LIN-3/EGF Promotes the Programmed Cell Death of Specific Cells in by Transcriptional Activation of the Pro-apoptotic Gene

Summary

Programmed cell death (PCD) is an evolutionarily conserved cellular process that is important for metazoan development and homeostasis. The epidermal growth factor (EGF) promotes cell proliferation, differentiation and survival during animal development. Surprisingly, we found that the EGF-like ligand LIN-3 also promotes the death of specific cells in Caenorhabditis elegans. We found that the LIN-3/EGF signal can be secreted from a cell to facilitate the demise of cells at a distance by activating the transcription of the PCD-promoting gene egl-1 in the doomed cells through the transcription factor LIN-1. LIN-1 binds to the egl-1 promoter in vitro and is positively regulated by the LIN-3/EGF, LET-23/EGF receptor, and the downstream MAPK signaling pathway. To our knowledge, LIN-3/EGF is the first extrinsic signal that has been shown to regulate the intrinsic PCD machinery during C. elegans development. In addition, the transcription factor LIN-31, which binds to LIN-1 and acts downstream of LIN-3/EGF, LET-23/EGF receptor, and the MAPK signaling pathway during vulval development, is dispensable for PCD. Thus, LIN-3/EGF promotes cell proliferation, differentiation, and PCD through common downstream signaling molecules but acts via distinct sets of transcription factors for different target gene expression.

Introduction

PCD is important for proper animal development and tissue homeostasis [1], [2] and its dysregulation can cause aberrant death or survival of cells, which may lead to developmental defects, degenerative diseases, or cancers [1], [2].

Caenorhabditis elegans is an excellent model for studying PCD because of its invariant cell lineage and the conserved cell death pathway [3], [4]. Throughout the development of the C. elegans adult hermaphrodite, 131 somatic cells undergo PCD in an essentially invariant temporal and spatial pattern [5], [6]. Genetic and molecular studies have identified four genes, egl-1 (BH3-only gene), ced-9 (bcl-2), ced-4 (apaf-1), and ced-3 (caspase), that function in the core PCD pathway [7]–[12]. In living cells, CED-9 interacts with, and sequesters, CED-4 at the surface of mitochondria to prevent the cells undergoing PCD [13]. In cells destined to die, EGL-1 binds to CED-9, resulting in a conformational change in CED-9 and the release of bound CED-4 [14]. The released CED-4 translocates from the mitochondrion to the perinuclear membrane and interacts with, and activates, the caspase CED-3, leading to the eventual demise of the cell [15]. A recent study in mid-embryos and the germline suggested the existence of an alternative cell death activation mechanism that does not involve a direct interaction between CED-4 and CED-9 [16]. The transcriptional regulation of egl-1 is a critical step in the induction of most PCD events in the embryo [17]. Several transcription factors controlling egl-1 transcription have been identified and shown to specify the PCD fate of specific cells [4], [18]. For example, two transcription factors HLH-2 and HLH-3 activate egl-1 transcription by direct binding to the egl-1 cis-regulatory region during the specification of the death fate of NSM sister cells [18], [19]. Like HLH-2 and HLH-3, cell death specification genes have been shown to transcriptionally regulate the components of the core PCD machinery in a cell-autonomous manner. It is unclear whether the PCD fate, like many other cell fates, may be regulated by an extrinsic signal.

Extrinsic signals are crucial for a variety of developmental processes and act through receptors to elicit specific biological functions in a cell-nonautonomous manner. One example of such a signal-receptor pair is epidermal growth factor (EGF) and the EGF receptor (EGFR), which are involved in cell proliferation, differentiation, migration, survival, and death [20]–[22]. EGF has long been considered as a growth factor, since it stimulates proliferation in cultured cells, animals, and humans [23]. It also protects cells from apoptosis, as shown in cultured cells and Drosophila [24]–[29]. EGF can signal through the RAS-ERK-mediated and/or PI3K-mediated pathway(s) to activate transcription of various anti-apoptotic proteins, such as Bcl-X_L and Mcl-1 [25]–[27], and also regulates post-transcriptional modifications, such as phosphorylation of BAD and caspase-9, to prevent apoptosis [28], [29]. However, in contrast to this cytoprotective function, it has also been shown to promote apoptosis, as exogenous EGF induces apoptosis in several cell lines, such as A431, MDA-MB-468, and MCF-7 [30]–[32]. It is not clear how it exerts different functions in different cells under different conditions or whether the apoptosis-promoting function plays a physiological role during animal development.

In C. elegans, lin-3 and let-23 encode, respectively, the sole EGF-like ligand and the EGFR and control many aspects of development, including ovulation, vulval differentiation, cell specification, and behavioral quiescence [33]–[38]. Activated LET-23 can recruit SEM-5 (orthologous to human Grb2) from the cytosol to the plasma membrane to activate LET-60, a member of the GTP-binding RAS family [39], [40]. LET-60 then triggers a kinase cascade involving the sequential phosphorylation of LIN-45 (a Raf ortholog), MEK-2 (a MAPK kinase kinase), and MPK-1 (an ERK ortholog) [41]–[45]. Once MPK-1 is phosphorylated, it translocates to the nucleus and regulates the transcription of numerous target genes by phosphorylation of specific transcription factors [46]. Besides activating the LET-60-MPK-1 pathway, LET-23 activates the PLC-γ-mediated signaling pathway to regulate ovulation and behavioral quiescence [36], [38]. In addition, the SEM-5-binding protein SOC-1 is involved in both the PLC-γ- and PI3K-mediated signaling pathways [47]. The LET-60-MPK-1- and PI3K-mediated signaling pathways have been shown to be involved in physiological and genotoxic stress-induced germline cell death [48]–[51]. However, how these pathways are regulated by upstream signals, how they are linked to the intrinsic PCD machinery to control germline cell death, and whether they play a role in somatic PCD remain unknown.

We now report that LIN-3 acts as an extrinsic signal to promote specific somatic PCDs and the identification of components that transmit the LIN-3 signal to the intrinsic PCD machinery. Our data indicate that secreted LIN-3 induces cell death through the receptor LET-23, which acts in a cell-autonomous manner during PCD. Our genetic and biochemical data suggest that the LET-60-MPK-1 pathway transmits the LIN-3 signal from LET-23 to the transcription factor LIN-1 for transcriptional up-regulation of egl-1, which leads to activation of PCD.

Results

lin-3 and let-23 promote specific PCDs in C. elegans

While screening to test the potential involvement of receptor tyrosine kinase genes in PCD (see Materials and Methods), we found that several let-23 mutants have reduced numbers of cell corpses (Figure 1A). The let-23 gene encodes EGFR [34], raising the intriguing possibility that LET-23 might mediate an outside-in signal to affect PCD. lin-3 encodes the only EGF in C. elegans [33] and we found that lin-3 mutants had a similar reduction in numbers of cell corpses as let-23 single mutants or the let-23; lin-3 double mutant at the same embryonic stages (Figure 1A). This shows that lin-3 and let-23 function in the same pathway to affect cell corpse number, probably by acting as a ligand-receptor pair. Inactivation of lin-3 or let-23 by RNA interference (RNAi) reduced the numbers of cell corpses to a similar extent as the lin-3 or let-23 mutations (Figure 1A). Because the lin-3 or let-23 RNAi treatment caused more than 50% rod-like larval lethality, phenocopying the strong loss-of-function lin-3 or let-23 mutants [52], the reduced numbers of cell corpses observed in the lin-3 or let-23 mutants mentioned above likely represented the phenotype caused by nearly complete elimination of lin-3 or let-23, respectively. Interestingly, in these mutants, only a partial reduction of cell corpse numbers from the comma stage to the 2-fold stage was observed, suggesting that lin-3 and let-23 may be essential for some but not all cell deaths.

To investigate whether lin-3 may affect specific cell deaths, we examined the first 13 cell deaths that occur in the AB lineage during embryogenesis [6], [53] in the lin-3(e1417) mutant by using four-dimensional DIC microscopy. In three out of thirteen lin-3 embryos we analyzed, only ABplpppapp or its lineal homolog ABprpppapp failed to die, and the other cell deaths occurred normally (Figure 1B and S1). ABplpppapp and ABprpppapp (these cells are collectively termed ABpl/rpppapp hereafter), which are aunt cells of hypodermal cells hyp8/9, are generated at approximately 270 min after fertilization and undergo PCD soon after birth [6]. Consistent with the DIC four-dimensional microscopy data, approximately 20% of surviving ABpl/rpppapp were observed in the tail of the lin-3(e1417) mutant (Figure 1C and D), indicating that the disappearance of ABpl/rpppapp corpses is due to the survival of these cells. A similar result was observed in the let-23(sy1) mutant and in the lin-3(e1417) mutant feeding let-23 RNAi (Figure 1D), showing that lin-3 and let-23 act in the same pathway to promote the death of ABpl/rpppapp.

We also measured the duration time of the first 13 cell corpses derived from the AB lineage in the wild-type and lin-3(e1417) mutant embryos and found that the corpse duration time was statistically undistinguishable in the wild-type and lin-3 embryos (Figure S2, P = 0.24, two-tailed t test). This result indicates that the kinetics of corpse removal is normal in the lin-3(e1417) mutant, and rules out the possibility that the decrease of cell corpse numbers in the lin-3 mutants may be attributed to enhanced cell corpse removal.

We next examined whether lin-3 or let-23 may also be important for the death of other PCDs. We scored the extra surviving cells in the pharynx of the lin-3 or let-23 mutant at the L3 and L4 stages. In wild-type animals, 16 cells in the anterior pharynx and 6 in the posterior pharynx undergo PCD during mid-embryogenesis. Eight out of these 22 cells (NSML/R, I2L/R, g1AL/R and g2L/R) undergo PCD during the comma to 2-fold stage. In virtually all wild-type embryos, no extra surviving cells are observed in these regions at the late larval stages [6], [11], [54]. In lin-3 and let-23 mutants, about 10% of the sister cells of g1A gland cells and I1 interneurons survived, and the death of other 18 cells in the pharynx region was essentially normal (Figure 1E). Furthermore, the percentage of animals with surviving g1A and I1 sisters was enhanced by the weak ced-3(n2427) mutation; however, essentially no enhancement was observed in other 18 cells examined (Table S1). This result shows that lin-3 and let-23 are important for the death of specific cells. Consistently, using an integrated transgene tpIs8, which expresses monomeric red fluorescent protein (mRFP) in the g1A, g1P, and g2 gland cells [55], we found that 10% of the lin-3 animals had extra gland cells, whereas wild-type worms had no extra gland cell and the strong ced-3(n717) mutants all had extra gland cell(s) (Figure 1F and G). This result indicates that, like ced-3, surviving cells in the lin-3 mutant adopt the terminal differentiation fate of their sister cells. A similar result was observed in the wild-type or lin-3(e1417) mutant feeding let-23 RNAi (Figure 1G), showing that lin-3 and let-23 also act in the same pathway to promote the death of gland cells. Our data show that lin-3 and let-23 are important for the death of specific cells, including g1A and I1 sister cells and ABpl/rpppapp and that these genes are dispensable for most cell deaths analyzed in this work.

Overexpression of lin-3 cause ectopic cell deaths

We next examined whether over-activation of lin-3 signaling could cause ectopic cell deaths during embryogenesis. We found that overexpression of lin-3 genomic DNA from the integrated transgene syIs1 [33] increased embryonic cell corpse numbers (Table S2). Using the integrated transgene tpIs8 to label g1 and g2 gland cells, we found that syIs1 caused the loss of one or two gland cells in 19% of transgenic worms (Figure 2A and B). This phenotype of gland cell loss was suppressed by the loss-of-function mutation in egl-1 or ced-3 (Figure 2A and B). This result shows that lin-3 overexpression induces ectopic gland cell death rather than alters gland cell fate, and that the cell death-promoting activity of lin-3 requires the core PCD pathway. In addition, the phenotype of ectopic gland cell death induced by syIs1 was also suppressed by let-23 RNAi (Figure 2B), showing that lin-3 acts through let-23 to promote the ectopic cell death of the gland cells. Consistently, the increased numbers of embryonic cell corpses induced by syIs1 was also suppressed by loss-of-function mutations in let-23 or in components of the core PCD pathway (Table S2).

let-23 is expressed in dying cells, including ABpl/rpppapp

Next, we examined whether there was a correlation between let-23 expression and PCD using the transgene syEx234[P_let-23::let-23::gfp] [56]. As shown in Figure 3A, the LET-23::GFP fusion protein was detected as a crescent on the surface of some dying cells (indicated by arrowheads). The cause and physiological significance of this crescent-shaped localization pattern is not clear. To be sure that let-23 was expressed in dying cells and not on the membranes of adjacent living cells, we generated the construct P_let-23::4Xnls::gfp and microinjected it into the wild-type to generate transgenic worms. P_let-23::4Xnls::gfp contains the let-23 promoter P_let-23 fused to the GFP cDNA containing four copies of the nuclear localization signal (4XNLS). The GFP signal was detected in the nuclei of the dying cells (Figure 3B), clearly showing that let-23 is expressed in dying cells. Notably, the GFP signal was observed in the ABpl/rpppapp corpses (Figure 3C).

LIN-3 can act as an extrinsic signal to promote PCD

Like its Drosophila and vertebrate homologs, lin-3 encodes a transmembrane precursor protein, consisting of an extracellular domain containing one EGF motif, a transmembrane domain, and a cytoplasmic domain [33], [57]. Genetic and laser ablation studies have suggested that the LIN-3 precursor protein can be cleaved to form a secreted product [33], [58], [59]. Using the transcriptional reporter P_lin-3::gfp [60] and P_let-23::4Xnls::mrfp, no P_lin-3::gfp-expressing cell was located adjacent to the P_let-23::4xnls::rfp-expressing cell corpses, as shown in Figure 3D. In addition, no P_lin-3::gfp signal was detected near ABpl/rpppapp corpses (Figure 3E). These observations raise a possibility that lin-3 may be secreted and act at a distance to promote PCD.

To test this hypothesis, we generated a DNA construct P_elt-2::lin-3::gfp, in which a lin-3::gfp fusion cDNA was expressed under the control of the elt-2 promoter P_elt-2 and introduced it to the wild-type and lin-3 mutant. P_elt-2 is used here because it is exclusively expressed in intestinal cells and their precursor cells [61] and no PCD occurs in these cells [6]. We found that, in the lin-3 mutant, the P_elt-2::lin-3::gfp transgene not only increased the number of embryonic cell corpses to the wild-type level (Figure 3F) but also rescued the inappropriate survival of gland cells (Figure 3G). Moreover, when introduced to the wild-type, the P_elt-2::lin-3::gfp transgene increased the embryonic cell corpse number (Figure 3F) and caused ectopic gland cell deaths (Figure 3H), as the transgene syIs1 did (Figure 2A, B and Table S2). These results show that LIN-3::GFP secreted from the intestine can act at a distance to induce cell death in other parts of the embryo, e.g. gland cells in the head region. In our study, the adult worms transgenic for P_elt-2::lin-3::gfp had a multi-vulva phenotype, resembling worms carrying syIs1 [33], reinforcing the notion that LIN-3::GFP expressed in the intestine is secreted and active.

lin-3 signaling up-regulates egl-1 transcription

Because the cell death-promoting activity of lin-3 requires the core PCD pathway (Figure 2A, B, and Table S2), we next examined whether lin-3 signaling promotes PCD by regulating the expression of egl-1, ced-9, ced-4, or ced-3. To this end, we measured the transcript levels of egl-1, ced-9, ced-4, or ced-3 in worms with different levels of lin-3 activity by quantitative real-time reverse transcription (RT)-PCR. As shown in Figure 4A, the abundance of egl-1 transcripts correlated well with the level of lin-3 activity, which, in turn, correlated well with the number of cell deaths. egl-1 transcripts were less abundant in lin-3 and let-23 mutants than in wild-type animals (average of 0.53-fold and 0.57-fold, respectively), while lin-3 overexpression using the integrated transgene syIs1 caused an increase in egl-1 transcript levels (average of 2.15-fold). Similarly, levels of the ced-4 transcript, but not those of the ced-9 and ced-3 transcripts, also showed a correlation with the level of lin-3 activity, although to a lesser extent than the egl-1 transcript. These data show that LIN-3 and LET-23 signaling activates transcription of egl-1 and, to a lesser extent, ced-4, in vivo. Interestingly, we found that the lin-3(e1417) mutation reduced the egl-1 transcript level in the ced-3(n717) mutant (average of 0.61-fold) to a similar extent as the lin-3(e1417) mutation did in the wild-type (Figure 4A). This result supports the model that the LIN-3-mediated up-regulation of egl-1 transcription occurs before ced-3 activation during PCD.

Next, we examined the effect of lin-3 on the egl-1 transcription activity in ABprpppapp using the transcriptional fusion reporter P_egl-1::gfp [62]. The egl-1(n1084n3082) mutation caused frequent survival of ABpl/rpppapp, showing that egl-1 is required for the death of ABprpppapp (Figure 1C and D). We measured and compared the fluorescence intensity of GFP in ABprpppapp of the ced-3 and lin-3 ced-3 mutants. The ced-3(n717) mutation blocks the death of ABprpppapp, hence allowing the observation of the P_egl-1::gfp transgene in the surviving ABprpppapp, and has no effect on the lin-3-mediated egl-1 transcription (Figure 4A). In the ced-3(n717) mutant, ABprpppapp survived and expressed P_egl-1::gfp at the comma stage (Figure 4B). The lin-3(e1417) mutation significantly decreased the intensity of P_egl-1::gfp to about 0.53-fold in ABprpppapp of the ced-3 mutant (Figure 4B, P = 0.0004), consistent with the result from quantitative real-time RT-PCR (Figure 4A). Notably, the lin-3(e1417) mutation did not affect the egl-1 transcription in two adjacent cells that also express egl-1 (Figure 4B, control experiment, P = 0.6839). Therefore, loss of lin-3 reduces the egl-1 transcription level in a cell-specific manner. These data and the aforementioned observation that the egl-1(lf) mutation blocked the syIs1-induced cell death (Figure 2A, B, and Table S2) together support the notion that lin-3 signaling promotes the death of ABprpppapp by transcriptional activation of egl-1.

LET-23 transduces the cell death-promoting signal via the LET-60-MPK-1 pathway

We next investigated how lin-3 and let-23 signaling may up-regulate egl-1 to promote PCD. Activated LET-23 has been shown to act via multiple signaling pathways, including the LET-60-MPK-1- and PLC-γ-mediated pathways, to control C. elegans development [36], [38]–[45]. SOC-1, a SEM-5-binding protein, has been reported to act not only in the LET-60-MPK-1 pathway, but also in the PLC-γ- and PI3K-mediated signaling pathways [47]. We found that mutants defective in the LET-60-MPK-1 pathway had reduced cell corpse numbers during mid-embryogenesis (Table S3), while mutants defective in either the PI3K-mediated pathway (Table S4A) or PLC genes (Table S4B) had normal cell corpse numbers throughout embryogenesis. We further analyzed whether specific PCDs were affected in the let-60 and mpk-1 mutants using the assays described above. We found that let-60 and mpk-1 are required for the death of ABpl/rpppapp (Figure 5A), glA sisters, and I1 sisters (Figure 5B) but essentially dispensable for most cell deaths examined in this work. These data show that mutants defective in the LET-60-MPK-1 pathway have a PCD-defective phenotype similar to lin-3 and let-23 mutants. Furthermore, the lin-3 RNAi treatment did not enhance the survival of ABpl/rpppapp in the let-60 or mpk-1 mutant (Figure 5A), showing that, at least in ABpl/rpppapp, lin-3 and let-23 transduce the cell death-promoting signal through the LET-60-MPK-1 pathway.

LET-23 transduces the cell death-promoting signal via the transcription factor LIN-1, but not LIN-31

Two transcription factors, the ETS domain-containing protein LIN-1 and the WH domain-containing protein LIN-31, transduce LET-60-MPK-1 signaling to control vulval differentiation [63], [64]. In the absence of the LET-60-MPK-1 signaling pathway, LIN-1 forms a complex with LIN-31 and inhibits vulval induction [46]. Activation of the LET-60-MPK-1 pathway leads to phosphorylation of LIN-1 and LIN-31 by MPK-1, and phosphorylation of LIN-31 results in dissociation of LIN-1 and LIN-31 from each other and subsequent vulval induction [46]. Interestingly, the lin-1(lf) mutation also caused the survival of ABpl/rpppapp (Figure 5A), g1A sisters, and I1 sisters (Figure 5B), whereas the lin-31(lf) mutation did not (Figure 5A and B). These results show that lin-1, but not lin-31, affects specific PCD, despite both genes being necessary for vulval differentiation. lin-1 mutants also displayed a reduction of embryonic cell corpse numbers to a similar extent as mutants defective in the LET-60-MPK-1 pathway, whereas lin-31 mutants had a normal cell corpse profile during embryogenesis (Table S5). The lin-3 RNAi treatment did not enhance the penetrance of ABpl/rpppapp survival in the lin-1 mutant, showing that lin-3 and lin-1 act in the same pathway to promote the death of ABpl/rpppapp (Figure 5A). Consistently, the cell corpse numbers of the lin-1(sy254) lin-3(e1417) double mutant showed no significant difference when compared to the lin-1(sy254) or lin-3(e1417) single mutant (Table S5). These data and the aforementioned genetic results support the notion that LIN-3 and LET-23 act through the LET-60-MPK-1 pathway and transcription factor LIN-1 to promote PCD.

LIN-1 activates egl-1 transcription by direct binding to the egl-1 promoter

We next studied the mechanism by which LIN-1 promotes PCD. As mentioned above, LIN-3 and LET-23 signaling activates transcription of egl-1 and, to a lesser extent, ced-4, in vivo (Figure 4A). This was confirmed by examining the expression level of the transcriptional fusion reporter P_egl-1::gfp in ABprpppapp (Figure 4B). Moreover, like lin-3 overexpression, heat shock-induced lin-1 overexpression in the wild-type caused the phenotype of gland cell loss, and this phenotype can be suppressed by an egl-1(lf) mutation (Figure 5C). These results support the model that LIN-1 acts upstream of egl-1 to promote the death of gland cells.

LIN-1 contains a highly conserved ETS domain [64], which binds to DNA with the minimal recognition sequence GGAA/T [65]. A potential LIN-1 binding sequence was found in the promoter region of egl-1, but not that of ced-4 (Figure 6A, see Materials and Methods for details), supporting the possibility that LIN-1 binds directly to egl-1. We therefore generated the GST::LIN-1(DBD) fusion protein, in which the DNA binding domain of LIN-1 was fused to glutathione S-transferase (GST), and tested its ability to bind to a 34 bp DNA fragment (named P3) containing the LIN-1 binding sequence in the electrophoretic mobility shift assay (EMSA). The results are shown in Figure 6B. The P3 fragment displayed similar mobility in the presence or absence of GST (lanes 1 and 2), confirming that GST does not bind to the DNA fragment. In contrast, the presence of GST::LIN-1(DBD) resulted in the dose-dependent appearance of a slowly migrating band (lanes 3–5), and this effect was eliminated by addition of an excess of non-labeled wild-type DNA competitor (lanes 6 and 7), but not with an excess of a DNA competitor with a mutation in the core GGAA/T motif (lanes 8 and 9). These results show that the slowly migrating band is a complex of GST::LIN-1(DBD) and P3 and that LIN-1 binds to the P3 region of egl-1 in vitro.

To test the physiological importance of the interaction between LIN-1 and the egl-1 promoter, we mutated the P3 sequence in egl-1 genomic DNA and assessed the ability of the mutant egl-1 gene to rescue the cell death defect of the egl-1 mutant. Specifically, we examined the cell death phenotype by scoring the surviving ABpl/rpppapp in the L1 tail (Figure 6C) and by counting the number of cell corpses at the 1.5-fold stage (Figure 6D). In the control experiment, egl-1 mutants carrying the wild-type egl-1 transgene showed no surviving ABpl/rpppapp (Figure 6C), indicating that the wild-type egl-1 transgene Ex[egl-1(+)] fully rescued the egl-1 cell death defect. When the LIN-1 recognition sequence CAGGAA in the P3 region of the egl-1 promoter was mutated into GCTACT [65], the mutant egl-1 transgene Ex[egl-1(mut)] only partially rescued the egl-1 cell death defect. In the two transgenic lines, the percentages of ABpl/rpppapp survival were 15% and 25%, similar to that of the lin-3 mutant (Figure 6C). The egl-1; Ex[egl-1(mut)] transgenic worms also had a similar number of embryonic cell corpses as the lin-3 mutant at the 1.5-fold stage (Figure 6D). These results show that the binding of LIN-1 to the egl-1 promoter is important for the cell-killing activity of egl-1.

To test whether this LIN-1-mediated cell-killing activity of egl-1 may depend on LIN-3 signaling, we examined whether a lin-3(lf) mutation may reduce the rescuing activity of the mutant egl-1 transgene by scoring surviving ABpl/rpppapp (Figure 6C) and embryonic cell corpses (Figure 6D). As shown in Figure 6C and 6D, the lin-3(lf) mutation reduced the rescuing activity of the wild-type egl-1 transgene, but failed to reduce the rescuing activity of the mutant egl-1 transgene. These results show that binding of LIN-1 to the egl-1 promoter is critical for LIN-3 signaling to promote PCD. These data support the idea that LIN-3 acts through LIN-1 to promote PCD by transcriptional activation of egl-1.

Discussion

LIN-3 signaling promotes PCD through LET-60-MPK-1 signaling and LIN-1 to activate egl-1 transcription

LIN-3-induced LET-60-MPK-1 signaling is important in multiple aspects of development in C. elegans [39]–[45]. In this study, we provide evidence that it is also essential for specific somatic PCD. Mutants defective in genes coding for proteins in this pathway showed reduced PCD in specific cells, ABpl/rpppapp, g1A sisters, and I1 sisters. In contrast, over-activation of this pathway resulted in ectopic deaths of specific cells, such as gland cells. These results show that LIN-3-induced LET-60-MPK-1 signaling is crucial for the death-versus-life fate of specific cells. In addition, the downstream transcription factor LIN-1, but not LIN-31, was shown to act in the LIN-3-induced LET-60-MPK-1 signaling pathway to regulate PCD via the target gene egl-1. Furthermore, we showed that LIN-1 bound directly to the egl-1 promoter in vitro and that the LIN-1 binding site of the egl-1 promoter was important for the cell death-promoting function of LIN-3 in vivo.

Out of 35 embryonic PCDs we analyzed, including the first 13 PCDs in the AB lineage and 22 pharyngeal PCDs, the lin-3-to-lin-1 signaling pathway is important for the death of ABpl/rpppapp, g1A sisters, and I1 sisters. ABpl/rpppapp and I1 sisters undergo PCD before the comma stage, and g1A sisters die during the comma stage [6]. Our embryonic cell corpse data revealed that mutants defective in the lin-3-to-lin-1 signaling pathway had less cell corpses at not only the comma stage, but also 1.5-fold and 2-fold stages. However, the identities of dying cells affected at the latter two stages have not yet been determined, partly because no cell-specific markers are available to test for their survival in the mutants. It is possible that those lineages where PCDs might be more severely affected have not yet been analyzed in our work and, if so, the lin-3 signaling pathway might play a predominant role in these PCDs as compared to the death of ABpl/rpppapp, g1A sisters, and I1 sisters. Since the mutants defective in the lin-3 signaling pathway showed a partially penetrant phenotype in the survival of ABpl/rpppapp, g1A sisters, and I1 sisters, other genes or pathways, in addition to the lin-3 signaling pathway, are required to efficiently promote the death of, at least, ABpl/rpppapp, g1A sisters, and I1 sisters.

Although we have not yet characterized all affected embryonic PCDs, the data obtained using either the specific cell death assays or the embryonic cell corpse counting assays were similar. For example, the lin-3, let-23, let-60, mpk-1 and lin-1 single mutants had the same phenotype as their respective single mutants in the lin-3 or let-23 mutant background or feeding lin-3 or let-23 RNAi (Figure 1A, 1D, 1G, 5A, and Table S5). Overexpression of lin-3 using its endogenous promoter or misexpression of lin-3 in the intestine resulted in ectopic gland cells and increased numbers of embryonic cell corpses during comma to two-fold stages. Both phenotypes could be suppressed by loss of let-23 or components of the core PCD pathway (Figure 2 and Table S2). In addition, the wild-type egl-1 genomic DNA fully rescued the egl-1 cell death defect, but the mutant egl-1 genomic DNA with a mutation in the LIN-1 binding site could only rescue the egl-1 cell death defect to a similar extent as the lin-3 mutant (Figure 6C and 6D). Furthermore, loss of lin-3 did not aggravate the PCD defect of the egl-1; Ex[egl-1(mut)] transgenic animals (Figure 6C and 6D). Therefore, the lin-3-to-lin-1 signaling pathway may promote the death of ABpl/rpppapp and g1A sisters as well as other yet uncharacterized doomed cells by the same molecular mechanism. However, we cannot rule out the possibility that let-23, mpk-1, let-60 or lin-1 may affect other aspects of PCD, rather than promoting the PCD fate, in the yet unidentified doomed cells.

In the genetic interaction experiments, the let-23(n1045) mutation does not completely eliminate the cell death-promoting function of syIs1: the mutation reduced the numbers of ectopic embryonic cell deaths (Table S2) to the wild-type level, but not to the let-23(n1045) mutant level. The let-23(n1045) allele may still have a residual let-23 activity [52] and therefore cannot fully block the lin-3 signaling from syIs1. Alternatively, syIs1 may exert its PCD-promoting function through both LET-23-dependent and LET-23 independent pathways to cause ectopic cell deaths.

Previous studies showed that LIN-1 is negatively regulated by the upstream LIN-3-induced LET-60-MPK-1 signaling pathway during vulval induction [46]. We found that the LIN-1 function is positively regulated by this signaling pathway to promote the PCD fate of specific cells. During vulval development, although lin-1 inhibits most target gene expression downstream of the LET-60-MPK-1 pathway, lin-1 is also required positively for the expression of several downstream genes, such as egl-17 [66]. Our data indicated that lin-1 activates egl-1 expression in promoting PCD of specific cells. Interestingly, both of these lin-1-mediated gene expression processes are positively regulated by the LET-60-MPK-1 pathway and do not require lin-31. It appears that LIN-1 could have either a positive or negative function on gene expression, depending on the cellular context and target genes.

LIN-3 signaling was also found to activate ced-4 transcription, but egl-1 transcription probably accounted for most of the LIN-3 signaling-mediated cell death, as, in the egl-1 mutant, the mutant egl-1 transgene lacking the LIN-1 binding site could only restore the penetrance of ABpl/rpppapp death and the embryonic cell corpse number to the levels seen in the lin-3 mutant.

A previous study showed that dying cells that inappropriately survive generally adopt the fate of the sister or lineal homolog if the latter cells are terminally differentiated [11]. We observed that the g1A sister, when survived in the lin-3 or ced-3 mutant, expressed the gland cell marker P_B0280.7::4Xnls::mrfp, indicating that the g1A sister adopted its sister cell fate. To rule out the possibility that the lin-3 mutation may cause cell fate transformation and hence block cell death, we examined surviving ABpl/rpppapp in detail in the lin-3 mutant. ABpl/rpppapp was chosen because the sister of ABpl/rpppapp divides once and generates a hypodermal cell and a phasmid sheath cell [6]. It has been shown that dying cells that inappropriately survive do not adopt their sister cell fates, if the latter cells undergo cell division and the resulting descendent cells then differentiate [11]. The observation that surviving ABpl/rpppapp in the lin-3 or egl-1 mutant did not express the hypodermal cell-specific genes dpy-7 and ajm-1, as assayed by their respective reporters P_dpy-7::yfp and ajm::gfp, suggesting that surviving ABpl/rpppapp did not further divide or differentiate to generate a hypodermal cell. This result argues against the possibility that the lin-3 mutation affects ABpl/rpppapp survival indirectly through cell fate transformation to its sister or niece cell.

An appropriate level of extrinsic LIN-3 signaling is important for the precise fine-tuning of the life-versus-death fate of cells

Our data support the following model explaining how different levels of LIN-3 expression or activity might regulate PCD (Figure 6E). In the wild-type, after binding LIN-3, LET-23 transduces the cell death-promoting signal through the LET-60-MPK-1 pathway, leading to LIN-1 activation, followed by egl-1 transcription, and the demise of specific cells. However, mutations in either the LIN-3-induced LET-60-MPK-1 signaling pathway or LIN-1 or a defect in the LIN-1-binding site of the egl-1 promoter reduces egl-1 transcription and thus the probability that a cell undergoes PCD. In contrast, overexpression of lin-3 enhances egl-1 transcription and leads to ectopic cell deaths. Therefore, LIN-3 signaling may act to specify the death of specific cells and fine-tunes the life-versus-death fate of these cells by modulating transcription of egl-1.

Although EGF has long been known to function in the process of cell proliferation and survival, some reports showed that it also promotes apoptosis [30]–[32]. In some cases, the difference in cellular outcome appears to depend on its concentration, with a low EGF concentration promoting cell proliferation and a high concentration inducing apoptosis [67], [68]. Cells expressing different levels of EGFR also show different responses to the same amount of EGF, as it has been reported that exogenous EGF attenuates cell death in rats, which express low levels of EGFR, but promotes cell death in mice, which express high levels of EGFR [69]. Thus, it is likely that overactivation of EGF signaling triggers apoptosis. This may be a conserved physiological mechanism for preventing unrestricted proliferation when cells respond to persistent or hyperactivated EGF signaling. In C. elegans, cells may use this strategy to fine-tune PCD: cells that are doomed to die may express a high level of LET-23 or its downstream signaling components. This may either avoid inappropriate survival of cells that occasionally escape from the cell death machinery or preset a cell in a sensitized stage to respond quickly when it receives the death signal. This may also explain why some cells are more sensitive to apoptotic stimuli than others.

Our data show that different cells appear to show different susceptibility to lin-3 signaling in PCD at, at least, three different levels. First, ABpl/rpppapp, g1A sisters, and I1 sisters require lin-3 signaling for PCD and are sensitive to the wild-type level of lin-3 signaling. Second, g1 and g2 cells are sensitive to a high level of lin-3 signaling in PCD, since they survive in the wild-type but can be induced to undergo PCD by the syIs1 transgene. Third, cells, such as PLM, survive even in the presence of the transgene syIs1 and are therefore insensitive to lin-3 signaling for their life vs death fate. These results together show a complexity of extrinsic signaling in PCD regulation.

The LET-60-MPK-1 pathway promotes germline and developmental cell deaths

Consistent with our data, mutants defective in ksr-1, a positive regulator of the RAS signaling pathway, have reduced embryonic cell corpse numbers [70]. The LET-60-MPK-1- and PI3K-mediated signaling pathways are also known to be involved in physiological or genotoxic stress-induced germ cell death in C. elegans [48]–[51], although the ligand and receptor for these pathways have not yet been identified. The LET-60-MPK-1 pathway regulates the apoptotic process of irradiation-induced germ cell death by restricting CEP-1 protein expression to cells in late pachytene. More importantly, MPK-1, the C. elegans ERK, is activated following irradiation and is required for egl-1 transcription in irradiation-induced germ cell death [50]. This result and our present findings show that the LET-60-MPK-1 pathway regulates egl-1 transcription to promote both physiological somatic PCD and genotoxic stress-induced germline cell death. In mammalian cells, activation of the RAS-ERK pathway upregulates Bcl-X_L transcription [26]. It has also been shown that EGF regulates the expression of Mcl-1, a member of the anti-apoptotic Bcl-2 family, through activation of MAPK signaling and the direct binding of Elk-1, a member of the ETS gene family, to the Mcl-1 promoter [71]. Thus, the interaction of the RAS-ERK pathway and Bcl-2 family members, such as egl-1, Bcl-X_L, and Mcl-1_, may be a conserved strategy for regulating cell death or survival in evolution.

Pathological and physiological roles of EGFR in promoting cell death

It has been reported that cisplatin, a widely used chemotherapeutic agent for human cancers, causes nephrotoxicity in patients and induces extensive death of the proximal tubules in mice or cultured proximal tubule cells [72]–[74]. Cisplatin activates EGFR and ERK to trigger apoptosis in cultured proximal tubule cells [74]. In addition, in the autoimmune cutaneous disease pemphigus vulgaris, the body produces autoantibodies that can induce EGFR activation and ERK phosphorylation, leading to activation of caspase-3 [75]. EGFR activation-induced RAS signaling has also been implicated in age-related brain degeneration in Drosophila [76]. Taken together, these results and our data show that EGFR activation-induced cell death is a conversed phenomenon and plays an important role in pathological and physiological conditions.

Mutations that cause constitutive activation of EGFR are commonly observed in human cancers. Specific antibodies or inhibitors against activated EGFR have been developed and used as targeted therapeutics [77]. However, previous studies in human cell lines [30]–[32] and our present results show that overactivation of EGF signaling can promote apoptosis, raising the concern that drugs that antagonize EGFR signaling might have an anti-apoptotic effect and thus a potential tumor-promoting side-effect, depending on the cellular context. Further information on the molecular basis of the distinct divergent cellular responses of cell proliferation, cell survival, and death to EGF signaling is important for our understanding of the physiological roles of EGF and EGFR signaling during development and homeostasis and for providing perspectives for optimizing the strategy of anticancer drug design.

Materials and Methods

General methods and strains

Unless otherwise stated, C. elegans strains were maintained at 20°C as described previously [78]. The N2 Bristol strain was used as the wild-type strain. Alleles used:

Linkage group (LG)I: aap-1(m889), aap-1(ok282), age-2(yw23), daf-16(m26), and mek-2(ku114).

LGII: age-1(hx546), age-1(m333), let-23(n1045), let-23(sy1), let-23(sy97), lin-31(gk569), lin-31(n301), lin-31(n1053), plc-3(tm1340), and rrf-3(pk1426).

LGIII: ced-4(n1162), ced-9(n1950), mpk-1(ku1), mpk-1(n2521), unc-32(e189), unc-79(e1068), and syIs107[P_lin-3::pes-10::gfp].

LGIV: ced-3(n717), let-60(n1046), let-60(sy93), lin-1(n1761), lin-1(n303), lin-1(n1047), lin-1(sy254), lin-3(e1417), lin-3(n378), lin-45(n2506), and plc-4(ok1215).

LGV: egl-1(n1084n3082), plc-2(ok1761), sos-1(cs41), and bcIs37[P_egl-1::his-24::gfp].

LGX: plc-1(rx1), sem-5(n1779), and syIs1[lin-3(+)].

All alleles are described in WormBase (http://www.wormbase.org/). tpIs8 is an integrated version of tpEx270[P_B0280.7::4Xnls::mrfp; unc-119(+)] (this report). P_B0280.7::4Xnls::mrfp is expressed in g1 and g2 gland cells [55].

RNA interference (RNAi)

There are about 40 receptor tyrosin kinase (rtk) genes in the C. elegans genome. To test the involvement of these genes in promoting PCD, RNAi of the individual rtk genes was performed by feeding [79], and the cell corpses of treated embryos were scored. In this screen, only let-23 RNAi significantly decreased cell corpse numbers at the comma and 1.5-fold stages. To characterize the effect of the lin-3 signaling pathway on the number of embryonic cell corpses, the rrf-3(pk1426) mutant, which is sensitive to RNAi [80], was treated by feeding RNAi at 20°C, as previously described [81]. In other RNAi experiments, animals were treated with RNAi without the rrf-3(pk1426) mutation. The lin-3 or let-23 RNAi treatment caused more than 50% rod-like larval lethality both in the wild-type and rrf-3(pk1426) mutant, confirming that lin-3 and let-23 RNAi works well in these experiments. RNAi plasmids were obtained from the J. Ahringer RNAi library. L4 hermaphrodites were put onto the RNAi plates, and F1 were scored for the number of embryonic cell corpses, surviving ABpl/rpppapp, or extra gland cells. The empty vector L4440 was used as negative control.

Transgenic animals

Germline transformation experiments were performed as described previously [82]. To observe the expression pattern of let-23, P_let-23::4Xnls::gfp (50 ng/µl) was injected into unc-119(ed3) worms with the coinjection markers unc-119 [83] (50 ng/µl) and P_ced-1::1Xnls::mrfp (50 ng/µl). To examine the expression pattern of lin-3 and let-23, P_let-23::4Xnls::mrfp (200 ng/µl) was injected into an integrated transgene syIs107[P_lin-3::pes-10::gfp] with the coinjection markers P_elt-2::gfp (50 ng/µl) [61]. To test whether secreted LIN-3 rescued the cell death defect of the lin-3 mutant and caused ectopic cell deaths, P_elt-2::gfp or P_elt-2::lin-3::gfp (50 ng/µl) was injected into the lin-3(e1417) mutant and wild-type with the coinjection marker sur-5::gfp (50 ng/µl). To test the effect of LIN-1 overexpression on PCD, plasmids pPD49.78 and pPD49.83 (containing P_hsp without an insert) or P_hsp::lin-1 (50 ng/µl) was co-injected with the coinjection marker sur-5::gfp (50 ng/µl) into animals carrying the integrated transgene tpIs8. To examine the importance of the LIN-1 binding site in the egl-1 promoter, wild-type (pBC08) [7] or mutant egl-1 (2 ng/µl) was injected into either egl-1(n1084n3082) or lin-3(e1417); egl-1(n1084n3082) mutants with the coinjection marker sur-5::gfp (50 ng/µl).

Heat shock treatment

Embryos were incubated at 20°C or heat shocked at 33°C for 10–60 min and then moved to 20°C. The number of g1 and g2 gland cells was scored at the L4 stage using the tpIs8 transgene after three days.

Molecular biology

Standard methods of cloning, sequencing, and PCR amplification were used. To generate P_let-23::4xnls::gfp, the let-23 promoter was amplified by PCR using primers 5′-gtctagagcatctgcacttggg-3′ and 5′-gggatccgcctcccag-3′ and the resulting PCR fragment was inserted into the pPD122.56 vector to generate pYW1242. Full-length lin-3 cDNA was obtained by RT-PCR using worm total RNA and the primers 5′-CCACCGGTATGCGGAAAATGCTAC-3′ and 5′-CCACCGGTTTTGTGTGTCGAATCATTGG-3′. To obtain P_elt-2::lin-3::gfp, the elt-2 promoter, previously amplified by PCR using 5′-TTGGATCCCGGTGAAACTCTCTTGG-3′ and 5′-CCGGATCCCAGTGGCACCTAAAACATC-3′, was cloned into the BamHI site of pPD95.75 and lin-3 cDNA was cloned into the AgeI site to generate pYW1243. To generate P_hsp::lin-1, a full-length lin-1 cDNA was amplified from the yk1150h12 plasmid by PCR using 5′-ccggatccatgaatcacattgaccttttg-3′ and 5′-ttccatggctacaaagttggcatttttatg-3′, and the resulting PCR fragment was inserted into the heat-shock vectors pPD49.78 and pPD49.83 to generate pYW1245 and pYW1246, respectively. The DNA fragment corresponding to the LIN-1 DNA binding domain was amplified by PCR using the yk1150h12 plasmid as template and 5′-CCGGATCCATGAATCACATTGACCTTTTG-3′ and 5′-TTGGATCCTTTCGTTGGCGGCTGCGG-3′ as primers [65], then the PCR product was inserted into vector pGEX-4T-1 to generate plasmid pYW1247. The pBC08 plasmid contains the egl-1 genomic fragment, which rescues the cell death defect of egl-1 mutants [7]. To generate mutant egl-1 with a defect in the LIN-1 binding site, two partial fragments of egl-1 from pBC08 were first amplified by PCR using primer pairs 5′-caatttatagtaaagttaacttca-3′ plus 5′-cacatacagtagcttgaattcaactatacatctg-3′ and 5′-aattcaagctactgtatgtgtcacttccaccacc-3′ plus 5′-ggccgctctagaactagtgg-3′. The products were then used as templates for fusion PCR to generate a larger fragment to replace the corresponding region in pBC08 to generate pYW1248. This mutant egl-1 was identical to wild-type egl-1 except that the core LIN-1 binding motif was changed from CAGGAA to GCTACT.

Cell death assays

Cell corpses were counted at the indicated developmental stages and the number of extra cells in the anterior and posterior pharynx of L4 hermaphrodites was scored using Nomarski optics as described previously [54]. To analyze the first 13 cell deaths that occur in the AB lineage during embryogenesis, embryos were mounted onto 4% agar pads and recorded using a four-dimensional DIC microscopy analysis as described previously [84]. The survival of ABpl/rpppapp was examined by its localization in the tail with the aid of sur-5::gfp. To count the number of gland cells in the pharynx, the integrated transgene tpIs8[P_B0280.7::4Xnls::mrfp] was used as a marker for these gland cells [55], and the number of the mRFP-positive cells was scored at L4 stage by fluorescence microscopy.

Quantitative real-time reverse transcriptase (RT)-PCR

Total RNA was isolated from embryos developing up to the comma stage and purified by chloroform extraction and isopropanol precipitation, then 500 ng was reverse transcribed into cDNA using RevertAid H Minus First Strand cDNA Synthesis Kits (Fermentas). Equal amounts of the different cDNAs were used as the template for gene-specific PCR with appropriate primers. Water was used as negative control and tbg-1 RT-PCR was performed as an internal control. The CT values of the test genes were normalized to that of tbg-1 and relative expression levels were derived using the comparative CT method [85].

Quantification of P_egl-1::gfp intensity in ABprpppapp

To measure the P_egl-1::gfp expression within ABprpppapp, fluorescence micrographs were captured using a Zeiss AxioImager M2 microscope (Zeiss) equipped with a charge-coupled device camera. GFP intensity was determined using Image J by measuring the fluorescence signals within ABprpppapp (Ia) and its neighboring region (Ib) of ced-3(717) and lin-3(e1417) ced-3(n717) mutants. As a control, the fluorescence signals within two cells near ABprpppapp (Ia) and its neighboring region (Ib) of ced-3(717) and lin-3(e1417) ced-3(n717) mutants was determined. The value of Ia minus the value of Ib was used as the value for relative GFP intensity, and the relative GFP intensity of the ced-3(n717) mutant was used for normalization.

Prediction of potential LIN-1 binding sites

The prediction of potential LIN-1 binding sites was performed at http://www.cbrc.jp/research/db/TFSEARCH.html by submitting the sequence from +174 to +5820 (5′-3′) downstream of the stop codon of the egl-1 gene and the sequence from −1914 to −837 (5′-3′) upstream of the stop codon [62]. Four binding sites for Elk-1, the mammal LIN-1 ortholog, were predicted; these were AGCTTTCCGGCTCA (P1), CGATCCGGAAATCC (P2), TAGTTGAATTCAACAGGAAGTATGTGTCACTTCC (P3), and CAATTTCCGGTAAT (P4). P2 and P3, but not P1 or P4, are conserved in C. briggsae. P2 has been shown to be important for the PCD of the neurosecretory-motor (NSM) neuron [19], but lin-3 and let-23 do not seem to be involved in the death of NSM sister cell (Figure 1D), we therefore focused on P3 in this study.

Electrophoretic mobility shift assay (EMSA)

Expression of GST and GST::LIN-1(DBD) was induced in Escherichia coli strain BL21 carrying the respective plasmids pGEX-4T-1 and pYW1247 by addition of 0.4 mm isopropyl thiogalactoside and incubation for 1 h at 37°C. The cells were then harvested and disrupted by sonication in phosphate-buffered saline containing protease inhibitor cocktail (Calbiochem) and 1% Triton X-100, and the proteins were purified on glutathione Sepharose (Amersham Biosciences) according to the manufacturer's protocol. The amount of protein was measured by the Bradford method (Biorad). The sense and antisense oligonucleotides containing the predicted LIN-1 binding sequences, which were synthesized and biotinylated at the 3′ end (Genomics), were mixed in annealing buffer (5 mM NaCl, 1 mM Tris-HCl, 1 mM MgCl₂, and 0.1 mM dithiothreitol, pH 7.9) at a final concentration of 1 µM, denatured at 98°C for 2 min, and allowed to anneal by slowly cooling the mixture to 25°C. Unlabeled probes were used as competitors in competition experiments. The sequences of the probes were: wild-type: 5′-gaattcaacaggaagtatgtgtcacttccaccac-3′ and 5′-gtggtggaagtgacacatacttcctgttgaattc-3′; mutant: 5′- gaattcaagctactgtatgtgtcacttccaccac-3′ and 5′-gtggtggaagtgacacatacagtagcttgaattc-3′. EMSA were performed using LightShift Chemiluminescent EMSA kits (Pierce), following the manufacturer's instruction. The mixture was separated at 100 V at 4°C on a 5.5% non-denaturing polyacrylamide gel, then the DNA was transferred to a charged nylon membrane (Millipore) at 150 mA for 30 min and cross-linked with UV light using the auto crosslink option on a UV Stratalinker 1800 (Stratagene). The membrane was then blocked, washed, and exposed to X-ray film following the manufacturer's protocol.

Statistical analysis

In histograms, the values are shown as the mean and standard deviation (SD). Data derived from different genetic backgrounds at a specific developmental stage were compared using Student's two-tailed unpaired t-test. Data were considered statistically different at P<0.05. Fisher's exact test was used when comparing the percentages of surviving or missing cells in different genetic backgrounds. Data were considered statistically different at P<0.05.

Supporting Information

Attachment 1

Attachment 2

Attachment 3

Attachment 4

Attachment 5

Attachment 6

Attachment 7