Sequential use of mushroom body neuron subsets during Drosophila odor memory processing

Abstract

Drosophila mushroom bodies (MB) are bilaterally symmetric multi-lobed brain structures required for olfactory memory. Previous studies suggested that neurotransmission from MB neurons is only required for memory retrieval. Our unexpected observation that Dorsal Paired Medial (DPM) neurons, which project only to MB neurons, are required during memory storage but not for acquisition or retrieval, led us to revisit the role of MB neurons in memory processing. We show that neurotransmission from the α′β′ subset of MB neurons is required to acquire and stabilize aversive and appetitive odor memory but is dispensable during memory retrieval. In contrast neurotransmission from MB αβ neurons is only required for memory retrieval. These data suggest a dynamic requirement for the different subsets of MB neurons in memory and are consistent with the notion that recurrent activity in a MB α′β′ neuron-DPM neuron loop is required to stabilize memories formed in the MB αβ neurons.

Introduction

The ephemeral nature of memory remains one of the greatest mysteries of modern biological research. Studies over the last century have determined that memory exists in time-dependent phases and is converted from a labile to a stable state after training by a process termed consolidation (Dudai, 2004). In mammals memory consolidation involves both parallel and sequential use of distinct brain regions. Consolidation initially requires the neural circuitry of the hippocampus and cortex but once the memory is consolidated, the requirement of the hippocampus is diminished. Hippocampal damage impairs the consolidation of new memories but leaves old memories intact suggesting that consolidated memories permanently reside in the cortex (Scoville and Milner, 1957).

In an animal model, learning and memory can be reduced to the novel association of two stimuli. This straightforward assay can be utilized in invertebrate models where the simpler anatomy and reduced complexity of the brain provide significant technical advantages for analysis of neural circuits involved in memory. Memory in invertebrates such as the sea slug Aplysia californica (Hawkins et al., 2006; Glanzman, 2006), the pond snail Lymnaea stagnalis (Lukowiak et al., 2003), the honeybee Apis mellifera (Menzel et al., 2006), and the fruit fly Drosophila melanogaster (Davis, 2005) share many of the same properties as their mammalian counterparts. Furthermore, studies in these “simple” systems suggest that the underlying molecular mechanisms have been conserved during evolution.

The most frequently studied learning and memory paradigm in Drosophila involves a pairing of one minute of odor (the conditioned stimulus, CS) with 12 electric foot shocks (the unconditioned stimulus, US) producing an aversive odor memory that lasts for several hours (Davis, 2005). This memory can be dissected into three distinct phases – short-term memory (STM), middle-term memory (MTM) and anesthesia resistant memory (ARM) (Folkers et al., 1993; Tully et al., 1994). Similar to vertebrate memory, earlier phases -STM and MTM-are labile and can be disrupted with a cold-shock anesthesia while the later phase, ARM, is anesthesia resistant and therefore represents a form of consolidated memory (Folkers et al., 1993; Tully et al., 1994). During this time frame (0~6 hours) memory is believed to be protein-synthesis independent (Tully et al., 1994). Although less studied, appetitive olfactory memory, formed by pairing sucrose (instead of shock) with an odor, shares similar temporal properties to aversive olfactory memory (Tempel et al., 1983; Schwaerzel et al., 2003; Keene et al., 2006).

A dependence for specific brain regions to process memory also appears to be generally conserved across taxa. The most heavily studied components of the fly olfactory memory circuit are the mushroom bodies (MBs), bilaterally symmetrical structures in the brain comprised of about 5000 neurons in total (Heisenberg, 2003). In addition to being critical for olfactory memory (Heisenberg et al., 1985; de Belle and Heisenberg, 1994), the MBs have also been implicated in other complex adaptive behaviors, including visual context generalization (Liu et al., 1999) and choice behavior (Tang and Guo, 2001), courtship conditioning (McBride et al., 1999) and sleep (Joiner et al., 2006; Pitman et al., 2006). The 2500 intrinsic neurons in each MB can be subdivided into at least three morphological subsets -αβ, α′β′ and γ- based on the bundling of their axonal projections in the region of the MBs called the lobes (Crittenden et al., 1998). Each MB neuron that contributes to the αβ subdivision bifurcates and sends one axon branch to the α lobe and one to the β lobe. Similarly, each neuron in the α′β′ lobe bifurcates and sends one axon branch to the α′ lobe and one to the β′ lobe. The significance of this morphological repertoire is poorly understood and as a result, conceptual models of MB function in olfactory memory do not clearly differentiate between αβ, α′β′ and γ neurons.

Experiments conducted with learning and memory defective mutant flies have led to the prevailing model where MB neurons associate the odor CS with 4 the shock, or sugar US, using potential coincidence detecting molecules like RUT adenylyl cyclase (Zars et al., 2000; McGuire et al., 2003; Mao et al., 2004) and store the aversive or appetitive associations within the specific neurons that are activated by a particular odor (Heisenberg, 2003; Davis, 2005). This model is supported by the demonstration that transient blockade of MB synaptic transmission during acquisition, storage, and/or retrieval indicates a requirement for MB output only during memory retrieval (Dubnau et al., 2001; McGuire et al., 2001; Schwaerzel et al., 2002; Davis, 2005). The apparent dispensability of MB neuron output for memory formation implies that memory could be represented at MB output synapses or synapses that are upstream of MB output synapses. Although there is ample evidence for a role of upstream antennal lobe circuits in memory in other insects (Stopfer and Laurent, 1999; Daly et al., 2004), particularly in the honeybee (Hammer and Menzel, 1998; Faber et al., 1999), only one live-imaging study in Drosophila has reported a short-term change in antennal lobe neural activity after aversive olfactory training (Yu et al., 2004).

The analysis of Dorsal Paired Medial (DPM) neurons suggests a more complex and dynamic process underlies olfactory memory. DPM neurons express the putative neuropeptide precursor encoded by the amn gene and their processes intermingle with the MB lobes indicative of a role in modulating the MB neuron ensemble (Waddell et al., 2000). Behavioral analyses have demonstrated that output from DPM neurons is critical after training for memory stability and is not required during acquisition or recall (Keene et al., 2004; Yu et al., 2005; Keene et al., 2006). Given that DPM neurons appear to be part of the MB circuit, one might expect that MB neurons would show a similar temporal requirement to DPM neurons during memory consolidation, rather than a role only in retrieval. However, the current literature does not address this discrepancy. DPM neuron projections to MB α′β′ lobe neurons appear to be sufficient to stabilize aversive and appetitive odor memory suggesting that a DPM neuron to MB α′β′ neuron connection could be critical for memory consolidation (Keene et al., 2006). These findings led us to investigate the role of MB α′β′ neuron output in memory processing. Here we show that stable memory requires the sequential involvement of different MB neuron subsets. α′β′ neurons are required during and after training to acquire and stabilize olfactory memory whereas, consistent with a previous report (McGuire et al., 2001), αβ neuron output is only required to retrieve the memory. Similar to mammals, memory processing in flies likely involves parallel and sequential use of distinct neural circuits.

Results

The c305a{GAL4} and c320{GAL4} enhancer-trap lines express in MB α′β′ neurons

Previous studies have reported that MB neuron output is dispensable during memory acquisition and storage (Dubnau et al., 2001; McGuire et al., 2001; Schwaerzel et al., 2002; Davis, 2005), while we have identified a requirement for DPM neuron activity during memory storage (Keene et al., 2004; Keene et al., 2006). One possible explanation for this observed difference between MB and DPM neuron temporal requirements is that the previously employed {GAL4} drivers provided incomplete coverage of all MB neuron subtypes. We therefore sought to examine in more detail where the previously employed MB{GAL4} drivers MB247 (Schwaerzel et al., 2002; Davis, 2005), c309, c747 (Dubnau et al., 2001; Schwaerzel et al., 2002) and c739 (McGuire et al., 2001), expressed in the MBs. We used the MB247, c309, c747 and c739 {GAL4} drivers to express a membrane-tethered GFP (uas-CD8::GFP). Projections of confocal stacks through the MBs of these GAL4 drivers are shown in Figure 1A1-D1. Gross inspection of the patterns revealed strong expression in the αβ and γ lobes and significantly less in the α′β′ lobe. This marked difference is most easily observed in the intertwined vertical α and α′ lobes (Figure 1A1-F1 and S1). MB247 (Figure 1A1), c309 (Figure 1B1) and c747 (Figure 1C1) strongly express in αβ and γ neurons, and in a few α′β′ neurons. c739, as previously described (McGuire et al., 2001), is the most restricted, expressing only in the MB αβ neurons; no GFP expression is visible in the α′ lobes (arrowheads, Figure 1D1).

Figure 1. MB α′β′ neuron expression in GAL4 driver lines. A1-F1.

Panels A1 through F1 show projections through the MB lobes of the respective GAL4 enhancer trap lines. Blue arrowheads indicate position of α′ lobe tip, red diamond indicates the center of the α lobe tips. Driver name is listed in the lower right hand corner of each panel. MB247(A1), c309(B1), c747(C1) and c739(D1) have little or no CD8::GFP expression in α′β′ neurons, but strong expression in αβ neurons. In contrast, c320(E1) and c305a(F1) show strong expression in α′β′ neurons. c305a is α′β′ neuron specific, while c320 expresses in both α′β′, αβ, and γ neurons. A2-F4. Panels A2 through F4 are single focal planes through the MBs showing colocalization of CD8::GFP (A2-F2) with FAS II (A3-F3) and TRIO (A4-F4), which mark αβ neurons and α′β′ neurons respectively. The GAL4 drivers used are the same as those shown as whole MB projections in the same row. Because α′β′, αβ, and γ neurons are all present in the optical section, the white dotted box in each panel highlights the portion of the left β′ lobe visible. A cartoon depicting the approximate lobe arrangement in these panels is shown in panel G1. In these figures we identify α′β′ neurons as being only TRIO positive, γ neurons as being only FAS II positive, and γ neurons as being both TRIO and FAS II positive. MB247(A2-A4), c309(B2-B4), c747(C2-C4) and c739(D2-D4) have relatively low CD8::GFP expression in β′ neurons in comparison to c320(E2-E4) and c305a(F2-F4). c305a appears to express in a larger fraction of β′ neurons than c320. G1. Cartoon depicting MB lobe arrangement in panels A2 through G4. G2-G4. A representative single focal plane showing the MB lobe arrangement seen in panels A2-F4 with FAS II and TRIO. Panels G3 and G4 show the expression pattern of FAS II (G3) and TRIO (G4) alone and G2 shows the merge. Panels G2-G4 are the same confocal section as panels F2-F4 for the c305a driver. Scale bar represents 10 μm.

To more precisely examine the enhancer trap expression patterns in the MB lobes, we co-labeled brains with anti-FASII antibody (Grenningloh et al., 1991) to mark the MB αβ neurons and anti-TRIO antibody to mark the MB α′β′ neurons (Awasaki et al., 2000) and examined the colocalization of CD8::GFP with these markers. Representative single confocal sections through the mushroom bodies for each of the aforementioned GAL4 lines are shown in Figure 1A2-1D4 and S1A1-D3. Our co-localization analysis confirmed the conclusions of our initial visual inspection. MB247 (Figure 1A2-A4, S1A1-3), c309 (Figure 1B2-B4, S1B1-3) and c747 (Figure 1C2-C4, S1C1-3) show little expression in α′β′ neurons and c739 (Figure 1D2-D4, S1D1-3) shows none. Based on our analysis we conclude that the previous studies that identified a requirement for MB output only during memory retrieval (Dubnau et al., 2001; McGuire et al., 2001; Schwaerzel et al., 2002; Davis, 2005) mostly blocked αβ and γ neuron activity and did not sufficiently address α′β′ neuron function.

To specifically investigate the role of α′β′ neurons we first screened a collection of enhancer-trap fly lines (www.fly-trap.org) for those that strongly expressed the GAL4 transcription factor in α′β′ neurons. We identified the c305a and c320 (Martini and Davis, 2005) lines as candidates and we verified the MB expression of these lines by crossing them to flies harboring a uas-CD8::GFP transgene (Figure 1E1, 1E2, 1F1, 1F2, S1E1-F3, S2 and S3) and subjecting the brains to the same analysis as the MB247, c309, c747, and c739 drivers.

Confocal microscopic analysis revealed CD8::GFP expression that was restricted to α′β′ neurons within the MBs in c305a (Figure 1F2-F4, S1F1-3), and α′β′ , αβ and a limited subset of γ neurons in c320 (Figure 1E2-E4, S1E1-3). The specificity of c305a for α′β′ neurons is particularly striking in single confocal sections: CD8::GFP shows almost no colocalization with FAS II (Figure 1F3, S1F2) and extensive colocalization with TRIO (Figure 1F4, S1F3). To further illustrate this specificity we have included the merged (Figure 1G2) and individual channels corresponding to FAS II (Figure 1G3) and TRIO (Figure 1G4) immunofluorescence shown in Figure 1F3 and 1F4 respectively. Other regions of significant expression in c305a include the antennal lobe, ring neurons in the ellipsoid body of the central complex and putative mechanosensory neurons in the antennal nerve (Figure 1F1 and S2). Although c305a and c320 also label other neurons in the brain we conclude that the most obvious common region of expression in the c305a and c320 lines is the MB α′β′ neurons (also see supplemental data).

Output from MB α′β′ neurons is required during and after training for consolidation of aversive odor memory

We used the c305a{GAL4} and c320{GAL4} lines (subsequently denoted c305a and c320) to examine the role of MB α′β′ neurons in aversive olfactory memory. Throughout this work we used the c739{GAL4} line (subsequently denoted c739) to compare the role of MB αβ neurons. We taught flies to associate an odor CS with a punitive electric shock US using standard protocols (Tully and Quinn, 1985; Keene et al., 2004). We first tested the role of MB α′β′ neurons in memory by blocking their output throughout the entire olfactory conditioning experiment (Figure 2B). We expressed the dominant temperature sensitive shibire^ts1 transgene (Kitamoto, 2001) in MB α′β′ neurons and performed memory experiments at either the permissive (25 °C) or the restrictive temperature (31 °C). At the restrictive temperature, shi^ts1 blocks vesicle recycling and thereby blocks synaptic vesicle release. At 25 °C odor memory scores of wild-type, uas-shi^ts1, c305a, c305a; uas-shi^ts1, c320, c320; uas-shi^ts1 and c739; uas-shi^ts1 flies were statistically indistinguishable (P>0.9) (Figure 2A). However, at 31 °C memory of c305a; uas-shi^ts1, c320; uas-shi^ts1 and c739; uas-shi^ts1 flies was severely reduced and was statistically different from wild-type, c305a, c320 and uas-shi^ts1 flies (c305a; uas-shi^ts1 all P<0.02; c320; uas-shi^ts1 all P<0.001; c739{GAL4}; uas-shi^ts1 all P<0.001) (Figure 2B). Therefore, MB αβ and α′β′ neuron synaptic release is necessary for odor memory.

Figure 2. Neurotransmission from MB α′β′ neurons is required for acquisition and consolidation of aversive odor memory whereas transmission from MB αβ neurons is only required for retrieval.

The temperature shift protocols are shown pictographically above each graph. A. The permissive temperature of 25°C does not affect 3 hour aversive odor memory of any of the lines used in this study. All genotypes were trained and tested for 3 hour memory at 25°C. B. Disrupting MB α′β′ or αβ neuron output at the restrictive temperature of 31°C impairs memory. All genotypes were trained and tested for 3 hour aversive odor memory at 31°C. C. Blocking MB α′β′ neuron output, but not MB αβ neuron output, during training impairs 3 hour aversive odor memory. Flies were incubated at 31°C for 15 minutes prior to and during training. Immediately after training they were returned to 25°C and tested for 3 hour memory. D. Blocking MB αβ neuron output, but not MB α′β′ neuron output, during testing disrupts 3 hour aversive odor memory. Flies were trained at 25°C and 165 minutes later they were shifted to 31°C. 15 minutes later 3 hour memory was tested at 31°C. E. Blocking MB α′β′ neuron output, but not MB αβ neuron output, immediately after training severely impairs 3 hour aversive odor memory. Flies were trained at 25°C and immediately after training they were shifted to 31°C for 60 minutes. Flies were then returned to 25 °C and tested for 3 hour aversive odor memory at 25°C. Error bars are s.e.m. Asterisks denote significant difference (P<0.05) from all other unmarked groups. All flies harbor two copies of the uas-shi^ts1 transgene.

The four previous reports that principally blocked output from MB αβ or MB αβ and γ neurons concluded that MB output was dispensable during memory acquisition and storage but was required during memory retrieval (Dubnau et al., 2001; McGuire et al., 2001; Schwaerzel et al., 2002; Davis, 2005). We therefore investigated whether MB α′β′ neurons were required during memory acquisition, consolidation or retrieval. We again blocked MB α′β′ neuron and MB αβ neuron output with uas-shi^ts1 but we restricted the inactivation to either the training, testing or storage period. Blocking MB αβ neurons during acquisition did not produce memory loss, consistent with the previous report (McGuire et al., 2001) (Figure 2C). Memory of c739; uas-shi^ts1 flies was comparable to wild-type and uas-shi^ts1 flies (both P>0.9). However, blocking α′β′ neuron output during acquisition impaired memory. Memory of c305a; uas-shi^ts1 flies was statistically different to wild-type (P<0.001), uas-shi^ts1 (P<0.01) and c739; uas-shi^ts1 flies (P<0.01). Memory of c320; uas-shi^ts1 flies was statistically different to wild-type (P<0.001), uas-shi^ts1 (P=0.001) and c739; uas-shi^ts1 flies (P<0.001). In contrast, blocking MB output during testing revealed that MB αβ neuron output is required for memory retrieval, consistent with the previous report (McGuire et al., 2001), whereas α′β′ neuron output is dispensable (Figure 2D). Memory of c739; uas-shi^ts1 flies was statistically different to wild-type (P=0.001), uas-shi^ts1 (P<0.001), c305a; uas-shi^ts1 (P<0.01) and c320; uas-shi^ts1 flies (P<0.01) whereas memory of c305a; uas-shi^ts1 and c320; uas-shi^ts1 flies was statistically indistinguishable from wild-type (P=0.75 and P=0.3 respectively) and uas-shi^ts1 flies (P=0.8 and P=0.4 respectively).

We next tested whether MB αβ neuron and α′β′ neuron output was required after training during memory storage. We trained flies at the permissive temperature and immediately after training we blocked MB α′β′ neuron or MB αβ neuron output for 60 min by shifting the flies to the restrictive temperature. We then returned flies to the permissive temperature and tested memory two hours later. Strikingly this manipulation severely impaired memory if α′β′ neuron output was blocked but did not affect performance if MB αβ neurons were blocked (Figure 2E). Memory of c305a; uas-shi^ts1 and c320; uas-shi^ts1 flies was statistically different from wild-type (both P<0.001), uas-shi^ts1 flies (both P<0.001) and c739; uas-shi^ts1 flies (both P<0.001) whereas memory of c739; uas-shi^ts1 flies was indistinguishable from wild-type (P=0.6) and uas-shi^ts1 flies (P=0.9). These data suggest that output from the MB α′β′ neurons is required for memory consolidation whereas, consistent with previous studies MB αβ neuron output is dispensable (Dubnau et al., 2001; McGuire et al., 2001; Schwaerzel et al., 2002; Davis, 2005). Furthermore these data are consistent with the notion that subsets of MB neurons have different roles in memory processing.

To conclude a memory-specific effect, it is necessary to determine that the experimental manipulation does not interfere with olfaction or shock avoidance. We tested the olfactory and shock acuity of c305a; uas-shi^ts1 and c320; uas-shi^ts1 flies at 25 °C and 31 °C (Table 1). The odor and shock acuity of c305a; uas-shi^ts1 flies is statistically indistinguishable from uas-shi^ts1 and wild-type flies at both temperatures (all P>0.5). However, c320; uas-shi^ts1 flies have normal shock acuity (P>0.8) but have a statistically significant olfactory acuity defect at 31 °C (P<0.05 for MCH). This acuity defect is somewhat surprising because c320; uas-shi^ts1 flies display normal memory performance when they are tested for memory retrieval at 31 °C (Figure 2D) suggesting that the flies can still discriminate between the odors. Nevertheless, the defective olfactory acuity of c320; uas-shi^ts1 flies questions the validity of the c320; uas-shi^ts1 acquisition block experiment where acuity is compromised during training. Hence, in the acquisition experiments we rely on the c305a; uas-shi^ts1 flies, whose acuity remains intact. Therefore our data suggest that stable memory requires neurotransmission from MB α′β′ neurons during and after training whereas memory retrieval exclusively depends on output from MB αβ neurons.

Table 1.

Olfactory Acuity and Shock Avoidance scores for strains used in this study. There are no statistical differences between the relevant groups other than c320{GAL4};uas-shi^ts1 flies which display lower MCH acuity at 31°C (denoted in bold).

	OCT		MCH		shock avoidance
	25°C	31°C	25°C	31°C	25°C	31°C
wild-type	0.52+/−0.07	0.65+/−0.07	0.88+/−0.02	0.89+/−0.04	0.42+/−0.02	0.68+/−0.04
uas-shi; uas-shi	0.53+/−0.03	0.70+/−0.04	0.90+/−0.03	0.94+/−0.02	0.39+/−0.04	0.73+/−0.05
uas-shi; c305a;uas-shi	0.59+/−0.05	0.62+/−0.03	0.90+/−0.02	0.92+/−0.01	0.45+/−0.06	0.74+/−0.03
uas-shi; c320;uas-shi	0.51+/−0.07	0.53+/−0.03	0.79+/−0.03	0.60+/−0.09	0.47+/−0.06	0.71+/−0.06
wild-type	0.64+/−0.06		0.70+/−0.04		0.40+/−0.02
uas-shi	0.68+/−0.04		0.75+/−0.07		0.41+/−0.03
c739;uas-shi	0.77+/−0.04		0.69+/−0.04		0.41+/−0.03
c305a/MBGAL80;uas-shi	0.64+/−0.03		0.75+/−0.07		0.41+/−0.03
c320/MBGAL80;uas-shi	0.71+/−0.08		0.76+/−0.09		0.41+/−0.04
c739/MBGAL80;uas-shi	0.70+/−0.07		0.77+/−0.08		0.41+/−0.04

Inhibiting MB expression in c305a and c320 reverses the uas-shi^ts1 induced memory loss

The c305a and c320 lines express in MB α′β′ neurons and in other neurons in the brain including the central complex and antennal lobes (Figure 1, 3, S2 and S3). Although we believe the only region of overlap between c305a and c320 is MB α′β′ neurons, we incorporated a MB-expressed GAL80 repressor of GAL4 (Lee and Luo, 1999) to determine if MB expression in the c305a and c320 lines was required for the memory loss in the previous experiments.

Figure 3. MB α′β′ neuron expression is required for the c305a and c320 dependent memory loss.

A. c305a{GAL4}; uas-CD8::GFP, B. c320{GAL4}; uas-CD8::GFP and C. c739{GAL4}; uas-CD8::GFP counterstained with anti-FASII to label MB αβ neurons. D., E. and F. Combining MB{GAL80} with c305a{GAL4}; uas-CD8::GFP, c320{GAL4}; uas-CD8::GFP and c739{GAL4}; uas-CD8::GFP eliminates MB neuron expression but expression elsewhere remains largely intact. Scale bar represents 10 μm. G. and H. Combining MB{GAL80} with c305a{GAL4}; uas-shi^ts1 and c320{GAL4};uas-shi^ts1 reverses the temperature induced memory loss observed when c305a{GAL4} and c320{GAL4} neurons are blocked G. during acquisition and H. for one hour after training. The temperature shift protocols are shown pictographically above each graph. Flies were tested for 3 hour aversive odor memory at 25°C. Error bars are s.e.m. Asterisks denote significant difference (P<0.05) from all other unmarked groups. All flies harbor one copy of the uas-shi^ts1 transgene.

The MB-specific GAL80, MB{GAL80} (a gift from H.Tanimoto), harbors a GAL80 transgene driven by a 247 base pair fragment from the D-Mef2 promoter which drives expression predominantly in the MB (Zars et al., 2000, Mao et al., 2004, Riemensperger et al., 2005). Our data presented here (Figure 3, S2 and S3) suggest that MB{GAL80} expresses GAL80 throughout the MB αβ, α′β′ and γ neurons (similar to MB247-dsRed in Riemensperger et al., 2005). In contrast, although the commonly used MB247{GAL4} line (Zars et al., 2000 and Figure 1A1-4) contains the same 247bp D-mef2 fragment fused to GAL4, it only expresses GAL4 in the MB αβ and γ neurons and does not express strongly in MB α′β′ neurons (Figure 1A1-4), presumably due to position-specific effects from the site of transgene insertion in the genome. We combined the MB{GAL80} insertion with the c305a, c320 and c739 drivers and uas-CD8::GFP and compared the resultant GFP expression patterns with that of the drivers without MB{GAL80}.

The presence of the MB{GAL80} transgene specifically abolished GAL4 activity in MBs but left expression elsewhere largely intact (Figure 3, S2 and S3). Compared to c305a; uas-CD8::GFP flies (Figure 3A and S2A-B), α′β′ neuron GFP expression was always eliminated in c305a/MB{GAL80}; uas-CD8::GFP flies (Figure 3D and S2C-D), antennal lobe expression was slightly reduced, and ellipsoid body expression appeared unchanged (see a more detailed discussion of the drivers and GAL80 effects in the supplementary information). In c320/MB{GAL80}; uas-CD8::GFP flies MB α′β′, αβ and γ GFP expression was eliminated while the CC and diffuse expression elsewhere remained (compare Figure 3B and 3D and see S3A-D). Lastly, in c739/MB{GAL80}; uas-CD8::GFP flies MB αβ neuron expression was removed while expression elsewhere appeared unchanged (compare Figure 3C and 3F). We therefore conclude that the MB{GAL80} transgene eliminated GAL4-mediated expression in the MBs.

We used the MB{GAL80} transgene in behavioral experiments with c305a, c320 and c739. We constructed a MB{GAL80}; uas-shi^ts1 fly line and crossed the flies to c305a, c320 and c739 flies to ask whether the MB α′β′ neurons were responsible for the memory loss when c305a and c320 neurons were blocked during training and for one hour after training. We trained c305a; uas-shi^ts1, c320; uas-shi^ts1 and c739; uas-shi^ts1 flies, with and without MB{GAL80}, at the restrictive temperature and immediately after training we returned the flies to the permissive temperature. We then tested memory three hours later. Strikingly this manipulation did not impair memory when α′β′ neuron expression was blocked by the presence of MB{GAL80} (Figure 3G). Memory of c305a/MB{GAL80}; uas-shi^ts1 and c320/MB{GAL80}; uas-shi^ts1 flies was statistically indistinguishable from wild-type (both P=1) and uas-shi^ts1 flies (both P=1) and was statistically different to that of c305a; uas-shi^ts1 (P=0.01 and P=0.02 respectively) and c320; uas-shi^ts1 flies (P<0.01 and P=0.01 respectively). We next trained c305a; uas-shi^ts1, c320; uas-shi^ts1 and c739; uas-shi^ts1 flies, with and without MB{GAL80}, at the permissive temperature and immediately after training we blocked neuron output for 60 min by shifting the flies to the restrictive temperature. We then returned flies to the permissive temperature and tested memory two hours later. Strikingly this manipulation did not impair memory when α′β′ neuron expression was blocked by the presence of MB{GAL80} (Figure 3H). Memory of c305a/MB{GAL80}; uas-shi^ts1 and c320/MB{GAL80}; uas-shi^ts1 flies was statistically indistinguishable from wild-type (P>0.9 and P=1 respectively) and uas-shi^ts1 flies (P>0.9 and P=1 respectively) and was statistically different to that of c305a; uas-shi^ts1 (P<0.005 and P=0.001 respectively) and c320; uas-shi^ts1 flies (P=0.03 and P<0.02 respectively). To control against non-specific effects on memory by MB{GAL80} we also combined MB{GAL80} with c739. Memory of c739/MB{GAL80}; uas-shi^ts1 flies was statistically indistinguishable from that of c739; uas-shi^ts1 flies (P=1). Although we acknowledge that MB{GAL80} causes a modest reduction in AL labeling of c305a flies, there is no obvious AL overlap between c305a and c320. Therefore, these data strongly suggest that MB α′β′ neurons are responsible for the memory loss observed following blockade of c305a and c320 neurons during and after training.

Output from MB α′β′ neurons is required during and after training for consolidation of appetitive odor memory

Flies can also be taught to associate an odor CS with an appetitive sugar reward US (Tempel et al., 1983) and MB output is also required to recall appetitive odor memory (Schwaerzel et al., 2003). However, a role for MB output in appetitive odor memory processing has not been reported. We therefore tested whether output from MB α′β′ neurons was also required for acquisition and consolidation of appetitive odor memory (Figure 4). We trained flies to associate odor with sucrose reward and tested whether blocking MB α′β′ neuron output either during training or for one hour after training impaired appetitive memory. At 25 °C (Figure 4A) odor memory performance of c305a; uas-shi^ts1, c320; uas-shi^ts1 and c739; uas-shi^ts1 flies was indistinguishable from that of wild-type and uas-shi^ts1 flies (all P 0.2). However, blocking α′β′ neuron output during training (Figure 4B) significantly impaired memory whereas blocking αβ neurons had no effect. Memory of c305a; uas-shi^ts1 and c320; uas-shi^ts1 flies was statistically different from wild-type (both P=0.02) and uas-shi^ts1 flies (both P<0.02) whereas memory of c739; uas-shi^ts1 flies was statistically indistinguishable from wild-type (P>0.9) and uas-shi^ts1 flies (P>0.9). We also tested for a role of α′β′ neurons during appetitive memory storage (Figure 4C). Immediately after training we blocked α′β′ neuron output for 60 min by shifting the flies to the restrictive temperature. We then returned flies to the permissive temperature and tested memory two hours later. Similar to aversive odor memory, this manipulation impaired memory when α′β′ neurons were blocked but not if αβ neurons were blocked. Memory of c305a; uas-shi^ts1 and c320; uas-shi^ts1 flies was statistically different from wild-type (both P<0.001), uas-shi^ts1 flies (P<0.002 and P<0.001 respectively) and c739; uas-shi^ts1 flies (P<0.002 and P<0.001 respectively). These data suggest that output from MB α′β′ neurons is required during training and storage for appetitive memory processing like it is for aversive odor memory.

Figure 4. MB α′β′ neuron output is required for stable appetitive odor memory.

The temperature shift protocol is shown pictographically above each graph. A. The permissive temperature of 25°C does not affect 3 hour appetitive odor memory of c305a{GAL4};uas-shi^ts1, c320{GAL4};uas-shi^ts1 or c739{GAL4}; uas-shi^ts1 flies. All genotypes were trained and tested for 3 hour memory at 25°C. B. Blocking MB α′β′ neuron (c305a and c320) output, but not MB αβ neuron (c739) output, during training impairs 3 hour aversive odor memory. Flies were incubated at 31°C for 15 minutes prior to and during training. Immediately after training they were returned to 25°C and tested for 3 hour memory. C. Blocking MB α′β′ neuron (c305a and c320) output, but not αβ neuron (c739) output, immediately after training severely impairs 3 hour appetitive odor memory. Flies were trained at 25°C and immediately after training they were shifted to 31°C for 60 minutes. Flies were then returned to 25°C and tested for 3 hour appetitive odor memory at 25°C. Error bars are s.e.m. Asterisks denote significant difference (P<0.05) from other unmarked groups. All flies harbor two copies of the uas-shi^ts1 transgene.

DPM neuron output to MB α′β′ lobe neurons is required after training for aversive odor memory consolidation

Prolonged Dorsal Paired Medial (DPM) neuron output after training is required to consolidate aversive and appetitive odor memory (Keene et al., 2004; Yu et al., 2005; Keene et al., 2006). Expressing a uas-DScam17-2::GFP transgene in DPM neurons (with the c316{GAL4} driver) selectively reduces DPM neuron projections to the MB α, β and γ lobes leaving DPM neurons that primarily project to the MB α′β′ lobes (Keene et al., 2006). Furthermore, flies with DPM neurons that primarily project to the MB α′β′ lobes retain the ability to consolidate both aversive and appetitive memory (Keene et al., 2006). Memory of uas-DScam17-2::GFP; c316 flies is indistinguishable from that of wild-type flies (Keene et al., 2006 and Figure 5, P>0.5). We asked whether DPM neurons projecting mostly to the MB α′β′ lobes have the same temporal requirement as wild-type DPM neurons and MB α′β′ neurons. We used the c316 driver to coexpress uas-DScam17-2::GFP and uas-shi^ts1 transgenes in DPM neurons. We blocked DPM output for one hour after training by shifting the flies from the permissive temperature (25 °C) to the restrictive temperature (31 °C). Blocking DPM output for 60 min after training significantly reduced 3hr aversive odor memory regardless of whether the uas-DScam17-2::GFP transgene is present or not (Figure 5). Memory of uas-DScam17-2::GFP; c316/uas-shi^ts1 flies is statistically different from wild-type (P<0.001), uas-shi^ts1 (P<0.0001) and uas-DScam17-2::GFP; uas-shi^ts1 flies (P<0.0001) and was statistically indistinguishable from c316/uas-shi^ts1 flies (P>0.3). These data suggest that output from DPM neurons to MB α′β′ neurons is required for memory consolidation.

Figure 5. Blocking synaptic transmission after training from DPM neurons that primarily project to the MB α′β′ lobes abolishes memory.

DPM neurons can be engineered to primarily project to the MB α′β′ lobes by expressing a uas-DScam17-2::GFP transgene in DPM neurons. Blocking output from the modified DPM projections for one hour after training severely impairs 3 hour odor memory. The temperature shift protocol is shown pictographically. Flies were trained at 25°C and immediately after training they were shifted to 31°C for 60 minutes. Flies were then returned to 25°C and tested for 3 hour aversive odor memory at 25°C. Error bars are s.e.m. Asterisks denote significant difference (P<0.05) from all other unmarked groups. All flies harbor one copy of the uas-shi^ts1 transgene.

Discussion

Mushroom bodies are required for acquisition, storage, and recall of olfactory memories

It is often said that form follows function. While this postulate would argue the striking multi-lobed arrangement of the insect mushroom bodies implies functional differences between the different types of MB neurons: αβ, α′β′ and γ, very limited data exists describing the individual function of these anatomical subdivisions. Although several complex behaviors in insects appear to require the MBs (Heisenberg et al., 1985; de Belle and Heisenberg, 1994; Martin et al., 1998; Liu et al., 1999; McBride et al., 1999; Tang and Guo, 2001; Pitman et al., 2006; Joiner et al., 2006) and a differential role for distinct MB neuron groups has been suggested (in memory by Zars et al., 2000; McGuire et al., 2001; McGuire et al., 2003; Akalal et al., 2006; and in sleep by Joiner et al., 2006), most conceptual models of memory treat the MBs as a single unit.

One of the most detailed examinations of MB function has been in the context of Drosophila aversive olfactory memory, where flies are trained to associate specific odors with the negative reinforcement of electric shock (Tully and Quinn, 1985). Genetic studies over the last thirty years have suggested that the MBs play an essential role in fly olfactory memory (Heisenberg, 2003; Davis, 2005), but the role of the MBs in memory acquisition, storage, and retrieval has only been examined recently. Taking advantage of a dominant, temperature-sensitive dynamin transgene, uas-shi^ts1, a number of laboratories concluded that MB output was required only for recall, but not acquisition or storage (Dubnau et al., 2001; McGuire et al., 2001; Schwaerzel et al., 2002; Davis, 2005). These and other findings have led to a simple model wherein Drosophila olfactory memory is formed and stored at MB output synapses.

Our functional studies of DPM neurons, MB extrinsic neurons that ramify throughout the MB lobes, demonstrated they were specifically required during consolidation, but not acquisition or storage (Waddell et al., 2000; Keene et al., 2004; Yu et al., 2005; Keene et al., 2006). Furthermore, genetically-modified DPM neurons that primarily innervate the MB α′β′ lobes retain function implying that MB α′β′ neurons might also have a similar function in memory consolidation (Keene et al., 2006).

Our examination of the GAL4 enhancer trap lines used to express the uas-shi^ts1 transgene in the earlier MB studies revealed that c309, c747, and MB247 only express in a few MB α′β′ neurons compared to αβ and γ neurons, while c739 expresses exclusively in αβ neurons. Thus it seems likely that prior studies utilizing these drivers did not observe requirements for MB activity during olfactory memory acquisition and storage because of insufficient expression in α′β′ neurons.

We subsequently identified two GAL4 enhancer traps that strongly express in MB α′β′ neurons to test this hypothesis. The expression of c305a appears to be entirely restricted to α′β′ neurons within the MBs whereas c320 expresses in α′β′, αβ, and a few γ neurons. Both of these lines also express in additional non-MB neurons so we employed a MB{GAL80} tool to more rigorously test the requirement for MB activity, when we manipulated the neurons labeled with either {GAL4} line. With these new reagents we investigated the role of MB α′β′ neurons in memory and found that MB α′β′ neuron output during and after training is critical to form, and consolidate, both appetitive and aversive odor memory from a labile to a more stable state. For comparison we also examined the requirements for MB αβ neurons using c739, confirming the results of McGuire et al. (2001). Thus, output from the MB α′β′ neuron subset is required for memory acquisition and stabilization whereas, as previously described, output from αβ neurons is apparently dispensable during training and consolidation but is required for memory retrieval (McGuire et al., 2001).

Based on our c305 and c739 data, we recognize that c320 flies, which express in both α′β′ and αβ neurons, might be expected to exhibit memory loss if MB neuron output was blocked during both the consolidation and recall time windows. However, it is possible that we did not observe a retrieval effect with c320 because it expresses GAL4 in fewer αβ neurons, or is in a different subset of αβ neurons, relative to the c739 driver.

Despite this caveat, we believe our data suggest that different lobes of the MB have different roles in memory and provide a significant shift in our understanding of the role of the MB in memory. Older models implied that MB αβ, α′β′ and γ neurons were largely interchangeable and that each of the MB neurons that responded to a particular odor received coincident CS and US input and modified their presynaptic terminals to encode the memory. The data presented here suggest that MB αβ and α′β′ neurons are functionally distinct.

In this study, we did not investigate the role of the unbranched γ lobe neurons. Previous work with c309, c747 and MB247 suggests that neurotransmission from γ neurons is likely dispensable for acquisition and consolidation (Dubnau et al., 2001; McGuire et al., 2001; Schwaerzel et al., 2002; Davis, 2005). In addition, a prior study indicated that γ neurons are minimally involved in MTM and ARM (Isabel et al., 2004). However, it is possible that experiments to date have not employed odors that require γ neuron activation. The response of γ neurons may be tailored to ethologically relevant odors such as pheromones. It is notable that fruitless, a transcription factor required for male courtship behavior, is expressed in MB γ neurons (Stockinger et al., 2005; Manoli et al., 2005) and blocking expression of the male-specific fru^M transcript in the MB γ neurons impairs courtship conditioning (Manoli et al., 2005). If the relevant odors can be identified, it will be interesting to determine if MB α′β′ neurons and DPM neurons are required to stabilize these odor memories in the γ neurons. Recent work by Akalal et al. (2006) is supportive of the idea that odor identity is an important factor in determining the requirement for the function of distinct subsets of MB neurons in olfactory learning.

Stable aversive and appetitive odor memory requires prolonged DPM neuron output during the first hour after training and DPM neuron output is dispensable during training and retrieval (Keene et al., 2004; Keene et al., 2006). DPM neurons ramify throughout the MB lobes but DPM neurons that have been engineered to project mostly to the MB α′β′ lobes retain wild-type capacity to consolidate both aversive and appetitive odor memory (Keene et al., 2006). In this study we have demonstrated that similar to wild-type DPM neurons, blocking output from these modified DPM neurons for one hour after training abolishes memory. Thus finding a specific role for both DPM neuron output to MB α′β′ lobes and MB α′β′ neuron output during the first hour after training is consistent with the notion that a direct DPM-MB α′β′ neuron synaptic connection is important for memory stability. It should be reiterated that the focus of this paper has been on protein-synthesis-independent memory and whether or not a similar processing circuit is utilized for protein synthesis-dependent LTM (Tully et al., 1994) remains an open question.

Beyond simply attributing an additional function to the MBs, when taken in conjunction with our work on the role of DPM neurons in memory (Waddell et al., 2000; Keene et al., 2004; Yu et al., 2005; Keene et al., 2006), the data presented here suggest a new model for how olfactory memories are processed within the MBs. We propose that olfactory information received from the second-order projection neurons is first processed in parallel by the MB αβ and α′β′ neurons during acquisition. Activity in α′β′ neurons establishes a recurrent α′β′ neuron-DPM neuron loop that is necessary for consolidation of memory in αβ neurons and subsequently memories are “stored” in αβ neurons, whose activity is required during recall. It is plausible that MB α′β′ neurons are directly connected to MB αβ neurons and/or that DPM neurons provide the conduit between MB neurons. However, our finding that DPM neurons that project primarily to MB α′β′ neurons are functional implies that only a few connections from DPM neurons to MB αβ neurons are necessary.

The requirement for α′β′ neuron output during training also potentially provides a source for the activity that drives DPM neurons. DPM neuron activity is not required during training (Keene et al., 2004; Keene et al., 2006) and our current data are consistent with the idea that olfactory conditioning triggers activity in MB α′β′ neurons that in turn elicits DPM neuron-dependent activity. We propose that after training recurrent MB α′β′ neuron-DPM neuron activity is self-sustaining for 60–90 minutes (Yu et al., 2005). This recurrent network mechanism is similar to models for working memory in mammals (Durstewitz et al., 2000). It is also conceivable that MB α′β′ neurons receive prolonged input after training from the antennal lobes (AL) via the projection neurons (PN). Olfactory conditioning has been reported to alter the odor response of Drosophila PNs in the AL but the observed effects were short-lived (Yu et al., 2004). Nevertheless, AL plasticity for a few minutes after training could contribute to the required MB α′β′ neuron activity. If continued activity from the AL is required for consolidation, blocking PN transmission with shibire^ts1 for one hour after training should abolish memory. The bee AL and MB are clearly involved in olfactory memory and may function somewhat independently in learning and memory consolidation respectively (Hammer and Menzel, 1998). However, biochemical manipulation of the bee AL can also induce LTM (Muller, 2000) and therefore it is possible that either plasticity in the AL alone can support LTM, or that the AL and MB interact during acquisition and consolidation. A differential role for the AL and MBs has also been suggested from neuronal ablation studies of courtship conditioning in Drosophila. Short-term courtship memory can be supported by the AL but memory lasting longer than 30 min requires the MBs (McBride et al., 1999).

Our work also has significant implications for the organization of aversive and appetitive odor memories in the fly brain. Stability of both appetitive and aversive memory is dependent on DPM neurons (Keene et al., 2006) and MB α′β′ neurons. It therefore appears that processing of aversive and appetitive odor memories may bottleneck in the MBs. Schwaerzel et al., (2003) demonstrated that aversive memory formation requires dopaminergic neurons whereas appetitive memory relies on octopamine providing a possible mechanism to distinguish valence. However, they also found that MB output is required to retrieve aversive and appetitive odor memory suggesting that both forms of memory involve MB neurons and that both US pathways may converge on MB neurons. It will be important to understand how the common circuitry is organized to independently process the different types of memory and to establish if, and how, such memories co-exist.

Our data taken with that of McGuire et al. (2001) and Isabel et al. (2004) imply that stable memory may reside in MB αβ neurons because blocking output from MB αβ neurons impairs retrieval of MTM and ARM (both components of 3-hour memory). We previously proposed that AMN peptide(s) released from DPM neurons cause prolonged cAMP synthesis in MB neurons that is required to stabilize memory (Waddell et al., 2000). Our finding that genetically-engineered DPM neurons mostly projecting to the MB α′β′ lobes, are functional (Keene et al., 2006) taken with the idea that stable memory resides in MB αβ neurons is somewhat inconsistent with the notion that crucial AMN-dependent memory processes occur in MB αβ neurons. However, it is plausible that AMN, or another DPM product that is released in a shibire-dependent manner, could diffuse locally from the aberrant DPM neurons to MB αβ neurons.

This work demonstrates that MB αβ neurons and α′β′ neurons have different roles in memory. Beyond gross structural and gene expression differences, it will be essential to establish the precise connectivity, relative excitability and odor responses of the different MB neurons. Future study may also reveal further functional subdivision within the MB lobes and it should be possible to refine our current MB α′β′ neuron GAL4 lines with appropriate GAL80 transgenes and FLP-out technology (Golic and Lindquist, 1989).

A neural circuit-based view of olfactory memory

In the mammalian brain memories that initially depend on the function of the hippocampus lose this dependence when they are consolidated. This transient involvement of the hippocampus has led to the idea that consolidation of memory results in the transfer of memory from the hippocampal circuits to the cortex. An alternate view is that aspects of the memory are always in the cortex but they are dependent on the hippocampus because recurrent activity from cortex to hippocampus to cortex is required for consolidation. Hence, disrupting hippocampal activity during consolidation leads to memory loss.

Our data suggest the simpler fruit fly brain similarly employs parallel and sequential use of different regions to process memory. MB α′β′ neuron activity is required to form memory, MB α′β′ neurons and DPM neurons are transiently required to consolidate memory and output from αβ neurons is exclusively required to retrieve memory. We therefore propose that aversive and appetitive odor memories are formed in MB αβ neurons and are stabilized there by recurrent activity involving MB α′β′, DPM neurons and the MB αβ neurons themselves.

It is becoming increasingly apparent that neural circuit analysis will play an important role in understanding how the brain encodes memory. The ease and sophistication with which one can manipulate circuit function in Drosophila, combined with the relative simplicity of insect brain anatomy should ensure that the fruit fly will make significant contributions to this emerging discipline.

Experimental Procedures

Fly strains

Fly stocks were raised on standard cornmeal food at 25°C and 40–50% relative humidity. The wild-type Drosophila strain used in this study is Canton-S. The uas-mCD8::GFP flies are described (Lee and Luo, 1999). We used flies carrying either a single insertion of the uas-shi^ts1 transgene (Kitamoto, 2001) on the third (Figure 3 and 5) or two insertions of the uas-shi^ts1 transgene on the X and third chromosome respectively (Figure 2 and 4). We previously described the DPM neuron-restricted c316{GAL4} (Waddell et al., 2000). The uas-DScam[exon17-2]::GFP flies, here designated as uas-DScam17-2::GFP are described (Wang et al., 2004). We generated flies expressing shi^ts1 in DPM neurons that primarily project to the MB α′β′ lobes by crossing uas-DScam17-2::GFP/CyO; uas-shi^ts1 males to homozygous w;c316{GAL4} females. All flies were trained and tested together and uas-DScam17-2::GFP; c316{GAL4}/ uas-shi^ts1 flies were sorted from CyO; c316{GAL4}/ uas-shi^ts1 flies after testing and were counted separately. c320{GAL4} flies were previously described (Martini and Davis, 2005). We generated c305a{GAL4}; uas-shi^ts1, c320{GAL4}; uas-shi^ts1 and c739{GAL4}; uas-shi^ts1 flies by crossing uas-shi^ts1 females to c305a, c320 and c739 male flies. We generated c305a{GAL4}/MB{GAL80}; uas-shi^ts1, c320{GAL4}/MB{GAL80}; uas-shi^ts1 and c739{GAL4}/MB{GAL80}; uas-shi^ts1 flies by crossing MB{GAL80}; uas-shi^ts1 female flies to c305a, c320 and c739 male flies. Heterozygote uas-shi^ts1/+, MB{GAL80}/+; uas-shi^ts1/+, c305a/+, c320/+ and c739/+ flies were generated by crossing uas-shi^ts1, MB{GAL80}; uas-shi^ts, c305a, c320 and c739 flies to wild-type flies. Unless stated otherwise all flies tested are heterozygote for the listed transgenes and a mixed population of sexes was tested for olfactory memory.

Behavioral analysis

The olfactory avoidance paradigm was performed as described previously (Tully and Quinn, 1985; Keene et al., 2004). The Performance Index (PI) is calculated as the number of flies avoiding the conditioned odor minus the number of flies avoiding the unconditioned odor divided by the total number of flies in the experiment. A single PI value is the average score from flies of the identical genotype trained with each odor (3-Octanol or 4-Methylcyclohexanol). Olfactory conditioning with sugar-reward was performed as previously described (Keene et al., 2006). Flies were starved for 16–20 hours before conditioning. The PI is calculated as the number of flies running toward the conditioned odor minus the number of flies running toward the unconditioned odor divided by the total number of flies in the experiment. A single PI value is the average score from flies of the identical genotype tested with each odor (3-Octanol or 4-Methylcyclohexanol). To reduce variation within experiments, all genotypes were tested in each experimental session.

We previously determined that the c316, uas-shi^ts1, c316; uas-shi^ts1, uas-DScam17-2::GFP and uas-DScam17-2::GFP; c316 flies strains tested in this study have normal odor, electric shock and sugar acuity (Waddell et al., 2000; Keene et al., 2004; Keene et al., 2006). We tested the odor, shock and sugar acuity of the remaining stocks (see Table 1) using previously reported methods (Keene et al., 2004; Keene et al., 2006).

Statistical analyses were performed using KaleidaGraph (Synergy Software). Overall analyses of variance (ANOVA) were followed by planned pairwise comparisons between the relevant groups with a Tukey HSD post-hoc test. Unless stated otherwise, all experiments are n≥8.

Immunocytochemistry

Adult brains expressing transgenic uas-mCD8::GFP were removed from the head capsule and fixed in 4% paraformaldehyde in Phosphate Buffered Saline (PBS), [1.86mM NaH₂PO₄, 8.41mM Na₂HPO₄, 175mM NaCl] for 15 min, and rinsed in PBS-T (PBS containing 0.25% Triton X-100). Brains were incubated with the following antibody concentrations; 1:4 mAb anti-TRIO (Awasaki et al., 2000), 1:4 mAb 1D4 anti-FASII (Grenningloh et al., 1991) (Hybridoma Bank, University of Iowa), 1:3000 Rb anti-FASII (gift from V. Budnik), 1:200 mAb anti-GFP (Invitrogen). They were then incubated with the appropriate fluorescent secondary antibodies (Jackson Laboratories). Confocal analysis was performed on a Zeiss LSM 5 Pascal confocal microscope. All samples to be compared were processed in parallel and images were acquired using identical microscope settings. Confocal stacks were processed using ImageJ and Adobe Photoshop.