function of the drosophila tgf-a homolog spitz is controlled by star

Function of the Drosophila TGF-a homolog Spitz is controlled byStar and interacts directly with Star

Frank Hsiunga, Eric R. Grif®sa, Amanda Pickupb, Maureen A. Powersa, Kevin Mosesa,*

aDepartment of Cell Biology, Emory University School of Medicine, 1648 Pierce Drive NE, Atlanta, GA 30322-3030, USAbMolecular, Cell, and Developmental Biology, University of California at Los Angeles, 2203 Life Sciences, 621 Young Drive South,

Los Angeles, CA 90095, USA

Received 6 April 2001; received in revised form 15 May 2001; accepted 17 May 2001

Abstract

Drosophila Spitz is a homolog of transforming growth factor a (TGF-a) and is an activating ligand for the EGF receptor (Egfr). It has been

shown that Star is required for Spitz activity. Here we show that Star is quantitatively limiting for Spitz production during eye development. We

also show that Star and Spitz proteins colocalize in Spitz sending cells and that this association is not coincident with the site of translation ±

consistent with a function for Star in Spitz processing or transmission. Finally, we have de®ned minimal sequences within both Spitz and Star

that mediate a direct interaction and show that this binding can occur in vivo. q 2001 Elsevier Science Ireland Ltd. All rights reserved.

Keywords: Drosophila; Star; Spitz; TGF-a; Growth factor; Secretion; Development; Signal transduction; EGF; EGF receptor; Retina

1. Introduction

The Drosophila homolog of the EGF receptor (Egfr)

mediates signal transduction during many aspects of devel-

opment, including regulation of the cell-cycle and pattern-

ing during oogenesis, zygotic and imaginal development

(Schweitzer and Shilo, 1997; Freeman, 1998; Nilson and

SchuÈpbach, 1999; Bier, 2000; Carpenter, 2000; Schles-

singer, 2000). In the developing eye, Egfr signaling

mediates the speci®cation of the eye itself, cell growth

and survival, the recruitment of cells to the developing

ommatidia, and the induction of development in the optic

lobes of the brain (Xu and Rubin, 1993; Sawamoto and

Okano, 1996; Dickson, 1998; Dominguez et al., 1998;

Huang et al., 1998; Kumar et al., 1998; Spencer et al.,

1998; Chen and Chien, 1999; Kumar and Moses, 2000,

2001; Yang and Baker, 2001).

The ®rst of several positively acting Egfr ligands known

in Drosophila was Spitz, which is a single EGF domain

growth factor homolog in the transforming growth factor

a (TGF-a)/glial growth factor (GGF) family (Aaronson,

1991; Rutledge et al., 1992; Schweitzer et al., 1995). spitz

acts in patterning the dorsal chorionic appendages, the

embryonic cuticle, central nervous system and the imaginal

discs (Mayer and NuÈsslein-Volhard, 1988; Schweitzer et al.,

1995; Golembo et al., 1996; O'Keefe et al., 1997; Stemer-

dink and Jacobs, 1997; Szuts et al., 1997; Wasserman and

Freeman, 1998). In the developing eye, Spitz is required for

the assembly of the ommatidia after the speci®cation of the

founding R8 photoreceptor cell (Freeman, 1994, 1996; Tio

et al., 1994; Tio and Moses, 1997). At least some of these

functions in the developing eye appear to be conserved in

vertebrates (Reh and Cagan, 1994; Bermingham-McDo-

nogh et al., 1996; Neumann and NuÈsslein-Volhard, 2000).

The Drosophila Star protein and another family of

proteins called Rhomboids are required in the Spitz sending

cell for the proper function of Spitz in all stages of devel-

opment, including the developing eye (Bier et al., 1990;

Heberlein and Rubin, 1991; Freeman et al., 1992; Heberlein

et al., 1993; Kolodkin et al., 1994; Schweitzer et al., 1995;

Sturtevant and Bier, 1995; Queenan et al., 1997; Schweitzer

and Shilo, 1997; Stemerdink and Jacobs, 1997; Freeman,

1998; Guichard et al., 1999; Wasserman et al., 2000). Star

is an integral membrane protein that is expressed in cells

that secrete Spitz and is localized to the early endoplasmic

reticulum (ER) and nuclear envelope (Kolodkin et al., 1994;

Pickup and Banerjee, 1999). However, while Rhomboid is

also an integral membrane protein, it has been localized to

apical patches only (Sturtevant et al., 1996). This difference

in subcellular localization suggests that Star may act at an

earlier point than Rhomboid in the pathway of Spitz matura-

tion. Schweitzer and co-workers showed that Star and

Mechanisms of Development 107 (2001) 13±23

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

PII: S0925-4773(01)00432-4

www.elsevier.com/locate/modo

* Corresponding author. Tel.: 11-404-727-8799; fax: 11-404-727-4717.

E-mail address: [email protected] (K. Moses).

Rhomboid function is required for the signaling activity of

ectopically expressed full-length Spitz (`membrane' or

`mSpitz') but not for the activity of a truncated form of

Spitz that does not require cleavage to release the factor

domain (`secreted' or `sSpitz', Schweitzer et al., 1995).

This suggests that Star and Rhomboid both act in the

production of a functional Spitz signal in the sending cell

and, furthermore, that they may function in the maturation

of the Spitz factor: in protein cleavage, glycosylation, secre-

tion or presentation. In a conceptually similar experiment,

the expression of Drosophila Spitz in a Xenopus animal cap

sandwich assay has been shown to require both Star and

Rhomboid (Bang and Kintner, 2000). The very fact that

Star nulls are haploinsuf®cient dominant mutations with a

rough eye phenotype indicates that Star is genetically dose

sensitive and suggests that Star function may be quantita-

tively limiting on the Spitz signal during eye development

(Harris et al., 1976; Mayer and NuÈsslein-Volhard, 1988).

Finally, the published subcellular localizations of Star

protein at the EM level and of Spitz by confocal microscopy

suggest that Star may interact directly with Spitz at some

point during its translation, maturation, secretion or presen-

tation (Tio and Moses, 1997; Pickup and Banerjee, 1999).

In this paper, we present data that support the direct inter-

action of Star with Spitz in vivo in the developing eye. We

show that Spitz and Star proteins colocalize in developing

retinal cells by confocal microscopy and that the quantita-

tive expression of Star can control the quantitative expres-

sion of Spitz ± in both loss and gain-of-function Star

genotypes in the developing eye. Furthermore, as previously

reported (Freeman, 1996), we con®rm that ectopic expres-

sion of mSpitz has no phenotypic consequence unless

accompanied by the overexpression of Star. We show that

the binding of Star to Spitz is mediated by the 48-residue

domain of Spitz and by a 19-amino-acid segment of Star

(which we call the `growth factor binding domain' or

GFBD) and that these two short protein segments can direct

the colocalization of `cargo' proteins to the same sub-cellu-

lar compartment when expressed in HeLa cells. These data

will be discussed in terms of models for the function of Star

protein in Spitz signaling in vivo.

2. Results

2.1. Spitz and Star proteins are colocalized within

developing photoreceptor cells

We and others have shown that Star acts genetically

upstream of spitz, that Star is required for the function of

full length (m)Spitz but not a truncated secreted form

(s)Spitz and that Star protein can be detected in the nuclear

envelope and early ER (Schweitzer et al., 1995; Golembo et

al., 1996; Pickup and Banerjee, 1999; Bang and Kintner,

2000). Furthermore, Star has an essential role in the devel-

oping eye (Heberlein and Rubin, 1991; Heberlein et al.,

1993). We thus examined the subcellular expression

patterns of both Spitz and Star proteins as well as spitz

RNA in the developing eye by immunolocalization and/or

RNA in situ hybridization and confocal microscopy (see

Fig. 1 and Section 4).

To detect Spitz protein, we used a mouse polyclonal anti-

serum that was raised against a portion of Spitz that lies N-

terminal to the fully processed growth factor-homologous

domain (Tio and Moses, 1997). Thus the Spitz antigen

detected in the experiments described below can only

include precursor forms of the Spitz protein within the

cells that express it ± we are unlikely to be visualizing the

mature secreted factor (`s-Spitz'). We ®nd that Spitz antigen

is found throughout the cytoplasm of the developing photo-

receptor cells (Fig. 1A,D, as published in Tio and Moses,

1997). The Spitz protein is granular rather than evenly

distributed and these granules are both perinuclear (Fig.

1D) and located in the apical microvillae (Fig. 1A). Star

protein is somewhat more evenly distributed (Fig. 1B,E,H)

but is most abundant in granules that colocalize with the

Spitz granules (Fig. 1C,F). This is consistent with an early

association of Star with nascent Spitz protein in the ER (as

published by Pickup and Banerjee, 1999) and the mainte-

nance of this association all the way to the site of signal

transmission (the apical microvillae). However, we cannot

draw any ®rm conclusion that Spitz and Star are adjacent

proteins in the ER in vivo from these data alone: while if

they are naturally adjacent, they are likely to be ®xed in

close proximity to each other by the action of the PLP

(2% paraformaldehyde, 0.01 M NaIO4, 0.075 M lysine,

0.037 M NaPO4, pH 7.4) ®xative but the normal subcellular

membranes are probably disrupted by the subsequent wash-

ing step that includes triton. To further explore this proposed

direct interaction between Spitz and Star, we have under-

taken biochemical experiments described below.

Moreover, we found that spitz mRNA, while showing a

granular distribution, does not colocalize with the Star/Spitz

protein granules (Fig. 1I). This suggests that Star is not

directly involved in spitz translation. The Star antigen

appears to be more abundant than Spitz in these images.

However, this experiment is far from quantitative and we

are reluctant to draw any conclusion from this observation.

2.2. Star function is the limiting factor for the quantity of

Spitz antigen in vivo

While it was clear from published work that Star is

required for the expression of Spitz, it was not obvious

that the quantity of Star is normally limiting for Spitz

expression. To test this, we examined the expression of

Spitz antigen in the developing eye in conditions of both

loss-of-function for Star and of Star ectopic expression (Fig.

2). In Star mutant (loss-of-function) mosaic clones Spitz

antigen is greatly reduced (Fig. 2A±C). That it is not entirely

absent may be due to the accumulation of some amount of

unprocessed Spitz precursor protein. We observe that Spitz

F. Hsiung et al. / Mechanisms of Development 107 (2001) 13±2314

protein level in the Star mutant clones was not visibly

reduced in the furrow itself (arrow in Fig. 2A,C), but only

in the later columns of developing ommatidia. It may be that

Star is not limiting for Spitz protein expression in the furrow

itself and that the later loss is due to a reduced number of

differentiating photoreceptor cells in the clone.

We expressed Star ectopically in the developing eye by

inducing mosaic clones with ey:FLP to drive the eye tissue

to consist entirely of recombinant cells (see Section 4).

FRT-mediated recombination is operationally a one-way

process: after mitosis, the two daughter cells are homozy-

gous. Under these conditions, FLPase is expressed in the eye

anlagen continuously. Thus as development proceeds, the

pool of un-recombined cells is progressively depleted. By

the third instar, the eyes are entirely composed of cells

homozygous for one or other of the two chromosomes.

These clones were visualized using negative marking

with b-galactosidase and the cells within these clones are

also homozygous for a Star cDNA driven by an hsp70 gene

promoter. Thus the third instar eye imaginal disc consists of

patches of cells that express b-galactosidase and are wild-

type for Star and patches of cells that are negative for b-

galactosidase and which express elevated levels of Star

ectopically. When Star protein is thus overexpressed in

mosaic clones, Spitz antigen now appears anterior to the

morphogenetic furrow (Fig. 2D±F). This indicates that the

quantity of Star is normally limiting for the expression of

Spitz ± at least in some parts of the developing eye (i.e.

anterior to the morphogenetic furrow). This must depend

on the observed low levels of spitz mRNA anterior to the

furrow (Tio et al., 1994; Tio and Moses, 1997).

Others have shown that Star loss-of-function mutant cells

die in the developing eye (Heberlein and Rubin, 1991;

Heberlein et al., 1993). Thus the loss of Spitz antigen in

our Star loss-of-function clones (see Fig. 2A±C) could be

trivially due to the apoptotic death of these cells. To test

F. Hsiung et al. / Mechanisms of Development 107 (2001) 13±23 15

Fig. 1. Spitz and Star protein, but not spitz mRNA colocalize in the developing eye. Confocal images of third instar eye imaginal discs showing Spitz and Star

antigens and spitz mRNA as indicated (see Section 4): (A±C) show an apical optical section, (D±F) are mid-level and (G±I) are basal. In (C,F), Spitz is red and

Star is green. In (I), spitz mRNA is red and Star protein is green. Note colocalization (C,F) of most intense level of stain (examples indicated by arrows). Scale

bar in (A) is 10 mm (all panels to the same scale). Anterior right.

this, we visualized apoptotic cells in Star loss-of-function

clones using TUNEL (Fig. 3A). We ®nd that there is cell

death in these clones, but that it occurs far posterior to the

furrow (about 10 ommatidial columns or 20 h after the

passage of the furrow). Furthermore, we were able to visua-

lize the near-normal expression of several other proteins in

these clones in the region of the furrow: Atonal (Ato in Fig.

3B) as well as Elav and Boss (data not shown). Thus we

conclude that living cells are present within the Star loss-of-

function clones for about 1 day after the passage of the

furrow. Furthermore, these cells are suf®ciently vigorous

to express at least three other proteins. Our observation

that they fail to express Spitz is thus not a simple conse-

quence of cell morbidity, but is a differential effect of loss of

Star function. This is consistent with a direct role for Star in

Spitz protein expression.

For further evidence that Star function is normally limit-

ing for Spitz function in vivo, we drove ectopic and elevated

levels of unprocessed full-length or `mSpitz' expression in

the developing eye using an GMR:Gal4 driver (see Section

4). We ®nd that the resulting adult compound eye is indis-

tinguishable from wild-type (Fig. 4A). This suggests that the

quantity of spitz mRNA expression is not normally limiting

for spitz function in the developing eye. We similarly over-

expressed Star and this results in a moderate rough eye

phenotype (Fig. 4B). This suggests that the quantity of

Star function may be limiting in normal development.

When we expressed both mSpitz and Star together in the

same animal, we observed a strong synergy (Fig. 4C). We

observed smooth eyes with little or no red pigment and few

mechanosensory bristles. This suggests a de®cit of acces-

sory cells. When we stained the third instar eye imaginal

discs from these animals, we observed that most or all nuclei

in the late developing ommatidial clusters stain for Elav

(Fig. 4D±F). This is consistent with their differentiation as

extra photoreceptor cells and may explain the reduced

number of accessory cells. Indeed, a very similar result

was observed for the function of these proteins in the devel-

oping wing (Pickup and Banerjee, 1999). The fact that

elevating levels of mSpitz has no gross effect on the devel-

oping eye until Star protein is also elevated is consistent

with a normal limiting function for Star on Spitz expression.


Fig. 3. Star loss-of-function induces late cell death. Confocal images of

third instar eye imaginal discs containing clones of cells mutant for Star

(see Section 4). TUNEL, Ato and LacZ as indicated. LacZ negatively marks

the homozygous Star2 cells. Position of the furrow indicated with arrow-

heads. Scale bar in (A) is 100 mm, (B) is to the same scale. Note that

TUNEL shows apoptotic cells in the Star clones ± but only about 10

columns, or 20 h after the passage of the furrow (arrow in A). Also note

that Ato expression is nearly normal in the clones (arrow in B). Anterior

right.

Fig. 2. Star function controls Spitz antigen. Confocal images of third instar eye imaginal discs containing clones of cells mutant for Star (see Section 4). (A±C)

Star loss-of-function, note that Spitz antigen is reduced in the clone (arrowheads). LacZ negatively marks the homozygous Star2 cells. (D±F) Star ectopic

expression clone, note that Spitz antigen is ectopically expressed both sides of the furrow (arrowheads). Spitz and LacZ antigens as indicated. Position of the

furrow indicated with arrows. Scale bars in (A) and (B) are 50 mm (panels in the same row to the same scale). Anterior right.

2.3. The EGF homologous `factor' domain of Spitz protein

mediates speci®c binding to Star in vitro

The colocalization of Spitz and Star proteins is consistent

with direct contact between these two proteins in vivo. To

approach this, we tested a series of fragments of Spitz fused

to glutathione-S-transferase (GST) for their ability to bind to

full-length (66 kDa) Star protein expressed in vitro and

labeled with 35S-methionine (see Section 4). We made

seven such protein fusions (SpitzA±GST through SpitzG±

GST, Fig. 5) and ®nd that the SpitzC fragment strongly

interacts with Star and that SpitzF interacts weakly (Fig.

5). However, we found that the weak binding of SpitzF is

not speci®c ± it also binds luciferase and all sub-fragments

of Star tested (see below). Thus we suggest that SpitzC

(residues 71±125) contains the major, high-speci®city bind-

ing domain for Star protein. An eighth Spitz fragment

(SpitzH, residues 75±122) is very similar to but slightly

smaller than SpitzC and behaves identically in this assay

(data not shown). SpitzC and SpitzH closely contain the

domain of Spitz, which is cleaved to release the six cysteine

containing EGF homologous `factor' domain (residues 78±

122). Thus our data are consistent with Spitz binding to Star

via the factor domain.

2.4. A 19-amino-acid fragment from Star protein mediates

speci®c binding to Spitz in vitro

We also sought to de®ne the minimal domain within Star

that is responsible for binding to Spitz. We repeated the

binding experiments described above using GST±SpitzC

and a series of seven sub-fragments of Star, 35S labeled in

vitro (called Star1 through Star7, see Fig. 6). Each Star

fragment was tested for binding to GST alone and to

GST±SpitzC. While Star1 and Star2 both bind GST±SpitzC,

they also bind equally well to GST alone. We thus cannot

demonstrate any speci®c SpitzC binding activity in Star1 or

Star2. However, both Star4 and Star5 do bind speci®cally to

SpitzC and these two fragments overlap over a short range

(residues 402±421). To con®rm this location for a speci®c

GFBD, we expressed a smaller fragment: Star7 (residues

390±420). As this fragment is too small to be labeled in

vitro by our protocol, we fused this to yellow ¯uorescent

protein (YFP). We ®nd that this Star7±YFP fusion is speci-

®cally precipitated by GST±SpitzC (Fig. 6). As a control,

we tested YFP alone and found that it is not precipitated by

either GST±SpitzC or GST alone. Our data are thus consis-

tent with a GFBD within Star that lies between the N-term-

inal residue of Star5 and the C-terminal amino acid of Star7

± a domain of only 19 residues.

2.5. The factor domain of Spitz and the GFBD of Star can

interact in vivo

Our in vitro experiments (above) suggest that Spitz binds

to Star via the factor domain and that Star binds to Spitz via

the 19-residue GFBD we de®ned. However, in vitro studies

like these may be prone to artifacts and require in vivo


Fig. 4. The quantity of Star is the limiting factor for Spitz function in vivo: (A±F) the protein(s) named in the lower right has been over-expressed in the

developing eye (see Section 4). (A±C) SEM of adult compound eyes, anterior right. (D±F) Confocal image of the edge of a third instar eye imaginal disc, red ±

Elav (neural nuclei), green ± actin. Overexpression of mSpitz has no effect and is indistinguishable from wild-type (A). Overexpression of Star has a moderate

effect (B). Overexpression of both Star and mSpitz together is synergistic (C). In the double overexpression (C), there is a de®cit of accessory cells resulting in a

smooth eye and an excess of photoreceptor neurons. Close to the furrow (arrowhead in D±F), there is a near normal array of single neurons. However, as

ommatidial assembly proceeds across the disc, excess numbers of neurons are recruited into each cluster (arrows in D±F). Scale bar in (A) is 100 mm (B,C to

the same scale). Scale bar in (D) is 25 mm (E,F to the same scale).


Fig. 6. GST±Spitz fragment C speci®cally recognizes a single region of the Star protein. At the top is a diagram of the structure of Star (597 amino acids total).

The N and C termini and the trans-membrane domain are indicated. Below are seven sub-fragments (1±7, see Section 4). These fragments were individually

labeled and precipitated with GST±Spitz fragment C (SpitzC) and glutathione±agarose beads complexed to GST alone (GST, as a control) as shown in the gels.

Protein fragment sizes indicated in kDa. Note that as Star7 is fused to YFP, it runs high on the gel (see Section 4). Note that Star fragment 2 is equally

precipitated by both Spitz and GST ± and we therefore consider this binding non-speci®c. Note that fragments 4, 5 and 7 de®ne a minimal sequence that is

speci®cally bound by Spitz fragment C.

Fig. 5. GST±Spitz fusion proteins can precipitate labeled full length Star. At the top is a diagram of the structure of Spitz (230 amino-acids in total). The N and

C termini and the signal sequence, mature factor and transmembrane domains are indicated. Below are seven sub-fragments that were fused to GST (A±G, see

Section 4). These were used to precipitate 35S-methionine labeled Star protein (66 kDa) as shown. Also shown are two controls: glutathione±agarose beads

alone (BDS) and glutathione±agarose beads bound to GST alone (GST). Note that fragment `C' strongly precipitates Star (dark shading) and fragment `F' does

so only weakly. However, Fragment `F' binds other targets such as luciferase (data not shown) and we therefore consider this binding non-speci®c. Fragment

`C' does not bind to a control target (luciferase) in this assay.

con®rmation. To approach this, we expressed two readily

detectable `cargo proteins' in tissue culture (HeLa) cells and

then used these to test for the ability of the Spitz factor

domain and the Star GFBD to cause the `cargo' proteins

to colocalize.

The `cargo' proteins were yellow and cyan ¯uorescent

proteins (YFP fused to another fragment to yield YFP±pyru-

vate kinase (PK) and CFP). When YFP±PK is transfected

into HeLa cells, it is evenly distributed between the nucleus

and the cytoplasm (Fig. 7A). We directed CFP to the

nucleus by adding a nuclear localization sequence from

SV40 virus large T antigen (CFP±nuclear localization signal

(NLS), Kalderon et al., 1984a,b) and when transfected into

HeLa cells, as expected CFP±NLS is nuclear (Fig. 7E).

When the two fusion proteins (YFP±PK and CFP±NLS)

are cotransfected into HeLa cells, YFP±PK remains gener-

ally distributed and CFP is nuclear. This shows that the two

fusion proteins do not normally bind signi®cantly to each

other in HeLa cells.

We then fused the 48 amino-acid SpitzH fragment

(containing the factor domain) to CFP±NLS to yield CFP±

SpitzH±NLS and as expected, this fusion protein is still

directed to the nucleus when expressed in HeLa cells (Fig.

7B,H). We fused the GFBD containing Star7 fragment to

YFP±PK to yield YFP±Star7±PK and, as expected the

protein remains evenly distributed (Fig. 7D). However,

when CFP±SpitzH±NLS and YFP±Star7±PK are coex-

pressed in HeLa cells, both the cyan and the yellow ¯uor-

escent labels colocalize predominantly to the cell's nuclei

(Fig. 7G±I). This suggests that the two proteins bind

together in the HeLa cells' cytoplasm and are then cotran-

slocated into the nucleus by virtue of the SV40 NLS on the

CFP±SpitzH±NLS partner. Control experiments show that

translocation of the yellow label (YFP) to the nucleus


Fig. 7. The Spitz factor domain can drive subcellular colocalization of a cargo protein via the Star GFBD in vivo. Confocal images of HeLa cells transfected to

express proteins as indicated (see Section 4): (A±C) shows a cell coexpressing YFP±PK and CFP±SpitzH±NLS. Note that YFP±PK is evenly distributed while

CFP±SpitzH±NLS is predominantly nuclear. (D±F) show a cell coexpressing YFP±Star7±PK and CFP±NLS. Note that YFP±Star7±PK is evenly distributed

while CFP±NLS is predominantly nuclear. (G±I) shows a cell coexpressing YFP±Star7±PK and CFP±SpitzH±NLS. Note that now both proteins are predo-

minantly nuclear. Compare the three YFP expressing cells (A,D,G), and note that YFP±PK is predominantly nuclear only when both the Star GFBD and the

Spitz factor domain are available to link the YFP±PK fusion protein to the CFP±NLS fusion protein. Scale bar in (A) is 20 mm.

requires both SpitzH and Star7 ± either one alone will not

mediate binding (Fig. 7A±F). It is important to note that we

do not suggest that either Spitz or Star are normally nuclear

proteins. We used this colocalization assay only to demon-

strate that the Spitz factor domain and the Star GFBD are

capable of signi®cant and stable binding in a living cell.

3. Discussion

Genetic analysis indicated that Star (with rhomboid) may

function upstream of the transmission of Spitz from sending

cells (Schweitzer et al., 1995; Bang and Kintner, 2000) and

Star protein has been shown to be localized to the nuclear

envelope and the early ER (Pickup and Banerjee, 1999).

Taken together, these data suggested to us that Star may

function in the translation, post-translational cleavage,

glycosylation, secretion or presentation of Spitz. To

approach this, we localized Star and Spitz proteins and

spitz mRNA in a series of pair-wise double stains at the

confocal level. We found that Spitz and Star proteins do

colocalize, but that spitz RNA does not. Furthermore Spitz

and Star proteins appear together in granular structures that

are perinuclear as well as apical in the cells. This suggests

that the Spitz±Star interaction persists through much or all

of the secretory pathway.

We found that controlling the quantity of Star directly

affects the quantity of Spitz antigen seen (in loss and gain-

of-function mosaic clones and by ectopic expression in the

entire eye). In short, less Star results in less Spitz and more

Star in more (and ectopic) Spitz. Consistent with this, we

found that overexpression of mSpitz alone has no phenotypic

effect, but that overexpression of Star does result in a moderate

rough eye suggesting that normally spitz RNA is in excess and

the quantity of the signal is limited by Star. Furthermore, over-

expressing both mSpitz and Star together results in a synergis-

tic effect ± and a grossly disordered eye, with a large excess of

photoreceptors and a de®cit of accessory cells. As the main

function of the Spitz/Egfr signal in the eye is to recruit cells to

the developing clusters and specify them as photoreceptor

neurons, this phenotype is consistent with a great increase in

the quantity of this signal. Our immuno-colocalization data

appear to suggest that there is more Star antigen than Spitz

in the developing eye (Fig. 1). However, it is very dif®cult to

draw any conclusions as to the actual relative abundance of

these proteins ± these experiments were not quantitative.

Taken together, all these data suggest that the quantity of

Star protein is the critical limiting factor for the Spitz/Egfr

signal, at least during the normal development of the

compound eye.

We have also used a series of GST-mediated in vitro

binding experiments to de®ne a single region each in

Spitz and Star that mediate their direct interaction (Fig.

8). In Spitz, this 48-residue segment (SpitzH) is virtually

identical to the `factor' domain ± that part of the protein that

contains the six cysteine residues and other features that

show homology to the small diffusible growth factors of

the TGF-a family (MassagueÂ, 1990; Derynck, 1992). In

Star, we have de®ned a 19-amino-acid GFBD responsible

for binding to Spitz. To con®rm these results, we conducted

a test in vivo: we fused SpitzH to CFP as well as a dominant

NLS. This CFP±SpitzH±NLS fusion when expressed in

HeLa cells is directed to the nucleus by virtue of the NLS

and a cyan-colored signal is detected there by confocal

microscopy. We also made a fusion protein in which an

evenly distributed YFP±PK protein was fused to the

GFBD from Star (YFP±Star7±PK). When YFP±Star7±PK

is expressed alone in HeLa cells, the yellow signal is evenly

distributed, but when coexpressed with CFP±SpitzH±NLS,

the yellow signal moves to the nucleus. This `cargo' experi-

ment con®rms that the Spitz factor domain and the Star

GFBD can bind in vivo. Taken together, the in vitro

`GST-pull down' experiments and the in vivo HeLa cell

`cargo' experiments are consistent with a direct interaction

in the living ¯y between the Spitz factor domain and the Star

GFBB. However, neither of these two experiments tested

this interaction in the secretory pathway.

It is interesting to note that the Spitz `factor' domain is N-

terminal to the Spitz trans-membrane domain ± and thus

presumed to lie outside of the plasma membrane (or in the

lumen of the organelles of the secretory pathway). The

GFBD in Star lies C-terminal to its trans-membrane domain

and thus would appear to lie on the wrong side of the plasma

or organelle membranes to interact directly with the Spitz

factor domain. However, structural features of Star have led

others to suggest that Star is actually a type II integral

protein, with its C-terminus outside and its N-terminus

inside (Kolodkin et al., 1994). This is therefore consistent


Fig. 8. Binding domains in Spitz and Star. See text. Six conserved cysteine residues in Spitz indicated by asterisks.

with a direct interaction between the Spitz factor domain

and the Star GFBD in vivo.

In summary, we have shown that Spitz and Star proteins

associate in living cells in the developing Drosophila

compound eye, that Star controls the quantity of Spitz signal

and that these proteins interact via the factor domain in Spitz

and the GFBD (we de®ned) in Star. Our data are consistent

with a role for Star in some stage or stages of Spitz signal

production subsequent to its translation. These conclusions

are very similar to those reached by others for Rhomboid

family proteins (Bier et al., 1990; Freeman et al., 1992;

Sturtevant et al., 1993; Schweitzer et al., 1995; Sturtevant

et al., 1996; Sapir et al., 1998; Guichard et al., 1999; Bang

and Kintner, 2000; Wasserman et al., 2000). While we can

draw no ®rm conclusions from our data, we suggest that Star

may be involved in a complex in the secretory pathway that

acts in the maturation of Spitz. Star could act before Rhom-

boid as Star has been localized early in the pathway (Pickup

and Banerjee, 1999) and Rhomboid has been localized to the

apical microvillae (Sturtevant et al., 1996) or they may act

together. There is no evidence to suggest that either Star or

Rhomboid are themselves proteases capable of cleaving

Spitz: perhaps they recruit one. Alternately Star may act

as a chaperone to route the pro-Spitz protein correctly

within the secretory pathway or it might be required for

the correct folding of Spitz or it may recruit glycosylation

enzymes. Indeed, our data suggest that Star can interact with

the Spitz factor domain in vitro in conditions in which it

may not be correctly folded. It is interesting to note that

anterior to the furrow, in the developing eye that Star

appears to be quantitatively limiting on Spitz expression.

It may be that Spitz pro-protein that is not correctly routed

or cleaved may be unstable.

While there are several known Rhomboid proteins, Star

appears to be unique in the Drosophila genome. While

homologs of Rhomboid have been detected in vertebrates

(Guichard et al., 1999; Wasserman et al., 2000), we have

been unable to detect any homolog of Star outside of Droso-

phila, either by searching for similarities to the entire

protein sequence or to the GFBD alone. We have shown

that Star is essential in Drosophila for the activation of an

otherwise inactive growth factor homolog (Spitz). There

may be proteins with similar functions in vertebrates,

which have conserved structure but which are too far

diverged at the primary sequence level to be found with

current computer searching algorithms.

4. Experimental procedures

4.1. Immunolocalization, RNA in situ hybridization and

microscopy

For immunolocalization, imaginal discs were prepared

and stained largely as described by Tomlinson and Ready

(1987) with minor modi®cations: discs were dissected in

0.1 M NaPO4, pH 7.4, then ®xed at room temperature for

30±45 min in `PLP'. This solution was made fresh before

use. After ®xation, discs were transferred to 0.1 M NaPO4,

pH 7.4, 0.1% Triton X-100 and washed for 1 h at room

temperature. They were then transferred to 0.1 M NaPO4,

pH 7.4, 0.1% Triton X-100, 10% normal goat serum to

block for 15 min. Then discs were transferred to the primary

antibody (dilution optimized for each antibody used) in

0.1 M NaPO4, pH 7.4, 0.1% Triton X-100, 10% normal

goat serum overnight at 48C. Then discs were transferred

to 0.1 M NaPO4, pH 7.4, 0.1% Triton X-100 and washed for

20 min, twice. Then discs were transferred to 0.1 M NaPO4,

pH 7.4, 0.1% Triton X-100, 10% normal goat serum and

washed and blocked for 20 min. Next discs were transferred

to secondary antibody (dilution optimized for each antibody

used) in 0.1 M NaPO4, pH 7.4, 0.1% Triton X-100, 10%

normal goat serum and incubated at room temperature for

3 h. For colocalization, primary antibodies were: mouse

anti-Spitz (Tio and Moses, 1997); rat anti-Star (Pickup

and Banerjee, 1999). Secondary antibodies were: HRP±

donkey anti-rat (ML, Jackson Lab); Cy5-rat anti-mouse

(ML, Jackson Lab). HRP-donkey anti-rat was ampli®ed

using a Tyramide kit (NEN/Life Science Products). In

mosaic clone experiments, b-galactosidase (LacZ) protein

was visualized using either Rabbit anti-b-galactosidase

(Cortex Biochem) or mouse anti-b-galactosidase (Promega)

primary antibodies and FITC±goat anti-rabbit (Jackson

Labs) or Cy5±goat anti-mouse (Jackson Labs) secondaries,

as appropriate. Atonal protein was visualized with rabbit

anti-ATO (Jarman et al., 1994) primary antibody and

Cy5±rat anti-mouse (ML, Jackson Labs) secondary anti-

body. Elav protein was visualized with rat anti-ELAV

(Developmental Studies, Hybridoma Bank) primary and

FITC±goat anti-rat (Jackson Labs). Boss protein was visua-

lized with mouse anti-Boss (KraÈmer et al., 1991) primary

and Cy5±rat anti-mouse (ML, Jackson Labs) secondary

antibody. For spitz RNA in situ hybridization, digoxi-

genin-labeled DNA probes were prepared by polymerase

chain reaction (PCR) from a spitz cDNA clone BA3 (Tio

et al., 1994). Hybridization and visualization were as

published (Kumar and Moses, 2001). Apoptotic cells were

detected by TUNEL stain (Intergen Company) using the

manufacturer's protocols. Filamentous actin was visualized

with rhodamine-conjugated phalloidin (Molecular Probes).

Fluorescent images were obtained using a Zeiss LSM510

confocal microscope. Scanning electron microscopy (SEM)

was as previously described (Moses et al., 1989).

4.2. Genetics and mosaic clone analysis

For colocalization of Spitz and Star proteins in the devel-

oping eye, imaginal discs from w; hs-S8/In(2LR)O third

instar larvae raised continuously at 258C were prepared as

previously described (Tomlinson and Ready, 1987). hs-S8

was as previously described (Kolodkin et al., 1994). Star

loss-of-function clones were obtained from third instar


larvae of the genotype: ey:FLP/1; S126 FRT40A/

p(conD)25A FRT 40A. p(conD)25A drives general expres-

sion of b-galactosidase in the eye disc (Tio and Moses,

1997). Star ectopic expression clones were obtained from

third instar larvae of the genotype: ey:FLP/1; hs-S8

FRT43D/FRT43D P(arm:LacZ). Star and mSpitz were

misexpressed using UAS-S4-3 (Golembo et al., 1996) and

UAS-mSpi (Freeman, 1996) driven by GMR:Gal4 (Pignoni

and Zipursky, 1997).

4.3. Protein binding assays

Fragments of Spitz were generated by PCR with restric-

tion enzyme linked primers (BamH1 and EcoR1), ligated in

frame into pGEX-2T vector (Amrad Corp.), and expressed

in BL-21(DE3) competent cells (Stratagene). Spitz frag-

ments are: (A) amino acids 1±50; (B) 28±70; (C) 71±125;

(D) 130±177; (E) 161±195; (F) 178±213; (G) 196±230; (H)

75±122 (see Fig. 5). Star fragments were generated by PCR.

All Star fragments were synthesized with T7 promoters and

Kozak sequences at their 5 0 ends, and stop codons and

poly(A) addition sites at their 3 0 ends. The PCR products

were used directly for in vitro protein synthesis using the

TNT-coupled transcription/translation kit (Promega). Full

length Star is 597 amino acids, Star fragments are: (1)

amino acids 1±210; (2) 140±279; (3) 244±360; (4) 305±

421; (5) 402±522; (6) 422±597; (7) 390±420 (see Fig. 6).

Fragment 7 is too small for in vitro labeling, it was therefore

fused to YFP to increase its size and then treated similarly to

Fragments 1±6. Binding assays were as previously

described (Dai et al., 2000). Protein products were resolved

on 12% SDS polyacrylamide gels and transferred to nitro-

cellulose membranes before exposing ®lm at 2808C. Equal

amount of GST fusion protein was loaded in each lane,

standardized by Western blot using a goat-anti-GST anti-

body (Amersham Pharmacia) and equal quantities of 35S in

vitro labeled full length Star was used for each lane in Fig. 5.

Equal amounts of 35S label for each Star fragment was used

for each lane in Fig. 6.

4.4. Cell culture experiment

CFP±NLS and CFP±Spitz fragment H-NLS were cloned

into the pECFP-C1 vector (Clonetech) between BglII and

BamH1 sites. CFP±NLS was made by PCR amplifying the

SV40 NLS sequence with BglII-linked 5 0 primer and

BamH1-linked 3 0 primer. CFP±SpitzH±NLS was made by

amplifying the Spitz fragment with a BglII-linked 5 0 primer,

and a BamH1-linked 3 0 primer that also includes the SV40

NLS sequence. SpitzH was tested and shown to have the

same binding capability to Star as SpitzC in a GST precipi-

tation assay as described above. Truncated PK (residues 15±

430) was PCR ampli®ed with enzyme site linked primers

and cloned into pEYFP-C1 (Clonetech) between BamH1

and Xba1 sites to generate the YFP±PK construct. Star frag-

ment 7 was re-ampli®ed with a BglII-linked 5 0 primer and

BamH1-linked 3 0 primer, and cloned into YFP±PK between

BglII and BamH1 sites to generate the YFP±Star7±PK

construct. HeLa cells (ATCC) were maintained in DMEM

media (Gibco) plus 10% FCS at 378C with 5% CO2 and

transfected with Fugene 6 (Roche) according to the manu-

facturer's procedure, then incubated overnight. Cells were

®xed in 4% paraformaldehyde and mounted in vectashield

(Molecular Probes).

Acknowledgements

This work was funded by a grant from the US National

Science Foundation (IBN-9807892). We thank N. Arnheim,

U. Banerjee, V. Fuandez, K. Hagedorn, A. Hodel, M. Stall-

cup, J. Tower and R. Warrior for advice and discussion and

B.-Z. Shilo for Drosophila stocks.

References

Aaronson, S.A., 1991. Growth factors and cancer. Science 254, 1146±1153.

Bang, A.G., Kintner, C., 2000. Rhomboid and Star facilitate presentation

and processing of the Drosophila TGF-alpha homolog spitz. Genes

Dev. 14, 177±186.

Bermingham-McDonogh, O., McCabe, K.L., Reh, T.A., 1996. Effects of

GGF/neuregulins on neuronal survival and neurite outgrowth correlate

with erbB2/neu expression in developing rat retina. Development 122,

1427±1438.

Bier, E., 2000. Drawing lines in the Drosophila wing: initiation of wing

vein development. Curr. Opin. Genet. Dev. 10, 393±398.

Bier, E., Jan, L.Y., Jan, Y.N., 1990. rhomboid, a gene required for dorso-

ventral axis establishment and peripheral nervous system development

in Drosophila melanogaster. Genes Dev. 4, 190±203.

Carpenter, G., 2000. The EGF receptor: a nexus for traf®cking and signal-

ing. BioEssays 22, 697±707.

Chen, C.-K., Chien, C.-T., 1999. Negative regulation of atonal in proneural

cluster formation of Drosophila R8 photoreceptors. Proc. Natl. Acad.

Sci. 96, 5055±5060.

Dai, H., Hogan, C., Gopalakrishnan, B., Torres-Vazquez, J., Nguyen, M.,

Park, S., Raftery, L.A., Warrior, R., Arora, K., 2000. The zinc ®nger

protein Schnurri acts as a Smad partner in mediating the transcriptional

response to decapentaplegic. Dev. Biol. 227, 373±387.

Derynck, R., 1992. The physiology of transforming growth factor-a. Adv.

Cancer Res. 58, 27±52.

Dickson, B.J., 1998. Photoreceptor development: breaking down the

barriers. Curr. Biol. 8, R90±R92.

Dominguez, M., Wasserman, J.D., Freeman, M., 1998. Multiple functions

for the EGF receptor in Drosophila eye development. Curr. Biol. 8,

1039±1048.

Freeman, M., 1994. The spitz gene is required for photoreceptor determina-

tion in the Drosophila eye where it interacts with the EGF receptor.

Mech. Dev. 48, 25±33.

Freeman, M., 1996. Reiterative use of the EGF receptor triggers differen-

tiation of all cell types in the Drosophila eye. Cell 87, 651±660.

Freeman, M., 1998. Complexity of EGF receptor signaling revealed in

Drosophila. Curr. Opin. Genet. Dev. 8, 407±411.

Freeman, M., Kimmel, B.E., Rubin, G.M., 1992. Identifying targets of the

rough homeobox gene of Drosophila: evidence that rhomboid functions

in eye development. Development 116, 335±346.

Golembo, M., Raz, E., Shilo, B.-Z., 1996. The Drosophila embryonic

midline is the site of Spitz processing, and induces activation of the

EGF receptor in the ventral ectoderm. Development 122, 3363±3370.

Guichard, A., Biehs, B., Sturtevant, M.A., Wickline, L., Chacko, J.,

Howard, K., Bier, E., 1999. rhomboid and Star interact synergistically


to promote EGFR/MAPK signaling during Drosophila wing vein devel-

opment. Development 126, 2663±2676.

Harris, W.A., Stark, W.S., Walker, J.A., 1976. Genetic dissection of the

photoreceptor system in the compound eye of Drosophila melanoga-

ster. J. Physiol. 256, 415±439.

Heberlein, U., Rubin, G.M., 1991. Star is required in a subset of photo-

receptor cells in the developing Drosophila retina and displays dosage

sensitive interactions with rough. Dev. Biol. 144, 353±361.

Heberlein, U., Hariharan, I.K., Rubin, G.M., 1993. Star is required for

neuronal differentiation in the Drosophila retina and displays dosage-

sensitive interactions with Ras 1. Dev. Biol. 160, 51±63.

Huang, Z., Shilo, B.Z., Kunes, S., 1998. A retinal axon fascicle uses spitz,

an EGF receptor ligand, to construct a synaptic cartridge in the brain of

Drosophila. Cell 95, 693±703.

Jarman, A.P., Grell, E.H., Ackerman, L., Jan, L.Y., Jan, Y.N., 1994. atonal

is the proneural gene for Drosophila photoreceptors. Nature 369, 398±

400.

Kalderon, D., Richardson, W.D., Markham, A.F., Smith, A.E., 1984a.

Sequence requirements for nuclear location of simian virus 40 large-

T antigen. Nature 311, 33±38.

Kalderon, D., Roberts, B.L., Richardson, W.D., Smith, A.E., 1984b. A short

amino acid sequence able to specify nuclear location. Cell 39, 499±509.

Kolodkin, A.L., Pickup, A.T., Lin, D.M., Goddman, C.S., Banerjee, U.,

1994. Characterization of Star and its interactions with sevenless and

EGF receptor during photoreceptor cell development in Drosophila.

Development 120, 1731±1745.

KraÈmer, H., Cagan, R.L., Zipursky, S.L., 1991. Interaction of bride of

sevenless membrane-bound ligand and the sevenless tyrosine-kinase

receptor. Nature 352, 207±212.

Kumar, J.P., Moses, K., 2000. Cell fate speci®cation in the Drosophila

retina. In: Fini, E. (Ed.). Vertebrate Eye Development. Results and

Problems in Cell Differentiation, Springer, Berlin, pp. 93±114.

Kumar, J.P., Moses, K., 2001. EGF receptor and notch signaling act

upstream of eyeless/Pax6 to control eye speci®cation. Cell 104 (in

press).

Kumar, J.P., Tio, M., Hsiung, F., Akopyan, S., Gabay, L., Seger, R., Shilo,

B.-Z., Moses, K., 1998. Dissecting the roles of the Drosophila EGF

receptor in eye development and MAP kinase activation. Development

125, 3875±3885.

MassagueÂ, J., 1990. Transforming growth factor-a. J. Biol. Chem. 265,

21393±21396.

Mayer, U., NuÈsslein-Volhard, C., 1988. A group of genes required for

pattern formation in the ventral ectoderm of the Drosophila embryo.

Genes Dev. 2, 1496±1511.

Moses, K., Ellis, M.C., Rubin, G.M., 1989. The glass gene encodes a zinc-

®nger protein required by Drosophila photoreceptor cells. Nature 340,

531±536.

Neumann, C.J., NuÈsslein-Volhard, C., 2000. Patterning of the zebra®sh

retina by a wave of sonic hedgehog activity. Science 289, 2137±2139.

Nilson, L.A., SchuÈpbach, T., 1999. EGF receptor signaling in Drosophila

oogenesis. Curr. Top. Dev. Biol. 44, 203±243.

O'Keefe, L., Dougan, S.T., Gabay, L., Raz, E., Shilo, B.Z., DiNardo, S.,

1997. Spitz, Wingless, emanating from distinct borders, cooperate to

establish cell fate across the Engrailed domain in the Drosophila epider-

mis. Development 124, 4837±4845.

Pickup, A.T., Banerjee, U., 1999. The role of Star in the production of an

activated ligand for the EGF receptor signaling pathway. Dev. Biol.

205, 254±259.

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

by Decapentaplegic. Development 124, 271±278.

Queenan, A.M., Ghabrial, A., SchuÈpbach, T., 1997. Ectopic activation of

torpedo/Egfr, a Drosophila receptor tyrosine kinase, dorsalizes both the

eggshell and the embryo. Development 124, 3871±3880.

Reh, T.A., Cagan, R.L., 1994. Intrinsic and extrinsic signals in the devel-

oping vertebrate and ¯y eyes: viewing vertebrate and invertebrate eyes

in the same light. Perspect. Dev. Neurobiol. 2, 183±190.

Rutledge, B.J., Zhang, K., Bier, E., Jan, Y.N., Perrimon, N., 1992. The

Drosophila spitz gene encodes a putative EGF-like growth factor

involved in dorsal±ventral axis formation and neurogenesis. Genes

Dev. 6, 1503±1517.

Sapir, A., Schweitzer, R., Shilo, B.-Z., 1998. Sequential activation of the

EGF receptor pathway during Drosophila oogenesis establishes the

dorsoventral axis. Development 125, 191±200.

Sawamoto, K., Okano, H., 1996. Cell±cell interactions during neural devel-

opment: multiple types of lateral inhibitions involved in Drosophila eye

development. Neurosci. Res. 26, 205±214.

Schlessinger, J., 2000. Cell signaling by receptor tyrosine kinases. Cell 103,

211±225.

Schweitzer, R., Shilo, B.-Z., 1997. A thousand and one roles for the Droso-

phila EGF receptor. Trends Genetics 13, 191±196.

Schweitzer, R., Shaharabany, M., Seger, R., Shilo, B.-Z., 1995. Secreted

Spitz triggers the DER signaling pathway and is a limiting component

in embryonic ventral ectoderm determination. Genes Dev. 9, 1518±

1529.

Spencer, S.A., Powell, P.A., Miller, D.T., Cagan, R.L., 1998. Regulation of

EGF receptor signaling establishes pattern across the developing

Drosophila retina. Development 125, 4777±4790.

Stemerdink, C., Jacobs, J.R., 1997. Argos and Spitz group genes function to

regulate midline glial cell number in Drosophila embryos. Develop-

ment 124, 3787±3796.

Sturtevant, M.A., Bier, E., 1995. Analysis of the genetic hierarchy guiding

wing vein development in Drosophila. Development 121, 785±801.

Sturtevant, M.A., Roark, M., Bier, E., 1993. The Drosophila rhomboid gene

mediates the localized formation of wing veins and interacts genetically

with components of the EGF-R signaling pathway. Genes Dev. 7, 961±

973.

Sturtevant, M.A., Roark, M., O'Neill, J.W., Biehs, B., Colley, N., Bier, E.,

1996. The Drosophila rhomboid protein is concentrated in patches at

the apical cell surface. Dev. Biol. 174, 298±309.

Szuts, D., Freeman, M., Bienz, M., 1997. Antagonism between EGFR and

Wingless signaling in the larval cuticle of Drosophila. Development

124, 3209±3219.

Tio, M., Moses, K., 1997. The Drosophila TGFa homolog Spitz acts in

photoreceptor recruitment in the developing retina. Development 124,

343±351.

Tio, M., Ma, C., Moses, K., 1994. spitz, a Drosophila homolog of trans-

forming growth factor-a, is required in the founding photoreceptor cells

of the compound eye facets. Mech. Dev. 48, 13±23.

Tomlinson, A., Ready, D.F., 1987. Neuronal differentiation in the Droso-

phila ommatidium. Dev. Biol. 120, 366±376.

Wasserman, J.D., Freeman, M., 1998. An autoregulatory cascade of EGF

receptor signaling patterns the Drosophila egg. Cell 95, 355±364.

Wasserman, J.D., Urban, S., Freeman, M., 2000. A family of rhomboid-like

genes: Drosophila rhomboid-1 and roughoid/rhomboid-3 cooperate to

activate EGF receptor signaling. Genes Dev. 14, 1651±1663.

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

adult Drosophila tissues. Development 117, 1223±1237.

Yang, L., Baker, N.E., 2001. Role of the EGFR/Ras/Raf pathway in speci-

®cation of photoreceptor cells in the Drosophila retina. Development

128, 1183±1191.


function of the drosophila tgf-a homolog spitz is controlled by star

Documents