Clathrin and AP2 Are Required for Phagocytic Receptor-Mediated Apoptotic Cell Clearance in Caenorhabditis elegans

Clathrin and the multi-subunit adaptor protein complex AP2 are central players in clathrin-mediated endocytosis by which the cell selectively internalizes surface materials. Here, we report the essential role of clathrin and AP2 in phagocytosis of apoptotic cells. In Caenorhabditis elegans, depletion of the clathrin heavy chain CHC-1 and individual components of AP2 led to a significant accumulation of germ cell corpses, which resulted from defects in both cell corpse engulfment and phagosome maturation required for corpse removal. CHC-1 and AP2 components associate with phagosomes in an inter-dependent manner. Importantly, we found that the phagocytic receptor CED-1 interacts with the α subunit of AP2, while the CED-6/Gulp adaptor forms a complex with both CHC-1 and the AP2 complex, which likely mediates the rearrangement of the actin cytoskeleton required for cell corpse engulfment triggered by the CED-1 signaling pathway. In addition, CHC-1 and AP2 promote the phagosomal association of LST-4/Snx9/18/33 and DYN-1/dynamin by forming a complex with them, thereby facilitating the maturation of phagosomes necessary for corpse degradation. These findings reveal a non-classical role of clathrin and AP2 and establish them as indispensable regulators in phagocytic receptor-mediated apoptotic cell clearance.


Introduction
Phagocytosis of apoptotic cells is critical to tissue remodeling, suppression of inflammation and control of immune responses [1,2]. During phagocytosis, cell corpses are firstly engulfed and subsequently degraded by phagocytes, both phases being controlled by evolutionarily conserved regulators. In the lifetime of a C. elegans hermaphrodite, 131 somatic cells and about half the germ cells undergo apoptosis and the resulting cell corpses are quickly removed by neighboring cells in the soma or by sheath cells encasing the germ line. The engulfment of cell corpses is essentially controlled by two partially redundant signaling pathways that induce the cytoskeletal reorganization of engulfing cells [3]. In one pathway, the intracellular molecules CED-2/ CrKII, CED-5/DOCK180, and CED-12/ELMO act through a protein interaction cascade to induce the activation of the small GTPase CED-10/Rac1, leading to the cytoskeleton reorganization necessary for engulfment [4][5][6][7]. In addition, the phosphatidylserine (Ptdser) receptor PSR-1 likely binds Ptdser, an ''eat me'' signal, and acts upstream of CED-2, -5, and -12 to regulate engulfment [4]. Two other signaling modules, INA-1/integrin-SRC-1/Src and UNC-73/TRIO-MIG-2/RhoG, were also found to function through the CED-5-CED-12 motility-promoting complex to facilitate CED-10 activation for corpse engulfment [8,9]. In addition, a non-canonical Wnt pathway consisting of the MOM-5 receptor, GSK-3 kinase and APC/APR-1 may act through CED-2 to regulate CED-10 activity for cell corpse engulfment during early embryo development [10]. In the other pathway, the phagocytic receptor CED-1, which shares homology with the human scavenger receptor SREC, LRP/CD91 and MEGF10, and Drosophila Draper and Six-microns-under (SIMU) [11][12][13][14][15], recognizes apoptotic cells by interacting with TTR-52, a PtdSer-binding protein secreted from engulfing cells [16]. The adaptor protein CED-6/Gulp likely acts downstream of CED-1 to transduce engulfing signals to other effectors including the large GTPase DYN-1/dynamin, resulting in cell corpse engulfment and formation of phagosomes [14,17,18]. In addition, the ABC transporter CED-7 is also required for cell corpse recognition by CED-1 in embryos [11,19]. Recent studies suggest that CED-7 acts with TTR-52 and NRF-5, another secreted PtdSer-binding protein, to mediate PtdSer transfer from cell corpses to phagocytes, thus promoting the recognition of cell corpses by CED-1 [20,21]. Subsequent to corpse internalization, CED-1 is recycled from the phagosome back to the plasma membrane by the retromer complex [22]. Phagosomes enclosing cell corpses then undergo a maturation process by dynamically fusing with endocytic organelles including early and late endosomes as well as lysosomes, leading to formation of phagolysosomes in which cell corpses are ultimately digested. It has been found that several molecules required for endocytic trafficking, such as DYN-1/ Dynamin, the phosphatidylinositol-3 kinase (PI3K) VPS-34, small GTPases and their regulators or effectors including RAB-2, RAB-5, TBC-2, RAB-7, RAB-14, and the HOPS complex, act in an ordered manner to regulate phagosome maturation [23][24][25][26][27][28][29]. As the phagolysosome forms, it is progressively acidified in order to activate lysosomal enzymes needed for cell corpse digestion [30].
The phagocytic receptor CED-1 plays a leading role in apoptotic cell clearance by recognizing cell corpses and transducing signals for engulfment and phagosome maturation. Nevertheless, it remains largely unknown how the CED-1-mediated signaling pathway triggers the cytoskeletal reorganization required for corpse internalization. In addition, the mechanisms governing the transition from corpse internalization to phagosome maturation are poorly understood. Interestingly, CED-1-mediated phagocytosis of cell corpses appears to resemble clathrin-mediated endocytosis (CME) of cell surface molecules in that both events cause receptor-dependent internalization of extracellular cargoes differing only in size [31]. In CME, recognition of the cytoplasmic domains of plasma membrane receptors by adaptor proteins triggers the formation of clathrin-coated vesicles (CCVs) with diameters ranging from 10-200 nm [32,33]. The formation of cargo-containing CCVs requires several protein module-mediated events, including FCH domain-only (FCHO) complex-mediated initiation, adaptor protein 2 (AP2)-dependent cargo selection and coat building, dynamin-mediated scission, and auxilin-and heat shock cognate 70 (HSC70)-dependent uncoating [32]. Recent studies revealed that some of the molecules required for CME are involved in phagocytosis of pathogens or the maturation of phagosomes containing apoptotic cells. For example, clathrin and the adaptor protein Dab2 were found to be important for phagocytosis of pathogenic bacteria by mammalian cells [34]. In C. elegans, DYN-1/dynamin regulates the initiation of phagosome maturation [23]. However, whether other components of the CME pathway play a role in apoptotic cell clearance remains unknown.
In this study, we investigated the mechanisms underlying CED-1-mediated cytoskeleton reorganization for phagocytosis and uncovered the role of CME regulators in apoptotic cell clearance. Our findings revealed that clathrin and the AP2 complex, the central players in CME, are critical to apoptotic cell removal in C. elegans. We found that clathrin and AP2 act downstream of CED-1 and CED-6 by forming a complex with them, which mediates the rearrangement of the actin cytoskeleton required for cell corpse engulfment. In addition, we demonstrated that LST-4, the C. elegans homolog of Snx9/18/33, functions at an early step of phagosome maturation by promoting phagosomal association of DYN-1/dynamin. Furthermore, we demonstrated that clathrin and AP2 also interact with LST-4 and DYN-1 to regulate the initiation of phagosome maturation required for cell corpse degradation. These findings suggest that clathrin and AP2 play essential roles in the phagocytic receptor CED-1-mediated apoptotic cell clearance pathway by regulating cytoskeletal reorganization and facilitating phagosome maturation.

Results
Clathrin and the AP2 complex are important for engulfment of germ cell corpses in C. elegans To explore how the phagocytic receptor CED-1 and its downstream adaptor CED-6 function to induce cytoskeletal reorganization for cell corpse engulfment, we firstly sought to identify proteins that are in complex with endogenous CED-1 and/or CED-6. Using antibodies specific for the C-terminus of CED-1 (CED-1C) [22] and CED-6, we performed immunoprecipitations in whole cell lysates of wild type (N2), ced-1(e1735) and ced-6(n1813) strong loss-of-function mutants, and analyzed proteins in the precipitates using liquid chromatography-coupled tandem mass spectrometry (LC-MS/MS). Interestingly, multiple peptides of the heavy chain of clathrin (CHC-1) were identified from proteins co-precipitated with CED-1 and CED-6 in the N2 lysate but not in lysates of ced-1(e1735) or ced-6(n1813) mutants ( Figure S1A and S1B). We therefore used RNA interference (RNAi) to deplete chc-1 and examined the persistence of cell corpses in C. elegans germ lines. We found that germ cell corpses accumulated significantly in an age-dependent manner in chc-1(RNAi) animals. A similar increase was observed at 25uC in a chc-1 temperature-sensitive mutant, b1025ts [35], though to a lesser extent than in chc-1(RNAi) animals ( Figure 1A). These results indicate that loss of clathrin function caused accumulation of apoptotic cells in C. elegans germ lines. Previously it was also reported that clathrin RNAi induced an elevation in the number of germ cell corpses [23], but how clathrin functions in this process remains unclear.
Given the central role of clathrin in CME, we went on to investigate whether inactivation of other regulators required for CME could also result in accumulation of apoptotic cells. We used RNAi to deplete C. elegans homologs of individual mammalian proteins involved in CME and examined the persistence of germ cell corpses. Compared to animals with control RNAi, a significant

Author Summary
Phagocytosis of apoptotic cells is an indispensable part of the cell death program. During phagocytosis, the evolutionarily conserved CED-1 family phagocytic receptors recognize cell corpses and transduce engulfment signals to induce the formation and maturation of phagosomes. However, it remains largely unknown how the CED-1 signaling pathway induces the cytoskeletal reorganization required for corpse internalization and initiates phagosome maturation. Interestingly, cell corpse phagocytosis appears to resemble clathrin-mediated endocytosis (CME) of cell surface molecules in that both events cause receptor-dependent internalization of extracellular cargoes differing only in size. In CME, the recognition of plasma membrane receptors by adaptor proteins such as the AP2 complex triggers the formation of clathrin-coated vesicles (CCVs) with diameters ranging from 10-200 nm. Nevertheless, it is not known whether clathrin and AP2 also play a role in phagocytosis of apoptotic cells that are much larger than CCVs. Here we provide genetic, cell biological, and biochemical experimental findings to demonstrate that clathrin and AP2 act downstream of CED-1 to regulate the actin cytoskeleton rearrangement required for cell corpse internalization and cell corpse degradation. Clathrin and AP2 form two types of complex with factors required for engulfment and phagosome maturation. These findings establish clathrin and AP2 as essential players in phagocytic receptor-mediated apoptotic cell clearance.
increase in germ cell corpses was observed in animals treated with RNAi of apa-2, apb-1 and dpy-23, which encode the a, b2 and m2 subunits of the AP2 complex, respectively (Table S1). A timecourse analysis confirmed that germ cell corpses increased significantly in an age-dependent manner in animals with RNAi of apa-2, apb-1, and dpy-23, but not aps-2, which encodes the s2 subunit of the AP2 complex ( Figure 1B). In addition, RNAi of lst-4 and dyn-1, which encode C. elegans homologs of mammalian sorting nexins 9/18/33 and dynamin, respectively, also led to a strong accumulation of germ cell corpses (Table S1) [18,22,36]. Clathrin and the AP2 complex are essential for formation of CCVs while sorting nexin 9 and dynamin are required for scission of CCVs from the plasma membrane during endocytosis [32,37].
To distinguish whether the increase in germ cell corpses in above RNAi animals resulted from excessive apoptosis or defective cell corpse clearance, we measured the duration of cell corpses using time-lapse recording. In animals with control RNAi, the average duration of germ cell corpses was 29.362.5 min (mean6-SEM, standard error of the mean). In contrast, the majority of germ cell corpses in chc-1(RNAi) and apb-1(RNAi) animals persisted longer than 120 min, and no cell corpses existed less than 60 min ( Figure 1C), suggesting that inactivation of clathrin or AP2 likely caused defective cell corpse clearance. To prove this, we performed transmission electronic microscopy (TEM) analysis to examine the engulfment of germ cell corpses. In chc-1(RNAi) animals, 28 out of 50 germ cell corpses examined (56%) from 5 gonad arms appeared not to be fully engulfed by sheath cells ( Figure 1D). Similarly, 12 of 25 corpses (48%) from 7 gonad arms of apb-1(RNAi) animals were found not to be internalized ( Figure 1D). In contrast, in animals treated with gla-3 RNAi, which induces excessive germ cell apoptosis without affecting cell corpse clearance, 100% of cell corpses were fully internalized by gonad sheath cells ( Figure 1D) [22,38]. These findings indicate that the engulfment of cell corpses was impaired when clathrin and the AP2 complex were down-regulated.
Clathrin and AP2 associate with phagosomes and act in the ced-1 engulfment pathway In C. elegans, the ced-1/6/7 and ced-2/5/12/10 signaling pathways act redundantly to mediate cell corpse engulfment [3]. As our findings revealed that CHC-1 and AP2 are involved in cell corpse engulfment, we tested whether depletion of individual AP2 components and chc-1 could exert an additive effect on defective corpse engulfment in mutants deficient in either engulfment pathway. In ced-1(e1735) and ced-6(n2095) strong loss-of-function mutants in the cell corpse recognition pathway, RNAi of apb-1, dpy-23, or chc-1 did not obviously change the numbers of germ cell corpses at all adult ages examined ( Figure 2A). In contrast, RNAi depletion of these three genes significantly enhanced the numbers of germ cell corpses in ced-2(n1994) and ced-5(n1812) strong loss-offunction mutants affecting the cytoskeletal reorganization pathway ( Figure 2B). These results suggest that chc-1 and genes of the AP2 complex likely act within the same genetic pathway as ced-1 and ced-6 to regulate cell corpse clearance.
CHC-1 and AP2 are required for actin rearrangement during phagocytosis Next we investigated whether loss of clathrin or AP2 function affects the rearrangement of the actin cytoskeleton, which is required for internalization of cell corpses. For this purpose, we generated transgenes expressing GFP-fused ACT-1, an actin isoform that controls cytoplasmic microfilament function, and GFP-tagged Drosophila Moesin (GFP::Moesin) [39], a filamentous actin (F-actin)-specific-binding protein, both of which were driven by the engulfing cell-specific ced-1 promoter. In wild-type animals, about 50% of germ cell corpses were surrounded by GFP::ACT-1.
We also tested the interaction of GST-fused CED-6 with individual AP2 components and CHC-1, and found that GST-CED-6 directly interacted with 35 S-labeled APA-2, APB-1 and DPY-23, and His6-tagged CHC-1C ( Figure 4A). These findings suggest the possibility that CED-1 and CED-6 form a complex with the AP2 complex and CHC-1. To prove this, we examined the interaction of CHC-1 and individual components of the AP2 complex with endogenous CED-1 and CED-6 by performing immunoprecipitations with CED-1C-or CED-6-specific antibodies. We found that endogenous CED-6 associated with CED-1 immunoprecipitated by the CED-1C antibody, providing direct evidence that CED-1 and CED-6 form a complex in C. elegans ( Figure 4C). Importantly, APA-2::GFP was co-immunoprecipitated with endogenous CED-1 and CED-6 ( Figure 4D and 4E), and similar co-immunoprecipitation of DPY-23::GFP and GFP::CHC-1 with endogenous CED-6 was observed ( Figure 4F and 4G). The specificity of these in vivo protein interactions was supported by the absence of any interaction, using the same immunoprecipitation assay, between CED-6 and GFP-fused LST-4 and DYN-1, two factors required for phagosome maturation (see below) ( Figure 4H and 4I). Thus the in vivo interactions of CHC-1 and individual AP2 components with CED-6 or CED-1 are consistent with their direct interactions in vitro, suggesting that the AP2 complex and CHC-1 likely fulfill their functions in cell corpse engulfment by forming a complex with CED-1 and CED-6.
Loss of chc-1 and AP2 function block phagosome maturation As EM analysis indicated that a significant proportion of germ cell corpses were still internalized by sheath cells in chc-1(RNAi) and apb-1(RNAi) animals, we wondered whether maturation of phagosomes containing cell corpses was affected in these animals. To assess this, we examined the acidification of phagosomes with LysoSensor Green DND-189, an indicator of normal progression of phagosome maturation. We found that germ cell corpses were mostly negative for LysoSensor Green DND-189 staining in chc-1(RNAi) and apb-1(RNAi) animals compared to animals with gla-3 RNAi that induces an elevation in apoptosis without affecting cell corpse clearance [38], suggesting that the maturation of phagosomes was inhibited ( Figure 5A). To corroborate this conclusion, we examined phagosomal recruitment of several effectors essential for phagosome maturation in apb-1(RNAi) and chc-1(RNAi) germ lines, including GFP-fused RAB-5 (GFP::RAB-5) and mCherry-fused RAB-14 (mCherry::RAB-14), two small GTPases required for phagosomal progression from early to late stages, and mCherry-fused NUC-1 (NUC-1::mCherry), a lysosomal DNase that indicates the formation of phagolysosomes [27]. We found that the labeling of cell corpses by these phagosomal markers in apb-1(RNAi) and chc-1(RNAi) animals was greatly reduced compared to that in wild type ( Figure 5B-5E), indicating that phagosomes in these animals arrested at an early stage of maturation. Taken together, these data indicate that clathrin and AP2 act at an early stage of phagosome maturation, impairment of which inhibited the progression of phagosomes from early to late stages.

LST-4 acts at an early stage of phagosome maturation
To elucidate how AP2 and CHC-1 function in phagosome maturation in addition to their role in cell corpse engulfment, we sought to determine their functional interactions with two other regulators identified in our screen, LST-4 and DYN-1, the C. elegans homologs of mammalian Snx9/18/33 and dynamin, respectively. Snx9 and dynamin act together with AP2 and clathrin to regulate the formation of CCVs in CME [32,40,41]. DYN-1 was previously shown to act at an early stage of phagosome maturation by forming a complex with VPS-34 and RAB-5 whereas LST-4 likely affects cell corpse degradation at a similar stage to DYN-1 [22,23,36,42]. As the first step towards our goal, we set out to clarify the role of LST-4 in phagosome maturation by comparing the cell corpse phenotype of two deletion mutants, tm2423 and qx159. We found that these mutants accumulate germ cell corpses to similar levels in an age-dependent manner ( Figure S5A and S5B). In addition, around 70% of lst-4(tm2423) germ cell corpses were found to be encircled by GFP::Moesin, compared with 60% in wild type, indicating that loss of lst-4 did not affect cell corpse internalization ( Figure S5C and S5H). However, germ cell corpses labeled by the early phagosome markers YFP::2xFYVE, GFP::RAB-5, and mCherry::RAB-14, and the late phagosome marker GFP::RAB-7, were greatly reduced in lst-4(tm2423) animals, indicating that loss of lst-4 inhibited the recruitment of factors required for phagosome maturation ( Figure S5D-S5H). Moreover, loss of lst-4 also blocked phagosome acidification as the majority of germ cell corpses were negative for LysoSensor Green DND-189 staining in either lst-4(tm2423) single mutants or double mutants of lst-4(tm2423) with vps-18(tm1125) that was previously shown to cause defective phagolysosome formation but not phagosome acidification [29] ( Figure S6A and S6B). This contrasts to the high proportion of corpses stained by the same dye in gla-3(RNAi) animals, in which cell corpses are normally removed, and in vps-18(tm1125) single mutants ( Figure S6A and S6B). Importantly, we found that LST-4 was recruited to phagosomes using LST-4::GFP or LST-4::mCherry fusions that fully rescued the cell corpse phenotype in lst-4(tm2423) mutants, even though they appeared cytoplasmic in several tissues ( Figure S6C-S6E; Figure S7A). The phagosomal association of LST-4 was blocked by loss of ced-1 and ced-6 but not RNAi depletion of dyn-1, rab-5 and rab-7, three genes required for phagosome maturation but not corpse engulfment ( Figure S6E and S6F). Together, these findings, in agreement with the results obtained by Almendinger et al. [36], establish that LST-4 acts at an early stage of phagosome maturation.

LST-4 functions through DYN-1 to promote phagosome maturation
We next characterized the functional interaction between LST-4 and DYN-1. In animals co-expressing LST-4::mCherry and DYN-1::GFP, which is able to rescue the defective cell removal in dyn-1(ky51) mutants, both proteins were found to colocalize on phagosomes ( Figure S7A and S7B). Time-lapse analysis revealed that both proteins were simultaneously recruited to the phagosome and quickly formed a crescent-like structure, before dissociating from the phagosome at the same time ( Figure 6A). Using immunoprecipitation we further found that these two proteins associated with one another in C. elegans ( Figure S7C, top panel) whereas they did not show detectable in vivo interaction with CED-6 ( Figure 4H and 4I). Consistent with this, His6-tagged recombinant LST-4 directly interacted with GST-fused DYN-1, which confirmed the in vitro interaction of these two proteins reported previously [42]. Nevertheless, no interaction of LST-4His6 or DYN-1His6 with CED-1C or CED-6 was detected in the same GST pull-down assay ( Figure S7C, bottom panel). Together these results indicate that LST-4 and DYN-1 form a complex to regulate phagosome maturation but do not act in complex with CED-1 or CED-6. To further determine the effect of LST-4- DYN-1 interaction on their association with phagosomes, we monitored the dynamic association of DYN-1::GFP with phagosomes in germ lines of wild-type and lst-4(tm2423) animals, and phagosomal association of LST-4::GFP in wild-type and dyn-1(RNAi) germ lines. In the wild type, DYN-1 was initially localized to the periphery of the phagosome and then quickly became enriched to form a large patch-like structure ( Figure 6B, 0-14 min). DYN-1 then became more evenly distributed on the phagosome before forming punctate structures ( Figure 6B, 14-56 min), which likely represent the dissociation of DYN-1 from the phagosome. Unlike in wild-type, DYN-1::GFP neither became sharply enriched nor formed an obvious patch on phagosomes in lst-4(tm2423) mutant germ lines ( Figure 6B), suggesting that loss of lst-4 likely affected the enrichment or stabilization of DYN-1 on phagosomes. These findings are in agreement with the observations made previously by Lu et al. that loss of lst-4 impaired the phagosomal association of DYN-1 during embryonic cell corpse removal [42]. In addition, we noticed that DYN-1::GFP was more enriched on phagosomes when co-expressed with LST-4::mCherry (compare Figure 6A and 6B). On the other hand, RNAi depletion of dyn-1 seemed not to affect the association of LST-4 with phagosomes, because no obvious difference in the dynamic association of LST-4::GFP with phagosomes was observed between dyn-1(RNAi) and control RNAi animals ( Figure 6C). These results, together with the findings made by Lu et al. and Almendinger et al. [36,42], establish that LST-4 promotes phagosomal activity of DYN-1. Importantly, we further found that lst-4(tm2423) mutants expressing DYN-1::GFP (qxIs139) displayed an obvious reduction in germ cell corpses compared with the same mutants without DYN-1::GFP expression ( Figure 6D). For example, lst-4(tm2423) animals expressing DYN-1::GFP (qxIs139) contained 6.360.6 (mean6SEM) and 7.660.6 corpses per gonad arm at adult ages of 36 and 48 h post L4, respectively, compared with 29.960.6 and 55.261.1 in lst-4(tm2423) mutants ( Figure 6D). In contrast, dyn-1 RNAi caused similar levels of germ cell corpse accumulation in both wild type and animals expressing LST-4::GFP (yqIs114) ( Figure 6E). Taken together, these findings provide strong evidence that LST-4 forms a complex with DYN-1 and acts through the latter to promote the initiation of phagosome maturation.

Discussion
During phagocytosis, the phagocytic receptor CED-1 recognizes cell corpses and transduces engulfment signals to the CED-6 adaptor. DYN-1/dynamin was also reported to participate in the ced-1 pathway for corpse engulfment, and likely acts downstream of CED-1 and CED-6 [18]. Nevertheless, it is not clear how these factors coordinate to induce the rearrangement of the actin cytoskeleton, a key event required for cell corpse internalization. Although it was previously proposed that the two engulfment pathways for cytoskeletal reorganization converged on the CED-10 GTPase, the molecular link between the phagocytic receptor CED-1 and CED-10 remains to be identified [44]. In this study, we explored the role of major regulators of CME, a process that internalizes cell surface materials by use of clathrin-coated vesicles, in phagocytosis of apoptotic cells. Our findings revealed that clathrin and the AP2 complex are essential players in the process of cell corpse engulfment. Inactivation of the clathrin heavy chain CHC-1 or individual components of AP2 resulted in accumulation of cell corpses in the C. elegans germ line. Moreover, RNAi of chc-1 or AP2 components significantly enhanced the engulfment defects in ced-2 and ced-5 strong loss-of-function mutants but not mutants deficient in ced-1 and ced-6, suggesting that the chc-1 and AP2 genes likely act within the same genetic pathway as ced-1 and ced-6. Our results demonstrated that CHC-1 and the AP2 complex associate with phagosomes containing cell corpses in an inter-dependent manner and their phagosomal recruitment requires CED-1 and CED-6. Importantly, loss of clathrin or AP2 function severely impaired the rearrangement of the actin cytoskeleton required for corpse engulfment. Altogether these findings provide strong evidence that clathrin and AP2 function downstream of CED-1 and CED-6 and likely mediate the cytoskeletal reorganization required for cell corpse internalization (Figure 8).  Our findings suggest that clathrin and the AP2 complex serve a dual role in the process of apoptotic cell removal (Figure 8). On one hand, clathrin and AP2 are important for cell corpse engulfment by acting downstream of CED-1 and CED-6 to mediate cytoskeletal rearrangement in engulfing cells (Figure 8). This function is likely achieved by forming a protein complex with CED-1 and CED-6. In support of this conclusion, we found that CED-1 indeed forms a complex with CED-6 in vivo. Remarkably, our immunoprecipitation and in vitro GST pull-down results revealed that the CED-6 adaptor protein directly interacts with CHC-1 and individual components of the AP2 complex in C. elegans. Interestingly, we found that the CED-1 receptor likely interacts with AP2 via the a subunit of the latter. Because clathrin can function as an actin organizer at large membrane interfaces that far exceed the size of conventional CCVs [45] and loss of CHC-1 and AP2 function caused defective recruitment of actin around cell corpses, we propose that the formation of a protein complex by CED-1, CED-6, AP2 and CHC-1 provides a hub for recruitment and assembly of actin for cell corpse engulfment. On the other hand, clathrin and the AP2 complex are essential for phagosome maturation following corpse internalization. Loss of CHC-1 and AP2 function abrogated the acidification of phagosomes and inhibited phagosomal recruitment of downstream effectors required for phagosome maturation. In addition, our data demonstrated that LST-4 interacts with DYN-1 to promote its association with phagosomes. Clathrin and AP2 facilitate phagosomal association of the LST-4-DYN-1 complex by interacting with them, thereby promoting the initiation of phagosome maturation ( Figure 8). Notably, whereas AP2 and CHC-1 were found to form complexes with either CED-1-CED-6 or LST-4-DYN-1, no protein interaction of CED-1 or CED-6 with LST-4 or DYN-1 was detected by either co-immunoprecipitation or GST pull-down assays. Thus clathrin and AP2 likely form two types of complex with factors required for engulfment and phagosome maturation, establishing them as a molecular link between engulfment and phagosome maturation in apoptotic cell clearance mediated by the phagocytic receptor CED-1 (Figure 8).
The recruitment of actin around germ cell corpses mediated by the complex of CED-1, CED-6, AP2 and clathrin may resemble the pathway used by mammalian cells to phagocytose pathogens.  35 S-labeled APA-2, APB-1, DPY-23 and His6-tagged CHC-1C were incubated with immobilized GST, GST-CED-9, GST-LST-4 and GST-DYN-1 (3 ug of each). Bound proteins were resolved on sodium dodecyl sulfate polyacrylamide gels and viewed by autoradiography or detected by immunoblotting with His6 antibody. GST and GST-fused proteins used for binding are shown in the right panel. (E and F) mCherry-tagged CHC-1 associated with DYN-1::GFP (E) and LST-4::GFP (F) in animals. IPs were performed with GFP antibody on lysates of animals expressing mCherry::CHC-1 (yqIs98) alone, animals co-expressing mCherry::CHC-1 and DYN-1::GFP (yqIs98;qxIs139), and animals co-expressing mCherry::CHC-1 and LST-4::GFP (yqIs98;yqIs114). Precipitates were detected by immunoblotting with antibodies for mCherry and GFP, respectively. (G and H) mCherry-tagged LST-4 associated with APA-2::GFP (G) and DPY-23 (H) in animals. IPs were performed with GFP antibody on lysates of animals expressing mCherry::LST-4 alone (yqIs119), animals co-expressing mCherry::LST-4 and APA-2::GFP (yqIs119;yqIs99), and animals co-expressing mCherry::LST-4 and DPY-23::GFP (yqIs119;yqIs120). Precipitates were detected by immunoblotting with mCherry and GFP antibodies. doi:10.1371/journal.pgen.1003517.g007 Figure 8. Schematic summary of the role of clathrin and the AP2 complex in both corpse engulfment and phagosome maturation during phagocytosis of apoptotic cells. In the cell corpse engulfment phase, clathrin and AP2 act downstream of CED-1 and CED-6 to promote actin rearrangement, which is required for phagocytosis. In the phagosome maturation phase, clathrin and AP2 promote phagosomal association of LST-4 and DYN-1, which initiates the maturation process. doi:10.1371/journal.pgen.1003517.g008 In mammalian cells, clathrin and some other regulators of CME are found to be essential for invasion of pathogenic bacteria, fungi and large viruses [34,[46][47][48][49][50][51]. For example, clathrin and dynamin were found to localize to bacterial entry foci during the invasion of Listeria monocytogenes and inactivation of major regulators of CME, such as Grb2, EPS15, CIN85 and CD2AP, severely inhibited bacterial internalization [34]. Further studies revealed that upon bacterial infection, the clathrin heavy chain CHC undergoes Srcdependent phosphorylation, which in turn initiates the accumulation of clathrin coats at bacterial adhesion sites. Through interaction of the clathrin light chain CLC with the actininteracting protein Hip1R, actin is recruited and assembled at bacteria-host adhesion sites, leading to bacterial internalization. Thus the clathrin-coated pits that accumulate at bacterial entry sites serve as platforms for the actin polymerization needed for phagocytosis [51]. Intriguingly, the clathrin adaptor Dab2, but not AP2, is critical for clathrin recruitment to L. monocytogenes entry sites [34]. In C. elegans, our findings indicate that clathrin is similarly required for the actin rearrangement needed for phagocytosis of apoptotic cells. Nevertheless, RNAi depletion of hipr-1 and clic-1, which encode C. elegans homologs of mammalian HipR1 and clathrin light chain, respectively, did not induce a similar level of corpse accumulation to that caused by chc-1 RNAi (Table S1). We also performed RNAi to deplete several other putative actinbinding proteins predicted by the STRING protein interaction prediction program (http://string-db.org/) but failed to detect an obvious accumulation of germ cell corpses (data not shown). Thus it is possible that multiple factors may function redundantly to mediate the recruitment of actin by the CED-1-CED-6-AP2clathrin complex. Further studies will be necessary to unveil the underlying mechanism. In addition, unlike clathrin recruitment during bacterial phagocytosis by mammalian cells, phagosomal recruitment of clathrin requires the AP2 complex in C. elegans. The requirement for different adaptors may be attributed to the use of different receptors for engulfment of apoptotic cells and bacteria. Besides, as the sizes of cell corpses in C. elegans are normally $1 mm, which is much larger than endocytic CCVs (,200 nm), it also remains to be determined how clathrin is assembled (i.e, clathrin per se, clathrin-coated vesicles, or clathrin-coated pits) when it associates with phagosomes. Moreover, the engulfment of cell corpses in C. elegans appears to involve fewer CME regulators compared with mammalian phagocytosis of pathogens. In our unbiased RNAi screen of C. elegans CME regulators, we found that only CHC-1 and AP2 components obviously affected cell corpse engulfment and degradation while LST-4/Snx9/18/33 and DYN-1/dynamin were essential for phagosome maturation; in contrast, RNAi inactivation of several major CME regulators has been shown to inhibit bacterial infection of mammalian cells. Thus, whereas both cell corpse engulfment in C. elegans and pathogen invasion in mammals make use of clathrin for actin rearrangement, other factors may differ owing to the requirement of distinct signaling mechanisms.
Remarkably, MEGF10, the mammalian ortholog of CED-1, was reported to interact with the m2 subunit of AP2 in a yeast 2hybrid screen, and the existence of a protein complex containing MEGF10 and AP2 subunits was further confirmed by a protein purification assay [52,53]. More recently, Drosophila Ced-6 was identified as a clathrin-associated sorting protein (CLASP) as it binds to clathrin and AP2 via the C-terminal region [54]. Furthermore, the phosphotyrosine-binding domain (PTB domain) of Drosophila Ced-6 specifically recognizes a noncanonical sorting signal in the vitellogenin receptor Yolkless. Thus Ced-6 participates in clathrin-mediated yolk uptake in Drosophila egg chambers [54]. In addition, the mammalian homolog of CED-6, Gulp, can also interact with both clathrin and AP2 [54,55]. In our study we found that clathrin and AP2 act in phagocytic receptor-mediated cell corpse removal by forming a protein interaction cascade with CED-1 and CED-6 to regulate the actin rearrangement required for engulfment and with LST-4 and DYN-1 to promote phagosome maturation needed for corpse degradation. Given that the major factors for apoptotic cell engulfment are evolutionarily conserved and the interactions of clathrin and AP2 with CED-6 and/or CED-1 similarly exist in C. elegans, Drosophila and mammals, our discovery that clathrin and AP2 play an essential role in removal of apoptotic cells suggests that the non-classical function of clathrin and its adaptor proteins in phagocytosis is likely conserved across diverse species.

Plasmid construction
The P chc-1 chc-1::gfp construct was generated by cloning a genomic DNA fragment containing a promoter region of 3 kb and the open reading frame (ORF) of the chc-1 gene in frame with GFP into the pPD95.77 vector. The P apa-2 apa-2::gfp construct was similarly generated by cloning a genomic fragment containing a promoter region of 2 kb and the ORF of apa-2. Genomic DNA containing the ORF of chc-1 was amplified and inserted into P ced-1 mCherry1 or P ced-1 gfp1 via the KpnI site to generate the P ced-1 mCherry::chc-1 and P ced-1 gfp::chc-1 constructs. To generate P ced-1 apa-2::gfp, a genomic fragment containing the apa-2 ORF was amplified and inserted into P ced-1 gfp3 via the KpnI site. To construct P ced-1 gfp::Moesin, the C-terminal of Moesin was amplified from the plasmid pJWZ6 [58] (provided by Dr. David R. Sherwood, Duke University) and inserted into P ced-1 gfp1 via the KpnI site. To generate P lst-4 lst-4(cDNA)::gfp, a DNA fragment containing a 2 kb promoter region and the first intron followed by the remaining cDNA sequence of the lst-4 isoform c was inserted between the HindIII and KpnI sites of the vector pPD95.77. P lst-4 lst-4(cDNA)::mCherry was derived from P lst-4 lst-4(cDNA)::gfp by replacing gfp with mCherry. To generate P lst-4 lst-4(gDNA)::gfp, a genomic fragment containing a 2 kb promoter region and the lst-4 genomic ORF were amplified and inserted between the XbaI and XmaI sites of the vector pPD95.77.

RNAi experiments
RNAi experiments were performed by using bacterial feeding assays as described previously [59]. In most cases, L4-stage animals were transferred to plates seeded with bacteria expressing either control double-stranded RNA (dsRNA) (L4440 empty vector) (Control RNAi) or dsRNA corresponding to the open reading frames of genes of interest. RNAi of apa-2 with its 39 UTR was performed by feeding animals with bacteria expressing dsRNA corresponding to the 39UTR of 516 bp. Germ cell corpses and other phenotypes were observed in adults of the progeny. For RNAi of chc-1, apb-1 and dyn-1, which may cause embryonic lethality in the progeny, L3-to L4-stage animals were transferred to plates seeded with bacteria expressing dsRNA of individual genes and phenotypes were observed in adults of the same generation.

Quantification of cell corpses
Cell corpses in synchronized animals were scored under Nomarski optics. To quantify germ cell corpses, cell corpses in the germline meiotic region of one gonad arm in each of at least 15 animals were scored at various adult ages (12,24,36,48 and 60 h after the L4 stage). The average numbers of germ cell corpses from one gonad arm were calculated for each adult age. Data derived from different genetic backgrounds were compared using unpaired t-tests. For cell corpse analysis of chc-1(b1025ts) mutants, animals were grown to L4 at 20uC and then shifted to 25uC. Germ cell corpses were scored at 12, 24, 36 and 48 h after the shift.

Immunofluorescence microscopy
To quantify the percentage of germ cell corpses labeled by various phagosomal markers, adult animals at 36 h after the L4 molt were mounted on agar pads in M9 buffer (1 litre contains: 3 g KH 2 PO 4 , 6 g Na 2 HPO 4 , 5 g NaCl, 1 mM MgSO 4 ) with 2 mM levamisole and then examined by fluorescence microscopy. To analyze the labeling of germ cell corpse by phagosomal markers in dyn-1(ky51ts) mutants, animals were grown to L4 stage at 20uC and then shifted to 25uC; cell corpses were analyzed 24 h after the shift. To view germ cell corpses in dyn-1 RNAi-treated animals, L4 larvae were cultured on RNAi plates and germ cell corpses were analyzed 24 h after the L4 molt.

Time-lapse analysis of cell corpses
To measure the duration of germ cell corpses, animals were mounted in M9 buffer containing 2 mM levamisole, sealed with beeswax and Vaseline (1:1), and recorded under Nomarski optics at 20uC. The gonadal region was recorded every 1 min at 1 mm/ section for 20 Z-sections. Images were captured using a Zeiss Axioimager M1 coupled with an AxioCam monochrome digital camera and Axiovision rel. 4.7 software. Animals were constantly examined for viability during recording.

Supporting Information
Table S1 Cell corpse phenotype caused by RNAi of C. elegans genes involved in clathrin-mediated endocytosis. C. elegans genes involved in clathrin-mediated endocytosis were identified by using sequences of individual human proteins to search for homologs in the C. elegans genome database. RNAi was performed as described in Methods. Germ cell corpses in one gonad arm of each animal were scored for at least 15 animals 60 h after the L4 stage. N/A indicates that hsp-1 RNAi caused defects in germline proliferation and no cell corpses could be scored. (DOC)