Liqun Luo1,* and John G. Flanagan2,*
1Howard Hughes Medical Institute, Department of Biological Sciences, Stanford University, Stanford, CA 94305, USA
2Department of Cell Biology and Program in Neuroscience, Harvard Medical School, Boston, MA 02115, USA
*Correspondence: email@example.com (L.L.), firstname.lastname@example.org (J.G.F.)
the visual retinotopic map, and discrete maps exemplified by the olfactory glomerular map. Here, we
review developmental mechanisms of retinotopic and olfactory glomerular mapping and discuss
gradients, axon-axon interactions, and the interplay between labeling molecules and neuronal activ-
ity in establishing these maps. Since visual retinotopic and olfactory glomerular maps represent two
ends of a continuum that includes many other types of neural map in between, these emerging gen-
eral principles may be widely applicable to map formation throughout the nervous system.
The use of space to encode information is a fundamental
organizational principle of the nervous system. Neural
maps are used in all sensory modalities, motor control,
and many places in between (see other articles in this is-
sue). Two qualitatively different kinds of neural maps
have been described: continuous and discrete (Figure 1).
In continuous neural maps, nearby neurons in the input
field connect with nearby neurons in the target field,
thereby preserving spatial order (Figure 1, top). An exam-
ple is the retinotopic map in the visual system, where
a two-dimensional image on the retina is recast in higher
visual processing areas in the brain. In discrete neural
maps, the spatial organization of the one field reflects
discrete qualities, rather than the spatial organization, of
neurons in the other field (Figure 1, bottom). An example
tory receptor neurons that express the same odorant re-
ceptors project to the same glomerular units in the brain,
creating a spatial map of discrete information in the target.
Most neural maps either resemble one of these two
extremes or fall in between. Auditory projections form
a continuous tonotopic map, which is comparable to the
retinotopic map. Taste systems use discrete channels
for different taste modalities, resembling the olfactory
map. Somatosensory and motor maps contain both con-
tinuous and discrete components; for example, somato-
sensory maps typically consist of a continuous represen-
tation of the body surface, but embedded in this are
discrete units such as whisker barrels. Higher visual
maps can contain both a continuous representation of vi-
sual space and discrete embedded columnar units to rep-
resent features such as ocular dominance and direction of
motion. Historically, the term ‘‘topographic’’ was used to
describe continuous maps exemplified by the retinotopic
map; but this term has become extended to other maps
with a spatial component. We use here the term ‘‘continu-
ous’’ to make a distinction with discrete maps. When we
use ‘‘topographic’’ in this review, we treat it according to
its historic meaning (i.e., synonymous with ‘‘continuous’’).
How are neural maps formed during development? Are
maps? Here, we review recent advances on mechanisms
of map development. We focus specifically on the retino-
topic map in the visual system (best-studied continuous
map) and the glomerular map in the olfactory system
(best-studied discrete map). By comparing these two sys-
tems that represent extremes in the spectrum of continu-
ous and discrete maps, we attempt to identify emerging
common principles, as well as diversity of solutions, that
may apply to the formation of other neural maps through-
out the nervous system.
2. Development of Retinotopic Maps
Visualmapsprovidetheprototypic exampleof continuous
topographic mapping. Light falling on the retina creates
formation is then transferred to the brain has long been of
interest, and the idea that the image would be transferred
in a spatially intact form to visual centers in the brain was
first proposed in the 17th century (Jacobson, 1991). This
transfer is now known to involve a topographic organiza-
tion of axons (Figure 2). In discussing mechanisms for
map formation, we will first focus on a vertebrate system,
the well-characterized projection from the retina to the
midbrain tectum (or its mammalian equivalent, the supe-
rior colliculus), followed by a discussion of retinotopic
development in Drosophila.
2.1 Retinotectal Map Formation: Graded
In explaining how a continuous topographic map could
develop, Sperry first proposed in his chemoaffinity theory
that maps such as the retinotectal projection could be
specified by complementary positional labels in gradients
(Sperry, 1943, 1963). These labels would mark position
across both the retina and tectum, and the projecting
Neuron 56, October 25, 2007 ª2007 Elsevier Inc.
retinal axons would find their correct position in the tectal
gradient, based on their position in the retinal gradient.
The ephrin-As and their EphA receptors were subse-
quently identified as molecules that fulfill all the major cri-
teria to be complementary graded mapping labels. First,
the receptors and ligands show appropriate expression
patterns in the retinotectal system, being in complemen-
tary gradients across both the retina and the tectum
ond, when tested in vitro the ephrins act as guidance cues
for retinal axons (Drescher et al., 1995; Nakamoto et al.,
1996). And third, when tested in vivo by both gain- and
loss-of-function approaches, both the ligands and recep-
tors can change the map position where retinal axons
project (Nakamoto et al., 1996; Frise ´n et al., 1998; Brown
et al., 2000; Feldheim et al., 2000). The ephrins currently
remain as the only molecules validated by all these tests
as graded mapping labels. Importantly, the functional as-
says all show differential responses by axons from differ-
entpositions in the projecting area,anessential propertyif
a molecule is to act as a topographic label.
For the development of a two-dimensional map, it is
necessary to have labels along at least two distinct
axes. The ephrin family appears to neatly provide a solu-
tion to this problem. Theephrinsand Eph receptors are di-
vided into A and B subfamilies, which show preferential
bindinginteractions withinasubfamily (EphNomenclature
Committee, 1997; Klein, 2004). While the A subfamily acts
the B subfamily has a labeling function along the dorso-
ventral axis (Figure 3A) (Holash and Pasquale, 1995;
Braisted et al., 1997; Hindges et al., 2002; Mann et al.,
Although topographic mapping is usually described in
terms of labeling along two axes at right angles, there is
no inherent necessity for the labels to be simplemonopha-
sic gradients along orthogonal axes. Indeed, in the original
chemoaffinity theory Sperry predicted that in binocular
species the anteroposterior axis would be mapped by
labels in a central-to-peripheral gradient. His reasoning
was that this would allow the temporal side of one eye
and the nasal side of the other eye—which in binocular
species view the same part of the visual field—to map to
the same position in the target (Figure 2B) (Sperry, 1963).
Studies on the human visual system have now shown that
retinal EphA receptors form a gradient that is low at nasal
and temporal extremes, rising to a high point in the central
retina (Lambot et al., 2005). In addition to providing an ex-
planation for binocular mapping, this result emphasizes
that it is worth bearing in mind that other topographic
maps may not be labeled by simple Cartesian coordinate
2.2 Positive and Negative Effects of Ephrins
While the general concept of chemoaffinity labels can
explain the formation of a continuous topographic map,
Gierer pointed out in the 1980s that a single type of guid-
a gradient were repellent or attractant, the result would
simply be accumulation of all the axons at one end of
the target (Gierer, 1987). Gierer therefore proposed that
axons must detect two opposing forces in the target.
These forces might take various forms, for example an
attractant and a repellent, opposing gradients of repel-
lents, or a single molecule with both attractant and repel-
lent properties. Each axon would then come to rest at the
point where the opposing forces cancel out, allowing for-
mation of a continuous topographic map where different
axons would terminate at different points.
Ephrin-As are distributed in the tectum in a posterior >
anterior gradient, and initial studies led to a model where
they would act as repellents (Flanagan and Vanderhae-
ghen, 1998; O’Leary et al., 1999). This model leads to
a prediction: if the counterbalancing force in the target
were a fully independent gradient, then removal of a repel-
lent ephrin-A gradient should cause a tendency for axons
to shift toward abnormally posterior positions, or in the
case of extreme nasal axons would cause no shift.
When ephrin-A gene knockouts were examined, temporal
retinal axons were indeed found to form ectopic arbors in
abnormally posterior locations, as anticipated (Frise ´n
et al., 1998; Feldheim et al., 2000). However, examination
parable degree of shift, but in the opposite, anterior direc-
against a simple model where afully independent gradient
provides the counterbalancing force.
Although the first tests had detected only repellent
effects of ephrins (Drescher et al., 1995; Nakamoto et al.,
1996)—and it is still sometimes assumed that ephrins
must act as repellents—many subsequent studies have
shown that, like a number of other guidance molecules,
Figure 1. Continuous and Discrete Neural Maps
(Top) Schematic of continuous neural map exemplified by the retino-
topic projections of the visual system. Nearby neurons inthe input field
connect with nearby neurons in the target field, thereby preserving the
spatial order of the visual image.
(Bottom) Schematic of discrete neural map exemplified by the glomer-
ular map in the olfactory system. Spatial organization in the target field
olfactory receptor neurons that express a common odorant receptor,
which converge their axonal projection to a common glomerulus.
Neuron 56, October 25, 2007 ª2007 Elsevier Inc.
ephrins are bifunctional. In assays of cell adhesion, cell
migration, and axon guidance, both negative/repellent
and also positive/attractant effects have been demon-
strated for both ephrin-As and ephrin-Bs (Huynh-Do
et al., 1999; Holmberg et al., 2000; McLaughlin et al.,
2003a; Weinl et al., 2003; Hansen et al., 2004; Matsuoka
et al., 2005; Weinl et al., 2005; Halloran and Wolman,
2006; Zimmer et al., 2007).
A crucial feature of this bifunctionality, in the case of the
ephrins, isthat itisconcentration dependent. When retinal
axons respond in vitro to ephrin-A2, low concentrations of
the ligand cause a several-fold promotion of outgrowth,
whereas high concentrations cause a complete inhibition.
Moreover, the balance of positive and negative effects
shows a systematically graded dependence on both eph-
ephrin-As would contribute both positive and negative
forces in mapping (Figure 3B) (Hansen et al., 2004).
Ephrin-Bs are also thought to have bifunctional actions
a phenotype consistent with attraction (Hindges et al.,
2002), whereas overexpression of patches of ephrin-B1
in chick caused repulsion (McLaughlin et al., 2003a).
Although the nature of the evidence for ephrin-B bifunc-
tionality was quite different from that for ephrin-As, it led
to a similar model, where axon branches located too
high or too low in the ephrin-B gradient would be respec-
tively repelled or attracted toward their topographically
correct D-V position (Figure 3B) (McLaughlin and O’Leary,
The molecular mechanism for bifunctionality could in
principle involve transduction into the axon of attractant
signals by some Eph receptors and repellent signals by
others. Alternatively, it may involve adhesion at low ligand
concentrations, and repellent signal transduction at high
ligand concentrations, and currentevidence appears con-
sistent with this second model (Holmberg et al., 2000;
Hansen et al., 2004; Matsuoka et al., 2005; Halloran and
Wolman, 2006). Signaling can also cause proteolytic dis-
ruption or endocytosis of the ligand-receptor complex,
and this may help to explain how the Eph-ephrin interac-
tion can promote adhesion at low concentrations (or low
oligomerization states) and cause repellent signaling at
high concentrations (Hattori et al., 2000; Zimmer et al.,
2003; Janes et al., 2005).
nized such that axons can be attracted by a low concen-
tration of ephrin and repelled by a high concentration (Fig-
ure 3B). However, a difference is that the high point on
the Eph gradient maps to the low point on the ephrin-A
gradient, but to the high point on the ephrin-B gradient
(Figure 3A). We suggest that this does not necessarily re-
flect a fundamental difference in the underlying molecular
Figure 2. Topographic Organization of Retinal Projections
tinuous topographic map, with spatial order preserved along both
anteroposterior (A-P; illustrated here) and dorsoventral (D-V; not
shown) axes. To avoid confusion in binocular species, the A-P axis of
the retina is commonly termed nasotemporal (N-T). (A) In species
such as mouse and chick, the two eyes view divergent parts of visual
mammalian equivalent, thesuperior colliculus),forming amap with na-
sal axons terminating in posterior tectum and temporal axons in ante-
rior tectum. In map development, EphA and ephrin-A molecules
(blue) act as graded complementary labels along the N-T and A-P
axes. EphB and ephrin-B molecules (red) serve an early function in se-
lective chiasm crossing (illustrated here), in addition to their later func-
tion in D-V mapping (see Figure 3). (B) In binocular species such as
humans, both eyes point in the same direction. The nasal half of each
ally. The two eyes form maps in register, with the temporal extreme of
one eye and the nasal extreme of the other eye—which view the same
point in visual space—mapping to the same position in the target.
During development, the retinal EphA gradient is highest in central
extremes, providing a coordinate system for the two eyes to map in
register. Diagrams are modified from Lambot et al., 2005.
(C and D) Retinotopic connections in Drosophila.The retina consists of
tains eight photoreceptor cells, R1–8, in a pattern with R8 located
below R7. The retina projects to two ganglia, the lamina and the me-
dulla, which also consist of orderly arrays of units, respectively called
cartridges and columns. Each cartridge receives input from the six
R1?6 neurons insix adjacent ommatidia, whichdetectlight from asin-
gle point in visual space. Projections to the medulla form a chiasm that
put from one R7 and one R8 neuron, as well as input from a cluster of
five interneurons located in the lamina, L1–5. The result of this highly
each receive input from a single point in visual space and form highly
ordered topographically mapped arrays.
Neuron 56, October 25, 2007 ª2007 Elsevier Inc.
explanation of bifunctionality. Instead, it may only reflect
the quantitative response curves of each receptor to ad-
hesion and repulsion. If increasing receptor concentration
the adhesion term to rise above the repulsion term, then
high receptor would map to high ligand.
Interestingly, a topographically specific guidance mole-
cule in a different family, Wnt3 (Figure 3A), which will be
discussed further below, also shows this property of con-
centration-dependent bifunctionality. Furthermore, like
ephrin-A2, the balance of positive versus negative effects
of Wnt3 in vitro was found to vary systematically with both
ligand concentration and retinal position (Schmitt et al.,
2005). For both ephrin-B1 (McLaughlin and O’Leary,
2005) and Wnt3 (Schmitt et al., 2005), the gradients ap-
pear to act in the same direction, with higher concentra-
tions located dorsally and proposed to cause repulsion,
in which case it does not seem that they act as gradient
forces opposed to oneanother inany simple way; instead,
they may act as a partially redundant system of cooperat-
ing labels to make dorsoventral mapping more robust.
Concentration-dependent bifunctionality, which is seen
in ephrin-As, ephrin-Bs, and Wnt3, can provide a unified
mapping mechanism along both A-P and D-V axes and
appears to be a common feature of graded mapping
cues. This can allow them to fulfill Gierer’s postulate by
not merely pushing or pulling axons in one direction, but
rather by specifying a position where each axon comes
to rest in the gradient.
2.3 Eph and Ephrin Countergradients
A notable feature of ephrin and Eph expression patterns is
that they generally take the form of countergradients
within an area. In the retina, for example, EphA receptors
are found in a temporal > nasal gradient, while ephrin-A
ligands are found in a nasal > temporal gradient
Figure 3. Gradients, Bifunctional
Guidance Mechanisms, and Neural
Activity in the Retinotectal Projection
(A) Gradients of guidance cues and their recep-
tors in the retinotectal system. Map topography
is shown at the left. Axons map from the retina
to the tectum of the midbrain, also known as the
superior colliculus in mammals. A, anterior; P,
distinct responses by axons from different parts
of the retina), although only the ephrins/Ephs
are known to be required as tectal guidance
on studies of chick, mouse, and Xenopus. Each
Eph/ephringradient illustratedistypically acom-
tectal ephrin-A gradient contains ephrin-A2,-A3,
and -A5. RGM and its receptor neogenin are in
complementary gradients. Wnt3 is graded in the
tectum, and its receptors are in the retina: Ryk
isatyrosine kinase receptorthat mediates repul-
transmembrane receptors and mediate the
attraction seen at lower Wnt3 concentrations.
En-2 attracts nasal and repels temporal axons
in vitro, apparently by crossing the cell mem-
tracellular retinal receptors.
(B) Bifunctional guidance mechanisms. Ephrin-As, ephrin-Bs, and Wnt3 all show bifunctional effects on retinal axons. Low concentrations cause positive/
attractant effects whereas high concentrations cause negative/repellent effects. This leads to a model for mapping along both A-P and D-V axes (left and
right diagrams) where axons find their final position at the neutral point between attraction (green bars) and repulsion (red bars). On the A-P axis, ephrin-
As are proposed to specify a neutral point that varies depending on the retinal position of axon origin (upper and lower diagrams) (Hansen et al., 2004).
high pointon the receptor gradient maps to thehigh point on the ligand gradient (Figure 3A). However, this need not reflect a fundamental difference in the
underlying molecular explanation for concentration-dependent bifunctionality (see text).
arborization at the topographically corresponding position in the target. In ephrin-A double and triple gene knockout mice, both nasal and temporal axons
form ectopic terminations scattered along the A-P axis, but they still arborize in tight foci. Mice mutant for the b-2 subunit of the nicotinic acetylcholine
receptor, which lack early spontaneous retinal waves, show arborizations that are abnormally diffuse, but located in the topographically correct area.
suggestingthepresenceof additionalgradedlabels.Theseresultssupport amodelwheretopographicorder isset upbygradedlabelsand refined bypat-
terned neural activity.
Neuron 56, October 25, 2007 ª2007 Elsevier Inc.
(Figure 3A). There are several potential roles for this coun-
inal neurons would interact in cis with EphA receptors,
downregulatingtheiractioninnasal retina. Thisideaissup-
ported by compelling evidence from several studies. Eph-
rin-A overexpression in the retina causes temporal axons
to lose their sensitivity to ephrin-A repulsion (Hornberger
et al., 1999). Correspondingly, ephrin-A gene knockout
greatly increases the sensitivity of explanted nasal retinal
axons to ephrin-A repulsion (Feldheim et al., 2000). Bio-
distinctfromthe binding sitefor trans receptor-ligand inter-
lating EphA receptors preferentially at the nasal extreme of
the retina, the effect of the cis interaction may be to make
the functional gradient of retinal EphA receptors steeper,
perhaps serving to enhance the precision of mapping.
A second role for the countergradients comes from the
ability of ephrins and Eph receptors to signal bidirection-
ally, with a ‘‘forward’’ signal transduced into the cell carry-
ing the Eph and a ‘‘reverse’’ signal transduced into the cell
carrying the ephrin (Klein, 2004). As a result, ephrins on
projecting axons and Eph receptors in the target can
also contribute to mapping. Along the retinotectal antero-
posterior axis, EphA7 is in an anterior > posterior gradient
in the target and can repel mouse retinal axons in vitro and
suppress arborization in vivo (Rashid et al., 2005). Along
the dorsoventral axis, EphB receptors are expressed in
a ventral > dorsal gradient in the Xenopus tectum and
can have an attractant effect in vivo and in vitro for dorsal
retinal axons (Mann et al., 2002).
Taken together, the previous two paragraphs raise
a question: if ephrins interact in cis with Ephs and down-
regulate them, can both Ephs and ephrins be active as re-
ceptors in the same axons? One potential resolution to
this paradox comes from a study of motor neurons, where
immunolocalization experiments provided evidence that
ephrin-As and EphAs do not interact in cis, but rather seg-
regate into separate microdomains where they may signal
independently (Marquardt et al., 2005). Superficially,
these results seem to contradict the studies described
above finding that interactions do occur in cis, including
at the cell surface. While the explanation for these differ-
ences is not yet clear, it seems possible that both models
are correct. For example, different neuronal types may
show cis or trans interactions. Alternatively, a subset of
the molecules on a single axon may interact in cis and
mediate downregulation, whileanothersubsetmaysegre-
gate and act independently as guidance receptors.
Finally, a third reason for the expression of both Ephs
and ephrins within the same area is that most regions of
the nervous system act both as projecting areas and as
target areas. The expression of countergradients of both
ephrins and Eph receptors may be important in allowing
a single area to serve as both the recipient of mapped in-
coming axons and the origin of mapped outgoing axons.
This feature may be critical for the serial (Feldheim et al.,
1998; Cang et al., 2005), parallel (Feldheim et al., 1998),
and reciprocal (Torii and Levitt, 2005) transfer of topo-
graphically mapped information among multiple intercon-
nected areas of the nervous system (Flanagan, 2006).
2.4 Axon-Axon Competition: Filling the Target
A notable feature of continuous topographic mapping is
that projecting axons tend to smoothly fill up the target.
This is seen notonlyin normal development, but even after
fairly drastic experimental manipulations such as removal
of half the retinaor tectum,which canbefollowed by grad-
Goodhill and Richards, 1999), or after genetic overexpres-
heim et al., 2000). These results show that mapping labels
trations along the gradients, but rather by determining the
relative position of axons. The tendency to fill the available
space is then explained by axon-axon competition. This
competition could in principle result from direct axon-
axon interactions or could result from axons competing
for one or more limiting factors in the target.
While the mechanism for competition is not well under-
stood at the molecular level, there are some candidates.
The neurotrophin BDNF and its receptor trkB are ex-
pressed in the tectum and retina, respectively, although
not in gradients. Increasing or decreasing the amount of
BDNF in the tectum causes expansion or reduction,
hen-Cory and Fraser, 1995). These properties seem to fit
the description of a limiting factor in the target that axons
get has not been shown directly. Another candidate is L1,
a member of the immunoglobulin-related cell adhesion
molecule (IgCAM) family. L1 gene disruption causes ab-
normalities in mapping of both temporal and nasal axons
when traced by focal dye injections. This mutant also has
axons do not fill the tectum smoothly, but instead leave
large irregularly shaped patches with a reduced density
of axon terminals (Demyanenko and Maness, 2003). L1 is
thus required to smoothly fill the target and is an excellent
candidate to mediate axon-axon competition.
It seems very possible that axon competition may be
mediated by several different molecules in combination,
potentially including IgCAMs, neurotrophins, and other
factors that may be limiting in the target or mediate repul-
sion (see sections 2.7 and 3.3 for molecules that may have
analogous functions). These molecules could act as ge-
netically specified labels or via activity-dependent mech-
anisms (see section 2.6); for instance, transcription of
BDNF is regulated by neural activity (West et al., 2001). In-
deed, a role for neural activity in competition is supported
by the finding that growth of retinotectal axon arbors is
inhibited when their activity is suppressed below that of
active neighbors (Hua et al., 2005). Whatever the specific
molecules and mechanisms involved, it is not difficult to
Neuron 56, October 25, 2007 ª2007 Elsevier Inc.
see the advantages of a mapping mechanism that oper-
ates by a combination of graded positional labels that de-
termine relative position, together with axon competition.
Thiscombination can providea flexible and robustsystem
to generate a smooth and orderly topographic map
throughout the target, despite changes in parameters
such as the concentration of the labeling molecules or
the size and shape of the map, when they vary during
development or evolution.
2.5 Additional Graded Molecules
in the Retinotectal Projection
Removal of all the known tectal ephrin-As by double and
triple gene knockout, in combination with activity disrup-
tion almost completely abolishes topographic order, con-
firming the importance of ephrins in mapping. However,
even in these conditions, some topographic bias remains,
supporting the presence of additional labeling mecha-
nisms (Feldheim et al., 2000; Pfeiffenberger et al., 2006).
One candidate is RGM, which is in a posterior > anterior
tectal gradient and has topographically specific repellent
effects on temporal axons in vitro (Monnier et al., 2002).
No retinotectal mapping phenotype was detected in
RGM knockout mice (Niederkofler et al., 2004), although
it remains possible that a role could have been masked
by redundancy with other cues, such as the ephrin-As.
Sema3D is expressed in the ventral zebrafish tectum,
and although it is not clear whether this distribution is
gradedordiscrete, changingthelevels ofSema3Dcauses
In addition to these cues that were initially identified as
guidance molecules, other studies have shown effects on
retinal axon guidance by two families of molecules that
had previously been best known for functions quite differ-
ent from guidance. The Wnts are cell-cell signaling mole-
cules that have been studied extensively for their effects
on cell fate, including actions as graded morphogens. In
the retinotectal system, Wnt3 is expressed in a dorsal >
ventral gradient in the target (Figure 3A). It was found
to have topographically specific effects on retinal axons
in vitro, with axon inhibition mediated by the Ryk receptor
and promotion mediated by Frizzled receptors, and it was
also found to affect mapping in vivo (Schmitt et al., 2005).
While their requirement has not yet been tested by gene
disruption, these studies support a role of Wnt3 and its
receptors in dorsoventral mapping.
An even more striking example of a molecule known for
very different functions having axon guidance activity
came from a study of the homeodomain transcription fac-
tor En-2. A number of previous studies had shown that En
proteins are expressed in the tectum in a posterior > ante-
rior gradient and regulate tectal cell fate, including the
downstream expression of ephrin-As (Logan et al., 1996;
Retaux and Harris, 1996). It had also been shown that
En misexpression causes mapping abnormalities, which
were explained in terms of its effect on ephrin-A expres-
sion (Friedman and O’Leary, 1996; Logan et al., 1996). Al-
though homeoproteins are best known as nuclear factors,
studies by Prochiantz and colleagues over the last 15
taken up by cells in culture, suggesting the potential to act
in cell-cell signaling (Joliot and Prochiantz, 2004). Soluble
En-2 was therefore tested for a direct effect on axons and
was found to act as a guidance factor for Xenopus retinal
axons in vitro. Moreover, it showed appropriate topo-
graphic specificity, attracting nasal axons and repelling
temporal axons (Brunet et al., 2005). In the future, it will
be interesting to see whether En proteins are transferred
to axons in intact tissues and to test the requirement for
En as a guidance cue in vivo, which may be inherently
challenging in view of its known function in regulating tec-
tal cell fate. Nevertheless, the studies on Wnt3 and En-2
provide new examples of the emerging principle that posi-
tional information gradients provide coordinate systems
that can be interpreted to regulate essentially any cell
function, depending on the receptor systems used to
decode this positional information (Osterfield et al., 2003).
2.6 Activity-Dependent Refinement
In explaining the development of neural connectivity, che-
moaffinity labels versus neural activity were historically
viewed as competing theories (Jacobson, 1991). How-
anisms are both important and play complementary roles.
In the retinotectal system, graded labels initially set up the
topographic order of the map, whereas neural activity re-
fines the map to increase its precision (Zhang and Poo,
2001; Ruthazer and Cline, 2004; McLaughlin and O’Leary,
2005; Torborg and Feller, 2005).
The theory underlying most studies of neural activity in
aptic neuron fire at similar times, the connection between
them is reinforced (Hebb, 1949). This could lead to refine-
ment of a continuous topographic map if neurons that are
adjacent in the projecting area fire in a correlated manner,
neuron and therefore strengthening their connection to it.
On the other hand, a neuron located more distantly would
show less correlated activity, and its connection to the
same postsynaptic neuron would not be strengthened, or
may even be weakened. The correlated activity necessary
for this Hebbian mechanism could result from visual
phibians, which are visually active while the map forms. In
birds and mammals, the correlation may instead be pro-
vided by waves of activity that are found to sweep across
the retina spontaneously during map development (Mei-
ster et al., 1991; Wong et al., 1993).
One line of evidence for the role of neural activity in the
retinotectal system has come from inhibitors of neural ac-
tivity. The use of activity blockers such as tetrodotoxin
showed that activity is not required to establish the basic
topographic layout of the map (Harris, 1980) but is re-
quired for refinement, narrowing down the area covered
by axonal connections in the target (Schmidt and Ed-
refinement (Cline, 1998).
Neuron 56, October 25, 2007 ª2007 Elsevier Inc.
While these studies show that activity is required, they
do not address whether its role is permissive or instruc-
tive. If the role is to be instructive, a key factor is to
show a requirement for suitable patterning of the neural
activity. One approach to this has been to force two
eyes in an amphibian to innervate the same tectal lobe.
The result is a segregation of their connections into ocular
dominance bands, and since the two eyes presumably
have the same chemoaffinity labels, it can be inferred
that the segregation results from correlated activity within
each eye (Constantine-Paton and Law, 1978; Ruthazer
et al., 2003). A second approach has been to raise fish
in stroboscopic illumination, so that retinal neurons fire
in a coordinated manner, but independently of their retinal
position, and this is found to cause a loss of map refine-
ment (Schmidt and Eisele, 1985). Another approach has
been to place electrodes simultaneously in the Xenopus
retina and tectum. In this way it could be shown that the
synaptic connection between the pre- and postsynaptic
cells is strengthened or weakened, respectively, when
the presynaptic neuron fires within a narrow time window
before or after the postsynaptic neuron, providing direct
evidence for the basic Hebbian proposal (Zhang et al.,
1998). These varied lines of evidence provide strong sup-
port for an instructive role of neural activity.
In assessing the relative contributions of labels and ac-
tivity, genetic studies have been informative (Figure 3C).
Following disruption of ephrin-A genes, retinal axons la-
beled by focal dye injection make connections that are
scattered along the anteroposterior axis of the target,
but nevertheless form tight termination zones (Frise ´n
et al., 1998; Feldheim et al., 2000). In contrast, mice with
a disruption of the b-2 subunit of the neuronal nicotinic
acetylcholine receptor, which do not display spontaneous
retinal waves during the early phase of map refinement,
form retinal axon connections in the topographically cor-
rect place, but abnormally diffuse (McLaughlin et al.,
2003b; Chandrasekaran et al., 2007). When the ephrin-A
mutations and the b-2 mutation are combined, the result
is labeling of a single diffuse termination zone covering
most of the target (Pfeiffenberger et al., 2006).
Current studies thus support a model where graded
labels and patterned neural activity have distinct instruc-
tive roles in respectively setting up topographic order
and refining it. However, the two processes have interest-
ing areas of overlap; for example, it was recently shown
during retinotectal map formation (Nicol et al., 2007), and
the ephrins are known to act in synapse formation and
may in turn regulate activity (Yamaguchi and Pasquale,
2004). Such observations suggest the potential for inter-
esting mechanistic links at the interface between label-
and activity-based mapping.
2.7 Retinotopic Mapping in Drosophila
Although the fly eye has a very different structure from its
vertebrate counterpart, it shares the hallmark property of
projecting to visual centers in the brain asa continuous to-
pographic map. Each eye in D. melanogaster consists of
a crystalline hexagonal array of approximately 750 omma-
tidia, each containing eight photoreceptor neurons, R1–8
(Figures 2C and 2D). Each ommatidium sends its axons
ing topographic maps. The lamina and medulla are them-
respectively as cartridges and columns. The axonal con-
nections between these units form in a highly stereotyped
pattern so that each cartridge and column corresponds to
andZipursky, 2002;Mastetal.,2006;TingandLee, 2007).
The fly visual system has been a rich source for the identi-
fication of axon guidance mechanisms; here, we focus on
mechanisms to specify topographic order.
The formation of a topographic map where neighboring
ommatidia project to neighboring targets in the lamina is
thought largely to reflect the spatiotemporal order of om-
matidial assembly. As the morphogenetic furrow sweeps
entiate andsend outsuccessive rows ofaxons. Theaxons
themselves organize the lamina, delivering two signaling
molecules, Hedgehog and Spitz. These signals induce
proliferation and differentiation of lamina neurons, which
then associate with the incoming axons to form cartridges
(Huang and Kunes, 1996; Huang et al., 1998). This mech-
anism fulfills at least two purposes. First, it can ensure
a matching number of units in the retina and its target.
And second, by maintaining the spatiotemporal order of
successive rows of axons, the result is generation, at least
along the anteroposterior axis, of a topographically map-
correct target units, evidence for an important spatial com-
ponent comes from embryological manipulations. For ex-
ing rotation in the projection pattern of its axons, indicating
mined by the direction of axon outgrowth rather than by
specific labels in the target (Clandinin and Zipursky, 2000).
A number of genes have been identified that are re-
quired for the axons to project in an organized manner
(Clandinin and Zipursky, 2002; Mast et al., 2006; Ting
and Lee, 2007). Of particular interest for their role in medi-
ating axon-axon interactions are cell-surface molecules in
cadherin, is required for the correct local topography of R
axon projections in the lamina and medulla and probably
acts at least in part by homotypic repulsive interactions
to ensure correct spacing of the axons (Lee et al., 2003;
Senti et al., 2003). In a recent study, Dscam2, a member
of the IgCAM family related to the highly diverse recogni-
tion molecule Dscam, was suggested to be a homophilic
repulsion molecule, and when the Dscam2 gene was spe-
cifically deleted from either the L1 axon or its neighbors,
the L1 axon no longer targeted to a single column in the
medulla but instead could innervate several neighboring
columns, disrupting the normal regularly tiled pattern of
connections (Millard et al., 2007). These studies show
Neuron 56, October 25, 2007 ª2007 Elsevier Inc.
a critical role for axon-axon interactions in maintaining
precise local topography.
Are global graded labels used in the fly visual system?
Although Drosophila has a single Eph receptor, no clear
evidence has been found for a guidance role in retinotopic
mapping (Boyle et al., 2006). A member of the Wnt family,
lamina and was required for ventral targeting of retinal
axons in an attractant mechanism dependent on the re-
ceptor Dfrizzled2. While this indicates a labeling mecha-
nism analogous to the retinotectal system, the distribution
of Dwnt4 appeared to have a sharp boundary, rather than
being graded (Sato et al., 2006). It thus remains an open
question whether fly retinotopy may use graded chemoaf-
finity labels. Instead of a primary role for global labels, fly
retinotopy may instead rely primarily on the repeated
use of local organizing cues, a strategy that may be well
suited to its regularly repeated matching arrays.
Finally, the role of neural activity seems to be a differ-
ence between fly and vertebrate retinal mapping. In the
fly, extensive testing of mutants has so far failed to find
a role of neural activity in setting up the pattern of retino-
topic connectivity (Hiesinger etal.,2006).Thus, compared
with vertebrate systems that exploit neural activity for
refinement and adaptation, the highly organized and ste-
reotyped retinal connections in the fly may be specified
primarily by precise genetic hardwiring.
3. Development of the Olfactory Glomerular Maps
The olfactory system detects odorants in the chemical
world to convey information regarding important matters
such as food, predators, and potential mates. How does
the brain represent this chemical world and process infor-
mation that is of special behavioral value? Compared to
centuries of studies on the visual map, key insights about
the olfactory system organization were obtained only after
the molecular cloning of odorant receptors less than two
decades ago (Buck and Axel, 1991). Here, we focus on
recent advances on the study of olfactory glomerular
maps in mice and flies.
3.1 Organization of the Olfactory Systems
in Mice and Flies
Following the cloning of odorant receptors, it was found
that each olfactory receptor neuron (ORN, also called
olfactory sensory neurons) expresses a single odorant
receptor (OR). ORNs that express the same odorant re-
ceptor (OR)—defined hereafter as belonging to the same
ORN class—converge their axonal projections onto the
same glomeruli. Remarkably, this organizational principle
applies from mammals to insects (Ressler et al., 1994;
Vassar et al., 1994; Mombaerts et al., 1996; Gao et al.,
2000; Vosshall et al., 2000) (Figures 4A and 5A). Thus,
the targets of ORN axons are organized as discrete units
and form a spatial map. These discrete units represent
discrete input channels at the periphery—different ORN
Furthermore, in most species examined to date there is
a clustering of glomeruli that respond to odorants bearing
particular chemical groups (Mori et al., 2006).
Cell bodies of a given ORN class are distributed widely
along the sensory epithelia, intermingled with cell bodies
of other ORN classes. One can envision that such
Figure4. Organization and Development
of the Mouse Olfactory Map
(A) (Top) Thousands of olfactory receptor neu-
rons (ORNs) in the olfactory epithelium (OE) that
express a common odorant receptor (OR) con-
that express a given OR are distributed within a
responding color-matched positions along the
dorsal-ventral (D-V) axis of the OB. According to
data from Miyamichiet al. (2005). Dotted rectan-
gles correspond to OB schematics in (B) and (C)
as indicated.A, anterior;P, posterior; D,dorsal;
V, ventral. The A-P axis in bottom left (nasal
epithelium) is orthogonal to the plane shown.
(B) Basal level G protein (three purple subunits)
coupling of individual ORs, through the adeny-
late cyclase activation of cAMP, PKA, and
CREB, induces gene expression that contrib-
utes to ORN axon targeting globally along the
anterior-posterior (A-P) axis. Specifically, puta-
tive axon guidance receptors have been found
to be targets of this signaling pathway and
exhibit graded distribution along the A-P axis
in the OB. According to data from Imai et al.
(2006). ‘‘?’’s denote that OR or cAMP/PKA
could regulate global A-P targeting through
mechanisms independent of gene regulation.
(C) OR-correlated and activity-dependent expression of homophilic cell adhesion molecules Kirrel2 and Kirrel3, and repulsive axon guidance ligand/
receptor pairephrin-Aand EphA,couldcontributetolocalaxonalconvergence and sorting.Accordingto datafromSerizawaet al.(2006). ‘‘?’’s denote
that OR and activity could play additional roles.
Neuron 56, October 25, 2007 ª2007 Elsevier Inc.
the literature that the distribution of cell bodies of ORNs
expressing a given OR is random. However, this is not
accurate for mice or flies. ORNs in the mouse are distrib-
uted in a convoluted two-dimensional nasal epithelium
that can nevertheless be simplified with two axes, ante-
rior-posterior (A-P) and dorsomedial-ventrolateral (DM-
VL) (Figure 4A). Interestingly, while the distribution of cell
bodies belonging to the same ORN class appears
‘‘random’’ along the A-P axis, the distribution along the
DM-VL axis is more orderly. Early in situ hybridization
experiments identified several discrete zones along the
in one of these discrete zones (Ressler et al., 1993; Vassar
et al., 1993). Recent experiments examining many more
ORN classes indicate that OR genes are expressed in
multiple overlapping bands along the DM-VL axis, which
Figure 5. Organization and Development of the Fly Olfactory Map
(A) Corresponding positions of ORN cell bodies in theantenna and maxillary palp (top) and theirglomerular targetsinthe antennal lobe (bottom). Each
colored dot represents a sensillum, which houses one to four distinct ORN classes and belongs to a specific type shown on the right. The coarse
organization of glomerular position according to sensillum type is evident. Taken from Couto et al. (2005).
(B) (Top) Olfactory projection neurons (PNs) derived from the anterodorsal neuroblast (adNb) project their dendrites to glomeruli that are complemen-
tary tothosefromthelateralneuroblast (lNb).(Bottom)SchematicsummaryofglomerulithataretargetsofPNsderived fromadNblineage(green) and
along the A-P axis are shown. Modified after Jefferis et al. (2001).
(C) Three examples showing that ORNs belonging to the same sensillum can project to glomeruli that are spatially close or far apart. Each ORN class
can be classified as Notch-On or Notch-Off. In the absence of Notch signaling, Notch-On ORNs project axons to glomeruli that are targets of the
Notch-Off ORN class within the same sensillum. Taken from Endo et al. (2007).
have extended their main axon trunks to higher brain centers but have just started to extend rudimentary dendrites into the area that will become the
adult antennal lobe. (2) Between 0 and 18 hr after puparium formation (APF), PN dendrites elaborate and expand within the antennal lobe while ORNs
are being born. By 18 hr APF, PN dendrites already occupy the approximate areas of the antennal lobe according to their future glomerular classes
(symbolized by different colors); pioneering antennal ORNs have just reached the edge of the antennal lobe. (3) At 32 hr APF, pioneering axons from
the maxillary palp (MP) ORNs reach the antennal lobe, which has been patterned by antennal ORN axons in addition to PN dendrites. (4) Synaptic
matching gives rise to the adult pattern of connectivity. See text for mechanisms that regulate each of these steps.
Neuron 56, October 25, 2007 ª2007 Elsevier Inc.
on average span one-quarter of the nasal epithelium. Fur-
ther, there is a strong correlation between the location of
the visual map. However, given that there are 1000 ORN
must be a lot of ‘‘salt and pepper’’ mixing of cell bodies of
different ORN classes even along the DM-VL axis.
In Drosophila, ORN cell bodies are organized into spe-
cialized structures called sensilla on each antenna and
maxillary palp. Each sensillum houses one to four ORNs
belonging to one to four specific ORN classes in a stereo-
typed fashion (de Bruyne et al., 1999, 2001). The cell body
positions and the glomerular targets for most of the ?50
ORN classes have been mapped (Couto et al., 2005; Fish-
ilevich and Vosshall, 2005). There is coarse organization
with regard to the nature of the odor and corresponding
sensillar types. For instance, ORNs belonging to the tri-
choid sensilla, which usually respond to pheromones,
cluster their cell bodies in one area of the antenna; basi-
conic ORNs, which usually respond to fruity odorants,
occupy other areas of the antenna. In the antennal lobe,
trichoid ORN targets are located in the lateral anterior re-
gion of the antennal lobe, whereas basiconic ORNs from
in other parts of the antennal lobe (Figure 5A). Beyond this
coarse organization, however, cell bodies of individual
ORN classes intermingle with those of other ORN classes
belonging to the same sensillar type. There is no clear to-
classes that pair their cell bodies within the same sensil-
lum target their axons to distinct glomeruli that can be
far apart in the antennal lobe (Figure 5C).
Information from ORNs is carried to higher olfactory
centers by second-order olfactory neurons: mitral cells
in vertebrates and projection neurons (PNs) in insects. In
both mice and flies, each mitral cell or PN sends dendrites
tecture of these second-order neurons is quite different
between the species. Specifically, mitral cells form a layer
underneath the glomerular layer in the vertebrate olfactory
bulb and tend to innervate a glomerulus close to their cell
bodies. Insect PN cell bodies are located adjacent to but
outside of the antennal lobe neuropil, which is composed
only of dendritic, axonal, and glial processes. In Drosoph-
ila, there is no discernable correspondence between the
positions of PN cell bodies and their glomerular targets.
Further, PN dendrite targeting appears to be specified
by their lineage and birth order, independent of their syn-
aptic partner ORN axons (Jefferis et al., 2001) (Figure 5B;
see below). Indeed, terminal arborization of PN axons in
one of the higher brain centers exhibit striking stereotypy
corresponding to their glomerular classes (Marin et al.,
of PNs independent of their input neurons could ensure
nected with specific downstream circuits that process
qualitatively distinct information, such as food and mating
pheromone (Jefferis et al., 2007).
In summary, in both flies and mice, the correspondence
of cell body and glomerular positions of ORNs can be de-
scribed as a mixture of two organizations. First, there is
a coarse topography of ORN classes according to cell
body positions along the DM-VL axis in the mouse and
for different sensillar groups in the fly. Second, there is
a position-independent component that correlates with
the identity of the ORNs; this applies to the course organi-
zation of the A-P axis and the fine-grained glomerular or-
ganization of both A-P and D-V axes in mice, and to
ORN classes belonging to the same sensillar group and
PNs in flies. It is this second organization that is qualita-
tively different from that of the continuous maps of the vi-
sual system. At least in flies there is also the additional
challenge that each PN needs to target its dendrites to
specific areas of the antennal lobe that will eventually de-
velop into specific glomeruli. Below, we review recent
studies that provide insights into the mechanisms of
how such wiring specificity is established.
3.2 Mechanisms of ORN Axon Targeting in Mice
In mice, each ORN expresses a single odorant receptor
(OR) from ?1000 OR genes. ORNs expressing the same
OR converge their axonal projections to a pair of mirror-
image glomeruli at the medial and lateral side of the olfac-
tory bulb (Figure 4A). Although there are local variations of
glomerular positions (of a few glomerular distance) corre-
sponding to a given ORN class from animal to animal or
even from two sides of the same animal (Strotmann
et al., 2000), overall targeting is remarkably precise: axons
from the same ORN class find each other and converge
Themostintensely studied molecules forORN axontar-
geting are the ORs themselves. By replacing the coding
sequence of one OR (recipient) with another (donor) using
gene targeting, glomerular positions of the resulting ORN
axons are altered, often to a position in between the glo-
meruli corresponding to those of the donor and the recip-
ient (Mombaerts et al., 1996; Wang et al., 1998). These
experiments indicate that ORs play an instructive role in
2004), suggesting that they might act as guidance recep-
torsorhavesomeother functionin thegrowthconeduring
targeting remain enigmatic.
Significant advances have been made recently by ge-
netic engineering of G protein and cAMP signaling path-
ways in ORNs. Constitutive activation of Gs is sufficient
to rescue ORN axon convergence in the absence of a
functional receptor (Imai et al., 2006; Chesler et al.,
2007). Further, the glomerular positions of a specific
ORN class along the A-P axis in the bulb are correlated
with different strengths of cAMP/PKA signaling engi-
neered into these ORNs. This difference in cAMP/PKA
Neuron 56, October 25, 2007 ª2007 Elsevier Inc.
signaling further results in differential gene expression
likely regulated by the CREB transcription factor (Imai
et al., 2006). Putative target genes include neuropilin 1,
a receptor for the axon guidance molecule Sema3A previ-
ously implicated in restricting ORN axon targeting
derived from ORN axon terminals, form an A-P gradient in
the olfactory bulb correlating with the strength of the
cAMP signaling in ORNs whose axons target to different
A-P positions (Imai et al., 2006). It has thus been postu-
lated that ORNs expressing a given OR have a defined
level of G protein signaling, which can be translated into
certain amount of guidance receptor expression, thereby
instructing their axons to target defined areas in the olfac-
tory bulb (Figure 4B). This appealing model can account
for several peculiar previous findings, including that alter-
ations of OR expression level cause projection shift and
that expression of an unrelated G protein-coupled recep-
tor can rescue ORN axon convergence in the absence of
The involvement of cAMP signaling in ORN axon target-
ing was supported by recent adenylate cyclase (Ac3)
knockout studies (Col et al., 2007; Zou et al., 2007). In
Ac3 mutant mice, gross targeting of ORNs and conver-
gence of specific classes are defective. Consistent with
the findings in Imai et al. (2006), targeting to posterior
bulb was more severely affected, and neuropilin 1 expres-
sion is drastically reduced in Ac3 mutant mice (Col et al.,
pilin 1 and/or other cAMP/CREB targeting genes cause
predictable shifts of glomerular position along the A-P
axis. In addition to regulating gene expression through
CREB, cAMP/PKA signaling is well known for regulating
axon guidance responses at the growth cone (Ming et al.,
1997).Since ORs arepresent atthe ORNaxons and termi-
environmental cues in different ORN classes.
Given that many values of A-P positioning need to be
whether a gradation of G protein/cAMP signaling levels
alone could accountfor suchspecification, orwhether ad-
ditional forces are necessary. Regarding the topographic
targeting along the D-V axis, consistent with the continu-
ous nature at the coarse level, several molecules, includ-
ing transcription factors and cell-surface receptors, are
expressed as gradients along this axis (Norlin et al.,
2001). A recent study provided functional evidence that
Slit-1 and Robo-2, a classic ligand-receptor pair for repul-
sive axon guidance, are required for axons from ORNs lo-
cated in dorsomedial OE (expressing high levels of Robo-
2) to avoid ventral OB (expressing high levels of Slit-1)
(Cho et al., 2007). It seems likely that Slit-1/Robo-2, per-
haps in combination with other ligand-receptor pairs that
are distributed in a graded fashion along the D-V axis,
topographic targeting along this axis.
after the initial coarse targeting make the daunting task of
specifying discrete addresses for 1000 ORN classes more
manageable. Sensory experience and neuronal activity
Yu et al., 2004; Zou et al., 2004). Alteration in the levels of
ephrin-A proteins, classic axon guidance molecules in-
volved in the formation of visual maps (section 2), resulted
in subtle changes of glomerular targeting position (Cutforth
above phenomena: levels of ephrin-A5, and its receptor
EphA5, are regulated by sensory activity. Whereas EphA5
transcription is upregulated by activity, ephrin-A5 is down-
regulated (Serizawa et al., 2006). Moreover, different OR
classes express different levels of ephrin-A (Cutforth et al.,
2003; Serizawa et al., 2006) and EphA5 (Serizawa et al.,
2006), so that neighboring glomeruli exhibit a complemen-
tary expression pattern of this pair of repulsive guidance
molecules. Interestingly, the levels of another pair of Ig-do-
main-containing homophilic adhesion molecules, Kirrel2
and Kirrel3, in ORNs also correlate with OR classes and
are also up- and downregulated by activity, respectively.
Furthermore, overexpressing Kirrel2 in half of ORNs of
a specific class causes the target glomerulus to split into
two adjacent glomeruli (Serizawa et al., 2006). Thus, the
receptor pairs can be used to ensure that axons belonging
to the same OR find each other and segregate from axons
belonging to other ORN classes near their target glomeruli
by Kirrel2/3, ephrin-A5/EphA5 and other such protein pairs
acting alone or as a combinatorial code and test whether
such local segregation in combination with global forces
(such as cAMP levels in the A-P axis) comes close to ex-
plaining the precise wiring of all ORN classes.
3.3 Mechanisms of ORN Axon Targeting in Flies
Despite the striking similarity in the organization of theirol-
factory systems (Hildebrand and Shepherd, 1997), an im-
portant difference between flies and mice is that ORs do
not participate in ORN axon targeting in flies (Dobritsa
et al., 2003; Wang et al., 2003). Indeed, the expression on-
set of most OR genes is after targeting is completed. The
participation of ORs in targeting may be a vertebrate in-
vention to accommodate the large increase in OR genes
and glomerular number. Identifying mechanisms that
allow fly ORNs to find their targets may uncover evolution-
arily ancient mechanisms that work together with, or are
co-opted by, OR-dependent mechanisms in mice.
When pioneering ORN axons enter the brain, the devel-
oping antennal lobe is already prepatterned by PN den-
drites. Different PN classes have already sent their den-
drites to the approximate area corresponding to their
future glomerular targets (Jefferis et al., 2004; see below).
Thus, ORN axon targeting can (1) use global cues in the
antennal lobe, (2) sort each other out by axon-axon
Neuron 56, October 25, 2007 ª2007 Elsevier Inc.
interactions similar to what has been postulated above for
the mouse, (3) use cues located on their future synaptic
partner PN dendrites.
been reported, but the following examples suggest their
usage. First, when the cell adhesion molecule N-cadherin
was removed from ORNs, their axons remain at the anten-
nal lobe surface without invading into the antennal lobe to
form proper synaptic contact with PN dendrites. Conse-
quently, glomeruli do not form. Yet N-cadherin mutant
ORNs of specific classes still target their axons to the sur-
face areas of the antennal lobe that roughly correspond to
the positions of their glomerular targets (Hummel and Zi-
pursky, 2004). These findings suggest that global target-
ing does not require interactions with PN dendrites or
the formation of glomeruli. Second, as discussed earlier,
pairs or multiple ORNs belonging to different classes
reside in the same sensillum but target their axons to dif-
ferent glomeruli. When Notch signaling is disrupted,
such differences disappear, and ORNs target to glomeruli
belonging to ‘‘Notch-OFF’’ ORN classes within the same
reotyped organization in the antennal lobe (Endo et al.,
2007)(Figure 5C). Thesimplest interpretation isthatNotch
signaling diversifies ORN cell types by conferring them
with expression of different guidance receptors, which
allow their axons to target to distinct areas in response
to the same global cues in the antennal lobe. This study
also provided strong evidence that coordination of ORN
axon targeting and OR expression later is constrained
by their lineages, upon which Notch signaling acts to di-
The involvement of axon-axon interactions in ORN axon
targeting can be inferred from studies analyzing mutants
in the cell adhesion molecule Dscam and transcription
factor Acj6. Mutant axons of the same ORN class often
form clusters in ectopic places, implying self-recognition
and/or stabilization of ORN axons that belong to the
same class (Hummel et al., 2003; Komiyama et al.,
2004). Furthermore, in mosaic animals, wild-type ORN
axons mistarget when other ORNs are mutant for Acj6,
suggesting that hierarchical interactions among different
ORN classes are necessary for axon targeting of certain
ORN classes (Komiyama et al., 2004). The transmem-
brane axon guidance molecule Sema-1a and its receptor
plexinA have recently been identified to play a role in such
axon-axon interactions (Lattemann et al., 2007; Sweeney
et al., 2007). Deprivation of Sema-1a from early-arriving
axons from antennal ORNs causes late-arriving wild-
type maxillary palp axons to mistarget to areas normally
occupied by antennal ORN axons, suggesting that
Sema-1a made by antennal axons acts as a repulsive li-
gand to constrain target choice of maxillary palp axons
through direct axon-axon interactions at the target (Swee-
ney et al., 2007) (Figure 5D). This strategy of temporal tar-
get restriction through axon-axon interactions may be
used in the development of other neural maps, including
the retinotopic map (see section 2.4).
The existence of cues on PN dendrites that are used for
ing. OverexpressingDscam in PNsthattarget dendrites to
adjacent glomeruli frequently causesone of these glomer-
uli (VA1d) to swap position with a neighboring glomerulus
get to VA1d or VA1lm always follow the mispositioned PN
tions (Zhu et al., 2006a). This observation implies that a di-
rect recognition of PN dendrites by ORN axons acts at
least at the final step to ensure correct synaptic matching.
what molecules mediate such synaptic matching.
In summary, existing evidence supports a multistep
targeting process whereby the identities of ORNs are
determined by a combination of their lineage and spatial
location in the peripheral sensory organs. This identity
cific guidance molecules that allow each ORN class to
uniquely respond to global cues in the antennal lobe,
cues fromaxons of other ORNclasses and fromtheirpart-
ner PN dendrites.
3.4 Mechanisms of PN Dendrite Targeting in Flies
A special feature of the fly olfactory map, not previously
described in any other neural circuits, is the active and
precise dendrite targeting of second-order olfactory pro-
jection neurons (PNs). (For recent examples of dendrite
targeting in vertebrates, see Mumm et al., 2006; Vrieseling
and Arber, 2006.) At the transition between larva and
pupa, PNs start to elaborate their dendrites. Within the
next 18 hr or so, these dendrite elaborations create the
markably, each of the ?50 classes of PNs have sorted out
proto-antennal lobe corresponding to their future glomer-
ular positions (Jefferis et al., 2004) (Figure 5D).
Clonal analysis suggested that PN dendrite targeting
(Jefferis et al., 2001) (Figure 5B). Indeed, candidate tran-
scription factors that carry lineage and birth-order infor-
mation have been identified. Two POU-domain transcrip-
tion factors, Acj6 and Drifter, are expressed specifically in
two major PN lineages, and loss- and gain-of-function ex-
periments suggest that they play an instructive role in
specifying PN dendrite targeting appropriate for their line-
ages (Komiyama et al., 2003). On the other hand, the zinc-
finger transcription factor Chinmo is expressed in multiple
neuroblast lineages, but Chinmo protein forms a temporal
from the same lineage have more Chinmo proteins than
those born later. When chinmo is deleted from early-
born PNs, their dendrites are targeted to glomeruli appro-
priate for their younger sisters (Zhu et al., 2006b). Despite
these advances, we are still at the beginning of identifying
the entire transcription factor code for PN dendrite target-
ing (Komiyama and Luo, 2007).
Neuron 56, October 25, 2007 ª2007 Elsevier Inc.
Transcription factor codes are presumably turned into
cell-surface receptor codes that allow PN dendrites of dif-
ferent classes to respond differentially to the same envi-
ronment. So far, only one cell-surface protein, Sema-1a,
has been shown to play an instructive role in PN dendrite
targeting. Interestingly, unlike in ORNs where Sema-1a
acts cell-nonautonomously as a ligand in axon-axon inter-
actions (section 3.3), Sema-1a acts cell-autonomously as
a receptor to direct PN dendrite targeting. Furthermore,
different PN classes express different amounts of Sema-
1a in their dendrites, such that Sema-1a protein forms
a gradient along the dorsolateral to ventromedial axis.
Loss- and gain-of-function experiments support a model
in which levels of Sema-1a instruct PN dendrite targeting
along this axis (Komiyama et al., 2007). Future experi-
ments identifying the ligand for Sema-1a, and other li-
gand-receptor pairs that instruct PN dendrite targeting
along this and other axes, will provide a more complete
picture of how the coarse dendrite map is generated.
The coarse dendrite map is refined into a discrete glo-
merular map after the arrival of ORN axons. The mecha-
nism by which this is achieved is unknown, but likely
involves complex cellular interactions. In addition to the
ORN-PN axon-dendrite and ORN axon-axon interactions
described above, dendrite-dendrite interactions among
PNs have also been shown to be essential to refine den-
drites of individual PNs within one glomerulus (Zhu and
Luo, 2004). A future challenge is to determine how these
forces act concertedly to help establishing the final glo-
merular map and how discreteness emerges.
On the surface, olfactory and retinotopic projections ap-
pear to be very different types of map, one toward the dis-
crete extreme and the other at the continuous extreme.
Can common themes nevertheless be discerned in the
similar principles to be used in other maps throughout the
One common theme in mapping is the existence of one
set of mechanisms to establish an initial rough map, fol-
lowed by another set for map refinement. Gradients of la-
bels are well established as a mechanism to set up initial
order in continuous topographic maps. While the role of
gradients of Ephs and ephrins has been especially well
studied in retinotopy, it also extends to many other contin-
uous maps. Interestingly, in olfactory maps, the use of
graded signaling systems and graded labels is also now
emerging as a developmental strategy. Why use gradients
in mapping? Gradients have two important advantages.
First, they can behighly economical, since a smallnumber
of graded molecules can specify many positional values
across an entire developmental field. Second, gradients
can provide information not only about final position, but
also about direction. Rather than axons having to search
the target by a random walk, gradients can be sensed at
distant points in the target and can be used by axons to
navigate toward their destination.
Gradients have the limitation, however, that they are an
inherently imprecise way to reliably distinguish nearby
points. Refinement mechanisms can increase the preci-
sion. In retinotopic mapping, refinement is achieved by
an instructive role for correlated neural activity. In the ol-
factory system, sharp discrimination can be achieved by
glomerulus-specific labels such as levels of Kirrels and
ephrins, which can segregate axons in a discrete manner
to the correct glomerulus. These seemingly different re-
finement strategies used in vision and olfaction have sim-
ilarities and might reflect similar underlying mechanisms.
In particular, expression of the glomerulus-specific labels
is found to be dependent on neural activity. In the future, it
will be interesting to know whether the role for activity in
olfactory mapping is instructive, using correlated activity
in a Hebbian-type mechanism, whether it is odorant-
evoked or spontaneous, and whether visual map refine-
ment may employ the same molecular segregation mech-
anisms that distinguish adjacent glomeruli.
Another common theme is the use of axon-axon inter-
actions to ensure filling of the target in an orderly manner.
An apogee of this principle is seen in the fly retinotopic
projections, where individual axons target to a unique car-
tridge or column. In this case, axon-axon interactions me-
diated by cell-surface molecules in the cadherin and
IgCAM families can determine the orderly guidance and
tiling of axons to form a stereotypic array in the target. In
the vertebrate retinotectal projection, axon-axon interac-
tions ensure that the entire target is filled with axons
smoothly and completely. While the molecular mecha-
nismsforthisarestill notwellunderstood, therequirement
of an IgCAM for smooth target filling is intriguing, since it
suggests that target filling in vertebrates and axon tiling
in flies may reflect the same underlying principle of map-
Axon-axon interactions are also used extensively in
both vertebrate and fly olfactory maps. In the mouse, ad-
hesive and repulsive axon-axon interactions are likely
used for local sorting of axon terminals correlating with
ORN identity. In the fly, repulsive axon-axon interactions
at the target are used to ensure that different ORN classes
occupy different areas of the antennal lobe.
ping strategies can reveal features specific to particular
types of map. For example, where patterning information
resides andhowmaps aresequentiallyassembled appear
to differ for different neural maps. The development of the
fly visual system is driven by the input field and involves
a spatiotemporal mechanism of local interactions where
retinal units develop sequentially and can induce the dif-
ferentiation and patterning of their corresponding target
units. While this mechanism seems especially well suited
system, similar spatiotemporal principles could be used in
other maps. Patterning information in the vertebrate reti-
notopic map and fly olfactory map resides in both input
and target fields. In the case of the fly olfactory map, pat-
terning of PN dendrite precedes the arrival of ORN axons.
Neuron 56, October 25, 2007 ª2007 Elsevier Inc.
In general, independent patterning allows target neurons
to coordinate their input with their own output as in the
case of fly olfactory system. Future studies will clarify to
what extent these differences reflect our partial under-
standing or reflect differences in size, developmental,
functional, and evolutionary constraints of each neural
Retinotopic and olfactory glomerular maps represent
two ends of a continuum that includes many other types
of neural map in between. We predict that the emerging
principles outlined above will be used in the development
of many neural maps described in this issue and in addi-
tional neural maps yet to be discovered in the nervous
We thank members of our laboratories and Tom Clandinin, David Feld-
heim, and Pierre Vanderhaeghen for critical comments and sugges-
tions. We specially thank Kazunari Miyamichi for making Figure 4
and for discussions about mouse olfaction; and Miriam Osterfield
and Yao Chen for discussions about gradient mechanisms. L.L. is an
investigator of the Howard Hughes Medical Institute. Research in our
laboratories is supported by the National Institutes of Health.
Barnea, G., O’Donnell, S., Mancia, F., Sun, X., Nemes, A., Mendel-
sohn, M., and Axel, R. (2004). Odorant receptors on axon termini in
the brain. Science 304, 1468.
Boyle, M., Nighorn, A., and Thomas, J.B. (2006). Drosophila Eph
receptor guides specific axon branches of mushroom body neurons.
Development 133, 1845–1854.
Braisted, J.E., McLaughlin, T., Wang, H.U., Friedman, G.C., Anderson,
D.J.,and O’Leary, D.D.M. (1997). Graded and lamina-specific distribu-
tions of ligands of EphB receptor tyrosine kinases in the developing
retinotectal system. Dev. Biol. 191, 14–28.
Brown, A., Yates, P.A., Burrola, P., Ortuno, D., Vaidya, A., Jessell,
T.M., Pfaff, S.L., O’Leary, D.D., and Lemke, G. (2000). Topographic
mapping from the retina to the midbrain is controlled by relative but
not absolute levels of EphA receptor signaling. Cell 102, 77–88.
Brunet, I.,Weinl, C., Piper, M.,Trembleau, A.,Volovitch,M.,Harris, W.,
Prochiantz, A., and Holt, C. (2005). The transcription factor Engrailed-2
guides retinal axons. Nature 438, 94–98.
Buck, L., and Axel, R. (1991). A novel multigene family may encode
odorant receptors: a molecular basis for odor recognition. Cell 65,
Cang, J., Kaneko, M., Yamada, J., Woods, G., Stryker, M.P., and Feld-
heim, D.A. (2005). Ephrin-as guide the formation of functional maps in
the visual cortex. Neuron 48, 577–589.
Carvalho, R.F., Beutler, M., Marler, K.J., Knoll, B., Becker-Barroso, E.,
Heintzmann, R., Ng, T., and Drescher, U. (2006). Silencing of EphA3
through a cis interaction with ephrinA5. Nat. Neurosci. 9, 322–330.
Chandrasekaran, A.R., Shah, R.D., and Crair, M.C. (2007). Develop-
mental homeostasis of mouse retinocollicular synapses. J. Neurosci.
Cheng, H.-J., Nakamoto, M., Bergemann, A.D., and Flanagan, J.G.
(1995). Complementary gradients in expression and binding of ELF-1
and Mek4 in development of the topographic retinotectal projection
map. Cell 82, 371–381.
Chesler, A.T., Zou, D.J., Le Pichon, C.E., Peterlin, Z.A., Matthews,
G.A., Pei, X., Miller, M.C., and Firestein, S. (2007). A G protein/cAMP
signal cascade is required for axonal convergence into olfactory glo-
meruli. Proc. Natl. Acad. Sci. USA 104, 1039–1044.
Cho, J.H., Lepine, M., Andrews, W., Parnavelas, J., and Cloutier, J.F.
(2007). Requirement for Slit-1 and Robo-2 in zonal segregation of ol-
factory sensory neuron axons in the main olfactory bulb. J. Neurosci.
Clandinin, T.R., and Zipursky, S.L. (2000). Afferent growth cone inter-
actions control synaptic specificity in the Drosophila visual system.
Neuron 28, 427–436.
Clandinin, T.R., and Zipursky, S.L. (2002). Making connections in the
fly visual system. Neuron 35, 827–841.
Cline, H.T. (1998). Topographic maps - developing roles of synaptic
plasticity. Curr. Biol. 8, R836–R839.
Cohen-Cory, S., and Fraser, S.E. (1995). Effects of brain-derived neu-
rotrophic factor on optic axon branching and remodelling in vivo. Na-
ture 378, 192–196.
Col, J.A., Matsuo, T., Storm, D.R., and Rodriguez, I. (2007). Adenylyl
cyclase-dependent axonal targeting in the olfactory system. Develop-
ment 134, 2481–2489.
Constantine-Paton, M., and Law, M.I. (1978). Eye-specific termination
bands in tecta of three-eyed frogs. Science 202, 639–641.
and functional organization of the Drosophila olfactory system. Curr.
Biol. 15, 1535–1547.
Cutforth, T., Moring, L., Mendelsohn, M., Nemes, A., Shah, N.M., Kim,
M.M., Frisen, J., and Axel, R. (2003). Axonal ephrin-As and odorant re-
ceptors: coordinate determination of the olfactory sensory map. Cell
de Bruyne, M., Clyne, P.J., and Carlson, J.R. (1999). Odor coding in
a model olfactory organ: the Drosophila maxillary palp. J. Neurosci.
de Bruyne, M., Foster, K., and Carlson, J.R. (2001). Odor coding in the
Drosophila antenna. Neuron 30, 537–552.
Demyanenko, G.P., and Maness, P.F. (2003). The L1 cell adhesion
molecule is essential for topographic mapping of retinal axons. J. Neu-
rosci. 23, 530–538.
Dobritsa, A.A., van der Goes van Naters, W., Warr, C.G., Steinbrecht,
R.A., and Carlson, J.R. (2003). Integrating the molecular and cellular
basis of odor coding in the Drosophila antenna. Neuron 37, 827–841.
Drescher, U., Kremoser, C., Handwerker, C., Loschinger, J., Noda, M.,
and Bonhoeffer, F. (1995). In vitro guidance of a retinal ganglion cell
axons by RAGS, a 25kDa tectal protein related to ligands for Eph
receptor tyrosine kinase. Cell 82, 369–370.
Endo, K., Aoki, T., Yoda, Y., Kimura, K., and Hama, C. (2007). Notch
signal organizes the Drosophila olfactory circuitry by diversifying the
sensory neuronal lineages. Nat. Neurosci. 10, 153–160.
Eph Nomenclature Committee (1997). Unified nomenclature for Eph
family receptors and their ligands, the ephrins. Cell 90, 403–404.
Feinstein, P., and Mombaerts, P. (2004). A contextual model for axonal
sorting into glomeruli in the mouse olfactory system. Cell 117, 817–
Feinstein, P., Bozza, T., Rodriguez, I., Vassalli, A., and Mombaerts, P.
(2004). Axon guidance of mouse olfactory sensory neurons by odorant
receptors and the beta2 adrenergic receptor. Cell 117, 833–846.
Feldheim, D.A., Vanderhaeghen, P., Hansen, M.J., Frisen, J., Lu, Q.,
Barbacid, M., and Flanagan, J.G. (1998). Topographic guidance labels
in a sensory projection to the forebrain. Neuron 21, 1303–1313.
Feldheim, D.A., Kim, Y.-I., Bergemann, A.D., Frisen, J., Barbacid, M.,
and Flanagan, J.G. (2000). Genetic analysis of ephrin-A2 and ephrin-
A5 shows their requirement in multiple aspects of retinocollicular map-
ping. Neuron 25, 563–574.
Neuron 56, October 25, 2007 ª2007 Elsevier Inc.
Fishilevich, E., and Vosshall, L.B. (2005). Genetic and functional subdi-
vision of the Drosophila antennal lobe. Curr. Biol. 15, 1548–1553.
Flanagan, J.G. (2006). Neural map specification by gradients. Curr.
Opin. Neurobiol. 16, 59–66.
Flanagan, J.G., and Vanderhaeghen, P. (1998). The ephrins and Eph
receptors in neural development. Annu. Rev. Neurosci. 21, 309–345.
Fraser, S.E., and Hunt, R.K. (1980). Retinotectal specificity: models
and experiments in search of a mapping function. Annu. Rev. Neuro-
sci. 3, 319–352.
Friedman, G.C., and O’Leary, D.D.M. (1996). Retroviral misexpression
of engrailed genes in the chick optic tectum perturbs the topographic
targeting of retinal axons. J. Neurosci. 16, 5498–5509.
Frise ´n, J., Yates, P.A., McLaughlin, T., Friedman, G.C., O’Leary,
D.D.M., and Barbacid, M. (1998). Ephrin-A5 (AL-1/RAGS) is essential
for proper retinal axon guidance and topographic mapping in the
mammalian visual system. Neuron 20, 235–243.
Gao, Q., Yuan, B., and Chess, A. (2000). Convergent projections of
Drosophila olfactory neurons to specific glomeruli in the antennal
lobe. Nat. Neurosci. 3, 780–785.
Gierer, A. (1987). Directional cues for growing axons forming the reti-
notectal projection. Development 101, 479–489.
Goodhill, G.J., and Richards, L.J. (1999). Retinotectal maps: moleculs,
models and misplaced data. Trends Neurosci. 22, 529–534.
Halloran, M.C., and Wolman, M.A. (2006). Repulsion or adhesion:
receptors make the call. Curr. Opin. Cell Biol. 18, 533–540.
Hansen, M.J., Dallal, G.E., and Flanagan, J.G. (2004). Retinal axon re-
sponse to ephrin-As shows a graded, concentration-dependent tran-
sition from growth promotion to inhibition. Neuron 42, 717–730.
Harris, W.A. (1980). The effects of eliminating impulse activity on the
development of the retinotectal projection in salamanders. J. Comp.
Neurol. 194, 303–317.
Hattori, M., Osterfield, M., and Flanagan, J.G. (2000). Regulated
cleavage of a contact-mediated axon repellent. Science 289, 1360–
Hebb, D.O. (1949). The Organization of Behavior (New York: John
Wiley and Sons).
Hiesinger, P.R., Zhai, R.G., Zhou, Y., Koh, T.W., Mehta, S.Q., Schulze,
K.L., Cao, Y., Verstreken, P., Clandinin, T.R., Fischbach, K.F., et al.
(2006). Activity-independent prespecification of synaptic partners in
the visual map of Drosophila. Curr. Biol. 16, 1835–1843.
Hildebrand,J.G., andShepherd,G.M. (1997).Mechanismsofolfactory
discrimination: converging evidence for common principles across
phyla. Annu. Rev. Neurosci. 20, 595–631.
Hindges, R., McLaughlin, T., Genoud, N., Henkemeyer, M., and
O’Leary, D.D. (2002). EphB forward signaling controls directional
branch extension and arborization required for dorsal-ventral retino-
topic mapping. Neuron 35, 475–487.
Holash, J.A., and Pasquale, E.B. (1995). Polarized expression of the
receptor protein tyrosine kinase Cek5 in the developing avian visual
system. Dev. Biol. 172, 683–693.
Holmberg, J., Clarke, D.L., and Frisen, J. (2000). Regulation of repul-
sion versus adhesion by different splice forms of an Eph receptor. Na-
ture 408, 203–206.
Hornberger, M.R., Dutting, D., Ciossek, T., Yamada, T., Handwerker,
ulation of EphA receptor function by coexpressed ephrinA ligands on
retinal ganglion cell axons. Neuron 22, 731–742.
Hua, J.Y., Smear, M.C., Baier, H., and Smith, S.J. (2005). Regulation of
axon growth in vivo by activity-based competition. Nature 434, 1022–
Huang, Z., and Kunes, S. (1996). Hedgehog, transmitted along retinal
axons, triggers neurogenesis in the developing visual centers of the
Drosophila brain. Cell 86, 411–422.
Huang, Z., Shilo, B.Z., and Kunes, S. (1998). A retinal axon fascicle
uses spitz, an EGF receptor ligand, to construct a synaptic cartridge
in the brain of Drosophila. Cell 95, 693–703.
Hummel, T., and Zipursky, S.L. (2004). Afferent induction of olfactory
glomeruli requires N-cadherin. Neuron 42, 77–88.
Hummel, T., Vasconcelos, M.L., Clemens, J.C., Fishilevich, Y., Vos-
shall, L.B., and Zipursky, S.L. (2003). Axonal targeting of olfactory re-
ceptor neurons in Drosophila is controlled by Dscam. Neuron 37,
Huynh-Do, U., Stein, E., Lane, A.A., Liu, H., Cerretti, D.P., and Daniel,
T.O. (1999). Surface densities of ephrin-B1 determine EphB1-coupled
activation of cell attachment through alpha(v)beta(3) and alpha(5)-
beta(1) integrins. EMBO J. 18, 2165–2173.
Imai, T., Suzuki, M., and Sakano, H. (2006). Odorant receptor-derived
cAMP signals direct axonal targeting. Science 314, 657–661.
Jacobson, M. (1991). Developmental Neurobiology,Third Edition (New
York, NY: Plenum press).
Janes, P.W., Saha, N., Barton, W.A., Kolev, M.V., Wimmer-Kleikamp,
S.H., Nievergall, E., Blobel, C.P., Himanen, J.P., Lackmann, M., and
module acts as a molecular switch for ephrin cleavage in trans. Cell
Jefferis, G.S.X.E., Marin, E.C., Stocker, R.F., and Luo, L. (2001).Target
Jefferis, G.S., Vyas, R.M., Berdnik, D., Ramaekers, A., Stocker, R.F.,
Tanaka, N.K., Ito, K., and Luo, L.(2004).Developmental origin of wiring
specificity in the olfactory system of Drosophila. Development 131,
Jefferis, G.S., Potter, C.J., Chan, A.M., Marin, E.C., Rohlfing, T.,
Maurer, C.R., Jr., and Luo, L. (2007). Comprehensive maps of Dro-
sophila higher olfactory centers: spatially segregated fruit and phero-
mone representation. Cell 128, 1187–1203.
Joliot, A., and Prochiantz, A. (2004). Transduction peptides: from tech-
nology to physiology. Nat. Cell Biol. 6, 189–196.
Klein, R. (2004). Eph/ephrin signaling in morphogenesis, neural devel-
opment and plasticity. Curr. Opin. Cell Biol. 16, 580–589.
Kobayashi, T., Nakamura, H., and Yasuda, M. (1990). Disturbance of
refinement of retinotectal projection in chick embryos by tetrodotoxin
and grayanotoxin. Brain Res. Dev. Brain Res. 57, 29–35.
Komiyama, T., and Luo, L. (2007). Intrinsic control of precise dendritic
targeting by an ensemble of transcription factors. Curr. Biol. 17, 278–
Komiyama, T., Johnson, W.A., Luo, L., and Jefferis, G.S. (2003). From
lineage to wiring specificity: POU domain transcription factors control
precise connections of Drosophila olfactory projection neurons. Cell
Komiyama, T., Carlson, J.R., and Luo, L. (2004). Olfactory receptor
neuron axon targeting: intrinsic transcriptional control and hierarchical
interactions. Nat. Neurosci. 7, 819–825.
Komiyama, T.,Sweeney,L.B.,Schuldiner,O.,Garcia,K.C., andLuo,L.
(2007). Graded expression of semaphorin-1a cell-autonomously di-
rects dendritic targeting of olfactory projection neurons. Cell 128,
Lambot, M.A., Depasse, F., Noel, J.C., and Vanderhaeghen, P. (2005).
Mapping labels in the human developing visual system and the evolu-
tion of binocular vision. J. Neurosci. 25, 7232–7237.
Lattemann, M., Zierau, A., Schulte, C., Seidl, S., Kuhlmann, B., and
Hummel, T. (2007). Semaphorin-1a controls receptor neuron-specific
Neuron 56, October 25, 2007 ª2007 Elsevier Inc.
axonal convergence in the primary olfactory center of Drosophila.
Neuron 53, 169–184.
Lee, R.C., Clandinin, T.R., Lee, C.H., Chen, P.L., Meinertzhagen, I.A.,
and Zipursky, S.L. (2003). The protocadherin Flamingo is required for
axon target selection in the Drosophila visual system. Nat. Neurosci.
Liu, Y., Berndt, J., Su, F., Tawarayama, H., Shoji, W., Kuwada, J.Y.,
and Halloran, M.C. (2004). Semaphorin3D guides retinal axons along
the dorsoventral axis of the tectum. J. Neurosci. 24, 310–318.
Logan, C., Wizenmann, A., Drescher, U., Monschau, B., Bonhoeffer,
F.,and Lumsden,A.(1996).Rostraloptic tectumacquirescaudalchar-
acteristics following ectopic Engrailed expression. Curr. Biol. 6, 1006–
Mann, F., Ray, S., Harris, W., and Holt, C. (2002). Topographic map-
ping in dorsoventral axis of the Xenopus retinotectal system depends
on signaling through ephrin-B ligands. Neuron 35, 461–473.
Marin, E.C., Jefferis, G.S.X.E., Komiyama, T., Zhu, H., and Luo, L.
(2002).Representation oftheglomerularolfactory mapintheDrosoph-
ila brain. Cell 109, 243–255.
Marquardt, T., Shirasaki, R., Ghosh, S., Andrews, S.E., Carter, N.,
Hunter, T., and Pfaff, S.L. (2005). Coexpressed EphA receptors and
ephrin-A ligands mediate opposing actions on growth cone navigation
from distinct membrane domains. Cell 121, 127–139.
Mast, J.D., Prakash, S., Chen, P.L., and Clandinin, T.R. (2006). The
mechanisms and molecules that connect photoreceptor axons to their
targets in Drosophila. Semin. Cell Dev. Biol. 17, 42–49.
Matsuoka, H., Obama, H., Kelly, M.L., Matsui, T., and Nakamoto, M.
(2005). Biphasic functions of the kinase-defective Ephb6 receptor in
cell adhesion and migration. J. Biol. Chem. 280, 29355–29363.
McLaughlin, T., and O’Leary, D.D. (2005). Molecular gradients and de-
velopment of retinotopic maps. Annu. Rev. Neurosci. 28, 327–355.
McLaughlin, T., Hindges, R., Yates, P.A., and O’Leary, D.D. (2003a).
Bifunctional action of ephrin-B1 as a repellent and attractant to control
bidirectional branch extension in dorsal-ventral retinotopic mapping.
Development 130, 2407–2418.
McLaughlin, T., Torborg, C.L., Feller, M.B., and O’Leary, D.D. (2003b).
Retinotopic map refinement requires spontaneous retinal waves dur-
ing a brief critical period of development. Neuron 40, 1147–1160.
nous bursts of action potentials in ganglion cells of the developing
mammalian retina. Science 252, 939–943.
Millard, S.S., Flanagan, J.J., Pappu, K.S., Wu, W., and Zipursky, S.L.
(2007). Dscam2 mediates axonal tiling in the Drosophila visual system.
Nature 447, 720–724.
Ming, G.L., Song, H.J., Berninger, B., Holt, C.E., Tessier-Lavigne, M.,
and Poo, M.M. (1997). cAMP-dependent growth cone guidance by
netrin-1. Neuron 19, 1225–1235.
Miyamichi, K., Serizawa, S., Kimura, H.M., and Sakano, H. (2005).
Continuous and overlapping expression domains of odorant receptor
genesintheolfactory epitheliumdeterminethedorsal/ventral position-
ing of glomeruli in the olfactory bulb. J. Neurosci. 25, 3586–3592.
Mombaerts, P., Wang, F., Dulac, C., Chao, S.K., Nemes, A., Mendel-
sohn, M., Edmondson, J., and Axel, R. (1996). Visualizing an olfactory
sensory map. Cell 87, 675–686.
Monnier, P.P., Sierra, A., Macchi, P., Deitinghoff, L., Andersen, J.S.,
Mann, M., Flad, M., Hornberger, M.R., Stahl, B., Bonhoeffer, F., and
Mueller, B.K. (2002). RGM is a repulsive guidance molecule for retinal
axons. Nature 419, 392–395.
Mori, K., Takahashi, Y.K., Igarashi, K.M., and Yamaguchi, M. (2006).
Maps of odorant molecular features in the Mammalian olfactory
bulb. Physiol. Rev. 86, 409–433.
Mumm, J.S., Williams, P.R., Godinho, L., Koerber, A., Pittman, A.J.,
Roeser,T., Chien, C.B., Baier, H.,and Wong, R.O. (2006). In vivo imag-
ing reveals dendritic targeting of laminated afferents by zebrafish reti-
nal ganglion cells. Neuron 52, 609–621.
Nakamoto, M., Cheng, H.-J., Friedman, G.C., McLaughlin, T., Hansen,
M.J., Yoon, C., O’Leary, D.D.M., and Flanagan, J.G. (1996). Topo-
graphically specific effects of ELF-1 on retinal axon guidance in vitro
and retinal axon mapping in vivo. Cell 86, 755–766.
Nicol, X., Voyatzis, S., Muzerelle, A., Narboux-Neme, N., Sudhof, T.C.,
Miles, R., and Gaspar, P. (2007). cAMP oscillations and retinal activity
are permissive for ephrin signaling during the establishment of the ret-
inotopic map. Nat. Neurosci. 10, 340–347.
Niederkofler, V., Salie, R., Sigrist, M., and Arber, S. (2004). Repulsive
sure but not retinal topography in the mouse visual system. J. Neuro-
sci. 24, 808–818.
Norlin, E.M., Alenius, M., Gussing, F., Hagglund, M., Vedin, V., and
Bohm, S.(2001). Evidence for gradients of gene expressioncorrelating
with zonal topography of the olfactory sensory map. Mol. Cell. Neuro-
sci. 18, 283–295.
O’Leary,D.D.M.,Yates, P.A.,and McLaughlin,T.(1999). Molecular de-
velopment of sensory maps: Representing sights and smells in the
brain. Cell 22, 255–269.
Osterfield, M., Kirschner, M.W., and Flanagan, J.G. (2003). Graded
positional information: interpretation for both fate and guidance. Cell
Pfeiffenberger, C., Yamada, J., and Feldheim, D.A. (2006). Ephrin-As
and patterned retinal activity act together in the development of topo-
graphic maps in the primary visual system. J. Neurosci. 26, 12873–
Rashid, T., Upton, A.L., Blentic, A., Ciossek, T., Knoll, B., Thompson,
I.D., and Drescher, U. (2005). Opposing gradients of ephrin-As and
EphA7 in the superior colliculus are essential for topographic mapping
in the mammalian visual system. Neuron 47, 57–69.
of odorant receptor gene expression in the olfactory epithelium. Cell
Ressler, K.J., Sullivan, S.L., and Buck, L.B. (1994). Information coding
in the olfactory system: evidence for a stereotyped and highly orga-
nized epitope map in the olfactory bulb. Cell 79, 1245–1255.
Retaux, S., and Harris, W.A. (1996). Engrailed and retinotectal topog-
raphy. Trends Neurosci. 19, 542–546.
Ruthazer, E.S., and Cline, H.T. (2004). Insights into activity-dependent
mapformationfrom theretinotectal system:amiddle-of-the-brainper-
spective. J. Neurobiol. 59, 134–146.
Ruthazer, E.S., Akerman, C.J., and Cline, H.T. (2003). Control of axon
branch dynamics by correlated activity in vivo. Science 301, 66–70.
Sato, M., Umetsu, D., Murakami, S., Yasugi, T., and Tabata, T. (2006).
DWnt4 regulates the dorsoventral specificity of retinal projections in
the Drosophila melanogaster visual system. Nat. Neurosci. 9, 67–75.
Schmidt, J.T., and Edwards, D.L. (1983). Activity sharpens the map
during the regeneration of the retinotectal projection in goldfish. Brain
Res. 269, 29–39.
Schmidt, J.T., and Eisele, L.E. (1985). Stroboscopic illumination and
dark rearing block the sharpening of the regenerated retinotectal
map in goldfish. Neuroscience 14, 535–546.
Schmitt, A.M., Shi, J., Wolf, A.M., Lu, C.C., King, L.A., and Zou, Y.
(2005). Wnt-Ryk signalling mediates medial-lateral retinotectal topo-
graphic mapping. Nature 439, 23–24.
Schwarting, G.A., Kostek, C., Ahmad, N., Dibble, C., Pays, L., and Pu-
schel, A.W.(2000).Semaphorin3A isrequired forguidanceofolfactory
axons in mice. J. Neurosci. 20, 7691–7697.
Neuron 56, October 25, 2007 ª2007 Elsevier Inc.
Senti, K.A., Usui, T., Boucke, K., Greber, U., Uemura, T., and Dickson,
B.J.(2003).Flamingo regulatesR8axon-axon andaxon-target interac-
tions in the Drosophila visual system. Curr. Biol. 13, 828–832.
Serizawa, S., Miyamichi, K., Takeuchi, H., Yamagishi, Y., Suzuki, M.,
and Sakano, H. (2006). A neuronal identity code for the odorant recep-
tor-specific andactivity-dependent axonsorting. Cell127,1057–1069.
Sperry, R.W. (1943). Visuomotor coordination in the newt (Triturus vir-
idescens) after regeneration of the optic nerve. J. Comp. Neurol. 79,
Sperry, R.W. (1963). Chemoaffinity in the orderly growth of nerve fiber
patterns and connections. Proc. Natl. Acad. Sci. USA 50, 703–710.
Strotmann, J., Conzelmann, S., Beck, A., Feinstein, P., Breer, H., and
Mombaerts, P. (2000). Local permutations in the glomerular array of
the mouse olfactory bulb. J. Neurosci. 20, 6927–6938.
Strotmann, J., Levai, O., Fleischer, J., Schwarzenbacher, K., and
Breer, H. (2004). Olfactory receptor proteins in axonal processes of
chemosensory neurons. J. Neurosci. 24, 7754–7761.
Sweeney, L.B., Couto, A., Chou, Y.H., Berdnik, D., Dickson, B.J., Luo,
ceptorneurons bySemaphorin-1a/PlexinA-mediated axon-axoninter-
actions. Neuron 53, 185–200.
Ting, C.Y., and Lee, C.H. (2007). Visual circuit development in Dro-
sophila. Curr. Opin. Neurobiol. 17, 65–72.
Torborg, C.L., and Feller, M.B. (2005). Spontaneous patterned retinal
activity and the refinement of retinal projections. Prog. Neurobiol. 76,
Torii, M., and Levitt, P. (2005). Dissociation of corticothalamic and tha-
lamocortical axon targeting by an EphA7-mediated mechanism. Neu-
ron 48, 563–575.
Vassar, R., Ngai, J., and Axel, R. (1993). Spatial segregation of odorant
receptor expression in the mammalian olfactory epithelium. Cell 74,
Vassar, R., Chao, S.K., Sitcheran, R., Nunez, J.M., Vosshall, L.B., and
Axel, R. (1994). Topographic organization of sensory projections to the
olfactory bulb. Cell 79, 981–991.
Vosshall, L.B., Wong, A.M., and Axel, R. (2000). An olfactory sensory
map in the fly brain. Cell 102, 147–159.
Vrieseling, E., and Arber, S. (2006). Target-induced transcriptional
control of dendritic patterning and connectivity in motor neurons by
the ETS gene Pea3. Cell 127, 1439–1452.
Wang, F., Nemes, A., Mendelsohn, M., and Axel, R. (1998). Odorant
receptors govern the formation of a precise topographic map. Cell
Wang, J.W., Wong, A.M., Flores, J., Vosshall, L.B., and Axel, R. (2003).
Two-photon calcium imaging reveals an odor-evoked map of activity
in the fly brain. Cell 112, 271–282.
Weinl, C., Drescher, U., Lang, S., Bonhoeffer, F., and Loschinger, J.
(2003). On the turning of Xenopus retinal axons induced by ephrin-
A5. Development 130, 1635–1643.
Weinl, C., Becker, N., and Loeschinger, J. (2005). Responses of tem-
poral retinal growth cones to ephrinA5-coated beads. J. Neurobiol.
West, A.E., Chen, W.G., Dalva, M.B., Dolmetsch, R.E., Kornhauser,
J.M., Shaywitz, A.J., Takasu, M.A., Tao, X., and Greenberg, M.E.
(2001). Calcium regulation of neuronal gene expression. Proc. Natl.
Acad. Sci. USA 98, 11024–11031.
Wong, R.O.L., Meister, M., and Shatz, C.J. (1993). Transient period of
correlated bursting activity during development of the mammalian ret-
ina. Neuron 11, 923–938.
Wong, A.M., Wang, J.W., and Axel, R. (2002). Spatial representation of
the glomerular map in the Drosophila protocerebrum. Cell 109, 229–
Yamaguchi, Y., and Pasquale, E.B. (2004). Eph receptors in the adult
brain. Curr. Opin. Neurobiol. 14, 288–296.
Yin, Y., Yamashita, Y., Noda, H., Okafuji, T., Go, M.J., and Tanaka, H.
(2004). EphA receptor tyrosine kinases interact with co-expressed
ephrin-A ligands in cis. Neurosci. Res. 48, 285–296.
Yu, C.R., Power, J., Barnea, G., O’Donnell, S., Brown, H.E., Osborne,
J., Axel, R., and Gogos, J.A. (2004). Spontaneous neural activity is re-
quired for the establishment and maintenance of the olfactory sensory
map. Neuron 42, 553–566.
Zhang, L.I., and Poo, M.M. (2001). Electrical activity and development
of neural circuits. Nat. Neurosci. Suppl. 4, 1207–1214.
critical window for cooperation and competition among developing
retinotectal synapses. Nature 395, 37–44.
Zhao, H., and Reed, R.R. (2001). X inactivation of the OCNC1 channel
gene reveals a role for activity-dependent competition in the olfactory
system. Cell 104, 651–660.
and axonal terminal arborization of olfactory projection neurons. Neu-
ron 42, 63–75.
Zhu, H., Hummel, T., Clemens, J.C., Berdnik, D., Zipursky, S.L., and
Luo, L. (2006a). Dendritic patterning by Dscam and synaptic partner
matching in the Drosophila antennal lobe. Nat. Neurosci. 9, 349–355.
Zhu, S., Lin, S., Kao, C.F., Awasaki, T., Chiang, A.S., and Lee, T.
(2006b). Gradients of the Drosophila Chinmo BTB-zinc finger protein
govern neuronal temporal identity. Cell 127, 409–422.
Zimmer, M., Palmer, A., Kohler, J., and Klein, R. (2003). EphB-ephrinB
bi-directionalendocytosis terminatesadhesion allowing contactmedi-
ated repulsion. Nat. Cell Biol. 5, 869–878.
Zimmer, G., Kastner, B., Weth, F., and Bolz, J. (2007). Multiple effects
of ephrin-A5 on cortical neurons are mediated by SRC family kinases.
J. Neurosci. 27, 5643–5653.
Zou, D.J., Feinstein, P., Rivers, A.L., Mathews, G.A., Kim, A., Greer,
C.A., Mombaerts, P., and Firestein, S. (2004). Postnatal refinement
of peripheral olfactory projections. Science 304, 1976–1979.
Zou, D.J., Chesler, A.T., Le Pichon, C.E., Kuznetsov, A., Pei, X.,
Hwang, E.L., and Firestein, S. (2007). Absence of adenylyl cyclase 3
perturbs peripheral olfactory projections in mice. J. Neurosci. 27,
Neuron 56, October 25, 2007 ª2007 Elsevier Inc.