Developmental Biology 374 (2013) 234–244
Contents lists available at SciVerse ScienceDirect
Developmental Biology
journal homepage: www.elsevier.com/locate/developmentalbiology
Genomes and Developmental Control
Pitx1 broadly associates with limb enhancers and is enriched on hindlimb
cis-regulatory elements
Carlos R. Infante a,1, Sungdae Park a,1, Alexandra G. Mihala a, David M. Kingsley b, Douglas B. Menke a,n
a
b
Department of Genetics, University of Georgia, Coverdell Building, Room 250A, 500 DW Brooks Drive, Athens, GA 30602, USA
Department of Developmental Biology and Howard Hughes Medical Institute, Stanford, University School of Medicine, Stanford, CA 94305, USA
a r t i c l e i n f o
a b s t r a c t
Article history:
Received 17 August 2012
Received in revised form
16 October 2012
Accepted 12 November 2012
Available online 27 November 2012
Extensive functional analyses have demonstrated that the pituitary homeodomain transcription factor
Pitx1 plays a critical role in specifying hindlimb morphology in vertebrates. However, much less is
known regarding the target genes and cis-regulatory elements through which Pitx1 acts. Earlier studies
suggested that the hindlimb transcription factors Tbx4, HoxC10, and HoxC11 might be transcriptional
targets of Pitx1, but definitive evidence for direct regulatory interactions has been lacking. Using ChIPSeq on embryonic mouse hindlimbs, we have pinpointed the genome-wide location of Pitx1 binding
sites during mouse hindlimb development and identified potential gene targets for Pitx1. We
determined that Pitx1 binding is significantly enriched near genes involved in limb morphogenesis,
including Tbx4, HoxC10, and HoxC11. Notably, Pitx1 is bound to the previously identified HLEA and HLEB
hindlimb enhancers of the Tbx4 gene and to a newly identified Tbx2 hindlimb enhancer. Moreover, Pitx1
binding is significantly enriched on hindlimb relative to forelimb-specific cis-regulatory features that
are differentially marked by H3K27ac. However, our analysis revealed that Pitx1 also strongly
associates with many functionally verified limb enhancers that exhibit similar levels of activity in
the embryonic mesenchyme of forelimbs and hindlimbs. We speculate that Pitx1 influences hindlimb
morphology both through the activation of hindlimb-specific enhancers as well as through the
hindlimb-specific modulation of enhancers that are active in both sets of limbs.
& 2012 Elsevier Inc. All rights reserved.
Keywords:
Pitx1
Tbx4
Tbx2
Enhancer
Limb
ChIP-Seq
Mouse
Introduction
The hindlimbs and forelimbs of mice exhibit substantial
differences in the relative size, shape and position of bones,
muscles, and tendons. Despite these differences in morphology,
the majority of limb patterning genes display highly similar
patterns of expression in embryonic forelimbs and hindlimbs
and are therefore not the primary drivers that specify forelimb or
hindlimb-specific morphologies (Duboc and Logan, 2011a). In
fact, relatively little is known regarding the developmental
mechanisms that determine limb-type morphology. Hindlimbspecific defects are observed in mouse knockouts of genes that
encode the Pitx1, Tbx4, Islet1, and HoxC10 transcription factors,
all four of which are predominantly or exclusively expressed in
the hindlimb rather than the forelimb of developing mouse
embryos (Lanctôt et al., 1999; Szeto et al., 1999; Naiche and
Papaioannou, 2003; Hostikka et al., 2009; Kawakami et al., 2011).
n
Corresponding author. Fax: þ1 706 583 0590.
E-mail address: [email protected] (D.B. Menke).
1
These authors contributed equally to this work.
0012-1606/$ - see front matter & 2012 Elsevier Inc. All rights reserved.
http://dx.doi.org/10.1016/j.ydbio.2012.11.017
However, to date only Pitx1 has been definitively shown to
specify hindlimb-type morphology.
Both loss and gain of function experiments demonstrate a role
for Pitx1 in directing the development of hindlimb-specific morphology. The hindlimbs of Pitx1 knockout mice are reduced in size
and exhibit a loss of hindlimb-specific features (Lanctôt et al., 1999;
Szeto et al., 1999). For instance, the ilium and the patella do not
form, and the relative size, position, and shape of hindlimb bones
take on more forelimb-like characteristics. Further evidence of a role
for Pitx1 in specifying hindlimb-type morphology comes from
ectopic expression experiments. Ectopic expression of Pitx1 in the
developing forelimb of mouse or chick embryos is sufficient to
induce the forelimbs to adopt a more hindlimb-like morphology
(Logan and Tabin, 1999; Delaurier et al., 2006). The ectopic expression of Pitx1 in embryonic forelimbs is also sufficient to induce the
expression of other hindlimb transcription factors, including Tbx4,
HoxC10, and HoxC11.
Despite the clear evidence that the Pitx1 transcription factor
promotes the formation of hindlimb-type morphology, our understanding of how Pitx1 regulates hindlimb morphology is limited.
One potential regulatory target of Pitx1 is the Tbx4 gene. Not only
can ectopic Pitx1 induce Tbx4 expression in the embryonic
forelimb, but Pitx1 knockout embryos also show reduced Tbx4
C.R. Infante et al. / Developmental Biology 374 (2013) 234–244
expression in the developing hindlimb (Lanctôt et al., 1999; Szeto
et al., 1999; Delaurier et al., 2006). Recent work has demonstrated
that restoring Tbx4 expression via a Tbx4 transgene can rescue the
ilium loss and hindlimb size reduction observed in Pitx1 knockout
mice, suggesting that Tbx4 may be an important target of Pitx1
function (Ouimette et al., 2010; Duboc and Logan, 2011b). Although
it is not currently known whether Pitx1 directly activates Tbx4
expression, highly conserved putative Pitx1 binding sites have been
identified in the HLEA and HLEB hindlimb enhancers of Tbx4 (Menke
et al., 2008). Regardless of whether Tbx4 is a direct target of Pitx1, it
is clear that many hindlimb features that are lost in Pitx1 null mice,
including formation of the kneecap, are not rescued by restoring
expression of Tbx4. Therefore, it is evident that Pitx1 cannot mediate
its effects solely through the up-regulation of Tbx4, and other
transcriptional targets of Pitx1 must confer additional aspects of
hindlimb morphology.
Here we investigate the in vivo binding sites of Pitx1 in the
developing mouse hindlimb to better understand how Pitx1
regulates hindlimb formation. We perform a global analysis of
Pitx1-bound regions via chromatin immunoprecipitation (ChIP)
and high-throughput sequencing to determine candidate transcriptional targets. We then investigate the function of specific
Pitx1 bound regions that are located in proximity to the Tbx2 and
Tbx4 genes. Finally, we examine whether Pitx1 associates with
known limb enhancers and establish whether Pitx1 is enriched on
hindlimb-specific cis-regulatory elements.
Materials and methods
Chromatin immunoprecipitation
Timed matings were performed with outbred ICR mice (Harlan
Laboratories) and embryonic hindlimb buds were collected at
E11.5. After cross-linking in 1% formaldehyde in PBS for 30 min,
hindlimb buds were rinsed and treated with trypsin for 5 min to
generate a single cell suspension. Samples were then sheared by
sonication to generate a chromatin size range of 200–600 bp.
PureProteomeTM Protein G Magnetic Beads (Millipore) were preincubated with Pitx1 antibody (Santa Cruz Biotechnology, sc18922) before incubating overnight with 400 mg of chromatin.
After washing, immune complexes were eluted from the beads,
and protein-DNA crosslinks were reversed by incubating at 65 1C
overnight. After treatment with RNase followed by proteinase K,
samples were purified with the GeneJETTM PCR Purification Kit
(Fermentas). Two independent biological replicates were used to
generate two Illumina ChIP-Seq and two control libraries. All ChIP
and input chromatin control libraries were produced using the
Illumina ChIP-Seq DNA Sample Prep Kit (IP-102-1001) as directed
by the manufacturer. Single-end libraries were sequenced on an
Illumina GA IIx, producing 36 bp reads (replicate 1), or a HiSeq
2000, producing 50 bp reads (replicate 2), at the HudsonAlpha
Institute for Biotechnology.
Quantitative PCR was used to evaluate the relative abundance
of HLEA and HLEB in input and Pitx1-ChIP DNA. qPCR assays were
performed in triplicate using MaximaTM SYBR Green ROX mix
(Fermentas) on an ABI7500 (Applied Biosystems). An unconserved
intergenic region upstream of HLEA that lacks matches to known
Pitx1 binding motifs was used as a control for normalization. The
2 DDt method was used for calculating enrichments for each target
in ChIP DNA relative to input. The primers used in qPCR were as
follows: HLEA-F: 50 -GAAATGGCGACCCTTGTCTG-30 ; HLEA-R: 50 -TCGAGCTGCAGCTGCAACTC-30 ; HLEB-F: 50 CTTCTGATTCGATCCACATG-30 ;
HLEB-R: 50 -CTGCTTTAGCATTTTCTGTG-30 ; Tbx4control-F: 50 -GATGGTGGCTGATCCTAATG-30 ; Tbx4control-R: 50 -ACGGATAGGATGTGAAGGAG-30 .
235
Peak calling and data analysis
Sequencing read quality was evaluated using FastQC (version
0.5.1, http://www.bioinformatics.bbsrc.ac.uk/projects/fastqc/). A
low quality read filter was then applied in which no reads with
more than six bases with a minimum Phred quality score of
twenty were retained. Sequencing reads from the ChIP and input
control libraries were aligned to the mouse genome (mm9) using
Bowtie v0.12.8 (Langmead et al., 2009) with parameters specified
to report the best alignment allowing no more than three
mismatches within either the first twenty bases (replicate 1) or
28 bases (replicate 2) on the high-quality end of the read and
excluding reads that aligned to more than one location in the
genome (–best -l 20(28) -n 3 -m 1). Pitx1 peaks were identified
using MACS v1.4.2 (Zhang et al., 2008) with default parameters
except for the effective genome size set for mouse ( g mm).
Subpeaks from the MACS output were called using PeakSplitter
(Salmon-Divon et al., 2010). Peaks with a false discovery rate
(FDR) greater than 5% and a fold enrichment less than three as
reported by MACS were discarded. Peaks were then sorted by
p-value and categorized into top 1000, top 5000, or all peaks for
further analysis. These top subsets are based on the observation
that overall MACS peak quality is most effectively evaluated by
ranking by reported p-value (Wilbanks and Facciotti, 2010). For
wiggle plots, sequences were scaled based on the number of
sequence tags per ten million reads. Peaks were annotated to
genes and transcription start sites (TSS) using CEAS v1.0.2 (Shin
et al., 2009) with an annotation table built from UCSC known
genes and the input control wiggle file produced by MACS. De
novo binding motifs were determined using MEME v4.8.1 (Bailey
et al., 2009) with non-default parameters ( maxsize 250,000
mod zoops nmotifs 10 minw 5 maxw 20 revcomp) on
sequences within750 bp of the summits of the top 1000 peaks
based on MACS p-value. The parameters for the MEME analysis
were chosen based on the recommendations of the algorithm’s
authors to balance sensitivity with the execution time of the
analysis for large datasets. De novo binding motifs were
matched to the JASPAR (Sandelin et al., 2004) and UniPROBE
databases (Newburger and Bulyk, 2009) using TOMTOM (Gupta
et al., 2007). Peaks were associated with Gene Ontology (GO)
terms (Ashburner et al., 2000) and Mouse Genome Informatics
(MGI) phenotypes (Eppig et al., 2012) using the Genomic Regions
Enrichment of Annotation Tool (GREAT) (Mclean et al., 2010) with
the whole genome as background. Significance was evaluated
using the reported region-based binomial test. To examine the
evolutionary conservation of Pitx1 binding regions, peak overlap
was compared to regions of conservation between mouse and
human as identified by the ECR Browser (Ovcharenko et al., 2004).
Additional annotation and analysis was performed using BEDTools v2.16.2 (Quinlan and Hall, 2010) and the BSGenome,
GenomicFeatures, rtracklayer, and ChIPpeakAnno packages in R
(http://www.bioconductor.org). Pitx1 ChIP-Seq data has been
deposited in the Gene Expression Omnibus (GEO Accession
Number GSE41591).
Previously published datasets of H3K27ac ChIP-Seq reads from
mouse E11.5 forelimb and hindlimb tissues (Cotney et al., 2012)
were downloaded from the Gene Expression Omnibus (GEO
Accession Numbers GSM875384, GSM875385, GSM875386, and
GSM875387). These paired-end reads were aligned with Bowtie
using the same parameters as the Pitx1 data. Insert size between
the paired-end reads was estimated from the data. This distance
was then used to specify the shift size in MACS in lieu of building
a shifting model ( nomodel shiftsize¼113). These settings
allow MACS to effectively identify the broad peaks characteristic
of histone marks (Feng et al., 2011). Two strategies for identifying
enriched peaks were used; the standard method in which the
236
C.R. Infante et al. / Developmental Biology 374 (2013) 234–244
input library for a tissue was used as the control, and an
alternative method in which the ChIP sample of one tissue was
used as the control for the other and vice versa.
Transgenic mice
Pronuclear injection into FVB or C57BL6/CBA F2 embryos was
performed to generate transient transgenic F0 mice (Stanford
Transgenic Research Facility, Xenogen Biosciences, and Cyagen
Biosciences). Prior to microinjection, BAC and plasmid DNAs were
purified as described (DiLeone et al., 2000). Microinjected embryos
were collected at E11.5 or E12.5 and stained with Xgal as described
(DiLeone et al., 1998). All mouse work was reviewed by the Stanford
University School of Medicine or the University of Georgia IACUC
and was performed under an approved Animal Use Protocol.
Tbx2 BAC modifications
Tbx2 mouse BACs RP24-209K13, RP24-376P4, and RP24-84E15
were modified by insertion of an IRES-bGeo into the 30 UTR of the
Tbx2 gene through the use of a bacterial recombineering system
(Lee et al., 2001). Briefly, 50 and 30 homology arms from the Tbx2
30 UTR were PCR amplified and cloned into pIPTGfTet to produce a
Tbx2 targeting cassette (Chandler et al., 2007). Successfully
targeted BAC clones had an IRES-bGeo and an FRT-flanked
tetracycline resistance gene (TetR) inserted 38 bp downstream
of the Tbx2 stop codon. Transient expression of FLPe recombinase
was used to remove the FRT-flanked TetR selection cassette.
Correctly targeted BAC clones were identified via PCR and were
verified by sequencing the BAC insertion site.
Hsp68 LacZ transgenes
Regions to be tested for enhancer activity were PCR amplified
using primers containing NotI restriction sites and were cloned
into the NotI site of p50 -Not-HspLacZ (DiLeone et al., 1998). In
cases where multiple tandem copies were desired, products were
amplified with primers containing XbaI and SpeI sites, cloned into
a modified pBS KSþ plasmid with two NotI sites flanking an XbaI
site, excised with NotI, and cloned into p50 -Not-HspLacZ.
Results and discussion
Pitx1 binding is enriched near gene promoters
A small number of potential transcriptional targets of Pitx1
have been identified through genetic gain and loss of function
experiments, but whether Pitx1 directly regulates these genes
remains unknown (Lanctôt et al., 1999; Logan and Tabin, 1999;
Szeto et al., 1999; Delaurier et al., 2006). Therefore, we performed
high-throughput sequencing of DNA fragments isolated via Pitx1
ChIP to reveal Pitx1 binding sites located around these potential
regulatory targets and across the mouse genome. Using a highly
specific Pitx1 antibody (Fig. S1), we performed two independent
ChIP experiments on chromatin isolated from E11.5 mouse
hindlimbs. For replicate 1, a total of 36.2 million and 35.2 million
single-end 36 bp reads were generated for the ChIP-Seq and input
control libraries, respectively. After quality filtering, 24.9 million
(71.3%) ChIP reads and 23.6 million (69.5%) input control reads
aligned uniquely to the mouse genome. These uniquely aligned
reads were further processed by MACS for redundancy, leading to
19.4 million treatment and 16.4 million control reads used in
peak calling. For replicate 1, MACS identified a total of 14,753
peaks, 14,734 of which were retained after filtering for further
analysis. For replicate 2, single-end sequencing generated 60.3
million 50 bp reads from the ChIP sample and 55.7 million reads
from the input control sample. After quality filtering, 29.1 million
(48.3%) ChIP reads and 29.8 million (53.5%) input control reads
aligned uniquely. Of these, 28.7 million ChIP reads and 29.4
million input control reads were used for peak calling. For
replicate 2, MACS identified a total of 25,027 peaks and 25,006
were retained after filtering based on FDR and fold enrichment.
To combine the results of the two biological replicates, we
focused further analysis on the intersection of the two peak sets
identified by MACS, our assumption being that ChIP peaks shared
by both replicates would represent the most robust signal
from the total data (Fig. S2). This intersection generated a list of
10,625 significant peaks. The genome-wide distribution of Pitx1
peak locations suggests that Pitx1 binding is enriched within
250 kb of annotated transcriptional start sites (TSS) (Fig. 1A).
Further analysis of regions within 10 kb of a TSS demonstrated
that Pitx1 peaks are significantly enriched in these regions, and
highly enriched ( 44 fold) in potential promoter regions within
the first 1000 bp upstream of TSS (Fig. 1B).
Like other bicoid-related homeodomain transcription factors,
Pitx1 has been shown to bind to the DNA sequence TAATCC and
related motifs (Lamonerie et al., 1996; Berger et al., 2008).
We determined whether this motif or other unrelated motifs
are associated with Pitx1 binding in embryonic mouse hindlimbs
by performing a de novo motif analysis with MEME to detect
sequences that are enriched750 bp from the summit of the top
1000 Pitx1 ChIP-Seq peaks. The most enriched motif closely
matches the in vitro defined Pitx1 binding motif (Fig. 1C).
An examination of the distribution of this motif among the top
10,625 peaks demonstrated that it was dramatically enriched at
the summits of Pitx1 peaks (Figs. 1D and S3). In all, 95.8% of Pitx1
peaks contain a match to this motif with a minimum score
threshold of 90%. Several other over-represented motifs were also
found in association with Pitx1 ChIP-seq peaks, and may represent binding sites for other transcription factors that collaborate
with Pitx1 to regulate gene expression (Fig. S3).
Pitx1 binding highlights functionally important subdomains of HLEA
Tbx4 was first identified as a possible regulatory target of Pitx1
based on gain and loss of function experiments performed in chick
and mouse more than a decade ago (Lanctôt et al., 1999; Logan and
Tabin, 1999; Szeto et al., 1999; Delaurier et al., 2006). More recently,
we identified putative Pitx1 binding sites within two distinct
hindlimb enhancers, HLEA and HLEB, which direct the hindlimb
specific expression of Tbx4 (Menke et al., 2008). These two enhancers are each capable of driving reporter gene expression in the
hindlimb field prior to hindlimb bud outgrowth and continue to
drive expression in the hindlimb late into embryogenesis (Fig. 2A
and B; (Menke et al., 2008)). In order to determine whether Pitx1
associates with either HLEA or HLEB in vivo, we assessed the relative
enrichment of HLEA and HLEB after Pitx1 ChIP by quantitative PCR
and found that Pitx1 associates with both enhancers (Fig. 2C).
However, the enrichment observed for HLEA was substantially
greater than that for HLEB.
We next investigated the functional importance of Pitx1 bound
regions of the HLEA element of Tbx4 using our high-resolution
ChIP-Seq data and DNA sequence conservation as our starting
point. A VISTA alignment (Mayor et al., 2000) of mouse HLEA
against the orthologous human sequence identified three distinct
regions of conservation across HLEA’s span (Fig. 3A). These
three regions are well-conserved in all mammalian genomes that
we have examined (30 placental mammals, 2 marsupials,
and a monotreme). Of the three regions, however, only the third
exhibits conservation beyond mammals with clear sequence
conservation to birds and crocodilians (data not shown). The
C.R. Infante et al. / Developmental Biology 374 (2013) 234–244
237
Fig. 1. Pitx1 binding is enriched near promoter regions and is associated with the canonical TAATCC bicoid-type binding motif. (A) Distribution of Pitx1 ChIP-seq peaks
relative to transcription start sites compared to 1000 replicates of randomized genomic regions. (B) Percentage of Pitx1 peaks found in promoter regions, regions
downstream of genes, or within the exons/introns of genes. (C) Comparison of the in vitro determined Pitx1 binding motif (Pitx1 2312.1) to the top de novo motif enriched
within 7 50 bp of Pitx1 ChIP-seq peak summits and (D) Distribution of de novo motif among all Pitx1 ChIP-seq peak summits.
HLEB
140
Relative Enrichment
HLEA
100
60
6
3
HLEA
HLEB
Fig. 2. Pitx1 associates with the HLEA and HLEB hindlimb enhancers of the Tbx4 gene. Whole-mount LacZ stained E11.5 mouse embryos carrying HLEA-hsp68LacZ
(A) or HLEB-hsp68LacZ (B) transgenes. (C) Quantitative PCR analysis reveals the relative enrichment of HLEA and HLEB after Pitx1 ChIP. Values presented are the mean and
standard error for eight (HLEA) or three (HLEB) independent ChIP experiments.
238
C.R. Infante et al. / Developmental Biology 374 (2013) 234–244
chr11
85,689,600
85,689,800
85,690,000
85,690,200
85,690,400
85,690,600
100%
Mouse:
Human
Region 1
Region 2
Region 3
50%
25
Pitx1
Chip
1
HLEA
R1+R2
R2+R3
R1+R3
R1+mR2
R1, 1x
R1, 4x
R2, 4x
R3, 4x
Fig. 3. Pitx1 binding highlights functionally important subdomains of HLEA. (A) Conservation plot comparing the sequence identity between mouse and human HLEA reveals three
regions of conservation. Aligned regions of at least 100 bp and Z70% sequence identity are shaded in pink. (B) Pitx1 ChIP-seq signal across HLEA. The y-axis indicates the number of
sequence tags per 10 million reads. The position and orientation of sequences that match Pitx1 binding motifs are denoted by blue and red arrowheads. (C) Schematic of HLEA
subregions tested for enhancer activity via Hsp68LacZ transgenes and (D)–(K) Representative whole-mount LacZ stained mouse embryos carrying different HLEA transgenes.
Forelimbs and hindlimbs are outlined in white.
three regions of conservation correspond to three distinct ChIPSeq subpeaks within a larger HLEA Pitx1 ChIP-Seq peak (Fig. 3B).
While there are 8 sites in mouse HLEA that match known Pitx1
binding motifs, only two of these sites lie directly under prominent Pitx1 summits, suggesting that only a subset of the potential
Pitx1 binding sites are bound by Pitx1 in vivo.
C.R. Infante et al. / Developmental Biology 374 (2013) 234–244
239
Pitx1 binding highlights the location of a Tbx2 hindlimb enhancer
Since our Pitx1 ChIP-Seq analysis indicated that Pitx1 ChIP
signal is greater over the first two conserved regions of HLEA than
the third conserved region, we tested the functional importance
of these three sub-domains (R1, R2, and R3) for enhancer activity
in vivo (Fig. 3C and K, Table 1, and Fig. S4). We deleted R3 from the
original HLEA-Hsp68LacZ transgene (Fig. 3C, transgene R1þR2)
and found that this transgene drives levels of LacZ expression
comparable to the full length HLEA (Fig. 3D and E). Embryos
carrying the HLEA or R1þR2 transgene both display strong
staining in the proximal and central regions of the hindlimbs
with reproducible, but much lower, levels of activity in the
forelimb. In contrast, deletion of R1 results in a complete or
nearly complete loss of enhancer activity, and removal of R2
dramatically reduces enhancer activity (Fig. 3F and G). Thus, Pitx1
ChIP-Seq signal overlaps functionally important regions (R1 and
R2) of HLEA. The R3 domain, which shows less Pitx1 ChIP-Seq
signal, appears to be of little or no importance for generating limb
enhancer activity at E11.5 in the context of the Hsp68LacZ
transgenes that we tested.
Our deletion analysis demonstrates that R1 is the most significant region of HLEA for conferring hindlimb enhancer activity
at E11.5. We therefore tested the ability of R1 to drive enhancer
activity in the absence of both R2 and R3. Production of R1
transgenic embryos established that this region is sufficient to
drive modest levels of hindlimb expression (Fig. 3I). Comparison of
the R1 transgene against the R1þR2 and R1þR3 transgenes
indicates that R2 acts to boost the enhancer activity of R1, while
R3 has little or no impact on R1 enhancer activity (Fig. 3E, G, and
I). As demonstrated previously, mutation of a highly conserved
Pitx1 motif, located directly underneath the Pitx1 ChIP-Seq summit of R2, strongly diminished the ability of R2 to stimulate R1
enhancer activity, suggesting that Pitx1 binding is important for
robust activation of this enhancer (Fig. 3E and H; (Menke et al.,
2008)).
While R2 acts jointly with R1 to increase enhancer activity, R2
alone has little or no effect despite the fact that our ChIP-Seq
analysis indicates that both regions are bound by Pitx1 (Fig. 3B
and F). To explore functional differences between R1 and R2, we
created a transgene construct that contained four tandem copies
of R1 in front the Hsp68LacZ cassette. This allowed us to determine whether multiple copies of R1 could increase enhancer
activity and functionally substitute for the presence of R2.
Including these multiple copies of R1 markedly increased enhancer activity (Fig. 3J). However, the hindlimb specificity of R1 was
lost as increased enhancer activity occurred in both the forelimb
and the hindlimb. Unlike R1, four tandem copies of either R2 or
R3 produced no detectable limb expression (Fig. 3K and data not
shown). Taken together, our results are consistent with R1
functioning to produce the basic pattern of hindlimb enhancer
activity and R2 acting to strengthen this activity.
In order to further investigate the relationship between Pitx1
binding sites and Tbx4 regulation, we examined our ChIP-Seq data
across the Tbx4 locus. In agreement with our Pitx1 ChIP qPCR
experiments, our ChIP-Seq data revealed a peak at HLEA, and a
small, but significant, peak at HLEB (Fig. 4A and B). In addition, we
detected strong peaks (in top 1000 category and fold enrichment
greater than 10) just downstream of Tbx4 and in a distant
upstream region located within an intron of the Bcas3 gene
(Fig. 4B). These additional peaks are located in well-conserved
non-coding regions and represent potential cis-regulatory elements for Tbx4 or its neighboring genes, Bcas3 and Tbx2.
We tested the strong Pitx1 ChIP-Seq peak located just 30 of
the Tbx4 locus for enhancer activity (Fig. 4B). The underlying
sequence associated with this Pitx1 peak is present in Tbx4 BAC
transgenes capable of driving hindlimb expression, but this small
sub-region had not been tested previously for enhancer activity
(Menke et al., 2008). To test this region specifically, we cloned
6.9 kb of sequence encompassing this Pitx1 peak and flanking
conserved sequences and inserted it in front of the Hsp68LacZ
cassette. We found that, unlike HLEA and HLEB, this region is not
sufficient to drive hindlimb enhancer activity in E11.5 transgenic
embryos (data not shown).
We decided next to explore the regulatory landscape around
the Pitx1 ChIP-Seq peaks that are located within a Bcas3 intron
(Fig. 4B). The strongest of these intronic peaks is located over
360 kb away from the promoter region of Bcas3 and is in much
closer proximity to the promoters of Tbx2 and Tbx4 (117 kb and
170 kb, respectively). Additionally, since Bcas3 does not exhibit
substantial expression in the limb buds at E11.5 and previous
regulatory analyses of Tbx4 BAC transgenes indicated that this
region is not required for robust Tbx4 hindlimb expression, we
hypothesized that Pitx1 binding events in this Bcas3 intron might
instead be associated with the regulation of Tbx2 (Siva and
Inamdar, 2006; Menke et al., 2008). The Tbx2 gene is expressed
along the anterior and posterior margins of the forelimbs and
hindlimbs (Chapman et al., 1996). Thus, this raises the intriguing
possibility that Pitx1 might modulate the hindlimb expression of
Tbx2, a limb patterning gene that is not hindlimb specific.
We evaluated the ability of three overlapping Tbx2 mouse
BACs to drive limb expression. Each of these BACs contained the
entire Tbx2 open reading frame with the ends of the BACs
staggered across the locus to test the regulatory importance of
different flanking regions (Fig. 4C). An IRES-bGeo reporter was
inserted into the 30 UTR of Tbx2 in each BAC clone via recombineering, and transient transgenic mouse embryos were generated
and stained with X-gal to detect b-galactosidase activity. All three
BACs drove common regions of Tbx2 expression in domains where
the native Tbx2 gene is expressed, including the dorsal eye and
Table 1
Relative enhancer activity of HLEA constructs in LacZ Positive F0 transgenic embryos.
Construct
No. embryos
Forelimb LacZ stain
þþþ
HLEA
R1 þR2
R1 þmR2
R2 þR3
R1 þR3
R1, 1x
R1, 4x
R2, 4x
R3, 4x
14
16
11
10
17
14
12
6
10
1/14
1/16
0/11
0/10
2/17
0/14
6/12
0/6
0/10
Hindlimb LacZ Stain
þþ
(7%)
(6%)
(0%)
(0%)
(12%)
(0%)
(50%)
(0%)
(0%)
5/14
5/16
4/11
1/10
1/17
0/14
2/12
0/6
0/10
þ
(36%)
(31%)
(36%)
(10%)
(6%)
(0%)
(17%)
(0%)
(0%)
4/14
6/16
3/11
5/10
5/17
7/14
0/12
1/6
2/10
(29%)
(38%)
(27%)
(50%)
(29%)
(50%)
(0%)
(17%)
(20%)
No expression
þþþ
4/14
4/16
4/11
4/10
9/17
7/14
4/12
5/6
8/10
7/14
7/16
1/11
0/10
2/17
0/14
5/12
0/6
0/10
(29%)
(25%)
(36%)
(40%)
(53%)
(50%)
(33%)
(83%)
(80%)
þþ
(50%)
(44%)
(9%)
(0%)
(12%)
(0%)
(42%)
(0%)
(0%)
2/16
4/16
3/11
1/10
1/17
1/14
2/12
0/6
0/10
No expression
þ
(14%)
(25%)
(27%)
(10%)
(6%)
(7%)
(17%)
(0%)
(0%)
5/14
4/16
4/11
4/10
13/17
9/14
1/12
1/6
1/10
(36%)
(25%)
(36%)
(40%)
(76%)
(64%)
(8%)
(17%)
(10%)
0/14
1/16
3/11
5/10
1/17
4/14
4/12
5/6
9/10
(0%)
(6%)
(27%)
(50%)
(6%)
(29%)
(33%)
(83%)
(90%)
240
C.R. Infante et al. / Developmental Biology 374 (2013) 234–244
Fig. 4. Pitx1 associates with limb cis-regulatory domains of the Tbx2 and Tbx4 genes. (A) Schematic of the genomic region surrounding Tbx2 and Tbx4. Regions of
conservation Z 100 bp with Z70% identity between mouse-human, mouse-chicken, and mouse-zebrafish are highlighted by colored peaks (blue ¼exons, pink ¼conserved
non-coding regions). The Tbx4 HLEA and HLEB elements are highlighted in orange, and the putative Tbx2 limb enhancer region is shaded green. (B) Pitx1 ChIP-Seq and
input control signal. An asterisk denotes the location of a Pitx1 peak just downstream of the Tbx4 open reading frame that was tested for enhancer activity. The y-axes of
Pitx1 ChIP and input control plots are scaled to indicate the number of sequence tags per ten million reads. (C) Genomic regions covered by mouse Tbx2 BAC and
Hsp68LacZ transgenes. Bars indicate the location of sequences contained within each transgene. Blue boxes within BAC transgenes indicate the insertion site of an IRESLacZ cassette. Yellow bar indicates the position of the transgene containing a putative Tbx2 hindlimb enhancer, Red bar indicates the transgene containing a putative Tbx2
forelimb/hindlimb enhancer and (D)–(I) Representative whole-mount LacZ-stained mouse embryos carrying Tbx2 BAC transgenes (D), (E), and (F) and Hsp68LacZ
transgenes (G), (H), and (I). Forelimbs and hindlimbs are outlined in white.
nasal region (Fig. 4D and F; (Chapman et al., 1996)). In contrast,
only the most proximal BAC, RP24-209K13, was capable of driving
limb expression in a Tbx2-like pattern with strong expression
observed in the posterior portion of both forelimbs and hindlimbs
(Fig. 4D). Tbx2-like limb expression was never observed in the
forelimbs or hindlimbs of transgenic embryos carrying the medial
or distal BACs (Fig. 4E and F; 0 out of 5 and 0 out of 6 transient
transgenic embryos, respectively). However, Tbx4-like hindlimb
expression was sometimes observed with both the medial and
distal BAC transgenes (Fig. S5 and Table S1). This hindlimbrestricted expression was likely a result of the presence of the
native HLEA and HLEB Tbx4 enhancers, which are contained
within the medial and distal BACs. We hypothesize that in the
context of a concatenated BAC transgene insertion, these Tbx4
hindlimb enhancers may, in some instances, be able to interact
with the promoter of Tbx2 and stimulate hindlimb-specific
expression.
The Tbx2 BAC scan implicated a 59 kb candidate limb enhancer
region that includes the strongest of the Bcas3 intronic Pitx1
peaks (Fig. 4). We designed five different Hsp68LacZ transgenes
from this region that together cover 498% of the sequence in this
interval that is conserved between the human and mouse
C.R. Infante et al. / Developmental Biology 374 (2013) 234–244
genomes (Fig. 4C). When tested in vivo, we found that two of the
five transgenes clearly exhibit limb enhancer activity in a Tbx2like pattern (Fig. 4G and I, Table S2). Remarkably, the subregion
that overlaps the Pitx1 peak reproducibly acts as a hindlimb
enhancer with posteriorly restricted expression (Fig. 4H). Notably,
an adjacent subregion that lacks significant Pitx1 peaks drives
posterior limb expression in both forelimbs and hindlimbs
(Fig. 4I). Thus, the 59 kb limb interval contains at least two
putative Tbx2 limb enhancers: one characterized by Pitx1 binding
and a high level of hindlimb specificity and a second which
functions in both forelimbs and hindlimbs and lacks detectable
Pitx1 binding. Our results suggest that Tbx2, a limb patterning
gene that has similar expression domains in both sets of limbs,
has at least partially partitioned hindlimb enhancer activity into a
specialized cis-regulatory domain.
Pitx1 preferentially binds to hindlimb cis-regulatory elements
As a hindlimb-restricted transcription factor, Pitx1 could
potentially regulate hindlimb gene expression patterns in a
number of different ways. For instance, Pitx1 could influence
target gene expression by interacting with hindlimb-specific
enhancer elements. A second possibility is that Pitx1 could act
in the hindlimb to repress the activity of forelimb-specific
enhancers, resulting in forelimb/hindlimb expression differences.
Alternatively, enhancer elements that are activated in both sets of
limbs could be bound by Pitx1 in the hindlimb, resulting in a
hindlimb-specific modulation of enhancer activity. These possible
mechanisms of Pitx1 action are not mutually exclusive. Consistent with the first scenario, the Pitx1 ChIP-Seq peaks around the
Tbx2/Tbx4 locus strongly correlate with the location of cis-regulatory elements with hindlimb-specific enhancer activity
(Fig. 2A and B, Fig. 4A C and H). Specifically, both of the two
known Tbx4 hindlimb enhancers overlap Pitx1 peaks, as does the
putative Tbx2 hindlimb enhancer. In contrast, other enhancers in
the region, including an intronic Tbx4 lung enhancer and the
putative Tbx2 enhancer that drives expression in the forelimbs
and hindlimbs, do not overlap Pitx1 peaks (Fig. 4 and (Menke
et al., 2008)). Moreover, when we examined the genomic region
around HoxC10 and HoxC11, two genes from the HoxC cluster that
exhibit hindlimb-restricted expression, we found a significant
Pitx1 peak in the intergenic interval between these two genes
(Fig. 5). This peak falls within a 10 kb region that has been shown
241
to contain regulatory sequences sufficient to drive hindlimbrestricted expression in transgenic mouse embryos (Papenbrock
et al., 2000).
To examine the relationship between limb-type specificity of
cis-regulatory elements and Pitx1 binding sites more comprehensively, we took advantage of recently published H3K27ac ChIPSeq data sets generated separately from E11.5 forelimbs and
hindlimbs (Cotney et al., 2012). The histone H3K27ac modification has been shown to mark active enhancers and promoters,
and ChIP-Seq against this histone mark is an effective means to
identify cis-regulatory elements that are active in a particular cell
type or tissue (Creyghton et al., 2010; Ernst et al., 2011; Zentner
et al., 2011). Cotney et al. (2012) demonstrated that the majority
of H3K27ac peaks are shared between embryonic forelimbs and
hindlimbs and are not limb-type specific, indicating that many of
the same cis-regulatory elements are active in both sets of limbs.
We examined the overlap of Pitx1 peaks with the location of
H3K27ac marks that are shared between forelimbs and hindlimbs,
or H3K27ac marks which are uniquely called as significant in only
the forelimbs or only the hindlimbs. To ensure consistency
between Pitx1 and H3K27ac peak-calling, we reanalyzed the
Cotney et al. H3K27ac ChIP-seq data using MACs v1.4.2 (Fig. S6;
see materials and methods). We found that Pitx1 peaks overlap a
greater proportion of hindlimb-specific elements than forelimbspecific elements (Fig. 6A; 19% vs. 7.2%, Fisher’s exact test,
p¼3.44 10 59).
Our initial comparison of Pitx1 peaks against those H3K27ac
marks that are uniquely found in only one limb type or the other
did not include cis-regulatory elements that are marked by
H3K27ac in both limb-types but which are marked more intensely in either the forelimbs or the hindlimbs. Previous work has
demonstrated that the strength of H3K27ac marks correlate with
gene transcription, with stronger H3K27ac signals associated with
greater levels of gene expression (Cotney et al., 2012). Consequently, H3K27ac marks that exhibit differences in strength
between forelimbs and hindlimbs may highlight enhancers that
are preferentially active in one set of limbs over the other. We
therefore further analyzed the Cotney et al. H3K27ac data with
MACS in order to identify H3K27ac marks that differ significantly
in signal strength between the forelimb and hindlimb data sets,
but which are not necessarily uniquely present in only one limb-type.
Our analysis identified 364 forelimb-biased and 648 hindlimb-biased
H3K27ac marked regions. When the Pitx1 peaks were compared
Fig. 5. A Pitx1 ChIP-Seq peak is associated with a HoxC10/11 hindlimb regulatory domain. (A) Conservation plot comparing the mouse and human HoxC gene clusters.
Regions of conservation Z100 bp with Z 70% identity between mouse-human and mouse-chicken are highlighted by colored peaks (blue ¼ exons, pink ¼conserved noncoding regions). (B) Pitx1 ChIP-Seq and input control signal. The y-axes of Pitx1 ChIP and input control plots are scaled to indicate the number of sequence tags per ten
million reads. A Pitx1 peak located in the intergenic region between HoxC10 and HoxC11 falls within a 10 kb region (orange shading) that has been previously
demonstrated to drive hindlimb-specific expression (Papenbrock et al., 2000).
242
C.R. Infante et al. / Developmental Biology 374 (2013) 234–244
histone acetyltransferase at E11.5 (Visel et al., 2009). The Pitx1
data showed significantly greater overlap with the P300 limb
enhancers (54%) compared to the forebrain (13%) or midbrain
Pitx1
6,373
80
Vista Positive
1,090
2,925
13,641
3,054
4,571
Forelimb
H3K27ac
Hindlimb
H3K27ac
Pitx1
Percentage of enhancers
237
P300
60
40
20
10,366
346
18
241
407
0
Limb
Heart
Forebrain
Midbrain
Hindbrain
-log (Binomial p-value)
0
Forelimb-biased
H3K27ac
Hindlimb-biased
H3K27ac
Fig. 6. Pitx1 binding is enriched on hindlimb cis-regulatory elements.
(A) Comparison of overlap between Pitx1 ChIP-Seq peaks from E11.5 hindlimbs
and H3K27ac ChIP-Seq peaks from E11.5 forelimbs and E11.5 hindlimbs (Cotney
et al., 2012). (B) Comparison of overlap between embryonic hindlimb Pitx1 ChIPSeq peaks and forelimb-biased and hindlimb-biased H3K27ac marks.
20
40
60
80
100
120
skeletal system development
limb development
limb morphogenesis
skeletal system morphogenesis
embryonic limb morphogenesis
against our stringent set of forelimb-biased and hindlimb-biased
H3K27ac marks, we found that Pitx1 overlapped the hindlimbbiased marks 7.5 more frequently than forelimb-biased marks
(Fig. 6B, Fisher’s exact test, p¼1.12 10 34). Moreover, Pitx1 peaks
are over-represented on hindlimb-biased H3K27ac regions and
under-represented on forelimb-biased H3K27ac relative to forelimb/
hindlimb common H3K27ac regions (18% of all forelimb/hindlimb
common H3K27ac regions overlap with Pitx1 ChIP-seq peaks
compared to 37% of hindlimb-biased H3K27ac regions and 5% of
forelimb-biased peaks). These results suggest that Pitx1 may influence target gene expression more often by interacting with hindlimb
enhancer elements or through the hindlimb-specific modulation of
elements that are activated in both sets of limbs, rather than by acting
to repress the activity of forelimb-specific enhancers in the hindlimb.
Pitx1 broadly associates with limb enhancers
cartilage development
embryonic organ morphogenesis
embryonic skeletal system development
embryonic skeletal system morphogenesis
regulation of mesenchymal cell proliferation
-log (Binomial p-value)
0
10
20
30
40
50
60
70
80
90
100 110
abnormal cartilage morphology
abnormal thoracic cage morphology
abnormal autopod morphology
abnormal joint morphology
We observed that approximately 86% of Pitx1 peaks overlap
non-coding regions that are well-conserved between mouse and
human, suggesting that many of these regions may be functionally important. Since many of these regions are likely to serve as
cis-regulatory elements, we investigated whether Pitx1 associates
with limb enhancers in the developing hindlimb. For this analysis,
we first used the VISTA Enhancer Browser to identify previously
characterized enhancers that drive expression specifically in
either E11.5 limb buds (n ¼111), the forebrain (n ¼142), the
midbrain (n¼ 101), the hindbrain (n ¼74), or the heart (n ¼69)
(Visel et al., 2007). Comparison of these VISTA positive enhancers
with Pitx1 ChIP-Seq peaks demonstrated significant enrichment
in the limb-specific dataset (Fisher’s exact test, p ¼4.02 10 40).
More than 67% of these known limb enhancers overlap a Pitx1
peak, compared to between approximately 1 to 9% of brain or
heart specific enhancers (Fig. 7A). We also tested the overlap
between our Pitx1 ChIP-Seq peaks and the locations of putative
limb, forebrain, or midbrain enhancers marked by the P300
abnormal vertebrae morphology
abnormal limb development
abnormal digit development
abnormal forelimb morphology
abnormal neurocranium morphology
abnormal limb/digits/tail development
Fig. 7. Pitx1 broadly associates with limb-specific enhancers and genes involved in
limb and skeletal development. (A) Percent of known limb, heart, forebrain, midbrain,
and hindbrain specific enhancers from the VISTA Enhancer Browser database (http://
enhancer.lbl.gov/) that overlap with Pitx1 ChIP-Seq peaks. Also shown are the
percentages of P300 ChIP-Seq peaks from E11.5 mouse limb, forebrain, and midbrain
(Visel et al., 2009) that overlap with Pitx1 peaks. (B) and (C) Results from GREAT
analysis. (B) Top 10 enriched terms from the GO Biological Processes database
associated with Pitx1 ChIP-Seq peaks and (C) Top 10 enriched terms from the Mouse
Phenotype database associated with Pitx1 ChIP-Seq peaks.
C.R. Infante et al. / Developmental Biology 374 (2013) 234–244
(11%) datasets (Fisher’s exact test, p ¼3.32 10 222; Fig. 6A). We
conclude that Pitx1 associates with many limb cis-regulatory
elements during hindlimb bud development.
Pitx1 binding is enriched near limb genes
To investigate genes that may be regulated by Pitx1, we compiled
a list of all genes that have a Pitx1 ChIP-Seq summit within 250 kb of
their TSS. This list of 4859 Pitx1-associated genes was compared to
lists of genes with forelimb-biased (58 genes) or hindlimb-biased
expression (127 genes) at E10.5 identified by RNA sequencing
(Cotney et al., 2012). We found no significant difference in the
number of Pitx1-associated genes that were in common with either
the forelimb or hindlimb-biased expression gene lists (Fisher’s exact
test, p¼0.5047). Of the genes with forelimb-biased expression, 37.9%
have a Pitx1 peak within 250 kb, compared to 32.3% of hindlimbbiased genes. We also compared our Pitx1-associated gene list to
datasets of forelimb and hindlimb expressed genes generated from
microarray data (Taher et al., 2011). In this dataset, 23 of 76 (30.3%)
forelimb up-regulated genes overlap Pitx1 associated genes, compared to 7 of 11 (63.6%) hindlimb up-regulated genes. There was a
higher percentage overlap with hindlimb up-regulated genes and this
association was marginally significant (Fisher’s exact test, p¼0.0423).
We next investigated whether Pitx1 binding site locations are
associated with particular developmental functions, pathways, or
gene classes using GREAT (Mclean et al., 2010). This analysis
demonstrated that Pitx1 peaks are strongly associated with genes
implicated in limb and skeletal development (Fig. 7B and C; Fig. S7).
For instance, 8 of the top 10 enriched terms under GO Biological
Process relate to gene groups involved in limb or skeletal development (Fig. 7B; p-values ranging from 10 59 to 10 134), and 6 of the
top 10 terms under Mouse Phenotype relate specifically to genes
required for limb development (Fig. 7C; p-values ranging from
10 76 to 10 117). Similar term associations are reported for all
10,625 peaks, the top 5000 peaks and the top 1000 peaks based on
p-value (Fig. S7). Therefore, Pitx1 likely regulates hindlimb morphology by acting upon many different transcriptional targets that
are involved in limb patterning and skeletal development.
243
has a relatively modest influence on the expression of many of its
target genes, since genes with subtle forelimb/hindlimb expression
differences are largely excluded from published forelimb and hindlimb gene lists, which focus on genes with large forelimb/hindlimb
expression differences (Taher et al., 2011; Cotney et al., 2012).
Alternatively, there is some evidence that Pitx1 can act as an
activator or repressor depending on its context (Island et al., 2002).
In this scenario, Pitx1 binding sites could act to reduce or increase
the expression of target genes in the hindlimb. A third possibility is
that only a small fraction of Pitx1 peaks are biologically relevant and
the influence of this subset of peaks on hindlimb gene expression is
obscured by binding sites that have little impact on gene transcription. Finally, it is possible that Pitx1 may exert stronger influences on
hindlimb expression at earlier developmental stages. ChIP-Seq analyses at additional stages of development and expression analyses of
wild-type and Pitx1 knockout mice will be critical in determining
which Pitx1 binding sites exert the most significant influence on
target gene expression during formation of the hindlimb.
In contrast to the absence of significant enrichment of Pitx1
binding near genes with hindlimb-biased expression, Pitx1 binding
was strongly enriched on putative cis-regulatory elements that are
marked by H3K27ac more intensely in the hindlimb than the
forelimb. These elements represent intriguing candidate regulatory
targets for Pitx1, and the influence of Pitx1 on the activity of these
elements requires further investigation. Our identification of a Pitx1bound hindlimb enhancer for Tbx2, a gene that exhibits similar
expression patterns in the forelimb and hindlimb, is consistent with
the hypothesis that Pitx1 promotes hindlimb morphology by
modulating the hindlimb expression of important limb patterning
genes. In a broader context, across the genome it is apparent that
Pitx1 is associating with a large number of limb cis-regulatory
elements. Even if Pitx1 only modulates the expression of a fraction
of these, there are likely many regulatory targets beyond Tbx4 and
HoxC10/11. The large number of limb genes with neighboring Pitx1
ChIP-Seq peaks suggests that Pitx1 influences hindlimb morphology
broadly and does not just act through a handful of target genes.
Acknowledgements
Conclusions
Our ChIP-Seq analysis of in vivo Pitx1 binding in E11.5 mouse
hindlimbs has helped to illuminate its role in the transcriptional
regulation of previously identified putative Pitx1 target genes.
Specifically, we have established that Tbx4 is, in fact, directly
regulated by Pitx1, and we have demonstrated that Pitx1 associated
regions highlight functionally important subdomains of the HLEA
element of Tbx4. A number of interesting potential targets have also
been identified that warrant further investigation. For instance, the
discrete HoxC10/11 binding site points to a role for Pitx1 in the
regulation of these posterior HoxC cluster genes. This is consistent
with gain of function studies that have demonstrated that ectopic
expression of Pitx1 in the embryonic mouse forelimb can induce
HoxC10 expression and ectopic expression in embryonic chick
forelimbs can induce the expression of HoxC10 and HoxC11 (Logan
and Tabin, 1999; Delaurier et al., 2006). Additionally, our analysis of
Pitx1 binding sites has led to our identification of a putative Tbx2
hindlimb enhancer. Further analyses will be needed to investigate
the relative importance of the different Pitx1 binding sites found
within these hindlimb regulatory elements.
Although we identified Pitx1 binding sites in the vicinity of genes
with hindlimb-biased gene expression, including Tbx4, HoxC10, and
HoxC11, in our global analysis Pitx1 ChIP-Seq peaks did not display
clear enrichment near genes that exhibit expression levels that are
greater in hindlimbs than forelimbs. This could be explained if Pitx1
This work was supported by start-up funds from the University of
Georgia and a National Science Foundation grant (#1149453)
awarded to D.B.M. This study was also supported in part by resources
and technical expertise from the University of Georgia Georgia
Advanced Computing Resource Center, a partnership between the
Office of the Vice President for Research and the Office of the Vice
President for Information Technology. Additional support to D.M.K.
was provided by National Institutes of Health grant P50 HG02568.
D.M.K. is an investigator of the Howard Hughes Medical Institute.
Appendix A. Supporting information
Supplementary data associated with this article can be found in
the online version at http://dx.doi.org/10.1016/j.ydbio.2012.11.017.
References
Ashburner, M., Ball, C.A., Blake, J.A., Botstein, D., Butler, H., Cherry, J.M., Davis, A.P.,
Dolinski, K., Dwight, S.S., Eppig, J.T., Harris, M.A., Hill, D.P., Issel-Tarver, L.,
Kasarskis, A., Lewis, S., Matese, J.C., Richardson, J.E., Ringwald, M., Rubin, G.M.,
Sherlock, G., 2000. Gene ontology: tool for the unification of biology. The Gene
Ontology Consortium. Nat. Genet. 25, 25–29.
Bailey, T.L., Boden, M., Buske, F.A., Frith, M., Grant, C.E., Clementi, L., Ren, J.,
Li, W.W., Noble, W.S., 2009. MEME SUITE: tools for motif discovery and
searching. Nucleic Acids Res. 37, W202–8.
244
C.R. Infante et al. / Developmental Biology 374 (2013) 234–244
Berger, M.F., Badis, G., Gehrke, A.R., Talukder, S., Philippakis, A.A., Peña-Castillo, L.,
Alleyne, T.M., Mnaimneh, S., Botvinnik, O.B., Chan, E.T., Khalid, F., Zhang, W.,
Newburger, D., Jaeger, S.A., Morris, Q.D., Bulyk, M.L., Hughes, T.R., 2008.
Variation in homeodomain DNA binding revealed by high-resolution analysis
of sequence preferences. Cell 133, 1266–1276.
Chandler, R.L., Chandler, K.J., McFarland, K.A., Mortlock, D.P., 2007. Bmp2 transcription in osteoblast progenitors is regulated by a distant 30 enhancer
located 156.3 kilobases from the promoter. Mol. Cell. Biol. 27, 2934–2951.
Chapman, D.L., Garvey, N., Hancock, S., Alexiou, M., Agulnik, S.I., Gibson-Brown, J.J.,
Cebra-Thomas, J., Bollag, R.J., Silver, L.M., Papaioannou, V.E., 1996. Expression
of the T-box family genes, Tbx1–Tbx5, during early mouse development. Dev.
Dyn. 206, 379–390.
Cotney, J., Leng, J., Oh, S., DeMare, L.E., Reilly, S.K., Gerstein, M.B., Noonan, J.P.,
2012. Chromatin state signatures associated with tissue-specific gene expression and enhancer activity in the embryonic limb. Genome Res. 22,
1069–1080.
Creyghton, M.P., Cheng, A.W., Welstead, G.G., Kooistra, T., Carey, B.W., Steine, E.J.,
Hanna, J., Lodato, M.A., Frampton, G.M., Sharp, P.A., Boyer, L.A., Young, R.A.,
Jaenisch, R., 2010. Histone H3K27ac separates active from poised enhancers
and predicts developmental state. Proc. Nat Acad. Sci. U.S.A. 107,
21931–21936.
Delaurier, A., Schweitzer, R., Logan, M.P.O., 2006. Pitx1 determines the morphology
of muscle, tendon, and bones of the hindlimb. Dev. Biol. 299, 22–34.
DiLeone, R.J., Marcus, G.A., Johnson, M.D., Kingsley, D.M., 2000. Efficient studies of
long-distance Bmp5 gene regulation using bacterial artificial chromosomes.
Proc. Nat. Acad. Sci. U.S.A. 97, 1612–1617.
DiLeone, R.J., Russell, L.B., Kingsley, D.M., 1998. An extensive 30 regulatory region
controls expression of Bmp5 in specific anatomical structures of the mouse
embryo. Genetics 148, 401–408.
Duboc, V., Logan, M.P.O., 2011a. Regulation of limb bud initiation and limb-type
morphology. Dev. Dyn. 240, 1017–1027.
Duboc, V., Logan, M.P.O., 2011b. Pitx1 is necessary for normal initiation of
hindlimb outgrowth through regulation of Tbx4 expression and shapes
hindlimb morphologies via targeted growth control. Development 138,
5301–5309.
Eppig, J.T., Blake, J.A., Bult, C.J., Kadin, J.A., Richardson, J.E., 2012. The Mouse
Genome Database (MGD): comprehensive resource for genetics and genomics
of the laboratory mouse. Nucleic Acids Res. 40, D881–6.
Ernst, J., Kheradpour, P., Mikkelsen, T.S., Shoresh, N., Ward, L.D., Epstein, C.B.,
Zhang, X., Wang, L., Issner, R., Coyne, M., Ku, M., Durham, T., Kellis, M.,
Bernstein, B.E., 2011. Mapping and analysis of chromatin state dynamics in
nine human cell types. Nature 473, 1–9.
Feng, J., Liu, T., Zhang, Y., 2011. Using MACS to identify peaks from ChIP-Seq data.
Curr. Protoc. Bioinf., Chapter 2, Unit 2.14.
Gupta, S., Stamatoyannopoulos, J.A., Bailey, T.L., Noble, W.S., 2007. Quantifying
similarity between motifs. Genome Biol. 8, R24.
Hostikka, S.L., Gong, J., Carpenter, E.M., 2009. Axial and appendicular skeletal
transformations, ligament alterations, and motor neuron loss in HoxC10
mutants. Int. J. Biol. Sci. 5, 397–410.
Island, M.L., Mesplede, T., Darracq, N., Bandu, M.T., Christeff, N., Djian, P., Drouin, J.,
Navarro, S., 2002. Repression by homeoprotein Pitx1 of virus-induced interferon a promoters is mediated by physical interaction and trans repression of
IRF3 and IRF7. Mol Cell Biol. 22, 7120–7133.
Kawakami, Y., Marti, M., Kawakami, H., Itou, J., Quach, T., Johnson, A., Sahara, S.,
O’Leary, D.D.M., Nakagawa, Y., Lewandoski, M., Pfaff, S., Evans, S.M.,
Belmonte, J.C.I., 2011. Islet1-mediated activation of the b-catenin pathway is
necessary for hindlimb initiation in mice. Development 138, 4465–4473.
Lamonerie, T., Tremblay, J.J., Lanctôt, C., Therrien, M., Gauthier, Y., Drouin, J., 1996.
Ptx1, a bicoid-related homeo box transcription factor involved in transcription
of the pro-opiomelanocortin gene. Genes Dev. 10, 1284–1295.
Lanctôt, C., Moreau, A., Chamberland, M., Tremblay, M.L., Drouin, J., 1999.
Hindlimb patterning and mandible development require the Ptx1 gene.
Development 126, 1805–1810.
Langmead, B., Trapnell, C., Pop, M., Salzberg, S.L., 2009. Ultrafast and memoryefficient alignment of short DNA sequences to the human genome. Genome
Biol. 10, R25.
Lee, E.C., Yu, D., Martinez de Velasco, J., Tessarollo, L., Swing, D.A., Court, D.L.,
Jenkins, N.A., Copeland, N.G., 2001. A highly efficient Escherichia coli-based
chromosome engineering system adapted for recombinogenic targeting and
subcloning of BAC DNA. Genomics 73, 56–65.
Logan, M.P.O., Tabin, C.J., 1999. Role of Pitx1 upstream of Tbx4 in specification of
hindlimb identity. Science 283, 1736–1739.
Mayor, C., Brudno, M., Schwartz, J.R., Poliakov, A., Rubin, E.M., Frazer, K.A., Pachter, L.S.,
Dubchak, I., 2000. VISTA: visualizing global DNA sequence alignments of arbitrary
length. Bioinformatics 16, 1046–1047.
Mclean, C.Y., Bristor, D., Hiller, M., Clarke, S.L., Schaar, B.T., Lowe, C.B., Wenger, A.M.,
Bejerano, G., 2010. GREAT improves functional interpretation of cis-regulatory
regions. Nat. Biotechnol. 28, 495–501.
Menke, D.B., Guenther, C., Kingsley, D.M., 2008. Dual hindlimb control elements in
the Tbx4 gene and region-specific control of bone size in vertebrate limbs.
Development 135, 2543–2553.
Naiche, L.A., Papaioannou, V.E., 2003. Loss of Tbx4 blocks hindlimb development
and affects vascularization and fusion of the allantois. Development 130,
2681–2693.
Newburger, D.E., Bulyk, M.L., 2009. UniPROBE: an online database of protein
binding microarray data on protein-DNA interactions. Nucleic Acids Res. 37,
D77–82.
Ouimette, J.-F., Jolin, M.L., L’honoré, A., Gifuni, A., Drouin, J., 2010. Divergent
transcriptional activities determine limb identity. Nat. Commun. 1, 1–9.
Ovcharenko, I., Nobrega, M.A., Loots, G.G., Stubbs, L., 2004. ECR Browser: a tool for
visualizing and accessing data from comparisons of multiple vertebrate
genomes. Nucleic Acids Res. 32, W280–6.
Papenbrock, T., Visconti, R.P., Awgulewitsch, A., 2000. Loss of fibula in mice
overexpressing HoxC11. Mech. Dev. 92, 113–123.
Quinlan, A.R., Hall, I.M., 2010. BEDTools: a flexible suite of utilities for comparing
genomic features. Bioinformatics 26, 841–842.
Salmon-Divon, M., Dvinge, H., Tammoja, K., Bertone, P., 2010. PeakAnalyzer:
genome-wide annotation of chromatin binding and modification loci. BMC
Bioinf. 11, 415.
Sandelin, A., Alkema, W., Engström, P., Wasserman, W.W., Lenhard, B., 2004.
JASPAR: an open-access database for eukaryotic transcription factor binding
profiles. Nucleic Acids Res. 32, D91–4.
Shin, H., Liu, T., Manrai, A.K., Liu, X.S., 2009. CEAS: cis-regulatory element
annotation system. Bioinformatics 25, 2605–2606.
Siva, K., Inamdar, M.S., 2006. Rudhira is a cytoplasmic WD40 protein expressed in
mouse embryonic stem cells and during embryonic erythropoiesis. Genet.
Express. Patterns 6, 225–234.
Szeto, D.P., Rodriguez-Esteban, C., Ryan, A.K., O’Connell, S.M., Liu, F., Kioussi, C.,
Gleiberman, A.S., Izpisúa-Belmonte, J.C., Rosenfeld, M.G., 1999. Role of the
Bicoid-related homeodomain factor Pitx1 in specifying hindlimb morphogenesis and pituitary development. Genes Dev. 13, 484–494.
Taher, L., Collette, N.M., Murugesh, D., Maxwell, E., Ovcharenko, I., Loots, G.G.,
2011. Global gene expression analysis of murine limb development. PLoS One
6, e28358.
Visel, A., Blow, M.J., Li, Z., Zhang, T., Akiyama, J.A., Holt, A., Plajzer-Frick, I.,
Shoukry, M., Wright, C., Chen, F., Afzal, V., Ren, B., Rubin, E.M., Pennacchio,
L.A., 2009. ChIP-seq accurately predicts tissue-specific activity of enhancers.
Nature 457, 854–858.
Visel, A., Minovitsky, S., Dubchak, I., Pennacchio, L.A., 2007. VISTA enhancer
browser—a database of tissue-specific human enhancers. Nucleic Acids Res.
35, D88–92.
Wilbanks, E.G., Facciotti, M.T., 2010. Evaluation of algorithm performance in ChIPSeq peak detection. PLoS One 5, 1–12.
Zentner, G.E., Tesar, P.J., Scacheri, P.C., 2011. Epigenetic signatures distinguish
multiple classes of enhancers with distinct cellular functions. Genome Res. 21,
1273–1283.
Zhang, Y., Liu, T., Meyer, C.A., Eeckhoute, J., Johnson, D.S., Bernstein, B.E., Nusbaum, C.,
Myers, R.M., Brown, M., Li, W., Liu, X.S., 2008. Model-based analysis of ChIP-Seq
(MACS). Genome Biol. 9, R137.