Identification of novel Hoxa1 downstream targets ...capecchi.genetics.utah.edu/.../03/179MakkiDevbio.pdf · control and Hoxa1 null embryos. For genomic profiling, tissue was microdissected

Developmental Biology 357 (2011) 295–304

Contents lists available at ScienceDirect

Developmental Biology

j ourna l homepage: www.e lsev ie r.com/deve lopmenta lb io logy

Identification of novel Hoxa1 downstream targets regulating hindbrain, neural crestand inner ear development

Nadja Makki, Mario R. Capecchi ⁎Howard Hughes Medical Institute and Department of Human Genetics, UT, USA

⁎ Corresponding author at: Howard Hughes Medical INorth 2030 East, Salt Lake City, UT 84112-5331, USA. Fa

E-mail address: [email protected] (

0012-1606/$ – see front matter. Published by Elsevierdoi:10.1016/j.ydbio.2011.06.042

a b s t r a c t

Article history:Received for publication 22 March 2011Revised 10 June 2011Accepted 29 June 2011Available online 18 July 2011

Keywords:Hoxa1MicroarrayNeural crestInner earRhombomere

Hox genes play a crucial role during embryonic patterning and organogenesis. Of the 39 Hox genes, Hoxa1 isthe first to be expressed during embryogenesis and the only anterior Hox gene linked to a human syndrome.Hoxa1 is necessary for the proper development of the brainstem, inner ear and heart in humans and mice;however, almost nothing is known about the molecular downstream targets through which it exerts itsfunction. To gain insight into the transcriptional network regulated by this protein, we performed microarrayanalysis on tissue microdissected from the prospective rhombomere 3–5 region of Hoxa1 null and wild typeembryos. Due to the very early and transient expression of this gene, dissections were performed on earlysomite stage embryos during an eight-hour time window of development. Our array yielded a list of around300 genes differentially expressed between the two samples. Many of the identified genes play a role in aspecific developmental or cellular process. Some of the validated targets regulate early neural crest inductionand specification. Interestingly, three of these genes, Zic1, Hnf1b and Foxd3, were down-regulated in theposterior hindbrain, where cardiac neural crest cells arise, which pattern the outflow tract of the heart. Othertargets are necessary for early inner ear development, e.g. Pax8 and Fgfr3 or are expressed in specific hindbrainneurons regulating respiration, e.g. Lhx5. These findings allow us to propose a model where Hoxa1 acts in agenetic cascade upstream of genes controlling specific aspects of embryonic development, thereby providinginsight into possible mechanisms underlying the human HoxA1-syndrome.

nstitute, University of Utah, 15x: +1 801 585 3425.M.R. Capecchi).

Inc.

Published by Elsevier Inc.

Introduction

Hox proteins constitute a family of transcription factors whichcontrol gene expression networks that regulate biological processessuch as neurogenesis, patterning, organogenesis and cancer (Alexanderet al., 2009; Capecchi, 1997). Mouse knockout studies revealed that Hoxgenes execute their role in a specific segment or domain of the embryo,often affecting several tissues at a given axial level (Mallo et al., 2010).Although many gain- and loss-of-function experiments have beencarried out, little is known about the molecular targets and thedevelopmental pathways regulated by Hox genes (Hueber andLohmann, 2008). In this study, we set out to identify the downstreamtargets of a specific Hox gene, Hoxa1. This gene affects the developmentof a diverse array of tissues in the anterior domain of the embryoincluding the brainstem, inner ear and heart.

Hoxa1 is strongly expressed in the neuroectoderm and mesodermat the level of the presumptive hindbrain (precursor of the brainstem)from mouse embryonic day (E) 7.75 to 8.5 (Murphy and Hill, 1991).Hoxa1 knockout mice die at or shortly after birth from breathingdefects, which are thought to result from mispatterning of the

hindbrain (Chisaka et al., 1992; Lufkin et al., 1991). During develop-ment, the hindbrain is subdivided into eight transient territoriestermed rhombomeres (r) (Lumsden and Krumlauf, 1996) andHoxa1−/− embryos exhibit abnormalities in r3–r5. Additionally, theotic vesicle (embryonic progenitor of the inner ear) forms but fails todifferentiate and cranial ganglia, condensations of sensory neurons inthe head, are smaller and do not connect properly with the brain(Mark et al., 1993). Cranial ganglia develop in part from cranial neuralcrest cells, which migrate from the dorsal hindbrain (Barlow, 2002),where Hoxa1 is expressed. So far it is unclear through whichmechanisms Hoxa1 regulates the development of neural crest cellsor the inner ear. Hoxa1 lineage analysis suggests that Hoxa1 mightplay a direct role in early patterning of the otic placode (precursor ofthe otic vesicle) and specification of neural crest cell precursors, whilethey reside in the neural tube (Makki and Capecchi, 2010).

More recently, humans with homozygous truncating mutations inHOXA1 have been identified (Athabascan Brainstem DysgenesisSyndrome and Bosley–Salih–Alorainy Syndrome). These patientssuffer from hypoventilation (requiring mechanical ventilation),deafness, facial weakness, vocal cord paralysis and swallowingdysfunction (Holve et al., 2003; Tischfield et al., 2005). In addition,patients display defects in the outflow tract of the heart, which havenot been described in mice so far. Notably, the development of thecardiac outflow tract depends on the influx of neural crest cells, which

296 N. Makki, M.R. Capecchi / Developmental Biology 357 (2011) 295–304

originate in the posterior hindbrain at the level of r6–r8 (Brown andBaldwin, 2006), where Hoxa1 is expressed.

Despite of what we know about the importance of Hoxa1 in properdevelopment of several embryonic tissues in humans and mice, almostnothing is known about the transcriptional network that is regulated bythis protein. In this study, we carried out a genome-wide microarrayanalysis to identify genes that are differentially expressed betweencontrol and Hoxa1 null embryos. For genomic profiling, tissue wasmicrodissected from the prospective rhombomere 3–5 region ofHoxa1Δ/Δ and wild type embryos at the 1–6 somite stage (ss). Ouranalysis identified novel targets of Hoxa1 that play a role in neuralcrest specification, otic placode patterning, and reticulospinal neurondevelopment.

Materials and methods

Gene targeting and genotyping

A 7.9 kb genomic DNA fragment containing the Hoxa1 locus wassubcloned into a conventional plasmid and an artificial AscI site wasplaced 36 bp downstream of the stop codon as described previously(Tvrdik and Capecchi, 2006). To generate the Hoxa1 conditional allele(Hoxa1c), one loxP site together with an EcoRI site were inserted200 bp upstream of the Hoxa1 transcription initiation site into a SwaIsite. The downstream loxP site along with an EcoRI site and a PolII-frt-Neo-frt selection cassette were inserted into the artificial AscI site 3′ ofthe Hoxa1 stop codon. Positive clones were identified by digestinggenomic DNA with EcoRI, Southern blotting and hybridization with a5′ external probe. Selected clones were further analyzed by digestionwith KpnI and hybridization with an exon1 and a Neo probe. PositiveES cell clones were injected into C57BL/6 blastocysts and chimericmaleswere crossed to C57BL/6 females. The neomycin resistance genewas removed by crossing the mice to anFlpe-deleter line (Rodriguezet al., 2000). The Hoxa1-deletion allele (Hoxa1Δ) was generated bycrossing Hoxa1 conditional mice to an Hprt-Cre deleter mouse (Tanget al., 2002). Recombination was verified by Southern analysis andPCR. Genotyping was performed using multiplex PCR. The followingprimers were used: wild type forward NM228 5′-TGAGGCTACTC-CAGCCCAACTC-3′, deletion forward NM230 5′-CTCTCACCTCTTGC-CAGTTCAGC-3′, reverse NM229 5′-CAATTGATGTGGACACCCGATG-3′,generating a 220 bp wild type, a 326 bp conditional and a 520 bpdeletion band.

Mouse breeding and tissue dissection

Hoxa1Δ/+ mice were maintained on a C57BL/6 background. Timedmatings were set up between Hoxa1Δ/+ mice and embryos wereharvested at E8.25. Deciduas were isolated in cold PBS and transferredinto HEPES-buffered DMEM with 5% FBS on ice. Each embryo wasisolated in a separate dish in PBS, extraembryonic tissues wereremoved and the number of somites counted. Using fine tungstenknives, the bulge region (rhombomere 3–5), including neuroecto-derm, mesoderm and otic ectoderm, was isolated and the tissuetrimmed by a horizontal cut at the level of the floorplate. The tissuewas then transferred into 40 μl of RLT buffer (Qiagen Micro-RNA Easykit), vortexed immediately for 1 min and stored on ice until allembryos were processed. The yolk sac was collected for DNA isolationand genotyping. Finally, the tissue was homogenized by vortexing for5 min followed by snap freezing in liquid nitrogen and storage at−80 °C. A total of 221 embryoswere collected and sorted according togenotype (verified at least twice) and somite stage. Twenty-four wildtype and 24 Hoxa1Δ/Δ embryos at the 1–6 somite stage were chosenfor analysis and pooled into four wild type and four mutant samples,containing one embryo of each somite stage.

RNA isolation, array hybridization and statistical analysis

RNA was isolated from the eight samples using the RNAqueous-Micro Kit (Ambion) with an on-column DNase treatment (Qiagen).The concentration and quality of the RNA was determined at theUniversity of Utah Microarray Core Facility using a Nanodrop andBioanalyzer (Agilent). The RNA Integrity Number (RIN) deduced fromthis analysis was 9.9–10 for all samples, which denotes an excellentRNA quality with no degradation (Schroeder et al., 2006). The finalconcentrations of total RNA varied from 15 to 20 ng/μl and 150 ng ofRNA from each pool was subjected to a single linear amplificationlabeling reaction with Cy3. RNA was hybridized to Agilent mousewhole genome 44 K microarray slides (Agilent), using the Agilentone-color gene expression hybridization protocol. Slides werescanned (Agilent G2505B) at 5 μm resolution using an extendeddynamic range protocol, and images were processed with AgilentFeature Extraction software 10.5.1.1. Within-array normalization wasperformed using the “Background detrending” software (Agilent). Thenonuniform outlier features (spots) were removed and the intensityvalues were transformed to a log base 2 scale. Signal density blotsshowed uniform ranges and distributions of intensity values fromeach array and no between-array normalization was necessary. Alleight array files were then compiled into a working directory andimported into the statistical analysis program “R” (Dudoit et al.,2003). Genes significantly differentially expressed were identifiedusing the Rank Products algorithm with the default setting of 100permutations (Breitling et al., 2004). Rank Products analysis waschosen because of its biologically meaningful emphasis on the foldchange of gene expression and the reproducibility in samples withsmall numbers of replicates. GO analysis was performed using DAVID(Dennis et al., 2003; Huang da et al., 2009) on significantlydifferentially expressed genes. In case of overlapping and similar GOterms, one representative is listed, and terms that are too generalwere not included. Data was hierarchically clustered with Spotfire(TIBCO) and heat maps for selected genes were generated. Themicroarray data can be found in the Gene Expression Omnibus (GEO)of NCBI through accession number GSE25868.

Quantitative real-time PCR (qPCR)

60 ng of total RNA was linearly amplified using the qScript cDNASuperMix (Quanta Biosciences). Reverse transcription and PCR condi-tions were essentially as described (Schmittgen and Livak, 2008) usingthe SYBR Green detection method. Primer pairs (Table S2) wereobtained from the PrimerBank database (Wang and Seed, 2003).Reactions were run on a 7900HT thermal cycler (Applied Biosystems)in the Genomics Core Facility at the University of Utah. For the finalexperiment, three wild type and three Hoxa1Δ/Δ cDNA samples(biological replicates) were analyzed individually in three replicates ofeach reaction (technical replicates) and the mean threshold cycle (CT)for each genewas derived. Relative expression levelswere calculated bythe ΔCT method (Schmittgen and Livak, 2008), normalizing to thehousekeeping gene β-actin, and data expressed as mean fold changerelative to wild type. Unpaired, two-tailed Student's t-test was used tocalculate P values between the Hoxa1 null and control samples.

Inner ear paint-filling and RNA in situ hybridization

For inner ear paint-filling, E15.5 embryoswere harvested and fixedovernight in Bodian's fixative. Embryos were washed in PBS,dehydrated in ethanol and cleared in methyl salicylate. Heads werehemisected and inner ears injected with 2% white latex paint inmethyl salicylate using a micropipette (Morsli et al., 1998). For RNA insitu hybridization, Digoxigenin-labeled antisense cRNA probes weregenerated from plasmids carrying cDNA fragments. The followingcloned mouse cDNAs were obtained, sequenced and used to prepare

297N. Makki, M.R. Capecchi / Developmental Biology 357 (2011) 295–304

riboprobes: Foxd3 (from T. Labosky) (Labosky and Kaestner, 1998),Hnf1b and Lhx5 (from Q. Ma) (Gray et al., 2004), Spry4 (from K. Shim/G. Martin) (Minowada et al., 1999), Pax8 (from A. Groves) (Ohyamaand Groves, 2004), Zic1 (from R. Arkell) (Elms et al., 2004), Lefty2(from Y. Saijoh/H. Hamada) (Meno et al., 1996), Hnf4a (fromY. Saijoh). Probes for Fzd8 and Fgfr3 (from L. Urness) were generatedfollowing direct PCR amplification of the 3′ UTR from genomic DNA. A28-base T7 RNA polymerase promoter (5′-GGATCCTAATACGACTCAC-TATAGGGAG-3′) was incorporated at the 5′ end of the reverse primer.Whole-mount in situ hybridization was performed on embryosisolated from timed pregnancies essentially as described (Henriqueet al., 1995).

Results

Hoxa1Δ/Δ mice exhibit the same phenotypes as previously describedHoxa1 null lines

A new Hoxa1 null allele (Hoxa1Δ) was created by flanking theHoxa1 coding region with loxP sites to generate a conditional allele(Fig. 1A) and then deleting the intervening sequence using Crerecombinase. Hoxa1 conditional (Hoxa1c) mice were generated fromtargeted ES cells (Fig. 1B) and then crossed to a Flpe-deleter line(Rodriguez et al., 2000) to excise the neomycin resistance gene. Micehomozygous for the Hoxa1 conditional allele are phenotypically wildtype. To generate a Hoxa1-deletion allele (Hoxa1Δ), Hoxa1 conditionalmice were crossed to an Hprt-Cre deleter line (Tang et al., 2002). Asexpected, mice with a homozygous deletion of Hoxa1 (Hoxa1Δ/Δ)resemble previously reported Hoxa1 null mice (Chisaka et al., 1992;Mark et al., 1993). Hoxa1Δ/Δ mice are born at normal Mendelian ratios

EC

loxP

E

EEEC

FrtloxP

E

3 kb2 kb

E C

exon1Hoxa1+

Hoxa1c

Hoxa1Δ

A

Hoxa1+

Hoxa1c

c/+c/+ +/+

B

Hoxa1+Hoxa1cc/c c/+ Δ/+ +/

+Hoxa1ΔΔ/Δ

Fig. 1. Hoxa1 targeting and phenotype analysis. (A) Depiction of Hoxa1 wild type (Hoxa1+),inserting a 5′ loxP site 200 bp upstream of the Hoxa1 transcription initiation site and a loxP-was removed by recombination, leaving one frt site behind. In the Hoxa1Δ allele, the entire Hboxes, UTRs. C, ClaI; E, EcoRI. (B) Upper panel: Southern blot analysis to identify positive Hgenerate an 8.3 kb wt and a 6 kb Hoxa1c band. Lower panel: PCR genotyping to identify the dview of paint-filled inner ears from E15.5 control (C) and Hoxa1Δ/Δ mice (C′). asc, anteendolymphatic sac; lsc, semicircular canal; s, saccule; u, utricle.

but die shortly after birth at perinatal day P0–P1 (n=34). We alsoexamined Hoxa1Δ/Δ embryos for inner ear defects using the inner earpaint-fill technique (Morsli et al., 1998) and found that the otic vesicleforms but does not differentiate (Figs. 1C, C′) (n=7), as was reportedin previous studies (Pasqualetti et al., 2001). Therefore, the Hoxa1Δ

allele represents a new Hoxa1 null allele, which was used in allsubsequent experiments.

Hoxa1 is expressed very transiently in its most anterior domain

Previous studies showed that Hoxa1 is most strongly expressed inthe anterior hindbrain (prospective r3–r5) and neighboring meso-derm (Makki and Capecchi, 2010; Murphy and Hill, 1991) and that allphenotypes resulting from loss of Hoxa1 function are associated withits most anterior expression domain (Chisaka et al., 1992; Lufkin et al.,1991). Therefore, we wanted to identify the exact embryonic timewindow, during which Hoxa1 is expressed in the prospective r3–r5region by carrying out RNA in situ hybridization at specific somitestages. As reference we visualized Krox20 expression, which can bedetected in r3 from the 4ss and in both r3 and r5 from the 7ss(Figs. 2A, B) (Mechta-Grigoriou et al., 2000). Our analysis revealedthat Hoxa1 is only expressed in its most anterior domain (theprospective r3/r4 boundary) from E7.75-2ss (data not shown). Fromthe 2ss (~E8.0), Hoxa1 expression starts retracting to posterior r4. Atthe 4ss (~E8.25), when Krox20 is first expressed as a single stripe in r3,Hoxa1 has retracted to r5 (Fig. 2A). When the second stripe of Krox20appears in r5 at the 7ss (~E8.5), Hoxa1 is no longer expressed in thisrhombomere (Fig. 2B). Thus, Hoxa1 is expressed for only around 12 h(E7.75-6ss) in its most anterior domain, which is muchmore transientthan previously believed (Murphy and Hill, 1991).

EE C

C

E C

exon2

1 kb

FrtloxP

control Hoxa1Δ/Δ

C C’cc

asc

psc lsc

us

co

lsc

es

ed

conditional (Hoxa1c) and deletion (Hoxa1Δ) alleles. The Hoxa1c allele was generated byfrt-PolII-Neo-frt cassette 36 bp downstream of the Hoxa1 stop codon. The Neo cassetteoxa1 promoter and coding region are deleted. Black boxes, Hoxa1 coding region; whiteoxa1c clones. DNA was digested with EcoRI and hybridized with a 5′ external probe toifferent Hoxa1 alleles. (C, C′) Abnormal inner ear morphology of Hoxa1mutants. Lateralrior semicircular canal; cc, common crus; co, cochlea; ed, endolymphatic duct; es,

D

5ss

r3

r5

r4ov

A

P

r3

r5

r4OP

6ss

wt Δ/Δ

Hoxa1

Hprt

Hoxb1

Kreisler

-

7ss7ss2ss 4 ss4ss

r3

posterior r4 r5 caudal to r5r3 r3 + r5

A B

C Ene e

m

nee

m

Krox20Krox20

Krox20 Pax2

Hoxa1 Hoxa1Hoxa1

r4r5

r3

r5r6

fp

Fig. 2. Temporal window of Hoxa1 expression and tissue dissection for array analysis. (A, B) Hoxa1 is expressed very transiently in its most anterior domain as visualized by RNA insitu hybridization in comparison to Krox20. The hindbrain “bulge” region (r3–r5) is marked by an open bracket. (A) At the 4ss, Krox20 labels r3 (the anterior border of the bulge).Hoxa1 is still expressed in posterior r4 at the 2ss but retracts to r5 at the 4ss. (B) At the 7ss, Krox20 also labels r5. At this stage,Hoxa1 is no longer expressed in this rhombomere. (C) Insitu hybridization was performed after cutting the hindbrain bulge region (arrowheads indicate cutting sites). Krox20 staining verified that the bulge corresponds to prospective r3–r5 and in situ for Pax2 demonstrates that the dissected region includes the entire otic ectoderm. (D) Schematic depiction (top) and brightfield image (bottom) of the dissected tissueused for RNA isolation. Embryos were flattened out and tissue was cut along the edges of the hindbrain bulge region. The tissue was then trimmed by a horizontal cut along thefloorplate of the neural tube (fp), generating a piece of tissue that contains neuroectoderm (ne), otic ectoderm (e) and mesoderm (m). (E) RT-PCR demonstrates that changes in theexpression of known downstream effectors of Hoxa1 can be detected in the dissected tissue of wild type and Hoxa1Δ/Δ embryos.

298 N. Makki, M.R. Capecchi / Developmental Biology 357 (2011) 295–304

Identifying and isolating the relevant tissue for microarray analysis

In order to identify genes regulated by Hoxa1, we set out to collecttissue from the prospective r3–r5 region of Hoxa1Δ/Δ and wild typeembryos for microarray analysis. Since our in situ experimentsrevealed that Hoxa1 is expressed in this region from around E7.75to the 6ss, we chose to collect embryos at the 1–6ss. This is anapproximately eight-hour time window (Tam, 1981), around andslightly after the peak of Hoxa1 expression, and before phenotypicmanifestations are apparent in Hoxa1 null embryos. Therefore, webelieve that our experimental setupwould allow identification of bothdirect and indirect targets of Hoxa1. Conveniently, at this time theprospective r3–r5 region is morphologically visible as a “bulge” thatforms in the future hindbrain (Fig. 2A, B open brackets). The bulgeregion was microdissected by performing two cuts along the edges ofthe bulge (Fig. 2C) and then trimming the tissue at the level of thefloorplate to include neuroectoderm, mesoderm and otic ectoderm atthe level of r3–r5 (Fig. 2D). To confirm that the bulge region includedthe entire r3–r5 region, we performed in situ hybridization for Krox20after cutting the tissue (Fig. 2C). Since Hoxa1 null mice also exhibitsevere inner ear defects, we wanted to include the otic ectoderm, theprecursor of the inner ear, which develops at the level of r4–r5(Ohyama and Groves, 2004). In situ staining for the otic marker Pax2on cut tissue confirmed that this region was included in our dissection(Fig. 2C). Finally, we wanted to verify that the tissue chosen fordissection would allow us to detect expression changes in knownHoxa1 downstream targets between wild type and Hoxa1 nullembryos. Therefore, we isolated RNA from a small number ofdissected embryos and performed RT-PCR on two of the few known

Hoxa1 targets, Hoxb1 and Kreisler (Mafb) (Pasqualetti et al., 2001). Wesaw clear changes in RNA levels of these two genes between the twogenotypes (Fig. 2E). This gave us confidence to carry out a large scaleanalysis using this technique (Fig. 3A). A total of 221 embryos from 52Hoxa1Δ/+ females were dissected and genotyped. TheMendelian ratiowas as follows: 21% homozygous, 51% heterozygous and 28% wildtype. For microarray analysis, embryos at the 1–6 somite stage werepooled into four wild type and four mutant samples, each containingone embryo per somite stage.

Microarray analysis reveals Hoxa1 candidate targets involved indifferent developmental processes

Toenable a comprehensive assessmentofHoxa1-regulatedgeneswecompared gene expression profiles of four Hoxa1Δ/Δ and four wild typesamples using genome-wide microarray analysis. Rank Productsanalysis (Breitling et al., 2004) yielded a list of 299 differentiallyexpressed genes (137 down-regulated and 162 up-regulated in themutant) with a≥2-fold change in expression at a false discovery rate of0.05andwithP-values≤0.0002(Table S1). As expected, themosthighlydown-regulated gene in this list is Hoxa1, with a fold change of 70. Thetwo known downstream targets of Hoxa1 were also among the down-regulated genes in the list:Mafb (Kreisler) (Pasqualetti et al., 2001) andHoxb1 (Barrow et al., 2000), with a 6.7-fold (third most highly down-regulatedgene) anda2.5-folddown-regulation, respectively. Inorder toidentify other “genes of interest”, we scanned the whole list of 299potential targets for genes that fulfill one of two criteria: (i) known toplay a role in a developmental process or (ii) expressed during earlyembryogenesis. Twelve of the 137 down-regulated and seven of the 162

AHarvest embryos from Hoxa1Δ/+ x Hoxa1Δ/+

at E8.25 (1-6ss)

Dissect tissue at the level of r3-r5

Collect in buffer RLT, homogenize, freeze, collecttissue for genotyping

Generate 4 pools of Hoxa1 null and wildtype embryos(6 embryos per pool), extract RNA with Ambion Micro kit

Hybridize samples to Agilent 4x44K Mouse WholeGenome Array

Statistical analysis using “Rank Products Test”identifies 137 down- and 162 up-regulated genes,

with > 2-fold change in expression

Experimental procedure

Analyze and validate candidate genes by qPCRand ISH

B

up-

regu

late

ddo

wn-

reg u

late

d

1 432

Dfna5

Spry4Fzd8

Wnt10b

Pax8

Hnf1b

Zic1

Sema3cLhx5Foxd3

Nodal

Clic5

Fgfr3

Left1

Apob

Tbx15

Mafb

Hoxb1

fold change

6.04.84.5

2.32.3

2.42.4

2.5

2.6

2.8

3.0

2.5

2.5

2.7

2.8

5.76.8

6.7

C

Gene Ontology TermFold of

Enrichment Genes

Embryonic organ development 3.9 Mafb, Hoxb1, Foxd3, Zic1, Tbx15, Hoxd3, Tshz1, Nodal, Clic5, Rbp4

Epithelium development 1.4 Sema3c, Npnt, Fgfr3, Nodal, Tcfcp2l1

Hindbrain development 8.2 Mafb, Hoxb1, Hnf1b, Lhx5, Smo

Inner ear development 5.6 Mafb, Pax8*, Zic1, Dfna5, Fgfr3, Clic5

Vasculature development 1.6 Sema3c, Smo, Arhgap24, Apob, Nodal

Hemopoietic or lymphoid organdevelopment

1.4 Bcl11a, Sox6, Tnfrsf11a, Vav1, Hba-x

Cardiac muscle tissue development 4.1 Csrp3, Rbp4, Ttn

Contractile fiber 6.8 Ankrd1 (cardiac), Mybpc3 (cardiac), Abra (heart tube), Acta2 (aorta),Csrp3, Myh4, Kcnj8, Ttn, Tmod1

Neuron differentiation 2 Lhx5, Dnfa5, Smo, Fezf2, Slitrk1, Fgfr3,Clic5, Dbx1

Muscle cell differentiation 4.7 Sox6, Dner, Met, Mybpc3 (cardiac), Musk, Ttn, Tmod1

Cell migration 2.3 Sema3c, Smo, Met, Nodal, Lefty1, Apoa1, Gab2

Negative regulation of apoptosis 1.6 Smo, 6030408C04Rik, Spp1, Sgk3, Il2rb

Retinol metabolism 3.4 Cyp3a13, Adh1, Ugt1a6b

TGF-beta signaling pathway 3.6 Dcn, Nodal, Lefty1, Lefty2

Wnt receptor signaling pathway 1.8 Wnt10b, Fzd8, Smo

Fig. 3.Microarray analysis identifies novel Hoxa1 targets involved in various developmental processes. (A) Flowchart showing the experimental procedure from embryo harvestingto validation of microarray targets. (B) Expression heat maps for relative expression of genes of interest obtained from four Agilent microarrays comparing Hoxa1Δ/Δ to controlembryos. Green indicates decreased and red increased expression in mutants. Note the reproducible direction and magnitude of the changes. Fold changes are log base 2; Pb0.0005.(C) Gene ontology (GO) analysis was performed on significantly differentially expressed genes using DAVID. Enriched GO terms for genes significantly down-regulated (green) orup-regulated (red), as well as fold of enrichment (compared to genome-wide background level) are listed. Asterisk indicates that the gene is involved in corresponding GO functionbut failed to be recognized by DAVID.

299N. Makki, M.R. Capecchi / Developmental Biology 357 (2011) 295–304

up-regulated genes were selected as potentially interesting candidates(Table 1). Themagnitude of expression changes of the selected genes ineach of the four samples is illustrated in the intensity map represen-tations (Fig. 3B). In order to identify biological processes that might beregulated by Hoxa1, we carried out gene ontology (GO) analysis ofsignificantly up- and down-regulated genes using the DAVID software(Database for Annotation, Visualization and Integrated Discovery)(Dennis et al., 2003; Huang da et al., 2009). This analysis identifiedcategories such as hindbrain development, inner ear development,vascular development, neuron differentiation and cell migration(Fig. 3C). Several of the genes listed under one of these categorieswere also selected as “genes of interest”.

Validation of microarray targets by quantitative PCR

We carried out two qPCR experiments to identify and validatenovel downstream targets of Hoxa1. First, we performed an initialqPCR screening of the 19 “genes of interest” (Table 1). For this, RNAfrom thepool of dissected tissue of three 4–5 somite and two7–8 somitewild type and Hoxa1Δ/Δ embryos was used. In this screening six genes(Foxd3, Lhx5, Hnf1b, Zic1, Pax8 and Fgfr3) were found to be differentiallyregulated between Hoxa1 null and wild type embryos in accordancewith themicroarray data. A second qPCR experiment was performed tofurther validate these six targets. qPCR validation was based on threebiological replicates, each containing dissected tissue from six 1–6

Table 1Differentially expressed genes of interest from Hoxa1 microarray.

Gene FC Gene name Proposed function Reported expression

Dfna5 6.0 Deafness, autosomal dominant 5 Inner ear receptor cell differentiation E10.5 (northern)

Foxd3 4.8 Forkhead box D3 Maintenance/induction of NCC E8.5, pre-migratory NCC

Lhx5 4.5 LIM homeobox protein 5 Respiration (reticulospinal), cardiac dev. E10.5

Sema3c 3.0 Semaphorin 3c NCC, nervous system, heart dev. E10.5, cardiac OT

Hnf1b 2.8 HNF1 homeobox b Hindbrain r5 dev., NCC E8.0, hindbrain, NCC, foregut

Spry4 2.6 Sprouty 4 Fgf signaling, craniofacial dev. E8.5, lateral to hindbrain

Fzd8 2.5 Frizzled 8 Wnt receptor E8.5, head, otic placode

Wnt10b 2.4 Wingless related 10b Wnt signaling E11, 1st arch

Tbx15 2.4 T-box 15 Craniofacial dev. E11, CNS

Pax8 2.3 Paired box gene 8 Inner ear (otic placode specification) ≥ 0ss, otic placode

Zic1 2.3 Zinc finger protein of the cerebellum 1 Neural plate patterning, NC specification ≥ 6ss, hindbrain, neural tube

Hoxd3 2.1 Homeobox d3 Cervical vertebrae dev., postnatal death E9, hindbrain r4/r5 border

Apob 6.8 Apolipoprotein B Artery morphogenesis E7.5 (RT-PCR)

Clic5 5.7 Chloride intracellular channel 5 Auditory receptor cell organization E16.5, cochlea

Lefty1 2.8 Left right determination factor 1 Left−right patterning 3−6ss, floor plate

Nodal 2.7 Nodal Left−right patterning 3−5ss, node, mesoderm

Hnf4a 2.7 Hepatic nuclear factor 4a Endodermal organ development E8.5, fore-midgut endoderm

Fgfr3 2.5 Fibroblast growth factor receptor 3 Inner ear dev. (hearing loss in humans) E9.0, otic placode

Lefty2 2.3 Left right determination factor 2 Left-right patterning 3−6ss, mesoderm

Nineteen candidate genes were selected from the total list of 299 differentially expressed genes for further analysis by qPCR and/or in situ hybridization, based on their publishedexpression pattern and/or proposed function during development as deduced from the Mouse Genome Informatics (MGI) webpage (http://www.informatics.jax.org/). Twelvecandidates were selected from the list of down-regulated genes (top) and seven from the list of up-regulated genes (bottom). Genes highlighted in green or redwere confirmed to bedown- or up-regulated, respectively. FC, fold change.

300 N. Makki, M.R. Capecchi / Developmental Biology 357 (2011) 295–304

somite embryos (the same samples used for microarray analysis).Unpaired, two-tailed Student's t-test was used to calculate P valuesbetween Hoxa1 null and control samples. In agreement with themicroarray results, Lhx5 and Foxd3 were ~5-fold down-regulated;Hnf1b, Pax8 and Zic1 were ~2-fold down-regulated and Fgfr3 was ~2-fold up-regulated compared to wild type (Fig. 4).

Validation of microarray targets by in situ hybridization

To further validate candidates from our “gene of interest” list, wecompared gene expression in somite-matched Hoxa1 null and controlembryos by whole mount in situ hybridization. Expression patterns ofthe following genes were examined: Fgfr3, Foxd3, Fzd8, Hnf1b, Hnf4a,Lefty2, Lhx5, Pax8, Spry4, Zic1. No obvious differences in expression ofSpy4, Fzd8 or Lefty2were seen betweenHoxa1Δ/Δ and control embryosat the 3-10ss (data not shown) and Hnf4a was not detected inembryonic tissue prior to E8.5. Interesting differences were found inthe expression patterns of Foxd3, Zic1 and Hnf1b, all genes known tobe expressed in neural crest precursors in the hindbrain. Foxd3expression in the hindbrain bulge region (prospective r4) was absentin Hoxa1 mutants and expression in the posterior hindbrain(prospective r6–r8) was reduced (Figs. 4A, A′). Similarly, expressionof Zic1 and Hnf1b in the posterior hindbrain (future r5–r8) of Hoxa1mutants was severely reduced (Figs. 4B, C, B′, C′). Moreover,expression of Pax8, a gene required for otic placode specification,was reduced in the placode of Hoxa1 null embryos as early as the 4ss(Figs. 4D, D′). Consistent with upregulation of Fgfr3 expression in themicroarray and by qPCR, in situ analysis revealed an anteriorexpansion of Fgfr3 expression from the r4/r5 boundary in wild typeembryos to the r3/r4 boundary in Hoxa1 null embryos (Figs. 4E, E′).Finally, we detected Lhx5 expression in the hindbrain bulge region(prospective r4) as early as E8.25 (6 somite stage) (Fig. 4F). Thisexpression domain was absent in Hoxa1 mutants (Fig. 4F′). Interest-ingly, three of the validated genes (Foxd3, Zic1, Hnf1b) are knownto play a role in neural crest development (Aruga, 2004; Barbacciet al., 1999; Dottori et al., 2001), two of the genes (Pax8, Fgfr3)are important for inner ear development (Mackereth et al., 2005;Pannier et al., 2009) and one gene (Lhx5) is expressed in hindbrainreticulospinal neuron precursors (Cepeda-Nieto et al., 2005; Grayet al., 2004) (Table 1).

Discussion

Although Hoxa1 is crucial for proper development of thehindbrain, inner ear and neural crest in humans and mice, little isknown about the downstream genes that are controlled by thistranscription factor. Here, we carried out microarray analysis of thisearly expressed Hox gene and compiled a list of 299 candidate targets.Through systematic analysis of this list, we validated an interesting setof Hoxa1 effector genes. These genes are known to control specificdevelopmental processes such as neural crest induction, inner earpatterning and hindbrain neuron specification and can now be placedin a gene cascade downstream of Hoxa1. This allows us to suggest anew model for how Hoxa1 might regulate the development of theabove tissues (Fig. 5) and opens up many new avenues for furtherinvestigation. To our knowledge, this is the first microarray analysisperformed as early as E8.25 to identify gene expression patterns in thedeveloping mammalian hindbrain and adjacent tissues.

Identification and validation of six novel downstream targets of Hoxa1involved in development of the neural crest, inner ear and hindbrainneurons

From the list of 299 putative Hoxa1 targets, we selected 19 genesfor further analysis. These genes were chosen based on theirexpression during early embryogenesis and/or a proposed functionin a developmental process or signaling pathway. Of the 19 genes, sixvalidated by qPCR and in situ hybridization. Three of the validatedHoxa1 targets, Foxd3, Zic1 and Hnf1b are involved in early neural crestdevelopment. Foxd3 is expressed in premigratory neural crest cells inthe hindbrain at around E8.5 (Labosky and Kaestner, 1998) and hasbeen shown to promote the development of neural crest from neuraltube progenitors (Dottori et al., 2001). Deletion of Foxd3 in neuralcrest cells using the Wnt1-Cre driver results in loss of neural crest-derived structures (Teng et al., 2008). In Foxd3c/−; Wnt1-Cre embryoscranial neural crest-derived ganglia and nerves are smaller. The samephenotype is seen in Hoxa1 null embryos, where cranial ganglia andtheir associated nerves are reduced in size (Mark et al., 1993),suggesting that Hoxa1 acts upstream of Foxd3 in neural crestdevelopment. The second gene, Zic1, is expressed in the neural tube,including the dorsal hindbrain from which neural crest cells arise(Elms et al., 2004; Gaston-Massuet et al., 2005; Nagai et al., 1997). Zic1

Hox

a1nu

llw

t

Zic1 Hnf1b Pax8 Fgfr3

EDCB

E’D’C’B’

7ss 7ss4ss8ss

7ss 7ss4ss8ss

A’

A

6ss

6ssFoxd3

F

F’

6ss

6ssLhx5

0

0.2

0.4

0.6

0.8

1

1.2

1.4

1.6

1.8

Lhx5 Foxd3 Hnf1b Pax8 Zic1 Fgfr3

rela

tive

expr

essi

on

wtHoxa1Δ/Δ

*

**

**

**

PCR FC

Array FC4.94.5

4.54.8

2.42.8

2.02.3

1.92.3

1.72.5

Fig. 4. Validation of novel Hoxa1 targets by RNA in situ analysis and quantitative PCR. Identification of six novel downstream targets of Hoxa1. Top panel: validation of Hoxa1microarray candidate genes by in situ hybridization in somite-matched control (A–F) and Hoxa1 null (A′–F′) embryos. (A, A′) Foxd3 expression in premigratory neural crest in thehindbrain is absent in future r4 (arrowhead) and strongly reduced in the posterior hindbrain (future r6–r8; open bracket) of Hoxa1mutants. (B, B′) Zic1 expression in the posteriorhindbrain is absent or reduced in Hoxa1 null embryos. (C, C′) Hnf1b expression is reduced in the posterior hindbrain. (D, D′) Pax8 expression in the otic placode (arrowhead) isreduced. (E, E′) Fgfr3 expression in the hindbrain is expanded anteriorly (black arrowhead: anterior border of Fgfr3 expression, white arrowhead: r2/r3 boundary). (F, F′) Lhx5expression is absent in the hindbrain bulge region (prospective r4; arrowhead) but is unaffected in the forebrain. 2–4 embryos per genotype were analyzed. Bottom panel: validationof candidate genes by quantitative PCR. Relative changes in gene expression levels were analyzed in three wt and three Hoxa1Δ/Δ samples (biological replicates). The mean thresholdcycle (CT) for each gene was derived from triplicate reactions for each sample. Relative expression levels were calculated by the ΔCT method, normalizing to the housekeeping geneβ-actin. Expression changes in Hoxa1Δ/Δ samples (white) are plotted as mean fold change relative to wt samples (black). Fold changes (FC) detected by qPCR are very similar to thefold changes found by microarray analysis. Data are represented as mean+/−SEM. *Pb0.02, **Pb0.002 by Student's two-tailed t-test.

301N. Makki, M.R. Capecchi / Developmental Biology 357 (2011) 295–304

was shown to play a role in early neural plate patterning, neural fateacquisition and neural crest specification in Xenopus (Aruga, 2004;Merzdorf, 2007), where it acts upstream of Pax3 and interacts withGbx2, the earliest factor in neural crest induction (Li et al., 2009).

Zic1−/− mice exhibit cerebellar abnormalities but neural crest defectshave not been studied in these mice. Besides Foxd3 and Zic1, whichplay a role in neural crest specification, we identified Hnf1b as adownstream target of Hoxa1. This gene is expressed in the hindbrain,

Hindbrain patterning

Neural crestNeural crest Hoxa1

Hnf1b

Krox20

Foxd3Zic1

Hnf1b

Cardiac outflow tractCardiac outflow tract

Inner ear

Fgfr3 RespirationRespiration

Reticulospinal neuronsReticulospinal neuronsLhx5

Pax8

Hindbrain patterning

Fig. 5. Proposed model for the regulation of hindbrain, inner ear and neural crest development by Hoxa1. Our data suggests that Hoxa1 influences hindbrain patterning throughHnf1b, which in turn activates Krox20. It also suggests that Hoxa1might regulate neural crest development, through Foxd3, Zic1 and Hnf1b, which could be the reason for the outflowtract defects in humans. In inner ear development, Hoxa1 acts upstream of Pax8 and Fgfr3. In addition, Hoxa1 might regulate Lhx5 expression in reticulospinal neuron precursors,which could contribute to the respiratory defects in Hoxa1 knockout mice. Whether the above effects are direct or through Hoxa1's influence on hindbrain patterning remains to beshown (as highlighted by the dotted arrows) and will be the ground for future investigations.

302 N. Makki, M.R. Capecchi / Developmental Biology 357 (2011) 295–304

neural crest cells and the foregut at E8.0 and is required for visceralendoderm specification and differentiation (Barbacci et al., 1999;Coffinier et al., 1999; Haumaitre et al., 2005). Because of the role invisceral endoderm development,Hnf1b null mice die at E7.5 (Coffinieret al., 1999) and the role of Hnf1b in mammalian hindbrain and neuralcrest development has not been studied. Hnf1b is, however, known toplay a role in hindbrain development in zebrafish (Choe et al., 2008),where loss of Hnf1b function results in complete absence of Krox20expression in r5. This is reminiscent ofHoxa1 knockoutmice, where r5is absent and the second stripe of Krox20 expression, which normallymarks this rhombomere, is missing (Lufkin et al., 1991). Analysis ofcis-regulatory sequences governing Krox20 expression identified aconserved enhancer containing a binding site for the Hnf1b transcrip-tion factor, which is necessary for the initiation of Krox20 expression(Chomette et al., 2006). Therefore, our findings suggest that Hoxa1acts upstream of Hnf1b in the initiation of Krox20 expression in r5(Fig. 5).

Interestingly, our in situ analysis revealed that Foxd3, Hnf1b andZic1 are strongly reduced in the posterior hindbrain (r6–r8) of Hoxa1null embryos. This region of the hindbrain is not mispatterned inHoxa1 mutants and was thought to be unaffected by loss of Hoxa1function. The posterior hindbrain gives rise to cardiac neural crestcells, which are important for remodeling of the cardiac outflow tractwhich is affected in humans with mutations in HoxA1. Therefore,reduction of Foxd3, Zic1 and Hnf1b, three neural crest markers, in theposterior hindbrain ofHoxa1 null mice suggests thatHoxa1might playa direct role in cardiac neural crest development and that this could bethe reason for the outflow tract defects in HoxA1-syndrome patients.

Two other confirmed genes, Pax8 and Fgfr3, are known to beimportant for inner ear development. Pax8 is expressed in the oticplacode starting at the pre-somite stage (Ohyama and Groves, 2004)and plays a role in otic placode induction and specification (Mackerethet al., 2005). It is interesting to find changes in Pax8 expression as earlyas the 4 somite stage, since gene expression profiles have not beenanalyzed in theotic placodeofHoxa1mutants prior toE9.25 (~20 somitestage),whenmorphological changes have already occurred (Pasqualettiet al., 2001). This suggests thatHoxa1 affects inner ear development at avery early stage, presumably during otic placode specification andmight, therefore, play a direct role in inner ear development. The onlyvalidated Hoxa1 downstream target that was up-regulated in Hoxa1mutants was Fgfr3. Expression of Fgfr3was found to be expanded in the

hindbrain of Hoxa1 null embryos, extending from its normal border atthe r5/r6 boundary anteriorly into r4. Fgf signaling in several tissues,including the hindbrain, is known to influence inner ear development(Zelarayanet al., 2007) and itwas shown that activating Fgfr3mutationscan cause hearing loss and inner ear defects in humans and mice(Mansour et al., 2009, Pannier et al., 2009). Since Hoxa1 is stronglyexpressed in r4, it is possible that it acts as an inhibitor of Fgfr3 in thehindbrain and that release of this inhibition leads to ectopic activation ofFgfr3, whichmight contribute to the inner eardefects inHoxa1nullmice.

Finally, Lhx5was identified as a novel downstream target of Hoxa1.Lhx5 expression in the hindbrain has previously been reported atE10.5 (Gray et al., 2004). Our in situ and qPCR data now show thatLhx5 is already expressed as early as E8.25 (6 somite stage).Interestingly, Lhx5 has been implicated in the determination ofreticulospinal neuron identity at E12.5 (Cepeda-Nieto et al., 2005).These neurons are involved in modulation of respiration andcardiovascular function both of which are affected by loss of Hoxa1.It is, therefore, possible that Hoxa1 acts upstream of Lhx5 in thedevelopment of hindbrain reticulospinal neuron precursors.

In conclusion, we identified Hnf1b, Foxd3 and Zic1 as Hoxa1downstream targets which are involved in hindbrain and early neuralcrest development. Interestingly, these markers were reduced in theposterior hindbrain, where cardiac neural crest cells originate suggest-ing that Hoxa1 might play a role in the development of these cells.Additionally, we identified changes in the expression patterns of Pax8and Fgfr3, two genes important for inner ear development, whichindicates that Hoxa1 affects otic placode specification. Whether it doesso directly or through signaling from the hindbrain remains to beshown. Finally, Lhx5, a gene expressed in hindbrain reticulospinalneuron precursors, was down-regulated in Hoxa1 mutants raising thepossibility thatHoxa1 acts upstreamof Lhx5 in the development of theseneurons (Fig. 5). Although our experiments do not allow us to concludeif the identified six genes are direct or indirect targets ofHoxa1, they arelikely to play important regulatory roles in the development of thetissues affected by loss of Hoxa1.

In addition to identifying effectors of Hoxa1 in neural crest, innerear and hindbrain development, our array provides a long list of novelpotential targets involved in other developmental and cellularprocesses such as cardiac and vascular development or neuron andmuscle cell differentiation (Fig. 3C), which will be the ground forfuture investigations.

303N. Makki, M.R. Capecchi / Developmental Biology 357 (2011) 295–304

Comparison of Hoxa1 microarray results to other published microarrayexperiments

Ten microarrays have been published which identified Hoxdownstream targets in themouse (reviewed by Hueber and Lohmann,2008). Of these, six have been carried out on mouse tissue thatexpresses the gene of interest, whereas the other four, including twoHoxa1 microarrays (Martinez-Ceballos et al., 2005; Shen et al., 2000),have been performed on cultured cell lines. We compared the list ofgenes identified in our microarray with the lists of the two publishedHoxa1 microarrays performed on cultured cells. None of the 28putative downstream effectors identified in the differential hybridi-zation screening of teratocarcinoma cells overexpressing Hoxa1 (Shenet al., 2000) were found in our microarray. In the second Hoxa1microarray study, which compared gene expression profiles of wildtype and Hoxa1−/− embryonic stem cells treated with retinoic acid,145 targets were identified (Martinez-Ceballos et al., 2005). Only 45of these targets are available online and again none of them wereidentified in our experiment. This is not surprising, since our micro-array and the previously published ones constitute very differentexperiments. The two previous microarrays identified Hoxa1 targetsin embryonic stem or cancer cells. Our study now adds a valuable newlist of downstream targets, which are controlled by Hoxa1 in thedeveloping embryo.

Microarrays were also performed on Hoxb1 (Tvrdik and Capecchi,2006), the paralog of Hoxa1 in mice and its ortholog hoxb1a inzebrafish (Rohrschneider et al., 2007). Since Hoxa1 and Hoxb1 areparalogous members and may share some downstream targets, wecompared our Hoxa1 dataset with the datasets from the above Hoxb1studies. The following genes were differentially expressed in both ourHoxa1 microarray as well as either the mouse Hoxb1 or zebrafishhoxb1a microarray and might represent common targets of the twogenes: Zinc finger protein of the cerebellum 1 (Zic1), delta/notch-likeEGF-related receptor (Dner), nephronectin (Npnt), transthyretin(Ttr), Sjogren syndrome antigen B (Ssb), Nik related kinase (Nrk),DEAD box polypeptide 3 (Ddx3y), leucine rich repeat containing 4(Lrrc4).

Since Hoxa1 is of profound importance to the development of avariety of tissues, analysis of some of the targets on our list allowed usto propose a model for how Hoxa1 might regulate specific aspects ofhindbrain, inner ear and neural crest development. Further investi-gation into the molecular mechanisms through which Hoxa1 orches-trates the development of these tissues will be necessary to betterunderstand the origin of the defects in HoxA1-syndrome patients. Webelieve that this study might provide a first stepping stone in thisdirection.

Supplementarymaterials related to this article can be found onlineat doi:10.1016/j.ydbio.2011.06.042.

Acknowledgments

We thank Petr Tvrdik for generating the Hoxa1c targeting vectorand members of our tissue culture and mouse facility, in particularSheila Barnett, Carol Lenz and Karl Lustig for ES cell culture, injectionand mouse care; Lisa Urness and Yukio Saijoh for their experimentalinput and advice, as well as sharing protocols and reagents; RuthArkell, Andy Groves, Patricia Labosky, XiaojingMa and Katherine Shimfor plasmids used to generate riboprobes and Brett Milash and BrianDalley from the University of Utah Microarray Core for assistance inanalyzing the microarray experiment. This manuscript was improvedby helpful comments from Anne Boulet and Daniel Kopinke. This workwas supported by grants from the NIH (NIH5R01GM021168-34) andHoward Hughes Medical Institute to M.R.C. and the BoehringerIngelheim Fonds PhD fellowship and the University of Utah GraduateResearch Fellowship to N.M.

References

Alexander, T., Nolte, C., Krumlauf, R., 2009. Hox genes and segmentation of thehindbrain and axial skeleton. Annu. Rev. Cell Dev. Biol. 25, 431–456.

Aruga, J., 2004. The role of Zic genes in neural development. Mol. Cell. Neurosci. 26,205–221.

Barbacci, E., Reber, M., Ott, M.O., Breillat, C., Huetz, F., Cereghini, S., 1999. Varianthepatocyte nuclear factor 1 is required for visceral endoderm specification.Development 126, 4795–4805.

Barlow, L.A., 2002. Cranial nerve development: placodal neurons ride the crest. Curr.Biol. 12, R171–R173.

Barrow, J.R., Stadler, H.S., Capecchi, M.R., 2000. Roles of Hoxa1 and Hoxa2 in patterningthe early hindbrain of the mouse. Development 127, 933–944.

Breitling, R., Armengaud, P., Amtmann, A., Herzyk, P., 2004. Rank products: a simple, yetpowerful, new method to detect differentially regulated genes in replicatedmicroarray experiments. FEBS Lett. 573, 83–92.

Brown, C.B., Baldwin, H.S., 2006. Neural crest contribution to the cardiovascular system.Adv. Exp. Med. Biol. 589, 134–154.

Capecchi, M.R., 1997. Hox genes and mammalian development. Cold Spring Harb.Symp. Quant. Biol. 62, 273–281.

Cepeda-Nieto, A.C., Pfaff, S.L., Varela-Echavarria, A., 2005. Homeodomain transcriptionfactors in the development of subsets of hindbrain reticulospinal neurons. Mol. Cell.Neurosci. 28, 30–41.

Chisaka, O., Musci, T.S., Capecchi, M.R., 1992. Developmental defects of the ear, cranialnerves and hindbrain resulting from targeted disruption of the mouse homeoboxgene Hox-1.6. Nature 355, 516–520.

Choe, S.K., Hirsch, N., Zhang, X., Sagerstrom, C.G., 2008. hnf1b genes in zebrafishhindbrain development. Zebrafish 5, 179–187.

Chomette, D., Frain, M., Cereghini, S., Charnay, P., Ghislain, J., 2006. Krox20 hindbraincis-regulatory landscape: interplay between multiple long-range initiation andautoregulatory elements. Development 133, 1253–1262.

Coffinier, C., Thepot, D., Babinet, C., Yaniv, M., Barra, J., 1999. Essential role for thehomeoprotein vHNF1/HNF1beta in visceral endoderm differentiation. Develop-ment 126, 4785–4794.

Dennis Jr., G., Sherman, B.T., Hosack, D.A., Yang, J., Gao, W., Lane, H.C., Lempicki, R.A.,2003. DAVID: database for annotation, visualization, and integrated discovery.Genome Biol. 4, P3.

Dottori, M., Gross, M.K., Labosky, P., Goulding, M., 2001. The winged-helix transcriptionfactor Foxd3 suppresses interneuron differentiation and promotes neural crest cellfate. Development 128, 4127–4138.

Dudoit, S., Gentleman, R.C., Quackenbush, J., 2003. Open source software for theanalysis of microarray data. Biotechniques, pp. 45–51. Suppl.

Elms, P., Scurry, A., Davies, J., Willoughby, C., Hacker, T., Bogani, D., Arkell, R., 2004.Overlapping and distinct expression domains of Zic2 and Zic3 during mousegastrulation. Gene Expr. Patterns 4, 505–511.

Gaston-Massuet, C., Henderson, D.J., Greene, N.D., Copp, A.J., 2005. Zic4, a zinc-fingertranscription factor, is expressed in the developing mouse nervous system. Dev.Dyn. 233, 1110–1115.

Gray, P.A., Fu, H., Luo, P., Zhao, Q., Yu, J., Ferrari, A., Tenzen, T., Yuk, D.I., Tsung, E.F., Cai, Z.,Alberta, J.A., Cheng, L.P., Liu, Y., Stenman, J.M., Valerius, M.T., Billings, N., Kim, H.A.,Greenberg, M.E., McMahon, A.P., Rowitch, D.H., Stiles, C.D., Ma, Q., 2004. Mousebrain organization revealed through direct genome-scale TF expression analysis.Science 306, 2255–2257.

Haumaitre, C., Barbacci, E., Jenny, M., Ott, M.O., Gradwohl, G., Cereghini, S., 2005. Lack ofTCF2/vHNF1 in mice leads to pancreas agenesis. Proc. Natl. Acad. Sci. U. S. A. 102,1490–1495.

Henrique, D., Adam, J., Myat, A., Chitnis, A., Lewis, J., Ish-Horowicz, D., 1995. Expressionof a Delta homologue in prospective neurons in the chick. Nature 375, 787–790.

Holve, S., Friedman, B., Hoyme, H.E., Tarby, T.J., Johnstone, S.J., Erickson, R.P., Clericuzio,C.L., Cunniff, C., 2003. Athabascan brainstem dysgenesis syndrome. Am. J. Med.Genet. A 120, 169–173.

Huang da,W., Sherman, B.T., Lempicki, R.A., 2009. Systematic and integrative analysis oflarge gene lists using DAVID bioinformatics resources. Nat. Protoc. 4, 44–57.

Hueber, S.D., Lohmann, I., 2008. Shaping segments: Hox gene function in the genomicage. Bioessays 30, 965–979.

Labosky, P.A., Kaestner, K.H., 1998. The winged helix transcription factor Hfh2 isexpressed in neural crest and spinal cord during mouse development. Mech. Dev.76, 185–190.

Li, B., Kuriyama, S., Moreno, M., Mayor, R., 2009. The posteriorizing gene Gbx2 is a directtarget of Wnt signalling and the earliest factor in neural crest induction.Development 136, 3267–3278.

Lufkin, T., Dierich, A., LeMeur, M., Mark, M., Chambon, P., 1991. Disruption of the Hox-1.6 homeobox gene results in defects in a region corresponding to its rostraldomain of expression. Cell 66, 1105–1119.

Lumsden, A., Krumlauf, R., 1996. Patterning the vertebrate neuraxis. Science 274,1109–1115.

Mackereth, M.D., Kwak, S.J., Fritz, A., Riley, B.B., 2005. Zebrafish pax8 is required for oticplacode induction and plays a redundant role with Pax2 genes in the maintenanceof the otic placode. Development 132, 371–382.

Makki, N., Capecchi, M.R., 2010. Hoxa1 lineage tracing indicates a direct role for Hoxa1in the development of the inner ear, the heart, and the third rhombomere. Dev. Biol.341, 499–509.

Mallo, M., Wellik, D. M., and Deschamps, J. Hox genes and regional patterning of thevertebrate body plan. Dev Biol.

Mansour, S.L., Twigg, S.R., Freeland, R.M., Wall, S.A., Li, C., Wilkie, A.O., 2009. Hearingloss in a mouse model of Muenke syndrome. Hum Mol Genet 18, 43–50.

304 N. Makki, M.R. Capecchi / Developmental Biology 357 (2011) 295–304

Mark, M., Lufkin, T., Vonesch, J.L., Ruberte, E., Olivo, J.C., Dolle, P., Gorry, P., Lumsden, A.,Chambon, P., 1993. Two rhombomeres are altered in Hoxa-1 mutant mice.Development 119, 319–338.

Martinez-Ceballos, E., Chambon, P., Gudas, L.J., 2005. Differences in gene expressionbetween wild type and Hoxa1 knockout embryonic stem cells after retinoic acidtreatment or leukemia inhibitory factor (LIF) removal. J. Biol. Chem. 280,16484–16498.

Mechta-Grigoriou, F., Garel, S., Charnay, P., 2000. Nab proteins mediate a negativefeedback loop controlling Krox-20 activity in the developing hindbrain. Develop-ment 127, 119–128.

Meno, C., Saijoh, Y., Fujii, H., Ikeda, M., Yokoyama, T., Yokoyama, M., Toyoda, Y., Hamada,H., 1996. Left-right asymmetric expression of the TGF beta-family member lefty inmouse embryos. Nature 381, 151–155.

Merzdorf, C.S., 2007. Emerging roles for zic genes in early development. Dev. Dyn. 236,922–940.

Minowada, G., Jarvis, L.A., Chi, C.L., Neubuser, A., Sun, X., Hacohen, N., Krasnow, M.A.,Martin, G.R., 1999. Vertebrate Sprouty genes are induced by FGF signaling and cancause chondrodysplasia when overexpressed. Development 126, 4465–4475.

Morsli, H., Choo, D., Ryan, A., Johnson, R., Wu, D.K., 1998. Development of the mouseinner ear and origin of its sensory organs. J. Neurosci. 18, 3327–3335.

Murphy, P., Hill, R.E., 1991. Expression of the mouse labial-like homeobox-containinggenes, Hox 2.9 and Hox 1.6, during segmentation of the hindbrain. Development111, 61–74.

Nagai, T., Aruga, J., Takada, S., Gunther, T., Sporle, R., Schughart, K., Mikoshiba, K., 1997.The expression of the mouse Zic1, Zic2, and Zic3 gene suggests an essential role forZic genes in body pattern formation. Dev. Biol. 182, 299–313.

Ohyama, T., Groves, A.K., 2004. Generation of Pax2-Cre mice by modification of a Pax2bacterial artificial chromosome. Genesis 38, 195–199.

Pannier, S., Couloigner, V., Messaddeq, N., Elmaleh-Berges, M., Munnich, A., Romand, R.,Legeai-Mallet, L., 2009. Activating Fgfr3 Y367C mutation causes hearing loss andinner ear defect in a mouse model of chondrodysplasia. Biochim. Biophys. Acta1792, 140–147.

Pasqualetti, M., Neun, R., Davenne, M., Rijli, F.M., 2001. Retinoic acid rescues inner eardefects in Hoxa1 deficient mice. Nat. Genet. 29, 34–39.

Rodriguez, C.I., Buchholz, F., Galloway, J., Sequerra, R., Kasper, J., Ayala, R., Stewart, A.F.,Dymecki, S.M., 2000. High-efficiency deleter mice show that FLPe is an alternativeto Cre-loxP. Nat. Genet. 25, 139–140.

Rohrschneider, M.R., Elsen, G.E., Prince, V.E., 2007. Zebrafish Hoxb1a regulates multipledownstream genes including prickle1b. Dev. Biol. 309, 358–372.

Schmittgen, T.D., Livak, K.J., 2008. Analyzing real-time PCR data by the comparative C(T)method. Nat. Protoc. 3, 1101–1108.

Schroeder, A., Mueller, O., Stocker, S., Salowsky, R., Leiber, M., Gassmann, M., Lightfoot,S., Menzel, W., Granzow, M., Ragg, T., 2006. The RIN: an RNA integrity number forassigning integrity values to RNA measurements. BMC Mol. Biol. 7, 3.

Shen, J., Wu, H., Gudas, L.J., 2000. Molecular cloning and analysis of a group of genesdifferentially expressed in cells which overexpress the Hoxa-1 homeobox gene.Exp. Cell Res. 259, 274–283.

Tam, P.P., 1981. The control of somitogenesis in mouse embryos. J. Embryol. Exp.Morphol. 65, 103–128 Suppl.

Tang, S.H., Silva, F.J., Tsark, W.M., Mann, J.R., 2002. A Cre/loxP-deleter transgenic line inmouse strain 129S1/SvImJ. Genesis 32, 199–202.

Teng, L., Mundell, N.A., Frist, A.Y., Wang, Q., Labosky, P.A., 2008. Requirement for Foxd3in the maintenance of neural crest progenitors. Development 135, 1615–1624.

Tischfield, M.A., Bosley, T.M., Salih, M.A., Alorainy, I.A., Sener, E.C., Nester, M.J., Oystreck,D.T., Chan, W.M., Andrews, C., Erickson, R.P., Engle, E.C., 2005. Homozygous HOXA1mutations disrupt human brainstem, inner ear, cardiovascular and cognitivedevelopment. Nat. Genet. 37, 1035–1037.

Tvrdik, P., Capecchi, M.R., 2006. Reversal of Hox1 gene subfunctionalization in themouse. Dev. Cell 11, 239–250.

Wang, X., Seed, B., 2003. A PCR primer bank for quantitative gene expression analysis.Nucleic Acids Res. 31, e154.

Zelarayan, L.C., Vendrell, V., Alvarez, Y., Dominguez-Frutos, E., Theil, T., Alonso, M.T.,Maconochie, M., Schimmang, T., 2007. Differential requirements for FGF3, FGF8 andFGF10 during inner ear development. Dev. Biol. 308, 379–391.