Programmed cell removal by calreticulin in tissue homeostasis and cancer

Abstract

Macrophage-mediated programmed cell removal (PrCR) is a process essential for the clearance of unwanted (damaged, dysfunctional, aged, or harmful) cells. The detection and recognition of appropriate target cells by macrophages is a critical step for successful PrCR, but its molecular mechanisms have not been delineated. Here using the models of tissue turnover, cancer immunosurveillance, and hematopoietic stem cells, we show that unwanted cells such as aging neutrophils and living cancer cells are susceptible to "labeling" by secreted calreticulin (CRT) from macrophages, enabling their clearance through PrCR. Importantly, we identified asialoglycans on the target cells to which CRT binds to regulate PrCR, and the availability of such CRT-binding sites on cancer cells correlated with the prognosis of patients in various malignancies. Our study reveals a general mechanism of target cell recognition by macrophages, which is the key for the removal of unwanted cells by PrCR in physiological and pathophysiological processes.

Fig. 1

Viable aged neutrophils are cleared through activation of program cell removal. a Analysis of neutrophil and macrophage cell populations after thioglycollate injection to WT and MRP8-Bcl2 mice. Peritoneal cells were collected and analyzed at 0, 2, 4, 8, 12, 24, and 72 h after thioglycollate injection. Similar lifespan was observed for WT and Bcl2 macrophages and neutrophils. Macs, macrophage;s WT, wild-type; PMN, neutrophil/polymorphonuclear cell. Results are representatives of three independent experiments. n = 3 mice for each time point. Error bars represent standard deviation. Cells were identified by surface markers such as CD11b⁺F4/80^-GR1⁺⁺ (Neutrophils), CD11b⁺F4/80⁺ (macrophages). Percentage reactive cells, the % of the total viable cells found in the peritoneum after induction of peritonitis. b WT but not Bcl2 neutrophils undergo cell death. WT and Bcl2 neutrophils were collected and cultured in vitro for 72 h. Cell viability was examined by AnnexinV and DAPI staining. Cells that were AnnexinV-DAPI- were considered as viable cells. n = 3. **P < 0.01 (t-test) for viability between WT and Bcl2 neutrophils. Error bars represent standard deviation. c A schematic showing peritonitis in WT and MRP8-Bcl2 mice. WT neutrophils undergo both programmed cell death (PCD) and programmed cell removal (PrCR) after maturation while Bcl2 neutrophils are resistant to PCD but maintain PrCR programs. d RNAseq anaylsis of 4 groups of neutrophils, including (1) WT neutrophils from bone marrow (BM), (2) WT neutrophils recruited to peritoneum (PT), (3) MRP8-Bcl2 neutrophils from bone marrow, and (4) MRP8-Bcl2 neutrophils recruited to peritoneum. PT, peritoneal; BM, bone marrow. e Distribution among different cellular pathways (y-axis) of: (blue) 333 genes that are associated with increased susceptibility to PrCR and (red) genes in the human genome. RNAseq analysis was performed on 4 groups of neutrophils, as described in d. RNAseq analysis was used to identify genes upregulated in the maturation process (BM to peritoneum) in both WT and MRP8-Bcl2 mice, which are likely associated with susceptibility to PrCR

Fig. 2

Cell surface CRT determines programmed cell removal of neutrophils in peritonitis. a, b Cell surface levels of CRT on macrophages and neutrophils after thioglycollate injection to MRP8-Bcl2 mice. Peritoneal cells were collected and analyzed at 0, 4, 6, 12, and 24 h after thioglycollate injection, by flow cytometry analysis. Results are representatives of three independent experiments. n = 3 mice for each time points. *P < 0.05 (paired t-test) for CRT staining between macrophages and neutrophils at the same time points. Error bars represent standard deviation. b Cell surface levels of CD47 on macrophages and neutrophils after thioglycollate injection to MRP8-Bcl2 mice. Peritoneal cells were collected and analyzed at 0, 4, 6, 10, and 24 h after thioglycollate injection, by flow cytometry analysis. Results are representatives of three independent experiments. n = 3 mice for each time point. n.s. (paired t-test) so significant differences for CD47 staining between macrophages and neutrophils at the same time points. Error bars represent standard deviation. c Binding of recombinant CRT to macrophages and neutrophils after thioglycollate injection to MRP8-Bcl2 mice. Peritoneal cells were collected and analyzed at 0, 4, 6, 10, and 24 h after thioglycollate injection. CRT-binding sites on macrophages and neutrophils were measured by incubating the cells with saturation concentration of recombinant CRT and analyzing by flow cytometry. Results are representatives of three independent experiments. n = 3 mice for each time point. *P < 0.05 (paired t-test) for rCRT staining between macrophages and neutrophils at the same time points. Error bars represent standard deviation. d An in vitro phagocytosis assay showing blockade of CRT inhibits phagocytosis of neutrophils, with WT and Bcl2 neutrophils as target cells and peritoneal macrophages. Neutrophils were collected 4 h after thioglycollate treatment and cultured for 24 h. Phagocytosis was normalized to the maximal response in the experiments. n = 3. **P < 0.01 (t-test) for phagocytosis between ctrl and CRT blocking Ab treatment. Error bars represent standard deviation. In a, b, and c, MFI, mean fluorescence intensity

Fig. 3

Macrophages are the sources of CRT and secrete CRT to label the target cells. a Expression of CRT measured by qRT-PCR in neutrophils and macrophages 8 h after thioglycollate. Macrophages and neutrophils were collected from MRP8-Bcl2 mice as described in “Experimental Procedures”. CRT mRNA level in macrophages were dramatically higher in macrophages as compared to neutrophils. *P < 0.05 (t-test) for expression of CRT between macrophages and neutrophils. Error bars represent standard deviation. b, c Immunofluorescent staining of CRT in mouse macrophages (b) and neutrophils (c). CRT is undetectable in neutrophils but abundant in macrophages. CRT localized to perinuclear regions, vesicles and cell surface of macrophages. Macrophages and neutrophils were collected from MRP8-Bcl2 mice. d ELISA assay showing the amount of CRT in medium (RPMI) with and without macrophages. n = 3. Error bars represent standard deviation. Macrophages were able to secrete CRT to the extracellular medium. e Expression levels of cell surface CRT on neutrophils cultured alone or with macrophages in a 0.4-μm Boyden chamber overnight, assayed by flow cytometry. Co-culture with macrophages led to a significant increase of cell surface CRT on neutrophils. n = 3. **P < 0.01 (t-test) for expression levels of cell surface CRT on neutrophils cultured alone or with macrophages. Error bars represent standard deviation. f A schematic showing the Click-iT assay to examine transfer of CRT from macrophages to neutrophils during co-culture. Macrophages proteins were labeled with a methionine analog AHA (l-azidohomoalanine) receptive for click chemistry. The proteins secreted by macrophages were all labeled with AHA. Because neutrophils and macrophages were cultured in a way independent of contact, the AHA-labeled proteins detected on neutrophils originate from the macrophages. g Neutrophils were cultured alone or together with AHA-treated macrophages in a 0.4-μm Boyden chamber. Cells were collected and CRT was immunoprecipitated and then subjected to click-chemistry adding in biotin to receptive methionine analogs. Samples were subjected to SDS-PAGE and then western blot for the presence of biotin. Detection of AHA-labeled CRT on neutrophils indicated that macrophage-secreted CRT to label neutrophils and this was independent of the contact between these two cell types

Fig. 4

Identification of CRT-binding ligands on aged and malignant cells. a Screening for CRT-binding glycans with carbohydrate microarray. Purified CRT-IgG-Fc proteins were used to probe a carbohydrate microarray containing a number of different types of glycans, including N-, O- and sulfated-glycans. Anti-IgG-Fc antibody was used for detecting binding of CRT to glycans. Glyco-antigens were used at 0.05 and 0.25 μg μl⁻¹ (left and right bars for each glycan). n = 3. Error bars represent standard deviation. b PHA-L binding to peritoneal neutrophils and macrophages at 0, 4, 6, 8, 24, and 72 h after thioglycollate injection in MRP8-Bcl2 mice. c, d Examination of cell surface CRT-binding sites on neutrophils. Bone marrow (c) or peritoneal neutrophils (d) were collected and treated with heat inactivated neuraminidase (Δneu) or neuraminidase (neu). Cells were incubated with PBS (control; black) or recombinant CRT proteins (red) and binding of CRT was then measured by flow cytometry with PE-conjugated anti-CRT antibody. rCRT binds to mature peritoneal neutrophils but not the immature bone marrow neutrophils. Treatment with neuraminidase led to the release of CRT-binding sites. e, f Immunofluorescent staining of CRT in HL60 (e) and SW620 (f) cells. CRT localized to perinuclear regions, vesicles, and cell surface. CRT was either expressed at a low level (e) or limited to the perinuclear regions (f), while a significant portion of PHA-L staining were observed on the cell surface (e and f).In a–d, MFI, mean fluorescence intensity

Fig. 5

Cell surface asialoglycans regulates CRT-mediated PrCR. a, b Treatment with neuraminidase led to the removal of sialic acids from the cell surface of HL60 cells. HL60 cells were treated with heat inactivated neuraminidase (Δneu) or neuraminidase (neu). Cell surface sialic acids were examined by staining with EBL (a) and MAL (b) by flow cytometry analysis. EBL, Elderberry Bark Lectin; MAL, Maackia Amurensis Lectin II. c, d Examination of cell surface CRT and PHA-L binding sites on cancer cells. HL60 cells were treated with heat inactivated neuraminidase (Δneu) or neuraminidase (neu). Recombinant CRT (c) and PHA-L (d) binding after treatment were measured by flow cytometry. e, f Phagocytosis of cancer cells with neuraminidase treatment. In vitro Phagocytosis assays were performed with HL60, K562, DLD-1, and SW620 cells treated with heat inactivated neuraminidase (Δneu) or neuraminidase (neu) as target cells. Mouse bone marrow-derived (e) and human peripheral blood monocyte-derived (f) macrophages were used for the assay. Phagocytosis was normalized to the maximal response in the experiments. n = 3. *P < 0.05, **P < 0.01 (t-test) for phagocytosis between heat inactivated neuraminidase (Δneu)- and neuraminidase (neu)-treated groups. Error bars represent standard deviation. g, h Effects of suppressing the expression of endogenous neuraminidases in cancer cells. Neu1–Neu4 gene knockout were performed with CRISPR in HL60 cells. In vitro Phagocytosis assays were performed with HL60 cells as target cells and mouse bone marrow-derived macrophages (h). Macrophages were treated with PBS (ctrl) or lipopolysaccharide (LPS). Neu4 knockout led to the decrease of cell surface CRT-binding sites and inhibited cancer cell phagocytosis. Phagocytosis was normalized to the maximal response in the experiments. n = 3. *P < 0.05, **P < 0.01 (t-test) for phagocytosis between ctrl and Neu4 KO groups. Error bars represent standard deviation

Fig. 6

Function of CRT-binding site in multiple malignancies and hematopoiesis. a, b In vivo tumorigenicity of cancer cells treated with neuraminidase. HL60 and DLD-1 cells were treated with heat inactivated neuraminidase (Δneu) or neuraminidase (neu), and injected subcutaneously into NSG mice. Growth of transplanted tumors was monitored by bioluminescence imaging. Tumor growth was normalized to bioluminescence signals of the injection day as fold changes. n = 5. *P < 0.05, **P < 0.01 (t-test) for tumor growth between heat inactivated neuraminidase (Δneu)- and neuraminidase (neu)-treated groups. Error bars represent standard deviation. c–f Genes related to the regulation of CRT-binding sites as a diagnostic marker for overall survival of cancer patients. Higher expression of NEU2 and NEU4 which induce the removal of sialic acids correlated with an improved survival while higher expression of ST3GAL1 and ST6GAL1 which enhance sialic acid expression correlated with a worse outcome. g Correlation between PHA-L binding and CRT levels on the cell surface of hematopoietic stem cells (HSC), Multipotential progenitor (MPP), leukemia stem cells (LSC), and blasts cells from primary AML patient samples; analysis by flow cytometry. In g, MFI, mean fluorescence intensity