Retromer Is Required for Apoptotic Cell Clearance by Phagocytic Receptor Recycling

See allHide authors and affiliations

Science  05 Mar 2010:
Vol. 327, Issue 5970, pp. 1261-1264
DOI: 10.1126/science.1184840


The cell surface receptor CED-1 mediates apoptotic cell recognition by phagocytic cells, enabling cell corpse clearance in Caenorhabditis elegans. Here, we found that the C. elegans intracellular protein sorting complex, retromer, was required for cell corpse clearance by mediating the recycling of CED-1. Retromer was recruited to the surfaces of phagosomes containing cell corpses, and its loss of function caused defective cell corpse removal. The retromer probably acted through direct interaction with CED-1 in the cell corpse recognition pathway. In the absence of retromer function, CED-1 associated with lysosomes and failed to recycle from phagosomes and cytosol to the plasma membrane. Thus, retromer is an essential mediator of apoptotic cell clearance by regulating phagocytic receptor(s) during cell corpse engulfment.

In Caenorhabditis elegans, cell corpse engulfment is controlled by two parallel pathways, one that recognizes and transduces engulfing signals, and the other that induces cytoskeleton reorganization (1). However, how components of these pathways are regulated and what other factors are involved remain unclear. To identify additional regulators of these pathways we performed genome-wide and candidate-based RNA interference (RNAi) screens (2) for genes whose inactivation greatly increased cell corpse numbers in the C. elegans germ line. Three genes, snx-1, snx-6, and lst-4, encoding homologs of mammalian sorting nexins 1/2, 5/6, and 9/18/33, respectively, were identified (figs. S1A and S2A and table S1). In mammals, sorting nexins 1/2 and 5/6 are essential components of the intracellular protein sorting complex retromer (35), whereas sorting nexins 9/18/33 regulate endocytosis (3). The deletion mutants snx-1(tm847) (6) and snx-6(tm3790), which probably represent strong loss-of-function mutations of snx-1 and snx-6 (figs. S1B and S2B), had increased numbers of germline and embryonic cell corpses, which persisted significantly longer than did the wild type (Fig. 1, A and B, and fig. S3). The number of cell deaths in snx-1(tm847) and snx-6(tm3790) embryos was indistinguishable from that of the wild type; thus, the increased cell corpse numbers result from defective corpse clearance rather than excessive apoptosis.

Fig. 1

C. elegans retromer affects clearance of apoptotic cells. (A) Quantification of germ cell corpses in N2 (wild type), snx-1(tm847), snx-6(tm3790), and snx-6(tm3790);snx-1(tm847) mutants. (B) Four-dimensional microscopy analysis of germ cell corpse duration in N2, snx-1(tm847), and snx-6(tm3790) mutants. Thirty germ cell corpses were recorded for each strain. (C) Quantification of germ cell corpses in vps-26(RNAi), vps-35(hu68), and vps-29(tm1320) animals. (D) Germ cell corpse phenotypes of control RNAi- and vps-29 RNAi-treated N2 and snx-1(tm847) mutants. In (A), (C), and (D), the y axis indicates the average number of germ cell corpses. Error bars represent the SEM. Comparisons were performed using unpaired t tests. **P < 0.001.

The retromer mediates retrograde transport of transmembrane cargoes from endosomes to the trans-Golgi network (4). SNX-1 and SNX-6 are essential subunits of the retromer sorting nexin dimer (4, 5). In snx-6(tm3790);snx-1(tm847) double mutants, germline and embryonic cell corpse numbers were indistinguishable from those of single mutants (Fig. 1A and fig. S3A), suggesting that these genes function in the same genetic pathway. In contrast, lst-4(tm2423);snx-1(tm847) double mutants exhibited significantly more germline and embryonic cell corpses than did single mutants (fig. S4). Thus, LST-4 probably acts separately from the retromer during cell corpse clearance. The retromer also contains a Vps26-Vps29-Vps35 trimer (4), so we examined vps-26(tm1523), vps-29(tm1320), and vps-35(hu68) deletion mutants (6). No corpse phenotype was observed in vps-26(tm1523) germ lines, probably because of abnormal germline development (6, 7). Partial RNAi inactivation of vps-26, however, significantly increased germ cell corpses (Fig. 1C). A similar increase was observed in vps-29(tm1320) and vps-35(hu68) germ lines (Fig. 1C). Intriguingly, cell corpse numbers in vps-26(tm1523), vps-29(tm1320), and vps-35(hu68) embryos were similar to those of the wild type. Thus, loss of individual retromer components affects corpse clearance to different extents. We next asked whether the VPS-26–VPS-29–VPS-35 and sorting nexin subcomplexes acted together in cell corpse removal. vps-29 RNAi on its own caused an increase in germ cell corpses, but did not further enhance the number of corpses in snx-1(tm847) germ lines (Fig. 1D), suggesting that snx-1 and vps-29 probably function in the same genetic pathway. Furthermore, in glutathione-S-transferase (GST) pull-down assays, SNX-1 and SNX-6 interacted with one another, and with GST-VPS-26, -29, and -35 (fig. S5). Thus, retromer components function together during cell corpse clearance in C. elegans.

We next examined germ cell corpse engulfment by transmission electron microscopy. In wild-type animals treated with gla-3 RNAi, which increases germ cell deaths without affecting corpse clearance (8), 10 cell corpses from 4 gonad arms were fully encircled by gonadal sheath cells (Fig. 2), consistent with previous findings that wild-type germ cell corpses are swiftly engulfed (9). Of 31 cell corpses from 2 gonad arms in snx-1(tm847) mutants, 25 (81%) were only partially encircled by sheath cells, while 6 (19%) were completely engulfed but not properly degraded [(Fig. 2, fig. S6, and supporting online material SOM) text]. Furthermore, engulfing cells required SNX-1/retromer activity for cell corpse removal, and SNX-1 and VPS-29 were recruited and colocalized on the cell corpse surface (SOM text and figs. S7 and S8). Thus, SNX-1/retromer is required for engulfment and degradation of cell corpses.

Fig. 2

Retromer functions in the ced-1/6/7 pathway. (A to F) Transmission electron micrographs of engulfed germ cell corpses in an N2 animal treated with gla-3 RNAi (A to C) and in snx-1(tm847) mutants (D to F). Traces of membranes are shown in (A) and (D). Boxed regions in (B) and (E) are magnified in (C) and (F), respectively. Black arrows indicate gonadal sheath cell membranes. Black and white arrowheads indicate cell corpse membranes and germline syncytium membranes, respectively. (G to J) Quantification of germ cell corpses in double mutants of snx-1(tm847) with ced-1(e1735) (G), ced-7(n1892) (H), ced-2(n1994) (I), and ced-5(n1812) (J) mutants. Data derived from two different genetic backgrounds at multiple stages were compared by unpaired t tests. *P < 0.05, **P < 0.001.

To determine in which pathway retromer acted, we examined double mutants of snx-1(tm847) with strong loss-of-function alleles of engulfment genes. In C. elegans, ced-1, ced-6 (GULP), ced-7 (ABC transporter), and dyn-1 (Dynamin) function in one pathway to recognize and transduce engulfing signals, whereas ced-2 (CrkII), ced-5 (Dock 180), ced-10 (Rac guanosine triphosphatase), ced-12 (ELMO), and psr-1 (phosphatidylserine receptor) act in the other pathway to activate cytoskeleton rearrangement (1, 10). Germline and embryonic cell corpse numbers in double mutants of snx-1(tm847) with ced-1(e1735) or ced-6(n2095) were indistinguishable from those of ced-1(e1735) and ced-6(n2095) single mutants (Fig. 2G and fig. S9, A to C). Intriguingly, both germline and embryonic cell corpses in ced-7(n1892);snx-1(tm847) double mutants were significantly increased compared to ced-7(n1892) single mutants (Fig. 2H and fig. S9D). Thus, snx-1/retromer probably acts in parallel to ced-7 in the same pathway as ced-1 and ced-6. Consistent with this, snx-1(tm847) significantly enhanced germline and embryonic cell corpse numbers in mutants affecting the other engulfment pathway (Fig. 2, I and J, and fig. S9, E to H).

During phagocytosis, the receptor CED-1 clusters in the phagocytic cup before quickly encircling the cell corpse. Recognition of apoptotic cells by CED-1 requires CED-7, loss of which disrupts clustering of CED-1 around embryonic cell corpses (11). Because retromer and CED-7 act in parallel in the ced-1 pathway, we investigated whether retromer regulates CED-1 by time-lapse chasing the association of CED-1::GFP (green fluorescent protein) expressed from an integrated array (smIs34) with cell corpses in snx-1(tm847) embryos, while simultaneously monitoring phagosomal recruitment of lysosomes labeled by the lysosome marker LMP-1::mCherry. In wild-type embryos, CED-1::GFP was quickly recruited to the phagocytic cup and encircled the cell corpse within 3 min (Fig. 3A). Between 3 and 9 min, CED-1 disappeared from the phagosome and reappeared on the plasma membrane of the engulfing cell. As the CED-1 ring formed, lysosomes were simultaneously recruited to the phagosome as indicated by formation of a LMP-1::mCherry ring, which remained associated with the phagosome after CED-1 was released (Fig. 3A). Thus, CED-1 is recycled swiftly from the phagosome to the engulfing cell membrane before the cell corpse is degraded. The mean duration of the CED-1 ring on the phagosome was 10.8 ±1.7 min (n = 6) in wild-type embryos, but in snx-1(tm847);smIs34 embryos it was 26.0 ± 1.8 min (n = 6) so that the ring remained associated with the phagolysosome until late in corpse degradation (Fig. 3A). Subsequently, when the corpses had adopted a pit-like structure, the proportion colabeled with CED-1::GFP and LMP-1::mCherry was <20% in wild-type embryos but >80% in snx-1(tm847) embryos (Fig. 3B). Thus, loss of retromer inhibits CED-1 recycling from the phagosome to the engulfing cell membrane, and phagosome-associated CED-1 is sent to lysosomes together with cell corpses. Furthermore, whereas smIs34-expressed CED-1::GFP was almost exclusively associated with the plasma membrane in wild-type embryos, in snx-1(tm847) embryos it displayed an intracellular punctate distribution pattern that colocalized with LMP-1::mCherry in addition to its membrane localization in several cell types such as hypodermal cells (Fig. 3C). Thus, CED- 1 is internalized from and recycled back to the cell membrane; loss of retromer function causes lysosomal accumulation of CED-1. Consistent with this, CED-1::GFP expression driven by the vps-33 promoter in macrophage-like coelomocytes was intracellular and partially overlapped with the lysosome marker mCherry::CUP-5 in snx-1(tm847) and snx-6(tm3790) mutants, whereas in the wild type it was mainly seen on the plasma membrane (SOM text and fig. S10A).

Fig. 3

snx-1(tm847) causes defective recycling of internalized CED-1. (A) Time-lapse chasing of CED-1::GFP (arrows) and LMP-1::mCherry (sharp arrowheads) on phagosomes in N2 and snx-1(tm847) animals. Clustering of CED-1::GFP on the phagocytic cup was set as 0 min. Blunt arrowheads indicate membranes of engulfing cells. Bars, 5 μm. (B) Representative images (top) and quantification (bottom) of CED-1::GFP and LMP-1::mCherry colocalization on cell corpses at a late stage of degradation in N2 and snx-1(tm847) embryos. At least 100 cell corpses in each strain were examined. Arrows indicate cell corpses. DIC, differential interference contrast. Bars, 5 μm. (C) Localization of CED-1::GFP and LMP-1::mCherry in hypodermal cells in N2 and snx-1(tm847) embryos. Bars, 10 μm.

The increased association of CED-1 with lysosomes suggests that CED-1 undergoes lysosome-mediated degradation. CED-1::GFP signal from smIs34 was reduced in snx-1(tm847) mutants compared to the wild type (fig. S10B). CED-1::GFP protein abundance was significantly lower in snx-1(tm847);smIs34 and snx-6(tm3790);smIs34 animals (Fig. 4A). CED-1::GFP was also reduced in vps-29(tm1320);smIs34 worms, though to a lesser extent. In contrast, CED-1::GFP was not reduced in lst-4(tm2423);smIs34 animals (Fig. 4A). Using a CED-1 C-terminal–specific antibody, we found that endogenous CED-1 was strongly reduced in snx-1(tm847) and snx-6(tm3790) animals, slightly reduced in vps-29(tm1320) mutants, and undetectable in ced-1(e1735) mutants containing an early stop codon in the ced-1 gene (Fig. 4B). Furthermore, endogenous CED-1 localized to the plasma membrane in wild-type embryos but not in snx-1(tm847) or ced-1(e1735) mutants (Fig. 4C). Thus, in the absence of retromer function, CED-1 fails to localize to the plasma membrane and is probably degraded in lysosomes, becoming limiting for cell corpse engulfment and causing accumulation of cell corpses. Consistent with this, RNAi knockdown of vps-37, a component of the ESCRT-I complex (endosomal sorting complex required for transport) (12), partially restored CED-1::GFP abundance in snx-1(tm847);smIs34 animals (fig. S11). Moreover, the increased germline and embryonic corpse numbers in snx-1(tm847) mutants were strongly reduced by overexpressing CED-1::GFP driven by the ced-1 promoter (smIs34) (Fig. 4D and fig. S12A). CED-1::GFP driven by the sheath cell-specific lim-7 promoter (bcIs39) similarly decreased the number of snx-1(tm847) germline corpses (Fig. 4D). Furthermore, smIs34 greatly reduced the numbers of both embryonic and germline cell corpses in snx-6(tm3790) single mutants as well as in snx-6(tm3790);snx-1(tm847) double mutants, but not in lst-4(tm2423) mutants (fig. S12, B to G), indicating that CED-1 overexpression specifically rescued the cell corpse phenotypes of retromer mutants and that retromer acts through CED-1 to affect cell corpse clearance. Finally, using the CED-1 C-terminal–specific antibody, we immunoprecipitated CED-1::GFP from cell lysates from wild-type animals expressing both CED-1::GFP and SNX-1::mCherry and found that SNX-1::mCherry was associated with CED-1::GFP, whereas no SNX-1::mCherry was coimmunoprecipitated from lysates of ced-1(e1735) worms expressing SNX-1::mCherry alone (Fig. 4E). Thus, CED-1 associates with retromer in C. elegans. In a pull-down assay, purified SNX-1 interacted with GST-fused CED-1 N terminus (CED-1N, amino acids 1 to 908) and C terminus (CED-1C, amino acids 933 to 1111), and SNX-6 interacted with GST-CED-1N (Fig. 4F). No obvious interaction was detected between CED-1 and VPS-35 (Fig. 4F). Thus, CED-1 cycling is probably achieved by direct interaction with specific retromer subunits.

Fig. 4

Retromer acts through CED-1. (A and B) Immunoblot analysis of CED-1::GFP expression by smIs34 (Pced-1ced-1::gfp) (A) and endogenous CED-1 protein abundance (B) in N2 and different mutants. The relative expression of CED-1 in each strain is indicated. (C) Localization of endogenous CED-1 in N2, ced-1(e1735), and snx-1(tm847) embryos. Merged images of CED-1–specific antibody staining (green) and nuclear staining (blue) of a four-cell-stage embryo (left) and a ~50-cell-stage embryo (right) are shown. Bars, 10 μm. (D) Quantification of germ cell corpses in N2, snx-1(tm847), snx-1(tm847);smIs34, and snx-1(tm847);bcIs39 (Plim-7ced-1::gfp) animals. **P < 0.001. (E) Immunoprecipitations (IP) were performed with CED-1–specific antibody from lysates of ced-1(e1735) mutants expressing SNX-1::mCherry (lane 1) and N2 worms expressing both CED-1::GFP and SNX-1::mCherry (lane 2). Proteins were detected with antibodies against GFP and mCherry, respectively. (F) His6-tagged SNX-1, SNX-6, and VPS-35 were incubated with immobilized GST, GST-CED-1C, and GST-CED-1N. Bound proteins were detected with antibody to His6 (left). GST fusion proteins used for binding are shown on the right.

The retromer mediates a wide range of processes (4), including transport of intracellular sorting receptors such as Vps10 in yeast and CI-MPR (cation-independent mannose 6-phosphate receptor) in mammals (13, 14), formation of Wnt gradients in C. elegans and Drosophila (6, 7, 1519), and transcytosis of the polymeric immunoglobulin receptor in polarized epithelial cells (20). Retromer-mediated cycling of receptors is probably achieved by interaction of cargoes with different retromer subunits (21) or between cargoes and a retromer partner like Grd19/Snx3 (22). Our findings establish a function of retromer in apoptotic cell clearance in mediating cycling of the phagocytic receptor CED-1 between the plasma membrane and intracellular organelles. CED-1 family phagocytic receptors may be similarly regulated in other organisms.

Supporting Online Material

Materials and Methods

SOM Text

Figs. S1 to S12

Table S1


  • * These authors contributed equally to this work.

References and Notes

  1. We thank S. Mitani, X. Wang, D. Xue, H. Fares, and C. elegans Genetic Center (CGC) for strains; X. Wang and H. Zhang for critically reading the manuscript; and I. Hanson for proofreading services. This research was supported by grants 2007CB947201, 30871266, and KSCX1-YW-R-70 from the Chinese government. C.Y. is supported by the Chinese Academy of Sciences 100-Talents Program.

Stay Connected to Science

Navigate This Article