The lateral line of zebrafish: a model system for the analysis of morphogenesis and neural development in vertebrates.

Abstract

The lateral line of the zebrafish has many of the advantages that made the sensory organs of Drosophila a very productive model system: 1) it comprises a set of discrete sense organs (neuromasts) arranged in a defined, species-specific pattern, such that each organ can be individually recognized; 2) the neuromasts are superficial and easy to visualize, and the innervating neurons are easy to label; 3) the sensory projection is simple yet reproducibly organized. Here we describe some of the tools that can be used to investigate the development of this system, and we illustrate their usefulness with specific examples. We conclude that the lateral line is uniquely suited among vertebrate sensory systems for a molecular, cellular and genetic analysis of pattern formation and of neural development.

Full text

Page 1

Review
The lateral line of zebrafish: a model system for the analysis
of morphogenesis and neural development in vertebrates
Christine Dambly-Chaudière *, Dora Sapède, Fabien Soubiran,
Kelly Decorde, Nicolas Gompel1,Alain Ghysen
Laboratoire de Neurogénétique, INSERM E343 Université Montpellier II, cc103 Place E. Bataillon 34095 Montpellier, France
Received 8 October 2003; accepted 17 October 2003
Abstract
The lateral line of the zebrafish has many of the advantages that made the sensory organs of Drosophila a very productive model system:
1) it comprises a set of discrete sense organs (neuromasts) arranged in a defined, species-specific pattern, such that each organ can be
individually recognized; 2) the neuromasts are superficial and easy to visualize, and the innervating neurons are easy to label; 3) the sensory
projection is simple yet reproducibly organized. Here we describe some of the tools that can be used to investigate the development of this
system,andweillustratetheirusefulnesswithspecificexemples.Weconcludethatthelaterallineisuniquelysuitedamongvertebratesensory
systems for a molecular, cellular and genetic analysis of pattern formation and of neural development.
© 2003 Elsevier SAS.All rights reserved.
Keywords: Cell labelling; Uncageing; BrdU; Time-lapse; DiI; Clonal analysis
1. Model systems
Morphogenesis reflects the ability of biological material
to spatially organize itself, to assume a distinct shape, to
generate a defined pattern. This process remains a central
concern of modern biology. Evidence has recently accumu-
lated to show that there is a widespread conservation of
developmental mechanisms and of the underlying sets of
genes in very distantly related animals. One important con-
sequence of this conservation is that the elucidation of a
developmental mechanisms in any one species will likely be
of general relevance. At the same time, the current conclu-
sion that all triplobasts derive from a single bilaterian ances-
tor species (Arendt and Nübler-Jung, 1994; De Robertis and
Sasai, 1996) raises the question of how the ancestral mor-
phology has undergone the extreme variation manifested by
its extant progeny, while most of its genes and the processes
they code for have remained highly conserved (Ghysen,
2003).
Understanding how morphological variation is generated
during the course of evolution presumably depends upon
building up a thorough understanding of a few developmen-
tal processes in species with very distinct morphologies. In
his remarkable analysis of human behaviour and science,
Adams (1980) elaborated on the idea that, since the effect of
gravity extends throughout the universe, the analysis of any
object-say,acrumbofcake-could,ifproperlydone,inform
us about the state of affairs of all galaxies. In a similar way,
onecouldsaythattheanalysisofanybiologicalobjectandof
its development will inform us about all other biological
objects and their development. This may explain why major
advances in understanding human disease have come from
such unexpected quarters as a yeast, a worm or a fly.
Thereasonwhyyeast,wormandflyhavebeensousefulis
that each of these systems had special qualities that turned
out to be most appropriate for a particular field of study, i.e.
they lent themselves to “properly done” analyses. One major
quality is, of course, the genetic tractability of these organ-
isms.Anotherqualityisthatprocessesandphenotypescanbe
studied at the level of the single cell. Since development is
the result of cells organizing themselves and their neigh-
bours, the description of developmental processes needs to
reach the single-cell resolution. It is no surprise, therefore,
thatcanonicalsystemssuchastheflyandthenematodeallow
the contribution of individual cells to be recognized and
* Corresponding author.
E-mail address: cdambly@univ-montp2.fr (N. Gompel).
1Present address: Howard Hughes Medical Institute R.M. Bock Labora-
tories, 1525 Linden drive Madison WI53706, USA.
Biology of the Cell 95 (2003) 579–587
www.elsevier.com/locate/bicell
© 2003 Elsevier SAS.All rights reserved.
doi:10.1016/j.biolcel.2003.10.005

Page 2

analysed - indeed, this was probably a condition of their
success.
The importance of single cell resolution in developmental
analysesisillustratedbytheremarkableprogressofdevelop-
mentalgeneticsoverthepast25years[ifweconsider(Lewis,
1978; Nüsslein-Volhard and Wieschaus, 1980), as the start-
ing blocks for this new era of biology]. This progress has
crucially depended on the analysis of the 24 hours old fly
embryo, where most visible structures are produced by iden-
tified cells. Further progress on fly developmental genetics
hasdependedonother,equallywell-definedsystems,suchas
the sense organs, with their invariant pattern, fixed cell num-
ber and defined lineages.
In contrast, there has been a lack of simple, defined and
experimentally accessible model systems in vertebrates, in
particular for the analysis of neural development. Yet the
need for such model systems is high, for several reasons.
First, vertebrates belong to one major branch of metazoans,
the deuterostomians, while flies and worms belong to the
other major branch, the prostostomians. Being able to draw
comparisons between proto- and deuterostomians may not
only lead to the identification of new mechanisms, but also
help us to understand how the same mechanisms have been
used to achieve very different effects.
Second, if this model system has undergone changes
within the vertebrate lineage, its study in different species
may help to understand how evolution can walk the many
interconnected steps that mediate the hazardous transition
from one species to another one. Third, being ourselves
vertebrates,weareunderstandablykeentohaveatleastsome
model systems closer to our own than, say, the nematode
vulva or the fly hairs. Here we will present one such system,
which presents qualities of accessibility, definition and trac-
tabilitysimilartothoseoftheflysenseorgans,butbelongsto
the vertebrate world: the zebrafish lateral line.
2. The fish lateral line: background
The lateral line is a sensory system of fish and amphibians
thatiscloselyrelated(andprobablyancestral)toourauditory
system. It comprises a set of discrete sense organs, the neu-
romasts, arranged on the body surface in species-specific
patterns, and a corresponding set of neurons that extend their
axons in the hindbrain. The neuromasts are arranged in lines
that extend on the body surface or sometimes in subepider-
mal canals. The neuromasts on the head form the so-called
anterior lateral line system (ALL) whose ganglion is located
between the ear and the eye, while the neuromasts on the
body and tail, including those on the caudal fin, form the
posterior lateral line system (PLL) whose ganglion is just
posteriortotheear(Fig.1A).Thisreviewwillconcentrateon
the posterior lateral line.
The lateral line responds to water movements and is in-
volved in a large variety of behaviours such as predator
avoidance, prey detection, or swimming in schools. It seems
to act mostly as a sense of distant touch, as proposed by
Dijkgraaf (1989). It has evolved into an electrosensory sys-
tem in some fish species, through a specialization of the hair
cell receptors, and has disappeared in terrestrial vertebrates.
The lateral line presents striking similarities with the sensory
hairs of insects, and it seems likely that both derive from a
common, ancestral mechanosensory organ (Adam et al.,
1998, Eddison et al., 2000, Fritzsch and Beisel, 2001).
Each neuromast (Fig. 1B) is made of hair cells (Fig. 1C)
whichresembleverymuchthoseinourinnerear,surrounded
by support cells. Neuromasts are generally exposed on the
body surface, but they may also be embedded in pits or in
subepidermalcanals.TheyareeasilyvisualizedwithNomar-
ski optics or after incubation of the fish with the vital fluores-
cent marker, 2-Di-4-Asp (Collazo et al., 1994), which accu-
mulates specifically in active hair cells (Fig. 2A).
Neuromasts are innervated by sensory neurons (Fig. 1B) as
well as by efferent fibers that control the sensitivity of the
system(Fig.1C).Thesensoryneuronsarebipolarandextend
a central projection into the hindbrain.
A most unusual aspect of the development of the lateral
line system is that the neuromasts are deposited by migrating
primordia which originate from cephalic placodes (Harrison,
1904;Stone,1922,1933).ThePLLplacodeformsatthelevel
of the hindbrain, just posterior to the otic placode. In ze-
brafish, the primordium begins its migration about 22 hours
after fertilization (haf), and the primary PLL is complete by
Fig. 1. A: Orgnisation of the lateral line system. The lateral line comprises
two parts: the anterior system (green) which includes all the neuromasts of
the head, and the posterior system (red), which includes the neuromast on
thebodyandtail.Thesensoryneuronsaregatheredrespectivelyinapre-otic
and a post-otic ganglion (dotted lines). B: Structure of a neuromast. The
mechanosensory hair cells are surrounded by support cells which secrete a
gelatinouscupulaensheathingthehaircellprocesses,andbyaringofmantle
cells. Each neuromast is innervated by at least two sensory neurons. C:
Function of the hair cell. The cell is depolarized by movement in one
direction, and hyperpolarized by the opposite movement. In addition to its
afferent (sensory) innervation, it receives two types of efferent input: an
inhibitory input coming from the hindbrain, and an excitatory input coming
from the forebrain. Figure reproduced from the “Encyclopedia of Neuros-
cience”, 3dedition, Elsevier, 2003..
580
C. Dambly-Chaudière et al. / Biology of the Cell 95 (2003) 579–587

Page 3

the end of embryogenesis, at 48 haf (Metcalfe et al., 1985).
This line comprises 5 regularly spaced neuromasts along
each flank, and 2-3 terminal neuromasts at the tip of the tail
(Fig. 3A). The PLL is progressively completed during larval
life, first through a reiteration of the embryonic process
(Sapède et al., 2002) that generates a reproducible pattern of
about 50 sensory organs arranged in four lines, then by a
budding process whereby each organ is transformed in a
“stitch” of 10-20 neuromasts aligned along the dorso-ventral
axis (Ledent, 2002).
3. Tools for studying cell migration and proliferation
The PLL placode is first detected at about 18-20 hours
after fertilization (Kimmel et al., 1995). It comprises two
unequal groups of cells: a rostral group of about 20 cells,
which stay put and form the ganglion (sensory neurons and
glial cells), and a caudal group of 100 to 120 cells, the
primordium. The latter group migrates along the horizontal
myoseptumallthewaytothetipofthetailataspeedofabout
150 µm/hour (Fig. 3B). During this migration the primor-
dium deposits in its wake the groups of cells that will even-
tually differentiate into neuromasts.
The transparency of the fish embryos allows one to follow
themigrationoftheprimordiumbytime-lapseimagingusing
Nomarski optics (Fig. 3C). Using this method we showed
that the position of a cell relative to its neighbours may show
small fluctuations, indicating that the cells migrate indepen-
dently of each other, but over a long distance the relative
positions are very much conserved. Time-lapse videos also
demonstrate that cells keep dividing as they migrate, and
keep migrating as they divide, i.e. the cytoskeletal rearrange-
ments associated with chromosome sorting and cytokinesis
arekeptseparatefromthecytoskeletalinvolvementinmigra-
tion (Gompel et al., 2001).
Fig. 2. Vital labelling of lateral line cells. Incubation in the vital dye
2Di4Asp specifically labels the hair cells of the neuromasts (and the olfac-
tory cells).A: three focal planes of a neuromast, to illustrate the diagram of
Fig. 1B. Figures taken with a Nipkow confocal system on a ZeissAxioplan.
B:After prolonged incubation, the dye is taken up by the neurons and labels
theircellbodiesandcentralprojection.C:Thecentralprojection(dottedbox
in panel B) can be visualized with considerable detail in a living embryo.
Projection of a set of focal planes taken with a water-immersion 40x Zeiss
objective, NA 1.2, through a Nipkow confocal system.
Fig. 3. Development of the embryonic line. A: The complete embryonic
pattern of the posterior line consists of 5 regularly spaced neuromasts
(named L1-L5) and 2-3 terminal neuromasts (ter). Note that the neuromasts
on the other side of the fish are also seen, due to complete transparency and
small width of the embryo. The anterior line extends around the eye and on
the jaw. B: The neuromasts (green dots) are deposited by a primordium that
moves from the otic region to the tip of the tail following a defined pathway
(red dotted line). The head neuromasts are also deposited by migrating
primordia. C: The posterior primordium as seen with Nomarski optics. The
primordium comprises about 100 cells. D: Mitotically quiescent cells as
detected by incubating the embryo in a solution of bromodeoxyuridine
(BrdU) at gastrula stage (8-10 haf) and revealing the presence of incorpora-
ted BrdU in the primordium at 30 haf. The BrdU labelling is diluted by
successive cell divisions in all but the mitotically quiescent cells, which can
thereforebeidentifiedbystrongerlabelling.E,F:Mitoticallyquiescentcells
in a neuromast at 48haf (E) and after 6 days (F). G: Gene expression in the
migratingprimordium.Insituhybridizationrevealsspecificpatternsofgene
expression within the primordium. In this case, the purple labelling reveals
the expression of cxcr4, a gene that codes for the homolog of the human
chemokinereceptorCXCR4.Thisgeneisdownregulatedinthetrailingcells
(left end of the primordium) which will soon stop migrating and will form a
neuromast.
581
C. Dambly-Chaudière et al. / Biology of the Cell 95 (2003) 579–587

Page 4

A more extensive analysis of cell proliferation can be
achievedbyvisualizingtheincorporationofbromodeoxyuri-
dine (BrdU).This is particularly easy with the zebrafish, as it
suffices to dip the embryo in a dilute BrdU solution to
observe incorporation. Incubating embryos for 30 min and
letting them grow for another hour (to allow for DNA syn-
thesis) reveals the presence of DNA replicating cells in the
migrating primordium, confirming the observations made on
time-lapse movies.
Just for curiosity we also performed BrdU incorporations
during the gastrula stage and examined the fish on the next
day, when the label has been largely diluted in most cells
through successive divisions. Our preliminary results sug-
gest that a substantial number of cells in the primordium
retain a high level of label (Fig. 3D), indicating that these
cellsstoppeddividingat,orsoonafter,thegastrulastage.The
implication that the primordium comprises two different cell
populations, one committed to proliferate while the other
stopped dividing at gastrula, should be examined more care-
fully.Labelisalsopresentinthedifferentiatedneuromastsof
48 haf embryos, except in the hair cells at the center of the
neuromast (Fig. 3E). This is not surprising as new hair cells
areconstantlyproduced(WilliamsandHolder,2000).Indeed
at later times the persistantly labelled cells become confined
to an outer rim within the neuromasts (Fig. 3F).
4. Tools for studying neuromast deposition
We had shown by time-lapse videos that the deposition of
a neuromast is a progressive process whereby a group of
about 15-20 cells at the trailing edge of the primordium
begins to slow down, becomes split from the bulk of the
primoridum which keeps migrating at a constant speed, and
eventually stops moving altogether. Based on the idea that
this difference in behaviour might entail a difference in gene
expression, we have screened a library of in situ hybridiza-
tions for genes that would be heterogeneously expressed
within the primordium, and found one gene whose expres-
sion parallels the migratory behaviour of the cells: it is
expressed at a high level in the migrating cells, at a lower
level in the trailing cells which are about to slow down, at a
very weak level in the cells that are being deposited, and not
at all in the neuromasts (Fig. 3G, Gompel et al., 2001a).
Subsequent work demonstrated that this gene, which codes
for the chemokine receptor CXCR4, indeed plays a crucial
role in the control of migratory behaviour (David et al.,
2002). Interestingly, this receptor also plays a prominent role
in the control of germ cell migration, not only in fish (Doitsi-
dou et al., 2002, Knaut et al., 2003) but also in mouse
(Molyneaux et al., 2003). Due to its visibility and reproduc-
ibility, the pattern of neuromasts may therefore provide a
very convenient and sensitive assay for CXCR4-mediated
cell migration - a process that has also been implicated in the
migration of cancer cells and the formation of metastases
(Homey et al., 2001).
While the presence of differentiated neuromasts is easily
assessed by DiAsp labelling, their absence in experimental
situations is less easy to interpret. It may result from the lack
of deposition, from normal deposition but a lack of differen-
tiation, or from normal deposition followed by cell death, or
from a number of other possibilities. There is fortunately a
simple method to examine cell deposition and survival inde-
pendently of differentiation. Injecting caged fluorescein in
the zygote will have no effect on its development. The caged
compound can be uncaged at any time and in any group of
cells, however, by a very brief pulse of UV light which does
noharmtotheirradiatedcells,butmakesthemfluorescent.If
uncageing is done on the primordium, then the migrating
cells will become fluorescent, and all their progeny can be
detected (Fig. 4A). This method is currently being used to
sort out various genetic or experimental treatments that alter
the patterning of the PLL.
Fluorochrome uncageing can also be used in the course of
normaldevelopment,todocumentveryslowmovementsthat
would be barely detectable with time-lapse video. This
method allowed us to demonstrate that proneuromast depo-
sition is not an abrupt cessation of movement, but rather a
progressive slowing down of the trailing cells in the migrat-
ing primordium, such that even after they have separated
from the primordium they are still capable of migrating one
or at most two somites before eventually stopping and differ-
entiating (Gompel et al., 2001).This result, together with the
observation that deposition occurs at regular intervals rather
than at defined positions, suggests that the deposition is
probably not triggered by external, positional cues, but by a
partitioning process intrinsic to the primordium itself.
5. Tools for clonal analysis
Theso-calledPLLplacode,canfirstbeobservedposterior
to the otic vesicle at about 18 haf (Kimmel et al., 1995). The
development of the PLL might begin much earlier, however,
since it has been reported that many of the sensory neurons
are postmitotic at gastrulation (Metcalfe, 1989), as we ob-
served for at least some of the primordium cells (see above).
Fate-mapping experiments have suggested that the ALL,
PLLandearoriginatefromcontiguousandpossiblyoverlap-
ping regions of the gastrula (Kozlowski et al. 1997). This
resultisconsistentwithobservationsontheembryosofother
fishes and in amphibians (e.g. Schlosser and Northcutt,
2000).
In an attempt to better characterize the early development
of the system, we have injected single cells with a vital tracer
at the blastula stage (Fig. 4B). The usefulness of such clones
may seem limited in view of the fact that the lineage of
blastodermcellsishighlyindeterminate,andnofatemapcan
be defined at this stage (Kimmel and Warga, 1987). As
gastrulationbegins,however,singlecellsgiverisetoprogeny
that are often confined to a given tissue (Kimmel et al., 1990)
and at least some of the cells become committed to specific
582
C. Dambly-Chaudière et al. / Biology of the Cell 95 (2003) 579–587

Page 5

fates (Ho and Kimmel, 1993).As a result, blastoderm clones
often give rise to subclones that include different elements of
the ear and lateral line system (Fig. 4C, D, Gompel et al,
2001).
The analysis of clones that label the primordium revealed
that the proportion of labelled cells is highly variable, rang-
ingfrom1or2cellstomorethan50%(Fig.4E,5).Giventhat
the primordium size is around 100 cells (Gompel et al.,
2001b), we evaluate that the precursor region comprises
about 30 cells at the time it is determined, and that the entire
PLL placode, including the neurons, could comprise about
50 cells.This corresponds roughly to the size of the region of
the gastrula defined as the minimum PLL territory by Ko-
zlowski et al. (1997) with fate-mapping experiments.
A more complete analysis of the sequence of determina-
tive events that lead to PLL formation will depend on injec-
tionsatthegastrulastage.Asinjectionismoredifficultatthis
stage, in part due to the smaller size of the cells, this part of
the analysis is still in progress.
6. Lineage relationships
Hair cells are the central element of mechanoreception in
the ear as well as in the lateral line organs. New hair cells can
Fig.4. Labellingoflaterallinecells.A:UncageingwasperformedwithasmallbeamofU.V.lightwhentheprimordiumwasatthepositionmarkedbythelarge
arrowhead. The cells exposed to the beam become fluorescent. Among them are myofibers, which do not move and mark the position of irradiation, and
primordium cells, which keep migrating and form neuromasts (L2 - L4) as well as the glial wrapping of the lateral line nerve (arrows). B. Single cell injection
at the blastoderm stage. C, D: the same embryo at 27 haf, at two magnifications.This represents a typical case where most of the labelling ends up in the ear and
lateral line. In the lateral line, this clone contributed to neuromast L1 and migrating primordium (prim) as well as glia surrounding the PLL nerve and several
cellsinthePLLganglion,atleastsomeofwhichhavealreadybeguntodifferentiateandtosendanaxonintothehindbrain(arrowheadinpanelD).Thepresence
of unlabelled cells in the ear and the primordium indicates that the injected blastomere did not contribute the entire structures. E: Nomarski view of the
primordium in a different embryo, to which the fluorescence pattern (red) is superimposed. In this case, the number of cells contributed by the injected
blastomere is much lower. Scale bars: C, D: 100 µm; E: 25 µm.
583
C. Dambly-Chaudière et al. / Biology of the Cell 95 (2003) 579–587