The neurogenic locus brainiac cooperates with the Drosophila EGF receptor to establish the ovarian follicle and to determine its dorsal-ventral polarity

Oogenesis involves direct interaction between germline cells and follicle cells for the successful development of fertile eggs (see Raven, 1961; McLaren and Wylie, 1983). In insects, these interactions take place within elongated tubes known as ovarioles. From the proximal end of the Drosophila melanogaster ovariole, where oogenesis begins with the formation of the egg chamber in the germarium, through the vitellarium where eggs mature, to the distal end of the ovariole, where eggs pass into the oviduct on their way to being laid, there are 12-13 increasingly mature egg chambers (King, 1970).

Egg chamber formation begins at the proximal tip of the germarium, where an oogonial stem cell divides to produce another oogonial stem cell and a cystoblast. The cystoblast undergoes four divisions with incomplete cytokinesis to give rise to a cluster of sixteen cystocytes interconnected by specializations called ring canals (Brown and King, 1964). Follicle cells (FCs) migrate in from the periphery of the germarium and completely surround each newly formed 16-cell cyst. The FCs along with the cyst form an egg chamber, or follicle (Koch and King, 1966). Newly formed follicles are displaced to the distal end of the germarium (by new cystocyte clusters formed more proximally), where the future oocyte becomes located at the posterior tip of the follicle. The remaining 15 daughter cells of the cystoblast differentiate as nurse cells (NCs), which provide most of the RNA and protein synthetic functions required for oogenesis (Koch and King, 1966). As the egg chamber leaves the germarium and enters the vitellarium, the FCs divide (Chandley, 1966) and change shape, forming a continuous, one-cell thick cuboidal epithelium that completely surrounds the egg chamber.

The newly formed egg chamber remains connected at its proximal end to the germarium and at its distal end to next older follicle by a one-cell thick stalk of 5-7 stalk cells. The follicle undergoes a phase of development in which both NCs and oocyte increase in volume. The follicular sheath increases from approximately 80 to 1100 cells during this growth phase (King and Vanoucek, 1960), maintaining a continuous follicular epithelium around the expanding egg chamber. Subsequently, at the inception of vitellogenesis, most of the FCs migrate posteriorly around the enlarging oocyte, leaving a thin layer of squamous FCs surrounding the NCs (King and Vanoucek, 1960). The cells overlying the rapidly growing oocyte become columnar. A clear dorsal-ventral asymmetry emerges in both the follicular epithelium and oocyte by this stage, as the oocyte nucleus comes to lie at the future anterior-dorsal edge of the oocyte, and the FCs on the dorsal side of the egg chamber become taller than those on the ventral side. After completion of these growth and polarization phases, most of the contents of the nurse cells (except the polyploid nuclei) flow into the oocyte, the vitelline membrane (or first layer of the eggshell) is completed, and the FCs switch their synthetic functions to the production of the outer eggshell or chorion (see King, 1970 and Mahowald and Kambysellis, 1980, for reviews of Drosophila oogenesis).

Interactions between germline cells and FCs have been shown to be involved in establishing the polarity of the ovarian follicle. Thus the dicephalic gene, which functions in locating the oocyte to the posterior end of the follicle (Lohs-Schardin, 1982), is required in both germline and somatic cells of the ovary (Frey and Gutzeit, 1986). The establishment of the dorsal-ventral polarity of the growing egg chamber requires germline, e.g. K10 and gurken (grk), and FC functions, e.g. torpedo (top). In the absence of K10, the oocyte and eggshell become dorsalized, whereas in the absence of either grk or top, both the eggshell and the embryo become ventralized (Wieschaus et al., 1978; Schüpbach, 1987). top mutations are alleles of faint little ball (flb), which has been shown to encode the Drosophila homolog of the EGF receptor (abbreviated DER, Schejter and Shilo, 1989; Price et al., 1989; Clifford and Schüpbach, 1989), a molecule known to play diverse roles in intercellular communication in vertebrates (for reviews see Hunter and Cooper, 1985; Yarden and Ullrich, 1988).

An influential role for intercellular communication is also apparent during insect neurogenesis. Cell ablation studies in the grasshopper (Doe and Goodman, 1985) have suggested that direct interactions between ectodermal cells occur in the early embryo, whereby neuroblasts (NBs) inhibit adjacent ectodermal cells from assuming a neuroblast fate. Inhibition of the NB fate is required to allow development of epidermoblasts, the precursors of the epidermis. In Drosophila melanogaster, the inhibitory signal appears to be mediated by a set of loci termed neurogenic (Hartenstein and Campos-Ortega, 1984; Lehmann et al., 1983). In the absence of neurogenic gene function, most or all of the cells within the neurogenic ectoderm adopt the NB fate. Two neurogenic loci, Notch(N) and Delta(Dl), encode large transmembrane proteins with EGF repeats (Wharton et al., 1985; Kidd et al., 1986; Vaessin et al., 1987). Two other neurogenic loci, pecanex and big brain, also encode large transmembrane proteins (La Bonne et al., 1989; Rao et al., 1990). One transcript within the E(spl) gene, another neurogenic locus, codes for a G-like protein subunit (Hartley et al., 1988), possibly involved in intracellular signal transfer. These molecular findings are consistent with a role for these loci in the production of a neuroblast inhibitory signal (reviewed by Campos-Ortega and Knust, 1990).

Neurogenic gene function has also been shown to be required in oogenesis. N and Dl appear to be required somatically for interaction between FCs leading to the development of specialized groups of FCs at the anterior and posterior poles of the egg chamber (Ruohola et al., 1991). The proper development of these FCs appears to be required for establishing the anterior-posterior polarity of the oocyte (Ruohola et al., 1991). In contrast to N and Dl, we show that brainiac (brn), a new neurogenic locus, is required for interaction between the germline and FCs. These interactions are necessary for the determination of the dorsal-ventral polarity of the follicle. In addition, our studies of brn demonstrate that interactions between germline and FCs begin very early in oogenesis to establish/maintain the ovarian follicle. brn acts in the germline in cooperation with somatically functioning DER to establish individual NC-oocyte complexes, maintain the follicular epithelium, and determine the dorsal-ventral polarity of the ovarian follicle. These are separable germline functions of the brn locus that appear to be needed for successive FC activities during oogenesis: migration, maintenance and dorsal-ventral cell fate determination. We propose that brn is necessary for the integrity of a germline signaling pathway differentially regulating DER-dependent FC activities throughout oogenesis.

The wild-type strain was Oregon R (OrR). brn^fs.107, previously fs(1)A107, and Fs(1)ovo^D1 were obtained in screens for female steriles (Gans et al., 1975). brn^l.6P6, previously l(1)6P6, was obtained in a screen for larval/pupal lethal mutations (Perrimon et al., 1989). A y w brn^l.6P6 chromosome, obtained as a recombinant with OrR, was used in most experiments. fs(1)K10 is described by Wieschaus et al. (1978). top¹ and top^CJ are described by Schüpbach (1987) and Clifford and Schüpbach (1989). grk^WG41, grk^HK36, and grk^HG21 are described by Schüpbach (1987). All top and grk chromosomes are marked with cn bw. Df(1)rb¹, Df(1)rb³³ and Df(1)rb⁴⁶ are described by Banga et al. (1986). Df(1)rb¹ probably removes brn completely since it also removes two loci on either side of brn, male-diplolethal (mdl) and mei-9 (Banga et al., 1986; Oliver et al., 1988). Df(1)GA102 and Dp(1;1)w^+64b13 are described by Craymer and Roy (1980). rugose (rg) and echinus (ec) were obtained from the Mid America Drosophila stock center, and are described by Lindsley and Grell (1968).

brn^fs.107 was mapped between polytene chromosome bands 3F7,8 and 4A3,6 because it is complemented by Df(1)rb⁴⁶ = Df(1)4A3,64C6-7 and Df(1)GA102 = Df(1)3D5; 3F78, whereas Df(1)rb³³ =Df(1)3F3-4; 4C15-16 fails to complement the neurogenic and fused dorsal appendage phenotypes associated with this allele. We are not able to deficiency map brn^l.6P6 because the brn locus lies immediately adjacent to the mdl locus. We therefore meiotically mapped brn^l.6P6 relative to ec (5.5) and rg (11.0). We obtained 11 out of 1972 y w ec⁺brn^l.6P6+ progeny from y w brn^l.6P6/ec females. We obtained 93/1761 y⁺w⁺brn^l.6P6+rg⁺ progeny from y w brn^l.6P6/rg females. All 104 of these recombinants were back crossed to the y w brn^l.6P6 chromosome, in order to determine that the fused dorsal appendage, neurogenic and fused chamber phenotypes mapped to the same locus as brn^l.6P6. All females were fully fertile and did not lay brown eggs or eggs with FDAs, indicating that all four phenotypes map to the same locus. Based on echinus recombinants, brn^l.6P6 maps to meiotic position 6.1; based on rugose recombinants, brn^l.6P6 maps to position 5.7. We assign brn^l.6P6 to position 5.9 on the meiotic map, which correlates with the chromosome band location of brn^fs.107. In addition, we obtained Dp(1;2)w^+64b13 in which the diplolethal locus was mutated by X-irradiation (Helen Salz, unpublished). This duplication complements the brn^fs.107 fused dorsal appendage and neurogenic phenotypes as well as the brn^l.6P6 lethality. With these mapping data, combined with the fact that brn^fs.107/brn^l.6P6 females lay eggs with fused dorsal appendages which develop neurogenic phenotypes, we conclude that brn^fs.107 and brn^l.6P6 are alleles of the same locus.

For analysis of maternal effect phenotypes, homozygous females were taken from balanced stocks. Heterozygous females were obtained from appropriate crosses. Double mutant females were taken from brn^fs.107/FM3; top/CyO stocks. Germline clones were induced by X-irradiation (1200 rads) of first instar larvae obtained from the following crosses: (a) brn^l.6P6/FM7 × Fs(1)ovo^D1/Y, (b) brn^fs.107/FM3;top^CJ/CyO × Fs(1)ovo^D1/Y;top^CJ and (c) brn^l.6P6/FM7;top^CJ/CyO × Fs(1)ovo^D1/Y;top^CJ. Informative females were identified by scoring appropriate markers, and females carrying germline clones (GLCs) were identified as described below. These females are referred to as (a) brn^l.6P6GLC, (b) brn^fs.107GLC;top^CJ and (c) brn^l.6P6GLC;top^CJ, respectively, in the text.

brn^fs.107;top/+ as well as brn^fs.107/+;top flies were analyzed as controls for brn^fs.107;top double mutant experiments. Females of both genotypes are highly fecund and never lay eggs having stronger ventralization than fusion of dorsal appendages, although the penetrance and expressivity of the FDA phenotype is increased in both cases (data not shown). Some ovarioles from these control females, nevertheless, contain chambers with more than one NC-oocyte complex. 6%, 4% and 6% of ovarioles from brn^fs.107;top^CJ/+, brn^fs.107/+;top¹ and brn^fs.107/+;top^CJ females, respectively, have at least one fused chamber. Discontinuities of the follicular epithelium are never observed in control egg chambers. In contrast, most egg chambers in every ovariole from double mutant females are abnormal. For germline clone experiments, brn^fs.107GLC;top^CJ/CyO females were analyzed as controls, and were found to have aberrant egg chambers at frequencies comparable to control females described above.

For the zygotic rescue experiment, brn^fs.107 females were mated to wild-type males, and embryos were collected for 5-6 hours. A set number of embryos was then transferred to another plate and, after aging for 30 hours, eggs that did not hatch were transferred to another agar plate and allowed to incubate at 25°C for 24 hours. The number of embryos that hatched was taken as the number of starting eggs minus the number of eggs transferred. Brown eggs (having melanized cuticle) were scored as not zygotically rescued, while white eggs, presumably unfertilized, were ignored.

Eggs were collected and aged at 25°C on yeasted agar-molasses plates.

Hoyer’s mounts of embryonic cuticles were prepared as described by Wieschaus and Nüsslein-Volhard (1986) and viewed using phase contrast microscopy. Cuticles were scored under phase contrast optics as intermediate or weak according to the criteria of Lehmann et al. (1983). If an embryo had any ventral cuticle remaining, it was scored as weak.

Drosophila embryonic nervous systems were stained with horse radish peroxidase (HRP) (Jan and Jan, 1982) according to the procedure of Fredieu and Mahowald (1989). Polyclonal antisera to HRP (Organon Teknika) were purified by HRP affinity chromatography, and detected with a FITC-conjugated secondary antibody. Embryos were mounted in 1:1 glycerol:ethanol and viewed by epifluorescence.

For histological preparations, embryos were removed from the vitelline membrane as for antibody staining and then placed overnight at 4°C in the fixative of Kalt and Tandler (1971). After dehydrating through an ethanol series, embryos were infiltrated with JB4 plastic overnight at 4°C, and then embedded in plastic. Serial 4-micron sections were stained with 1% methylene blue in a 1:1 mixture of 1% boric acid:70% ethanol, and viewed in bright field optics.

Eggs were collected at 18°C or 25°C on yeasted agar-molasses plates. All eggs in Tables 2 and 3 were scored using a dissecting scope. Follicle cell imprints on double mutant eggs were analyzed by phase contrast microscopy of Hoyer mounted eggs (prepared as described for embryos).

brn^fs.107 flies were grown at either 18°C or 25°C and females collected over a two day period. Eggs were collected for 3-4 days at 12-24 hour intervals. Flies were shifted to either 18°C or 25°C and eggs were collected at intervals as designated in Fig. 3. All eggs laid following the shift were scored. Six bottles (40 females/bottle) were used per experiment. Results were consistent between bottles.

All experimental ovaries were dissected at 1-1.5 days post eclosion, to avoid analyzing secondary phenotypes. For example, egg chambers are commonly observed fusing within the vitellarium of 3-day old double mutant females, but we do not observe these events in 1to 1.5-day old females. However, chambers having supernumerary NC-oocyte complexes as well as discontinuities in the follicular epithelium are observed in newly eclosed brn^fs.107; top double mutant females. Since 1to 1.5-day old females do not lay eggs, ovaries in germline clone experiments that had ovarioles with egg chambers developed past stage 3-4 (the point at which Fs(1)ovo^D1 blocks oogenesis [Busson et al., 1983]) were scored as containing germline clones. In most females, only one of the two ovaries contains a clone, and the contralateral ovary can be used as a control. Egg chambers in ovarioles from such contralateral controls do not have supernumerary oocyte NC complexes or follicular epithelium discontinuities as do the experimental ovaries.

For Hoechst staining, ovaries were dissected in Drosophila Ringers solution and then stained in 1% Hoechst # 33528 dye (obtained from Sigma, prepared in Drosophila Ringers) for 10 minutes, rinsed in Ringers, then fixed for 5 minutes in acetone. Ovaries were then mounted on a slide in 60% glycerol, the ovarioles teased apart using tungsten needles, and examined by epifluorescence. Oocyte nuclei were distinguished from NC nuclei by size, and then confirmed by differential interference contrast (DIC) microscopy. NC and oocyte numbers were scored by focusing through the egg chamber. In all cases, the numbers were confirmed by rescoring. For chambers containing 16 to approximately 40 cystocytes, counts were typically repeated once, and the repeated count typically fell within +/− 1 of the original count. Sometimes counts were repeated more than this, depending on the difficulty of the sample. For chambers containing >40 cystocytes, counts were repeated 3–4 times, with counts falling within +/− 2 of the noted count.

Ring canals were stained with phalloidin according to a modification of Warn et al. (1985). After dissecting ovaries in Ringer’s, they were fixed in 4% formaldehyde in phosphate-buffered saline (PBS) for 30 minutes, then extracted with acetone for 5 minutes at −20°C before incubating in rhodamine-phalloidin (0.16 μg/ml PBS) for an hour. Ovaries were then mounted on a slide in 60% glycerol and the ovarioles teased apart. Ring canals were revealed using epifluorescence as above. For confocal microscopy, a Zeiss confocal laser scan microscope was used (courtesy of Dr. T. Karr) to produce a 3-D reconstruction of phalloidin-stained egg chambers.

Hoechst-stained preparations were analyzed for holes in the follicular epithelium. Since acetone is a harsh fixative and can cause tissue distortion, we confirmed this phenotype by viewing both living tissue as well as glutaraldehyde-fixed tissues using DIC microscopy. Following fixation in 2% glutaraldehyde (in PBS) for 10 minutes, ovaries were removed, washed and the ovarioles were teased apart with tungsten needles. Ovarioles were either mounted in Aquamount (Lerner) for temporary preparations, or dehydrated through an ethanol series, cleared in xylene and mounted in Permount for permanent preparations. The percentage of chambers having holes was determined using Hoechst-stained preparations only.

Ovaries were fixed and then stained with Fas III monoclonal antibody as described by Ruohola et al. (1991), except that antiFas III was used at 1:5 dilution with a FITCconjugated antibody. Ovarioles were teased apart and mounted in 60% glycerol, and fluorescently labelled follicle cells were examined using epifluorescence as above.

Embryonic pole cells were transplanted by the method of M. Zalokar and P. Santamaria as described by Lehmann and NüssleinVolhard (1987), with minor differences. Cloroxed hosts and donors were stage-selected by posterior clearing to mid-late cellular blastoderm during agar alignment and mounted on double stick tape (3M #666). Hosts were desiccated over silica gel for a fixed time, empirically giving about 5% leaking eggs. Injections were performed under Halocarbon Product’s Series 700 oil and the embryos allowed to hatch on plain agar under Series 27 oil in a humid chamber. Processing, injection and hatching were carried out at 17°C, then hatchings were transferred en masse to standard vials. Isolated females were mated on eclosion to brn^fs.107 males, observed for egg laying, and dissected after 5 days.

The brainiac (brn) locus maps between polytene bands 3F7,8 and 4A3,6 and to meiotic map position 1-5.9 (see Materials and methods). The maternal effect phenotypes described in this paper are based on analysis of two partial loss-of-function alleles of the brn locus, brn^fs.107 and brn^l.6P6. brn^fs.107 females are viable, whereas brn^l.6P6 animals die as pharate adults. Analysis of the maternal effect phenotypes associated with brn^l.6P6 was accomplished either by induction of germline clones (GLCs) in females heterozygous for brn^l.6P6 (referred to as brn^l.6P6 GLC females, see Materials and methods), or by analysis of eggs derived from females doubly heterozygous for brn^fs.107/brn^l.6P6.

brainiac females produce embryos with epidermal hypoplasia (Fig. 1A, B) and neural hyperplasia (Fig. 1C, D) (Perrimon et al., 1986, 1989; our results). As for other neurogenic mutants (Lehmann et al., 1983), this phenotype appears to arise during hours 4-7 of embryogenesis due to an excessive commitment of ectodermal cells to the NB lineage (Fig.1E, F). The strongest brn phenotype resembles the “intermediate” level (as classified by Lehmann et al., 1983) of neurogenic transformation.

brn embryos are zygotically rescuable. Whereas 2% of embryos hatch when brn^fs.107 females are mated to brn^fs.107 males (N=328), 51% hatch (N=607) when brn^fs.107 mothers are mated to wild-type males (embryos from brn^fs.107/brn^l.6P6, brn^fs.107/Df(1)rb¹, and brn^l.6P6 GLC females are also zygotically rescuable, to a large but undetermined extent). As for the well-characterized zygotic neurogenic loci, brainiac gene activity is thus required zygotically to prevent all cells of the neurogenic region from differentiating as NBs. The brn^l.6P6 allele shows greater reduction in the neurogenic component of gene activity than brn^fs.107, the former behaving as a complete loss of gene activity (Table 1).

The chorion of the Drosophila egg is secreted by the FCs during the final stages of oogenesis and contains several polarized structures that reflect the asymmetries arising during the earlier phases of development (King, 1970; Mahowald and Kambysellis, 1980; Wieschaus, 1979). For example, the respiratory appendages identify the anteriordorsal surface (Fig. 2A) of the egg. The chorions produced by brn females have dorsal appendages located at the normal anterior position, but fused along the dorsal midline (Fig. 2). The fused dorsal appendage (FDA) phenotype is similar to the phenotype associated with weak alleles of two loci known to be required for the determination of the dorsal-ventral axis of the embryo and eggshell, torpedo (top) and gurken (grk) (Fig. 2; Schüpbach, 1987). Stronger alleles of these loci are associated with a complete loss of the dorsal appendages. These phenotypes have been interpreted as resulting from a “ventralization” of FC fates, such that FCs located on the dorsal side of the egg have adopted the fates normally adopted by FCs located more ventrally. Zygotes developing within ventralized grk and top eggshells show a corresponding increase in embryo ventralization with greater loss of maternal gene function, developing greater amounts of ventrally derived mesoderm with loss of dorsal ectoderm (Schüpbach, 1987).

In contrast to the effect on early neurogenesis, brn^fs.107appears to represent a greater loss of brn gene activity for dorsal appendage patterning than brn^l.6P6 (Tables 1 and 2), suggesting that the brn chorion and embryo phenotypes are independent. This idea is substantiated by brn embryos showing no evidence of a shift in dorsal-ventral polarity (Fig.1) and being zygotically rescuable (see previous section). In contrast, the embryonic ventralization phenotype of top and grk correlates with the chorion phenotype and is strictly dependent on the genotype of the mother. Further, the ventralized FDA phenotype associated with top¹ and top^CJ homozygous females is increased in severity in top¹/Df(2)top^3F18 or top^CJ/Df(2)top^3F18 hemizygous females, which lay severely ventralized eggs completely lacking dorsal appendages (Clifford and Schüpbach, 1989). In contrast, neither homozygous brn^fs.107 nor hemizygous brn^fs.107/Df(1)rb¹ females (18°C) lay more severely ventralized eggs.

Does the brn FDA phenotype represent a defect in dorsalventral patterning or another developmental process? This question was addressed by constructing females doubly mutant for brn^fs.107 and top. top is required for the dorsalventral patterning of the ovarian follicle; if brn is required likewise, then top function should depend on brn function, and females double mutant for brn and weak top alleles should have more strongly ventralized eggs. Most of the eggs laid by double mutant brn^fs.107; top females completely lack dorsal appendages (Fig. 2D; Table 3), have a more rounded dorsal side than wild-type eggs, and have FC imprints around their circumference characteristic of the ventral surface of the wild-type egg (data not shown). As demonstrated in genetic mosaic experiments (Table 6), brn is required in the germline for determination of FC fates.

The FDA phenotype of brn^fs.107 eggs shows a cold temperature sensitivity (Table 2) which is first detected at stages 6–7 of oogenesis (Fig. 3). This is the point at which the egg chamber first begins overt polarization. This phenocritical period is consistent with a role for brn in determining the dorsal-ventral axis of the ovarian follicle.

brn^l.6P6 GLC females lay only 3-7 eggs over their lifetime (Perrimon et al.,1989, and this report). Similarly, brn^fs.107; top double mutant females lay only a few or no eggs (Table 3), in contrast to females mutant for either brn^fs.107 or top, which lay hundreds of eggs. Although double mutant females show a reduction in viability (Tables 4 and 5), reduced egg production is not caused by this reduced viability. The poor fecundity of these females must be due to a specific ovarian interaction of brn and top, since the same ovarian phenotype is obtained from brn^fs.107; top^CJ double mutant females and from brn^fs.107GLC; top^CJ females, in which only the germ cells, and not the whole animal, are mutant for brn^fs.107.

The cause of the reduced fecundity of doubly mutant brn^fs.107GLC; top^CJ females appears to be the same as that found in brn^l.6P6GLC females. Most follicles (see below) have more than one oocyte and more than 15 NCs. These follicles are formed within the germarium (Fig. 4E, F and below) and can proceed to late stages of oogenesis (Fig. 4G, H). Occasionally such chambers will proceed with chorion synthesis and form an “egg” in the shape of a ball, lacking overt polarity, presumably because competition for polarity among several oocytes is balanced. In most cases, however, such chambers degenerate around stages 9-11. Development of normal egg chambers depends on brn function in the germline, since in these germline clone experiments only the germ cells are homozygous mutant for brn (there is a normal copy of brn in the FCs).

Fig. 4. brainiac and the Drosophila EGF receptor are required to establish the ovarian follicle. Tissues are stained with Hoechst and the nurse cell nuclei (large, polyploid nuclei), oocyte nuclei (small nuclei not visible in most of the egg chambers shown; arrowheads in (B)) and follicle cell nuclei are revealed with flourescence microscopy. (A) Ovarioles from a brn^fs.107 female. As for top¹, top^CJ or top¹/top^CJ females, egg chambers (arrows) within the vitellarium are completely normal, consistent with the high fecundity of females of all four genotypes. Bar, 50 μm. (B) All ovarioles from brn^l.6P6 germ-line clone(GLC) females have aberrant egg chambers containing supernumerary oocytes and nurse cells. The arrow points to the germarium. (C) All ovarioles from brn^fs.107GLC;top^CJ females resemble those from brn^l.6P6GLC females (B). In addition, egg chambers are formed with large discontinuities in the follicular epithelium (arrowheads). (D) Egg chambers containing over six presumptive nurse cell-oocyte complexes can occupy essentially the whole ovariole of brn^l.6P6GLC;top^CJ females. The egg chamber forming here appears to be continuous with, or just pinching from the germarium. Note the graded increase in nurse cell nuclear size from the proximal to distal end of this egg chamber, indicating the earlier birthdate of oocyte-nurse cell complexes found more distally. In contrast to chambers found within brn^fs.107GLC;top^CJ ovarioles(C), egg chambers as large as those shown here do not have discontinuities within the follicular epithelium. Note that follicle cells are not found within egg chambers from brn^l.6P6GLC or double mutant females. (E and F) brn^l.6P6GLC germarium enlarged 5 times from B. An egg chamber is pinching from the germarium (arrows) containing over 40 cystocytes. Similar formations are found for double mutant combinations (also see Fig. 7). Bar, 5 μm. (G and H) Stage 10 egg chamber from a brn^fs.107GLC;top^CJ female. This egg chamber has five oocytes (G, arrowheads) that are able to take up yolk and approximately 75 nurse cells. In addition, these oocytes are able to attract migratory follicle cells (H, arrowheads). Note the starred oocyte (G) “pulling” follicle cells toward it (H). The cap of follicle cells surrounding the oocyte at the bottom of the figure is out of focus. Experiments were performed at 25°C. Ovarioles were dissected from 1to 1G-day old females.

The synergistic interaction between brn and top is not allele-specific. Flies doubly mutant for brn^fs.107 and three top allele combinations (cf. Table 3) display similar phenotypes. Further, brn^l.6P6 GLC; top^CJ ovaries have a strikingly enhanced phenotype compared to brn^l.6P6 GLC ovaries, showing in some instances chambers with more than 100 NCs (Fig. 4D). We compared the phenotypes of these different allelic combinations by scoring the frequency with which chambers are found with different numbers of NCs and oocytes (Fig. 5). These results show that these different allele combinations fall into a series of decreasing severity: brn^l.6P6 GLC; top^CJ > brn^fs.107GLC; top^CJ > brn^l.6P6 GLC.

We consider two manners in which follicles containing supernumerary oocytes and NCs might originate in the germarium. Egg chambers with supernumerary cystocytes could result from aberrant patterns of cystocyte division as postulated for ovarian tumor mutations (King et al., 1957). On the other hand, these chambers might result from the failure of FCs to gather around and correctly separate individual 16-cell cysts. Both of these mechanisms depend on brn function in the germline, since the results presented in the previous section were obtained in germline clone experiments in which only the germ cells, but not the remaining cells of the animal, were homozygous mutant for brn.

Follicles with supernumerary NCs and oocytes might be produced by additional cystoblast divisions, producing follicles with multiples of 16 cells. A number of observations make this unlikely. First, the cells of a wild-type 16-cell cluster remain interconnected via ring canals remaining after each incomplete cystocyte division (Brown and King, 1964). The oocyte and one of the NCs have four ring canals. Extra rounds of cell division would be expected to give rise to oocytes and NCs having 5, 6 and 7 ring canals. We have stained egg chambers with phalloidin to observe the actin ring of the canals (Warn et al., 1985) and we do not observe oocytes or NCs with more than 4 canals. Further, all cystocytes have at least one ring canal. This contrasts with the ovarian tumor phenotype, which is thought to result from the over-proliferation of cystocytes (King et al., 1957). In these mutants, most cystocytes undergo complete division (Johnson and King, 1972; King and Riley, 1982), and do not have ring canals interconnecting adjacent cystocytes. Further, the supernumerary oocytes found in the tumorous egg chambers of benign gonial cell neoplasm females remain clustered together at the normal posterior position yet fail to take up yolk (Gutzeit and Strauss, 1989), in contrast to the varied distributions of the vitellogenic oocytes we observe (e.g. Fig. 4G, H). Second, in follicles recently derived from the germarium in which there is no size differential between NCs, the ring canal pattern clearly establishes the separate origin of chambers (Fig. 6). In this instance, the size of the ring canals for the posterior oocyte establishes its greater age (cf. Koch and King, 1969). Third, within follicles having extra sets of NCs and oocytes, the NC clusters are clearly of different sizes, suggesting that they are at different levels of ploidy, and are thus derived from cystoblasts with different birthdates. In the most extreme case found in a brn^l.6P6 GLC; top^CJ female, a clear gradient in NC size can be seen from apparent germarial cystocytes at the proximal end of the egg chamber to apparent stage 10-type polyploid NCs at the distal end of the egg chamber (Fig. 4D). A similar distribution of NC sizes can be seen distributed between more egg chambers in ovarioles from brn^fs.107 GLC;top^CJ and brn^l.6P6 GLC females shown in Fig. 4. Since no FCs are observed between clusters of NCs, these chambers are most easily explained by the inclusion of multiple 16-cell cysts, each derived from a single cystoblast, within one follicular epithelium. These results suggest that mutant egg chambers result from the failure of the FCs to separate adjacent 16-cell cysts.

The number of supernumerary NCs within a given egg chamber is not always a multiple of 15. We frequently observe, however, that the total number of NCs in two adjacent chambers is a multiple of 15. For example, in brn^l.6P6 GLC females, one chamber had 17 NCs and an oocyte and the next chamber had 13 NCs and an oocyte. In another instance, one chamber had 35 NCs and 2 oocytes, while an adjacent chamber had 25 NC and 2 oocytes. In both of these cases, the combined NC to oocyte ratio between pairs of chambers is 15:1. A cluster of NCs in such a chamber containing more NCs than 15 X (no. of oocytes) appears to be approximately the size of the NCs in the neighboring egg chamber with fewer NCs than 15 X (no. of oocytes). This complementarity between chambers suggests that some cystocytes somehow become separated from their sister cells and are included within adjacent follicles.

Alternatively, complementary chambers might originate by incomplete cystocyte division and subsequent enclosure of multiple incomplete cysts within one egg chamber. This type of mechanism would predict that the average NC:oocyte ratio among several chambers would be less than the normal 15. The total number of NCs and oocytes for 160 different chambers was calculated. The average NC:oocyte ratio was 14.6 for 106 chambers from brn^l.6P6 GLC females, 14.7 for 24 chambers from brn^fs.107 GLC; top^CJ females, and 15.1 for 30 chambers from brn^l.6P6 GLC; top^CJ females. These findings are consistent with normal patterns of cystocyte division.

We stained germaria with Fasciclin III (FasIII) antibody (Patel et al., 1987). In wild-type germaria, FasIII-positive FCs (Fig. 7A) can be observed to line the walls of and to span across the germarium (Brower et al., 1981; Ruohola et al., 1991), separating individual cysts. In brn^l.6P6 GLC; top^CJ females, Fas III-positive FCs are found to be lining the walls of the germarium in regions 2 and 3, but in 69% of germaria (N=30) FCs fail to migrate within the germarium (Fig. 7C). For the remaining 31%, a single band of FCs can be observed to span the germarium, enclosing multiple sets of cystocytes (Fig. 7B). FCs are always found in germaria from brn^fs.107GLC; top^CJ females. brn^l.6P6 GLC; top^CJ females, which have the most severe phenotype in terms of the number of NC-oocyte complexes per egg chamber (Fig. 5), also have the most severe disruption of FC migration, suggesting that the primary reason follicles are formed with supernumerary NC-oocyte complexes is a failure of FCs to respond to germline signals necessary for enveloping each 16-cell cyst.

In addition to supernumerary NC-oocyte complexes, egg chambers from brn^fs.107 GLC; top^CJ females frequently (66%, N=195) have discontinuous follicular epithelia (Fig. 4C). These discontinuities are judged to range from as large as one third of the surface of the follicle to as small as 78 cells. The phenotype is not allele specific for top, because brn^fs.107; top¹ (45%, N=190) as well as brn^fs.107; top¹/top^CJ (% not determined) females develop egg chambers with the same phenotype. The phenotype is allele-specific for brn, however, as neither brn^l.6P6 GLC nor brn^l.6P6 GLC; top^CJ females develop similar phenotypes. Thus, the normal packaging of NC-oocyte complexes and development of a complete follicular epithelium appear to be separate functions of the brn locus.

Since the phenotype is apparent in brn^fs.107GLC; top^CJ females, in which somatic cells have a wild-type copy of brn, the formation of a continuous follicular epithelium appears to be a specific germline-dependent function of the brn locus. Discontinuities clearly do not arise as a result of overstretching of the follicular epithelium in unusually large chambers (containing multiple NC-oocyte complexes), because egg chambers of brn^l.6P6 GLC; top^CJ females as large as those shown in Fig. 4D do not develop holes. Even normal sized chambers from brn^fs.107GLC; top^CJ females containing single NC-oocyte complexes may possess holes.As many as 4 consecutive chambers, each having a hole, can be observed within the vitellarium of one day old females. Such configurations may contribute to the fusion of adjacent chambers observed in older females.

Significantly, we also find that gurken females that lay completely ventralized eggs (grk^WG41, grk^WG41/grk^HK36, or grk^HG21, data not shown) also have egg chambers with holes in the follicular epithelium similar to those in Fig. 4C. Thus, three loci required for the determination of the dorsal fate of the FCs are also required for the formation of the follicular epithelium. In contrast to these loci, neither fs(1)K10, which is required for determination of ventral FC fates, nor doubly mutant fs(1)K10, brn^fs.107 females develop incom plete follicular epithelia.

Phenotypes developed by brn^l.6P6 GLC as well as brn^fs.107 GLC; top^CJ females indicate that brn gene function is required in the germline for FC migration, formation/maintenance of the follicular epithelium, and determination of dorsal-ventral FC fates. These results, however, leave open the question as to whether the DER is required in the germline and/or the soma for early oogenesis. Considering that the DER has been shown to be required in the FCs for dorsal-ventral patterning (Schüpbach, 1987), and that growth factors and their receptors are required for cell migration and cell division in vertebrates (Grotendorst et al., 1981; Blay and Brown, 1985; Postlethwaite et al., 1987; Yarden and Ullrich, 1988; White, 1990), we considered it an attractive possibility that the DER is required in the FCs for their migration into the germarium and for their proliferation over the egg chamber during the early growth phase. We constructed mosaic females by pole cell transplantation (Illmensee, 1973) in which the germ cells are mutant for brn^fs.107, while the surrounding host, including the somatically derived FCs, is mutant for top^CJ. These females have normal brn function in the FCs and normal DER function in the germline. As shown in Fig. 8 and Table 6, these females develop egg chambers having multiple NC-oocyte complexes and discontinuous follicular epithelia, demonstrating the somatic (presumably FC) requirement for top function during the early phases of oogenesis.

All “maternal” neurogenic loci, like brn, show some degree of zygotic rescuability (Shannon, 1972; Perrimon et al., 1984, 1989). All “zygotic” neurogenic loci, with the exception of big brain, show a maternal effect (Jimenez and Campos-Ortega, 1982; Campos-Ortega, 1985). The distinction between “maternal” and “zygotic” neurogenic loci appears to be a reflection of the degree to which a given neurogenic gene product is contributed by the mother, not a reflection of an intrinsic developmental mechanism as, for example, distinguishes the maternal dorsal-ventral loci from the zygotic dorsal-ventral loci (Simpson, 1983). Indeed, the zygotic rescuability of brn function strongly indicates that the brn requirement for dorsal-ventral FC fate determination is independent of the brn requirement in the early embryo. The brn requirement for dorsal-ventral FC fate determination occurs at the inception of vitellogenesis, many hours before zygotic genome activation. Further, brn embryos show no indications of altered gastrulation as seen in mutants for top, grk, fs(1)K10, cappuccino and spire, which are required maternally for determination of FC dorsal-ventral fates (Wieschaus et al., 1978; Schüpbach, 1987; Manseau and Schüpbach, 1989).

Thus, brn appears to be unique among loci required for dorsal-ventral patterning during oogenesis. The DER is required in the FCs and grk, fs(1)K10, cappuccino and spire are required in the germline for the determination of both follicle and embryonic dorsal-ventral cell fates (Wieschaus et al., 1978; Schüpbach, 1987; Manseau and Schüpbach, 1989). nudel, pipe and windbeutel are required in the soma (presumably the FCs) for the determination of embryonic, but not FC fates (Stein et al., 1991), demonstrating that a flow of information from soma to germline exists which can be separated from FC fate determination. brn, on the other hand, is required in the germline for the determination of follicle, but not embryonic, dorsal-ventral cell fates, demonstrating that a flow of information from germline to FCs exists which can be separated from determination of embryonic cell fates. Consistent with this notion, the overall shape of brn oocytes is not altered as are top and grk oocytes (see Fig. 2).

nudel, pipe and windbeutel appear to be components of a FC signaling pathway required for production of a ligand that specifies embryonic dorsal-ventral cell fates (Stein et al., 1991). The Toll gene product, which is evenly distributed in the membrane surrounding the embryo, appears to encode the receptor for the ligand (Hashimoto et al., 1988; Stein et al., 1991). The DER is expressed continuously during oogenesis on the surface of the FCs (and not in the germline), including the period during which the egg chamber has obtained dorsal-ventral asymmetry (R. Schweitzer, N. Zak, and B. Shilo, personal communication). Further, the genetic mosaic experiments presented in Table 6 demonstrate that the dorsal-ventral pattern activity, which requires DER activity in the follicle cells, is also dependent on germline functions, at least one of which is brn. Thus, the DER appears to be a receptor for (a) germline signal(s) (cf. Manseau and Schüpbach, 1989).

We have not used null alleles of either brn or torpedo in these studies, and it is therefore not possible to deduce whether brn and the DER act in the same or a parallel genetic pathway, i.e. to decide whether brn might be a component of a germline signaling pathway required for DER dorsal-ventral patterning activity. Although our results do not rule out that brn and the DER may be in parallel pathways, the mosaic studies clearly show that the DER depends on germline functions, presumably a ligand(s), to dorsalventral pattern the follicle. Acting in a parallel pathway to the DER, brn may be required in a more general fashion for reception of DER signals. For example, brn may be required for proper contact/adhesion between germline and follicle cells necessary to ensure proper germline signaling for follicle cell dependent DER function.

Both the chorion and embryonic phenotypes associated with brn^fs.107 and brn^l.6P6 are not as strong as the strongest phenotypes associated with prototypical ventralizing or neurogenic loci. Since both brn^fs.107 and brn^l.6P6 behave like deficiencies, or a complete loss of gene function, for dorsalventral patterning and neurogenesis, respectively, a hypothetical null brn allele would not be associated with stronger phenotypes. brn function may be redundant, either at the gene family level, or at the level of gene pathways (Rykowski et al., 1981; Ferguson and Horvitz, 1989). Functional redundancy is a common feature associated with growth factors and their receptors (cf. Nicola and Metcalf, 1991).

The brn^fs.107 temperature-sensitive period for dorsal-ventral patterning occurs during stages 6-7 of oogenesis, well after the follicle has been established. This temperature-sensitive period is consistent with the period in oogenesis in which the oocyte nucleus has been found to be required for dorsal-ventral patterning, stages 6-9 (Montel et al., 1991). This indicates that the brn gene product is needed at the time dorsal-ventral patterning takes place during oogenesis, and thus separates this requirement for brn gene function from the requirement to establish the egg chamber. In contrast to brn^l.6P6, brn^fs.107 retains the early brn oogenesis functions. However, when somatic DER function is reduced simultaneously, germline brn^fs.107 function is revealed as inadequate for packaging 16-cell cysts and forming a complete follicular epithelium. This synergistic phenotype demonstrates that intercellular communication between germline cells and somatic follicle cells starts very early in oogenesis.

Our data indicate that egg chambers containing multiple NC-oocyte complexes result from a failure of FCs to migrate into the germarium and enclose individual 16-cell cysts. Pole cell transplantation experiments demonstrate that germline brn function is required with somatic DER function for the enclosure of individual cysts. Although the level of brn and top activity retained by either brn^fs.107 or top^CJ mutations is sufficient for formation of individual follicles, germline brn^fs.107 activity is not sufficient for normal follicle cell activities when DER function is simultaneously reduced in the FCs. This synergistic phenotype resembles that found for brn^l.6P6GLC females (in the presence of normal DER follicle cell function); thus, the level of brn activity retained by brn^l.6P6 is not sufficient for follicle formation, presumably due to disruption of a signal from the germline to the follicle cells. This notion is strengthened by the finding that reduction of top function dramatically enhances the brn^l.6P6GLC phenotype. The expression of the DER on the surface of FCs and not in the germline (R. Schweitzer, N. Zak, and B. Shilo, personal communication), and the involvement of tyrosine kinase receptors and their growth factors in regulating cell migration in vertebrates (Grotendorst et al., 1981; Blay and Brown, 1985; Postlethwaite et al., 1987; White, 1990) are consistent with the idea that brn and the DER are necessary for the integrity of an intercellular signaling pathway required for FC migration. Both the discovery of complementary sets of NCs in adjacent follicles, and the finding of an average NC-oocyte ratio among a large number of mutant egg chambers of approximately 15, suggest that abnormal numbers of NCs are not due to faulty cystocyte divisions. Normal packaging of 16-cell cysts by the FCs may be needed for the stabilization of the cyst. Koch and King (1969) found that prior to FC enclosure cystocytes are more loosely attached than the FC-enveloped germarial cystocytes, which form a very compact mass. The normal FC enclosure may be required to prevent stretching and breakage of the ring canal interconnections between cystocytes. On the other hand, the FCs may play a more active role in breaking cysts apart. Mahowald and Strassheim (1970) found that FCs always migrate around, and not between cystocytes. Further, Koch and King (1966) found that invading mesodermal cells apply as much of their surface as possible to interconnected cystocytes. If FCs are unable to recognize putative specialized properties of the cystocytes, or if such specialized properties are abnormally distributed over the cystocyte surfaces, then FCs might migrate between, instead of around, the cystocyte clusters.

Discontinuity of the FC epithelium is a novel phenotype not previously described in Drosophila mutants affecting oogenesis. The discontinuities associated with brn, top and grk mutations do not appear to be due to FC death, because pycnotic FC nuclei are only observed late when the whole chamber is necrosing. Discontinuities are already evident on some chambers as they leave the germarium (Fig. 8), so that it is reasonable to assume that the cause of the discontinuities is due to a defect evident already in the germarium. One interpretation is that in some instances the invading follicular epithelium failed to complete the envelopment of a follicle, perhaps because of the absence of an essential adhesive property in some cyst cells. Since the holes persist and may also arise during the early growth phase of the egg chamber, this adhesive property appears to be required even during the subsequent growth stages of oogenesis.

An alternate explanation for the origin of the follicular discontinuities is a failure of sufficient cell division, beginning already in the germarium. The predominant activity of the FCs as the egg chamber leaves the germarium (Chandley, 1966) and during the early growth phases of the egg chamber (King and Vanoucek, 1960) is mitosis: the FCs increase from 10-20 in number to approximately 1100 cells. Without the normal number of follicle cells, a gap might arise in the follicular epithelium. Since epithelia can ordinarily stretch over large surfaces, as happens in stage 10 when the follicle cells forms a squamous epithelium over the nurse chamber, this explanation is less likely. However, the presence of DER on the surface of all FCs during the early egg chamber growth phase (R. Schweitzer, N. Zak, and B. Shilo, personal communication) and the known role of tyrosine kinase receptors in regulating cell division in vertebrates (Yarden and Ullrich, 1988) are consistent with this explanation. These hypotheses are not mutually exclusive. Proper FC adherence to the germline may be needed for successful engagement of mitotic signals, or successful completion of mitosis might be needed to trigger follicle cell adhesion functions, or the phenotype might arise due to a simultaneous disruption of both processes.

The association of discontinuous follicular epithelia with egg chambers from brn^fs.107; top double mutant females, but not egg chambers from either brn^l.6P6GLC or brn^l.6P6GLC; top^CJ females suggests that these are separable functions of the brn locus. This indicates that brn and the DER are not only needed for signals within the germarium responsible for follicle cell migration, but are also required for the adherence and/or division of the follicle cells to and/or around the 16-cell cyst. brn and the DER would presumably be needed for these adherence and/or division functions throughout oogenesis. That the DER is expressed on the apical surface of all follicle cells throughout oogenesis (R. Schweitzer, N. Zak, and B. Shilo, personal communication) is consistent with this proposition.

Mutations in three other neurogenic loci show significant phenotypes in both oogenesis and neurogenesis. Mutation of another late zygotic lethal locus with a neurogenic maternal effect, zw4, is associated with the development of egg chambers having “tumorous-like overgrowth of NCs” (Perrimon et al., 1989). N and Dl encode large transmembrane proteins with EGF repeats (Wharton et al., 1985; Kidd et al., 1986; Vaessin et al., 1987), although neither has been shown to interact with the DER. Both, however, are required in subpopulations of FCs (Ruohola et al., 1991). Reduction of either function causes a hypertrophy of polar precursor and polar follicle cells and a corresponding hypotrophy of stalk and flanking follicle cells, respectively, comparable to the hypertrophy of NBs at the expense of epidermoblasts during neurogenesis (Ruohola et al., 1991). These findings suggest that N and Dl are required for lateral inhibition between follicle cells, and thus appear to have parallel roles in neurogenesis and oogenesis. This similarity has led to the suggestion that N and Dl as well as other neurogenic loci may act as a “cassette” to produce the intercellular signalling required for lateral inhibition in more than one developmental context (Ruohola et al., 1991).

In contrast to this view, brn has distinct functions in oogenesis from N and Dl. brn is required in the germline for interactions between the germline and surrounding FCs, necessary for development and maintenance of the follicular epithelium around individual cysts as well as for determination of dorsal-ventral follicle cell fates. None of these functions is shared by N and Dl. Possibly only a subset of the genes involved in the lateral inhibition by neuroblasts play a similar role in the development of ovarian polar cells. The failure of a deletion of the neurogenic locus pecanex to disrupt oogenesis (LaBonne et al., 1989) is consistent with this notion. We propose that N and Dl may utilize other signaling pathways for effecting inhibition between FCs. Comparison and contrast of the manner in which neurogenic functions are used during oogenesis and neurogenesis should aid in understanding the role these functions play for both processes.

brn appears to be required for normal tyrosine kinase function during oogenesis, a biochemical function not associated with any of the six cloned neurogenic loci. Our preliminary experiments indicate that brn is in the same genetic pathway as Notch and other neurogenic loci. The neurogenic phenotypes associated with mutations in all of these loci (with the exception of the neurogenic gene big brain, which is not in the same genetic pathway) are suppressed by mutations in the proneural gene daughterless (Brand and Campos-Ortega, 1988). Likewise, daughterless mutations suppress the brn neurogenic phenotype (S. G., unpublished data).

top function is almost certainly required only in the FCs throughout oogenesis. Our genetic mosaic experiments show that top function is required in the FCs during the early phases of oogenesis, consistent with the observed expression of DER only on the surface of the FCs and not in the germline. brn is clearly a germline function, although the genetic mosaic studies do not rule out the possibility that brn function is also required in the FCs. These genetic mosaic experiments, however, suggest that the DER is dependent on (a) germline signal(s) throughout oogenesis (Fig. 9).

We propose that, in the germarium, both the migration and division and/or adhesion functions of brn are “ON” in the germline and are required for a signal to the (pre)follicle cells received by the DER. Prefollicle cell division is observed in this region of the germarium and appears to be required to replenish migratory follicle cells (Johnson and King, 1972). At the posterior of the germarium, brn migration function is turned “OFF” by an unknown mechanism (?, Fig.9). Interestingly, this appears to be the time at which the neurogenic loci Notch and Delta are required for the development of the stalk cells that connect egg chambers throughout oogenesis (Ruohola et al., 1991). brn division and/or adhesion function remains “ON” for follicle cell division and/or adhesion around the growing egg chamber at the posterior of the germarium and throughout the early growth phase.

At the end of the early growth phase, when the egg chamber begins to polarize, the brn gene product is needed for dorsal-ventral patterning, as determined by the brn^fs.107 temperature-sensitive period. fs(1)K10, required for determination of ventral FC fates, is expressed at approximately the same time (Haenlin et al., 1987). In addition to the finding that neither fs(1)K10 nor fs(1)K10, brn^fs.107 females develop discontinuous follicular epithelia, it appears that fs(1)K10 is specifically required for dorsal-ventral patterning. We suggest that fs(1)K10 may modify continuously required brn, grk and DER function at the end of the early growth phase, altering their function from a cell division and/or adhesive mode to a cell determination mode (Fig. 9). This idea would be consistent with the model for dorsalventral patterning gene function proposed by Manseau and Schüpbach (1989), in which fs(1)K10 is required upstream of top and grk for the initial polarization of the oocyte, as well as with our suggestion that brn is required upstream of the DER as part of a signaling pathway. If fs(1)K10 expression is turned on according to an intrinsic developmental clock, such as NC ploidy or number of FC divisions (Fig. 9), this would explain how dorsal-ventral polarity is always established at the proper time, in an environment of continuously acting functions and variable rates of egg chamber development. At least part of the switch may involve increasing the level of activity of these continuously acting functions (thickening line, Fig. 9), since we observe that all females laying completely ventralized eggs (brn; top double mutants, grk) also develop discontinuous follicular epithelia. Females, laying less severely ventralized eggs having fused dorsal appendages (brn, top), never develop these discontinuities, suggesting that more of these functions are required for cell fate determination than for the proposed follicle cell division. The apparent complexity of brn and DER activity in regulating follicle cell behavior during Drosophila oogenesis is reminiscent of complexities described for regulation of cell behavior by growth factors and their receptors in other developmental systems (e.g. see Noble et al., 1988; Raff et al., 1988; Flanagan and Leder, 1990; Flanagan et al., 1991).

Neurogenic loci are all thought to be required for the specific neuroblast to epidermoblast inhibitory signal (Lehmann et al., 1983). Most of these loci appear to be required for the transmission of the signal rather than the intracellular response (Technau and Campos-Ortega, 1987). Similarly, at least half of the known loci required to determine ventral embryonic cell fates appear to act upstream of the putative Toll receptor (Stein et al., 1991). Our studies of brn also indicate a considerable complexity in the signals provided to the FCs, even though these signals are apparently received via “the same” DER function. Much of the current focus in understanding the function of tyrosine kinase receptors centers on finding phosphorylated molecules utilized in transferring surface signals to the nucleus (Yarden and Ullrich, 1988). Continued study of brn should illuminate how signals can be regulated to give the varied set of intracellular responses needed for regulation of varied cellular processes.

The early phases of this work were completed in the Department of Genetics at Case Western Reserve University. We thank Brian Oliver and Peter Harte for reading and commenting on the manuscript and for sharing their knowledge and interest throughout this project. Thanks to Charlie Rudin for reading and commenting on the manuscript. We acknowledge John Fredieu for help with embryo antibody staining. We thank Helen Salz for the dpl Dp(1;2)w+64B13 stock prior to publication. We thank Trudi Schüpbach for sending top and grk stocks, and thank Jim Price and Robert Clifford for helpful discussions concerning brn and brn-top interactions. Thanks to Hannele Ruohola for sending and discussing the Fas III antibody. We are indebted to Tim Karr for generating confocal images and sharing his knowledge of confocal microscopy. We thank Jim Shapiro for taking pictures of Drosophila eggs. We thank Gerry Grofman for help with making figures. We thank Macintosh guru Dr. Mark D. Garfinkel for help with the computer. This work has been supported by a grant from the American Cancer Society (A.P.M.) and NIH fellowships (#T32 HD7104 and #T32 GM07197) to S.G.