3.3 The Main Olfactory System
3.3.4 Projections towards the Olfactory Bulb
In the MOE, ORs are expressed in topographically overlapping zones (Ressler et al., 1993; Vassar et al., 1993; Strotmann et al., 1994b). When leaving the sensory epithelium, unmyelinated axons sent by OSNs form small diameter bundles (5µm) that distally coalesce into larger ones (20µm), to finally asso-ciate into the olfactory nerve (or cranial nerve I). This latter passes through the cribriform plate and reaches the OB, where it defasciculates allowing ax-ons to reach the external layer of the OB. There, like-axax-ons coalesce and form spherical structures called glomeruli (Nedelecet al., 2005). Like-axons means fibers emanating from OSNs expressing the same OR. To each glomerulus will thus correspond a given OR gene (Ressler et al., 1994; Vassar et al., 1994). Two glomeruli per bulb are innervated by fibers expressing a given
receptor. They are organized symmetrically between the medial and lat-eral parts of the bulb, and between bulbs. Thus, to each OR, in a given animal, correspond four glomeruli (Mombaertset al., 1996). The spatial seg-regation between medial and lateral glomeruli reflects the division present in the MOE: septal OSNs usually project their axons towards medial glomeruli whereas OSNs in the turbinates form lateral glomeruli. The position of the projections appears to be somewhat dependent on the OR expressed since OSNs that transcribe structurally-related ORs send their axons to neighbor-ing glomeruli (Tsuboi et al., 1999). A remarkable tool allows to visualize the precise map corresponding to specific ORs. It consists in several knock-in mouse lknock-ines, knock-in which different ORs are tagged with bicistronic markers (Mombaerts et al., 1996) (Figure 3.3 on page 18).
Figure 3.3: Olfactory sensory neurons project towards the olfactory bulb. The class II M71 OR gene has been tagged with the taulacZ marker. (A) Lateral view of an M71 -IRES-taulacZ mouse, with the MOE on the left and the OB on the right. (B) Dorsal view of the OB. Modified from Feinsteinet al., 2004.
In the OB, glomeruli synapse with periglomerular cells and second-order neurons, the mitral and tufted cells (Shepherd, 1972). The cell bodies of periglomerular cells encircle glomeruli, while the ones of mitral cells are lo-cated deeper in the bulb. Tufted cells, which are similar but smaller than mitral cells, project into the “inner plexiform layer”: they are responsible for
the interactions between glomeruli innervated by OSNs expressing the same OR (Belluscioet al., 2002; Lodovichiet al., 2003). The major neurotransmit-ter involved in the communication between these cells isγ-aminobutyric acid (GABA). It is released by periglomerular cells and causes three effects: a) in-hibition of apical mitral/tufted cells; b) retrograde inhibition of the sensory input and c) lateral signaling into neighboring periglomerular cells. There is a sort of “negative feedback loop” starting after activation of mitral cells, since the connections between external tufted cells can modulate mitral and tufted cell activation (Belluscioet al., 2002; Lodovichiet al., 2003; Zhou and Belluscio, 2008). Moreover, granule cells that do not possess axon-like struc-tures laterally inhibit the activation of neighbouring secondary neurons. It seems that this loop is used to “promote” mitral cells exhibiting high activity (Shepherd et al., 2007).
Axon Guidance towards the Olfactory Bulb
It is still unclear how axons sent by OSNs find their way into well-defined zones in the OB. A correct targeting is of course of major importance since the coding strategy used by the olfactory system depends entirely on this wiring.
To end up in a defined position, OSNs must direct their axons along three main axes: the dorsoventral, anteroposterior and mediolateral axes. Different molecules are apparently involved in this process. These include semaphorin and tyrosine kinase receptors, cAMP, homophilic adhesive molecules and, naturally, the ORs themselves.
We must keep in mind that the question of OSN guidance is at least double. First, how do like-fibers end their course in a restricted part of the OB? And second, how do like-axons coalesce into a glomerulus?
Robo2 is a receptor for the Slit chemorepellents. Several studies indi-cate that this ligand/receptor couple plays a major role in axon growth and guidance in the brain of zebrafish (Lee et al., 2001; Miyasaka et al., 2005), Drosophila (Jhaveri et al., 2004) and rodents (Marillat et al., 2002). During
development and adulthood, Robo2 expression forms a high dorsomedial to low ventrolateral gradient across the MOE, a gradient which is reflected in the OB by the Robo2 ligand, Slit1. Knock-out mice for both receptor and lig-and result in mistargeting of a subset of dorsally projecting OSNs during the development of the lateral olfactory tract. Axonal projections coming from these neurons ends up forming glomeruli in more ventral regions of the bulb, suggesting that interactions between Robo2 and Slit1 regulate the dorsoven-tral position of axons coming from at least a subset of OSNs (Fouquetet al., 2007; Cho et al., 2007; Nguyen-Ba-Charvet et al., 2008).
Two members of the neuropilin family, Neuropilin-1 and Neuropilin-2, are also expressed in a graded manner by OSNs which project their axons to spe-cific zones of the OB. Neuropilin-1 and -2 are membrane-bound receptors for secreted class II semaphorins (Sema3), Sema3A and 3F, respectively. Axons, according to the level of Neuropilin-1 expressed by their corresponding OSNs, form glomeruli along the anteroposterior axis (Schwartinget al., 2004). Sim-ilarly, the expression of Neuropilin-2 forms a gradient in the MOE, being abundant in the ventrolateral and almost absent in the dorsomedial regions (Norlin et al., 2001). In the OB, glomeruli corresponding to Neuropilin-2-expressing OSNs are located in two ventral areas: the first being at the most anterior part and the second in close contact with the telencephalon (Walz et al., 2002). In absence of their ligands, neuropilin-expressing OSNs are mistargeted to regions where they normally do not project (Schwart-ing et al., 2004). However, it seems that a lack of Sema3A expression also affects glomeruli which are not innervated by axons coming from Neuropilin-1-expressing OSNs: there is a reduction in both their number (Taniguchi et al., 2003) and size, as well as problems in their formation (Schwarting et al., 2004). Similarly, the misrouting in absence of Neuropilin-2 seems to be caused by a loss of repulsion between mitral and tufted cells secreting Sema3B and Sema3C (Walz et al., 2002).
cAMP levels are important for axonal targeting. This is true also in the
main olfactory system, where the deletion of ACIII leads to major targeting aberrations. These include innervation of novel bulbar territories by OSNs, and more importantly, incomplete convergence and innervation of glomeruli by mixed OSN populations (Zou et al., 2007; Dal Col et al., 2007). The lack of ACIII directly affects Neuropilin-1 transcription, since no neuropilin is expressed by OSNs in these mutants. In the MOE, cellular responses mediated by cAMP involve among other molecules the Protein Kinase A (PKA). Mice expressing a dominant-negative PKA have difficulties forming a proper glomerular map along the anteroposterior axis, a situation reminiscent of the ACIII knock-out mice (Imaiet al., 2006; Chesleret al., 2007; Zouet al., 2007; Dal Col et al., 2007).
Based on these results, the following model has been proposed: the intrin-sic activity of an OR determines a specific intracellular cAMP level, which in turn controls the relative expression of guidance molecules, such as neu-ropilin, via both PKA and the cAMP response element (CREB) (Imai et al., 2006; Imai and Sakano, 2009) (Figure 3.4 on page 22).
Eph receptors constitute the largest group of the tyrosine kinase family.
They bind specifically to membrane-associated ligands, the ephrins, and are known to play a role in the formation of topographic maps in sensory and motor systems (Wilkinson, 2000). Several Ephs have been reported to be expressed in the main olfactory system, at different times and in different locations, together with their ligands (St John et al., 2000; St John and Key, 2001; St John et al., 2002). Eph4, for example, is present on olfactory ax-ons only during development, although it is expressed throughout the entire life in the MOE. EphA7 and EphB2 are exclusively expressed in OSN cell bodies, but never in axons. Two ligands, ephrin-A2 and ephrin-B1, are con-tinuously present in cell bodies, but their presence on axons is only transient or restricted to embryonic life, respectively. Both axons and glomeruli of adult mice express only two ephrins, ephrin-A3 and -A4 (Zhang et al., 1996;
St John et al., 2002; Cutforth et al., 2003).
In glomeruli, ephrins are transcribed at different levels depending on
v v v
Figure 3.4: OR/cAMP-directed glomerular positioning along the anteroposterior axis.
During the growth of OSN axons, the intrinsic activity of each OR generates a unique level of cAMP signals. This signal is then converted to a specific level of axon guidance molecules via cAMP-dependent protein kinase A (PKA) and cAMP response element binding protein (CREB). Several guidance molecules, such as Neuropilin-1, navigates OSN axons along the AP axis according to their expression levels. Modified from Imai and Sakano, 2009.
the glomerulus analysed. Homotypic glomeruli (i.e. innervated by OSNs expressing the same OR) exhibit similar ephrin levels. Overexpression of ephrin-A5 along with the class II P2 OR in heterozygous compound mice (P2 -IRES-GFP ×P2 -IRES-ephrin-A5-IRES-taulacZ) causes the glomeruli corresponding to the overexpression to be shifted anteroventrally with re-spect to normalP2 glomeruli (Cutforth et al., 2003). These two genes share another mechanism of differential regulation, this time linked to neuronal activity: experimental data showed that, following naris occlusion (which should reduce neuronal activity in OSNs), ephrin-A5 transcription is upreg-ulated, while Eph-A5 is downregulated (Serizawa et al., 2006).
Kirrel-2 and Kirrel-3 are members of the immunoglobulin superfamily and were shown to represent adhesive cell-recognition molecules in Drosophila.
They are both present in the MOE and in glomeruli, and show a
comple-mentary expression: to high levels of 2 correspond low levels of Kirrel-3, and vice versa (Serizawa et al., 2006). Because of the nature of these molecules, they may mediate homophilic interactions between two neighbor-ing glomeruli or two different areas of the same glomerulus. As for Eph and ephrins, neuronal activity affects the transcription of both genes (Serizawa et al., 2006).
Finally, the OR itself may help axons to find their correct position into the OB, or to coalesce. This view is supported by the presence of ORs in axonal endings (Barnea et al., 2004; Feinstein et al., 2004; Strotmann et al., 2004). Various experiments have been conducted to test this hypothesis.
These include: a) the swap of OR coding regions with other ORs or GPCRs.
Depending on the proximity of the swapped receptor, like-axons form novel glomeruli, or coalesce into the glomerulus corresponding to the endogenous swapped allele. Divers GPCRs have been swapped, including the vomero-nasal receptor V1RB2 and the β2 adrenergic receptor (Wang et al., 1998;
Feinstein and Mombaerts, 2004; Feinstein et al., 2004). b) The deletion or mutation of OR coding regions, which lead to a complete disorganization of the projection map, likely reflect the coexpression of novel OR alleles (Mom-baerts et al., 1996; Serizawa et al., 2003; Lewcock and Reed, 2004; Barnea et al., 2004; Feinstein et al., 2004; Strotmann et al., 2004; Chesler et al., 2007). Because of this second choice mechanism, it is difficult to obtain
“empty” neurons and thus to investigate the fate of neurons lacking an OR.
While the bases of the mechanism involving the OR in guidance are still unknown, it seems that the expression of a given OR directs the coalescence of like-axons. We however know that this mechanism is dependent on the coupling of the OR to the G-protein. OSNs from transgenic mice in which the DRY motif of a given receptor (mandatory for G-protein coupling) is changed into RDY show no coalescence (Imaiet al., 2006). In contrast with other OR mutant mice, in this case a second OR is not expressed. Axonal fibers do not enter the glomerular layer and instead stop in the anterior part of the OB. Supporting the role of a G-protein-mediated cascade, the constitutive
expression of the active form of the Gαs-protein (which, similarly to Gαolf is associated with an OR) can rescue the inability to converge in the bulb (Imai et al., 2006). Thus, the OR activity can apparently regulate axon guidance and convergence. The OR role could, as previously indicated, determine a specific level of cAMP, which would in turn be responsible for the expression of other molecular determinants such as neuropilins (Imai et al., 2006; Imai and Sakano, 2009) (Figure 3.4 on page 22). It should be noted that this hypothesis would require an extremely precise regulation of cAMP levels, with no overlap between OSNs expressing very related ORs. Knowing that thousands of OSNs expressing a given receptor must find their way towards the bulb during development, and that extremely rare targeting “mistakes”
are observed, it seems quite unlikely that such a simple mechanism is at work.
In favor of a direct role played by the OR, or at least of an “attractive identity” provided by the expressed receptor, is the observation that OSNs expressing transgenically a given OR and converging to a place usually not innervated by OSNs expressing the endogenous OR “attract” the fibers ex-pressing the endogenous allele to the ectopic glomerulus (Vassalliet al., 2002;
Rothman et al., 2005).
Arguing against a direct role played by ORs in guidance are a few intrigu-ing observations. First, as indicated previously, expression of an active Gαolf or Gαs with different ORs leads to convergence of non-like-axons (Imaiet al., 2006; Chesler et al., 2007). Second, the constitutive expression of an active Gαs with a non-functional OR also leads to convergence (Imai et al., 2006).
Third, different levels of expression of a given OR lead to the generation of distinct glomeruli (Feinstein et al., 2004).
Today, what regulates convergence of axons coming from OSNs expressing the same OR is therefore not known.