Download Cardiovascular development in the zebrafish

31
Development 119, 31-40 (1993)
Printed in Great Britain © The Company of Biologists Limited 1993
Cardiovascular development in the zebrafish
I. Myocardial fate map and heart tube formation
Didier Y. R. Stainier, Robert K. Lee and Mark C. Fishman
Cardiovascular Research Center, Massachusetts General Hospital, 149 13th Street, Charlestown, MA 02129, USA and
Department of Medicine, Harvard Medical School, Boston, MA 02115, USA
SUMMARY
We have analyzed the origin of cardiac progenitors in
the zebrafish embryo by injection of single blastomeres
with a lineage tracer dye, and examined the formation
of the zebrafish heart tube by serial sectioning of
immunostained embryos. At the 512-cell stage (early
blastula), most cardiac progenitors lie in a marginal zone
that extends from 90° longitude (midway between the
future dorsal and ventral axis) through 180° longitude
(the future ventral axis) to 270° longitude. By focusing
on myocardial progenitors located at 90° (and 270°)
longitude, we found that a single cell injected in the early
blastula can contribute progeny to both the atrium and
ventricle. A cell injected in the midblastula contributes
progeny to either the atrium or ventricle, but not both.
This analysis suggests that, at least for these myocardial
progenitors, the atrial and ventricular lineages separate
in the midblastula.
Precardiac cells involute early during gastrulation and
turn towards the animal pole with other early involuting
cells. These cardiogenic cells reach the embryonic axis
around the 8-somite stage, and there they coalesce to
form a pair of myocardial tubular primordia on either
side of the midline. By the 21-somite stage, the
tropomyosin-immunoreactive myocardial tubes have
moved closer to each other, and a distinct group of cells,
the endocardial progenitor cells, sits medially between
them. The myocardial tubes then fuse to enclose the
endocardial cells and form the definitive heart tube. By
22 hours postfertilization (26-somite stage), the heart
tube is clearly beating. The regionalization of cardiac
myosin heavy chain expression distinguishes the cardiac
chambers at this stage, although they are not morphologically delineated until 36 hours.
This work shows that cardiogenic regions can be identified in the early blastula, and that chamber restriction
seems to arise in the midblastula. Additionally, it
provides the basis for embryological perturbation at the
single cell level, as well as for the genetic analysis of heart
tube formation in the zebrafish.
INTRODUCTION
opmental milestones (reviewed by DeHaan, 1965).
Although the chick and amphibian have provided most of
our current knowledge about cardiac development
(reviewed by Litvin et al., 1992), the zebrafish offers several
advantages. Because the embryo is transparent, it is feasible
to follow the progeny of individually tagged blastomeres
(Kimmel, 1989). Since the zebrafish is so small and fertile,
and the embryonic heart so prominent and easily scored,
genetic screens for cardiac mutations are feasible (Stainier
and Fishman, 1993).
In the chick, tissue capable of cardiac differentiation
forms a pair of mesodermal crescents on either side of the
primitive streak (Rawles, 1936; De Haan, 1963; Rosenquist,
1966; Gonzalez-Sanchez and Bader, 1990). In the neurulating amphibian, heart rudiments lie just lateral to the presumptive hindbrain (Wilens, 1955; Jacobson, 1961). In all
vertebrates, these cardiogenic precursors appear to migrate
medially as part of the lateral plate mesoderm, and to
generate a pair of tubular primordia, one on either side of
the midline. These tubular primordia then fuse to generate
During metazoan development, individual cells progressively acquire their differentiated state through a combination of signals derived from their lineage and their environment. The genetic analysis of invertebrate development has
provided much insight into how a single cell comes to
assume a specific fate (Horvitz and Sternberg, 1991;
Greenwald and Rubin, 1992). The highly synchronized
differentiation of groups of cells to form specific organs is
less well understood. Organogenesis requires, in addition to
single cell fate decisions, the assembly of cells of different
embryological origins, the coordination of function of the
different tissues within the organ, and the integration with
other systems for effective physiological homeostasis.
The heart is an ideal model system for both embryological and genetic approaches to organogenesis since it
develops early, is readily accessible for observation and
manipulation, is composed of a limited number of cell types,
and has been well described with regard to particular devel-
Key words: lineage, cardiac progenitors, heart tube, myocardium,
endocardium, heart beat, looping, cardiac chambers
32
D. Y. R. Stainier, R. K. Lee and M. C. Fishman
the definitive heart tube, which consists of an inner, endocardial tube and an outer, myocardial tube. The definitive
heart tube is subsequently divided into separate chambers.
The fish heart consists of four chambers in series: the sinus
venosus, the atrium and the ventricle (the most prominent
chambers), and the bulbus arteriosus, the latter being continuous with the ventral aorta (Santer, 1985; Stainier and
Fishman, 1992).
It is not known whether separate lineages contribute to the
different chambers of the definitive heart tube. In previous
experiments, by treating early zebrafish embryos with low
doses of retinoic acid, we found that myocardial progenitors
appear to acquire their axial coordinates (which seem to
determine their chamber assignment) during early gastrulation (Stainier and Fishman, 1992). We were therefore interested in the possibility that the atrial and ventricular lineages
might separate at a very early stage.
This investigation focuses on the earliest events in cardiac
Fig. 1. Lineage analysis. Single
blastomeres in the early blastula
(256/512-cell stage) or the midblastula
(2000-cell/high stage) were injected
with rhodamine dextran (at left is
diagrammed an injection at 90°
longitude). V, ventral; D, dorsal. At 36
hpf (right), the embryos were
examined for the presence of fluorescent cells in the beating heart. At this time, the distinction between chambers and between myocardial
and endocardial layers is readily apparent. A, atrium; V, ventricle; arrows mark the direction of blood flow.
Fig. 2. Single blastomeres injected at the midblastula stage exhibit cardiac chamber restriction. Computer enhanced images. (A,B) A
single first tier DEL blastomere at 90° longitude injected in the early blastula (A) populates both cardiac chambers (B). (C,D) A single
third tier DEL blastomere at 90° longitude injected in the midblastula (C) populates a single cardiac chamber (D), in this case, the
ventricle. m, blastoderm margin; A, atrium; V, ventricle; e, eye.
Zebrafish heart development
morphogenesis in the zebrafish embryo. Specifically, by
injection of single blastomeres with a lineage tracer dye, we
have analyzed the origin of cardiac progenitors and observed
the separation of the atrial and ventricular lineages during
early embryonic stages.
MATERIALS AND METHODS
Zebrafish embryos
Zebrafish were raised and handled as described in Westerfield
(1989). Developmental time at 28.5°C was determined from the
morphological features of the embryo as described by Kimmel
(1990): somitogenesis starts around 10 hours post-fertilization
(hpf), proceeds to reach 18 somites around 18 hpf and 26 somites
around 22 hpf. At 24 hpf, embryos have between 29 and 30
somites. For the lineage analysis, healthy embryos were selected,
placed in agarose-coated Petri dishes (one embryo per dish) and
dechorionated using fine watchmaker’s forceps in preparation for
blastomere injections at the early blastula (256/512-cell stage) or
midblastula (2000-cell/high stage) period.
33
stage of development, although the animal and vegetal poles are
easily distinguishable, it is not possible to determine the dorsalventral position, or longitude of a cell. Nevertheless, if the embryo
is allowed to free-fall from the Pasteur pipette, it is likely to land
with the lateral margin facing the surface, and the dorsal (0°) and
ventral (180°) poles positioned at the edges (Warga and Kimmel,
personal communication). We assumed that the blastomere closest
to the surface was positioned at either 90° or 270° longitude (see
Kimmel et al., 1990 for coordinate assignments), and subsequently
confirmed that this was true in more than 85% of the cases.
The blastomeres of the zebrafish embryo are of two types:
enveloping layer (EVL) and deep layer (DEL) cells. DEL cells are
round while EVL cells have a flattened shape (Kimmel et al.,
1990). Since EVL cells only give rise to periderm, a specialized
superficial epithelium, we took precautions to inject only DEL
cells. Impalement of a DEL cell was accompanied by a rapid
voltage change. The lineage tracer was expelled by controlling the
pressure from an Eppendorf Microinjector until the entire
cytoplasm was filled with dye. By simultaneously monitoring the
membrane potential, it was possible to assess the state of the cell
during dye injection and to determine if the microelectrode had
Blastomere injections
Glass microelectrodes were made from borosilicate capillaries (1.0
xx outer diameter, 0.75 mm inner diameter; WPI) with an internal
filament, which were pulled to a fine tip with a Brown-Flaming
puller. Microelectrodes were backfilled with 1-2 µl of tetramethylrhodamine-isothiocyanate dextran (or fluorescein dextran
for the double-label experiments) (Molecular Probes; 10×103 Mr;
5% in 0.2 M KCl) as the lineage tracer; the butt was then filled
with 0.2 M KCl so that the same microelectrode could be used for
dye injections and monitoring of membrane potentials.
Prior to dye injections, the embryo was transferred from its dish
to a depression slide using a fire-polished Pasteur pipette. At this
Table 1. Analysis of heart clones
Period of
injection
n
% with cells
in one chamber
% with cells
in both chambers
Early blastula
Midblastula
23
22
61
100
39
0
n=number of injected embryos with labeled myocardial cells.
256
512
1K
2K
mesoderm
atrium (and
other mes.)
ventricle (and
other mes.)
Fig. 3. Derived composite model for myocardial cell lineage. In
the early blastula (256-512-cell stage), a blastomere can contribute
progeny to both the atrium and the ventricle, while in the
midblastula, a blastomere contributes progeny either to the
ventricle or the atrium. As shown, these blastomeres also
contribute progeny to other mesodermal tissues (mes.) such as
head and trunk mesoderm (and perhaps also endodermal tissues;
see Warga and Kimmel, 1990).
Fig. 4. The migrating anterior lateral plate mesoderm gives rise to
2 tubular primordia. Transverse sections of a 15-somite stage
embryo at the level of the cardiac primordia. Embryos were
stained with HNK-1 to determine the location of the trigeminal
ganglia which approximates the anterior extent of the cardiac
primordia (data not shown). Bilateral tubular primordia (arrows)
are in direct contact with the yolk syncytial layer (y) ventrally.
Dorsally, a thin endodermal (e) layer spans the width of the
embryo. (A) Rostral section; (B) caudal section. Scale bar, 50 µm.
34
D. Y. R. Stainier, R. K. Lee and M. C. Fishman
Zebrafish heart development
Fig. 7. The zebrafish heart in the form of a cone. Wholemount
staining of a 24-somite stage embryo with CH1. (A) CH1 staining;
(B) Nomarski image. At this stage, the heart appears as a cone
with its base sitting on the yolk (Y). Dorsal is up and anterior is to
the right.
penetrated another cell. A series of injections at low pressure, as
opposed to a single high pressure injection, was found to be preferable for maintaining intracellular recordings and healthy cells. At
the end of injections, the microelectrode was rapidly withdrawn
from the cell. The embryo was briefly observed with epifluorescence; those in which more than one DEL cell was injected were
discarded from subsequent lineage analysis. The injected embryo
was returned to its Petri dish for continued development.
Lineage analysis
This analysis is schematized in Fig. 1. At 36 hours post-fertilization, the embryo was transferred to a depression slide for analysis
of the clonal progeny of the injected blastomere. The embryo is
usually stationary but, when necessary, it was anesthetized with
0.004% 3-amino benzoic acidethylester (Sigma). The embryo was
illuminated with transmitted light and epifluorescence optics. A
Silicon-Intensified-Target (SIT) video camera was used to amplify
the fluorescent signal.
For the lineage separation studies, we concentrated on the
myocardial progenitors located at 90° (and 270°) longitude. We
recorded (1) the frequency with which fluorescent cells were found
in the heart, (2) whether they resided in one or both chambers, and
35
Fig. 8. (A,B) Horizontal sections through the apex of the heart
cone. Tropomyosin immunoreactive myocardial cells (arrows)
surround the endocardial progenitor cells as the heart cone is
forming. Anterior is to the right. Scale bar, 50 µm.
(3) their number in each chamber. The mean (x) and standard error
(s.e.m.) were calculated from these numbers. Occasionally,
although fluorescent cells were clearly present in the heart, they
were too dim to be seen as discrete entities. In these cases, it was
not possible to quantify their number and these cases were
excluded from the statistical analysis.
Wholemount immunohistochemistry and histology
Embryos were staged according to somite number, then stained in
0.1% saponin as described in Stainier and Gilbert (1990). Monoclonal antibody (mAb) CH1 (anti-tropomyosin; Lin et al., 1985)
was obtained from the Developmental Studies Hybridoma Bank.
mAb CC8 is a mouse IgM that stains the notochord and the apical
surface of the floor plate (Stainier, Boulos and Fishman, unpublished data). Double-labeled embryos were stained with mAb’s
CH1 (IgG1) and CC8 (IgM) followed by fluoresceinated goat antimouse IgG1, and rhodaminated goat anti-mouse IgGM. Trigeminal ganglia were visualized by HNK-1 staining (Metcalfe et al.,
1990). After peroxidase staining with CH1 or HNK1, some of the
wholemounts were postfixed in 4% formaldehyde (EM grade;
Polysciences) and then embedded in JB4 (Polysciences) and
sectioned at 4 µm. Sections were counterstained with methylene
blue. Our wholemount staining protocol leads to the gentle washing
away of the yolk, which makes it easier to observe the heart, and
also to the detachment of the periderm (as can be observed in some
of the plastic sections).
36
D. Y. R. Stainier, R. K. Lee and M. C. Fishman
RESULTS
Lineage analysis
In the zebrafish, a fate map emerges in the late blastula, as
deep layer (DEL) cell lineages become restricted (Kimmel
et al., 1990). We wanted to determine whether cardiogenic
cells were concentrated in specific regions prior to this time,
and so injected single DEL cells with a lineage tracer dye
at, or near, the margin of the early blastula (256- and 512cell stage embryos) according to procedures developed by
Kimmel and Warga (1986). DEL cells located near the blastoderm margin develop mesodermal and endodermal fates,
whereas DEL cells near the animal pole develop ectodermal
fates (Kimmel et al., 1990).
By sampling DEL cells located near the margin, we found
that, in the early blastula, cardiac progenitors lie between
90° longitude (midway between the future dorsal and ventral
axis) and 180° longitude (the future ventral axis; see
methods). Blastomeres located around 90° longitude give
rise to myocardial cells but never to endocardial cells
whereas ventral blastomeres exhibit a more complex lineage
pattern and give rise to myocardial, endothelial, including
endocardial, and blood cells (Lee, Stainier and Fishman,
unpublished data). Dorsal blastomeres give rise to noncardiac mesoderm such as hatching gland and notochord, as
previously described by Kimmel et al. (1990). At 90°
longitude in the early blastula, a first tier DEL cell contributed to the myocardium in 53.5% of the injected embryos
(n=43), a third tier DEL cell in 13.6% (n=22), and a fifth
tier DEL cell in 4.8% (n=21). Higher tier blastomeres gave
rise to ectodermal tissues, as previously described by
Kimmel et al. (1990). To examine the question of bilateral
contribution to the heart, a 90° blastomere and a 270° blastomere were injected in the same embryo, one with
rhodamine dextran, the other with fluorescein dextran.
These gave doubly labeled hearts in 2 cases (n=18; data not
included). These results indicate that the zebrafish heart has
a bilateral origin, and that most cardiac progenitors lie near
the margin between 90° and 270° longitude.
To study the question of cardiac chamber lineage restriction, we elected to focus on the myocardial progenitors
located at 90° (and 270°) longitude. Single DEL blastomeres
from this region injected at the early blastula stage can contribute progeny to both major cardiac chambers. An example
is shown in Fig. 2A,B, in which a single blastomere injected
in the early blastula (Fig. 2A) contributed to both the atrium
and the ventricle (Fig. 2B). At 36 hours postfertilization
(hpf), the number of labeled cells observed in the beating
heart after injecting a single blastomere in the early blastula
varied from 2 to 22 cells (x=5.7, s.e.m.=1.3).
When we injected third tier DEL blastomeres at 90° (or
270°) longitude in the midblastula, 18% of the embryos
(n=101) had labeled myocardial cells, whereas a DEL blastomere located at the margin at 90° longitude gave rise to a
myocardial clone in only 4.3% of the injected embryos
(n=92). In contrast to the early blastula injections, individual blastomeres injected in the midblastula contributed only
to a single chamber. An example of this chamber restriction
is shown in Fig. 2C,D, in which a single blastomere injected
in the midblastula (Fig. 2C) contributed only to the ventricle
(Fig. 2D). A single blastomere of the midblastula contributed between 2 and 12 cells to the 36-hour heart (x=3.9,
s.e.m.=0.8).
The chamber distribution of heart clones after injection of
single blastomeres in the early and mid-blastula is documented in Table 1. Individual blastomeres injected in the
midblastula contributed to a single chamber in 100% of the
cases (n=22). Fig. 3 diagrams a derived composite model for
myocardial cell lineage: the atrial and ventricular lineages,
still mixed in the early blastula period, separate in the midblastula. Note that these blastomeres also contribute progeny
to other tissues. We did not examine carefully the specific
location of all the non-cardiac progeny, but our observations
agree with the analysis of Kimmel et al. (1990) that most of
them are mesodermal.
Heart tube formation
Cardiac progenitors are among the first mesodermal cells to
involute, sometimes even interspersed with involuting endodermal cells (Warga and Kimmel, 1990). After involuting,
they reverse their direction of movement, along with the
other early involuting cells, and turn towards an area about
30°-60° posterior to the animal pole (Stainier and Fishman,
1992). They begin to arrive at the embryonic axis around the
8-somite stage. During these 7 to 8 hours of postinvolution
migration, cardiac progenitor cells divide once or twice.
These cells coalesce at the embryonic axis, and by the 15somite stage, form two tubular primordia in the lateral plate
mesoderm (Fig. 4, arrows). At this stage, the bilateral
cardiac primordia extend from the caudal end of the trigeminal ganglia, rostrally, to the otic vesicles, caudally (data not
included). Cardiac progenitors at the ventral side of the tubes
appear to be migrating directly on the syncytial layer of the
yolk cell (y). A thin layer of endoderm (e) sits dorsomedially to the cardiac primordia.
The marker we have found to be the first to reveal the
cardiac primordia is tropomyosin. mAb CH1 (antitropomyosin) staining outlines this structure starting at the
18-somite stage (Stainier and Fishman, 1992, 1993). Fig. 5
shows a 20-somite stage embryo double labeled with monoclonal antibodies CH1 (in green) and CC8 (in red), the
latter primarily labels the notochord. The tubular cardiac
primordia (arrowheads) as well as the somitic muscles are
CH1-positive. As observed in the wholemount, and
confirmed by serial sectioning, the caudal extent of the
cardiac primordia (identified by CH1 immunoreactivity)
corresponds closely to the rostral extent of the notochord
(although CH1-negative bilateral tubular structures extend
beyond this point).
At the 21-somite stage, transverse sections show that the
tubular cardiac primordia, which now form complete tubes,
have moved closer to each other (Fig. 6). This is more pronounced medially (Fig. 6B), a region where a group of CH1negative cells sits between the CH1-positive cardiac tubes.
Swaen and Brachet (1901) have studied this group of cells
in the trout embryo and named it the portion moyenne du
mésoblaste and have suggested that it gives rise to the endocardium (see also Senior, 1909). We have followed these
cells on serial sections of zebrafish embryos at the 21-, 24and 27-somite stage (see below) and confirmed that they
give rise to the endocardium.
Zebrafish heart development
The fusion of the two tubular cardiac primordia seen at
the 21-somite stage (Fig. 6) leads to the formation of the
definitive heart tube by 24 hpf. The wholemount staining of
intermediate stage embryos with CH1 reveals that the intermediate stage heart is shaped like a cone, with its base sitting
on the yolk (Fig. 7; see also Senior (1909) for a description
of the early teleost heart as a cone). The apex of the cone,
the future arterial end (see Fig. 5C,D in Stainier and Fishman
(1992) for the staining of a 26-somite stage embryo with
chamber specific mAbs) is continuous with the ventral aorta.
The base of the cone, the future venous end, is attached to
the visceral pericardium. Fig. 8 shows horizontal sections
through the apex of the cone as it is forming. Tropomyosin
immunoreactive myocardial cells (arrows) surround the
endocardial progenitor cells. Fig. 9 shows a rostrocaudal
sequence of transverse sections of the heart cone at the 27somite stage. Fig. 9A is at the anterior portion of the base
of the cone, which consists of a continuous and single sheet
of myocardial cells. Fig. 9B-E shows sections through the
apex of the cone. Fig. 9D is in the middle of the apex of the
cone showing the endocardial cells, and Fig. 9B,C,E shows
sections through the walls of the apex of the cone. The endocardial cells (arrows) appear to migrate from the apex of the
heart cone in both anterior and posterior directions, and by
this stage have reached the cone’s posterior extremity (Fig.
9F) but not its anterior one (Fig. 9A).
Chamber development
This study focused primarily on the earliest stages of heart
tube formation. Subsequent developmental milestones
resemble those of chick heart development (DeHaan, 1965).
For the purpose of future genetic analyses, we were interested in determining the exact timing of these events in the
wild-type zebrafish embryo. Individual myocardial cells
begin to contract irregularly around the 22-somite stage (as
visualized during lineage tracking) and the heart beats in a
coordinated fashion, at a rate of about 25 beats/minute, by
the 26-somite stage. By 24 hpf, the zebrafish heart tube beats
at about 90 contractions/minute, and, shortly thereafter, circulation begins. Morphological differentiation progresses in
an arterial to venous direction (Stainier and Fishman, 1992),
and, as in the chick, there is a corresponding shift in the
location of the pacemaker to the venous end, and an acceleration in beat rate accompanied by an increase in contractility. There is, at this stage, no external indication of separation into bulbus arteriosus, ventricle, atrium and sinus
venosus, although the regionalization of cardiac myosin
heavy chain (MHC) expression defines chamber boundaries
by the 26-somite stage (see Fig. 5C,D in Stainier and
Fishman, 1992). An early regionalization of cardiac MHC
expression has also been shown in the chick embryo (Zhang
et al., 1986; Sweeney et al., 1987).
By 30 hpf, the heart tube is beating at about 140
beats/minute, and has started to loop to the embryo’s righthand side. This direction of looping of the heart tube is a
feature that has been conserved through all of vertebrate
evolution from the cyclostomes (Santer, 1985). By 36 hpf,
the heart tube has looped. It is beating at about 180
beats/minute, and provides a strong circulation to the trunk
and head. Clear indentations mark chamber boundaries, and
looping leaves the atrium sitting on the left and brings the
37
ventricle to the right side of the embryo. Two minor
chambers, the sinus venosus and the bulbus arteriosus can
also be distinguished at the extremities of the heart tube
(Stainier and Fishman, 1992).
DISCUSSION
The zebrafish heart may provide an ideal system for
combining embryological and genetic approaches to the
complex processes of vertebrate organogenesis. The
formation of bilateral cardiac tubes and their subsequent
fusion, the lining of the myocardium by the endocardium,
the morphological differentiation of the heart in an arterial
to venous direction, paralleled by an acceleration of the heart
beat, all contribute to the formation and proper functioning
of the heart, and all can be readily assayed in the zebrafish
embryo by visual inspection. The zebrafish provides distinct
advantages for studying heart development because of its
accessible embryology and its genetics. Because early fish
heart development has not been well characterized, we
undertook this study to provide the histological framework
for further embryological and genetic manipulations.
Lineage analysis
Previous work in the chick and in the amphibian has mapped
the general regions with cardiogenic potential during
embryogenesis. In the gastrulating chick embryo, bilateral
regions of cardiac progenitors are located in the mesodermal layer just lateral to the tip of the primitive streak
(reviewed by Litvin et al., 1992). A vital dye fate mapping
of the Xenopus embryo at the onset of gastrulation indicates
that the two patches of heart mesoderm are located within
the dorsolateral regions of the deep mesoderm, to either side
of the prospective head mesoderm (Keller, 1976; see also
Sater and Jacobson (1990) for the examination of heart
forming potency in Xenopus embryos).
One advantage of the transparent zebrafish embryo is that
this process can be followed at the single cell level, and
lineages can be established starting with a single progenitor
(Warga and Kimmel, 1990). Our single blastomere injections in the zebrafish show that in the early blastula, most
cardiac progenitors lie in a marginal zone that extends from
90° to 270° longitude.
We found that, in the early blastula, individual myocardial progenitors located at the margin around 90° longitude
contributed cells to both cardiac chambers. When injected
in the midblastula, individual progenitors from 90°
longitude contributed cells only to a single cardiac chamber.
This suggests that, at least for these myocardial progenitors,
the atrial and ventricular lineage separate during the midblastula (as diagrammed in Fig. 3). Previously, we found
that low concentrations of retinoic acid applied as early as
the onset of gastrulation truncate the zebrafish heart tube
from its arterial end to its venous end without visibly
affecting the rest of the embryo (Stainier and Fishman,
1992). Further analysis of this effect suggested that cardiac
progenitors acquire their axial coordinates during early gastrulation, a time when DEL cell lineages are known to
become tissue-restricted (Kimmel et al., 1990). The
observed lineage separation between the atrium and
38
D. Y. R. Stainier, R. K. Lee and M. C. Fishman
Fig. 9. Transverse sections through the heart cone. Rostrocaudal sequence of transverse sections of a 27-somite stage embryo stained with
CH1. (A) The anterior portion of the base of the cone where the originally tubular primordia now form a continuous and single sheet of
CH1-positive cells, which lies basally, adjacent to the yolk syncytial layer (arrowhead). (B-E) The apex of the cone. (B,C,E) Sections
through the walls of the apex of the cone. (D) Medial section through the apex of the cone showing the CH1-negative endocardial cells
(arrow). The endocardial cells (arrows, B-F) have migrated further inside the heart cone and by this stage have reached the cone’s
posterior extremity (F) but not its anterior one (A). Scale bar, 50 µm.
ventricle in the midblastula, reported in this study, suggests
that myocardial progenitors at 90° longitude may acquire
positional information at an even earlier stage. In this model,
by the time that cells have become restricted to the cardiac
lineage at the onset of gastrulation, they may already bear
positional information restricting them to the atrium or
ventricle. Alternatively, the observed lineage separation
may be explained by limited spatial rearrangements of the
cardiac progenitors as they migrate after the midblastula
period, and/or by a limited number of divisions of these cells
between the midblastula and gastrula periods. Time-lapse
observation of labeled myocardial progenitors may help us
analyze these issues further.
Heart tube formation
The cardiac progenitors migrate towards the embryonic axis
as part of the anterior lateral plate mesoderm. The rostrocaudal extent of the cardiac primordia within the lateral plate
mesoderm can be defined by the 18-somite stage, a time
when cardiac progenitors start expressing tropomyosin. The
mechanisms by which the anteroposterior extent of the
cardiac primordia is demarcated are unclear. It is interesting
that, in the zebrafish embryo, the posterior extent of the
cardiac primordia closely corresponds to the anterior extent
of the notochord (Fig. 5). This suggests the intriguing possibility that the notochord, which plays an important
inductive role in patterning the nervous system (Watterson,
1965; Yamada et al., 1991), may also act to set the posterior
limit of the cardiac primordia. Previous explant and extirpation experiments have suggested that presumptive neural
tissue inhibits heart formation (Jacobson, 1960), but the
influence of the notochord has not been investigated.
We have characterized heart tube formation in the
zebrafish embryo as a prelude to future embryological and
genetic manipulations. The zebrafish heart is generated from
a pair of tubular primordia. At the 21-somite stage, all cells
forming the tubular primordia are immunoreactive for
tropomyosin, suggesting that they will contribute to the
myocardium (Fig. 6). We have not addressed the dynamic
cellular changes by which these two complete tubes fuse to
generate a single myocardial tube, and by which a second
endocardial tube comes to be enclosed within it. One
potential model for this progression is shown diagramatically in Fig. 10. The bilateral cardiac primordia fuse
Zebrafish heart development
39
heart development and should help refine this model. Also,
the isolation and careful characterization of mutations
affecting this process should clarify it.
The formation of the definitive heart tube from bilateral
cardiac primordia in the zebrafish embryo is essentially
similar to that reported in other species, including the chick
(DeHaan, 1965; Virágh et al., 1989), and mouse (DeRuiter
et al., 1992). The cardiac primordia, complete tubes in the
zebrafish, incomplete tubes in the chick and mouse, fuse to
form the definitive heart tube. The details of the process by
which the endocardial cells come to line the inside of the
definitive heart tube remain to be explored in any species.
Myosin heavy chain expression delineates the cardiac
chambers by 22 hpf, which corresponds to the initiation of
the heart beat but precedes obvious chamber formation
(Stainier and Fishman, 1992). As the heart starts looping at
30 hpf, distinctive chambers become evident, demarcated by
narrower rings of tissue at their junction, and each chamber
starts beating as a synchronous unit, in veno-arterial
sequence. The looping of the heart tube to the embryo’s right
hand side is a conserved feature within vertebrates and it is
likely to be under tight genetic control. Indeed, a mouse
mutation, iv, randomizes the direction of looping (Hummel
and Chapman, 1959).
In a preliminary screen for zebrafish heart mutants, we
have recovered a number of interesting mutations, including
a spontaneous mutation that affects the proper development
of the endocardium (Stainier and Fishman, unpublished
data). We expect that the characterization of this and other
mutations affecting heart development, combined with the
unique embryological advantages of the zebrafish, will lead
to a better understanding of the development of cardiac form
and function.
Fig. 10. Model for the formation of the definitive heart tube from
two bilateral cardiac tubes. At the 21-somite stage (top), the
bilateral tubular primordia form complete tubes that sit on either
side of the endocardial progenitor cells (in red). By the 27-somite
stage (middle), a cone is formed, with its (venous) base sitting on
the yolk. Medially, the myocardial cells surrounding the
endocardial cells form the apex of the cone. As the head lifts
dorsally, the cone is rotated so that by 24 hpf (bottom), its
(venous) base sits anteriorly. At this stage, an outer myocardial
layer and an inner endocardial layer form the definitive heart tube.
between the 21-somite stage and the 27-somite stage. At
their extremities, the originally tubular structures fuse to
form a single cellular sheet. Around the portion moyenne
(the endocardial progenitor cells, in red), the tubular structures fuse to form the apex of the cone. As the head lifts
dorsally, the heart cone rotates so that its (venous) base sits
anteriorly by 24 hpf. High resolution time-lapse videomicroscopy of labeled myocardial progenitors may help
establish their exact positions during the different stages of
We thank Chris Simpson for expert technical help with the
histology, Colleen Boggs for zebrafish care and supervision,
Rachel Warga for showing us how to inject single blastomeres for
lineage analysis and pointing us to 90° longitude, Wolfgang
Driever, Megan Grether, Alexander Schier, Lila Solnica-Krezel,
and Brant Weinstein for comments on the manuscript. This work
is supported by a grant from the Helen Hay Whitney Foundation
(to D. Y. R. S.) and by NIH RO1 grant HL49579 (to M. C. F.) and
a sponsored research agreement from Bristol-Myers Squibb (to M.
C. F.).
REFERENCES
DeHaan, R. L. (1963). Migration patterns of the precardiac mesoderm in
the early chick embryo. Exp. Cell Res. 29, 544-560.
DeHaan, R. L. (1965). Morphogenesis of the vertebrate heart. In
Organogenesis (ed. R. L. DeHaan and H. Ursprung), pp. 377-420. New
York: Holt, Rinehart and Winston.
DeRuiter, M. C., Poelmann, R. E., VanderPlas-de Vries, I., Mentink, M.
M. T. and Gittenberger-de Groot, A. C. (1992). The development of the
myocardium and endocardium in mouse embryos. Anat. Embryol. 185,
461-473.
Gonzalez-Sanchez, A. and Bader, D. (1990). In vitro analysis of cardiac
progenitor cell differentiation. Dev. Biol. 139, 197-209.
Greenwald, I. and Rubin, G. M. (1992). Making a difference: the role of
cell-cell interactions in establishing separate identities for equivalent
cells. Cell 68, 271-281.
Horvitz, H. R. and Sternberg, P. W. (1991). Multiple intercellular
40
D. Y. R. Stainier, R. K. Lee and M. C. Fishman
signalling systems control the development of the Caenorhabditis elegans
vulva. Nature 351, 535-541.
Hummel, K. P. and Chapman, D. B. (1959). Visceral inversion and
associated anomalies in the mouse. J. Hered. 50, 9-13.
Jacobson, A. G. (1960). The influences of ectoderm and endoderm on heart
differentiation in the newt. Dev. Biol. 2, 138-154.
Jacobson, A. G. (1961). Heart determination in the newt. J. Exp. Zool. 146,
139-151.
Keller, R. E. (1976). Vital dye mapping of the gastrula and neurula of
Xenopus laevis. I. Prospective areas and morphogenetic movements in the
deep region. Dev. Biol. 51, 118-137.
Kimmel, C. B. (1989). Genetics and early development of zebrafish. Trends
Genet. 5, 283-288.
Kimmel, C. B. (1990). Stages of Early Development of the Zebrafish.
Oregon, USA: University of Oregon Press.
Kimmel, C. B. and Warga, R. M. (1986). Tissue-specific cell lineages
originate in the gastrula of the zebrafish. Science 231, 365-368.
Kimmel, C. B., Warga, R. M. and Schilling, T. F. (1990). Origin and
organization of the zebrafish fate map. Development 108, 581-594.
Lin, J. J. C., Chou, C. S. and Lin, J. L. C. (1985). Monoclonal antibodies
against chicken tropomyosin isoforms: production, characterization, and
application. Hybridoma 4, 223-242.
Litvin, J., Montgomery, M., Gonzalez-Sanchez, A., Bisaha, J. G. and
Bader, D. (1992). Commitment and differentiation of cardiac myocytes.
Trends Cardiovasc. Med. 2, 27-32.
Metcalfe, W. K., Myers, P. Z., Trevarrow, B., Bass, M. B. and Kimmel,
C. B. (1990). Primary neurons that express the L2/HNK-1 carbohydrate
during early development in the zebrafish. Development 110, 491-504.
Rawles, M. E. (1936). A study in the localization of organ forming areas of
the chick blastoderm of the head process stage. J. Exp. Zool. 72, 271-315.
Rosenquist, G. C. (1966). A radioautographic study of labeled grafts in the
chick blastoderm. Carnegie Inst. Wash. Publ. Contrib. Embryol. 625, 71110.
Santer, R. M. (1985). Morphology and innervation of the fish heart. Adv.
Anat. Embryol. Cell Biol. 89, 1-102.
Sater, A. K. and Jacobson, A. G. (1990). The restriction of the heart
morphogenetic field in Xenopus laevis. Dev. Biol. 140, 328-336.
Senior, H. D. (1909). The development of the heart in shad. With a note on
the classification of Teleostean embryos from a morphological
standpoint. Am. J. Anat. 9, 212-276.
Stainier, D. Y. R. and Gilbert, W. (1990). Pioneer neurons in the mouse
trigeminal sensory system. Proc. Natn. Acad. Sci. USA 87, 923-927.
Stainier, D. Y. R. and Fishman, M. C. (1992). Patterning the zebrafish
heart tube: acquisition of anteroposterior polarity. Dev. Biol. 153, 91-101.
Stainier, D. Y. R. and Fishman, M. C. (1993). Cardiac morphogenesis in
the zebrafish, patterning the heart tube along the anteroposterior axis. In:
Molecular Basis of Morphogenesis, 51st Annual Symposium of the
Society for Developmental Biology (ed. M. Bernfield), pp. 77-89. New
York: John Wiley & Sons, Wiley-Liss Division.
Swaen, A. and Brachet, A. (1901). Etude sur les premières phases du
développement des organes dérivés du mésoblaste chez les poissons
Téléostéens. Arch. Biol. Liege et Paris 18, 73-190.
Sweeney, L. J., Zak, R. and Manasek, F. J. (1987). Transitions in cardiac
isomyosin expression during differentiation of the embryonic chick heart.
Circ. Res. 61, 287-295.
Virágh, S., Szabo, E. and Challice, C. E. (1989). Formation of the
primitive myo- and endocardial tubes in the chicken embryo. J. Mol. Cell.
Cardiol. 21, 123-137.
Warga, R. M. and Kimmel, C. B. (1990). Cell movements during epiboly
and gastrulation in zebrafish. Development 108, 569-580.
Watterson, R. L. (1965). Structure and mitotic behavior of the early neural
tube. In Organogenesis (ed. R. L. DeHaan and H. Ursprung), pp. 377-420.
New York: Holt, Rinehart and Winston.
Westerfield, M. (1989). The Zebrafish Book. University of Oregon press.
Wilens, S. (1955). The migration of heart mesoderm and associated areas in
Amblystoma punctatum. J. Exp. Zool. 129, 576-605.
Yamada, T., Placzek, M., Tanaka, H., Dodd, J. and Jessell, T. M. (1991).
Control of cell pattern in the developing nervous system: polarizing
activity of the floor plate and notochord. Cell 64, 635-647.
Zhang, Y., Shafiq, S. A. and Bader, A. (1986). Detection of a ventricularspecific myosin heavy chain in adult and developing chicken heart. J.
Cell. Biol. 102, 1480-1484.
(Accepted 25 May 1993)