Heterotrimeric G(o) is one of the most abundant proteins in the brain, yet relatively little is known of its neural functions in vivo. This study demonstrates that G(o) signaling is required for the formation of associative memory. In Drosophila, pertussis toxin (PTX) is a selective inhibitor of G(o) signaling. The postdevelopmental expression of PTX within mushroom body neurons robustly and reversibly inhibits associative learning. The effect of G(o) inhibition is distributed in both γ- and α/β-lobe mushroom body neurons. However, the expression of PTX in neurons adjacent to the mushroom bodies does not affect memory. PTX expression also does not interact genetically with a rutabaga adenylyl cyclase loss-of-function mutation. Thus, G(o) defines a new signaling pathway required in mushroom body neurons for the formation of associative memory (Ferris, 2006).

An associative memory is one that links external stimuli to particular events, such that the stimuli come to predict the events. In the negatively reinforced olfactory associative learning assay of Drosophila, flies are presented with an odor (conditional stimulus paired, CS⁺) paired with an electric shock (unconditioned stimulus, US). The flies are then presented with a second odor (conditioned stimulus unpaired, CS^-). The associative memory is measured as the conditioned avoidance of the CS⁺ in a T-maze. The disruption of the cyclic AMP (cAMP) signaling pathway within Drosophila leads to reduced learning scores. The effect of cAMP disruption has been mapped back to the mushroom body neurons through the targeted expression of a constitutively active G(s)α and by rescuing the rutabaga type I adenylyl cyclase (rut) phenotype with targeted expression of a rut cDNA. It is thought that the cAMP pathway controls the association between the CS⁺ and the US within the mushroom body neurons (Ferris, 2006).

The G(o) heterotrimeric protein is thought to be the most abundant membrane protein in the vertebrate brain and is activated both by numerous G protein–coupled receptors (GPCRs) and by amyloid precursor protein. Although G(o) can participate in diverse signaling pathways, only a few specific in vivo functions have been ascribed to this molecule. In Drosophila, G(o)α47A is the only gene encoding the alpha subunit of G(o), and it is expressed throughout the adult brain. The G(o) protein is much more abundant in the heads of rutabaga (rut) and dunce learning mutants than in the heads of wild-type flies, suggesting a possible role for G(o) in memory formation (Ferris, 2006).

The S1 subunit of PTX from Bordetella pertussis catalyzes the transfer of an ADP-ribose onto the Gα subunit of the vertebrate G(i/o/t) heterotrimeric G proteins, preventing these proteins from binding to activated GPCRs. In Drosophila, PTX is a selective enzymatic inhibitor of G(o) signaling: Drosophila does not have a transducin homolog, and the G(i)α65A protein does not contain the PTX recognition site, whereas G(o)α does; PTX will ADP-ribosylate a single protein in Drosophila, as seen in western blots and after isoelectric focusing; and PTX comigrates with G(o)α and is immunoprecipitated by independent G(o)α-specific antibodies (Ferris, 2006).

To determine if G(o) is a mediator of associative memory, a PtxA transgene was expressed within the mushroom body neurons. The P{UASPTX}16 transgenic line was selected because the basal expression of PTX is low in this line and because PTX can be induced by Gal4, albeit in small amounts. G(o)α47A loss-of-function mutant embryos die during embryogenesis owing to defects in nervous system and mesoderm development. In keeping with this result, it was found that the induction of PTX within the developing mesoderm or nervous system also results in embryonic lethality, indicating that this toxin is functional when expressed early in development (Ferris, 2006).

The role of G(o) in associative memory was examined by inducing PTX expression within the adult mushroom bodies with the P{MBSwitch}12 Gene-Switch driver. The resulting induction abolished the immediate associative memory 3 min after training, which is frequently taken as a measure of learning. Although the PTX-uninduced P{MBSwitch}12/P{UASPTX}16 flies also showed reduced learning, their scores were not significantly lower than the PTX-uninduced P{UASPTX}16/+ control group. The induction of PTX within the mushroom body did not alter naïve sensitivities to either odorants or electric shock, indicating that PTX expressed in the mushroom bodies does not affect the perception of the stimuli (Ferris, 2006).

The severity of the PTX learning phenotype might result from the death of the mushroom body neurons. This hypothesis was tested by examining the integrity of the mushroom bodies after the induction of PTX and by establishing whether the associative learning phenotype was reversible. Because Gene-Switch has slow off-rate kinetics, the Gal80^ts system was used with the P247 Gal4 driver. P247 drives expression in ~700 α/β- and γ-lobe mushroom body neurons. Two independent Gal80^ts transgenes were used to ensure more complete inhibition of Gal4 at 18°C. After inducing PTX for 12 h at 32°C, 3-min memory was almost entirely abolished. Using antibodies to downstream of receptor kinase (DRK), which preferentially mark the mushroom body, it was found that PTX induction did not alter either the gross structure of the mushroom bodies or the expression of DRK. Similar results were found using antibodies to cAMP-dependent protein kinase 1 (DCO). It was also found that the effect of PTX was reversible: although a 2-h induction of PTX within mushroom body neurons produced significant inhibition of 3-min memory, this effect was completely reversed after 6 d. Therefore, the effect of PTX on learning is not due to the death of the mushroom body neurons (Ferris, 2006).

Next, whether the effect of PTX on learning was specific to the mushroom bodies was examined. PTX was induced in the R3 and R4d neurons of the ellipsoid body and separately in the dorsally paired medial (DPM) neurons, which innervate the mushroom bodies. The induction of PTX with the Gal80^ts system in either set of neurons did not affect performance in the learning assay, suggesting that PTX is cell autonomous. Moreover, PTX induction in the DPM neurons had no effect on 60-min memory. The inhibition of neurotransmission in DPM neurons by the shibire^ts transgene completely blocks 60-min memory but has no effect on 3-min memory. Thus, PTX and shibire^ts have different effects in the DPM neurons, indicating that PTX is not a general inhibitor of neurotransmission (Ferris, 2006).

Experiments were conducted to see whether the requirement for G(o) signaling in olfactory associative learning is dispersed throughout the different neurons of the mushroom body lobes, or if the requirement is limited to a subset of these neurons. Several genes have been identified that are preferentially expressed in the different mushroom body lobes, indicating that these lobes have distinct molecular repertoires; however, direct tests for lobe function have yet to provide unequivocal and differentiated roles for the constituent neurons in associative learning. The c772 Gal4 line drives expression in ~800 neurons of the α/β and γ lobes. It was found that a 12-h induction with c772 was sufficient to ablate the associative memory, whereas a 2-h induction was not. There were no differences in the naïve avoidance of odor or shock between the c772/Gal80^ts20; PTX/Gal80^ts2 PTX-induced and PTX-uninduced experimental groups. There were, however, some differences in naïve odor avoidance between the c772/Gal80^ts20; PTX/Gal80^ts2 PTX-induced group and the control genotypes, suggesting that PTX induction in non-mushroom-body neurons by c772 may affect odor perception or discrimination and that the Gal80^ts inhibition may not be complete in these neurons. The differences in odor avoidance may also participate in the severe c772/PTX phenotype, although it is unlikely to have a major effect on learning as naïve avoidance scores were not significantly different in the within-genotype control group. It is likely that differences in expression levels between c772 and P247 account for the different time courses in the inhibition of learning by PTX between these two lines. The 12-h induction of PTX in the γ-lobe neurons marked by 1471 caused a substantial, but not complete, loss of 3-min memory, as did the expression of PTX in the α/β-lobe neurons marked by c739. Thus, G(o) signaling is required for 3-min memory in both the γ and α/β neurons of the mushroom body as defined by the 1471 and c739 drivers, respectively. In contrast, PTX driven by the α/β-lobe driver 17d did not have an observable effect on 3-min memory. The mushroom body neurons defined by 17d are most likely the core neurons of the α/β lobe, which may be functionally distinct from the other neurons of the α/β lobe, since they are insensitive to the effects of PTX in associative memory and have no effect on the rescue of the rut learning phenotype. The fact that associative memory formation was affected by PTX induction in the α/β- and γ-lobe neurons, but not in the putative α/β core neurons, defines a new requirement for G(o) signaling in these lobes for learning and memory and should further help dissect the memory process in these neurons (Ferris, 2006).

Next it was considered whether the G(o) pathway interacts genetically with the rut adenylyl cyclase in associative memory formation. The persistent activation of vertebrate G(o) may initially lead to the short-term inhibition of type I adenylyl cyclase, followed by the increased responsiveness of this enzyme to G(s)α stimulation, known as heterologous sensitization or supersensitization. Thus, PTX may be interfering with the down regulation of rut activity by G(o), resulting in neurons with too much adenylyl cyclase activity. Alternatively, PTX may inhibit the heterologous sensitization of rut by G(o), leaving the neurons with too little cAMP after G(s) activation. The former hypothesis predicts that a reduction in rut activity may partially suppress the PTX phenotype, whereas the latter suggests that the reduction in rut may act synergistically with PTX. These predictions were tested by looking for a genetic interaction between a mild induction of PTX in the subset of mushroom body neurons defined by P247 and a single copy of rut²⁰⁸⁰. This rut mutation demonstrates a semidominant haploinsufficiency, indicating that learning is extremely sensitive to the activity levels of this enzyme. It was found that the performance of the rut²⁰⁸⁰/+; Gal80^ts20/+; PTX/P247, Gal80^ts2 PTX-induced flies was reduced, but not significantly, as compared to that of the Gal80^ts20/+; PTX/P247, Gal80^ts2 flies. This result suggests an additive interaction between PTX and the rut²⁰⁸⁰ heterozygote, but there was evidently neither suppression nor a synergistic relationship between PTX and one copy of rut²⁰⁸⁰. The independence of G(o) function during learning from rut was further assessed in rut homozygotes. The performance of the rut²⁰⁸⁰; Gal80^ts20/+; PTX/P247, Gal80^ts2 PTX-induced flies was significantly worse than that of either the PTX-induced flies or the rut homozygous flies. Thus, G(o) signaling has functions in olfactory learning and memory within the mushroom body neurons defined by P247 that are independent of rut (Ferris, 2006).

This study has shown, through the postdevelopmental induction of PTX expression within mushroom bodies, that activation of G(o) is required during the physiological events, which lead to associative memory formation. The severity of the learning phenotype in PTX-induced flies coupled with the lack of genetic interaction with rut²⁰⁸⁰ strongly suggests that the function of G(o) in associative learning and memory is largely independent of the cAMP pathway. Additional members of this new associative learning pathway are currently unknown. One possibility is that, similar to the role of the G(o) in the vertebrate dorsal root ganglia, the Drosophila G(o) may participate in learning through the inhibition of voltage-gated Ca²⁺ channels (VGCCs). These Ca²⁺ channels are thought to be activated by the odor-induced depolarization of the mushroom body neurons, leading to the release of synaptic vesicles and the CS pathway activation of rut. It is plausible that the negative regulation of the VGCCs may be necessary to restrict the number of activated synapses during learning. Nevertheless, it is now clear that the in vivo functions of G(o) include the formation of associative memories in Drosophila (Ferris, 2006).