Cell death during Drosophila melanogaster metamorphosis is controlled by the steroid hormone 20-hydroxyecdysone (20E). Elements of the signaling pathway that triggers death are known, but it is not known why some tissues, and not others, die in response to a particular hormone pulse. We found that loss of the tissue-specific transcription factor Fork head (Fkh) is both required and sufficient to specify a death response to 20E in the larval salivary glands. Loss of fkh itself is a steroid-controlled event that is mediated by the 20E-induced BR-C gene, and that renders the key death regulators hid and reaper hormone responsive. These results implicate the D. melanogaster FOXA orthologue Fkh with a novel function as a competence factor for steroid-controlled cell death. They explain how a specific tissue is singled out for death, and why this tissue survives earlier hormone pulses. More generally, they suggest that cell identity factors like Fkh play a pivotal role in the normal control of developmental cell death.
Programmed cell death (PCD) is essential for the normal development of most, if not all, metazoans. The developmental time at which specific cells or tissues are removed is often specified by the release of systemic or locally acting signaling molecules. During amphibian metamorphosis, for instance, thyroid hormone signals cell death that leads to resorption of the tadpole tail and other larval tissues (Tata, 1994; Shi et al., 2001). During vertebrate limb development, separation of the limb digits requires death of the interdigital regions that is controlled by BMP signaling (Zuzarte-Luis and Hurle, 2005). Although much is known about the temporal aspect of regulation in these and other systems, it is less well understood why some cells and tissues, but not others, die in response to widespread signals (Vaux and Korsmeyer, 1999). A system that is particularly well suited to address this question is the removal of larval tissues by PCD during insect metamorphosis. In particular, the larval salivary glands of Drosophila melanogaster have been extensively used to unravel signaling pathways that control developmental cell death (Baehrecke, 2003; Yin and Thummel, 2005).
Death of the larval salivary glands takes place in the early pupa and is triggered by a pulse of the steroid hormone 20-hydroxyecdysone (20E). The salivary glands survive an earlier 20E pulse that leads to the destruction of the larval midgut (Jiang et al., 1997). The two consecutive hormone pulses that trigger these stage-specific responses are referred to in this study as the late-larval and the prepupal 20E pulse (Fig. 7). Salivary gland death is foreshadowed by transcriptional activation of the death genes reaper (rpr) and head involution defective (hid; Jiang et al., 2000). The protein products of both genes kill by interfering with caspase inhibition by the D. melanogaster inhibitor of apoptosis protein (IAP) 1 (DIAP1). A critical target of DIAP1 is the apical caspase Dronc, which is required for execution of salivary gland death (for review see Kornbluth and White, 2005). The mammalian cell death regulators Smac/Diablo and Omi/HtrA2, which are related to hid and rpr, act in a similar way by antagonizing IAP function (Du et al., 2000; Verhagen et al., 2000; Hegde et al., 2002). Loss of hid, but not rpr, leads to salivary gland persistence (Peterson et al., 2002; Yin and Thummel, 2004). However, rpr has been shown to synergize with hid in bringing about salivary gland death (Yin and Thummel, 2004). Induction of both hid and rpr requires the up-regulation by 20E of transcription factors encoded by E93, Broad-Complex (BR-C), and E74A (Jiang et al., 2000; Lee et al., 2000, 2002). In addition, full induction of rpr depends on direct binding of the 20E receptor EcR/Usp to a salivary gland enhancer of the gene (Jiang et al., 2000). Proper expression of the early hormone response genes and salivary gland death require the transient expression of the nuclear receptor βFtz-F1 in mid–prepupae (Broadus et al., 1999). Thus, βFtz-F1 has the properties of a competence factor for stage-specific hormone signaling (Woodard et al., 1994; Broadus et al., 1999). However, βFtz-F1 expression is observed in almost all larval tissues (Yamada et al., 2000), leaving the question open of how the tissue specificity of salivary gland death is achieved.
Tissue-restricted expression of 20E-regulated genes in the larval salivary glands has been shown to require coregulation by the transcription factor Fork head (Fkh; Lehmann and Korge, 1996; Mach et al., 1996). fkh is already expressed in the salivary glands during embryogenesis, and is required for the proper development of this organ (Weigel et al., 1989; Myat and Andrew, 2000). Expression of fkh during larval development is restricted to the salivary glands and a small number of other tissues, including the lymph glands and Malpighian tubules (Weigel et al., 1989; Kuzin et al., 1994; Wang et al., 2004). The mammalian counterparts of Fkh are the FOXA1, 2, and 3 proteins (also known as HNF3α, β, and γ; Mazet et al., 2003; Lee and Frasch, 2004), which are members of the larger family of Fkh/HNF or Fox transcription factors (Weigel and Jäckle, 1990; Gajiwala and Burley, 2000; Kaestner et al., 2000). Similar to fkh, FOXAs play a role in specifying tissue-specific responses to steroid signaling, suggesting that aspects of FOXA function are evolutionarily conserved (Friedman and Kaestner, 2006).
In fkh mutants, the embryonic salivary glands undergo extensive apoptosis, which is foreshadowed by rpr and hid expression. Whereas this indicates that the presence of Fkh is required for survival of the embryonic salivary glands, other data suggest that the protein has an independent developmental role in secretory cell invagination (Myat and Andrew, 2000). Thus, fkh is part of a long list of developmental genes that cause ectopic cell death when impaired in their function. It has been estimated that nearly 20% of all D. melanogaster genes can cause PCD when mutated (White et al., 1994). It is difficult to establish whether these genes normally participate in the control of apoptosis, or whether activation of the default death pathway is an indirect result of aberrant development (Abrams, 1996).
We show that fkh plays a key role in specifying a cell death response to steroid signaling during normal development. Fkh is lost from the larval salivary glands at the onset of metamorphosis, and this loss is required for the subsequent steroid-induced removal of the tissue. Ectopic expression of fkh rescues the salivary glands and premature knockdown of fkh leads to the premature 20E-induced activation of PCD and of the death genes hid and rpr. Transcription of fkh is down-regulated in a BR-C–mediated response to the late-larval 20E pulse, followed by a loss of the Fkh protein during prepupal development. These data indicate that Fkh protects the salivary glands from hormone-induced death until a stage-specific, hormone-induced loss of the protein earmarks the tissue for destruction in response to future hormone exposure.
Salivary gland expression of fkh ceases at the onset of metamorphosis
We previously showed that fkh is transcriptionally down-regulated in the salivary glands in response to the late-larval 20E pulse, and that this response is mediated, at least in part, by the early 20E response gene BR-C (Renault et al., 2001). However, it was not clear whether this down-regulation is a transient event followed by a resumption of fkh expression, or whether fkh expression remains low or absent in prepupal and pupal salivary glands. To resolve this point, we dissected salivary glands from staged larvae, prepupae, and early pupae. Total RNA extracted from these glands was analyzed for fkh expression by Northern blot hybridization (Fig. 1 A). Consistent with our earlier results, we found that fkh is expressed in the salivary glands of early and mid-third instar larvae (Fig. 1 A, lanes P and -18), and that expression is turned off in an apparent response to the late-larval 20E pulse (Fig. 1 A, lanes -8 and -4). Our Northern analysis did not detect fkh mRNA in early prepupae (Fig. 1 A, lane 0), and it did not indicate that expression resumes at any time before the salivary glands die at ∼14 h after puparium formation (APF). These data suggested that Fkh protein is lost from the larval salivary glands before the tissue is removed in response to the prepupal 20E pulse. To test this prediction and to determine the temporal profile of Fkh protein expression at the larval–prepupal transition, we stained salivary glands dissected from staged animals with a Fkh antibody (Fig. 1 B). Strong immunostaining was observed in the cell nuclei at −4 h APF, a time at which fkh mRNA has almost disappeared from the salivary glands (Fig. 1 A). Fkh protein is still present in considerable amounts in the glands of freshly formed prepupae (0 h APF). 2 h later, the concentration has greatly diminished, and by 4 h APF the protein is reduced to very low levels. Collectively, these data show that fkh is transcriptionally down-regulated in response to the late-larval 20E pulse, and that the protein is still present in the cell nuclei in substantial amounts at the larval–prepupal transition. Subsequently, the protein is lost from the salivary glands during prepupal development.
Loss of Fkh is required for steroid induction of PCD in the salivary glands
We next asked whether the down-regulation of fkh might be required for the salivary glands to undergo PCD in response to the prepupal 20E pulse. When fkh is ectopically expressed from a heat-inducible transgene in the transformant P[hs-Fkh111], the Sgs4 gene, which is normally repressed by the late-larval 20E pulse, fails to be down-regulated (Renault et al., 2001). We used the same transgenic line to express fkh ectopically at 10 h APF, shortly before the prepupal 20E pulse signals salivary gland destruction. Pupae of the fkh-expressing line and heat-shocked control pupae (w1118) were dissected 20 h APF, which is ∼6 h after the salivary glands are normally destroyed. All fkh-expressing pupae still possessed larval salivary glands at this time (n > 37; penetrance = 100%), whereas no salivary glands could be found 20 h APF in any of the heat-shocked w1118 control pupae or in non–heat-shocked P[hs-Fkh111] pupae. The structure of the cells and cell nuclei of most rescued salivary glands appeared well preserved, and the overall morphology of the glands was very similar to that of salivary glands before the onset of PCD (Fig. 2). In some of the rescued salivary glands, the cells appeared to be more round in shape and contained large vacuole-like structures, suggesting that the cell death program had been initiated in these glands. However, in contrast to dying cells at 14 h APF, which were largely depleted of filamentous actin, these structures were still supported by portions of a well-developed actin cytoskeleton. At 26 h APF, the number of persisting salivary glands was substantially reduced (11% persistence; n = 18), which is consistent with the interpretation that some of the glands at 20 h APF had entered the death pathway. We suspected that this was caused by a waning effect of the ectopic Fkh. Therefore, we tested whether survival of the salivary glands could be further prolonged by sustained expression of fkh. After the first heat shock at 10 h APF, we applied a second heat shock at 16 h APF, which resulted in 67% salivary gland persistence at 26 h APF (n = 21). This suggests that activation of the cell death program can indeed be continuously suppressed by maintained fkh expression. Collectively, these data demonstrate that Fkh is sufficient to prevent PCD in the larval salivary glands. Importantly, they suggest that the down-regulation of Fkh before the prepupal 20E pulse provides competence to the salivary glands to respond to this pulse by activation of the death pathway.
We wondered whether rescue of the larval salivary glands by fkh was a tissue-specific effect or whether ectopic fkh would block PCD in other tissues as well. To address this question, we ectopically expressed fkh shortly before the larval midgut normally dies. This did not lead to a delay in the initiation or execution of PCD in this tissue (Fig. S1). Thus, our results support a model in which fkh specifically functions in the larval salivary glands in the developmental control of death.
Fkh represses expression of the death genes hid and rpr
In an attempt to identify genes that might mediate the survival function of Fkh, we analyzed the effect of fkh on the expression of genes that had previously been implicated in the control of salivary gland death. RNA was isolated from the salivary glands of P[hs-Fkh111] and w1118 control animals at different times after heat treatment at 10 h APF. Northern blots of this RNA were first hybridized to detect expression of genes of the 20E-controlled signaling pathway (Fig. 3). The primary hormone response genes E93, E74A, and BR-C, which are required for proper salivary gland death (Jiang et al., 2000; Lee et al., 2000, 2002), were all expressed in the presence of ectopic Fkh. However, the amount of RNA and the timing of expression differed from the controls. Expression of all three genes started earlier when Fkh was present, and E74A and BR-C were also expressed more strongly. The E93 mRNA level appeared to be somewhat diminished, but did not go down by 16 h APF, as it did in the control (Fig. 3). BR-C failed to be down-regulated in 16 h APF salivary glands as well. These data indicate that salivary gland survival in the presence of ectopic Fkh is not likely to be caused by a reduced expression of upstream regulators of steroid-induced death.
Next, we tested whether expression of the downstream death activators rpr or hid was changed by Fkh. We found that rpr mRNA was diminished when compared with the control, but still detectable. More strikingly, expression of hid, which was strongly expressed in the control glands, appeared to be completely suppressed by the ectopic Fkh (Fig. 3). As hid is known to be required for proper salivary gland death, this suggested that repression of hid was at least partially responsible for the suppression of salivary gland death by Fkh.
In fkh mutants, the transcription factor Senseless (Sens) is not properly expressed and, similar to fkh mutants, sens mutants exhibit embryonic salivary gland apoptosis (Chandrasekaran and Beckendorf, 2003). We were therefore interested to also determine the expression of sens in the presence and absence of Fkh. Hybridization with a sens probe showed that induction of fkh was followed by a brief burst of sens transcription at 12 h APF (Fig. 3), confirming that sens is a target gene of Fkh. However, although massive overexpression of sens from a sens transgene delayed salivary gland death, it did not affect the transcript levels of hid or rpr (unpublished data). Thus, sens uses a different pathway than fkh to protect cells from PCD (unpublished data). This conclusion is consistent with the finding that forced expression of sens in fkh mutants does not rescue the embryonic cell death phenotype of fkh (Chandrasekaran and Beckendorf, 2003).
Collectively, our results show that at least part of the effect of Fkh on cell death is mediated by repression of the death genes hid and rpr. They further suggest a crosstalk between fkh and another survival pathway that acts through sens. We also note that fkh is epistatic to the cell death regulator E93 (Lee et al., 2000), which is not sufficient to initiate activation of the death pathway as long as fkh is expressed.
fkh coordinately represses IAP inhibitors and affects other apoptosis-related genes
To provide a broader foundation for our conclusions, we performed a microarray analysis of gene expression at 14 h APF in the presence and absence of ectopic Fkh. Expression of fkh was induced by heat shock in P[hs-Fkh111] animals, and RNA extracted from the salivary glands of these and heat-shocked w1118 control animals was hybridized to Affymetrix Drosophila Genome Arrays. The microarray analysis confirmed that hid and rpr were down-regulated in response to Fkh (16- and 2-fold, respectively). In addition, it revealed that another known IAP inhibitor, Jafrac2 (Tenev et al., 2002), was down-regulated. Overall, the microarray analysis identified 55 genes annotated as functioning in apoptosis whose expression was at least 1.5-fold changed by Fkh (Tables I and S1). hid was among the four most strongly down-regulated genes, as was Companion of reaper, a gene that has been reported to synergize with hid to promote apoptosis (Meier and Silke, 2003). Genes encoding the D. melanogaster Apaf-1 orthologue Ark and the apical caspase Dronc (Nc) were down-regulated by approximately twofold by Fkh. As these proapoptotic proteins are known to be required for the destruction of the larval salivary glands (Cakouros et al., 2004; Daish et al., 2004; Waldhuber et al., 2005; Mills et al., 2006), down-regulation of the corresponding genes is likely to contribute to the antiapoptotic effect of fkh. The two Bcl-2 family members of D. melanogaster, Buffy and debcl, were 2- and 2.5-fold up-regulated by fkh. We found that fkh-independent overexpression of either Buffy or debcl using the UAS/Gal4 system did not affect salivary gland death (unpublished data). These genes therefore have no, or at least no essential, role in mediating the effect of fkh. Interestingly, among the genes strongly up-regulated by fkh was the D. melanogaster PDK1 orthologue. PDK1 is an essential activator of the protein kinase Akt (Rintelen et al., 2001). As signaling through the PI3K–Akt pathway can protect salivary glands from PCD (Liu and Lehmann, 2006), PDK1 is likely to contribute to the survival function of fkh. In summary, the microarray data suggest that Fkh ensures survival by the coordinated repression of IAP antagonists and the regulation of other apoptosis-related genes. In particular, they confirm that the death gene hid, which is required for salivary gland death (Yin and Thummel, 2004), is a prime candidate for a target of the survival function of fkh.
Loss of fkh is sufficient to create competence for a death response to steroid signaling
Our results showed that the down-regulation of fkh in response to the late-larval 20E pulse is required for proper induction of hid and rpr in response to the prepupal 20E pulse and subsequent death of the salivary glands. This prompted us to ask whether down-regulation of fkh is also sufficient to specify these responses to steroid signaling. To address this question, we down-regulated fkh prematurely in early third instar larvae using RNAi. We generated transgenic fly stocks that use a heat-shock promoter to drive expression of a double-stranded (ds) fkh RNA (Lam and Thummel, 2000). Northern analysis confirmed that these lines strongly expressed fkh dsRNA upon heat treatment (unpublished data), followed by a dramatic decline in the amount of Fkh protein (determined by Western analysis; Fig. S2). To further ascertain that expression of the dsRNA led to an effective knockdown of fkh activity in the salivary glands, we examined the expression of known target genes of fkh by Northern blot hybridization. RNA was isolated from the salivary glands of staged fkh dsRNA-expressing animals and w1118 controls. A Northern blot of this RNA was hybridized to detect expression of the Sgs4 and sage genes (Fig. 4). Sgs4 is a well-characterized target gene of fkh that encodes a glue protein and is strongly expressed in third-larval instar salivary glands until puparium formation (Lehmann and Korge, 1996). As expected, large amounts of Sgs4 mRNA were detected in the salivary glands of the heat-shocked control larvae. In striking contrast, Sgs4 mRNA was absent from the salivary glands of the fkh dsRNA-expressing larvae. sage encodes a bHLH transcription factor that is enriched in larval salivary glands and down-regulated at the larval–prepupal transition (Li and White, 2003). Recently, it has been shown that sage is directly activated by fkh, although some sage expression is still observed in a fkh mutant (Abrams et al., 2006). Consistent with these findings, sage RNA was severely, but not completely, reduced after fkh dsRNA expression (Fig. 4). These data indicate that fkh function in the larval salivary glands is severely compromised by RNAi. Importantly, the RNAi knockdown had no effect on overall development and growth of the larvae, as heat-shocked P[hs-fkhRNAi-4] larvae pupariated at the same time and formed prepupae of similar size as heat-shocked w1118 control larvae (unpublished data).
Next, we inspected salivary glands that had been treated with fkh dsRNA for signs of PCD. Staining with acridine orange revealed no difference between fkh RNAi and control glands before puparium formation (Fig. 5, A and B). This indicated that the premature down-regulation of fkh by itself was not sufficient to cause cell death. However, salivary glands after puparium formation exhibited a progressive loss of tissue integrity and showed strong nuclear acridine orange staining (Fig. 5, D and E). The salivary glands of heat-shocked control animals showed no nuclear acridine orange staining, or any signs of tissue disintegration, and died at the normal time in response to the prepupal 20E pulse (Fig. 5 C). The nuclei of the salivary glands of fkh dsRNA-treated animals were stainable with acridine orange as early as 4 h APF (Fig. 5 D). At 9 h APF, salivary glands could still be dissected from some, but not all, prepupae. These glands contained acridine orange–positive cell nuclei, but the cell boundaries, which were clearly discernable in the control glands, had disappeared (Fig. 5 E). Salivary glands could occasionally also be found at later time points. However, the cellular structure and shape of these glands were severely compromised. This suggests that, although the death program had been prematurely activated, the final steps of tissue disintegration and removal could not be performed properly. We overexpressed hid in early prepupae from a heat-inducible transgene and found that the salivary glands responded in a similar way, showing nuclear acridine orange staining, but persisting for several more hours (unpublished data). Collectively, these results strongly suggested that loss of fkh was sufficient to specify a cell death response to 20E signaling.
To further test this conclusion, we asked whether death gene expression was changed after the knockdown of fkh, and, if yes, whether this change occurred in response to the late-larval 20E pulse or earlier. Northern blot hybridization revealed that hid and rpr were strongly activated at or shortly before puparium formation in salivary glands that had been treated with fkh dsRNA. hid reached a peak in expression at 4 h APF, whereas rpr was induced earlier, already reaching a very high transcript level at 0 h APF (Fig. 4). This profile is strikingly similar to the temporal profile of hid and rpr expression observed in response to the late-prepupal 20E pulse. Also at this time, rpr shows maximal expression earlier than hid, which peaks in expression several hours after the early hormone response genes E74A and BR-C are first detected (Jiang et al., 2000). This suggests that, similar to the response in late-prepupal glands, hid induction after premature loss of fkh is a secondary hormone response mediated by E74A and BR-C. This hypothesis is supported by the expression profiles of E74A and BR-C, which are very similar in late-larval and -prepupal salivary glands (Jiang et al., 2000). Hybridization with an E74A probe confirmed that this early response gene is induced at the normal time after knockdown of fkh (Fig. 4). Collectively, the morphological and gene expression data demonstrate that loss of fkh leads to a premature activation of the death program in response to the late-larval 20E pulse. They identify hid and rpr as two key death regulators whose hormone responsiveness is controlled by Fkh.
fkh is derepressed in BR-C mutants that are defective in salivary gland death
The response of fkh to 20E signaling at the end of larval development is mediated, at least in part, by the early 20E-inducible gene BR-C (Renault et al., 2001). The down-regulation of fkh that is normally observed at this time does not occur in 2Bc mutants of BR-C. The continued expression of fkh in these mutants is sufficient to maintain expression of the Sgs4 gene in prepupal salivary glands. 2Bc mutants also show defects in salivary gland death, which are even more pronounced in mutants of the rbp subfunction of BR-C (Restifo and White, 1992; Jiang et al., 2000). Collectively, these observations raised the possibility that a derepression of fkh might be responsible for the persistence of the larval salivary glands in BR-C mutants. To test this possibility, we performed a Northern analysis of fkh expression in the salivary glands of late prepupae and early pupae of the rbp5 mutant (Fig. 6). As expected, the salivary glands of control animals did not show expression of fkh shortly before their destruction, at 12 or 14 h APF. In striking contrast, salivary glands of hemizygous rbp5 mutant animals exhibited strong expression of fkh mRNA both at these times and also in 16-h pupae. In 16-h control pupae, the disintegration of the salivary glands was too far advanced to obtain RNA for a Northern analysis. These data indicate that fkh is indeed derepressed in the persisting salivary glands of BR-C mutants. Hybridization of the Northern blot with a hid probe confirmed that this death gene was not properly induced in the rbp5 mutant (Jiang et al., 2000). The expression of fkh in rbp5 provides an explanation for the persistence of larval salivary glands in BR-C mutants and identifies fkh as at least one critical target of the function of BR-C in programmed cell death. These results lend further support to a model in which fkh is tied into the hormonal signaling pathway through the 20E-controlled BR-C gene (Fig. 7).
Developmental cell death in invertebrates and vertebrates is often controlled by systemic signals, which provide the trigger for cell and tissue destruction (Jacobson et al., 1997; Baehrecke, 2002). However, it is not well understood why these signals induce death only at a particular time and only in some cells and tissues, but not in others. The experimental data presented in this study support a model that explains how a specific tissue of the fruit fly D. melanogaster is singled out for destruction in response to the steroid hormone 20E (Fig. 7). It explains why the larval salivary glands are destroyed in response to a particular 20E pulse, the prepupal pulse, and why they survive earlier pulses of the same hormone. It thus provides a framework for our understanding of how tissue-specific developmental cell death is precisely timed.
In more general terms, our data suggest that a key event in acquiring competence for a cell death response to a systemic signal is the loss of a tissue-specific survival factor. This loss occurs in response to a temporal signal that precedes the death-inducing signal. In the salivary glands, the tissue-specific survival factor is Fkh, and the signal that leads to the loss of Fkh is provided by the late-larval 20E pulse. Our data show that loss of Fkh is required for the death response to the prepupal 20E pulse (Fig. 2). They further suggest that the salivary glands survive all previous hormone pulses because they are protected by the presence of Fkh (Fig. 1). Thus, Fkh not only plays a key role in determining the tissue selectivity of salivary gland death, but also in the proper timing of this event. Elimination of Fkh is mediated by the 20E-induced transcription factor BR-C (Renault et al., 2001). We find that fkh is strongly expressed at the normal time of salivary gland death in the rbp5 mutant of BR-C, which demonstrates that BR-C is required for the continued repression of fkh beyond the larval–prepupal transition (Fig. 6). The derepression of fkh in rbp5 is sufficient to explain why the salivary glands do not die in the mutant.
Loss of the survival factor renders critical death regulators responsive to the death-inducing signal. In the salivary glands, these death regulators are the IAP antagonists hid and rpr, which together are required for salivary gland death (Yin and Thummel, 2004). In the absence of Fkh, the two genes are inducible by hormone, as shown by the premature induction of hid and rpr after RNAi knockdown of fkh (Fig. 4). Importantly, loss of fkh by itself is not sufficient to activate hid and rpr, or to kill the salivary glands within the ∼36 h between fkh knockdown and the late-larval 20E pulse. Strong activation of hid and rpr and death only occur in response to the hormonal signal. After elimination of Fkh, the hormone induces expression of hid and rpr at a level that is sufficient to kill (Fig. 5). This observation supports the conclusion that there are no other repressors present in prepupal salivary glands that are sufficient to prevent hormonal induction of cell death. All that seems to be needed to induce death is the hormone 20E and one or more 20E-induced transcriptional activators.
Previous work has shown that E74A and BR-C play the role of hormone-induced activators of hid and rpr in late-prepupal salivary glands (Jiang et al., 2000). Both E74A and BR-C are required for the induction of hid, which has the characteristics of a secondary-response gene. Intriguingly, the activation of hid after premature loss of fkh shows the same secondary-response characteristics, suggesting that the same 20E-induced transcription factors are responsible for the activation of hid by the late-larval 20E pulse (Fig. 4). Full induction of rpr depends not only on BR-C but also on direct binding of the hormone receptor EcR/Usp to the gene (Jiang et al., 2000). Thus, rpr has characteristics of both a primary- and secondary-response gene, leading to an earlier induction of the gene in response to the prepupal 20E pulse. Again, premature activation in response to the late-larval pulse shows the same temporal characteristics (Fig. 4). This suggests that E74A and BR-C are responsible for the premature activation of hid and rpr after knockdown of fkh, and that such an activation is normally prevented by the presence of Fkh. Our immunostaining data support this conclusion by showing that Fkh protein is still present in the larval salivary glands at the time when the two genes are active. It only disappears from the tissue 2–4 h APF (Fig. 1 B). These data explain why E74A and BR-C mediate a death response exclusively to the prepupal 20E pulse, despite a very similar induction pattern of the two genes in response to the preceding late-larval pulse.
Our results exclude that repression of hid and rpr is mediated by the fkh target sens. Repression may thus be mediated by another downstream target of fkh or by direct binding of Fkh to transcriptional control regions of hid and rpr. In support of the latter possibility, we found that the first intron of hid contains a cluster of 13 Fkh binding sites. One of these sites exhibits strong binding of Fkh in in vitro DNA-binding assays, whereas the other sites have weak to moderate binding affinity (unpublished data; de Banzie, J., personal communication). Although this region may function as a silencer of hid expression in vivo, lacZ reporter gene assays in transgenic flies did not reveal that it has an enhancer function. We were not able to identify a similar binding site cluster in rpr.
Our microarray data identify other apoptosis-related genes that are down- or up-regulated by Fkh. Therefore, it is likely that Fkh protects cells from death by interfering with the cell death program at multiple levels. Regulation of genes such as Ark, Dronc, or PDK1, is likely to mediate a general function of fkh as a survival factor. This function appears to be required for the survival of the developing salivary glands during embryogenesis (Myat and Andrew, 2000). However, it is not essential for the survival of postembryonic salivary glands, as demonstrated by the failure of the glands to die in the absence of Fkh during prepupal development. Our data confirm this conclusion by showing that the salivary glands fail to undergo PCD within the ∼36 h between the premature knockdown of fkh and the steroid induction of death. They separate a general protective function of Fkh from a specific function that Fkh has in the control of steroid-induced developmental PCD.
Tissue-specific developmental cell death controlled by steroid hormone plays an important role not only in insects but also in humans and other vertebrates. Glucocorticoids, for instance, control the development of the immune system by killing specific types of thymocytes (Ashwell et al., 2000). Many genes regulated by glucocorticoids are coregulated by the vertebrate FOXA counterparts of Fkh (Friedman and Kaestner, 2006). It will be interesting to see whether FOXAs have evolutionarily conserved functions in glucocorticoid-induced death and in other types of developmental cell death.
Materials And Methods
Plasmid construction and P element transformation
For construction of the P[fkhRNAi] transformation plasmid, a segment of the coding region of fkh (corresponding to amino acid positions 189–435) was amplified by PCR from the plasmid fkh-pET3b. Two copies of the product were then sequentially cloned in a head–head orientation into the transformation vector pCaSpeR-hs-act (Thummel et al., 1988), leaving a 130-bp spacer between the copies. Five independent transformant lines were obtained, which all expressed fkh dsRNA upon heat shock; a detailed description of the construction steps can be obtained upon request). P element injections were performed by BestGene, Inc.
Developmental timing and Northern blot hybridizations
Third instar larvae were staged using the blue gut method as previously described (Andres and Thummel, 1994). For the −4-h time point (clear gut) in Fig. 1 A, only salivary glands that had completed glue protein secretion into the gland lumen were used. For the −4-h time point in Fig. 4, this additional criterion could not be used because glue production was severely affected in heat-shocked P[hs-fkhRNAi-4] animals. Prepupae and pupae were staged by collecting freshly formed prepupae within 30 min of puparium formation and keeping them on damp filter paper at 25°C for the indicated lengths of time.
For the Northern analysis of fkh expression in a BR-C mutant background, we used the y rbp5/Binsn stock (Belyaeva et al., 1980). Hemizygous mutant male larvae of this stock (y rbp5/Y) can be distinguished from males carrying the Binsn X balancer (Binsn/Y) by the yellow color of their mouth hooks and denticles (y phenotype; Binsn carries y+ allele). The males were separated based on this phenotype, and salivary glands were dissected for RNA extraction.
RNA extraction, fractionation by gel electrophoresis, transfer to nylon membranes, and hybridization with radioactive DNA probes were performed as previously described (Lehmann et al., 2002). Probes were derived by restriction digest from the following plasmids: fkh, 848-bp AflII–BglII fragment from RE03865; Sgs4, 1-kb EcoRI–HindIII fragment from pOW3Sal; sens, 1.4-kb EcoRI fragment from pBS-sens (provided by H. Bellen; Baylor College of Medicine, Houston, TX); BR-C, 480-bp StuI–PvuII fragment from paaDM527; and E93, 1.6-kb AccI fragment from E93 cDNA (provided by C. Thummel; University of Utah, Salt Lake City, UT). hid, rpr, E74A, and rp49 probes were prepared as previously described (Lehmann et al., 2002).
Ectopic expression and RNAi experiments
For ectopic expression of fkh, prepupae of the transformant line P[hs-Fkh111] and w1118 control prepupae were collected at 9.5 h APF and incubated for 30 min in a 37°C water bath. The animals were transferred to damp filter paper in a Petri dish and kept at 25°C until the salivary glands were dissected for RNA extraction or microscopic analysis. For RNAi knockdown of fkh, third instar larvae of P[hs-fkhRNAi-4] and w1118 were collected within 3 h of the second–third larval instar molt and transferred to fresh yeast paste. Larvae were kept at 25°C and subjected to heat shocks, as described in the previous section, at 12, 26, and 40 h after collection. The salivary glands of larvae expressing fkh dsRNA were defective in glue protein production and showed a reduced size compared with w1118 control glands (Fig. 5). Glue proteins make up >30% of the total gland protein in wandering third instar larvae (Korge, 1977), suggesting that the reduced organ size was caused by smaller cell size and not by reduced cell number. Staining of the salivary glands with the nuclear dye Hoechst 33342 confirmed this prediction. Impaired glue production was an expected result of the knockdown of fkh, as expression of all glue protein genes, including Sgs4 (Fig. 4), depends on fkh (Mach et al., 1996; Roth et al., 1999; our microarray data).
Histochemical staining and Western analysis
Acridine orange staining of salivary glands and midguts was performed as previously described (Jiang et al., 1997). Independent of the genotype of the animals, the salivary glands of both heat-shocked and non–heat-shocked wandering larvae had a speckled appearance after acridine orange staining (Fig. 5). Microscopic analysis of single nuclei at high magnification revealed that this was caused by staining of the nucleoli, which stain brightly with acridine orange (Nash and Plaut, 1964). The polytene chromosomes did not show staining above background (Fig. S3).
Staining of salivary glands with Fkh antibodies, FITC-labeled phalloidin (Alexis Biochemicals), and Hoechst 33342 (AnaSpec, Inc.), was performed using standard procedures. The anti-Fkh antibody was affinity purified and used at a dilution of 1:500 (Lehmann and Korge, 1996). Bound Fkh antibodies were detected with a Cy2-conjugated goat anti–guinea pig secondary antibody (1:400; Jackson ImmunoResearch).
For Western blot analysis (Fig. S2), Fkh antiserum was used at a dilution of 1:2,000, and bound antibody was detected using a peroxidase-conjugated goat anti–guinea pig secondary antibody (1:5,000; Jackson ImmunoResearch Laboratories).
Images of the fluorescently labeled tissues shown in Figs. 1, 2, and 5 were taken with an inverted confocal microscope (TCS SP2; Leica), overlaying a z series of 30–50 sections. Tissues were mounted in PBS, and images were acquired with a 10×/0.40 NA (Figs. 1 and 5) or 20×/0.70 NA (Fig. 2) HC PL APO objective (eyepiece lens 10×/22 HC PLAN; Leica). Bright-field images shown in Fig. S1 were taken with a 3–charge-coupled device digital imaging camera (KY-F75U; JVC) using a microscope (Axioskop 2 Plus; Carl Zeiss MicroImaging, Inc.) and Auto-Montage imaging software (Syncroscopy). The differential interference contrast and fluorescent images shown in Fig. S3 were taken with a camera (AxioCam MRm; Carl Zeiss MicroImaging, Inc.) using an Axioskop 2 Plus microscope and AxioVision 4.1 software. 2.5×/0.075 NA (Fig. S1) and 40×/0.75 NA (Fig. S3) Plan-Neofluar objectives (Carl Zeiss MicroImaging, Inc.) were used for image acquisition (10×/23 eyepiece lens). Images were processed using Photoshop 7.0 (Adobe), with uniform adjustments made to brightness and contrast, and were assembled using Illustrator CS2 (Adobe). All images were taken at room temperature.
P[hs-Fkh111] prepupae and w1118 control prepupae were heat shocked at 9.5 h APF for 30 min at 37°C and the salivary glands dissected 4 h later. Samples were prepared in three replicates from the P[hs-Fkh111] glands and in two replicates from the w1118 control glands. Total RNA was isolated using Trizol (Invitrogen) and purified on RNAeasy columns (QIAGEN). Hybridization to Affymetrix Drosophila Genome Arrays was performed by the microarray facility of the University of Maryland Biotechnology Center. Raw data provided by the Center were normalized, pooled, and compared using dChip (Li and Wong, 2001). Analysis was performed using the PM-only model with outlier detection. The datasets were filtered for genes that showed an at least 1.5-fold relative change in their mean expression and an absolute expression change of at least 400. Query for apoptosis-related genes was performed based on annotation using dChip and Microsoft Access.
Online supplemental material
Fig. S1 shows that misexpression of fkh does not affect PCD in the larval midgut. Fig. S2 shows a Western blot analysis of Fkh protein expression after knockdown of the gene by RNAi and after overexpression from P[hs-Fkh111]. Fig. S3 shows that acridine orange strongly stains the nucleoli of salivary gland nuclei of normal w1118 third instar larvae. Table S1 lists all apoptosis-related genes identified by microarray analysis that showed an at least 1.5-fold response to fkh.The online version of this article is available.
We thank Carl Thummel, Hugo Bellen, and the Bloomington Stock Center for providing fly stocks; Hugo Bellen for providing sens plasmids; Alvaro Godinez and Kirst King-Jones for help with the microarray data collection and analysis; members of the Spiegel and Henry laboratories for help with the use of shared equipment; John de Banzie for helpful discussions during the course of this work; and Carl Thummel for critical reading of the manuscript.
This work was supported by National Institutes of Health grant P20RR15569 from the Centers of Biomedical Research Excellence program of the National Center for Research Resources, and by an Arkansas Biosciences Institute grant ABI-511.
Abbreviations used in this paper: 20E, 20-hydroxyecdysone; APF, after puparium formation; ds, double stranded; hid, head involution defective; DIAP, D. melanogaster IAP; IAP, inhibitor of apoptosis protein; PCD, programmed cell death.