Dpp and Hedgehog mediate neuron–glia interactions in Drosophila eyedevelopment by promoting the proliferation and motility of subretinal glia

Radha Rangarajan1, Helene Courvoisier1, Ulrike Gaul*

Laboratory of Developmental Neurogenetics, Rockefeller University, 1230 York Avenue, New York, NY 10021, USA

Received 12 July 2001; accepted 18 July 2001

Abstract

Neuron–glia interactions are crucial for the establishment of normal connectivity in the nervous system during development, but the

molecular signals involved in these interactions are largely unknown. Here we show that differentiating photoreceptors in the developing

Drosophila eye influence the proliferative and migratory behavior of the subretinal glia through the diffusible factors Decapentaplegic (Dpp)

and Hedgehog (Hh). We demonstrate that proliferation and migration of the glia are separable processes, and that Dpp promotes both the

proliferation and motility of the glia, whereas Hh appears to promote only their motility; neither specifies the direction of migration. We

present evidence that Dpp and Hh act on the glia in parallel and through the regulation of transcription. Finally, we show that ectopic

migration of subretinal glia can result in the ectopic projection of photoreceptor axons. Our study suggests a novel function for Hh in

regulating migratory behavior and provides further evidence for a complex mutual dependence between glial and neuronal cells during

development. q 2001 Elsevier Science Ireland Ltd. All rights reserved.

Keywords: Dpp; Hedgehog; Subretinal glia; Glial migration; Visual system development; Drosophila

1. Introduction

Glial cells are major constituents of the nervous system

and play an important role during its development in both

vertebrates and invertebrates (reviewed by Auld, 1999;

Birling and Price, 1995; Silver, 1993; Tear, 1999; Tessier-

Lavigne and Goodman, 1996; Thomas, 1998). They have

been shown to help control the patterning of neuronal differ-

entiation, to provide guidance for migrating neurons and

growing axons, and to prompt fasciculation and defascicu-

lation of axons. Later on, glia are essential for wrapping and

insulating neurons, providing them with trophic and survi-

val factors, and effecting their homeostasis.

In order for glial cells to fulfill their role, they must be

matched in number and positioned correctly with respect to

the neurons they support. For many glial populations, this

means proliferation of a limited pool of precursor cells and

migration over many cell diameters within a specific time

period during development. In vertebrates, this is true for

the oligodendrocytes of the CNS, which arise from a prolif-

erating pool of motile precursor cells, as well as for the

peripheral glia, which originate from the neural crest

(Barres and Raff, 1999). In the Drosophila CNS, most

glial cells arise from a relatively small number of neuro-

glioblasts, then migrate and spread to reach their final desti-

nation (Hartenstein et al., 1998; Ito et al., 1995); most PNS

glia are similarly derived from proliferative pools of motile

precursor cells (Choi and Benzer, 1994; Giangrande et al.,

1993; Perez and Steller, 1996). The regulation of glial cell

proliferation and survival is well studied, in particular in the

vertebrate system (for review see Barres and Raff, 1999),

however, relatively little is known about the molecular

mechanisms involved in glial migration.

The present study investigates the proliferation and

migration of the subretinal glia in the developing Droso-

phila visual system. Subretinal glial precursors originate

in the optic stalk and migrate into the eye imaginal disc

concomitant with the onset of differentiation of photorecep-

tors (Choi and Benzer, 1994). Their presence in the eye disc,

in turn, is necessary for guiding photoreceptor axons from

the eye disc into the optic stalk (Rangarajan et al., 1999).

Here we examine the effects of the diffusible factors Deca-

pentaplegic (Dpp) and Hedgehog (Hh), which have an

important role in photoreceptor differentiation, on the devel-

opment of the glial precursor cells.

Dpp belongs to the TGF-b superfamily of ligands, which

include the bone morphogenetic proteins (BMPs), the

Mechanisms of Development 108 (2001) 93–103

0925-4773/01/$ - see front matter q 2001 Elsevier Science Ireland Ltd. All rights reserved.

PII: S0925-4773(01)00501-9

www.elsevier.com/locate/modo

* Corresponding author. Tel.: 11-212-327 7621; fax: 11-212-327 7923.

E-mail address: gaul@rockvax.rockefeller.edu (U. Gaul).1 These authors contributed equally to this work.

growth and differentiation factors (GDFs), and various

forms of transforming growth factor-b, Activin and Nodal

(for review see Massague, 1998; Massague and Chen,

2000). Produced by diverse cell types, these factors have

been shown to regulate a wide range of processes, including

cell migration, adhesion, proliferation, fate specification,

and survival, although effects on cell migration and adhe-

sion have mostly been studied in culture. In Drosophila,

Dpp regulates the growth and patterning of the imaginal

discs (Burke and Basler, 1996; Capdevila and Guerrero,

1994; Heberlein et al., 1993). In the eye disc, it is one of

the key players in the initiation of the morphogenetic

furrow, which marks the propagating front of photoreceptor

differentiation (Greenwood and Struhl, 1999; Heberlein et

al., 1993).

Dpp transmits its signal by binding to Type I serine threo-

nine kinase receptors, Saxophone and Thick veins (Tkv),

and the Type II receptor Punt. Upon ligand binding, the

Type I receptors become phosphorylated by the Type II

receptor, resulting in the activation of the receptor I kinase.

This kinase then phosphorylates the Receptor-Smad Mad,

which in turn binds the Co-Smad, Medea. The Smad

complex translocates to the nucleus, where it activates target

gene expression (Massague and Chen, 2000).

The secreted protein Hedgehog and its vertebrate homo-

logs have been shown to regulate a wide range of develop-

mental processes, involving growth control, patterning, and

organ morphogenesis (for review see Ingham, 1998). In the

developing Drosophila eye, Hh controls the initiation and

propagation of the morphogenetic furrow, leading to photo-

receptor differentiation (for review see Treisman and Heber-

lein, 1998). In addition, Hh is transported along the

photoreceptor axons into the optic lobe, where it triggers

the proliferation and differentiation of neuronal precursor

cells that form the lamina (Huang and Kunes, 1996,

1998). In the vertebrate nervous system, Sonic Hedgehog

(Shh) acts as a morphogen in the specification of neuronal

cell fates in the ventral spinal cord (Briscoe and Ericson,

1999) and as a mitogen for a variety of cell types, including

astrocytes in the rodent optic nerve (Wallace and Raff,

1999). Effects of Hh on the migratory behavior of cells

have not been reported.

Hh transduces its signal by binding to its receptor Patched

(Ptc), which acts as a negative regulator of the seven-trans-

membrane protein Smoothened (Smo); binding of Ptc by Hh

removes this inhibitory effect and allows Smo to activate the

transcriptional effector Cubitus interruptus (Ci). In the

absence of Hh, Ci is proteolytically cleaved and acts as a

transcriptional repressor of Hh target genes. Smo activity

prevents the proteolysis of Ci, resulting in the accumulation

of full-length Ci, which acts as a transcriptional activator of

Hh target genes (for review see McMahon, 2000).

In this study, we show that Dpp and Hh, in addition to

controlling the propagation of photoreceptor differentiation

in the eye disc, also have a role in regulating glial migration

and proliferation. Both Dpp and Hh promote the overall

motility of the subretinal glia, while only Dpp stimulates

their proliferation. We provide evidence that both factors

exert their effects through transcriptional regulation. Our

findings identify Dpp and Hh as important factors in coor-

dinating neuronal and glial cell development during the

establishment of the Drosophila visual system.

2. Results

In the developing Drosophila visual system, the eye disc

is connected to the optic lobe by the optic stalk. Photore-

ceptor cells are generated in the eye disc in a posterior-to-

anterior progression; the wave of differentiation is marked at

its front by an indentation of the disc epithelium called the

morphogenetic furrow. The photoreceptors extend axons

into the basal layer of the eye disc, where they turn and

proceed posteriorly to exit the disc through the optic stalk

(Fig. 1B,C). The subretinal glia (also known as retinal basal

glia) originate from precursor cells in the optic stalk; the

glial precursors in the stalk already express the glial-specific

marker Repo (Halter et al., 1995). The glia begin to migrate

into the eye disc with the onset of the differentiation of

photoreceptors (Choi and Benzer, 1994), but do not require

the presence of axons in the stalk to find their way into the

eye disc (Rangarajan et al., 1999). Once in the eye disc, the

glia migrate anteriorly on the basal surface of the eye epithe-

lium (Fig. 1A); the anterior border of their migration lies

typically 2–4 rows of ommatidia posterior to the morpho-

genetic furrow, which is where axons sent out by the differ-

entiating photoreceptors begin to turn posteriorly (Fig.

1B,C). Thus, glia are present only in the axonal portion of

the eye disc. The glial cells associate closely with the photo-

receptor axons and develop extensive processes that

surround the axons and fill the intervening space (Choi

and Benzer, 1994). The development of the subretinal glia

thus involves proliferation, migration, and terminal differ-

entiation events, and it is closely tied, both spatially and

temporally, to the development of the photoreceptors.

2.1. Pattern of glial proliferation and terminal

differentiation

In order to get a better understanding of the sequence of

events in the development of the subretinal glia, we sought

to determine the pattern of glial proliferation and cell shape

change. We marked proliferating cells using antibodies

against phosphorylated histone H3 (Hendzel et al., 1997).

These antibodies recognize cells between late G2 and meta-

phase of the cell cycle and are reliable markers for mitosis.

We observe dividing glial cells in the optic stalks of early

and late third instar larval stages. Interestingly, we also find

dividing glial cells in the eye discs of these animals (Fig.

1D). The number of labeled cells is generally very small,

between one and ten, which is probably due to the relatively

small window of expression of phosphorylated histone H3

and a lack of synchrony in glial cell divisions. The cell

R. Rangarajan et al. / Mechanisms of Development 108 (2001) 93–10394

divisions show no consistent spatial distribution pattern.

These data show that the glial cell population continues to

proliferate throughout the third instar larval stage, both in

the optic stalk and after entering the eye disc. During this

expansion period, there is no significant amount of cell

death in the glial cell population, as determined by using

acridine orange and human anti-caspase antibodies as

markers (data not shown).

To relate the morphology of the subretinal glia to their

position within the eye disc, we generated small clones of

glial cells expressing GFP (see Section 4). We find that in

the anterior, all cells are small and rounded, whereas in the

posterior a range of cell morphologies coexist. In addition to

small cells with simple morphology, which may represent

migrating cells, we find large cells that are elongated to

varying degrees and show a complex morphology (Fig.

1E–J). This indicates that terminal differentiation of the

subretinal glia (e.g. wrapping of axons) does not wait until

the eye disc is completely developed and filled with glia, but

rather takes place concomitantly with the progression of the

morphogenetic furrow and while other glia are still migrat-

ing.

How are the proliferation and migration of the subretinal

glia regulated; and what is the molecular basis for the coor-

dination between photoreceptor and glial cell development?

In an effort to address these questions, we examined the role

of the diffusible factors Dpp and Hh.

2.2. Role of Dpp in subretinal glial development

2.2.1. Loss of Dpp signaling impairs the accumulation of

glia in the eye disc

Dpp is an important factor in the initiation and propaga-

tion of the morphogenetic furrow. It is expressed in the

posterior and lateral margins of the eye disc in early third

instar animals (Fig. 2A) and in the morphogenetic furrow in

later stages (Fig. 2B) (Blackman et al., 1991; Heberlein et

al., 1993). It is not significantly expressed anywhere else in

the eye disc or in the glial cells, as confirmed by enhancer

trap expression analysis and RNA in situ hybridization (Fig.

2C).

Since removal of Dpp affects overall eye development,

we sought to use the FLP-FRT system (Duffy et al., 1998;

Xu and Rubin, 1993) to generate mosaic animals in which

mad, an essential component of the Dpp signaling pathway

(Wiersdorff et al., 1996), is removed from the glia, but not

from cells of the overlying photoreceptor epithelium. Since

the glial population is small compared to that of neuronal

cells, it is impossible to use hsFLP to generate glial clones

without also greatly affecting the other relevant tissues. We

therefore decided to specifically target the glial cells for

somatic recombination using ombGAL4 driving a UASFLP

transgene; ombGAL4 is expressed in the glia and in the

margins of the eye disc throughout larval development,

but not in the photoreceptors (see Section 4) (Brand and

Perrimon, 1993; Duffy et al., 1998; Lecuit et al., 1996;

Rangarajan et al., 1999). A UAS nuclear-GFP transgene

serves as a negative marker for the mutant glial population.

To determine the efficiency of recombination, we generated

clones with an FRT chromosome bearing no mutations and

counted the percentage of cells lacking GFP. The average

percentage of GFP-negative glial cells, i.e. cells having lost

the marker through recombination, in such eye discs is

35.2% (^2.7; n ¼ 29) (Fig. 2F), indicating that the effi-

ciency of recombination between the FRT chromosomes is

lower than the theoretical limit. Since glial cells are still able

to divide after arriving in the eye disc (see above), this

R. Rangarajan et al. / Mechanisms of Development 108 (2001) 93–103 95

Fig. 1. Organization and morphology of subretinal glia in third instar larval

eye discs. (A) Wild-type eye disc, showing the distribution of subretinal

glia nuclei, labeled with anti-Repo antibodies (red), in the basal layer of the

eye disc and in the optic stalk. (B) Same disc as in (A), with photoreceptor

cell bodies and axons labeled with anti-HRP antibodies (green) and glial

nuclei labeled with anti-Repo antibodies (red). (C) Schematic drawing

showing the progression of development along the anterior–posterior axis

and the relative positions of the photoreceptor cell bodies and axons (green)

and glia (red) along the apical-basal axis. (D) Eye disc labeled with anti-

pH3-antibodies (green) to reveal mitotic cells and with propidium iodide to

reveal all nuclei (red). Two glial cells in the basal layer of the eye disc are

undergoing mitosis (yellow, arrows); the other two mitotic cells are part of

the eye disc margin. (E,G–J) show the morphologies of individual glial cells

expressing cytoplasmic GFP (green) at different positions along the ante-

rior-posterior axis of the eye disc. The glial nuclei are labeled with anti-

Repo antibodies (red). In the anteriormost row of glia, the cells are small,

round, and have a simple morphology (G); more posteriorly, glial cells are

larger and show a more complex and elongated morphology (E,H,J).

However, cells of simple morphology (I) are also observed in the posterior.

(F) Schematic drawing showing the relative positions of the cells depicted

in (G–J) within the eye disc. In all figures except (C) posterior is down,

dorsal to the right; in (C), posterior is to the left. Scale bars in (B) for

(A,B,D): 30 mm; in (E) and (J) for (G–J): 20 mm.

suggests that at least a fraction of cells migrate into the eye

disc (and possibly reach the anterior boundary) as hetero-

zygotes before undergoing recombination to become homo-

zygous mutant.

In mad mosaic animals, the average percentage of GFP

negative glial cells in the eye disc is significantly lower than

for the wild type clones, at 17.2% (^1.8; n ¼ 29) (Fig.

2G,H). This indicates that the loss of mad function impairs

the accumulation of mutant glia in the eye disc. The total

number of glia in the eye disc, however, appears unchanged,

suggesting that wild-type glia are able to compensate for the

loss of mad mutant cells. The relative distribution of GFP

negative cells within the eye disc is also unaffected: mutant

cells are found at all positions within the differentiating

portion of the eye disc. When a dominant negative form

of the Dpp receptor Tkv is expressed in the entire glial

cell population, again using the UAS/GAL4 system and

ombGAL4 as a driver, the total number of glia in the eye

disc is reduced, but again without effect on their spatial

distribution (Fig. 2E).

These findings show that Dpp signaling is required for the

accumulation of glial cells in the eye disc, but, partly

because of the limitation in the mosaic technique, they do

not allow us to determine whether this results from an effect

on glial proliferation, survival, or on migration. The follow-

ing gain of function experiments address this question.

2.2.2. Dpp stimulates proliferation and motility of the

subretinal glia

We created different situations of overactive Dpp signal-

ing, by expressing Dpp in ectopic positions (either with the

GMRGAL4 driver or using Act . FLPout . GAL4) and by

expressing a constitutively active version of the Tkv recep-

tor in the glial cells (Das et al., 1998; Hay et al., 1994;

Nellen et al., 1996; Pignoni and Zipursky, 1997). In all

three situations, we find a pronounced increase in the

density of glial cells in the entire system, that is both in

the optic stalk and in the eye disc (Fig. 3B–E). Given that

there is no significant cell death in the subretinal glia at this

stage in wild type, this increase in cell number has to be the

R. Rangarajan et al. / Mechanisms of Development 108 (2001) 93–10396

Fig. 3. Increased Dpp signaling stimulates the accumulation of glia in the

eye disc. Third instar larval eye discs, glia are labeled with anti-Repo

antibodies (red). (A) Wild type. (B) y,hs FLP122/Act . FLPout . GAL4;

UAS dpp, containing random, unmarked clones of dpp expression. (C)

GMRGAL4; UAS dpp, with dpp expression in the differentiating portion

of the eye disc. (D) ombGAL4; UAS tkvact, with constitutive activation of

the Dpp pathway in all glial cells. (E) Histogram showing the average

number of glial cells per row of ommatidia (mean ^ SEM) in eye discs

of wild type, GMRGal4;UASdpp, ombGAL4;UAStkvact; GMRGal4;UAShh,

and ombGAL4;UASCiact, respectively. Sample sizes are shown in paren-

theses, *P , 0:01 as compared to wild type.

Fig. 2. Loss of Dpp signaling disrupts subretinal glial development. (A) In

early third instar larval eye discs, dpp-lacZ expression is visible in the

posterior and lateral margins. In late third instar larvae, dpp-lacZ (B) and

dpp RNA (C) are expressed in the morphogenetic furrow, with no expres-

sion in the glial layer of the eye disc. (D–G) late third instar larval eye discs,

glial nuclei labeled with anti-Repo antibodies (red). (D) wild type. (E)

ombGAL4; UAS tkvDGSK, showing a significant reduction in the number

of glial cells. (F) ombGAL4; FRT40 UAS GFP/FRT40; UAS FLP, having

undergone FRT-UAS FLP mediated mitotic recombination to generate

clones of glial cells that lack GFP but are otherwise wild type. Cells that

have not undergone mitotic recombination are GFP positive and appear

yellow, whereas those that have appear red. (G) ombGAL4; FRT40 UAS

GFP/FRT40mad12; UAS FLP. Glia lacking mad function are GFP-negative

(red). (H) Histogram showing the average percentage of GFP-minus cells

(mean ^ SEM) in eye discs when clones of wild type, mad12, smo3, and

mad12/smo3, respectively, are generated in the glia. Sample sizes are shown

in parentheses; (a) indicates P , 0:0001 as compared to wild type, (b)

indicates P , 0:01 as compared to the single mutant conditions. Scalebar

in (D) for (D–G): 30 mm.

result of increased proliferation, rather than of a suppression

of apoptosis.

Interestingly, Dpp overactivity affects not only the

number but also the positioning of the glia. In eye discs in

which dpp is expressed broadly in the entire differentiating

portion with GMRGAL4, the glial population migrates

beyond its normal axonal boundary along the entire front

(11/55). In the milder cases, the glia migrate only a few rows

beyond the normal anterior boundary of 2–4 rows posterior

to the morphogenetic furrow. In a few cases, however, they

migrate even beyond the morphogenetic furrow (Fig. 4B).

Given the mitogenic effects of Dpp on the glia, this

migratory phenotype could simply be a secondary conse-

quence of increased proliferation: a larger number of cells

competing for space might push the boundary of migration

forward. To test this possibility, we expressed String (Stg)

and CyclinE (CycE) proteins in the glia using ombGAL4 as a

driver. These proteins are G2-M and G1-S phase cell cycle

regulators, respectively; simultaneously increasing their

expression leads to faster cell cycles and overproliferation

(Neufeld et al., 1998). We find that glial cells overexpres-

sing Stg and CycE undergo excessive proliferation and

overcrowd the eye disc; in no instance, however, do they

cross the axonal boundary (n ¼ 15) (Fig. 4C,D). When we

overexpress CycE and Stg in an eyes absent (eya) mutant

background, in which no photoreceptor differentiation takes

place and the glia fail to enter the eye disc (Fig. 4K), we

similarly find severe overproliferation of glia in the optic

stalk, but still no migration into the eye disc (Fig. 4L). Both

experiments show that overproliferation by itself is not suffi-

cient to force ectopic migration of the subretinal glia. This

suggests that the forward shift of the boundary of glial

migration that we observe in eye discs expressing high

levels of Dpp is not an indirect consequence of an increase

in cell number, but rather the result of a separate and direct

effect of Dpp on the migratory capacity of the glia.

2.2.3. Dpp acts as a motogen but does not provide

directional cues

Dpp stimulates the motility of the subretinal glia, but does

it also provide directional cues for their migration? To

address this question we generated random clones of Dpp

expressing cells in the eye disc using the FLPout technique

(see Section 4) (Pignoni and Zipursky, 1997; Struhl and

Basler, 1993) and asked whether the glia migrate toward

such local sources of Dpp. A complication of this experi-

ment is that ectopic expression of Dpp in the anterior

portion of the eye disc can lead to the initiation of an ectopic

morphogenetic furrow with concomitant ectopic differentia-

tion of photoreceptors (Pignoni and Zipursky, 1997;

Rangarajan et al., 1999). In eye discs without such ectopic

photoreceptor differentiation (18/37), glial cells do over-

shoot anteriorly, either in a broad front or a narrow stream

(9/18). This mismigration is not preferentially directed

toward the Dpp clones, however, indicating that Dpp does

not specify the direction of migration (Fig. 4E–H). In eye

R. Rangarajan et al. / Mechanisms of Development 108 (2001) 93–103 97

Fig. 4. Increased Dpp signaling stimulates the motility of subretinal glia.

Third instar larval eye discs, photoreceptors are labeled with anti-HRP

antibodies (green) and glia with anti-Repo antibodies (red). (A) In wild

type, the anterior border of glial migration (arrowhead) lies 2–4 rows

behind the anteriormost row of photoreceptors (arrow). (B) GMRGAL4;

UAS dpp. The anterior boundary of the glial migration (arrowhead) has

moved ahead of the anteriormost row of photoreceptors (arrow). (C,D)

ombGAL4; UAS stg, UAS cycE. Despite the massive overproliferation of

glia in the optic stalk and the eye disc (C), the anterior boundary of glial

migration (arrowhead) remains within the normal range. (E–J) y,hsFLP122/

Act . FLPout . GAL4; UAS lacZ; UAS dpp eye discs, containing random

clones of Dpp expressing cells. (E,F and G,H) Two discs without ectopic

photoreceptor differentiation. Clones are visualized with anti-b-galactosi-

dase antibodies (blue, E,G). Glial cells overshoot anteriorly, either in a

stream (E,F) or a broader front (G,H), but they do not preferentially migrate

towards or into the clones. (I,J) An eye disc with an ectopic island of

differentiating photoreceptors in the anterior of the disc, with glial cells

migrating specifically towards the island (J). (K–M) Glial migration in an

eya mutant background. (K) eya. No photoreceptor differentiation takes

place, and the subretinal glia fail to enter the eye disc; in some cases,

glia migrate ectopically along the larval optic nerve. (L) ombGAL4; eya;

UAS stg, UAS cycE. Despite massive overproliferation, the glia fail to enter

the eye disc. (M) ombGAL4; eya; UAS Ciact. Increased Hh activity is

equally insufficient to induce migration of glia into the eye disc. Scale

bar in (M) for (K–M): 30 mm.

discs forming ectopic patches of differentiating photorecep-

tors, on the other hand (19/37), glial cells migrate specifi-

cally to such patches (17/19) (Fig. 4I,J). Differentiating

photoreceptors thus attract the subretinal glia, but Dpp by

itself cannot mimic this effect. As is the case for the effects

on proliferation, the effects of Dpp on the migratory beha-

vior of the subretinal glia can be phenocopied by activation

of the Dpp pathway in the glia themselves: in eye discs in

which constitutively active Tkv is expressed in the glia with

the ombGAL4 driver, we also find anterior overshooting of

glia, albeit less pronounced (10/53) (data not shown).

In summary, our analysis of Dpp overactivity shows that

Dpp signaling promotes both the proliferation and the moti-

lity of the subretinal glia, but without specifying the direc-

tion of migration. The motogenic effect is separable from

the effect on proliferation.

2.3. Role of Hh in subretinal glial development

Another candidate for mediating the interaction between

photoreceptors and glial cells in eye development is Hh. Hh

is expressed in the posterior margin of early third instar

larval eye discs prior to photoreceptor differentiation, and

subsequently in all differentiating photoreceptors; it is not

expressed in the glial cells (Heberlein et al., 1993; Ma et al.,

1993) (Fig. 5A). The Hh receptor Patched (Ptc), which is not

expressed in the photoreceptors, is found in all glial cells

within the eye disc, though not in the optic stalk (Fig. 5B).

Since ptc expression has been shown to be upregulated in

response to Hh signaling (Ingham, 1998), this finding indi-

cates that glial cells in the eye disc are exposed to the Hh

ligand.

2.3.1. Loss of Hh signaling

Following the procedure outlined above for the Dpp

effector Mad, we generated loss of function clones of the

Hh receptor Smo (Alcedo et al., 1996) in the glial cell

population. In such smo mosaic animals, there is no signifi-

cant reduction in the percentage of homozygous mutant

cells, and the relative spatial distribution of the glia in the

eye disc appears unaffected (Figs. 2H and 5D). Expression

of a dominant negative form of the Hh effector Ci (Methot

and Basler, 1999) in the glial population results in a mild

reduction in the total number of glia in the eye disc (Fig.

5C), but no effects on the spatial distribution. To uncover

potential redundancy with Dpp signaling, we generated loss

of function clones in which both smo and mad are removed

from the glia. In such animals, there is a small, though

statistically significant, reduction in the percentage of

mutant cells compared to mad single mutant clones (Figs.

5E and 2H). Taken together, these findings suggest at best a

mild requirement for Hh signaling in controlling the accu-

mulation of subretinal glial cells in the eye disc.

2.3.2. Hh acts as a strong motogen

As with Dpp, we created different situations of overactive

Hh signaling, by expressing the Hh ligand in ectopic posi-

tions (using Act . FLPout . GAL4) and by expressing a

constitutively active version of the Ci transcription factor

in the glial cells (Das et al., 1998; Hay et al., 1994; Nellen et

al., 1996; Pignoni and Zipursky, 1997). In both situations,

we observe normal numbers of glial cells in the eye disc and

the optic stalk, suggesting that Hh, in contrast to Dpp, has no

significant role in glial cell proliferation or survival (Figs.

6B–D and 3E).

We do, however, observe significant effects on the migra-

tory behavior of the subretinal glia. In eye discs containing

random clones of Hh expressing cells, we see robust anterior

overshooting of glial cells (14/40) past the axonal boundary,

usually in streams a few cells wide, and without regard to

the position of the clone (Fig. 6A,B). When an activated

form of the Hh effector Ci is expressed in the glial cells,

they similarly overshoot anteriorly (17/38), in narrow

streams that originate near the equator (Fig. 6C,D). These

results show that Hh acts as a strong motogen for the subret-

inal glia, without providing directional information. The

motogenic effect of Hh, however, is not sufficient to over-

come the lack of differentiation in the eye disc: if activated

Ci is expressed in the glia in an eya background, the glia fail

to enter the eye disc, as they do in eya mutants (Fig. 4M).

Interestingly, we observe that the overshooting glial cells

are frequently accompanied by photoreceptor axons that

follow them into the anterior portion of the eye disc, instead

of projecting posteriorly as in wild type (Fig. 6C,D). Since

R. Rangarajan et al. / Mechanisms of Development 108 (2001) 93–10398

Fig. 5. Role of Hh signaling in subretinal glial development. Third instar

larval eye discs. (A) Expression of hh-lacZ is found in all differentiating

photoreceptors. (B) Expression of ptc-lacZ (green) and Repo (red). Only

glial cells in the eye disc are ptc-positive (yellow). (C) ombGAL4; UAS Cidn

disc, showing a mild reduction in the number of glial cells compared to

wildtype (cf. Fig. 2D). (D) ombGAL4; FRT40,UAS GFP/FRT40smo3; UAS

FLP, labeled with anti-Repo antibodies (red). The number of GFP negative

( ¼ smo mutant) cells (red) in such eye discs is slightly lower than in discs

with wild-type clones (cf. Fig. 2F). (E) ombGAL4; FRT40, UAS GFP/

FRT40 mad12,smo3; UAS FLP, labeled with anti-Repo antibodies. The

number of GFP negative ( ¼ mad and smo double mutant) cells (red) is

lower than in discs with single mutant clones (cf. Figs. 5D and 2G).

this occurs when gene function is perturbed in the glia,

through expression of activated Ci, and since we find cases

of glial overshooting without concomitant axon misprojec-

tion, but never the reverse, we conclude that the ectopic

migration of glia is the primary effect and the cause of the

misrouting of photoreceptor axons. This phenomenon

demonstrates that glial cells can be sufficient to guide

axons and thus underscores the importance of the spatial

restrictions on subretinal glial migration that are found in

wild type.

3. Discussion

3.1. Dpp and Hh help coordinate glial with photoreceptor

development

In this study we have explored the molecular basis for the

regulation of glial cell development in the Drosophila eye.

We show that Dpp and Hh, factors that are known to orches-

trate the temporal progression of photoreceptor differentia-

tion within the eye disc, have a role in regulating the

proliferation and migration of the subretinal glia. Dpp and

Hh thus mediate between neurons and glia and help to

ensure that the development of the two cell populations is

closely coordinated.

The coordination of neuronal and glial development is

crucial for the establishment of functional circuitry in the

visual system. Normal photoreceptor axon projections

depend on the proper positioning of the subretinal glia:

Glia have to be present in the eye disc near the entrance

to the optic stalk for photoreceptor axons to find their way

into the stalk (Rangarajan et al., 1999); as we have shown

here, if glia are induced to mismigrate to ectopic positions in

the anterior, undifferentiated portion of the eye disc, axons

frequently follow. To prevent anterior misprojection of

axons it is therefore imperative that the migration of the

subretinal glia be restricted to the differentiating portion

of the eye disc. In addition, the number of glia in the eye

disc has to be roughly proportional to the number of omma-

tidia to ensure proper wrapping and insulation of the photo-

receptor axons. This means that the glial population has to

expand quickly in response to the growing number of

ommatidia progressively forming in the eye disc. This

matching of cell numbers requires the appropriate modula-

tion of glial proliferation and migration.

Neuron–glia interactions have also been shown to play a

role in oligodendrocyte and astrocyte development in the

vertebrate visual system. Retinal ganglion cell axons signal

to the glia in various ways to promote their proliferation and

survival (Barres and Raff, 1993; Barres et al., 1993; Burne

and Raff, 1997), thereby ensuring precise matching of the

two cell populations. However, there has been no evidence

to date for neuronal regulation of glial migration other than

through haptotaxis (Barres and Raff, 1999).

3.2. Subretinal glial proliferation

Our Dpp loss of function experiments show a substantial

reduction in the number or percentage of mutant glia in the

eye disc, indicating that a disruption of Dpp signaling impairs

the accumulation of subretinal glia in the eye disc. This

phenotype could be the result of an effect of Dpp on the

proliferation, the survival, or the migration of the glia. To

determine which aspect of glial development is affected, we

have taken a gain of function approach. We find that Dpp

overactivity leads to a pronounced increase in the number of

glial cells in both eye disc and optic stalk, and to ectopic

migration of the glia beyond their normal anterior boundary.

The increase in cell number, coupled with the fact that there is

no appreciable cell death in wild-type glia at this stage of eye

development, suggests that Dpp stimulates glial cell prolif-

eration. This result is in agreement with previous studies

showing a role for Dpp in the growth of wing and eye imagi-

nal disc cells (Burke and Basler, 1996; Capdevila and Guer-

rero, 1994).

For Hh, by contrast, our loss of function data show only a

mild effect on the accumulation of glia in the eye disc, and

the gain of function results confirm that there is no signifi-

cant effect of Hh on glial proliferation. This is somewhat

surprising, given that Hh is required for the growth of the

Drosophila eye disc and the proliferation of the neurons in

the lamina (Heberlein et al., 1995; Huang and Kunes, 1996)

and in view of its mitogenic effects on many cell types of the

vertebrate nervous system, including astrocytes in the optic

nerve of rodents (Jensen and Wallace, 1997; Wallace and

Raff, 1999; Wechsler-Reya and Scott, 1999).

R. Rangarajan et al. / Mechanisms of Development 108 (2001) 93–103 99

Fig. 6. Increased Hh signaling stimulates the motility of subretinal glia.

Labeling of clones with anti-b-galactosidase antibodies (blue), of glial cells

with anti-Repo antibodies (red), and of photoreceptors/axons with anti-HRP

antibodies (green). (A,B) y,hsFLP122/Act . FLPout . GAL4; UAS lacZ;

UAS hh eye disc, expressing Hh in random patches. The position of the

clone is visualized in (A) and the ectopically migrating glia (arrow) in (B).

The glia overshoot anteriorly, without regard to the position of the clone; in

this example, they migrate away from the clone. (C,D) Two examples of

ombGAL4; UAS Ciact eye discs, showing ectopic glial migration; in both

cases, the overshooting glia are accompanied by axons misrouting ante-

riorly (arrows).

3.3. Subretinal glial migration

Subretinal glia migrate from the optic stalk into eye discs

only when photoreceptors have begun to differentiate in the

eye disc (Choi and Benzer, 1994), but they do not require

axons as a substrate for their migration (Rangarajan et al.,

1999). They also migrate towards ectopic islands of differ-

entiating photoreceptors, again without axonal substrate.

These findings suggest that differentiating photoreceptors

regulate the migration of the subretinal glia over a distance,

either by secreting diffusible molecules or by preferentially

stabilizing far-reaching glial filopodia (Rangarajan et al.,

1999). Our present study demonstrates that the diffusible

factors Dpp and Hh are among the signals the differentiating

photoreceptors send to promote subretinal glial motility.

Our gain-of-function experiments show that overactivity

of Dpp as well as of Hh signaling leads subretinal glial cells to

overshoot their normal anterior boundary, either in a rela-

tively narrow stream or in a broad front. In the case of Dpp,

this ectopic migration of the glia is accompanied by severe

overproliferation, and we therefore had to consider the possi-

bility that the changes in migratory behavior are merely a

secondary consequence of the increase in cell number.

However, we find that severe overproliferation of glia in

otherwise wild type eye discs does not lead to ectopic migra-

tion. This is true not only when overproliferation is induced

by acceleration of the cell cycle, but also when it is caused by

Ras overactivity (Rangarajan and Gaul, unpublished obser-

vation), which presumably mimics the effects of a wide range

of growth factors. Similarly, overproliferation of glia in the

optic stalk of eya mutants, which lack photoreceptors, does

not induce migration into the eye disc. Conversely, we do

observe ectopic migration of glia in eye discs with Hh over-

activity, i. e. in discs in which the number of glial cells is

largely normal. These results demonstrate that overprolifera-

tion in itself is not sufficient to induce abnormal migration of

glial cells and that proliferation and migration are in fact

independently regulated processes.

The ectopic migrations of glial cells that we observe are

not preferentially directed towards the sources of Dpp or

Hh. Moreover, similar migratory phenotypes can be induced

by expression of constitutively active components of the

Dpp and Hh pathways within the glial cells. Such cell-

autonomous activation of the pathway precludes the use

of information about ligand distribution as a positional

cue. Both findings therefore argue that Dpp and Hh exert

their effect on the glia not by providing positional informa-

tion, but by stimulating motility. This motogenic effect may

be due to an enhanced ability to migrate or to an impaired

ability to respond to an inhibitory signal (see below). In any

case, both Dpp and Hh appear to act by regulating transcrip-

tion within the glial cells, through their canonical signal

transduction machinery.

One of the best understood motogens is scatter factor (SF)

(for review see Birchmeier and Gherardi, 1998), which

disperses cohesive colonies of epithelial cells by stimulating

random motility. However, it can also act as a mitogen

(hepatocyte growth factor), morphogen, trophic factor, or

chemoattractant depending on the cellular context; in all

cases, its effects are mediated by the receptor tyrosine

kinase c-Met. The actual scattering induced by SF occurs

only 4–6 h after addition of SF and is inhibited by cyclo-

heximide, suggesting that regulation of gene expression is

involved (Ridley et al., 1995). Thus, Dpp and Hh resemble

SF both in terms of phenotypic effect and with regard to

their mechanism of action.

The function of Hh as a motogen is novel. Although the

molecule is being studied in many contexts, a role in cell

motility has not been attributed to it to date. Dpp and its

vertebrate counterparts, on the other hand, have been impli-

cated in the control of migratory processes during develop-

ment. Dpp plays a role in tracheal cell migration along the

dorso-ventral body axis (Vincent et al., 1997) and in the

dorsal migration of the ectoderm during dorsal closure of

the Drosophila embryo (for review see Stronach and Perri-

mon, 1999). During wound healing, the migration of various

cell types, including astrocytes, depends on TGF-b (Martin,

1997).

Why don’t we observe more clear-cut effects on the

migratory behavior of the subretinal glia in our loss of func-

tion experiments? Our mosaic data confirm at least for Dpp

that the molecule is required for the normal accumulation of

glia in the eye disc. But because of the limitation of our

mosaic technique it is difficult to distinguish between effects

on proliferation and on motility and to discern specific

migratory effects: the failure of ombGAL4 to drive recom-

bination to completion prior to entry of the glia into the eye

disc means that at least some glia arrive in the eye disc and

migrate anteriorly as heterozygotes, i.e. before undergoing

recombination to become homozygous mutant.

Another major reason for the lack of a more pronounced

loss of function effect is most likely that the Dpp and Hh

signal transduction pathways are (partially) functionally

redundant with other pathways in regulating glial cell devel-

opment. As we have shown previously, Ras signaling, possi-

bly triggered by diffusible RTK ligands, is very likely to

play a role in regulating both the proliferative and migratory

behavior of the subretinal glia (Rangarajan et al., 1999).

Generally, Dpp and Hh must be part of a more complex

network of signals regulating glial development in the eye

disc. For example, we do not yet know how the sharp ante-

rior boundary of glial migration behind the morphogenetic

furrow is established. Possible mechanisms include an

attractive substrate or signal in the posterior, or an inhibitory

signal in the anterior. Since in wild type the distribution of

both Dpp and Hh extends further to the anterior than the

glial cells migrate, this mechanism also has to be able to

counteract the motogenic effects of Dpp and Hh.

3.4. Signaling by Dpp and Hh: mechanism of action

How do Dpp and Hh exert their effects on subretinal glial

R. Rangarajan et al. / Mechanisms of Development 108 (2001) 93–103100

migration? Since Hh and Dpp are known to regulate each

other’s expression in other tissues (Ingham, 1998), it would

be conceivable that they act in sequence. However, neither

dpp nor hh is expressed in the subretinal glial cells, which

rules out the possibility that they mutually regulate each

other’s transcription and implies that Dpp and Hh both act

in a paracrine fashion. It therefore seems likely that Dpp and

Hh signaling converge to control the transcription of genes

required for cell motility (or possibly for reading the stop

signal). These targets could include genes coding for cytos-

keletal components, proteins regulating cell adhesion, or

enzymes to degrade the extracellular matrix. The idea that

such genes could be transcriptionally controlled by Dpp or

Hh is supported by the recent finding that in the Drosophila

wing compartmental cell sorting at the anterior/posterior

boundary is under the opposing transcriptional control of

Hh and En, suggesting that these molecules regulate the

expression of a single cell adhesion molecule (Dahmann

and Basler, 2000). The identification of transcriptional

targets of Dpp and Hh signaling will provide important

insights into the mechanism by which these signals regulate

glial cell motility.

4. Experimental procedures

4.1. Drosophila stocks

The GMRGAL4 line was kindly provided by M. Freeman,

ombGAL4 by S. Cohen, UAS dpp and UAS hh by J. Treis-

man, Act . FLPout . GAL4 by S.L. Zipursky, UAS tkvact

by R. Padgett (chromosome II) and G. Struhl (chromosome

III), UAS tkvDGSK (dominant negative form) by M. O’Con-

nor, UAS stg and UAS cycE by B. Edgar, UAS Ciact by T.

Orenic, UAS Cidn by K. Basler, UAS FLP by the Blooming-

ton Stock Center, FRT40A mad12 by E. Matunis, and

FRT40A smo3 by S. DiNardo.

4.2. Construction of UAS . FLPout . GFP

A UAS GFP transgene was constructed by cutting out a

0.7 kb Xba fragment containing the GFP coding sequence

from PeGFP (Clontech) and cloning it into the Xba site of

the pUAST vector (Brand and Perrimon, 1993). Next, the

FLPout cassette, consisting of DRaf sequences flanked by

FRT sites, was excised as a 7.0 kb Kpn fragment from

plasmid D237 (gift of G. Struhl) (Struhl and Basler, 1993)

and cloned into the Kpn site of the polylinker of the UAS

GFP construct located between the UAS and GFP

sequences to generate the UAS . DRaf . GFP transgene.

4.3. Tracking morphologies of individual glial cells

expressing GFP

y,hsFLP122; Adv/SM6-TM6B females were crossed to

ombGAL4; UAS . DRaf . GFP males. The progeny of

the cross were heat-shocked at 48–96 h of development at

358C for 30 min and female larvae were dissected the next

day. The morphologies of these cells was visualized by the

cytoplasmic GFP; glial nuclei were labeled with anti-Repo

antibodies.

4.4. Generation of dpp- and hh- expressing FLPout clones

This was done as described in Pignoni and Zipursky,

(1997). y,hsFLP122; UASdpp/TM6B males were crossed

to Act . FLPout . GAL4; UASlacZ females. The progeny

of the cross were heat shocked at 24–36 h of development

for 30 min at 358C. Female larvae were dissected at the late

third instar larval stage and stained with 22C10, FITC anti-

HRP, anti-b-galactosidase, and anti-Repo antibodies; or

with anti-pH3 (phosphorylated histone 3), FITC-coupled

anti-HRP antibodies, and propidium iodide, a DNA dye.

For hh-expressing clones, y,hsFLP122; UAShh/TM6B

females were crossed to Act . CD2 . GAL4; UASlacZ

males. The progeny of the cross were heat shocked at 24–

36 h of development for 30 min at 398C.

4.5. Generation of loss of function clones in the glia

This was done essentially by the FRT-UAS FLP method

described before (Duffy et al., 1998; Xu and Rubin, 1993),

using ombGAL4 as a driver to restrict recombination to the

glial population. A UAS GFP nuclear transgene inserted on

the non-mutant FRT chromosome (Courvoisier and Gaul,

unpublished) served to negatively mark mutant glial cells.

To generate homozygous mutant clones of smo3 and mad12,

and double mutant clones of smo3 and mad12 in the glial

cells, ombGAL4; FRT40A,UAS GFP males were crossed to

FRT40A smo3; UAS FLP/SM6-TM6B females, FRT40A

mad12;UAS FLP/SM6-TM6B females, or FRT 40A mad12,

smo3; UAS FLP/SM6-TM6B females, respectively. Female

third instar larvae were dissected and stained with anti-Repo

antibodies and propidium iodide. The numbers of mutant or

GFP minus cells were counted and are reported as percen-

tages of the total number of glial cells. Average percentages

of GFP minus cells are given as mean ^ SEM. Due to the

non-Gaussian distribution of the data, differences between

groups have been statistically assessed by the Mann–Whit-

ney test (Altman, 1991).

4.6. Histology

For RNA in situ hybridization, antisense digoxigenin

labeled RNA probes for dpp were generated according to

the manufacturer’s specifications (Boehringer). Whole-

mount in situ hybridization was carried out essentially as

described by Tear et al. (1996). For antibody stainings, eye–

brain complexes were dissected and stained as described in

Rangarajan et al. (1999); anti-H3 phosphorylated histones

(Upstate Biotechnology) were used at 1:100 dilution. To

stain samples with propidium iodide, fixed eye brain

complexes were digested with 0.1 mg/ml of RNase A for 1

h, followed by two quick washes with PBS-TX and incuba-

R. Rangarajan et al. / Mechanisms of Development 108 (2001) 93–103 101

tion with propidium iodide (Sigma) at a 1:500 dilution for 1

h. Specimens were imaged on a Zeiss LSM 510 confocal

microscope.

Acknowledgements

We thank K. Basler, S. Cohen, S. DiNardo, B. Edgar, M.

Freeman, E. Matunis, M. O’Connor, T. Orenic, R. Padgett,

G. Struhl, J. Treisman, S. L. Zipursky, and the Bloomington

Stock Center for stocks and reagents, and we thank S.L.

Zipursky for sharing results prior to publication. We are

grateful to S. Cohen, N. Heintz, J. Hudspeth, and members

of the Gaul laboratory for critically reading the manuscript,

and to H.R. Chung for help in preparing Fig. 1. This work

was supported by a graduate fellowship from the Rockefel-

ler University (R.R.), by a postdoctoral fellowship from the

Association pour la Recherche sur le Cancer (H.C.), by a

grant from the National Institutes of Health (U.G.), and by

Klingenstein Fellowship and McKnight Scholar Awards

(U.G.).

References

Alcedo, J., Ayenzon, M., Von Ohlen, T., Noll, M., Hooper, J.E., 1996. The

Drosophila smoothened gene encodes a seven-pass membrane protein,

a putative receptor for the Hedgehog signal. Cell 86, 221–232.

Altman, D.G., 1991. Practical Statistics for Medical Research, Chapman

and Hall, London.

Auld, V., 1999. Glia as mediators of growth cone guidance: studies from

insect nervous systems. Cell. Mol. Life Sci. 55, 1377–1385.

Barres, B.A., Jacobson, M.D., Schmid, R., Sendtner, M., Raff, M.C., 1993.

Does oligodendrocyte survival depend on axons? Curr. Biol. 3, 489–

497.

Barres, B.A., Raff, M.C., 1993. Proliferation of oligodendrocyte precursor

cells depends on electrical activity in axons. Nature 361, 258–260.

Barres, B.A., Raff, M.C., 1999. Axonal control of oligodendrocyte devel-

opment. J. Cell Biol. 147, 1123–1128.

Birchmeier, C., Gherardi, E., 1998. Developmental roles of HGF/SF and its

receptor, the c-Met tyrosine kinase. Trends Cell Biol. 8, 404–410.

Birling, M.C., Price, J., 1995. Influence of growth factors on neuronal

differentiation. Curr. Opin. Cell Biol. 7, 878–884.

Blackman, R.K., Sanicola, M., Raftery, L.A., Gillevet, T., Gelbart, W.M.,

1991. An extensive 3 0 cis-regulatory region directs the imaginal disk

expression of decapentaplegic, a member of the TGF-b family in

Drosophila. Development 111, 657–666.

Brand, A.H., Perrimon, N., 1993. Targeted gene expression as a means of

altering cell fates and generating dominant phenotypes. Development

118, 401–415.

Briscoe, J., Ericson, J., 1999. The specification of neuronal identity by

graded Sonic hedgehog signalling. Cell Dev. Biol. 10, 353–362.

Burke, R., Basler, K., 1996. Hegehog-dependent patterning in the Droso-

phila eye can occur in the absence of Dpp signaling. Dev. Biol. 179,

360–368.

Burne, J.F., Raff, M.C., 1997. Retinal ganglion cell axons drive the prolif-

eration of astrocytes in the developing rodent optic nerve. Neuron 18,

223–230.

Capdevila, J., Guerrero, I., 1994. Targeted expression of the signalling

molecule decapentaplegic induces pattern duplications and growth

alterations in Drosophila wings. EMBO J. 13, 4459–4468.

Choi, K.-W., Benzer, S., 1994. Migration of glia along photoreceptor axons

in the developing Drosophila eye. Neuron 12, 423–431.

Dahmann, C., Basler, K., 2000. Opposing transcriptional outputs of Hedge-

hog signaling and engrailed control compartmental cell sorting at the

Drosophila A/P boundary. Cell 100, 411–422.

Das, P., Madzuia, L.L., Wang, H., Finelli, A.L., Cho, S., Smith, M.M.,

Padgett, R.W., 1998. The Drosophila gene medea demonstrates the

requirement for different Smads in dpp signaling. Development 125,

1519–1528.

Duffy, J.B., Harrison, D.A., Perrimon, N., 1998. Identifying loci required

for follicular patterning using directed mosaics. Development 125,

2263–2271.

Giangrande, A., Murray, M.A., Palka, J., 1993. Development and organiza-

tion of glial cells in the peripheral nervous system of Drosophila mela-

nogaster. Development 117, 895–904.

Greenwood, S., Struhl, G., 1999. Progression of the morphogenetic furrow

in the Drosophila eye: the roles of Hedgehog, Decapentaplegic and the

Raf pathway. Development 126, 5795–5808.

Halter, D.A., Urban, J., Rickert, C., Ner, S.S., Ito, K., Travers, A.A., Tech-

nau, G.M., 1995. The homeobox gene repo is required for the differ-

entiation and maintenance of glia function in the embryonic nervous

system of Drosophila melanogaster. Development 121, 317–332.

Hartenstein, V., Nassif, C., Lekven, A., 1998. Embryonic development of

the Drosophila brain. II. Pattern of glial cells. J. Comp. Neurol. 402, 32–

47.

Hay, B.A., Wolff, T., Rubin, G.M., 1994. Expression of bacculovirus P35

prevents cell death in Drosophila. Development 121, 2121–2129.

Heberlein, U., Wolff, T., Rubin, G.M., 1993. The TGF-b homolog Dpp and

the segment polarity gene Hedgehog are required for the propagation of

a morphogenetic wave in the Drosophila retina. Cell 75, 913–926.

Heberlein, U., Singh, C.M., Luk, A.Y., Donohoe, T.J., 1995. Growth and

differentiation in the Drosophila eye coordinated by Hedgehog. Nature

373, 709–711.

Hendzel, M.J., Wei, Y., Mancini, M.A., Van Hooser, A., Ranalli, T., Brink-

ley, B.R., Bazett-Jones, D.P., Allis, C.D., 1997. Mitosis specific phos-

phorylation of histone H3 initiates primarily within pericentromeric

heterochromatin during G2 and spreads in an ordered fashion coinci-

dent with mitotic chromosome condensation. Chromosoma 106, 348–

360.

Huang, Z., Kunes, S., 1996. Hedgehog, transmitted along retinal axons,

triggers neurogenesis in the developing visual centers of the Drosophila

brain. Cell 86, 411–422.

Huang, Z., Kunes, S., 1998. Signals transmitted along retinal axons in

Drosophila: Hedgehog signal reception and the cell circuitry of lamina

cartridge assembly. Development 125, 3753–3764.

Ingham, P., 1998. Transducing Hedgehog: the story so far. EMBO J. 17,

3505–3511.

Ito, K., Urban, J., Technau, G.M., 1995. Distribution, classification and

development of Drosophila glial cells in the late embryonic and early

larval ventral nerve cord. Roux’s Arch. Dev. Biol. 204, 284–307.

Jensen, A., Wallace, V.A., 1997. Expression of Sonic hedgehog and its

putative role as precursor cell mitogen in the developing mouse retina.

Development 124, 363–371.

Lecuit, T., Brook, W.J., Ng, M., Calleja, M., Sun, H., Cohen, S.M., 1996.

Two distinct mechanisms for long-range patterning by Decapentaplegic

in the Drosophila wing. Nature 381, 387–393.

Ma, C., Zhou, Y., Beachy, P.A., Moses, K., 1993. The segment polarity

gene hedgehog is required for the progression of the morphogenetic

furrow in the developing Drosophila eye. Cell 75, 927–938.

Martin, P., 1997. Wound healing–aiming for perfect skin regeneration.

Science 276, 75–81.

Massague, J., 1998. TGF-beta signal transduction. Annu. Rev. Biochem.

67, 753–791.

Massague, J., Chen, Y.G., 2000. Controlling TGF-beta signaling. Genes

Dev. 14, 627–644.

McMahon, A.P., 2000. More surprises in the Hedgehog signaling pathway.

Cell 100, 185–188.

Methot, N., Basler, K., 1999. Hedgehog controls limb development by

R. Rangarajan et al. / Mechanisms of Development 108 (2001) 93–103102

regulating the activities of distinct transcriptional activator and repres-

sor forms of Cubitus interruptus. Cell 96, 819–831.

Nellen, D., Burke, R., Struhl, G., Basler, K., 1996. Direct and long-range

action of a Dpp morphogen gradient. Cell 85, 357–358.

Neufeld, T.P., de la Cruz, A.F., Johnston, L.A., Edgar, B.A., 1998. Coordi-

nation of growth and cell division in the Drosophila wing. Cell 93,

1183–1193.

Perez, S.E., Steller, H., 1996. Migration of glial cells into retinal axon target

field in Drosophila melanogaster. J. Neurobiol. 30, 359–373.

Pignoni, F., Zipursky, S.L., 1997. Induction of Drosophila eye development

by decapentaplegic. Development 124, 271–278.

Rangarajan, R., Gong, Q., Gaul, U., 1999. Migration and function of glia in

the developing Drosophila eye. Development 126, 3285–3292.

Ridley, A.J., Comoglio, P.M., Hall, A., 1995. Regulation of Scatter Factor/

Hepatocyte Growth Factor responses by Ras, Rac and Rho in MDCK

cells. Mol. Cell. Biol. 15, 1110–1122.

Silver, J., 1993. Glia–neuron interactions at the midline of the developing

mammalian brain and spinal cord. Perspect. Dev. Neurobiol. 1, 227–

236.

Stronach, B.E., Perrimon, N., 1999. Stress signaling in Drosophila. Onco-

gene 18, 6172–6182.

Struhl, G., Basler, K., 1993. Organizing activity of Wingless protein in

Drosophila. Cell 72, 527–540.

Tear, G., 1999. Neuronal guidance. A genetic perspective. Trends Genet.

15, 113–118.

Tear, G., Harris, R., Sutaria, S., Kilomanski, K., Goodman, C.S., Seeger,

M.A., 1996. commissureless controls growth cone guidance across the

CNS midline in Drosophila and encodes a novel membrane protein.

Neuron 16, 501–514.

Tessier-Lavigne, M., Goodman, C.S., 1996. The molecular biology of axon

guidance. Science 274, 1123–1133.

Thomas, J.B., 1998. Axon guidance: crossing the midline. Curr. Biol. 8,

R102–R104.

Treisman, J.E., Heberlein, U., 1998. Eye development in Drosophila:

formation of the eye field and control of differentiation. Curr. Top.

Dev. Biol. 39, 119–158.

Vincent, S.E.R., Greider, N., Chen, C.K., Haerry, T., Schuh, R., Affolter,

M., 1997. Dpp controls tracheal cell migration along the dorsoventral

body axis of the Drosophila embryo. Development 124, 2741–2750.

Wallace, V.A., Raff, M.C., 1999. A role for Sonic hedgehog in axon-to-

astrocyte signaling in the rodent optic nerve. Development 126, 2901–

2909.

Wechsler-Reya, R.J., Scott, M.P., 1999. Control of neuronal precursor

proliferation in the cerebellum by Sonic Hedgehog. Neuron 22, 103–114.

Wiersdorff, V., Lecuit, T., Cohen, S.M., Mlodzik, M., 1996. Mad acts

downstream of Dpp receptors, revealing a differential requirement for

dpp signaling in initiation and propagation of morphogenesis in the

Drosophila eye. Development 122, 2153–2162.

Xu, T., Rubin, G.M., 1993. Analysis of genetic mosaics in developing and

adult Drosophila tissue. Development 117, 1223–1237.

R. Rangarajan et al. / Mechanisms of Development 108 (2001) 93–103 103