A rt i c l e s

The CD46-Jagged1 interaction is critical for human TH1 immunity

Gaëlle Le Friec1, Devon Sheppard2, Pat Whiteman3,14, Christian M Karsten4,14, Salley Al-Tilib Shamoun5,14, Adam Laing1, Laurence Bugeon6, Margaret J Dallman6, Teresa Melchionna1, Chandramouli Chillakuri3, Richard A Smith1, Christian Drouet7, Lionel Couzi8, Veronique Fremeaux-Bacchi9,10, Jörg Köhl4,11, Simon N Waddington12, James M McDonnell13, Alastair Baker5, Penny A Handford3, Susan M Lea2 & Claudia Kemper1

CD46 is a complement regulator with important roles related to the immune response. CD46 functions as a pathogen and is a potent costimulator for the induction of interferon-g (IFN-g)-secreting effector T helper type 1 (TH1) cells and their subsequent switch into interleukin 10 (IL-10)-producing regulatory T cells. Here we identified the Notch family member Jagged1 as a physiological for CD46. Furthermore, we found that CD46 regulated the expression of Notch receptors and ligands during T cell activation and that disturbance of the CD46-Notch crosstalk impeded induction of IFN-g and switching to IL-10. Notably, CD4+ T cells from CD46-deficient patients and patients with hypomorphic in the gene encoding Jagged1 () failed to mount appropriate TH1 responses in vitro and in vivo, which suggested that CD46-Jagged1 crosstalk is responsible for the recurrent infections in subpopulations of these patients.

CD46 (MCP) was initially discovered as a complement-regula- Notch–Notch ligand interactions. Given the similarities between these 1 tory , then was identified as a cell-entry receptor ‘hijacked’ two evolutionarily old systems (complement and Notch) in TH1 biology, by several viruses to promote infection2 and is now emerging as an we investigated the possibility of a functional connection between immunomodulatory molecule with vital functions in the costimu- CD46 and Notch and if this potential complement––5 lation and regulation of human T helper type 1 (TH1) cells . The system crosstalk is required for the effector function of TH1 cells. ligands for CD46 and their binding sites in CD46 have been defined We not only demonstrate here that the activation of CD46 on for the first two activities; complement-activation fragments C3b CD4+ T cells regulated the expression of Notch and its ligands but © 2012 Nature America, Inc. All rights reserved. America, Inc. © 2012 Nature and C4b bind to the complement-control protein (CCP) domains also identify Jagged1 as an additional physiological ligand for CD46. CCP2, CCP3 and CCP4 (CCP2–CCP4)1, whereas viral ligands such The Jagged1-binding site in CD46 was located in the amino-terminal as adenovirus knob proteins or measles virus hemagglutinin com- CCP domains CCP1 and CPP2, and interference with the CD46- npg monly interact with domains CCP1 and CCP2 (refs. 6,7). However, Jagged1 interaction resulted in substantially less induction of TH1 not all activities of CD46 can be explained by its interaction with cells in vitro. We obtained support for the in vivo importance of this the ligands known at present; the egg-sperm fusion event mediated protein interaction by establishing that patients with mutations in the by CD46 requires CCP1 (ref. 8), and although the intrinsic genera- genes encoding CD46 (refs. 10,11) or Jagged1 (patients with Alagille tion of T cell–derived C3b is required for CD46 stimulation of CD4+ syndrome)12 shared key features. These patients suffered recurrent T cells4, the molecular basis of CD46-mediated costimulatory activity infections, and although T cell proliferation and the effector function is unknown. Similarly, whereas the importance of the Notch system in of TH2 cells was unaffected, the in vitro induction (or regulation) of the induction of cells of the TH1 and TH2 lineages is undisputed and TH1 cells was absent or severely compromised. Mechanistically, the + key signaling events mediated by the activation of Notch on CD4 faulty induction of TH1 cells seemed to involve altered responsive- T cells have been identified9, many functional aspects of this system ness to cytokines of the interleukin 2 (IL-2) family, as all patients had in T cell biology remain unknown and cannot be explained solely by considerable deviation from normal expression of the IL-7 receptor

1Division of Transplantation Immunology and Mucosal Biology, MRC Centre for Transplantation, King’s College London, Guy’s Hospital, London, UK. 2Sir William Dunn School of Pathology, University of Oxford, Oxford, UK. 3Department of Biochemistry, University of Oxford, Oxford, UK. 4Institute for Systemic Inflammation Research, University of Lübeck, Lübeck, Germany. 5Child Health Clinical Academic Grouping, King’s Health Partners, Denmark Hill Campus, London, UK. 6Division of Cell and Molecular Biology, Department of Life Sciences, Imperial College London, London, UK. 7Université Joseph Fourier, Groupe de Recherche et d’Etude du Processus Inflammatoire–Age and Imaging Research Unit, Centre National de la Recherche Scientifique FRE3405, CHU de Grenoble, Grenoble, France. 8Nephrology- Transplantation, CHU Bordeaux, Bordeaux, France. 9Cordeliers Research Center, Inserm Unite Mixte de Recherche en Sante 872, Paris, France. 10Hopital Europeen Georges Pompidou, Service d’Immunologie Biologique, Assistance Publique–Hopitaux de Paris, Paris, France. 11Division of Cellular and Molecular Immunology, Cincinnati Children’s Hospital Medical Center and University of Cincinnati College of Medicine, Cincinnati, Ohio, USA. 12Institute for Women’s Health, Gene Transfer Technology Group, University College London, London, UK. 13Randall Division of Cell & Molecular Biophysics, King’s College London, UK. 14These authors contributed equally to this work. Correspondence should be addressed to C.K. ([email protected]).

Received 15 August; accepted 19 September; published online 21 October 2012; doi:10.1038/ni.2454

nature immunology VOLUME 13 NUMBER 12 DECEMBER 2012 1213 A rt i c l e s

a CD46 immobilized b Jagged1 immobilized c WT RBCs + CD46-TG RBCs + CD46-TG RBCs + 3.0 0.6 J-1(DSL-EGF3) J-1(DSL-EGF3) BSA *** 180 10 ) 2.5 ** 42.8 ) 0.5 ) 450 450 2.0 * 2 *** 0.4 7.5 0 39.7 0.02 1.5 0.3 90 Isotype 5.0 1.0 *** 0.2 Events CD46-TG

Binding ( A 2.5 FSC ( × 10 0.5 Binding ( A 0.1 RBC 0 0 0 0 100 101 102 103 104 100 101 102 103 104 C3b BSA sCR1 BSA rhuJ-1 sCD46 CD46 Bound J-1(DSL-EGF3)

N-1(11–13) N-1(11–13) J-1(DSL-EGF3)

Figure 1 Jagged1 is a ligand for CD46. (a,b) Enzyme-linked immunosorbent assay of the interaction of immobilized soluble CD46 (CCP1–CCP4) with the soluble proteins C3b, recombinant human Jagged1 (rhuJ-1), J-1(DSL-EGF3), N-1(11–13) or bovine serum albumin (BSA; a) or of immobilized J-1(DSL-EGF3) with the soluble proteins N-1(11–13), CD46, (sCR1) or BSA in Ca2+ buffer (b; protein details, Supplementary Fig. 1). Results are presented as absorbance at 450 nm (A450). *P < 0.05, **P < 0.005 and ***P < 0.001, versus BSA binding (Student’s one-tailed t-test and Bonferroni correction for multiple comparisons). (c) Expression of CD46 on the surface of RBCs from CD46-transgenic mice (far left); number above bracketed line indicates percent CD46+ cells. Isotype, istoype-matched control antibody. Right, binding of biotinylated J-1(DSL-EGF3) (middle) or BSA (far right) to CD46 on RBCs from wild-type (WT) and CD46-transgenic (CD46-TG) mice, measured with fluorochrome-labeled streptavidin. Numbers above outlined areas indicate percent cells with bound J-1(DSL-EGF3). FSC, forward scatter. Data are representative of five experiments (a,b; mean and s.d.) or are from one representative of three independent experiments (c).

α-chain (CD127) or the common γ-chain (CD132) or both. Notably, To further characterize the CD46-Jagged1 interaction, we measured patient-derived T cells with defective TH1 induction in vitro were the binding affinity of J-1(DSL-EGF3) and truncated CD46 constructs also unable to undergo such induction in vivo when assessed in a by surface plasmon resonance and mapped the Jagged1-binding site humanized mouse graft-versus-host disease model. Our data identify on CD46 by nuclear magnetic resonance spectroscopy. The binding a functional connection between the complement and Notch systems of J-1(DSL-EGF3) to CD46 constructs composed of CCP1–CCP4, that is critical for the induction and regulation of human TH1 cells in CCP1–CCP3 and CCP1–CCP2 all had similar interaction affinity infection and homeostasis of the immune system. and kinetics, but J-1(DSL-EGF3) did not bind to the closely related mouse complement-regulatory molecule Crry16 (Fig. 2a–d). These RESULTS experiments established that the two amino-terminal domains of Jagged1 binds to CCP1 and CCP2 of CD46 CD46 were fully able to bind Jagged1. The dissociation constant (Kd) Because several of the biological activities of CD46 cannot be explained for the CD46–J-1(DSL-EGF3) interaction was about 8 µM, within by its interaction with C3b or C4b, the existence of another physiologi- the less-tight affinity range of 1 µM noted for the interaction between cal ligand has long been suspected. Using an initial screen based on an CD46 and C3b (C. Harris, personal communication) but tighter enzyme-linked immunosorbent assay with recombinant soluble CD46 than the assumed Kd for interactions between soluble N-1(11–13) (Supplementary Fig. 1a) and available recombinant proteins of the and J-1(DSL-EGF3), for which protein concentrations in excess of Notch receptor and Notch ligand families (Supplementary Fig. 1b,c), 50 µM have been required to observe such an interaction and the © 2012 Nature America, Inc. All rights reserved. America, Inc. © 2012 Nature we identified Jagged1 as a CD46-binding protein. CD46 bound to C3b interaction is not saturated with protein concentrations up to 160 µM (positive control), full-length Jagged1 and a Jagged1 construct contain- (ref. 13). The use of purified, soluble proteins to characterize interac- ing the Notch1-binding site (the disulfide-rich Delta–Serrate–Lag-2 tions has demonstrated that most cell-surface proteins interact with npg (DSL) domain and the first three epidermal growth factor (EGF)-like each other with Kd values in the micromolar range and concomitant domains (J-1(DSL-EGF3)))13 (Supplementary Fig. 1c) but not to fast on and off rates. This has often been interpreted as facilitating Notch1 or a recombinant soluble construct composed of Notch1 EGF- the fine-tuning of interactions by the avidity effects either due to like domains 11–13 containing the Jagged1-binding site (N-1(11–13)); the sharing of many weak interactions across two interacting cells or Supplementary Fig. 1b) or other proteins of the Notch family (Fig. 1a increases in avidity resulting from the need for cell-surface molecules and data not shown). Conversely, Jagged1 bound CD46 but not soluble on the same cell diffuse in only two dimensions rather than three recombinant complement receptor 1, which shared with CD46 the dimensions to find interacting partners. This means that even appar- ability to bind C3b or C4b (Fig. 1b). We also did not observe binding ently transient interactions will occur in vivo more frequently and for of Jagged1 to factor H or C4bp, two fluid-phase complement regulators a longer duration than solution studies indicate, and it suggests that that contain C3b- or C4b-binding sites (data not shown). Therefore, similar characteristics would be important in the interaction between the interaction between CD46 and Jagged1 was specific. Furthermore, CD46 and Jagged1. our data obtained with immobilized CD46 or J-1(DSL-EGF3) indi- We mapped the interaction site on CD46 by nuclear magnetic reso- cated that the CD46- and Notch1-binding sites in Jagged1 were in that nance chemical-shift perturbation. We did backbone assignment of same region (Fig. 1a,b). We confirmed the binding of Jagged1 to CD46 the construct of the two amino-terminal domains of CD46 by stand- expressed on the surface of red blood cells (RBCs) from mice with ard triple-resonance methods17. The Jagged1-binding site mapped to transgenic expression of human CD46 (CD46-transgenic mice). These an interaction surface composed of CCP1 and CCP2 (Fig. 2e–g). The mice express CD46 on all cells14, whereas wild-type mice lack CD46 Jagged1-interaction surface was distinct from that for C3b and C4b expression on somatic cells15. Notably, RBCs also lack expression of (which use CD46 domains CCP2–CCP4) but was in the same two Notch receptors and Notch ligands. With this system, we observed domains used by measles virus and adenovirus serotypes 11, 21 and 35 that J-1(DSL-EGF3) bound to RBCs from CD46-transgenic mice but (refs. 6,7; Supplementary Fig. 1d). These data identified Jagged1 as a not those from wild-type mice (Fig. 1c), which confirmed that physi- physiological and not pathogen-derived ligand for CD46 and mapped ological, cell-expressed CD46 was able to bind Jagged1. the binding site of Jagged1 to CCP1 and CCP2 of CD46.

1214 VOLUME 13 NUMBER 12 DECEMBER 2012 nature immunology A rt i c l e s

Figure 2 Jagged1 binds to CCP1 and CCP2 a b c of CD46. (a) Surface plasmon resonance CD46(1–4) CD46(1–3) CD46(1–2) 50 of the binding of J-1(DSL-EGF3) to CD46 120 120 100 10 µM 100 containing CCP1–CCP4 (CD46(1–4)) coupled 40 100 31 µM 80 100 80 29 µM 80 60 80 60 on the surface of the chip with a Kd of ~8 µM 30 5 µM 16 µM 40 15 µM 40 60 60 (normalized by subtraction of the mock-coupled 20 8 µM 7 µM 20 2.5 µM 20 K = 6.8 µM Binding (RU) K = 8.4 µM 40 4 µM d 40 4 µM d channel). (b,c) Binding of CD46 containing 10 2 M 0 2 M 0 1.25 µM µ Response units (RU) 0 10 20 30 40 µ 0 5 10 15 20 25 30 35 1 µM 20 CCP1–CCP3 (CD46(1–3); b) or the two amino- 20 Concentration (µM) Concentration (µM) 0 0 terminal domains of CD46 (CD46(1–2); c) 0 –10 –20 Binding to J-1(DSL-EGF3) (RU) Binding to Jagged1 surface (RU) 0 0 50 100 150 200 250 300 Binding to Jagged1 surface (RU) 0 50 100 150 200 250 300 to immobilized J-1(DSL-EGF3), normalized 50 100 150 200 250 300 350 400 450 500 as in a. Inset, equilibrium values of binding Time (s) Time (s) Time (s) and Kd fit. (d) Binding of CD46 containing CCP1–CCP4 (21 µM), CCP1–CCP3 (25 µM) or d e 9.5 9.0 8.5 8.0 7.5 7.0 f the two amino-terminal domains (19 µM) after 1.6 110 110 1.4 injection over immobilized J-1(DSL-EGF3); 115 115 Domain 1 Domain 2 1.2 Crry containing CCP1–CCP4 (20 M) serves as µ 1.0 120 120 0.8 a negative control. Results are normalized for N (p.p.m.) 0.06 construct molecular weight. Below, values. 0.6 15 125 125 Kd 0.4 0.05 (e) Nuclear magnetic resonance spectroscopy 0.2 130 130 Binding to Jagged1 0.04 1 15 0 overlay of the H, N-heteronuclear single- 9.5 9.0 8.5 8.0 7.5 7.0 ∆δ 1 0.03 CD46 CD46 Crry H (p.p.m.) quantum coherence (HSQC) of CD46 containing 1–3) CD46 (1–4) ( (1–2) 0.02 the two amino-terminal domains (black), K d M M M 121.0 V26 0.01 showing the chemical-shift perturbation after 121.5 9.4 µ 1.9) 6.2 µ 8.0 µ I115 0 (± (±2.2) (±3.5) 122.0 N94 A79 0 20 40 60 10080120 140 the addition of unlabeled J-1(DSL-EGF3) R62

N (p.p.m.) 122.5 C93 Residue (green) at a molar ratio of 0.6. Bottom, enlargement of area outlined above. 15 123.0 (f) Chemical-shift perturbation by residue for those unambiguously assigned 8.9 8.8 8.7 8.6 8.5 1 and baseline-resolved residues in e. (g) Surface structure of CCP1 and CCP2 H (p.p.m.) ( accession code, 3O8E), showing residues with unambiguous assignment g (dark gray) and chemical-shift perturbation of >0.15 p.p.m. (red) after the addition of J-1(DSL-

EGF3). Blue, N-linked glycosylation sites. Data are from one experiment with six replicates of two CCP1 independent sets of proteins (a), twelve replicates of three independent sets of proteins (b) or nine replicates of two independent sets of proteins (c), are representative of six experiments (d; mean ± CCP2 s.d. of triplicates) or are from one representative of three independent experiments (e,f). CCP3

CD46 regulates expression of Notch and Notch ligand CCP4 Although CD46 was initially discovered as a complement-regulatory molecule, it transmits intracellular signals after engagement at the cell surface and can modulate the function of several immunologically transcription was lower at 36 h in cells activated with immobilized competent cell types18. For example, CD46 is a T cell costimulator3 anti-CD3 and anti-CD46 and was even less than the transcription and regulates the production of interferon-γ (IFN-γ) and IL-10 by in nonactivated cells. Similarly, crosslinking of CD46 led to rapid © 2012 Nature America, Inc. All rights reserved. America, Inc. © 2012 Nature human TH1 cells, as follows: coengagement of the coreceptor CD3 loss of CD46 itself from the cell surface (Supplementary Fig. 2a). and CD46 in the presence of a low concentration of IL-2 induces Downregulation of CD46 after stimulation has been observed widely, + − 18 proinflammatory TH1 cells (IFN-γ IL-10 ), whereas the presence but its biological importance is unknown . We detected NOTCH3 npg of high environmental concentration of IL-2 initiates coexpres- and NOTCH4 transcripts, as well as DLL3 and DLL4 transcripts, in sion of IFN-γ and IL-10 with a switch to a regulatory resting cells, but this remained unaltered by any antibody-mediated (IFN-γ+IL-10+) and, finally, a shutdown of IFN-γ expression activation conditions tested (data not shown). We also observed all (IFN-γ−IL-10+)4,18,19. This CD46- and IL-2-driven (self-)regula- these changes at the level of cell surface–expressed proteins (Fig. 3d). tory pathway is defective in T cells from patients with rheumatoid These results demonstrated that activation of CD46 induced a specific arthritis, which fail to switch4. Similarly, signaling events initiated pattern of expression of Notch receptors and ligands on CD4+ T cells by the members of the family of human Notch receptors and ligands that was typified by higher expression of Notch1 and Notch2, as well (Notch1–Notch4, Jagged1 and Jagged2, and Delta-like 1 (DLL1), as Jagged1 and Jagged2, but loss of DLL1 and CD46. DLL3 and DLL4; Supplementary Fig. 1b,c) have a crucial role in 9,20 induction of the TH1-TH2 lineage and, specifically, in the coex- CD46-Notch crosstalk is vital for TH1 induction in vitro 21 pression of IFN-γ and IL-10 by TH1 cells . Consequently, we next must be processed successively by the metallopro- determined whether the CD46-Jagged1 interaction was important in tease ADAM10 or ADAM17 and then the presenilin–γ-secretase the regulation of IFN-γ and IL-10 in TH1 cells. complex to become signaling competent, and the contributions We first measured the expression of CD46, Jagged1 and other mem- of Notch-mediated signals have been assessed with inhibitors of bers of the Notch family on resting T cells and on T cells that had ADAM and γ-secretase9,20. That approach cannot be used to address been activated with immobilized antibodies to CD3 (anti-CD3) and whether CD46-Notch crosstalk is a requisite for the regulation of TH1 anti-CD46. The coactivation of CD46 by antibody crosslinking rap- cells, as activation of and signaling by CD46 on CD4+ T cells also idly increased expression of the genes encoding Notch1 (NOTCH1), required processing by ADAM and γ-secretase22 (Supplementary Notch2 (NOTCH2), Jagged1 (JAG1) and Jagged2 (JAG2), similar to Fig. 2a,b). CD46 exists in four isoforms that arise through splicing activation with anti-CD3 alone or with anti-CD3 and anti-CD28 (data of a glycosylated extracellular region and the two possible intracel- not shown), and this expression remained high in CD46-induced lular tails, CYT1 and CYT2 (ref. 1; Supplementary Fig. 1a). Human TH1 cells and IL-10-switched TH1 cells (Fig. 3a–c). In contrast, DLL1 Jurkat T cells stably transfected to express the CD46-CYT1 isoform

nature immunology VOLUME 13 NUMBER 12 DECEMBER 2012 1215 A rt i c l e s

Figure 3 CD46 regulates the expression of Notch receptors and ligands 4 c a 10 8 on human CD4+ T cells. (a) Expression of IFN-γ and IL-10 by human + 103 CD4 T cells activated for 36 h with anti-CD3 and anti-CD46 plus IL-2. 6 NS + Outlined areas indicate three distinct subpopulations: IFN-γ (top left), 102 *** IFN-γ+IL-10+ (top right) and IL-10+ (bottom right). (b,c) Expression of 4 NS 101 and mRNA (b) and , and mRNA (c) NS NOTCH1 NOTCH2 JAG1 JAG2 DLL1 2 IFN- γ 100 by nonactivated T cells (NA), T cells activated for 2 h with anti-D3 (α- 0 1 2 3 4

10 10 10 10 10 JAG1 mRNA (relative) 0 CD3) and anti-CD46 (plus 50 U/ml recombinant human IL-2) and T cells IL-10 of the subpopulations in a; results are presented relative to 18s mRNA b 6 15 * expression. (d) Expression of Notch1, Jagged1 and DLL1 protein on * non-subsorted bulk CD4+ T cells after stimulation for 36 h with anti-CD3 * 4 and anti-CD46. MFI, mean fluorescence intensity. NS, not significant; 10 *** *P < 0.05, **P < 0 0.005 and ***P < 0.001, versus nonactivated *** *** 2 NS cells (Student’s one-tailed t-test and Bonferroni correction for multiple 5 **

comparisons). Data are representative of four experiments (a–d; mean JAG2 mRNA (relative)

and s.d. in b,c). NOTCH1 mRNA (relative) 0 0 5 3 NS NS 4 * 2 (Jurkat-BC1 cells) produce IL-10 after activation with anti-CD3 and 3 NS * 2 anti-CD46, whereas untransfected Jurkat cells or Jurkat cells trans- 1 fected to express CD46-CYT2 (Jurkat-BC2) are unable to express 1 *** **

DLL1 mRNA (relative) ***

IL-10 (ref. 4). To investigate whether the CD46-CYT1–mediated NOTCH2 mRNA (relative) 0 0 + + + + + + NA α-CD3 IFN-γ IFN-γ IL-10 NA α-CD3 IFN-γ IFN-γ IL-10 stimulation involves subsequent Notch1 signaling, we disrupted + + α-CD46 IL-10 α-CD46 IL-10 Notch1 signaling through the use of an inhibitory monoclonal anti- (2 h) α-CD3 + α-CD46 (2 h) α-CD3 + α-CD46 body or soluble N-1(11–13) to compete with cell surface–expressed (36 h) (36 h) 200 Notch1 for Notch ligands. Each treatment abrogated the CD46- d MFI 2 MFI 1 MFI 3 Isotype MFI 10 MFI 5 MFI 7 NA ­mediated production of IFN-γ and diminished IL-10 secretion by MFI 48 MFI 8 MFI 70 MFI 50 MFI 9 MFI 78 α-CD3 100 >50% in Jurkat-BC1 cells (Supplementary Fig. 2c,d), which indicated MFI 80 MFI 11 MFI 19 α-CD3 + that CD46 and Notch signaling pathways indeed intersected in the Events α-CD28 α-CD3 + production of TH1 cytokines and switching to IL-10. 0 α-CD46 Consistent with that hypothesis, the addition of soluble CD46, 100 101 102 103 104 Notch1 Jagged1 DLL1 DLL1 or Jagged1 also resulted in much less switching of purified CD4+ T cells from IFN-γ to IL-10, mediated by anti-CD3, anti-CD3 and anti-CD28, or anti-CD3 and anti-CD46, in each case in the ­downregulation of CD46 was accompanied by 50% less production of presence of IL-2 (Fig. 4a,b). This was probably due to interfer- IFN-γ and IL-10 (Supplementary Fig. 4e), whereas the proliferation ence with temporally regulated changes in interactions of members and viability of cells were unaffected (data not shown). Although of the CD46 and Notch system and, thus, signaling events during these data suggested that α-E-catenin participated in CD46- T cell activation. Super-resolution imaging experiments demonstrated mediated signaling events in CD4+ T cells, we cannot exclude the © 2012 Nature America, Inc. All rights reserved. America, Inc. © 2012 Nature that >95% of Jagged1 on resting CD4+ T cells localized together with ­possibility that the changes in cytokine production were secondary CD46, whereas we observed only negligible interactions between to inapparent additional effects of the knockdown of α-E-catenin. Jagged1 and Notch1 (Fig. 4c). After activation with anti-CD3 and We were not able to inhibit the observed downregulation of DLL1 npg anti-CD46, which induced CD46 downregulation and the ‘release’ during T cell activation without treating CD4+ T cells with an inhibi- of Jagged1, a substantial proportion (>50%) of Notch1 molecules tor of ADAMs. However, we noted that the CD46-mediated down- were bound to Jagged1 (Fig. 4c). These data suggested that our affin- regulation of DLL1 was less efficient than on Jurkat-BC1 cells than on ity measurements obtained with limited recombinant fragments primary CD4+ T cells (Supplementary Fig. 4f ), which might explain of CD46, Notch1 and Jagged1 (Fig. 2) extended to intact proteins why Jurkat-BC1 cells produce relatively small amounts of IL-10. In expressed on the cell surface. Further support for the possibility of agreement with that, transfection of Jurkat-BC1 cells with short hair- regulated successive crosstalk of the CD46 and Notch system dur- pin RNA targeting DLL1 mRNA resulted in lower DLL1 expression in ing the induction of TH1 cells (model, Supplementary Fig. 3) was activated Jurkat-BC1 cells with proportionally greater IL-10 produc- provided by the observation that activation of CD46 alone, without tion (Supplementary Fig. 4f). direct antibody-mediated engagement of Notch, induced consider- In sum, these data suggested that the presence of CD46 on T cell able transcription of the Notch target gene HES1 and transcription surfaces restricted interactions of Notch1 with Jagged1. They also of the gene encoding the Notch signaling mediator RBPJ20 (Fig. 4d). indicated that engagement of CD46 during T cell activation led to Conversely, inadequate downregulation of CD46 hinders TH1 induc- α-E-catenin-dependent downregulation of CD46 and α-E-catenin- tion. A member of the E-cadherin network, α-E-catenin, binds to the independent downregulation of DLL1. Disturbance in this spatially cytoplasmic portion of CD46 in human intestinal epithelial cells23. and temporally regulated crosstalk between complement and Notch We found that α-E-catenin also interacted with CD46 in primary proteins led to deregulated TH1 responses in vitro. human CD4+ T cells (Supplementary Fig. 4a), and knockdown of α-E-catenin protein (Supplementary Fig. 4b) impaired the down- CD46 deficiency causes defective TH1 function in vivo regulation of CD46 without having an effect on the expression of Rodents (mice, rats and guinea pigs) lack CD46 expression on somatic Notch, Jagged1 or DLL1 protein (Supplementary Fig. 4c) or that tissues, and this restricted expression pattern impedes direct evaluation of additional molecules vital for T cell activation, such as CD3, of the in vivo importance of the CD46-Jagged1 interaction with a small CD25, CD28 or CD69 (Supplementary Fig. 4d). Inhibition of the animal model15. Although the mouse Crry protein compensates for

1216 VOLUME 13 NUMBER 12 DECEMBER 2012 nature immunology A rt i c l e s

Figure 4 Undisturbed crosstalk by the CD46 a Med sDLL1 sJ-1(DSL-EGF3) sCD46 b 5 and Notch system is required for normal *** 4.1 1.1 4.0 0.9 1.0 0.2 24.1 1.5 4 switching of human TH1 cells from IFN-γ to α-CD3 1.1 1.0 0.7 0.9 IL-10. (a) Expression of IFN- and IL-10 by 3 γ ** CD4+ T cells activated with anti-CD3 alone 2 ** (top), anti-CD3 and anti-CD28 (middle) or IFN- γ /IL-10 1 anti-CD3 and anti-CD46 (bottom) in the 5.7 1.5 7.8 1.7 5.7 1.2 25.2 1.6 α-CD3 + 1.6 1.6 1.5 1.9 0 presence of recombinant human IL-2 (25 U/ml) α-CD28 Med and treated with medium alone (Med) or soluble sDLL1 sCD46 (s) DLL1, J-1(DSL-EGF3) or CD46. Numbers in 104 plots indicate subpopulations (as in Fig. 3a). 7.2 9.7 14.6 14.3 14.9 13.8 80.9 3.0 sJ-1(DSL-EGF3) 103 7.3 4.4 4.6 0.2 α-CD3 + α-CD46 (b) Ratio of IFN-γ to IL-10 secreted into the α-CD3 + 2 α-CD46 10 media of cells treated as in a, bottom. 1 10 d 50 0 (c) Super-resolution confocal microscopy and IFN- γ 10 ** 0 1 2 3 4 three-dimensional analysis of nonactivated 10 10 10 10 10 40 IL-10 * T cells and T cells activated with anti-CD3 30 and anti-CD46, stained with anti-CD46, c CD46 Jagged1 CD46-Jagged1 anti-Notch1 or anti-Jagged1 to assess molecular 20 * * colocalization (white areas, far right). Outlined 10

area (bottom left), staining for CD46. Original HES1 mRNA (relative) 0 magnification, ×600. (d) Quantitative PCR + + + NA analysis of the transcription of HES1 and RBPJ -CD46 (2 h) IFN-γ + IL-10 IL-10 mRNA in purified CD4+ T cells (as in Fig. 3b,c). NA γ Notch1 Jagged1 Notch1-Jagged1 US IFN- *P < 0.05, **P < 0.005 and ***P < 0.001 α-CD3 α (Student’s one-tailed t-test and Bonferroni α-CD3 + α-CD46 (36 h) correction for multiple comparisons), versus 14 * medium alone (b) or nonactivated cells (d). 12 ** * Data are representative of six experiments 10 (a,b,d; mean and s.d.) or two independent 8 experiments (c). Notch1 Jagged1 Notch1-Jagged1 6 ** CD46 4 2 the complement-regulatory function of CD46, α-CD3 + RBPJ mRNA (relative) 16 0 it does not regulate TH1 responses . For that α-CD46 + + + + NA ­reason, we obtained CD4 T cells from patients (2 h) IFN-γ + IL-10 IL-10

with mutations in either CD46 or JAG1 and IFN- α-CD3 α-CD46 γ assessed their ability to mount TH1 responses α-CD3 + α-CD46 in vitro and in vivo. Mutations in CD46 that (36 h) affect protein expression or complement- regulatory function cause atypical hemolytic uremic syndrome24. At an IFN-γ- and IL-10-coexpressing phenotype, and these patients thus present, fewer than ten patients worldwide have been identified as having lacked a major T cell population key to the prevention and clearance © 2012 Nature America, Inc. All rights reserved. America, Inc. © 2012 Nature homozygous mutations in CD46 but, notably, over 50% of those patients of infections caused by intracellular pathogens25 (Fig. 5b). In contrast, have common variable immunodeficiency and recurrent chest infec- T cells from patient CD46-2 produced normal amounts of IFN-γ and tions, which indicates that CD46 mutations can indeed induce immuno­ IL-10 after activation with anti-CD3 or with anti-CD3 plus anti-CD28 npg logical defects10,11. The following three patients with homozygous (although they lacked the usual CD46-mediated increase in IFN-γ pro- CD46 mutations participated in this study (Fig. 5a): patient CD46-1, duction and switch to IL-10). The reason for this is unclear. Notably, who has a splice-site alteration between 1 and 2 that leads to B cells from patient CD46-1 are fully functional26, but that has not been only 10% of normal CD46 expression on peripheral blood mononu- confirmed for patients CD46-2 and CD46-3. clear cells10; patient CD46-2, who has a similar splice-site alteration As predicted, lack of CD46 resulted in altered regulation of the that causes aberrant mRNA transcripts and loss of CD46 expression expression of Notch1 and Jagged1, as nonactivated T cells from on >90% of peripheral blood mononuclear cells11 and normal expres- patients CD46-2 and CD46-3 had higher expression of Jagged1 than sion on the remaining 10% of those cells; and patient CD46-3, with two did those from healthy donors (Table 1 and Supplementary Fig. 5a) mutations in 2 and absence of detectable cell-surface expression of but then failed to upregulate the expression of Jagged1 protein after CD46 (Supplementary Table 1 and Supplementary Fig. 5a). Although activation with anti-CD3 and anti-CD46. Similarly, T cells from hospital records for patient CD46-1 are unavailable, patients CD46-2 patient CD46-3 were defective in activation-induced upregulation of and CD46-3 have suffered confirmed recurrent infections, have been Notch1, whereas T cells from patient CD46-2 overexpressed Notch1 diagnosed with common variable immunodeficiency and are being after activation (data on Notch1 and Jagged1 expression not avail- treated with intravenous immunoglobulin infusion. All three patients able for patient CD46-1; Table 1 and Supplementary Fig. 5a). We have normal numbers of B cells and CD4+ and CD8+ T cells10,11 (data also assessed the expression of additional key cell-surface markers + not shown), and activation of their purified CD4 T cells showed that required for normal TH1 responses on resting and activated T cells they proliferated at normal rates (data not shown) and mounted strong from healthy donors and the patients with CD46 mutations. We found TH2 responses (Fig. 5b); this suggested that an intrinsic -derived no substantial differences among those cells in the expression and reg- defect in induction of the T cell lineage was unlikely. However, activa- ulation of CD3, CD11a (α-chain of LFA-1), CD28, CD69, CD122 and tion with anti-CD3, with anti-CD3 plus anti-CD28 or with anti-CD3 CCR7 (data not shown). Although all three patients showed a trend and anti-CD46 did not induce IFN-γ secretion in cells from patients toward less upregulation of CD25 and downregulation of CD62L CD46-1 and CD46-3; consequently, the cells also failed to switch to after activation with anti-CD3 or with anti-CD3 plus anti-CD28,

nature immunology VOLUME 13 NUMBER 12 DECEMBER 2012 1217 A rt i c l e s

Figure 5 T cells from CD46-deficient patients have defective in vitro TH1 a CCP TM induction. (a) Location of CD46 mutations (‘d’; bottom) in CD46-deficient SP domains STP region ? domain CYT1 CYT2 patients CD46-1, CD46-2 and CD46-3 (including exon structure); above, 1 2 3 4 ST ST ST Protein A B C corresponding protein domains of CD46. SP, signal peptide; STP (STA,

STB, STC), serine-threonine-proline–rich regions; ?, region of unknown Exons 1 2 3 4 5 6 7 8 9 10 11 12 13 14 function; TM, transmembrane; CYT1 and CYT2, cytoplasmic tail. (b) Secretion of cytokines by CD4+ T cells purified from freshly drawn dCD46-1 dCD46-3 blood samples from two healthy donors (HD3 and HD4; representative dCD46-2 of 12 age- and sex-matched donors) and patients CD46-1, CD46-2 and HD3 HD4 CD46-1 CD46-2 CD46-3 HD3 HD4 CD46-1 CD46-2 CD46-3 CD46-3 (above plots), then left nonactivated or activated for 36 h with b 35 20 various combinations of immobilized anti-CD3, anti-CD28 and anti-CD46 30 25 15 (horizontal axes) in the presence of recombinant human IL-2 (25 U/ml). 20 10 TNF, tumor-necrosis factor; ND, not detectable. Data are representative of 15 three experiments with duplicate samples (mean). 10 5 IL-10 (ng/ml) IFN- γ (ng/ml) 5 0 ND ND ND 0 ND 10 80 + − + as well as a small memory T cell pool (CD4 CD45RA CD45RO 8 60 6 cells), these results were within the range of normal donor varia- 40 tion (data not shown). We discovered, however, that T cells from 4 20 IL-4 (pg/ml) TNF (ng/ml) 2 all patients showed considerable deregulation of CD127 and CD132, ND ND ND ND ND ND NDND 27 0 0

which together form the IL-7 receptor : T cells from patients lacked NA NA NA NA NA 0.5 the CD127 downregulation usually induced by activation with anti- α -CD3 α -CD3 α -CD3 α -CD3 α -CD3 0.4 CD3 plus anti-CD46, whereas CD132 was overexpressed on resting 0.3

or activated T cells (Table 1 and Supplementary Fig. 5a). α -CD3 + -CD28 α -CD3 + -CD46 α -CD3 + -CD28 α -CD3 + -CD46 α -CD3 + -CD28 α -CD3 + -CD46 α -CD3 + -CD28 α -CD3 + -CD46 α -CD3 + -CD28 α -CD3 + -CD46 0.2

The observed defect in in vitro TH1 induction of T cells lacking IL-5 (pg/ml) 0.1 normal CD46 expression also extended to an in vivo model of graft- 0 ND ND ND 28 NA NA NA NA NA

versus-host disease . We activated T cells from three healthy donors α -CD3 α -CD3 α -CD3 α -CD3 α -CD3 or patient CD46-1 with anti-CD3 and anti-CD28 and then injected those cells into β -microglobulin-deficient mice of the nonobese

2 α -CD3 + -CD28 α -CD3 + -CD46 α -CD3 + -CD28 α -CD3 + -CD46 α -CD3 + -CD28 α -CD3 + -CD46 α -CD3 + -CD28 α -CD3 + -CD46 α -CD3 + -CD28 α -CD3 + -CD46 diabetic–severe combined immunodeficiency strain and monitored engraftment of human T cells by the presence of human CD45RA+ cells in blood and human IFN-γ in the serum. The engraftment of Alagille syndrome causes defective TH1 function in vivo T cells from patient CD46-1 and healthy donors was similar. In con- Complete deficiency in members of the Notch family has not been trast, and in line with the data generated in vitro, we detected human described so far; this is probably because the Notch system has a cen- IFN-γ only in serum from mice injected with T cells from healthy tral role in cell-cell communication during tissue morphogenesis and donors (Supplementary Fig. 5b,c). Furthermore, whereas T cells organ development29. Heterozygous mutations of JAG1 are inherited from healthy donors induced graft-versus-host disease, as indicated in an autosomal dominant way and cause Alagille syndrome, which by weight loss (Supplementary Fig. 5d) and immunohistochemi- is characterized by developmental problems that affect mainly the © 2012 Nature America, Inc. All rights reserved. America, Inc. © 2012 Nature cal analysis of intestinal tissue (data not shown), mice injected with , , eye and skeleton12,30. Although it is known that ~25% of T cells from patient CD46-1 developed no disease. Blood samples patients with Alagille syndrome also suffer recurrent infections of from patients CD46-2 and CD46-3 could not be obtained for this the ears and respiratory tract31, the immunological defects underly- npg experiment. Thus, CD46-mediated signaling events were required ing these infections are entirely unexplored. Given the discovery of not only for the switch of TH1 cells into an IL-10-coexpressing phe- an interaction between CD46 and Jagged1 and the lower or absent 4 + notype but also for the initial induction of a TH1 phenotype in CD4 TH1 responses in CD46-deficient patients, we hypothesized that T cells in vivo. Furthermore, CD46 participated in the regulation of the recurrent infections in patients with Alagille syndrome are also the expression of CD127 and CD132 on TH1 cells but was not required rooted in altered function of TH1 cells caused by deregulated cross- for the normal effector function or proliferation of TH2 cells. talk of the CD46-Notch system.

Table 1 Expression of surface markers on CD4+ T cells from CD46-deficient patients Jagged1 Notch1 CD46 CD127 CD132 NA CD3,CD28 CD3,CD46 NA CD3,CD28 CD3,CD46 NA CD3,CD28 CD3,CD46 NA CD3,CD28 CD3,CD46 NA CD3,CD28 CD3,CD46 HD3 + ↑ ↑ + ↑ ↑ + ↔ ↓ + ↓ ↓ + ↑ ↑ HD4 + ↑ ↑ + ↑ ↑ + ↔ ↓ + ↓ ↓ + ↑ ↑ CD46-1 NDA NDA NDA NDA NDA NDA 10% of ↔ ↓ + ( ) ↓ ↔* + ( ) ↑ ↔† normal CD46-2 + ( ) ↔ ↔† + On 10% ↔ ↓ + ↔ + ( ) ↔ ↔† of cells CD46-3 + ( ) ↔ ↔† + ↔ ↔‡ − − − + ↓ ↔* + ( )* ( )* Expression of surface markers on CD4+ T cells from healthy donors and CD46-deficient patients, left nonactivated or activated with anti-CD3 and anti-CD28 (CD3,CD28) or with anti-CD3 and anti-CD46 (CD3,CD46): +, present on resting T cells; −, not present on resting (or activated) T cells; ↑, higher expression after activation; ↓, lower expression after activation; ↔, no change in expression; (↑) and (↑↑), higher baseline expression(≤250% and >250%, respectively) than that of cells from healthy donors; ↑ and ↑↑, more upregu- lation (≤250% and >250%, respectively) on activated cells than that on cells from healthy donors; ↔, no change in expression relative to that of nonactivated cells. NDA, no data available. Data are representative of three experiments. *Expression remains higher than that of cells from healthy donors. †Expression similar to that of activated T cells from healthy donors. ‡Expression remains lower than that of cells from healthy donors. Raw data, Supplementary Figure 5.

1218 VOLUME 13 NUMBER 12 DECEMBER 2012 nature immunology A rt i c l e s

Figure 6 T cells from patients with Alagille syndrome have defective a DSL CR TM PDZ SP ? domain EGF-like repeats domain domain domain in vitro TH1 induction. (a) Location of JAG1 mutations in patients AP1– AP4 with Alagille syndrome (presented as in Fig. 5a). CR, cysteine-rich Protein 1 2 3 4 5 6 7 8 9 10 11 12 1314 1516 region; PDZ, conserved scaffolding structural domain. (b) Secretion of 11 19 cytokines by CD4+ T cells purified from freshly drawn blood samples from Exons 1 2 3 4 5 6 7 8 9 10 12 13 14 15 16 17 18 20 21 222324 25 26 two healthy donors (HD1 and HD2) and patients AP1–AP4 (above plots), AP4 AP2 then left nonactivated or activated for 36 h with various combinations of AP3 AP1

immobilized anti-CD3, anti-CD28 and anti-CD46 (horizontal axes) in the HD1 HD2 AP1 AP2 AP3 AP4 HD1 HD2 AP1 AP2 AP3 AP4 presence of recombinant human IL-2 (25 U/ml). Data are representative of b 20 8 four experiments with duplicate samples (mean). 15 6 10 4 5 2 IFN- γ (ng/ml) We studied four patients with Alagille syndrome (AP1–AP4) who IL-10 (ng/ml) ND ND ND ND ND ND ND ND had mutations in exon 3, 4, 18 or 19 of JAG1 and suffered recur- 0 0 rent and persistent otitis media and chest infections (Fig. 6a and 5 120 4 100 80 Supplementary Table 2). Because Notch–Notch ligand interactions 3 + + 60 participate in development of the CD4 and CD8 T cell lineage in 2 40

9 IL-4 (pg/ml) the thymus , we first assessed the composition of peripheral blood TNF (ng/ml) 1 20 0 ND ND ND 0 ND ND ND ND mononuclear cell populations and found no deviation in the fre- 6

NA

NA

NA

NA

NA

NA 1.6

-CD3 α

-CD3 α

-CD3 α

-CD3 α

-CD3 α quency of key lymphocyte subpopulations in samples from patients -CD3 α -CD46 α 1.4 32 1.2 AP1–AP4 relative to those from healthy donors (Supplementary 1.0 0.8

-CD4 α + -CD3 α

-CD46 α + -CD3 α

-CD46 α + -CD3 α

-CD46 α + -CD3 α

-CD46 α + -CD3 α Table 2). However, when we compared the expression of TH1 and TH2 + -CD3 α 0.6 + 0.4 cytokines by activated purified CD4 T cells from patients AP1–AP4 IL-5 (ng/ml) 0.2 and their counterparts from healthy donors, we observed a profile 0 ND ND ND ND NA NA NA NA NA NA α -CD3 α -CD3 α -CD3 α -CD3 α -CD3 reminiscent of that of CD46-deficient patients (Fig. 6b). Cell pro- α -CD3 liferation and secretion of tumor-necrosis factor was normal for all patients, but T cells from patients AP1 and AP2 produced no IFN-γ α -CD3 + -CD46 α -CD3 + -CD46 α -CD3 + -CD46 α -CD3 + -CD46 α -CD3 + -CD46 α -CD3 + -CD46 (and did not switch to IL-10 production) after activation with anti- CD3 alone or with anti-CD3 plus anti-CD28 (date not shown) or activation with anti-CD3 and anti-CD46, whereas patients AP3 and ­efficiently downregulate CD46 after activation with anti-CD3 and AP4 had much less TH1 induction (≤50%). T cells from patients AP3 anti-CD46 (Table 2 and Supplementary Fig. 6a). and AP4 also had notably enhanced TH2 responses (Fig. 6b). Similar to T cells from patient CD46-1, T cells from patients AP1 Although basal Notch1 expression and upregulation of Notch1 and AP3 were also unable to induce TH1 responses in vivo or cause expression after activation was unaffected, patients AP1–AP4 had graft-versus-host disease when injected into IL-2 receptor (IL-2R) unexpectedly higher expression of Jagged1 protein on resting T cells γ-chain–deficient mice of the nonobese diabetic–severe combined (Table 2 and Supplementary Fig. 6a). In contrast to results obtained immunodeficiency strain (Supplementary Fig. 6b–d). However, this for the CD46-deficient patients, however, Jagged1 was further upregu- failure to cause graft-versus-host disease might have been the result of lated after activation (Table 2 and Supplementary Fig. 6a). Whereas poor engraftment, because in contrast to T cells from patient CD46- © 2012 Nature America, Inc. All rights reserved. America, Inc. © 2012 Nature the expression and regulation of CD3, CD11a, CD25, CD28, CD69, 1, which engrafted at a ‘reasonable’ rate (Supplementary Fig. 5b), CD12 and CCR7 was also normal in patients AP1–AP4 (data not T cells from patients with Alagille syndrome failed to engraft in the shown), we found substantial deviation (similar to that observed mice. Notch1 signaling on human CD4+ T cells has been shown to npg for the CD46-deficient patients) in the regulation of CD127 and regulate the adhesion, migration and chemotaxis of these cells via CD132 by T cells from the patients with Alagille syndrome. Cells modulation of GTPases of the Rho family33. Thus, defects in Notch from patients AP1 and AP4 completely lacked downregulation of system signaling may affect not only TH1 cytokine production in CD127 after activation with anti-CD3 and anti-CD46, and whereas all patients with Alagille syndrome but also the homing ability of their patients had more CD132 on resting T cells than did healthy donors, T cells. In summary, patients with Alagille syndrome with recurrent after activation with anti-CD3 and anti-CD46, T cells from patients infections had a T effector cell phenotype similar to that of CD46- AP1, AP3 and AP4 upregulated CD132 expression well beyond the ­deficient patients, characterized by defective induction of TH1 cells normal expression achieved by such activation. Furthermore, we and deregulation of the expression of CD127 and CD132 but unaf- observed that T cells from patients AP2 and AP3 were unable to fected function of TH2 cells.

DISCUSSION Table 2 Expression of surface markers on CD4+ T cells from patients with Alagille syndrome Here we have identified Jagged1 as a physi- Jagged1 Notch1 CD46 CD127 CD132 ological ligand for CD46 and have dem- NA CD3,CD46 NA CD3,CD46 NA CD3,CD46 NA CD3,CD46 NA CD3,CD46 onstrated that coordinated CD46-Jagged1 HD1 + + + + + ↑ ↑ ↓ ↓ ↑ crosstalk was required for T 1 responses. HD2 + ↑ + ↑ + ↓ + ↓ + ↑ H AP1 + ( ) ↑ ( ) + ↑ + ↓ + + ( ) ( ) Activation of the Notch system is control- AP2 + ( ) ↑ ( ) + ↑ + (50%) + ↓ + ( ) led by spatial and temporal restriction of AP3 + ( ) ↑ ( ) + ( ) ↑ + + ↓ + ( ) the availability of receptors and ligands dur- AP4 + ( ) + ( ) ↑ + ↓ + ( ) + ( ) * ing cell-cell interactions9,29, and our study Expression of surface markers on CD4+ T cells from healthy donors and patients with Alagille syndrome, left nonac- has suggested that CD46 participates in tivated or activated with anti-CD3 and anti-CD46: symbols as in Table 1; (↓), lower baseline expression than that of cells from healthy donors. Data are representative of four experiments. this process. We propose a model in which *Expression remains higher than that of cells from healthy donors (raw data, Supplementary Fig. 6). CD46 sequesters Jagged1 on resting T cells,

nature immunology VOLUME 13 NUMBER 12 DECEMBER 2012 1219 A rt i c l e s

thereby limiting interactions between Jagged1 and Notch1 and favor- cytokine family other than IL-7. IL-2-mediated signaling through the ing Notch1 and DLL1 cis interactions, which inhibit T cell activa- high-affinity receptor for IL-2 (CD25, CD122 and CD132) is needed 34 27 tion . Thus, similar to DLL1 expression, in the absence of antigen for cell-activation induction of TH1 responses , and chemical inhibi- or danger signals, CD46 expression on T cells may function as the tion of Notch1 signaling impedes normal IL-2R expression and TH1 ‘brake’. After engagement of the T cell antigen receptor, the CD46 induction40. IL-2 is also linked to CD46 as follows: CD46 negatively ligand C3b (ref. 34) is generated locally. Binding of C3b to CD46 ini- regulates IL-2 expression but also integrates IL-2R signals for IL-10 4,5 tiates CD46-mediated signaling events, including the migration and and IFN-γ coexpression in TH1 cells . We therefore speculate that cluster formation of T cells35, downregulation of CD46 and DLL1 and T cells from CD46-deficient patients and patients with Alagille syn- maintenance of the surface availability of Notch1 and Jagged1. This drome may be unable to induce TH1 responses, at least in part because change in surface expression of CD46 and Notch proteins releases of aberrant IL-2R signaling. the brake and allows orchestrated Notch1 and DLL1 interactions in The immunomodulatory function of CD46 is probably one reason trans (that generate IFN-γ9), as well as binding of Notch1 and Jagged1 for the use of CD46 as a receptor by several human pathogens. CD46- in cis or trans (necessary for IL-10 induction9). The role of IL-2 and interacting viruses target CCP1 and CCP2 of CD46 (refs. 2,6,7), which potential functions of generated soluble CD46 and members of the contain the Jagged1-binding site. Structures of CCP1 and CCP2 in Notch family remain to be integrated into this model. In support complex with viral proteins that bind CD46 have demonstrated sub- of this model is our observation that both CD46-deficient patients stantial reorientation of these two CCPs relative to each other, which and patients with JAG1 mutations who suffer recurrent infections suggests that their arrangement is highly ligand specific. The interac- did not generate normal TH1 responses. Both patient groups also tion surfaces identified here for the binding of Jagged1 to CD46 sug- shared additional key features in their T cell phenotype as additional gested that a Jagged1-specific conformation of CCP1-2 was required evidence that an overlapping CD46 and Notch pathway is affected. for binding and indicated that Jagged1-bound CD46 would not be able First, in line with published observations that CD46-mediated sig- to bind viral ligands simultaneously without displacement of Jagged1 19 nals are specifically needed for TH1 induction , TH2 responses and from CD46. Hence, the observation that the binding of adenovirus tumor-necrosis factor were induced. Moreover, both patient groups serotype 35 to CD46 on human CD4+ T cells induced downregulation showed a trend toward exaggerated TH2 responses, which may of CD46 but less production of IL-2 and IFN-γ could have been due explain why patients with Alagille syndrome also suffer a greater to interference with the coordinated CD46-Notch system signaling 41,42 prevalence of TH2-driven conditions, including otitis media, asthma events during T cells activation . and eczema12,25. Furthermore, and consistent with our model, CD4+ Our observations may provide a platform from which to advance T cells from C3-deficient patients (which cannot produce the CD46 understanding of the complex signaling networks that underlie the ligand C3b locally) were also unable to assume a TH1 phenotype and biology of TH1 cells as well as differences in the human and mouse had deregulated IL-2R expression but produced large amounts of TH2 systems in the induction and regulation of TH1 responses. Future analy­ cytokines (data not shown). sis of the effect of the binding of virus to CD46 and its effect on the The mutations in patients AP2 and AP3 are predicted to lead to interaction of Jagged1 with CD46 may also provide new insights into nonsense-mediated decay of JAG1 mRNA that results in the expres- how CD46-binding pathogens may interfere with the CD46-Jagged1– sion of only wild-type Jagged1 on the cell surface. Furthermore, mediated normal TH1 induction to foster infection. Furthermore, the expression studies of additional Alagille syndrome–associated mis- identification of the surface expression of CD46 as a ‘stop signal’ offers sense mutations in cell lines have shown that they led to retention of the following hypothesis to explain the counterintuitive downregula- © 2012 Nature America, Inc. All rights reserved. America, Inc. © 2012 Nature the mutant protein in the endoplasmic reticulum (data not shown), tion of CD46 on most activated cell types: it provides a ‘go’ signal which suggests that is the pathogenic mechanism when immune activation is apparent5,23. Finally, as the Notch system that operates in most cases. The patients with Alagille syndrome also has fundamental roles in tissue morphogenesis and renewal, we npg studied here unexpectedly had higher Jagged1 expression on rest- anticipate that the CD46-Jagged1 interaction may be important in ing T cells; we have no explanation for this observation at present. these biological processes as well. Nonetheless, each patient with Alagille syndrome and CD46-deficient patient had distinct deviations in the expression of Jagged1, Notch1 or Methods CD46 or a combination thereof. However, the most notable phenotype Methods and any associated references are available in the online in terms of cell-surface receptors involved in TH1 biology was the version of the paper. considerable deregulation of CD127 and CD132 (which together form the receptor for IL-7) on T cells from each patient group. Notably, not Reagent requests should be addressed to P.A.H. (penny.handford@ only is IL-7 required for T cell homeostasis and the enhancement of bioch.ox.ac.uk) or S.M.L. ([email protected]). 36 TH1 and TH17 responses but the gene encoding CD127 has also been identified as a strong risk not linked to the major histo- Note: Supplementary information is available in the online version of the paper. compatibility complex for the T cell–driven disease multiple sclero- Acknowledgments sis37,38. Similarly, deregulation of the expression of CD46 isoforms has We thank the CD46-deficient patients and patients with Alagille syndrome for been connected with the progression of multiple sclerosis39. Future their support, and A. Hayday for data discussions. Supported by the Medical studies should assess whether T cells from patients with CD46 muta- Research Council (G1002165 to C.K.), the European Union Framework Programme 7 (Innovative Medicines Initiative “Be The Cure” project with C.K. as tions have an altered responses to IL-7. However, CD132 is also an lead researcher), the Medical Research Council Centre for Transplantation (Guy’s essential component of the receptors for IL-2, IL-4, IL-7, IL-9, IL-15 Hospital, King’s College), the Department of Health, the National Institute for and IL-21 (the IL-2 receptor family)27. Most members of this family Health Research Biomedical Research Centre (for Guy’s & St. Thomas’ National are involved in the normal function of T cells and B cells as well as Health Service Foundation Trust in partnership with King’s College London and King’s College Hospital National Health Service Foundation Trust), the Wellcome natural killer cells, and CD132 deficiency is linked to X-linked severe Trust (097928/A/08/Z to S.M.L. and P.A.H.), the German Research Foundation 27 combined immunodeficiency . Deregulation of CD132 would there- (GRK1727 TP8 and SFB/TR22 A21 to J.K.) and the European Research Council fore also affect the responsiveness of T cells to members of the IL-2 (‘SomaBio’ to S.N.W.).

1220 VOLUME 13 NUMBER 12 DECEMBER 2012 nature immunology A rt i c l e s

AUTHOR CONTRIBUTIONS 15. Tsujimura, A. et al. Molecular cloning of a murine homologue of membrane cofactor A.B. and P.A.H. contributed equally to this work. G.L.F. designed and did protein (CD46): preferential expression in testicular germ cells. Biochem. J. 330, experiments and wrote the manuscript; D.S. did surface plasmon resonance and 163–168 (1998). nuclear magnetic resonance spectroscopy experiments; P.W. and C.C. generated 16. Fernández-Centeno, E., de Ojeda, G., Rojo, J.M. & Portoles, P. Crry/p65, a membrane complement regulatory protein, has costimulatory properties on mouse T cells. recombinant Notch and Jagged1 proteins; C.M.K. and J.K. did the super-resolution J. Immunol. 164, 4533–4542 (2000). microscope studies and edited the manuscript; S.A.-T.S., A.B., C.D., L.C. and 17. Grzesiek, S. & Bax, A. type determination in the sequential assignment V.F.-B. provided blood samples from patients and discussed the data; A.L. did the procedure of uniformly 13C/15N-enriched proteins. J. Biomol. NMR 3, 185–204 graft-versus-host disease experiments and discussed data; L.B. and M.J.D. designed (1993). the RT-PCR experiments and discussed the data; T.M. and R.A.S. generated soluble 18. Kemper, C. & Atkinson, J.P. T-cell regulation: with complements from innate CD46 and soluble complement receptor 1 and discussed data; S.N.W. provided immunity. Nat. Rev. Immunol. 7, 9–18 (2007). + mice with transgenic expression of human CD46 and edited the paper; J.M.M. did 19. Kemper, C. et al. Activation of human CD4 cells with CD3 and CD46 induces a surface plasmon resonance experiments and edited the paper; P.A.H. provided T-regulatory cell 1 phenotype. Nature 421, 388–392 (2003). 20. Kopan, R. & Ilagan, M.X. The canonical : unfolding the recombinant Notch and Jagged proteins, designed experiments and edited the activation mechanism. Cell 137, 216–233 (2009). paper; S.M.L. designed the surface plasmon resonance and nuclear magnetic 21. Rutz, S. et al. Notch regulates IL-10 production by T helper 1 cells. Proc. Natl. resonance experiments, provided recombinant CD46 proteins and edited the Acad. Sci. USA 105, 3497–3502 (2008). manuscript; and C.K. conceived of and designed the study, did experiments and 22. Ni Choileain, S. et al. The dynamic processing of CD46 intracellular domains provides edited the manuscript. a molecular rheostat for T cell activation. PLoS ONE 6, e16287 (2011). 23. Cardone, J., Al-Shouli, S. & Kemper, C. A novel role for CD46 in wound repair in. Front Immun. 2, 28 (2011). COMPETING FINANCIAL INTERESTS 24. Fang, C.J. et al. Membrane cofactor protein mutations in atypical hemolytic uremic The authors declare no competing financial interests. syndrome (aHUS), fatal Stx-HUS, C3 glomerulonephritis, and the HELLP syndrome. Blood 111, 624–632 (2008). 25. Romagnani, S. Th1/Th2 cells. Inflamm. Bowel Dis. 5, 285–294 (1999). Published online at http://www.nature.com/doifinder/10.1038/ni.2454. 26. Fuchs, A., Atkinson, J.P., Fremeaux-Bacchi, V. & Kemper, C. CD46-induced human Reprints and permissions information is available online at http://www.nature.com/ Treg enhance B-cell responses. Eur. J. Immunol. 39, 3097–3109 (2009). reprints/index.html. 27. Liao, W., Lin, J.X. & Leonard, W.J. IL-2 family cytokines: new insights into the complex roles of IL-2 as a broad regulator of T helper cell differentiation. Curr. Opin. Immunol. 23, 598–604 (2011). 1. Liszewski, M.K., Post, T.W. & Atkinson, J.P. Membrane cofactor protein (MCP or 28. Nervi, B. et al. Factors affecting human T cell engraftment, trafficking, and CD46): newest member of the regulators of complement activation gene cluster. associated xenogeneic graft-vs-host disease in NOD/SCID beta2mnull mice. Exp. Annu. Rev. Immunol. 9, 431–455 (1991). Hematol. 35, 1823–1838 (2007). 2. Cattaneo, R. Four viruses, two bacteria, and one receptor: membrane cofactor 29. Liu, J., Sato, C., Cerletti, M. & Wagers, A. Notch signaling in the regulation of stem protein (CD46) as pathogens’ magnet. J. Virol. 78, 4385–4388 (2004). cell self-renewal and differentiation. Curr. Top. Dev. Biol. 92, 367–409 (2010). 3. Astier, A., Trescol-Biemont, M.C., Azocar, O., Lamouille, B. & Rabourdin-Combe, 30. Oda, T. et al. Mutations in the human Jagged1 gene are responsible for Alagille C. Cutting edge: CD46, a new costimulatory molecule for T cells, that induces syndrome. Nat. Genet. 16, 235–242 (1997). p120CBL and LAT phosphorylation. J. Immunol. 164, 6091–6095 (2000). 31. Quiros-Tejeira, R.E. et al. Variable morbidity in Alagille syndrome: a review of 43 4. Cardone, J. et al. Complement regulator CD46 temporally regulates cytokine production cases. J. Pediatr. Gastroenterol. Nutr. 29, 431–437 (1999). by conventional and unconventional T cells. Nat. Immunol. 11, 862–871 (2010). 32. Robinson, M. et al. An analysis of the normal ranges of lymphocyte subpopulations 5. Cope, A., Le Friec, G., Cardone, J. & Kemper, C. The Th1 life cycle: molecular in children aged 5–13 years. Eur. J. Pediatr. 155, 535–539 (1996). control of IFN-γ to IL-10 switching. Trends Immunol. 32, 278–286 (2011). 33. Bhavsar, P.J., Infante, E., Khwaja, A. & Ridley, A.J. Analysis of Rho GTPase 6. Casasnovas, J.M., Larvie, M. & Stehle, T. Crystal structure of two CD46 domains reveals expression in T-ALL identifies RhoU as a target for Notch involved in T-ALL cell an extended measles virus-binding surface. EMBO J. 18, 2911–2922 (1999). migration. Oncogene (2012). 7. Arnberg, N. Adenovirus receptors: implications for tropism, treatment and targeting. 34. del Álamo, D., Rouault, H. & Schweisguth, F. Mechanism and significance of Rev. Med. Virol. 19, 165–178 (2009). cis-inhibition in Notch signalling. Curr. Biol. 21, R40–R47 (2011). 8. Riley, R.C., Tannenbaum, P.L., Abbott, D.H. & Atkinson, J.P. Cutting edge: inhibiting 35. Alford, S.K., Longmore, G.D., Stenson, W.F. & Kemper, C. CD46-induced measles virus infection but promoting reproduction: an explanation for splicing and immunomodulatory CD4+ T cells express the adhesion molecule and chemokine tissue-specific expression of CD46. J. Immunol. 169, 5405–5409 (2002). receptor pattern of intestinal T cells. J. Immunol. 181, 2544–2555 (2008).

© 2012 Nature America, Inc. All rights reserved. America, Inc. © 2012 Nature 9. Amsen, D., Antov, A. & Flavell, R.A. The different faces of Notch in T-helper-cell 36. Bikker, A., Hack, C.E., Lafeber, F.P. & van Roon, J.A. Interleukin-7: a key mediator differentiation. Nat. Rev. Immunol. 9, 116–124 (2009). in T cell-driven autoimmunity, inflammation, and tissue destruction. Curr. Pharm. 10. Fremeaux-Bacchi, V. et al. Genetic and functional analyses of membrane cofactor Des. 18, 2347–2356 (2012). protein (CD46) mutations in atypical hemolytic uremic syndrome. J. Am. Soc. 37. Gregory, S.G. et al. Interleukin 7 receptor α chain (IL7R) shows allelic and functional Nephrol. 17, 2017–2025 (2006). association with multiple sclerosis. Nat. Genet. 39, 1083–1091 (2007). npg 11. Couzi, L. et al. Inherited deficiency of membrane cofactor protein expression and 38. Hafler, D.A. et al. Risk alleles for multiple sclerosis identified by a genomewide varying manifestations of recurrent atypical hemolytic uremic syndrome in a sibling study. N. Engl. J. Med. 357, 851–862 (2007). pair. Am. J. Dis. 52, e5–e9 (2008). 39. Astier, A.L., Meiffren, G., Freeman, S. & Hafler, D.A. Alterations in CD46-mediated 12. Vajro, P., Ferrante, L. & Paolella, G. Alagille syndrome: An overview. Clin. Res. Tr1 regulatory T cells in patients with multiple sclerosis. J. Clin. Invest. 116, Hepatol. Gastroenterol. 36, 275–277 (2012). 3252–3257 (2006). 13. Cordle, J. et al. A conserved face of the Jagged/Serrate DSL domain is involved in 40. Adler, S.H. et al. Notch signaling augments T cell responsiveness by enhancing Notch trans-activation and cis-inhibition. Nat. Struct. Mol. Biol. 15, 849–857 CD25 expression. J. Immunol. 171, 2896–2903 (2003). (2008). 41. Adams, W.C. et al. Attenuation of CD4+ T-cell function by human adenovirus type 14. Greig, J.A. et al. Influence of coagulation factor x on in vitro and in vivogene 35 is mediated by the knob protein. J. Gen. Virol. 93, 1339–1344 (2012). delivery by adenovirus (Ad) 5, Ad35, and chimeric Ad5/Ad35 vectors. Mol. Ther. 42. Adams, W.C. et al. Adenovirus type-35 vectors block human CD4+ T-cell activation 17, 1683–1691 (2009). via CD46 ligation. Proc. Natl. Acad. Sci. USA 108, 7499–7504 (2011).

nature immunology VOLUME 13 NUMBER 12 DECEMBER 2012 1221 ONLINE METHODS anti-CD62L (559772), anti-CD69 (555530), anti-CD127 (557938), anti-CD132 Healthy donors and patients. Purified T cells were obtained from buffy coats (555900) and anti-mouse CD45RB (16A) were all from BD Biosciences. The (National Blood Service) or blood samples from healthy volunteers. Informed broad-spectrum matrix metalloproteinase inhibitor TAPI-2 was from Merck consent was obtained from all subjects, and blood was collected and proc- Chemicals; marimastat was from Tocris Bioscience; and the presenilin and essed with the approval of and in accordance with the King’s College Ethics γ-secretase inhibitor L-685,458 was from Sigma-Aldrich. Committee guidelines (06/Q0705/20). Adult patients with CD46 deficiencies were recruited in France under appropriate institutional guidelines; two cases Enzyme-linked immunosorbent assay. After 96-well microplates were coated have been described10,11. Six Caucasian children between 2 and 12 years of age overnight at 4 °C with 5 µg/ml of protein (soluble CD46, human C3b, J-1(DSL- were recruited, of whom four were diagnosed with Alagille syndrome and two EGF3) or recombinant CR1), they were blocked with 1% BSA and then incub­ were healthy (Review Board of National Research Ethics Services Committee ated for 1.5 h at 37 °C with protein samples diluted to a concentration of London: 09/H0711/38). Patients with Alagille syndrome all had repeated infec- 0.5 µg/ml (and roughly equimolar amounts) in 4% BSA, 0.005% Tween20, tions and/or allergies and food intolerances. None of the patients were taking 0.25% NP-40, 20 mM HEPES and 10 mM CaCl, pH 7.4. Because recombinant immunosuppressants or had undergone liver transplantation. Blood samples mouse C3b and C4b are commercially unavailable, 5% mouse serum was used were processed within a maximum of 3 h from the time of collection. as source for mouse C3b and C4b. Bound proteins were detected with the appropriate primary mAbs, followed by horseradish peroxidase–linked sec- Mice and graft-versus-host disease model. CD46-transgenic mice were gen- ondary antibodies or streptavidin and subsequent visualization with OPD erated by backcrossing an established CD46-transgenic line onto outbred MF1 substrate (O-phenylenediamine dihydrochloride; Sigma-Aldrich). mice14. Mice were handled and samples were obtained and processed under UK Home Office license 70/6906. Mice of the nonobese diabetic–severe combined Surface plasmon resonance. All data were collected with a Biacore T100 (GE immunodeficiency strain that were deficient in β2-microglobulin (Taconic Healthcare) with Jagged1 or CD46 immobilized through primary amine- Farms) or IL-2R common γ chain (Charles River) were used for the injection of ­coupling to the surface of the Biacore CM5 Chip. CD46 or J-1(DSL-EGF3) T cells from patient CD46-1 or from patients with Alagille syndrome, respec- constructs were passed over the chip in a solution of 10 mM HEPES (pH 7.4), tively, and were maintained under pathogen-specific sterile conditions. Graft- 150 mM NaCl, 3 mM EDTA and 0.005% surfactant P20. Multiple titrations versus-host disease was induced as described28. Peripheral blood mononuclear were done over a concentration range of 0.1 µM to 40 µM, with flow rates from cells from healthy donors, CD46-deficient patients or patients with Alagille 25 µl/min to 40 µl/min at 25 °C. Data were processed with the manufacturer’s syndrome were activated for 72 h with immobilized monoclonal antibody BIAevaluation software and were fit with a Langmuir 1:1 equilibrium model (mAb) to CD3 and mAb to CD28 before adoptive transfer via injection into or, where possible, kinetic analysis with simultaneous fits of the on and off the tail vein (1 × 107 T cells: 80–85% CD4+ and 15–20% CD8+). Engraftment rates (SigmaPlot). of human cells was monitored by counting of human CD45+ cells (ratio of human CD45+ cells to mouse CD45+ cells) and measurement of human IFN-γ Nuclear magnetic resonance spectroscopy. A sample containing 65 µM of a in mouse blood at various time points. Body weight was monitored and mice construct of the two amino-terminal domains of CD46 uniformly enriched in 15 were culled when they reached the humane end point of a decrease of 15% in N in 25 mM sodium acetate (pH 5.5) and 5% D2O was used for collection of body weight. Disease was further confirmed by immunohistological analysis sensitivity-enhanced 1H,15N-HSQC48 on a 500-MHz Bruker Avance (Bruker of intestinal tissue. U) equipped with a cryoprobe. Another 1H,15N-HSQC was collected with the addition of 38 µM unlabeled Jagged1 BirA (as described13). In addition, T cell isolation and activation. T cells were isolated and activated as described4. the 1H,15N-HSQC was repeated with a sample of unlabeled Jagged1 without 1 15 The human embryonic kidney HEK293T and Jurkat cell lines were cultured the BirA tag (data not shown). For the H, N-HSQC, acquisition times for t1 according to the manufacturer’s protocol (American Type Culture Collection). were 42 ms with 256 complex data points. Data were processed and analyzed Jurkat cells (including those stably transfected to express either CD46-CYT1 with NMRPipe NMR data-processing software and Sparky NMR spectra- 4

© 2012 Nature America, Inc. All rights reserved. America, Inc. © 2012 Nature (Jurkat-BC1) or CD46-CYT2 (Jurkat-BC2)) were activated as described for ­display software. Chemical-shift perturbation values were calculated with the purified CD4+ T cells but for 5 d with IL-2 supplementation every 2 d. following equation:

2 Recombinant proteins. Serum-purified C3b was from Complement  g  ∆d = ∆ d 2 + N ∆d npg ()H  N  Technologies, and recombinant complement receptor 1 (CR1) was produced as g H  published43. Recombinant human DLL1 and Jagged1 and recombinant mouse Jagged1 were from R&D Systems. N-1(11–13) and (J-1(DSL-EGF3)) with Assignments were done on [U-15N, 13C, 1H] the construct of the two carboxy-terminal biotinylation were produced as described13. CD46 constructs amino-terminal domains of CD46 with standard triple resonance correla- (Adprotech) were subcloned into the pET14b vector and were transformed tion experiments. HNCA/HN(CO)CA, HNCO/HN(CA)CO and HNCACB/ into B834 cells. Labeled proteins were produced as described44 and proteins HNCA(CO)CB experiments were done as described49. were refolded by an established protocol45. Recombinant Crry (containing CCP1–CCP4) was generated as described46. High-resolution microscopy and co-localization analysis. T cells were treated and then were stained for 25 min at 4 °C with anti-CD46, anti-Jagged1 or anti- Antibodies and inhibitors. The following cell-stimulating monoclonal anti- Notch1. Cells were then mounted with Fluoromount-G (SouthernBiotech). bodies were used: anti–human CD28 (CD28.2), anti–mouse CD3 (145-2C11), Images were obtained by confocal fluorescence microscopy with a laser- anti–mouse CD28 (37.51), anti-CD3 purified from a specific hybridoma line ­scanning microscope (Fluorview 1000; Olymbus) with a 60× oil objective with (OKT-3; all from from BD Biosciences); and anti-CD46 (TRA-2-10; generated a numerical aperture of 1.35. For three-dimensional image analysis, z-stacks in house)47. Expression and binding of Notch1 was assessed with mAb 527425 were obtained with an interval of 0.1 µm with the confocal microscope with (R&D Systems) or mAb A6 (Thermo Fisher Scientific). Human Jagged1 was 20–30 slices per stack to visualize cells in their full extension. Stacks were detected with mAb 188331 (R&D Systems), and DLL1 was detected with mAb then used for image analysis with IMARIS software (version 7.4.2; Bitplane), 251127 (R&D Systems). Biotinylated J-1(DSL-EGF3) and N-1(11–13) were operating with IMARIS Surpass (volume and isosurface rendering analysis), detected with allophycocyanin-labeled streptavidin (BD Biosciences). CD46 to visualize and locate points of interest (expression of Jagged1, CD46 and expression was assessed with anti-CD46 (E4.3; BD Biosciences). Human Notch1). For colocalization studies, data sets were analyzed with IMARIS C3b and CR1 were detected with mAb ab17453 (Abcam) and mAb E11 (BD software. Data of colocalization events were determined with the statistical Biosciences), respectively. The mAb to α-E-catenin (ab19446) was from modules of the colocalization software of the IMARIS package. Abcam. Anti-CD122 (FAB224A) and anti-CCR7 (FAB197F) were from R&D Systems, and anti-CD4 (555349), anti-CD8 (555635), anti-CD11a (555379), Cytokine measurements. Cytokines from cell cultures or mouse serum were anti-CD25 (555431), anti-CD46RA (555488), anti-CD45RO (559865), measured with human TH1/TH2 Cytometric Bead Arrays (BD Biosciences) or

nature immunology doi:10.1038/ni.2454 the human IFN-γ and IL-10 Cytokine Secretion Assay Kits (Miltenyi Biotec) FuGENE 6 Transfection Reagent (Roche Diagnostics). After 48 h, medium in combination according to the manufacturer’s protocol. was collected and filtered and was added to Jurkat cell cultures. Virus-infected cells were selected by puromycin. Knockdown of DLL1 protein was consist- Quantitative real-time RT-PCR. Primers used to quantify mRNA tran- ently above 50%. scription in CD4+ T cells were as follows: NOTCH1 forward, 5′-CG CACAAGGTGTCTTCCAG-3′, and reverse, 5′-AGGATCAGTGGCGTC Statistical analysis. Statistical analyses were done with the Student’s one- GTG-3′; NOTCH2 forward, 5′-TTGAGAGTTATACTTGCTTGTGTGC-3′, tailed t-test and Bonferroni correction for multiple comparisons (Excel soft- and reverse, 5′-GATACACTCGTCAATGTCAATGG-3′; JAG1 forward, ware; Microsoft). 5′-AGCCTTGTCGGCAAATAGC-3′, and reverse, 5′-AGCCTTGTCGGC AAATAGC-3′; JAG2 forward, 5′-CGACCAGTACGGCAACAA-3′, and reverse, 43. Gibb, A.L., Freeman, A.M., Smith, R.A., Edmonds, S. & Sim, E. The interaction of soluble human complement receptor type 1 (sCR1, BRL55730) with human ′ ′ ′ 5 -GGAGCAAATTACACCCTTGTTTA-3 ; DLL1 forward, 5 -GTGGGG complement component C4. Biochim. Biophys. Acta 1180, 313–320 (1993). AGAAAGTGTGCAA-3′, and reverse, 5′-TCACAAAATCCATGCTGCTC- 44. Marley, J., Lu, M. & Bracken, C. A method for efficient isotopic labeling of 3′; HES1 forward, 5′-GAAGCACCTCCGGAACCT-3′, and reverse, 5′-GT recombinant proteins. J. Biomol. NMR 20, 71–75 (2001). CACCTCGTTCATGCACTC-3′; and RBPJ forward, 5′-GAAGTACCATGG 45. White, J. et al. Biological activity, membrane-targeting modification, and crystallization of soluble human decay accelerating factor expressed in . ′ ′ ′ E. coli CGTGGATT-3 , and reverse, 5 -TTTCGCATAGCTTCCCTAGTAAGT-3 . Protein Sci. 13, 2406–2415 (2004). 46. Roversi, P. et al. Structures of the rat complement regulator CrrY. Acta Crystallogr. RNA silencing. Indocarbocyanine-labeled small interfering RNA target- Sect. F Struct. Biol. Cryst. Commun. 67, 739–743 (2011). ing human α-E-catenin (s3718) and negative control siRNA were from 47. Wang, G., Liszewski, M.K., Chan, A.C. & Atkinson, J.P. Membrane cofactor protein 4 (MCP; CD46): isoform-specific tyrosine phosphorylation. J. Immunol. 164, Ambion; these experiments done as described . Transfection efficiency and 1839–1846 (2000). cell viability was consistently above 80% and 75%, respectively, and protein 48. Schleucher, J. et al. A general enhancement scheme in heteronuclear knockdown peaked at 24–36 h after transfection. For short hairpin RNA– multidimensional NMR employing pulsed field gradients. J. Biomol. NMR 4, mediated silencing of DLL1 in Jurkat T cells, the appropriate lentivirus was 301–306 (1994). 49. Yamazaki, T., Lee, W., Arrowsmith, C.H., Muhandiram, D.R. & Kay, L.E. A Suite of generated by cotransfection of HEK293T cells with the packaging plasmid triple-resonance NMR experiments for the backbone assignment of N-15, C-13, psPAX2 (Addgene), envelope plasmid pMD2.G (Addgene) and pLKO.1 vector H-2 labeled proteins with high-sensitivity. J. Am. Chem. Soc. 116, 11655–11666 containing short hairpin RNA targeting DLL1 (Abgene) through the use of (1994). © 2012 Nature America, Inc. All rights reserved. America, Inc. © 2012 Nature npg

doi:10.1038/ni.2454 nature immunology