Postnatal Refinement of Peripheral Olfactory Projections

See allHide authors and affiliations

Science  25 Jun 2004:
Vol. 304, Issue 5679, pp. 1976-1979
DOI: 10.1126/science.1093468


Axonal projections from the olfactory epithelium to the olfactory bulb are organized into glomeruli according to the expressed odorant receptor. Using gene-targeted mice, we show that glomerular maturation proceeds along different time courses for two similar receptors and requires sensory input during distinct sensitive periods. During early development, some glomeruli are innervated by axons of neurons that do not express the same receptor. These heterogeneous glomeruli normally disappear with age, but they persist in adults deprived of sensory input by unilateral and permanent naris closure. Persistence may be due, in part, to prolonged survival of olfactory sensory neurons.

The decoding of environmental stimuli detected by specialized sensory cells requires the formation of precise projections from the periphery to the brain. In mammalian olfactory systems, the connection between periphery and brain occurs over a single synapse. The axons of olfactory sensory neurons (OSNs) expressing the same odorant receptor (OR) project to spatially conserved regions of the olfactory bulb, where they coalesce into glomeruli (15). This process uses molecular determinants, but it remains controversial to what extent it also relies on sensory activity (611).

We examined the postnatal formation of glomeruli for ORs M71 and M72 in genetargeted M71-IRES-taulacZ and M72-IRES-taulacZ mice (M71 and M72 mice) during development (12, 13). The expression of the axonal marker tau–β-galactosidase (β-gal) encoded by the gene taulacZ is under the direct control of the endogenous M71 or M72 OR promoter (8, 14, 15). ORs M71 and M72 are highly homologous, with 96% identity in their amino acid sequences. The axons from OSNs expressing these ORs (referred to as M71 or M72 axons) each form at least one medial and one lateral glomerulus (M71 or M72 glomerulus) in adult olfactory bulbs. M71 and M72 glomeruli are located within a few hundred microns of each other.

The positions of specific glomeruli are bilaterally symmetrical between olfactory bulbs, each of which possesses an internal symmetry as well. There are thus four “half-bulbs” in an individual. In newborn mice (Postnatal Day 0, PD0), M71 and M72 axons had reached two spatially conserved posterior-dorsal areas in each bulb (one each on the medial and lateral half-bulbs), but the axons were widely scattered, and M71 and M72 glomeruli were not yet identifiable (8, 16). By PD10 these glomeruli were clearly defined, but there were often multiple M71 and M72 glomeruli at either the medial or lateral half-bulbs (whole-mount observations, Fig. 1A), whose precise locations varied within a spatially conserved region. In older animals (≥PD40), single M71 and M72 glomeruli were generally observed in both the medial and lateral half-bulbs (Fig. 1, B and C). Between PD10 and PD60, the proportion of half-bulbs with multiple glomeruli decreased significantly [P < 0.01 (17)] from 52% (M71, n = 112) to 16% (n = 80) and from 55% (M72, n = 88) to 17% (n = 76), with a corresponding increase in single glomeruli (Fig. 1D). These observations suggest a process of OSN axon projection refinement, which may reflect a threshold of an axonal population needed to maintain a glomerulus (18).

Fig. 1.

Maturation of M71 and M72 glomeruli. (A to C) Examples of M72 glomeruli in whole-mount olfactory bulbs (OB) stained with X-gal. (A) Multiple glomeruli (arrows) in a young animal (PD10). (B) and (C) Single glomerulus in the medial half-bulb of the right OB (B) and lateral half-bulb of the left OB (C) from an older animal (PD40). Orientation for the whole mounts: D, dorsal; V, ventral; A, anterior; P, posterior; M, medial; L, lateral. Scale bar in (A), and in (B) for (B) and (C), 0.5 mm. (D) Proportion of half-bulbs with multiple glomeruli in whole mounts between PD10 and PD60 for M71 (blue) and M72 (red) glomeruli. (E) Time courses for the maturation of M71 (blue) and M72 (red) glomeruli in whole mounts. The average number of M71 glomeruli per half-bulb was 1.19 ± 0.04 at >PD90 (mean ± SEM, n = 140) and 1.18 ± 0.05 at PD60 [n = 80; not different from value at >PD90, P > 0.05 (19)]. The average number of M72 glomeruli per half-bulb was 1.17 ± 0.04 at >PD90 (n = 116) and 1.29 ± 0.06 at PD20 [n = 100; not different from value at >PD90, P > 0.05 (19)].

M71 and M72 glomeruli matured differently (Fig. 1E). M71 glomeruli underwent a prolonged maturation, and the average number of M71 glomeruli per half-bulb did not reach the mature level (>PD90) until PD60. In contrast, the average number of M72 glomeruli per half-bulb rapidly decreased to the level of mature animals (>PD90) by PD20. Thus, glomerular maturation can follow distinct time courses for even closely related ORs.

A hallmark of mature glomeruli is that they are innervated exclusively by axons from OSNs expressing the same OR (5). To establish whether this is the case for glomeruli in young animals, we used a double-label immunohistochemical protocol on serial coronal cryosections with antibodies for olfactory marker protein (OMP) to stain all mature OSN axons, and antibodies for β-gal to label M71 (or M72) axons. In homogeneous M71 (or M72) glomeruli, immunoreactivity for these two markers should largely overlap. At PD10, when multiple glomeruli are commonly observed in each half-bulb, most M71 (or M72) glomeruli exhibited colocalization of OMP and β-gal immunoreactivity (Fig. 2A and Table 1). But many M71 and M72 glomeruli also exhibited OMP-positive, β-gal–negative axons (Fig. 2B and Table 1), suggesting that these immature glomeruli are innervated also by axons from OSNs expressing other ORs. In contrast, in PD60 animals, which predominantly have a single glomerulus in each half-bulb, glomeruli were homogeneously innervated by M71 (or M72) axons, with few exceptions (Fig. 2C and Table 1).

Fig. 2.

Developmental changes in the organization of OSN axons within M71 and M72 glomeruli. M71 or M72 glomeruli were detected by immunostaining in serial coronal cryosections through the bulb. Red, olfactory marker protein (OMP) immunoreactivity labels olfactory sensory axons. Green, β-galactosidase (β-gal) immunoreactivity labels M71 or M72 axons. Blue, nuclear staining with TOTO-3 marks the glomeruli. (A) Multiple homogeneous M71 glomeruli in the medial half-bulb at PD10. (B) One heterogeneous M72 glomerulus in the medial half-bulb at PD10. (C) Single homogeneous M71 glomerulus in the medial half-bulb at PD60. (D) Proportion of heterogeneous M71 (blue) and M72 (red) glomeruli in half-bulbs between PD10 and PD60. Scale bar in (A) for (A) to (C), 50 μm.

Table 1.

Homogeneous and heterogeneous glomeruli during development and after naris closure. The organization of OSN axons within M71 and M72 glomeruli was examined by OMP and β-gal double immunostaining in serial coronal sections through the bulbs of both normally developed and naris-closed animals. Animals were naris closed at PD0 to PD5 (M71) or PD0 (M72) and examined at PD40.

Glomeruli Age/side Number of half-bulbs Total glomeruli per half-bulb (min—max) Homogeneous glomeruli per half-bulb Heterogeneous glomeruli per half-bulb
Normally developed animals
M71 PD10 18 1.78 ± 0.26 (1-5) 1.33 ± 0.14 0.44 ± 0.25
PD60 14 1.29 ± 0.13 (1-2) 1.21 ± 0.15 0.07 ± 0.07
M72 PD10 16 1.81 ± 0.21View inline (1-3) 1.19 ± 0.19 0.63 ± 0.18View inline
PD60 18 1.11 ± 0.08 (1-2) 0.94 ± 0.10 0.17 ± 0.09
Naris-closed animals
M71 closed 14 2.43 ± 0.17View inline (2-4) 1.14 ± 0.15 1.29 ± 0.19View inline
open 14 1.07 ± 0.07 (1-2) 0.93 ± 0.07 0.14 ± 0.14
M72 closed 12 2.25 ± 0.13View inline (2-3) 0.92 ± 0.19 1.33 ± 0.19View inline
open 12 1.33 ± 0.19 (1-3) 1.17 ± 0.21 0.17 ± 0.17
  • Statistically significant (P < 0.05, run on SPSS program):

  • View inline* Comparing normally developed animals between PD10 and PD60, Mann-Whitney test.

  • View inline Comparing naris-closed animals between bulbX and bulbO, Wilcoxon signed-ranks test.

  • Between PD10 and PD60, the proportion of half-bulbs with heterogeneous M72 glomeruli significantly [P < 0.05 (17)] decreased. M71 glomeruli showed a similar tendency (Fig. 2D). However, the average number of homogeneous M71 or M72 glomeruli per half-bulb did not change significantly [P > 0.05 (19), Table 1]. These data point to a maturation process that may occur through rearrangement or selective degeneration of OSN axons (9) innervating the heterogeneous glomeruli.

    We sought to determine whether manipulations of sensory input could influence the maturation process of M71 and M72 glomeruli. To reduce levels of sensory input to one olfactory bulb, we performed permanent unilateral naris closures, a method for introducing olfactory sensory deprivation [reviewed by Brunjes (20)]. In the bulb ipsilateral to the closed naris (bulbX), both spontaneous and odorant-induced activity are dramatically reduced (21). Because of the ipsilateral projection pattern of OSN axons, the contralateral open side bulb (bulbO) can serve as an internal control.

    Naris closure was performed at PD0, when M71 and M72 glomeruli begin to form. When animals were examined at PD40 to PD90, predominantly single M71 and M72 glomeruli were found in the whole-mount medial and lateral control half-bulbsO (Fig. 3A), similar to what is found in normally developed animals. In contrast, in the half-bulbsX, multiple M71 and M72 glomeruli were observed (Fig. 3B). The precise positions of the multiple M71 and M72 glomeruli were variable, but they occurred within conserved areas and were separated by no more than 300 μm. The proportion of half-bulbsX with multiple M71 or M72 glomeruli was greater [P < 0.001 (17)] than that of contralateral half-bulbsO (Fig. 3C). Thus the maturation of M71 and M72 glomeruli is influenced by naris closure, which appears to preserve the immature condition into adulthood.

    Fig. 3.

    Naris closure differentially perturbs the maturation of M71 and M72 glomeruli. (A and B) X-gal stained M71 glomeruli in PD40 whole mount with left naris closed since PD0. A single glomerulus was seen in the lateral half-bulbO (A), while two glomeruli (arrows) occurred in the lateral half-bulbX (B). Scale bar in (A) for (A) and (B), 0.5 mm. (C) Animals were naris closed at PD0 and examined with X-gal staining at PD40 to PD90. The proportions of half-bulbs with multiple glomeruli were compared between the half-bulbsX and half-bulbsO for M71 (blue, n = 44) and M72 (red, n = 40) glomeruli. (D and E) The average number of M71 [(D), blue] or M72 [(E), red] glomeruli per half-bulbX (filled circles) was compared with that of the contralateral half-bulbsO (open circles) for each age of naris closure.

    Sensory activity often modulates the development of sensory projections within a defined sensitive period, during which the brain is particularly plastic. We performed permanent naris closure in groups of M71 and M72 mice starting from PD0 to PD30 in 5-day increments. Whole-mount bulbsX and bulbsO were examined when animals reached PD40 to PD90 and the naris had been closed for at least 20 days. For M71 glomeruli, when closed at PD25 (or later), a similar number of glomeruli were observed in both half-bulbsX and half-bulbsO (Fig. 3D) [1.55 ± 0.09 and 1.42 ± 0.09, respectively; n = 60; P > 0.05 (22)], suggesting that the sensitive period for M71 glomeruli terminates around PD25. In contrast, the sensitive period for M72 glomeruli ended around PD15 (Fig. 3E) [1.31 ± 0.07 glomeruli per half-bulbX and 1.24 ± 0.08 glomeruli per half-bulbO; n = 42; P > 0.05 (22)]. Mirroring the later, prolonged maturation of M71 glomeruli and the earlier, rapid maturation of M72 glomeruli during normal development, the ends of the sensitive periods are correlated with the distinct developmental time courses for these glomeruli.

    Is the decreasing frequency of heterogeneous glomeruli during normal development influenced by sensory activity? To address this issue, we again used the protocol of OMP and β-gal double immunostaining to examine the organization of OSN axons in PD40 animals that underwent naris closure at PD0 to PD5 for M71 glomeruli, or at PD0 for M72 glomeruli. In the half-bulbsO, the vast majority of M71 and M72 glomeruli were homogeneous (Fig. 4A). In contrast, almost all half-bulbsX contained at least one heterogeneous M71 and M72 glomerulus (Fig. 4B). Thus, the proportion of half-bulbsX with heterogeneous glomeruli was significantly greater [P < 0.001 (17)] than that of contralateral half-bulbsO in the same animals (Fig. 4C).

    Fig. 4.

    The organization of M71 and M72 glomeruli in naris-closed animals. Animals were naris closed at PD0 to PD5 (M71) or PD0 (M72) and examined at PD40. (A and B) The organization of OSN axons in M72 glomeruli was examined by OMP (red) and β-gal (green) double immunostaining in serial coronal sections through OBs. Blue, nuclear staining with TOTO-3. The same medial half-bulbX contained one homogeneous M72 glomerulus (A) and one heterogeneous M72 glomerulus (B), 195 μm caudal to the homogeneous glomerulus. (C) Proportion of heterogeneous M71 (blue) and M72 (red) glomeruli in the half-bulbsX and half-bulbsO from the same naris-closed animals. (D) Dendrites of bulb neurons (MAP-2 immunoreactivity, red) intermingled with M71 axons (β-gal immunoreactivity, green) in multiple M71 glomeruli in the half-bulbX.(E) Colocalization of immunoreactivity specific for synapsin I (red) and β-gal–positive axons (green) in one of the multiple M71 glomeruli in the half-bulbX. (F) Ultrastructural view of β-gal–positive OSN axon terminals in one of the multiple M71 glomeruli in the half-bulbX. One unlabeled dendrite, most likely a mitral/tufted cell dendrite, formed synapses with two olfactory axon terminals immunolabeled for β-gal. Den, dendrite; ONt, olfactory nerve terminal. Direction of arrows indicates polarity of synapse. Scale bar in (A) for (A) and (B), 50 μm; in (D), 50 μm; in (E), 25 μm; in (F), 0.5 μm.

    Consistent with the observations in whole mounts, the average number of M71 (or M72) glomeruli per half-bulbX in serial cryosections was nearly double that found in the contralateral half-bulbsO (Table 1). The increased number of M71 (or M72) glomeruli in half-bulbsX can be accounted for by the significantly greater number of heterogeneous glomeruli [P < 0.05 (22)] (Table 1), whereas the number of homogeneous glomeruli remained the same, roughly one per half-bulbX [P > 0.05 (22)] (Table 1). Thus it appears that sensory activity determines the fate of heterogeneous glomeruli but does not influence several other processes involved in glomerular formation, such as the targeting of OSN axons to the proper region in the bulb and the coalescence of axons specific for the same OR into a tight glomerular organization. Indeed, the occurrence of only a small number of heterogeneous glomeruli in young or sensory-deprived animals lends further support to a major role of molecular determinants in axon guidance. Without such determinants, we would have expected a large and variable population of heterogeneous glomeruli.

    We further examined whether the cellular organization of multiple glomeruli observed in serial sections of bulbsX is similar to that of the normally developed mature glomeruli. In the bulbsX, the dendritic marker microtubule-associated protein 2 (MAP2) showed positive staining in all glomeruli. Individual MAP2-positive dendritic arbors from bulb neurons intermingled with β-gal–positive M71 (or M72) axon terminals (Fig. 4D), as in normal glomeruli (23), suggesting that dendrites of bulb neurons innervate all of the M71 (or M72) glomeruli in the bulbX. Immunoreactivity to the synaptic vesicle protein synapsin I was observed in all of the β-gal–positive M71 (or M72) glomeruli in the bulbsX (Fig. 4E), consistent with the notion that synapses were formed between the M71 (or M72) axons and dendrites of bulb neurons and that the extra glomeruli in the bulbsX may be functional. Finally, ultrastructural examination of glomeruli in the bulbsX by electron microscopy showed normal synapses formed between the incoming M71 (or M72) axons and dendrites of bulb neurons (Fig. 4F).

    To begin defining the mechanisms underlying the sensory-dependent glomerular remodeling, we considered three possibilities—axonal pruning, axonal rearrangement, and selective neuronal cell death. Because OSN axons do not branch before reaching their glomerulus (2426), pruning seemed unlikely, as did wholesale rearrangement of axonal projections over several hundred microns. We therefore focused on cell death. The olfactory epithelium renews itself throughout life, with a balance between the generation of new OSNs from a basal stem cell population and continuous OSN cell death. The kinetics of this process are affected by naris closure. Whereas the overall cell number remains constant (2729), the generation of OSNs is known to decrease on the closed side (28, 29), suggesting that the survival of cells on the closed side may be prolonged. These altered dynamics point to a change in cell turnover as a possible mechanism of glomerular remodeling. To investigate this further, we labeled the dividing epithelial cells with 5-bromo-2′-deoxyuridine (BrdU) at PD10 and simultaneously performed naris closure. At PD20, the ratio of BrdU-positive profiles between the closed and the open side in the septum was 1.32 ± 0.17 [n = 6, P < 0.05 (17)], demonstrating that significantly more OSNs born on or after PD10 survived on the closed side (Fig. 5). Although it is not possible to determine the OR expressed in the BrdU-labeled cells or track their axons to the olfactory bulb, the prolonged survival of sensory-deprived OSNs could underlie the preservation of heterogeneous glomeruli in the bulbX.

    Fig. 5.

    Prolonged survival of OSNs in naris-closed animals. (A) An example of BrdU-positive cells in the septum detected by immunostaining at PD20. The animal was injected with BrdU, and the left naris was closed at PD10. (B and C) Enlarged views of left (closed side) and right (open side) boxes in (A), respectively. BrdU-immunopositive profiles, indicated by arrows in the OSN layer, were of the type included in our analysis. BrdU profiles indicated by arrowheads in the basal cell layer (B) and supporting cell layer (C) were excluded from our analysis. Scale bar in (A), 50 μm; in (C) for (B) and (C), 25 μm.

    Taken together, our results establish four principles of olfactory system development. First, the absence of sensory activity perturbs glomerular maturation; second, there is a sensitive period during which activity influences the maturation of glomerular organization; third, sensitive periods occur asynchronously, with the specific timing dependent on the OR; fourth, glomeruli may be heterogeneously innervated by more than one OSN population early in development or in the absence of activity.

    Molecular determinants are necessary for the initial formation of glomeruli. Evidence for activity-independent formation of the glomerular array has emerged from studies using genetic ablations of components in the transduction pathway to reduce odor-driven activity. However, these manipulations typically leave one or more residual processes intact. In contrast, naris closure reduces all activity associated with normal passage of air through the nasal cavity— biochemical, electrical, and toxic—providing a more revealing challenge to the system. The increased survival of OSNs on the closed side suggests that the dynamics of cell turnover in the epithelium provides a mechanism for the refinement of OSN axon projections. A decrease in OSN turnover, such as occurs in an epithelium that is deprived of activity, may provide less opportunity for the removal of cells that project axons to heterogeneous glomeruli. Unlike in the visual system (30), selective pruning of axon branches is less likely to operate in the olfactory system because axons do not have interglomerular axon collaterals. Further, we deem it unlikely that coordinated activity evoked by natural odors, which are typically complex mixtures of odorants, over relatively short preadult stages would provide enough discrimination as the driving force to untangle 1000 populations of axons into discrete glomeruli. A more complex chain of events may underlie the dynamic effects observed here.

    Supporting Online Material

    Materials and Methods


    References and Notes

    View Abstract

    Stay Connected to Science

    Navigate This Article