PROTEIN KINASE Cα MEDIATES A NOVEL FORM OF PLASTICITY IN THE ACCESSORY OLFACTORY BULB

Abstract

Modification of synapses in the accessory olfactory bulb (AOB) is believed to underlie pheromonal memory that enables mate recognition in mice. The memory, which is acquired with single-trial learning forms only with coincident noradrenergic and glutamatergic inputs to the AOB. The mechanisms by which glutamate and norepinephrine (NE) alter the AOB synapses are not well understood. Here we present results that not only reconcile the earlier, seemingly contradictory, observations on the role of glutamate and NE in changing the AOB synapses, but also reveal novel mechanisms of plasticity. Our studies suggest that initially, glutamate acting at Group II metabotropic receptors and NE acting at α₂-adrenergic receptors inhibit N-type and R-type Ca²⁺ channels in mitral cells via a G-Protein. The N-type and R-type Ca²⁺ channel inhibition is reversed by activation of α₁-adrenergic receptors and protein kinase Cα (PKCα). Based on these results, we propose a hypothetical model for a new kind of synaptic plasticity in the AOB that accounts for the previous behavioral data on pheromonal memory. According to this model, initial inhibition of the Ca²⁺ channels suppresses the GABAergic inhibitory feedback to mitral cells, causing disinhibition and Ca²⁺ influx. NE also activates phospholipase C (PLC) through α₁-adrenergic receptors generating inositol 1,4,5-trisphosphate and diacylglycerol (DAG). Calcium and DAG together activate protein kinase Cα (PKCα) which switches the disinhibition to increased inhibition of mitral cells. Thus, PKCα is likely to be a coincidence detector integrating glutamate and NE input in the AOB and bridging the short-term signaling to long-term structural changes resulting in enhanced inhibition of mitral cells that is thought to underlie memory formation.

The ability to change the strength of synapses allows organisms to learn and form memories. Much of our knowledge of such plasticity comes from studies on long-term potentiation and depression (Malenka and Bear, 2004), which are thought to underlie memory formation during multiple-trial learning. Synaptic plasticity underlying single-trial learning, on the other hand, is less well studied. We have investigated the mechanism of synaptic plasticity in the accessory olfactory bulb (AOB) that is associated with the formation of the pheromonal memory underlying mate recognition in mice.

Female mice form a robust, long-lasting memory for the pheromones of their mate during a sensitive period around the time of mating. Exposure to pheromones from a male that is genetically dissimilar to her mate causes a recently mated female to return to estrus, blocking her pregnancy (Brennan and Keverne, 1997). The recognition of her mate's pheromones prevents the mate from causing abortion of his own offspring, and is therefore vital for reproductive success. The male pheromones are detected by the sensory neurons in the vomeronasal (VN) organ, which project to mitral/tufted cells (henceforth referred to as `mitral cells') in the AOB (Mori, 1987; Shepherd, 2004; Ma, 2007), via the VN nerve. Mitral cells project from the AOB to the corticomedial amygdala, which in turn projects via the bed nucleus of the stria terminalis to the medial hypothalamus. Activation of this pathway by pregnancy blocking pheromones from an unfamiliar male leads to a decrease in prolactin secretion from the pituitary and a consequent reduction in circulating progesterone and failure of embryo implantation. Mitral cell activity in the AOB is regulated by inhibitory feedback from granule cell interneurons at reciprocal dendrodendritic synapses (Shepherd, 2004). Memory formation is hypothesized to result from long-lasting increases in the inhibitory feedback on those mitral cells that respond to the mating male's pheromones (Brennan and Keverne, 1997), preventing them from activating the pregnancy blocking pathway. In support of this hypothesis, GABA release in the AOB has been shown to increase during and after pheromonal memory formation (Brennan et al., 1995).

Memory formation requires the association of pheromonal input mediated by glutamate release from the VN nerve onto mitral cells and the release of NE in the AOB from locus ceruleus projections, which signal that mating has occurred (Brennan and Keverne, 1997). Infusion of Group II (mGluR₂/mGluR₃; henceforth referred to as mGluR_2/3) metabotropic glutamate receptor agonist DCG-IV into the AOB obviates the requirement for NE and induces memory formation during exposure to male pheromones, in the absence of mating (Kaba et al., 1994). DCG-IV has been found to presynaptically inhibit GABA release from granule cells in AOB slices in vitro (Hayashi et al., 1993). The decrease in GABA release caused by mGluR_2/3 activation is inconsistent with the long term increase in mitral cell inhibition that is hypothesized to underlie mate recognition. Thus far no satisfactory explanation has been obtained for the mechanism by which activation of mGluR_2/3 or adrenergic receptors in the AOB causes increased inhibition of mitral cells which is thought to underlie memory formation. We therefore investigated the cellular effects of glutamatergic and adrenergic signaling using acute AOB slices with a view to identify the mechanisms of synaptic plasticity in the AOB that might explain pheromonal learning.

EXPERIMENTAL PROCEDURES

Recordings from AOB slices

Balb/c mice were obtained from Charles River (Wilmington, MA) and used for experiments using a protocol approved by the Institutional Animal Care and Use Committee of Wake Forest University Health Sciences. Female mice between 2–4 weeks of ages were used for optimal patch-clamp experiments in AOB slices. Our previous studies have shown that in these mice all the major components of the glutamatergic and adrenergic signaling pathway are expressed at levels comparable to the AOB in adult female mice (Skinner et al., 2008). Sagittal sections (200 μM) were prepared using a vibratome in an oxygenated and chilled solution (incubation solution) containing 124 mM NaCl, 5 mM KCl, 10 mM Glucose, 2 mM CaC1₂, 2.9 mM MgSO4, 3 mM NaH₂PO₄ and 23 mM NaHCO₃ (pH 7.4). The slices were recovered for at least 1 h in the incubation solution. The extracellular solution contained 105 mM NaCl, 2.5 mM KCl, 20 mM Glucose, 2 mM CaC1₂, 1 mM MgCl₂, 1.25 mM NaH₂PO₄ and 26 mM NaHCO₃ (pH 7.4). Whole cell recordings were carried out using a patch electrode pulled from borosilicate glass with a resistance of 4–6 MΩ, filled with a solution containing CsC1 130 mM, 2 mM MgCl₂, 20 mM N-[2-hydroxy-ethyl] piperazine-N'-2-ethanesulfonic acid (HEPES), 4 mM ATP, 0.5 mM GTP, 11 mM EGTA, 1 mM CaCl₂; pH 7.2 (Edwards et al., 1989; Hayashi et al., 1993; Jia et al., 1999). Mg-ATP was used in the internal solution in order to minimize the run-down of Ca²⁺ channels (Kepplinger KJF and Romanin C, 2005). Voltage-clamp recordings were done with a Multiclamp 700A amplifier. The microscope was equipped with infrared differential interference optics and a charge-coupled device camera which allowed the experimenter to visually identify the AOB neurons and to guide the patch electrodes. Mitral cell and granule cell layers were readily distinguishable in the AOB slices. Mitral cells, which are considerably larger than granule cells, were identified based on their size and morphology. After a gigaohm seal and whole cell access was achieved, the series resistance was between 10 and 20 mΩ and was compensated (60–80%) by the amplifier. Cells with more than 20% variation in input resistance were discarded. High threshold Ca²⁺ currents were recorded under voltage-clamp in the presence of tetrodotoxin (TTX) (1 μM) and triethylammonium (20 mM). For the study of current-voltage relationships, mitral cells were held at −80 mV and currents were recorded in response to +10mV steps.

Spontaneous and evoked IPSC measurements from mitral cells were made with a pippette solution containing (in mM): 125 KCl, 2MgCl₂, 2 CaCl₂, 10 EGTA, 2 NaATP, 0.5 NaGTP, and 10 HEPES, adjusted to pH 7.3 with KOH. All data were filtered at 1 kHz and digitized at a sampling rate of 10 kHz. Data were required at a holding potential of 0mV in the presence of 1μM TTX (Schoppa et al., 1998). For stimulation of the glomerular layer for measurement of evoked IPSCs, a bipolar tungsten stimulating electrode (Frederick Haer & Co, Brunswick, ME), was used. Stimulus pulses were generated by Digidata1322A (Axon Instruments, Sunnyvale, CA) which triggered a stimulus isolation unit (Maximum 2 mA stimulation with duration of 100 μsec).

While space clamp concerns cannot be eliminated from studies of neurons in slice preparations, we took the following steps to assess and minimize it. Poor space clamp is indicated by an increase in the delay to onset of inward current with increasing magnitude of the command pulse, or by rapidly inactivating inward currents that were elicited by steps from depolarizing test potentials back to the holding potential. Cells that included these response features were eliminated from our analysis. In addition, all reported quantitative measurements were taken at steady state (250 msec after the test pulse) to minimize the effect of inadequate space clamp on the transient phase of the current. Voltage output of the amplifier was recorded in parallel with current traces to monitor escape voltage transients (Halliwell et al., 1994; Graef et al., 2009).

Drug application

The concentrations of the pharmacological and other reagents used were as follows:

BAPTA (1,2-Bis [2-aminophenoxy] ethane-N,N,N',N'-tetraacetic acid): 10 mM; Bisindolylmaleimide I (Bis I): 2 μM; Bisindolylmaleimide V (Bis V): 2 μM; Chelerythrine: 10 μM; Cirazoline: 10 μM; Clonidine: 5 μM; DCG-IV: 0.5 μM; GDβPS: 100 μM; GTPγS: 100 μM; Glutamate: 500 μM; HBDDE (2, 2', 3, 3' 4, 4'-Hexahydroxy-1, 1'biphenyl6-6'-dimethanol dimethyl ether: 50 μM; L-CCG-I: 10 μM; LY333531 (low): 10 nM; LY333531 (high): 1 μM; LY379268: 10 μM; NE: 30 μM; Nifedipine: 10 μM; Picrotoxin: 200 μM; PMA (phorbol myristate acetate): 500 nM; PTX (pertussis toxin): 1 μg/ml; Recombinant PKCα: 2 μM; Recombinant PKCγ: 2 μM; PKCα inhibitor (Compound 6): 20 μM; SNX-482: 100 nM; UK14304: 5 μM; U73122: 10 μM; U73343: 10 μM; ω-conotoxin-GVIA: 1 μM; ω-conotoxin-MVIIC: 1 μM. Stock solutions of Bis I, Bis V, LY333531, nifedipine, picrotoxin, PMA, U73122,U73343, UK14304 were prepared in dimethyl sulphoxide (DMSO). For these reagents the same volume of DMSO alone was used for controls. All other reagents were dissolved in distilled water.

For testing the effect of PTX, we used the A-protomer (the subunit containing enzymatic activity) (Biomol, Plymouth Meeting, PA) in the patch pipette. The A-protomer, because of its small size (28 K) can be efficiently dialysed into cells and has been used successfully in several studies (Elmslie, 1992; Kurachi et al., 1986; Xie and Lewis, 1997). The A-protomer was pre-incubated with 10 mM dithiothreitol at 37°C for 10 min to convert it to its enzymatically active form and was diluted in the internal solution (1 μg/ml final concentration) along with 1 mM nicotinamide adenine dinucleotide (NAD). For testing the effect of protein kinase C isoforms, purified recombinant PKCα and PKCγ (Upstate USA Inc., Charlottesville, VA) were diluted in the internal solution (2 μM final). Heat-inactivated PTX and PKC isoforms were used as controls. PKCα Compound 6 (kindly provided by David Larsen, Albert Einstein School of Medicine, New York) and the peptides (Quality Controlled Biochemicals, Hopkinton, MA) were included in the patch pipette.

For testing the effect of ω-conotoxin-GVIA, ω-conotoxin-MVIIC, SNX-482 and picrotoxin, the AOB slices were incubated in ACSF containing the toxin for 15 min.

Data were collected from the same cells while using drugs that could be washed out. For experiments with drugs that could not be washed out or those with irreversible effects (PTX, ω-conotoxin-GVIA, ω-conotoxin-MVIIC, SNX-482, picrotoxin, GDβPS and GTPγS, recombinant PKCα and PKCγ, and PKCα, PKCγ), different cells were used for data collection with each reagent and the experiments were carried out under comparable conditions.

Data analysis

Data were analyzed using one-way analysis of variance (ANOVA) followed by a post-hoc Tukey test (pairwise multiple comparison procedures) while comparing multiple groups for treatment effect. In these instances, the F and the P values for the overall comparison are given first and then the P values from the post-hoc Tukey test are given next to the groups being compared. For comparison between two groups, Student's t-test was used. Cumulative probability distributions of IPSC amplitude and frequency were analyzed using Kolmogorov-Smirnov test. The values are expressed as mean ± standard error. The P-values between 0.05 and 0.01 are indicated by one asterisk (*) or a number sign (#) in the figures; P-values less than 0.01 are indicated by two asterisks (**) or number signs (##). The P-values indicate significant difference with respect to controls unless otherwise stated in the text or otherwise indicated in the figures. Comparisons between other groups are stated in the text where necessary and in the figures such comparisons are indicated by lines between bars that are being compared.

RESULTS

mGluR_2/3 and α₂-adrenergic receptors (α₂-ARs) inhibit Ca²⁺ currents in the mitral cells of the AOB

In the AOB, mGluR_2/3 receptors disinhibit mitral cells by presynaptic inhibition of GABA release from the granule cell side of the reciprocal synapses. The mechanism underlying disinhibition is unclear. One possibility is that mGluR_2/3 close voltage-gated Ca²⁺ channels (Swartz, 1993; Catterall, 2000), thus controlling glutamate release from mitral cells or GABA release from granule cells, either of which leads to disinhibition of the mitral cells. The presence of mGluR_2/3 in mitral cells has been demonstrated through previous immunocytochemical experiments (Ohishi et al., 1998; Sahara et al., 2001). To test the idea that mGluR_2/3 reduces the Ca²⁺ current, we isolated the high-threshold Ca²⁺ current in mitral cells in the whole cell mode under voltage-clamp conditions and added mGluR_2/3 agonist L-CCG-I (10μM) (Hayashi et al., 1992). Typical Ca²⁺ currents in untreated (control) and L-CCG-I-treated mitral cells are shown in Fig. 1 A & B. We also added two other mGluR_2/3 agonists LY379268 (Monn et al., 1999) or DCG-IV (Hayashi et al., 1993)(10μM each) to the bath. All the three agonists L-CCG-I (56.7 ± 7.6%; P <0.01; n = 11), LY379268 (56.7 ± 9.9%; P <0.01; n = 6), and DCG-IV (58.3 ± 14.5%; P <0.05; n = 5) significantly reduced the Ca²⁺ current relative to controls (100%) (Fig. 1C).

Fig. 1. mGluR_2/3 and α₂-AR agonists reduce Ca²⁺ currents through a G-protein.

(A) Ca²⁺ currents were evoked by 800-ms pulses from a holding potential of −80 mV to various test potentials. Current traces recorded without any treatment (Control) after depolarizations to potentials between −50 mV and −20 mV. (B) Superimposed traces recorded 5 min after application of 10 μM L-CCG-I with the same protocol as in A. The voltage steps and the calibration bars apply to both A and B. (C) Application of mGluR_2/3 agonists L-CCGI (10 μM), DCG-IV (0.5 μM), and LY379268 (10 μM) causes reduction in Ca²⁺ current amplitude relative to controls (black bars). Separate control recordings were made for each agonist. (D) Application of α₂-AR agonists clonidine (5 μM) and UK 14304 (5 μM) reduces Ca²⁺ current amplitude relative to controls (black bars). Representative intracellular traces and cumulative data are shown for each agonist. (E) GDPβS (100 μM) and PTX (1 μg/ml) increase the Ca²⁺ current while GTPγS (100 μM) decreases the Ca²⁺ current. In the presence of GDPβS or PTX, L-CCG-I fails to reduce the Ca²⁺ current. (F) Time course of GDPβS and GTPγS effect on the Ca²⁺ current. GDPβS and GTPγS were included in the patch pipette. The current was allowed to stabilize for 4 min after which the time course was determined (*P<0.05; **P<0.01).

The current exhibited characteristic properties of Ca²⁺ currents: we were able to substitute Ba²⁺ as charge carrier, and the current was blocked by Cd²⁺ (1.6 ± 1.4%, P <0.001 relative to control [100%]; n = 9) (Supplementary Fig. S1). The run-down (decrease over time) of Ca²⁺ channels is a commonly observed phenomenon as has been reported previously (Kepplinger KJF and Romanin C, 2005). When we recorded the Ca²⁺ current in several mitral cells (n=17) without any treatment, we observed an average of 7.5% run down at the end of a 15-min period.

To test whether α₂-ARs control Ca²⁺ channels (Li et al., 1998) in the AOB, we isolated the Ca²⁺ current and applied α₂-AR agonists clonidine or UK14304 (5 μM each) (Andorn et al., 1988) to the bath. Both clonidine (52.5 ± 9.4%; P <0.01; n = 10) and UK14304 (48.3 ± 13.3%; P <0.01; n = 6) significantly inhibit the Ca²⁺ current relative to controls (100%) (Fig. 1D)

Inhibition of Ca²⁺ channels by mGluR_2/3 is mediated by a G-protein

The effect of mGluR_2/3 on Ca²⁺ channels could be mediated by direct coupling via G-proteins as has been shown in cultured cortical neurons (Swartz, 1993). If the reduction in Ca²⁺ current were mediated by a G-protein, reagents that inhibit G-proteins should prevent the reduction in the Ca²⁺ current caused by L-CCG-I. Inclusion of inhibitors of G-proteins GDPβS or pertussis toxin (PTX) (Swartz, 1993; Ikeda et al., 1995) in the patch pipette not only reduced the inhibition but significantly increased the Ca²⁺ current relative to controls but inclusion of a G-protein activator GTPγS significantly reduced the Ca²⁺ current. In the presence of GDPβS or PTX, mGluR_2/3 agonist L-CCG-I failed to decrease the Ca²⁺ current (Fig 1E) (see Supplementary Table S1 for details). These results indicate that mGluR_2/3 acts on Ca²⁺ channels via an inhibitory G-protein. Time course experiments with GDPβS showed an increase in the Ca²⁺ current relative to controls whereas time course with GTPγS showed a decrease in the Ca²⁺ current further supporting the role of a G-protein in Ca²⁺ channel inhibition (Control at 4 min: 100%; At 10 min, Control: 96 ± 2.1%, GDPβS: 135.6 ± 8.1%, P <0.01; GTPγS: 62.0 ± 8.7%, P<0.01; At 20 min: Control: 96 ± 1.4%, GDPβS: 145.4 ± 7.9%, P <0.01; GTPγS: 36.0 ± 6.8%; P<0.01) (Fig. 1F). In the presence of GDPβS and PTX, the Ca²⁺ currents were higher than untreated controls indicating that these reagents removed tonic inhibition (Swartz, 1993; Yang and Tsien, 1993; Zhu and Ikeda, 1994) of Ca²⁺ channels by G-proteins. Further increase in the Ca²⁺ current with the addition of L-CCG-I suggests that perhaps mGluR_2/3 activation causes a cycle of G-protein activation and formation of additional hetero-trimers because of Gα association with Gβγ which in turn should increase the efficacy of GDPβS and PTX. Overall, the results indicate that mGluR_2/3 receptors inhibit Ca²⁺ channels by direct coupling through G-proteins.

Mutual occlusion of mGluR_2/3 and α₂-AR effects on the Ca²⁺ current

Since agonists of both mGluR_2/3 and α₂-AR cause reduction in the Ca²⁺ current, we tested whether these receptors exert their inhibitory effect on the Ca²⁺ current through similar or dissimilar mechanisms. We first administered L-CCG-I, the mGluR_2/3 agonist followed by clonidine, the α₂-AR agonist. As before, L-CCG-I reduced the Ca²⁺ current (62.8 ± 8.5%; P <0.05) relative to controls (100%) but clonidine did not have an additive effect on Ca²⁺ current (58.8 ± 7.5%; P = 0.1873; n = 8) (Fig. 2A) The converse experiment (administration of clonidine followed by L-CCG-I) yielded similar results (Control: 100%; Clonidine: 53.9 ± 5.9%, P <0.01; Clondine + L-CCG-I: 51.1 ± 6.0%; P = 0.38 Clondine + L-CCG-I compared to Clonidine alone; n = 7; paired t-test) (Fig. 2B). Thus mGluR_2/3 and α₂-AR occlude each other's effect. It is possible, however, that occlusion could occur if either L-CCG-I or clonidine had caused maximal inhibition of Ca²⁺ current possible under the recording conditions.

Fig. 2. Mutual occlusion of L-CCG-I and clonidine effect on the Ca²⁺ currents.

(A) L-CCG-I occludes the inhibitory effect of clonidine of Ca²⁺ currents. Application of L-CCG-I (10 μM) causes significant (*P<0.05) reduction in the Ca²⁺ current compared to controls (100%). Clonidine (5 μM) application following L-CCG-I application does not reduce Ca²⁺ current further. (B) Clonidine occludes the inhibitory effect of L-CCG-I on Ca²⁺ currents. Clonidine reduces the Ca²⁺ current significantly (**P <0.01) relative to controls (100%) but application of L-CCG-I application following clonidine does not additionally decrease the Ca²⁺ current.

The Ca²⁺ current in mitral cells is mediated by N- and R- type Ca²⁺ channels

Next, we sought to characterize the type of Ca²⁺ channels that mediate the Ca²⁺ currents in mitral cells. The Ca²⁺ current was reduced to 43.2 ± 3.70% compared to controls (100%) in the presence of ω-conotoxin-GVIA (1 μM), an N-type Ca²⁺ channel blocker (Tsien et al., 1988) (Fig. 3). Although the inhibition of the Ca²⁺ current by ω-conotoxin-GVIA was significant (P<0.01; n=6) a portion of the current remained resistant to the toxin. ω-conotoxin-MVIIC (P/Q-type Ca²⁺ channel blocker, 1μM) and nifedipine (L-type Ca²⁺ channel blocker, 10 μM) are without effect (Supplementary Fig. S1) suggesting that the Ca²⁺ current is not likely to be mediated by P/Q or L-type channels. Therefore, we tested the effect of SNX-482, an R-type Ca²⁺ channel blocker (Newcomb et al., 1998). SNX-482 reduced the Ca²⁺ current to 59.9 ± 5.3% relative to controls (100%) (Fig. 3). To test whether N-type and R-type channels carry bulk of the Ca²⁺ current, we applied both ω-conotoxin-GVIA and SNX-482. Together these two reagents reduced the Ca²⁺ current to 3.1 ± 1.2% relative to controls (100%) (Fig. 3).

Fig. 3. The Ca²⁺ current is mediated by N-type and R-type Ca²⁺ channels.

The Ca²⁺ current is significantly (**P <0.01) but not completely inhibited by an N-type Ca²⁺ channel inhibitor ω-conotoxin-GVIA (1 μM) and an R-type Ca²⁺ channel inhibitor SNX-482 (100 nM) (*P <0.05) relative to control (100%). Together, the two drugs almost completely inhibit the Ca²⁺ current (*P <0.05) relative to control.

PKC activation reverses mGluR_2/3- and α_2-AR- mediated decrease in the Ca²⁺ current

Activation of mGluR_2/3 or α_2-AR reduces the Ca²⁺ influx via a G-protein and thus could cause a reduction in glutamate release from mitral cells or GABA release from granule cells. Either or both of these effects would lead to disinhibition of mitral cells in the AOB. In order for pheromonal memory to form, however, disinhibition has to be converted to increased inhibition. A molecule that can potentially reverse the action of mGluR_2/3 or α_2-AR is PKC, which is known to uncouple G-proteins from N-type (Swartz, 1993; Zamponi et al., 1997) and R-type (Bannister et al., 2004) Ca²⁺ channels. To test whether PKC has an effect on Ca²⁺ currents, we applied an activator of PKC, phorbol 12-myristate-13-acetate (PMA). We observed that PMA (500 nM) reverses the decrease in the Ca²⁺ current by L-CCG-I (ANOVA: F_(2,30) = 14.72, P < 0.001; Control: 100%; L-CCG-I: 56.5 ± 7.3%, P <0.05; L-CCG-I + PMA: 145.0 ± 18.8%; P <0.01 compared to L-CCG-I and P <0.01 compared to control; n = 11) (Fig. 4A). To test whether PMA acts through a Ca²⁺-sensitive PKC, we carried out the experiments with a Ca²⁺ chelator BAPTA (10 mM). When BAPTA was present in the internal solution, PMA application did not reverse the L-CCG-I-mediated inhibition of the Ca²⁺ current (ANOVA: F_(2,18) = 8.134, P=0.003; Control: 100%; L-CCG-I: 55.5 ± 7.7%, P <0.01; L-CCG-I + PMA: 57.0 ± 13.3, P=0.992 L-CCG-I + PMA compared to L-CCG-I alone) (Supplementary Fig. S2). We studied the current-voltage relationships of L-CCG-I and PMA. We noticed that PMA leads to rapid activation of the Ca²⁺ current at more negative test pulses and the increase in current amplitude occurs over the entire range of test pulses (Fig 4B top panel). Activation of Ca²⁺ current at negative test pulses has been shown to be enhanced because of removal of tonic inhibition of Ca²⁺ channels by G-proteins (Swartz, 1993). We also tested the effect of L-CCG-I and PMA in time course experiments. In these experiments L-CCG-I was maintained in the bath for 10 min after which PMA was added to bath for 4 min. L-CCG-I decreased the Ca²⁺ current and PMA reversed the effect of L-CCG-I and increased Ca²⁺ current. The increase in the Ca²⁺ current was maintained as long as PMA was present in the bath and gradually decreased after washout of PMA (At 15 min, L-CCG-I: 42.9 ± 7.9%, P< 0.01; at 19 min, L-CCG-I + PMA: 131.1± 10.2%; P< 0.05, L-CCG-I + PMA at 19 min compared to L-CCG-I at 15 min; n = 8) (Fig. 4B bottom panel).

Fig. 4. PKC activator PMA reverses the inhibition of Ca²⁺ current by mGluR_2/3 and α₂-AR.

(A) L-CCG-I (10 μM) reduces the Ca²⁺ current significantly (**P <0.01) compared to control and PMA (500 nM) reverses L-CCG-I-mediated inhibition (^##P <0.01). (B) Current-voltage relationship (top) and time course (bottom) of PMA-mediated reversal of inhibition of Ca²⁺ current by L-CCG-I. (C) Clonidine reduces the Ca²⁺ current significantly (*P <0.05) compared to controls and PMA reverses clonidine-mediated inhibition (^##P <0.01). (D) Current-voltage relationship (top) and time course (bottom) of PMA-mediated reversal of inhibition of Ca²⁺ current by clonidine.

We carried out similar experiments on the effect of α₂-AR agonist clonidine on the Ca²⁺ current. The current-voltage relationships of clonidine and PMA and the time course studies with these reagents were similar to that observed with L-CCG-I and PMA. (ANOVA: F_(2,21) = 13.22, P < 0.001; Control: 100%; Clonidine: 52.8 ± 12.8%, P <0.05; Clonidine + PMA: 132.0 ± 13.5%, P <0.01 Clonidine + PMA compared to Clonidine alone; P <0.05 Clonidine + PMA compared to control; n = 8) (Time course: at 15 min, Clonidine: 35.5 ± 6.8%; P< 0.01; at 19 min, Clonidine + PMA: 125.5 ± 8.4%; P< 0.05, Clonidine + PMA at 19 min compared to Clonidine alone at 15 min; n = 8) (Fig. 4C & D).

The time course of PMA effect on the Ca²⁺ current is similar to that observed in frog sympathetic neurons (Yang and Tsien, 1993). The extent of the Ca²⁺ current with L-CCG-I + PMA and Clonidine + PMA is comparable to the effect of PMA alone (see next section) suggesting reversal of Ca²⁺ channel inhibition which is in agreement with recent studies on rat sympathetic neurons (Diaz-Cardenas et al., 2008).

The α isoform of PKC mediates the effect on the Ca²⁺ currents in mitral cells

There are twelve mammalian PKC isoforms (Mellor and Parker, 1998) and different PKC isoforms appear to have specific roles in other systems (Sajikumar et al., 2005; Sossin, 2007; Sacktor, 2008). To understand the precise role of PKC in AOB plasticity, it is important to identify the PKC isoform critical for uncoupling mGluR_2/3 and α₂-AR from Ca²⁺ channels and for increases in Ca²⁺ current. To identify the PKC isoform, first we standardized our experiment using bath application of chelerythrine (10 μM) which inhibits all isoforms of PKC. We found that chelerythrine blocked the PMA-mediated increase in the Ca²⁺ currents. We narrowed down the list of PKC isoforms for our study by reasoning that one of four Ca²⁺ -and diacylglycerol (DAG)- dependent (classical) PKC isoforms (α, βI, βII, γ) might act on Ca²⁺ channels because glutamate receptors can cause a rise in intracellular Ca²⁺ and adrenergic receptors can stimulate production of DAG. Blockade of the PMA effect in reversing the L-CCG-I-mediated inhibition of the Ca²⁺ current in the presence of BAPTA (see Supplementary Fig. S2) ruled out participation of novel PKCs, which are DAG-dependent but Ca²⁺-insensitive (Mellor and Parker, 1998), further supporting a role for classical PKC isoforms. An inhibitor specific to PKCα and PKCγ HBDDE (Kashiwada et al., 1994; Mathur and Vallano, 2000) significantly inhibited the increase in Ca²⁺ current caused by PMA whereas an inhibitor of βI and βII isoforms of PKC, LY 333531 (10 nM), had no effect (Fig. 5A). At higher concentrations LY333531 can inhibit PKCα and PKCγ (Shen et al., 2003). Application of 1 μM LY333531 inhibited the Ca²⁺ current (Fig. 5A). Furthermore, a highly potent and specific inhibitor of PKCα, Compound 6 (20 μM) (Lee et al., 2004) (P <0.05; n = 11), blocked the PMA effect on Ca²⁺ current (Fig. 5B) (see Supplementary Table S2 for details). Moreover, inclusion of purified recombinant PKCα but not PKCγ in the patch pipette resulted in an increase in Ca²⁺ current (At 10 min, Control: 98.3 ± 1.7%; PKCα: 190.7 ± 11.4%, n = 7, P < 0.01; PKCγ: 102.5 ± 2.4%, P = 0.279; At 15 min, Control: 97.5 ± 1.78; PKCα: 200.4 ± 11.0%, P < 0.01; PKCγ: 98.3 ± 1.7%, n = 6, P = 0.752) (Fig. 5C, top panel). Application of a PKC inhibitor chelerythrine inhibited the increase in Ca²⁺ current caused by purified PKCα and inhibitory effect is reversed after the washout of the drug (At 10 min [just before chelerythrine application], 195.0 ± 9.0%; P < 0.01 relative to baseline; At 15 min (just before washout of chelerythrine): 109.8 ± 2.3%, P < 0.01, 15 min compared to 10 min; at 20 min [i.e. after washout of chelerythrine for 5 min], 152.5 ± 8.0%; P <0.01, 20 min compared to 15 min) (Fig. 5C, bottom panel).

Fig. 5. PKCα is the PKC isoform that uncouples mGluR_2/3 and α₂-AR from Ca²⁺ channels.

(A) Inhibition of PMA-mediated increase in Ca²⁺ current (*P <0.05 or **P<0.01 versus control) by chelerythrine (10 μM), HBDDE (50 μM) and high concentration (1 μM) of LY333531 (^##P <0.01 versus PMA) but not low concentration (10 nM) of LY333531. In the cumulative data shown in the bar graph, the third bar in each cluster represents the effect of the inhibitor. Representative intracellular traces are shown for each condition (1: Control; 2: PMA; 3: inhibitor). (B) Inhibition of PMA-mediated increase in Ca²⁺ current by a specific PKCα inhibitor. (C) Top Panel: Increase in Ca²⁺ current by purified PKCα but not PKCγ included in the patch pipette. Values for control time course overlap with that for PKCγ and hence are not shown in this graph. The traces at left show increase in the Ca²⁺ current with PKCα while the traces for PKCγ (right) show no change. Bottom Panel: Time course PKCα-mediated increase in Ca²⁺ current and its inhibition by chelerythrine. (D) PKCα C-terminal peptide but not PKCγ C-terminal peptide inhibits PMA-mediated increase (**P <0.01 versus control) in the Ca²⁺ current.

Our finding that PKCα but not the closely related isoform PKCγ causes an increase in Ca²⁺ current suggested that a unique region of PKCα might play a role in mediating its effect on Ca²⁺ channels. The C-terminus of PKCα is known to interact with a PDZ domain containing protein PICK1 (protein interacting with C-kinase 1) (Staudinger et al., 1997) which is believed to localize PKCα close to the target of phosphorylation. If this idea is correct, then a PKCα peptide that would compete with the endogenous protein should block the PMA-induced increase in Ca²⁺ current. We tested a peptide with the unique C-terminal 11 amino acids of PKCα. As control of specificity, we also tested 11-amino acid C-terminal peptide from PKCγ. We found that PKCα peptide significantly inhibited the PMA-mediated increase in the Ca ²⁺ current relative to controls (P < 0.01; n = 7) while PKCγ peptide had no effect (P=0.903; n = 7) (Fig. 5D) (see Supplementary Table S2 for details).

α₁-adrenergic receptors increase the Ca²⁺ current via phospholipase C (PLC) and PKC

Our results on reversal of Ca²⁺ current inhibition by PKCα suggest that the kinase must be activated by stimuli that induce pheromonal memory. Classical isoforms of PKC such as PKCα require Ca²⁺ and DAG for activation (Mellor and Parker, 1998). We hypothesized that in addition to its action on α₂-AR, NE also acts on α₁-adrenergic receptor (α₁-AR) activating PLC. Activated PLC produces DAG that can activate PKCα. We tested this hypothesis using cirazoline, an agonist of α₁-AR (Gellai and Ruffolo, Jr., 1987). We found that cirazoline caused an increase in the Ca²⁺ current. This increase was abolished by PLC inhibitor U73122 (Bleasdale et al., 1990) (ANOVA: F_(2,21) = 56.48, P < 0.001; Control: 100%, Cirazoline: 168.8 ± 9.7%, P< 0.001; Cirazoline + U73122: 65.0 ± 7.3%, P <0.001 Cirazoline + U73122 compared to Cirazoline alone, n = 8) whereas an inactive analogue U73343 had no effect on the increase in the Ca²⁺ current caused by cirazoline application (Control: 100%, Cirazoline: 157.3 ± 9.4%; P<0.05; Cirazoline + U73343: 156.7 ± 8.8; P>0.05 Cirazoline + U73343 compared to Cirazoline alone) (Fig. 6A). The PLC inhibitor U73122 alone did not have an effect on the Ca²⁺ current suggesting that inhibition caused by U73122 was through a specific inhibitory effect on PLC (Control: 100%, U73122: 98.8 ± 5.8%, P=0.97, n = 7) (Supplementary Fig. S3). Therefore, PKC activation is likely to be mediated by PLC activation through α₁-AR. Activation of PKC would in turn increase the Ca²⁺ current. If α₁-AR-mediated increase on the Ca²⁺ current is mediated by PKC, inhibition of PKC should decrease the α₁-AR-mediated increase in the Ca²⁺ current. Indeed, the PKC inhibitor chelerythrine significantly reduced the effect of cirazoline on the Ca²⁺ current (ANOVA: F_(2,14) = 8.73, P=0.002; Control: 100%, Cirazoline: 151.9 ± 8.4%, P <0.01; Cirazoline + Chelerythrine: 96.0 ± 9.1%, P <0.01 Cirazoline + Chelerythrine compared to Cirazoline alone) (Fig. 6B). To obtain additional evidence for a role of PKC, we used the PKC inhibitor Bis I and an inactive analogue Bis V. We observed that Bis I significantly reduced cirazoline-mediated increase in the Ca²⁺ current (ANOVA: F_(2,21) = 12.18, P < 0.001; Control: 100%, Cirazoline: 164.8 ± 11.4%, P <0.05; Cirazoline + Bis I: 90.9 ± 14.1%, P <0.05 Cirazoline + Bis I compared to Cirazoline alone) whereas Bis V did not have a significant effect (Control: 100%, Cirazoline: 148.8 ± 11.9%, P <0.05; Cirazoline + Bis V: 145.2 ± 12.0%; P>0.05 Cirazoline + Bis V compared to Cirazoline alone, n=7) (Fig. 6C). These results suggest that α₁-AR-mediated increase in the Ca²⁺ current is mediated by PLC and PKC.

Fig. 6. α₁-AR increases Ca²⁺ current through PLC and PKC.

(A) Cirazoline (10 μM), an agonist of α₁-AR significantly (**P <0.01 versus control) increases the Ca²⁺ current amplitude relative to control and a PLC inhibitor U73122 (10 μM) inhibits (##P <0.01 versus cirazoline), whereas an inactive analog U73343 (10 μM) does not inhibit the cirazoline effect. In the cumulative data shown in the bar graph, the third bar in each cluster represents the effect of the reagent as indicated (in panels A & C). Representative intracellular traces are shown for each reagent. (B) Cirazoline increases the Ca²⁺ current amplitude and PKC inhibitor chelerythrine inhibits (#P <0.05 versus cirazoline) the cirazoline effect. (C) A PKC inhibitor Bis I (2 μM) inhibits (##P <0.01 versus cirazoline), but the inactive analog Bis V (2 μM) does not inhibit (P>0.05) the cirazoline-mediated increase (*P <0.05 relative to control) in the Ca²⁺ current.

NE and glutamate modulate inhibitory postsynaptic currents in mitral cells

Inhibitory transmission from granule to mitral cells controls the mitral cell output. We therefore investigated the effect of mGluR_2/3 agonist L-CCG-I and PKC activator PMA on spontaneous inhibitory postsynaptic potentials (IPSCs) observed in the mitral cells. L-CCG-I reduced both the frequency and amplitude of IPSCs, and PMA reversed the L-CCG-I effect (Amplitude, ANOVA: F_(2,30) = 13.24, P < 0.001; Control: 100%, L-CCG-I: 69.3 ± 8.2%, P< 0.01; L-CCG-I + PMA: 396.2 ± 82.8%, P <0.01 L-CCG-I + PMA compared to L-CCG-I alone; Frequency, ANOVA: F_(2,30) = 22.88, P < 0.001; Control: 100%; L-CCG-I: 50.8 ± 10.4%, P <0.01, L-CCG-I + PMA: 144.5 ± 13.4%, P <0.05 L-CCG-I + PMA compared to L-CCG-I alone) (Fig. 7). The IPSCs are significantly reduced with picrotoxin, a GABA_A receptor antagonist (Supplementary Fig. S4). Although PMA-mediated reversal of Ca²⁺ current inhibition caused by L-CCG-I suggests a role for PKC based on our other experiments, it is possible that PMA could have an effect on IPSCs independent of PKC. To test this possibility, we measured the effect of PMA alone on IPSCs. PMA increased the amplitude as well as frequency of IPSCs. The PMA effect was significantly inhibited by the PKC inhibitor Bis I (Amplitude, ANOVA: F_(2,15) = 28.16, P < 0.001; Control: 100%, PMA: 203.7 ± 13.0%, P <0.001; PMA + Bis I: 135.2 ± 11.3%, P <0.001 PMA + Bis I compared to PMA alone; Frequency, ANOVA: F_(2,15) = 12.80, P < 0.001; Control: 100%, PMA: 360.7 ± 59.8%, P <0.001; PMA + Bis: 190 ± 56.6%, P <0.05 PMA + Bis I compared to PMA alone) (Supplementary Fig. S5). These results suggest that the PMA effect on IPSCs is mediated by PKC. The PMA and PKC effect on IPSCs most likely occurs through the PKC effect on Ca²⁺ channels although additional direct PKC effect on neurotransmitter release cannot be ruled out.

Fig. 7. mGluR_2/3 agonist L-CCG-I decreases inhibitory transmission in the AOB and PKC activator PMA reverses the L-CCG-I effect.

(A) Cumulative probability distribution showing a decrease in spontaneous IPSC amplitude in L-CCG-I treated mitral cells (significant shift to left relative to control; P <0.01, Kolmogorov – Smirnov [KS] test) and an increase in amplitude in cells treated L-CCG-I + PMA (significant shift to right relative to control, P<0.01, KS test). (B) Summary graph showing the decrease of IPSC amplitude with L-CCG-I (**P<0.01 versus control) and reversal of L-CCG-I effect by PMA (##P<0.01: L-CCG + PMA versus L-CCG-I). (C) Cumulative probability distribution of spontaneous IPSC frequency showing decrease in frequency in L-CCG-I-treated cells (significant shift to right relative to control; P< 0.01, KS test) and increase in frequency in cells treated with L-CCG-I + PMA (significant shift to left relative to control; P <0.01, KS test). (D) Summary graph showing the decrease of IPSC frequency with L-CCG-I (**P < 0.01 versus control) and reversal of L-CCG-I effect by PMA (#P < 0.05: L-CCG-I + PMA versus L-CCG-I).

Next, we tested the effect of L-CCG-I and PMA on IPSC evoked in mitral cells by stimulating the glomerular layer, which evoked IPSCs because of the reciprocal dendrodendritic synapses between mitral and granule cells. When glomerular stimulation excites a mitral cell it releases glutamate onto the granule cell. Excitation of the granule cell causes it to release GABA back onto the mitral cell which causes the IPSC (Schoppa et al., 1998). We observed that L-CCG-I decreased the IPSC evoked in mitral cells by stimulating the glomerular layer, and PMA increased the evoked IPSC (ANOVA: F_(2,15) = 17.79, P < 0.001; Control: 100%; L-CCG-I: 53.3 ± 6.7%, P < 0.01; L-CCG-I + PMA; 208.0 ± 32.5%, P <0.05 L-CCG-I + PMA compared to L-CCG-I alone) (Fig. 8A). Furthermore, application of NE reduced the evoked IPSCs Since PKC is likely to be activated by Ca²⁺ influx through glutamate receptors, we tested application of glutamate together with NE. Glutamate and NE together cause an increase in evoked IPSC (ANOVA: F_(2,15) = 31.96, P < 0.001; Control: 100%, NE: 31.3 ± 7.1%, P < 0.05, NE + Glutamate: 212.0 ± 27.1%; P < 0.01 NE + Glutamate compared to NE alone) (Fig. 8B).

Fig. 8. Regulation of inhibitory transmission in the AOB as measured by evoked IPSCs.

(A) L-CCG-I decreases evoked IPSCs (*P < 0.05 versus control) and PMA reverses the L-CCG-I effect (#P < 0.05: L-CCG-I + PMA versus L-CCG-I). (B) NE (30 μM) decreases evoked IPSCs (*P < 0.05 versus control) and application of NE (30 μM) + glutamate (500 μM) increases the evoked IPSCs (##P < 0.01: NE + Glutamate versus NE).

DISCUSSION

The acquisition of mate recognition memory in female mice, through single-trial learning, depends on the association of glutamate release from AOB mitral cells, resulting from pheromonal input, with the mating-induced NE release in the AOB. This is postulated to result in long-lasting increase in granule cell inhibitory feedback on the AOB mitral cells that respond to the mating male (see Supplementary Fig. S6). The cellular and molecular mechanisms underlying the convergence of the glutamate and NE signals have yet to be established, however. Our findings suggest that activation of α₂-AR and mGluR_2/3 pathways leads to a blockade of N- and R- type Ca²⁺ channels. This most likely leads to a decreased release of glutamate onto granule cells causing decreased GABA release from granule cells which in turn results in disinhibition of mitral cells. Based on our data we infer that this disinhibition of mitral cells is switched to increased inhibition by α₁-AR-mediated activation of PKCα, which blocks the inhibitory effect of mGluR_2/3 and α2-ARs on Ca²⁺ channels in mitral cells. This study has identified the pathway for α₁-AR activation of PKC via PLC, and is the first study to unequivocally establish the isoform of PKC (PKCα) that mediates effects on Ca²⁺ channels in a physiological context. These results also demonstrate a novel role for cross-talk between PKC and G-proteins in regulating Ca²⁺ channels critical for synaptic plasticity.

mGluR_2/3 and α2-AR cause disinhibition of mitral cells in the AOB

Our findings that activation of mGluR_2/3 receptors results in a reduction in GABA-mediated spontaneous and stimulus-evoked IPSCs in mitral cells is consistent with the previous finding that activation of mGluR_2/3 receptors on granule cells results in disinhibition of mitral cell activity, by a presynaptic inhibition of GABA release (Hayashi et al., 1993). The observations of the present study, however, suggest that the mGluR_2/3 pathway might also cause disinhibition by a presynaptic inhibition of glutamate release from mitral cells onto granule cells at the reciprocal synapses. Our functional studies are supported by previous immunohistochemical studies that have found high levels of mGluR_2/3 expression in mitral cells of the AOB (Ohishi et al., 1998; Sahara et al., 2001). The effectiveness of the G-protein activator GTP-γ-S in reducing the Ca²⁺ current and of the G-protein inhibitors GDP-β-S and PTX in increasing it suggest that the mGluR_2/3 effect on the N-and R- type Ca²⁺ channels in mitral cells is mediated through a G-protein. G-protein-mediated modulation of N-type and R-type Ca²⁺ channels is well established (Ikeda and Dunlap, 1999; Meza et al., 2007), and similar inhibitory effects of mGluR_2/3 receptors on Ca²⁺ channels have been reported in cultured hippocampal, cortical (Swartz, 1993), and sympathetic (Ikeda et al., 1995) neurons.

Our experiments revealed that α₂-ARs also act via a G-protein to decrease the N-type and R-type Ca²⁺ currents recorded from mitral cells, in a similar manner to mGluR_2/3 receptors. High levels of α₂-AR expression in the AOB of bullfrog (Bachman et al., 1998) and mice (Skinner et al., 2008) have been shown. This is consistent with the report that NE inhibited Ca²⁺ channels in mitral cells and therefore reduced glutamate release onto granule cells (Kaba and Huang, 2005). It is also in agreement with the previous work showing that the α₂-AR agonist clonidine inhibited Ca²⁺ currents in mitral cells from accessory and main olfactory bulbs of Xenopus laevis (Czesnik et al., 2001). Others have observed α₂-AR-mediated inhibition of Ca²⁺ currents in cultured neurons from the rat main olfactory bulb (Trombley and Shepherd, 1992; Trombley, 1992) and neocortex (Timmons et al., 2004). Several studies have reported a similar disinhibitory effect of NE in the olfactory bulb (Jahr and Nicoll, 1982; Ciombor et al., 1999; Hayar et al., 2001) as well as other parts of the brain such as hippocampus (Doze et al., 1991).

We found that the effects of mGluR_2/3 and α₂-AR activation on mitral cell Ca²⁺ currents showed mutual occlusion, suggesting that mGluR_2/3- and α₂-AR- mediated pathways converge at some point. This convergence resulting in mitral cell disinhibition may provide an explanation for how pheromonal learning can be induced by local infusions of the mGluR_2/3 agonist DCG-IV into the AOB in the absence of mating and the associated NE release (Kaba et al., 1994).

PKCα converts the disinhibition to increased inhibition of mitral cells in the AOB

The current hypothesis explaining mate recognition postulates a long-lasting increase in inhibitory feedback on the sub-population of mitral cells receiving input from the mating male's pheromones, thus preventing the signal from being transmitted centrally to block pregnancy (see Supplementary Fig. S6 for details). Based on our results, it can be deduced that activation of PKC converts the disinhibition of mitral cells to increased inhibition. Mechanistically this could come about by PKC-mediated reversal in mitral cells of the inhibitory effect of both mGluR_2/3 and α₂-AR on N-type Ca²⁺ channels or stimulation of R-type channels by PKC superimposed on N-type channel inhibition. In other systems, enhancement of R-type Ca²⁺ current by PKC, independent of G-protein-mediated inhibition, has been reported (Meza et al., 2007). PKC has been shown to decouple the signaling pathway from neurotransmitter receptors to Ca²⁺ channels in cultured sympathetic neurons (Zhu and Ikeda, 1994) and non-neuronal cells (Zamponi et al., 1997). The efficacy of the specific inhibitor of PKCα in blocking the effect of PKC activation by PMA on Ca²⁺ channels, and the effectiveness of recombinant PKCα but not a closely related isoform PKCγ in stimulating the Ca²⁺ current, suggest a role for the α-isoform of PKC. Moreover, a peptide corresponding to a unique C-terminal region of PKCα (but not an equivalent PKCγ peptide) blocked the PMA effect suggesting a role for the PDZ domain-containing PICK1 (Staudinger et al., 1997), which is thought to localize PKCα in close proximity to Ca²⁺ channels, and requires further investigation.

Classical isoforms of PKC, such as PKCα require both DAG and Ca²⁺ for activation. Our finding that α₁-AR agonist cirazoline increased the mitral cell Ca²⁺current via PLC suggests that the α₁-AR pathway is capable of activating PKCα in the AOB mitral cell. Expression of α₁-AR is considerable in the AOB (Skinner et al., 2008). The α₁-AR activation of PKCα may therefore play a critical role in reversing the inhibitory effect of mGluR_2/3 and α₂-ARs on glutamate release from mitral cells, allowing the inhibitory GABAergic feedback onto mitral cells to increase. We found that NE decreased the magnitude of evoked IPSCs in AOB mitral cells, but when combined with glutamate the IPSC amplitude was increased. This could be explained by the excitatory action of glutamate on mitral cells leading to elevated intracellular Ca²⁺, which when combined with α₁-AR-mediated production of DAG (as well as release of Ca²⁺ from intercellular stores through IP3), would activate PKCα and increase mitral cell Ca²⁺ currents and glutamate release. Increases in glutamate release could in turn cause increased GABA release and increases in IPSC amplitude. Our findings on α₁-AR-mediated effects on mitral cells are in agreement with the recent report that NE increased inhibition of mitral cells in AOB slices via an α₁-AR pathway, although action of NE was attributed only to increased GABA release from granule cells and the mechanism of α₁-AR effect was not investigated (Araneda and Firestein, 2006). The reason these authors did not find a disinhibitory effect of NE on mitral cell activity could be due to the use of mice of C57/Bl6 strain, which have been found to show differences in electrophysiological properties compared to mice of other inbred strains (Nguyen et al., 2000a; Nguyen et al., 2000b). Although it is also possible that disinhibition was insubstantial in their experimental conditions, as the overall effect of adrenergic activity on the membrane potential of mitral cells was reported to be rather small (Araneda and Firestein, 2006).

A hypothetical model of cellular and molecular mechanisms of AOB plasticity underlying pheromonal memory

The results described here represent our initial attempts to identify the signaling events that occur in the AOB in response to glutamatergic and noradrenergic input. Based on our data, we propose that, in the short-term (on a milliseconds to seconds time scale), glutamate acting through mGluR_2/3 and NE acting through α₂-ARs, inhibit N-type/R-type Ca²⁺ channels via inhibitory G-proteins. This results in decreased glutamate release from mitral cells and consequently decreased GABA release from granule cells, which would be greatest in those mitral cells receiving pheromonal input, due to the mGluR_2/3 influence. These short-term events depend mainly on regulation of presynaptic N-type/R-type channels (Catterall and Few, 2008; Xu et al., 2007) and are akin to those underlying presynaptic short-term plasticity in other parts of the brain (Zucker and Regehr, 2002; Lee et al., 2008).

In the intermediate-term, the disinhibition of mitral cells would lead to increased Ca²⁺ influx through possibly through extrasynaptic L-type Ca²⁺ channels (Tippens et al., 2008) as has been shown by others (Miura et al., 1997; Hsu et al., 1999), which combined with α₁-AR-mediated activation of PLC and DAG (Zhong and Minneman, 1999) would lead to the activation of PKCα. Because α₁-AR effect occurs through the generation of second messengers, NE action of α₁-AR would be expected to occur over a slower time course (hundreds of seconds) (Bhalla, 2002), compared to the action of NE on α₂-ARs. Activation of PKCα through α₁-ARs would decouple the inhibitory effect of mGluR_2/3 and α₂-ARs on N-type/R-type Ca²⁺ channels, thus increasing glutamate release from mitral cells, and GABA release from their associated granule cells. The consequence would be an increase in inhibitory gain of the reciprocal synapses of those mitral cells receiving pheromonal input (see Supplementary Fig. S7).

Thus our results provide a novel alternative mechanism for memory induction, in which the association between glutamatergic pheromonal input and mating-induced NE is explained by the role of PKCα in coincidence detection in mitral cells. Moreover, activation of PKCα provides a mechanism for the maintenance of enhanced feedback inhibition onto mitral cells, during the hours following mating, until the protein synthesis-dependent changes that underlie the long-lasting, and NE-independent, maintenance of memory for the mating male have become established. Thus, activation of PKCα through PLC (Rosenzweig et al., 1993; Steidl et al., 2003; Stough et al., 2006) might extend the short-term molecular events of memory induction and bridge them to the long-term changes brought about by gene induction. In support of this idea, we have found that PLC- and PKC- mediated pathway induces gene expression in the AOB (Skinner et al., 2008). The validity of the proposed mechanisms and the role of these signaling molecules ultimately need to be tested using behavioral experiments.

The proposed mechanism has the advantage that the locus of memory induction is in the presynaptic mitral cells that actually receive pheromonal input, rather than their associated postsynaptic granule cells, and therefore potentially allows greater specificity. Persistent presynaptically mediated transmitter release has been observed in other systems such as hippocampal neurons (Arancio et al., 1995). It does not exclude, however, a role for postsynaptic induction at the granule cell side of the reciprocal synapse by conventional mechanisms of synaptic plasticity involving ionotropic glutamatergic transmission (Wilson et al., 1996). Indeed this is supported by the finding of NMDA-dependent long-term potentiation of the glutamatergic synapse from mitral to granule cells (Kaba and Huang, 2005) and increased length of the postsynaptic density on granule cells following mating (Matsuoka et al., 2004). Moreover, it is possible that NE also has an effect on granule cells through α₂ and α₁ ARs leading to increased GABA release and thus enhanced inhibition of the mitral cells. A recent study showed an excitatory effect of α₁ ARs on granule cells (Smith et al., 2009). Additional detailed investigations will be necessary to elucidate the mechanism of plasticity in granule cells.

The coincidence detecting role of PKCα in AOB synaptic plasticity is likely to be broadly relevant to how NE influences synaptic plasticity in many other areas of the brain. The neural module representing feedback inhibition in the AOB exists in the MOB (Hildebrand and Shepherd, 1997), where it is critical for several forms of social learning (Kendrick et al., 1992; Sanchez-Andrade and Kendrick, 2008). This neural module, in slightly different forms, is also seen in other brain regions (Doze et al., 1991). Adrenergic projections are widespread in the brain (Moore and Bloom, 1979). Although the role of β-adrenergic receptors in synaptic plasticity and memory has been extensively investigated (Watabe et al., 2000; Frey et al., 2001; McGaugh and Roozendaal, 2002), much less is known regarding the mechanisms by which α-adrenergic receptors contribute to memory. The role of both α1-ARs and α2-ARs in memory formation is well established (Gibbs and Summers, 2002; McGaugh, 2004) but the signaling pathways mediating the role of these receptors is are not well understood. NE-mediated disinhibition acting through α2-ARs is likely to enhance neuronal sensitivity to diverse sensory stimuli, increasing intracellular Ca²⁺ levels and priming them for activation of PKCα by NE acting at α1-ARs. PKC is known to activate gene expression through transcription factors such as Egr-1. Thus by activating gene expression cascades, PKCα can convert the transient action of NE into long-term synaptic plasticity. The model we have proposed with PKCα as a coincidence detector provides a potential new framework in which to understand strong memory formation mediated by α-adrenergic receptors.

Abbreviations

AOB

Accessory Olfactory Bulb

α₁-AR

α₁-adrenergic receptor

α₂-AR

α₂-adrenergic receptor

BAPTA

1,2-Bis [2-aminophenoxy] ethane-N,N,N',N'-tetraacetic acid

Bis I

Bisindolylmaleimide I

Bis V

Bisindolylmaleimide V

DAG

Diacylglycerol

HBDDE

2, 2', 3, 3' 4, 4'-Hexahydroxy-1, 1'biphenyl6-6'-dimethanol dimethyl ether

mGluR_2/3

mGluR₂/mGluR₃

NE

Norepinephrine

NMDA

N-methyl-D-aspartic acid

PKC

Protein Kinase C

PKCα

protein kinase Cα

PLC

Phospholipase C

TTX

tetrodotoxin

PTX

pertussis toxin

VN

vomeronasal