Two distinct mechanisms silence chinmo in Drosophila neuroblasts and neuroepithelial cells to limit their self-renewal

Limitation of stem cell self-renewal during development ensures that organs reach their appropriate size. However, little is known about the temporal cues and downstream effectors that control stem cell activity during the early steps of tissue building. Recently, the chromatin-associated high mobility group protein HMGA2 has been shown to promote progenitor self-renewing potential in various mammalian tissues during development (Copley et al., 2013; Nishino et al., 2008; Parameswaran et al., 2014). During embryonic and early fetal stages, the RNA-binding proteins (RBPs) Imp1 and Lin28 post-transcriptionally promote Hmga2 expression in mouse cortical progenitors. In contrast, during late fetal stages, the microRNA let-7 promotes the progressive silencing of Hmga2 facilitating the termination of self-renewal in the cortex (Nishino et al., 2013; Yang et al., 2015). A similar post-transcriptional mechanism regulates Hmga2 and self-renewal in fetal hematopoietic progenitors (Copley et al., 2013). In addition, in such progenitors, the transcription factor RUNX1 is also known to silence Hmga2 during development (Lam et al., 2014). Thus, both transcriptional and post-transcriptional mechanisms operate to regulate the temporal expression of Hmga2 in the various progenitors allowing limited and controlled self-renewal during development. A strict control of these processes is essential as the deregulation of Hmga2 can promote unlimited self-renewal and tumorigenesis in these tissues (Fusco and Fedele, 2007). Yet, the mechanisms that regulate the temporal expression of Lin28, Imp1, Let-7 or Runx1 in NSCs and other progenitors during fetal development are still unclear. Moreover, although Hmga2 appears quite widely expressed in the central nervous system (CNS) during early mammalian development, it is still unclear whether the same temporal and regulatory mechanisms operate in the various regions of the CNS to limit NSC self-renewal.

The development of the Drosophila CNS is simpler than its mammalian counterpart, and is better understood. As such it represents a good model to investigate the basic principles limiting NSC self-renewal (Homem and Knoblich, 2012). Like Hmga2 in mammalian cortical progenitors, the BTB zinc-finger gene chinmo is highly expressed in Drosophila asymmetrically dividing NSCs of the ventral nerve cord (VNC) and central brain (CB), called neuroblasts (NBs), during early development, and its silencing during late development is necessary to limit NB self-renewing potential (Narbonne-Reveau et al., 2016). Interestingly, chinmo is also regulated at the post-transcriptional level in mushroom body neurons by the RNA-binding proteins (RBPs) Imp and Syncrip, and the let-7 miRNA (Kucherenko et al., 2012; Liu et al., 2015; Wu et al., 2012; Zhu et al., 2006). Similarly, Imp and Lin28 promote chinmo expression in NB tumors (Narbonne-Reveau et al., 2016). The post-transcriptional regulation of chinmo may be a general feature of NBs, as they also co-express Imp and lin28 during early larval stages, and Syncrip at later stages (Narbonne-Reveau et al., 2016; Syed et al., 2017). In particular, both chinmo and Imp need to be silenced during development to allow timely termination of NB self-renewal before adulthood (Narbonne-Reveau et al., 2016). Upstream of chinmo, Imp and lin28 lies a series of sequentially expressed transcription factors, known as temporal transcription factors for their ability to specify the birth-order of the various NB progeny generated upon successive asymmetric divisions (Isshiki et al., 2001; Kambadur et al., 1998). The sequential expression of temporal transcription factors is used as a timing mechanism to schedule, during late larval stages, the end of the Lin28⁺/Imp⁺/Chinmo⁺ expression window (Maurange et al., 2008; Narbonne-Reveau et al., 2016). Indeed, blocking sequential expression of temporal transcription factors leads to aberrant maintenance of Chinmo, Imp and Lin28, triggering unlimited NB self-renewal in adults (Maurange et al., 2008; Narbonne-Reveau et al., 2016).

Concomitant with NB asymmetric divisions in the VNC and CB during larval stages, a neuroepithelium (NE) first expands and then undergoes differentiation into neural progenitors that will form the optic lobes (OLs) in the brain. Part of this NE, named the outer proliferation center (OPC), will be converted into short-lived NBs that will generate the neurons of the medulla (Egger et al., 2007; Lanet et al., 2013; Yasugi et al., 2008). The NE-to-NB conversion in the OPC is initiated around mid-L3 by high levels of ecdysone produced by the ring gland after the larva reaches a critical weight (Lanet et al., 2013; Lanet and Maurange, 2014). Indeed, in addition to committing larvae to metamorphosis, ecdysone in the brain triggers the rapid progression of a differentiation wave throughout the NE, allowing the rapid differentiation of all NE cells into NBs (Lanet et al., 2013; Yasugi et al., 2010, 2008). Inactivation of ecdysone signaling in NE cells leads the unlimited persistence of a proliferative NE in adults. By limiting the self-renewal capacity of NE cells and promoting their rapid differentiation in NBs, ecdysone therefore limits the number of medulla NBs produced, consequently allowing the optic lobe to reach an appropriate final size (Lanet et al., 2013). Moreover, because ecdysone is produced in large quantities once the larvae has reached a critical mass, this mechanism coordinates the initiation of differentiation with organismal growth (Lanet and Maurange, 2014; Layalle et al., 2008; Mirth et al., 2005).

Thus, both cell-intrinsic and systemic signals are used to limit neural progenitors self-renewing potential in the different regions of the developing CNS in Drosophila. Yet it remains unclear whether similar effectors downstream of ecdysone or of the temporal transcription factor series operate in the various types of neural progenitors.

Here, we find that Chinmo not only regulates self-renewal in NBs but also in NE cells during early development. However, while the temporal regulation of chinmo in NBs relies on a post-transcriptional mechanism mainly controlled by a cell-intrinsic timer, its regulation in the NE is transcriptional and controlled by ecdysone. This bi-modal regulation of Chinmo allows NSC self-renewal to be promoted by the same master gene but controlled by different temporal strategies in the various regions of the brain.

While investigating the role of Chinmo in NBs of the VNC and CB, we noticed that it was also expressed and temporally regulated in the medulla NE. We performed a precise time course to investigate the temporal dynamics of chinmo expression in NBs relative to the NE. In VNC and CB NBs, chinmo is expressed from larval hatching up to the early L3 stage (Fig. 1A-C). However, we find that the silencing of chinmo in these NBs is not synchronous, suggesting a NB-intrinsic timing mechanism that is not coordinated between NBs (Fig. 1B). In the NE, chinmo is expressed from larval hatching, but remains expressed longer than in most NBs, undergoing a rapid and synchronous silencing around mid-L3 stages (between 12 and 24 h after the L2/L3 molt, Fig. 1C,D). Chinmo is also expressed for a short period of time in the first few medulla NBs converted from the NE around this time (Fig. 1C, asterisk). Thus, Chinmo in NE cells is silenced synchronously, suggesting that a systemic signal may coordinate synchronous Chinmo silencing in this region (Fig. 1E). Together, these observations suggest that the timing of chinmo silencing in the NE and NBs is controlled by different temporal mechanisms (Fig. 1F).

chinmo expression is known to be post-transcriptionally silenced in mushroom body NBs and neurons (Liu et al., 2015; Zhu et al., 2006). However, its mode of regulation is unclear in most NBs of the VNC and CB, as well as in the NE. We found, using fluorescent in situ hybridization, that in late L3, chinmo RNA can be detected in VNC NBs (data not shown) and in CB NBs, and in their surrounding late-born neurons (Fig. 2A, box 1), whereas the protein Chinmo is not produced at this stage. Moreover, the use of a lacZ enhancer trap inserted in the chinmo first exon and previously used to assess chinmo transcriptional activity (Flaherty et al., 2010; Zhu et al., 2006) indicated consistent lacZ expression in early and late L3 NBs (Fig. 2B-D). Together, this shows that chinmo is transcriptionally active in NBs throughout larval stages. Thus, a post-transcriptional mechanism operates to silence chinmo in most, if not all, late L3 NBs of the VNC and CB. To further investigate this, we generated transgenic Drosophila that allowed the conditional expression of a construct in which the mcherry-coding sequence is flanked by the 5′ and 3′ UTRs of chinmo (named UAS-mCherry^chinmoUTRs) (Fig. 2B). When transcribed in the VNC and CB NBs using nab-GAL4 UAS-mCherry^chinmoUTRs, we observed by immunostaining the strong expression of mCherry in NBs and their progeny up to early/mid-L3 (Fig. 2E). mCherry levels then rapidly decrease and the signal becomes almost undetectable in late L3 NBs (Fig. 2F). This contrasts with a GFP transgene (without the chinmoUTRs) that is concomitantly expressed at a constant level in NBs throughout larval stages (Fig. 2E,F). Finally, we misexpressed two different chinmo transgenes: UAS-chinmo^FL, which contains the ORF and the UTRs; and UAS-HA-chinmo, which contains only the ORF and lacks the 5′ and 3′ UTRs (Fig. 2B). Consistently, Chinmo is absent in most NBs of late L3 larvae when the UAS-chinmo^FL is expressed using nab-GAL4 (Fig. 2G). In contrast, Chinmo is highly expressed in NBs of late L3 larva when the UAS-HA-chinmo is expressed, leading to an amplification of NBs (Fig. 2H). Thus, the silencing of chinmo in late NBs of the VNC and CB is mainly mediated by a post-transcriptional mechanism through the UTRs.

We then tested whether chinmo was also regulated by a post-transcriptional mechanism in the NE. In contrast to VNC and CB NBs, chinmo mRNA is not detected by fluorescent in situ hybridization in NE cells and medulla NBs in late L3 larvae (Fig. 2A, box 2). In addition, when assessing expression of the chinmo-lacZ transgene, we found downregulation of LacZ around mid-L3 in the NE, coinciding with the downregulation of endogenous Chinmo (Fig. 3A,B). Thus, chinmo appears to be transcriptionally silenced in late L3 in the NE and medulla NBs. Moreover, when the mCherry^chUTRs transgene was transcribed in the NE using ogre-GAL4, we observed a strong mCherry expression at all stages of larval development (Fig. 3C,D). mCherry also persisted in the converted NBs from the NE. Thus, in contrast to VNC and CB NBs, post-transcriptional repression of chinmo is not operating in NE cells and medulla NBs (Fig. 2D). Finally, both the misexpression of UAS-chinmo^FL and UAS-HA-chinmo in the NE using ogre-GAL4 lead to high levels of Chinmo protein the NE (Fig. 3E,F), showing that there is no post-translational regulation of chinmo expression. All together, these data therefore demonstrate that chinmo is regulated by distinct mechanisms in two different regions of the brain. It is post-transcriptionally silenced in most, if not all, NBs of the VNC and CB, whereas it is transcriptionally silenced in the medulla NE and NBs.

Because Chinmo downregulation in the NE coincides with the critical weight (CW) and the subsequent production of ecdysone to initiate metamorphosis, we next tested whether ecdysone signaling could cell-autonomously silence chinmo. The ecdysone receptor (EcR) is continuously expressed in the NE throughout larval stages (Fig. 5E). Strikingly, misexpression of two dominant-negative forms of EcR (EcR^DN) (UAS-EcR.B1-DeltaC655.F645A and UAS-EcR.B1-DeltaC655.W650A), which are known to efficiently counteract ecdysone signaling (Cherbas et al., 2003), throughout the NE using ogre-GAL4 or in clones using MARCM (Lee and Luo, 1999), both led to the maintenance of Chinmo in the NE in late L3 larvae (Fig. 4A-C). Thus, ecdysone signaling cell-autonomously silences chinmo around mid-L3 in the NE. In contrast, expression of EcR^DN forms in MARCM and FLP-out clones or using nab-Gal4 did not lead to the persistence of Chinmo in late L3 NBs, although a slight delay in Chinmo downregulation was observed around early/mid-L3 (Fig. 4D,E, Fig. S1A,B). Similar results were obtained by abrogating CW-mediated ecdysone pulses using molting defective (mld^dts3) mutant larvae switched at 29°C from late L2 (Holden et al., 1986). In this case, we observed the persistence of Chinmo in the NE cells of late L3 larvae, but not in VNC and CB NBs (Fig. 4F,G compared with Fig. 1E). Thus, in contrast to the NE, ecdysone is not necessary for chinmo silencing in late L3 VNC and CB NBs, although it appears to facilitate the timely transition to a chinmo⁻ state (Fig. 4H). These experiments demonstrate that chinmo in NBs and in the NE is regulated by different mechanisms. In VNC and CB NBs, it is silenced at the post-transcriptional level by a NB-intrinsic temporal mechanism encoded by the sequential expression of temporal transcription factors. In NE cells, it is silenced at the transcriptional level by mid-L3 pulses of ecdysone produced after CW is reached.

We next sought to determine the function of Chinmo in the NE. We noticed that Chinmo downregulation correlates with the initiation of NE-to-medulla NB conversion triggered by ecdysone after the CW. Thus, shortly after the L2/L3 molt (before the CW), no or rare NE-to-NB conversion is observed (Fig. 5A,F). In contrast, upon downregulation of chinmo expression by RNAi in the NE from larval hatching using ogre-GAL4 UAS-chinmo^RNAi, we observed precocious NE-to-NB conversion in L2 and early L3 (Fig. 5B,F and Fig. S2A-C). This is accompanied by precocious medulla neuron production (Fig. 5C,D). Of note, similar results were obtained using two different RNAi lines provided by TRiP and NIG-Fly, although phenotypes were less penetrant with the NIG-FLY RNAi (Fig. S2B-D). Premature NE-to-NB conversion during early larval stages upon chinmo knockdown could be due to the precocious establishment of the signaling pathways responsible for the pro-neural wave. Alternatively, these pathways may be pre-established during early larval stages and free to operate when chinmo is knocked down from early L2. To investigate this issue, we stained the NE for PointedP1 (PntP1), which is downstream of the EGFR pathway and is required to initiate and propagate the proneural wave (Yasugi et al., 2010), and for Lethal of Scute [L(1)sc], which labels NE cells at the wavefront (Yasugi et al., 2008). Strikingly, both markers are already expressed in the NE of wild-type L2 larvae, demonstrating that the signaling for NE-to-NB differentiation is pre-established early on, before chinmo downregulation (Fig. 5E). Thus, Chinmo in the early NE prevents precocious NE differentiation by blocking the propagation of the pre-established proneural front.

We also found that chinmo knockdown led to smaller and less proliferative NE cells in early L3, showing that Chinmo is required for cell growth and to boost mitotic activity (Fig. 5G-I and Fig. S2D). Consequently, downregulation of chinmo in the early NE led to a smaller NE and fewer medulla NBs in late L3 (Fig. 5G,I-M), resulting in a smaller optic lobe in adults (Fig. 5N-P). Thus, together these data show that Chinmo promotes NE expansion before the CW by stimulating cell growth and proliferation, and by preventing precocious differentiation (Fig. 5Q).

We next investigated the impact of a temporal deregulation of chinmo expression on the NE. When generating MARCM clones misexpressing chinmo (chinmo^OE) from early L2 to late L3 stages, we found a delayed conversion of the NE compared with the surrounding tissue (Fig. 6A). A similar repression of NE conversion was also observed when chinmo is misexpressed in the whole NE (OPC) throughout development (ogre>chinmo^OE, Fig. 6B,C,E). However, under such conditions, the chinmo^OE NE was only slightly larger than the wild-type NE in late L3 (Fig. 6B-D). This small difference seemed inconsistent with the strong repression of NE differentiation that is observed (Fig. 6B,C). We detected high levels of apoptosis in ogre>chinmo^OE, possibly explaining this phenotype (Fig. 6F). Apoptosis inhibition by misexpressing Baculovirus p35 (ogre>chinmo^OE, p35) led to the massive overgrowth of NE in late L3 with few medulla NBs being converted, consistent with the ability of chinmo to prevent differentiation and boost cell growth (Fig. 6G).

Strikingly, although the NE is normally eliminated during metamorphosis due to its complete conversion in medulla NBs, misexpression of chinmo in L2-induced MARCM clones prevented the elimination of the NE, leading to perdurance of a proliferative NE in adult optic lobes (Fig. 6H). Thus, downregulation of Chinmo during development is necessary to allow efficient NE-to-NB conversion, leading to NE elimination by the end of development.

Medulla NBs converted from the NE sequentially express five temporal transcription factors [Homothorax (Hth), Eyeless (Ey), Sloppy-paired (Slp), Dichaete (D) and Tailless (Tll)] as they age, allowing the generation of a repertoire of neurons (Li et al., 2013). We have observed that the very first medulla NBs generated around the CW transiently express Chinmo (Fig. 7A). We decided to investigate the temporal identity of these NBs and detected Hth in the NBs, which were early born (Fig. 7A). Thus, Chinmo does not need to be downregulated in medulla NBs to initiate temporal patterning. Moreover, premature NBs induced in early L3 by chinmo knockdown equally initiate and progress throughout the temporal series as we find that they can express Hth, D and Tll (Fig. 7B,G,H). In addition, when chinmo is misexpressed in the NE, the few medulla NBs that are converted at a low rate appear also able to initiate and progress throughout temporal patterning as they express the temporal factors D and Tll (Fig. 7G,H). Because the conversion is delayed when chinmo is misexpressed, only very few medulla NBs express the last factor Tll in late L3 compared with the oldest wild-type medulla NBs at the same time (Fig. 7H). All these results show that Chinmo downregulation is not necessary to initiate or to progress through the temporal series in medulla NBs.

Here, we show that self-renewal in the different types of neural progenitors present in the Drosophila CNS is promoted during early development by the same transcription factor, Chinmo. However, the expression of chinmo is controlled by different regulatory strategies. This system allows a core self-renewing program to be temporally regulated by distinct intrinsic and extrinsic cues in the various regions of the Drosophila brain.

We have previously shown that misexpression of chinmo in NBs is sufficient to promote their unlimited self-renewal and that aberrant expression of chinmo in NB tumors induced by dedifferentiation is responsible for their unrestricted growth potential (Narbonne-Reveau et al., 2016). These data indicate that Chinmo confers an unlimited self-renewing potential to NBs of the VNC and CB. Silencing of chinmo by progression of temporal transcription factors is therefore necessary to limit NB self-renewal during development. Here, we report that chinmo is also expressed during early larval stages in the expanding NE of the OPC that will form the medulla region of the optic lobe in the brain. In the NE, Chinmo favors cell growth and proliferation, and appears to counteract differentiation. Indeed, loss of Chinmo in the NE of L2 and early L3 larvae is sufficient to induce premature NE-to-NB transition. We have observed that, during normal development, L1sc, which labels NE cells at the front of the proneural wave, and PntP1, which is downstream of the EGFR pathway and is required for the initiation and progression of the pro-neural wave, are already expressed in the NE of L2 larvae, before medulla NBs are being produced. Thus, Chinmo appears to counteract the pre-established pro-neural wave in order to prevent precocious NE-to-NB transition, therefore allowing NE expansion from L1 to early L3. The mode of action of Chinmo remains unknown. The knockdown of cell cycle genes has been shown to promote precocious NE-to-NB conversion similar to chinmo knockdown (Zhou and Luo, 2013). Further work aiming at identifying Chinmo transcriptional targets should help elucidate whether Chinmo prevents differentiation by promoting cell cycle progression and/or by interfering with targets of the EGFR, JAK/STAT, Notch and Hippo signaling pathways that regulate proneural wave progression (Egger et al., 2010; Reddy et al., 2010; Yasugi et al., 2010, 2008).

From mid-L3 stages, Chinmo is then silenced. This triggers the sudden acceleration of the pro-neural wave, leading to the progressive exhaustion of NE cells through their differentiation into medulla NBs – which generate a much shorter lineage than VNC and CB NBs. In contrast, continuous misexpression of chinmo in the NE induces its continuous expansion throughout larval stages and maintenance of NE self-renewal in the adult brain. Although Chinmo acts as a brake on the NE-to-NB conversion, it does not seem to interfere with the establishment and progression of the series of temporal transcription factors in medulla NBs. Indeed, loss of Chinmo in NE gives rise to precocious medulla NBs that progress through the Hth→Slp→Ey→D→Tll series, similar to the few NBs that can be converted from NE cells misexpressing chinmo. Consistently, the few NBs that are produced and still express chinmo in early L3 also exhibit Hth expression.

Together, these data indicate that both in the NE of the OPC and in NBs of the VNC and CB, Chinmo expression confers unlimited self-renewal, and its silencing during larval stages ensures the timely elimination of NBs and NE by the end of development. The general role of a single ‘master’ transcription factor in promoting self-renewal in different types of neural progenitors suggests that the same core set of target genes governs self-renewal, independently of the progenitor type.

Because Chinmo is temporally regulated during development and promotes self-renewal of neural progenitors, it appears to have a role reminiscent of some mammalian oncofetal genes, such as Hmga2, Mycn or Zbtb17. These genes all promote neural progenitor proliferation during early development, albeit through different mechanisms. HMGA2 formats chromatin structure, MYCN is a transcription factor that activates genes required for protein biogenesis, and MIZ1 is a Myc co-factor that transforms Myc into a transcriptional repressor of differentiation genes (Boon et al., 2001; Kerosuo and Bronner, 2016; Kishi et al., 2012; Knoepfler et al., 2002; Nishino et al., 2008).

In addition, like Chinmo in VNC and CB NBs, both HMGA2 and MYCN are regulated by RBPs of the IMP (also known as IGF2BP) and LIN28 families in neural progenitors (Bell et al., 2015; Copley et al., 2013; Li et al., 2012; Liu et al., 2015; Molenaar et al., 2012; Narbonne-Reveau et al., 2016; Nishino et al., 2013; Yang et al., 2015). This emphasizes the striking conservation throughout evolution of the post-transcriptional regulation of self-renewal genes by IMP and LIN28 proteins during early development and tumorigenesis.

Therefore, even though no clear orthologs of Chinmo in mammals and of MYCN, HMGA2 and MIZ1 in insects have been identified (although MIZ1 is a BTB transcription factor with 32% homology with Chinmo), elucidating the mode of action of Chinmo should help to reveal ancestral and generic mechanisms underlying the transcriptional control of stem cell self-renewal.

Our work indicates that chinmo is under different modes of regulation in NBs and NE cells. In VNC and CB NBs, chinmo is post-transcriptionally regulated. Important players in this post-transcriptional regulation could be the RBPs Imp, which could promote chinmo expression in early larval NBs, and Syncrip, which could repress chinmo in late larval NBs. Both RBPs indeed antagonistically regulate chinmo expression in mushroom body neurons and are, respectively, expressed in early and late NBs (Liu et al., 2015; Syed et al., 2017; Yang et al., 2017). In contrast, chinmo is mainly regulated at the transcriptional level in the NE of the optic lobe that expands during early larval development. We had previously shown that ecdysone signaling is strongly activated in the NE shortly after the CW (about 12 h after the L2/L3 molt) (Lanet et al., 2013). We now demonstrate that one of the main roles of ecdysone signaling at this stage is to transcriptionally silence chinmo, therefore limiting NE self-renewal and allowing progression of the pro-neural wave. Chinmo downregulation by ecdysone in the NE does not seem to involve the JAK/STAT pathway, as we did not observe any upregulation of JAK/STAT activity, by measuring levels of Stat92E (Flaherty et al., 2010), in the EcR^DN context (data not shown). In addition, it is likely that ecdysone also regulates other targets in parallel, such as Delta (Lanet et al., 2013) because manipulation of chinmo expression alone did not affect Delta expression (data not shown).

Blockage of ecdysone signaling through the misexpression of different forms of EcR^DN, or prevention of ecdysone production in the mld^DTS3 context, systematically led to the permanent maintenance of Chinmo in the NE of late L3 larvae. In contrast, under similar conditions, chinmo silencing is only delayed by a few hours in VNC and CB NBs. A similar role for ecdysone signaling in promoting chinmo silencing in NBs has also been recently described (Syed et al., 2017). It can be noted, however, that in the former study, the delay in chinmo silencing appeared more pronounced than we observed, despite the use of the same UAS-EcR^DN constructs. The underlying reasons for the differences observed with our study are unclear, but may, for example, reside in different fly food compositions or experimental temperatures, both of which can strongly influence developmental transitions. Further studies will be needed to clarify this point. At this stage, we favor a model in which EcR signaling in VNC and CB NBs is dispensable but facilitates chinmo silencing that is triggered by temporal series progression. The timely silencing of chinmo promoted by the conjunction of temporal patterning and ecdysone signaling may be important for the precise regulation of glial cell numbers in some CB lineages (Syed et al., 2017).

Bi-modal (transcriptional and post-transcriptional) regulation of chinmo allows stem cell self-renewal to be under the control of the same master transcription factor (Chinmo), and therefore under the same transcriptional program, while being regulated by different cell-intrinsic or extrinsic cues in the different regions of the brain (Fig. 8). Consequently, self-renewal in the NE appears to be directly controlled by environmental cues, such as nutritional conditions and hormones, whereas other regions of the CNS, such as the CB and VNC may be less affected. Interestingly, the regulation of self-renewal by ecdysone signaling in the OL also reveals a mechanism by which endocrine disruptors may affect more heavily the development of specific regions of the brain (Préau et al., 2015).

Whether the temporal regulation of genes promoting neural progenitor self-renewal during early mammalian development is under the control of a NSC-encoded series of transcription factors or relies on hormonal or more localized cues is still unclear. Interestingly, retinal progenitors in mice have recently been shown to sequentially express orthologs of Drosophila temporal transcription factors that specifies the fate of their progeny (Mattar et al., 2015), and NE cells in the developing cortex relies on retinoic acids produced by surrounding meninges (Siegenthaler et al., 2009). Thus, as for Drosophila, different temporal mechanisms may govern the self-renewing potential of neural progenitors in the various regions of the developing nervous system in mammals.

Drosophila stocks were maintained at 18°C on standard medium (8% cornmeal, 8% yeast, 1% agar).

Confocal images were acquired on a Zeiss LSM780 microscope. FIJI was used to process confocal data, and to compile area and volume data.

For each experiment, at least three biological replicates were performed. Biological replicates are defined as replicates of the same experiment with flies being generated by different parent flies. For all experiments, we performed a Mann–Whitney test for statistical analysis, except for Fig. S2D, where a t-test was performed. No data were excluded. Statistical tests were performed with the online BiostaTGV (http://marne.u707.jussieu.fr/biostatgv/). Results are presented as dot plots, also depicting the median in red and a boxplot in the background (Whisker mode: 1.5IQR). The sample size (n), the mean±s.e.m. and the P value are reported in the figure legends (****P≤0.0001, ***P≤0.001, **P≤0.01 and *P≤0.05).

Experiments were performed at 29°C. yw line is used as a control. For generating UAS-EcR.B1-DeltaC655.F645A (in this study called UAS-EcR^DN) (Cherbas et al., 2003) MARCM clones (Lee and Luo, 1999), we used w, tub-GAL4, UAS-nlsGFP::6xmyc::NLS, hsFLP¹²²; FRT82B, tubP-GAL80/TM6B crossed to UAS-EcR^DN/CyO_ActGFP; FRT82B/TM6 (from Bloomington #6869). The progeny of the above crosses was heat-shocked 1 h at 37°C just after larval hatching and raised at 29°C. Similar results are obtained with UAS-EcR.B1-DeltaC655.W650A (Bloomington #6872). Flip-out clones were generated using hs-FLP; Act5c>CD2>GAL4, UAS-GFP (from N. Tapon, The Francis Crick Institute, London, UK) with UAS-chinmo^FL (Bloomington #50740) or UAS-EcR^DN. The progeny of these crosses was heat-shocked 1 h at 37°C just after larval hatching and raised at 29°C. The GAL4 lines used were nab-GAL4 (#6190 from Kyoto DGRC; Maurange et al., 2008) and ogre-GAL4 (GMR29C07-GAL4, Bloomington #49340), which is active in the OPC, at all larval stages. wor-GAL4; ase-GAL80 (a gift from J. Knoblich, Institute of Molecular Biotechnology, Vienna, Austria) is only active in type II NBs in the central brain. The UAS lines used were: UAS-chinmo^FL, UAS-HA-chinmo (Flaherty et al., 2010), UAS-EcR^DN and UAS-chinmo^RNAi (TRiP #HMS00036, Bloomington #33638 or NIG-Fly #17156R-1). UAS-dicer2 (Bloomington #24650 and #24651) was used in combination with GAL4 lines in order to improve RNAi efficiency. UAS-mCD8::GFP (Bloomington #32186 and #2185) and UAS-mCD8::RFP (Bloomington #27399) were used to follow the driver expression. The progeny of the above crosses was raised at 29°C. The chinmo-lacZ (Bloomington #10440) line was used to monitor chinmo transcription. mld^dts3 mutant was used to abrogate CW-induced ecdysone pulses (Bloomington #3014).

The larval stages are standardized using morphological criteria. Early L3 are selected just after the L2/L3 molt: the early L3 larvae are the same size as late L2 larvae, but have everted spiracles. Late L3 were selected 48 h after L2/L3 molt.

The 5′ and 3′ untranslated transcribed regions (UTRs) of chinmo were cloned on both sides of the mcherry-coding sequence, under the regulation of upstream activation sequences (UAS) using the In-Fusion HD Cloning protocol (In-Fusion HD Cloning kit, Clontech). The entry vector used was pUASTattB-PmeI (a gift from Jean-Marc Philippe, Institut de Biologie du Développement de Marseille, France). The chinmo 5′UTR and 3′UTR sequences were obtained from the EST clone pFLC1-RE59755 (Berkeley Drosophila Genome Project, GOLD collection). The mcherry reporter gene was obtained from the pBPGUw-mCherry plasmid (a gift from I. Lohmann, University of Heidelberg, Germany) (Sorge et al., 2012). The primers used were: Ch-5′UTR_F (F1) ATTCGTTAACAGATCTAGTCAAAAAGAAACTGCCGTG; Ch-5′UTR_R (R1) GCCCTTGCTCACCATGGTGCCAGCAGTGATGCT; mCherry_F (F2) ATGGTGAGCAAGGGCGAG; mCherry_R (R2) TGTTGCGGCTGCTTCTTACTTGTACAGCTCGTCCATGC; Ch-3′UTR_F (F3) GAAGCAGCCGCAACAGCA; and Ch-3′UTR_R (R3) ACAAAGATCCTCTAGAGGTGAATTTTCATTTGTACGAAGAA. Bold and underlines indicate sequences used for PCR amplification and overlapping sequences for In-Fusion, respectively.

Dissected tissues were fixed 5 to 15 min in 4% formaldehyde/PBS depending on the primary antibody. Staining was performed in 0.5% Triton/PBS with antibody incubations separated by several washes. Tissues were then transferred in Vectashield with or without DAPI for image acquisition. Primary antibodies were: chicken anti-GFP (1:1000, Aves #GFP-1020), rabbit anti-RFP (1:500, Rockland #600-401-379), rat anti-RFP (1:500, Chromotek #5F8), mouse anti-Miranda (1:50, A. Gould, Francis Crick Institute, London, UK), rabbit anti-PH3 (1:500, Millipore #06-570), rat anti-PH3 (1:500, Abcam #AB10543), rat anti-Elav (1:50, DSHB #9F8A9), rat anti-DECadherin (1:50, DSHB #DCAD2), mouse anti-Repo (1:200, DSHB #8D12), mouse anti-EcR^com (1:7, DSHB #Ag10.2 and #DDA2.7), rabbit anti-βGalactosidase (1:1000, Cappel #559562), rabbit anti-ζPKC (1:100, Santa Cruz Biotechnology #sc-216), rat anti-L(1)sc (1:50, A. Carmena), rabbit anti-PntP1 (1:500, J. B. Skeath, Washington University School of Medicine, St Louis, MO, USA), rabbit anti-cleaved Dcp-1 (1:500, Cell Signaling #9578), rabbit anti-Tll (1:100, J. Reinitz, University of Chicago, OH, USA), guinea-pig anti-D (1:50; Maurange et al., 2008), rabbit anti-Hth (1:50, A. Saurin, Institut de Biologie du Développement de Marseille, France), rat anti-Chinmo (1:500, N. Sokol, Indiana University, Bloomington, USA) and guinea-pig anti-Chinmo (1:500, N. Sokol). Adequate combinations of secondary antibodies (Jackson ImmunoResearch) were used to reveal expression patterns.

Sens and antisense digoxigenin-labeled riboprobes (DIG RNA Labeling MIX, Roche) against one exonic region of the chinmo transcript (CG31666 RD) were generated (Zhu et al., 2006). The primers used were: Chinmo_probe (F1): TAATACGACTCACTATAGGACGACCAAGCTGGACAAGAAGCC; Chinmo_probe (F2): TAATACGACTCACTATAGGGTTTGGTTTGGTTTGGTTTGGATTTG. Bold and underlines indicate sequence used for chinmo amplification and the T7 promoter sequence, respectively.

The labeled RNAs were detected by anti-DIG-POD antibody (1:1000, Roche) and visualized with Cy3-tyramide (1:500, PerkinElmer), as previously described (Daul et al., 2010). Tissues were also immunostained using chicken anti-GFP (1:1000, Aves #GFP-1020) and rabbit anti-ζPKC (1:100, Santa Cruz Biotechnology #sc-216) antibodies.

We are grateful to E. A. Bach, A. Carmena, A. Gould, J. Knoblich, T. Lecuit, F. Schnorrer, J. B. Skeath, J. Reinitz, A. Saurin and N. Sokol for flies, plasmids and antibodies. We also acknowledge the Bloomington Drosophila Stock Center (NIH P40OD018537), the Vienna Drosophila RNAi Center (VDRC), TRiP at Harvard Medical School (NIH/NIGMS R01-GM084947), the Berkeley Drosophila Genome Project and the Kyoto DGRC Stock Centers, and the Developmental Studies Hybridoma Bank (DSHB). We thank France-BioImaging/PICsL infrastructure (ANR-10-INSB-04-01), E. Jullian and V. Thomé for technical advice, and C. Gaultier and S. Genovese for critical reading of the manuscript.