The influence of retinal innervation on neurogenesis in the first optic ganglion of drosophila

Neuron, Vol. 6, 83-99, January, 1991, Copyright The Influence Neuronenesis of Droiophila 0 1991 by Cell Press of Retinal Innervation on in the First Optic Ganglion Scott B. Selleck and Hermann Steller Howard Hughes Medical Institute Department of Brain and Cognitive and Department of Bioiogy Massachusetts Institute of Technology Cambridge, Massachusetts 02139 Sciences Summary We have examined the influence of retinal innervation on the development of target neurons in the first optic ganglion, the lamina, of D. melanogaster. Mitotically active lamina precursor cells (LPCs), which normally produce lamina neurons, are absent in mutants that lack retinal innervation, while other proliferative centers appear unaffected. Reducing the number of innervating photoreceptor axons results in fewer mitotic LPCs. In g/ass mutants photoreceptors project to abnormal locations and LPCs are found adjacent to these aberrant projections. We conclude that the arrival of photoreceptor axons in the larval brain initiates, directly or indirectly, cell division to produce lamina neurons. Our results provide an explanation for how the synchronous development of these two interacting systems is coordinated. Introduction The development of a nervous system requires the generation of many distinct cell types and their precise interconnection with each other. One important aspect of generating functional neuronal circuits is to match the relative number of cells in communicating populations of neurons. In many instances the exact number of neurons is not genetically predetermined, and competition reduces an excess of neurons by regulatory cell death (reviewed in Purves and Lichtman, 1985; Williams and Herrup, 1988). Most of our knowledge about developmental interactions between preand postsynaptic elements stems from studies on vertebrates, in which the influence of innervation on the numberoftargetcells has beenextensivelydescribed. For example, removal of the eye in newborn mice and avian embryos produces degeneration of retinal target neurons (DeLong and Sidman, 1962; Heumann and Rabinowicz, 1980; Levi-Montalcini, 1949). The effect of disrupting sensory neuron projections is not restricted to their direct synaptic partners. Damaging sensory neurons in the hair follicles of newborn rodents results in the loss of cortical neurons that are three synaptic relays removed (Van der Loos and Woolsey, 1973; Woolsey et al., 1981). A similar dependence of target neurons in the CNS on their peripheral sensory cells is also observed in the visual system of Drosophila melanogaster. Mutations that disrupt the development or connectivity of photoreceptors also produce structural abnormalities in the optic ganglia, the CNS components of the visual system (Power, 1943; Meyerowitz and Kankel, 1978; Fischbach, 1983; Fischbach and Technau, 1984; Steller et al., 1987). In the total absence of photoreceptor innervation, the optic ganglia are greatly reduced. This effect is more severe on those parts of the optic ganglia that normally receive direct input from photoreceptors. The first optic ganglion, the lamina, is completely missing in eyeless flies (Power, 1943; Fischbach, 1983). For several different mutations that affect both eye and brain, studies of genetically mosaic flies have shown that the optic lobe abnormality is a result of the abnormal gene function in the eye (Meyerowitz and Kankel, 1978; Fischbach and Technau, 1984). These data suggest that it is the aberrant or deficient photoreceptor projections that produce the optic lobe phenotype. While it is clear that optic ganglion development is dependent on the eye, little is known about the underlying cellular or molecular mechanisms. Innervation could be required forthegeneration,differentiation, or maintenance of optic ganglion neurons. Previous studies have shown that at least some neurons in the optic ganglia differentiate autonomously but require retinal input for their continued survival (Fischbach, 1983; Fischbach and Technau, 1984; Steller et al., 1987). However, excessive neuronal cell death has not been documented in the developing lamina of eyeless mutants, despite the fact that no lamina neurons (L-neurons) are found in the adult brains of these flies (Fischbach and Technau, 1984). In the crustacean Daphnia, differentiation of target neurons in the firstopticganglion dependson ingrowth of photoreceptor axons (Lopresti et al., 1973; Macagno, 1979), and a similar role of retinal innervation for the differentiation of L-neurons in Diptera has been previously suggested (Meinertzhagen, 1974). In the present study we have asked whether innervation is required for the birth and/or differentiation of first order interneurons in the lamina (L-neurons). To examine the generation and differentiation of L-neurons, we investigated the pattern of cell division and the expression of neuronal markers in the CNS at that time in development when photoreceptor axons are entering the brain. We find that expression of early neuronal markers in the presumptive lamina depends on retinal innervation. Moreover, the wave of mitotic activity that generates L-neurons is absent in mutants that completely lack photoreceptor projections. Reducing the number of innervating photoreceptor axons results in a decreased number of mitotitally active lamina precursor cells (LPCs). In another mutation, glass 0, photoreceptor axons project to abnormal locations in the brain. In this case, the spatial distribution of mitotic LPCs is abnormal and corresponds to the aberrant position of photoreceptor projections. These results suggest that ingrowth of Neuron 84 photoreceptor axons induces neuronal divide, thereby generating the target outer photoreceptor cells. precursors neurons for to the Results Expression of Neuronal Markers during Wild-Type Lamina Development We have examined the expression of several early neuronal differentiation markers during optic lobe development to determine precisely where and when relative to retinal innervation the differentiation of L-neurons begins. We find that L-neurons differentiate only after photoreceptor axons arrive in the brain. In accordance with the posterior to anterior progression of retinal innervation (Meinertzhagen, 1973,1974), expression of neuronal markers was first detected in the posterior region of the lamina, advancing anteriorly as development proceeds. The expression of one neuronal marker, elav (Campos et al., 1987; Robinow et al., 1988; Robinow and White, 1988), during different stages of visual system development is shown in Figure 1. This marker is especially useful, since it is expressed very early during neural differentiation, and its nuclear localization marks the positions of cell bodies (Campos et al., 1987; Robinow and White, 1988; Bier et al., 1988). In Figure lA,acryostatsection througha hrpupawasdouble labeled with anti-elav (green fluorescence) and MAb24B10, a monoclonal antibody against chaoptin (Zipursky et al., 1984,1985; Van Vactor et al., 1988) that stains photoreceptor axons (red fluorescence). At this stage the visual system appeared fully developed, and its structure could be easily related to the welldescribed adult pattern (see Trujillo-Cenoz, 1965, Meinertzhagen, 1973; Strausfeld, 1976; Kankel et al., 1980; Frohlich and Meinertzhagen, 1982; Shaw and Meinertzhagen, 1986; Fischbach and Dittrich, 1989). In the adult, each of the approximately800 ommatidia in a compound eye contains eight photoreceptor neurons, which, based on their spectral sensitivity and projection pattern, can be divided into three classes (for recent reviews see Tomlinson, 1988; Ready, 1989; Rubin, 1989)Theouter photoreceptor cells, RI-6, project to the first optic ganglion, the lamina (Figure 1). The two central photoreceptors, R7 and R8, send their axons to different regions of the second optic ganglion, the medulla (Figure 1). Like the eye, the lamina is composed of repeated units, called neuro-ommatidia or cartridges. Each cartridge contains five laterally located monopolar neurons, Ll-5 (L-neurons), which project to the medulla, as well as different glial elements, and a more centrally located amacrine neuron (reviewed in Strausfeld, 1976; Fischbach and Dittrich, 1989). The lamina cortex containing the cell bodies of L-neurons is visible as a distinct band of green fluorescing nuclei at the lateral margin of the pupal brain in Figure IA. The RI-6 axons penetrate this cell body layer and terminate deeper (Figure IA, arrowheads), somewhat lateral to the medulla cortex. Synaptic connections between photoreceptor axons and lamina neurons occur in the lamina neuropil, the region between the L-neuron cell bodies and the RI-6 termini (see, for example, Strausfeld, 1976; Frohlich and Meinertzhagen, 1982; Shaw and Meinertzhagen, 1986; Fischbach and Dittrich, 1989). In Figure IA, the lamina neuropil shows diffuse red staining. This is due to extensive ramification of photoreceptor axons in the region where they synapse onto L-neurons (see, for example, Fischbach and Dittrich, 1989). Through analysis of progressively earlier developmental stages (Figures IB and IC), we were able to examine L-neuron differentiation in late larvae, when photoreceptor axons are growing into the brain from the eye disc (Figure ID). The relationship between photoreceptor axons and the developing lamina is illustrated in the confocal images of third instar larval brains stained with anti-HRP antibody (Figures IF and IG), which stains membranes of all neurons (Jan and Jan, 1982). Photoreceptor axons encompass that region of the brain that contains differentiating L-neurons. At all stages, L-neurons could be readily identified bytheir position lateral to the RI-6termini and by their ability to express elav protein. Figure ID shows a horizontal optical section through a whole mount preparation of a late third instar larval eye disc and brain. This preparation was stained with MAb44C11, a monoclonal antibody that detects e/a\/ protein (red fluorescence; Bier et al., 1988; Robinow et al., 1988), and anti-HRP antibody to visualize photoreceptor axons (green fluorescence). The RI-6 axon endings are seen as a line of green fluorescence (Figure ID, arrowheads), resulting from their terminal arborizations. L-neurons are visible as two groups of elav-positive cells (Figure ID, arrows) lateral to the RI-6 endings. Significantly, no MAb44Cllstaining was detected in the anterior region of the prospective iamina, i.e., in the area that has very recently received retinal input or that hasyetto be innervated during later development (see below). Similar results have been obtained with other lamina differentation markers (IacZ markers generated by “enhancer trapping” [O’Kane and Gehring, 1987; K. Ressler, R. Chadwick, K. White, and H. Steller, unpublished data]). Our results on the relationship of photoreceptor axons and the expression of neuronal markers in the developing lamina are summarized in Figure IE. This figure also incorporates results from confocal microscopy analyses (examples are shown in Figures IF and ‘IQ, Di-I injections (data not shown), as well as previous work from other laboratories on visual system structure and development (for reviews see Meinertzhagen, 1973,1974; Kankel et al., 1980, Tomlinson, 1988; Ready, 1989). In late third instar larvae, photoreceptor cells have formed in the posterior half of the eye disc and project axons into the larval brain. As new photoreceptors are added in the anterior part of the eye disc Innervation-Dependent 05 Lamina Neurogenesis (Figure IE, indicated by orange), their axons project to the anterior edge of the developing lamina. The expression of neuronal markers was first detected in the posterior region of the lamina, spreading progressively more anterior as development proceeded. Neuronal differentiation in the lamina therefore follows the posterior to anterior arrival of photoreceptor axons (Meinertzhagen, 1973, 1974). However, elav-expression lagged behind axon arrival by several hours. Consequently, for any given receptor fascicle, L-neurons were not yet differentiated at the time when photoreceptor axons innervated the brain. Our results demonstrate that neuronal differentiation in the lamina, like that in the eye, proceeds in a posterior to anterior direction, as has been previously suggested by Meinertzhagen (1973, 1974). Expression of Neuronal Markers in the Lamina Depends on Retinal Innervation The previous results suggested that L-neurons differentiate only after the arrival of photoreceptor axons. To determine whether retinal innervation was necessary for L-neuron differentiation, we examined the expression of early neuronal markers in mutants lacking retinal innervation. One mutation, sine oculis (so), reduces the amount of photoreceptor neurons to a variable degree, often resulting in the complete elimination of all photoreceptors (Fischbach, 1983). Previous genetic mosaic studies have shown that the so mutation acts in theeye, and that theso+gene product is not required autonomously in L-neurons for their development (Fischbach and Technau, 1984). On the basis of these studies we can safely conclude that defects in the optic ganglia of so larvae are a consequence of abnormalities in eye development. Figure 2 depicts whole-mount nervous systems from wild-type and so late third instar larvae stained with MAb44Cll (red) and anti-HRP antibodies (green). so larvae devoid of photoreceptors completely lacked any cells expressing detectable levels of elav protein in the presumptive lamina (Figures 2B and 2E, compare with wild type in Figures 2A and 2D or Figure ID). In so larvae with a reduced number of photoreceptors (Figures 2C and 2F), there was a correspondingly smaller number of elav-positive cells in the developing laminacompared with wild type (Figures2Aand 2D). These findings demonstrate that in so larvae the number of cells in the developing lamina that express detectable levels of elav protein corresponds to the extent of photoreceptor innervation. We were also unable to detect expression of neuronal markers in the prospective lamina of larvae bearing the eyes absent (eya) and disconnected (disco) mutations, both ofwhich prevent retinal innervation (data not shown). These results indicate that L-neurons fail to develop in the absence of retinal input. However, we could not determine from this analysis whether retinal innervation dictated the mere differentiation of lamina cells, or their generation from neuronal precursors. Eyeless Mutants lack a Discrete Subset of Dividing Cells in the Brain To determine whether the production of L-neurons from precursor is affected by eye development, we have investigated the pattern of cell divisions in the CNS when photoreceptor axons are entering the brain. Late third instar larvae were injected with the thymidine analog BUdR,which is incorporated into replicating DNA. Previous studies have shown that DNA replication in third instar larvae represents neuroblast and ganglion mother cell division (Nordlander and Edwards, 1969a, 1969b; White and Kankel, 1978; Truman and Bate, 1988; Hofbauer and Campos-Ortega, 1990). Following a period of time to allow incorporation, the larval brainsweredissected,fixed,and incubatedwith an antibody specific for BUdR (Truman and Bate, 1988). anti-BUdR staining of whole-mount preparations allowed us to determine both the number and relative position of dividing cells in the entire brain. Figure 3A shows the pattern of BUdR incorporation for a wild-type third instar larva. In a lateral view of the brain hemisphere three arc-shaped domains of labeled cells were detected (Figures 3A-3C). The large broad stripe of labeled cells extending around the entire perimeter of the hemisphere represents mitotitally active cells in the outer proliferative center (OPC). The smaller, inner belt of labeled cells identifies the inner proliferative center (IPC). Previous studies of [3H]thymidine-labeled brains showed that the OPC produces medullary neurons, whereas neurons of the lobula complex and inner medulla derive from the IPC (White and Kankel, 1978; Hofbauer and Campos-Ortega, 1990). Between the OPC and IPC is a narrow stripe of approximately 100-150 BUdR-labeled cells, indicated with arrows in Figure 3A, which wecall LPCs. Pulse-chase experiments show that this mitotic domain generates L-neurons (see below). Hofbauer and Campos-Ortega (1990) have previously described this mitotic domain at the lateral margin of the OPC and reported that it forms the lamina. We have performed BUdR pulse-chase experiments that confirm these findings and demonstrate that L-neurons derive from LPC mitoses (see below). We refer to these dividing cells as LPCs to describe their properties and to distinguish them from other dividing cells of the OPC. To determine whether the generation of L-neurons is affected by retinal innervation, we have examined the pattern of BUdR incorporation in late third instar larval brains for different mutants that are completely devoid of photoreceptor axons. Figure 3G shows the pattern of cell division detected by BUdR incorporation in so third instar larval brains. In so third instar larvae without any photoreceptors, as judged by antiHRP antibody staining (Figure 3H), mitotic LPCs are absent, while the pattern of mitosis elsewhere in the proliferation centers appears normal (Figure 3G). Our inability to identify the LPC mitotic domain in these brains could be due either to the absence of LPC proliferation or to the misplacement of mitotic LPCs, for Innervation-Dependent Lamina Neurogenesis 87 example, their”fusion”with the broad zoneof mitotitally active cells in the OPC. We have performed a number of experiments to distinguish between these possibilities, and we will return to this point after describing in more detail the properties of the LPC mitotic domain. We have carried out a similar analysis for the mutation eya (Sved, 1986). eya eye discs display no photoreceptors at any developmental stage (Renfranz and Benzer, 1989; Figure 3F). We were unable to identify the mitotic LPCs in this mutant as well (compare Figures 3C and 3E). For the larva shown in Figure 3E, the general pattern of mitosis in the proliferative centers appeared unaffected. Thus, for two different eye mutations, the LPC mitotic domain is selectively lost in the absence of photoreceptor projections to brain. We have noted that approximately 10% of eya third instar larvae have a slightly abnormal OPC in addition to the loss of mitotic LPCs. We do not know the source of this variability. We have examined the pattern of cell division in third instar larvae of another mutant lacking photoreceptor innervation, disco (Steller et al., 1987). In disco mutants photoreceptor cells form but typically fail to project to the optic ganglia (unconnected phenotype). We were unable to identify mitotic LPCs in disco lar- Figure 1. Relationship of Photoreceptor Axons to L-Neurons during vae of the unconnected phenotype. However, as the other proliferative centers were also abnormal, the interpretation of these results is difficult (data not shown). In both eya and so brains without imaginal photoreceptors, a small number (10-20) of scattered, BUdRpositive nuclei areevident in thegeneral region where LPCs are usually found (Figures 3E-3H). Wild-type brains show a number of scattered mitotically active cells in addition to the LPCs (see, for example, Figures 4A and 48; Figures 5A and 5B). Their developmental fate is unknown, and it is possible that the few remaining BUdR-positive cells in the eyeless mutants correspond to these cells. Scattered mitotic cells have been observed in the developing lamina of the butterfly Danaus and reported to produce glial cells (Nordlander and Edwards, 1969a, 196913). Lamina Precursor Cells Produce L-Neurons, the Target Cells for Photoreceptor Axons The eyeless mutants we have examined (Figure 3) displaythe selective loss of adiscrete set of S-phase cells, those reported to produce the lamina (Hofbauer and Campos-Ortega, 1990). However, this earlier study did not distinguish between thedifferentcell typeswithin lamina. To confirm that LPCs do in fact produce L-neu- Development (A-C) Horizontal sections of wild-type pupal heads. Anterior is at the top. Neuronal nuclei appear green (anti-elav staining), and photoreceptor axons are stained red (MAb24BlO staining). All photographs are double exposures superimposing two individual fluorescent images. (A) Section through a 54 hr pupa (stage P8; Ashburner, 1989a). Axons from the outer photoreceptor cells, RI-6, penetrate the L-neuron cell body layer (green fluorescing nuclei marked with an arrow; la) and terminate at the point marked with the closed arrowheads. Axons from R7 and R8 axons project to the medullary neuropil (me). The lamina neuropil, where RI-6 axons established connections with their target cells, shows diffuse red staining. (B and C) Sections through a (B) 26 hr (stage P6) and (C) 18 hr (stage P5) pupa. RI-6 axons terminate in an S-shaped red line (marked with arrowheads). At these stages, no lamina neuropil can be detected. L-neurons expressing elav (green fluorescence; marked with an arrow) are seen in the region lateral to the RI-6 termini. The anterior-posterior (AP) axis of the R7 and R8 projections in the medulla is perpendicular to the AP axis in the lamina (see also Figure IE). A 90° rotation of the medulla during pupal development (Shatoury, 1956) gives rise to the first optic chiasm (compare [B] and [Cl to [A]). (D) Horizontal optical section through the brain of a late third instar larval whole-mount preparation. Neuronal nuclei were labeled with MAb44Cll (red fluorescence), and axons were visualized with FITC-conjugated anti-HRP (green fluorescence). Photoreceptors axons from the eye disc (ed) project through the optic stalk into the brain. Due to their terminal arborizations, the RI-6 axon termini are seen as a green line (labeled with an arrowhead). Two distinct groups of elav-positive L-neurons in the developing lamina are marked with arrows. Expression of elav is first detected in the posterior part of the lamina, 2-4 rows behind the most anterior fascicles of photoreceptor axons (marked with an open arrow in (D). R7 and R8 axons project to the medullary neuropil (me). (E) Diagramatic representation of retinal projections and L-neuron differentiation in third instar larvae. Photoreceptor axons from each ommatidium in the eye disc (ed) project as fascicles retinotopically to the brain (see for example, Trujillo-Cenoz and Melamed, 1973). Axons from the outer photoreceptor cells, RI-6, terminate in the lamina (green line labeled RI-6). Photoreceptor axons from posterior ommatidia in the eye disc (indicated in blue) project to the posterior lamina, and axons from anterior segments of the eye disc (orange) projecttoanteriorpositions inthelamina. R7and R8axonsprojecttothemedulla,withaxonsfrom moreposteriorommatidiaterminating deeper in the brain (indicated by “A+“; the projection pattern of anterior R7/8 axons is indicated in yellow, that of posterior axons in light blue; see also Meinertzhagen, 1973). Like differentiation of photoreceptors, L-neuron differentiation proceeds in a posterioranterior direction. Expression of early neuronal markers in L-neurons is first detected 2-4 columns of cells behind the most anterior photoreceptor axons, corresponding to 3-6 hr after their arrival in the brain (compare to Figure ID). (F and C) Confocal images of photoreceptor axon projections in third instar larval brains. Whole-mount preparations were stained with FITC-conjugated anti-HRP antibody to visualize photoreceptor axons. Both (F) and (G) are lateral views, with anterior to the left of the photograph. (F) shows photoreceptor axons entering the brain from the eye disc (ed), across the optic stalk. In (G) the eye disc has been removed. The “crescent moon” shaped region encompassed by photoreceptor axon projections defines the developing lamina (arrow marked la). The continuous white line (labeled with an open arrow in [F] and [G]) marks a furrow near the surface of the brain that defines the anterior boundary of the developing lamina. Abbreviations are as follows: A-P, anterior-posterior axis; eye, compound eye; ed, eye disc; la, lamina cortex; me, medullary neuropil; RI-6, outer photoreceptor axons. Bar in (A), 50 urn for (A)-(C); bars in (D), (F), and (C), 25 pm. Innervation-Dependent 89 Lamina Neurogenesis rons, we have conducted a series of BUdR pulse-chase experiments. To identify L-neurons these preparations were simultaneously stained with anti-elav antibody. Figures 4A and 4B show horizontal sections of a third instar larval brain pulse-labeled with BUdR for 2 hr. The plane of section relative to BUdR-labeled cells is indicated in the cartoon inset. BUdR-labeled LPCs (marked with open arrow, Figure 4B) are located at the anterior limit of the developing lamina, along the lateral margin of the brain. The group of BUdR-incorporating cells between the LPCs and the OPC do not form a continuous band along the dorsal-ventral axis as do the LPCs and are recognized as a scattered collection in the whole-mount preparations (see Figures 5A and 5B). Significant levels of elav expression are only found posterior to the LPC mitotic domain. elavexpression within the lamina is graded, with cells in the most posterior segments displaying the most intense signal (see also Figure ID). Notice that with the pulse-label there are only a few scattered BUdRincorporating cells within the body of the lamina itself. To follow the fate of LPCs, we injected climbing third instar larvae with BUdR and allowed them to continue development for various periods of time. Using this protocol, we obtain BUdR labeling of a limited number of cells. Evidently, the injected BUdR is bioavailable only for a relatively short period, permitting a discrete pulse to be delivered. Figures 4C and 4D show horizontal sections of two different prepupae, dissected and fixed 17-18 hr after injection as climbing third instar larvae. LPCs labeled during the late third instar larval period move as a block into the cortex of the lamina as development proceeds and occupy positions where e&expressing cells are found. Figure 3. Patterns of Cell Division in the Brain of Wild-Type In fact, LPC progeny reside along the entire depth of elav-expressing cells (along the medial to lateral axis), suggesting that all the neurons in lamina that express elav ultimately derive from LPC divisions. BUdR delivered during the third instar larval stage and chased to the adult labels a contiguous set of L-neurons, providing further evidence that these cells are indeed generated at this stage in development (data not shown; see also Hofbauer and Campos-Ortega, 1990). Lamina Precursor Cells Divide Adjacent to the Entry Point of Photoreceptor Axons The results from eyeless mutants indicate that photoreceptor axons entering the brain influence the pattern of cell division in the developing optic ganglia. We have examined wild-type late third instar larval brains in more detail to better define the anatomical relationship between photoreceptoraxon projections and mitotic LPCs. We find that the mitotically active LPCs are closely associated with those axons that have entered the brain most recently (Figure 5). LPCs abut the forward margin of the developing lamina, adjacent to the most recent photoreceptor axons to enter the brain (Figures 5A and 5B). A furrow located on the surface of the brain, at the anterior margin of the developing lamina (Meinertzhagen, 1973; White and Kankel, 1978) is visjble as a thin arc of fluorescence in the confocal image (Figure 5B, open triangles). Horizontal optical sections of the developing lamina reveal that LPCs divide at the surface of the brain, immediatelyanteriortotheentrypointof photoreceptor axons (Figures 5C and 5D). The RI-6 axons terminate somewhat deeper, where their endings give rise to a distinct green line (Figures 5C and 5D, arrowheads; see also Figure ID). L-neuron cell bodies reside and Eyeless Third lnstar Larvae Late third instar larvae were labeled with BUdR to visualize mitotically active cells. BUdR incorporation was detected using HRP immunohistochemistry. The panels in the left column are Nomarski images of wild type (A and C) and eyeless mutants (E: eya, G: so). The same brains were also stained with a FITC-conjugated anti-HRP antibody to label axons, and the corresponding images are shown in the right column (D, F, and C). All pictures are lateral views; anterior is to the left and ventral toward the bottom. (A and C) Wild-type larvae. (A) The large domains of BUdR-labeled cells represent cell divisions in the outer (OPC) and inner (IPC) proliferative centers. Proliferating LPCs are seen as a narrow ring of labeled cells (marked with arrows) between IPC and OPC. To the left of the brain hemisphere is the eye disc, which contains a stripe of BUdR incorporation just posterior to the morphogenetic furrow (out of focal plain). (B) Diagramatic representation of the proliferative centers and their relationship to the eye disc (ed), morphogenetic furrow (mf), and developing lamina (la). The black dots in (B) represent dividing LPCs, located at the anterior margin of the developing lamina (la; hatched). As development proceeds, the “wave” of proliferating LPCs sweeps from posterior to anterior in the direction indicated by arrows. The outer and inner proliferative centers are shown in black. The stippled region behind the morphogenetic furrow (mf) in the eye disc (ed) marks the region of the disc where ommatidia have already differentiated. (C and D) Pattern of mitosis relative to the photoreceptor axon projections in a wild-type larva. Photoreceptor axons project through the optic stalk (marked with a closed triangle in [D]) into the developing lamina. Mitotically active LPCs are found along the anterior margin of the photoreceptor projections. (E-H) BUdR incorporation pattern and anti-HRP antibody staining for the eyeless mutants eya (E and F), and so (G and H), respectively. Note the absence of mitotic LPCs (E and G) in these mutants without photoreceptor projections to the brain (F and H). The anti-HRPstained projections in the optic stalk of the eya larva ([F]; indicated with an open triangle) derive from the larval photoreceptor organ. The fluorescent signal in the brain (F and H) is due to higher order neurons that are out of the focal plain. No imaginal photoreceptor cells were detected in the eye discs of the eya or so brains shown here. The BUdR-labeled nuclei block the fluorescence signal and are seen as dark spots in the fluorescence photomicrographs. Abbreviations are as follows: ad, antenna1 disc; ed, eye disc; ipc, inner proliferation center, la, lamina; mf, morphogenetic furrow; opt, outer proliferation center. Bar, 50 urn. Innervation-Dependent 91 Lamina Neurogenesis in the the more posterior region, between the RI-6 termini and the lateral margin of the brain (compare Figures ID and IE). Figure 5E summarizes our results in diagramatic form. The Number of Dividing Lamina Precursor Cells Depends on the Amount of Retinal Innervation The pattern of BUdR incorporation in the developing optic ganglia of eyeless mutants indicates that retinal innervation influences the pattern of cell division in the larval brain. The close association of mitotic LPCs with newly arrived photoreceptor axons and the lack of adiscrete LPC mitoticdomain in the brain of eyeless mutants are consistent with the model that photoreceptor axons induce cell division of LPCs upon entering the brain. However, the apparent absence of a distinct LPC mitotic domain in eyeless mutants could also result from the “fusion” with another proliferation zone, e.g., mitotically active cells in the OPC. For example, one could imagine that normally retinal innervation represses the mitotic activity of cells between LPCs and the broad zone of proliferation in the OPC, resulting in their merger when photoreceptor axons are absent. To distinguish between these possibilities we first examined brains receiving reduced retinal innervation. If photoreceptor innervation serves Figure 2. elav Expression in Wild-Type to induce cell division of LPCs, we expected to find a reduced number of mitotic cells if the number of photoreceptor axons projecting to the brain was reduced. In this case, patches of BUdR-labeled cells should be found well separated from the OPC around isolated photoreceptor axon fascicles. This was indeed observed (Figure 6). In so larvae with reduced numbers of ommatidia, correspondingly fewer BUdRpositive cells were found at the anterior border of the photoreceptor projections (Figures 6A and 6B). In so larvae with very few, isolated clusters of ommatidia in the eye disc, discontinuous patches of mitotic cells were observed (Figures 6C and 6D). These results strongly support the idea that photoreceptor innervation induces cell division of LPCs (see Discussion). The Position of Mitotic LPCs Corresponds to the Location of Aberrant Photoreceptor Axons in gl Mutants The experiments we have described thus far make use of mutations that reduce the number of photoreceptor axons that project to the optic ganglia. The mutation gl allowed us to examine the relationship between axon ingrowth and cell division in a different way. Flies bearing the mutation gl have abnormal eye and brain morphology (Bridges and Morgan, 1923; Meyer- and so Larvae Horizontal optical sections of whole-mount late third instar larval brains stained with MAb44Cll (which recognizes elav protein in neuronal nuclei; red fluorescence) and anti-HRP (which stains axons; green fluorescence). Anterior is to the left with the lateral margin toward the bottom of the photograph. (A and D) Wild type. (B and E) so completely lacking photoreceptor cells. (C and F) so with a small number of ommatidia. (D-F) are enlargements of the pictures shown in (A)-(C). (A and D) In a wild-type larva, L-neurons expressing elav (red fluorescence; marked with arrows in [D]) are readily detected lateral to the RI-6 axon endings (green line marked with open arrow) and anterior to neurons derived from the IPC. The posterior margin of the lamina is marked with a small triangle. The intense MAb44Cll staining immediately posterior (to the right) of the lamina is due to neurons from the inner medulla and lobula complex. (B and E) In a so larva without any ommatidia in the eye disc (ed), no MAb44Cll staining can be detected in the region that normally becomes lamina. This region is located anterior to neurons derived from the IPC (to the left of the triangle) and is traversed by the larval optic nerve (marked with a thin arrow in (B); see Steller et al., 1987X (C and F) A so larva with a small number of ommatidia in the eye disc (labeled with a thin arrow in [Cl). In this case, only a few elav-positive L-neurons (arrow) can be detected beneath the RI-6 axon termini (marked with an open arrow). Bar in (A), 50 pm for (A)-(C); bar in (D), 20 pm for (D)-(F). Figure 4. Pulse-Chase Analysis of Lamina Precursor Cells Thefateof LPCswasdetermined byapuIse-chaseanalysisofBUdR-labeledcells.Thepositionof L-neuronswasassessed bysimultaneous staining with anti-elav antibody. (A) Horizontal section of climbing third instar larva pulse-labeled with BUdR in vitro for 2 hr. The cartoon inset shows a lateral view of a BUdR pulse-labeled CNS and eye disc. The bar indicates the approximate plane of section. BUdR-labeled cells display a brown immunohistochemical product, while elav-expressing cells are seen in black. Anterior is to the left and the lateral margin of the brain is toward the bottom of the photograph. (B) High magnification view of LPCs and developing lamina seen in (A). The position of the lamina furrow is indicated with an arrowhead. This represents the anterior margin of the developing lamina. The posterior margin of the lamina is marked with a small arrow. LPCs (open arrow) are found at the anterior border of the developing lamina. elav-immunoreactive L-neurons (stained black) are found posterior to mitotic LPCs. Lamina neurons show a graded expression of elav, with cells in the posterior segment containing the highest level of immunoreactive elav. The BUdR-incorporating cells within segments of the outer and inner proliferative centers are labeled OPC and IPC, respectively. The BUdR-labeled cells between the OPC and LPCs occupy a position along the lamina furrow. These cells do not form a continuous group along the dorsal-ventral axis (the axis perpendicular to the page) and are seen as a scattered collection in whole-mount preparations (see Figure 5). (C and D) Position of LPCs following a 17-18 hr chase of BUdR pulse-labeled cells. (C) and (D) are horizontal sections with the same orientation as (A) and (B) above, taken from two different prepupae. The plane of section is at the level of the optic stalk, somewhat moreventral than thatfor (A)and (B).Thelaminafurrowand posterior boundaryofthelaminaareindicatedasabove. BUdR-incorporating cells, which with a pulse label were located at the anterior margin of the lamina, are found within the body of the lamina following 17-18 hr of chase. These cells occupy the depth of the lamina cortex and overlap elav-expressing cells in the more posterior segments of the lamina. Bars in (A) and (C), 25 pm; bar in (B), 10 pm. (C) and (D) are the same magnification. Figure 5. Anatomical Relationship of Mitotically Active Lamina Precursor Cells to Photoreceptor Axons Late third instar larvae were pulse-labeled with BUdR and double-stained with FITC-conjugated anti-HRP and anti-BUdR antibodies to visualize photoreceptor axons (green in [A, C, and D]; blue in [B]) and mitotic cells (red), respectively. (A and B) Lateral views of the brain hemisphere taken with fluorescence (A) and confocal (B) microscopes. Anterior is to the left, and dorsal is toward the top of the photograph. Dividing LPCs (red nuclei; marked with an arrow) reside immediately anterior to the region innervated by photoreceptor axons. The high magnification confocal microscope image of the developing lamina (B) represents a lateral viewwith the focal plane just below the surface of the brain. The blue line between the open triangles represents a furrow at the anterior boundary of the developing lamina. (C and D) Horizontal optical sections of a whole-mount larval brain, with anterior to the left and lateral toward the bottom of the photograph. (D) is an enlargement of(C), showing the region of the developing lamina. Mitotic LPCs (marked with an arrow in [Cl) are located at the periphery of the brain, just in front of the most anterior photoreceptor axons. The RI-6 axons terminate somewhat deeper, forming a green line (marked with arrowheads in [C] and [D]). (E) Diagramatic representation of a horizontal section through a third instar larval brain. LPCs divide where photoreceptor axons enter the brain, in front of the most anterior axon fascicles (indicated in orange). Differentiation of L-neurons (indicated by yellow circles) is first detected two to four columns of cells posterior from that point (compare to Figure ID and 1E). Abbreviations are as in Figure 3. Bar in (A), 50 pm; bars in (B) and CD), 25 pm. (C) is the same magnification as (A). Innervation-Dependent 93 Lamina Neurogenesis normally reside (Figures 7A-7C). Simultaneous staining of gl brains for BUdR-labeled cells and with antiHRP antibody revealed that these S-phase cells are found at the anterior margin of the gl photoreceptor projections, and their position mirrors the abnormal location of the photoreceptor axons (Figures 7C and 7D; compare, for example, to Figure 3D or Figures 6A and .6B). The position of these cells relative to the photoreceptor axons and the other proliferative centers supports their assignment as LPCs. Their position is in stark contrast to that found in wild type, where they constitute a precise arc-shaped group of BUdRincorporating cells (see Figures 3A and 3C). The correspondence between the location of BUdR-incorporating cells and the aberrant photoreceptor projections in gl larvae provides positive evidence that photoreceptor ingrowth serves to signal the production of L-neurons. These data also provide an explanation for the finding that ommatidia bearing the gl mutation disrupt the structure of genetically wild-type optic ganglia (Meyerowitz and Kankel, 1978). Figure 6. Proliferation of Lamina Precursor with a Reduced Number of Photoreceptor Cells in so Larvae Projections to Brain Late third instar larvae were pulse-labeled with BUdR and double stained with anti-BUdRand anti-HRPantibodies.All photographs show lateral views with anterior to the left. (A and B) Photographs of the same so third instar larval brain, which had one sizable patch of ommatidia in the eye disc (ed). (A) is the Nomarski image of the BUdR-positive cells, and (B) shows the position of the photoreceptor axons, detected by anti-HRP staining. A reduced number of mitotic LPCs (marked with an arrowhead in [A]) can be seen along the anterior margin of the photoreceptor projections ([B]; compare to Figures 3C and 3D). (C and D) BUdR incorporation pattern from two different so larvae with very small, scattered patches of developing ommatidia. (C)An eye disc containing only two ommatidia (not shown) produced two distinct patches of mitotic lamina precursor cells (marked with arrowheads). (D) For this brain a few ommatidia were found in the dorsal portion of the eye disc only (data not shown). Clusters of BUdR-positive LPCs (arrows) can be recognized between the OPC and IPC in the dorsal half of the hemisphere only (dorsal is at the top). Bars, 25 urn. (C) and (D) are the same magnification. owitz and Kankel, 1978; Moses et al., 1989). Genetic mosaic experiments demonstrated that optic lobe abnormalities are a consequence of abnormal gl gene function in the eye, not the optic lobe (Meyerowitz and Kankel, 1978). Previous studies have shown that the gl mutation prevents terminal differentiation of photoreceptor cells (Zipursky et al., 1984; Ready et al., 1986; Moses et al., 1989). We have found that in addition to these well-characterized defects, photoreceptor axons in gl mutants project to the brain aberrantly (see Figure 7). In third instar larvae gl photoreceptor axons cross the optic stalk but project to abnormal positions in the brain (Figures 7D and 7F). This phenotype is variable in both of the presumed null alleles we have examined (g/’ and g/@i; Moses et al., 1989). BUdR labeling experiments of gl third instar larvae show a variable pattern of S-phase cells located between the OPC and IPC, in the region where LPCs Discussion In Drosophila, the normal development of the optic ganglia is dependent upon signals from the eye (see for example, Power, 1943; Meyerowitz and Kankel, 1978; Fischbach, 1983; Fischbach and Technau, 1984; Steller et al., 1987). In the absence of retinal innervation, the first optic ganglion (the lamina) is absent and the second ganglion (the medulla) is greatly reduced. We are interested in determining the events in optic ganglion development that require signals from the eye and the precise nature of these signals. In this study, we have examined the role of retinal innervation for the development of photoreceptor target neurons (L-neurons) in the first optic ganglion, the lamina. Two general models have been proposed to account for the absence of optic ganglion neurons in eyeless mutants. Retinula fibers could be required for the differentiation of L-neurons (Meinertzhagen, 1974). This phenomenon has been demonstrated for the crustacean Daphnia (Lopresti et al., 1973; Macagno, 1979). Alternatively, innervation may be required for the maintenance of autonomously differentiating L-neurons. It is clear that in the absence of innervation from the eye, some optic ganglion neurons degenerate (Fischbach, 1983; Fischbach and Technau, 1984; Steller et al., 1987). However, because excessive neuronal cell death has not been documented in the lamina, we investigated the possibility that retinal innervation is required for lamina neurogenesis. To determine whether mutations that disrupt eye development influence the production of L-neurons, we examined the pattern of cell divisions in larval brains usingthethymidineanalog, BUdR(Truman and Bate, 1988). Previous studies have shown that DNA replication in third instar larvae represents neuroblast and ganglion mother cells divisions (Nordlander and Edwards, 1969a, 1969b; White and Kankel, 1978; Tru- Figure 7. Patterns of BUdR Incorporation and Photoreceptor Projections in gl Late Third lnstar Larvae Late third instar larvae were pulse-labeled with BUdR and double-stained with antiBUdR and anti-HRP antibodies. All photographs show lateral views with anterior to the left. (A-C) Nomarski images of three different gW brains (lateral views) to demonstrate the variability in the position of mitotic LPCs. BUdR-labeled cells presumed to represent dividing I-PCs are marked with arrowheads. Notice the abnormal position of these ceils with respect to the wild type (see Figures 3A and 3C). (C and D) Nomarski (C) and anti-HRP fiuorescence (D) images of the same gW brain showing the location of mitotic LPCs with respect to photoreceptor axons. Note the abnormal pattern of photoreceptor projections compared with wild type (Figure 30) and the correspondence between the position of the photoreceptor axons and the BUdR-labeled LPCs. The BUdR-labeled nuclei appear as dark spots, because the histochemical stain blocks the background fluorescence. (E and F) Paired images as for (C and D) above of a gl’ brain. In this individual, the photoreceptor projections (labeled with a closed triangle) failed to branch out and did not terminate in the region of the lamin’s but proceeded to more posterior segments of the b:ain (F). No mitotic WCs could be detected between the IPC and OPC (E). Bars in (A) and (B), 50 ym; bars in (C)-(F), 25 Wm. man and Bate, 1988; Hofbauer and Campos-Ortega, 1990). In agreement with earlier autoradiography studies of sectioned material (White and Kankel, 1978; Hofbauer and Campos-Ortega, 1990), we detect three domains of proliferative activity (S-phase cells) in wild-type third instar larval brains. L-neurons derive from a wave of mitotic acitvity at the lateral margin of the OPC (see Figure 4; see also Hofbauer and Campos-Ortega, 1990). We refer to these cells as LPCs. Our studies of cell division patterns in wild type and mutants deprived of retinal innervation strongly suggest that ingrowth of photoreceptor axons induces cell division of LPCs, thereby generating L-neurons. The evidence in support of this conclusion may be summarized as follows: First, in wild-type larvae, LPCs divide immediately adjacent to the most recent photoreceptor axons to innervate the brain. Consequently, LPCs enter mitosis when photoreceptor axons are in close proximity and could readily receive local signals from them. Second, mitoticaily active LPCs are selectively absent in mutants that completely lack photoreceptors. In these mutants, L-neurons never appear, as judged by the failure to express early neuronal differentiation markers (see below). Third, reducing the amount of retinal innervation (in so individuals with a reduced number of ommatidia) results in a corresponding decrease of LPC mitotic activity. Photoreceptor axons projecting from so eye discs containing isolated ommatidia produce discontinuous patches of mitotic LPCs (see Figures 6C and 60). Because these patches indicate where the “normal” LPC mitotic domain would be located, these results strongly suggest that the lack of mitotic LPCs in eyeless mutants is not the result of their fusion with other mitotic domains. Fourth, in gl larvae, mitotic LPCs are found in abnormal positions, coincident with the aberrant location of photoreceptor axons characteristic of this mutant. This indicates that the mitotic response of LPCs can be q!jite flexible and that the position where cells divide, within limits, is dictated by the position of photoreceptor axons. Taken together these results implicate photoreceptor axons in the induction of lamina precursor cell divisions. While the lack of identifiable mitotic LPCs in eyeless mutants could be the result of their misplacement or fusion with several lines Qf evidence tion. A mechanism that other argue would proliferation against explain, this centers, interpretafor example, Innervation-Dependent 95 Lamina Neurogenesis the fusion of LPCs with the OPC in eyeless mutants cannot readily account for the association of S-phase cells with aberrant photoreceptor axons in gl larvae or with axons from patches of ommatidia in so mutants. In this context, it is important to note that we donotfindanydiscontinuityintheOPCofg/mutants or brains receiving partial innervation, which would be expected if mitotic LPCs fused with the OPC. Furthermore, as judged by lamina-specific enhancer trap markers (K. Ressler, R. Chadwick, K. White, and H. Steller, unpublished data), there is no evidence of an abnormally located lamina in eyeless mutants, which would result if LPCs were simply displaced. These markers also demonstrate that small clusters of L-neurons differentiate in so brains receiving partial innervation (data not shown; see also Figures 2C and 2F). Likewise, in gl larvae, L-neurons express laminaspecific markers but are found in aberrant positions (data not shown). The abnormal organization of the lamina in gl larvae is consistent with the hypothesis that photoreceptor axons direct lamina neurogenesis. While these data strongly indicate that photoreceptor axons stimulate proliferation of lamina precursor cells, we do not have evidence that photoreceptor axons affect LPC divisions directly. It is possible that other cells, e.g., glial cells, participate in the transmission of an inductive signal from photoreceptor axons to LPCs. Whatever the mechanism, it should be noted that the mitotic response of LPCs to retinal innervation is very localized and prompt. Several of the points cited above in support of the idea that ingrowth of photoreceptor axons induces the division of lamina precursor cells are derived from studies of mutants, and a few comments concerning the interpretation of these datawith respect to normal development are warranted. First, for two of the mutations we have examined, so and gl, genetic mosaic analyses have shown that mutant ommatidia can disrupt the development of genotypically normal optic ganglion neurons (Meyerowitz and Kankel, 1978; Fischbach and Technau, 1984). Our results provide an explanation for these findings. Second, the gl gene has been cloned (Moses et al., 1989) and is active in eye discs of third instar larvae but apparently not in developing optic ganglia at this stage (K. Moses and G. Rubin, unpublished data). Finally, genetic mosaic analyses for the so mutation show that the so+ gene product is not required in L-neurons or other components of the lamina for their normal development (Fischbach and Technau, 1984). Therefore, any abnormalities in the lamina, such as deficiencies in the number of mitotic LPCs, are almost certainly a consequence of defects in eye development and cannot be due to an autonomous effect of the mutation in the lamina. Our studies indicate that the disruption of photoreceptor development selectively affects the division of lamina precursor cells, but not the general pattern of mitoses in the developing optic ganglia. However, given the resolution of our analyses, we cannot be certain that the other proliferative centers are completely normal. Yet, even accepting the possibilitythat themutationswehaveexamined maysubtlyinfluence the OPC and IPC does not detract from the conclusion that innervation induces division of LPCs. It is important to note that several different eyeless mutants have the same effect on the mitotic activity of LPCs, without apparently altering the other proliferative centers. Furthermore, the position of the aberrant retinal projections in gl larvae predicts the location of mitotic LPCs. A global effect of these mutations on cell division in the developing optic ganglia could not account for these observations. We have also investigated the role of retinal innervation for lamina development by following the expression of early neuronal differentiation markers. Our results indicate that the expression of neuronal markers in the laminadependson retinal innervation. In wild-type third instar larvae, elav-positive cells appear in the presumptive lamina following innervation from the eye. elav-expressing neurons are first found in the posterior segments of the developing lamina. As development proceeds, progressively more anterior cells in the lamina express elav protein, demonstrating a posterior-to-anterior gradient of lamina differentiation (see also Meinertzhagen 1973, 1974). so larvae devoid of photoreceptors completely lack any cells in the presumptive lamina that express detectable elav protein. In so larvae with a reduced number of photoreceptors,there isacorresponding reduction of elav-positive L-neurons. We interpret these findings as an obvious consequence of the failure to stimulate LPC mitotic activity in the absence of retinal innervation. However, we do not know whether L-neurondifferentiation proceeds independentlyafter their generation from precursor cells or requires the continued presence of photoreceptor axons. We have noticed some striking similarities between the morphogenesis of the lamina and the retina in Drosophila. These similarities and our current view of the interaction between these tissues are schematically illustrated in Figure 8. Eye development proceeds from posterior to anterior (reviewed by Tomlinson, 1988; Ready, 1989; Rubin, 1989). Likewise, differentiation of L-neurons (Figure 8), as judged by the expression of neuronal markers, advances along a posterioranterior gradient, with the posterior regions being the furthest along in their development (see also Meinertzhagen, 1973, 1974; Trujillo-Cenoz and Melamed, 1973; White and Kankel,, 1978). The morphogenetic furrow in the eye disc (Figure 8) defines the boundary between cells assembling into ommatidia and those that have not yet begun to differentiate into photoreceptors. As photoreceptors develop they extend axons across the optic stalk into the brain. The most recent photoreceptor axons arrive at the anterior margin of the developing lamina (Meinertzhagen, 1973; Trujillo-Cenoz and Melamed, 1973). The proliferating LPCs are directly adjacent to these newly arriving axons. These divisions are found just posterior to a fur- Nf2”VXl 96 developmental time F A* l P mf medulla R-axons Figure during 8. Model for Development the Interaction between Eye and Lamina Schematic representation of a horizontal view of the developing Drosophila eye disc and lamina. Photoreceptor neurons begin to differentiate in the eye disc posterior to a dorsoventral indentation, the morphogenetic furrow (mf). As development proceeds, the morphogenetic furrow moves anteriorly, generating one new ommatidial column approximately every90 min (Ready et al., 1976; Campos-Ortega and Hofbauer, 1977; Basler and Hafen, 1989). As a result, the temporal sequence of photoreceptor differentiation is laid out spatially along the anterior-posterior (AP) axis of the eye disc (indicated by the arrow marked “developmental time”; reviewed in Tomlinson, 1988; Ready, 1989; Rubin, 1989). Likewise, differentiation of L-neurons (LN) progresses along the AP axis, advancing in synchrony with eye development. We propose that this synchrony is achieved by the innervation-dependent birth of L-neurons (LN). Soon after photoreceptor cells have emerged behind the morphogenetic furrow in the eye disc, they project axons through the optic stalk into the larval brain. This induces lamina precursor cells (LPCs) to divide, thereby generating L-neurons (LN), which differentiate and ultimately establish synaptic connections with RI-6 axons. Lamina precursor cells derive from the OPC (cross hatched circles) and are produced independent of retinal innervation. The small arrows indicate that cell division from the OPC also produces medullary neurons and more neuroblasts to maintain the OPC (see White and Kankel, 1978; Hofbauer and CamposOrtega, 1990). Abbreviations are as follows: A-P, anterior-posterior axis; If, lamina furrow; LN, lamina neurons; LPC, lamina precursor cell; mf, morphogeneticfurrow; OPC, outer proliferativecenter; R-axons, photoreceptor axons. row, which we suggest is the lamina counterpart of the eye morphogenetic furrow. The furrows found in eye and lamina anlagen share several characteristics. First, both are found at increasingly anterior positions as development proceeds. In fact the rate of movement of eye and lamina furrows is the same, as judged by the synchronous progression of eye and lamina proliferation zones (Hofbauer and Campos-Ortega, 1990). Second, each defines the anterior margin of a differentiating field. Finally, a discrete region of cell division is located just behind the furrow in eye and lamina. The mitotic activity of LPCs depends on photo- receptor innervation. We believe that each group of photoreceptor axons that enter the brain triggers the production of a set of L-neurons and by this means controls the number of target neurons. It is possible that photoreceptor axon-induced production of target neurons serves not only to control cell number, but also to facilitate the establishment of correct retinotopic connections between photoreceptor axons and L-neurons. Becauseof the repetitive nature of both eye and lamina, photoreceptor axons have to find and recognize their correct synaptic partners from among an abundance of candidate target cells. RI-6 axons from an individual ommatidium ultimately project to L-neurons in six different cartridges (see, for example, Trujillo-Cenoz and Melamed, 1966; Braitenberg, 1967; Kirschfeld, 1967). One can imagine that the coordinated posterior-to-anterior progression of both eye and famina differentiation may actually simplify the development of proper connectivity. Cells along the posterior-to-anterior axis represent different stages of photoreceptor and lamina neuron development. Consequently, the anterior-posterior spatial dimension is also represented as a gradient in developmental time. The distinctions between cells in different developmental stages along the anteriorposterior axis may promote recognition between the appropriate L-neurons and photoreceptor axons. Such a mechanism would obviously require that the developmental “clocks” in the retina and lamina are precisely coupled with each other. Previous studies have indicated remarkable synchrony in the progression of eye and lamina morphogenesis (TrujilloCenoz, 1965; White and Kankel, 1978; Hofbauer and Campos-Ortega, 1990). Our finding that the generation of L-neurons is under control of retinal innervation suggests an obvious explanation for how this synchrony is achieved. The control of neurogenesis in the Drosophila lamina by retinal innervation may serve as an economic alternative to regulatory cell death for determining neural cell number. In other systems, neuron-induced cell division has been described for the interaction between Schwann cells and dorsal root ganglion processes in vitro (Ratner et al., 1988). In this case, the mitogen is associated with the plasma membrane and requires contact between dorsal root ganglion neurons and the responding Schwann cells. There is some evidence implicating interactions between neurons in the control of cell division in the vertebrate CNS. Removal of the eye prior to axon outgrowth in the frog produces a decrease in mitotic figures within the developing tectum (Kollros, 1982). Indirect evidence from several studies (reviewed in Williams and Herrup, 1988) suggests that Purkinje cells in the mammalian cerebellum induce cell division in the external granule layer. Very recently it has been reported that peripheral organs stimulate the production of CNS neurons in the leech (Baptista et al., 1990). However, substantially more work is required to establish whether the control of neurogenesis by innervation is a wide- Innervation-Dependent 97 Lamina Neurogenesis spread means of determining veloping nervous system. Experimental cell number in the de- Procedures Drosophila Culture All fly strains were grown on standard cornmeal medium (Cline, 1978) at 18OC or 25’C. Canton-S served as the wild-type strain. gP and #were kindly provided by Drs. Kevin Moses and G. M. Rubin. Cryostat Sections and Antibody Staining For frozen sections of pupal stages, the pupal case was opened and pupae were fixed in paraformaldehyde-lysine-periodate (McLean and Nakane, 1974) for 1 hr at room temperature. The tissue was washed and freeze-protected by overnight submersion in 20% sucrose in phosphate buffer at 4OC. Heads were mounted in OCT on cryostat chucks, and 10 urn sections were cut on a Reichart-Jung model 2800-Frigocut microtome at - 18’-‘C. Sections were picked up on subbed slides, air-dried, and postfixed in 2% paraformaldehyde. Antibody staining was performed as previously described (Steller et al., 1987). Antibody dilutions were I:100 for both anti-HRP (Cappel) and anti-elav (Robinow et al., 1988) and I:1 for MAb24BlO (Zipursky et al., 1984). Whole-Mount Antibody Staining and Confocal Laser Scanning Microscopy Larvae were dissected in 0.1 M phosphate buffer (0.1 M sodium phosphate [pH 7.21). For staining with MAb44Cll (diluted 1:5; Bier et al., 1988), dissected nervous systems were incubated in 1 mg/ml collagenase for 15 min at room temperature to aid antibody penetration. Subsequently, the tissue was briefly rinsed in phosphate buffer, fixed in 2% paraformaldehyde, and stained with antibodies as previously described (Steller et al., 1987). For confocal microscopy analyses, dissected eye discbrain complexes were stained with an anti-HRP antibody conjugated with FITC (diluted 1:200, Capel). For simultaneous detection of BUdR incorporation and photoreceptor axons with the confocal microscope, brains were incubated with anti-BUdR monoclonal antibody as described below, followed by biotinconjugated goat anti-mouse IgG (Cappel] and Texas red-conjugated avidin. The goat secondary antibody was used at a I:100 dilution and incubated at 4OC overnight. The Texas red-avidin was added at I:50 dilution and incubated for 20 min at room temperature, followed by four washes in balanced salt solution (Ashburner, 1989b) prior to mounting. Incubation with the antiHRP antibody was at the same time as the goat secondary antibody. Samples were viewed on an MRC 500 confocal scanning laser microscope. Image processing was performed with software provided by the manufacturer. BUdR Labeling of Third lnstar larvae We employed the procedure described by Truman and Bate (1988) with some minor modifications as follows. Climbing third instar larvae were washed briefly in distilled Hz0 to remove debris. They were then injected with 0.1-0.2 ~.rl of 100 ug/ml BUdR (Boehringer Mannheim) in injection buffer (0.1 mM phosphate buffer [pH 6.8],5 mM KCI) with a drawn out 5 ul microcapillary attached to a syringe. Larvae were injected in the posterior half, on the ventral surface to avoid damage to the brain hemispheres. Injected larvae were placed on moist Whatman’s filter in a petri dish for 3 hr at room temperature to permit incorporation of label. Larvae were then dissected in ice-cold PBS (130 mM NaCI, 7 mM Na2HP0,, 3 mM NaHzP04). Brain-eye disc complexes were placed in 2% paraformaldehyde or Carnoy’s fixative for l-l.5 hr. After fixation in Carnoy’s fixative, brains were rehydrated to 1 x PBS plus 0.3% Triton X-100 through 5 min washes in 80%, 60%, and 40% ethanol. Typically the brains were stored overnight in 1 x PBS, 0.3% Triton X-100 at 4OC. Brains were then incubated in 2 N HCI, 1 x PBS, 0.3% Triton X-100 for 60 min and rinsed twice in 1 x PBS, 0.3% Triton X-108. Brains were then washed twice in balancedsaltsolutionfor5mineach, priortoa45min incubation at room temperature in balanced salt solution including 5% goat serum and O.l%-0.3% Triton X-100. Antibody staining was as previously described (Steller et al., 1987). Primary antibodies were detected using HRP immunocytochemistry or immunofluorescence with FITC- or RITC-conjugated antibodies. For the BUdR pulse-chase analysis, climbing third instar larvae were injected with BUdR as described above and allowed to develop for up to 18 hr at 25OC. The CNS and eye disc were dissected as a unit from the larvae or prepupae and stained with anti-BUdRantibodyasoutlinedabove. Following thehistochemical detection of BUdR, the whole-mount CNS was postfixed in formaldehyde-acetic acid-ethanol (Blest and Davie, 1980) plus 1% glutaraldehyde for 1 hr at room temperature, infiltrated, and embedded in paraffin following dehydration. The paraffin blocks were sectioned at 6 urn thickness and placed on subbed slides. Sections were stained with anti-elav antibody according to the procedure of Robinow and White (unpublished data; see also Robinow, 1990), with the exception that the histochemical reaction was performed in the presence of 0.2 mg/ml NiCI, which results in a black histochemical product readily distinguished from the usual, brown product generated from diaminobenzidine in the absence of NiCI. The pulse-labeled third instar larva shown in Figure 4 was labeled in vitro by incubating dissected brains in Schneider’s medium containing 20 uglml BUdR. Acknowledgments We would like to thank Drs. E. Bier, Y. N. Jan, L. Jan, S. Robinow, K. White, and S. Benzer for kindly providing neuron-specific antibodies used in this study. We are grateful to Don Doering for introducing us to the use of the confocal microscope. We thank Drs. C. Bargmann, R. Hynes, R. MacKay, S. Robinow, D. Vollrath, T. Orr-Weaver, K. White, and our colleagues in the Steller lab for their comments on a draft of this manuscript. We also thank Drs. A. Hofbauer and J. Campos-Ortega for sharing results prior to publication. Some of the BUdR pulse-chase analyses were conducted in the laboratory of Dr. K. White, Brandeis University. This research was supported by National Institutes of Health grant ROl-NS26451 and a Pew Scholars Award to H. S. S. 8. S. is a Burroughs Wellcome Fund Fellow of the Life Sciences Research Foundation. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked “advertisement”in accordance with 18 USC Section 1734 solely to indicate this fact. Received June 19, 1990; revised October 2, 1990. References Ashburner, M. (1989a). Drosophila. (Cold Spring Harbor, NewYork: Cold A Laboratory Spring Harbor Ashburner, M. (1989b). Drosophila. A Laboratory Spring Harbor, New York: Cold Spring Harbor Handbook. Laboratory). Manual. (Cold Laboratory). Baptista, C. A., Gershon, T. R., and Macagno, E. E. (1990). eral organs control central neurogenesis in the leech. 346, 855-858. Basler, K., and Hafen, enless: position-dependent 931-934. PeriphNature E. (1989). Ubiquitous expression of sevspecification of cell fate. Science243, Bier, E., Ackerman, L., Barbel, S., Jan, L., and Jan, Y. N. (1988). Identification and characterization of a neuron-specific nuclear antigen in Drosophila. Science 240, 913-916. Blest, A. D., and Davie, P. S. (1980). Reduced silver impregnations derived from the Holmes technique. In Neuroanatomical Techniques, N. J. Strausfeld and T. A. Miller, eds. (New York: Springer-Verlag). Braitenberg, V. (1967). Patterns of optic projections in the visual system of the fly. I. Retina-lamina projections. Exp. Brain Res. 3, 271-298. Bridges, C. B., and Morgan, T. H. (1923). The third chromosome group of mutant characters of Drosophila gie Inst. Wash. Publ. 327, 188190. melanogaster. Carne- nection patterns in the Symp. 29, 265-288. visual systems of insects. Ciba Found. Campos,A. R., Rosen, D. T., Robinow, S. N., and White, K. (1987). Molecular analysis of the locus elav in Drosophila melonogaster: a gene whose expression is neural specific. EMBO J. 6425-431. Meyerowitz, E. M., and Kankel, D. (1978). visual system development in Drosophila Biol. 62, 112-142. Campos-Ortega, J. A., and Hofbauer, A. (1977). Cell clones and pattern formation: on the lineage of photoreceptor cells in the compound eye of Drosophila. Roux’s Arch. Dev. Biol. 787, 227245. Moses, K., Ellis, M. C., and Rubin, C. (1989). The glass gene encodes a zinc-finger protein required by Drosophila photoreceptor cells. Nature 340, 531-536. Cline, T. W. (1978). Two closely linked melanogaster that are lethal to opposite daughterless. Genetics 90, 683-698. mutations in Drosophila sexes and interact with A genetic analysis of melanogaster. Dev. Nordlander, R. H., and Edwards, J. S. (1969a). Postembryonic brain development in the monarch butterfly, Danaus piexippus plexippus, L. I. Cellular events during brain morphogenesis. W. Roux’s Arch. Dev. Biol. 762, 197-217. DeLong, G. R., and Sidman, R. L. (1962). Effects of eye removal at birth on histogenesis of the mouse superior colliculus: an autoradiographic analysis with tritiated thymidine. J. Comp. Neurol. 778, 205-219. Nordlander, R. H., and Edwards, J. S. (1969b). Postembryonic brain development in the monarch butterfly, Danaus plexippus plexippus, L. Ii. The optic lobes. W. ROUX’S Arch. Dev. Biol. 763, 197-220. Fischbach, K. F. (1983). Neural sensory deprivation in the optic ter. Dev. Biol. 95, 1-18. O’Kane, C. j., and genomic regulatory USA 84, 9123-9127. cell types surviving lobes of Drosophila congenital melanogas- Fischbach, K. F., and Dittrich, A. P. M. (1989). The optic Drosophila melanogaster. I. A Golgi analysis of wild-type ture. Cell Tissue Res. 258, 441-475. lobe of struc- Fischbach, K. F., and Technau, C. (1984). Cell degeneration in the developing optic lobe of the sine oculis and small-optic lobes mutants of Drosophila melanogaster. Dev. Biol. 104, 219-239. Gehring, elements W. J. (1987). in Drosophila. Power, M. E. (1943). The effect of reduction in numbers of ommatidia upon the brain of Drosophila melanogaster. J. Exp. Zool. 94, 33-71. Purves, opment D., and Lichtman, j. W. (1985). Principlesof (Sunderland, MA: Sinauer Assoc.). Frohlich, A., and Meinertzhagen, I. A. (1982). Synaptogenesis in the first optic neuropile of the fly’s visual system. J. Neurocytol. 77, 159-180. Ratner, N., Hong, D., Lieberman, (1988). The neuronal cell-surface Schwann cells isa heparin-binding USA 85, 6992-6996. Heisenberg, M., and Wolf, Springer-Verlag). Ready, D. F. (2989). A multifaceted approach ment. Trends Neurosci. 72, 102-110. R. (1984). Vision in Drosophila (Berlin: Detection rn situ of Proc Natl. Acad. Sci. Neural Devel- M., Bunge, R., and Glaser, L. molecule mitogenic for protein. Proc. Natl.Acad. Sci. develop- Heumann, D., and Rabinowicz, T. (1980). Postnatal development ofthedorsal lateral geniculate nucleus in the normal and enucleated albino mouse. Exp. Brain Res. 38, 75-85. Ready, D. F., Hanson,T. of the Drosophiia retina, 217-240. Hofbauer, A., and Campos-Ortega, J. A. (1990). Proliferation pattern and early differentiation of the optic lobes in Drosophila melanogaster. Roux’s Arch. Dev. Biol. 798, 264-274. Ready, D. F., Tomlinson, A., and Lebowitz, R. M. (1986). Building an ommatidium: geometry and genes. In Development of Order in the Visual System, S. R. Hilfer and J. B. Sheffield, eds. (New York: Springer-Verlag), pp. 97-125. Jan, L. Y., and Jan, Y. N. (1982). Antibodies to horseradish peroxidase as specific neuronal markers in Drosophilaand in grasshopper embryos. Proc. Nati. Acad. Sci. USA 79, 2700-2704. Kankel, D. R., Ferrus, A., Caren, S. H., Harte, P. J., and Lewis, P. E. (1980). The structure and development of the nervous system. In The Genetics and Biology of Drosophila, Vol. 2D, M. Ashburner and T. R. F. Wright, eds. (London: Academic Press). Kirschfeld, K. (1967). Die Projektion der optischen das Raster der Rhabdomere im Komplexauge von Brain Res. 3, 248270. Kollros, J. J. (1982). Peripheral in the frog. J. Comp. Neurol. control of midbrain 205, 171-178. Umwelt Musca. auf Exp. E., and Benzer, a neurocrystalline to neural 5. (1976). Development lattice. Dev. Biol. 53, Renfranz, P. J., and Benzer, S. (1989). Monocional antibody probes discriminate early and late mutant defects in development of the Drosophila retina. Dev. Biol. 736, 411-429. Robinow, S. (1990). The elav gene of Drosophila melanogaster encodes a neuron-specific RNA binding protein which is required for the development and maintenance of the nervous system. Ph.D. thesis, Brandeis University, Waltham, Massachusetts. Robinow, S., and White, K. (1988). l-he locus e/a~ of Drosophi!a melanogaster is expressed in neurons at all developmental stages. Dev. Biol. 726, 294-303. mitoticactivity Levi-Montalcini, R. (1949). The development of the acousticovestibular centers in the chick embryo in the absence of afferent root fibers and of descending fiber tracts. J. Comp. Neural. 97, 209-241. Lopresti, V. A., Macagno, E. R., and Levinthal, C. (1973). Structure and development of neuronal connections in isogenic organisms: cellular interactions in the development of theoptic lamina of Daphnia. Proc. Natl. Acad. Sci. USA 70, 433-437. Robinow, S., Campos,A. R., Yoa, K.-M., and White, K. (1988). The elavgene product of Drosophila, required in neurons, has three RNP consensus motifs. Science 242, 1570-1572. Rubin, ductive C. M. (1989). Development of the Drosophila retina: inevents studied at single cell resolution. Cell 57,519-520. Shatoury, H. H. E. (1956). Differentiation and metamorphosis of the imaginal optic glomeruli in Drosophila. J. Embryol. Exp. Morphol. 4, 240-247. Macagno, E. R. (1979). Cellular interactionsand pattern formation in the development of the visual system of Daphnia magna (Crustacea, Brachiopoda). I. Interactions between embryonic retinular fibers and laminar neurons. Dev. Biol. 73, 206-238. Shaw, S. R., and Meinertzhagen, gression at synaptic connections gous neurons. Proc. Natl. Acad. McLean, I. W., and Nakane, P. K. (1974). Periodate-lysineparaformaldehyde fixative. A new fixative for immunoelectron microscopy. J. Histochem. Cytochem. 22, 1077-1083. Steller, ii., Fischbach, K. F., and Rubin, G. M. (1987). nected: a locus required for neuronal pathway formation visual system of Drosophila. Cell 50, 1139-1153. Meinertzhagen, I. A. (1973). Development of the compound eye and optic lobe of insects. In Developmental Neurobiology of Arthropods, D. Young ed. (Cambridge: Cambridge University Press), pp. 51-104. Strausfeld, N. J. (1976). Springer-Verlag). Meinertzhagen, I. A. (1974). The development of neuronal con- Sved, J. (1986). Tomlinson, Drosphila Atlas eyes absent. I. A. (1986). Evolutionary promade by identified homoloSci. USA 83, 7961-7965. of an Dros. Insect inf. Serv. A. (7988). Cellular interactions eye. Development 104, 183-193. Brain (New disconin the York: 63, 169. in the developing Innervation-Dependent 99 Lamina Neurogenesis Tomlinson, A., and Ready, D. F. (1987). Neuronal differentiation in the Drosophila ommatidium. Dev. Biol. 7.20, 366-376. Trujillo-Cenoz, 0. (1965). zation of the intermediate 73, l-33. Some aspects of the structural organiretina of Dipterans. J. Ultrastruct. Res. Trujillo-Cenoz, O., and Melamed, J. (1966). Dipterans: anatomical basis for integration-an scope study. J. Ultrastruct. Res. 76, 395-398. Trujillo-Cenoz, the retina-lamina 42, 554-581. Compound electron O., and Melamed, J. (1973). The development complex in Muscoid flies. J. Ultrastruct. Truman, J. W., and Bate, M. (1988). Spatial of neurogenesis in the central nervous melanogaster. Dev. Biol. 725, 145-157. eye of microof Res. and temporal patterns system of Drosophila Van der Loos, H., and Woolsey, T. A. (1973). Somatosensory cortex: structural alterations following early injury to sense organs. Science 179, 395-398. Van Vactor, D., Jr., Krantz, D. E., Reinke, R., and Zipursky, S. L. (1988). Analysis of mutants in chaoptin, a photoreceptor cellspecific glycoprotein in Drosophila, reveals its role in cellular morphogenesis. Cell 52, 281-290. White, K., and Kankel, D. R. (1978). Patterns of cell division and cell movement in the formation of the imaginal nervous system of Drosophila melanogaster. Dev. Biol. 65, 296-321. Williams, number. R. W., and Herrup, Annu. Rev. Neurosci. K. (1988). The 17, 423-453. control of neuron Woolsey, T. A., Durham, D., Harris, R. M., Simons, D. J., and Valentino, K. L. (1981). Somatosensorydevelopment. In Development of Perception, Vol. 1, R. N. Aslin, J. R. Alberts, and M. R. Petersen, eds. (New York: Academic Press, Inc.), pp. 259-292. Zipursky, S. L., Venkatesh, T. R., Teplow, D. B., and Benzer, S. (1984). Neuronal development in the Drosophila retina: monoclonal antibodies as molecular probes. Cell 36, 15-26. Zipursky, S. L., Venkatesh, T. R., and Benzer, S. (1985). From monoclonal antibody to gene for a neuron-specific glycoprotein in Drosophila. Proc. Natl. Acad. Sci. USA 82, 1855-1859.