|Home | About | Journals | Submit | Contact Us | Français|
Innate sensing mechanisms trigger a variety of humoral and cellular events that are essential to adaptive immune responses. Here we describe an innate sensing pathway triggered by Plasmodium infection that regulates dendritic cell (DC) homeostasis and adaptive immunity via Flt3L release. Plasmodium-induced Flt3L release requires toll-like receptor activation and type I interferon production. We find that type I interferon supports the up-regulation of xanthine dehydrogenase, which metabolizes the xanthine accumulating in infected erythrocytes to uric acid. Uric acid crystals trigger mast cells to release soluble Flt3L from a pre-synthesized membrane-associated precursor. During infection Flt3L preferentially stimulates expansion of the CD8α+/CD103+ DC subset or its BDCA3+ human DC equivalent and has a significant impact on the magnitude of T cell activation, mostly in the CD8+ compartment. Our findings highlight a new mechanism that regulates DC homeostasis and T cell responses to infection.
The Flt3 receptor tyrosine kinase and its ligand Flt3L, regulate DC development in the steady state1–6. Flt3 is first expressed in multipotent haematopoetic progenitors (CD150−Flt3+) that develop into lymphoid and myeloid cells7, 8, but expression is lost by all lineages except DCs during development.
In the steady state, DC development proceeds from precursors that give rise to monocytes and DCs (MDP)4, 9, to common DC progenitors that lack monocyte potential (CDP)5, 6, 10, 11, and give rise to circulating pre-DCs11–13. All of these cell types are dependent on Flt3L in the steady state. During inflammation, pattern recognition receptors, such as Toll-Like Receptors, RIG-like helicases and the inflammasome, mediate pathogen sensing and induce the maturation of differentiated DCs, a process controlling the initiation of immune responses. However, little is known about the mechanisms that regulate the development of DCs in response to infection and microbial innate sensing.
In this report we examine how the DC pathway is regulated during acute malaria infection. We find that Plasmodium infection is associated with increased serum levels of Flt3L and increased DC production in both mice and humans. Moreover, DC expansion in response to P. chabaudi infection requires Flt3L, which has a predominant effect on the CD8a+ DC subset in mice and BDCA3+ DC subset in humans. We identify mast cells (MCs) as an important source of the Flt3L, and show that they produce it in response to uric acid accumulation. This response is facilitated by TLR9-dependent type I IFN induction of Xanthine dehydrogenase, which supports hyper-uricemia during Plasmodium infection. Ultimately, the DCs induced by inflammatory Flt3L help activate T cell responses to the parasites.
To examine DC regulation during Plasmodium infection, we infected C57BL/6 mice with blood stage P. chabaudi and measured Flt3L1–3. We found that soluble Flt3L was increased from 300 to 1000 pg.ml−1 in serum as early as d2 and that it remained elevated up to six days after infection (Fig.1a). Consistent with previous reports14, 15, we found that spleen DCs (DEC205+CD8α+ and 33D1+CD8α−) undergo transient contraction and then expansion (Fig.1b, c, d, Supplementary Fig.1a, b). To determine if DC expansion during P. chabaudi infection was dependent on Flt3L and its receptor Flt3, we infected wild type or Flt3−/−16 or Flt3L−/−1 mice and quantified spleen DCs (Fig.1c, d). As previously noted, both subsets of spleen DCs were reduced in naive Flt3−/−4 and Flt3L−/− mice1. In addition, expanded DC populations, especially CD8α+ DCs, remained Flt3- and Flt3L-dependent (Fig. 1c, d).
To determine which aspect of DC development is dependent on Flt3 during P. chabaudi infection, we analyzed DC progenitors in the BM. Haematopoietic stem cell (HSC)-containing fraction (Lin− CD117(c-kit)highSca-1+) undergoes a progressive expansion as early as two days after infection (Fig.1e, Supplementary Fig.2a)17. Both Lin−CD117(c-kit)highSca-1+CD150+ and CD150− subsets expand, but only Flt3-expressing CD150− cells are critically regulated by Flt3 (Fig.1f, Supplementary Fig.2b)7, 8. Similar to DCs in the periphery, MDP and CDP undergo a transient contraction followed by expansion as determined by analyzing Lin−CD117+Sca1−CD115+ cells4, 9, 10(Fig.1g, Supplementary Fig.2c). Importantly, the infection-induced MDP/CDP expansion is compromised in Flt3−/− and Flt3L−/− mice (Fig.1h, Supplementary Fig.2d) demonstrating the entire DC developmental pathway responds to infection by a Flt3-dependent mechanism.
We used langerin-DTR mice to deplete CD8α+/CD103+DCs18, 19, which is the DC subset most affected by acute increase in Flt3L (Fig.2a). P. chabaudi infection triggers activation of both CD4+ and CD8+ T cells as measured by CD44highCD62Llow effector/memory T cells 7 days after infection (Fig.2b–g). Depletion of spleen CD8α+/CD103+ DCs did not modify the percentage of activated CD4+ T cell in the spleen (Fig.2b,c), nor their phenotype (Supplementary Fig.3d), however their number was reduced (Fig. 2d). In contrast, CD8α+/CD103+ DC depletion impaired CD8+ T cell activation both at the level of percentages and absolute numbers (Fig.2e,f,g). We conclude that Flt3L-regulated CD8α+ DCs support T cell responses most importantly in the CD8+ T cell compartment. Depletion of CD8α+/CD103+DCs increased parasitemia and compromised survival to P. chabaudi infection (Supplementary Fig.3a–c).
In contrast to P. chabaudi20, CD8+ T cell responses to P. berghei ANKA induce lethal brain immunopathology21–23. Depletion of CD8α+/CD103+ DCs during to P. berghei ANKA infection using langerin-DTR mice delayed lethality without impacting parasitemia (Supplementary Fig. 4a), reduced brain pathology (Supplementary Fig.4b) and CD8+ T cell accumulation (Supplementary Fig.5a). Similar results were found in Flt3L−/− mice (Supplementary Fig.6). We conclude that Flt3L-regulated CD8α+/CD103+ DCs are required for the onset of CD8+ T cell-dependent brain immunopathology during P. berghei ANKA infection.
To determine the source of inflammatory Flt3L we produced reciprocal BM chimeras between WT and Flt3L−/− mice (Fig.3a). Analysis of Flt3L serum levels revealed that the increase in serum Flt3L in response to P. chabaudi infection was dependent on the radio-resistant compartment. Lethally irradiated Flt3L−/− recipients of WT or Flt3L−/− BM were defective in their serum Flt3L responses to P. chabaudi infection, whereas WT recipients of WT or Flt3L−/− BM were not (Fig. 3a). We conclude that a radio-resistant cell type releases Flt3L during P. chabaudi infection.
Analysis of expression database (http://biogps.gnf.org) indicate that radioresistant24 mast cells (MCs, and Supplementary Fig 7), express high levels of Flt3L in the steady state. To analyze the potential contribution of MCs to Flt3L increase, we analyzed the membrane bound precursor of Flt3L on BM-derived MC (BMMC) before and after degranulation (Supplementary Fig.8). MC degranulation (probed by lamp1 upregulation), decreases surface Flt3L levels while soluble Flt3L increases dramatically in supernatants (Supplementary Fig.8). Thus, BMMCs express Flt3L precursor, which is released as a soluble cytokine upon degranulation. To address the relevance of this process in vivo, we measured serum Flt3L upon FcεRI-triggered MC degranulation (Supplementary Fig. 9a-d). Naïve peritoneal MCs express high levels of membrane bound Flt3L (Supplementary Fig.9a), but other immune cells do not (Supplementary Fig.9b). Upon administration of anti-IgE, MCs degranulate (lamp-1 up-regulation) and show decreased cell surface Flt3L (Supplementary Fig. 9a). Concomitantly, Flt3L serum levels rapidly increase in anti-IgE but not in MC-deficient KitW-sh mice25 (Fig.3f) or control Ig-treated mice (Supplementary Fig.9c–d). We conclude that MC constitutively display cell surface Flt3L and that MC degranulation triggers the release of soluble Flt3L.
We next examined peritoneal MCs from P. chabaudi-infected mice. MCs underwent degranulation as measured by increased surface lamp-1 (Supplementary Fig.9e), and decreased membrane Flt3L (Fig. 3b, quantified in 3c) concomitantly with Flt3L increase in the serum (Fig.1a, ,3d).3d). Furthermore, the increase in serum Flt3L in response to P. chabaudi (and P. berghei ANKA) was dependent on MCs since it was nearly absent in MC-deficient KitW-sh mice25 despite normal levels of steady state Flt3L (Fig.3d, Supplementary Fig.9f). We conclude that MCs are activated during malaria infection and they release membrane Flt3L into the circulation.
In order to determine whether MCs are required for the observed increases in DCs during malaria infection, we analyzed spleen DC populations in MC-deficient KitW-sh mice (Fig. 3e–g). Although steady state numbers of CD8α+ DCs are similar in WT and KitW-sh mice, CD8α+ but not CD8α− DCs fail to increase in the spleen of P. chabaudi-infected KitW-sh mice (Fig.3e–g). Importantly, administration of recombinant Flt3L to P. chabaudi-infected KitW-sh mice restored CD8α+ DC expansion (Fig.3e,f) showing that the KitW-sh mutation does not impair DC responsiveness to Flt3L. MC deficiency also lead to a significant impairment in CD8+ T cell activation (Fig.3h,i) but not had no effect on CD4+ T cells (Fig.3j). Again, this defect is circumvented by administration of recombinant Flt3L, thus excluding a default in KitW-sh CD8+ T cells. Taken together, these results show that MCs promote CD8+ T cell activation by controlling Flt3L serum levels and spleen CD8α+/CD103+ DCs expansion during P. chabaudi infection.
P. chabaudi infection might induce Flt3L release from MCs by inflammasome-dependent cytokines (IL-1/IL-18), or ATP-sensing receptor P2X7R, but we found no evidence that these pathways mediate Flt3L release from MCs (Supplementary Fig.10a–c). An alternative possibility is that MC activation is mediated directly by Uric Acid. It has been shown in vitro that hypoxanthine released from Plasmodium-infected RBCs can be converted to uric acid by phagocyte-derived xanthine dehydrogenase26, 27 (Xdh) and uric acid crystals can in turn be sensed by various immune cells by a mechanism that requires direct activation of Syk28–30. To test the possibility that uric acid induces Flt3L, we injected mice with uric acid crystals (Fig.4a–c). Whereas serum Flt3L increased rapidly after the injection, MCs showed decreased levels of membrane bound Flt3L (Fig.4a–c). Control, allopurinol crystals had no effect on MC activation (Supplementary Fig.11a). By contrast, oxonic acid, a specific inhibitor of uricase inducing hyper-uricemia in vivo also increases serum Flt3L (Supplementary Fig.11b). We conclude that uric acid induces rapid increases in circulating Flt3L through MC activation.
To determine whether uric acid triggers MC release of Flt3L during Plasmodium infection, we measured serum uric acid levels during P. chabaudi infection (Fig.4d). We found that P. chabaudi infection (but not control injection, Supplementary Fig.11c) induced a significant and transient hyperuricemia, which is also a hallmark of experimental cerebral malaria in mice31, 32 and severe malaria in humans33, 34. To test the contribution of uric acid to serum Flt3L responses, we blocked uric acid accumulation by administration of allopurinol (an inhibitor of Xdh) and uricase (Supplementary Fig.11d). Consistent with the idea that uric acid triggers Flt3L release from MCs, Flt3L serum levels were significantly decreased in treated mice (Fig.4e). MC activation was also decreased (Lamp1 exposure and Flt3L release: Fig.4f). Uric acid inhibition also led to a significant decrease in CD8a+ DCs expansion in the spleen of P. chabaudi infected mice (Fig.4g) and CD8+ T cell activation (d8, Fig.4h). We conclude that uric acid participates in MC activation and Flt3L release during malaria infection, which results in DC expansion and promotes T cell activation.
Toll like receptor (TLRs) are of paramount importance in innate sensing of Plasmodium parasites35–38. We found that Myd88−/−/TRIF−/−, Myd88−/− and TLR9−/− mice but not TRIF−/− mice had an impaired Flt3L response to P. chabaudi blood stage infection (Fig.5a). In accordance with TLR9 involvement35–38, we found that P. chabaudi infection triggers a transient and systemic IFN-α response that was fully abrogated in Myd88−/−/TRIF−/− mice (Fig.5b) (and pDCs-depleted, Supplementary Fig12a,b). Mice deficient in type I IFN receptor (IFNα/βRI−/−) failed to produce a systemic increase in Flt3L in response to P. chabaudi (Fig.5c) or P.berghei ANKA (Supplementary Fig.12c). However, this effect was indirect because type I IFN failed to induce Flt3L secretion from BMMCs (Supplementary Fig.10a,b), and the requirement for type I IFN receptor was on radiosensitive cells (Fig.5d), which are not the main source of inflammatory Flt3L (Fig.3a). Since uric acid sensing participates to MC activation and Flt3L release during Plasmodium infection (Fig.4), we asked whether type I IFN signaling was involved in regulating uric acid production. To test this possibility, we measured the uric acid producing enzyme Xanthine dehydrogenase (Xdh). We found that Xdh mRNA, its enzymatic activity and lung uric acid levels (Fig.5e) were highly up-regulated in the first day after P. chabaudi or P.berghei ANKA (Sup. Fig.12d) infection in a manner that was IFNα/βRI-dependent.
Consistent with the role of uric acid in triggering MC activation and Flt3L release (Fig.4), we found that MC activation was impaired in IFNα/βRI−/− both in terms of Lamp1 exposure and Flt3L release (Fig.5f). As a result of impaired Flt3L release, both type I IFNα/βRI−/− (Fig.5g) and Myd88−/−TRIF−/− (Supplementary Fig.13a) showed impaired spleen CD8α+ DC expansion in response to P. chabaudi infection despite normal responsiveness to recombinant Flt3L (Supplementary Fig.13b). CD8+, but not CD4+ T cell activation was also impaired in IFNα/βRI−/− mice (Fig.5h). We conclude that type I IFN signaling downstream TLR-dependent type I IFN secretion controls XdH production by radiosensitive cells. This acute increase in XdH synthesis supports uric acid production, which in turn induces MCs to release Flt3L in response to Plasmodium infection (Supplementary Fig. 14 for diagram).
Severe human malaria is associated with increased numbers of circulating BDCA3+ DCs39, which correspond to murine CD8α+/CD103+ DCs18, 40–42. To determine whether Flt3L is elevated in humans infected with P. falciparum we measured levels of this growth factor in the blood of Kenyan children with mild or severe malaria. Flt3L was not detectable in the serum of age matched un-infected controls (n=47), but was present in n=5/27 cases of mild and n=32/92 cases of severe malaria (Fig. 6a). The increase in Flt3L plasma concentration with increasing disease severity was statistically significant (Kruskal Wallis test p<0.0001). For a subgroup of patients, we had previously determined the frequency of peripheral blood DC subsets39. Patients with detectable levels of Flt3L in plasma had elevated frequencies of circulating BDCA3+ but not BDCA1+ or BDCA2+ DCs (Fig. 6b–d).
To determine whether human Flt3L increases the numbers of circulating BDCA3+ DCs we analyzed the effect of human Flt3L administration in mice reconstituted with human HSCs (Fig. 6e,f, Supplementary Fig15a). We found that Flt3L administration was sufficient to induce an increase the human DC compartment in blood (Fig6e) and the spleen (Fig.6f). Importantly, the effect was more pronounced in the BDCA3+ DC subset in both blood and spleen compartments (Fig. 6e,f, Supplementary Fig15b).
To determine whether the increased levels of Flt3L and DCs are also associated with enhanced levels of CD8+ T cell activation in infected individuals, we analyzed T cell activation status by measuring CD69 expression levels by flow cytometry. We found that Flt3L positive patients showed a similar proportion of activated CD4+ T cells (Fig.6g) and an increased proportion of activated CD8+ T cells (Fig. 6h). We conclude that the host response to human malaria includes a systemic increase in Flt3L, BDCA3+ DC expansion and CD8+ T cell activation.
Here we demonstrate that innate sensing by MCs during Plasmodium infection leads to a systemic elevation of Flt3L levels which impacts on host responses to the parasite by increasing the number of DCs and promoting T cell activation.
Flt3L is an essential growth factor for DCs in vitro and in vivo1, 4, 43 that can have tolerogenic effects in the absence of systemic inflammation44, 45. During MCMV infection, Flt3L production is upregulated46. We have shown that Flt3L levels in serum also increase in response to P. chaubaudi and P. berghei ANKA infections in mice, and P. falciparum in humans. Having found that inflammatory Flt3L originates from the radio-resistant compartment, we identified radio-resistant MCs as major producers of soluble Flt3L.
MCs and their products have been implicated in regulating DCs47, 48. MCs release of pre-synthesized TNF-α during inflammation promotes DC maturation and migration from skin to lymph nodes47–49. Our experiments uncover a previously unappreciated TNF-α–independent role for MCs in regulating serum Flt3L and DC populations during inflammation. MCs express high levels of the membrane Flt3L, which is rapidly released into the circulation upon degranulation resulting in preferential expansion of CD8α+CD103+ DCs. This new function for activated MCs in adaptive immunity is complementary their role in DC migration47–49 and down-regulation of Treg function50.
Our experiments show that Flt3L release by radio-resistant MCs in response to Plasmodium infection requires type I IFN signaling by radio-sensitive cells. Type I IFN is produced in response to TLR-9/Myd88 sensing during Plasmodium infection35–38. Moreover, mice deficient for the type I IFN receptor or TLRs display impaired immune responses to P. chabaudi infection36 and are more resistant to P. berghei immunopathology51, 52. Type I IFN mediates most of its effects on Flt3L indirectly, i.e., by acting on the radiosensitive compartment. Type I IFN induces Xanthine dehydrogenase (Xdh), an enzyme that converts hypoxanthine and xanthine that accumulates in Plasmodium-infected erythrocytes to uric acid26, 27. In addition uric acid can also precipitates inside infected RBCs providing an inflammatory stimulus for CD80/CD86 upregulation and MHC II down-regulation in DCs30. Uric acid, released from dead cells36 or Plasmodiuminfected erythrocytes26, 27, 30 is known to be pro-inflammatory and hyper-uricemia is a hallmark of malaria infection31, 32 including severe malaria in humans33, 34. Our experiments reveal that, in addition to direct DC activation29, 30, 53, uric acid contributes to Flt3L release from MCs leading to increased DCs numbers.
In mice, the CD8α+CD103+ subset is the more sensitive to acute increases in serum Flt3L. This phenomenon is mirrored by the increase of their human equivalent, BDCA3+ DCs18, 40–42 in the circulation during P. falciparum infection39. Although, only limited data was available for infected children, the association between increased Flt3L and increased numbers of human BDCA3+ DCs was confirmed by mechanistic experiments in humanized mice showing Flt3L alone is able to preferentially increase the number of circulating BDCA3+ DCs.
Despite toxic effects of infection54, DC function is relevant for protective immunity to Plasmodium blood stage infection55. Experiments with CD11c-DTR mice where CD11c-expressing cells were depleted with DT show that these cells are essential for protective immunity to P.yoelii liver phase56 and for induction of experimental cerebral malaria in mice infected with the blood stage of P. berghei ANKA57. Our experiments with Langerin-DTR mice show that the loss of CD8α+CD103+ DCs, which are acutely responsive to inflammatory Flt3L, results in impaired CD8+ T cell activation19, 58, 59. In addition, the same CD8α+CD103+ DC subset is essential to induce pathogenic CD8+ T cells that are responsible for lethal brain inflammation during P. berghei ANKA infection. Finally we show that, the proportion of activated CD8+ T cells (but not CD4+ T cells) in peripheral blood correlates with the magnitude of the Flt3L response in P. falciparum infection in severe malaria.
Taken together, out data uncover an evolutionarily conserved pathway that regulates DC development and T cell activation during infection.
A detailed version of methods I available in the Supplementary Online Materials including detailed cellular preparation protocoles and flow cytometry analysis.
C57BL/6 mice were purchased from Jackson. Flt3−/− mice were a kind gift of I. Lemischka16. Flt3L−/− mice were obtained from Taconics 1. KitW-sh/W-sh (KitW-sh/HNihrJaeBsmJ)25 and all other stains are described in S.O.M. Langerin-DTR mice60 were obtained from B. Malissen and hosted at the Mount Sinai Immunology Institute. BM chimeric mice were produced as previously described4. irradiated NOD Rag1−/− IL2Rγnull (the Jackson Laboratory) were reconstituted by injecting human CD34+ fetal liver-derived hematopoietic stem cells. Human fetal livers were obtained from Advanced Bioscience Resources (ABR), Inc. (Alameda, CA) or the Human Fetal Tissue Repository (HFTR, Bronx, NY) and hematopoietic stem cells were isolated as detailed in the S.O.M. All mice were maintained in specific pathogen–free conditions, and all mouse protocols were approved by the Rockefeller University Institutional Animal Care and Use Committee (IACUC). P. chabaudi chabaudi clone AS and P. berghei ANKA was obtained through the Malaria Research Reagents and Resource Center (MR4, http://www.mr4.org). Infections with Plasmodium berghei ANKA were performed as described previously23. See S.O.M for more details.
We analyzed the concentration of FLT3L in plasma by ELISA from children presenting with severe malaria at the Kilifi District Hospital on the Kenyan Coast between 2003 and 2007, those presenting with mild malaria in 2004–2009 and in healthy children identified during cross-sectional surveys in 2004 and 2009. Children admitted to Kilifi District Hospital with severe malaria had a parasite-positive blood film by microscopy and suffered from one or more of the following symptoms: impaired consciousness (Blantyre Coma Score <5), coma (Blantyre Coma score <3) severe anemia (Hb <5 g.dl−1 and at least 10,000 parasites.ml−1 blood) or severe respiratory distress (deep breathing and chest recession or nasal flaring). Children with mild malaria were identified in cohorts under active surveillance for acute malaria episodes and had a parasite-positive blood film by microscopy and temperature >37.5C. Healthy children had no fever or history of fever over the last 7 days and were negative for Plasmodium ssp. by microscopy and rapid diagnostic test (Optimal). For 21 healthy children and 27 children with severe malaria, data on the frequency of peripheral blood DC subsets were available as reported in Urban et al.39. Parents or Guardians of the children provided informed consent before participation in the studies. The studies received ethical approval from the Kenyan National Ethics Review Committee (SSC nos. 669, 883 and 1131) and the Oxford Tropical Research Ethics Committee (protocol nos. 005-02 and 30-06).
Mouse serum or culture supernatants Flt3L or IFN-α and human plasma Flt3L were measured by ELISA kit (RnD systems Cat#MFK001 and cat#DFK00, interferonsource Cat# 42120-1). Serum uric acid or from lung extracts (prepared by smashing lungs in sucrose 250mM, tris 50mM pH7.4, followed by two cycle of freeze/thaw before to be centrifuged 15mn at 15000rpm at 4degC to get rid of debris) was assessed using a recombinant Xdh-based colorimetric assay (Invitrogen, Cat# A22182). XdH was assayed using hypoxanthine as a substrate using a colorimetric assay (Invitrogen, Cat# A22181). For Xdh PCR, lungs from infected mice were chopped in Trizol LS reagent (Invitrogen, Cat# 10296-028), frozen on dry ice and stored at −80°C. RNA was purified according manufacturer’s protocol. cDNA was prepared using SuperScript II Reverse Transcriptase (Invitrogen, Cat# 18064-014) with random 6-mer primers according manufacturer’s protocol. Quantitative PCR (Q-PCR) was performed using Brilliant SYBR Green QPCR Master Mix (Agilent Technologies, Cat# 600548) according manufacturer’s protocol. Reactions were performed in triplicate and analyzed with an MX3000P Q-PCR machine (Stratagene). Results were normalized to GAPDH. Primers used were as follows: Xdh forward, 5’ TGCTATTTCCTCTGATCATCTGC 3’, Xdh reverse, 5’ CCAGGACACTTTCCACCAAG 3, GAPDH forward, 5’ TGA AGC AGG CAT CTG AGG G 3', GAPDH reverse, 5' CGA AGG TGG AAG AGT GGG AG 3'.
Acknowledgments and contributions
We are grateful to R. Steinman, M. P. Longhi, E. Pamer, Leiner I., K. Marsh for helpful discussion and reagents or critical reading of the manuscript, Celldex for hrFlt3L. We thank the children and their guardians for participation. This paper is published with the permission of the Director of the Kenya Medical Research Institute. This work was supported in part by NIH grant number AI051573 and by the Agency for Science, Technology and Research (Singapore).
P.G. is a CNRS investigator. B.C.U. is a Wellcome Trust Senior Research Fellow (grant no. 079082). M.C.N. is an HHMI investigator. M.C.N., P.G., B.C.U., M.M., Y.S., F.G., L.R., J.H., C.C., S.D. participated to experimental design, G.B., H. A.S., N. F.-S., E.B., M.D., C.M. R., A. P., F. K. were involved in the production and analysis of NSG humanized mice, M. S., M. C. provided BDCA2-DTR mice, P.G., J.H., C.C., S.D., H.K., A.G., G.D.-J., S.B.T., , A. O. K., M. M., R. N., C.T., F.M., A. H., D.B., T.E. conducted most immunological experiments in mice.