The Drosophila retained/dead ringer gene and ARID gene family function during development.

Abstract

The recently discovered ARID family of proteins interact with DNA through a phylogenetically conserved sequence termed the A/T Interaction Domain (ARID). The retained/dead ringer (retn/dri) gene of Drosophila melanogaster is a founding member of the ARID gene family, and of the eARID subfamily. This subfamily exhibits an extended region of sequence similarity beyond the core ARID motif and a separate conserved domain termed the REKLES domain. retn/dri is involved in a range of developmental processes, including axis patterning and muscle development. The retn/dri ARID motif has been shown by in vitro studies to exhibit sequence-specific DNA binding activity. Here we demonstrate that the ARID domain is essential for the in vivo function of retn/dri during embryonic development by showing that a mutant form of RETN/DRI, deleted for part of the ARID domain and unable to bind DNA in vitro, cannot rescue the retn/dri mutant phenotype. In the presence of wild-type RETN/DRI this construct acts as a dominant negative, providing additional support for the proposal that RETN/DRI acts in a multiprotein complex. In contrast, we are yet to find an in vivo role for the REKLES domain, despite its clear evolutionary conservation. Finally, we have used germline clone analysis to reveal a requirement for retn/dri in the Drosophila preblastoderm syncytial mitoses.

Full text

Page 1

The Drosophila retained/dead ringer gene and ARID gene
family function during development
TETYANA SHANDALA1, R. DANIEL KORTSCHAK1 and ROBERT SAINT1,2,*
1Centre for the Molecular Genetics of Development and Dept. of Molecular Biosciences, Adelaide University, Adelaide, SA, Australia and
2Research School of Biological Sciences, Australian National University, Canberra, ACT, Australia
ABSTRACT
genetically conserved sequence termed the A/T Interaction Domain (ARID). The retained/dead
ringer (retn/dri) gene of Drosophila melanogaster is a founding member of the ARID gene family,
and of the eARID subfamily. This subfamily exhibits an extended region of sequence similarity
beyond the core ARID motif and a separate conserved domain termed the REKLES domain. retn/
dri is involved in a range of developmental processes, including axis patterning and muscle
development. The retn/dri ARID motif has been shown by in vitro studies to exhibit sequence-
specific DNA binding activity. Here we demonstrate that the ARID domain is essential for the in vivo
function of retn/dri during embryonic development by showing that a mutant form of RETN/DRI,
deleted for part of the ARID domain and unable to bind DNA in vitro, cannot rescue the retn/dri
mutant phenotype. In the presence of wild-type RETN/DRI this construct acts as a dominant
negative, providing additional support for the proposal that RETN/DRI acts in a multiprotein
complex. In contrast, we are yet to find an in vivo role for the REKLES domain, despite its clear
evolutionary conservation. Finally, we have used germline clone analysis to reveal a requirement
for retn/dri in the Drosophila preblastoderm syncytial mitoses.
The recently discovered ARID family of proteins interact with DNA through a phylo-
KEY WORDS: ARID motif, gene regulation, cell cycle control, DNA binding proteins
Int. J. Dev. Biol. 46: 423-430 (2002)
0214-6282/2002/$25.00
© UBC Press
Printed in Spain
www.ijdb.ehu.es
*Address correspondence to: Prof. Robert Saint. Centre for the Molecular Genetics of Development, Research School of Biological Sciences, Australian National
University, Canberra, ACT 0200, Australia. Fax: +61-2-6125-8294. e-mail: robert.saint@anu.edu.au
Abbreviations used in this paper: ARID, AIT-rich interaction domain; EMS,
ethylmethanesulphonate.
Introduction
The development of multicellular organisms requires tight spa-
tial and temporal control of switches in the activity of key regulatory
genes. Tissues are specified by transcriptional regulatory cas-
cades that respond to a variety of extracellular and intracellular
cues. It has become clear in recent years that the regulators of
tissue-specific gene expression, the transcription factors, are
assembled into multi-protein complexes. These complexes in-
clude DNA-binding proteins, which recruit other components that
are not able to bind DNA directly. The composition of complexes
assembled on a specific promoter can be transient, as seen in the
cascade of gene regulation in the early Drosophila embryo. Curi-
ously, many transcription factors fall into a relatively small number
of phylogenetically conserved families, although different family
members can assume very different developmental roles in differ-
ent organisms (see, for example, Lall and Patel, 2001). As a result,
studies of members of conserved protein families very often
provide new approaches to understanding mechanisms that regu-
late a variety of developmental events.
One of the most recent of the phylogenetically conserved
families to have been identified is the ARID family (see Kortschak
et al., 2000). The family was discovered simultaneously by the
characterisation of two proteins, mouse Bright (Herrscher et al.,
1995; reviewed in Webb, 2001) and Drosophila Dead ringer
(Gregory et al., 1996). The dead ringer gene has since been shown
to correspond to the retained (retn) gene (Schüpbach and
Weischaus, 1991; Ditch, Pitman, Finley, Edeen and McKeown,
unpublished observations) and is here referred to as retn/dri. Bright
and RETN/DRI were found to contain a conserved sequence-
specific DNA binding motif, termed the A/T Interaction Domain
(ARID) (Herrscher et al., 1995; Gregory et al., 1996). Members of
this family have since been found in Saccharomyces cerevisiae,
Caenorhabditis elegans, Danio rerio, mouse and human (Fig. 1A,
Kortschak et al., 2000). Sequence-specific DNA binding appears
not to be an intrinsic feature of the ARID motif, since the ARID motif
proteins human SMARCF1 and Drosophila OSA (orthologs of
yeast SWI1) exhibit non-sequence-specific DNA binding (Dallas et
al., 2000; Collins et al., 1999).
More detailed sequence comparisons revealed that the ARID
family proteins could be divided into three subfamilies. The first of

Page 2

424 T. Shandala et al.
these, which contains Drosophila RETN/DRI and mouse Bright,
exhibits an extended region of similarity either side of the core
ARID motif. We have termed this larger region of similarity the
extended ARID (eARID) motif (Kortschak et al., 2000). eARID
subfamily members also share an independent conserved domain
termed the REKLES domain (see below). The second subfamily is
defined on the basis of the presence of several C-terminally located
conserved domains, including PHD finger domains and a highly
conserved PLU-domain. This sub-family includes Ustilago maydis
Rum1 (Quadbeck-Seeger et al., 2000), mouse and human SmcX/
Y (Agulnik et al., 1994a; Agulnik et al., 1994b), mouse Jumonji and
Desrt (Takeuchi et al., 1995; Lahoud et al., 2001), and human PLU-
1, XE169, RBP1, RBP2, MRF1 and MRF2 (Lu et al., 1999; Wu et
al., 1994; Fattaey et al., 1993; Huang et al., 1996). The third
subfamily of ARID proteins includes yeast SWI1 (Quinn et al.,
1996), Drosophila OSA/ELD (Treisman et al., 1997), a
Caenorhabditis elegans ortholog (GeneBank accession
no.U80439), mouse Osa1 and human SMARCF1/OSA1/p270/
B120/BAF250 (Takeuchi et al., 1997; Kozmik et al., 2001) proteins.
In addition to the ARID motif, these proteins have N-terminal Osa
Homology Domains 1 and 2 (OHD1, OHD2) containing four LXXLL
motifs, which are implicated in nuclear hormone receptor binding
(NR-box).
From the outset, it appeared that members of this gene family
would play roles in developmental regulation. For example, retn/dri
exhibits a highly regulated temporal and spatial pattern of expression
(Fig. 2, Gregory et al., 1996; Shandala et al., 1999), while Bright, a
B cell regulator of IgH transcription also exhibits tissue-restricted
patterns of expression (Herrscher et al.,1995; Webb et al., 1998).
Evidence for a developmental role for
ARID family genes has also come from ge-
netic analysis. The SWI1 gene of Saccharo-
myces cerevisiae is part of the Swi/Snf com-
plex required for the regulation of many
genes involved in simple developmental pro-
cesses such as mating type switching (see
Sudarsanam and Winston, 2000, for a re-
view). More recently, the regulator U. maydis
1 (rum1) gene has been shown to be essen-
tial for normal development of Ustilago
maydis spores (Quadbeck-Seeger et al.,
2000). The effect on sporulation appears to
be mediated by downregulation of a very
defined set of genes, including egl1, dik1,
lga2, hum2, that are normally induced by two
homeobox genes, bE and bW, encoded by
the U. maydis mating type locus.
A role for the Drosophila retn/dri gene in
embryonic development was revealed by
characterisation of retn/dri EMS-induced al-
leles (Shandala et al., 1999). retn/dri was
found to be essential for several aspects of
embryogenesis, including anterior/posterior
and dorsal/ventral patterning. One of the
targets of retn/dri transcriptional regulation
in the blastoderm embryo appears to be the
dorsal-specific gene zerknüllt (zen). zen is
normally expressed in a narrow dorsally
located stripe along the embryo. In retn/dri
mutant embryos, zen expression fails to refine to its normal
amnioserosa-specific expression (Valentine et al., 1998). The
activity of RETN/DRI in this case is to convert the normal transcrip-
tional activator Dorsal (DL) into a repressor by binding adjacent to
DL and creating a surface to which the Groucho (GRO) WD-40
repeat co-repressor protein binds (Valentine et al., 1998). GRO
has been found to interact with the N-terminal tail of histone H3 and
with the histone deacetylase Rpd3, an enzyme involved in chroma-
tin condensation and gene inactivation (Palaparti et al., 1997; Chen
et al., 1999).
Repression of huckebein along the ventral side of the embryo
trunk is also mediated by RETN/DRI and DL, through recruitment
of GRO to the ventral repression element of hkb (Hader et al.,
2000). huckebein is a Drosophila terminal gap gene expressed in
the anterior and posterior caps. It is activated along the ventral side
of the embryo by DL. For both zen and hkb, there is a difference
between the behaviour of the endogenous genes and of the
minimal repression element derived from them, the minimal ele-
ments exhibiting much greater sensitivity to retn/dri regulation.
This is presumably the result of a level of redundancy in the
repression of the wild-type gene that is lost in the reduction to the
minimal regulatory element.
In addition to its role as a repressor, evidence suggests that
RETN/DRI can act as an activator. retn/dri is required for activation
of the argos gene in two near-terminal domains of expression in the
blastoderm embryo, accounting for the head skeleton phenotype
observed in retn/dri mutant embryos (Shandala et al., 1999).
None of these studies addressed the significance of the various
RETN/DRI motifs, including the ARID motif, in the in vivo function
A
B
Fig. 1. Sequence comparison of the ARID and REKLES motifs of selected ARID family proteins.
(A) eARID and ARID alignment showing the extended region of homology of RETN/DRI and Bright,
both N and C-terminal to the core ARID which is conserved across all family members. (B) Alignment
of amino acid sequences of the REKLES region from all known eARID proteins.

Page 3

Drosophila ARID Family Developmental Function 425
A
CD
E
B
of the protein. This issue is made more significant by the observa-
tion that the ARID motif is dispensable for the in vivo function of the
S. cerevisiae SWI1 protein (C. Peterson, personal communica-
tion). Here we provide evidence that the ARID domain is essential
for the in vivo function of retn/dri. In addition, we explore the
function of the separate REKLES domain within retn/dri. Finally, we
show that retn/dri is required for normal mitosis during the syncytial
cleavage stage of Drosophila development.
Results
The ARID Domain is Necessary for DNA Binding
To assess the importance of the RETN/DRI ARID motif, in vitro
mutagenesis was used to generate a retn/dri construct, termed
ARIDδH5, in which the third α-helix of the ARID motif was mutated.
This helix was selected on the basis that it is highly conserved and
includes an invariant tryptophan residue found in all ARID do-
mains, indicating that it should be essential for ARID function. The
RETN/DRI eARID, spanning the sequence from amino acid 258 to
410, has previously been shown to be sufficient for sequence-
specific DNA binding to an oligonucleotide consisting of repeats of
a consensus Engrailed binding site termed NP (Gregory et al.,
1996). To determine whether helix 5 of the RETN/DRI ARID is
necessary for binding to the NP sequence, electrophoretic mobility
shift assays were conducted using both the wild-type ARID (GST-
RETN/DRIARID), as has been previously described (Gregory et al.,
1996), and the mutant ARID (GST- RETN/DRIARIDδH5) lacking the
helix 5 sequence. Using elevated protein levels to detect any
potential binding by the mutant, GST- RETN/DRIARID was able to
retard the available NP6 while GST-RETN/DRIARIDδH5 was not able
to retard the NP6 DNA (Fig. 3A). We conclude that the RETN/DRI
ARID is necessary for DNA binding in vitro.
The ARID Motif is Essential for RETN/DRI Function In Vivo
To test the in vivo importance of the ARID motif, the ARIDδH5
construct was placed downstream of a yeast GAL4 upstream
activator sequence (UAS) and transformed into the Drosophila
germline. Genetic crosses were used to place this construct under
the transcriptional control of endogenous retn/dri enhancers via a
retn/dri::GAL4 enhancer-trap P-element insertion line (Shandala
et al., 1999). When a wild-type retn/dri construct is expressed in this
way, embryonic lethality of retn/dri1 homozygotes is rescued
(Shandala et al., 1999). However, when the ARIDδH5 construct
was tested in the same assay, no rescue was observed. We
conclude that the ARID domain is essential for RETN/DRI function
in vivo.
RETN/DRIARIDδH5 Acts as a Dominant Negative
Failure of the retn/dri ARID deletion mutant to rescue the mutant
phenotype demonstrated the in vivo requirement for the ARID
domain. The RETN/DRIARIDδH5 construct used in this experiment had
the capacity to interact with co-factors involved in RETN/DRI func-
tion. If so, the construct should act in an antimorphic, or dominant
negative, way in the presence of the wild-type gene. To test for the
antimorphic potential of this construct, wild-type RETN/DRI and
RETN/DRIARIDδH5 lacking DNA-binding activity was expressed in the
wild-type retn/dri pattern as described above, but now in the pres-
ence of the wild-type retn/dri gene. No flies were observed to carry
both the UAS::retn/driARIDδH5 transgene and the retn/dri::GAL4
driver, indicating that these genes are synthetically lethal, consistent
with an antimorphic function.
In order to confirm the antimorphic activity of retn/driARIDδH5, we
examined flies heterozygous for an amorphic retn/dri allele which
were also expressing retn/driARIDδH5 in the wing, a non-essential
retn/dri-expressing tissue. Examination of larval retn/dri expression
Fig 2. Developmentally
regulated pattern of
retn/dri expression in
Drosophila embryos
and larvae. Immu-
nochemical staining of
whole mount embryos (A-
C) or larval tissues (D,E)
with rat-anti-DRI (red) (A-
E). Images were captured
using epifluorescence mi-
croscopy, except (D)
which is a confocal im-
age. Samples are coun-
terstained with rabbit
anti-REPO (green) (A),
rabbit anti-βgal (green)
(B,C), mouse anti-CUT
(green) (D) and rabbit anti-
DILP (green) (E). Embry-
onic tissues that express
retn/dri include the posterior region of the developing brain (A, long thin arrows), lateral glia and neurons of the ventral nerve cord (B, white arrowhead),
Drosophila endocrine tissue of the ring gland (corpus allatum) (B, asterisk), pharyngeal muscles (A-C, short thin arrows), rings of the cells at major junctions
along the gut (A-C, short thick arrows) and specific rows of cells that form part of the hindgut epithelium (A,B, yellow arrowheads). Expression in most
of these tissues continues into larval development (e.g. cells in the brain in E, long thin arrows). Most of these cells are of unknown lineages and function.
For instance, we failed to colocolise brain retn/dri positive cells with available markers such as the glial specific homeodomain protein REPO (A) and
Drosophila Insulin-Like Peptide (DILP) (E). (D) Anti-Cut (green) and anti-RETN/DRI (red) staining in the larval wing imaginal disk. White arrowheads indicate
the location of the sense organ precursor cells expressing both Cut and RETN/DRI.

Page 4

426 T. Shandala et al.
UAS:: retn/driARIDδH5 expression was induced in developing
wing cells using the GAL471B enhancer trap, which expresses
GAL4 in the wing-blade anlagen of the wing imaginal disk (Brand
and Perrimon 1993). Expression of UAS::retn/driARIDδH5 under
GAL471B control resulted in variable wing vein defects and
losses of campaniform sensilla (Fig. 4 A-C), in regions consistent
with the endogenous expression pattern of RETN/DRI (see Fig.
2D).
To establish whether retn/driARIDδH5 acts antimorphically, the
dose of endogenous retn/dri was halved by placing GAL471B and
P[UAS::retn/driARIDδH5] in a retn/dri heterozygous mutant back-
ground. Enhancement of the severity of the phenotype in retn/dri1
heterozygotes was observed (Fig. 4, compare D with E), confirming
that the retn/driARIDδH5 construct was acting as an antimorphic
form of the gene.
Other Effector Domains in eARID-Containing Proteins: the
REKLES Domain
The eARID family proteins contain two additional conserved
motifs, the REKLESα and REKLESβ domains (Fig. 1B, Kortschak
et al., 2000). The REKLES motif is located immediately carboxy-
terminal to the eARID. It has a size of a little over 100 amino acids
in the C. elegans and vertebrate RETN/DRI homologs, but in
Drosophila there is a non-conserved insertion of 237 amino acids
within the domain, dividing it into the α and β subdomains. A role
for the REKLES domain in eARID protein function has been
suggested by the observation that C-terminal deletions of the
Bright protein, removing regions that include the REKLES do-
main, fail to tetramerise in gel retardation assays (Herrscher et al.,
1995).
The REKLESβ Region is not Required for Self-Association
To test whether the most conserved region of the REKLESβ
domain was responsible for the homomerisation of RETN/DRI, the
most highly conserved region of the RETN/DRI REKLES domain
(amino acids 792 to 807), which includes the invariant residues of the
REKLESβ region, was deleted by in vitro site-directed mutagenesis.
A western blot of wild-type REKLES (GST–RETN/DRIREKLES) and
mutant REKLES (GST– RETN/DRIREKLESδβ), was probed with in vitro
Fig. 4. Effects of mutant and wild-type retn/
dri expression under GAL471B control in the
wing-blade anlagen of the wing imaginal disk.
(A-F) Photomicrographs of wings from females
of the following genotypes: (A) w1118 (full
wing). (B) P[UAS::retn/driARIDδH518.1]/
w1118;+;GAL471B/+ (full wing), (C1) and (C2)
higher magnifications of (A) and (B) respectively
(bottom boxed area), showing the region around
the anterior cross vein (black arrowhead in C1),
(C3) and (C4) higher magnifications of (A) and (B)
respectively (top boxed area), showing the re-
gion around the twin campaniform sensilla of the
margin (black arrow heads indicate the presence
of campaniform sensilla), (D) w1118; +; GAL471B
P[UAS::retn/driARIDδH5 17.6]/+ and (E) w1118;
retn/dri1/+; GAL471B P[UAS:: retn/driARIDδH5
17.6]/+, showing an enhanced phenotype when
the retn/dri gene dosage was halved.
Fig. 3. In vitro analysis of eARID deletions. (A) Gel electrophoretic
mobility shift assay using GST, GST-RETN/DRIARID and GST-RETN/DRIARIDδH5
proteins incubated with labelled NP6 oligonucleotide. (B) Protein blot
assay for RETN/DRI REKLES domain self-association. GST-RETN/DRIREKLES
and GST-RETN/DRIREKLESδβ were electrophoresed and transferred to nitro-
cellulose membranes. Membranes were probed with either anti-DRI
antibody or radiolabelled full-length RETN/DRI protein, produced by in vitro
transcription and translation. The 29kDa GST protein unfused to RETN/DRI
and probed with radiolabelled RETN/DRI protein is also shown. Note that
the GST-RETN/DRIREKLES and GST-RETN/DRIREKLESδβ proteins appear as
two bands, both of which are less than the size predicted for a GST-RETN/
DRIREKLES protein. The reason for this has not been established.
patterns revealed that it is expressed in a subset of wing imaginal disk
cells that express the homeodomain protein Cut (Fig. 2D). These cut
and retn/dri-expressing cells are known to be the precursors of a
group of sense organs, known as the campaniform sensilla, located
on the wing (Huang et al., 1991; Blochlinger et al., 1993). Additionally,
retn/dri appears to be expressed at low levels throughout the wing
imaginal disk. Thus the wing was chosen as a target tissue for
confirmation of retn/driARIDδΗ5 antimorphism.
AB
AB
C
D
E
1234

Page 5

Drosophila ARID Family Developmental Function 427
shock promoter was used to induce recombination between non-
sister chromatids of the two homologs. This resulted in the generation
of retn/dri mutant cells that no longer carried the dominant female
sterile allele ovoD1 and could therefore produce eggs.
Many early embryos derived from germline retn/dri- clones, and
therefore lacking maternally-derived RETN/DRI, exhibited prolif-
eration defects. The most pronounced of these defects was the
loss of synchrony of cell divisions across the early syncytial embryo
(Fig. 6 A-F, 7A-E). Affected embryos exhibited numerous anaphase
bridges (Fig. 7 C,D) leading to the formation of aberrant nuclei,
including micronuclei and the fusion of nuclei (Fig. 7 C-E). Cell
cycle defects of this nature lead to a loss of cells during the syncytial
divisions (Sullivan et al., 1993), so it is not surprising that some
embryos had fewer cells while still undergoing mitosis (Fig. 6 D-F,
7B,D). However, as previously observed for the zygotic retn/dri
phenotypes, the germline clone embryonic proliferation defects
were highly variable, with approximately 65% of blastoderm stage
embryos exhibiting a normal distribution of nuclei during embryo-
genesis.
Discussion
The ARID family of genes are involved in a wide variety of
transcriptional regulatory mechanisms and in a diversity of bio-
logical processes. The retn/dri gene of Drosophila melanogaster
is perhaps the best characterised ARID family gene with respect
to its role in development. In this report we extend this
characterisation by describing studies of the in vitro and in vivo
functions of conserved domains in the RETN/DRI protein. Gel
mobility shift assays using the Engrailed consensus binding NP6
DNA sequence revealed that disruption of the ARID motif by
deletion of the fifth α-helix disrupts DNA-binding. Similarly, dele-
Fig. 5. RETN/DRI distribution during mitosis RETN/DRI distribution, detected using polyclonal rat
anti-RETN/DRI (red, A-C) during a wave of mitoses occurring in a syncytial embryo. (A) Speckles of
RETN/DRI are observed in interphase nuclei. (B) At prophase/metaphase RETN/DRI can be visualised
both on the chromosomes and diffusely at a low level in cytoplasm. (C) During anaphase, RETN/DRI
appears to have degraded, before a high level of protein re-accumulates in the telophase nuclei. (A’-
C’). The same sections stained forDNA with Hoechst 33258. (A’’-C’’) merged images.
transcription/translated, 35S-methionine-labelled RETN/DRI (Fig. 3B,
arrow). No difference in signal intensity was observed between the
wild-type and mutant forms, indicating that the REKLESβ region is
not necessary for self-association.
retn/dri Lacking the REKLESβ Region is able to Rescue retn/dri
Function In Vivo
In order to examine the in vivo role of the REKLESβ region of
RETN/DRI, rescue experiments similar to those described for full
length retn/dri and the ARID mutant form of retn/dri were performed.
Unlike the ARID motif mutation, constructs lacking the REKLESβ
region could still rescue the mutant phenotype. This shows that
although it represents the most conserved part of the REKLES motif,
the REKLESβ region of RETN/DRI is not essential for function during
stages of embryonic and imaginal development rescued by the
transgene. However, in vivo expression of retn/dri::GAL4>UAS::retn/
driREKLESδβ in a heterozygous background caused a small but
significant reduction in viability (data not shown), suggesting that this
construct may have acted as a mild dominant negative form of the
protein.
ARID Family Genes and Cell Cycle Regulation: A Role for retn/
dri in Drosophila Preblastoderm Mitoses
retn/dri maternal products are uniformly distributed in the early
Drosophila embryo during the preblastoderm cleavage stage, where
nuclei are undergoing mitoses in a syncytium (Fig. 5, Gregory et al.,
1996). At interphase, the RETN/DRI protein is present in discrete
apically located nuclear foci not completely overlapping with the
highly condensed DNA foci stained preferentially by Hoechst 33258
dye (Fig. 5A). At metaphase, RETN/DRI can be visualised both on
the chromosomes and at a low level in the cytoplasm (Fig. 5B). The
cytoplasmic RETN/DRI gradually disappears during the anaphase/
telophase transition, after which high levels
of protein accumulate in the newly re-formed
nuclei (Fig. 5C). After this stage, retn/dri is
expressed in a range of terminally differen-
tiating cells that have ceased proliferation
(Fig. 2). Analysis of the zygotic retn/dri
phenotype showed that retn/dri plays roles
in many and perhaps all of the differentiat-
ing cells in which it is expressed (Shandala
et al., 1999; Shandala, Sibbons and Saint,
unpublished observations).
To determine a function for retn/dri dur-
ing the syncytial divisions, it is necessary to
eliminate the maternal retn/dri product from
the egg. This was achieved using mitotic
recombination to produce retn/dri mutant
clones of cells in the developing oocytes,
using the method of the FLP-FRT-ovoD1
system (Chou and Perrimon, 1996). retn/
dri mutant alleles were recombined onto
chromosomes carrying FRT recombina-
tion sites at the base of chromosome arm
2R and placed in trans to a chromosome
carrying a dominant female sterile muta-
tion, ovoD1, that prevents the production of
eggs. A transgene expressing the FLP
recombinase under the control of the heat
A
A'
A''
B
B'
B''
C
C'
C''