Apoptotic Cell-Mediated Immunoregulation of Dendritic Cells Does Not Require iC3b Opsonization¹

Dying cells need to be carefully handled by the immune system so as not to provoke an immune response against self-Ags. Apoptosis, the programmed mode of cell death, is carefully orchestrated to be an anti-inflammatory event. Both the apoptotic cells themselves (1), as well as the cells clearing away the apoptotic debris, contribute to this anti-inflammatory environment (2). This anti-inflammatory environment comes in the form of functional suppression of APCs via inhibition of proinflammatory cytokine production. This suppression prevents the immune system from initiating a response against the multiple neo-self-Ags created during the apoptosis process. In the archetypal autoimmune disease systemic lupus erythematosus (SLE),³ an impaired clearance of apoptotic cells, or their failure to induce tolerance to self Ags, are hypothesized as a pathogenic mechanism (3).

There are many receptors for apoptotic cells, found in various combinations depending on cell type, which may provide a specific way for phagocytes to recognize dying cells and induce the appropriate anti-inflammatory environment. The α_mβ₂ integrin complement receptor 3 (CR3) recognizes apoptotic cells by binding to the serum protein iC3b bound to their surface. Opsonization of iC3b is thought to be driven in part by the specific exposure of phosphatidylserine molecules by apoptotic cells (4).

Three recent genome-wide association studies have found that a particular polymorphism of the gene encoding the α-chain of CR3, ITGAM, is associated with high risk to develop SLE (5, 6, 7). The molecular and functional mechanisms underlying this risk factor still remain to be discovered.

We and others have shown that CR3 ligation suppresses inflammatory cytokine production by dendritic cells (DCs). However, in each of these experimental systems, the ligand used to produce the CR3 effect has been artificial, either using a mAb against the α-chain (8, 9), or RBC that have been opsonized with iC3b (10). Although apoptotic cells have certainly been demonstrated to suppress DC inflammatory cytokine production (11), the physiological contribution of CR3/iC3b interactions toward this effect remains in question. Furthermore, there are multiple conflicting reports on the ability of CR3/iC3b interactions to enhance the phagocytosis and internalization of apoptotic cells (10, 12, 13, 14). Because apoptosis is an asynchronous event, it is difficult to generate homogenous populations of apoptotic cells. Using an IL-3-dependent cell line, we are able to generate a relatively homogenous population of apoptotic cells that can be opsonized with serum proteins. Using this system, we show that although apoptotic cell coculture results in cytokine suppression, this effect does not require CR3/iC3b interactions. We furthermore present data that CR3/iC3b interactions do not contribute to the amount or kinetics of apoptotic cell phagocytosis. Understanding the physiologic role of CR3 in apoptotic cell processing will help us to evaluate its potential pharmacologic utility.

C57BL/6, C57BL/6-Rag1^−/−, and ITGAM^−/− (deficient in CR3) mice were purchased from Jackson ImmunoResearch Laboratories. C3^−/− mice were generously provided by Dr. Wenchao Song (University of Pennsylvania, Philadelphia, PA). All mice were bred and maintained in accordance with guidelines of the Institutional Animal Care and Use Committee of The Children’s Hospital of Philadelphia, an American Association for the Accreditation of Laboratory Animal Care accredited facility. Anti-TNF-α, IL-12, IL-10, and CR3 Abs were purchased from BD Biosciences. Anti-iC3b mAb (clone 2/11. rat IgG1) was purchased from HyCult Biotechnology and FITC-anti-Rat IgG1 (BD Biosciences) was used as a secondary Ab. IL-10 receptor blocking Ab 1B1.3A was used at 10 μg/ml.

Bone marrow-derived DCs were generated as previously described (15). In brief, bone marrow precursors from Rag1^−/− mice were cultured for 6 days in complete IMDM containing 3.3 ng/ml GM-CSF (BD Biosciences). Generating DCs from RAG-KO BM does not require depletion of T- and B-cells and RAG-KO BM-DCs behave identically as those from normal mice. To generate BMDCs from C3^−/−, CR3^−/− mice and littermate controls, bone marrow precursors were depleted of lymphocytes using anti-Thy1.2 and anti-B220 directly conjugated magnetic beads.

Mice were injected i.p. with 2 ml of 3% aged Brewer’s thioglycollate. Three days after injection, mice were sacrificed and cells collected from peritoneal lavage. Peritoneal cells were placed in plastic tissue culture dishes and nonadherent cells were removed after 1 h. The remaining cells were >95% macrophages as assessed by surface marker phenotype.

Thymocytes were prepared from 6- to 8-wk-old mice and immediately irradiated with 500 rads of gamma irradiation to induce apoptosis. Irradiated thymocytes were cultured in RPMI 1640 medium with 10% FCS for 16 h before use. BaF3 cells were grown in RPMI 1640 with 10% FCS supplemented with 1 ng/ml of murine IL-3 (PeproTech). BaF3 cells were rendered apoptotic by 48 h of culture in medium lacking IL-3. Both types of apoptotic cells were incubated for 30 min at 37°C with normal C57BL/6 mouse serum or serum from C3^−/− mice (gift of Dr. Wenchao Song, University of Pennsylvania) to achieve serum opsonization. Apoptotic cells were used immediately after opsonization. For experiments in which tracking of the apoptotic cells was necessary, BaF3 cells were incubated with 12.5 μg/ml of the vital dye TAMRA for 30 min at 37°C before apoptosis induction.

BMDCs were cocultured with apoptotic GFP-BaF3 cells for 3 h and then stimulated cells with LPS (100 ng/ml) in the presence of the Golgi transport inhibitor monensin (BD Biosciences) for 5 h. After the 5 h stimulation, BMDCs were harvested in cold PBS and immediately stained for flow cytometric analysis. For phagocytosis experiments, TAMRA-labeled apoptotic BaF3 cells were incubated with BMDCs for the time points shown. BMDCs were then harvested and stained with anti-CD11c Ab conjugated to Alexa 647 (Invitrogen). These cells were then analyzed by flow cytometry, and percent of phagocytosis was calculated as CD11c⁺TAMRA⁺ events/total CD11c⁺ events.

BMDCs were washed in cold PBS, incubated with rat anti-mouse CD16/CD32 (clone 2.4G2) mAb for 10 min to block FcγR, and then stained for 30 min with protein-G purified Ab from supernatants from the N418 hybridoma (anti-CD11c, American Type Culture Collection) conjugated to Alexa 647 (Invitrogen). We conducted all stainings on ice. Intracellular staining for TNF-α and IL-12 was performed using the Cytofix/Cytoperm kit (BD Biosciences) according to the manufacturer’s instructions. Apoptotic cells were stained with an anti-iC3b Ab for 30 min, followed by washing and incubation with a FITC-conjugated secondary Ab. Nonspecific staining was determined by incubating apoptotic cells with the secondary Ab only. Cells were collected on a FACSCanto flow cytometer (BD Biosciences) and the data were analyzed using FlowJo software (Tree Star).

We examined BMDCs under an Axioplan-2 microscope with a ×63 oil objective and we captured and analyzed digital images including 3-D reconstructions using SlideBook software.

Two-tailed Student’s t test or ANOVA testing was performed where appropriate to determine the statistical significance of our findings. The p values of <0.05 were considered to represent statistical significance.

It has been previously reported that iC3b opsonization is predominately a late apoptotic cell event in human cells (16). We performed experiments to examine at which stage iC3b was deposited on murine apoptotic cells. Gamma irradiation of thymocytes followed by 14 h of culture results in a mixed population of apoptotic and necrotic cells as identified by Annexin V/7-AAD staining (Fig. 1,A). Following the convention described by Gaipl et al. (16), we divided cells into live (Annexin V^neg, 7-AAD^neg), early apoptotic (Annexin V^pos, 7-AAD^neg), late apoptotic (Annexin V^pos, 7- AAD^lo,subG1), and necrotic (Annexin V^pos, 7-AAD^high). The forward (FSC) and side scatter profiles of these populations also reflected these designations (data not shown). We incubated apoptotic thymocytes in mouse serum for 30 min at 37°C after which we stained with an Ab specific for mouse iC3b and we detected iC3b deposition by flow cytometry on both early and late apoptotic cells, while we found no iC3b bound to necrotic or live cells (Fig. 1,B). No iC3b deposition was detected on cells that were not incubated in serum, or on cells incubated in serum from C3^−/− mice, confirming the specificity of the Ab used. Late apoptotic cells were uniformly positive, while early apoptotic cells showed a bimodal distribution of iC3b deposition. Within this early Annexin V^pos, 7-AAD^neg population, the iC3b-negative cells had a FSC × side scatter profile more consistent with live cells, while the iC3b-positive population had a lower FSC, consistent with further progression in apoptosis (Fig. 1 C). This suggests that iC3b deposition is an event occurring during the development of early apoptosis, but not at its initial onset.

Because we desired a more homogenous apoptotic cell population for our experiments, we tested multiple models of apoptotic cell death. IL-3 deprivation of the IL-3 dependent BaF3 cells (17) for 48 h resulted in the most homogenous cell death, with ∼70% of cells in the early apoptotic stage (Fig. 1,D). Incubation with serum resulted in iC3b deposition on both early and late apoptotic BaF3 cells (Fig. 1 E). Neither live nor necrotic BaF3 cells fixed iC3b to their surface.

Thus, we find in two different models of murine apoptotic cell death that iC3b deposition occurs both in early and late apoptotic cells, while it does not occur in necrotic cells.

Apoptotic cells can suppress the LPS-stimulated production of proinflammatory cytokines by DCs. We and others (8, 9, 10) have shown that CR3 activation can have the same effect. However, in all of these reports, CR3 activation was obtained with artificial ligands such as mAbs or iC3b coated RBC. Therefore, the physiologic contribution of CR3 to the inhibition of the proinflammatory cytokines induced by apoptotic cells still remains to be determined. To do so, we used BaF3 cells that were transduced with the GFP protein (GFP-BaF3) to generate apoptotic cells that could be easily differentiated from BMDCs. We cocultured BMDCs for 3 h with apoptotic GFP-BaF3 cells that had been previously incubated with no serum, wild type serum, or C3^−/− serum for 30 min. We then stimulated these BMDCs with LPS in the presence of monensin for 5 h and analyzed for cytokine production by intracellular staining and flow cytometry. Because GFP-BaF3 cells lost their GFP positivity as they became apoptotic, likely due to proteolysis of the GFP during apoptosis (data not shown), we used the GFP label to gate early apoptotic BaF3 cells out and identified BMDCs as CD11c⁺GFP⁻ cells within the appropriate forward and side scatter gate for DCs. We found that apoptotic cells inhibited both LPS-induced TNF-α and IL-12 expression by DCs as measured by median fluorescence intensity (MdFI) (Fig. 2), confirming previous reports that apoptotic cells inhibit LPS-induced up-regulation of proinflammatory cytokines in human (18) and murine systems (11). Apoptotic cells exposed to serum in general tended to have a slightly more pronounced effect on the overall MdFI of both TNF-α and IL-12 that did not reach statistical significance, while the presence or absence of C3 in the serum (and thus iC3b on the surface of the apoptotic cells) did not make a difference.

The technique of intracellular staining provides the advantage of detection of cytokines at a single cell level and we discovered that BMDCs do not activate homogeneously. Four populations of BMDCs were present after LPS stimulation, cells that produced no cytokines, cells that were single positive for either TNF-α or IL-12, and cells that were positive for both cytokines (Fig. 3,A, top left panel). Apoptotic cells reduced the IL-12 single-positive and double-positive BMDCs while leaving the TNF-α single positive population unaffected (Fig. 3 B). The addition of serum did not have a significant effect on the extent of inhibition, suggesting that this was a CR3/iC3b independent event.

It remained possible that the lack of a serum effect was specific only to high dose LPS stimulation. We performed similar experiments decreasing the dose of LPS over a three log range (1 ng/ml to 100 ng/ml) and found no difference in the ability of serum opsonized apoptotic cells to inhibit cytokines over nonopsonized apoptotic cells (data not shown). These data demonstrate that apoptotic cells reduce the IL-12 single-positive and double-positive BMDCs while leaving the TNF-α single positive population unaffected upon TLR stimulation. We therefore hypothesize that the effect of the apoptotic cells is not unique to a particular cytokine, but rather targets distinct functional subsets of BMDCs.

We have previously shown that CR3 ligation with the specific mAb decreases the cytokine production induced by the TLR9 ligand CpG (8). We have now found that apoptotic cells can also inhibit the CpG-induced production of proinflammatory cytokines in DCs. Serum opsonization did not affect the inhibition of cytokines caused by apoptotic cells in CpG stimulated DCs, demonstrating serum independence for multiple TLR ligands (Fig. 3 C). Interestingly, in this case, apoptotic cells inhibited not only the IL-12 single positive and double positive DCs, but also the TNF-α single-positive cells. Thus, apoptotic cells may attenuate different DC populations depending on which TLR agonist follows them.

Because DCs are capable of local C3 production (19), we hypothesized that lack of C3 in the C3^−/− serum could have been compensated for by BMDC-derived C3 secretion. This would account for the lack of differences in cytokine suppression seen between the wild type and the C3^−/− serum. We cocultured apoptotic TAMRA labeled BaF3 cells with C3^+/+ and C3^−/− BMDCs for 3 h. We identified the apoptotic cells as CD11c⁻TAMRA⁺ events and determined the surface levels of iC3b by the MdFI within this population. No differences were seen in iC3b levels in the apoptotic cells exposed to wild type and C3^−/− BMDC, suggesting that local C3 production by BMDCs is not playing a role in iC3b opsonization of apoptotic cells in our culture system (Fig. 4,A). Furthermore, no differences in the pattern of inhibition of cytokines were seen in BMDC sufficient or deficient in C3 exposed to apoptotic BaF3 cells, confirming that local C3 production is not playing a role in the cytokine inhibition caused by apoptotic cells (Fig. 4 B).

It has been previously shown that apoptotic cells inhibit the production of proinflammatory cytokines induced by LPS in macrophages (20). We tested whether the opsonization with mouse serum, deficient or not of C3, affected this inhibition. Peritoneal macrophages responded more homogeneously to activation than BMDCs, because the large majority produced TNF-α upon LPS stimulation (Fig. 5); only a small percentage of them (5%) produced IL-12, confirming previous reports that macrophages do not have the same potential to produce IL-12 as DCs (21). Apoptotic cells completely obliterated IL-12 production in macrophages, and they also induced a small reduction of percentage of TNF-α-producing macrophages. Serum opsonization did not alter macrophage LPS induced cytokines (Fig. 5). These results show that serum opsonization is also dispensable for apoptotic cell mediated inhibition of macrophage cytokine production.

The anti-inflammatory cytokine IL-10 is secreted by phagocytes upon the ingestion of apoptotic cells (22). In our hands, the exposure to apoptotic cells caused an increase in IL-10 production in both BMDCs and peritoneal macrophages, but this effect was not enhanced by serum opsonins (Figs. 2 and 6, A and B). Although we have shown that ligation of CR3 via mAb actually inhibits IL-10 production (8), no difference was seen in IL-10 levels when wild type or C3^−/− serum was used. When we analyzed BMDCs for expression of IL-10 with respect to IL-12 and TNF-α, surprisingly we found that IL-10 was produced equally by all four populations of BMDCs (Fig. 6,A). These results suggest that the capacity to produce IL-10 is independent of the state of production of inflammatory cytokines. It is noteworthy that IL-10 was induced specifically by the apoptotic cells, because at this early time point (5 h) LPS did not yet induce any IL-10, while apoptotic cells without LPS resulted in IL-10 production (Fig. 6 C). These data are consistent with the previously described kinetics of cytokine production in DCs (23).

Opsonization with C3^+/+ or C3^−/− serum did not affect the induction of IL-10 by apoptotic cells, either in BMDCs or in peritoneal macrophages (Fig. 6, A and B). These results indicate that in the murine system, apoptotic cells induce IL-10 production both in dendritic cells and macrophages through a mechanism independent from serum opsonins.

Importantly, the IL-10 induced by apoptotic cells was not responsible for the inhibition of TNF-α and IL-12 because pretreatment with an IL-10 receptor blocking Ab did not affect TNF-α/IL-12 inhibition induced by apoptotic cells with or without iC3b opsonization (Fig. 6 D). The efficacy of IL-10 receptor blockade is demonstrated by the increase in baseline LPS induced TNF-α/IL-12 production in the IL-10 receptor blocked DCs compared with isotype Ab-treated cells.

Because there are conflicting reports in the literature about the role of serum opsonins and iC3b especially in facilitating phagocytosis of apoptotic cells, we further investigated whether serum opsonization of apoptotic cell enhances the extent or kinetics of phagocytosis in mouse macrophages and DCs. We incubated apoptotic TAMRA labeled BaF3 cells in serum as described above. We cocultured apoptotic cells with either BMDCs or peritoneal macrophages and measured phagocytosis at various time points (Fig. 7). No differences were seen in either the amount or the rate of phagocytosis when we used serum opsonized apoptotic cells or apoptotic cells without serum. Furthermore, phagocytes from mice deficient in CR3 (integrin α_m^−/−) showed no defect in phagocytosis (data not shown). These results are in agreement with Ren et al. (36) suggesting that complement opsonization of apoptotic cells does not enhance phagocytosis in otherwise normal phagocytes.

Given the lack of effect of iC3b opsonization on cytokine production and phagocytosis, we confirmed that iC3b on the surface of apoptotic cells does interact with CR3 on dendritic cells. We have previously observed that triggering of CR3 with a specific Ab causes CR3 internalization (8). We used immunofluorescence microscopy and 3-D reconstruction to determine whether iC3b-opsonized apoptotic cells could induce the same CR3 internalization, as sign of CR3/iC3b interaction. To do so, we cocultured TAMRA-labeled apoptotic BaF3 cells, with or without iC3b opsonization, and BMDCs for 1 h. BMDCs were then harvested, permeabilized, and stained for CR3. BMDCs that had not interacted with an apoptotic cell showed a rim, surface staining pattern of CR3 (Fig. 8,A). BMDCs were able to internalize nonopsonized apoptotic cells, however CR3 remained on the surface of the BMDC and did not colocalize with the vesicle containing the apoptotic cell (Fig. 8,B). In contrast, phagocytosis of iC3b opsonized apoptotic cells resulted in the internalization of CR3 from the surface, colocalizing around the vesicle containing the apoptotic cells (Fig. 8 C). Thus, iC3b on apoptotic cells does interact with CR3 and induces its internalization even though it does not enhance phagocytosis or cytokine suppression.

Much attention has recently been given to CR3 activation as an anti-inflammatory event. These data, including our own, have mostly been generated using artificial means of activating CR3 to model the presumed effects of iC3b opsonized apoptotic cells binding to the receptor. Data examining the role of iC3b/CR3 interactions in the phagocytosis of apoptotic cells is conflicting (10, 12, 13, 14). In this report, we provide evidence suggesting that CR3 is not necessary for either apoptotic cell phagocytosis or apoptotic cell mediated DC cytokine suppression. The BaF3 model provides an excellent means to study early, apoptotic cells in a homogenous population, something that is not easily accomplished using other systems we attempted, such as irradiated thymocytes or splenic B cell, and ceramide treated fibroblasts (data not shown). Because it was the early apoptotic population of BaF3 that had the highest levels of iC3b deposition (Fig. 1), this was an ideal model for identifying the optimal iC3b effect. Of course, caution should be applied in generalizing to other apoptotic cell models where more heterogeneous death is seen, and other cell specific factors may be in effect.

The lack of iC3b effect is likely due to the redundancy of the apoptotic cell receptor system, which includes other molecules also known to inhibit inflammatory cytokine production including mer (24) and CD36 (25), as well as many different phagocytic receptors (26, 27, 28). Indeed, C1q is the more important mediator in the in vivo clearance of apoptotic cells compared with C4/C3, particularly in the absence of inflammation (29). It is also possible that CR3 triggering induces unique and as of yet uncharacterized factors important in the anti-inflammatory response that we are simply unable to assay, such as directing the apoptotic cell to a particular endosomal compartment for alternate processing (30). Finally, because of the results of the microscopy experiments, we exclude the possibility that, in our culture system, iC3b opsonization simply does not trigger CR3 on DCs. We speculate that physiologic triggering of CR3 by iC3b opsonized apoptotic cells does not result in the same DC inhibition induced by ligation with a mAb due to differences in either the strength or qualitative nature of the signals.

In vitro studies of serum opsonization have been limited to phagocytosis (10, 12, 13, 14) and offer conflicting reports. Apoptotic cells in serum-free conditions have been recently been shown to be equally effective as apoptotic cells with serum opsonins in their ability to inhibit LPS induced IL-12 (31). Our observation that serum opsonization of apoptotic cells, regardless of the presence or absence of C3, does not affect their in vitro cytokine inhibition or phagocytosis also suggests that serum factors in general are redundant for apoptotic cell processing by DCs.

The heterogeneous nature of cytokine secretion by CD11c⁺ BMDCs was surprising because it suggests that there may be different subsets of DCs with varying capacity to respond to TLR stimulation in a population that otherwise would appear fairly homogenous. These same subsets would appear to also respond differently to apoptotic cell exposure with respect to suppression of inflammatory cytokines because in the context of LPS, TNF-α single producing cells are less affected by apoptotic cells than IL-12 single producers or TNF-α/IL-12 double producers. Interestingly, all the three subpopulations of cytokine producing DCs were inhibited by apoptotic cells upon stimulation with CpGs, suggesting that these functionally different subsets of BMDCs have different sensitivity to apoptotic cells depending on which TLR agonist is used. In contrast, the effect of apoptotic cells on the production of IL-10 was independent of the TNF-α/IL-12 status of the BMDC, suggesting that some apoptotic cell responses occur homogenously within DCs. Further characterization of these DC subsets may reveal new factors that determine TLR responsiveness as well as participate in apoptotic cell processing.

Our data show in two different murine systems of apoptosis that iC3b deposition occurs both on early and late apoptotic cells. This is in contrast to an earlier report in human cells that described iC3b binding as a predominantly late apoptotic event (16). Certainly this may reflect both differences in the organisms (human vs mouse) and cell types (polymorphonuclear cells and peripheral blood lymphocytes vs thymocytes and a leukemic cell line). Because in vivo clearance of apoptotic cells usually occurs before the apoptotic cells reach the late stage (reviewed in Ref. 32), only early deposition of iC3b would allow the binding of iC3b-opsonized apoptotic cells to CR3 receptors in physiologic conditions. Moreover, the absence of iC3b deposition on necrotic cells suggest that this molecule could be partially responsible for the distinct effects that apoptotic vs necrotic cells have on the cells of the immune system (33).

Clearly, CR3 can have a major effect on the maintenance of immune tolerance as evidenced by the strong association of CR3 polymorphisms with the development of SLE (5, 6, 7). Interestingly, there are also many associations of defective apoptotic cell processing and the development of autoimmunity in humans and mice (3, 34, 35, 36). However, given our data, it would appear that the functional absence of CR3 would be unlikely to result in any significant deficits in apoptotic cell processing. Furthermore, in the report of Nath et al. (7), the functional variant of ITGAM associated with SLE is not in the iC3b binding region of CR3, again suggesting that the association of CR3 with SLE may not be dependent on apoptotic cell processing. It is possible that other physiologic ligands of CR3, for example fibronectin (37) or ICAM-1 (38), may also result in cytokine suppression and that these interactions are more physiologically relevant for maintaining immune tolerance than its role as an apoptotic cell receptor. It is also possible that the role of CR3 in SLE is related to its many other roles: adhesion (39), respiratory burst (40), or NO production (41). Further study of CR3 in SLE models may help distinguish these possibilities.

In conclusion, it would appear that CR3 is dispensable in the phagocytosis of apoptotic cells and in the suppression of cytokines that they induce. Nonetheless, our data do not in any way diminish the possibility of using artificial ligation of CR3 as a means of manipulating DC function, because supraphysiologic activation of CR3 can clearly modulate DCs toward an anti-inflammatory phenotype (8, 9, 10). Data already exist to suggest that the activation of CR3 via mAb may be useful in treating psoriasis and inflammatory colitis in mouse models of these diseases (42). However, it is important to put the role of CR3 in a physiologic context with respect to apoptotic cell processing and continuing to study the physiologic roles of CR3 will help us to understand how to best pursue its potential pharmacologic utility.

We thank Debra K. Shivers for her assistance in maintaining the animal colonies and Mark Ma for technical support with the microscopy equipment.

The authors have no financial conflict of interest.