|Home | About | Journals | Submit | Contact Us | Français|
Anaphylaxis is a life-threatening hyperacute immediate hypersensitivity reaction. Classically, it depends on IgE, FcεRI, mast cells, and histamine. However, anaphylaxis can also be induced by IgG antibodies, and an IgG1-induced passive type of systemic anaphylaxis has been reported to depend on basophils. In addition, it was found that neither mast cells nor basophils were required in mouse models of active systemic anaphylaxis. Therefore, we investigated what antibodies, receptors, and cells are involved in active systemic anaphylaxis in mice. We found that IgG antibodies, FcγRIIIA and FcγRIV, platelet-activating factor, neutrophils, and, to a lesser extent, basophils were involved. Neutrophil activation could be monitored in vivo during anaphylaxis. Neutrophil depletion inhibited active, and also passive, systemic anaphylaxis. Importantly, mouse and human neutrophils each restored anaphylaxis in anaphylaxis-resistant mice, demonstrating that neutrophils are sufficient to induce anaphylaxis in mice and suggesting that neutrophils can contribute to anaphylaxis in humans. Our results therefore reveal an unexpected role for IgG, IgG receptors, and neutrophils in anaphylaxis in mice. These molecules and cells could be potential new targets for the development of anaphylaxis therapeutics if the same mechanism is responsible for anaphylaxis in humans.
Anaphylaxis is a systemic hyperacute allergic reaction (1) responsible for more than 1,500 deaths per year in the US (2). Anaphylaxis is associated with intense vasodilatation and bronchoconstriction, severe laryngeal edema, drop of cardiac pressure, and hypothermia. As anaphylaxis is a life-threatening medical emergency, the mechanisms thought to be responsible for anaphylaxis have been mostly investigated in animal models. Two types of models have been developed since the initial description of anaphylaxis in dogs (3): active anaphylaxis, in immunized animals, and passive anaphylaxis, in nonimmunized animals injected with antibodies. Indeed, susceptibility to anaphylaxis can be transferred by serum from immunized donors or by purified antibodies.
IgE-induced passive systemic anaphylaxis (PSA) is elicited by injecting mice systemically with IgE antibodies 24–48 hours before an i.v. challenge with specific antigen. The anaphylactic shock that develops within minutes can be easily assessed by monitoring the decrease in body temperature. IgE-induced PSA observed in WT mice was abrogated in mice deficient for FcεRI, the high-affinity IgE receptor expressed by mast cells and basophils (4), and in mast cell–deficient W/Wv mice (5). It was also abrogated in histidine decarboxylase–deficient mice, which lack histamine (6), and in mice injected with histamine receptor antagonists (7). Anaphylactic symptoms could be induced by an i.v. injection of histamine (6). These findings together demonstrate the mandatory role of mast cells and of FcεRI in IgE-induced PSA, and they emphasize the contribution of histamine, contained in mast cell granules that are rapidly released during exocytosis. This mechanism has been widely accepted as a paradigm of the anaphylactic reaction.
IgG-induced PSA is elicited by injecting mice systemically with IgG antibodies 2–3 hours before an i.v. challenge with specific antigen. Alternatively, preformed IgG immune complexes (IC) can be injected i.v. Similar symptoms develop, with comparable kinetics, during IgE- and IgG-induced PSA. IgG1 is the dominant antibody subclass raised during humoral responses to protein antigens in mice, and passively administered IgG1-IC are sufficient to induce anaphylaxis. Because the low-affinity IgG receptor FcγRIIIA was shown to trigger mast cell activation in vitro (8) and passive cutaneous anaphylaxis in vivo (9), it has been generally accepted that these receptors account for IgG1-induced PSA. No published paper formally demonstrated this assumption, but we confirmed that, indeed, IgG1-induced PSA was abrogated in FcγRIIIA-deficient mice (P. Bruhns and M. Daëron, unpublished observations). Surprisingly, IgG1-induced PSA was not abrogated in mast cell–deficient mice (5), but it was reported to be abrogated in basophil-depleted mice (10). This suggests that mouse basophils express FcγRIIIA. FcγRIIIA are also expressed by other myeloid cells. Upon activation, mouse basophils rapidly release granular mediators, including histamine, but also lipid-derived mediators such as platelet-activating factor (PAF). Like histamine, PAF could, by itself, reproduce the clinical signs of an anaphylactic shock when injected in animals (11). PAF, but not histamine, was shown to be responsible for IgG1-induced PSA (10). These findings together indicate that IgG1-IC trigger anaphylaxis through the release of PAF, probably by aggregating FcγRIIIA on basophils.
Active systemic anaphylaxis (ASA) is elicited by an i.v. injection of antigen into mice immunized with that antigen. Similar symptoms develop with comparable kinetics during ASA and PSA in WT mice. More mice, however, die during ASA than during PSA. Different adjuvants can be used for immunization. It is generally accepted that alum favors IgG1 and IgE antibodies, whereas CFA favors IgG2 antibodies. In both cases, however, IgG1 antibodies are the most abundant and IgE the less abundant. ASA was not affected in Cε-deficient mice, which produce no IgE (12). Antibodies other than IgE are therefore sufficient to induce ASA. Supporting this conclusion, ASA was not altered in FcεRI-deficient mice, but it was abrogated in FcRγ-deficient mice (5), which express no activating receptors for IgE (FcεRI) or for IgG (FcγRI, FcγRIIIA, FcγRIV). Activating FcRs are therefore mandatory for ASA. ASA was not altered in mast cell–deficient Wsh/Wsh or W/Wv mice (5, 13), in basophil-deficient mice (14), or in basophil-depleted WT mice (10). Neither mast cells nor basophils are therefore mandatory for ASA. Furthermore, ASA-induced decrease in body temperature still occurred in the absence of both cell types, but ASA-associated mortality was abolished (10). Basophils and mast cells can therefore contribute to ASA. Reduced anaphylactic shock was observed in PAF receptor–deficient (PAF-R–deficient) mice (15) and in mice injected with PAF-R antagonists (16). These findings together indicate that antibodies other than IgE, activating FcRs other than FcεRI, cell types other than basophils and mast cells, and mediators other than histamine contribute to ASA. These antibodies, FcRs, cells, and mediators are unknown.
We unravel here an unexpected role of neutrophils and an underestimated contribution of IgG and IgG receptors to anaphylaxis. Neutrophils and basophils both contributed to ASA in WT mice. Neutrophils, however, were sufficient for ASA in genetically modified mice expressing no activating IgG or IgE receptors on mast cells and basophils. Importantly, like murine neutrophils, human neutrophils restored ASA in ASA-resistant mice, suggesting that these cells can contribute to anaphylaxis in humans. Neutrophil-dependent ASA was mediated by PAF. Neutrophils contributed also to IgG-induced PSA. Finally, 2 IgG receptors, FcγRIIIA and FcγRIV, accounted for ASA in WT mice.
An i.v. antigen challenge induced a decrease of body temperature in WT or in mast cell–deficient Wsh mice immunized with antigen in CFA/incomplete Freund adjuvant (IFA), but not in FcRγ–/– (officially referred to as Fcer1g–/–) mice, which lack all activating IgG and IgE receptors (data not shown). ASA indeed depends on antibodies. Immunizations in CFA/IFA lead to the production of comparable IgG1 and IgG2 antibody levels in WT and FcRγ–/– mice and, as expected (5), of higher levels of IgE antibodies in FcRγ–/– mice than in WT mice (Supplemental Figure 1A; supplemental material available online with this article; doi: 10.1172/JCI45232DS1). Nonimmunized WT, but not FcRγ–/–, mice underwent anaphylaxis upon challenge with antigen if injected with serum from mice immunized with BSA in CFA/IFA (Supplemental Figure 1B).
ASA was not altered in Fcer1a/Fcer2a–double-deficient (FcεRI/II–/–) or in Fcgr3a-deficient (FcγRIIIA–/–) mice compared with WT mice (Figure (Figure1A).1A). Unexpectedly, ASA was also unaltered in FcγRI/FcγRIIB/FcγRIIIA/FcεRI/FcεRII–/– (5KO) mice (Figure (Figure1B1B and Supplemental Table 1). 5KO mice lack all IgE and IgG receptors except the activating IgG2 receptor FcγRIV (17). Immunizations did not significantly modify the expression of FcγRIV (Supplemental Figure 1C). Anti-FcγRIV blocking mAbs abolished ASA in 5KO mice (Figure (Figure1C).1C). FcγRIV can therefore trigger anaphylaxis.
FcγRIV is expressed by monocytes/macrophages and neutrophils (17, 18), but not by basophils, mast cells, and eosinophils (Figure (Figure1D).1D). Monocytes and macrophages can be depleted by injecting toxic liposomes in vivo (19). Both control and toxic liposomes had the same effect on ASA, reducing ASA-associated mortality without altering ASA-associated temperature drop (Figure (Figure1E).1E). Toxic liposomes, but not control liposomes, however, depleted monocytes (example shown in Supplemental Figure 1D). Depletion of monocytes/macrophages induced by toxic liposomes, therefore, did not alter ASA (Figure (Figure1E).1E). Supporting this result, ASA was unaffected in 5KO mice injected with gadolinium, which inhibits monocyte/macrophage function (Supplemental Figure 1E). Monocyte/macrophages are therefore not mandatory for ASA in 5KO mice. Noticeably, ASA-associated temperature drop and mortality were abrogated by neutrophil depletion in 5KO mice (Figure (Figure1F).1F). Anti-Gr1 injections depleted neutrophils, but did not reduce basophil numbers (Supplemental Figure 2, A and B). ASA was restored within 7 days in neutrophil-depleted mice, and it was correlated with blood neutrophil recovery (Figure (Figure1G).1G). Neutrophils, but not monocytes/macrophages, are therefore required for FcγRIV-dependent ASA. Importantly, purified bone marrow neutrophils from 5KO mice, but not from FcRγ–/– mice, restored ASA-associated temperature drop in BSA-immunized FcRγ–/– recipients (Figure (Figure2A).2A). Neutrophils are therefore necessary and sufficient for FcγRIV-dependent ASA. Mediators released and/or secreted by activated neutrophils should therefore be responsible for the anaphylactic shock observed. Among them, PAF (20–22) was shown to mimic anaphylaxis when injected in mice (11). We detected PAF in the supernatant of neutrophils purified from 5KO, but not from FcRγ–/–, mice stimulated with IgG2b-IC in vitro. Anti-FcγRIV blocking mAbs abrogated PAF production by neutrophils from 5KO mice (Figure (Figure2B).2B). PAF may therefore be responsible for FcγRIV-induced ASA in 5KO mice.
Upon activation, blood neutrophils release granules that contain mediators and enzymes, including myeloperoxidase (MPO). MPO can be released within minutes by activated neutrophils (data not shown and ref. 20) and, under inflammatory conditions, by monocytes/macrophages (23). MPO can oxidize luminol, and oxidized luminol emits luminescence. Luminescence could, indeed, be detected in vivo in WT, but not in Mpo–/–, mice injected with luminol and challenged with PMA or LPS (24). Immunized, but not naive, 5KO mice exhibited a massive, systemic luminescence emission following antigen challenge and luminol injection (Figure (Figure2C2C and Supplemental Figure 2C). Luminescence was detectable within minutes after antigen challenge and lasted for at least 20 minutes (Figure (Figure2D2D and Supplemental Figure 2D). These results suggest that neutrophils are systemically activated within minutes in ASA.
We investigated whether neutrophil depletion or FcγRIV blocking, which both affect ASA, may also affect an IgG-induced PSA in immunodeficient Rag2–/– 5KO mice. These mice, indeed, lack endogenous IgG that might compete with IC for FcγRIV binding. FcγRIV is indeed a high-affinity IgG receptor (17). An i.v. injection of monoclonal IgG2b-IC induced a significant, although nonlethal, temperature drop in Rag2–/– 5KO mice, but not in Rag2–/–FcRγ–/– mice (Figure (Figure3A).3A). FcγRIV blockade or neutrophil depletion abolished monoclonal IgG2b-induced PSA in Rag2–/– 5KO mice (Figure (Figure3B).3B). FcγRIV can therefore induce IgG2b-induced PSA that depends on neutrophils. Immunocompetent 5KO mice also developed IgG2b-induced PSA (data not shown). Because neutrophils from 5KO mice secrete PAF when stimulated in vitro with IgG2b-IC (Figure (Figure2B),2B), we measured PAF in the plasma of 5KO mice undergoing neutrophil-dependent IgG2b-induced PSA. We detected elevated PAF levels in the plasma of challenged, but not in nonchallenged (no detectable PAF; not shown), 5KO mice (Figure (Figure3C).3C). FcγRIV blockade abolished the increase in circulating PAF levels. Elevated PAF levels in plasma therefore correlate with FcγRIV-triggered, neutrophil-dependent PSA. Surprisingly, FcγRIV blockade or neutrophil depletion abolished monoclonal IgG2b-induced PSA in WT mice also (Figure (Figure3D).3D). FcγRIV is therefore responsible for IgG2b-induced neutrophil-dependent PSA in WT mice.
A nonlethal PSA could also be observed in Rag2–/– 5KO mice injected with IC made of GPI and K/BxN serum (GPI/anti-GPI-induced PSA) (Figure (Figure3E),3E), which contains polyclonal IgG1 and IgG2 anti-GPI antibodies (25, 26). Neutrophil depletion abolished this GPI/anti-GPI–induced PSA in Rag2–/– 5KO mice (Figure (Figure3F).3F). As expected, FcγRIV blockade only slightly reduced GPI/anti-GPI–induced PSA in WT mice. Surprisingly, however, the depletion of neutrophils, which express FcγRIV and FcγRIIIA, was sufficient to abolish GPI/anti-GPI–induced PSA in these mice (Figure (Figure3G).3G). Similar results were observed in 5KO mice (data not shown). Whereas basophils have been reported to mediate monoclonal IgG1-induced PSA (10), neutrophil depletion, but not basophil depletion, abolished GPI/anti-GPI–induced PSA in the same experiment (Figure (Figure3H).3H). Similarly, whereas mast cells have been reported to contribute to monoclonal IgG1-induced PSA (5), mast cell–deficient Wsh mice showed unaltered GPI/anti-GPI–induced PSA. Neutrophil depletion abolished GPI/anti-GPI–induced PSA in Wsh mice as efficiently as in WT mice, whereas basophil depletion had no effect (Figure (Figure3I).3I). Together, these results demonstrate that neutrophils, but not mast cells or basophils, are mandatory for polyclonal IgG–induced PSA in nonimmunized WT mice.
We next investigated the contribution of activating Fc receptors to ASA in WT mice. WT mice express 1 activating IgE receptor, FcεRI, and 3 activating IgG receptors, FcγRI, FcγRIIIA, and FcγRIV. No anti-FcεRI– or anti-FcγRI–blocking antibodies are available. The recently described (27) mAb 275003, which is specific for FcγRIIIA (Figure (Figure4A4A and Supplemental Figure 3, A and B), and which can block the binding of IgG-IC to FcγRIIIA in vitro (Supplemental Figure 3C), abolished FcγRIIIA-dependent (9) IgG-induced PSA (Figure (Figure4B).4B). FcγRIIIA blockade also significantly reduced ASA-associated temperature drop and mortality in WT mice (Figure (Figure4,4, C and D). FcγRIV blockade reduced ASA-associated temperature drop similarly to FcγRIIIA blockade, but reduced mortality less efficiently (Figure (Figure4D).4D). Concomitant FcγRIIIA and FcγRIV blockade, however, were necessary to abolish ASA in WT mice (Figure (Figure4,4, D and E). FcγRIIIA and FcγRIV therefore account for ASA in WT mice.
FcγRIIIA and FcγRIV are both expressed by neutrophils, monocytes, and macrophages. FcγRIIIA, but not FcγRIV, is expressed also on basophils, eosinophils, and mast cells (Figure (Figure5A5A and Figure Figure1D).1D). All these cell types can potentially contribute to ASA. Inhibition of monocyte/macrophages (Figure (Figure5B)5B) or depletion of eosinophils (Supplemental Figure 3D) did not affect ASA in WT mice. Importantly, neutrophil depletion reduced temperature drop and prevented ASA-associated death in WT mice (Figure (Figure5C).5C). Basophil depletion (example shown in Supplemental Figure 3E) had no effect unless neutrophils were also depleted, suggesting a dominant role for neutrophils in ASA (Figure (Figure5C).5C). Similar results were obtained in mast cell–deficient Wsh mice (Figure (Figure5D)5D) and in FcγRIIIA–/– mice that express no activating FcγRs on mast cells and basophils (Figure (Figure5E).5E). ASA in WT mice is therefore unaffected by a deficiency in a single cell population, except that of neutrophils.
We observed that the proportion of blood neutrophils significantly increased following immunizations with antigen in CFA/IFA. Neutrophilia is not specific to this adjuvant, as it was also observed following immunization with alum or with alum plus Pertussis toxin (Figure (Figure6A).6A). This observation could possibly explain the dominant contribution of neutrophils to ASA following CFA/IFA immunizations. To test this hypothesis, we delayed the antigenic challenge for several weeks until neutrophil numbers dropped back to baseline. Six weeks after immunization, neutrophil numbers in 5KO mice immunized with BSA in CFA/IFA were comparable to neutrophil numbers in naive mice (Figure (Figure6B),6B), while the levels of anti-BSA antibodies in the serum remained high (Figure (Figure6C).6C). When challenged with BSA, these 5KO mice developed ASA, as expected, and ASA-associated temperature drop and mortality were both dependent on neutrophils (Figure (Figure6D).6D). The dominant role for neutrophils in ASA is therefore not due to higher neutrophil numbers in immunized mice.
Because neutrophils play a dominant role in ASA in 5KO mice and in WT mice, because neutrophils secrete PAF in vitro (Figure (Figure2B),2B), and because PAF concentration is elevated in plasma during neutrophil-dependent PSA (Figure (Figure3C),3C), we analyzed the contribution of PAF to ASA in 5KO and in WT mice. Two PAF-R antagonists (Figure (Figure7A),7A), but not a histamine receptor-1 antagonist (Figure (Figure7B),7B), markedly reduced ASA in 5KO mice. PAF-R antagonists also abolished ASA-associated death and inhibited temperature drop in WT mice, whereas histamine receptor-1 antagonist had a much milder effect (Figure (Figure7C).7C). PAF production requires cytosolic phospholipase A2 (cPLA2) (28–30). ASA was strongly inhibited in cPLA2-deficient mice (Figure (Figure7D).7D). PAF may therefore account for neutrophil-dependent ASA in 5KO mice and contributes to ASA in WT mice.
Like murine neutrophils (Supplemental Figure 4A), human neutrophils express FcRs. The only activating FcRs expressed by neutrophils from normal donors are FcγRIIA (Figure (Figure8A).8A). Human neutrophils, indeed, do not express FcγRI, FcεRI, or FcγRIIIA. They, however, express the GPI-anchored FcγRIIIB (31, 32). FcγRIIA, but not FcγRIIIB, could bind IC made with mouse IgG1, IgG2a, or IgG2b (Figure (Figure8B).8B). Human neutrophils could be activated in vitro by IC made with mouse polyclonal IgG (Figure (Figure8C)8C) or mouse monoclonal IgG1 or IgG2b (Supplemental Figure 4B), as could murine neutrophils from WT and 5KO, but not from FcRγ–/–, mice (Supplemental Figure 4C). Neutrophils purified from normal human donors could restore ASA when injected i.v. into immunized FcRγ–/– mice prior to antigen challenge (Figure (Figure8,8, D–F). Importantly, the same neutrophils purified from individual donors induced no anaphylaxis upon antigen challenge in nonimmunized mice (Figure (Figure8,8, E and F). The intensity of the shock was proportional to the number of human neutrophils transferred into FcRγ–/– mice. Human neutrophils can therefore substitute for murine neutrophils in ASA.
We show here that IgG antibodies, activating FcγRs, neutrophils, and PAF are the main players in ASA, whereas IgE antibodies (12), FcεRI (5), mast cells (5, 13), and histamine do not play a major role.
Several cell types have been involved in anaphylaxis; among them are mast cells and basophils. We demonstrate that neutrophils are not only sufficient to induce ASA, but play a dominant role. Indeed, neutrophil depletion, but not basophil depletion, eosinophil depletion, monocyte/macrophage inhibition or mast cell deficiency, abrogated ASA-associated death, and reduced temperature drop in WT mice. Importantly, transfer of mouse neutrophils expressing FcγRIV, but not of neutrophils expressing no activating FcγRs, restored ASA in FcRγ–/– mice. Neutrophil activation could be visualized in vivo within minutes in mice undergoing ASA. Although monocytes/macrophages were reported to be involved in a model of anaphylaxis induced by an i.v. injection of goat IgG in mice immunized with goat IgG anti-mouse IgD (33), we could not detect any role for monocytes/macrophages in ASA in mice immunized with antigen in CFA/IFA. The depletion of eosinophils did not impair ASA either. Basophil depletion, which did not affect ASA by itself, further reduced ASA when combined with neutrophil depletion. Basophils therefore contribute to ASA together with neutrophils, but neutrophils play a dominant role. One possible explanation is that neutrophils are much more numerous than basophils in blood.
Noticeably, we observed neutrophilia following immunization of mice with antigen in CFA/IFA. We excluded that neutrophilia accounted for the dominant contribution of neutrophils to ASA by delaying antigen challenge until neutrophil numbers in immunized mice were comparable to those in naive mice. Under these conditions, neutrophils were also mandatory for ASA in 5KO mice. We found that neutrophils also predominantly contributed to IgG-induced PSA (whether induced by monoclonal IgG2b-IC or polyclonal-IC) in nonimmunized WT mice that have normal numbers of neutrophils. We also observed neutrophilia following immunizations with alum or with alum plus Pertussis toxin. In preliminary experiments, neither neutrophil depletion nor basophil depletion significantly reduced temperature drop, but a depletion of both cell types abolished ASA in mice immunized with antigen in alum (our unpublished observations). Taken together, our results demonstrate that neutrophils contribute to anaphylactic shock in 2 models of ASA, i.e., following immunization in CFA/IFA or in alum, and in 2 models of PSA, i.e., induced by polyclonal IgG-IC or monoclonal IgG2b-IC.
Two mediators, PAF and histamine, were found to play a critical role in experimental anaphylaxis. Mast cells, basophils and, apparently, neutrophils (34) can release histamine. Histamine accounts for IgE-induced PSA but not for IgG1-induced PSA, whereas PAF accounts for IgG1-induced PSA but not IgE-induced PSA (10). We found that PAF has a dominant role in ASA. Indeed, PAF-R antagonists, but not histamine receptor H1 antagonists, markedly reduced temperature drop and prevented death in ASA in mice immunized with antigen in CFA/IFA. In agreement with these results, ASA-associated heart rate and arterial pressure reduction were strongly impaired in PAF-R–deficient mice (15), suggesting that ASA-associated temperature drop and mortality may also be inhibited in these mice. Indirectly supporting this assumption, ASA was virtually abrogated in cPLA2-deficient mice (our results and ref. 29), which cannot generate several lipid-derived mediators, including PAF. Activated neutrophils (20, 22), monocytes/macrophages (35), and eosinophils (36) all produce PAF, but neutrophils were reported to be major producers (21). We found, indeed, that neutrophils secrete PAF when stimulated by IgG2b-IC, but we also found elevated PAF levels in plasma of mice undergoing neutrophil-dependent PSA. Together our results suggest that PAF released upon neutrophil activation during ASA and PSA is responsible for anaphylactic shock.
ASA depends on activating receptors (5) that associate with and whose expression and signaling depend on FcRγ. These are FcγRI, FcγRIIIA, FcγRIV, and FcεRI (37). ASA was reported in IgE-deficient mice (12), and we observed that it was comparable in WT and FcεRI-deficient mice. Similarly, the deletion of FcεRI did not affect ASA in mice immunized with antigen in alum (5, 38). While blocking either FcγRIIIA or FcγRIV reduced ASA, blocking both receptors abrogated ASA in WT mice. IgG antibodies seem therefore more important than IgE antibodies in ASA following immunization in CFA/IFA. FcγRIV contributed to ASA in WT mice, and it was necessary and sufficient to induce ASA in 5KO mice. Although FcγRIV are expressed by neutrophils and by monocytes/macrophages (18), only neutrophil FcγRIV accounts for ASA in 5KO mice. A single activating FcγR on a single cell population is therefore sufficient to induce ASA. Our results demonstrate that FcγRIIIA or FcγRIV is responsible for ASA following CFA/IFA immunizations, and that each could substitute for the other to induce ASA, provided that IgG2 antibodies are produced.
That IgG1 can induce anaphylaxis was demonstrated by PSA. The only activating FcR having an affinity for IgG1 is FcγRIIIA (17, 18). Depletion of basophils using specific mAbs abrogated IgG1-induced PSA (10), although mast cells and neutrophils also express FcγRIIIA. A model of basophil-deficient mice, however, challenges this result (14). Mast cells are necessary for IgE-induced PSA, although basophils also express FcεRI. Noticeably, IgE-induced PSA was abolished in 5KO mice (data not shown). FcγRIV that can bind IgE with a low affinity (17) is therefore insufficient to trigger this reaction. IgG1 is the predominant antibody subclass following immunization in alum, but also in CFA/IFA. IgG1 antibodies are likely to act as the main players in ASA by engaging FcγRIIIA. BSA immunizations in alum induced specific IgG1, but not specific IgG2 (Supplemental Figure 4D), indicating that FcγRIIIA, but not FcγRIV, were engaged during ASA following this immunization protocol. Supporting this conclusion, FcγRIV was not sufficient to induce ASA in 5KO mice following immunization in alum (Supplemental Figure 4E).
We found, however, that not only IgG1, but also IgG2, antibodies could induce PSA. IgG2-induced PSA could develop in 5KO mice, and neutrophils contributed to the shock. IgG2 can therefore contribute to anaphylaxis. IgG2-IC that may form in vivo upon antigen challenge following immunization in CFA/IFA may be responsible for the predominant role of neutrophils as IgG2 IC can engage FcγRIV (but also FcγRIIIA) on these cells. In the absence of IgG2 antibodies, as in ASA following immunization with antigen in alum, IgG1-IC may trigger FcγRIIIA-expressing basophils and neutrophils. It follows that basophils contribute to ASA and to PSA when IgG1-IC are present (this report and ref. 10), but also other cells (14), among which are neutrophils. Our results indicating that neutrophils are mandatory for polyclonal IgG-induced PSA differ from previous reports implicating mast cells (5) and basophils (10) in IgG1-induced PSA. They are, however, not contradictory. Indeed, monoclonal IgG1-IC can selectively engage FcγRIIIA, while IgG-IC made of IgG1 and IgG2 isotypes will engage both FcγRIIIA and FcγRIV, to induce PSA. FcR-expressing cells involved in each type of PSA may therefore be different. Taken together, these data suggest that IgE, IgG1, and IgG2 can all induce anaphylaxis when engaging FcεRI, FcγRIIIA, and FcγRIV on mast cells, basophils, and neutrophils, respectively. Although likely, it was not formally demonstrated that FcεRI or FcγRIIIA alone could induce ASA. Using the 5KO model, we provide evidence that FcγRIV alone can. Because FcγRIV is a receptor for IgG2, but not for IgG1, and is expressed by neutrophils, but not by basophils or mast cells, we are able to demonstrate here a role for the most unexpected antibody, receptor, and cell type in ASA in mice.
Human anaphylaxis is essentially active. IgG1 is the most abundant IgG subclass in human plasma, and the majority of antibodies raised by vaccinations (generally in alum) belong to this subclass. Human IgG1 binds to all activating human FcγRs (39). Both specific IgE and IgG antibodies are found in the serum of allergic patients, but the relative concentration of the various classes is poorly known and rarely investigated. FcγRIV has no human ortholog, and FcγRIIIA are not expressed by human neutrophils, basophils, and mast cells (40). These 3 cell types, nevertheless, express activating Fc receptors. All of them express another activating IgG receptor, FcγRIIA. Mast cells and basophils also express the high-affinity IgE receptor FcεRI in normal individuals as well as neutrophils in atopic patients (41). These 3 cell types can produce PAF upon activation, especially neutrophils (21). PAF could play an important role in human anaphylaxis. Indeed, plasma PAF concentration has been correlated with the severity of shocks in patients (42). Although the cellular source of PAF in human anaphylaxis was not identified, the above-mentioned results endow neutrophils with a critical role. Supporting this assumption, IgG-IC could activate human neutrophils in vitro, and a transfer of human neutrophils restored ASA in FcRγ-deficient mice.
In conclusion, we demonstrate here a previously unexpected role of neutrophils in anaphylaxis. An IgG2-induced, FcγRIV-dependent, PAF-mediated active anaphylaxis, contingent on neutrophils, could be unraveled using immunized multiple FcR-deficient mice. IgG-induced, FcγR-dependent, PAF-mediated ASA was also observed in WT mice, to which neutrophils contributed. Anaphylaxis induction is therefore a property of neutrophils, which is not trivial considering that neutrophils are the most numerous cells among blood leukocytes in humans. This may have important therapeutic consequences if indeed neutrophils can induce IgG-dependent anaphylactic reactions in humans.
5KO (N6 B6) mice have been described (17). cPla2–/– mice (129/B6) were provided by J. Bonventre (Brigham and Women’s Hospital, Harvard Institutes of Medicine, Boston, Massachusetts, USA); KRN transgenic mice were provided by D. Mathis and C. Benoist (Harvard Medical School, Boston, Massachusetts, USA) and the Institut de Génétique et de Biologie Moleculaire et Cellulaire (Illkrich, France). WT C57BL/6J and NOD mice were purchased from Charles River, and Wsh/Wsh, FcγRIIIA–/– and FcRγ–/– C57BL/6J mice from Jackson Laboratories. Rag2–/– mice were used to generate Rag2–/– 5KO and Rag2–/–FcRγ–/– mice by intercrosses. All mouse protocols were approved by the Animal Care and Use Committees of Paris, Île de France, France.
PBS- and clodronate-liposomes were prepared as previously described (19). Alum hydroxide gel, pertussis toxin, BSA, DNP-HSA, rabbit GPI, gadolinium(III) chloride, CFA and IFA, CV-3988, ABT-491, cyproheptadine, and luminol were obtained from Sigma-Aldrich; anti-mouse CD11b, CD4, CD19, Gr1, SiglecF, CD117, IgE, and anti-human FcγRI, FcγRIII, CD62L were from BD Biosciences; anti-mouse DX5, mFcεRI, and anti-hFcεRI were from eBioscience; anti-mFcγRI (290322) and mFcγRIIIA (275003) were from R&D Systems; anti–hFcγRIIA mAbs (IV.3) were from StemCell Technologies; mouse IgG3 anti-DNP were from Serotec; and the MPO ELISA kit was from HyCult Biotech. The hybridomas producing mAbs anti-mFcγRIV (9E9) were provided by J.V. Ravetch (Rockefeller University, New York, New York, USA); anti-Gr1 (RB6-8C5) was provided by R. Coffman (DNAX Research Institute, Palo Alto, California, USA); mIgG2a anti-platelet (6A6) was provided by R. Good (University of South Florida College of Medicine, Tampa, Florida, USA); and mIgG1 and mIgG2b anti-DNP was provided by B. Heyman (Uppsala Universitet, Uppsala, Sweden). Purified mAbs anti-hFcγRIIB/C (GB3) were provided by U. Jacob (SuppreMol GmbH); anti-CCR3 was provided by J.J. Lee (Mayo Clinic, Scottsdale, Arizona, USA). Anti-CD200R3 (Ba103) was produced as described (10). CHO K1 cells stably transfected with FLAG-tagged mouse FcγRs (17) or human FLAG-tagged FcγRs (39) were cultured as described.
Cells were stained with indicated fluorescently labeled mAbs for 30 minutes at 0°C.
Mouse IC were preformed by incubating 10 μg/ml DNP16-BSA-biotin with 15 μg/ml anti-DNP mAbs for 1 hour at 37°C. 2 × 105 cells, preincubated or not with indicated blocking mAbs, were incubated with IC for 2 hours at 4°C, which were detected using neutravidin-PE at 2 μg/ml for 30 minutes at 4°C.
6- to 9-week-old male mice were used for ASA. They were injected i.p. at days 0, 14, and 28 with 200 μg BSA, either once in CFA and twice in IFA, or 3 times in alum, or 3 times in alum plus Pertussis toxin. BSA-specific IgG1 and IgG2a/b/c antibodies in serum were titered by ELISA at day 30. Briefly, BSA-coated plates were sequentially incubated with dilutions of serum, HRP-labeled anti-Ig isotype antibodies, and SIGMAFAST OPD solution. Internal negative (serum from naive C57BL/6J mice) and positive (pool of serum of BSA-immunized C57BL/6J mice) controls were added to all test plates to define a background value and a “value 1” in arbitrary units, respectively. High titers of IgG1 antibodies were found in mice immunized using either CFA/IFA or alum immunization protocol. IgG2 antibodies were detected in mice immunized using CFA and IFA only. Absolute counts and proportions of granulocytes in peripheral mouse blood were determined using an ABC Vet automatic blood analyzer (Horiba ABX).
Mice with comparable antibody titers were challenged i.v. with 500 μg BSA (unless otherwise specified) 10 days after the last immunization. Note that the amount of antigen that induced ASA (Supplemental Figure 5A) was similar to that reported to be required for IgG-induced PSA (43). Central temperature was monitored using a digital thermometer (YSI), and time of death was recorded. Except in Figure Figure1,1, E and F, and in Figure Figure2A,2A, data presented in each figure correspond to 1 representative experiment. Supplemental Table 1 lists the total number of mice that were used for indicated experimental conditions, their grades of anaphylactic shock, and the mortality rate.
Mice were injected i.v. with preformed IC made of 20 μl K/BxN serum and 50 μg GPI (Figure (Figure4B);4B); 500 μg IgG2b mAb anti-DNP and 200 μg DNP-HSA (Figure (Figure3,3, A, B, and D); and IC made of 100 μl K/BxN serum and 40 μg GPI (Figure (Figure3,3, E–I). Alternatively (Supplemental Figure 1B), 500 μl of serum from BSA-immunized mice collected on day 30 was injected i.v. into naive mice and mice were challenged 3 hours later with an i.v. injection of 500 μg BSA. Central body temperature was recorded using a digital thermometer (YSI).
Blood samples were taken retroorbitally before and at indicated time points after the i.v. injection of 5 μg of anti-platelet mAb 6A6. Platelet counts were determined using an ABC Vet automatic blood analyzer (Horiba ABX).
200 μg/mouse of anti-FcγRIV mAbs were injected i.v. 30 minutes before challenge. 50 μg/mouse of anti-FcγRIIIA mAbs, 2.1 mg/mouse of PBS- or clodronate-liposomes, 300 μg/mouse of anti-Gr1 or anti-CCR3 mAbs, and 30 μg/mouse of anti-CD200R3 mAbs or 1 mg/mouse of gadolinium were injected 24 hours before challenge. 66 μg/mouse CV-3988, 25 μg/mouse ABT-49, or 50 μg/mouse cyproheptadine was injected i.v. 10, 20, or i.p. 30 minutes before challenge, respectively.
Depletion of cell populations was ascertained for specificity of the depletion using flow cytometry on blood samples taken during or after the experiment (as represented in the figures, or in examples shown in Supplemental Figure 1D, Supplemental Figure 2, A and B, and Supplemental Figure 3, D and E). Depletion of specific cell populations or efficiency of blocking antibodies was also controlled using the macrophage-dependent experimental immune thrombocytopenia (ITP) model: ITP in 5KO mice was inhibited by FcγRIV blockade (Supplemental Figure 5B), as expected (18), was inhibited by monocyte/macrophage depletion using toxic liposomes (Supplemental Figure 5C), as expected (44), and was unaltered by neutrophil depletion (Supplemental Figure 5D).
Bioluminescence from depilated and anesthetized mice injected i.p. with 15 mg/mouse luminol was acquired on an IVIS 100 (Caliper Life Sciences) using 2–5 minute exposure times with medium binning. Average radiance (photons/seconds/cm2/surface radiance) and total photon flux of indicated regions of interest (Figure (Figure2C)2C) were calculated using Living Image software.
For in vitro analysis, murine neutrophils were purified (>95% purity) from bone marrow suspensions using anti-Ly6G microbead kits (Miltenyi Biotec). 1.4 × 105 neutrophils were challenged with plate-bound stimulants or PMA plus thapsigargin for 1 hour. CD62L expression was assessed by flow cytometry and MPO by ELISA on cell supernatants.
For in vivo transfer, neutrophils were recovered from bone marrow suspensions following double-gradient centrifugation (Histopaque-1119, Ficoll-Paque PLUS: 50/50 v/v). 3 × 106 neutrophils were injected i.v. into BSA-immunized mice 20 minutes before challenge.
Purified neutrophils (1 × 106 cells in 0.1 ml) were incubated on plate-bound DNP-HSA alone (Ag; at 10 μg/ml) or plate-bound DNP-HSA (at 10 μg/ml) plus anti-DNP IgG2b GORK (IC; at 100 μg/ml) in HBSS buffer supplemented with 1.26 mM CaCl2 and 0.9 mM MgCl2. When indicated, neutrophils were preincubated with 20 μg/ml of anti-FcγRIV mAbs. Culture supernatants were collected after 20 minutes at 37°C.
100 μl of plasma was collected from mice undergoing monoclonal IgG2b PSA (IC made of 1 mg IgG2b mAb anti-DNP and 300 μg DNP-HSA) at t = 20 minutes following challenge. When indicated, mice were preinjected with 200 μg/ml of anti-FcγRIV Ab 30 minutes before challenge.
Culture supernatants or plasma were mixed 1:10 v/v in methanol, centrifuged at 16,100 g for 10 minutes; supernatants were frozen at –20°C. After adding deuterized PAF (PAF-d4; Cayman Chemical) as an internal standard to supernatants, PAF-containing lipids were extracted with Oasis HLB solid phase extraction cartridge (10 mg/ml; Waters) as described (45). PAF was quantified using liquid chromatography-tandem mass spectrometry on a TSQ Quantum Ultra mass spectrometer (Thermo Fisher Scientific) that was operated in negative electrospray ionization. HPLC conditions and mass spectrometry parameters were described previously (45). The selected reaction monitoring (SRM) modes was used to monitor 568.5 to 59 m/z transitions for PAF and 572.5 to 59 m/z transitions for PAF-d4. Commercial PAF (Cayman Chemical) mixed with PAF-d4 was used to draw calibration curves, and data were processed using the Xcalibur 2.0 software (Thermo Fisher Scientific). PAF values of less than 5 pg should be considered under the lower detection limit (Figure (Figure2B2B and Figure Figure3C). 3C).
Human granulocytes were purified (>95% purity) from normal heparinized blood by double-gradient centrifugation (Histopaque-1119, Ficoll-Paque Plus: 50/50 v/v) and using anti-CD15 microbeads (Miltenyi Biotec). For transfer experiments, neutrophils were injected i.v. 15 minutes prior to BSA challenge. For in vitro analysis, 2 × 105 neutrophils were stimulated for 1 hour at 37°C with 100 nM PMA or IC (30 μg/ml GPI and 1/100 dilution of KxB/N serum) or neutrophils with plate-bound IC (DNP-HSA and indicated mouse anti-DNP antibodies).
We are thankful to F. Hamano (The University of Tokyo, Tokyo, Japan) for help with PAF measurements, to A.-M. Nicola and the Plate-Forme d’Imagerie Dynamique (Institut Pasteur, Paris) for help with the bioluminescence experiments, to C. Detchepare for administrative help, and to G. Eberl and P. Bousso (Institut Pasteur, Paris) for critical reading of the manuscript. We are thankful to our colleagues for their generous gifts: S. Verbeek (Leiden University Medical Center, Leiden, The Netherlands), J.-P. Kinet (Harvard Institutes of Medicine, Boston, Massachusetts, USA), M. Lamers (Max-Planck-Institut für Immunbiologie, Freiburg, Germany), D. Mathis and C. Benoist (Harvard Medical School, Boston, Massachusetts, USA) and the Institut de Génétique et de Biologie Moléculaire et Cellulaire (Illkirch, France) for mice; R. Coffman, R. Good, B. Heyman, J.J. Lee, and J.V. Ravetch for antibodies. We thank J. Bonventre and E. O’Leary (Brigham and Women’s Hospital, Harvard Institutes of Medicine, Boston, Massachusetts, USA) for cPLA2–/– mice, and F. Bourgade and K. Sebastien (Centre des Opérations Sanitaires de l’Animalerie Centrale, Institut Pasteur, Paris, France) for their breeding and housing. We thank M. Aubier (Service de Pneumologie, Hôpital Bichat, Paris, France) for advice on human neutrophil transfer, A. Herbelin (INSERM U935, Hôpital Paul Brousse, Villejuif, France) for advice on alum immunization protocols, and X. Zhang (Institut Pasteur, Paris) for advice on neutrophil purification. Cl2MDP was a gift of Roche Diagnostics GmbH. This work was supported by the Institut Pasteur, the Institut National de la Santé et de la Recherche Médicale (INSERM), the Agence Nationale de la Recherche (ANR) (grants 05-JCJC-0236-01 and GENOPAT-09-GENO-014-01), the Fondation pour la Recherche Médicale (FRM) (Programme Allergie 2007), by funding under the Sixth Research Framework Programme of the European Union, Project MUGEN (MUGEN LSHG-CT-2005-005203), by the Société Française d’Allergologie (SFA) (Soutien de la recherche en allergologie 2010) and the Balsan company. F. Jönsson was financially supported by ANR, MUGEN, and is currently a recipient of a fellowship from the FRM. D.A. Mancardi was financially supported by FRM and is currently a recipient of a fellowship from the Institut Pasteur (Bourse Roux). T. Shimizu and Y. Kita are supported in part by Grants-in-Aid from the Ministry of Education, Science, Culture, Sports and Technology of Japan.
Conflict of interest: The authors have declared that no conflict of interest exists.
Citation for this article: J Clin Invest. 2011;121(4):1484–1496. doi:10.1172/JCI45232.
See the related Commentary beginning on page 1260.