GoKawiilscRNA-seq data generation
To discriminate technical from biological variability, we used flies from different genetic backgrounds19,77, enabling us to generate scRNA-seq libraries of VNCs of different sex and developmental stages and separate them after sequencing through single-nucleotide polymorphisms (SNP)-based sample demultiplexing (see below)78 (Extended Data Fig. 1). Randomized samples from different genotypes were used for the same sex and developmental stages across experiments. In total, data were collected from 92 individual animals, covering four developmental stages and both sexes. Each stage–sex condition is represented by 5 to 18 biological replicates, covering between 5 and 11 different genotypes. Some experimental preparations were used to create 2 or 3 independent libraries.
Dissociation
For each stage–sex condition, several genotypes and individuals were used, as summarized in Supplementary Table 1. Flies of the genotypes indicated in Supplementary Table 2 were allowed to mate and their larvae developed at 25 °C until white pupae stage. Then pupae were collected with forceps, sexed under a dissecting scope and put in either male or female vials to develop until the correct stage, that is, 6, 24, 36 or 48 h.a.p.f. VNCs were dissected in ice-cold DPBS (Dulbecco’s PBS, Gibco). A maximum of 8 VNCs, each of a different genotype, was pooled in a low-bind Eppendorf tube containing 100 µl of DPBS on ice. VNCs from animals of two or three different developmental stages and sex were pulled in the same tube (6 + 24 h.a.p.f., 24 + 48 h.a.p.f. or 24 + 36 + 48 h.a.p.f.). Four brain samples were included in the last round of experiments, but later excluded from the analysis of VNC. After dissections, tissues were collected at the bottom of the tube by centrifugation (800g, 5 min, 4 °C), DPBS was replaced with 50 µl of dispase I (3 mg ml−1, Sigma-Aldrich, D4818, reconstituted in 50 mM HEPES/KOH pH 7.4, 150 mM NaCl) and 75 µl of collagenase I (100 mg ml−1, Invitrogen, 17100-017, reconstituted in HBSS). VNCs were dissociated in a thermo mixer (Eppendorf) at 25 °C at 500 rpm for 30 (6 + 24 h.a.p.f. samples) or 40 min (24 + 36 + 48 h.a.p.f. samples). Dissociation was aided by pipetting up and down 10 times every 10 min ensuring that the tip touched the bottom of the tube to increase the mechanical stress. Dissociated cells were centrifuged (800g, 5 min, 4 °C), washed once with DPBS and resuspended in 180 µl of DPBS with 0.04% BSA. The cell suspension was filtered through a 10 µm filter (pluriStrainer, Cambridge Bioscience, 43-50010-03). Viability, presence of debris and cell number were accessed using the Countess II automated cell counter (Invitrogen).
Library preparation and sequencing
Samples were collected on 3 different days. Each day, a total of 4 different cell suspensions (independent samples) was generated, each containing different combinations of genotypes, ages and sex. Single-cell expression libraries were generated at the CRUK-CI sequencing facility using Chromium Next GEM Single Cell 3′ HT v3.1. We aimed to load each sample in duplicate, 20,000 cells per lane. Before sequencing, libraries were inspected on the TapeStation. Sequencing was performed on the NovaSeq6000 machine (Illumina) with the following sequencing parameters: 28 regular cycles of which 16 are 10x barcode and 12 are unique molecular index, 10 i7-index cycles, 10 i5-index cycles, 90 regular cycles.
scRNA-seq data processing
No formal experimental blinding was applied for the generation and analysis of the scRNA-seq atlas; however, computational analyses were performed in an unsupervised manner, and clustering and dimensionality reduction were performed without using sample identity.
CellRanger
Data from the NovaSeq6000 sequencer were processed using CellRanger v.8.0.0 using the cellranger count function with the default parameters. CellRanger reference index genome was built on genome assembly BDGP6.32 supplemented with the coding sequences for the expected transgenes (Supplementary Table 6). Before quality control, cell ranger total output was 805,970 cells. The median sample number of reads per cell was 39,142, the number of detected genes per cell 1,511 (min = 183, max = 8,543) and the median number of unique molecular identifiers (UMIs) per cell 4,893 (min = 666, max = 847,868).
... continue reading