Requirement for DCP-1 Caspase During Drosophila Oogenesis

See allHide authors and affiliations

Science  09 Jan 1998:
Vol. 279, Issue 5348, pp. 230-234
DOI: 10.1126/science.279.5348.230

This article has a correction. Please see:


Caspases, a class of cysteine proteases, are an essential component of the apoptotic cell death program. DuringDrosophila oogenesis, nurse cells transfer their cytoplasmic contents to developing oocytes and then die. Loss of function for thedcp-1 gene, which encodes a caspase, caused female sterility by inhibiting this transfer. dcp-1 nurse cells were defective in the cytoskeletal reorganization and nuclear breakdown that normally accompany this process. Thedcp-1 phenotype suggests that the cytoskeletal and nuclear events in the nurse cells make use of the machinery normally associated with apoptosis and that apoptosis of the nurse cells is a necessary event for oocyte development.

Apoptosis, a form of programmed cell death, is a mechanism used by organisms to remove cells that are superfluous, abnormal, or no longer needed (1,2). The failure of cells to undergo apoptosis at the appropriate time during development can lead to abnormal differentiation of tissues and the death of the organism (3). Large numbers of cells of the developing germ line of vertebrates and invertebrates are lost as a result of apoptosis (4); however, it is not clear if apoptosis is a required event in the germ line of some organisms.

The Drosophila ovary consists of individual egg chambers, each of which contains 16 sister germline cells that remain interconnected because of incomplete cytokinesis (5,6). One germ cell becomes an oocyte, and the rest develop into nurse cells that are linked to the oocyte by cytoplasmic bridges (ring canals). The nurse cells become polyploid and synthesize large amounts of RNA and protein, which are rapidly transferred to the oocyte through the ring canals as a result of actin- and myosin-based contraction of the nurse cells late in oogenesis (7,8). Several hours later, the nurse cell nuclei degenerate, and the oocyte proceeds through the final stages of oogenesis. Mammalian ovarian development also includes the formation of intercellular bridges between germ cells at a period of massive germline cell death (9).

Caspases, a class of cysteine proteases, are an essential component of the apoptotic machinery (10). Upon activation, caspases cleave a variety of substrates including structural and regulatory proteins (10, 11). ThreeDrosophila caspases have been identified (12,13), with mutations isolated in one of them, DCP-1. Homozygous dcp-1 mutants die as larvae (12), yet homozygous embryos do not show substantial defects in cell death. However, a role for dcp-1 in early embryonic cell death cannot be ruled out because dcp-1 mRNA is synthesized in the ovary and maternally supplied to early embryos (12).

To generate a complete loss of function for dcp-1, we removed the maternal component by generating homozygousdcp-1 clones in the female germ line (14). Females carrying thedcp-1 germline clones (GLCs) were sterile. Although the dcp-1 GLC females could lay a few eggs, these embryos failed to develop properly, and most embryos were arrested within the first few mitotic divisions. The early arrest of dcp-1 GLC embryos indicated that oogenesis was not proceeding normally. Indeed, dissected ovaries from dcp-1GLC females contained many abnormal late-stage egg chambers (Fig.1 and Table 1). These egg chambers showed a “dumpless” phenotype (6) in which nurse cell contents were not transferred into the oocyte. Nurse cells die by apoptosis late in oogenesis (15), because they stain positively for TUNEL (Tdt-mediated deoxyuridine triphosphate nick end labeling) and acridine orange (AO), two markers for apoptotic cells (16). Using these methods, we examined whether apoptosis occurred in the dcp-1 GLC nurse cells (Fig. 1). In control egg chambers, the nurse cell nuclei stained positively for TUNEL and AO during stages 12 and 13 of oogenesis (15) [stages according to (5)]. In contrast, in dcp-1 GLC egg chambers, TUNEL staining was frequently delayed and was found in stage 14 egg chambers. Therefore, the loss of dcp-1 function did not completely inhibit nurse cell apoptosis.

Figure 1

Dumpless egg chambers fromdcp-1 GLC females with a delay in apoptosis. Egg chambers are labeled with TUNEL (A to D) or AO (E to H) (16). All panels are oriented with anterior to the left. (A) Stage 13 control egg chamber shows TUNEL-positive staining in the nurse cell cluster. N, nurse cell cluster; O, oocyte. (B) By stage 14, the TUNEL-positive material has been cleared from the egg chamber. Arrowheads indicate dorsal appendages that are fully developed at stage 14. (C and D) Stage 14 dcp-1 GLC egg chambers have TUNEL-positive nurse cell nuclei. The morphology of these egg chambers is abnormal because a substantial amount of nurse cell cytoplasm (arrows) has failed to be transferred to the oocyte. (E) Stage 12 control egg chamber shows AO-positive nurse cell nuclei. (F) Stage 13 control egg chamber with one AO-positive nucleus. (G) Stage 14 dcp-1 GLC egg chamber has a substantial amount of AO-positive staining. (H) Another stage 14 dcp-1 GLC egg chamber with a severe dumpless phenotype stains positively for AO. Scale bars, 50 μm.

Table 1

Expressivity of the dcp-1 phenotype. The severity of the dcp-1 phenotype is likely to be underestimated because of a low frequency of recombination that has been observed between the dcp-1 mutant chromosome and the CyO balancer and because of incomplete penetrance of the Cy phenotype. Variability of the dcp-1 phenotype may be due to the perdurance of dcp-1 mRNA or protein or the activity of a related caspase, drICE, which is present during oogenesis (13). Frequency and strength of the stage 14 dumpless phenotype are given. Egg chambers were scored as strong (for example, Fig. 1D), weak (for example, Fig. 1C), or normal. chic, a dumpless mutant (8), was used as a control.

View this table:

To further analyze the nurse cell phenotype in the dcp-1 GLC egg chambers, we used a nuclear β-galactosidase (β-Gal) marker carried on the dcp-1 mutant chromosome (17). Ovaries from dcp-1 heterozygotes and dcp-1 GLC females were stained with the use of X-gal as a substrate for β-Gal (Fig. 2). Heterozygous females displayed wild-type egg chambers with β-Gal activity that was tightly localized to nurse cell nuclei at early stages. Beginning in stage 10B, nurse cell nuclei became permeable, and β-Gal diffused into the cytoplasm. β-Galactosidase was then transferred to the oocyte during the dumping stage. The dcp-1 GLC ovaries had normal β-Gal expression in early egg chambers. However, the dcp-1 GLC nurse cell nuclei were not permeable during stage 10B, and β-Gal remained nuclear even in some late stage 14 egg chambers (Table 2). Additionally, β-Gal was frequently not transferred to the oocyte. In this assay, β-Gal serves as a marker for the likely fate of other nurse cell nuclear proteins. A failure to transfer critical nuclear proteins may explain the early mitotic arrest of the dcp-1GLC embryos.

Figure 2

Nurse cell nuclear permeability blocked in dcp-1 egg chambers. Heterozygous dcp-1/+ (A to D) anddcp-1 GLC (E to H) egg chambers were stained with X-gal to detect β-Gal activity (17). Heterozygotes had egg chambers that appeared wild type. (A) Stage 10A egg chamber displays nuclear β-Gal staining in the nurse cells and oocyte. Follicle cells (fc) have a columnar shape. (B) Stage 10B egg chamber. β-Galactosidase has diffused out of nurse cell nuclei that are most proximal to the oocyte. Follicle cells (fc) have flattened over the oocyte. (C) Late stage 11 egg chamber. β-Galactosidase has diffused out of all nurse cell nuclei, except the most distal ones, and has begun to be transferred to the oocyte. (D) Stage 14 egg chamber. All β-Gal has been transferred to the oocyte. (E) Stage 10A dcp-1 GLC egg chamber. (F and G) Stage 10B (F) and 12 (G)dcp-1 GLC egg chamber. β-Galactosidase remains nuclear and the follicle cells fail to flatten. The size of the mutant egg chambers at late stages was generally smaller than that of the wild-type egg chambers. (H) Stage 14 dcp-1 GLC egg chamber, with chorion (c). Scale bars, 50 μm.

Table 2

Frequency of egg chambers with discrete nuclear β-Gal staining.

View this table:

Breakdown of the nuclear envelope is a central event during apoptosis and is accompanied by caspase-mediated cleavage of nuclear lamins (18, 19). Nuclear lamins are intermediate filament proteins that assemble next to the inner nuclear membrane to form the nuclear lamina (20). To address whether nuclear lamins were degraded as the nurse cell nuclei became permeable, we examined egg chambers for the distribution of lamin Dm0, aDrosophila homolog of mammalian lamin B (21) (Fig. 3). A monoclonal antibody (mAb) to Dm0 revealed the loss of lamin Dm0 signal in control nurse cells. Whereas early egg chambers showed sharp nuclear envelope staining of nurse cells, by stage 11, lamin staining was a diffuse cytoplasmic cloud around the nuclei. In contrast to the control, dcp-1 GLC nurse cells continued to show distinct nuclear envelope staining even as late as stage 14. Thus,dcp-1 mutants were defective in the cleavage or disassociation, or both, of nuclear lamins. This failure in lamin breakdown is a likely cause of the defect in nuclear permeability revealed by the β-Gal marker. Lamin breakdown may be directly due to DCP-1 protease activity because purified DCP-1 protein is capable of cleaving lamin Dm0 in vitro (22). In mammalian tissue culture cells, mutation of the caspase cleavage site in nuclear lamin A causes abnormal morphology and a delay in the nuclear dissolution of cells undergoing apoptosis (19). Although this effect on cultured cells is transient because the cells still die, in a developing tissue such as the Drosophila egg chamber precise timing may be critical. The failure to cleave correct substrates at the appropriate time in development may lead to severe abnormalities.

Figure 3

Inhibition of nuclear lamin breakdown in dcp-1 nurse cell nuclei. Confocal images of control (A to C) and dcp-1 GLC (D to F) egg chambers labeled with mAb 101, which detects nuclear lamin Dm0 (21). (A andD) In early stage 10A, mutant and wild-type egg chambers are indistinguishable. (B and C) Nuclear lamin staining becomes diffuse in stage 11 (B) and 12 (C) control egg chambers. Note the difference in intensity between nurse cells (nc) and follicle cells (fc). (E and F)dcp-1 GLC egg chambers continue to show distinct nuclear staining at stages 11 (E) and 14 (F). Arrowhead indicates dorsal appendage. Scale bar, 50 μm.

The process of nurse cell dumping is accompanied by alterations in the actin cytoskeleton (7, 8, 23). To examine whether dcp-1 mutants were defective in the cytoskeletal reorganization in nurse cells, we stained egg chambers with rhodamine-phalloidin, which binds to filamentous actin (Fig.4) (24). Actin was localized to the plasma membrane during early stages in control anddcp-1 GLC egg chambers. During stage 10B in control egg chambers, actin bundles formed throughout the cytoplasm, connecting the nuclei and plasma membrane. In contrast, actin in many dcp-1GLC egg chambers remained associated with the plasma membrane, even in stage 14 egg chambers (Table 3). Therefore, dcp-1 activity is required for the proper formation of cytoplasmic actin bundles in nurse cells.

Figure 4

Failure to reorganize filamentous actin in dcp-1 egg chambers. Confocal images of control (A to C) and dcp-1 GLC (D to F) egg chambers labeled with rhodamine-phalloidin (24). (A and D) Stage 10A egg chambers show subcortical actin staining. (B and C) Cytoplasmic actin bundles have formed in stage 10B (B) and 12 (C) egg chambers. (E and F) Actin bundles fail to form in dcp-1 GLC egg chambers at stages 11 (E) and 14 (F). Scale bar, 50 μm.

Table 3

Frequency of egg chambers with cytoplasmic actin bundles.

View this table:

Mutations in certain genes, such as chickadee andsinged, lead to dumpless phenotypes (8,25) similar to dcp-1 GLCs. Molecular analysis of these genes revealed that they encode Drosophila homologs of the actin-binding proteins profilin and fascin, respectively. Mutants lacking these genes showed a disruption of the process of actin polymerization and bundling and are thought to display dumpless phenotypes because the nurse cell nuclei become lodged within the ring canals during nurse cell dumping. The nuclei within the ring canals then obstructed the flow of cytoplasm to the oocyte. Nurse cell nuclei within the dcp-1 GLC egg chambers remained well spaced and did not appear to block the ring canals, suggesting thatdcp-1 activity may be required for the actin- and myosin-based contraction that drives the dumping process.

These cytoskeletal changes seen in the nurse cells are similar to events that occur in cells undergoing classical apoptosis (1, 26, 27). In cultured cells, cytoskeletal alterations lead to the formation of blebs and apoptotic bodies. Cytoskeleton-associated proteins such as fodrin, Gas2, and PAK2 may play critical roles in cytoskeletal reorganization because these proteins are cleaved during apoptosis or in vitro by caspases (11, 28). Disruption of actin polymerization or PAK2 function blocks the formation of apoptotic bodies (11,27).

In the Drosophila ovary, the process of nurse cell cytoplasmic dumping and degeneration occurs in a highly ordered and reproducible manner. Our results show that a caspase is required for multiple events during this process. The dcp-1 GLC nurse cells were defective in nuclear breakdown, cytoskeletal reorganization, and membrane contraction. Although these morphological changes are characteristic of cells undergoing apoptosis, there are several unusual aspects to this nurse cell death. DNA fragmentation in the nurse cells occurs after the majority of cytoplasm has been lost. It is likely that apoptotic effectors, such as active DCP-1 protein, are transferred to the oocyte during the dumping process. However, the oocyte escapes damage, suggesting the existence of a protective mechanism. Whereas current views of apoptosis invoke extensive degradation of cellular proteins and organelles (10, 29), nurse cells contribute functional proteins and mitochondria to the oocyte (30). Our findings imply that the activity of caspases in this system is inhibited or restricted, perhaps by cellular compartment or substrate availability. In this way, precise surgical cuts are made of only the appropriate targets. Finally, this process is a clear example of how a single cell, the oocyte, uses the death of its sister cells to develop properly.


Stay Connected to Science

Navigate This Article