Antagonistic roles of Wnt5 and the Drl receptor in patterning the

ARTICLES
© 2007 Nature Publishing Group http://www.nature.com/natureneuroscience
Antagonistic roles of Wnt5 and the Drl receptor in
patterning the Drosophila antennal lobe
Ying Yao1,5, Yuping Wu2,5, Chong Yin2,5, Rie Ozawa1, Toshiro Aigaki3, Rene R Wouda4,
Jasprina N Noordermeer4, Lee G Fradkin4 & Huey Hing1,2
Numerous studies have shown that ingrowing olfactory axons exert powerful inductive influences on olfactory map development.
From an overexpression screen, we have identified wnt5 as a potent organizer of the olfactory map in Drosophila melanogaster.
Loss of wnt5 resulted in severe derangement of the glomerular pattern, whereas overexpression of wnt5 resulted in the formation
of ectopic midline glomeruli. Cell type–specific cDNA rescue and mosaic experiments showed that wnt5 functions in olfactory
neurons. Mutation of the derailed (drl) gene, encoding a receptor for Wnt5, resulted in derangement of the glomerular map,
ectopic midline glomeruli and the accumulation of Wnt5 at the midline. We show here that drl functions in glial cells, where it
acts upstream of wnt5 to modulate its function in glomerular patterning. Our findings establish wnt5 as an anterograde signal that
is expressed by olfactory axons and demonstrate a previously unappreciated, yet powerful, role for glia in patterning the Drosophila
olfactory map.
The perception of odors requires that stimulus information be systematically organized in a ‘map’ in the olfactory bulb. Indeed, axons of
olfactory receptor neurons (ORNs) expressing a given odorant receptor
sort out from other axons and terminate specifically in one of a field of
discrete synaptic structures, termed glomeruli, in the olfactory
bulb1. Previous studies have shown that the pattern of glomeruli, or
olfactory map, is a direct result of precise axon pathfinding and
synaptogenesis2,3. Recent work in several species has identified a
number of molecules that are necessary for the proper development
of the olfactory map. These molecules include transmembrane
proteins such as Ncam-180, Neuropilin-1, Plexin A, Robo, the odorant
receptors, Dscam and Cadherin, as well as cytoplasmic signaling
molecules such as Src, Fyn, Dock and Pak4–11. An attempt to
integrate these diverse mechanisms has led to a hierarchical model
of ORN axon targeting, wherein ORN axons are sequentially
guided toward their postsynaptic targets with an increasing degree
of precision5,8.
A wealth of evidence shows that ingrowing ORN axons are important for directing the development of the glomeruli in Drosophila and
other species. First, genetic or surgical disruption of the ORN axons
blocks the formation of the glomeruli, indicating that ORN axons are
necessary for glomerular development9,12,13. Second, when the LIM
kinase 1 (Limk1) gene is overexpressed in the ORNs, or when the ORNs
are transplanted to ectopic sites, the axons direct the formation
of glomeruli in ectopic positions, indicating that the presence of
ORN axons is also sufficient for glomerular development13–16. It has
been suggested that ORN axons have an intrinsic ability to specify
glomerular development in their target tissues17, but the underlying
mechanisms for this are unknown.
The Drosophila olfactory system is highly suited for use in unraveling
the mechanisms of olfactory map development. The anatomy and
development of the fly antennal lobe closely resembles that of the
olfactory bulb of mammals, but is vastly simpler, containing only 43
glomeruli compared with B2,000 in the mouse. We conducted an
overexpression screen for genes that disrupted the stereotyped anatomy
of the Drosophila antennal lobe. We identified wnt5 as a candidate
regulator of antennal lobe development. The wnt5 gene encodes a
member of the large, evolutionarily conserved Wnt family of secreted
proteins, which have well-established roles in embryonic patterning,
cell proliferation and cell differentiation18,19. Recent studies have
shown that Wnt5 functions as a repulsive cue that routes axons
through the correct commissures of the embryonic ventral midline20
and regulate their fasciculation21, and is required for stabilizing a subset
of axons in the adult brain22. We show here that wnt5 functions in
ORNs to regulate the patterning of the olfactory glomeruli. In the
absence of wnt5, ORN axons do not target properly, and the glomerular
map is disorganized. We also provide evidence that Drl, a Ryk-family
receptor tyrosine kinase previously demonstrated to act as a
repulsive neuronal Wnt5 receptor in the embryonic CNS20, functions
in glia to inhibit Wnt5 signaling via its extracellular Wnt inhibitory
factor (WIF) domain. Our results demonstrate that Wnt5 is an
anterograde signal by which ORN axons specify the organization of
the olfactory map and that Drl, presented by glial cells, is a powerful
modulator of this signal.
1Neuroscience
Program and 2Cell and Developmental Biology, University of Illinois at Urbana-Champaign, 601 South Goodwin Avenue, Urbana, Illinois 61801, USA.
Sciences, Tokyo Metropolitan University, 1-1 Minami-osawa, Hachioji-shi, Tokyo 192-0397, Japan. 4Laboratory of Developmental Neurobiology, Department of
Molecular Cell Biology, Leiden University Medical Center, Einthovenweg 20, 2300 RC Leiden, The Netherlands. 5These authors contributed equally to this work.
Correspondence should be addressed to H.H. ([email protected]).
3Biological
Received 10 July; accepted 17 September; published online 14 October 2007; doi:10.1038/nn1993
NATURE NEUROSCIENCE ADVANCE ONLINE PUBLICATION
1
ARTICLES
© 2007 Nature Publishing Group http://www.nature.com/natureneuroscience
RESULTS
ORNs overexpressing wnt5 disrupt antennal lobe structure
From a screen of 3,996 P{GS} Drosophila lines23, we found that the
overexpression of wnt5 (P{GS1}1192) under the control of the SG18.1Gal4 driver, which is preferentially expressed in the ORNs9,24,
profoundly altered the anatomy of the antennal lobe (Fig. 1). In the
wild-type adult expressing GFP under the control of SG18.1-Gal4, the
antennal lobes were spherical (B70 mm diameters) and were connected
by a commissure B20 mm thick (Fig. 1a). In the wnt5-overexpressing
animals, the antennal lobes were malformed and glomerulus-like
a
b
structures appeared in the commissure in 79% (42/53) of the antennal
lobes, substantially enlarging the structure to B60 mm in thickness
(Fig. 1b). We observed that a large percentage of the Or22a, Or43b and
Or47b glomeruli in the wnt5-overexpressing animals were split into
smaller subunits, many of which were displaced into the midline
(Supplementary Fig. 1 online). The termination of ORN axons in
ectopic positions indicates that ORN axons were misguided in the
wnt5-overexpressing animals. Examination of the projection neurons
showed that their dendritic arbors were disrupted, and a subset of them
invaded the midline, where they innervated the ectopic glomeruli
c
P{GS1}1024, 1039 and 1192 insertion
-glutamyltranserase gene
wnt5 gene
wnt5 deletion line
WT
SG18.1 > P{GS1} 1192
e
d
f
wnt5
WT
Or43b
Or47a & Or47b
D
wnt5
g
h
WT
Or71a
l
Or43b
o
wnt5
60
50
40
30
20
10
0
Between
two
glomeruli
*
***
**
Or71a
p 120
WT
wnt5
From glomerulus to
the top of AL
*** ***
***
*
Splitting in wnt5
Crossing defect
in wnt5
100
Or47a & Or47b
n
720K
wnt5
q
r
WT
wnt5
Or47a & Or47b
60
40
20
a
b
59
c
42
a
72
0K
47
b
67
d
67
b
71
43
d
b
wnt5
WT
47
67
a
c
a-
47
42
b
59
a
67
b
71
720K
80
0
43
WT
Or47b in WT
Or47b in wnt5
j
m
Percentage (%)
wnt5
Or47a in WT
Or47a in wnt5
i
Or43b
k
Distance (µm)
Or43b in WT
Or43b in wnt5
M
Figure 1 The olfactory map is disrupted in the wnt5 mutant. (a–r) Preparations of wild-type (WT) flies and wnt5 mutants were stained with nc82 (magenta)
and antibody to GFP (green). Antennal lobes of wild-type (a), wnt5-overexpressing (P{GS1}1192) (b) and wnt5400 (d) animals were visualized using GFP driven
by SG18.1-Gal4. Commissures are indicated with arrows. (c) Schematic diagram of the wnt5 locus showing the P{GS} lines recovered from our screen and
the wnt5400 deletion breakpoints. (e) Superimposed outlines of antennal lobes from wild-type flies and the wnt5400 mutants (n ¼ 12). The wnt5400 mutant
antennal lobes were flattened dorsomedially (D, dorsal; M, medial), producing a characteristic heart-shape appearance. (f) Plots of the positions of Or43b,
Or47a and Or47b glomeruli in the wild type and wnt5400 mutant. Preparations of wild-type (g–j) and wnt5400 mutant (k–n) antennal lobes in which specific
glomeruli are highlighted with GFP. In the wnt5400 mutant, the glomeruli were closer to the antennal lobe dorsal edge or were split into smaller structures
(arrowheads), compared with wild type. ORN axons frequently took meandering paths (arrow) to their targets and failed to cross the midline (double
arrowheads). (o) Histograms showing the distances between various glomeruli and the antennal lobe dorsal edge or between two glomeruli in the same
antennal lobes (ALs). Values are mean ± s.e.m. (p) Quantification of the glomerular-splitting and midline-crossing defects in the wnt5400 mutant. Single
confocal sections showing the projection neuron dendritic arrangement in the wild type (q) and wnt5400 mutant (r), labeled by GH164-Gal4 driving
UAS-mGFP. For each group, n ¼ 25–40. Scale bar, 20 mm.
2
ADVANCE ONLINE PUBLICATION NATURE NEUROSCIENCE
a
WT
b
26 h
wnt5
c
© 2007 Nature Publishing Group http://www.nature.com/natureneuroscience
WT
e
Percentage (%)
ARTICLES
26 h
120
100
80
60
40
20
0
wnt5
f
Crossing midline
Project dorsally
Absent
40–50 hAPF
50 h
g
h
j
k
18
4
50
18
0
22
0
28
0
0
0
9
18 h
i
Adult
Adult
40–50 hAPF
(n = 18) (n = 26) (n = 50) (n = 55)
d
50 h
WT wnt5
18 h
(Supplementary Fig. 1). Thus, wnt5 overexpression in the ORNs
disrupted the stereotyped structure of the antennal lobe and elicited
the formation of ectopic glomeruli at the midline.
wnt5 is necessary for antennal lobe structure
To determine whether wnt5 normally functions in the antennal lobes,
we examined the antennal lobes of the wnt5400 null mutant (Fig. 1c),
which is homozygous viable21. In the wnt5400 mutant expressing GFP
under the control of SG18.1-Gal4, many antennal lobes (37/55 antennal
lobes, B67%) were not connected by commissures (Fig. 1d,p) and
appeared misshapen; being flattened dorsally and elongated ventrally,
producing a characteristic heart shape (Fig. 1d,e,k–n).
We investigated the glomerular arrangement of eight ORN subclasses. In the wild type, these glomeruli are located at stereotyped
positions in the antennal lobe25,26 (Fig. 1g–j). In the wnt5400 mutant,
the dorsal edge of the antennal lobe appeared to be closer to many of
these glomeruli (Fig. 1f,k–n,o). To test the possibility that the antennal
lobes were distorted, we measured the distance between glomeruli in
the same antennal lobe (Or47a to Or47b, and Or67d; Fig. 1j,n,o). In the
wild type, the distance between the Or47a and Or47b glomeruli was
22.51 ± 1.60 mm (mean ± s.e.m., n ¼ 32). In the wnt5400 mutant, this
distance decreased to 8.70 ± 2.73 mm (n ¼ 32, P ¼ 0.0001 compared
with wild type). Thus, the wnt5400 antennal lobes showed a characteristic ‘collapse’ of the dorsomedial region of the antennal lobe (Fig. 1e).
Examination of the ORN fibers also showed that ORN axons frequently
took meandering paths toward their targets (arrows, Fig. 1k,l), looped
back on the ipsilateral glomeruli (double arrowheads, Fig. 1n), stalled
before the commissure, projected aberrantly to dorsal regions of the
NATURE NEUROSCIENCE ADVANCE ONLINE PUBLICATION
Figure 2 Wnt5 functions during antennal lobe
development and localizes to projection neuron
dendrites. (a–d) Preparations of wild type (a,c)
and wnt5400/Y hemizygote (b,d) in which GFP was
driven by SG18.1-Gal were stained with nc82
(magenta) and GFP (green) antibodies. Figures
show projected confocal sections. (a,c) In the wild
type, contralateral axons crossed the midline
(arrow) and the antennal lobes had spherical
shapes. (b,d) In the mutant, contralateral axons
projected aberrantly toward dorsal regions of
the brain (arrows). (e) Quantification of the
commissural defects seen in the wnt5400 mutant
at 40–50 hAPF or adulthood. (f–h) Preparations of
18-hAPF pupal brains expressing GFP (green)
under the control of the pan-olfactory driver pblGal4 were stained with antibodies to Wnt5 (red)
and Elav (blue). Figures show single confocal sections. Asterisks indicate three (of the four) spheroidal structures that accumulated Wnt5 and are
surrounded by masses of ORN axons. (i–k) Preparations of 18-hAPF pupal brains expressing GFP
(green) under the control of GH146-Gal4 were
stained with antibodies to Wnt5 (red) and Elav
(blue). Projection neuron dendrites overlapped
largely with the niduses of Wnt5 staining (red).
Dorsal is up and lateral is to the right in f–k. Scale
bars, 20 mm in a–d and 10 mm in f–k.
brain (arrows, Fig. 2) or terminated on ectopic glomeruli (arrowheads, Fig. 1k,l,n,p). In
summary, the loss of wnt5 resulted in aberrant
targeting of the ORN axons and characteristic
shifts of dorsomedial glomeruli leading to the
antennal lobes acquiring a distinct heartshaped appearance.
We also examined the projection neuron dendrites by expressing
GFP under the control of the GH146-Gal4 driver. In the wild type,
GH146-Gal4 labeled a subset of projection neuron dendrites, revealing
an invariant pattern in the antennal lobe neuropil (Fig. 1q). In the
wnt5400 mutant, this pattern was disrupted and many dendritic arbors
were displaced ventrally (Fig. 1r). Thus, mutation of the wnt5 gene
disrupted the targeting of the ORN axons and the positioning of the
projection neuron dendrites.
wnt5 functions in antennal lobe development
To determine when the observed defects arise in the wnt5400 mutant, we
evaluated the antennal lobe structure of the wnt5400 mutant during the
pupal stages. In the control wnt5400/+ heterozygotes at 26 h after
puparium formation (hAPF), the antennal lobe had a smooth neuropil
(Fig. 2a). ORN axons had arrived at the outer surface of the antennal
lobes and were projecting across the midline in a commissure that was
B5 mm in width. At 50 hAPF, the antennal lobe neuropil was well
partitioned into glomeruli and a thick commissure connected the left
and right antennal lobes (Fig. 2c). In the hemizygous wnt5400/Y
mutant, the antennal lobes were oval shaped at 26 hAPF, as in the
wild type, although in many brains, axons appeared to project dorsally
instead of across the midline (Fig. 2b). At 50 hAPF, the antennal lobes
appeared misshapen, with many showing the distinctive heart shape
seen in the adult (Fig. 2d). Contralateral axons failed to decussate and
instead projected dorsally in 85% of the brains examined (22/26;
Fig. 2e). By the adult stage, many of the brains no longer showed
dorsal projections, indicating that the wayward axons had likely
retracted or degenerated (Fig. 2e). These results indicate that wnt5
3
ARTICLES
a
WT
WT
wnt5/Y;
SG > wnt5
wnt5
wnt5/Y;SG > wnt5
f
wnt5/Y;720K > wnt5
h
g
Or47b in WT
Or47b in wnt5/Y mutant
Or47b in rescue of wnt5/Y mutant
D
M
*** ***
*
WT
wnt5/Y
wnt5/Y;
wnt5/Y;
SG > wnt5 72OK > wnt5
j
wnt5
wnt5/Y;
SG > wnt5
WT and wnt5/Y
wnt5/Y;
72OK > wnt5
wnt5/Y;GH146 > wnt5
Splitting in WT
k
l
WT
wnt5
WT
d
WT
wnt5
50
48
46
44
42
40
**
Percentage (%)
i
30
25
20
15
10
5
0
wnt5/Y
wnt5/Y;
720K > wnt5
c
Distance between 720K
and the top of AL (µm)
Distance between
Or47b and the
top of AL (µm)
© 2007 Nature Publishing Group http://www.nature.com/natureneuroscience
e
b
80
Splitting in wnt5
Crossing defect
in WT
Crossing defect
in wnt5
60
40
20
0
Figure 3 wnt5 functions in the ORNs. (a–h) Antennal lobes from wild-type flies (a), wnt5400 mutants (b) and wnt5400 mutants expressing UAS-wnt5 under
the control of various drivers (c,d,h) were stained with antibody to GFP (green), to reveal the Or47b axons, and nc82 (magenta). Expression of UAS-wnt5 with
SG18.1-Gal4 (c) and 72OK-Gal4 (d) rescued the wnt5400 mutant phenotype, but expression with GH146-Gal4 did not (h). (e) Histograms show the distances
between the Or47b glomerulus and the top edge of the antennal lobe in the various genotypes. (f) Superimposed antennal lobe outlines from various genotypes
(n ¼ 12). (g) Plots showing the Or47b glomerulus positions in the various genotypes. (i,j) Representative antennal lobes with MARCM clones of wild-type and
wnt5400 mutant 72OK neurons. In antennal lobes with wnt5400 mutant 72OK axons, the 72OK glomerulus was split into smaller structures (arrowheads) and
contralateral axons (arrows) looped back on the ipsilateral targets instead of crossing the midline. (k) A plot of the positions of the 72OK glomerulus in the
wild-type and wnt5400 mosaic animals. (l) Quantification of the distance between the 72OK glomerulus and the top edge of the antennal lobe, the splitting of
glomerulus and the absence of the contralateral tract in wild-type and wnt5400 mosaic animals. Values are mean ± s.e.m. All figures show projected confocal
sections. SG ¼ SG18.1-Gal4; 72OK ¼ 72OK-Gal4. Scale bar, 20 mm.
functions during development to regulate ORN axon projection and
antennal lobe development.
Wnt5 protein is localized in the developing antennal lobes
To ascertain the distribution of Wnt5 protein, we stained pupal brains
with an antibody to Wnt5 (ref. 21) that did not stain wnt5400 mutant
brains (Supplementary Fig. 2 online), which demonstrates its specificity. We stained pupae expressing GFP under the control of pbl-Gal4,
a pan–olfactory neuron driver11. We observed ORN axons penetrating
into the antennal lobe as early as 18 hAPF (arrows, Fig. 2f,g). At this
time, Wnt5 immunolabeling was seen in four spheroidal structures
(B7 mm in diameter, asterisks, Fig. 2h), each surrounded by profuse
axonal terminals (Fig. 2f). This pattern of Wnt5 expression remained
relatively unchanged up to 42 hAPF (Supplementary Fig. 2). To
identify the spheroidal structures, we stained 18-hAPF animals expressing GFP under the control of GH146-Gal4, which labels the projection
neuron dendrites, with the Wnt5 antibody. Wnt5 immunoreactivity
largely coincided with the projection neuron dendrites, indicating that
Wnt5 was localized to projection neuron dendrites during glomerular
development (Fig. 2i–k). Besides the dendritic localization, we show
below that Wnt5 also accumulated in the region of the antennal
commissure in the drl2 mutant. Wnt5 staining was greatly reduced
by 70 hAPF, when glomerular development was largely complete (data
not shown). Our immunolocalization results indicate that Wnt5 is
localized to incipient glomeruli during the period of ORN axon
targeting and glomerular development.
4
wnt5 is required in the ORNs
To delineate the cell type in which wnt5 functions, we first restored
wnt5 function to specific cell types in the wnt5400 mutant background.
We employed a UAS-wnt5 transgene21, which expresses wnt5 at a low
level and therefore does not disrupt antennal lobe development
(Supplementary Fig. 1). Three criteria were specifically assessed: the
shape of the antennal lobe, the position of the Or47b glomerulus and
the presence or absence of the commissure. In the wnt5400 mutant, the
antennal lobe was heart shaped (compare Fig. 3a,b), the Or47b
glomerulus was located 14.44 ± 0.8 mm (n ¼ 14) from the dorsal
edge of the antennal lobe compared with the wild-type glomerulus,
which was located 24.66 ± 1.3 mm (n ¼ 20) from the dorsal edge
(P o 0.0001; Fig. 3), and the commissure was present in only 30% of
the brains. When we expressed UAS-wnt5 using the GH146-Gal4 driver,
which is expressed in projection neurons, the mutant phenotype was
not rescued (Fig. 3h). No rescue was observed when UAS-wnt5 was
expressed under the control of MZ317-Gal4, which is expressed in glia
(data not shown). When UAS-wnt5 was expressed under the control of
SG18.1-Gal4, the wnt5400 mutant phenotype was rescued to a significant degree. The rescued antennal lobes were spherical, instead of
heart-shaped (Fig. 3c,f), the Or47b glomerulus was located 20.12 ±
1.4 mm from the dorsal edge of the antennal lobe (n ¼ 24, P ¼ 0.0102
compared with mutant; Fig. 3e,g), and the antennal lobes were
connected by a commissure in 55% (11/20) of the brains. Expression
of UAS-wnt5 under the control of the 72OK-Gal4 driver, which is
expressed in a subset of ORNs7, also significantly rescued the mutant
ADVANCE ONLINE PUBLICATION NATURE NEUROSCIENCE
ARTICLES
Or43b
Or67b
Or42a
b
Or71a
c
d
g
h
l
WT
a
e
f
drl
*
*
j
k
m
n
o
q
r
WT
p
*
drl
s
Percentage (%)
© 2007 Nature Publishing Group http://www.nature.com/natureneuroscience
drl
i
120 drl- mutant phenotype
100
Targeting
Unbalance
D
t
D
80
L
60
M L
M
71a
43b
40
Figure 4 The olfactory map is disrupted in the drl
mutant. (a–r) Antennal lobes of wild type (a–d,
m–o) and drl2 mutants (e–l,p–r) were stained with
antibody to GFP (green) and nc82 (magenta).
(a–d) Wild-type antennal lobes expressing GFP in
the Or43b, Or67b, Or42a and Or71a glomeruli.
(e–l) drl2 mutant antennal lobes showing the
corresponding glomeruli. Arrowheads indicate the
split glomeruli, arrows indicate the wandering
ORN axons and asterisks denote the ectopic
midline glomeruli. Antennal lobes from wild-type
flies (m) and drl2 mutants (p) expressing GFP
(green) under the control of SG18.1-Gal4 were
stained with nc82 (magenta). Ectopic protrusions
were visible at the midline of the drl2 mutant
(asterisk). In the wild-type adult, projection
neuron dendrites, visualized with GH146-Gal4
driving GFP (green), were arranged in a
stereotyped pattern (n) and never terminated at
the midline (o). In the drl2 mutant, the dendritic
pattern was severely deranged (q) and dendrites
innervated ectopic glomeruli in the midline
(asterisk, r). (s) Quantification of the targeting
(dark bars) or the unbalanced (light bars) defects
in the various ORN subtypes. (t) Plots of the
positions of the ectopic Or43b and Or71a
glomeruli (magenta dots) in the drl2 mutant. All
panels, except for n, o, q and r, show projected
confocal sections. Scale bars, 20 mm in a–d and
10 mm in e–h.
20
0
22a 43b 59c 42a 92a 47a 33c 67b 46a 71a 43a 47b
Medial
Lateral
V
phenotype. The rescued antennal lobes had a wild-type shape
(Fig. 3d,f), the Or47b glomerulus was located 19.68 ± 1.2 mm from
the dorsal edge of the antennal lobe (n ¼ 16, P ¼ 0.001 compared with
mutant; Fig. 3e,g) and the antennal lobes were connected by a
commissure in 77% (10/13) of the brains.
Next, we investigated whether wnt5 is required in the ORNs for
proper antennal lobe development. We employed the mosaic analysis
with a repressible cell marker (MARCM) system27 to induce and
examine clones of wild-type or wnt5400 homozygous 72OK cells.
Large clones were examined, as small clones were likely to be rescued
by Wnt5 secreted from wild-type cells adjacent to the clones. In control
animals with wild-type clones, the antennal lobes had a spherical shape
(Fig. 3i). The wild-type 72OK ORNs projected to their expected positions (49.14 ± 3.5 mm from dorsal surface of antennal lobe, n ¼ 14;
Fig. 3k,l) and formed glomeruli with stereotyped shapes. Their
contralateral axons formed a distinct fascicle that projected normally
across the midline. In animals bearing wnt5400 clones, the antennal
lobes were misshapen, with a heart-shaped appearance (Fig. 3j). The
mutant 72OK ORNs projected to a slightly more dorsal position than
that of the control (44.16 ± 4.5 mm, n ¼ 18, P ¼ 0.002; Fig. 3k,l) and
formed distorted and split glomeruli (Fig. 3j,l). Their contralateral
axons frequently appeared defasciculated and (72.2%, 13/18 brains)
failed to project across the midline (Fig. 3j,l). Our cell type–specific
cDNA rescue and mosaic experiments therefore indicated that wnt5
functioned in the ORNs for appropriate antennal lobe patterning.
drl is necessary for patterning the glomerular map
The capacity of Wnt5 to regulate glomerular patterning focused our
attention on its potential downstream signaling pathway. Drl has been
shown to act as a receptor for the repulsive Wnt5 cue in axons that cross
NATURE NEUROSCIENCE ADVANCE ONLINE PUBLICATION
V
the embryonic midline20. To determine
whether drl also functions in the antennal
lobe, we examined the antennal lobes of the
drl2 null mutant28 (Fig. 4). In 39% (22/56) of the mutant brains,
ectopic glomeruli developed at the midline (asterisks, Fig. 4a–f,m,p,r),
whereas abnormal protrusions extended from the dorsal-medial corner
of the antennal lobes of the remainder (61%). In each of the 12 subclasses of ORNs that we analyzed, ectopic midline glomeruli were seen
at low frequencies, suggesting that all subclasses have an equal tendency
to terminate at the midline. Plotting the positions of the Or43b and
Or71a glomeruli in different antennal lobes revealed no consistent
pattern, indicating that the glomeruli were randomly situated in the
mutant (Fig. 4t). In addition to the glomerular defects, ORN axons
frequently took circuitous routes to their targets (arrows, Fig. 4g,i–k).
Strikingly, a number of axons terminated unilaterally (confirmed by
unilateral antennal ablations, Supplementary Fig. 3 online), resulting
in the loss of presynaptic structures from a single antennal lobe
(Fig. 4i–k). To better assess the data, we grouped the glomerular
positioning and splitting defects as ‘targeting defects’ (Fig. 4e–h,l) and
the unilateral loss (or reduction) of glomeruli as ‘unbalanced defects’
(Fig. 4i–k). Both types of defects were present in each of the 11 ORN
subclasses (Fig. 4s).
We also inspected the organization of the projection neurons in the
drl2 mutant using the GH146-Gal4 marker. Unlike in the wild-type
adult, where GH146-Gal4-expressing cells arborized in specific glomeruli, the organization of the dendritic arbors appeared chaotic in the drl2
mutant adult (Fig. 4n,q), and a subset of arbors even invaded the
midline, where they targeted the ectopic midline glomeruli (Fig. 4o,r).
Loss of drl therefore leads to disruption of the glomerular arrangement
and midline invasion by projection neuron dendrites. This midline
invasion of projection neuron dendrites is highly reminiscent of the
wnt5 overexpression phenotype (Supplementary Fig. 1). When we
examined the antennal lobes of the drl2 mutant at 40 hAPF (Fig. 5), we
5
ARTICLES
a
WT
Figure 5 Drl functions during antennal lobe development and localizes to
TIFR glia. (a) In the wild type at 40 hAPF, projection neuron dendrites,
visualized with GH146-Gal4 driving GFP (green), were confined to the
antennal lobes. (b) An equivalent section of drl2 mutant showed projection
neuron dendrites extending toward the midline (arrows). (c–g) Frontal optical
sections of a wild-type 30-hAPF pupal brain expressing GFP (green) under
the control of Repo-Gal4 stained with an antibody to Drl (magenta). Antennal
lobes are indicated by the dotted outlines. (c–e) The Drl antibody intensely
stained the TIFR (brackets). (f,g) A more posterior section, at the level of the
antennal commissure, shows a bundle of glial processes that connects the
two antennal lobes and also expressed the Drl protein (arrows). (h) No Drl
staining was found in the drl2 mutant brain. (i–k) Antennal lobe from a
30-hAPF wild-type animal expressing GFP (green) under the control of
GH146-Gal4 stained with antibodies to Drl (red) and Elav (blue). Drl
distribution colocalized largely with the projection neuron dendrites.
Dorsal is up and lateral is to the right in i–k. All the panels show
single confocal sections. Scale bars, 10 mm.
b
40 h
drl
40 h
c
d
e
f
g
h
i
j
k
glial cells. At 30 hAPF, GFP was highly expressed by the transient
interhemispheric fibrous ring (TIFR)29, a toroidal glial structure lying
in the mid-sagittal plane of the brain (brackets, Fig. 5c,e). The ventral
edge of the TIFR is closely associated with and bridges the antennal
lobes. In more posterior sections, at the level of the antennal commissure, a thick bundle of glial processes was seen to pass from one
antennal lobe to the other (Fig. 5f). The TIFR was strongly stained by
the Drl antibody (Fig. 5d), including the bundle of glial processes
linking the antennal lobes (Fig. 5g). We next stained brains from
animals expressing GFP under the control of GH146-Gal4. At 30 hAPF,
GFP highlighted the projection neuron cell bodies and their dendritic
arbors in the antennal lobe neuropil (Fig. 5j). Drl immunolabeling
largely overlapped with the dendritic GFP staining, indicating that
Drl was localized to a large subset of the projection neuron dendrites
(Fig. 5i–k). The Drl antibody did not stain the drl2 null mutant
brains (Fig. 5h).
30 h
observed fine dendritic processes growing into the midline, which did
not occur in the wild type, indicating that the defects arose during
development (Fig. 5a,b).
Drl is localized to glial cells and projection neurons
To ascertain the localization of the Drl protein in the olfactory system,
we stained developing brains with an antibody to Drl. We first stained
animals expressing GFP under the control of Repo-Gal4, which labels
Or46a
Or43b
Or22a
Or47b
b
c
d
e
f
g
h
i
j
k
l
drl
drl
WT
a
Repo > drl
n
Repo > drl
Repo > drl
o
p
drl
m
Repo > drl
Repo > drl ∆cyto
Repo > drl ∆cyto
r
Repo > drl ∆cyto
s
Repo > drl ∆cyto
Or43b
drl
q
Repo > drl ∆wif
v
u
40
WT
6
drl
FL
GH146
FL
K371A
Repo
∆INTRA
43b
46a
47b
22a
43b
46a
43b
47b
22a
43b
46a
43b
46a
47b
22a
43b
46a
0
47b
20
22a
GH146 > drl
60
43b
GH146 > drl
Targeting
Unbalance
80
46a
t
Repo > drl K371A
100 drl-mutant phenotype
Percentage (%)
Repo > drl ∆wif
drl
© 2007 Nature Publishing Group http://www.nature.com/natureneuroscience
30 h
∆WIF
Figure 6 Drl functions in glial cells and its WIF
domain is essential. (a–u) Preparations of wildtype flies (a–d) and drl2 mutants (e–u) expressing
Or46a-mGFP, Or43b-mGFP, Or22a-mGFP and
Or47b-mGFP were stained with antibody to GFP
(green) and nc82 (magenta). (a–d) Glomeruli had
predictable shapes and were located in invariant
topographic positions in the wild type. (e–h) In the
drl2 mutant, glomeruli were ectopically positioned,
split or missing. (i–l) Expression of UAS-drl under
the control of the Repo-Gal4 driver strongly
rescued the mutant phenotype. Glomeruli
were located in their stereotyped positions.
(m–p) Expression of a truncated Drl protein
lacking the cytoplasmic domain using the
Repo-Gal4 driver rescued the mutant phenotype.
(q,r) Expression of a truncated Drl protein lacking
the extracellular WIF domain failed to rescue the
mutant phenotype. Glomeruli were split or
missing. (s) Expression of a ‘kinase-dead’ Drl
protein (K371A) also rescued the mutant
phenotype. (t,u) Expression of UAS-drl under
the control of GH146-Gal4 partially rescued
the mutant phenotype. Although the glomerular
defects remained, protrusions from the antennal
lobes were absent. (v) Quantification of the mutant
defects found in the various genotypes. Dark bars
represent targeting defects, whereas light bars
represent unbalanced defects. FL, full-length Drl
protein. All images are projections of confocal
z-series. Scale bars, 20 mm. n Z 15.
ADVANCE ONLINE PUBLICATION NATURE NEUROSCIENCE
a
*
Drl’s extracellular domain regulates glomerular patterning
Our ability to rescue the mutant phenotype allowed us to probe the
functions of the different domains of the Drl protein in glomerular
patterning. To determine whether the Drl kinase activity is required for
antennal lobe development, we mutated the conserved K371 residue,
which is essential for kinase activity32, to an alanine. Expression of the
UAS-drlK371A transgene under the control of Repo-Gal4 strongly
rescued the drl mutant phenotype (26% midline defects and 11%
Or43b axon targeting defects compared with 100% and 35%, respectively, in the drl2 mutant; Fig. 6s), indicating that kinase activity is not
NATURE NEUROSCIENCE ADVANCE ONLINE PUBLICATION
*
c
*
drl –/–
drl –/–;
SG18.1 > wnt5
SG18.1 > wnt5
d
e
f
*
*
*
g
h
i
WT
wnt5/Y
+/Y;drl
j
k
l
–/–
m
wnt5/+;drl
p
–/–
100
50
44%
38%
12 18
8
18 13
0
–/–
n
WT
*
/–
wnt5/Y;drl
drl functions in glial cells
To delineate the cell types in which drl functions, we sought to restore
drl function to specific cell types in the drl2 mutant background.
Because Drl localized, in part, to projection neuron dendrites, we
employed the GH146-Gal4 driver, which is active in two thirds of
all projection neurons30,31, to drive the expression of UAS-drl. We
examined several criteria to assess genetic rescue (Fig. 6): the presence
or absence of ectopic midline structures and the position and integrity
of the Or22a, Or43b, Or46a and Or47b glomeruli (Fig. 6a–d). In the drl2
mutant, 100% of the brains showed aberrant midline structures (either
midline glomeruli or protrusions) and 38%–94% of the brains showed
defects in the positioning or integrity of the glomeruli (Fig. 6e–h,v).
Expression of UAS-drl under the control of GH146-Gal4 partially
rescued the mutant phenotype (58% targeting defects; Fig. 6t,u,v).
To determine whether drl is required in the projection neurons, we
generated and examined drl-mutant projection neuron MARCM
single-cell clones. Loss of drl from single projection neurons, however,
did not disrupt the projection neuron dendrite morphology or the
stereotyped arrangement of the glomeruli in the antennal lobes
(Supplementary Fig. 4 online). To further evaluate whether drl
functions in neurons during antennal lobe development, we expressed
UAS-drl under the control of elav-Gal4, a pan-neuronal driver, but this
did not rescue the drl2 mutant phenotype (data not shown). Next, we
tested whether expression of drl in glia would rescue the drl2 phenotype.
Expression of UAS-drl under the control of the pan-glial driver RepoGal4 completely eliminated the midline defects (0%) and strongly
reduced the occurrence of targeting defects to 11–24% for different
glomeruli (Fig. 6i–l,v). We conclude that drl functions in the glia for
the proper development of the antennal lobes.
b
W
w T
n
+/ t5/Y
Y;
dr
w l –/–
nt
dr 5/Y
l– ;
w /–
n
dr t5/+
l– ;
Figure 7 drl functions upstream of wnt5 to inhibit wnt5 function.
(a–f) Preparations of various genotypes expressing GFP under the control of
SG18.1-Gal4 were stained for GFP (green) and nc82 (magenta). (a) Antennal
lobes of the drl2 mutant. (b) Antennal lobes of the drl2 mutant expressing
UAS-wnt5 under the control of SG18.1-Gal4. (c) Antennal lobes of
the wild type expressing UAS-wnt5 under the control of SG18.1-Gal4.
(d–f) Corresponding deep sections of a–c, revealing the large ectopic
glomeruli (asterisks) of b and e compared with a and d. (g) Wild-type
antennal lobes with invariant shape and position of Or47b glomeruli. (h) The
wnt5400 antennal lobes had a characteristic heart shape and the Or47b axons
failed to cross the midline (arrow). (i) In the drl2 antennal lobes, the Or47b
glomerulus was split and ectopic glomeruli appeared at the midline (asterisk).
(j) In the wnt5;drl double mutant, the antennal lobes had a heart-shaped
appearance and the Or47b contralateral fibers failed to cross the midline
(arrow). (k) Removal of a copy of wnt5 suppressed the drl mutant phenotype.
The antennal lobes had wild-type shapes and ectopic midline structures
were absent. (l) Quantification of the percentage of antennal lobes showing
the failure of Or47b axons to cross the midline. (m–q) Superimpositions of
the antennal lobe outlines from the various genotypes (n ¼ 6 for each
genotype). Midline protrusions are indicated with an asterisk. a–c and
g–k are projected confocal sections and d–f are single confocal sections.
Scale bar, 20 mm.
Midline
crossing (%)
© 2007 Nature Publishing Group http://www.nature.com/natureneuroscience
ARTICLES
o
wnt5/Y
+/Y; drl –/–
q
wnt5/Y; drl –/–
wnt5/+; drl –/–
essential for Drl function in antennal lobe development. To ascertain
whether the cytoplasmic domain is necessary for antennal lobe development, we used the UAS-drlDcyto transgene33, which encodes a
truncated Drl protein bearing only the extracellular and transmembrane domains. Notably, UAS-drlDcyto retained substantial ability to
rescue the drl mutant phenotype (33% midline defects, 23% targeting
defects, Fig. 6m–p,v). To determine whether the Drl WIF domain is
needed for antennal lobe development, we generated a UAS-drlDWIF
transgene, which encodes a Drl protein that only lacks the WIF domain.
The loss of the WIF domain abolished the ability of drl to rescue the
mutant phenotype (89% midline defects and 56% targeting
defects; Fig. 6q,r,v). Taken together, these results indicate that Drl
regulates antennal lobe development largely through its Wnt5-binding
WIF domain.
wnt5 acts downstream of drl in antennal lobe patterning
We next investigated the genetic relationship between wnt5 and drl. The
similarity of the midline defects seen in the drl loss-of-function and
wnt5 gain-of-function mutants suggested that the two genes might act
antagonistically in antennal lobe development. To test this idea, we
examined the effect of expressing a low level of wnt5 in the drl2 mutant
background (Fig. 7). In the drl2 mutant, midline glomeruli were found
in 39% of the animals, with most showing only protrusions from the
dorsomedial corner of the antennal lobes (Fig. 7a,d). Expression of a
single copy of UAS-wnt5 under the control of SG18.1-Gal4 in the wildtype background did not alter the antennal lobe structure (Fig. 7c,f and
Supplementary Fig. 1). In contrast, expression of a single UAS-wnt5 in
the drl2 mutant background resulted in the induction of large midline
glomeruli in 83% of the animals (Fig. 7b,e). Thus, wnt5 function is
7
ARTICLES
Figure 8 Wnt5 is localized in the antennal lobe
WT
drl
drl;Repo > drl
neuropil and midline commissure in the drl
mutant animals. (a–f) Preparations of 36-hAPF
pupae of wild-type flies (a,d) and drl2 mutants
(b,c,e,f) expressing UAS-GFP under the control
WT
drl
AL
of Repo-Gal4 were stained for Wnt5 (red), GFP
AL
AL
AL
AL
AL
drl; Repo > drl
(green) and Elav (blue). (c,f) drl2 mutant
120
***
coexpressing UAS-drl under the control of
100
***
Repo-Gal4. (a–c) Anterior optical sections
80
showing Wnt5 staining in the antennal lobe
60
neuropil. (d–f) Posterior sections showing Wnt5
staining in the commissure (dashed lines). The
40
TIFR is indicated by a bracket in a. (g) QuantiAL
20
AL
AL
AL
AL
AL
fication of the levels of Wnt5 in the antennal lobe
0
36 h
36 h
36 h
Superficial Commissure
neuropils and commissures of the various geno30 h
36 h
types. (h–k) Preparations of wild-type (h,i) and
drl2 mutant (j,k) pupae expressing UAS-GFP
under the control of GH146-Gal4 were stained
for Wnt5 (red), GFP (green) and Elav (blue).
Micrographs were taken at equivalent section
planes (at the level of the commissure) and the
antennal lobes are outlined with dashed lines.
(j) At 30 hAPF in the drl2 mutant, Wnt5 was
located medial to the antennal lobes (arrowheads), whereas Wnt5 remained in the confines of
the antennal lobes in the wild type (h). (k) At
36 hAPF in the drl2 mutant, Wnt5 accumulation
increased the antennal commissure (arrowhead).
Projection neuron dendrites (arrows) began to transgress the antennal lobe boundaries and projected toward the region of Wnt5 accumulation, which was not
seen in the wild type (i). All figures are of single confocal optical sections. Scale bar, 10 mm.
b
c
Superficial
a
e
f
i
j
k
WT
h
drl
© 2007 Nature Publishing Group http://www.nature.com/natureneuroscience
Commissure
d
Percentage intensity of
Wnt5 mAb staining (%)
g
strongly increased in the drl2 mutant background, which is consistent
with the notion that drl antagonizes wnt5 during glomerular development and antennal lobe patterning.
To determine the hierarchical order of wnt5 and drl in the genetic
pathway regulating antennal lobe development, we constructed animals
bearing null mutations in both genes and expressing GFP in the Or47b
axons. Heterozygosity for drl2 had no effect on the homozygous
wnt5400 (wnt5/Y; drl/+) mutant phenotype (data not shown). In
contrast, heterozygosity for wnt5400 markedly suppressed both the
dorsomedial antennal lobe protrusions and the glomerular defects
that were typical of the drl2 mutant (wnt5/+; drl/drl; Fig. 7i,k,o,q).
Homozygosity for both the wnt5400 and drl2 null mutations resulted
in antennal lobes having the characteristic wnt5 mutant phenotype (wnt5/Y; drl/drl; Fig. 7g–j). The dorsal-medial corners of the
antennal lobes were collapsed, producing a heart-shaped appearance
(Fig. 7m–p), and the Or47b contralateral axons failed to cross the
midline in 56% of the animals (compared with 0% in drl2 mutants and
62% in wnt5400 mutants; Fig. 7j,l). wnt5400 is thus epistatic to drl2,
suggesting that wnt5 functions downstream of and is repressed by drl in
the signaling pathway that regulates antennal lobe development.
Wnt5 protein accumulates at the midline in the drl mutant
The antagonistic relationship between drl and wnt5 led us to ask how
they might function together to regulate antennal lobe development.
The strong midline defects of the wnt5 and drl mutants and the
localization of Drl to the TIFR suggest a role for these proteins at the
midline (Fig. 8). As mentioned above, we observed Wnt5 immunoreactivity in the wild-type antennal lobe neuropil at 36 hAPF (100 ±
5.5%, n ¼ 8; Fig. 8a,g), but no detectable staining in the antennal
commissure (47.0 ± 1.7%, n ¼ 8; Fig. 8d,g). We then stained for the
Wnt5 protein in the drl2 loss-of-function mutant. Of note, Wnt5 was
found in the antennal lobe neuropil (96.0 ± 8.9%, n ¼ 16, P ¼ 0.5165
compared with wild-type neuropil; Fig. 8b,g) and the antennal
8
commissure, a structure enwrapped by processes of the TIFR glia
(75.9 ± 8.5%, n ¼ 16, P o 0.0001 compared with wild-type commissure; dashed lines, Fig. 8e,g). This result indicates that drl downregulates Wnt5 protein in the antennal commissure. To determine
whether the downregulation of Wnt5 at the midline is a result of drl
acting in glia, we restored drl specifically to glia in the drl2 mutant. In
drl2 mutants expressing UAS-drl under the control of the Repo-Gal4
driver, Wnt5 staining in the antennal commissure was no longer
observed (43.7 ± 4.7%, n ¼ 16, P ¼ 0.25 compared with wild-type
commissure; Fig. 8f,g).
The midline accumulation of Wnt5 in the drl2 mutant prompted us
to ask how it might affect glomerular development in the mutant. At
30 hAPF, as Wnt5 began to accumulate in the dorsomedial corners of
the antennal lobes in the drl2 mutant (arrowheads in Fig. 8j, compare
with Fig. 8h), projection neuron dendrites were observed lateral to the
domains of Wnt5 staining, confined to the antennal lobes (Fig. 8j). At
36 hAPF, as high levels of Wnt5 accumulated in the commissure
(arrowhead in Fig. 8k, compare with Fig. 8i), the projection neuron
dendrites began to transgress the antennal lobe boundaries and
projected medially toward the region of ectopic Wnt5 enrichment
(arrows in Fig. 8k). Our results show that in the absence of drl, Wnt5
accumulates in the commissure, leading to the formation of ectopic
midline glomeruli.
DISCUSSION
The mechanisms by which ingrowing axons sort into precise maps,
such as those found in the olfactory glomeruli or the somatosensory
barrels, are poorly understood. Deafferentation and transplantation
experiments revealed that ingrowing axons are important for specifying
the maps in the initially homogenous structures15,16,34. However, little
is known about how the ingrowing axons carry out these feats. In this
report, we show that ingrowing ORN axons express Wnt5, which
contributes to organizing the glomerular pattern of the Drosophila
ADVANCE ONLINE PUBLICATION NATURE NEUROSCIENCE
© 2007 Nature Publishing Group http://www.nature.com/natureneuroscience
ARTICLES
olfactory system. We also show that the Drl receptor tyrosine kinase
acts in glial cells to modulate Wnt5 signaling. This previously unknown
interaction between ORN axons and glia reveals an important function
of ORN axon-glia interactions in regulating the precise neural circuitry
of the Drosophila antennal lobes.
The wnt5 mutant had characteristic disruptions of the olfactory
map. Many dorsomedial glomeruli were displaced ventrally (resulting
in heart-shaped antennal lobes) and the antennal commissure failed to
form. In contrast to the loss-of-function defects, overexpression of
wnt5 led to the displacement of glomeruli into the midline. Examination of the ORN axons in the wnt5 mutant showed that they take
circuitous paths to their targets and frequently misprojected to dorsal
regions of the brain. Consistent with a role for wnt5 in antennal lobe
development, the antennal lobe defects appeared during the pupal
stage, when ORN axon targeting and glomerular development occur.
Our genetic mosaic and cell type–specific rescue experiments indicated
that wnt5 is required in the ORNs. Antibody stainings indicated that
the Wnt5 protein was enriched on the dendrites of the projection
neurons, where it presumably accumulated subsequent to its secretion
by ORNs. In addition to the projection neuron dendrites, Wnt5 also
accumulated in the antennal commissure in the drl2 mutant. We
propose that Wnt5 is a signal by which ingrowing ORN axons direct
the development of their target field.
Mutation of the drl gene also produced disruptions of the olfactory
map. However, unlike the stereotyped shifts of glomeruli seen in the
wnt5 mutant, the glomeruli were randomly positioned in or missing
from one antennal lobe in the drl mutant. Furthermore, there was a
strong tendency for glomeruli to form at the midline. As in the wnt5
mutant, ORN axons took indirect routes to their targets. That drl
functions in development is supported by the observation that antennal
lobe defects were visible at 40 hAPF, the time when ORN axon targeting
and glomerular development take place.
Antibody staining showed that the Drl protein was highly expressed
by the projection neurons and TIFR glia, cells that are intimately
associated with the ingrowing ORN axons. In the projection neurons,
Drl was enriched in the dendrites of nascent glomeruli, four of which
also appeared to accumulate Wnt5. The TIFR is a donut-shaped midsagittal structure located between the antennal lobes. Our histological
studies showed that TIFR glial processes were closely associated with
ORN axons that were projecting across the midline. Several observations indicated that drl functions in the TIFR to regulate wnt5 function.
First, removal of drl from single projection neuron clones did not
disrupt the development and morphology of the projection neurons.
Second, neuronal expression of drl in the drl2 mutant background did
not rescue the mutant phenotype. Third, expression of UAS-drl under
the control of Repo-Gal4 strongly rescued the drl mutant phenotype,
suggesting that drl functions in glial cells. Although we cannot rule out
roles for Drl in the projection neurons, collectively, our observations
suggest that drl functions predominantly in glial cells to regulate
antennal lobe development.
The phenotypic similarities between the drl loss-of-function and the
wnt5-overexpressing mutants raise the intriguing possibility that the
two genes act antagonistically in antennal lobe development. Indeed,
expression of a weak wnt5 transgene in the ORNs, which has no effect
in the wild type, triggers the formation of ectopic glomeruli in the drl2
mutant. Thus, wnt5 and drl function in opposition to each other in
antennal lobe development. To ascertain the relative positions of wnt5
and drl in this signaling pathway, we generated animals carrying null
mutations in both genes. We found that the wnt5400;drl2 double
mutants had the characteristic wnt5 phenotype. The wnt5 gene is
therefore epistatic to the drl gene, indicating that wnt5 functions
NATURE NEUROSCIENCE ADVANCE ONLINE PUBLICATION
downstream of drl in antennal lobe development. This conclusion is
also supported by the observation that, although the removal of a copy
of the wnt5 gene strongly suppressed the drl homozygous mutant
phenotype, the removal of a copy of the drl gene had no effect on the
wnt5 homozygous mutant phenotype. The genetic data that drl downregulates wnt5 function is further supported by our observation that
the Wnt5 protein significantly accumulates in the commissure in the
absence of Drl. Taken together, our genetic and histological data
indicate that drl acts to inhibit the activity of wnt5 during antennal
lobe development.
To probe the molecular mechanisms by which Drl regulates antennal
lobe development, we mutated the various domains of Drl. We
observed that neither disruption of the kinase activity nor deletion of
the intracellular domain significantly impaired rescue by the drl
transgene. In contrast, deletion of the extracellular WIF domain
completely abolished Drl’s ability to rescue the mutant phenotype.
These results suggest that Drl regulates antennal lobe patterning
predominantly through its extracellular WIF domain. How might
Drl inhibit the function of Wnt5? One possibility is that Drl inhibits
Wnt5 function simply by promoting Wnt5’s sequestration or endocytosis, thus limiting its interaction with another as yet unidentified
receptor (Supplementary Fig. 5 online). This receptor might be one of
the other Drosophila receptor tyrosine kinases or a member of the
Frizzled family, one of which, frizzled 2 (fz2), interacts genetically with
wnt5 to stabilize axons of the Drosophila visual system22. Alternatively,
Drl may directly interact with another receptor and Wnt5, as has been
observed previously for its mammalian ortholog Ryk and members of
the Wnt and Frizzled families35. This interaction could inhibit or alter
the signal transduced from the membrane. However, we did not
observe a requirement for Drl’s cytoplasmic domain, suggesting that
transduction of the Wnt5 signal by Drl alone is unlikely to have a major
role in patterning the antennal lobes.
How do glial cells interact with the ORN axons to specify the
olfactory map? Our data suggest that the ingrowing ORN axons
contribute to antennal lobe patterning through secretion of Wnt5
and that glial cells locally regulate Wnt5 actions through Drl. We
propose the following working model for how Wnt5-Drl signaling
might regulate glomerular patterning. Ingrowing ORN axons express
Wnt5, which is important for the precise organization of the glomeruli
and pathfinding of the ORN axons, such as those crossing the midline
or projecting to the dorsomedial region of the antennal lobes. Normal
antennal lobe development requires that the Wnt5 signal be locally
attenuated by the TIFR glial cell–expressed Drl protein. In the wnt5
mutant, the lack of Wnt5 signaling results in the failure of ORN axons
to cross the midline and the establishment of glomeruli in more ventral
positions. In the drl mutant, Wnt5 accumulates at the midline and
presumably inappropriately signals through another receptor, resulting
in aberrant ORN axon targeting to the midline and the formation of
ectopic glomeruli at the dorsomedial corner of the antennal lobe and at
the midline. Further studies will hopefully help to unravel the precise
mechanisms by which Wnt5 and Drl act together to specify the
patterning of the Drosophila olfactory map.
METHODS
Experimental animals. The P{GS} Drosophila lines were generated by the
Drosophila Gene Search Project (Metropolitan University, Tokyo)36. The
Or-Gal4 and Or-MCD8-GFP (Or-mGFP) lines were from L. Vosshall (The
Rockefeller University) and B. Dickson (IMBA, Austria), the UAS-drlDcyto line
was from J. Dura (Institut de Genetique Humaine)33 and the 72OK-Gal4 line
was kindly provided by K. Ito (The University of Tokyo). The construction of
9
ARTICLES
the GH-mGFP line has been described14. Other stocks, referenced throughout
the text, were obtained from Bloomington Drosophila Stock Center.
© 2007 Nature Publishing Group http://www.nature.com/natureneuroscience
Transgenes. Full-length drl cDNA (from Open Biosystems) was subcloned into
the pUAST vector to generate UAS-drl. To create UAS-drlK371A, we generated a
fragment of the drl coding region (base pairs 1079–1830) bearing the K371A
mutation by PCR, fused it in frame with the remainder of the drl coding region
and then cloned it into pUAST. To generate UAS-drlDWIF, we generated a DNA
fragment (nucleotide 472–1830) that lacked the WIF-encoding sequences by
PCR, fused it with the first 60 nucleotides of the drl coding region, and
then subcloned it into pUAST. Transgenic animals were generated by
standard procedures.
Immunohistochemistry. Adult (1 to 2 d old) or pupal brains were quickly
dissected in cold phosphate-buffered saline (PBS, 130 mM NaCl, and 10 mM
Na2HPO4, pH 7.2) and fixed in PLP fixative (2% paraformaldehyde, 0.25%
sodium periodate, 75 mM lysine-HCl and 37 mM sodium phosphate, pH 7.4)
for 1 h. The fixed brains were washed with PBST (PBS with 0.5% Triton X-100)
and stained with primary antibodies overnight. For Wnt5 staining, dissected
brains were directly stained with antibody to Wnt5 in PBS (2.5 h at 4 1C),
washed with PBS and goat serum, and fixed in PLP (1 h, 25 1C). Antibodies
and dilutions are described here. mAb nc82 (1:20 dilution) was a gift from
A. Hofbauer37, rabbit antibody to GFP (1:100 dilution) was obtained from
Molecular Probes, and rat antibody to mCD8 (1:100 dilution) was from Caltag.
Affinity-purified rabbit antibody to Wnt5 (ref. 21) and rabbit antibody to Drl
were both used at 1:100 dilutions. The anti-Drl antiserum was raised against a
GST fusion protein that included Drl amino acids 123–222 and was affinity
purified against the same protein coupled to a column. The secondary
antibodies, FITC-conjugated goat antibodies to rabbit, Cy3-conjugated goat
antibodies to mouse and FITC-conjugated goat antibodies to rat, were obtained
from Jackson Laboratories and were used at 1:100 dilution.
Statistic analysis. Statistical comparisons were performed using Microsoft
Excel and GraphPad Prism4. Tests on independent groups were two-tailed
Student’s t-test. Values and error bars indicate mean ± s.e.m. *, ** and ***
denote P o 0.05, 0.01 and 0.001, respectively.
Note: Supplementary information is available on the Nature Neuroscience website.
ACKNOWLEDGMENTS
We thank the Bloomington Drosophila Stock Center for the fly lines,
A. Hofbauer for the generous gift of the nc82 antibody and W. Zhou for
construction of the UAS-drl transgenic fly line. This work was supported by
grants from the US National Institutes of Health and the US National Institute
on Deafness and other Communication Disorders (DC5408-01), the Roy J.
Carver Charitable Trust (#03-27) (H.H.), ASPASIA (Netherlands Organization
for Scientific Research) and Pionier grants (J.N.), and a Genomics grant (L.F.
and J.N.) from the Nederlandse Organisatie voor Wetenschappelijk Onderzoek.
AUTHOR CONTRIBUTIONS
Y.Y., Y.W., C.Y. and R.O. conducted experiments in the laboratory of H.H.
R.R.W. conducted experiments in the laboratory of J.N.N. L.G.F., Y.Y., Y.W.
and H.H. analyzed the data. T.A. provided the P{GS} lines screened by the
H.H. lab. H.H. and L.G.F. wrote the manuscript with contributions from the
other authors.
Published online at http://www.nature.com/natureneuroscience
Reprints and permissions information is available online at http://npg.nature.com/
reprintsandpermissions
1. Mombaerts, P. Axonal wiring in the mouse olfactory system. Annu. Rev. Cell Dev. Biol.
22, 713–737 (2006).
2. Dynes, J.L. & Ngai, J. Pathfinding of olfactory neuron axons to stereotyped glomerular
targets revealed by dynamic imaging in living zebrafish embryos. Neuron 20,
1081–1091 (1998).
3. Potter, S.M. et al. Structure and emergence of specific olfactory glomeruli in the mouse.
J. Neurosci. 21, 9713–9723 (2001).
4. Miyasaka, N. et al. Robo2 is required for establishment of a precise glomerular map in
the zebrafish olfactory system. Development 132, 1283–1293 (2005).
10
5. Lin, D.M. & Ngai, J. Development of the vertebrate main olfactory system. Curr. Opin.
Neurobiol. 9, 74–78 (1999).
6. Hummel, T. et al. Axonal targeting of olfactory receptor neurons in Drosophila is
controlled by Dscam. Neuron 37, 221–231 (2003).
7. Hummel, T. & Zipursky, S.L. Afferent induction of olfactory glomeruli requires
N-cadherin. Neuron 42, 77–88 (2004).
8. Key, B. & St John, J. Axon navigation in the mammalian primary olfactory pathway: where
to next? Chem. Senses 27, 245–260 (2002).
9. Ang, L.H., Kim, J., Stepensky, V. & Hing, H. Dock and Pak regulate olfactory axon
pathfinding in Drosophila. Development 130, 1307–1316 (2003).
10. Lattemann, M. et al. Semaphorin-1a controls receptor neuron-specific axonal
convergence in the primary olfactory center of Drosophila. Neuron. 53, 169–184
(2007).
11. Sweeney, L.B. et al. Temporal target restriction of olfactory receptor neurons by
Semaphorin-1a/PlexinA-mediated axon-axon interactions. Neuron. 53, 185–200
(2007).
12. Oland, L.A., Orr, G. & Tolbert, L.P. Construction of a protoglomerular template by
olfactory axons initiates the formation of olfactory glomeruli in the insect brain.
J. Neurosci. 10, 2096–2112 (1990).
13. Stout, R.P. & Graziadei, P.P. Influence of the olfactory placode on the development of the
brain in Xenopus laevis (Daudin). I. Axonal growth and connections of the transplanted
olfactory placode. Neuroscience 5, 2175–2186 (1980).
14. Ang, L.H. et al. Lim kinase regulates the development of olfactory and neuromuscular
synapses. Dev. Biol. 293, 178–190 (2006).
15. Graziadei, P.P. & Kaplan, M.S. Regrowth of olfactory sensory axons into transplanted
neural tissue. 1. Development of connections with the occipital cortex. Brain Res. 201,
39–44 (1980).
16. Schneiderman, A.M., Matsumoto, S.G. & Hildebrand, J.G. Trans-sexually grafted
antennae influence development of sexually dimorphic neurones in moth brain. Nature
298, 844–846 (1982).
17. Oland, L.A. & Tolbert, L.P. Multiple factors shape development of olfactory glomeruli:
insights from an insect model system. J. Neurobiol. 30, 92–109 (1996).
18. Moon, R.T., Bowerman, B., Boutros, M. & Perrimon, N. The promise and perils of Wnt
signaling through beta-catenin. Science 296, 1644–1646 (2002).
19. Cadigan, K.M. & Nusse, R. Wnt signaling: a common theme in animal development.
Genes Dev. 11, 3286–3305 (1997).
20. Yoshikawa, S., McKinnon, R.D., Kokel, M. & Thomas, J.B. Wnt-mediated axon guidance
via the Drosophila Derailed receptor. Nature 422, 583–588 (2003).
21. Fradkin, L.G. et al. The Drosophila Wnt5 protein mediates selective axon fasciculation in
the embryonic central nervous system. Dev. Biol. 272, 362–375 (2004).
22. Srahna, M. et al. A signaling network for patterning of neuronal connectivity in the
Drosophila brain. PLoS Biol. 4, e438 (2006).
23. Zhang, D. et al. Misexpression screen for genes altering the olfactory map in Drosophila.
Genesis 44, 189–201 (2006).
24. Jhaveri, D., Sen, A. & Rodrigues, V. Mechanisms underlying olfactory neuronal connectivity in Drosophila-the atonal lineage organizes the periphery while sensory neurons
and glia pattern the olfactory lobe. Dev. Biol. 226, 73–87 (2000).
25. Fishilevich, E. & Vosshall, L.B. Genetic and functional subdivision of the Drosophila
antennal lobe. Curr. Biol. 15, 1548–1553 (2005).
26. Couto, A., Alenius, M. & Dickson, B.J. Molecular, anatomical, and functional organization of the Drosophila olfactory system. Curr. Biol. 15, 1535–1547 (2005).
27. Lee, T. & Luo, L. Mosaic analysis with a repressible neurotechnique cell marker
for studies of gene function in neuronal morphogenesis. Neuron 22, 451–461
(1999).
28. Dura, J.M., Taillebourg, E. & Preat, T. The Drosophila learning and memory gene linotte
encodes a putative receptor tyrosine kinase homologous to the human RYK gene
product. FEBS Lett. 370, 250–254 (1995).
29. Simon, A.F., Boquet, I., Synguelakis, M. & Preat, T. The Drosophila putative kinase
linotte (derailed) prevents central brain axons from converging on a newly described
interhemispheric ring. Mech. Dev. 76, 45–55 (1998).
30. Jefferis, G.S., Marin, E.C., Stocker, R.F. & Luo, L. Target neuron prespecification in the
olfactory map of Drosophila. Nature 414, 204–208 (2001).
31. Stocker, R.F., Heimbeck, G., Gendre, N. & de Belle, J.S. Neuroblast ablation in
Drosophila P[Gal4] lines reveals origins of olfactory interneurons. J. Neurobiol. 32,
443–456 (1997).
32. Hanks, S.K., Quinn, A.M. & Hunter, T. The protein kinase family: conserved features and
deduced phylogeny of the catalytic domains. Science 241, 42–52 (1988).
33. Taillebourg, E., Moreau-Fauvarque, C., Delaval, K. & Dura, J.M. In vivo evidence for a
regulatory role of the kinase activity of the linotte/derailed receptor tyrosine kinase, a
Drosophila Ryk ortholog. Dev. Genes Evol. 215, 158–163 (2005).
34. Schlaggar, B.L. & O’Leary, D.D. Potential of visual cortex to develop an array of
functional units unique to somatosensory cortex. Science 252, 1556–1560
(1991).
35. Lu, W., Yamamoto, V., Ortega, B. & Baltimore, D. Mammalian Ryk is a Wnt coreceptor
required for stimulation of neurite outgrowth. Cell 119, 97–108 (2004).
36. Toba, G. et al. The gene search system. A method for efficient detection and rapid
molecular identification of genes in Drosophila melanogaster. Genetics 151, 725–737
(1999).
37. Hofbauer, A. Eine Bibliothek monoklonaler Antikorper gegen das Gehirn von Drosophila
melanogaster. (University of Wurzburg, Wurzburg, 1991).
ADVANCE ONLINE PUBLICATION NATURE NEUROSCIENCE