|Home | About | Journals | Submit | Contact Us | Français|
Using functional optical imaging in vivo, we demonstrate that the γ mushroom body (MB) neurons of Drosophila melanogaster respond with axonal calcium influx when odors or electric shock stimuli are presented to the fly. Pairing of odor and electric shock stimuli in a single training trial or multiple, massed training trials failed to modify the odor-evoked calcium signal when flies were tested at several different times after training. In contrast, animals that received multiple but spaced odor-shock pairings exhibited a robust increase in calcium influx into the MB axons when tested between 18 and 48 hr after training. This time window for the γ neuron memory trace is displaced relative to the modifications that occur between 9 and 24 hr after training in the α branch of the α/β MB neurons. The α/β and the γ neuron long-term memory traces were both blocked by expressing a repressor of the transcription factor Creb or a CaMKII hairpin RNA. These results demonstrate that behavioral long-term olfactory memory is encoded as modifications of calcium influx into distinct MB neurons during overlapping but different windows of time after training.
Drosophila exposed to odors paired with electric shock learn this association and display their memory as a selective avoidance of the odor paired with the negative reinforcer. The strength and duration of the memory is dependent on the the training protocol. For instance, a single cycle of conditioning generates robust short-term memory that decays over the period of about 1 day (Tully et al., 1994; Beck et al., 2001; Pascual and Preat, 2001). Multiple cycle conditioning with no rest between each cycle (massed conditioning) generates robust initial memory that decays over a few days, whereas introducing a rest between cycles (spaced conditioning) generates memory that persists for 4–7 days and is dependent upon normal protein synthesis and Creb (cyclic AMP response element-binding protein) activity at the time of conditioning (Tully et al., 1994; Perazzona et al., 2004; Yu et al., 2006). The different temporal forms of memory generated by different conditioning protocols are mechanistically distinct and must therefore generate distinct cellular memory traces that underlie the conditioned behavior.
Functional optical imaging has provided important insights into the neuroanatomy of cellular memory traces in the Drosophila brain that involve an increase or decrease of calcium influx into specific neurons, or increased synaptic transmission, in response to the conditioned odor after training (Yu et al., 2004, 2005, 2006; Liu and Davis, 2009). The memory traces observed are strong correlates of behavioral memory: they are generated only by conditioning protocols that produce behavioral performance gains and insults that disrupt behavioral memory also disrupt the memory traces. These traces include a short-lived memory trace in the antennal lobe (Yu et al., 2004), a middle-term and branch-specific memory trace in the dorsal paired medial (DPM) neurons (Yu et al., 2005), and a long-term memory trace that forms in the α branch of the α/β mushroom body (MB) neurons (Yu et al., 2006).
Many studies have identified the MBs as crucial for olfactory associative learning and memory in insects (for review see Davis, 2005). Yet, the α/β MB neurons comprise only one of three distinct classes of MB intrinsic neurons (α/β, α′/β′ and γ) whose axons extend to form five lobes of neuropil (α, β, α′, β′ and γ). The functional diversity of the different MB neurons has long been recognized based on differential expression of gene and protein markers (Yang et al., 1995; Crittenden et al., 1998) and behavioral transgenic rescue experiments that target expression of rescuing transgenes in specific neuronal subsets (Zars et al., 2000; McGuire et al., 2003). Here, we describe the properties of a recently discovered memory trace that forms in the γ neurons. This memory trace is similar to the previously described α/β MB memory trace in that it is detected as increased calcium influx into the MB axons and forms only after multiple, spaced conditioning trials. However, it is distinct with respect to its onset and persistence, forming by 18 hr after conditioning and being detectable up to 48 hr. Both memory traces mechanistically employ Drosophila Creb and calcium-calmodulin dependent protein kinase II (CaMKII).
Flies were cultured on standard medium at room temperature and transferred overnight to a 25°C incubator before training. Flies carrying Uas transgenes containing the Drosophila Creb2-b repressor coding region (Uas-dCreb2-b, [Yu et al., 2006]), CaMKII hairpin RNA (Uas-CaMKIIhpn, [Ashraf et al., 2006]) or G-CaMP (Uas-G-CaMP, [Wang et al., 2003; Colbran and Brown, 2004]) were employed along with the 1471-Gal4 (expression pattern described in [Isabel et al., 2004]), and the c739-Gal4 driver (c739-Gal4 flies were from K. Kaiser, Division of Molecular Genetics, University of Glasgow, Scotland). The w(CS10) flies (CS flies carrying the w1118 mutation) served as a wild-type control in certain experiments.
Except for the data presented in Figure 5G–I, olfactory learning was assayed using a modified olfactory classical conditioning procedure that allows the performance gains occurring from the pairing of an odor with electric shock to be measured relative to naive flies (Yu et al., 2006). A mixed population of male and female flies was exposed to two odors in succession, one odor (the conditioned stimulus, CS+) paired with electric shock pulses (unconditioned stimulus; US) followed by a counter odor (the CS−) without electric shock during the training phase. After training, the flies were incubated at 25°C for the indicated times (3, 9, 18, 24, 48, or 96 hr) before testing. For testing in the T-maze, separate groups of animals were tested during each time point with each group only being tested once. During testing, the flies were presented with both odors in a T-maze and their avoidance of the CS+ was quantified. A corresponding naïve control for each trained group underwent all of the manipulations as the trained flies except that they were not administered odor or electric shock. The Performance Indices were then calculated for both the naïve and trained group and the ΔP.I. was obtained by subtracting the naïve score from the score of the corresponding trained group. In all cases, only experiments where the naïve flies exhibited naïve performance scores that were not significantly different from zero (t-test) were used. This modification allowed us to make meaningful comparisons among groups since performance indices were normalized to naïve performance (Yu et al., 2006). In addition, the assay allowed us to obtain an index of the performance gains due to conditioning with each specific odor as the CS+, so that these gains could be compared to the results obtained after optically imaging individual flies. Data for Figure 5G were obtained using the more traditional assay (Tully and Quinn, 1985) in which two groups of flies are trained to two different odors as the CS+ (Oct and Ben, for instance) and their performance scores subsequently averaged. For assaying long-term memory, multiple spaced training trials were performed, which are required to induce a longer-lasting memory (Tully et al., 1994). We used 5X spaced training to generate long-term memory, which persists for more than 4 days. Spaced training was performed with an interval of 15 min between each training cycle whereas massed training was performed with no intertrial interval. In all behavioral experiments, the performance indices were subjected to a one-way ANOVA with genotype as the main effect, followed by Post-hoc analysis with the Bonferroni/Dunn test unless otherwise indicated. Error bars are the standard error of the mean. Prior studies have shown that behavioral performance indices are normally distributed. (Tully et al., 1994).
We performed functional imaging procedures according to previously described protocols (Yu et al., 2004, 2005, 2006). Flies containing both a Gal4 driver and Uas-G-CaMP with or without an additional transgene of interest (Uas-dCreb2-b or Uas-CaMKIIhpn) were separated prior to behavioral testing from the remainder of the trained flies. The bulk of the trained flies were tested for behavioral memory as described above. Those removed for functional imaging were mounted in pipette tips and their exposed heads secured to the tip opening with silicon cement. To expose the brain, a small region of cuticle was removed from the top of the head capsule and the exposed area was covered with a piece of plastic wrap. Confocal imaging was performed by mounting the flies under the 20X objective of a Leica TCS confocal microscope and imaged with a 488 nm excitation line. The emitted light was collected from 520±15 nm. Two criteria were used to ensure that the same volume of the γ MB lobe was imaged between flies. First, the complete medial-lateral extent of the γ lobe needed to be visible in the baseline image for functional imaging to continue. If not, the fly was discarded and another prepared. Second, this volume was scanned in the z-plane to find the most intense focal plane, which we believe occurred when the focus was centered on the midpoint of the γ lobe in the dorsal/ventral axis. Odorants were spread on a small piece of filter paper inside a syringe barrel that was placed in line with the pressurized air flowing at a rate of 100 mls per min. Concentrated odorants were diluted in mineral oil. Odorant delivery was accomplished using a 3-way teflon valve under the control of a programmable timer, such that fresh air could be delivered to the animals for a determined period with an instantaneous switch to odor-laced air without altering the overall flow rate. Electric shock pulses were applied to the fly’s abdomen. A total of 12 pulses of electric shock at 90V were delivered with each shock lasting 1.25 sec. Conditioned flies were collected after training and tested at 3, 9, 18, 24 and 48 hours after training and tested for calcium influx into the MB axons when the CS+ and CS-odors were delivered at 5-minute intervals.
Images were acquired at ~5 frames/per second at a resolution of 256 × 256 pixels followed by image data analysis as previously described (Yu et al., 2004, 2005, 2006). Regions of interest were circumscribed and a pseudo-color image of the %ΔF/Fo ratio produced. The value Fo was calculated for each pixel within the region of interest as the fluorescence prior to odor application as averaged over 5 successive frames. The value ΔF was calculated for each pixel within the region of interest as the difference between the maximum average intensity during the 3 sec odor application for 5 successive frames and Fo.
Previously, we showed that the axons of the α/β MB neurons responded with calcium influx following odor delivery or exposure to 90V shock pulses to the fly (Yu et al., 2006). A memory trace forms in these neurons between 3 and 9 hr after spaced conditioning and exhibits axon branch specificity, forming only in the α axon branch of the α/β MB neurons. In addition, we showed that the memory trace is dependent on normal protein synthesis at the time of conditioning and the activity of the transcription factor, Creb. In this study, we first tested whether γ MB axons would respond similarly to odors delivered to the antennae and electric shock pulses delivered to the fly’s abdomen. Flies carrying the calcium reporter transgene Uas-G-CaMP expressed by 1471-Gal4 were prepared for in vivo functional imaging of the brain by mounting them stably under a laser-scanning confocal microscope to detect basal fluorescence and the resulting change in fluorescence following the application of odor or electric shock. The Gal4 driver, 1471-Gal4, is an enhancer detector element that drives expression specifically in the γ MB neurons (Isabel et al., 2004; Aso et al., 2009). The cell bodies of the γ MB neurons reside in the dorsal and posterior cellular cortex. In contrast to the branched α/β and α′/β′ neurons, each γ neuron possesses an unbranched axon (but see Discussion) that projects to the anterior face of the brain as a bundled peduncle with the axons of other MB neurons. The axons of the γ MB neurons turn medially in the anterior brain to form the γ lobe, which is oriented mediolaterally in the head capsule and horizontally relative to the long axis of the fly. We collected imaging data across time at the depth of the Drosophila brain at which the γ neurons are visible in order to visualize calcium influx into the axons of the γ MB neurons (Figure 1A,B).
Stimulation of the flies with the odors 3-octanol (Oct) or benzaldehyde (Ben) elicited significant, reproducible and transient increases in calcium content in axons of the γ MB neurons (Figure 1A–C). Figure 1A illustrates the transient response that is concurrent with the odor presentation. Figure 1B illustrates a false color representation of the percent change in fluorescence that occurs following odor stimulation in the axons of the γ MB neurons. The group data show that the change in fluorescence is measurable and reproducible with amplitudes typically between 6 and 7% and statistically significant (Wilcoxon matched pairs test) for both odors compared with zero (Figure 1C). Similarly, application to the fly’s abdomen of electric shock pulses of the same intensity, duration, and frequency as those used for behavioral conditioning (Roman and Davis, 2001) produced significant calcium increases in the axons of γ MB neurons (Figure 1D,E). These data therefore show that γ MB neurons respond with calcium increases as detected by fluorescence changes in G-CaMP to odor or electric shock stimuli delivered to the fly.
Flies carrying one copy of the calcium reporter G-CaMP plus a copy of 1471-Gal4, a Gal4 driver that promote expression in the γ MB neurons, were trained to associate an odor with electric shock using three experimental protocols (Figure 2A): (a) a single cycle conditioning protocol (1x forward), (b) a massed conditioning protocol of 5 consecutive single conditioning cycles (5x massed forward), and (c) a spaced conditioning protocol consisting of 5 single conditioning cycles with a 15 min intertrial interval (5x spaced forward). Each conditioning protocol was likewise performed in a backward fashion for control groups, wherein the CS+ odor was presented after the onset of the electric shock stimuli. Backward conditioning protocols generally fail to produce the robust excitatory behavioral conditioning that occurs with forward conditioning (Rescorla, 1988). In addition, each of the two odorant used, 3-octanol (Oct) and benzaldehyde (Ben), were used as the CS+ in one set of experiments and as the CS− in the other. Each group of flies that received olfactory conditioning was partnered with a corresponding group of naïve flies that were passed through the same manipulations except that they did not receive the associative conditioning of odor with electric shock stimuli. The performance of the flies due to associative conditioning with the CS+ odor was then computed as a Δ Performance Index, (ΔPI, see Materials and Methods) by subtracting the performance of each naïve group from its corresponding conditioned group (Yu et al., 2006).
Flies conditioned using the spaced forward conditioning protocol exhibited a significantly higher level of associative memory when tested at 24 hr after conditioning than flies conditioned using any of the backward conditioning protocols (Figure 2B). This was observed for flies trained with either Oct or Ben as the CS+ odor. Flies conditioned using the 1x forward conditioning protocol and Oct as the CS+ exhibited associative memory indistinguishable from flies conditioned using the 5x massed forward conditioning protocol. Previous experiments have shown a significant difference in 24-hr performance of single- and mass-trained flies if 10 conditioning cycles are employed (Tully et al., 1994). Unexpectedly, 1471-Gal4; UAS-G-CaMP flies receiving 5x massed forward conditioning using Ben as the CS+ and Oct as the CS− failed to exhibit memory at 24 hr. Whether this odor-specific effect is due to genetic background or the expression of G-CaMP in the neurons defined by 1471-Gal4 is unknown, although it is inconsequential to our results identifying a late-phase, long-term memory trace in the γ neurons after 5x spaced forward conditioning. In all cases, the corresponding backward conditioned group failed to show any significant performance gains. These data are consistent with prior studies that established these conditioning protocols and indicate that spaced forward conditioning induces robust 24 hr memory (Tully et al., 1994), although it should be noted that our experiments utilized 5x conditioning protocols in contrast to the 10x protocols performed in earlier experiments (Tully et al., 1994).
Following the various conditioning protocols (Figure 2A), flies from each group were removed prior to behavioral testing at 24 hr and functionally imaged for calcium responses to odor. We observed that all of the groups save one exhibited CS+ odor-responses in the γ MB neurons that were not quantitatively different from those observed typically for naïve flies (ΔF/F that is in the 6–7% range as shown in Figure 1C), and that the responses to the CS+ odor were indistinguishable from the CS− odor (Figure 3A,C). This held true for the 1x (forward and backward), 5x massed (forward and backward), and 5x spaced backward groups. For the 5x spaced forward group we detected odor-elicited calcium response nearly twice as large as any of the other groups (Figure 3A–C), and this increase was specific to the CS+ odor compared to the CS-odors (Figure 3A,C,D). Therefore, spaced forward conditioning induces cellular changes that lead to an increased calcium influx into axons of the γ MB neurons in response to the CS+ odor when measured at 24 hr after conditioning.
To delimit the time of γ MB trace formation, we assayed flies carrying both 1471-Gal4 and UAS-G-CaMP at various times after spaced forward conditioning. At 3 hr post-training, behavioral memory is robust for animals trained with either Oct or Ben as the CS+. The scores show a gradual decay for Oct as the CS+ with flies tested at 3 and 9 hr showing significantly higher performance compared to flies tested at 24 hr and beyond (Figure 4A). For Ben as the CS+, flies tested at 3 hr showed significantly higher scores compared to all of the other groups with the performance dropping abruptly by 9 hr and remaining relatively constant until 48 hr. This abrupt drop in performance is probably not due to a difference in how flies remember the two different odors (Oct vs Ben), as other groups of flies trained with Ben fail to show the abrupt performance drop (Figure 4D). Performance was lowest at 96 hr with the scores from flies trained with Oct as the CS+ being significantly lower compared to 3, 9, and 18 hr, whereas for flies trained with Ben as the CS+, the 96 hr scores were significantly lower than scores at 3, 9, and 24 hr. In contrast, the cellular memory trace formed in response to the CS+ exhibits the opposite trend, such that at 3 and 9 hr after training the response in the γ MB neurons was at the naïve level for both Ben and Oct as the CS+ (Figure 4B,C). An enhanced calcium response was observed starting at 18 hr and this persisted until at least 48 hr following conditioning. The response dropped back to naive levels for flies tested at 96 hr post-conditioning. Calcium responses to the CS− were indistinguishable from naïve responses for both odors tested.
The time window for detection of the γ MB neuron memory trace after spaced conditioning was different from that previously reported for the α branch-specific memory trace, which appears by 9 hr after training with Oct as the CS+ (Yu et al., 2006). This suggested that two distinct long-term memory traces are formed in different sets of MB neurons after spaced forward conditioning. To confirm and extend the data indicating their distinction, we examined the α/β MB neuron memory trace using Oct and Ben as the CS+ and added additional time points for assay after training (18, 48, and 96 hr). Flies carrying c739-Gal4, an enhancer detector element that drives expression from Uas-transgenes specifically in the α/β MB neurons using anonymous enhancers in the vicinity of the hr39 gene (unpublished data), were used to examine the effects on the α/β MB trace. Following 5x spaced forward conditioning, the performance of flies carrying c739-Gal4 and Uas-G-CaMP showed a gradual decrease with time for both odors tested (Figure 4D). For Oct (CS+), the performance of the flies gradually decreased across time and flies tested at 3 hr performed significantly better than flies tested at 24, 48, or 96 hr. Flies tested at all other time points (9, 24, 48 and 96 hr) demonstrated significantly different performance scores compared to one another except for the scores of flies tested at 48 hr vs 96 hr. We observed the same gradual decrement in memory for flies trained using Ben as the CS+ with the flies tested at 3 hr performing significantly better than flies tested at 24, 48, and 96 hr. Flies tested at 9 hr were not significantly different from flies tested at 24 hr but the performance score was significantly higher than flies tested at 48 and 96 hr. When the trained c739-Gal4/Uas-G-CaMP flies were assayed using functional optical imaging, we discovered that the α/β MB neuron memory trace forms by 9 hr after training and persists until at least 24 hr, but that it decays to baseline levels by 48 hr after training (Figure 4E,F). This was observed for both odors tested and there was no significant difference in response to the CS− odor among the different time points. Thus, the combined data indicate that the γ MB neurons form a memory trace that is detected between 18 and 48 hr after spaced forward conditioning, whereas the α/β MB neurons form a memory trace that is detected between 9 and 24 hr after spaced forward conditioning. The distinct time windows for the existence of the two long-term memory traces suggest that there are mechanistic differences underlying their formation and their persistence, or that there is a differential delay in the transit of the necessary information to spur the formation of the traces in the two sets of neurons.
The transcription factor Creb (cAMP response element binding protein) has been identified as being important for the formation of long-term memory (Yin et al., 1994; Kandel, 2001; Perazzona et al., 2004; Yu et al., 2006). We previously showed that expression of a Uas-dCreb2-b transgene, which encodes a repressor isoform of dCreb2, in the α/β MB neurons blocks the formation of long-term behavioral memory (Yu et al., 2006). Behavioral experiments revealed that control flies carrying the dCreb2-b transgene without the Gal4 driver have 24 hr memory after spaced forward conditioning comparable to wild-type flies. In contrast, flies additionally carrying the c739-Gal4 driver element are impaired in long-term memory but have wild-type levels of 3 min memory, confirming that expression of the repressor in the α/β neurons specifically impairs long-term memory measured at 24 hr. Functional imaging experiments revealed that the expression of the dCreb2-b repressor using the c739-Gal4 driver abolished the branch specific long-term memory trace that is normally observed with either Oct or Ben as the CS+ odor (Yu et al., 2006). A second molecule that has well-described roles for mammalian and fly synaptic plasticity and memory is CaMKII (Lisman et al., 2002; Griffith et al., 2003; Elgersma et al., 2004; Griffith, 2004). Changes in CaMKII levels (Ashraf et al., 2006) and phosphorylation state (Mehren and Griffith, 2004) have been implicated in Drosophila synaptic plasticity.
To examine the requirement for these molecules in the MB neurons for behavioral plasticity and the memory traces that form in these neurons, we expressed the dCreb2-b repressor using the γ driver 1471-Gal4. We also expressed a CaMKII hairpin RNA that has been previously shown (Ashraf et al., 2006) to attenuate long-term behavior when expressed with a CaMKII-Gal4 driver, which promotes expression broadly throughout the brain. The expression of dCreb2-b and CaMKII hairpin RNA in neurons defined by the 1471-Gal4 driver severely attenuated or abolished long-term behavioral memory following spaced forward conditioning for both CS+ odors tested compared to control flies carrying only the Gal4 driver and Uas-G-CaMP (Figure 5A). Functional imaging revealed a significant difference in the measured calcium trace for both CS+ odors 24 hr after spaced forward conditioning between flies carrying only the expressed reporter (1471-Gal4; Uas-G-CaMP) and flies that carried the expressed reporter along with Uas-dCreb2-b or Uas-CaMKIIhpn (Figure 5B, C). The response to the CS- was similar for the control flies and the dCreb2-b- and CamKIIhpn -expressing flies. Thus, the expression of a dominant negative for the transcription factor Creb or a transgene that inhibits CaMKII expression in the γ MB neurons blocks the formation of the γ MB neuron memory trace in parallel with long-term behavioral memory as measured 24 hours later.
When we expressed the CaMKII hairpin RNA in the α/β MB neurons using c739-Gal4 we observed a similar abolishment of behavioral long-term memory following spaced forward conditioning for both CS+ odors compared to control flies (Figure 5D). In addition, there was a significant difference in the measured α branch memory trace in response to the CS+ odors 24 hr after spaced forward conditioning between flies carrying only the expressed reporter (c739-Gal4; Uas-G-CaMP) and flies that carried the expressed reporter along with Uas-CaMKIIhpn (Figure 5E,F). There was no significant difference in response to the CS− among the genotypes (Figure 5E,F).
Performance at 24 hr after spaced conditioning in Drosophila is due to at least two distinct components, a protein-synthesis-dependent component and a protein-synthesis-independent component (Tully et al., 1994), the latter also being known as anesthesia-resistant memory due to its resistance to cold anesthesia presented immediately after training. Massed conditioning generates the protein-synthesis independent component of long-term memory but not the protein-synthesis-dependent component. To determine whether Creb and CaMKII are functionally important for protein-synthesis-independent long-term memory, and further relate the γ MB neuron memory trace to the distinct long-term memory components, we used mass conditioning protocols to train flies expressing either the dCreb2-b repressor or the CaMKII hairpin RNAi in the α/β MB or γ MB neurons. Flies expressing either transgene in either type of MB neuron exhibited 24 hr memory that was not significantly different from controls (Figure 5G,H). This indicates that protein-synthesis-independent memory does not require these biochemical functions, at least to the level perturbed by these disruptive reagents. Furthermore, these data provide additional evidence that the γ MB neuron memory trace is relevant only to protein-synthesis-dependent long-term memory, since the dCreb2-b repressor and the CaMKII hairpin disrupt behavioral performance after spaced conditioning in parallel with the memory trace.
The various genotypes were also tested for 3 min performance following single cycle conditioning to determine whether the transgenes impaired short-term memory. Flies expressing the dCreb2-b repressor or the CaMKII hairpin RNAi in combination with G-CaMP using the 1471-Gal4 or c739-Gal4 drivers exhibited short-term memory performance that was not significantly different from control w(CS10) flies (Figure 5I). Thus, expression of the dCreb2-b repressor or the CaMKII hairpin RNAi in either the γ or α/β MB neurons specifically impairs long-term, protein-synthesis-dependent behavioral memory
MBs play a central role in Drosophila learning and memory and there is significant data indicating that the different classes of MB neurons are biochemically, developmentally, morphologically and functionally distinct (Yang et al., 1995; Crittenden et al., 1998; Zars et al., 2000; McGuire et al., 2001, 2003; Akalal et al., 2006; Krashes et al., 2007). Here, we identify a cellular memory trace that forms in the γ MB neurons following spaced forward olfactory conditioning. This memory trace, registered as increased calcium influx to the CS+ odor after training, was not generated by other conditioning protocols including spaced backward conditioning, single-cycle conditioning in both forward and backward configurations, and massed conditioning in both forward and backward configurations. Furthermore, we demonstrate that this long-term memory trace requires the normal activity of the transcription factor Creb and CaMKII in the γ MB neurons, like long-term behavioral memory. The parallel requirements of the γ MB cellular memory trace and long-term behavioral memory for spaced forward conditioning, normal Creb activity, and intact CaMKII signaling provides strong evidence that the γ MB memory trace is intimately tied to long-term behavioral memory.
The γ MB memory trace is similar in some respects to a recently reported long-term memory trace that forms in only one axonal branch of the α/β MB neurons. Like the γ MB neuron memory trace, the memory trace that forms in α/β MB neurons requires spaced forward conditioning, normal Creb activity (Yu et al., 2006), and as shown here, intact signaling through CaMKII. The finding that Creb and CaMKII signaling in either the α/β or γ MB neurons is necessary for long-term memory shows that there is some overlap in the molecular mechanisms that underlie the formation of these long-term memory traces in these subsets of MB neurons. However, the memory traces are distinct in that the α branch-specific MB trace forms by 9 hr and persists until 24 hr after 5x spaced forward conditioning, while the γ memory trace forms by 18 hr and persists up to 48 hr after conditioning. Thus, the memory traces occur in distinct windows of time and are therefore described as a long-term memory trace (α/β) and a late-phase, long-term memory trace (γ).
The discovery of a role for the γ MB neurons in late-phase, long-term memory is surprising given the prior evidence ascribing a role for these neurons in short-term memory and the emerging evidence attributing long-term memory primarily to the α/β MB neurons. The short-term memory deficit of rutabaga (rut) mutants is partially rescued by the expression of a wild-type rut transgene with Gal4 drivers that express in the γ MB neurons (Zars et al., 2000; Akalal et al., 2006), indicating a requirement for rut gene function in these neurons for short-term memory. The neuroanatomical mapping of short-term memory formed from courtship conditioning has been attributed largely to the γ MB neurons (Joiner and Griffith, 1999). Long-term memory is disrupted in alpha lobes absent (ala) flies that are missing the vertical lobes of the MBs and consequently the α axons of the α/β MB neurons (Pascual and Preat, 2001). A long-term memory trace also forms after 5X spaced forward conditioning in the α axons of the α/β MB neurons (Yu et al., 2006), consistent with a role of the α/β neurons in long-term memory processes. Finally, prior studies using Uas-Shibirets1 (Uas-Shits) combined with c739-Gal4 or 1471-Gal4 to functionally inactivate synaptic transmission using 30°C as the restrictive temperature prior to testing after spaced forward conditioning provided evidence that neurotransmission from the α/β neurons, but not γ neurons, is required for the full expression of 24 hr behavioral memory (Isabel et al., 2004). However, we attempted to replicate these results using Uas-shits1 and 1471-Gal4 but found that this combination produced locomotor and postural defects at restrictive temperatures of ≥32°C that were not observed with the α/β driver, c739-Gal4 (Supplemental Figure 1). This is presumably due to inactivating synaptic transmission from neurons that express 1471-Gal4 and that are required for normal posture and locomotor activity. Thus, the prior experiments performed to test the role for γ MB neuron synaptic transmission in long-term behavior may have failed to sufficiently inactivate these neurons to uncover their role in long-term behavioral memory. Furthermore, recent experiments have revealed that the function of orb2 is required in the γ MB neurons for long-term courtship conditioning (Keleman et al., 2007). The combined observations along with our new results suggest that both the α branch of the α/β MB neurons and the γ MB neurons are involved in long-term memory formation.
What is the significance of having two distinct, olfactory long-term memory traces occurring in distinct time windows that form in two different sets of MB neurons? The most attractive general explanation is that behavioral memory, which is robust initially after training and decays continuously over time, is not due to a single memory trace that decays over time in one type of neuron. Rather, behavioral memory is underlain by multiple, olfactory memory traces that form in many different types of neurons within the olfactory nervous system with each memory trace forming from network interactions and through molecular mechanisms intrinsic to each type of neuron. This would provide for differential kinetics in the formation and decay of individual cellular memory traces. The idea that different temporal forms of memory are stored in different sets of neurons is not new. The traditional view of mammalian long-term memory involves hippocampal-dependent consolidation of labile short-term memory into stable, longer-lasting memory stored in the neocortex (Squire and Alvarez, 1995; McGaugh, 2000; Wittenberg and Tsien, 2002; Squire et al., 2004; Wang et al., 2006). Testing whether the Drosophila long-term memory traces form in serial or parallel requires additional tools, including but not limited to, the generation of new gene promoter systems that are independent of the UAS-Gal4 system to allow for expression of a reporter for functional imaging in some neuron types while blocking synaptic transmission from other neurons types by expressing Uas-Shibire.
The results presented here along with those in Yu et al. (2006), reveal that both Creb and CaMKII activities are required in the α/β and the γ MB neurons for normal long-term behavioral memory. The specific role for Creb in long-term memory is not surprising, although the combined results indicate that Creb activity is required in at least these two types of MB neurons for normal long-term behavioral memory. Studies in Aplysia, Drosophila, and rodents indicate that Creb is important for the conversion of short-term to long-term memory [For reviews see (Frank and Greenberg, 1994; Silva et al., 1998; Mayford and Kandel, 1999)]. In Aplysia, Creb inhibitors can selectively block long-term facilitation (LTF) but not short-term facilitation (Dash et al., 1990). In Drosophila, a repressor isoform of Drosophila Creb (dCreb2-b) has been shown to reduce the long-term performance of trained flies without affecting 3-min memory, as well as to abolish the long-term memory calcium trace normally observed in the α branch of MB α/β neurons (Yu et al., 2006). Reducing the level of Creb activity in mice also impairs certain forms of long-term memory (Bourtchuladze et al., 1994; Balschun et al., 2003). Data presented here indicate that disrupting Creb in the γ MB neurons impairs long-term behavioral memory while sparing short-term memory.
It is surprising that CaMKII has a specific role in Drosophila long-term olfactory memory generated by classical conditioning. We demonstrate here that disruption of CaMKII signaling using a CaMKII hairpin RNA in either the α/β or the γ MB neurons specifically impairs long-term behavioral memory. A prior report (Ashraf et al., 2006) also demonstrated that broad expression of the CaMKII hairpin RNA in Drosophila using a CaMKII-Gal4 driver impairs long-term behavioral memory. However, early memory after courtship conditioning requires CaMKII activity in the antennal lobes (Joiner and Griffith, 1999) and this is enhanced following expression of a constitutively activated form of CaMKII (Mehren and Griffith, 2004). Furthermore, CaMKII has a well-accepted role in the induction phase of LTP [For reviews see (Lisman et al., 2000; Griffith, 2004; Colbran and Brown, 2004)]. Further studies are required to dissect the differential roles of CaMKII in various temporal forms of memory.
Using a combination of behavioral assays, functional imaging, and transgenic disruption experiments, our combined results indicate that there are parallel and partially overlapping temporal memory traces that form in distinct sets of MB neurons and may represent protein-synthesis-dependent, long-term behavioral memory and late-phase, long-term behavioral memory after spaced olfactory conditioning in Drosophila. These differentially timed long-term memory traces in the MB neurons along with the immediate and short-lived AL projection neuron memory trace (Yu et al., 2004) and the medium-term memory trace in the DPM neurons (Yu et al., 2005) lend support to the concept that behavioral memory is the sum of multiple cellular memory traces that form in discrete subpopulations of neurons.
Expressig UAS-shits1 with 1471-Gal4 produces locomotor/postural defects.
A total of 50 flies carrying one copy of either 1471-Gal4 or c739-Gal4 and one copy of UAS-shits1 were collected in empty vials and placed in an incubator set to temperatures between 30–34°C. Control flies were w(CS10) crossed with flies carrying UAS-shits1. Photographs were taken at each time point and the number of flies located at the bottom of each vial was counted.
(A) At restrictive temperatures of 32°C and 34°C, flies expressing UAS-shits1 using the Gal4 driver 1471 showed a significant difference in the percentage of flies located at the bottom of the vials, demonstrating reduced mobility/activity compared to control flies. Error bars are the standard error of the mean. Mann-Whitney pair-wise comparisons, *p<0.05, **p<0.01, ***p<0.0001.
(B) At restrictive temperatures of 32°C and 34°C, flies expressing UAS-shits1 using the Gal4 driver c739 showed no significant difference in the percentage of flies that were located at the bottom of the vials when compared to control flies. Mann-Whitney pair-wise comparisons, p≥0.6310 at 32°C and p≥0.1735 at 34°C. n=6 for all groups. Error bars are the standard error of the mean.
Funding: Supported by grants NS19904 and NS52352 from the NINDS.
Author contributions: DBA, DY, and RLD conceived and designed the experiments. DBA performed the behavioral experiments and DY performed the imaging experiments. They also analyzed their data. DBA, DY, and RLD wrote the paper.
Competing Interests: The authors have declared that no competing interests exist.