Visualizing Long-Term Memory Formation in Two Neurons of the Drosophila Brain

Long-Term Memory (LTM): Requires the synthesis of new proteins Two Neurons: Located at the dorsal-anterior-lateral (DAL) of protocerebrum

Abstract

Disruption of protein synthesis in DAL neurons results in defects in LTM formation

Comparison of 1-day memory after spaced training is always coupled with that after massed training to confirm certain effect is LTM specific

Background

1. LTM is achieved by spaced training in Drosophila, where new proteins are synthesized, as a result of cAMP mediated regulation

2. According to past findinds, mushroom body (MB) is suggested to be the primary site in the brain that is responsible for both LTM and STM

3. Lack of visual evidence indicating de novo protein synthesis takes place in MB during LTM formation

I. Visualizing de novo protein synthesis in targeted neurons

KAEDE: GFP, which irreversibly changes to RFP upon UV radiation ---> used to measure de novo protein synthesis 1. 2. 3. 4. per-Gal4>UAS-kaede Photoconversion of KAEDE every 4 hours Exhibited diurnal cycle In parallel to the oscillation of per RNA

I. Visualizing de novo protein synthesis in targeted neurons

Cycloheximide (CXM): protein synthesis inhibitor

1. CXM was fed to the flies 2. Significant reduction of KAEDE (Green Fluorescence) synthesis 3. Red Fluorescence remained constant
4. KAEDE conversion is non-spontaneous and irreversible

I. Visualizing de novo protein synthesis in targeted neurons

RICIN: cytotoxic protein that inactivates eukaryotic ribosomes ---> used to prevent de novo protein synthesis under the control of temperature OK107-Gal4>UAS-ricinCS (OK107: nearly all MB neurons; CS = Cold Sensitive) At 30C: Active RICIN, causing hydrolization of ribosome and thus leading to suppression of protein synthesis and deformation of MB At 18C: Inactive RICIN, no defects caused in MB Activity of RICINCS was confirmed (i.e. strictly regulated by temperature) again by following KAEDE synthesis in per-Gal4>UAS-kaede/UAS-rincinCS construct Similar inhibitory effect of RICIN was observed with the same constructs but using OK07-Gal4 and Cha-Gal4 instead

II. Behavioral screen for neurons involved in protein synthesis-dependent memory formation

1. Expression of ricin during the 24 hours after training impairs 1-day memory 2. Only affected spaced training 3. Feeding of CXM did not change the outcome 4. All of above suggest impairment is specific to LTM

II. Behavioral screen for neurons involved in protein synthesis-dependent memory formation

1. Past evidence suggest all learning processes are either directly or indirectly mediated via MB 2. However, inhibition of protein synthesis in a subsets of neurons which primary include MB neurons did not result in defects in 1-day memory retention

3. Inhibition of protein synthesis in acetylcholine-producing neurons, with the exception of those in MB (achieved by Cha-Gal4 and MB-Gal80>UASrincinCS), resulted in 1-day memory impairment after spaced training

4. In contrast to original thoughs, the above results indicate de novo protein synthesis during LTM formation takes place in Cha-Gal4 neurons outside MB

II. Behavioral screen for neurons involved in protein synthesis-dependent memory formation

Screen of RICINCS expression pattern indicates LTM formation is dependent on the de novo protein synthesis in the targeted neurons of 8 Gal4 strains, which are star marked in the figure, alone with ChaGal4, where some of them covering the neurons from MB

III. Identification of individual neurons with protein synthesis during LTM formation

1. PER, but not other circadian, protein is necessary for LTM formation since 1-day memory is impaired after spaced training but not massed training 2. Blocking of neurotransmission with UAS-shits in per neurons causes defects in retrieving of 1-day memory after spaced training 3. Induction of RINCIN in per neurons also caused impairment of 1-day memory 4. Protein synthesis is required only in the first 12 hours of LTM formation 5. Activation of RINCIN in MB neurons did not cause the same problem

III. Identification of individual neurons with protein synthesis during LTM formation Analysis of the previously mentioned 9 Gal4 strains revealed two DAL neurons that are likely to be the mediators of protein synthesis during LTM formation as the expression patterns of these neurons are overlapped among these Gal4 strains. Also they both express N-methyl-D-asparatate receptors, which are required for LTM formation

III. Identification of individual neurons with protein synthesis during LTM formation 1. Limiting the expression of RICIN in Cha-Gal4 and per-Gal4 by inducing cryGal80 did not lead to impairment of 1-day memory 2. Each of the Gal4>UAS-ricinCS itself can induce 1-day memory impairment itself 3. Neurons whose activities are governed by all of the three genes together are important for LTM formation, which includes the two DAL neurons

cry per cha

III. Identification of individual neurons with protein synthesis during LTM formation

1. By using another three Gal4 drivers (E0946, G0338 and G0431) to minimize the number of neurons being affected, but still including the two DAL neurons, impairment of 1-day memory was detected when RICIN was expressed during the first 12 hours after spaced training

2. Retrieval, but not formation, of LTM failed as neurotransmission was blocked from these two neurons by inducing UAS-shiTS during the test trial 1 day after spaced training

3. The fact that blocking of neurotransmission from the two neurons only led to impairment of LTM retrieval suggests there are neurons, other than those covered by the three Gal4 drivers, involved in protein synthesis for LTM formation

III. Identification of individual neurons with protein synthesis during LTM formation

1. The combination of cry-Gal80 and cer-Gal4 which limited the spatial expression of RICIN immediately after spaced training for 24 hours leads to impairment of 1-day memory, indicating the LTM formation is also dependent on these cer neurons

2. Same expression subtraction using other Gal4 driver did not yield the same result, and therefore limiting the candidates to cer neurons

3. So far, target regulatory neurons are limited down to the intersection of the expression patterns of per, cha, cry and cer

IV. DAL axons are structurally connected with MB calyx 1. The exploitation of Dscam-GFP as dendritic marker and synaptotagminGFP as axon marker helped locating the projection pattern of the DAL neurons, whose connectives span the superior dorsofrontal protocerebrum, dorsolateral protocerebrum and inferior dorsofrontal protocerebrum

2. As well as pointing out that DAL neurons might make contact with MB calyx at K5 region
3. This is confirmed by doing GFP reconstitution across synaptic partners labeling where DAL neurons (G0431-Gal4) are shown to make close contact with the neurons (L5275-LexA) of MB

V. Identification of newly synthesized proteins in DAL neurons during LTM formation 1. Immunostaining revealed key proteins expressed in DAL neurons: DDC, PER, dNR1, dNR2, CaMKII, TEQ 2. Induction of UAS-perPAS-IR in four different Gal4 backgrounds (per, Ddc, G0338 and G0431) led to 1-day memory impairment after spaced training

3. Using RNAi to suppress synthesis of the above detected proteins in G0431Gal4 background also showed the same result
4. Suppression of RNAi effect during development under the control of tubGal80ts, while inhibiting the production of corresponding protein at adult stage showed same impairment in 1-day memory, indicating phenotypes observed are not caused by misdevelopment

VI. CREB2 activity in DAL neurons, but not MB neurons, is required for LTM formation 1. Repressing of CREB2 in MB neurons achieved by c739-Gal4>UAS-dcreb2-b caused impairment in 1-day memory after spaced training as previously suggested 2. Learning process was inhibited only when CREB2 was co-expressed in two additional MB driver lines: OK107-Gal4 and c247-Gal4. However, morphological analysis of each separate case suggested this might be caused by developmental defects as formation of certain parts of MB was affected 3. Therefore, expression of CREB2 repressor was inhibited by tub-Gal80ts until adult stage in the same Gal4 background. No defects in memory were observed in these flies, while MB was still severely damaged 4. Same concept was applied but using Cha-Gal4 as driver led to impairment of 1day memory after spaced training 5. The above results suggests target neurons are not MB specific but covered by Cha 6. Antagonizing of CREB2 activity (UAS-dcreb2-b) and down regulation of CREB2 expression in DAL neurons (RNAi) both led to impairment of 1-day memory after spaced training

VII. Visualizing transcriptional activity in identified neurons during LTM formation 1. LTM formation is dependent on adult-specific expression of the following eight genes: per, dNR1, dNR2, CaMKII, Teq, Ddc, Trh and dcreb2 in Dal neurons

2. The use of RICINTS and KAEDE in CaMKII-Gla4 background revealed that disruption of protein synthesis in MB did not affect memory consolidated from spaced training, while constant protein synthesis driven by CaMKII-Gla4 in DAL, but not MB, is required for spaced training
3. KAEDE synthesis driven by per-Gal4 was also elevated in only DAL neurons after spaced training

VII. Visualizing transcriptional activity in identified neurons during LTM formation

1. Elevation of CaMKII and per activities mainly occurred in the first 8 hours after spaced training 2. Suppression of CREB2 repressor in adult stage by Gal80TS led to reduction of CaMKII and per activities after spaced training 3. On the other hand, spaced training did not lead to upregulation of cry, Ddc and Trh, as verified by KAEDE synthesis analysis

Summary

de novo protein synthesis in DAL neurons in protocerebrum are required for LTM formation, but not during acquisition and consolidation