Patterned epidermal cell death in wild-type and segment polarity mutant Drosophila embryos

Abstract

Programmed cell death plays an essential role in the normal embryonic development of Drosophila melanogaster. One region of the embryo where cell death occurs, but has not been studied in detail, is the abdominal epidermis. Because cell death is a fleeting process, we have used time-lapse, fluorescence microscopy to map epidermal apoptosis throughout embryonic development. Cell death occurs in a stereotypically striped pattern near both sides of the segment border and to a lesser extent in the middle of the segment. This map of wild-type cell death was used to determine how cell death patterns change in response to genetic perturbations that affect epidermal patterning. Previous studies have suggested that segment polarity mutant phenotypes are partially the result of increased cell death. Mutations in wingless, armadillo, and gooseberry led to dramatic increases in apoptosis in the anterior of the segment while a naked mutation resulted in a dramatic increase in the death of engrailed cells in the posterior of the segment. These results show that segment polarity gene expression is necessary for the survival of specific rows of epidermal cells and may provide insight into the establishment of the wild-type epidermal cell death pattern.

Full text

INTRODUCTION
Pattern formation during Drosophila embryogenesis requires
specific numbers of cells to be established in each tissue
primordia. The developing embryo must maintain a delicate
balance between cell division and cell death to generate the
appropriate number of cells. The patterns of cell division in
the Drosophila embryo is very well understood (Hartenstein
and Campos-Ortega, 1985; Foe, 1989). The proteins that
regulate the Drosophila cell cycle are also well characterized
(reviewed by Edgar and Lehner, 1996). However, little is
known about the mechanism for regulating the global mitotic
pattern, which gives rise to regions of synchronously
dividing cells, known as mitotic domains. It is believed that
these mitoses produce an excess of cells required for a viable
embryo and that programmed cell death, or apoptosis, is
necessary to remove these extraneous cells (reviewed by
Jacobson et al., 1997). Cell death is essential for embryo
survival and a number of genes that function in the apoptotic
pathway are known. Embryos that are deficient for cell death
fail at head involution (Grether et al., 1995) and show signs
of hypertrophy in the central nervous system (White et al.,
1994). Cell death also plays a role in repairing pattern
defects. Mis-patterned embryos with increased numbers of
abdominal cells, are repaired by the elimination of excess
cells by apoptosis (Namba et al., 1997). There are two major
gaps in our understanding of cell death in Drosophila
embryos: (1) what is the wild-type pattern of cell death and
(2) what are the signals that initiate the apoptotic
programme?
One tissue where apoptosis is present, but has not been
studied in detail, is the embryonic abdominal ectoderm,
which gives rise to the highly patterned larval cuticle with
its characteristic bands of ventral denticles and dorsal hairs.
The arrangement of these structures is generated by the
precise expression of patterning genes in the embryo.
Segment polarity genes play a key role in establishing the
fate of the cuticle secreting cells through cell signaling
interactions between neighboring rows of cells in the
ectoderm (reviewed by DiNardo et al., 1994; Martinez Arias,
1993). Establishing the appropriate numbers of cells in the
epidermis is believed to be essential to produce a normal
cuticle pattern. Too few or too many cells may result in
abnormally sized structures, pattern disruption and potential
lethality. Mutations in segment polarity genes lead to
shortened embryos and increased epidermal cell death
(Klingensmith et al., 1989; Perrimon and Mahowald, 1987).
Given the intimate role of segment polarity genes in
establishing embryonic epidermal fates, it is of interest to
determine the role segment polarity genes play in initiating
the apoptotic cell death that leads to the pruning of excess
cells in the epidermis.
3427
Development 125, 3427-3436 (1998)
Printed in Great Britain © The Company of Biologists Limited 1998
DEV7637
Programmed cell death plays an essential role in the
normal embryonic development of Drosophila
melanogaster. One region of the embryo where cell death
occurs, but has not been studied in detail, is the abdominal
epidermis. Because cell death is a fleeting process, we have
used time-lapse, fluorescence microscopy to map epidermal
apoptosis throughout embryonic development. Cell death
occurs in a stereotypically striped pattern near both sides
of the segment border and to a lesser extent in the middle
of the segment. This map of wild-type cell death was used
to determine how cell death patterns change in response to
genetic perturbations that affect epidermal patterning.
Previous studies have suggested that segment polarity
mutant phenotypes are partially the result of increased cell
death. Mutations in wingless, armadillo, and gooseberry led
to dramatic increases in apoptosis in the anterior of the
segment while a naked mutation resulted in a dramatic
increase in the death of engrailed cells in the posterior of
the segment. These results show that segment polarity gene
expression is necessary for the survival of specific rows of
epidermal cells and may provide insight into the
establishment of the wild-type epidermal cell death pattern.
Key words: Apoptosis, Epidermis, Segment polarity genes,
Drosophila melanogaster, wingless, naked
SUMMARY
Patterned epidermal cell death in wild-type and segment polarity mutant
Drosophila
embryos
Todd M. Pazdera1, Prem Janardhan2 and Jonathan S. Minden1,*
1Department of Biological Sciences and the Center for Light Microscope Imaging and Biotechnology and 2Department of
Computer Science, Carnegie Mellon University, Pittsburgh, PA 15213, USA
*Author of correspondence (e-mail: minden@cmu.edu)
Accepted 27 May; published on WWW 6 August 1998

3428
In order to gain a better understanding of the role of
apoptosis in epidermal patterning, a map of the cell death in
wild-type embryos was generated. Apoptosis first appears at
embryonic stage 11 and persists throughout the remainder of
embryonic development (Campos-Ortega and Hartenstein,
1985). Previous studies have shown that apoptosis is a
relatively short-lived process where the dead cells are
engulfed either by neighboring cells or migrating
macrophages. Images of embryos that were fixed and stained
for cell death hint at a segmentally repeated cell death pattern,
but these snapshots are unable to capture the full pattern of
this dynamic process. We have used time-lapse microscopy to
record the spatial and temporal changes in cell death. These
recordings provide a precise mapping of cell death over time,
which clearly shows that epidermal cell death is repeated
segmentally. In order to determine the location of these cell
deaths within the segment, apoptosis was mapped with respect
to segment borders and segment polarity gene expression in
live embryos. We find that cell death in the epidermis occurs
predominantly near the segment borders, near or in engrailed
(en)-expressing cells and to a lesser extent in the middle of
the segment.
The wild-type cell death map was used to compare the cell
death profiles of a number of segment polarity mutant embryos.
Mutations in the wg signaling pathway lead to the demise of
several rows of cells in the anterior-most portion of the
segment. Mutations in the segment polarity gene naked (nkd)
also lead to a dramatic increase in apoptotic death in en-
expressing abdominal cells. These results show that segment
polarity mutants have an altered cell death pattern in a
stereotypical manner, suggesting that these genes are important
in determining epidermal cell survival.
MATERIALS AND METHODS
Stocks
Oregon R and white Canton S were used as wild-type stocks. A
temperature sensitive allele of wingless (wg1-12) (Nüsslein-Volhard
and Wieschaus, 1984), was provided by Steve DiNardo. The wg1-12
allele was maintained with a blue balancer chromosome carrying an
eve:lacZ construct, which was used to determine homozygous mutant
embryos. The armadillo temperature sensitive allele, armH8.6,was
provided by Norbert Perrimon. Null mutants of gooseberry
(Df(2R)gsb) (Li and Noll, 1993) and naked (nkd2) (Perrimon, 1989)
were provided by the Bloomington Stock Center.
Monitoring cell death, gene expression and segment
borders in vivo
Stage 3 embryos were collected, mounted and injected as described
previously (Namba et al., 1997). Cell death was visualized by
injecting the embryos with a solution of 0.5 mg/ml acridine orange
(AO) in PBS. To visualize lacZ expression, embryos were injected
with a 1.5 mM solution of RGPEG (Minden, 1996). Segment
borders were visualized by injecting embryos in the intervitelline
space with RGPEG which fluoresces brightly when it accumulates
in crevices on the embryo surface. Fluorescent, time-lapse
recordings of injected embryos were generated starting at late stage
11. Each recording consisted of stacks of 4-6 images taken 5 µm
apart, taken at 5-8 minute intervals for up to 12 hours. Embryo
collection, maturation and recording temperatures are detailed in the
Results section.
Identifying homozygous mutant embryos
Two methods were used to identify the genotype of mutant embryos:
(1) mutants were identified by inspection of the cuticle pattern of
stage 17 embryos. Cuticle preparations of injected embryos were
performed in situ on the coverslip to maintain the relative position of
the embryos to correlate cuticle phenotype with time-lapse images.
Late stage embryos were injected in the intervitelline space with a
4:1 mixture of acetic acid: glycerin, incubated on a heating block at
65˚C for 30 minutes and then incubated overnight at room
temperature. The halocarbon oil was scraped from the coverslip,
covered with 1:1 CMC10 (Masters chemical):lactic acid, covered
with a second coverslip and incubated overnight at 60˚C. (2) wg
mutant embryos were identified by injecting stage 4 embryos with
RGPEG into the syncytial cytoplasm to detect eve:lacZ expression
from the balancer chromosome. Homozygous wg embryos lacked
eve:lacZ expression.
Generation of cell death maps in wild-type embryos
The position of apoptotic cells when they first become AO positive
was determined. For each time point of the time-lapse recording, 2-
3 optical sections, taken 5 µm apart, were superimposed
electronically. These images were contrast enhanced by applying a
spot detection function, consisting of a scanning comparison box of
the dimensions of a nucleus, which identifies small, bright circular
objects – the fluorescent nuclei of apoptotic cells. This spot
detection routine allows one to easily identify AO-positive cells and
follow them over time. To generate a composite image showing the
lifetime of each dying cell, 5-10 sequential spot detected images
were then overlaid using a projection function (Deltavision, Applied
Precision Issaqua, WA). This time projection method produces a
tracing of the dying cell over time. The initial position of each dying
cell was mapped onto the appropriate stage embryo (Hartenstein,
1993).
In situ hybridization
rpr cDNA was isolated and labeled as described previously (Namba
et al., 1997). Embryos were processed as whole mounts as described
by Tautz and Pfeifle (1989). Complexed probe was detected using an
alkaline-phosphatase-conjugated antibody against the digoxigenin
dUTP (Boehringer Mannheim). For double labeling experiments with
rpr in situ probe and EN antibody, the in situ protocol was carried out
first. The antibody staining was performed after dehydration with
ethanol without xylene clearing.
Antibody staining
Embryos were processed as described by Bomze and Lopez (1994)
using a anti-en/inv mouse monoclonal antibody diluted 1:1000 to
visualize EN. Primary antibodies were visualized by using a goat anti-
mouse secondary antibody and the ABC kit from Vector Labs.
TUNEL
Embryos were dechorionated and fixed for 20 minutes in 4%
paraformaldehyde in PBS:heptane 1:1. Embryos were devitellinized
by shaking in 80% ethanol and gradually rehydrated in PBS.
Embryos were then washed once in reaction buffer consisting of 2.5
mM CoCl2and 1×terminal transferase (TdT) buffer (Boehringer
Mannheim). For the terminal transferase reaction, embryos were
incubated for 3 hours at 37°C in 2.5 mM CoCl2, 1×TdT buffer, 5
µM dUTP, 5 µM CY3-dUTP (Amersham), 50 units TdT. Embryos
were washed three times for 10 minutes in PBS plus 0.3% Triton X-
100 and then processed with the ABC kit following standard
protocols. After the reaction, embryos were washed four times for 15
minutes in PBS plus 0.3% Triton X-100 and probed with the EN
antibody as described above. Embryos were viewed using a Texas
Red filter to detect Cy3-labeled DNA fragments in the dying cells.
T. M. Pazdera, P. Janardhanand J. S. Minden

3429
Drosophila
epidermal cell death
In these fluorescent images the en HRP-labeled cells appeared black
on a gray background.
RESULTS
In vivo mapping of cell death in wild-type
embryos
Embryonic cell death is a very dynamic process
that starts during late stage 11 and continues
through hatching of the embryo into a larva.
Given the short lived nature of these apoptotic
events, it is essential to use time-lapse
microscopy to map cell death. Apoptosis can be
monitored in living embryos by using the
fluorescent vital dye acridine orange (AO)
(Abrams et al., 1993). AO is stored in the
lysosomes of living cells in a quenched state until
the cell undergoes apoptosis. When a cell dies by
apoptosis the AO is released from the lysosome
and enters the nucleus where it binds to duplex
DNA and becomes highly fluorescent (Delic et
al., 1991). Time-lapse, fluorescence microscopy
of apoptosis allows one to gain a complete picture
of the spatial and temporal locations of all the cell
deaths in a single embryo. To ensure that only
epidermal cell death was scored, focal planes
within 10 µm of the embryo surface were used for
analysis. Fig. 1A shows a series of images from
a wild-type embryo that has been injected with
AO. Dying cells appear as small white spots in
the black background. Scavenger cells or
macrophages also become fluorescent after they
engulf dying cells.
There were two problems associated with
generating the cell death maps: the persistence of
dead cells and interference from macrophages that
engulf dying cells. The cell death map is intended
to show the position of a cell when it first becomes
AO positive. However, the AO-positive cells can
persist in the epithelium up to 45 minutes before it
is engulfed by neighboring cells or macrophages.
This means that the same apoptotic cell will appear
in up to 10 sequential images of a time-lapse
recording. To prevent counting the same cell more
than once, one must track each apoptotic cell over
time. Tracking apoptotic cells is made more
complicated by morphogenetic movement and
engulfment by neighboring cells. To track AO-
positive cells we used spot detection software and
multiple time-point projections. The spot detection
software identified all of the AO-positive cells in
each image (Fig. 1B). Overlaying sequential spot-
detected images revealed a tracing of AO-positive
cells over time (Fig. 1C). The first point on this
tracing provided the map location of when the each
cell first became AO-positive and subsequent time-
points in the track were used to avoid re-mapping
the same cell. Tracking the apoptotic cells over
time allowed us to follow these cells during
germband retraction and dorsal closure, and their
movement as they are engulfed by macrophage or neighboring
cells.
The second problem with cell death mapping was that
highly motile macrophages became fluorescent when they
engulfed dying cells and thus complicated the images.
Although macrophage are not as abundant in the ectoderm as
Fig. 1. Cell death mapping in wild-type embryos. (A) A series of 5 consecutive
fluorescence, time-lapse images of an AO-injected wild-type embryo mounted
laterally. Images were taken at 6 minute time intervals. Dying cells appear as
small white spots in the dark background and macrophages appear as larger
irregular shape spots. In this and all the remaining figures, except Fig. 4, anterior
is to the left and dorsal is up. The autofluorescent yolk has been overlaid with a
light-gray mask. (B) The same series of images as shown in A after being
processed with the spot detection software. Each spot in these images represents
either a dying cell or macrophage. (C) An overlay of the images from the five
successive time points shown in B. Many of the dying cells move with the
surrounding tissue, over time, forming a line of spots (posterior two arrows),
while other dying cells move only slightly over time (anterior two arrows).

3430
in the head and central nervous system, some are present. The
macrophages can be distinguished from the dying cells
because they are highly motile and larger than the dying cells
and form paths that are not continuous or switch direction in
the time-projected images. Macrophages can also be
distinguished by the fact that they often have numerous bright
spots within them, representing several engulfed apoptotic
nuclei. These characteristics of macrophage have been
described previously in fixed embryos (Tepass et al., 1994).
Creating tracks of AO-positive
objects allowed us to create a
spatiotemporal map of the
patterns of cell death in the
embryonic epidermis.
In order to generate relational
cell death maps, a variety of
markers were used to indicate
specific regions of the embryo.
en:lacZ expression, which can be
detected in living embryos by
RGPEG injection (Minden,
1996), was used to indicate the
posterior margin of the
abdominal segments. Time-lapse
recordings were made of en:lacZ
embryos coinjected with AO and
RGPEG (Fig. 2). Approximately
three-quarters of the dying cells
appeared in or immediately
adjacent to the en stripe. The rest
of the apoptotic nuclei were
located in the middle of the
segment. There was also an
apparent clustering of apoptotic
nuclei at specific locations along
the dorsal-ventral axis (Fig. 2E).
Examination of multiple embryos
showed that the pattern of cell
death in each embryo was similar, however the number of
dying cells in each embryo and each segment varied from
embryo to embryo (Fig. 2F). This variation indicates a
plasticity in development, where each developing embryo can
eliminate different numbers of cells depending on its own
developmental circumstances. Approximately 12-16 cells die
in the lateral region of each epidermal segment during stages
12-14. After stage 14, cell death in the abdominal ectoderm
diminished.
T. M. Pazdera, P. Janardhanand J. S. Minden
Fig. 2. Mapping cell death relative to en:lacZ
expression. en:lacZ embryos were coinjected with
AO and RGPEG. (A-C) Images from a time-lapse
recording of an embryo at stage 12, 13 and 14;
red shows en expression and green shows AO-
positive cells. Using these landmarks and the spot
detection procedure, maps of cell death spanning
stages 12 to 14 were created. (D) A map of dying
cells relative to en expression (the red stripes)
prior to and during stage 12, superimposed on a
sketch of a stage 12 embryo (Hartenstein, 1993).
(E) A map of dying cells relative to en expression
(the red stripes) during stages 13 and 14,
superimposed on a sketch of a stage 14 embryo
(Hartenstein, 1993). Arrow, dorsal cluster;
arrowhead, mid-lateral cluster; concave
arrowhead, ventrolateral cluster) (F) Comparison
of cell death in three wild-type embryos at stage
13. Shown here are the results from three separate
cell death mapping experiments using different
embryos from that in A-E. Each colour represents
a different embryo.
Fig. 3. Mapping cell death relative to segment borders. Embryos were injected with AO into the
syncytial cytoplasm (green) and then injected with RGPEG into the intervitelline space to mark the
segment borders (red). The large shaded area masks the signal due to yolk autofluorescence.
(A-C) Images from a time-lapse recording of an embryo at stage 12, 13 and 14. (D) A map of
abdominal epidermal cell deaths during stages 11-12. (E) A map of abdominal ectodermal cell death
during stages 13-14. Arrows and arrowheads indicate the three clusters of death along the D-V axis.

3431
Drosophila
epidermal cell death
Since β-galactosidase is a very stable protein, the early broad
en expression domain remained at later stages and exaggerated
the width of en expression. To corroborate the en:lacZ data,
apoptosis was mapped with respect to the segment borders,
which were visualized by injecting the embryos in the
intervitelline space with RGPEG, which is cleaved by an
uncharacterized β-galactosidase activity (Minden, 1996) (Fig.
3). Approximately three-quarters of the dying cells mapped
adjacent to both sides of the segment border while the
remainder were in the middle of the segment. In addition to the
segmentally repeated cell death pattern, the majority of AO-
positive cells appeared in roughly three clusters along the
dorsal ventral axis (Fig. 3E).
Cells die in stripes in the ventral ectoderm
The limited axial resolution of the microscope prevents one
from monitoring all dying cells throughout the embryo in a
single recording. In order to map ectodermal apoptosis
adjacent to the ventral midline, time-lapse recordings were
made of ventrally oriented wild-type embryos. Intervitelline
RGPEG recordings revealed that two rows of cells on either
side of the segment border died in each segment of the ventral
ectoderm during stages 12-14 (Fig. 4). There were
approximately 6-8 dying cells in each row per hemi-segment.
Similar to the lateral ectoderm, the precise location and number
of dying cells was variable.
Accounting of cell death figures per segment
Quantitative analysis of the number of cell deaths per
segment showed a definite bias toward apoptosis at the
segment boundaries. We counted the number of cell deaths
occurring within the 2-3 cells either side of the segment
boundary and compared this with 5-6 cells in the middle of
the segment. Considering that a segment is 10-12 cells across,
this accounting partitions the segment into two equal-sized
groups; the segment border cells and mid-segment cells. The
segment border cells included the en cells plus one or two
rows of cells anterior and two or three cell rows posterior to
the en stripe. Analysis of four time-lapse recordings of wild-
type embryos, where eight segments were scored per embryo,
showed that 73% of cell death occurred in the segment border
cells and the remaining 27% occurred in the mid-segment
cells.
TUNEL labeling in the epidermis
The TUNEL method was used to label dying cells in fixed
embryos to corroborate the in vivo results (White et al.,
1994). The embryos were double labeled for en expression
Fig. 4. Cell death in the ventral ectoderm. (A-C) Time-lapse images
of a ventrally oriented embryo doubly injected with cytoplasmic AO
and intervitelline RGPEG, at stage 12, 13 and 14. Macrophages
appear as large cells predominantly along the ventral midline. Some
of the macrophages appear to have taken up some of the intervitelline
RGPEG as well as AO-positive cell bodies and appear as yellow cells
containing both fluorophores. Anterior is to the left.
Fig. 5. Comparison of AO-positive cell death pattern to TUNEL and
rpr-positive cell death patterns. (A-D) Images of mid stage 12, late
stage 12, stage 13 and stage 14 embryos. The embryos have been
double labeled for TUNEL (bright white spots) and en expression
(black stripes). The bracket in D indicates the ventral cell death. (E-
H) rpr expression (blue) and EN protein (brown) in wild-type
embryos at stage 11, mid stage 12, stage 13 and stage 14,
respectively. The bracket in H indicates out of focus rpr expression
in the ventral ectoderm and CNS. The various types of arrow and
arrowheads indicate the same D-V clusters of apoptotic cells as seen
in the preceding Figs.

3432
to highlight segment boundaries. TUNEL-positive nuclei
were located throughout the segment and exhibited
clustering along the dorsoventral axis similar to the in vivo
results (Fig. 5A-D). The number of TUNEL-positive nuclei
at a single time point was similar to that seen in individual
AO images except in the ventral epidermis where there were
fewer TUNEL-positive nuclei. The earliest cell deaths, at
stage 12, occurred more frequently in the mid-segment cells
(Fig. 5A). However, when scoring for cell death over time,
we found about two-thirds of the dying cells were located in
the segment border cells; one-third of the apoptotic nuclei
were seen in the mid-segment, where 12 embryos between
stages 12-14 were analyzed. Considering that TUNEL
labeling is a static technique, we cannot determine if the
location of the dying cells in these images is the precise
position at which the cell first becomes apoptotic. In the
time-lapse AO recordings, apoptotic cells that originated in
the en:lacZ stripes were seen to move anteriorly relative to
the en stripe, indicating the engulfment of the dead cells by
their anterior neighbors. Epidermal cells have been shown
previously to engulf dead neighbors (Tepass et al., 1994).
The engulfment of the segment border cells by their mid-
segment neighbors could explain the relative increase in mid-
segment TUNEL-positive cells since TUNEL also labels
engulfed nuclei. These results demonstrate that programmed
cell death occurs in a stereotypical pattern in the lateral
ectoderm.
Cells adjacent to the epidermal segment border
express
rpr
Previous experiments have indicated that expression of rpr
correlates with the pattern of dying cells in the embryo and
precedes AO signal by 2 hours (White et al., 1994). In situ
hybridization to rpr mRNA was performed in an attempt to
correlate rpr expression with the in vivo cell death maps. rpr
expression at early stage 11 was in large patches in the ventral
epidermis (Fig. 5E). The number of cells expressing rpr at
stage 11 was greater than the number of AO-positive cells that
were seen throughout epidermal development. Lack of
correlation between the number of rpr-expressing cells and
AO-positive cells at this stage can be explained in two ways.
(1) Some cells in the epidermis that express rpr at stage 11
do not become AO-positive, but still apoptose. (2) Some of
the stage 11 rpr-expressing cells do not enter apoptosis and
become AO-positive. We prefer the latter explanation, given
that the AO and TUNEL results correlate. At late stage 12
and stages 13-14 the expression of rpr in the epidermis
correlated fairly well with the cell death pattern observed
with AO (Fig. 5 F-H). Quantifying these results showed a
good spatial correlation between later rpr expression and AO-
positive cells both in the lateral ectoderm. In the lateral
ectoderm during late stage 12 to stage 14, approximately two-
thirds of the rpr-positive cells were in the segment border
region; 30 embryos were scored. Although relative numbers
of rpr-positive and AO-positive cells were similar at similar
time points, an exact quantitative comparison cannot be
performed due to the static nature of the rpr analysis.
Previous results have suggested that rpr expression precedes
cell death by 2 hours (White et al., 1994), however our results
show a better correlation with AO data at similar
developmental stages.
Cell death in segment polarity mutants of the
wg
class
Since the pattern of cell death correlated with segment polarity
gene expression, we examined the role that these genes play in
establishing the abdominal epidermal cell death pattern. Strong
mutations in segment polarity genes lead to the production of
larvae that are smaller than wild-type (Perrimon and Mahowald,
1987; Klingensmith et al., 1989). It has been proposed that this
reduction in larval size is a result of increased cell death in the
embryo. If there is extra cell death in segment polarity mutants,
then the segment polarity genes must play a role in maintaining
epidermal cell viability. To investigate this further we examined
apoptosis in segment polarity mutants.
The development of the epidermis can be divided into two
phases. Signaling between WG and EN cells is required during
stages 9 and 10 to stabilize segment polarity gene expression
throughout the segment. During stages 11 and 12, WG
signaling is required to establish cell fates throughout the
segment (Bejsovec and Martinez Arias, 1991; DiNardo et al.,
1994). Time-lapse analysis of strong alleles of wg,en and arm
showed extensive cell death throughout the abdominal
epidermis as well as other embryonic tissues (data not shown).
Since these mutations altered epidermal development early in
the stabilization phase, we analyzed cell death in temperature
sensitive mutants to monitor the effects of loss of WG and
ARM function during either the stabilization phase or the fate
specification phase.
wg1-12 (referred to as wgts) produces functional WG at the
permissive temperature of 17°C. At the non-permissive
temperature, >20°C, WG protein fails to be secreted
(Gonzalez et al., 1991). wgts embryos were collected at the
permissive temperature and syncytial embryos were injected
with AO. The embryos were then incubated at the permissive
temperature until stage 9, approximately 4.5 hours after
injection. The embryos were then shifted to the non-
permissive temperature of 25°C, and time-lapse recordings
were made 2 hours later at 22°C. A series of time-lapse images
from such a movie is shown in Fig. 6A-C. Massive cell death
was seen throughout the embryo. The most obvious increase
in cell death was seen at the tip of the germband and in the
epidermis. It is important to note that although there was a
dramatic increase in cell death throughout the epidermis,
many regions of the epidermis exhibited little or no cell death.
Under these conditions, WG secretion was inhibited during
the stabilization phase of epidermal patterning, which affects
overall patterning of the epidermis. The widespread cell death
was likely to be an indirect effect of general pattern disruption
and not a direct result of the WG signal promoting cell
survival.
In order to determine if WG plays a role in providing a
trophic signal to specific cells, wgts embryos were shifted to
the non-permissive temperature at stage 11, during the fate
specification phase of epidermal development. In order to
examine cell death during this stage of epidermal
development, wgts embryos were injected with AO and
incubated at the permissive temperature until stage 11, 6.5
hours after injections and then time-lapse recordings were
immediately started at the non-permissive temperature.
Under these conditions, there was a significant increase in
cell death that appeared as stripes about 2-3 cells in diameter
in each segment during stages 12-14 (Fig. 6D-F).
T. M. Pazdera, P. Janardhanand J. S. Minden

3433
Drosophila
epidermal cell death
Intervitelline RGPEG injections showed that the dying cells
were located in the anterior half of the segment near the
segment border (Fig. 6G). Given the abundance and
concentration of cell deaths in these mutants, we were unable
to perform an exact time-lapse quantitative analysis using the
tracking method. However, a comparison of the wild-type and
mutant embryos at the same stage showed a 5-fold increase
in the number of dying cells in the anterior of the segment.
The level of cell death in the remainder of the segment
appeared to be unaffected. The embryos were allowed to
develop to late stage 17 and their cuticle patterns were
analyzed to confirm that embryos with the increased cell
death pattern were wg mutants. Embryos with increased cell
death at the segment border consistently exhibited a wg
cuticle phenotype.
Cell death in an
armadillo
temperature sensitive
mutant
ARM, a β-catenin homolog, functions both in the adhesion
junctions and in WG signal transduction (Peifer and Wieschaus,
1990). In order to examine the role of ARM in cell survival, we
examined cell death in arm mutant embryos. A temperature
sensitive mutation was used to precisely control the time at
which ARM signaling was removed. The temperature sensitive
mutant, armH8.6 (referred to as armts) is primarily defective in
WG signal transduction (Klingensmith et al., 1989). armts
embryos were collected at room temperature (22°C), injected
with AO and immediately shifted to the restrictive temperature
of 29°C and aged until stage 11, at which point time-lapse
recordings were initiated at room temperature. Cell death
increased in the same rows of cells as seen in wgts embryos that
were shifted to the restrictive temperature at stage 11 (Fig. 7A-
B). Cell death in the rest of the segment was similar to wild-
type. The width of the stripe of cell death expanded posteriorly,
toward the middle of the segment, in embryos that were
collected at 25°C, instead of 22°C (data not shown). These
embryos also had a slight increase in cell death in the posterior
of the segment in the en-expressing cells. The homozygous
armts embryos that were collected at 25°C had more severe
cuticle defects than those collected at 22°C, which correlated
with the width of the cell death stripe. These results are
consistent with the wg result and indicate that the transduction
of the WG signal through the ARM pathway is required to
promote cell survival of rows of cells in the anterior half of the
segment.
Cell death in
gsb
mutant embryos
Previous studies have shown that gsb, which encodes for a
transcription factor, functions in an autoregulatory loop
required for the maintenance of WG expression (Li and Noll,
1993). More recently, it has been proposed that GSB is
specific to the ventral epidermis for the maintenance of WG
expression and that another transcription factor, ladybird,
provides the corresponding function in the dorsal epidermis
(Jagla et al., 1997). It was reasoned that loss of gsb may only
lead to the demise of cells in the ventral epidermis. This was
examined by time-lapse microscopy of AO and intervitelline
RGPEG-injected embryos that had both gsbd and gsbp genes
deleted. Mutant embryos were identified by cuticle analysis of
the mounted embryos. gsb mutant embryos displayed an
increase in apoptotic nuclei in the anterior portion of the
segment similar to that seen in wgts and armts embryos (Fig.
7C-D). However, the increase in cell death was restricted to
the ventral and ventrolateral surface of the embryo. There was
also a slight increase in cell death in the ventral EN-expressing
cells.
Cell death in
nkd
mutant embryos
Mutations in the segment polarity gene, nkd, also resulted in
increased cell death that appeared in a striped pattern in the
abdominal epidermis (Fig. 7E). Intervitelline RGPEG
indicated that the cells that were dying were in the extreme
posterior of the segment where en is expressed. The segment
borders were shallow and in some cases ran together between
adjacent segments. We used TUNEL to localize the cell death
in nkd mutants relative to en expression to show that the
majority of the increased cell death was in the en-expressing
cells (Fig. 7F-G). The nkd embryos had about a 6-fold increase
in apoptosis in EN cells compared to similar staged wild-type
embryos. The domain of EN expression was expanded
anteriorly in nkd mutants, which was where the increased cell
death was located. There was also a slight increase in cell death
throughout the segment.
DISCUSSION
Cell death in wild-type embryos is segmentally
repeated
Programmed cell death in Drosophila embryos starts at late
stage 11 and continues throughout embryonic development.
Using time-lapse microscopy of AO-injected embryos, we
have analyzed the cell death patterns in the abdominal
epidermis to show that approximately 40-45 cells die in the
ectoderm of each segment during stages 12-14. The number
of dying cells varies from embryo to embryo although the
relative location of the dying cells is consistent. The fact that
cells die in similar positions in different embryos indicates
that precise mechanisms are employed to select the doomed
cells. The variation in cell death numbers may reflect
fluctuations in the number of cells in the abdomen of
different embryos and variability in pattern formation. It is
believed that the role of cell death is to remove excess cells
at the end of a patterning process, such as is the case in
Drosophila eye development (Cagan and Ready, 1989). We
have previously shown that increasing or decreasing the
number of cells fated to the abdomen was repaired by
increased or decreased abdominal cell death, respectively,
which resulted in normalized abdominal patterning (Namba
et al., 1997).
Double labeling for cell death and segment boundaries was
used to spatially locate dying cells within each segment. The
majority of the apoptotic nuclei lay within a few cell diameters
of either side of the segment border. This corresponds to the
region in or adjacent to en expression. Mapping cell death with
respect to en expression provided similar results where
approximately three quarters of the dying cells mapped in or
adjacent to en expression. This bias in the location of cell death
was independent of whether dying cells were identified by
TUNEL, rpr expression or AO. The segmental pattern of cell
death implicated the segment polarity genes in establishing the
wild-type cell death pattern. The segment polarity gene

3434
products are known to be essential in establishing cell-cell
communication that is responsible for generating epidermal
fates (reviewed by DiNardo et al., 1994). Perhaps this cell-cell
communication is also required for cell survival as well as
determining cell fate.
Segment polarity mutants of the WG class have
increased cell death in the anterior of the segment
Increased cell death has often been implicated in mutations that
affect segment polarity genes resulting in smaller larvae
(Perrimon and Mahowald, 1987; Klingensmith et al., 1989).
Time-lapse microscopy of AO-injected embryos mutant for
genes in the WG signaling pathway was performed to
determine precisely when and where cells die in these mutants.
In order to gain better control over the stage at which the WG
expression was lost, we used a temperature
sensitive mutation of wg. Cell death in wgts mutant
embryos shifted to the restrictive temperature at
stage 9 was increased dramatically in many
regions of the embryo, making it difficult to
ascertain any specific cell death pattern. This
abundant cell death is likely to be a result of an
overall defect in pattern establishment. In spite of
the early stage 9 loss of wg expression, the
increase in cell death was not observed until stage
12, during germband retraction. The restriction of
cell death to after stage 11 or 12 in these mutants
and in wild-type embryos indicates developmental
regulation of cell death competence. Despite the
extensive cell death in wgts embryos that were
shifted to the restrictive temperature at stage 9,
many cells did not undergo apoptosis, indicating
that the survivors were either unaffected by the
loss of WG signaling or they were resistant to
apoptosis.
When WG function was disrupted during stage
11, the fate specification phase of epidermal
development, approximately two rows of cells
died in the anterior-most portion of each segment
during stages 12-14 (Fig. 6). These dying cells
were approximately 6 rows of cells away from the
WG-secreting cells. An arm mutation that
disrupted WG signaling also eliminated the same
rows of cells. These results show that WG
signaling is required at a distance to promote the
survival of cells in the anterior of the segment. It
is important to note that the cells expressing WG
prior to the temperature shift and those cells
immediately anterior to WG-secreting cells did
not die and, therefore, may be more resistant to
apoptosis than those cells at a distance. Mutations
in the segment polarity genes gsb also led to the
death of cells in the anterior of the segment.
However this death was more restricted to the
ventral surface.
Apoptosis in
nkd
mutants
Strong mutations in nkd results in a cuticle
phenotype opposite to that of wg mutations.
Previous genetic studies have suggested that the
nkd gene product is necessary to suppress the
domain of WG function. nkd mutants also have increased cell
death, of which the majority is in the expanded EN-expressing
domain. There is however no increase in cell death in the
anterior of the segment similar to that seen in WG mutants.
Taken together these results suggest that two separate systems
may be involved in promoting cell survival in the embryonic
ectoderm: a nkd-dependent system that keeps posterior cells
alive and a wg-dependent system that plays a similar role in
the anterior of the segment. These two systems may not be
mutually exclusive. It has been suggested that cell death in a
wg mutant is suppressed by a nkd mutation (Bejsovec and
Wieschaus, 1993). These results provide evidence that segment
polarity gene interactions play an intimate role in epidermal
cell survival. However, much more work is needed to further
our understanding of these processes.
T. M. Pazdera, P. Janardhanand J. S. Minden
Fig. 6. Cell death in wgts embryos shifted to the restrictive temperature at stage 9,
during the stabilization phase. Embryos from wgts/Cy (ftz:lacZ) were injected with
AO, aged at the permissive temperature for 4.5 hours and then shifted to the
restrictive temperature. (A-C) Time-lapse images of the same embryo at stage 12,
13 and 14. Cell death in wgts embryos shifted to the restricted temperature at stage
11. wgts embryos were injected with AO and aged at the permissive temperature for
6.5 hours until stage 11 and then shifted to the restrictive temperature.
(D-F) Selected images from a time-lapse recording of a wgts embryo at stage 12, 13
and 14. Only the abdomen is shown. The red arrows indicate continuous stripes of
cell death. (G) Double injection of wgts embryos with intervitelline RGPEG (red)
and cytoplasmic AO (green) shows that the increased cell death is near the anterior
segment border in approximately two rows of cells along the dorsal-ventral axis.

3435
Drosophila
epidermal cell death
How does the removal of WG cause cell death in the
anterior of the segment?
As stated above, there is a significant separation between the
WG-secreting cells and the cells that die in response to loss
of WG signaling. One possible explanation is that WG acts
directly on these cells by diffusing over a large distance to
promote cell survival. When WG is removed, the cells die.
The cells adjacent to the WG cells may still receive sufficient
signal to survive. There is some debate as to how far the WG
signal can be transmitted across the segment. Previous results
suggest that the WG signal is only able to directly affect the
fate of cells adjacent to the WG secreting cells, only one or
two rows away (Vincent and Lawrence, 1994). Lawrence et.
al (1996) proposed a model where a steep gradient of secreted
WG plays a role in promoting cell survival. Another possible
explanation for the long-range WG effect is that WG is acting
through an intermediate signal that is produced by cells
receiving a short-range WG signal – this is a relay
mechanism. When WG-responding cells adjacent and
anterior to WG-secreting cells are disturbed, there is a
disruption of a signal that promotes the survival of even more
anterior cells. The fact that mutations in arm lead to the same
cell death phenotype as wgts mutants and that there is no
indication of nuclear ARM accumulation in the anterior-most
cell of the segment, suggests that the survival of the anterior-
most cells may be dependent on a different factor that is
secreted in an ARM-dependent fashion.
The time and location of epidermal cell death in the wild-
type embryo indicates that cells are selected to die during the
stage that WG and EN cells are signaling to each other to
establish epidermal cell fates. Our results indicate that one of
the roles of this signaling is to establish cell survival. Cells that
do not receive the appropriate signal or signals at the
appropriate time die by apoptosis. Furthermore, our results
suggest that complex cell-cell interactions are essential to
promote cell survival within segments of the abdominal
epidermis. Cell-cell interactions also seem to play a role in
promoting cell survival in the wing disc. Cell ablation in one
region of the wing disc leads to apoptosis of cells in adjacent
territories (Milán et al., 1997). The results presented here only
examine the effects of removing signals from a few rows of
epidermal cells. Interactions between other rows of epidermal
cells will certainly be important for promoting cell survival.
Future experiments will need to be performed to further
address the role of cell-cell communication in the survival of
abdominal ectoderm cells. These results also provide a basis in
furthering our knowledge on how epidermal mutant
phenotypes evolve and how the segment polarity genes
function in normal embryos. wg mutants are devoid of smooth
cuticle and have several rows of type four denticles. Our results
Fig. 7. Cell death patterns in arm, gsb and nkd mutant
embryos. (A-B) Images from a time-lapse recording of
an armts embryo at stage 12 and 13 shifted from 22°C to
the restricted temperature at stage 5. AO-positive signal,
or dying cells are shown in green and the segment
borders are in red. This embryo is slightly rotated to
show more of the dorsal surface. (C-D) Images from a
time-lapse recording of a gsb embryo at stage 12 and
13. Notice the increased cell death at the anterior
segment border. The majority of the ectopic cell death
was restricted to the ventral half of the embryo. (E) An
image of the abdomen of a stage 13 nkd mutant embryo
injected with AO. (F-G) A nkd mutant embryo at late
stage 12 and 13 that has been double labeled with
TUNEL and anti-EN antibody.

3436
suggest that one requirement for this phenotype is the death of
anterior row cells, which produce type two and three denticles.
Future experiments will examine the role of cell death in
generating these mutant phenotypes.
We would like to thank Steve DiNardo for the wgts flies and the
EN/INV antibody and Norbert Perrimon for the armts fly stock. We
would also like to thank Brian McNally and Wladimir Labeikovsky
for their technical help and Minako Pazdera for her support. T. P. was
supported in part by an NSF graduate training Grant to the Science
and Technology Center for Light Microscope Imaging and
Biotechnology, BIR 9256343. J. M. is a Lucille P. Markey Scholar,
and this work was supported in part by grants from the Lucille P.
Markey Charitable Trust and NIH grant HD31642.
REFERENCES
Abrams, J. M., White, K., Fessler, L. I. and Steller, H. (1993). Programmed
cell death during Drosophila embryogenesis. Development 117, 29-43.
Bejsovec, A. and Martinez Arias, A. (1991). Role of wingless in patterning
the larval epidermis of Drosophila. Development 113, 471-485.
Bejsovec, A. and Wieschaus, E. (1993). Segment polarity gene interactions
modulate epidermal patterning in Drosophila embryos. Development 119,
501-517.
Bomze, H. M. and López, A. J. (1994). Evolutionary conservation of the
structure and expression of alternatively spliced ultrabithorax isoforms from
Drosophila. Genetics 136, 965-977.
Cagan, R. L. and Ready, D. F. (1989). The emergence of order in the
Drosophila pupal retina. Dev. Biol. 136, 346-362.
Campos-Ortega, J. A. and Hartenstein, V. (1985). The embryonic
development of Drosophila melanogaster. New York, Berlin, Heidelberg,
Tokyo: Springer-Verlag.
Delic, J., Coppey, J., Magdelenat, H. and Coppey-Moisan, M. (1991).
Impossibility of acridine orange intercalation in nuclear DNA of the living
cell. Exp. Cell. Res. 194, 147-153.
DiNardo, S., Heemskerk, J., Dougan, S. and O’Farrell, P. H. (1994) The
making of a maggot: patterning the Drosophila embryonic epidermis. Cur.
Opin. Gen. Dev. 4, 529-534.
Edgar, B. A. and Lehner, C. F. (1996). Developmental control of cell cycle
regulators: a fly’s perspective. Science. 274:5293, 1646-1652.
Foe, V. E. (1989). Mitotic domains reveal early commitment of cells in
Drosophila embryos. Development. 107, 1-22.
Gonzalez, F., Swales, L., Bejsovec, A., Skaer, H. and Martinez Arias, A.
(1991). Secretion and movement of wingless protein in the epidermis of the
Drosophila embryo. Mech. Dev. 35, 43-54.
Grether, M. E., Abrams, J. M., Agapite, J., White, K. and Steller, H.
(1995). The head involution defective gene of Drosophila melanogaster
functions in programmed cell death. Genes Dev. 9, 1694-1708.
Hartenstein, V. and Campos-Ortega, J. A. (1985). Fate-mapping in wild-
type Drosophila Melanogaster I. The pattern of embryonic cell divisions.
Wilhelm Roux’s Arch. Dev. Biol. 194, 181-195.
Hartenstein, V. (1993). In Atlas of Drosophila Development. Cold Spring
Harbor, NY: Cold Spring Harbor Laboratory Press.
Jacobson, M. D., Weil, M., and Raff, M. C. (1997). Programmed cell death
in animal development. Cell 88, 347-354.
Jagla, K., Jagla, T., Heitzler, P., Dretzen, G., Francois, B., and Bellard, M.
(1997) ladybird, a tandem of homeobox genes that maintain late wingless
expression in terminal and dorsal epidermis of the Drosophila embryo.
Development. 124, 91-100.
Klingensmith, J., Noll, E., and Perrimon, N. (1989). The segment polarity
phenotype of Drosophila involves differential tendencies toward
transformation and cell death. Dev. Biol. 134, 130-145.
Lawrence, P. A., Sanson, B., and Vincent, J. P. (1996). Compartments,
wingless and engrailed: patterning the ventral epidermis of Drosophila
embryos. Development. 122, 4095-4103.
Li, X. and Noll, M. (1993). Role of the gooseberry gene in Drosophila
embryos: maintenance of wingless expression by a wingless-gooseberry
autoregulatory loop. EMBO J. 12, 4499-4509.
Martinez-Arias, A. (1993). Development and patterning of the larval
epidermis of Drosophila. In The Development of Drosophila melanogaster
(ed. Bate, M. and Martinez-Arias, A.), pp. 517-608. Cold Spring Harbor,
NY: Cold Spring Harbor Laboratory Press.
Milán, M., Campuzano, S., and García-Bellido, A. (1997). Developmental
parameters of cell death in the wing disc of Drosophila. Proc. Natl. Acad.
Sci. USA. 94, 5691-5696.
Minden, J. S. (1996). Synthesis of a new substrate for detection of lacZ gene
expression in live Drosophila embryos. BioTechniques 20, 122-129.
Namba, R., Pazdera, T. M., Cerrone, R. L., and Minden, J. S. (1997)
Drosophila embryonic pattern repair: how embryos respond to bicoid
dosage alteration. Development 124, 1393-1403.
Nüsslein-Volhard, C. and E. Weischaus. (1984). Mutations affecting the
pattern of the larval cuticle in Drosophila Melanogaster. I. Zygotic loci
on the second chromosome. Wilhelms Roux’s Arch. Dev. Biol. 193, 267-
282.
Peifer, M. and Wieschaus, E. (1990). The segment polarity gene armadillo
encodes a functionally modular protein that is the Drosophila homolog of
human plakoglobin. Cell 63, 1167-1178.
Perrimon, N. and Mahowald, A. (1987). Multiple functions of segment
polarity genes in Drosophila. Dev. Biol. 119, 587-600.
Perrimon, N. (1989) Multiple functions of a Drosophila homeotic gene
zeste-white 3, during segmentation and neurogenesis. Dev. Biol. 135, 287-
305.
Tautz, D. and Pfeifle, C. (1989). A non-radioactive in situ hybridization
method for the localization of specific RNA’s in Drosophila embryos reveals
translational control of the segmentation gene hunchback. Chromosoma 98,
81-85.
Tepass, U., Fessler, F. I., Aziz, A., and Hartenstein, V. (1994). Embryonic
origin of hemocytes and their relationship to cell death in Drosophila.
Development. 120, 1829-1837.
Vincent, J. P. and Lawrence, P. A. (1994). Drosophila wingless sustains
engrailed expression only in adjoining cells: evidence from mosaic
embryos. Cell. 77, 909-915.
White, K., Grether, M. E., Abrams, J. A., Young, L., Farrell, K. and Steller,
H. (1994). Genetic control of programmed cell death in Drosophila. Science
264, 677-683.
T. M. Pazdera, P. Janardhanand J. S. Minden