The Drosophila brain on cocaine at single-cell resolution

Whereas the neurological effects of cocaine have been well documented, effects of acute cocaine consumption on genome-wide gene expression across the brain remain largely unexplored. This question cannot be readily addressed in humans but can be approached using the Drosophila melanogaster model, where gene expression in the entire brain can be surveyed at once. Flies exposed to cocaine show impaired locomotor activity, including climbing behavior and startle response (a measure of sensorimotor integration), and increased incidence of seizures and compulsive grooming. To identify specific cell populations that respond to acute cocaine exposure, we analyzed single-cell transcriptional responses in duplicate samples of flies that consumed fixed amounts of sucrose or sucrose supplemented with cocaine, in both sexes. Unsupervised clustering of the transcriptional profiles of a total of 86,224 cells yielded 36 distinct clusters. Annotation of clusters based on gene markers revealed that all major cell types (neuronal and glial) as well as neurotransmitter types from most brain regions were represented. The brain transcriptional responses to cocaine showed profound sexual dimorphism and were considerably more pronounced in males than females. Differential expression analysis within individual clusters indicated cluster-specific responses to cocaine. Clusters corresponding to Kenyon cells of the mushroom bodies and glia showed especially large transcriptional responses following cocaine exposure. Cluster specific coexpression networks and global interaction networks revealed a diverse array of cellular processes affected by acute cocaine exposure. These results provide an atlas of sexually dimorphic cocaine-modulated gene expression in a model brain.


Drosophila stock
Canton S (B) flies (Norga et al. 2003) were maintained on standard cornmeal/yeast/molassesagar culture medium at 25C on a 12:12 hour light:dark cycle with 50% humidity in controlled adult density vials to prevent overcrowding. Briefly, 5 males and 5 females were placed into a vial and allowed to mate for two days before being cleared. Progeny from these vials were collected after eclosion and aged for 3-5 days before experimentation.

Cocaine exposure
Cocaine.HCl was obtained from the National Institute on Drug Abuse under Drug Enforcement Administration license RA0443159. To expose flies to cocaine, we performed a modified version of the capillary feeder (CAFÉ) assay (Ja et al. 2007). We collected 200 Canton S (B) flies between 3 and 5 days old using CO2 anesthesia, sexes separately. We placed them individually in culture vials containing cornmeal/yeast/molasses/agar culture medium (Genesee Scientific, Inc., San Diego, CA) and allowed them to recover for 24 hours before experimentation. Between 3:00-5:00 PM on the day before the assay, we transferred the flies to vials containing 1.5% agar (Sigma Aldrich, St. Louis, MO) in which a capillary (VWR International, Radnor, PA: 12.7 cm long, 5 μl total volume) filled completely with a solution of 4% sucrose (Sigma Aldrich) and 1% yeast (Fisher Scientific, Hampton, NH) was inserted. The next morning, we replaced the sucrose capillaries for 100 flies with capillaries containing 4% sucrose supplemented with 1 µg/µL of cocaine and 1% yeast; for the other 100 flies, we replaced the sucrose capillaries with fresh 4% sucrose and 1% yeast with no drug. A droplet of mineral oil (Sigma Aldrich) was added to the top of each capillary to minimize evaporation. We collected the first 40 flies that consumed 0.53 µL of cocaine and the first 40 flies that consumed 0.53 µL of sucrose, corresponding to an 8 mm reduction in the height of the solution in the capillary. All experiments were carried out between 8 AM and 11 AM. Flies were allowed to feed for no more than 2 hours.

Behaviors
We measured negative geotaxis and startle response of individual flies within a 10-minute timeframe immediately following acute exposure to cocaine in the CAFÉ assay. We quantified grooming and seizures in addition to measuring the behavioral response in each assay. Excessive grooming was defined as more than 10 seconds of constant grooming (Video S2). Seizure activity was defined as severe muscle tremors that prevented the fly from moving normally (Video S3).
Negative geotaxis: Following acute cocaine consumption, we placed each fly in a 14.8 cm-tall clear glass vial with its circumference marked 7.5 cm up the vial. Flies were given 30 seconds to acclimate to the vials. We then tapped the flies to the bottom of the vial and recorded the time taken for each fly to cross the 7.5 cm mark, with a maximum allowed climb time of 30 seconds.
Flies that did not pass the mark within 30 seconds were designated as "did not finish". The numbers of flies tested are indicated in the legend to Figure 1. Significant differences from control were assessed using one-tailed Student's t-test. Grooming and seizure activity were also scored at this time and differences between control flies and flies exposed to cocaine were assessed using Fisher's exact test.
Startle response: Following acute cocaine consumption, we tested single flies in their vials for acute startle response. To ensure the same amount of mechanical stimulation for all trials, we constructed a 'fly drop tower' in which all vials were dropped 42 cm and then secured in a horizontal position. As soon as the vials attained a horizontal position the flies were observed for 45 seconds and the total time each fly spent moving was recorded (Video S1). The numbers of flies tested are indicated in the legend to Figure 1. Significant differences from control were assessed using one-tailed Student's t-test. Grooming and seizure activity were also scored at this time and differences between control flies and flies exposed to cocaine were assessed using Fisher's exact test. While grooming, flies were stationary.

Brain dissection and dissociation
Brains were dissected from each fly immediately after it consumed the designated amount of sucrose or cocaine solution. We used a dissociation protocol modified from Croset et al. (2018) and Davie et al. (2018). We dissected brains in cold D-PBS (Gibco, Thermo Fisher Scientific, Waltham, MA) and collected them into 1.7 ml tubes in cold Schneider's medium (Gibco). We collected 20 brains per sample within one hour. We collected eight samples of 20 brains from males and females exposed to cocaine or sucrose, with two biological replicates per treatment and sex. We replaced the D-PBS in the dissection dish after dissecting 2 brains to ensure that it stayed cold and we used separate drops of buffer for decapitation and brain dissection to avoid contaminating the brain samples. We centrifuged the samplesn at 300xg at 4 o C for 5 min and removed the supernatant. We then added 450µl of collagenase solution (50 ul of fresh 25mg/ml collagenase (Gibco) in sterile water + 400µl of Schneider's medium), flicked the tube gently and allowed the brains to incubate at room temperature for 30 min. We replaced the collagenase solution after centrifugation with PBS + 0.04% BSA (NEB, Ipswich, MA). We mechanically dissociated the brains slowly and gently, using stepwise trituration -P200 pipette 5 times, 23G needle pre-wetted with PBS + BSA 5 times, and 27G pre-wetted needle 5 times. We passed the suspension through a pre-wetted 10µm strainer (Celltrics, Görlitz, Germany) aided by gentle tapping. We added 50ul of PBS+BSA to aid flow of the suspension through the strainer. We counted live cells using a hemocytometer with trypan blue exclusion. We proceeded with GEM generation using the Chromium controller (10X Genomics, Pleasanton, CA) if we had a live cell count of > 500 live cells/µl.

Library preparation and sequencing
We made libraries after GEM generation in accordance with 10X Genomics v3.1 protocols. We determined fragment sizes using Agilent Tapestation kits (Agilent, Santa Clara, CA) -d5000 for amplified cDNA and d1000 for libraries. However, the samples were not diluted at either step since these were not high concentration libraries. We measured the concentrations of amplified cDNA and the final libraries using a Qubit 1X dsDNA HS kit, also without dilution. In addition to Qubit, we quantified the final library concentrations using a qPCR based library quantification kit (KAPA Biosystems, Roche, Basel, Switzerland) in order to measure the concentration of fragment sizes of interest in accordance with the manufacturer's recommendations. We used 12 cycles for the cDNA amplification and 14 cycles for indexing PCR. We sequenced the final libraries on an S1 flow cell using a Novaseq (Illumina, Inc., San Diego, CA) according to the manufacturer's instructions.

FASTQ generation, demultiplexing and alignment
The mkfastq pipeline within Cell Ranger v3.1 (10X Genomics, Pleasanton, CA) was used to convert BCL files from the sequence run folder to demultiplexed FASTQ files. Release 6 version of the Drosophila melanogaster reference GCA_000001215.4 from NCBI Genbank was indexed using the mkref pipeline and used for alignment using the count pipeline within Cell Ranger v3.1 with the expected cell count parameter set to 5,000 cells. The sequencing and alignment summary is given in Supplemental Table S2.

Preprocessing, integration and cell-type clustering
Raw expression counts output for each sample from the Cell Ranger pipeline was imported and analyzed using the Seurat v3 package in R (Butler et al. 2018). Genes expressed in less than 5 cells and cells with less than 300 or greater than 2500 RNA features were filtered out.
Normalization and subsequent integration were performed using scTransform pipeline (Hafemeister and Satija, 2019). To identify the cell-type clusters within the dataset, unsupervised clustering using the FindClusters function and a resolution of 0.8 was used. Cluster marker genes were identified using FindAllMarker function (min.pct=0.25, logfc.threshold = 0.5, only.pos = TRUE).The top three genes with positive expression for each cluster were extracted and used for cell-type characterization.

Differential expression
Differential expression was performed for each cluster in two ways: (i) after combining male and female samples together to test for effects of cocaine that are common to both sexes; and (ii) testing for effects of cocaine in males and females separately to identify sexually-dimorphic responses. The Pearson residuals output from scTransform pipeline was used as input for differential expression (DE) calculation (Hafemeister and Satija, 2019). The MAST algorithm was used as the testing methodology in the FindMarkers function (test.use = "MAST", assay = "SCT", slot = "scale.data") for each cluster to calculate DE. Clusters with sufficient number of DEGs were subjected to pathway enrichment analysis using the statistical overrepresentation test using the PantherDB (Thomas et al. 2003) and Reactome databases (Fabregat et al. 2016). Pathways with BH-FDR adjusted P values < 0.05 were considered statistically enriched.

Simulation of bulk RNAseq response
The results from DE calculation from the combined dataset were used to determine which genes were consistently upregulated and downregulated, respectively, across all clusters as a result of exposure to cocaine. The top 50 ranked differentially upregulated genes for each cluster and the top 20 ranked differentially downregulated genes for each cluster were input into TopKLists R package (Schimek et al. 2015).

Cluster-specific co-expression networks
The scaled data from the scTransform pipeline for differentially expressed genes from clusters 16 and 22 were extracted for the male samples. These scaled data were used as input for filtering through Random Matrix Theory (RMT; Gibson et al. 2013). The correlations that passed the filtering process were visualized using Cytoscape version 3.7.2. The MCODE algorithm (Bader et al. 2003) was utilized to identify highly interconnected modules within the larger cluster network.
Genetic interaction networks were constructed by converting the gene IDs to gene names/symbols using the FlyBase Consortium's 'Query-by-symbols/ID' tool and calculating interactions between gene products using the stringApp plugin within Cytoscape (Doncheva et al.  Tables   Table S1: Raw behavioral data of flies exposed to cocaine. Refer to Supplemental_Table_S1.xlsx            correlations.