Subpopulations in the Main Olfactory Epithelium

Dans le document Vomeronasal receptors: from monogenic expression to axon guidance (Page 25-32)

3.3 The Main Olfactory System

3.3.2 Subpopulations in the Main Olfactory Epithelium

Conventional OR-expressing Neurons

Most mature OSNs express ORs. OR transcription is limited to zones in the MOE, whose position depends on the expressed OR. Each of these zones covers about a third of the MOE, and is parallel to the cribriform plate.

OR37-expressing Neurons

Some OSNs escape this band-like expression pattern such as for example OR37-expressing neurons which are localized in patches (Strotmann et al., 1992, 1994b,a). OR37 members differ from classical ORs by their extracel-lular loop (EC) 3, which contains a small amino acid insertion. Apart from this peculiarity, they behave like conventional ORs since they show monoal-lelic (and probably monogenic as well) expression (Strotmann et al., 2000), and their timing of expression resembles that of other OR genes (Conzel-mann et al., 2000). In addition, axons corresponding to OR37-expressing OSNs converge into ventral and homogeneously innervated glomeruli (Strot-mann et al., 2000). The role played by OR37-expressing OSNs is unclear, but since the OR37 family is conserved among several mammalian species (Kubick et al., 1997; Hoppe et al., 2006), neurons expressing these receptors may represent an important and specialized class of chemosensors.

Guanylyl Cyclase D-expressing Neurons

A posterior region of the MOE, near the cribriform plate, harbors a distinct population of OSNs. These do not express ORs nor elements of the canoni-cal olfactory transduction pathway on their cilia. Instead, key players of the guanylyl cyclase cascade are present: the guanylyl cyclase D (GC-D), the cGMP-stimulated phosphodiesterase PDE2 and the cGMP-dependent chan-nel CNGA3 (F¨ulle et al., 1995; Juilfs et al., 1997; Meyeret al., 2000). Due to the lack of known chemoreceptors expressed by these cells, it is thought that GC-D may play this sensory role. This view is consistent with the topology of GC-D, since guanylyl cyclases are single transmembrane domain proteins containing an extracellular ligand binding moiety and an intracellular part with two domains (a protein-kinase-like and a cyclase catalytic domain).

GC-D neurons project their axons to a very peculiar region of the OB called the “necklace glomerular complex”, which resembles a circular chain formed by fibers from both lateral and medial surfaces of the OB (Zheng and Jourdan, 1988; Shinoda et al., 1989; Juilfs et al., 1997). A subset of

the GC-D-expressing glomeruli, called the “modified glomerular complex”, has been reported to play a role in suckling behavior of rat pups (Teicher et al., 1980; Greer et al., 1982; Shinoda et al., 1989, 1993), thus suggesting that GC-D neurons may be able to recognize pheromone-like compounds in young animals. However, no evidence supports this hypothesis. On the con-trary, recent experiments indicate that Gucy2d (the GC-D gene) knock-out mice are not deficient in suckling or mating behaviors (Leinders-Zufallet al., 2007). Two alternative roles potentially played by GC-D neurons have been suggested. First, these neurons express carbonic anhydrase type II (CAII) and can detect atmospheric concentrations of CO2, a characteristic which is lost in CAII knock-out mice (Hu et al., 2007; Sun et al., 2009). Second, GC-D neurons are possibly involved in the evaluation of the metabolic status of an individual, since they can detect two intestinal hormones, uroguanylin and guanylin (Leinders-Zufall et al., 2007).

The projection pattern of GC-D glomeruli is, unlike what is observed for OR-expressing OSNs, not dependent on the expression of GC-D (Leinders-Zufall et al., 2007). Their adequate formation is however dependent on the expression of Neuropilin-2 (Walz et al., 2002).

Trace Amine-Associated Receptor-expressing Neurons

A subpopulation of OSNs expressing trace amine-associated receptors (TAARs) have been recently described. All members but one of the TAAR family are exclusively expressed in the olfactory system (Liberles and Buck, 2006). These receptors are coexpressed with Gαolf and share very low amino acid identity with ORs, lacking typical OR motifs. In general, they are mu-tually exclusively expressed in specific subset of OSNs, and are usually not coexpressed with ORs (Liberles and Buck, 2006). TAARs respond to three volatile biogenic amines: isoamylamine, trimethylamine and β-phenylethyl-amine. These molecules are present in urine of sexually mature male mice and may therefore act as sex pheromones (Liberles and Buck, 2006).

3.3.3 Odorant Receptors

ORs are 7TM G-protein-coupled receptors and are encoded by the largest gene superfamily present in the rodent genome (Buck and Axel, 1991): to date, the mouse repertoire comprises more than 1000 intact OR genes and around 300 pseudogenes, grouped into over 150 families (Zhanget al., 2004a, 2007). They are present on most autosomes and members of a given family cluster along the genome. This situation reflects, at least partly, their com-mon duplicative genomic origin (Zhang and Firestein, 2002; Godfrey et al., 2004).

Based on phylogenetic analyses, ORs can be further subdivided into two main groups, class I and class II. Class I ORs are “older” genes and are present in aquatic species. They recognize water-solubles molecules (Malnic et al., 1999; Kobayakawa et al., 2007). Class II ORs, on the contrary, are predominant in terrestrial species and recognize airborne molecules (Grus and Zhang, 2008). The mouse repertoire is composed primarily of class II ORs (Shi and Zhang, 2007). This division at both the phylogenetic and odorant-binding levels is also reflected in the mouse nasal cavity: class I ORs are only expressed in the dorsal zone of the MOE (Zhang et al., 2004b).

As previously mentioned, in the MOE, most of the ORs are expressed after birth (Zhang et al., 2004b). OR transcription starts in immature post-mitotic neurons and continues in mature OSNs, a population that can be easily identified as it expresses the olfactory marker protein (OMP) (Margo-lis, 1972). Each OR is transcribed randomly and neurons expressing this OR form a distinct band along the dorsomedial-ventrolateral axis of the MOE (Miyamichi et al., 2005). To each OR corresponds a specific probability to be chosen. However, ORs are also expressed in other tissues such as testes (Spehret al., 2003). Although OR genes contains multiple exons, their coding sequence is usually fully included in the last exon. Variable 50 and 30 UTRs have been observed for some OR genes, a variability obtained via alternative splicing or differential polyadenylation (Young et al., 2003; Michaloskiet al., 2006).

OR Gene Transcription

At the base of our current understanding of olfactory perception is the ex-pression by OSNs of a single or a very limited number of OR genes. This leads to the establishment of multiple parallel lines, each composed of a few thousand individual but functionally similar sensory neurons. Not only are ORs monogenically transcribed, but they are also monoallelically expressed (Chesset al., 1994; Serizawaet al., 2000; Ishiiet al., 2001; Vassalliet al., 2002;

Feinstein and Mombaerts, 2004; Roppolo et al., 2007). This mechanism is also referred to as the “one neuron, one receptor rule” or as “monoallelic” and

“monogenic” expression. Despite few exceptions described (Rawson et al., 2000; Serizawaet al., 2000, 2003), a strict and still not understood mechanism is active to ensure that only one receptor is expressed per neuron.

In contrast to the immune system, genomic rearrangements are not re-sponsible for this situation: mice cloned from a postmitotic OSN show a nor-mal OR expression pattern and not a monoclonal one (Eggan et al., 2004;

Li et al., 2004). Asynchronous replication for the ORs has been observed (Chesset al., 1994), but is thought to reflect the organisation of ORs in clus-ters (Gimelbrant and Chess, 2006). Two OR genes can be coexpressed, but usually only if one of them is non-functional (Serizawaet al., 2003; Feinstein et al., 2004; Shykind et al., 2004).

Thus, in a given OSN, some sort of signal apparently blocks the expres-sion of a second OR when another one is already functionally expressed.

Exceptions to this rule, i.e. cotranscription of two OR genes, were observed in the following conditions: a) a transgenic non-functional OR allele and the endogenous intact counterpart (Qasba and Reed, 1998; Serizawaet al., 2003);

b) an endogenous pseudogenes and a transgenic functional allele (Qasba and Reed, 1998; Serizawa et al., 2003); c) a transgene inserted in an OR clus-ter and the neighboring OR genes (Pyrski et al., 2001; Lewcock and Reed, 2004); d) functional and non-functional M71 (Feinstein et al., 2004) and e) any OR when the expression of the second one is indirectly dependent of a tetracycline-transactivator-based system (Nguyen et al., 2007; Fleischmann

et al., 2008).

Models for OR Gene Regulation

Two models have been proposed to explain these results: “negative feedback”

and “negative selection”. In the former case and as discussed, as soon as an OR is expressed, it emits a signal which blocks the transcription of any other OR (Serizawa et al., 2003; Lewcock and Reed, 2004). In the latter model, cells can naturally and randomly coexpress multiple OR genes, but they are never detected because they die (Mombaerts, 2004). This is not the case if between the two OR genes only one is functional: in this case, the cell does not die (since it is not negatively selected) and is thus detectable. While both models are based on the effects of an OR-mediated signal (prevention of novel OR transcription or cell death, respectively), the timeframe is different: in the negative feedback model, coexpressed genes are transcribed one after the other, while in the negative selection model everything happens at the same time. Both models suffer from weaknesses, but a major problem with the second one is the exclusion between parental alleles. How would the neuron make the difference between one or the other, knowing that transcription levels are variable between ORs and in time (Young et al., 2003)?

A third model involves the notion of a transcription quality step. An experiment using OSNs expressing the Cre recombinase under the control of an OR promoter allowed to show that a transient expression of ORs exists in maturating OSNs. When crossed to a reporter mouse line, some neurons which activated the reporter did not express the recombinase anymore, but a novel OR gene (Shykind et al., 2004). This mechanism, called “gene switch-ing”, was increased when a non-functional allele was the first chosen. Thus, a mechanism ensures that each mature OSN expresses a single functional OR, which is an important task given the putative transcription of pseudo-genes (the pseudogene repertoire accounts for around 20% of the whole OR repertoire) (Shykind et al., 2004). The switching model could be considered as part of the negative feedback hypothesis: sensory neurons can express

another OR gene only during a certain timeframe. This preserves a) the neuron from affecting axonal wiring in the OB, which is dependent on the expressed OR as we will see later, andb) the response profile of a given neu-ron (Shykindet al., 2004). However, it is still not clear which signal mediates both negative feedback and gene switching.

Interesting experimental data, involving the overexpression of a single OR in the MOE, have been recently reported (Nguyen et al., 2007). In this study, all attempts to express an OR under the direct control of theOMP or Gγ8 promoters failed. OMP being a late promoter, the explanation could lie in the negative effect of ORs expressed before activation of the transgene.

However, this explanation is not applicable to the Gγ8 promoter, which is active in immature neurons (Ryba and Tirindelli, 1995), where OR expression is barely detectable. Thus, in this latter case, the suppressed expression is not due to the effect of endogenous ORs. Expression of a transgenic OR was achieved only when the promoter and the coding sequence (CDS) were physically separated. The following observations were made using this tool:

a) expression of the transgenic OR prevents the transcription of endogenous ORs, andvice versa. b) Overexpression of a given OR is possible, if the choice of the transgenic OR is achieved early during development. c) Expression of a non-functional OR in which the DRY motif (ORs contain a conserved motif Asp-Arg-Tyr, DRY, which is mandatory for G-protein coupling) has been mutated to ALE still suppresses the transcription of endogenous ORs. This confirms a previous report (Imai et al., 2006). d) Multiple functional ORs can be expressed in a given OSN if they are encoded by the same transcript.

Taken together, these data, although very interesting, do not provide any model for OR gene regulation.

OR Transduction Pathway

ORs are GPCRs. The OR transduction cascade is therefore linked to G-proteins. G-proteins are composed by three subunit (α, β and γ). In ma-ture OSNs, ORs couple to a Gαs-family subunit, Gαolf: upon binding, the

G-protein activates adenylyl cyclase 3 (ACIII), which in turn reduces intra-cellular adenosine triphosphate (ATP) into cyclic adenosine monophosphate (cAMP). This second messenger then opens the multimeric cyclic nucleotide-gated channel CNGA2/CNGB1b, leading to a cation influx and successive opening of Ca2+-dependent channels. These channels promote the exit of Cl -and finally the depolarization of the OSN (Imai -and Sakano, 2008) (Figure 3.2 on page 17).

Odorant

ACIII

cAMP ATP

CNG Ca2+

Golf α βγ

α

Na+

Ca2+ Cl –

Cl – channel

Figure 3.2: Odorant receptor transduction cascade. Modified from Rodriguez, 2003.

Dans le document Vomeronasal receptors: from monogenic expression to axon guidance (Page 25-32)