SEARCH

SEARCH BY CITATION

Keywords:

  • inner ear;
  • epibranchial;
  • placode;
  • Pax2;
  • Pax8

Abstract

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. RESULTS
  5. DISCUSSION
  6. EXPERIMENTAL PROCEDURES
  7. Acknowledgements
  8. REFERENCES

Background: The inner ear and epibranchial ganglia of vertebrates arise from a shared progenitor domain that is induced by FGF signalling, the posterior placodal area (PPA), before being segregated by Wnt signalling. One of the first genes activated in the PPA is the transcription factor Pax2. Loss-of- and gain-of function studies have defined a role for Pax2 in placodal morphogenesis and later inner ear development, but have not addressed the role Pax2 plays during the formation and maintenance of the PPA. Results: To understand the role of Pax2 during the development of the PPA, we used over-expression and repression of Pax2. Both gave rise to a smaller otocyst and repressed the formation of epibranchial placodes. In addition, cell cycle analysis revealed that Pax2 suppression reduced proliferation of the PPA. Conclusions: Our results suggest that Pax2 functions in the maintenance but not the induction of the PPA. One role of Pax2 is to maintain proper cell cycle proliferation in the PPA. Developmental Dynamics 241:1716–1728, 2012. © 2012 Wiley Periodicals, Inc.


INTRODUCTION

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. RESULTS
  5. DISCUSSION
  6. EXPERIMENTAL PROCEDURES
  7. Acknowledgements
  8. REFERENCES

The inner ear and epibranchial ganglia form from a common precursor field, known as the posterior placodal area (PPA; Ladher et al., 2010). This progenitor domain is specified by the action of members of the fibroblast growth factor (FGF) family of secreted signalling molecules (Phillips et al., 2001; Leger and Brand, 2002; Maroon et al., 2002; Wright and Mansour, 2003; Ladher et al., 2005), and subsequently the inner ear placode and epibranchial placodes emerge from this domain (Ohyama et al., 2006; Freter et al., 2008). One of the first responses to FGF signalling in the chick PPA is the induction of Pax2 expression (Martin and Groves, 2006).

Pax2 is part of an evolutionarily related family of transcription factors, characterized by a highly conserved amino terminal domain, the paired box, which binds DNA (Dressler et al., 1990; Treisman et al., 1991; Holland et al., 1995). Pax genes are found in a wide variety of organisms including the placozoans, cnidarians, insects, mollusks, tunicates, and vertebrates (Vorobyov and Horst, 2006). In vertebrates there are nine Pax genes, which can be split into different classes based on sequence homology and presumed ancestral relationships (Mansouri et al., 1996; Stuart and Gruss, 1996; Vorobyov and Horst, 2006). Pax2 groups with Pax5 and Pax8, and in many regions of the embryo either two or all three of the Pax2/5/8 sub-family, are co-regulated (Krelova et al., 2002). In most vertebrates, the mid-hindbrain boundary expresses all three genes, although the kidney and inner ear primordia only express Pax2 and Pax8 (Plachov et al., 1990; Stoykova and Gruss, 1994; Lun and Brand, 1998; Pfeffer et al., 1998; Funahashi et al., 1999; Kozmik et al., 1999; Bouchard et al., 2002; Krelova et al., 2002; Hans et al., 2004), with Pax5 not present in these tissues (Funahashi et al., 1999; Kwak et al., 2006). The overlap between Pax2 and Pax8 in the PPA has confounded functional characterization of this Pax subfamily due to their presumed redundancy.

The role of Pax genes during the development of the inner ear primordium has been investigated. In mice, zebrafish and frogs Pax8 is expressed earlier than Pax2 during inner ear formation (Plachov et al., 1990; Pfeffer et al., 1998; Heller and Brandli, 1999; Bouchard et al., 2004). Here, the expression of Pax genes was considered to be restricted to the inner ear. However, lineage labeling of the Pax2 expression domains in chick and either Pax2- or Pax8-expressing regions in mouse suggest that it is more appropriate to consider these genes as markers of both inner ear and epibranchial placode progenitors (Bouchard et al., 2004; Ohyama and Groves, 2004; Streit, 2004). Functional analysis has not provided conclusive evidence of the mechanistic role of Pax2/8 genes during inner ear and epibranchial development. Despite its earlier expression, mice mutant for only Pax8 do not show an overt inner ear phenotype (Bouchard et al., 2002). Pax2 mutants do show a later otic phenotype with malformations of the inner ear apparent in mice at E12, well after the inner ear has invaginated and closed (Torres et al., 1996; Burton et al., 2004). The absence of an early phenotype is thought to result from redundancy with Pax8. However, in mice mutant for both Pax2 and Pax8, the inner ear does form although its development does not extend past the otocyst stage (Bouchard et al., 2010). In chick, reduction of Pax2 does not affect the early induction of the PPA. However, morphogenesis and the integrity of the later otocyst, is disrupted (Christophorou et al., 2010). In zebrafish, the roles of Pax2 and pax8 are slightly different. Epistatic analysis indicates that pax8 is upstream of the duplicated Pax2 genes, pax2a and pax2b: pax8 repression down-regulates Pax2 genes and inner ear induction is reduced, but not absent (Hans et al., 2004; Mackereth et al., 2005). Mutations of both zebrafish Pax2 genes cause later defects in the inner ear, with a slight reduction of hair cell number. When both Pax2 genes together with pax8 are repressed, initial induction is reduced, similar to that seen in single pax8 loss-of-function animals. Furthermore, the reduced inner ear is not maintained and eventually degenerates (Mackereth et al., 2005; Padanad and Riley, 2011). Thus, mouse and zebrafish experiments suggest both Pax2 and Pax8 function redundantly during early inner ear formation, but not in the initial induction of its progenitor domain. In zebrafish, recent experiments have suggested a subsequent role for different levels of pax2a in segregating the otic and epibranchial progenitors within the PPA. Wnt signalling acts on the PPA increasing the expression of pax2a in the otic portion of the PPA (McCarroll et al., 2012). In addition, pax2 and pax8 have a non-cell-autonomous effect on epibranchial formation via the induction of fgf24 in the inner ear (Padanad and Riley, 2011).

In this study, we investigate the role of the Pax2/8 family in the formation of the chick posterior placodal area (PPA) and its derivatives. We find that in diapsids (lizards and birds) the Pax8 gene has been lost, leaving Pax2 as the sole Pax gene functioning during early inner ear development. This allows us a more complete insight into the role of Pax genes during this process. In agreement with previous studies, we find that neither loss-of-function using shRNA and repressor approaches nor gain-of-function affects the induction of the PPA. Reducing Pax2 expression only mildly effects inner ear formation. However, we find that the differentiation of the epibranchial ganglia is inhibited. In addition, reduction of Pax2 reduces proliferation of PPA progenitors. These results suggest Pax2 functions not in induction but in maintenance of the PPA.

RESULTS

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. RESULTS
  5. DISCUSSION
  6. EXPERIMENTAL PROCEDURES
  7. Acknowledgements
  8. REFERENCES

Pax8 Has Been Lost From the Chicken Genome

In most vertebrates, Pax8 is considered to be one of the earliest markers of otic fate (Bouchard et al., 2002; Hans et al., 2004; Mackereth et al., 2005). We thus attempted to clone Gallus Pax8 from the 4ss/HH8 presumptive inner ear region cDNA using RT-PCR. Despite varied attempts using a number of degenerate primers, almost all of the clones that were sequenced were Pax2 (data not shown). We next attempted to identify the Pax8 gene in the chicken genome; BLAST searches toward whole genome assemblies (version 2.1 available at the Ensembl Genome Browser; versions 3.0, 3.1, and 3.2 available at The Genome Institute of Washington University; International Chicken Sequencing Consortium, 2004) using both human, zebrafish, and mouse Pax8 peptide sequences as queries only identified the chick Pax2 and Pax5 loci previously annotated. We also analysed genome assemblies of other diaspid species (zebra finch, turkey, and anole lizard) available at Ensembl (Warren et al., 2010; Alfoldi et al., 2011) and garter snake genome sequences (Schwartz et al., 2010) as well as publicly available transcriptome sequences (Tzika et al., 2011). In all these searches, no Pax8 orthologous sequence could be placed into the Pax8 clade in the molecular phylogenetic tree (Fig. 1A). It is, therefore, likely that the Pax8 gene was lost before the radiation of extant diapsids.

thumbnail image

Figure 1. The Pax8 ortholog is absent from the chicken genome. A: Molecular phylogenetic tree of the vertebrate Pax2/5/8 genes. The tree was inferred with the maximum-likelihood (ML) method using the program PhyML (Guindon and Gascuel, 2003), assuming the WAG+I+G model. Support values at nodes are bootstrap probabilities based on 100 resamplings with the ML and neighbor-joining methods in order. Note that there is no sauropsidan gene included in the Pax8 clade, in spite of thorough in silico searches as well as RT-PCR with degenerate primers (see text). B: Syntenic comparison of the genes around Pax8 and Pax2. While not all genes are shown, immediate neighbours as well as upstream and downstream representative genes around the human Pax8 and human Pax2 loci are depicted. Blue blocks represent conserved synteny around Pax8; other colours represent blocks of chromosomes without syntenic homology. Maroon represents conserved synteny around Pax2.

Download figure to PowerPoint

We next asked if we could identify a region in the chicken genome that harboured the Pax8 ortholog, by analysing conserved synteny. We compared the order of genes flanking the human Pax8 in 11 completed amniote genomes represented in Ensembl (Fig. 1B). The chromosome structure upstream of Pax8 is well conserved amongst mammals; however, it is apparent that immediately upstream (telomeric) to Pax8, translocations occur with high frequency, such that even phylogenetically closely related genomes show a differing genomic arrangement in this region. In birds and reptiles, synteny is only preserved in the region downstream (centromeric) to the presumed Pax8 location. However, no Pax8 orthologs could be detected. In contrast, there is good syntenic conservation around Pax2 in all amniote species inspected. This suggested that Pax8 had been lost from the genome of birds and reptiles. Finally, as expression data indicated that the other member of this Pax gene subfamily, Pax5, is not expressed in the early inner ear (Funahashi et al., 1999; Kwak et al., 2006), these data suggested that Pax2 is the sole Pax gene functioning during early PPA development in the chick.

Pax2 Does Not Regulate PPA Identity

The PPA is a shared progenitor domain for both the inner ear and epibranchial placodes induced by fibroblast growth factor signalling (Sun et al., 2007; Freter et al., 2008; Ladher et al., 2010) and characterized by the expression of Pax2 (Martin and Groves, 2006). Previous reports in zebrafish and mouse have shown that Pax2/8 knockdown in the PPA does not completely obliterate inner ear formation; these data suggest that early PPA development may not require Pax gene activity. We investigated this further by manipulating Pax2 function in chick embryos.

Firstly, Pax2 levels were repressed using short-hairpin expressing constructs. These were electroporated into the ectoderm of HH4/5 chick embryos (Fig. 2A). Such electroporated embryos were cultured until 7–13ss (HH9–11). While control (shScr) electroporated embryos did not affect Pax2 expression (n=4/4), shPax2 electroporation caused endogenous Pax2 gene expression to be repressed to levels undetectable by in situ hybridization (Fig. 2B, C; n=10/11). However, Dlx3, another marker for the PPA (Brown et al., 2005) was unchanged in shPax2 electroporated embryos (Fig. 2D, E; n=4/6). A second approach to reduce activity was the construction of a repressor form of Pax2; fusing the open reading frame of Pax2 to the powerful transcriptional repressor module from the engrailed transcription factor (Pax2-EnR) (Jaynes and O'Farrell, 1988). As Pax2 normally functions as an activator, such a fusion construct acts to completely repress endogenous Pax2 activity (Conlon et al., 1996; Brent and Tabin, 2004). Despite this suppression, endogenous Pax2 transcription was not repressed by the introduction of Pax2-EnR (Fig. 2F, G; n=4/5 unaffected). The expression of Dlx3 was similarly unaffected after Pax2-EnR electroporation (n=5/5 unaffected).

thumbnail image

Figure 2. Pax2 is not required for the induction of the posterior placodal area (PPA). A: The presumptive PPA was targeted by unilateral electroporation at HH5 and assayed at 7–13ss. In all experiments, the unelectroporated side is on the left, the electroporated side is presented on the right. B: Pax2 expression is repressed by shPax2 electroporation. C: Section showing absence of Pax2 transcripts and extent of electroporation after anti-RFP immunostaining. D: Dlx3, a PPA marker at stage HH10, is not reduced after introduction of shPax2. E: Sections show expression in thickened ectoderm as well as RFP localization. F: Pax2 PPA expression is maintained after electroporation of repressive Pax2-EnR. G: Sections show normal Pax2 expression and widespread unilateral anti-RFP immunoreactivity. H: Over-expression of mouse Pax2 does not alter endogenous Pax2 expression. I: Sections show that Pax2 is maintained in the whole PPA together with widespread immunoreactivity for the GFP tracer. J: Dlx3 is expressed normally following Pax2 over-expression. K: Section shows normal Dlx3 expression despite widespread GFP immunoreactivity.

Download figure to PowerPoint

We next asked if Pax2 activity was sufficient to induce a broader or ectopic PPA. The mouse Pax2 coding region was fused to green fluorescent protein (eGFP) cDNA under the control of the β-actin promoter and this construct was electroporated into the ectoderm of HH4/5 chick embryos and cultured until 10–13ss. Neither Pax2 expression (Fig. 2H, I; n=3/3) nor Dlx3 (Fig. 2J, K; n=5/6) was expanded as a result of mouse Pax2 over-expression. Taken together, these data indicate that Pax2 alone does not determine PPA identity.

Pax2 Influences Early Inner Ear Development

Although Pax2 repression did not affect PPA specification, we hypothesized that differentiation of the inner ear may be affected. One of the first genes expressed in the otic placode proper is Soho1 (Deitcher et al., 1994). Its expression is first detected at around 9/10ss and is concomitant with the ability of an isolated otic placode to differentiate hair cells (Freter et al., 2008). Embryos were electroporated at HH5 with shPax2 and assayed for Soho1 expression at two time-points; at 10ss/HH10 and at 22ss/HH14. At 9–10ss, Soho1 expression was abolished by Pax2 knockdown (Fig. 3A, B), although control shScr electroporations were unaffected (data not shown). By 22ss/HH14, Soho1 expression could be detected; however, the otic placode was smaller by 40% (n=8) when compared to stage-matched control electroporations (Fig. 3C, D). These results suggest that inner ear development is temporally delayed and attenuated in the absence of Pax2.

thumbnail image

Figure 3. Pax2 knock-down causes moderate repression of the inner ear. Pax2 activity was repressed using shPax2 and assayed at 10–22ss; in all cases the experimental side is on the right in dorsal views of the embryos. A: shPax2 electroporation represses Soho1 expression at HH10. B: Fluorescent image of the RFP tracer showing the extent of electroporation. C: By HH13/14, Soho1 expression is observed; however, it describes a smaller domain. D: Section shows that the smaller otic placode (op) retains its pattern of Soho1 expression despite good RFP tracer immunoreactivity. E: The expression domain of Foxg1 is smaller after shPax2 electroporation. F: Section shows Foxg1 expression pattern is retained, marking the neurogenic portion of the otocysts and in delaminating neuroblasts (arrowheads), even on the electropoarted side. G: Gbx2 is unaffected in the smaller otic vesicle after Pax2 down regulation. H: Section shows Gbx2 expression after electroporations. I: Nkx5.1 is expressed in the smaller otic vesicle after Pax2 reduction. J: Section shows Nkx5.1 expression. K: The expression of Dlx5 within the otic vesicle is reduced after Pax2 knockdown. L: Section shows Dlx5 expression is maintained in the dorsal otocyst. M: Tbx1 expression is maintained in shPax2 electroporated embryos. N: Section shows Tbx1 shifted medially in the smaller otic cup.

Download figure to PowerPoint

Pax2 function has been shown to play a role in the patterning of the otocyst; mouse mutants for Pax2 show defects in formation of ventral regions of the otocyst (Burton et al., 2004). Thus, we asked whether normal otocyst patterning was affected following Pax2 repression in the chick system. The expression patterns of Foxg1 (Fig. 3E, F; n=6/8) and Gbx2 (Fig. 3G, H; n n=5/5) were not substantially affected in the smaller otocyst found in electroporated embryos. Similarly, Nkx5.1 (Fig. 3I, J; n=8/8) and Dlx5 (Fig. 3K, L; n=9/10), which normally mark a dorsal domain in the otocyst (Herbrand et al., 1998; Brown et al., 2005), were also unchanged. Tbx1, which is normally expressed laterally (Raft et al., 2004), showed a slight ventro-medial expansion of its expression domain (Fig. 3M, N; n=6/8). We noted expression of Foxg1 in neuroblasts delaminating from the inner ear is retained in shPax2 electroporated embryos, further indicating relatively normal cyto-differentiation (arrowheads, Fig. 3F).

To determine the effect of Pax2 over-expression on the differentiation of the otic lineage of the PPA, CAG-mPax2 was unilaterally electroporated into the ectoderm of HH4/5 chick embryos, and cultured until 16ss/HH13–22ss/HH15. Pax2 over-expression failed to enlarge the otocyst, or cause the formation of ectopic otocysts. Control, pCAG-GFP electroporations did not affect Soho1 expression (data not shown). However, CAG-mPax2 electroporation caused a reduction in the expression domains of otic markers: Soho1 (Fig. 4A–C; n=8/10), Nkx5.1 (Fig. 4D–F; n=7/7), Foxg1 (Fig. 4G–I; n=8/9), Dlx5 (Fig. 4J–L; n=6/7), Gbx2 (Fig. 4M–O; n=3/5), and Tbx1 (Fig. 4P–R; n=5/5). Occasionally, the otic placode failed to invaginate and remained as a flat placode. In such cases, apical enrichment of F-actin (Sai and Ladher, 2008) was also diminished (Fig. 4S and T; n=3/4). Expression of the cleaved, and thus active form of caspase 3 revealed that pCAG-mPax2 electroporated cells were not more prone to cell death than unelectroporated or pCAG-GFP electroporated controls.

thumbnail image

Figure 4. Sustained Pax2 over-expression represses inner ear commitment. Exogenous Pax2 was introduced using a β-actin driven construct and assayed at 16–22ss; in all cases, the experimental side is on the right in these dorsal views. A: Expression of Soho1 is reduced after CAG-mPax2 electroporation. B: Section shows a reduced uninvaginated otic placode. C: GFP positive electroporated cells contribute evenly to the ectoderm. D: Nkx5.1 expression showing reduction of the inner ear on the electroporated side. E: Section shows a smaller but thicker placode. F: GFP immunoreactivity shows extent of electroporation. G: Foxg1 expression is reduced after CAG-mPax2 electroporation. H: Section shows an uninvaginated placode, although some Foxg1-positive neuroblasts can be detected (arrowheads). I: GFP tracer is detected in ectoderm, but not delaminating neuroblasts. J: Dlx5 expression is reduced after CAG-mPax2 electroporation. K: Section shows reduced Dlx5 within an uninvaginated placode. L: GFP immunoreactivity can be detected throughout the ectoderm. M: Gbx2 expression highlights a delay in otic vesicle closure. N: Section shows Gbx2 expression in a smaller otic placode. O: CAG-mPax2 electroporated cells contribute to the forming placode. P: Expression of the ventral ear marker Tbx1 is reduced. Q: Section shows Tbx1 expression in the electroporated otic placode. R: GFP immunoreactivity shows the extent of electroporation. S: At 16ss, F-actin is normally apically enriched. The right electroporated otic placode is smaller and F-Actin is present at both apical and basal sides of the placode. T: Electroporation targets one side of the embryo.

Download figure to PowerPoint

Pax2 Is Necessary for Differentiation of Epibranchial Neurons

As the PPA also generates epibranchial placodes (Ladher et al., 2010), we investigated if Pax2 repression affected their specification and differentiation. Embryos electroporated at HH4/5 with the shPax2 construct were cultured to either 10–13ss (HH10–11) or 22–31ss (HH14–17). Epibranchial formation was assessed using the following markers: Sox3 is expressed in the neurogenic regions of the PPA, which gives rise to otic and epibranchial neurons (Abu-Elmagd et al., 2001; Abello et al., 2010). Foxi2 was used as a marker of the non-otic lineage within the PPA (Freter et al., 2008), Phox2a is expressed in committed epibranchial neuroblasts both prior to and during their delamination, Phox2b is expressed in delaminated migrating neuroblasts (Begbie et al., 2002).

Pax2 repression did not affect the expression of Sox3 (Fig. 5A, B; n=5/5 unaffected), nor did it affect the peri-otic expression of Foxi2 (Fig. 5C–E; n=8/11 unaffected). However, suppression of Pax2 dramatically reduced the expression of Phox2a (Fig. 5F, G; n=6/8), Phox2b (Fig. 5H–I; n=6/8), and neurofilament1 (Fig. 5J, K; n=3/4). In most cases, differentiation of all epibranchial placodes was repressed, although in a few samples, traces of the petrosal and nodose epibranchial placodes could be detected. We noted that the development of the bilobular trigeminal ganglion was not affected (see Fig. 5J, K).

thumbnail image

Figure 5. Repression of Pax2 allows initiation of the epibranchial field but inhibits epibranchial differentiation. A: Unilateral electroporation of shPax2 does not affect Sox3 expression. B: RFP expression showing extent of electroporation after 18 hr, at 13ss/HH11. C: Unilateral electroporation of shPax2 does not affect Foxi2 expression. D: Section analysis shows Foxi2 localized to the peri-otic and lateral ectoderm. E: Tracer analysis reveals that Foxi2 is expressed after electroporation. F: Lateral view of the control side of a HH17 electroporated embryo showing Phox2a expression in epibranchial placodes and delaminating neuroblasts. G: Lateral view of the shPax2 electroporated (right) side of a HH17 embryo shows reduced Phox2a expression in sparse individual neurons in the geniculate and petrosal area. H: Lateral view of the control side of a HH17 electroporated embryo showing normal Phox2b expression in the epibranchial placodes and delaminating neuroblasts. I: Lateral view of the shPax2 electroporated (right) side of a HH17 embryo showing reduced/absent Phox2b expression in epibranchial-derived neuroblasts. J: Lateral view of the control side of an electroporated embryo showing Neurofilament expression in epibranchial neurons and the ophthalmic lobe of the trigeminal complex. K: Lateral view of the shPax2 electroporated (right) side showing reduced Neurofilament expression in epibranchial-derived neurons; the ophthalmic portion of the trigeminal complex is unaffected. opt, ophthalmic; g, geniculate; p, petrosal; n, nodose.

Download figure to PowerPoint

Sustained expression of Pax2 led to an inhibition of inner ear differentiation. To determine if Pax2 exerted a similar inhibitory action to epibranchial differentiation, CAG-mPax2 was introduced into the ectoderm of HH4/5 chick embryos. Such electroporated embryos were incubated until 10–13ss (HH10–11) or 22–31ss (HH14–17), at which point the extent of epibranchial differentiation was assessed. At early stages, over-expression of Pax2 did not affect the expression of Sox3 (Fig. 6A, B; n=4/4 unaffected). Foxi2 mRNA expression was repressed in the most lateral regions of its normal domain (Fig. 6C–E; n=10/13), although the peri-otic expression domain was maintained. Similar to Pax2 repression, sustained Pax2 expression repressed the differentiation of the epibranchial placodes as determined by Phox2a (Fig. 6F, G; n=4/5).

thumbnail image

Figure 6. Sustained Pax2 expression inhibits epibranchial neurogenesis. A: Unilateral electroporation of CAG-mPax2 does not affect Sox3 expression at 12ss. B: GFP expression showing the extent of electroporation. C: Unilateral electroporation of CAG-mPax2 does not affect peri-otic Foxi2 expression at 14ss, although more lateral regions are affected. D: Section analysis showing the presence of Foxi2 in peri-otic ectoderm after electroporation. E: GFP immunoreactivity can be detected throughout the ectoderm. F: Lateral view of the control (left) side of a HH17 CAG-mPax2 electroporated embryo showing Phox2a expressed in epibranchial placodes and delaminating neuroblasts. G: Lateral view of the CAG-mPax2 (right) electroporated side of a HH17 embryo showing reduced Phox2a expression localized to sparse neurons in the petrosal area. H: The control (left) side of a HH18 embryo stained with an antibody against the medium chain of neurofilament (NF-M) after epibranchial electroporation at HH10. I: Electroporated (right) side shows misrouting of axons, particularly those of the petrosal and nodose placodes. J: Section shows that in some cases, the Pax2 over-expressing epithelium fails to be internalized but can still produce some neurons. K: Boxed area in J at higher magnification shows neurons projecting from uninvaginated or delaminated superficial ectoderm. opt, opthlamic lobe of the trigeminal; g, geniculate; p, petrosal; n, nodose.

Download figure to PowerPoint

To determine the effect of Pax2 over-expression on the delamination and projections of the epibranchial neurons, we performed in ovo electroporations at later stages (10ss). This did result in a milder epibranchial phenotype. Whole mount neurofilament immunostaining revealed that although some epibranchial neurons were present, axonal patterning was disrupted (Fig. 6H, I; n=16/21). Regions of ectoderm in which CAG-Pax2 was expressed were often not internalised, resulting in ectopic patches of thickened placodal ectoderm. These superficial patches were labeled by neurofilament antibodies and seemed to project towards the hindbrain (Fig. 6J, K), indicating that the defects in axonal projection were most likely caused by the errant superficial location of the affected epibranchial placodes and neuronal precursors.

Pax2 Influences Cell Cycle

We next sought to clarify the mechanisms underlying the reduction in otocyst size and the number of epibranchial neuroblasts after inhibition of Pax2 activity. Cell death was not a major contributory factor, as immunostaining for cleaved caspase 3 did not show any differences between control shScrambled and shPax2 electroporations (data not shown). A second possibility is that Pax2 may regulate the length of the cell cycle, thus blocking Pax2 function could reduce proliferation of PPA progenitors, which would result in the generation of fewer inner ear and epibranchial cells. To test this, embryos were unilaterally electroporated with shPax2 and allowed to develop to 7–10ss/HH9–10. At this point, immunoreactivity to the phosphorylated form of histone H3 (pHH3), a specific marker for cells in M-phase, was used to assess the number of cells undergoing mitosis (Hendzel et al., 1997; Hans and Dimitrov, 2001). We found that the number of cells showing immunoreactivity to pHH3 (and thus undergoing mitosis) decreased after shPax2 (Fig. 7A, B). The reduction in mitotic cells was confirmed by using consecutive pulses of the thymidine analogues, BrdU and EdU. shPax2 electroporation reduced the number of PPA cells that had taken up BrdU, EdU, or both when compared to the control shScr construct (Fig. 7C–E).

thumbnail image

Figure 7. Pax2 repression reduces the number of proliferating PPA cells. A: shPax2 reduces the number of pHH3-positive cells in the PPA (boxed area). B: Graph showing the difference in the number of pHH3-positive PPA cells between control, shPax2 and mPax2 electroporated embryos. In control electroporated embryos, the electroporated side of the embryo has 1.42% fewer pHH3-positive cells (n=881 in 20 embryos) than the unelectroporated (n=922 in 20 embryos). Twenty-two percent fewer pHH3-positive cells (n=279 in 11 embryos) are found in the shPax2 electroporated side of these embryos than the non-electroporated side (n=359 in 11 embryos) **P<0.01. The reduction observed by CAG-mPax2 over-expression of 3.7% on the treated side (n=573 in 19 embryos) compared to the untreated side (n=608 in 19 embryos) was not statistically significant. C, D: Sections of control shScr (C) and shPax2 (D) electroporated embryos treated with first BrdU and then EdU; electroporated cells are false coloured in blue, BrdU is in red and EdU in green. E: Graph showing the percentage of electroporated cells also positive for BrdU (red), EdU (green), or both (blue). Pax2 repression causes less BrdU and EdU incorporation (n=9 separate samples, compared to 3 separate control samples). **P < 0.01.

Download figure to PowerPoint

A similar analysis was performed after Pax2 over-expression using CAG-Pax2-GFP (Fig. 7B). Neither the number of pHH3-positive cells nor the numbers of the BrdU/EdU-positive cells (data not shown) were significantly changed after Pax2 over-expression.

DISCUSSION

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. RESULTS
  5. DISCUSSION
  6. EXPERIMENTAL PROCEDURES
  7. Acknowledgements
  8. REFERENCES

Pax2 is one of the earliest responses to FGF signalling during PPA induction. However, its role during this process is ambiguous. Our analysis suggests both over-expression and repression of Pax2 blocks the formation of the otic and epibranchial lineage to varying extents. Cell cycle analysis reveals one possible mechanism for this inhibition: we find that Pax2 suppression reduces the number of cells undergoing mitosis, suggesting a role for Pax2 in maintaining the proliferation of PPA progenitors.

Loss of Pax8

Pax2, -5, and -8 form a subgroup amongst the vertebrate Pax genes, based on both sequence comparison and ancestral relationship (Vorobyov and Horst, 2006). During inner ear induction in zebrafish, Xenopus, and mouse, Pax8 and Pax2 are both expressed in the inner ear, and their function in early inner ear induction is thought to be redundant (Heller and Brandli, 1999; Bouchard et al., 2002; Hans et al., 2004; Mackereth et al., 2005). We find that in the lineage that leads to birds and lizards, the Pax8 gene has been lost. Our sequence comparison found that the only region of the chick and zebrafinch genome that aligns with human, mouse, or even platypus Pax8, corresponds to the Pax5 gene. Similarly, the only region of the lizard, Anolis, genome that aligned with mammalian Pax8 corresponded to the Anolis Pax2 locus. Of course, this analysis did not rule out the possibility that the part of the genome containing the Pax8 gene in chick, zebrafinch, and Anolis had not been sequenced. However, it is unlikely that in all three genomes, the same region would not have been covered. Syntenic comparison suggests a mechanism by which Pax8 may have been lost. In mammals, the region immediately downstream of Pax8 in mammals is a conserved chromosomal break-point (Becker and Lenhard, 2007) and the organization of genes in this region differs even amongst closely related mammals. We can thus speculate that in the lineage leading to birds and lizard, a chromosomal rearrangement occurred that led to an inactivation of the Pax8 gene. Functional redundancy with Pax2 in this lineage may then have permitted the loss of Pax8. However, it is not clear how diapsids have released the constraint on Pax8 retention. In other tetrapods, Pax8 and Pax2 are required non-redundantly for kidney development and in neural patterning, and thus it is possible that other genes may have taken over this role in diaspids. More developmental insights will undoubtedly shed further light on the mechanisms that have led to Pax8 loss in birds and lizards.

Pax2 Control of Inner Ear and Epibranchial Development

Despite Pax2 being an early response to FGF-mediated induction of the PPA, general markers of the PPA, such as Dlx3 and Pax2, together with genes expressed in the neurogenic and epibranchial portions of the PPA, Sox3 and Foxi2, were not affected when Pax2 was repressed. This corroborates previous experiments in zebrafish showing relatively normal PPA development after pax8 and both zebrafish pax2 genes were repressed (Hans et al., 2004; Mackereth et al., 2005; Padanad and Riley, 2011). It is thus likely that other transcription factors, induced in response to FGF signalling, mediate the specification of the PPA, and these factors control the expression of Pax2. Indeed, this idea has been suggested from experiments in zebrafish. Here, pax2 or pax8 over-expression can only expand the inner ear in an FGF-dependent manner (Padanad et al., 2012).

While induction of the PPA does not appear to be affected by Pax2 repression, specification of the inner ear and epibranchial placodes are repressed; otic development is reduced and epibranchial neurogenesis blocked. Previous studies using morpholinos to down-regulate Pax2 in the chick have suggested that the repression of otic differentiation results from the inhibition of placodal morphogenesis (Christophorou et al., 2010). However, we do not observe such a dramatic phenotype on PPA cells. Instead, we find that after repression by either short-hairpin-encoded constructs or the repressive form of Pax2, a smaller otocyst is observed. This is reminiscent of the phenotype of the mouse Pax2/Pax8 double knockout, which also shows a smaller, but well-formed otocyst (Bouchard et al., 2010). The apparent difference between phenotypes may be due to the extent of Pax2 repression. Zebrafish studies suggest that reduced levels of Pax2 favor an epibranchial fate. In morpholino-mediated knockdowns, residual Pax2 may act to transform otic placode cells into epibranchial neuroblasts resulting in a perturbation of cell identity and morphology within the otic placode. We suggest that as a result of short-hairpin-mediated knockdown or introduction of a repressive form of Pax2, the smaller otocyst is due to changes in the proliferation of PPA cells. This is likely to be the underlying cause of the smaller otocyst in Pax2/Pax8 mutant mice (Bouchard et al., 2010).

Pax2 down-regulation repressed the differentiation of the epibranchial-derived neuroblasts. In zebrafish, it has been suggested that this is due, in part, to a cell-autonomous requirement of pax2 in the otic placode controlling the expression of fgf24, which is required for epibranchial induction (Padanad and Riley, 2011) as well as a cell-autonomous role for pax2 in the specification of the zebrafish epibranchial placodes (McCarroll et al., 2012). In the chick, otic development still occurs after Pax2 repression, but epibranchial formation is nonetheless curtailed. This argues against the non-cell-autonomous role for Pax2 in epibranchial development. However, the paralogue of zebrafish fgf24, chick Fgf18 (Itoh and Konishi, 2007; Wotton et al., 2008), is not expressed in the otic placode. Fgf18 is found in ectoderm lateral to the otic placode, at the site of the epibranchial placodes (Karabagli et al., 2002). Thus, it is possible that Pax2 repression in chick does inhibit Fgf18 expression, similar to its proposed role in zebrafish (Padanad and Riley, 2011). Although the function of Fgf18 in epibranchial formation has yet to be determined, we hypothesize that the more severe epibranchial phenotype is a combination of reduced progenitor production in the PPA, aberrant specification of the epibranchial portion of the PPA (McCarroll et al., 2012), and reduced autocrine Fgf18 action during epibranchial specification.

We find that Pax2 over-expression also leads to an inhibition of both inner ear and epibranchial differentiation. The suppression of inner ear differentiation by Pax2 over-expression is more extreme than that seen for Pax2 repression. This is not due to increased cell death; we do not observe increased numbers of apoptotic cells. One possible explanation could be that over-expression leads to high levels of Pax2 protein that are far in excess of its possible DNA binding sites and thus sequesters the necessary co-factors that endogenous Pax2 needs for its activity. However, there are several qualitative differences between Pax2 gain of and loss of expression. The clearest is the effect on the cell cycle, where Pax2 over-expression shows little effect on the rate of division of PPA cells in this experimental design. In addition, the effect on tissue morphogenesis is also different, with Pax2 over-expression frequently affecting otic invagination, as has already been reported (Christophorou et al., 2010). Over-expression data from zebrafish suggest that Pax2 over-expression expands early otic territories (Padanad et al., 2012; McCarroll et al., 2012). In the zebrafish pax2 expression is expanded anteriorly (Padanad et al., 2012; McCarroll et al., 2012). However, this expansion of the early otic territory does not translate into an expanded otic terrain, or ectopic otic vesicles. Importantly, the transient over-expression in zebrafish does not inhibit endogenous otic vesicle formation while, as we and others have shown, sustained high-level Pax2 over-expression in the chick does (Christophorou et al., 2010). This suggests that the temporal regulation of Pax2 is as important as the exact levels of expression for inner ear and epibranchial formation; too little and the cells of the PPA do not proliferate adequately, whereas too much, or for too long, prevents the differentiation of the PPA into its derivatives. This is reminiscent of the action of Fgf, the signalling molecule responsible for Pax2 induction. Here sustained Fgf action, like Pax2 over-expression, is inhibitory to the differentiation of the inner ear (Freter et al., 2008).

Control of Cell Division by Pax2

A function for Pax genes in progenitor maintenance is not without precedence. Indeed, numerous studies have pointed to a role for Pax genes in the maintenance of progenitor cells. For example, Pax7 is expressed in skeletal muscle satellite cells, a resident progenitor population that can generate new myoblasts (Seale et al., 2000; Zammit et al., 2004). In Pax7 mutants, satellite muscle cells are formed, but in reduced numbers that fall further over time (Zammit et al., 2004). In addition, forced expression of Pax7 in satellite cells prevents their difference into myoblasts (Olguin and Olwin, 2004). Similarly the activity of Pax6 is involved in conferring progenitor status to neuronal populations in the spinal cord, cerebral cortex, adult hippocampus, and olfactory bulb (Grindley et al., 1995; Kohwi et al., 2005; Bel-Vialar et al., 2007; Tucker et al., 2008). Similarly, the expression of Pax2 and Pax8 is associated with proliferating progenitor populations in the developing kidney (Bouchard et al., 2002) and in the sensory patches of the inner ear (Li et al., 2004; Warchol and Richardson, 2009). Thus, the activity of Pax2 in the PPA is consistent with the generalized model of Pax function that has emerged (Blake et al., 2008). The question of how Pax2 maintains the PPA progenitor population is undoubtedly important and identification of interacting partners and downstream targets will illuminate these pathways.

The mechanism by which Pax2 mediates cell cycle control is not known. However, recent data from studies on Pax8 provide important insights. In human cancer cell lines, Pax8 regulates the G1/S transition by regulating the transcription of the transcription factor E2F, which is important for entry into S-phase, as well as protecting retinoblastoma (Rb) protein from degradation (Li et al., 2011). Given the functional similarity to Pax8, we speculate that the control of PPA proliferation exerted by Pax2 is likely to be mediated by a similar mechanism.

EXPERIMENTAL PROCEDURES

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. RESULTS
  5. DISCUSSION
  6. EXPERIMENTAL PROCEDURES
  7. Acknowledgements
  8. REFERENCES

Chicken Embryos

Fertilised hens' eggs (Shiroyama Farms, Kanagawa, Japan) were incubated in a humidified chamber at 38°C until embryos reached the desired stage. Embryos were staged by counting somites (ss), or by using the stage series of Hamburger and Hamilton (HH; Hamburger, 1951). Following manipulations, embryos were incubated to the required stage, fixed in 4% paraformaldehyde (PFA), and rinsed in PBS.

Constructs and Electroporation

Knockdown of Pax2 was performed using shRNA-encoding constructs. Briefly, putative shRNA sequences designed against 20 bases starting from position 395 of Pax2 were inserted into pRFPRNAi (Das et al., 2006). As control for non-specific effects, a scrambled version of the sequence used for shPax2 was constructed. To verify shRNA knockdown, we also used a repressive form of Pax2. Here the mouse Pax2 coding sequence was cloned in frame with the transcriptional repressor domain from Engrailed (Conlon et al., 1996). To over-express Pax2, the coding region of mouse of Pax2 was cloned in frame with GFP, under the control of the chick β-actin promoter.

Vectors were electroporated unilaterally into the cranial ectoderm, either ex ovo at HH5 or in ovo at HH9/10 (-10ss). After electroporations, embryos cultured ex ovo were placed in a humidified CO2 incubator at 37°C for 10–49 hr. In ovo electroporated embryos were resealed and placed into a humidified incubator at 37°C for 2 to 3 days.

Whole-Mount In Situ Hybridization and Immunohistochemistry

Whole-mount in situ hybridization was performed as described before, using chick probes for: Dlx3, Dlx5, Foxi2, Gbx2, Nkx5.1, Pax2, Phox2a, Phox2b, Soho1, and Tbx1 (Begbie and Graham, 2001; Brown et al., 2005; Freter et al., 2008). The neurofilament template was synthesized by PCR (primer details available on request). Probes were labeled with digoxigenin (DIG) and detected using an alkaline phosphatase-coupled antibody (1:2,000, Roche, Basel, Switzerland) followed by NBT/BCIP (Roche, Basel, Switzerland) staining. Some embryos were cryosectioned following in situ hybridization. Immunohistochemistry on frozen sections or on whole embryos was as previously described (Teraoka et al., 2009). The following primary antibodies were used; anti-Ds-Red (Clontech, Mountain View, CA), anti-GFP (Invitrogen, Carlsbad, CA), anti-neuro-filament medium chain (BD Biosciences, San Jose, CA), the cleaved (and active) form of caspase3 (BD Biosciences, San Jose, CA), and anti-phospho-histone H3-Ser10 (pHH3; Millipore, Billerica, MA.). Following PBS washes, the secondary antibodies were added: anti-rabbit or anti-mouse Alexa 594 (both 1:1,000, Invitrogen, Carlsbad, CA) to be visualized under fluorescence, or anti-mouse-HRP (1:1,000, Dako, Glostrup, Denmark) visualized with VIP substrate system (Vector Labs, Burlingame, CA).

EdU and BrdU Incorporation

Electroporated embryos were cultured ex vivo for 6 hr. A pulse of bromodeoxyuridine (BrdU) was applied by dropping 10 μl of 10 μM BrdU in ringer's solution onto the embryo, followed by incubation for 90 min at 37°C. The embryo was then washed in Ringers solution, and placed in culture for 60 min. Then 10 μl of 50 μM ethynyl deoxyuridine (EdU) in Ringer's solution was applied onto the embryo, incubated for 90 min, and then washed out using fresh Ringer's solution. The embryo was then incubated for 12 hr.

After fixation for 30 min in 4%PFA, permeabilization with 0.5% Triton in PBS, and antigen retrieval using 2M hydrochloric acid for 20 min, BrdU was detected using an anti-BrdU antibody, following the manufacturer's instructions. Alexa-488 or 594 anti-mouse secondary antibodies were used. The manufacturer's instructions were followed for EdU detection, using either Alexa-488 or Alexa-657. Embryos were then sectioned, and labelled cells in the PPA were counted.

Statistical Analysis

Embryos were electroporated on either the left or right sides. To measure the size of the otic placode, morphology as well as Soho1 staining was used, and the otic area from 8 different embryos measured using Image J.

pHH3-positive cells were counted and the normal difference between the electroporated and non-electroporated sides of 9 control electroporated embryos between 10ss and 13ss calculated, as described previously (Freter et al., 2008). The difference in pHH3-positive cells between control sides and electroporated sides of 9 shPax2-RFP and 19 CAG-Pax2 electroporated embryos were similarly calculated. P values were calculated using Student's t-tests.

BrdU and EdU incorporation in electroporated cells was determined, and P values calculated using a two-tailed t-test.

Bioinformatic Analysis

To verify that the Pax8 gene was missing in birds and lizard genomes, we adopted an in silico approach. Using the UCSC genome browser (http://genome.ucsc.edu), we first identified the region of the human genome containing PAX8 and compared it to mouse, platypus, Xenopus, zebrafish, zebra finch, lizard, and chicken genome net tracks (full mode). Each one of the best match (level 1) items from the genomic net tracks was then manually analyzed. For chicken, the level 1 item corresponded to genomic coordinates chrZ: 74379713–74390847 of galGal3 assembly. This region contained the first 4 exons of chicken Pax5. For zebra finch, the corresponding region of level 1 genome net track alignment had the genomic coordinates chrZ: 72609752–72621819 (taeGut1 assembly). The best alignment in this region was with Pax5 (acc. no. NM_016734) followed by Pax2 (NM_003990). Similarly, the best matching item at level 1 for the lizard (scaffold_508:852540–853832, anoCar1 assembly) was Pax2 protein (NM_003990). For other species, the level 1 items all corresponded to Pax8. Phylogenetic comparison was performed using the program PhyML (Guindon and Gascuel, 2003), assuming the WAG+I+G model.

Syntenic comparison was performed using the ENSEMBL genome browser, by first identifying the coordinates for human Pax8 and then using the functionality of the browser to inspect syntenies in 11 amniote species.

Acknowledgements

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. RESULTS
  5. DISCUSSION
  6. EXPERIMENTAL PROCEDURES
  7. Acknowledgements
  8. REFERENCES

We thank members of the Laboratory for Sensory Development for a critical reading of this manuscript, and to Yoko Matsuoka for excellent technical support. This work was supported by a RIKEN CDB intramural grant, RIKEN FPR fellowship (P.O.), and a JSPS post-doctoral fellowship (S.F.).

REFERENCES

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. RESULTS
  5. DISCUSSION
  6. EXPERIMENTAL PROCEDURES
  7. Acknowledgements
  8. REFERENCES