key: cord-0001628-7vsdqhi0
authors: Burgess, Stacey L.; Buonomo, Erica; Carey, Maureen; Cowardin, Carrie; Naylor, Caitlin; Noor, Zannatun; Wills-Karp, Marsha; Petri, William A.
title: Bone Marrow Dendritic Cells from Mice with an Altered Microbiota Provide Interleukin 17A-Dependent Protection against Entamoeba histolytica Colitis
date: 2014-11-04
journal: mBio
DOI: 10.1128/mbio.01817-14
sha: 0afa3ea846396533c7ca515968abcfea3f895082
doc_id: 1628
cord_uid: 7vsdqhi0

There is an emerging paradigm that the human microbiome is central to many aspects of health and may have a role in preventing enteric infection. Entamoeba histolytica is a major cause of amebic diarrhea in developing countries. It colonizes the colon lumen in close proximity to the gut microbiota. Interestingly, not all individuals are equally susceptible to E. histolytica infection. Therefore, as the microbiota is highly variable within individuals, we sought to determine if a component of the microbiota could regulate susceptibility to infection. In studies utilizing a murine model, we demonstrated that colonization of the gut with the commensal Clostridia-related bacteria known as segmented filamentous bacteria (SFB) is protective during E. histolytica infection. SFB colonization in this model was associated with elevated cecal levels of interleukin 17A (IL-17A), dendritic cells, and neutrophils. Bone marrow-derived dendritic cells (BMDCs) from SFB-colonized mice had higher levels of IL-23 production in response to stimulation with trophozoites. Adoptive transfer of BMDCs from an SFB(+) to an SFB(−) mouse was sufficient to provide protection against E. histolytica. IL-17A induction during BMDC transfer was necessary for this protection. This work demonstrates that intestinal colonization with a specific commensal bacterium can provide protection during amebiasis in a murine model. Most importantly, this work demonstrates that the microbiome can mediate protection against an enteric infection via extraintestinal effects on bone marrow-derived dendritic cells.

port neutrophil infiltration in inflammatory-disease models (10) (11) (12) (13) . Neutrophils play a central role in cell-mediated clearance of parasites and may be important in clearance of E. histolytica (14) . IgA was also implicated in protection against E. histolytica infection in a childhood cohort and thus may be involved in immunity against the parasite (15) . Given that IL-17A and downstream mediators are important in immunity to E. histolytica, components of the intestinal microbiota that elicit exogenous IL-17A in the absence of intestinal pathology might prove to be protective during amebiasis.

Segmented filamentous bacteria (SFB), or "Candidatus Savagella," are genetically and morphologically unique members of the Clostridia. They represent an uncultivable component of the mammalian intestinal microbiota, reported to colonize both humans and mice (16) (17) (18) . It has been demonstrated that SFB colonization induces a potent Th17 helper response in the intestine that is dependent on intestinal dendritic cells (19, 20) . SFB colonization is associated with higher intestinal levels of both IL-17A and the damage-associated molecular pattern molecule and antimicrobial peptide serum amyloid A (SAA) (20) . SFB infection influences many models of intestinal and extraintestinal inflammatory disease, suggesting that it may have a systemic influence on the immune response (21) (22) (23) . Indeed, a growing body of literature suggests that intestinal colonization with commensal microorganisms can have extraintestinal effects on dendritic-cell precursors that influence susceptibility to pathogens (24) (25) (26) . It is also quite possible that mediators induced by the intestinal microbiota in the serum might influence the bone marrow in such a way as to prime DCs to provide protection against, or exacerbate, enteropathogen infections (27, 28) .

Bone marrow dendritic cells (BMDCs) are known to be able to recapitulate effector functions of some in vivo antigen-presentingcell populations, and lipopolysaccharide (LPS)-matured BMDCs have been effectively utilized in adoptive transfer experiments to examine the influence of dendritic cells in infection models (28) (29) (30) . Thus, to begin to examine the influence of changes in the microbiota on resistance to amebic infection, we utilized a murine model and BMDCs to explore what influence colonization with SFB had on intestinal infection with E. histolytica. Here we dem-onstrate that colonization with SFB provided protection against E. histolytica colitis and that bone marrow dendritic cells (BM-DCs) derived from SFB-colonized mice were able to recapitulate protection in mice that had not been colonized with the bacteria. Protection mediated by BMDCs from SFB ϩ mice was IL-17A dependent. These data suggest that intestinal colonization with a commensal bacterium can alter bone marrow in such a way as to provide protection against parasite infection.

Cohousing Charles River (SFB ؉ ) and Jackson mice (SFB ؊ ) protected Jackson CBA mice from E. histolytica infection. To test if alteration if the intestinal microbiota could alter susceptibility to E. histolytica infection, mice from two different animal vendors, Charles River and Jackson Laboratories, were cohoused (31, 32) . The SFB status of mice after 3 weeks of cohousing was measured via qPCR and Gram stains. The cohoused mice were then challenged with E. histolytica trophozoites (2 ϫ 10 6 ) via intracecal injection. Seven days later, E. histolytica and SFB burden were measured by qPCR. Cohousing mice transferred SFB (Fig. 1A) , and potentially many other bacteria, and provided protection against E. histolytica infection (Fig. 1B) .

Segmented filamentous bacteria specifically protect against E. histolytica and induce increased IL-17A, IL-23, neutrophils, and dendritic cells in the intestine and SAA in the blood. As cohousing of mice both transferred SFB and was protective against E. histolytica infection, we sought to determine if SFB specifically provided protection against the ameba. We directly colonized CBA/J mice from Jackson Laboratories with SFB by orogastric gavage with SFB-monoassociated feces resuspended in phosphate-buffered saline (PBS), provided by the Yakult Central Institute for Microbiological Studies, 1 week prior to E. histolytica infection. Mice became colonized with SFB following gavage ( Fig. 2A) and were protected from infection with the ameba (Fig. 2B) . Additionally, there was increased IL-17A and IL-23 expression before ( onized mice after E. histolytica infection. We also observed increased SAA in serum in SFB-colonized mice after ameba infection (Fig. 2G ). BMDCs derived from SFB-colonized mice have an increased capacity to produce IL-23 that is partially recapitulated with SAA treatment. IL-23 is a key cytokine in the generation and maintenance of IL-17A-producing cells (33) . SAA can directly induce IL-23 from both DCs and macrophages (20, 34) and is known to upregulate an epigenetic mediator, JMJD3 (33) , that specifically increases IL-23 production in a macrophage cell line. Thus, as SFB increased frequency of intestinal DCs, IL-23, IL-17A expression, and circulating serum SAA, we examined the capacity of bone marrow DCs from SFB-infected mice to produce IL-23 in response to trophozoites. BMDCs were generated from SFB-free and SFB-infected mice and treated with trophozoites for 24 h. BMDCs and trophozoites were cocultured in RPMI with 10% fetal bovine serum (FBS) under aerobic conditions rather than in ameba culture medium under anaerobic conditions, as both DCs and trophozoites were viable after 24 h in these media; however, some cell death had occurred, as observed via trypan blue. Cytokines in the supernatants were then determined by enzyme-linked immunosorbent assay (ELISA) (IL-23; R&D Systems). BMDCs derived from mice that were colonized with SFB produced significantly more IL-23 in response to trophozoites (Fig. 3A) . These data suggested that SFB altered bone marrow cells in such a way as to favor increased IL-23 production. We hypothesized that SAA induced in the serum by intestinal SFB colonization could influence BMDCs to produce increased IL-23. Therefore, we wanted to test if the presence of SAA during the differentiation of SFB-free BMDCs might alter them in such a way that they also would produce more IL-23. Thus, SAA (10 g/l) was added for the first 4 days of bone marrow DC culture. Addition of SAA to BMDC culture from SFB-free mice partially recapitulated the increased IL-23 seen in BMDCs derived from SFB ϩ mice (Fig. 3B ). This suggested that SAA might be at least partially responsible for alteration of bone marrow DCs in this model. BMDCs from SFB ؉ but not SFB ؊ mice migrated to the intestine. To test if the increase in DCs in the gut of SFB ϩ mice was due to a direct effect on the bone marrow, BMDCs from SFBcolonized and SFB-free mice were adoptively transferred to SFB Ϫ mice. For in vivo transfers, day 6 BMDC from SFB Ϫ and SFB ϩ mice were matured with LPS (1 g/ml), and 16 h later, cells were stained with carboxyfluorescein succinimidyl ester (CFSE; 10 nM), and 5 ϫ 10 5 cells were administered via intraperitoneal injection. CFSE ϩ BMDC were detected in the lamina propria but only from SFB ϩ mice (Fig. 4) . The ability of SFB ϩ BMDCs to home to the gut may help explain the increased numbers of DCs seen in the lamina propria of SFB-colonized mice during E. histolytica infection (Fig. 2E) . These results were consistent with other studies that have shown that adoptive transfer of LPS-matured BMDCs results in DC trafficking to the intestine (28, 29, 35) .

Adoptive transfer of BMDCs from SFB colonized mice was sufficient to provide protection against infection in an IL-17Adependent manner. We hypothesized that BMDCs from SFBcolonized mice may provide protection against amebiasis and that the mechanism of this protection might be via downstream induction of IL-17A. To explore this idea, we adoptively transferred BMDCs (5 ϫ 10 5 cells) from SFB Ϫ or SFB ϩ mice to SFB Ϫ mice prior to E. histolytica infection. Indeed, we found that BMDCs from SFB ϩ mice were sufficient both to confer protection (Fig. 5A) and to recapitulate the increase in neutrophils (Fig. 5B ) observed with SFB colonization alone. Blockade of IL-17A during transfer abrogated protection (Fig. 5C ) and neutrophil influx (Fig. 5D ) and led to a decrease in IgA induction (Fig. 5E ). We concluded that IL-17A and downstream innate and perhaps adaptive immune responses underlie the protection observed during adoptive transfer of BMDCs from SFB ϩ mice.

We have demonstrated that alteration of the microbiota via introduction of the commensal bacteria SFB impacts susceptibility to amebic infection in a murine model, likely via its impact on the mucosal immune system (36) . There was an increase in IL-17A and IL-23 expression in the intestines of SFB-colonized mice prior to and following amebic infection and a relative increase in neu- trophils after amebic infection. Our laboratory and others have shown that these effectors may provide protection against E. histolytica (14) . We also observed increased CD11c ϩ MHCII ϩ cells in the intestines of mice, suggesting that dendritic cells induced by SFB may also help mediate protection against infection. This is not surprising given that Goto et al. recently demonstrated that major histocompatibility complex class II (MHC-II)-dependent presentation of SFB antigens by intestinal DCs is crucial for Th17 cell induction (37) . Interestingly, protection in our model appeared to involve not only the mucosal immune response in the gut but also the bone marrow. Resistance to amebic colitis was adoptively transferred to SFB-free mice with bone marrow-derived dendritic cells (BM-DCs) from SFB-colonized mice. This suggested that SFB alteration of bone marrow cells might underlie protection against amebic colitis. BMDCs derived from SFB colonized mice produced more IL-23 in response to E. histolytica trophozoites than BMDCs from mice that were not colonized with the bacteria. However, this response was not specific to E. histolytica, as increased IL-23 production from SFB ϩ BMDCs was also seen with LPS and SAA treatment (data not shown). This suggests that SFB colonization of the intestine has conditioned bone marrow cells to be more responsive to stimulation with pathogen-associated molecular patterns (PAMPs) or damage-associated molecular patterns (DAMPs), such as SAA, or, alternatively, that there were shifts in the populations of dendritic-cell precursors present in the bone marrow prior to BMDC culture in SFB-colonized mice.

Therefore, in this model of SFB-induced alteration of the microbiome, it is not yet clear how SFB altered bone marrow cells. Future studies will be required to examine the possibility of shifts in dendritic-cell precursors or changes in responsiveness to DAMPS and PAMPS. In this context, intestinal inflammation or colonization with commensal organisms has been shown to alter populations of bone marrow progenitor cells (25, 26, 38) . However, exactly how cross talk occurs between the intestine and bone marrow is not well understood. Recent work has shown that gut microbiota metabolism of dietary fiber can increase the concentration of circulating short-chain fatty acids (SCFAs) and that treatment of mice with the SCFA propionate led to alterations in bone marrow hematopoiesis (39) . A similar mechanism might underlie protection in our model of E. histolytica/SFB coinfection. However, SFB-mediated alteration of the metabolome has not been explored. SFB do, however, have an intimate association with the gut intestinal mucosa, likely due to their limited metabolic capability (40, 41) . These bacteria are thus known to alter host gene expression in the intestine (41) . SAA has been shown to be upregulated in the intestines of mice colonized with SFB (20) , and we have demonstrated that SAA is also increased in the serum of E. histolytica/SFB-infected mice. SAA is known to be produced by many cell types, including epithelial cells (3), and SFB tightly associate with the intestinal mucosa (4). SAA-dependent induction of Jmjd3-mediated epigenetic regulation of inflammatory cytokine gene expression has also been described (33) , including IL-23, in SAA-stimulated macrophages. It is thus not hard to imagine that circulating SAA might act on bone marrow cells to alter IL-23 induction via epigenetic mechanisms (33) . Indeed, we cultured BMDCs from SFB Ϫ mice in the presence of SAA for the first 4 days of culture and then removed the mediator, and the resulting differentiated BMDCs produced significantly more IL-23 than BMDCs from SFB Ϫ mice alone. However, this did not fully recapitulate the increase in IL-23 production seen in BMDCs from SFB-colonized mice. This study is not conclusive, however, and multiple mechanisms, perhaps including alteration of metabolic by-products, may underlie the increased IL-23 production seen in BMDCs from SFBcolonized mice. We propose a model in which intestinal colonization with SFB leads to induction of systemically circulating me-diators, perhaps including SAA, that alter bone marrow dendriticcell precursors so that they provide protection against E. histolytica via induction of downstream effectors, which may include IL-17A, neutrophils, and IgA (Fig. 6) .

In conclusion, we have demonstrated that colonization with a specific component of the intestinal microbiota, SFB, provided protection against infection with the protozoan pathogen E. histolytica and that BMDCs and downstream IL-17A induction recapitulated this protection. This work therefore suggests that alteration of the microbiome can mediate resistance to intestinal infection via extraintestinal effects on bone marrow cells. Future studies will be required to describe and understand the underlying mechanisms of these microbiome-induced changes in bone marrow dendritic-cell precursors.

Male 5-week-old CBA/J mice (Jackson Laboratories, Charles River) were housed in a specific-pathogen-free facility in micro-isolator cages and provided autoclaved food (Lab Diet 5010) and water ad libitum. All procedures were approved by the Institutional Animal Care and Use Committee of the University of Virginia.

E. histolytica culture and intracecal injection. Animal-passaged HM1:IMSS E. histolytica trophozoites were cultured from cecal contents of infected mice in complete trypsin-yeast-iron (TYI-33) medium supplemented with Diamond vitamin mix (JRH Biosciences), 100 U/ml of penicillin and streptomycin, and bovine serum (Sigma-Aldrich). Prior to injection, trophozoites were grown to log phase, and 2 ϫ 10 6 parasites were suspended in 150 l culture medium and injected intracecally (42) .

SFB colonization. Mice were colonized with SFB by cohousing for 3 weeks with SFB ϩ age-matched mice or directly colonized with SFB by orogastric gavage with SFB-monoassociated feces (a kind gift from the Yakult Central Institute for Microbiological Studies) 1 week prior to E. histolytica infection. SFB-monoassociated feces were independently confirmed to be associated with SFB only, with no presence of other bacteria or fungal contamination, by both Y. Umesaki and us via qPCR, Sanger sequencing, and culture (43) . Quantitative real-time RT-PCR. SFB and E. histolytica colonization were measured by real-time PCR. For SFB colonization, qPCR with Sybr green was performed, and data were normalized to expression of a conserved Eubacteria 16S RNA gene (EUB). Primer sequences are as follows: EUB forward, 5=-ACTCCTACGGGAGGCAGCAGT-3=; EUB reverse, 5=-ATTACCGCGGCTGCTGGC-3=; SFB forward, 5=-GACGCTGAGGCAT GAGAGCAT-3=; SFB reverse, 5=-GACGGCACGGATTGTTATTCA-3=. A 6-carboxyfluorescein (FAM)-labeled probe (44), a standard curve prepared from trophozoites, and quantitative PCR (qPCR) were utilized for E. histolytica quantification; the primer and probe sequences are as follows: Eh-f, AAC AGT AAT AGT TTC TTT GGT TAG TAA AA; Eh-r, CTT AGA ATG TCA TTT CTC AAT TCA T; Eh-YYT, ATT AGT ACA AAA TGG CCA ATT CAT TCA-dark quencher; IL-17A F, 5=-ACTACCTCAA CCGTTCCACG-3=; IL-17A R, 5=-AGAATTCATGTGGTGGTCCAG-3=. Cytokines were measured via qPCR with Sybr green, and data were normalized to expression of S14: IL-23p19F, 5=-GACCCACAAGGACTCAA GGA-3=; IL-23p19 R, 5=-CATGGGGCTATCAGGGAGTA-3=; S14 F, 5=-T GGTGTCTGCCACATCTTTGCATC-3=; S14 R, 5=-AGTCACTCGGCAG ATGGTTTCCTT-3=. Melting temperatures were 60°C for EUB, SFB, IL-17A, and IL-23 and for E. histolytica (44) . Primers and probes were purchased from Integrated DNA Technologies, Coralville, IA, USA.

Serum SAA was measured by ELISA (ab157723; Abcam, Cambridge, England).

Bone marrow-derived-dendritic-cell (BMDC) culture. Bone marrow cells were harvested from 4 weeks old male CBA/J mice from Jackson Laboratories that had been treated by gavage with PBS or SFB the week prior. Cells were cultured in RPMI with 10% FBS supplemented on days 0 and 3 with granulocyte-macrophage colony-stimulating factor (GM-CSF) (10 ng/ml; Peprotech) and harvested on day 6. For in vitro experiments, 2.5 ϫ 10 5 cells were plated per well of a flat-bottom 96-well dish and treated with trophozoites for 24 h. Cytokines in the supernatants were determined by ELISA (IL-23; R&D Systems).

DC transfer. For in vivo transfers, day 6 BMDCs were stimulated with LPS (1 g/ml); 16 h later, cells were washed in PBS, and 5 ϫ 10 5 cells were administered intraperitoneally. For CFSE experiments, cells were stained with CFSE in PBS for 10 min (10 nM) and then washed twice in PBS prior to adoptive transfer.

IL-17A blockade. Four-week-old CBA/J mice were treated with 200 g rat anti-mouse IL-17A monoclonal antibody (MAb) (clone M210; Amgen) or rat IgG2a isotype control (clone 2A3; BioXcell) at 3 days before and on the day of BMDC transfer (500,000 intraperitoneally from mice receiving SFB gavage or those not receiving SFB gavage). Mice were challenged intracecally with 2 ϫ 10 6 trophozoites 7 days after BMDC transfer.

In vitro culture and flow cytometry of intestinal cells. For intracellular staining, intestinal tissue was digested in Liberase TL (0.17 mg/ml Roche) and DNase (0.5 mg/ml; Sigma) for 45 min at 37°C and processed into a single-cell suspension. A total of 1 ϫ 10 6 cells were incubated with CD16/CD32 MAb (BD Biosciences) to block Fc binding, followed by staining with CD11V-BV421, MHCII-FITC, TCR-beta-APC, B220-APC, CD45.2-APC, Ly6G-PE, Ly6C-Percp Cy 5.5, or CD11b-Pacific Blue and run on an LSR Fortessa cell analyzer (BD Biosciences). The data were analyzed with FlowJo (Tree Star Inc.).

Statistical analysis. An analysis of variance (ANOVA) followed by the Tukey-Kramer test was used for analysis of differences among multiple groups. Student's t test was used for comparisons between two groups. P values of less than 0.05 were considered significant. Results of representative experiments are shown; all experiments were replicated 2 to 4 times, with 4 to 12 individuals per group.

Supplemental material for this article may be found at http://mbio.asm.org /lookup/suppl/doi:10.1128/mBio.01817-14/-/DCSupplemental. Figure S1 , EPS file, 1.8 MB.

WHO/PAHO informal consultation on intestinal protozoal infections. World Health Organization

Update on human infections caused by intestinal protozoa

Contribution of enteric infection, altered intestinal barrier function, and maternal malnutrition to infant malnutrition in Bangladesh

Burden and aetiology of diarrhoeal disease in infants and young children in developing countries (the Global Enteric Multicenter Study, GEMS): a prospective, case-control study

The dynamic interdependence of amebiasis, innate immunity, and undernutrition. Semin. Immunopathol

Entamoeba histolytica-associated diarrheal illness is negatively associated with the growth of preschool children: evidence from a prospective study

Interactions between parasites and microbial communities in the human gut

CD4 ϩ and CD8 ϩ T cell-and IL-17-mediated protection against Entamoeba histolytica induced by a recombinant vaccine

Th17 cells upregulate polymeric Ig receptor and intestinal IgA and contribute to intestinal homeostasis

Th17 cytokines and the gut mucosal barrier

Regulation of airway MUC5AC expression by IL-1beta and IL-17A; the NF-kappaB paradigm

Cxcr2 and Cxcl5 regulate the IL-17/G-CSF axis and neutrophil homeostasis in mice

Th17 cells: biology, pathogenesis of autoimmune and inflammatory diseases, and therapeutic strategies

Resistance to intestinal Entamoeba histolytica infection is conferred by innate immunity and Gr-1ϩ cells

Amebiasis and mucosal IgA antibody against the Entamoeba histolytica adherence lectin in Bangladeshi children

Singlecell sequencing provides clues about the host interactions of segmented filamentous bacteria (SFB)

Comparative analysis of the distribution of segmented filamentous bacteria in humans, mice and chickens

Habitat, succession, attachment, and morphology of segmented, filamentous microbes indigenous to the murine gastrointestinal tract

Segmented filamentous bacteria antigens presented by intestinal dendritic cells drive mucosal Th17 cell differentiation

Induction of intestinal Th17 cells by segmented filamentous bacteria

Naturally transmitted segmented filamentous bacteria segregate with diabetes protection in nonobese diabetic mice

Gut-residing segmented filamentous bacteria drive autoimmune arthritis via T helper 17 cells

Proinflammatory T-cell responses to gut microbiota promote experimental autoimmune encephalomyelitis

Gut microbiota promote hematopoiesis to control bacterial infection

Commensal bacteria-derived signals regulate basophil hematopoiesis and allergic inflammation

Dysregulated hematopoietic stem and progenitor cell activity promotes interleukin-23-driven chronic intestinal inflammation

The intestinal metabolome: an intersection between microbiota and host

Modulation of dendritic cell differentiation in the bone marrow mediates sustained immunosuppression after polymicrobial sepsis

Distinct roles for IL-6 and IL-12p40 in mediating protection against Leishmania donovani and the expansion of IL-10 ϩ CD4 ϩ T cells

Evasion by stealth: inefficient immune activation underlies poor T cell response and severe disease in SARS-CoV-infected mice

Variation in the gut microbiota of laboratory mice is related to both genetic and environmental factors

Temporal stability of the mouse gut microbiota in relation to innate and adaptive immunity

Jmjd3-mediated epigenetic regulation of inflammatory cytokine gene expression in serum amyloid A-stimulated macrophages

Serum amyloid A activates the NLRP3 inflammasome and promotes Th17 allergic asthma in mice

Dendritic cell immunization route determines integrin expression and lymphoid and nonlymphoid tissue distribution of CD8 T cells

Human nutrition, the gut microbiome and the immune system

Segmented filamentous bacteria antigens presented by intestinal dendritic cells drive mucosal Th17 cell differentiation

Gut microbiota promote hematopoiesis to control bacterial Infection

Gut microbiota metabolism of dietary fiber influences allergic airway disease and hematopoiesis

The genome of Th17 cell-inducing segmented filamentous bacteria reveals extensive auxotrophy and adaptations to the intestinal environment

The lifestyle of the segmented filamentous bacterium: a non-culturable gut-associated immunostimulating microbe inferred by whole-genome sequencing

The mouse model of amebic colitis reveals mouse strain susceptibility to infection and exacerbation of disease by CD4 ϩ T cells

Segmented filamentous bacteria are indigenous intestinal bacteria that activate intraepithelial lymphocytes and induce MHC class II molecules and fucosyl asialo GM1 glycolipids on the small intestinal epithelial cells in the ex-germ-free mouse

Realtime-PCR assay for diagnosis of Entamoeba histolytica infection

This work was supported by NIH grant 5R01 AI026649-25 to W.P. We thank Yoshinori Umesaki, from the Yakult Central Institute for Microbial Research (Tokyo, Japan) for supplying the SFB-monocolonized feces. We also gratefully acknowledge the gift of anti-IL-17A monoclonal antibodies from Alison Budelsky at Amgen Incorporated (Seattle, WA) and Florent Ginhoux, Brian Kelsall, and Tom Braciale for helpful discussions.S.B., W.P., and M.W.-K. contributed to the design, analysis and interpretation of the data. S.B., E.B., M.C., C.C., C.N., and Z.N. conducted experiments and critiqued the work. S.B. and W.P. wrote the manuscript.