Developmental Cell 10, 635–646, May, 2006 ª2006 Elsevier Inc.DOI 10.1016/j.devcel.2006.03.003
A C-Terminal Motif Targets Hedgehog to Axons,
Coordinating Assembly of the Drosophila
Eye and Brain
Tehyen Chu,1Michael Chiu,1Elisa Zhang,1
and Sam Kunes1,*
1Department of Molecular and Cellular Biology
Cambridge, Massachusetts 02138
The developmental signal Hedgehog is distributed to
two receptive fields by the photoreceptor neurons of
the developing Drosophila retina. Delivery to the retina
field.Transport along photoreceptoraxons inducesthe
development of postsynaptic neurons in the brain.
Here, we showthatthe N-terminal domain is targetedto
the retinawhen synthesized inthe absenceofthe C-ter-
minal domain. In contrast to studies that have focused
on cholesterol as a determinant of subcellular localiza-
tion, we find that the C-terminal domain harbors a con-
served motif that overrides retinal localization, sending
most of the autocleavage products into vesicles bound
for growth cones or synapses. Competition between
targeting signals at the opposite ends of Hedgehog
apparently controls the match between eye and brain
The precise spatial and temporal delivery of a morpho-
genic signal is critical to its activity in a receptive field.
Toachievethisprecision, amorphogen maybereleased
from a particular subcellular location, such as the apical
or basal pole of a cell, or from an appendage, such as
cytoneme or axon (Huang and Kunes, 1996; Ramirez-
Weber and Kornberg, 1999; Cho et al., 2000; Gibson
and Schubiger, 2000). From a specialized point of re-
lease, a morphogen encounters unique environments
that determine its distribution and shape its develop-
mental activity. This is the case for the morphogen
Hedgehog (Huang and Kunes, 1996; Burke et al., 1999;
Ma et al., 2002; Gallet et al., 2003), but how the protein
Hedgehog is an unusual protein, composed of N-ter-
minal signaling and C-terminal protease domains that
dissociate in an autocatalyzed cleavage reaction. The
protease domain covalently attaches cholesterol to the
N-terminal polypeptide fragment (denoted Hh-Np),
which harbors signaling activity (Lee et al., 1994; Porter
et al., 1995, 1996a; Mann and Beachy, 2004). Little is
known of where in the cell this reaction occurs or how
it is regulated. Yet the reaction and its consequences
nant form ofHh, consisting ofonly the N-terminal signal-
ing domain (denoted Hh-Nu), is synthesized without
autocleavage and cholesterol attachment (Porter et al.,
1996b; Burke et al., 1999). This aberrant Hh isoform has
properties distinct from Hh-Np. These include reduced
signaling activity (Porter et al., 1996b) and indepen-
dence from the transporter-like protein Dispatched
and extracellular proteoglycans (Bellaiche et al., 1998;
The et al., 1999; Burke et al., 1999; Ma et al., 2002; Tian
et al., 2005). In the epidermis of the Drosophila embryo,
the wild-type patterning activity of Hh-Np is directed to
both anterior and posterior receptive cells; the choles-
terol-negative Hh-Nu displays asymmetrical activity di-
rected mainly to the posterior (Gallet et al., 2003). Thus,
Hh morphogenic activity depends on the peculiarities of
its intracellular sorting and subcellular release.
In the developing retina, Hedgehog is synthesized by
photoreceptor neurons and distributed to two receptive
cellpopulations tocoordinate andcontrol theirdevelop-
ment. Release into the retina propagates a wave of om-
matidial differentiation, while transport along the photo-
receptor axons into the brain induces the first steps of
postsynaptic neuron differentiation (Figure 1) (Huang
and Kunes, 1996; reviewed by [Kunes, 2000]). How is
Hh partitioned for release from opposite poles of the
photoreceptor neuron? We show that the cholesterol-
negative Hh-Nu remains almost exclusively in the retina.
Targeting to axons, growth cones, and synapses relies
on a conserved motif at the Hh C terminus that sends
the autocleavage products into axon bound vesicles.
Loss of this targeting motif renders Hh deficient in post-
synaptic signaling. Thus, a competition between signals
at the opposite ends of Hedgehog controls the relative
induction of Drosophila eye and brain development.
A Role for the Hedgehog C Terminus
in Axon Targeting
Hedgehog (Hh) is synthesized by photoreceptor neu-
rons as they form nascent ommatidial clusters in the de-
veloping retina (Figure 1) (Lee et al., 1992). As they begin
to differentiate, the neurons deliver Hh to the anterior
disc epithelium, triggering further ommatidial develop-
ment, and transport Hh along their axons to initiate the
development of lamina postsynaptic neurons in the
brain (Huang and Kunes, 1996). Hedgehog is found con-
centrated in the photoreceptor cell bodies, especially in
their axons in the optic stalk and growth cones in the
To conveniently assess the cellular distribution of
wild-type and mutant Hh isoforms, we monitored the
distribution of transgene-encoded variants tagged by
the hemaglutinin repeat (HA; Figures 1B and 1D). We uti-
lized a Hh variant with an HA tag inserted just upstream
of the autocleavage site (Hh-FNHA; NHA for N-terminal
HA) (Burke et al., 1999). This protein undergoes normal
terol-modified N-terminal fragment (Hh-Np) with wild-
type properties of tissue distribution and signaling
activity (Figure 2A) (Burke et al., 1999). Animals bearing
the hh1mutation lack eye and lamina development;
this phenotype was rescued by the eye-specific expres-
sion of Hh-FNHA(data not shown). When expressed in
photoreceptor neurons, Hh-FNHAwas found in the photo-
receptor cell bodies, axons, and growth cones (Figures
1B, 1D, and 2A), like endogenous Hh (see Figure 6E).
Studies of Hh trafficking have utilized a truncated hh
transgene (hh-NNHA) that produces HA-tagged Hh-Nu,
a polypeptide equivalent to Hh-Np but for the choles-
terol moiety added by autocleavage (Figure 2B) (Burke
et al., 1999; Lewis et al., 2001; Gallet et al., 2003).
When hh-NNHAwas expressed specifically in the eye,
and was nearly absent from the photoreceptor axons
and growth cones (Figures 2B and 2F). Hh-Nu was con-
centrated in puncta like those seen with Hh-Np (Fig-
ure 2B00). In contrast, a fusion protein composed of the
Hh secretory signal sequence (amino acids 63–83) and
EGFP (ssGFP) was distributed uniformly along the pho-
toreceptor cell membrane, including the axons and
growth cones (Figure 4B). These observations are con-
sistent with the notion that a signal for apical (retinal)
localization is located in the N-terminal domain.
In order to quantify the relative distribution of Hh iso-
forms in cell bodies, axons, and growth cones, the hh
transgenes were expressed under control of the neu-
ron-specific elav-GAL4 driver (Luo et al., 1994), which
restricted expression in the retina to the photoreceptor
neurons. Specimens were coordinately stained and im-
aged by confocal microscopy under quantitative condi-
tions (Figure 2) (see Experimental Procedures). Average
pixel intensities were determined in selected areas
(Figure 2F, inset) and were expressed as ratios between
retina and optic stalk (retina/os) or R1–R6 growth cones
and optic stalk (gc/os). The quantitative analysis
(Figure 2F) revealed that Hh-Nu (the hh-NNHAproduct)
was 7-fold more concentrated in the retina relative to
the optic stalk than was Hh-Np. The Hh-Nu protein
was 2-fold less concentrated in the growth cones
relative to the optic stalk than was Hh-Np. There was
thus a dramatic reduction of Hh-Nu at growth cones
relative to the wild-type Hh product.
A possible explanation for the failure of Hh-Nu to lo-
calize to axons and growth cones would be the absence
of the cholesterol moiety, as has been proposed to
explain the distinct properties of Hh-Nu. However, it is
also the case that Hh-Nu is produced from a truncated
mRNA and is never joined to the C-terminal domain.
We explored these other possible influences on Hh lo-
calization with an additional pair of transgenes. One
contained the complete hh cDNA with stop codons in-
troduced at the autocleavage site (hh-SNHA; Figure 2C).
This transgene produced Hh-Nu that was, in principle,
the same as the hh-NNHAproduct, but from an essen-
tially normal mRNA. The second construct contained
the full hh cDNA with a Cys-to-Ala mutation at the auto-
cleavage site (hh-UNHA; Figure 2D). This transgene en-
codes a full-length polypeptide that does not undergo
autocleavage or acquire a cholesterol moiety (Porter
et al., 1995). The production of stable full-length Hh by
hh-UNHAand a truncated Hh-Nu polypeptide by hh-SNHA
was confirmed by Western analysis (Figure 5C; data
In photoreceptor neurons, both transgenes produced
Hh distributions distinct from the Hh-Nu encoded by
hh-NNHA(Figures 2C, 2D, and 2F). The Hh-Nu encoded
by hh-SNHAhad greater localization to axons than the
product of hh-NNHA(Figures 2B0and 2C0), but consider-
ably less than that of wild-type Hh-Np; the localization
to growth cones was particularly poor (Figures 2C00and
2F). In contrast, the cleavage site mutant isoform en-
coded by hh-UNHAwas distributed like the wild-type
Hh-Np (Figures 2D0and 2F). Indeed, colocalization anal-
ization of Hh-UNHAand Hh-FNHAon axons and growth
cones (Figures 2G–2J and 3). Thus, the 30region of the
but a more robust determinant of growth cone localiza-
tion apparently resides in the Hh C-terminal domain.
Figure 1. Hedgehog Localization in Photore-
ceptor Neurons of the Developing Visual
(A and B) The visual system shown from the
lateral perspective. Ommatidial clusters of
photoreceptor neurons differentiate in a tem-
poral order, following the progression of the
rior (right to left) across the eye disc (Eye) ep-
ithelium. The neurons send their axons as an
ommatidial fascicle (the R1–R8 axons) into
the brain via the optic stalk, an epithelial
tube that connects the developing eye and
brain. The axons spread to dorsal and ventral
retinotopic positions (dorsal is up). The R1–
R6 axons terminate in the lamina (lam), while
R7 and R8 axons terminate in the deeper me-
dulla ganglion (see [C]). In the micrograph in
(B), the distribution of HA-tagged Hh protein,
expressed specifically in the retina from the
transgene UAS-hh-FNHAand the GMR-GAL4
driver, is visualized with anti-HA antibody staining. Hh is sequestered in puncta in the retinal cell bodies, axons in the optic stalk, and growth
cones in the brain.
(C and D) The visual system from the horizontal perspective. Hh secreted from developing photoreceptor neurons is required for both eye and
lamina development (arrows in [C]). In this horizontal view, Hh (UAS-hh-FNHA), expressed as in (B), is found to be highly concentrated at R1–R6
growth cones (R1-6 ter. in [C]; R1-6 in [D]) in the lamina (lam) and the R7/R8 axon termini (R7/8 ter. in [C]; R7/8 in [D]) in the medulla.
The Hh C Terminus Is Concentrated at
the Growth Cones
If the Hh C-terminal domain carries an axon-targeting
signal, it might be particularly well localized to axons
and growth cones. To visualize the disposition of the C
terminus, several HA insertion constructs were made
in which the tag was placed in the C-terminal domain
at sites chosen to avoid interference with autocleavage
(Figure 2E) (see Experimental Procedures). Western
tions at three different sites, which produced an HA-la-
beled C-terminal product (Figure 5C). When expressed
inphotoreceptorneurons ofthedeveloping retina,these
C-tagged isoforms (Hh-FCHA; CHA for C-terminal HA)
Figure 2. Distinct Localization Properties of the Hh N-Terminal and C-Terminal Domains
(A–E00) For each transgenic construct depicted in the top row (A–E), corresponding late third instar-stage specimens are shown stained for HA
immunoreactivity from (A0–E0) lateral and (A00–E00) horizontal perspectives in the columns beneath. (A–E) The yellow box indicates the position of
the HA tag (NHA, N-terminal domain, red color; CHA, C-terminal domain, gray color). In (A) and (E), the autocleavage site is indicated by ‘‘Cut,’’
while the mutant cleavage is indicated in (D) (C–A). (A0–E0) The indicated hh transgenes were driven specifically in the retina by GMR-GAL4. Anti-
HA immunoreactivity is shown in all panels at a focal plane that permits simultaneous visualization of ommatidia in the retina (ed), photoreceptor
axons in the optic stalk (os), and axons spreading to dorsoventral locations in the lamina. In (E0), a slightly deeper focal plane is shown in order to
reveal the concentrated HA immunoreactivity in the R1–R6 growth cones (gc) observed for Hh-FCHA. (A00–E00) The indicated hh transgenes were
driven in all neurons by elav-GAL4. The UAS-CD8::GFP transgene was also present to permit visualization of neuronal landmarks (not shown),
including the position ofthe R1–R6growth cones (gc, yellowarrowheads).Expression of HA-taggedHhin laminaneuronsisdistinctly positioned
from the growth cones and is marked if evident (lam, red arrowhead).
(F) Quantification of Hh isoforms distributed to the retina, axon, and growth cone. Micrographs in (A00)–(E00) were collected from 6–14 specimens
for each transgene. Average pixel intensities were calculated from selected regions (red outline in inset): retina, optic stalk, and growth cone. To
normalize, ratios were calculated for retina versus optic stalk (retina/os) and growth cone versus optic stalk (gc/os). The bar graph reports the
average ratio for each transgene shown. Error bars indicate SEM.
(G–J) High-magnification images of the (G and H) optic stalk and the (I and J) R1–R6 growth cone layer (R1-R6 gc) of the lamina for animals ex-
pressing (G and I) UAS-hh-FNHAor (H and J) UAS-hh-UNHAunder GMR-GAL4 control. The wild-type HA-tagged protein Hh-FNHAis localized to
puncta, similar to the puncta observed with the cleavage site mutant Hh-UNHA.
The scale bar in (A) represents 20 mm for (A)–(E), the scale bar in (G) represents 2 mm for (G) and (H), and the scale bar in (I) represents 10 mm for (I)
A C-Terminal Motif Targets Hedgehog to Axons
were found to be strongly concentrated at the photore-
ceptor growth cones (Figures 2E0, 2E00, and 2F). Notably,
the tagged C terminus displayed a 5-fold greater con-
centration at growth cones relative to axons (3.0 gc/os)
than did HA-tagged Hh-Np (0.60 gc/os; Hh-FNHA).
In axons and growth cones, the C-tagged protein dis-
played a punctate distribution overlapping the N-termi-
nal domain (see Figure 6). Hence, the Hh C terminus is
tion of the N-terminal domain remains in the retina.
Figure 3. Hh Transport in Synaptotagmin-Containing Vesicles
(A–F0) Horizontal views of the late third instar visual system in animals expressing UAS-synaptotagmin::eGFP (Syt::GFP) along with either (A, C,
and D) UAS-hh-FNHAor (B, E, and F) UAS-hh-UNHAunder the control of the eye-specific GMR-GAL4 driver. Both the Hh-FNHAand Hh-UNHAprod-
ucts are found in puncta (red color in [A] and [B]; shown alone in [C], [D], [E], and [F]) that extensively overlap Syt::GFP-positive puncta (green
color in [A] and [B]; shown alone in [C0], [D0], [E0], and [F0]). High-magnification views of the specimens in (A) and (B) are shown in (C)–(F). Yellow
arrowheads in (C)–(F) highlight puncta with overlapping localization of Hh and Syt::GFP.
(G and H) A fusion protein composed of the Dpp signal sequence and GFP (sGFP) is restricted to the photoreceptor cell bodies, absent from
photoreceptor axons, and weakly detected at the R1–R6 growth cones (green color in [G] and [H]; shown alone in [G00] and [H00]). In the retina
([H], high-magnification view), sGFP is colocalized (arrowheads) with Hh-FNHAimmunoreactivity (red color in [G] and [H]; shown alone in [G0]
(I) When expressed in the retina of a young adult under control of the GMR-GAL4 driver, Hh-FNHA(red color in [I], shown alone in [I0]) is concen-
trated at the presynaptic terminals of R1–R6 photoreceptors in lamina cartridges (the area between the horizontal bars). Hh is colocalized with
the synaptic vesicle-associated Cysteine String Protein (green color in [I], shown alone in [I00]).
The scale bar in (A) represents 15 mm for (A), (B), and (G), and the scale bar in (C) represents 10 mm for (C), (E), (H), and (I).
Hedgehog Is Transported in Synaptotagmin-Marked
To gain insight into the nature of the Hh-containing
puncta, we examined their colocalization with secretory
pathway markers and other proteins localized to axons
and growth cones (Figure 3), including proteins with
functions in vesicle trafficking (Clathrin::GFP) (Chang
et al. 2002) and membrane proteins (Fasciclin II) (Davis
et al. 1997). Robust colocalization was found in only
three cases. Two were presynaptic vesicle proteins,
Synaptotagmin (Syt) (Perin et al., 1990) and Cysteine
String Protein (CSP) (Zinsmaier et al., 1990). A Syt::GFP
fusion protein (Zhang et al., 2002) labeled puncta in the
photoreceptor cell bodies, axons, and growth cones.
In larval motor neurons, Syt::GFP puncta displayed
properties of transported vesicles (Miller et al., 2005).
In photoreceptor axons, the Syt::GFP fluorescent
puncta robustly overlapped thepuncta ofcoexpresssed
Hh-FNHA(Figures 3A, 3C, and 3D). A similar colocaliza-
tion was observed in adult animals, in which the photo-
receptor axon termini had developed into presynaptic
terminals (data not shown). In these terminals, Hh-FNHA
colocalized with the synaptic vesicle protein CSP
(Figure 3I). The cleavage site mutant, Hh-UNHA, dis-
played a similar colocalization with Syt::GFP (Figures
3B,3E,and3F).Thus, irrespective ofwhether autocleav-
age occurred, Hh was transported to growth cones and
presynaptic terminals in association with vesicle pro-
teins involved in neurotransmitter release.
It has been suggested that the cholesterol-modified
Hh-Np may be transported in association with lipid rafts
(Rietveld et al., 1999), membrane subdomains rich in
cholesterol and proteins bearing glycosylphosphoinosi-
tol (GPI) modification. However, we did not observe ro-
bust colocalization of Hh-Np with the lipid raft marker
UAS-GFP-gpi (data not shown) (Greco et al., 2001). A
lack of Hh colocalization with this lipid raft marker has
also been observed in embryonic epithelia (Gallet
et al., 2003). A protein containing a portion of the mor-
phogen Decapentaplegic (DPP) fused to GFP (denoted
sGFP) (Entchev et al., 2000) displayed colocalization
with Hh-positive puncta early in the secretory pathway
(Gallet et al., 2003), but not at the later stages of delivery
to apical or basal locales. We likewise found that sGFP
was localized to the photoreceptor cell bodies in the ret-
ina, where it colabeled Hh-Np-positive puncta (Figures
3G and 3H). Thus, the secretory mechanisms for Hh dis-
tribution in photoreceptor neurons appear to be similar
to those in other cell types, at least in the early steps
of the pathway.
Localization of a C-Terminal Axon-Targeting Signal
To localize polypeptide sequences that target Hh to
growth cones, we performed a deletion analysis that re-
moved various extents of the C terminus. Nonsense co-
dons were introduced at several locations in the con-
struct that encodes the cleavage site mutant, Hh-UNHA
(Figure 4A),which was used inorder toavoid variable ef-
fects of autocleavage efficiency with different deletions.
As shown in Figure 4A, all of the deletions reduced Hh
axon and growth cone targeting to a level well below
that of the full-length Hh-UNHA. Notably, these deletion
isoforms displayed apical membrane localization like
Hh-Nu, the polypeptide produced by a nonsense codon
at the autocleavage site (Hh-SNHA, top panels).
ficient for axonal targeting, it was fused to a heterolo-
gous protein, EGFP. The Hh secretory signal sequence
(amino acids 63–83) was fused to the N terminus of
EGFP to create ‘‘ssGFP’’ (Figure 4B). Either the full Hh
C-terminal domain (amino acids 258–471) or the most
C-terminal 30 amino acids (442–471) was added to the
ssGFP fusion protein. The Hh secretory signal alone
conferred uniform membrane localization to cell bodies,
axons, and growth cones (Figure 4B, top panel). When
either the full Hh C terminus or the C-terminal 30 amino
acids was present, the fusion protein was localized to
puncta in axons and growth cones, like Hh-Np
(Figure 4B, middle and bottom panels). The complete
C-terminal domain was more effective in growth cone
targeting and sequestration into puncta than the C-ter-
minal 30 amino acids alone (gc/os = 3.4 versus gc/os =
2.1, respectively; Figure 4B, right panels; Figure 4C). In
sum, a signal at the Hh C terminus appears necessary
and sufficient for at least the partial sequestration of
Hh into growth cone-targeted vesicles, but full localiza-
tion activity required the entire C-terminal domain.
A Targeting Motif at the C Terminus
Hh is a widely conserved protein, with family members
known from fly to human. Alignment of this family re-
vealed a nearly invariant motif, G*HWY (where ‘‘*’’ is I,
M, or V), and well-conserved adjacent residues within
the C-terminal 30 amino acid region (Figure 5A). We
reckoned that a tyrosine in this motif, Y452, might be
a critical residue, perhaps a target for regulatory phos-
phorylation. To test this notion, the tyrosine was re-
placed by alanine in the cleavable Hh isoform, Hh-F,
taggedwith HA ineither theN-terminal orC-terminal do-
mains (Hh-FNHA-Y452A and Hh-FCHA-Y452A). These
Y452 mutant proteins underwent efficient autocleavage
when expressed in the retina, producing N-terminal or
C-terminal HA-tagged fragments of an appropriate size
(Figure 5C). When examined for their distribution in the
developing visual system, the Y452 mutants were found
to be concentrated in the photoreceptor cell bodies,
with little distribution to axons or growth cones (Figures
5B and 5D). Moreover, the Y452 mutant C-terminal do-
main was retained within the retina, rather than concen-
trated in growth cones like the wild-type C-terminal do-
main. A similar result was obtained when Y452 was
mutated to glutamic acid, an indication that phosphory-
lationmay not havea roleintheactivity oftheconserved
motif. These observations demonstrate that a signal at
the Hedgehog C terminus, rather than autocleavage
and cholesterol modification, is required for efficient
axonal targeting of Hh.
Autocleavage Precedes Axon Transport
How the C-terminal motif might function in axon target-
ing depends on whether autocleavage precedes or fol-
lows axon transport. To address this question, the Hh
polypeptide was examined directly by Western analysis
of larval-stage photoreceptor cell bodies, axons, and
growth cones, which were surgically separated by sev-
ering the optic stalk connecting the retina and brain.
This was done with animals expressing Hh-FNHA
A C-Terminal Motif Targets Hedgehog to Axons
driver. With the intact larval eye-brain complex (Fig-
ure 6C), nearly all Hh was observed in the autocleaved
form. Similarly, when the retina and brain were sepa-
rated, nearly all Hh in either isolate was in the auto-
cleaved state (Figure 6C, lanes 1 and 2). Therefore, it
Figure 4. Localization of an Axon-Targeting Signal to the Hedgehog C Terminus
(A) The cleavage site mutant Hh-UNHAwas modified by the insertion of stop codons to create a set of C-terminal deletions, indicated graphically
above each pair of micrographs. The position of the HA tag is shown in yellow, the N-terminal domain is shown in red, and the C-terminal domain
is shown in gray. In the left column of micrographs, anti-HA antibody was used to visualize the axonal distribution of the tagged protein, which
was expressed specifically in the retina under GMR-GAL4 driver control. In the right column, anti-HRP staining of the same specimen is shown.
The largest deletion, Hh-SNHA, produces Hh-Nu (lacking the entire C-terminal domain; top panels). With no deletion (Hh-UNHA, bottom panels),
there is strong axonal localization of HA immunoreactivity. Even the shortest C-terminal deletion (Hh-UNHA441) is deficient in Hh localization to
axons and growth cones. Os, optic stalk. The scale bar in (A) is 20 mm.
(B)TheHhsignalsequence(AAs 63–83)wasfusedtothe EGFPN terminus tomakessGFP.Twoadditionalconstructs containeither theentire Hh
C-terminal domain (ssGFP-HhC) or the C-terminal 30 amino acids (ssGFP-Hh30) fused at the C terminus of ssGFP (shown schematically). When
expressed with the elav-GAL4 driver, either C-terminal Hh sequence enhanced the localization of GFP to puncta in axons (right panels, high-
magnification view of optic stalk) and growth cones (yellow arrowheads, left panels). The scale bar in (B) is 15 mm for the left panels and 2 mm
for the right panels.
(C) Fluorescence for specimens as in (B) was quantified (see Figure 2F). With the full C terminus present (ssGFP-HhC), GFP was considerably
more concentrated in the R1–R6 growth cones than with the ssGFP control. With only the Hh C-terminal 30 amino acids fused to ssGFP
(ssGFP-Hh30), GFP localization in the retina and growth cones was increased relative to the optic stalk. The C-terminal 30 amino acids were
less potent than the full C terminus in targeting Hh to axons and localizing the protein to axonal puncta (see [B]).
Figure 5. A Short C-Terminal Amino Acid Motif Mediates Hh Axonal Targeting
(A) A total of 20 Hh sequences from diverse species were aligned by using ClustalW; 8 are shown (zfish, zebrafish; Dhydei, Drosophila hydei;
in all cases.
(B) The role of tyrosine (Y) 452 in the G*HWY motif was explored by mutation to alanine or glutamine in cleavable Hh isoforms tagged with either
N-terminal or C-terminal HA insertions, as indicated schematically (yellow bar). These constructs were expressed, along with UAS-CD8::GFP, in
developingphotoreceptorcellswiththeelav-GAL4driver(top panels;GFP,green color;anti-HA,redcolor,shown aloneinbottompanels).When
Y452 was mutated to alanine (Y452A), the Hh N terminus was nearly absent from the photoreceptor axons and growth cones (left two panels, ar-
rowheads indicate the position of the growth cones). When the C-terminal domain was tagged by HA insertion (right two panels), the Y452A mu-
(C) Hh isoforms harboring the Y452A mutation undergo normal autocleavage. Adult retinas expressing the UAS-hh-FNHAY452A or UAS-hh-
FCHAY452A transgenes under control of the GMR-GAL4 driver were examined by Western analysis with anti-HA antibody. Two blots are shown.
On the left, the product of hh-FNHA(FNHA, left lane) is shown beside the product bearing the Y452A mutation (FNHAY-A, right lane). The N-terminal
HA-taggedproduct migrates atw26kDaandis efficiently formed inthe Y452Amutant.Onthe right is acomparison of theproductsof C-terminal
HA-tagged isoforms, with (FCHAY-A) and without (FCHA) the Y452A mutation. The C-terminal HA-tagged product migrates at w32 kDa and is
site mutant isoform (top band, above degradation products).
fold (1.7 versus 0.3, retina/os) difference in the relative distribution of the tagged Hh N terminus between retina and optic stalk, and a 3.9-fold
reduction in the amount found at the growth cone, relative to the retina. Error bars indicate the SEM.
A C-Terminal Motif Targets Hedgehog to Axons
Figure 6. The Hh Autocleavage Products Share a Transport Vehicle to the Brain
(A and B) The micrographs in (A) and (B) show a section of the optic stalk in the developing visual system, where Hh was visualized in the pho-
toreceptor axons. The scale bar in (A) is 2 mm and represents both (A) and (B). (A) Photoreceptor neurons that express Hh-FCHA(HA-tagged
C terminus) were stained with anti-HA (red color, [A]; alone in [A0]) and anti-Hh-N (green color, [A]; alone in [A00]), which recognizes the N-terminal
fragmentofHh.Thecorrelationplot(rightpanel)showsthatthelabeledNandCtermini punctaextensivelyoverlap,indicating thattheautocleav-
age products are typically transported in the same vehicle.
(B) When two tagged polypeptides were expressed from different transgenes, UAS-Hh-FNHAand UAS-ssGFP-HhC, the HA-tagged N terminus
(red color in [B], alone in [B0]) was located in puncta that usually harbored the C-terminal fusion to GFP (green color in [B], alone in [B00]). A cor-
relation plot is shown in the right panel.
(C) Western analysis of Hh in the developing visual system. The eye-brain complex was isolated from third instar larvae expressing the Hh-FNHA
polypeptide under control of the GMR-GAL4 driver. In lanes 1 and 2, the eye-brain complex was dissected to separate the eye disc (FNHAed, lane
1) from the optic stalk and brain (FNHAbrain, lane 2). Lanes 3–5 display control isolates, including the eye-brain complex without transgene ex-
pression(ywe+b,lane3)andthe intacteye-braincomplexexpressingHh-FNHA(FNHAe+b,lane4).TheC-terminalautocleavageproduct istagged
in lane 5 (FCHAe+b) by retinal expression of the protein Hh-FCHA. Only a small fraction of material is detected as uncleaved precursor product;
most Hh in the axon and growth cones is in the form of autocleaved product. Control specimens in lanes 6 and 7 are, respectively, the N-terminal
appears that, at most, a minor fraction of the polypep-
tide could be transported to growth cones prior to auto-
That autocleavage evidently occurs prior to axon
transport raises the question of whether the N- and
C-terminal domains travel down the axon together in
the same vesicles. To address this issue, both cleavage
products were simultaneously visualized. In animals ex-
pressing Hh-FCHAin photoreceptor neurons, the C-ter-
minal and N-terminal domains were respectively labeled
by anti-HA and anti-N-terminal domain antibodies (anti-
optic stalk and at the R1–R6 growth cones were exten-
sively colabeled, indicating that the autocleavage prod-
ucts were usually transported together (Figure 6A; cor-
relation R = 0.76, right panel). A significant fraction of
puncta did not appear, however, to contain both cleav-
age products. In a parallel approach, two proteins
were expressed from different transgenes—the Hh
C-terminal domain fusion to GFP (ssGFP-HhC; Fig-
ure 4B) and Hh-FNHA(HA-tag in the N terminus)—to
assess colocalization of the HA-tagged N terminus with
the GFP-labeled C-terminal domain. Most GFP-positive
puncta were HA positive (Figure 6B; R = 0.69, right
panel), but, as described above, some puncta appeared
to harbor one or the other protein. Thus, both of the
autocleavage products are usually transported along
the same pathway, but there may also be additional
minor pathways that transport the N- and C-terminal
A Role for the C-Terminal Targeting Motif in
Developmental Signaling to the Brain
terminal motifhasaroleintargeting Hhsignalingactivity
to the brain. In the first approach, the wild-type Hh-FCHA
and Y452 mutant, Hh-FCHA-Y452A, were examined for
rescue of the visual system-specific defect of hh1ani-
mals. In hh1animals, retinal development arrests after
assembly of the first 11–13 columns of ommatidia, and
lamina development fails to occur; the lamina neurons
(L1–L5) are absent (Huang and Kunes, 1996). We found
that transgenic rescue of the hh1phenotype depended
on the strength of the GAL4 driver. With the strong eye-
specific driver GMR-GAL4, both the wild-type and
Y452 mutant transgenes fully rescued eye and lamina
development (data not shown). On the other hand, with
the weaker driver eyeless-GAL4, only wild-type Hh-FCHA
rescued lamina development, while both the wild-type
and mutant transgenes produced a similar rescue of
eye development (Figure 6D). With the Y452 mutation,
(Figure 6D, right panel, blue color) but lacked the ability
to induce lamina develoment; the defect in Hh-FCHA-
of developmental signaling to the brain.
In the second approach, we examined chromosomal
hh alleles for differential retina and lamina signaling ac-
tivity. Among several hh hypomorphic alleles that par-
lele hh2displayed disproportionately poor rescue of
lamina development. As a homozygote, hh2has a weak
segment polarity phenotype (Perrimon and Mahowald,
1987). In combination with hh1(e.g., hh2/hh1), the adult
retina was indistinguishable from the wild-type in that it
displayed a similar number of ommatidial columns (Fig-
ure 6G, top panels). Lamina development was, however,
tected with an antibody against the homeodomain pro-
tein BSH (Jones and McGinnis, 1993), were absent (Fig-
L1 and L2 neurons, revealed by the marker 21D-GAL4,
dle panel, green color). In contrast, with the hhQ50allele
(Tashiro et al., 1993), which rescued eye development
in hhQ50/hh1animals (Figure 6G, bottom right panel). In
the third instar larval stage, hh2/hh1animals also dis-
played less than half the normal number of Dac-positive
lamina precursor cells, which are induced by axonal Hh
signaling (Figures 6E and 6F). Hh antigen accumulated
in the retina of these heterozygotes (anti-Hh staining,
Figure 6E). The hh2product was thus poorly targeted to
photoreceptor axon termini and was deficient in signal-
ing to the brain.
The hh2allele was cloned and sequenced. A stop co-
don was found at the position of amino acid 426, which
would delete the C-terminal 45 amino acids of the
and C-terminal HA-tagged autocleavage products isolated from adult heads in animals with retina-specific expression of the respective trans-
genes. Each lane contains material isolated either from 30–50 larvae or 4–6 adult heads.
(D) The Hh Y452A mutant is deficient in signaling to the brain. The transgenes UAS-hh-FCHAand UAS-hh-FCHA-Y452A were expressed in the ret-
ina under control of eyeless-GAL4 in the hh1background. As in hh1, few or no Dacshund-positive (green color) lamina precursor cells are present
withtheexpression ofHh-FCHA-Y452A (rightpanel),whilethewild-typeHh-FCHAinducesarelativelynormalpopulationofDac-positivecells(lam,
left panel). Both the wild-type and mutant Hh rescue eye development, as indicated by the similar ingrowth of photoreceptor axons (anti-HRP
antibody; blue color). White lines in the left panel indicate the joining of two micrographs of different focal planes. lob, lobula.
(E) A point mutation in hh2introduces a stop codon at amino acid position 426 (TGG to TGA). In the micrographs, photoreceptor cells and axons
were stained with anti-Hh antibody (red color), and lamina precursor cells were visualized with anti-Dac antibody (green color). In the hh2/hh1
specimen (right panel), Hh immunoreactivity is concentrated in the retina and is reduced on the photoreceptor axons (retina/optic stalk, wild-
type 0.90 versus hh2/hh10.70). The frequency of Dac-positive lamina precursor cells (green color) is reduced by about 2-fold (see analysis in
[F]). Dac-positive cells are also present in the adjacent lobula.
and imaged in serial confocal Z-sections. The total number of Dac-positive cells was enumerated from three-dimensional Dac-positive images
constructed by using Velocity. The hh2/hh1animals, on average, displayed less than half as many Dac-positive cells as the wild-type (wt, 580
versus hh2/hh1, 279). The left bar graph shows the total Dac-positive cell count, while the right bar graph shows the data normalized by the num-
ber of ommatidial columns (wt, 35 versus hh2/hh1, 18).
(G) Both the hh2(hh2/hh1) and hhQ50(hhQ50/hh1) alleles rescue the hh1eye phenotype to the wild-type ommatidial number (top panels). In the
lamina (lower panels), hh2/hh1displays less than 50% of the normal number of lamina neurons. Anti-BSH antibody staining (red color) reveals
that L4 and L5 neurons are absent in hh2/hh1. Labeling L1 and L2 with the marker 21D-GAL4, UAS-CD8::GFP (green color) reveals a reduction
to w40% of wild-type. In contrast, the lamina of hhQ50/hh1(lower right panel) is normal. The scale bar is 50 mm.
A C-Terminal Motif Targets Hedgehog to Axons
polypeptide, including the G*HWY targeting motif. It is
not clear whether the hh2product undergoes autocleav-
age.However, we note that autocleavage is not required
for Hh signaling activity, as the uncleavable isoform
Hh-U fully rescued both eye and lamina development
in the hh1background when expressed in the eye with
GMR-GAL4 (data not shown). Thus, a chromosomal
mutation that would delete the Hh C terminus yields a
predicted product deficient in axonal targeting and sig-
naling activity in the brain.
The photoreceptor neurons of the Drosophila retina pro-
vide Hedgehog to receptive cells in both the eye and
ceptive and processing circuit. Release into the retina
occurs via an apical pathway, while access to the brain
requires transport along axons, a basal pathway (Izad-
doost et al., 2002; Pellikka et al., 2002; Djiane et al.,
2005; Longley and Ready, 1995). One might suppose
that this pattern of release would arise by a mechanism
that distributes Hh generally in the cell. However, our
terminants at opposite ends of the polypeptide. Hh is
composed of N-terminal signaling and C-terminal prote-
ase domains that dissociate in an autoproteolytic reac-
tion that attaches cholesterol to the N-terminal signaling
fragment (Hh-Np) (reviewed in Mann and Beachy, 2004).
We found that the N-terminal product localized to the
retina when it was synthesized in the absence of the
C-terminal domain. This localization was not conferred
by the Hh secretory signal motif alone (Figure 4), and
hence must reflect the activity of a targeting signal
elsewhere in the N-terminal domain. Conversely, the
C-terminal domain, synthesized with or without the
N-terminal domain, was strongly localized to the growth
cones (Figures 2 and 4). Thus, the cellular distribution of
the N-terminal signaling domain would appear to reflect
opposing determinants; a targeting motif at the C termi-
nusdirectsmost oftheN-terminal cleavageproductinto
an axonal pathway (Figure 7).
The apical localization of the N-terminal domain was
observed with Hh-Nu, the product of a 30-truncated
explanations for this localization; that axonal targeting
requires: (1) the absent 30mRNA sequences, (2) a poly-
peptide signal in the C-terminal domain, or (3) the cho-
lesterol moiety added by Hh’s autocleavage. We found
that the 30region of the hh mRNA contributed modestly
to axonal localization (e.g., hh-SNHA; Figure 2) and was
relatively ineffective in mediating growth cone localiza-
tion (Figure 2C00). This region of mRNA may direct the
mRNA and its translation to sites that enhance mem-
brane localization of the product, as has been observed
with other morphogens (Wilkie and Davis, 2001; Sim-
monds et al., 2001). In contrast, indelibly appending
the C-terminal domain to the N terminus by mutating
the autocleavage site produced strong axonal and
growth conelocalization inpunctasharedwith Synapto-
tagmin, a fate like that of wild-type Hh (Figures 2 and 3).
Though we cannot preclude all other explanations for
the localization properties of these Hh isoforms, the
most straightforward is thatthe C-terminal domain sorts
Hh into a pathway for axonal transport.
The C-terminal axon-targeting signal was mapped to
a 30 amino acid region (Figure 4) where the Hh family
shares a well-conserved five amino acid motif (Figure 5).
When an invariant tyrosine in this motif (Y452) was
mutated to either phenylalanine or glutamic acid, auto-
cleavage appeared normal, but the two products re-
mained, for the most part, in the retina (Figure 5B).
When another well-conserved tyrosine residue outside
the motif at amino acid 457 was similarly mutated, there
was no effect on axon transport (unpublished data). The
N-terminal domain thus relies on a signal in the C-termi-
nal domain for its axonal localization, though autocleav-
age separates the two polypeptides prior to transport
(Figure 6). It is therefore evident that targeting the N-ter-
minal domain to axon terminals does not require auto-
cleavage or cholesterol modification.
A number of studies have examined the intracellular
trafficking, release, and extracellular movement of Hh
or its mammalian counterpart, Sonic Hedgehog. They
have relied on comparison between the N-terminal
product of Hh autocleavage, Hh-Np, and the choles-
terol-negative isoform, Hh-Nu, to draw conclusions on
the role of cholesterol in intracellular trafficking and
Figure 7. Distribution of Hedgehog to Retina
Targeting signals at the N and C termini com-
pete to target Hh to apical and basal destina-
tions, controlling the relative induction of eye
and lamina development. When liberated
from the N-terminal domain by autocleavage,
the C-terminal domain (orange color) is tar-
geted to the axonal pathway by the C-termi-
nal signal (yellow star) and is concentrated
at the photoreceptor axon terminals. Axonal
targeting evidently precedes autocleavage,
such that the N-terminal fragment can enter
the axonal pathway.
extracellular signaling activity. Our observations raise
doubt but do not directly address the role of cholesterol
modification in other developing tissues and organisms.
We moreover do not address the role of lipid modifica-
tion in Hh movement outside of the cell, where, for ex-
ample, cholesterol underlies the protein’s assembly
into extracellular lipoprotein transport particles (Pana-
kova et al., 2005).
In the brain, Hh acts over several cell diameters to ex-
pand the pool of lamina precursors in rough proportion
to the number of photoreceptor axon fascicles that ar-
nation between the differentiation programs of the eye
and brain and, eventually, a numerical match between
sensory axons and postsynaptic neurons. Is the parti-
tif necessary to produce this match? In support of this
view, we found that the C-terminal motif Y452 mutant
it is similar to wild-type Hh in the induction of retinal de-
velopment. The chromosomal mutant hh2was also
proficient for eye development and deficient in axonal
localization and lamina induction (Figure 6). The hh2al-
lele, unlike other alleles that affect the eye and lamina
equivalently, is predicted to yield a truncated product
lacking the C-terminal targeting motif (Figure 6). The
C-terminal motif is conserved in the mouse Sonic
Hedgehog (SHH; Figure 5), which regulates visual sys-
tem development via delivery from retinal ganglion cell
bodies and axons (Wallace and Raff, 1999; Dakubo
et al., 2003). Adult neuronal expression and anterograde
2001). Therefore, the C-terminal motif may have a con-
served role in localizing Hedgehog, thus regulating its
signaling activity in the development and function of
Strains and Reagents
The UAS-hhFNHAand UAS-hhNNHAtransgenes were described by
Burke et al. (1999). The following transgenes were used as secretory
pathway markers: UAS-Grasp65-GFP (Golgi complex), UAS-EGFP-
Clc, UAS-GFP-gpi (Greco et al., 2001), UAS-nsybGFP, UAS-
syt.eGFP, UAS-sGFP (Entchev et al., 2000), UAS-domeGFP (Ghi-
glione et al., 2002), UAS-Khc-lacZ. The 21D-GAL4 strain was kindly
provided by A. Keller and M. Heisenberg (Wu ¨rzburg). The following
hh alleles were examined in combination with hh1: hh2, hh3, hh4,
hh5, hh6, hh9, hh18, hh19, hh20, and hh21.
The UAS-hh-UNHAand UAS-hh-SNHAtransgenes were derived from
UAS-hh-FHA(pDA519) (Burke et al.,1999). For UAS-hh-UNHA, the co-
don for cysteine at position 258 was changed from ‘‘TGC’’ to ‘‘GCC.’’
For UAS-hh-SNHA, ‘‘TGC’’ for C258 was changed to ‘‘TGA,’’ a stop
codon. All deletion constructs were derived from UAS-hh-UNHA.
Two stop condons, ‘‘TGATGA,’’ were inserted at the site of the cor-
responding codon, as indicated in Figure 4A. The UAS-hh-FCHAcon-
structs were derived from hh cDNA clone 11 (Lee et al., 1992). Using
the crystal structure (Hall et al., 1997) as a guide, a 33HA tag was in-
serted into the C-terminal domain at the following amino acid (AA)
positions: UAS-hh-FC1HA, AA 337; UAS-hh-FC2HA, AA 361; UAS-hh-
FC3HA, AA 268. To construct GFP fusion proteins, the Hedgehog
signal sequence (AA 63–83) was fused to EGFP in the pEGFP-C2
vector. In UAS-ssGFP-Hh30, the last 30 amino acids of Hh (AA
442–471) were fused to the C-terminal end of EGFP. For UAS-
ssGFP-HhC, the entire C-terminal fragment of Hh (AA 258–471)
was fused at the C terminus of EGFP.
Tissue was prepared from either20 adult heads or 50dissected eye-
brain complexes from late third instar larvae and was boiled in 100 ml
13 SDS sample buffer. Where the eye disc was separated from the
brain, dissection was done on ice in Shields and Sang M3 medium
(Sigma), and either 100 eye discs or 100 larval brain hemispheres
were boiled in 100 ml 13 SDS sample buffer. Samples were run on
a 12%–15% polyacrylamide gel and transferred onto PVDF mem-
brane (Bio-Rad). The membrane was blocked with 5% milk in PBT
(13PBS, 0.05% Tween 20) and probed with primary mouse anti-
HA antibody (12CA5 clone, Roche) at a 1:5000 dilution, followed
by secondary HRP-conjugated anti-mouse antibody (Amersham)
at a 1:5000 dilution. Bands were visualized by chemiluminescence
Antibody staining was performed as described by Kunes et al.
(1993). Primary antibodies were used at the following dilutions: rab-
bit a-BSH, 1:500; mouse a-CSP (mAb6D6), 1:10; mouse a-Dac (Mar-
don et al., 1994), 1:20; rat a-Elav, 1:25; mouse a-FasII (1D4), 1:10;
mouse a-FasIII (7G10), 1:20; rabbit a-GFP (A11122, Molecular
Probes), 1:200; rabbit a-HA (Santa Cruz), 1:400; mouse a-HA
bita-Hh(P. Ingham, ICRF), 1:500;mouse a-neuroglian(BP104), 1:20;
mouse a-neurotactin (BP106), 1:10; Cy3-goat a-mouse (Jackson),
1:200; Cy3-goat a-rabbit (Jackson), 1:500; Cy5-goat a-mouse (Jack-
son), 1:200; Cy5-donkey a-rabbit (Jackson), 1:500.
Microscopy and Data Analysis
Specimens were viewed on a Zeiss LSM510 Meta confocal micro-
scope, with constant acquisition settings when comparing speci-
mens within a given experiment. Quantification was performed
with Image J, which computed the mean pixel intensity of a selected
region from 6–14 specimens of each genotype. Lamina neuron
quantification in the third instar visual system was performed via
an analysis of z-stacks with Velocity version 3.5 (Improvision) soft-
ware. Specimens with a comparable number of ommatidial columns
ton, M. Gonzalez-Gaitan, A. Keller, S. Noselli, M. Ramaswami, and
T. Tabata for sharing reagents. We are indebted to the Drosophila
Stock Center (Bloomington) for strains. This work was supported
by National Institutes of Health-National Eye Institute grant
EY10112 to S.K.
Received: September 25, 2005
Revised: February 1, 2006
Accepted: March 13, 2006
Published: May 8, 2006
ila homologue of the putative tumour suppressor EXT-1 and is
needed for Hh diffusion. Nature 394, 85–88.
Burke, R., Nellen, D., Bellotto, M., Hafen, E., Senti, K.A., Dickson,
B.J., and Basler, K. (1999). Dispatched, a novel sterol-sensing do-
main protein dedicated to the release of cholesterol-modified
hedgehog from signaling cells. Cell 99, 803–815.
Chang, H.C., Newmyer, S.L., Hull, M.J., Ebersold, M., Schmid, S.L.,
and Mellman, I. (2002). Hsc70 is required for endocytosis and cla-
thrin function in Drosophila. J. Cell Biol. 159, 477–487.
Cho, K.O., Chern, J., Izaddoost, S., and Choi, K.W. (2000). Novel sig-
naling from the peripodial membrane is essential for eye disc pat-
terning in Drosophila. Cell 103, 331–342.
Dakubo, G.D., Wang, Y.P., Mazerolle, C., Campsall, K., McMahon,
A.P., and Wallace, V.A. (2003). Retinal ganglion cell-derived sonic
hedgehog signaling is required for optic disc and stalk neuroepithe-
lial cell development. Development 130, 1302967–1302980.
A C-Terminal Motif Targets Hedgehog to Axons
Davis, G.W., Schuster, C.M., and Goodman, C.S. (1997). Genetic
analysis of the mechanisms controlling target selection: target-de-
rived Fasciclin II regulates the pattern of synapse formation. Neuron
Djiane, A., Yogev, S., and Mlodzik, M. (2005). The apical determi-
nantsaPKC anddPatj regulate Frizzled-dependentplanar cellpolar-
ity in the Drosophila eye. Cell 121, 621–631.
Entchev, E.V., Schwabedissen, A., and Gonzalez-Gaitan, M. (2000).
Gradient formation of the TGF-b homolog Dpp. Cell 103, 981–991.
Gallet, A., Rodriguez, R., Ruel, L., and Therond, P.P. (2003). Choles-
terol modification of hedgehog is required for trafficking and move-
ment, revealing an asymmetric cellular response to hedgehog. Dev.
Cell 4, 191–204.
Ghiglione, C., Devergne, O., Georgenthum, E., Carballes, F., Medi-
ceptor Domeless controls border cell migration and epithelial polar-
ization during oogenesis. Development 129, 5437–5447.
Gibson, M.C., and Schubiger, G. (2000). Peripodial cells regulate
proliferation and patterning of Drosophila imaginal discs. Cell 103,
Greco, V., Hannus, M., and Eaton, S. (2001). Argosomes: a potential
vehicle for the spread of morphogens through epithelia. Cell 106,
Hall, T.M., Porter, J.A., Young, K.E., Koonin, E.V., Beachy, P.A., and
Leahy, D.J. (1997). Crystal structure of a Hedgehog autoprocessing
domain: homology between Hedgehog and self-splicing proteins.
Cell 91, 85–97.
axons, triggers neurogenesis in the developing visual centers of the
Drosophila brain. Cell 86, 411–422.
Izaddoost, S., Nam, S.C., Bhat, M.A., Bellen, H.J., and Choi, K.W.
(2002). Drosophila Crumbs is a positional cue in photoreceptor ad-
herens junctions and rhabdomeres. Nature 416, 178–182.
Jones, B., and McGinnis, W. (1993). A new Drosophila homeobox
gene, bsh, is expressed in a subset of brain cells during embryogen-
esis. Development 117, 793–806.
Kunes, S. (2000). Axonal signals in the assembly of neural circuitry.
Curr. Opin. Neurobiol. 10, 58–62.
Kunes, S., Wilson, C., and Steller, H. (1993). Independent guidance
of retinal axons in the developing visual system of Drosophila.
J. Neurosci. 13, 752–767.
Lee, J.J., von Kessler, D.P., Parks, S., and Beachy, P.A. (1992). Se-
cretion and localized transcription suggest a role in positional sig-
naling for products of the segmentation gene hedgehog. Cell 71,
Lee, J.J., Ekker, S.C., von Kessler, D.P., Porter, J.A., Sun, B.I., and
Beachy, P.A. (1994). Autoproteolysis in hedgehog protein biogene-
sis. Science 266, 1528–1537.
Lewis, P.M., Dunn, M.P., McMahon, J.A., Logan, M., Martin, J.F.,
St-Jacques, B., and McMahon,A.P. (2001). Cholesterol modification
of sonic hedgehog is required for long-range signaling activity and
effective modulation of signaling by Ptc1. Cell 105, 599–612.
Longley, R.L., Jr., and Ready, D.F. (1995). Integrins and the develop-
ment of three-dimensional structure in the Drosophila compound
eye. Dev. Biol. 171, 415–433.
netic functions of similar small GTPases: Drosophila Drac1 is in-
volved in axonal outgrowth and myoblast fusion. Genes Dev. 8,
Ma, Y., Erkner, A., Gong, R., Yao, S., Taipale, J., Basler, K., and
Beachy, P.A. (2002). Hedgehog-mediated patterning of the mamma-
lian embryo requires transporter-like function of Dispatched. Cell
Mann, R.K., and Beachy, P.A. (2004). Novel lipid modifications of se-
creted protein signals. Annu. Rev. Biochem. 73, 891–923.
Mardon, G., Solomon, N.M., and Rubin, G.M. (1994). dachshund en-
codes a nuclear protein required for normal eye and leg develop-
ment in Drosophila. Development 120, 3473–3486.
Miller, K.E., DeProto, J., Kaufmann, N., Patel, B.N., Duckworth, A.,
and Van Vactor, D. (2005). Direct observation demonstrates that
Liprin-alpha is required for trafficking of synaptic vesicles. Curr.
Biol. 15, 684–689.
Panakova, D.,Sprong, H.,Marois,E., Thiele,C.,and Eaton,S. (2005).
Lipoprotein particles are required for Hedgehog and Wingless sig-
naling. Nature 435, 58–65.
Pellikka, M., Tanentzapf, G., Pinto, M., Smith, C., McGlade, C.J.,
Ready, D.F., and Tepass, U. (2002). Crumbs, the Drosophila homo-
logue of human CRB1/RP12, is essential for photoreceptor morpho-
genesis. Nature 416, 143–149.
Perin, M.S., Fried, V.A., Mignery, G.A., Jahn, R., and Su ¨dhof, T.C.
(1990). Phospholipid binding by a synaptic vesicle protein homolo-
gous to the regulatory region of protein kinase C. Nature 345, 260–
Perrimon, N., and Mahowald, A.P. (1987). Multiple functions of seg-
ment polarity genes in Drosophila. Dev. Biol. 119, 587–600.
Porter, J.A., von Kessler, D.P., Ekker, S.C., Young, K.E., Lee, J.J.,
Moses, K., and Beachy, P.A. (1995). The product of hedgehog auto-
proteolytic cleavage active in local and long-range signaling. Nature
Porter, J.A., Young, K.E., and Beachy, P.A. (1996a). Cholesterol
modification of hedgehog signaling proteins in animal development.
Science 274, 255–259.
Porter, J.A., Ekker, S.C., Park, W.J., von Kessler, D.P., Young, K.E.,
Chen, C.H., Ma, Y., Woods, A.S., Cotter, R.J., Koonin, E.V., and
Beachy, P.A. (1996b). Hedgehog patterning activity: role of a lipo-
philic modification mediated by the carboxy-terminal autoprocess-
ing domain. Cell 86, 21–34.
Ramirez-Weber, F.A., and Kornberg, T.B. (1999). Cytonemes: cellu-
lar processes that project to the principal signaling center in Dro-
sophila imaginal discs. Cell 97, 599–607.
Rietveld, A., Neutz, S., Simons, K., and Eaton, S. (1999). Association
sophila raft lipid microdomains. J. Biol. Chem. 274, 12049–12054.
Simmonds, A.J., dosSantos, G., Livne-Bar, I., and Krause, H.M.
(2001). Apical localization of wingless transcripts is required for
wingless signaling. Cell 105, 197–207.
Takei, Y., Ozawa, Y., Sato, M., Watanabe, A., and Tabata, T. (2004).
Three Drosophila EXT genes shape morphogen gradients through
synthesis of heparan sulfate proteoglycans. Development 131,
Tashiro, S., Michiue, T., Higashijima, S., Zenno, S., Ishimaru, S., Ta-
kahashi, F., Orihara, M., Kojima, T., and Saigo, K. (1993). Structure
and expression of hedgehog, a Drosophila segment-polarity gene
required for cell-cell communication. Gene 124, 183–189.
The, I., Bellaiche, Y., and Perrimon, N. (1999). Hedgehog movement
fate proteoglycan. Mol. Cell 4, 633–639.
Tian, H., Jeong, J., Harfe, B.D., Tabin, C.J., and McMahon, A.P.
(2005). Mouse Disp1 is required in sonic hedgehog-expressing cells
for paracrine activity of the cholesterol-modified ligand. Develop-
ment 132, 133–142.
Traiffort, E., Moya, K.L., Faure, H., Hassig, R., and Ruat, M. (2001).
High expression and anterograde axonal transport of aminoterminal
sonichedgehog inthe adult hamster brain. Eur. J.Neurosci.14,839–
Wallace, V.A., and Raff, M.C. (1999). A role for Sonic hedgehog in
axon-to-astrocyte signalling in the rodent optic nerve. Development
Wilkie, G.S., and Davis, I. (2001). Drosophila wingless and pair-rule
transcripts localize apically by dynein-mediated transport of RNA
particles. Cell 105, 209–219.
Zhang, Y.Q., Rodesch, C.K., and Broadie, K. (2002). Living synaptic
vesicle marker: synaptotagmin-GFP. Genesis 34, 142–145.
Zinsmaier, K.E., Hofbauer, A., Heimbeck, G., Pflugfelder, G.O.,
Buchner, S., and Buchner, E. (1990). A cysteine-string protein is ex-
pressed in retina and brain of Drosophila. J. Neurogenet. 7, 15–29.