The scavenger receptor SCARA1 (CD204) recognizes dead cells through spectrin

Scavenger receptor class A member 1 (SCARA1 or CD204) is an immune receptor highly expressed on macrophages. It forms homotrimers on the cell surface and plays important roles in regulating immune responses via its involvement in multiple pathways. However, both the structure and the functional roles of SCARA1 are not fully understood. Here, we determined the crystal structure of the C-terminal SRCR domain of SCARA1 at 1.8 Å resolution, revealing its Ca2+-binding site. Results from cell-based assays revealed that SCARA1 can recognize dead cells, rather than live cells, specifically through its SRCR domain and in a Ca2+-dependent manner. Furthermore, by combining MS and biochemical assays, we found that cellular spectrin is the binding target of SCARA1 on dead cells and that the SRCR domain of SCARA1 recognizes the SPEC repeats of spectrin in the presence of Ca2+. We also found that macrophages can internalize dead cells or debris from both erythrocytes and other cells through the interaction between SCARA1 and spectrin, suggesting that SCARA1 could function as a scavenging receptor that recognizes dead cells. These results suggest that spectrin, which is one of the major components of the cytoskeleton, acts as a cellular marker that enables the recognition of dead cells by the immune system.

Scavenger receptors (SRs) 2 were discovered in late 1970s through the uptake of the modified low-density lipoprotein (LDL) by macrophages (1,2). Over the past decades, a large number of cell-surface receptors have been identified as SRs and are categorized into more than 10 subfamilies or classes based on the sequence and structural similarities (3,4). SRs are typically expressed on antigen-presenting cells such as macrophages and dendritic cells and are involved in the regulation of both innate and adaptive immune pathways (5,6). The ligands of SRs are quite diverse, including both self-and nonself targets (3,7), which is not surprising as numerous substances such as metabolites and dead cells need to be removed promptly to maintain body homeostasis, and evidence over the past years has shown that SRs are widely associated with diseases, including autoimmunity, cardiovascular diseases, and cancer (4,8,9).
SR class A (SR-A) has five known members, including SCARA1 (CD204, SR-A, SR-A1, MSR, etc.), SCARA2 (MARCO), SCARA3 (CSR), SCARA4 (SRCL), and SCARA5 (10). These molecules are type II transmembrane proteins and share similar structural features, including a small N-terminal cytoplasmic region, a transmembrane helix, and a large C-terminal extracellular portion (Fig.  1A), but their functions appear to be diverse based on the published data (11)(12)(13). The ectodomains of SR-A members have been predicted to form homotrimers on the cell surface and may contain coiled-coil regions (CC), a collagen-like region (CL), and a C-terminal globular domain (Fig. 1A). Among them, SR class A member 1, SCARA1/CD204, is highly expressed on macrophages and dendritic cells (14). The ectodomain of SCARA1 contains a CC region, a CL region, and a scavenger receptor cysteine-rich (SRCR) domain ( Fig. 1A) (15). The SRCR domain is typically found in SRs and may contain cation-binding sites, which are usually associated with ligand recognition (16). Previous studies have shown that SCARA1 is involved in multiple biological pathways and is associated with diseases, including cancer (17), cardiovascular disease (18), and Alzheimer's disease (19). A number of ligands of SCARA1 have been reported (14,20), and evidence has also shown that it might be involved in phagocytosis of dead cells by macrophages (21)(22)(23). However, the mechanisms of SCARA1 in these pathways are not fully understood.
To maintain body homeostasis, dead cells (apoptotic and necrotic cells) need to be removed promptly to prevent autoimmunity and other potential diseases (24 -26). Phagocytes such as macrophages and dendritic cells play critical roles in the clearance of dead cells, which recognize the specific markers on dead cells by the surface receptors (27). Among the known dead-cell markers, phosphatidylserine (PS) has been identified decades ago as a typical marker on the surface of apoptotic cells and can be recognized by several receptors and mediates apoptotic cell clearance (26,28). Recently, it has been shown that other cellular markers can also mediate the recognition of dead cells through different receptors on macrophages or dendritic cells (29 -31), suggesting that the efferocytosis pathways could be more complex than previously thought (25,27,32,33).
Spectrin was initially identified in red blood cells and is known as a "ghost protein" because it is located on the medial side of erythrocyte membrane (34) and is important for maintaining the biconcave shape of erythrocytes (35,36). Later, it was found that spectrin was also widely expressed in nonerythrocytes as a general component of the cytoskeleton (37,38). Spectrin has an ␣-chain and a ␤-chain, and the sequence of the ␣-chain contains a number of repetitive domains (SPEC domains) as well as an SH3 domain and an EF hand domain (37,39,40). As a part of cytoskeleton, spectrin interacts with other cytoskeleton components such as actin, ankyrin, adducin, myosin, and flotillin, thus forming a cytoskeleton network in cytoplasm, which is important for cell growth, tissue patterning, and organ development (37,(41)(42)(43), and it has also been linked to a number of diseases (44). However, the functional roles of spectrin other than the cytoskeleton remain unclear.
Here, we determined the structure of the C-terminal SRCR domain of SCARA1 by crystallography and explored the function of SCARA1 using biochemical and biophysical methods, and we found that SCARA1 could recognize dead cells specifically through cellular spectrin in a Ca 2ϩ -dependent manner, suggesting a novel pathway for the immune recognition of dead cells.

Crystal structure of the C-terminal SRCR domain of SCARA1
Because the intact ectodomain of SCARA1 contains three regions with flexible linkers in-between (Fig. 1A), the structural determination of the whole ectodomain might be difficult. Therefore, we expressed the C-terminal SRCR domain of mouse SCARA1 (residues 347-454) in insect cells and purified the domain by Ni-NTA affinity chromatography and size-exclusion chromatography (SEC) (Fig. 1B). The purified SRCR domain was crystallized, and the structure was solved by molecular replacement using the SRCR domain of the Mac-2binding protein (M2BP) (PDB entry 1BY2) as a search model and refined to 1.8 Å resolution (Table S1). The SRCR domain of SCARA1 has the typical SRCR fold (Fig. 1C), similar to the SRCR domains from other SRs (13,15,45). The SRCR domain of SCARA1 has a cation-binding site coordinated with residues Asp-379, Asp-380, and Glu-446 (Fig. 1D), which is modeled as Ca 2ϩ as it is included in the crystallization buffer. The residues around the Ca 2ϩ -binding site are conserved for the mouse and the human SRCR domain of SCARA1, and a similar cationbinding site has also been identified in the SRCR domain of SCARA2 (13). Although some other SRCR domains, for example the SRCR domain of Mac-2-binding protein (M2BP) (PDB entry 1BY2) (15) and the SRCR domain of CD6 (PDB entry 5A2E) (45), do not contain the similar Ca 2ϩ -binding site, the local conformations around the site are relatively conserved, suggesting that Ca 2ϩ binding is largely contributed by the coordinating residues on protein surface.

Ca 2؉ -dependent recognition of dead cells by SCARA1
To explore the potential scavenging function of SCARA1, the ectodomains of both mouse and human SCARA1 fused with GFP were expressed in insect cells and purified by Ni-NTA affinity chromatography and SEC. The purified proteins were used for cell-binding assays and monitored by flow cytometry. The results showed that mouse SCARA1 had no binding to live Jurkat cells or NIH 3T3 cells ( Fig. 2A and Fig. S1, A-C). However, when these cells were treated with actinomycin D (ActD) to induce apoptosis and necrosis, mouse SCARA1 exhibited binding activities to the dead cells in the presence of Ca 2ϩ ( Fig.  2A and Fig. S1, A-C), and the binding is pH-independent ( Fig.  2A and Fig. S1, A-C). Similar binding characteristics were also observed for human SCARA1 with Jurkat cells (Fig. 2B and Fig.  S1D), and ELISA data also showed that SCARA1 could bind to Jurkat cell lysates (Fig. 2C). Moreover, confocal microscopy showed that SCARA1 could bind the permeabilized NIH 3T3 and HEK293 cells in the presence of Ca 2ϩ (Fig. 2, D and E, and Fig. S1, E and F), suggesting that the ligands of SCARA1 might be the naturally-existing cellular components, rather than cell death-induced products. In addition, the binding data showed that human and mouse SCARA1 appeared to have cross-activities in recognizing dead cells from humans and mice; this is not unexpected as the sequence identity between human and mouse SCARA1 is higher than 70%, and the Ca 2ϩ -binding site is also conserved for the SRCR domain from humans and mice. Furthermore, Ca 2ϩ was replaced by Mg 2ϩ , Na ϩ , or K ϩ in the binding assays, and the results showed that Mg 2ϩ could also mediate the recognition of SRCR with dead cells, but no binding was detected in the cases of Na ϩ and K ϩ (Fig. S1G), suggesting that other divalent cations may also induce the dead-cell recognition by the SRCR domain of SCARA1.
To further characterize the binding of SCARA1 with dead cells, both viable cells and the ActD-treated Jurkat cells were stained by the GFP-fused SCARA1, annexin V-APC, and propidium iodide (PI) in the presence of Ca 2ϩ (Fig. 2F), and the FACS results showed that SCARA1 had no binding to viable cells (annexin V Ϫ PI Ϫ ) but was able to recognize both apoptotic (annexin V ϩ PI Ϫ ) and necrotic cells (annexin V ϩ PI ϩ ), implying that the cellular ligands of SCARA1 were exposed at apoptotic and necrotic stages during cell death.

SCARA1 recognizes dead cells through its C-terminal SRCR domain
Based on the sequence of SCARA1, the ectodomain of SCARA1 is predicted to have a CC region, a CL region, and an SRCR domain (Fig. 1A). To locate the dead cell-recognizing region of SCARA1, we expressed a series of truncation mutants for binding assays. The results showed that the fragment containing the CC and the CL regions of SCARA1 had no binding to dead cells either in the presence or absence of Ca 2ϩ (Fig. 3A and Fig. S2A). By contrast, a truncation mutant containing both the CL region and the SRCR domain showed strong binding to dead cells in the presence of Ca 2ϩ at both acidic and basic pH values ( Fig. 3B and Fig. S2B). These results were also confirmed by dot-blot assays with cell lysates (Fig. 3D) and confocal microscopy (Fig. 3, F and G, and Fig. S2, C-F).

SCARA1/CD204 recognizes dead cells through spectrin
Because the Ca 2ϩ -binding site on the SRCR domain is conserved for humans and mice, we mutated the Ca 2ϩ -binding sites of the mouse SRCR domain by generating a single mutant (E446S) and a double mutant (D379S/D380S) (Fig. 1D), which were expressed in insect cells and purified similarly. The binding data showed that both mutants exhibited no binding affinities to the ActD-treated Jurkat cells in the presence of Ca 2ϩ ( Fig. 1E and Fig. S3C). Similar results were obtained for the mutants with the frozen-thawed NIH 3T3 cells ( Fig. 1F and Fig.  S3D), confirming that the cation-binding site is indispensable in recognizing dead cells by the SRCR domain.
Moreover, ELISA data showed that the CL-SRCR fragment had similar binding affinities to dead cells with Ca 2ϩ concen-trations ranging from 2 to 20 mM (Fig. S1I), whereas no obvious binding was detected at 2 M Ca 2ϩ (Fig. S1H), which roughly corresponds to the Ca 2ϩ concentration at endosomes (46).

Trimeric form of SCARA1 increases the binding affinity to dead cells
A previous structural prediction (14) suggested that the ectodomain of SCARA1 might form a homophilic trimer on the cell surface; therefore, we expressed the ectodomain of mouse SCARA1(88 -454) in insect cells (Fig. S3A), and the purified SCARA1 ectodomains were cross-linked by glutaraldehyde and loaded onto SDS-PAGE, and indeed, it showed a band with triple molecular weight (Fig. S3B), suggesting the formation of

SCARA1/CD204 recognizes dead cells through spectrin
homotrimers. Because the SRCR domain alone is a monomer as shown in the crystal structure, the SCARA1 homotrimer is formed through the CC and the CL regions (Fig. 1A). Interestingly, FACS data showed that the monomeric SRCR domain only had weak binding to dead cells (Fig. 3, H and I, and To validate the effect of trimerization of SCARA1 on deadcell recognition, we expressed an artificial homotrimer of SRCR by using a de novo designed trimeric tag, PDB entry 5J0J (47), to replace the CC-CL region of SCARA1 (Fig. S2G). The results

SCARA1/CD204 recognizes dead cells through spectrin
showed that the artificial SRCR trimer was able to recognize dead cells efficiently ( Fig. 3C and Fig. S2H) and had similar binding activities to the trimeric SCARA1 ectodomain (Fig. 2, A and B, and Fig. S1, A-D) and the CL-SRCR fragment ( Fig. 3B and Fig. S2B); thus, supporting the formation of the SCARA1 homotrimer could increase the binding affinity to dead cells significantly.

SCARA1 recognizes protein ligands on dead cells
To identify the ligands of SCARA1 on dead cells, we treated dead cells with a number of enzymes, including protease K, DNase I, and RNase A. The results showed that only the treatment of protease K could block the binding of SCARA1 to dead cells (Fig. 3J), suggesting that SCARA1 might recognize dead cells through protein ligands. We also tested the potential lipidbinding activities of SCARA1 by dot-blot assays, and no obvious binding was detected (Fig. 3E). Furthermore, we treated HEK293 cell lysates with protease K at different concentrations and with different incubation times in the dot-blot assays (Fig.  S4A). The results showed that the binding of SCARA1 to dead cells diminished gradually as the protease K concentration and incubation time increased, confirming that SCARA1 may recognize protein ligands on dead cells.

SCARA1 recognizes dead cells through cellular spectrin
To isolate the protein ligands of SCARA1, purified human SCARA1 ectodomain was incubated with Jurkat cell lysates in the presence of Ca 2ϩ and then pulled down with Ni-NTA beads. The pulldown assays were also performed in the absence of Ca 2ϩ in parallel as controls. The eluates from Ni-NTA beads were analyzed by mass spectrometry (MS). Although hundreds of proteins were detected by MS, the potential ligands of SCARA1 could be identified by comparing the abundance of different proteins found in the eluates in the presence of Ca 2ϩ with the controls in the presence EDTA. Indeed, the MS results showed that cellular spectrins could be pulled down by SCARA1 in the presence of Ca 2ϩ , and the spectrin ␣-chain was the major protein showing large differences over the controls (Fig. S4B), suggesting that spectrin might be the binding target of SCARA1 on dead cells.
Furthermore, we did dot-blot assays using the purified GFP-hCL-SRCR fragment (fused with a His-tag) with the cell lysates from Jurkat cells. The cell lysates were incubated with the Ni-NTA beads bound with the purified GFP-hCL-SRCR fragments in the presence of Ca 2ϩ or EDTA, and the pulldown products were spotted onto nitrocellulose membranes and detected by anti-spectrin or anti-GFP antibodies. The results showed that spectrin could be pulled down by the GFP-hCL-SRCR fragment from the cell lysates only in the presence of Ca 2ϩ (Fig. 3K).
To verify the interactions between SCARA1 and spectrin, we expressed the ␣-chain of human nonerythrocytic spectrin in three fragments: F1 (residues 1-785), F2 (residues 784 -1549), and F3 (residues 1544 -2476) (Fig. 4A) in Escherichia coli and purified by Ni-NTA-affinity chromatography and SEC. The ELISA data showed that all three spectrin fragments were able to bind to SCARA1 in a Ca 2ϩ -dependent manner with similar binding profiles (Fig. 4B). Moreover, all three spectrin fragments were able to inhibit the binding of the CL-SRCR fragment of SCARA1 to dead cells efficiently (Fig. 4, F and G, and Fig. S5, A and B), suggesting that spectrin is the binding target of SCARA1 on dead cells. In addition, the binding data also showed that both human and mouse SCARA1 could bind to human ␣-spectrin, which is not surprising as ␣-spectrins are well-conserved between humans and mice.
To assess the exposure of spectrin during cell death, the binding of antibodies against spectrin with dead cells was monitored by FACS at different stages of cell death, and the results showed that the antibodies had no binding to viable cells (annexin V Ϫ PI Ϫ ) but could recognize both apoptotic (annexin V ϩ PI Ϫ ) and necrotic cells (annexin V ϩ PI ϩ ) (Fig. 3L), similar to the SCARA1-binding results shown above (Fig. 2F), suggesting that cellular spectrins could be exposed at both apoptotic and necrotic stages. This seems not surprising as spectrins locate at the intracellular side of the plasma membrane and therefore could be exposed when membrane asymmetry is lost during apoptosis.

SRCR domain of SCARA1 recognizes the SPEC domain of spectrin specifically
To identify the spectrin-binding domain of SCARA1, three fragments of SCARA1, including the SRCR, the CL-SRCR, and the CC-CL fragment, were expressed in insect cells and applied for dot-blot assays ( Fig. 4, C-E). The results showed that both the SRCR and the CL-SRCR fragments could bind to all three spectrin fragments in the presence of Ca 2ϩ , whereas the CC-CL fragment had no binding to the three fragments, suggesting that the SRCR domain might be the spectrin-binding domain of SCARA1. Sequence analyses suggest that the ␣-chain of human spectrin contains 19 SPEC domains (SPEC1 H, monomeric SRCR domain of mSCARA1 shows weaker binding to the ActD-treated Jurkat cells than the trimeric mSCARA1 ectodomain. GFP is applied as a control. I, monomeric SRCR domain of hSCARA1 shows weaker binding to the ActD-treated Jurkat cells than the trimeric hSCARA1 ectodomain. J, binding of mSCARA1 to dead cells is abolished by the protease K treatment. K, dot-blot assays show that spectrin can be pulled down by GFP-hCL-SRCR fragment from the Jurkat cell lysates. The cell lysates were incubated with the Ni-NTA beads bound with the purified GFP-hCL-SRCR fragments in the presence of Ca 2ϩ or EDTA, and the pulldown products were spotted onto nitrocellulose membranes and detected by anti-spectrin or anti-GFP antibodies. The pulldown products from the empty Ni-NTA beads as well as the GFP-hCL-SRCR fragment alone were also spotted as controls. L, staining of viable (left) and the ActD-treated (middle) Jurkat cells by annexin V-APC, PI, and anti-spectrin antibodies. The binding of anti-spectrin antibodies to the gated subsets of viable and the ActD-treated Jurkat cells is shown on the right. Mock represents untreated viable cells.

SCARA1/CD204 recognizes dead cells through spectrin
to SPEC19), an SH3 domain, and an EF-hand domain (Fig. 4A) (37,48). The sequence identity among the individual SPEC domains is around 30%. To narrow down the binding region of SCARA1 on spectrin, we randomly chose three representative SPEC domains, including SPEC1, SPEC11, and SPEC9 (SPEC9 was chosen as it has a longer sequence among the SPEC domains), as well as the SH3 domain and the EF hand domain for binding assays (Fig. 4A). The ELISA results showed that both mouse and human SCARA1 could bind to SPEC9 and SPEC11 in the presence of Ca 2ϩ , rather than the SH3 or the EF-hand domain of spectrin ( Fig. 5A and Fig. S6A). Furthermore, the FACS data showed that both SPEC9 and SPEC11 could block the binding of SCARA1 to dead Jurkat cells almost completely ( Fig. 5B and Fig. S6, B-D), but the SH3 and the EF hand domains had no inhibition to dead-cell recognition, confirming that the SPEC domains of spectrin are the binding targets of SCARA1 on dead cells. The binding/inhibition experiments were also repeated using HEK293 and NIH 3T3 cells, and similar results were obtained (Fig. 5, C and D, and Fig. S6, E and F). Moreover, two mutants of the CL-SRCR fragment, D379S/ D380S and E446S, where the Ca 2ϩ -binding sites on the SRCR domain were mutated, were also applied for ELISAs, and the results showed that both mutants had no detectable binding to the SPEC domains (Fig. 5E), thus validating the Ca 2ϩ -dependent binding of the SPEC domain by the SRCR domain of SCARA1. Furthermore, the binding affinities between SRCR (monomer), CL-SRCR (trimer), and 5J0J-SRCR (artificial trimer, see Fig. S2G) with the SPEC domain were estimated based on the ELISAs (Fig. 5H). The results showed that the trimerized SRCR domains, CL-SRCR and 5J0J-SRCR, had higher affinities for spectrin than the monomeric SRCR, consistent with the FACS results shown above.
Because spectrins are usually categorized as erythrocytic spectrins and nonerythrocytic spectrins (49), we also expressed the SPEC domains, SPEC9 e and SPEC11 e , from human erythrocytes for binding assays. The ELISA results showed that both human and mouse SCARA1 could bind to SPEC9 e and SPEC11 e , similar to the SPEC domains from nonerythrocytic spectrin (Fig. 5F). The FACS data also showed that both SPEC9 e and SPEC11 e could block the binding of SCARA1 to dead cells efficiently as the nonerythrocytic SPEC domains ( Fig. 5G and  Fig. S6, G and H), suggesting that SCARA1 might be a generic spectrin receptor for both erythrocytes and nonerythrocytes.

SCARA1-transfected cells recognize spectrin though the SPEC domain
To test the recognition of spectrin by SCARA1 at cellular level, the SCARA1-transfected HEK293 cells were incubated with either the SPEC11 e from erythrocytes or the SPEC17-18 fragment from nonerythrocytes. The FACS data showed that the HEK293 cells transfected with the full-length SCARA1 could recognize SPEC11 e and SPEC17-18 specifically, and similar results were obtained for both human and mouse SCARA1 (Fig. 6, A and B), whereas the nontransfected cells had no binding to SPEC11 e and SPEC17-18 (Fig. 6A), suggesting that SCARA1 expressed on the cell surface can recognize spectrin in the presence of Ca 2ϩ . Furthermore, the fluorescent images of the SCARA1-transfected cells by confocal microscopy showed that the SPEC domains and SCARA1 were well co-localized on the cell surface and cytoplasm in the presence of Ca 2ϩ (Fig.  6C), suggesting that the SPEC domains can be internalized by SCARA1.

Macrophages internalize dead cells from both erythrocytes and nonerythrocytes through SCARA1-spectrin interactions
Before endocytosis assays, we tested the binding of SCARA1 with the purified mouse erythrocytes (Ͼ99%) (Fig. S7). The results showed that mouse SCARA1 had no binding to the healthy erythrocytes (Fig. S7, A and B), but the frozen-thawed erythrocytes could be recognized by both mouse and human SCARA1 in a Ca 2ϩ -dependent manner (Fig. S7, A and B). The confocal images of the permeabilized erythrocytes stained by mouse or human SCARA1 revealed the recognizable pattern of spectrin on erythrocytes (Fig. S7, C and D) (36,50). Moreover, FACS data showed that SPEC11 e could block the binding of SCARA1 to dead erythrocytes almost completely (Fig. S7, E and  F), suggesting that the SPEC domains of spectrin are the binding targets of SCARA1 on dead erythrocytes.
To examine the internalization of dead cells by the interaction between SCARA1 and spectrin, macrophages were applied for endocytosis assays. The expression of SCARA1 on the surface of macrophages was confirmed by the binding of anti-SCARA1 antibodies using flow cytometry (Fig. S8), which is consistent with the published data showing that SCARA1 is highly expressed on macrophages (14). Then either the GFPtagged spectrin fragments or the ultra-sonication-treated erythrocytes or nonerythrocytes (HEK293 cells or Jurkat cells) were fed to macrophages for internalization. The FACS data showed that both the spectrin fragments and the dead cells could bind to macrophages (Fig. 6, D and E), and the binding could be blocked by the CL-SRCR fragment of SCARA1 almost completely (Fig. 6, D and E). Moreover, the confocal images showed that both the GFP-tagged spectrin fragments and the spectrins from the dead cells could be identified inside macrophages, and the internalization could be inhibited by the preincubation with the CL-SRCR fragment of SCARA1 (Fig. 6F), thereby validating the endocytosis of dead cells via the SCARA1spectrin interaction by macrophages.  (F1, F2, and F3) expressed for binding assays are labeled. B, ELISAs show the binding of the spectrin fragments to hSCARA1. Sumo is applied as a control. C, dot-blot assays show the binding of spectrin F1 fragment to the SRCR and the CL-SRCR fragments of mSCARA1 or hSCARA1, whereas the F1 fragment has no detectable binding to the CC-CL fragment of SCARA1. SCARA1 fragments were spotted onto nitrocellulose membranes, and then the spectrin F1 fragment was incubated with the membranes in the binding buffer, including Ca 2ϩ or EDTA. Anti-GFP or anti-spectrin antibodies were used for detection. Similar procedures were followed for F2 (D) and F3 (E) fragments. D, dot-blot assays show the binding of spectrin F2 fragment to the SRCR and the CL-SRCR fragments of mSCARA1 or hSCARA1, whereas the F2 fragment has no detectable binding to the CC-CL fragment of SCARA1. E, dot-blot assays show the binding of spectrin F3 fragment to the SRCR and the CL-SRCR fragments of mSCARA1 or hSCARA1, whereas the F3 fragment has no detectable binding to the CC-CL fragment of SCARA1. F, spectrin fragments (F1, F2, and F3) block the binding of the CL-SRCR fragment of mSCARA1 to dead cells. Sumo is applied as a control. G, spectrin fragments (F1, F2, and F3) block the binding of the CL-SRCR fragment of hSCARA1 to dead cells.

Discussion
Although SRs were initially identified as receptors for modified LDL (2), evidence has shown that the ligands of SRs could be rather diverse, which is not surprising as numerous substances such as dead cells and metabolites need to be cleaned promptly to maintain homeostasis. Despite that PS receptors are widely-known for recognizing apoptotic cells, recent evidence shows that several SRs are able to recognize dead cells through different ligands. For example, Clec9A binds F-actin of the damaged cells (30,31), and DEC205/CD205 recognizes dead cells through keratins at the acidic environment (29). Previous studies have shown that SCARA1 might be involved in the phagocytosis of dead cells (14), but the mechanisms remain unclear. Here, we find that SCARA1 can recognize dead cells specifically though spectrin with its C-terminal SRCR domain, representing a novel pathway of dead-cell recognition. The biochemical data show that the trimerization of SCARA1 enhances the dead-cell-binding affinity significantly, and this is not entirely unexpected as spectrins are assembled into long fibrous structures and integrated into the cytoskeleton network, and thus the trimeric SCARA1 may have advantages in spectrin uptake by providing higher binding affinities and larger mechanic force during dead-cell clearance (Fig. 6G).
SRCR domains are commonly found in SRs with a relativelyconserved fold (16), and the cation-binding sites on SRCR are usually associated with ligand recognition. For example, the cation-binding sites on the SRCR domain of SCARA2, another member of SR-A, are involved in the binding with modified LDL (13), and the Ca 2ϩ -binding site on the SRCR domain of CD163 is important for recognizing the hemoglobinhaptoglobin complexes (51). Therefore, it is not surprising that Ca 2ϩ is required for the ligand binding of the SRCR domain of SCARA1, and it might be involved in ligand recognition directly at the active site. In addition, cell death is often associated with the change of Ca 2ϩ levels (52,53). The extracellular Ca 2ϩ is usually at millimolar levels (46,54), which would allow the spectrin recognition by SCARA1. However, it has been shown that the Ca 2ϩ levels at endosomes could be much lower (46), implying that the ligands would be released from SCARA1 at endosomes, which is consistent with the binding data shown above.
Spectrin is an essential component of the cytoskeleton and is expressed in most of the eukaryotes (37,41). Spectrin usually locates near the intracellular side of the plasma membrane and forms a network by associating with other cytoskeleton components in the cytoplasm. It is known that during apoptosis, the plasma membrane asymmetry is lost and leads to the exposure of the inner leaflet components such as PS; therefore, it would be expected that spectrin could also be exposed at this stage, as has been confirmed by the antibody-binding data shown above. This is in contrast with other known intracellular dead-cell markers such as keratins and F-actins, which are largely exposed at the necrotic stage of cell death (29 -31). Considering the wide distribution of spectrin in eukaryotes and the high expression of SCARA1 on macrophages, the SCARA1-spectrin interaction might be a generic pathway for dead-cell recognition and clearance under physiological conditions (Fig. 6G). Moreover, the findings of spectrin, together with actins and keratins as cellular markers for dead cells, suggest that the cytoskeleton is not only a scaffold for cellular structures, but also functions as universal markers for dead cells.
Erythrocytes have a life span of about 120 days, and the removal of aged or dead erythrocytes is largely controlled by macrophages (55). However, it is not entirely clear how erythrocytes are cleaned by the interactions with macrophages, as erythrocytes are lacking the organelles normally found in eukaryotes and do not have the usual apoptotic pathways. Spectrin is one of the major proteins expressed in erythrocytes and is critical for maintaining the membrane structure of erythrocytes. It has been shown that the loss of membrane asymmetry also occurs for erythrocytes during eryptosis (56), and therefore, spectrin could be exposed and mediate the interactions with macrophages. In addition, it has been shown that during eryptosis, erythrocytes would undergo membrane blebbing and Ca 2ϩ leakage (57), which would allow the binding of spectrin by SCARA1 on macrophages during the clearance of erythrocytes.
Because billions of cells are renewed every day in the human body, dead-cell clearance is critical for homeostasis. In the past decades, researchers have been largely focused on the recognition and clearance of apoptotic cells, especially the interactions between the PS receptors and apoptotic cells (26,28). The emerging evidence has shown that efferocytosis could be much more complex than previous thought, and the receptors such as Clec9A, DEC205/CD205, and SCARA1/CD204 are involved in the clearance of dead cells by recognizing intracellular protein markers, which could be exposed at apoptotic or necrotic stages. This is not surprising as more cell-death pathways such as necroptosis and pyroptosis have started to be characterized (58,59), suggesting that a large number of dead cells could be generated under different circumstances, including inflammation or infection, which often lead to the exposure or release of intracellular components. Therefore, the receptors recognizing intracellular components would be necessary to remove dead cells or cell debris efficiently in these cases. However, because the expression of the intracellular dead-cell markers could vary for different cell types, these receptors may play different roles in the clearance of dead cells in different tissues or environments. Therefore, clarifying the mechanisms of these pathways

SCARA1/CD204 recognizes dead cells through spectrin
would not only improve the understanding of efferocytosis, but would also facilitate the therapeutic strategies against related diseases in the future.

Protein expression and purification
The ectodomain of SCARA1 contains a CC region (residues 88 -262 for mouse SCARA1; residues 84 -262 for human SCARA1), a CL region (residues 262-352 for mouse SCARA1; residues 263-347 for human SCARA1), and a SRCR domain (residues 348 -454 for mouse SCARA1; residues 348 -451 for human SCARA1). Constructs encoding the ectodomain of SCARA1, CL-SRCR, CC-CL, SRCR, and 5J0J-SRCR were all subcloned into the pFastBac vectors with a melittin signal sequence and an N-terminal His 6 -tag and expressed in insect cells (Invitrogen). The GFP-tagged SCARA1 and its truncation mutants were also subcloned with a melittin signal sequence and an N-terminal His 8 -tag into the pFastBac vectors for expression. The Sf9 cells were used for generating recombinant baculoviruses, and High Five cells were used for protein production (Invitrogen). The infected cells were cultured for 3 days in a 27°C humidified incubator. The supernatants of the infected High Five cells were buffer-exchanged with 50 mM Tris, 150 mM NaCl, pH 8.0, and then applied to Ni-NTA chromatography following the manufacturer's instruction (Ni-NTA Superflow, GE Healthcare). The imidazole eluates were further purified by gel-filtration chromatography with a Superdex 200 column or a Superdex 6 column (GE Healthcare).
The full length of SCARA1 (mouse SCARA1, residues 1-454; human SCARA1, residues 1-451) was subcloned into the pTT5 expression vectors with mCherry inserted either between the ectodomain and the transmembrane domain of hSCARA1 or at the N terminus of mSCARA1. The constructs were transiently transfected into HEK293F cells with FreeStyle 293 Expression Medium (Gibco). The SCARA1-transfected cells were cultured in a humidified CO 2 incubator at 37°C for 24 h before flow cytometry and confocal microscopy.
The truncation mutants of human spectrin F1, F2, and F3 were expressed in E. coli BL21 DE3 cells (Novagen) using the pET28a expression vector with an N-terminal Sumo-tag and a His 6 -tag, and then purified as soluble proteins from the supernatants of cell lysates by Ni-NTA chromatography followed by gel-filtration chromatography with a Superdex 200 column (GE Healthcare). The individual domains or fragments of human spectrin, including SPEC9 e and SPEC11 e from erythrocytes and SPEC9, SPEC11, SPEC17-18, SH3, and EF hand from nonerythrocytes, fused with an N-terminal Sumo or GFP-tag and a His 6 -tag were also expressed and purified similarly.

Cross-linking experiment
The purified mSCARA1 ectodomain (25 mM Hepes, 150 mM NaCl, pH 7.4) was treated with 0.75% (w/v) glutaraldehyde for 45 min at room temperature and then 2 mM glycine was added to terminate the cross-linking reaction. The samples were resuspended in 25 l of SDS loading buffer, boiled for 10 min, and loaded onto SDS-PAGE (Bio-Rad) for separation and detection.

Crystallization and structural determination
The SRCR domain of mSCARA1 purified from insect cell supernatants was buffer-exchanged into 10 mM Tris, 150 mM NaCl, 10 mM CaCl 2 , pH 7.4, at 10 mg/ml. Crystal screening was performed by the hanging-drop vapor diffusion method, and crystals were obtained at 4°C in a solution containing 20% (v/v) PEG 300, 5% (w/v) PEG 8000, 10% glycerol, and 0.1 M Tris, pH 8.5. Diffraction data were collected at BL18U beamline at Shanghai Synchrotron Radiation Facility (SSRF) and processed using the HKL-3000 package (60). The structure was solved by molecular replacement using the structure of M2BP (PDB entry 1BY2) as a search model. The iterative-build OMIT map procedure is applied for removing model bias (61). Coot (62) and PHENIX (63) were used for structural refinement. The crystallographic statistics are listed in Table S1. Figures were made using UCSF Chimera (64).

Apoptotic and necrotic cell preparation
Jurkat cells were cultured in RPMI 1640 medium (Gibco) supplemented with 10% (v/v) FBS (HyClone Laboratories). NIH 3T3 cells were cultured in DMEM (Gibco) supplemented with 10% (v/v) FBS (HyClone Laboratories). To induce apoptosis and necrosis, Jurkat cells or NIH 3T3 cells were both incubated in tissue culture flasks for 12-16 h with 1 g/ml ActD until use. For inducing apoptosis and necrosis of HEK293 cells, the cells were cultured in FreeStyle 293 medium (Gibco), including apoptosis inducer A (Apopida) (1:1000 (v/v)) (Beyotime) for 16 h. The frozen-thawed cells (3T3 cells or HEK293 cells) were prepared by incubating in a dry-ice ethanol bath for 10 min and then transferring immediately into a 37°C water bath for 10 min and repeated three times.

Flow cytometry
Apoptotic and necrotic cells were monitored using annexin V apoptosis detection kit APC (eBioscience, Inc.). Briefly, cells were washed three times with the washing buffer (10 mM Hepes at pH 7.4, 150 mM NaCl), then with the binding buffer (10 mM Hepes at pH 7.4, 150 mM NaCl, 10 mM CaCl 2 ), then resuspended in the binding buffer including 5 l of annexin V-APC  , 5 m). G, cartoon representation of dead cell recognition by SCARA1/CD204 via spectrin.

SCARA1/CD204 recognizes dead cells through spectrin
at 1-5 ϫ 10 6 cells/ml, and incubated at 4°C for 20 min. Then the cells were washed in the binding buffer and resuspended in 400 l of binding buffer including 5 l of PI-staining solution and analyzed by flow cytometry.
For the triple-staining assays, the cells were washed three times in the buffer (25 mM Hepes, 150 mM NaCl, 10 mM CaCl 2 , pH 7.4), then resuspended in the binding buffer (10 mM Hepes at pH 7.4, 150 mM NaCl, 10 mM CaCl 2 ) including the GFPtagged mSCARA1 and 5 l of annexin V-APC solution, and incubated at 4°C for 30 min. Then the cells were washed in the binding buffer and stained with PI similarly as described above for flow cytometry.
For GFP staining, the cells were washed as described above and then washed three times with the corresponding buffers (25 mM Hepes, 150 mM NaCl, 10 mM CaCl 2 or 1 mM EDTA, pH 7.4, or 25 mM BisTris, 150 mM NaCl, 10 mM CaCl 2 or 1 mM EDTA, pH 6.0) for different assays. Buffers with different Ca 2ϩ concentrations (2 M, 2-10 mM) were used for Ca 2ϩ gradient assays. The cells were incubated with the GFP-tagged mouse or human SCARA1 fragments in the buffers (pH 7.4 or 6.0) at room temperature for 20 min, then washed with the buffers (pH 7.4 or 6.0) three times, and stained with PI for flow cytometry.
For the cation assays, the ActD-treated Jurkat cells were washed three times with buffer A (25 mM Hepes, 150 mM NaCl, 1 mM EDTA, pH 7.4) and three times with buffer B (25 mM Hepes, 150 mM NaCl, pH 7.4). Then the buffers (25 mM Hepes, 150 mM NaCl, pH 7.4) containing Ca 2ϩ or Mg 2ϩ or Na ϩ or K ϩ at 5 mM concentrations were added to the cells for binding assays. The cells were incubated with the GFP-tagged human SCARA1 in the corresponding buffers at room temperature for 20 min, then washed with the corresponding buffers three times, and stained with PI for flow cytometry.
For the enzymatic treatment assays, the cells were washed three times with the buffer (25 mM Hepes, 150 mM NaCl, pH 7.4) and then treated with DNase I, RNase A, or protease K at concentrations of 10 g/ml for 30 min, respectively. After washing three times with the binding buffer (25 mM Hepes, 150 mM NaCl, 10 mM CaCl 2 at pH7.4), the cells were incubated with the GFP-tagged mouse SCARA1 in the binding buffer at room temperature for 20 min. After washing three times with the binding buffer, the cells were then stained with PI for flow cytometry.
For the spectrin inhibition assays, the cells were washed three times with the buffers (25 mM Hepes, 150 mM NaCl, 10 mM CaCl 2 , or 1 mM EDTA, pH 7.4, or 25 mM BisTris, 150 mM NaCl, 10 mM CaCl 2 , or 1 mM EDTA, pH 6.0) and incubated with the GFP-tagged mouse or human SCARA1 fragments (10 g/ml) with or without the fragments of spectrin (20 g/ml) at room temperature for 20 min. After washing three times with the corresponding buffer, the cells were stained with PI for flow cytometry.
For the endocytosis assay, HEK293 cells were transiently transfected with the full-length SCARA1 fused with an mCherry-tag. After 24 h, 10 g of GFP or GFP-tagged SPEC fragments (SPEC11 e or SPEC17-18) was added to the culture medium containing 10 mM Ca 2ϩ . Similar conditions were applied for the nontransfected HEK293 cells as controls. After 2-4 h, cells were washed twice with the washing buffer (25 mM Hepes, 150 mM NaCl, 10 mM CaCl 2 ) and stained with PI for flow cytometry.
FACS data were acquired using a BD Biosciences FACS Caliber flow cytometer with CellQuest TM software. Data analysis was performed using FlowJo software (Tree Star, Inc.).

Dot-blot assay
Cell lysates were prepared by ultra-sonication and spotted onto nitrocellulose membranes (Whatman) according to the manufacturer's instruction. The membranes were air-dried at room temperature for 2 h and blocked in blocking buffer (25 mM Hepes, 150 mM NaCl, 5% (w/v) BSA, 0.1% Tween 20, pH 7.4) for at least 1 h. Then, the GFP-tagged SCARA1 fragments (10 g/ml) were applied to the membranes and incubated with the mouse anti-GFP antibody (Abcam, ab184601) for 1 h, then incubated with the goat anti-mouse IgG secondary antibody HRP conjugates (SAB, L3032-2) for 1 h, and detected with the DAB reagent. Between every two steps, the membranes were washed six times with the washing buffer (25 mM Hepes, 150 mM NaCl, 10 mM CaCl 2 , 0.1% Tween 20, pH 7.4) for 5 min each.
For the lipid strip dot-blot assays, the phospholipids were purchased from Echelon Biosciences. About 2 g of Jurkat cell lysates were also spotted as a positive control. The strips were incubated overnight in a blocking buffer (25 mM Hepes, 150 mM NaCl, 5% (w/v) BSA, 0.1% Tween 20, pH 7.4) at 4°C, then transferred into a blocking buffer containing the GFP-tagged SCARA1 (10 g/ml) for 2 h at room temperature, and then the similar staining procedures were carried out as described above.
For the SCARA1 ligand detection, about 2 g of HEK293 cell lysates prepared by ultra-sonication were treated with or without protease K at different conditions and different times and then applied for dot-blot assays with similar procedures and reagents as described above.
For the dot-blot assays of the SCARA1 fragment pulldown products, the Jurkat cell lysates were incubated with the Ni-NTA beads bound with the purified GFP-hCL-SRCR fragments (fused with a His-tag) in the presence of Ca 2ϩ or EDTA. Then, the pulldown products were spotted onto nitrocellulose membranes (Whatman) and blocked similarly as described above. The pulldown products from the empty Ni-NTA beads as well as the GFP-hCL-SRCR fragment alone were spotted as controls. Then, the membranes were incubated with mouse anti-GFP antibody (Abcam, ab184601) or mouse anti-spectrin antibody (Abcam, ab11755) for 1 h at room temperature and then incubated with goat anti-mouse IgG secondary antibody HRP conjugates (SAB, L3032-2) for 1 h following the similar staining procedures described above.

SCARA1/CD204 recognizes dead cells through spectrin
For the binding between the purified SCARA1 fragments and spectrin fragments, SCARA1 fragments were spotted onto nitrocellulose membranes (Whatman) and blocked similarly as described above. Then spectrin fragments (10 g/ml) were applied to the membranes in the binding buffer (25 mM Hepes, 150 mM NaCl, 10 mM CaCl 2 , or 1 mM EDTA, 0.1% Tween 20, pH 7.4), incubated with mouse anti-GFP antibody (Abcam, ab184601) or mouse anti-spectrin antibody (Abcam, ab11755) for 1 h, and then stained with the similar procedures described above.

Mass spectrometry
The solution sample from the pulldown experiments was precipitated and resolved by 8 M urea and then treated with 5 mM tris(2-carboxyethyl)phosphine and 10 mM iodoacetamide to reduce the disulfide bonds and alkylate the resulting thiol groups, sequentially. The mixture was digested for 16 h at 37°C by trypsin at an enzyme-to-substrate ratio of 1:50 (w/w). The trypsin-digested peptides were loaded on an in-house packed capillary reverse-phase C18 column (15 cm length, 100-m inner diameter ϫ 360-m outer diameter, 3 m particle size, 100 Å pore diameter) connected to a Thermo Easy-nLC1000 HPLC system. The samples were analyzed with a 90-min HPLC gradient from 0 to 80% of buffer B (buffer A: 0.1% formic acid in water; buffer B: 0.1% formic acid in acetonitrile) at 300 nl/min. The eluted peptides were ionized and directly introduced into a Q-Exactive mass spectrometer using a nano-spray source. Survey full-scan MS spectra (from m/z 300 to 1,800) were acquired in the Orbitrap analyzer with resolution r ϭ 70,000 at m/z 200. Protein identification was done with Proteome Discoverer2.1 (65,66). The PSM value represents the number of secondary mass spectrum identified in the protein group, which can be used to estimate the relative content of the protein.

ELISA experiments
For the interaction of mSCARA1 with cell lysates, Jurkat cell lysates were prepared by ultra-sonication and coated onto 96-well plates with ϳ2 g of protein per well at 4°C overnight. The plates were then blocked with the blocking buffers (25 mM Hepes, 150 mM NaCl, 0.1% Triton X-100, 2% (w/v) BSA, pH 7.4) at 37°C for 3 h. The purified mSCARA1 was serially diluted and added to each well in the binding buffer (25 mM Hepes, 150 mM NaCl, 10 mM CaCl 2 or 1 mM EDTA, 0.1% Triton X-100, 2 mg/ml BSA, pH 7.4). In the calcium concentration gradient experiment, different Ca 2ϩ concentrations (2, 5, 10, and 20 mM) were used to replace 10 mM Ca 2ϩ . After 2 h of incubation at 37°C, the plates were washed five times with the washing buffer (25 mM Hepes, 150 mM NaCl, 10 mM CaCl 2 or 1 mM EDTA, 0.1% Triton X-100, pH 7.4) and then incubated with the mouse anti-GFP antibody (Abcam, ab184601) for 1 h. After washing, the plates were incubated with the goat anti-mouse IgG secondary antibody HRP conjugates (SAB, L3032-2) for 1 h. After washing, 100 l of chromogenic substrate (1 g/ml tetramethylbenzidine, 0.006% H 2 O 2 in 0.05 M phosphate citrate buffer, pH 5.0) was added to each well and incubated for 30 min at 37°C. Then, 50 l of H 2 SO 4 (2.0 M) was added to each well to stop the reactions. The plates were read at 450 nm on a Synergy Neo machine (BioTek Instruments). For the Ca 2ϩ gradient assays, 2-20 mM CaCl 2 were used in the ELISA experiments.
For the interactions between the SCARA1 fragments and spectrin fragments, about 1 g of the purified SCARA1 fragments were coated onto each well of 96-well plates, and the spectrin fragments were serially diluted and added to each well. Mouse anti-spectrin antibody (Abcam, ab11755) and goat antimouse IgG secondary antibody HRP conjugates (SAB, L3032-2) were used for binding detection following the similar procedures described above.
For the interactions between the SCARA1 fragments (mSRCR, mCL-SRCR, and 5J0J-mSRCR) with the SPEC domain of spectrin, about 1 mol of the purified SPEC1 was coated onto each well of 96-well plates, and the SCARA1 fragments were serially diluted and added to each well. Mouse anti-GFP antibody (Abcam, ab184601) and goat anti-mouse IgG secondary antibody HRP conjugates (SAB, L3032-2) were used for binding detection following the similar procedures described above. The K D values were calculated based on the fitting of the sigmoidal curves using software GraphPad Prism 6 (67, 68).

Confocal microscopy
HEK293 or NIH 3T3 cells grown on coverslips were fixed by 4% paraformaldehyde in TBS (pH 7.4). After washing with the buffer (25 mM Hepes, 150 mM NaCl, 10 mM CaCl 2 , or 1 mM EDTA, pH 7.4, or 25 mM BisTris, 150 mM NaCl, 10 mM CaCl 2 , or 1 mM EDTA, pH 6.0), the cells were permeabilized with 0.25% Triton X-100 in the corresponding buffers mentioned above. After washing twice with 0.1% Triton X-100 in the buffers, the cells were stained with the GFP-tagged SCARA1 fragments for 2 h in the binding buffer (25 mM Hepes, 150 mM NaCl, 10 mM CaCl 2 , or 1 mM EDTA, 0.1% Triton X-100, pH 7.4, or 25 mM BisTris, 150 mM NaCl, 10 mM CaCl 2 , or 1 mM EDTA, 0.1% Triton X-100, pH 6.0). Then, the cells were washed twice with the buffers, incubated with 5 M DAPI for 30 min, and then washed twice again for confocal microscopy. The confocal images were taken on a Leica SP8 microscope.
For cell-binding assays of spectrin, HEK293 cells were transfected with the full-length SCARA1 fused with a mCherry tag in six-well plates. After 24 h of transfection, 10 g of GFP-tagged SPEC domains (SPEC11 e from erythrocyte and SPEC17-18

SCARA1/CD204 recognizes dead cells through spectrin
from nonerythrocyte) were added to the plates with 10 mM Ca 2ϩ . After 2-4 h of incubation, the cells were washed three times with the buffer (25 mM Hepes, 150 mM NaCl, 10 mM CaCl 2 ) and incubated with 5 M DAPI for 30 min. Then the plates were washed again for confocal microscopy with a Leica SP8 microscope.

Erythrocyte assays
Isolated fresh mouse erythrocytes (99%) were purchased (Bersee, Beijin, China) and stored in the Archer's fluid. Before the experiment, the erythrocyte cells were cultured in RPMI 1640 medium (Gibco) supplemented with 10% (v/v) FBS (HyClone Laboratories) overnight. The frozen-thawed cells were prepared by incubating cells in a dry-ice ethanol bath for 10 min and then transferring immediately into a 37°C water bath for 10 min and repeated three times. For GFP staining, the cells were washed three times with the washing buffer (10 mM Hepes at pH 7.4, 150 mM NaCl) and then washed three times with the binding buffers (25 mM Hepes, 150 mM NaCl, 2 mM CaCl 2 , pH 7.4). Then cells were blocked with blocking buffer (25 mM Hepes, 150 mM NaCl, 2 mM CaCl 2 , 5% (v/v) FBS, 3% BSA, 0.1% Tween 20, pH 7.4) for 1 h and stained with the GFPtagged mouse or human SCARA1 fragments, anti-Ly76 antibody (Abcam, ab91113), and goat anti-rat IgG H&L (Abcam, ab150160) in the binding buffer at room temperature for 1 h, and then washed with the buffers (pH 7.4) three times and analyzed by flow cytometry. For inhibition experiments, both mouse and human SCARA1 are blocked with SPEC11 e for about 4 h in 4°C.
For confocal microscopy assay, erythrocytes on coverslips were fixed by 4% paraformaldehyde in TBS (pH 7.4). After washing with the buffer (25 mM Hepes, 150 mM NaCl, 2 mM CaCl 2 or 1 mM EDTA at pH 7.4), the cells were permeabilized with 0.25% Triton X-100 in the corresponding buffers mentioned above. After washing twice with 0.1% Triton X-100 in the buffers, the cells were stained with the GFP-tagged SCARA1 fragments for 2 h in the binding buffer (25 mM Hepes, 150 mM NaCl, 2 mM CaCl 2 or 1 mM EDTA, 0.1% Triton X-100, at pH 7.4). Then the cells were washed twice with the buffers, incubated with 5 M DAPI for 30 min, and then washed twice again for confocal microscopy. The confocal images were taken on a Leica SP8 microscope.
Mouse erythrocytes, HEK293, cells or Jurkat cells were treated by ultra-sonication with low intensity and centrifuged briefly to remove fractions that may interfere with imaging, and then incubated with anti-spectrin antibody (Abcam, ab11755) and goat anti-mouse IgG H&L Alexa Fluor 488 (Abcam, ab150113) before feeding to macrophages blocked with 10% (v/v) FBS, 3% BSA for 1 h and incubated in the binding buffer (DMEM supplemented with 10% (v/v) FBS) at 37°C for 4 h before flow cytometry and confocal microscopy. The GFPtagged spectrin fragments (SPEC11 e and SPEC 17-18) were also applied for endocytosis assays. For flow cytometry, the macrophages were washed three times with the washing buffer (25 mM Hepes, 150 mM NaCl, 10 mM CaCl 2 ) and stained with PI similarly as described above. For confocal microscopy, the macrophages on coverslips were fixed by 4% paraformaldehyde in TBS (pH 7.4). After washing with the buffer (25 mM Hepes, 150 mM NaCl, 2 mM CaCl 2 ), the cells were permeabilized with 0.25% Triton X-100. After washing twice with 0.1% Triton X-100 in the buffers, the cells were incubated with 5 M DAPI for 30 min, and then washed once again for confocal microscopy. The confocal images were taken on a Leica SP8 microscope. For endocytosis inhibition assays, the GFP-tagged spectrin fragments or the ultra-sonication-treated cells were incubated with mCL-SRCR fragment of SCARA1 at 4°C for 4 h before feeding to macrophages.

Data availability
The structure of the SRCR domain of mouse SCARA1 was deposited in the PDB with accession number 6J02.