Hindawi BioMed Research International Volume 2017, Article ID 8932583, 16 pages https://doi.org/10.1155/2017/8932583

Research Article An Integrative Developmental Genomics and Systems Biology Approach to Identify an In Vivo Sox Trio-Mediated Regulatory Network in Murine Embryos

Wenqing Jean Lee,1 Sumantra Chatterjee,2 Sook Peng Yap,3 Siew Lan Lim,1 Xing Xing,1 Petra Kraus,4 Wenjie Sun,1 Xiaoming Hu,1 V. Sivakamasundari,5 Hsiao Yun Chan,1 Prasanna R. Kolatkar,6 Shyam Prabhakar,1 and Thomas Lufkin4

1 Genome Institute of Singapore, 60 Biopolis Street, Singapore 138672 2McKusick-Nathans Institute of Genetic Medicine, Johns Hopkins University School of Medicine, 733 N. Broadway, Baltimore, MD 21205, USA 3Apta Biosciences Pte Ltd, 31 Biopolis Way, Singapore 138669 4Department of Biology, Clarkson University, 8 Clarkson Avenue, Potsdam, NY 13699, USA 5The Jackson Laboratory for Genomic Medicine, 10 Discovery Drive, Farmington, CT 06030, USA 6Qatar Biomedical Research Institute, Hamad Bin Khalifa University, Qatar Foundation, P.O. Box 5825, Doha, Qatar

Correspondence should be addressed to Thomas Lufkin; [email protected]

Received 16 January 2017; Accepted 16 April 2017; Published 28 May 2017

Academic Editor: Ruijin Huang

Copyright © 2017 Wenqing Jean Lee et al. This is an open access article distributed under the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.

Embryogenesis is an intricate process involving multiple and pathways. Some of the key transcription factors controlling specific cell types are the Sox , namely, Sox5, Sox6,andSox9, which play crucial roles in organogenesis working in a concerted manner. Much however still needs to be learned about their combinatorial roles during this process. A developmental genomics and systems biology approach offers to complement the reductionist methodology of current developmental biology and provide a more comprehensive and integrated view of the interrelationships of complex regulatory networks that occur during organogenesis. By combining cell type-specific transcriptome analysis and in vivo ChIP-Seq of the Sox trio using mouse embryos, we provide evidence for the direct control of Sox5 and Sox6 by the transcriptional trio in the murine model and by Morpholino knockdown in zebrafish and demonstrate the novel role of Tgfb2, Fbxl18,andTle3 in formation of Sox5, Sox6,andSox9 dependent tissues. Concurrently, a complete embryonic gene regulatory network has been generated, identifying a wide repertoire of genes involved and controlled by the Sox trio in the intricate process of normal embryogenesis.

1. Introduction complex system, fulfilling critical functions such as move- ment, hematopoiesis, and the protection of vital organs. The TheSoxfamilyofproteinsthatareencodedbyatleast20 incidence of skeletal diseases is estimated to be 1 in 4000 genesinmammalsplaysamyriadofrolesduringembryonic births with half of them being early lethal but the real fre- cell type specification and organogenesis [1–8]. Three of quency is probably higher as some disorders develop later in the Sox genes, Sox5, Sox6,andSox9, while having unique life [18]. Genes controlling skeletogenesis are often involved roles in specific cell types, together play a concerted role in other developmental processes, resulting in complex syn- during embryonic skeletogenesis, being absolutely neces- dromes with skeletal disorders being one of the outcomes. sary for proper chondrogenesis [9–17]. The skeleton is a To develop greatly needed therapies for human diseases, it 2 BioMed Research International is necessary to understand the molecular basis behind them. molecular mechanisms behind postmesenchymal condensa- Deciphering the fundamentals of skeletogenesis will aid in tion processes in chondrogenesis. the identification of the molecular mechanisms behind the prevalent skeletal diseases. 2. Materials and Methods Chondrocytes are the first skeleton-specific cell type to appear during development and defects in chondrogenesis; 2.1. Generation of Transgenic Mice. BAC clones, RP24- an essential event during endochondral ossification (respon- 248D4, RP23-82L9, and RP23-403L18 containing regions of sible for the formation of most of the skeleton) and post- the genome for Sox9, Sox5,andSox6,respectively,were natal cartilage maintenance lead to chondrodysplasias and ordered from BACPAC, Resources Centre, CHORI, Oakland, osteoarthritis [19, 20]. The commitment of progenitor cells to USA (http://bacpac.chori.org) [35, 36]. BAC modification the chondrogenic fate is largely determined by the expression to create the transgenes was done using the Quick and of the , Sox9 [21]. Heterozygous mutations Easy BAC Modification Kit (Gene Bridges) according to in and around Sox9 were shown to cause campomelic dyspla- the manufacturer’s protocol. The modified BACS were then sia, a severe form of human chondrodysplasia often accom- subcloned to the pBSSK+ (Stratagene) or minimal vector panied by male sex reversal and defects in other nonskeletal (Gene Bridges) as targeting vectors for homologous recom- organs, highlighting its critical role in chondrogenesis and bination in ESC. All vectors used were verified by restriction other tissues [22, 23]. Early chondrocyte differentiation and enzyme mapping and sequencing [37, 38]. Correctly modified subsequent maturation are controlled by Sox9 and its family ESC lines were microinjected into 2- to 8-cell stage mouse members, Sox5 and Sox6 [24, 25]. The expression of the embryos to generate chimeric embryos and chimeric mice trio in nonchondrogenic cells has been shown to induce that were subsequently used to generate mouse colonies cartilage formation [26]. Sox5 and Sox6 which are highly [39]. The Neo cassettes in the transgene were removed by related to each other are expressed downstream of Sox9 crossing with mice expressing Flpe recombinase driven by [27]. They exhibit functional redundancy for each other the ROSA26 promoter (Stock number 3946, The Jackson during chondrogenesis with the double-gene knockout mice Lab) or Cre recombinase driving by the Zp3 promoter as displaying severe chondrodysplasia as compared to the single previously described [40, 41]. All animal procedures were knockouts, which showed mild skeletal defects [28]. Recent conducted according to IACUC guidelines. Further details molecular studies done in vitro have shown that the trio can be found in Supplementary Material available online at synergistically activates the expression of cartilage-specific https://doi.org/10.1155/2017/8932583. genes such as Col2a1, Acan,andMatn1 [29, 30]. 2.2. Ethical Statement. All animal procedures were per- ∗ The genetic and molecular events underlying a biological formed according to the Singapore A STAR Biopolis Bio- process such as chondrogenesis are often complex and the logical Resource Center (BRC) Institutional Animal Care complete understanding of the gene regulatory network and Use Committee (IACUC) guidelines which are set by (GRN) involved requires a comprehensive developmental the National Advisory Committee for Laboratory Animal genomics and systems biology approach [31]. Sox9, Sox5, Research (NACLAR) for the ethical treatment of animals. The and Sox6 were previously found to be the major transcrip- IACUC protocols employed were reviewed and approved by tion factors regulating chondrogenesis [27]. The targets of the aforementioned committee before any animal procedures the Sox trio in chondrogenesis uncovered by conventional were undertaken for this study described here (IACUC experimental approaches previously mentioned are probably Protocol numbers 110689 and 110648). The mouse strains only a portion of the network involved and only partially used in this study were provided, housed, and maintained in ∗ recapitulate what is happening in vivo. In order to have a IVC cages by the A STAR Biopolis Biological Resource Cen- comprehensive view of what is occurring during skeleton ter following the aforementioned guidelines for the ethical development, classical mouse genetics in combination with treatment of animals. expression profiling and chromatin immunoprecipitation- sequencing (ChIP-Seq) were carried out for Sox9, Sox5,and 2.3. Gene Targeting in ESC. V6.4 and R1 mouse ESC were ∘ Sox6 using E13.5 mouse embryos, a stage representing the used for gene targeting. Briefly, they were cultured at 37 C second stage of chondrocyte differentiation preceding the with 5% CO2 on gelatin-coated feeder plates with culture transition of chondrocytes into hypertrophic chondrocytes medium made from DMEM (Gibco) supplemented with 15% [27, 32]. Though the roles of these three have been heat-inactivated ES-grade fetal bovine serum (FBS) (Gibco), studied at E13.5 previously with their individual targets [27, 500 U/ml LIF (Chemicon), 0.1 mM 𝛽-mercaptoethanol 28, 33, 34], there is presently no comprehensive systems (Invitrogen), 4 mM L-glutamine (Invitrogen), 1 mM sodium biologyinvivogenelistestablishedfortheSoxtrioatthis pyruvate (Invitrogen), and 40 𝜇g/ml gentamicin (Sigma) stage. Here we report the identification of direct and indirect [42, 43]. 10–15 𝜇g of the linearized vector was used for targets of the transcriptional Sox trio in vivo, generating an each electroporation at 125 𝜇F, 0.4kV into approximately 10 extensive GRN at E13.5, separating cohorts of genes regulated million ESC. Individual G418-resistant ESC colonies were by Sox9 alone or together with Sox5 and/or Sox6. We have picked after 10 days of selection with 200–400 𝜇g/ml G418 also verified Tgfb2, Tle3,andFbxl18 as novel chondrogenic and screened by Southern blot analysis. targets regulated by the Sox trio. The modes of action of the Sox trio and the discovery of their additional target genes in 2.4. Genotyping. Mouse ESC, tail biopsies, and embryonic this valuable dataset have increased the understanding of the tissues were digested with Proteinase K (Sigma). Genomic BioMed Research International 3

DNA was extracted using phenol-chloroform followed by [46, 47]. RNA SISH was carried out with 10 𝜇m paraffin- ethanol precipitation and used for genotyping [44]. All mod- embedded sections as previously described [40, 48]. The ified ESC and founder mice from the transgenic lines were following cDNAs were used as templates for synthesiz- confirmed with Southern blot analysis. Southern blot analysis ing the antisense probes: 0.5 kb Col2a1 [46]; 0.7 kb Matn4 wascarriedoutaspreviouslydescribed[40]using10–20𝜇g (IMAGE Clone ID: 1366191); 4.7 kb Hapln1 (IMAGE Clone of genomic DNA digested with the indicated restriction ID: 30430221); 0.5 kb 3110079O15Rik (IMAGE Clone ID: enzyme. Routine genotyping was done by PCR [45]. 40104070); 0.6 kb Sox5 (IMAGE Clone ID: 40047865); and 0.5 kb Sox9 (Sox9 exon 1 sequence from BAC clone RP24- 2.5. FACS. E13.5 mouse embryos were harvested in ice- 248D4). Alcian Blue staining was performed as described cold Leibovitz’s L-15 (Gibco). EGFP-expressing embryos were [46, 47, 49]. Crossing of the Col2a1-Cre transgenic line (Stock identified using a fluorescent dissection microscope (Leica) number 3554, The Jackson Lab) with the conditional Sox9 and dissected such that all internal organs were removed, knockout line was performed as described [50]. All sections leaving the axial and appendicular skeleton intact. Tissue were photographed using Zeiss Axio Imager Z1. dissociation into single cells was carried out by mechanical pipetting using an enzymatic solution comprising 100 U/ml 2.9. ChIP-Seq and Peak Calling. Tissue from limbs and tails of E13.5 CD1 wildtype embryos were used for Sox6 and Sox9 Collagenase I and Collagenase II (Invitrogen), 50 U/ml 𝐻𝐴/𝐻𝐴 DNase I (Invitrogen), and 0.05% Trypsin (Invitrogen) in ChIP. The same tissues were dissected from E13.5 𝑆𝑜𝑥5 Leibovitz’s L15. The enzymes were stopped by the addition of embryos for the Sox5 ChIP. Crosslinking, chromatin isola- 20% FBS (Gibco) in Leibovitz’s L15. The dissociated cells were tion, sonication, and immunoprecipitation were carried out filtered through a 100 𝜇Mfollowedbya40𝜇Mcellstrainer. as previously described [51]. Preclearing was done with rabbit Cells were pelleted down and resuspended in 5% FBS, 4 mM or goat IgG (sc-2027 and sc-2028, Santa Cruz). Immunopre- EDTA in Leibovitz’s L15 for sorting using a 4-laser FACSAria cipitation was done using an anti-Sox9 antibody (AF3075, Cell Sorter (BD Biosciences). R&D Systems), anti-Sox6 antibody (ab30455, Abcam), and anti-HA antibody (AP09230PU-N, Acris). All antibodies 2.6. Microarrays. Total RNA was isolated from the sorted employed have been shown to be specific for their specific cells using the RNeasy Micro Kit (Qiagen). The RNA target and not to cross-react with any other protein extracted from all the samples was quantified using Ribo- (e.g., Figure S1). 10–15 ng of purified ChIP DNA from each green (Invitrogen) and their RNA integrity was assessed to sample was used to synthesize the sequencing library as ensure the high quality using the Bioanalyzer (Agilent Tech- instructed by the ChIP-Seq DNA sample Prep Kit (Illumina). nologies). For the Sox9 microarray, 50 ng of total RNA per The libraries were then subjected to the Solexa sequencing sample was labeled using the Illumina TotalPrep RNA Ampli- according to Illumina’s instruction. Sequence reads produced ficationKit(Ambion)andhybridizedontheMouseWG- by the Illumina Genome Analyzer II/IIx that passed the signal 6 v1.1 Expression BeadChip (Illumina). For the Sox5/Sox6 purity filtering were mapped to the mouse genome mm9, microarray, 25 ng of total RNA per sample was labeled using using the Illumina Genome Analyzer Pipeline. All uniquely the TargetAmp-Nano Labeling Kit for Illumina Expression mapped reads that are with two or fewer mismatches were BeadChips (Epicentre) and hybridized on the MouseWG-6 retained. Genomic binding sites in the ChIP-Seq datasets v2.0 Expression BeadChip (Illumina). Microarray hybridiza- were identified using the peak calling algorithm MACS (version 1.4.0 beta) with default settings (band width = 300, tion was carried out according to Illumina’s guidelines. The 𝑝 1.00𝑒 − 05 microarray data can be found in GEO under the accession model fold = 10, 30, value cutoff = ,andrange number GSE33173. for calculating regional lambda = 1000 and 10000 bps) [52]. The corresponding control libraries were used for all the peak callings. The datasets can be found at GEO under the 2.7. Microarray Data Analysis. Raw data was extracted with accession number GSE33419. background subtraction employed using Genome Studio (Illumina). Further analysis was done with Genespring 11 2.10. Motif Analysis. Briefly, binding sites within 1 kb up- (Agilent). log2 transformation and percentile shift normal- stream and downstream of the transcription start site (TSS) 𝑝 ization at 75th percentile were performed. Default detection of the gene were considered in the TSS region. Binding sites valuerangeassuggestedbythesoftwareforpresent,marginal, 1–5 kb upstream of the TSS were considered to be in the and absent flags was used. Entities with present and marginal promoter region. Intragenic binding sites were defined as flags were further filtered by the percentile of their expression beingwithinthegenebutoutoftheTSSandpromoter using the default settings before statistical analysis. Unpaired region. Proximal binding sites were binding sites out of the 𝑡 -test with the Benjamini-Hochberg multiple test correction genebutwithin5–10kbupstreamoftheTSSand1–10kb was performed for the Sox9 microarray and the Sox5/6 downstream of the TSS. Distal binding sites were defined as 𝑝 microarray, respectively. A corrected value cutoff of 0.05 10–100 kb upstream and downstream of the TSS and out of with a minimum fold change of 1.5 was used to find the the gene. Other binding sites found more than 100 kb away differentially expressed probes. fromtheTSSandoutofthegenewereclassifiedasothers. Peaks called by MACS were ranked according to the total ∘ 2.8. Histology. Mouse embryos were fixed overnight at 4 Cin tagscountasdefinedintheMACSoutputfile.Thetop200 4% paraformaldehyde and processed as previously described peaks were used for motif analysis and the repeat masked 4 BioMed Research International genome sequences +/−50bpfromthesummitofthese200 3. Results peaks were downloaded from the UCSC genome browser (http://genome.ucsc.edu/). After masking repeats to 𝑁,we 3.1. In Vivo Identification of the Genes Involved in the Sox Trio- performed de novo motif finding using MEME ver. 4.3.0 with Associated Functions. To isolate the Sox9-expressing cells the sequences. MAST was used to scan for the occurrences of directly from developing mouse embryos for in vivo com- the primary de novo motif obtained using all the sequences parative expression profiling, the endogenous Sox9 was +/−50 bp of the ChIP-Seq peak summit. The cutoff for motif targeted with the enhanced green fluorescent protein (EGFP) 𝑝 = 1.0𝑒 −4 reporter gene via homologous recombination, generating matchinMASTusedwasthedefault . +/+ −/− Sox9-wildtype (𝑆𝑜𝑥9 (EGFP)) and Sox9-null (𝑆𝑜𝑥9 (EGFP)) 2.11. Zebrafish Assays. Fish were maintained in the GIS gene targeted lines (Figures 1(a)–1(c)) for fluorescence- +/+ zebrafish facility according to the standard of IACUC guide- activated cell sorting (FACS). 𝑆𝑜𝑥9 (EGFP) was made with lines. The stages of embryos were indicated as hpf (hours the reporter linked via the foot-and-mouth disease virus 2A 󸀠 after fertilization) or dpf (days after fertilization) [53]. Mor- peptide (F2A) at the 3 end of the Sox9 locus (Figure 1(b)), pholinos were injected into about 200 1-cell stage embryos forming a bicistronic system with a single open reading +/+ at a concentration of 1.2 picomolar. To reduce cell death due frame[45,57].The𝑆𝑜𝑥9 (EGFP) mice generated were viable to off-target effects 1.8 picomolar of Morpholino was and fertile, indicating that the gene targeted mice are nor- coinjected as previously described [54]. Embryos were main- mal given the lethality of the Sox9 heterozygous mutation. ∘ +/+ tained in a 28 C incubator overnight and dead embryos were Western blot analysis using E13.5 𝑆𝑜𝑥9 (EGFP) and E13.5 removed the next morning before scoring for morphants. wildtype embryos verified that the Sox9 and EGFP proteins Alcian Blue staining was performed on 24 hpf wildtype were expressed individually and Sox9 protein levels in our +/+ and morphant embryos and embryos were sectioned in a 𝑆𝑜𝑥9 (EGFP) weresimilartothatofthewildtype(Figures transverse orientation as previously described [38, 54]. All S1). Sox9 heterozygotes mice die at birth and thus cannot sections were imaged with the Zeiss Axio Imager. be intermated to generate Sox9 loss-of-function animals. −/− To overcome this problem, 𝑆𝑜𝑥9 (EGFP) embryonic stem 2.12. Luciferase Assay. Using an oligo cloning strategy, the cells (ESC) were generated by sequential inactivation of binding site sequences were cloned between the BamHI each Sox9 allele in ESC via gene targeting and used to and XhoI sites of the pGL4.23 luciferase vector (Promega 4 generate high-percentage chimeras for FACS (Figure 1(c)). Corporation, USA). NIH/3T3 or HEK293 cells (5-6 × 10 +/+ −/− Both 𝑆𝑜𝑥9 (EGFP) and 𝑆𝑜𝑥9 (EGFP) embryos showed EGFP cells/well) were cultured in DMEM supplemented with 10% expression in the Sox9-expressing domains, indicating that FBS and 40 𝜇g/ml of gentamycin sulphate. For the assays the reporter is replicating the Sox9 expression endogenously. conductedfortheSox9bindingsitesalone,cellswere Likewise, the Sox5 and Sox6 loci were tagged with the transfected with 500 ng of luciferase vector containing the EGFP reporter in mice for FACS. The EGFP reporter gene was binding site, 300 ng of Sox9 expression vector, and 5 ng of inserted into the endogenous Sox5 loci after the translation Renilla luciferase vector (transfection control) using 6 𝜇lof start site (Figure 1(d)), disrupting the longer protein isoform FuGENE HD transfection reagent (Roche Diagnostic, USA), of Sox5 which is the predominant form in chondrocytes [58]. in 100 𝜇l of OPTI-MEM I medium (Invitrogen, USA). For Similarly, the EGFP reportergenewasusedtoinactivatethe the assays validating the Sox trio binding sites, 300 ng of Sox6 allele (Figure 1(e)). Viable and fertile lines of heterozy- luciferase vector containing the binding sites, 200 ng of each +/− gous mice were generated from the targeted 𝑆𝑜𝑥5 and of Sox9, Sox5, and Sox6 expression vectors, and 2 ng of Renilla +/− luciferase vector (transfection control) using 6 𝜇l of FuGENE 𝑆𝑜𝑥6 ESC, respectively, and their phenotypes were similar HD transfection reagent (Roche Diagnostic, USA), in 100 𝜇l to what was previously described [28]. As Sox5 and Sox6 are of OPTI-MEM I medium (Invitrogen, USA), were transfected considered functionally redundant during chondrogenesis into cells. The ratios between the different vectors were the [33], the two mouse lines were crossed to generate double same for all transfections except where indicated. The cells homozygous embryos with both Sox5 and Sox6 inactivated −/− −/− were grown for 48 hours and the luminescence was measured (𝑆𝑜𝑥5 𝑆𝑜𝑥6 ) for the transcriptome analysis. using Dual Luciferase Reporter Assay System on a Glomax Sox9 is expressed specifically during embryogenesis in Multidetection System Luminometer as per manufacturer’s a variety of organs such as the male gonad, heart, nervous instructions (Promega Corporation). system, and kidney [9]. Sox5 and Sox6 are expressed in the neurons, oligodendrocytes, and further cell types other 2.13. GO Analysis and GRN Construction. GO analysis was than chondrocytes [59, 60]. For enriching skeletal cells carried out using DAVID [55]. The differential genes were expressing the Sox trio, the E13.5 embryos had their internal ranked based on expression fold change and if the number organs removed, leaving the axial and appendicular skeleton of genes analyzed exceeded 3000, the 3000 genes with the intact before dissociation into single cells for FACS. The highestfoldchangewereusedforGOanalysis.TheGO- remaining tissues present expressing the Sox trio would be terms for biological process were filtered using the corrected from the skeletal and central nervous system. It was not 𝑝 value < 0.05 adjusted by the Benjamini-Hochberg multiple possible to dissect out the central nervous system without testcorrectionandrankedaccordingtofoldenrichmentas compromising cell viability; hence the results here represent determined by DAVID. The Sox trio-mediated GRN was both chondrogenic and neurogenic cells types. Cells from a generated using Cytoscape [56]. wildtype E13.5 embryo were used to set the gating for FACS BioMed Research International 5

7.2 kb

1 kb ATG TGA E (a) E H E + 㰀 WT Sox9 allele 1 2 3 3 UTR Probe 1 Probe 2 X X

ATG F2A E E H E 㰀 Sox9 targeting construct 1 2 3 EGFP Neo 3 UTR FRT FRT +EGFP+Neo+ Sox9 HR 8.0 kb

ATG F2A E E E H E +EGFP+Neo+ 㰀 (b) Sox9 allele 1 23EGFP Neo 3 UTR FRT FRT In vivo FLPe recombinase ATG F2A E E H E +EGFP+ 㰀 Sox9 allele 1 2 3 EGFP 3 UTR FRT TGA E loxP loxP E −EGFP+Neo+ 㰀 (c) Sox9 allele // EGFP Neo 1 2 3 3 UTR

−EGFP+ Sox9 TGA E E −EGFP+ 㰀 Sox9 allele // EGFP 1 233 UTR loxP TGA E FRT FRT E −Neo+ 㰀 Sox9 allele // Neo 1 2 3 3 UTR

TGA E E − 㰀 Sox9 allele // 1 233 UTR FRT ATG + 㰀 (d) −EGFP+ WT Sox5 allele Sox5 5 UTR // Sox5

loxP loxP −EGFP+Neo+ 㰀 Sox5 allele Sox5 5 UTR EGFP Neo //

−EGFP+ 㰀 Sox5 allele Sox5 5 UTR EGFP // loxP ATG + (e) −EGFP+ WT Sox6 allele Sox6 CDS // Sox6

loxP loxP −EGFP+Neo+ Sox6 allele EGFP Neo Sox6 CDS //

−EGFP+ Sox6 allele EGFP Sox6 CDS // loxP Figure 1: Sox9, Sox5,andSox6 gene targeting and their E13.5 transgenic embryos generated for FACS and expression Profiling. (a) Wildtype Sox9 allele (Sox9+) is indicated with exons depicted as black boxes. Translation start site (ATG) as indicated. (b) Sox9 with the F2A-EGFP- FRTNeo inserted in exon 3 (𝑆𝑜𝑥9+(EGFPNeo)) and after FLPe recombination (𝑆𝑜𝑥9+(EGFP)). 𝑆𝑜𝑥9+/+(EGFP) E13.5 embryo shown on the right. − − (c) Sox9-null allele with the EGFP-loxPNeo inserted after the ATG (𝑆𝑜𝑥9 (EGFPNeo)) and after CRE recombination (𝑆𝑜𝑥9 (EGFP)). Sox9-null − − −/− allele with FRTNeo inserted in exon 1 (𝑆𝑜𝑥9 (Neo)) and after FLPe recombination (𝑆𝑜𝑥9 ). 𝑆𝑜𝑥9 (EGFP) E13.5 chimeric embryo shown on the − right. (d) Exon containing the ATG of wildtype Sox5 (Sox5+)shown.EGFP-loxPNeo inserted after the ATG (𝑆𝑜𝑥5 (EGFPNeo)) and after CRE − +/− recombination (𝑆𝑜𝑥5 (EGFP)). 𝑆𝑜𝑥5 (EGFP) E13.5 embryo shown on the right. (e) Exon containing the ATG of wildtype Sox6 (Sox6+)shown. − − +/− EGFP-loxPNeo inserted after the ATG (𝑆𝑜𝑥6 (EGFPNeo)) and after CRE recombination (𝑆𝑜𝑥6 (EGFP)). 𝑆𝑜𝑥6 (EGFP) E13.5 embryo shown with a wildtype embryo (WT) under same lighting condition. EGFP,green boxes; F2A, blue ovals; FRT sites, red triangles; loxP sites, yellow triangles (see also Figure S1 and Table S1). 6 BioMed Research International

(Figure S1). Postsort analysis of the collected EGFP-positive 3.2. Tissue ChIP-Seq Shows Sox9 Acting as a Homodimer and fraction from all genotypes routinely showed that the EGFP- Chondrogenic Activator. The pool of differentially expressed expressing cells in the sorted population make up more than genes from the microarray results consisted of target genes 95% as compared to the presorted population with 0.5–3% that were regulated directly by Sox9 and other indirect target EGFP-expressing cells, indicating that the Sox-expressing genes whose expression was altered due to the loss of Sox9 cells have been successfully isolated from the nonexpressing affecting their upstream regulators. To find genes involved in cells (Figure S1). Total RNA from the sorted cells was then chondrogenesis directly controlled by Sox9, ChIP-Seq with used for microarray analysis. an anti-Sox9 antibody using dissected tissues from limbs and 5742 genes were found to be differentially expressed tails of wildtype E13.5 embryos, naturally enriched for the +/+ −/− between 𝑆𝑜𝑥9 (EGFP) and 𝑆𝑜𝑥9 (EGFP) (Table S1). The loss chondrogenic lineage, was performed (Figure 3(a)), with IgG of Sox9 resulted in the upregulation of 3105 genes and ChIP as background control. Using the model-based analysis downregulation of 2637 genes. Genes that were upregulated for ChIP-Seq (MACS) [52] algorithm, 3260 peaks were represented the putative genes that are normally repressed identified. ChIP-quantitative PCR validation on randomly by Sox9 and those that are downregulated represented the selected peaks identified from the analyzed ChIP-Seq results putative targets activated by Sox9. RNA section in situ showed good correlation between the two datasets, indicating hybridization (SISH) was carried out on 25 differentially that the peaks called were specific (Figure S3; Table S3). expressed genes identified, including that of a novel gene, To investigate the function of the binding sites, the peaks 3110079O15Rik, and it was ascertained that their expres- were annotated based on their positional information relative sion domains overlapped with that of Sox9 (Table S2). To tothenearestgene(Figure3(b)).Thebreakdownofthe further validate our Sox9 microarray results, RNA SISH distribution of binding sites showed 77% within 100 kb of the +/+(EGFP) transcriptionstartsiteofageneorwithinthegeneitselfwith on some of the targets was performed using 𝑆𝑜𝑥9 the majority present in distal and intragenic regions. embryos and conditional Sox9 knockout embryos previously The Sox9 tissue ChIP-Seq identified peaks at previously described [50]. A reduction in the expression of Col2a1,a characterized chondrocyte-specific enhancers in vitro such known Sox9 target, was observed as expected when Sox9 as Col2a1 and Col9a1 [62, 63]. Luciferase assays were carried was inactivated (Figure S2). Downregulated genes such as out on some of the identified Sox9 binding sites in or Matn4, Hapln1, and the novel gene, 3110079O15Rik,were nearesttotheputativetargetgeneslikeMatn4, Mia1, Hapln1, observed to have reduced expression whereas upregulated Acan, Papss2, Sox5,andSox6, validating the transactivation genes like Pax1 showed an increase in expression, con- function of the binding sites (Figure S4; Table S3). firming the expression profiling results, giving rise toa To study the in vivo Sox9 binding sequence, de novo list of genes associated with the function of Sox9 at the motif finding was performed using Multiple EM for Motif postmesenchymal condensation stage (Figure S2) and con- Elicitation (MEME) [64] with the top 200 Sox9 peaks with currently, indicating that the experimental approach was the highest tags count as defined by MACS. The primary robust. motif identified confirmed in vivo that Sox9 functions as ComparativeprofilingatE13.5wascarriedoutforthe −/− −/− a homodimer with the consensus Sox motifs in opposing sorted cells from 𝑆𝑜𝑥5 𝑆𝑜𝑥6 embryos, against sorted +/− orientation (Figure 3(c)) as reported in in vitro studies previ- cells from 𝑆𝑜𝑥6 embryos, which have previously been ously [65]. Motif Alignment and Search Tool (MAST), part of reported to be of a wildtype phenotype [28]. In the absence the MEME suite, was then used to scan for the primary motif of Sox5 and Sox6, a total of 4780 genes were found to identified in all the Sox9 binding sites. About 37% (1197 out of be differentially expressed with 1446 genes downregulated 3260) of the Sox9 binding sites were found to have the Sox9 and3334genesupregulated(TableS1).ThefactthatSox9 primary homodimer motif with the most common spacer was not found to be differentially expressed validates pre- number of four base pairs found between them (Figure S3). vious findings describing Sox9 as an upstream regulator of The spacing requirement for Sox9 binding has been reported Sox5 and Sox6, unlikely to be regulated by Sox5 and Sox6 in in vitro experiments previously [65] and it has also [27]. been shown that certain patients with campomelic dysplasia To assess the major biological functions associated with possess a mutant Sox9 incapable of homodimerization [66]. the differentially expressed genes from the microarray anal- Our study supports recent findings on genomewide surveys ysis, (GO) analysis was carried out for the of the homodimer motifs and spacing in vivo [7, 8, 67, 68] upregulated and downregulated genes using the Database reviewed in [69, 70]. Interestingly, in the study by Ohba et al. for Annotation, Visualization and Integrated Discovery [7] they suggest that the Sox9 protein can have two different (DAVID)[61].ThetopGO-termsforbiologicalprocesses mechanisms of action, one where it interacts directly with the that were overrepresented in the downregulated gene sets for TSS and a second one where it binds as a homodimer to its both the Sox9 and the Sox5/6 microarray were enriched for DNA recognition motifs [7] reviewed in [69, 70]. terms associated with skeletogenesis such as skeleton system morphogenesis and cartilage development (Figures 2(a) and 3.3. Regulation of Sox5 and Sox6 Expression by the Sox 2(b)). The top GO-terms for the upregulated genes did not Trio. Sox5 and Sox6 are known to act downstream of Sox9 show skeletal-related terms (Figure S2), supporting previous during chondrogenesis. Together, these three transcription observations that the Sox trio acts as chondrogenic activators factors are thought of as the chondrogenic master regulators [1]. [24, 33, 71] and Sox9 binding sites were previously found BioMed Research International 7

GO:0048736 ∼ appendage development GO:0048705 ∼ skeletal system morphogenesis GO:0030326 ∼ embryonic limb morphogenesis GO:0035113 ∼ embryonic appendage morphogenesis GO:0035107 ∼ appendage morphogenesis GO:0035108 ∼ limb morphogenesis GO:0051216 ∼ cartilage development GO:0051329 ∼ interphase of mitotic cell cycle GO:0060021 ∼ palate development GO:0006888 ∼ ER to Golgi vesicle-mediated transport

0 0.5 1 1.5 2 2.5 3 3.5 4

Fold enrichment (a)

GO:0048706 ∼ embryonic skeletal system development GO:0048705 ∼ skeletal system morphogenesis GO:0032259 ∼ methylation GO:0016569 ∼ covalent chromatin modification GO:0016570 ∼ histone modification GO:0051216 ∼ cartilage development GO:0043414 ∼ biopolymer methylation GO:0048704 ∼ embryonic skeletal system morphogenesis GO:0021953 ∼ central nervous system neuron differentiation GO:0001502 ∼ cartilage condensation

0 12345678

Fold enrichment (b)

Figure 2: Differentially expressed genes from the Sox9 microarray and Sox5/Sox6 double-null microarray. (a) Top 10 GO-terms for genes that are downregulated when Sox9 is inactivated. (b) Top 10 GO-terms for genes that are downregulated when Sox5 and Sox6 are inactivated (see also Figure S2 and Table S2).

at the Sox5 locus in a ChIP-on-chip experiment using rat that the HA-tag was not interfering with the Sox5 protein 𝐻𝐴/𝐻𝐴 +/− chondrosarcoma cells [72]. None have been reported for Sox6 function, the 𝑆𝑜𝑥5 mice were crossed to the 𝑆𝑜𝑥5 𝐻𝐴/− so far. In this study, multiple Sox9 binding sites in vivo were gene targeted line to generate 𝑆𝑜𝑥5 mice, which were found in both Sox5 and Sox6 loci (Figure S3). Sox5 and indistinguishable from wildtype mice. E13.5 wildtype limbs Sox6 were both downregulated in the microarray by 5.6-fold and tails immunoprecipitated with HA antibody were used and 2.5-fold average, respectively, when Sox9 was inactivated. as a negative control. To determine the Sox6 binding sites, Taken together, this confirms that Sox9 activates Sox5 and anti-Sox6 antibody was used to immunoprecipitate limbs and Sox6 expression, confirming them as direct Sox9 targets at tails of E13.5 wildtype embryos with the IgG ChIP serving as E13.5. anegativecontrol. A previous in vitro study showed that the DNA binding A total of 632 and 299 peaks were identified for Sox5 of Sox5 and Sox6 was not restricted to their consensus and Sox6, respectively, by the MACS analysis. Motif analysis Sox motifs but instead they can bind to variable motifs for the Sox5 and Sox6 ChIP-Seq dataset showed that the 󸀠 󸀠 on cartilage-specific enhancers on genes such as Acan [29], consensus Sox motif 5 -(A/T)(A/T)CAA(A/T)G-3 [73] was suggesting that in silico approaches are not the ideal way enriched (Figures 3(d) and 3(e)), highlighting the robustness to find the binding sites. Thus, to determine the in vivo oftheinvivoChIP-Seq.Thepeakswerethenannotatedwith binding profile, tissue ChIP-Seq was done for Sox5 and Sox6 respect to the nearest genes and classified as described for the using E13.5 mouse embryos. Sox5 ChIP-Seq was carried out 𝐻𝐴/𝐻𝐴 Sox9 ChIP library (Figures 3(f) and 3(g)). The distribution using limbs and tails from E13.5 embryos of 𝑆𝑜𝑥5 of the binding sites found in the TSS and the promoter 𝐻𝐴/𝐻𝐴 mice previously described [40]. In these 𝑆𝑜𝑥5 mice, region of the gene for the Sox5 and Sox6 ChIP library was only the longer isoform of Sox5, relevant in chondrogene- similar to those of the Sox9 ChIP library. It was observed sis, was tagged with haemagglutinin (HA). To demonstrate that most of the Sox5 and Sox6 binding sites were not 8 BioMed Research International

TSS, 169, 5% Promoter, 108, 3%

Others, 73 23%

Intragenic, 1056, 32% 2

Distal, 108 33% 1 (Bits)

Proximal, 116, 0 4% 1 2 3 4 5 6 7 8 9 1011121314151617181920 (a) (b) (c)

2 2

1 1 (Bits) (Bits)

0 0 1234567891011 1 234567891011 (d) (e)

TSS, 37, 6% TSS, 13, 4% Promoter, Promoter, 5, 2% 7, 1%

Intragenic, Proximal, 7, Intragenic, 64, 10% 1% 42, 14% Proximal, 2, 1% Distal, 45, 7% Distal, 24, 8% Others, 47 Others, 213, 75% 71%

(f) (g)

Figure 3: ChIP-seq results for Sox9, Sox5, and Sox6. (a) Dissected tissues of the E13.5 embryo used for ChIP are indicated by the dotted blue lines. (b) Distribution of the Sox9 binding sites with respect to their nearest genes. (c) Primary Sox9 motif found shows Sox9 acts as a homodimer with the binding sites in opposite orientation. (d) Primary motif found from the Sox5 binding profile. (e) Primary motif found from the Sox6 binding profile. (f) Distribution of the Sox5 binding sites with respect to their nearest genes. (g) Distribution of theSox6 binding sites with respect to their nearest genes (see also Figure S3 and Table S3).

found in the proximity of the genes unlike what was seen not identical results to those described here have previously for the Sox9 binding sites. In particular, multiple Sox5 and been described [67] and reviewed in [69, 70]. Sox6 binding sites were found in their loci (Figure S3), suggesting that there is a self-regulatory feedback mechanism 3.4. Genes That Are Directly Regulated by the Sox Trio. in addition to the Sox9 regulation. Complementary although To find genes that are regulated by Sox9, Sox5, and Sox6, BioMed Research International 9

GO:0048736 ∼ appendage development GO:0048562 ∼ embryonic organ morphogenesis GO:0035107 ∼ appendage morphogenesis GO:0035108 ∼ limb morphogenesis GO:0022610 ∼ biological adhesion GO:0007155 ∼ cell adhesion GO:0048568 ∼ embryonic organ development GO:0048598 ∼ embryonic morphogenesis GO:0051216 ∼ cartilage development GO:0001501 ∼ skeletal system development 0 24681012

Fold enrichment (a) GO:0051216 ∼ cartilage development GO:0045935 ∼ positive regulation of nucleobase, nucleoside, nucleotide… GO:0006355 ∼ regulation of transcription, DNA-dependent GO:0051173 ∼ positive regulation of nitrogen compound metabolic process GO:0031328 ∼ positive regulation of cellular biosynthetic process GO:0010557 ∼ positive regulation of macromolecule biosynthetic process GO:0009891 ∼ positive regulation of biosynthetic process GO:0051252 ∼ regulation of RNA metabolic process GO:0045449 ∼ regulation of transcription GO:0006350 ∼ transcription 0 5 10152025

Fold enrichment (b)

GO:0045893 ∼ positive regulation of transcription, DNA-dependent GO:0051254 ∼ positive regulation of RNA metabolic process GO:0045944 ∼ positive regulation of transcription from RNA polymerase II promoter GO:0010604 ∼ positive regulation of macromolecule metabolic process GO:0010557 ∼ positive regulation of macromolecule biosynthetic process GO:0031328 ∼ positive regulation of cellular biosynthetic process GO:0009891 ∼ positive regulation of biosynthetic process GO:0045165 ∼ cell fate commitment GO:0001501 ∼ skeletal system development GO:0051216 ∼ cartilage development

0 10 20 30 40 50 6070 80 90

Fold enrichment (c)

Figure 4: (a) Top 10 GO-terms for genes that are activated by Sox9. (b) Top 10 GO-terms for genes that are activated by Sox5. (c) Top 10 GO-terms for genes that are activated by Sox6 (see also Figure S4 and Table S4).

the differentially expressed genes from the Sox9 microarray rather than a repressor during this stage [27, 74]. To assess were overlapped with the Sox9 ChIP-Seq library and the the direct targets of Sox9 in relation to their biological differentially expressed genes from the Sox5/6 microarray relevance, GO analysis was used. The upregulated genes when were overlapped with the Sox5 and Sox6 ChIP-Seq data. 490 Sox9 was inactivated did not have any GO-terms with a 𝑝 genes, 113 genes, and 52 genes were found to be controlled by value < 0.05. The downregulated genes, however, were clearly Sox9, Sox5, and Sox6, respectively (Table S4). enriched for skeletal-related genes with the GO-term skeletal More than two-thirds of the 490 genes with Sox9 binding system development and cartilage development topping the sites were downregulated when Sox9 was inactivated, reiter- list (Figure 4(a)), indicating that Sox9 activates skeletal genes ating that Sox9 is more of an activator for in vivo and that the dataset generated was robust and specific. 10 BioMed Research International

More genes were found to be upregulated rather than phenotype compared to the control that was injected with downregulated by Sox5 and Sox6 during the overlap, oppos- the scramble Morpholino at 24 hpf (Figure 5(c); Table S6). ing the trend observed for Sox9.Thedownregulatedgenes Transverse sections of the Morpholino-injected embryos during Sox5 inactivation with Sox5 binding sites were over- showed extensive defects in the emerging sclerotome as represented by GO-terms for transcription and cartilage demonstrated by the reduced Alcian Blue staining and development (Figure 4(b)), showing that Sox5 activates car- loss of structural integrity, indicating that the three genes tilage genes. Similarly, genes shown to be activated by Sox6 played a key role in developing chondrocytes (Figure 5(c)). were enriched for cartilage development and skeletal system Knockdown of Rad51c in the zebrafish was inconclusive, as development (Figure 4(c)). Despite the larger number of the mutants do not survive after Morpholino injection. In upregulated genes for both Sox5 and Sox6, which overlap, summary, the in vivo results from the microarray and ChIP- none of the GO-terms has a significant adjusted 𝑝 value < Seq done in mice, together with the zebrafish knockdowns 0.05. and in vitro luciferase assays performed, indicate that Fbxl18, Tgfb2,andTle3 are involved in sclerotome formation and are 3.5.Fbxl18,Tgfb2,andTle3AreTargetsoftheSoxTrio. As transcriptionally regulated by the Sox trio. This also provides observed earlier, a large portion of the Sox5 and Sox6 binding validation to the in vivo datasets generated in this study, sites was present more than 100 kb away from the TSS or allowing the construction of the most comprehensive Sox trio outside the gene. To understand the biological relevance of GRN done to date (Figure 6). these binding sites, another approach was taken in which differentially expressed genes on both sides of the Sox5, 4. Discussion Sox6, and Sox9 binding sites were considered (Table S5). Previously, only the nearest gene next to the binding site was It is becoming increasingly clear that the complete under- used for analysis. 21 genes were found to be differentially standing of any biological process in vertebrates will require expressed in both sets of microarray with binding sites for a comprehensive approach looking at multiple transcription all three of the Sox proteins. These included targets like factors simultaneously and incorporating systems biology Sox5 and Sox6 themselves, Tgfb2, Tle3,andFbxl18 (Figure approaches [77–79]. Coupling of genomewide binding site S3, Figure 5). To confirm the transcriptional activity of the data with expression profiling will give a detailed snapshot binding sites found for the Sox trio near some of these of the network of genes involved in a particular process [80, targets, luciferase assay was carried out for the Sox trio-bound 81]. Previously only was subjected to a comprehensive enhancers of Fbxl18, Rad51c, Sox11, Sox5, Sox6, Tgfb2,and genomewide systems approach [82] but this was limited to Tle3. Luciferase reporters were constructed with the element in vitro cell culture lines which often fail to recapitulate bearing the binding sites for the three Sox proteins of each theactualinvivofunction.Ourstudylookedatthree of the target genes upstream of a minimal promoter (Table major transcription factors involved in organogenesis in the S3). The results showed that the DNA elements bound by all developing embryo to understand the molecular and genetic three Sox proteins were activated in the presence of the Sox events underlying the process of chondrogenesis. Previously trio, validating the functionality of the identified binding sites there had been numerous attempts to look at the role of (Figure S5). the Sox trio in gene regulation. These studies often involved The different Sox proteins were then further tested for analyzingsinglegenelociininvitrosystems[29,30,83]. their individual contribution in the transcriptional activation Our three-factor ChIP-Seq coupled with extensive expression of the Sox trio-bound regulatory regions found near the profiling performed using developing mouse chondrocytes genes Fbxl18, Rad51c, Tgfb2,andTle3 (Figures 5(a) and in an in vivo context has made it possible to perform a 5(b)). These four genes were shortlisted as little has been global analysis, deciphering the genomewide interactive roles reported about their role in chondrogenesis. It was observed of Sox9, Sox5, and Sox6 in regulating the genes involved in from the luciferase assay that the absence of Sox9 alone chondrogenesis thus providing an extensive Sox trio-driven caused the biggest drop in the reporter activity of all the gene regulatory network in vivo for the first time. Sox-bound enhancers whereas the loss of Sox5 and/or Sox6 The approach of enriching for specific chondrogenic resulted in a smaller loss of activity. The results indicated and tissues in both ChIP-Seq and transcriptome profiling exper- supported the hypothesis that Sox9 is the main activator for iments has allowed us to specifically look at genes and the regulatory regions of chondrogenic genes with Sox5 and pathways involved primarily in chondrogenesis. By analyzing Sox6 supporting it. the gene expression profiles together with the identified To investigate the biological function of these Sox trio- binding sites for the Sox trio, we identified a wide repertoire of regulated genes in particular with respect to skeletal forma- genes that are regulated directly by Sox9 alone such as Prelp, tion, Morpholino-mediated knockdowns of Fbxl18, Rad51c, Hapln1, and the novel target 3110079O15Rik as well as those Tgfb2,andTle3 wereperformedinzebrafishbyinjecting that are regulated by Sox5 and/or Sox6 alone or together with the translation blocking Morpholino into the 1-cell stage of Sox9 during embryogenesis (Figure 6). zebrafish embryos. Tle3 has two homologues in zebrafish, To provide biological insight, hypothesis generating groucho1 (gro1) and groucho2 (gro2).Sincegro1 is expressed screens such as ChIP-Seq experiments need to be combined primarily in the central nervous and hematopoietic system with meaningful assays for downstream functional valida- [75, 76], we chose to focus on gro2. Reduced expression of tion. For this study, we took advantage of the zebrafish Fbxl18, Tgfb2,andTle3 (gro2) resulted in a marked midline model as it allows high-throughput knockdown analysis BioMed Research International 11

Sox5 35 Sox6 30 Sox9 25 Fbxl18 20 Sox5 15 Sox6 10 Sox9 5

Tgfb2 0 Sox5 NC Fbxl Fbxl Fbxl Fbxl Tle3 Tle3 Tle3 Tle3

Sox6 Tgfb2 Tgfb2 Tgfb2 Tgfb2 Rad51c Rad51c Rad51c Rad51c Sox9 Sox9 −+−+++−+++−+++−++ Sox5 −++−+++−+++−+++−+ Tle3 Tle3 Tle3 Sox6 −+++−+++−+++−+++− (a) (b)

WT Fbxl18 Tgfb2 Tle3

NT NT NT NT

(c)

Figure 5: Genes regulated by the Sox trio. (a) Binding profile of Sox9, Sox5, and Sox6 for Fbxl18, Tgfb2,andTle3.(b)Transactivationassay for the Sox binding sites found for Fbxl18, Tgfb2,andTle3. Different Sox proteins were transfected as indicated with the enhancers containing the Sox trio binding sites and the luciferase activity was then measured. (c) Top panel shows the whole mount picture of the 24 hpf zebrafish embryos injected with the Morpholinos for Fbxl18, Tgfb2,andTle3 along with the wildtype (WT) injected with the scramble Morpholino as a negative control. Bottom panel shows the respective transverse sections of the Morpholino-injected embryos stained with Alcian Blue. TheWTembryoshowsnormalneuraltube(NT)andawell-definedemergingsclerotome(blackarrow).Themorphantsatthesamestage show extensive midline phenotype with defects in both neural tube and the sclerotome as seen by reduced Alcian Blue staining and loss of structural integrity (see also Figure S5 and Table S6).

in vivo in the context of the entire developing organism. providing evidence of the autoregulatory role of the Sox trio By employing the multifactor ChIP-Seq and microarray, in vivo. The network also includes genes that were previously together with the zebrafish validation, three new genes Fbxl18, not thought to play any role in chondrogenesis like Fbxl18, Tgfb2,andTle3 were discovered to be coregulated by the Tle3, Rad51c,andBach2 andwehavevalidatedtwoofthese Sox trio, highlighting the biological relevance of the network novel targets in this study. generated. Other genes previously identified to be part of The knockdowns of Fbxl18, Tgfb2,andTle3 showed clear skeletogenesis but not known to be controlled by the Sox sclerotome defects thus providing evidence for their role trio were also uncovered in our network. These include in skeletal under the control of the Sox trio enhancers. Notch4,partoftheNotch-Deltasignalingpathway,andSox11 A recent report linked the of Tgfb2 (Tgfbr2) to whose homozygous mutants display craniofacial and skeletal intervertebral disk formation [85]. Our data clearly shows abnormalities [84]. In addition, multiple Sox binding sites that the ligand Tgfb2 is directly controlled by the Sox trio. in and around Sox5 and Sox6 were found and validated, The transcriptional activation assay with Sox trio-bound 12 BioMed Research International

Col9a3 Sema3d Ech1 5730508B09Ri k Spcs2 Ndufb4 Dmrta1 Cdh8 Kat2b Vwf Pcyt1a Gin1 Tbc1d19 Actb Tpp1 Ak1 Pex7 Snx3 Cytip Smarcc1 Hspb8 Rnf182 Ppp2r2a Sp3 Hadh Mtdh Gng12 Rgmb Brd2 Ptpn12 Pitpnc1 Micall2 Mcf2lDlgap4 Myl4 Smarca5 Hook1 Grb10 Gpr153 Snrpa Clk4 Evl Zfp827 Ppp4r1 Art3 Slc29a1 Ccdc66 Ccdc109a Adamts18 Laptm4b Postn Yipf1 Smox Prg4 Dab1 9130024F11RiAdk k Col9a2 1110058L19Rik Maml2 Col5a2 Pax1 Fgd4 Cdyl Pxdn Arglu1 Cic Anapc1 Tasp1 Mycbpap Pex3 Cyr61 Lbh Nfe2 Lgr5 Spsb1 Vill Ptpn1 Zfp191 Xpo4 St8sia2 Spsb 4 Cep170 Zcchc6 Eya1 Tpd52 Erlin2 Frmd6 Tubb2b Tmem14c Cox7b Zdhhc17 Rasgef1b Tmpo Trps1 Camk1d Tmed3 Dlc1 Srpk2 Rbm6 Tmcc2 Tom1l1 Sox8 Slc30a5 Rpl29 Lmcd1 Plce1 Fam107a Fam101a B3galnt 1 Ephb2 Pcnt Dusp6 Nkx6−1 Errfi1 Jun Rbpjl Ebf3 Slc39a3 Lrrc1 Lrch1 Camta1 Lcorl Ptx3 Fgfr2 Zfp618 Slc25a4 Cdk6 Rcl1 Nup214 Cdk8 Scap Smyd2 Deb1 Lbr Vamp3 Ppp3ca Gcnt1 Gadd45g Taf13 Fstl1 Tmem17 Ctnnd1 1110034A24Rik Zmiz2 Tuba3b Chst8 Pik3r1 Cbll1 Sec16a Bmp4 Bcas2 Rala Mbp Cmtm5 Ptprb Itga11 Col14a1 Comp Dnajc24 Stau2 Dcn Hopx Lsm6 Idua Slc4a3 Dtx3l Gas2 Fign Slc12a4 Dach1 Stam2 Melk Ube2v1 Adamts9 Papola Sfrp2 Nfix Extl1 Zfp335 Cnot2 Lrrk2 Camk2a Elovl6 Fbp2 Fn1 Lum Chd6 C80913 Sema3e Gria2 Vasn Arrdc4 Map4k3 Ywhaz Ppp2r1a Kif26b Zfp184 Zfand1 A830080D01RikPmepa1 Hapln1 Aspscr1 Sgcb Pdzd8 Tpm4 Ptpn23 Por Wnk1 Lect1 Tspan2 Chd2 Sh3rf1 Utrn Mnat1 Fscn1 1500015O10Rik Slc18a2 Irf2 2010106G01Rik Actr10 Gja5 Cltc Rabl5 Scamp3 Ints10 Plekhf2 Itgb8 Nbea Tomm34 Vgll4 Ptn Snx4 2900011O08Rik Col27a1 Nup54 Glrx3 Clasp2 Rhobtb3 Klf12 Mapk9 Ift57 Marco Mkks P2ry1 Ctgf Wbscr17 Dck Vps4b Gnai2 Fut8 Ube4b Ppap2b Crim1 Barx2 Rev3l Fxyd3 Aaas Picalm Npepl1 Bsg A430107O13Rik St8sia4 1190002N15Rik H2−T23 Zfp36l1 Hlcs 6720401G13Rik Lonp2 Dhx36 Gm2a Fam108b Ncf2 Akap2 Pcdh17 Moxd1 Eps8 Prosc Mtbp Pfkm Tbx4 Igf2bp3 Snx2 Pcca Cxadr RarsSec61b C030044B11Rik Ppic Sobp Foxf1a Trim33 Lrp1 Tox Rock1 Tlr13 Stk39 Kcnma1 Nfu1 Tle1 Galnt14 Bdh1 Gli2 Hs3st1 Clint1 Tac1 Cd9 Hs3st3b1 Mbnl2 Cep70 Bsdc1 Acvr1b Col12a1 Spnb2 Flrt1 Col23a1 Cog6 2410018L13Rik Ppcdc Alg14 Fubp1 Cdc5l Chd7 Lrmp Gadd45a Spred 1 Palld Col2a1 1810011O10Rik Glt25d2 Ttc8 Sgk1 Entpd6 Pigp Znhit6 Tln2 Pcdh10 Fat3 Jag1 Foxp4 Snx1Sema6a Gorasp2Ppt1 Flrt3 Fmnl2 Nrp2 Prdm6 Mpeg1 Phlda1Ezh2 Upf1 Bmpr1a Tanc1 Commd6 Cdkn2aip Atp5c1 Slc35f1 Thsd7b Tssc4 Ptprf Mgst1 Col8a1 Fndc3b Zfp113 Rhbdd2 Sec22b Bmp5 Rbbp9 Rgs10 Bcl2l1 Anln Mkl1 Myo1e Acpl2 Etv1 Exoc1 Smc2 Ilf3 Aff1 Mmp16 Tsc22d1 Gnb5 Bmf Gphn Fkbp14 Fip1l1 Fcrls 4930402H24Rik Cblb Wwtr1 Rarb 1700030C10Rik Diap2 Tbx5 Yipf6 Stim1 Ttk Gmds 5930412G12RikTnrc6a Gse1 Gpr125 Asph Prmt5 Glul Fgl1 Malt1 Ect2 Foxf2 Pou3f1 Lmnb1 Sars 4933424B01Rik Npm3−ps1 Mrpl13 Tbc1d2b Efnb2 Rtkn Slc35e3 Glis2 Egfl7 Yeats4 Gli3 Tyw1 Dhx40 Tgoln1 Ust Cbln1 Hist1h2be P4ha3 6430573F11Ri k Tbx15 Ppard Dapk1 Wbp2 Kcnj2 Kirrel3 Tjp1 Has2 Bbs9 Slamf9 Zfp330 Zfp507 Hmgcs2 Lrrc8d Emb Syncrip Utp23 Tns3 Gfpt2 Dmxl1 1700034H14Rik Arl5b Csgalnact1 Xkr5 Arid3b Ccnd2 Csf1 Apip Cat Pxn Wif1 Gtf2i Sesn 3 Gria1 Snx7 Tnfrsf21 Bmp1 Mgll Slc39a8 Rasa1 Fancc Kcnk1 Gtf3c6 Rrs1 Ddah1 Sox9 Slc35a3 Ss18 Mylip Nt5c2 Atp1a2 Erg Rab11a Dnajb6 Igsf11 Wwp2 Kras Dpagt1 Pde6d Dact1 Prkar2b Crabp2 2700060E02Rik Rbms1 Cspg4 Fzd5 Gramd3 Camk2d E430025E21Rik BC027231 Itpkb AI593442 Hoxa9 Cobll1 Ldb2 Ung Sec24d Maea Zc3hc1 Hspa4 Tcof1 Zeb2 Mia1 Etv5 Fam114a2 Bbs12 Scamp5 Stard10 Sorbs2 Myst4 Atg10 Tgm2 Cd24a Psip1 Cdadc1 Gbas Usp9x Ssh1 Akap1 Ttc30b A730069N07Rik Ndst1 E330033B04Rik Pthlh Snx13 Itgb1 Mpdz Lonrf1 Gabpb 1 Rhou Trim62 Ttc15 Lrp12 Arhgef2 Rangap 1 Mrpl48 Pygb Dab2 Pcsk6 Snx24 Sec14l 1 Trappc3 9030625A04Rik Pitx2 Glcci1 Cald1 Lmbr1l Pdzrn3 Fam122b Tbxas1 Erbb2ip Msi2 Ube2e3 Ets2 Fam181b Ylpm1 Comt Eapp Opa1 Col1a1 Minpp1 Bbox1 Pdzrn4 Cdon Nrip1 Bcl11b Plekha2 Sec23a Prpf40b Ccnd1 Ndufa10 Foxp2 Pabpc1 Pafah2 Sh3kbp 1 Hmgb1 Ing3 Snx9 Ranbp6 Dst Klf6 Il17d Nob1 Dct Sall1 Hipk2 Vat1 Dlg2 Esm1 Pbx3 Sertad 4 Cd109 Trio Actn4 Rab28 Cttnbp2nl 1190002H23Rik Kank1 Ahi1 Slc8a1 Stmn4 Isca2 Acot7 Ncam1 Btg1 Stard3 Lsamp Zfc3h1 Sugt1 Fbn1 Pde7a Cntn1 Igfbp3 Bdh2 Gls Tlk2 Wnt11 Cpsf6 Lef1 Gulp1 H47 Psma5 Ube2e1 Klhdc10 Rgs3 2810008D09Rik Alox5ap Pfn2 Adra2a Runx1 Efnb3 Cep78 Nrg1 Plekhb1 Lrp1b Smc3 Ppp1r3b Rere Atp5j

Umps9430023L20Rik Clk2 Il1rap Tmem161b Tns4 Cthrc1 Hsf2 Cap1 Efcab2 Fzd8 Rps24 Akap9 Igsf6 0610007P08Rik Dchs1 Adam11 Prelp Bcor Chic2 Hdac4 Crat Pde1a P2ry6 Amotl2 Dcc Msrb3 Nudt9 Dleu7 Calcoco1 Mif Robo2 Prdx6 Psmb7 Tssc1 Nosip Slc25a30 Cpe Unc13c Lyplal1 Cyp26b1 Fzd9 2900097C17Rik Etnk1 0610007P14Ri k Chst3 Cyld Ptk7 Ifnar2 Nfatc4 Evi2a Trib2 Mical1 Dok1 Slc10a7 Map4k4 Dnm3os Barx1 Atp6ap2 Zfp704Gpbp1 Lrp6 Zfp825 Zfp238 Slit3 Atic Ascc3 Angptl1 T Fndc3a Sgms1 Pcmtd1 Asns Nt5dc3 Uncx Etfa Slc14a1 Usp47 Gch1 Thbs2 Txnl1 Irf8 Peg10 Psat1 Rarg Cxxc5 1300014I06Rik Tspan8 Acan Cpm Gpam Shox2 Tpd52l1 Hsp90ab 1 Spg20 Kcns1 Itga9 Tsen15 Sulf2 Osr1 Dab2ip 2810001G20Rik Ccdc99 Kif21a Dkk3 Gata3 Plekha1 Klf15 Usp8 Ino80c Rab8a Bahcc1 Dhrs4 C8b Prickle1 Nog Myh9Hs3st3a1 Tmbim6 Elac2 Nedd4l Arhgef1 Flnb Fez2 Sphk1 Fgfr3 Akap13 Olfml3 Prrc1 Cetn3 Tex2 Ror1 Fbn2 Atp2b1 Gnaq Chd1 Ptprd Ppm1l Aars2 Cdc42ep 3 Olig1 Nudt4 Stk32a Zfp687 Synpo Thbs1 Tspan18 Aqr Mylk Anp32e Lats2 Fam20bHdac11 Cdk7 Mccc2 Zfp521 Adpgk Lat Riok3 Azi2 Lca5 Scrg1 Gnal Tcfec Lrrc33 Mtus1 Zfpm2 St3gal 1 Ralb A330049M08Rik Mest Slc6a15 Hgf Pdia5 Nmi Jdp2 Zbed3 Plxna4 Slit2 BC030307 Bcl11a Rbm5 Pcbp3 Grasp Srrm1 Vcan Hoxa2 Tgfbi Pcdh7 Fgfr1op Mon1b Lyl1 Pms1 Gdf10 Jmy Dyrk1a Atp9b Zmynd11 Ptpre Pak3 Scrn1 Wrnip1 Capn6 Arhgef19Tnrc6b Fbln2 Grid2Rapgef 2 Six2 Neo1 Sdccag8 Susd5 Hace1 Slc44a2 Scin 3110079O15Rik Sumf1 Ccdc77 Cul2

Mbnl1 1300010F03RiCdh5 k Sox4 Yipf5Nfatc1 Hmcn1 Bach2 Txndc11 Syne2 Zfp326Chst11 Myh10 Lphn3 Myo1b Notch4 Il15 Setd4 Mtap1b Frmd4b Plcg2 Pde3a Tgfb2 Tpcn2 Col9a1 Arhgap28 Nsmce2 Hs1bp3 Fbxl18 Mme Nab1 Tmem2Phldb2 Pmp22 Hmga2 Aebp2 Papss2 Col11a1 Tcf12Ptprz1 Srbd1Thsd4 Col4a1 Ednra Epha4Zfp462 Tgif1 Tgfbr2 Ptprm Rasl11b Unc5c Krr1 Gpm6a Sox11 St6galnac6 Clic4 Robo1 Rhob Trpm3 Ntng1 Edil3 Col25a1 Gjb6 Mtf2 Tle3 Zfhx3 Ggta1 Tbx1 Mef2c Odz2

Rad51c

Cdh6 Rnf144a

Svep1 Trim16Actn2 1700020O03Rik Prpf3 Iqgap1 Stoml1 B3gat2 6030458C11RikAcaa2 Map3k7 Rarres2 Irx2 Nat10 Rbms3 Fstl5 Map3k5 Skil Mllt3 Itm2b Bcas1 Arhgap12 Rab34 Rab31 Rara Sgip1 Gm1587 Ccdc6 Ythdc2 Twist2 Phyhipl Elf2 Pfdn4 Etv4 Ube2t Zmiz1 Dusp10 Dbx1 Uso1 Sox10 2010300C02Rik Ypel5 Tshz3 Gja4 Grhl2 Adamts2 Shroom3 St8sia1 Phf21b Ttc28 Sox5 Sox6 Rnpc3 Rgag 4 Zfp668 Atrnl1 Cnot6l2610001J05Ri k Col22a1 Galc Smo Csf2ra Rapgef 4 Epb4.1l2 Ccdc46 Fmo5 Trim45 E130114P18Rik Palmd Glipr2 Gm1943 Cyp2j9 Atp9a Hhat Ets1 Zscan20 Birc2 Egln3 Mpped1 Robo4 Mtx2 Mtap2 Zdhhc21 S1pr3 Fbxo9 2010111I01Rik Maf Fam13c Mcoln3 Prmt6 Nfib Osbpl10 Mapk10 Nfkb1 Mvk Bach1 Zfp703 Tmc6 Mapre2 Eya2 Matk Zfp397 Csnk1g3 Sirpa Tcf4 Snx30 Thnsl1 Asxl3 A230050P20Rik Mll1 Igsf3 F2rl1 Lrrc7 Gad1 Ank3 Gipc1 Grik1 Neurog1Tnfsf13b Matn1 Dscr3 Ndfip2 Cited2Slc24a3 B3galtl Klf7 Ing4 Auts2 Rap2b AI427809 Gpsm3 BC006965 Ripk4 Tmprss2Slmo2 Reep5Tnfrsf19 Ski Shq1Pdlim3 Ssbp3 Dok5 Lyst Arhgap24 Laptm4a Magi1 Nedd4 Insc Stk3 Tia1 Ptpru Slc7a11 Krt7 Cdc123 Sorcs1 Stard13 Myo10 Sorl12610035D17Rik Ppapdc1b Mef2b Ercc5 Kif26a Foxo1 Dnahc11 Cpa6 Parva Cldn1 Snn Helz Mki67 Tmem38b Qpct Leprel1 Rtkn2 Ajap1 Rgs16Fhl2 Pin4 Mllt4 Scara5 Cgnl1 Tshz1 Lifr Frs2 Sort1 Sorbs1 Lypd1 Epm2aip1 Nos3 Nfia Oxr1Col18a1 Tbx2 BC057079 Atf7ip Ppm1b Pard3 Traf4 Cux1 Cdkn2b Fbxl12 G3bp2 Grip1 Sdk1 Trmt1 Phox2b Arhgef16 Tshz2 1810058I24Rik Dlgap1 Pdzd2 Chchd3 Abi3bp Cxcl14 Olfr267 Mgmt Hbs1l4930473A06Rik Gatad2bD030016E14Rik Rimbp2 Wdfy3 Dnajc5bSlc44a1 Tax1bp1 Fam173a Ppp2ca Tnrc6c Rasl12 Jazf1 Efna5 Cnr1 Epha7 Tmco5 Dysf Stat4 Cartpt Rftn1 Nfe2l3 Piwil1 Id2 D16Ertd472e Foxk1

Figure 6: Sox trio-regulated network of genes. Blue dots represent genes activated (green lines) or repressed (red lines) by the transcription factors Sox9, Sox5, and Sox6 (see also Table S5).

regulatory regions proves and supports the hypothesis that on mouse embryonic tissue confirm this observation in vivo in the process of chondrogenesis Sox9 plays a primary and and further detects the presence of Sox5 and Sox6 binding compensatory role with Sox5 and Sox6 complimenting it. sites along with Sox9 near many genes, helping to uncover This may explain why Sox5 and Sox6 single null mutants have new genes controlled by the Sox trio with a potential role in a mild phenotype up to a very late stage in development as chondrogenesis. Sox9 can compensate to a large extent for their loss. One of the defining features of these Sox-bound Sox9 was previously shown to bind to pairs of inverted enhancers that we found is that they are at a fairly large Sox motifs from studies done on campomelic dysplasia distance from the TSS. There have been increasing numbers patients and in cell culture [29, 66]. Our Sox9 ChIP-Seq data of publications from genomewide studies indicating that BioMed Research International 13 most of the transcription factor binding sites identified are Song Jie, Geraldine Leong, Jie Wei Goh, and Valerie Tan for found in the intragenic region rather than at the TSS [79, 86]. technical assistance. Similar binding profiles are observed for other transcription factors in mid gestation mouse embryos (unpublished References data)leadingustobelievethattheprimaryroleforthese transcription factors at these stages is not just to initiate [1] V. Lefebvre, “The SoxD transcription factors—Sox5, Sox6, and transcriptionofdownstreamgenesbutmainlytodirectthe Sox13—are key cell fate modulators,” International Journal of tissue specificity by controlling a subset of their regulatory Biochemistry and Cell Biology,vol.42,no.3,pp.429–432,2010. elements simultaneously in a fine-tuned manner in a way we [2] A. Sarkar and K. Hochedlinger, “The SOX family of transcrip- are beginning to understand. tion factors: versatile regulators of stem and progenitor cell fate,” This study has provided an extensive network of genes Cell Stem Cell,vol.12,no.1,pp.15–30,2013. that are directly and indirectly controlled by Sox9, Sox5, [3] Y.Kamachi and H. Kondoh, “Sox proteins: regulators of cell fate and Sox6 in the context of cell type specification in mouse specification and differentiation,” Development,vol.140,no.20, pp. 4129–4144, 2013. embryos. Evidence supporting the role that the Sox trio plays in this biological process has been previously reported. This [4] S. D. Castillo and M. Sanchez-Cespedes, “The SOX family of however is the first systems biology genomewide in vivo genes in cancer development: Biological relevance and oppor- tunities for therapy,” ExpertOpiniononTherapeuticTargets,vol. analysis done and reported, validating some of the known 16, no. 9, pp. 903–919, 2012. players such as Sox5 and Sox6 while at the same time dis- [5] L.-J. Chew and V. Gallo, “The Yin and Yang of Sox proteins: covering and verifying novel ones like Tle3.Theinformation activation and repression in development and disease,” Journal generated from this extensive study will allow for future work of Neuroscience Research,vol.87,no.15,pp.3277–3287,2009. tobeconductedonsomeofthedownstreamgenesidentified, [6] V. Harley and V. Lefebvre, “Twenty Sox, twenty years,” Interna- paving the way for a comprehensive understanding of the tional Journal of Biochemistry and Cell Biology,vol.42,no.3,pp. vertebrate chondrogenic gene regulatory network that will be 376-377, 2010. insightful in the treatment of human skeletal diseases. [7]S.Ohba,X.He,H.Hojo,andA.P.McMahon,“Distinct transcriptional programs underlie regulation of the mam- Conflicts of Interest malian chondrocyte,” Cell Reports,vol.12,no.2,pp.229–243, 2015. The authors declare that there are no conflicts of interest and [8] M. Kadaja, B. E. Keyes, M. Lin et al., “SOX9: a stem cell trans- no competing financial interests regarding the publication of criptional regulator of secreted niche signaling factors,” Genes this paper. and Development,vol.28,no.4,pp.328–341,2014. [9] H. Akiyama, “Control of chondrogenesis by the transcription factor Sox9,” Modern Rheumatology,vol.18,no.3,pp.213–219, Authors’ Contributions 2008. [10] J. C. Kiefer, “Back to basics: Sox genes,” Developmental Dynam- Thomas Lufkin conceived the project. Wenqing Jean Lee and ics,vol.236,no.8,pp.2356–2361,2007. Sook Peng Yap designed and made the targeting constructs [11] S. P. Henry, S. Liang, K. C. Akdemir, and B. De Crombrugghe, andperformedthegenetargetinginESCforgeneratingthe “The postnatal role of Sox9 in cartilage,” Journal of Bone and transgenic mice, FACS, and microarray profiling. Sumantra Mineral Research,vol.27,no.12,pp.2511–2525,2012. Chatterjee performed the luciferase assays and the zebrafish [12] P. Dy, W. Wang, P. Bhattaram et al., “Sox9 directs hypertrophic work. Wenqing Jean Lee and Siew Lan Lim performed the maturation and blocks differentiation of growth plate tissue ChIP-Seq. Xing Xing assisted in cloning, genotyping, chondrocytes,” Developmental Cell,vol.22,no.3,pp.597–609, cell culture, and FACS. Petra Kraus and Hsiao Yun Chan 2012. performed ESC microinjection. Petra Kraus provided all [13]V.Y.L.Leung,B.Gao,K.K.H.Leungetal.,“SOX9gov- in vivo material. Prasanna R. Kolatkar helped with data erns differentiation stage-specific gene expression in growth interpretation and provided material. V. Sivakamasundari plate chondrocytes via direct concomitant transactivation and performed the histology. Wenjie Sun, Xiaoming Hu, and repression,” PLoS Genetics, vol. 7, no. 11, Article ID e1002356, Shyam Prabhakar conducted the ChIP-Seq analysis. Wen- 2011. jie Sun, Xiaoming Hu, Prasanna R. Kolatkar, and Shyam [14] Z. H. Wang, X. L. Li, X. J. He et al., “Delivery of the Sox9 gene Prabhakar helped with computational analysis. Wenqing Jean promotes chondrogenic differentiation of human umbilical Lee, Sumantra Chatterjee, and Thomas Lufkin wrote the cord blood-derived mesenchymal stem cells in an in vitro manuscript. model,” Brazilian Journal of Medical and Biological Research,vol. 47,no.4,pp.279–286,2014. [15] W. Sun, K. Zhang, G. Liu et al., “Sox9 gene transfer enhanced Acknowledgments regenerative effect of bone marrow mesenchymal stem cells on the degenerated intervertebral disc in a rabbit model,” PLoS This work was supported by the Agency for Science, Tech- ∗ ONE,vol.9,no.4,ArticleIDe93570,2014. nology and Research (A STAR) of Singapore and the Bayard [16] C.-D. Oh, Y. Lu, S. Liang et al., “SOX9 regulates multiple genes and Virginia Clarkson Endowment to Thomas Lufkin. The in chondrocytes, including genes encoding ECM proteins, ECM authors are grateful to Huck Hui Ng for advice on the ChIP modification enzymes, receptors, and transporters,” PLoS ONE, protocol. They thank Patricia Ng for useful discussions and vol. 9, no. 9, Article ID e107577, 2014. 14 BioMed Research International

[17]N.Ahmed,J.Iu,C.E.Brown,D.W.Taylor,andR.A.Kandel, [33] P. Smits, P. Dy, S. Mitra, and V. Lefebvre, “Sox5 and Sox6 are “Serum- and growth-factor-free three-dimensional culture sys- needed to develop and maintain source, columnar, and hyper- tem supports cartilage tissue formation by promoting collagen trophic chondrocytes in the cartilage growth plate,” Journal of synthesis via Sox9-Col2a1 interaction,” Tissue Engineering - Part Cell Biology,vol.164,no.5,pp.747–758,2004. A, vol. 20, no. 15-16, pp. 2224–2233, 2014. [34] Y. Nakamura, K. Yamamoto, X. He et al., “Wwp2 is essential [18] V. Lefebvre and P. Bhattaram, “Vertebrate skeletogenesis,” Cur- for palatogenesis mediated by the interaction between Sox9 rent Topics in Developmental Biology,vol.90,pp.291–317,2010. and mediator subunit 25,” Nature Communications,vol.2,no. [19] A. Superti-Furga, S. Unger, P. Beighton et al., “Nosology and 1, article 251, 2011. classification of genetic skeletal disorders: 2006 Revision,” [35] X. Chen, X. Li, W. Wang, and T. Lufkin, “Dlx5 and Dlx6: American Journal of Medical Genetics, Part A,vol.143,no.1,pp. An evolutionary conserved pair of murine genes 1–18, 2007. expressed in the embryonic skeleton,” Annals of the New York Academy of Sciences,vol.785,pp.38–47,1996. [20] E. Zelzer and B. R. Olsen, “The genetic basis for skeletal dis- eases,” Nature,vol.423,no.6937,pp.343–348,2003. [36] P. Kraus and T. Lufkin, “Mammalian Dlx homeobox gene control of craniofacial and inner ear morphogenesis,” Journal [21]W.Bi,J.M.Deng,Z.Zhang,R.R.Behringer,andB.deCrom- of Cellular Biochemistry, supplement 32-33, pp. 133–140, 1999. brugghe, “Sox9 is required for cartilage formation,” Nature [37] S. Chatterjee, G. Bourque, and T. Lufkin, “Conserved and Genetics,vol.22,no.1,pp.85–89,1999. non-conserved enhancers direct tissue specific transcription in [22] J. W. Foster, M. A. Dominguez-Steglich, S. Guioli et al., ancient germ layer specific developmental control genes,” BMC “Campomelic dysplasia and autosomal sex reversal caused by Developmental Biology, vol. 11, article 63, 2011. mutationsinanSRY-relatedgene,”Nature,vol.372,no.6506, [38] S. Chatterjee, L. Min, R. K. M. Karuturi, and T. Lufkin, “The pp. 525–530, 1994. role of post-transcriptional RNA processing and plasmid vector [23] T. Wagner, J. Wirth, J. Meyer et al., “Autosomal sex reversal and sequences on transient transgene expression in zebrafish,” campomelic dysplasia are caused by mutations in and around Transgenic Research,vol.19,no.2,pp.299–304,2010. the SRY-related gene SOX9,” Cell, vol. 79, no. 6, pp. 1111–1120, [39] P. Kraus, G. Leong, V. Tan et al., “A more cost effective and 1994. rapid high percentage germ-line transmitting chimeric mouse [24] H. Akiyama and V. Lefebvre, “Unraveling the transcriptional generation procedure via microinjection of 2-cell, 4-cell, and 8- regulatory machinery in chondrogenesis,” JournalofBoneand cell embryos with ES and iPS cells,” Genesis,vol.48,no.6,pp. Mineral Metabolism,vol.29,no.4,pp.390–395,2011. 394–399, 2010. [25] T. Ikeda, H. Kawaguchi, S. Kamekura et al., “Distinct roles [40] W. J. Lee, P. Kraus, and T. Lufkin, “Endogenous tagging of of Sox5, Sox6, and Sox9 in different stages of chondrogenic the murine transcription factor Sox5 with hemaglutinin for differentiation,” Journal of Bone and Mineral Metabolism,vol. functional studies,” Transgenic Research,vol.21,no.2,pp.293– 23,no.5,pp.337–340,2005. 301, 2012. [26] T. Ikeda, S. Kamekura, A. Mabuchi et al., “The combination [41] M. Lewandoski, K. M. Wassarman, and G. R. Martin, “Zp3- of SOX5, SOX6, and SOX9 (the SOX trio) provides signals cre, a transgenic mouse line for the activation or inactivation of sufficient for induction of permanent cartilage,” Arthritis and IoxP-flanked target genes specifically in the female germ line,” Rheumatism,vol.50,no.11,pp.3561–3573,2004. Current Biology,vol.7,no.2,pp.148–151,1997. [42] Z.-X. Wang, J. L. L. Kueh, C. H.-L. Teh et al., “Zfp206 is a [27] H.Akiyama,M.-C.Chaboissier,J.F.Martin,A.Schedl,andB.de transcription factor that controls pluripotency of embryonic Crombrugghe, “The transcription factor Sox9 has essential roles stem cells,” Stem Cells,vol.25,no.9,pp.2173–2182,2007. in successive steps of the chondrocyte differentiation pathway and is required for expression of Sox5 and Sox6,” Genes and [43] P. Kraus and T. Lufkin, “Dlx homeobox gene control of mam- Development,vol.16,no.21,pp.2813–2828,2002. malian limb and craniofacial development,” American Journal of Medical Genetics, Part A,vol.140,no.13,pp.1366–1374,2006. [28] P.Smits,P.Li,J.Mandeletal.,“ThetranscriptionfactorsL-Sox5 [44] W.Wang,P.Lo,M.Frasch,andT.Lufkin,“Hmx:anevolutionary and Sox6 are essential for cartilage formation,” Developmental conserved homeobox gene family expressed in the developing Cell,vol.1,no.2,pp.277–290,2001. nervous system in mice and Drosophila,” Mechanisms of Devel- [29] Y. Han and V. Lefebvre, “L-Sox5 and Sox6 drive expression of opment,vol.99,no.1-2,pp.123–137,2000. theaggrecangeneincartilagebysecuringbindingofSox9toa [45] H. Y. Chan, V. Sivakamasundari, X. Xing et al., “Comparison of far-upstream enhancer,” Molecular and Cellular Biology,vol.28, IRES and F2A-based locus-specific multicistronic expression in no. 16, pp. 4999–5013, 2008. stable mouse lines,” PLoS ONE,vol.6,no.12,ArticleIDe28885, [30] A. Nagy, E. Kenesi,´ O. Rentsendorj et al., “Evolutionarily con- 2011. served, growth plate zone-specific regulation of the matrilin-1 [46] C. Tribioli and T. Lufkin, “The murine Bapx1 homeobox gene promoter: L-Sox5/Sox6 and Nfi factors bound near TATA finely plays a critical role in embryonic development of the axial tune activation by Sox9,” Molecular and Cellular Biology,vol.31, skeleton and spleen,” Development,vol.126,no.24,pp.5699– no. 4, pp. 686–699, 2011. 5711, 1999. [31] T. Lufkin, “Transcriptional control of Hox genes in the ver- [47] F. Zhao, T. Lufkin, and B. D. Gelb, “Expression of Tfap2d, the tebrate nervous system,” Current Opinion in Genetics and gene encoding the transcription factor Ap-2𝛿 during mouse Development,vol.6,no.5,pp.575–580,1996. embryogenesis,” Gene Expression Patterns,vol.3,no.2,pp.213– [32] W. Bi, W. Huang, D. J. Whitworth et al., “Haploinsufficiency 217, 2003. of Sox9 results in defective cartilage primordia and premature [48] C. Tribioli, R. F. Robledo, and T. Lufkin, “The murine fork head skeletal mineralization,” Proceedings of the National Academy of gene Foxn2 is expressed in craniofacial, limb, CNS and somitic Sciences of the United States of America, vol. 98, no. 12, pp. 6698– tissues during embryogenesis,” Mechanisms of Development, 6703, 2001. vol. 118, no. 1-2, pp. 161–163, 2002. BioMed Research International 15

[49]D.Sheng,D.Qu,K.H.H.Kwoketal.,“DeletionoftheWD40 [66] P.Bernard,P.Tang,S.Liu,P.Dewing,V.R.Harley,andE.Vilain, domain of LRRK2 in zebrafish causes parkinsonism-like loss “Dimerization of SOX9 is required for chondrogenesis, but not of neurons and locomotive defect,” PLoS Genetics,vol.6,no.4, for sex determination,” Human Molecular Genetics,vol.12,no. Article ID e1000914, 2010. 14, pp. 1755–1765, 2003. [50]S.P.Yap,X.Xing,P.Kraus,V.Sivakamasundari,H.Y.Chan, [67] C.-F. Liu and V. Lefebvre, “The transcription factors SOX9 and and T. Lufkin, “Generation of mice with a novel conditional null SOX5/SOX6 cooperate genome-wide through super-enhancers allele of the Sox9 gene,” Biotechnology Letters,vol.33,no.8,pp. to drive chondrogenesis,” Nucleic Acids Research,vol.43,no.17, 1551–1558, 2011. pp. 8183–8203, 2015. [51] D. Schmidt, M. D. Wilson, C. Spyrou, G. D. Brown, J. Hadfield, [68] V. C. Garside, R. Cullum, O. Alder et al., “SOX9 modulates the and D. T. Odom, “ChIP-seq: using high-throughput sequencing expression of key transcription factors required for heart valve to discover protein-DNA interactions,” Methods,vol.48,no.3, development,” Development (Cambridge),vol.142,no.24,pp. pp.240–248,2009. 4340–4350, 2015. [52] Y. Zhang, T. Liu, C. A. Meyer et al., “Model-based analysis of [69] C. Liu, W. E. Samsa, G. Zhou, and V. Lefebvre, “Transcriptional ChIP-Seq (MACS),” Genome Biology,vol.9,no.9,articleR137, control of chondrocyte specification and differentiation,” Semi- 2008. nars in Cell & Developmental Biology,vol.62,pp.34–49,2017. [70] H.Hojo,A.P.McMahon,andS.Ohba,“Anemergingregulatory [53] C. B. Kimmel, W. W. Ballard, S. R. Kimmel, B. Ullmann, and T. landscape for skeletal development,” Trends in Genetics,vol.32, F.Schilling, “Stages of embryonic development of the zebrafish,” no.12,pp.774–787,2016. Developmental Dynamics,vol.203,no.3,pp.253–310,1995. [71] H. Akiyama, H. S. Stadler, J. F. Martin et al., “Misexpression of [54] M. E. Robu, J. D. Larson, A. Nasevicius et al., “p53 activation Sox9 in mouse limb bud mesenchyme induces polydactyly and by knockdown technologies,” PLoS Genetics,vol.3,no.5,article rescues hypodactyly mice,” Matrix Biology,vol.26,no.4,pp. e78, 2007. 224–233, 2007. [55] W. da Huang, B. T. Sherman, and R. A. Lempicki, “Systematic [72] C.-D. Oh, S. N. Maity, J.-F. Lu et al., “Identification of SOX9 and integrative analysis of large gene lists using DAVID bioin- interaction sites in the genome of chondrocytes,” PLoS ONE, formatics resources,” Nature Protocols,vol.4,no.1,pp.44–57, vol. 5, no. 4, Article ID e10113, 2010. 2009. [73] V.R. Harley, R. Lovell-badge, and P.N. Goodfellow, “Definition [56] M. E. Smoot, K. Ono, J. Ruscheinski, P. L. Wang, and T. Ideker, of a consensus DNA binding site for SRY,” Nucleic Acids “Cytoscape 2.8: new features for data integration and network Research, vol. 22, no. 8, pp. 1500-1501, 1994. visualization,” Bioinformatics,vol.27,no.3,pp.431-432,2011. [74] V.Lefebvre,W.Huang,V.R.Harley,P.N.Goodfellow,andB.De [57] G. Trichas, J. Begbie, and S. Srinivas, “Use of the viral 2A peptide Crombrugghe, “SOX9 is a potent activator of the chondrocyte- for bicistronic expression in transgenic mice,” BMC Biology,vol. specific enhancer of the pro𝛼1(II) collagen gene,” Molecular and 6, article 40, 2008. Cellular Biology,vol.17,no.4,pp.2336–2346,1997. [58]V.Lefebvre,P.Li,andB.DeCrombrugghe,“Anewlong [75] S. Scholpp, A. Delogu, J. Gilthorpe, D. Peukert, S. Schindler, form of Sox5 (L-Sox5), Sox6 and Sox9 are coexpressed in and A. Lumsden, “Her6 regulates the neurogenetic gradient and chondrogenesis and cooperatively activate the type II collagen neuronal identity in the thalamus,” Proceedings of the National gene,” The EMBO Journal,vol.17,no.19,pp.5718–5733,1998. Academy of Sciences of the United States of America,vol.106,no. [59] C. C. Stolt, A. Schlierf, P. Lommes et al., “SoxD proteins 47, pp. 19895–19900, 2009. influence multiple stages of oligodendrocyte development and [76] F. Dayyani, J. Wang, J.-R. J. Yeh et al., “Loss of TLE1 and modulate SoxE protein function,” Developmental Cell, vol. 11, TLE4 from the del(9q) commonly deleted region in AML no. 5, pp. 697–709, 2006. cooperates with AML1-ETO to affect myeloid cell proliferation [60] F. Connor, E. Wright, P. Denny, P. Koopman, and A. Ashworth, and survival,” Blood, vol. 111, no. 8, pp. 4338–4347, 2008. “The Sry-related HMG box-containing gene Sox6 is expressed [77]A.He,S.W.Kong,Q.Ma,andW.T.Pu,“Co-occupancyby in the adult testis and developing nervous system of the mouse,” multiple cardiac transcription factors identifies transcriptional Nucleic Acids Research, vol. 23, no. 17, pp. 3365–3372, 1995. enhancers active in heart,” Proceedings of the National Academy of Sciences of the United States of America,vol.108,no.14,pp. [61] G. Dennis, B. T.Sherman, D. A. Hosack et al., “DAVID:database 5632–5637, 2011. for annotation, visualization, and integrated discovery,” Genome Biol,vol.4,no.5,p.P3,2003. [78] E. Soler, C. Andrieu-Soler, E. De Boer et al., “The genome-wide dynamics of the binding of Ldb1 complexes during erythroid [62] P.Zhang, S. A. Jimenez, and D. G. Stokes, “Regulation of human differentiation,” Genes and Development,vol.24,no.3,pp.277– COL9A1 gene expression: activation of the proximal promoter 289, 2010. region by SOX9,” Journal of Biological Chemistry,vol.278,no.1, [79] M. R. Tijssen, A. Cvejic, A. Joshi et al., “Genome-wide analysis pp. 117–123, 2003. of simultaneous GATA1/2, RUNX1, FLI1, and SCL binding in [63] M. A. Genzer and L. C. Bridgewater, “A Col9a1 enhancer megakaryocytes identifies hematopoietic regulators,” Develop- element activated by two interdependent SOX9 dimers,” Nucleic mental Cell,vol.20,no.5,pp.597–609,2011. Acids Research,vol.35,no.4,pp.1178–1186,2007. [80]X.Fang,J.-G.Yoon,L.Lietal.,“TheSOX2responseprogram [64]T.L.Bailey,M.Boden,F.A.Buskeetal.,“MEMESUITE:tools in glioblastoma multiforme: an integrated ChIP-seq, expression for motif discovery and searching,” Nucleic Acids Research,vol. microarray, and microRNA analysis,” BMC Genomics,vol.12, 37, no. 2, pp. W202–W208, 2009. article 11, 2011. [65] L. C. Bridgewater, M. D. Walker, G. C. Miller et al., “Adjacent [81] O. J. Tamplin, B. J. Cox, and J. Rossant, “Integrated microarray DNA sequences modulate Sox9 transcriptional activation at and ChIP analysis identifies multiple Foxa2 dependent target paired Sox sites in three chondrocyte-specific enhancer ele- genes in the notochord,” Developmental Biology,vol.360,no.2, ments,” Nucleic Acids Research,vol.31,no.5,pp.1541–1553,2003. pp. 415–425, 2011. 16 BioMed Research International

[82] E. Engelen, U. Akinci, J. C. Bryne et al., “Sox2 cooperates with Chd7 to regulate genes that are mutated in human syndromes,” Nature Genetics, vol. 43, no. 6, pp. 607–611, 2011. [83]C.-J.Liu,Y.Zhang,K.Xu,D.Parsons,D.Alfonso,andP.E. Di Cesare, “Transcriptional activation of cartilage oligomeric matrix protein by Sox9, Sox5, and Sox6 transcription factors and CBP/p300 coactivators,” Frontiers in Bioscience,vol.12,no. 10, pp. 3899–3910, 2007. [84] E.Sock,S.D.Rettig,J.Enderich,M.R.Bosl,¨ E. R. Tamm, and M. Wegner, “Gene targeting reveals a widespread role for the high- mobility-group transcription factor Sox11 in tissue remodeling,” Molecular and Cellular Biology,vol.24,no.15,pp.6635–6644, 2004. [85] P. Sohn, M. Cox, D. Chen, and R. Serra, “Molecular profiling of the developing mouse axial skeleton: a role for Tgfbr2 in the development of the intervertebral disc,” BMC Developmental Biology,vol.10,article29,2010. [86] S. R. Holmstrom, T. Deering, G. H. Swift et al., “LRH-1 and PTF1-L coregulate an exocrine pancreas-specific transcriptional network for digestive function,” Genes and Development,vol.25, no. 16, pp. 1674–1679, 2011. International Journal of Peptides

Advances in BioMed Stem Cells International Journal of Research International International Genomics Hindawi Publishing Corporation Hindawi Publishing Corporation Hindawi Publishing Corporation Hindawi Publishing Corporation Hindawi Publishing Corporation http://www.hindawi.com Volume 2014 http://www.hindawi.com Volume 2014 http://www.hindawi.com Volume 2014 Virolog y http://www.hindawi.com Volume 2014 http://www.hindawi.com Volume 2014

Journal of Nucleic Acids

=RRORJ\ International Journal of Hindawi Publishing Corporation Hindawi Publishing Corporation http://www.hindawi.com Volume 201 http://www.hindawi.com Volume 2014

Submit your manuscripts at https://www.hindawi.com

Journal of The Scientific Signal Transduction World Journal Hindawi Publishing Corporation Hindawi Publishing Corporation http://www.hindawi.com Volume 2014 http://www.hindawi.com Volume 2014

Genetics Anatomy International Journal of Biochemistry Advances in Research International Research International Microbiology Research International Bioinformatics Hindawi Publishing Corporation Hindawi Publishing Corporation Hindawi Publishing Corporation Hindawi Publishing Corporation Hindawi Publishing Corporation http://www.hindawi.com Volume 2014 http://www.hindawi.com Volume 2014 http://www.hindawi.com Volume 2014 http://www.hindawi.com Volume 2014 http://www.hindawi.com Volume 2014

Enzyme International Journal of Molecular Biology Journal of Archaea Research Evolutionary Biology International Marine Biology Hindawi Publishing Corporation Hindawi Publishing Corporation Hindawi Publishing Corporation Hindawi Publishing Corporation Hindawi Publishing Corporation http://www.hindawi.com Volume 2014 http://www.hindawi.com Volume 2014 http://www.hindawi.com Volume 2014 http://www.hindawi.com Volume 2014 http://www.hindawi.com Volume 2014