Phosphorylation via PKC Regulates the Function of the Drosophila Odorant Co-Receptor

Insect odorant receptors (ORs) have a unique design of heterodimers formed by an olfactory receptor protein and the ion channel Orco. Heterologously expressed insect ORs are activated via an ionotropic and a metabotropic pathway that leads to cAMP production and activates the Orco channel. The contribution of metabotropic signaling to the insect odor response remains to be elucidated. Disruption of the Gq protein signaling cascade reduces the odor response (Kain et al., 2008). We investigated this phenomenon in HEK293 cells expressing Drosophila Orco and found that phospholipase C (PLC) inhibition reduced the sensitivity of Orco to cAMP. A similar effect was seen upon inhibition of protein kinase C (PKC), whereas PKC stimulation activated Orco even in the absence of cAMP. Mutation of the five PKC phosphorylation sites in Orco almost completely eliminated sensitivity to cAMP. To test the impact of PKC activity in vivo we combined single sensillum electrophysiological recordings with microinjection of agents affecting PLC and PKC function and observed an altered response of olfactory sensory neurons (OSNs) to odorant stimulation. Injection of the PLC inhibitor U73122 or the PKC inhibitor Gö6976 into sensilla reduced the OSN response to odor pulses. Conversely, injection of the PKC activators OAG, a diacylglycerol analog, or phorbol myristate acetate (PMA) enhanced the odor response. We conclude that metabotropic pathways affecting the phosphorylation state of Orco regulate OR function and thereby shape the OSN odor response.


INTRODUCTION
Olfaction in nematodes and vertebrates utilizes G protein-coupled receptor (GPCR) signaling. In insects, odorant receptor (OR) proteins share the seven-transmembrane topology of GPCRs but retain no sequence-similarity (Benton et al., 2006). Furthermore, they form heterodimers of a ligand-binding OR and an ubiquitous co-receptor such as Dmel/Orco (previously Or83b in Drosophila; Vosshall and Hansson, 2011), both of which are inversely oriented in the membrane compared to GPCRs (Benton et al., 2006;Lundin et al., 2007). On the other hand, G proteins are expressed in the dendrites of olfactory sensory neurons (OSNs) bearing the ORs (Boto et al., 2010), and Drosophila mutants with disturbed G protein signaling cascades show impaired odor processing (reviewed in Hansson et al., 2010). Here we concentrate on OR22a as ligandbinding receptor. This is probably the most well-investigated OR of D. melanogaster. It is expressed in large basiconic sensilla and is tuned to fruit volatiles (ethyl butyrate, ethyl hexanoate) emitted by, e.g., pineapple (Stensmyr et al., 2003;Hallem and Carlson, 2006).
Studies on insect OR function in heterologous expression systems provided evidence for a G protein-independent, ionotropic mode of action (Sato et al., 2008;Wicher et al., 2008). However, an additional metabotropic pathway has been shown to stimulate cAMP production, which in turn activates the ion channel-forming co-receptor protein Orco (Wicher et al., 2008). Manipulating the cAMP level changed the kinetics of the odor response in heterologous cells (Smart et al., 2008). Genetic manipulation of G protein signaling in flies produced comparatively mild effects on odor responses (Yao and Carlson, 2010). However, the important role of stimulatory G proteins and subsequent cAMP signaling was recently demonstrated both in flies and in the heterologous expression system (Deng et al., 2011). A consensus model of these controversial results suggests a modulation of the ionotropic response by metabotropic pathways (Nakagawa and Vosshall, 2009).
In addition to cAMP signaling, DAG/IP 3 signaling may also play a role in insect olfaction (Krieger and Breer, 1999). Mutations in the Drosophila dgq gene encoding the G q α subunit produces flies with reduced responses to odor stimulation (Kain et al., 2008). The responses were further attenuated by additional mutations in plc21C, a gene encoding for a PLCβ. In the present study, we investigate the effect of G q protein downstream signaling on heterologously expressed Orco proteins and demonstrate the relevance of these results using single sensillum electrophysiology in Drosophila OSNs combined with microinjection of compounds affecting the G q protein signaling cascade.

Stable transfection
The open reading frame of Orco was PCR-amplified using gene specific primers with restriction sites for XhoI and HindIII and cloned into the pcrII TA-cloning vector (Invitrogen, Carlsbad, CA, USA). The identity of the insert was sequenced in full length to verify identity and integrity and subcloned into the pcDNA3.1(+) expression vector via the integrated restriction sites. The resultant construct was verified by sequencing. Flp-In ™ -T-Rex ™ 293 cells held in DMEM (high glucose, with l-Glutamine)/Ham's F12 (with l-Glutamine; PAA, Pasching, Austria) +10% FCS were transfected with OR 83b pc DNA 3.1 using Fugene (Invitrogen) according to the manufacterer's protocol. 24-h post-transfection 800 mg/l of G418 were added to the medium. After 12 weeks, clones were separated and tested for activity using measurements of channel activity as described (Wicher et al., 2008). The clone exhibiting the highest sensitivity to 8-bromo-cAMP was maintained under antibiotic selection and used for this study.

PATCH-CLAMP ELECTROPHYSIOLOGY
Ion currents in HEK293 cells were measured at room temperature using whole-cell patch-clamp with appropriate compensation of series resistance and of capacitive currents. Additional experiments were performed in the inside-out configuration. Current measurements and data acquisition were performed using an EPC9 patch-clamp amplifier controlled by PatchMaster software (both HEKA Elektronik, Lambrecht, Germany). Patch-clamp pipettes were fabricated from borosilicate capillaries. Pipettes for whole-cell recordings had resistances of 2-4 MΩ for excised-patch recordings the pipette resistance was up to 15 MΩ.
With the exception of GTP-γ-S, and GDP-β-S, which were applied via the patch pipette, all substances were applied to the bath using either a bath perfusion system (BPS4 from ALA, NY, USA) or a rapid solution changer (RSC160 from Biologic, Claix, France) which were controlled by the PatchMaster software (HEKA Elektronik).
For data analysis the software IgorPro (WaveMetrics, Lake Oswego, OR, USA) or Prism 4 (Graph Pad Software, San Diego, CA, USA) were used.

EXTRACELLULAR SINGLE SENSILLUM RECORDING AND MICROINJECTION
Recording and injection protocols were performed on Or22a-GAL4; UAS-CD8-GFP Drosophila melanogaster flies expressing membrane tagged GFP in Or22a-OSNs. Two-to 5-day-old adults were fixed dorsally to a microscope slide. Compounds and concentrations for injection were diluted in receptor lymph solution (Kaissling and Thorson, 1980) as follows: U73122 (0.5 mM), Gö6976 (0.5 mM), OAG (0.1 mM), PMA (0.1 mM). Note that due to a dilution effect, concentrations of injected agents were 100× the concentration used in whole-cell preparations. A microinjection setup consisting of a dual-pump system was used to inject agents via air pressure through the microelectrode holder and into the sensillum lymph. For odor stimulation, 10 μl of ethyl butyrate (99%, Sigma, Munich, Germany) in hexane (10 ng/μl; 99%, Fluka Analytical, Buchs, Switzerland) was pipetted onto 1 cm diameter filter paper disks and placed in disposable Pasteur pipettes. Odor stimuli were delivered at 0.5 l/min into a 1.0 l/min humidified air stream.
Sensilla were localized at 1000× magnification and an Ag/AgCl coated silver wire inserted into a sharpened glass capillary used to detect the extracellular analog signals originating from the OSNs. Action potentials were extracted digitally according to top-top amplitudes using Syntech Auto Spike 32 software. Cell activities were recorded for approximately 20 s before an initial 0.5 s stimulation with ethyl butyrate. Microinjection commenced at 100 s, and cells were again stimulated with an 0.5-s odor pulse after approximately 300 s. Responses of the larger amplitude Or22a-carrying cell were analyzed for 1500 ms after stimulus onset. For response kinetics, spike frequency ratios were analyzed as peri-stimulus time histograms (PSTHs) in 25 ms bins by dividing each 25 ms instantaneous spike frequency by the average pre-stimulus frequency over 2 s to give a normalized ratio for each time point. Areas under the PSTH curve were calculated for the stimulus (500 ms) Frontiers in Cellular Neuroscience www.frontiersin.org and total response (1350 ms) windows respectively, adjusting for a 150 ms mechanical stimulus delay. These values were divided by time to establish a normalized frequency average for each response. Mann-Whitney U tests compared treatments with the control (receptor lymph ringer) after injection. All analyses were performed using PASW (SPSS) v. 18 software.

RESULTS
Mutant flies with disrupted G q protein/phospholipase C (PLC) signaling cascade show reduced odor responses (Kain et al., 2008). When seeking the molecular mechanism by which PLC inhibition affects the odor response, the most parsimonious assumption is that PLC targets the OR complex itself. As multiple receptors are affected by G q protein disruption (Kain et al., 2008), the ubiquitous Orco would be a good target candidate. In human embryonic kidney (HEK293) cells co-expressing Or22a and Orco, we have previously observed that intracellular application of the non-hydrolysable G protein inhibitor GDP-β-S reduces the sensitivity of the receptor dimers to ethyl butyrate, a key ligand (Wicher et al., 2008). G protein inhibition prevented odor-induced cAMP production and consequent activation of Orco via the metabotropic pathway. Here, we asked whether inhibition of G proteins could affect the sensitivity of Orco to cAMP. Using the voltage-clamp technique in the whole-cell configuration, we perfused HEK293 cells expressing Orco with the non-hydrolysable GDP analog GDP-β-S via a patch pipette. Stimulation of Orco by bath application of the membrane-permeable cAMP analog 8-bromo-cAMP could -even at the highest concentrationsinduce only a weak membrane current ( Figure 1B). Without GDP-β-S in the pipette, 8-bromo-cAMP induced a current in a concentration-dependent manner ( Figure 1B). Unexpectedly, permanent stimulation of G proteins with the non-hydrolysable GTP analog GTP-γ-S induced a current even in the absence of 8-bromo-cAMP (Figures 1A,B). Subsequent application of 8bromo-cAMP further enhanced this current. However, there was less current production by 8-bromo-cAMP than under control conditions (287 pA vs. 550 pA; Figure 1B), indicating that the pool of channels available for activation by 8-bromo-cAMP is reduced due to pre-activation by GTP-γ-S. To assess whether G proteins have a direct effect on Orco and whether there is any subtype-specificity, we co-expressed Orco together with the β 2 -adrenergic receptor (β 2 -AR), which activates G s proteins, and with the α 1 -adrenergic receptor (α 1 -AR), which activates G q proteins. Stimulation of β 2 -AR with 10 μM isoproterenol induced currents of comparable size to those obtained after application of 1 μM 8-bromo-cAMP (relative current: 1.2 ± 0.2; n = 5). Preincubation of cells with the adenylyl cyclase inhibitor Whole-cell current response in a non-transfected HEK293 cell on a voltage ramp from −100 to +100 mV after breaking into the cell (Control) and 2 min after application of 100 μM 8-br-cAMP (top), and a HEK293 cell expressing Orco after breaking into the cell (Control), after perfusion with GTP-γ-S (500 μM) to activate G proteins, and 2 min after application of 100 μM 8-br-cAMP (bottom). Note that 8-br-cAMP fails to produce a current in non-transfected cells. (B) Concentration-response curve for 8-br-cAMP-induced Orco currents, measured at −100 mV in the whole-cell mode with standard pipette solution (Control) and a solution containing GTP-γ-S (500 μM) to activate or GDP-β-S (500 μM) to inactivate G proteins, respectively. The current "before" is the difference between GTP-γ-S and Con described in (A). (Control, n = 11; GTP-γ-S, n = 8, **P to Control < 0.01; GDP-β-S, n = 9, ***P < 0.001). (C) Test for the role of G s protein activation on Orco. Current responses in a cell expressing Orco and the β 2 adrenergic receptor (β 2 AR) to isoproterenol stimulation (10 μM) in the absence (ISO) and 6 min after application of SQ22536 (200 μM). Curves represent differences (2 min ISO -before ISO). (D) Test for the role of G q protein activation on Orco. Currents evoked in a cell expressing Orco and the α 1 adrenergic receptor (α 1 AR) by norepinephrine (NE) stimulation (1 μM) in the absence and 8 min after application of the PLC inhibitor U73122 (10 μM). Curves represent differences (2 min NE -before NE). SQ22536 (200 μM) prevented current production by isoproterenol (relative current: −0.1 ± 0.1; n = 5; Figure 1C). Thus, neither the G s α subunit nor the βγ subunit complex was able to activate Orco in the absence of cAMP.
Stimulation of α 1 -AR with 1 μM norepinephrine induced currents of 1.6 ± 0.3 (n = 5), normalized to the 1 μM 8-bromo-cAMP response. To test for a direct G protein effect, PLC was inhibited with U73122 (10 μM). Under these conditions, norepinephrine failed to elicit a current (relative current: −0.2 ± 0.2; n = 7; Figure 1D), indicating that neither the G q α subunit nor the βγ subunit complex could activate Orco alone. The activation of coexpressed Orco upon stimulation of α 1 -AR is remarkable as it for the first time demonstrates that Orco can be metabotropically activated independent of cyclic nucleotides. There are thus at least two independent signaling pathways capable of producing Orco currents.
To examine whether inhibition of PLC activity could account for the reduced cAMP-sensitivity of Orco with GDP-β-S, we tested the effect of 8-bromo-cAMP in the presence of U73122. We then observed only marginal responses, even at the highest concentration of 8-bromo-cAMP (Figures 2A,C). U73343, an Frontiers in Cellular Neuroscience www.frontiersin.org inactive analog of U73122, did not suppress the 8-bromo-cAMP effect (P = 0.91; n = 8). The G protein activity-dependence of the Orco response to cAMP is therefore related to either PLC activity or a downstream process, i.e., the response of Orco to cAMP requires some basal PLC activity. To maintain catalytic activity of PLC, physiologically free Ca 2+ levels are necessary (Rebecchi and Pentyala, 2000). We thus tried to stimulate Orco using a Ca 2+ -free pipette solution, and we recorded only marginal responses to 8-bromo-cAMP (5 μM; 31 ± 17 pA; n = 9) or forskolin (10 μM; 95 ± 41 pA; n = 8) compared with those obtained with standard pipette solution (cAMP: 398 ± 67 pA; n = 11; forskolin: 697 ± 67 pA; n = 9). Inhibition of PLC activity prevents PIP 2 cleavage and subsequent IP 3 and DAG production. We asked which of these effects could account for the observed depression of current production by cAMP. Mimicking PIP 2 accumulation due to PLC inhibition using bath application of PIP 2 with inside-out patches did not significantly reduce the cAMP effect (5 μM cAMP enhanced the background current of 3.6 ± 0.7 pA by 10 ± 2.7 pA; n = 8). Thus, the PIP 2 cleavage products may instead be critical for the cAMPsensitivity of Orco. Even though IP 3 activated a tiny current (∼0.5 pA at 5 μM), this current was too weak to rescue the
The concentration-response curve for whole-cell current activation by 8-bromo-cAMP indicates that Orco is highly sensitive to cAMP at physiological Ca 2+ levels (Figures 2B,C). To demonstrate this directly we tested the effect of cAMP on inside-out patches. Intriguingly, current activation even after fast application of cAMP develops slowly and with a delay (Figure 3). The activation process thus differs from classical gating such as in CNG channels. The concentration-response curve obtained was characterized by EC 50 = 0.7 nM cAMP and a Hill coefficient = 0.40 ( Figure 2D). Therefore, Orco is at least four orders of magnitude more sensitive to cAMP than the most cAMP-sensitive CNG channel (Dhallan et al., 1990). As in the whole-cell experiments, GDP-β-S, U73122, and Gö6976 drastically reduced the responses to cAMP (Figure 2D).
These results suggest that PKC activity plays a central role in controlling Orco function, and especially the sensitivity to cAMP. The Orco protein bears five PKC phosphorylation sites (Figure 4A), three in the intracellular loop 2 (IC2), T250, S289 and T327, and two with putative extracellular localization (S159 in EC2; T371 in EC3). To confirm the role of PKCmediated phosphorylation, we produced an Orco PKC mutant with serine/threonine to asparagine replacements at all PKC sites ( Figure 4A). This mutant was expected to mimic the Orco wild type in the presence of PLC or PKC inhibitors for both background activity and stimulation by cAMP. Odorant receptor heterodimers such as Or22a/Orco show some background activity even in the absence of odor stimuli (Wicher et al., 2008). Inside-out patches from HEK293 cells expressing Orco conducted a significantly higher resting current than those from non-transfected cells ( Figure 4B), indicating that Orco is spontaneously active even in the absence of cAMP. The resting current was reduced by GDP-β-S, U73122, and Gö6976 ( Figure 4B). For the PKC phosphorylation mutant Orco PKC, the resting current was similar to nontransfected cells and Orco-expressing cells in the presence of these inhibitors, and was significantly lower than for Orco-expressing cells under control conditions ( Figure 4B). Compared to native Orco stimulation, activation of the mutant with 8-bromo-cAMP in the whole-cell configuration or cAMP in inside-out patches produced very weak responses (Figures 4C, 5 and 6). Similarly, PMA failed to elicit a current in the absence of 8-bromo-cAMP ( Figure 5C).
To confirm that impaired membrane targeting of Orco PKC did not account for the small current production by cAMP, we tested the distribution of Orco and Orco PKC immunofluorescence in the HEK293 cells ( Figure 4D). Comparison of immunostaining with plasma membrane staining indicated that both Orco and Orco PKC were localized within the membrane (Figures 4D,E).
We also designed two partial mutants (Figure 4A), mutant 1 (S159N, T250N, S289N) and mutant 2 (T327N, T371N) which both showed no resting activity. We next asked how their response to 8-bromo-cAMP compared to that of Orco and Orco PKC (Figure 4C). The weakest response was seen in the complete mutant Orco PKC; the response obtained with mutant 2 was significantly stronger than that of mutant 1 and Orco PKC itself ( Figure 4C). This indicates that the various

Frontiers in Cellular Neuroscience
www.frontiersin.org FIGURE 6 | cAMP responses in Orco and Orco PKC. All-point histograms for the data shown in Figure 5B. Currents from inside-out patches from a cell expressing Orco (top) and Orco PKC (bottom) after breaking into the cell (Control) and 2 min after application of cAMP at 500 pM and 100 μM, respectively. Patches were held at −60 mV, currents were recorded for 5 s (20 repetitions). phosphorylation sites contribute to cAMP-sensitivity in additive manner, and the extent of Orco response to cAMP stimulation may thus be regulated by the degree of its phosphorylation via PKC.
To test whether PKC phosphorylation affects the odor response of OSNs in the fly, we combined extracellular recording of OSN activity upon odor stimulation with injection of compounds affecting PLC/PKC activity. For these experiments, a microelectrode was inserted into the antenna near the base of large basiconic ab3 sensilla housing OSNs expressing the receptor protein Or22a. These neurons were localized under fluorescence using the GAL4-UAS system to drive GFP expression in Or22a expressing neurons (Dobritsa et al., 2003). Stimulation with ethyl butyrate (Etb; log[−5] dilution) accelerated the firing frequency of the Or22a expressing neuron (A in Figures 7A,B; f norm = 3.72 ± 0.73, stimulus window, f norm = 2.37 ± 0.22, total response, n = 11) without affecting the other neuron B), as shown by Hallem et al. (2004). Injection of the PLC inhibitor U73122 into sensilla diminished the response to Etb in comparison to sham injection (Figures 7A,C; f norm = 2.52 ± 0.43, stimulus window, n = 13; P = 0.014, Mann-Whitney U ). By contrast, the inactive analog U73343 had no effect (f norm = 4.67 ± 1.04, stimulus window, n = 11, P = 0.870, Mann-Whitney U ). The PKC inhibitor Gö6976 reduced the odor response in a manner similar to U73122 (Figures 7A,C; f norm = 2.57 ± 0.43, stimulus window, n = 12; P = 0.023, Mann-Whitney U ). By contrast, injection of the PKC activators OAG or PMA caused a robust increase of the Etb response (Figures 7B,D; f norm = 7.06 ± 1.45 and 7.91 ± 1.17, respectively, total response, n = 8 and 10; P = 0.003 and <0.001, Mann-Whitney U ).

DISCUSSION
Here we present evidence that the function of the odorant coreceptor Orco is controlled by its phosphorylation state via PKC. Regulation of ligand sensitivity by PKC phosphorylation has earlier been observed in, e.g., CNG channels. PKC activity can either enhance cGMP sensitivity (Müller et al., 1998) or reduce it (Müller et al., 2001). In the latter case the phosphorylation site was localized within the cGMP binding domain. Under normal physiological conditions the basal activity of PLC and PKC in HEK293 cells is sufficient to maintain Orco sensitivity to cAMP stimulation. Conditions leading to inhibition of these enzymes such as low free Ca 2+ concentration suppress the activation of Orco by cAMP and thus may affect the odorant response. By contrast, high PLC/PKC activity would activate Orco independently of cAMP.
Inhibition of PLC function by GDP-β-S in HEK293 cells expressing Orco indicates constitutive activity of Orco-G q protein pairs. Constitutive activity of GPCR-G protein pairs was, for example, reported in other receptors such as thromboxane receptors (Chillar et al., 2010) and mutations in adrenergic receptors (Cotecchia, 2010). In insect ORs, constitutive activity causing a receptor current in the absence of stimuli occurs in various heterologously expressed receptor heterodimers (Sato et al., 2008;Wicher et al., 2008), as well as in solely expressed Orco ( Figure 4B). For Orco, background activity of PLC or PKC seems to be sufficient to maintain a phosphorylation state required for a constitutive activity ( Figure 4B).
Odor stimulation of Or22a, either solely expressed or coexpressed with Orco in HEK293 cells, activated G s proteins but not G q proteins (Wicher et al., 2008). If this is a general rule Frontiers in Cellular Neuroscience www.frontiersin.org (which remains to be shown), odorant responses would not induce G q downstream signaling. This signaling cascade would thus be available to modulate OR function, for example by neuromodulators. The experiments using combined single sensillum recordings and microinjection demonstrate the significance of up-and down-regulation of PLC/PKC activity for the odor response of OSNs. Enzyme inhibition reduced the frequency dynamics of the OSN response, while stimulation of PKC produced a more robust and prolonged OSN response (Figure 7). A recent study utilizing similar extracellular recordings in transgenic flies with various G protein mutations failed to see any effect of G proteins on the in vivo olfactory response (Yao and Carlson, 2010). However, genetically manipulated animals could contain some counter-regulation of the metabotropic effects such as adaptation or up-regulation of regulating enzymes (e.g., phosphodiesterases). This can even occur during transient expression of constitutively active G proteins. An unexpected result of the experiments with excised patches from HEK293 cells expressing Orco was the slow activation kinetics of the current after fast cAMP stimulation (Figure 3). The heterologous system may lack components that in vivo accelerate its activation. However, in the case that this slow time course of Orco activation would be similar in the OSNs, metabotropic signaling would not be able to contribute to the fast odor response of these neurons. For example, a 0.5-s odor pulse gives rise to a response terminating after 2 s (Figure 7), whereas the response of Orco to cAMP took tens of seconds to develop (Figure 3).
It must be noted that the compound microinjection mimicked metabotropic signaling processes initiated in a temporal domain before the odor stimulation.
Our study was not designed to determine whether insect ORs are mixed ionotropic and metabotropic receptors (Wicher, 2010) or metabotropically modulated ionotropic receptors (Nakagawa and Vosshall, 2009). However, the PKC mutant of Orco provides a useful tool to address this question in future investigations that assess the relationship between ionotropic and metabotropic signaling.

CONCLUSION
We have demonstrated a significant impact of the metabotropic pathway on olfactory response both in heterologous in vitro studies as well as in the Drosophila fly itself. Orco phosphorylation via PKC regulates OR sensitivity to cAMP, and therefore to odorants, and it may activate the receptor even in the absence of cAMP. This provides a powerful mechanism to adapt OR sensitivity not only via G q proteins but also via free [Ca 2+ ] i levels.