The C-terminal acidic region of calreticulin mediates phosphatidylserine binding and apoptotic cell phagocytosis

Abstract

Calreticulin is a calcium-binding chaperone that is normally localized in the endoplasmic reticulum (ER). Calreticulin is detectable on the surface of apoptotic cell under some apoptosis-inducing conditions, where it promotes phagocytosis and immunogenicity of dying cells. However, the precise mechanism by which calreticulin, a soluble protein, localizes to the outer surface of the plasma membrane of dying cells is unknown, as are the molecular mechanisms that are relevant to calreticulin-induced cellular phagocytosis. Calreticulin comprises three distinct structural domains; a globular domain, an extended arm-like P-domain and a C-terminal acidic region containing multiple low affinity calcium binding sites. We show here that calreticulin, via its C-terminal acidic region, preferentially interacts with phosphatidylserine (PS) compared to other phospholipids, and that this interaction is calcium-dependent. Additionally, exogenous calreticulin binds apoptotic cells via a higher affinity calcium-dependent mode that is acidic-region-dependent. Exogenous calreticulin also binds live cells, including macrophages, via a second, lower affinity P-and globular domain-dependent, but calcium-independent binding mode that likely involves its generic polypeptide-binding site. Truncation constructs lacking the acidic region or arm-like P-domain of calreticulin are impaired in their abilities to induce apoptotic cell phagocytosis by murine peritoneal macrophages. Taken together, the results of this investigation provide the first molecular insights into the phospholipid-binding site of calreticulin as a key anchor point for the cell surface expression of calreticulin on apoptotic cells. These findings also support a role for calreticulin as a PS-bridging molecule that co-operates with other PS-binding factors to promote the phagocytosis of apoptotic cells.

Introduction

Phagocytosis is a vital process in the maintenance of tissue homeostasis by initiating the clearance of dying cells (reviewed in (1)). Failure to clear apoptotic cells have been associated with autoimmunity (2), leading to necrosis and pathologic inflammatory responses triggered by the release of pro-inflammatory molecules from dying cells (reviewed in (3)). It is thus important to understand the cellular and molecular factors that promote the uptake of dying cells, while preserving their live counterparts.

To evade phagocytic uptake, live cells and cancer cells express anti-phagocytic ‘don’t eat me’ signals (such as CD47) on their cell surfaces, counterbalancing any pro-phagocytic markers and shifting the dynamic balance between pro- and anti-phagocytic indicators to prevent cellular uptake (reviewed in (4)). On the other hand, apoptotic cells express numerous pro-phagocytic ‘eat me’ signals (reviewed in (5)). The best-characterized factor that mediates the phagocytosis of apoptotic cells is the phospholipid phosphatidylserine (PS). While healthy cells maintain an asymmetric distribution of PS and phosphatidylcholine (PC)-containing lipids on their cell surfaces (with PS being sequestered within the inner leaflet of the plasma membrane), a hallmark of apoptosis is the exposure of PS on the outer leaflet of the cell membrane (6). Exposed PS is a potent ‘eat me’ signal, acting either via direct interaction with PS receptors on phagocytes or through bridging molecules (reviewed in (7)). A putative PS-bridging molecule is the ER chaperone calreticulin (CRT), previously shown to be expressed on the surface of apoptotic cells under certain apoptosis-inducing conditions and to induce the phagocytosis of apoptotic or dying cells (8, 9). Moreover, calreticulin-dependent phagocytic uptake of dying cancer cells is suggested to contribute to enhanced CD8⁺ T-cell responses against tumors (9). However, the molecular and structural basis for calreticulin-induced phagocytosis remains poorly defined.

Calreticulin is a soluble ER chaperone containing three domains (Fig. 1A (top left panel)); a globular domain comprising a lectin fold (10, 11); (ii) a flexible, proline-rich, arm-like domain (termed the P-domain) (12) that binds co-chaperones (13, 14); and (iii) a C-terminal acidic region of unknown structure that contains several low-affinity calcium binding sites (15) and functions as a calcium buffer in the ER (16). In the ER, calreticulin interacts with monoglucosylated glycoproteins via a glycan-binding site present within its globular domain (11, 17, 18) (Fig. 1A, top left panel; glycan-binding site). In cooperation with co-chaperones such as ERp57, calreticulin facilitates the folding of nascent client proteins such as major histocompatibility complex (MHC) class I molecules (reviewed in (19)). In addition to its defined glycan-binding role, calreticulin also contains a polypeptide-binding site in the vicinity of its glycan-binding site (Fig. 1A, top left panel; polypeptide-binding site), with the P-domain contributing to stabilization of calreticulin-polypeptide complexes (10, 20, 21).

Figure 1. Expression and characterization of calreticulin truncation constructs.

(A) (Upper left panel) Model for the structure of mCRT(1-351) based on published structures of the globular and P-domains of calreticulin and calnexin (10-12, 51) (Upper right panel) Representative SDS-PAGE gel depicting the migration positions of purified forms of various mCRT constructs used in this investigation. (Lower panel) Clustal-W2 alignments of the amino acid sequences of the murine calreticulin (mCRT) constructs used in this study. The constructs are: mCRT(WT), wild type mCRT; mCRT(ΔC), mCRT lacking the C-terminal acidic domain (residues 340-399); mCRT(ΔP), mCRT lacking the arm-like P-domain (residues 187-283); mCRT(1-351), mCRT lacking the C-terminus of the acidic C-domain (residues 352-399); mCRT(1-351 ΔP), mCRT containing the globular domain alone (lacking both the P-domain (residues 187-283) as well as the C-terminal region of the acidic C-domain (residues 352-399); GB1-mCRT(P), mCRT containing the P-domain alone (residues 187-283) expressed as a fusion with the B1 domain of protein G (GB1); GB1-mCRT(C), mCRT containing the acidic C domain alone (residues 340-399) expressed as a fusion with GB1. (B) Representative bio-layer interferometry analysis (from 2 independent experiments) depicting the binding of GB1-mCRT(P) to varying ERp57 concentrations. The calculated steady-state binding affinities (K_D) for GB1-mCRT(P) and mCRT(WT) are 0.101 ± 0.008 and 0.282 ± 0.038 μM respectively (C) Representative far-UV CD spectra (out of 2 independent experiments) of GB1-mCRT(C) or GB1 in the presence or absence of calcium. (D) Representative ITC thermograms (from 2 independent experiments) showing binding of GB1-mCRT(C) or GB1 to calcium. Plots show the raw titration curves (above) and corresponding curve fits (below). Calculated thermodynamic parameters are shown in Table I.

Currently, the interactions that mediate the cell surface expression of calreticulin in apoptotic cells are unknown. Previous studies have shown that calreticulin can bind externalized PS leaflets in a calcium-dependent manner (22, 23). Given the presence of multiple calcium binding sites within the C-terminal acidic region of calreticulin (15, 24), we hypothesized that the acidic region of calreticulin interacts with phospholipids via calcium-mediated interactions. Furthermore, acidic region-dependent interactions could serve as a mechanism to localize calreticulin to the surface of dying cells and, in turn, leave the globular and P-domains of calreticulin free to interact with phagocytic receptors. To examine these models, using various truncation constructs and isolated domains of calreticulin, we investigated the molecular basis for calreticulin-PS interactions, and for the binding of calreticulin to live and apoptotic cells. We also present data on the structural basis for calreticulin-induced phagocytosis of apoptotic cells.

Materials and Methods

Supplies and antibodies

Unless indicated, all reagents were purchased from Sigma. FITC was purchased from Fisher Scientific. Palmitoyl-oleoyl phosphatidylcholine (POPC) and palmitoyl-oleoyl phosphatidylserine (POPS) were purchased from Avanti. Oxaliplatin, cisplatin and nocodazole were purchased from Sigma. Annexin V, 4',6-diamidino-2-phenylindole (DAPI) and 7-Aminoactinomycin D (7AAD) were purchased from BD Biosciences. Rat anti-mouse CD11b conjugated to PerCP-Cy5 (550393) or Alexa Fluor 647 (557686) (clone M1/70) were purchased from BD Biosciences. Chicken anti-CRT was purchased from Thermo Scientific (PA1-902A). Donkey anti-chicken-PE and unconjugated murine IgG (Fc block) was purchased from Jackson Immunoresearch Laboratories. Goat anti-rat IgG-Texas Red (T-6392) was purchased from Invitrogen. CellTracker Green 5-chloromethylfluorescein diacetate (CMFDA) was purchased from Invitrogen.

Protein expression and purification

The globular domain of calreticulin (mCRT(1-351 ΔP)) was constructed using the Quickchange-II site directed mutagenesis kit (Agilent) as described earlier (24) to introduce a stop codon at position 352, with mCRT(ΔP) in the pMCSG7 vector serving as the template. The following primers were used:

Forward: 5`-GAG GAG CAG AGG CTT AAG TAA GAA GAA GAG GAC AAG-3`; Reverse: 5`-CTT GTC CTC TTC TTC TTA CTT AAG CCT CTG CTC CTC-3`.

The isolated arm-like P-domain and the C-terminal acidic region of calreticulin were expressed as GB1 fusions in the pGB1 vector (25) via ligation independent cloning (LIC) as described (17, 26). The following primers were used:

GB1-C-terminal acidic region (GB1-mCRT(C)): Forward: 5`- TAC TTC CAA TCC AAT GCT ATG AAG GAC AAG CAG GAT GAG GAG CAG −3`; Reverse: 5`- TTA TCC ACT TCC AAT GTT ACA GCT CAT CCT TGG CTT −3`

GB1-P-domain (GB1-mCRT(P)): Forward: 5`- TAC TTC CAA TCC AA TG CCC CAC CCA AGA AGA TAA AGG ACC CTG ATG CTG CCA AGC CG −3`; Reverse: 5`- TTA TCC ACT TCC AAT GCT AGG AGT ATT CAG GGT TGT CAA TTT CTG GGT G −3`

All calreticulin constructs expressed N-terminal hexahistidine tags and TEV cleavage sites to enable removal of the hexahistidine tag. Calreticulin was purified by nickel affinity chromatography, followed by TEV cleavage in certain experiments (Fig. 1) and size-exclusion chromatography as described previously (17, 24). Following elution from the Ni-NTA resin, GB1-mCRT(C) was purified to homogeneity via anion-exchange chromatography using DEAE sepharose beads (GE) as follows: GB1-mCRT(C) was dialyzed overnight against 20 mM HEPES pH 7.5, 70 mM NaCl and 5 mM CaCl₂. GB1-mCRT(C) was allowed to bind 5 mL DEAE resin for 2 hours at 4 °C and subsequently eluted on a NaCl step gradient ranging from 100 mM NaCl to 500 mM NaCl. Ovalbumin was purified via Ni-affinity chromatography as described earlier (17). GB1 was purified via Ni-affinity chromatography using the same protocol as calreticulin without the subsequent size-exclusion chromatography step. Purified calreticulin was conjugated to FITC at a 1:1 molar ratio as described previously for other fluorophores (21).

Isothermal titration calorimetry (ITC)

Small unilamellar vesicles (SUVs) of POPS or POPC were made in the following manner: POPC or an equimolar mixture of POPS and POPC were re-suspended in 20 mM HEPES (pH 7.5) and 10 mM NaCl to a final concentration of 3 mM. The resulting lipid solution was sonicated at room temperature in a sonicating water bath for 2-10 min until the solution cleared. Prior to use, 5mM CaCl₂ was added as needed.

ITC measurements (at 25 ºC) for calreticulin interacting with calcium were undertaken as described earlier (24) with calreticulin at a concentration of 43 μM in 20 mM HEPES (pH 7.5) and 10 mM NaCl. A stock solution of 5 mM CaCl₂ was titrated into the protein sample and the changes in enthalpy were measured. To assess lipid binding, 10 μl injections of 1.3 mM (POPC), 0.63 mM (POPS), or 3 mM (POPS or POPC) lipid stock solutions were titrated into 43 μM CRT (960 μl) in 20 mM HEPES (pH 7.5), 10 mM NaCl ± 5 mM CaCl₂, and the changes in enthalpy were measured. Twenty-five total injections were performed and the data analyzed using NanoAnalyze (ver 2.2.0) (TA instruments) via non-linear least squares regression using an independent ligand binding model.

Circular dichroism (CD) spectroscopy

Far-UV CD measurements (at 21 ºC) of GB1-mCRT(C) or GB1 alone were undertaken as described earlier for other calreticulin constructs (24). In these analyses, GB1-mCRT(C) or GB1 was at a concentration of 0.1 mg/mL in 50 mM NaHPO₄ pH 7.5 and 500 mM NaF, with the addition of 5 mM CaCO₃ as needed.

Bio-layer interferometry

Interaction of mCRT(WT) and GB1-mCRT(P) with ERp57 were assessed at 25 °C via bio-layer interferometry as described previously (21). In these analyses calreticulin was immobilized on streptavidin sensors and allowed to associate with 0.125, 0.25, 0.5 or 1 μM ERp57 for 250 seconds followed by a 500 second dissociation in 20 mM HEPES pH 7.5, 10 mM NaCl, 5 mM CaCl₂.

Animals

For phagocytosis studies, C57BL/6 (H-2b) mice were purchased from The Jackson Laboratory. Euthanized B6D2F1 mice (a cross between a female C57BL/6 mouse and male DBA/2 male) were obtained from the University of Michigan Transgenic core. Mice were maintained in a specific pathogen-free facility and cared for according to protocols that were reviewed and approved by the University Committee on Use and Care of Animals. Mice were euthanized by CO₂ asphyxiation.

Cell culture

Calreticulin-deficient (K42) mouse embryonic fibroblasts (MEFs) (16) were cultured in RPMI-1640 (Gibco) supplemented with 10% (v/v) heat-inactivated fetal bovine serum (FBS) (Gibco) and antibiotic-antimycotic (1x from a 100x stock) (Gibco). Cells were cultured in 100 mm plates at 37 ºC in 5% CO₂. K42 MEFs were reconstituted with mCRT(WT) via retroviral infection as described previously (17, 27, 28).

Resident macrophages were isolated from the peritonea of naïve 7-8-week-old mice as follows: Following euthanization via CO₂ asphyxiation, the abdomen was cut using blunt tipped scissors. The peritoneum was exposed and flushed 5-6 times with 1.5 mL ice-cold sterile PBS containing 0.5 mM EDTA (Invitrogen) to obtain 8-10 mL of cells in suspension. The cells were washed twice in leukocyte culture media (LCM; RPMI 1640 containing 25 mM HEPES pH 7.5 (Gibco), 1 mM sodium pyruvate (Gibco), antibiotic-antimycotic (1x from a 100x stock) (Gibco), 10% (v/v) heat inactivated FBS (Gibco) and 50 μM β-mercaptoethanol). Cells were plated in LCM in 12-well plates or on coverslips placed in 6-well plates. Following incubation for 60 min at 37 °C, non-adherent cells were removed by washing. Macrophages were allowed to rest overnight in LCM at 37 ºC in 5% CO₂ prior to use.

Binding assays by flow cytometry

Measurement of the binding of calreticulin to cells was undertaken as follows: Apoptotic K42 MEFs were made via UV-treatment as described below. Apoptotic MEFs or live cells (K42 MEFs or murine peritoneal macrophages) were incubated with 0 - 48 μM FITC-labeled calreticulin in the dark for 20 min at room temperature in binding buffer (20 mM HEPES pH 7.5 and 140 mM NaCl), with the addition of 5 mM CaCl₂ as indicated. Excess calreticulin was removed via two rounds of washing. Cells were suspended in binding buffer with 7AAD or DAPI to screen dead cells and analyzed via flow cytometry.

Inhibition experiments were undertaken similarly, with the exception that 24 μM FITC-labeled mCRT(WT) was incubated with cells in the presence of varying concentrations (0 - 150 μM) of an unlabeled inhibitor (GB1-mCRT(C), GB1-mCRT(P) or mCRT(1-351 ΔP)). All inhibition experiments were undertaken in 20 mM HEPES pH 7.5, 140 mM NaCl and 5 mM CaCl₂.

Assessment of the cell-surface expression of endogenous calreticulin in drug-treated cells was undertaken as described previously (29, 30). Briefly, CRT^−/− (K42) MEFs or K42 cells reconstituted with mCRT(WT) were treated with 300 μM oxaliplatin (OXP) (for 16 hours), 150 μM cisplatin (CDDP) (for 16 hours) or 1 μM nocodazole (NOCO) (for 20 or 48 hours) at 37 °C in RPMI-1640 supplemented with 10% (v/v) FBS. Adherent and non-adherent cells were harvested with 5 mM EDTA in PBS and washed in 20 mM HEPES pH 7.5, 140 mM NaCl and 5 mM CaCl₂. Cell-surface calreticulin was detected via flow cytometry as described earlier (30) using a chicken anti-calreticulin antibody (PA1-902A; Thermo Scientific) (2.5 μg/ml) or a PE-conjugated mouse anti-calreticulin antibody (FMC 75; Abcam) (1:100 dilution) with analyses gated on the 7AAD⁻ population to exclude dead cells.

Sequestration of externalized PS on apoptotic MEFs by calreticulin was assessed as follows: UV-treated CRT^−/− (K42) MEFs were incubated with 24 μM ovalbumin for 20 min at room temperature in 20 mM HEPES pH 7.5, 140 mM NaCl and 5 mM CaCl₂. Afterwards, the cells were washed and incubated with 24 μM FITC-labeled mCRT(WT), mCRT(ΔC) or mCRT(ΔP) for 20 min at room temperature in 20 mM HEPES pH 7.5, 140 mM NaCl and 5 mM CaCl₂. Following incubation with calreticulin, the cells were washed and incubated with 0.43 μM PE-Annexin-V for 20 min at room temperature in 20 mM HEPES pH 7.5, 140 mM NaCl and 5 mM CaCl₂ to mark externalized free (unsequestered) PS. Cells were washed and suspended in 20 mM HEPES pH 7.5, 140 mM NaCl and 5 mM CaCl₂ with 7AAD to screen dead cells and analyzed via flow cytometry.For all experiments, flow cytometry data were collected on 10,000 cells using a FACSCanto flow cytometer (BD Biosciences). Data were analyzed in FlowJo (Treestar), with samples gated on the 7AAD⁻ or DAPI⁻ population to exclude dead cells. Binding affinities and inhibition constants were calculated in GraphPad Prism ver. 6.05 (GraphPad Software) via non-linear least squares regression.

Fluorescence microscopy

Visualization of the co-localization of endogenous calreticulin and externalized PS on the surface of apoptotic MEFs was undertaken as follows: CRT^−/− (K42) MEFs reconstituted with mCRT(WT) were treated with 1 μM nocodazole for 20 hours at 37 °C in RPMI-1640 supplemented with 10% (v/v) FBS. Following drug-treatment, adherent and non-adherent cells were harvested with 5 mM EDTA in PBS and washed in 20 mM HEPES pH 7.5, 140 mM NaCl and 5 mM CaCl₂. The cells were labeled with a 100 μL PE-conjugated mouse anti-calreticulin antibody (FMC 75; Abcam) (1:100 dilution) and 0.43 μM FITC-Annexin V (BD Biosciences) for 20 min at 4 °C in 20 mM HEPES pH 7.5, 140 mM NaCl and 5 mM CaCl₂. Following labeling, cells were washed and fixed with 4% formalin in PBS for 15 min at room temperature following which, the cells were washed twice with PBS to remove any excess formalin. The fixed cells were resuspended in 10 μL PBS and mounted on slides using Prolong Gold anti-fade reagent (Invitrogen). Microscopy slides were cured overnight at room temperature and visualized using a Zeiss Apotome upright fluorescent microscope fitted with an exfo-illumination system with fluorescent filters for DAPI, GFP, TRITC, and Ds-red/Cy5. Images were captured using an attached high resolution Axiocam camera system.

Phagocytosis assays

In vitro phagocytosis assays were undertaken as described earlier (30). Briefly, target cells (CRT^−/− (K42) MEFs, or those reconstituted with mCRT(WT)) were labeled with 1 μM CMFDA at 37 ºC for 20 minutes in RPMI-1640 supplemented with 10% FBS. Following removal of excess CMFDA, MEFs were treated with 1 μM for nocodazole for 48 hours at 37 ºC in RPMI-1640 supplemented with 10% FBS. Alternatively, apoptosis was induced via exposure of MEFs to UV light for 5 minutes followed by a 16-18 hour incubation at 37 ºC in RPMI-1640 supplemented with 10% FBS. Non-adherent cells were harvested and coated with 0-40 μM calreticulin, its mutants or ovalbumin for 20 min at room temperature in 20 mM HEPES pH 7.5, 140 mM NaCl and 5 mM CaCl₂. Following binding, cells were washed to remove any unbound calreticulin. 0.2-1 x10⁶ target cells were fed to 0.2-1 x10⁶ macrophages plated in 12-well plates (for flow cytometry-based analyses) or attached to coverslips (for microscopy-based assays) for 1 hour at 37 ºC. Target cells were fed to macrophages in RPMI-1640 (containing 0.424 mM Ca²⁺) supplemented with 10% (v/v) FBS.

Following incubation of target cells with macrophages, the macrophages were washed with PBS and fixed with 1% formalin (Fisher) in PBS as described earlier (30). For flow cytometry-based analyses, macrophages were detached with 5 mM EDTA in PBS and washed once with flow cytometry buffer (2% (v/v) FBS in PBS) following which, macrophages were stained with anti-CD11b-PerCP-Cy5.5 (1:250 dilution) for 20 minutes at 4 ºC. Cells were washed twice and data was collected using a FACSCanto flow cytometer (BD Biosciences) For all flow-cytometry based in vitro phagocytosis assays, fluorescence data was collected on 10,000 cells and analyzed in FlowJo, with phagocytosis defined as the %CMFDA⁺ cells within the macrophage (CD11b⁺) gate. Phagocytic events were distinguished from adhesion by comparing the co-staining of the CD11b⁺ and CMFDA⁺ signals at 37 ºC relative to those at 4 ºC (a temperature at which phagocytic ingestion is inhibited).

For microscopy-based analyses, formalin-fixed macrophages were incubated with blocking buffer (1% BSA in PBS) for 30 minutes at 37 ºC and stained with anti-CD11b (1 mg/ml; diluted in blocking buffer) for 2 hours at 37 ºC. Cells were washed three times and incubated with a goat Texas red-conjugated secondary antibody for 1 hour at 37 ºC. Coverslips were washed three times with blocking buffer and mounted on slides using Prolong Gold anti-fade reagent (Invitrogen). 200 macrophages were counted per condition, with phagocytosis defined as the %CMFDA⁺ cells co-localized with the counted macrophages. Microscopy slides were cured overnight at room temperature and visualized using a Zeiss Apotome upright fluorescent microscope fitted with an exfo-illumination system with fluorescent filters for DAPI, GFP, TRITC, and Ds-red/Cy5. Images were captured using an attached high resolution Axiocam camera system.

In vivo phagocytosis assays were undertaken as follows: CRT^−/− (K42) MEFs or K42 MEFs retrovirally reconstituted with mCRT(WT) were labeled with 1 μM CMFDA as described above for the in vitro assays and treated with 1 μM nocodazole for 48 hours. Non-adherent cells were harvested and 2×10⁶ target cells were injected intra-peritoneally. The peritoneum was lavaged with 0.05% EDTA in PBS, 60 minutes after injection. The lavage cells were fixed with formalin, incubated with murine IgG to bind and block Fc receptors, and stained with anti-CD11b-PerCP-Cy5.5 for 20 min at 4 °C prior to data collection by FACSCanto flow cytometer. 100,000 cells were analyzed by flow cytometry, with phagocytosis defined as the %CMFDA⁺ cells within the macrophage (CD11b⁺) gate.

Bioinformatics and Statistical Analyses

The primary amino acid sequence for murine calreticulin was obtained from UniProt (31). ClustalW sequence alignments were undertaken using the EMBL ClustalW2 server (32, 33) or Lasergene MegAlign (ver. 12.1) (DNAstar).

All statistical analyses were undertaken in GraphPad Prism (ver. 6.05) (GraphPad Software) with grouped data represented as mean ± SEM. Statistically significant differences between means were assessed either via paired t-tests or one-way ANOVA followed by a Dunnett’s post-hoc test. For all analyses, statistical significance was assessed at an alpha value of 5%.

Results

Expression and purification of calreticulin truncation constructs and demonstration of calcium binding by the acidic C-terminal region of calreticulin

To better understand the nature of calcium-mediated interactions between calreticulin and phospholipids, we expressed and purified various truncated calreticulin constructs (Fig. 1A). We have previously described the expression and characterization of N-terminally histidine-tagged versions of wild-type murine calreticulin (mCRT(WT)) (17); mCRT(ΔC), which lacks 60 residues of the C-terminal acidic region (residues 340-399) (17); mCRT lacking residues 352-399 [mCRT(1-351)] (24), which truncates most of the acidic C-terminal region, but retains the first 12 residues that form an alpha helix contiguous with the C-terminal helix of the globular domain (24); and mCRT(ΔP), which lacks the P-domain (residues 187-283) (17) (a structural model for mCRT(1-351) is shown in Fig. 1A).

We also expressed and purified individual mCRT domains. A construct comprising only the globular domain of CRT (Fig. 1A) was characterized via x-ray crystallography (11) and corresponds to mCRT(1-351 ΔP) (Fig. 1A). A construct containing only the P-domain (residues 187-283) was expressed and purified as a fusion protein with the immunoglobulin-binding domain B1 of streptococcal protein G (GB1) (Fig. 1A; GB1-mCRT(P)). As expected, based on previous studies (13), GB1-mCRT(P) is able to bind the co-chaperone ERp57 with a similar affinity as mCRT(WT), whereas mCRT(ΔP) does not (Fig. 1B).

We also expressed the acidic C-terminal region of calreticulin (residues 340-399) as a fusion protein with GB1 (Fig. 1A; GB1-mCRT(C)). Analysis of the secondary structure content of this construct using circular dichroism (CD) spectroscopy revealed a pronounced signal at 220 nm that was distinct from the signal obtained with GB1 alone (Fig. 1C). These findings are suggestive of a well-folded protein with a defined secondary structure. Furthermore, consistent with previous findings with mCRT(WT) (24), the secondary structure of GB1-mCRT(C) is enhanced in the presence of 5 mM Ca²⁺, while no calcium-induced changes in secondary structure were seen with GB1 alone (Fig. 1C). Additionally, it is noteworthy that GB1-mCRT(C) maintains a defined secondary structure even in the absence of calcium (Fig. 1C).

The acidic C-terminal region of calreticulin is estimated to contain multiple calcium binding sites (15). Using isothermal titration calorimetry (ITC), we were previously able to measure ~4 enthalpy-driven calcium binding sites for mCRT(WT), ~2 sites for mCRT(1-351), but were unable to measure calcium binding by mCRT(ΔC) (24) (Table I). GB1-mCRT(C) is also able to bind calcium with a steady-state affinity similar to mCRT(WT) (Fig. 1D and Table I). Taken together, these findings suggest that the GB1-mCRT(C) construct is folded, able to bind calcium, and structurally responsive to calcium. However, the estimates of fewer enthalpy-driven calcium binding sites relative to mCRT(WT) (Table I) suggest that other regions of calreticulin may contribute to cooperative calcium binding by the C-terminal acidic region.

Table I.

Interactions of calreticulin with calcium and phospholipids, as assessed by isothermal titration calorimetry. For calcium binding by GB1-mCRT(C), data show mean ± SEM of 2 independent measurements, and no binding was detectable when GB1 alone was present (single measurement). All POPS and POPC binding measurements shown were undertaken in the presence of 5 mM CaCl₂. For POPS binding, data show mean ± SEM of 6 (mCRT(WT)), 4 (GB1-mCRT(C) and mCRT(ΔP)), 3 (mCRT(1-351ΔP), mCRT(ΔP)) or 2 (mCRT(1-351)) independent measurements. For POPC binding, data show mean ± SEM of 3 (mCRT(WT), mCRT(1-351ΔP)) or 2 (for mCRT(1-351) and mCRT(ΔP)) independent measurements. No POPS or POPC binding was detectable in 4 measurements (for mCRT(AC)), and no POPC binding was detectable in 4 of 4 (for GB1-mCRT(C)), 2 of 5 (for mCRT(WT)) or 2 of 4 (for mCRT(ΔP)) measurements. In the absence of CaCl₂, no POPS or POPC binding was detected in 4 (mCRT(WT), GB1-mCRT(C)) or 3 (mCRT(1-351ΔP)) independent measurements.

mCRT construct	Calcium Binding			POPS Binding			POPC Binding
	KD (μM)	ΔH (KJ/mol)	n (mol)	KD (μM)	ΔH (KJ/mol)	n (mol)	KD (μM)	ΔH (KJ/mol)	n (mol)
WT (1-399)	590.6 ± 88 a	9.4 ± 0.4a	3.8 ± 0.5 a	12.12 ± 4.49	-26.03 ± 7.72	1.16 ± 0.73	158.03 ± 37.34	-53.47 ± 18.40	1.22 ± 0.31
mCRT (ΔC)	No binding a			No binding			No binding
mCRT (ΔP)		Not determined		7.50 ± 5.69	−202.02 ± 152.83	1.50 ± 0.66	92.46 ± 10.96	−19.13 ± 6.17	1.85 ± 1.02
mCRT (1-351)	661.4 ± 28.5a	4.6 ± 0.7a	2.1 ± 0.003a	32.04 ± 5.54	−33.05 ± 4.69	0.62 ± 0.25	53.4 ± 13.80	−11.56 ± 1.96	1.99 ± 0.55
mCRT(1-351 ΔP)	Not determined			20.17 ± 4.08	−15.71 ± 5.72	2.94 ± 0.73	42.59 ± 11.48	−8.45 ± 3.64	4.73 ± 1.39
GB1-mCRT(C)	414.6 ± 142.1	13.4 ± 1.8	2.02 ± 0.22	28.23 ± 4.30	28.68 ± 9.83	0.90 ± 0.15	No binding
GB1	No binding			No binding			No binding

^aCalcium binding affinities and stoichiometries for some constructs were previously published in Wijeyesakere et al. (2011) (24).

The acidic C-terminal acidic region of calreticulin interacts with phospholipids in a calcium-dependent manner

To understand the nature of the interaction between calreticulin and lipids, we measured the thermodynamics of calreticulin-liposome interactions using ITC. Previous investigations have shown that, in the presence of calcium, calreticulin can interact with liposomes enriched in PS-containing phospholipids, with an affinity of ~15 μM (22). Indeed, we observed that mCRT(WT) displayed favorable binding to liposomes containing palmitoyl-oleoyl phosphatidylserine (POPS) (K_D = 12.12 ± 4.49 μM) (Fig. 2A and Table I). Furthermore, the affinity of mCRT(WT) towards POPS was an order of magnitude higher than that towards liposomes comprised of palmitoyl-oleoyl phosphatidylcholine (POPC) (K_D = 158.03 ± 37.34 μM) (Fig. 2B and Table I). Moreover, binding to both POPS and POPC is abrogated by the absence of calcium (Fig. 2A and B). Previous studies have reported an inability to detect the binding of a water-soluble PC derivative (06:0 PC) to calreticulin via SPR analysis (22). The ability to measure binding of mCRT(WT) towards POPC (Fig. 2B and Table I) could be due to differences in measurement techniques (ITC vs. SPR) or the specific type of lipid used (POPC liposomes vs. 06:0 PC). On the other hand, mCRT(ΔC), a construct that lacks the acidic C-terminal region, was unable to interact with either POPS or POPC, even in the presence of calcium (Fig. 2C, Table I). Furthermore, mCRT(1-351) and mCRT(1-351 ΔP) are able to bind both POPS and POPC in a calcium-dependent manner (Fig. 2D and E and Table I). However, based on the calculated affinities, the specificities of mCRT(1-351) and mCRT(1-351 ΔP) for PS relative to PC are reduced compared to mCRT(WT) (Table I). Additionally, for mCRT(1-351 ΔP), the number of POPC and POPS binding sites is increased, combined with reductions in binding enthalpies (Table I), suggesting the possibility of non-specific lipid associations that are induced by the combined P-domain and C-terminal truncations.

Figure 2. Calreticulin binds lipids in a calcium- and C-terminal acidic region-dependent manner.

Figure shows representative ITC thermograms (from of 3-5 independent experiments) showing the interaction of the indicated calreticulin constructs with liposomes containing POPC or POPS in the presence or absence of 5 mM CaCl₂. Plots show the raw titration curves (above) and corresponding curve fits (below). Unless indicated otherwise, mCRT-lipid interactions were measured in the presence of 5 mM CaCl₂. Calculated thermodynamic parameters are described in Table I.

Similar to our findings with mCRT(WT), GB1-mCRT(C) was able to bind POPS with an affinity of 28.23 ± 4.30 μM in a calcium-dependent manner (Fig. 2F and Table I). However, GB1-mCRT(C) is unable to bind POPC, even in the presence of calcium (Fig. 2F and Table I), suggesting that additional residues contained within the globular domain are required for stable POPC binding. Taken together, these findings indicate that the interaction between calreticulin and PS is mediated by the C-terminal acidic region of calreticulin via conjugated calcium ions. The lipid-binding site appears to be located within the N-terminus of the acidic region of calreticulin (residues 340-351). On the other hand, the C-terminal end of the acidic region (residues 352-399) is important for conferring specificity for PS, and for destabilizing binding to PC, a neutral lipid.

Calcium- and acidic region-dependent binding of calreticulin to apoptotic cells

Our studies thus far indicated that calreticulin can bind to synthetic PS in a calcium- and acidic region-dependent manner (Fig. 2). Since PS exposure is a hallmark of apoptosis (6), we examined the modes of acidic region-dependent binding of calreticulin to UV-treatment-induced apoptotic cells.

GB1-mCRT(C) bound apoptotic calreticulin-deficient MEFs in a calcium-dependent manner (Fig. 3A), but binding to live cells was not detectable, even in the presence of calcium (Fig. 3B). A different calcium-dependent binding pattern was seen for interactions between mCRT(WT) and apoptotic cells, where the presence of calcium increased the binding affinity by ~3-fold relative to the calcium-depleted condition (Fig. 3C). Similar results were observed with mCRT(ΔP), which retains the acidic C-terminal region (Fig. 3D). On the other hand, preferential calcium-dependent binding to apoptotic cells was not observable with mCRT(ΔC), which lacks the acidic C-terminal region (Fig. 3E). Together, these findings are consistent with the presence of distinct modes of interaction between mCRT(WT) and apoptotic cells: (i) a higher-affinity calcium and acidic region-dependent binding to apoptotic cells; and (ii) a lower affinity calcium and acidic region-independent binding mode. The observation of the higher affinity calcium- and acidic-region dependent mode of binding requires apoptotic cells, as mCRT(WT) binding to live cells was calcium-independent (Fig. 3F). Thus, calreticulin can engage in preferential calcium- and acidic region-dependent interactions with apoptotic cells, an observation consistent with the ability of calreticulin to bind PS-enriched liposomes in a calcium and acidic region-dependent manner (Fig. 2 and Table I).

Figure 3. Calcium-bridged interactions with the acidic C-terminal region of calreticulin mediates mCRT binding to apoptotic, but not live cells.

Flow cytometric analysis of the binding of FITC-labeled GB1-mCRT(C) (A and B), mCRT(WT) (C and F), mCRT(ΔP) (D) and mCRT(ΔC) (E) to apoptotic (A, C-E) or live (B and F) mouse embryonic fibroblasts (MEFs) in the presence or absence of calcium as indicated. Figures show representative plots from 3 (A), 2 (B), 5 (C), 2 (D and E) and 4 (F) independent experiments. In all analyses, binding was measured in the 7AAD⁻ or DAPI⁻ populations by flow cytometry.

Distinct modes of calreticulin binding to apoptotic MEFs compared to live MEFs and macrophages

To further characterize the modes by which calreticulin binds to live and apoptotic cells, inhibition assays were conducted in the presence of purified GB1-mCRT(C), mCRT(1-351 ΔP), as well as GB1-mCRT(P) (Fig. 1A). In the presence of 5 mM Ca²⁺, GB1-mCRT(C) was a more potent inhibitor of mCRT(WT)-apoptotic cell interactions (IC₅₀ = 23.49 ± 4.82 μM), compared with mCRT(1-351 ΔP) and GB1-mCRT(P) (Fig. 4A and Table II). While both mCRT(1-351 ΔP) and GB1-mCRT(C) were able to bind PS with similar affinities, mCRT(1-351 ΔP) also bound PC with a similar affinity as PS (Table I). Thus, it is possible that sequestration of mCRT(1-351 ΔP) by cell surface PC or other cell membrane components makes it a less potent inhibitor of mCRT(WT)-PS interactions on apoptotic cells. Taken together, the findings thus far support a model where interactions between mCRT(WT) and apoptotic cells involve engagement of cell-surface PS by the acidic C-terminal region of calreticulin, although additional factors may also be relevant.

Figure 4. Distinct modes of calreticulin binding to apoptotic cells compared to live cells or macrophages.

Inhibition assays (panels A, B and D) or direct binding assays (panel C) are shown. Panels A, B and D show representative IC₅₀ plots for the dose-dependent inhibition of the binding of FITC-labeled mCRT(WT) to apoptotic MEFs (A), live MEFs (B) or live macrophages (D) by the indicated mCRT domains. Calculated inhibition constants and data replicates are described in Table II. (C) Flow cytometric analysis of the dose-dependent binding of FITC-mCRT(WT) (shown as filled circles) or GB1-mCRT(C) (shown as filled squares) to macrophages in the presence or absence of calcium as indicated. Panel shows representative plots from 6 (mCRT(WT) in 5 mM Ca²⁺) or 2 (mCRT(WT) in 0 mM Ca²⁺ and GB1-mCRT(C)) independent experiments. In all analyses, binding was measured in the 7AAD⁻ or DAPI⁻ populations by flow cytometry.

Table II.

GB1-mCRT(P), GB1-mCRT(C) and mCRT(1-351 ΔP)-mediated inhibition of mCRT(WT) interactions with live and apoptotic MEFs and murine peritoneal macrophages. With live MEFs, the data represent the mean ± SEM of 7 (for GB1-mCRT(P) and GB1-mCRT(C)) or 4 (for GB1-mCRT(1-351 ΔP)) independent measurements. With apoptotic MEFs, the data represent the mean ± SEM of 7 (for GB1-mCRT(C), 6 (GB1-mCRT(P)) or 3 (for GB1-mCRT(1-351 ΔP)) independent measurements. With macrophages, the data represent mean ± SEM of 5 (for GB1-mCRT(P)), 3 (for GB1-mCRT(C)) or 2 (mCRT(1-351 ΔP)) independent measurements, and inhibition by GB1-mCRT(C) in 2 of 5 measurements was insufficient for accurate IC₅₀ estimates. All inhibition studies were undertaken in the presence of 5 mM CaCl₂.

Inhibitor	IC50 (μM) a
	Live MEFs	Apoptotic MEFs	Macrophages
GB1-mCRT(P)	18.09 ± 2.29	118.0 ± 33.9	15.76 ± 3.97
GB1-mCRT(C)	268.3 ± 109.4	23.49 ± 4.82	112.8 ± 14.13
mCRT(1-351 ΔP)	31.50 ± 11.62	110.0 ± 65.07	7.64 ± 0.38

^aData represent Mean ± SEM

In contrast to findings with apoptotic MEFs, interactions between mCRT(WT) and live MEFs were more effectively inhibited by GB1-mCRT(P) and mCRT(1-351 ΔP) than by GB1-mCRT(C) (Fig. 4B and Table II). While calreticulin is best characterized as an ER chaperone that binds monoglucosylated glycans on nascent proteins (reviewed in (34-36)), it also has a generic polypeptide-specific binding mode (17, 27, 37-39). We have previously shown that the polypeptide-binding site of calreticulin is relevant to its cell-surface interaction with factors present on live thapsigargin-treated cells (27). Moreover, our recent studies indicate that the generic polypeptide binding site of calreticulin is contained within the globular domain, and stabilized by a conformational change in the P-domain (21). Taken together, the results thus far indicate two distinct modes of calreticulin binding to apoptotic and live MEFs: (i) a higher affinity calcium- and acidic-region dependent interaction with apoptotic MEFs that likely involves interactions with cell-surface PS; and (ii) a lower affinity calcium-independent, globular and P-domain-dependent interaction with live MEFs that could involve interactions with cell surface proteins via the generic polypeptide-binding site of calreticulin.

To further understand the mode of interaction between calreticulin and phagocytes, we also undertook calreticulin-macrophage binding assays (Fig. 4C and D). In the presence of calcium, mCRT(WT) was able to bind murine peritoneal macrophages (K_D = 8.12 ± 1.54 μM) (Fig. 4C), albeit with a lower affinity than that observed for mCRT(WT)-apoptotic cell interactions (Fig. 3C). Furthermore, similar to findings with live MEFs (Fig. 3F), mCRT(WT)-macrophages interactions are independent of both, the acidic C-terminal region and calcium (Fig. 4C). Additionally, compared to GB1-mCRT(C), GB1-mCRT(P) was a stronger inhibitor of mCRT(WT) binding to macrophages (Fig. 4D and Table II). The globular domain (mCRT(1-351 ΔP)) also inhibited mCRT(WT) binding to macrophages with a similar IC₅₀ as that for the GB1-mCRT(P) (Table II). Taken together, these findings indicate that calreticulin binds macrophages via a calcium-independent binding mode resembling that seen for live MEFs (Table II).

Calreticulin and PS are simultaneously detectable on the surface of apoptotic cells, where calreticulin promotes apoptotic cell phagocytosis

It is known that externalized PS is a potent ‘eat me’ signal that promotes efferocytosis of dying cells (reviewed in (7)). As such, the binding of apoptotic cell-surface calreticulin to externalized PS leaflets could block PS-accessibility to other PS-binding factors, and induce PS-independent mechanisms of cellular phagocytosis. Alternatively, simultaneous exposure of calreticulin and PS could synergize with other PS-binding factors to potentiate apoptotic cell phagocytosis. The latter model is supported by previous studies, which have shown that calreticulin and externalized PS are co-localized within distinct regions on the surface of apoptotic cells (8). To extend those findings and confirm the simultaneous exposure of calreticulin and PS under physiologically-relevant concentrations of calreticulin exposure, we reconstituted calreticulin-deficient MEFs with mCRT(WT) via retroviral transduction. We have previously shown that this procedure reconstitutes mCRT to levels similar to those present in wild type mouse fibroblasts (17, 27). Previously, we were unable to observe the cell-surface expression of endogenous calreticulin in UV-treated apoptotic MEFs (30). Surface induction of calreticulin on apoptotic (PS-exposed) cells was thus assessed in response to three previously-studied drugs: oxaliplatin, cisplatin, and nocodazole (Fig. 5A). Oxaliplatin, but not the related platinum-based drug cisplatin, has been shown to induce immunogenic cell death in murine tumor cells by inducing surface calreticulin exposure (40). However, oxaliplatin did not induce significant cell surface calreticulin levels in MEFs (Fig. 5A), suggesting a cell-line specific induction of cell surface calreticulin by the drug.

Figure 5. Calreticulin surface exposure in nocoadazole-treated MEFs and calreticulin-mediated phagocytosis induction in vitro and in vivo.

(A) CRT^−/− (K42) MEFs or K42 cells reconstituted with mCRT(WT) were treated with 300 μM oxaliplatin (OXP) (for 16 hours), 150 μM cisplatin (CDDP) (for 16 hours) or 1 μM nocodazole (NOCO) (for 48 hours) and analyzed for surface calreticulin expression by flow cytometry. Panels depict representative histograms (from 3 independent experiments) showing cell-surface calreticulin expression with the analyses gated on the apoptotic (Annexin-V⁺/7AAD⁻) population. (B) Upper left panel shows representative flow cytometry plots depicting the externalization of calreticulin and PS (detected via Annexin V staining) in nocodazole-treated (top row) or untreated MEFs (bottom row), with cells gated on the 7AAD⁻ population. There is low level of non-specific binding of the anti-CRT antibody to a cell-surface factor that is upregulated following drug treatment. Upper middle panel shows the quantification of the percentage of CRT⁺ MEFs within the Annexin V⁺ population. Upper right panel shows the percentage of total Annexin V⁺ cells following nocodazole-treatment of CRT^−/− MEFs or those reconstituted with mCRT(WT). Averaged data ± SEM from 4 independent experiments are shown in the middle and right panels, with statistical significance assessed via paired t-tests and statistically significant differences indicated (*). Lower panel shows representative fluorescence microscopy images (original magnification ×40) (from 2 independent experiments) depicting the co-localization of mCRT(WT) and Annexin V (to mark PS) on the surface of nocodazole-treated MEFs. (C and D) Phagocytosis of nocodazole-treated MEFs by murine peritoneal macrophages as assessed in vitro via fluorescence microscopy (C) or in vivo via flow cytometry (D). In D, CMFDA-labeled, nocodazole-treated apoptotic CRT^−/− cells or those reconstituted with mCRT(WT) were injected i.p. and macrophages were harvested by peritoneal lavage 60 minutes post-injection. The percentages of macrophages (CD11b⁺ cells) positive for the target cell marker (CMFDA) are shown. ‘PBS’ represents a control where mice were injected with sterile PBS alone in the absence of apoptotic target cells. Left-hand panels in C and D show representative fluorescence images or flow cytometry data used to measure the uptake of nocodazole-treated CMFDA-labeled mCRT(WT)-expressing MEFs by murine peritoneal macrophages (CD11b⁺ cells) (C). Right-hand panels in C and D show the mean phagocytic uptake ± SEM from 4 independent experiments, with 2 mice per data point shown for the in vivo experiments (panel D). Statistical significance was assessed via one-tailed paired t-tests and statistically significant differences indicated (*).

Consistent with previous findings in murine tumor cells (41), the microtubule inhibitor nocodazole induced significant levels of calreticulin on the surface of apoptotic (Annexin V⁺/7AAD⁻) MEFs, with >90% of mCRT(WT)-expressing apoptotic cells being CRT⁺ (Fig. 5A and B (upper left and middle panels)). Moreover, under this condition, surface exposure of mCRT(WT) did not result in a significant decrease in the proportion of Annexin V⁺ cells (Fig. 5B (upper right panel)), a finding consistent with the previous report (8). Thus, in nocodazole-treated apoptotic cells, externalized calreticulin and unsequestered PS are both observable on the cell surface. Furthermore, consistent with previous findings (8), fluorescence microscopy confirmed extensive regions of co-localization between externalized calreticulin and Annexin V (Fig. 5B, lower panel).

We also assessed whether nocodazole-induced cell-surface expression of endogenous calreticulin enhances phagocytosis of dying MEFs by murine peritoneal macrophages in vitro and in vivo. CRT^−/− MEFs or those reconstituted with mCRT(WT) were labeled with CMFDA and treated with nocodazole to induce cell death and cell-surface calreticulin exposure. After 48 hours of drug treatment, dying/dead non-adherent cells were used as target cells for phagocytosis by murine peritoneal macrophages in vitro, with phagocytosis assessment via fluorescence microscopy to measure the proportion of macrophages (CD11b⁺ cells) that were also positive for the target cell marker (CMFDA).

Cell-surface exposure of endogenous mCRT(WT) following nocodazole treatment (Fig. 5A and B) translated to enhanced phagocytic uptake relative to apoptotic CRT^−/− cells (Fig. 5C). Additionally, significant calreticulin-independent phagocytosis was observable, with surface expression of mCRT(WT) potentiating uptake by ~40% relative to CRT^−/− cells (Fig. 5C). Similar results were obtained using an in vivo phagocytosis assay where nocodazole-treated dying cells expressing mCRT(WT) were phagocytosed more efficiently than those lacking calreticulin, following injection into the mouse peritoneum (Fig. 5D). Taken together, these findings indicate that both cell surface calreticulin and PS are detectable and co-localized on the surface of nocodazole-treated apoptotic cells. Additionally, externalized calreticulin potentiates the efferocytosis of nocadazole-treated apoptotic cells. Further, as previously shown (Fig. 5A and (9, 30)), not all apoptotic conditions resulted in the display of detectable levels of cell-surface calreticulin by flow cytometric assays.

Contributions of the acidic and P domains to calreticulin-induced phagocytosis of apoptotic cells

mCRT(ΔC) is secreted into cell supernatants at increased levels compared to mCRT(WT) ((42) and our unpublished findings). Furthermore, mCRT(ΔC) was poorly induced on the surface of nocodazole-treated cells (data not shown). Thus, we used apoptotic CRT^−/− MEFs opsonized with saturating concentrations of mCRT(WT), mCRT(ΔC), mCRT(ΔP) or the control protein ovalbumin, to better understand the functional role of calreticulin domains in binding apoptotic cells and promoting efferocytosis.

When CMFDA-labeled apoptotic MEFs were fed to murine peritoneal macrophages (CD11b⁺ cells) at 37 ºC, the uptake of mCRT(WT)-coated cells was higher than uncoated cells (Fig. 6A and B), a result consistent with findings using mCRT(WT)-expressing nocodazole-treated MEFs (Fig. 5C and D). Furthermore, mCRT(WT)-coated apoptotic MEFs were phagocytosed more efficiently than those coated with ovalbumin, mCRT(ΔC) or mCRT(ΔP) (Fig. 6A and B). Ovalbumin and the mCRT truncation constructs were purified from the same E. coli source as mCRT(WT), mCRT(ΔP) and mCRT(ΔC). Taken together, these findings suggest that the pro-phagocytic role of calreticulin is dependent upon both the arm-like P-domain as well as the C-terminal acidic region.

Figure 6. Roles for calreticulin domains in the phagocytic uptake of apoptotic cells and in binding to apoptotic cells.

(A-C) Phagocytosis by murine peritoneal macrophages of apoptotic (UV-treated) MEFs coated with indicated calreticulin constructs, assessed by flow cytometry (A, C) or microscopy (B). Phagocytosis measurements were done at 37 °C, with 4 °C incubations serving as adhesion controls. In all cases, phagocytosis was assessed as the %CMFDA⁺ cells within the macrophage (CD11b⁺) population. The left-hand panel in A shows a representative flow cytometric analysis. The left-hand panel in B shows representative fluorescence images (original magnification ×40) of the uptake of CMFDA-labeled mCRT(WT)-coated apoptotic MEFs by murine peritoneal macrophages. Bar graphs (right panels) in panels A and B show the mean phagocytic uptake ± SEM from 5 (A) or 7 (B) independent experiments. Statistical significance within each temperature group was assessed via repeated measures one-way ANOVA followed by a Dunnett’s post-hoc test (panel A) or paired t-tests (panel B) and statistically-significant differences are indicated (*). Panel C shows a representative EC₅₀ plot (from 3 independent experiments) depicting dose-dependent changes in the efferocytosis of apoptotic MEFs coated with 0-40 μM mCRT(WT) by murine peritoneal macrophages as assessed via flow cytometry. (D) Assessment of sequestration of externalized PS by calreticulin. Left panel shows representative flow-cytometry plots (measured within the 7AAD⁻ population) showing the binding of the indicated mCRT-FITC constructs (24 μM) followed by Annexin V-PE (to mark externalized PS) to apoptotic cells. Other panels depict quantifications of the percentage (± SEM) of total Annexin V-PE⁺ cells (second panel), total CRT⁺ cells (third panel) and CRT⁺ cells within the Annexin V⁺ population (fourth panel) from 3 independent experiments, with statistical significance assessed via paired t-tests and statistically significant differences indicated (*).

The observed mCRT(WT)-induced uptake of apoptotic MEFs was dose-dependent, with an EC₅₀ of 6.13 ± 0.68 μM (Fig. 6C), a value close to the steady-state affinity of mCRT(WT) towards apoptotic MEFs (Fig. 3C). Additionally, it is noteworthy that there are no significant differences in the co-localization of CD11b⁺ and CMFDA⁺ signals at 4 ºC (Fig. 6A and B), a condition that inhibits phagocytosis, but not binding. Taken together, these results suggest that cell-surface calreticulin can promote uptake of apoptotic cells, but is unable to mediate their adhesion to phagocytes.

To better understand whether the impairment in phagocytosis of mCRT(ΔC)-coated cells resulted from altered/impaired binding of mCRT(ΔC) to apoptotic MEFs, we used flow cytometry to quantify the binding of Annexin V to cells pre-coated with mCRT(WT) or the mCRT(ΔC), mCRT(ΔP) mutants. Both calreticulin and Annexin V were able to simultaneously bind to apoptotic cells (Fig. 6D first panel), indicating that not all surface-exposed PS is sequestered, even when exogenous calreticulin is bound at saturating concentrations (24 μM). Pre-coating apoptotic MEFs with each of the tested mCRT constructs prior to incubation with Annexin V reduced the total number of detectable Annexin V⁺ cells (Fig. 6D second panel), indicating that the reduction is independent of the acidic C-terminal region of calreticulin (Fig. 6D second panel), and suggesting calreticulin-mediated steric hindrance of interactions between Annexin V and exposed PS leaflets at the high concentrations of exogenous calreticulin used (24 μM). Nonetheless, it is noteworthy that the majority of CRT⁺ cells were also Annexin V⁺ when cells were coated with mCRT(WT) or mCRT(ΔP) (Fig. 6D last panel). Additionally, total CRT⁺ cells as well as CRT⁺AnnV⁺ cells were reduced in the case of cells coated with mCRT(ΔC) compared to those coated with mCRT(WT) or mCRT(ΔP) (Fig. 6D last two panels). Thus, mCRT(ΔC), which is unable to bind PS (Table I), interacts less efficiently with apoptotic MEFs, suggesting an altered mode of binding compared to mCRT(WT), a finding consistent with the previous binding studies (Figs. 3 and 7). Moreover, the P-domain is important for calreticulin-mediated phagocytosis (Figs. 6A, 6B and 7), even though apoptotic cell binding by mCRT(ΔP) is very similar to that seen with mCRT(WT) (Fig 3C-D and 6D).

Figure 7. Proposed model for the pro-phagocytic role of calreticulin.

(A and D) The acidic C-terminal regions of mCRT(WT) anchors it to exposed phosphatidylserine on the surface of the apoptotic target cell via calcium-bridged interactions. The arm-like P-domain of calreticulin interacts with receptors on the phagocyte (such as LRP-1), allowing calreticulin to function as a pro-phagocytic signal and promote uptake of the dying cell. Phagocytosis models in which the globular domain engages the phagocyte receptor (A) or apoptotic cell co-receptor (D) are both consistent with the data. (B and C) Proposed models for the PS-independent (B) and dependent modes (C) of interaction between apoptotic target cells and mCRT(ΔC) and mCRT(ΔP) respectively.

Discussion

In this study, we identify the acidic region of calreticulin as the mediator of calcium-dependent PS binding (Figs. 2 and 3 and Tables 1 and II). This region also plays a role in the calcium-dependent mode of binding between calreticulin and apoptotic cells (Fig. 7A). mCRT(ΔC), which is unable to bind PS (Fig. 2C and Table I), interacts with apoptotic cells via a calcium-independent binding mode (Figs. 3E and 7B). The inability of mCRT(ΔC) to promote phagocytosis of apoptotic cells (Fig. 5) can be explained by this alteration in the mode of apoptotic cell engagement (Figs. 3C, 3E, 6D and 7B). On the other hand, the P- and globular domains preferentially mediate interactions between calreticulin and factors present on live cells, including macrophages (Figs. 3, 4 and Table II). The latter findings are consistent with a model where macrophage/live cell binding is mediated by the generic polypeptide-binding site of calreticulin (Fig. 1A), which has recently been shown to involve the globular and P-domains of calreticulin (10, 20, 21).

The presence of a PS binding site within the acidic domain in mCRT(ΔP) (Table I) allows it to bind to apoptotic MEFs in a calcium-dependent manner similar to mCRT(WT) (Figs. 3D, 6D and 7C). The inability of mCRT(ΔP) to promote phagocytosis (Fig. 6) can be attributed to its inability to engage a phagocyte receptor (Fig. 7C). Apoptotic MEFs (which display externalized PS) can be phagocytosed in the absence of cell-surface calreticulin (Figs. 5 and 6), which may in part relate to the functional activity of macrophage cell surface calreticulin. It is noteworthy that this constitutive calreticulin-independent efferocytosis is unaffected by opsonization of apoptotic target cells with mCRT(ΔP) (Fig. 6A and B), which is able to bind both PS (Table 1) and apoptotic cells (Fig. 3D). These findings suggest that constitutive calreticulin-independent phagocytosis must involve PS-bridging factors that are endogenous to macrophages/MEFs or those present in the media. Recent studies have reported that murine peritoneal macrophages phagocytose apoptotic cells using Tim4 and MerTK as co-receptors, with serum-derived Protein-S serving as the PS-binding bridging molecule (43). Since the phagocytosis assays (Figs. 5 and 6) were undertaken in the presence of fetal bovine serum, it is possible that the observed basal phagocytosis is mediated via serum-derived Protein-S moieties that are pre-bound to the apoptotic cells. Compared to mCRT, the predicted higher affinity of Protein-S towards apoptotic cells (~220 nM) (44) may impede its displacement by exogenous calreticulin.

Overall, our findings support a bridging model for calreticulin-induced efferocytosis wherein te C-terminal acidic region of calreticulin anchors the protein to exposed PS rafts on the outer membrane of the dying cell (via conjugated Ca²⁺ ions), while the P-domain bridges an interaction with receptors on the phagocyte, thereby stimulating uptake (Fig. 7A). It is likely that the globular domain of calreticulin is also important for the bridging interaction, through binding the phagocyte receptor (Fig. 7A) or an apoptotic cell co-receptor (Fig. 7D). Calreticulin can engage one or more cell-surface proteins on macrophages and other live cells via its generic polypeptide-binding site (Figs. 3F and 4), but it is likely that specific spatially-constrained receptor interactions within an area of the cell enriched in PS-binding factors and their receptors (termed the phagocytic synapse) dictates calreticulin-mediated potentiation of phagocytosis. Given the presence of both, surface calreticulin and unsequestered PS on the surface of apoptotic cells (Figs. 5 and 6), we propose that calreticulin co-operates with other PS-dependent modes of phagocytic uptake (such as Protein-S (43)) to potentiate the clearance of dying cells. Furthermore, while the low density lipoprotein receptor-related protein 1 (LRP-1) is suggested to be a phagocyte receptor that engages externalized calreticulin ((8) and reviewed in (19)), further studies are needed to assess the involvement of this protein or other phagocyte receptors as well as apoptotic cell co-receptors.

The identified role for calreticulin in potentiating the internalization of apoptotic cells appears to be independent of the mechanism of capture that tethers dying cells to a phagocyte, as no significant differences in co-staining of mCRT(WT)-coated apoptotic cells and macrophages were observed at 4 ºC (Fig. 6A and B). Since membrane internalization is inhibited at 4°C, these results suggest that opsonization of an apoptotic cell with calreticulin is insufficient to promote tethering to phagocytes. Thus, other adhesion molecules may play a role in the capture and anchoring of an apoptotic cell to a phagocyte, with calreticulin serving to promote uptake.

In addition to apoptotic cells, recent studies have shown that externalized calreticulin promotes the uptake of certain live cells, including cancer cells (45). Cancer cells are shown in some cases to display elevated levels of PS (46) and endothelial cells of tumor-draining vessels are also known to display elevated levels of anionic phospholipids (47). Thus, the mechanisms described here may also be relevant to the uptake of cancer cells with elevated PS exposure, although additional studies are needed to understand the extent of PS exposure in different in vivo cancer models, and their relevance to cellular phagocytosis.

Previously, cell-surface calreticulin expression has been suggested to enhance cellular phagocytic uptake of live cancer cells, with increased levels of “don’t eat me signals” such as CD47 serving to offset this effect (45). A follow-up study from the same laboratory using blocking anti-calreticulin antibodies suggested that phagocytosis is mediated by macrophage-cell surface calreticulin rather than cancer cell-surface calreticulin (48). Our findings suggest that calreticulin-mediated engagement of both, the phagocyte and target cell are required for calreticulin-mediated efferocytosis (Fig. 6). Furthermore, our findings suggest that specific orientations and interactions of calreticulin on the target cell and macrophage are important. Consequently, cell-surface calreticulin on either the target cell or the phagocyte can be rendered functionally inert through non-specific protein interactions that occur via the generic polypeptide-binding site of calreticulin. Further studies are needed to understand the mechanisms by which calreticulin engages cancer cells during phagocytosis, as well as the receptors that are relevant to calreticulin-macrophage binding during phagocytosis.

There has been heightened recent interest in calreticulin expression in cancer cells following reports of mutant calreticulin constructs in myeloproliferative neoplasms lacking JAK2 mutations (49, 50). These calreticulin mutations target the C-terminus of the protein, resulting in novel basic C-termini that contain more positively charged residues. These mutations are likely to alter both the calcium and lipid binding properties of calreticulin, which remain areas for further study.

It is well known that calreticulin can induce the phagocytosis of apoptotic cells (8) as well as stressed cells and certain cancer cells (9, 30, 45) under conditions where its cell-surface expression is induced. The results of this investigation provide the first structural insights into the PS-binding site of calreticulin as a key anchor point for the expression of extracellular calreticulin under apoptotic conditions that induce expression of cell-surface calreticulin. A more complete understanding of phagocyte receptors that engage the P-domain of calreticulin will yield valuable insights into pathways that are linked to immunogenic forms of cell death.