Apoptotic Cells Induce Immunosuppression through Dendritic Cells: Critical Roles of IFN-γ and Nitric Oxide¹

One of the most significant characteristics of cell death by apoptosis is the quick and quiet removal of the cell corpse by surrounding cells (1, 2, 3, 4). Unlike the eradication of necrotic cells, which is often associated with local inflammation, the clearance of apoptotic cells has been shown to suppress immune responses (2, 3, 5, 6). Therefore, it is highly likely that the interaction of phagocytes with apoptotic cells results in different consequences compared with their interaction with necrotic cells (3). Phagocytosis conducted by unicellular organisms is believed to be involved primarily in food up-take (7). Phagocytosis exists throughout the metazoan subkingdom. In the lower nematodes, such as Caenorhabditis elegans, one of the main functions of phagocytosis is the removal of apoptotic cells by neighboring cells (8). In higher animals, phagocytosis of apoptotic cells is largely accomplished by cells of the innate immune system such as macrophages and dendritic cells (DC)⁵ (9). Therefore, it is possible that during evolution, phagocytosis arose first for nutritional purposes in unicellular organisms, and then evolved for the removal of apoptotic cells. The phagocytic functions of innate immune cells may therefore have evolved from cells with the capability of apoptotic cell removal.

Although a century has passed since Élie Metchnikoff first discovered that the clearance of regressing tissue during amphibian morphogenesis occurs without inflammation (10), the link between immune tolerance and phagocytosis of apoptotic cells was made only in recent years. Several groups have studied how apoptotic cells signal for their clearance without eliciting inflammatory responses by the engulfing phagocytes (5, 9). The hallmarks of apoptotic cell death are plasma membrane blebbing, cell shrinkage, and the loss of membrane phosphatidylserine (PS) asymmetry. The exposure of PS on the outer leaflet of the cytoplasmic membrane is recognized by a phagocyte PS receptor, milk fat globule epidermal growth factor 8 (11). In addition, a number of PS-binding proteins, such as plasma-protein β₂-glycoprotein I and protein S, can act as a bridge between apoptotic cells and phagocytes. In previous studies, we showed that blocking the interaction between PS and phagocytes with annexin V prevents apoptotic cell-induced tolerance for allogeneic heart transplants (6). In addition, recent studies have shown that carbohydrate changes on the surface of apoptotic cells may be important in triggering recognition by lectin receptors (12). C1q has also been found to bind to the blebs of apoptotic cells and interact with collectins during apoptotic cell phagocytosis (13). Furthermore, apoptotic cells display a carbohydrate group that binds to CD14 (14). Thus, phagocytosis is mediated by a number of different receptors and ligands.

Investigations in the last few years have indicated a strong correlation between defects in apoptotic cell phagocytosis and the development of autoimmunity in both humans and genetically modified mice (12, 13, 14, 15, 16, 17). Apoptotic cells are important modulators of immune responses and are believed to be a primary source of tolerogens (3, 18). There are several examples showing the induction of tolerance after injection of apoptotic cells and the role of APCs, such as macrophages and DCs, in the induction and maintenance of tolerance after phagocytosis of dead cells (6, 19, 20, 21, 22). Because DC are the most potent APCs capable of activating naive T lymphocytes, the processing of apoptotic cells by DC is likely to be important for apoptotic cell-mediated immune regulation (19, 21). Indeed, several studies have demonstrated that phagocytosis of apoptotic cells by DC leads to the generation of regulatory T cells (23) and the induction of IL-10 production (24). Additionally, TGFβ (19, 25) and PGE₂ (26) are also believed to play a role.

We have reported that, in an allogeneic rat heart transplant model, a single transfusion of a few million apoptotic splenocytes from the donor strain before transplant resulted in dramatically prolonged survival of heart allografts. This effect was prevented by administration of gadolinium chloride, which disrupts phagocyte function, and by annexin V, which blocks the binding of exposed phosphatidylserine to its receptor on phagocytes (6). In this study, we investigated the molecular mechanism controlling this apoptotic cell-mediated immunosuppression. We found no role for IL-10, TGF-β, PGE₂, CTLA-4, BTLA-4, or regulatory T cells. Instead, DC exposed to apoptotic cells (DC_ap) are affected in at least two ways: reduced ability to prime T cells to produce IFN-γ and increased propensity for IFN-γ-induced production of NO. In the presence of both DC_ap and normal DC, T cells are activated by normal DC to produce IFN-γ, which in turn induces DC_ap to release NO and thus suppress immune responses. Therefore, our studies reveal a novel mechanism through which apoptotic cells induce immunosuppression.

C57BL/6 mice (6–8 wk) were obtained from National Cancer Institute. IFN-γR1^−/− and iNOS^−/− mice (6–8 wk), both in a C57BL/6 background, were purchased from The Jackson Laboratory. All mice were maintained in the Robert Wood Johnson Medical School Vivarium, with food and water provided ad libitum. Animals were used 1 wk after receipt and were matched for gender and age in each experiment, all approved by the Institutional Animal Care and Use Committee.

Recombinant mouse IFN-γ, TNF-α, IL-1α, anti-mouse CTLA-4, anti-mouse BTLA-4, and PE-conjugated anti-mouse CD11c, CD40, CD80, CD86, and MHC-II were from eBioscience. Abs against IL-10 and TGF-β were from R&D Systems. Ab against IFN-γ was produced by Harlan Sprague Dawley. Biotin-conjugated anti-mouse IFN-γR1 was from Biolegend. Indomethacin and NG-monomethyl-l-arginine were purchased from Sigma-Aldrich.

Mice were euthanized and the femurs were isolated. Bone marrow was flushed out with RPMI 1640 medium using a 1 ml syringe (27G). The cells were washed twice with PBS and then cultured in complete RPMI 1640 medium supplemented with GM-CSF (20 ng/ml) and IL-4 (10 ng/ml). On day 3, 75% of the medium was changed. On day 5, the suspended cells were removed and the loosely adherent cells were collected as immature DC (CD11c⁺ cells were >80%). For mature DC, the immature DC were stimulated with LPS (1 μg/ml) for 24 h.

Apoptotic cells were generated by treating splenocytes with dexamethasone (1.0 μM) for 12 h. At least 80% of the resultant cells were apoptotic, as determined by 2-color staining with FITC-conjugated annexin V and propidium iodide, as well as DNA content analysis (27). Cells were washed twice with PBS before use. Immature DC_ap were generated by culturing apoptotic splenocytes with immature DC at a 10:1 ratio for 24 h followed by removal of buoyant splenocytes. Mature DC_ap were obtained from immature DC_ap by LPS (1 μg/ml) stimulation for 24 h.

CD4⁺ T cells were isolated by negative selection using a mouse CD4⁺ T cell isolation kit (Miltenyi Biotec). Cells were cultured in 0.1 ml medium in 96-well plates for the indicated times. To assay de novo cell proliferation, 0.5 μCi of [³H]thymidine (Tdr, GE Healthcare BioSciences) was added to each well 6 to 8 h before termination of the cultures by freezing. Plates were then thawed, harvested, and incorporated [³H]Tdr was counted using a Wallac Microbeta scintillation counter (PerkinElmer).

NO was detected using a modified Griess reagent (Sigma-Aldrich). In this method, all NO₃ is converted into NO₂ by nitrate reductase, and the total amount of NO₂ is detected as a colored azo dye product of the Griess reaction.

For cell surface Ag staining, DC were blocked with anti-CD16/CD32 and stained for CD40 (1C10), CD80 (16-10A1), CD86 (GL1), and MHC class II (M5/114.15.2) by direct immunofluorescence. Cells were evaluated by flow cytometry, using a FACScan instrument and CellQuest software for acquisition and analysis (BD Immunocytometry Systems). For the detection of IFN-γR1 expression, DC were first stained with biotin-conjugated anti-mouse IFN-γR1, which was in turn detected by PE-conjugated streptavidin. CD4⁺Foxp3⁺ regulatory T cells were tested by a mouse regulatory T cell staining kit (eBioscience).

Cytokines in the culture supernatant were assayed by a Luminex microbead-based multiplexed assay using commercially available kits according to the manufacturer’s protocol (BioPlex, Bio-Rad).

Mature DC or DC_ap (1 × 10⁶ in 50 μl PBS) preincubated with 50 μg/ml OVA for 6 h were injected into the right footpad of mice. After 7 days, draining popliteal lymph nodes were isolated and single-cell suspensions were prepared. Cell activation was determined by culturing the cells (1 × 10⁶ cells/ml) with 10 μg/ml OVA, and measuring proliferation after 48 h by [³H]thymidine incorporation.

RNA was isolated from cell pellets using an RNeasy mini kit. First-strand cDNA synthesis was performed using the Sensiscript RT Kit with random hexamer primers (all kits from Qiagen). mRNA were quantitated by real-time PCR (MX-4000 from Stratagene) using SYBR Green Master Mix (Applied Biosystems). The amount of mRNA was normalized to endogenous β-actin mRNA. Primer sequences for STAT1 gene were: forward, 5′-CTCCTTCTGGCCTTGGATTG-3′; reverse, 5′-GCCCATAATGCACCCATCATT-3′.

Differences in treatment groups were assessed by two-tailed Student’s t test or by ANOVA for repeated measures, as specified. Statistical significance was determined as p values: ∗, p < 0.05; ∗∗, p < 0.01; and ∗∗∗, p < 0.001.

Previous studies in various systems have shown that apoptotic cells can induce immune tolerance (3, 5, 9). We have reported that in vivo administration of apoptotic cells prolongs engraftment of allogeneic heart transplants (6). To investigate the molecular mechanisms controlling this immunosuppression, we examined the effect of apoptotic cells on DC, APCs that are critical for the activation of resting T cells. DC were derived from mouse bone marrow and exposed to syngeneic apoptotic splenocytes (DC_ap) in vitro. Both normal DC and DC_ap were matured by LPS stimulation and then used for the following experiments.

The DC_ap, normal mature DC, or a mixture of both DC types, were incubated with OVA in vitro to load OVA peptide and then injected into mouse hind footpads. In this in vivo model, the DC would migrate from the site of injection to the draining lymph nodes (dLNs) and cause swelling (28). Seven days after injection, lymphocytes in the popliteal dLNs were isolated and restimulated with OVA ex vivo. As shown in Fig. 1,A, mice immunized with OVA-loaded normal DC showed a dramatic increase in cellularity in the dLNs, as compared to a PBS control, while those receiving DC_ap-presented OVA exhibited much smaller increases. Importantly, when normal DC and DC_ap were injected together, the increment in dLN cellularity was significantly less than with normal DC alone, indicating that DC_ap inhibited stimulation by normal DC. To directly assess the effect of apoptotic cells on T cell stimulation by DC, we examined the response of dLN lymphocytes to OVA stimulation in vitro. Equal numbers of dLN cells were stimulated with OVA, and proliferation and cytokine production determined 48 h later. As expected, dLN cells from mice stimulated with OVA-loaded normal DC exhibited vigorous proliferation and high levels of several cytokines. In contrast, dLN cells from mice primed with DC_ap or with mixed DC_ap and normal DC showed much less Ag-specific proliferation (Fig. 1,B). They also produced less of the Th1 cytokines, IFN-γ and IL-2, as well as the Th2 cytokine, IL-4 (Fig. 1 C). Cytokines could not be detected from the lymphocytes isolated from PBS treated mice (data not shown). Thus, DC exposed to apoptotic splenocytes exhibited an immunosuppressive function both in vivo and ex vivo.

Our in vivo data clearly demonstrate that apoptotic cells dramatically affect a specific Ag-induced immune response through their interaction with DC. Therefore, we explored the changes that occur in DC after exposure to apoptotic lymphocytes. We first examined the expression of several costimulatory molecules. As shown in Fig. 2,A, after stimulation with LPS, DC_ap showed a significantly less mature phenotype compared with normal DC: surface expression levels of CD40, CD80, CD86, and MHC-II were all down-regulated. We also examined the function of DC_ap by stimulating them with LPS and measuring proinflammatory cytokine production. We found that DC_ap produced less proinflammatory cytokines, such as TNF-α, IL-1α, and IL-12, but more IL-10, than do normal DC (Fig. 2 B). Taken together, these results show that coculture with apoptotic cells, causes mature DC to exhibit significant functional and phenotypic changes.

We next investigated the capability of DC_ap to support anti-CD3-stimulated proliferation of splenic CD4⁺ T cells in vitro. To our surprise, DC_ap were almost equally as efficient as normal DC in promoting T cell proliferation at a 1:5 ratio of DC to T cells (Fig. 3,A), a big discrepancy from our ex vivo results (Fig. 1). Examination of the induction of T cell expression of cytokines by DC_ap revealed that, while IL-2 was still produced at the same level as that induced by normal DC, IFN-γ production was not induced by DC_ap (Fig. 3,B). Surprisingly, in the presence of both DC_ap and normal DC (DC_ap:DC = 1:1), T cell proliferation was almost completely inhibited (Fig. 3,A). This strongly suggests that an interaction between normal DC and DC_ap leads to the suppression of T cell responses induced by either type of DC alone. It is important to point out that the inhibition is very strong. Significant inhibition could be observed at a total DC: T cell ratio of 1:10 and 80% inhibition was achieved at 1:5 (Fig. 3,C). We also found that maximal inhibition was at a DC_ap:DC ratio of 1:1 (Fig. 3 D). Therefore, cells were cocultured at the ratios of 1:5 total DC to T cells and 1:1 DC to DC_ap to dissect the suppressive mechanism in the following experiments.

To identify the mechanisms by which the interaction between DC_ap and normal DC leads to such potent immunosuppression, we treated DC_ap with the supernatant from a culture of CD4⁺ T cells stimulated with normal DC, and vice-versa. The T cell-stimulatory capacity of these DC was then assayed by adding them to cultures of fresh CD4⁺ T cells and determining T cell proliferation in response to soluble anti-CD3. We found that DC_ap exposed to supernatant from the coculture of CD4⁺ T cells with normal DC largely lost their ability to subsequently activate T cells. In contrast, supernatant from DC_ap-activated CD4⁺ T cells had little effect on the stimulatory ability of normal DC (Fig. 4 A). These results suggest that CD4⁺ T cells activated by normal DC secrete some factor(s) responsible for the induction of the immunosuppressive function of DC_ap.

To probe the responsible factor(s) involved in immunosuppression by the combination of normal DC and DC_ap, we added various blocking Abs and inhibitors during the stimulation of CD4⁺ T cells and observed the effect on proliferation. First, the involvement of several costimulatory molecules was excluded by adding neutralizing Abs against CD80, CD86, CD40L (data not shown), CTLA-4, and BTLA-4 (Fig. 4,B); none of these had an effect. We then tested the effects of anti-IL-10, anti-TGF-β, and indomethacin (a COX-2 inhibitor that blocks PGE₂ production) but again found no effect (Fig. 4,B). CD4⁺Foxp3⁺ regulatory T cells were also found to have no role in this process (Fig. 4,C). Inhibition of inducible NO synthase (iNOS) by the arginine analog, NG-monomethyl-l-arginine, however, largely reversed the inhibition of T cell proliferation by the combination of DC_ap and normal DC (Fig. 4 B). Other iNOS inhibitors, 1400W and l-NAME, also showed the same effect (data not shown), indicating that iNOS may be involved in the immunosuppression by DC_ap.

NO has been shown to suppress immune responses at high concentrations. In cells of the hematopoietic system, NO is generated exclusively by iNOS. Because IFN-γ has been shown to be one of the best inducers of iNOS expression, we next determined the role of IFN-γ in the induction of NO production by DC_ap. Although we have shown that DC_ap are inefficient in inducing IFN-γ production by T cells (Fig. 3,B), normal DC are quite effective. Accordingly, we added neutralizing Ab against IFN-γ during the activation of CD4⁺ T cells with anti-CD3 by normal DC and DC_ap, and found complete reversal of the inhibition of T proliferation (Fig. 5 A). Therefore, it is possible that IFN-γ-induced NO production is responsible for the immunosuppression induced by the combination of normal DC and DC_ap.

As shown in Fig. 3, exposure to apoptotic cells down-regulated the ability of DC_ap to stimulated T cells to produce IFN-γ. However, our data also show that IFN-γ is required for the immunosuppression induced by the DC + DC_ap coculture. It is thus possible that normal DC support T cells to produce large amount of IFN-γ which in turn induces NO production by DC_ap cells. To test this possibility, we directly compared the differential responses of normal DC and DC_ap to exogenous IFN-γ. Because IL-1 and TNF-α have been reported to be NO inducers, their effects were also determined with and without IFN-γ. Normal DC or DC_ap were incubated with IFN-γ with or without IL-1 and TNF-α for 24 h and NO production was assayed. Surprisingly, IFN-γ alone induced dramatic NO production by DC_ap, but only a low level in normal DC (Fig. 5 B). The inclusion of IL-1 or TNF-α had no significant effect. It should be pointed out that high levels of NO are required to suppress T cell proliferation, while DC normally produce only low levels.

To further examine the role of IFN-γ-induced NO, normal DC and DC_ap were generated from mice deficient in iNOS (iNOS^−/−) or IFN-γ receptor 1 (IFN-γR1^−/−). Consistent with our prediction, after stimulation with IFN-γ, DC_ap derived from either iNOS^−/− or IFN-γR1^−/− mice produced little NO (Fig. 5 C).

To verify that NO is indeed produced by DC_ap in response to IFN-γ in the mixed coculture of DC and DC_ap, we mixed normal DC and DC_ap derived from iNOS-deficient, IFN-γR1-deficient, or wild-type mice. We found that both types of DC derived from all mice were capable of inducing T cell activation. When DC_ap from iNOS^−/− or IFN-γR1^−/− mice were cocultured with wild-type normal DC, immunosuppression was not observed (Fig. 6,A), indicating that IFN-γ receptor-mediated signaling and NO production in DC_ap are the key to immunosuppression. However, when the respective deficient normal DC were mixed with wild-type DC_ap, immunosuppression still occurred (Fig. 6,B) because normal DC from iNOS^−/− or IFN-γR1^−/− mice did not lose the ability to support T cells to produce IFN-γ (data not shown). We also excluded the possibility that IFN-γ produced from T cells by DC_ap at early time points could lead to NO production at later time points. By analyzing IFN-γ production, NO release, and T cell proliferation on day 1, 2, and 3, we found that, DC_ap did not support the production of large amounts of IFN-γ by T cells (Fig. 6 C). Thus, these results demonstrate that in the mixed culture of DC and DC_ap, it is the DC_ap that produce NO in response to IFN-γ that is provided mainly by normal DC.

To extend these in vitro results to an in vivo system, we tested the effect of DC_ap on an Ag-specific immune response. DC or DC_ap from wild-type, IFN-γR1^−/− and iNOS ^−/− mice were injected into the footpads of wild-type mice, using the same protocol as in Fig. 1. Unlike DC_ap derived from wild-type mice, DC_ap derived from IFN-γR1^−/− or iNOS^−/− mice were capable of activating dLN lymphocytes to the same extent as normal DC, as shown by the effects on cell number (Fig. 7,A), Ag-specific lymphocyte proliferation (Fig. 7,B) and T cell cytokine production (Fig. 7 C). These results indicate a critical role for IFN-γ and NO in immunosuppression by apoptotic cell-educated DC in vivo.

The greatest difference between DCs that have or have not been exposed to apoptotic cells lies in their distinct responses to IFN-γ. Therefore, we next investigated the molecular mechanism of these responses. Because it has been reported that some microbes can alter the IFN-γ-response by modulation of IFN-γR1 expression (29), we examined whether DC_ap express greater levels of IFN-γR1. We found no difference in IFNgR1 expression between mature DC_ap and mature DC (Fig. 8,A). We also found that addition of exogenous IFN-γ did not alter the expression pattern of IFN-γR1 (data not shown). We then compared the expression of STAT1 in IFN-γ-treated DC and DC_ap, since STAT1 signaling is the most prevalent pathway of IFN-γ-induced NO production (30, 31). We found that STAT1 induction after treatment with IFN-γ was significantly higher in DC_ap than in normal DC, which may explain the greater IFN-γ response of DC_ap cells (Fig. 8 B). Further molecular investigations should elucidate the mechanism by which distinct IFN-γ signaling occurs in DC_ap cells.

In conclusion, our study reveals a novel mechanism by which apoptotic cells mediate immunosuppression. Upon exposure to apoptotic cells, DC lose their capacity to induce IFN-γ production by T cells, while at the same time they become capable of producing large amounts of NO in response to IFN-γ. With IFN-γ provided by T cells stimulated by other normal DC, the high levels of NO produced by DC_ap can suppress T cell responsiveness, leading to immunosuppression.

Cellular apoptosis occurs throughout the lives of higher metazoan animals (32). Although the extent of apoptosis in different tissues varies depending on the tissue type, age, and physiological and ambient conditions, it has been shown in mice and humans that deficiency in the removal of apoptotic cells can lead to autoimmune disorders such as systemic lupus erythematosus (33). In addition, apoptotic cells by themselves can actually lead to immunosuppression (3, 18, 20, 34, 35, 36, 37, 38). It was reported that decreased phagocytosis of apoptotic cells contributes significantly to the development of systemic lupus erythematosus in mice and humans (39). Thus, removal of the remnants of apoptotic cells by phagocytes not only scavenges the degradation products of apoptotic cells, but can also result in neutral responses from phagocytes or active suppression of inflammation through production of anti-inflammatory mediators. It has also been reported that apoptotic cells trigger tolerogenic responses in the adaptive immune system. Apoptotic cells loaded with exogenous Ag induce specific tolerance to the Ag when injected i.v. into mice (40, 41). We have reported that apoptotic cell administration also prevents the rejection of allogeneic heart transplants (6). Various immunosuppressive factors such as IL-10, TGF-β, and PGE₂ have been proposed to be responsible for these effects (42). In this study, we examined the effects of apoptotic cells on DC and found that, although DC_ap showed little decrement in their ability to activate T cells, they were strongly immunosuppressive when combined with normal DC. Our further analysis revealed that DC exposed to apoptotic cells lost their ability to induce IFN-γ production by T cells, while enabling DC to be highly responsive to IFN-γ by producing large amounts of NO. It is this NO that is responsible for T cell suppression in this system. We also examined IL-10, TGFβ, and PGE₂ and found no role for these factors. Therefore, we report a novel mechanism through which apoptotic cells affect immune responses.

Several studies have found that apoptotic cells exhibit a dramatic immunosuppressive effect in vivo (6, 43, 44). Our studies with heart transplantation showed that a single injection of apoptotic cells could extend successful engraftment for >40 days on the average. When tested in vitro, however, we found that apoptotic cell-pulsed DC lost little of their capacity to activate T cells, in terms of T cell proliferation and IL-2 production, except for decreased secretion of IFN-γ. In fact, most studies of immunosuppression by apoptotic cells have been demonstrated using in vivo systems; few have been reported in vitro. Considering the dramatic immunosuppressive effects of apoptotic cells in vivo, we reasoned that it is highly probable that an interaction exists between apoptotic cell-primed phagocytes and other cells. By combining apoptotic cell-treated DC with normal DC, we found that the copresence of both cell types significantly inhibits T cell activation. This novel finding has significant ramifications for our understanding of the critical role of apoptotic cells in the maintenance of immune homeostasis, autoimmunity and cancer immunology, especially in pathophysiological processes with significant apoptosis.

This study puts NO at the center of the immunosuppression mediated by apoptotic cells. This novel conclusion was based on studies with iNOS inhibitors and iNOS-deficient mice. In fact, it has been shown that iNOS^−/− mice are more responsive to immunological challenge and are more prone to development of autoimmune diseases (45). Although NO is known to be very labile and can act only at short range, it is astonishing that such strong immunosuppressive effects can be achieved by this unstable gaseous molecule. Interestingly, we found that IFN-γ also dramatically up-regulated the expression of several T cell chemokines, more so in DC_ap than in DC. Blockade of the chemokine receptors, CXCR3 and CCR5, reduced the immunosuppressive effect of DC_ap (data not shown). Therefore, upon stimulation by IFN-γ produced by T cells that have been activated by normal DC, in turn DC_ap produce NO and chemokines. It is presumed that the latter attracts T cells to the near vicinity of DC_ap, where the locally high concentrations of NO suppress T cell responsiveness.

The removal of apoptotic cells by phagocytes involves a complex array of still poorly defined molecules. Besides those on the cell surface, soluble factors that link apoptotic cells and phagocytes are also involved. Although the detailed molecular mechanisms are unknown, our data shows that apoptotic cells induce two major changes in DC cells: DC_ap lose their capability to stimulate T cells to produce IFN-γ and become highly responsive to IFN-γ to produce high levels of NO. This is a novel mechanism by which apoptotic cells induce immunosuppression. Because several molecules have been reported to be involved in the phagocytosis of apoptotic cells, more studies are needed to understand which are involved in these changes in DC_ap. In addition, further studies of recently identified molecules such as calreticulin, which has been shown to be critical in converting apoptotic cells from tolerogenic to immunogenic, may shed new light on our understanding of autoimmunity and lead to novel tactics for cancer immunotherapy.

We express gratitude to Drs. Douglas R. Green, Arnold Rabson, and David I. Denhardt for comments on this manuscript, and for support from the Chinese Academy of Sciences.

The authors have no financial conflict of interest.