Developmental Cell, Vol. 7, 401–412, September, 2004, Copyright 2004 by Cell Press
Directional Cell Migration Establishes
the Axes of Planar Polarity
in the Posterior Lateral-Line Organ of the Zebrafish
The first mitosis, which is invariably oriented parallel to
the anteroposterior axis of the fly, is controlled by Friz-
zled and other components of the planar polarity signal-
ing pathway (Gho and Schweisguth, 1998).
Although the proteins involved in interpretation of the
polarizing signals have been widely studied, the cell-
biological mechanisms that lead to planar polarization
are not well understood (Keller, 2002). The mechanore-
ceptive hair cell of the vertebrate internal ear provides
ment in amenable to investigation (Lewis and Davies,
2002). A hair cell takes its name from the hundred or so
cylindrical processes constituting the hair bundle that
projects from the cellular apex (for a review see Huds-
peth, 1989). A single true cilium, the kinocilium, occurs
at the tall edge of the bundle. The remaining processes,
the actin-rich stereocilia, are organized in order of in-
creasing height across the apical surface of each cell,
so that the hair bundle’s top surface is bevelled like a
hypodermic needle. The axis of morphological polarity
corresponds to the cell’s axis of excitability (Shotwell
et al., 1981). A positive deflection, which moves the
bundle toward its tall edge, opens transduction chan-
nels, allows cations to flow into the cell, and thereby
depolarizes the plasma membrane. Deflection in the op-
posite direction leads to hyperpolarization. The senses
of hearing and equilibrium thus rely on the exquisite
precision with which tens to thousands of hair cells are
oriented relative to one another across the sensory epi-
Some aquatic vertebrates sense directional water
closely related to that of the ear (Montgomery et al.,
1997). This organ comprises a stereotyped array of sen-
sory clusters called neuromasts (Metcalfe et al., 1985;
Montgomery et al., 1997). Haircells and supporting cells
form the core of each neuromast, which is innervated
by sensory neurons projecting to the central nervous
system (Metcalfe, 1985; Alexandre and Ghysen, 1999).
In the zebrafish, there are two main components, the
anterior andposterior lateral-line organs,which develop
from cephalic epithelial thickenings called neurogenic
ectodermal cells (Gompel et al., 2001). A group of 30
cells forms the precursors of a ganglion and of one
neuromast (L1), whereas the remainder of the cells con-
stitute a first primordium that initiates a posteriorward
migration around 20 hr postfertilization (hpf) (Sapede et
al., 2002). Cells of this primordium express the CXCR-
4b chemokine receptor, which allows them to follow the
horizontal myoseptum along a path formed by the Sdf-
1a chemokine (David et al., 2002; Li et al., 2004). By 40
hpf, the first primordium has completed its journey to
the tail of the fish, leaving behind seven to nine pro-
neuromasts. Soon thereafter, a second, smaller primor-
dium begins to migrate along the same path, depositing
a few pro-neuromasts over the rostral part of the trunk.
The entire complement of neuromasts later migrates
ventrally to reach its final location several neuromast
Herna ´n Lo ´pez-Schier, Catherine J. Starr,
James A. Kappler, Richard Kollmar,1
and A.J. Hudspeth*
Laboratory of Sensory Neuroscience and
Howard Hughes Medical Institute
The Rockefeller University
1230 York Avenue
New York, New York 10021
The proper orientation of mechanosensory hair cells
along the lateral-line organ of a fish or amphibian is
essential for the animal’s ability to sense directional
water movements. Within the sensory epithelium, hair
cells are polarized in a stereotyped manner, but the
mechanisms that control their alignment relative to
the body axes are unknown. We have found, however,
that neuromasts can be oriented either parallel or per-
pendicular to the anteroposterior body axis. By char-
acterizing the strauss mutant zebrafish line and by
romasts of these two orientations originate from, re-
spectively, the first and second primordia. Further-
more, altering the migratory pathway of a primordium
reorients a neuromast’s axis of planar polarity. We
to the body axes is established through an interaction
between directional movementby primordial cells and
the timing of neuromast maturation.
The coordinated orientation of polarized cells within the
plane of an epithelium is termed planar cell polarity. The
stepsproceeding fromglobalcellorientation tointracel-
lular and cytoskeletal rearrangements along the axis of
polarization. Extensive genetic analyses in Drosophila
have identified several proteins involved in the control
elled form part of the core of the signaling pathway
(Strutt, 2003). These proteins integrate signals leading
to the alignment of hairs and bristles in the wing, notum,
and abdomen, as well as the orientation of ommatidia
in the eye (for reviews see Adler, 2002; Tree et al., 2002;
Uemura and Shimada, 2003). The global orientation of
cellular asymmetry relative to the main body axes is
exemplified by the cell divisions of the sensory-organ
precursors inthe Drosophila pupalnotum. Thefour cells
that form each sensory organ originate through two
rounds of mitosis from a single precursor cell called pI.
1Present address: Department of Molecular and Integrative Physiol-
ogy, University of Illinois at Urbana-Champaign, Urbana, Illinois
diameters below the horizontal myoseptum (Ledent,
2002; Sapede et al., 2002). The developing posterior
lateral-line organ therefore undergoes a net posterior
and ventral movement. By the fifth day of development,
all neuromasts contain functional hair cells, but those
generated by the second primordium can still be identi-
fied by their location and smaller size.
Numerous studies have described the developmental
and molecular basis of the formation and innervation of
the lateral-line organ in wild-type zebrafish and other
species. Few, however, have focused attention on the
development of planar polarity by hair cells in neuro-
masts (Rouse and Pickles, 1991). In particular, nothing
is known about the cell-biological basis of hair cells’
global orientation along an animal’s body. We report
here the initial results of our investigation of this issue,
which indicate that the trajectory of the primordial cells
bine to establish this orientation.
with a fluorescent compound, 4-Di-2-ASP, that enters
transducing hair cells (Figures 2A–2D). Using rhoda-
mine-phalloidin to mark filamentous actin in stereocilia
(Figures 1G and 2E) and an antibody against acetylated
tubulin to label each hair cell’s soma and kinocilium
(Figures1Gand 2F)(HaddonandLewis, 1996),wefound
that every neuromast possesses an axis along which all
the constituent hair cells align in parallel but opposing
orientations (Figures 2F and 2G) (Lewis and Davies,
2002). The orientation of this axis is highly conserved in
each specific neuromast. For more than 100 wild-type
fish analyzed at 6 dpf, hair bundles in neuromast L1 and
in L5–L9 were always aligned with the anteroposterior
axis of the body; we designate neuromasts of this con-
figuration “parallel.” By contrast, the hair bundles in
neuromasts L2, L3, and often L4 oriented their hair bun-
dles orthogonally to this axis (Figure 2H); we term such
neuromasts “perpendicular.” In about 20% of the ani-
mals analyzed, however, hair bundles in L4 were polar-
ized parallel to the anteroposterior axis.
rior lateral-line organ displays three levels of planar cell
polarity. The first level, defined by the eccentric location
of the kinocilium in the hair bundle, corresponds to the
vectorial mechanosensitivity of the individual hair cell
reflects the axis of hair-cell orientation within a neuro-
mast, that is, the orientation of individual hair cells with
respect to one another (Flock and Wersa ¨ll, 1962; Lewis
and Davies, 2002). The final level, which characterizes
each neuromast along the anteroposterior axis of the
animal, specifies the orientation of each neuromast with
respect to the animal’s main body axes. The highly con-
served organization of planar polarity along the sensory
organ implies that its establishment must be precisely
Morphogenesis of the Posterior Lateral-Line Organ
The first primordium of the posterior lateral-line organ
begins to migrate along the horizontal myoseptum at
20 hpf, depositing discrete groups of cells, the pro-
neuromasts, in a stereotyped pattern along the animal’s
body to its tail (Ledent, 2002; Sapede et al., 2002). Sen-
sory axons of the octavolateralis system closely follow
this moving primordium, pathfinding for glial cells of
neural-crest origin (Gilmour et al., 2002).
We have investigated the development of the lateral-
line organ with an antibody against the tight-junctional
component claudin b, whose transcript is highly ex-
et al., 2001). The antibody marks all the cells of the
migrating primordium, the pro-neuromasts, and a trail
of cells connecting them (Figures 1A–1C). To confirm
that these trailing cells derive from the primordium, we
injected single-cell embryos with caged fluorescein-
dextran and used ultraviolet-light irradiation to uncage
the fluorophore in primordial cells at the onset of migra-
tion. The following day, animals were fixed and the clau-
din b protein was immunolabeled. All green-fluorescent
trailingcells werealsointensely labeledby theantibody,
confirming their primordial origin (Figure 1D).
Using claudin b as a marker, we also confirmed the
observation that the sensory nerve fibers trace the tra-
jectory of the primordium (Gilmour et al., 2002, 2004).
The claudin b-positive cells lying between pro-neuro-
masts frequently contact the axons and sometimes en-
wrap them in a manner suggestive of Schwann cells
(Figures 1Eand 1F).They maytherefore representa glial
subtype of placodal origin that ensheathes the accom-
panying axons during migration (David et al., 2002). At
4 days postfertilization (dpf), claudin b becomes re-
marker for these cells identified in the zebrafish.
Neuromast Innervation Is Dispensable for Planar
Polarization of Hair Cells
volateralis system confers neuromast identity along the
anteroposterior axis, and thus sets the orientation of
different neuromasts. To do this, we used an antisense
morpholino directed against neurogenin-1 transcripts.
Neurogenin-1 is necessary for the differentiation of all
cranial ganglia, including the octavolateralis ganglion,
but not for the formation of neuromasts and hair cells
(Andermann, et al., 2002). Although neurogenin-1 mor-
phant larvae lacked the acetylated tubulin-positive ax-
ons that normally lie beneath neuromasts, planar polar-
ization by hair cells was unaffected (data not shown).
Thus, innervation appears dispensable for neuromast
strauss Mutant Fish Lack Perpendicular Neuromasts
To investigate the requirements for planar polarity in
the posterior lateral-line organ, we sought mutations
affecting hair-cell orientation in mutant strains identified
during a mutagenic screen. We identified one line,
strauss, that has no defects in the first two levels of
ularly oriented neuromasts. strauss mutants also de-
velop fewer neuromasts and have smaller eyes than
wild-type fish (Figures 3A–3D).
Development of Planar Polarity in Hair Cells
Functional hair cells in the posterior lateral-line organ
begin to appear in an anterior-to-posterior progression
starting at 48–52 hpf, as assessed by labeling animals
Cell Migration and Planar Polarity
Figure 1. Claudin b Labeling of Primordial Cells and Mature Supporting Cells in the Posterior Lateral-Line Organ
(A) In a confocal section at 28 hpf, antiserum against claudin b (Cldn b) strongly labels all cells of the first primordium. The trail of cells
between the primordium and the most recently deposited pro-neuromast is also marked.
(B) The leading neurites, labeled for acetylated tubulin (Ac tub), invade the first primordium at an early stage of migration.
(C) In a confocal section of neuromasts at 48 hpf, axons of the octavolateralis nerve extend along the trail of claudin b-positive cells.
(D) Labeling of the first primordium with uncaged fluorescein-dextran (Fl-dex) marks neuromasts and trailing cells, which are also positive for
(E and F) Claudin b-positive cells enwrap the extending sensory axons adjacent to a neuromast (E) and along the developing octavolateralis
nerve (F). Note several points of apparent contact between the claudin b-positive cells and the axons (arrowheads).
(G) Actin-rich hair bundles are evident at hair-cell apices (asterisk) in a lateral view of a 5 dpf neuromast. Anti-claudin b specifically labels
mature supporting cells in all neuromasts, whereas acetylated tubulin marks the hair-cell somata and kinocilia (not evident in this image) as
well as neurons delaminating from the basal epithelial surface (arrowhead).
(H) In an intensity profile of a top view of a neuromast, peripheral supporting cells accumulate high levels of claudin b, which is excluded
from the center of the organ where mature hair cells label for acetylated tubulin and rhodamine-phalloidin. The intensity scale is linear but
its units are arbitrary. In these and all subsequent illustrations, the animal’s anterior is oriented to the left and its dorsum is situated to the top.
Scale bars: 50 ?m in (A)–(D), 10 ?m in (E) and (F).
The initial development of the posterior lateral-line
organ is unaffected by the strauss mutation. Wild-type
and mutantanimals formfirst primordiaof identicalsize,
which migrate along the horizontal myoseptum at the
same time and speed (data not shown). There is no
difference in the number of neuromasts deposited by
fish, the first functional hair cells appear in pairs aligned
parallel to the animal’s anteroposterior axis (Figures 3E
and 3F). The axis of hair-bundle polarity is also consis-
tently aligned with this body axis (Figures 3G and 3H);
there are no perpendicular neuromasts at this stage.
During the fourth day of development, wild-type-fish
begin to form hair cells in two or three additional neuro-
masts near the rostral end of the organ (Figure 3C).
roposterior body axis (Figure 3I). In strauss mutant lar-
vae, by contrast, additional neuromasts rarely appear
Distinct Origins of Parallel and Perpendicular
Our observation that parallel and perpendicular neuro-
masts develop sequentially suggests that the latter ma-
ture more slowly or that they derive from different pri-
mordia. To distinguish between these possibilities, we
Figure 2. Hair-Cell Development and Polarity in the Posterior Lateral-Line Organ of Wild-Type Larvae
(A) When treated with the fluorescent compound 4-Di-2-ASP at 48–52 hpf, only one neuromast on the head (left arrowhead) and one on the
trunk (right arrowhead) display functional hair cells.
(B) By 60 hpf, mature hair cells appear in additional neuromasts in an anterior-to-posterior progression.
(C and D) In lateral (C) and ventral views (D) of a 6 dpf larva, all the neuromasts contain several functional hair cells. The L1 neuromasts are
out of the plane of focus in (D). Note the difference in the positions of equivalent neuromasts on the two sides of the transparent larva.
(E) A surface view of a preparation labeled with rhodamine-phalloidin shows hair bundles protruding from the centers of two neuromasts.
(F) The axis of hair-bundle polarity in neuromast L1 is demonstrated by labeling with rhodamine-phalloidin, which marks the actin in the
stereocilia, and with antiacetylated tubulin, which marks the eccentric kinocilia.
(G) Labeling with rhodamine-phalloidin alone demonstrates parallel cellular polarity by revealing a notch in each actin-rich cuticular plate
corresponding to the position of the kinocilium.
(H) Similar labeling reveals the perpendicular orientation of hair bundles in neuromast L2.
Scale bars: 100 ?m in (A)–(D), 10 ?m in (E)–(G).
used laser-mediated cell ablation to eliminate first-pri-
mordial cells at the onset of migration. As assessed by
labeling at 60 hpf with 4-Di-2-ASP, unilaterally treated
two or three neuromasts situated between somites 6
were always perpendicular. The control side of the fish
displayed a normal organization with both neuromast
polarities. In a converse experiment, elimination of the
out perpendicular neuromasts on their treated sides
(data not shown).
These results suggest that all parallel neuromasts
originate from the first primordium, whereas perpendic-
ular neuromasts stem from the second. To confirm this
Cell Migration and Planar Polarity
Figure 3. Absence of Perpendicular Neuromasts in strauss Mutant Larvae
(A and B) Differential-interference-contrast images of a wild-type larva (A) and a homozygous strauss larva (B) at 4 dpf demonstrate the small
eyes of the mutant.
(C and D) Treatment of the same animals with 4-Di-2-ASP reveals the presence of functional hair cells in mature neuromasts. Wild-type fish
form additional neuromasts (dashed circle) in the rostral part of the organ, whereas strauss mutant animals do not.
(E) At 48–60 hpf, the hair cells of neuromasts appear in pairs (arrowheads).
(F and G) When labeled with 4-Di-2-ASP (F) or with an anti-parvalbumin 3 antiserum and rhodamine-phalloidin (G), these early neuromasts
are seen to be aligned parallel to the anteroposterior body axis.
(H) A higher-magnification view of a neuromast shows the opposing orientations of the two hair bundles whose axis of planar polarity runs
parallel with the axis of hair-cell development.
(I) A high-magnification view of the neuromast circled in (C) reveals the perpendicular development of the first two hair cells.
Scale bars: 100 ?m in (A)–(E), 1 ?m in (F)–(I).
inference, we marked first-primordial cells by uncaging
fluorescein-dextran in them at 20–24 hpf. At 5 dpf, we
consistently found that green-fluorescent cells popu-
lated only parallel neuromasts and the trailing cells con-
necting them, whereas perpendicular neuromasts were
pendicular neuromasts originate exclusively from the
second primordium. Variability in the number of pro-
neuromasts deposited by the second primordium ex-
plains why we observe two (L2–L3) or sometimes three
Figure 4. Origin of Parallel and Perpendicular Neuromasts
(A) After ablation of the first primordium on one side of a wild-type larva, treatment with the fluorescent compound 4-Di-2-ASP at 60 hpf
labels neuromasts on only the untreated side (compare Figure 2B).
(B) At 6 dpf, the treated side develops a single mature neuromast in the rostral part of the posterior lateral-line organ (arrowhead).
(C) Uncaging of fluorescein-dextran in the first primordium labels a neuromast (L1) and the trail of cells left by that primordium (green), whereas
a neuromast derived from the second primordium (L2) is not labeled.
(D and E) Actin labeling of the neuromasts depicted in (C) shows the parallel orientation of hair bundles in the fluorescein-positive neuromast
L1 (D) and the perpendicular orientation of those in fluorescein-negative neuromast L2 (E).
Scale bars: 10 ?m in (C)–(E).
(L2–L4) perpendicular neuromasts in more mature lar-
vae. Together with our previous results, these observa-
tions also suggest that strauss mutant fish are defective
in the formation or viability of the neuromasts derived
from the second primordium.
linear arrangement of neuromasts along the trunk to the
tail (data not shown). These observations indicate that
hedgehog signaling is dispensable for the orientation of
neuromasts and for the planar polarization of hair cells.
When the firstprimordium migrated atypically, however,
the orientationof theaxis ofplanar polarityfor individual
neuromasts was randomized relative to the main body
axes. Most neuromasts nonetheless remained aligned
with the axonal tracks (Figures 7A). Similar observations
were made in wild-type fish treated overnight with 30
?M cyclopamine (data not shown), a potent and highly
specific inhibitor of hedgehog signaling that phenocop-
ies the slow muscle omitted mutation (Hammond et al.,
2003). The uncoupling of neuromast orientation with re-
spect to the main body axes is unlikely to be explained
by body mispatterning: neuromasts remain aligned with
the anteroposterior axis in parachute/Ncad mutant lar-
vae, which are misshapen and present defects in the
development of the subset of muscles that fail to form
in slow muscle omitted mutants (Figures 6F–6H) (Cortes
et al., 2003).
We confirmed these results using fused somites mu-
tant fish, in which abnormal migration of the first primor-
dium results in the mislocalization of some neuromasts
along the trunk of the larva (Gilmour et al., 2002, 2004).
Although its basis is unknown, we took advantage of
in the mutants. At 80 hpf, a fraction of the neuromasts
were mislocalized and no longer aligned with the main
body axes (Figures 6C–6E). Instead, they were more
7B). We also altered the migratory behavior of the pri-
mordium by blocking the formation of the endogenous
sdf1a chemokine and simultaneously creating ectopic
sources of sdf1b, which has recently been shown to
efficiently attract primordial cells (Li et al., 2004). Eggs
were injected with a morpholino specific to sdf1a to-
gether with a plasmid for expression of a fusion protein
of sdf1b and green-fluorescent protein (GFP) driven by
a heat-shock promoter (Li et al., 2004). Several larvae
lacking posterior lateral-line organs developed neuro-
masts in ectopic locations on the trunk or over the yolk
sac (Figure 6I). Many of these ectopic neuromasts had
an axis of polarity that was no longer aligned with the
anteroposterior axis, but that was oriented parallel to
the accompanying axons (Figures 6I–6K). These data
demonstrate that neuromast orientation can be uncou-
pled from the body axes by altering the migratory path
of the primordium, but that most neuromasts remain
Differential Ventral Migration of First-
and Second-Primordial Neuromasts
Although the foregoing experiments demonstrate that
the first and second primordia generate, respectively,
the parallel and perpendicular neuromasts, they do not
explain why these axes of polarity are orthogonal. To
examine this issue, we studied the movements of the
primordia with greater spatial and temporal resolution.
dium migrates along thehorizontal myoseptum just dor-
sal to the path taken by the first primordium (Figures
quently moves ventrally, beginning with the more rostral
neuromasts (Ledent, 2002; Sapede et al., 2002). We un-
expectedly found that pro-neuromasts derived from the
second primordium begin their ventral movement even
before they are fully dislodged from the migrating pri-
mordium (Figures 5A–5D). As they move downward,
these pro-neuromasts deflect the trail of claudin b-posi-
tive cells connecting first-primordial neuromasts (Fig-
ures 4C, 5C, and 5D). The ventrally migrating pro-neuro-
masts of second-primordial origin harbor no mature hair
cells, whereas the neuromasts deposited by the first
primordium already bear mature hair cells but remain
at the level of the horizontal myoseptum (Figure 5B).
By the time first-primordial neuromasts begin to move
ventrally, neuromasts L2 and L3 are already located one
tum(Figures 5Eand 5F).Unlike wild-typeanimals, strauss
mutants do not form ventrally migrating neuromasts
(Figures 5G and 5H), a deficiency reflected in the ab-
sence of perpendicular neuromasts in older larvae.
Altering the Trajectory of Primordial Cells
These observations raise the intriguing possibility that
neuromast orientation is causally related to the migra-
esis, we used fish mutant for slow muscle omitted,
whose loss of smoothened function blocks hedgehog
signaling (Barresi et al., 2000). Mutant fish lack the hori-
zontal myoseptum and in turn the trail of Sdf-1a chemo-
kine along which primordial cells normally migrate. The
first primordium accordingly moves aberrantly, depos-
iting a variable number of neuromasts in a haphazard
pattern (David et al., 2002). We examined neuromast
orientation in larvae at 72–96 hpf, a period during which
we can still unambiguously distinguish neuromasts de-
rived from the two primordia on the basis of their hair-
cell number. In this interval, the nerve fibers can still be
used to trace the trajectory of a primordium, for the
neuromasts have not yet begun to move away from the
main nerve bundle.
In slow muscle omitted larvae (Figures 6A and 6B),
was affected (Figure 6B, inset). Moreover, all first-pri-
mordial neuromasts were aligned parallel to the antero-
posterior body axis in those mutant fish that showed a
We have observed an unexpected degree of complexity
in the patterning of the zebrafish’s posterior lateral-line
organ, whose neuromasts are oriented either parallel
or perpendicular to the anteroposterior body axis. The
generation of neuromasts with different orientations
probably has biological significance, for the response
ulation coincides with the axis of maximal sensitivity
of the hair cells (Mohr and Go ¨rner, 1996). Information
Cell Migration and Planar Polarity
Figure 5. Migration of Pro-Neuromasts Derived from the Second Primordium in Wild-Type and strauss Larvae
(A) Labeling for claudin b marks the rostral end of the posterior lateral-line organ in a wild-type fish at 56 hpf. The arrow marks the initial
location of the second primordium a few cell diameters posterior and dorsal to that of the first. The initial neuromast derived from the second
primordium migrates ventrally and depresses the trail of claudin b-positive cells connecting adjacent first-primordial neuromasts (arrowhead).
(B) Labeling for parvalbumin 3 demonstrates mature hair cells only in neuromasts derived from the first primordium (L1 and L3), whereas
claudin b immunoreactivity additionally marks the second primordium and a pro-neuromast (L2) derived from it.
(C) In a detailed view of the initial neuromast deposited by the second primordium, the arrowheads indicate the trail of claudin b-positive
cells left by the first primordium.
(D) Prior to its ventral migration, a neuromast deposited at the trailing edge of the second primordium already contacts the trail of claudin
b-positive cells derived from the first primordium (arrowheads).
(E and F) Of the five rostral neuromasts in a wild-type fish (dashed circles in [F]), two (L2 and L3) lie about two neuromast diameters below
the horizontal myoseptum.
(G and H) strauss mutants lack these ventrally located neuromasts; instead, all neuromasts (dashed circles in [H]) remain located close to the
Figure 6. ReorientationofNeuromastsAsso-
ciated with the Trajectory of Primordial Cells
(A) Neuromasts assume random positions in
a slow muscle omitted mutant larva labeled
path followed by the primordium, which had
initially migrated toward the posterior, then
turned back anteriorly. In the inset, labeling
the first two levels of planar cell polarity de-
(C) A fused somites mutant fish, labeled for
acetylated tubulin (green) and actin (red),
eral-line organ (arrow) with a single neuro-
mast at the end (arrowhead).
(D) In a high-magnification view of the same
neuromast, axons labeled for acetylated tu-
bulin (green) reveal the trajectory of the pri-
(E) A higher-magnification view of the neuro-
perpendicular to the anteroposterior body
axis but are oriented along the trajectory of
(F and G) Comparison of a wild-type larva (F)
with a parachute/Ncad mutant larva (G) at 4
dpf demonstates that the mutant fish is
shorter, misshapen, and displays defects in
its trunk musculature.
(H) Actin labeling reveals that the planar po-
larization of neuromasts in the mutants is
romasts of the posterior lateral-line organ
was produced by coinjection of an antisense
morpholino against endogenous sdf1a and a
DNA construct forexpression of a sdf1b-GFP
the animal displays functional hair cells (ar-
rowheads); note the absence of neuromasts
along the trunk of the fish. Neuromasts of the
anterior lateral-line organ appear not to be
affected by the treatment. Some sdf1b-GFP-expressing cells (green) can still be seen at this stage (broken circle).
(J) One such neuromast located over the yolk sac is labeled for actin (red) and acetylated tubulin (green) to demonstrate, respectively, hair
bundles and axonal tracks.
(K) A higher-magnification view of this neuromast reveals that hair cells are aligned parallel with the axonal tracks.
derived from neuromasts of the two orthogonal orienta-
tions likely enhances an animal’s ability to evaluate wa-
ter movements during swimming.
The precise conservation of the pattern of planar cell
polarity in the posterior lateral-line organ suggests that
tional information used to establish an animal’s body
plan. Although this hypothesis has not been tested di-
rectly in ourstudy, we believe thatthe local environment
lent neuromasts do not have a fixed location along the
trunk of the fish (Figure 2D), yet they are accurately
ablation, and cell marking, we have ascertained that
the different orientations of parallel and perpendicular
neuromasts instead reflect their independent origins
from, respectively, the first and second primordia. This
early during the third day of development, whereas per-
pendicular neuromasts do not form mature hair cells
until a day later.
Determinants of Global Neuromast Orientation
When the primordium of the posterior lateral-line organ
is deflected from its normal trajectory, the orientation
of theresulting neuromasts israndomized relativeto the
main body axes. Most neuromasts nonetheless remain
aligned with the trajectory of the primordium. A possible
explanation for this behavior is that the direction of mi-
gration of primordial cells at a particular time in their
ity. Because both primordia normally move parallel to
the anteroposterior axis, the presence of perpendicular
neuromasts at first appears to be at odds with this idea.
Our results indicate, though, that pro-neuromasts de-
rived from the second primordium turn ventrally through
Cell Migration and Planar Polarity
Figure 7. Quantification of Neuromast Orien-
tation along the Anteroposterior Axis and the
Trajectory of First-Primordial Cells
(A) Two histograms indicate the frequencies
roposterior axis (left) and to axonal tracks
(right). In wild-type larvae (blue), the axonal
tracks are directed along the anteroposterior
axis, and neuromasts are oriented parallel
with both. The neuromast distributions with
respect to the axis and to the tracks differ
significantly from chance by a ?-square
respectively). For slow muscle omitted mu-
tants (red), the neuromast orientation is es-
sentially random with respect to the main
body axes (p ? 0.7), but largely accords with
the direction of growth of the associated
nerve fibers (p ? 10?9).
(B) Histograms compare the frequencies of
neuromast orientationsfor slowmuscle omit-
For mutants of both types, neuromasts are
ior body axis (purple) than with the axonal
tracks (orange). By a ?-square test for homo-
geneity of categorical data, the distributions
with respect to the anteroposterior axis and
to the axonal tracks differ significantly (p ?
10?4and p ? 0.002, respectively).
90? relative to their original trajectory along the antero-
hair cells. Perpendicular neuromasts may form because
these second-primordial cells remain able to polarize
while migrating ventrally. It has been suggested that the
combination of directional migration by primordial cells
and the timing of neuromast maturation accounts for
the two-dimensional patterning of the posterior lateral-
line organ (Sapede et al., 2002). We propose that the
combination of these two factors also forms part of the
along the animal’s main body axes.
hair cells or move to random locations, suggesting that
hair cells are born symmetrical and that subsequent
signaling controls the asymmetrical localization of the
kinocilium to a specific edge of the cell. It remains un-
clear, however, when asymmetry is first established and
what cellular events occur during the orientation of hair
cells across the sensory epithelium.
Our findings support the idea that directional cell mi-
gration is an early event in the specification of hair-
cell orientation. Although our results do not exclude
the involvement of a morphogen, they suggest that the
global orientation of hair cells in the posterior lateral-
line organ of the zebrafish embryo occurs without the
necessity for long-range signaling. It remains plausible
that a molecular concentration gradient acts simultane-
ously with cell migration to facilitate the interpretation
of the polarizing vector. One intriguing possibility is that
the molecules directing primordial cell migration also
pathway participates in several instances of long-range
cell migration and cell polarization, such as lymphocyte
homing, cancer metastasis, and germ-cell migration
(Doitsidou et al., 2002; Knaut et al., 2003; Molyneaux et
al., 2003). The Sdf1a chemokine could in principle be a
convergence point for polarized cell migration and the
ality, the substance could additionally polarize primor-
dial cells along the migratory path (Reichman-Fried et
al., 2004). Because primordial cells do not survive in
the absence of Sdf1-CXCR4 signaling (our unpublished
Initiation of Global Hair-Cell Orientation
planar polarity has received considerable attention ever
since it was hypothesized to involve a molecular con-
centration gradient (Lawrence, 1966). It has recently
sic cellular properties combine to control the establish-
ment of planar polarity across fields of cells (Tree et al.,
2002; Strutt, 2003; Uemura and Shimada, 2003). The
analysis of mice mutant for loop-tail, a loss-of-function
mutant allele of strabismus, and for Celsr1, a mutation
in flamingo (Kibar et al., 2001; Murdoch et al., 2001;
Curtin et al., 2003), provides compelling evidence that
the core components of the planar cell polarity system
control stereociliary orientation (Curtin et al., 2003;
Montcouquiol et al., 2003). In the cochleas of these mu-
tants, kinocilia either remain centered on the apices of
with a DNA construct containing a fusion between the sdf1b gene
and that encoding GFP. This fusion gene (Li et al., 2004), which is
of the heat-shock 70 promoter (pHsp70/4:sdf1bEgfp). Injected em-
bryos were left to develop at 28.5 C until 12–16 hpf and were subse-
quently heat-shocked by incubation at 39.5 C for 8–10 hr. After
this treatment, misshapen fish and those not expressing the fusion
protein were discarded and the remainder left to develop for an
additional 3 days. Uninjected embryos maintained identically were
used as controls.
data), this hypothesis unfortunately cannot be tested
A Role for Oriented Cell Division in Maintaining
the Polarity Vector
An important question raised by these findings is when
pro-neuromasts acquire vectorial information. Although
it is unknown whether primordial cells are polarized be-
fication of vectorial information occurs during migration
and, furthermore, that the axis of this polarity subse-
quently becomes fixed to maintain neuromast orienta-
tion during organ growth and hair-cell regeneration.
Such a mechanism would explain the observation that
the orientation of hair cells in neuromasts does not
change ontogenetically (Webb, 2000). Mitotically active
supporting cells at the periphery of each neuromast
continually produce new hair cells during development
and regeneration in the zebrafish (Williams and Holder,
2000). That hair cells are born in pairs suggests that
they have a common precursor, a finding consistent
with previous observations on two teleosts in families
distinct from that of the zebrafish (Rouse and Pickles,
1991). Because we found that pairs of hair cells are
produced parallel to the future axis of planar polarity in
the neuromast, we anticipate that hair-cell polarity is
specified by the orientation of the mitotic spindle in the
dividing precursors. Frizzled and other core compo-
nents of the pathway control mitotic-spindle orientation
relative to the main body axes in the pI cell during the
notum (Gho and Schweisguth, 1998). Because the ze-
brafish fz7a gene is highly expressed in migrating pri-
mordia and mature neuromasts (Sumanas et al., 2002),
tant to test this possibility and to identify the molecular
mechanism that controls mitotic-spindle orientation.
Another challenge is to determine how the orientation
of cell division is translated into hair-bundle polarity
and whether hair cells of a pair influence each other’s
orientation. The present work provides a framework for
further investigations into the cell-biological and molec-
ular bases of planar polarity in hair cells.
The 3? end of the coding sequence for claudin b (Kollmar et al.,
2001) was amplified in a PCR with the primers 5?-GAGAGAATTCC
CCGAAAAATCAG-3? and 5?-GAGACTCGAGTTACACAAAGTTC-3?
and inserted between the EcoRI and XhoI sites of the vector pGEX-
5X-1 (Amersham). The resulting fusion protein, which comprised
glutathione-S-transferase and the 30 amino acids at the carboxyl
terminus of claudin b, was expressed in Escherichia coli BL21, puri-
fied on a glutathione column (Amersham), and used to immunize
between the EcoRI and XhoI sites of the vector pET-32a(?) (Nova-
gen). The resulting thioredoxin fusion protein with a hexahistidine
tag was purified from E. coli BL21 (DE3)pLysS under denaturing
conditions on a Ni-NTA His·BindSuperflow column (Novagen). After
covalent coupling to an AminoLink Pluscolumn (Pierce), the peptide
was used to purify antibodies against the carboxyl terminus of clau-
din b from rabbit and rat sera. The specificity of the purified antisera
was demonstrated with immunoblots of bacterial lysates containing
the thioredoxin-claudin b fusion protein, the corresponding thiore-
Labeling Procedures and Imaging
For immunohistochemistry, manually dechorionated zebrafish em-
bryos or free-swimming larvae were collected and fixed overnight
at 4?C in a solution of 4% paraformaldehyde in phosphate-buffered
saline solution containing 1% Tween-20. Samples were washed in
the same solution without fixative and blocked with 10% bovine
serum albumin at room temperature. Primary- and secondary-anti-
body incubations were conducted overnight at 4?C in phosphate-
buffered saline solution with 0.2% Tween-20. Labeling with rhoda-
mine-phalloidin (Molecular Probes) was conducted during the first
wash after secondary-antibody incubation, either overnight at 4?C
or for 4–8 hr at room temperature. Antibodies were used at the
following dilutions: rabbit anti-claudin b, 1/500; rat anti-claudin b,
1/200; mouse antiacetylated tubulin (clone 6-11B-1, Sigma), 1/1000;
and rabbit anti-parvalbumin 3 (Heller et al., 2002), 1/2000. Fluores-
Cy5- and Texas Red-labeled donkey anti-rabbit immunoglobin sec-
ondary antibodies (Jackson Labs) were used at 1/200.
For vital labeling of hair cells, larvae were immersed in a 200
?M solution of 4-(4-(diethylamino)styryl)-N-methylpyridinium iodide
(4-Di-2-ASP; Molecular Probes) for 1 min at room temperature in
the dark. Treated larvae were washed briefly to remove excess
fluorophore, anesthetizedwith 3-aminobenzoicacid ethylester, and
mounted in 3% methylcellulose on a glass slide.
Specimens were examined with a Zeiss Axioplan II microscope
and images were acquired with a Nikon Coolpix-995 digital camera.
Confocal images were obtained with a BioRad MRC-1024ES scan
head mounted on a Zeiss Axiovert microscope or with a Zeiss Meta-
Zebrafish Strains and Husbandry
Zebrafish were maintained in our facility under standardized condi-
mel et al., 1995), and maintained in system water at a density of 50
embryos perPetri dishat 28.5?C. Wild-typefish wereTu ¨bingen Long
Fin (TL),albino, orAB. Animals mutantfor fusedsomites, parachute/
Ncad, and slow muscle omitted carried the fssti1, pac/Ncadm117, and
smub641alleles, respectively. The straussru891mutation was identified
during secondary screening of lines generated in an F3screen con-
ducted in our laboratory by standard procedures (Starr et al., 2004)
following mutagenesis of TL zebrafish with ethylnitrosourea.
Scoring of Neuromast Orientation
Actin- and acetylated tubulin-labeled neuromasts were imaged us-
ing a confocal microscope. Orientations of hair cells in neuromasts
were determined relative to the anteroposterior body axis and nerve
bundles and classified into three categories, each representing the
same angularrange: parallel(0?–30? and150?–180?), oblique(30?–60?
and 120?–150?), and perpendicular (60?–120?). For the left histogram
in Figure 7A, 558 neuromasts were analyzed in 100 wild-type larvae
and 80 neuromasts were examined in 28 mutant larvae. For the right
histogram, 131 neuromasts were analyzed in 25 wild-type larvae
and 91 neuromasts were examined in 47 mutant larvae. For the left
Morpholino Oligonucleotides and DNA Constructs
Morpholino antisense oligonucleotides to sdf1a (5?-CTACTACGAT
CACTTTGAGATCCAT-3?) and to neurogenin-1 (Andermann et al.,
2002) (5?-ACGATCTCCATTGTTGATAACCTGG-3?) were purchased
from Gene Tools (Corvalis, OR). They were diluted in water and
injected into one- to four-cell embryos at a concentration of 0.5 mM.
To block endogenous sdf1a and simultaneously create an ectopic
source of the chemokine, we injected sdf1a morpholino together
Cell Migration and Planar Polarity
were analyzed, and for the right histogram (fss), 90 (purple) and 91
(orange) neuromasts were examined.
and causes severe neural tube defects in the mouse. Curr. Biol.
David, N.B., Sapede, D., Saint-Etienne, L., Thisse, C., Thisse, B.,
basis of cell migration in the fish lateral line: role of the chemokine
receptor CXCR4 and of its ligand, SDF1. Proc. Natl. Acad. Sci. USA
Doitsidou, M., Reichman-Fried, M., Stebler, J., Koprunner, M., Dor-
ries, J., Meyer, D., Esguerra, C.V., Leung, T., and Raz, E. (2002).
Cell 111, 647–659.
Flock, A˚., and Wersa ¨ll, J. (1962). A study of the orientation of the
sensory hairs of the receptor cells in the lateral line organ of fish,
with special reference to the function of the receptors. J. Cell Biol.
Gho, M., and Schweisguth, F. (1998). Frizzled signalling controls
orientation of asymmetric sense organ precursor cell divisions in
Drosophila. Nature 393, 178–181.
Gilmour, D.T., Maischein, H.-M., and Nu ¨sslein-Volhard, C. (2002).
Migration and function of a glial subtype in the vertebrate peripheral
nervous system. Neuron 34, 577–588.
Gilmour, D., Knaut, H., Maischein, H.-M., and Nu ¨sslein-Volhard, C.
(2004). Towing of sensory axons by their migrating target cells
in vivo. Nat. Neurosci. 7, 491–492.
Gompel, N., Cubedo, N., Thisse, C., Thisse, B., Dambly-Chaudiere,
C., and Ghysen, A. (2001). Pattern formation in the lateral line of
zebrafish. Mech. Dev. 105, 69–77.
Haddon, C., and Lewis, J. (1996). Early ear development in the em-
bryo of the zebrafish, Danio rerio. J. Comp. Neurol. 365, 113–128.
Hammond, K.L., Loynes, H.E., Folarin, A.A., Smith, J., and Whitfield,
T.T. (2003). Hedgehog signalling is required for correct anteroposte-
rior patterning of the zebrafish otic vesicle. Development 130, 1403–
Parvalbumin 3 is an abundant Ca2?buffer in hair cells. J. Assoc.
Res. Otolaryngol. 3, 488–498.
Hudspeth, A.J. (1989). How the ear’s works work. Nature 341,
Hudspeth, A.J., and Corey, D.P. (1977). Sensitivity, polarity, and
conductance change in the response of vertebrate hair cells to
controlled mechanical stimuli. Proc. Natl. Acad. Sci. USA 74, 2407–
Keller, R. (2002). Shaping the vertebrate body plan by polarized
embryonic cell movements. Science 298, 1950–1954.
Kibar, Z., Vogan, K.J., Groulx, N., Justice, M.J., Underhill, D.A., and
Gros, P. (2001). Ltap, a mammalian homolog of Drosophila Strabis-
mus/Van Gogh, is altered in the mouse neural tube mutant Loop-
tail. Nat. Genet. 28, 251–255.
Kimmel, C.B., Ballard, W.W., Kimmel, S.R., Ullmann, B., and Schil-
ling, T.F. (1995). Stages of embryonic development of the zebrafish.
Dev. Dyn. 203, 253–310.
Knaut, H., Werz, C., Geisler, R., and Nu ¨sslein-Volhard, C. (2003). A
zebrafish homologue of the chemokine receptor Cxcr4 is a germ-
cell guidance receptor. Nature 421, 279–282.
Kollmar, R., Nakamura, S.K., Kappler, J.A., and Hudspeth, A.J.
(2001). Expression and phylogeny of claudins in vertebrate primor-
dia. Proc. Natl. Acad. Sci. USA 98, 10196–10201.
Lawrence, P.A. (1966). Development and determination of hairs and
bristles in the milkweed bug, Oncopeltus fasciatus (Lygaeidae, He-
miptera). J. Cell Sci. 1, 475–498.
Ledent, V. (2002). Postembryonic development of the posterior lat-
eral line in zebrafish. Development 129, 597–604.
Lewis, J., and Davies, A. (2002). Planar cell polarity in the inner ear:
how do hair cells acquire their oriented structure? J. Neurobiol.
Li, Q., Shirabe, K., and Kuwada, J.Y. (2004). Chemokine signaling
regulates sensory cell migration in zebrafish. Dev. Biol. 269,
Laser Cell Ablation and Cell Marking
Wild-type embryos at 20–22 hpf were manually dechorionated,
mounted in methylcellulose, and viewed with a 60? water-immer-
sion objective lens. To ablate primordial cells, we used a Micropoint
Laser System (Photonic Instruments, Inc.) adapted to a Zeiss Axio-
plan II microscope. One side of each animal was used as control.
After the treatment, embryos were transferred to fresh system water
and raised at 28.5?C. The success of the ablation was tested the
following day by treating fish with 4-Di-2-ASP and scoring the num-
ber of neuromasts on both sides. Each fish was photographed and
labeled at 5 dpf.
For labeling of the first primordium, embryos at the one- to four-
cell cell stage were injected with 4% 10 kDa dextran conjugated to
4,5-dimethoxy-2-nitrobenzyl-caged fluorescein (Molecular Probes)
in Danieau’s solution. Animals were left to develop at 28.5?C in the
dark. At 20 hpf, they were mounted as described for laser ablation.
After the first primordium had been located, uncaging was accom-
plished by illumination through a 60? water-immersion objective
lens for 1–2 s with the microscope’s fluorescence-excitation system
at full power. The pattern of uncaging was regulated by the field
diaphragm of the illumination system; the degree of uncaging was
determined by observation through a 20? objective lens.
We thank S. Devoto, S. Holley, and the Zebrafish International Re-
source Center for mutant zebrafish lines, J. Kuwada and Q. Li for
DNA constructs, K. Hammond and T. Whitfield for advice on cyclo-
pamine treatment, A. Carmany-Rampey and E. Chiappe for advice
on lasercell ablation,Y. Castellanosand C.Denis for thepurification
of antisera, A. Afolalu and P. Espitia for expert maintenance of
the zebrafish colony, and M. Farin ˜as for statistical consultation.
R. Benton, S. Desbordes, M. Gonza ´lez-Gaita ´n, and the members of
our research groupprovided valuable comments onthe manuscript.
This research was supported by grant DC00241 from the National
Institutes of Health. H.L.-S. was supported by a Wellcome Trust Inter-
national Travelling Research Fellowship; A.J.H. is an Investigator of
Howard Hughes Medical Institute.
Received: April 21, 2004
Revised: June 30, 2004
Accepted: July 6, 2004
Published: September 13, 2004
Adler, P.N. (2002). Planar signaling and morphogenesis in Drosoph-
ila. Dev. Cell 2, 525–535.
Alexandre, D., and Ghysen, A. (1999). Somatotopy of the lateral line
projection in larval zebrafish. Proc. Natl. Acad. Sci. USA 96, 7558–
Andermann, P., Ungos, J., and Raible, D.W. (2002). Neurogenin1
defines zebrafish cranial sensory ganglia precursors. Dev. Biol.
Barresi, M.J., Stickney, H.L., and Devoto, S.H. (2000). The zebrafish
slow-muscle-omitted gene product is required for Hedgehog signal
ment 127, 2189–2199.
Cortes, F., Daggett, D., Bryson-Richardson, R.J., Neyt, C., Maule,
J., Gautier, P., Hollway, G.E., Keenan, D., and Currie, P.D. (2003).
Cadherin-mediated differential cell adhesion controls slow muscle
cell migration in the developing zebrafish myotome. Dev. Cell 5,
Curtin, J.A., Quint, E., Tsipouri, V., Arkell, R.M., Cattanach, B., Copp,
Mutation of Celsr1 disrupts planar polarity of inner ear hair cells
Developmental Cell Download full-text
Metcalfe, W.K. (1985). Sensory neuron growth cones comigrate with
posterior lateral line primordial cells in zebrafish. J. Comp. Neurol.
Metcalfe, W.K., Kimmel, C.B., and Schabtach, E. (1985). Anatomy
of the posterior lateral line system in young larvae of the zebrafish.
J. Comp. Neurol. 233, 377–389.
Mohr, C., and Go ¨rner, P. (1996). Innervation patterns of the lateral
tion during metamorphosis. Brain Behav. Evol. 48, 55–69.
Molyneaux, K.A., Zinszner, H., Kunwar, P.S., Schaible, K., Stebler,
J., Sunshine, M.J., O’Brien, W., Raz, E., Littman, D., Wylie, C., and
Lehmann, R. (2003). The chemokine SDF1/CXCL12 and its receptor
CXCR4 regulate mouse germ cell migration and survival. Develop-
ment 130, 4279–4286.
Montcouquiol, M., Rachel, R.A., Lanford, P.J., Copeland, N.G., Jen-
kins, N.A., and Kelley, M.W. (2003). Identification of Vangl2 and
Scrb1 as planar polarity genes in mammals. Nature 423, 173–177.
Montgomery, J., Baker, C., and Carton, G. (1997). The lateral line
can mediate rheotaxis in fish. Nature 389, 960–963.
Murdoch, J.N., Doudney, K., Paternotte, C., Copp, A.J., and Stanier,
P. (2001). Severe neural tube defects in the loop-tail mouse result
tion. Hum. Mol. Genet. 10, 2593–2601.
Reichman-Fried, M., Minina, S., and Raz, E. (2004). Autonomous
modes of behavior in primordial germ cell migration. Dev. Cell 6,
Rouse, G.W., and Pickles, J.O. (1991). Paired development of hair
cells in neuromasts of the teleost lateral line. Proc. R. Soc. Lond.
B. Biol. Sci. 246, 123–128.
Sapede, D., Gompel, N., Dambly-Chaudiere, C., and Ghysen, A.
(2002). Cell migration in the postembryonic development of the fish
lateral line. Development 129, 605–615.
Shotwell, S.L., Jacobs, R., and Hudspeth, A.J. (1981). Directional
sensitivity of individual vertebrate hair cells to controlled deflection
of their hair bundles. Ann. N Y Acad. Sci. 374, 1–10.
Starr, C.J., Kappler, J.A., Chan, D.K., Kollmar, R., and Hudspeth,
A.J. (2004). Mutation of the zebrafish choroideremia gene encoding
Rab escort protein 1 devastates hair cells. Proc. Natl. Acad. Sci.
USA 101, 2572–2577.
Strutt, D. (2003). Frizzled signalling and cell polarisation in Drosoph-
ila and vertebrates. Development 130, 4501–4513.
Sumanas, S., Kim, H.J., Hermanson, S.B., and Ekker, S.C. (2002).
Lateral line, nervous system, and maternal expression of frizzled 7a
during zebrafish embryogenesis. Mech. Dev. 115, 107–111.
Tree, D.R., Ma, D., and Axelrod, J.D. (2002). A three-tiered mecha-
nism for regulation of planar cell polarity. Semin. Cell Dev. Biol.
Uemura, T., and Shimada, Y. (2003). Breaking cellular symmetry
Webb, J.F. (2000). Mechanosensory lateral line: microscopic anat-
omy and development. In The Handbook of Experimental Animals—
The Laboratory Fish, G.K. Ostrander, ed. (San Diego: Academic
Williams, J.A., and Holder, N. (2000). Cell turnover in neuromasts of
zebrafish larvae. Hear. Res. 143, 171–181.