Introduction

The transient receptor potential (TRP) genes constitute a growing superfamily of ion channels with a wide variety of important physiological functions [31]. One major subfamily within this group contains the TRPC genes (classical, canonical or short TRPs), which, in mammals, consist of seven structurally related members: TRPC1–TRPC7 (for recent reviews, see [11, 42]). There is currently an enhanced interest in the functional role of these genes, because evidence is accumulating that they play key roles in phospholipase C (PLC)-regulated Ca2+-signaling pathways.

This review will be concerned with a summary of recent progress that enabled us to define the specific function of one particular member of the TRPC genes, TRPC2 (also known as TRP2), which is specifically expressed in sensory neurons of the vomeronasal organ (VNO) in the mammalian accessory olfactory system. It is now clear that TRPC2 plays a fundamental role in the signal transduction machinery that is necessary for the sensing of pheromonal and other chemical social recognition signals in the rodent VNO. As such, this work has become an excellent example for the usefulness of TRPC-deficient mouse models in determining the biological role of TRPC channels, not only in native cells but also at the systems level in behaving animals.

Functional organization of the vomeronasal organ

In mice, odorants are recognized by two anatomically and functionally distinct sensory organs, the main olfactory epithelium (MOE) and the VNO [10, 30, 51]. The VNO is located at the base of the nasal septum, anterior and ventral to the MOE (Fig. 1a). Chemosensory stimuli gain access to the VNO through its single rostral opening which, in rodents, opens into the nasal cavity. The main structural elements of the VNO are best seen in coronal sections (Fig. 1b, c) and include a central lumen, a medially located crescent-shaped sensory epithelium, and a laterally located nonsensory epithelium together with an extensive network of blood sinuses surrounded by a band of cavernous erectile tissue (Fig. 1c, d). These latter components provide the hardware for a vomeronasal pump that aids in stimulus delivery [52].

Fig. 1
figure 1

Organization of the sense of smell in the mouse. a Midsaggital view of the rodent nasal cavity (NC) and forebrain. Sensory neurons in the main olfactory epithelium (MOE) project their axons to glomeruli in the main olfactory bulb (MOB). These neurons employ cyclic nucleotide-gated channels for sensory transduction. Vomeronasal sensory neurons (VSNs) in the vomeronasal organ (VNO), which is located at the base of the nasal septum (S), project to the anterior (red) or posterior (green) accessory olfactory bulb (AOB). b–d Coronal sectioning through the nasal cavity reveals the typical crescent-shaped organization of the VNO sensory epithelium (SE). The region delimited by the black box in c is shown at higher magnification in d. L Lumen, NSE nonsensory epithelium, V vein, SC sustentacular cell. Original sections in c and d are Nissl stains courtesy of X-H Li. Bars = 100 μm (c) and 50 μm (d). e The sensory epithelium is segregated into two distinct zones, both of which express a unique set of transduction-related molecules: (1) an apical zone (red VSNs) that expresses the G protein Gαi2 as well as members of the V1R family of vomeronasal receptors, and (2) a basal zone that characteristically contains VSNs (green) that express Gαo and members of the V2R receptor family. Both neuron types express TRPC2. f Distribution and axonal projection pattern of a subpopulation of gene-targeted VSNs, showing that axon bundles (vomeronasal nerves) terminate in the AOB (arrow). To visualize this topography, a targeted mutation was generated that led to coexpression of taulacZ in all VSNs expressing a distinct V1R receptor (from [37]). g Electronmicrograph of a transverse section through two mouse VSNs showing their dendritic process (D), which enlarges into a dendritic tip or knob (DT). From the knob emanate various microvilli (MV) (from [8]). h Visualization by confocal microscopy of the two principal expression zones of the VNO sensory epithelium using double-label immunohistochemistry with antibodies to phosphodiesterase type 4A (PDE4A, red) and Gαo (green). Scale bar = 100 μm (adapted from [18])

Full size image

The VNO sensory epithelium consists primarily of three cell types: (1) elongated glia-like supporting cells (sustentacular cells), (2) bipolar sensory neurons known as vomeronasal sensory neurons or VSNs, and (3) scarce basal stem cells (Fig. 1d). The VSNs, which are responsible for the detection of chemostimuli, extend a dendritic process to the luminal surface, where it enlarges into an apical dendritic knob (Fig. 1f). Unlike olfactory sensory neurons (OSNs), the dendritic knob of VSNs lacks cilia but contains up to 100 microvilli (Fig. 1f). The primary chemotransduction events are thought to take place in these microvilli. As in OSNs [50], cellular organelles that could serve as intracellular Ca2+ stores are spatially compartmentalized in VSNs, being present in the dendritic knob but not in the microvilli [8, 52] (Fig. 1f). The axonal process of each VSN emerges from the basal aspect of the soma and exits the epithelium through the basal lamina. VSN axons coalesce to form the vomeronasal nerves that course along the nasal septum, pass through the cribriform plate, and terminate in the glomeruli of the accessory olfactory bulb (AOB) where they synapse onto second order mitral cells. This axonal projection pattern is best observed in gene-targeted mouse strains in which subpopulations of VSNs coexpress specific markers such as taulacZ or tauGFP [37] (Fig. 1e).

An intriguing feature of the VNO is its segregation into two distinct zones, both of which express a unique set of transduction-related molecules ([10, 30, 52]; Fig. 1g, h): (1) an apical (superficial) zone that expresses the G protein Gαi2 as well as members of the V1R family of vomeronasal receptors (~150 genes), and (2) a basal (deep) zone that characteristically contains VSNs that express Gαo and members of the V2R receptor family (>150 genes). Apical VSNs function as highly sensitive, narrowly tuned detectors of small urine-derived volatiles [4, 17]. A chemosensory function of basal VSNs was proven only very recently when it was shown that they detect a large family of nonvolatile immune system molecules, major histocompatibility complex peptide ligands, which likely convey information about genetic individuality [18].

Identification of TRPC2

By the mid 1990s, a transduction model for OSNs in the main olfactory system was well-established, involving a cAMP-mediated second messenger cascade and activation of an olfactory cyclic nucleotide-gated (CNG) cation channel [49]. Therefore, early efforts aimed at understanding the signal transduction machinery in VSNs focused on cyclic nucleotides and CNG channels. However, it became rapidly clear that mammalian VSNs do not employ a cyclic nucleotide-activated channel for sensory signaling. For example, unlike OSNs [49], mammalian VSNs do not produce large membrane currents in response to cyclic nucleotides [2, 21]. As a direct consequence of these studies, the focus of interest then shifted toward the other major families of ion channels involved in sensory signaling, the TRP channels [22].

TRPC2 was first discovered in the laboratory of Craig Montell when the search of an expressed sequence tag data base for human orthologues of the Drosophila TRP channel led to the initial identification of the human TRPC2 gene [44]. This was followed by cloning of mouse [13, 41, 46, 48], bovine [45], and rat [22] orthologues. Whereas the primary structure of hTRPC2 contains stop codons, full-length transcripts were found in mouse and rat, leading to the suggestion that TRPC2 is a pseudogene in man but not in rodents. Phylogenetic analysis of the bovine orthologue suggested a closer relationship to mTRPC2 than to hTRPC2 [45]. Initial sequence analysis predicted an ion channel containing six membrane-spanning domains, a putative pore region between the fifth and sixth regions, and an intracellular N terminus with ankyrin repeats ([13, 22, 41]; Fig. 2a). However, a systematic analysis of the structure of TRPC2 has not yet been carried out (for a discussion of structure-function relationships of the TRPC subfamily, see [42]).

Fig. 2
figure 2

a Structure of the transient receptor potential canonical 2 (TRPC2) ion channel. The rat TRPC2 protein contains 885 amino acids, with an ankryn-repeat domain in the N terminus and a coiled-coil domain in the C terminus. The six-transmembrane (tm) domains are thought to fold like those of a K+ channel, with a proposed pore region between the fifth and sixth tm domains (from [25]). b Phylogenetic tree of mammalian TRPC channels. TRPC2 is the most divergent of the mammalian TRPC channels (from [25]). c Expression of TRPC2 in the VNO. Labeling of a section of VNO from an adult rat with a digoxigenic antisense probe directed against TRPC2 reveals strong expression (dark reaction product) of the TRPC2 mRNA in all vomeronasal sensory neurons (N) (from [22]). d In a singly dissociated VSN, TRPC2 immunoreactivity (red) is clearly seen in the tuft of microvilli at the distal end of the dendrite (adapted from [22])

Full size image

A major advance occurred with the cloning of TRPC2 from rat VNO by Liman et al. [22]. This work provided three particularly striking results. First, Northern analysis showed expression of rTRPC2 in the VNO but not in the main olfactory system (except for some sparse cells seen by in situ hybridization) or brain, suggesting a highly specific role for the gene product. Second, in situ hybridization revealed an expression pattern that was restricted to all VSNs (in both expression zones, Fig. 2c). Third, immunolabeling of TRPC2 demonstrated intense staining only in a highly limited region at the dendritic tip of the sensory neurons containing the sensory microvilli, the proposed site of sensory transduction (Fig. 2d). Based on these findings, TRPC2 was proposed to encode an ion channel participating in VNO sensory transduction ([22]; see also [13]). This was further supported by immunoelectron microscopic localization of TRPC2 [29].

Generation of mice deficient for TRPC2

To investigate the precise role of TRPC2 in vomeronasal function, TRPC2−/− mouse lines were generated independently by two laboratories [19, 39]. In both cases, the construct strategy was aimed at deleting functionally critical regions of the protein that include multiple transmembrane domains and the putative channel pore, although details of the constructs differed between the two laboratories [19, 39]. The electrophysiological responses to natural and synthetic stimuli and sexual and social behaviors were then examined in TRPC2 mutant mice.

TRPC2-deficient mice reveal an essential role for TRPC2 in VNO pheromone detection

To demonstrate a functional involvement of TRPC2 in VNO sensory transduction, both groups [19, 39] took advantage of recent developments allowing for simultaneous recording of neural activity from large populations of mouse VSNs, thus overcoming the problem of identifying a sufficient number of responsive cells. These recording methods, which had been developed only 2 years earlier [15, 17], were based, in one case, on the registration of local field potentials from the microvillous surface of an intact mouse VNO sensory epithelium [17], in combination with the use of a newly developed VNO coronal slice preparation, which allowed for optical recording of sensory responses in large numbers of individual VSNs by means of confocal Ca2+ imaging, as well as for patch clamp recording from single, visually identified VSNs [17]. In an alternative approach, the VNO sensory epithelium was peeled off the basement membrane, flattened, and mounted on an array of 61 extracellular electrodes [15]. Action potential activity was then recorded using the array of electrodes, or in some cases, by penetrating the neuroepithelium with a bundle of wire electrodes [15, 39].

At the neurophysiological level, the TRPC2-deficient mice clearly revealed an important role of TRPC2 in the generation of sensory responses induced by pheromonal cues in the VNO [19, 39]. In the absence of TRPC2, pheromone-induced VNO field potential responses were either absent or strongly diminished, depending on the stimulus concentration ([19]; Fig. 3a–d). Similarly, extracellular action potential recordings showed that individual VSNs from TRPC2−/− mice are electrically active but unable to respond to cues present in dilute urine [39]. Both studies concluded that TRPC2 is crucial for the generation of electrical responses in VSNs to sensory stimulation, i.e., functions as an essential component of the VSN signal transduction machinery. However, an important difference between these two studies was that one clearly showed the existence of residual sensory activity in TRPC2-deficient VNO ([19]; Fig. 3), whereas the other concluded that genetic ablation of TRPC2 eliminates the sensory response in VSNs [39].

Fig. 3
figure 3

a–d Generation of TRPC2-deficient mice and electrophysiological characterization of VNO responses. a, b In situ hybridization on VNO sections with antisense TRPC2 probe. Signal is detected in all wild-type VSNs in both zones (a), but is absent in TRPC2−/− tissue (b). c, d Stimulus-induced field potential responses to 2-heptanone and dilute urine at two concentrations each in wild type and TRPC2−/− VNO. Adapted from [19]

Full size image

In contrast to its function in the VNO, it is still unclear whether TRPC2 plays important roles in other cell types, especially in spermatocytes. Although low levels of TRPC2 transcripts were detected in testis [13, 45], immunohistochemistry failed to confirm this [39]. Other results suggested a specific role for TRPC2 in the sperm acrosome reaction [16]. Because TRPC2−/− mice are fertile and do not differ significantly from wild-type mice in the number of offspring [19, 39], the exact role of TRPC2 in sperm physiology requires further investigation.

TRPC2-deficient mice exhibit striking defects in social behaviors

TRPC2-deficient mice have afforded an excellent opportunity to define the role of the VNO in the generation of innate sexual and social behaviors in mice. This work has shown that deletion of only a single gene from the genome can lead to striking defects in complex social behaviors [10, 19, 25, 39, 53]. Two key results have emerged from these investigations thus far. First, TRPC2 is essential for pheromone-evoked male–male aggression. In a resident-intruder assay, which tests for intermale aggression, resident TRPC2−/− males fail to initiate attack behavior, although they are physically and neurologically capable of displaying aggressive interactions ([19, 39]; Fig. 4a, b). Interestingly, presumably as a result of the lack of an aggressive response, TRPC2−/− males fail to establish dominance hierarchies also and instead display urine marking behavior typical of subordinate males [19]. Female mice are usually not aggressive toward intruders unless they are lactating. However, we found that aggressive behavior is also severely attenuated in lactating female TRPC2−/− mice that are confronted with a male intruder, indicating that signals transduced by the VNO initiate aggressive behavior in both males and females ([19]; Fig. 4c).

Fig. 4
figure 4

Behavioral phenotyping of TRPC2-deficient mice. Lack of aggressive behavior in TRPC2−/− males (a, b) and lactating females (c). Adapted from [19]

Full size image

Second, a striking defect is also seen in the sexual behavior of TRPC2−/− males. Although TRPC2−/− males mate normally with females, they display increased sexual behavior toward other males, i.e., mounting other males at a much higher rate [19, 39]. This behavior had not been observed previously in animals in which the VNO was surgically ablated. This unexpected result has been interpreted as evidence that TRPC2-mediated signaling may be essential for gender discrimination [39], although other researchers have questioned this idea [33]. One possible model consistent with these data is that mounting is an innate behavior that is inhibited by male pheromones acting through the VNO. TRPC2−/− males, therefore, persist in mounting other males. Because of the absence of obvious defects in male–female sexual behavior in TRPC2−/− mice, pheromones, and other chemosignals essential for mating may not be detected by the VNO, but rather by other sensory systems such as the MOE.

Hence, it is now clear that TRPC2 is essential for the detection of male-specific cues in the VNO that, in turn, regulate the expression of complex behavioral repertoires including aggression, dominance hierarchies, and sexual interactions.

TRPC2 as a genetic marker for the evolution of VNO-dependent pheromone sensing

Because TRPC2 is expressed uniquely in the VNO and is essential for VNO function, the loss of a TRPC2 gene can potentially serve as a marker for the loss of VNO function. Based on this reasoning, Liman and Innan [24], Zhang and Webb [47], and Webb et al. [43] examined sequences of the TRPC2 gene from a large number of primate species (Fig. 5). The human TRPC2 gene has six mutations that generate premature stop codons, resulting in a severely truncated protein. The earliest mutation is a nonsense mutation that is shared by all old world (OW) monkeys and apes and that is predicted to generate a protein that is missing much of its C terminus [23, 25, 47]. As this mutation occurs in a well-conserved region of the protein, it is likely to impair functioning. Thus, based on the observation that this mutation is found in all OW monkeys and apes but not in new world (NW) monkeys, one can date the loss of a functional TRPC2 gene in the human lineage to 25–40 million years ago. This dating is further supported by an examination of selective pressure on the TRPC2 gene, which was also relaxed at this time [23].

Fig. 5
figure 5

The earliest mutation in the TRPC2 gene occurred in the common ancestor of old world (OW) monkeys and apes. a Sequence of TRPC2 from monkeys was examined for the presence of stop codons or frameshift mutations. Mutations are indicated by numbers 1–9, placed at the point in the phylogenetic tree where the mutation is inferred to have occurred. Note that mutation 9 was found in OW monkeys and gibbons but not other apes, indicating that it either arose twice or that there was a reversion event (indicated by a white 9 on a black background). b Schematic representation of TRPC2, indicating the position of each mutation. Black bars represent transmembrane domains. Adapted from [24, 25]

Full size image

A diacylglycerol-activated cation channel in VSNs

An important question that was until recently unresolved is whether TRPC2 is a part of functional ion channel in VSNs, and if so, how sensory stimulation leads to channel activation. Early findings using heterologous expression systems yielded somewhat conflicting results. Whereas one set of studies showed that TRPC2 is not detectable in the plasma membrane but rather accumulates in intracellular organelles when expressed in a variety of cell lines [13, 14], other studies concluded that TRPC2 encodes a store depletion-activated capacitative Ca2+ entry channel [41, 42]. The latter conclusions were based on measurements of the intracellular Ca2+ concentration, which are sometimes difficult to interpret because Ca2+ represents, in most cells, a complex downstream signal resulting from the interaction of multiple ion channel types.

To resolve this matter and to define the nature of the signal transduction mechanism in VSNs, we have focused our attention on an experimental strategy leading to the identification and characterization of a native ion channel dependent on TRPC2 [27]. This experimental design was based on four assumptions:

  1. 1.

    The native channels are likely to be found at the dendritic tip of VSNs.

  2. 2.

    They are likely to be regulated, directly or indirectly, by products of PLC activity.

  3. 3.

    They are likely to be defective in TRPC2−/− VSNs.

  4. 4.

    They are expected to exhibit functional properties consistent with those of the pheromone-induced conductance.

By using a combination of inside-out and whole-cell patch clamp recordings in VSNs from wild-type and mutant mice, we were able to identify an ion channel that meets all of these criteria [27]. We first performed inside-out patch recordings using plasma membrane patches that were excised from the dendritic tips of freshly dissociated VSNs (Fig. 6a). To test whether a conductance exists in these patches that is controlled by products of PLC activity, we explored a potential role of diacylglycerol (DAG), one of the two second messengers generated by PLC. Addition of the endogenous DAG analogue 1-stearoyl-2-arachidonoyl-sn-glycerol (SAG) to the cytosolic face of these patches produced an increase in channel activity, giving rise to a sustained inward current (Fig. 6b). This effect was accompanied by an increased noise level, reflecting fluctuations in the activity of single channels. The presence of ATP, GTP, Ca2+, or Mg2+ in the internal solution was not required for activation of this current, and the current could not be activated by Ca2+ alone, i.e., without added SAG. We also examined the effect of the second product of PLC activity, Ins(1,4,5)P3. However, in no case did we observe noticeable channel activity in response to Ins(1,4,5)P3, whereas the DAG-activated channels were found in almost every patch [27].

Fig. 6
figure 6

Identification of a diacylglycerol (DAG)-activated cation channel in inside-out membrane patches from the dendritic tip of VNO neurons. a Micrograph showing an isolated mouse VSN. MV microvilli, D dendrite, S soma. Plasma membrane patches were taken from the encircled area denoted as dendritic tip (DT). Scale bar = 3 μm. b Addition of the endogenous DAG analogue 1-stearoyl-2-arachidonoyl-sn-glycerol (SAG, 100 μM) to the bathing medium produced an increase in channel activity that gave rise to a sustained inward current. This effect was accompanied by an increased noise level that reflects fluctuations in the activity of single channels (holding potential, −80 mV). c, d Current–voltage relationship of SAG-induced conductance in response to a voltage-ramp obtained with Na+ (c) or Ca2+ (d) as the sole intracellular cation. Extracellular pipet solution contained Na+ as the sole cation. e, f Examples of single-channel events before (e) and after (f) application of SAG (10 μM). The SAG-activated channel has a unitary conductance of 42 pS in divalent cation-free solutions with Na+ as the sole cation and shows none of the hallmarks of capacitative Ca2+ entry channels (holding potential, −80 mV). Adapted from [27]

Full size image

The DAG-activated channel from VSN dendrites shows a striking resemblance in some of its biophysical properties to the DAG-gated cation channels formed by hTRPC3 or hTRPC6 in heterologous cells [12]. Analysis of current–voltage curves using different ionic conditions revealed an almost linear, nonselective cation conductance in VSNs that is permeable for Na+, Cs+, and Ca2+, but not for N-methyl-D-glucamine ([27]; Fig. 6c, d). Our work provided also a first analysis of DAG-activated single-channel currents in VSNs, revealing a slope conductance of 42 pS in symmetrical 150 mM Na+ solution (Fig. 6e, f). These experiments clearly demonstrated the existence of a previously unknown cation channel in VSN dendrites that can be activated by DAG in a membrane-delimited fashion. Importantly, this channel shows none of the hallmarks [20, 34] of store depletion-activated capacitative Ca2+ entry channels.

TRPC2 is required for a functional DAG-gated channel in VSN dendrites

We hypothesized that the TRPC2 gene is required for normal function of the newly identified DAG-gated channel [27]. This was tested by comparing the properties of DAG-activated whole-cell currents in wild-type and TRPC2−/− VSNs (Fig. 7a–f). This analysis revealed a striking defect in the activation of the DAG-gated channel. The DAG-activated whole-cell current was strongly diminished at all concentrations tested by as much as 90% ([27]; Fig. 7f). Hence, the DAG-activated channel clearly depends on an intact TRPC2 gene. Thus far, no other defective ion channel has been identified in TRPC2−/− VSNs. We concluded, therefore, that TRPC2 is a major component, if not the principal subunit, of the DAG-gated cation channel.

Fig. 7
figure 7

a–f TRPC2−/− VSNs are severely defective in the activation of a DAG-gated conductance. a Family of whole-cell currents to a series of depolarizing and hyperpolarizing voltage steps (as indicated in the figure) recorded from an isolated wild-type VSN. Experiments were performed in the presence of 1 μM tetrodotoxin to block voltage-gated Na+ channels; voltage-activated K+ channels were blocked by using a Cs+-based pipet solution. Dotted line Zero current level. b A prominent DAG-activated conductance was observed in these cells following application of SAG (100 μM). c, d In VSNs from TRPC2−/− mice, SAG application failed to activate a large conductance. However, a drastically diminished residual response to SAG still exists in these cells. e Steady-state current–voltage relationships of the SAG-induced responses shown in b and d, respectively; WT mice (black rectangles), TRPC2−/− (red rectangles). f Dose dependence of averaged SAG-induced currents (at −70 mV) from multiple WT (black bars) and TRPC2−/− VSNs (red bars). Adapted from [27]

Full size image

An interesting finding was that the deletion of TRPC2 did not fully abolish activation of the DAG-gated channel in VSNs but left a small but significant residual conductance intact ([27]; Fig. 7d, f). The presence of a residual conductance in TRPC2−/− VSNs suggests that other channel subunits activated by DAG may exist in these cells, although with strongly reduced efficacy. It remains to be seen whether these predicted channel subunits associate with TRPC2 to form a heteromultimeric channel complex or whether they form independent channels. A search for additional members of the TRP channel family in VSNs has consistently failed [27]. It is interesting to note, in this respect, that TRPC2 does not seem to interact with any known TRPC protein [14], although other researchers, using primary erythroblasts isolated from mouse spleen, provided evidence for an interaction of TRPC2 with TRPC6 [7].

VSNs use the phosphoinositide pathway for pheromone transduction

Having identified DAG as a ligand for the TRPC2 channel, we asked whether endogenously produced DAG is responsible for activation of the pheromone-induced conductance and whether the DAG-activated channel can account for the properties of this conductance. The use of pharmacological inhibitors of enzymes participating in phospholipid and phosphatidylinositide second-messenger cascades revealed that the sensory current depends critically on PLC but not on DAG lipase, phosphatidylinositol 3-kinase, or phospholipase A2 ([27]; Fig. 8a–c). These results supported the argument against arachidonic acid and other polyunsaturated fatty acids [38] as endogenous activators of the sensory current. At the same time, they established PLC as a key enzyme in the activation mechanism of the sensory current, consistent with the notion that an increase in the concentration of DAG initiates this current. We also explored a potential involvement of Ins(1,4,5)P3 [5], the second product of PLC activity, in the sensory current but found no evidence that generation of Ins(1,4,5)P3 represents a primary step for initiation of the pheromone conductance [27]. However, our results do not rule out a role for Ca2+-activated channels [23], possibly stimulated via Ins(1,4,5)P3-dependent Ca2+ release, as secondary amplifiers of the primary response in some VSNs [27].

Fig. 8
figure 8

Evidence that VSNs use the phosphoinositide pathway for pheromone transduction. a Pharmacological dissection of key enzymes participating in phospholipid and phosphatidylinositide second messenger cascades. b VSN sensory currents induced by brief application of dilute urine (DU) are inhibited by the phospholipase C (PLC) antagonist U73122 (10 μM) but not by an antagonist of DAG lipase, RHC-80267 (50 μM). c Summary of the effects on urine-evoked currents of enzyme inhibitors of phospholipid and phosphatidylinositide signaling cascades. Each response was normalized to its own control response obtained before application of each drug as shown in (a). Adapted from [27]

Full size image

Thus, a direct link between pheromone-induced PLC activity in VSNs, gating of the TRPC2 channel by DAG, and the pheromone-sensitive current has been established.

A DAG kinase regulates activity of the TRPC2 channel in VSNs

Independent evidence that endogenously produced DAG may be important in activation of the TRPC2-dependent channel and thus, in the pheromone response, comes from pharmacological blockade of DAG kinase ([DGK] [27]; Fig. 9). DGK phosphorylates DAG to produce phosphatidic acid and is, therefore, a potential terminator of DAG signaling ([28, 40]; Fig. 9a). In VSNs, we found that application of an antagonist of Ca2+-activated DGK isotypes, R59949, produces sustained activation of the TRPC2-dependent channel (Fig. 9b–d). A likely explanation for this effect is that, even in the absence of sensory stimulation, PLC exhibits a small but significant basal activity, thus continuously producing DAG. Hence, even partial blockade of DGK can lead to DAG build-up which, in turn, results in channel activation. In support of this idea, we have shown that the R59949-evoked current is abolished upon inhibition of PLC (Fig. 9b). These findings resemble the constitutively active TRP channels in the Drosophila DGK mutant rdgA [35]. They are also reminiscent of the effect of phosphodiesterase inhibitors on the activity of CNG channels in OSNs [49].

Fig. 9
figure 9

DAG kinase (DGK) regulates DAG concentration in VSNs. a Dissection of DGK signaling in VSNs. b Application of the type I DGK inhibitor R59949 (20 μM) produces a sustained inward current in voltage-clamped VSNs, an effect that is reversed by the PLC inhibitor U-73122 (10 μM). However, U-73122 has no effect when the current is directly activated by SAG [27], indicating that the R59949-evoked conductance results from DAG built-up due to basal activity of PLC occurring in the absence of receptor stimulation. c, d The R59949-evoked conductance is indistinguishable in its properties from that induced by SAG or pheromones. The conductance has a reversal potential close to 0 mV and is inhibited by 2-APB (50 μM). Using the dose-response relationship of SAG-evoked currents in WT mice VSNs as a calibration (Fig. 7f), we estimate that VSNs produce less than 10 μM DAG during a pheromone response. Adapted from [27]

Full size image

Pharmacology of the TRPC2 channel

The pharmacology of TRPC channels is somewhat unsettled. An interesting but unexpected result was that 2-aminoethoxydiphenyl borate (2-APB) functions as a blocker of the DAG-gated channel in VSNs ([27]; Fig. 9). 2-APB is sometimes considered as an inhibitor of Ins(1,4,5)P3 signaling and store-operated Ca2+ entry [3]. Although 2-APB clearly inhibited the DAG-gated channel, it did not affect Ins(1,4,5)P3-activated currents, which are present in a subset of mouse VSNs [27]. Hence, the action of 2-APB seems to differ significantly in different cell types, and an effect of 2-APB on a cellular response cannot be deemed to be definitive proof for the involvement of an Ins(1,4,5)P3-dependent mechanism.

A model for VNO pheromone transduction

Identifying the signal transduction mechanism in pheromone-sensitive VSNs represents an important step in understanding how the accessory olfactory system encodes social and reproductive information that is essential for reproductive fitness. Based on the work summarized here, we have proposed a model in which the TRPC2 subunit functions as an essential part of a primary conductance pathway in a pheromone-stimulated second messenger cascade of mouse VSNs ([27]; Fig. 10a). Our results strongly support the notion that pheromones and other chemosignals stimulate specific seven transmembrane-domain receptors [9], leading to the activation of distinct G-protein isoforms. In turn, this stimulates phospholipase C, which generates DAG and Ins(1,4,5)P3. Our results [27] reinforce the argument that it is the generation of DAG (or its endogenous analogues) that initiates—by gating of the TRPC2 cation channel—the sensory current that underlies the rapid depolarizing receptor potential of VSNs. Termination of DAG signaling occurs, at least in part, through the activity of a DAG kinase ([27]; Fig. 10b). This model differs significantly from the mechanism underlying sensory transduction in the vertebrate main olfactory system, which employs cation channels directly gated by cyclic nucleotides [51]. Hence, distinct molecular cascades have evolved in the mammalian main and accessory olfactory systems for the detection of chemosensory signals.

Fig. 10
figure 10

a, b Schematic representation of pathways that might regulate the activity of the TRPC2 channel in the dendritic tip of a mammalian VSN. (See text for rationale.) The model shown in a refers to VSNs of the apical zone of the VNO. Whether this applies also to VSNs of the basal zone remains to be investigated.

Full size image

It must be noted, however, that this transduction model is far from being complete and that additional refinement will be required. Numerous important questions remain to be answered: What is the role of Ins(1,4,5)P3? How does Ca2+ entry through the TRPC2 channel contribute to sensory signaling? Which receptors are activated by which ligands, and how do they couple to the second messenger cascade? Does phosphatidylinositol-4,5-bisphosphate play any role in channel activation? Which subunits form the DAG-gated channel? What is the molecular identity of other components of this signaling cascade including G-proteins, PLC, and DAG kinase? Can the phenotype in TRPC2-deficient mice be rescued by TRPC2 gene delivery? Does TRPC2 form a functional DAG-activated channel in a heterologous cell type? What is the molecular gating mechanism induced by DAG in this new family of ion channels?

Conclusions

The past few years have delivered substantial progress in understanding the functional role of TRPC2 in the mammalian vomeronasal (accessory olfactory) system. On the basis of this work, we can now assign a clear function to TRPC2 at both the cellular and systems levels, showing that TRPC2 occupies a critical role in the detection of pheromonal signals in the VNO and thus is essential for social recognition of conspecifics. The experimental strategy employed in the mouse VNO now serves as a powerful model for examining the native functions of other TRP genes. This work also serves as a prime example for the importance of investigating the function of TRP channels within their native cellular environment and with respect to their biological roles at the whole-animal level.

Our work has shown that TRPC2 is essential for a DAG-gated cation channel in VSNs. In fact, this DAG-gated channel represents the first native DAG-gated channel in the mammalian nervous system. Hence, our findings may have important implications beyond the sense of smell in that they provide a general mechanism by which PLC signaling can produce neuronal excitation and Ca2+ entry, independently of protein kinase C, Ins(1,4,5)P3, and Ca2+ stores. TRPC2 does not seem to be expressed in other neurons outside the VNO, unlike the closely related subunits TRPC3, TRPC6, and TRPC7 [36]. Because all three proteins can function as DAG-gated channels [12, 26, 32], it is reasonable to propose that channel activation by DAG may serve as a general signal transduction mechanism in the brain, as well as in other nonneuronal cell types, e.g., portal vein myocytes [1]. Only very recently has it been recognized that DAG not only activates its classical target, protein kinase C, but is in fact involved in a multitude of interconnected signaling processes [6]. Ca2+-permeable, DAG-activated cation channels of the TRPC family can now be added to the “nonclassical, non-PKC” targets of DAG.