Download Embryonic Wnt gene expression in the nitrofen-induced Hajime Takayasu , Paula Murphy

Journal of Pediatric Surgery (2010) 45, 2129–2135
www.elsevier.com/locate/jpedsurg
Embryonic Wnt gene expression in the nitrofen-induced
hypoplastic lung using 3-dimensional imaging
Hajime Takayasu a , Paula Murphy b , Hideaki Sato a , Takashi Doi a,c , Prem Puri a,c,⁎
a
The Children's Research Centre, Our Lady's Children's Hospital, Dublin, Ireland
Department of Zoology, School of Natural Sciences, Trinity College, Dublin, Ireland
c
School of Medicine and Medical Science and Conway Institute of Biomolecular and Biomedical Research,
University College Dublin, Dublin, Ireland
b
Received 22 January 2010; revised 24 June 2010; accepted 28 June 2010
Key words:
Congenital diaphragmatic
hernia;
Optical projection
tomography;
Whole-mount in situ
hybridization;
Wnt;
Nitrofen
Abstract
Purpose: Wnts have been reported to play a key role in the lung morphogenesis. We have previously
reported that pulmonary gene expression of Wnt2 and Wnt7b is downregulated on day 15 of gestation in
the nitrofen-induced congenital diaphragmatic hernia (CDH) model. However, the distribution pattern of
gene expression of Wnts in the very early lung development remains unclear. Optical projection
tomography (OPT) is a new technique for 3-dimensional imaging of small developing organs and gene
distribution combined with whole-mount in situ hybridization. We designed this study to investigate the
distribution pattern of Wnts gene expression in lung buds of nitrofen-induced CDH model using OPT.
Methods: Embryos from normal and nitrofen-treated dams were harvested on embryonic day 10 (E10),
and divided into controls and nitrofen group, respectively. Whole-mount in situ hybridization to detect
transcripts of Wnt2 and Wnt7b was performed, analyzed, and reconstructed using OPT.
Results: The expression of Wnt2 transcripts was detected in the lung bud mesenchyme and markedly
diminished in nitrofen group compared to controls, whereas Wnt7b transcripts were expressed in the
mesoderm of bronchi and the lung bud with no detectable difference between 2 groups.
Conclusion: We provide evidence for the first time that Wnt2 expression is downregulated at lung bud
stage in the nitrofen model. Optical projection tomography is potentially a useful approach to visualize
both gene expression and morphology during very early stages of lung development.
© 2010 Elsevier Inc. All rights reserved.
Despite improved understanding of the pathophysiology
of congenital diaphragmatic hernia (CDH) and advances in
perinatal care, the prognosis of infants with severe CDH
remains poor [1,2]. Hypoplastic lung and persistent pulmo⁎ Corresponding author. The Children's Research Centre, Our Lady's
Children's Hospital, Dublin 12, Ireland. Tel.: +353 1 4096420; fax: +353 1
4550201.
E-mail address: [email protected] (P. Puri).
0022-3468/$ – see front matter © 2010 Elsevier Inc. All rights reserved.
doi:10.1016/j.jpedsurg.2010.06.046
nary hypertension are the principle causes of the high
morbidity and mortality in infants with CDH [1-3]. The
affected infants show developmental retardation of lung,
including fewer bronchial branches and alveoli, and retardation of alveolar development [4-6]. The pathogenesis of
pulmonary hypoplasia associated with CDH is not fully
understood. In the nitrofen-induced CDH animal model,
some investigators reported that nonmechanical factors
directory mediated by nitrofen play a significant role in the
2130
pathogenesis of lung hypoplasia, causing abnormal regulation of early lung development [7,8]. Keijzer et al [8]
proposed the dual-hit hypothesis to explain the observations
on pulmonary hypoplasia in this model. This hypothesis
proposes that the early retardation in lung development that
occurs before the development of the diaphragmatic defect is
caused by nitrofen, whereas the late-gestational increase in
lung hypoplasia is caused by mechanical compression from
herniated viscera. Recently, a role for Wnt signaling in lung
development was suggested by the observation that several
Wnt genes are expressed in the developing lung mesenchyme
and/or epithelium [9-11]. Studies in several model systems
have shown that Wnt signaling regulates multiple steps in
organogenesis, including cell proliferation, differentiation,
and lineage specification [12,13]. In particular, Wnt gene
products have been shown to act as mediators of epithelialmesenchymal interactions in the developing lung [13] where
Wnt7b is required for normal lung mesenchymal proliferation
in a narrow window of development before E15.5 in mice
[14,15]. Wnt7b inactivation decreased airway branching,
caused pulmonary hypoplasia, and decreased lung smooth
muscle [14,15]. Wnt 7b null mutant mice died of respiratory
failure at birth [14]. Other Wnt genes, such as Wnt2 and
Wnt5a are expressed in the mesenchyme of the developing
lung [16,17]. Wnt2 is expressed in mesoderm during
branching morphogenesis and approximately 50% of Wnt2
knockout mice die soon after birth of respiratory failure [18].
We previously reported that the expression level of Wnt2
and Wnt7b genes was altered in nitrofen-treated hypoplastic
lungs compared to normal lungs at early-gestational stages
(E15), providing evidence that the Wnt signaling pathway is
downregulated in the early stages of lung development [19].
However, the level of expression is only one aspect of the
potential patterning activity of signaling molecules, with
the precise spatiotemporal distribution of transcripts within
the emerging tissue playing an important role in morphogenetic outcomes. Therefore, we hypothesized that not only the
amount of expression but also the distribution of expression
of Wnt7b and Wnt2 may be affected in the hypoplastic lungs
of nitrofen-treated embryos. To test this hypothesis, we
evaluated the 3-dimensional (3D) distribution pattern of Wnt
genes expression in normal and hypoplastic lungs in the very
early stages of lung development using a recently developed
technique, optical projection tomography (OPT).
Optical projection tomography is a new approach for 3D
imaging of small biologic specimens [20]. The principals of
OPT is similar to that of x-ray computed tomography. It
fills an imaging gap between magnetic resonance imaging
and confocal microscopy, being most suited to specimens
that are from 1 to 10 mm across [20,21]. In addition, it is the
only technique that can record both colorimetric and
fluorescent staining simultaneously with tissue morphology
and therefore allows imaging of the 3D distribution of gene
expression patterns as revealed by in situ techniques
[20-22] to detect either RNA (in situ hybridization) or
protein (immunolabeling). Application of the technique to
H. Takayasu et al.
mutant or disrupted embryos has the added potential of
revealing anatomical abnormalities in detail including
subtle aspects of the phenotype not readily detected from
sectioning specimens where sections may be lost or
distorted or simply may not reveal 3D details [20]. Thus,
we performed OPT to study the precise expression pattern
of Wnt2 and 7b genes in the context of the normal and
disrupted developing lung in the nitrofen-induced CDH
animal model with a view to elucidating the molecular
mechanisms involved in lung hypoplasia.
1. Materials and methods
1.1. Animal model
Fetal pulmonary hypoplasia with coexistent diaphragmatic hernia was induced by gavaging time-mated pregnant
CD-1 mice [23]. Females caged with males were checked
each morning for the presence of a vaginal plug as an
indication of mating and the time designated as 0.5 days
postcoitum (dpc). At 8.5 dpc, 25-mg nitrofen (WAKO
Chemical, Osaka, Japan) dissolved in olive oil was given as a
single dose via a stomach tube under anesthesia. In control
animals, the same dose of olive oil was given without
nitrofen. Cesarean delivery was performed on 10.5dpc of
gestation under general anesthesia. Embryos were harvested
by laparotomy. Twenty nitrofen-treated embryos and 20
control embryos were subjected to OPT study (10 for Wnt2,
10 for Wnt7b for each group).
Embryos were collected in ice-cold phosphate buffered
saline (PBS), separated from surrounding tissue and fixed
overnight in 4% paraformaldehyde in PBS at 4°C before
dehydration through a series of methanol washes (25%, 50%,
75% in PBS and 100%) and storage at −20°C. The
Department of Health and Children approved all the animal
experiments (reference B100/3697) under the Cruelty to
Animal Act, 1876, as amended by European Communities
Regulations 2002.
1.2. In situ hybridization
Antisense RNA probes to detect gene expression and
control sense probes were generated by in vitro transcription
from linearized plasmid DNA containing complementary
DNA clones corresponding to Wnt7b and Wnt2 genes. The
gene sequences used as probes correspond to nucleotide 93 to
1581 on genbank sequence NM_009528 (Wnt7b) and
nucleotide 19 to 1493 on genbank sequence BC026373
(Wnt2), both kind gifts of A. McMahon, Harvard. RNA probes
labeled with digoxigenin (DIG)-uridin triphosphate (UTP)
were synthesized using the DIG RNA labeling mix and
appropriate RNA polymerases (Roche Molecular Biochemicals, Mannheim, Germany) according to the manufacturer's
recommendations. Sense control probes were routinely used in
Embryonic Wnt gene expression
each experiment. In situ hybridization was carried out largely
according to Wlikinson [24] with minor adjustments.
Hybridization times were extended to 2 to 3 days at 65°C.
Final posthybridization washes were in 0.2% saline-sodium
citrate (SSC)/0.1% cholamidopropyl dimethlammonio porpanesulfonate (CHAPS), 3 times for 20 minutes at 65°C. The
digoxigenin-labeled RNA was localized within the embryonic
tissue using an alkaline phosphatase conjugated antidigoxigenin antibody (Roche Molecular Biochemicals, Mannheim,
Germany). Colorimetric detection of alkaline phosphatase
activity was performed with substrate solution containing 225
μg/mL of nitroblue tetrazolium and 87.5 μg/mL of 5-bromo-4chloro-3-indolyl phosphate. The level of stain was monitored
carefully, and the reaction was stopped by placement in PBS
and fixation in 4% paraformaldehyde in PBS when the color
had developed to a clearly visible level and background
staining was still low, usually less than 2 hours. Note that very
darkly stained specimens are not suitable for OPT scanning as
light will not be transmitted through the tissue. Specimens were
photographed before OPT scanning.
2131
(Fig. 1B and D). Wnt7b expression in the developing lung
buds is clear (Fig. 1A arrow). Wnt7b expression was also
observed in the future brain, in the developing spinal cord, and
the optic and otic vesicles. Superficially, Wnt2 expression was
observed in a more widespread and diffuse domain in the
region of the developing lungs (Fig. 1C, arrow) as well as in
the heart, septum transversum, allantois, umbilicus, and in the
lateral mesenchyme of the body wall between the fore and
hind limb buds. Although expression of Wnt7b appeared to be
similar in nitrofen-treated and control embryos, the expression
of Wnt2 appeared to be less extensive in some nitrofen-treated
embryos (Fig. 1C and D).
The OPT scanning of the specimens and 3D reconstruction
of the expression patterns allowed a more thorough
examination. Fig. 2 shows reconstructions that allow 3D
representations of the lung-associated patterns of both Wnt7b
and Wnt2 in normal embryos. These images are a combination
of volume representations (left) highlighting the expression
domains in white/light gray and surface representations at
different threshold levels showing the expression domains
(center) in isolation and the surface of the whole embryo
1.3. Optical projection tomography
Hybridized and photographed specimens were rinsed
briefly in distilled water and embedded in an appropriate
orientation in molten (below 37°C) low melting point agarose
(1% in water). Once the agarose had solidified at room
temperature, blocks were cut, glued to a metal mount, and
dehydrated in methanol at room temperature over night.
Specimens were then optically cleared in benzyl alcohol/
benzyl benzoate (1:2) for a minimum of 5 hours and up to
2 days. The specimens were scanned as described by Sharpe
et al [20] with a prototype OPT scanner built and provided by
the MRC Human Genetics Unit, Edinburgh, installed in the
Zoology Department, Trinity College Dublin. A Q imaging
Retiga Exi camera was used to record images through a 360°
rotation of the specimen viewed through a Leica MZ FLIII
microscope with a plan 0.5× objective. Scans were performed
under visible light provided by a 20 W halogen lamp (brightfield channel) using 700 or 750 nm filters and under UV light
to capture autoflourescence from the specimen to represent
embryo morphology. The 3D reconstructions were performed
as described by Sharpe et al [20] using programs provided by
the Edinburgh Mouse Atlas Project. The 3D reconstructions
were viewed and analyzed using software (MA3DView and
MAPaint) also provided by Edinburgh Mouse Atlas Project.
Virtual sections shown here were captured directly from the
3D reconstructions using MAPaint.
2. Results
Superficial examination of whole mount in situ hybridized
embryos showed the overall expression patterns of Wnt7b and
Wnt2 in control (Fig. 1A and C) and nitrofen-treated embryos
Fig. 1 Whole mount in situ analysis of Wnt7b expression in
control (A) and nitrofen-treated (B) E10.5 embryos. Arrows
indicate the Wnt7b expression in the developing lungs. Wnt7b
was expressed in an increasing gradient from the proximal-to-distal
airway. (C) An E10.5 mouse embryo whole-mount stained for
expression of the Wnt2 gene. An arrow indicates the Wnt2
expression in the lung. (D) The Wnt2 transcripts were localized to
the outer layer. The Wnt2 expression was not detected in the lungs
of E10.5 nitrofen-treated embryos (arrow).
2132
H. Takayasu et al.
Fig. 2 Definition of Wnt7b (A) and Wnt2 (B) expression in control embryo. The isosurface has been rendered transparent so that the 3D shapes
of Wnt7b and Wnt2 mRNA expression patterns can be seen with in the embryo. (A) Strong and uniform Wnt7b expression was observed in the
developing lung and trachea (demarcated with pink line). Wnt7b expression was also observed in the future brain and in the developing spinal
cord. (B) Wnt2 expression was observed in the lateral side of the lungs and surrounding mesenchyme (demarcated with pink line).
(right). Using this visualization, the shape and size of the
expression domains of Wnt7b (Fig. 2A) and Wnt2 (Fig. 2B)
are clear; the domains of Wnt7b in the endoderm of the lung
buds and Wnt2 in the mesenchyme around the lung buds have
been outlined. This extraction of the patterns allows complex
3D domains to be viewed more readily.
The 3D computer reconstructions of scanned embryos
can also be rotated and virtually sectioned in any plane to
view the internal distribution of expressing cells within the
morphological context of the embryo. Figs. 3 and 4 shows
sections of normal (A-C) and nitrofen-treated embryos (DF) stained to reveal expression of Wnt7b (Fig. 3) and Wnt2
(Fig. 4). Coronal (B) and a series of transverse sections (C)
show expression (white/light gray) domains in the lung
buds for Wnt7b and mesenchyme surrounding the lungs for
Wnt2 in normal embryos. This visualization of the patterns
allows a more detailed comparison between treated and
control embryos. This approach showed that there was no
detectable difference in the distribution of Wnt7b expres-
sing cells between control and treated embryos. On the
other hand, nitrofen-treated embryos showed variability in
the expression domains of Wnt2, specifically in the lungassociated mesenchyme.
In 4 of 10 nitrofen-treated embryos, Wnt2 expression was
not detected in the lung mesenchyme (compare Fig. 4C and
F) although expression in other sites such as the lateral body
wall and the allantois are still clearly visible. In another
4 embryos, Wnt2 expression was very weak in the lungassociated mesenchyme, whereas in 2 of 10 nitrofen-treated
embryos, the expression level of Wnt2 was similar to that in
normal developing lungs.
3. Discussion
Despite intensive research, the molecular mechanisms
involved in lung hypoplasia associated with CDH are still
Embryonic Wnt gene expression
2133
Fig. 3 OPT analysis of the Wnt7b expression pattern of control (A, B, C) and nitrofen-treated embryos (D, E, F). Photomicrographs of
lateral view of whole mount in situ hybridized specimen of control embryo (A) and nitrofen-treated embryo (D). Virtual sections of control
embryo (B and C) and nitrofen-treated embryo (E and F) taken through the computer reconstructions after OPT scanning. (B) and (C) are
sections indicated by the lines in (A), and (E) and (F) are sections taken in the planes indicated by lines in (B). Arrows show the expression
domain of Wnt7b observed in the lungs.
unclear. According to the dual-hit hypothesis proposed by
Keijzer et al [8] to explain the nitrofen-induced animal model
of CDH, the early retardation in lung development that
occurs before the development of the diaphragmatic defect is
caused directly by nitrofen, whereas lung hypoplasia in late
gestation is caused by mechanical compression from
herniated viscera. To elucidate the mechanisms involved in
hypoplastic lung in the early stages of gestation, we focused
on Wnt7b and Wnt2.
Although the expression of Wnt7b and Wnt2 have been
previously described in the developing lung [14,15,17], the
detailed distribution of expressing cells in 3D had not
previously been shown. The 3D patterns shown here
emphasize the complementary expression of these 2 Wnt
signaling molecule encoding genes in neighboring tissues of
the endoderm and mesenchyme of the developing lungs.
Gene inactivation studies have implicated both genes in the
correct development of the lungs, although the link is clearer
for Wnt7b where knockout mice die of respiratory failure at
birth (16), whereas 50% of Wnt2 inactivated mice die soon
after birth, probably because of respiratory failure (18). It is
therefore likely that reciprocal Wnt signaling between these
adjacent tissues is needed to guide correct development.
The use of OPT allowed a more detailed study of potential
changes in the expression of these Wnt genes in response to
nitrofen exposure; specifically, 3D reconstruction allowed
the distribution of expressing cells to be analyzed. Because
these genes encode signaling molecules that influence the
development of surrounding cells, any changes in the
localization of signal production could have profound effects
on developmental processes.
The expression pattern of Wnt7b in all 10 nitrofen-treated
embryos was similar to the expression pattern observed in
the control group. This would suggest that nitrofen
influences the amount of gene product but not the population
of cells expressing Wnt7b gene; Wnt7b is still expressed at a
level detectable by in situ hybridization in the lung bud
epithelium. It is difficult to draw conclusions about the
significance of this finding because we do not know the
functional significance of different levels of Wnt7b transcripts. On the other hand, lack or reduced expression of
Wnt2 was observed in the lung bud mesenchyme of nitrofentreated embryos. Our results suggest that nitrofen altered the
localization of Wnt2 production during very early stages of
lung development. Although the precise role of Wnt2
expression in lung development is still unclear, these
findings indicate that the impact of nitrofen on lung
development may be at least in part mediated by alterations
in the local production of Wnt2 signal.
This study illustrates the use of OPT to analyze and
compare gene expression in an animal model of a
congenital abnormality and shows the potential to simultaneously reveal morphology and gene expression changes.
In this study, analysis was focused on early stages of
2134
H. Takayasu et al.
Fig. 4 OPT analysis of the Wnt2 expression pattern of control (A, B, C) and nitrofen-treated embryos (D, E, F). Photomicrographs of lateral
view of whole mount in situ hybridized specimen of control embryo (A) and nitrofen-treated embryo (D). Virtual sections of control embryo
(B and C) and nitrofen treated embryo (E and F) taken through the computer reconstructions after OPT scanning. (B) and (C) are sections
indicated by the lines in (A), and (E) and (F) are sections taken in the planes indicated by lines in (B). Arrows show the expression domain of
Wnt2 observed in the lungs. The OPT clearly visualized reduced Wnt2 expression in the lungs in nitrofen-treated embryo (F), whereas the
expression pattern of other sites were not changed.
gestation, 1.5 days after nitrofen treatment, with the view to
observing direct gene expression effects. It therefore
observed changes before the closure of the diaphragm
occurs. Because the anomalies that occur in CDH are
complex (lung hypoplasia, defect of diaphragm, herniation
of abdominal organ, and others), the technique of 3D
visualization of the developing organs will provide
additional information at later stages that may help to
elucidate the mechanisms involved, for example, during
formation of diaphragm. Although OPT holds great
potential to look at morphological changes at later stages
of development in the model, there are some challenges to
observing gene expression in significantly larger embryos.
This is related to the penetration of the molecular
components required for in situ hybridization in more
complex tissues. However, we have successfully visualized
expression in visceral tissues up to E13.5 in embryos where
the trunk has been dissected and imaged separately.
In conclusion, our findings demonstrate that OPT is a
powerful approach to visualizing gene expression and
morphology during very early stages of lung development.
This technique may provide insights into molecular
mechanisms involved not only in lung hypoplasia but also
in other anomalies occurring in the nitrofen-induced CDH
model. It has great potential for wider use in the analysis of
development in animal models of congenital abnormalities.
Acknowledgment
We thank Kristen Summerhurst for her help and guidance
with molecular techniques and in carrying out OPT. We
thank many members of the Edinburgh Mouse Atlas Project
team for computing tools, help, and advice.
References
[1] Thibeault DW, Haney B. Lung volume, pulmonary vasculature, and
factors affecting survival in congenital diaphragmatic hernia. Pediatrics 1998;101:289-95.
[2] Skari H, Bjornland K, Haugen G, et al. Congenital diaphragmatic
hernia: a meta-analysis of mortality factors. J Pediatr Surg 2000;35:
1187-97.
[3] Sweed Y, Puri P. Congenital diaphragmatic hernia: influence of
associated malformations on survival. Arch Dis Child 1993;69:68-70.
[4] Kitagawa M, Hislop A, Boyden EA, et al. Lung hypoplasia in
congenital diaphragmatic hernia. A quantitative study of airway,
artery, and alveolar development. Br J Surg 1971;58:342-6.
[5] Okazaki T, Sharma HS, McCune SK, et al. Pulmonary vascular
balance in congenital diaphragmatic hernia: enhanced endothelin-1
gene expression as a possible cause of pulmonary vasoconstriction.
J Pediatr Surg 1998;33:81-4.
[6] Sabharwal AJ, Davis CF, Howatson AG. Post-mortem findings in fetal
and neonatal congenital diaphragmatic hernia. Eur J Pediatr Surg 2000;
10:96-9.
Embryonic Wnt gene expression
[7] Guilbert TW, Gebb SA, Shannon JM. Lung hypoplasia in the nitrofen
model of congenital diaphragmatic hernia occurs early in development.
Am J Physiol Lung Cell Mol Physiol 2000;279:1159-71.
[8] Keijzer R, Liu J, Deimling J, et al. Dual-hit hypothesis explains
pulmonary hypoplasia in the nitrofen model of congenital diaphragmatic hernia. Am J Pathol 2000;156:1299-306.
[9] Shu W, Guttentag S, Wang Z, et al. Wnt/beta-catenin signaling acts
upstream of N-myc, BMP4, and FGF signaling to regulate proximaldistal patterning in the lung. Dev Biol 2005;283:226-39.
[10] Eberhart CG, Argani P. Wnt signaling in human development: betacatenin nuclear translocation in fetal lung, kidney, placenta, capillaries,
adrenal, and cartilage. Pediatr Dev Pathol 2001;4:351-7.
[11] Warburton D, Bellusci S, De Langhe S, et al. Molecular mechanisms
of early lung specification and branching morphogenesis. Pediatr Res
2005;57:26R-37R.
[12] Nusse R. Wnt signaling in disease and in development. Cell Res 2005;
15:28-32.
[13] Pongracz JE, Stockley RA. Wnt signalling in lung development and
diseases. Respir Res 2006;7:15.
[14] Shu W, Jiang YQ, Lu MM, et al. Wnt7b regulates mesenchymal
proliferation and vascular development in the lung. Development
2002;129:4831-42.
[15] Wang Z, Shu W, Lu MM, et al. Wnt7b activates canonical signaling in
epithelial and vascular smooth muscle cells through interactions with
Fzd1, Fzd10, and LRP5. Mol Cell Biol 2005;25:5022-30.
2135
[16] Li C, Hu L, Xiao J, et al. Wnt5a regulates Shh and Fgf10 signaling
during lung development. Dev Biol 2005;287:86-97.
[17] Levay-Young BK, Navre M. Growth and developmental regulation of
Wnt-2 (irp) gene in mesenchymal cells of fetal lung. Am J Physiol
1992;262:L672-83.
[18] Monkley SJ, Delaney SJ, Pennisi DJ, et al. Targeted disruption of the
Wnt2 gene results in placentation defects. Development 1996;122:
3343-53.
[19] Takayasu H, Nakazawa N, Montedonico S, et al. Down-regulation of
Wnt signal pathway in nitrofen-induced hypoplastic lung. J Pediatr
Surg 2007;42:426-30.
[20] Sharpe J, Ahlgren U, Perry P, et al. Optical projection tomography as a
tool for 3D microscopy and gene expression studies. Science 2002;
296:541-5.
[21] Sharpe J. Optical projection tomography as a new tool for studying
embryo anatomy. J Anat 2003;202:175-81.
[22] Miller SF, Summerhurst K, Runker AE, et al. Expression of Plxdc2/
TEM7R in the developing nervous system of the mouse. Gene Expr
Patterns 2007;7:635-44.
[23] Leinwood MJ, Tefft JD, Zhao J, et al. Nitrofen inhibition of pulmonary
growth and development occurs in the early embryonic mouse.
J Pediatr Surg 2002;37:1263-8.
[24] Wilkinson DG, Nieto MA. Detection of messenger RNA by in situ
hybridization to tissue sections and whole mounts. Methods Enzymol
1993;225:361-73.