Main

The Drosophila caspases include three initiators with long prodomains – dronc (Drosophila Nedd-2-like caspase), dredd (Death-related ced-3/Nedd2-like) and strica, and four effectors – dcp-1 (death caspase-1), drice (Drosophila ice), damm and decay.1 Mutations have been reported in dronc, dredd, decay, dcp-1 and drice.1, 2, 3, 4, 5, 6, 7, 8, 9, 10 Embryos maternally and zygotically mutant for dronc display head involution defects and a significant decrease in apoptotic cells.4, 5 Mosaics and escapers exhibit defects in eyes, wings, spermatid individualization and programmed cell death (PCD) of larval salivary glands.4, 5, 6, 7, 11 dredd, however, functions primarily in the immune response.2 drice mutant embryos show decreased PCD, as well as defects in eye, aristal and wing development.9, 10 dcp-1 is required for germline PCD in mid-oogenesis.3 While some adults contain extra bristles12 and there is reduced apoptosis in eye discs,8 mutants show no other defects in PCD.

Caspases have been shown to display redundancy, with one caspase compensating for loss of another. For example, single mutations in mammalian caspases-3 and -7 resulted in healthy mice, whereas double knockouts displayed cardiac defects and died shortly after birth.13 In the Drosophila eye, RNA interference (RNAi) depletion of drice and dcp-1 suppressed head involution defective (hid)-induced death better than removal of either caspase alone.14 dcp-1; drice double mutants have extra aristal branches and decreased levels of embryonic apoptosis, compared to single mutants.9, 10 The failure of dronc mutants to display a complete loss of PCD raises the possibility of redundancy with another caspase.5 Clues toward the identity of this caspase may come from RNAi studies showing that dronc and strica suppressed hid-induced PCD better than single RNAi depletions.14

PCD in the Drosophila ovary is less well understood than in other tissues. Individual egg chambers are composed of an oocyte and 15 germline-derived nurse cells, surrounded by approximately 650 somatic follicle cells.15 Egg chambers proceed through 14 stages (st) of development, and toward the end of oogenesis transport (dump) their cytoplasmic contents into the growing oocyte. After dumping, fragmented and condensed nurse cell nuclear remnants are phagocytosed by the surrounding follicle cells.16

In addition to developmental nurse cell death during late oogenesis, entire egg chambers can undergo degeneration during mid-oogenesis in response to limited nutrients, certain chemicals, defects in hormone signaling, or poor environmental conditions.17 In contrast to developmental nurse cell PCD which occurs normally during the production of mature egg chambers, PCD during mid-oogenesis is thought to be the outcome of a checkpoint, where there is a final opportunity to eliminate defective egg chambers before vitellogenesis.17

PCD during oogenesis appears to be regulated differently than cell death in other Drosophila tissues. Reaper, Hid and Grim, which are required for almost all embryonic PCD,1 are not required for nurse cell death.18 Although caspases are expressed at high levels just before nurse cell death,19, 20, 21, 22, 23 studies utilizing caspase reporters and inhibitors suggest that they do not play a role in developmental PCD.24, 25 Caspase-3 activity is extremely high throughout egg chambers undergoing mid-oogenesis degeneration.24 We have previously shown that the effector caspase dcp-1 is essential for germline PCD in mid-oogenesis.3 Starved dcp-1 mutants show st7-8 egg chambers with uncondensed nurse cells and a lack of follicle cells,3 suggesting that follicle cells died but nurse cells persisted. Overexpression of Drosophila inhibitor of apoptosis protein 1 (diap1) or the baculovirus caspase inhibitor p35 results in mid-oogenesis defects similar to dcp-1 mutants,24, 25 supporting a requirement for caspases during mid-oogenesis PCD.

Here, we further analyze the role of caspases and their inhibitors in Drosophila oogenesis. Neither dronc nor dredd mutants were found to disrupt PCD in the ovary at either mid- or late oogenesis. Examination of a third initiator caspase, strica, was performed by creating deletions in the gene. Whereas small deletions in strica resulted in moderate defects during oogenesis, a deletion of the entire gene resulted in only a few abnormal egg chambers. Double mutants removing strica and dronc demonstrated a high frequency of egg chambers defective in both mid- and late oogenesis. Additionally, dcp-1 and drice act redundantly during late oogenesis. Contrary to previous reports, our findings indicate that caspases do participate in developmental nurse cell death. We have revealed a novel pathway of PCD in the ovary that utilizes strica, dronc, dcp-1 and drice, and illustrates significant redundancy among the caspases.

Results

Dynamic regulation of DIAP1 during oogenesis

To investigate the regulation of caspase activity in oogenesis, we examined the expression of the caspase inhibitor DIAP1. Immunostaining of wild-type ovarioles with anti-DIAP1 revealed that DIAP1 expression was high from the germarium until st7. It then decreased during st7–8 before increasing again at st9–11 (Figure 1a). Expression decreased again at st11, and localization of DIAP1 shifted from nuclear to cytoplasmic (Figure 1b and c). Nuclear localization in nurse cells has also been observed with a different DIAP1 antibody (K White, personal communication). These results are consistent with previous findings on diap1 transcript levels by in situ hybridization.18 Interestingly, these stage-specific decreases in diap1 expression coincide with the stages when PCD occurs.

Figure 1
figure 1

DIAP1 expression and localization during oogenesis. (a–g) Wild-type egg chambers. (a) Anti-DIAP1 staining (red) is high from the germarium to st7 and at st9. DIAP1 expression is low during st7–8 (arrow). (b) DIAP1 is localized to the nurse cell nuclei before st11. (c) A st12 egg chamber demonstrating that DIAP1 becomes cytoplasmic. (d) 4′,6 diamidino-2-phenylindole (DAPI) staining of a st8 degenerating egg chamber (arrow) and a healthy st10 egg chamber (arrowhead). (e) The same egg chambers as in (d), stained with anti-DIAP1. The degenerating egg chamber (arrow) displays no nuclear and very little cytoplasmic staining of DIAP1, while the healthy egg chamber (arrowhead) shows high DIAP1 staining in the nurse cell nuclei. (f) DAPI staining of additional degenerating (arrow) and healthy (arrowhead) mid-stage egg chambers. (g) The egg chambers in (f) stained with anti-active caspase-3 (red) show a pattern of staining reciprocal to DIAP1. High levels of active caspase-3 are seen in the degenerating chamber (g, arrow), while staining is present only at very low levels in the cytoplasm of a healthy egg chamber (g, arrowhead). (h) NGT/UASp-diap1 9-4; nosGAL4VP16 ovarioles showed higher and mostly cytoplasmic anti-DIAP1 staining at st10 compared to wild type. They also demonstrated a drastic reduction in staining during mid-oogenesis (arrows), similar to wild-type ovaries (a). (i) The same image as (h) with brightness enhanced in Adobe Photoshop. (j) NGT40/UASp-lacZ; nosGAL4VP16 ovarioles stained with anti-β-galactosidase showing expression during st7–10. Arrow indicates mid-stage egg chamber. All images except (d, f and j) were visualized on the confocal microscope, while (d, f and j) were taken on a conventional fluorescent microscope

Full size image

To further examine regulation by DIAP1, we compared DIAP1 and active caspase-3 staining. In healthy egg chambers at mid-oogenesis, active caspase-3 and DIAP1 stained reciprocally. DIAP1 was nuclear, while caspase-3 activity was found at low levels in the cytoplasm (Figure 1d–g, arrowheads). However, a startling contrast was observed in degenerating egg chambers, where very low levels of DIAP1 were found to be solely cytoplasmic and excluded from nurse cell nuclei (Figure 1d and e). Conversely, active caspase-3 in degenerating egg chambers displayed intense staining throughout the egg chambers (Figure 1f and g, arrows).

Overexpression of diap1 or p35 inhibits germline PCD during mid- and late oogenesis

We also investigated the role of caspases in oogenesis by overexpressing diap1 in the germline with the UASp/nanos-Gal4:VP16 system.26 As expected, egg chambers from UASp-diap1; nanos-Gal4:VP16 transgenic flies displayed a significant increase in DIAP1 expression relative to controls (Figure 1a and h, i). Interestingly, overexpression lines showed a dramatic decrease in DIAP1 at st7–8 in healthy egg chambers, similar to wild-type flies (Figure 1a and h, i, arrows). nanos-Gal4:VP16 can drive expression of UASp-lacZ during st7–8 (Figure 1j)26; thus the lack of DIAP1 overexpression driven by Gal4 at this stage indicates that downregulation of DIAP1 is post-transcriptional. This may occur through the autoubiquitination of DIAP1 via its RING domain acting as a ubiquitin ligase.27

Overexpression of diap1 or p35 resulted in defective egg chambers during mid-oogenesis (Figure 2b) as reported previously,24, 25 resembling dcp-1 mutants.3 The overexpression lines displayed 93–100% defective mid-stage egg chambers, while control ovaries contained only 2% defective egg chambers (Table 1).

Figure 2
figure 2

Overexpression of diap1 and p35 inhibits mid-oogenesis nurse cell death, and partially disrupts late-oogenesis nurse cell death. (a) NGT40; nanos-Gal4:VP16 control ovaries contain degenerating egg chambers characterized by condensed nurse cell nuclei at st7–9 (arrows). (b) UASp-p35 ovaries display defective egg chambers during mid-oogenesis, characterized by a lack of follicle cells but containing large, uncondensed nurse cell nuclei. (c) NGT40; nanos-Gal4:VP16 egg chambers undergo condensation and phagocytosis of nurse cell nuclei by st14. (d) Approximately 20% of UASp-p35 st14 egg chambers show uncondensed, persisting nurse cell nuclei (arrow). Egg chambers are stained with DAPI

Full size image
Table 1 Phenotypic quantification of diap1 and p35 over-expression
Full size table

In diap1 and p35 overexpression lines, cytoplasmic transport occurred normally, indicating that caspases do not play a major role in dumping. However, UASp-diap1 and UASp-p35 ovaries exhibited persisting nurse cell nuclei in up to 21% of st14 egg chambers (Figure 2d, Table 1), suggesting that caspases are involved in the final stages of nurse cell PCD. Combined overexpression of diap1 and p35 resulted in 31% persisting nuclei (Table 1). This increased percentage in the double transgenic lines may reflect an additive effect of inhibiting two separate pathways, suggesting that p35 and diap1 can inhibit different caspases. Overexpression of diap1 in the ovary did not appear to affect embryo viability but egg production was reduced in older females (data not shown).

dronc and dredd are not required for PCD during oogenesis

We wished to determine which caspases were required for PCD in oogenesis. Previous studies have demonstrated that dronc is required for most PCDs at multiple developmental stages.4, 5 Furthermore, mutations in dronc suppressed degeneration of the ovary induced by a partial loss of diap1.5 Based on these findings, dronc was a strong candidate to play a role in PCD during oogenesis.

droncI24 and droncI29 are both semi-lethal nonsense mutations.5 Surviving flies displayed extra aristal branches on the posterior axis in comparison to wild type (Figure 3a and b), demonstrating a requirement for dronc in antennal development. This phenotype is similar to that seen in hid mutants,28 suggesting that dronc acts downstream of hid in the arista.

Figure 3
figure 3

Mutations in dronc result in increased number of aristal branches but do not have a major effect on oogenesis. (a) yellow white (yw) flies display aristae with approximately five posterior (p) branches. (b) droncI24 homozygous flies contain approximately 6–8 extra branches on the posterior. In the ovary of both healthy yw (c) and droncI24 (d) flies, there are occasional degenerating egg chambers (arrows) seen at mid-oogenesis. (e) In droncI29 ovaries, both normally degenerating egg chambers (not shown) and abnormal egg chambers (arrowhead) at mid-oogenesis are occasionally seen, characterized by condensed nurse cells and few or no follicle cells. (c–e) were stained with DAPI

Full size image

Despite the requirement for dronc in other tissues, only minor defects during oogenesis were seen in either the few homozygous survivors or in germline clones (GLCs; Figure 3c and d). droncI29 GLC ovaries displayed slight defects in mid-oogenesis PCD, where some egg chambers showed condensed nurse cell nuclei but no somatic follicle cells (Figure 3e). Normally the dying nurse cells are removed before the follicle cells die. Slight defects were also seen in late oogenesis, as 10% of dronc GLC st14 egg chambers contained persisting nurse cell nuclei. The lack of a major phenotype, however, indicates that dronc is not required for PCD in oogenesis. Consistent with this, we found no phenotype with germline expression of dominant negative (DN) dronc constructs (data not shown). However, the possibility of redundancy between dronc and another caspase(s) still existed.

Given the absence of any significant oogenesis defects in dronc mutants, we analyzed a possible role for another initiator caspase, dredd. dreddB118 flies are viable,2 and examination of their ovaries revealed no abnormal phenotype during oogenesis. This suggests that dredd, on its own, has no significant role in oogenesis. To determine if dredd shares a redundant role with dronc, dreddB118; UASp-droncDN/nanos-Gal4:VP16 flies were generated. Again, no ovary defects were observed (data not shown) indicating that dronc and dredd do not have overlapping functions. Furthermore, dreddB118dronc GLC were examined and no defects in mid-oogenesis were observed (data not shown). This suggests that either an initiator caspase may not be required during oogenesis, or the third initiator, strica,22 is the key to transducing the apoptotic signal.

Generation of strica alleles through imprecise P-element excision

To generate mutations in strica, imprecise P-element excision was performed. We used the P-element insertion, P{XP}d06491, which was located 11 bp (basepairs) upstream of the predicted start of strica transcription (Figure 4a). We identified four of 295 excision lines that carried deletions in strica (Figure 4). While the first three deletion alleles still leave part of the coding region, strica4 is a 2.63 kb deletion removing 355 bp upstream of the P-element insertion site, the entire 527aa coding region and the all of the transcribed region except the final 70 bp (Figure 4a). All alleles of strica are viable and fertile, and do not appear to disrupt the flanking genes.

Figure 4
figure 4

Genomic organization of strica. (a) The genomic organization at the strica locus for the P-element insertion line, P{XP}d06491, and the four deletion lines. The three exons of strica are labeled I, II and III, and the first and last (527) amino acid positions are indicated. Deletions are indicated by dotted lines. Everything is drawn to scale except for the size of the P-element. strica1 is a 230 bp deletion removing the region from the P-element insertion site through the first 18aa. strica2 is a 1084 bp deletion removing the first 118aa and 549 bp upstream of the P-element insertion, while strica3 is a 1047 bp deletion removing the region from the P-element insertion site to the first 291aa. strica4 is a 2.63 kb deletion removing 355 bp upstream of the P-element insertion site, the entire 527aa coding region and all of the transcribed region except the final 70 bp. (b) Quantitative real time RT-PCR of strica mutants. Each bar represents the average of three separate biological replicates (RNA isolations). Expression levels are shown as mean normalized mRNA expression values, taking into account primer efficiency. Error bars represent the standard error of measurement (S.E.M.)

Full size image

strica mRNA levels in adult flies were examined by quantitative reverse transcriptase polymerase chain reaction (RT-PCR), and confirmed that strica4 flies are null. Furthermore, strica2 and strica3 were each found to express approximately 2.2-fold less strica mRNA than control (Figure 4b). Decreased strica expression was also seen in P{XP}d06491 flies (Figure 4b). It is unclear whether the mRNA expressed in strica2 and strica3 flies is translated. There are several potential translational start sites that could put the truncated proteins into their normal frame. The similarity in phenotype to P{XP}d06491 (see below) suggests that these alleles have some residual function.

strica participates during germline PCD in mid-oogenesis

To determine whether strica plays a role in oogenesis, ovaries from strica mutants were examined. The original insertion line, P{XP}d06491, and two partial deletion alleles, strica2 and strica3, showed clear defects in mid-oogenesis. Approximately 30% of dying mid-stage egg chambers in strica2 and strica3 (Figure 5b and f) showed defects in germline PCD, resembling dcp-1Prev1 mutants.3 Additionally, a small percentage of partially defective egg chambers resembling those seen in droncI29 were found (Figure 5c and f). Surprisingly, no abnormal mid-stage egg chambers were seen in ovaries of strica4, the null allele (Figure 5f). The phenotypes observed in the partial deletion alleles are not due to a background mutation on the starting chromosome, as precise excision lines did not show the defective phenotypes. Additionally, strica4 in trans to a deletion of the region (Df(2R)nap9) (Df, deficiency) also appeared similar to wild type, while strica2/Df(2R)nap9 and strica3/Df(2R)nap9 displayed approximately 35% abnormal egg chambers (Figure 5f; data not shown). Furthermore, the presence of abnormal mid-stage egg chambers in strica2/strica3 (Figure 5f) and P{XP}d06491 (not shown) ovaries suggest that partial loss of strica is responsible for the mid-oogenesis defects. However, the possibility of truncated proteins suggests that they could be neomorphic or antimorphic alleles.

Figure 5
figure 5

strica mutants show defects in PCD. (a–e) Egg chambers from strica2 flies stained with DAPI. (a) Dying mid-stage egg chambers from strica2 flies displayed phenotypes including (a) normal degenerating egg chambers, (b) defective mid-stage egg chambers characterized by large intact nuclei but a loss of follicle cells and (c) egg chambers with condensed nurse cell nuclei with a loss of follicle cells. During late oogenesis, strica2 flies exhibited both (d) st14 egg chambers that developed normally, with no remaining nurse cell nuclei and (e) st14 egg chambers with persisting nurse cell nuclei (arrow). (f) The percentage of mid-stage egg chambers degenerating or failing to die normally was quantified in both strica mutants and flies that underwent precise excision of d06491. Pwop (peas without pods) refers to the egg chamber seen in (b), while pwop-like refers to that seen in (c)

Full size image

strica alleles that displayed abnormal phenotypes during mid-oogenesis also exhibited minor abnormalities during late oogenesis. While no significant defects were seen in cytoplasmic transport, strica2 displayed 12% of st14 egg chambers with persisting nurse cell nuclei (Figure 5e). Conversely, just as strica4 showed no significant defects during mid-oogenesis, 98% of st14 egg chambers in the strica4 null mutant appeared normal.

Despite the phenotype of strica2 in oogenesis, we failed to see any defects in PCD during embryogenesis in strica2 and strica4 using acridine orange to reveal apoptotic cells or a slit-lacZ enhancer trap which labels midline glia (data not shown). A recent study utilizing RNAi depletion has suggested that strica is required for the death of larval salivary glands.14 However, we did not observe any significant degree of persisting salivary glands in strica4 mutants. Additionally, strica mutants had minimal defects in the pattern of lattice cell apopotosis in the pupal retina (C Brachmann, personal communication).

strica and dronc display redundant functions during mid- and late oogenesis

The lack of any clear phenotypes in strica4 compared to strica2 and strica3 suggested that strica could share a redundant function with another caspase that could compensate when strica is completely deleted. dronc was a strong candidate to display redundancy with strica based on its critical role in many other tissues that undergo PCD. To determine whether strica displays redundancy with dronc, we generated double mutants null for both caspases. strica4; droncI24 and strica4; droncI29 GLC females were fertile but their ovaries contained 31 and 46% of mid-stage egg chambers with missing follicle cells, but large surviving nurse cells (Figure 6b and f) resembling dcp-1 mutants.3 Many of these contained excessive numbers of nurse cells (Figure 6c). Among st6 egg chambers, we found that 18% (n=117) contained excessive nurse cells, indicating a possible defect in PCD in the germarium.17, 29 Double mutants also showed an increase in mid-stage egg chambers with condensed nurse cell nuclei but lacking follicle cells (Figure 6f). Such phenotypes were not seen at any significant level in strica4, droncI24 or droncI29 alone (Figure 6f). Furthermore, no significant phenotypes were seen in strica2/strica4; droncI29 GLCs (Figure 6f), suggesting that residual Strica activity in the strica2 allele is sufficient for normal PCD. Egg chambers from strica4; dronc GLCs that degenerated normally also showed normal levels of activated caspase-3 staining (data not shown), suggesting that downstream caspases could still be activated in the absence of dronc and strica.

Figure 6
figure 6

strica and dronc act redundantly during PCD in oogenesis. (a–e) Egg chambers stained with DAPI. (a) Dying mid-stage egg chamber from strica4 flies showing normal degeneration of the nurse cells before death of the follicle cells. (b and c) Defective mid-stage egg chambers from strica4; droncI24 or I29 GLCs characterized by large intact nurse cell nuclei but a lack of follicle cells. These egg chambers occasionally contained more than 15 nurse cell nuclei, as seen in (c). (d) A strica4 st14 egg chamber developed normally, with no remaining nurse cell nuclei. (e) A strica4; droncI24 or I29 GLC st14 egg chamber displays persisting nurse cell nuclei (arrow). (f) The percentage of mid-stage egg chambers degenerating or failing to die normally was quantified in both single strica and dronc mutants, and GLCs removing both caspases. Pwop (peas without pods) refers to the egg chamber seen in (b and c), while pwop-like refers to the phenotype shown previously in Figure 5c

Full size image

In late oogenesis, strica4; dronc GLC ovaries contained 16–21% st14 egg chambers with persisting nurse cell nuclei (Figures 6e and 7). Persisting nurse cell nuclei were not observed at any significant level in strica4 alone, and in only 10% of droncI24 or droncI29st14 egg chambers (Figure 7). To verify that these phenotypes were due to the combined loss of strica and dronc, double mutants carrying a dronc+ genomic rescue transgene were generated. These flies showed completely normal degeneration during both mid- and late oogenesis (Figure 6f; data not shown).

Figure 7
figure 7

Caspases display redundant functions during late oogenesis. The percentage of st14 egg chambers appearing normal or containing persisting nurse cell nuclei was quantified in both caspase mutants and flies overexpressing caspase inhibitors

Full size image

To determine if dredd also acted redundantly with strica, we examined dreddB118strica4 double mutants and found that they did not display any defects during oogenesis (data not shown). Additionally, dreddB118; strica4; droncI29 GLCs were generated and phenotypes were not more severe than strica4; droncI29 GLCs alone (data not shown).

dcp-1 and drice display redundancy during late oogenesis

Because dronc and strica demonstrated redundancy during oogenesis, we hypothesized that other caspases may also display overlapping functions in nurse cell death. Recent studies have shown dcp-1 and drice to be redundant in the embryo and arista,9, 10 implicating them as good candidates for redundancy in the ovary as well. dcp-1Prev1 mutants display 10% persisting nuclei during late oogenesis, while drice17 mutants show only 5% persisting nuclei. However, dcp-1Prev1drice17 double-mutant GLCs displayed 21% persisting nurse cell nuclei (Figure 7). Thus, dcp-1 and drice also appear to play redundant functions during late oogenesis.

Discussion

Our previous findings with dcp-1 and overexpression of diap1 indicated that caspases are required for PCD during mid-oogenesis.3, 24 This was supported by a subsequent study overexpressing p35 in the female germline.25 However, the identity of the initiator caspase(s) responsible for dcp-1 activity was unclear, as was the role of caspases during late oogenesis. Here we have shown that the ovary utilizes a novel pathway of caspase activation during mid- and late oogenesis that depends largely on redundancy among caspases.

We examined the three initiator caspases, with no significant defects seen in dronc or dredd mutants. The lack of a phenotype in dronc mutants was particularly surprising, considering its importance in most tissues and its ability to suppress the ovarian degeneration caused by loss of diap1.4, 5, 6, 7, 8, 14 Partial deletion strica mutants contained abnormal egg chambers during mid-oogenesis, characterized by absent or degenerating follicle cells and intact nurse cells, similar to those seen in dcp-1 mutants.3 This indicated a role for strica as an initiator caspase participating in the regulation of cell death during mid-oogenesis. Interestingly, strica4, a null allele of strica, did not display any defects during mid-oogenesis. We hypothesized that this may be due to another caspase compensating for the absence of strica.

dronc was a strong candidate to exhibit redundancy with strica based on the central role of dronc in most PCD. Additionally, a previous study showed that strica and dronc display redundancy in the eye.14 Indeed, strica; dronc double mutants contained a high percentage of defective mid-stage egg chambers. Leulier et al.14 have previously proposed that Strica acts in a pathway parallel to a Dronc-Drice cascade. Consistent with this model, we found that strica; drice double mutants also produced egg chambers defective in PCD in mid-oogenesis (data not shown). Because strica; dronc double mutants do not display complete inhibition of mid-oogenesis PCD, as is seen in dcp-1Prev1 and overexpression of the caspase inhibitors, a third caspase may act redundantly with strica and dronc. We tested whether dredd might perform this role, but triple mutant GLCs did not show a stronger phenotype than strica; dronc GLCs. Alternatively, dcp-1 may be activated through a novel mechanism of caspase activation. The drastic downregulation of DIAP1 activity during mid-oogenesis indicates that lower initial levels of caspase activity may be sufficient to initiate the degeneration of an egg chamber. Such low levels of caspase activity would be inhibited at other stages by the higher level of DIAP1.

During late oogenesis, nurse cell nuclei normally degenerate and are removed from the egg chamber by st14. However, we found persisting nurse cell nuclei in several caspase mutant combinations. Importantly, we have found redundancy in late oogenesis between strica and dronc, and between dcp-1 and drice. strica4; dronc GLC ovaries contained 16–21% of st14 egg chambers with persisting nurse cell nuclei, while dcp-1Prev1drice17 GLC ovaries contained 21%. Persisting nuclei were observed at much lower frequencies in single mutants of strica, dronc, dcp-1 and drice (Figure 7), whereas comparable levels were found when diap1 and p35 were overexpressed. Defective PCD in late oogenesis was likely not detected in previous studies with the inhibitors,24, 25 because persisting nuclei only become apparent with increased Gal4 copy number.

Unlike mid-oogenesis, we found that caspase inhibition was able to disrupt developmental PCD in a maximum of 31% of st14 egg chambers. It is possible that a third caspase that is p35 and DIAP1 resistant may also be involved. Alternatively, late oogenesis may be regulated by both apoptotic and non-apoptotic forms of cell death. Redundancy between apoptosis and other forms of cell death has been demonstrated in the mammalian death receptor pathway, which normally leads to apoptosis through caspase-8. Cell death still takes place in certain cell types following inhibition of caspase-8, through a switch to either autophagy or necrosis.30 Persisting st14 nuclei have also been reported in mutants lacking the transmembrane protein Spinster.31 Cell death induced by the human ortholog to Spinster, Hspin1, appears to include the formation of autophagic vacuoles.32 While autophagic vacuoles have been reported previously in nurse cells,33 a recent study using GFP fusion proteins failed to see any evidence of such vacuoles.25 Further study of autophagy in a background where caspases are inhibited is warranted.

Overall, our findings demonstrate that caspases are required for mid-oogenesis PCD, and play a partial role in late oogenesis PCD. During mid-oogenesis, dronc, strica and potentially another factor act redundantly, likely by activating dcp-1. This may occur through Strica and Dronc binding a common adaptor molecule such as Dark (Drosophila Apaf-1-related killer). Thus in the absence of Strica, increased Dronc binding to the adaptor may take place. Interestingly, dark GLCs show a phenotype similar to strica;dronc double mutants in late oogenesis but have normal PCD in mid-oogenesis (BP Bass and KM, unpublished), suggesting there could be a distinct adaptor that functions in mid-oogenesis. Alternatively, redundancy may occur through upstream components that normally initiate PCD via divergent Strica and Dronc pathways. In this model, upon disruption of Strica, signaling through Dronc is upregulated. Although the effect on late oogenesis is not as strong as it is in mid-oogenesis, it makes economical sense that the ovary utilizes a similar mechanism of caspase activation at both stages, in which dronc and strica appear to activate dcp-1, drice and perhaps other effector caspases. Future studies may reveal a caspase-independent pathway that likely plays a role in developmental PCD in the ovary.

Materials and Methods

Fly strains and crosses

Yellow white flies were used as a wild-type control. P{XP}d06491 used in the creation of strica alleles was obtained from the Exelixis Drosophila stock collection at Harvard. UASp-diap1, UASp-p35, UASp-lacZ and UASp-droncDN were crossed to the nanos-Gal4-tubulin (NGT40); nanos-Gal4:VP16 driver. NGT40; nanos-Gal4:VP16 flies were obtained from Rachel Cox34 and UASp-lacZ flies were from Pernille Rorth.23 drice17 (strong loss-of-function), droncI24 (null) and droncI29 (null) flies were received from Andreas Bergmann,5, 10 dreddB118 (null) and dronc+ transgenic flies were obtained from John Abrams.2, 4 dcp-1Prev1 (null) mutants and UASp-diap1 lines were generated in our laboratory.3, 24 All other lines were obtained from the Bloomington stock center. Fly lines were maintained on a standard cornmeal, agar, molasses and yeast food and raised at 25°C. Because of the reduced viability of dronc mutants, GLCs were generated as described previously using the FLP/FRT/ovoD system.35, 36 To generate double mutant GLCs, strica4/CyO; dronc FRT/TM6B females were crossed to HSflp; strica4/CyO; ovoDFRT/MKRS males and progeny were heat shocked twice at 37°C for 1 h during the third instar larval stage. The GLC-carrying strica4/strica4; ovoD FRT/dronc FRT females were collected and conditioned as described below. For triple mutant GLCs, the dreddB118 mutation was crossed into the strica4/CyO; dronc129FRT/TM6B stock or recombined onto the HSflp chromosome (BP Bass, unpublished). dreddB118; strica4/CyO; dronc129FRT/TM6B females were crossed to dreddB118 HSflp; strica4/CyO; ovoDFRT/MKRS males and treated as described above. The dronc+ genomic transgene was recombined onto the strica4 chromosome and crossed into the dronc129FRT/TM6B background. To confirm that the stocks were correct, the strica4 and dreddB118 mutations were verified by PCR and DNA sequencing, respectively. The presence of the dronc mutation was visualized by its pupal lethal phenotype (either on its own or outcrossed to another dronc allele).

Transgenic flies

Three dominant-negative dronc constructs, pro-dronc C>A, ΔN dronc C>A and CARD-only (gifts from Pascal Meier, Institute of Cancer Research, London), were subcloned from pUASt into pUASp via EcoRI and XbaI sites in both vectors to yield UASp-droncDN constructs. For generating the UASp-p35 construct, the pSH36-1 vector (a gift from S Shaham, the Rockefeller University) was digested with BamHI and EagI, and the fragment containing the p35 coding region was ligated into the BamHI and NotI sites of a modified version of the UASp vector, pUASp-2 (a gift from A Hudson, Yale University School of Medicine). Standard Drosophila techniques were used to generate transgenic lines from these constructs.

Imprecise P-element excision

P{XP}d06491 homozygous flies were crossed to Sp/CyO; Sb 2–3/TM6 to mobilize the P-element. Single d06491, w+/CyO; Sb Δ2–3/+ males were crossed to CyO/Gla females. Single male progeny of the genotype d06491 revertant (w−)/CyO were then crossed to CyO/Gla females to create stable lines. Genomic DNA was amplified from single flies as described.37 Revertant homozygotes were screened by PCR for deletions first using primers GGATCGGTCACCAGAAGAAA and GAGGTGGTCGTCGTAGTGGT flanking the insertion site, which produced a 720 bp product in wild-type flies. If no PCR product resulted, PCR reactions were performed with primers TTGTGGGCGTTAAAGTAGGC and CAATTCCGACCCCAGACTAA, which resulted in a 2.7 kb band in wild-type flies, or primers CTCGTTAAGCCGGTGTTTTC and CAATTGCTGGGACGATTTTT, which resulted in a 4.3 kb product in wild-type flies. Both the 2.7 and 4.3 kb PCR reactions were performed with Platinum Taq Polymerase High Fidelity (Invitrogen, Carlsbad, CA, USA).

Aristal dissections and mounting

Antennae were dissected in isopropanol and transferred to a slide. Following the evaporation of isopropanol, a 1 : 1 ratio of Canadian Balsam and methylsalicylate (Sigma, Highland, IL, USA) was added to mount aristae. The tissue was covered with a coverslip and examined under the light microscope.

RNA isolation and RT-PCR

Isolation of RNA was performed using TRIZOL (Invitrogen, Carlsbad, CA, USA). For each sample, RNA was isolated from 15 male flies homogenized in 1 ml of TRIZOL Reagent. cDNA synthesis was performed with the First-Strand cDNA Synthesis Kit (Amersham Biosciences, Piscataway, NJ) using a random hexamer pd(N)6 primer. cDNA was made from three different RNA preparations.

To determine relative levels of cDNA, quantitative real-time PCR was performed. A 30 μl reaction consisted of 15 μl of SYBR Green PCR master mix (Applied Biosystems, Foster City, CA, USA), 3 μl of each primer, 1 μl of the cDNA reaction and 8 μl of water. strica primers ACGAATTCCGAAAGGGAAGT and TCGTCCAGGGAGTAGTCGTC flanking the first intron were used at 5 pmol/μl. Control lamin primers GATCGATATCAAGCGTCTCTG and GCCTGGTTGTATTTGTTGTTC were used at 3 pmol/μl. Samples were divided into triplicates and run on a 7900HT Sequence Detection System (Applied Biosystems). A dilution set was also performed to take into account primer efficiency. Samples were run using the cycle 50°C at 2 min, 95°C at 10 min (40 × (95°C at 15 s, 60°C at 1 min)), 60°C at 1 min and 95°C at 15 s. Analysis of real-time PCR was performed using Q-Gene (Biotechniques), to obtain the mean normalized expression levels.38 Creation of graphs and statistical analysis were performed using GraphPad Prism 4 (GraphPad Software, San Diego, CA, USA).

Ovary dissections and antibody staining

Flies were conditioned on yeast paste for 4–6 days before dissection, and ovaries were dissected in Drosophila Ringers. Egg chambers were stained as described previously.39 To quantify persisting nurse cell nuclei, only st14 egg chambers with fully formed dorsal appendages were counted. Degenerating mid-stage egg chambers occurred spontaneously or were induced by nutrient deprivation as described.24 Primary antibodies included rabbit anti-cleaved caspase-3 diluted 1 : 100 (Cell Signaling, Danvers, MA, USA), mouse anti-β-galactosidase diluted 1 : 400 (Promega) and mouse anti-DIAP1 diluted 1 : 300 (a gift from Bruce Hay). Secondary antibodies included Cy-3-conjugated goat anti-mouse or goat anti-rabbit (Jackson ImmunoResearch Laboratories, West Grove, PA, USA), diluted 1 : 200. Samples were mounted in Vectashield with 4′,6 diamidino-2-phenylindole (Vector Labs, Burlingame, CA, USA). Samples were viewed using an Olympus (Melville, NY, USA) BX60 fluorescent microscope and digital camera (Olympus model U-TVO.5XC). Images were taken using MagnaFIRE SP and processed in Adobe Photoshop. Confocal images were taken on an Olympus Fluoview confocal microscope, and images were processed in Adobe Photoshop.