Activated microglia cause reversible apoptosis of pheochromocytoma cells, inducing their cell death by phagocytosis

ABSTRACT

Some apoptotic processes, such as phosphatidylserine exposure, are potentially reversible and do not necessarily lead to cell death. However, phosphatidylserine exposure can induce phagocytosis of a cell, resulting in cell death by phagocytosis: phagoptosis. Phagoptosis of neurons by microglia might contribute to neuropathology, whereas phagoptosis of tumour cells by macrophages might limit cancer. Here, we examined the mechanisms by which BV-2 microglia killed co-cultured pheochromocytoma (PC12) cells that were either undifferentiated or differentiated into neuronal cells. We found that microglia activated by lipopolysaccharide rapidly phagocytosed PC12 cells. Activated microglia caused reversible phosphatidylserine exposure on and reversible caspase activation in PC12 cells, and caspase inhibition prevented phosphatidylserine exposur and decreased subsequent phagocytosis. Nitric oxide was necessary and sufficient to induce the reversible phosphatidylserine exposure and phagocytosis. The PC12 cells were not dead at the time they were phagocytised, and inhibition of their phagocytosis left viable cells. Cell loss was inhibited by blocking phagocytosis mediated by phosphatidylserine, MFG-E8, vitronectin receptors or P2Y₆ receptors. Thus, activated microglia can induce reversible apoptosis of target cells, which is insufficient to cause apoptotic cell death, but sufficient to induce their phagocytosis and therefore cell death by phagoptosis.

Summary: NO from microglia causes reversible caspase activation and phosphatidylserine exposure by PC12 cells, resulting in their cell death by phagocytosis.

INTRODUCTION

Apoptosis was originally characterised as a form of cell death accompanied by nuclear condensation, cell shrinkage and subsequent phagocytosis and digestion of the cell by phagocytes (Kerr et al., 1972). Later, it was found that the molecular machinery driving apoptosis was activation of BH3-domain-containing proteins, caspase activation and phosphatidylserine exposure on the surface of the cell – the latter inducing phagocytosis of the apoptotic cell (Galluzzi et al., 2012). This way of characterising apoptosis, that is, as coupled to phagocytosis of the cell, leaves it unclear as to whether phagocytosis contributes to the death of the cell. In C. elegans, it was found that inactivating mutations of phagocytic genes did not prevent the apoptotic death of most cells programmed to die during development (Hedgecock et al., 1983). However, it was subsequently found that phagocytosis did contribute to the death of some of these cells, and this contribution was greater if the caspase activation was weak or the cells were subjected to a variety of sub-lethal stresses (Hoeppner et al., 2001; Neukomm et al., 2011; Reddien et al., 2001). In vitro, in mammalian cells, strong induction of apoptosis causes cell death in the absence of phagocytes. However, more recently, it has been found that some of the molecular processes of apoptosis can be reversible and/or occur in viable cells, for example, caspase activation (D'Amelio et al., 2012), chromatin condensation (Neher et al., 2013) and phosphatidylserine exposure (Hammill et al., 1999; Mackenzie et al., 2005; Stowell et al., 2009). Furthermore, mild or transient induction of apoptosis can result in the reversible induction of apoptotic markers, such as cell shrinkage, mitochondrial fragmentation, nuclear condensation, cytoplasmic shrinkage, caspase activation and DNA damage, without resulting in cell death (Tang et al., 2012, 2009). This raises the question as to whether apoptosis itself can be reversible (Tang et al., 2012, 2009), that is, can the molecular processes of apoptosis be reversibly induced without resulting in apoptotic cell death, and if so whether in the presence of phagocytes they can drive cell death by phagocytosis.

A variety of cells, in a variety of conditions, can die as a result of being phagocytosed by other cells, a form of cell death that we have called ‘phagoptosis’ (Brown and Neher, 2012). This might be a common cause of cell death physiologically, as there is strong evidence that phagoptosis is responsible for the turnover of erythrocytes (the most abundant form of cell death physiologically), and increasing evidence showing that it contributes to turnover of neutrophils and lymphocytes (Brown and Neher, 2012). Phagoptosis also contributes to innate and adaptive immunity to pathogens through phagocytosis of opsonised pathogens by phagocytes. In addition, cancer development could be limited by phagoptosis given that most human cancers expose the ‘eat-me’ signal calreticulin (Chao et al., 2010) and overexpress the ‘don't-eat-me’ signal CD47 (Chao et al., 2012), and that blocking CD47 allows phagoptosis and subsequent clearance of multiple cancers in animal models (Chao et al., 2010, 2012; Edris et al., 2012; Goto et al., 2014; Gül et al., 2014; Willingham et al., 2012). Despite the potential ubiquity and importance of phagoptosis, its mechanisms and functions are poorly understood.

Microglia can phagoptose neurons (Neher et al., 2012) following activation by the bacterial Toll-like receptor agonists lipoteichoic acid and lipopolysaccharide (LPS) (Neher et al., 2011), the Alzheimer's-disease-related peptide amyloid β (Neniskyte et al., 2011), the pesticide rotenone (Emmrich et al., 2013) or the P2Y₆ receptor agonist UDP (Neher et al., 2014). Microglial phagoptosis of otherwise viable neurons can also contribute to neuronal loss in animal models of stroke (Neher et al., 2013), Parkinson's disease (Barcia et al., 2012), amyotrophic lateral sclerosis (Liu et al., 2012) and during neurogenesis (Cunningham et al., 2013), and thus might impact on brain development, inflammation and neurodegeneration (Brown and Neher, 2010, 2014).

Microglial phagocytosis of neurons can be mediated by neuronal exposure of the ‘eat-me’ signal phosphatidylserine (PtdSer), bound by the opsonin milk fat globule-EGF factor 8 (MFG-E8), which then induces phagocytosis by also binding the microglial vitronectin receptor (VNR; an α_vβ₃ or α_vβ₅ integrin) (Fricker et al., 2012a; Neher et al., 2011; Neniskyte and Brown, 2013). Calreticulin exposure on neurons can act as another ‘eat-me’ signal, which induces phagocytosis through the low-density lipoprotein receptor-related proteins (LRPs) on microglia (Fricker et al., 2012b). UDP release from stressed neurons can act as a final engulfment signal by activating P2Y₆ receptors on microglia (Koizumi et al., 2007; Neher et al., 2014).

BV-2 cells are a murine microglia cell line (Blasi et al., 1990; Bocchini et al., 1992). PC12 cells are a rat pheochromocytoma cell line that can be differentiated into cells of a neuronal phenotype (Greene and Tischler, 1976). Apoptotic PC12 cells can be phagocytosed by BV-2 cells; this process can require LRP-mediated recognition of exposed calreticulin (Fricker et al., 2012b) and/or extracellular annexin A1 binding both PtdSer on PC12 cells and formyl peptide receptor 2 on the BV-2 cells (McArthur et al., 2010). Moreover, BV-2 cells have been suggested to phagocytose ‘live’ PC12 cells when activated with LPS (McArthur et al., 2010) or PMA (Hornik et al., 2014), although it was not tested whether the PC12 were in fact alive when phagocytosed, or whether blocking phagocytosis prevented cell death. Because LPS-activated BV-2 cells produce nitric oxide (NO) from inducible NO synthase (iNOS, also known as NOS2) (Choi and Park, 2012), NO can induce apoptosis of PC12 cells (Bal-Price and Brown, 2000) and supernatants from activated BV-2 cells can induce apoptosis of PC12 cells (Ye et al., 2013), it might be that BV-2 microglia simply induce apoptosis of co-cultured PC12 cells. We therefore decided to use this system to investigate whether activated microglia could kill the PC12 cells by (1) apoptosis, (2) phagocytosis of dead cells or (3) phagocytosis of reversibly apoptotic cells.

RESULTS

LPS-activated BV-2 microglia rapidly phagocytose PC12 cells

We found that BV-2 cells (a microglial cell line) could rapidly phagocytose apparently healthy undifferentiated PC12 cells (a pheochromocytoma cell line): after 3 h co-incubation, 24±2% (mean±s.e.m.) of microglia contained PC12 cell debris (example Fig. 1A). This was increased 2-fold by pre-activation of microglia with lipopolysaccharide (LPS; 100 ng/ml) for 24 h prior to adding the PC12 cells, the phagocytosis being measured either by using microscopy [the number of LPS-treated BV-2 cells containing PC12 cell debris was 204±12% relative to untreated BV-2 cells (set at 100%); P<0.001; Fig. S1A] or flow cytometry [the average PC12 stain per BV-2 was 200±11% for LPS-treated BV-2 cells relative to untreated BV-2 cells (set at 100%); P<0.001; Fig. S1B; an example flow cytometry result is shown in Fig. 1B]. Thus, inflammatory activation of microglia increases their ability to phagocytose apparently healthy target cells.

Fig. 1.

Microglia phagocytose live and dead PC12 and cause loss of PC12 from co-cultures. (A) Phase-contrast and fluorescent images of BV-2 cells, labelled green with IB₄, having phagocytosed PC12 cells, stained red with TAMRA. Nuclei are stained blue with Hoechst 33342. Phagocytosis is visible as punctate TAMRA staining in IB₄-stained cells. The white arrowhead indicates a PC12 cell. Representative images are shown. Scale bars: 20 µm. (B) Typical flow cytometry results showing gating used to select IB₄-labelled (green; FL1) BV-2 cells (green arrowheads) to measure their average FL3 fluorescence (Av FL3) and thereby phagocytosis of TAMRA-stained (red; FL3) PC12 cells (red arrowheads). I, untreated BV-2 cells; II, untreated BV-2 cells that have phagocytosed TAMRA-stained PC12 (red; FL3) during 3 h of co-incubation; III; as II, but with 24 h pre-activation with LPS (100 ng/ml), which increases BV-2 cell phagocytosis of PC12 cells. (C) LPS pre-activation increases BV-2 cell phagocytosis of live and cycloheximide-treated (100 µg/ml) (dead) PC12 cells over 3 h. BV-2 cells phagocytose dead PC12 cells more readily than live PC12 cells. ***P<0.001 versus untreated live PC12 cells; ^###P<0.001 versus untreated dead PC12 cells. Note that 24±2% of non-activated microglia phagocytosed untreated PC12 cells. n=3. (D) PC12 cells were added to culture wells (with or without platted BV-2 cells and LPS) and the percentage of PC12 recovered from the well (with or without chromatin condensation) was counted after 0 or 3 h. LPS-pre-activated BV-2 cells removed healthy PC12 cells. Note that not all cells added to an empty well were recovered after 3 h due to some binding to the well. Healthy; ***P<0.001 versus seeded 0 h; ###P<0.001 versus BV-2 untreated. Chromatin condensed; ^&&&P<0.001 versus seeded 0 h; ^$$$P<0.001 versus no BV-2; ^£££P<0.001 versus BV-2 untreated. n=4. (E) Cytochalasin D (Cytoch D; 0.5 µM) decreases the proportion of BV-2 cells phagocytosing PC12 cells. **P<0.01, ***P<0.001 versus untreated; ^###P<0.001 versus LPS. n=5. (F) Cytochalasin D prevents loss of PC12 cells from LPS-pre-activated BV-2 co-cultures and increases the proportion of PC12 cells recovered after co-culture with untreated BV-2 cells. ***P<0.001 versus untreated; ^###P<0.001 versus LPS. n=5. Data shown are means±s.e.m. Statistical significance was quantified by one-way ANOVA and Bonferroni test.

Pre-addition of the LPS-binding peptide polymyxin B inhibited the LPS-induced activation of phagocytosis (Fig. S1C), indicating that the stimulation was due to LPS itself. Uptake into the microglia was roughly proportional to the number of target PC12 cells added (Fig. S1D), indicating that the microglia were not saturated with PC12 cells at the 1:1 ratio used. The absence or presence of serum during the uptake assay had relatively little effect on the amount of phagocytosis (1.3-fold greater in 0.5% than 0% serum; Fig. S1E).

In order to test whether dead PC12 cells were more readily taken up than live PC12 cells, we pre-treated PC12 cells with 100 µg/ml cycloheximide for 24 h, resulting in 95±1% of PC12 cells being necrotic, and then added these dead PC12 to the BV-2 cells as above. Unactivated BV-2 cells took up 3.6-fold more dead than live PC12 cells, and LPS-pre-activated microglia took up 2.9-fold more dead than live PC12 cells (Fig. 1C). Thus LPS pre-activation of microglia does not enhance or impair their ability to distinguish between live and dead PC12 cells for phagocytosis, but increases their overall phagocytic capacity. Note that the ‘live’ PC12 culture contained very few dead cells: 0.7±0.1% of cells were necrotic and 1.2±0.2% showed condensed chromatin at the time of addition to microglia, and 3.5±0.4% of cells were necrotic and 1.8±0.3% showed condensed chromatin at the end of the uptake assay. Thus, the amount of necrotic and apoptotic cells in the ‘dead’ cultures was ∼25-fold higher than in the ‘live’ cultures, and therefore if BV-2 cells only phagocytosed dead or dying PC12 cells then phagocytosis of ‘dead’ PC12 cells should be ∼25-fold greater than of ‘live’ PC12, rather than 3-fold higher. However, whether the PC12 were phagocytosed alive is more directly tested below.

If sufficient PC12 cells are taken up by microglia, then we should observe a decrease in PC12 cell numbers recovered from co-cultures with microglia, and indeed we observed this when the microglia were pre-activated with LPS to enhance uptake (Fig. 1D). Increasing the number of PC12 targets added resulted in a corresponding increase in PC12 cells recovered (Fig. S1F).

We next tested whether the uptake of PC12 cells into microglia was due to phagocytosis. Incubation at 4°C, a temperature that supresses phagocytosis (Peterson et al., 1977), completely prevented uptake of both live and dead PC12 cells (Fig. S2A). Uptake was also decreased by pre-treatment with 0.5 µM cytochalasin D, an inhibitor of actin polymerisation that prevents phagocytosis (Elliott and Winn, 1986) (Fig. 1E). This was not due to induction of apoptosis in the BV-2 cells as viability was not affected [viable BV-2 cells were 107±14% when treated with cytochalasin D relative to untreated cells (set as 100%)]. This inhibition was dependent on the concentration of cytochalasin D for the uptake of both healthy PC12 cells (Fig. S2B) and carboxylate-modified beads (which mimic the negatively-charged surface of cells exposing phosphatidylserine; Fig. S2C). Pretreatment with cytochalasin D also prevented the loss of PC12 cells resulting from co-incubation with microglia (Fig. 1F). We therefore conclude that microglia, particularly activated microglia, take up healthy PC12 cells by phagocytosis.

As PC12 cells can be differentiated into neuronal cells by addition of nerve growth factor and reduction of the serum concentration (Greene, 1978), we investigated whether this affected their phagocytosis by microglia. To avoid trypsin treatment and centrifugation of the differentiated PC12 cells, we added microglia to PC12 cells that had been differentiated for 3 days, and quantified loss and uptake of PC12 cells over a further 3 days (Fig. 2A). Addition of microglia in the absence of LPS activation resulted in no significant loss of PC12 cells (Fig. 2B). However, addition of LPS to the co-cultures resulted in significant loss of PC12 cells, which was increased by adding more microglia up to 12,500 per well, causing loss of 26±5% of the PC12 cells (Fig. 2B). Similar cell loss was observed when PC12 cells were differentiated for 7 days prior to microglia addition (Fig. S2D). Loss could be prevented by addition of cytochalasin D, leaving healthy PC12 cells (Fig. 2C). This suggests that phagocytosis must be the cause of death, because if the PC12 cells had died and subsequently been phagocytosed, then inhibition of phagocytosis would have left dead cells, rather than the live cells we observed. PC12 phagocytosis was seen in 16±4% of non-activated microglia (example in Fig. 2A), and this was increased 2-fold by LPS activation and blocked by cytochalasin D (Fig. S2E). LPS-activated microglia thus phagocytose healthy differentiated PC12 cells as well as naïve PC12 cells. To make a more direct comparison of this in the same conditions, we added either undifferentiated or differentiated PC12 cells to BV-2 cells. We found that the addition of PC12 cells differentiated for 7 days resulted in 4.5-fold more phagocytosis by microglia than addition of naïve PC12 cells (Fig. 2D). Thus, differentiated PC12 cells appear to be more readily phagocytosed than naïve PC12 cells. Note, however, that this differentiation and subsequent trypsin treatment also increased the proportion of necrotic PC12 cells to 14±2% in differentiated and trypsinised PC12 cells compared with 1.7±0.3% in naïve and untrypsinised PC12 cells.

Fig. 2.

LPS-activated microglia phagocytose differentiated PC12 over 72 h. (A) Phase-contrast and fluorescent images of 7-day differentiated PC12 cells (dPC12), stained with 50 µM 5-(and-6)-carboxytetramethylrhodamine succinimidyl ester (TAMRA; red; PC12), co-cultured with BV-2 cells (labelled with IB₄; green; BV-2) for 96 h and stained with Hoechst 33342 (Hoechst; blue; nuclei). After 24 h of LPS treatment (100 ng/ml), PC12 cell loss and uptake into BV-2 over 72 h is seen, visible as punctate TAMRA staining in IB₄-stained cells (white arrowheads). Representative images are shown. Scale bars: 50 µm. (B) 24 h of LPS treatment to activate BV-2 cells cause loss of 3-day differentiated PC12 cells (both healthy and with condensed chromatin) in a concentration-dependent manner during 72 h of co-incubation. Healthy; *P<0.05, ***P<0.001 versus no BV-2. Chromatin condensed; ^&&P<0.01, ^&&&P<0.001 versus no BV-2. n=3. (C) Cytochalasin D (0.2 µM) prevents LPS-activated BV-2-induced loss of healthy and chromatin condensed 3-day differentiated PC12 cells over 72 h. Healthy; ***P<0.001 versus untreated; ^##P<0.01 versus LPS. Chromatin condensed; ^&&P<0.01 versus untreated; ^$$$P<0.001 versus LPS. n=3. (D) Untreated and LPS-activated BV-2 cells more readily phagocytose differentiated PC12 cells than naïve PC12 during 3 h of co-incubation. ***P<0.001 versus naïve PC12 untreated; ^###P<0.001 versus naïve PC12 with LPS; ^&&&P<0.001 versus differentiated PC12 untreated. n=4. Data shown are means±s.e.m. Statistical significance was quantified by one-way ANOVA and Bonferroni test.

Phagocytosed PC12 cells are alive, but are induced to reversibly expose phosphatidylserine and activate caspases

To investigate whether the PC12 cells were alive or dead at the time of phagocytosis by microglia, we added propidium iodide to the co-cultures during the phagocytosis assay, using PC12 cells stained fluorescent green with 5(6)-carboxyfluorescein diacetate-N-succinimidyl ester (CFSE; Fig. 3A). Propidium iodide only enters necrotic cells, due to plasma membrane rupture, and then stains the nuclei of such cells fluorescent red, thus, if PC12 were phagocytosed after death then the phagocytosed PC12 should also be stained red with propidium iodide. When dead PC12 (killed with cycloheximide) were incubated with microglia, 99.9% of microglia containing PC12 cell debris also stained with propidium iodide (associated with the PC12 debris, not the microglial nucleus), indicating that the PC12 cells were taken up when necrotic (Fig. 3B). However, when live PC12 cells were incubated with microglia, of all the PC12 phagocytosis events observed, only 1.6% co-stained for propidium iodide (Fig. 3B), indicating that virtually all the PC12 cells were ‘eaten’ with an intact plasma membrane.

Fig. 3.

PC12 cells phagocytosed are alive but are induced by microglia to reversibly expose PtdSer. (A) Phase-contrast and fluorescent images of BV-2 cells having phagocytosed PC12 cells stained with 5 µM 5(6)-carboxyfluorescein diacetate-N-succinimidyl ester (CFSE; green; PC12) while incubated with the vital dye propidium iodide (PI; red; necrosis) and later stained with Hoechst 33342 (Hoechst; blue; nuclei). Live PC12, BV-2 phagocytosis of live PC12 cells, visible as punctate CFSE staining without propidium iodide. Dead PC12, BV-2 phagocytosis of necrotic PC12 cells, visible as punctate CFSE and propidium iodide co-staining. Representative images are shown. Scale bars: 20 µm. (B) Healthy PC12 co-cultured with BV-2 cells for 3 h are not necrotic when phagocytosed by untreated or LPS-pre-activated (24 h, 100 ng/ml) BV-2 cells, in contrast with killed PC12 cells, which are. Live PC12 (CFSE+ PI−), ***P<0.001 versus live PC12 untreated; dead PC12 (CFSE+ PI+), ^###P<0.001 versus live PC12 untreated; ^&&&P<0.001 versus dead PC12 untreated. n=4. (C) LPS-pre-activated BV-2 cells induce PC12 PtdSer exposure during 3 h of co-incubation (3 h), which is reversed after 24 h in untreated medium (3 h+24 hr untreated), without increasing the proportion of cells with condensed chromatin. Not chromatin condensed; **P<0.01, ***P<0.001 versus 3 h untreated; ^###P<0.001 versus 3 h LPS. Chromatin condensed; ^&&&P<0.001 versus 3 h untreated; ^$$$P<0.001 versus 3 h LPS. Note that 11±3% of untreated PC12 cells exposed PtdSer. n=4. (D) Phase-contrast and fluorescent images of 50 µM TAMRA-stained PC12 cells (red; PC12) recovered from 3 h co-incubation with BV-2 (unstained), and later stained with annexin-V–FITC (annexin V; green; exposed PtdSer) and Hoechst 33342 (blue; nuclei). Untreated, PC12 cells from co-culture with untreated BV-2 cells show little PtdSer exposure. LPS-treated, PC12 cells from co-culture with LPS-pre-activated BV-2 cells strongly expose PtdSer. Representative images are shown. Scale bars: 20 µm. (E) Annexin V (100 nM) decreases LPS-pre-activated BV-2 cell phagoptosis of PC12 during 3 h of co-incubation. ***P<0.001 versus untreated; ^###P<0.001 versus LPS. n=4. (F) Annexin V prevents loss of PC12 cells (both healthy and with condensed chromatin) from LPS-pre-activated BV-2 co-cultures. Healthy; **P<0.01 versus untreated; ^###P<0.001 versus LPS. Chromatin condensed; ^&&&P<0.001 versus untreated; ^$$$P<0.001 versus LPS. n=3. Data shown are means±s.e.m. Statistical significance was quantified by one-way ANOVA and Bonferroni test.

As phagoptosis of cells by macrophages and microglia is often mediated by exposure of the ‘eat-me’ signal PtdSer on the target cell (Neher et al., 2011), we tested whether the PC12 cells were exposing PtdSer. We found that 11±3% (mean±s.e.m.) of untreated PC12 cells exposed PtdSer, which was increased 1.7-fold by addition of microglia. Co-incubation with LPS-pre-activated microglia further increased the PtdSer exposure of PC12 cells by 2.5-fold (Fig. 3C,D). However, this PtdSer exposure was reversed if the PC12 were removed from the microglia for 24 h, and these cells remained viable (Fig. 3C). Thus co-incubation with activated microglia caused PC12 cells to reversibly expose PtdSer on their surface.

In order to test whether the exposure of PtdSer on PC12 cells was responsible for their phagocytosis, we added the PtdSer-binding protein annexin V to the co-incubation to block exposed PtdSer and prevent its recognition by microglia (van Genderen et al., 2008). Addition of annexin V inhibited uptake of PC12 into LPS-pre-activated microglia (Fig. 3E) and prevented loss of PC12 in the presence of LPS-pre-activated microglia (Fig. 3F). Thus the PtdSer exposure on PC12 cells induced by activated microglia appears to be responsible for their uptake.

PtdSer exposure by cells can be caused either by Ca²⁺-activated PtdSer scramblases (Suzuki et al., 2010) or by caspase-activated scramblases (Bevers and Williamson, 2010). We therefore tested whether co-incubation of PC12 cells with activated microglia caused caspase activation in the PC12 cells. Activated microglia induced a 1.8-fold increase in the measured caspase-3 activity of PC12 (Fig. 4A). For comparison, cycloheximide induced a 7.6-fold increase in caspase-3 activity in PC12 cells, thus the increase caused by activated microglia is mild. Furthermore, the mild increase in caspase-3 activity induced by activated microglia returned to basal levels after 24 h (Fig. 4A). In order to test whether the caspase activation induced by activated microglia was responsible for the PtdSer exposure on the co-incubated PC12 cells, we added the pan caspase inhibitor Z-Val-Ala-DL-Asp(OMe)-fluoromethylketone (zVAD). zVAD prevented the PtdSer exposure by the PC12 cells (Fig. 4B) without affecting PC12 viability (necrotic PC12 cells when untreated, 4.8±0.7%, and when treated with zVAD, 4.1±0.5%). Thus the activated-microglia-induced PtdSer exposure on PC12 cells appears to be mediated by the mild caspase activation in the PC12 cells. In order to determine whether the caspase activation in the PC12 cells was also responsible for their phagocytosis, we tested whether caspase inhibition by zVAD blocked their phagocytosis. zVAD had no effect on phagocytosis by non-activated microglia, but decreased the phagocytosis of PC12 cells by activated microglia by ∼20%, so that the LPS-induced increase in phagocytosis was decreased by about 40% (Fig. 4C). This indicates that the mild caspase activation in PC12 cells is partially responsible for their phagocytosis by activated microglia. zVAD did not affect microglia viability or microglia phagocytosis of beads (Fig. S3A,B). Thus, activated microglia appear to eat live PC12 cells, which have been ‘stressed’ by the activated microglia to reversibly expose PtdSer, which is partially due to the mild caspase activation in the co-incubated PC12 cells.

Fig. 4.

Microglia and NO cause reversible apoptosis of PC12, inducing phagocytosis. (A) After 24 h of LPS treatment (100 ng/ml), BV-2 cells induce caspase-3 activation in PC12 cells during 3 h of co-incubation (3 h), which is reversed after 24 h (3 hr+24 hr untreated). ***P<0.001 versus no BV-2; ^###P<0.001 versus LPS 3 h. n=4. (B) zVAD (100 µM) prevents PC12 PtdSer exposure induced by 3 h co-culture with untreated and LPS-pre-activated BV-2 cells. **P<0.01, ***P<0.001 versus untreated; ^###P<0.001 versus LPS; n=4. (C) zVAD decreases PC12 phagocytosis by LPS-pre-activated BV-2. ***P<0.001 versus untreated; ^##P<0.01 versus LPS; n=4. (D) 1400 W (20 µM) decreases PC12 phagocytosis by untreated and 24 h LPS-pre-activated (100 ng/ml) BV-2 during 3 h of co-incubation. ***P<0.001 versus untreated; ^###P<0.001 versus LPS. n=4. (E) Phase-contrast and fluorescent images of PC12 stained with annexin-V–FITC (annexin V; green; exposed PtdSer) and Hoechst 33342 (Hoechst; blue; nuclei). Untreated, untreated PC12 cells do not expose PtdSer and do not have condensed chromatin. DETA-NO, PC12 cells treated with DETA-NO for 3 h expose PtdSer and have condensed chromatin. Representative images are shown. Scale bars: 20 µm. (F) 3 h treatment (3 h) of PC12 cells with DETA-NO (1 mM) causes PtdSer exposure and chromatin condensation, which is reversed after 24 h in untreated medium (3 hr+24 hr untreated). Not chromatin condensed; **P<0.01 versus 3 h untreated; ^#P<0.05 versus 3 h DETA-NO. Chromatin condensed; ^&&&P<0.001 versus 3 h untreated; ^$$$P<0.001 versus 3 h DETA-NO; ^£££P<0.001 versus 3 h+24 h untreated. n=4. (G) DETA-NO treatment during 3 h co-incubation of PC12 cells with BV-2 cells induces PC12 PtdSer exposure in the absence of chromatin condensation and causes healthy and chromatin condensed PC12 cell loss. Healthy; ***P<0.001. PtdSer exposed; ^###P<0.001. Chromatin condensed; ^&&&P<0.001. n=3. (H) Treatment of co-cultures with DETA-NO increases PC12 phagocytosis by BV-2 cells. ***P<0.001. n=5. Data shown are means±s.e.m. Statistical significance was quantified by one-way ANOVA and Bonferroni test, except for data shown in panels in G and H, where student's t-test was used.

NO from microglia induces PtdSer exposure on PC12 cells and their subsequent phagocytosis

We next investigated what the activated microglia might release to cause PtdSer exposure and phagocytosis of the PC12. LPS-activated microglia are known to express iNOS and release NO (Kumar et al., 2014; Wen et al., 2011). We therefore tested whether blocking iNOS with the specific inhibitor N-(3-(aminomethyl)benzyl)acetamidine dihydrochloride (1400 W) prevented uptake of live PC12 cells, and found that it partially inhibited this phagocytosis (Fig. 4D). Addition of 1400 W to microglia did not inhibit their phagocytosis of beads (Fig. S3C), indicating that NO from iNOS did not affect the phagocytic capacity of BV-2 microglia.

We then tested whether NO could cause PtdSer exposure on PC12 cells and their subsequent phagocytosis by microglia. A 3-h treatment of PC12 cells with the NO donor DETA-NONOate (Feelisch, 1998) (DETA-NO; 1 mM) caused a 4.7-fold increase in PtdSer exposure as well as a 22-fold increase in chromatin condensation (Fig. 4E,F). This transient DETA-NO treatment did not affect PC12 viability (after 3 h treatment, necrotic PC12 cells in untreated samples, 4.2±0.6%; necrotic cells in DETA-NO-treated samples, 4.3±0.4%; mean±s.e.m.), and both chromatin condensation and PtdSer exposure returned to basal levels after 24 h (Fig. 4F), indicating that the NO-induced stress is reversible. Addition of DETA-NO to co-cultures of PC12 cells and microglia also increased PC12 PtdSer exposure 2.4-fold, resulting in loss of ∼20% of PC12 cells (Fig. 4G) and a 2.5-fold increase in the uptake of PC12 cells into microglia (Fig. 4H). The effect of DETA-NO was apparently on the PC12 cells, not the microglia, as pre-treating the microglia with DETA-NO then changing the medium prior to PC12 cell addition, resulted in virtually no effect of the DETA-NO (Fig. S3D), and DETA-NO did not affect phagocytosis of 1-µm beads [beads per cell in DETA-NO-treated was 92±12% relative to untreated (100%)]. Thus, NO from iNOS in activated microglia appears to be necessary and sufficient to induce PC12 PtdSer exposure and subsequent phagocytosis by microglia.

PtdSer binding of the vitronectin receptor through MFG-E8 is required for microglial phagoptosis of PC12

PtdSer-exposed cells can be recognised by a variety of opsonins and receptors, including MFG-E8 released by activated microglia (Fuller and Van Eldik, 2008) and its receptor, VNR, which is found on microglia (Akiyama et al., 1991). Phagocytosis of PC12 cells by LPS-activated microglia was decreased by H-Arg-Gly-Asp-Ser-OH (RGDS) (Haubner et al., 1996), a VNR antagonist (Fig. 5A). A more specific VNR inhibitor, cyclo(Arg-Gly-Asp-D-Phe-Val) (cRGDfV), reduced phagoptosis by 30% in non-activated cultures and by 60% in LPS-pre-activated cultures (Fig. 5B). cRGDfV also fully prevented the loss of PC12 cells in the presence of LPS-pre-activated microglia, and the cells rescued in this way were viable and remained viable 24 h later (Fig. 5C), further confirming death was by phagoptosis. In the differentiated PC12 system, RGDS completely prevented phagoptosis, inhibiting loss of both 3-day and 7-day differentiated PC12 cells (Fig. 5D), as well as uptake of 3-day differentiated PC12 cells into microglia (Fig. S3E). VNR couples to PtdSer-exposed cells through the opsonin MGF-E8, which binds to both PtdSer through a C2 domain and to VNR through a RGD sequence. We used a function-blocking antibody against MFG-E8 to test its role, and found that phagocytosis of PC12 cells by LPS-pre-activated microglia was decreased (Fig. 6A). Thus PtdSer recognition by MFG-E8 and VNR appears to mediate microglial phagoptosis of PC12 cells.

Fig. 5.

The vitronectin receptor is required for microglial phagocytosis of PC12. (A) RGDS (50 µM) decreases LPS-pre-activated (100 ng/ml for 24 h) BV-2 phagocytosis of PC12 cells during 3 h of co-incubation. ***P<0.001 versus untreated; ^##P<0.01 versus LPS. n=4. (B) cRGDfV (50 µM) decreases the proportion of untreated and LPS-pre-activated BV-2 cells taking up PC12 cells by phagocytosis. **P<0.01, ***P<0.001 versus untreated; ^###P<0.001 versus LPS. n=3. (C) cRGDfV prevents loss of healthy and dead (chromatin condensed propidium-iodide-positive cells are labelled as ‘PI +ve - necrotic’) PC12 cells during a 3 h co-incubation with 24 h LPS-pre-activated (100 ng/ml) BV-2 cells. PC12 cells rescued from phagocytosis by cRGDfV remain viable after a further 24 h (3 h+24 h). Healthy; **P<0.01, ***P<0.001 versus untreated equivalent time; ^###P<0.001 versus LPS equivalent time; ^&&P<0.01, ^&&&P<0.001 versus 3 h equivalent condition. Chromatin condensed PI +ve; ^$$P<0.01, ^$$$P<0.001 versus untreated equivalent time; ^£££P<0.001 versus LPS equivalent time; ^∼∼∼P<0.001 versus 3 h equivalent condition. n=4. (D) RGDS (50 µM) prevents LPS-activated BV-2-induced loss of healthy and chromatin condensed 3- and 7-day differentiated PC12 cells during 72 h of co-incubation. Healthy; ***P<0.001 versus untreated; ^###P<0.001 versus LPS. Chromatin condensed; ^&&P<0.01 versus untreated; ^$$P<0.01, ^$$$P<0.001 versus LPS. n=6 (3-day differentiated PC12 cells); n=3 (7-day differentiated PC12 cells). Data shown are means±s.e.m. Statistical significance was quantified by one-way ANOVA and Bonferroni test.

Fig. 6.

MFG-E8 and P2Y₆ receptor mediate microglial phagocytosis of PC12 and differentiated PC12 cells. (A) Anti-MFG-E8 antibody (5 µg/ml), but not control rabbit IgG (5 µg/ml), decreases LPS-pre-activated BV-2 phagocytosis of PC12 cells. ***P<0.001 versus untreated; ^###P<0.001 versus LPS. n=4. (B) MRS 2578 (1 µM) decreases LPS-pre-activated (100 ng/ml for 24 h) BV-2 phagocytosis of PC12 cells during 3 h of co-incubation. ***P<0.001 versus untreated; ^###P<0.001 versus LPS. n=4. (C) UDP (100 µM) increases untreated and LPS-pre-activated BV-2 phagocytosis of PC12 cells. *P<0.05, ***P<0.001 versus untreated; ^###P<0.001 versus LPS. n=5. (D) MRS 2578 prevents 24 h LPS-pre-activated (100 ng/ml) BV-2-induced loss of healthy and chromatin condensed 3- and 7-day differentiated PC12 cells during 72 h of co-incubation. Healthy; ***P<0.001 versus untreated; ^###P<0.001 versus LPS. Chromatin condensed; ^&&P<0.01 versus untreated; ^$$P<0.001 versus LPS. n=6 (3-day differentiated PC12 cells); n=3 (7-day differentiated PC12 cells). Data shown are means±s.e.m. for at least three independent experiments. Statistical significance was quantified by one-way ANOVA and Bonferroni test.

P2Y₆ receptor activation mediates microglia phagoptosis of PC12

As P2Y₆ receptor activation has been shown to mediate microglial phagoptosis of neurons (Neher et al., 2014), we tested this in our system. Treatment with the P2Y₆ receptor antagonist MRS 2578 (Mamedova et al., 2004) decreased phagoptosis of PC12 cells by LPS-pre-activated microglia (Fig. 6B), whereas the P2Y₆ receptor agonist UDP (Koizumi et al., 2007) increased bead phagocytosis [beads per cell untreated in UDP-treated was 147±12% relative to untreated (set at 100%); P<0.001] and PC12 phagoptosis 1.3-fold (Fig. 6C). Note that the effects of the P2Y₆ receptor agonist and antagonist are small but significant in this system. However, activated microglial phagoptosis of 3- and 7-day differentiated PC12 cells over a 72-h period was completely prevented by MRS 2578; both loss of healthy PC12 (Fig. 6D) and uptake into microglia (Fig. S3F) were inhibited. P2Y₆ receptor activation is therefore an essential step in PC12 phagoptosis by microglia in this system.

BV-2 phagocytose mouse neuroblastoma N2A cells

In order to test whether microglia could phagocytose other tumour cell lines, we used the mouse neuroblastoma cell line N2A. We found that co-incubation with BV-2 microglia for 3 h resulted in significant phagocytosis of N2A cells as measured by flow cytometry (the percentage of BV-2 cells containing N2A cells was 17±4%, n=4; mean±s.e.m.), and pre-activation of the microglia with either LPS or TNF-α increased the phagocytosis (Fig. 7A). Fluorescence imaging was used to confirm that the N2A cells or parts of N2A cells were inside the BV-2 cells (Fig. 7B,C; Fig. S4). Thus, microglia can also phagocytose neuroblastoma cells.

Fig. 7.

Activation of BV-2 with LPS and TNF-α increases phagocytosis of live N2A cells. (A) Microglial phagocytosis of N2A cells measured by flow cytometry after a 3 h of co-culture of BV-2 cells with N2A cells; BV-2 cells were pre-activated for 24 h with LPS (100 ng/ml) or TNF-α (50 ng/ml), which increased BV-2 phagocytosis of N2A cells. Data shown are means±s.e.m. for four independent experiments; *P<0.05 (statistical significance was quantified by one-way ANOVA and Bonferroni test). (B) Projection of a confocal z-stack images of co-cultured cells, and (C) separate z-stack images showing N2A cells (red), BV-2 cells (green) and BV-2 cells with engulfed N2A cells in different stages of digestion (yellow).

DISCUSSION

Addition of PC12 cells to activated microglia, or addition of activated microglia to differentiated PC12 cells, resulted in loss of PC12 cells and uptake into microglia. This uptake and loss of PC12 was due to phagocytosis by microglia as it was inhibited at 4°C and by the phagocytosis inhibitor cytochalasin D (Elliott and Winn, 1986). Although microglia had a greater capacity to phagocytose dead PC12 cells than live PC12 cells, they still phagocytosed live PC12 cells at a significant rate, particularly after activation by LPS. When microglia were added to differentiated PC12 cells, significant PC12 cell loss only occurred when microglia were activated with LPS.

That the PC12 cells were not necrotic when phagocytosed is shown by the finding that the dye propidium iodide was taken up with the PC12 cells when dead PC12 cells were added but not when live PC12 cells were added. Thus, the plasma membrane of the PC12 cells was intact at the time they were phagocytosed. Crucially, to determine whether phagocytosis was the cause of death, we inhibited phagocytosis and tested whether the PC12 cells rescued remained viable. We found that the ‘saved’ PC12 cells were viable when phagocytosis was inhibited with annexin V or cRGDfV, and that these cells remained healthy when cultured for a further 24 h. When activated microglia were added to differentiated PC12 cells, inhibition of phagocytosis with cytochalasin D, RGDS or MRS 2578 also rescued viable PC12 cells. If the activated microglia had killed the PC12 cells first and subsequently phagocytosed them, inhibition of phagocytosis would leave dead cells. However, we found that all the PC12 cells lost as a result of phagocytosis were viable when phagocytosis was inhibited, indicating that phagocytosis was the cause of death, that is, the PC12 died by phagoptosis.

Phagocytosis of cells is normally mediated by PtdSer exposure, and we found that PC12 cells co-cultured with activated microglia did indeed expose PtdSer. However, this PtdSer exposure was fully reversible if the PC12 cells were separated from the microglia cells, and these PC12 cells remained viable. Thus, activated microglia induced the reversible PtdSer exposure on neurons. Reversible PtdSer exposure on viable cells has been shown in many different cell types and conditions (Hammill et al., 1999; Mackenzie et al., 2005; Stowell et al., 2009). To test whether the PtdSer exposure was responsible for the phagocytosis, we blocked the exposed PtdSer by adding the PtdSer-binding protein annexin V (van Genderen et al., 2008), and found that this inhibited the PC12 cell uptake and loss. Thus the PtdSer exposure on PC12 induced by activated microglia was responsible for the phagocytosis of the PC12.

Microglial phagocytosis of PtdSer-exposed neurons is mediated by the bridging opsonin MFG-E8 binding to PtdSer-exposed neurons and to the integrin vitronectin receptor on phagocytes through an RGD motif in MFG-E8 (Fricker et al., 2012a). We found that phagocytosis of PC12 by activated microglia was inhibited by anti-MFG-E8 antibodies and by RGD-based inhibitors of the VNR (Haubner et al., 1996). Thus, phagocytosis of the PC12 cells appears to be mediated by a PtdSer–MFG-E8–VNR pathway, as occurs with primary microglia and neurons. UDP activation of microglial P2Y₆ receptors can act as a final engulfment signal (Koizumi et al., 2007), and we found that addition of UDP increased phagocytosis of PC12 cells, whereas MRS 2578 inhibition of P2Y₆ receptors (Mamedova et al., 2004) mildly inhibited phagocytosis. This UDP activation of P2Y₆ receptors can serve as an engulfment signal in the microglial phagocytosis of PC12 cells, as in primary microglia and neurons (Neher et al., 2014).

PtdSer exposure can be caused either by caspase activation or by Ca²⁺ and oxidant elevation (Chong et al., 2004; Zwaal et al., 2005). We found that caspase activity was mildly but reversibly increased in PC12 cells co-cultured with activated microglia, and that the caspase inhibitor zVAD prevented the PtdSer exposure and partially inhibited the phagocytosis of the PC12 cells. Ca²⁺-activated PtdSer scramblases, such as TMEM16F, can cause reversible PtdSer exposure, but as zVAD fully prevented the PtdSer exposure, it appears that a caspase-activated scramblase is most important for PtdSer exposure in the conditions used. Thus, it appears that the activated microglia cause a mild reversible caspase activation in the PC12 cells that is responsible for the reversible PtdSer exposure on the PC12 cells, and partially responsible for their phagocytosis.

LPS-activated microglia express iNOS and release NO (Bal-Price and Brown, 2001), which can cause caspase activation in PC12 cells (Bal-Price and Brown, 2000). We found here that inhibiting iNOS with 1400 W partially prevented microglial phagocytosis of PC12 cells, and that the NO donor DETA-NO was sufficient to induce reversible PtdSer exposure on PC12 cells and their uptake into microglial cells.

PC12 cells are derived from a pheochromocytoma and N2A cells from a neuroblastoma. The research here suggests the possibility that activated microglia protect against tumours by ‘stressing’ the cancer cells and then phagocytosing these cells. And this is supported by evidence that microglia can phagocytose live glioma cells (Kopatz et al., 2013). However, it would be important to determine whether phagocytosis can protect against brain tumours in vivo.

Overall, it appears that activated microglia phagocytose stressed-but-viable PC12 cells, resulting in their death by phagoptosis – a possible mechanism is depicted in Fig. 8. These results have the implication that apoptosis can be ‘reversible’, and that, in this case, caspase activation can be insufficient to induce cell death on its own but sufficient to induce phagocytosis of the cell and therefore cell death by phagoptosis. Cell death in C. elegans has been shown to be partly mediated by phagocytosis in conditions where caspase activation is partial (Hoeppner et al., 2001; Neukomm et al., 2011; Reddien et al., 2001). Caspase activation by apoptotic pathways can occur in viable neurons and mediate physiological processes (D'Amelio et al., 2012). Thus apoptotic activation of caspases does not always result in apoptotic cell death, but rather, where the caspase activation is mild, can result in cell death by phagoptosis.

Fig. 8.

Possible mechanism of microglial phagoptosis of PC12. LPS, rendered inactive by polymyxin B (PMX), activates BV-2 through TLR4. This causes production of NO by iNOS, which can be inhibited by 1400 W. NO from iNOS or DETA-NO induces mild and reversible caspase-3 activation in PC12 cells (which is inhibitable by zVAD), causing reversible exposure of PtdSer (PS, which is blocked by annexin V). Exposed PtdSer is detected by VNR (which is blocked by RGDS or cRGDfV peptides) on the BV-2 cells through the secreted factor MFG-E8 (which can be blocked by specific antibodies). Stressed PC12 cells might secrete UDP, activating their engulfment by BV-2 through P2Y₆ receptors (P2Y6R, blocked by MRS 2578, MRS). Subsequent uptake is prevented by cytochalasin D inhibition of actin polymerisation. As PC12 caspase-3 activation and PtdSer exposure are reversible, inhibition of phagocytosis leaves viable PC12 cells.

MATERIALS AND METHODS

Materials

Lipopolysaccharide from Salmonella enterica serotype typhimurium (LPS) and 5(6)-carboxyfluorescein diacetate-N-succinimidyl ester (CFSE) were purchased from Sigma, MRS 2578 and UDP from Tocris, N-(3-(aminomethyl)benzyl)acetamidine dihydrochloride (1400 W), DETA NONOate (DETA-NO), 7-amino-4-methylcoumarin (AMC) and Ac-DEVD-AMC from Enzo Life Sciences, cyclo(Arg-Gly-Asp-D-Phe-Val) (cRGDfV), H-Arg-Gly-Asp-Ser-OH (RGDS) and Z-Val-Ala-DL-Asp(OMe)-fluoromethylketone (zVAD) from Bachem, isolectin-B₄–Alexa-Fluor-488 from Griffonia simplicifolia (IB₄) and 1-µm fluorescent-carboxylate-modified microspheres were from Invitrogen, 5-µm fluorescent carboxyl particles were from Spherotech, 5-(and-6)-carboxytetramethylrhodamine succinimidyl ester (TAMRA) were from Biotium Inc., annexin-V–FITC was from Immunotools (Friesoythe, Germany), annexin V was from BioVision, anti-MFG-E8 (G-17) antibody and control IgG were from Santa Cruz Biotechnology, and F(ab’)2 anti-IgG was from Jackson ImmunoResearch Laboratories. Unless otherwise indicated, all other materials were purchased from Sigma.

Cell culture

All tissue culture medium was supplemented with 100 units/ml penicillin G and 100 µg/ml streptomycin sulphate (Invitrogen) or 100 µg/ml gentamicin (Invitrogen). All cells were kept at 37°C and 5% CO₂ in 75-cm² flasks (Nunc Thermo Scientific; Massachusetts, USA) and seeded in 24-well plates (Nunc Thermo Scientific).

Cell lines

The murine microglial cell line BV-2 (Blasi et al., 1990; Bocchini et al., 1992) (passage <30) was maintained in Dulbecco's modified Eagle's medium (DMEM; Invitrogen) supplemented with 10% fetal bovine serum (FBS; Invitrogen). At confluence, cells were harvested using 0.5% trypsin (Invitrogen) in phosphate-buffered saline pH 7.2 (PBS; Invitrogen) and seeded at 4×10⁴ cells/well for microscopy or 5×10⁴ cells/well for flow cytometry in DMEM supplemented with 0.5% FBS (0.5% glial medium). Rat pheochromocytoma cells (PC12) (Greene and Tischler, 1976) were maintained in Roswell Park Memorial Institute (RPMI)-1640 medium (Invitrogen) supplemented with 10% horse serum (Invitrogen) and 5% FBS, in flasks coated with 0.5 mg/ml collagen type IV. For differentiated PC12 cells, cells were harvested at 80% confluence using 0.5% trypsin in PBS, seeded on collagen at 5×10⁴ cells/well in RPMI-1640 supplemented with 0.5% horse serum and 100 ng/ml nerve growth factor 7S (Invitrogen), and left to differentiate for 3 or 7 days. Unless stated otherwise, the PC12 cells used were naïve. N2A (Neuro-2A) cells are derived from a mouse neuroblastoma, were a kind gift of Bazbek Davletov, University of Sheffield, UK, and were cultured in DMEM plus 10% FBS. These cell lines were not recently authenticated or tested for contamination.

Microscopy

Cells were imaged using a Leica DMI6000 microscope (Leica Microsystems; Wetzlar, Germany). Four microscopic fields (each 1.9×10⁵ µm²) per well in at least two wells per condition were quantified for a single experiment. Cultures were stained with the nuclear stains Hoechst 33342 (4 µg/ml; blue channel) and propidium iodide (4 µg/ml; red channel) and the microglial-specific dye IB₄ (1 µg/ml; green channel) as indicated. Dead or dying cells were identified by nuclear morphology (cells with condensed chromatin were considered apoptotic) or by whether they had a permeable plasma membrane (staining with propidium iodide; positive staining meant the cells were considered necrotic). All cell quantification and image manipulation was carried out using ImageJ.

Phagocytosis assays

BV-2 treatment for phagocytosis

BV-2 cells were allowed to adhere for 24 h then stimulated with LPS (100 ng/ml) for 24 h. Where indicated, the phagocytosis inhibitors RGDS (50 µM), cRGDfV (50 µM), MRS 2578 (1 µM), annexin V (100 nM) and cytochalasin D (0.5 µM), the inhibitors zVAD (100 µM) and 1400 W (20 µM) or the antibodies anti-MFG-E8 antibody (5 µg/ml) and rabbit control IgG (5 µg/ml) were then added for 30 min. The activators UDP (100 µM) and DETA-NO (1 mM) were added 48 h after seeding for 30 min or 1 h, respectively.

Where indicated, LPS (100 ng/ml) was blocked with polymyxin B (1000 units/ml) for 30 min at 37°C prior to treatment. Antibodies were Fc blocked by incubation for 30 min at 4°C with a 3.3-fold molar excess of the host-species Fc region specific F(ab’)2 fragment.

Phagocytosis measured by flow cytometry

PC12 cells were harvested using 0.5% trypsin in PBS, stained for 15 min with TAMRA (50 µM) in PBS, then washed with PBS. PC12 cells were seeded on BV-2 cells treated as above (BV-2 treatment) either immediately at 3×10⁵ cells/well (healthy PC12 cells with 0.7±0.06% necrotic and 1.2±0.2% apoptotic; mean±s.e.m.) or at 2.5×10⁵ cells/well after 24 h death induction by treatment with staurosporine (1 µM) or cycloheximide (100 µg/ml; dead PC12; proportion necrotic=95±0.9%). After 3 h, cultures were stained with IB₄ and washed with ice-cold PBS to remove excess PC12 cells. BV-2 cells and remaining PC12 cells were harvested with 0.5% trypsin in PBS, spun down at 130 g for 5 min at 4°C and resuspended in PBS on ice.

The FL1 (excitation 488 nm, emission 530±15 nm; green; IB₄ in BV-2 cells) and FL3 (excitation 640 nm, emission >670 nm; red; TAMRA in PC12 cells) fluorescence of the cells was measured using a BD Accuri C6 flow cytometer (BD Biosciences; California, USA). FL1 fluorescence was used to positively select BV-2 cells and eliminate PC12 and debris. 1×10⁴ BV-2 events per well, from three separate wells per condition, were recorded for a single experiment. The average FL3 fluorescence of this BV-2 population, reflecting TAMRA-stained PC12 phagocytosis, was determined. The background FL3 intensity of IB₄-stained BV-2 cells alone was subtracted from the mixed BV-2 and PC12 culture FL3 intensities. Flow cytometry analysis was performed using BD Accuri C6 software.

Where indicated, TAMRA-stained live or dead PC12 cells were pre-treated with zVAD for 1 h, washed with PBS to remove excess reagent and seeded as above. To inhibit phagocytosis, plates were kept at 4°C for the duration of co-incubation. For serum concentration experiments, 24 h after seeding and prior to LPS stimulation, BV-2 cells were washed with PBS and fresh DMEM or 0.5% glial medium added. For medium change experiments, BV-2 cells were pre-treated with DETA-NO for 1 h then washed with PBS and fresh untreated 0.5% glial medium added.

Phagocytosis of TAMRA-stained N2A cells by IB₄-stained BV-2 cells was measured similarly by flow cytometry after 3 h of co-incubation. BV-2 were pre-activated for 24 h with LPS (100 ng/ml) or TNF-α (50 ng/ml).

Phagocytosis measured by microscopy

Healthy TAMRA-stained PC12 cells were seeded on BV-2 cells at 2×10⁵ cells/well, and after 3 h, stained with Hoechst 33342 and IB₄, washed twice with ice-cold PBS to remove excess PC12 cells and imaged. The proportion of BV-2 cells staining with TAMRA (indicating phagocytosis of PC12) was quantified.

To assess viability of phagocytosed PC12 cells, live or dead PC12 cells were stained with CFSE (5 µM), washed with PBS and seeded on BV-2 cells at 2×10⁵ cells/well. Propidium iodide was added concomitantly, and mixed cultures stained with Hoechst 33342 after 3 h. BV-2 cells were identified by nuclear morphology and phase-contrast images, and the proportion of CFSE-stained BV-2 cells (healthy PC12 cells phagocytosed), CFSE and propidium-iodide-stained BV-2 cells (necrotic PC12 cells phagocytosed) and propidium-iodide-stained BV-2 cells (necrotic BV-2 cells phagocytosed) were quantified.

Phagocytosis of TAMRA-stained N2A cells by IB₄-stained BV-2 cells was measured after 3 h of co-incubation using an Olympus FV1000 Upright confocal microscope after fixation of the cells with 4% paraformaldehyde.

PC12 cells recovered from phagocytosis co-cultures

Experiments were performed as for flow cytometry measurement of PC12 phagocytosis, except that after 3 h of co-incubation the TAMRA-stained-PC12-rich supernatant was collected and transferred to collagen-coated wells. Cells were allowed to adhere for 1.5 h, stained with Hoechst 33342 and IB₄ and imaged. The number of healthy and chromatin condensed PC12 cells was quantified. Because PC12 cells were stained with TAMRA, propidium iodide could not be used concurrently to image necrosis; chromatin condensation was therefore used as an overall measure of cell death (apoptotic and necrotic) as all propidium-iodide-positive cells also presented with condensed chromatin. To quantify necrosis, selected experiments were repeated with unstained PC12 cells and stained with propidium iodide prior to imaging.

To measure PtdSer exposure, TAMRA-stained PC12 cells were treated, collected and adhered as above, but stained with Hoechst 33342 and annexin-V–FITC (4 µg/ml; green channel). Cultures were fixed at 22°C with 4% paraformaldehyde, quenched with glycine (30 mM) in PBS then washed three times with PBS. Cultures were imaged and the number of healthy, chromatin condensed and annexin-V–FITC-positive (PtdSer-exposed) PC12 cells quantified. Alternatively, an equal volume of fresh PC12 medium was added and cells left for a further 24 h before staining and fixing. To investigate the effect of DETA-NO on PC12 PtdSer exposure, PC12 cells were treated with DETA-NO for 3 h then processed as above.

For cytochalasin D recovery experiments in PC12 cells, unstained PC12 cells were used and those washed off counted by flow cytometry. The FL1 (IB₄ in BV-2 cells) fluorescence of the cells was measured and used to eliminate BV-2 cells. The number of PC12 phagocytosis events in 100 μl of supernatant from three separate wells per condition was recorded for a single experiment.

Phagocytosis of differentiated PC12 cells

BV-2 cells were seeded at 1.25×10⁴ cells/well on TAMRA-stained (50 µM) differentiated PC12 cells. Where indicated, the phagocytosis inhibitors cytochalasin D (0.2 μM), MRS 2578 (1 µM; added daily) or RGDS (50 µM) were added concomitantly. Mixed cultures were stimulated with LPS (100 ng/ml) after 24 h. After a further 72 h, cultures were stained with Hoechst 33342, propidium iodide and IB₄ and imaged. The number and viability of PC12 cells, as well as the number of TAMRA-positive BV-2 cells (BV-2 having phagocytosed PC12) was quantified.

Bead phagocytosis

BV-2 cells were treated as described in the section ‘BV-2 treatment for phagocytosis’, then incubated with a 0.002% (w/v) solution of 1-µm carboxylate-modified microspheres for 1 h or a 0.025% (w/v) solution of 5-µm carboxyl particles for 2 h. The medium was removed and cells washed with ice-cold PBS to remove unattached beads. Cells were stained with Hoechst 33342, imaged, and the number of beads per cell quantified in healthy cells (not chromatin condensed).

PC12 caspase activity

Experiments were carried out as for flow cytometry measurement of PC12 phagocytosis, except the mixed culture supernatant was collected after phagocytosis. The supernatant was spun down at 130 g for 5 min and cells washed with PBS. Cells were resuspended and lysed for 5 min in a lysis buffer consisting of 20 mM HEPES pH 7.4, 250 mM sucrose, 10 mM KCl, 1.5 mM MgCl₂ (VWR; Pennsylvania, USA), 0.5 mM EDTA, 0.5 mM EGTA, 0.1% (w/v) CHAPS, 0.1% Triton X-100 (VWR) and a 1 in 1000 dilution of protease inhibitor cocktail. The lysate was centrifuged at 10,000 g at 4°C for 5 min and the supernatant collected. Samples were diluted 1 in 4 with an assay buffer consisting of 20 mM HEPES pH 7.4, 10% (w/v) sucrose, 5 mM DTT, 1 mM EDTA, 0.1% (w/v) CHAPS, a 1 in 1000 dilution of protease inhibitor cocktail and 26.67 μM Ac-DEVD-AMC. Fluorescence (excitation 340 nm, emission 510 nm) was measured every minute for 3 h using a Fluostar Optima plate reader (BMG Labtech; Orternberg, Germany), and the average of the last 15 min recorded. AMC standards were used to calculate the amount of AMC produced and, hence, lysate caspase activity, which was adjusted for the measured protein concentration. Caspase activity was measured in duplicate in supernatants from three wells per condition for a single experiment. PC12 cells were treated with cycloheximide (100 µg/ml) for 3 h as a positive control. For caspase activity after 24 h, supernatants were transferred to new wells, an equal volume of fresh PC12 medium added, and cells left for a further 24 h before collection and lysis.

Statistical analysis

All data shown are expressed as the mean±s.e.m. for at least three independent experiments. Normality of data was verified by a Kolmogorov–Smirnov test. Means were compared by Student's t-test or by one-way ANOVA and post-hoc Bonferroni test (normally-distributed data) and by Kruskal–Wallis and Mann–Whitney U test (data that was not normally distributed). P<0.05 was considered significant. SPSS software (IBM) was used for statistical analysis.