Download A role for the Drosophila Bag-of-marbles protein in

Development 121, 2937-2947 (1995)
Printed in Great Britain © The Company of Biologists Limited 1995
2937
A role for the Drosophila Bag-of-marbles protein in the differentiation of
cystoblasts from germline stem cells
Dennis McKearin* and Benjamin Ohlstein
Department of Biochemistry, University of Texas-Southwestern Medical Center, 5323 Harry Hines Blvd., Dallas, TX 75235-9038,
USA
*Author for correspondence (e-mail: [email protected])
SUMMARY
Cell differentiation commonly dictates a change in the cell
cycle of mitotic daughters. Previous investigations have
suggested that the Drosophila bag of marbles (bam) gene is
required for the differentiation of germline stem cell
daughters (cystoblasts) from the mother stem cells, perhaps
by altering the cell cycle. In this paper, we report the preparation of antibodies to the Bam protein and the use of those
reagents to investigate how Bam is required for germ cell
development. We find that Bam exists as both a fusome
component and as cytoplasmic protein and that cytoplasmic and fusome Bam might have separable activities. We
also show that bam mutant germ cells are blocked in differentiation and are trapped as mitotically active cells like
stem cells. A model for how Bam might regulate cystocyte
differentiation is presented.
INTRODUCTION
germline stem cell divides to produce two daughters (Fig. 1A).
One daughter remains a stem cell while the other becomes a
cystoblast. The cystoblast divides precisely four times with
incomplete cytokinesis at each mitosis to produce a cluster of
sixteen interconnected sister cells called cystocytes. Cystocytes remain connected to one another by stable intercellular
bridges, termed ring canals (King, 1970; Robinson et al.,
1994), that form at the site of contact between the mitotic
cleavage furrow and the spindle equator (King, 1970; Telfer,
1975). One cystocyte becomes the oocyte while its fifteen
sisters become nurse cells that supply the developing oocyte
with nutrients and biomolecules for oocyte maturation and subsequent embryonic development.
Each round of cystocyte mitosis is accompanied by the
growth of a germ cell-specific organelle called the fusome
(King, 1970; Telfer, 1975; Storto and King, 1989; Lin et al.,
1994). It is found as a a spherical dot in the cystoblast and
grows to become an elongated, branched structure in the
cystocyte cluster (i.e. the cyst). Stem cells also contain a
similar organelle which Lin et al. (1994) have designated the
spectrosome. Fig. 1B shows a drawing of a fusome connecting a cluster of eight cells in the germarium of a wasp
(Maziarski, 1913); Drosophila fusomes are essentially indistinguishable from wasp fusomes. The fusome acts as a pole of
each mitotic spindle by capturing a centriole and, in this
capacity, serves to orient the planes of cell division at each
cystocyte mitosis.
Since the syncytium produced by four rounds of cystocyte
divisions represents a clear example of asymmetric differentiation among the mitotic sisters (i.e. oocyte versus nurse cell
differentiation), it has been hypothesized that cystocyte
Stem cell divisions are inherently asymmetric since they
produce mitotic daughters from which specialized cell types
will differentiate (Horvitz and Herskowitz, 1992; Hall and
Watt, 1989). Mutations in genes required for asymmetric segregation of differentiation factors or in genes that are targets
of such factors could block cellular differentiation and produce
mitotic daughters that would inappropriately continue to divide
like stem cells. Examples of these kinds of mutations might
account for some types of leukemias and other neoplasms that
are composed of blast-like cells (Lynch, 1995; Gondos,
1987a,b; Hall and Watt, 1989).
Genetic screens for mutations that cause the accumulation
of specific cell types have proven useful for identifying genes
required for cell differentiation (Horvitz and Herskowitz, 1992;
Hartwell and Weinert, 1989). In many cases, careful analysis
of the points of arrest have provided insight into when the
affected gene products begin to act and what biochemical
functions they might influence to regulate steps of a differentiation pathway.
The Drosophila germ cell lineage
The germ cell lineage of Drosophila (Spradling 1993a,b) is an
attractive model to identify the mechanisms of asymmetric
cytoplasmic distribution during cell division since it affords
genetic and molecular approaches. For example, mutations in
many genes required for germline development are viable but
sterile and mutations that produce apparent arrest of germ cell
development have been identified.
The process of oogenesis in the adult fly begins when a
Key words: Drosophila, bag of marbles, cytoblast, stem cell, cell
cycle, mitosis, germ cell, fusome
2938 D. McKearin and B. Ohlstein
Fig. 1. (A) Schematized view of germ cell lineage from stem cells
through the formation of 16 interconnected cystocytes. (B) A
drawing of the fusome, which appears as the dark branched organelle
at the boundaries of interconnected cystocytes of wasp germ cells.
Five mitotic spindles and the complements metaphase chromosomes
are also illustrated. The original drawing appears in Mazurski, 1915;
this copy was reproduced from Telfer, 1975. (C) A schematic view
of a wild-type germarium from Drosophila melanogaster. The
numbers 1, 2A, 2B, 3 refer to Germarial Regions as described in
King (1970). Region 1 contains stem cells, cystoblasts and 2-cell, 4cell and 8-cell clusters of cystocytes. In region 2A lie the recently
completed 16-cell clusters while region 2B contains lens-shaped
clusters of cystocytes that stretch across the full width of the
germarium as they are becoming surrounded by migrating follicle
cells. Germarial region 3 contains a spherical egg chamber with a
complete complement of follicle cells arranged in an epithelial
monolayer surrounding 15 nurse cells and an oocyte. TF, terminal
filament cells.
divisions are inherently asymmetric with respect to distribution
of cytoplasmic determinants (King, 1970; Spradling, 1993b).
In this context, it has been exciting to learn that fusome distribution at the time of cystoblast mitosis is asymmetric suggesting that the fusome might contribute to marking cell fate
(Lin and Spradling, 1995).
Recent investigations have established that Spectrin (Spc;
Lee et al., 1993) and Hu-li tai shao (Hts) protein, a Drosophila
adducin-like protein, are fusome components and that hts is
required for normal fusome assembly and cyst formation (Lin
et al., 1994). Cystocytes in hts mutant females fail to complete
four rounds of mitosis and consequently produce egg chambers
that contain between two and ten nurse cells and frequently
lack an oocyte (Yue and Spradling, 1992). Finally, no fusome
forms in hts germ cells (Lin et al., 1994).
Germ cell divisions take place in the germarium at the
anterior tip of the ovary (Fig. 1C). The germarium has been
divided into four regions (Mahowald and Kambysellis, 1980)
that correspond to the various stages of cyst development.
Stem cell divisions occur at the anterior end in region 1 and
cysts are forced posteriorward as they mature. Upon
completion of the fourth round of mitosis, a cyst becomes
surrounded by a monolayer of somatic cells that migrate
inward from the somatic cells that line the walls of the
germarium.
The bag-of-marbles (bam) gene was identified in a Pelement mutagenesis screen (Cooley et al., 1988) as a gene
required for progression through the early steps of the germ
cell lineage (McKearin and Spradling, 1990). The predicted
Bam sequence suggested a novel protein except for a weak
similarity to the Drosophila Otu protein and a motif near the
protein’s C terminus that matched the consensus for PEST
domains (McKearin and Spradling, 1990). The presence of a
PEST domain suggested that Bam might have a short intracellular half-life (Rogers et al., 1986).
Mutations in the bam gene arrest germ cell differentiation at
very early stages of both oogenesis and spermatogenesis in a
manner that suggests bam is required for the differentiation of
germline stem cell daughters (McKearin and Spradling, 1990).
In the absence of bam function, these cells continue to divide
like stem cells producing germ cell tumors that characterize the
mutant phenotype. In this paper, we report results that corroborate these hypotheses. We show that Bam is a fusome protein.
We find that germ cells require Bam to progress beyond the
cystoblast and, in its absence, germ cells become trapped as
mitotically active non-differentiating cells. Finally, we present
evidence that the initial requirement for Bam is not dependent
on an intact fusome. Instead we suggest that it is associated
with cytoplasmic Bam that begins to accumulate when the cystoblast initiates mitosis and disappears after the fourth
cystocyte division.
MATERIALS AND METHODS
Fly stocks
All fly stocks were maintained under standard culture conditions.
y;ry506 and w1118 stocks (Lindsley and Zimm, 1992) were used as
wild-type controls. bam86 mutation is a near complete deletion of the
bam gene coding sequence and will be described elsewhere. The hts1
and hts2 mutations have been described by Yue and Spradling (1992)
and Lin et al. (1994). Balancer chromosomes SM6b and TM3 are
described in Lindsley and Zimm (1992).
hts;bam mutant flies were constructed by crossing hts1/SM6b;ry506
males to +/+;bam86/TM3 virgin females. hts1/+;bam86/+ males were
recovered in the F1 progeny and mated to +/SM6b;+/TM3 virgin
females. Sib-mating hts1/SM6b;bam86/TM3 males and virgin females
recovered from progeny established a stock from which hts1;bam86
Role of Drosophila Bag of marbles protein 2939
females could be isolated for ovary examinations. The same mating
strategy was used to establish stocks of hts2/SM6b; bam86/TM3.
Preparation of antibodies
Bacterially expressed protein was produced from a pET vector
(Novagen) containing a fragment of the bam gene. The recombinant
protein carried a 10-residue poly(His) tag that was used for Ni2+column affinity purification of the fusion protein according to manufacturers instructions (Novagen). The PmlI-SacII fragment from a
near full-length bam cDNA clone was ligated as a blunt-ended
fragment into BamHI site of the pET16b polylinker; BamHI ends were
made blunt by Klenow fill-in prior to ligation. The resulting clone
encodes an open reading frame that includes 23 vector-encoded amino
acids and residues 5-140 of the predicted Bam sequence (McKearin
and Spradling, 1990). A second His-tagged recombinant protein was
constructed by ligating the PmlI-HindIII cDNA fragment into
pET16b. This fusion protein contains codons 5-405 of the predicted
Bam polypeptide.
Affinity-purified recombinant protein was injected into mice in the
amount of 75 µg protein/injection. Antisera were tested for
immunoreactivity on immunoblots against fusion protein, fulllength
in vitro translated Bam and ovarian extracts from wild-type and bam86
flies. Antisera recovered from mice injected with either recombinant
Bam antigen produced similar immunoreactivities. A total of seven
mice produced Bam-positive antisera (15 total immunized).
Immunoblots were performed as described previously (Christerson
and McKearin, 1994); antisera were used at 1:7500 dilution.
Secondary antibodies conjugated to alkaline phosphatase and the
chemiluminescent substrate CSPD were used according to manufacturers instructions (Tropix) to detect membrane-bound antigens.
Immunohistochemistry
Ovaries from w1118 or bam mutant females were dissected into
Drosophila Ringer’s and the tips of the ovaries were teased open to
enhance antibody penetration of the tissue. Ovaries were fixed for 2030 minutes at 24°C in 0.3 ml of 1× PBTF/0.9 ml heptane. 1× PBTF
is 3 mM NaH2PO4, 7 mM Na2HPO4, 130 mM NaCl, 0.125% TWEEN
20 and 4% formaldehyde (methanol-free; Pella Scientific). Ovaries
were then rinsed 3× in PBT and washed 3× for 10 minutes/wash in
PBT. The tissues were incubated for 60 minutes in PBTA (PBT+1.5%
BSA). This solution was replaced with fresh PBTA containing the
desired antibodies at an appropriate dilution (anti-Bam antisera were
used at 1:2000) and incubated overnight at 4°C with gentle agitation.
To test antisera specificity by soluble antigen competition, antisera
(1:1000) were incubated with 0.5 µg of affinity-purified His-tagged
Bam or 0.5 µg of His-tagged fusion protein for another Drosophila
ovarian protein for 30 minutes at 24°C before addition of fixed
ovaries. The reactions were performed as described above from this
point on. We have noted that detection of Bam antigen, especially
fusome associated Bam, is sensitive to fixation conditions and perhaps
other factors since a fraction of germaria in a given sample are not
stained by Bam antisera.
BrdU incorporations were carried out essentially as described in
Gönczy (1995). Briefly ovaries or testes were dissected from animals
into 1× Drosophila Ringer’s solution. When sufficient organs had
been collected, the buffer was removed and replaced with fresh
Drosophila Ringer’s/20 µM BrdU and incubated for the appropriate
time in the dark. At the end of the period of incorporation, the
Ringer’s/BrdU was removed, organs were washed three times for 10
minutes in fresh Drosophila Ringer’s and fixed in 1 volume of 1×
Drosophila Ringer’s/4% formaldehyde and 3 volumes of heptane with
shaking for 20 minutes. After washing the fixation solution from the
organs, they were incubated for 30 minutes in 2 N HCl. HCl was
removed and the tissues were neutralized by incubating for 2 minutes
in 100 mM Na2B2O4. The organs were then washed three times for
10 minutes each in 1× PBS/0.1% Triton X-100. Incorporated BrdU
was detected by incubating the processed organs with anti-BrdU
according to manufacturer’s suggestions and detecting with antimouse secondary antibodies conjugated to the fluorophore Cy3
(Jackson Immunoresearch Inc.).
Microscopy
Microscopy was performed on a Zeiss Axiophot microscope equipped
with phase and differential interference contrast objectives. Confocal
images were collected using the MRC 600 system (Bio-Rad Microsciences Division) attached to a Zeiss Axioplan microscope. Confocal
images were photographed using a Polaroid FreezeFrame Video
Recorder. The images recording ovaries double-labeled with antiSpectrin and anti-Bam shown in Fig. 4C were collected on a BioRad
MRC1000 confocal microscope. To examine the cytology of ovaries,
tissue was dissected in 1× PBS, and fixed with 4% paraformaldehyde
in 1× PBS; ovarioles were teased apart to expose the egg chambers.
Electron microscopy was performed on a JEOL 120 kV electron
microscope at the UT-Southwestern Center for Microscopy.
RESULTS
Antibodies against Bam
Previous investigations suggested that Bam was required for
progression of germ cells beyond the cystoblast/2-cell
cystocyte stage (McKearin and Spradling, 1990). As a first step
to identify specific aspects of early germ cell differentiation
that are dependent on Bam activity, we prepared antibodies
against Bam. Polyclonal antisera were collected from mice
immunized with recombinant proteins that contained segments
of the predicted Bam sequence (see Material and Methods).
Bam-positive antisera reacted with a single band of ~Mr 55,000
(Fig. 2, lane 3) in extracts derived from coupled in vitro tran-
Fig. 2. Immunoblot of anti-Bam.m2 antiserum (1:7500) reacted
against lanes containing protein from bam86 ovaries (lane 1), wildtype ovaries (lane 2), in vitro translation (IVT) extract programmed
with bam mRNA (lane 3), IVT extract programmed with mRNA for
an irrelevant ovarian protein (lane 4) and yeast cells expressing a
LexA-Bam fusion protein (lane 5). The very slight difference in
migration between the bands in lanes 2 and 3 probably does not
represent authentic molecular mass differences since the shift is not
reproducible on other immunoblots of similar samples. The faint
band visible in lane 3 at ~ 30,000 Mr represents reaction of the
mouse polyclonal serum against a contaminating protein; a band of
the same Mr and that comigrated with lysozyme was detected in
molecular mass marker lanes (not shown). The approximate
positions of molecular mass marker proteins is indicated on the right
side of the figure.
2940 D. McKearin and B. Ohlstein
µ
µ
µ
µ
scription-translation extracts that had been programmed with a
bam cDNA clone. These antisera also reacted with a band of
~Mr 80,000 (Fig. 2, lane 5) in extracts derived from yeast cells
expressing a LexA-Bam fusion protein; the larger size is consistent with the predicted molecular mass of the LexA-Bam
protein.
Fig. 2, lane 2 shows reaction of Bam antiserum against
extracts from wild-type ovaries and shows that Bam appears
as a prominent band at ~Mr 55,000. This is slightly larger than
the calculated molecular weight of ~Mr 50,000 (McKearin and
Spradling, 1990) but comigrates with the protein produced
from in vitro translation extracts (Fig. 2, lane 3). Reaction of
the antisera against immunoblots containing protein extracts
µ
prepared from testes and 0-3 hour embryos showed that a
single Mr 55,000 protein band was detected (data not shown).
The presence of Bam in testes and embryos recapitulates bam
mRNA expression in those tissues (McKearin and Spradling,
1990). In addition, genetic investigations of bam mutations
have documented that Bam is required for normal spermatogenesis (McKearin and Spradling, 1990).
Immunolocalization characterization
Incubation of anti-Bam antisera with fixed ovaries revealed
that Bam could be found in two distinct cellular locations.
Immunoreactive Bam was found in the cytoplasm (designated
BamC) of mitotically active cystocytes and in the fusome/spec-
Role of Drosophila Bag of marbles protein 2941
Fig. 3. Anti-BamF immunostaining. In the succeeding panels, the
anterior end of the germarium is to the left. (A) A stack of optical
sections that shows the reaction of an anti-Bam antiserum that
detected both BamC and BamF protein (anti-Bam2.0). Cytoplasmic
staining (BamC) is clearly visible in a cluster of 8 cystocytes (5
visible) towards the top of the micrograph (labeled no. 3) as well as
fainter staining in a 2-cell cluster (no. 2) towards the bottom.
Fusome-associated antigen appears as bright spherical dots at the
germarium anterior end in a putative stem cell (no. 1), as a slightly
elongated dot connecting the 2 cystocytes labeled no. 2, and as a
branched fusome (no. 4). (B) The reaction of anti-Bam.m13
antiserum against wild-type ovaries. The germarial anterior end is
marked by an arrow. This antiserum reacts strongly with fusomeassociated Bam but only weakly with cytoplasmic Bam, which is not
detectable in this micrograph. The high magnification series of
optical sections displays the reaction against BamF antigen in
germarium region 1 and shows staining of a spectrosome in a stem
cell (closed arrow) and most of the fusome of an 8-cystocyte cluster
(open arrow). (C) A wild-type germarium double stained with antiSpectrin and anti-Bam.m13 antisera. Spectrin staining is red while
Bam staining is green. The region of overlap, which is the
spectrosomes and fusomes, are therefore yellow. This micrograph
was taken at approximately the same magnification as A. (D) A stack
of 21 optical sections of a wild-type germarium plus a stage 2 egg
chamber stained with anti-Bam.m13. Cytoplasmic Bam, as detected
with this particular antiserum, is labeled BamC (arrow). The faint,
string-like staining in the lower right corner of the micrograph are
fusome remnants that remain in a stage 2 egg chamber that has
emerged from the posterior end of the germarium. (E) The reaction
of the antiserum is specific for Bam. Ovaries (Ov and arrows) from
bam null mutant females (bam86 flies) were reacted in the same tube
as wild-type ovaries with the anti-Bam.m13 antiserum used in D. All
reaction against ovarian antigens was eliminated in bam86 ovaries.
(F) Reaction of anti-BamF antiserum against a wild-type testis; again
fusomes are strongly detected. The spermatocyte fusomes are thinner
than those found in cystocytes and the bulbous ends tend to be more
exaggerated. Note also that Bam antigen appears in many small dots
that are not associated with formal fusomes. The striated structure in
the center of the image (arrow) is a bundle of elongated sperm that
fills the lumen of the testis; fluorescent signal from these sperm tails
represents background level staining.
trosome (designated BamF) in all germ cell in the germarium
and in nurse cells of young egg chambers. Independently
derived antisera reacted with antigen at both subcellular
locations although the relative strengths of immunoreactivity
varied. Antisera specificity for Bam in both cellular compartments was confirmed by demonstrating that cytoplasmic and
fusome staining was eliminated in ovaries from females that
were homozygous for a null allele of bam, bam86 (Materials
and Methods). In addition, both staining patterns were eliminated when antisera were preadsorbed with affinity-purified
histidine-tagged Bam produced in E. coli (see Material and
Methods). Preincubation of the same antisera with an irrelevant affinity-purified histidine-tagged protein had no effect on
anti-BamC or anti-BamF staining. Our current data about
BamC and BamF do not allow us to determine if proteins found
in these cellular locations represent covalently modified
isoforms of Bam or differential localization of a single form of
the protein. We will describe the distinct immunocytochemical staining patterns separately.
Fig. 3A shows reaction of Bam antiserum with a wild-type
germarium. Bam was detected in the spectrosome of a probable
stem cell (1) and the antiserum stained both BamC and BamF
in cystocyte clusters 2 and 3. For example, in the pair of cystocytes labeled ‘2’ in the micrograph, Bam was detected as a
halo of antigen distributed throughout the cytoplasm and was
concentrated in the fusome that connects the two cells. Cluster
3 (probably an 8-cell cyst) contained a high concentration of
BamC and the intensity of staining in the cytoplasm obscured
the signal from fusome-associated Bam. Finally a branched
fusome was stained (4) in a cluster of cystocytes in germarial
region 2a. Note that BamC protein was no longer found in
these cystocytes.
A more complete analysis of BamF protein is presented in
Fig. 3B-F using an antiserum that recognizes BamF strongly
but BamC only weakly. High magnification optical sections
through the anterior of a wild-type germarium showed that
BamF was recognized in a probable stem cell spectrosome
(Fig. 3B, closed arrow) and the fusome of an 8-cell cyst (Fig.
3B, open arrow). Fusome branches terminated in an intensely
stained knob within the cytoplasm of each cystocyte.
Confirmation that the structure stained by anti-Bam.m13 in
wild-type germaria was the fusome was obtained by co-localization of staining with anti-Spectrin antibodies (Fig. 3C). In
Fig. 3C, a wild-type germarium has been stained with both
anti-Spectrin (shown in red) and anti-Bam (shown in green)
antisera. All spectrosomes and fusomes in the micrograph
appear yellow because Bam and Spectrin staining overlap completely in the spectrosome/fusome. Further evidence of fusome
recognition by Bam antibodies was obtained by noting that
reaction of anti-BamF.m13 antiserum against hts1 germ cells,
which do not form fusomes (Lin et al., 1994), detected only
cytoplasmic Bam antigen (data not shown).
Fig. 3D presents 21 superimposed optical sections through
a wild-type germarium stained with anti-Bam.m13. Fusomeassociated Bam was detected in all cystocytes throughout the
germarium; remnants of the fusome can be seen in individual
egg chambers as mature as stage 2 (bottom right of Fig. 3D).
Staining appeared most intense at the knob-like ends of the
fusome branches, which may reflect greater accessibility of the
antigen at those termini or may be due to an enrichment of Bam
at these sites. Staining of BamC appeared as a diffuse cloud of
antigen (arrow labeled BamC) and was confined to cystocytes
in germarial region 1.
Fig. 3E shows reaction of the anti-Bam.m13 antisera against
bam86 ovaries to demonstrate that immunoreaction is specific
for Bam. The failure to detect any signal above background
can be attributed to specific loss of Bam antigen, and not
simply elimination of fusomes, since we can show by alternate
means that bam mutant germ cells contain fusomes (Figs 5, 6).
In testes, anti-Bam antisera also reacted with cytoplasmic
(not shown) and fusome-associated Bam (Fig. 3F). This is consistent with genetic results that indicated that bam+ is required
for spermatogenesis and that the protein might serve the same
or a closely related function in both male and female germ cells
(McKearin and Spradling, 1990).
Fig. 4A,B shows reaction of one of the Bam antisera (antiBam.m2) against wild-type ovaries. Anti-Bam.m2 reacts only
with BamC in wild-type ovaries but recognizes fusome-associated Bam in at least three mutant backgrounds (not shown). We
hypothesize that this reflects increased access of antibodies in
this serum to Bam epitopes in the fusomes of those mutants; data
from experiments that probe fusome structure in various mutant
genotypes using Bam antisera will be published separately.
2942 D. McKearin and B. Ohlstein
µ
µ
Fig. 4. Staining of cytoplasmic Bam in wild-type germaria. The
anterior end of germarium is in the upper left and is marked by a
curved arrow in A and a closed arrow in B. Anti-Bam.m2 antiserum
was reacted against wild-type ovaries and prepared for confocal
viewing using fluorescent secondary antibodies. (A) Two Bampositive cells; outlines of the other cells in this germarium can be
seen faintly and represent background level signal. (B) Eight
cystocytes in region 1 that are positive for BamC; the beginning of
region 2a is marked by an open arrow. The micrograph presented in
B was overexposed so that the outline of the germarium can be seen
clearly. In all cases examined, BamC-positive cells were confined to
germarium region 1. Cells in the anteriormost position, putative stem
cells, were never positive.
The specificity of anti-Bam.m2 for BamC protein in wildtype germ cells can be exploited to document clearly the distribution of cytoplasmic antigen. BamC is distributed throughout the cytoplasm of the cystocytes and is largely excluded
from the nuclei of those cells. The number of Bam-positive
cystocytes varied for different germaria (Fig. 4A,B) from as
few as 1 positive cell (rarely) to as many as ~20. Most
commonly a germarium had 6-10 positive cells. Cells that were
positive for BamC were always confined to germarial region
1, the region that contains stem cells, cystoblasts and mitotically active cystocytes (see Fig. 1C). The most anterior cells
in the germarium, the probable stem cells, were never positive
for BamC antigen. Likewise cystocytes in region 2a and
beyond, where completed 16-cell cysts are found, lacked
BamC antigen. Thus, accumulation of BamC protein could be
correlated with mitotically active cystocytes. Approximately
10% of germaria were negative for BamC antigen, suggesting
Fig. 5. Anti-Fusome staining in bam86 mutants with anti-Hts.
Anterior end of the germarium is in the upper left corner of A and B.
(A) Wild-type fusomes visualized by anti-Hts antisera. The fusomestaining pattern is essentially indistinguishable from that described
for anti-Bam antisera; note that anti-Hts also stains the membrane
cytoskeleton of the somatic and germline cells. An example of a
small dot of Hts-positive material not associated with fusome or
membrane cytoskeleton is indicated by the open arrow. Closed arrow
indicates the anterior end of germarium immediately adjacent to the
terminal filaments. Also see Lin et al. (1994) for a description of
anti-Hts immunohistochemistry. (B) The reaction of Hts antibodies
against bam86 germarium; the micrograph is a projection of optical
sections assembled through 5 µm of tissue. The fusomes recognized
by Hts antiserum appear as dots or dumbbell-shaped fusomes (dbF).
This germarium also shows an example of a highly elongated fusome
(long arrow) that can sometimes be found connecting bam86 germ
cells.
that some germaria did not contain any cystocytes in mitotically active stages at the time of ovary dissection. Experiments
following the incorporation of BrdU into wild-type germ cells
(see below) and independent experiments carried out by H. Lin
(personal communication) and Carpenter (1981) support the
hypothesis that some germaria lack mitotically active cystocytes at any given time.
Effects of bam mutation on fusome formation
Lin et al. (1994) demonstrated that formation of an intact
fusome was dependent on the Hts since strong mutations in the
hts gene eliminated the fusome from germ cells. Since Bam
was also a fusome protein, we tested the effects of elimination
Role of Drosophila Bag of marbles protein 2943
of Bam function on fusome structure by reacting anti-Spc and
anti-Hts antibodies against bam mutant ovaries.
Anti-Hts staining of wild-type ovaries revealed that Hts is
in spectrosomes and branched fusomes in germ cells and that
Hts is associated with the membrane cytoskeleton in somatic
cells (Fig. 5A; also Lin et al., 1994). The open arrow in Fig.
5A indicates small dots of Hts antigen that may represent
aggregates of fusome proteins. Similar small dots can be found
in cystocytes and spermatocytes using anti-Bam antisera.
Reaction of anti-Hts (Fig. 5B) or anti-Spc (not shown)
against bam86 ovaries revealed that nearly every cell contained
a fusome. Most often (~75%) the organelles were spherical
while ~25% of fusomes appeared as two spheres apposed so
closely that they touched (i.e. dumbbell-shaped; dbF in Fig.
5B). Based on the appearance of fusomes in wild-type 2-cell
cysts (for example, see cluster 2 in Fig. 4A), we suspected that
dumbbell-shaped fusomes in bam86 samples probably passed
through ring canals connecting two cells. Sometimes, long
unbranched fusomes were found scattered throughout bam
germaria; an example can be seen in Fig. 5B (arrow). Counterstaining nuclei with DAPI (not shown) or by locating nuclei
as Spectrin-negative spheres in the faintly stained cystoplasm
of bam germ cells (Fig 5B) revealed that such linear fusomes
stretched between two cells. Older bam mutant cells occasionally were negative for anti-Hts or anti-Spc fusome staining
probably reflecting eventual loss of the organelle in the oldest
cells.
Electron microscopy
Strikingly, branching fusomes that would mark multicellular
syncytia of cystocytes were not detected in bam mutant germ
cells. This suggested that bam cells were unable to execute successive mitoses with incomplete cytokinesis to produce interconnected clusters of cells. However, there remained the
formal possibility that bam cells divided with incomplete
cytokinesis but that the fusome did not mark the event by
extending branches through newly formed ring canals.
Therefore we examined fusomes and ring canals in bam mutant
cells in the electron microscope. A second goal of this analysis
was to examine the ultrastructure of bam fusomes to determine
if the organelle was normally formed.
Fig. 6 shows a transmission electron microscope (TEM)
micrograph of several germ cells and a fusome in germarial
region 1. The fusome in Fig. 6A appears as a spherical region
devoid of mitochondria adjacent to the nucleus (labeled ‘fu’ in
Fig. 6B) The organelle is characterized by a tangled network
of membranous material that fills its central region and by the
exclusion of other cellular organelles such as mitochondria.
Close examination has revealed that the fusome also excludes
ribosomes such that the density of ribosomes within the fusome
is only 10% that of surrounding cytoplasm (King, 1970). This
aspect of fusome ultrastructure is difficult to see in wild-type
cells because of interference by the membranous network. In
some EM thin sections of extended and branched fusomes, the
organelle has a fibrillar appearance similar to that depicted in
the drawing in Fig. 1B; the biochemical composition of the
fibrils is unknown.
Electron micrographs of germ cells from bam86 ovaries
showed cytoplasmic regions that resembled fusomes because
of their spherical shape (as predicted by anti-Hts staining) and
exclusion of ribosomes and organelles (Fig. 6C,D). Examina-
tion of hundreds of bam germ cells in thin sections revealed
that (1) most bam mutant fusomes are simple spheres and are
contained within a single cell and (2) dumbbell-shaped
fusomes that pass through ring canals can be identified in
bam86 germ cells (see inset, Fig. 6C). Significantly, whenever
a ring canal connected bam cells, a fusome filled the canal’s
lumen. Thus, the relationship between ring canal formation and
penetration of the canal by a fusome was preserved in bam
mutant cells.
On the basis of anti-Hts and anti-Spc-immunostained bam
ovaries, we had concluded that branched fusomes did not form
in bam cells. The fact that only single cells with spherical
fusomes or pairs of cells connected by a ring canal and a
fusome could be found in thin sections suggested that most
bam cells were not connected to a neighboring cell and that
none were connected to more than one cell. This conclusion
was corroborated by serial reconstruction of thin sections of
bam germaria. Stacked sections passed through 1-1.5 µm at
intervals of 0.35 µm; in one case, sections that passed through
6 µm of a bam germarium were examined. All fusomes seen
(>40) in these sections were either spherical and contained
within one cell or passed through a single ring canal that
connected a two cystocyte pair.
The examination of bam fusomes in the EM allowed comparison of fusome ultrastructure to that found in wild-type
germ cells. Lin et al. (1994) had previously examined fusome
morphology in a hypomorphic bam mutant, bam1, and
concluded that bam fusomes had normal ultrastructure.
However, in bam86 mutant cells, the density of the vesicular
material in the center of the fusome was greatly reduced
compared to wild type (Fig. 6C). The vesicular material present
in the mutant fusomes was sparsely scattered within the body
of the fusome or was occasionally missing altogether. In wildtype germ cells, the density of the vesicular material in fusomes
increases as cystocytes mature. Thus stem cell spectrosomes
have the lowest density of vesicles of all germ cells. However,
the amount of the membranous component in bam fusomes
was reduced even below that seen in wild-type spectrosomes.
Furthermore, bam fusomes had low vesicular density irrespective of their position in the mutant germarium. Thus aberrant
fusome ultrastructure did not correlate with the age of bam
germ cells. The difference between fusome structure in bam1
germ cells (Lin et al., 1994) and bam86 fusomes is probably
because bam86 represents a complete loss of bam+ function
while bam1 is partial loss-of-function allele (Ohlstein and
McKearin, unpublished).
Functional analysis by epistasis tests
The experiments described above demonstrated that, like the
Hts, Bam is a fusome component. Yet mutations in these two
genes produced remarkably different phenotypes. Ovaries in
hts− females contain egg chambers with fewer than sixteen cystocytes, frequently produce nurse cells and occasionally an
oocyte (Yue and Spradling, 1992). Ovaries from bam− females,
in contrast, contain a proliferating population of undifferentiated germ cells. The fact that Bam was found outside of the
fusome (i.e. BamC) suggested a possible explanation for phenotypic differences; perhaps Bam’s activity is not limited to
the fusome. This hypothesis was tested by making females
mutant for both hts (which would genetically eliminate the
fusome; Lin et al., 1994) and bam. If only fusome-associated
2944 D. McKearin and B. Ohlstein
Fig. 6. EM micrographs of wildtype and bam fusomes. (A) A
transmission electron microscope
micrograph at 10,000×
magnification of parts of several
cystocytes in anterior germarial
region 1. A fusome that lies
between the nucleus and the plasma
membrane occupies the center of
the micrograph. The inset in the
lower left corner of A shows a ring
canal (the letter R marks the lumen
of the canal; the rims of the canal
lie above and below the letter R)
and the accompanying fusome that
passes through the canal’s opening.
(B) A line drawing of the
micrograph in A highlights the
fusome (Fu) and several
surrounding organelles. (C,D) A
transmission electron microscope
micrograph and accompanying line
drawing of fusomes found in bam86
germ cells. One fusome can be seen
in the center of the micrograph near
the nucleus as a spherical region of
reduced electron density. A second
fusome can be seen in the upper
portion of the micrograph. It
appears as an oblong and is again
marked by reduced electron density.
The cells carrying these fusomes
are a considerable distance from the
germarial tip yet the fusomes are
confined to single cells. Note that
bam86 fusomes manifest a very
sparse density of vesiculated
bodies. The inset shows a high
magnification view of a ring canal
produced within a bam86 germ cell
and the fusome that passes through the canal. The fusome seen in the inset is an example of a fusome that would be detected as a dumbbellshaped organelle by anti-Hts immunohistochemistry (see Fig. 4).
Bam can be active, then hts; bam ovaries will resemble hts
ovaries since this represents the fusome− phenotype. Therefore
we constructed flies that carried mutant alleles of both hts and
bam (Materials and Methods). Examination of the ovaries of
such hts; bam double mutants revealed that they were indistinguishable from those produced by single mutant bam
females.
Thus, in genetic terms, bam is epistatic to hts. Stated differently, loss of bam+ function is epistatic to loss of the fusome
and suggests that Bam protein is active independent of association with the fusome. A clear caveat to this interpretation is
the extent to which hts mutations cause a complete loss of
fusome function. Although the female sterile hts1 and hts2
alleles represent only partial loss of Hts function (Yue, Lin and
Spradling, personal communication), Lin et al. (1994) showed
with anti-Spc antibodies and TEM that no fusome could be
detected in the germ cells of hts1 and hts2 animals. Furthermore, we confirmed by anti-Spc antibodies that no fusome
could be detected in the germ cells of hts; bam86 double
mutants.
bam mutant germ cells are mitotically active
The results of hts; bam double mutants showed that bam was
epistatic to hts but did not order the timing of action of the two
genes. However comparison of the phenotypes produced by
mutant hts and bam alleles suggest that germ cell differentiation is disrupted earlier by loss of bam function. For
example, hts cystocytes divide only a limited number of times
and can make nurse cells and occasionally an oocyte while bam
germ cells overproliferate and don’t produce any differentiated
cell types.
Several hypotheses could explain the germ cell hyperplasia
that characterizes bam ovaries: (1) loss of Bam might cause
germline stem cells to divide much more frequently than wildtype counterparts, (2) bam mutation might cause a limited
expansion of the number of cells within a germarium that serve
as stem cells or (3) most or all bam stem cell daughters might
continue to divide indefinitely. To distinguish between these
models, we assayed the mitotic activity of bam mutant cells by
following DNA replication-dependent incorporation of the
nucleotide analog, BrdU (Materials and Methods). Fig. 7
Role of Drosophila Bag of marbles protein 2945
recently completed DNA replication. Genome polyploidization
cannot explain BrdU incorporation since DAPI staining of
nuclear DNA and measurement of nuclear size (not shown)
indicated that bam cells were diploid. Additional experiments
demonstrated that the percentage of cells within a bam cyst that
were labeled by BrdU was dependent on the length of time that
the ovaries were incubated in the reagent before fixation. These
results suggest that cells from any position within a bam cyst
can accomplish mitosis and argue against the idea that only a
small subset of bam cells is responsible for germ cell proliferation.
DISCUSSION
Fig. 7. Mitotic S-phase assayed by incorporation of BrdU. (A) The
extent of incorporation of BrdU in wild-type ovaries after a 1.75
hour incubation in the reagent. In this example, cystocytes in region
1 are positive as well as nurse cells and follicle cells in more mature
egg chambers positioned posterior to the germarium. Small BrdUpositive dots are scattered throughout the tissue; these may represent
incorporation of the nucleotide analog into mitochondrial genomes
although this hypothesis has not been tested directly. The anterior
end of the germarium is in the upper left corner of the micrograph.
(B) BrdU incorporation into bam cells. Mutant germ cells throughout
a bam germarium and a tumorous cyst have taken up BrdU. An
arrowhead marks the division between the germariuma and tumorous
cyst. As in A, the anterior end of the germarium is in the upper left.
shows an example of BrdU incorporation in wild-type and
bam86 ovaries that were incubated together for 1.75 hours in
the reagent. A cluster of BrdU-positive cystocytes can be seen
in region 1 of a wild-type ovariole in Fig. 7A ; stem cells in
this germarium did not take up BrdU in the course of the incubation. BrdU-positive nurse cells and follicle cells in more
mature egg chambers were probably undergoing DNA replication associated with polyploidization of their genomes.
Small BrdU-positive dots could also be detected. We hypothesize that these represent incorporation of the nucleotide
analog into mitochondrial genomes since we could occasionally place the dots in cytoplasm when the corresponding
nucleus (especially nurse cell nuclei) was also labeled (not
shown).
Incubation of bam mutant ovaries in BrdU produced
evidence of very active DNA synthesis. An example of a bam86
germarium and tumorous egg chamber is shown in Fig. 7B and
demonstrates that cells in every region of the germarium and
egg chamber were actively replicating genomic DNA or had
Cellular compartmentalization of Bam protein
We have shown by immunocytochemistry that anti-Bam antibodies can recognize Bam in two distinct cellular compartments, the fusome (BamF) and the cytoplasm (BamC). One
explanation that we have considered is that cytoplasmic Bam
is a pool of protein that is depleted as the fusome expands. The
preference that different antisera showed for BamC or BamF
could be explained if some epitopes exposed in cytoplasmic
Bam were less accessible in fusome-associated antigen. An
alternative model is that cytoplasmic Bam carries post-translational modifications that distinguish it from fusome-associated
Bam. Such modifications could have special significance in
light of the results obtained with hts; bam double mutants
which suggested that at least some aspects of Bam function do
not require the fusome. However, our present data do not allow
us to distinguish between the alternate models.
BamF protein was detectable in stem cells despite the fact
that bam transcripts were not detectable in these cells
(McKearin and Spradling, 1990). Perhaps Bam in stem cells is
derived from bam gene transcription in preadult germ cells or
in stem cells at levels below those detectable by our methods
of RNA in situ hybridization. We have preliminary evidence
that bam is transcribed at very low levels in both preadult germ
cells and adult stem cells since β-Galactosidase activity is
detectable in both cell types from lacZ genes driven by a bam
promoter (Ohlstein and McKearin, unpublished data).
Bam is required to promote incomplete cytokinesis
Most bam cells were single and unconnected to any neighbors,
and remained mitotically active long after they were born (Fig.
7B). Occasionally, interconnected pairs of cells could be identified by the morphology of their fusomes or seen directly in
EM micrographs. This mixture of single and pairs of cells
could arise if most bam cell divisions were complete but a
fraction produced stable pairs. If this explanation is correct, the
fact that branched fusomes were not seen would suggest that
an interconnected pair of bam cells became mitotically inactive
or at least incapable of additional incomplete cytokineses. An
alternative and perhaps simpler explanation for a mixture of
single and paired cells is that all bam germ cell divisions
include formation of ring canals but the canals between mitotic
sisters break; pairs of cells could simply represent a steadystate level of sisters that have not yet broken apart. We believe
that the examples of elongated fusomes stretched between two
bam germ cells (Fig. 5B) might represent pairs of bam cells in
the process of pulling apart at their ring canal. Precedence for
2946 D. McKearin and B. Ohlstein
Stem cell
Stem cell
Cystoblast
2 cystocytes
Bam required for switch from SC to Cb
mitosis; promotes incomplete cytokinesis
and activates fusome growth.
4 cystocytes
8 cystocytes
16 cystocytes
Destruction of BamC blocks fusome
growth and signals Cy withdrawal from
mitotic cell cycle.
such divisions has been described in Drosophila; ring canal
formation and subsequent breakage accompanies divisions
between wild-type spermatogenic stem cells and spermatoblasts (Hardy et al., 1979) and oogenic stem cells and cystoblasts (Carpenter, 1981). Although we cannot currently distinguish between the alternate views for how bam mutant cells
divide completely, either explanation implies that Bam is
required to promote stable incomplete cytokinesis during cystoblast division.
How might Bam regulate incomplete cytokinesis in a cystoblast? A structural role in ring canal formation could explain
how loss of Bam function could cause complete cytokinesis.
However, bam germ cells can make ring canals (Fig. 6C) and
anti-Bam antibodies place Bam in the cytoplasm and the
fusome but not in the ring canal. Likewise, it is not likely that
loss of fusome-associated Bam could account for complete
cytokinesis of bam mutant cells because the fusome is not
required for arrest of the contractile ring (Lin et al., 1994). Furthermore, hts; bam double mutants demonstrated that Bam
does not require fusome association to exert its effects on cystoblast mitosis. Thus we postulate that BamC is responsible for
promoting incomplete cytokinesis in the cystoblast.
The differences between cystoblast and stem cell divisions
imply that cystoblast differentiation includes coordinated
changes for regulating cell division. Arrest of the contractile
ring, growth of the fusome and changes in fusome components
are examples of features that distinguish cystoblasts from stem
cells. BamC may be required, by itself or together with other
cytoplasmic partners, to signal a switch from stem cell to cystoblast mitoses; consequences of such a switch could include
modifications of the contractile ring, promotion of fusome
growth and delivery of membranous material to the fusome.
The very sparse density of membranous network within bam
fusomes suggests that this aspect of fusome assembly may be
dependent on Bam function. As bam mutations are epistatic to
hts, we propose that Bam’s association with the expanding
fusome in the cystoblast follows its initial action in the
cytoplasm.
This model leaves open the question of how germ cells are
signaled to withdraw from the mitotic cell cycle after four
rounds of division. If BamC is required to carry out the first
cystocyte-like division, it may be required to maintain these
mitoses. Since the expression and implied turnover of BamC
correlates with the mitotically active stages of cystocytes, an
intriguing possibility is that degradation of BamC after the
fourth cystocyte division blocks further mitosis. Perhaps
Fig. 8. This figure summarizes how Bam
might be required during the germ cell lineage
to regulate the switch from stem cell mitoses
to cystoblast mitoses. Our model also proposes
that Bam is required to continue cystocyte
divisions and that degradation of the BamC
protein in completed cysts could stop further
divisions.
BamC-dependent functions, such as the delivery of membranous material to the fusome, are necessary for fusome growth
and other aspects of cystocyte divisions. Fig. 8 presents
features of this model in the context of the germ cell lineage.
The marked disappearence of BamC from postmitotic cystocytes suggests that the protein’s cytoplasmic half-life might
be regulated. We have noted previously that Bam has a PEST
domain near the protein’s C-terminal end (McKearin and
Spradling, 1990). Similar domains have been shown in several
instances to be essential for rapid protein turnover (Rogers et
al., 1986; Salama et al., 1994; Belvin et al., 1995); we are
currently testing if this motif destabilizes BamC in cystocytes.
Bam is probably not a primary germline sex
determination factor
Several genes that mutate to produce tumorous egg chambers
have been shown to affect germ cell sex determination (Pauli
and Mahowald, 1990; Steinmann-Zwicky, 1992; Pauli et al.,
1993). To what extent the tumorous egg chamber phenotype
predicts a gene involved in regulating germ cell sex determination remains an open question. As a representative of the
tumorous egg chamber phenotype bam does not seem well
positioned to serve as a germline sex determination factor. The
bam gene does not express sexually dimorphic forms of bam
mRNA (McKearin and Spradling, 1990), bam mutants express
the sex appropriate form of several sexually dimorphic markers
such as Sxl (Bopp et al., 1993), otu (Sass et al., 1995) and orb
(McKearin and Christerson, 1994) and bam mutants are not
rescued by constitutive expression of Sxl (McKearin and Christerson, 1994). The data presented in this paper identify Bam as
a component of a germ-cell-specific organelle found in both
males and females. Furthermore, our data suggest a role more
consistent with regulating a choice between types of cell
divisions rather than a direct involvement in decisions of sex
determination. Perhaps the tumorous egg chamber phenotype
is more a manifestation of blockage of germ cell differentiation
than of germline sex determination specifically. Thus germline
sex determination would be one of the steps that germ cells
need to complete to progress beyond the earliest stages of
differentiation.
The authors thank Dr Alasdair MacDowall for expert assistance with
the preparation and viewing of ovaries in the electron microscope. The
authors also extend appreciation to Lynn Cooley for generous assistance in preparing anti-Bam antibodies and Haifan Lin and Allan
Spradling for generously sharing data and reagents prior to publica-
Role of Drosophila Bag of marbles protein 2947
tion. D. Branton graciously provided polyclonal anti-Spc antisera. The
authors thank H. Krämer for his patient assistance on the Biorad
MRC1000 for the double exposure in Fig. 4C. Mary Kuhn is responsible for the drawings in Fig. 6 and her artistic efforts, as well as very
capable technical assistance, are greatly appreciated. S. Wasserman, F.
Katz, L. Avery, L. Cooley, A. Spradling, members of McKearin laboratory and anonymous reviewers provided valuable comments on the
manuscript. D. M. also thanks A. Carpenter, A. Spradling and H. Lin
for useful discussions about the Drosophila fusome and L. Cooley for
conversations about numerous oogenesis-related topics. The work
described in this manuscript was supported by NIH Training Grant
T32-GM08014 to B. O. and NIH award RO1-GM45820 to D. M.
REFERENCES
Belvin, M. P., Jin Y. and Anderson K. V. (1995). Cactus protein degradation
mediates Drosophila dorsal-ventral signaling. Genes Dev. 9, 783-793.
Bopp, D., Horabin, J. I., Lersch, R. A., Cline, T. W. and Schedl, P. (1993).
Expression of the Sex-lethal gene is controlled at multiple levels during
Drosophila oogenesis. Development 118, 797-812.
Carpenter, A. T. C. (1981). EM autoradiographic evidence that DNA
synthesis occurs at recombination nodules during meiosis in Drosophila
melanogaster females. Chromosoma 83, 59-80.
Christerson, L. and McKearin, D. M. (1994). orb is required for
anteroposterior and dorsoventral patterning during Drosophila oogenesis.
Genes Dev. 8, 614-628.
Cooley, L., Kelley R. and Spradling A. C. (1988). Insertional mutagenesis of
the Drosophila genome with single P elements. Science 239: 1121-1128.
Gönczy, P. (1995). Towards a molecular genetic analysis of spermatogenesis in
Drosophila. Ph. D. thesis. The Rockefeller University.
Gondos, B. (1987a). Comparative studies of normal and neoplastic ovarian
germ cells: 1. Ultrastructure of oogonia and intercellular bridges in the fetal
ovary. Int. J. Gynec. Path. 6, 114-123.
Gondos, B. (1987b). Comparative studies of normal and neoplastic ovarian
germ cells: 2. Ultrastructure and pathogenesis of dysgerminoma. Int. J.
Gynec. Path. 6, 124-131.
Hall, P. A. and Watt, F. M. (1989). Stem cells: the generation and
maintainance of cellular diversity. Development 106, 619-633.
Hardy, R. W., Tokuyasu, K. T., Lindsley, D. L and Garavito, M. (1979).
The germinal proliferation center in the testis of Drosophila melanogaster. J.
Ultrastruct. Res. 69, 180-190.
Hartwell, L. H. and Weinert, T. A. (1989). Checkpoints: controls that ensure
the order of cell cycle events. Science 246, 629-634.
Horvitz, H. R. and Herskowitz, I. (1992). Mechanisms of asymmetric cell
division: two Bs or not two Bs, that is the question. Cell 68, 237-255.
King, R. C. (1970). In Ovarian Development in Drosophila melanogaster.
New York: Academic Press.
Lee, J. K., Coyne, R. S., Dubreuil, R. R., Goldstein, L. S. B. and Branton, D.
(1993). Cell shape and interaction defects in a-Spectrin mutants of
Drosophila melanogaster. J. Cell Biol. 123, 1797-1809.
Lin, H., Yue, L. and Spradling, A. C. (1994). The Drosophila fusome, a
germline specific organelle, contains membrane skeletal proteins and
functions in cyst formation. Development 120, 947-956.
Lin, H. and Spradling, A. C. (1995). Fusome asymmetry and oocyte
determination in Drosophila. Devel. Genet. 16: 6-12.
Lindsley, D. L. and Zimm, G. G. (1992). In The genome of Drosophila
melanogaster. San Diego: Academic Press.
Lynch, R. G. (1995). Differentiation and cancer: the conditional autonomy
phenotype. Proc. Natl. Acad. Sci. USA 92, 647-648.
Mahowald, A. P. and Kambysellis, M. P. (1980). Oogenesis. In Genetics and
Biology of Drosophila 2d. (eds. M. Ashburner and T. R. F. Wright), pp. 141223. New York: Academic Press.
Maziarski, S. (1913). Sur la persistance des residus fusariaux pendant les
nombreusesgenerations cellulaires au cours de l’ovogenese de Vespa
vulgaris. Arch Zellf., Leipzig 10, 507-532.
McKearin, D. M. and Spradling, A. C. (1990). bag-of-marbles; A Drosophila
gene required to initiate male and female gametogenesis. Genes Dev. 4,
2242-2251.
McKearin, D. M. and Christerson, L. (1994). Molecular genetics of early
germ cell differentiation during Drosophila oogenesis. In CIBA Foundation
Symposium; Germline Development. (eds. J. Marsh and J. Goode), pp. 210222. New York: John Wiley and Sons Ltd.
Pauli, D., A. P. Mahowald. (1990). Germ line sex determination in
Drosophila. Trends in Genetics 6, 259-264
Pauli, D., Oliver, B. and Mahowald, A. P. (1993). The role of the ovarian
tumor locus in Drosophila-melanogaster germ-line sex determination.
Development 119, 123-134.
Robinson, D. N., Cant, K. and Cooley, L. (1994). Morphogenesis of
Drosophila ovarian ring canals. Development 120, 2015-2025.
Rogers, S., Wells R. and Rechsteiner M. (1986). Amino acid sequences
common to rapidly degraded proteins: the PEST hypothesis. Science 234,
364-368.
Salama, S. R., Hendricks, K. B. and Thorner, J. (1994) G1 cyclin
degradation: the PEST motif of yeast Cln2 is necessary, but not sufficient, for
rapid protein turnover. Mol. Cell Biol. 14, 7953-7966.
Sass, G. L., Comer, A. R. and Searles, L. L. (1995). The Ovarian Tumor
Protein isoforms of Drosophila melanogaster exhibit differences in function,
expression and localization. Dev. Biol. 167, 201-212.
Spradling, A. C. (1993a). Developmental genetics of oogenesis. In Drosophila
Development (eds. M. Bate and A. Martinez-Arias), pp. 1-70. Cold Spring
Harbor, NY: Cold Spring Harbor Press.
Spradling, A. C. (1993b). Germline cysts: Communes that work. Cell 72, 649652.
Steinmann-Zwicky, M. (1992). How do germ cells choose their sex?
Drosophila as a paradigm. Bioessays 14, 513-518.
Storto, P. D. and King, R. C. (1989). The role of polyfusomes in generating
branched chains of cystocytes during Drosophila oogenesis. Dev. Genet. 10,
70-86.
Telfer, W. H. (1975). Development and physiology of the oocyte-nurse cell
syncytium. In Advances in Insect Physiology, Vol. 11. (eds. J. E. Treherne,
M. J. Berridge and V. B. Wigglesworth), pp. 223-320. New York: Academic
Press.
Yue, L. and Spradling, A. C. (1992). hu-li tai shao, a gene required for ring
canal formation during Drosophila oogenesis, encodes a homolog of
adducin. Genes Dev. 6, 2443-2454.
(Accepted 13 June 1995)