Introduction

Billions of apoptotic cells are produced daily in multicellular organisms, which are efficiently removed by a highly conserved process called cell corpse clearance (efferocytosis) [1, 2]. Efferocytosis is essential for development, tissue homeostasis, and mitigation of inflammatory responses under physiological conditions [3], and depends on recognition of specific surface ligands, termed “eat me” signals, that distinguish apoptotic from healthy cells [4]. Phosphatidylserine (PS), the most well-known and conserved “eat me” signal, is exposed on the outer surface of apoptotic cells and binds to macrophage PS receptors to trigger engulfment of apoptotic cells [5]. In healthy cells, PS primarily localizes to the inner surface rather than the outer surface of the PM [6], and the exposure of PS on the cell surface results from scramblase activation. Apoptosis signals induce caspase-mediated cleavage of scramblases, especially Xkr8, allowing apoptotic PS externalization from the inner to outer PM leaflets to initiate efferocytosis [6]. The ubiquitously expressed Xkr8 localizes to the PM through heterodimerization and chaperoning by two type I membrane proteins, Basigin (BSG) and Neuroplastin (NPTN), which form a complex following caspase cleavage to promote apoptotic PS exposure [7]. Xkr8 deficiency has been linked to systemic lupus erythematosus (SLE)-type autoimmune diseases and male infertility[8]. However, whether Xkr8 associates with other proteins to regulate PS exposure during apoptosis remains unclear.

PS is mainly synthesized in the endoplasmic reticulum (ER) and therefore requires transport to other membrane-bound cellular compartments [9]. Lipids can be transferred from the ER to other organelles via vesicular transport along the secretory and endocytic pathways [10, 11]. However, such transport alone cannot account for the compositional heterogeneity of different membranes. Mounting evidence suggests that non-vesicular transport provides an alternative, faster transport mechanism, particularly for the PM, which rapidly changes composition in response to cellular status [6, 12, 13]. Non-vesicular transport of PS from ER to PM is mediated by oxysterol-binding protein (OSBP) and OSBP-related proteins (ORPs) located at ER-PM contact sites, which recruit various protein players to promote lipid exchange between ER and PM, which is essential for many physiological processes [14, 15]. In particular, ORP5 and ORP8 have recently been shown to mediate countertransport of phosphatidylinositol 4- phosphate (PI4P) and PS between the ER and PM, transferring PS from the ER to the PM in concert with PI4P from the PM to the ER, in a PI4P gradient-dependent manner. Importantly, PI4P delivered to the ER is immediately dephosphorylated by the ER- resident phosphatase, suppressor of actin mutations 1-like (Sac1), to maintain an extremely low PI4P gradient level in the ER [16].

In Drosophila melanogaster, apoptotic cells are removed during efferocytosis by nonprofessional phagocytes (e.g., epithelial cells) or professional phagocytes (i.e., macrophages and glial cells), thereby exposing PS and recruiting Drosophila macrophages via a mechanism similar to that in mammals [17]. In Drosophila, PS is exposed on the surface of degenerating or pruned neurons and recognized by receptors on the surface of phagocytes, such as Draper and SIMU [18, 19], thereby mediating the clearance of neuronal fragments. Meanwhile, some bridging molecules, such as Onion and NimB1, also positively or negatively regulate the recognition and phagocytosis of apoptotic cells via binding to PS [20, 21]. Exposed PS can be detected on the cell surface of damaged or apoptotic cells in flies, indicating that PS exposure is specifically regulated by apoptotic signals and is crucial for subsequent efferocytosis [22–24]. Because the Drosophila homolog of Xkr8, Xkr, was considered to lack a caspase recognition site compared with the mouse Xkr8 and C. elegans CED-8 amino acid sequences [25], this has raised a long-standing question as to whether Xkr regulates PS exposure as a phospholipid scramblase during apoptosis in Drosophila. However, a recent study showed that Xkr can be constitutively activated to disrupt the asymmetric distribution of phospholipids, thereby altering the deformability of cell membranes [26], suggesting that Xkr may have participated in apoptotic PS exposure or the following efferocytosis.

In this study, we confirmed that Xkr exhibited scramblase activity to mediate PS exposure during apoptosis; Xkr deficiency led to a significant reduction in PS exposure and efferocytosis both in Drosophila S2 cells and embryos. Through mass spectrometry screening and immunoprecipitation, we identified the transmembrane protein, TM9SF4, and the ER phospholipase Sac1, which regulated the PS distribution; disordered PS distribution also caused reduced PS exposure and inefficient efferocytosis. Additionally, we found that the Drosophila ORP, dORP9 (CG1513), directly interacted with Xkr via its pleckstrin homology domain, regulating apoptotic PS exposure in Drosophila S2 cells, while Xkr accumulated specifically at ER-PM junctions. Our results demonstrated that Xkr cooperated with dORP9 to mediate non-vesicular PS transport, thus exhibiting PS externalization to mediate efferocytosis in flies. Our work uncovers a new caspase-independent apoptosis response mechanism mediated by the phospholipid scramblase Xkr, which regulates non-vesicular PS transport from the ER to the PM, responsible for efficient PS exposure and apoptotic cell clearance conserved from insects to mammals.

Results

Xkr exhibits scramblase activity, participating in PS exposure and apoptotic cell clearance in vitro and in vivo

Activation of Xkr8 in mammals and CED-8 in Caenorhabditis elegans both require a caspase-3 cleavage site to expose apoptotic PS. However, in Drosophila, Xkr lacks both N- and C-terminal domain caspase recognition motifs (Figure S1A) [27]. To investigate how Xkr mediates scramblase activity, we first expressed N- and C-terminal fusion reporters, GFP-Xkr and Xkr-GFP, in Drosophila S2 cells, wherein Xkr primarily localized to the plasma membrane with a lesser proportion also present in the cytoplasm (Figures 1A and 1B). To test whether Xkr could indeed expose PS during apoptosis, we treated S2 cells with xkr-RNAi or control-RNAi for staining with FITC-labeled annexin V before inducing apoptosis. Subsequent flow cytometry analysis showed that xkr knockdown significantly reduced the levels of exposed PS (Figure 1C).

Figure 1.

To define the precise role of Xkr in PS exposure during apoptosis, we used CRISPR/Cas9 to generate a mutant S2 cell and mutant flies harboring an early translation termination codon in Xkr (Figures 1D and 1E). Xkr antibody immunofluorescence staining confirmed that wild-type (WT) Xkr primarily localized on the plasma membrane, whereas no Xkr signal was detected in xkr^ko S2 cells (Figure S1B). Flow cytometry analysis indicated that the ability of xkr^ko cells to expose PS on the outer plasma membrane was significantly reduced. At the same time, re-expression of Xkr in the mutant restored the PS exposure (Figure 1F), thus demonstrating that Xkr was important for apoptotic PS exposure in S2 cells. To further assess Xkr scramblase activity during apoptosis, WT and xkr mutant cells were treated with a fluorescently labeled phospholipid analog, NBD-PC, which was internalized after inducing apoptosis. NBD-PC levels on the inner leaflet were significantly lower in xkr^ko cells (Figure 1G), showing that scramblase activity was suppressed in the absence of functional Xkr. These results indicate that Xkr facilitates phospholipid scrambling in the plasma membrane and mediates PS exposure during apoptosis, thus supporting a conserved role of Xkr in PS exposure between insects and mammals. As apoptotic cells are recognized and cleared by macrophages in a PS-dependent manner, we hypothesized that loss of scramblase function in S2 cells could also impair the recognition and clearance of apoptotic cells (ACs). To test this possibility, we applied Cell Tracker™ Red CMTPX to label both WT and xkr^ko cells, induced apoptosis by exposing them to actinomycin D (AcD) for 1 h, and then added them to non-apoptotic WT S2 cells to initiate efferocytosis. Confocal microscopy showed that a single phagocyte contained approximately 3.8 CMTPX-labeled ACs in assays with apoptotic S2 cells as prey. In contrast, only 1.8 CMTPX-positive ACs were detected on average using xkr^ko cells as prey (Figures 1H and 1I, and figures S1C and S1D).

In Drosophila reared at 25°C, ACs begin to appear at stage 11 of embryogenesis[28], followed by macrophage engulfment originating in the cephalic mesoderm and spreading throughout the embryo. To further investigate the role of Xkr in cell corpse engulfment in Drosophila, we generated a xkr^ko fly line using CRISPR/Cas9 (Figures 1D and 1E). We confirmed the null allele mutant status by immunofluorescence staining (Figures S1E and S1F). We then employed the phagocytic index (PI), i.e., number of apoptotic particles per macrophage, to quantify macrophages’ capacity to engulf apoptotic bodies [29, 30]. We found no significant differences in the number of macrophages at stage 13 between WT and xkr^ko flies (Figures 1J and 1K), indicating that the embryo development was unaffected by Xkr deletion. However, we observed a higher number of unengulfed ACs in the xkr^ko mutant. In comparison, the number of engulfed ACs per macrophage was approximately 50% lower in xkr^ko flies compared to that in WT (average ACs/macrophage: 1.6 vs 2.6, xkr^ko vs WT; Figures 1J to 1L). In Drosophila, the hemolectin (Hml) protein is specifically expressed in embryonic and larval hemocytes, particularly in plasmatocytes and crystal cells, making it a suitable marker for labeling macrophages [31]. To further confirm that Xkr-mediated PS exposure in apoptosis promoted macrophage recruitment and corpse recognition, we used an Hml-GAL4 driven UAS-GFP (Hml>GFP) reporter fly strain for labeling macrophages. Fluorescence microscopy revealed that ACs in xkr^ko embryos were generally unengulfed or found at low levels within macrophages. In contrast, the majority of ACs were engulfed by GFP-positive macrophages in control Hml>GFP embryos (Figures 1M to 1O). Taken together, these results demonstrated that Xkr-mediated PS exposure plays a prominent role in the recognition and clearance of ACs by macrophages.

TM9Sf4 is identified to interact with Xkr to regulate PS distribution for exposure in apoptosis

Since the above result indicates that Xkr regulates apoptotic PS exposure in a caspaseindependent manner, it remains unclear whether Xkr activity is specifically activated by apoptotic signals. We detected the mRNA and protein level of Xkr after apoptosis induction, and found that Xkr protein level increased after apoptosis induction, whereas xkr mRNA was unaffected (figures S2A to S2C). And, we conducted subcellular colocalization assays by expressing the Xkr-mCherry reporter in a panel of S2 cell lines with various GFP-labeled organelles. These assays revealed that Xkr could colocalize with, i.e., physically contact, membranous organelles such as endosomes or the ER (Figures S2D to S2H). To explore the function of cytoplasmic Xkr and its regulatory mechanism in efferocytosis, we performed immunoprecipitation-mass spectrometry proteomics (IP-MS) to screen the candidate proteins interacting with Xkr (Figure 2A), and identified 81 putative Xkr-interacting proteins (Data S1). Considering the properties and distribution of Xkr in cells, we selected the transmembrane 9 family protein, TM9SF4, which reportedly contributes to the regulation of innate immunity mediated by hemocyte phagocytosis in Drosophila, for closer examination [32] (Figure S3A). Subsequent Co-IP assays indicated that Xkr interacted with TM9SF4 in S2 cells (Figure 2B), and the co-expression of Xkr-GFP with TM9SF4-mCherry in S2 cells indicated that they mainly colocalized in the plasma membrane, with a proportion also co-localized in the cytoplasm (Figure 2C). Yeast two-hybrid (Y2H) assays and bimolecular fluorescence complementation (BiFC) assays in HEK 293T cells revealed that Xkr likely directly interacted with TM9SF4 (Figure 2D; figure S3B), suggesting that TM9SF4 may chaperone Xkr to the plasma membrane, which aligns with its reported role in other protein interactions[33].

Figure 2.

TM9SF4 encodes a nonaspanin-type protein, which has been reported to be involved in cellular immunity and cell adhesion [34] in both humans and Drosophila. TM9SF4 is primarily localized to the Golgi apparatus and early endosomes and has been shown to directly interact with the phospholipase Sac1 [35]. We performed IP-MS of Flag-Sac1 in S2 cells and identified TM9SF4 as a candidate interacting protein (Figures S3C to S3E, Data S2). Subsequent Co-IP experiments and BiFC assay in HEK 293T cells further verified direct TM9SF4-Sac1 interaction (Figure 2E; figure S3F). Co-expression of TM9SF4-mCherry and Sac1-GFP in S2 cells showed that a proportion of intracellular TM9SF4-mCherry signals colocalized with Sac1-GFP puncta (Figure 2F). As a lipid phosphatase, Sac1 is known to dephosphorylate PI4P in the ER to maintain the lipid gradient and facilitate continuous PS delivery to the PM or other organelles [16]. Although Y2H assays showed that Sac1 did not directly interact with Xkr (figure S3H), silencing tm9sf4 or sac1 both resulted in a similar reduction in apoptotic PS exposure to xkr^ko (Figure 2G), indicating that the transport of phospholipids may play a role in PS exposure.

Based on our discovery of the interaction between Xkr and TM9SF4, we assessed whether TM9SF4 or Xkr affected the subcellular distribution of PS. As the C2 domain of lactadherin (Lact::C2) can specifically bind to the biologically relevant PS isomer, phosphatidyl-L-serine, and thus serve as a probe to label PS [36], we used a mCherry-Lact::C2 fusion reporter to visualize cellular PS in S2 cells [37]. Fluorescence microscopy analysis showed that mCherry-Lact::C2 was primarily localized in the PM, and to a significantly lower extent in cytoplasmic organelles, especially the ER, in S2 cells (Figure 2H). Subsequently, induction of apoptosis led to PS enrichment at the PM (Figure S3I). By contrast, PS levels were significantly lower at the cell membrane of xkr^ko cells compared to those in S2 cells (Figures 2H’ and 2I). In contrast, knockdown of either tm9sf4 or sac1 resulted in obviously reduced PS localization to the PM, but increased localization in intracellular membrane structures (Figures 2H’’ and 2H’’’). If Xkr deficiency disrupted PS transport, more PS would be retained in the ER. When we co-expressed PS indicator (mCherry-Lact::C2) and ER indicator (Calr-GFP), we observed greater signal overlap in xkr^ko cells (Figures 2J and 2K), indicating that Xkr regulated PS exposure via PS transport from the ER to the PM. Furthermore, this disrupted PS localization to the PM could not be restored by apoptosis induction (Figures S3I and S3J), indicating that apoptotic PS exposure depended on PS delivery from the ER to the PM. Sac1 has been reported to regulate PS transport by dephosphorylating PI4P on the ER, to establish a linkage mechanism for the opposite transport of PI4P and PS. Since tm9sf4 and sac1 knockdown both resulted in impaired PS exposure similar to that in xkr^ko cells, sac1 RNAi in ACs only slightly declined efferocytosis with no significant difference (Figures 2L and 2M). Therefore, we assume that there is a more direct and efficient manner to facilitate apoptotic PS transport and efferocytosis.

Non-vesicular PS transport is required for apoptotic PS exposure and efferocytosis

As the above results revealed a role for Xkr in both distribution and translocation of PS at the PM, as well as in mediating apoptotic PS exposure, we next investigated whether components involved in non-vesicular PS transport might also participate in PS exposure during apoptosis. ORP5/ORP8 (oxysterol-binding protein (OSBP) - related protein, ORP) were located at membrane contact sites (MCS) between the ER and PM, which were recently shown to play a crucial role in regulating phosphatidylinositol 4- phosphate (PI4P) and PS levels in the PM [38]. Studies in yeast and human cells have shown that ORPs form MCSs between the ER and the PM, thus regulating lipid homeostasis via facilitating non-vesicular lipid exchange [37]. There are five ORPs in Drosophila (Figure 3A) with uncertain roles in PS transport at the MSCs between different organelles, while ORPs and its homologs are conserved between yeast (Osh family) to mammals (ORP family), and share a lipid binding domain (ORD) structure in the C-terminal region, and a pleckstrin homology (PH) domain in the N-terminal region with high affinity for PM-associated PI4P (Figure S4A) [37, 39]. The phylogenetic analysis of ORPs from yeast, Drosophila, mice, and humans (Figure S4B) resulted in the annotation of CG3860 as dORP2 and CG1513 as dORP9. Most Drosophila ORPs lack an ER transmembrane domain, suggesting that the ER integral membrane proteins may play more essential roles in recruiting Drosophila ORPs to the ER membrane. VAP (VAMP-associated protein, VAP33 in Drosophila) has been reported as an ER-anchored protein that interacts with various lipid transport proteins at ER-organelles MCSs [40].

Figure 3.

Therefore, we generated knockdown S2 cell lines by RNAi for five orp genes, as well as vap33, to assess the role of ORPs in facilitating PS exposure during apoptosis. Annexin V-PS analysis by flow cytometry showed that dOrp9, dOrp2, or cg9205 knockdown each resulted in significantly decreased PS exposure to 29.10%, 44.94%, and 31.18%, respectively, compared with 75.55% in RNAi non-targeted control cells. Additionally, vap33 knockdown also inhibited apoptotic PS exposure (Figure 3B; figure S4C), indicating that PS transport via the PM-ER MCS promotes PS exposure in ACs. We then determined whether PS transport via PM-ER MCS was required for efferocytosis and found that silencing tm9sf4, dorp2, or dorp9 could also reduce ACs recognition (Figures 3C and 3D), similar to xkr^ko. These results indicate that non-vesicular PS transport via the PM-ER MCS is essential for efferocytosis. Co-IP, Y2H, and BiFC assays indicated that Xkr likely directly interacts with both dORP2 and dORP9 (Figures 3E and 3F; figure S4D), and the co-localization of Xkr-GFP and mCherry-dORP2/mCherry-dORP9 in S2 cells showed that Xkr and dORP9 had more overlapping areas in cytoplasmic vesicles (Figure 3G; figure S4E). The endogenous Xkr can be immunoprecipitated by dORP9-Flag (figure S4F), which indicated that the interaction was not artificial or due to excessive protein expression. Meanwhile, we found that the cytoplasmic TM9SF4 also overlapped with dORP9 (figure S4G). To investigate whether knockdown of these genes affected apoptosis, we measured the levels of cleaved Drosophila Dcp-1(a caspase-3 homolog in Drosophila), which only recognizes endogenous levels of the cleaved Dcp-1 fragment during apoptosis. We found that Dcp-1 could be detected by adding actinomycin D to S2 cells, and neither xkr^ko nor tm9sf4, sac1, nor dorp9 prevented apoptosis (Figure S4H). These results suggested that Xkr did not affect apoptosis, that the apoptosis signal induced Xkr levels without caspase activation, and that increased Xkr may accelerate non-vesicular PS transport for efficient apoptotic PS exposure.

To investigate whether Xkr and TM9SF4 transferred apoptotic PS via ER-PM MCS, we employed a GFP-MAPPER or mCherry-MAPPER construct to monitor ER-PM contact sites selectively. Co-localization of MAPPER and Calr indicated that MAPPER was appropriate for monitoring ER-PM MCS(figures S5A and S5B) [41]. Confocal microscopy of S2 cells co-expressing GFP-MAPPER with TM9SF4-mCherry, or mCherry-MAPPER with Xkr-GFP, revealed Xkr and TM9SF4 localization at the plasma membrane, with an intracellular proportion of the signal for each protein overlapping with MAPPER (Figure 3H). Furthermore, the induction of apoptosis increased ER-PM MCSs and the accumulation of Xkr at ER-PM MCSs (figure S5C), supporting the role of Xkr in regulating PS transport via ER-PM MCS during apoptosis. Co-expression of Lact::C2 with MAPPER in S2 and xkr^ko cells showed that PS levels were reduced in the PM of xkr^ko cells, while fewer ER-PM MCSs were detected in xkr^ko cells than in WT (Figure 3I). These results suggest that apoptotic PS exposure was driven by PS transport via PM-ER MCS.

dORP9 serves as a bridge that strengthens ER-PM MCS through interactions with Xkr and VAP33

Protein structure analysis on the Swiss-Model website revealed that both dORP2 and dORP9 lack an ER TM domain. Therefore, we hypothesized that OSBP requires VAP33 to maintain ER stability in Drosophila. Although VAP33 was immunoprecipitated with Xkr, no apparent direct interactions were observed between them in Y2H assays. Further Co-IP analysis of possible VAP33 interactions with dORP9/dORP2, followed by BiFC experiments, indicated that VAP33 could directly interact with dORP2 and dORP9 (Figures S5D and S5E). dORP2 lacks the PH domain, and Xkr showed a more overlapping expression region with dORP9; meanwhile, the interaction between Xkr and dORP9 was enhanced after apoptosis induction, whereas inhibiting apoptosis significantly reduced the interaction (Figure S5F). We hypothesized that the interaction between Xkr and dORP9 was more specific to apoptosis. Structural analysis of dORP9 identified a PH domain that reportedly binds PI4P and an ORD domain that functions in lipid transport (Figure 4A) [39]. To narrow down the scope of regions potentially required for the dORP9-Xkr interaction, we generated a series of dORP9 truncation variants. Co-IP, Y2H, and BiFC assays indicated slight differences in the interacting domains between Xkr and dORP9, whereas all assays supported the direct interaction between Xkr and the dORP9 PH domain (Figures 4B to 4D). To enhance persuasiveness, we used Alphafold3 to predict binding sites through structural analysis, and the computational model confirmed that the intracellular loop of Xkr (D298, T301, E304, Y305, between the 6th and 7th transmembrane domains) substantially bound to the dORP9 PH domain (1-150AA) (Figure 4E). Similarly, the FFAT (diphenylalanine in an acidic tract, EFFDAxE) motif of ORP proteins can bind to the ER anchoring protein VAP, connecting OSBP to the ER membrane and mediating its structural role in supporting ER-PM MCSs [11]. Co-IP, Y2H, and BiFC assays demonstrated that the dORP9 FFAT motif (EEDFYDAYDD) likely shared direct interactions with VAP33 (Figures 4F to 4H), and Alphafold3 calculated that dORP9 likely interacts with VAP33 via the FFAT-interacting domain (Y310, D311, A312, Y313, D314, Figure 4I). Consequently, we constructed the dORP9^MU-Flag vector (mutated the amino acids of the FFAT motif into Ala) and co-expressed with HA-VAP33 in S2 cells; no longer interaction was detected (Figure 4J; figure S5G), confirming VAP33 as an ER-anchor to reinforce ER-PM MCS structures.

Figure 4.

To explore the role of Xkr-dORP9-VAP33 complex in maintaining ER-PM MCSs, we knocked down tm9sf4, sac1, dorp9 and vap33 in mCherry-Xkr and GFP-MAPPER expressing S2 cells, respectively, and we found dorp9 and vap33 disrupted the ER-PM MCSs structure, which resulted in less colocalization of Xkr and ER-PM MCSs (Figure S6), while cytoplasmic Xkr was still located at the ER-PM MCSs with a low expression of tm9sf4 and sac1. This finding indicated that depleting Xkr resulted in the destabilization of the protein complex at the ER-PM junction, and consequently, the disappearance of the ER-PM MCS and ultimately decreasing PS accumulation in the PM. Collectively, these results demonstrated that Xkr can directly interact with dORP9 at ER-PM MCS to facilitate non-vesicular PS transport during apoptosis.

Xkr cooperates with non-vesicular PS transport proteins to mediate efferocytosis in vivo

Given that Xkr interacted with dORP9 and VAP33 at ER-PM MCSs, TM9SF4 interacted with Sac1, we hypothesized that Xkr, TM9SF4, and dORP9 could exhibit a similar phagocytic phenotype in the same genetic pathway. In light of our above results showing that respectively suppressing tm9sf4, sac1, or dorp9 in S2 cells led to an obvious reduction in efferocytosis, we next tested whether these proteins regulated efferocytosis in vivo. To this end, we characterized the efferocytosis phenotypes of tm9sf4¹, sacl^EY02269, and dorp9^S-532 mutants from the Bloomington Drosophila Stock Center. Relative expression analysis by RT-qPCR showed that all three mutants were likely null alleles (Figure S7A), and the numbers of macrophages and apoptotic cells in mutant embryos (at stage 13) did not significantly differ from those in WT embryos, suggesting that they do not regulate apoptosis or macrophage development in Drosophila (Figures S7B to S7G). The tm9sf4¹ and dorp9^s-532 embryonic macrophages both engulfed fewer apoptotic cells than macrophages in WT embryos, with a higher proportion of unengulfed than engulfed apoptotic cells, indicating impaired recognition and clearance of apoptotic cells (Figures 5A to 5H). However, the sac1 mutant only exhibited slightly lower capacity for efferocytosis, with PIs of 2.05 ± 0.59. As tm9sf4 RNAi displayed an efferocytosis phenotype more similar to that of xkr^ko in vitro, we speculated that Sac1 deficiency could decrease the efficiency of PS translocation from ER to PM without completely abolishing it.

As our data showed that Xkr could physically interact with TM9Sf4 and dORP9, we next tested whether Xkr also genetically interacted with TM9SF4 and dORP9 in mediating cell corpse clearance by combining these mutations in vivo. To this end, we crossed xkr^ko mutant flies with tm9sf4¹ and dorp9^S-532 flies to obtain respective heterozygous double mutants. Efferocytosis assays in double mutant embryos indicated that xkr^ko/tm9sf4¹ and xkr^ko/dorp9^S-532 heterozygotes exhibited similarly impaired efferocytosis compared to WT flies, as that observed in homozygous single mutants (Figures 5D and 5G). Alternatively, macrophages in xkr^ko/tm9sf4¹ and xkr^ko/dorp9^S-532 heterozygous embryos failed to engulf more apoptotic cells (Figure 5F). Together, these results provide evidence of genetic interactions among xkr, tm9sf4, and dorp9 in the efferocytosis of embryonic macrophages.

Figure 5.

Human ORPs Xkr8 is required for the apoptotic PS exposure

In mammalian cells, Xkr8 undergoes C-terminal cleavage by caspase 3, leading to the release of activated Xkr8 fragments, which subsequently form heterotetramers with either basigin (BSG) or neuroplastin (NPTN). These heterotetramers then promote phospholipid scrambling during apoptosis [25]. And in some unidentified processes, Xkr8 can constitutively expose PS without an apoptosis stimulus in a phosphorylationdependent manner [42]. To assess whether PS transport from the ER to PM influences apoptotic PS exposure in mammals, we suppressed ORP expression in human HEK 293T cells using xkr8 silencing by RNAi as a positive control. Flow cytometry assays with FITC-annexin V staining in HEK 293T cells treated with staurosporine (STS) to induce apoptosis showed that xkr8, tm9sf4, or osbpl8 silencing resulted in significantly decreased PS exposure (Figure 6A, figure S8A). These results suggest that non-vesicular PS transport from the ER to the PM may also be required to regulate apoptotic PS exposure in mammals. Therefore, we investigated whether PS exposure mediated by non-vesicular PS transport was required for efferocytosis in vitro. For this experiment, we induced apoptosis in Jurkat cells with shRNA knockdown of xkr8, htm9sf4, sacm1l, osbpl5, and osbpl8 encoded on a constitutive GFP expression vector, and then incubated these reporter mutants with RAW264.7 macrophages [43]. Quantification of phagocytic engulfment of each Jurkat knockdown cell line indicated that xkr8 and htm9sf4 knockdown resulted in reduced efferocytosis compared to that in untargeted RNAi control cells; osbpl8 knockdown resulted in slightly reduced efferocytosis (Figures 6B and 6C, figure S8B).

Figure 6.

As OSBP5/8 has been found to play important roles in non-vesicular PS transport via ER-PM MCS [38], and our above experiments showed that dORP9 could interact with Xkr in Drosophila, we sought to determine whether Xkr8 functions in conjunction with these ORP proteins to facilitate PS translocation from the ER to PM during apoptosis in mammalian cells. Co-IP assays in HeLa cells revealed that Xkr8 could interact with human OSBPL8 or OSBPL9, as well as human TM9SF4 (Figure 6D). This finding confirmed that transport of PS via ER-PM contact sites leads to efficient PS exposure and that interactions between Xkr8 and ORP5/8 regulate the entire process, ensuring PS is exposed in response to apoptotic signals.

Discussion

Our findings discover that the Xkr8 homolog in Drosophila melanogaster, Xkr, regulates non-vesicular transport of PS from the ER to the PM through direct interaction with the Drosophila ORPs (dORP9) to ensure efficient PS exposure upon apoptosis induction, triggering cell corpse recognition and clearance. Here, we show that Xkr protein level is increased in response to apoptotic signaling stimulation in a caspaseindependent manner, and the enhanced Xkr is enriched at ER-PM MCSs. The increased ER-PM MCSs may lead to the reinforced interaction between Xkr and dORP9, which in turn promotes the non-vesicular transport of PS from the ER to the PM through ER-PM contact sites, ultimately leading to PS externalization and efficient clearance of apoptotic cells. This mechanism is conserved from Drosophila to human macropahge (Figure 7).

Figure 7.

As a classic and essential phospholipid scramblase, it has been clearly elucidated that Caenorhabditis elegans CED-8 and mammalian Xkr8 mediate apoptotic PS exposure in a caspase-dependent manner, respectively. Cleavage of an N-terminal EQED site of CED-8 or a C-terminal DXLD caspase recognition site of Xkr8 is required for their scramblase activation and the subsequent apoptotic cell clearance [44, 45]. The caspase activation is essential for the apoptotic PS exposure, whereas Xkr8 homolog in Drosophila, Xkr, lacks caspase recognition sites according to sequence alignment, and in native S2 cells, both C- and N-terminal eGFP-tagged Xkr showed a similar localization, mainly in the PM, partially in the cytoplasm, suggesting that Xkr activation may not require cleavage. Our study demonstrates that Xkr functions as a scramblase in the exposure of PS during apoptosis, a process essential for efficient efferocytosis both in vivo and in vitro. Previous studies have shown that in Drosophila, degenerated or damaged neurons expose PS to cause recognition and phagocytosis of dendrites by epidermal cells [46]. Overexpression of Xkr leads to ectopic PS exposure and a reduction of dendrites, suggesting that Xkr is involved in the PS exposure of damaged dendrites; but this PS exposure-mediated phagocytic process is independent of caspase activity, though it depends on the phagocytic receptor Draper on epidermal cells [23]. In addition, mouse Xkr8 activity in inducing apoptotic PS exposure has been reported to be stimulated by kinase-mediated phosphorylation, independent of caspase-mediated cleavage, suggesting that Xkr scramblase activity may also be regulated through an unidentified mechanism. Consistent with their finding, we found that apoptosis signal-induced increase in Xkr levels was regulated post-transcriptionally without caspase cleavage. Further research is required to investigate how apoptosis signals regulate Xkr amounts at the post-transcriptional level.

Based on these findings, we performed IP-MS analysis to identify proteins interacting with Xkr in regulating apoptotic PS exposure, and TM9SF4 was identified as a direct interaction partner of Xkr by Co-IP, Y2H, and BiFC assays. In human immune cells, TM9SF4 has been shown to increase phagocytic activity in metastatic tumor cells [47]. In Drosophila, tm9sf4 mutant macrophages and tm9sf4 knockdown S2 cells show a striking defect in phagocytosis of Escherichia coli-derived particles [34]. However, whether and how TM9SF4 is involved in apoptosis remains unclear. Our results showed that Xkr and TM9SF4 generally colocalize in the PM and, to a lesser extent, in the cytoplasm. Knockdown of tm9sf4 in S2 cells disrupted the subcellular distribution of PS, inhibiting its exposure during apoptosis and subsequently impairing efferocytosis, whereas direct binding between TM9SF4 and Xkr promoted apoptotic PS exposure in vitro and in vivo. These data clearly demonstrate that TM9SF4 can regulate PS exposure, thus providing a previously undocumented caspase-independent mechanism of apoptotic PS exposure in efferocytosis.

As TM9SF4 has not been reported to participate in phospholipid transport, previous studies have shown that TM9SF4 can interact with the phospholipid phosphatase, Sac1, in human cells [35], which is consistent with our results that also identified TM9SF4 as a direct Sac1 interaction partner, whereas Sac1 does not directly interact with Xkr. Sac1, which primarily converts PI4P into phosphoinositides (PIs) at the MCS, plays a crucial role in establishing lipid gradients required for ORPs-mediated non-vesicular lipid transport [16]. We found that knockdown of xkr, tm9sf4, or sac1 resulted in significantly increased PS localization to intracellular membrane structures (i.e., the ER and vacuoles) and reduced PS localization at the PM, whereas apoptosis signals could not restore the disorder PS localization. Furthermore, RNAi silencing of tm9sf4 or sac1 inhibited apoptotic PS exposure and phagocyte recognition. These cumulative results suggest that Xkr and TM9SF4 may regulate apoptotic PS exposure via non-vesicular PS transport at ER-PM contact sites. Our preliminary Sac1 knockdown data hint at a possible role for phosphoinositide regulation, but further studies are required.

Non-vesicular lipid transfer between membranes of different organelles is crucial for maintaining membrane homeostasis [48], with ORPs regulating intermembrane lipid transport in eukaryotes [15]. Non-vesicular PS transport was first reported in the homologous Osh/ORP proteins, Osh6 and Osh7, which localize to ER-PM contact sites in yeast. However, unlike vesicular transport, these proteins remain the only known lipid transporters with a demonstrated preference for PS substrate [49, 50]. In contrast, their mammalian orthologs, ORP5 and ORP8 [37, 50], are recruited to PM-ER contact sites to mediate PS-PI4P exchange between the PM and ER [51, 52]. Interestingly, our RNAi screen of OSBP homologs in Drosophila showed that knockdown of an OSBPL2 homolog, CG3860 (dORP2), and an OSBPL9 homolog, CG1513 (dORP9), but not ORP5/ORP8 homologs, resulted in obvious defects in PS exposure and efferocytosis in S2 cells, implying a role in non-vesicular PS transport.

We found that Xkr directly interacts with dORP9. Previous studies have shown that ORP9, along with the ORP5/8 ortholog ORP10/11, localizes to ER-trans Golgi network (TGN) contact sites and influences PS levels in the TGN[53]. Two recent studies found that these proteins could mediate PS-PI4P exchange between the ER and endocytic compartments for downstream processes such as recycling endosome fission [54] or lysosomal repair [55]. These findings support the possibility that ORP9 participates in lipid exchange at ER-endosome contact sites, while our results show that dORP9 regulates PS exposure during apoptosis via PS transport at ER-PM contact sites. dORP9 is expressed at ER-PM contact sites, and structural analysis revealed that dORP9 harbors both PH and ORD domains, which reportedly bind to PI4P in the PM and transport lipids, respectively [56]. As dORP9 lacks an ER TM domain, we hypothesized that other proteins might be required to stabilize its localization in the ER. We found that dORP9 directly interacts with VAP33 in the ER via its FFAT motif (EEDFYDAYDD), whereas Xkr directly interacts with its PH domain. However, Xkr showed no direct interaction with VAP33, suggesting dORP9 might serve as a bridge to connect VAP33 with Xkr, further reinforcing PM-ER contact sites to mediate PS transport. This similar mechanism was also observed in mice. In mouse WR19L T cells, another member of the Xkr family, Xk, participates in the exposure of PS mediated by ATP-P2X7 in a caspase-independent manner. Xk was found to interact with the lipid transporter protein VPS13A. VPS13A is primarily located at the contact sites between the ER and intracellular organelles, mediating phospholipid transport at the organelle membrane contact sites. Xk directly binds to the PH domain of VPS13A via its intracellular P—hairpin structure, enabling Xk to open the channel to scramblase P2X7- induced PS exposure. Defects in Xk and vps13a both lead to neuroacanthocytosis associated with McLeod syndrome, which may be caused by disrupted PS homeostasis [57]. However, the mechanism by which P2X7 activates the Xk-Vps13A complex to scramblase PS remains unclear, just as we are not clear about how the apoptosis signal activates the Xkr-dORP9 complex to expose apoptotic PS, although we proved that the interaction between Xkr and dORP9 is enhanced after apoptosis induction.

Taken together, this study reveals for the first time that apoptotic signaling induces post-translational upregulation of Drosophila Xkr, the assembly of the XKR-DORP9- VAP33 complex at the ER-PM contact site through direct binding to the PH domain of dORP9, and drives the non-vesicular trafficking of PS from the ER to the PM. At the same time, Xkr cooperated with TM9SF4 module to achieve PS eversion and efficient phagocytosis of apoptotic cells. However, how apoptosis signals specifically “sense” and post-translationally regulate Xkr protein levels needs to be further illustrated. Meanwhile, as Xkr8 forms heterodimers with either Basigin (BSG) or Neuroplastin (NPTN), further studies are necessary to determine whether these molecules also participate in non-vesicular PS transport. Finally, we demonstrated that this PS transport mechanism, in which Xkr8/Xkr functions as a scramblase to expose PS while also mediating PS transfer from the ER to the PM is conserved between flies and mammals, which holds major implications for developing targeted therapeutics in several human diseases associated with defects in apoptosis and efferocytosis.

Materials and methods

Cell culture and transfection

Drosophila melanogaster Schneider’s S2 cells were cultured in Sf-900TM II SFM (Gibco #10902088) containing 1% penicillin/streptomycin. Cells were incubated at 25 °C and transfected at 60-75% density with a DDAB suspension (250 μg/ml) [58].

Human embryonic kidney cell line HEK 293T and RAW 264.7 mouse macrophages were cultured in DMEM (Gibco) containing 10% FBS and 1% penicillin/streptomycin, and Jurkat cells were cultured in RPMI-1640 (Gibco) containing 10%FBS and 1% penicillin/streptomycin. Mammalian cells were incubated at 37°C and 5% CO2. Cell transfection was performed using Lipofectamine™ LTX and Plus Reagent (Invitrogen# 15338100), according to the manufacturer’s instructions.

Double-stranded RNA production and RNAi treatment

Primers for generating specific double-stranded RNA (dsRNA) were obtained from the D. melanogaster RNAi Screening Center (DRSC), DRSC/TRiP Functional Genomics Resources (harvard.edu), and the T7 RNA polymerase sequence was added to the 5′ end of each primer. dsRNA was generated using 1 μg of PCR product, followed by the High Yield RNA Synthesis Kit (NEB #E2040S) dsRNA synthesis. dsRNA sequences are shown in Table S1. S2 cells were treated with 10μg/ml dsRNA and incubated for 48h to get high RNAi efficiency. Cells were harvested to extract RNA or proteins, and apoptotic cells were added to observe phagocytic phenotypes using a confocal microscope (C2, Nikon Instruments).

Total RNA preparation and Quantitative Real-time PCR

Total RNA was extracted from S2 cells and flies using TRIzol reagent (Invitrogen, #15596026) and the Direct-zol RNA miniprep kit (Zymo Research, #R2052). RNA was reverse transcribed using the HiScript II 1st Strand cDNA Synthesis Kit (Vazyme #R211-02), and qPCR was performed using ChamQ Universal SYBR qPCR Master Mix (Vazyme #Q711-02) in an ABI Step One Plus instrument. Data represent the ratio or relative ratio of target mRNA levels to those of an internal reference gene (Rpl32). All experiments were performed in triplicate, with mRNA samples extracted from at least three biological replicates. The relative gene expression was quantified using the comparative method (ΔΔCT). The qPCR primers used for real-time quantitative PCR are listed in Table S2.

DNA Constructs

pAC-mCherry-Lact::C2 was used to indicate PS distribution. The pAC-mCherry-Lact::C2 were constructed first by PCR amplifying the bovine C2 domain (245-380aa) of mouse Lactadherin[59] and subsequently cloned into the pAC-mCherry vectors, using the Xhol and Xbal sites.

pAC-GFP-MAPPER or pAC-mCherry-MAPPER was used to indicate the ER-PM MCS. The MAPPER sequence was cloned as described by previous research [41]. Briefly, MAPPER contains: 1) a signal peptide (1-22aa) of human STIM1 in the N-terminal; 2) GFP or mCherry followed by STIM1 transmembrane (TM) domain (195- 240aa) with a downstream (GA)*6 linker sequence [60]; 3) FKBP12-rapamycin-binding (FRB) with a downstream (SAGG)*3 linker sequence; 4) the polybasic tail (PB, 199-219aa) from human Rit tail [61]. To generate MAPPER, two helical linkers, (EAAAR)*4 and (EAAAR)*6, were cloned into the upstream and downstream flanking regions of the FRB, respectively.

Efferocytosis assay in vitro

Efferocytosis assays were performed as described previously [62]. Briefly, S2 cells were incubated with 5 μM CellTracker™ Red CMTPX (ThermoFisher#C34552) at 25°C for 1 h and then treated with 0.25 μg/mL Actinomycin D (AcD, MCE#HY-17559) at 25°C for 1 h, which has been demonstrated to induce apoptosis in S2 cells and exposure of PS on the surface of the dying cells[63]. Cells were collected by centrifugation at 1000 × g for 3 min and suspended in Sf-900TM II SFM. CMTPX-labeled prey cells (3×10⁵ cells) were incubated with 1×10⁵ S2 cells at 25°C for 60 min, and analyzed by confocal microscopy on glass-bottom 96-well plates (Thermo Fisher).

NBD-labeled phospholipid incorporation

Cells were stimulated with apoptotic stimuli, as described above. Centrifuge (1000×g, 3 min, room temperature) and wash the cells with 1 ml HBSS. The cells were centrifuged and washed with 1 ml HBSS/CaCl₂ (containing 1 mM CaCl₂), and then resuspended in 1 ml of cold HBSS/CaCl₂. Transfer 600 ml of the cell suspension to a 1.5-ml tube. Incubate the mixture in an aluminum block on ice for 7 min. Next, 600 ml of 200 nM NBD-PC (Avanti # 810130) in HBSS/CaCl₂ to the cell suspension (NBD-PC final concentration, 100 nM). Take 150 μl aliquots at 1, 2, 4, 6 and 8 min, mixed with 150 μl of lipid extraction buffer (HBSS/CaCl₂, 5 mg/ml fatty acid-free Bovine Serum Albumin) containing 500 nM Sytox Blue (Invitrogen, # S34857), and kept on ice. The Sytox Blue-negative population was analyzed for the incorporation of fluorescent lipids by flow cytometry.

Generation of xkr^ko S2 mutant cells

xkr sgRNA sequences were selected with an NGG PAM sequence in the genome. Two oligonucleotides for each target-Xkr sgRNA 1F: GATTTGCATCAACTGGAGCC; Xkr sgRNA 2R: GGCTCCAGTTGATGCAAATC were synthesized and annealed as previously described [64], ligations were performed with approximately 50 ng pAc-sgRNA-Cas9 vector and 2 ml annealed diluted ligonucleotides with T4 DNA ligase (NEB) in a 10 ml reaction volume for 2 h at 18 °C, and transformed into chemically competent E. coli DH5a. Positive clones were selected by colony PCR and sequenced with U6F. pAc-xkr sgRNA-Cas9 vector was transfected into S2 cells and selected by 5mg/ml puromycin (Sigma), a single cell was screened by 96-well plates, followed by temperature gradient PCR to identify xkr knockout cells.

Fly strains and genetics

The following D. melanogaster strains were obtained from the Bloomington Drosophila Stock Center: TM9SF4¹(BDSC#26653), Sac1^EY02269(BDSC#19651), CG1513^MB07184(BDSC#25259), CG3860^{CR00915-TG4.1} (BDSC#81155), Hml-GAL4>2xEGFP, Chr2(BDSC#30140), Hml-GAL4>2xEGFP, Chr3(BDSC#30142), vasa-cas9(BDSC#51323), and Crq-GAL4 (ffl) was a gift from Dr. Zongzhao Zhai (Hunan Normal University). The pUAS-Xkr were constructed in our laboratory. Briefly, selected genes were amplified by PCR and cloned into the Invitrogen pENTR/D-TOPO vector (Thermo Fisher # K242020). The Gateway™ LR Clonase™ II kit (Thermo Fisher #11791100) was used to transfer selected genes from the pENTR/D-TOPO vector to pPWG following the manufacturer’s instructions. xkr^ko fly mutant was obtained by our lab. The xkr sgRNA constructs were constructed using xkr sgRNA oligos obtained from http://targetfinder.flycrispr.neuro.brown.edu/, as shown in Table S2. The constructs were verified by sequencing, and transgenic flies were generated by Unihuaii Inc. using the PhiC31 targeted integration system. The sgRNA transgenic flies were crossed with the vasa cas9 strain to generate the Cas9-editing mutants, a single fly mutant was verified by qPCR and sequencing.

Antibody generation

The Xkr antibodies were obtained by immunizing mice in our laboratory. We synthesized two polypeptides of Xkr (from GL Biochem), 5-20AA and 194-206AA, respectively, and then mixed them with Freund’s complete adjuvant, getting the antigen. Antibodies were raised by immunizing mice (BALB/c, 6-8 weeks old).

Protein extraction and western blotting

Protein samples were extracted and homogenized on ice in lysis buffer (50 mM Tris-HCl [pH 7.4], 1% TritonX-100, 0.15 M NaCl, 1 mM EDTA) with protease inhibitors. Lysates were centrifuged at 20000 xg for 30 min, and the supernatants were collected and quantified using the bicinchoninic acid assay. Equal amounts of protein were separated by SDS-PAGE on a 10-15% gradient gel, and transferred to a PVDF membrane (Millipore), which was blocked with fresh milk (5% in TBST) and incubated overnight with primary antibodies.

The antibodies were used at the following dilutions: mouse anti-actin (CST), 1:1000; mouse anti-Flag (Sigma) 1:1000; mouse anti-HA (CST) 1:1000; rabbit anti-Dcp-1 (CST, Cleaved Drosophila Dcp-1 (Asp215)) 1:400; rabbit anti-V5 (Beyotime) 1:1000; mouse anti-Crq 1:500, rabbit anti-Crq 1:400; mouse anti-Xkr 1:400. Anti-rabbit or anti-mouse horseradish peroxidase (HRP)-coupled secondary antibodies (Jackson Immuno Research Laboratories) were used at a 1:10000 dilution; ECL detection was performed according to the manufacturer’s protocol (Pierce). All western blotting assays needed for quantitative analysis were repeated three times.

Immunoprecipitation and LC-MS/MS protein preparation

Plasmids encoding the differently expressed tagged proteins were co-transfected into S2 cells or 293T cells as previously described, and immunoprecipitation was carried out at 4 °C. Briefly, transfected cells were lysed in 50 mM Tris-Cl [pH 7.4], 1% TritonX-100, 0.15 M NaCl, 1 mM EDTA with protease inhibitor, and centrifuged. The supernatants were incubated with anti-FLAG M2 magnetic beads (Sigma#M8823) or anti-HA magnetic beads (CST#11846) for 2 h at 4 °C, beads were washed three times with TBS (50 mM Tris, 150 mM NaCl), and boiled in 2xSDS loading buffer for 5 min. For Co-IP assays, the supernatants were analyzed by western blotting. For LC-MS/MS analysis, 5 μl of the supernatants were loaded onto an SDS-PAGE gel, followed by silver staining. The rest of the samples were sent to HOOGEN BIOTECH company (Shanghai, China) for LS-MS/MS analysis. The MS/MS analysis was performed using three independent biological repeated samples.

Yeast two-hybrid assays

For binary yeast two-hybrid assays, transmembrane proteins use the split-ubiquitin system, which inserts genes of interest into Cub or Nub as vectors and transforms vectors into yeast THY.AP4 [65]. Yeast competent cells were obtained using LiAc solution, as previously described, and transformed with target gene-Cub or Nub vectors. The mixture was stirred and incubated at 30 °C for 30 min, followed by heat shock treatment at 42 °C for 40 min. The yeast cells were then centrifuged at 2000 ×g for 1 min, washed with 100 μL of sterile water, resuspended in 50 μL of sterile water, plated on synthetic-defined (SD)-Leu/Trp dropout plates, and grown for 3-4 days at 30 °C. Finally, three representative colonies were selected and diluted for spot testing. The interaction was detected using SD-Ade/His/Trp/Leu plates.

Bimolecular fluorescence complementation (BiFC)

BiFC analysis enables the direct visualization of protein interactions in living cells. The two fluorescein fragments of mVenus were fused to two candidate interacting proteins, and the fragments formed a fluorescent complex when the two proteins interacted with each other [66]. The two proteins whose interactions were to be detected were cloned into pcDNA3.1-mVenusN and pcDNA3.1-mVenusC vectors, coding for the N-terminal (VN) and C-terminal (VC) fragments of mVenus, respectively, and transfected into 293T cells using the PEI method. After 48 h of culture, the fluorescence complementarity results were observed by confocal laser microscopy. If the two proteins interact, they excite yellow fluorescence.

Embryo immunostaining

Immunostaining was performed as previously described [67, 68]. Briefly, Drosophila embryos at stage 13 were fixed in 4% paraformaldehyde and stained with relevant antibodies, i.e., Crq specifically labels macrophages while Dcp-1 labels ACs. The goat anti-rabbit Alexa Fluor Plus 594-coupled or anti-mouse Alexa Fluor Plus 488-coupled secondary antibodies (Jackson Immuno Research Laboratories) were used at a 1:1000 dilution. Stained embryos were mounted in Vecta Shield medium, observed using confocal laser microscopy (C2, Nikon Instruments), and photographed at 20* and 100* magnifications. Images were processed using the Adobe Illustrator software. The phagocytosis ratio was quantified as the number of engulfed ACs compared to the total ACs, and PI was quantified as previously described [68, 69]. The PI was calculated from confocal images taken from five embryos and three confocal image stacks per embryo.

Cell immunofluorescence assay

S2 cells were plated on a cover glass, fixed in 4% paraformaldehyde for 5 min, incubated in 1% bovine serum albumin (BSA) in 0.1% PBS-Tween for 1 h, and then incubated with the antibody overnight at 4 °C in PBS containing 1% BSA and 0.1% Tween. Alexa Fluor 488-conjugated goat anti-mouse antibody was used at a 1:500 dilution for 1 h. Hoechst 33342 was used to stain the cell nuclei. The samples were imaged using confocal laser microscopy (C2; Nikon Instruments).

Molecular modeling of Xkr, Vap33, and dORP9

The protein sequences of Xkr, Vap33, and dORP9 were downloaded from Flybase(https://flybase.org/), and the structures of the Xkr, Vap33, and dORP9 proteins were calculated using the AlphaFold3 server[70]. The output prediction model was rigorously evaluated for quality. Finally, the prediction results were visualized in three dimensions and analyzed in depth for the interaction interface using PyMol software (version 3.1.4). In brief, AlphaFold 3 generated five alternative models along with their corresponding confidence scores. We mainly relied on the interface template modeling score (ipTM > 0.6) to select the optimal model for subsequent analysis. The optimal complex structure model predicted by AlphaFold 3 was imported into the PyMOL molecular graphics system for visualization and interaction analysis. After the nonpolymer ligands were deleted and only the protein backbone was retained in PyMOL; we predicted the binding interface of protein A and protein B through hydrogen bond and salt bridge analysis, hydrophobic interaction analysis, and interface area calculation. Further ray-tracing rendering is used to generate high-resolution illustrations.

Apoptosis induction of Jurkat cells and efferocytosis assay by Raw264.7 cells

Jurkat human leukemia T cells treated with the broad-spectrum protein kinase inhibitor staurosporine (Sts) were used as a model for apoptosis, which causes activation of executioner caspases [71]. Briefly, Jurkat cells transfected with shRNA-pGreen vectors were treated with 10nM Sts at 37°C for 60 min. The cells were then collected by centrifugation at 500 g for 5 min and suspended in DMEM-10% FCS. Apoptotic Jurkat cells were added to Raw264.7 cells at a ratio of 10:1 and incubated at 37°C for 6h. After washing with PBS, the phagocytes were suspended in 100 μl of PBS containing 2% FCS and analyzed by confocal microscopy on 96-well glass-bottom plates (Thermo Fisher Scientific).

Phylogenetic tree

Firstly, collect homologous protein sequences through public databases (such as Flybase, NCBI) to make a fasta file, and then perform sequence alignment using Mega11 software. The phylogenetic tree was generated by the maximum likelihood method.

Statistical analysis

Statistical analyses were performed using the GraphPad Prism software. Data are presented as the mean ± SEM. One-way analysis of variance (ANOVA) or Student’s t- test was used to determine statistical significance. “*” stands for p < 0.05, “**” for p < 0.01, “***” for p < 0.001, and “****” for p < 0.0001. For efferocytosis statistics in flies, at least five embryos from each genotype were tested, unless otherwise indicated.

Data availability

All data needed to evaluate the conclusions in the paper are present in the paper and/or the Supplementary Materials. The mass spectrometry proteomics data have been deposited to the ProteomeXchange Consortium via the PRIDE partner repository with the dataset identifier PXD058895.

Supplementary information

Table S1.

Table S2.

Table S3.

Fig. S1.

Fig. S2.

Fig. S3.

Fig. S4.

Fig. S5.

Fig. S6.

Fig. S7.

Fig. S8.

Acknowledgements

We thank the Bloomington Drosophila Stock Center for the Drosophila melanogaster strains. This work was supported by the list funding:

• National Natural Science Foundation Key Project of China grant 91954114 (XH)

• National Natural Science Foundation of China grant 32370799 (XH)

• National Natural Science Foundation of China Youth Program grant 31801164 (ZQ) Natural Science Foundation of Shaanxi Province grant 2024JC-YBMS-638 (WH) Program of Innovative Research Team for the Central Universities grant GK202302003 (XH)

Additional information

Ethics Statement

All experiments were performed in accordance with the approved guidelines. The animal experiments were performed in accordance with the National Institute of Health’s Guide for the Care and Use of Laboratory Animals and were approved by the Animal Care and Use Committee of Shaanxi normal university (Xi’an, China).

Author contributions

Conceptualization and experimental designing: XH, ZQ

Experiments performation: ZQ, MXY, XT, WAH, LFQ

Methodology and materials preparation: MXY, YL, WH, YQ

Data analysis and experimental advices: ZQ, LFQ, YL, WH

Writing - original draft: ZQ, XH

Writing - review & editing: ZQ, XH, WH

Funding acquisition: XH, ZQ, WH

Additional files

Data file S1. The screened proteins of LC-MS/MS preyed by Xkr-FLAG.

Data file S2. The screened proteins of LC-MS/MS preyed by Sac1-FLAG.