Drosophila optic lobe neuroblasts triggered by a wave of proneural gene expression that is negatively regulated by JAK/STAT

In order to populate the central nervous system (CNS) with the appropriate number of neurons and glia, both the proliferation of neural precursor cells and timing of neural differentiation must be controlled. In the Drosophila CNS, neuroblasts (NBs) have been extensively studied as a model for neural stem cell development(Campos-Ortega, 1993; Goodman and Doe, 1993). Embryonic NBs delaminate as single cells from an epithelium called the ventral neuroectoderm. Neuroectodermal cells divide symmetrically in the plane of the neuroectoderm to generate identical daughter cells, but, upon differentiation into NBs, their axis of division rotates to a vertical plane (perpendicular to the neuroectoderm). NBs divide asymmetrically to generate a self-renewing NB and a ganglion mother cell (GMC), which divides again and typically generates two post-mitotic neurons (Fuerstenberg et al., 1998; Yu et al.,2006). Similar mechanisms have been described for vertebrate CNS development; progenitor cells proliferate through symmetric division in which one cell gives rise to identical daughter cells, followed by the neurogenesis in which a subset of cells becomes restricted to a neuronal or glial lineage(Anderson, 2001; Gotz and Huttner, 2005). These cells undergo asymmetric cell division in which one cell is maintained as a multipotent progenitor cell, while the other is fated to differentiate into a neuron or glia within a few rounds of cell division.

Mechanisms that underlie the neuroectoderm to progenitor NB transition in the CNS of flies and vertebrates have been difficult to identify, in part because the transitions are not well ordered in space and time. By contrast,we find and describe here, that the development of Drosophila medulla neurons is a process that can be precisely described because the transition from neuroepithelial (NE) cells to NBs progresses in a synchronized and ordered manner.

The Drosophila visual system is composed of the retina and the optic lobe. The latter contains three optic ganglia: lamina, medulla and lobula. During embryonic development, the optic lobe invaginates from a region of head epidermis called optic lobe placode(Green et al., 1993). The optic lobe loses contact with the outer surface of the embryo and forms an epithelial vesicle attached to the brain(Green et al., 1993), and soon after larval hatching, its cells start to proliferate and separate into an outer optic anlagen (OOA) and an inner optic anlagen (IOA)(Hofbauer and Campos-Ortega,1990). Towards the end of the first instar, the OOA adopts a crescent shape, with the opening of the crescent pointing posteriorly(Nassif et al., 2003). The OOA generates the outer medulla and the lamina neurons, while the IOA generates the inner medulla, the lobula and the lobula plate neurons. The epithelial part of the OOA is composed of a single layer of NE cells. During first and second instar stages, NE cells of the OOA proliferate by symmetric cell division. NE cells differentiate into medulla NBs and lamina precursor cells at the medial and the lateral edge, respectively(Fig. 1A-C). Medulla NBs divide asymmetrically along apico-basal axis and produce GMCs, which divide again and become medulla neurons (Fig. 1C) (Egger et al.,2007; Nassif et al.,2003; Toriya et al.,2006).

The mechanisms underlying neurogenesis have been most intensely studied in the development of external sense organs and embryonic CNS of Drosophila. In these systems, NBs are induced from among NE cells in a `proneural cluster' that express `proneural genes' such as atonal,achaete (ac), scute (sc) and lethal of scute [l(1)sc] (Cabrera et al., 1987; Jarman et al.,1993; Jarman et al.,1994; Martin-Bermudo et al.,1991; Skeath and Carroll,1992). These proneural genes encode basic helix-loop-helix (bHLH)transcription factors that dimerize with another bHLH protein Daughterless(Da) (Jarman et al., 1993; Murre et al., 1989a; Murre et al., 1989b; Villares and Cabrera, 1987). Single or several NBs are selected from each cluster by the mechanism known as lateral inhibition (Artavanis-Tsakonas and Simpson, 1991; Hassan and Vaessin, 1996). In contrast to the external sense organs and embryonic CNS, the differentiation from NE cells to medulla NBs is well ordered (Egger et al., 2007). We found a `proneural wave' of differentiation starts from the medial edge of the NE sheet and sweeps the optic lobe from medial to lateral during third instar (L3) stage; l(1)sc is expressed transiently at the wave front and plays an important role in differentiation of NBs.

We also found that the Janus kinase/signal transuducer and activator of transcription (JAK/STAT) pathway negatively regulates proneural wave progression. JAK/STAT pathway is a well-conserved signaling pathway that was first identified in mammals. The Drosophila JAK/STAT signaling pathway is composed of several major factors: the three ligands Unpaired (Upd)(Harrison et al., 1998), Upd2(Gilbert et al., 2005; Hombria et al., 2005) and Upd3(Agaisse et al., 2003); the transmembrane receptor Domeless (Dome)(Brown et al., 2001; Chen et al., 2002); the JAK homolog Hopscotch (Hop) (Binari and Perrimon, 1994); and the STAT homolog Stat92E(Hou et al., 1996; Yan et al., 1996). JAK/STAT signaling is involved in many developmental processes, including segmentation in embryogenesis, eye morphogenesis, hematopoiesis and stem cell maintenance(Arbouzova and Zeidler, 2006; Luo and Dearolf, 2001; Zeidler et al., 2000). Here,we report a novel mechanism underlying NB (neural progenitor) formation and discuss its role in establishing a precise topographic map in the visual system.

Flies were grown at 25°C otherwise noted. y w and Canton-S flies were used as wild-type controls. The following mutant and transgenic strains were used in this study. tkv-lacZ is a reporter construct that has an insertion of a P element carrying lacZin the promoter region of the thickveins gene(Tanimoto et al., 2000). ap-lacZ is a P element enhancer trap insertion just 5′ of the apterous gene and expressed in ∼50 % of medulla neurons(Cohen et al., 1992). upd-Gal4 is an enhancer trap line of upd(Halder et al., 1995; Tsai and Sun, 2004) and its expression was visualized by crossing with UAS-GFP.nls flies. 10xSTAT-GFP is an in vivo detector that reflects the activation of JAK/STAT signal (Bach et al.,2007). UAS-l(1)sc was described previously(Carmena et al., 1995). UAS-l(1)sc was overexpressed by crossing with NP6099 driver(Hayashi et al., 2002; Yoshida et al., 2005). Overexpression clones of UAS-upd(Zeidler et al., 1999) and UAS-hop^Tum-l (Harrison et al., 1995) were induced by hs-flp; AyGal4 flies(Ito et al., 1997). eya¹ is a mutant allele of eyes absent described by Bonini et al. (Bonini et al.,1993). Deficiency chromosome of Df(1)260-1, Df(1)sc10-1,Df(1)sc19 or Df(1)ase1 was recombined onto FRT19A. These clones were induced by NP7340(Hayashi et al., 2002). UAS-flp. da¹⁰ is an amorphic allele of the damutant (Caudy et al., 1988). Clones of da¹⁰ were induced by NP6099 UAS-flp. hop² is a null allele of hop(Perrimon and Mahowald, 1986). Stat92E^85C9 is a strong hypomorphic allele of Stat92E (Silver and Montell,2001) and Stat92E⁶³⁴⁶ is a putative null allele of Stat92E (Hou et al.,1996). Stat92E^F is a temperature-sensitive allele of Stat92E (Baksa et al.,2002). Stat92E^85C9/Sta92E^F and Stat92E ⁶³⁴⁶/Stat92E^F flies were raised at 29°C. To assess the function of Stat92E, we generated clones in a Minute back ground with hsflp; FRT82 ubi-GFP M(3)^w124 (Ferrus,1975).

Immunohistochemistry was performed as described(Huang and Kunes, 1996; Takei et al., 2004). The following antibodies were provided by the Developmental Studies Hybridoma Bank(DSHB): mouse anti-Dac (mAbdac2-3, 1:1000), mouse anti-Arm (N2 7A1, 1:40) and rat anti-Elav (7E8A10, 1:50). Rabbit anti-Tll (1:600) was provided by East Asian Distribution Center for Segmentation Antibodies. We also used rat anti-L(1)sc (1:800, A. Carmena), guinea pig anti-Dpn (1/1000, J. Skeath),rabbit anti-Ase (1:2000, Y. N. Jan), rabbit anti-Dlg (1:1000, T. Uemura), goat anti-Horseradish Peroxidase (HRP, 1:100, Accurate Chemical and Scientific),rabbit anti-cleaved Caspase 3 (1:100, Cell Signaling Technology), mouse anti-β-gal (1:250, Promega) and rabbit anti-β-gal (1:2000, Cappel). Secondary antibodies (Jackson) were used at the following dilutions:anti-mouse Cy3, 1:200; anti-mouse Cy5, 1:200; anti-mouse FITC, 1:200;anti-guinea pig Cy3, 1:200; anti-guinea pig Cy5, 1:200; anti-rat Cy3, 1:200;anti-rat Cy5, 1:200; anti-rabbit FITC, 1:200; anti-rabbit Cy5, 1:200;anti-rabbit Alexa 546, 1:200 (Molecular Probes). Specimens were mounted with vectashield mounting media (Vector) and viewed on a Zeiss LSM510 confocal microscope.

In situ hybridization was performed as described previously(Nagaso et al., 2001). DNA template for the upd probe has been described previously(Tsai and Sun, 2004).

During larval development, photoreceptor cells (R cells) in the eye disc project axons to the optic lobe in the brain(Fig. 1A). Axons of R1-R6 terminate in the lamina, and axons of R7 and R8 terminate in the medulla(Fig. 1C). Lamina neurons are in the lateral most aspect of the optic lobe; medulla NBs that produce large numbers of medulla neurons are on the medial side(Fig. 1B,C,E,F). Both medulla and lamina neurons derive from NE cells that constitute the surface of the OOA(Fig. 1B-F)(Nassif et al., 2003). Medulla neurons derive from NBs on the medial side of the NE sheet; lamina neurons derive directly from the lateral side of the OOA without formation of NBs(Fig. 1B-F)(Nassif et al., 2003). Medulla NBs are generated during L3 stages both by symmetric divisions of the OOA NE cells and progressively from the medial edge(Fig. 1G-J)(Egger et al., 2007; Nassif et al., 2003). NB generation is synchronized, forming a one-cell wide band of newly differentiated NBs to the inner slope of the U-shaped developing OOA. The NBs divide asymmetrically with their division plane oriented perpendicular to the surface and along the apico-basal axis; they produce medulla neurons basally and expand the volume of the optic lobe(Fig. 1C,F)(Ceron et al., 2001; Egger et al., 2007; Toriya et al., 2006). As the swath of NBs widens during maturation of the optic lobe, the expanse of NE cells decreases (Fig. 1G-J). On the lateral most side, NE cells receive Hedgehog (Hh) signals from innervating R axons and differentiate into lamina neurons(Fig. 1C,F)(Huang and Kunes, 1996; Huang and Kunes, 1998).

We determined that the proneural protein L(1)sc(Hinz et al., 1994; Jimenez and Campos-Ortega,1990; Martin-Bermudo et al.,1991) is transiently expressed in a narrow band of 1-2 cells at the medial edge of the NE sheet (Fig. 2A-G). This band precedes NB formation and moves laterally throughout the L3 period. The expression pattern of L(1)sc led us to test the idea that L(1)sc makes NE cells competent to differentiate into medulla NBs. We overexpressed l(1)sc in most of the NE cells during the stages when medulla neurogenesis begins, and observed that, at early stages, the number of Deadpan (Dpn)-expressing cells on the surface of the optic lobe increased; at later stages, only medulla neurons were observed and lamina was not formed (Fig. 2K-M; 100%; n=20, compare with Fig. 2H-J, see also Fig. S1 in the supplementary material). These results suggest that L(1)sc activity is sufficient to induce NBs, and the medulla and lamina neuron precursors derive from the limited number of cells in a common primordium. As L(1)sc-expressing cells lead a front that sweeps across the optic lobe and induces differentiation of NBs, we refer to the band of L(1)sc-expressing cells as the proneuroal wave(Fig. 2G). The proneural wave converts all NE cells to NBs in an ordered manner from the medial side. This program contrasts with the neurogenesis of the Drosophila sense organs or embryonic CNS in which only one or several NBs are selected from the NE cells.

We asked whether L(1)sc is required for NB formation. As no mutant lacking only l(1)sc is available, we first examined da mutants. da encodes a bHLH transcription factor that is an indispensable co-factor for all known bHLH proneural proteins(Murre et al., 1989b). In clones of cells mutant for da¹⁰, an amorphic allele of da, expression of Dpn was strongly reduced(Fig. 3D; n=18). As Dpn often remained in the medial cells of the clones, we infer that Dpn expression is delayed in mutant cells. However, L(1)sc expression was not affected in mutant cells (Fig. 3E,K; n=13). We conclude that proneural function is required to time NB differentiation but not for progression of the proneural wave. This result also suggests that the progression of the proneural wave is independent of NB differentiation.

In order to assess the requirement for other proneural genes, we used deficiency chromosomes that delete various regions of the achaete-scute complex (AS-C; Fig. 3F). The AS-Ccontains four bHLH proneural genes: ac, sc, l(1)sc and asense (ase). In wild type, Ac expression was not detected;both NE cells and NBs express Sc (Egger et al., 2007) and Ase expression was detected in NBs(Fig. 3A-C). Mutant clones deficient for all the proneural genes [Df(1)260-1(Hinz et al., 1994)] caused the delay in Dpn expression by 4-6 rows of cells, which is indistinguishable from the da mutant clones, suggesting that only these four proneural genes are crucial for the proneural function in the optic lobe(Fig. 3G,K; n=29). As clones deficient for both ac and sc [Df(1)sc10-1(Campuzano et al., 1985)] did not affect Dpn expression (Fig. 3H; n=10), we conclude that ac and scare not essential for NB differentiation. However, Dpn expression was delayed by 1-2 rows of cell in clones deficient for ac, sc and l(1)sc [Df(1)sc19(Carmena et al., 1995)],suggesting that l(1)sc is required for the timely onset of NB formation (Fig. 3I,K; n=36). Ase expression was not affected in these clones (data not shown). Mutant clones deficient for ase(Brand et al., 1993) initiated Dpn expression normally (Fig. 3J; n=19), suggesting that ase is required for timing of the normal onset of NB formation when l(1)sc is impaired. Sc might also have a redundant function but appropriate deficiency chromosomes are not available at present to unambiguously study the role of Sc. We also examined whether the onset of NB formation is regulated by inputs from the retinal axons because lamina development depends on arriving retinal axons (R axons) (Selleck and Steller,1991). The `eyeless' mutant, eya¹(Bonini et al., 1993), did not affect L(1)sc expression or medulla NB differentiation, suggesting that medulla neurogenesis is independent of R axon projection at least at this early phase (see Fig. S2A,B in the supplementary material). However, the medulla neuropil of this mutant was poorly organized in the later pupal stage partly because of excess cell death (see Fig. S2C,D in the supplementary material).

We next searched for genes that regulate proneural wave progression and determined that upd is expressed in NE cells(Fig. 4). Analysis of upd-Gal4 (Halder et al.,1995; Tsai and Sun,2004) suggests that the pattern of upd expression is dynamic. In the early L3 stage, upd-Gal4 is expressed in some NE cells that express high levels of Armadillo (Arm)(Fig. 4A)(Hayden et al., 2007). upd-Gal4 expression was restricted to the lateral side of the NE in mid L3 (Fig. 4B) and was in the lamina neuron precursors in late L3 (Fig. 4C, see Fig. S3A in the supplementary material). In the early to mid L3, expression pattern of upd mRNA was similar to that of upd-Gal4 (see Fig. S3B,C in the supplementary material). But in late L3, upd mRNA was specifically expressed in the lamina furrow that is located at the most lateral NE cells (see Fig. S3D,E in the supplementary material). The different patterns observed in the enhancer trap expression and mRNA distribution might be caused by perdurance of the Gal4 and/or GFP reporter protein (Tsai and Sun,2004).

To determine where the JAK/STAT signal is activated in the optic lobe, the expression of a 10xSTAT-GFP reporter construct(Bach et al., 2007) was examined. GFP fluorescence was observed from early L3(Fig. 4D), and was detected in the lateral side of the NE cells in mid L3(Fig. 4E). In late L3, the GFP fluorescence was weak in the NE cells and stronger in the lamina(Fig. 4F). These results suggest that JAK/STAT signaling is activated in the NE cells at least in early to mid L3, and that it is low medially and higher in the lateral NE cells(Fig. 4B,E).

To determine whether JAK/STAT signaling has a role in NB development, we examined loss of function phenotypes for components of the JAK/STAT pathway,including Hop and Stat92E. In mid L3 optic lobes of hop²,a null allele of hop (Perrimon and Mahowald, 1986), both NE cells and NBs were found to express high levels of Arm and Dpn (Fig. 5A,B; 100%; n=9). In the wild type, they are co-expressed only in cells at the transition from NE cells to NBs. Late L3 mutant optic lobes were smaller, had few NBs, the number of neurons was fewer and lamina neurons were absent (Fig. 5C,D;100%; n=20). We infer that NBs developed prematurely in the hop² mutant and that in the absence of the JAK/STAT signal, there was insufficient proliferation of the NE cells prior to transition to the NB fate. Similar phenotypes were observed in Stat92E^85C9/Stat92E^F or Stat92E⁶³⁴⁶/Stat92E^F loss-of-function mutants(see Fig. S4 in the supplementary material and data not shown). To examine whether Elav-expressing neurons observed in hop² mutant were differentiated medulla neurons, optic lobes are stained by ap-lacZ (Fig. 5E,F). Neurons in hop² mutant expressed ap-lacZ, which indicates that though the number of NBs in hop² was fewer than wild-type flies, NBs in the mutant optic lobe produce differentiated medulla neurons (Fig. 5F; 100%; n=15).

To further investigate the role of JAK/STAT signaling, Stat92Emutant clones were generated. Near clones of Stat92E^85C9,which is a strong hypomorphic allele of Stat92E(Silver and Montell, 2001),both L(1)sc and Dpn were expressed in more lateral cells than in surrounding Stat92E^85C9/+ cells(Fig. 6A,H; n=29),expression of Ase was present at more lateral locations, and ectopic Elav expression was observed within and near Stat92E⁶³⁴⁶(putative null allele of Stat92E)(Hou et al., 1996) clones(Fig. 6B; n=21). These results suggest that loss of Stat92E function leads to faster progression of proneural wave and earlier initiation of medulla differentiation. Interestingly, these phenotypes were not cell-autonomous. Ectopic expression of L(1)sc was observed not only in the mutant cells but, in some cases, in adjacent Stat92E^85C9/+ cells as well(Fig. 6A′, arrowheads). When Stat92E^85C9 clones were extended into the putative lamina region, Dac-expressing lamina precursors disappeared and Dpn-expressing medulla NBs occupied the region (Fig. 6C; n=27). This suggests that the number of lamina neurons and medulla neurons is balanced by the JAK/STAT signal.

We next generated clones of cells expressing Hop^Tum-l, an active form of Hop (Harrison et al.,1995). Clones in the NE cells were associated with more medial expression of L(1)sc and delayed expression of Dpn(Fig. 6D,H; n=17). Cells that did not express Dpn expressed high Arm, indicating that they remained in an undifferentiated NE state(Fig. 6E; n=20). When upd was overexpressed (as above), a similar phenotype was observed(Fig. 6F,G; n=20 for F; n=10 for G). These results are consistent with the expectation that Upd acts as an extracellular activator of the JAK/STAT pathway. Note that the phenotypes were not cell autonomous and the proneural wave was continuous,even when it extended more medially (with elevated JAK/STAT signaling) or laterally (with decreased signaling). These results suggest that JAK/STAT signaling negatively regulates the progression of the proneural wave and that it does not directly regulate expression of the proneural genes(Fig. 6I). The alteration in proneural wave progression described here cannot be attributed solely to growth defects because any differences in proneural wave progression were associated with slowly growing Minute clones (see Fig. S5 in the supplementary material).

We report a novel mechanism of neural stem cell (NB) formation in Drosophila medulla development. We show that medulla development provides a novel experimental system in which stem cells are generated in a step-by-step manner. The entire development from neuroepithelium, NBs and neurons proceeds sequentially as spatially ordered arrays of progressively aged cells, and we identified a new role of JAK/STAT signaling in NB formation. We found a `proneural wave' that sweeps unidirectionally in the NE sheet to trigger NB differentiation. This differs from the well-known but not yet fully understood mechanism called `lateral inhibition' that singles out NBs from NE cells.

NE cells are programmed to differentiate into NBs from the medial edge of the developing optic lobe. The wave of differentiation progresses synchronously in a row of cells from medial to lateral optic lobe sweeping across the entire NE sheet; it is preceded by the transient expression of the proneural gene l(1)sc. As the NBs at the medial edge are oldest and the more lateral ones are youngest, developmental process of medulla neurons can be viewed as an array of progressively aged cells across optic lobe mediolaterally. This contrasts with NB formation in the embryonic CNS in which a small number of cells are selected from NE cells to become NBs, leaving the majority of NE cells to develop into non-neural cells. The optic lobe proneural wave is reminiscent of the morphogenetic furrow that moves across the developing eye imaginal disc. The morphogenetic furrow is the site where differentiation from neuroepithelium to photoreceptor neurons is initiated(Ready et al., 1976). The progression is driven by the secreted Hh expressed in the differentiated photoreceptor cells (Heberlein and Moses,1995; Heberlein et al.,1993; Ma et al.,1993). By contrast, the proneural wave still progresses even when NB differentiation is impaired, suggesting that its progression is not driven by a factor emanating from differentiated NBs. We failed to observe progression-defective phenotypes when Hh or Decapentaplegic (Dpp) signaling was reduced (T.T., unpublished). We favor the model that the proneural wave progression is driven by an intrinsic mechanism such as a segmentation clock and is negatively regulated by JAK/STAT pathway(Fig. 7). As the JAK/STAT ligand Upd is expressed only by the most lateral NE cells, proliferation of the NE cells moves the source of ligand laterally and as a consequence releases more medial NE cells from negative regulation and allows the proneural wave to progress laterally. Alternatively, distribution of the Upd ligand and/or the response to Upd changes as the NE cells age as graded 10xSTAT-GFP activities are more prominent in the early stage. Non-autonomous action of JAK/STAT signal indicates that it does not directly regulate L(1)sc expression and there are second signal(s) that regulate the expression of L(1)sc under the control of JAK/STAT signal.

Three out of the four AS-C genes [sc, l(1)sc and ase] are expressed during medulla neurogenesis. l(1)sc is expressed in NE cells and ase in NBs, while sc is expressed both in NE cells and NBs (Egger et al.,2007). Deleting all AS-C genes causes as significant delay as da in NB formation but does not completely eliminate NB formation, suggesting that Da-dependent proneural gene activities are required for timely onset of NB formation. Mutation for sc or asealone does not affect NB formation, but the simultaneous deletion of sc and l(1)sc causes the delay in NB formation and the additional deletion of ase further delays NB formation. aseexpression is not altered in the absence of l(1)sc and l(1)sc is not altered in the absence of ase, indicating that l(1)sc and ase both contribute to the differentiation from NE cells to NBs. Although the contribution of Sc cannot be formally excluded,the highly specific expression pattern led us to infer that L(1)sc plays a major role in the proneural wave.

JAK/STAT signaling is known to regulate stem cell maintenance in the adult germline of Drosophila (Arbouzova and Zeidler, 2006; Fuller and Spradling, 2007). In the male testis, germline stem cells (GSCs)attach to a cluster of somatic support cells at the tip (hub) of the testis. When a GSC divides, the daughter retaining contact with the hub maintains self-renewing GSC identity, while the other daughter differentiates into gonialblast. Upd is specifically expressed in the hub cells and activates JAK/STAT signal in the GSCs to maintain stem cell state(Kiger et al., 2001; Tulina and Matunis, 2001). In the female ovary, JAK/STAT signaling is required in the somatic escort stem cells whose daughters encase developing cysts(Decotto and Spradling, 2005). Here, we show that in the optic lobe development, JAK/STAT signaling maintains NE cells in an undifferentiated state. We suggest that a common mechanism operates in both these developmental systems. Loss of Hop or Stat92E function decreases number of stem cells and ectopic expression of Upd results in over proliferation of undifferentiated cells. The cell fate may be determined by the distance of the cells from the source of ligand; the cells farther from the source commence to differentiate.

In the vertebrate CNS, NE cells first proliferate by symmetric cell divisions and differentiate into neurons and glia in later developmental stages (Anderson, 2001; Gotz and Huttner, 2005; McKay, 1997). JAK/STAT signaling has been implicated in maintenance of neural precursor cells(Yoshimatsu et al., 2006), but there is no clear evidence that those cells are in the same developmental stage as we describe here for Drosophila. Further study of JAK/STAT signaling will reveal whether a common mechanism underlies stem cell development in both Drosophila and vertebrates, and should give new insights into vertebrate CNS neurogenesis.

Development of a precise topographic map (retinotopic map) in Drosophila is known to involve regulation of lamina neuron development with respect to the incoming R axons(Selleck and Steller, 1991). The lateral NE sheet is continuous with a groove called the lamina furrow where NE cells are arrested at G1/S phase(Fig. 1C)(Selleck et al., 1992). The arriving R axons deliver Hh and liberate the arrested NE cells to proliferate and develop into lamina neuron precursors(Fig. 1C)(Huang and Kunes, 1996; Huang and Kunes, 1998). And,thus, R axons can induce the development of their synaptic partners in their vicinity to balance the number of R axonal termini and lamina neurons. However, medulla development does not depend on inputs from the R axons in the early phase. As we have shown here, both lamina and medulla neurons are derived from the continuous NE sheet. Large clones of cells mutant for the JAK/STAT signaling cause immature proliferation of medulla NBs at the expense of lamina neurons, suggesting that the number of NE cells serves as the limiting factor to generate precursors for lamina and medulla neurons. Thus,the number of medulla neurons is roughly regulated at the level of NBs whose generation might be balanced indirectly with the number of lamina neurons through regulating proneural wave progression by JAK/STAT signaling. JAK/STAT signaling therefore plays an important role in the formation of a precise retinotopic map in the visual center.

We thank T. B. Kornberg and W. Chia for critical comments on the manuscript, and members of Tabata laboratory for helpful discussions. We also thank S. Kato for the use of the confocal microscope and Y. Maeyama for her excellent technical help. We are grateful to A. Carmena for kindly providing us with the antibody against L(1) sc and fly strains. We are also grateful N. Perriomon, D. Strutt, G.-H. Baeg, H. Sun, K. Ito, J. Modolell, the Bloomington Stock Center, the National Institute of Genetics, the Drosophila Genetic Resource Center Kyoto, and Szeged Drosophila Stock Center for fly strains, and J. Skeath, Y. N. Jan, T. Uemura, East Asian Distribution Center for Segmentation Antibodies, and the Developmental Studies Hybridoma Bank for antibodies. This work was supported by Grants-in-Aid from Ministry of Education, Science and Culture of Japan, and by Toray Science Foundation to T.T. Support for D.U. was provided by a COE program from Ministry of Education, Science and Culture of Japan. Support for T.Y. was provided by a Predoctoral Fellowship from the Japan Society for the Promotion of Science for Japanese Junior Scientists.