Differential and overlapping expression patterns of X-dll3 and Pax-6 genes suggest distinct roles in olfactory system development of the African clawed frog Xenopus laevis

The African clawed frog Xenopus laevis belongs to the Pipidae family of frogs, all of which have an aquatic lifestyle in adulthood. They rarely leave the water and have several adaptations to aquatic life including loss of the tongue (Sokol, 1969) and the presence of lateral line organs (Russel, 1976). Furthermore, in X. laevis, the olfactory system has adapted to accommodate this primarily aquatic lifestyle. Together with a nasal cavity that is exposed to air, the main olfactory system of the adult has an additional cavity that is exposed to water (Altner, 1962; Venus et al., 1998). The formation of this new adult nasal cavity occurs at metamorphosis concomitant with the remodeling of the larval nasal cavity (Föske, 1934; Paterson, 1939; Paterson, 1951).

The formation of the olfactory epithelium involves embryonic, larval and metamorphic phases. The olfactory epithelium develops during embryogenesis from a thickening of the anterior ectoderm that comprises neuronal and non-neuronal ectodermal cells. This thickening forms the olfactory placode that will give rise to all the cells forming the olfactory epithelium: basal cells, olfactory receptor neurons and supporting cells (Klein and Graziadei, 1983). The olfactory epithelium of the larval principal cavity (PC; medial diverticulum) is exposed to water and is thought to respond only to water-borne odorants throughout larval life (Elepfandt, 1996; Venus et al., 1998). At metamorphosis, the PC undergoes major cellular and molecular transformations (Föske, 1934; Reiss and Burd, 1997a) apparently in preparation for sensing air-borne odorants. Also at metamorphosis, the middle cavity (MC; lateral diverticulum) develops de novo, is exposed to water (Venus et al., 1998) and is likely to sense water-borne odorants (Elepfandt, 1996; Venus et al., 1998). The functional segregation, however, is not as well defined as the anatomical one, as it appears that olfactory receptor neurons from the MC are able to respond to both water-borne and air-borne odorants (Iida and Kashiwayanagi, 1999).

The olfactory epithelium of the adult MC displays similar cellular and biochemical characteristics to those of the olfactory epithelium of the larval PC, which suggests a conserved functional role for these two cavities. At the cellular level, both epithelia contain ciliated and microvillar olfactory receptor neurons whose axons project to the ventral olfactory bulb (Reiss and Burd, 1997b; Hansen et al., 1998). At the biochemical level, both epithelia display identical expression patterns for the lectin soybean agglutinin (SBA) (Key and Giorgi, 1986; Hofmann and Meyer, 1991; Reiss and Burd, 1997b) and the monoclonal antibody E7 (Petti et al., 1999). X. laevis possess a gene repertoire encoding two classes of olfactory receptor gene; one class is related to the olfactory receptor genes of fish and is called Class I and the other class is related to the olfactory receptor genes of mammals and is called Class II (Freitag et al., 1995). Finally, at the molecular level, both epithelia express the fish-like Class I olfactory receptor genes (Freitag et al., 1995). In contrast, the olfactory epithelium of the adult PC has only ciliated olfactory receptor neurons whose axons project to the dorsal bulb (Fritz et al., 1996; Reiss and Burd, 1997b; Hansen et al., 1998), does not stain with either SBA or E7 (Key and Giorgi, 1986; Hofmann and Meyer, 1991; Petti et al., 1999) and expresses the mammalian-like Class II olfactory receptor genes (Freitag et al., 1995). The cellular, biochemical and molecular differences between olfactory epithelia that are exposed to water or to air are likely to be the result of differential expression of developmental control genes. We therefore hypothesized that differences in expression of developmental control genes exist between the formation of the larval PC and adult MC and the remodeling of the adult PC.

The Distal-less and Pax-6 genes are homeobox genes whose differential expression has been shown to mediate patterning, morphogenesis and histogenesis in vertebrate development (Callaerts et al., 1997; Bendall and Abate-Shen, 2000). In vertebrates, there are several members of the Distal-less gene family. Unfortunately, the nomenclature for vertebrate Distal-less genes is confusing. Distal-less genes are expressed in the olfactory placodes of zebrafish, chick, mouse and Xenopus laevis but different numerical names have been used for homologous genes. For example, the homologous Distal-less genes expressed in the nasal tissues are zdfdlx4 in zebrafish (Akimenko et al., 1994), Dlx3 in chick (Pera and Kessel, 1999) and Dlx5 in mouse (Simeone et al., 1994). Expression of Dlx5 is required for the development of the olfactory epithelium, as demonstrated by the phenotypes of homozygous Dlx5 mouse mutants lacking exons I and II (Depew et al., 1999; Acampora et al., 1999). These mice display a reduction or absence of olfactory epithelium. Since the construct used by Acampora et al. (Acampora et al., 1999) lacked the start codon, Dlx5 protein synthesis should be abolished in these mutant mice. Similarly, Pax-6 is expressed in the olfactory placode and developing olfactory bulb of human, mouse, chick, quail and zebrafish (Ton et al., 1991; Walter and Gruss, 1991; Goulding et al., 1993; Martin et al., 1992; Puschel et al., 1992). In humans and mice, homozygotic mutants lacking the Pax-6 gene fail to form olfactory tissues, resulting in the absence of both olfactory epithelium and olfactory bulb (Glaser et al., 1994; Grindley et al., 1995).

In X. laevis, both Distal-less and Pax-6 are expressed in the developing olfactory system. X-dll3, the orthologue of the mouse Dlx5, and Pax-6 are both expressed early during embryogenesis in the anterior ectoderm, tissue that gives rise to the prospective olfactory placodes (Papalopulu and Kintner, 1993; Swiergiel et al., 1994; Burd et al., 1995; Hirsch and Harris, 1996). Since X-dll3 and Pax-6 are in a position to specify the early olfactory placode in X. laevis embryos, we predicted that these two genes might also be involved in remodeling the olfactory epithelium at metamorphosis. Using a combination of in situ hybridization and immunocytochemical procedures, we show that X-dll3 is expressed in all regions of olfactory epithelium throughout embryogenesis and metamorphosis, whereas Pax-6 is expressed only in a subset of structures at specific stages in the embryonic olfactory placode, larval PC and vomeronasal organ and adult MC. This differential expression pattern suggests that X-dll3 is likely to be involved in the patterning of olfactory epithelia exposed to both air and water, whereas Pax-6 involvement seems to be restricted to the formation of olfactory epithelia exposed to water.

Xenopus laevis (Daudin) embryos were obtained from gonadotropin-induced breedings (HCG; Sigma), and tadpoles and adult frogs were purchased from NASCO (Fort Atkinson, WI, USA). Animals were reared at 22°C on a 12h:12h L:D cycle in 6l polypropylene tanks containing X. laevis rearing solution (Burd, 1991). Animals were given daily feedings of either boiled nettle (tadpoles; Wunderlich-Diez Corp., Hasbrouck Heights, NJ, USA) or frog brittle (frogs; NASCO). Animals were staged according to established developmental criteria (Nieuwkoop and Faber, 1994).

Digoxigenin-labeled RNA probes were synthesized from linearized DNA templates using either T3 or T7 RNA polymerase. A partial-length X-dll3 clone containing base pairs +1 to +753 was obtained from Dr Nancy Papalopulu (Papalopulu and Kintner, 1993). The X-dll3 antisense probe was obtained by digestion with EcoRI and subsequent transcription with T7 RNA polymerase in the presence of digoxigenin-labeled UTP following recommended protocols (Boehringer Mannheim); the X-dll3 sense probe was obtained as above using BamHI and T3 RNA polymerase. For Pax-6 probes, a partial-length Pax-6 clone was isolated from a X. laevis cDNA library provided by Dr Richard Harland (Hemmati-Brivanlou et al., 1991). The probe used to screen the library was a radiolabeled 1035basepair EcoRI–NdeI restriction fragment of the zebrafish Pax-6 cDNA clone pn108 (Puschel et al., 1992). The Pax-6 antisense probe was obtained by digestion with BstXI and subsequent transcription with T7 RNA polymerase in the presence of digoxigenin-labeled UTP following recommended protocols (Boehringer Mannheim); the Pax-6 sense control probe was obtained as above using EcoRV and T3 RNA polymerase. The nucleotide identity of the probes was confirmed by sequencing the clones used to generate the probes. All probes were hydrolyzed for 10–15min in 40mmoll^-1 sodium bicarbonate/60mmoll^-1 sodium carbonate at 60°C.

Whole-mount in situ hybridization using digoxigenin-labeled RNA probes was performed as described by Harland (Harland, 1991). Briefly, embryos were fixed in MEMFA (0.1moll^-1 MOPS, pH7.4, 2mmoll^-1 EGTA, 1mmoll^-1 MgSO₄, 3.7% formaldehyde) for 1–2h. For storage, embryos were kept at –20°C in 100% methanol and were then rehydrated by washes in 75% ethanol, 50% ethanol and 25% ethanol in PBS-T (0.1moll^-1 phosphate-buffered saline with 0.1% Tween 20) for immediate use. Embryos were then incubated for 10–15min in 10mgml^-1 proteinase K (Fisher), washed in 0.1moll^-1 triethanolamine (TEA, pH7–8) and incubated in 0.1moll^-1 TEA, 0.25% acetic anhydride. Embryos were then washed in PBS-T and prehybridized for at least 6h at 60°C in 50% formamide, 5× SSC (1× SSC is 0.15moll^-1 sodium chloride, 0.015moll^-1 sodium citrate), 1mgml^-1 torula RNA, 100mgml^-1 heparin, 1× Denhardt’s solution (0.02% Ficoll 400, 0.02% polyvinyl pyrrolidone, 0.02% bovine serum albumin), 0.1% Tween 20, 0.1% CHAPS, 5mmoll^-1 EDTA. Hybridization was carried out at 60°C overnight in prehybridization solution containing 0.5–1mgml^-1 digoxigenin-labeled probes. Following hybridization, embryos were washed at 60°C in prehybridization solution, rinsed at 60°C in 2× SSC, treated at 37°C in 20mgml^-1 RNAse A and 10mgml^-1 RNAse T1 in 2× SSC, and rinsed at 60°C in 0.2× SSC. Embryos were then rinsed in maleic acid buffer (MAB; 100mmoll^-1 maleic acid, 150mmoll^-1 NaCl, pH7.5), blocked in MAB containing 20% heat-inactivated lamb serum and 2% blocking reagent (Boehringer Mannheim) at room temperature (24°C) for 2h, and incubated overnight in the same solution containing a sheep alkaline-phosphatase-conjugated, anti-digoxigenin antibody (1:2000 dilution of the Fab fragment; Boehringer Mannheim).

For the colorimetric detection, embryos were rinsed in MAB and washed three times in alkaline phosphatase (AP) buffer (100mmoll^-1 Tris-HCl, pH9.5, 50mmoll^-1 MgCl₂, 100mmoll^-1 NaCl, 0.1% Tween 20 and 5mmoll^-1 levamisol). The colorimetric reaction was performed using 0.34mgml^-1 Nitro Blue Tetrazolium (NTB; Sigma) and 0.18mgml^-1 5-bromo-4-chloro-3-indolyl-phosphate (BCIP; Sigma). The staining was stopped by washing the embryos several times in AP buffer. Embryos were then refixed in MEMFA and stored in methanol at −20°C. Stained animals were photographed, and selected embryos were embedded in paraffin, sectioned at 10mm, cleared in xylene, coverslipped and photographed with a Nikon Biophoto microscope.

In situ hybridization on tissue sections was performed as described by Strahle et al. (Strahle et al., 1994). Animals were anesthetized using 0.02% 3-aminobenzoic acid ethyl ester (MS222, Sigma), fixed by perfusion with 4% paraformaldehyde, 4% (w/v) sucrose in PB buffer (0.1moll^-1 phosphate buffer, pH7.4) and stored in the fixative solution overnight at 4°C. After a series of rinses in fixative solution minus paraformaldehyde, tissues were embedded in 4% agarose and soaked overnight at 4°C in PB containing 30% (w/v) sucrose. Alternatively, tissues were frozen in Tissue-Tek (OCT compound, Miles Inc., Elkhart, IN, USA) for immediate sectioning. The tissues were sectioned at 14mm using a cryostat (Microm, Heidelberg) and collected on either TEPSA- (Sigma) or Vectabond-coated (Vector Labs) slides. Sections were air-dried at room temperature for at least 1h and stored at −20°C for subsequent analysis. Sections were directly hybridized overnight at 58°C with 0.5–1mgml^-1 RNA probe in 50% formamide, 0.3moll^-1 NaCl, 10mmoll^-1 NaH₂PO₄, 10mmoll^-1 EDTA, 10mmoll^-1 Tris-HCl, pH7.5, 10% dextran sulfate, 1mgml^-1 rRNA, 1× Denhardt’s solution. Slides were coverslipped and placed into a sealed box containing Whatman filter paper soaked with the hybridization solution minus the RNA probe. The box was then placed in an incubator overnight at 58°C. The coverslips were then removed, and the sections were washed first at 65°C in 25% formamide, 1× SSC, 0.5× PBS and then at room temperature in PBS. Sections were then blocked in PBS-T, 0.2% (w/v) bovine serum albumin, 2% heat-inactivated lamb serum for 2h at room temperature, incubated in alkaline-phosphatase-conjugated anti-digoxigenin antibody, rinsed and reacted with the colorimetric solutions as described above for the whole-mount procedure except that, for some reactions, the BM Purple AP Substrate (Boehringer Mannheim) replaced the NTB/BCIP solution. The staining was analyzed with a Nikon Biophoto microscope.

Localization of the Distal-less protein was carried out using indirect immunofluorescence. The cross-reacting polyclonal antibody against the Distal-less protein was generated against the butterfly Precis coenia and subsequently affinity-purified. This antibody was provided by Dr G. Panganiban (University of Wisconsin, Madison, WI, USA). Tissue sections were prepared as described above for in situ hybridization on frozen sections. Sections were treated for 30min at room temperature in blocking solution (3% normal goat serum, 0.4% Triton X-100) in PB (0.1moll^-1 sodium phosphate buffer, pH7.4). Sections were then incubated for 2h at room temperature in blocking solution containing the polyclonal antibody against Distal-less (0.45mgml^-1 final concentration). After three washes in PB, the sections were incubated for 1h at room temperature in PB containing the Cy3-conjugated goat anti-rabbit antibody (1:500 dilution; Jackson ImmunoResearch Laboratories). In all experiments, sections were counterstained for 15min at room temperature with the nuclear stain 4′-6-diamidino-2-phenylindole (DAPI; Sigma). The slides were then coverslipped using AquaPolymount (Polysciences Inc). The fluorescence was analyzed under a Zeiss Universal epifluorescence microscope. Negative controls consisted of staining in the absence of the primary antibody. The positive control for Distal-less staining was the demonstration of staining in tissues known to contain Distal-less (e.g. larval limb tissues); the positive control for Pax-6 staining was the demonstration of staining in the olfactory bulb with our antibody and a Pax-6 antibody generated by Davis and Reed (Davis and Reed, 1996).

Animals and tissues were photographed onto transparency films. Films and images were digitized and figures were processed using Adobe Photoshop (Adobe Systems Incorporated).

The olfactory placodes can be fate-mapped from cells located in the anterior lateral portion of the anterior neural ridge, a strip of tissue that lies anterior to the open neural plate at stages 14–16 (Eagleson and Harris, 1989; G. D. Burd, A. Collazo and S. E. Fraser, unpublished observations) (Fig.1A). When the neural tube closes, the anterior neural ridge folds back over the anterior neural tube; as the right and left sides of the anterior ectoderm fuse over the anterior neural tube at stage 20, the sense plate is formed (Burd, 1999) (Fig.1A). At stage 23, the olfactory placodes form from bilateral thickenings in the sense plate (Nieuwkoop and Faber, 1994; Klein and Graziadei, 1983). The olfactory placode remains a flattened thickening of tissue until stage 31, when the nasal pit begins to form (Nieuwkoop and Faber, 1994) (Fig.1A). At stage 37/38, the nasal pit begins to differentiate into the principal cavity (PC) and the vomeronasal organ (VNO) (Nieuwkoop and Faber, 1994; Burd, 1999). At stage 52, the onset of metamorphosis, the middle cavity (MC) forms de novo (Burd, 1999). After metamorphosis and in the adult, the nasal cavity contains the PC, MC and VNO (Fig.1B).

The tissue fated to give rise to olfactory placodes (Eagleson and Harris, 1989; Burd et al., 1995) expresses X-dll3 by stage 15, the open neural plate stage of development (Papalopulu and Kintner, 1993) (Fig.2A). The expression pattern continues in pre-olfactory placode tissue during closure of the neural tube at stages 18–19 (Fig.2B). X-dll3 is widely expressed in the sense plate between stages 20 and 25 before and during the initial formation of the olfactory placodes (Fig.2C,J), but staining is more intense in the developing olfactory placodes. Ultimately, X-dll3 staining is lost from the sense plate by stage 35, but remains in the olfactory placodes and developing olfactory epithelium (Fig.2D,L). In tissue sections, it is clear that X-dll3 expression is present in both the non-neuronal and neuronal layers of the sense plate at stage 26 (Fig.2J). In stage 35 animals, staining for X-dll3 expression is present throughout all cell types of the olfactory placode (Fig.2L). Cells fated to give rise to the olfactory bulb in the anterior neural tube do not express X-dll3 at stage 26 (Fig.2J), but begin to express X-dll3 by stage 27, at a time before olfactory axons contact the neural tube (Byrd and Burd, 1991). In embryos, staining for X-dll3 occurs throughout the neural tube cells that form the olfactory bulb (Fig.2L), but becomes restricted to a subset of bulb cells in the larval stages (see below).

Pax-6 expression is first seen at the small yolk plug stage (stage 12) as patches of stain dorsal to and flanking the receding yolk plug (data not shown). This region corresponds to the posterior sensory layer of the neuroectoderm (Nieuwkoop and Faber, 1994). At the open neural plate stage (stage 15), Pax-6 is expressed as a broad stripe in the anterior neural ridge and anterior neural plate, tissue fated to give rise to the olfactory placodes, olfactory bulb, eyes and various forebrain tissues (Fig.2E). Two posterior stripes of stained cells are fated to give rise to the spinal cord and hindbrain (Fig.2E). The gap in staining in the neural tube corresponds to the presumptive midbrain (Eagleson and Harris, 1989). Pax-6 expression is observed primarily in the most anterior sense plate cells at the neural tube stage (stage 21) (Fig.2F), while the presumptive olfactory placodes are fated to form from all regions of the sense plate at this stage (G. D. Burd, A. Collazo and S. E. Fraser, unpublished observations). Pax-6 expression is very light during the early stages of olfactory placode development (Fig.2G,K). Pax-6 expression increases in the olfactory placodes by stage 39 but remains light (Fig.2H). Throughout the early embryonic stages, precursor cells that are fated to become olfactory bulb cells stain for Pax-6 (Fig.2K). However, by stage 37, Pax-6 expression in the precursor cells around the edge of the ventricles is reduced, and most of the Pax-6 expression in the developing olfactory bulb is in differentiating neurons (Fig.2M).

Metamorphosis of the olfactory epithelium includes the remodeling of the larval PC and the de novo formation of the adult MC. The remodeling of the larval PC appears to start at stage 49 with the expression of the genes encoding the mammalian-like Class II odorant receptors (Mezler et al., 1999). The formation of the adult MC begins at stage 51 as a small bud at the anterior end of the larval PC (Föske, 1934; Petti et al., 1999).

Analysis of the spatio-temporal pattern of X-dll3 expression during early larval development shows a similar pattern to that seen at embryonic stages 35 and 37 (Fig.2D,L). The PC in young larvae expresses X-dll3 throughout the cavity including the supporting, neuronal and basal cell layers (data not shown). The onset of metamorphosis (stage 51) marks the beginning of a series of changes in the pattern of expression of X-dll3. At stage 51, X-dll3 expression is lost from the supporting cell layer of the PC, as shown by the absence of purple staining from the apical-most layer of the olfactory epithelium (Fig.3B). X-dll3 expression in the PC remains the same until premetamorphosis (stage 54) (Fig.3E). By the onset of metamorphic climax (stage 58), X-dll3 expression has become restricted to the deepest cell layer of the PC that contains basal cells and some olfactory receptor neurons (Petti et al., 1999) (Fig.3H). By the end of metamorphosis (stage 64), only clusters of cells close to the basal lamina of the PC express X-dll3 (Fig.3K). These deep clusters are likely to be basal cells and a subpopulation of olfactory receptor neurons (Petti et al., 1999). The expression pattern of X-dll3 within the developing MC is similar to that of the remodeling PC during metamorphosis; however, the expression is delayed temporally, as illustrated in Fig.3D,G,J. At stage 54, X-dll3 is expressed primarily in the deeper layers of the MC epithelium (Fig.3D), but by stage 59 X-dll3 expression is restricted to the neuronal and basal cell layers, as illustrated by the absence of purple staining from the supporting cell layer of the MC (Fig.3G). By the end of metamorphosis, X-dll3 expression becomes lighter and but remains in the deeper layer of the MC (Fig.3J). In addition, we analyzed X-dll3 expression within the developing olfactory bulb. X-dll3 is expressed mainly around the lateral ventricles and in the granular cell layer of the olfactory bulb, and the pattern of expression does not change during metamorphosis (Fig.3C,F,I,L).

The potential functional role for Distal-less in the development of the olfactory epithelium, as revealed by the expression pattern of X-dll3 mRNA, was supported by the finding that Distal-less protein has a similar expression pattern. Using immunocytochemistry, we show that Distal-less protein is expressed throughout the MC and in the neuronal and basal cell layers of the PC at stage 53 (Fig.4A,B). At the adult stage, Distal-less is expressed primarily in the neuronal and basal cell layers of both the MC and PC, and the overall pattern of expression has moved towards the basal lamina of both cavities (Fig.4C–F).

The pattern of expression of Pax-6 during early larval development and metamorphosis was analyzed using in situ hybridization. Analysis of Pax-6 mRNA distribution during early larval development shows that Pax-6 expression decreases in the larval PC around stage 44 and disappears from the PC by stage 49 (Fig.5B). Interestingly, the ability of the olfactory placode to regenerate when removed is also lost at approximately stage 44 (Stout and Graziadei, 1980; Byrd and Burd, 1993). Pax-6 expression remains absent from the PC throughout metamorphosis, as illustrated for stages 58, 61 and 66 (Fig.5D,F,H). In contrast, Pax-6 is expressed in the MC at the onset of its formation (data not shown) and during metamorphic climax (Fig.5A,E). In the MC, Pax-6 is expressed primarily in the neuronal and basal cell layers (Fig.5A,E), whereas in the VNO, Pax-6 is expressed only in the supporting cell layer (Fig.5A). In the developing forebrain, Pax-6 is expressed in the granular cells and in some mitral cells in the olfactory bulb throughout metamorphosis. Ventricular cells do not express Pax-6 during this same period (Fig.5C,G,I). Apparently, the amount of Pax-6 protein produced in the olfactory epithelium during metamorphosis is low since we were unable to demonstrate protein expression using immunocytochemistry with a Pax-6 antibody generated against the last 18 amino acid residues at the carboxy terminus; the peptide we used was identical to that used by Davis and Reed (Davis and Reed, 1996). Both our antibody and that of Davis and Reed gave the same pattern and intensity of staining in the olfactory bulb.

A direct comparison of the expression patterns of X-dll3 and Pax-6 during the embryonic, larval and metamorphic phases of olfactory system development is illustrated in Fig.6. Fig.6A shows that X-dll3 is expressed in all cells of the olfactory placode and in all cells of the young larval PC. As the larval PC remodels, X-dll3 expression becomes restricted to the deeper layers of the olfactory epithelium. By the end of metamorphosis, X-dll3 is expressed only in deep neurons and basal cells. A similar spatial pattern of expression is found in the MC; X-dll3 is expressed in all cells of the newly forming MC, and expression becomes restricted to deep neurons and basal cells at the end of metamorphosis. Fig.6B shows that Pax-6 is expressed in neurons and basal cells of larval PC and metamorphic MC. Pax-6 expression then completely disappears from the larval PC as functional remodeling begins at stage 49.

The olfactory system of the African clawed frog Xenopus laevis is a valuable model in which to study the pattern of expression of developmental control genes because there are two developmental stages. During early embryonic development, the body plan is established. Later, at metamorphosis, many of the same regions are remodeled and other structures are formed de novo. From the perspective of developmental biology, it is interesting to compare these two developmental stages to determine whether any of the molecular signals used to establish a particular tissue are reused to remodel that tissue at metamorphosis. Clearly, the major signal involved in the initiation of metamorphosis is thyroid hormone, which stimulates the adaptive body changes associated with the normal amphibian transition from aquatic to terrestrial life (White and Nicoll, 1981). Thyroid hormone exerts its effect on metamorphosis through its nuclear receptors, the thyroid hormone receptors (Evans, 1988). These receptors are thyroid-hormone-dependent transcription factors that induce metamorphosis by activating a cascade of changes in gene expression in most tissues of the larvae (Shi, 1996). While thyroid hormone is responsible for stimulating tissue changes at metamorphosis, it does not play a role in early embryonic development in frogs. Thus, we were interested in determining whether similar patterning genes might participate in the formation of specific olfactory tissues in the absence and presence of thyroid hormone.

The molecular mechanisms involved in embryonic patterning have been extensively studied in recent years and have been shown to involve at least four types of molecule: the signaling molecules, the growth factors, the cell adhesion molecules and the transcription factors. Examples of these four types of molecule have been shown to participate in the embryonic development of the olfactory system (see Reiss and Burd, 1997a). In contrast, little is known about the molecular mechanisms involved in post-embryonic patterning. In this report, we show that genes encoding the transcription factors X-dll3 and Pax-6 are expressed during both embryonic and post-embryonic development of the olfactory system. Subtle differences in their spatio-temporal expression patterns suggest that they play distinct roles in forming the adult peripheral olfactory system.

Several studies have suggested that X-dll3 orthologues are involved in the patterning of the olfactory system through expression in the developing olfactory placodes of X. laevis, mice, zebrafish and chicks (Papalopulu and Kintner, 1993; Akimenko et al., 1994; Simeone et al., 1994; Pera and Kessel, 1999). Recently, the generation of mice homozygous for a targeted deletion of the Dlx5 gene has confirmed the role of these genes in the development of olfactory placodes (Depew et al., 1999). In the most severe cases, mice with Dlx5 deletions have nearly complete aplasia of the nasal capsule, and in these mice the olfactory epithelium and vomeronasal organs are absent (Depew et al., 1999). During early development (E8.0–E8.5) in mice, Dlx5 is expressed in a thin stripe of cells at the neural/non-neural boundary of the neural plate that contains olfactory and otic placodes (Depew et al., 1999). At a similar developmental stage (stage 15), this thin stripe of cells, termed the neural ridge in X. laevis, displays a similar pattern of expression when stained with the X-dll3 orthologue. Our report supports the involvement of X-dll3 in the early patterning because we show that X-dll3 is expressed in cells fated to form the olfactory placodes as early as stage 15 (Eagleson and Harris, 1989). In addition, we show that X-dll3 is expressed in the MC when it starts to develop at stage 51, suggesting that X-dll3 is involved not only in embryonic patterning but also in the morphological changes that occur at metamorphosis. In both larval PC and MC, the spatio-temporal expression of X-dll3 is similar, suggesting a common role in the formation of these two cavities. These results also suggest that X-dll3 expression is not entirely under the control of thyroid hormone since X-dll3 is expressed in the absence (embryogenesis) and in the presence (metamorphosis) of thyroid hormone.

In vertebrates, the Distal-less genes have been correlated with specific functions throughout development (for a review, see Bendall and Abate-Shen, 2000), ranging from neural induction (Papalopulu and Kintner, 1993; Yang et al., 1993; Pera et al., 1999) to cell differentiation (Porteus et al., 1994; Liu et al., 1997). The pattern of X-dll3 expression in the peripheral olfactory system changes during development, suggesting that X-dll3 may have multiple functions in establishing the adult olfactory system. Indeed, during embryogenesis, X-dll3 is expressed in both supporting and neuronal cell layers, suggesting that X-dll3 has a role in patterning the epithelium.

However, during metamorphosis, X-dll3 expression is restricted to the deeper cell layers, suggesting that X-dll3 is involved in neurogenesis. In addition, the intensity of the expression decreases as the PC and MC cavities form, correlating with the decrease in the number of new neurons generated at these later stages. Specifically, X-dll3 expression in the PC becomes restricted to the olfactory receptor neurons and basal cells by mid-metamorphosis. Similarly, X-dll3 is expressed throughout the olfactory epithelium of the developing MC, and its expression later becomes restricted to olfactory receptor neurons and basal cells at metamorphic climax. We also show that the Distal-less protein is expressed in young olfactory receptor neurons and basal cells of both adult PC and MC. Taken together, these results suggest that X-dll3 is first involved in the patterning and genesis of all cells forming the peripheral olfactory tissue and is later involved in neuronal maturation and neuronal turnover in putative water- and air-sensing epithelia. Indeed, in most vertebrates, the development of the olfactory system requires initial patterning and continuous replacement of the olfactory neurons throughout adult life (Graziadei and Metcalf, 1971; Graziadei and Monti Graziadei, 1979).

Unlike X-dll3, which seems to be involved in the formation and maintenance of all nasal cavities, Pax-6 seems to have a narrower role in the development of the peripheral olfactory system of the frog X. laevis. Our analysis of the expression pattern at stage 15 suggests that Pax-6 is likely to be involved in early patterning of the olfactory system, but the level of expression during subsequent stages of early olfactory placode development is very low. In contrast, during these same stages of placode development, there is strong expression of X-dll3. In addition, while X-dll3 is expressed in the sensory epithelium of the PC, MC and VNO during and following metamorphosis, suggesting a role in development and maintenance, Pax-6 expression appears to be restricted to the formation of olfactory epithelia since it is expressed only during placodal stages in the PC, and in the developing MC. Since these two olfactory tissues are responsible for sensing odors in water, Pax-6 appears to be involved in forming water-sensing olfactory epithelia. Alternatively, Pax-6 could be involved in early and de novo non-remodeling events during the formation of the main olfactory epithelium, independent of water-sensing function.

The pattern of Pax-6 expression in the VNO also differs from the pattern of VNO X-dll3 expression. In addition, the patterns of expression of X-dll3 and Pax-6 differ in the olfactory bulb. X-dll3 expression is restricted to bulb precursor cells and neurons in the early stages of differentiation both in embryos and through metamorphosis, while the expression pattern for Pax-6 is primarily in differentiating and mature neurons during embryonic development, throughout metamorphosis and into adulthood. Thus, the spatial and temporal patterns of expression suggest different roles for these two transcription factors in the development of the olfactory epithelium and olfactory bulb.

Pax-6 expression in the olfactory epithelium differs between mice and X. laevis. In mice, Pax-6 protein is expressed only in basal and supporting cells of the main olfactory epithelium and in the proliferative zone of the VNO (Davis and Reed, 1996; Behrens et al., 2000), leading to the suggestion that Pax-6 is involved in the regulation of non-neuronal, lineage-specific genes (Davis and Reed, 1996). Our findings show that, in contrast, Pax-6 is expressed in the supporting cells of the VNO and in basal cells and olfactory receptor neurons of the larval PC and MC, the equivalent tissues to the mammalian olfactory epithelium. Thus, Pax-6 appears to play different roles in the developing olfactory epithelium of X. laevis and mice.

During the development of the olfactory system of X. laevis, there is an interesting spatio-temporal correlation between the patterns of expression of Pax-6 and at least one of the fish-like Class I odorant receptor genes. The expression of fish-like odorant receptor genes begins at stage 32 in the larval PC (Mezler et al., 1999). Although diminishing by stage 50, their expression remains in the PC (Mezler et al., 1999) until the onset of the cellular remodeling of the PC at stage 57 (Reiss and Burd, 1997a; Reiss and Burd, 1997b; Hansen et al., 1998). Although some fish-like olfactory receptor genes continue to be expressed in the PC until stage 57 (Mezler et al., 1999), the neurons expressing them are probably the neurons that are fated to die to allow a complete remodeling of the PC (Hansen et al., 1998; Higgs and Burd, 2001). At stage 49, overlapping with the loss of the fish-like Class I olfactory receptor genes in the PC is the onset of expression of the mammalian-like Class II odorant receptor genes in the larval PC (Mezler et al., 1999). The absence of Pax-6 expression during this period of remodeling in the PC suggests that Pax-6 is not involved in establishing the air-sensing olfactory epithelium of the PC. In contrast, both Pax-6 and the fish-like olfactory receptor genes (Freitag et al., 1995) are expressed in the MC, a cavity presumably specialized for sensing odorants in water (Venus et al., 1998). Taken together, these observations suggest that the fish-like receptor genes are co-expressed with Pax-6 in the presumed water-sensing epithelia and that Pax-6 expression never overlaps with expression of the mammalian-like receptor genes. This correlation in gene expression favors a role for Pax-6 in the specific formation of presumed water-sensing epithelium.

Thyroid hormone plays a causative role in amphibian metamorphosis and has been shown to have a major influence over the remodeling of the olfactory system in the frog X. laevis (Burd, 1990; Burd, 1992; Petti and Burd, 1995; Burd, 1999). Thyroid hormone function is mediated through the regulation of gene expression (Shi, 1996), and many thyroid-hormone-regulated genes have been identified during hind limb transformation (Buckbinder and Brown, 1992), intestine remodeling (Shi and Brown, 1993) and tail resorption (Wang and Brown, 1991; Wang and Brown, 1993; Brown et al., 1996). Another gene, the transcription factor NF-I, is regulated by thyroid hormone (Shi and Brown, 1993) and has been reported to participate in establishing olfactory neuron phenotype in the rat (Baumeister et al., 1999; Behrens et al., 2000). These results suggest that thyroid hormone is likely to be involved in regulating the transcription factors associated with olfactory system plasticity. Thus, it is likely that thyroid hormone is involved in regulating some of the changes in expression of Pax-6 and X-dll3 in the olfactory epithelium at metamorphosis. However, Pax-6 and X-dll3 cannot be regulated by thyroid hormone during embryogenesis since this hormone is not present until approximately stage 54, at the onset of metamorphosis (Leloup and Buscaglia, 1977).

This report provides new insight in the involvement of two transcription factors, X-dll3 and Pax-6, in the patterning the adult olfactory system of the frog X. laevis. Our results suggest that X-dll3 is involved in the formation and maintenance of olfactory epithelia exposed to either air or water and that Pax-6 is specifically involved in the formation of olfactory epithelium exposed to water. That Pax-6 expression can be correlated with the onset of expression of the fish-like Class I receptor genes in the larval PC and adult MC and that Pax-6 expression in the MC is coincident with rising levels of thyroid hormone suggest that Pax-6 is likely to be involved in a network of transcriptional control involving thyroid hormone at metamorphosis.

The authors thank Dr G. Panganiban for the Distal-less antibody, Dr Randall Reed for one of the Pax-6 antibodies, Dr N. Papalopulu for the X-dll3 clone, Drs C. Kintner, N. Papalopulu and A. Chitnis for assistance with the embryonic in situ hybridization, Dr Richard Harland for the stage 32 head cDNA library and Dr B. Sander for critical reading of the manuscript. This work was supported by grants to G.D.B. from NIH-NINDS NS37147 and NIH-NIDCD DC03905.