A novel mutant phenotype implicates dicephalic in cyst formation in the Drosophila ovary

Abstract

The establishment of polarity in Drosophila requires the correct specification of the oocyte in early stages of oogenesis, its positioning at the posterior of the egg chamber, and signalling events between the oocyte and the adjacent posterior follicle cells. As a consequence, the anterior-posterior and the dorsal-ventral axes are fixed. The posterior localisation of the oocyte depends on cadherin-mediated adhesion between the oocyte and the follicle cells. Here we show that dicephalic mutants affect the posterior positioning of the oocyte without interfering with oocyte specification in the germarium. Unlike other mutants that also affect the posterior placement of the oocyte, dicephalic mutants affect neither gurken expression nor karyosome formation during meiosis. By analysing in detail the mutant phenotypes of dicephalic, we find that cyst formation in mutant germaria is defective and that it shares some similarities with cysts that lack DE-cadherin in the germline cells. We propose a model in which dicephalic is involved in the proper adhesion between the oocyte and the somatic follicle cells. Developmental Dynamics 235:908–917, 2006. © 2005 Wiley-Liss, Inc.

INTRODUCTION

The formation of body axes is a phenomenon that occurs in multicellular organisms and is concomitant with the development of asymmetry in the egg and early embryo. Drosophila, like most organisms, has two clearly defined, main body axes: the anterior-posterior or AP axis and the dorsal-ventral or DV axis. It is known that these axes are established during oogenesis before the egg is fertilized by the asymmetric localisation of bicoid, oskar, and gurken mRNAs (St Johnston and Nüsslein-Volhard, 1992; Riechmann and Ephrussi, 2001). The localisation of these mRNAs depends on earlier steps in oogenesis, and involves the determination of the oocyte, its positioning at the posterior of the egg chamber, and signalling events between the germline and the somatically derived follicle cells (reviewed in Huynh and St Johnston, 2004). A central signaling molecule in the establishment of the AP and DV axes is Gurken, as this protein induces the follicle cells adjacent to the posteriorly placed oocyte to adopt a posterior fate. These posterior follicle cells then send a signal back to the oocyte to trigger a re-organization of the microtubule cytoskeleton that is ultimately responsible for the anterior-dorsal localization of the oocyte nucleus. Since gurken mRNA localizes above the oocyte nucleus at this stage, and since Gurken is also responsible for the induction of dorsal follicle cell fates, both major axes of the Drosophila egg chamber depend on Gurken signaling (González-Reyes et al., 1995; Roth et al., 1995).

Mutations in a number of genes that lead to the mispositioning of the oocyte have demonstrated its importance for the polarization of the oocyte. For instance, mutations in spindle (spn) genes—spn-A, spn-B, spn-D, spn-E, and mus301 (also known as spn-C)—vasa, or in armadillo germline clones and shotgun germline or follicle cell clones, give rise to egg chambers in which the oocyte, instead of lying posterior to the nurse cells, can be located anywhere in the egg chamber (Peifer et al., 1993; González-Reyes and St Johnston, 1994, 1998; Gillespie and Berg, 1995; González-Reyes et al., 1997; Oda et al., 1997; Godt and Tepass, 1998; Styhler et al., 1998; Tinker et al., 1998; Tomancak et al., 1998; White et al., 1998). However, and in spite of this phenotypic similarity, the mechanisms by which these mutants affect oocyte placement within the egg chamber are quite distinct. The spn genes and vasa are members of the spindle-group genes, as they all affect many aspects of patterning in the egg chamber and their oocytes fail to form a normal karyosome, the hollow sphere of highly condensed chromatin characteristic of wild-type oocytes (González-Reyes and St Johnston, 1994; Gillespie and Berg, 1995; González-Reyes et al., 1997; Styhler et al., 1998; Tinker et al., 1998; Tomancak et al., 1998). The defective karyosome phenotype has been shown to correlate with a failure to progress through meiosis correctly, thus implicating this group of genes in both meiosis and pattern formation (Boyd et al., 1981; Ghabrial et al., 1998; Ghabrial and Schüpbach, 1999; Abdu et al., 2003; Jang et al., 2003; Staeva-Vieira et al., 2003). armadillo and shotgun, on the other hand, encode components of the cadherin-adhesion complex, as Armadillo is the Drosophila β-catenin and shotgun encodes the Drosophila orthologue of E-cadherin (Peifer et al., 1992; Oda et al., 1994; Tepass et al., 1996). Eliminating the function of this adhesion complex at the time when the oocyte comes to lie at the posterior of the cyst disrupts the normal localization of the oocyte and prevents the initial establishment of the AP axis (Godt and Tepass, 1998; González-Reyes and St Johnston, 1998).

The dicephalic gene (dic) was named because of its mutant embryonic phenotype where the abdomen is replaced by a head and thorax of inverted polarity, giving rise to an embryo with a mirror type duplication of the anterior of the body (Lohs-Schardin and Sander, 1976; Lohs-Schardin, 1982). The only mutant allele identified to date, dic¹, arouse spontaneously. It is characterized as a semi-dominant maternal effect mutation of low penetrance that maps to a single locus on the third chromosome: position 46 ± 1.0, between cp and in ri (75D4 and 77C). Further investigation of dic¹ revealed that the embryonic phenotype emerges because mutant females produce during oogenesis “bipolar” egg chambers with the oocyte in the middle and nurse cells on both sides (Lohs-Schardin, 1982). As a result, the oocyte is not correctly polarized, with no posterior being specified, and instead two anterior regions develop. The generation of chimaeric egg chambers showed that dic is necessary in both the follicle cells and the germline cells (Frey and Gutzeit, 1986).

We decided to investigate the dic mutant phenotypes further to see if dic shared any similarities, other than a role in oocyte positioning, with the spn-group genes or whether it could be classified as an “adhesion” gene. We find that the oocyte is selected correctly and forms a normal karyosome in mutant ovaries. We also describe that the oocyte is mispositioned in a proportion of mutant egg chambers, and that this aberrant location of the oocyte can be traced back to cyst formation in germarial stages. In addition, we show that dic¹ cysts in the germarium adopt a rounded shape, reminiscent of those observed when the cadherin-mediated adhesion between the germline cells and the follicle cells is blocked, suggesting a role for Dic in adhesion between these two cell types. Finally, analysis of double mutants between dic and spn-B, spn-D, or mus301 showed a strong genetic interaction between dic and these spn genes.

RESULTS

An Overview of Early Oogenesis

Drosophila has a pair of ovaries each composed of 16–20 ovarioles. Ovarioles usually consist of 6 to 7 sequentially more mature egg chambers, each separated by short chains of interfollicular stalk cells and subdivided into 14 stages (S1 to S14). Each ovariole is contained within a muscular sheath that helps maintain the linear order of the egg chambers and helps to move them posteriorly with time. An ovariole is divided into an anterior germarium and a posterior vitellarium, with two to three stem cells located at the anterior of each ovariole. These stem cells divide asymmetrically to produce a daughter stem cell and a cystoblast. The cystoblast undergoes four rounds of mitosis, with incomplete cytokinesis at each mitotic division, resulting in the formation of a cyst of 16 cells or cystocytes, interconnected by cytoplasmic bridges called ring canals (Mahowald and Kambysellis, 1980). Two of the cells of the cyst possess 4 ring canals and behave like “pro-oocytes,” since they both form complete synaptonemal complexes and accumulate oocyte specific markers. Finally, one of the 4-ring canal cells becomes the oocyte while the other 4-ring canal cell and the remaining 14 cystocytes develop as nurse cells performing nutritive roles for the oocyte. The cyst of 16 cells becomes surrounded by a layer of somatic follicle cells to form an egg chamber that exits from the germarium and progresses through the vitellarium, maturing as it does so (Spradling, 1993b).

It is within the germarium that the critical events of choosing the oocyte and its positioning posterior to the nurse cells occur. The germarium is divided into several region: region 1 contains the germline stem cells, the mitotically dividing cystoblasts, and the 2- to 8-cell cysts. Region 2a contains 16-cell cysts, with two or more occupying the width of the germarium. In region 2b, the cysts flatten to form a 1-cell thick lens-shaped disc, which extends across the whole width of the germarium. By this stage, the oocyte has been chosen and lies in the centre of the cyst; ∼2 somatic stem cells are found in this region of the germarium and they divide and start to migrate to surround the cyst to separate it from the preceding cyst. The most posterior region of the germarium, region 3 (also known as S1), contains only a single cyst, completely surrounded by somatic follicle cells. The cyst is now spherical in shape with the oocyte positioned at the posterior. This positioning is the first visible sign of AP polarity. The cyst, or egg chamber as it is now called, is ready to enter the vitellarium, where it matures. The development from region 1 to region 3 takes approximately 3 days (King, 1970; Spradling, 1993a; Margolis and Spradling, 1995; González-Reyes and St Johnston, 1998).

The Oocyte Is Selected Correctly in dic Mutants

The penetrance of the oocyte mispositioning phenotype was analyzed by raising dic¹ females at 25°C, dissecting their ovaries, and staining them with Rhodamine-phalloidin to visualize filamentous actin. Eleven percent of egg chambers contain a misplaced oocyte (n = 368), with the majority of these egg chambers being of the bipolar type. Egg chambers with the oocyte in a lateral position or at the anterior pole are seen too, but less frequently than bipolar egg chambers (Fig. 1A,B). The percentage of misplaced oocytes in dic mutants is much higher than that seen in spn mutants where it is about 2%.

Figure 1. Mutant phenotypes displayed by dic¹ egg chambers. A, B: Egg chambers stained with Rhodamine-phalloidin to visualise F-actin. The wild-type egg chamber in A shows the oocyte at the posterior. B: Mutant egg chamber with the oocyte in a lateral position. C, D: Anti-Gurken staining of S9 egg chambers. Both the mutant and the wild-type exhibit normal levels and localisation of Gurken protein. E, F: Egg chambers stained with Rhodamine-phalloidin in red and OliGreen to label DNA in green. Oo, oocyte; arrowheads indicate the karyosome. In this and all following Figures, the orientation of S2 and older egg chambers is with the anterior to the left and dorsal up.

It has been shown that in spn mutants meiosis is disrupted, leading to activation of a meiotic checkpoint that then prevents the translation of Gurken mRNA and, as a result, a disruption of the polarization of both the AP and DV axes (González-Reyes et al., 1997; Ghabrial et al., 1998; Ghabrial and Schüpbach, 1999; González-Reyes, 1999; Abdu et al., 2002, 2003; Staeva-Vieira et al., 2003). Although in dic mutants the DV axis is not disrupted, as the eggs laid carry both dorsal respiratory appendages, eggs are laid that have a micropyle at both ends (Lohs-Schardin, 1982). Because the micropyle is a structure found only at the anterior of a wild type egg, these eggs most likely arise from the bipolar egg chambers in which the arrangement of the nurse cells and the oocyte is disrupted. Further confirmation that the DV axis is not affected comes from the analysis of the pattern of expression of Gurken. dic mutant S9-10 egg chambers stained with anti-Gurken antibody exhibit wild type levels of Gurken (Fig. 1C,D). In addition, and since Gurken is also expressed normally in early egg chambers, the AP axis of dic mutant egg chambers in which the oocyte is localized at the posterior is likely to be correctly established (data not shown; González-Reyes et al., 1995; Roth et al., 1995).

Failure to form a karyosome correctly in spn mutants reflects a disruption of meiosis (Ghabrial et al., 1998; Ghabrial and Schüpbach, 1999; Abdu et al., 2003; Staeva-Vieira et al., 2003). Examination of the karyosome phenotype in dic mutants revealed that 100% of egg chambers (n > 300) contain a wild type karyosome (Fig. 1E,F). This indicates that meiosis is most likely not disrupted in dic mutants.

Therefore, the only shared phenotype between spn mutants and dic mutants is a failure to correctly position the oocyte. In spn mutants, it is thought that the mispositioning of the oocyte is a consequence of the delay in choosing one pro-oocyte to become the oocyte (González-Reyes et al., 1997). To investigate if dic mutants also show misplaced oocytes because of a defect in oocyte selection, germaria from dic mutants were examined for the accumulation of oocyte specific markers, such as Orb, Bic-D, and Cup proteins (Wharton and Struhl, 1989; Suter and Steward, 1991; Christerson and McKearin, 1994; Lantz et al., 1994; Keyes and Spradling, 1997). dic mutants show no delay in choosing the oocyte, since Bic-D, Orb, and Cup proteins all accumulate in the oocyte in the majority of region 2B cysts, as in wild type (Fig. 2C,D and data not shown).

Figure 2. dic¹ germaria do not form cysts properly. Germaria of different genotypes stained with Rhodamine-phalloidin (red) and anti-Orb (green). The oocyte is the cell that accumulates more Orb protein and, when visible, is labelled with an asterisk. A: Wild type germarium showing two cysts in region 2B adopting a lens shape and spreading across the width of the germarium, one cyst in transit from region 2b to region 3 and one region 3 cyst. B:shg^IG29 germline clones showing the cysts in region 2 remaining rounded with the oocyte occupying random positions within the cyst. C, D: Two different dic¹ germaria showing an abnormal organisation of the cysts. The oocyte is randomly positioned within the cyst in region 2 but, despite this, some oocytes become correctly positioned by region 3 (D). Dotted lines delineate region 2b and 2b/3 cysts. In this and all following Figures, the orientation of germaria is such that anterior is up.

Mutations in dic Disrupt the Morphology of the Germarium

It has been shown that cadherin-mediated adhesion between the oocyte and the adjacent follicle cells in the germarium is required to correctly position the oocyte (Godt and Tepass, 1998; González-Reyes and St Johnston, 1998). Drosophila E-cadherin is encoded by shotgun (shg) and β-catenin is encoded by armadillo (arm) (Peifer and Wieschaus, 1990; Tepass et al., 1996; Uemura et al., 1996). β-catenin and α-catenin are required to link Cadherin to the actin cytoskeleton (Kemler, 1993). In a large proportion of arm germline clones and shg germline or follicle cells clones, the oocyte is mispositioned (Peifer et al., 1993; Oda et al., 1997; Godt and Tepass, 1998; González-Reyes and St Johnston, 1998; White et al., 1998). In addition, chimaeric germaria bearing shg germline clones are disrupted, as mutant region 2b cysts remain rounded instead of flattening and adopting a lens shape that spans the width of the germarium (Fig. 2A,B) (González-Reyes and St Johnston, 1998). Drosophila α-catenin maps to 80F3 but no known mutants exist. It is expected that mutants in α-catenin share a similar phenotype to that seen in arm and shg germline clones.

In dic mutant germaria, the organization of the cysts is aberrant, such that the cysts remain rounded instead of extending across, perpendicular to the AP axis of the germarium to acquire a lens shape (52% of germaria show at least one mutant cyst, n = 64; Fig. 2C,D). This phenotype is reminiscent of that seen in shg germline clones, raising the exciting possibility that dic might be required for the cadherin-mediated adhesion between the oocyte and the follicle cells.

Furthermore, since both dic and α-catenin had been mapped close to the centromere on the third chromosome, dic to 75–77 and α-catenin to 80, it is possible that dic¹ is a mutant allele of α-catenin. In this case, the localisation of α-catenin could be disrupted in dic mutants. To test this possibility, antibody stainings were done with anti-α-catenin and anti-Arm on dic mutant germaria. We find that in dic mutants DE-cadherin, α-catenin and Arm accumulate in the follicle cells and in germline cysts in which the oocyte is placed at the posterior as in wild-type (Fig. 3 and data not shown). These results suggest that cadherin-mediated adhesion is not disrupted in dic mutants. From these observations, we conclude that dic mutants disrupt oocyte positioning most likely independent of the cadherin-mediated adhesion pathway. Our results, however, sustain a role for Dic in some adhesion process. First, the germarial phenotype of dic resembles that of shg germline clones and, like shg, dic is required in both the germline and the follicle cells (Frey and Gutzeit, 1986). Second, dic¹ also affects the migration of the follicle cells as they surround the forming cysts. In dic¹ germaria, the follicle cells often accumulate in large aggregates in a manner that never occurs in wild type germaria (Fig. 2C,D). This strongly suggests that mutant follicle cells are unable to engulf the cyst and adhere to the oocyte to position it, a possibility further supported by our finding that sometimes dic egg chambers are disrupted such that they contain too few or too many nurse cells (data not shown), indicating a malfunction during cyst encapsulation in the germarium.

Figure 3. DE-cadherin-mediated adhesion is not affected in dic¹ ovaries. α-catenin (A, C, E, F) and anti-armadillo (B, D) staining of wild type and dic¹ ovaries. α-catenin and Armadillo expression in dic¹ mutants is unchanged with respect to controls. A-D: Germaria. E, F: S3-6 egg chambers.

Attempts to Map dic to a Deficiency

To advance further in its genetic and molecular characterization, we tried to map the dic gene to a deficiency. A total of nine deficiencies from the third chromosome deficiency kit, spanning 73A to 79B, were crossed to dic homozygotes (see Experimental Procedures section) and the female progeny were scored for misplaced oocytes. None of the deficiencies in trans to dic gave misplaced oocytes, suggesting that the deficiencies used do not uncover dic. This result indicates that either the meiotic mapping of dic is not accurate and does not fall within the 75D–77C region, or that dic lies in one of the gaps between the deficiencies utilized in our mapping effort. Another reason for our failure to map dic to a deficiency could be that the penetrance of the misplaced oocyte phenotype is weaker in trans to a deficiency than when homozygous, hence the frequency of misplaced oocytes might become too low to detect. In an attempt to enhance the dic oocyte misplacement phenotype, we recombined the mus301⁰⁹⁴ allele onto the dic chromosome hoping that the removal of one wild-type copy of mus301 would show a genetic interaction with dic. Hemizygous mus301⁰⁹⁴ ovaries give rise to ∼2% misplaced oocytes (Table 1). mus301⁰⁹⁴dic¹/dic¹ ovaries show no enhancement of the oocyte misplacement phenotype when compared to dic¹ controls (not shown). However, when the mus301⁰⁹⁴dic¹ double homozygotes were analyzed, a strong enhancement of the oocyte misplacement phenotype was seen.

Table 1. Frequency of Misplaced Oocytes in dic spn Double Mutantsa

Genotype	mus301^094/Df	mus301^094dic^1	spn-B^056/Df	dic^1spn-B^056	spn-D^349/Df	dic^1spn-D^349
a n, The number of egg chambers scored.
% misplaced oocytes (n)	3 (296)	>20 (447)	<2 (203)	10 (144)	<2 (335)	9 (232)

Analysis of the mus301 dic Double Mutants

mus301⁰⁹⁴dic¹ double homozygotes display the following phenotypes:

• 1
  Similar to dic homozygotes, the double mutants have very disrupted germaria, where the cysts remain rounded instead of forming a lens shape spanning the germarium (Fig. 4A,B);
• 2
  100% of egg chambers (>200 scored) have a mutant karyosome (Fig. 4C);
• 3
  mus301 mutants exhibit a delay in choosing one pro-oocyte to become the oocyte (González-Reyes and St Johnston, 1998; McCaffery et al., unpublished data). Although the frequency of this phenotype is the same in mus301⁰⁹⁴ homozygotes and mus301⁰⁹⁴dic¹ double homozygotes, the length of the delay is enhanced in the double mutants, which sometimes show two oocyte-like cells persisting up until S5–7, as seen by the presence of failed karyosomes or by Bic-D accumulation in both cells with four ring canals (39%, n = 49; Fig. 4D,E). Although the double mutants show a late oocyte selection up to S5–7, this process does not seem to be blocked, as older egg chambers often display only one of the pro-oocytes accumulating oocyte markers, indicating that the final selection of the oocyte has been made (Fig. 4F). These phenotypes are reminiscent of those seen in double mutant combinations of spn genes, such as mus301 spn-B or mus301 spn-E (González-Reyes et al., 1997).
• 4
  The frequency of misplaced oocytes can range from anywhere between 20% up to 54% (n = 447; Table 1). Why such variability occurs is unknown, but mus301⁰⁹⁴ mutants themselves show some variability in the penetrance of the misplaced oocyte phenotype. Seventy percent of the misplaced oocytes in the double mutants are bipolar egg chambers (Fig. 4G).

Figure 4. Some ovarian phenotypes of homozygous mus301⁰⁹⁴dic¹ females. A, B: Germaria stained with Rhodamine-phalloidin (red) and anti-BicD (green) showing region 2b and region 3 cysts with a rounded, aberrant shape. The oocyte in these cysts is neither positioned nor selected correctly, as demonstrated by the accumulation of BicD protein in both pro-oocytes (asterisks). C, D: Egg chambers stained with Rhodamine-phalloidin (red) and OliGreen (green) displaying the typical mus301 mutant karyosome (arrowheads). D: Frequently, both 4-ring canal cells contain a mutant karyosome. E, F: Egg chambers stained with Rhodamine-phalloidin (red) and anti-BicD (green). E: BicD occasionally accumulates in both cells with 4-ring canals, even at S6, indicating that the oocyte has still not been chosen. F: Often, when BicD is found only in the oocyte at S6–7, the other 4-ring canal cell is smaller than all the other nurse cells. G: Bipolar egg chamber stained with Rhodamine-phalloidin in which the oocyte (oo) occupies a central position.

In the course of the analysis of mus301 dic double mutants, a totally novel phenotype was observed. The frequency of ventralized eggs laid by mus301⁰⁹⁴ hemizygotes is 49% (n = 553); mus301⁰⁹⁴dic¹ females lay 62% of wild-type eggs, 29% of centralized eggs and 9% of eggs with extra dorsal appendages (n = 236; Fig. 5A,B and data not shown). These dorsalized eggs are never observed in mus301 mutants. Since mus301 is required for the proper accumulation of Gurken protein in a proportion of the egg chambers (González-Reyes et al., 1997), and in order to investigate the origin of the dorsalized eggs produced by mus301dic mutants, we analyzed Gurken expression in these double mutant ovaries. In good correlation with the percentage of wild-type eggs laid, the majority of S9–10 egg chambers presents a normal distribution of Gurken protein above the oocyte nucleus (Neuman-Silberberg and Schüpbach, 1993; data not shown). In addition, in some egg chambers two distinct regions of Gurken protein accumulation are seen (Fig. 5C). This interesting phenotype was examined further and shown to be due to the presence of two nuclei in one oocyte. Transmission images overlaid with fluorescent anti-Gurken staining clearly show each region of Gurken localisation to be associated with a nucleus (Fig. 5, C1, C2). Double staining with Rhodamine-phalloidin and anti-Gurken showed that both nuclei lie in the oocyte as no cell membrane separates them (data not shown). These egg chambers probably arise from those where two mutant karyosomes are seen in one 4-ring canal cell, the future oocyte, with the adjacent cell with four-ring canals containing no nucleus (35%, n = 49; Fig. 5D,D'). The localization of two oocyte nuclei in a single cell does not seem to be produced by a breakdown of the cytoplasmic membrane shared by both 4-ring canal cells, as F-actin staining shows no such disruption. In fact, the second nucleus seems to move to the oocyte through the ring canal connecting both cells (Fig. 5E,E').

Figure 5. A proportion of mus301⁰⁹⁴dic¹ egg chambers show bi-nucleated oocytes. A: Wild-type egg laid by a double mutant female. It possesses two anterior-dorsal appendages. B: Two dorsalized eggs from homozygous mus301⁰⁹⁴dic¹ females. C: Anti-Gurken antibody staining of a S9 egg chamber showing two distinct regions of Gurken localisation. Magnifications of two different focal planes to image the upper and lower regions of Gurken staining are shown in C1 and C2, respectively. The transmission images of the egg chamber overlaid with the fluorescent staining from the anti-Gurken antibody (in green) show that each region of Gurken expression is associated with a germinal vesicle (indicated by a black arrowhead). D, D': Rhodamine-phalloidin (red) and OliGreen (green) staining of an egg chamber where the oocyte contains two mutant karyosomes and the adjacent cell with 4-ring canals contains no nucleus. D': Magnification of the two 4-ring canal cells in D. E, E': Two confocal sections from an egg chamber stained with Rhodamine-phalloidin (red) and OliGreen (green). They show the oocyte containing a mutant karyosome (E) and the mutant karyosome from the adjacent cell passing through the adjoining ring canal (enlarged in E'). Asterisks label the 4-ring canal cells; arrowheads point towards mutant karyosomes; the open arrowhead indicates the karyosome passing through the ring canal.

Analyzing Other spn dic Double Mutants

Since dic can enhance the frequency of misplaced oocytes in mus301 mutants, we decided to see if this enhancement was general in other spn mutants too. Therefore, the following double mutants were made: dic¹spn-A⁰⁵⁷, dic¹spn-B⁰⁵⁶, dic¹spn-D³⁴⁹. dic¹spn-A⁰⁵⁷ ovaries are greatly distorted and could not be scored reliably. The frequency of misplaced oocytes for the remaining combinations was calculated and is presented in Table 1. These results show that in all cases combining dic with a spn mutation enhances the frequency of misplaced oocytes. In addition, and in good correlation with the dic mutant phenotypes, in all the dic spn double mutants analyzed there was a proportion of egg chambers that either contained too many or too few nurse cells, suggesting that the budding off of the egg chambers from the germarium is disrupted. Similarly, in the above dic spn double mutants, germaria resemble that of dic homozygotes and are severely disrupted (data not shown).

The analysis of the phenotypes shown by dic¹ ovaries has been done on homozygous flies, raising the possibility that the mutant phenotypes found are not associated directly with dic¹ and are instead a consequence of the genetic background. In this regard, our attempts to map dic to a deficiency failed, thus precluding the study of dic¹ in hemizygous conditions. However, the original ru th stdic¹e chromosome was recombined during the construction of the spn dic double mutants, allowing the study of cleaned dic¹ chromosomes. stdic¹e/ru th stdic¹e or ru th stdic¹/ ru th stdic¹e females show the same phenotypes and frequencies as homozygous ru th stdic¹e (not shown). Although it is possible that a modifier of dic¹ resides within the regions of the ru th stdic¹e chromosome that have not been exchanged, the above results suggest that the phenotypes reported in this work do correspond to dic¹.

DISCUSSION

The Role of dic in Oocyte Positioning

The posterior positioning of the oocyte within the egg chamber is a crucial step for the polarization of the oocyte and the embryo (Riechmann and Ephrussi, 2001). Several genes are known to contribute to this process. Among them, the spindle genes (spn-A, spn-B, spn-D, spn-E, and mus301) and mutations that affect the cadherin-mediated adhesion between the follicle cells and the germline are some of the best characterized (González-Reyes et al., 1997; Godt and Tepass, 1998; González-Reyes and St Johnston, 1998). There are, however, clear differences in the requirements for these genes in the process. For instance, the spn genes have been proposed to affect oocyte positioning by delaying the selection of the oocyte in the germarium. Thus, cysts mutant for mus301 or doubly mutant for any of the spn genes show the two pro-oocytes accumulating oocyte markers until much later than wild-type cysts. Although the decision as to which 4-ring canal cell will become the oocyte is finally reached, it is thought that this late oocyte selection is responsible for the misplaced oocytes observed in spn mutants (González-Reyes et al., 1997). The spn genes give name to a group of genes that show defects in the patterning of the egg chamber as well as in progression through meiosis. The spn-group comprises the 5 spn genes, vasa, okra, aubergine, and maelstrom (Gillespie and Berg, 1995; Wilson et al., 1996; González-Reyes et al., 1997; Ghabrial et al., 1998; Styhler et al., 1998; Tomancak et al., 1998; Abdu et al., 2003; Findley et al., 2003; Staeva-Vieira et al., 2003). When mutated, these genes fail to form a proper karyosome and do not localize Gurken protein correctly. Since several of these genes encode proteins involved in DNA repair, and since the patterning defects and the abnormal karyosome phenotype of spn-A, spn-B, spn-D, mus301, and okra can be suppressed by mutations in mei-W68, which block the formation of dsDNA breaks, or in mei-41, a component of the DNA damage meiotic checkpoint, the primary role of these spindle-group genes seems to be the repair of recombinational double-strand DNA breaks (Ghabrial and Schüpbach, 1999; González-Reyes, 1999).

The analysis of the early steps in oocyte positioning led to the discovery that E-cadherin-mediated adhesion is necessary for the posterior placement of this cell during cyst formation in the germarium. In this context, eliminating the function of two of the archetypical components of cadherin complexes, DE-cadherin or Armadillo, in the germarium prevents the localization of the oocyte in a high percentage of cysts. The selection of the oocyte, the accumulation of oocyte-specific markers, and progression through meiosis are not affected in these mutant conditions, demonstrating that, even if the oocyte is correctly specified, it can be misplaced if the adhesion between the germline and the somatic follicle cells is disrupted (Godt and Tepass, 1998; González-Reyes and St Johnston, 1998). In an attempt to understand the role of dic in the posterior positioning of the oocyte, and in light of the current understanding of the process, we decided to perform a detailed phenotypic characterization of a mutant condition of the gene.

When dic¹ egg chambers were compared to that of the spn genes, the only common phenotype seen was oocyte mispositioning, and the oocytes from dic¹ mutant ovaries develop normally in all other respects. An explanation for the “dicephalic” phenotype came when we analyzed the first steps in cyst formation in mutant ovaries. dic mutant germaria show a severe disruption of the morphology of germarial germline cysts and the follicle cells that surround them, as the germline clusters often adopt a rounded shape and the follicle cells form aggregates in the germarium. These phenotypes resemble chimaeric cysts in which the cadherin adhesion required for the stretching of the cysts in region 2b is impaired, as shown for shg or arm (González-Reyes and St Johnston, 1998). Altogether, these data support a model in which Dic is needed to sustain proper adhesion between germline and follicle cells and/or for the migration of the follicle cells to envelop the germline cells. This adhesion is most likely independent of the cadherin system, as Arm, DE-cadherin, and α-catenin localize normally in dic¹ ovaries. Further support for this model comes from the fact that dic is required in both germline and follicle cells for the posterior localisation of the oocyte (Frey and Gutzeit, 1986). Interestingly, two genes coding for cadherin superfamily proteins map to the dic region, fat-like and Cad74A (Hill et al., 2001; Castillejo-Lopez et al., 2004), raising the possibility that one of them could correspond to dic. Hence more than one type of cadherin may be required to position the oocyte.

mus301 dic Double Mutants Give Rise to Binucleated Oocytes

The finding that all spn dic double mutants display a strong enhancement of the oocyte mispositioning phenotype suggests that the spn genes and dic act in different parallel pathways regarding oocyte positioning. This is especially clear in the case of mus301 dic double mutants, which exhibit germaria with both defects that account for the misplacement of the oocyte: the delay in oocyte specification typical of mus301 mutants and the disrupted cyst morphology observed in dic¹. Unexpectedly, these double mutants produce egg chambers where two mutant karyosomes are found in the oocyte, while the adjacent 4-ring canal cell has no nucleus. Since these egg chambers contain only one binucleated cell, the oocyte, this phenotype is not produced by a defect in cystocyte division, as has been reported for the spaghetti-squash gene (Roth et al., 1999). Rather, this novel phenotype seems to be a consequence of the migration of the second nucleus into the oocyte through the intercellular bridge shared by both cells with 4-ring canals. What causes this phenotype is unknown, but it seems to be specific of the “losing” pro-oocyte and unrelated to the actin cytoskeleton, as we never observed mutant egg chambers in which the nurse cell nuclei are associated to the ring canals, a phenotype shared by several mutants that affect the actin cytoskeleton of germline cells (Mahajan-Miklos and Cooley, 1994). It would be interesting to investigate whether this double mutant combination interferes with any of the known steps in oocyte nucleus positioning (Koch and Spitzer, 1983; Januschke et al., 2002).

Although the molecular characterization of dic remains elusive, the results reported here have revealed a previously unknown role for dic in cyst formation in the germarium consistent with a requirement for Dic in the adhesion between the germline and the follicle cells. Considering the conservation throughout the animal kingdom of the formation of germline cysts (Pepling et al., 1999), shedding light on the mechanisms responsible for the positioning of the oocyte in the Drosophila ovary might provide new cues as to how the polarity is established in Drosophila and, perhaps, in some vertebrates (Huynh and St Johnston, 2004).

EXPERIMENTAL PROCEDURES

Fly Stocks

Germline clones were generated using the DFS technique (Chou et al., 1993). mus301⁰⁹⁴, spn-B⁰⁵⁶, and spn-D³⁴⁹ are strong loss-of-function alleles and have been described elsewhere (Lohs-Schardin and Sander, 1976; Tearle and Nüsslein-Volhard, 1987; González-Reyes et al., 1997; McCaffrey et al., unpublished data). Df(3L)66C-G28, Df(3R)red³¹, and Df(3R)5B-RXP are deficiencies that uncover mus301, spn-B, and spn-D, respectively (González-Reyes et al., 1997). shg^IG29 is a strong loss-of-function allele of the DE-cadherin gene shotgun (Tepass et al., 1996). The dic¹ allele has been reported elsewhere (Lohs-Schardin and Sander, 1976). Its genetic nature could not be confirmed, as we failed to obtain a deficiency for dic. The list of deficiencies from the Bloomington Drosophila stock Centre were used to try to map dic and are presented in Table 2.

Table 2. Deficiencies Used to Map dic

Deficiency	Cytology
Df(3L)81K19	73A03-74F
Df(3L)W10	74A6-75C1
Df(3L)VW3	76A3-76B2
Df(3L)kto2	76B-76D
Df(3L)rdgC-co2	77A1-77D1
Df(3L)ri-79C	77B-77F
Df(3L)Pc-MK	78A02-78C09
Df(3L)31A/Dp(3;3)C126	78A-79B

Scoring Egg Shell Phenotypes

Eggs laid overnight were collected on apple juice plates and scored as being wild type, weakly ventralized (the dorsal appendages fused) or fully ventralized (no dorsal appendages). Eggs were photographed after mounting in Hoyer's medium.

Staining Procedures

Antibody, DNA, and rhodamine-phalloidin stainings were performed according to standard procedures. Detailed protocols are available upon request. DNA was counterstained with Oligreen™ (Molecular Probes, Eugene, OR; 1:2,000 dilution in phosphate buffer containing 5 mg/ml RNaseA) or DAPI (Sigma, St. Louis, MO; 5 mg/ml). Antibodies were used at the following concentrations: mouse monoclonal anti-Gurken (Ghabrial and Schüpbach, 1999), 1/10; mouse monoclonal anti-Orb antibodies 4H8 and 6H4 (Lantz et al., 1994) from the Developmental Studies Hybridoma Bank (University of Iowa), 1/200 each; mouse monoclonal anti-BicD (Suter and Steward, 1991), 1/10; rat anti-cup (Keyes and Spradling, 1997), 1/2,000; rabbit anti-Arm (Riggleman et al., 1990), 1/100; rat monoclonal anti-α -catenin (Oda et al., 1993), 1/20. FITC, Cy2, Cy3, and Cy5 conjugated secondary antibodies (Jackson Laboratories) were used at a final concentration of 1/200. Images were collected using Bio-Rad (Richmond, CA) 1024, Bio-Rad Radiance or Leica TCS-SP2 scanning Confocal microscopes. Images were assembled using Adobe Photoshop and labelled in Adobe Illustrator.

Acknowledgements

We thank Hiroki Oda, Trudi Schüpbach, Beat Suter, Eric Wieschaus, the Developmental Studies Hybridoma Bank (University of Iowa), and the Bloomington Stock Centre for fly stocks and reagents. Financial support is acknowledged from the EC Marie Curie Programme to R.M.; from the Medical Research Council (UK), the Spanish Ministerio de Ciencia y Tecnología (BMC2003-01512) and the Junta de Andalucía (CVI-280) to A.G.-R.; and from a Wellcome Trust (UK) Principal Research Fellowship to D.St J.