Competence, specification and commitment to an olfactory placode fate

Cranial sensory placodes arise as transient thickenings of embryonic head ectoderm (van Wijhe, 1883; von Kupffer, 1891) and invaginate/ingress to form components of the nose, lens and inner ear(Bailey and Streit, 2006; Baker and Bronner-Fraser, 2001; Schlosser, 2006). Of the three sensory placodes, the lens and otic are relatively easy to distinguish and manipulate early in development. Therefore, tissue interactions and signals guiding their induction have been studied extensively (reviewed by Brown et al., 2003; Chow and Lang, 2001; Donner et al., 2006; Grainger, 1992; Noramly and Grainger, 2002; Riley and Phillips, 2003; Saha et al., 1989; Streit, 2001; Torres and Giraldez,1998).

By contrast, development of the olfactory sensory system has been primarily investigated at later stages. An attractive model for studying neurogenesis,its myriad of derivatives include the regenerative odorant sensing olfactory neurons in the olfactory epithelium (Calof et al., 1998; Graziadei and Graziadei, 1979; Graziadei and Metcalf, 1971; Graziadei and Monti Graziadei, 1983), the ingressing gonadotropin-releasing hormone neurons (reviewed by Parhar,2002; Wray, 2002)[for a contrary view see Whitlock(Whitlock, 2005)], the pheromone-detecting vomeronasal organ(Dulac, 1997) and other neuromodulatory and neuroendocrine cells(Northcutt and Muske, 1994; Tarozzo et al., 1995; Yamamoto et al., 1996). It is the only placode to give rise to glial cells that ingress and migrate along the olfactory nerve towards the brain(Chuah and West, 2002; Ramon-Cueto and Avila, 1998). Cues that guide differentiation along these different pathways and the molecular mechanisms underlying the patterned wiring of olfactory sensory neurons have been examined extensively (reviewed by Baker and Margolis, 2002; Balmer and LaMantia,2005).

Contrasting with this wealth of data, inductive events that initiate olfactory development remain ambiguous, partially because the placode is morphologically invisible until relatively late (HH13+ in birds). Transplantation studies show that precursors straddle the lateral anterior neural folds and adjacent ectoderm in the neurula(Couly and Le Douarin, 1985),but their exact location at intervening stages was unclear. Recent cell marking studies in zebrafish and chick have examined the origin of the olfactory placode at higher resolution and at intervening stages(Bhattacharyya et al., 2004; Whitlock and Westerfield,2000). Olfactory precursors are initially distributed in a broad region spanning the anterior neural folds and overlapping with lens precursors. Over time, they become progressively restricted to the most anterior ectoderm, finally resolving to a discrete olfactory pit. Knowledge of the location of precursors at different stages has facilitated manipulation of their development.

To understand how and when the nasal epithelium is first induced, we examined competence, specification and commitment of ectoderm toward an olfactory fate. This information is a necessary prerequisite for interpreting results of functional perturbations at the molecular level. First, we correlated molecular markers with a more complete fate-map(Bhattacharyya et al., 2004)and found that Dlx3, Dlx5 and Pax6 can serve as molecular guideposts for olfactory placode precursors at different stages. Although an initially broad region of head ectoderm is competent to form olfactory placode, this ability is restricted with time. In vitro explant assays suggest that placode induction by adjacent tissues occurs by ∼HH10, when olfactory precursors have segregated from their neighbors. In vivo heterotopic transplantation studies demonstrate that commitment occurs with a time lag of 12 hours. This suggests a period of developmental plasticity, perhaps accounting for the regulative nature of early olfactory placode development. This step-wise progression of placode formation favors flexibility in fate until such time as the placode becomes morphologically apparent.

Fertilized chick and quail eggs were incubated at 38°C to the desired stage (Hamburger and Hamilton, 1951). Host embryos were visualized using India Ink or dye, and the vitelline membrane removed. Before HH8, clean quail ectoderm was obtained without enzymatic treatment. Between HH10-16, donor ectoderm was separated from mesenchyme and forebrain using 1 mg/ml dispase+0.25% trypsin in DMEM (Boehringer Mannheim; 30 minutes on ice, 10 minutes at 37°C) or +0.2-0.25% collagenase (Worthington) in PBS (20 minutes at 37°C). Mid/hindbrain level ectoderm was dissected from neural crest using 1 mg/ml dispase. Ectoderm pieces were washed and allowed to recover before grafting and positioned in the correct inside-outside orientation; embryos were incubated for 48-72 hours and fixed in 4%paraformaldehyde. Seventy out of 264 surviving chimaeras showed appropriate graft incorporation.

Quail embryos (HH8,10,14) were collected in Ringer's solution. Small pieces of HH8 anterior ectoderm or HH10-14 presumptive olfactory placode were dissected as above and stored in PB1 medium on ice until use. Collagen (90μl; Collaborative Research) +10 μl of 10× DMEM (Invitrogen) in 0.375% sodium bicarbonate was solidified in four-well plates (Nunc); tissue pieces were transferred, overlaid with 5 μl collagen mix, immersed in 0.5 ml of the defined medium, F12+N2 (Invitrogen), incubated for 24-48 hours at 37°C with 5% CO₂, then fixed in 4% paraformaldehyde for 30 minutes at room temperature.

pCIG-GFP plasmid DNA (Megason and McMahon, 2002) was injected at a concentration of 2.5 μg/μl in Ringer's solution with 50 ng/μl Fast Green between the embryo's ventrum and extra-embryonic ectoderm at HH10-11. Platinum electrodes (4 cm long and 4 mm apart) and the electroporator were custom-built in our laboratory and by the electronics shop, respectively. The positive electrode was placed on top and the negative electrode under the anterior tip of the embryo and five pulses of 8-10V for 50 ms at 100 ms intervals were applied. Embryos were reincubated until HH18 or 27 (23/47 and 14/29 survivors, respectively),collected, fixed in 4% paraformaldehyde overnight at 4°C and prepared for cryosectioning.

Embryos and collagen gels were embedded in 7.5% gelatin in 15% sucrose and cryosectioned. Sections were degelatinized, stained with antibodies in blocking solution, rinsed and mounted in Fluoromount-G (Southern Biotechnology) or Permount containing 10 μg/ml DAPI. Primary antibodies were: QCPN (DSHB; 1:1), QCPN conjugated to Alexa-568 (at the University of Oregon; 1:50), Pax6 rabbit polyclonal (Covance; 1:100), Pax7 monoclonal (DSHB;1:20), Dlx3 goat polyclonal (1:1500), pan-Dlx rabbit polyclonal (Dr Jhumku Kohtz; 1:70), Hu monoclonal (Molecular Probes; 1:250), chick GnRH-I rabbit polyclonal (Dr James Millam; 1:100) and mouse GnRH-I rabbit polyclonal (Abcam;1:100). Antibodies against the mouse and chick GnRH-I peptides, which differ at amino acid position 8, were used concurrently to maximize staining efficiency. Secondary antibodies used include Molecular Probes Alexa 488 and Alexa 568 (1:1000-1:2000), and Jackson ImmunoResearch AMCA and Cy5(1:200-1:400).

Tri-color images were acquired on a Zeiss Axioskop2 using an Axiocam Mrm camera and Axiovision software or a Zeiss LSM 510 Meta confocal microscope.

GST-C-terminal chick Dlx3 fusion protein (amino acids 185-278) was produced in BL-21 cells and purified using Glutathione Sepharose beads (Pharmacia)(Sambrook and Russell, 2001). Goats were injected with 5 mg Dlx3 protein (Alpha Diagnostic). Immunoblots and immunostaining confirmed specificity of each bleed.

Whole-mount in situ hybridization was performed as described(Streit and Stern, 2001) at a hybridization temperature of 70°C using cDNAs for Dlx3 (Dr Andy Groves),Pax6 and Dlx5 (Dr Michael Kessel).

Avian olfactory placode development was followed from dispersed cells in the ectoderm to the morphologically thickened and invaginated olfactory pit,taking advantage of the spatiotemporal expression of several transcription factors that cumulatively reflect acquisition of olfactory fate. Chick Dlx5 and Pax6 transcripts are detected as early as HH6 (neurula stage) in olfactory and lens precursors in anterior cranial ectoderm(Bhattacharyya et al., 2004). Their overlap is maintained until just before HH10, when Dlx5 is upregulated and Pax6 is simultaneously downregulated in presumptive olfactory placode cells (Fig. 1B,C,E,F)(Bailey et al., 2006). At HH8,both Pax6 and Dlx5 are expressed in a `salt and pepper' pattern in ectoderm abutting the neural folds containing nasal precursors(Fig. 1B,B″,C,C′)(Bhattacharyya et al., 2004);anterior-most ectoderm expresses Dlx5 uniformly(Fig. 1B′). Pax6 expression is also observed within anterior neural folds, which contribute to olfactory epithelium and telencephalon(Fig. 1C). Conversely, Dlx3 is absent from anterior neural folds and low in adjacent ectoderm at HH8(Fig. 1A,A′), but strongly upregulated in olfactory placode precursors at HH10(Fig. 1D,D′)(Pera and Kessel, 1999). Expression refines over time to the medioventral region of the head ectoderm between the forming eyes (Fig. 1G,G′). By HH11, olfactory precursors occupy a discrete region in the ventral ectoderm(Bhattacharyya et al., 2004),where Dlx5 is expressed continuously (Fig. 1E,E′), but Pax6 is extinguished(Fig. 1F,F′). Interestingly, Pax6 is re-expressed once the olfactory placode has formed at HH14 (Fig. 1I,I′),contiguous to corneal Pax6 expression (Fig. 1I′, arrow) and subsequently spreads anteromedially. The thickened placode shows continued expression of Dlx3 and Dlx5 transcripts at HH14+ (data not shown) (Fig. 1H,H′). The first co-expression of all three transcripts is coincident with morphological appearance of the olfactory placode.

As expected, no marker gene is expressed exclusively in the mature olfactory placode/pit by HH19-20 (Fig. 2A-C′). While Dlx3 is found in olfactory and otic placodes(Fig. 2A,A′), Dlx5 is also expressed in the adenohypophysis and both are expressed in subpopulations of neural crest (Fig. 2A,B;data not shown). Dlx3 is expressed in cranial crest populating distal elements of branchial arches 1-2 (Fig. 2A) (Depew et al.,2005; Pera and Kessel,1999), whereas Dlx5 is more widely expressed in branchial arches 1-4 and the trigeminal ganglion (Fig. 1H,H′; Fig. 2B) (Pera et al.,1999). Dlx5 is also evident in newly emergent trunk neural crest(Fig. 3B). Like the Dlx genes,Pax6 is expressed at multiple sites, including lens and nasal placodes(Fig. 2C,C′). However,combined expression of Pax6 and Dlx genes by HH19 is unique to the olfactory placode.

Unlike Dlx3 and Pax6, Dlx5 is expressed in a few cells in the mesenchyme as the olfactory placode thickens (Fig. 1H′, arrow), even continuing to HH20(Fig. 2B′, arrowhead). This raises the intriguing possibility that it marks a subset of neuronal cells ingressing from the olfactory placode. To test this, HH10 ventral ectoderm was electroporated with a GFP construct (survivors n=37) and examined for expression of Dlx and GnRH neuropeptide. At HH18, GFP+ cells are Dlx+ within the olfactory epithelium (Fig. 2D,d). Moreover, ingressing GFP+/Hu+ neurons are also Dlx+(Fig. 2d′,d″). At HH27, by which time GnRH neurons are detectable(Fig. 2G)(Mulrenin et al., 1999), some GnRH+/GFP+ cells were located within a ganglion-like structure between the epidermis and ventral forebrain (Fig. 2e,f), unequivocally demonstrating their placodal rather than neural crest origin. Adjacent sections also had GFP+/Dlx+ cells(Fig. 2e′,f′),leading to the conjecture that some GnRH cells may co-express Dlx; direct double-labeling was not possible as both antibodies were raised in the same species.

To detail temporal expression of Dlx3 protein in the developing olfactory placode, we generated a polyclonal antibody against the C terminus of chick Dlx3, a region of minimal similarity between the six Dlx family members. The polyclonal antibody recapitulates Dlx3 mRNA expression, though the protein has more widespread expression in ectoderm than the mRNA (compare Fig. 2A′ with Fig. 3C). To ensure against crossreactivity with Dlx5, which has highest sequence homology and maximum expression overlap, we compared localization of Dlx5 and Dlx3 transcripts and protein. At HH11, the otic placode at the rhombomere (r) 5 level begins to express Dlx5 weakly (Fig. 3A),while strongly expressing Dlx3 (not shown). HNK1+ neural crest cells atop the neural tube, however, express Dlx5 robustly (compare outlined domains in Fig. 3A,A′). The Dlx3 antibody clearly recognizes otic placode cells (also HNK1+) but not dorsal neural tube, conclusively demonstrating lack of crossreactivity with Dlx5(compare Fig. 3A and Fig. 3A′). Similarly,emigrating trunk neural crest expresses Dlx5 but not Dlx3 (compare Fig. 3B with Fig. 3B′). To ensure against crossreactivity with other Dlx family members, we co-labeled sections through heads of 3-day-old chick embryos with Dlx3 and pan-Dlx antibodies(Panganiban et al., 1995). Dlx3 is the only family member not expressed in the CNS for the first three days (data not shown) (Beanan and Sargent,2000; Liu et al.,1997; Panganiban and Rubenstein, 2002; Pera and Kessel, 1999; Price,1993; Robinson and Mahon,1994) (see also Dhawan et al.,1997; Zhu and Bendall,2006). However, it is expressed robustly in the olfactory epithelium at HH18-19 (Fig. 3C). Other Dlx family members expressed in the olfactory epithelium include Dlx5/6 (Brown et al.,2005; Pera and Kessel,1999). The pan-Dlx antibody recognizes all chick Dlx proteins;therefore, strong co-localization of both the Dlx3 and pan-Dlx antibodies is noted in the nasal pit (Fig. 3C′; Fig. 3C″, yellow). Dlx1, 2, 5 and 6 are expressed in the telencephalon at various times in many vertebrates, including chick(Eisenstat et al., 1999; Liu et al., 1997; Panganiban and Rubenstein,2002), and are detected by the pan-Dlx but not the Dlx3-specific antibody (Fig. 3C″). These data indicate that the chick Dlx3 antibody is specific and does not crossreact with closely related family members.

We next investigated colocalization of Dlx3 and Pax6 in cells fated to form olfactory placode at single cell resolution. Just prior to olfactory placode thickening (HH13), Dlx3 protein is present at high levels in forebrain-level ectoderm (Fig. 3D,D″),whereas Pax6 protein is in scattered cells more laterally(Fig. 3D′,D″). The number of cells co-expressing both progressively increases such that by HH19-21, robust expression of Dlx3 and Pax6 proteins is noticeable in the now pseudostratified placode epithelium (Fig. 3E-E″). Significant overlap is seen in the cup-shaped olfactory epithelium (Fig. 3E″) though domains of solely Dlx3+ or Pax6+ cells are also observed. Similar results were seen in quail embryos (data not shown).

The first signs of differentiation occur at ∼HH14, when earliest ingressing placode cells express the RNA-binding protein Hu, which marks post-mitotic neurons (Fornaro et al.,2001; Fornaro et al.,2003). By pit stages, Hu labels abundant neurons within the epithelium, which express Pax6 and/or Dlx3, as well as those migrating along the olfactory nerve towards the telencephalon(Fig. 3F,G). Ultimately, Pax6 expression is only maintained in a subpopulation of non-neuronal cells in the nascent olfactory epithelium (Davis and Reed, 1996). Dlx3+ neurons are unique to the olfactory system at least at HH19, although Dlx3 expression has been reported in the ventral forebrain at HH27-29 (Zhu and Bendall,2006). Accordingly, we observed Dlx3+ but Hu-cells in the midline between the two telencephalic vesicles at HH19 (not shown).

Interactions between placodal ectoderm and frontonasal mesenchyme, which is largely neural crest derived(Osumi-Yamashita et al., 1994; Serbedzija et al., 1992), play a pivotal role in orchestrating development and differentiation of the nasal epithelium and capsule (Bailey et al.,2006; Firnberg and Neubüser, 2002; LaMantia et al., 2000). To assess the contribution of neural crest cells to olfactory placode specification and commitment, we used Pax7 as a marker for neural crest and the lateral domain of frontonasal mesenchyme(Fig. 3H)(Firnberg and Neubüser,2002).

Fate maps (Couly et al.,1985; Bhattacharyya et al.,2004) suggest that the chick nasal placode arises from around the anterior neural folds at HH8. Accordingly, isotopic and isochronic quail/chick grafts of HH8 anterior neural fold (Fig. 4a) or adjacent ectoderm (Fig. 4b) gave rise to quail cells [distinguished using quail cell peri-nuclear (QCPN) antigen] in the olfactory placode 2-3 days later(Fig. 4A′,B′). Donor cells expressed Dlx3, Pax3 (data not shown), Pax6 (n=6)(Fig. 4A-A″,B-B″)and differentiated as neurons. Because olfactory precursors closely associate with both lens and forebrain progenitors(Bhattacharyya et al., 2004),quail cells were additionally found in the telencephalon(Fig. 4A′, arrow; Fig. 4A″) or ectoderm adjacent to the olfactory placode including parts of the cornea and lens(Fig. 4B′; data not shown).

To determine the earliest time when ectodermal precursors are capable of responding to olfactory placode-inducing cues, HH4-5 (gastrula to early neurula) anterior or lateral epiblast (adjacent to the area opaca) was heterochronically grafted into ectoderm adjacent to the anterior neural fold at HH8. This most anterior or lateral epiblast does not contribute to neural ectoderm but eventually forms non-neural ectoderm, neural crest and otic placode (Fernandez-Garre et al.,2002; Garcia-Martinez et al.,1993; Rosenquist,1966; Schoenwolf and Alvarez,1991), none of which co-express Pax6/Dlx3 and form Dlx3+ neurons. However, this epiblast can express Pax6 and Dlx3 in the olfactory placode when grafted to anterior neural fold region (n=3)(Fig. 4c,C-C″; data not shown). Interestingly, grafted tissue underwent thickening and invagination to form another placode-like structure when incorporated aside the endogenous olfactory placode (Fig. 4C′, boxed area). A few graft-derived cells differentiated into Dlx3+ olfactory neurons in the thickened epithelium (compare Fig. 4C″ and Fig. 4D″). Additionally,we noted many quail Hu+ cells lining the olfactory nerve(Fig. 4D″, inset). This demonstrates that though epiblast tissue is regionally specified(Wilson et al., 2001), it nonetheless is plastic and retains competence to respond to diverse inducing signals and differentiate accordingly.

We determined the competence of ectoderm at different axial levels to contribute to olfactory epithelium and express olfactory markers as a function of time by grafting surface ectoderm from HH8-13 donor quails, to the prospective nasal territory of HH7-9 chick hosts. Chimaeras were allowed to survive until the olfactory pits were distinctly visible (HH18: ∼48-72 hours later).

Midbrain-level ectoderm, which is fated to contribute to trigeminal ganglion (D'Amico-Martel and Noden,1983; Xu et al.,2008), robustly expresses Pax3 from HH8(Stark et al., 1997), as well as low levels of Pax6 at HH8 (Fig. 1C) and Dlx3 at HH10 (not shown). However, their expression is not maintained in the ophthalmic trigeminal (opV) placode(Fig. 2A,C). Therefore, we looked for prolonged expression of these markers and concomitant neurogenesis in midbrain level ectoderm grafted adjacent to anterior neural folds(Fig. 5b,c) (n=15). HH8 presumptive opV placode ectoderm expresses both Dlx3 and Pax6 at high levels following incorporation into the olfactory placode and robust expression is maintained in grafted ectoderm taken from HH10 embryos(Fig. 5B-B″). Similarly,opV placode ectoderm that simultaneously incorporates into cornea and lens,expresses high levels of Pax6 in these novel locations(Fig. 5E,F). Additionally,quail neurons contribute to the migratory mass lining the olfactory nerve(Fig. 5C-C″). Shortly after HH10, Pax6 expression is extinguished in grafted cells. In fact,ectoderm grafts taken from HH12 and onwards show no Pax6 expression in peripheral ectoderm that is ordinarily Pax6+. Weak expression is observed in a few cells only if the graft incorporates into the placode (data not shown).

Ectoderm adjacent to r1-6, including presumptive otic placode ectoderm(Couly and Le Douarin, 1990; Groves and Bronner-Fraser,2000; Streit,2002) was tested for its potential to contribute to olfactory placode (n=16). Hindbrain-level ectoderm does not express Pax6 at early stages in situ; however, Dlx3 is expressed at high levels by otic ectoderm from HH9 onwards and at low levels in ectoderm rostral to otic ectoderm (Pera and Kessel,1999) (data not shown). Hindbrain-level ectoderm (HH8-) can replace ablated nasal progenitors and express both Dlx3(Fig. 5A-A″) and Pax6 plus generate placode-derived neurons that line the olfactory nerve. Shortly thereafter, Pax6 expression is no longer induced in hindbrain-level ectoderm,even when it spans the corneal ectoderm, which usually expresses Pax6 at high levels (Fig. 5G). However,three cells that contribute to the lens express Pax6 at this stage (HH8+)(Fig. 5G, arrowhead). Dlx3 is only weakly induced at these stages (data not shown). Neither gene is expressed in grafts of pre-otic ectoderm from HH10-11 embryos; instead,otic-derived ectoderm expresses Pax2 (an otic placode marker) in its new location. Thus, hindbrain-level ectoderm loses capacity to respond to olfactory placode-inducing signals with age more rapidly than midbrain level ectoderm.

Trunk level ectoderm does not express Dlx3 except in the flank that forms the apical ectodermal ridge of the limbs. It never expresses Pax6 nor gives rise to placodes. To test for placodal competence, trunk level ectoderm(n=5) at the level of the 5th-9th somites was heterotopically transplanted adjacent to the anterior neural folds of HH8 to HH8+ embryos. Although the grafted ectoderm could not express Pax6 even at the earliest ages tested (HH8-) (Fig. 5D-D″), it could express Dlx3 up until HH9-. Thus,competence to express olfactory markers was particularly short-lived in trunk ectoderm.

In conclusion, all levels of head ectoderm (which have some degree of placodal bias) are competent to form Dlx3+/Pax6+ cells and Hu+ neurons in olfactory epithelium and along the olfactory nerve. However, the level of competence in embryonic ectoderm is graded such that ectoderm closest to olfactory progenitors remains competent for the longest duration.

Classically, specification is defined as the potential of tissue explanted and grown without additional activating/inhibitory differentiation signals,indicating their degree of progression along a particular developmental pathway. To determine the time by which olfactory induction had occurred,presumptive olfactory ectoderm was cultured without serum for 24-72 hours in three-dimensional collagen gels. Expression of Pax6, Dlx3 and neuron formation was assessed in isolates from HH8-14 embryos(Fig. 6).

At HH8, olfactory precursors intermingle with lens and epidermal precursors(Bhattacharyya et al., 2004),precluding isolation of a homogenous population of nasal progenitors. Bearing this in mind, the explanted presumptive olfactory-lens region is both Pax6+and Dlx3+ at 24 hours or 72 hours of culture (n=18/18)(Fig. 6A-A″). Interestingly, this ectoderm shows spatially segregated zones of gene expression, with the Dlx3+ cell cluster at one end of the explant(Fig. 6A) and Pax6+ cells at the other, forming a lentoid-like structure(Fig. 6A′, arrowhead). Pax6+/Dlx3+ cells are found at the intersection of the two domains(Fig. 6A″, arrows). This cell sorting is reminiscent of Dlx3 and Pax6 expression in the presumptive placodal region at HH13 (Fig. 3D″). Consistent with the possibility that placodal development in HH8 explants cannot progress beyond this stage in vitro,neither thickening nor neuronal differentiation were observed in these explants even after 3 days (Fig. 6A-A″) (Bailey et al.,2006). Although this appears contradictory to the findings of Sjodal et al. (Sjodal et al.,2007) where over 10% of cells in HH8 olfactory explants were keratin+/Hu+, these explants also expressed neural markers such as L5, HuC/D and Nkx2.1 in a distinct region. Similarly, in our hands, ∼11% of ectodermal cells in HH8 explants were Hu+ when cultured with the anterior neural ridge (Bailey et al.,2006). By HH10, after olfactory and lens precursors have separated(Bhattacharyya et al., 2004),nasal progenitors are specified to express Dlx3 and Pax6 (n=20/20)and 20% of these explants differentiate to give rise to neurons(Fig. 6B,C,C′). Although Dlx3+ neurons can be unambiguously assigned to an olfactory fate(Fig. 6B, arrow), Dlx3-neurons(Fig. 6B, arrowhead) could either be olfactory (Fig. 3G)or forebrain-derived, as a few forebrain cells could have been explanted along with prospective placode. The segregation of Dlx3+, Pax6+ and Dlx3+/Pax6+zones continues (Fig. 6C′). Although some explants exhibit thickening(Fig. 6C,C′), most contain cell aggregates with no particular structure. However, prospective nasal placode explanted at the cusp of thickening (HH13-14; n=25)forms a thickened epithelium and undergoes significant neurogenesis(Fig. 6D,E). Cumulatively,these results suggest that nasal placode precursors are specified to express Dlx3 and Pax6 as early as HH8; however, neuronal specification, as assayed by Hu expression, is noted only once nasal precursors have completely segregated from neighboring ectodermal populations and is complete just prior to acquiring placodal identity.

Operationally, a tissue is considered intrinsically committed to its fate if it recapitulates that fate when challenged by transplantation to a novel embryonic environment. As a rigorous test to determine irreversible commitment of olfactory placode ectoderm, its developmental potential was challenged by grafting it in place of non-placode forming trunk ectoderm. To this end,presumptive olfactory placode ectoderm derived from HH8-16 was transplanted to lateral plate ectoderm at the level of the most recently formed somites in HH8-9 chicks (Fig. 7A-C′). Even at HH8, grafts were committed to express Pax6 and Dlx3, paralleling their specification status(Fig. 7D). However, only a single graft from this stage expressed Hu, suggesting that neurogenesis was not yet fixed in most cases. At HH10, close apposition of the presumptive olfactory placode ectoderm with the forebrain makes their separation difficult. However, we cleanly dissected placodal ectoderm away from the forebrain in a few cases. The results show that, by HH10, olfactory precursors are not only capable of expressing Pax6/Dlx3 when grafted to the trunk, but a few graft-derived cells even differentiate as Hu+ neurons(Fig. 7E). We next heterotopically grafted older ectoderm (HH13) just prior to its thickening to form a morphologically distinct olfactory placode. In all cases, adherent mesenchyme was grafted with the presumptive placode ectoderm to the ectoderm overlying the lateral plate mesoderm. In two out of seven cases, the presumptive placodal ectoderm (identified by both Pax6 and Dlx3 expression)underwent thickening to acquire morphology characteristic of the olfactory placode and neuronal differentiation as assayed by Hu staining(Fig. 7F,F′). Isolated neurons appeared to migrate towards the dorsal aorta(Fig. 7F,F′, arrowheads). In the remaining five out of seven cases, no donor-derived neurons were detected. In these cases, although some of the grafted quail cells looked ectodermal and are Dlx3+/Pax6+, the majority of grafted cells appeared mesenchymal.

When placodal ectoderm is grafted (HH14-15; n=4), the ectoderm maintains its morphology, expresses Dlx3 and Pax6, and shows robust neuronal differentiation (Fig. 7G-G″). Interestingly, neurons that ingress from placodal ectoderm at this stage form a coherent group unlike grafts taken from younger donors (compare Fig. 7F,F′,G′,G″). Furthermore, graft-derived neurons are never found migrating through the Pax7+ mesenchyme, consistent with the endogenous situation where Hu+ cells migrate out of the olfactory epithelium in close association with each other and skirt the Pax7+ mesenchyme.

Cumulatively, our data suggest that once olfactory precursors segregate away from other ectodermal and placodal precursors, they are committed to express Dlx3 and Pax6 but have not yet received all signals for neuronal differentiation. Shortly before acquiring placodal morphology, transplanted olfactory ectoderm can express olfactory markers and differentiate; however,the neurons that ingress are not cohesive. Once the olfactory placode has acquired its characteristic morphology, differentiation occurs normally.

Olfactory progenitors arise within the anterior neural folds and adjacent ectoderm where they intermingle with lens precursors(Bhattacharyya et al., 2004). We have previously demonstrated that olfactory induction requires inhibition of lens fate by FGF at neurula stages and subsequently by signal(s) from the neural crest (Bailey et al.,2006). Here, we explore the spatiotemporal framework within which olfactory progenitors are precisely and reproducibly generated from an apparently homogenous field of anterior ectoderm. To this end, we define intrinsic properties of cranial ectoderm precursors that result in eventual irreversible acquisition of olfactory fate by a subset of these cells.

Comparing fate maps and gene expression patterns enables identification of molecular markers that foreshadow olfactory fate. Using this approach, we identified three transcription factors, Dlx3, Dlx5 and Pax6, the combinatorial expression of which uniquely demarcates progenitors of the chick olfactory epithelium. Current literature suggests significant species-specific differences in the co-expression of Dlx3, Dlx5 and Pax6 in the olfactory epithelium. For example, Dlx3 expression in the mouse is undetectable at E8.5 and only weakly detectable by E9.5 (Quint et al., 2000) or at E14.5(Merlo et al., 2007), but appears to be conserved between chick and zebrafish(Akimenko et al., 1994; Ellies et al., 1997). By contrast, the expression pattern of Dlx5 is consistent between chick, mouse and frog (Papalopulu and Kintner,1993; Pera et al.,1999; Yang et al.,1998); interestingly, Dlx5 expression is not observed in the gastrula stage zebrafish embryo and is detected only when olfactory precursors coalesce to form the placode (Akimenko et al., 1994; Ellies et al.,1997; Quint et al.,2000). Pax6 is expressed in the mouse(Grindley et al., 1995; Walther and Gruss, 1991),zebrafish (Krauss et al.,1991; Nornes et al.,1998; Puschel et al.,1992) and frog developing olfactory placodes(Franco et al., 2001),although a detailed timeline of its placodal expression is missing in the zebrafish.

These transcription factors not only serve as markers for developing nasal placodes, but also have functional importance. We show that some ingressing olfactory placode cells are Dlx+ and can become GnRH neurons. This is similar to the situation in mouse, where ∼70% of the Dlx5+ cells become GnRH+neurons that line the olfactory nerve(Merlo et al., 2007). In Dlx5 knockout mice, GnRH neuroendocrine cells persist but are mislocalized and express GnRH at reduced levels (Merlo et al., 2007). Dlx5-null mice have greatly reduced olfactory epithelia, with few sensory neurons therein, and rudimentary vomeronasal organs in addition to grossly impaired otic vesicle derivatives(Depew et al., 1999; Long et al., 2003; Merlo et al., 2007). Similarly, nasal placodes and cavities are absent in small eye(Sey) mice and rats lacking Pax6(Fujiwara et al., 1994; Grindley et al., 1995; Hogan et al., 1986; Matsuo et al., 1993). Chimeric analysis shows a cell-autonomous requirement for Pax6 for full contribution to the lens and nasal epithelium (Collinson et al., 2000; Collinson et al., 2003); Pax6 appears to be required for the transition from nasal ectoderm to placode (Grindley et al., 1995). Likewise, in the zebrafish deletion mutant, b380,which lacks Dlx3a and Dlx3b, there is no induction of otic or olfactory placodes (Solomon and Fritz,2002).

Evidence from mouse and chick supports an olfactory placode origin for basal forebrain localized GnRH neurons(Wray, 2002). In both organisms, detailed immunocytochemical analysis revealed GnRH+ cells in the olfactory epithelium, later along the olfactory nerve and finally in the forebrain (Murakami et al.,1991; Norgren and Lehman,1991; Sullivan and Silverman,1993; Wray et al.,1989a; Wray et al.,1989b). Furthermore, birth-dating forebrain GnRH neurons suggests they leave the mitotic cycle shortly before GnRH expression is seen in the olfactory epithelium (Mulrenin et al.,1999; Schwanzel-Fukuda and Pfaff, 1989; Wray et al.,1989a). More direct evidence for an olfactory placode origin for these neurons have come from dye labeling(Murakami and Arai, 1994),transplantation (Yamamoto et al.,1996) and ablation studies(Akutsu et al., 1992; Norgren and Gao, 1994). However, it has been argued that cells with non-olfactory placode origins could have mixed in with olfactory placode cells in these studies(Whitlock, 2008). Additional evidence in support of an olfactory placode origin has come from analysis of the Sey mutant, which lacks olfactory placodes and has no GnRH neurons (Dellovade et al.,1998; Skynner et al.,1999). Contrary to these studies, unilateral ablation of ectoderm immediately adjacent to, but not including, the olfactory placode at an early stage resulted in loss of forebrain-localized GnRH neurons(el Amraoui and Dubois,1993).

Advanced teleost fishes (atherinomorpha, percomorpha), including zebrafish,have two distinct types of forebrain GnRH neurons: terminal nerve ganglia GnRH neurons and hypothalamic GnRH neurons(Parhar, 2002; Whitlock et al., 2003). Of these, the terminal nerve ganglia GnRH neurons were considered to originate from olfactory placode (von Bartheld and Baker, 2004). However, recent dye labeling and lineage analysis in zebrafish suggest that these neurons are neural crest derived(Whitlock et al., 2003). Furthermore, a severe reduction in the number of hypothalamic GnRH neurons was observed in the you-too mutant, which lacks most of the pituitary but has normal olfactory placodes (Whitlock et al., 2003). Our data are consistent with the hypothesis that at least some GnRH neurons in the forebrain region are olfactory placode ectoderm derived. However, we cannot rule out the possibility that other GnRH neurons might originate from embryonic populations such as the adenohypophyseal placode or the neural crest.

Induction of structures at predetermined locations can occur by restricted localization of inducing signals and/or strict delimitation of responding tissue. For trigeminal placodes, dorsal neural tube at all axial levels have inducing ability, but competence to respond is relegated to cranial ectoderm(Baker et al., 1999). The converse occurs for otic placodes where inducing signals are localized at r2-7 while competence to respond is broad(Groves and Bronner-Fraser,2000). Work from many laboratories suggests that growth-factors such as Fgf8, Bmp4, Shh and retinoic acid are crucial for induction of the olfactory placode (Bailey et al.,2006; Bhasin et al.,2003; Calof et al.,2002; Kawauchi et al.,2005; LaMantia et al.,2000). Accordingly, competence to co-express Pax6/Dlx3 and differentiate as olfactory cells is restricted and decreases steadily in an anterior-posterior gradient, with trunk ectoderm possessing negligible capacity to contribute to the nasal placode. Additionally, we find good correlation between the loss in competence to express olfactory markers and commitment to differentiate as other derivatives. For example, ability to express Pax6 is lost in grafts of HH10+ hindbrain level ectoderm as it is committed to an otic placode fate (Groves and Bronner-Fraser, 2000).

In comparing chick trigeminal, otic and olfactory placode competence(Baker et al., 1999; Groves and Bronner-Fraser,2000) (and this study), trunk ectoderm has either no capacity to form placodes or only briefly retains this ability. Consistent with this,cranial ectoderm expresses epibranchial placode markers in response to BMP7 from pharyngeal endoderm but trunk ectoderm does not do so(Begbie et al., 1999). Competence to form lens is also restricted to lateral and ventral head ectoderm in amphibians and aves (Jacobson,1966; Sullivan et al.,2004). In Amblystoma punctatum, as in chick, competence to form olfactory placode extends, at early stages, over a wide area of head ectoderm, but not to anterior flank ectoderm after stage 15(Haggis, 1956). Furthermore,misexpression of signaling molecules or transcription factors involved in placodal development usually results in a partial recapitulation of the placode developmental program adjacent to the endogenous placodal territory or placodes forming in conjunction with neural tissue(Altmann et al., 1997; Barembaum and Bronner-Fraser,2007; Begbie et al.,1999; Köster et al.,2000; Ladher et al.,2000; Taylor and Labonne,2005; Vendrell et al.,2000). No experimental manipulation has so far led to the de novo generation of sensory or neurogenic placodes at all axial levels in the trunk ectoderm.

Recently, Martin and Groves (Martin and Groves, 2006) suggested a two-step model of placode induction wherein cranial ectoderm must first acquire generic properties of the pre-placodal region before responding to more specific placode-inducing signals. Wnt signals from lateral/posterior mesoderm and within trunk ectoderm have been hypothesized to restrict placode formation to a horseshoe-shaped domain in cranial ectoderm; however, Wnt inhibition alone cannot induce ectopic pre-placodal markers (Litsiou et al., 2005). Instead, intermediate levels of Bmp and Wnt inhibition in conjunction with Fgf signaling elicit pre-placodal markers such as Six1 and Six4 (Ahrens and Schlosser,2005; Brugmann et al.,2004; Litsiou et al.,2005). Thus, signaling pathways can prime cranial but not trunk ectoderm to respond to placode-inducing cues.

The capacity of presumptive olfactory ectoderm to express Pax6/Dlx3 and form neurons was assayed in 3D collagen gels. We find that presumptive olfactory placode ectoderm is specified to express Pax6 and Dlx3 by HH8(Bailey et al., 2006) (this paper). Neuronal specification begins around HH10-11, implying that ectoderm has seen signals that will direct its fate even before it is morphologically visible as a placode, analogous to trigeminal, otic and lens placode specification (Baker et al.,1999; Groves and Bronner-Fraser, 2000; Sullivan et al., 2004).

We also determined the time when presumptive olfactory placode ectoderm is irreversibly committed by challenging its fate in a new environment via grafting over lateral plate ectoderm of HH8-9 chick embryos. Distant from the endogenous placode, this site probably lacks inducing signals. Interestingly,commitment to express Pax6 and Dlx3 and form some neurons occurred by HH10,before the placode is self-evident, consistent with early experiments showing that grafts of the anterior tip of HH11 embryos to the chorioallantoic membrane or trunk produced relatively normal olfactory organs in a small percentage of grafts (Street,1937). However, sustained olfactory development only occurred once the placode acquired its characteristic morphology(Street, 1937). In Amblystoma punctatum, there is some discrepancy regarding when olfactory determination occurs. Although one study corroborates our findings(Haggis, 1956), another observes commitment to an olfactory fate at neurula stages(Carpenter, 1937). One possibility is that neural tissue was present in grafts of olfactory precursors taken at neurula stages, which would then continue to provide signals necessary for growth and differentiation. Another possibility suggests that the site of transplantation might have been in range of olfactory-inducing stimuli (Haggis,1956).

We show that co-expression of Dlx3, Dlx5 and Pax6 demarcates olfactory precursors prior to acquiring placodal status. Competence to form olfactory placode resides within head ectoderm up until HH9-10, but is largely absent from trunk ectoderm. Olfactory fate specification and commitment are complete just before acquiring overt placodal morphology and prior to differentiation into olfactory epithelium. This may highlight a general trend in placodal development.

We thank Drs Andy Groves and Michael Kessel for cDNA clones, Dr Jhumku Kohtz for the pan-Dlx antibody, Dr James Millam for the chick GnRH-I antibody,Dr Andy Kowalczyk for use of his microscope, and Matt Jones, David Arce and Andrea Manzo for technical assistance. We are grateful to Dr Tatjana Sauka-Spengler for critical reading of the manuscript and to Dr Andrea Streit for many helpful discussions. This work was supported by DE16459 to M.B.-F. and partly by a HHMI Predoctoral Fellowship to S.B.