Noses and neurons: Induction, morphogenesis, and neuronal differentiation in the peripheral olfactory pathway


  • Curtis W. Balmer,

    1. Department of Cell and Molecular Physiology and UNC Neuroscience Center, The University of North Carolina School of Medicine, Chapel Hill, North Carolina
    Search for more papers by this author
  • Anthony-Samuel LaMantia

    Corresponding author
    1. Department of Cell and Molecular Physiology and UNC Neuroscience Center, The University of North Carolina School of Medicine, Chapel Hill, North Carolina
    • Department of Cell and Molecular Physiology and UNC Neuroscience Center, CB #7545, The University of North Carolina School of Medicine, Chapel Hill, NC 27599
    Search for more papers by this author


Nonaxial mesenchymal/epithelial (M/E) induction guides peripheral olfactory pathway differentiation using cellular and molecular mechanisms similar to those in the developing limbs, aortic arches, and branchial arches. At each of these bilaterally symmetric sites off the midline axis, a thickened ectodermal epithelium is apposed to a specialized mesenchyme derived largely, but not exclusively, from the neural crest. The capacity of M/E interaction in the olfactory primordia (the combined olfactory placodal epithelium and adjacent mesenchyme) to induce a distinct class of sensory receptor neurons—olfactory receptor neurons—suggests that this mechanism has been modified to accommodate neurogenesis, neurite outgrowth, and axon guidance, in addition to musculoskeletal differentiation, chondrogenesis, and vasculogenesis. Accordingly, although the olfactory primordia share signaling molecules and transcriptional regulators with other bilaterally symmetric, nonaxial sites such as limb buds, their activity may be adapted to mediate distinct aspects of cellular differentiation and process outgrowth during the initial assembly of a sensory pathway—the primary olfactory pathway—during early forebrain development. Developmental Dynamics 234:464–481, 2005. © 2005 Wiley-Liss, Inc.


The olfactory primordia, which give rise to the nose as well as the neuronal elements of the peripheral olfactory sensory pathway, are remarkably prominent on the anterolateral aspect of the developing head. Nevertheless, when considered aside their more thoroughly studied neighbors, the developing eye and ear, these rudiments of the developing nose seem somewhat neglected. Initially, the bilaterally symmetric olfactory placodes are the most recognizable component of the nascent peripheral olfactory pathway. The placodes themselves are local thickenings of the lateral surface epithelium of the head; however, these specialized epithelial domains are intimately associated with mesenchymal cells of the frontonasal mass. During midgestation, the placodes and adjacent mesenchyme grow, invaginate to form olfactory pits, and subsequently translocate from the lateral to the medial aspect of the developing face. The medial aspects of the primordia fuse to form the cartilaginous septum of the nose, and the lateral aspects give rise to the two external nares. Concurrently, the invaginated ectoderm of the olfactory pit gives rise to the olfactory as well as respiratory epithelium that lines the nasal cavity. Several distinct cellular processes, including nasal chondrogenesis, maturation of ciliated epithelial cells of the respiratory epithelium, morphogenesis of the vomeronasal and septal organs, gonadotropin releasing hormone (GnRH) neurogenesis, and differentiation of olfactory receptor neurons (ORNs) occur within the tissues of the olfactory primordia (Fig. 1; for review, see Cuschieri and Bannister, 1975; Schwanzel-Fukuda and Pfaff, 1989; Wray et al., 1989; LaMantia et al., 1993; De Carlos et al., 1995; Tarozzo et al., 1998; Cappello et al., 1999; Weiler and Farbman, 2003; Pavlov et al., 2003). The embryological, cell biological, and molecular mechanisms that mediate these diverse processes in the developing nose have not been as well studied as similar mechanisms in other sensory organs—especially the retina and cochlea. Over the past decade, however, several laboratories have identified essential tissue–tissue and cell–cell interactions, molecular signals, and transcriptional activators that guide nasal and olfactory differentiation. One central observation that has emerged from work in our laboratory is that interaction between the olfactory placodal epithelium and the associated mesenchyme is essential to establish cellular and molecular diversity as well as appropriate axon trajectories in the peripheral portion of the olfactory pathway.

Figure 1.

The early development of the peripheral olfactory pathway. A: At embryonic day (E) 9.0 in a mouse embryo, the frontonasal surface epithelium (arrows), seen in a lateral view of the embryo's head, is relatively undifferentiated. There is no evidence of neuronal differentiation in the presumptive olfactory epithelium; however, differentiated neurons, labeled with an antibody to β-tubulin (TuJ1), can be seen in the central nervous system (asterisk) and nascent cranial ganglia (arrowheads). The box indicates the level of the schematized coronal section shown in B. B: A schematic of the frontonasal mass and forebrain at E9.0, showing the forebrain neuroepithelium (fb), frontonasal mesenchyme (fnM), and the presumptive olfactory epithelium, which has not yet begun to thicken into an olfactory placode (pOP). C: At approximately E10.0, a region of the frontonasal surface epithelium has thickened to become the olfactory placode (arrows), and differentiated neurons can be detected using antibodies against TuJ1. The box indicates the level of the schematized coronal section shown in D. D: A schematic coronal section through the frontonasal region of an E10.0 mouse embryo, showing the expanded frontonasal mesenchyme as well as the thickened olfactory placode, which already has begun to generate immature neurons (purple cells). E: By E11.0, in a lateral view, the olfactory primordia are visible as cup-like structures extending laterally from the head. Neurons and axons, labeled for neural cell adhesion molecule (NCAM), can be seen in the placode and nascent olfactory nerve (arrows). The box indicates the approximate level of the coronal section shown in F and the schematic shown in G. F: Coronal view of the olfactory primordium/pit at E11.0, showing the initial projection of the nascent olfactory nerve (labeled for NCAM) extending from the medial aspect of the olfactory epithelium, anteriorly to the medial ventral region of the telencephalic vesicles. G: A schematic illustrates the organization of the E11.0 primary olfactory pathway (purple), including the olfactory nerve. H: At slightly later stages (E14.5), ciliated respiratory epithelial cells (shown at higher magnification in inset) labeled by a reporter transgene for the Forkhead1 gene (green; Forkhead indicator transgenic line courtesy L. Ostrowski, UNC Cystic Fibrosis Center) can be seen at the boundary of the respiratory and olfactory (neural) epithelium. I: Olfactory receptor neurons in the olfactory epithelium at E16.5 (labeled for olfactory marker protein, OMP, green) have a single apical dendrite oriented toward the luminal surface that terminates in numerous cilia specialized for signal transduction, labeled here with an antibody to the olfactory receptor neuron–selective adenylate cyclase ACIII (red).


The mechanisms that initially specify the ectodermal and mesenchymal components of the olfactory primordia are unclear (reviewed by Farbman, 1992; Reiss and Burd, 1997; Baker and Bronner-Fraser, 2001). Two hypotheses might explain the origin of the placodal epithelium from the frontonasal surface epithelium: first, local signals may induce indifferent ectoderm to become neurogenic; second, primary neuroectodermal induction may occur within the anterolateral neural plate ectoderm followed by planar displacement of these neurogenic cells to the placodal region in the lateral surface of the head (Couly and Le Douarin, 1985). Fate mapping in amphibians, birds, and fish suggests that the olfactory placodal ectoderm is established fairly early in embryogenesis from regions within the neural plate or tube (Zwilling, 1940; Jacobson, 1963a–c; Eagelson et al., 1995; Whitlock and Westerfield, 2000). Furthermore, surrounding cells and tissues, including the endoderm, can influence the identity of the placode. Thus, there must be some ectodermal patterning and positional constraints that confer neural competence upon the presumptive olfactory placodal epithelium quite early in embryogenesis.

The frontonasal mesenchyme adjacent to the olfactory placodal epithelium is primarily derived from neural crest cells that originate at the anterior boundary of the dorsal neural tube at the mesencephalic/diencephalic junction (Serbedzija et al., 1992; Osumi-Yamashita et al., 1994; see below). In addition, the expression of mesodermal markers in the frontonasal mesenchyme (reviewed by O'Rourke and Tam, 2002) suggests a mesodermal contribution to the frontonasal mesenchyme. This evidence is not definitive, however, because many of these markers are also expressed in neural crest cells. Thus, the precise derivation of the initial population of frontonasal mesenchymal cells remains unclear. It is possible that the initial establishment of both the olfactory placodal epithelium and adjacent mesenchyme may be similar to that for the lens, epibranchial, and otic placodes (Bhattacharyya et al., 2004) as well as the placode-like organization in limb buds (Bhasin et al., 2003). In each case, specification of cellular fields capable of distinct, yet parallel, inductive interactions may depend upon direct apposition of specialized ectoderm and mesenchyme influenced by local positional cues along the anterior–posterior (a-p) embryonic axis, as well as the arrival of neural crest cells from distinct positions along the a-p axis of the neural tube. In the olfactory primordia, subsequent interactions between placodal ectoderm and adjacent mesenchyme, modified by additional local signals, may influence critically the differentiation of the olfactory epithelium and nerve as well as the development of the nares and associated craniofacial structures.


In the 1920s, Balinsky, presumably inspired by other transplantation and ectopic induction experiments in amphibians and birds being performed at the time (for review, see Hamburger, 1988), demonstrated that transplantation of the olfactory primordium (there is no indication that any effort was made to remove the underlying mesenchyme from the olfactory placodal epithelium) beneath the flank ectoderm of a newt could induce a supernumerary limb. This experiment, all but forgotten until a decade ago (Slack, 1995), suggests three important conclusions: first, the olfactory primordium has inductive capacity; second, significant mechanistic similarities must exist between inductive interactions in the primary olfactory pathway and those in the limb; and third, the anterior surface ectoderm (i.e., olfactory placodal ectoderm) is distinct from flank ectoderm based upon its neurogenic rather than epidermogenic capacity. The morphogenesis of a limb from flank ectoderm interacting with tissues of the olfactory primordium suggests anteroposterior positional information resident in the host ectoderm significantly influences the outcome of local inductive interactions. Nevertheless, Balinksy did not assess the independent contribution of placodal epithelium vs. the associated frontonasal mesenchyme; thus, it is not clear to what degree each component contributes to the differentiation of the ectopic supernumerary limb.

In the 1970s, Graziadei and colleagues began another series of transplantation experiments in the frog, and later the rat, that suggested that the olfactory primordium can “induce” the neural tube to become a specialized target for olfactory axons (Stout and Graziadei, 1980; Graziadei and Monti Graziadei, 1986, 1992; reviewed by Dryer and Graziadei, 1994). Indeed, an olfactory primordium (once again, the extent to which the epithelial vs. mesenchymal component were separated is not made clear), when placed at diverse locations along the anterior posterior axis of the developing central nervous system (CNS), can generate ORNs. Moreover, the axons of these ectopic ORNs can innervate heterologous targets and initiate the formation of iterated circuit elements resembling the glomeruli of the olfactory bulb. Despite the ability of supernumerary ORNs to innervate ectopic locations, there is significant regulation provided by the target (Byrd and Burd, 1993). Thus, in the olfactory bulb of a frog, the relationship between the number of axons and glomeruli is not strictly linear; beyond a certain level, supernumerary ORN axons do not induce a parallel addition of glomeruli. At the time of their publication, these intriguing observations were consistent with the hypothesis that sensory pathways are first specified in the periphery and then impose that specificity upon central structures (reviewed by Killackey et al., 1995). This interpretation, however, is debatable. There may be mechanisms that specify the olfactory target independent of arrival of axons (see Yoshida et al., 1997). Furthermore, the differentiation of ORNs in ectopic locations, and the retention of essential characteristics like the ability to form glomeruli in heterologous targets, remains unexplored. Graziadei and colleagues, similar to Balinsky, did not address the independent contributions of frontonasal mesenchyme and olfactory placodal epithelium to this inductive capacity. Thus, it is difficult to determine whether their results reflect only the ingrowth of axons from ectopically generated, prespecified ORNs, or the local influence of frontonasal mesenchyme on the neural tube and olfactory placodal epithelium. These potential mesenchymal/epithelial (M/E) interactions may be important for establishing the identity of the ORNs, the competence of their axons to innervate a target, and essential aspects of the “olfactory” identity of the target, including the ability to accommodate ORN axons whose terminals eventually segregate into glomeruli.

Some of the experimentally defined inductive capacity of frontonasal tissues (defined here as the olfactory primordia) recognized by Balinsky, Graziadei, and others no doubt reflect, in part, the cellular interactions and molecular signals that normally mediate olfactory pathway differentiation. Indeed, Balinsky's observation can be interpreted as the first indication that induction in the peripheral olfactory pathway is similar to other bilaterally symmetric, nonaxial structures like the limbs, branchial arches, and aortic arches. In the peripheral olfactory pathway, these M/E inductive events result not only in initial morphogenesis and cellular differentiation, but differentiation of neural precursor populations that retain the ability to generate new ORNs throughout the life of the animal. The observations of Graziadei suggest that this inductive potential may play a specific role in establishing subsequent organization in the olfactory pathway, including some permissive influence on presumptive CNS targets for peripheral ORN axons to segregate into appropriate circuit units (glomeruli). Until recently, however, the mechanisms of this apparent inductive capacity on differentiation of the peripheral component of the olfactory pathway itself (i.e., the olfactory epithelium and nerve) have been largely unexplored. Recent observations, summarized below, define cellular, molecular and genetic bases for the differentiation of the olfactory placodal epithelium, by means of interactions with the adjacent mesenchyme, into the diverse cell types of the peripheral olfactory pathway.


Morphogenesis of the olfactory pits from the relatively undifferentiated frontonasal mass is a major event during the early development of the vertebrate head. In the mouse, on approximately the ninth day of gestation (embryonic day [E] 9, determined by identifying the day of a vaginal plug as E0.5), the tissues that constitute the olfactory primordia have not undergone significant morphological differentiation (Fig. 1A,B). There is no indication of neuronal differentiation, despite neurogenesis in other regions of the peripheral and central nervous system (Fig. 1A). The placodal epithelium has not begun to thicken (Fig. 1B, compare with Fig. 1D), and the underlying mesenchyme is relatively undifferentiated. Within an additional day (E10), just before initial invagination of the olfactory pit, the placodal epithelium is visibly thickened and has acquired some neural characteristics (Fig. 1C,D; LaMantia et al., 2000). Furthermore, the underlying mesenchyme has expanded so that the placodal epithelium and associated mesenchyme protrude noticeably from the surface of the head (Fig. 1D; reviewed by Farbman, 1992; Croucher and Tickle, 1989; Serbedzija et al., 1992; Osumi-Yamashita et al., 1994; LaMantia et al., 2000). Within less than a day (approximately E10.5), the thickened ectoderm invaginates to define the epithelium of the olfactory pit, and concurrently the medial and lateral nasal processes that constitute the rudiments of the nasal septum and nares become recognizable (Fig. 1E–G). The epithelium of the olfactory pit will generate cell biologically and molecularly distinct classes of respiratory and olfactory cells (Fig. 1H,I). There are distinct axes established in both the mesenchymal and epithelial components of the olfactory primordia during the initial stages of olfactory pit formation, and they constrain all subsequent events in the initial assembly of the olfactory pathway. These axes can be recognized by the location of differentiated neurons in the invaginating epithelium (defining the medial domain), directed growth of the nascent olfactory nerve (at this stage, toward the anterior aspect of the forebrain; Fig. 1E–G), and patterned expression of several markers in the lateral mesenchyme (LaMantia et al., 2000; see also Fig. 4).

During the subsequent 2 days (E11.5–E12.5), morphogenesis in the invaginated epithelium of the olfactory pit becomes progressively more elaborate, resulting in definition of the vomeronasal organ and rudimentary olfactory turbinates. Concurrently, subsets of epithelial cells delaminate and enter the mesenchyme of the medial nasal process, adjacent to the presumptive olfactory epithelium. The earliest migratory cells express general neuroepithelial markers and a later migratory population expresses gonadotropin/luteinizing hormone releasing hormone (GnRH/LHRH) as well as odorant receptors (Wray et al., 1989; Schwanzel-Fukuda and Pfaff, 1989; Fonaro et al., 2001; Schwarzenbacher et al., 2004). Some of these cells will migrate into the diencephalon (primarily the GnRH neurons), whereas others remain part of the migratory mass: a coherent group of neuroepithelial cells that coalesce within the mesenchyme in association with the nascent olfactory nerve (see below). In parallel with early neural differentiation, the olfactory pits are translocated from the lateral to medial aspect of the developing head. They fuse, and cartilage condensation and chondrogenesis establishes the nasal septum, resulting in a single nasal prominence with two nares at the craniofacial midline. Although the eventual morphogenesis of the nose is a critical step for proper craniofacial development (reviewed by Carstens, 2002), the remainder of this review will focus on the establishment of the neuronal components of the olfactory pathway by means of mesenchymal/epithelial (M/E) induction within the olfactory primordia as it transforms from undifferentiated frontonasal surface epithelium and mesenchyme to an olfactory pit and eventually a single nose on the facial midline.


Induction reflects a series of molecular signaling events constrained by the proximity and identity of relevant cells and tissues. Thus, aside from assessing the inductive capacity of the placodal epithelium and associated mesenchyme, it is essential to identify both signaling molecules as well as potential downstream targets that regulate differentiation in response to inductive interactions. A series of elegant transplantation/recombination experiments in the late 1980s and early 1990s (Richman and Tickle, 1989, 1992) re-emphasized the inductive capacity of the frontonasal epithelium and mesenchyme and invited further comparisons between craniofacial inductive interactions involving the olfactory primordia and those in the limb. Particularly striking were similarities between the inductive potential of craniofacial epithelium and mesenchyme (including that of the olfactory primordia), the limb, and the signaling molecule retinoic acid (RA). RA, a member of the Steroid/Thyroid superfamily of nonpeptide hormones, signals through a family of receptor-transcription factors that elicit changes in gene expression in response to ligand binding (reviewed by Evans, 1988; Tsai and O'Malley, 1994). Accordingly, RA can serve as a model inductive signal—the local availability of the ligand causes a change in gene expression in an adjacent target. Several observations, including direct biochemical identification, showed that RA is a potent local inductive signal in the limb bud (Tickle et al., 1982; Thaller and Eichele, 1987). Accordingly, based on cellular and molecular comparisons with RA signaling in limb induction (for review, see Tickle and Eichele, 1994), as well as observations showing olfactory pathway dysmorphogenesis after maternally administered RA (Shenefelt, 1972), it seemed possible that RA, locally available in the epithelium or mesenchyme of the olfactory primordia, plays a role in the early development of the olfactory pathway.

Using a combination of in vitro assays for local, restricted, RA production as well as transgenic reporter mice to detect RA signaling in embryos (Fig. 2A; Balkan et al., 1992; Colbert et al., 1993), we identified a local source of RA in the frontonasal mesenchyme (Fig. 2B–G; LaMantia et al., 1993). Thus, when the entire E10.5 frontonasal prominence (epithelium and mesenchyme) or the associated forebrain neuroepithelium is cultured on a monolayer of clonally derived RA-reporter cells, the frontonasal mesenchyme uniquely elicits an RA-response in the underlying indicator cells (Fig. 2B). Neither the presumptive olfactory epithelium nor the forebrain neuroepithelium can activate RA-mediated gene expression in the indicator cells (Fig. 2C,D). This local production of RA is established quite early; by at least E9.0 (just as the headfolds are closing), the mesenchyme adjacent to the presumptive olfactory epithelium can generate an RA response in indicator cell monolayers (Fig. 2E). Furthermore, mesenchymal production of RA is apparently autonomous because frontonasal mesenchyme continues to produce RA when presumptive olfactory epithelium is removed (Fig. 2F,G). The distribution of RA receptors and binding proteins in the olfactory placode/pit and forebrain (Osumi-Yamashita et al., 1990; Ruberte et al., 1991, 1993; Anchan et al., 1997; Whitesides et al., 1998; Bhasin et al., 2003) suggest that the local mesenchymal source of RA might activate RA-mediated changes in gene expression in adjacent epithelia. Using transgenic mice engineered to detect transcriptional change by means of RA activation of its receptors (Balkan et al., 1992), we found local, restricted RA-mediated gene expression in the epithelium of the olfactory placode and pit as well as the neuroepithelium of the forebrain—both are immediately adjacent to the RA-producing mesenchyme (Fig. 2A). Moreover, these RA-activated epithelial regions are sites of the early neuronal differentiation as well as restricted expression of several cell adhesion molecules (LaMantia et al., 1993; Whitesides and LaMantia, 1995, 1996). Thus, RA is a local M/E inductive signal in the olfactory primordium and forebrain: it is produced in one tissue compartment (M) and signals in the adjacent compartment (E). These observations provided an initial outline for the cellular and molecular mechanism of M/E induction in the olfactory pathway.

Figure 2.

There is a close relationship between retinoid signaling, the neural crest, and the olfactory epithelium. A: The pattern of local retinoic acid (RA) -mediated gene expression in the head is restricted to the presumptive olfactory epithelium (poe) and ventrolateral forebrain neuroepithelium (fb) but excluded from the frontonasal mesenchyme (fnm, hatched area) in an embryonic day (E) 10.5 mouse. In the two panels, identical patterns of RA-sensitive transgenic reporter expression in two independent transgenic lines (Tg1 and 2) are shown. B: The frontonasal mesenchyme is the local source of RA, based upon the activation of RA-mediated gene expression (indicated by blue cells) in a monolayer of RA-responsive fibroblasts cocultured with an olfactory primordium, with the mesenchymal surface placed on the cells. C: An olfactory primordium, placed epithelium side down on the same RA indicator L-cells does not elicit an RA response. D: Forebrain neuroepithelium does not produce RA, based upon lack of activation in the same L-cell indicator assay. E: In a similar assay, using a green fluorescent protein (GFP) -expressing RA indicator cell line, the E9.0 olfactory placode (placed mesenchyme side down) produces RA. F: Isolated E9.0 placodal epithelium (mesenchyme removed) does not produce RA. G: Isolated E9.0 frontonasal mesenchyme produces RA. H: The neural crest component of the frontonasal mesenchyme (blue cells), demonstrated by an enhancer trap indicator transgene (βgeo6, see LaMantia et al., 2000) is seen in the frontonasal mass before the closing of the headfolds, approximately E8.5. I–L: Serial sections through the embryo shown in H, demonstrate the contribution of the neural crest to the frontonasal mesenchyme at this early stage.

The localization of RA to the frontonasal mesenchyme from at least E9.0 raised an obvious question: is the frontonasal mesenchyme distinct in its derivation or differentiation, and might such a distinction be related to the local production of RA? This question seemed especially relevant because other nonplacodal head mesenchyme did not seem to produce RA. An important initial answer came from migration and lineage tracing studies in birds and rodents. All of these studies showed that the cranial neural crest, particularly that from anterior regions of the neural tube where Hox genes are not expressed, is an early and significant contributor to the frontonasal mesenchyme (for recent review, see Kulesa et al., 2004; Le Douarin, 2004). Indeed, neural crest-derived cells are among the earliest to arrive in the mesenchyme adjacent to the preplacodal frontonasal surface epithelium, approximately between E8.5 and 9.0, before the ectoderm has begun to thicken (Serbedzija et al., 1992; Osumi-Yamashita et al., 1994; Fig. 2H–L). Accordingly, our observations suggest that the onset of RA signaling in the olfactory primordium coincides with the arrival of the neural crest. Additional observations at other sites of nonaxial M/E interaction—particularly the branchial and aortic arches (Sulik et al., 1988; Lumsden, 1988; Creazzo et al., 1998)—suggest that teratogenic (chemical) or physical ablation of the neural crest results in anomalies that resemble those seen after RA-deprivation. Thus, it seems possible that the neural crest derived component of the frontonasal mesenchyme might account for (or contribute to) the establishment of an endogenous local source of RA in the developing olfactory pathway.

We used several approaches to evaluate the potential relationship between neural crest-derived mesenchyme and RA production. In particular, we focused on analyses of RA signaling in the Pax6 mutant mouse small eye (Pax6Sey/Sey; Fig. 3) because of several dramatic phenotypes in the olfactory pathway. In homozygous small eye mice as well as Pax6 mutant rats, the olfactory placode does not differentiate, the olfactory epithelium and bulb fail to develop, and neural crest migration to the frontonasal process is abolished (Hill et al., 1991; Matsuo et al., 1993; Grindley et al., 1995). Thus, if neural crest is responsible for local RA production in the frontonasal mesenchyme, this local RA source as well as RA-mediated gene expression in the olfactory epithelium and forebrain should be lost in small eye homozygous mice. RA signaling is absent in the remnant of the frontonasal surface epithelium and ventral forebrain in homozygous Pax6Sey/Sey embryos (Fig. 3A,B), as is the capacity of the remaining frontonasal mesenchyme to produce RA (Fig. 3C,D; Anchan et al., 1997; see also Enwright and Grainger, 2000). Moreover, several neural-specific markers associated with the neural crest, including GAP43 and Pax7, are absent in the frontonasal mesenchyme of Pax6Sey/Sey mice (Anchan et al., 1997; LaMantia et al., 2000). Furthermore, by crossing a neural crest-selective enhancer trap mouse (βgeo6; see Fig. 2H–L) into Pax6Sey/Sey mice, we confirmed that the crest-derived component is missing from the remaining frontonasal mesenchyme in mutant embryos (Fig. 3K–L; LaMantia et al., 2000). These βgeo6-labeled, presumably neural crest-derived cells of the frontonasal mesenchyme express GAP43, nestin, and the retinoid synthetic enzyme retinaldehyde dehydrogenase 2 (RALDH2; McCaffery et al., 1991; Zhao et al., 1996; Bhasin et al., 2003; see also Figs. 4, 5), which has been implicated in RA-mediated signaling during limb, heart, and craniofacial induction based upon phenotypes seen in RALDH2 null mice (Niederreither et al., 1999). The neural crest-derived, RALDH2-expressing population coincides with the local source of RA from the olfactory placode mesenchyme. In a modified indicator cell assay in which frontonasal processes from E10.5 βgeo6 indicator transgenic embryos are cultured on monolayers of RA-responsive cells that express green fluorescent protein as a reporter, βgeo6 cells in the lateral nasal mesenchyme are in register with RA-activated cells in the underlying monolayer (Fig. 3E–J; LaMantia et al., 2000; Bhasin et al., 2003). Apparently, RA-mediated signaling in the olfactory placode depends upon the normal migration and differentiation of the neural crest. Neural crest-derived RA signals alone are not likely to be responsible for olfactory placode induction or subsequent olfactory pathway differentiation. Nevertheless, RA, a molecular signal associated with induction at several other embryonic sites, is specifically produced and localized in the olfactory primordium, and altered RA signaling accompanies disruptions of initial olfactory pathway differentiation.

Figure 3.

Disrupted retinoic acid (RA) signaling is associated with altered neural crest migration to the olfactory placode. A: Lateral view of a wild-type E10.5 embryo carrying an RA indicator transgene demonstrates domains of RA-mediated gene expression in the ventral forebrain (arrowheads) and olfactory primordium (op), as well as the eye (e) and spinal cord (sc). B: Lateral view of a Pax6Sey/Sey mutant littermate in which RA-mediated gene expression in the olfactory primordium and forebrain are no longer detected; however, this activity is still present in the spinal cord. C: RA is produced by the wild-type frontonasal mass based upon activation of an RA-sensitive transgene in a monolayer of stably transfected RA indicator L-cells. D: The frontonasal mesenchyme of a Pax6Sey/Sey mutant does not produce RA, based upon the RA-indicator L-cell coculture assay. E,F: The normal relationship between neural crest-derived frontonasal mesenchyme, visualized by the βgeo6 enhancer trap (E), and domains of RA-mediated gene expression detected with the RA-indicator transgene (F). G: The βgeo6-labeled cells are confined to the apparent lateral nasal process (lnp) of the explant; the mesenchyme of the medial nasal process (mnp) has no labeled cells. H: Corresponding image showing RA-activation in a monolayer of RA detector cells (in this case with a green fluorescent protein [GFP] reporter) in a coculture assay using the explant shown in G. I,J: Plots of β-gal (neural crest; blue dots, I) and RA reporter L-cells (green dots, J) demonstrate spatial coincidence of the neural crest mesenchymal population and local RA production. K,L: Wild-type (K) and littermate Pax6Sey/Sey (L) embryos carrying the βgeo6 transgene demonstrate the location of the neural crest. In the Pax6Sey/Sey embryo, all other crest derivatives are seen, including the mesenchyme in the first and second branchial arches (bI and bII), the maxillary process (mx), and the fifth and eighth cranial ganglia (V and VIII). In contrast, the crest-derived mesenchyme in the frontonasal process (fnm), prominent in the wild-type embryo (arrow in K), is absent in the Pax6Sey/Sey mutant (arrow in L), in register with the failure of RA production from the frontonasal mesenchyme.

Figure 4.

An in vitro assay for studying mesenchymal/epithelial (M/E) interactions in the olfactory placode. Top panel: Schematic of the dissection and subsequent separation of mesenchymal (M) and epithelial (E) components of the embryonic day (E) 9 frontonasal mass/forebrain shows the approximate origin of the starting tissues and the basic approach of separation and recombination. In the inset, a recombined olfactory primordium (epithelium and associated mesenchyme) is shown in which epithelium comes from a wild-type mouse, and mesenchyme from a ROSA26 mouse in which a βgal reporter (blue) labels all cells. The asterisk indicates an accumulation of epithelial cells and processes assumed to represent the migratory mass, which in vivo defines the advancing front of the nascent olfactory nerve as it grows from the epithelium to the forebrain. Middle rows: Images showing expression of two neuronal markers (β-tubulin and neural cell adhesion molecule [NCAM]) in the medial aspect of the olfactory pit and two primarily mesenchymal markers (retinaldehyde dehydrogenase 2 [RALDH2] and Pax7) limited to the lateral vs. medial nasal processes. In each case, the expression and pattern of the marker in the embryo and the olfactory primordium in vitro are the same. Bottom row: With the exception of RALDH2 (a retinoic acid [RA] synthetic enzyme), expression of the remaining markers relies upon M/E interaction, because they are not seen in isolated frontonasal epithelium (e) or mesenchyme (m) cultured for the same period as the recombined explants. The asterisks indicate the presumed migratory mass, which is labeled by two markers (b-tubulin and NCAM) that also stain axons and cells of the migratory mass in vivo.

Figure 5.

Local signals are correlated with distinct axes and influence gene expression and cellular differentiation in the olfactory primordium. Left hand panels. First row: Immunocytochemical or in situ hybridization localization of four signals associated with the lateral (retinoic acid [RA]), medial (fibroblast growth factor [FGF], sonic hedgehog [shh]) and posterior (bone morphogenetic protein 4 [BMP4]) axes of the olfactory placode; these signals are expressed in similar axial patterns in the olfactory primordium in vitro as those seen in the embryo. Second Row: Expression of the mesenchymal marker Pax7 is shifted in response to adding various signaling molecules to the medium. In each case, this pharmacological gain of function results in an expression change in register with the axis that the signal represents. Thus, there expansion of Pax7 in response to RA, suggesting “lateralization” of the placode; constriction of Pax7 in response to FGF8, suggesting “medialization,” and an almost complete elimination of Pax7 in response to BMP4, consistent with “posteriorization.” Third Row: Pharmacological gain of function has complimentary effects on neural differentiation, associated with the medial axis of the placode, visualized here using neural cell adhesion molecule (NCAM) immunohistochemistry. NCAM neurons are limited to a small region of the epithelium, and their number declines in response to RA (see LaMantia et al., 2000), consistent with suppressing the medial axis. In response to FGF8, the distribution of NCAM neurons expands, and their number increases (LaMantia et al., 2000), consistent with expanding this medial characteristic. In response to BMP4, neuronal differentiation is abolished in the epithelium, and ectopic NCAM-expressing cells with randomly oriented processes are seen in the mesenchyme. Right panel: A schematic of the location of four local signals in register with axes in the olfactory placode and forebrain at embryonic day (E) 9, before placode differentiation, and at E11, when the basic organization of the primary olfactory pathway is initially recognizable.


The apposition of mesenchyme and epithelium, the participation of neural crest, the availability of an established inductive signal (RA), and the coincidence of olfactory pathway anomalies with those in the limbs, heart, and face (reviewed by LaMantia, 1999; Maynard et al., 2001) all suggest that M/E induction in the olfactory primordia may resemble that at other nonaxial sites, including the limb buds, aortic arches, and branchial arches. To evaluate this hypothesis further, we analyzed directly the mechanism and consequences of M/E induction in the frontonasal mass (Fig. 4) using a novel in vitro assay based on techniques used previously in limb buds and branchial arches (Neubuser et al., 1997). When frontonasal mesenchyme and surface epithelium from E8.5 to E9.0 mouse embryos are separated and then re-apposed in vitro, in the absence of a forebrain target or systemic cues, several cardinal features of peripheral olfactory pathway development are recapitulated. Maintenance of donor tissue segregation (E or M) from distinct genotypes (ROSA26 vs. wild-type) in these explant cultures indicates that this process reflects molecular signaling, rather than cell migration, between the two tissue compartments (Fig. 4, top panel, far upper right).

In the explants, neuronal- and mesenchymal-specific molecules are expressed in patterns similar to those seen in the olfactory primordium in vivo (Fig. 4, middle rows). Thus, β-tubulin, which recognizes the initial complement of postmitotic neurons in the medial olfactory pit epithelium in vivo (Fig. 4, first row, far left), is most prominent in the apparent medial aspect of thickened epithelial domain of frontonasal explants. In addition, β-tubulin–labeled cells and processes can be seen in the cellular extension at the apparent anterior end of the explant (asterisk in Fig. 4, first row, far left). RALDH2, the RA synthetic enzyme associated with the local source of RA from the lateral frontonasal mesenchyme in vivo (Fig. 4, first row, middle left), is restricted to an apparently lateral mesenchymal domain in explants (Fig. 4, second row, middle left). Neural cell adhesion molecule (NCAM), associated with ORNs that are actively extending an axon in vivo (Fig. 4, first row, middle right; Caloff and Chikaraishi, 1989; Miragall et al., 1989), is expressed in the medial aspect of the thickened epithelium as well as in processes that define a coherent nerve terminating in a cellular mass (asterisk in Fig. 4, second row, middle right). Pax7, whose expression in the periphery is restricted to the lateral frontonasal mesenchyme (Fig. 4, first row, far right), is similarly restricted to the apparent lateral mesenchyme in explants. Finally, expression of most of these markers relies upon M/E interaction (Fig. 4, third row). The one exception is expression of the RA synthetic enzyme RALDH2, which remains in isolated frontonasal mesenchyme, consistent with the autonomous production of RA from this tissue (Bhasin et al., 2003; see also Fig. 2G).

Apparently, molecular and cellular differentiation in the explants follows cardinal axes seen in vivo: neuralization of the placodal epithelium predominates medially, the medial and lateral frontonasal mesenchyme are distinct, and axons fasciculate and grow anteriorly toward a forebrain target that—in vitro—is not there (Fig. 4, lower panels). Accordingly, we surmised there are mediolateral and anteroposterior axes in each olfactory primordium that constrain initial differentiation of ORNs and the olfactory nerve, similar to the proximodistal and anteroposterior axes within which limb and craniofacial development occur. It seemed possible that such local axial information might be established or maintained by M/E interactions. Thus, M/E induction, mediated by local signals, may be an initial determinant of ORN genesis as well as the trajectory and target of the peripheral portion of the olfactory pathway, analogous to its role in determining skeletal, digital, and odontogenic organization in limbs and face.

Established inductive signaling molecules, including RA, fibroblast growth factor 8 (FGF8), sonic hedgehog (Shh), and bone morphogenetic protein 4 (BMP4), all of which are associated with local axes in limbs, branchial arches, and heart (Johnson and Tabin, 1997; Farrell et al., 1999; Schneider et al., 1999), might also be associated with local axes in the olfactory primordia. Accordingly, we manipulated these signals in our in vitro assay to evaluate their specific role in establishing axes and influencing initial differentiation of the peripheral olfactory pathway. As at other inductive sites, the expression or availability of RA, FGF8, shh, and BMP4 are associated with distinct axes of the olfactory primordia in vivo as well as in vitro (LaMantia et al., 2000; Bhasin et al., 2003; Fig. 5, top left panels). Moreover, alterations in RA, FGF8, and BMP4 signaling by means of direct addition of each molecule or functional antagonists shift differentiation in explants along the axis with which each signal is normally associated (LaMantia et al., 2000; Fig. 5, bottom left panels). For example, RA, a presumed lateral signal, when added to the medium, expands expression of Pax7, a lateral mesenchymal marker, and reduces the numbers of NCAM-labeled neurons normally restricted to the medial epithelium (Fig. 5, first column). In contrast, addition of FGF8, a presumed medial signal, results in diminished, restricted expression of Pax7 and an expansion of the number and distribution of NCAM-expressing neurons (Fig. 5, second column). Surprisingly, Shh had general growth promoting activity as has been suggested for other cell populations (Wechsler-Reya and Scott, 1999; Zhu et al., 1999) but did not shift axes or alter neuronal differentiation in the in vitro assay (see below; Balmer and LaMantia, 2004). BMP4, an apparent posterior signal, results in the loss of most anterior structures, including the nascent olfactory epithelium; instead, there is apparent neural or glial differentiation in the mesenchyme, consistent with BMPs influence on neural and glial differentiation from neural crest (Jin et al., 2001). Thus, initial differentiation of the peripheral olfactory pathway is, at least in part, a local inductive event. It occurs within a set of axial coordinates that are either established or maintained by interactions between the frontonasal mesenchyme and surface epithelium and relies upon signaling molecules with distinct local activities.

The established inductive signaling molecules assessed thus far in the olfactory primordia—RA, FGF8, BMP4, and Shh (Fig. 5, right panel)—are shared by the limb buds, branchial arches, and aortic arches where they have similar influences on establishment or maintenance of local axes. Considering these similarities, it is not surprising that, when placed underneath flank ectoderm, as in Balinsky's classic experiment, the tissues of the olfactory primordium (placodal epithelium and associated mesenchyme) can induce a supernumerary limb. Indeed, at least three of these molecules (RA, FGF8, and Shh), when presented by means of slow-release microspheres (rather than from a complex ectopically placed tissue source like the olfactory primordium) can also induce a supernumerary limb (Tickle et al., 1982; Riddle et al., 1993; Cohn et al., 1995). A central question that remains, however, is how (or whether) these common molecular signals are adapted to mediate specific aspects of neuronal differentiation in the olfactory pathway. It is possible that these signals—RA, FGF8, BMP4, and Shh—as well as others directly influence the acquisition of ORN characteristics or the establishment of distinct precursors in the ORN lineage. Alternately, the shared signals may serve a more generic role, modulating widely expressed transcriptional regulators or influencing cell proliferation or survival without any specificity for establishing cell identity. The contribution of these shared signals to establishing the unique cellular characteristics of the ORN, as well as their capacity to project to the forebrain during initial development of the olfactory pathway, remains to be determined.


ORNs must use extrinsic cues, presumably available locally from the adjacent mesenchyme as well as at a distance from the telencephalon, to guide the growth of their axons from the periphery into the forebrain, de novo. In contrast, other primary sensory receptor cells, like photoreceptors, are generated within the CNS, make local synapses, and do not have axons that relay sensory information to other processing centers in the brain. Dorsal root and cranial sensory ganglion cells originate from neural crest cells; their cell bodies are displaced from the CNS by small distances in the embryo, and their centrally projecting processes most likely follow migratory routes of their neural crest precursors. Hair and taste cells are specialized neuroepithelial cells without axons—they relay their information to the CNS by means of specialized synapses onto sensory ganglion cells. Thus, among the full range of sensory transduction cells, only ORNs have axons that must pioneer a pathway from the periphery to the central nervous system in a region through which neuroepithelial cells or neuronal processes have not migrated. In mammals, the entry of axons (as well as population of migratory cells) from the olfactory epithelium (OE) into the mesenchyme is thought to be simultaneous; however, in fish, there is a population of migratory neuroepithelial cells that apparently pioneer the pathway from the olfactory placode to forebrain (Whitlock and Westerfield, 1998; Whitlock, 2004). These events rely upon the integrity of the same tissue compartments that interact during initial differentiation of the olfactory primordium: the epithelium and the mesenchyme. At this point in mammalian olfactory development, however, the mesenchyme becomes further specialized to support fasciculated axon growth as the nascent olfactory nerve forms. This dimension of mesenchymal differentiation may be related to the distribution and activity of inductive signals (e.g., RA and BMP4) that are also localized within specialized mesenchymal populations (Whitesides et al., 1998; Shou et al., 2000), or it may rely upon other signaling and adhesion molecules. Accordingly, it seems possible that continued interactions between mesenchyme and maturing epithelium ensure the normal completion of this process of constructing an olfactory nerve that relays chemosensory information from the periphery to the forebrain.

One of the most striking features of the nascent olfactory nerve is its coherence. Despite originating from neurons scattered throughout the medial epithelium of the olfactory pit, the initial cohort of ORN axons coalesce into a single nerve that projects exclusively to the ipsilateral telencephalic vesicle shortly after the olfactory pit has formed (Marin-Padilla and Amieva, 1989; Gong and Shipley, 1996; Whitesides and LaMantia, 1996). These early growing axons are associated with a heterogeneous population of cells called the migratory mass, much of which is apparently derived from the OE rather than the mesenchyme (LaMantia et al., 2000; see Fig. 4). This complex cellular aggregate includes olfactory ensheathing cells (OECs), a unique population of glia that wrap fascicles of ORN axons, and GnRH neurons that migrate from the placode to the ventral hypothalamus (Hinds, 1972; Doucette, 1984; Wray et al., 1989; Schwanzel-Fukuda and Pfaff, 1989; Monti-Graziadei, 1992; Valverde et al., 1992). Upon reaching the ventromedial aspect of the ipsilateral telencephalic vesicle, ORN axons, tightly fasciculated into a single nerve and associated with the migratory mass, are divided into smaller fascicles that penetrate the ventromedial forebrain through fenestrations in a limited region of the basal lamina (Marin-Padilla and Amieva, 1989; Treloar et al., 1996; Whitesides and LaMantia, 1996; Balmer and LaMantia, 2004). As the number of axons entering the forebrain increases, the basal lamina at the site of entry disintegrates completely (Marin-Padilla and Amieva, 1989); large numbers of ORN axons and OECs enter the forebrain, and these axons eventually constitute the olfactory nerve fiber layer of the rudimentary olfactory bulb. Thus, the initial assembly of the olfactory nerve reflects migration of epithelial-derived cells as well as concomitant extension of ORN axons into the frontonasal mesenchyme. These epithelial cells and processes are maintained in a restricted region of the mesenchyme before entering the forebrain at a distinct ventromedial location. Their presence within the mesenchyme creates a novel opportunity for additional cell–cell interactions between mesenchymal- and epithelial-derived components of the nascent olfactory pathway. As development proceeds, the olfactory epithelium becomes larger and more complex, and the olfactory nerve is subdivided into multiple fascicles that travel through the cribriform plate—a cranial element that differentiates from the frontonasal mesenchyme—into the olfactory bulb. It is not know whether apparent adhesive constraints and cellular interactions that influence initial construction of the olfactory nerve continue to operate in the maturing nerve or during ongoing regeneration after ORN turnover in the adult.


During initial assembly of the olfactory nerve, several adhesion molecules and related signaling factors apparently contribute to the striking cohesiveness of ORN axonal growth to the forebrain (Fig. 6A,B). Throughout the 1990s, several laboratories mapped the distribution of Ca++-independent adhesion molecules (CAMs) in the pre- and postnatal olfactory pathway. Several CAMs, including the non-polysialylated and polysialylated forms of the neural cell adhesion molecule NCAM (Fig. 6A) as well as the L1 CAM are expressed in the nascent olfactory pathway (from E10.5 onward in mouse; Miragall et al., 1989; Gong and Shipley, 1996; Whitesides and LaMantia, 1996). Other CAMs, including the mammalian homologue of the Drosophila axon adhesion molecule Fasciclin II (also known as OCAM) as well as distinct carbohydrate moieties are seen at slightly later stages of olfactory nerve development (Hamlin et al., 2004; Storan et al., 2004). In addition, early ORN axons express nonreceptor tyrosine kinases src and fyn that are thought to influence axon growth and guidance by mediating CAM signaling (Maness et al., 1996; Whitesides and LaMantia, 1996; Morse et al., 1998). Null mutations in src and fyn, or compound loss of function of both genes disrupts axon fasciculation in the mesenchyme (Morse et al., 1998). Furthermore, there is evidence that several members of the Ca++-dependent adhesion molecules family (cadherins) are expressed in the olfactory epithelium early in development (Whitesides and LaMantia, 1996; Liu et al., 2004), and several Ephrins can be localized to the epithelium as well as ORN axons at early stages of development (St. John et al., 2000, 2002). Thus, cell surface adhesion molecules, many of which influence axon pathfinding or fasciculation in other developing pathways, are expressed by ORNs during the initial assembly of the olfactory nerve (summarized in Fig. 6B).

Figure 6.

The initial assembly of the olfactory nerve reflects different growth cone behaviors in distinct domains of adhesion molecule expression in the frontonasal mesenchyme or forebrain epithelium. A: A photomontage of the primary olfactory pathway at embryonic day (E) 12 in the mouse, showing the olfactory epithelium (OE), the olfactory nerve/migratory mass (ON/mm), the zone of ON entry into the forebrain, and the territory of the presumptive olfactory bulb (Fb/pOB). The olfactory receptor neurons (ORNs) and their axons have been labeled immunohistochemically for neural cell adhesion molecule (NCAM, red), and the mesenchyme adjacent to the olfactory nerve as well as the tissues surrounding forebrain is labeled for the extracellular matrix molecule laminin (green). B: A schematic of the nascent olfactory pathway showing domains of adhesion molecule expression and behavior of ORN growth cones in those domains, as well as in response to individual adhesion molecules assayed in vitro. C,D: Examples of ORN growth cones labeled using the carbocyanine dye 1,1′, di-octadecyl-3,3,3′,3′,-tetramethylindo-carbocyanine perchlorate (DiI) in vivo upon their entry into the forebrain vesicle (C) and when grown on purified L1 (D), which is found on processes as well as cells in the ventrolateral forebrain. E,F: Examples of the fasciculated ORN axons and individual simple growth cones from the olfactory nerve within the mesenchyme (E) and ORN axons grown on purified laminin in vitro (F). G: Higher magnification image of the olfactory epithelium (OE) showing ORNs axons as they initially exit the OE.

In addition to these cell surface molecules, several secreted extracellular matrix adhesion molecules (ECMs) can be found in distinct relationships to the nascent olfactory nerve. Laminin as well as type IV collagen delineate the trajectory of the nerve in a limited region of the mesenchyme, suggesting a role for these molecules in directing and constraining ORN axon outgrowth in a manner similar to that for peripheral sensory and motor nerves from the spinal cord and hindbrain (Rogers et al., 1986; Riggott and Moody, 1987; Letourneau et al., 1988; Liesi and Silver, 1988; Gong and Shipley, 1996; Treloar et al., 1996; Whitesides and LaMantia, 1996; Julliard and Hartmann, 1998). This patterning of ECM may reflect the secretion of laminin (as well as other tropic and trophic signals) by early OECs associated with the migratory mass (Doucette, 1990; see below) and, thus, may resemble similar patterning and signaling by early migrating Schwann cells associated with peripheral spinal or cranial sensory and motor nerves (reviewed by Reichardt and Tomaselli, 1991; Jessen and Mirsky, 1999; Chernousov and Carey, 2000). Our studies suggest that the differential distribution of some of these adhesion molecules, perhaps in concert with factors expressed in the mesenchyme (see below), influence the complexity and pathfinding behavior of the earliest ORN growth cones (Whitesides and LaMantia, 1996; LaMantia et al., 2000; Fig. 6B–G). Moreover, at least one inductive signal, RA, can potentiate ORN axon growth and fasciculation in a substrate-selective manner (on laminin but not fibronectin), suggesting that local mesenchymal signals may indeed influence axon growth as well as initial patterning and cellular differentiation (Whitesides et al., 1998). Thus, the distribution of adhesion molecules and perhaps local signaling molecules, mediated in part by the interaction of epithelial-derived cells of the migratory mass with the adjacent mesenchyme, may result in distinct fasciculation or defasciculation of ORN axons at “decision” regions in the early developing olfactory pathway, as is the case in other systems (Tosney and Landmesser, 1985; Bovolenta and Mason, 1987).

One intriguing aspect of the relationship between ORN axons and the frontonasal mesenchyme is where ORN axons do not go. Additional extracellular matrix adhesion molecules within the frontonasal mesenchyme may contribute to this aspect of olfactory pathway development. Fibronectin, in contrast to laminin, tends to be absent from regions of axon growth (Riggott and Moody, 1987). Before ORN axon outgrowth, fibronectin is expressed diffusely throughout the frontonasal mesenchyme but is then gradually eliminated from regions through which olfactory axons project (Croucher and Tickle, 1989; Gong and Shipley, 1996; Whitesides and LaMantia, 1996). In addition, the secreted adhesion molecule netrin (which has some homology to ECM molecules, particularly the B1 laminin; Serafini et al., 1994) and its receptor, deleted in colorectal cancer (DCC; Keino-Masu et al., 1996), have dynamic patterns of expression during initial phases of ORN axon growth (Astic et al., 2002). Similarly, the axon repulsive semaphorins are expressed in the telencephalon before olfactory bulb differentiation and are thought to prevent entry into the forebrain before maturation of the olfactory bulb rudiment by signaling through neuropilin receptors expressed by ORNs (Pasterkamp et al., 1999; Renzi et al., 2000). Other ECM and related molecules may influence where ORN axons do and do not grow in the mesenchyme. ORN axons in vitro fasciculate and grow preferentially on laminin (Whitesides and LaMantia, 1996; Kafitz and Greer, 1997; Fig. 6E,F). In contrast, fibronectin has substantially lower growth-promoting activity for ORN axons and thereby may constrain ORN axon trajectory and prevent innervation of ectopic regions in the frontonasal mesenchyme, perhaps by graded rather than absolute differences in adhesion signaling (Fig. 6; Whitesides and LaMantia, 1996). The expression of a variety of cell surface and extracellular matrix adhesion molecules on ORN axons and in the frontonasal mesenchyme suggests that initial guidance to the forebrain includes regulation of both epithelial and mesenchymal gene expression and cellular differentiation. Remarkably, there is little data on the expression of integrins in ORNs, even though these molecules act as receptors for ECM adhesion molecules, interact with CAMs, and may influence the signaling context for cadherins (for review, see Panicker et al., 2003; Lee and Juliano, 2004). There is some indication that target-derived guidance signals, perhaps in concert with ECM components, influence axons and associated glial cells (including olfactory ensheathing cells) growing though mesenchyme to targets, whether skin, teeth, or the forebrain (Liu et al., 1995; Patapoutian et al., 1999; O'Connor and Tessier-Lavigne, 1999). Thus, the trajectory of ORN axons through the frontonasal mesenchyme and the adhesive mechanisms that constrain this process may be shared among multiple sites of nonaxial M/E interaction.


A critical test of apparent similarities of induction in the olfactory primordia and other nonaxial sites is whether mutations compromise the developing olfactory pathway, limbs, heart, and face concomitantly. Moreover, it is essential to assess whether phenotypes at each site are restricted to similar non-neural elements (e.g., common myogenic, skeletogenic, or angiogenic mechanisms) or disrupt neuronal differentiation in parallel with their effects on other tissues. Over the past decade, several single gene mutations—both naturally occurring and produced by means of homologous recombination—have been identified that compromise olfactory development as well as that at other M/E inductive sites. To assess relationships between olfactory phenotypes and those in limbs, hearts, and faces, we examined the development of the aberrant frontonasal region of small eye (Pax6; Pax6Sey/Sey) at later stages of development, and we analyzed the consequences of two additional mutations: extra toes (Gli3; extra toes, Jackson Labs: XtJ) and a null allele of sonic hedgehog (Shh; Shh−/−).

The small eye mutation (Pax6Sey/Sey) in the murine Pax6 gene is best known for its phenotypic consequences for eye and lens development in both heterozygotes and homozygotes (Ashery-Padan and Gruss, 2001; reviewed by Treisman, 2004). Failure to form the olfactory placode most likely reflects disruption of the epithelial and mesenchymal compartments of the frontonasal mass because Pax6 is expressed in the placodal epithelium (Grindley et al., 1995), and indirectly maintains the integrity of the frontonasal mesenchyme (Anchan et al., 1997; LaMantia et al., 2000; see above). Mutant placodal epithelial cells do not readily acquire neuronal characteristics in chimeric experiments both in vitro and in vivo, suggesting that Pax6 ectoderm is not easily neuralized, nor can neurons derived from Pax6Sey/Sey frontonasal surface ectoderm acquire characteristics of mature ORNs (Quinn et al., 1996; LaMantia et al., 2000; Collinson et al., 2003). In the absence of any nasal structures and olfactory neurons, remaining craniofacial elements of Pax6Sey/Sey embryos are highly dysmorphic (Anchan et al., 1997). Thus, we asked whether expression or patterning of mesenchymal ECM molecules associated with olfactory nerve development is retained in this altered environment. The residual cranial mesenchyme continues to express fibronectin throughout the facial remnant, and the fibronectin component of the basal lamina surrounding the telencephalon is continuous (Fig. 7A). Similarly, laminin is seen around blood vessels, as well as around small axon fascicles of the presumed trigeminal nerve in the highly atrophic maxillary process. There is no indication, however, that laminin remains in a domain similar to that seen when ORN axons and OECs migrate through the mesenchyme, nor is there any fenestration of the laminin component of the telencephalic basal lamina (Fig. 7B). These observations in Pax6Sey/Sey embryos suggest that in the absence of a second phase of M/E interactions—those between the maturing frontonasal mesenchyme and epithelial-derived ORN axons and OECs—appropriate patterning of ECM components does not occur.

Figure 7.

Mesenchymal extracellular matrix adhesion molecule (ECM) patterning and axon trajectories are altered in Sey/Sey, XtJ/XtJ, and Shh−/− mutant embryos. A,B: Montages of coronal sections through the remnant of the face (there is no nose) and forebrain in a Sey/Sey embryo demonstrate changes in ECM patterning in the absence of olfactory receptor neuron (ORN) axons and migratory mass cells. A: Fibronectin is distributed throughout the remaining cranial mesenchyme with no apparent discontinuities in the cranial portion or in the fibronectin component of the telencephalic basal lamina (asterisks). Fibronectin is diminished in regions where small fascicles of presumed fascicles of the maxillary branch of the trigeminal nerve innervated the remaining maxillary process (arrowheads). B: Laminin is seen around blood vessels (arrows), associated with presumed maxillary trigeminal fascicles (arrowheads), and throughout the telencephalic basal lamina (asterisks). C: The normal trajectory of the nascent olfactory nerve at E12, especially its apposition to the ventromedial forebrain, can be seen in a sagittal section. The migratory mass is seen in the mesenchyme ventral to the telencephalon and as it approaches the presumptive olfactory bulb (POB) the axons enter the forebrain (arrows), and the basal lamina is fragmented. D: In a sagittal section through an embryonic day (E) 12 XtJ/XtJ embryo, the olfactory nerve diverges from a trajectory oriented toward the ventral forebrain (dotted and solid arrows), and the basal lamina remains intact at the normal site of ORN axon entry. E: In a coronal section through an E12 XtJ/XtJ embryo, the olfactory nerve can be seen extending toward the dorsal telencephalic vesicle, where ORN axons are not normally found. F: One of two trajectories seen for the single olfactory nerve that extends through the proboscis in a Shh−/− embryo at E12. In this trajectory, shown here in a sagittal section, axons from the olfactory epithelium (OE) coalesce into an olfactory nerve (ON) that grows caudally toward the remnant of the dorsal telencephalic vesicle (TV). G: The second trajectory, also shown here in sagittal section, is directed toward the ventral aspect of the telencephalic remnant, where the ORN axons encounter and apparently fasciculate with axons from the fused, ectopic trigeminal ganglion. Some axons defasciculate, turn, and grow toward the ventral telencephalon (arrows and inset). H: In coronal section, the single OE of an E12 Shh−/− embryo is seen with a few presumed ORN axon fascicles (arrow), distal from the OE and ORN fascicles, there is a peripheral array of fascicles presumed to originate from the trigeminal ganglion (arrowheads).

The heterozygous extra toes-Jackson (Xtj) mutation (a 53-kb deletion in the murine Gli3 gene; Hui and Joyner, 1993; Maynard et al., 2002) is known primarily for its eponymous polydactyly. Homozygous (XtJ/XtJ) embryos have a more severe polydactylous phenotype and lack olfactory bulbs; however, the olfactory epithelium and nerve remain (Johnson, 1967; Schimmang et al., 1992). Null mutants of the signaling molecule sonic hedgehog (Shh−/−), which is transcriptionally repressed by Gli3 at several embryonic sites (Wang et al., 2000; Litingtung and Chiang, 2000; Litingtung et al., 2002), have severe craniofacial, limb, and heart abnormalities, as well as loss of ventral forebrain structures, including the olfactory bulbs (Chiang et al., 1996). Thus, both XtJ/XtJ and Shh−/− embryos have a spectrum of phenotypes at nonaxial M/E inductive sites as well as in the olfactory pathway. Accordingly, we asked whether the activity of these two genes influences any aspect of M/E interactions that mediate early olfactory pathway development.

The constitutive loss of Gli3 and Shh function in XtJ/XtJ and Shh−/− embryos results in distinct, localized disruption of ORN axon trajectories. The most consistent alterations in ORN axon growth are seen in XtJ/XtJ embryos. During early development of the primary olfactory pathway in the XtJ/XtJ mutant, ORN axons project aberrantly into the intertelencephalic mesenchyme (rather than approaching and fenestrating the ventral telencephalon), and their association with laminin is disrupted (Fig. 7D,E). In Shh−/− embryos, the assembly and initial trajectory of a single coherent olfactory nerve through an elongated proboscis with a single nare is surprisingly normal, and the relationship between ORN axons and ECM components within the mesenchyme of the proboscis is not significantly altered (Fig. 7F–H). As the axons approach the mutant forebrain, however, their growth is disrupted. In fact, there are at least two distinct abnormal trajectories: in some Shh−/− embryos, the single olfactory nerve ends in a neuroma at the dorsal remnant of the forebrain; in others, the olfactory nerve extends toward the ventral telencephalon where it fasciculates with the misplaced trigeminal nerve (Fig. 7F,G). These differences may reflect the variability in the morphogenetic relationship between the fused proboscis and forebrain remnant. The axon guidance phenotypes in XtJ/XtJ and Shh−/− embryos are not likely to be cell-autonomous. Neither Gli3 nor Shh is expressed in ORNs: Gli3 is found primarily in the frontonasal mesenchyme as well as in the telencephalon (Hui and Joyner, 1993), whereas Shh is expressed in a narrow, non-neural zone of the olfactory placodal epithelium as well as in a distinct, limited domain of the ventromedial forebrain (Platt et al., 1997; LaMantia et al., 2000). Accordingly, the disrupted axon trajectories are most likely secondary to changes in craniofacial mesenchymal or forebrain differentiation that result from loss of Gli3 transcriptional regulation or Shh signaling. Within this altered craniofacial environment, cell–cell interactions between ORN axons, OECs, and other epithelial cells and the frontonasal mesenchyme may fail to provide adequate guidance information to ensure appropriate formation and termination of the olfactory nerve.

Surprisingly, despite disrupted ORN axon trajectories in Gli3 and Shh mutants, several aspects of ORN differentiation are preserved. Cell biological and genetic experiments over the past decade have defined a set of transcriptional regulators associated with distinct populations of ORN precursors (Fig. 8, top panel). These regulators include the Sox B1 factors, particularly Sox2 (Ellis et al., 2004), associated with early neural precursors, and the homeodomain transcription factor MEIS1, which is expressed in neuroepithelial and epithelial cells in the lateral aspect of the developing OE (Toresson et al., 2000; Balmer and LaMantia, 2004). Subsequently, the achaete scute homologue Mash1 is expressed in an intermediate ORN precursor, and Mash1 function is necessary for the genesis of a normal complement of ORNs (Guillemot et al., 1993; Gordon et al., 1995). The basic helix–loop–helix/neurogenic genes Ngn1 and NeuroD are expressed in immediate ORN precursors (transit-amplifying precursors) or early postmitotic ORNs, and are thought to be downstream in a cascade of sequential gene expression in the ORN lineage initiated by Mash1 (Cau et al., 1997, 2002). Several general markers of maturing neurons, including GAP-43 and NCAM, are next expressed in early differentiating ORNs (i.e., those extending axons; Calof and Chikaraishi, 1989; Verhaagen et al., 1989). Finally, the ORN-selective olfactory marker protein (OMP) is expressed in mature ORNs, presumably those whose axons have reached the olfactory bulb (Miragall and Monti Graziadei, 1982; Baker et al., 1989), and the expression of the ORN-associated adenyl cyclase ACIII indicates functional maturation (Menco et al., 1992). In both XtJ/XtJ and Shh−/− embryos, these aspects of ORN differentiation are not obviously compromised. There are apparently normal populations of MEIS1, Mash1, and Ngn1 precursors, and normal expression of several markers of ORN differentiation and maturation, including GAP43, NCAM, OMP, and ACIII (Fig. 8; Balmer and LaMantia, 2004). Thus, disruption of Gli3 transcriptional regulation or shh signaling does not compromise M/E inductive mechanisms that influence ORN genesis or the establishment of molecularly distinct precursor populations. Gli3 and Shh influence cellular differentiation in the spinal cord (reviewed by Jacob and Briscoe, 2003); however, they do not seem to be critically involved in the initial differentiation of ORNs or related precursors. Moreover, it is unlikely that only Gli3 disrupts the final steps of functional ORN differentiation, because subsets of odorant receptors remain expressed in the mutant OE of XtJ/XtJ embryos (Sullivan et al., 1995).

Figure 8.

Top panel: A summary of current data on the association of molecular markers with distinct populations of olfactory receptor neuron (ORN) precursors. Bottom panels: Essential molecular characteristics of olfactory epithelium (OE) precursors and ORNs are preserved in Gli3 (Xtj) and shh null mutants. MASH1, NGN1, and neural cell adhesion molecule (NCAM) are associated with early precursors, transit amplifying precursors, early neuroblasts, and immature ORNs, respectively. The olfactory marker protein OMP and the ORN-selective adenylate cyclase ACIII characterize mature ORNs, and the low affinity nerve growth factor receptor P75 is associated with the olfactory ensheathing cells that migrate from the OE in association with ORN axons within the olfactory nerve.

The olfactory phenotypes in Pax6Sey/Sey, XtJ/XtJ, and Shh−/− embryos indicate that there may be distinct signaling and transcriptional mechanisms that mediate the early assembly of the olfactory pathway. These mechanisms operate in parallel with those in the limbs, face, and heart; however, it is not clear to what extent phenotypic change at each site represents equivalent functions of Pax6, Gli3, or Shh. Appropriate patterning of ECM components may only occur when ORN axons and OECs enter the cranial mesenchyme during a subsequent phase of epithelial/mesenchymal interaction. When the axons are not present, as is the case in Pax6Sey/Sey embryos, there is no subsequent patterning of ECM as well as related molecules in the remaining frontonasal mesenchyme. Mutations in Gli3 and Shh also disrupt this aspect of olfactory pathway development, perhaps due to an altered mesenchymal environment as well as a dysmorphic forebrain target that no longer provides appropriate local or long-range signals. Such disruptions may compromise interactions in which ORN axons, ensheathing cells and other epithelial-derived migratory cells respond to mesenchymal cues and modify the local organization of ECM to define the trajectory of the nascent nerve. Although these interactions may not be strictly inductive, they nevertheless depend upon the integrity of the two tissue compartments that mediate initial morphogenesis and differentiation in the olfactory primordia. Thus, there may be at least two molecularly distinct types of cell–cell interactions for olfactory pathway differentiation: the first uses M/E induction to drive initial morphogenesis of the nasal processes as well as neuralization of placodal epithelium. The second results in the patterning of the mesenchyme and forebrain in register with ORN axon growth and additional cell migration. This later set of interactions facilitates the proper growth of ORN axons from the periphery to their forebrain targets during the initial formation of the primary olfactory pathway.


Precursor or stem cells with the capacity to generate new ORNs may be established initially in the olfactory primordia and maintained throughout life to mediate the ongoing turnover of ORNs in the mature OE. Thus, the sequence of cellular interactions and differentiation established by M/E induction may be preserved and repeated throughout life. It is also possible that only a subset of critical steps of this inductive mechanism are recapitulated in less well-defined neural stem cells in the mature OE to facilitate de novo differentiation of ORN precursors and their subsequent maturation to functional ORNs. Alternately, the generation of ORNs in the adult may use a completely independent mechanism; however, this possibility seems unlikely because several transcription factors, neuronal growth associated molecules, and inductive signals are maintained in maturing or adult regenerating olfactory epithelium (Whitesides et al., 1998; Murray and Calof, 1999; Manglapus et al., 2004). Some of these signals, including several BMPs and other transforming growth factor beta family members continue to be expressed in the mesenchyme or OE, at least through late prenatal life (Shou et al., 2000; Wu et al., 2003). Similarly, several molecules associated with retinoid signaling, including RA receptors and binding proteins, are also expressed in either the frontonasal mesenchyme or olfactory epithelium at later prenatal stages, once the initial population of ORNs is established (Whitesides et al., 1998). In addition, RA signaling by means of the DR5 RA response element (RARE) continues throughout postnatal development and in the adult OE in cells with characteristics of maturing olfactory receptor neurons (Whitesides et al., 1998). This specific type of RA signaling occurs in at least two rudimentary regions of the nascent olfactory pathway—the placode and the anterior ventral forebrain—and remains in mature regions that retain precursors with neurogenic potential (Whitesides et al., 1998; Thompson Haskell et al., 2002; Haskell and LaMantia, 2005). Furthermore, in the anterior forebrain, we have found that a distinct class of precursor cell, the slowly dividing multipotent SVZ precursor, is specifically RA-activated by means of DR5-RARE–mediated mechanisms (Haskell and LaMantia, 2005). Whereas at most suggestive, these observations are consistent with a continued role for inductive signals, used initially for morphogenetic M/E interactions, in a full or partial recapitulation of mechanisms that specify ORNs or their precursors during ongoing regeneration in the adult olfactory pathway.


The induction of the olfactory pathway relies upon many of the same cellular and molecular mechanisms that guide patterning and morphogenesis in the limbs, heart, and face (and to a lesser extent the ears and eyes). At each site, there are shared imperatives: local morphogenesis of skeletal, muscular, epidermal, and microvascular elements must occur, and afferent or efferent axons must grow to appropriate targets. Nevertheless, a remarkable event occurs in the olfactory primordia: the local generation of sensory receptor neurons with unique morphology, physiology, and molecular identity, including the singular expression of distinct odorant receptor genes (reviewed by Serizawa et al., 2004; Mombaerts, 2001). Our data indicate that molecules shared by the olfactory primordia, limbs, heart, and face may contribute to ORN induction and differentiation. There are most likely additional signals that modify M/E interactions and subsequent developmental events in the placodal epithelium and adjacent mesenchyme that ensure the unique differentiation of ORNs. If nonaxial M/E induction in the olfactory primordia, branchial arches, aortic arches, and limb buds is a singular process, modified based upon anterior–posterior position but nevertheless mechanistically similar, then it is possible that this general mode of interaction has evolved to ensure morphogenesis of bilaterally symmetric appendages along the entire axis of the organism—including noses, and the neurons that define their chemosensory capacity.


John Whitesides provided critical contributions to the early phases of this work. The RA detector mice and cell lines were produced and characterized in collaboration with Elwood Linney. Tom Maynard produced all of the schematic drawing used in the figures of this review. Eric Tucker reviewed in detail several drafts of the manuscript, Dan Meechan and Amanda Peters provided helpful comments, and Josh Smith assisted with proofreading and referencing. This work has been supported over the past decade by NICHD (HD029178 to A.-S.L.) as well as The Sloan Foundation, The March of Dimes, and The National Down Syndrome Society.