Multiple neurons encode CrebB dependent appetitive long-term memory in the mushroom body circuit

To investigate the spatial requirement of CrebB for LTM formation within the mushroom body circuit we generated a CrebB conditional knockout allele (CrebB^cKO) using the CRISPR/Cas9 system (Bassett and Liu, 2014). In short, two FRT sites and an N-terminal eGFP tag were introduced into the CrebB locus. The first CRISPR site was selected immediately upstream of the CrebB start codon and the second used CRISPR site was located before the last exon of CrebB thus replacing the coding region of the locus with a donor template sequence for homology-directed repair (HDR) containing two FRT sites flanking the coding sequence of eGFP and the CrebB genomic sequence (Figure 1A,B). The successful generation of the CrebB^cKO allele was confirmed by sequencing of the 3’ and 5’ introduced GFP and FRT sites. The FRT flanked CrebB::GFP locus allows visualization of CrebB protein localization and proof of flippase recombinase mediated removal of the fusion protein by the absence of GFP expression. Using different Gal4 driver lines and Gal80^ts, a temperature sensitive inhibitor of Gal4, spatially and temporally controlled UAS-FLP expression was performed (McGuire et al., 2003). In FLP expressing cells, GFP-tagged CrebB is deleted from the genome, creating CrebB mutant cells (Figure 1B).

We first assessed the engineered FRT- GFP::CrebB -FRT locus for proper CrebB protein expression by generating a guinea-pig polyclonal anti-CrebB antibody raised against the CrebB full-length protein sequence of isoform F. CrebB::GFP is expressed in virtually all cells of the brains of CrebB^cKO flies. Co-localization of anti-CrebB and anti-GFP antibody shows successful tagging of CrebB with GFP and further supports the pan-neuronal expression of CrebB (Figure 1C and Figure 1—figure supplement 1). UAS-FLP; nSyb-Gal4 was used to remove CrebB::GFP in all neurons. To prevent developmental defects, we temporally restricted FLP expression using a tubGal80^ts transgene (McGuire et al., 2003). Knockout of CrebB was induced after hatching by moving animals to 29°C for 6 days before antibody staining or conditioning experiments. Brains from flies with induced pan-neuronal CrebB deletion were dissected and antibody staining performed. Expression of CrebB::GFP in the brain was completely lost in neurons and only detectable in Repo-positive glia cells confirming the efficiency of the created CrebB conditional knockout (Campbell et al., 1994; Xiong et al., 1994; Halter et al., 1995) (Figure 2A,B).

To study the role of CrebB in memory formation, we used a classical olfactory conditioning paradigm, in which flies learned to approach an odor that was previously associated with sugar reward (Krashes and Waddell, 2008; Colomb et al., 2009). We tested flies directly after conditioning to measure short-term memory (STM), after 3 h to measure middle-term memory (MTM) or after 24 h to measure long-term memory (LTM). We first tested flies with induced pan-neuronal CrebB knockout along with the corresponding parental lines for STM. Since CrebB is located on the X chromosome, we always calculated memory performance indices of male and female animals separately. Male offspring flies of CrebB^cKO; tubGal80^ts females crossed with +;UAS-FLP; nSyb-Gal4 males have only the CrebB^cKO allele (CrebB^cKO/Y) and therefore are CrebB null mutants after flippase induced excission. Female progeny of the cross have in addition to CrebB^cKO a wild type allele of CrebB and served as control group together with male flies of the parental lines. The tested groups did not show significantly different memory scores measured immediately after training, confirming dispensability of CrebB for STM (Figure 2C). We next assessed MTM and found that also for this memory phase CrebB is not required in neurons (Figure 2D). To evaluate LTM, flies were tested 24 h after conditioning. Unflipped male CrebB^cKO/Y flies showed intact LTM and did not perform significantly different from female CrebB^cKO/+, which further confirmed that conducted genome editing in the CrebB locus does not interfere with the function of CrebB in LTM formation. However, males with induced CrebB knockout in all neurons showed drastically reduced 24 h LTM performance (Figure 2E). The memory performance of non-induced flies kept continuously at 18°C was similar to that of genetic controls (Figure 2—figure supplement 1).

We next tested the requirement of CrebB in the mushroom body, a central structure for olfactory associative memory (Heisenberg et al., 1985; de Belle and Heisenberg, 1994; Dubnau et al., 2001; Crittenden et al., 1998). To remove CrebB from the entire MB we used the OK107-Gal4 line, which is expressed in most of the KCs across the different lobes (Aso et al., 2009). To confirm the CrebB removal we stained with the KC marker Eyeless (Ey), which showed complete absence of CrebB::GFP from the entire MB six days after knockout induction (Kurusu et al., 2000) (Figure 3B). Next, we tested MB-specific CrebB knockout flies for their memory performance. While control animals showed normal LTM, animals lacking CrebB in the MBs displayed impaired LTM, indicating that the MB is essential for CrebB mediated LTM formation (Figure 3C). In contrast, STM was not affected by loss of CrebB in KCs (Figure 3—figure supplement 1) and non-induced flies displayed normal LTM (Figure 3—figure supplement 2A).

We next examined the effect of CrebB knockout on LTM in each of the three major MB lobe subclasses using lobe specific Gal4 drivers (c739-Gal4 for α/β, c305a-Gal4 for α′/β′ and 5-HTR1B-Gal4 for γ neurons) (Aso et al., 2009; Yuan et al., 2005; Xie et al., 2013). We observed a significant reduction in LTM of animals with CrebB knockout in MB α/β neurons compared to control groups (Figure 3D). Similarly, flies with a CrebB knockout in MB α’/β’ neurons also exhibited impaired long-term memory performance (Figure 3E). However, removing CrebB from MB γ neurons did not affect LTM formation. Performance indices did not differ between the tested groups (Figure 3F). Effective knockout of CrebB from MB γ neurons could be confirmed by antibody staining (Figure 3—figure supplement 3). LTM performance was also tested in flies with non-induced CrebB knockout. Non-induced control flies of the KC subtype experiments did not show LTM defects (Figure 3—figure supplement 2B–D). These findings show that appetitive LTM formation requires CrebB activity in α/β and α′/β′ KCs but not in γ KCs.

For distinct forms of learning, different types of MB intrinsic and extrinsic neurons are required. Appetitive memories require dopaminergic input from the PAM cluster neurons (Burke et al., 2012; Liu et al., 2012). We used the GMR58E02-Gal4 line that strongly labels the PAM cluster neurons to mediate CrebB knockout in those DANs. However, flies with CrebB null mutant PAM neurons were able to form LTM indistinguishable of the concurrently tested control groups (Figure 4A). To verify that CrebB was successfully removed from PAM neurons, antibody staining experiments were conducted. After induction of CrebB knockout in PAM neurons, CrebB::GFP could not be detected anymore in most TH-labeled dopaminergic neurons in the brain region where PAM neurons are located (Figure 4—figure supplement 1).

We next assayed if CrebB is needed in MB efferent neurons. Two pairs of cholinergic mushroom body output neurons MBON α3 (also called MB-V3) were reported to be necessary for appetitive LTM (Plaçais et al., 2013). The G0239-Gal4 line is a highly specific driver that expresses Gal4 exclusively in MBON α3 in the adult brain (Chiang et al., 2011). We used this line to express FLP and induce CrebB deletion from the genome of MBON α3. Interestingly, memory measured 24 h after appetitive olfactory conditioning was impaired in these flies (Figure 4B). Furthermore, LTM performance was not affected under low temperature conditions, in which FLP expression is not induced (Figure 4—figure supplement 2). Thus, while dopaminergic input neurons do not require CrebB, MBON α3 requires CrebB for appetitive LTM. This finding supports that LTM is not only encoded within the MB-lobe system, but also involves a memory trace in the MBON extrinsic element.

Flies (Drosophila melanogaster) were generally kept at 25°C and subjected to a 12 h light – 12 h dark cycle. A cornmeal medium supplemented with yeast, fructose and molasses was used to rear the flies. Canton-S was used as wild-type (courtesy of R. Stocker). c739-Gal4 was obtained from Hiromu Tanimoto (Tohoku University) and OK107-Gal4 (106098) was received from Kyoto stock center. nSyb-Gal4 (51635), c305a-Gal4 (30829), 5-HTR1B-Gal4 (27636), GMR58E02-Gal4 (41347), G0239-Gal4 (12639), tubGal80^ts (7019), nos-Cas9 (54591) and UAS-FLP (55804, 55806) were obtained from Bloomington stock center.

Genomic DNA of nos-Cas9 flies was used as template for PCR of the CrebB genomic fragments. A 3.3 kb fragment of the CrebB encoding region including introns was amplified with primers ‘CrebB noStart R1 fw’ (gcgaattcGACAACAGCATCGTCGAGGAGAACG) and ‘CrebB intr RV re’ (gagataTCCTGCCAAGTCGCAACTAAAGGC). The resulting sequence starts with the second codon (Asp GAC) of CrebB just after the EcoRI restriction site, and ends within the last intron 45 bp downstream of the facultative exon of isoforms PG, PI, PN and PQ followed by an EcoRV restriction site. For the upstream homology arm a 2.3 kb fragment upstream of the CrebB Start codon was amplified with primers ‘CrebB CR Not fw’ (gggcggcCGCGGAGGTAATGCGGATTTGG) and ‘CrebB 5’UTR Spe re’ (ttactagtCCTGGCGATCTTCAGCAGCACC). This fragment starts 2312 bp upstream of the CrebB start codon within the sequence of the RNA expressing CR43686 gene, and ends 30 bp upstream of the CrebB Start codon within the 5’UTR. For the downstream homology arm a 2.1 kb fragment including the last exon of CrebB was amplified with primers ‘CrebB intr Xho fw’ (ggaccactcgagAATCGAACTGGAATCGAGGGTCTATC) and ‘CrebB 3’UTR Kpn re’ (gtggtaCCGTCCCTTCGTCTCTTTTCTACC). This fragment starts within the last intron of CrebB 56 bp downstream of the facultative exon of isoforms PG, PI, PN and PQ, and ends within the 3’UTR of the long CrebB isoforms 1131 bp downstream of the Stop codon. All three genomic fragments were subcloned into pBluescript with the appropriate restriction enzymes. After verification of the sequences, the three fragments were assembled in a pBluescript derivative generated in our lab in which we have inserted two FRT sites and the coding sequence of GFP into the multiple cloning site allowing us to generate FRT-flanked and GFP-tagged CRISPR templates. The first coding exon of CrebB lacking the CrebB Start codon was fused in frame with GFP. One of the FRT sites is located immediately upstream of the GFP coding sequence between the 2.3 kb upstream homology arm and GFP, the other FRT site is located between the 3.3 kb CrebB coding region and the 2.1 kb downstream homology arm. Thus, upon FRT mediated recombination the GFP-tagged CrebB coding region from the first coding exon to the second last coding exon will be removed.

For the first guide RNA oligos ‘CRISPR CrebB Start sense’ (cttcGATCGCCAGGATCGGCAACA) and ‘CRISPR CrebB Start antisense’ (aaacTGTTGCCGATCCTGGCGATC) were annealed and ligated into BbsI digested pU6-BbsI-chiRNA vector. This guide RNA will target Cas9 to cut 2 bp upstream of the CrebB Start codon. For the second guide RNA oligos ‘CRISPR CrebB intr sense’ (cttcGGACCACTCGTAAATCGAAC) and ‘CRISPR CrebB intr antisense’ (aaacGTTCGATTTACGAGTGGTCC) were annealed and ligated into BbsI digested pU6-BbsI-chiRNA vector. This guide RNA will target Cas9 to cut 61 bp downstream of the facultative exon of isoforms PG, PI, PN and PQ. The CRISPR sites are missing in the template DNA, where they were replaced with the FRT sites. A mix containing 0.4 µg/µl template and 0.2 µg/µl of each guide RNA plasmid was injected into nos-Cas9 expressing flies. The injected flies were crossed with w¹¹¹⁸ mutants and their offspring screened for GFP expressing larvae. GFP positive flies were crossed with FM7 balancer flies to establish a stable line. The presence of the GFP-tag and the FRT sites was verified by PCR and sequencing of the genomic DNA of the homozygous CrebB^cKO stock.

CrebB cDNA was PCR amplified from EST clone RT01009 and cloned into the pGEX-6P-1 plasmid to produce a Gst-tagged version in bacteria. The resulting Gst-CrebB coding sequence had a small in-frame deletion removing some Glycine residues of the N-terminal Glycine stretch but the rest of the sequence was unaltered. Bacterially expressed Gst-CrebB was purified with Glutathione Sepharose beads (GE healthcare) and injected into guinea pigs for antibody production (eurogentec).

Flies, in which CrebB knockout was induced, were moved for 6 days to 29°C before antibody staining experiments. Male adult brains were dissected in phosphate-buffered-saline (PBS) and fixed at room temperature for 30 min with a 3.7% formaldehyde solution (in PBS). Brains were washed at least five times with PBST (PBS with 0.3% triton X-100) before primary antibodies were added for overnight incubation at 4°C. The following primary antibodies were used: rabbit anti-Eyeless (1:400, courtesy of Uwe Waldorf), chicken anti-GFP (1:1000, Abcam ab13970), rabbit anti-GFP (1:1000, Invitrogen A-6455), guinea pig anti-CrebB (1:400), rabbit anti-Tyrosine hydroxylase (1:100, Merck AB152), mouse anti-Repo (1: 20, Developmental Studies Hybridoma Bank 8D12), mouse anti-Discs large (1:50, Developmental Studies Hybridoma Bank 4F3). Brains were washed again before the overnight incubation at 4°C with secondary antibodies. The used secondary antibodies were conjugated with Alexa fluorescent proteins (488, 567 or 647; Molecular Probes) and diluted 1:200. After a last washing step, brains were mounted in Vectashield H-1000 or Vectashield with DAPI H-1200 (Vector Laboratories) on a microscope slide. Samples were imaged with Leica SP5 confocal microscope and the images were processed with Imaris Bitplane 9.2 and Adobe Photoshop CS6.

Memory experiments were performed at 23–25°C and 70–75% relative humidity. Conditioning was carried out in dim red light and tests were done in darkness. The used conditioning apparatus is based on Tully and Quinn, 1985 and was modified to perform four experiments in parallel (Schwaerzel et al., 2002); CON-Elektronik, Greussenheim, Germany). The odors benzaldehyde (Fluka, 12010) and limonene (Sigma-Aldrich, 183164) were used. 60 μl of benzaldehyde was applied in plastic containers measuring 5 mm in diameter and 85 μl of limonene was applied in plastic containers measuring 7 mm in diameter. Odor delivery was effected with a vacuum pump adjusted to a flow rate of 7 l/min. Filter papers were soaked with distilled water or with a 1.5 M sucrose (Sigma-Aldrich, 84100) solution the day before the experiment and left to dry at room temperature overnight.

19–21 h before conditioning groups of 60–100 flies were put into plastic vials with wet cotton wool on the bottom for starvation. For appetitive olfactory conditioning, flies were loaded in tubes lined with water filter papers. After an acclimatization period of 90 s, a first odor was presented for 2 min. Then, the odor was removed and animals were transferred within 60 s to tubes lined with sucrose filter papers. Subsequently, the second odor was presented for 2 min.

For the memory tests, flies were moved to a two-arm choice point where they could choose between limonene and benzaldehyde for 2 min. After this period, the number of flies within each arm was counted and a preference index was calculated.

PREF = ((N_{paired odor} – N_{control odor}) 100)/N_total

One memory experiment consisted of two groups with reciprocal conditioning, in which the sucrose paired odor was exchanged. The preference indices from these two groups were averaged to calculate a performance index (PI).

To measure short-term memory, flies were tested immediately after conditioning. Middle-term memory was examined 3 h after conditioning. For those experiments, animals were put back into starvation vials after training and were starved until the test. Flies tested for 24 h memory were put in food vials after conditioning for 3–5 h and then transferred to starvation vials until the test.

Flies used for memory experiments were reared at 18°C and collected after hatching (0–3 d old flies). To induce expression of flippase, flies were moved for 6 d to 29°C. For the starvation period before conditioning and until the test, those animals were kept at 25°C. Flies used for the non-induced FLP expression experiments were kept at 18°C after collection (for 6 days) and during the starvation time. Memory experiments with those flies were performed at 20–22°C.

To compare PIs between two groups the Welch two sample t-test was used. Statistical analyses and graphical representation of the data were performed using R version 3.4.1.

All data is included in the manuscript.