BIOLOGICAL OVERVIEW

Learning in Drosophila involves the cyclic AMP second messenger system. Stimulation of adenylyl cyclase (Rutabaga) by neural excitation results in the production of cyclic AMP, which in turn activates Protein kinase A. PKA acts on downstream targets to bring about cellular changes that are required for learning and memory.

G proteins activate and inhibit elements in the learning pathway. Heterotrimeric G proteins transduce signals from cell surface 7 pass transmembrane serpentine receptors to downstream effectors, such as ion channels, cytoskeletal elements and other signal transduction proteins. Classification of G proteins is based historically on the functional interaction of G protein alpha subunits with specific effector proteins. For example, transducins are responsible for the activation of the cGMP phosphodiesterase in the retina. Gs alpha stimulates adenylate cyclase, while Gi alpha acts as an inhibitor.

How does the excitation of neurons stimulate the cyclic AMP second messenger system? In eukaryotic cells, serpentine receptors, excited by neurotransmitters are coupled to the activation of adenylyl cyclase by the heterotrimeric GTP binding protein complex Gs. In this pathway, the initial binding of extracellular ligands to these receptors results in the activation of the Gs complex (the heterotrimeric G protein complex) by promoting the exchange of GDP for GTP on the alpha subunit, and the dissociation of Gs alpha from beta/gamma. The GTP-bound, activated Gs alpha then mediates the activation of adenylyl cyclases, resulting in the elevation of the intracellular levels of the second messenger, cAMP. Termination of the signal occurs when GTP bound by the alpha subunit is hydrolyzed to GDP by an enzymatic activity intrinsic to the alpha subunit (Neer, 1995). For more information on the consequences of neural excitation see Cyclic AMP Second Messenger System - The Learning Pathway.

Evidence in Drosophila points to the involvement of Gs alpha in the process of associative learning. However, before examining the involvement of Gs alpha in learning, an example will be give of a developmental pathway involving Gs alpha that does not involve the cyclic AMP second messenger system.

Gs alpha can be altered by mutation to become constitutively activated. Mutation of Gs alpha in the putative S box of its guanine nucleotide-binding domain (Quan, 1991), results in Gs alpha*, which has a greatly reduced hydrolytic rate for GTP. This results in receptor independent activation of the alpha subunit and consequently in constitutive activation of downstream components coupled to G protein activation.

When Gs alpha* is expressed in flies under control of various promoters, a variety of phenotypes result, including alteration of wing morphology, smaller than normal adults and lethality. One consistent effect of Gs alpha* expressed during late pupal periods is formation of wing blisters. Phenotypes arising from activation of Gs alpha pathways would be expected to depend on the activation of PKA, given the traditional scheme developed in mammals. Consistent with this prediction, many studies have demonstrated just such a critical reliance on PKA in the response of cultured cells to activation of this pathway, through the expression of mammalian Gs alpha*. To test whether this pathway is acting in wing epithelial cells to generate blistering, a dominant-negative form of the regulatory subunit of PKA was employed. PKA functions in Drosophila wings to repress the expression of signaling molecules (for example, Decapentaplegic) that mediate subsequent growth and pattern formation in the developing wing. Thus, in the absence of PKA, inappropriate expression of dpp leads to anterior wing duplication. Expression of dominant negative PKA results, as expected, in wing duplications. By constructing flies that carry both Gs alpha* and dominant negative PKA (dnPKA), activation of Gs alpha pathways and inhibition of PKA occurs simultaneously within cells of the wing epithelium. Coexpression of both altered proteins in wing cells results in a superimposition of the Gs alpha phenotype on the dnPKA phenotype. Thus, each phenotype occurs independently when these proteins are coexpressed, indicating that PKA activity is not required to generate the blistering observed on activation of the Gs alpha pathway in wing epithelial cells (Wolfgang, 1996).

The independence of PKA expression from Gs alpha stimulation is not completely without precedence. In mammalian epithelial cells, transport of influenza virus hemagglutinin protein from the apical surface is retarded by G protein specific reagents, such as drugs or toxins. Treatment of cells with reagents known to influence the Gs class of G proteins specifically affect the apical pathway. Also, antibodies against the N-terminal domain of the alpha-subunit of Gs inhibits the transport of hemaglutinin from the trans-Golgi network to the apical surface, but not between the endoplasmic reticulum and the Golgi complex. Addition of cAMP to cells has no effect on the transport of hemaglutinin to the apical surface. Thus the effects of Gs on apical transport must be mediated by another downstream effector (Pimplikar, 1993). It seems, therefore, that Gs can act independently of the cyclic AMP second messenger pathway both in Drosophila wing blistering and in mammalian vesicular transport. For more information on Drosophila wing blistering, see Myospheroid and Serum response factor.

To test whether Gs alpha signaling is involved in learning in Drosophila, Gs alpha* was targeted either to the mushroom bodies (MB) or alternatively, to the central complex (CC) of the fly brain. To examine associative learning in these flies, a Pavlovian olfactory conditioning assay was used (see Dunce for a more complete description of this procedure). Briefly, flies are trained by exposure to electroshock paired with one odor (octanol or methylcyclohexanol) and subsequently exposured to a second odor without electroshock. Immediately after training, learning is measured by forcing flies to choose between the two odors used during training. No preference between odors results in a performance index of zero (no learning), as is the case for naive flies. Avoidance of the odor previously paired with electroshock, however, yields a performance index greater than O.

When Gs alpha* is expressed in mushroom bodies, learning is completely abolished. As a control, wild type Gs alpha was expressed in mushroom bodies, and no effect on learning was detected. In the learning-impaired lines, olfactory responses to the odors is normal, demonstrating that Gs alpha* does not affect naive sensorimotor response to electroshock or odors. When Gs salpha* is expressed in the ellipsoid body or fan-shaped body of the CC, learning is unaffected. Pan-neural expression of Gs alpha* during development produces neither lethality nor overt behavioral phenotypes, suggesting that perturbation of Gs signaling does not significantly affect basic neuronal function. Gross morphology appears normal when Gs alpha* is expressed in the MBs. Thus the abolition of learning does not appear to result from maldevelopment of underlying structures. Therefore Gs alpha* specifically impairs learning, implicating wild type Gs alpha in stimulation of the learning pathway.

Null alleles of dunce and rutabaga, two genes involved in the cyclic AMP second messenger pathways produce only a partial impairment in learning and not the complete impairment observed for Gs alpha* expression. Thus, disruption of all adenylyl cyclase regulation by Gs alpha* expression seems to have more drastic effects on signaling than removal of one form of adenylyl cyclase (as in rutabaga) or cyclic nucleotide phosphodiesterase (as in dunce). Alternatively, Gs alpha could exert signaling effects other than through the cAMP pathway, as it does in the blistering phenotype described above (Connolly, 1996).

Distinct memory traces for two visual features in the Drosophila brain

The fruit fly can discriminate and remember visual landmarks. It analyses selected parts of its visual environment according to a small number of pattern parameters such as size, colour or contour orientation, and stores particular parameter values. Like humans, flies recognize patterns independently of the retinal position during acquisition of the pattern (translation invariance). The central-most part of the fly brain, the fan-shaped body, contains parts of a network mediating visual pattern recognition. Short-term memory traces have been identified of two pattern parameters -- elevation in the panorama and contour orientation. These can be localized to two groups of neurons extending branches as parallel, horizontal strata in the fan-shaped body. The central location of this memory store is well suited to mediate translational invariance (Liu, 2006).

A fly tethered to a torque meter, with its head (and hence its eyes) fixed in space, can control its orientation with respect to the artificial scenery in a flight simulator. In this set up, the fly is conditioned to avoid certain flight directions relative to virtual landmarks and recognizes these visual patterns for up to at least 48 h. Visual pattern recognition in Drosophila has been studied in some detail. Flies store values of at least five pattern parameters: size, colour, elevation in the panorama, vertical compactness, and contour orientation. Moreover, they memorize spatial relations between parameter values. The neuronal substrate underlying visual pattern recognition is little understood in any organism (Liu, 2006).

In Drosophila, memory traces can be localized to groups of neurons in the brain. Using the enhancer GAL4/UAS expression system, short-term memory traces of aversive and appetitive olfactory conditioning have been assigned to output synapses of subsets of intrinsic neurons of the mushroom bodies (MBs). The Rutabaga protein -- a type 1 adenylyl cyclase that is regulated by Ca²⁺/Calmodulin and G protein, and is considered a putative convergence site of the unconditioned and conditioned stimulus in olfactory associative learning, selectively restores olfactory learning if expressed in these cells in an otherwise rutabaga (rut)-mutant animal. Moreover, expressing a mutated constitutively activating Galpha_s protein (Galpha_s*) in the MBs interferes with olfactory learning. Blocking the output from these neurons during memory retrieval has the same effect, while blocking it during acquisition has no effect. Interestingly, memory traces for other learning tasks seem to reside in other parts of the brain: for remembering its location in a dark space, the fly seems to rely on a rut-dependent memory trace (Zars, 2000) in neurons of the median bundle and/or the ventral ganglion (Liu, 2006).

The present study localizes short-term memory traces for visual pattern recognition to the fan-shaped body (FB), the largest component of the central complex (CX; also called the central body in other species). The CX is a hallmark of the arthropod brain. It has been characterized functionally as a pre-motor centre with prominent, but not exclusive, visual input. In the locust, large-field neurons sensitive to the e-vector orientation of polarized light have been described in the CX. Because of its repetitive structure and the precisely ordered overlay of fiber projections from the two hemispheres in the FB, neighbourhood relations of visual space might still be partially preserved at this level (retinotopy). Using the genetic approach, this study shows that a small group of characteristic stratified neurons in the FB house a memory trace for the pattern parameter 'elevation', and a different set of neurons forming a parallel stratum contain a memory trace for 'contour orientation' (Liu, 2006).

Of ten mutants with structural abnormalities in the CX, all were impaired in visual pattern recognition. They were able to fly straight and to avoid heat, yet they failed to remember the patterns. Did they really lack the memory or had they lost their ability to discriminate between patterns? Fortunately, individual flies often display spontaneous preferences for one of the patterns. In three lines, these preferences were consistent enough to reveal intact pattern discrimination, suggesting that aberrant circuitry of the central complex can affect visual learning independent of visual pattern discrimination (Liu, 2006).

Since the developmental and structural defects in these mutants are not well characterized, the GAL4/UAS system was used to acutely interfere with CX function. A GAL4 driver line (c205-GAL4) was used with expression in parts of the CX and, the gene for tetanus toxin light chain (CntE) was used as the effector. CntE blocks neurons by cleaving neuronal Synaptobrevin, a protein controlling transmitter release. For temporal control, the temperature-sensitive GAL4-specific silencer GAL80 was added under the control of a tubulin promoter (tub-GAL80^ts). Flies (UAS-CntE/+; tub-GAL80^ts/c205-GAL4) were raised at 19 °C, and were transferred for 14 h to the restrictive temperature (30 °C) just before the behavioural experiment to induce GAL4-driven toxin expression. Flies kept at the low temperature showed normal memory scores, while after inactivation of GAL80^ts no pattern memory was observed. Again, flight control and heat avoidance were normal, and Fourier analysis confirmed that flies at the high temperature had retained their ability to tell the patterns apart. As with the structural mutants, interrupting the circuitry of the CX by tetanus toxin expression seemed to specifically interfere with visual pattern memory. In addition, the use of tub-GAL80^ts excluded the possibility that toxin expression in unknown tissues during development might cause the memory impairment in the adult. These results do not, as yet, address the question of memory localization (Liu, 2006).

Visual pattern memory in the flight simulator requires an intact rut gene. Mutant rut flies (rut²⁰⁸⁰) showed normal visual flight control, heat avoidance and pattern discrimination. To confirm that the defect was indeed due to the mutation in the rut gene rather than an unidentified second-site mutation, rut was rescued by the expression of the wild-type rut cDNA (UAS-rut⁺) using the pan-neuronally expressing driver line elav-GAL4. Indeed, flies of the genotype rut²⁰⁸⁰/Y;elav-GAL4/UAS-rut⁺ have normal memory (Liu, 2006).

Visual pattern memory in the flight simulator has been shown to depend upon at least two kinds of behavioural plasticity: (1) an associative classical (pavlovian) memory trace is formed linking a particular set of values of pattern parameters to heat; (2) the fly's control of the panorama operantly facilitates the formation of this memory trace (Brembs, 2000). Either of the two processes might depend upon the Rut cyclase (Liu, 2006).

To address this issue, rut mutant flies were tested in a purely classical variant of the learning paradigm. During training, panorama motion was uncoupled from the fly's yaw torque and the panorama was slowly rotated around the fly. Heat was made contingent with the appearance of the 'punished' pattern in the frontal quadrant of the fly's visual field. All other parameters were kept as described. For testing memory, panorama motion was coupled again to yaw torque and the fly's pattern preference was recorded as usual. Even in the absence of operant facilitation, visual pattern memory required the intact rut gene. Therefore, the rut-dependent memory trace investigated in this study represents the association of a property of a visual pattern with the reinforcer (Liu, 2006).

As a first step in localizing the memory trace, it was asked in which neurons of the rut mutant expression of the wild-type rut gene would be sufficient to restore learning. To this end, a total of 27 driver lines expressing GAL4 in different neuropil regions of the brain was used to drive the UAS-rut⁺ effector gene in the rut mutant background. The parameter 'elevation' was measured. With seven of the driver lines, pattern memory was restored (104y, 121y, 154y, 210y, c5, c205 and c271) (Liu, 2006).

Comparison of the expression patterns of the 27 lines allowed the putative site of the memory trace to be narrowed down to a small group of neurons in the brain. The seven rescuing lines all showed transgene expression in a stratum in the upper part of the FB. In three of them staining is rather selective. It comprises, in addition to the FB, only a layer in the medulla, several cell clusters in the subesophageal ganglion and a few other scattered neurons (Liu, 2006).

Evidently, rut⁺ expression in the MBs is neither necessary (104y, c5, c205, 154y) nor sufficient for rescue. This result is in line with the earlier observation that elimination of more than 90% of the MBs by hydroxyurea treatment of first-instar larvae has no deleterious effect on visual pattern memory. The MBs were ablated in one group of rescue flies (rut²⁰⁸⁰/Y;UAS-rut⁺/ +;c271/+). They showed full visual pattern memory (Liu, 2006).

Although GAL4 expression in the optic lobes is prominent in all seven rescuing lines, it occurs in distinctly different layers that do not overlap. For instance, in 104y expression is restricted to layer 2, whereas in 210y it is found only in the serpentine layer (layer 7). A similar situation is found for the s ganglion, although there the staining patterns are more difficult to evaluate. Finally, expression in the ellipsoid body is again not necessary (104y, c5, c205, 154y) or sufficient (c232, 78y, 7y, and so on) for rescue. Thus, the expression patterns favour the conclusion that the neurons of the upper stratum of the FB might be the site of the memory trace for the parameter 'elevation' in visual pattern memory (Liu, 2006).

Neurons in this stratum, labelled in all seven rescuing lines, have a very characteristic shape. Their cell bodies are located just lateral to the calyces. Their neurites run slightly upward in an antero-medial direction, forming an upward-directed tufted arborization just behind the alpha/alpha'-lobe of the MB. From there, the fiber turns sharply down and backward towards the midline just in front of the FB. Finally, it turns horizontally backward, spreading as a sharp stratum through all of the FB across the midline. These neurons have been described in Golgi preparations. They belong to a larger group of tangential FB neurons called F neurons. Besides the stratum in FB, most of them have an arborization in a particular part of the unstructured neuropil. The layer stained in 104y, and the other six rescuing lines, is tentatively classify as layer 5 (from bottom upward), and hence provisionally the neurons are called F5, although, without further markers, it is difficult to reliably number the layers. In summary, expression of Rut cyclase in F5 neurons rescues the rut-dependent memory defect for pattern elevation, whereas no rescue effect is observed in any of 20 strains without expression of Rut cyclase in F5 neurons (though Rut cyclase was expressed in other regions of the brain). Hence, a rut-dependent memory trace for pattern elevation may reside in F5 neurons (Liu, 2006).

This finding does not exclude the possibility that memory is redundant, and that other rut-dependent memory traces for pattern elevation might be found elsewhere. Therefore, it was asked whether plasticity in the F5 neurons is necessary for visual pattern memory. The Rut cyclase is regulated by G protein signaling, and olfactory learning/memory can be blocked by a constitutively active form of the Galpha_s protein subunit (Galpha_s*). The Galpha_s* mutant protein was expressed in the FB using the driver line c205, and the flies were tested for their memory of 'elevation'. Memory was fully suppressed. Since in olfactory learning, overexpression of the wild-type protein does not interfere with learning, these results support the hypothesis that continuous upregulation of Rut cyclase in the F5-neurons interferes with visual short-term memory, implying that F5 neurons are the only site of a rut-dependent memory trace for pattern elevation (Liu, 2006).

The patterns used in the experiments so far exclusively addressed the parameter 'elevation' (upright and inverted Ts or horizontal bars at different elevations). It was of interest to discover whether the mutant defect in rut and the Rut rescue in the F5 neurons affects only this parameter, or whether it applies to other pattern parameters as well. Therefore, the study looked at to two further parameters: 'size' and 'contour orientation'. Three driver lines -- c205, NP6510 and NP2320 -- were chosen showing different expression patterns in the FB. In the line NP6510, as in c205, a group of F neurons is marked. They are putatively classified as F1, since their horizontal stratum lies near the lower margin of the FB. Their cell bodies form a cluster in the dorso-frontal cellular cortex above the antennal lobes. Like the F5 neurons, they have large arborizations in the dorsal unstructured neuropil. The line NP2320 expresses the driver in columnar neurons running perpendicular to the strata of F neurons, with their cell bodies scattered singly or in small groups between the calyces. Since they seem to have no arborizations outside the FB, they are tentatively classified as pontine neurons (Liu, 2006).

Initially, it was shown that pattern memory requires the rut gene for each of the three parameters. Next, the Rut rescue flies were studied (for example, rut²⁰⁸⁰/Y;c205/UAS-rut⁺). In the line c205, memory was restored only for 'elevation', not for 'size' or 'contour orientation'. Correspondingly, the memory impairment by expression of dominant-negative Galpha_s* in this driver line should be specific for 'elevation', as is indeed the case. With the driver line NP6510, memory was not restored for either 'elevation' or for 'size,' but memory was restored for 'contour orientation'. The third driver line, NP2320, labelling columnar neurons of the FB, did not restore the memory for any of the three pattern parameters. Among the 27 GAL4 lines, a second was found with a very similar expression pattern as NP6510 (NP6561). The P-element insertions in the two lines are only 124 nucleotides apart from each other. Like NP6510, NP6561 restores the memory for 'contour orientation' but not for 'size' or 'elevation'. These results strongly suggest that memory traces for distinct visual pattern parameters are located in different parts of the FB, and that, in addition to the memory trace in F5 neurons, a memory trace for the parameter 'contour orientation' is located in F1 neurons (Liu, 2006).

A pertinent question in rescue experiments is whether the rescue is due to the provision of an acute function in the adult or to the avoidance of a developmental defect. Therefore, the tub-GAL80^ts transposon was added to the system. The driver lines c205 and NP6510 were chosen. Groups of adult males (for example, rut²⁰⁸⁰/Y;+/tub-GAL80^ts;NP6510/UAS-rut⁺), raised at 19°C, were kept as adults for 14 h at 19°C or 30°C. Afterwards, pattern memory for the corresponding pattern parameter was tested. In both cases, flies that had been kept at 30°C showed normal memory, indicating that Rut cyclase induced just a few hours before the experiment had restored an immediate neuronal function rather than preventing a developmental defect. This conclusion was further supported by the finding that Galpha_s* expression in the adult (using tub-GAL80^ts) was sufficient to disrupt memory (Liu, 2006).

Several conclusions can be drawn from the above results. Memory traces in Drosophila are associated with specific neuronal structures: odor memories with the MBs, visual memories with the CX, and place memory (tentatively) with the median bundle. Memory traces are not stored in a common all-purpose memory centre. Even within the visual domain, memories for distinct pattern parameters are localized within distinct structures: a rut-dependent short-term memory trace for the pattern parameter 'elevation' to F5 neurons, and a corresponding memory trace for 'contour orientation' to F1 neurons. Moreover, if the constitutively activating Galpha_s* protein indeed interferes with the regulation of Rut cyclase, it follows that the brain contains no other redundant rut-dependent memory traces for these pattern parameters. The Rut-mediated plasticity is necessary and sufficient, at least in F5 neurons. As in the earlier examples, the memory traces are confined to relatively small numbers of neurons. At least in flies, and probably in insects in general, memory traces appear to be part of the circuitry serving the respective behaviour (Liu, 2006).

This study provides a first glimpse of the circuitry within a neural system for visual pattern recognition. Though the picture is far from complete, it invites (and may guide) speculation. The FB is a fiber matrix of layers, sectors and shells. The F1- and F5-neurons form two sharp parallel horizontal strata in this matrix. If the width of the FB represents the azimuth of visual space as has been proposed, the horizontal strata of the F neurons would be well suited to mediate translation invariance. In any case, it is satisfying to find a translation invariant memory trace in the CX where visual information from both brain hemispheres converges. These first components of the circuitry may encourage modelling efforts for pattern recognition in small visual systems (Liu, 2006).

GENE STRUCTURE

The exon-intron structure of the Drosophila Gs alpha shows substantial similarity to that of the human gene for Gs alpha. Alternative splicing of intron 7 in Drosophila, involving either the use of an unusual TG or consensus AG3' splice site, results in transcripts which code for either a long or short form of Gs alpha. These subunits differ by inclusion or deletion of three amino acids and substitution of a Ser for a Gly. The two forms of Drosophila Gs alpha differ in a region where no variation in the primary sequence of vertebrate Gs alpha subunits has been observed. Additional Gs alpha transcript heterogeneity refects the use of multiple polyadenylation sites (Quan, 1990).

cDNA clone length - 1345

Bases in 5' UTR - 303

Exons - 9

Bases in 3' UTR - 189

PROTEIN STRUCTURE

Amino Acids - 385

Structural Domains

The Drosophila G protein salpha 60A is 71% homologous to the long form of bovine Gs alpha. The level of homology to Gi alpha and Go alpha and transducin is lower but still significant (41-44%) (Quan, 1989). The similarity is highest in the four regions of homology that have been identified in the G alpha subunits, the ras oncogene proteins, and bacterial elongation factor Tu. These highly conserved regions (A, C, E and G) are thought to be responsible for guanine nucleotide binding and hydrolysis. In mammals, the alignment of the deduced amino acid sequence of the various G alpha polypeptides has shown that Gs alpha is the most divergent member of the G alpha family. Vertebrate Gs alpha is structurally heterogeneous, existing as at least two species with apparent molecular weights of 45,000 and 52,000 in SDS polyacrylamide gels. These proteins are formed by alternative splicing. The Drosophila proteins (both long and short forms) correspond most closely to the smaller vertebrate form (Quan, 1989).

The crystal structure of Gsalpha, the heterotrimeric G protein alpha subunit that stimulates adenylyl cyclase, was determined at 2.5 A in a complex with guanosine 5'-O-(3-thiotriphosphate) (GTPgammaS). Gsalpha is the prototypic member of a family of GTP-binding proteins that regulates the activities of effectors in a hormone-dependent manner. Comparison of the structure of Gsalpha.GTPgammaS with that of Gialpha.GTPgammaS suggests that their effector specificity is dictated primarily by the shape of the binding surface formed by the switch II helix and the alpha3-beta5 loop, despite the high sequence homology of these elements. In contrast, sequence divergence explains the inability of regulators of G protein signaling to stimulate the GTPase activity of Gsalpha. The betagamma binding surface of Gsalpha is largely conserved in sequence and structure to that of Gialpha, whereas differences in the surface formed by the carboxyl-terminal helix and the alpha4-beta6 loop may mediate receptor specificity (Sunahara, 1997).

date revised: 20 August 2006
 

Home page: The Interactive Fly © 1995, 1996 Thomas B. Brody, Ph.D.

The Interactive Fly resides on the
Society for Developmental Biology's Web server.