Long-Term Memory Formation in Drosophila Depends on the 3′UTR of CPEB Gene orb2

Activation of local translation in neurites in response to stimulation is an important step in the formation of long-term memory (LTM). CPEB proteins are a family of translation factors involved in LTM formation. The Drosophila CPEB protein Orb2 plays an important role in the development and function of the nervous system. Mutations of the coding region of the orb2 gene have previously been shown to impair LTM formation. We found that a deletion of the 3’UTR of the orb2 gene similarly results in loss of LTM in Drosophila. As a result of the deletion, the content of the Orb2 protein remained the same in the neuron soma, but significantly decreased in synapses. Using RNA immunoprecipitation followed by high-throughput sequencing, we detected more than 6000 potential Orb2 mRNA targets expressed in the Drosophila brain. Importantly, deletion of the 3′UTR of orb2 mRNA also affected the localization of the Csp, Pyd, and Eya proteins, which are encoded by putative mRNA targets of Orb2. Therefore, the 3′UTR of the orb2 mRNA is important for the proper localization of Orb2 and other proteins in synapses of neurons and the brain as a whole, providing a molecular basis for LTM formation.


Introduction
Synaptic plasticity is one of the key mechanisms underlying learning and memory formation. The mechanism is based on the ability of neurons to alter the strength and efficiency of signal transmission through synapses in response to stimulation [1]. Transmission of the signal can be amplified (a process of potentiation) or weakened (a process of depression), and changes in signal transmission can be of short duration or can persist for a longer period [2]. Short-term alterations in synaptic function and the formation of short-term memory can be induced without protein or mRNA synthesis, while long-lasting alterations in synapse function and the formation of long-term memory (LTM) require mRNA and protein syntheses [3]. It is thought that mRNA and protein syntheses alter the functional properties of individual synapses or induce new synaptic connections. However, these changes in synaptic function are most likely not restricted to a single synapse, but rather occur throughout a network of neurons that are responsible for capturing memory [4].
One of the mechanisms that alter local synaptic functionality and/or induce de novo assembly of new synaptic junctions is based on local translation and synthesis of proteins

Fly Strains
The following Drosophila fly strains were used in this study: a strain with a 3 UTR deletion of the orb2 gene (orb2 R [26]), which was obtained using the CRISPR/Cas9 system; the parental Cas9-expressing fly strain 51324 (w [1118]; PBacVK00027) from the Drosophila Bloomington stock center (DBSC, USA) was used as a control to measure the mRNA and protein levels and to perform whole mount staining; learning and memory formation are similar in 51324 and WT Canton S (CS) flies, and the CS strain was therefore used as a control strain in behavioral experiments; a strain that expressed the Orb2-GFP fusion instead of the WT orb2 allele (a kindly gift from Kristina Keleman) was used in immunoprecipitation experiments.

Locomotion Assays
In a climbing assay, 10 flies from each strain (WT and orb2 R ) were placed in a standard empty Drosophila-culture vial with a marked 6 cm distance from the bottom. Flies were knocked to the bottom of vial with series of 10 strong hits on the table. The number of flies that climbed 6 cm in 10 s was counted. A total of 100 flies of each genotype were tested and the percentage of flies that climbed up in the established time was calculated. In a bang-sensitivity assay, 10 flies were placed in an empty plastic vial and shaken on an microspin FV-2400 (500 g) for 10 s, and those that returned to a standing position in 10 and 20 s were counted. A total of 100 flies of each genotype were tested and the percentage of flies that climbed up in the established time was calculated. In a heat-shock assay, flies grown at room temperature were placed in an empty preheated vial and heat shocked at 37 • C in cultural thermostat for 10 min. The number of flies that were still standing after a 10-min heat shock was counted.

Behavioral Assay
A courtship-suppression paradigm was used to check memory formation [27]. Flies were raised on standard Drosophila medium (cornmeal, agar, molasses, yeast, and nipagin) at 25 • C and 60-70% humidity with a 12-h light-dark cycle. Males of the strain under study were collected without anesthesia 0-10 h after eclosion and kept separately in small food vials until the tests were carried out. Training was performed using 5-day-old CS females fertilized 1 day before in the experimental chamber (15 mm in diameter, 5 mm in height). Learning and testing were performed at an age of 5 days. For training, males without experience in courtship behavior were placed in a vial with a fertilized female for 30 min. For LTM formation, male and female flies were kept in the same vial for 5 h.
Short-term memory was assessed immediately after training, and mid-term memory, after 180 min. LTM was assessed 2 days after training, using newly fertilized 5-day-old CS females. Males without courtship behavior experience were used as a control. In each group, 20 males were tested. Behavior of males was assessed immediately, 3 h, and 2 days after learning (30-min training). The time spent in courtship (orientation, following, wing vibration, licking and attempted copulation) was recorded for 300 s when tested males were placed in an experimental chamber with a previously mated female. The courtship index (percentage of time spent in courtship) was calculated for each male. A courtship index (CI) was calculated as a ratio of the male courtship time to the entire observation time [27]. A learning index (LI) was calculated as LI = [(CIn − CIexp)/CIn] × 100%, where CIn is the average CI of two independent samples of males without courtship behavior experience, and CIexp is the average CI of two independent samples of males after training. Statistical comparisons of behavioral data were made by a randomization test, by directly calculating the probability of rejection of the null hypothesis α R . A sampled randomization test with 10,000 permutations was used. The null hypothesis was rejected at α R < 0.05. A two-sided test was used to compare the courtship indices. A one-sided test was used to compare learning indices.

Antibodies
We used the following antibodies against Orb2: 4G8 (1:50 dilution) for Western blotting and 2D11 for whole brain staining (1:100 dilution). The antibodies were deposited in the Developmental Studies Hybridoma Bank (DSHB) by P. Schedl. Anti-GFP antibodies DSHB-GFP-2G6, 4C9, 8H11, and 12A6 were deposited in the DSHB by DSHB investigators (DSHB Hybridoma Product DSHB-GFP-2G6, DSHB-GFP-4C9, DSHB-GFP-8H11 and DSHB-GFP-12A6). An antibody against the eyes absent (Eya) protein (10H6) was deposited in the DSHB by S. Benzer and N.M. Bonini. The DCSP-1 (ab49) antibody against the cysteinestring protein (CSP) was deposited in the DSHB by E. Buchner and A. Hofbauer. An antipolychaetoid (PYD1) antibody was deposited in the DSHB by A.S. Fanning. Antibodies were used at a 1:500 dilution in Western blotting and a 1:200 dilution in whole brain staining. The 3C11 antibody against Synapsin was deposited in the DSHB by B.R.E. Klagge and colleagues and used at 1:200 dilution. The ADL67.10 antibody against the lamina protein (Adl) was deposited in the DSHB by P.A. Fisher and was used at a 1:1000 dilution in Western blotting. Alexa 488-conjugated goat anti-mouse IgG antibody (Invitrogen, USA, catalog number A-11001) was used as a secondary antibody at a 1:1500 dilution. HRP-conjugated goat anti-mouse IgG antibody (Jackson ImmunoResearch, UK, catalog number 115-035-174) was used as a secondary antibody in Western blotting at 1:2000.

RNA Immunoprecipitation
To collect an appropriate amount of fly heads, about 10 g of flies of the Orb2-GFP strain were frozen in liquid nitrogen, vortexed, and sifted through two 800-and 380-µm sieves to separate heads from bodies. All the following steps were carried out at 4 • C. Using a Teflon pestle, heads were homogenized in a polysome buffer (10 mM HEPES, pH 7.4, 100 mM KCl, 5 mM MgCl 2 , 0.5% NP-40, 1 mM DTT) supplemented with a protease-inhibitor cocktail (Roche, catalog number 11836145001) and a ribonuclease inhibitor (Ribolock, Thermo Fisher Scientific, catalog number EO0382). The lysate was centrifuged twice at 1000× g for 5 min to pellet the debris. Of the final volume (2 mL), 100 µL of the clarified lysate was used for RNA isolation and 10 µL, for Western blot analysis. The samples were designated as Input.
Further, the initial lysate was incubated with an anti-GFP antibody mix (4C9, 8H11, 2G6, and 12A6 with a concentration of 30 ng/µL each) for 24 h at +4 • C. A negative control sample contained immunoglobulin G (IgG) from a pre-immune serum. After incubation, 10 µL of protein G Sepharose (Invitrogen, catalog number 101241) was added to each sample, incubated for 2 h at room temperature, and washed three times with the polysome buffer for 15 min. The resulting Sepharose was resuspended in 60 µL of the polysome buffer; 2 µL was used in Western blot analysis, and the rest of the suspension was used for RNA isolation to assess the quality of the precipitation reaction.

RNA Isolation, Reverse Transcription, qPCR
To isolate RNA, we used Trizol reagent according to the manufacturer's recommendations (Molecular Research Center, USA, catalog number TR118). The final RNA concentration was measured with a RNA assay kit (Molecular probes, USA, catalog number Q32852). Equal amounts of Input RNA, immunoprecipitate (IP) obtained with anti-GFP antibodies, and mock IP were used to synthesize cDNA with a RevertAid kit (Invitrogene). The resulting cDNA was used for qPCR in three replicates of each sample. To assess the quality of the IP reaction, the enrichment of Orb2 target mRNAs was measured using primers act5c, Orb2, tub56D (positive control), 28S, and RplP0 (negative control) RNAs.
Five independent reaction sets were tested, and the most enriched IP sample obtained with anti-GFP antibodies was used for high-throughput sequencing.
To measure the orb2 mRNA level, extracts from 20 fly heads of the 51324 and Orb2 R strains were used to isolate RNA; cDNA was synthesized, and qPCR was carried out with primers to Orb2 and gapdh2 as a reference gene. We performed five independent biological replicates and used Student's t-test (p < 0.05).

Next-Generation Sequencing Data Analysis
RNA isolation and preparation of cDNA libraries from Input, IP with antibodies, and mock IP were performed using a NEB E7770S kit (New England Biolabs, USA) according to the manufacturer's recommendations. The cDNA libraries were sequenced on the Illumina 2000 platform. Reads were mapped to the Drosophila genome using TopHat software [28]. Next, the number of reads that mapped to exons was calculated using bedtools followed with htseq-count [29] calculation of reads for each gene; this value was used to calculate CPM (counts per million). Raw reads were deposited in the BioProject database with accession No. PRJNA899720 (SRA accessions are SRR22244251, SRR22245477, and SRR22254174). GO-term analysis was performed by the service PANTHER [30]. Enrichment by published gene sets was performed using the service FlyMine [31].

Crude Synaptosome Preparation
For each genotype, 20 fly heads were homogenized in 50 µL of 0.32 M sucrose buffer (5 mM HEPES, pH 7.4, 145 mM NaCl, 5 mM KCl, 2 mM CaCl 2 , 1 mM MgCl 2 , 5 mM glucose) supplemented with a protease inhibitor cocktail. Crude synaptosome preparations were obtained essentially as described in [32] with an extra centrifugation step added after collecting the first pellet.

Semiquantitative Western Blot Analysis
To assess the distribution of the proteins of interest in Drosophila nervous cells, protein concentrations in a cell soma fraction and a synaptosome fraction were measured using a Qubit protein assay kit (Molecular probes, catalog number Q33211). Equal amounts of each sample were diluted by factors of 2, 4, and 6 and loaded onto precast stain-free PAAG (Biorad, USA, cat. number 456-8095); bands were visualized in a Chemidoc system to quantify the protein concentration. Chemiluminescence was measured with the same system.

Whole Mount Immunostaining
The brains of 5-day-old flies were dissected, washed twice with ice-cold PBS supplemented with 0.05% Triton-X100 (PBSt), fixed with methanol for 20 min, and treated with acetone for 5 min for permeabilization. After two wash steps, the brains were incubated in 4% skim milk with 1% BSA overnight (+4 • C). A primary antibody was added, and incubation continued for 24 h with rotation (12 rpm) at +4 • C. After 3× washing with PBSt for 30 min each, a secondary antibody was added, and incubation was continued for 2 h at room temperature with rotation (12 rpm). After washing with PBSt for 1 h, the samples were placed in a mounting medium with DAPI (Vector Labs, USA). Images were obtained using a Leica STELLARIS 5 confocal microscope. The stained samples were scanned in multichannel mode using a 40× oil objective with numerical aperture 1.3. Images with a frame size of 2048 × 2048 pixels and a z resolution of 1 µm were taken at a scan speed of 400. At least twenty brain images were obtained for each genotype, and the selected images were processed using LazX software (Leica Application Suite X, version 3.7.25997.6., Leica Microsystems, Germany)

Deletion of orb2 3'UTR Does Not Affect Locomotion
The Orb2 protein is widely distributed in the Drosophila brain, including the ellipsoid and fan-shaped body structures [17], which are responsible for walking, assessing distance to an object, and orientation in space. In previous studies, a piggyBac transposon inserted into the first intron of the orb2 gene (orb2 6090 ) was found to significantly decrease the Orb2 protein levels and to impair the ability of a fly to orient in space and to walk [17]. To assess the role of the extended orb2 3 UTRs in motor activity, we used the orb2 R -deletion mutant generated previously. As described above, the mutation deletes a 4.5-kb DNA sequence encoding the larger orb2B mRNA isoforms and leaves the orb2B 3 UTR sequence that includes only five CPEs [26] ( Figure 1A). Flies were tested for the abilities to climb after a strong impact (climbing assay), to restore orientation in space after intense shaking (vortex assay), and to stand up after heat shock.

Long-Term, but Not Short-or Mid-Term, Memory Formation Is Disrupted in the orb2 R Mutant
To investigate the effects of orb2 R on learning and memory retention, we used conditioned courtship suppression paradigm in males, which was tested after interact with non-receptive females [27]. The training session was 30 min in the short-and m term memory assays. The behavior of males was assessed immediately after training check short-term memory and 3 h later to check mid-term memory. We found no sign cant difference in courtship index between orb2 R and WT males in the case of short-a mid-term memory ( Figure 1D). The results are consistent with published data that a co plete deletion of the orb2 gene, a deletion of Orb2A (orb2 ΔA ), or a deletion of the polyg tamine domain of the Orb2 protein (orb2 ΔQ ) have no effect on the formation of short mid-term memory [19].
To assess LTM formation, male flies were first trained for 5 h and the courtship in was measured two days later. As shown in Figure 1E, orb2 R males had a pronounced d show the CPE sequences. Different colors of lines correspond to different CPE sequences (see [26] for details). The red dashed lines show the deleted part of the 3 UTR. (B,C) The percentage of WT and orb2 R flies that could climb after strong shaking (climbing assay, B) and restore the normal body position after intense shaking (vortex assay, C). The standard deviation is shown, t-test, n = 100; nsnonsignificant (D,E) Learning indices of WT and orb2 R flies as calculated for short-, mid-(D), and long-term (E) memory by assessing the suppression of male sexual behavior after interaction with a previously fertilized female. Assessment of differences between learning indices based on calculated courtship indices (two-side randomization test, p < 0.05, n = 20).
Assessment of physical activity using the climbing assay showed an insignificant decrease in the speed of movement of orb2 R flies after impact ( Figure 1B). We did not find significant differences in the ability of orb2 R flies to restore the body position after vortexing ( Figure 1C). Likewise, the mutant flies were as resistant as WT flies to a brief heat treatment (10 min at 37 • C). Thus, orb2 R flies clearly differed from flies carrying the null allele orb2 6090 piggyBac in these assays.

Long-Term, but Not Short-or Mid-Term, Memory Formation Is Disrupted in the orb2 R Mutant
To investigate the effects of orb2 R on learning and memory retention, we used the conditioned courtship suppression paradigm in males, which was tested after interacting with non-receptive females [27]. The training session was 30 min in the short-and mid-term memory assays. The behavior of males was assessed immediately after training to check short-term memory and 3 h later to check mid-term memory. We found no significant difference in courtship index between orb2 R and WT males in the case of short-and midterm memory ( Figure 1D). The results are consistent with published data that a complete deletion of the orb2 gene, a deletion of Orb2A (orb2 ∆A ), or a deletion of the polyglutamine domain of the Orb2 protein (orb2 ∆Q ) have no effect on the formation of short-or mid-term memory [19].
To assess LTM formation, male flies were first trained for 5 h and the courtship index was measured two days later. As shown in Figure 1E, orb2 R males had a pronounced deficit in LTM; their courtship index was comparable to that of naïve untrained WT males. Thus, the effects of the 3 UTR deletion are similar to those observed for the orb2 ∆A and orb2 ∆Q mutations [7,19]: no alteration occurred in short-term or mid-term memory, but LTM was strongly impaired.

Orb2 Protein Levels in the Brain Are Altered in orb2 R
While transcription of the orb2 gene is not likely to be affected by the 3 UTR deletion, there could be alterations in the yield of fully processed orb2B mRNAs or changes in their stability. To test these possibilities, we used RT-qPCR with primers corresponding to an exon common to all orb2 mRNA isoforms. The gapdh2 mRNA was used as a reference. As shown in Figure 2A, the level of the orb2 mRNA extracted from the heads of 5-day-old homozygous orb2 R flies did not differ from that of WT flies. stability. To test these possibilities, we used RT-qPCR with primers corresponding to an exon common to all orb2 mRNA isoforms. The gapdh2 mRNA was used as a reference. As shown in Figure 2A, the level of the orb2 mRNA extracted from the heads of 5-day-old homozygous orb2 R flies did not differ from that of WT flies. The levels of the Orb2B protein were examined next. Note that Orb2A is not detectable in Western blots of fly heads [17]. Total lysates from the heads of 5-day-old WT and orb2 R flies were analyzed by Western blotting with an anti-Orb2 antibody. To assess the relative amount of the Orb2B protein, we normalized the chemiluminescent signal detected by Western blotting to the total protein level in the sample. The Orb2B level in orb2 R was about 2/3 of that in WT ( Figure 2B,C). This finding suggests that the 3′UTR deletion reduces the stability or translation efficiency of the orb2B mRNAs.

Orb2 Accumulation in Synaptic Fractions Is Affected in orb2 R Mutant
A modest reduction in the amount of the Orb2 protein in head lysates seemed inconsistent with the clear deficit in LTM formation. Because 3′UTR sequences are believed to be important not only for translational regulation, but also for the mRNA localization in The levels of the Orb2B protein were examined next. Note that Orb2A is not detectable in Western blots of fly heads [17]. Total lysates from the heads of 5-day-old WT and orb2 R flies were analyzed by Western blotting with an anti-Orb2 antibody. To assess the relative amount of the Orb2B protein, we normalized the chemiluminescent signal detected by Western blotting to the total protein level in the sample. The Orb2B level in orb2 R was about 2/3 of that in WT ( Figure 2B,C). This finding suggests that the 3 UTR deletion reduces the stability or translation efficiency of the orb2B mRNAs.

Orb2 Accumulation in Synaptic Fractions Is Affected in orb2 R Mutant
A modest reduction in the amount of the Orb2 protein in head lysates seemed inconsistent with the clear deficit in LTM formation. Because 3 UTR sequences are believed to be important not only for translational regulation, but also for the mRNA localization in different subcellular compartments, we wondered if the distribution of the Orb2 protein within neurons is altered in mutant flies. To explore this possibility, we fractionated the total-head homogenate into neuronal cell body and synaptic fractions using procedures described previously [32]. A total-head homogenate (L) was first centrifuged through a 0.32 M sucrose solution to obtain a low-speed pellet fraction (LP) enriched in neuronal cell bodies (soma) ( Figure 3A). The supernatant was then centrifuged through a 0.8 M sucrose solution. The resulting pellet fraction (P1) contained proteins from neuronal synapses together with a mixture of cytosolic proteins, while other cytosolic proteins were left in the supernatant (S1). The pellet was resuspended in a 0.8 M sucrose buffer and centrifuged again through a 0.8 M sucrose solution. The pellet fraction (P2) contained mostly synaptic proteins. To assess the purity of the synaptic fractions, we used antibodies against Lamin of the nuclear envelope and the cysteine string protein (Csp), which is involved in synaptic transmission [33]. Western blot analysis of the fractions showed that Lamin was present in the LP fraction but absent in the final synaptic fraction P2. On the contrary, Csp was enriched in both the P1 and P2 synaptic fractions ( Figure 3A).  The Orb2B distribution through the fractions was compared between extracts prepared from WT and orb2 R flies. For this purpose, the chemiluminescent signals from anti-Orb2 antibody staining were normalized to the total protein levels for each fraction. Orb2 protein levels in soma were similar in orb2 R and WT flies ( Figure 3B). At the same time, the Orb2B protein in synapses of orb2 R flies was almost three times less abundant than in WT.
To confirm that the distribution of the Orb2 protein in neurons was altered in the orb2 R mutant, we probed the brain whole mounts of WT and orb2 R with an anti-Orb2 an- The Orb2B distribution through the fractions was compared between extracts prepared from WT and orb2 R flies. For this purpose, the chemiluminescent signals from anti-Orb2 antibody staining were normalized to the total protein levels for each fraction. Orb2 protein levels in soma were similar in orb2 R and WT flies ( Figure 3B). At the same time, the Orb2B protein in synapses of orb2 R flies was almost three times less abundant than in WT.
To confirm that the distribution of the Orb2 protein in neurons was altered in the orb2 R mutant, we probed the brain whole mounts of WT and orb2 R with an anti-Orb2 antibody. Figure 4A shows that there was a general decrease in immunofluorescent signal in the orb2 R samples. Previously, the Orb2 protein has been shown to occur both in the perinuclear space (rings around nuclei) and axons of neurons in the fly brain [19]. In agreement with this, we found Orb2 in the neuron bodies and the neuropile region in WT (marked in Figure 4A,B). In orb2 R samples, the Orb2 protein was detected predominantly in the vicinity of neuronal nuclei, forming a characteristic ring distribution ( Figure 4C). The intensity of perinuclear staining was comparable between WT and orb2 R , while the signal from neuropile region was significantly reduced in orb2 R flies ( Figure 4A,B).
Therefore, both biochemical and immunostaining approaches indicate that the amount of Orb2B localized in axons and synapses was decreased in the orb2 R mutant. The abnormalities observed in Orb2B localization could potentially account for LTM impairment in orb2 R flies.

Identification of Orb2 mRNA Targets in the Nervous System
A reduction of the amount of Orb2B localized in the synaptic zone would be expected to impact the localization and/or translation of mRNAs that are normally located in this subcellular compartment. To identify the potential neuronal mRNA targets that might be impacted by a reduction of Orb2B protein levels, we used RNA immunoprecipitation followed by next-generation sequencing. Because the available anti-Orb2 antibodies efficiently precipitated Orb2 from head extracts, we used a fly strain in which the orb2 gene encodes an Orb2-GFP fusion protein. It has been shown that orb2-gfp flies are fully viable and fertile and have normal LTM memory [19]. Antibodies against GFP were used to pull down Orb2-GFP and associated target mRNAs. The anti-GFP antibodies ensured a significant enrichment of the Orb2-GFP fusion protein in the experimental IP, while the Orb2-GFP protein was not detected in a mock IP (obtained with IgG of preimmune serum). We also found that only 10-15% of the Orb2-GFP protein was left in the output fraction after precipitation.
We assessed the efficiency of precipitation of Orb2 target mRNAs (act5C, orb2, tub56D transcripts) using qPCR. The targets showed moderate excess of presence in IP obtained with the anti-GFP antibodies as compared to the initial lysate, while lower levels were observed in the mock IP. Negative control transcripts (28S rRNA and RpLP0 mRNA) showed a similar precipitation efficiency in both IP with the anti-GFP antibodies and mock IP ( Figure 5B). Thus, our protocol showed the efficiency and selectivity of the precipitation of target mRNAs.
The sample of RNA coprecipitated with the anti-GFP antibodies was further analyzed by high-throughput sequencing. Reads were mapped onto the genome, and protein-coding genes whose transcripts were detected in both Input and IP samples were left in the list. The enrichment of each item was checked by two parameters, a ratio of CPM values in experimental IP to CPM in Input and a ratio of CPM in experimental IP to CPM in mock IP. We discarded genes with enrichment values lower than those observed for the orb2, act5C and tub56D mRNAs. Thus, we obtained a list of 6415 genes (Table S1).
We compared our list with those described previously. Among 5298 targets found in S2 cells [34], about 60% were also present in our list. Among 27 Orb2 targets found in the nervous system [35], 21 were detected. The list included the orb and apkc genes, whose mRNAs have previously been identified as Orb2 targets [18,36].
Many genes on the list are specific to the nervous system; about 40% of the genes show moderately high levels of expression in fly heads [37]. Analysis of GO terms of the set showed enrichment of genes that encode proteins of axons and synapses; proteins of neuron bodies were also present. Genes of the list are involved in multiple biological processes related to nervous system functions, including synapse assembly and activity, learning, and memory ( Figure 5C,D). Thus, we generated a list of putative Orb2 target mRNAs, which could be used for further analysis.

Intracellular Distribution of Proteins in Neurons is Altered in orb2 R Flies
A plausible cause of LTM loss in orb2 R flies is dysregulated translation of mRNAs synapses because of the depletion and improper localization of the Orb2 protein. To plore this possibility, we selected genes whose mRNAs were recovered as potential o targets in our RNA-IP experiment and had been implicated in LTM formation. We cho cysteine string protein (csp) [38], eyes absent (eya) [39], and polychaetoid (pyd) [40].
First, we estimated the distribution of these proteins in synapses and the soma usi a biochemical approach. The Csp protein was detectable in both somatic and synap fractions of WT and orb2 R flies. Although the Csp amount in the soma relative to the to protein in orb2 R flies was similar to that in WT flies, this was not the case with the synap fraction. As shown in Figure 6A, the Csp level in the synaptic fraction was nearly doub in orb2 R compared to WT. In the case of eya, we observed an approximately two-fold duction in the amount of the Eya protein in the soma fraction, while a very limited amou of the Eya protein in the synaptic fraction was unchanged ( Figure 6A). In the case of Pyd protein, we detected a slight decrease in protein amount in neuron bodies, while th was a marked (~three-fold) increase in the relative amount of the protein in the synap fraction.

Intracellular Distribution of Proteins in Neurons Is Altered in orb2 R Flies
A plausible cause of LTM loss in orb2 R flies is dysregulated translation of mRNAs in synapses because of the depletion and improper localization of the Orb2 protein. To explore this possibility, we selected genes whose mRNAs were recovered as potential orb2 targets in our RNA-IP experiment and had been implicated in LTM formation. We chose cysteine string protein (csp) [38], eyes absent (eya) [39], and polychaetoid (pyd) [40].
First, we estimated the distribution of these proteins in synapses and the soma using a biochemical approach. The Csp protein was detectable in both somatic and synaptic fractions of WT and orb2 R flies. Although the Csp amount in the soma relative to the total protein in orb2 R flies was similar to that in WT flies, this was not the case with the synaptic fraction. As shown in Figure 6A, the Csp level in the synaptic fraction was nearly doubled in orb2 R compared to WT. In the case of eya, we observed an approximately twofold reduction in the amount of the Eya protein in the soma fraction, while a very limited amount of the Eya protein in the synaptic fraction was unchanged ( Figure 6A). In the case of the Pyd protein, we detected a slight decrease in protein amount in neuron bodies, while there was a marked (~three-fold) increase in the relative amount of the protein in the synaptic fraction.
Next, we performed immunostaining of whole mount Drosophila brain samples with antibodies against these proteins. Staining with the antibody against Eya showed that Eya was found in nuclei and neuronal cell bodies in all parts of the central and lateral brain and the optic lobes. There was no overall change in Eya distribution in the mutant; however, as expected from the Western results, the signal in WT nuclei was greater than in Orb2 R nuclei. In addition, a group of nuclei that lacked the Eya protein in Orb2 R flies was observed in the basal part of the brain between the mushroom body calyces (a white oval in Figure 7). Next, we performed immunostaining of whole mount Drosophila brain samples w antibodies against these proteins. Staining with the antibody against Eya showed that E was found in nuclei and neuronal cell bodies in all parts of the central and lateral br and the optic lobes. There was no overall change in Eya distribution in the mutant; ho ever, as expected from the Western results, the signal in WT nuclei was greater than Orb2 R nuclei. In addition, a group of nuclei that lacked the Eya protein in Orb2 R flies w observed in the basal part of the brain between the mushroom body calyces (a white o in Figure 7).   Next, we performed immunostaining of whole mount Drosophila brain samples with antibodies against these proteins. Staining with the antibody against Eya showed that Eya was found in nuclei and neuronal cell bodies in all parts of the central and lateral brain and the optic lobes. There was no overall change in Eya distribution in the mutant; however, as expected from the Western results, the signal in WT nuclei was greater than in Orb2 R nuclei. In addition, a group of nuclei that lacked the Eya protein in Orb2 R flies was observed in the basal part of the brain between the mushroom body calyces (a white oval in Figure 7).

Figure 7.
Whole-mount staining of the Drosophila brain in WT and orb2 R flies. Staining was performed with antibodies against the Eya, Csp, and Pyd proteins. A dashed line shows the board between the optic lobe and the central brain, and a solid line shows the board of the central brain. White ovals show a group of cells whose nuclei were stained with the antibody against the Eya Figure 7. Whole-mount staining of the Drosophila brain in WT and orb2 R flies. Staining was performed with antibodies against the Eya, Csp, and Pyd proteins. A dashed line shows the board between the optic lobe and the central brain, and a solid line shows the board of the central brain. White ovals show a group of cells whose nuclei were stained with the antibody against the Eya protein in WT. For each antibody staining, the same exposure was used for samples of WT and Orb2 R flies.
The antibody against the Csp protein stained the central brain regions enriched in axons and dendrites, similar to published data [41]. The distribution of Csp protein in the WT brain was similar to that in Orb2 R flies, but the signal was higher in Orb2 R flies. The Pyd protein showed a perinuclear localization in neurons of the central brain and optic lobes and was mostly excluded from the neuropile region. We did not observe a significant change in staining pattern in Orb2 R flies, but the overall signal was higher, as in the case of the Csp protein.

Discussion
Previous studies have implicated Orb2 in learning and memory retention [7,8,19]. The two Orb2 isoforms, Orb2A and Orb2B, are believed to have different functions in LTM formation. Orb2A is thought to function as a catalyst of proto-prion formation [42]. Upon synaptic activity, Orbr2A primes LTM by undergoing a structural transition of the N-terminal polyQ domain. This enables Orb2A to form heteromeric complexes with the much more abundant Orb2B isoform. The structural transition from a proto-prion to a prion-like structure of the Orb2A polyQ domain generates a self-propagating feedforward loop, which ensures that any newly synthesized or transported Orb2 protein in the local neighborhood of the affected synapse will be captured in these heteromeric complexes. This mechanism provides a basis for long-lasting molecular changes that preserve memory over time.
Critical to this mechanism proposed for LTM formation, the Orb2 protein must be localized to the synapses responsible for memory formation. Based on the previous studies, several potential ways of targeting Orb2 to synapses could be in play before the events leading to memory formation take place. One of these would be translation of the orb2 mRNAs in the soma and then either diffusion or active transport of the protein into the neurite. Another mechanism would be on-site translation of the orb2 mRNAs in proximity to synaptic connections of neurons [43]. In this mechanism, the orb2 mRNAs would be transported from the soma to neurites in a translationally repressed state. Once on site, they could potentially be stored for future use or translated immediately. In both cases, an active process would be required to convert the translationally repressed mRNA into an active form. Both transport and translational regulation of mRNA typically depend on sequences in the 3 UTR [44]. These UTR sequences are likely to be especially important in neurons, as suggested by the observation that neurons differ from other somatic cell types in that many mRNAs have extended 3 UTRs [45].
The orb2 gene produces multiple mRNAs with different 3 UTRs. Noteworthy, only a single mRNA encodes Orb2A, and this mRNA has a short UTR of only 395 nt without any CPE-binding sequences. The localization and translation of the orb2A mRNA are therefore likely to be independent of the Orb2A/B proteins. In contrast, the mRNAs encoding Orb2B have 3 UTRs of 583, 1581, 3829, and 5794 nt. The two larger mRNA isoforms have over 25 CPEs, while the mRNAs with the two shorter UTRs have 2 and 7 CPEs. To determine whether the long UTRs are important for the orb2 function in the nervous system, we generated a deletion that removed all but the first five CPEs from the 3 UTR (the five CPEs are present in the 1581-nt isoform) [26]. This mutation did not affect the 3 UTR of the mRNA isoform encoding Orb2A.
Although we did not find reductions in viability or locomotion, striking defects were observed in LTM retention in orb2 R flies. This finding indicates that the 3 UTR of the long orb2 isoforms plays an important role in LTM. The deletion does not appear to affect orb2 transcription or mRNA stability because the levels of the orb2 mRNAs in orb2 R are similar to those in WT. However, the accumulation of the Orb2B protein is altered; i.e., we found that the total amount of the Orb2B protein in orb2 R heads was approximately 2/3 of that in WT heads. Although this reduction is modest, it is conceivable that what is hypothesized to be a key step in memory consolidation, the assembly of heteromultimers between Orb2A and Orb2B, might depend upon the concentration of Orb2B in the vicinity of critical synapses. This idea is supported by our analysis of the distribution of the Orb2B protein in WT and orb2 R neurons. Fractionation experiments indicate that Orb2B levels in the soma and synaptic fractions are close to equal in WT. In orb2 R , the Orb2B levels in the soma are similar to those in WT; however, the protein amount in the synaptic fraction is nearly three times lower. Therefore, the reduction in Orb2B protein levels in orb2 R appears to be largely restricted to the subcellular compartment that is expected to be intimately involved in the formation of LTM dependent on mRNA translation.
Using the RNA IP approach combined with high-throughput sequencing, we obtained a list of 6415 genes whose mRNAs are potential targets for binding and regulation by the Orb2 protein in the Drosophila nervous system. Analysis of the list of pulled-down mRNAs indicates that it fits with previously published results. The list is enriched in genes thought to be important for learning and LTM formation [46][47][48][49][50]. Moreover, Orb2 has previously been shown to participate in asymmetric cell division in neuroblasts [17]. Regulators of this process [51][52][53] are significantly enriched in our list of pulled-down mRNAs.
The list includes genes for proteins interacting with the microtubule cytoskeleton [95,96] and cytoskeletal machinery of neurons during axonal growth and maintenance [97]. At the same time, the list contains many genes encoding transcription factors [98,99], regulators of RNA metabolism [100,101], and components of signal-transduction systems [102,103]. Thus, Orb2 apparently mediates diverse molecular mechanisms of synaptic plasticity, and our list of putative Orb2 targets provides a useful tool for further studies in the field.
We evaluated the influence of the orb2 R mutation on the intracellular distribution of proteins whose mRNAs were identified as Orb2 targets. The Orb2 protein has been shown to bind the pyd and csp mRNAs, but not the eya (found in our RNA IP experiment) mRNAs in CLIP experiments [34]. Although the first two mRNAs are broadly expressed in flies and the cell culture used in CLIP analysis, eya expression is limited to a subset of cells, including neurons. The csp gene encodes a synaptic vesicle chaperone [104], which is expected to occur in dendrites and synapses. However, Csp is found in the soma as well. The pyd protein product is found in different parts of the cell, is involved in cell-to-cell junctions, and interacts with cytoskeletal proteins [105,106]. Finally, eya encodes a transcription factor, which is expected to occur in the nucleus (soma) [107]. It should be noted that all mentioned genes carry multiple CPE sequences in their 3 UTRs. We found that the orb2 R mutation changes the intracellular distribution of the Pyd and Csp proteins. The levels of both of these proteins increased in the synaptic fraction in the mutant compared to the WT control. Given the reduction in Orb2B protein in the synaptic fraction, this finding would suggest that Orb2B normally functions to repress translation of the pyd and csp mRNAs in the synaptic compartment. A different result was observed for eya. The accumulation of the Eya protein in the somatic fraction (nuclei) is reduced in the orb2 R mutant, indicating that orb2B likely functions as a translational activator of this mRNA in the soma compartment.
Supplementary Materials: The following supporting information can be downloaded at: https://www.mdpi.com/article/10.3390/cells12020318/s1, Figures S1-S6: Full-size images of the panels shown in Figure 4C. DAPI and anti-Orb2 staining of WT (Figures S1-S3) and mutant orb2 R (Figures S4-S6) fly brain preparations are shown. The white rectangles are the areas shown in Figure 4C. Scale bar 100 µm. Table S1: List of putative Orb2 targets in the nervous system. CPM values in In, IP, and mock IP samples and logFC values are given for each gene. The second tab presents the indicated values after the initial mapping of reads of the genome and their assignment to individual genes.

Institutional Review Board Statement: Not applicable.
Informed Consent Statement: Not applicable.

Data Availability Statement:
The data that support the findings of this study are openly available in the BioProject database.