The brain is consisted of diverse neurons arising from a limited number of neural stem cells. Drosophila neural stem cells called neuroblasts (NBs) produces specific neural lineages of various lineage sizes depending on their location in the brain. In the Drosophila visual processing centre - the optic lobes (OLs), medulla NBs derived from the neuroepithelium (NE) give rise to neurons and glia cells of the medulla cortex. The timing and the mechanisms responsible for the cessation of medulla NBs are so far not known. In this study, we show that the termination of medulla NBs during early pupal development is determined by the exhaustion of the NE stem cell pool. Hence, altering NE-NB transition during larval neurogenesis disrupts the timely termination of medulla NBs. Medulla NBs terminate neurogenesis via a combination of apoptosis, terminal symmetric division via Prospero, and a switch to gliogenesis via Glial Cell Missing (Gcm), however, these processes occur independently of each other. We also show that temporal progression of the medulla NBs is mostly not required for their termination. As the Drosophila OL shares a similar mode of division with mammalian neurogenesis, understanding when and how these progenitors cease proliferation during development can have important implications for mammalian brain size determination and regulation of its overall function.

Fly stocks and husbandry

The fly strains used were detailed in the extended Materials and Methods in the paper. Fly stocks were reared on standard media at 25^oC. For larval dissection, brains were dissected at wandering L3 stages (approximately 120h ALH). For pupal dissection, white pupae were selected and allowed to age to the desired stages.

 Knockdown and overexpression experiments using eyR16F10-GAL4 were moved to 29^oC after overnight embryo collection until dissection. Experiments using tub-GAL80^ts;eyR16F10-GAL4, embryos were reared at 25^oC for one day, larvae were then moved to repressive 18^oC for 5 days to inhibit GAL4 activity, and transferred to permissive 29^oC to induce GAL4-mediated transgene expression for 2 days prior to dissection. Knockdown and overexpression flip-out clones were induced at 24h ALH at 37^oC for 10 min and then allowed to develop at 29^oC until dissection except for pros RNAi clones that were let developed at 25^oC to allow a mild knockdown of pros. FRT42D and FRT42D;UAS-tll MARCM clones were induced at 24h ALH (Figure 2A-D) or 48h ALH (Figure 5-figure supplement 2A-B) at 37^oC for 15 min. FRT82B and FRT82B,pros¹⁷ MARCM clones were induced at 24h ALH at 37^oC for 8 min.

Immunostaining

Larval and pupal brains were dissected in phosphate buffered saline (PBS), fixed in 4% formaldehyde for 20 min and rinsed three time in 0.3% or 0.5% PBST (PBS + 0.5% Triton), respectively at room temperature. For immunostaining, brains were incubated in primary antibodies overnight at 4^oC, followed by two washes in 0.3-0.5% PBST and then an overnight secondary antibody incubation at 4^oC or 3h at room temperature. Brains were afterwards rinsed two times in 0.3-0.5% PBST and incubated in 50% glycerol in PBS for 20 min. Samples were mounted in 80% glycerol in PBS, on glass slides and sealed with coverslips (22x22 mm, No.1.5, Knittel) for image acquisition. The primary antibodies used were: rat anti-Dpn (1:200, Abcam 195172), chick anti-GFP (1:1000, Abcam 13970), mouse anti-EcR^common (1:50, DSHB Ag10.2), rabbit anti-Dcp-1 (1:100, Cell Signalling 95785), rat anti-Mira (1:100, Abcam 197788), rat anti-Pros (a gift from Fumio Matsuzaki), rabbit anti-ß-Galactosidase (1:200, a gift from Helena Richardson), rabbit anti-Ase (a gift from Lily Jan and Yuh Nung Jan), mouse anti-Repo (1:50, DSHB 8D12), rabbit anti-PatJ (1:500, a gift from Helena Richardson), anti-Ey (1:50, DSHB), guinea pig anti-Slp (1:200, a gift from Kuniaki Saito), rabbit anti-Tll (1:200, a gift from Kuniaki Saito). Secondary donkey antibodies conjugated to Alexa 555 and Alexa 647, and goat antibodies conjugated to Alexa 405, 488, 555, and 647 (Molecular Probes) were used at 1:500.

Image acquisition and processing

Images were acquired using Olympus FV3000 confocal microscope with 40x (NA 0.95, UPLSAPO) and 60x (NA 1.30, UPLSAPO) objectives. Δz = 1.5 μm. Images were processed using Fiji (https://imagej.net/Fiji).

 Quantification was performed in Fiji or via 3D reconstruction in Imaris (Bitplane). To quantify NB numbers (Figure 1C, 4E, Figure 1-figure supplement 1N, 2E, Figure 3-figure supplement 1G), the numbers of Dpn⁺ cells in the OPC were measured. To quantify the nuclear sizes of the medulla NBs (Figure 1E), nuclear diameters identified by Dpn were measured. To quantify NB and NE volumes in the OPC (Figure 2A), the Dpn⁺ and PatJ⁺ volumes were measured for the NBs and the NE, respectively. To quantify the ratio of NBs (Figure 2J, 2M, 3P, 4I, 5I, 5P, Figure 1-figure supplement 1E, Figure 5-figure supplement 2I) or glia (Figure 5D) in clones, GFP⁺ clones were first detected and GFP⁺ volumes were measured. The GFP⁺ areas were then utilised to make a mask for Dpn⁺ or Repo⁺ cells within the clones, which was in turn used to measure NB or glial cell numbers, respectively. The NB or glial ratio in clones is calculated as the ratio of the number of GFP⁺Dpn⁺ or GFP⁺Repo⁺ cells over the GFP⁺ volume. To quantify the distributions of apoptotic NBs expressing Ey, Slp, and Tll (hereby referred as tTF) (Figure 3K), Dpn::GFP⁺Dcp-1⁺ cells were first identified and used as a mask for cells that are positive for tTFs. The ratio of each tTF is calculated as the ratio of the number of GFP⁺Dcp-1⁺tTF⁺ cells over the number of GFP⁺Dcp-1⁺ cells. To quantify the number of glia in the medulla (Figure 5-figure supplement 2H), the total numbers of Repo⁺ cells in the deep section of the medulla were measured. Images were assembled in Affinity Publisher 2. Schematics were created in Affinity Design 2. Scale bars = 10 μm or 50 μm per indicated in figures.

Statistical analyses

The biological replicates in our experiments are represented by a minimum of three animals per genotype. The n number reflects brain lobe numbers unless otherwise specified. Statistical analyses were performed, and graphs were plotted in GraphPad Prism 9. In graphs, data is represented as mean ± standard error of the mean (SEM). For comparisons between two conditions, P-values were calculated by non-parametric Mann-Whitney tests when data if not normally distributed. For comparisons between more than two experimental conditions, P-values were calculated by ordinary one-way ANOVA tests for normally distributed data or data with small sampling size, i.e. n<5 (Figure 1C, 2E). For non-normally distributed data, Kruskal-Wallis tests were employed. Holm-Šídák's or Dunn’s tests were used to correct for multiple comparisons following one-way ANOVA and Kruskal-Wallis tests, respectively. For the comparisons of apoptotic NBs expressing different temporal factors at 12-16h APF (Figure 3K), ordinary two-way ANOVA test was used, followed by Šídák's test for multiple comparisons. (*) p<0.05 (**) p< 0.01, (***) p< 0.001, (****) p < 0.0001, (ns) p>0.0. If quantifications were not performed, the number of samples in which specific phenotypes were observed, is mentioned in the text.