Central projections of photoreceptor axons originating from ectopic eyes in Drosophila

Abstract

Ectopic expression of the retinal determination gene eyeless (ey) induces the formation of supernumerary eyes on antennae, legs, wings, and halteres. These ectopic eyes form ommatidia that contain photoreceptors and accessory cells and respond to light. Here, we demonstrate that ectopic eyes on antennae and legs extend axonal projections to the central nervous system. Furthermore, electroretinograms and morphological evidence indicate that the photoreceptor axons of at least the antennal ectopic eyes can form completely constituted ectopic synapses with foreign postsynaptic elements and suggest that transmission at these sites may be functional. However, the ectopic axons do not connect to their correct optic lobe targets and do not project deeply into the neuropile, but rather form synapses at superficial positions in the neuropils. By means of confocal and electron microscopy we show that these ectopic synapses resemble normal synapses, albeit with some distinct morphological differences. Our data strongly suggest that the developmental programs controlling photoreceptor synaptogenesis and visual map formation depend to a considerable extent on presynaptic and thus photoreceptor-autonomous steps. Our data also suggest that photoreceptor axon projections and the establishment of the highly stereotypical neural circuitry in the optic lobe, the normal target neuropil, may depend on target-specific cues that appear to be absent from the antennal lobe and thoracic ganglion.

The compound eyes of the Drosophila visual system relay visual information to the underlying optic lobes via four successive neuropils, the distalmost lamina, medulla, and proximal lobula and lobula plate (1). The compound eyes consist of ≈800 ommatidia, each of which contains eight photoreceptors and 11 accessory cells (2). Photoreceptors R1–R6 project to the lamina, whereas R7 and R8 axons terminate in the medulla (reviewed in refs. 1 and 3).

In the past decade, considerable insight has been gained in the genetic basis of early eye development, photoreceptor determination and ommatidium formation, and circuit assembly in the optic lobe. Early eye development is controlled by a genetic cascade comprising the Pax6 genes eyeless (ey) and twin of eyeless (toy), considered master control genes acting as the first step in the hierarchy, the SIX family members sine oculis (so) and optix, eyes absent (eya), dachshund (dac), and the Pax6-related gene eyegone (eyg). These genes act in complex gene networks and are integrated with signaling pathways (reviewed in refs. 4–6). Some of these genes individually, or in combination with others, are capable of inducing the formation of ectopic eyes (7–12).

Photoreceptor differentiation is initiated at the posterior margin of the eye imaginal disc with the formation of the morphogenetic furrow and involves the integration of the decapentaplegic, hedgehog, and wingless signaling pathways with the retinal determination gene network (reviewed in refs. 5 and 13). Signaling by the Notch and Epidermal Growth Factor Receptor (EGFR) pathways then leads to R8 photoreceptor selection and sequential addition of all photoreceptor and accessory cells that constitute an ommatidium (reviewed in ref 14).

Once photoreceptors have differentiated, proper connections in the brain need to be established. R8 photoreceptor axons first enter the lamina where they induce both a final cell division among lamina precursor cells and the differentiation of lamina neurons (15). These processes require Hedgehog (Hh) and EGFR signaling, respectively (16–18). R8 axons also establish a scaffold along which glia migrate (19). Glia then direct the correct targeting of R1–R6 axons in the lamina (20), a process that requires a combination of signaling and transcription factor-dependent events (reviewed in ref. 3). Finally, R1–R6 target selection results in the stereotypical pattern seen in the lamina and involves three cell surface molecules, the classical cadherin, N-cadherin, the receptor tyrosine phosphatase LAR, and the nonclassical cadherin flamingo. These cell surface molecules with PTP69D are also required for layer-specific targeting in the medulla (reviewed in ref. 3).

Starting in the second half of pupal development, the terminals of R1–R6 begin to form synapses when prospective postsynaptic dendrites contact nascent presynaptic sites over the surface of the terminal. Each site eventually accumulates a postsynaptic tetrad of postsynaptic elements. These invariably contain contributions from two lamina neurons, L1 and L2, and two contributions from three other lamina cells, two neurons (L3 and amacrine cells), and a glial cell, in variable combinations (21). These elements accumulate one by one, in a nondeterminate sequence that excludes at any one time incorrect or supernumerary participants (22). The assembly sequence implies the action of a concatenated sequence of cell recognition steps, but molecular details for these are lacking. The presynaptic site is marked by the accumulation of synaptic vesicles and the appearance of a presynaptic density, the T-bar ribbon (23, 24), first as a pedestal and then surmounted by a platform (25).

In summary, currently available data all indicate that the highly organized anatomy of the visual system is the result of a precisely organized cascade of molecular events, many of which still remain to be discovered.

In the current study, we determine the extent to which ectopically induced photoreceptors are able to establish synaptic contacts and whether any of the ordered organization of the optic lobe can be generated when ectopic photoreceptor axons project to novel targets.

Results

Electrical Activity of Ectopic Eyes.

We have previously reported that the ectopic expression of ey results in ectopic compound eyes with a full complement of cell types and light-evoked activity (10). To determine whether ectopic photoreceptor neurons form synapses, we started by measuring electroretinograms (ERGs) of ectopic eyes. A normal ERG consists successively of a positive “on” transient, a sustained negative potential, and a negative “off” transient (see Fig. 1A). The on and off transients are observed at the beginning and end of the light stimulus and originate in the lamina (26). The negative sustained potential represents the primary light-evoked response of the photoreceptor neuron cell bodies (27, 28). We recorded ERGs of ectopic eyes from females with the following three genotypes: (i) E132/w;UE10/+, (ii) MS1096/w;UE10/+, and (iii) w/w;dpp^blinkGAL4/UE10. MS1096/w;UE10/+ leads to the formation of supernumerary eyes on the dorsal part of the wing (29), whereas the other genotypes have ectopic eyes on antennae, legs, and wings (10). A small negative sustained photoreceptor potential was measured for: w/E132;UE10/+ adults (wings, 7 of 10; antennae, 6 of 6; leg, 1 of 1); MS1096/w;UE10/+ flies (wings, 2 of 6)(see Fig. 1 B and C for examples). The most comprehensive set of ERG recordings was obtained for w/w;dpp^blinkGAL4/UE10 flies. Because these flies have distorted legs and fail to hatch, all recordings were made on adults that were manually removed from the pupal case. For a limited number of ectopic eyes on the antennae, we were able to record on and off transients. To control for possible leakage from the adjacent normal eye, we covered the ipsilateral eye completely with a coating of correction fluid (TippEx) and eventually covered the complete head with the exception of the ectopic antennal eye with the same coating. In all cases, we continued to observe the on and off transients (see Fig. 1D). Results for w/w;dpp^blinkGAL4/UE10 flies are summarized in Table 1.

Fig. 1.

Electroretinograms of normal and ectopic eyes (A–D) and confocal images of ectopic photoreceptor axon projections (E–O). (A) ERG of a WT eye consists of a positive on transient (left arrowhead), a negative sustained potential of photoreceptor origin, and a negative off transient (right arrowhead). The transients coincide with the onset and offset of the light. (B) ERG of an ectopic eye on the leg of a w/w;dpp^blinkGAL4/UE10 fly. Only the photoreceptor potential can be detected. (C) ERG of an ectopic eye on the wing of a MS1096/w;UE10/+ fly. A small photoreceptor potential is recorded. Also note the slow repolarization after lights off. (D) ERG of an ectopic eye on the antenna of a w/w;dpp^blinkGAL4/UE10 fly. A small photoreceptor potential is observed. In addition, on and off transients were recorded. The bar beneath the ERG recording indicates the duration of the light stimulus. (E) Schematic overview of ectopic photoreceptor axons (marked with asterisk and arrow) projecting to the antennal lobe (AL) glomeruli. (F) Confocal images corresponding to the overview in E. The photoreceptor axons immunolabeled with monoclonal anti-DLG antibody 4F3 (magenta) enter the brain and project to the antennal lobe immunolabeled with anti-Bruchpilot antibody nc82 (green). (G–I) Higher magnification of boxed area in F, with H and I representing single confocal channels. (J–0) Schematic overview (J) of the ventral ganglion shown in K–N. T1-T3, thoracic neuromeres 1–3; A, abdominal neuromere. (K) Confocal image corresponding to boxed area in J. Thoracic ganglion immunolabeled with anti-HRP antibody (magenta) and nc82 (green). (L) Higher magnification of the boxed area in K. (M and N) Single confocal channels of L. (O) Boutons of ectopic photoreceptors in the thoracic ganglion.

Table 1.

Summary of ERG recordings of normal eyes, and ectopic eyes on leg, wing, and antennae in w/w;dpp^blinkGAL4/UE10 flies

Fly part	Total measured, n	Normal polarity with transients, n	Reversed polarity, n	Photoreceptor potential, n	No ERG, n
Eye	25	25	–	–	–
Antenna	17	7	4	5	1
Leg	20	–	4	12	4
Wing	10	–	–	2	8

Ectopic Photoreceptor Axons Project to the CNS.

We next analyzed the projections of ectopic photoreceptor axons in third-instar larvae, four pupal stages, and pharate adults of the genotype w/w;dpp^blinkGAL4/UE10 by means of immunocytochemistry with the photoreceptor-specific antibody 24B10 or labeling with anti-HRP to detect HRP driven in the photoreceptors by a GMR element (30). Ectopic photoreceptors were seen in eye-antennal, leg, wing, and haltere imaginal discs of third-instar larvae. The photoreceptor axons displayed elaborate projections within the imaginal discs, but in no case did they project to the larval CNS (results not shown). We observed ectopic photoreceptor axons terminating in the brain and the ventral ganglia starting in pupae of 48 h (50% pupal development) and in all later pupal stages and pharate adults (see Table 2 for summary). In pharate adults, we found no evidence for antennal ectopic photoreceptor axons projecting toward the normal eye and ultimately to the lamina. Rather, antennal ectopic eyes projected their axons along the antennal nerve, eventually projecting to the glomeruli of the antennal lobe (Fig. 1 E–I). Ectopic photoreceptors originating in the legs projected to the thoracic ganglion, where they terminated superficially without penetrating deeply into the neuropil (Fig. 1 J–O). Sometimes photoreceptors of as many as five (of six) leg ectopic eyes sent axonal projections to the thoracic ganglion. Ectopic photoreceptors in the wing were seen projecting in the wing blade (results not shown).

Table 2.

Projection of ectopic photoreceptor axons into the brain or ventral ganglion of pupae and 1-day old adults

Genotype	Tissue	Developmental stage
		25% pupae	50% pupae	75% pupae	100% pupae	1-Day adult
UE10, dppGAL4	Brain	0/8	10/21	7/11	2/3	10/25
	Ventral ganglion	0/7	8/13	6/6	1/1	5/5
UE11, dppGAL4	Brain	0/6	2/3	1/2	ND	ND
	Ventral ganglion	0/5	5/5	3/3	ND	ND

The numbers (x/y) indicate the number of cases (x) in which axons enter the brain or ventral ganglion and the total number of brain or ganglia analyzed (y). ND, not done.

A similar analysis of early pupal stages of the w/w;UE11/+;dpp^blinkGAL4/+ genotype gave comparable results (see Table 2 for summary).

Ectopic Photoreceptor Axons Form Synapses Within the CNS.

In the ventral ganglion, we frequently observed the presence of structures resembling synaptic boutons along the ectopic photoreceptor axons, suggesting that the axons established ectopic synapses (Fig. 1 K–O). To verify the presence of synapses and investigate whether the synapses resembled those formed in the lamina and medulla of the optic lobe by the photoreceptors of the normal compound eye, we needed to identify the ectopic photoreceptors in ultrathin sections via electron microscopy (EM). To accomplish this, we generated the pGMR-CD2-HRP expression vector to target photoreceptor-specific expression by using the glass multimer reporter (GMR; ref. 30) of membrane-associated CD2-HRP (31). HRP was chosen as a marker, because it produces an electron-dense deposit after reacting with 3,3′ diaminobenzidine (DAB). Transgenic flies of the genotype GMR-CD2-HRP/+;dpp^blinkGAL4/UE10 exhibited strong labeling of both the endogenous and ectopic photoreceptors as revealed when reacted with DAB, and these anti-HRP labelings overlapped perfectly with anti-24B10 labeled axons (data not shown), indicating that the photoreceptors alone expressed HRP.

To investigate the ultrastructure of the CD2-HRP-labeled ectopic photoreceptors, we examined leg and antennal ectopic eyes. HRP reaction product confirmed that EM profiles were from photoreceptors of the ectopic eye. For both leg and antennal eyes, the labeled profiles of photoreceptor axons invariably ran along the surface of the corresponding ganglion. We were able to locate synaptic features of ectopic photoreceptor axons in antennal eyes, but not in leg-thorax specimens containing leg ectopic eyes.

Axon bundles from ectopic eyes were confined and not extensive. Those from the antenna fasciculated in the antennal nerve and entered the brain's deuterocerebrum, where their labeled profiles often faded out, making it difficult to trace axons further to see whether they formed synapses. Once within the neuropil, the axons grew preferentially between the cortex and neuropil. Lack of labeling at the nerve tip may have resulted because HRP membrane turnover at the axon tip diluted the label. However, incomplete penetration of DAB reagents in regions still enclosed by cuticle seems likely to have limited the visibility of labeled fibers in all preparations. For example, in projections from antennal eyes, labeled axon profiles were only visible once the nerve left the antenna (which was not opened up). Because further dissection to reveal enclosed regions impaired the quality of EM preservation, we chose a compromise with limited dissection.

Synaptic Profiles.

As visible by light microscopy, many axons swelled or even terminated en route, forming large profiles in the nerve (Fig. 2 A and B). Synapses were not seen in such profiles, however. After analyzing several preparations, we succeeded in finding synaptic contacts in the axon profiles of an antennal eye once these had entered the superficial strata of the antennal lobe neuropil (Fig. 2 A and C–E). The profiles had HRP reaction product on the presynaptic membrane and were filled with synaptic vesicles of normal size, ≈25 nm in diameter (21). Synaptic sites were visible from the presence of a presynaptic T-bar ribbon (24), comprising a platform surmounting a narrow pedestal, very similar in form to those in photoreceptor axon terminals in either the lamina (21) or medulla (32). The ectopic T-bar ribbons were smaller than normal, however, with a platform ≈138 nm wide (compare 210 nm for a normal photoreceptor synapse in the lamina). Although the synaptic profiles were unmistakable, the clarity of EM images was not optimal because of technical limitations inherent to the preparation.

Fig. 2.

Ectopic photoreceptor terminals form synaptic contacts in foreign territories. (A) A stained photoreceptor axon bundle from an antennal eye off the left hand side of the section can be seen extending through the scape (scp), along the antennal nerve (AN; right rectangle) and into the antennal lobe (AL; left rectangle) in a light micrograph of an unstained 2-μm semithin section before re-embedding. (B–D) Electron micrographs from the same section as in A. (B) Bundle of axon profiles in the antennal nerve (right rectangle in A) with HRP-labeled plasma membranes. (C) Profiles of axons with labeled membranes in the superficial layers of the antennal lobe, beneath the cortex, enlarged from left rectangle in A. Three labeled axon profiles enclosed in the rectangle are further enlarged in D. (D) Three profiles of synaptic terminals, containing mitochondria (m), have a presynaptic profile (rectangle, enlarged in E). (E) Presynaptic T-bar ribbon (arrowhead) in terminal, opposite profiles of two of the postsynaptic elements (asterisks). The presynaptic element has many profiles of synaptic vesicles (arrow), one of which appears to be labeled with HRP reaction product possibly endocytosed from the plasma membrane (short arrow), but lacks the characteristic glial invaginations at capitate projections.

The postsynaptic elements at synaptic sites were not identified, but not surprisingly did not resemble the profiles of normal L1/L2 targets in the lamina. The latter typically have a characteristic endoplasmic cisterna beneath the postsynaptic membrane (21). Nevertheless, postsynaptic elements formed dyads in single sections that cut the midregion of the ribbon. There was at least a third profile contacting the release site in adjacent sections. Thus, many such ectopic synapses upon foreign target neurons appeared to be triads. This arrangement compares with the normal photoreceptor synapses, which are tetrads for R1–R6 terminals in the lamina (21) and mostly triads for R7 and R8 terminals in the medulla (32). The targets, whether neurons or glia, could not be identified further, but we infer that their postsynaptic geometry is attributable to the presynaptic site, the photoreceptor.

Ectopic photoreceptor terminals contained no capitate projections, i.e., normal glial invaginations into photoreceptor terminals in both the lamina (33) and medulla (T. N. Edwards and I. A. Meinertzhagen, unpublished work). Their absence signals the failure to find the correct glia in the antennal lobe.

Discussion

We have demonstrated that ey-induced ectopic photoreceptors on legs and antennae can successfully extend axonal projections to the CNS. Furthermore, morphological evidence and ERGs demonstrate that the photoreceptor axons of at least antennal ectopic eyes form completely constituted ectopic synapses with foreign postsynaptic elements and suggest that transmission at these sites may be functional.

The on and off transients are generated by cells in the lamina (28) reported to be the large monopolar cells (26). We do not know the precise identity of the cells postsynaptic at ectopic photoreceptor axons, but these cannot be lamina neurons, and our observations therefore suggest there could be a wide range of target cells that can serve as postsynaptic partners at photoreceptor synapses. Further, the fact that we can measure ERGs with on and off transients suggests that the extracellular resistance pathways at the ectopic antennal eye may be similar to those at the normal eye (28). The lack of the correct geometry or electrical resistances may also explain why we did not observe on and off transients for ectopic eyes on the legs, despite the fact that many boutons appear to form from their axons.

Sustained negative photoreceptor potentials were observed in 72% of all ectopic eyes analyzed. This number includes eyes with photoreceptor potential only, and those with on and off transients. The high incidence with which we observe photoreceptor potentials can be explained by the fact that in most cases the ectopic compound eye is morphologically normal with photoreceptors displaying normal morphological and physiological development. It has previously been established in the housefly Musca domestica (34) and the fleshfly Sarcophaga bullata (35) that photoreceptor potentials are generated even in the absence of synaptic contacts. One previous report describes photoreceptor axon projections of ectopic eyes in the Drosophila extra eye mutant. Almost all photoreceptor axons terminate in a tissue mass lying beneath the ectopic ommatidia that give rise to them, but in two cases rare projections to the ipsilateral lamina were observed when the ectopic eyes had fused with the normal eye and thereby presumably gained access to the correct guidance pathway (36). The photoreceptor axon projections we observed were all from ectopic eyes totally isolated from the normal eyes of the flies. Our observation that ectopic boutons formed in the thoracic ganglion and the antennal lobe and that we could find clear synaptic profiles in the latter suggests that synapse formation could be preprogrammed by the photoreceptor neuron in a manner independent of both the host environment or the identity of the postsynaptic cells. Our observation that photoreceptors form synapses independent of the particular postsynaptic partners supports findings in which such independence was observed in photoreceptors that are induced to mistarget by genetic means (37). Likewise, photoreceptor axons beneath transplanted ectopic eyes form presynaptic sites upon glial cells (38).

A related issue concerning ectopic photoreceptor axon projections is the mechanism by which photoreceptor axons in legs and antennae navigate in an abnormal environment to form synapses in the CNS. Our observation that retinal axonal outgrowth occurs in all imaginal discs suggests that the extracellular environment of all discs is permissive for photoreceptor axon outgrowth. The first ectopic photoreceptor axons enter the brain 48 h after puparium formation, at a time when most normal sensory neurons of the host discs have fully established their axonal projections. This timing suggests that ectopic photoreceptor axons may use these existing normal pathways to project to the CNS. Pathway formation from receptor neurons in the leg, wing, and potentially all imaginal discs is independent of any central influence and highly reproducible. In leg discs, several neurons are present before metamorphosis and serve as pioneers for later neurons (39). In wing and haltere discs, pioneer neurons appear after the onset of metamorphosis (39, 40), whereas antennal disc neurons, by contrast, do not use pioneer pathways (39). In leg and eye-antennal, but not wing discs, some of the larval precursor nerves connecting the periphery to the CNS persist and serve as pioneer neurons (39, 41, 42). The neuronal pathways depend on at least two sources of guidance: the intrinsic property to fasciculate and use pioneer neurons and the responsiveness to polarity and positional guidance cues (43–47). All imaginal discs appear to share a common mechanism for axonal outgrowth and establishment of central connections (39), with the general property that axons from a few early neurons establish the initial nerve pathway. However, although these early pioneer neurons probably facilitate axonal pathfinding, there is ample evidence that normal nerve routes can also establish in the absence of pioneers (48). The possibility of a general lack of specificity in pathway selection among sensory axons in imaginal discs is also supported by transplantation experiments (49) demonstrating that the L3 pathway in the developing wing can be used by neurons from transplants of the eye, antennal, or leg discs. Kunes et al. (50) also observed that photoreceptor axons can make certain pathfinding decisions independent of pre-existing retinal axons. In summary, our data are consistent with the axons of sensory neurons being relatively nonspecific in pathway selection and show that all imaginal discs are permissive for photoreceptor outgrowth. Furthermore, synapse formation at least for the presynaptic organelles seems to be a property intrinsic to the terminal of a photoreceptor neuron.

In contrast to their ability to find a path to enter the CNS, the projections of photoreceptor axons in the thoracic ganglion or antennal lobe are all superficial. In no case did we observe a projection pattern with similarities to the highly ordered and stereotypical pattern found in the optic lobes. Several explanations can be advanced for this lack. (i) Photoreceptor innervation may be too late. Normal photoreceptors start entering the optic lobes in the third-larval instar. In contrast, we observed the entry of ectopic photoreceptors mainly around midpupal development. (ii) The establishment of the highly regular photoreceptor connections in the optic lobe involves a cascade of inductive events (reviewed in refs. 1 and 3) and responsiveness of the target tissue to inductive cues such as Hh for photoreceptors (16). Our results suggest that the ectopic target is not responsive at the time when it receives photoreceptor axon innervation. (iii) The establishment of synaptic contacts requires several cell surface molecules, including N-cadherin, flamingo, PTP69D, and LAR. It is therefore conceivable that the absence of certain cell adhesion molecules in the ectopic target may preclude the formation of synaptic contacts as seen in the normal optic lobe.

In conclusion, we have shown that axons of ectopic photoreceptors navigate imaginal disc environments to enter regions of the CNS that reflect their ectopic location, ectopic photoreceptors in antennae projecting to antennal lobes, etc., without regard to the normal target territories in the optic lobe. Ectopic photoreceptor terminals then establish synapses with presynaptic sites similar to those at normal photoreceptor synapses and with a postsynaptic organization that incorporates multiple elements (24), albeit with the absence of specializations such as glial invaginations (33). Our observations are consistent with the autonomy of the photoreceptor in pathfinding and synaptogenesis. Furthermore, the absence in the ectopic target of the highly stereotypical architecture normally seen in the optic lobes indicates the lack or mismatch in the host territory of the antennal lobe or thoracic ganglion of the molecular signaling events required to establish that architecture.

Materials and Methods

Drosophila Genetics and Induction of Ectopic Eyes.

Ectopic eyes were induced as described (10). The following GAL4 driver lines were used: dpp^blinkGAL4 (w/w;Sp/CyO;40C.6/TM6B: ref. 51), MS1096 (29), and E132 (10). The UAS-ey stocks used in these experiments were UE10 (homozygous viable on 3) and UE11 (homozygous viable on 2; ref 10).

ERGs.

ERGs were recorded with a glass microelectrode filled with 3 M KCl solution inserted in conductive gel topically applied on either the normal eye that served as a control or ectopic eyes. The ground electrode was inserted in a similar manner on the abdomen. The electrodes were connected to a WPI-707 amplifier, and the signal was visualized on a Tectronix storage oscilloscope. A photocell was also connected to the amplifier and used to monitor the exposure to light. The flies were dark-adapted for at least 1 min before ERG recordings. The eyes were illuminated with a focused light source from a fiber optic light (Volpi 5000). The light beam was interrupted by the shutter of a camera or manually. The oscilloscope record was photographed with a handheld Polaroid camera.

Histology.

Brains and ventral ganglia from pupae and adults were dissected in 1× PBS and kept on ice for up to 1 h until fixation. The tissue was fixed in a 4% formaldehyde solution in PBS for 15 min. After washing, PBS was removed and the tissue incubated with PAXD (PBS containing 5% BSA, 0.3% Triton X-100, and 0.3% sodium deoxycholate; ref 52) for 1 h, followed by overnight incubation at 4°C with primary antibody in PAXD. After washes in PAXD at room temperature for 4–6 h with several changes, the tissue was incubated overnight at 4°C with HRP-, FITC-, or Cy3-conjugated secondary antibodies. The next day, the tissue was washed for 4–6 h with PAXD. HRP was detected with DAB solution with 0.3% H₂O₂, and the tissue was mounted in glycerol/PBS (70:30). FITC- or Cy3-labeled tissue was mounted in Vectashield (Vector Laboratories). Preparations were analyzed with Olympus AX70 and BX61 microscopes. Confocal microscopy used a Leica TCS SP2 microscope.

Antibodies.

The 24B10 antibody (photoreceptor-specific anti-Chaoptin) (53) was used at a dilution of 1:30 in PAXD. The 4F3 (anti-DLG) antibody was used at a dilution of 1:20. The nc82 antibody (anti-Bruchpilot) was used at a dilution of 1:10 and labels all neuropils (54). Goat-anti-peroxidase was used at a dilution of 1:2,000. Secondary antibodies were used at dilutions of 1:200.

24B10, 4F3 and nc82 antibodies were obtained from the Developmental Studies Hybridoma Bank, University of Iowa, Iowa City. Goat-anti-peroxidase was from Sigma. HRP-, FITC-, or Cy3-conjugated secondary antibodies were from Jackson Immunoresearch.

Generation of pGMR-CD2-HRP.

The DNA fragment coding for the CD2-HRP fusion protein (31) was cloned into the EcoRI site of pGMR (30) to form pGMR-CD2-HRP. The construct was verified by sequencing and injected into w¹¹¹⁸ embryos as a service (Duke University Model Systems Genomics) to obtain transgenic fly stocks. Several independent transgenic lines were generated. For two of these, stocks were generated that also have the UE10 chromosome to overexpress Eyeless. The resultant genotypes were: GMR-HRP-23B/GMR-HRP-23B;UE10/UE10, and GMR-HRP-33A/GMR-HRP-33A;UE10/UE10.

EM.

To examine axon profiles of ectopic eyes at the EM level, we crossed virgins of 23B and 33A GMR-HRP lines with male dpp-Gal4. Because more F₁ arose from the 33A cross, these were used for the actual selection of specimens with ectopic eyes. To reveal HRP activity, we used the DAB method (55) and assisted post-pharate flies to emerge from their puparium, choosing those that had ectopic eyes on either the leg or antenna.

We dissected the preparation in fixative [4% paraformaldehyde and 0.5% glutaraldehyde in 0.1 M phosphate buffer (PB), pH 7.3 at 4°C]. As much of the cuticle as possible was removed with forceps and Vannas scissors, and the brains were then fixed for 1 h on ice. Brains were washed in 0.1 M PB 2× for 30 min (or overnight), transferred to a sieve bucket, washed again, and then treated with freshly made 1% sodium borohydride in 0.01 M PBS (pH 7.3) at 4°C for 20 min. They were then washed again, four times for 15 min each in 0.01 M PBS (0.01 M PB, pH 7.4 in 137 mM NaCl and 2.68 mM KCl) with 0.07% Triton-X (PBS-T), and then three times for 10 min each in 0.01 M Tris·Cl at pH 7.6 in 137 mM NaCl and 2.68 mM KCl with 0.07% Triton-X (TBS-T). Brain samples were preincubated protected from direct light, for 30 min at 22°C, in filtered DAB solution of 20 ml of 0.01 M TBS-T (with 0.04% Triton-X) containing a 10-mg tablet of DAB powder, to which 8% NiCl₂, was added to a final concentration of 0.064%, and then incubated in DAB solution containing 0.03% H₂O₂. The reaction took ≈40 min at 22°C until fine axons from the retina to the lamina and from the lamina to the medulla were visible beneath the normal compound eye, which also expressed the GMR-controlled CD2-HRP transgene. These served as markers to control the final incubation for the ectopic eye. Preparations were then washed in 0.01 M TBS three times each for 10 min and then postfixed for EM in 1% OsO₄ for 30 min at 4°C. Finally, preparations were washed in distilled water or 30% ethanol for 30 min (or overnight), dehydrated, and infiltrated with PolyBed 810 for ultramicrotomy.

Semithin 3-μm sections were cut and their images were captured from dry sections with either a ×40/0.75 or ×20/0.50 Plan Neofluar objective. Appropriate sections were then re-embedded, ultrathin sections were cut at 70 nm, and unstained sections were viewed in a Philips Tecnai 12 electron microscope operated at 60 kV.