Drep-2 is a novel synaptic protein important for learning and memory

Abstract

CIDE-N domains mediate interactions between the DNase Dff40/CAD and its inhibitor Dff45/ICAD. In this study, we report that the CIDE-N protein Drep-2 is a novel synaptic protein important for learning and behavioral adaptation. Drep-2 was found at synapses throughout the Drosophila brain and was strongly enriched at mushroom body input synapses. It was required within Kenyon cells for normal olfactory short- and intermediate-term memory. Drep-2 colocalized with metabotropic glutamate receptors (mGluRs). Chronic pharmacological stimulation of mGluRs compensated for drep-2 learning deficits, and drep-2 and mGluR learning phenotypes behaved non-additively, suggesting that Drep 2 might be involved in effective mGluR signaling. In fact, Drosophila fragile X protein mutants, shown to benefit from attenuation of mGluR signaling, profited from the elimination of drep-2. Thus, Drep-2 is a novel regulatory synaptic factor, probably intersecting with metabotropic signaling and translational regulation.

Keywords: CIDE-N protein family; D. melanogaster; fragile X syndrome; learning and memory; metabotropic glutamate receptors; mushroom body; neuroscience; synapses.

Figure 1.. Expression and mutants of drep-2.

(A) Genetic scheme of the drep-2 locus on chromosome IIR. The neighboring genes mad-1 and myd88 extend beyond the sequence displayed. The cDNA labeled drep-2-RA was used for rescue experiments. Blue: untranslated regions; green: exons; black lines: deleted regions in the mutants. (B) In situ hybridization of drep-2 reveals a neuronal expression pattern (stage 17). (C) Western blot of adult fly head extracts using the anti-Drep-2^C-Term antibody. Drep-2 isoforms are predicted to run at 52 and 58 kDa. The signal is absent in both the drep-2^ex13 and the drep-2^ex27/Df^w45-30n mutant. DOI: http://dx.doi.org/10.7554/eLife.03895.003

Figure 1—figure supplement 1.. Drep protein alignment.

Sequence alignment of all four Drosophila Dff proteins, as well as human (HS) and murine (MM) Dff40. Drep-4 has the strongest similarity to Dff40, yet Drep-2 also shows conserved motifs in addition to the CIDE-N domain. The alignment was created using Geneious v5.3.6. (http://www.geneious.com) DOI: http://dx.doi.org/10.7554/eLife.03895.004

Figure 1—figure supplement 2.. Reduced lifespan of drep-2^ex13 mutants.

Comparison to isogenic w¹¹¹⁸ control flies: 50% of mutant flies were dead after 21.5 days. Mutant: n = 10 vials (each containing 25 flies), control: n = 11. DOI: http://dx.doi.org/10.7554/eLife.03895.005

Figure 2.. Synaptic Drep-2 staining in the CNS.

(A–B) Confocal frontal sections of adult Drosophila brains. Anti-Drep-2^C-Term and Brp^Nc82 immunostaining; the latter marks all synaptic active zones. Synaptic Drep-2^C-Term signal is visible throughout the brain of wild-type flies (A). Complete loss of the anti-Drep-2^C-Term staining can be observed in drep-2^ex13 mutants (B). Scale bars: 20 µm. (C–E) Frontal sections of wild-type brains, anti-Drep-2^C-Term, and Brp^Nc82 staining. Scale bars: 5 µm. (C) Posterior–dorsal detail showing strong Drep-2 staining in MB calyces (arrow). (D) Anterior frontal section with antennal lobes and MB lobes. (E) Ellipsoid body in the central complex and bulbs (lateral triangles) (E′: magnification of strong Drep-2 staining in bulbs). DOI: http://dx.doi.org/10.7554/eLife.03895.006

Figure 3.. No evidence for a role of Drep-2 in regulation of apoptosis.

(A) Synaptosome-like preparation of adult wild-type head extracts (Depner et al., 2014), probed with Drep-2^C-Term. Drep-2 is concentrated in fractions containing synaptic membranes. S = supernatant, P = pellet, L = (after) lysis. Please see the protocol by Depner et al. (2014) for a more detailed explanation of the fractionation procedure. (B) Mutants (drep-2^ex13) did not show a rough eye phenotype. The facet eyes of flies, highly ordered structures, are often affected in apoptosis mutants. By contrast, the eyes of drep-2 mutants appeared normal. (C) The number of mb247-positive KCs does not differ between drep-2^ex13 mutants and controls. GFP was expressed using the MB KC driver mb247-Gal4. GFP-positive cell bodies were counted and compared between genotypes. No significant difference was found between mean cell body counts (Mann–Whitney U test, p = 0.886). Average cell body counts were in the expected range: control = 651, mutant = 669, published = 700 (Schwaerzel et al., 2002). (D) Purified Drep-2 does not degrade linearized plasmid DNA. Left: SDS-PAGE of the final elusion profile of purified Drep-2, loaded onto a HighLoad Superdex S200 16/60 column. Right: Nuclease activity assay of purified Drep-2 analyzed by 1% (wt/vol) agarose gel. Drep-2 was incubated in a time course experiment with linearized plasmid DNA. No nuclease activity could be detected. Instead, Drep-2 seemed to precipitate DNA, as evidenced by high-molecular DNA not entering into the agarose gel when incubated with Drep-2 (arrow). DOI: http://dx.doi.org/10.7554/eLife.03895.007

Figure 4.. Drep-2 is enriched at KC postsynapses.

(A–B) Drep-2^C-Term and Brp^Nc82 staining in animals expressing the construct mb247::Dα7^GFP that marks acetylcholine receptors in MB KCs. (A) Detailed image of the MB calyx. Scale bar: 2 µm. (B) Detail of a single microglomerulus in the calyx. Drep-2^C-Term overlaps with postsynaptic mb247::Dα7^GFP and not with presynaptic Brp. Scale bars: 1 µm. (C) Localization of Drep-2 relative to choline acetyltransferase (ChAT, presynaptic cytosol, C), the postsynaptic ACh receptor subunit Dα7 (antibody staining, C′), and the postsynaptic scaffolding protein Discs large (Dlg, C″). Drep-2 colocalizes with postsynaptic markers. Scale bars: 1 µm. (D) Post-embedding immunoelectron microscopy of Drep-2^C-Term in the calyx. Arrows: Clusters of postsynaptic Drep-2^C-Term. Scale bars: 100 nm. DOI: http://dx.doi.org/10.7554/eLife.03895.008

Figure 4—figure supplement 1.. Drep-2 localizes to postsynaptic membranes of KCs in the calyx.

(A) STED microscopy superresolution recording of Drep-2^C-Term; the Brp^Nc82 channel is in normal confocal mode. The Drep-2 signal does not overlap with presynaptic Brp. Scale bar: 1 µm. (B–E) Expression of drep-2 constructs in KCs yields a label resembling the Drep-2 antibody staining. Comparison to Brp^Nc82; all scale bars: 1 µm. (B) Pan-neural overexpression. Elav^c155-Gal4 and UAS-Drep-2^mStrawberry; mStrawberry signal is shown. (C) KC-specific overexpression. C305a-Gal4, UAS-Drep-2^mStrawberry, and UAS-Dα7^GFP; mStrawberry and GFP signals are shown. D-2 = Drep-2^mStrawberry, D7 = Dα7^GFP. (D) PN-specific overexpression. Gh146-Gal4 and UAS-Drep-2^mStrawberry; diffuse mStrawberry is shown. (E) KC-specific expression of UAS-Drep-2 in the drep-2^ex13-mutant background. Mb247-Gal4 and untagged UAS-Drep-2; Drep-2^C-Term staining is shown. DOI: http://dx.doi.org/10.7554/eLife.03895.009

Figure 5.. Drep-2 is required in KCs for olfactory short- and intermediate-term memory.

(A) Flies mutant for drep-2 sense electric shock and the odors 4-methyl-cyclohexanol (4-MCH) and 3-octanol (3-OCT) normally; there is no difference in mean performance indices between mutants and isogenic w¹¹¹⁸ control flies (Mann–Whitney U tests (MWU)). Sample sizes n are indicated with white numbers; grey bars show SEMs. (B) Both mutants drep-2^ex13 and drep-2^ex27/Df^w45-30n are deficient in aversive olfactory conditioning, 3 min STM in a T-maze. The graph shows mean learning indices and SEMs. Mutants performed significantly worse than isogenic controls (MWU: p = 0.00001 for both comparisons, Bonferroni-corrected significance level α = 0.0167, 3 tests). (C) Re-expression of drep-2 cDNA with elav^III-Gal4 (pan-neural), 30y-Gal4 (MB KCs), or mb247-Gal4 (MB KCs) restores the deficit to normal levels. Heterozygous drep-2^ex13 mutants do not display a significant deficit. MWU for individual comparisons showed a significant difference between these groups (α = 0.0042, 12 tests): w¹¹¹⁸ and drep-2^ex13 (p < 0.00001), drep-2^ex13/drep-2^ex13 and drep-2^ex13/+ (p < 0.00001), drep-2^ex13 and drep-2^ex13;uas-drep-2/elav^III-gal4 (p < 0.00001), drep-2^ex13 and drep-2^ex13;uas-drep-2/30y-gal4 (p < 0.00001), drep-2^ex13 and drep-2^ex13;uas-drep-2/mb247-gal4 (p < 0.00001). None of the differences indicated as not significant had a p < 0.12, except for w¹¹¹⁸ and drep-2^ex13/+ (p = 0.03851; not significant in the case of α = 0.0042). (D) Intermediate-term memory (ITM = ASM + ARM) performance. Mutants (drep-2^ex13) are defective in ASM, but not in ARM. The defect can be restored with elav^III-Gal4 or mb247-Gal4 (30y-Gal4 was not used here). Statistical tests were run separately for ITM and ARM. For ITM, MWU for individual comparisons showed a significant difference between these groups (α = 0.00625, 8 tests): w¹¹¹⁸ and drep-2^ex13 (p < 0.0001), drep-2^ex13 and drep-2^ex13;uas-drep-2/elav^III-gal4 (p < 0.0001), drep-2^ex13 and drep-2^ex13;uas-drep-2/mb247-gal4 (p < 0.0001). For assessing differences in ARM, ITM and ARM performances of each genotype were compared with MWU. The following genotypes showed a significant difference between ITM and ARM (α = 0.0071, 7 tests): w¹¹¹⁸ (p < 0.0001), drep-2^ex13;uas-drep-2/elav^III-gal4 (p = 0.0002), drep-2^ex13;uas-drep-2/mb247-gal4 (p = 0.0006). None of the differences indicated as not significant had a p < 0.11. DOI: http://dx.doi.org/10.7554/eLife.03895.010

Figure 5—figure supplement 1.. PN-KC synapses appear morphologically normal in drep-2 mutants.

(A) Absence of major neuroanatomical defects in drep-2^ex13 mutant brains. MB lobes, Fasciclin II (FasII) staining, maximum intensity projections. Scale bar: 10 μm. (B) Antibody staining of w¹¹¹⁸ control and drep-2^ex13 mutant brains, using antibodies against the postsynaptic ACh receptor subunit Dα7 and presynaptic Brp^N−Term. Focus on microglomeruli of PN-KC synapses in the MB calyx. Microglomeruli of mutants appear structurally normal. Scale bar: 1 µm. (C) Electron microscopy of w¹¹¹⁸ control and drep-2^ex13 mutant brains. Microglomeruli and postsynaptic KC profiles of mutants appear structurally normal. Scale bars: 100 nm. (D) The number of synapses (active zones) in the MB calyx does not significantly differ between drep-2^ex13 mutants and w¹¹¹⁸ controls. Syd-1-positive spots were counted and compared between genotypes as described (Kremer et al., 2010). No significant difference was found between the number of spots (MWU, p = 0.62). Average synapse counts were in the range expected (28,000–30,000 [Kremer et al., 2010]). DOI: http://dx.doi.org/10.7554/eLife.03895.011

Figure 6.. Functional overlap between Drep-2 and mGluR in olfactory conditioning.

(A) Wildtype adult MB calyces stained with Drep-2^C-Term and DmGluRA^7G11 (first row) or with Drep-2^C-Term and anti-Homer (second row). Drep-2 colocalizes tightly with both proteins. The insets show single microglomeruli. Scale bars: 2 µm. (B) Flies carrying the mutation dmGluRA^112b are deficient in aversive olfactory conditioning STM when compared to isogenic dmGluRA^2b controls that do express DmGluRA; MWU: p = 0.043, α = 0.05. The graph shows mean learning indices and SEMs; sample sizes n are indicated with white numbers. (C) The drep-2^ex13 phenotype in olfactory STM can be rescued by raising animals on food containing the DmGluRA agonist 1S,3R-ACPD (ACPD). Food was supplemented throughout development and adulthood with either the DmGluRA receptor antagonist MPEP (9.7 µM) or the agonist ACPD (72.2 µM) diluted in H₂O (label: dev+ad). Control animals received only H₂O. One group of animals was transferred to food supplemented by ACPD only after eclosion and not during development; the corresponding experiments are indicated by the label +ACPD adult. MPEP lowered the w¹¹¹⁸ performance significantly (MWU p = 0.0003). MPEP did not alter drep-2^ex13 indices (p = 0.8772) and ACPD did not change w¹¹¹⁸ performance (p = 0.1145). ACPD rescued the drep-2 mutant phenotype to control levels if fed during both development and adulthood (comparison of drep-2^ex13 +ACPD dev+ad to untreated drep-2^ex13: p < 0.00001; comparison to untreated w¹¹¹⁸: p = 0.0945). ACPD did not rescue the mutant phenotype if fed only during adulthood (+ACPD adult, no significant difference to untreated drep-2^ex13 (p = 0.2281), significant difference to mutants treated with ACPD during both development and adulthood (p < 0.00001)). The difference between untreated w¹¹¹⁸ and drep-2^ex13 flies was also significant (p < 0.00001). Significance level α = 0.005 (10 tests). (D) Phenotypes of drep-2^ex13; dmGluRA^112b double mutants were non-additive. Both drep-2^ex13 and dmGluRA^112b single mutants showed significantly lower olfactory STM than isogenic controls (MWU, p = 0.00008 for both comparisons). Double mutants showed similar learning indices (comparison to w¹¹¹⁸: p = 0.00018). The two single mutants and the double mutant did not significantly differ from each other (p > 0.178). α = 0.0083 (6 tests). (E) Loss of drep-2 antagonizes dfmr1 phenotypes in olfactory conditioning STM. Both homozygous drep-2^ex13 mutants and heterozygous dfmr1^B55/+ mutants are deficient in olfactory learning STM, but double mutants carrying both alleles do learn. The graph shows mean learning indices and SEMs. MWU for individual comparisons (α = 0.01, 5 tests): w¹¹¹⁸ and drep-2^ex13 p < 0.00001, w¹¹¹⁸ and dfmr1^B55/+ p = 0.00069, w¹¹¹⁸ and drep-2^ex13; dfmr1^B55/+ p = 0.83751, drep-2^ex13 and drep-2^ex13; dfmr1^B55/+ p < 0.00001, dfmr1^B55/+ and drep-2^ex13; dfmr1^B55/+ p = 0.00071. DOI: http://dx.doi.org/10.7554/eLife.03895.012

Figure 7.. Quantitative mass spectrometry: Drep-2 and FMRP were found in a common protein complex.

(A) Strategy for the identification of Drep-2 interactors by quantitative mass spectrometry. UAS-Drep-2^GFP was overexpressed using the pan-neural driver line elav^c155-Gal4. (B) Volcano plot showing proteins from Drep-2^GFP flies binding to anti-GFP and/or plain control beads. A hyperbolic curve (set at an FDR of 1%) separates GFP-enriched proteins (light pink) from background (grey). Proteins enriched in the control are shown in blue. Proteins that were significantly enriched, both in Drep-2^GFP flies and in independent control experiments with wild-type flies, are colored magenta (n = 35). Drep-2 and GFP are shown as green dots. (C) Classification of the 35 core network proteins; multiple counts were allowed. (D) Network of the 35 proteins that were significantly and reproducibly enriched in GFP pulldown experiments (at an FDR of 1%, magenta-colored dots in B). Additional putative interactors of the core network (FDR set at 10%) are shown in white (Supplementary file 2). The circle (node) and font size correspond to the rank within the results (indicated in Supplementary files 1 and 2). The line (edge) width and shade correspond to the number of interactions each of the significantly enriched proteins has with others. The line/edge length is arbitrary. (E) Anti-FMRP probing confirmed the specific presence of FMRP in Drep-2^GFP complexes. Head extracts of flies expressing Drep-2^GFP or the presynaptic protein Syd-1^GFP were processed in parallel. FMRP was only enriched in preparations of Drep-2^GFP extracts. Immunoprecipitations were performed using either GFP-Trap-A beads (lanes labeled IP) or blocked agarose beads as binding control (labeled Blank beads). DOI: http://dx.doi.org/10.7554/eLife.03895.013

Figure 7—figure supplement 1.. Drep-2^GFP colocalizes with endogenous Drep-2.

(A) Pan-neural overexpression of UAS-drep-2^GFP by elav^c155-Gal4. MB calyx stained with anti-GFP, Brp^N-Term and Drep-2^C-Term. The Drep-2^GFP label does not differ from the Drep-2^C-Term antibody staining, compare also to Figure 4 and Figure 4—figure supplement 1. Scale bars: 10 µm and 1 µm (insets). (B) KC-specific expression of UAS-drep-2^GFP by mb247-Gal4 in drep-2^ex13 mutants. Staining and scale bars as in (A). DOI: http://dx.doi.org/10.7554/eLife.03895.014

Author response image 1.