Mechanisms of Development 122 (2005) 157–173
www.elsevier.com/locate/modo
Intestinal growth and differentiation in zebrafish
Kenneth N. Wallacea,1, Shafinaz Akhtera, Erin M. Smitha, Kristin Lorenta, Michael Packa,b,*
a
Department of Medicine, University of Pennsylvania School of Medicine, Rm 1212, BRB 2/3, 421 Curie Blvd., Philadelphia, PA 19104, USA
b
Department of Cell and Developmental Biology, University of Pennsylvania School of Medicine, Rm 1212,
BRB 2/3, 421 Curie Blvd., Philadelphia, PA 19104, USA
Received 23 July 2004; received in revised form 13 October 2004; accepted 15 October 2004
Available online 10 November 2004
Abstract
Intestinal development in amniotes is driven by interactions between progenitor cells derived from the three primary germ layers. Genetic
analyses and gene targeting experiments in zebrafish offer a novel approach to dissect such interactions at a molecular level. Here we show
that intestinal anatomy and architecture in zebrafish closely resembles the anatomy and architecture of the mammalian small intestine. The
zebrafish intestine is regionalized and the various segments can be identified by epithelial markers whose expression is already segregated at
the onset of intestinal differentiation. Differentiation of cells derived from the three primary germ layers begins more or less
contemporaneously, and is preceded by a stage in which there is rapid cell proliferation and maturation of epithelial cell polarization.
Analysis of zebrafish mutants with altered epithelial survival reveals that seemingly related single gene defects have different effects on
epithelial differentiation and smooth muscle and enteric nervous system development.
q 2004 Elsevier Ireland Ltd. All rights reserved.
Keywords: Zebrafish; Intestine; Development; Differentiation
1. Introduction
The intestinal tract of higher vertebrates (amniotes) is
comprised of two contiguous organs, the small intestine and
large intestine. The small intestine is the principal site of
nutrient absorption whereas the primary function of the
large intestine is water and salt absorption. Each organ also
plays an important role in immunity and functions as a
barrier to pathogens and other environmental toxins.
Intestinal function is dependent upon the interactions of
cells derived from the three embryonic germ layers.
Understanding how these cellular interactions are established is an important question of developmental biology
that is also relevant to biomedical research.
The cellular anatomy of the small and large intestine is
organized in a nearly identical manner. Each organ is lined
by a simple epithelium that is surrounded by connective
* Corresponding author. Tel.: C1 215 573 4145; fax: C1 215 898 9841.
E-mail address: [email protected] (M. Pack).
1
Present address: Department of Biology, Clarkson University, Postdam,
NY, USA.
0925-4773/$ - see front matter q 2004 Elsevier Ireland Ltd. All rights reserved.
doi:10.1016/j.mod.2004.10.009
tissue, smooth muscle and enteric nerves. Together these
cells are arranged in four distinct concentric layers; the
epithelium, subjacent connective tissue and thin layer of
smooth muscle (muscularis mucosa) form the innermost
layer, the mucosa. The layer beneath the mucosa, the
submucosa, contains blood vessels, lymphatics, and plexes
of enteric nerves, which innervate the epithelium, within a
thick layer of supporting connective tissue. Surrounding the
submucosa are the inner circular and outer longitudinal
smooth muscle layers of the muscularis. Enteric neurons
within plexes between the two muscle layers regulate
intestinal motility. Surrounding the muscularis in most
intestinal regions is an outermost layer, the serosa.
Digestive physiology is a function of the coordinated
interactions of cells within each of the intestinal layers.
Interactions between the endodermal derived epithelium of
the primitive gut tube and its surrounding mesenchyme
(derived from lateral plate mesoderm) drive intestinal
differentiation and morphogenesis (de Santa Barbara, 2003;
Kaestner et al., 1997; Karlsson et al., 2000; Kedinger et al.,
1998; Pabst et al., 1999). Mesenchymal signals also regulate
the differentiation of enteric neurons (Gershon, 1999;
158
K.N. Wallace et al. / Mechanisms of Development 122 (2005) 157–173
Gianino et al., 2003; Nosrat et al., 1996; Shin et al., 1999;
Widenfalk et al., 1997). The intestinal anlage is patterned by
signals exchanged between endoderm and surrounding
mesenchyme and this patterning drives region-specific
differentiation. Roles in gut patterning and subsequent
intestinal epithelial and smooth muscle differentiation have
been identified for Hox genes (Kapur et al., 2004; Roberts et
al., 1998), the BMP (Chalazonitis et al., 2004; Roberts et al.,
1998), Notch (Jensen et al., 2000), Wnt (Korinek et al., 1998;
van den Brink et al., 2004) and Hedgehog (Ramalho-Santos
et al., 2000; Roberts et al., 1998) signaling pathways.
Importantly, these pathways appear to function in the
juvenile and adult intestine and their regulation may be
altered in intestinal diseases (Amiel and Lyonnet, 2001;
Fodde, 2002; Haramis et al., 2004; Kuhnert et al., 2004).
Although most of our knowledge of vertebrate intestinal
development derives from embryological and gene targeting
studies performed in rodent and avian model systems,
forward genetic analyses, such as can be performed with
zebrafish, offer a complementary approach to this topic.
Mutagenesis screens in zebrafish have identified mutations
that alter differentiation and maintenance of intestinal
epithelia at embryonic or larval stages (Chen et al., 1996;
Mayer and Fishman, 2003; Pack et al., 1996). Molecular
characterization of mutants discovered in these and
subsequent screens are expected to identify genetic pathways that regulate intestinal development and physiology.
Such work will complement developmental studies performed in other model systems and may lead to the
characterization of common pathways that regulate intestinal organogenesis in a large number of vertebrate species.
Integration of molecular pathways that direct zebrafish
intestinal development into such a framework requires a
comparative understanding of intestinal development and
anatomy in zebrafish and other vertebrate organisms. In
previous work, we described formation of the zebrafish
digestive tract and accessory organs (Wallace and Pack,
2003). Here we define subsequent stages of intestinal
development as well as intestinal anatomy in adult
zebrafish. We also show that mutations that alter intestinal
epithelial survival have variable effects on epithelial,
smooth muscle and enteric nervous system development.
Through these studies, we define a simple two-stage model
of intestinal development. The first is characterized by rapid
expansion of the intestinal anlage coupled with maturation
of the polarized epithelial phenotype. Smooth muscle and
enteric nerve progenitors populate the intestine at this stage.
The second stage is characterized by epithelial, smooth
muscle and enteric nervous system differentiation along
with morphogenesis of the folded epithelium. Mutant
analyses show that single gene defects can differentially
affect these latter processes. Molecular characterization of
such mutants is predicted to identify genetic pathways that
direct tissue interactions in the developing and adult
intestine.
2. Results
2.1. Anatomy and architecture of the zebrafish
adult intestine
Previous studies have shown that intestinal anatomy and
architecture in cyprinid teleost fish is closely related to
mammals (Curry, 1937; Rogick, 1931). We performed
histological and immunohistochemical analyses to determine whether such features are conserved in zebrafish.
2.2. Gross anatomy
The adult zebrafish intestine is a folded tube that
occupies the majority of the abdominal cavity. Like other
cyprinids, zebrafish are stomachless. The anterior intestine,
often referred to as the intestinal bulb, has a wider caliber
than the lumen of the posterior intestine and thus may
function as a reservoir. Illumination of the intestinal wall,
which is transparent, reveals the presence of large randomly
shaped epithelial folds (Fig. 1A). Proportionally, these folds
are significantly larger than the finger-like intestinal villi of
mammals and other amniotes. Many folds are oriented
circumferentially, but a significant percentage of folds are
randomly organized. Fold height is smaller in the mid vs.
anterior intestine (Fig. 1B,C). The shortest folds that are
oriented longitudinally define the posterior intestinal
segment (Fig. 1D).
2.3. Epithelial architecture
Histological sections through the anterior, mid and
posterior intestinal segments reveal that the folds are
comprised of a single layer of epithelium that rests upon a
connective tissue core similar to the lamina propria of the
mammalian small intestine (Fig. 1B–D). The epithelium at
the base of the folds varies in width and lacks crypts of
Lieberkuhn (arrow Fig. 1B). Laminin immunohistochemistry identifies a prominent basement membrane beneath the
epithelium and at the base of the folds (Fig. 1G). Three of
the four principal cell types within the mammalian small
intestinal epithelium are present within the anterior zebrafish intestine. Columnar-shaped absorptive enterocytes that
we identify by the presence of apical sodium phosphate cotransporter protein (NaPi) (Graham et al., 2003) are the most
numerous (Fig. 1H). Goblet cells are the second most
populous epithelial cell type. They are identified by their
large apical mucin containing theca, that also bind
fluorescently labeled wheat germ agglutinin lectin (not
shown) (Fig. 1C,F). Enteroendocrine cells, the third
epithelial cell type, are identified by cytoplasmic pancreatic
polypeptide hormone, as previously described (Langer
et al., 1979) (Fig. 1I). Based on the distribution of pancreatic
polypeptideCcells, we rarely identify enteroendocrine cells
beyond the anterior zebrafish intestine. By contrast,
goblet cells are identified in all intestinal regions, whereas
K.N. Wallace et al. / Mechanisms of Development 122 (2005) 157–173
159
Fig. 1. Adult zebrafish intestine: (A) Whole-mount, interior view of the adult zebrafish anterior intestine reveals irregularly shaped epithelial folds. The long
axis of most folds is oriented circumferentially; others are oriented randomly (double arrow defines anterior–posterior axis). (B–D) Histological cross-section
through the folds of the anterior (B), mid (C) and posterior (D) intestine. Note progressive diminution of the height of mid and posterior folds. A connective
tissue core containing blood vessels and nerves underlies the epithelium of each fold. Note absence of crypts at base of folds (arrow in B). (E,F) Enlargement of
region adjacent to arrowheads in (B,C) showing columnar-shaped enterocytes and goblet cells. Arrow in (E) points to an erythrocyte in a blood vessel in the
connective tissue of the intestinal fold (G) Laminin immunohistochemistry identifies the basement membrane separating the epithelium from the underlying
connective tissue (lamina propria). Laminin also encircles blood vessels within the lamina propria (arrow). Four principal cell types are identified within the
epithelium. (H) Enterocytes are identified by apical sodium phosphate transporter (NaPi). (I) Enteroendocrine cells are identified by pancreatic polypeptide.
(C,F) Goblet cells are identified histologically by their mucinous theca (arrow in C). (J,L) Antigen presenting cells within segment II of the mid-intestine are
identified by their distinctive supra-nuclear vacuole. (K,M) Desmin (green) identifies inner circular (arrow) and outer longitudinal (arrowhead) smooth muscle
layers beneath the lamina propria. (N) Acetylated tubulin identifies axonal fibers within the muscularis and the lamina propria. (O) Hu immunohistochemistry
identifies a solitary enteric neuron cell body within the muscularis (arrow). (P) Schematic representing the anterior (red), mid (green) and posterior (blue)
intestinal segments. Stippled region of the mid segment represents location of specialized enterocytes. (Q–S) Intestinal epithelial cell renewal revealed by BrdU
immunohistochemistry. S-phase cells are present at the base of the intestinal folds 1 h after BrdU injection (Q). 4-day post-injection, S-phase cells are identified
at the base and in the mid fold region (R). 7-day post-injection, S-phase cells are only identified at the tip of the folds (S). (T) TUNEL assay identifies apoptotic
cells at the tips of intestinal folds.
NaPiCenterocytes do not extend beyond the mid intestine.
As reported for other cyprinid fish, Paneth cells were not
identified histologically in any intestinal segment.
Histological analyses revealed the presence of a fourth
epithelial cell type within a posterior segment of the
zebrafish mid intestine. These NaPiC enterocytes contain a
prominent supranuclear vacuole in which pinocytosed
luminal contents can be stored (Fig. 1J,L). They are
believed to function as antigen presenting cells and may
be analogous to the M-cells of the adult mammalian
intestine (Gebert et al., 1996; Jang et al., 2004; Rombout
et al., 1985). These specialized enterocytes can also be
identified histochemically in fish that have ingested
horseradish peroxidase protein (discussed below). Goblet
cells are interspersed between adjacent specialized enterocytes within the posterior midgut. Histological sections
reveal that nearly all of the NaPiC cells within the posterior
segment of the mid intestine are specialized enterocytes.
160
K.N. Wallace et al. / Mechanisms of Development 122 (2005) 157–173
These cells are largely restricted from the base of the
intestinal folds as are enterocytes, goblet cells and
enteroendocrine cells in other intestinal regions (Fig. 1B,
C,H,I). Fig. 1P shows the relative positions of the anterior,
mid and posterior regions of the adult intestine as viewed
ventrally.
2.4. Zebrafish intestinal epithelial cell renewal
Previous studies have described epithelial cell renewal
within the intestine of adult and larval cyprinid fish
(Rombout et al., 1984; Stroband and Debets, 1978). These
studies identified a compartment of proliferating epithelial
cells at the base of the intestinal folds. We defined this
compartment in zebrafish and recorded the rate at which
cells traverse the epithelial folds using BrdU immunohistochemistry. Detection of BrdU 1 h following pulselabeling identified proliferative cells at the base of the
intestinal folds (Fig. 1Q). BrdU detection at 24-h intervals
showed that epithelial cells exit the base of the folds and
migrate luminally (Fig. 1P–R). Labeled cells reach the
apical region of anterior intestinal folds within 5–7 days,
whereas it took 7–10 days to traverse the folds of the mid
intestine. Using PCNA immunohistochemistry (not shown),
we found that the proliferative compartment comprised
approximately 20% of intestinal epithelial cells, as
described previously for larval zebrafish (Rawls et al.,
2004). As expected, the TUNEL assay identified apoptotic
cells at the tips of many intestinal folds (Fig. 1T).
2.5. Smooth muscle, connective tissue and enteric nervous
system architecture
Histological and immunohistochemical analyses
revealed that the overall organization of the enteric nervous
system and the anatomy of connective tissue and smooth
muscle layers within the zebrafish intestine are comparable
to mammals. However, several differences were identified.
First, the lamina propria separating the epithelium from the
underlying smooth muscle layers (muscularis) in zebrafish
is less complex than the corresponding mammalian
connective tissue layers (lamina propria and submucosa).
Further, a smooth muscle sublayer comparable to the
mammalian muscularis mucosa is not identified in zebrafish.
Instead, the epithelium and lamina propria are only
surrounded by circular and longitudinal smooth muscle
layers (Fig. 1J,K).
A second distinctive feature of the zebrafish intestine is
the location of enteric ganglia. In mammals, they are located
in the submucosa and between the circular and longitudinal
smooth muscle layers of the muscularis. Using a panneuronal anti-Hu antibody (Kelsh and Eisen, 2000;
Marusich et al., 1994), we identified only individual or
pairs of enteric nerve cell bodies between the circular and
longitudinal smooth muscle layers of the adult intestine
but not within the adjacent connective tissue (Fig. 1M).
As expected, an anti-tubulin antibody revealed nerve fibers
within this region and in the connective tissue layer beneath
and within the epithelial folds (Fig. 1K). This anatomical
arrangement of enteric nerves appeared similar in all
intestinal segments we analyzed although the density of
HuCcell bodies was greatest in the posterior intestine.
A schematic comparing the architecture of the adult
zebrafish and mammalian intestine is shown in Fig. 2.
2.6. Intestinal development
2.6.1. Intestinal growth and proliferation
The zebrafish digestive tract forms in a segmental
fashion beginning at mid somite stages (Wallace and
Pack, 2003). The intestinal anlage, also referred to as the
gut tube, forms first, followed by the pharynx and
esophagus, which arise in situ within endoderm anterior to
the gut. The gut tube is comprised of polarized epithelial
cells that express apical alkaline phosphatase and beta-actin,
and basolateral cadherin, but lack digestive enzymes and
other components of the apical junctional complex.
Between completion of gut tube morphogenesis (34 h
post-fertilization (hpf)) and the onset of exogenous feeding
(120 hpf), there is a dramatic increase in the size of the
intestine and the appearance of the epithelial cells in
histological sections (Fig. 3A–I). Throughout this period,
the epithelium is comprised of a single layer of cells that is
never stratified. We charted proliferative rates within the
epithelium as a first step in understanding the molecular
signals that regulate intestinal growth and fold morphogenesis in zebrafish. We also correlated this pattern with
maturation of the polarized epithelial phenotype.
Proliferating epithelial cells were identified using BrdU
immunohistochemistry. Dissected intestines were analyzed
as whole-mounts using a confocal microscope, or histologically. The 1-h labeling data revealed that the gut epithelium
is highly proliferative through 74 hpf, but that this rate drops
precipitously between 74 hpf and 120 hpf, the onset of
exogenous feeding (Fig. 3J). At all time points analyzed,
proliferative rates were uniform within anterior, mid and
posterior segments of the developing intestine. With these
experiments, we found that proliferating cells appeared to
be restricted to the base of the folds. However, the relatively
small percentage of cells in S-phase in 5-day postfertilization (dpf) larvae with evident folds complicated
this interpretation.
Gut microbiota in zebrafish and mammals have been
shown to influence various aspects of intestinal development (Bry et al., 1996; Rawls et al., 2004; Uribe et al.,
1997). Zebrafish larvae reared in the presence of normal
commensal bacteria have a four-fold higher rate of cell
proliferation at 6 dpf (24 h BrdU labeling) compared with
larvae raised under germ-free conditions. We observed that
intestinal BrdU incorporation in larvae raised in
our conventional embryo media was identical to that of
larvae colonized by commensal bacteria reported by
K.N. Wallace et al. / Mechanisms of Development 122 (2005) 157–173
161
Fig. 2. Mammalian and teleost intestinal architecture: this schematic depicts intestinal layers within the mammalian (A) and teleost (B) intestine. Note that
teleosts lack a muscularis mucosa (black line between lamina propria and submucosa in (A). As a result, only a thin layer of connective tissue separates the
epithelium at the base of the folds from the inner circular layer of smooth muscle. The teleost intestinal epithelium is arranged in broad, irregular folds rather
than villi, and lacks crypts of Lieberkuhn. Individual myenteric neuronal cell bodies seen in zebrafish are also depicted in (B).
Rawls et al. (2004) (nZ19% BrdUCcells; 3447 total cells;
24 h labeling). Thus, the decline in cell proliferation rates
we observe in larvae raised in our conventional media is
likely to be the result of a normal physiological program.
The increased percentage of S-phase cells identified with
24-hour BrdU labeling (as opposed to a 1-h pulse) also
allowed a more accurate determination of where such cells
reside on the nascent folds. This showed that over 70% of
proliferating cells that could be assigned a position (nZ381
of 496 BrdUCcells; 2476 total cells analyzed) were
restricted to the inter-fold region. These data suggest that
progenitor cell specification begins at larval stages.
2.6.2. Maturation of polarized epithelial cells
Epithelial cells within the intestine are joined apically
through a complex set of junctional complexes (tight
junctions, adherens junctions and desmosomes). These
complexes restrict the movement of membrane components
between apical and basolateral cellular domains, and serve
as a barrier to the paracellular space through which luminal
contents may otherwise enter the organism. Our previous
work showed that basolateral cadherin is present within gut
epithelial cells at 34 hpf (Wallace and Pack, 2003). This
distribution is considered evidence for the presence of
adherens junctions and is supported by ultrastructural
analyses (not shown). The tight junction marker ZO-1 is
absent at this stage, but at 50 hpf scattered ZO-1
immunoreactivity is observed in the apical region of most
of the intestinal epithelia (Fig. 4A). This staining pattern is
pronounced at 74 hpf (Fig. 4B). At this stage, desmosomes
are first evident in electron micrographs (Fig. 4C). Further,
sodium–potassium ATPase, a transporter that is normally
restricted to the lateral cell membrane and may be
required for tight junction and desmosome formation
(Rajasekaran et al., 2001), is first apparent throughout the
intestine at this stage (Fig. 4D–F). These data show that
162
K.N. Wallace et al. / Mechanisms of Development 122 (2005) 157–173
Fig. 3. Biphasic intestinal growth in zebrafish embryos and larvae: (A–D) Lateral views of 50, 74, 96 and 120 hpf zebrafish. The intestine is prominent at 96 and
120 hpf (arrows C, D) but difficult to identify at 50 (A) and 72 hpf (B). (E–H) Corresponding histological cross-sections through the anterior (left) and posterior
(right) intestine of zebrafish depicted in (A–D). Note appearance of folds and columnar-shaped epithelial cells between 72 and 96 hpf. Folds are cut in crosssection (red arrow) and tangential to the long fold axis (black arrow). (I) High power view of intestinal folds in H shows cuboidal cells at base of fold (red
arrow) and adjacent columnar cells. (J) Epithelial cell proliferation is biphasic; nadir is reached at 120 hpf and cell proliferation does not increase within the
first 24 hpf of feeding (not shown). Arrowheads identify levels of histological sections. Arrows in histological sections point to the intestine unless otherwise
noted. (K,L) Transmission electron micrographs showing the apical region of a 74 (K) and 120 hpf (L) intestinal epithelial cell.
epithelial maturation, as measured by the appearance of
junctional complexes, is completed by 74 hpf, the time point
when the rate of epithelial proliferation begins to fall.
Interestingly, the sequence of junctional complex formation
in the developing zebrafish intestine (adherens junction
prior to tight junction and desmosome) is identical to that
reported to occur in cultured mammalian epithelial cells
(Gumbiner et al., 1988).
2.6.3. Coordinate development of enteric neuron
and smooth muscle progenitors
Enteric nervous system development in mammals has
been studied extensively. Recent work has shown that
development of the zebrafish enteric nervous system is
regulated by conserved molecular programs (Dutton et al.,
2001; Pingault et al., 2002; Shepherd et al., 2001, 2004).
Enteric neuron progenitors (cranial neural crest) populate
K.N. Wallace et al. / Mechanisms of Development 122 (2005) 157–173
163
Fig. 4. Step-wise polarization of developing intestinal epithelia: (A) cross-section through the anterior intestine of a 50 hpf embryo processed for ZO-1
immunohistochemistry. Apical ZO-1 (arrowheads) is first evident at this stage. (B) Apical ZO-1 staining in the anterior intestine is pronounced at 74 hpf.
(C) Transmission electron micrograph (TEM) shows well-defined tight junction (arrowheads) and desmosomes (arrow) in an intestial epithelial cell of a 74 hpf
larva. (D–F) Basolateral Na/K–ATPase is present in the 74 hpf anterior intestinal epithelium, but there is minimal staining in the posterior intestine at this stage
(E). Pronounced Na/K–ATPase is present in the posterior intestine at 5 dpf (F). (P) Pronephric ducts.
the mesenchyme surrounding the larval zebrafish intestinal
epithelium between 36 hpf and 74 hpf (Shepherd et al.,
2004). Here we show that intestinal smooth muscle
progenitors are identified around the same time. Based
upon the expression of the smooth muscle myosin heavy
chain gene these cells are first evident in the anterior
intestine at 50 hpf (Fig. 5A). Between 58 (Fig. 5B) and
72 hpf (not shown), these cells proliferate and populate all
intestinal segments, as do enteric neuron progenitors.
Timing and pattern of enteric neuron and smooth muscle
progenitor appearance along the anterior–posterior intestinal axis suggests that the development of these cell lineages
might be linked (discussed below). However, whether
smooth muscle progenitors migrate in an anterior to
posterior direction within the developing intestine, as
described for the enteric neuron progenitors, could not be
addressed by our in situ experiments.
2.6.4. Epithelial differentiation
Differentiated cells within the mammalian intestine are
post-mitotic. We defined the timing of epithelial, smooth
muscle and neuronal differentiation in developing larvae to
determine if the onset of cytodifferentiation coincided with
the decline in intestinal cell proliferation. As described
below, these studies suggest a link between differentiation
and progenitor cell proliferation.
Differentiation of larval epithelial cell types was
analyzed using markers used to identify enterocytes, goblet
cells and enteroendocrine cells within the adult intestine.
Enterocytes and goblet cells were the first differentiated cell
type we identified. The NaPi co-transporter first localized to
the apical border of epithelial cells at 74 hpf (Fig. 5D); its
expression was pronounced at 5 dpf (Fig. 5E). Enterocytes
expressing aminopeptidase in the anterior and mid intestine
were identified at identical time points (not shown).
NaPiCcells were located throughout the anterior and mid
intestine but were excluded from the posterior intestine.
Thus, the distribution of NaPiCenterocytes within the adult
intestine is already established in the zebrafish larvae at the
onset of enterocyte differentiation.
By contrast, goblet cells, which are present within all
adult intestinal segments, are present only within the larval
mid intestine (Fig. 5K). Like the NaPiCenterocytes, they
are first identified at 74 hpf, either histologically, or by the
binding of a fluorescently labeled wheat germ agglutinin
lectin (Fig. 5G). Their number and the size of their
mucinous theca increase significantly between 74 and
120 hpf (Fig. 5H). Pancreatic polypeptideCenteroendocrine cells within the anterior intestine and specialized
enterocytes (antigen presenting cells) within the posterior
segment of the mid intestine were first identified at 96 hpf
(Fig. 5F,I–K). The specialized enterocytes were identified
either histologically (not shown), or by the uptake of
horseradish peroxidase that had been injected into the
intestinal lumen. As with enterocytes, the distribution of
both cell types within the larval and adult intestine was
nearly identical. The relative distributions of the various
epithelial cell subtypes along the anterior–posterior axis of
the larval zebrafish intestine are depicted in Fig. 5C.
Epithelial morphology developed in concert with the
differentiated cell types. Columnar cell shape developed
progressively between 50 and 120 hpf, but these changes
were most pronounced between 74 and 120 hpf (Fig. 3F–H).
Apical membrane development, as determined by the
density of the microvillus brush border in electron
micrographs, also progressed markedly during these stages
(Fig. 3K,L). Maturation of the columnar epithelial morphology was coincident with the formation of the epithelial
folds. Interestingly, the position of cells within the fold
could be distinguished morphologically; cells at the tip of
164
K.N. Wallace et al. / Mechanisms of Development 122 (2005) 157–173
Fig. 5. Smooth muscle progenitors and epithelial differentiation: (A,B) Whole-mount RNA in situ showing intestinal smooth muscle myosin heavy chain
expression at 50 (A, arrowheads) and 58 hpf (B). (C) Cartoon depicting anterior (red), mid (green) and posterior (blue) segments of the 5 dpf larval intestine.
Distal region of the mid intestine that contains specialized enterocytes is stippled (arrowhead). (D,E,G,H) Cross-sections through the larval mid intestine. (D,E)
Enterocytes contain the NaPi transporter within their apical cell membrane. NaPiC enterocytes are first identified at 74 hpf (D); NaPi levels are pronounced at
120 hpf (E). (G) Goblet cell mucin, revealed here by rhodamine-dextran labeled WGA, is first evident at 78 hpf. (H) Mucin is much more abundant at 120 hpf
and is easily recognized in histological specimens at this stage (not shown). (J) Cross-section through the anterior intestine of a 96 hpf larva. Pancreatic
polypeptideC enteroendocrine cells are first identified at this stage. Note slender apical cytoplasm characteristic of these cells. (K) Lateral view of the 120 hpf
intestine showing regionalized distribution of differentiated epithelial cells. Pancreatic polypeptide containing enteroendocrine cells (green) are restricted to
the anterior intestine (intestinal bulb) whereas goblet cells identified with WGA (red) are restricted to the mid intestine. NaPiC enterocytes are present
throughout the anterior and mid intestine, but not the posterior intestine (not shown). (F) Lateral view of a portion of the mid intestine from a 96 hpf larva that
has ingested horseradish peroxidase protein (HRP). Following pinocytosis, HRP can be detected histochemically within the apical cytoplasm of specialized
enterocytes of the mid intestine (segment 2) that also express NaPi (not shown). (I) Sagittal histological section through the mid intestine of the larva shown in
(F), reveals HRP within the apical enterocyte cytoplasm.
the fold had a high columnar shape, whereas cells at the base
had a more cuboidal appearance (Fig. 3I). This observation
suggests that fold morphogenesis may be driven by changes
in cell morphology, rather than by outgrowth of underlying
connective tissue, which has been reported to direct
mammalian villus formation (Karlsson et al., 2000).
Consistent with this idea, a developing central connective
tissue core was rarely identified within the intestinal folds at
larval stages.
2.6.5. Smooth muscle and enteric neuron differentiation
Zebrafish enteric neuron differentiation is reported to
begin around 72 hpf (Kelsh and Eisen, 2000). The number
and distribution of HuCneurons within the intestine
increases between 96 and 120 hpf (Kelsh and Eisen, 2000;
Shepherd et al., 2004). By contrast, the timing of smooth
muscle differentiation has not been reported. Given that the
progenitors for both cell types populate the intestine at
similar stages, we predicted that the timing of smooth
muscle and enteric neuron differentiation would coincide.
Although smooth muscle progenitors express smooth
muscle myosin heavy chain at 58 hpf, antibody stainings for
myosin and desmin proteins are negative at this stage (not
shown). At 74 hpf, a discontinuous layer of circularly
aligned smooth muscle cells was evident in all intestinal
segments (Fig. 6A–C). Differentiated zebrafish enteric
neurons express the Hu antigen at 74 hpf (Kelsh and
Eisen, 2000) and have few axonal projections (Fig. 6G,H).
Enteric nerve cell bodies are restricted to the lateral
intestinal wall at this stage (Kelsh and Eisen, 2000;
Shepherd et al., 2001). At 96 hpf, the desminCcircular
smooth muscle cells form a continuous ring around the
epithelium and longitudinally aligned smooth muscle cells
are easily identified (Fig. 6D–F). The number and
distribution of enteric neurons also increase significantly
at this stage (96 hpf) and at 120 hpf ((Shepherd et al., 2001);
K.N. Wallace et al. / Mechanisms of Development 122 (2005) 157–173
Fig. 6I,K). Further, they develop complex axonal projections (Fig. 4J,L).
2.7. Mutations that disrupt epithelial, smooth muscle
and enteric nervous system differentiation
Mutations altering the appearance of the intestinal
epithelium have been identified in zebrafish morphologybased mutagenesis screens (Chen et al., 1996; Pack et al.,
1996). Epithelial detachment and/or degeneration are
165
phenotypic features shared by several mutants within this
class. We analyzed three of these mutants, slimjim (slj),
flotte lotte (flo), and piebald, (pie) using epithelial, smooth
muscle and neuronal markers to determine whether these
mutations alter epithelial differentiation or the development
of other intestinal lineages.
2.7.1. Epithelial differentiation defects
The appearance of the intestine of slj, flo, pie and wild
type cannot be reliably distinguished in live 74 hpf larvae.
Fig. 6. Intestinal smooth muscle development is coincident with enteric neuron differentiation: (A) Circular smooth muscle cells are first identified at 74 hpf by
desmin immunohistochemistry (lateral view of a the mid intestine; whole-mount projection with anterior–posterior axis aligned horizontally). (B) Crosssections at this stage reveal a discontinuous layer of desminC smooth muscle cells. (C) Enlarged inset from (B) (arrowhead) shows absence of longitudinal
fibers. (D) Lateral view of the 120 hpf mid intestine, desmin immunohistochemistry; same orientation as A. Well-defined longitudinal smooth muscle cells are
now evident. (E,F) Cross-sections reveal a nearly continuous inner layer of circular smooth muscle fibers and individual longitudinal smooth muscle cells
peripheral to the circular muscle layer. Arrowhead points to area from E that is enlarged in (F). (G) Whole-mount view of a 74 hpf intestine processed for Hu
immunohistochemistry shows enteric neuron cell bodies along the lateral borders of anterior, mid and posterior intestine. (H) Acetylated tubulin
immunohistochemistry at this stage shows early axonal projections of these cells. (I–L) Hu and tubulin stainings at 96 (I,J) and 120 hpf (K,L) show
proliferation and medial migration of enteric neurons and elaboration of their axonal projections.
166
K.N. Wallace et al. / Mechanisms of Development 122 (2005) 157–173
By 84–96 hpf, the size of the anterior intestine in each
mutant is decreased; this size difference, while variable
between individual mutant larvae, is pronounced in all
intestinal segments in all 120 hpf larvae (Fig. 7). Histological sections through the anterior intestine at this stage
(120 hpf) reveal that epithelial morphology is altered to a far
greater extent in slj and flo, in which cells appear cuboidal
and folds are absent, than in pie, in which the epithelium is
columnar and fold formation has initiated (Fig. 7A,C,E,G).
Ultrastructural analyses identified normal junctional complexes in each mutant (Pack et al., 1996 and not shown).
Immunohistochemical and functional assays of 5 dpf slj,
flo and pie larvae revealed that epithelial differentiation is
altered in each mutant. In flo, NaPiCenterocytes and goblet
cells are rare or absent and the number of pancreatic
polypeptideCenteroendocrine cells is reduced (Fig. 7F,N).
However, the relative number of endocrine cells appears
slightly increased when accounting for the reduced number
of epithelial cells in the anterior flo intestine (Table 1).
Injection of horseradish peroxidase into 4 and 5 dpf flo
larvae revealed only rare specialized enterocytes in 2 of 20
larvae (not shown). Thus, only a small number of
differentiated cells other than enteroendocrine cells are
found within the flo intestine.
Less pronounced epithelial defects are present within the
slj and pie intestine. NaPiC enterocytes were distributed in
a normal pattern in both mutants (Fig. 7D,H). As in flo, the
total number of pancreatic polypeptideCenteroendocrine
cells in slj and pie are decreased (Fig. 7L,P) but the relative
percentage of enteroendocrine cells appeared to be mildly
increased in both mutants (Table 1). Similarly, the
percentage of goblet cells within the posterior intestinal
epithelium was elevated relative to wild type (Table 1).
Finally, specialized enterocytes develop normally in the pie
mid intestine, but do not form in slj larvae (nZ5 larvae
analyzed; not shown).
2.7.2. Smooth muscle and enteric neuron defects
Signals exchanged between gut endoderm and mesenchyme direct intestinal smooth muscle development (Kedinger
et al., 1998; Ramalho-Santos et al., 2000). Accordingly,
smooth muscle defects were identified in mutants with the
most severe epithelial defects, slj and flo. Analysis of wholemount specimens and tissue sections (not shown) of slj and
flo larvae revealed reduced circular smooth muscle (slj and
flo) and increased longitudinal smooth muscle (flo)
compared with WT and pie larvae (Fig. 8A–D). Interestingly, smooth muscle myosin heavy chain expression
appeared normal in 84 hpf slj and flo larvae (not shown).
Thus, the smooth muscle defects in these mutants do not
appear to arise from a primary deficit of progenitor cells.
Smooth muscle defects in slj and flo were associated with
abnormalities of the enteric nervous system. In both
mutants, enteric neuron axons project in a predominantly
longitudinal direction and branching is highly reduced
(Fig. 8E,F). By contrast, this pattern is complex in WT
and pie larvae (Figs. 6L,8G). The distribution of the enteric
neuron cell bodies in flo and slj is also abnormal. Enteric
neurons in flo and slj, do not undergo the normal
circumferential migration from their original position
within the lateral intestinal wall (Fig. 8H,I). Further, there
are significantly less enteric neurons in both mutants. Using
Hu immunohistochemistry, we identified 383 enteric neuron
cell bodies per WT intestine (nZfour 5 dpf larvae), 401 cell
bodies per pie larvae (nZfour 5 dpf larvae), but only 122
and 225 enteric neurons per flo (nZfive 5 dpf larvae) and
slj (nZfour 5 dpf larvae), respectively. The number of
HuCneurons is already reduced approximately 50% at the
onset of smooth muscle differentiation in both slj and flo
larvae (WTZ102, sljZ56 and floZ47; nZfour 78 hpf
larvae for each). Consistent with these findings, intestinal
motility was altered in both mutants; food accumulated in
the buccopharynx of slj and flo larvae fed at 5 dpf and did
not enter the anterior intestine, whereas it was present in all
intestinal segments of WT and pie larvae (not shown).
Table 2 lists epithelial, smooth muscle and enteric nerve
defects in 5 dpf slj, flo and pie larvae.
3. Discussion
3.1. Zebrafish intestinal anatomy
Intestinal anatomy in amniotes is highly conserved.
Here, we identify a related architecture in the zebrafish
intestine that is also similar to other cyprinid fish.
The zebrafish intestine is joined to the anterior digestive
tract by a short muscular esophagus. Three intestinal
segments may be defined based upon the histological
appearance of the epithelial folds and the distribution of
differentiated epithelial cell types. Consistent with its
primary role in nutrient absorption, digestive enzymes are
strongly expressed in the anterior segment and the epithelial
surface area (fold height) is greatest in this region. We
define the mid intestine as the region with a related
epithelial architecture that lacks enteroendocrine cells.
Expression of solute transporters and digestive enzymes
(Pack, unpublished) within enterocytes of this segment
support a role in nutrient absorption. At this time it is not
known whether there are qualitative differences of nutrient
absorption or other intestinal functions between the anterior
and mid segments. However, quantitative differences of
fat and amino acid absorption within analogous intestinal
regions of other cyprinid fish have been reported
(Dabrowski, 1986; Noaillac-Depeyre and Gas, 1974).
Consistent with this idea, the zebrafish intestinal fatty acid
binding protein is most prominently expressed in the
anterior half of the zebrafish larval intestine (Her et al.,
2004; Mudumana et al., 2004).
The terminal region of the mid intestine is comprised of
specialized enterocytes that appear to play a role in mucosal
immunity. This region of the mid intestine may be
K.N. Wallace et al. / Mechanisms of Development 122 (2005) 157–173
167
Fig. 7. Mutations that alter epithelial maintenance also affect epithelial differentiation: (A–H) Cross-sections through the mid intestine of 5 dpf WT and
intestinal mutants processed for routine histology and NaPi immunohistochemistry. (A) The WT intestine has a folded epithelium comprised of columnar
epithelial cells. (B) Apical NaPi in the WT mid intestine. (C) Although the slj intestine appears undifferentiated, a similar NaPi pattern is present (D). (E) The
epithelium in flo also appears undifferentiated, but lacks apical NaPi staining (F). (G) The epithelium of pie larvae is columnar with some folds and thus appears
more developed than in slj or flo. (H) Apical NaPi is present in the pie epithelium. (I,K,M,O) Lateral views of the WT, slj, flo and pie intestine. (J,L,N,O)
Similar views following detection of enteroendocrine cells (pancreatic polypeptide immunohistochemistry; green) and goblet cells (fluorescent conjugated
wheat germ agglutinin lectin; red). (I,J) The WT intestine is folded; pancreatic polypeptideCenteroendocrine cells are restricted to the anterior intestine
whereas goblet cells are located in the mid intestine. (K–P) The size of the anterior intestine and the number of enteroendocrine cells is reduced in slj (K,L), flo
(M,N) and pie (O,P) mutants, although intestinal morphology is most severely altered in flo and slj compared with pie. Goblet cells appear normal in slj (L) and
pie (P) but are absent in flo (N). Arrowheads point to the slj (K) and flo (M) intestine.
168
K.N. Wallace et al. / Mechanisms of Development 122 (2005) 157–173
Table 1
Relative amount of enteroendocrine and goblet cells in slj, flo and pie larvae
WT
slj
flo
pie
PPC cells
per larva
Epith. cellsa
Relative number
of PPC cellsb
Goblet cellsc
Epith. cellsc
Relative number
of goblet cellsd
60 (45–78)
18.7 (14–28)
18 (11–28)
22.3 (19–28)
1102 (984–1209)
182 (153–216)
186 (171–202)
242.3 (209–285)
1
1.9
1.8
1.7
15 (11–19)
20.3 (18–24)
NA
30 (26–38)
305.3 (268–331)
93.3 (74–116)
NA
140.3 (123–150)
1
4.4
NA
4.3
PP—pancreatic polypeptide; NA—not analyzed.
a
Total number of cells in eight histological cross-sections through the anterior intestine: nZ3 WT, 3 slj, 3 flo and 3 pie larvae.
b
Mutant/WT PPC cells: mutant/WT total epithelial cells.
c
Goblet cell and total epithelial cells per larvae in eight histological cross-sections through the mid intestine: nZ3 WT, 3 slj and 3 pie larvae.
d
Mutant/WT goblet cells: mutant/WT total epithelial cells.
analogous to the region of the mammalian ileum where
antigen presenting epithelial cells (M cells) and submucosal
lymphoid aggregates known as Peyer’s patches are located.
The short posterior segment begins immediately posterior to
this region. Here the epithelial folds are short and
longitudinally arrayed. The architecture and absence of
absorptive enterocytes in this region suggest it may be
analogous to the colon of higher vertebrates. Ultrastructural
studies performed in other cyprinid fish and analyses of
intestinal motility in zebrafish larvae support this idea
(Holmberg et al., 2003; Stroband and Debets, 1978).
Histological analyses show that the adult zebrafish
intestine is arranged in concentric layers similar to those
present within the mammalian intestine. However, our
studies show that the supporting connective tissue layers are
less complex in zebrafish as reported for other cyprinid fish
(Curry, 1937; Rogick, 1931). Zebrafish and other teleosts
also lack a smooth muscle layer comparable to the
Fig. 8. Circular smooth muscle and enteric neuron defects accompany epithelial alterations in slj and flo, but not in pie mutants. (A–D) Whole-mount view of a
segment of the mid intestine of WT, flo, slj and pie larvae processed for desmin immunohistochemistry. The WT intestine (A) contains short longitudinal
smooth muscle fibers and longer, underlying circular smooth muscle fibers. In the flo (B) and slj (C) intestine the longitudinal fibers predominate. By contrast,
smooth muscle appears normal in pie larvae (D). (E–G) Axonal projections of enteric neurons in the mid intestine of 5 dpf flo, slj and pie larvae. Axonal
projections of flo (E) and slj (F) are less complex than WT (Fig. 6L) even at 96 hpf. Axonal projections are predominantly aligned along the anterior–posterior
axis in both mutants. (G) Axonal projections in pie larvae appear normal. (H–J) Whole-mount images of 5 dpf flo, slj and pie larvae processed for Hu
immunohistochemistry. Compared with WT larvae (Fig. 6I,K) fewer flo and slj enteric neurons have migrated from the lateral intestinal borders, whereas the
pie pattern of HuC cells more closely resembles WT (Fig. 6K). The number of HuC cells is reduced in flo and slj and fewer flo HuC cells have migrated to the
posterior intestinal segment.
K.N. Wallace et al. / Mechanisms of Development 122 (2005) 157–173
169
Table 2
Epithelial, smooth muscle and enteric nerve defects in zebrafish intestinal mutants
flo
slj
pie
NaPi
P. polypepa
Gobleta
Spec. ent.
Folds
CSM
LSM
ENS
Absent
Normal
Normal
[
[
[
YYYY
[
[
YYYY
YYYY
Normal
Absent
Absent
Initiated
YYY
YYY
Normal
[
[
Normal
YYY
YY
Normal
Number and direction of arrows signifies magnitude and quantity of various cell types compared with wild type siblings. NaPi, Sodium phosphate transporterC
enterocytes; P. Polypeptide, Pancreatic polypeptideC enteroendocrine cells; Spec. Ent., specialized enterocytes; CSM, circular smooth muscle; LSM,
longitudinal smooth muscle; ENS, enteric nervous system.
a
Refers to the number of enteroendocrine and goblet cells per total epithelial cells in mutant vs. wild type larvae.
muscularis mucosa. Contraction of these longitudinally
aligned smooth muscle fibers enables the epithelium to
move independently of the intestinal wall and thereby aids
digestion. We hypothesize that this thin layer of smooth
muscle is not required in zebrafish because of the thin
caliber of the intestinal wall. These differences notwithstanding, the overall distribution of blood vessels and the
surrounding smooth muscle layers was nearly identical to
that of mammals. Finally, our studies identified cell bodies
of enteric neurons between the smooth muscle layers but not
the adjacent connective tissue. We speculate that enteric
nervous system architecture is less complex in zebrafish
than in mammals because of the closer proximity of the
smooth muscle and epithelial layers.
The organization of the intestinal epithelium into broad
irregular folds rather than finger-like villi, and the presence
of proliferative cells at the base of these folds, rather than in
specialized glands (crypts), is another distinctive feature of
the zebrafish intestine. This configuration of the proliferative compartment within the zebrafish intestine resembles
the organization of proliferative cells within the mammalian
embryonic intestine (Korinek et al., 1998). The absence of
fully differentiated cell types at the base of the epithelial
folds supports the idea that some type of undifferentiated or
less differentiated progenitor cell population exists within
the adult zebrafish intestine. Studies of other cyprinid fish
have shown that proliferative cells function in nutrient
absorption and pinocytosis of macromolecules (Rombout
et al., 1984; Stroband and Debets, 1978). Whether such cells
are multipotent is unknown. Interestingly, preliminary
studies with a zebrafish ortholog of the atonal-1 gene
(unpublished), which in mammals directs differentiation of
goblet, enteroendocrine and Paneth cell intestinal lineages
from a multipotent progenitor (Yang et al., 2001) were
unrevealing, as we did not observe intestinal expression of
this gene in zebrafish larvae and gene knockdowns did not
alter epithelial differentiation (unpublished). However, it is
conceivable that such a role is played by an unidentified
zebrafish atonal gene. Alternatively, it is conceivable that
renewal of each epithelial lineage in the zebrafish intestine
is regulated by a distinct progenitor cell, as suggested for
other cyprinid fish (Rombout et al., 1984).
Several other distinctive features of the zebrafish
intestine we report may be related to the markers used in
this study. The distribution and type of enteroendocrine
cells are one example. A previous survey of 11 teleost fish
identified pancreatic polypeptide as the only hormone
produced by zebrafish enteroendocrine cells (Langer et al.,
1979). Our own studies confirmed the presence of such cells
within the anterior intestine, and that this pattern is present
at the outset of intestinal epithelial differentiation. Although
these data may be a function of the antisera used for this
study, these reagents recognized a wide range of hormone
secreting cells in closely related teleost fish. Thus, it seems
plausible that pancreatic polypeptide is the only common
hormone produced by zebrafish enteroendocrine cells. This
hypothesis will be tested directly when the zebrafish
orthologs of other mammalian enteroendocrine markers
are identified.
Similarly, our histological survey did not identify cells
resembling mammalian Paneth cells. Although some
mammals and other vertebrates have been reported to also
lack intestinal Paneth cells, future studies using molecular
markers for defensins, which have not been cloned in
zebrafish, or other genes expressed by Paneth cells will help
further define this question. Finally, we also did not identify
c-kit positive pacemaker cells (interstitial cells of Cajal)
within the muscularis layer (unpublished). Whether a
second c-kit ortholog in zebrafish accounts for this disparity
can be addressed in future studies.
3.2. Coordinated differentiation programs within
the developing zebrafish intestine
The markers we used to identify differentiated cell
types within the adult zebrafish intestine allowed us to
chart the timing of epithelial, smooth muscle and enteric
neuron differentiation in zebrafish larvae. We defined
differentiated cells based upon the appearance of protein
markers rather than gene expression, which we consider
as identifying proto-differentiated cells. Our studies
revealed several interesting findings. First, as judged by
the appearance of these markers, differentiation of cell
types derived from the three germ layers begins more or
less contemporaneously (w74 hpf). Second, this wave of
cell differentiation occurs in the setting of a pronounced
decline in epithelial cell proliferation. Third, regionspecific differentiation of enterocytes and enteroendocrine
cells in the zebrafish larvae presaged their distribution in
the adult intestine, whereas the distribution of goblet
170
K.N. Wallace et al. / Mechanisms of Development 122 (2005) 157–173
cells expands during intestinal maturation. Whether the
presence of goblet cells in the adult anterior and
posterior intestine results from cell migration or the
appearance of a new cell lineage in this region can be
addressed in future studies.
The idea that epithelial, smooth muscle and enteric
neuron differentiation in the zebrafish intestine are linked is
not surprising given that tissue recombination and gene
targeting experiments have identified related interactions
during intestinal development in other vertebrates (Kaestner
et al., 1997; Karlsson et al., 2000; Kedinger et al., 1998;
Roberts et al., 1998). Whether these and other genes that
regulate epithelial morphogenesis in mammals (Karlsson
et al., 2000; Pabst et al., 1999; Ramalho-Santos et al., 2000;
Saotome et al., 2004) play a related role in formation of the
zebrafish intestinal folds is not known. Conservation of such
programs would be noteworthy because villus formation in
mammals involves transition from a flat, stratified architecture to a folded, single-layered columnar architecture,
whereas zebrafish retain a single-layered intestinal epithelium throughout embryonic and larval development.
3.3. Stages of intestinal development
The finding that cytodifferentiation within the developing zebrafish intestine occurs in the setting of reduced cell
proliferation suggests that differentiation programs are
more likely to be initiated upon exit from the cell cycle.
The distinction between cell proliferation and differentiation, albeit somewhat arbitrary, allows us to define two
broad stages of intestinal development following formation
of the intestinal anlage (gut tube). In the first stage
we define, there is widespread proliferation of progenitor
cells and maturation of the polarized phenotype. In the
second stage, the percentage of proliferative cells declines
and differentiation of epithelial cells, smooth muscle, and
enteric neurons begins. The report that the RNA binding
protein gene nil per os (npo), which is strongly expressed
in the intestine at early stages, regulates intestinal size
and polarization supports this distinction (Mayer and
Fishman, 2003). Preferential expression of RNA binding
protein genes by progenitor cells within mammalian
intestinal crypts also supports this hypothesis (Stappenbeck
et al., 2003).
3.4. Intestinal mutants help define tissue interactions during
organ development
Phenotypic analyses of mutants presented in this report
support the importance of tissue interactions during
zebrafish intestinal development. Defects in intestinal cell
types derived from the three primary germ layers were
identified in two mutants, slj and flo. Given the importance
of epithelial–mesenchymal interactions during intestinal
development in mammals and birds, defective signaling
between these cell types in developing slj and flo larvae
seems likely. Alternatively, it is also possible that the slj
and flo genes function autonomously in the affected tissues,
or they may function in a stage-dependent manner, as
described for the cloche gene in developing erythrocytes
(Parker and Stainier, 1999). Preliminary transplantation
experiments suggest that the flo epithelial defects arise
non-autonomously (unpublished). We hypothesize that this
signal (putative) most likely arises from a mesenchymal
cell rather than a crest-derived enteric neuron or glial
cell because intestinal morphology is normal in zebrafish
larvae with severe enteric nervous system defects
(Shepherd et al., 2004).
Phenotypic analyses of intestinal mutants also offer
insight into the hierarchical relationships of developmental
programs. Our studies show that slj and flo have similar
smooth muscle and enteric nerve defects yet different
epithelial phenotypes. Differentiated enterocytes and goblet
cells, as defined by the markers used for this study, are rare
or absent in flo larvae whereas enterocytes develop normally
and goblet cells are increased relative to wild type siblings
in slj larvae. These data suggest that epithelial differentiation may not be fully dependent upon normal smooth
muscle and enteric nerve development. Differences in the slj
and flo epithelial phenotypes may arise from the fact that the
flo mutation disrupts an early mesenchymal signal required
for enterocyte and goblet cell differentiation whereas the slj
gene encodes a mesenchymal signal that is required at a
later developmental stage. Alternatively, the slj mutation
may disrupt a gene that functions autonomously within the
epithelium and non-autonomously regulates smooth muscle
and enteric nerve development. Consistent with the latter
model, studies in chicken embryos show that endodermal
cells possess genetic information required for regionspecific epithelial differentiation (Duluc et al., 1994;
Roberts et al., 1998).
In contrast to epithelial differentiation, epithelial
morphology is altered in a similar fashion in slj and flo,
whereas it is more advanced in pie, a mutant with largely
normal smooth muscle and enteric nerves. Thus, it is
possible that normal smooth muscle and/or enteric nerves
are required for epithelial cells to develop a columnar
morphology and arrange into folds. Consistent with this
idea, targeting of mesenchymal genes alters villus
formation in mammals (Calabi et al., 2001; Haramis
et al., 2004; Kaestner et al., 1997; Karlsson et al., 2000;
Pabst et al., 1999). In addition, tissue recombination
experiments have consistently demonstrated the ability of
heterologous mesenchymes to redirect epithelial morphogenesis in a variety of fetal endoderms (Duluc et al., 1994;
Mizuno and Yasugi, 1990). Neuronal paucity and axonal
defects in slj and flo may also arise from perturbation of
mesenchymal signals. Based on studies in mammals, such
defects may result from the altered migration, proliferation
or differentiation of crest-derived enteric neuron progenitors. Epithelial signals have also been shown to
either positively or negatively regulate enteric neuron
K.N. Wallace et al. / Mechanisms of Development 122 (2005) 157–173
development (Crone et al., 2003; Ramalho-Santos et al.,
2000; Sukegawa et al., 2000) and alteration of such signals
may also account for slj and flo enteric nerve defects.
Recently, gene-targeting experiments have begun to
define molecular mechanisms of mammalian intestinal
organogenesis. These studies have identified roles for
signaling pathways that function in other aspects of
vertebrate development. For example, Hedgehog signaling
has been shown to play a role in smooth muscle
development, enteroendocrine differentiation and villus
morphogenesis (Ramalho-Santos et al., 2000). Interference
with smooth muscle and goblet cell development also
occurs following disruption of the gene encoding the
laminin-a5 chain (Bolcato-Bellemin et al., 2003). BMP
signaling regulates epithelial progenitor cell identity, in part
through activation of the Wnt signaling pathway (Haramis
et al., 2004; Korinek et al., 1998; van de Wetering et al.,
2002). Further, roles for BMPs in smooth muscle and enteric
nerve development have been reported (Chalazonitis et al.,
2004). Finally, disruption of Notch signaling increases the
numbers of enteroendocrine and goblet cells relative to
enterocytes. One goal of future studies will be to determine
whether the expression of genes that participate in
orthologous zebrafish signaling pathways is altered in
mutants with epithelial and smooth muscle defects.
Similarly, expression of neurotrophin-3, GDNF, neurturin
or other signaling molecules that direct mammalian enteric
neuron development through the Ret signaling system may
also be altered in intestinal mutants. Related enteric nerve
defects in slj, flo and morphants in which GDNF signaling is
targeted (Shepherd et al., 2004) supports this possibility.
4. Experimental Procedures
4.1. Fish stocks
Fish maintenance and matings were performed as
described (Westerfield, 1993). AB strain wild type fish
(Westerfield, 1993) were used for histological, immunohistochemical, and in situ analysis. Mutant alleles used for
analysis were floti262c (Chen et al., 1996), sljm74 and pie m497
(Pack et al., 1996).
4.2. Histology
Embryos and adult intestines were processed as previously described (Wallace and Pack, 2003).
171
and incubated with secondary antibody (1:500 dilution) for
2 h at room temperature. Analysis of whole-mount immunostainings was performed using a Nikon E600 microscope.
For histological analysis, embryos were embedded in JB4
(Polysciences) and sectioned by microtome (5 mm). Primary
antibodies used were rabbit anti-pancreatic polypeptide
(1:250 dilution) (a gift of J. Polak), rabbit anti-type IIb
sodium-phosphate co-transporter (1:100 dilution) (a gift of
A. Werner), rabbit anti-desmin (1:100 dilution) (Sigma),
mouse anti-acetylated tubulin (1:100 dilution) (Sigma),
mouse anti-HuC/D (1:50 dilution) (Molecular Probes),
mouse anti-Zo1 (1:100 dilution) (gift of Dr S. Tsukita and
T. Obara), rabbit anti-sodium/potassium ATPase (1:100
dilution) (Developmental Studies Hybridoma Bank).
Secondary antibodies were Alexa Fluor 488- or 564conjugated anti rabbit or mouse Ig (1:500) (Vector
Laboratories). Goblet cell mucin was detected with
rhodamine conjugated wheat germ agglutinin (1:100
dilution) (Vector Laboratories) incubated 2 h to overnight.
TUNEL assay was used to detect apoptotic cells (Oncor).
4.4. RNA in situ hybridization
Whole-mount RNA in situ hybridization was performed
as previously described (Wallace and Pack, 2003). Antisense
probe was transcribed from a cDNA for zebrafish smooth
muscle myosin heavy chain.
4.5. 5-Bromo-2 0 -deoxy-uridine (BrdU) and Horseradish
Peroxidase (HRP) incorporation
BrdU (30 mM) (Sigma) was microinjected into the
peritoneal cavity of adult and larval fish or the yolk sac of
embryos. After 1 h the animals were fixed in 4% paraformaldehyde for 2 h to overnight. Alternatively, embryos were
incubated in embryo medium containing 160 mg/ml BrdU for
24 h. Fixed embryos were pretreated with 0.1% Collagenase
(Sigma) in PBS for 20 min followed by 0.2N HCL for 30 min
at room temperature. Mouse anti-BrdU (1:100 dilution)
(Roche) was incubated overnight at 4oC. Embryos were
washed and incubated with biotin-conjugated anti-mouse Ig
(Vector Laboratories) (1:500 dilution) for 2 h at room
temperature. Histochemical detection was performed with
Vectastain/Vector SG (Vector Laboratories).
Embryos were either microinjected or incubated with
HRP (10 mg/ml) for 2 h and fixed in 4% paraformaldehyde
for 2 h to overnight. Embryos were treated with 0.1%
Collagenase (Sigma) in PBS for 20 min and incorporated
HRP was detected with Vector SG (Vector Laboratories).
4.3. Immunohistochemistry
Embryos and adult intestines were fixed in 4%
paraformaldehyde for 2 h to overnight. Fixed embryos
were pretreated with 0.1% Collagenase (Sigma; C-9891) in
PBS for 20 min at room temperature and incubated in
primary antibody overnight at 4 8C. Embryos were washed
Acknowledgements
This work was supported by NIH grants DK54942 (MP)
and DK61142 (MP) and core facilities provided NIH Center
Grant (P30) DK50306.
172
K.N. Wallace et al. / Mechanisms of Development 122 (2005) 157–173
References
Amiel, J., Lyonnet, S., 2001. Hirschsprung disease, associated syndromes,
and genetics: a review. J. Med. Genet. 38, 729–739.
Bolcato-Bellemin, A.L., Lefebvre, O., Arnold, C., Sorokin, L., Miner, J.H.,
Kedinger, M., et al., 2003. Laminin alpha5 chain is required for
intestinal smooth muscle development. Dev. Biol. 260, 376–390.
Bry, L., Falk, P.G., Midtvedt, T., Gordon, J.I., 1996. A model of host–
microbial interactions in an open mammalian ecosystem. Science 273,
1380–1383.
Calabi, F., Pannell, R., Pavloska, G., 2001. Gene targeting reveals a crucial
role for MTG8 in the gut. Mol. Cell Biol. 21, 5658–5666.
Chalazonitis, A., D’Autreaux, F., Guha, U., Pham, T.D., Faure, C.,
Chen, J.J., et al., 2004. Bone morphogenetic protein-2 and -4 limit the
number of enteric neurons but promote development of a TrkCexpressing neurotrophin-3-dependent subset. J. Neurosci. 24, 4266–
4282.
Chen, J.N., Haffter, P., Odenthal, J., Vogelsang, E., Brand, M., van
Eeden, F.J., et al., 1996. Mutations affecting the cardiovascular
system and other internal organs in zebrafish. Development 123,
293–302.
Crone, S.A., Negro, A., Trumpp, A., Giovannini, M., Lee, K.F., 2003.
Colonic epithelial expression of ErbB2 is required for postnatal
maintenance of the enteric nervous system. Neuron 37, 29–40.
Curry, E., 1937. The histology of the digestive tube of the carp (cyprinus
carpio communis). J. Morphol. 65, 53–78.
Dabrowski, K., 1986. Protein digestion and amino acid absorption along the
intestine of the common carp (Cyprinus carpio L.), a stomachless fish:
an in vivo study. Reprod. Nutr. Dev. 26, 755–766.
de Santa Barbara, P., Van den Brink, G.R., Roberts, D.J., 2003.
Development and differentiation of the intestinal epithelium. Cell
Mol. Life Sci. 60, 1322–1332.
Duluc, I., Freund, J.N., Leberquier, C., Kedinger, M., 1994. Fetal endoderm
primarily holds the temporal and positional information required for
mammalian intestinal development. J. Cell Biol. 126, 211–221.
Dutton, K.A., Pauliny, A., Lopes, S.S., Elworthy, S., Carney, T.J.,
Rauch, J., et al., 2001. Zebrafish colourless encodes sox10 and specifies
non-ectomesenchymal neural crest fates. Development 128, 4113–
4125.
Fodde, R., 2002. The APC gene in colorectal cancer. Eur. J. Cancer 38,
867–871.
Gebert, A., Rothkotter, H.J., Pabst, R., 1996. M cells in Peyer’s patches of
the intestine. Int. Rev. Cytol. 167, 91–159.
Gershon, M.D., 1999. Endothelin and the development of the enteric
nervous system. Clin. Exp. Pharmacol. Physiol. 26, 985–988.
Gianino, S., Grider, J.R., Cresswell, J., Enomoto, H., Heuckeroth, R.O.,
2003. GDNF availability determines enteric neuron number by
controlling precursor proliferation. Development 130, 2187–2198.
Graham, C., Nalbant, P., Scholermann, B., Hentschel, H., Kinne, R.K.,
Werner, A., 2003. Characterization of a type IIb sodium–phosphate
cotransporter from zebrafish (Danio rerio) kidney. Am. J. Physiol.
Renal Physiol. 284, F727–F736.
Gumbiner, B., Stevenson, B., Grimaldi, A., 1988. The role of the cell
adhesion molecule uvomorulin in the formation and maintenance of the
epithelial junctional complex. J. Cell Biol. 107, 1575–1587.
Haramis, A.P., Begthel, H., Van den Born, M., Van Es, J., Jonkheer, S.,
Offerhaus, G.J., et al., 2004. De novo crypt formation and juvenile
polyposis on BMP inhibition in mouse intestine. Science 303, 1684–
1686.
Her, G.M., Chiang, C.C., Wu, J.L., 2004. Zebrafish intestinal fatty acid
binding protein (I-FABP) gene promoter drives gut-specific expression
in stable transgenic fish. Genesis 38, 26–31.
Holmberg, A., Schwerte, T., Fritsche, R., Pelster, B., Homgren, S., 2003.
Ontogeny of intestinal motility in correlation to neuronal development
in zebrafish embryos and larvae. J. Fish Biol. 63, 318–331.
Jang, M.H., Kweon, M.N., Iwatani, K., Yamamoto, M., Terahara, K.,
Sasakawa, C., et al., 2004. Intestinal villous M cells: an antigen entry
site in the mucosal epithelium. Proc. Natl Acad. Sci. USA 101, 6110–
6115.
Jensen, J., Pedersen, E.E., Galante, P., Hald, J., Heller, R.S., Ishibashi, M.,
et al., 2000. Control of endodermal endocrine development by Hes-1.
Nat. Genet. 24, 36–44.
Kaestner, K.H., Silberg, D.G., Traber, P.G., Schutz, G., 1997. The
mesenchymal winged helix transcription factor Fkh6 is required for
the control of gastrointestinal proliferation and differentiation. Genes
Dev. 11, 1583–1595.
Kapur, R.P., Gershon, M.D., Milla, P.J., Pachnis, V., 2004. The influence of
Hox genes and three intercellular signalling pathways on enteric
neuromuscular development. Neurogastroenterol. Motil. 16 (Suppl. 1),
8–13.
Karlsson, L., Lindahl, P., Heath, J.K., Betsholtz, C., 2000. Abnormal
gastrointestinal development in PDGF-A and PDGFR-(alpha) deficient
mice implicates a novel mesenchymal structure with putative
instructive properties in villus morphogenesis. Development 127,
3457–3466.
Kedinger, M., Duluc, I., Fritsch, C., Lorentz, O., Plateroti, M., Freund, J.N.,
1998. Intestinal epithelial–mesenchymal cell interactions. Ann. NY
Acad. Sci. 859, 1–17.
Kelsh, R.N., Eisen, J.S., 2000. The zebrafish colourless gene regulates
development of non-ectomesenchymal neural crest derivatives. Development 127, 515–525.
Korinek, V., Barker, N., Moerer, P., van Donselaar, E., Huls, G.,
Peters, P.J., et al., 1998. Depletion of epithelial stem-cell compartments
in the small intestine of mice lacking Tcf-4. Nat. Genet. 19, 379–383.
Kuhnert, F., Davis, C.R., Wang, H.T., Chu, P., Lee, M., Yuan, J., et al.,
2004. Essential requirement for Wnt signaling in proliferation of adult
small intestine and colon revealed by adenoviral expression of
Dickkopf-1. Proc. Natl Acad. Sci. USA 101, 266–271.
Langer, M., Van Noorden, S., Polak, J.M., Pearse, A.G., 1979. Peptide
hormone-like immunoreactivity in the gastrointestinal tract and
endocrine pancreas of eleven teleost species. Cell Tissue Res. 199,
493–508.
Marusich, M.F., Furneaux, H.M., Henion, P.D., Weston, J.A., 1994. Hu
neuronal proteins are expressed in proliferating neurogenic cells.
J. Neurobiol. 25, 143–155.
Mayer, A.N., Fishman, M.C., 2003. Nil per os encodes a conserved RNA
recognition motif protein required for morphogenesis and cytodifferentiation of digestive organs in zebrafish. Development 130, 3917–
3928.
Mizuno, T., Yasugi, S., 1990. Susceptibility of epithelia to directive
influences of mesenchymes during organogenesis: uncoupling of
morphogenesis and cytodifferentiation. Cell Differ. Dev. 31, 151–159.
Mudumana, S.P., Wan, H., Singh, M., Korzh, V., Gong, Z., 2004.
Expression analyses of zebrafish transferrin, ifabp, and elastaseB
mRNAs as differentiation markers for the three major endodermal
organs: liver, intestine, and exocrine pancreas. Dev. Dyn. 230, 165–173.
Noaillac-Depeyre, J., Gas, N., 1974. Fat absorption by the enterocytes of
the carp (Cyprinus carpio L.). Cell Tissue Res. 155, 353–365.
Nosrat, C.A., Tomac, A., Lindqvist, E., Lindskog, S., Humpel, C.,
Stromberg, I., et al., 1996. Cellular expression of GDNF mRNA
suggests multiple functions inside and outside the nervous system. Cell
Tissue Res. 286, 191–207.
Pabst, O., Zweigerdt, R., Arnold, H.H., 1999. Targeted disruption of the
homeobox transcription factor Nkx2-3 in mice results in postnatal
lethality and abnormal development of small intestine and spleen.
Development 126, 2215–2225.
Pack, M., Solnica-Krezel, L., Malicki, J., Neuhauss, S.C., Schier, A.F.,
Stemple, D.L., et al., 1996. Mutations affecting development of
zebrafish digestive organs. Development 123, 321–328.
Parker, L., Stainier, D.Y., 1999. Cell-autonomous and non-autonomous
requirements for the zebrafish gene cloche in hematopoiesis. Development 126, 2643–2651.
K.N. Wallace et al. / Mechanisms of Development 122 (2005) 157–173
Pingault, V., Girard, M., Bondurand, N., Dorkins, H., Van Maldergem, L.,
Mowat, D., et al., 2002. SOX10 mutations in chronic intestinal pseudoobstruction suggest a complex physiopathological mechanism. Hum.
Genet. 111, 198–206.
Rajasekaran, S.A., Palmer, L.G., Moon, S.Y., Peralta Soler, A.,
Apodaca, G.L., Harper, J.F., et al., 2001. Na,K-ATPase activity is
required for formation of tight junctions, desmosomes, and induction of
polarity in epithelial cells. Mol. Biol. Cell 12, 3717–3732.
Ramalho-Santos, M., Melton, D.A., McMahon, A.P., 2000. Hedgehog
signals regulate multiple aspects of gastrointestinal development.
Development 127, 2763–2772.
Rawls, J.F., Samuel, B.S., Gordon, J.I., 2004. Gnotobiotic zebrafish reveal
evolutionarily conserved responses to the gut microbiota. Proc. Natl
Acad. Sci. USA 101, 4596–4601.
Roberts, D.J., Smith, D.M., Goff, D.J., Tabin, C.J., 1998. Epithelial–
mesenchymal signaling during the regionalization of the chick gut.
Development 125, 2791–2801.
Rogick, M.D., 1931. Studies on the comparative histology of the digestive
tube of certain teleost fishes. J. Morphol. 52, 1–25.
Rombout, J.H., Lamers, C.H., Helfrich, M.H., Dekker, A., TaverneThiele, J.J., 1985. Uptake and transport of intact macromolecules in the
intestinal epithelium of carp (Cyprinus carpio L.) and the possible
immunological implications. Cell Tissue Res. 239, 519–530.
Rombout, J.H., Stroband, H.W., Taverne-Thiele, J.J., 1984. Proliferation and
differentiation of intestinal epithelial cells during development of Barbus
conchonius (Teleostei Cyprinidae). Cell Tissue Res. 236, 207–216.
Saotome, I., Curto, M., McClatchey, A.I., 2004. Ezrin is essential for
epithelial organization and villus morphogenesis in the developing
intestine. Dev. Cell 6, 855–864.
Shepherd, I.T., Beattie, C.E., Raible, D.W., 2001. Functional analysis of
zebrafish GDNF. Dev. Biol. 231, 420–435.
Shepherd, I.T., Pietsch, J., Elworthy, S., Kelsh, R.N., Raible, D.W., 2004.
Roles for GFRalpha1 receptors in zebrafish enteric nervous system
development. Development 131, 241–249.
Shin, M.K., Levorse, J.M., Ingram, R.S., Tilghman, S.M., 1999. The
temporal requirement for endothelin receptor-B signalling during
neural crest development. Nature 402, 496–501.
173
Stappenbeck, T.S., Mills, J.C., Gordon, J.I., 2003. Molecular features of
adult mouse small intestinal epithelial progenitors. Proc. Natl Acad. Sci.
USA 100, 1004–1009.
Stroband, H.W., Debets, F.M., 1978. The ultrastructure and renewal of the
intestinal epithelium of the juvenile grasscarp. Ctenopharyngodon
idella (Val.). Cell Tissue Res. 187, 181–200.
Sukegawa, A., Narita, T., Kameda, T., Saitoh, K., Nohno, T., Iba, H., et al.,
2000. The concentric structure of the developing gut is regulated by
Sonic hedgehog derived from endodermal epithelium. Development
127, 1971–1980.
Uribe, A., Alam, M., Midtvedt, T., Smedfors, B., Theodorsson, E., 1997.
Endogenous prostaglandins and microflora modulate DNA synthesis
and neuroendocrine peptides in the rat gastrointestinal tract. Scand.
J. Gastroenterol. 32, 691–699.
van de Wetering, M., Sancho, E., Verweij, C., de Lau, W., Oving, I.,
Hurlstone, A., et al., 2002. The beta-catenin/TCF-4 complex imposes a
crypt progenitor phenotype on colorectal cancer cells. Cell 111, 241–
250.
van den Brink, G.R., Bleuming, S.A., Hardwick, J.C., Schepman, B.L.,
Offerhaus, G.J., Keller, J.J., et al., 2004. Indian Hedgehog is an
antagonist of Wnt signaling in colonic epithelial cell differentiation.
Nat. Genet. 36, 277–282.
Wallace, K.N., Pack, M., 2003. Unique and conserved aspects of gut
development in zebrafish. Dev. Biol. 255, 12–29.
Westerfield, M., 1993. The Zebrafish Book. University of Oregon Press,
Eugene, OR.
Widenfalk, J., Nosrat, C., Tomac, A., Westphal, H., Hoffer, B., Olson, L.,
1997. Neurturin and glial cell line-derived neurotrophic factor receptorbeta (GDNFR-beta), novel proteins related to GDNF and GDNFRalpha with specific cellular patterns of expression suggesting roles in the
developing and adult nervous system and in peripheral organs.
J. Neurosci. 17, 8506–8519.
Yang, Q., Bermingham, N.A., Finegold, M.J., Zoghbi, H.Y., 2001.
Requirement of Math1 for secretory cell lineage commitment in the
mouse intestine. Science 294, 2155–2158.