Maturation of the Olfactory Sensory Neuron and Its Cilia

Abstract

Olfactory sensory neurons (OSNs) are bipolar neurons, unusual because they turn over continuously and have a multiciliated dendrite. The extensive changes in gene expression accompanying OSN differentiation in mice are largely known, especially the transcriptional regulators responsible for altering gene expression, revealing much about how differentiation proceeds. Basal progenitor cells of the olfactory epithelium transition into nascent OSNs marked by Cxcr4 expression and the initial extension of basal and apical neurites. Nascent OSNs become immature OSNs within 24–48 h. Immature OSN differentiation requires about a week and at least 2 stages. Early-stage immature OSNs initiate expression of genes encoding key transcriptional regulators and structural proteins necessary for further neuritogenesis. Late-stage immature OSNs begin expressing genes encoding proteins important for energy production and neuronal homeostasis that carry over into mature OSNs. The transition to maturity depends on massive expression of one allele of one odorant receptor gene, and this results in expression of the last 8% of genes expressed by mature OSNs. Many of these genes encode proteins necessary for mature function of axons and synapses or for completing the elaboration of non-motile cilia, which began extending from the newly formed dendritic knobs of immature OSNs. The cilia from adjoining OSNs form a meshwork in the olfactory mucus and are the site of olfactory transduction. Immature OSNs also have a primary cilium, but its role is unknown, unlike the critical role in proliferation and differentiation played by the primary cilium of the olfactory epithelium’s horizontal basal cell.

The olfactory epithelium and its neural progenitor cells

The major selective force driving the evolution of the mammalian olfactory epithelium is the detection of volatile chemicals, known as odorants. This epithelium evolved an unusual multiciliated neuron to serve as the sensor for odorants. These neurons, called olfactory sensory neurons (OSNs), have dendrites that reach the surface of the epithelium, ending in a knob from which extend multiple non-motile cilia into the mucus that covers the epithelium. At the surface of these cilia is where odorants encounter receptor proteins capable of transducing odorant binding into biochemical events leading to the electrical responses necessary for OSNs to signal to the brain. These OSNs are the most abundant type of cell in the olfactory epithelium. In a juvenile mouse, each side of the nose has about 5.2 million OSNs (Kawagishi et al. 2014; Bressel et al. 2016). The patterns of activation across this population of OSNs are responsible for encoding the identity of each odor encountered. In fact, OSNs appear to have evolved to maximize the encoding of odor identity, fundamental to odor discrimination. Each mature OSN expresses not only just one odorant receptor gene, but only one of the 2 alleles of that gene (Chess et al. 1994; Malnic et al. 1999; Mombaerts 2004; Monahan and Lomvardas 2015). Therefore, each OSN responds only to the odorant agonists of a single receptor protein, thereby allowing the OSN inputs to the olfactory bulb to unambiguously represent patterns of receptor response. This organization should maximize the ability of the olfactory system to generate distinct patterns of activity in response to different odors and thereby optimize the animal’s capacity for odor discrimination.

OSNs are also unusual in a second respect. They can be replaced throughout life (Moulton 1974; Graziadei and Graziadei 1979). Across the entire mammalian nervous system, only certain types of interneurons in the olfactory bulb and in the dentate gyrus of the hippocampus share this robust capacity for lifelong replacement (Kaplan and Hinds 1977). Unlike the neural plasticity function served by the addition of new interneurons in the adult brain (Hardy and Saghatelyan 2017; Toda and Gage 2018), OSN turnover probably evolved to cope with the consequences of damage by external agents such as chemicals, microorganisms, and particulate matter. Exposure to these damaging agents is necessary in order to allow OSNs access to odor molecules. In fact, OSNs are likely more exposed to environmental damage than any other type of neuron. The olfactory epithelium’s ability to replace OSNs is the ultimate defense against damage to its neurons. A link between damage and OSN lifespan was established by experiments that reduced access of damaging agents to the olfactory epithelium, resulting in increased OSN lifespan (Hinds et al. 1984; Mackay-Sim and Kittel 1991a, 1991b). A consensus estimate for OSN lifespan has been difficult to achieve, however, not only because housing conditions and levels of damaging agents differ across institutions, but also because OSN lifespan varies tremendously even within individuals. In mice housed in the institutional animal facilities where exposure to damaging agents is minimal, most OSNs survive for 1–3 months, but the lifespan distribution also has a tail that extends out more than a year (Hinds et al. 1984; Kondo et al. 2010; Holl 2018).

The continuous turnover of OSNs is possible because the olfactory epithelium contains all of the cell types responsible for adult neurogenesis (Figure 1). These cell types were originally identified by morphology and position in the epithelium (Farbman 1992), but molecular markers specific to each cell type are now the standards used to unambiguously identify and distinguish olfactory epithelium cell types, as well as some of the cell types of neighboring tissues also found in most preparations of olfactory mucosal tissue (Keller and Margolis 1975; Verhaagen et al. 1989; Schwartz Levey et al. 1991; Chen et al. 1992; Cau et al. 2002; Yu et al. 2005; McIntyre et al. 2010; Packard et al. 2011). Most cell types in the olfactory epithelium have their cell bodies located in layered strata, but because the processes of OSNs and sustentacular cells span the epithelium, it is considered a pseudostratified epithelium. The olfactory epithelium lies apical to the lamina propria, which separates the olfactory epithelium from the bones of the septum and olfactory turbinates, contains the blood vessels serving the epithelium, the converging fascicles of the olfactory nerve, the olfactory ensheathing cells that enwrap these fascicles, and other more poorly studied cell types. Most preparations of olfactory epithelium also contain this lamina propria, as well as neighboring respiratory epithelium in many cases, and for these reasons, such preparations are properly called olfactory mucosa.

The most basally located cell bodies in the olfactory epithelium are horizontal basal cells (HBCs). These cells are rich in tonofilaments and directly contact the basal lamina that defines the boundary between the olfactory epithelium proper and the underlying lamina propria (Holbrook et al. 1995). HBCs can be identified by expression of keratins Krt5 and Krt14 (Schwartz Levey et al. 1991; Packard et al. 2011). Quiescent HBCs, which are the majority of HBCs under normal conditions, can be identified by expression of Trp63, which encodes a transcription factor that along with Wnt signaling represses mitosis and the ability of HBCs to differentiate (Fletcher et al. 2011, 2017; Packard et al. 2011). Mitotically quiescent HBCs serve as a reserve population of tissue-level stem cells. Under normal conditions, HBCs only contribute to a fraction of the production of new OSNs. Most OSNs originate from a population of multipotent globose basal cells (GBCs), but if the epithelium is so badly damaged that GBCs are lost, some HBCs become more active and can regenerate the entire epithelium (Huard and Schwob 1995; Iwai et al. 2008; Leung et al. 2007; Gadye et al. 2017).

At least 3 subtypes of GBCs, which lie apical to the HBCs, are known and all contain Sec8, an exocyst complex protein not expressed by other cell types in the olfactory epithelium (Joiner et al. 2015). As mentioned above, multipotent GBCs are most often the source of newly differentiating cells in the olfactory epithelium, especially neurons, microvillar cells, sustentacular cells, and Bowman’s gland cells (Goldstein and Schwob 1996; Chen et al. 2004; Fletcher et al. 2017). Sustentacular cells can also arise directly from HBCs and this may be their most common source (Fletcher et al. 2017). Sustentacular cells are capable of mitosis sufficient to keep pace with the expansion of the olfactory epithelium with age but any additional need for sustentacular cells, such as after damage to the olfactory epithelium, appears to be due to de novo production from basal cells (Weiler and Farbman 1997). Multipotent GBCs expressing both Sox2 and Pax6 include both quiescent cells and activated cells that give rise to transit-amplifying forms of GBCs (Schwob et al. 2017). The neurally fated GBCs produced from multipotent GBCs express Ascl1 (Mash1), and they divide to produce an immediate neuronal precursor type of GBC marked by expression of Neurog1 (Cau et al. 2002). As implied by their name, immediate neuronal precursors give rise to OSNs.

The first recognizable neuron produced from GBCs is the nascent OSN, marked by expression of Cxcr4 and the initial extension of a basal neurite and an apical neurite (Figure 2A) (McIntyre et al. 2010). Nascent OSNs rapidly differentiate into immature OSNs, whose canonical marker is Gap43 (Figure 2B) (Verhaagen et al. 1989). With some minor differences depending on age, immature OSNs require about a week to differentiate into mature OSNs (Rodriguez-Gil et al. 2015; Liberia et al. 2019), whose canonical marker is Omp (Figure 2C) (Keller and Margolis 1975). This review focuses on the postmitotic events that transform nascent OSNs into mature OSNs. Excellent reviews emphasizing the biology of the basal cells of the olfactory epithelium can be found elsewhere (Calof et al. 2002; Schwob 2002; Schwob et al. 2017).

Nascent OSNs represent the initial transition into a neuronal phenotype

As might be expected of newly born neurons, nascent OSNs are identifiable by expression of genes whose protein products are associated with the initiation of neurite extension, specifically Cxcr4 and Dbn1 (McIntyre et al. 2010). In situ hybridization for the mRNAs of these genes identifies a thin layer of cells lying just apical to the basal cells (Figure 2A). Half of Cxcr4⁺ nascent OSNs express neither the canonical marker of immediate neuronal precursors, Neurog1, nor the canonical marker of immature OSNs, Gap43. Cxcr4 is best known for its role as a coreceptor for HIV, but it is also an important mediator of the initiation of neurite growth and branching, especially in axons (Pujol et al. 2005). Consistent with this, Cxcr4⁺ nascent OSNs have a basal neurite extending into the lamina propria and an apical neurite that often has not reached the epithelial surface (Figure 3A). These features identify a population of cells that have neurites but are not yet identifiable as immature OSNs. Because more than half of these nascent OSNs do not yet express canonical markers of more differentiated OSNs, recognition of these cells as neurons has been slow. For example, a recent single-cell RNA-seq analysis of cell lineages in the olfactory epithelium (Fletcher et al. 2017) describes these cells as immediate neuronal precursors (Figure 3B,C), but because Cxcr4⁺ cells extend neurites and therefore have a neuronal morphology, they must be considered neurons.

Single-cell RNA-seq analysis has provided additional evidence about the phenotype of nascent OSNs (Hanchate et al. 2015; Fletcher et al. 2017). Two clusters of cells, which Fletcher et al. (2017) labeled as immediate neuronal precursor 2 (INP2) and INP3 (Figure 3B,C), are enriched in Cxcr4 transcripts. INP2 cells are equivalent to Cxcr4⁺ nascent OSNs that lack detectable expression of Gap43, whereas INP3 cells are equivalent to nascent OSNs coexpressing Cxcr4 and Gap43, as these cells transition into immature OSNs. Functional bioinformatics of the mRNAs that distinguish INP2 cells identifies overrepresented annotations for neural development, microtubules, cytoskeleton, transcriptional regulation, and synapses. The same annotation categories, with the exception of transcriptional regulation, are overrepresented among the mRNAs that distinguish INP3 cells, but many of the mRNAs differ, indicative of progress toward a more differentiated state. These findings of gene expression characteristic of neural development coupled with the immunodetection of Cxcr4⁺ neurites suggest that nascent OSNs are a key transition during OSN differentiation where a shift in transcriptional regulation leads to the initial steps in the acquisition of a neuronal phenotype.

Nascent OSNs must be capable of rapidly becoming immature OSNs given that proliferating basal cells labeled during DNA replication at mitosis require as little as 24 h to become Gap43⁺ immature OSNs (Coleman et al. 2017; Liberia et al. 2019; Savya et al. 2019). This conclusion is supported by evidence that a fraction of Cxcr4⁺ OSNs coexpress Gap43 and that sections of olfactory bulb show no detectable Cxcr4 immunoreactivity in the olfactory nerve layer—arguing that OSN axons do not reach the bulb until after OSNs have reached the immature stage (McIntyre et al. 2010). Taken together, these data suggest that nascent OSNs become immature OSNs within a day or two of beginning to extend neurites.

Very little else is yet known about nascent OSNs. The main endogenous agonist for Cxcr4, a G-protein-coupled receptor (GPCR), is Cxcl12 (also known as SDF-1). Cxcl12 is expressed by as yet unidentified cells in the lamina propria at birth and in the underlying bone at age postnatal day 21 (McIntyre et al. 2010), suggesting that nascent OSNs might be cueing off a basal-apical gradient of Cxcl12. The dependence of newborn OSNs on Bmp4 (Shou et al. 2000) suggests the hypothesis that nascent OSNs may depend on Bmp4 to trigger or sustain their differentiation. Expression pattern data (Nickell et al. 2012) predict that Col9a2, a collagen, and Gng2, a heterotrimeric G-protein subunit, are additional markers of nascent OSNs. These predictions are confirmed by single-cell RNA-seq data (Fletcher et al. 2017), but why Col9a2 and Gng2 might be important to newly born OSNs is as yet unknown.

Immature OSNs undergo a week-long maturation process

Newborn OSNs require about a week to become Omp⁺ mature OSNs and most of this period is spent as immature OSNs (Miragall and Monti Graziadei 1982; Kondo et al. 2010; Kikuta et al. 2015; Rodriguez-Gil et al. 2015; Liberia et al. 2019; Savya et al. 2019). The time from birthdate marking of dividing cells to the appearance of the first new mature OSNs is 6 days at age postnatal day 7 and 7–8 days at 3–4 weeks of age (Kondo et al. 2010; Rodriguez-Gil et al. 2015; Liberia et al. 2019). The period that an OSN spends in the immature stage is necessary to transition from a neurally fated progenitor cell to a neuron whose dendrite ends in a knob from which projects non-motile cilia at the apical surface of the epithelium and whose axon has formed mature glutamatergic synapses with targets in the olfactory bulb. The changes that occur in Gap43⁺ immature OSNs are so extensive that the transcriptome of newly formed immature OSNs is distinct from that of late-stage immature OSNs (Heron et al. 2013; Hanchate et al. 2015). This was first discovered by expression profiling of the olfactory epithelium at 5 days after unilateral removal of an olfactory bulb, a time point when nearly all mature OSNs have been lost and increased basal cell proliferation has begun producing new immature OSNs in the ipsilateral olfactory epithelium (Heron et al. 2013). The transcripts previously identified in immature OSNs (Nickell et al. 2012) were separated into 2 distinct populations by this experiment. Immature OSN transcripts associated with neuritogenesis and regulation of transcription increased in abundance, whereas immature OSN transcripts associated with neuronal homeostasis functions such as energy production and protein catabolism were decreased (Figure 4). The former events are a continuation of the extensive molecular changes needed to convert a neurally fated progenitor cell into a neuron, whereas the latter events represent the expression of proteins needed for functions that carry over into mature OSNs.

The phenotypic transitions that occur as OSNs differentiate are driven by widespread changes in gene expression. The number of genes whose transcripts have been detected by transcriptome profiling of samples of olfactory mucosa ranges from 15 409 to 17 983 (Nickell et al. 2012; Ibarra-Soria et al. 2014). Of these, 1750 encode proteins whose functional annotations (ontologies) include a role in transcription via RNA polymerase II (Supplementary Table S1). Using the changes in transcript abundance that occur after olfactory bulbectomy to compare against transcript abundance measured in samples enriched in mature OSNs or immature OSNs (Nickell et al. 2012; Heron et al. 2013; Saraiva et al. 2015), these 1750 genes can be divided into groups associated with stages in the OSN lineage. They include transcripts from 133 genes that increase after bulbectomy because they are expressed primarily in cell types other than OSNs (Supplementary Table S2). These transcripts should mostly consist of transcription factors expressed in neurally fated GBCs responding to the loss of mature OSNs. Consistent with this prediction, the 133 genes include Ascl1, Foxg1, Neurog1, Neurod1, Runx1, Six1—all encoding transcription factors expressed in GBCs and known to be critical for the production of OSNs (Cau et al. 2002; Theriault et al. 2005; Ikeda et al. 2007; Duggan et al. 2008; Kawauchi et al. 2009; Packard et al. 2011). In addition, the 133 genes include 10 of 13 additional transcription factors known to be expressed in GBCs or in both GBCs and sustentacular cells: Dlx3, Id1, Isl1, Msx1, Pax6, Pitx1, Six4, Sox2, Tead2, and Tgif1 (Shetty et al. 2005; Chen et al. 2009; Guo et al. 2010; Parrilla et al. 2016). However, the nearly complete absence of the most abundant cell type in the olfactory epithelium, mature OSNs, at 5 days after bulbectomy predicts that increased proportions of other cell types in these preparations could cause transcripts specific to these cell types to be included among these 133 transcriptional regulator mRNAs, irrespective of their importance to olfactory neurogenesis. This prediction is not consistently true, however. The 133 mRNAs do not include 3 homeobox transcription factors specific to cells in the lamina propria (Alx1, Alx3, and Pax3), or 2 homeobox transcription factors specific to Bowman’s glands or ducts (Dmbx1 and Msx2), or the microvillar cell transcription factor Pou2f3, but they do include 2 homeobox transcription factors specific to sustentacular cells (Epas1 and Meis2), the HBC transcription factor Trp63, and Ascl3, a transcription factor expressed by activated HBCs and progenitors of microvillar cells and Bowman’s gland cells (Shetty et al. 2005; Chen et al. 2009; Guo et al. 2010; Parrilla et al. 2016). These data indicate that only a small proportion of these 133 transcriptional regulators are expressed primarily by cells outside the OSN cell lineage. The majority are instead expressed by cells in the OSN cell lineage, which predicts many of them have a role in driving the changes in gene expression necessary to produce nascent OSNs from GBCs.

Eight hundred and seven of the 1750 annotated transcriptional regulators detected in samples of olfactory mucosa are expressed predominantly in immature OSNs (Supplementary Table S3). They show a gradient of response to olfactory bulbectomy that is inversely correlated with their enrichment in samples of Omp⁺ mature OSNs (Figure 5). Of these 807 transcripts, 74 increase in response to olfactory bulbectomy (Supplementary Table S4 and set A in Figure 5). This pattern identifies these 74 transcriptional regulators as candidates to help drive the changes in gene expression responsible for the early phase of immature OSN differentiation. Very few of the proteins encoded by these 74 transcripts have been investigated. Of those genes whose expression has been tested, what emerges is a consistent pattern of expression in GBCs and immature OSNs, with some also expressed in sustentacular cells. This is true for Bach2, Hes6, Mycn, Satb1, Zfhx3, and Dlx5 (Levi et al. 2003; Long et al. 2003; Suzuki et al. 2003; Gussing and Bohm 2004; Shetty et al. 2005; Parrilla et al. 2016). Direct evidence that these 74 transcriptional regulators are important for the production of immature OSNs is lacking, with one exception. The absence of Dlx5 results in deficient production of OSNs, and the OSNs that are able to develop have axons that fail to reach the olfactory bulb (Levi et al. 2003; Long et al. 2003).

The majority of transcriptional regulators whose transcripts are abundant in immature OSNs continue to be expressed in mature OSNs. These 500 transcripts (Supplementary Table S5 and set B in Figure 5) are detectable in samples enriched for immature OSNs, but their abundance tends to be unaffected at 5 days after bulbectomy, evidence of greater abundance in immature OSNs compared with GBCs and mature OSNs (Heron et al. 2013). The enrichment of these transcripts in immature OSNs has been confirmed for Nhlh1, Sox11, Hopx, Pou6f1, Lhx2, Tshz1, Cux1, Pbx2, Zhx1, Zhx2, and Zhx3 (Shetty et al. 2005; Parrilla et al. 2016). Tests of the functions of their encoded proteins in OSNs have been done for a few of these 500 genes. Bcl11b is expressed by OSNs expressing Class II odorant receptors, strongly in such OSNs in the ventral domain of the olfactory epithelium and more weakly in the dorsal domain. Bcl11b encodes a repressor of an enhancer responsible for driving expression of Class I odorant receptors, thereby playing an important role in restricting Class I odorant receptor expression to the dorsal olfactory epithelium (Enomoto et al. 2019). Bptf has been implicated in the control of OR enhancer networks that drive expression of OR genes (Markenscoff-Papadimitriou et al. 2014). Lhx2 is critical for the highly elevated expression of a single odorant receptor allele (Zhang et al. 2016; Monahan et al. 2019) and the consequent final maturation of OSNs, discussed below. In addition, the critical transcription factor controlling expression of genes encoding the enzymes of the cholesterol synthesis pathway, Srebf2, is among these 500 transcriptional regulators (Shimano 2002; Tarr and Edwards 2008). Cholesterol is important for neurite outgrowth, but its synthesis is energetically expensive. In the developing brain, cholesterol synthesis occurs in immature neurons, but not in mature neurons, which obtain their cholesterol from lipoprotein particles produced by neighboring cells, especially astrocytes (Lopes-Cardozo et al. 1986; Saito et al. 1987; Boyles et al. 1989; Hayashi et al. 2004; Ko et al. 2005). The olfactory epithelium appears to behave in the same way, as expression of Srebf2 and cholesterol synthesis genes is lost when OSNs mature (Nickell et al. 2012; Heron et al. 2013).

Only 81 of the transcriptional regulators identified as enriched in immature OSNs decrease in response to bulbectomy (Heron et al. 2013). Transcripts that behave in this way do so because they are also expressed by mature OSNs (Supplementary Table S6 and set C in Figure 5). Even if they are more abundant in individual immature OSNs than in individual mature OSNs, they can show an overall decrease when mature OSNs are lost because mature OSNs greatly outnumber immature OSNs in adult olfactory epithelia. Evidence consistent with this interpretation is found in the expression patterns of Emx2, Pbx3, Pknox2, Tshz2, Tshz3, and Zbtb7b (McIntyre et al. 2008; Michaloski et al. 2011; Parrilla et al. 2016). The expression patterns of these transcriptional regulators predict that they have roles in processes important for both immature and mature OSNs. What little is known about them is consistent with this prediction. Zbtb7b may be part of a repressor complex at the promoter elements of some odorant receptor genes (Michaloski et al. 2006). Emx2 is important for the differentiation of OSNs and has a minor role—as yet poorly defined—in odorant receptor gene expression (McIntyre et al. 2008; Zhang et al. 2016).

In immature OSNs, both neurites develop to almost their mature states. Immature OSN dendrites reach the apical surface and form a dendritic knob with abbreviated cilia (Menco and Farbman 1985b; Schwob et al. 1992). The full extension of cilia and the final maturation of synapses are accomplished after OSNs become mature. The dendrites of immature OSNs are not yet enwrapped by sustentacular cells (Liang 2020). However, the need for the newly forming neurites to interact with neighboring cells is met by increased expression of focal adhesion proteins in immature OSNs (Chacón and Fazzari 2011; Chacón et al. 2012; Heron et al. 2013). Immature OSN axons exuberantly form synapses in olfactory bulb glomeruli and only later after the OSN transition to maturity are these scaled back (Marcucci et al. 2011). OSN synapse formation depends partly on the copper transporter, Atp7a, and OSN survival is decreased when Atp7a is absent (El Meskini et al. 2007). In what may be a related process, the survival of immature OSNs depends on the prion protein, probably due to impaired formation of synaptic connections (Parrie et al. 2018).

OSN differentiation is largely an intrinsic process

Although the rate of production of OSNs from basal cells is sensitive to the loss of mature OSNs, the 6- to 8-day time period required for a newly born OSN to complete differentiation does not appear to be sensitive to manipulations that increase the need for more OSNs (Miragall and Monti Graziadei 1982; Kondo et al. 2010; Kikuta et al. 2015; Rodriguez-Gil et al. 2015; Liberia et al. 2019; Savya et al. 2019). The timing of appearance of mature OSNs after manipulations that evoke synchronous replacement of OSNs—such as ablation of one olfactory bulb—requires more than a week (Verhaagen et al. 1990; Schwob et al. 1992). Similarly, the time required for newly born OSNs to reach maturity in phenotypically normal olfactory epithelia in mice at age 3–4 weeks is 7–8 days (Kondo et al. 2010; Liberia et al. 2019). This consistency in the period required for progression from newborn OSN to mature OSNs even in the face of the loss of nearly all mature OSNs argues that factors external to OSNs have only a small effect on the speed of OSN differentiation. We hypothesize that once nascent OSNs are formed their differentiation follows an intrinsic program. However, the period an OSN spends at the immature stage can be lengthened by interruptions in the processes that drive the transition to maturity, discussed in the next section.

The characteristic developmental progression of expression and action of transcriptional regulators described above is the major force driving the intrinsic program of OSN differentiation. The actions of these proteins should drive a similarly characteristic program of expression of structural genes in OSNs. This hypothesis has been best explored in an assessment of the appearance of presynaptic proteins and their mRNAs in the OSN cell lineage (Marcucci et al. 2009). Some of them, such as piccolo, bassoon, Munc18-1, and syntaxin 1A, begin to be detectable in neurally fated GBCs and continue to be expressed in all cell types further along the lineage. These proteins share a common function of managing the formation and dynamics of synaptic vesicles. Several additional presynaptic proteins and their mRNAs become detectable in immature OSNs, such as α-neurexins, synapsin 1, Vamp2, Snap25, synaptotagmin 1, and synaptophysin. These proteins promote synapse formation, help assemble the secretory apparatus, and support exocytosis in presynaptic terminals. Finally, the neurotransmitter transporter necessary for OSNs to employ glutamatergic synaptic transmission, Slc17a6 (Vglut2), becomes expressed in mature OSNs. This sequence of events is not dependent on OSN axons forming synapses in the olfactory bulb, nor does it depend on the presence of adenylyl cyclase III (Adcy3) and normal olfactory transduction, though the convergence of OSN axons and their paths to their glomerular targets is aberrant in OSNs lacking Adcy3 (Zou et al. 2007; Marcucci et al. 2009). This analysis of the dynamics of synaptic maturation provides evidence that OSN differentiation does not depend on trophic support from the olfactory bulb, unlike OSN survival (Schwob et al. 1992).

Transition to maturity depends on the singular expression of one odorant receptor

The critical change driving the transition from immature OSN to mature OSN is the exceptionally strong expression of one odorant receptor gene locus and continued repression of all other odorant receptor loci. Repression via the formation of heterochromatin at odorant receptor gene loci begins early in the OSN lineage at basal cells and increases as cells differentiate further along the OSN cell lineage (Monahan et al. 2019). Immature OSNs contain low levels of one or more odorant receptor transcripts, but mature OSNs have very high levels of just one odorant receptor transcript (Hanchate et al. 2015; Saraiva et al. 2015). This shift occurs when Lhx2 and Ldb1 seed interchromosomal interactions between enhancers of odorant receptor genes, a rare event that forms a super-enhancer capable of greatly increasing expression of an odorant receptor gene locus (Monahan et al. 2019). When an immature OSN begins to greatly increase expression of an odorant receptor the accumulation of odorant receptor protein triggers an ER stress signaling pathway, resulting in expression of Adcy3 and decreased expression of the histone demethylase Kdm1a (Lyons et al. 2013). Without Kdm1a to relieve epigenetic repression of odorant receptor gene loci, an OSN can no longer rescue odorant receptor gene loci from repression, thereby preventing expression from other odorant receptor loci and in effect locking in expression of the one odorant receptor allele whose increased expression triggered these events.

When the events driving the massive expression of a single odorant receptor are interrupted, OSNs fail to reach maturity. For example, Kdm1a is required for OSN maturation (Coleman et al. 2017) and conditional deletion of the homeodomain transcription factor Lhx2 in immature OSNs interferes with odorant receptor expression (Zhang et al. 2016; Monahan et al. 2019), resulting in delayed OSN maturation and buildup of immature OSNs (Zhang et al. 2016). The latter data extend evidence of stronger effects of germline deletion of Lhx2, which causes the olfactory epithelium to have great difficulty producing mature OSNs (Hirota and Mombaerts 2004; Kolterud et al. 2004; Hirota et al. 2007; Berghard et al. 2012). Conditional deletion of Adcy3, which encodes an enzyme important for preventing release of repression of additional odorant receptor genes (Lyons et al. 2013), also causes a buildup of immature OSNs (Zhang et al. 2017). Similar effects occur when the function of the Ebf family of transcription factors is altered. Ebf transcription factor binding sites are a canonical feature of the enhancers and promoters of odorant receptor genes and other (relatively) olfactory-specific genes (Monahan et al. 2019), and these transcription factors are known to support expression of genes important for OSN maturation (Wang et al. 1997, 2004). Zfp423/Roaz, which inhibits dimerization of the Ebf transcription factors and disrupts target gene activation, is normally found only in immature OSNs, but if its expression is prolonged OSN differentiation is substantially impaired (Tsai et al. 1994; Tsai and Reed 1997; Roby et al. 2012). Together, these findings are evidence that achieving strong expression of a single odorant receptor gene locus is a critical trigger for the transition to a mature OSN phenotype.

External factors might also help control the transition to maturity, but evidence for this is limited. One external factor involved is leukemia inhibitory factor, a cytokine that regulates the transition to maturity via PI3K, Bcl2, and Stat3 (Moon et al. 2002, 2009). Whether other local paracrine factors help to control the transition to maturity is as yet unknown. However, a few proteins intrinsic to OSNs and important for OSN maturation have been identified. Omp is necessary for the final maturation of electrophysiological response properties of OSNs and the increased odorant selectivity of OSNs that accompanies final maturation (Lee et al. 2011). The transcriptional regulator Mecp2 is involved in the transition—or perhaps in the earlier phases of the differentiation of immature OSNs—because its absence causes a transient delay in the final maturation of OSNs (Matarazzo et al. 2004).

The processes described in this section that drive the transition from immature to mature OSN represent dynamic changes that ride on top of a much larger continuity of phenotype that carries over from immature OSNs to mature OSNs, as mentioned in the previous section (Figure 4). At least 6761 genes are expressed by both immature and mature OSNs (Nickell et al. 2012). The functions significantly overrepresented among the proteins encoded by these genes include energy production, axonal growth, and protein metabolism (Heron et al. 2013). By the time an immature OSN transitions into a mature OSN, the vast majority of the OSN phenotype is established. Only a relatively small set of additional genes comprising less than 8% of the genes expressed by mature OSNs (not including odorant receptors) become expressed during or after the transition to maturity (Nickell et al. 2012).

What happens when an immature OSN fails to successfully transition to maturity? The hypothesis that these OSNs suffer apoptosis is supported by the substantial increases in OSN apoptosis observed in mouse mutants that fail to produce normal numbers of mature OSNs. For example, selective deletion of Lhx2 in immature OSNs causes a buildup of immature OSNs and increased OSN apoptosis (Zhang et al. 2016), as does deletion of Adcy3 (Zhang et al. 2017). Similarly, mice with a germline deletion of Emx2 also show increased OSN apoptosis (McIntyre et al. 2008). Even in phenotypically normal mice, the transition from immature to mature appears to be associated with increased apoptosis because the peak location in the olfactory epithelium for OSN cell bodies immunoreactive for active caspase 3, a marker of apoptosis, is at the junction between the immature and mature OSN cell body layers in mice at postnatal day 26 (Figure 6). A second peak occurs at the junction between the basal cell and immature OSN cell body layers, another location where cells are making substantial phenotypic changes. Other studies show similar evidence of peaks of apoptosis in the basal third of the olfactory epithelium (Savya et al. 2019). However, failure to make synaptic connections may also be a significant cause of apoptosis for OSNs just reaching maturity. In bulbectomized rats, newly produced OSNs survive only 5–14 days due to absence of trophic support from the olfactory bulb (Schwob et al. 1992), and this same temporal profile occurs for newborn OSNs in phenotypically normal mice in institutional housing conditions. Once mice reach adulthood and the growth and expansion of the olfactory epithelium is complete, the number of newborn OSNs achieving maturity and surviving decreases (Kondo et al. 2010). This pattern of OSN apoptosis may be caused by competition for synaptic connections in the olfactory bulb, a mechanism related to activity-dependent survival of OSNs (Zou et al. 2004). This pruning of excess OSNs due to failure to capture trophic support from the olfactory bulb would seem to be a more likely cause of early apoptosis of OSNs than failed maturation under normal conditions, but further experiments are necessary to distinguish these possible causes.

Mature OSNs, refining synaptic transmission, and the final maturation of cilia

Once an OSN becomes mature, its lifespan can be quite variable and depends on environmental conditions, as mentioned above (Hinds et al. 1984; Mackay-Sim and Kittel 1991a, 1991b). Early studies estimated OSN lifespan at about a month in rodents (Moulton 1974; Graziadei and Graziadei 1979) but more recent work demonstrates that some OSNs live for months, even more than a year in rare instances (Kondo et al. 2010; Holl 2018).

Mature OSNs express nearly 10 000 genes, according to early microarray data from purified Omp⁺ OSNs (Sammeta et al. 2007; Nickell et al. 2012), and more recent RNA-seq of samples enriched for Omp⁺ OSNs gives a similarly large number of genes expressed by mature OSNs (Saraiva et al. 2015). A more refined estimate using criteria of normalized RNA-seq counts of >5 in these Omp⁺ OSN samples and enrichment in Omp⁺ OSN samples compared with whole olfactory mucosa (fold enrichment > 1) identifies 9941 genes expressed in mature OSNs, and this includes the vast majority of the odorant receptors (Supplementary Table S7).

The features that distinguish mature OSNs from immature OSNs can be deduced from the functions of the products of the 691 genes (excluding odorant receptors) specifically expressed by mature OSNs (Nickell et al. 2012). These proteins are largely involved in establishing the mature structure and function of the 2 ends of the OSN—its cilia and its synapses (Figure 4). Only in mature OSNs is the full complement of cilia-related genes expressed, consistent with the growth of cilia to their final, mature lengths. This includes several genes whose encoded proteins are involved in olfactory signal transduction, such as Gnal, Adcy3, Cnga2, and Omp (Nickell et al. 2012). In fact, specific expression in mature OSNs was an important criterion in the identification of genes involved in olfactory transduction, discoveries that have been fundamental to our understanding of odor detection (Jones and Reed 1989; Bakalyar and Reed 1990; Dhallan et al. 1990). However, these olfactory transduction proteins and the broader set of genes whose expression is needed to complete the maturation of cilia constitute only a fraction of the genes whose expression is needed for the final maturation of OSN axons and synapses. Most of the mRNAs specific to mature OSNs encode proteins destined for the axon and its synapses. These include proteins involved in ion flux, electrical responses, and glutamatergic synaptic transmission (Heron et al. 2013). The ability to detect odors, transduce detection into a receptor potential, and efficiently encode the receptor potential in a train of action potentials that evoke glutamatergic neurotransmission to postsynaptic neurons in the olfactory bulb are the final steps in OSN maturation.

Even though OSN axons form synapses in the olfactory bulb in the immature stage, the transition to maturity does not result in loss of expression of all axon guidance genes. Although many cell adhesion molecules and receptors for axon guidance cues do not continue to be expressed in mature OSNs and a set of more than 10 intracellular signaling proteins that control behavior of growth cones are specific to immature OSNs, several receptors for inhibitory guidance cues are found in mature OSNs and some, such as Plnxa1, Plxna4, Nrp2, and Efna5, are detected in mature, but not immature, OSNs (McIntyre et al. 2010). The demonstration that mature OSN synapses undergo activity-dependent remodeling well into adulthood provides a likely explanation for the continued expression of guidance cue receptors in mature OSNs (Cheetham et al. 2016).

The pseudostratified organization of the olfactory epithelium separates the stages of the OSN cell lineage in such a way that cells beginning to assume a neural fate are located basally and the cell bodies of the most differentiated cells in the lineage, mature OSNs, are located apically. Because the mature OSN layer is several cell bodies deep, this organization might indicate that a basal to apical age gradient exists within the mature OSN cell body layer. However, although new mature OSN cell bodies enter the mature OSN layer from the basal end, no proof exists that mature OSN cell bodies progress apically as they age. In fact, the dynamics of mature OSN cell body location are unknown. Gradients of labeling within the mature OSN cell body layer for 2 transcripts, Umodl1 (also known as N8) and Kcnc4, have been detected and could reflect OSN age (Yu et al. 2005; Sammeta et al. 2010), but the appearance of stronger labeling in the most apical of the mature OSN cell bodies could instead be a response to position or life history (damage or amount of activity) rather than age.

The role of cilia in OSN function and olfactory epithelium maturation

The use of cilia by OSNs as surfaces for detecting odorants is a unique example of the multitude of functions performed by cilia. Cilia are organelles that can be found projecting from the surface of numerous mammalian cell types (Wheatley et al. 1996). Cilia perform many roles, including propulsion and fluid movement, regulating cell homeostasis and cell cycle, and cellular signaling and detection of external stimuli such as growth factors, light, and odors (Berbari et al. 2009; Goetz and Anderson 2010). Although it has commonly been assumed that ciliary structures and proteins are highly conserved, there is growing evidence showing that differences exist between species and between cilia types within one organism. We gained much of our knowledge of basic ciliary biology from invertebrates (i.e., Caenorhabditis elegans, Chlamydomonas) and mammalian primary cilium, which can be notably different from OSN cilia, as it relates to ciliary structure, function, and protein composition. Cilia are often classified based on different criteria; for example, they can be categorized structurally (microtubule organization 9 + 2 vs. 9 + 0), functionally (motile vs. non-motile), or numerically (primary vs. multiciliate) (Satir and Christensen 2007; Jenkins et al. 2009; Takeda and Narita 2012). OSNs cilia are unique hybrids in this categorization, and as such this makes OSNs fertile ground for investigating the unique properties of this organelle. Even though our understanding is still evolving, in many ways, cilia on OSNs are the best characterized of the neuronal cilia. Very little is known about primary cilia on neurons in the brain, and therefore much of the information is extrapolated from the primary cilia literature. We have reviewed olfactory cilia in detail in the past (McEwen et al. 2008; Jenkins et al. 2009; Uytingco et al. 2019), and there exist numerous review articles detailing primary cilia structure and function (Goetz and Anderson 2010; Wheway et al. 2018; Anvarian et al. 2019). Here, we will focus on differences between primary cilia on neurons of the brain and the multiple cilia of mature OSNs.

OSN cilia differ from the typical primary cilia found in most neurons in several different aspects. Electron microscopy studies and en face imaging show that mature OSNs have longer cilia and more cilia than most other ciliated cells (Figure 7). OSNs project from their dendritic knobs approximately 20–35 cilia of up to ~100 µm in length, forming a meshwork in the mucus layer that covers the surface of the olfactory epithelium (Jenkins et al. 2009; Williams et al. 2014). These projections of cilia into the nasal cavity make OSNs the only neurons in direct contact with the external environment. The axoneme of OSN cilia is arranged in a (9 + 2) configuration with 9 pairs of microtubule doublets concentric to a single central pair of microtubules (Menco 1984). This microtubule configuration is normally found in motile cilia but OSN cilia lack the dynein arms necessary for movement and are thus rendered immotile (Menco 1984). Also unique to OSNs, the 9 + 2 configuration exists only in the proximal segment (~1–2 µm) and then in short distance tapers to microtubule singlets as the cilia extend to their full length (on average, ~20 µm in length) (Williams et al. 2014). In contrast, the primary cilium on most neurons in the brain is a solitary organelle that projects from the cell body into the interstitial space. It is relatively short (1–10 µm) and has a 9 + 0 configuration over nearly its full length, breaking down only at the tip (Satir and Christensen 2006).

There is some controversy about whether olfactory cilia length is uniform across the different regions of the olfactory epithelium. By immunostaining in fixed tissue, Challis et al. (2015) showed that within the olfactory septum, olfactory cilia vary in length depending on the OSN location. However, using a different approach in a live oxygenated tissue preparation, Williams et al. (2017) showed that olfactory cilia have uniform OSN cilium lengths and numbers across various turbinate surfaces of the olfactory epithelium, which is consistent with previous reports (Menco 1997, 1980; Strotmann et al. 2004). Also, mixed reports from one research group showing that the absence of Adcy3, a key protein in the olfactory signal transduction cascade, has region specific effects on cilia length (Challis et al. 2015, 2016) were not substantiated in a recent study by another group utilizing scanning electron microscopy (Zhang et al. 2017). This lack of agreement complicates the interpretation of the role of Adcy3 in ciliogenesis and indicates the nonuniformity of olfactory cilia remains an unanswered question. Given the importance of cilia in providing a receptive field for both odor detection and mechanosensation, these questions about OSN cilia length and the maintenance of cilia warrant further investigation.

The multiple cilia of OSNs extend from basal bodies that are organized in a ring-like fashion around the dendritic knob. In most cell types with primary cilia, including neurons, the basal bodies reside in a ciliary pocket. This is an invaginated membrane domain from where endocytic and exocytic vesicles are formed and which can provide an interface to the actin-cytoskeleton (Molla-Herman et al. 2010; Ghossoub et al. 2011). Although not conclusive, it appears from electron microscopy data that OSN cilia do not have a ciliary pocket (Menco 1988; Benmerah 2013). Similarly, the ciliary rootlet is a cytoskeleton structure found originating from the basal body in many ciliated cells and links the base of the cilium to the cell body, which provides structural support for the cilium (Yang et al. 2002). Although OSNs have been shown to express components of the ciliary rootlet (Yamamoto 1976; McClintock et al. 2008), it is still unknown whether olfactory cilia have a rootlet.

Because cilia lack the necessary machinery for protein synthesis, proteins are transported from the cell body through the transition zone, which is believed to function as a selectivity filter (Chih et al. 2011), to the cilia by an evolutionarily conserved process, termed intraflagellar transport (IFT). IFT moves cargo with IFT-A and IFT-B subcomplexes, which bidirectionally traverse the ciliary axoneme via association with kinesin and dynein motors (Taschner et al. 2012). Studies in C. elegans sensory cilia show that the Kif17 homologue OSM-3 is solely responsible for distal singlet microtubule biogenesis and heterotrimeric kinesin-2 mediates IFT in cooperation with OSM-3 only along the proximal segment (Perkins et al. 1986; Snow et al. 2004). However, proximal and distal axonemes of olfactory cilia show no bias toward IFT kinesin-2 choice and that Kif17 homodimer is dispensable for distal segment IFT (Williams et al. 2014). This highlights unique aspects of how OSN cilia are built and maintained. Interestingly, as the cargo adaptor for IFT, Bardet–Biedl syndrome proteins (BBSome) seem to have a different function in olfactory cilia than primary cilia of neurons in the brain. Lack of ciliary localization of GPCRs (somatostatin receptor type 3 and melanin-concentrating hormone receptor 1) in neurons is observed in mice lacking the Bbs4 gene (Berbari et al. 2008). However, odorant receptors, members of GPCR superfamily, are still be able to properly localize to olfactory cilia of Bbs4^−/− mice (Uytingco et al. 2019), suggesting that olfactory cilia have a unique mechanism in regulating ciliary protein transport. The BBSome is shown to play a role in basal body and transition zone maintenance (Goetz and Anderson 2010; Karmous-Benailly et al. 2005; Uytingco et al. 2019). The organization of the transition zone can differ between species and cilia types within one organism (Gonçalves and Pelletier 2017; Jana et al. 2018). The composition of the transition zone in olfactory cilia largely remains unknown. Several studies have demonstrated that the loss of transition zone proteins (Cep290, Mks1, and Mks3) shows an olfactory phenotype (McEwen et al. 2007; Pluznick et al. 2011). However, the conditional loss of Cc2d2a (Mks6), a core transition zone component in other types of cilia, results in kidney and retinal phenotypes but does not result in olfactory deficits (Lewis et al. 2019), which highlights the unique property of olfactory cilia. The ciliary necklace is a specialized structure within the transition zone that consists of several parallel strands of intramembrane particles (Szymanska and Johnson 2012). Olfactory cilia typically have more strands per cilium than primary cilia or other multiciliated cells such as the respiratory cilia (Menco 1980). The formation of the ciliary necklace occurs early in ciliogenesis as a patch of membrane and in malformed cilia necklace-like structures persist (Menco 1980; Carson et al. 1981). More work is necessary to definitively demonstrate the organization, composition, and function of the transition zone in olfactory cilia.

It is clear that OSN cilia have a proteome distinct from the rest of the cell (Jenkins et al. 2009). As the initial binding site for inhaled odorants, OSN cilia are enriched with olfactory signaling proteins as showed by early immune electron microscopy studies (Asanuma and Nomura 1991; Menco et al. 1992; Menco 1997). Odorant signal transduction is initiated when odorants bind with specific odorant receptors located within these cilia (Lancet and Ben-Arie 1993). Then the receptors subsequently activate olfactory-specific G-protein (Golf) and, in turn, activate adenylate cyclase (Adcy3), resulting in the generation of cyclic AMP (cAMP) (Sklar et al. 1986; Ronnett et al. 1993). The increased cAMP in the cilia opens a cyclic nucleotide-gated (CNG) channel, resulting in an influx of Na⁺ and Ca²⁺ (mostly Ca²⁺), that increases the activity of the calcium-activated chloride channel and depolarizes the OSN (Nakamura and Gold 1987; Firestein and Werblin 1989; Leinders-Zufall et al. 1997, 1998). This olfactory transduction pathway is one of the most well-defined signaling mechanisms that take place in mammalian olfactory cilia. The signaling proteins, including Golf, Adcy3, and CNG channels, appear to preferentially localize to the long distal segment of OSN cilia where OSN can make its first contact with odorants (Menco 1997; Matsuzaki et al. 1999; Flannery et al. 2006; Jenkins et al. 2009), which presumably improves the efficiency of odorant signal transduction. Importantly, mutations in genes associated with the structural integrity or function of cilia can cause defects in odor detection and lead to olfactory dysfunction (Williams et al. 2017; Green et al. 2018; Uytingco et al. 2019), which further confirm that OSN cilia function as the primary site of odorant binding.

Studies of the penetrance of cilia specific diseases, termed ciliopathies, in the olfactory system have provided valuable insight into the identification of several proteins important to the establishment, maintenance, and the function of OSN cilia (Kulaga et al. 2004; McEwen et al. 2007; Tadenev et al. 2011; McIntyre et al. 2012; Williams et al. 2017; Uytingco et al. 2019). In mouse ciliopathy models such as the Oak Ridge Polycystic Kidney mouse, as well as in knockouts of Bbs1, Bbs4, Bbs8, or Ift88, OSN ciliation was either absent or significantly reduced with decreased cilium length and number. In addition, these mice exhibit a loss of olfactory function and significantly reduced olfactory bulb activity. Defects in OSN axon targeting in the olfactory bulb were also observed in Bbs8 knockout and Ift88 OSN knockout mice (Tadenev et al. 2011; Williams et al. 2017; Green et al. 2018; Uytingco et al. 2019). Furthermore, loss of Ift88 in OSNs results in aberrant odor-guided behavior in mice (Green et al. 2018). Importantly, intranasal adenoviral-mediated gene delivery of wild-type Bbs1, Bbs4, and Ift88 to Bbs1^osnKO, Bbs4^KO, and Ift88^osnKO mice restored ciliary morphology and olfactory function, respectively. Interestingly, this recovery of sensory input was sufficient to restore olfactory bulb activity (Williams et al. 2017; Green et al. 2018; Uytingco et al. 2019) and even re-established the odor-guided behaviors in Ift88^osnKO mice (Green et al. 2018). Together, these preclinical proof-of-concept studies show that targeted gene therapy potentially could be used to treat olfactory dysfunction originating from ciliopathies.

Although these studies highlight the necessity for understanding the structure and function of cilia in the olfactory system, these represent only a fraction of the proteins that must be necessary for the formation and function of cilia. Lack of knowledge of the full set of cilia proteins and their potential interactions is a major impediment to understanding cilia. Recognition of this fact that has led to numerous studies using a variety of approaches to identify or predict cilia genes or proteins (Su et al. 2004; Sammeta et al. 2007; Klimmeck et al. 2008; Mayer et al. 2008; McClintock et al. 2008; Mayer et al. 2009; Tadenev et al. 2011). Many of these used in silico bioinformatics methods, including one effort based on the identification of the full set of genes expressed by mature OSNs (McClintock et al. 2008). It mined mature OSN genes for evidence that their transcripts were also unusually abundant in other tissues enriched in cells with cilia or flagella: lung, trachea, and testis. It identified 99 genes, including 18 genes encoding a protein whose function was already known to be important to cilia. Most of the remaining 81 genes encoded proteins that had never been studied directly. This data set has proved to be extremely accurate at predicting importance to cilia according to a recent meta-analysis of predicted cilia-related genes (van Dam et al. 2019). At least 19 more of the 99 genes have proved to be important to cilia, mostly by studies performed on ciliated cells other than OSNs (Table 1).

It is interesting to consider the timing of cilia gene expression and its intersection with the maturation of OSNs (Figure 8). Developing OSNs possess primary cilia during embryogenesis (Menco and Farbman 1985a, 1985b); however, the function of these cilia is currently unknown. The initial stages of ciliogenesis occur by embryonic day 11 (E11) and coincide with morphological changes to OSNs, including the formation of the dendritic knob (Cuschieri and Bannister 1975; Jenkins et al. 2009). Centriole duplication in the perinuclear region of the neuron is a critical step (Jenkins et al. 2009). One day later, the centrioles migrate toward the dendritic knob and eventually disperse singly around the knob periphery where they anchor with the plasma membrane to presumably form basal bodies. Ciliogenesis commences when a single, primary cilium of approximately 1 µm extends toward the apical surface and into the nasal cavity (Schwarzenbacher et al. 2005; Jenkins et al. 2009). By E14, multiple cilia up to 2 µm in length can be seen extending from a single dendritic knob (Figure 8). Prior to and continuing after birth, existing cilia extend their microtubule axoneme and additional cilia form. It is important to note the timing of this process is not uniform across all OSNs and likely reflects the nonsynchronous maturation of neurons. This is reminiscent of the differential temporal expression of proteins necessary for odor detection. This process appears to begin with odor receptor expression as early as E11 (Saito et al. 1998; Schwarzenbacher et al. 2004; Schwarzenbacher et al. 2005) and continues until cyclic nucleotide-gated channel expression at E19 (Margalit and Lancet 1993). Although ciliogenesis and olfactory transduction protein expression appear to share an overlapping temporal pattern, it is unclear if there exists a dynamic reciprocity between the 2 processes in regulating OSN maturation. Similarly, numerous questions regarding the primary cilium on OSNs remain, including their exact function, associated signaling pathways, accompanying genetic programs, and the consequences of cilia loss or perturbation during and after development.

Until recently, it was widely believed that OSNs were the sole ciliated cell type in the olfactory epithelium. However, the presence of primary cilia on a quiescent population of basal stem cells, the HBCs, has recently been described (Joiner et al. 2015). It is assumed, although clearly not proven, that HBC cilia are analogous, in terms of structure and maintenance, to primary cilia on neurons in the brain as described above. This is visibly true in terms of number and length. Compared with the multiple cilia of terminally differentiated OSNs, HBC primary cilia are likely more comparable to the single cilium found on neuroblasts migrating from the subventricular zone to the olfactory bulb (Matsumoto et al. 2019). However, the precise function of these developmental cilia remains unclear. Genetic deletion of HBC cilia showed that cilia may play a role in signal dependent HBC activation, proliferation, and/or differentiation in the olfactory epithelium (Joiner et al. 2015). Importantly, regeneration is more limited when the HBCs are genetically deciliated, and this also alters neurogenesis during the development of the olfactory epithelium. This is consistent with an established role for primary cilia in cell proliferation, differentiation, and regulation of the cell cycle (Irigoín and Badano 2011). Studies show that primary cilia regulate neurogenesis and/or proper differentiation of adult stem cells into amplifying progenitor cells within the brain (Amador-Arjona et al. 2011; Kumamoto et al. 2012; Tong et al. 2014). As a result, when cilia are disrupted in these systems, cilia-mediated signaling pathways, such as sonic hedgehog and Wnt signaling, are also disrupted (Kumamoto et al. 2012; Tong et al. 2014). These findings demonstrate the important role that cilia play in the signaling pathways that are essential for proper cell differentiation during development and adult homeostasis in a neuronal tissue. Even with the discovery of the importance of HBC cilia, however, much about HBC primary cilia remains to be understood. We know much less about the cilia gene expression profile in these cells compared with OSNs and the structure, function, and signaling pathway associated with primary cilia in HBCs remain to be elucidated.

Conclusions and unanswered questions

An explosion of transcriptome data in the past decade has produced an in-depth understanding of what genes are expressed by OSNs, and whether these genes are expressed by nascent OSNs, immature OSNs, mature OSNs, or some combination of these stages of neural differentiation. These data reveal what phenotypes are possible in OSNs, at which stage of differentiation they are functional, and predict which genes are responsible for these phenotypes. Future work is necessary to test these predictions and explore their importance for olfactory biology. Furthermore, it is still true that a majority of mammalian genes encode proteins whose functions have never been tested directly. At best, their functions are predicted from similarity to proteins that have been studied, often in invertebrates. OSNs express many of these genes. The accessibility of the olfactory epithelium and its capacity of lifelong replacement of OSNs offer many advantages for discovering the functions of unstudied genes and their protein products both in a mature neuron and in the differentiation processes responsible for producing the neuron.

In addition, the molecular bases for several phenomena known to be important to OSNs remain to be discovered. For example, what are the signals from the olfactory bulb necessary to support OSN survival? OSNs express many types of receptors, including GPCRs other than odorant receptors (Sammeta et al. 2007; Nickell et al. 2012; Ibarra-Soria et al. 2014; Saraiva et al. 2015). Of these, the D2 dopamine receptor (Drd2) and the GABA-B receptor (Gabbr1) are important for presynaptic inhibition at OSN synapses in the olfactory bulb (Nickell et al. 1994; Ennis et al. 2001), but what roles do the other receptors play and what are the sources of the signaling molecules activating them?

Cilia are critical for the primary function of OSNs and are the most exposed parts of the OSN. Does damage to cilia occur—and if so, is it easily repaired or is it important in determining OSN survival? In addition, the precise role of cilia in the maturation of OSNs remains unclear. It is tempting to speculate that there is a dynamic reciprocity between cilia signaling and OSN maturation. Do primary cilia on emerging immature OSNs influence genetic programs that regulate or control the transition of OSNs to the mature state? In addition, the role of cilia in the maintenance of the olfactory epithelium is also important to consider. Recent work suggested that primary cilia on the cell body of neurons in brain can regulate axon extension and synaptic formation (Higginbotham et al. 2012; Kumamoto et al. 2012; Ferent et al. 2019). Are primary cilia important for OSN axon extension, convergence, or coalescence in the bulb? The resorption of cilia is necessary for cell cycle-dependent cell division in other cells. The role of cilia in maintaining basal cell quiescence in the olfactory epithelium also needs further investigation.

What we know about OSN differentiation is largely based on adult neurogenesis. Whether there are differences between these adult-born OSNs and the earliest OSNs born embryonically and perinatally was long uncertain. The discovery of a critical perinatal period in the formation of the map of homotypic glomeruli in the olfactory bulb could be due to a distinctive feature of embryonic OSNs, or instead it could be a function of a distinctive perinatal environment experienced by OSN axons (Ma et al. 2014; Tsai and Barnea 2014). Further work reveals that early “navigator” neurons in the olfactory epithelium have a distinctive transcriptome and phenotype (Wu et al. 2018). These navigators have short lifespans, show greater abundance of mRNAs encoding proteins involved in cell metabolism and mitochondrial morphogenesis, and lower abundance of mRNAs encoding proteins involved in olfactory transduction. More importantly, their axon pathfinding is distinctively exuberant and exploratory compared with that of mature OSNs formed just a few days later. They appear to be unrelated to the immature OSNs that serve as embryonic pioneer neurons whose axons are the first to reach the telencephalon from the olfactory epithelium (Gong and Shipley 1995). These data suggest that except for a small number of OSNs formed early in the development of the olfactory epithelium, the differentiation of OSNs is fundamentally the same in embryos, newborns, and adults.

Identifying essentially all genes expressed by immature OSNs and mature OSNs has provided at best a gross overview of the transitions necessary to produce a mature neuron. Studies to date mostly revolve around the effects of the loss of a transcription factor or the dynamics of changes of abundance of specific cell types. They provide significant information about the details and timing of the progression of events during OSN differentiation, but what they achieve is far from a complete understanding of how differentiation is accomplished. The evidence that OSN differentiation is primarily an intrinsic process potentially simplifies study of these events and interpreting the findings of such studies.

This detailed understanding of OSN genomics, and that of neighboring cells as well, also increases the rigor with which we can investigate the biology of the olfactory epithelium. No longer are distinctions between cell types and shifts in processes dependent on single markers but rather can—and should—be documented rigorously using sets of markers capable of better placing cells within the OSN cell lineage. Furthermore, basing conclusions about cell-type distinctions or shifts in phenotype on expression of 1 or 2 marker genes is revealed as somewhat artificial by global analyses (Figure 3B, C). Cell lineages progress toward a differentiated state via networks of processes that occur at varying speeds and last for varying durations. The use of 1 or 2 markers of states achieved in this progression tends to emphasize states that linger at the expense of more transient states that are often equally important. A broader view of the shifting networks of molecular events that drive the process of OSN differentiation is now available (Fletcher et al. 2017) and should be employed when assessing the phenotypes of OSNs and their precursors.

Supplementary material

Supplementary data are available at Chemical Senses online.

Table S1. List of 1750 transcriptional regulator transcripts detected in samples of olfactory mucosa. P(in) values are the probabilities of expression in mature OSNs, immature OSNs, or the pool of other cell types in such samples (Nickell et al. 2012). OBX fold diff is the difference in transcript abundance between ipsilateral and contralateral olfactory mucosal samples 5 days after unilateral olfactory bulbectomy (OBX) (Heron et al. 2013). Mat OSN index is the ratio of transcript abundance between samples enriched for Omp⁺ OSN samples and whole olfactory mucosa (Saraiva et al. 2015).

Table S2. List of 133 transcriptional regulator transcripts that increase in samples of olfactory mucosa 5 days after ablation of the ipsilateral olfactory bulb. P(in) values are the probabilities of expression in mature OSNs, immature OSNs, or the pool of other cell types in such samples (Nickell et al. 2012). OBX fold diff is the difference in transcript abundance between ipsilateral and contralateral olfactory mucosal samples 5 days after unilateral olfactory bulbectomy (OBX) (Heron et al. 2013). Mat OSN index is the ratio of transcript abundance between samples enriched for Omp⁺ OSN samples and whole olfactory mucosa (Saraiva et al. 2015).

Table S3. List of 807 transcriptional regulator transcripts expressed primarily in immature OSNs. P(in) values are the probabilities of expression in mature OSNs, immature OSNs, or the pool of other cell types in such samples (Nickell et al. 2012). OBX fold diff is the difference in transcript abundance between ipsilateral and contralateral olfactory mucosal samples 5 days after unilateral olfactory bulbectomy (OBX) (Heron et al. 2013). Mat OSN index is the ratio of transcript abundance between samples enriched for Omp⁺ OSN samples and whole olfactory mucosa (Saraiva et al. 2015).

Table S4. List of 74 transcriptional regulator transcripts, expressed primarily in immature OSNs, that increase in samples of olfactory mucosa 5 days after ablation of the ipsilateral olfactory bulb. P(in) values are the probabilities of expression in mature OSNs, immature OSNs, or the pool of other cell types in such samples (Nickell et al. 2012). OBX fold diff is the difference in transcript abundance between ipsilateral and contralateral olfactory mucosal samples 5 days after unilateral olfactory bulbectomy (OBX) (Heron et al. 2013). Mat OSN index is the ratio of transcript abundance between samples enriched for Omp⁺ OSN samples and whole olfactory mucosa (Saraiva et al. 2015).

Table S5. List of 500 transcriptional regulator transcripts expressed abundantly in immature OSNs but also expressed in mature OSNs. P(in) values are the probabilities of expression in mature OSNs, immature OSNs, or the pool of other cell types in such samples (Nickell et al. 2012). OBX fold diff is the difference in transcript abundance between ipsilateral and contralateral olfactory mucosal samples 5 days after unilateral olfactory bulbectomy (OBX) (Heron et al. 2013). Mat OSN index is the ratio of transcript abundance between samples enriched for Omp⁺ OSN samples and whole olfactory mucosa (Saraiva et al. 2015).

Table S6. List of 81 transcriptional regulator transcripts, expressed in both immature and mature OSNs, that decrease in samples of olfactory mucosa 5 days after ablation of the ipsilateral olfactory bulb. P(in) values are the probabilities of expression in mature OSNs, immature OSNs, or the pool of other cell types in such samples (Nickell et al. 2012). OBX fold diff is the difference in transcript abundance between ipsilateral and contralateral olfactory mucosal samples 5 days after unilateral olfactory bulbectomy (OBX) (Heron et al. 2013). Mat OSN index is the ratio of transcript abundance between samples enriched for Omp⁺ OSN samples and whole olfactory mucosa (Saraiva et al. 2015).

Table S7. List of 9,941 genes expressed in mature OSNs. Criteria for inclusion are normalized RNA-seq counts of >5 in these Omp⁺ OSN samples and enrichment in Omp⁺ OSN samples compared to whole olfactory mucosa (fold enrichment > 1) mucosa (Saraiva et al. 2015) or mature OSN P(in) values > 0.5 (Nickell et al. 2012). OSN avg, average normalized FPKM values for samples enriched for GFP⁺ OSNs from Omp-GFP mice. WOM avg, average normalized FPKM values for samples of whole olfactory mucosa (WOM). OSN index, the ratio of OSN avg/WOM avg. P(in) values are the probabilities of expression in mature OSNs, immature OSNs, or the pool of other cell types in such samples (Nickell et al. 2012). OBX fold diff is the difference in transcript abundance between ipsilateral and contralateral olfactory mucosa samples 5 days after unilateral olfactory bulbectomy (OBX) (Heron et al. 2013).

Funding

This work was supported by the National Institute on Deafness and other Communication Disorders of the National Institutes of Health awards (R01 DC014468 to T.S.M. and R01 DC009606 to J.R.M.).

Acknowledgments

We thank Matthew Hazzard, Tom Dolan, and Julien C. Habif for assistance with figures.

Conflict of interest

T.S.M. has an equity interest in a company based on technologies used to measure responses to odors.

References