Temporal and spatial requirements for Hoxa3 in mouse embryonic development

Author’s Accepted Manuscript Temporal and spatial requirements for Hoxa3 in mouse embryonic development Jena L. Chojnowski, Heidi A. Trau, Kyoko Masuda, Nancy R. Manley www.elsevier.com/locate/developmentalbiology

PII: DOI: Reference:

S0012-1606(16)30001-X http://dx.doi.org/10.1016/j.ydbio.2016.05.010 YDBIO7117

To appear in: Developmental Biology Received date: 1 January 2016 Revised date: 9 May 2016 Accepted date: 9 May 2016 Cite this article as: Jena L. Chojnowski, Heidi A. Trau, Kyoko Masuda and Nancy R. Manley, Temporal and spatial requirements for Hoxa3 in mouse embryonic development, Developmental Biology, http://dx.doi.org/10.1016/j.ydbio.2016.05.010 This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting galley proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

Title: Temporal and spatial requirements for Hoxa3 in mouse embryonic development Authors: Jena L. Chojnowski1, Heidi A. Trau2, Kyoko Masuda3, and Nancy R. Manley* Department of Genetics, Paul D. Coverdell Center, 500 DW Brooks Drive, University of Georgia, Athens, GA, 30602

*Author for correspondence: [email protected], fax 706-583-0691. 1

Current address: Department of Natural Sciences, University of South Carolina Beaufort, One University Boulevard, Bluffton, SC 29909, USA 2

Current address: Department of Physiological Sciences, Eastern Virginia Medical School, 700 W Olney Road, Lewis Hall, Norfolk, VA 23507, USA 3

Current address: The Institute of Frontier Medical Sciences, Kyoto University, 53 Shogoin-Kawahara-cho, Sakyo-ku, Kyoto-shi, Kyoto, 606-8397 Japan Abstract Hoxa3null mice have severe defects in the development of pharyngeal organs including athymia, aparathyroidism, thyroid hypoplasia, and ultimobranchial body persistence, in addition to defects of the throat cartilages and cranial nerves. Some of the structures altered in the Hoxa3null mutant embryos are anterior to the described Hoxa3 gene expression boundary: the thyroid, soft palate, and lesser hyoid horn. All of these structures develop over time and through the interactions of multiple cell types. To investigate the specific cellular targets for HOXA3 function in these structures across developmental time, we performed a comprehensive analysis of the temporal and tissuespecific requirements for Hoxa3, including a lineage analysis using Hoxa3Cre. The combination of these approaches showed that HOXA3 functions in both a cell autonomous and non-cell autonomous manner during development of the 3rd and 4th arch derivatives, and functions in a neural crest cell (NCC)-specific, non-cell autonomous manner for structures that were Hoxa3-negative by lineage tracing. Our data indicate that HOXA3 is required for tissue organization and organ differentiation in endodermal cells (in the tracheal epithelium, thymus, and parathyroid), and contributes to organ migration and morphogenesis in NCCs. These data provide a detailed picture of where and when HOXA3 acts to promote the development of the diverse structures that are altered in the Hoxa3null mutant. Data presented here, combined with our previous studies, indicate that the regionally restricted defects in Hoxa3 mutants do not reflect a role in positional identity (establishment of cell or tissue fate), but instead indicate a wider variety of functions including controlling distinct genetic programs for differentiation and morphogenesis in different cell types during development.

Key words: HOXA3, temporal regulation, pharyngeal arches, organogenesis, thymus, thyroid

Introduction HOX proteins are a highly conserved family of transcription factors that play essential roles at multiple stages of the regulatory program that defines regional identity and organogenesis during metazoan development. Hox genes are expressed in a regionspecific, combinatorial and partly redundant manner along the anterior-posterior axis of the embryo to specify axial patterning that generates different regional characteristics during development of a tissue (reviewed in Montavon and Soshnikova, 2014). The combination of tissue-specific HOX proteins and other transcription factors leads to the specific activation of downstream genes, including other transcription factors and signaling pathway components (Rezsohazy et al., 2015). HOX proteins work in concert with their targets to regulate development in an organ- and tissue-specific manner within those regions. Classically, HOX proteins are defined by their specification of positional identity, or the establishment of cell fates specific to their location within the developing embryo. This function is tightly linked to their nested expression pattern along the anterior-posterior (A-P) axis during embryonic development, and is expressed in segmented structures via the generation of homeotic transformations after mutation. In addition, HOX expression within a region varies with embryonic time, creating a potential for temporally delimited functions, and functions beyond initial establishment of positional identity. Expression and function have also been reported in later embryonic and adult stages. Examples include the accessory gland in Drosophila (Gligorov et al.), the postnatal brain in Drosophila (Kuert et al.) and mice (Hutlet et al.), and in nonsegmented structures such as blood (Lebert-Ghali et al.), indicating a wide range of potential Hox gene functions beyond positional identity. Thus, relatively broad expression of Hox genes can lead to developmental control on a very fine scale. In vertebrates, the duplication of the HOX clusters has led to both segregated and redundant functions among paralogous HOX genes (reviewed in Duboule, 2007). Hoxa3, the first Hox gene to be mutated in mice by homologous recombination, has two paralogs in mammals, including mice and humans, Hoxb3 and Hoxd3. Single and multiple mutants of these paralogous genes in mice were among the first to be generated, and showed that while single mutants had non-overlapping phenotypes, double and triple mutants for these three genes had progressively more severe phenotypes, including deletions of specific structures (Chisaka and Capecchi, 1991; Condie and Capecchi, 1993; Condie and Capecchi, 1994; Manley and Capecchi, 1995, 1997, 1998). For example, Hoxa3 single mutants have loss of expression of patterning genes and deletions of 3rd pharyngeal pouch-derived organs (Chisaka and Capecchi, 1991; Manley and Capecchi, 1995), and double mutants between any of these paralogs result in deletions of specific vertebral structures (Condie and Capecchi, 1994; Manley and Capecchi, 1997). However, more recent data from our lab showed that the deletion of 3rd pouch-derived organs in Hoxa3 single mutants is a not due to failure to establish regional identity, but is rather a delay in patterning and failure of these organs to survive (Chojnowski et al., 2014). Hoxa3 has an anterior expression limit at the 3rd pharyngeal arch and rhombomere 5 at embryonic day 10.5 (E10.5) in mice. Hoxa3 is expressed broadly in this domain, including endoderm, ectoderm, mesoderm, and neural crest cells (NCCs) (Chisaka and Capecchi, 1991; Chojnowski et al., 2014; Gaufo et al., 2003; Gaunt, 1987, 1988; Kameda

et al., 2004; Kameda et al., 2002; Kameda et al., 2003; Manley and Capecchi, 1995, 1998). Hoxa3null mice, which have been well characterized, have severe defects in the development of pharyngeal organs including athymia, aparathyroidism, thyroid hypoplasia, and ultimobranchial body persistence, in addition to defects of the throat cartilages and cranial nerves (Chisaka and Capecchi, 1991; Kameda et al., 2002; Manley and Capecchi, 1995)(summarized in Table 1). Some of the structures altered in the Hoxa3null mutant embryos (the thyroid, soft palate, and lesser hyoid horn) are anterior to the described Hoxa3 gene expression boundary at E10.5. The soft palate and the lesser hyoid horn develop from the 2nd arch NCCs (Rodriguez-Vazquez et al., 2011) and the thyroid diverticulum develops from the 2nd arch ventral endoderm, although it does migrate through 3rd and 4th arch Hoxa3 positive NCCs before forming the thyroid lobes and isthmus at its final resting place in the adult throat (Fagman and Nilsson, 2011; Romert and Gauguin, 1973). Thus, the origin of these defects and whether they are directly related to Hoxa3 expression or a downstream effect are unclear. The precise spatiotemporal control of Hox gene expression during establishment of the vertebrate A-P axis has been recently linked to cell movements and interactions during gastrulation, specifically linked to Bmp signaling (Durston, 2015; Durston and Zhu, 2015). However, later expression of Hox genes is also dynamic. The spatiotemporal expression pattern of Hoxa3 changes during the development of the pharyngeal region, suggesting distinct tissue-specific and temporal roles for HOXA3 in development. For example, our previous work showed that Hoxa3 expression levels are similar in all expressing tissues in the pharyngeal region at E10.5, but is selectively reduced in the endoderm and pharyngeal pouches compared to the surrounding NCCs by E11.5, while expression in the NCCs does not decline until E12.5 (Fig. 1A-C)(Chojnowski et al., 2014). These data suggest that HOXA3 acts early in development in the endoderm and NCCs and is potentially involved only in NCC-specific functions later in development. For example, HOXA3 in the endoderm temporally regulates early thymus- and parathyroid-specific markers, while HOXA3 in NCCs regulates later events such as organ separation that occur well after cell fate has been established. Although these results showed that HOXA3 does not establish 3rd pharyngeal arch and pouch regional identity, HOXA3 is required in both pharyngeal endoderm and NCCs for thymus and parathyroid survival, and acts independently in each of these two cell types for organ patterning and early organogenesis. In the current study we expanded on these previous findings to perform a comprehensive analysis of the temporal and tissue-specific requirements for HOXA3 in the development of the diverse structures affected in the Hoxa3null mutant phenotype. We evaluated cell type-specific functions of HOXA3 by generating tissue-specific deletions in the endoderm or NCCs. As Hoxa3 has different patterns of expression during the temporal progression of organogenesis, we determined the distinct time windows during which HOXA3 is required for the development of each structure by deleting Hoxa3 globally at different developmental time points. In addition, we tested whether defects in structures anterior to the known Hoxa3 expression pattern were due to a direct role for HOXA3 in these structures by performing a lineage analysis using Hoxa3Cre. The combination of these approaches showed that HOXA3 is required at earlier stages for the development of the most anterior structures that require its function. HOXA3 functions in a largely cell autonomous manner during development of the 3rd and 4th arch derivatives,

indicating that it primarily acts to directly affect the differentiation and behavior of cells in which it is expressed. However, it functions in a cell non-autonomous manner with structures derived from the 2nd pharyngeal arch, soft palate, thyroid diverticulum, and lesser hyoid horn, since these structures were Hoxa3-negative by lineage tracing. Our data also indicate that the cellular functions of HOXA3 differ between endodermal and NC derived cells. HOXA3 expression in endodermal cells affects tissue organization and organ differentiation (in the tracheal epithelium, thymus, and parathyroid), while in NCCs it regulates morphogenesis, including migration and tissue separation/fusion events of the IXth cranial ganglion (CG) and pharyngeal organs. These data provide a detailed picture of both where and when an individual HOX gene acts to promote the development of diverse structures across embryonic time. Materials and Methods Generation of Mice and Genotype Analysis CAGG-CreERTM (Hayashi and McMahon, 2002), Wnt1-Cre (Danielian et al., 1998), Foxa2CreERT2 (Park et al., 2008), R26R (Soriano, 1999), Hoxa3Cre (Macatee et al., 2003), Gcm2-EGFP (Gong et al., 2003), Hoxa3null (Chisaka and Capecchi, 1991; Chojnowski et al., 2014), and Hoxa3fx conditional allele (Chojnowski et al., 2014) strains were maintained and genotyped as described. All colonies were maintained on a majority C57BL6/J genetic background. Yolk sac or tail DNA was genotyped by PCR. Embryos or newborns carrying the tissue-specific deletion (fx/-;Cre/+) were termed „mutant‟. Conditional mutants were generated by crossing Hoxa3fx/fx females with Hoxa3+/-; -Cre/+ males. R26R females were crossed with Hoxa3Cre males to generate mice in which Bgal is expressed in all Hoxa3-expressing cells and their progeny. Embryonic age was estimated by considering noon of the day of a vaginal plug as E0.5; stages were confirmed by somite number and morphology. To stimulate nuclear translocation of the inducible Cre, pregnant mice were treated with single intraperitoneal injection of tamoxifen (Sigma) with a dose of 3mg tamoxifen (Sigma-Aldrich) per 40g mouse in sterile corn oil (Sigma-Aldrich) at E7.5-E12.5 for the CAGG-CreERTM mice or at E6.5 for the Foxa2CreERT2 mice. There were no gross abnormalities in wild type embryos after TI. Both +/fx; Cre/+ and +/fx;+/+ embryos were used as controls (Chojnowski et al., 2014) and were indistinguishable in all assays. The presence of the deleted and undeleted Hoxa3fx alleles was detected by PCR using primers specific to the floxed allele that distinguish these two alleles (Chojnowski et al., 2014) every six hours up to 24 hours from extracted whole embryo genomic DNA used as a template. All experiments involving animals were carried out with the approval of the UGA Institutional Animal Care and Use Committee. Histology and immunohistochemistry Standard hematoxylin and eosin (H&E) histological staining was performed on transverse paraffin sections of embryos and newborns. Immunohistochemistry (IHC) was performed on sectioned (7µm) paraffin-embedded or frozen embryos fixed in 4% PFA. Paraffin-embedded tissue was washed in xylene and rehydrated through an ethanol gradient to dH2O, then boiled in antigen-retrieval buffer (10mM Na3Citrate pH6, 0.05% Tween20) for 30 minutes. Slides were incubated overnight in primary antibody, 10% donkey serum, and 0.05% Triton-X in PBS at 4°C. Slides were washed in PBS, then

incubated with secondary antibody in PBS for 30 minutes at room temperature in the dark. Slides were washed twice in PBS containing 1:10,000 dilution of DAPI, and then mounted with EMS-Fluorogel. Antibodies recognized FOXN1 (Santa Cruz G-20) or GCM2 (Abcam). Secondary antibodies were DyLight-conjugated (Jackson Immunoresearch). E11.5 embryos were collected in PBS for neurofilament immunostaining following (Davis et al., 1991). In short, they were fixed in 1ml methanol:DMSO (4:1) overnight at 4°C. Fixative solution was replaced with methanol:DMSO:30%H2O2 (4:1:1) for 4 hours at room temperature and then stored at -20°C in 100% methanol. Embryos were rehydrated in a PBST:methanol series and blocked by incubating twice in PBST:2% skim milk:1% DMSO (PBSTMD) for 1 hr at room temperature. Embryos were incubated overnight at 4°C in a 4:1 ratio of PBSTMD and neurofilament monoclonal antibody ZH3 cell supernatant (DSHB) and an HRP-conjugated goat anti-mouse secondary antibody (Jackson Immunobiologicals) diluted 1:100 in PBSTMD subsequently. Embryos were washed with PBSTMD between the primary and secondary antibody incubations. Color development occurred after 1 hr in DAB and then 0.03% H2O2 in PBST for 1-2 minutes. Embryos where then rinsed in PBST:0.2%BSA, dehydrated through a methanol series, and then put in BABB before imaging. X-gal staining X-gal staining was performed on embryos from multiple stages of Hoxa3Cre and R26R crosses as described (DasGupta and Fuchs, 1999). In short, frozen sections (10um) were fixed with 0.2% glutaraldehyde for 2 min and then washed with PBS. Slides were then placed in stain solution with the X-Gal substrate at a final concentration of 1mg/ml for 6hr-overnight at 37°C. Imaging of slides Images were acquired using a Zeiss LSM510 confocal microscope and Zeiss image acquisition software. Between 2 and 5 embryos were analyzed for each stage and genotype; representative images are shown. N-values for each experiment are provided in the text and figure legends. Skeletal preparations Skeletons have skin and internal organs removed before fixation in 95% ethanol for 5 days. Lipids are removed in acetone for 2 days and then the skeletons were stained with an alcian blue 8GX and alizarin red S solution for 10 days at 37°C. After rinsing with water the skeletons were cleared in a 30% borax:1% trypsin solution at 37°C for 4-6 hours and then put through a 1% KOH:glycerol dilution series over 2-3 weeks until they reached 100% glycerol for storage. Skeletons were imaged on a Zeiss SteREO Discovery.V8 with SPOT Imaging SolutionsTM. Results Temporal or tissue-specific Cre-mediated deletion of the Hoxa3 gene The tissue and temporal-specific requirements for Hoxa3 during embryogenesis were resolved for the structures with known mutant phenotypes in the Hoxa3null mouse.

We previously described a Hoxa3 conditional allele (Hoxa3fx/fx) in which the second exon is flanked by LoxP sites producing a null Hoxa3 allele after Cre-mediated recombination (Chojnowski et al., 2014). We used a transgenic mouse line (CAGG-CreERTM) that produces a ubiquitously expressed tamoxifen-inducible CreERTM protein (Hayashi and McMahon, 2002) in conjunction with Hoxa3fx/fx for temporal clarification of Hoxa3. Also, we determined the endoderm and neural crest cell (NCC)-specific roles of HOXA3 with Hoxa3fx/fx and the transgenic mouse lines Foxa2CreERT2 or Wnt1-Cre, respectively (Chen et al., 2010; Chojnowski et al., 2014). First we tested the timing and efficiency of the CAGG-CreERTM -induced recombination of the Hoxa3fx allele by crossing male CAGG-CreERTM mice to female Hoxa3fx/fx mice and harvested embryos at 0, 6, 12, 18, and 24 hours after a single intraperitoneal tamoxifen injection (TI) to pregnant females at E10.5. After 6 hours, the floxed allele was clearly reduced and was only barely detectable by 18 hours post TI, and undetectable by 24 hours (Fig. 1B). Allowing an additional six hours for any residual HOXA3 protein to degrade (Beslu et al., 2004; Sandoval et al., 2013; Sharova et al., 2009; Yang et al., 2003), and given the inherent variability in plug time and different fertilization times for different embryos within a litter, we concluded that for the purpose of this study, HOXA3 is essentially absent 24-36 hours after TI. For example, we interpreted our results so that any structure that had a mutant phenotype after an E9.5 TI time point required HOXA3 before E11.0 for normal development. To establish the temporal-specific requirements for HOXA3, we induced Cre recombinase activity by TI in pregnant Hoxa3fx/fx females that had been crossed with Hoxa3+/-;CAGG-CreERTM males at each day between E7.5-E12.5 and examined embryos at E11.5, E12.5, E15.5, and/or E18.5. In addition, to determine the tissue-specific roles of HOXA3 we examined E11.5 and E18.5 embryos from Hoxa3fx/fx pregnant females crossed to either Hoxa3+/-;Wnt1-Cre males (for NCCs), or Hoxa3+/-;Foxa2CreERT2 males (for the endoderm) (Chen et al., 2010; Chojnowski et al., 2014). Endoderm-specific deletion was induced by TI at E6.5. Embryos that contained both Cre alleles, and thus had both tissues deleted in the same embryo (double deletion) were examined at E18.5. Finally, we used embryos in which Hoxa3Cre was used to activate the R26R lacZ indicator to determine the lineage history of cells expressing Hoxa3 at any time prior to that of analysis. Hoxa3 plays multiple roles in thymus and parathyroid survival and morphogenesis We previously showed that Hoxa3 expression is necessary for the survival of the thymus and parathyroids; both organs undergo coordinated apoptosis before E13.5 in null embryos (Chojnowski et al., 2014). To determine when HOXA3 is necessary for thymus and parathyroid survival we tested for the presence of the organs at E12.5, E15.5, or E18.5 after deleting Hoxa3 at specific times (Table 1). At a TI time point of E8.5 or earlier, the thymus and parathyroids were absent, similar to the null (not shown). We examined four E9.5 TI time point embryos. Two embryos exhibited the null phenotype of athymia and aparathyroidism (Fig. 2A), one had both organs still attached to each other and the thymus still attached to the pharynx (not shown), and one embryo lacked the left thymus lobe and parathyroid while the right thymus and parathyroid were present but attached to each other (Fig. 2E) and to the pharynx (not shown). We never saw the presence of one organ without the other on the same side. These results indicate that there

is a sharp temporal boundary for the requirement for Hoxa3 for thymus and parathyroid survival at about E11. In addition to thymus and parathyroid survival, HOXA3 is involved in the processes of thymus-pharynx separation and organ migration. HOXA3 is independently required in endoderm and NCC for thymus and parathyroid separation and migration (Chojnowski et al., 2014; Gordon and Manley, 2011). The thymus detaches from the pharynx via coordinated endodermal apoptosis induced by signals from the surrounding Hoxa3-positive NCCs between E11.5 and E12.5. The parathyroid and thymus lobes separate from each other between E12.5-E13.5, during migration, by a process that does not involve apoptosis (Chojnowski et al., 2014; Gardiner et al., 2012; Gordon et al., 2004). We determined when HOXA3 is necessary for these multiple functions by deleting it at different time points (Table 1). When deletion was induced by TI at E11.5 the thymus and parathyroids were normal, consistent with our previous data that showed deleting Hoxa3 from the developing thymus with a Foxn1Cre driver at E11.5 results in normal thymus development (Chojnowski et al., 2014). At the E10.5 TI time point, both organs were present but displayed a suite of defects in separation and subsequent migration (Fig. 2B,C,F,G). We analyzed 11 embryos, for a total of 22 thymic lobes. Three right lobes had normal positioning with the left lobes in ectopic positions (Fig. 2B), seven were still attached to the pharynx (Fig. 2G), and 12 were detached from the pharynx but positioned in ectopic locations near the pharynx (Fig. 2F). All of the mutant embryos that exhibited asymmetric thymus migration had ectopic left thymic lobes, consistent with our previously published data showing that the left thymus lobe separates from the pharynx and migrates slightly later than the right lobe (Gordon et al., 2010). For each thymus the parathyroid had a similar phenotype; if the thymus was still attached to the pharynx then the parathyroid was still attached to the thymus, and if the thymus had partially migrated then the parathyroid was also ectopic (Fig. 2E-G). No embryo had both extremes of the spectrum, with one attached to the pharynx and one that had completed migration, which implies that the timing of the separation of the two primordia from the pharynx is relatively close. To better resolve the timing of separation and migration, we examined an E11.0 TI time point (Fig. 2D,H). Of four mutant embryos, two left thymi were ectopic, none still attached, and the rest normal. These data show that the temporal requirement for HOXA3 during thymus-pharynx separation and organ migration is between E12-E12.5, which overlaps the initial stages of normal thymus-pharynx separation (E11.5-E12.5), but is prior to thymus-parathyroid separation and migration (E12.5-E14.5). Endodermal Hoxa3 is required before E13 for development of the tracheal epithelial lining The epithelial lining of the trachea is derived from Hoxa3-positive cells of the ventral endodermal epithelium of the embryonic pharynx (Fig. 3A)(Gaunt, 1988), and in the Hoxa3null embryos, the epithelium is thicker than normal and its layers of cells were disorganized compared to a well organized single layer. In addition, the tracheal epithelium appeared to have lost its columnar epithelial morphology (Fig. 13 in Manley and Capecchi, 1995). To determine when HOXA3 is necessary to organize the tracheal epithelial lining and if HOXA3 is necessary only in the endoderm we deleted Hoxa3 at different time points specifically within endoderm, NCCs, or in both cell types during

embryogenesis (Table 1). The same mutant phenotype of a disorganized, thicker epithelium was seen in embryos from E11.5 or earlier TI time points, in the endoderm deletion mutant (Fig.3 F,I), and also in the double deletion mutant (data not shown). The tracheal epithelium showed a normal phenotype at E12.5 TI time points and in the NCC mutants. These data indicate that HOXA3 is necessary in the endoderm for correct tracheal epithelium organization before E13.0 (Fig. 3G,H). Proper fusion of the ultimobranchial body and thyroid lobe requires Hoxa3 in NCCs before E12. The ultimobranchial body is formed from the 4th pharyngeal pouch endoderm, then migrates to and fuses with the thyroid diverticulum, which also migrates ventrally from its original position on the pharynx floor at the 2nd pharyngeal arch (Romert and Gauguin, 1973). The thyroid anlage buds at E9.5-E10.5 and then migrates to its final position in the throat before growing laterally into a bilobed organ with a connecting isthmus between E11.5-E13.5 (Fagman and Nilsson, 2011). The ultimobranchial body starts to detach from the pharynx at E11.5, migrates to the thyroid diverticulum by E13.5, and then the lateral lobes of the thyroid diverticulum and the ultimobranchial bodies fuse at E13.5-E14.5 (Fagman and Nilsson, 2011). After fusion, the thyroid diverticulum gives rise to the thyroxin-producing cells of the thyroid, and the ultimobranchial body-derived cells distribute throughout the organ giving rise to the parafollicular cells of the thyroid that produce calcitonin (Ozaki et al., 2011). Lineage tracing of Hoxa3-expressing cells was completed to determine if Hoxa3 was expressed in the ultimobranchial bodies, the thyroid diverticulum, or both during development. Our results showed that the ultimobranchial bodies but not the thyroid diverticulum were labeled at E13.5 (Fig. 3B). At E18.5, the thyroid lobes contained scattered Hoxa3-lineage cells of ultimobranchial body origin, while the thyroid-diverticulum-derived follicular cells and the thyroid isthmus were negative for the Hoxa3 lineage tracer (Fig 3C-D). Therefore, Hoxa3expressing cells are only found in the ultimobranchial body and the cells that derive from the ultimobranchial body after fusion. The Hoxa3null mutants, exhibited multiple thyroid phenotypes. These included missing or reduced thyroid isthmus, thyroid hemiagenesis, failure of the ultimobranchial body to fuse with the thyroid lobes, and cystic ultimobranchial bodies, although both components formed and the ultimobranchial body migrated properly (Manley and Capecchi, 1995, 1998). To determine when HOXA3 is involved in the different aspects of thyroid development, including thyroid migration, fusion, and isthmus integrity we deleted Hoxa3 at different time points in development (Table 1). The ultimobranchial body null mutant phenotype was seen in embryos from TI time points E10.5 or earlier and in the NCC deletion embryos (Fig. 3J-N). TI up to E9.5 resulted in two out of 16 missing right thyroid lobes but most were primarily normal (14 of 16) (Fig. 3L ), and the thyroid isthmus was missing (4 of 8), reduced (3 of 8) or normal (1 at TI E9.5). Two embryos exhibited normal, bilateral thyroid lobes but were missing the isthmus. The other missing thyroid isthmuses were in embryos that were also missing a thyroid lobe. Hoxa3-lineage cells were not found in the endodermal cells of the thyroid diverticulum. However, the thyroid does migrate through Hoxa3-positive NCCs, raising the possibility that mutant phenotypes could appear in the NCC or double deletion embryos but not the endoderm deletion embryos (Fig. 3B). Our results showed that there

were no thyroid abnormalities in NCC deletion (n=10) endoderm deletion (n=8), or double deletion (n=2) embryos (Table 1). These data suggest that this phenotype requires absence of HOXA3 from both cell types simultaneously, similar to our previous data on survival of the thymus (Chojnowski et al.). However, the number of double deletion mutant embryos was too small to potentially see the full range of phenotypes. Based on these results, we propose that HOXA3 is required for thyroid development before E11.0, and acts redundantly in endoderm or NCCs before E12.0 for proper fusion of the ultimobranchial body and the thyroid primordium, well before fusion starts to occur at E13.5. Hoxa3 expression in NCCs drives soft palate, epiglottis, and IXth nerve development The soft palate develops from the 2nd arch NCCs that are Hoxa3 negative at E10.5, while the epiglottic bud develops from a mesenchymal condensation derived from the 3rd and/or 4th arch NCCs that are Hoxa3 positive at E10.5 (Chojnowski et al., 2014; Katori et al., 2011). However, both structures were truncated in the Hoxa3null mutant (Chisaka and Capecchi, 1991). We used lineage tracing to determine whether the known defects in these structures in the Hoxa3null mutants are due to direct or indirect roles for Hoxa3. Hoxa3-lineage cells were present in the epiglottis but not the soft palate at E18.5 (Fig. 4A). Hoxa3-lineage cells were present at the fusion point between the soft palate and the nasal cavity (Fig. 4B-D); however, the fusion point developed normally in all mutants analyzed (Fig. 4F-H). Hoxa3 is necessary for normal soft palate development between E9.0 and E12.0, since TI at E10.5 resulted in a normal phenotype, but an E7.5 TI time point resulted in a mutant phenotype (Table 1)(Fig. 4I,J,M,N). Furthermore, deletion of Hoxa3 in NCCs but not in the endoderm resulted in similar phenotypes to the null in both the palate and epiglottis, consistent with their NCC origins (Table 1)(Fig. 4K,L,O,P). These results indicated that both the cell-autonomous epiglottal and non-cell-autonomous palate phenotypes are due to the activity of Hoxa3 in NCCs. The Hoxa3null mutant has a reduced IXth cranial ganglion (CG), which is fused to the Xth CG or truncated and disconnected from its dorsal nerve bundle (Chisaka and Capecchi, 1991; Manley and Capecchi, 1997)(Fig. 5B). The IXth CG originates from the 3rd arch NCCs and placode-derived cells (Barlow, 2002). Hoxa3 is required for the migration of the placode-derived cells, leading to the suggestion that these migration defects are the origin of IXth CG defects in Hoxa3null mutants (Watari et al., 2001). We tested whether Hoxa3 expression in NCC or endoderm is required for the Hoxa3null IXth CG mutant phenotype (Table 1). In the NCC deletion mutant the IXth CG was disconnected from its dorsal nerve bundle in 47% of mutant embryos examined, or fused to the Xth CG in 34% of the mutant embryos (n=32), similar to that seen in the Hoxa3null mutant (54% and 29% respectively; n=24) (Fig. 5). All embryos with endoderm-specific deletion of Hoxa3 showed a normal phenotype (n=6) (data not shown). These data show that Hoxa3 expression in NCCs is required for normal CG formation. This phenotype was not examined by temporal deletion. Hoxa3 is required throughout the development of the pharyngeal skeleton In the Hoxa3null mutant, the lesser horns of the hyoid bone are reduced or missing and the greater horn of the hyoid is fused with the superior horn of the thyroid cartilage (Condie and Capecchi, 1994; Manley and Capecchi, 1997)(Fig. 6E,F). The 2nd arch

NCCs give rise to the lesser horn of the hyoid bone while the rest of the hyoid bones originate from the third arch NCCs, and the throat cartilages arise from the fourth arch NCCs (Rodriguez-Vazquez et al., 2011). Since Hoxa3 is not expressed in the 2nd arch NCCs at E10.5 (Chojnowski et al., 2014), we performed lineage tracing to see if any Hoxa3-lineage cells incorporate into the lesser horn of the hyoid bones. Hoxa3-lineage cells were not present in 2nd arch NCCs at E10.5 or in the lesser horn at E18.5, although they were found throughout the greater horn and the thyroid cartilages (Fig. 6A-C). Consistent with these skeletal elements being NCC-derived, the NCC deletion resulted in mutant phenotypes similar to the early global Hoxa3 deletions (Fig. 6M,N). There is also strong evidence for an endodermal role in pharyngeal cartilage patterning and formation (Couly et al., 2002; Crump et al., 2004; Peters et al., 1998; Ruhin et al., 2003). However, these skeletal elements were normal after endoderm-specific deletion of Hoxa3, indicating that Hoxa3 is not required for this endodermal function. Analysis of the hyoid bones and thyroid cartilage after deleting Hoxa3 in early development (TI times of E7.5-E12.5) showed that there was not a clear temporal boundary between the Hoxa3null and wild type phenotypes (Table 1). Instead, there was a progressively stronger mutant phenotype as deletion of Hoxa3 occurred earlier (Fig. 6GL). An E12.5 TI time point generated a phenotypically normal pharyngeal skeleton, while E10.5 and E11.5 TI time points generated normal lesser horn development with some embryos exhibiting projections from the greater horn of the hyoid bone that were never seen in the control or null embryos (Fig. 6J-L). At the E9.5 TI time point, the greater horn was completely fused to the thyroid cartilage, as in the Hoxa3null and the lesser horn also failed to form correctly (Fig. 6I). The posterior portion of the lesser horn was missing, and the resulting structure consisted of the thin projection from the body of the hyoid ending in a ball of cartilage. This phenotype was further exacerbated in the embryos with TI time points of E7.5 and E8.5, which had only a thin projection of cartilage from the body of the hyoid ending in a very small ball of cartilage and extra connections to the greater hyoid (Fig. 6G,H). No mutant embryos from any temporal deletion completely resembled the Hoxa3null phenotype (Fig. 6E,F). Overall, these data demonstrate that HOXA3 is required before E10.0 for the lesser horn to separate from the greater horn, before E11.0 for the correct development of the posterior lesser horn and correct separation of the greater horn and the thyroid cartilage, and before E13.0 for correct greater horn morphology. Discussion This study presents the most comprehensive analysis of the tissue-specific and temporal requirements for a vertebrate HOX gene performed to date. Our results show that Hoxa3 plays multiple tissue- and temporal-specific roles in pharyngeal region development, consistent with the diversity of structures affected in the Hoxa3null and the complexity of its expression pattern during development. The combination of lineage tracing and tissue-specific deletion demonstrated that Hoxa3 has both cell autonomous and cell non-autonomous roles, and is required in both NCC and endoderm for specific functions. For example, 2nd pharyngeal arch-derived structures affected in the Hoxa3null never express Hoxa3. Therefore, the defects in these structures are due to a cell nonautonomous role for HOXA3. In contrast, Hoxa3 is required cell-autonomously in the

tissue of origin for the development of most structures derived from the 3rd and 4th pharyngeal arches, which corresponds to its most anterior limit of expression. Hoxa3 is required earlier for structures derived from the 2nd pharyngeal arch than for those structures derived from the 3rd and 4th pharyngeal arches (Fig. 7). In most cases the role of Hoxa3 is straightforward. For example, it has a cell autonomous, endoderm-specific role in the organization of the tracheal epithelium lining, and a cell autonomous, NCCspecific role in IXth CG development. In some cases, though, its role appears to be more complex. For example, in ultimobranchial body-thyroid lobe fusion it has a cell nonautonomous function that is needed days prior to the actual event, as global temporal deletions showed that it is required prior to E12 for a fusion event that occurs after E13.5. In addition, in the pharyngeal skeleton Hoxa3 appears to have multiple functions over several days of development. Global temporal deletions generated morphologically different and progressively more severe phenotypes with earlier deletion time points, rather than displaying a clear cutoff beyond which Hoxa3 was no longer required. . Hoxa3-negative structures have mostly NCC-specific phenotypes Several of the structures affected in the Hoxa3 null mutant are derived from the 2nd pharyngeal arch, which does not express Hoxa3 at E10.5. These structures include the lesser horn of the hyoid, the soft palate, and the thyroid lobes and isthmus. Our lineage tracing showed that these structures never express Hoxa3 and so these defects are due to a cell non-autonomous function. The lesser horn of the hyoid and the soft palate exhibit defects when Hoxa3 is specifically deleted from NCCs, or is globally deleted before E11.0. Therefore, the development of these structures requires Hoxa3 expression in more posterior NCC prior to E11. These results reveal a required interaction between 2nd arch and more posterior NCC for the proper formation of these skeletal elements. The mechanism underlying these interactions remains to be determined. Regulation of signaling molecules that act on neighboring tissues is one obvious possibility. A recent study has also provided evidence for HOX proteins being able to directly move between neighboring cells, then inducing its own expression, “copying” Hox gene expression (Bardine et al.). Our lineage analysis would identify any cell that has turned on the locus at any point in its history, whether that happened by patterning mechanisms or by HOX protein transfer. Nonetheless, if the HOXA3 protein were transferred without upregulating its own expression, we could not exclude that possibility. Deleting Hoxa3 from either NCCs or the endoderm results in normal thyroid development, while deleting Hoxa3 globally prior to E11 results in thyroid defects. These data suggest that Hoxa3 has a redundant function in these two cell types for thyroid development. We cannot discount the possibility that Hoxa3 expression in another tissue could be involved in the thyroid phenotype, but there is no known data supporting involvement of other cell types in thyroid development. A similar form of HOXA3 redundant function in the NC and endoderm also occurs in thymus survival, where deletion from either cell type results in thymus survival, but deletion in both cell types recapitulates the null phenotype (Chojnowski et al., 2014). In the thymus, we proposed that HOXA3 could promote the expression of survival signals in both the endoderm and NCCs, with either source being sufficient for survival. As the thyroid phenotype does not appear to be due to cell death, it is at this point unclear what this redundant Hoxa3 function could be.

Soft palate truncation is often a secondary defect due to malformations of other facial structures, but HOXA3 appears to be required specifically for secondary palatogenesis since no other facial malformations are seen in Hoxa3 mutants. The soft palate forms from cell migration and proliferation of the palatal shelves towards the midline and then mesenchymal-directed fusion of the epithelium between E12.5 and E15.5 (Bush and Jiang, 2012). Since Hoxa3 is not expressed in the migrating and proliferating cells during palatogenesis but instead in the original cells that form the lateral edges, Hoxa3 most likely has a cell non-autonomous role for the migration and proliferative growth of the palatal shelves and this function occurs before medial growth begins. Hoxa3 in NCC regulates organ separation Thymus-parathyroid separation occurs via NCCs intercalating between the two organs (Gordon et al., 2010). Our data shows that Hoxa3-expressing NCCs are necessary for this process. In the NCC-specific deletion of Hoxa3 there are no organized mesenchymal cells that insert between the two organs, similar to what is seen in the Foxg1-Cre;Bmp4 mutants that also have a connected organ phenotype (Gordon et al., 2010). However, Hoxa3 regulates Bmp4 expression in the endoderm, but not in NCCs (Chojnowski et al., 2014). Therefore, this NCC function for HOXA3 is likely independent of BMP4. Hoxa3-specific thyroid defects are cell non-autonomous Our phenotypic data is consistent with a model in which HOXA3 regulates separation of the thyroid diverticulum from the pharynx floor in a cell non-autonomous manner. The mechanism for thyroid separation from the pharynx floor is unknown but there has been progress in determining the molecular factors involved. Shhnull and Tbx1null mice show the same left lateralization of the thyroid that we see in some of our Hoxa3null and early TI Hoxa3 deletion mice (Alt et al., 2006; Fagman et al., 2004; Fagman et al., 2007). Shh, SHH signaling, Tbx1, and Hoxa3 are not expressed in the developing thyroid and only Tbx1 is found in the surrounding mesenchyme. The Hoxa3 mutant embryos more closely resemble the Tbx1null embryos since they both have correct development of the arteries (Fagman et al., 2007; Washington Smoak et al., 2005), while the Shhnull embryos do not. These arteries provide the framework necessary for correct migration of the thyroid diverticulum and without them the thyroid forms only a single, lateralized lobe. Shh and Hoxa3 independently regulate Tbx1 in the 3rd pharyngeal pouch (Chojnowski et al., 2014; Grevellec et al., 2011; Moore-Scott and Manley, 2005). Our results here provide evidence that these independent pathways may also control the development of the thyroid through Tbx1, although both are acting in a distant, early, and cell non-autonomous manner. Our data suggest a model in which HOXA3 directs the timing of the thyroid diverticulum separation from the pharynx floor, which indirectly disrupts migration and morphogenesis, resulting in a single, ectopic left lobe. The only cell-autonomous function of Hoxa3 that affects thyroid development is via its role in ultimobranchial body fusion to the thyroid diverticulum-derived lobes. The thyroid diverticulum cells engulf the ultimobranchial body and then the ultimobranchial body cells diffuse within the now combined organ (Fagman and Nilsson, 2011). Hoxa3positive NCCs are necessary for the fusion of the thyroid diverticulum lobes and the

ultimobranchial body in a cell non-autonomous manner. However, this requirement is days before fusion occurs. HOXA3 is not involved in ultimobranchial body development or migration since both are normal in the Hoxa3null mutant. The temporal displacement between the requirement for HOXA3 and the event itself suggest that HOXA3 may set in motion a cascade of gene expression whose downstream effect occurs much later. This phenotype will therefore require further knowledge of the molecular and cellular mechanisms underlying the fusion event before the specific role of HOXA3 may be understood. Acknowledgements We thank Julie Gordon and Brian Condie for project advice. We thank Brian Condie, James Lauderdale and Scott Dougan for helpful edits of the manuscript. This project was supported by funds provided to NRM from the UGA Office of the Vice President for Research.

References Alt, B., Elsalini, O.A., Schrumpf, P., Haufs, N., Lawson, N.D., Schwabe, G.C., Mundlos, S., Gruters, A., Krude, H., Rohr, K.B., 2006. Arteries define the position of the thyroid gland during its developmental relocalisation. Development 133, 3797-3804. Bardine, N., Lamers, G., Wacker, S., Donow, C., Knoechel, W., Durston, A., 2014. Vertical signalling involves transmission of Hox information from gastrula mesoderm to neurectoderm. PLoS One 9, e115208. Barlow, L.A., 2002. Cranial nerve development: placodal neurons ride the crest. Curr Biol 12, R171-173. Beslu, N., Krosl, J., Laurin, M., Mayotte, N., Humphries, K.R., Sauvageau, G., 2004. Molecular interactions involved in HOXB4-induced activation of HSC selfrenewal. Blood 104, 2307-2314. Bush, J.O., Jiang, R., 2012. Palatogenesis: morphogenetic and molecular mechanisms of secondary palate development. Development 139, 231-243. Chen, L., Zhao, P., Wells, L., Amemiya, C.T., Condie, B.G., Manley, N.R., 2010. Mouse and zebrafish Hoxa3 orthologues have nonequivalent in vivo protein function. Proc Natl Acad Sci U S A 107, 10555-10560. Chisaka, O., Capecchi, M.R., 1991. Regionally restricted developmental defects resulting from targeted disruption of the mouse homeobox gene hox-1.5. Nature 350, 473-479. Chojnowski, J.L., Masuda, K., Trau, H.A., Thomas, K., Capecchi, M., Manley, N.R., 2014. Multiple roles for HOXA3 in regulating thymus and parathyroid differentiation and morphogenesis in mouse. Development 141, 3697-3708. Condie, B.G., Capecchi, M.R., 1993. Mice homozygous for a targeted disruption of Hoxd-3 (Hox-4.1) exhibit anterior transformations of the first and second cervical vertebrae, the atlas and the axis. Development 119, 579-595. Condie, B.G., Capecchi, M.R., 1994. Mice with targeted disruptions in the paralogous genes hoxa-3 and hoxd-3 reveal synergistic interactions. Nature 370, 304-307. Couly, G., Creuzet, S., Bennaceur, S., Vincent, C., Le Douarin, N.M., 2002. Interactions between Hox-negative cephalic neural crest cells and the foregut endoderm in patterning the facial skeleton in the vertebrate head. Development 129, 1061-1073. Crump, J.G., Swartz, M.E., Kimmel, C.B., 2004. An integrin-dependent role of pouch endoderm in hyoid cartilage development. PLoS Biol 2, E244. Danielian, P.S., Muccino, D., Rowitch, D.H., Michael, S.K., McMahon, A.P., 1998. Modification of gene activity in mouse embryos in utero by a tamoxifen-inducible form of Cre recombinase. Curr Biol 8, 1323-1326. DasGupta, R., Fuchs, E., 1999. Multiple roles for activated LEF/TCF transcription complexes during hair follicle development and differentiation. Development 126, 4557-4568. Davis, C.A., Holmyard, D.P., Millen, K.J., Joyner, A.L., 1991. Examining pattern formation in mouse, chicken and frog embryos with an En-specific antiserum. Development 111, 287-298. Duboule, D., 2007. The rise and fall of Hox gene clusters. Development 134, 2549-2560. Durston, A.J., 2015. Time, space and the vertebrate body axis. Semin Cell Dev Biol 42, 66-77.

Durston, A.J., Zhu, K., 2015. A time space translation hypothesis for vertebrate axial patterning. Semin Cell Dev Biol 42, 86-93. Fagman, H., Grande, M., Gritli-Linde, A., Nilsson, M., 2004. Genetic deletion of sonic hedgehog causes hemiagenesis and ectopic development of the thyroid in mouse. Am J Pathol 164, 1865-1872. Fagman, H., Liao, J., Westerlund, J., Andersson, L., Morrow, B.E., Nilsson, M., 2007. The 22q11 deletion syndrome candidate gene Tbx1 determines thyroid size and positioning. Hum Mol Genet 16, 276-285. Fagman, H., Nilsson, M., 2011. Morphogenetics of early thyroid development. J Mol Endocrinol 46, R33-42. Gardiner, J.R., Jackson, A.L., Gordon, J., Lickert, H., Manley, N.R., Basson, M.A., 2012. Localised inhibition of FGF signalling in the third pharyngeal pouch is required for normal thymus and parathyroid organogenesis. Development 139, 34563466. Gaufo, G.O., Thomas, K.R., Capecchi, M.R., 2003. Hox3 genes coordinate mechanisms of genetic suppression and activation in the generation of branchial and somatic motoneurons. Development 130, 5191-5201. Gaunt, S.J., 1987. Homeobox gene Hox1.5 expression in mouse embryos: earliest detection by in situ hybridization is during gastrulation. Development 101, 51-60. Gaunt, S.J., 1988. Mouse homeobox gene transcripts occupy different but overlapping domains in embryonic germ layers and organs: a comparison of Hox-3.1 and Hox-1.5. Development 103, 135-144. Gligorov, D., Sitnik, J.L., Maeda, R.K., Wolfner, M.F., Karch, F., 2013. A novel function for the Hox gene Abd-B in the male accessory gland regulates the long-term female post-mating response in Drosophila. PLoS Genet 9, e1003395. Gong, S., Zheng, C., Doughty, M.L., Losos, K., Didkovsky, N., Schambra, U.B., Nowak, N.J., Joyner, A., Leblanc, G., Hatten, M.E., Heintz, N., 2003. A gene expression atlas of the central nervous system based on bacterial artificial chromosomes. Nature 425, 917-925. Gordon, J., Manley, N.R., 2011. Mechanisms of thymus organogenesis and morphogenesis. Development 138, 3865-3878. Gordon, J., Patel, S.R., Mishina, Y., Manley, N.R., 2010. Evidence for an early role for BMP4 signaling in thymus and parathyroid morphogenesis. Dev Biol 339, 141154. Gordon, J., Wilson, V.A., Blair, N.F., Sheridan, J., Farley, A., Wilson, L., Manley, N.R., Blackburn, C.C., 2004. Functional evidence for a single endodermal origin for the thymic epithelium. Nat Immunol 5, 546-553. Grevellec, A., Graham, A., Tucker, A.S., 2011. Shh signalling restricts the expression of Gcm2 and controls the position of the developing parathyroids. Developmental biology 353, 194-205. Hayashi, S., McMahon, A.P., 2002. Efficient recombination in diverse tissues by a tamoxifen-inducible form of Cre: a tool for temporally regulated gene activation/inactivation in the mouse. Dev Biol 244, 305-318. Hutlet, B., Theys, N., Coste, C., Ahn, M.T., Doshishti-Agolli, K., Lizen, B., Gofflot, F., 2016. Systematic expression analysis of Hox genes at adulthood reveals novel patterns in the central nervous system. Brain structure & function 221, 1223-1243.

Kameda, Y., Arai, Y., Nishimaki, T., Chisaka, O., 2004. The role of Hoxa3 gene in parathyroid gland organogenesis of the mouse. J Histochem Cytochem 52, 641-651. Kameda, Y., Nishimaki, T., Takeichi, M., Chisaka, O., 2002. Homeobox gene hoxa3 is essential for the formation of the carotid body in the mouse embryos. Dev Biol 247, 197-209. Kameda, Y., Watari-Goshima, N., Nishimaki, T., Chisaka, O., 2003. Disruption of the Hoxa3 homeobox gene results in anomalies of the carotid artery system and the arterial baroreceptors. Cell Tissue Res 311, 343-352. Katori, Y., Takeuchi, H., Rodriguez-Vazquez, J.F., Kitano, H., Murakami, G., Kawase, T., 2011. Fetal development of the human epiglottis revisited: appearance of GFAP-positive mesenchymal cells and fibrous connections with other laryngeal and lingual structures. Ann Anat 193, 149-155. Kuert, P.A., Bello, B.C., Reichert, H., 2012. The labial gene is required to terminate proliferation of identified neuroblasts in postembryonic development of the Drosophila brain. Biol Open 1, 1006-1015. Lebert-Ghali, C.E., Fournier, M., Kettyle, L., Thompson, A., Sauvageau, G., Bijl, J.J., 2016. Hoxa cluster genes determine the proliferative activity of adult mouse hematopoietic stem and progenitor cells. Blood 127, 87-90. Macatee, T.L., Hammond, B.P., Arenkiel, B.R., Francis, L., Frank, D.U., Moon, A.M., 2003. Ablation of specific expression domains reveals discrete functions of ectoderm- and endoderm-derived FGF8 during cardiovascular and pharyngeal development. Development 130, 6361-6374. Manley, N.R., Capecchi, M.R., 1995. The role of Hoxa-3 in mouse thymus and thyroid development. Development 121, 1989-2003. Manley, N.R., Capecchi, M.R., 1997. Hox group 3 paralogous genes act synergistically in the formation of somitic and neural crest-derived structures. Dev Biol 192, 274-288. Manley, N.R., Capecchi, M.R., 1998. Hox group 3 paralogs regulate the development and migration of the thymus, thyroid, and parathyroid glands. Dev Biol 195, 1-15. Montavon, T., Soshnikova, N., 2014. Hox gene regulation and timing in embryogenesis. Seminars in cell & developmental biology 34, 76-84. Moore-Scott, B.A., Manley, N.R., 2005. Differential expression of Sonic hedgehog along the anterior-posterior axis regulates patterning of pharyngeal pouch endoderm and pharyngeal endoderm-derived organs. Dev Biol 278, 323-335. Ozaki, T., Nagashima, K., Kusakabe, T., Kakudo, K., Kimura, S., 2011. Development of thyroid gland and ultimobranchial body cyst is independent of p63. Lab Invest 91, 138-146. Park, E.J., Sun, X., Nichol, P., Saijoh, Y., Martin, J.F., Moon, A.M., 2008. System for tamoxifen-inducible expression of cre-recombinase from the Foxa2 locus in mice. Developmental dynamics : an official publication of the American Association of Anatomists 237, 447-453. Peters, H., Neubuser, A., Kratochwil, K., Balling, R., 1998. Pax9-deficient mice lack pharyngeal pouch derivatives and teeth and exhibit craniofacial and limb abnormalities. Genes Dev 12, 2735-2747.

Rezsohazy, R., Saurin, A.J., Maurel-Zaffran, C., Graba, Y., 2015. Cellular and molecular insights into Hox protein action. Development 142, 1212-1227. Rodriguez-Vazquez, J.F., Kim, J.H., Verdugo-Lopez, S., Murakami, G., Cho, K.H., Asakawa, S., Abe, S., 2011. Human fetal hyoid body origin revisited. J Anat 219, 143-149. Romert, P., Gauguin, J., 1973. The early development of the median thyroid gland of the mouse. A light-, electron-microscopic and histochemical study. Z Anat Entwicklungsgesch 139, 319-336. Ruhin, B., Creuzet, S., Vincent, C., Benouaiche, L., Le Douarin, N.M., Couly, G., 2003. Patterning of the hyoid cartilage depends upon signals arising from the ventral foregut endoderm. Developmental dynamics : an official publication of the American Association of Anatomists 228, 239-246. Sandoval, P.C., Slentz, D.H., Pisitkun, T., Saeed, F., Hoffert, J.D., Knepper, M.A., 2013. Proteome-wide measurement of protein half-lives and translation rates in vasopressin-sensitive collecting duct cells. J Am Soc Nephrol 24, 1793-1805. Sharova, L.V., Sharov, A.A., Nedorezov, T., Piao, Y., Shaik, N., Ko, M.S., 2009. Database for mRNA half-life of 19 977 genes obtained by DNA microarray analysis of pluripotent and differentiating mouse embryonic stem cells. DNA Res 16, 45-58. Soriano, P., 1999. Generalized lacZ expression with the ROSA26 Cre reporter strain. Nat Genet 21, 70-71. Washington Smoak, I., Byrd, N.A., Abu-Issa, R., Goddeeris, M.M., Anderson, R., Morris, J., Yamamura, K., Klingensmith, J., Meyers, E.N., 2005. Sonic hedgehog is required for cardiac outflow tract and neural crest cell development. Dev Biol 283, 357372. Watari, N., Kameda, Y., Takeichi, M., Chisaka, O., 2001. Hoxa3 regulates integration of glossopharyngeal nerve precursor cells. Dev Biol 240, 15-31. Yang, E., van Nimwegen, E., Zavolan, M., Rajewsky, N., Schroeder, M., Magnasco, M., Darnell, J.E., Jr., 2003. Decay rates of human mRNAs: correlation with functional characteristics and sequence attributes. Genome Res 13, 1863-1872.

Figure legends Fig. 1. Hoxa3 expression is selectively reduced in the NCCs and endoderm and reduction of the Hoxa3 floxed allele following Cre-mediated recombination. (A-C) Sagittal sections of E10.5, E11.5, and E12.5 wild-type embryos that had been subjected to ISH with a Hoxa3 riboprobe. Staining for all sections was performed together and for an equal length of time. The 3rd and 4th pharyngeal pouches are outlined in the E10.5 and E11.5 sections, and the parathyroid and thymus are outlined in the E12.5 section. Scale bars: 40μm. (D) PCR results of pooled genomic DNA extracted at the time indicated post TI at E10.5 from Hoxa3fx/+;CAGG-CreERTM (control, Cre+) embryos. Somite (s) numbers for all embryos were counted and only embryos with the same somite number were included for each time point (T=0, 37s; T=6, 39s; T=12, 41s; T=18, 44; T=24, 47s). There were at least three embryos for each group in each time point. The 470 bp band is the floxed allele. Primers were set inside the floxed sites so the loss of the band shows the loss of the genomic region within the floxed sites. m=mutant. c=control. Fig. 2. The thymus and parathyroids show a range of separation and migration defects after TI at E9.5 to E11.0. (A-D,F-H)) H&E staining of transverse sections of Hoxa3fx/;CAGG-CreERTM (control, Cre+) embryos. The right lobe of (B) is representative of a normal phenotype. (D) is a representative of a normal phenotype. (E) An E15.5 transverse section that had been immunostained with DAPI, an antibody against Foxn1, and imaged for Gcm2-EGFP expression (Hoxa3fx/-;CAGG-CreERTM;Gcm2-EGFP). (G) Inset is closer view of thymus and parathyroid attachment. Injection time point and time of collection indicated for all panels. In all sections, dorsal is towards the top of the image. thy, thymus; ub, ultimobranchial body; thr isth, thyroid isthmus; pt, parathyroid. Scale bars: 200μm. Fig. 3. HOXA3 acts cell autonomously in the tracheal epithelium but non-cell autonomously in thyroid development and thyroid-ultimobranchial body fusion. (A-D) Sagittal sections of the pharyngeal region at E18.5 (A,C,D) and E13.5 (B) of control embryos stained for Hoxa3lacZ expression. (E-N) H&E staining of E18.5 transverse sections of Hoxa3fx/-;CAGG-CreERTM (E-G, J-L), Hoxa3fx/-;Wnt1-Cre (H,M), Hoxa3fx/;Foxa2-CreERTM (I,N) embryos. (M) Inset shows clearer view of ultimobranchial body and thyroid lack of fusion. Scale bar is represented for each panel following it in a row. In all sections, dorsal is towards the top of the image. tr epi, tracheal epithelium;UB, ultimobranchial body; thr div, thyroid diverticulum; thy, thymus; tr, trachea; thr isth, thyroid isthmus; pth, parathyroid; thr, thyroid; Endo, endoderm. Scale bars 50 or 80μm. Fig. 4. The soft palate and epiglottis are truncated early in embryos with a NCC-specific deletion of Hoxa3. (A-D) Sagittal sections of the soft palate in E18.5 wild-type embryos stained for R26R expression. All sections come from the same embryo and given that the embryo was cut at an angle the two sides show different anatomical structures. Both sides are serial sections. (E) Diagram of human cleft palate in the coronal plain. (F-P) H&E sagittal sections of the soft palate in E18.5 Hoxa3fx/-;CAGG-CreERTM embryos that had TI at E7.5 (F-J), Hoxa3fx/-;Foxa2-CreERTM embryos (K,L), Hoxa3fx/-;CAGG-CreERTM embryos that had TI at E10.5 (M,N), and Hoxa3fx/-;Wnt1-Cre embryos (O,P). Circles identify lateral soft palate fusion points. F-H are serial sections. R., right; L., left; sp, soft palate; ep, epiglottis; ton, tongue; *, missing or truncated structure. Scale bar (200μm) in A is representative for all panels.

Fig. 5. Hoxa3-expressing are NCCs necessary for IXth CG nerve formation. Immunostaining for ZH3, a neurofilament marker, in E11.5 wild-type (A), Hoxa3null (B), and Hoxa3fx/-;Wnt1-Cre (C-E) embryos. D and E represent the two mutant phenotypes seen. Anterior is towards the left of the panel. IX, ninth cranial ganglion; X, tenth cranial ganglion. Fig. 6. NCC-specific Hoxa3 mutant skeletal phenotypes are progressively worse as Hoxa3 is deleted earlier in development. (A) An E10.5 wild-type embryo stained for Hoxa3lacZ expression showing the anterior border and NCCs labeled with Hoxa3 in the pharyngeal arches and the rhombomeres. (B,C) Sagittal sections of E18.5 wild-type embryos stained for Hoxa3lacZ expression highlighting the hyoid bones and pharyngeal cartilage. (D-N) Pharyngeal skeletons from E18.5 control (D), Hoxa3null (E,F), TI time point E7.5-E12.5 (G-L), Hoxa3fx/-;Wnt1-Cre (M), and Hoxa3fx/-;Foxa2-CreERTM(N) embryos. The two null panels (E,F) represent the two mutant phenotypes seen in the Hoxa3null. pa, pharyngeal arch; NC, neural crest; LHB, lesser hyoid bone; GHB, greater hyoid bone; TC, thyroid cartilage; Thr, thyroid; CC, cricoid cartilage; Es, esophagus; Endo, endoderm; *, lesser hyoid bone; arrowheads, points of fusion; arrows, extra projections. Fig. 7. Scheme of crucial time intervals of HOXA3 requirement for pharyngeal structures. The structures are listed in order of anterior-posterior tissue of origin. The terms in the parentheses describe the tissue of origin, including both the endoderm and NCCs in those cases, and then the pharyngeal arch of origin. Both the hyoid and the throat cartilage originate from NCCs but the hyoid is from the 3rd pharyngeal arch while the throat cartilage is from the 4th pharyngeal arch. The hyoid lesser horn is absent in the Hoxa3null but our data cannot conclude if early Hoxa3 is necessary for survival or initiation of the structure. Blue lines represent when HOXA3 is required for the noted functional attribute. pa, pharyngeal arch; pp, pharyngeal pouch; pth, parathyroid; thy, thymus.

Table 1. A comprehensive table of phenotypic outcomes after deletion of Hoxa3 from the endoderm, NCCs, endoderm and NCCs, or globally at different time points during development. Phenoty pe

Tissu e Origi n

HoxA3 Null

Soft Palate

2nd Arch NCC

Truncat ed

Hyoid Lesser Horn

Thyroid Lobes

Thyroid Isthmus

2nd Arch NCC

2nd Arch Ventr al Endo derm 2nd Arch Ventr al Endo derm

Tracheal Epitheliu m

3rd Arch Ventr al Endo derm

Parathyr oids

3rd Pouc h Endo derm

Thymus

cgIX

Absent

NCC deletion

Endode rm deletio n

Truncat ed

Normal

n=2

n=3

n=1

Reduce d/ Fused

Normal

Reduce d/ Fused

Reduce d/ Fused

n=5

n=8

n=4

Normal

Normal

Normal

n=10

n=8

Normal

Double deletio n

NA

TI E7.5

Truncat ed

TI E8.5

TI E9.5

NA

NA

TI E10.5

TI E11.5

Normal

Normal

TI E12 .5

NA n=1

n=1

Reduce d/ Fused

Normal

Normal

Nor mal

n=1

n=1

n=2

n=2

n=1

Normal

Normal

Absent/ Normal

Normal

Normal

Nor mal

n=2

n=2

n=2

n=4

n=4

n=4

n=2

Normal

Normal

Reduce d

Absent

Absent/ Normal

Normal

Normal

Nor mal

n=10

n=8

n=2

n=2

n=2

n=4

n=4

n=4

n=2

Normal

Disorga nized

Disorga nized

Disorga nized

Disorga nized

Disorga nized

Disorga nized

Disorga nized

Nor mal

n=10

n=8

n=2

n=2

n=4

n=4

n=4

n=4

n=2

Absent

Absent

Absent/ Attache d

Attache d/ Ectopic

Normal

Nor mal

n=2

n=2

n=4

n=11

n=10

n=2

Absent

Absent

Absent/ Attache d

Attache d/ Ectopic

Normal

Nor mal

n=2

n=2

n=4

n=11

n=10

n=2

NA

NA

NA

NA

NA

NA

NA

Absent/ Normal

Absent/ Reduce d

Disorga nized

Absent

3rd Pouc h Endo derm

Absent

3rd Arch NCC

Disconn ected/ Fused with cgX

Ectopic*

Ectopic*

Disconn ected/ Fused with cgX

Ectopic *

Ectopic *

Normal

Absent *

Absent *

NA

n=32 3rd and 4th Arch NCC

Reduce d/ Fusions

Epiglotti s

3rd and 4th Arch NCC

Absent/ Truncat ed

Ultimobr anchial Body

4th Pouc h Endo derm

Hyoid/ Throat Cartilage

Persiste nt

n=6 Reduce d/ Fusion s

Reduce d/ Fusion s

Fusion s

Normal/ Projecti ons

Normal/ Projecti ons

Nor mal

n=8

n=4

n=1

n=1

n=2

n=2

n=1

Absent

Normal

Absent

Normal

Normal

n=2

n=3

n=1

n=1

Persiste nt

Normal

Persist ent

Persist ent

Persist ent

Persist ent

Persist ent

Normal

Nor mal

n=10

n=8

n=2

n=2

n=2

n=3

n=4

n=4

n=2

Reduce d/ Fusions

Normal

n=5

NA

NA

*Data obtained from Chojnowski et al. 2014.

NA

NA

n=1

NA