Type 3 inositol 1,4,5-trisphosphate receptor is dispensable for sensory activation of the mammalian vomeronasal organ

Signal transduction in sensory neurons of the mammalian vomeronasal organ (VNO) involves the opening of the canonical transient receptor potential channel Trpc2, a Ca2+-permeable cation channel that is activated by diacylglycerol and inhibited by Ca2+-calmodulin. There has been a long-standing debate about the extent to which the second messenger inositol 1,4,5-trisphosphate (InsP3) and type 3 InsP3 receptor (InsP3R3) are involved in the opening of Trpc2 channels and in sensory activation of the VNO. To address this question, we investigated VNO function of mice carrying a knockout mutation in the Itpr3 locus causing a loss of InsP3R3. We established a new method to monitor Ca2+ in the endoplasmic reticulum of vomeronasal sensory neurons (VSNs) by employing the GFP-aequorin protein sensor erGAP2. We also performed simultaneous InsP3 photorelease and Ca2+ monitoring experiments, and analysed Ca2+ dynamics, sensory currents, and action potential or field potential responses in InsP3R3-deficient VSNs. Disruption of Itpr3 abolished or minimized the Ca2+ transients evoked by photoactivated InsP3, but there was virtually no effect on sensory activation of VSNs. Therefore, InsP3R3 is dispensable for primary chemoelectrical transduction in mouse VNO. We conclude that InsP3R3 is not required for gating of Trpc2 in VSNs.

The mammalian olfactory system has evolved two major signaling systems for chemoelectrical transduction: one that depends on cyclic nucleotide-gated (CNG) channel activation and involving cAMP or cGMP signaling, respectively, and another that depends on transient receptor potential (TRP) channel activation, mainly involving the Trpc2 cation channel 1, 2 . Trpc2 is a central transduction element in sensory neurons of the mouse vomeronasal organ (VNO) [3][4][5] which play important roles in the detection of socially-relevant molecular cues such as pheromones and kairomones [6][7][8] . The recent finding that Trpc2 is also expressed in sensory neurons of the main olfactory epithelium (MOE) 9,10 where it is required in type B cells for the detection of low environmental oxygen 11 has sparked renewed interest in its function. Despite two decades of research, the second messenger signaling mechanisms underlying activation of Trpc2 and its corresponding sensory responses are still debated and there is presently no single agreed-upon mechanism for its activation 7,[12][13][14][15] . Moreover, the behavioral phenotypes of Trpc2 mutant mice are more complex than previously thought 16 . A complete understanding of Trpc2 signaling mechanisms will be required to fully appreciate the role of Trpc2 in chemical communication and social behaviors.
To resolve these problems and to further define the nature of the signal transduction mechanism in VSNs, we investigated VNO function in a gene-targeted mouse strain that carries a knockout mutation in the Itpr3 locus and thus lacks InsP 3 R3 40,41 . We used a recently developed method to monitor Ca 2+ in the endoplasmic reticulum (ER) by employing a novel, genetically-encoded Ca 2+ sensor protein specifically targeted to the ER (erGAP2) 42,43 , and we performed simultaneous InsP 3 photorelease and Ca 2+ monitoring experiments. By applying a combination of state-of-the-art Ca 2+ imaging and electrophysiological methods using wild-type and InsP 3 R3-deficient VSNs, we asked the following: (1) Are intracellular Ca 2+ stores significantly mobilized during VSN sensory activation? (2) Does InsP 3 induce Ca 2+ signals in VSNs? (3) Is InsP 3 R3 necessary for stimulus-evoked Ca 2+ signaling, the generation of sensory currents, and action potential or field potential responses in VSNs? With this approach, we provide compelling evidence that sensory activation of the mouse VNO is largely independent of InsP 3 R3, ruling out a crucial role for InsP 3 signaling in the primary chemotransduction process of mouse VSNs.  2+ ] ER in terms of Ca 2+ affinity, organelle selectivity and/or dynamic range. To overcome these difficulties, we used a novel low-Ca 2+ affinity sensor, erGAP2, that can be targeted specifically to the ER lumen 42 . erGAP2 is a fluorescent, genetically-encoded Ca 2+ indicator based on the fusion of two jellyfish proteins, GFP and aequorin. erGAP2 has been optimized for measurements in high Ca 2+ concentration environments (K d for Ca 2+ = 407 µM) due to two substitutions on the EF hands of the Ca 2+ -binding domain (Fig. 1a). Targeting to the ER is achieved through a combination of the calreticulin signal peptide and the ER retention sequence KDEL flanking the sensor (Fig. 1a). This reporter enables ratiometric imaging using two excitation peaks at 405 and 470 nm and single emission at 510 nm (Fig. 1b) 43 .

Monitoring [Ca
We expressed erGAP2 in freshly dissociated VSNs from C57BL/6 (wild-type, WT) mice using a herpes simplex virus type 1 (HSV-1) amplicon vector 44 (Fig. 1b). A 24-h incubation period was sufficient to observe robust erGAP2 expression in infected VSNs with intact cell viability (Fig. 1c). We performed ratiometric [Ca 2+ ] ER imaging on infected VSNs and analysed stimulus-induced fluorescence signals (Fig. 1d). We compared response patterns to several previously established VSN chemostimuli: a mix of sulfated steroids (E1100, E0893, E0588, and E1050, each at 100 µM), and the high molecular weight fraction of mouse urine (HMW; 1:300) which contains major urinary proteins (MUPs). These stimuli have been reported to activate VSNs of both the apical (sulfated steroids) and basal (HMW) layers of the VNO neuroepithelium [45][46][47][48][49] and their responses are known to depend on Trpc2 45,46 . We also applied caffeine (Caf, 50 mM), a potent activator of the ubiquitous ryanodine receptors 50 , to induce Ca 2+ release from the ER. We measured F 470 /F 405 ratios in 368 cells that expressed detectable erGAP2 fluorescence. All cells showed a decrease of the F 470 /F 405 ratio in response to a 60-s caffeine stimulus, indicative of a sharp decrease of [Ca 2+ ] ER . In contrast, stimulation with the mix of sulfated steroids (E mix) or HMW did not induce major changes in F 470 /F 405 ratio. Imaging of a representative cell and its F 470 /F 405 ratio trace as a function of time is displayed in Fig. 1d and e, respectively. On average, both E mix and HMW induced no or only very minor reductions in F 470 /F 405 ratio (−0.0117 ± 0.0033 and −0.0155 ± 0.0035, respectively) whereas caffeine produced a robust, 15-fold larger decrease in F 470 /F 405 ratio (−0.175 ± 0.0071; Mann-Whitney test, n = 368, P = 1.6 × 10 −85 and 5.6 × 10 −83 ).

Simultaneous [Ca 2+ ] ER and [Ca 2+ ] C imaging in VSNs.
To determine whether VSNs activated by HMW or steroid ligands show Ca 2+ release from the ER, we performed simultaneous measurements of [Ca 2+ ] ER and [Ca 2+ ] C using combined imaging of erGAP2 and the cytosolic Ca 2+ dye fura-2. For this approach, we imaged erGAP2-expressing cells loaded with fura-2 and performed triple illumination with 340 nm, 380 nm and 470 nm excitation light and single channel emission at 510 nm (Fig. 1f). Changes in [Ca 2+ ] C were estimated by calculating the 340/380 nm ratio (R 340/380 ), and [Ca 2+ ] ER was estimated by using single F 470 dynamics because of the partial F 405 overlap with fura-2. We imaged a total of 599 erGAP2-expressing cells and applied 60-s stimuli of either caffeine, HMW, or the sulfated steroid estradiol-3,17-disulfate (E1050; 100 µM). We observed robust [Ca 2+ ] C increases, measured as a rise in R 340/380 , in 15 cells responding to HMW and in 8 cells responding to E1050. These cells also showed [Ca 2+ ] C increases in response to caffeine that were accompanied by simultaneous reductions in F 470 , indicative of massive Ca 2+ release from the ER (Fig. 1f).

Removal of extracellular Ca 2+ abolishes ligand-induced VSN responses.
To assess whether extracellular Ca 2+ is necessary to generate elevations of [Ca 2+ ] C following stimulation with HMW and E1050, we performed fura-2 imaging in freshly dissociated, non-infected VSNs 45,47,51 . We first established that multiple, 60-s stimulations with HMW or E1050 that were separated by 4-min interstimulus intervals are sufficient to produce robust and repeatable increases in [Ca 2+ ] C ( Fig. 2a and b). Next, we used a protocol in which a Ca 2+ -free extracellular medium containing 0.5 mM EGTA was applied during the second stimulation with either HMW or E1050. Ca 2+ responses were completely absent during the second stimulus application in Ca 2+ -free medium for all cells that responded to any of the two stimuli during the first application (HMW, n = 3; E1050, n = 8; Fig. 2a and b). When extracellular Ca 2+ was re-introduced, [Ca 2+ ] C increases recovered during a third application ( Fig. 2a and b). To verify that Ca 2+ stores were not depleted under these conditions, we applied caffeine in Ca 2+ -free medium and observed robust Ca 2+ transients (Fig. 2e). These responses were very similar to caffeine responses obtained in Ca 2+ -containing medium after a 4-min recovery period ( Fig. 2e and f; Wilcoxon signed-rank test, P = 0.166) indicating that internal Ca 2+ stores were not depleted. Therefore, extracellular Ca 2+ and Ca 2+ entry is required to generate cytosolic Ca 2+ transients in response to activation with HMW and E1050, consistent with previous results 47, 52 . Depletion of intracellular Ca 2+ stores does not affect ligand-induced Ca 2+ responses. To further explore the impact of intracellular Ca 2+ stores on stimulus-evoked VSN responses, we applied cyclopiazonic acid (CPA), a potent inhibitor of sarcoplasmic-endoplasmic reticulum Ca 2+ -ATPases (SERCAs), in order to deplete intracellular Ca 2+ stores. VSNs were incubated with 30 µM CPA for 20 min in an extracellular medium containing Ca 2+ . We then performed fura-2 imaging during stimulation with HMW or E1050 in the presence of CPA. We observed robust increases of [Ca 2+ ] C in 4/140 cells stimulated with HMW and in 7/250 cells stimulated with E1050 ( Fig. 2c). Unlike other SERCA inhibitors such as thapsigargin, the effect of CPA can be reversed after washout. We used 60-s caffeine pulses (50 mM) to determine that a 5-min washing period after CPA incubation was sufficient to restore [Ca 2+ ] ER and enable robust Ca 2+ release (Fig. 2d). Caffeine-induced Ca 2+ responses increased drastically after CPA washout (Wilcoxon signed-rank test, P < 0.001; Fig. 2d). We compared the cell-to-cell peak amplitudes of HMW and E1050 during CPA treatment (0.2788 ± 0.0579, HMW; 0.5153 ± 0.0745, E1050) vs. washout (0.3018 ± 0.0766, HMW; 0.4058 ± 0.0611, E1050). Although E1050 responses tended to be slightly larger in the presence of CPA, we found no significant differences between the two conditions (Wilcoxon signed-rank test, P = 0.58, HMW; P = 0.07, E1050). To determine the effect of CPA application on intracellular Ca 2+ , we recorded the first 5 min after CPA incubation and observed a slow and transient Ca 2+ increase (n = 14; Fig. 2g), consistent with sustained, store-dependent Ca 2+ release. Together, these results indicate that depletion of intracellular Ca 2+ stores does not prevent the generation of cytosolic Ca 2+ transients following activation with HMW and E1050. Thus, intracellular Ca 2+ stores are not critical for VSN cytosolic Ca 2+ transients to these sensory stimuli and extracellular Ca 2+ must be the main source of cytosolic Ca 2+ for these Ca 2+ responses.

InsP 3 R3 is expressed in mouse VSNs. InsP 3 R3 immunoreactivity has been reported in rat VSNs 28 and
transcripts of all three InsP 3 R isoforms were identified in whole mouse VNO tissue 53 . To obtain independent evidence for the presence and function of InsP 3 Rs in mouse VSNs, we investigated the expression of InsP 3 R3 in VNO cryosections of adult WT mice using specific anti-InsP 3 R3 antibodies. We observed InsP 3 R3 immunoreactivity in the entire vomeronasal neuroepithelium, especially in VSN somata, dendrites and possibly in dendritic endings, whereas immunoreactivity was weaker in supporting cell somata and non-sensory parts of the VNO (Fig. 3a). Importantly, this labeling was absent in InsP 3 R3 −/− VNO sections, confirming antibody specificity (Fig. 3a). Next, we performed RT-PCR on cDNA libraries prepared from single-cell RNA. We dissociated VNOs from OMP-GFP mice 54 to obtain single, isolated and fluorescent VSNs that were individually collected using a microcapillary pipette. In OMP-GFP mice, GFP serves as a marker for mature (olfactory marker protein-expressing) VSNs. We picked 10 GFP+ cells and 2 GFP-cells. We prepared RNA from each cell, generated single-cell cDNA libraries 10,44 , and assessed for gene expression by PCR with gene-specific primers. We amplified Omp PCR products in 8/10 GFP+ cells (Fig. 3b), indicative of a 80% success rate for this method. We further screened for Itpr3 gene expression in these 8 GFP+ cells and obtained Itpr3 PCR products in 4 cell samples (Fig. 3c), demonstrating that InsP 3 R3 is indeed expressed in at least a fraction of VSNs. Full-length gels are presented in Supplementary Fig. S1. We found no detectable expression of Itpr1 (InsP 3 R1) or Itpr2 (InsP 3 R2) genes in all samples, except for one of the GFP− control cells that was positive for Itpr2 (not shown). For comparison, we also sampled 18 cells from InsP 3 R3 −/− VNO (which are GFP−), 8 of which were positive for the Omp PCR (Fig. 3b). We did not amplify Itpr3 PCR products from any of these cells (Fig. 3c). We found Itpr2 expression in 5 of the cells that were negative for Omp (not shown). These results indicate that InsP 3 R3 is the predominant isoform expressed in VSNs, whereas InsP 3 R1 and InsP 3 R2 are not co-expressed in these cells. InsP 3 R2 is likely expressed by supporting and other non-sensory cell types. There was no evidence for upregulation of Itpr1 or Itpr2 in InsP 3 R3 −/− VSNs.

Photolysis of caged InsP 3 shows that InsP 3 R3 is required for Ca 2+ release. To determine whether
InsP 3 is capable to evoke Ca 2+ elevations in VSNs and, if so, whether this signal would require InsP 3 R3, we performed flash photolysis experiments with caged InsP 3 in InsP 3 R3 +/− and InsP 3 R3 −/− VSNs (Fig. 3d-h). Freshly dissociated VSNs were co-loaded with a photoactivatable and membrane-permeant caged InsP 3 propionyloxymethyl ester 55 as well as the Ca 2+ indicator fluo-4 (see Methods for details). We used confocal laser-scanning microscopy to monitor changes in cytosolic Ca 2+ and to locally deliver ultraviolet light (UV, 355 nm) stimulation which, in turn, caused photoliberation of caged InsP 3 and thereby released the active InsP 3 molecule. Cells exhibiting Ca 2+ transients in response to either E1050 or HMW were identified as functional VSNs and the regions circumscribed by each cell were outlined and stimulated with UV light (Fig. 3d). A typical UV stimulation for a single cell with a diameter of 10 µm consisted of 10 individual scans with a pixel dwell time of 1.54 µs, resulting in a total UV exposure time of 1.76 ms. With these conditions, InsP 3 R3 +/− VSNs showed striking Ca 2+ transients in response to UV stimulation ( Fig. 3d and e) demonstrating that InsP 3 is indeed capable to induce increases of cytoplasmic Ca 2+ in these cells. Consistent with previous studies using InsP 3 photorelease in other cell types 56 , uncaging efficiency in InsP 3 R3 +/− VSNs was between 30% and 40%: 14/38 (37%) of E1050-activated cells and 15/48 (31%) of HMW-activated cells showed Ca 2+ transients after UV exposure ( Fig. 3f and g). By contrast, InsP 3 R3 −/− VSNs, which showed normal response rates for HMW and E1050 (Fig. 3h), were largely insensitive to InsP 3 photorelease (Fig. 3d,e,g and i), indicating that InsP 3 R3 is required for InsP 3 -induced Ca 2+ release in these VSNs. For example, from a total of 32 cells responding to E1050 and 43 additional cells responding to HMW, we observed only a single cell in each group that could be activated by UV light (Fig. 3f and g). Furthermore, when we analyzed only those cells (n = 16 InsP 3 R3 +/− VSNs and n = 13 InsP 3 R3 −/− VSNs, respectively) that were stimulated successively with a chemostimulus (HMW or E1050), UV light, and a second chemostimulus (Fig. 3i), UV-evoked responses were completely absent in InsP 3 R3 −/− versus InsP 3 R3 +/− VSNs (Mann-Whitney, P < 0.001). Together, these results show that photolysis of InsP 3 can induce transient Ca 2+ elevations in VSNs heterozygous for the InsP 3 R3 mutation but that these responses are absent or strongly reduced in InsP 3 R3-deficient VSNs. Therefore, InsP 3 R3 is the predominant isoform mediating InsP 3 -evoked Ca 2+ release in these sensory neurons.  (Fig. 4a). Average values of these responses (−0.193) were not significantly different from WT VSNs (Mann-Whitney test, P = 0.103; Fig. 4b). Thus, ER Ca 2+ loading and caffeine-induced Ca 2+ release seem to be intact in these InsP 3 R3 −/− cells. E mix and HMW induced modest changes in F 470 /F 405 ratio (−0.0126 and −0.015 a.u., respectively), not significantly different from WT VSNs (Kruskal-Wallis test, P = 0.48; Fig. 4b).
Although not significant, [Ca 2+ ] C amplitudes tended to be somewhat higher in InsP 3 R3 −/− VSNs. Interestingly, this tendency is reversed in erGAP2-infected cells after 24 h (Fig. 4e). The reasons for this are not known but argue further against a major contribution of InsP 3 R3-mediated Ca 2+ release to the cytosolic signal under these conditions. We also note that the number of responding cells cannot be compared directly in freshly dissociated versus erGAP2-infected cells. Together, these results indicate that InsP 3 R3 and Ca 2+ release from the intracellular stores are not critically required for ligand-evoked cytosolic Ca 2+ transients and sensory activation of mouse VSNs.

InsP 3 R3 is not required for activation of VSN sensory currents. To strengthen our results obtained
with Ca 2+ imaging, we carried out whole-cell patch-clamp recordings in voltage-clamped VSNs 5, 57 . These experiments used acute VNO tissue slices 52,58,59 in which the cellular VSN architecture is preserved, enabling recordings from optically identified VSNs located in apical or basal layers of the sensory epithelium. VSN sensory currents depend, in full or in part, on the activation of Trpc2 cation channels [3][4][5] . Trpc2 can form a protein-protein interaction complex with InsP 3 R3 and it has been proposed that this interaction contributes to the electrical response of VSNs to chemostimulation 28,29 . If so, activation of VSN sensory currents should be severely disrupted in InsP 3 R3 −/− mice. To assess this, we recorded sensory currents in InsP 3 R3 +/− vs. InsP 3 R3 −/− VSNs (Fig. 5). Cells were initially stimulated with a mixture of diluted urine from male and female mice (DU, 1:100), a rich source of natural pheromones. This stimulus induced small but reliable inward currents in 19/51 (37%) of InsP 3 R3 +/− VSNs and in 20/48 (42%) of InsP 3 R3 −/− VSNs, consistent with previous reports of WT VSNs 33,35,38 . The properties of urine-evoked sensory currents were analysed in more detail using 5-s urine applications, a holding potential of −70 mV, and a KCl-based intracellular solution (Fig. 5a-d). Under these conditions, we found no significant difference in peak amplitude, time-to-peak, or decay time constants between the two genotypes ( Fig. 5a-d and Table 1). Ca 2+ -activated K + channels can contribute to urine-evoked currents in VSNs 38 . To determine whether large K + currents would mask the contribution of InsP 3 R3 at resting membrane potential, we recorded urine-evoked currents at a holding potential of −100 mV, close to the K + reversal potential (Fig. 5e-h). This resulted in considerably larger peak amplitudes of urine-evoked currents, but again there was no significant difference between the two genotypes in peak amplitude, time-to-peak, or decay time constant (Fig. 5e-h and Table 1). Furthermore, urine-evoked sensory currents (holding potential, −100 mV) were fully preserved when the tissue was pretreated for 5 min with the SERCA inhibitor thapsigargin (1-5 µM) to deplete intracellular Ca 2+ stores irreversibly, and there was no significant difference between InsP 3 R3 +/− vs. InsP 3 R3 −/− VSNs in their peak amplitudes (Fig. 5i and Table 1). These results indicate that Ca 2+ release from intracellular stores is not essential for urine-evoked VSN sensory currents.
We also analysed sensory currents evoked by the sulfated steroid E1050 (Fig. 6 and Table 1). This stimulus routinely evoked large inward currents with peak amplitudes of several hundred picoamperes, even at a concentration of only 10 nM (Fig. 6a and b). However, we found no significant difference in peak amplitude, time-to-peak and decay time constant between the two genotypes ( Fig. 6b-d and Table 1). Furthermore, current-voltage curves obtained at the peak of E1050-evoked responses (CsCl-based intracellular solution) where very similar between the two genotypes and showed a reversal potential close to 0 mV (InsP 3 R3 +/− : V rev = 4.3 ± 2 mV; InsP 3 R3 −/− : 1.4 ± 1.2 mV, n = 3 each) (Fig. 6e). Therefore, InsP 3 R3 is not essential for the activation of VSN sensory currents. supported by extracellular loose-patch recordings from VSNs in tissue slices measuring stimulus-evoked action potential sequences (Fig. 7a-f). These represent VSN output signals that are ultimately transmitted to the olfactory forebrain. Using 1-s pulses of diluted urine as stimulus, action potential responses could be readily evoked and repeated multiple times in both InsP 3 R3 +/− and InsP 3 R3 −/− VSNs ( Fig. 7a and b). Post-stimulus time histograms (PSTHs) obtained from group data of such recordings revealed no significant difference between the two genotypes (P = 0.07-0.97) (Fig. 7c and d). The same basic result was also observed for action potential sequences in response to E1050 (10 nM, P = 0.06-1) (Fig. 7e and f).

InsP 3 R3 is not required for Ca 2+ -calmodulin-dependent VNO adaptation. The N-terminus of
Trpc2 binds CaM in a Ca 2+ -dependent manner 60 . We showed previously that Ca 2+ -CaM feedback mediates sensory adaptation and inhibits DAG-activated Trpc2 currents in VSNs 32 . For other TRPC channels such as Trpc3, an activation model has been proposed in which an InsP 3 R binds to a site that partially overlaps with the Ca 2+ -CaM site, thereby displacing CaM and thus causing channel activation 61 . We tested whether activation and Ca 2+ -CaM-dependent VSN adaptation is altered in the absence of InsP 3 R3 by performing extracellular field potential recordings from the surface of the sensory epithelium using an intact VNO wholemount preparation 32, 52, 58, 62, 63 (Fig. 7g-j). We recorded isobutylamine-evoked potentials (0.1 µM), which depend on type 1 vomeronasal receptors 62 , and potentials to SYFPEITHI (1 nM), a major histocompatibility complex (MHC) binding peptide 58 . SYFPEITHI-evoked responses require type 2 vomeronasal receptors and the G protein Gαo 47, 59 but persist in Trpc2-deficient mice 64 . We found that both ligands evoked robust, phasic-tonic field potentials in InsP 3 R3 +/− and InsP 3 R3 −/− VNO that underwent time-dependent desensitization with prolonged stimulation (Fig. 7g). We analysed field potential peak amplitudes, the ratio between plateau and peak as a measure of the extent of adaptation, and the adaptation onset time constant 32 (Fig. 7h-j). Despite a slight trend for lower plateau-peak ratios with isobutylamine, there was no significant difference between the two genotypes for both stimuli in any of these parameters (unpaired t-test: P = 0.09-0.93). These results indicate that InsP 3 R3 is neither required for the activation of ligand-evoked field potentials in the VNO nor for their Ca 2+ -CaM-dependent desensitization.

Discussion
The role of the second messenger InsP 3 in vertebrate olfaction has been discussed controversially for almost 30 years, but no gene deletion studies have been performed to address this critical problem in any of the olfactory subsystems for any vertebrate species. To overcome this limitation, we undertook a series of investigations employing InsP 3 R3-deficient mice. We focused on sensory neurons of the VNO because these have been shown previously to express InsP 3 R3 but not InsP 3 R1 or InsP 3 R2 28 . In agreement with these results, our knockout-controlled immunolabeling and single-cell RT-PCR experiments confirmed the presence of InsP 3 R3 RNA and protein in mouse VSNs but found no evidence for other InsP 3 Rs in these cells. We used a confocal, laser-scanning-controlled approach to photorelease InsP 3 and simultaneously monitor cytosolic Ca 2+ in isolated VSNs. These experiments demonstrated for the first time that InsP 3 evokes transient, intracellular Ca 2+ rises in VSNs and that InsP 3 R3 is functionally required for this effect, providing a robust foundation for investigating the role of InsP 3 R3 in sensory activation of the mouse VNO. We applied a wide range of vomeronasal chemostimuli and used multiple recording techniques and VNO preparations to address a potential role of InsP 3 R3. Our approach included Ca 2+ imaging as well as voltage-clamp, loose-patch, and field-potential recordings in isolated VSNs, acute VNO tissue slices, and a VNO wholemount preparation. All of these experiments led to the same basic result, namely that InsP 3 R3 is dispensable for sensory activation of the VNO and, therefore, does not contribute crucially to the primary, chemoelectrical transduction process of VSNs. We also demonstrated that InsP 3 R3 is not required for Ca 2+ -CaM-dependent VSN adaptation. As a whole, these experiments call for a   revision of current schemes suggesting that InsP 3 R3 has a critical function in VSN sensory activation. We note that our experiments do not rule out a potential role for InsP 3 signaling or InsP 3 R3 function in secondary or modulatory signaling events of mammalian VSNs, nor do they rule out distinct VSN signaling mechanisms in lower vertebrates 65 . We also cannot exclude that primary signaling of noncanonical VSNs, such as those expressing formyl peptide receptor 66 or odorant receptor genes 67 , would require InsP 3 R3. Future experiments will be needed to address these questions. Our results have important implications for the gating mechanism of Trpc2 cation channels and, hence, for the TRP channel field in general. Most importantly, our results rule out any gating model in which a physical link of InsP 3 R3 with Trpc2, either directly or indirectly, mediates the opening of the Trpc2 channel in VSNs. Consequently, these results also rule out any model in which the displacement of inhibitory Ca 2+ -CaM by InsP 3 R3 causes Trpc2 activation. The present results are fully consistent with and strengthen further our previous findings that Trpc2 is a DAG-activated, Ca 2+ -permeable cation channel that can be inhibited by Ca 2+ -CaM and does neither require InsP 3 nor Ca 2+ release from intracellular stores for its activation in VSNs 5,7,18,32 . These experiments leave the activation of Trpc2 by DAG, possibly in conjunction with PIP 2 , as the most plausible VSN primary transduction mechanism. It remains to be seen whether this model applies also to other Trpc2-expressing cells outside the VNO where cells likely exhibit different cellular architecture or express other Trpc2 splice variants and interacting proteins. It has been suggested that modes of Trpc2 activation are cell-specific and require different interactions and activators in different cell types 14 . Our results should also be of general interest to the activation models of other DAG-sensitive TRP channels such as TRPC3, TRPC6, and TRPC7 68,69 .
There is considerable evidence that Ca 2+ -activated chloride channels in VSNs, for which TMEM16A/anoc-tamin1 is an essential component 37 , serve to amplify the sensory response and cause further VSN depolarization because of elevated intracellular Cl − concentrations 33,35,70,71 . In dorsal root ganglia, activation of anoctamin1 by localized Ca 2+ signals requires coupling with the type 1 InsP 3 receptor 72 . Similarly, it has been proposed that activation of VSN chloride channels is triggered by Ca 2+ release from intracellular stores 35 . Our electrophysiological recordings in InsP 3 R3-deficient VSNs found no evidence for a significant reduction of the size of sensory responses. This argues strongly that chloride channel activation in VSN sensory responses is not mediated by InsP 3 R3-dependent Ca 2+ release.
Several interesting results emerge from our study with respect to the role of sulfated steroids as VSN ligands. Initially, these molecules were tested at relatively high concentrations, at 100 or 200 µM, using extracellular spike recordings 46 or Ca 2+ imaging 48 , but the primary sensory currents evoked by these ligands have not been analyzed previously. Using patch-clamp recordings under voltage-clamp, we found that VSNs in VNO slices generate surprisingly large currents at only 10 nM of E1050 (Fig. 6) and also evoke action potentials at this concentration (Fig. 7). These low thresholds are supported by a previous study identifying V1rj receptors that selectively respond to sulfated steroids (including E1050) at 10 nM 73 . Our response rates for E1050 are also consistent with a report that found approximately 0.2-5.5% of the VSNs to be responsive to individual sulfated steroid components including E1050 74 . We cannot yet determine whether the differences in amplitude or response kinetics reflect specific differences in the underlying mechanisms of the sensory currents but, importantly, there was no obvious effect of the InsP 3 R3 deletion on responses to E1050, neither at low nor at high stimulus conditions. Our experiments demonstrate for the first time the suitability of a novel, genetically-encoded Ca 2+ sensor protein, erGAP2, for monitoring Ca 2+ dynamics in the ER of mammalian VSNs. erGAP2 is a ratiometric low-affinity Ca 2+ sensor of the GFP-aequorin protein family that has been optimized for measurements in high Ca 2+ concentration environments and that can be targeted to intracellular organelles 42,43 . These experiments, together with pharmacological manipulations of store-dependent Ca 2+ mobilization, provided clear evidence that Ca 2+ release from the ER does not play a critical role in the primary transduction mechanism of mouse VSNs, results that are fully consistent with previous conclusions 5 .
In summary, the present study provides important new insights into the primary mechanisms underlying chemoelectrical signal transduction of the mammalian VNO and are necessary for advancing our understanding of Trpc2 activation. Our data show conclusively that InsP 3 R3 is not required for these functions. These results advance substantially our understanding of the molecular mechanisms mediating sensory activation of the mammalian VNO.

Materials and Methods
Mice. All animal protocols complied with the ethical guidelines for the care and use of laboratory animals issued by the German Government and were approved by the Animal Welfare Committee of Saarland University School of Medicine. All methods were carried out in accordance with the relevant guidelines and regulations. Mice were housed in ventilated cages under a 12:12 hour light/dark cycle with food and water available ad libitum. Generation of mice that carry a knockout mutation in the Itpr3 locus and thus lack InsP 3 R3 (InsP 3 R3 −/− ) has been described 40,41 . This strain was intercrossed with C57BL/6 mice for at least twelve times before use 41 . C57BL/6 mice (denoted as wild-type, WT) or InsP 3 R3 +/− littermates were used as reference mice. Some experiments were performed on OMP-GFP mice (heterozygous for the mutation) in which all cells expressing olfactory marker protein (OMP) are genetically labeled and show robust GFP fluorescence 54 . All experiments were performed on adult, 6-18 weeks old mice (both sexes).
Live-cell Ca 2+ imaging. Ca 2+ imaging of freshly dissociated VSNs was performed as described 45,47,51 . VNO epithelium was detached from the cartilage and minced in PBS at 4 °C. The tissue was incubated (20 min at 37 °C) in PBS supplemented with papain (0.22 U/ml) and DNase I (10 U/ml; Fermentas), gently extruded in DMEM (Invitrogen) supplemented with 10% FBS, and centrifuged at 100 × g (5 min). Dissociated cells were plated on coverslips previously coated with concanavalin-A type V (0.5 mg/ml, overnight at 4 °C; Sigma). Cells were used immediately for fura-2 imaging after loading them with fura-2/AM (5 µM; Invitrogen) for 60 min, or Photorelease of InsP 3 . VNO cells were dissociated and plated on coverslips as described above and loaded with 3 μM caged InsP3/PM [D-2,3-O-isopropylidene-6-O-(2-nitro-4,5-dimethoxy)benzyl-myo-inositol 1,4,5-trisphosphate-hexakis (propionoxymethyl)ester; Enzo Life Sciences, Switzerland] mixed with the same volume of Pluronic F127 in DMSO (10%) in Hepes-HBSS buffer. Cells were loaded with caged InsP3/PM for 30 min at room temperature in the dark followed by an additional 30 min incubation of 2.5 μM fluo-4/AM (Invitrogen) and InsP3/PM. Stock solutions were made in DMSO and kept for up to 1 week stored at −20 °C. The final DMSO concentration did not exceed 0.5%. Coverslips containing VSNs were placed in a laminar-flow chamber (Luigs and Neumann) and constantly perfused with extracellular Hepes-buffered solution. We used an upright scanning confocal microscope (Zeiss LSM 880 Indimo) equipped with a standard Argon laser for excitation at wavelength of 488 nm (fluo-4 excitation) and a UV laser (Coherent) emitting 355 nm (InsP3 uncaging). Images were acquired at 0.5 Hz. Emitted fluorescence was collected between 500 and 560 nm. All scanning head settings were kept constant during each experiment. The UV laser light was coupled to the confocal microscope and focused onto the image plane through a 20 × 1.0 NA Plan-Apochromat water immersion objective (Zeiss). The depth of focus was 16 µm which ensured, together with the region of interest (ROI) diameter, illumination of individual cells. Before starting photolysis of caged InsP 3 , UV laser light was optimally focused using 18 µm thick brain tissue sections loaded with Hoechst 33342 (1:10000; ThermoFisher) and the semi-automated correction tool of the Zen software (Zeiss). Photolysis of caged InsP 3 was achieved by directing UV laser light (1.036 mW) on preselected ROIs (spot diameter ~10 µm) using the Zen software (Zeiss) before reverting back to the visible wavelength laser to resume monitoring of fluo-4 fluorescence. Photorelease of caged InsP 3 with ROI spot illumination was performed on single cells previously identified to respond to HMW or E1050. In some experiments, we used UV whole-field illumination (ROI area, 425 × 425 µm) to photorelease InsP 3 in a larger area containing multiple fluo-4 loaded cells, in order to record also from IP 3 R3-deficient cells that did not respond to HMW or E1050 but were potentially sensitive to InsP 3 (such as non-VSN cell types present in the VNO and expressing IP 3 R1 or IP 3 R2, serving as positive controls). The estimated intracellular InsP 3 concentration after photolysis was expected to be in the 0.1-5 μM range 56 . Ca 2+ changes were generally expressed as relative fluorescence changes, i.e. ΔF/F 0 (F 0 was the average of the fluorescence values of 5-10 frames before stimulation). Images were acquired at 0.5 Hz and analysed using ImageJ (NIH). Peak Ca 2+ signals evoked by photoreleased InsP 3 were calculated from the temporal profiles of ROI values during the first 10-20 s after a UV stimulus.
Virus production. Virus production, packaging of herpes simplex virus type 1 (HSV-1) vectors and VSN infection was performed as described 44 . This virus-based amplicon delivery system was initially developed to overexpress vomeronasal receptors of the V1r, V2r, and Fpr families in VSNs 44 . To produce viral vectors, a HindIII/EcoRI fragment containing the erGAP2 cDNA was cloned into the herpes simplex virus plasmid pHSVpUC. Packaging and titration of virus particles were performed as reported earlier 75 . VSN cultures were infected with a multiplicity of infection (moi) ranging between 0.01 and 0.1 one day before use.
Immunohistochemistry. Mice were deeply anesthetized with CO 2 prior to decapitation and VNOs were removed, fixed for 3 h in 4% paraformaldehyde, equilibrated overnight in PBS containing 30% sucrose, embedded in OCT (Tissue-Tek), and snap-frozen in a dry ice/2-methylbutane bath. Frozen tissue sections (16 µm) were collected on glass slides (Superfrost Plus, Polysciences) and stored at −80 °C until use. Sections were post-fixed 15 min in 4% paraformaldehyde, washed 3 times in PBS (10 min each), incubated in blocking solution (0.5%