Minibrain drives the Dacapo-dependent cell cycle exit of neurons in the Drosophila brain by promoting asense and prospero expression

Cell proliferation, specification and terminal differentiation must be precisely coordinated during brain development in order to ensure the correct balance between the number of cells in different populations and the subsequent formation of functional circuits. When the coordination between these processes is impaired, aberrant tissue development, neuropathologies and tumorigenesis may ensue (Zhu and Skoultchi, 2001; Buttitta and Edgar, 2007; Malumbres and Barbacid, 2009; Hindley and Philpott, 2012).

Given its intermediate complexity and the sophisticated genetic tools available, the larval central nervous system (CNS) of Drosophila provides a suitable experimental model in which to study the genetic basis underlying neurodevelopment (Gonzalez, 2007; Sousa-Nunes et al., 2010; Brand and Livesey, 2011; Homem and Knoblich, 2012). The larval CNS is composed of two brain hemispheres and the ventral ganglia. The primordium of the adult central brain (CB) develops in the medial region of each hemisphere, whereas the adult optic lobes (OLs) develop from the primordia located laterally in the hemispheres (summarized in Fig. S1).

Most OL neurons are generated at the end of larval development from neural progenitors (NPs) called neuroblasts (NBs) that delaminate progressively from two neuroepithelia during the third instar larval period – the OPC (outer proliferation center) and the IPC (inner proliferation center). The OPC is located laterally on the OL surface and it gives rise to the lamina precursor cells (LPCs) and medulla neurons, whereas the IPC, which is located inside the lobe, generates the lobula neurons (Hofbauer and Campos-Ortega, 1990; reviewed by Meinertzhagen and Hanson, 1993; Apitz and Salecker, 2014). The adult CB originates from ∼100 NBs located medially in each hemisphere. These NBs are born in the procephalic region during embryogenesis, and, after generating the functional neurons of the larval brain, most of them enter quiescence at the end of embryogenesis. These CB NBs re-enter the cell cycle progressively from the first to third larval instar and they stop proliferating on the first day of pupation (White and Kankel, 1978; Truman and Bate, 1988; Hofbauer and Campos-Ortega, 1990; Ito and Hotta, 1992; Datta, 1995; Ebens et al., 1993).

Based on their cellular and molecular properties, two main types of larval NBs have been described. The ventrally located Type I CB NBs and OPC NBs proliferate like embryonic NBs, although they produce more cells with more restricted fates within each lineage. Thus, they divide asymmetrically several times to generate a new NB and an intermediate progenitor called ganglion mother cell (GMC), which divides terminally to generate two post-mitotic ganglion cells (GCs) that most frequently differentiate into neurons (Ito et al., 1997; Ceron et al., 2001; Colonques et al., 2007). Type II NBs are located in the dorso-posterior medial region and they self-renew by asymmetrical division, budding-off intermediate progenitors that undergo multiple divisions (Bello et al., 2008; Boone and Doe, 2008; Bowman et al., 2008).

Adult mutant minibrain (mnb) flies have a smaller brain but with no gross alterations in its overall organization (Fischbach and Heisenberg, 1984; Tejedor et al., 1995). This change in size is particularly evident in the ventro-anterior CB and the OL, and is due to neuronal deficits that have been related to altered proliferation in the larval neurogenic centers, particularly in the OPC, suggesting that Mnb regulates neural proliferation and neurogenesis during post-embryonic CNS development (Tejedor et al., 1995). Nevertheless, the cellular and molecular mechanisms underlying this phenotype remain mostly unknown.

The mnb gene encodes a dual protein kinase that is strongly conserved from insects to humans (reviewed by Galceran et al., 2003), and it has been implicated in vertebrate neurodevelopment (reviewed by Tejedor and Hämmerle, 2011) and cell cycle regulation (reviewed by Becker, 2012). Interestingly, the truncation of DYRK1A in humans causes microcephaly (Møller et al., 2008) and autosomal dominant mental retardation 7 (MRD7; OMIM 614104; van Bon et al., 2011), which is reminiscent of the mnb loss-of-function (LoF) phenotype in Drosophila. Remarkably, the human DYRK1A gene lies in the so-called Down syndrome (DS) critical region on chromosome 21 (Guimerâ et al., 1996) and it has been related to several neurodevelopmental pathological aspects of DS (reviewed by Tejedor and Hämmerle, 2011). Thus, the MNB/DYRK1A kinase is currently considered a DS therapeutic target (Becker et al., 2014). Therefore, defining the cellular processes and molecular mechanisms in which MNB/DYRK1A is involved could be crucial for the design of such therapeutic strategies.

There is compelling evidence that Dyrk1A is a crucial regulator of the transition from mitotically active precursors to terminally differentiated neurons in the vertebrate CNS (reviewed by Tejedor and Hämmerle, 2011). In particular, we found that Dyrk1A regulates the cell cycle exit of vertebrate CNS neurons by upregulating the expression of the cyclin-dependent kinase inhibitor (CKI) p27^Kip1 (Cdkn1b) (Hämmerle et al., 2011). Given the strong evolutionary conservation of Mnb/Dyrk1A, we have performed genetic, cellular and molecular analyses in the Drosophila larval CNS to unravel the molecular mechanisms underlying this function.

Given the phenotype of mnb mutants, this study focused on the OPC and the ventral-anterior CB.

To define the developmental stage at which neuronal deficits are generated in mnb mutants, we carried out in vivo pulse/chase bromodeoxyuridine (BrdU) labeling experiments (Fig. S2). As a decrease in BrdU⁺ cells was only detected when pulses were applied at late larval stages, we concluded that the neuronal deficit of mnb flies is mainly generated during late post-embryonic neurogenesis. Indeed, we previously found that changes in the pattern of BrdU labeling were evident in the OL of mnb late third-instar larval brains after a 1 h BrdU pulse, although the number of BrdU-labeled cells was often not substantially altered (Tejedor et al., 1995). As we subsequently found the cell cycle of OL NBs to be shorter than 1 h (Ceron et al., 2001), we applied 5-10 min BrdU or 5-ethynyl-2′-deoxyuridine (EdU) pulses in vitro to dissected late larval brains. Surprisingly, under these conditions there were more proliferating (BrdU- or EdU-labeled) cells in the larval brain hemispheres of mnb mutants than in those of wild-type (wt) controls (Fig. 1A-C; Fig. S3), particularly in the OPC and CB. We also observed a similar increase in cells expressing other cell cycle markers such as cyclin E (Fig. 1D,E). Conversely, we found that Mnb overexpression inhibited the proliferation of NBs in the OPC and CB, reflected by a reduction in the number of BrdU/EdU-labeled cells (Fig. 1C,F-K) and weaker cyclin expression (Fig. 1L,M). These results demonstrate that Mnb negatively regulates neural proliferation during post-embryonic neurogenesis.

The increase in proliferating cells in the late larval brain of mnb mutants is intriguing given the neuronal deficits in the adult brain and the decreased BrdU labeling in pulse/chase experiments from late larvae to adults (Fig. S2E,F). This apparent paradox could be explained by a high incidence of apoptotic cell death in the OPC and CB of mnb mutants, as determined by terminal deoxynucleotide transferase-mediated dUTP nick-end labeling (TUNEL) analysis (Fig. 1N-P), Acridine Orange staining, electron microscopy (Fig. S5A-D) and activated caspase-3 immunolabeling (not shown).

We studied mnb expression in the late larval CNS in detail by fluorescence in situ hybridization (FISH) and immunohistochemistry (IHC) in combination with markers for NBs/GMCs [Grainyhead (Grh)], mitotic NBs/GMCs [phosphorylated histone-3 (PH3)], newborn GCs [Prospero (Pros); Colonques et al., 2011] and early differentiating GCs/neurons (Elav; Robinow and White, 1991). We found strong mnb mRNA expression in scattered OPC and CB cells (Fig. 2A). These cells were located beneath the NB/GMC layers in the OPC and close to CB NBs/GMCs (Grh⁺, PH3⁺) that lacked mnb mRNA (Fig. 2C-G), suggesting that mnb is mainly expressed by GCs. As the GC markers did not work properly after FISH, this was assessed by IHC with a specific Mnb antiserum that labeled cells throughout the larval CB and OPC (Fig. 2H). This cellular pattern was more extensive than the restricted distribution of mnb mRNA-expressing cells, probably reflecting the higher sensitivity of IHC, as the specificity of both patterns was evident by the widespread reduction in mnb hypomorph alleles (Fig. 2A,B,H,I). Indeed, consistent with the expression of Mnb in GCs, Mnb immunostaining was detected in the progeny close to each NB, partially overlapping that of Pros and Elav (Fig. 2K). Weak immunolabeling was also observed in NBs (Fig. 2K,L). Hence, mnb appears to be weakly expressed in NBs and upregulated in newborn GCs, and the Mnb protein persists as these cells differentiate into neurons and move away from the parental NB (Fig. 2M).

Given this pattern, the cell type on which mnb acts to regulate proliferation should be clarified. Interestingly, the excess proliferation in mnb mutants was mainly due to an increase in small proliferating (BrdU⁺ or EdU⁺) cells located beneath the OPC NBs and near to the CB NBs (Fig. 3B-D), where putative GMCs and GCs lie. This did not appear to be due to supernumerary GMCs because, as in wt brains, most frequently, only one midsize Mira⁺ cell (GMC) was associated with each CB NB in mnb brains (Fig. S5E). Double Mira/EdU labeling revealed three types of EdU-labeled cells in wt CB lineages: (1) numerous strong Mira⁺ NBs/GMCs (large and medium cells, respectively); (2) pairs of small newborn GCs (weakly expressing Mira) attached to a Mira⁺ NB; (3) few Mira⁻ cells. Remarkably, there was a much higher proportion of these Mira⁻/EdU⁺ cells located near NBs/GMCs in mnb mutants, concomitant with a decrease in Mira⁺/EdU⁺ NBs/GMCs (Fig. 3E-G). Hence, we concluded that the overproliferation in mnb mutants was not due to supernumerary NBs/GMCs but rather to perturbations in the prospective GCs. Indeed, apoptotic (TUNEL⁺ and pyknotic) cells were located near Mira⁺ CB and OPC NBs/GMCs (Fig. 3H,I; Fig. S5C,D). Together, these results are consistent with the notion that many GCs continue to proliferate rather than exiting the cell cycle in mnb mutants, and that they eventually die by apoptosis.

The data described above strongly suggest that Mnb is required for the cell cycle exit of GCs. Indeed, and as we previously reported for vertebrate Mnb/Dyrk1A (Hämmerle et al., 2011), we found that Mnb regulates the expression of dacapo (dap; Fig. 4A-C), the Drosophila homolog of p27^Kip1. To investigate the molecular mechanisms underlying this regulation, we focused on Pros. This TF is segregated into the GMC during the asymmetric division of embryonic NBs. In GMCs, Pros acts as a key cell fate determinant and it is then immediately downregulated in the newborn GCs (Hirata et al., 1995; Knoblich et al., 1995; Spana and Doe, 1995). By contrast, Pros is expressed weakly (or sub-threshold) in the majority of larval type I NBs and it is robustly upregulated in newborn GCs (Ceron et al., 2001; Colonques et al., 2011), driving their cell cycle exit by promoting dap expression (Colonques et al., 2011). We found that Mnb and Pros proteins are co-expressed by GCs although Mnb expression appears to precede that of Pros (Fig. 2L,M). As reported previously, Pros expression precedes that of Elav (Colonques et al., 2011) but it is downregulated earlier than Mnb and Elav as GCs move away from the NB (Fig. 2K,O).

We next investigated whether Mnb regulated pros expression. Thus, we found substantially weaker Pros immunolabeling in the OPC and CB of mnb LoF mutant brains and, conversely, stronger labeling when Mnb was overexpressed (Fig. 5A-D,G,H; >90% of brains; n>50; Fig. S6). This regulation appears to take place at the transcriptional level, as Mnb overexpression increased Pros FISH labeling (Fig. 5E,F; 8/9 brains). Remarkably, the increase in Pros expression in response to mnb gain of function (GoF) took place in GCs whereas Pros did not apparently change in NBs, even when Mnb protein accumulated strongly following Gal4 driver induction (Fig. 5I-J″). These results suggest that Mnb promotes pros expression in newborn GCs, indicating that Mnb acts upstream of pros in driving the cell cycle exit of GCs.

We previously found that the activation of dap expression by Pros is mediated through the expression of Deadpan (Dpn; Colonques et al., 2011), an essential pan-neural basic helix-loop-helix (bHLH) TF (Bier et al., 1992) that represses dap expression in the larval brain (Wallace et al., 2000). Consistent with Mnb acting upstream of pros, mnb GoF reduced Dpn immunolabeling (Fig. 6A-B″; 8/11 brains). Remarkably, the overexpression of mnb in NBs substantially dampened Dpn expression in the absence of pros upregulation (Fig. 6A′-B″), suggesting that Mnb might regulate dpn in NBs independently of Pros. Conversely, Dpn expression was enhanced in mnb mutants, concomitant with Pros downregulation (Fig. 6C,D; 23/23 brains; quantification in Fig. S7), and small Dpn⁺ cells were frequently found beneath the OPC NBs (Fig. 6E,F; 15/23 brains). As these extranumerary Dpn⁺ cells lacked or very weakly expressed Mira (Fig. 6E,F), and no significant increase in Dpn⁺ NBs was seen in the CB (Fig. 6G), dpn appears to be upregulated in prospective GCs of mnb mutants without transforming them into NBs.

The observation that Mnb GoF suppresses NB proliferation (Fig. 1) without inducing Pros expression (Fig. 5I; Fig. 6B′) strongly suggests that Mnb can promote cell cycle exit by an additional mechanism independent of Pros. Thus, we focused our attention on Asense (Ase), a bHLH TF of the achaete-scute proneural gene complex (Campuzano et al., 1985) that is involved in OL development (Gonzalez et al., 1989). Although Ase is extensively used as a marker of larval NBs, it has intriguingly been reported to promote the expression of dap in the larval brain (Wallace et al., 2000). Interestingly, we found that in addition to its expression in NBs (Dpn⁺ cells) Ase was also expressed strongly by newborn GCs in the OPC, as identified by their position inside the OL, their lack of Dpn and the co-expression of Pros (Fig. 7A,B). In the CB, Ase is expressed in scattered cell clusters that contain one NB (Dpn⁺, Mira⁺) and two or four small neighboring Dpn⁻/Mira⁻ cells (Fig. 7A,C,D). Accordingly, these small Ase⁺ cells appear to be newborn GCs. Nevertheless, and unlike in the OPC, these cells lacked Pros and Elav or expressed them very weakly (Fig. 7C). Notably, these small Ase⁺ cells accumulated more Ase than mitotic NBs and GMCs (Fig. 7D,E). Together, these results strongly suggest that ase expression is downregulated during NB division, upregulated in newborn GCs and quickly downregulated again.

These results prompted us to assess the possible regulation of ase expression by Mnb. Thus, the overexpression of Mnb in NBs (c831-Gal4 driver, 8/15 brains) and GCs (pros-Gal4 driver, 7/14 brains; see driver pattern in Fig. S4C,D) substantially enhanced Ase protein labeling (Fig. 8A-C; quantification in Fig. S8A) without apparently increasing the number of Ase⁺ GCs (Fig. 8D,E). Conversely, we found a strong decrease in Ase protein in mnb LoF mutants (Fig. S8D-G), both in NBs and newborn GCs (Fig. 8F,G; 9/19 brains). This regulation of Ase expression appeared to occur at the transcriptional level given the enhanced FISH staining in mnb GoF mutant brains (Fig. S8B,C; 13/15 brains).

Because Mnb expression persists in Elav⁺ GCs that move away from the parental NB after Ase and Pros expression decays (Fig. 2M; Fig. 7F), and Elav is a factor required for neuronal differentiation (Robinow et al., 1988; Yao et al., 1993; Lai et al., 2012), we wondered whether Mnb promoted neuronal differentiation of GCs in addition to triggering cell cycle exit. Consistent with this idea, we found weaker Elav expression in the GCs of mnb LoF mutants (Fig. 8F,G; 14/15 brains; Fig. S9A-D) and after Mnb RNAi (Fig. S9E,F). Conversely, Elav expression was enhanced by Mnb overexpression in GCs driven by elav-Gal4 (data not shown) or pros-Gal4 (Fig. 8A′-C′; n=13/15; quantification in Fig. S9G), whereas its overexpression driven by c831-Gal4 did not induce Elav expression in NBs (Fig. S9H,I), even after long-term overexpression that produced prolonged cell cycle arrest (Fig. 8H,I). Accordingly, we concluded that Mnb is necessary and sufficient to promote Elav expression in GCs but not in NBs. By contrast, pros (Fig. S10; 7/13 brains) or ase (Fig. S11; 19/21 brains) GoF did not stimulate or even mildly repress Elav expression.

The decision of neuronal precursors to exit the cell cycle and commence their terminal differentiation at the right time is a crucial aspect of brain development. CKIs play a central role in the regulation of cell cycle exit during development (Zhu and Skoultchi, 2001; Buttitta and Edgar, 2007). Drosophila melanogaster lacks Ink-type CKIs and it has a single Cip/Kip-type CKI, Dap (de Nooij et al., 1996; Lane et al., 1996). Hence, like p27^Kip1 in vertebrates (Sherr and Roberts, 1999; Nguyen et al., 2006), Dap plays a central role in controlling the cell cycle exit of CNS neurons. Thus, dap LoF mutants are characterized by overproliferation in the larval brain (Wallace et al., 2000), where the expression of dap is promoted by Ase and repressed by Dpn. Accordingly, ase LoF and dpn GoF mutants experience larval brain overproliferation (Wallace et al., 2000). We previously showed that the expression of dap in the larval brain is also controlled by Pros and, thus, dap is upregulated in newborn GCs in response to the transient expression of Pros, which suppresses dpn (Colonques et al., 2011). This is also consistent with the requirement for pros to terminate dpn expression in embryonic NBs (Vaessin et al., 1991) and to end proliferation of post-embryonic NBs (Maurange et al., 2008).

We show here that mnb inhibits proliferation in the larval brain, where it is expressed weakly in NBs and more strongly in newborn GCs shortly after division of the GMC. Mnb was seen to be necessary and sufficient to promote pros expression in GCs, and, consequently, to upregulate dap expression. We also found that Mnb promotes ase expression in newborn GCs. Thus, Mnb can upregulate dap expression through two convergent mechanisms mediated by Ase and Pros (summarized in Fig. 9). Redundant mechanisms regulate cell cycle exit in a number of organisms (reviewed by Buttitta and Edgar, 2007), yet as Ase and Pros are expressed sequentially in GCs, they may also play different roles. Indeed, Ase and Pros appear to regulate antagonistically several gene clusters in gene expression arrays (Southall and Brand, 2009). Thus, we speculate that after the downregulation of ase in newborn GCs, Pros helps to maintain dap expression, avoiding re-entry of these neuronal precursors into the cell cycle and facilitating their terminal differentiation.

Given the key roles of Dpn in the specification, self-renewal and proliferation of larval NBs (San-Juán and Baonza, 2011; Zhu et al., 2012), the capacity of Mnb to repress dpn expression in NBs independently of pros might explain the arrest of proliferation induced by Mnb overexpression in NBs. Hence, Mnb may participate in the control of larval NB proliferation, an issue that merits further study. However, it appears unlikely that Mnb significantly influences NB/GMC specification. Together, our data fit best with the idea that the main actions of Mnb leading to cell cycle exit and terminal differentiation take place in newborn GCs. Accordingly, many of these GCs fail to withdraw correctly from the cell cycle in mnb mutants, continuing to proliferate rather than differentiating into neurons, and ultimately dying, possibly explaining the neuronal deficits in the adult mnb mutant brain (Fischbach and Heisenberg, 1984; Tejedor et al., 1995).

We previously found that Dyrk1A, the mammalian ortholog of mnb, is also transiently expressed in newborn neuronal precursors in the embryonic mouse brain (Hämmerle et al., 2008). Furthermore, Mnb/Dyrk1A is necessary and sufficient for p27^Kip1 expression in the chick and mouse embryo (Hämmerle et al., 2011), strongly suggesting an evolutionarily conserved function in the cell cycle exit of CNS neurons.

Abundant data from diverse experimental systems have implicated Prox1, the vertebrate ortholog of Pros (Oliver et al., 1993), in cell cycle exit, neuronal differentiation and tumorigenesis (reviewed by Elsir et al., 2012). Remarkably, Prox1 suppresses proliferation by inducing p27^Kip1 expression in neuroblastoma cells (Foskolou et al., 2013). Similarly, Dpn is a homolog of the vertebrate Hes TFs that fulfill essential roles in proliferation and specification of neural stem cells (Kageyama et al., 2008). Ascl1 (also known as Mash1), the closest Ase homolog, is a major regulator of vertebrate neurogenesis and, like Ase at the NB to GC transition, Ascl1 is expressed in an oscillatory manner in the transition from NPs to neuronal precursors (Imayoshi et al., 2013). Also, like Ase, genome-wide characterization of Ascl1 target genes identified both positive and negative regulators of the cell cycle (Castro et al., 2011). Indeed, Ascl1 LoF in mice is associated with neuronal deficits, whereas its overexpression in NPs induces cell cycle exit and neuronal differentiation (reviewed by Bertrand et al., 2002; Castro and Guillemot, 2011). Remarkably, the expression of Ascl1 is sufficient to induce neuronal differentiation of uncommitted mammalian cells in vitro, preceded by the expression of p27^Kip1 (Farah et al., 2000). Moreover, p27Xic1 is required for Ascl1-induced neurogenesis in Xenopus embryos (Ali et al., 2014). Interestingly, Ascl1 and Prox1 expression are regulated at the transition from NP proliferation to terminal differentiation in the rodent forebrain (Torii et al., 1999). Therefore, it will be interesting to determine whether Ascl1 and Prox1 act downstream of Dyrk1A to regulate the cell cycle exit of vertebrate CNS neurons.

In contrast to the anti-proliferative effect of Mnb in the CNS of Drosophila described here, Mnb is thought to promote the growth of Drosophila imaginal discs by regulating the Salvador-Warts-Hippo pathway (Degoutin et al., 2013). These contrasting data may reflect the different functions of Mnb in neural and non-neural Drosophila progenitor cells. Intriguingly, the vertebrate Mnb/Dyrk1A orthologs (Yabut et al., 2010; Litovchick et al., 2011; Hämmerle et al., 2011; Soppa et al., 2014) and other DYRKs (reviewed by Becker, 2012) appear to act as cell cycle repressors in both neural and non-neural mammalian cells.

Cell cycle exit and terminal cell differentiation must be tightly coupled during normal development in order to avoid premature differentiation of proliferating progenitors, and to ensure that terminally differentiating cells become refractory to proliferative signals. The capacity of Mnb to upregulate elav suggests that Mnb may promote the differentiation of newborn GCs in addition to driving cell cycle exit. Elav is a neuronal-specific RNA-binding protein required for the differentiation and maintenance of neurons in Drosophila (Robinow et al., 1988; Yao et al., 1993; Lai et al., 2012) where it regulates alternative splicing (Koushika et al., 1996; Lisbin et al., 2001). Interestingly, Mnb/Dyrk1A has also been implicated in the regulation of splicing (Álvarez et al., 2003; de Graaf et al., 2004; Shi et al., 2008) and, thus, it is tempting to speculate that Mnb can promote neuronal differentiation through mechanisms affecting RNA regulation.

Given that Mnb is expressed in differentiating GCs after ase and pros are downregulated, and that, unlike Pros or Ase GoF alone, it promotes the expression of Elav, Mnb could regulate Elav expression independently of Pros and Ase. This might also explain why mnb mutants exhibit both GC overproliferation and cell death, whereas only overproliferation is evident in ase and pros mutants (Wallace et al., 2000; Colonques et al., 2011). Nevertheless, we cannot rule out the possibility that Pros and Ase, together or along with additional factors, may act downstream of Mnb to promote Elav expression.

Together, our data suggest a possible role for Mnb in coupling the termination of GC proliferation by regulating dap via Ase and Pros, to the beginning of neuronal differentiation through Elav (and possibly other unknown factors). We previously found that Mnb/Dyrk1A is also transiently expressed in vertebrate neuronal precursors (Hämmerle et al., 2002; Hämmerle et al., 2008), where it promotes both cell cycle exit and neuronal differentiation (Hämmerle et al., 2011). Thus, our data suggest an evolutionarily conserved function that couples cell cycle exit to neuronal differentiation during CNS neurogenesis.

The presence of the human mnb ortholog DYRK1A on chromosome 21 (Guimerâ et al., 1996), its overexpression in the developing DS brain (Guimera et al., 1999) and its neurodevelopmental functions have implicated this kinase in DS neuropathology (reviewed by Tejedor and Hämmerle, 2011). Among the neurological alterations caused by DS, we must highlight the neuronal deficit in certain brain regions (Becker et al., 1991). As neural proliferation and neurogenesis are altered in the forebrain of fetuses with DS and in embryos of DS mouse models, this deficit most likely originates during early brain development (reviewed by Contestabile et al., 2010). Hence, we hypothesized that the overexpression of DYRK1A in fetuses with DS could cause premature cell cycle exit and differentiation, and, consequently, the depletion of NP pools, thereby contributing to the neuronal deficits in DS (Tejedor and Hämmerle, 2011). Remarkably, a specific inhibitor of the Mnb/Dyrk1A kinase rescues the premature neuronal differentiation phenotype displayed by NPs isolated from DS mouse brains (Mazur-Kolecka et al., 2012).

In conjunction, our results support the idea that Mnb acts on multiple factors that link the mechanisms regulating neurogenesis, cell cycle control and terminal differentiation, ensuring the precise generation of neurons. Genetic modifications of Mnb/Dyrk1A, such as those that occur in microcephaly and DS, will disrupt this precise coordination, altering neurogenesis during brain development.

Fly stocks (Drosophila melanogaster) were raised at 25°C on standard medium, except for flies carrying the Gal4/UAS transgenes, which were kept at 17°C until the time of induction when the temperature was raised to 29°C. We generated several transgenic w; P[w⁺UAS-mnb] lines by inserting the coding regions of the mnb C transcript (Tejedor et al., 1995) into the pUAST plasmid (Brand and Perrimon, 1993) following standard procedures for embryo transformation. The wt strains used were Berlin and Oregon-R. The following mnb alleles were used: mnb¹, mnb², mnb³, mnb^e02428, mnb^EY14320 and UAS-mnb RNAi (see details in Table S1). Other mutant stocks were: pros-Gal4.C20, UAS-pros (F. Matsuzaki, Center for Developmental Biology, Kobe, Japan); Elav-Gal4, UASmCD8::GFP, hs-Gal4/TM3 (Bloomington Stock Center); c820-Gal4, c831-Gal4 (Manseau et al., 1997); UAS-ase (Y. N. Yan, Howard Hughes Medical Institute, San Francisco, USA).

The overexpression of Mnb was carried out by crossing UAS-mnb flies with several Gal4 drivers. These flies were grown at 16-18°C from embryo until mid-late third instar larvae. Tests were carried out to assess leaky activation of the driver. Only drivers that remain silent during the low temperature period were used. Then, the temperature was shifted to 29°C for 8-12 h for the activation of Gal4. In the case of crosses with the hs-Gal4 driver, this last step was preceded by a heat shock by immersing the larvae tubes in a water bath at 37°C for 20 min.

hsp70-Flp; Actin5C <yellow, [stop]> Gal4, UAS-GFP (for control clones) or hsp70-Flp; Actin5C <yellow, [stop]> Gal4, UAS-GFP / UAS-mnb RNAi (stock P{KK102642} from the Vienna Drosophila RNAi Center) were grown at 17°C until 24 h before pupae formation. Then, a heat shock of 20 min at 37°C was applied and incubation continued at 29°C until wandering larvae stage. Brains were dissected and IHC performed as previously described (Colonques et al., 2007; Colonques et al., 2011).

For in vivo pulse/chase BrdU labeling, larvae were fed for 2.5 h with BrdU (pulse) at the desired developmental time and then allowed to develop until (chase) white pupae or 1 day after adult eclosion. In vitro BrdU labeling of dissected larval brains was essentially performed as described previously (Tejedor et al., 1995; Ceron et al., 2001) but with BrdU pulses of 5-10 min. For cellular analysis, BrdU-labeled larval brains were embedded in plastic and processed, as described elsewhere (Rogero et al., 1997). In vitro EdU labeling was also carried out by giving pulses of 6-30 min to dissected larval brains followed by IHC analysis. See supplementary Materials and Methods for detailed descriptions.

We raised an antiserum (MNB-NTH) against the N-terminal region of all Mnb protein isoforms following a procedure similar to that described elsewhere (Tejedor et al., 1995). Its specificity was validated by IHC of mnb mutants (Fig. 2H,I). See supplementary Materials and Methods for details.

Immunohistochemical analysis of late larval brains was performed as described previously (Ceron et al., 2001; Colonques et al., 2007, 2011). See supplementary Materials and Methods for details of antibodies and quantitative image analysis.

Digoxigenin (DIG) DNA-labeled probes were synthesized by PCR. For mnb, we used either a 237 bp fragment that corresponds to positions 1617-1853 or a 321 bp fragment corresponding to positions 1000-1320 of the mnb cDNA sequence (GenBank BT015255). FISH of Dap was performed as described elsewhere (Colonques et al., 2011). A DIG-RNA probe for Ase was prepared from a cDNA obtained by subcloning a 0.9 kb cDNA fragment of clone RE15472 (nucleotides 1541-2446) inserted in the pFLC1 vector. Antisense and sense probes were synthesized with T3 and T7 RNA polymerases, respectively. All hybridized probes were detected with a horseradish peroxidase-coupled anti-DIG antibody (Roche), and visualized by tyramide signal amplification (TSA) with Cy2 or Cy3 (Perkin Elmer).

For TUNEL analysis, larval brains were dissected in cold PBS and immediately fixed with 4% paraformaldehyde/1% Triton X-100 in PBS for 1 h at room temperature. The tissue was then washed sequentially with 0.3% Triton X-100/0.02% NaN₃ in PBS 3×10 min, once quickly in PBS, then in citrate buffer (pH 6)/0.1% Triton X-100 for 1 min at 65°C and finally with PBS again. The nick-end labeling reaction was performed as recommended by the supplier (Roche). Cell death was also monitored by Acridine Orange staining, electron microscopy and by activated caspase 3 IHC. See supplementary Materials and Methods for details.

Experiments were repeated at least three times. Sample size (n=number of brains) was adjusted according to the penetrance and variability of phenotype. A minimum of six brains was used in each experiment. Only damaged or disoriented samples were excluded from the analysis.

For statistics, mean and standard deviation (s.d.) values were calculated by standard methods and the statistical significance of differences was determined using the Student's t-test, one-way ANOVA or the Mann–Whitney Rank Sum test, depending on the experimental characteristics.

We are grateful to J.J. Cabanes and D. Tornero for assistance in generating the transgenic flies, to M.J. Chulia and E. Llorens for expert technical assistance, and to the other lab members for interesting discussions. We are indebted to A. Baonza, H. Bellen, S. Campuzano, C. Doe, C. Gonzalez, I.K. Hariharan, Y.-N. Jan, A. Jarman, F. Matsuzaki, H. Reichert, H. Richardson, S. Selleck, J. Skeath, T. Uemura, H. Vaessin, the VDRC, the Bloomington Stock Center, and the Developmental Studies Hybridoma Bank for providing us with flies, antisera and other molecular tools.