Many networks of epithelial tubes develop from isolated
branching units which interconnect by fusion of tubes from
different units. In the development of the vertebrate vascular
system, for example, blood islands form and then coalesce to
establish the network of major vessels (Risau and Flamme,
1995). Later in development, new tubules sprout from these
vessels and grow out and fuse with other vessels to further
interconnect the vascular network. Fusion of epithelial tubes
also occurs during development of the kidney: the developing
nephron fuses with the extending ureteric duct system to
establish continuity between these tubes, so that urine can flow
from the nephron to the ureters (Bard et al., 1994). Fusion of
epithelial tubes requires that growing tubes locate a fusion
partner, adhere and establish a patent connection between
them. Little is known of the cellular mechanisms or molecular
control of such epithelial tube fusion events.
The Drosophila tracheal (respiratory) system is a branched
tubular epithelial network that transports oxygen throughout
the body. The twenty tracheal metameres arise independently
from sacs of ~80 tracheal cells that undergo a series of sequen-
tial branching events (Manning and Krasnow, 1993;
Samakovlis et al., 1996). While most tracheal branches
continue forming new branches throughout embryonic and
larval life, five branches in each metamere cease branching
during embryogenesis and grow towards and fuse to branches
from the neighboring hemisegments to interconnect the
tracheal network (Fig. 1A-D). Each fusion event is mediated
by a single, specialized cell in each of the fusing branches,
which expresses a set of fusion markers that were identified in
a P[lacZ] transposon enhancer trap screen (Samakovlis et al.,
1996; see below).
In this paper, we describe the cellular dynamics of a tracheal
fusion event and identify a gene regulatory hierarchy that
controls it. Each fusion cell and its partner undergo a sophis-
ticated morphogenetic program involving cytoplasmic
outgrowth, cell adhesion and formation of an intracellular
lumen, generating a connecting joint composed of two
doughnut-shaped cells. Fusion markers are expressed in a
specific sequence that anticipates the cellular events of fusion.
One of the two early markers is identified as the escargot gene,
which encodes a zinc finger transcription factor that has pre-
viously been found to function in the development of imaginal
histoblasts (Ashburner et al., 1990; Whiteley et al., 1992;
Hayashi et al., 1993; Fuse et al., 1994). escargot is an activator
of the fusion program, as well as a repressor of branching, that
can drive ectopic tracheal fusion events and repress terminal
branching when misexpressed.
MATERIALS AND METHODS
The enhancer trap markers 1-eve-1 (Tracheal-1), Fusion-1 to 4,
Fusion-7 (Branch-2), and Terminal-1 and 2 have been described
Development 122, 3531-3536 (1996)
Printed in Great Britain © The Company of Biologists Limited 1996
During development of tubular networks such as the
mammalian vascular system, the kidney and the Drosophila
tracheal system, epithelial tubes must fuse to each other to
form a continuous network. Little is known of the cellular
mechanisms or molecular control of epithelial tube fusion.
We describe the cellular dynamics of a tracheal fusion
event in Drosophila and identify a gene regulatory
hierarchy that controls this extraordinary process. A
tracheal cell located at the developing fusion point
expresses a sequence of specific markers as it grows out and
contacts a similar cell from another tube; the two cells
adhere and form an intercellular junction, and they
become doughnut-shaped cells with the lumen passing
through them. The early fusion marker Fusion-1 is identi-
fied as the escargot gene. It lies near the top of the regula-
tory hierarchy, activating the expression of later fusion
markers and repressing genes that promote branching.
Ectopic expression of escargot activates the fusion process
and suppresses branching throughout the tracheal system,
leading to ectopic tracheal connections that resemble
certain arteriovenous malformations in humans. This
establishes a simple genetic system to study fusion of
Key words: tube fusion, epithelial morphogenesis, branching
morphogenesis, trachea, escargot, Drosophila
Genetic control of epithelial tube fusion during Drosophila tracheal development
Christos Samakovlis1,2, Gerard Manning1, Pär Steneberg2, Nir Hacohen1, Rafael Cantera3
and Mark A. Krasnow1,*
1Department of Biochemistry, Beckman Center, Stanford University School of Medicine, Stanford, CA 94305, USA
2Umeå Center for Molecular Pathogenesis, Umeå University, S-90187 Umeå, Sweden
3Department of Zoology, University of Stockholm, S-10691 Stockholm, Sweden
*Author for correspondence (e-mail: firstname.lastname@example.org)
(Perrimon et al., 1991; Samakovlis et al., 1996). Fusion-1 alleles
l(2)07082 and l(2)05730 were generated in the laboratory of Alan
Spradling (Spradling et al., 1995), and the B7-2-22 allele was
generated in the laboratory of Y. N. Jan (Hartenstein and Jan, 1992).
The Fusion-5 and Fusion-6 markers are P[lacZ] insertions at cyto-
logical positions 92E and 99D, respectively. The escargot mutant
strains used were the null escargot allele G66 (Whiteley et al., 1992)
and the strong EMS-induced alleles VS2 and VS8 (Ashburner et al.,
1990; Hayashi et al., 1993). The chromosomal deficiency used was
Df(2L)osp29 which removes the escargot locus.
Embryo fixation and staining
Embryo fixation and staining, and light and confocal fluorescence
microscopy were as described (Samakovlis et al., 1996). The lumen-
specific antibodies used were TL-1 and mAb2A12. The anti-escargot
antibody was a rat polyclonal antiserum from Shigeo Hayashi (Fuse
et al., 1994). The anti-DSRF monoclonal antibody was mAb 2-161
from Michael Gilman (Ariad Corporation, Boston, MA) and the anti-
coracle guinea pig polyclonal antiserum was from Rick Fehon (Fehon
et al., 1994). Embryo staging was according to Campos-Ortega and
Tracheal cell counts and TUNEL staining to detect dead or dying
cells in the lateral trunk were performed as described (White et al.,
1994; Samakovlis et al., 1996). Cell counts were done on the lateral
trunk of tracheal metameres Tr5 and 6 in an escargot hemizygote
(escargotVS8/Df(2L)osp29) carrying the Tracheal-1 or Tracheal-2
markers; the comparison strain was an escargot heterozygote.
TUNEL staining was followed by staining with the TL1 tracheal
lumenal marker to identify labelled cells in the lateral trunk.
Specimens were prepared for electron microscopy as described
(Tepass and Hartenstein, 1994) except that the fixative was 25% glu-
taraldehyde, 4% paraformaldehyde and 3% tannic acid in 0.1 M
phosphate buffer. Specimens were viewed with a JEOL100CX
Molecular mapping of Fusion-1 alleles was carried out by polymerase
chain reaction of genomic DNA prepared from the enhancer trap
mutants using primers representing either positions 472-490 or 1484-
1466 in the escargot sequence (Whiteley et al., 1992) in conjunction
with a primer from the P element inverted repeat.
GAL4 strains and ectopic expression of escargot
The driver strains used, TrGal4 and C38, express GAL4 in most
tracheal cells beginning at stage 13 (Guillemin et al., 1996). The UAS-
escargot construct has been described (Fuse et al., 1994). Embryos
carrying a GAL4 driver and the UAS-escargot construct were
collected at 18°C for 10 hours, aged at 22°C for 4 hours and then
transferred to 29°C for 5 hours to maximize GAL4 activity. Embryos
were fixed and stained with mAb2A12 alone or mAb2A12 and DSRF
mAb 2-161 to examine tracheal morphology.
Cell dynamics of a tracheal fusion event
Fifty tracheal fusion events occur during embryonic develop-
ment to interconnect the tracheal network. Dorsal and lateral
trunk fusion events link up neighboring tracheal metameres on
each side of the embryo, while dorsal and ventral anastomosis
fusion events connect the two sides of the tracheal system (Fig.
1A-D). Each fusion event is mediated by a specialized cell at
the tip of each fusion branch which expresses the Fusion-1
marker and other fusion markers and undergoes a similar mor-
phogenetic program that results in formation of a bi-cellular
fusion joint (Fig. 1E).
We analyzed in detail the cellular and molecular events of
fusion of the dorsal branches (DB) to form a dorsal anasto-
mosis (DA) in the second through ninth tracheal metameres.
Initially each DB fusion cell (DB2 cell) lies at or near the end
of the outgrowing DB and is indistinguishable from neighbor-
ing tracheal cells. The fusion cell extends a cytoplasmic
process that grows out and contacts a similar outgrowth from
its partner (Fig. 2A,B). After contact, the cells contract and
their cell bodies move toward each other, and a lumen extends
through each cell (Fig. 2C,D). The cells align their lumens and
the lumens fuse end-on, forming a complete passage through
each fusion cell and establishing the connection between
branches (Fig. 2E). The entire process takes about 3 hours.
The two fusion cells do not form a syncytium in the process,
C. Samakovlis and others
Fig. 1. Tracheal fusion and the expression of fusion markers at
developing fusion joints. (A) The left side of the tracheal system at
stage 13 before branch fusion (anterior left, dorsal up). There are ten
independent tracheal hemisegments, Tr1 to Tr10. The embryo was
stained with a lumen-specific antiserum. (B) Three hours later at
stage 15 when specific branches in each hemisegment have fused to
branches in the neighbouring hemisegments to form the dorsal and
lateral tracheal trunks. The dorsal and ventral anastomoses (not
shown), which interconnect the left and right sides of the tracheal
system, form slightly later at stage 16. Bar, 25 µm. (C) Schematic
diagram of Tr5 and Tr6 (lateral view). The five positions where
branches of Tr 6 have fused to branches in the neighboring tracheal
hemisegments to form the dorsal anastomosis (DA), dorsal trunk
(DT) and lateral trunk (LT) are indicated by arrowheads. Ventral
anastomoses (not shown) arise from the ganglionic branches (GB) in
Tr1 and Tr2. DB, dorsal branch; VB, visceral branch. (D) Schematic
diagram of the left and right Tr6 showing the DA that connects them
(dorsal view, anterior at top). (E) Expression of the Fusion-1
enhancer trap marker in both fusion cell partners before (top panels)
and after (bottom panels) branch fusion to form the DA, DT and LT.
Embryos carrying the Fusion-1 marker were double-stained for a
tracheal lumenal antigen (TL-1, top panels, mAb2A12, bottom
panels) and the marker protein (β-galactosidase) which is expressed
in the nucleus of all fusion cells. White dots are placed adjacent to
the fusion cell nuclei. Bar, 2 µm.
3533 Tracheal tube fusion
but become tightly attached by an intercellular junction that
can be visualised in electron micrographs or by staining with
an anti-coracle antibody that labels septate junctions (Fig. 3A).
The structure of the fusion cells is unlike other DB cells, which
are curled up into a tube and have an autocellular junctional
seam running along their length (Samakovlis et al., 1996).
Junctional immunostains (Fig. 3A) and electron micrographs
(Fig. 3B) show that the fusion cells lack a junctional seam and
the tracheal lumen passes through each cell. In cross section,
each fusion cell looks like a doughnut with the lumen forming
the hole in the center (Fig. 3B).
Fusion markers are expressed in a specific
sequence during fusion
During DB fusion, six fusion markers were activated in a fixed
sequence that anticipated the cellular events of fusion (Fig. 4).
The earliest markers (Fusion-1 and Fusion-2) turned on at stage
13, about 2 hours before the first morphological events of
fusion (Fig. 1E). The other fusion markers turned on over the
next 1-3 hours, as the fusion partners approached each other.
The order of fusion marker activation in other tracheal fusion
events was similar, although two of the markers were not
expressed in all fusion cells (see legend to Fig. 4). The corre-
lation of fusion marker expression with the cellular events of
fusion and the finding that mutations in these markers disrupt
fusion (C. S., G. M., and N. H., unpublished results) suggested
that the marker genes might comprise a hierarchical genetic
regulatory program that controls the fusion process. Our
molecular genetic analysis of Fusion-1, described below,
supports this hypothesis.
The Fusion-1 marker is escargot
Three Fusion-1 P[lacZ] inserts were mapped to cytological
position 35C,D (Samakovlis et al., 1996). We carried out com-
plementation tests with existing mutations in the region which
showed that the homozygous lethal Fusion-1 alleles l(2)07082
Fig. 2. Cell dynamics of a tracheal fusion event.
Confocal images of DB fusion cells forming a dorsal
anastomosis at successive stages of fusion, showing
tracheal cells (1-eve-1 marker, pseudocolored red) and
lumen (mAb2A12; pseudocolored green) (dorsal view,
anterior left). The cell bodies of the DB1, 2 and 3 cells in
the two dorsal branches are labeled (1, 2, 3 and 1′, 2′, 3′).
Each fusion cell (2 and 2′) extends a cytoplasmic process
towards its partner (arrowheads in A). After contact
(arrowhead in B), their cell bodies move toward each
other (C,D) and a lumen extends through each cell and
connects to the lumen of its partner (E). The lumen
expands slightly at the fusion point just prior to lumen
fusion and the expansion persists afterwards (arrowhead
in E). Bar, 5 µm.
Fig. 3. Structure of a
(A) Confocal projection
through a fully formed
DA fusion joint (dorsal
view, anterior at top)
showing tracheal cells
pseudocolored red) and
cell junctions (anti-
The DB fusion cells (2
and 2′) that form the
fusion joint are attached
by an intercellular
junctions with the DB1 and DB3 cells are also indicated
(arrowheads). There is no coracle staining in the part of the fusion
cell adjacent to the fusion point, indicating the absence of
autocellular junctions. Analysis of serial confocal sections (not
shown) confirmed that the DB2 fusion cells are seamless tubes. Bar,
2 µm. (B) Electron micrograph of a cross section through a DB2
fusion cell at stage 16. The cell is doughnut-shaped with the lumen
(LU) forming a hole in the center of the cell. Note that there is no
autocellular junction, which would appear as a double membrane
connecting the plasma membrane to the membrane surrounding the
lumen. Bar, 0.22 µm
Fig. 4. Time course of fusion marker expression. The onset of
expression of the different fusion markers in DB fusion cells during
formation of the dorsal anastomosis was determined by
immunostaining different enhancer trap strains for the beta-
galactosidase marker and for a tracheal lumenal antigen (TL-1 or
mAb2A12) to accurately assess developmental stage. The results are
represented on a developmental time line that also shows the timing
of the cellular events of fusion (see Fig. 2). The same general order
of marker activation and cellular events was observed for the other
fusion cells. The Fusion-7 marker, however, was not expressed in
DA or LT fusion cells and the Fusion-6 marker was not expressed in
DT fusion cells.
10 11 12 13 14 hours
and l(2)05730 failed to complement mutations in escargot.
Molecular mapping of Fusion-1 alleles l(2)07082 and B7-2-22
by polymerase chain reaction (see Materials and Methods)
demonstrated that the P[lacZ] transposon had inserted in an
exon of the escargot gene, ~200 and ~400 base pairs, respec-
tively, upstream of the assigned translation start site. Embryo
staining with an anti-escargot antiserum and an mRNA probe
confirmed that escargot is expressed in all fusion cells in the
same pattern as the Fusion-1 marker, in addition to its previ-
ously described expression in imaginal histoblasts and in other
embryonic tissues (data not shown). Fusion-1 is therefore
allelic to escargot.
escargot is required for fusion and inhibits terminal
branching by DB fusion cells
The tracheal phenotype and the expression of fusion markers
were examined in embryos homozygous and hemizygous (over
Df(2L)osp29) for the null escargotG66allele (Whiteley et al.,
1992) and the strong loss-of-function alleles escargotVS2and
escargotVS8(Ashburner et al., 1990; Hayashi et al., 1993). The
results for all three alleles were indistinguishable and are
therefore described together. In the mutants, the DB cells that
normally mediate fusion instead formed discrete branches that
never met or joined (Fig. 5A). These cells went on to form
additional branches that resembled normal terminal branches
in the larva, unlike normal fusion cells which undergo no
further branching (Fig. 5E). Marker expression studies showed
that the mutant fusion cells failed to express the Fusion-4, 5
and 6 markers, while the Fusion-2 and 3 markers continued to
be expressed normally (Fig. 5C and data not shown). The
mutant cells also inappropriately expressed the terminal
branching markers Terminal-1 (DSRF) and Terminal-2 (Fig.
5D); these genes are normally expressed and function only in
tracheal cells that undergo terminal branching, such as the DB1
cell (Guillemin et al., 1996; Samakovlis et al., 1996). Thus,
escargot is required to activate specific fusion genes, as well
as to repress expression of genes that control terminal
branching, in DB fusion cells.
In ventral anastomosis fusion events, the escargot gene
played a role very similar to the one just described for dorsal
anastomosis fusion: it activated expression of the same fusion
markers and repressed terminal marker expression in the fusion
cells. The dorsal and lateral trunk fusion events, however, had
different requirements for escargot. Lateral trunk fusion
branches were completely missing in escargot mutants, leaving
a gap in each segment of the lateral trunk (Fig. 5B). The cells
(LTa4 and LTp8) that normally form these branches apparently
died by apoptosis in the mutants, because (i) no expression of
any fusion markers was detected in the lateral trunk, (ii) there
were two fewer cells than normal in each segment of the lateral
trunk (9.1±1.2 cells in the escargot mutant (n=28 metameres),
compared to 11.5±1.1 cells in the control strain (n=36)), and
(iii) TUNEL analysis (White et al., 1994), which transiently
labels the DNA of apoptotic cells showed no staining in wild-
type embryos (n=10), but labelled the lateral trunk fusion cells
in 6 of 13 mutant embryos. Unlike the defects in the lateral
trunk, ventral anastomosis, and the dorsal anastomosis which
were fully penetrant and expressive in the second through ninth
tracheal metameres, dorsal trunk fusion proceeded almost
normally with only sporadic breaks (~5% of fusion points)
observed in the mutants.
C. Samakovlis and others
DA, Fusion-6 marker
DA, Terminal-1 marker
Larval DA, esg
Fig. 5. Tracheal fusion defects
and altered marker expression in
escargot mutants. (A) The DA in
a wild-type (left) and an
stained with mAb2A12 (dorsal
view, anterior left). The two
fusion branches (arrowheads) fail
to connect in the mutant.
(B) Lateral view of two
hemisegments in wild-type (left)
embryos (right). In the mutant
there is a gap at the fusion point
in each hemisegment of the
lateral trunk (asterisks). Although
in the mutant there are sporadic
breaks in the dorsal trunk (see
text), the portion of the dorsal
trunk shown is intact. Bar, 5 µm. (C) The Fusion-6 marker is
expressed in the DB2 cells (2 and 2′) of wild-type (left panel) but not
in the DB2 cells (asterisks) of the mutant (right panel). (D) DSRF
protein (Terminal-1 marker) is inappropriately expressed in the DB2
cell of the mutant (2 and 2′, right panel). In the wild type (left panel),
DSRF is expressed exclusively in terminal tracheal cells such as DB1
(1 and 1′), where it regulates formation of an extensive array of
terminal branches (Guillemin et al., 1996). Bar (A,C,D), 5 µm.
(E) DA fusion branch (arrowhead) in an escargotmutant larva viewed
by Nomarski optics (dorsal view, anterior left). Note the ramification
into fine branches that resemble normal terminal branches like those
formed by the neighboring DB1 cell (arrows). The escargot mutant
shown is homozyogus for a P element insert into an upstream region
of the escargot locus; it is an apparent null allele for tracheal
expression and function but leaves other escargot expression intact
(C. S. and T. Heino, unpublished data). Other escargot alleles die
earlier in development but the occasional larval survivor displays the
same tracheal phenotype shown. Bar, 10 µm. Dashed lines,
continuations of tracheal branches out of the plane of focus.
3535 Tracheal tube fusion
Ectopic expression of escargot causes ectopic
tracheal fusion events
To further test the function of escargot in branch fusion, we
analyzed the consequences of expressing escargot in tracheal
cells that do not normally mediate fusion. The GAL4 indirect
expression system (Brand and Perrimon, 1993) was used to
drive escargot expression throughout the tracheal system
beginning at stage 13, when expression of the gene is normally
restricted to just the five fusion cells in each tracheal metamere.
This resulted in ectopic expression of the two fusion markers
analyzed and formation of ectopic tracheal connections,
linking up tracheal branches that normally never join (Fig. 6A-
C). Misexpression of escargot also repressed terminal marker
expression and inhibited terminal branching throughout the
tracheal system (Fig. 6D,E).
The tracheal branch fusion process that we have described is
a sophisticated morphogenetic program executed by a special-
ized cell in each fusing branch. The process is symmetric: both
partners express the same markers and undergo the same mor-
phogenetic program in concert. Initially the fusion cells extend
a cytoplasmic process that grows out until finding a similar
extension from its partner. After contact, the cells adhere and
begin to establish an intercellular junction. As their cell bodies
draw together, each cell develops an intracellular lumen. The
net result is a connecting joint composed of two doughnut-
shaped fusion cells, attached face to face, with the lumen
running through them.
The initial stages of the fusion process are reminiscent of
other oriented cytoplasmic outgrowths, such as shmoo
formation during mating in yeast (Drubin and Nelson, 1996)
and neurite extension in animals (Tanaka and Sabry, 1995).
Although the signals that guide outgrowth of the fusion cells
are unknown, it seems likely that the growing cells either signal
each other or use a common guidepost provided by a sur-
rounding tissue in order to find each other. The cellular events
following contact of the fusion cells are more unusual, as the
cells develop an intracellular lumen spanning the length of the
cell. One appealing mechanism for this intracellular lumen
formation is an oriented vesicular fusion process. Vesicles
might line up in the cytoplasm and fuse, generating an elongate
vesicle spanning the length of the cell and ultimately fusing
with the plasma membrane to establish the external openings
that allow air to flow through the cell. Whatever the
mechanism, in order to establish a functional connection the
lumen has to be directly in line with the lumen of its fusion
partner and continuous with the lumen of the preexisting
A gene regulatory hierarchy with seven extant members
controls the fusion process. The genes are expressed sequen-
tially in the fusion cells beginning two hours before the first
morphological signs of fusion and anticipating the morpho-
logical events of fusion. escargot, the Fusion-1 marker, is one
of the two earliest markers and lies near the top of the
hierarchy. It is a key regulator that activates expression of some
of the later fusion markers and can induce ectopic marker
expression and ectopic branch fusion events when misex-
pressed. Since the escargot protein is a DNA-binding tran-
scription factor, some of these regulatory effects may be direct.
If so, then escargot can function as a transcriptional activator,
as well as a transcriptional repressor as was previously shown
(Fuse et al., 1994). The late genes in the regulatory hierarchy
presumably execute the cellular events of fusion including cell
adhesion and morphogenesis into doughnut-shaped cells.
Further analysis of the hierarchy and the other fusion genes
should reveal how they control and execute the remarkable cell
dynamics of fusion and how the hierarchy is normally activated
just in fusion cells.
The cessation of branching is another important aspect of
the branch fusion program which is also regulated by escargot.
In the absence of escargot, the DB fusion cells inappropriately
expressed terminal cell markers and they went on to form fine
branches that resembled the normal networks of terminal
branches. Furthermore, when escargot was expressed through-
out the developing tracheal system, it suppressed terminal
Fig. 6. Expression of
escargot throughout the
tracheal system causes
ectopic tracheal fusions
and suppresses terminal
branching. The GAL4
system was used to
system, starting at stage 13. (A-C) Arrowheads show ectopic
connections between the stalk of visceral branch VB6 and the end of
VB7 and between the stalk of VB5 and the end of VB6 (A), between
the stalks of the Tr7 and Tr8 dorsal branches (B) and between the
Tr5 and Tr6 ganglionic branches (C). None of these branches
normally connect. (D,E) DSRF (Terminal-1 marker) expression in
the DB of a wild-type embryo (D) and an embryo expressing
escargot throughout the tracheal system (E). Note the inhibition of
DSRF expression (brackets) and stunted terminal branches
(asterisks) in the DB1 terminal cells in E. Suppression of terminal
branching throughout the tracheal system and ectopic fusion events
were observed in most embryos and were more pronounced in
embryos carrying two copies of the UAS-escargot construct. Bars in
A,B,E, 5 µm; C, 10 µm.
marker expression and terminal branching everywhere.
escargot thus has two functions in fusion: it activates
expression of fusion markers and promotes branch fusion, and
it represses terminal marker expression and prevents terminal
We have focused here on the fusion program controlling
formation of the dorsal anastomoses in the second through
ninth tracheal metameres. While the other pairs of tracheal
fusion cells express a similar set of fusion genes and undergo
a superficially similar morphogenetic process, their regulatory
programs are not all identical. Two of the fusion markers were
not expressed in some sets of fusion cells. Moreover, not every
fusion cell responded to loss of escargot function in the same
way as the dorsal branch fusion cells. Lateral trunk fusion cells
died in escargot mutants and dorsal trunk fusion cells were
only sporadically affected. Thus, other factors must influence
the function of escargot and the fusion program at different
fusion positions in the embryo.
The simple genetic system established here to study fusion
of tracheal tubes may help understand other epithelial tube
fusion processes and how they go awry, as in the human
vascular system where there are sometimes anomalous con-
nections between arteries and veins that resemble the ectopic
tracheal fusions that we observed (Young, 1988). Blood vessel
fusion also appears to occur by cytoplasmic outgrowth and cell
adhesion (Flamme et al., 1993), and there is evidence that
genesis of an intracellular lumen and formation of doughnut-
shaped cells might also be involved (Wolff and Bär, 1972;
Folkman and Haudenschild, 1980). And, since vertebrate
homologs of escargot are known (Nieto et al., 1992) and can
drive tracheal fusion in Drosophila (P. S. and C. S., unpub-
lished data), the process could be controlled by similar genes.
We thank S. Hayashi, J. Kassis, R. Fehon, M. Gilman, J. Roote,
Alan Spradling, Todd Laverty and the Berkeley Drosophila Genome
Project for strains and antibodies, and D. Andrew for experimental
advice and assistance. We thank the members of our laboratories for
their comments on the manuscript. G. M. was supported by a Howard
Hughes predoctoral fellowship and C. S. by a long term EMBO fel-
lowship during the early stages of this work. This work was supported
a Swedish Research Council (NFR) grant to C. S., and an NIH grant
and an NSF Presidential Young Investigator award to M. A. K.
Ashburner, M., Thompson, P., Roote, J., Lasko, P. F., Grau, Y., el, M. M.,
Roth, S. and Simpson, P. (1990). The genetics of a small autosomal region
of Drosophila melanogaster containing the structural gene for alcohol
dehydrogenase. VII. Characterization of the region around the snail and
cactus loci. Genetics 126, 679-94.
Bard, J. B. L., McConnell, J. E. and Davies, J. A. (1994). Towards a genetic
basis for kidney development. Mech. Dev. 48, 3-11.
Brand, A. H. and Perrimon, N. (1993). Targeted gene expression as a means
of altering cell fates and generating dominant phenotypes. Development118,
Campos-Ortega, A. J. and Hartenstein, V. (1985). The Embryonic
Development of Drosophila melanogaster. New York: Springer-Verlag.
Drubin, D. and Nelson, W. (1996). Origins of cell polarity. Cell 84, 335-344.
Fehon, R. G., Dawson, I. A. and Artavanis-Tsakonas., S. (1994). A
Drosophila homologue of the membrane-skeleton protein 4.1 is associated
with septate junctions and is encoded by the coracle gene. Development 120,
Flamme, I., Baranowski, A. and Risau, W. (1993). A new model of
vasculogenesis and angiogenesis in vitro as compared with vascular growth
in the avian area vasculosa. The Anatomical Record 237, 49-57.
Folkman, J. and Haudenschild, C.(1980). Angiogenesis in vitro. Nature288,
Fuse, N., Hirose, S. and Hayashi, S. (1994). Diploidy of Drosophila imaginal
cells is maintained by a transcriptional repressor encoded by escargot. Genes
Dev. 8, 2270-2281.
Guillemin, K., Groppe, J., Ducker, K., Treisman, R., Hafen, E., Affolter,
M. and Krasnow, M. A. (1996). The pruned gene encodes the Drosophila
serum response factor and regulates cytoplasmic outgrowth during terminal
branching of the tracheal system. Development 122, 1353-1362.
Hartenstein, V. and Jan, Y. N. (1992). Studying Drosophila embryogenesis
with P-lacZ enhancer trap lines. Roux’s Arch. Dev. Biol. 201, 194-220.
Hayashi, S., Hirose, S., Metcalfe, T. and Shirras, A. D. (1993). Control of
imaginal cell development by the escargotgene of Drosophila. Development
Manning, G. and Krasnow, M. A. (1993). Development of the Drosophila
tracheal system. In The Development of Drosophila (ed. A. Martinez-Arias
and M. Bate), Vol. 1, pp. 609-685. Cold Spring Harbor, New York: Cold
Spring Harbor Laboratory Press.
Nieto, M. A., Bennett, M. F., Sargent, M. G. and Wilkinson, D. G. (1992).
Cloning and developmental expression of Sna, a murine homologue of the
Drosophila snail gene. Development 116, 227-37.
Perrimon, N., Noll, E., McCall, K. and Brand, A.(1991). Generating lineage-
specific markers to study Drosophila development. Dev. Genet. 12, 238-52.
Risau, W. and Flamme, I. (1995). Vasculogenesis. Annu. Rev. Cell Dev. Biol.
Samakovlis, C., Hacohen, N., Manning, G., Sutherland, D., Guillemin, K.
and Krasnow, M. A. (1996). Branching morphogenesis of the Drosophila
tracheal system occurs by a series of morphologically distinct but genetically
coupled branching events. Development 122, 1395-1407.
Spradling, A.C., Stern, D.M., Kiss, I., Roote, J., Laverty, T. and Rubin,
G.M. (1995). Gene disruptions using P transposable elements: An integral
component of the Drosophilagenome project. Proc. Natl. Acad. Sci. USA92,
Tanaka, E. and Sabry, J. (1995). Making the connection: cytoskeletal
rearrangements during growth cone guidance. Cell 83, 171-6.
Tepass, U. and Hartenstein, V. (1994). The development of cellular junctions
in the Drosophila embryo. Dev. Biol. 161, 563-596.
White, K., Grether, M. E., Abrams, J. M., L. Young, K. F. and Steller, H.
(1994). Genetic control of programmed cell death in Drosophila. Science
Whiteley, M., Noguchi, P. D., Sensabaugh, S. M., Odenwald, W. F. and
Kassis, J. A. (1992). The Drosophila gene escargot encodes a zinc finger
motif found in snail-related genes. Mech. Dev. 36, 117-27.
Wolff, J. R. and Bär, T. (1972). ‘Seamless’ endothelia in brain capillaries
during development of the rat’s cerebral cortex. Brain Research 41, 17-24.
Young, A. E. (1988). Pathogenesis of Vascular Malformations. In Vascular
Birthmarks: Hemangiomas and Malformations (ed. J. B. Mulliken and A. E.
Young), pp. 107-113. Philadelphia: W.B. Saunders.
(Accepted 1 August 1996)
C. Samakovlis and others