mice [Mills et al., 1999;Yang et al., 1999]. In the absence of p63, animalsdo not survive and exhibit a dramatic loss ofepithelial structures such as the epidermis andmultiple epithelial appendages including teeth,mammary glands, hair follicles, etc.
� 2008 Wiley-Liss, Inc.
This article contains supplementary material, whichmay be viewed at the Journal of Cellular Biochemistrywebsite at http://www.interscience.wiley.com/jpages/0730-2312/suppmat/index.html.
Abbreviations used: ChIP, chromatin immunoprecipita-tion; EMSA, electrophoretic mobility shift assay; HPRT,hypoxanthine guanine phosphoribosyl transferase; K14,Keratin14; ORS, outer root sheath; Swg, sweat gland; TK,thymidine kinase; PMK, primary mouse keratinocytes.
Grant sponsor: NIH; Grant number: AR049238.
*Correspondence to: Satrajit Sinha, PhD, 121 Farber Hall,3435 Main Street, Buffalo, NY 14214.E-mail: [email protected]
Received 19 October 2007; Accepted 17 December 2007
DOI 10.1002/jcb.21700
Our understanding of the molecular mecha-nisms that underlie p63 function in develop-mental pathways has been hampered partly bythe complexity of the p63 isoforms. The p63isoforms are derived from multiple transcriptsgenerated from the p63 gene by alternatepromoter usage and differential splicing [Yanget al., 1998]. Transcripts (TAp63) that encodefor a longer version include a transactivationdomain and are generated from a proximalpromoter. On the other hand, an internalpromoter located within intron 3 directs expres-sion of transcripts that lack this domain(DNp63). In addition, alternative splicingresults in a, b and g isoforms of p63 that areunique at their C-terminus. Despite thesedifferences, all six versions of the p63 proteinshare common DNA-binding and oligomeriza-tion domains, which are analogous to that ofp53.
Since the DNp63 isoforms lack the amino-terminal sequences, it was originally proposedthat these proteins might act primarily asdominant negative regulators. However recentstudies indicate that DNp63 proteins haveintrinsic biological activity and can transcrip-tionally activate or repress target gene expres-sion [King and Weinberg, 2007]. In addition, it isclear that DNp63 is the predominant isoformexpressed in most epithelial cells. In contrast,the expression profile of TAp63 remains miredin controversy with conflicting results fromdifferent laboratories [Mikkola, 2007]. Althoughboth DNp63 and TAp63 show similar biochem-ical properties, in vivo studies have begun toaddress the unique and independent functionsof these proteins [Koster et al., 2004; Suh et al.,2006]. For example, siRNA mediated knock-down of DNp63 in the skin of transgenicmice has demonstrated that DNp63 proteinsare essential for maintaining basement mem-brane integrity and are required for epidermalmorphogenesis and homeostasis [Koster et al.,2007]. Similarly, another study has shown thatthe skin phenotype of the p63 null animals canbe partially rescued by expressing DNp63 butnot TAp63 [Candi et al., 2006]. Taken together,these data strongly suggest a dominant role forDNp63 during the epithelial development anddifferentiation program.
The primary role of p63 involves regulation ofa wide cohort of target genes at the transcrip-tional level. Despite considerable sequence con-servation between the DNA-binding domains of
p63 and p53, it has become apparent that thereare distinct target genes of p63 independentfrom those of p53 [Vigano and Mantovani,2007]. Most of the p63 target genes has beenidentified by global microarray profiling and byChIP-Chip experiments [Barbieri et al., 2006;Carroll et al., 2006; Vigano et al., 2006; Yanget al., 2006]. In keeping with the complex bio-logical role of p63, these studies have uneartheda fascinating and diverse range of p63 targetgenes that are involved in a myriad of cellularprocesses. The discovery of genomic targets ofp63, however, remains far from complete sinceeach search strategy has inherent limitations.What is even more remarkable is the factthat to date none of the p63-target discoveryexperiments have been performed using mousemodels, which have been a steadfast resourcefor functional studies of p63 over the pastseveral years. This is important and relevantin view of growing evidence for species-specificdifferences in binding sites for highly conservedtranscription factors between human andmouse [Odom et al., 2007]. Thus, continuedsearch for p63 targets is warranted.
We have recently developed and utilized amodified ChIP approach to successfully identifytarget genes of DNp63 in HaCaT cells [Birkayaet al., 2007]. This modified ChIP procedureentails amplification of the immunoprecipitatedDNA by ligation-mediated PCR technique fol-lowed by selective enrichment. Here, we havetaken a similar approach to identify targetgenes of DNp63 in primary mouse keratino-cytes. Our screening approach revealed over100 potential DNp63-response elements associ-ated with a variety of target genes. Althoughsome of the genes we identified as potentialDNp63 targets turned out to be the mousecounterpart of human genes previously identi-fied, several novel candidates also emerged.Finally, as a proof of principle to demonstratethe validity of our approach and to furtherelucidate the role of DNp63 in keratinocytes,we chose to focus our studies on another tran-scription factor, Runx1, which emerged as adirect target gene of DNp63.
Runx1 belongs to a small family of transcrip-tional regulators (Runx1-3 in mammals), whichare characterized by a highly conserved DNA-binding domain named after the DrosophilaRunt gene and are evolutionary conserved inall metazoans [Levanon and Groner, 2004].These factors play roles in proliferation versus
DNp63 Regulates Runx1 1205
differentiation decisions in various tissues, instem cell maintenance and in organogenesisfunctioning as activators or repressors oftarget genes [Levanon and Groner, 2004;Mikhail et al., 2006]. In skin, Runx1 is prom-inent in several of the epithelial cell layers ofthe hair follicles and in the bulge, which is thestem cell niche for these appendages [Ravehet al., 2006].
We show that DNp63 binds to a conservedintronic enhancer present in the Runx1 geneand that the expression level of Runx1 isaffected by the levels of DNp63 in keratinocytesin culture as well as in mouse skin. Our studyestablishes that DNp63 and Runx1 are co-expressed in specific cell types within variousskin appendages, such as hair follicles andsweat glands suggesting that these two tran-scription factors might be interlinked in regu-lating the expression of downstream genesimportant for development of these cell types.Collectively our ChIP approach has identifiednumerous mouse target genes ofDNp63 and hasoffered insight into the regulatory pathwayscontrolled by p63 during skin morphogenesis.
MATERIALS AND METHODS
Cell Culture
The mouse keratinocytes utilized in thesestudies have been well characterized anddescribed before [Romano et al., 2006]. Kerati-nocytes were maintained in low calcium mediasupplemented with 10% fetal bovine serum and100 U/ml penicillin and 100 mg/ml streptomycin.Cells were routinely passaged at 90% conflu-ency. PTK2 (rat kangaroo kidney epithelial)cells were grown in minimal essential mediumsupplemented with 10% fetal bovine serum,1% MEM non-essential amino acid solution,100 U/ml penicillin and 100 mg/ml streptomycin.
Chromatin Immunoprecipitation (ChIP) Assaysfor Cloning of Immunoprecipitated Products andConfirmation of p63-Binding to Runx1 Enhancer
Chromatin immunoprecipitation (ChIP)experiments were performed with anti-p63antibodies using mouse keratinocytes. Cellswere grown to 80% confluency and then cross-linked with 1% formaldehyde (Sigma) at 378Cfor 10 min, rinsed two times with cold PBS, andthen harvested in PBS with protease inhibitorsby centrifugation for 5 min at 2,000g. Thecloning of DNA fragments subsequent to the
ChIP procedure was performed as previouslydescribed [Birkaya et al., 2007]. For thispurpose immunoprecipitated DNA was ampli-fied by PCR after addition of a linker element.Linker DNA was removed by digestion withHind III restriction enzyme and purified withPCR purification kit (Qiagen) and clonedinto pBluescript vector. Plasmid DNAs thatcontained inserts that were at least 100 bp werechosen for sequencing at the Roswell ParkSequencing Facility. For in vivo ChIP, newbornmouse skin samples were isolated and treatedwith Dispase II (Roche) overnight at 48C. Theepidermis was separated, finely chopped andwashed with PBS followed by crosslinking with1% formaldehyde in PBS at room temperaturefor 15 min. To stop crosslinking, 0.125 M glycinewas added for 3–5 min followed by washeswith PBS and standard ChIP protocols werefollowed. Binding of DNp63 to Runx1 enhancerwas evaluated by independent ChIP assayutilizing immunoprecipitated DNA from twodifferent anti-p63 antibodies, H-129 (SantaCruz) and RR-14. Primers used are listed inSupplemental Table 2. Real time PCR condi-tions are described for siRNA knockdown of p63.
In Silico Data Analysis
The clones obtained from the ChIPed DNAcloning were sequenced and compared againstmouse genome database by BLAST search ofvarious database including UCSC genomebrowser (http://genome.ucsc.edu/), ENSEMBL(http://www.ensembl.org/index.html), andNCBI(http://www.ncbi.nlm.nih.gov/blast/) to identifypotential target genes. Gene Ontology fromNCBI was used to identify the function of theidentified genes. We utilized the GenomeVISTA(http://pipeline.lbl.gov/cgi-bin/genomevista) pro-gram to obtain the ortholog sequences frommultiple species and generate the VISTA plot.Multiple sequence alignment was done usingCLUSTALW (http:www.ebi. ac.uk/clustalw).The program Transcription element searchsystem (TESS; http://www.cbil. upenn.edu/cgi-bin/tess/tess) was used to identify putativetranscription factor binding sites located withinthe conserved DNA sequence of the Runx1enhancer.
Preparation of Nuclear Extracts andElectrophoretic Mobility Shift Assay (EMSA)
Mouse keratinocytes were grown as describedabove and nuclear extracts prepared by stand-
1206 Ortt et al.
ard methods as described before [Romano et al.,2006]. EMSAs were performed with 5 mg ofnuclear extract and end-labeled double-strandedoligonucleotides. Oligonucleotide sequences arelisted in Supplemental Table 2. For competi-tion assays increasing amounts of unlabeledoligonucleotides were incubated with nuclearextract and labeled oligonucleotides. Protein–DNA complexes were resolved by gel electro-phoresis on 5% non-denaturing polyacrylamidegels in 1�TBE buffer at room temperature. Thegels were dried and visualized by autoradiog-raphy after electrophoresis. Anti-p63 antibod-ies used for supershift experiments have beenpreviously described [Romano et al., 2007].
Plasmids
Recombinant plasmids were constructedusing standard molecular biology protocols.ChIP-DNA sequence corresponding to Runx1enhancer was PCR amplified and cloned intoKpnI and NheI sites of the pGL3-basic vector(Promega) containing the thymidine kinase(TK) minimal promoter. Mutations were intro-duced by PCR amplification. All constructs wereverified by sequencing.
Transient Transfections and Reporter Assays
PTK2 cells were seeded in 6-well platesthe day before transfection. Transfections wereperformed using Fugene 6 reagent (Roche)according to the manufacturer’s protocol. Onemicrogram of each luciferase reporter constructwas transfected per well along with 0.25 mg ofpCMVLacZ plasmid to serve as an internalcontrol for transfection efficiency. Twenty-fourhours posttransfection cell lysates were pre-pared by rinsing cells two times in PBS followedby extracting proteins in 200 ml Lysis Buffer(Promega). Each well was scraped and collectedinto microfuge tubes, vortexed and the debriswas pelleted by centrifugation at 48C. Super-natant solutions were subsequently used inb-galactosidase (Galacto-Light, Applied Bio-system) and luciferase activity assay systems(Promega) following the manufacturer’s proto-cols using a Lumat LB 9501 Luminometer(Berthold). Means and standard deviationswere calculated based on data from threeindependent transfection experiments.
Reverse Transcriptase-PCR
Total RNA from mouse keratinocytes wasisolated and purified by Trizol (Invitrogen)
according to established protocols. Two micro-grams of total RNA from mouse keratinocytes,mouse skin, and spleen was reverse transcribedwith the Thermoscript cDNA synthesis kit(Invitrogen). Primer sequences used are listedin Supplemental Table 2. Jump Start TaqPolymerase (Sigma–Aldrich) was used forPCR amplifications.
siRNA Knockdown of p63 in Mouse Keratinocytes
Mouse keratinocytes were seeded 24 h priorto transfection in 6-well plates. Transfectionswere performed with cells at 30–40% conflu-ency with 10 nM Trp63 siRNA, 10 nM GapdhsiRNA, and 10 nM Negative siRNA (Ambion)using Lipofectamine 2000 (Invitrogen). Cellswere collected 72 h after transfection andanalyzed for knockdown of p63 by Western blotanalysis and quantitative PCR. Anti-p63 anti-bodies(RR-14) were used for detection of DNp63protein levels and the membrane was strippedand re-probed with antibodies against b-tubulinto confirm equal loading. For estimating trans-cript levels, RNA was extracted from p63,Gapdh, and negative siRNA transfected mousekeratinocytes and reverse transcribed with theThermoscript cDNA synthesis kit (Invitrogen).Quantitative PCR was performed with 1ml of 1:5diluted cDNA, 10 mM of each primer and thesuggested amount of SYBER Green I dye (Bio-Rad) according to the manufacturer’s instruc-tions. Cycling parameters were as follows 958Cfor 8 min followed by 35 cycles at 958C for 15 sand 608C for 1 min. Fluorescent data werenormalized to the reference gene b-2-micro-globulin. Primer sequences used are listed inSupplemental Table 2. All reactions wererepeated at least three times.
RNA Isolation From Wild Type and p63þ/� Skin
The p63 heterozygous animals were obtainedfrom Jackson Laboratory (strain B6.129S7-Trp63tm1Brd/J) and were mated to generate wildtype and heterozygote mutant mice. Animalswere genotyped using PCR as described [Millset al., 1999]. Skin from E18.5 wild type andp63þ/� embryos was isolated and processed forRNA isolation using Trizol (Invitrogen) accord-ing to established protocols. Two micrograms oftotal RNA from mouse skin was reversed tran-scribed with the Thermoscript cDNA synthesiskit (Invitrogen). Real-time PCR was performedas described before.
DNp63 Regulates Runx1 1207
X-Gal and Immunohistochemical Stainingof the Runx1þ/lz Animals
The Runx1þ/lz mice utilized in the local-ization studies have been described before[Raveh et al., 2006]. Skin sections (8 mm) werefixed in paraformaldehyde (4% in PBS, 10 min)washed five times in PBS, and incubated in X-gal staining solution at room temperatureovernight. After serial washings, sections weretreated with 1.5% hydrogen peroxide for block-ing endogenous peroxidase activity and incu-bated for 30 min with blocking solutioncontaining 0.05% Triton X-100, 2.5% normalgoat serum (NGS), 1% BSA and 0.25% glycine inPBS prior to 2 h incubation with RR-14 primaryantibodies, diluted 1:200 in 50% blockersolution. Biotinylated anti-rabbit secondaryantibodies were applied diluted in 1.5% NGSPBS for 1 h and detected by the avidin-biotinperoxidase technique (ABC, Vectastain, Vectorlaboratories), using DAB reagent (Sigma).Stained sections were observed and photo-graphed using an Olympus BX51 microscopeand a Magnafire SP digital camera. Confocalimages were taken on a Bio-Rad MRC-1024confocal microscope.
RESULTS
An Improved ChIP Approach to IdentifyDNp63-Response Elements in Primary
Mouse Keratinocytes
Identifying the target genes of DNp63 willbetter elucidate the role of this transcriptionfactor in keratinocyte development and differ-entiation. One approach to identify these targetgenes is to locate the specific genomic region towhich DNp63 is bound in vivo since theseregions are likely to be cis regulatory elementscontrolling the expression of these targets. Suchprior studies for p63 have been limited by thefact that these experiments utilized trans-formed cell lines of human origin and antibodiesthat recognize both DNp63 and TAp63. Toovercome such caveats, we utilized a ChIP-based screening strategy with primary mousekeratinocytes, which are known to express highlevels of DNp63. Furthermore, to specificallyprobe the DNp63 proteins, we used the RR-14antibody, which has been developed inour laboratory and is directed against an N-terminal peptide unique to this isoform. Wehave utilized this antibody successfully in prior
ChIP experiments in both human and mousecells [Birkaya et al., 2007; Romano et al., 2007].Primary mouse keratinocytes were grown in amedium containing low concentration of Caþþ
to ensure that the cells do not differentiate andstratify. The keratinocytes were processed forChIP experiments when they were in activestate of proliferation (�80% confluent), since atthis stage these cells contain the highest levelsof DNp63 (unpublished data). The strategy toenrich, clone and sequence the immunoprecipi-tated DNA has been described in detail before[Birkaya et al., 2007]. Briefly, we utilized aligation-mediated PCR technique, to amplifythe genomic immunoprecipitated DNA, whichwas then enriched and purified by incubationwith agarose beads containing GST-DNp63protein to reduce the contamination with non-specific DNA fragments. This enrichment pro-cedure allowed us to select for DNA fragmentsthat are more likely to contain binding-sitesfor DNp63. Following immunoprecipitation andpurification, the DNA fragments were clonedinto the pBluescript vector. The average size ofthe DNA fragment obtained after the ChIPprocedure was �250 bp. We sequenced 280individual clones that represented the library ofDNA fragments immunoprecipitated byDNp63.
The p63-ChIPed Sequences Correspondto Known and Novel In Vivo Target Genesof DNp63 in Primary Mouse Keratinocytes
We analyzed the sequences for everyretrieved ChIP fragment by mapping them tothe mouse genome database by using a varietyof search programs of University of CaliforniaSanta Cruz genome browser, ENSEMBL, or theBLAST program at NCBI. This allowed us todetermine the location of the DNp63 immuno-precipitated DNA fragments in relation toknown or predicted genes. This study revealedthat 109 DNA fragments immunoprecipitatedby DNp63 antibodies were embedded within orlocated near known, annotated, or predictedgenes. A small number of sequences were ofunknown origin since their identity could not beverified in the databases, while a few sequenceswere of bacterial origin, possibly due to con-tamination. We chose to assign the DNA frag-ments to a specific gene if the sequence matchedto the intragenic region or a segment within100 kb upstream or downstream. The 109 gene-associated DNp63-binding fragments identifiedby this approach are listed in Supplemental
1208 Ortt et al.
Table 1 with their precise genomic coordinates.Approximately 16% of the genes we identifiedin this study were the mouse counterpartsof human genes identified by other studiesdemonstrating the validity of our approach(Table I).
In most cases, the ChIPed DNA sequencesmapped to a genomic neighborhood that wasinhabited by or in close proximity to a singlegene making it easy to identify the putativeDNp63 target. However, in some cases the DNAfragment that was recovered was located inbetween two genes. For these cases, we havelisted both of the genes as DNp63 targets. Forexample, one of the immunoprecipitated DNAfragments mapped to a region that is 16 kb fromthe 50 end of Cdh1 (gene encoding for P-Cadherin) and 30 kb from the 30 end of Cdh3(gene encoding for E-Cadherin). In view of theknown biological role of P-Cadherin and E-Cadherin in epidermal and hair development, itis quite possible that the p63-response elementlocated in the intergenic region may coordin-ately regulate the expression of both thesegenes. One of the interesting target sitesuncovered by our screen corresponded to apreviously described enhancer for the p63 geneitself, located within an intron �42 kb down-stream from the DNp63 transcriptional startsite. It has been shown by ChIP experimentsthat the enhancer is bound by p63 suggesting anauto-regulatory role for p63 [Antonini et al.,2006]. Our data confirms the previous findingsand provides further evidence for the efficiencyof our approach in uncovering bona-fide p63bound genomic fragments.
We found that 44% percent of the DNA frag-ments that were immunoprecipitated mappedwithin the region spanning 100 kb upstream ordownstream of candidate target genes (Fig. 1A).Although many sites were found in the 50 and 30
end of the genes, p63-binding sites were alsofound within a gene, particularly in intronicsegments. Indeed, we found that 45% of theDNA fragments immunoprecipitated werelocated within an intron of known or predictedgenes, with an overwhelming preference for thefirst intron (Supplemental Table 1 and Fig. 1B).This finding is consistent with our previous dataof DNp63 target genes in HaCaT cells and withthe emerging trend of transcription factor-binding sites across the genome. This suggeststhat DNp63 may regulate the gene expressionof numerous target genes by associating and
binding to enhancer elements embedded withinthese genes. Intriguingly, only a few bindingsites were found in exons corresponding toprotein-coding domains, suggesting that poten-tial transcription factor-binding sites in exonsmay be under negative selection and thusnot associated with cis-regulatory functions.Examination of these potential target genesrevealed that these genes are involved in amyriad of cellular processes including celladhesion, transcription, and signal transduction(Supplemental Table 1 and Fig. 1B). Finally, itis important to note that a small percent of theDNA fragments were found to be in gene desertsand thus could not be assigned to any known orpredicted gene. Notwithstanding the obviouspossibility of false positives, these sites mayalso reflect enhancer elements located far awayor regulatory elements associated with un-annotated genes, non-coding RNAs or noveltranscriptional units.
The Immunoprecipitated DNA Fragment Mappingto the Runx1 Gene Is Highly Conserved
Among the putative p63 target genes thatwere identified, we chose to focus our attentionon the transcription factor Runx1, which isknown to play an important role in the develop-ment and differentiation of several tissuesincluding the hair follicles. First, we examinedthe sequence of the DNA fragment that wasimmunoprecipitated and recovered from theDNp63 ChIP screening. Interestingly, this DNAfragment corresponding to an intronic region ofthe Runx1 gene is highly conserved acrossspecies, suggesting that this element mayfunction as a potential enhancer (Fig. 2A,B).Two promoters drive the expression of differentisoforms of Runx1 [Ghozi et al., 1996]. Whilea distal promoter P1 directs expression ofRunx1c, the proximal promoter P2 drivesexpression of Runx1b and Runx1a isoforms.Runx1b is the major isoform in most cell types,and in some cases alternative splicing generatesRunx1a (Fig. 2A). The DNA fragment immuno-precipitated in our ChIP approach is embeddedwithin an intron common to both Runx1b andRunx1c making it possible that both isoformsare regulated by DNp63. To test this, wedesigned primers specific for each isoform,indicated by the arrows in Figure 2A. We foundthat only Runx1b is expressed in mouse kera-tinocytes grown in culture and in the epidermalcompartment of the skin, strongly suggesting
DNp63 Regulates Runx1 1209
TA
BL
EI.
Lis
to
fS
ele
cte
dD
Np
63
Mo
use
Ta
rg
et
Gen
es
Iden
tifi
ed
by
aM
od
ified
Ch
IPA
pp
ro
ach
Ch
rom
Gen
eD
escr
ipti
onL
ocati
onA
cces
sion
#F
un
ctio
nS
tart
En
dR
efer
ence
s
4C
7M
ap
1lc
3M
icro
tubu
le-a
ssoc
iate
dp
rote
in1
ND
NG
_005200
Un
kn
own
105616612
105617006
Yan
get
al.
[2006]
4D
2.3
Ibrd
c3IB
Rd
omain
con
tain
ing
330en
dX
M_2
04030
Pro
tein
-bin
din
g128598215
128598341
Yan
get
al.
[2006]
3F
1Iv
lIn
vol
ucr
in50en
dN
M_0
08412
Str
uct
ura
lp
rote
in92665045
92665110
Yan
get
al.
[2006]
2E
2F
jx1
Fou
rjo
inte
dbox
1(D
roso
ph
ila)
30en
dN
M_0
10218
Sig
nali
ng
102234945
102235357
Yan
get
al.
[2006]
7F
1P
lek
ha7
Ple
ckst
rin
hom
olog
yd
omain
con
tain
ing,
fam
ily
mem
750en
dN
M_1
72743
Un
kn
own
116047718
116047842
Yan
get
al.
[2006]
5G
2A
uts
2A
uti
smsu
scep
tibil
ity
can
did
ate
2In
tron
5N
M_1
77047
Un
kn
own
131860963
131861133
Yan
get
al.
[2006]
17C
Dd
r1D
isco
idd
omain
1re
cep
tor
fam
ily
150en
dN
M_0
07584
Cel
lad
hes
ion
35380228
35380413
Osa
da
etal.
[2005],
Carr
oll
etal.
[2006],
Yan
get
al.
[2006],
Bir
kaya
etal.
[2007]
18
A1
Arh
gap
12
Rh
oGT
Pase
act
ivato
rp
rote
in12
30en
dN
M_0
2977
Sig
nal
tran
sdu
ctio
n5948618
5948762
Tru
ong
etal.
[2006]
12
A2-A
3D
nm
t3a
DN
Am
eth
ylt
ran
sfer
ase
Intr
on1
NM
_007872
Tra
nsc
rip
tion
3848079
3848465
Yan
get
al.
[2006]
11
E1
Sm
urf
2S
MA
Dsp
ecifi
cE
3u
biq
uit
inp
rote
inli
gase
2ex
11/i
n11
NM
_025481
Pro
tein
fold
ing
106657541
1066557820
Vig
an
oet
al.
[2006]
16
C3.3
Mrp
l32
Mit
och
ond
rial
ribos
omal
pro
tein
L39
30en
dN
M_0
17404
Ele
ctro
ntr
an
spor
t84577487
84577657
Yan
get
al.
[2006]
10
A2
Gp
r126
Gp
rote
in-c
oup
led
rece
pto
r126
30en
dN
M_0
01002268
Sig
nali
ng
14055593
14055746
Yan
get
al.
[2006]
16
B1
Trp
63
Tra
nsf
orm
ati
onre
late
dp
rote
in63
Intr
on2
NM
_011641
Tra
nsc
rip
tion
25759148
25759340
Barb
ieri
etal.
[2006]
an
dY
an
get
al.
[2006]
12
D3
Sto
n2
Sto
nin
2In
tron
1N
M_1
75367
Tra
nsp
ort
921515653
92152038
Yan
get
al.
[2006]
8E
1C
dh
13
Cad
her
in13
Intr
on3
NM
_019707
Cel
lad
hes
ion
121597853
121597928
Carr
oll
etal.
[2006]
XA
2-A
3G
6p
dX
Glu
cose
-6-p
hos
ph
ate
deh
yd
rogen
ase
X-l
ink
edIn
tron
2N
M_0
08062
Met
abol
ism
70676886
70676929
Yan
get
al.
[2006]
5E
5C
ds1
CD
Pd
iacy
lgly
cero
lsy
nth
ase
1In
tron
5N
M_1
73370
Met
abol
ism
102043506
102043628
Yan
get
al.
[2006]
1A
3X
kr9
XK
ell
blo
odgro
up
pre
curs
orre
late
dfa
mil
ym
ember
Intr
on1
NM
_001011873
Mem
bra
ne
pro
tein
135865082
135865154
Yan
get
al.
[2006]
3G
1P
tbp
2P
olyp
rim
idin
etr
act
-bin
din
gp
rote
in2
Intr
on6
NM
_019550
Pro
tein
-bin
din
g119736125
119736459
Yan
get
al.
[2006]
16
C4
Ru
nx1
Ru
nt
rela
ted
tran
scri
pti
onfa
ctor
1In
tron
3N
M_0
09821
Tra
nsc
rip
tion
92515494
92515729
Tru
ong
etal.
[2006]
an
dY
an
get
al.
[2006]
14
D3
Ccd
c122
Coi
led
-coi
ld
omain
con
tain
ing
122
Intr
on1
NM
_175369
Un
kn
own
75772117
75772382
Yan
get
al.
[2006]
8B
1.1
Fat1
Fat
tum
orsu
pp
ress
orh
omol
og(D
roso
ph
ila)
Intr
on2
XM
_885736
Cel
lad
hes
ion
46490981
46491956
Carr
oll
etal.
[2006]
13
A4
Gfo
d1
Glu
cose
-fru
ctos
eox
idor
edu
ctase
dom
ain
con
tain
ing
Intr
on1
NM
_001033399
Met
abol
ism
43242969
43243250
Yan
get
al.
[2006]
16
B2
Cld
n1
Cla
ud
in1
50en
dN
M_0
16674
Tra
nsp
ort
26349815
26350037
Carr
oll
etal.
[2006],
Tru
ong
etal.
[2006],
Yan
get
al.
[2006]
16
B2
Cld
n16
Cla
ud
in16
50en
dN
M_0
53241
Tra
nsp
ort
26349815
26350037
Yan
get
al.
[2006]
18
B3
Cen
td3
Cen
tau
rin
,d
elta
350en
dN
M_1
39206
Sig
nal
tran
sdu
ctio
n38123967
38124462
Yan
get
al.
[2006]
11
B1.3
An
krd
34
An
kyri
nre
pea
td
omain
43
50en
dN
M_1
83173
Pro
tein
-bin
din
g53333445
533333592
Yan
get
al.
[2006]
6A
2S
t7S
up
pre
ssio
nof
tum
orig
enic
ity
750en
dN
M_0
22332
Un
kn
own
17668908
17669384
Yan
get
al.
[2006]
12
F1
Rage
Ren
al
tum
oran
tigen
Intr
on9
NM
_011973
Kin
ase
act
ivit
y111256343
111256729
Yan
get
al.
[2006]
2A
1G
ata
3G
ata
-bin
din
gp
rote
in3
30en
dN
M_0
08091
Tra
nsc
rip
tion
9682830
9683136
Yan
get
al.
[2006]
8D
3C
dh
3C
ad
her
in3
30en
dN
M_0
01037809
Cel
lad
hes
ion
109461118
109461449
Carr
oll
etal.
[2006]
an
dY
an
get
al.
[2006]
8D
Cd
h1
Cad
her
in1
50en
dN
M_0
09864
Cel
lad
hes
ion
109461118
109461449
Carr
oll
etal.
[2006]
an
dY
an
get
al.
[2006]
14
A3
Arh
gef
3R
ho
gu
an
ine
nu
cleo
tid
eex
chan
ge
fact
or(G
EF
)3
Intr
on3
NM
_027871
Sig
nal
tran
sdu
ctio
n2619197589
26197814
Yan
get
al.
[2006]
17
A2-A
3S
yn
j2S
yn
ap
toja
nin
2In
tron
2N
M_0
11523
RN
A-b
ind
ing
5942962
5943160
Yan
get
al.
[2006]
1210 Ortt et al.
Fig. 1. A: Location of the identified p63 ChIPed DNA sites relative to known genes. Examination of 124potential target genes demonstrates the p63-binding sites are located within introns as well as the 50 and 30
end of these target genes. B: Target genes of DNp63 identified by ChIP are involved in a myriad of cellularprocesses.
Fig. 2. DNA fragment isolated by ChIP is embedded within anintron of the Runx1 gene. A: Shown is the genomic locus of twodifferent isoforms of Runx1 that are generated from two distinctpromoters (P1 and P2). The putative enhancer element maps toan intron common to both isoforms and is conserved acrossspecies. This enhancer element contains a putative p63-bindingsite shown in the box. Also shown are predicted response
elements for additional transcription factors. B: Genome Vista of5 kb of the Runx1 intron containing the putative enhancerelement. The enhancer segment indicated by the box is highlyconserved across several species. C: RT-PCR demonstratesexpression of Runx1b in skin, primary mouse keratinocytes(PMK), and spleen, while Runx1c is expressed only in spleen.HPRT was used as a loading control.
DNp63 Regulates Runx1 1211
that DNp63 regulates expression of onlyRunx1b (Fig. 2C). It is possible that in othercell types, the expression of Runx1c is alsounder control of this enhancer.
Independent ChIP Assay DemonstratesBinding of DNp63 to Runx1
To clearly demonstrate that Runx1 is a directtarget gene of DNp63 we performed an inde-pendent ChIP assay (Fig. 3). We used twodifferent anti-p63 antibodies, the DNp63 N-terminal specific antibody RR-14, as well as analpha-specific C-terminal antibody, H-129 toimmunoprecipitate crosslinked chromatin frommouse keratinocytes as well as mouse skintissue. We designed a set of primers to amplifythe putative Runx1 enhancer element identifiedin our screen and as a negative control, we useda set of primers that amplify a genomic segmentin the Gapdh gene. As shown by PCR data inFigure 3, specific enrichment of the Runx1enhancer was observed after immunoprecipita-tion with both antibodies against p63 ascompared to the IgG control. In contrast, therewas no localization of DNp63 to the Gapdhgenomic locus. Three independent ChIP experi-ments yielded similar results. In addition, weperformed quantitative PCR to examine therelative fold enrichment of the Runx1 enhancerwhen immunoprecipitated with anti-p63 anti-bodies (Fig. 3B). We determined the relative foldenrichment of Runx1 by comparing the amountof Runx1-specific PCR product amplified from
the anti-p63 ChIP to that from the negativecontrol (IgG) ChIP samples. The significantenrichment we observed when we immunopre-cipitated with the two anti-p63 antibodiesconfirms that the Runx1 enhancer is occupiedbyDNp63 protein in mouse keratinocytes grownin culture and more importantly in skin tissuein vivo.
DNp63 Directly Binds to the p63-ResponseElement Located Within the Intronic
Region of the Runx1 Gene
The results of the ChIP assays prompted us totest if DNp63 directly binds to the enhancerelement. Therefore, we scanned the enhancerelement for the presence of a p63 consensusDNA-binding site as defined by several recentstudies [Ortt and Sinha, 2006; Perez et al.,2007]. The Runx1 intronic element that wasimmunoprecipitated byDNp63 contains a bona-fide p63 consensus site, with the core sequencebeing completely conserved among severalspecies (Fig. 2). To determine if DNp63 bindsto this response element, we designed anoligonucleotide probe (WT) for this site andperformed electrophoretic mobility shift assay(EMSA). Upon incubation of the labeled oligo-nucleotide probe containing the putative Runx1p63-response element and nuclear extractfrom mouse keratinocytes, a specific complexwas observed (Fig. 4, lane 1). Competitionassays with the p63-response element fromthe K14 enhancer, which has previously been
Fig. 3. Independent ChIP assay demonstrates binding of DNp63 to the Runx1 putative enhancer element.ChIPwas performed inmousekeratinocytes andmouse skinwithN-terminalRR-14andC-terminalH-129anti-p63 antibodies. A: Enrichment of Runx1 enhancer element relative to IgG on an agarose gel. B: Real-time PCRdemonstrating the relative enrichment of Runx1 enhancer element with p63 antibodies and IgG.
1212 Ortt et al.
demonstrated to contain a bona-fide p63-bind-ing site, as well as a probe containing aconsensus p53-binding site were used to dem-onstrate specificity [Romano et al., 2007]. Whenwe incubated nuclear extract with increasingamounts of unlabeled oligonucleotide probescontaining the K14 p63-binding site, a consen-sus p53-binding site, or the wild type sequences,the complex was abolished (lanes 2–7). On theother hand, excess amounts of oligonucleotideprobes containing two different mutations inthe Runx1 p63-binding site (MT1,CTTG>ATTC; MT2, CTTG>ATTG) had noeffect (lanes 8–11). Finally, the addition of anti-DNp63 specific antibodies (but not preimmuneserum) resulted in a distinct supershift of thecomplex, demonstrating that this complex con-tains DNp63 (lanes 12–13). Our EMSA studiesclearly demonstrate that the Runx1 enhancerelement contains a p63-binding site and thatendogenous DNp63 present in mouse keratino-cytes can bind to this site.
DNp63 Transcriptionally Regulatesthe Runx1 Enhancer Element
Having demonstrated thatDNp63 can directlybind to the putative Runx1 enhancer, our nextgoal was to determine if the enhancer elementwas transcriptionally responsive to DNp63 intransient transfection experiments. We utilizedPTK2 cells for these experiments since thesecells have been shown to lack endogenous p63expression. The Runx1 enhancer element wascloned upstream of the heterologous minimalTK promoter. The reporter plasmid was co-transfected with either an expression plasmidencoding for HA-tagged DNp63 a, b, or g or anempty HA control vector. Expression ofDNp63 a,b, or g resulted in increased levels of reporteractivity (four to sixfold) as demonstrated inFigure 5. Interestingly, TAp63g, the most potenttransactivating p63 isoform, also resulted in anincrease level of reporter activity. In contrast,when a mutation was introduced in the p63-binding site (MT1, the mutation that abolishesbinding of p63 as shown in EMSA experiments),all three DNp63 isoforms and TAp63g failed toactivate the reporter construct. Collectively,these data strongly suggest that p63 activatesthe Runx1 enhancer element through its bindingto the p63-binding site.
Fig. 4. EMSA demonstrates binding of DNp63 to the conservedp63-binding site embedded within the Runx1 putative enhancerelement. EMSAs were performed with radiolabeled oligonucleo-tide probe containing the p63-binding site corresponding to theRunx1 enhancer element. Probe was incubated with nuclearextract from mouse keratinocytes. Oligonucleotides containingknown p63 response elements, compete while mutant oligonu-cleotides have no effect at 20- and 100-fold excess amounts.Supershift with anti-p63 antibodies confirms that the complexcontains DNp63. PIS, preimmune serum
Fig. 5. Runx1 enhancer element is responsive to all isoforms ofDNp63 and TAp63g. Luciferase construct containing the Runx1enhancer element cloned upstream of a heterologous TKpromoter demonstrates high levels of activity when co-trans-fected with expression plasmids encoding DNp63 a, b, g orTAp63g in PTK2 cells. Luciferase values were determined andnormalized against b-galactosidase values. Corrected luciferasevalues for cells transfected with empty vector pCMV-HA were setat 1. TK, thymidine kinase.
DNp63 Regulates Runx1 1213
Knockdown of p63 Expression in MouseKeratinocytes Affects Runx1 Gene Expression
Having shown that DNp63 can bind andactivate the Runx1 enhancer, we next wantedto demonstrate whether DNp63 is directlyinvolved in regulating expression of the Runx1gene. First we examined the effect of knockingdown p63 expression in mouse keratinocytes byemploying synthetic dsRNA (siRNA) to reducep63 expression. We observed a significantreduction in both the mRNA and protein levelsof p63 in mouse keratinocytes transfected withthe p63 siRNA (Fig. 6A,B). We synthesizedcDNA from three independent RNAi experi-ments and performed quantitative RT-PCRassays to examine Runx1 gene expression inmouse keratinocytes that had reduced p63levels. As a reference standard, we used thehousekeeping gene B2M (beta2 microglobulin)whose expression is not thought to be dependenton p63 or Runx1. Expression of Runx1 wasdramatically reduced (�60%) in keratinocytestransfected with p63 siRNA, whereas cellstransfected with control siRNA showed no
alteration in Runx1 levels. As demonstratedbefore, expression of K14, another direct targetof DNp63 was also reduced in the p63 siRNAtreated keratinocytes [Romano et al., 2007].
Since p63 null animals die soon after birthand exhibit a complete dissolution of skin tissue,it is difficult to assess the alterations of putativep63 target genes in these mice. However, wereasoned that examination of skin samples fromthe p63 heterozygous animals, which appearphenotypically normal, should reveal whetherreduction in p63 levels affects Runx1 geneexpression. For this purpose, we isolated RNAfrom skin tissue of newborn p63 heterozygousand control wild type animals. By quantitativeRT-PCR, we show that p63 levels are 62% lowerin the p63 heterozygous animals as expected(Fig. 6C). Interestingly, in agreement with ourp63 knockdown experiments with keratinocytesin culture, p63 heterozygous animals alsoexhibit lower levels of Runx1 and K14 geneexpression as compared to wild type. In light ofour ChIP, EMSA and transient transfectionexperiments, these data argues that Runx1 is abona-fide target gene of DNp63 in mouse skin.
Fig. 6. Runx1 levels are reduced when p63 expression isknocked down by siRNA in mouse keratinocytes and in p63heterozygous skin. Mouse keratinocytes were transfectedwith negative control siRNA, Gapdh siRNA, and p63 siRNA.A: Western Blot using the RR-14 antibody demonstratesdecreased expression of DNp63. b-tubulin was used as a loading
control. B: Expression of Runx1 was analyzed by real-time PCR.Negative control siRNA-transfected cells were used as a control.C: Skin from wild type and p63þ/� embryos was isolated.Expression levels of DNp63 and Runx1 in skin were analyzed byreal-time PCR. Gapdh and K14 were used as controls. Asterisksindicate statistical significance derived from Student’s t-test.
1214 Ortt et al.
DNp63 and Runx1 Are Co-Localized inthe Outer Root Sheath of the Hair Follicle
Our biochemical data strongly suggests thatRunx1 is a direct target gene of DNp63. We nextsought to examine the in vivo expressionpattern of Runx1 and DNp63 to determine ifthey are co-expressed in skin epithelium and inparticular in hair follicles. This is of particularinterest since we have recently reported thatanimals with skin-specific knockout of Runx1show aberrations in hair follicle morphogenesisassociated with specific structural defects.Hence to analyze the distribution patterns ofRunx1 and DNp63, we investigated theirexpression in mouse skin during embryogenesisand postnatal development. For this purpose,we utilized mice that have the LacZ reportergene integrated into the Runx1 genomiclocus such that X-gal staining mimics endoge-nous Runx1 expression. As shown before, atembryonic day 14.5 (E14), the expression of
Runx1 is restricted to the dermal compartment(Fig. 7A). This dermal region is distinct from theoverlying epidermal layer, which shows strongexpression of DNp63 as detected by DNp63-specific RR-14 antibody. However, at a laterstage of development (E18.5), when hair fol-licles formation has progressed further, nuclearDNp63 expression is co-localized with Runx1along the outer root sheath (ORS) and in thematrix of the hair follicles (Fig. 7A,B). Thereis also strong DNp63 expression in the basalcells of the epidermis, however no blue stainingis observed in the interfollicular epidermissuggesting the lack of Runx1 expression in thisregion of the skin. The pattern of co-localizationin the hair follicle is even more apparent inhair follicles at birth (NB) and postnatally(Fig. 7A,B). In view of the intriguing expressionprofile of DNp63 and Runx1 in hair follicle, wedecided to investigate their expression in otherappendages of the skin. As shown in Figure 7C,in the nail, DNp63 is expressed more in basal
Fig. 7. DNp63 and Runx1 are co-expressed in the outer rootsheath (ORS) of the hair follicle. Expression of Runx1 (X-galstaining, Blue) and DNp63 (immunostaining, brown) during hairfollicle morphogenesis and in nails and sweat glands. A: Atembryonic day 14.5 (E14), Runx1 dermal expression is distinctfrom the epidermal expression of DNp63. At E18.5, nuclearDNp63 expression is co-localized with Runx1 along the ORS andin the hair matrix of the hair follicles, while only DNp63 is alsoexpressed in the basal cells of the epidermis (indicated by the
arrows). This pattern of co-localization is more apparent at birth(NB). B: In adult back skin and whiskers, DNp63 is co-localizedwith Runx1 in the lower portion of the ORS (secondary germzone; indicated by the arrows). C: In the nail, DNp63 is expressedin the matrix and the nail bed in more differentiated cells.In eccrine sweat glands of the toes (Swg), DNp63 and Runx1 areco-expressed in the proximal loops. Scale bars¼50mm (except forNail and Swg¼100 mm).
DNp63 Regulates Runx1 1215
cells in the matrix and the nail bed compared toRunx1, which is expressed in more differenti-ated cells, suggesting that there is little overlapof these two transcription factors in this tissue.On the contrary, in the eccrine sweat glands ofthe toes, DNp63 and Runx1 were found to be co-expressed in the proximal loops (Fig. 7C). Takentogether, these data suggest that DNp63 andRunx1 expression overlap in some regions of theskin tissues, but not all. In particular Runx1and p63 only partially co-localize in the mousehair follicle, with p63 mainly detected in theproliferating cells of the ORS and hair bulbmatrix, while Runx1 expression is broader andextends into differentiated cells of the hairshaft.
DISCUSSION
Identification of binding sites for p63 and thecorresponding target genes is essential fordeciphering the regulatory circuits throughwhich this transcription factor controls variouscellular processes in the epithelium. Althoughglobal studies such as microarray profilingassays to probe the transcriptome status inp63 over expressing, knockdown or null celllines have provided a glimpse into p63-regu-lated pathways and networks, these types ofapproaches have inherent limitations. Forexample, these studies fail to discriminatedirect from indirect effects of transcriptionfactors and a priori, do not reveal the locationof regulatory response elements that controltarget gene expression. Recent progress inisolation and mapping of target transcriptionfactor-binding sites by ChIP-based technologyhave proven to be useful in overcoming someof these shortcomings [Kim and Ren, 2006].One such strategy, ChIP-chip identifies target-binding regions directly by probing the immu-noprecipitated DNA using a DNA microarraythat tiles extensive regions of the genome.However, with improved ultrahigh-throughputsequencing capabilities, direct DNA sequencingof ChIP fragments (ChIP-Seq) is gainingmomentum and has been successfully per-formed for some transcription factors.
The network of potential p63 targets hasbecome extensive primarily due to ChIP-chipstudies that have been performed recently[Vigano et al., 2006; Yang et al., 2006]. Althoughthese studies have provided an exhaustive list ofp63-response elements and putative target
genes that are activated and repressed by p63,the database is still evolving and far fromcompletion. The fact that ChIP experimentshave only been performed on transformedhuman cell lines undermines the physiologicalrelevance of the data. In addition, antibodiesutilized in these experiments recognize both theTA andDN isoforms of p63, further adding to thecomplexity in data interpretation. It is verylikely that the distribution of p63 across the vastarray of genomic segments and the consequenttranscriptional effects on a battery of targetgenes is a dynamic event that is dictated by theinherent nature of the cellular environment.The ChIP-based approach can thus provide onlya snapshot of the p63-DNA interactions, whichdepending upon cell type and experimentalconditions may vary significantly. Hence, it isimportant to probe the dynamics of p63 and itstarget DNA interaction under a variety ofexperimental conditions to achieve a compre-hensive view of its biological function.
With that in mind, in this study we haveutilized ChIP in combination with sequencingand mapping of isolated genomic DNAs tolocate binding sites for DNp63 and its potentialtarget genes in primary mouse keratinocytesin vivo. Although sequencing ChIPed DNAfragments (ChIP-Seq) is potentially straightforward, limiting amounts of DNA obtainedduring the procedure and the overwhelmingexcess of non-specifically precipitated DNA isa complicating issue. Hence, there is a need toexplore experimental parameters important forenhancing the performance of ChIP-seq strat-egy particularly for complex genomes such asthose of mammalian cells. We have previouslydemonstrated a modified and improved ChIPapproach to successfully identify target genesof DNp63 in HaCaT cells, which are immortal-ized human keratinocytes [Birkaya et al.,2007]. Since the typical ChIP procedure gen-erates only minute amounts of immunoprecipi-tated genomic DNA, which is often associatedwith non-specific DNA contamination, directsequencing and identification of the genomicDNA presents a number of technical chal-lenges. We have overcome these limitations byincorporating a PCR-based amplification andGST-purification step to enrich the immuno-precipitated genomic DNA. As we demonstratein this report, this strategy allowed us toidentify numerous target genes of DNp63 inprimary mouse keratinocytes.
1216 Ortt et al.
A careful evaluation of the �100 DNp63-reponse elements and their correspondinggenes lead to numerous interesting observa-tions. A number of the genes identified in ourstudy were the mouse counterparts of humangenes identified in previous studies. There arealso a significant number of genes that havethus far not been reported to be direct targets ofp63. These might represent a subset of p63targets that are unique to DNp63 since manyof the prior studies did not discriminatebetween theDN and TA isoforms. Alternatively,differences in cell type, growth conditions, andantibodies may account for the lack of completeoverlap in p63 targets obtained from variousstudies. We believe that most of the ChIPedDNA fragments represent true p63-responsetargets since these DNA segments contain oneor more p63-binding sites as identified bysequence comparison with the p63 consensusbinding site. This coupled with the fact thatmany of the target genes are linked to pathways(such as the Notch pathway) that have beenimplicated to be under the control of p63 or haveappeared in multiple other screens for p63targets (such as DDR1), strongly suggests thefunctional relevance of these segments. Inter-estingly, most of the response elements werefound to be located in intragenic regions or insegments that are located far away from thepromoters demonstrating the propensity forDNp63 to regulate gene expression throughthe utilization of distal enhancer elements. Thisseems to be a recurring theme for a vastmajority of transcription factors and shouldbe an important consideration when tryingto decipher locations of transcription factor-binding sites.
It is important to identify the DNp63 targetgenes to better understand its role in thedevelopment of stratified epidermis and asso-ciated appendages. We have focused our studyon Runx1 to illustrate the validity of our ChIPbased search approach for DNp63 target genesand to evaluate its role as a potential target ofDNp63 in keratinocytes. Runx1 is a memberof a family of three transcription factors inmammals that are involved in the control ofproliferation and differentiation in varioustissues. The role of Runx1 as a regulator ofhematopoiesis is well established, particularlyfor megakaryocytic maturation and lymphocytedifferentiation [Mikhail et al., 2006]. However,it is becoming increasingly clear that Runx1 can
also play a role in the homeostasis of othertissues such as the nervous system and skeletalmuscles [Wang et al., 2005; Kramer et al., 2006;Marmigere et al., 2006]. Of the three Runxproteins, Runx1 is prominently expressed inmouse skin, exhibiting a dynamic expressionpattern particularly in the hair follicles. Usingepidermal-targeted Runx1 knockout mice wehave shown that Runx1 affects hair structureand morphogenesis through its involvement inthe differentiation program of the hair shaft[Raveh et al., 2006].
Here we identify a putative enhancer elementwithin the Runx1 gene that is evolutionarilyconserved. We show that this putative enhancerelement is occupied by DNp63 in vivo and canbe transcriptionally activated by all isoformsof DNp63 in transient transfection assays.Importantly, DNp63 is a key transcriptionalmediator of Runx1, since expression of Runx1 isdecreased in mouse keratinocytes that havereduced p63 levels. The Runx1 gene gives rise totwo major mRNA products, Runx1c and Runx1bthat result from transcriptional initiation ateither a distal P1 or proximal P2 promoter.The novel enhancer described in our study isembedded within the Runx1 gene in an intronicregion, which is common to both Runx1b andRunx1c. However, since we observe thatRunx1b is the predominant isoform expressedin mouse skin and keratinocytes in culture,we speculate that this enhancer is likely toinfluence the activity of the proximal P2promoter, which is relatively closer as comparedto the distal P1 promoter. Prior studies evaluat-ing the transcriptional regulation of Runx1have been focused primarily on B cells and haveimplicated Ets sites, CCAAT boxes and Runx1elements present in the proximal promoter[Ghozi et al., 1996]. Interestingly we find thatthere are no p63-response elements in thepromoter regions of Runx1 (unpublished data).It will be interesting to test if the enhancerreported here also plays a role in other cell typesand if p63 or any of its closely related familymembers influence the enhancer activity inother cell types as it does in keratinocytes.
One interesting observation from our studiesis the fact that Runx1 and p63 do not completelyoverlap in the various keratinocyte cell types inthe mouse skin. This is clearly evident by thefact that whereasDNp63 is present primarily inboth the proliferating cells of the interfollicularepidermis and the ORS of the hair follicles,
DNp63 Regulates Runx1 1217
Runx1 expression is restricted to follicularepithelial cells that also include the differ-entiated cells of the inner root sheath. In hairfollicle development, there is evidence for a rolefor GATA-3 and Sox9, both of which are not onlydirect transcriptional targets of p63 but alsohave been functionally linked with Runx1[Kaufman et al., 2003; Vidal et al., 2005; Zhouet al., 2006; Naoe et al., 2007]. In this regard, it isimportant to note that the Runx1 intronicenhancer identified in this study contains bind-ing sites for both GATA and Sox family oftranscription factors. Thus an interesting pos-sibility is that the development and differ-entiation of hair follicle keratinocytes involvesextensive crosstalk and feedback among thesecritical transcription factors.
There are a number of skin diseases andsyndromes involving Ectrodactyly, EctodermalDysplasia and alopecia in human patients thatare associated with mutations in the p63 gene[van Bokhoven and McKeon, 2002]. In view ofour findings, it is tempting to speculate that thehair follicle phenotype in these patients couldbe a consequence of de-regulated Runx1 expres-sion. Our identification of a wide range of in vivotarget genes of DNp63 provides only a glimpseof the function of p63 during epithelial deve-lopment and maintenance. The target genesidentified are involved in numerous cellularpathways further illustrating the complexityand importance of this transcription factor. Therole of p63 in hair follicle morphogenesis is aninteresting aspect and unearthing the targetgenes that may be involved in this process, suchas Runx1, is one step into revealing themolecular mechanisms by which p63 master-fully controls this intricate process. Finally, thesuccess of our improved ChIP strategy inidentifying bona-fide target genes of DNp63suggests its usefulness as a genomic tool forother transcription factors.
ACKNOWLEDGMENTS
We are especially grateful to Dr. Lee AnnGarrett-Sinha and Rose-Anne Romano forhelpful discussion and advice. We also thankKirsten W. Smalley for excellent technicalassistance in mouse husbandry and othermembers of our laboratory for useful commentson this study. We would like to thank NancySpeck for providing the Runx1þ/lz mice. Thiswork was supported by a grant from NIH
R01AR049238 (S.S.) and in part by a grantsupport from United States-Israel BinationalScience Foundation grant 2005197 (U.G. andS.S.).
REFERENCES
Antonini D, Rossi B, Han R, Minichiello A, Di Palma T,Corrado M, Banfi S, Zannini M, Brissette JL, Missero C.2006. An autoregulatory loop directs the tissue-specificexpression of p63 through a long-range evolutionarilyconserved enhancer. Mol Cell Biol 26:3308–3318.
Barbieri CE, Pietenpol JA. 2006. p63 and epithelial biology.Exp Cell Res 312:695–706.
Barbieri CE, Tang LJ, Brown KA, Pietenpol JA. 2006. Lossof p63 leads to increased cell migration and up-regulationof genes involved in invasion and metastasis. Cancer Res66:7589–7597.
Birkaya B, Ortt K, Sinha S. 2007. Novel in vivo targets ofDeltaNp63 in keratinocytes identified by a modifiedchromatin immunoprecipitation approach. BMC Mol Biol8:43.
Blanpain C, Fuchs E. 2007. p63: Revving up epithelialstem-cell potential. Nat Cell Biol 9:731–733.
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. 2006. Differentialroles of p63 isoforms in epidermal development: Selectivegenetic complementation in p63 null mice. Cell DeathDiffer 13:1037–1047.
Carroll DK, Carroll JS, Leong CO, Cheng F, Brown M, MillsAA, Brugge JS, Ellisen LW. 2006. p63 regulates anadhesion programme and cell survival in epithelial cells.Nat Cell Biol 8:551–561.
Ghozi MC, Bernstein Y, Negreanu V, Levanon D, Groner Y.1996. Expression of the human acute myeloid leukemiagene AML1 is regulated by two promoter regions. ProcNatl Acad Sci USA 93:1935–1940.
Kaufman CK, Zhou P, Pasolli HA, Rendl M, Bolotin D, LimKC, Dai X, Alegre ML, Fuchs E. 2003. GATA-3: Anunexpected regulator of cell lineage determination inskin. Genes Dev 17:2108–2122.
Kim TH, Ren B. 2006. Genome-wide analysis of protein-DNA interactions. Annu Rev Genomics Hum Genet 7:81–102.
King KE, Weinberg WC. 2007. p63: Defining roles inmorphogenesis, homeostasis, and neoplasia of the epi-dermis. Mol Carcinog 46:716–724.
Koster MI, Roop DR. 2004. The role of p63 in developmentand differentiation of the epidermis. J Dermatol Sci 34:3–9.
Koster MI, Kim S, Mills AA, DeMayo FJ, Roop DR.2004. p63 is the molecular switch for initiation ofan epithelial stratification program. Genes Dev 18:126–131.
Koster MI, Dai D, Marinari B, Sano Y, Costanzo A, KarinM, Roop DR. 2007. p63 induces key target genes requiredfor epidermal morphogenesis. Proc Natl Acad Sci USA104:3255–3260.
Kramer I, Sigrist M, de Nooij JC, Taniuchi I, Jessell TM,Arber S. 2006. A role for Runx transcription factorsignaling in dorsal root ganglion sensory neuron diversi-fication. Neuron 49:379–393.
1218 Ortt et al.
Levanon D, Groner Y. 2004. Structure and regulatedexpression of mammalian RUNX genes. Oncogene 23:4211–4219.
Marmigere F, Montelius A, Wegner M, Groner Y, ReichardtLF, Ernfors P. 2006. The Runx1/AML1 transcriptionfactor selectively regulates development and survival ofTrkA nociceptive sensory neurons. Nat Neurosci 9:180–187.
Mikhail FM, Sinha KK, Saunthararajah Y, Nucifora G.2006. Normal and transforming functions of RUN X1:Aperspective. J Cell Physiol 207:582–593.
Mikkola ML. 2007. p63 in skin appendage development.Cell Cycle 6:285–290.
Mills AA, Zheng B, Wang XJ, Vogel H, Roop DR, Bradley A.1999. p63 is a p53 homologue required for limb andepidermal morphogenesis. Nature 398:708–713.
Naoe Y, Setoguchi R, Akiyama K, Muroi S, Kuroda M,Hatam F, Littman DR, Taniuchi I. 2007. Repression ofinterleukin-4 in T helper type 1 cells by Runx/Cbf betabinding to the Il4 silencer. J Exp Med 204:1749–1755.
Odom DT, Dowell RD, Jacobsen ES, Gordon W, DanfordTW, MacIsaac KD, Rolfe PA, Conboy CM, Gifford DK,Fraenkel E. 2007. Tissue-specific transcriptional regu-lation has diverged significantly between human andmouse. Nat Genet 39:730–732.
Ortt K, Sinha S. 2006. Derivation of the consensus DNA-binding sequence for p63 reveals unique requirementsthat are distinct from p53. FEBS Lett 580:4544–4550.
Osada M, Park HL, Nagakawa Y, Yamashita K, FomenkovA, Kim MS, Wu B, Nomoto S, Trink B, Sidranksy D. 2005.Differential recognition of response elements determinestarget gene specificity for p53 and p63. Mol Cell Biol 25:6077–6089.
Perez CA, Ott J, Mays DJ, Pietenpol JA. 2007. p63consensus DNA-binding site: Identification, analysisand application into a p63MH algorithm. Oncogene.
Raveh E, Cohen S, Levanon D, Negreanu V, Groner Y, GatU. 2006. Dynamic expression of Runx1 in skin affectshair structure. Mech Dev 123:842–850.
Romano RA, Birkaya B, Sinha S. 2006. Defining theregulatory elements in the proximal promoter ofDeltaNp63 in keratinocytes: Potential roles for Sp1/Sp3,NF-Y, and p63. J Invest Dermatol 126:1469–1479.
Romano RA, Birkaya B, Sinha S. 2007. A functionalenhancer of keratin14 is a direct transcriptional targetof deltaNp63. J Invest Dermatol 127:1175–1186.
Suh EK, Yang A, Kettenbach A, Bamberger C, MichaelisAH, Zhu Z, Elvin JA, Bronson RT, Crum CP, McKeon F.
2006. p63 protects the female germ line during meioticarrest. Nature 444:624–628.
Truong AB, Kretz M, Ridky TW, Kimmel R, Khavari PA.2006. p63 regulates proliferation and differentiation ofdevelopmentally mature keratinocytes. Genes Dev 20:3185–3197.
van Bokhoven H, McKeon F. 2002. Mutations in the p53homolog p63: Allele-specific developmental syndromes inhumans. Trends Mol Med 8:133–139.
Vidal VP, Chaboissier MC, Lutzkendorf S, Cotsarelis G,Mill P, Hui CC, Ortonne N, Ortonne JP, Schedl A. 2005.Sox9 is essential for outer root sheath differentiation andthe formation of the hair stem cell compartment. CurrBiol 15:1340–1351.
Vigano MA, Mantovani R. 2007. Hitting the numbers:The emerging network of p63 targets. Cell Cycle 6:233–239.
Vigano MA, Lamartine J, Testoni B, Merico D, Alotto D,Castagnoli C, Robert A, Candi E, Melino G, Gidrol X,Mantovani R. 2006. New p63 targets in keratinocytesidentified by a genome-wide approach. EMBO J 25:5105–5116.
Wang X, Blagden C, Fan J, Nowak SJ, Taniuchi I, LittmanDR, Burden SJ. 2005. Runx1 prevents wasting, myofi-brillar disorganization, and autophagy of skeletalmuscle. Genes Dev 19:1715–1722.
Yang A, Kaghad M, Wang Y, Gillett E, Fleming MD, DotschV, Andrews NC, Caput D, McKeon F. 1998. p63, a p53homolog at 3q27-29, encodes multiple products withtransactivating, death-inducing, and dominant-negativeactivities. Mol Cell 2:305–316.
Yang A, Schweitzer R, Sun D, Kaghad M, Walker N,Bronson RT, Tabin C, Sharpe A, Caput D, Crum C,McKeon F. 1999. p63 is essential for regenerativeproliferation in limb, craniofacial and epithelial develop-ment. Nature 398:714–718.
Yang A, Kaghad M, Caput D, McKeon F. 2002. On theshoulders of giants: p63, p73 and the rise of p53. TrendsGenet 18:90–95.
Yang A, Zhu Z, Kapranov P, McKeon F, Church GM,Gingeras TR, Struhl K. 2006. Relationships betweenp63 binding, DNA sequence, transcription activity, andbiological function in human cells. Mol Cell 24:593–602.
Zhou G, Zheng Q, Engin F, Munivez E, Chen Y, Sebald E,Krakow D, Lee B. 2006. Dominance of SOX9 functionover RUNX2 during skeletogenesis. Proc Natl Acad SciUSA 103:19004–19009.
DNp63 Regulates Runx1 1219