Chem. Senses 27: 245-260,
2002
© Oxford University Press 2002
Axon Navigation in the Mammalian Primary Olfactory Pathway
Where to Next?
1 Department of Anatomy and Developmental Biology, School of Biomedical Sciences, University of Queensland, Brisbane 4072, Australia 2 Centre for Functional and Applied Genomics, University of Queensland, Brisbane 4072, Australia
Correspondence to be sent to: Brian Key, Department of Anatomy and Developmental Biology, School of Biomedical Sciences, University of Queensland, Brisbane 4072, Australia. e-mail: brian.key{at}uq.edu.qu
 |
Abstract |
|---|
 Top Abstract IntroductionAxon navigation to the...Axon sorting in the...Axon homing to glomerular...Working models of axon...Where to next?References |
|---|
The process of establishing long-range neuronal connections can be divided into at least three discrete steps. First, axons need to be stimulated to grow and this growth must be towards appropriate targets. Second, after arriving at their target, axons need to be directed to their topographically appropriate position and in some cases, such as in cortical structures, they must grow radially to reach the correct laminar layer. Third, axons then arborize and form synaptic connections with only a defined subpopulation of potential post-synaptic partners. Attempts to understand these mechanisms in the visual system have been ongoing since pioneer studies in the 1940s highlighted the specificity of neuronal connections in the retino-tectal pathway. These classical systems-based approaches culminated in the 1990s with the discovery that Eph—ephrin repulsive interactions were involved in topographical mapping. In marked contrast, it was the cloning of the odorant receptor family that quickly led to a better understanding of axon targeting in the olfactory system. The last 10 years have seen the olfactory pathway rise in prominence as a model system for axon guidance. Once considered to be experimentally intractable, it is now providing a wealth of information on all aspects of axon guidance and targeting with implications not only for our understanding of these mechanisms in the olfactory system but also in other regions of the nervous system.
|
Introduction |
|---|
|
TopAbstractIntroductionAxon navigation to the...Axon sorting in the...Axon homing to glomerular...Working models of axon...Where to next?References |
|---|
In this commentary we will highlight insights into axon navigation within the mammalian primary olfactory system provided by recent molecular studies while at the same time controversies that surround some of the underlying data will be pointed out. We will finish our discussion with a description of a model that may help to explain axon targeting in the olfactory bulb. The pathway followed by olfactory sensory axons can be treated as consisting of three distinct sub-regions: the olfactory nerve connecting the olfactory neuroepithelium to the olfactory bulb; the outer olfactory nerve fibre layer; and finally the inner olfactory nerve fibre and glomerular layers. Since axons course in each of these regions they appear to depend on distinct molecular and cellular cues. For instance, in mice deficient in the SRC-family tyrosinase kinases, p59fyn and pp60c-src, axons defasciculate in the olfactory nerve (Morse et al., 1998
)
while in N-CAM-180 null mutant mice, olfactory sensory axons do not correctly
sort out in the nerve fibre layer (Treloar
et al., 1997). In mice lacking the P2 odorant receptor
gene, P2 axons fail to exit the nerve fibre layer and do not enter the
underlying glomerular layer (Wang et
al., 1998). Although we will examine each of these different
sub-regions in turn, our main focus will be on the homing of axons to specific
glomerular targets in the olfactory bulb. |
Axon navigation to the telencephalon |
|---|
|
TopAbstractIntroductionAxon navigation to the...Axon sorting in the...Axon homing to glomerular...Working models of axon...Where to next?References |
|---|
Little is known about the signals that stimulate axon formation and outgrowth from olfactory sensory neurons. Early studies suggested that the mesenchyme lying between the presumptive placode and the rostral surface of the neural tube produced inductive signals. Retinoic acid is expressed by this mesenchyme (LaMantia et al., 1993
) and it has been implicated in both olfactory sensory neuron
generation and subsequent axon outgrowth
(Whitesides et al.,
1998; LaMantia et
al., 2000). In the absence of appropriate olfactory neuronal
cell lines, studies have focused on understanding olfactory sensory
neuron-specific gene expression rather than describing mechanisms controlling
the initiation of axon growth (Kudrycki
et al., 1998; Qasba
and Reed, 1998; Behrens et
al., 2000). However, a recent study has identified NELF, a
novel 50 kDa protein expressed by olfactory sensory axons which has neurite
outgrowth promoting activity (Kramer and
Wray, 2000). Inhibition of expression of this protein using
antisense oligonucleotides demonstrated a remarkable attenuation of outgrowth
of olfactory axons from explant cultures of olfactory mucosa. An understanding
of the function of this protein in development of the olfactory nerve pathway
awaits the targeted mutagenesis of this molecule.
There is an unwritten law that once generated `axons prefer to grow in
a forward direction unless directed to do otherwise'. This is clearly
true in vitro where axons are observed to grow radially away from
explants of neural tissue. Axons do not spontaneously turn-about, form whorls
or make radical directional changes. In some cases the forward growth of an
axon is restricted to a particular pathway by physical constraints. For
instance, in the retino-tectal pathway axons are initially guided by the
physical restraints imposed by the optic stalk. Although the olfactory system
does not possess a similar conduit connecting the olfactory nasal pit to the
rostral telencephalon it appears to generate its own conduit de novo.
Olfactory axons navigate through the frontonasal mesenchyme by following a
glial bridge that forms between the nasal pit and the rostral surface of the
brain (Farbman and Squinto,
1985; Doucette,
1989; Tennet and Chuah, 1996). These glia are born in the
olfactory neuroepithelium of the nasal pit and migrate towards the
telencephalon probably under the influence of chemotropic molecules released
by the presumptive olfactory bulb (Liu
et al., 1995) or surrounding frontonasal mesenchyme
(LaMantia et al.,
2000). This distance is initially only 150-250 µm in mouse
(Marin-Padilla and Amieva,
1989), which is appropriate for the action of soluble growth
factors (Nellen et al.,
1996).
The glia in the olfactory nerve pathway are the presumptive ensheathing
cells of the olfactory nerve and nerve fibre layer of the olfactory bulb.
These ensheathing cells are excellent promoters of olfactory axon growth
in vitro (Ramón-Cueto and
Valverde, 1995; Kafitz and Greer,
1998b,
1999). In fact, these cells
are now credited with the extraordinary growth potential of olfactory axons
throughout life and appear to facilitate regeneration of spinal axons when
transplanted into lesions of the spinal cord
(Ramón-Cueto and Avila,
1998). If given a choice of extracellular matrices or ensheathing
cells, olfactory axons prefer to grow on ensheathing cells
(Tisay and Key, 1999). Thus,
the initial guidance of olfactory axons to the brain should really be thought
of in terms of mechanisms guiding migratory glia. The frontonasal mesenchyme
is rich in chondroitin sulphates whereas the olfactory nerve pathway is devoid
of these molecules (Treloar et
al., 1996a). As ensheathing cells do not migrate on a
substrate of chondroitin sulphates (Tisay
and Key, 1999), the chondroitin sulphates may act to restrict
migration of glia and hence confine axons to defined channels just as the
optic stalk provides a physical conduit for retinal axons. In comparison,
ensheathing cells readily migrate away from explants of olfactory
neuroepithelium plated on laminin and as the olfactory nerve pathway also
contains laminin (Gong and Shipley,
1996; Treloar et al.,
1996a; Kafitz and Greer,
1998a) it probably acts in an autocrine loop to stimulate both the
migratory activity and axongrowth promoting properties of ensheathing cells
(Tisay and Key, 1999). The
dorso-lateral surface of the olfactory nerve is bounded by a region of neural
crest-derived mesenchyme that selectively expresses Pax-7
(LaMantia et al.,
2000). When the area of the Pax-7 expressing cells are
experimentally reduced in vitro the trajectory of olfactory axons
arising from neuroepithelial explants is severely disrupted
(LaMantia et al.,
2000). These results suggest that this specialized domain of
mesenchyme provides specific guidance cues for the migratory mass, although
the identity of these molecules remains to be determined. It is clearly not
laminin or CS-56 chondroitin sulphates since this region of the frontonasal
mesenchyme is devoid of these extracellular matrix molecules
(Treloar et al.,
1996a). The importance of the frontonasal mesenchyme in formation
of the olfactory pathway was confirmed, at least in vitro, when this
tissue was prevented from forming in explant cultures. In this case olfactory
sensory axons failed to grow away from olfactory neuroepithelium and instead
whorled on top of the explant (LaMantia
et al., 2000). Interestingly, a similar phenotype is also
observed when ensheathing cells fail to migrate away from explants grown on a
chondroitin sulphate substrate (Tisay and
Key, 1999).
|
Axon sorting in the nerve fibre layer |
|---|
|
TopAbstractIntroductionAxon navigation to the...Axon sorting in the...Axon homing to glomerular...Working models of axon...Where to next?References |
|---|
Upon reaching the rostral telencephalon the olfactory sensory axons and glial cells do not immediately enter into the neuroepithelium of the telencephalon, rather the axons defasciculate and branch (Whitesides and LaMantia, 1996
) and form a transient tissue aggregate referred to as the
migratory mass (Valverde et al.,
1992). This mass is clearly separated from the telencephalon and
its formation occurs at a time when the marginal zone and outer surface of the
presumptive olfactory bulb expresses a `wall' of CS-56 chondroitin sulphates
(Treloar et al.,
1996a). The migratory mass only fuses with the bulb when the
expression of these chondroitin sulphates is downregulated. At this time the
olfactory axons penetrate the marginal zone of the telencephalon and form the
presumptive nerve fibre layer. Although the purpose of this delay in fusion of
the migratory mass with the telencephalon is unknown, one possibility is that
it may allow sufficient time for axon subpopulations to sort out into discrete
fascicles as demonstrated in the insect olfactory system
[(Rössler et al.,
1999); see reviews by (Oland
and Tolbert, 1996; Tolbert,
1998)]. However, in the absence of appropriate markers of axon
subpopulations at these earlier ages in mammals, this hypothesis remains
untested.
After the olfactory axons fuse with the telencephalon their tangential
spread is restricted to the surface of the presumptive olfactory bulb. The
mechanisms limiting the spread of olfactory axons on the telencephalic surface
remains unknown. However, the territory on the rostral surface of the
telencephalon that is destined to become the olfactory bulb is delineated by
the expression of two transcriptional factors, Brn-4 and Tst-1
(Alvarez-Bolado et al.,
1995) and by anosmin-1, an extracellular matrix protein
(Hardelin et al.,
1999). The presumptive olfactory bulb division is also contained
within a much broader domain of neuroepithelium expressing retinoic acid
response elements which may activate specific differentiation events in
response to retinoic acid in the surrounding frontonasal mesoderm
(LaMantia et al.,
1993). Although the significance of these transcriptional factors
in development of the olfactory system remains unknown, retinoic acid has been
implicated in a Pax-6 dependent signalling pathway associated with olfactory
bulb formation (Anchan et al.,
1997). Anosmin-1 is the protein encoded by the gene KAL-1 that is
disrupted in the X chromosome-linked form of Kallmann syndrome
(Hardelin, 2001). In Kallmann
syndrome the olfactory bulbs fail to develop after the olfactory axons are
unsuccessful in penetrating the presumptive olfactory bulb region of the
telencephalon (Hardelin,
2001). Anosmin-1 has both neurite growth promoting as well as cell
adhesion activities, the latter of which is dependent on trans-interactions
with cell surface heparan sulphate and chondroitin sulphate glycosaminoglycans
(Soussi-Yanicostas et al.,
1996,
1998;
Robertson et al.,
2001). Since olfactory sensory axons are rich in heparan sulphate
proteoglycans (Watanabe et al.,
1996) as well as the chondroitin sulphate proteoglycan neurocan
(Clarris et al.,
2000), it is possible that the entry of axons into the presumptive
olfactory bulb is dependent on interactions between these molecules and
anosmin-1.
There is considerable defasciculation and sorting occurring in the
olfactory nerve fibre layer, particularly in its outer region. Subpopulations
of olfactory axons are typically intermixed in the olfactory nerve and only as
they enter the nerve fibre layer do they sort out into molecularly distinct
bundles such as those expressing cell surface carbohydrates
(Figure 1a)
(Key and Akeson, 1993;
Riddle et al., 1993;
St John and Key, 2001a). One
possibility is that the nerve fibre layer is a sorting zone for olfactory
axons in the same way that there is a sorting zone in the olfactory pathway of
insects (Rössler et al.,
1999). This sorting seems to be a necessary prerequisite for
subsequent homing to specific glomerular targets. For instance, the failure of
distinct subpopulations of axons to correctly fasciculate in the nerve fibre
layer of either N-CAM-180 (Treloar et
al., 1997) or galectin-1
(Puche et al., 1996)
deficient mice leads to aberrant glomerular formation. Evidence from other
systems indicates that chemorepulsive forces play an important role in sorting
axons into specific bundles (Simpson
et al., 2000) or defasciculating axons at specific sites
in pathways (Hentschel and van Ooyen,
1999). Little attention has been paid to the role of
chemorepulsive molecules and their receptors in the defasciculation and
sorting of axons in the nerve fibre layer. Defasciculation of axons, for
instance at target sites, is often thought to be mediated by the presence of
local chemorepulsive ligands acting on receptors present on the axons. One
possibility in the olfactory pathway is that chemorepulsive cues may be
expressed by ensheathing cells in the nerve fibre layer and that these
molecules act on receptors expressed by the axons. For instance, the
chemorepulsive receptor neuropilin-1 is expressed by a subpopulation of
olfactory sensory axons that appears to be diverted from ensheathing cells
residing in the ventral olfactory nerve fibre layer which selectively express
the secreted neuropilin ligand Sema3A
(Crandall et al.,
2000). There is still some debate concerning the precise
expression pattern of Sema3A in rat with reports that it is either expressed
solely by olfactory sensory neurons
(Williams-Hogarth et al.,
2000) or also by both plial cells and second-order mitral cells in
the bulb (Pasterkamp et al.,
1998). Nonetheless, olfactory sensory axons are clearly responsive
to Sema3A and display chemorepulsive behaviour to substrate-bound Sema3A
in vitro (Crandall et
al., 2000;
William-Hogarth et al.,
2000). A subsequent and more detailed analysis of expression of
Sema3A in embryonic mice revealed that this ligand was expressed by mitral
cells as well as by ensheathing cells in the outer region of the nerve fibre
layer, in the region where olfactory axons are known to sort out
(Schwarting et al.,
2000). More importantly, analysis of Sema3A homozygous mutant mice
revealed that the sorting of axons within the nerve fibre layer was disrupted
and axons terminated in topographically inappropriate glomeruli
(Schwarting et al.,
2000). Thus, chemorepulsive interactions between axons and
ensheathing cells in the outer region of the nerve fibre layer mediate sorting
out of axons and this sorting is a necessary prerequisite for subsequent
homing of axons to correct glomerular targets. These results are consistent
with the aberrant targeting in N-CAM-180 mutant mice where the loss of
polysialic acid (which regulates the degree of axon fasciculation) in the
nerve fibre layer inhibited sorting of axons in the nerve fibre layer
(Treloar et al.,
1997).
|
While axon sorting is clearly modulated by interactions between axon
receptors and chemorepulsive ligands in the local environment it is also
possible that axons co-express both ligand and receptor. By regulating the
activity of these molecules either directly or indirectly, it is possible to
control the extent of defasciculation in axon tracts
(Lin et al., 1994;
Tang et al., 1994;
Yu et al., 2000).
Interestingly, olfactory sensory neurons express the chemorepulsive receptor
Robo 1 and its ligands slit 1, slit2 and slit3
(Ba-Charvet et al.,
1999; Yuan et al.,
1999). The relative levels of these molecules are known to
influence the fasciculation patterns of axons
(Simpson et al.,
2000). The role of these molecules and of other co-expressed
chemorepulsive receptors and ligands such as the Ephs and ephrins
(St John et al.,
2000; St John and Key,
2001b) needs to be better understood in the olfactory system.
It is becoming clear in other systems that the combinatorial role of
multiple cues is essential for axon navigation
(Winberg et al.,
1998). The relative balance between repulsive and attractive cues
is important for correct guidance and this is achieved by cis-interactions
between receptor domains both at the cell surface and at cytoplasmic face of
the plasma membrane as well as by cross-talk between transmembrane signalling
pathways. For instance, chemoattractive responses mediated by interactions
between the axon receptor DCC and its secreted ligand netrin-1 are silenced by
slit activation of Robo through a Robo/DCC complex
(Stein and Tessier-Lavigne,
2001). Since DCC is expressed by olfactory neurons early in
development (Schwarting et al.,
2001) it is possible that this receptor modulates Robo mediated
behaviour. The role of these and other interactions such as those occurring in
NCAM/axonin-1/NgCAM (Rutishauser,
2000) and neuropilin/plexin/sema
(Takahashi and Strittmatter,
2001) complexes needs to be clarified in the olfactory pathway. It
is intriguing that binding of EphB receptors to ephrin-B ligands can silence G
protein signalling downstream of seven transmembrane G-coupled receptors
(Schmucker and Zipursky,
2001). One of the consequences of this blockage is the silencing
of signalling downstream of the CXCR4 cytokine receptor, resulting in abnormal
granule cell migration in the developing cerebellum
(Lu et al., 2001). It
is tempting to speculate that EphB—ephrin-B interactions in olfactory
axons could modulate signalling downstream of the G-coupled odorant receptors
in growth cones and affect olfactory sensory axon navigation.
|
Axon homing to glomerular targets |
|---|
|
TopAbstractIntroductionAxon navigation to the...Axon sorting in the...Axon homing to glomerular...Working models of axon...Where to next?References |
|---|
The degree to which distinct subpopulations of olfactory sensory axons converge to specific glomeruli in the olfactory bulb was not fully realized until after the cloning of olfactory receptor genes. By using in situ hybridization for detection of receptor mRNA in axons it was possible to show that neurons expressing the same receptor projected specifically to a small number of glomeruli in each olfactory bulb (Figure 2) (Vassar et al., 1994
). Axons typically terminated in a glomerulus on both the
medial and lateral surfaces of the olfactory bulb. The positions of these
glomeruli appeared to be highly conserved between bulbs, both within the same
animal as well as between different animals. This raised the possibility that
odorant receptors were subserving two distinct roles, one at the level of the
nasal cavity in signal transduction and the other in the growth cone as a
guidance receptor to enable axons to target specific glomeruli. The targeting
of axons expressing the same receptor to individual glomeruli was subsequently
more clearly defined using mice engineered to express reporter molecules
specifically in a subpopulation of olfactory sensory neurons
(Figure 1b)
(Mombaerts et al.,
1996). By using homologous recombination, the P2 coding region was
replaced with a construct consisting of the P2 coding region linked to a
tandem array of an internal ribosome entry site recognition (IRES) sequence, a
signalling sequence from the tau gene that traffics proteins into axons, and
finally the LacZ gene. In this P2-IRES-tau-LacZ line of mice,
ß-galactosidase was expressed in the axons of neurons expressing the P2
receptor (P2 axons). Thus, enzyme histochemical staining enabled the complete
trajectory of P2 axons from the olfactory neuroepithelium to two principal
glomeruli in each olfactory bulb to be clearly depicted.
|
A knock-in approach was subsequently used to directly test the hypothesis
that receptor proteins were involved in axon targeting to specific glomeruli.
By replacing the P2 coding region with the M12 coding region it was possible
to examine whether axon targeting was dependent on the type of receptor a
neuron expressed (Mombaerts et
al., 1996). A line of mice was generated in which the P2
coding region was replaced by an M12-IRES-tau-LacZ construct (abbreviated as
M12
P2). If the receptor was instructing the axons to target a specific
glomerulus in the olfactory bulb then P2 neurons should inappropriately
project to the M12 glomerulus in these M12P2 mice. If the receptor was
not involved in targeting then the P2 neurons in the M12P2 mice should
project to the P2 glomerulus regardless of the fact that they were expressing
M12 rather than P2. Instead, it was found that P2 neurons projected somewhere
in-between the P2 and M12 glomeruli, indicating that receptors play a role in
axon targeting but that there were certainly other molecules that also
contributed to the guidance of olfactory axons.
There was controversy concerning the interpretation of these results. The
very notion that receptors involved in transduction of sensory stimuli could
also be involved in targeting was unsettling. Questions were raised as to
whether the knock-in approach disrupted the regulatory sequences which
non-specifically or indirectly affected axon targeting. This was of particular
concern since the insertion of IRES-tau-LacZ into the M12 locus caused
aberrant axon targeting of M12 axons
(Mombaerts et al.,
1996). However, these concerns were partly allayed following
multiple repeats of these swap experiments using both distant and closely
related members of receptor sub-families
(Wang et al., 1998).
Altogether four distinct receptor lines were generated and in all cases axons
targeted inappropriate glomeruli. However, all lines were generated by
modifying the P2 locus and as yet these results have not yet been confirmed
with knock-ins into other receptor genes.
It is still not clear from these experiments whether the odorant receptors are either directly or indirectly involved in axon guidance. One possibility is that the choice of receptor expressed by an olfactory neuron could determine the expression of other guidance receptors. Several possible mechanisms could explain this regulation. First, the coding sequence of receptor genes may contain remote enhancer elements that act at a distance with promoters for guidance genes. These enhancers would have to be present in the coding region since in the receptor knock-in mice only the coding regions were replaced. Second, the odorant receptor mRNA itself may directly modulate expression of other guidance cues, either at the transcriptional or post-transcriptional level. Third, odorant receptor protein or its signalling pathways may regulate guidance receptor activity. However, none of these mechanisms seem very likely and axon targeting is probably best explained by a direct role of the odorant receptor as a guidance receptor at the cell surface. To date, there is still no evidence that odorant receptors are expressed on the growth cones where they would be expected to be functionally active as guidance receptors nor are there any reports that these molecules are neurite-outgrowth promoting using any of the classical substrate-bound in vitro assays. Just how these odorant receptors act as guidance receptors is not at all clear. If they act as receptors for ligands expressed in the olfactory bulb then mutational analysis of extracellular domains in transgenic knock-in mice might provide a means of identifying the binding sites.
The importance of an odorant receptor in axon convergence and glomerular
formation was clearly demonstrated in transgenic mice whose P2 coding region
was either replaced by an IRES-tau-LacZ construct or contained three point
mutations which resulted in expression of truncated receptors
(Wang et al., 1998).
In these mice the axons of P2 neurons lacking a functional P2 receptor were
easily visualized by LacZ histochemical staining. P2 axons in these mice
failed to form glomeruli. They appeared to reach the vicinity of their
topographically correct target site and only then failed to converge, instead
remaining in the nerve fibre layer. Similar results were noted with deletion
of the M12 receptor although this data was not published
(Wang et al., 1998).
Interestingly, the P2 axons in the P2 deletion mice were confined to the nerve
fibre layer in 2-week-old knockout mice in a similar distribution to that
displayed by P2 axons during normal embryonic development
(Royal and Key, 1999). At
E15.5, P2 axons are diffusely distributed in the nerve fibre layer before they
begin to condense and form presumptive glomerular-like loci at E17.5
(Royal and Key, 1999). In the
absence of the P2 receptor, the P2 axons appear to fail to converge and form
these protoglomeruli. It would seem that the receptor protein is critical at
least for the latter stages of axon targeting and convergence. However, it was
never resolved whether the failure of P2 axons to converge was a result of the
upregulation of inappropriate receptors following either deletion of P2 or
expression of truncated receptor.
Although we are concerned here only with axon navigation in the main
olfactory system it is important to briefly examine the effect of
mis-expression of odorant receptors in the vomeronasal sensory neurons on axon
targeting in the accessory olfactory bulb. Deletion of the putative pheromone
receptor VRi2 causes the axons of neurons that should be
expressing this receptor to fail to converge and form specific glomeruli
(Rodriguez et al.,
1999). Instead these axons now broadly terminate throughout the
glomerular layer of the accessory olfactory bulb. When the
VRi2 receptor coding region is replaced by the coding
region of the unrelated M71 odorant receptor the convergence of axons to
specific glomeruli is restored. Although in this case the glomeruli are
distinct from the wild-type VRi2 glomeruli, they are
restricted to a topographically appropriate zone in the accessory olfactory
bulb (Rodrigeuz et al., 1999). The important point here is that the
coding region of M71 is sufficient to allow convergence of axons albeit to the
wrong target sites. These results suggest that the receptor protein is
autonomously responsible for convergence, most likely through
receptor—receptor interactions. It is hard to imagine that the M71
receptor could be binding heterophically with molecules selectively expressed
by subpopulations of axons present in both the main and accessory olfactory
systems. Both odorant receptor and other cues appear to be necessary for
correct targeting after convergence. If odorant receptor type was not involved
in targeting then one would predict that these axons would have innervated
their correct glomeruli after they had converged. Thus, the odorant receptor
appears to have a dual role in both convergence and targeting. It would be
interesting to see whether a VRi2P2 swap would both
maintain convergence and perturb targeting of P2 axons in the main olfactory
system.
What is the cellular source of putative guidance cues for targeting in the
olfactory bulb? One possibility is the undifferentiated neuroepithelial cells
in the presumptive olfactory bulb. When the migratory mass fuses with the
telencephalon, olfactory sensory axons penetrate into the outer telencephalic
wall through holes that emerge in its basement membrane
(Treloar et al.,
1996a). Axons then interact directly with the endfeet of
undifferentiated neuroepithelial cells that span the width of the
telencephalic wall at this stage
(Marin-Padilla and Amieva,
1989). These neuroepithelial cells are prime candidates for
providing topographical cues. Another possibility is that the cues are present
on early differentiating projection neurons, the mitral cells. To begin to
address the role of mitral cells and interneurons in axon targeting Bulfone
et al. examined the trajectory of P2 axons in two lines of mice
lacking either Tbr-1 or Dlx-1 and Dlx-2
(Bulfone et al.,
1998). The Tbr-1 deficient mice had a reduced number of
mitral/tufted cells while the Dlx-1/Dlx-2 deficient mice had a reduced
complement of interneurons. In both mouse lines P2 axons continued to converge
and form glomeruli at positions similar to those of wild-type animals.
Although it is clear that interneurons do not play a role in axon targeting
since they are not generated until well after P2 axons have begun to converge
(Hinds, 1968), the role of
mitral cells still remains uncertain. The Tbr-1-/- mice had a
reduced number of mitral cells that still may have been sufficient to produce
appropriate guidance cues.
Three recent papers (Bailey et
al., 1999; Treloar et
al., 1999; Puche and
Shipley, 2001) have examined in detail the cellular interactions
between olfactory sensory axons and bulb cells occurring during glomerular
formation. These studies have demonstrated that glomerular formation precedes
the compartmentalization of dendritic arborizations of second-order
mitral/tufted cells into glomeruli. Bailey et al. revealed that
olfactory axon arborization appears to be secondary to the formation of tufts
of radial glial cell processes (Bailey
et al., 1999). They proposed that radial glial cells
induced the formation of protoglomeruli independent of any involvement of
mitral cells. However, this observation does not preclude the role of
second-order neurons in the initial targeting events but only in the process
of glomerularization. The cellular basis of targeting remains to be clarified
and increasing attention should be directed at deciphering the role of
undifferentiated neuroepithelial cells, radial glia and mitral cells in this
process.
How specific is targeting by olfactory axons? The view that is generally
exhorted is that all axons expressing the same odorant receptor typically
project to two glomeruli, one on the medial surface and one on the lateral
surface of each olfactory bulb (Mombaerts
et al., 1996; Wang
et al., 1998). The positions of these glomeruli are
considered fixed between bulbs within the same animal as well as between
different animals. Convergence of axons begins as early as E16.5
(Mombaerts et al.,
1996), which led to the idea that olfactory axons form glomeruli
without error. This has been supported by evidence from time-lapse imaging of
single axons arborizing without extensive overshooting in the zebra fish
olfactory bulb (Dynes and Ngai,
1998). Interestingly, while Golgi staining in the early postnatal
rat olfactory bulb failed to detect exuberant axon growth or extra-glomerular
branching (Klenoff and Greer,
1998) a similar analysis in rabbit showed extra-glomerular
branching to inappropriate glomeruli (Yilmazer-Hamke et al., 2000).
Some of the differences in results obtained between these studies could be
attributed to the species analysed as well as to problems with the use of
Golgi staining to observe thin immature axons. The view that axon targeting is
initially highly specific is not supported by dye-tracing experiments.
Detailed analysis of the projections of olfactory axons in the early postnatal
period in both rat and mouse revealed pre-glomerular branching, axons passing
through the glomerular layer without branching, axon branching into two
glomeruli and axons arborizing in glomeruli as well as growing into deeper
layers (Tenne-Brown and Key,
1999). This overshooting and arborization in more than one
glomerulus was also observed for P2 axons by LacZ histo-chemistry
(Royal and Key, 1999).
Moreover, P2 axons were found in a second extra glomerulus at one surface of
the olfactory bulb in 85% of adult animals. The prevalence of these extra
glomeruli was more marked on the medial surface of the olfactory bulb, an
observation subsequently verified in a different colony of P2-IRES-tau-LacZ
mice (Ebrahimi and Chess,
2000). The position of extra glomeruli was highly variable as
opposed to the more constant location of the principal glomerulus
(Royal and Key, 1999;
Schaeffer et al., 2001). In addition, P2 axons sometimes only
partially innervated the extra glomerulus. This result clearly revealed that
glomeruli can be innervated by axons expressing more than a single receptor
protein, an idea initially suggested several years earlier from observations
of a line of transgenic mice expressing LacZ in a subpopulation of olfactory
neurons (Treloar et al.,
1996b) which appears to be under the control of an odorant
receptor gene promoter (Pyrski et
al., 2001). These extra glomeruli are not only formed by P2
neurons; they have also been observed for neurons expressing M72
(Zheng et al.,
2000), M50 and M71 (Lin et
al., 2000), OR-Z6 (Pyrski
et al., 2001) and mOR37
(Strotmann et al.,
2000). Taken together, the overshooting of axons and the
innervation of more than one glomerulus by axons expressing the same receptor
suggests that once axons have reached the vicinity of their target their
ultimate site of convergence can be modulated by additional mechanisms. One
possibility is that neuronal activity and/or level of receptor protein may
influence this process. In transgenic mice, olfactory sensory neurons
mis-expressing an exogenous copy of the odorant receptor MOR28 innervated an
independent glomerulus located near to the wild-type MOR28 glomerulus
(Serizawa et al.,
2000). The distance separating these two glomeruli was determined
by both the genetic background of the endogenous and transgene alleles and by
whether the alleles were tagged with reporter molecules
(Ishii et al., 2001).
For instance, when endogenous alleles were homozygous from the 129 strain and
the transgene was from a C57/B16 background the glomeruli were 
300-400
µm apart. This phenotype is similar to that observed for the P2 extra
glomeruli when P2 axons are tagged with LacZ
(Royal and Key, 1999).
However, the modification of the P2 locus by the knock-in of LacZ is not
solely responsible for the formation of extra glomeruli since these structures
are also present in wild-type animals as revealed by in situ
hybridization (Pyrski et al.,
2001). It seems that differences in genetic background between the
two alleles contributes to the formation of these extra glomeruli.
Interestingly, when one endogenous MOR28 allele was modified to co-express
GFP, axons expressing this allele segregated from axons expressing the
unmodified endogenous allele within the same target glomerulus
(Ishii et al., 2001).
This segregation became more prominent when the maternal and paternal alleles
were from different genetic backgrounds. Thus, it appears that the final
target site is fine tuned according to the genetic background of the receptor
allele which may reflect either differences in amino acid sequence and/or
differences in level of receptor expressed. In each case glomerular targeting
could be affected by resultant differences in either target recognition and/or
neuronal activity dependent mechanisms.
It is now clear that during the early stages of glomerular formation both
olfactory sensory axons (Tenne-Brown and
Key, 1999) and the dendrites of mitral cells
(Malun and Brunjes, 1996)
branch and contribute to multiple glomeruli. Over a period of several days
these branches are withdrawn until each axon and dendrite terminates in a
single glomerulus. Although little attention has been directed towards
understanding the molecular basis of these rearrangements it has been
suggested that interactions between the chemorepulsive Eph receptors and their
ephrin ligands may be involved (St John
et al., 2000). During this period of plasticity, EphA5 is
differentially expressed by mitral cells while its ligands are differentially
expressed by olfactory sensory axons leading to the presence of subpopulations
of glomeruli with high and low levels of these molecules
(St John et al.,
2000). It was proposed that axons expressing high levels of
ligands innervated glomeruli containing dendrites expressing low levels of
EphA5 and vice versa (St John et
al., 2000). However, without knowing the full complement of
Ephs and ephrins expressed by these neurons this hypothesis will be difficult
to test.
What is the role of neuronal activity in axon targeting? This question was
recently addressed in two studies that examined the targeting of P2, M50 and
M72 olfactory neurons in mice deficient in the olfactory cyclic
nucleotidegated channel OCNC-1 (Lin et
al., 2000; Zheng et
al., 2000). While the trajectory of P2 and M50 axons were
unaffected by loss of neuronal activity there were modest defects in the
targeting of M72 axons (Zheng et
al., 2000). In the absence of OCNC-1 there was an increase in
the number of small ectopic glomeruli innervated by M72 axons in comparison to
control littermates. It remains to be determined whether this
activity-dependent fine tuning of targeting occurs for many different
olfactory neuron subpopulations or whether it is instead restricted to only a
small subpopulation. None-theless, the take-home lesson appears to be that not
all neurons are equally dependent on neuronal activity for correct targeting.
One therefore has to be careful when extrapolating from mechanisms identified
for either P2 or M72 to the whole population of olfactory sensory neurons.
Zheng et al. (Zheng et
al., 2000) extended their analysis from complete loss of
neuronal activity to a paradigm where M72 neurons with different levels of
neuronal activity were forced to compete for the same glomerular space. By
taking advantage of the phenomenon of monoallelic inactivation of odorant
receptor genes by olfactory sensory neurons, Zheng et al.
(Zheng et al., 2000)
were able to generate mice with M72 olfactory neurons that were either
OCNC1-negative and tagged with GFP or OCNC1-positive and tagged with LacZ in
an OCNC1-/- background. In these mice the GFP and LacZ positive
axons segregated into distinct glomeruli indicating that the activity of the
M72 neurons governed their final targeting. Whether this is the case for all
olfactory neuron subpopulations remains to be determined. Considering that
addition of reporter tags can affect axon targeting it is possible that this
separate innervation of glomeruli occurs independently of differences in
activity. In an important control, Zheng et al. examined the
targeting of M72 alleles tagged with either LacZ or GFP
(Zheng et al.,
2000). Despite the presence of these reporters, M72 axons
converged to a single glomerulus, indicating that the level of neuronal
activity was the important determining factor in the segregation of axons.
Thus, differences in neuronal activity may explain why olfactory sensory
neurons expressing an odorant receptor MOR28 transgene innervate a glomerulus
adjacent to the glomerulus innervated by wild-type MOR28 neurons when all
alleles are from the same genetic background
(Serizawa et al.,
2000).
The role of neural activity in organization of the olfactory pathway was
also demonstrated in females heterozygous for a mutant allele of the X-linked
OCNC1 (Zhao and Reed, 2001).
In these mosaic animals approximately half of the olfactory sensory neurons
would be expected to express wild-type OCNC1 due to random inactivation of the
gene on one of the two X chromosomes. In neonates, two distinct populations of
glomeruli innervated by either wild-type or OCNC1 negative neurons co-existed.
In contrast, essentially all the glomeruli in adult female heterozygous mice
were wild type and contained OCNC1. There were few OCNC1-negative glomeruli
(which were tagged with LacZ) since the OCNC1-negative neurons were
selectively lost with increasing age. Thus, competition between subpopulations
of neurons with either all or no activity also influences neuronal survival.
However, one should remember that the all-or-none effect of neural activity in
these mice models is artificial. Perturbations to targeting and survival were
only demonstrated in an extreme competitive environment; these results may not
be extrapolated to variations in physiological activity that may arise during
normal experience. It remains unclear whether different physiologically
relevant odour environments can modulate the activity of subpopulations of
neurons sufficiently to affect targeting.
|
Working models of axon navigation |
|---|
|
TopAbstractIntroductionAxon navigation to the...Axon sorting in the...Axon homing to glomerular...Working models of axon...Where to next?References |
|---|
Taken together, the current available data support a hierarchical model of axon navigation to explain formation of the olfactory pathway (Figure 3). First, widely expressed molecules such as N-CAM and TAG-1 probably ensure fasciculation along the olfactory nerve and in the olfactory nerve fibre layer. Second, molecules such as OCAM (Alenius and Bohm, 1997
; Yoshihara et
al., 1997; Nagao et
al., 2000), which are expressed in several zones of the
olfactory neuroepithelium, may be involved in the partitioning of axons into
large spatially defined bundles (e.g. ventrolateral vs ventromedial). Third,
adhesion molecules may also be expressed by all neurons in a single zone where
they could contribute to a zonal partitioning of the olfactory pathway.
Although such zone-specific adhesion molecules have not yet been identified,
there is evidence of the restricted expression of other molecules
(Miyawaki et al.,
1996). However, such zonal-specific expression may be unnecessary
since it is possible that combinations of several adhesion molecules
distributed across several zones could generate sufficient specificity for
zonal identity. The whole purpose of this hierarchy of cell adhesion molecules
is to ensure that olfactory sensory axons arising from one region of the nasal
cavity are guided to a complementary region in the olfactory bulb.
Superimposed on this selective sorting through adhesion interactions are
chemorepulsive interactions that drive the growth of axons away from
inappropriate bulbar regions (Nagao et
al., 2000; Schwarting
et al., 2000). Fourth, subpopulations of axons in each
zone-specific bundle express distinct cell surface carbohydrates such as NOCs
(Dowsing et al.,
1997; Pays and Schwarting,
2000; St John and Key,
2001a). NOCs could form a glycocode which would enable sorting of
axons into discrete smaller fascicles in the nerve fibre layer
(St John and Key, 2001a)
through interactions with carbohydrate-binding proteins such as galectin-1
(Puche et al., 1996;
St John and Key, 1999;
Tenne-Brown et al.,
1998). This step in the hierarchy of interactions would ensure
finer partitioning of the axons, a necessary prerequisite for their subsequent
convergence. Finally, subpopulations of axons expressing specific receptor
proteins then converge and target specific glomeruli.
|
|
This model of hierarchical sorting of axons does not require all axons expressing the same receptor to selectively fasciculate into a single bundle before entering its target glomerulus. What this model predicts is that axons expressing other cell surface recognition molecules (e.g. carbohydrates) will target topographically fixed glomeruli. This prediction was recently confirmed for NOC-3, a unique glycoform of N-CAM (St John and Key, 2001a
). Although axons expressing the same receptor
must interact there is no a priori reason to suppose that these axons
must self-fasciculate. In fact, during embryonic development the glomerulus
forms over several days as the arbors of loosely dispersed and intermixed
subpopulations of axons sort out on the basis of receptor expression and
condense at a specific locus (Royal and
Key, 1999). As a result of the dynamics of this condensation,
axons enter the glomerulus from multiple points. As development proceeds,
axons appear to self-organize into fascicles as they continue to enter
glomeruli. The spatial arrangement of axon entry points into a glomerulus
appears stochastic and hence different for every glomerulus and for the same
glomerulus in different bulbs.
What is the nature of the cue in the bulb that determines the target site?
It is unlikely that there will be 1000 unique molecular markers for each of
the glomerular pairs within the bulb. It is not that this number is too large
since there are already 1000 odorant receptors—it is just that these
molecules would then need to be expressed at topographically invariant points
in each bulb, a highly uneconomical scenario in terms of molecular
interactions. The molecular markers of glomerular position could not be
stochastically expressed in the bulb, like the receptors are in the olfactory
neuroepithelium, since glomerular position is highly stereotypical. By
observing the massive convergence of P2 axons in adult animals it is tempting
to speculate the presence of a soluble tropic substance emanating from
glomeruli and attracting axons to their target. However, as discussed above,
axons do not home to a point source during embryonic development but rather
reach a general address and then condense gradually into a glomerulus. We
instead favour the postulate that glomerular target sites are defined by the
expression of two overlapping gradients of ligands or cues that are
distributed over the glomerular surface of the olfactory bulb
(Gierer, 1998).
In the retino-tectal pathway, target cues exist as linear gradients across the ventrodorsal and anterioposterior axes of the optic tectum. If a similar gradient model was present on the olfactory bulb, duplicate gradients would need to exist on both the medial and lateral surfaces since olfactory axons expressing the same receptor target glomeruli on both of these surfaces. We propose a variant of this model based on gradients originally set up from point sources. In this model, two point sources positioned above one another in the dorsoventral axis on either side of the rostrolateral surface of the telencephalon would create a symmetrical co-ordinate system of radially dispersed cues, with lower levels of cues present at positions more distant from the source (Figure 4). Every glomerular position on the surface of the presumptive olfactory bulb (e.g. see arrowheads pointing to blue glomeruli in Figure 4) would correspond to a particular level of each of the two cues (coloured red and green in Figure 4). Since the cues are radially dispersed, there would be two points in space represented for every combination of the two cues, one on the medial surface and one on the lateral surface of the presumptive olfactory bulb (see arrowheads pointing to blue glomerular targets in Figure 4). This two-dimensional topographical map of glomerular targets is initially present on the telencephalic surface but is converted into a three-dimensional map as the bulb evaginates. If the midline of the bulb evaginates from a line connecting the two point sources, near identical gradient cues would be present on both the medial and lateral surfaces of the bulb. In this way, complementary blue glomeruli would be present on either surface of the bulb.
The final stereotactic position of glomeruli would depend on the ultimate
shape of the bulb. If one side of the bulb grew disproportionately more than
the other, there would be a skew in the absolute glomerular positioning
between the medial and lateral surfaces. This appears to be the case since
identified glomeruli never appear to be at the same rostrocaudal or
dorsoventral position on the medial and lateral surfaces of the bulb. In fact,
for all glomeruli examined to date [P2
(Mombaerts et al.,
1996); M72 (Zheng et
al., 2000); MOR10, MOR18, MOR28, MOR83 and A16
(Tsuboi et al.,
1999); P3 (Wang et
al., 1998)], the medial glomerulus is always located slightly
more caudally along the rostrocaudal axis than its counterpart on the lateral
surface. It appears that the gradient has uniformly shifted caudally on the
medial surface. The two point source gradient model also predicts
that the relative positioning of identified glomeruli will be the same on
either surface of the bulb, which is supported by the detailed mapping of five
glomeruli in the mouse olfactory bulb
(Tsuboi et al.,
1999). In addition, this model predicts that the midline will
contain glomeruli which are not duplicated since at this position the
intersection of gradients between the two point sources produces only
a single level of each ligand. To date, available data confirms this
prediction. Two distinct subpopulations of olfactory sensory neurons
expressing either mOR37 (Strotmann et
al., 2000) or OR-Z6
(Pyrski et al., 2001)
innervate single glomeruli in the ventral midline of the olfactory bulb. There
was some variability in the number of these glomeruli and for mOR37A 20%
of bulbs contained two glomeruli. This would be expected for axons targeting
the midline since small variations in the morphology of the bulb could easily
lead to a flattening of the gradient in this midline region and hence result
in multiple glomeruli. Consequently, one would also expect to have some
variability in the relative positioning of glomeruli in the midline, which
proved to be the case when glomerular targets of several members of the mOR37
receptor subfamily were mapped (Strotmann
et al., 2000). All these observations are consistent with
the duplication of gradients on both the medial and lateral surfaces of the
bulb as predicted by the hypothesis of radial gradients arising initially from
two point sources. Is there any evidence for point sources of gene
expression on the rostrolateral surface of the telencephalon? Two
transcriptional factors Brn-4 and Tst-1 are both expressed as discrete patches
in the region of the presumptive olfactory bulb in the early embryonic
telencephalon (Alvarez-Bolado et
al., 1995). However, it remains to be determined whether
Brn-4 and Tst-1 are expressed in overlapping gradients or whether they
regulate the release of factors that set up gradients across the surface of
the telencephalon, either of which would support the two point source
hypothesis.
The hierarchical and two point source models of guidance attempt
to explain the ability of olfactory sensory axons to sort out and converge to
form specific glomeruli. Although correct sorting of axons in the nerve fibre
layer appears to be necessary for the subsequent receptor-mediated homing of
axons to their topographic target site in the olfactory bulb
(Puche et al., 1996;
Treloar et al., 1997;
Schwarting et al.,
2000), it is interesting to ask why this sorting is necessary in
the first place. It could be argued that sorting is not needed and is merely a
passive by-product of axons actively homing to target signals present in the
bulb. This appears to be the case at least in the retino-tectal pathway where
fibre order in the optic nerve and tract as well as their point of entry of
retinal axons into the tectum does not affect the topographic targeting of
these axons (DeLong and Coulombre,
1967; Fujisawa, 1981; Trowe
et al., 1996). Provided retinal axons reach the tectum
they seem capable of correct targeting due to the existence of complementary
gradients of guidance receptors on retinal axons and ligands on tectal cells.
Thus, retinal axons expressing a distinct level of guidance receptor will
either migrate up or down concentration gradients of ligands in the tectum
with little concern for their point of entry into that gradient in order to
reach their target site.
If there is a ligand gradient over the glomerular surface of the bulb why
then do axons depend on prior sorting in the nerve fibre layer to reach their
target? Why don't olfactory axons behave like retinal axons which do not
require sorting cues to correctly home in on their target site in the tectum?
One possibility is that guidance receptors on olfactory axons are only able to
respond to a narrow range of ligand levels in this gradient. Hence axons need
to be sorted and directed to the vicinity of the target site first by other
guidance cues. Thus, when axons find themselves in the wrong region they are
incapable of navigating back to their correct target. This could explain why
in the receptor swap experiments
(Mombaerts et al.,
1996) M12P2 axons never reach the M12 glomerulus but instead
converge prematurely at inappropriate sites. Since the hierarchical sorting of
these axons is dictated by their original P2 identity, M12P2 axons will
be guided to the broad vicinity of the P2 glomerulus. The receptor swap
experiments in the accessory olfactory system have taught us that receptors
act autonomously to cause convergence of axons expressing the same receptor.
Hence, M12P2 axons will converge but at inappropriate sites. However, if
the M12P2 axons could not read the local ligand gradient why do they
still terminate in topographically fixed positions? In fact, these
inappropriately targeted glomeruli appear to be in the same relative position
with respect to the P2 glomerulus on both the medial and lateral surface of
the olfactory bulb (Mombaerts et
al., 1996; Wang et
al., 1998). Thus, it is most likely that the M12P2
axons probably can read the local gradient but are forced to adopt a new
position due to the confines placed on them by other guidance molecules
expressed by the P2 neurons.
An alternative explanation for why axons need to be presorted in the
olfactory nerve fibre layer in order to reach their correct target has to do
with the location of the putative guidance cues in the olfactory bulb. In the
tectum the guidance cues seem to be present on at least the astroglial cells
present in the outer plexiform layer—a location that facilitates direct
interactions with retinal axons (Davenport
et al., 1996; Braisted
et al., 1997; Stier
and Schlosshauer, 1999). However, the outer olfactory nerve fibre
layer of the olfactory bulb, which is derived from the olfactory
neuro-epithelium, seems to lack astroglial cells during targeting
(Bailey et al., 1999;
Treloar et al., 1999;
Puche and Shipley, 2001).
Within the olfactory nerve, axons expressing the same odorant receptor are
widely separated across different filia olfactoria (bundles of axons enwrapped
by the processes of individual ensheathing cells) as well as across different
fasciculi (bundles of filia olfactoria). These fasciculi pass through the
numerous foramina in the roof of the nasal cavity and then enter the olfactory
bulb at multiple points along its rostrocaudal length. As the olfactory nerve
fibre layer grows in thickness the later growing axons would reach the surface
of the bulb lacking any guidance cues associated with astroglia, unlike in the
tectum where axon—glial interactions continue throughout formation of
the retino-tectal pathway (Vanselow
et al., 1989). Therefore defasciculation and sorting of
olfactory sensory axons would be essential for widely dispersed axons to reach
cues lying deeper in the bulb.
|
Where to next? |
|---|
|
TopAbstractIntroductionAxon navigation to the...Axon sorting in the...Axon homing to glomerular...Working models of axon...Where to next?References |
|---|
Although most of the cellular and molecular events underlying axon targeting in the mammalian olfactory system remain unknown, the models outlined here at least provide a framework to formulate predictions that can be empirically tested. For instance, what is now needed is a systematic search for molecules expressed from point sources on the surface of the rostral telencephalon. To start with, Brn-1 and Tst-1 are two possible candidates and mis-expression of these molecules should alter the topography of the olfactory projection if they are somehow involved in the formation of gradients of guidance cues. Alternatively, the two point source hypothesis predicts that there will be differences in expression of regulatory molecules between the ventral and dorsal regions of the presumptive olfactory bulb region in the telencephalon so it may be possible to examine for differential expression across this axis. The role of putative targeting cues in the retino-tectal pathway has been tested by mis-expressing molecules in the tectum using micro-injected retroviruses (Nakamoto et al., 1996
). A similar approach could be used in the adult olfactory
bulb by taking advantage of the unique regenerative capacity of olfactory
neurons. There are now genetic means of synchronizing the ablation and
subsequent regeneration of select subpopulations of olfactory sensory neurons
expressing reporter molecules (Gogos
et al., 2000). Since these regenerating olfactory axons
are able to enter the adult olfactory bulb and target the site of their
original glomerulus they could be used as an assay system to examine the role
of numerous target derived guidance molecules. The alternative approach is to
mis-express putative guidance receptors or dominant negative forms of these
molecules on the surface of olfactory axons in transgenic mice using a global
promoter such as that for OMP (Walters
et al., 1996). Promoters for zone-specific molecules such
as OCAM and neuropilin-1 have yet to be defined but when available they should
provide alternate means of regulating the spatial expression pattern of
guidance molecules in the olfactory system. This approach could be used to
test the role of candidate guidance molecules such as cell surface
carbohydrates, OCAM and neuropilin-1. Ideally, it would be more appropriate if
promoters for various odorant receptors were available so that molecules could
be mis-expressed in specific subpopulations of olfactory neurons by
transgenesis. Evidence to date indicates that the most of the regulatory
elements are within 7 kb upstream of the coding region of the odorant receptor
(Qasba and Reed, 1998).
Although this fragment recapitulates random expression in a subpopulation of
olfactory neurons, it does not correctly specify zonal expression and it does
not limit expression to the correct subpopulation. Because of this
unpredictability in expression pattern this 7 kb promoter region is not
appropriate for studies looking at targeting of specific subpopulations of
axons.
Interestingly, in ovo electroporation was recently used to
mis-express dominant negative neuropilin-1 in chick olfactory axons
(Renzi et al., 2000).
Although this technique in chick does not provide the same level of resolution
capable by transgenesis in mouse, it at least demonstrates proof of principle
for the mis-expression approach. The advantage of using a knock-in strategy
rather than transgenesis (apart from the obvious fact that promoters do not
have to be defined) is that it is possible to create a competitive environment
where wild-type axons are competing with genetically modified axons for the
same glomerular space. Since odorant receptors are monoallelically expressed
(Chess et al., 1994),
each olfactory neuron subpopulation consists of a mosaic of neurons randomly
expressing either one allele or the other but never both. Consequently in mice
heterozygous for a knock-in in one receptor gene, approximately half of the
neurons from that subpopulation would be expressing the wild-type gene while
the other half would be expressing from the genetically modified locus. In
this way it is possible to knock-in dominant negative forms of guidance
molecules in half of the subpopulation of, for example, P2 neurons. Such a
competitive environment is advantageous since it also allows one to determine
whether axon guidance molecules act cell autonomously. By directly comparing
axon trajectories between wild-type and mutated axons in the same animal
critical guidance points could also be readily identified. One c