Integrin α PAT-2/CDC-42 Signaling Is Required for Muscle-Mediated Clearance of Apoptotic Cells in

Summary

Clearance of apoptotic cells by engulfment plays an important role in the homeostasis and development of multicellular organisms. Despite the fact that the recognition of apoptotic cells by engulfment receptors is critical in inducing the engulfment process, the molecular mechanisms are still poorly understood. Here, we characterize a novel cell corpse engulfment pathway mediated by the integrin α subunit PAT-2 in Caenorhabditis elegans and show that it specifically functions in muscle-mediated engulfment during embryogenesis. Inactivation of pat-2 results in a defect in apoptotic cell internalization. The PAT-2 extracellular region binds to the surface of apoptotic cells in vivo, and the intracellular region may mediate signaling for engulfment. We identify essential roles of small GTPase CDC-42 and its activator UIG-1, a guanine-nucleotide exchange factor, in PAT-2–mediated cell corpse removal. PAT-2 and CDC-42 both function in muscle cells for apoptotic cell removal and are co-localized in growing muscle pseudopods around apoptotic cells. Our data suggest that PAT-2 functions through UIG-1 for CDC-42 activation, which in turn leads to cytoskeletal rearrangement and apoptotic cell internalization by muscle cells. Moreover, in contrast to PAT-2, the other integrin α subunit INA-1 and the engulfment receptor CED-1, which signal through the conserved signaling molecules CED-5 (DOCK180)/CED-12 (ELMO) or CED-6 (GULP) respectively, preferentially act in epithelial cells to mediate cell corpse removal during mid-embryogenesis. Our results show that different engulfing cells utilize distinct repertoires of receptors for engulfment at the whole organism level.

Introduction

During the development of multicellular organisms, cells that are unnecessary, damaged, or harmful undergo programmed cell death (apoptosis) [1]. Apoptotic cells are recognized and subsequently internalized by engulfing cells [2], [3]. Improper engulfment of apoptotic cells has been linked to diseases: too little engulfment may cause inflammation, autoimmune diseases, and cancers [4]–[6], whereas too much engulfment has been implicated in degenerative diseases [7]–[9].

In flies and mammals, engulfment of apoptotic cells is mediated by “professional” phagocytes, such as mobile macrophages and dendritic cells, or by “amateur” phagocytes, such as muscle cells, epithelial cells, and endothelial cells [10]–[14]. Several mammalian receptors involved in apoptotic cell engulfment have been identified and characterized. Receptors such as BAI1 [15], stabilin-1 [16], stabilin-2 [17], TIM-1 [18], TIM-3 [19], TIM-4 [18], integrins [20], [21], and receptor tyrosine kinase Mer [22] bind to the “eat me” signal, externalized phosphatidylserine (PS) [23], [24], on the surface of apoptotic cells either directly [25], [26] or through bridging molecules [27], [28]. BAI1, integrins, and Mer then signal through the conserved DOCK180/ELMO1/RAC GTPase signaling module to promote the internalization of apoptotic cells [15], [29]–[31], whereas stabilin-1 and stabilin-2 do so through the intracellular adaptor GULP [16]. Other “eat me” signal and receptor pairs for engulfment have been reported. For example, lectin receptors bind to altered sugars on apoptotic cells [32], scavenger receptors to oxidized LDL-like moieties [33], and CD14 to ICAM3 [34]. The in vivo role of most of these receptors in the clearance of apoptotic cells and the tissues in which they act at the whole organism level have not been defined.

During the development of a C. elegans adult hermaphrodite, 1090 somatic cells are generated, 131 of which undergo apoptosis [35]–[37]. The apoptotic cells are removed by their neighboring cells [35], [38]. Cell types such as hypodermal cells (which constitute the external epithelium), pharyngeal muscle cells, and intestinal cells have been shown to function as engulfing cells [37], [38]. Three partially redundant pathways that regulate the engulfment process have been identified. The first pathway is mediated by two cell-surface proteins CED-1 (mammalian homologue MEGF10) and CED-7 (ABCA1) [39], [40]. CED-1 binds to an apoptotic cell through secreted molecule TTR-52 (transthyretin) and transduces the engulfment signal through the adaptor protein CED-6 (GULP) and DYN-1 (dynamin) to promote the engulfment and degradation of apoptotic cells [41]–[43]. The second pathway is regulated by at least three engulfment receptors, phosphatidylserine receptor PSR-1 [44], Frizzled MOM-5 [45], and integrin INA-1/PAT-3 [46], all of which signal through the adaptor protein CED-2 (CRKII) and the bipartite GEF complex CED-5 (DOCK180)/CED-12 (ELMO) for CED-10 (RAC1) GTPase activation [47]–[52]. Phosphoinositide phosphatase MTM-1 (myotubularin) negatively regulates this pathway by inhibiting the recruitment of CED-12 to the plasma membrane [53], [54]. These two engulfment pathways may converge at CED-10 GTPase, which promotes the actin-based cytoskeleton rearrangement required for phagocytosis of apoptotic cells in engulfing cells [55]. CED-10 activity is negatively regulated by GTPase activating protein SRGP-1 during the engulfment process [56]. Compared to these two major pathways, little is known about the third pathway, which is negatively regulated by the cytoskeletal regulator ABL-1 (Abl), which inhibits the engulfment of apoptotic cells by inhibiting ABI-1 (Abl-interacting protein) and acts independently of CED-10 [57].

Integrins are transmembrane αβ heterodimers that make connections to the extracellular matrix and cytoskeleton and activate several signaling pathways required for multiple cellular processes, including cell adhesion, cell migration, and cell survival [58], [59]. C. elegans has two integrin α subunits, INA-1 and PAT-2, and a single β subunit, PAT-3 [60]–[62]. Integrin PAT-2/PAT-3 is a component of muscle attachment complexes and is essential for sarcomere assembly [63], [65] and also acts to direct muscle arm extension [66] and distal tip cell migration [67]. We have recently shown that integrin INA-1/PAT-3 functions as an engulfment receptor for apoptotic cells [46]. Intriguingly, the pat-3 knockout mutant has a stronger defect in cell corpse engulfment than the ina-1 mutant [46], raising the possibility that pat-2 may also mediate the removal of apoptotic cells. In this study, we examined and characterized the role of pat-2 in cell corpse engulfment and showed that it functions in the muscle-mediated internalization of apoptotic cells and acts through a pathway distinct from the previously known pathways.

Results

pat-2 loss-of-function results in an increased number of embryonic cell corpses

pat-2(st567) mutants [64] and worms treated with pat-2 RNAi are embryonic lethal and show a phenotype of paralyzed arrest at the two-fold stage (Pat), as PAT-2 plays an essential role in body wall muscle assembly and function during embryogenesis [63]–[65]. We tested the involvement of pat-2 in apoptosis by counting the number of apoptotic cells at the comma and 1.5-fold stages, the two stages at which the majority of embryonic apoptosis occurs [37] and pat-2 mutant embryos are still developing normally, and found that both pat-2(st567) and pat-2(RNAi) embryos had a Ced (cell death abnormal) phenotype with increased numbers of apoptotic cells (Table 1). The Ced phenotype of the pat-2(st567) mutant was rescued by the transgene P_pat-2pat-2::gfp, in which the pat-2::gfp translational fusion construct is expressed under the control of the endogenous pat-2 promoter P_pat-2 (Table 2), confirming that the Ced phenotype of the pat-2(st567) mutant was specifically caused by pat-2 loss of function. The P_pat-2pat-2::gfp transgene also rescued the Pat phenotype of the pat-2(st567) mutant (Table 3).

ced-3(n717) and ced-4(n1162), strong mutations in the pro-apoptotic genes ced-3 and ced-4 that block almost all programmed cell death [68], suppressed the phenotype of an increased number of cell corpses in pat-2(st567) or pat-2(RNAi) embryos (Table 1), showing that the extra cell corpses observed in the pat-2 mutants were generated by programmed cell death. In contrast, the Pat phenotype of the pat-2(st567) or pat-2(RNAi) mutants was not suppressed by either the ced-3 or ced-4 mutation (Table 1). The fact that the Pat and Ced phenotypes can be uncoupled shows they are probably due to the loss of different pat-2 functions.

pat-2 is required for the removal of embryonic cell corpses

To determine the cause of the Ced phenotype of the pat-2 mutant, we performed a time-lapse differential interference contrast (DIC) microscopy analysis of cell corpses in the wild-type and pat-2(st567) mutant during embryogenesis prior to the 2-fold stage. We found that, although the timing and number of cell death events were similar in the two types of embryo (Figure 1A), the length of time that the cell corpses persisted was significantly different (Figure 1B). In the wild-type, approximately 97% of cell corpses disappeared within 40 minutes and no cell corpses persisted longer than 60 minutes, whereas, in the pat-2 embryo, more than 50% of the cell corpses persisted longer than 40 minutes and about 25% longer than 100 minutes. These results demonstrate that pat-2 functions in the clearance of cell corpses.

pat-2 does not function in any previously known pathway for apoptotic cell removal

In addition to PAT-2, C. elegans has another integrin α subunit, INA-1, which forms a complex with the single integrin β subunit, PAT-3, on the cell surface [62] and is also required for the clearance of embryonic cell corpses [46]. RNAi inactivation of either ina-1 or pat-3 results in increased numbers of apoptotic cells at the comma and 1.5-fold stages [46] (Table 1). The pat-2(st567) mutation further increased the number of cell corpses in the ina-1(RNAi) mutant, but not in the pat-3 mutant (Table 1). These results support the notions that PAT-2, like INA-1, acts together with PAT-3 during cell corpse clearance and that the two integrins, PAT-2/PAT-3 and INA-1/PAT-3, function in a partially redundant fashion during the clearance process.

We next determined whether pat-2 functioned together with previously identified genes to promote cell corpse removal. The two major pathways that regulate cell corpse removal are mediated, respectively, by ced-1, ced-6, and ced-7 or ced-2, ced-5, and ced-12 [2]. We therefore generated and analyzed double mutants containing either the pat-2(st567) or pat-2(RNAi) mutation and strong loss-of-function or null alleles of the engulfment ced genes for the two pathways. Interestingly, the pat-2(st567) or pat-2(RNAi) mutation further enhanced the engulfment defect in mutants defective in either pathway (Table 4). This suggests two possibilities. First, pat-2 could act in both pathways, with the combination of a pat-2 mutation and a defect in either pathway having an additive effect. Alternatively, pat-2 could act in a separate pathway that is partially redundant with these two pathways. To distinguish between these two possibilities, we tested whether the pat-2(st567) or pat-2(RNAi) mutation could increase the engulfment defect in double mutants between the two pathways, such as the ced-1(e1735); ced-5(n1812) and ced-12(tp2); ced-7(n1892) double mutants. In the first case in which pat-2 would act in both pathways, we would expect that pat-2(st567) or pat-2(RNAi) would not enhance the engulfment defect of the double mutants. However, we found that the pat-2(st567) or pat-2(RNAi) mutation significantly increased the number of cell corpses in both double mutants (Table 4). We therefore conclude that pat-2 probably functions in a pathway distinct from these two pathways to promote the engulfment of cell corpses.

A recent study has shown that abi-1 (abl-1 interactor 1) acts in parallel to the two major engulfment pathways during cell corpse removal [57]. Our analysis showed that pat-2(RNAi) or pat-3(st564) enhanced the engulfment defect in the abi-1(null) or abi-1(RNAi) embryo (Table 4), showing that pat-2 and pat-3 function independently of abi-1 to mediate the removal of apoptotic cells.

pat-2 functions in muscle cells to mediate the removal of apoptotic cells

We next examined the localization of PAT-2 using the P_pat-2pat-2::gfp or P_pat-2pat-2::mcherry transgene, which rescued the Ced and Pat phenotypes of the pat-2 mutant (Table 3 and data not shown). PAT-2::GFP or PAT-2::mCherry was found to be expressed in body wall muscles and hypodermal cells during embryogenesis (Figure 2A, 2B and Figure S2A). Notably, a strong GFP signal was observed along the surface of apoptotic cells adjacent to muscle cells (Figure 2A and 2E), whereas a relatively weak GFP signal was seen around apoptotic cells engulfed by hypodermal cells (Figure 2B and 2F). Our analysis of embryos expressing the transgene P_pat-2nls::gfp, in which GFP was expressed under the control of P_pat-2 and was predominantly localized to the nucleus, indicated that P_pat-2 expression was absent from cell corpses during embryogenesis (Figure 2I, 2K, Figure S1 and Text S1). Thus, the PAT-2::GFP signal surrounding apoptotic cells likely originated from the surface of the engulfing cells. The localization of PAT-2::GFP around the surface of apoptotic cells near body wall muscles was not affected by the ced-1;ced-5 double mutation (Figure S3), consistent with our genetic data showing that pat-2 acts in parallel to ced-1 and ced-5 to mediate cell corpse engulfment. When PAT-2::mCherry and PAT-3::GFP were co-expressed under the control of their respective endogenous promoters, PAT-2::mCherry and PAT-3::GFP were co-localized on apoptotic cells and to the dense bodies (Z-disks) and M-lines of muscle cells (Figure S2A), in agreement with the idea that PAT-2 and PAT-3 form a complex.

In contrast to flies and mammals, C. elegans does not have macrophage-like motile phagocytes; instead, apoptotic cells are removed by their neighboring cells [36], [38]. Hypodermal cells, pharyngeal muscle cells, and intestinal cells have been shown to function as engulfing cells [37], [38]. Since pat-2 was expressed in both hypodermal cells and body wall muscles, we tested whether it functioned in muscle cells and/or hypodermal cells for the removal of apoptotic cells. To this end, we expressed pat-2 cDNA under the control of the P_ajm-1 or P_unc-54 promoter and examined the ability of each transgene to rescue the Ced phenotype of the pat-2 mutant. P_ajm-1 is expressed in all epithelia including hypodermal cells [69], whereas P_unc-54 is expressed in body wall muscles [70]. We found that expression of pat-2 by P_unc-54, but not by P_ajm-1, fully rescued the Ced phenotype of the pat-2(st567) embryo (Table 2). P_unc-54 did not appear to express in apoptotic cells (Figure S4A). These results support the notion that pat-2 acts in muscle cells to mediate cell corpse removal during embryogenesis.

We next co-expressed PAT-2::mCherry and PAT-2::GFP by the P_pat-2 and P_unc-54 promoters, respectively, to monitor the PAT-2-mediated and muscle-mediated engulfment processes simultaneously in the 1.5-fold embryos. Approximately 22.7% and 18.1% of the cell corpses were enclosed by the PAT-2::mCherry and PAT-2::GFP circles, respectively, and most of the mCherry and GFP circles were co-localized, suggesting that most, if not all, PAT-2-mediated engulfment involves muscle cells (Table S1).

CED-1 and INA-1 may function in the hypodermal cell-mediated engulfment of apoptotic cells

CED-1 and INA-1/PAT-3 are engulfment receptors that are expressed in multiple cell types, including hypodermis and body wall muscles [40], [46], but act in pathways in parallel to that involving PAT-2 (Table 1 and Table 4). We therefore tested whether CED-1 and INA-1 acted in specific cell types to mediate the engulfment of apoptotic cells. Expression of ced-1 in hypodermal cells using the P_ajm-1 promoter fully rescued the ced-1 engulfment defect, whereas expression of ced-1 in body wall muscles using the P_unc-54 promoter did not (Table 2). Similarly, expression of ina-1 in hypodermal cells, but not muscle cells, rescued the ina-1 engulfment defect (Table 2). These data suggest that CED-1 and INA-1 preferentially act in hypodermal cells, at least during the comma and 1.5-fold stages. This observation, together with our result that PAT-2 predominantly functions in muscle cells, indicates that different engulfing cells may utilize different engulfment receptors to mediate cell corpse removal.

Removal of apoptotic MSpppaaa cells by muscle cells requires PAT-2

The mesodermal MSpppaaa cell is generated in the head region about 250 minutes after the first cleavage of a zygote [37]. After circumferential migration to the dorsal midline, the MSpppaaa cell is located near the anterior dorsal muscle cells. Approximately 400 minutes after its generation, the MSpppaaa cell undergoes apoptosis at the four-fold stage [37]. To examine whether the apoptotic MSpppaaa cell was removed by a muscle cell, we expressed membrane-bound monomeric red fluorescent protein (mRFP) in muscle cells using P_unc-54. MSpppaaa cell corpses were found inside the adjacent muscle cells (Figure 2J and 2K), showing that MSpppaaa cell corpses were engulfed by muscle cells. To confirm this result, we tagged PAT-2 with mCherry and GFP and co-expressed PAT-2::mCherry in body wall muscles using P_unc-54 and PAT-2::GFP in hypodermal cells using P_ajm-1 in wild-type embryos. The PAT-2::mCherry, but not PAT-2::GFP, signal was observed around apoptotic MSpppaaa cells (Figure S5), confirming that apoptotic MSpppaaa cells are engulfed by muscle cells, but not hypodermal cells. The basement membrane between muscle and hypodermis may limit the access of hypodermal cells to the apoptotic MSpppaaa cell for engulfment.

We next examined whether the clearance of apoptotic MSpppaaa cells is defective in pat-2 embryos by scoring apoptotic MSpppaaa cells in wild-type and pat-2 embryos at specific embryonic stages. The two stages during which the pharyngeal grinder has just formed and the pharynx is pumping were chosen. The MSpppaaa cell undergoes apoptosis during pharyngeal grinder formation. Pharyngeal pumping begins about 1.5 hours after grinder formation is complete. At the time when pharyngeal grinder formation had just finished, only 19.6% of wild-type embryos contained the MSpppaaa cell corpse and none remained at the time when pharyngeal pumping occurred (Table 5). However, 74.6% of pat-2 embryos contained the MSpppaaa cell corpse at the time when pharyngeal grinder formation had just finished, and 52.4% still contained a corpse when pharyngeal pumping had started (Table 5). This shows that pat-2 is important for the removal of apoptotic MSpppaaa cells. In addition, the muscle-specific expression of pat-2 by P_unc-54 significantly reduced the percentage of pat-2 mutant embryos containing the MSpppaaa cell corpse at the time when pharyngeal grinder formation had just finished or pharyngeal pumping had started (Table 5). In contrast, the hypodermal cell-specific expression of pat-2 by P_ajm-1 failed to do so (Table 5). These results show that pat-2 functions in muscle cells to mediate the removal of the MSpppaaa cell corpse.

Three cells C1, C2 and C3, called as in reference [42], are generated in the ventral side of embryos at approximately 300–350 min after first cleavage and subsequently undergo apoptosis. Their cell corpses are engulfed by the ventral hypodermal cells ABplaapppp, ABpraapppa and ABplaapppa, respectively [42]. The duration time of these apoptotic cells appeared normal in the pat-2 mutants (Table S2). This result further supports the notion that pat-2 preferentially acts in muscle cells but not hypodermal cells for cell-corpse removal. In contrast to pat-2, ced-1 has been previously shown to be important for the hypodermal cell-mediated removal of C1, C2 and C3 cell corpses [42]. In the ced-1 mutants, the duration time of MSpppaaa cell corpses was slightly longer than that of the wild-types: approximately 19% of MSpppaaa cell corpses still persisted in the ced-1 mutants during pharyngeal pumping, whereas none remained in the wild-type embryos at this stage (Table 5). Thus, ced-1 also plays a minor role in the muscle-mediated removal of apoptotic MSpppaaa cells in late embryogenesis.

Muscle cell-mediated internalization of apoptotic cells is defective in pat-2 mutants

The persistence of cell corpses in the pat-2 mutant could be due to a defect in either the internalization or the degradation of the corpse. We used the P_unc-54myri::mrfp translational reporter to express the MYRI::mRFP fusion protein on the surface of muscle cells and followed the membrane processes of a muscle cell around an apoptotic cell using the time-lapse fluorescence microscopy analysis. The comma stage was chosen because the pat-2 embryos at this stage show an increased number of cell corpses and the embryos at this stage do not move around. In wild-type embryos, MYRI::mRFP fusion protein appeared to localize to the growing pseudopods, which eventually formed a circle around an apoptotic cell upon the completion of the internalization process (Figure 3A). The MYRI::mRFP circle formed around an apoptotic cell in approximately 6 minutes. However, in the pat-2(st567) mutant embryos, the MYRI::mRFP circle formation took approximately 21 minutes to complete, more than three times longer than that in the wild-type, indicating that the internalization process was compromised.

We further examined the internalization of MSpppaaa cell corpses in the pat-2 mutant using the P_unc-54myri::mrfp transgene. As shown in Table 5 and Figure 3B, in wild-type embryos, all apoptotic MSpppaaa cells showed the MYRI::mRFP circle at the time when grinder formation had just finished and their cell corpses were cleared at the time of pharyngeal pumping. However, in pat-2 mutants, no apoptotic MSpppaaa cells had the MYRI::mRFP circle at the time when grinder formation had just finished, and nearly half of MSpppaaa cell corpses still persisted and showed no MYRI::mRFP circle at the time of pharyngeal pumping (Table 5). These data indicate that pat-2 is required for the internalization of an apoptotic MSpppaaa cell by a muscle cell.

The extracellular domain of PAT-2 recognizes and binds to apoptotic cells in vivo

We then examined whether PAT-2 recognized apoptotic cells to trigger their internalization. To this end, we generated the transgene P_hsppat-2(ex)::mcherry, in which the coding sequence for the PAT-2 extracellular domain with a signal sequence [PAT-2(ex)] was fused to that of mCherry under the control of the heat-shock promoter P_hsp. The transgene was then introduced into the wild-type and ced-1(e1735); ced-5(n1812) double mutant embryos. The ced-1(e1735); ced-5(n1812) double mutant embryos contain many persistent apoptotic cells, resulting in a greater chance of seeing PAT-2(ex)::mCherry binding, especially during late embryogenesis when very few cell corpses are present in the wild-type embryos. We found that secreted PAT-2(ex)::mCherry clustered on the surface of apoptotic cells (Figure 2M), albeit with a weaker fluorescence intensity compared to that of PAT-2::GFP (Figure 2A). Approximately 15.2% of apoptotic cells displayed the PAT-2(ex)::mCherry circle in the ced-1(e1735); ced-5(n1812) double mutant embryos at the 4-fold stage. A similar percentage (16.6%) was observed in the wild-type embryos at the 1.5-fold stage (Table S1). Thus, PAT-2(ex) recognizes and binds to the surface of apoptotic cells. However, we do not know if binding is direct or indirect.

We further examined whether PAT-2(ex)::mCherry bound to specific apoptotic cells. Approximately 20% of the MSpppaaa cell corpses had a PAT-2(ex)::mCherry circle (Table S1 and Figure S6A), but none of C1, C2 and C3 cell corpses, which are engulfed by hypodermal cells, had a PAT-2(ex)::mCherry circle (Table S1 and Figure S6C and S6D). This result indicates that PAT-2(ex) binds to specific apoptotic cells and may explain why PAT-2 is required for the removal of MSpppaaa (Table 5), but not C1, C2 or C3, cell corpses (Table S2) and why only a subset, but not all, of apoptotic cells are labeled with PAT-2(ex)::mCherry (Figure 2L and 2M). Furthermore, we used the P_unc-54 promoter to express PAT-2(ex)::mCherry and observed PAT-2(ex)::mCherry clustering around apoptotic cells including the apoptotic MSpppaaa cell (Figure S4B), despite the fact that the PAT-2(ex)::mCherry signal was not as strong as that expressed by the P_hsp promoter. This result shows that PAT-2(ex)::mCherry expressed in and secreted from muscle cells recognizes, and binds to, the surface of apoptotic cells. Embryos with PAT-2(ex)::mCherry overexpression by either P_hsp or P_unc-54 resulted in increased numbers of apoptotic cells at the comma and 1.5-fold stages (Table 3) and delayed the removal of MSpppaaa cell corpses (Figure S6A), indicating that PAT-2(ex)::mCherry overexpression interferes with the normal process of cell corpse engulfment.

The INA-1 extracellular domain, termed INA-1(N) as in reference [46], has been shown to recognize apoptotic cells [46]. To test whether PAT-2 and INA-1 may recognize the same apoptotic cells, we co-expressed PAT-2(ex)::mCherry and INA-1(N)::GFP in embryos using the transgenes P_hsppat-2(ex)::mcherry and P_hsp ina-1(N)::gfp. We found that PAT-2(ex)::mCherry was co-localized with INA-1(N)::GFP on some apoptotic cells (Figure S6E). Thus, the extracellular domains of PAT-2 and INA-1 can recognize identical apoptotic cells, although they preferentially function in different cell types for apoptotic cell removal.

The cytoplasmic domain of PAT-2 is required for cell-corpse engulfment, but not for muscle assembly or contraction

As shown above, PAT-2 binds to apoptotic cells and may serve simply to tether the corpse to an engulfing cell or also initiate a signaling pathway for engulfment. To distinguish between these two possibilities, we tested whether the cytoplasmic domain of PAT-2 was essential for signaling by deleting it and examining the ability of truncated PAT-2Δcyto to rescue pat-2(st567) mutants. We generated pat-2(st567); P_pat-2pat-2Δcyto::gfp transgenic animals that expressed the PAT-2Δcyto::GFP fusion protein under the control of the pat-2 promoter. To our surprise, PAT-2Δcyto::GFP fully rescued the Pat phenotype of the pat-2(st567) mutant, but failed to rescue the Ced phenotype (Table 3). The pat-2(st567); P_pat-2pat-2Δcyto::gfp transgenic embryos contained increased numbers of cell corpses (Table 3) at comma and 1.5-fold stages, and some of the cell corpses persisted longer than those of the wild-type embryos (Figure 1B), showing an engulfment defect. The non-Pat phenotype of the transgenic embryos allowed us to count cell corpses at and beyond the 2-fold stage, which is impossible for homozygous pat-2(st567) embryos because of developmental arrest. The pat-2(st567) embryos carrying the P_pat-2 pat-2Δcyto::gfp transgene also had increased numbers of cell corpses at the 2-, 3- and 4-fold stages (Table S3). Thus, pat-2 is required for the engulfment of apoptotic cells throughout embryogenesis. In addition, MSpppaaa cell corpses also persisted longer in the pat-2(st567); P_pat-2 pat-2Δcyto::gfp transgenic embryos than those in wild-type embryos. Approximately 61% of the MSpppaaa cell corpses still persisted at the stage of pharyngeal pumping, while none remained in the wild-type embryos at this stage (Table 5).

PAT-2Δcyto::GFP was localized to the surfaces, dense bodies and M-lines of muscle cells in the wild-type and pat-2 mutants (Figure S7), similar to PAT-2::GFP or PAT-2:: mCherry (Figure S2, S7 and data not shown), suggesting that the cytoplasmic domain is not required for PAT-2 localization. We then used the PAT-2Δcyto::GFP signal to monitor the internalization of MSpppaaa cell corpses by muscle cells. In the wild-type embryos, approximately 66.6% of MSpppaaa cell corpses were enclosed by the PAT-2Δcyto::GFP circle at the time when grinder formation had just finished, whereas no MSpppaaa cell corpses were enclosed by the PAT-2Δcyto::GFP circle in the pat-2(st567) embryos at this stage (Table 5 and Figure S8). Thus, the internalization of MSpppaaa cell corpses is defective in the pat-2(st567); P_pat-2pat-2Δcyto::gfp embryos. This result and the aforementioned data together indicate that the PAT-2 cytoplasmic domain is required for the muscle-mediated engulfment of apoptotic cells, but is dispensable for its subcellular localization in muscle cells and function during muscle development.

Expression of the PAT-2Δcyto::GFP fusion protein in the wild-type embryos resulted in increased numbers of cell corpses at the comma and 1.5-fold stages comparable to those seen in pat-2 mutants, but failed to induce the Pat phenotype (Table 3). In addition, PAT-2Δcyto::GFP also prolonged the duration time of the MSpppaaa cell corpses and interfered with the internalization of the MSpppaaa cell corpses (Table 5). For example, in the wild-type embryos expressing PAT-2Δcyto::GFP, 39% of the MSpppaaa cell corpses persisted at the time when grinder formation had just finished, and 66.6% of these cell corpses were internalized by engulfing muscle cells. However, in the control wild-type embryos, only 16.6% of the MSpppaaa cell corpses were present at this stage, and all the remaining cell corpses were internalized by engulfing muscle cells (Table 5). This result reinforces the specific function of the PAT-2 cytoplasmic domain in the engulfment of apoptotic cells. PAT-2Δcyto::GFP may compete with the endogenous PAT-2 for binding of PAT-3 or apoptotic cells, but fail to initiate signaling for the internalization of cell corpses.

PAT-2 functions through UIG-1 and CDC-42 to promote the engulfment process

During the engulfment of apoptotic cells, cytoskeletal rearrangement occurs as an engulfing cell extends pseudopods around an apoptotic cell [71]. Rho-family GTPases are important regulators of the actin cytoskeleton [72]. Although CED-10 (RAC1) GTPase is required for cell corpse engulfment [55], [72], our data showing that a ced-10 mutation enhanced the engulfment defect in the pat-2(RNAi) mutant (Table 4) and that ced-10 overexpression failed to rescue the engulfment defect of the pat-2(RNAi) mutant (Table 6) suggest that pat-2 may act independently of ced-10 in promoting the cytoskeletal rearrangement required for the internalization process.

We then examined whether actin filament assembly occurred during the muscle-mediated internalization of apoptotic cells. MOESIN has been used in Drosophila and C. elegans to specifically mark the filamentous form of actin [73], [74]. Analysis of embryos expressing P_unc-54moesin::gfp in which MOESIN::GFP was expressed in muscle cells using P_unc-54 showed a MOESIN::GFP circle around apoptotic cells (Figure 2C), demonstrating that filamentous actin assembly occurs as an engulfing muscle cell extends pseudopods along the surface of an apoptotic cell.

A previous study showed that the Rho-family GTPase CDC-42 and UIG-1, a guanine nucleotide exchange factor (GEF) specific for CDC-42, function downstream of PAT-2/PAT-3 signaling for muscle assembly [75]. We next tested the involvement of uig-1 and cdc-42 in PAT-2-mediated cell corpse engulfment. The cdc-42(gk388) allele has a deletion that eliminates part of the 5′ regulatory sequence, the entire first exon, and part of the first intron of the cdc-42 gene (C. elegans Gene Knockout Consortium), while the uig-1(ok884) mutation deletes the region coding for the DH/PH domain, which has the GEF activity [76]. Both alleles are likely to be null. We found that uig-1 or cdc-42 embryos contained increased numbers of cell corpses at the comma and 1.5-fold stages (Table 4), similar to those observed in the pat-2 mutants. A four-dimensional DIC analysis of cell corpse persistence in the cdc-42(gk388) embryos revealed a cell-corpse engulfment defect. In the wild-type, 97% of cell corpses disappeared within 40 minutes and none lasted longer than 60 minutes (Figure 1C). However, in the cdc-42(gk388) mutants, only 40% of the cell corpses disappeared within 40 minutes and almost 40% persisted longer than 60 minutes (Figure 1C).

We next tested whether cdc-42 acted in the same pathway as pat-2 to promote apoptotic cell engulfment. We found that neither cdc-42 nor uig-1 increased the number of cell corpses in pat-2 mutants at the comma and 1.5-fold stages (Table 4), suggesting that cdc-42 and uig-1 both act in the same genetic pathway as pat-2. Like PAT-2::GFP, the GFP::CDC-42 fusion protein expressed using the transgene P_unc-54gfp::cdc-42 was localized in muscle pseudopods around apoptotic cells (Figure 2D). In addition, the transgene P_unc-54gfp::cdc-42 rescued the engulfment defect of the cdc-42(gk388) mutants, whereas P_ajm-1gfp::cdc-42, which expressed GFP::CDC-42 in hypodermal cells, did not (Table 6). This result suggests that cdc-42 acts in muscle cells to mediate the engulfment of apoptotic cells. This muscle-specific function of cdc-42 is further supported by the observation that loss of cdc-42 results in a defect in the engulfment of apoptotic MSpppaaa cells (Table 5), but not C1, C2 or C3 cells (Table S2). When P_unc-54gfp::cdc-42 and P_pat-2pat-2::mcherry were co-expressed, GFP::CDC-42 was co-localized with PAT-2::mCherry along the surface of apoptotic MSpppaaa cells (Figure 2N–2Q). Furthermore, the P_unc-54gfp::cdc-42 transgene rescued the engulfment defect of the pat-2(RNAi) embryos (Table 6). These results support the model that cdc-42 functions downstream of pat-2 in the muscle-mediated engulfment of apoptotic cells.

CED-10 and CDC-42 are not exchangeable in function during cell-corpse engulfment

Our genetic data suggested that cdc-42 acts with pat-2 in the same pathway, whereas ced-10 functions in a partially redundant manner with pat-2 to mediate cell corpse engulfment (Table 4). We then tested whether the functions of ced-10 and cdc-42 in engulfment were exchangeable when ubiquitously overexpressed. We found that overexpression of ced-10 or constitutively active ced-10 V12 by the heat-shock promoter P_hsp, which rescued the ced-10(n1993) or ced-10(tm597) mutant ([50], Table 6), failed to rescue the engulfment defect of the pat-2(RNAi) embryos (Table 6). Reciprocally, overexpression of cdc-42 by P_hsp, which rescued the pat-2 mutant (Table 6), did not rescue the engulfment defect of the ced-10(n3246) mutants (Table 6). Moreover, overexpression of cdc-42 by the P_hsp or P_ajm-1 promoter also failed to rescue the defective engulfment of the ced-1(e1735) or ina-1(gm144) mutants, respectively (Table 6). These data suggest that the mechanisms by which cdc-42 and ced-10 mediate the actin-based cytoskeletal rearrangement during cell-corpse engulfment are distinct and that their functions are not exchangeable.

INA-1 and PAT-2 can partially substitute for each other in cell-corpse engulfment

As shown in Figure S6E, the INA-1 and PAT-2 extracellular domains can recognize the same apoptotic cells. We found that overexpression of pat-2 by the P_ajm-1 promoter in hypodermal cells partially rescued the defective engulfment of the ina-1 mutants and, reciprocally, overexpression of ina-1 by the P_unc-54 promoter in muscle cells also partially rescued the engulfment defect of the pat-2 mutants (Table 2). This result indicates that ina-1 and pat-2 can partially substitute for each other in cell corpse engulfment. Interestingly, overexpression of ina-1 by the P_unc-54 promoter, which partially rescued the engulfment defect of the pat-2 mutants, did not rescue that of the cdc-42(gk388) mutants (Table 6) suggests that cdc-42 may be necessary for ina-1 overexpression-induced engulfment in muscle cells. If so, ina-1 overexpression in muscle cells may promiscuously activate cdc-42 signaling, which, in turn, leads to the engulfment of apoptotic cells. Nevertheless, ina-1 overexpression does not efficiently activate the phagocytosis machinery for cell corpse removal in muscle cells since only partial rescue of the pat-2 engulfment defect by ina-1 overexpression was observed. Similarly, overexpression of pat-2 dose not efficiently induce the phagocytosis machinery in hypodermal cells. This may, in part, explain why pat-2 and ina-1 preferentially function in different cell types for cell corpse removal, despite that they are expressed in both cell types.

In contrast, overexpression of ced-1 under the control of the P_unc-54 promoter failed to rescue the defective engulfment of the pat-2 mutants (Table 2). In a reciprocal experiment, overexpression of pat-2 by the P_ajm-1 promoter also failed to rescue the defective engulfment of the ced-1 mutants (Table 2). These results indicate that ced-1 and pat-2 functions are distinct and not exchangeable in cell corpse engulfment.

Discussion

The engulfment of apoptotic cells requires the recognition and subsequent internalization of apoptotic cells by the engulfing cells. Here, we showed that, in C. elegans, the integrin α subunit PAT-2 functions in both the recognition and internalization steps. pat-2 loss of function resulted in an increased number of embryonic cell corpses due to a defect in cell corpse removal. Our data showed that PAT-2 bound to apoptotic cells through its extracellular domain and initiated downstream signaling via its cytoplasmic domain. We characterized the pat-2 signaling pathway and identified a previously unassigned function of cdc-42 and uig-1 in cell corpse engulfment. We further showed that PAT-2 predominantly functioned in muscle cells to mediate the engulfment process. We propose that binding of PAT-2 to an apoptotic cell results in the recruitment of UIG-1 and the subsequent activation of CDC-42 GTPase, which, in turn, regulates cytoskeletal rearrangement as a muscle cell extends pseudopods around an apoptotic cell.

The finding that truncated PAT-2 lacking the cytoplasmic domain (PAT-2Δcyto::GFP ) fully rescued the Pat phenotype of the pat-2 mutants, but failed to rescue the engulfment defect (Table 3) argues against the possibility that the engulfment defect of the pat-2 mutant is a secondary effect caused by abnormality of muscle assembly or organization during embryogenesis. PAT-2 is co-localized with PAT-3 to the dense bodies and M-lines (Figure S2A), which are platforms serving as anchoring sites for signaling/adapter proteins in muscle attachment and organization [65]. Because the deletion of the cytoplasmic domain of PAT-2 affects neither its localization pattern (Figure S7) nor its function in muscle development (Table 3), muscle development probably requires only the transmembrane and extracellular domains, which are likely sufficient for the interaction of PAT-2 with PAT-3 and/or the extracellular matrix. Thus, the PAT-2/PAT-3 integrin likely mediates the intracellular signaling and/or adaptor protein binding through the cytoplasmic domain of PAT-3, but not that of PAT-2, for muscle development. In contrast, the requirement of the cytoplasmic domain of PAT-2 for cell-corpse removal suggests that this domain is important for signaling and/or serves as an anchorage site for adapter proteins during cell-corpse internalization.

The pat-2(st567) mutants show a weak engulfment phenotype compared with that of the ced-1 mutants during embryogenesis (Table 2). One possible explanation for the weak engulfment defect is that pat-2 predominantly functions in muscle cells, while only a small fraction of cell corpses (e.g. ∼20% of cell corpses in the 1.5-fold stage embryos, as shown in Table S1) are removed by muscle cells. Secondly, other engulfment receptor(s) may act redundantly with pat-2 in muscle-mediated engulfment. For instance, both PAT-2 and CED-1 are involved in the muscle-mediated internalization of MSpppaaa cells, despite that PAT-2 plays a bigger role than CED-1 (Table 5).

The PAT-2 extracellular region bound to apoptotic cells when PAT-2(ex) was fused to mCherry and overexpressed using the heat shock promoter (Figure 2M). Because exposed PS is detected on the surface of apoptotic MSpppaaa cells (Figure S1B), which are then removed by the PAT-2-mediated engulfment process (Table 5), PAT-2 might recognize exposed PS on apoptotic MSpppaaa cells. Mammalian integrins α_vβ₃ and α_vβ₅ have been shown to bind to apoptotic cells via the secreted bridging molecule MFG-E8 [25], [48]. In addition, integrin α_vβ₃ binds synergistically with the cell-surface protein CD36 to apoptotic cells through the bridging molecule thrombospondin, an extracellular matrix glycoprotein [50]. MFG-E8 binds to integrin α_v through its RGD domain. On the basis of amino acid sequence, PAT-2 is more closely related to the RGD-binding integrins than to the laminin-binding integrins. However, C. elegans does not appear to have an MFG-E8 homolog. A screen for RGD-containing molecules may be helpful in testing the involvement of RGD-containing molecules in the binding of PAT-2(ex)::mCherry to apoptotic cells.

Like PAT-2, CED-1 and INA-1 are expressed in muscle and hypodermal cells [40], [46]. It is intriguing that different receptors are preferentially used in different cell types when they are all present in these cells. This may be, in part, because some receptors preferentially bind to a subset, but not all, apoptotic cells. For example, PAT-2(ex)::mCherry binds to MSpppaaa, but not C1, C2 or C3, cell corpses (Figure S6A, S6C and S6D) and PAT-2 is required for the removal of MSpppaaa, but not C1, C2 or C3, cell corpses. In contrast, INA-1 and CED-1 receptors recognize apoptotic C1, C2 and C3 cells and mediate the engulfment of these cells [42], [46]. However, some cell corpses can be recognized by multiple receptors. For example, PAT-2(ex)::mCherry and INA-1(N)::GFP are co-localized on some apoptotic cells (Figure S6E). Therefore, additional factor(s) other than the receptor-cell corpse interaction determines the cell-type specificity of engulfment receptors. The observation that CDC-42 preferentially functions in muscle cells but not hypodermal cells for engulfment suggests that the downstream molecule(s) are important for the cell-type specificity of engulfment receptors. CED-10 and CDC-42 are important for the actin-based cytoskeleton rearrangement [72], which occurs during engulfment of apoptotic cells ([55], Figure 2C). Hypodermal cells and muscle cells appear to have different requirement for CED-10 and CDC-42 in cell corpse removal, although the two GTP ases are expressed in both cell types [76]. Hypodermal cells and muscle cells may utilize different actin-based phagocytosis mechanisms or employ different ways to activate the phagocytosis machinery for cell corpse removal.

We showed that the muscle-mediated internalization of cell corpses took apporximately 6 mintues (Figure 3A). Similarly, the internalization of the C3 cell corpse by the hypodermal cell ABplaapppp took about 6 minutes [42], [46]. However, previous studies by Wang et al. and Zou et al. [43], [53] showed that the time for the CED-1::GFP-mediated internalization of apoptotic cells may take about 18 or 25 minutes, respectively, although the identities of the engulfing cells are unclear. Therefore, the time necessary for the entire internalization process to occur from initiation to completion appears different for different engulfing cells. Nonetheless, at least some engulfment processes mediated by the muscle cell and hypodermal cell proceeds with similar kinetics.

Recently, conditional deletion of integrin α_v in the mouse immune system revealed that this protein is essential for the engulfment of apoptotic cells by gut-associated macrophages and dendritic cells [4]. In addition, mice lacking MFG-E8, which mediates apoptotic cell clearance through integrin α_v, are defective in the removal of apoptotic B cells by tingible body macrophages in the spleen germinal centers [77]. However, little was previously known about whether integrin α or other engulfment receptors were involved in apoptotic cell removal mediated by amateur phagocytes, such as muscle cells. C. elegans provides a good system for studying the amateur phagocyte-mediated engulfment of apoptotic cells, as it does not have professional phagocytes. Our data showed that PAT-2 acts in muscle cells and transduces the engulfment signal through a novel signaling pathway for apoptotic cell removal. Recently, a mouse lacking ELMO1 showed a defect in Sertoli cell-mediated engulfment of apoptotic germ cells, but no engulfment defect was detected in macrophages or fibroblasts [31]. This result, together with our data, suggest that, like professional phagocytes [78], amateur phagocytes in different tissues utilize different sets of engulfment receptors and signaling molecules for apoptotic cell engulfment at the whole organism level.

Materials and Methods

Strains and alleles

The N2 Bristol strain was used as the reference wild-type strain. All strains were maintained on nematode growth medium (NGM) plates with Escherichia coli OP50 as the food source at 20°C unless otherwise noted [79]. The following mutations were used: linkage group I (LGI), ced-1(e1735), ced-12 (n3261, tp2); LGII, cdc-42(gk388), mIn1[mIs14 dpy-10(e128)]; LGIII, abi-1(ok640), ced-4(n1162), ced-6(n1813), ced-7(n1892, n1996), unc-79(e1068), pat-2(st567), dpy-17(e164), pat-3(st564), qC1 dpy-19(e1259) glp-1(q339); LGIV, ced-2(n1994), ced-3(n717), ced-5(n1812), ced-10(n1993, n3246, tm597); LGIV, uig-1(ok884). dpy-17(e164) was used to balance unc-79(e1068) pat-2(st567). The homozygous unc-79(e1068) pat-2(st567) mutants were also maintained using the extrachromosomal P_pat-2pat-2::gfp or P_pat-2pat-2::mcherry transgene. In either of the transgenic lines, the unc-79(e1068) pat-2(st567) mutant embryos that lost the transgene were both Pat and Ced. The numbers of cell corpses in the unc-79(e1068) mutant at the comma and 1.5-fold stages were similar to those of the wild-type at the same stage, indicating that unc-79(e1068) does not affect apoptosis. qC1 dpy-19(e1259) glp-1(q339) was used to balance pat-3(st564). mIn1 was used to balance cdc-42(gk388). The pat-2(st567) allele has a G441D alternation in the extracellular domain. The information of the transgenic strains used in this work is listed in Table S4.

Plasmid construction

Three vectors were used to generate constructs expressing fluorescent proteins, the gfp vector pPD95.75, the mcherry vector pYW806, and the mrfp vector pYW897. pYW806 and pYW897 were respectively generated by replacing gfp in pPD95.75 with mcherry or mrfp via the KpnI/EcoRI sites. To generate P_pat-2pat-2::gfp (pYW903) or P_pat-2pat-2::mCherry (pYW950), the 10 kb fragment containing the 4 kb upstream sequence and the full-length pat-2 coding region without the stop codon was PCR-amplified from the cosmid F54F2 (Sanger Institute, Cambridge, United Kingdom) using the primers TCCCCCCGGGTTTATGACTCACAGAC and GGGTACCGATGCATTTGTC CGTGACGT, and cloned into pPD95.75 or pYW806, respectively, via the KpnI site. The full-length pat-2 cDNA construct (pYW901) was generated by inserting into yk616b4 (Dr. Yuji Kohara) via the PstI site a 0.6 kb pat-2 cDNA which was amplified by RT-PCR using the primers AACTGCAGATGCGAGAGGGTAGTTTTCC and GATTCTTCTTTCCTGGAACTGCAGC. To generate P_pat-2nls::gfp (pYW949) or P_pat-2gfp (pYW903), the 4 kb upstream sequence of pat-2 was first amplified by PCR from the cosmid F54F2 using the primers TCCCCCCGGGTTTATGACTCACAGAC and TCCCCCCGGGATCTACTGG AAATTTG and inserted into pPD95.67 or pPD95.75, respectively. The P_unc-54gfp (pYW899) or P_unc-54mcherry (pYW900) construct was generated by inserting the 1 kb HindIII/KpnI fragment of pPD30.38 containing P_unc-54 into pPD95.75 or pYW806, respectively. The P_ajm-1gfp (pYW902) was generated by inserting into pPD95.75 via the SalI/BamHI sites a 5.5 kb P_ajm-1 fragment which was amplified by PCR from pBR980 [69] using the primers CGTCGACCGATTTGACCGTTCGATAAG and CGGATCCTCGTCGGTA GTACTCGTCC. P_ced-1gfp (pYW898) was generated by inserting into pPD95.75 via the SphI site a 5 kb P_ced-1 fragment which was PCR-amplified from genomic DNA using the primers GGCATGCATACCTCCTGATATG GGGTGA and GCATGCTTGC GGCTGCAAAAAAACAGGG. P_ced-1ced-1::gfp (pYW904) was generated by three-piece ligation. The 6 kb PCR-amplified fragment from P_ced-1ced-1::gfp [40] using the primers AGGTACCATGCGTCTCATTCTCCTTGTGC and GGTCGA CGTGATTGTTCAGATGA and the 2.4 kb fragment PCR-amplified from genomic DNA using the primers CGTCGACCTCTATTAGAAGAGCATGACG and TGGTACCGAGGTGTACAAATTGTCCTGAGC were inserted into pYW898 via the SalI and KpnI sites. P_ced-1ced-1::gfp fully rescued the engulfment defect of the ced-1(e1735) mutant (data not shown). To generate pYW901 containing pat-2 cDNA without the stop codon, pat-2 cDNA was PCR-amplified from the full-length pat-2 cDNA clone using the primers CGGTACCATGCGAGAGGGTAGTTTTCC and GGGTACCGATAGCATTTGTC CGTGACGT and inserted into the pGEM-T Easy vector (Promega) via the KpnI site. To generate P_unc-54pat-2::gfp (pYW913) or P_unc-54pat-2::mCherry (pYW916), the 3.7 kb KpnI pat-2 fragment from pYW901 was inserted into pYW899 or pYW900, respectively, via the KpnI site. To generate P_unc-54ced-1::gfp (pYW905), the 8.5 kb KpnI ced-1 fragment from pYW904 was inserted into pYW899 via the KpnI site. To generate P_unc-54 ced-1::mrfp (pYW941), a 8.6 kb SpeI/BamHI fragment, containing P_unc-54 and the first 4.9 Kb of ced-1, and a 3.6 kb BamHI/KpnI fragment, corresponding to the rest ced-1 sequence, of pYW905 were fused to a 1.5 kb KpnI/SpeI fragment, containing the mrfp sequence, of pYW897 by three-piece-ligation. To generate P_unc-54moesin::gfp (pYW940), the 0.6 kb SmaI/NcoI fragment containing the moesin actin-binding sequence was inserted into pYW899 via the NheI/NcoI sites. The moesin plasmid was a gift from Dr. Fabio Piano [75]. To generate P_unc-54 gfp::cdc-42 (pYW906), two plasmids gfp_Ntag_TA and cdc-42_Ntag_TA were generated first. To generate the gfp_Ntag_TA plasmid, PCR-amplified gfp, using oligonucleotides CGGTACCATGAGTAAAGGAG AAGAACT and GTCTAGATTTGTATAGTTCATCCATGCC as primers and pPD95.75 as template was inserted to the vector pGEM T-Easy. To generate the cdc-42_Ntag_TA plasmid, PCR-amplified cdc-42 from k1101h01(Y. Kohara) using primers GTCTAGAATGCAGACGATCAAGTGCG and CGTTAACCTAGAGAATATTGCACTTCTTC was inserted into the pGEM T-Easy vector. The 1 kb HindIII/KpnI fragment of pPD30.38 containing P_unc-54, the KpnI /XbaI gfp fragment from the gfp_Ntag_TA plasmid, and the XbaI/EcoR1 cdc-42 fragment from the cdc-42_Ntag_TA plasmid were inserted to the pPD95.75 vector previously digested with HindIII and EcoRI in a four-piece-ligation reaction to generate pYW906. To generate P_ajm-1pat-2::gfp (pYW948) or P_ajm-1ced-1::gfp (pYW914), the 3.7 kb KpnI pat-2 sequence from pYW901 or the 8.5 kb KpnI ced-1 sequence from pYW904 was inserted, respectively, into pYW902. To generate P_ced-1pat-2::gfp (pYW942), the 3.7 kb KpnI pat-2 sequence from pYW901 was inserted into pYW898 via the KpnI site. To generate P_pat-2pat-2Δcyto::gfp (pYW964), pat-2Δcyto was PCR-amplified from pYW901 using the primers CGGTACCATGCGA GAGGGTAGTTTTCC and TTGGT ACCGTC CTATAGAATAATGCAA and cloned into pYW903 via the KpnI site. To generate Pemphtypepat-2(ex)::mcherry (pYW917), the DNA fragment corresponding to pat-2(ex) was PCR-amplified from pYW901 using the primers CGGTACCATGCGAGAGG GTAGTTTTCC and CGGTACCAGATCTCTTCC TTCTTCAGA and cloned into pYW900 via the KpnI site. To generate P_hsppat-2(ex)::mcherry (pYW966), the 4.2 kb NheI/PvuI fragment of the P_unc-54 pat-2(ex)::mCherry plasmid containing pat-2(ex)::mCherry was inserted into pPD49.78 via the NheI/PvuI sites. To generate P_hspgfp::cdc-42 (pYW959), the 1.5 kb KpnI/HpaI fragment of pYW906 corresponding to gfp::cdc-42 was inserted into pPD49.78 and pPD49.83 (different tissue specificity) via the KpnI/EcoRV sites. To generate the P_unc-54ina-1::gfp plasmid, P_unc-54 was PCR-amplified using pYW913 as template and oligonucleotides GCATCCGCCAAGCTTGTCTTCTTC and GGATCCGGTACCGT CGACGCTAC as primers and used to replace the P_ced-1 region of P_ced-1ina-1::gfp via the SphI/BamHI sites. To generate the P_pat-3pat-3::gfp (pYW1091) plasmid, the 10.2 kb pat-3 genomic DNA was amplified by PCR and subsequently cloned into the pPD95.75 vector via the XmaI/KpnI sites. To generate the P_unc-54myri::mrfp (pYW1092) construct, the myri::mrfp cDNA of P_unc-86myri::gfp (from C. Bargmann) was inserted into the P_unc-54mrfp vector via the KpnI site. To generate the P_ajm-1gfp::cdc-42 (pYW1093) plasmid, the gfp region of the pPD95.75 plasmid was replaced by gfp::cdc-42 from the P_hspgfp::cdc-42 construct via the KpnI site, and the resulting construct was then digested with BamHI and SalI and subsequently inserted with the BamHI and SalI fragment containing the P_ajm-1 promoter from pYW902.

RNAi experiments

To generate the pat-2 RNAi clone, the 1.3 kb PstI/HindIII fragment of pat-2 cDNA was inserted into the pPD129.36 vector. The pat-3 RNAi plasmid was obtained from the J. Ahringer RNAi library. All RNAi experiments were carried out using a bacterial feeding protocol [80]. In brief, L4 larvae were laid out on control plates (pPD129.36) or the indicated RNAi plates and cultured at 20°C for 48 h, then the F1 embryos were picked for phenotypic analysis.

Transgenic animals

Transgenic animals were generated by microinjection of the indicated plasmid(s) (3–50 ng/µL), using the pRF4[rol-6 (su1006)], pTG96[sur-5::gfp] or P_sur-5rfp plasmids (50 ng/µL) as coinjection markers [81], [82]. The injection procedure was performed as described previously [8]. The resulting transgenes and genetic backgrounds of the strains were listed in Table S4. P_hspced-10V12 was injected with the coinjection marker pTG96[sur-5::gfp] to unc-79(e1068) pat-2(st567)/dpy-17(e164), and no rescue of the Pat phenotype was observed. To score cell corpses, transgenic embryos carrying the transgene P_hspced-10V12 and pTG96[sur-5::gfp] were scored and only the data for those which later exhibited the Pat phenotype were used.

Microscopy

Embryos were mounted on a 4% agar pad in M9 buffer in the presence (four-fold stage embryos) or absence (comma and 1.5-fold stage embryos) of 30 mM sodium azide at 20°C. Cell corpses were analyzed using DIC microscopy, as previously described [46]. A Zeiss Axioplan 2 microscope equipped with a digital camera (AxioCam; Carl Zeiss, Inc.) and 4.7 AxioVision imaging software was used. To obtain the cell corpse data of homozygous pat-2(st567) embryos derived from the heterozygous unc-79(e1068) pat-2(st567)/dpy-17(e164) mothers, embryos at the indicated developmental stages were analyzed and only the data for those which later exhibited the Pat phenotype were used. To obtain the cell corpse data of homozygous pat-2(st567) embryos derived from unc-79(e1068) pat-2(st567) mothers carrying either P_pat-2pat-2::gfp or P_pat-2pat-2::mcherry transgene, non-fluorescent embryos at the indicated developmental stages were analyzed. To obtain the cell corpse data of homozygous cdc-42(gk388) embryos derived from the heterozygous cdc-42(gk388)/mIn1[mIs14 dpy-10(e128)] mothers, GFP-negative embryos at the indicated developmental stages were analyzed. To analyze cell death events during embryogenesis, wild-type or mutant embryos were filmed during the period 200–460 min after the first cleavage and the time point at which each cell corpse appeared was noted and was reported relative to the first cell death for comparison between the wild-type and mutants. About 40–50 serial Z-sections were recorded at 0.4 µm intervals every 1 min. To measure the duration of cell corpses in the wild-type and mutants, cell corpses appearing between 360 to 410 min after the first cleavage during embryogenesis were followed, and about 50–60 serial Z-sections were recorded at 0.3 or 0.4 µm intervals every 1 min. To monitor the muscle-mediated internalization of apoptotic cells, the fluorescence images of wild-type or mutant embryos carrying the P_unc-54myri::mrfp transgene were recorded using the DeltaVision microscope (GE Healthcare company) equipped with a digital camera (Photometrics Cascade II 512 EMCCD) at 1 (for wild-type) or 3 (for mutant)-minute intervals for about 120 minutes.

Heat shock experiments

For the heat shock rescue experiments, transgenic embryos were subjected to heat shock at 33°C for 30 min and transferred to 20°C to recover for 2 hours, then cell corpses in embryos at the indicated stages were counted. To test the binding of PAT-2(ex)::mCherry to apoptotic cells and its effect on cell-corpse engulfment when overexpressed, embryos carrying the transgene P_hsppat-2(ex)::mcherry or embryos carrying the transgenes P_hsppat-2(ex)::mcherry and P_hspina-1(N)::gfp were subjected to heat shock at 33°C for 60 min and transferred to 20°C to recover for 4–5 hours, then embryos were examined using fluorescence microscopy (for mCherry and/or INA-1(N)::GFP signal) or DIC microscopy (for cell corpses). To overexpress GFP::CDC-42, CED-10V12 or Annexin V::mRFP by heat-shock, the embryos carrying the respective transgene were subjected to heat shock at 33°C for 30 min and transferred to 20°C to recover for 2 hours, then embryos were examined using fluorescence microscopy (mCherry signal) or DIC microscopy (cell corpses).

Supporting Information

Attachment 1

Attachment 2

Attachment 3

Attachment 4

Attachment 5

Attachment 6

Attachment 7

Attachment 8

Attachment 9

Attachment 10

Attachment 11

Attachment 12

Attachment 13