Control of Relb During Dendritic Cell Activation

© 2012 Nature America, Inc. All rights reserved. Kyoto, Japan. San Diego, La Jolla, California, USA. California, USA. 1 the determines activation RelB of extent the Indeed, the RelB-controlling kinase NIK results in deficient T cell responses DC whereas mice, null these in hyperactive are cells notably, T phenotype: opposite RelB- functionally to which leads in deficiency types cell other by masked be may DCs in RelB of in mice have splenic DC subsets deficiencies responses T cell antigen-specific alters sion the radically DC maturation process and results in blunted Specifically, small interfering RNA–mediated silencing of RelB exprescells antigen-presenting as function cells antigen-presenting in therapies immune-based modulating or engineering for attention responses immune innate DC in underlies functions interferon and (TNF) factor necrosis tumor as such duction of such molecules inflammatory as nitric oxide and cytokines pro addition, In activation. cell T and presentation, and processing uptake, antigen effective enable that factors intracellular of program expression gene a to addition in IL-23) 23; interleukin example, (for co cell T (MHC), molecules complex histocompatibility major of sion response immune adaptive role in the their that determines a program undergo maturation DCs (PAMPs),patterns molecular immature gens or pathogen-associated When patho functions. encountering immune-surveillance perform and lymphoid nonlymphoid both populate that ture DCs imma become to differentiate progenitors DC immunity. adaptive DCs are sentinel specialized immune cells essential in both innate and T cell responses. potential utility of systems analyses in guiding the development of combination therapeutics for modulating DC-dependent control. Canonical pathway control of RelB regulated pathogen-responsive gene expression programs. This work illustrates the control points of this unexpected cell type–specific regulation. Fibroblasts that we engineered accordingly showed DC-like RelB maturation but enabled rapid pathogen-responsive maturation. Computational modeling of the NF- pathway, but as a RelB-p50 dimer regulated by canonical I in disease. Here we report that RelB promoted DC activation not as the expected RelB-p52 effector of the noncanonical NF- The NF- Jeffrey D Kearns Vincent F-S Shih integrates canonical and noncanonical NF- Control of RelB during dendritic cell activation nature immunology nature Received 28 November 2011; accepted 29 August 2012; published online 21 October 2012; Signaling Systems Laboratory, Department of Chemistry and Biochemistry and San Diego Center for Systems Biology, University of California, San Diego, La Jolla, - The transcription factor NF- factor transcription The stimulatory molecules (CD40, CD80 or CD86) and cytokines cytokines and CD86) or CD80 (CD40, molecules stimulatory k B protein RelB controls dendritic cell (DC) maturation and may be targeted therapeutically to manipulate T cell responses 5 Present address: Merrimack Pharmaceuticals, Cambridge, Massachusetts, USA. Correspondence should be addressed to A.H. ( 2 Division of Biological Sciences, University of California, San Diego, La Jolla, California, USA. 1 1 , , , Jeremy Davis-Turak 5 , , Tony Yu aDV 1 5 . A hallmark . of A is maturation DC expres hallmark and is critical for DC maturation, DC DC maturation, DC for critical is and A 4 Innovation Center for Immunoregulation and Therapeutics, Graduate School of Medicine, Kyoto University, Yoshida-Konoe, Sakyoku, NCE ONLINE PUBLIC ONLINE NCE κ in in vitro B protein RelB is highly expressed expressed highly is RelB protein B 1 , , Riku Fagerlund 6 and DC-mediated immunity. immunity. DC-mediated and 2 and and , 3 . DCs have thus attracted attracted thus have DCs . 8 , in in vivo 9 but roles other critical - 1 specific deletion of of deletion specific , , Monica Macal A 7 TION . RelB-deficient . RelB-deficient

tolerance or or tolerance 1

tissues and tissues

, , Masataka Asagiri k Bs, I 4 1 . 0 - - - - - . 2 k , , Jenny Q Huang B of iterative computational modeling and quantitative experimental experimental quantitative and modeling computational iterative of approach biology systems We a DCs. used in RelB regulating for ble absence. their in DCs of maturation spontaneous PAMPsor limit must but pathogens to exposure after programs tion matura of induction decisive and rapid enable to expression RelB DCs, the molecular control mechanisms must provide for constitutive led in DCs may be different than what has in been described MEFs. In that suggest reports by the mechanism is RelB which activity control essential be to shown been has TRAF6 component pathway (LPS) lipopolysaccharide and TNF stimuli pathway canonical by DCs in activated rapidly be to found functions DC proper for required and NIK protein signaling the as such components pathway signaling noncanonical DCs, in RelB NF- synthesized NF- ‘noncanonical’ the via but pathway signaling ‘canonical’ dependent, NEMO- the via pool cytoplasmic latent a from activated not is RelB NF- classical have that unlike shown MEFs in studies biochemical Detailed studies. signaling many for system model useful a as served have (MEFs) fibroblasts embryonic Mouse unclear. remained have DCs in activity its control that mechanisms autoimmune for DCs RelB-silenced using therapies diseases cell-based of investigations prompted cells T regulatory or activated associated of balance the determining by transplants organ allogenic of rejection a In this study, we elucidated the molecular mechanisms responsi mechanisms molecular the elucidated we study, this In molecular the RelB, of importance clinical potential the Despite and I κ B pathway that involves proteolysis and processing of newly newly of processing and proteolysis involves that pathway B doi:10.1038/ni.244 k 1 B , 4 « , Elina , I Elina Zuniga . I k κ B control of RelB minimized spontaneous B2 p100 B2 1 3 , , Julia Ponomarenko San Diego Supercomputer Center, University of California, 6 12– 1 4 2 Nfkb2 . Consistent with the critical role of of role critical the with Consistent . & Alexander Hoffmann 1 k 10 1 16– B signaling module identified . k , 1 gene have been reported to be be to reported been have gene 5 1 . However, RelB has also been been also has RelB However, . 9 B pathways , and the canonical signaling signaling canonical the and , κ [email protected] B (the RelA-p50 dimer), dimer), B RelA-p50 (the 3 7 , . These insights have have insights These .

ticles e l c i rt A

9 k . These These . ). 1 B - - -

© 2012 Nature America, Inc. All rights reserved. trol in MEFs in trol parameter set ensemble; it recapitulates well-documented NF- single a to model the constrain that measurements made newly and published on based parameters kinetic unique 53 and reactions 132 Note ( kinetics mass-action of terms in dimers RelB and RelA of regulation and formation math the a describes that model developed ematical we manner, quantitative a in DCs in signaling is implicated in and cell survival maturation. To examine NF- p100 to p52 and generation of the RelB-p52 transcription factor, which of processing triggers and IKK1, NIK pathway,kinases the involving matory and proliferative gene expression programs. The noncanonical RelA-containing and c-Rel-containing NF- I classical of NF- degradation triggers IKK, kinase dependent ( pathways NF- of view established The NF for model a DC-specific Developing RESULTS responses. immune control DC-orchestrated pathway activity programs by canonical the regulated expression gene RelB-dependent that revealed profiling accordingly to produce RelB DC-like control. Finally, gene expression pathway, and we demonstrated by their sufficiency engineering MEFs control points to by that render RelB subject regulation the canonical DC-specific two identified studies pathway. Modeling canonical the activity was limited by classical I NF- the of analyses s.d.; bars, (error experiments independent three least at of representative are Data types. cell indicated the from prepared in described are methods Quantification fraction. nuclear and from derived parameters DC-specific NF- describing 5.0–MEF version model mathematical the using distribution immunoblotting. p52 and p100 IKK1, to subjected were 1–10) (days course ( MEFs. in value respective the to relative graphed BMDCs, and BMDMs MEFs, resting in cell per numbers (right); immunoblot by proteins p52 and p100 RelB, LT or LPS by induced respectively) nRelB, ( clarity of sake for I (The translocation. nuclear RelB-p52 for allows which processing, p100 in results that complex kinase NIK-IKK1 activate signals Developmental nucleus. the into I and I triggers that complex kinase NEMO-containing MEFs NF- noncanonical and canonical ( DCs. in regulation RelB 1 Figure s e l c i rt A lymphotoxin delayed b

) Computational simulations using the MEF-based kinetic model version 5.0–MEF ( 5.0–MEF version model kinetic MEF-based the using simulations ) Computational a

Canonical TNF orLPS κ Ø ). The first version of the model involves 41 molecular species, species, molecular 41 involves model the of version first The ). 1 B 2 NEMO . Inflammatory signals lead to activation of the the of activation to lead signals . Inflammatory d IK ε

( ) IKK1 and p52 abundance increase during DC differentiation. Whole-cell extracts prepared from BMDC culture during a differentiation time time a differentiation during culture BMDC from prepared extracts Whole-cell differentiation. DC during increase abundance p52 and ) IKK1 K A MEF-based kinetic model does not account for for account not does model kinetic A MEF-based α α βε κ , , Fig. 1 Fig. Bs: I Bs: Inflammation β 20– and and A 2 2 κ 50 2 1 a B , such as prompt LPS-induced RelA activation and and activation RelA LPS-induced prompt as such , ). A, RelA; B, RelB; 50, p50; 52, p52; and Ø, sink (from which proteins are synthesized and into which they are degraded). degraded). are they which into and synthesized are proteins which (from sink Ø, and p52; 52, p50; 50, RelB; B, RelA; A, ). ε ) ), degradation and the release of RelA-p50 RelA-p50 of release the and degradation ), 1 α κ 2 , I , . The canonical pathway, involving the NEMO- the pathway, involving canonical The . B signaling network in DCs to reveal that RelB RelB that reveal to DCs in network signaling B Organogenesis β κ a B –mediated RelB activation ( activation RelB –mediated ) Schematic of the distinct distinct the of ) Schematic B 52 β and I and p100 κ LT κ NI IK B pathways identified in identified B pathways Cytoplasm K B signaling comprises two separate separate two comprises signaling B K κ β Nucleus Bs, I κ c B , Ø κ d ε B ( κ . Resulting activation of latent latent of activation Resulting . δ Supplementary Note Supplementary B β Noncanonical pathway is not shown shown not is pathway stimulation. ( stimulation. α and I κ k b B dimers controls inflam κ B B signaling nRelA (nM) nRelA (nM) B 100 100 50 50 α κ , , I B 0 0

ε κ

, and regulated via κ B Fig. 1 Fig. Supplementary Supplementary B inhibitors, the B inhibitors, β c Time (h) Time (h) 1 8 2 1 ) Quantification of of ) Quantification

LPS

LT β ). Supplementary Figure 1 Figure Supplementary b 6 (f ).

) Quantification of RelB molecules per wild-type (WT) BMDC distributed in cytoplasmic cytoplasmic in distributed BMDC (WT) wild-type per molecules RelB of ) Quantification κ κ

B B RelB 24 B B con 3 d β -actin p100 IKK1 Day:

p52 nRelB (nM) nRelB (nM) 100 100 - - - 50 50 κ 0 0 B activation in MEFs as in in as MEFs in B activation β 3 1 Rela and and ( analyzed types cell the among abundance protein p100 the in showed but difference little immunoblotting quantitative BMDCs, more 3.5-fold We observed BMDCs. in expression RelB enhanced with correlated expression p100 whether tested fore the by encoded ( BMDMs and MEFs more sixfold ( correlated types cell amount in of protein these and relative the RelA and MEFs, BMDMs housekeeping the gene of expression the to Relative (BMDMs). macro phages marrow–derived bone and MEFs in that to comparison in NF- key of sion threefold, which increased the abundance of RelB but not its its not but RelB of abundance the increased which threefold, sion DCs in control RelB ( recapitulate to model computational the to NF- noncanonical the of activity constitutive elevated but also of RelB expression increased only not involves tion microRNAs of control the ( p52 to processing p100 concomitant with NF- cal expression of IKK1, the kinase determining the activity of noncanoni and degradation and p100 processing to p52 may occur p100 in BMDCs complete ( both that suggests which BMDCs, in p52 of amount the in increase We 2.5-fold a noted BMDCs. in elevated be may tion degrada p100 that suggested RNAand protein abundance p100 tive -actin served as a loading control. ( control. a as loading served -actin Supplementary Note Supplementary To adapt the model to DCs, we first measured the expression expression the measured first we DCs, to model the adapt To Based on the measurements, we made specific modifications modifications specific made we measurements, the on Based Time (h) Time (h) 1 8 2 1 , , Supplementary Supplementary Fig. 1b LPS Supplementary Fig. 1b Fig. Supplementary LT 1 6 Relb β β -actin -actin ( Supplementary Note Supplementary . κ 0 (g 6 and and B pathway, gradually increased during DC differentiation differentiation DC during increased gradually pathway, B ) RelB, RelA and p100 immunoblots of cytoplasmic extracts extracts cytoplasmic of immunoblots p100 and RelA ) RelB, aDV e κ 24

Computational 3 B proteins in bone marrow–derived DCs (BMDCs) (BMDCs) DCs marrow–derived bone in proteins B Nfkb2 Actb Relb RelB (nM) A 10 20 30 40 NCE ONLINE PUBLIC ONLINE NCE 0 Nfkb2 c ), ), expression of mRNA and protein expression in BMDCs than than BMDCs in expression protein and mRNA Nfkb2 mRNA Relb mRNA Rela mRNA transcripts by quantitative RT-PCR RelA, of and (left) quantitative by transcripts ME

Nucleus Cytoplasm (fold) (fold) (fold) 0 1 2 3 4 0 1 2 3 4 0 1 2 3 4 Fig. 1 Fig. D F i. 1 Fig. ). First, we increased increased we First, ). b ee i kon o nii Rl. e there We RelB. inhibit to known is gene, MEF or in model version 5.0–DC incorporating incorporating 5.0–DC version model in or BMDC ) of nuclear RelA or RelB activity (nRelA and and (nRelA activity RelB or RelA ) nuclear of C

2 BMDM 3 ). ). Consistent with this hypothesis, protein c . Our data indicate that DC differentia DC that indicate data Our . ). Lack of correlation between the rela the between correlation of Lack ). c ). In contrast, we observed threefold to to threefold observed we contrast, In ). and and f

e Experimental RelB ) ) Rela (per nucleus/cytoplasm In silico In

upeetr Fg 1a Fig. Supplementary 3 p100 protein RelB protein RelA protein A × 10 ) 120 TION

mRNA was similar in BMDCs, (fold) (fold) (fold) 30 60 90 0 2 4 6 8 0 2 4 6 8 0 2 4 6 8 0 Nucleus Cytoplasm n MEF = 3).

simulation of RelB cellular cellular RelB of simulation BMDC

Fig. 1 Fig. BMDM nature immunology nature Relb κ B signaling pathway. signaling B and and β g d -actin p100 RelA RelB ), potentially via via potentially ), Nfkb2 p52 protein Nfkb2 (fold) MEF 0 2 4 6 8

mRNA in in mRNA MEF

BMDC BMDC expres ). p100, p100, ).

Fig. 1 Fig. BMDM Fig. 1 BMDM

c c ------

© 2012 Nature America, Inc. All rights reserved. RelB from BMDC whole-cell lysates and analyzed the associated associated the analyzed ( proteins and lysates whole-cell immunoprecipitated BMDC we from DCs, RelB in RelB of inhibitors for search To I cytoplasm. the in RelB sequester that mechanisms regulatory undescribed as-yet be may there that gested of RelB control, did not reproduce our experimental observations sug the known that encodes which mechanisms model, the mathematical of than BMDCs of MEFs or BMDMs, p100 was not ( Fig. 1c ( cytoplasmic was protein RelB total the of 75% than more that found but extracts nuclear and cytoplasmic in DCs primarily localizes into the nucleus, we separated BMDCs into ( activity RelB nuclear IKK1- in as the ( expression by MEFs p100 p100 comparable achieve to destabilized pathway we dependent Then localization. nuclear * Rel (GM-CSF)–cultured factor colony-stimulating granulocyte-macrophage MHCII of frequency the of cells in respective quadrants are indicated. Dot plots indicate Percentages (KO). knockout by proteins indicated for depleted MHCII be ( BMDCs. (WT) wild-type in activity RelB total to relative ( total the to normalized and summed were (bottom) analysis antibody-ablation the of intensities Band EMSA. by revealed as (KO) knockout by proteins indicated for depleted BMDCs immature in RelB-p52 and RelB-p50 activity, Rel Nfkbia from collected extracts nuclear ( data. the in variance the represents overlap apparent the and number represents the average of several independent analyses in signals FT and signals co-IP of analyses I with interacting not or complex I with interacting RelB ( controls. negative as served members I other with immunoblotting RelB; and proteins ( control. specificity antibody an as served immunoprecipitates IgG (FT). flow-through and IN) (input; lysates cell whole to IP) was compared to (immunoprecipitates; bound RelB proteins of Fraction partners. interaction for indicated probed DC extracts ( 2 Figure nature immunology nature no leukocyte reactions produced for each of two independent BMDC cultures. Error bars, s.e.m.; s.e.m.; bars, Error cultures. BMDC independent two of each for produced reactions leukocyte in Data ratio. cell DC:T the of as a function graphed cytokine, activation-associated indicated the for positive T cells of fraction bottom, and, cells divided of fraction Middle, production. IL-2 ( peptide a f d k P ) Fractions of CD11c of ) Fractions ) Immunoblots of RelB immunoprecipitates from whole whole from immunoprecipitates RelB of ) Immunoblots ) ) NF- < 0.01. ( < 0.01. B n −/− −/− p105 p100 RelB I b a I I κ κ p50 κ canonical regulator, was associated with RelB. Unexpectedly, Unexpectedly, RelB. with associated was regulator, canonical a B B B

Tnf α restrains RelB-p50 and spontaneous DC maturation DC spontaneous and RelB-p50 restrains β Relb ε −/− κ , d

B RelB DNA binding activities revealed by EMSA with with EMSA by revealed activities binding DNA B RelB Fig. h Tnf −/− ). ). Indeed, whereas RelB was more abundant in the cytoplasm IP RelB I hi IgG κ ) or OVA protein ( OVA ) or protein Fig. 2 Fig. CD86 −/− B cells ( cells g

−/− IP

α – 1 BMDCs (right). ( (right). BMDCs IN binding RelB-p50 limits spontaneous DC maturation. maturation. DC spontaneous limits RelB-p50 binding i

c ) T cell proliferation in DC–T cell cocultures using using cocultures cell DC–T in proliferation ) T cell Rel ). This change resulted in a substantial increase of of increase substantial a in resulted change This ). FT hi a and MHCII and ). As expected, p100, the known RelB inhibitor and and inhibitor RelB known the p100, expected, As ). n −/− = 5), = 5), b p100 RelB I I I ) Co-immunoprecipitation of I of ) Co-immunoprecipitation κ BMDCs (middle) and and (middle) BMDCs κ κ + B B κ B BMDCs determined by FACSby to determined BMDCs hi α β B ε Fig. CD86 IP p100 α IP

Nfkbia , , I i IN aDV ). Top, raw FACS data of CFSE-labeled T cells stained for IL-2, showing proliferation-associated dye dilution and and dilution dye proliferation-associated showing IL-2, for stained T cells Top,). CFSE-labeled FACS of raw data

κ 1 lo FT B hi e e CD86 Rel ε ) Quantification of total RelB RelB total of ) Quantification dendritic cells derived from from derived cells dendritic ). To test experimentally whether RelB RelB whether To experimentally ). test κ −/− A , p100 or I or , p100 IP IP Bs based on quantitative quantitative on based Bs NCE ONLINE PUBLIC ONLINE NCE −/− Tnf IN I κ lo B FT Tnf in controls and in cells cells in and controls in α −/− c ) Fractions of ) Fractions IP IP a −/−

Rel , IN b I d a κ BMDCs (left), BMDCs Nfkbia κ , and and ) and graphed graphed ) and Fig. 1 Fig. B FT d B −/− β , δ e

κ –containing –containing IP ( are representative of at least two independent experiments. Data in in Data experiments. independent two least at of representative are IP B family B family n b IN −/− I = 6), and and = 6), . The . The f κ κ and and B FT B Tnf ε

−/− A

associated Fig. Fig. 1 Supplementary Supplementary Not TION

Nfkbia κ

p100/I B-

). ). The fact

IL-2 g 17% 10 10 10 10 Tnf 41% 0 2 3 4 5 κΒδ −/− CFSE 8.54 0 + + + + CFSE dilution 0 CD4 TNF cells CD4 IL-2 cells 10 −/−

27% 2 15% Ctrl Tnf

(%) 10

(%) (%) 100 3 20 40 60 80 20 40 60 80 20 40 60 80 , , 10 0 0 0 I 91.5 4 Nfkbia κ I −/− 10 κ -

1:32 0 1:32 1:32 B 5 B

α Medium ε 15.8 0.319 I

1:16 κ Rel 1:16 1:16 B I I Ctrl DC/T cell DC/T cell DC/T cell which prompted us to investigate the function of this interaction. this of I function the with investigate to us prompted which associated (19–34%) RelB of proportion substantial A that 37–45% of RelB was associated with p100 and 12–17% with I ( in BMDCs protein distribution RelB I with various immunoprecipitation after flow-through in the and remaining tured ciatedwith each RelB molecule. Analyses of the amounts of RelB cap and used of that only the antibodies one I specificity immun ( expected as ( I with associated also but p100 with acted NF- ous cytoplasmic ( primarily was complex this and RelB in found immunoprecipitates, were pathway, canonical the in RelA of partner ing bind the as known p50, of amounts Substantial not. were p105 and NF- I α κ κ Fig. Supplementary Fig. 2a Supplementary κ B B -K

−/− 1:8 1:8 α α

−/− 1:8 B O 80.4 3.51 &RelB-K -K f

α 1:4 Tnf 1:4 O κ

1:4 Relb MHCII 12.5 0 and I and 2 B pathway, were also immunoprecipitated with RelB, but I but RelB, with immunoprecipitated also pathway, B were RelB-KO 1:2 1:2 10 10 10 10 10 I b 1:2 κ −/− 0 1 2 3 4 o 10 B O and κ CD86 precipitates did not contain other I α 58 0 −/− and and & 10 87.2 B inhibitors confirmed that RelB not only directly inter directly only not RelB that confirmed inhibitors B

0.3 hi hi 1 Ctrl κ

( MHCII CD86 10 B Supplementary Supplementary Fig. 2c n

(%) 2 κ Supplementary Fig. 2d Fig. Supplementary Nfkbia 27.9 10 = 3) bone marrow cells in individual experiments. experiments. individual in cells marrow bone = 3) 20 40 60 ε h n IL-2 0 B antibodies provides a quantitative understanding of provides a B understanding quantitative antibodies , the classical I classical the , 10 10 10 10 3 = 4. 10 0 2 3 5 4 + + + + CFSE dilution d CFSE 21.3 19.3

CD4 TNF cells CD4 IL-2 cells 4 0 (%) (%) (%) 10 2 100 Ctrl 10 −/− 42.3 RelB EMSA 20 40 60 80 20 40 60 80 20 40 60 80 I 3 0 0 0 κ * 10 56.7 2.68 B Tnf 1:32 1:32 1:32 4 10 α 5 -KO , OVA pep b 1:16 1:16 1:16 20.1 59.5 I −/− κ DC/T cell DC/T cell DC/T cell ). Reciprocal ). immunoprecipitation of Reciprocal vari * B 42

1:8 α Relb 1:8 1:8 -K O 13.2 7.18 1:4 I κ 1:4 1:4 κ B B inhibitors regulating the canonical canonical the regulating inhibitors B 64.2

−/− 1:2 1:2 α 1:2 23.4 8.46 RelB-KO &RelB-KO I κ BMDCs exposed to medium ( medium to exposed BMDCs B g ), ), and RelA associated with I α – 18.6 & 65.4 2.72 i Fig. 2 Fig. are the average of duplicate duplicate of average the are ). The observations that I that observations The ).

IL-2 i κ κ 10 10 10 10 + + + + CFSE dilution Ab 0 3 c B isoforms confirmed the 2 5 CD4 TNF cells CD4 IL-2 cells 4 B e 0 CFSE ). This analysis revealed revealed ). analysis This

(%) 7.89 0.263 (%) (%) 10 α 2 20 40 60 20 40 60 80 20 40 60 80 RelB activity 10

Ctrl – 0 0 0 and I and 3 RelB (fold)

1:32 10 1:32 1:32 Ctrl 0 1 2 3 4.87 4 p50 87 10

1:16 s e l c i rt A 1:16 1:16 5 DC/T cell κ p52 17.2 38 I DC/T cell DC/T cell κ RelB-p50 RelB-p52 Total Ctrl κ OVA B isoform B asso isoform 1:8 1:8 B p50+p52

1:8 α B -K

ε 1:4 1:4 O 1:4 29.3 15.4 I in BMDCs BMDCs in κ I

κ 1:2 1:2 – B

1: B 9.3 0.744 α 2 RelB-KO RelB α I -KO 5 κ

-KO B p50 g α ), OVA),

& p52 88.3 1.6 κ κ κ κ 5

B p50+p52 B B κ B α α ε B β - - - - - , . ,

© 2012 Nature America, Inc. All rights reserved. the partial dependence on ( MHC of RelB expression surface dependence the partial but largely on not ( RelB dependent entirely (OVA)-responsive T cells to exposed OVA ovalbumin peptide, and with this effect was cocultures BMDC in functions antigen-presenting cell DC–T in cells c T antigen-specific proliferation of activate production to cytokine ability and their testing by DCs of functions ( phenotype the the of deletion compound as RelB, on ent I of phenotype maturation DC tion ( stimuli MHCII of frequency ( stimuli nal (42% versus 28%) of MHCII I II. MHC CD86 and markers activation the of expression surface DCs by matured indicated as of frequency the monitoring by misregulation RelB ( BMDCs NF- predominant the RelB-p50 ing render dimer, RelB-p50 active in increase substantial a in resulted of I ablation activity, RelB-p52 constitutive primarily contained p52 and ( p50 for specific be to shown were that antibodies twofold with sis than more was I in activity RelB elevated that found we Using tools, EMSA. RelB these specific a in resulting RelA, for antibodies ing with (EMSA) assay shift mobility electrophoretic standard the modified ( background we c-Rel–deficient First, a onto activity. mice the RelB bred on analysis experimental our focus to gies I of advantage took s.d.; bars, (error experiments independent two of representative are in shown Data (right). cells resting respective to relative graphed NF- ( proteins. indicated the in (KO) deficient were that 5.1–DC version on based models, mathematical in activation RelB CpG-induced with Immunoprecipitation BMDCs. ( proteins. indicated to antibodies with immunoblotting to subjected were BMDCs WT from prepared extracts Whole-cell CpG. by induced profiles ( (bottom). activity basal respective their to relative stimulation CpG after and before activities RelB-p52 and in shown activity RelB nuclear total to up sum that stimulation LPS upon activities RelB-p52 and ( MEFs. and BMDCs in activity peak RelA similar reveal to adjusted was images of exposure and loaded were MEFs or BMDCs from proteins nuclear of amounts Equal EMSA. to subjected and collected were stimuli indicated by activated MEFs WT or BMDCs (WT) wild-type from extracts Nuclear EMSA. by monitored NF- ( 5.1–DC. version model mathematical refined the using course a 3 h time during activity RelB and RelA ( maturation. DC TLR-mediated during 3 Figure s e l c i rt A protein, ovalbumin to cocultures these exposed we when However, c a upeetr Fig. Supplementary o

) Computational simulations of RelB-p50 RelB-p50 of simulations ) Computational LPS-induced of simulations ) Computational cultures ( cultures To test whether I whether test To κ κ B RelB DNA binding activities in BMDCs depleted for indicated proteins by knockout (KO), monitored by EMSA (left). Signals were quantified and and quantified were Signals (left). EMSA by monitored (KO), knockout by proteins indicated for depleted BMDCs in activities binding DNA B RelB activities binding DNA (right) B RelB κ B site–containing probes ( probes site–containing B

RelB-p50 is rapidly activated activated rapidly is RelB-p50 a e (top). Quantification of RelB-p50 RelB-p50 of Quantification (top). Fig. ) Association of I of ) Association Fig. 2g Fig. Supplementary Fig. 2f Supplementary Fig. 2 Fig. κ

2 B κ b e Fig. 2 Fig. B α d ) ) NF- ). We examined the functional consequences of of consequences functional the examined We ). α ) ) I dfcet MC ( BMDCs -deficient – f κ deficiency resulted in an increased percentage percentage increased an in resulted deficiency ). Although RelB deficiency did not affect the the affect not did deficiency RelB Although ). κ i κ B ). We). I that found hi B protein expression expression B protein κ B f

α CD86 B RelA (left) and and (left) B RelA 2e ). We then examined the antigen-presenting antigen-presenting the examined We then ). α -deficient mice -deficient may inhibit RelB activity in BMDCs, we we BMDCs, in activity RelB inhibit may ) revealed that whereas control BMDCs BMDCs control whereas that revealed ) κ B hi α hi CD86 to RelB monitored during a CpG time course by examining RelB immunoprecipitates (IP) from CpG-stimulated WT CpG-stimulated from (IP) immunoprecipitates RelB examining by course time a CpG during monitored RelB to BMDCs before exposure to matura to exposure before BMDCs κ Relb B κ α

hi B EMSA) to include shift-ablat include to EMSA) B -deficient BMDCs was depend was BMDCs -deficient ), ), the inappropriate spontaneous

−/− BMDCs in the absence of exter κ extracts (C) serves as a control, indicating specificity of RelB antibody. ( antibody. RelB of specificity indicating a control, as serves (C) extracts atvt i I in activity B κ 2 i. 2 Fig. B 2 and developed two strate two developed and α Fig. Fig. 2 deficiency increased the the increased deficiency Relb d . uesit analy Supershift ). gene fully reversed reversed fully gene a h c

), correlating with ), correlating Experimental Computational nRelA (nM) 100 50 Rel RelB activity RelB activity 0 κ (fold) (nM) 1 0 100 0 1 2 3 −/− B 50 0 α 1 0 ); then we we then ); -deficient -deficient Time (h 1 0 RelB-p50 RelB-p52 b Fig. LPS Time (h , d RelB-p50 RelB-p52 n , Time (h e = 3). = 3). 2 κ )

LPS , 2 g B ) f are representative of at least three independent experiments. Data shown in in shown Data experiments. independent three least at of representative are α ). ). ------2 )

with anti-LTßR ( anti-LTßR with stimulated when points time later at RelB activated MEFs although MEFs, in not but BMDCs in stimuli TLR to response in activation MEFs stimulated with TLR ligands ( ( proteins Rel domain–containing activation for antibodies ablating EMSA RelA developed newly the NF- RelB-containing and RelA-containing of NF- cal noncanoni the induce LTßRto to antibody agonistic an as well as Pam TLR9 ligand the TLR2 the with CpG, ligand MEFs and BMDCs stimulated we experimentally, ( RelB of also but RelA of only not activation substantial and rapid indicated simulations Such input. measured an as experimentally activity kinase IKK NEMO-dependent the of data using time-course maturation DC (TLR)-induced Note protocol ( optimization parameter multidimensional in as a results constraints immunoprecipitation the quantitative used rated them into the model mathematical as rate kinetic equations and interactions I RelB-p50 with of consequences regulatory the explore To NF canonical the via RelB-p50 activate TLRs maturation. spontaneous inappropriate prevent to activity immature it role DCs in RelB has also functional restraining a critical of the expression restrains RelB by RelA or controlling c-Rel NF- classical the that demonstrate data Together, these cells. of antigen-presenting program processing and uptake antigen the regulating in RelB for function specific a gesting DCs RelB-deficient of studies previous with correlating presented, being before processed and T cell activation showed up a near absolute dependence on RelB taken ( be must which nRelB (nM) Supplementary Fig. Supplementary 100 3 50 d g 0 α Time (h 1 0 Time (h -tubulin . e iuae NF- simulated We ). p105 p100 I I κ I κ κ κ B B B B κ α β ε Time (h ) B pathway. To specifically examine the activation profiles profiles activation the examine specifically Topathway. B α ) LPS aDV and I and 0 0 0.5 0.5 2 ) CpG Rel A CpG 1 1 NCE ONLINE PUBLIC ONLINE NCE B κ Fig. 3 Fig. 1.5 1.5 B 3 ε 3 proteins during DC maturation, we incorpo we maturation, DC during proteins 3

BMDC 3b b ME b e I I WT Time (h κ κ ). ). RelA activation was similar in BMDCs and and and B B F α βε RelB RelB I I κ κ -K Ig Pa Pa κ δ B B -K G O α α B regulation during Toll-like receptor receptor Toll-like during regulation B C ) m m Time (h O Supplementary Fig. 3c Fig. Supplementary 3 3 IP RelB CSK CSK Cp LP LP Cp 1 0 G S S 4 4 ) f

RelB activity G ) Computational simulations of simulations ) Computational 3 2 Fig. Fig. 3 A Fig. 3 Fig. CSK 2 (fold) 0 n Rl ES uig shift- using EMSA RelB and TION Input IP 0 1 2 3 0. 0 κ 5 4 RelA B inhibitor, I inhibitor, B b and the TLR4 ligand LPS LPS ligand TLR4 the and f a 1 1

). ). We rapid observed RelB

). To test this prediction prediction this test To ). RelB activity nature immunology nature 1 Time (h (fold) .5 1. 1. 2. κ 0 5 0 3 0 B dimers, we used used we dimers, B g k ) 3 2 ) CpG-induced ) CpG-induced B B pathway Supplementary Supplementary 0 0 ). Similarly, we we Similarly, ). κ 2 1 Time (h I I WT B κ κ .5 B B α 6 RelB α βε , and sug and , , not only only not , -K 1 1 δ 2 I I WT -K ) O κ κ Fig. 4 B B , , but in O α βε .5 -K δ

-K O 3 2 c O i 3

), ), - - -

© 2012 Nature America, Inc. All rights reserved. activation ( activation and I activity RelA (ref. PS-1145 inhibitor the way,by Fig. 3h of ( RelB pool of a pre-existing translocation nuclear stimulus-responsive of indicative stimulation, CpG after cytoplas RelB of mic disappearance by accompanied was RelB of mulation accu nuclear DCs in that confirmed immunoblotting contrast, In activation. RelB CpG-inducible for required not is synthesis protein ( blocked I of resynthesis whereas activation, RelB induced CpG- not block did by cycloheximide of synthesis protein inhibition ( fourfold than more induced or RelB of expression or p50, protein in increases detect not did we DCs responsive NF- is activation pathway.RelB canonical by the regulated maturation DC during that suggest data These Fig. ( activity RelB-p50 increased primarily were present under unstimulated conditions but that CpG stimulation activities and RelB-p52 RelB-p50 both that revealed extracts nuclear than dimer ( RelB-p52 rather of RelB-p50 consists activity RelB induced this that suggested or Pam rapidobserved s.d.; bars, (error experiments independent two of representative are here I via stimuli canonical to responsive becomes and p50, also but p52 only not dimerize will RelB levels steady-state high at MEFs; in reported as stimuli noncanonical to responsive is RelB levels, steady-state low at RelB: activate stimuli which determines activity pathway noncanonical and expression RelB of control steady-state type–specific Cell stimuli. canonical to responsive become and p50 with dimerize or stimuli noncanonical stimulus-induced to response in p52 with dimerize either may RelB stimuli. canonical or noncanonical by RelB of regulation the or control of stimulation TNF to response in protein fusion RelB-GFP expressed ( activity. basal dimers’ RelB-containing respective to relative ( (RelB-TG). or control from collected extracts nuclear with monitored were LPS by induced activities DNA-binding (bottom) ( protein. RelB endogenous represents band bottom the whereas or (EV) or control from collected extracts whole-cell ( RelB of rate synthesis mRNA the and ( NIK of half-life the modulating when course time induction LPS an during RelB-p50 from derived results The halflife. NIK and synthesis RelB a function is RelB of inducibility LPS-mediated how 4 Figure nature immunology nature of abundance the We monitored activation. RelB for required is ing NIK halflife a A hallmark of canonical signaling is the release of a pre-existing pre-existing a of release the is signaling canonical of hallmark A κ (fold) 3e 10 B dimer, whereas noncanonical signaling B involves dimer, the signaling stimulus- noncanonical whereas 1 5 Relb ), unlike LTßR stimulation of MEFs, which induces RelB-p52. LTßR RelB-p52. ), unlike induces of which MEFs, stimulation ). ). Inhibition of IKK2 activity, a hallmark of the canonical path 3 0

1 CSK Relb Determinants of RelB’s responsiveness to canonical signals. ( signals. canonical to responsiveness RelB’s of Determinants upeetr Fg 3g Fig. Supplementary transgene (RelB-TG). The top band in the RelB blot represents exogenous protein, protein, exogenous represents blot RelB the in band top The (RelB-TG). transgene d de novo de 100 Supplementary Fig. 3g Fig. Supplementary nRelB-p50 (nM) ) Quantification of nuclear (n)RelB-p50 and RelB-p52 activities in LPS-stimulated LPS-stimulated in activities RelB-p52 and (n)RelB-p50 nuclear of ) Quantification mRNA, whereas whereas mRNA, 4 ( (fold) Rel

Supplementary Fig. 3d Fig. Supplementary activation activation of RelB in splenic DCs stimulated with CpG 5 b 200 κ generation of the dimer the of generation B α Fig. Fig. 3 in silico in protein degradation but also in reduced RelB RelB in but reduced also degradation protein Traf3 300

1 0 aDV −/− c TRAF3 ). Experimentally, supershift analyses of ). analyses Experimentally, supershift β Supplementary Fig. 3f Fig. Supplementary b -actin A simulations of peak abundance of nuclear DNA encoding encoding DNA nuclear of abundance peak of simulations MEFs transduced with empty vector (EV) or a or (EV) vector empty with transduced MEFs p100 Rel p52 NI Nfkbia NCE ONLINE PUBLIC ONLINE NCE K B ), suggesting that that suggesting ), x 26) resulted not only in reduced reduced in only not resulted 26) , Ctrl i axis). ( axis). EV ), suggesting that IKK2 signal IKK2 that suggesting ), mRNA, encoding I encoding mRNA, ). Computational simulations simulations Computational ). RelB-TG Fig. 3 Fig. Traf3 Traf EV b 12 ) Immunoblots with indicated antibodies of antibodies indicated with ) Immunoblots 3 , –/ −/− 2

c RelB-TG – 5 and and . In CpG-responding In CpG-responding . MEFs reconstituted with empty vector vector empty with reconstituted MEFs κ A Time (h c Traf Traf Supplementary Supplementary Supplementary Supplementary B TION ). Furthermore, Furthermore, ). α 3 3 Ctrl Ctrl e novo de –/ –/– protein was was protein – e ) ) Single-cell data at indicated time points time indicated at data ) Single-cell

0 0 κ c ) ) NF- B α .5 RelB RelB , was was , EV 1 1 κ a B RelB (top) and RelA RelA and (top) B RelB ) Heat map depicting depicting map ) Heat n - - - - .5 = 3). = 3).

0 3 RelA RelB mature but unstimulated DCs are abundant in basal RelB expression expression RelB in basal are abundant DCs mature but unstimulated We previously showed that hallmarks of the NF- control RelB DC-like show MEFs Engineered DCs. in activation RelB CpG-induced I that evidence mechanistic and genetic provide ( kinetics I activation was robust in ( by immunoblotting products Nfkbib I we used predictions, modeling activation diminished I greatly an in to opposed as activation, pathway 3k Fig. I I of in case the except effect, had little inhibitors of individual deletion I the of simulate effect computationally to model mathematical the used we I pre-existing I of degradation that suggest ( CpG to response in decreased RelB I of amount the assays, coimmunoprecipitation ( activation RelB of kinetics activation the with unaltered; however, I remained p105 and p100 inhibitors RelB potential the of abundance NF- known κ κ κ Relb Traf3 B B B To investigate the role of I of role the investigate To α ε α 0. and I and ( -deficient BMDCs showed a diminished increase and delayed delayed and increase diminished a showed BMDCs -deficient transgene transgene RelB-TG 1 5 −/− Supplementary Fig. 3j Fig. Supplementary −/−

κ y ) showed robust RelB activation in response to canonical canonical to response in activation RelB robust showed )

axis) axis) B MEFs (RelB-TG). Scale bars, 10 10 bars, Scale (RelB-TG). MEFs Fig. 3 Fig. Nfkbie α κ 1. κ -deficient model ( model -deficient B deletions on RelB activation. We found that that found We activation. RelB on deletions B

κ 3 5 B

B inhibitor proteins during the CpG time course: the the course: time CpG the during proteins inhibitor B κ

B (which elevated basal RelB activity; activity; RelB basal elevated (which g

−/− α and and Traf3 d -RelB complexes. -RelB f

Nfkb2 nRelB (fold) 8 0 4 κ Canonical Supplementary Fig. 3m Fig. Supplementary B −/− ∅ Time (h) RelB-p50 RelB-p52 1 0 PAMPs α Nfkbib NEM 7 ( IK and I of the nuclear localization of a retrovirally a retrovirally of localization nuclear the of −/− Relb K O mice. We confirmed the lack of protein protein of lack the We confirmed mice. α ε κ κ transgene) MEFs; signals were graphed graphed were signals MEFs; transgene) ). Even compound deficiency of I of deficiency compound Even ). B −/− Fig. 3 Fig. (min (min κ Time Time Supplementary Fig. 3l Fig. Supplementary B e α B κ α C Nfkbie allows for the release of RelB from from RelB of release the for allows ε B ) ) κ in TLR-induced RelB activation, activation, RelB TLR-induced in were rapidly degraded, correlating α B f α -deficient mice -deficient ). To test these computational computational these test To ). and I and 3 0 3 0 50 Fig. 3 Fig. −/− µ m. ( m. Nfkb2 κ Traf B B f e ) Schematic depicting depicting ) Schematic ). Together, these data data these Together, ). Ctrl +RelB-TG ε ). Together, these data data Together, ). these steady state Cell-specific κ control. Data shown shown Data control. 6 0 6 0 3 κ Fig. 3 Fig. B signaling system in –/ −/− κ – B +RelB-TG 52 B ticles e l c i rt A α BMDCs, whereas α 2 Supplementary Supplementary associated with with associated 2 is required for for required is ). Indeed, RelB RelB ). Indeed, and generated generated and d Cytoplasm 9 0 9 0 ). Notably, in Notably,in ). p100 NI IK Nucleus K K in silico in ∅

0 0 κ Noncanonical B β ,

© 2012 Nature America, Inc. All rights reserved. these stimuli remained unchanged. Neither single genetic alteration alteration genetic single Neither unchanged. remained stimuli these by activation RelA and not, did MEFs control parental the whereas ( LPS to response in tion 4a Fig. ( MEFs untransduced in that to relative retroviral a we transduced then by simulations, As model the suggested naling. trolling NIK degradation con E3-ligase an TRAF3, in deficient MEFs of advantage Wetook RelB. of regulation canonical DC-like for allow to sufficient be may MEFs into mechanisms two these engineering genetically whether activity. IKK elevated noncanonical and synthesis mRNA constitutively DC-specific, by explained be can tion stimula pathway canonical upon activation RelB rapid DC-specific, ( elevated concomitantly were parameters both when only occurred enhancement substantial but increased, was parameter either when increased mildly RelB of These half-life. protein NIK basal governing values range of parameter for a activation of RelB simulations computational we be, performed may contributions relative their what and are sufficient mechanisms and basal noncanonical pathway activity. To investigate whether these in in defect expression severe most the with genes 50 the for determined was deficiency RelB by caused phenotype expression The genes. of sets ( courses. time across gene given the of value expression highest normalized denotes ‘1.0’ scale Color (SAM). microarray of analysis significant by identified BMDCs in genes downregulated significant 157 of scale induction fold (log a in experiment one from pattern expression the shows map Heat points. Pam and CpG Rel and WT from analysis expression mRNA ( signals. input to relative graphed and genes indicated of region promoter the to corresponding primers with qPCR by performed was precipitated DNA of Quantification CpG. with stimulation after min 75 or (−) before collected BMDCs WT from extracts cell using antibodies control IgG or RelB with analyses ( from sequence the containing probes DNA using BMDCs WT stimulated CpG- from collected extracts nuclear ( cells. resting respective to relative graphed were Signals qRT-PCR. Pam and ( FACS. by analyzed were h 24 for CpG with treated or (Med) untreated Cells CpG. to response and (WT) wild-type in ( mediators. inflammatory and markers 5 Figure s e l c i rt A RelB enhanced that indicating activation, RelB substantial produced f d a

) RelB and c-Rel regulate overlapping overlapping regulate c-Rel and RelB ) ) Analysis of cell surface marker expression expression marker surface cell of Analysis ) ) Chromatin immunoprecipitation immunoprecipitation Chromatin ) a To test this model-derived hypothesis experimentally, we asked asked we experimentally, hypothesis model-derived this test To −/− – Relb 3 d CSK Relb are representative of at least three independent experiments (error bars, s.e.m.; s.e.m.; bars, (error experiments independent three least at of representative are b ). The engineered MEFs exhibited substantial RelB activa RelB substantial exhibited MEFs engineered The ).

) Gene expression analyses of WT WT of analyses expression Gene ) −/− RelB regulates DC activation activation DC regulates RelB 4 −/− Relb for the indicated time course by by course time indicated the for BMDCs stimulated with CpG or or CpG with stimulated BMDCs BMDCs stimulated with with stimulated BMDCs 3 CSK κ transgene to increase RelB expression about threefold threefold about expression RelB to increase transgene Tnf B site containing promoter promoter containing site B or or 4 for indicated time time indicated for Rel Il23a Relb −/− Rel c e Relb ) EMSA with with EMSA ) 2 −/− ) Microarray Microarray ) Fig. 4 Fig. Fig. 4 Fig. gene. gene. 7 −/− , , to increase constitutive noncanonical sig BMDCs in in BMDCs Relb in silico in −/− BMDCs. The list of genes was sorted according to expression differences between WT and and WT between differences expression to according sorted was genes of list The BMDCs. c a

) or TNF ( TNF or ) −/− ). Our simulations suggest that the the that suggest simulations Our ).

) analyses showed that activation activation that showed analyses

Fig. 4 Fig.

Relb Supplementary Fig. 4b Fig. Supplementary b mRNA synthesis and mRNA synthesis and and Relb d c b a WT Supplementary Supplementary –/– Time (h)

Tnf (fold) 140 Tnf (fold) 20 40 70 Cells 0 0 Tnf Tnf 10 MHCII 0

(% input) 10 1 0 1 0 0.01 0.02 0.03 0 1 Pam 10 Time (h) Time (h) 0 0.5 ** ** 2 Relb CpG 10 RelB CpG 3 CSK 1 3 – ** ** 3 3 ), ), 10 - - - 1.5

4 4 CpG IgG ** ** ** 8 8 10 3 CD40 vation by stimuli such as CpG that engage the canonical pathway pathway canonical ( the engage that CpG as acti such RelB-p50 stimuli by for vation allow and DCs) in (found characterize state activity steady another pathway noncanonical and expression RelB ( pathway noncanonical by such as stimuli LTactivation for allows RelB-p52 and in MEFs) (found state one steady noncanonical characterizes and activity pathway analysis expression RelB Our constitutive low signals. that indicates canonical NEMO-dependent I by by bound regulated be may RelB p50, with dimer in a pathway; noncanonical the by control to subject is RelB p52, or ( pathways canonical or noncanonical NF- the which in model a the in lation ( model upon TNF stimu nuclear translocation revealed into also cells single computational the by of Overexpression transduced a protein fusion retrovirally RelB-GFP predicted and DCs observed as in (twofold) dimer dimer RelB-p52 RelB-p50 the than the rather activated (sevenfold) primarily signaling canonical that ( analysis depletion way and render it responsive to TLR agonists. Antibody supershift and cally, as bypredicted the model, to push RelB into synergisti the path canonical function activity pathway noncanonical and expression 0 Supplementary Fig. 4e Fig. Supplementary 10 These iterative computational-experimental studies support support studies computational-experimental iterative These 1 Time (h) α 10 -RelB

II23a 2 II23a

II23a (fold) II23a (fold) 10 140 100 200

(% input) 70 3 0.01 0.02 0.03 0 0 10 n 4 0 = 3). * = 3). aDV 0 1 0 3 1 0 10 0. Pam CD80 Traf3 0 Time (h) Time (h) – 1 5 ** ** 10 RelB CpG A CpG 1 3 NCE ONLINE PUBLIC ONLINE NCE 10 CpG CSK 1.5 ** ** ** P 3 2 −/− < 0.05 and ** and 0.05 < 10 Fig. 4 Fig. 4 3 CpG Me 3 ** * 8 8 context but not in control cells ( cells control in not but context 10 d 4 10 Supplementary Fig. 4d Fig. Supplementary ). CD86 d 0 10 and and κ f 1 B protein RelB may function in either either in function may RelB protein B Relb WT 10

–4 –2 Phenotype 2 0 2 4 6 8 10 Supplementary Fig. 4c Fig. Supplementary –/ P 3 – 10 < 0.01. 0.01. < 4 A WT versus WT versus Time (h) e TION Fig. 4 Fig.

Relb –1.0 (relative tomaximumintensity) f 1 0 nature immunology nature Re Relb ). In a dimer with p100 p100 with In a dimer ). WT Cd40 Tnf κ 2 8 −/− l B CpG Cxcl2 –/ 0 4 ). High constitutive constitutive High ). –/ – α BMDCs. Data shown shown Data BMDCs. Relb Relb – Re and I and Expression 8 1 β II1b l –/ that the engage –/ –/ 24 – Fig. 4 Fig. – – 1 0 Pam WT ) confirmed confirmed ) κ 2 8 II23a B i. 3 Fig. 0 4 e 3 ε CSK Relb ). Rel and is is and 8 1 –/ 4 –/ 24 – 1.0 – c ). ). - - - -

© 2012 Nature America, Inc. All rights reserved. the signaling system. Whereas RelA and c-Rel protein expression expression protein c-Rel and RelA Whereas system. signaling the within relationship their investigated we programs, expression gene ( activating in c-Rel and of RelB both RelB and c-Rel. Quantitative RT-PCR validated the requirements RelB and c-Rel between dimers. specificities interaction an DNA suggesting in overlap genes), RelB-independent and RelB-dependent (of tinuous of spectrum RelB rather dependency than two distinct classes ( dependency RelB of degree increasing in sorted was genes of order the and DCs, null and wild-type between calculated were induction in fold phenotypes Expression TLR ligands. with stimulated BMDCs in genes c-Rel–RelB-dependent severely most 50 the of type of genes, we RelB these the in expression pheno examined activating in ( downregulated substantially were that Pam and CpG ligands TLR by RelB activated profiling and expression c-Rel Genome-wide DCs. in doubly-deficient expression gene examining matu by DC program the ration regulating in functions overlapping have RelB and ( cultures marrow CD11c in defects overt show not did outs cells stromal splenic in RelB to ascribed previously sequences unconventional the than rather c-Rel, NF- canonical known the with ated ( assay immunoprecipitation chromatin of regions. regulatory vivo these In with interact directly can RelB-p50 that κ the to DNAcontaining probes bound RelB-p50 activated In EMSAs, in reduced were and activation RelB induced Tnf Fig. Supplementary in CD40 and CD80 CD86, MHCII, ers Pam ligand mark activation of DC expression surface TLR2 reduced observed indeed the the or with CpG stimulation ligand After TLR9 markers. activation DC or of regulators expression the controls it whether wondered we maturation, DC during PAMPs by induced is RelB-p50 that Given maturation DC TLR-induced in cooperate c-Rel and RelB s.e.m.; bars, (error experiments independent three of representative are shown Data EMSA. by monitored BMDCs gene-deficient indicated in LPS by induced RelA and Rel of BMDCs. from indicated prepared cell extracts 6 Figure nature immunology nature were similarly abundant in and BMDCs wild-type those lacking RelB, Fig. 5 Fig. Supplementary Fig. 5c Fig. Supplementary Fig. B sites found in the the in found sites B α Relb Given overlapping functions of c-Rel and RelB in regulating DC DC regulating in RelB and c-Rel of functions overlapping Given We noted that RelB bound to consensus a -tubulin Tnf −/− and and c-Rel RelB RelA

BMDCs, relative to wild-type (WT) cells. ( cells. (WT) wild-type to relative BMDCs, 5 and and transcripts compared by quantitative RT-PCR with mRNA collected from wild-type and and wild-type from RT-PCRcollected mRNA with quantitative by compared transcripts e f , we observed recruitment of RelB to the promoter regions regions promoter the to RelB of recruitment observed we ,

and and and and Tnf RelB RelB may in cRel mediate functions DCs. ( Il23a Relb

and and –/– Il23a Supplementary Table2 Supplementary Supplementary Table 1 Supplementary WT , correlated with the kinetics of CpG or Pam or CpG of kinetics the with correlated , Rel Il23a –/– genes after DC maturation with CpG using the the using CpG with maturation DC after genes

Supplementary Fig. 5d Fig. Supplementary Rel –/–

Relb were identified in this analysis as regulated by regulated as analysis this in identified were 5a Tnf –/–

, ). b and and aDV c b ). Expression of proinflammatory genes, genes, proinflammatory of Expression ). Relb mRNA (AU) 0. 0. 1. A Il23a 3 Cxcl2 0 4 8 2 NCE ONLINE PUBLIC ONLINE NCE 2 CSK 9

or MEFs or WT Re * l –/ promoters ( promoters

κ – 4 , , ). To delineate the contribution contribution Tothe ). delineate ). This analysis revealed a con a revealed analysis This ). revealed a group of 157 genes genes 157 of group a revealed B pathway effectors, RelA and and RelA effectors, pathway B Cd40 Fig. 5 Fig. Time (h) Time (h) α Relb κ ), we tested whether c-Rel c-Rel whether tested we ), 2 + -tubulin served as served -tubulin a ( control. loading and and B B site sequences 0 c cell generation in bone bone in generation cell . Because single knock single Because . ) ) NF- Rel d Relb −/− a 3 ). 0 Il1b CSK ) Immunoblot ) for RelA, RelB Immunoblot and c-Rel of whole- 0 −/− Fig. 5 Fig. DCs ( DCs 0. κ 0. −/− WT B DNA binding activities of RelB, c-Rel c-Rel RelB, of activities binding B DNA 5 WT A 5 Relb gene expression expression gene TION 1 1 4 1 1 DCs ( DCs

for 24 h, we we h, 24 for inflammatory inflammatory c Rel .5 .5 ), indicating indicating ), −/− Fig. 5 Fig.

B 0 0 0 0 BMDCs BMDCs 2 Fig. 5 Fig. 8 Re Rel .5 .5 associ 3 Relb CSK l a b n –/ 1 1 1 1 –/ and and – = 4). * = 4). – −/− b .5 Rel .5 4 ). ). ------A

0 0 0 0 P Re .5 genes in in genes responsive gene expression by comparing the expression of RelB target tion programs. tion activa DC in c-Rel of mediator downstream a as acts RelB which in cytokine ( inflammatory expression and marker surface of reductions showed NF- program. gene target only the signaling pathway that RelB is responsive to but also the RelB production, not determines of RelB partner dimerization TNF the that suggest studies in RelB for role a GM-CSF and Bcl-xl expression has been reported and RelA-p50, to similarly 1 factor (SDF-1 cell–derived stromal and (BLC) chemokine cell blastoid lympho B (ELC), chemokine lymphoid ligand EBI1 (SLC), secondary chemokine tissue as such organogenesis, in lymphoid involved secondary chemokines regulating in function RelB-p52–specific of range broader a RelB-p52 recognize allows to that DNA with contact base additional an makes distinct may be p52 and of RelB- RelB-p50 Similarly, the DNAcharacteristics interaction specificities of protein-interaction studies to previous relevance I to surfaces molecular ferent system. signaling by steady state is of the effectors determined physiological the NF- relevant functionally are they whether pathways; nonca signaling and nonical canonical the of effectors potential are RelB and RelA both Thus, pathway. signaling noncanonical the of effector an RelA I of recent the discovery with Conversely, stimuli. maturation to exposure upon pathway canonical the via activity RelB-p50 of activation rapid for I the by inhibited is p50 RelB- DCs, immature in nuclear mostly is which RelB-p52, dimer. Unlike RelB-p50 the of formation in but dimer RelB-p52 expected the in only not resulted activity pathway noncanonical steady-state elevated and increased, expression RelB differentiation DC DCs. During in pathway canonical the of effector an also is RelB that here showed we However, organogenesis. lymphoid secondary in factor transcription a as RelB-p52 role on its pathway, based noncanonical the of NF- canonical effectors as considered previously been had c-Rel and RelA DISCUSSION < 0.01. .5 WT l –/ 1 1 Why then, would DCs use RelB as an effector of the canonical canonical the of effector an as RelB use DCs would then, Why dif present RelB-p52 and RelB-p50 that imply observations Our – 1 1 κ .5 b B signaling pathway B along with RelA and signaling One possibility c-Rel? c-Rel .5 ) ) Amount α 0 0 Re ) 0 0 Rel l 20 .5 –/ Rel – Rel , .5 2 −/− Supplementary Fig. Supplementary 1 1 b 9 b –/ . In contrast, RelB-p50 interacts with DNA sequences sequences DNA with interacts RelB-p50 contrast, In . 1 1 κ κ –/ – .5 DCs and and DCs B – B signaling pathway, and RelB as the effector of the the of effector the as RelB and pathway, signaling B .5 δ 21 32 , 2 , 2 3 , chronic inflammatory conditions rendered rendered conditions inflammatory chronic , 3 . RelB residue Arg125 in the RelB-p52 dimer dimer RelB-p52 the in Arg125 residue . RelB deficient DCs may also be defective in RelB- in defective be may also DCs deficient c-Rel– whether tested We therefore stimuli. sure of differentiating cells to c-Rel–inducing ated but immature DCs may the reflect expo suggests that expression of RelB in differenti by controlled architecture circuit fact feed-forward The c-Rel. in is expression, RelB namely pathway, canonical the by control cate that one of of RelB the key determinants in response to LPS ( RelB DNA-binding in activity diminished severely in resulted ( BMDC wild-type in ~40% by reduced were ( expression tein Rel κ Rel B proteins, I proteins, B −/− −/− BMDCs exhibited RelB decreased pro κ Relb B proteins, providing physiological physiological providing proteins, B κ

6a B sites. This may account for the the for account may This sites. B −/− , b ). These data support a model model a support data These ). DCs. Indeed, Indeed, DCs. κ B i. 6 Fig. α Fig. 6 Fig. and I and Fig. Fig. 6 a 34 c ). ). b κ , ). ). These data indi ticles e l c i rt A 3 ). This reduction reduction This ). B Rel 5 Relb . . Together, these ε Rel , which allows allows which , Rel −/−

activation of activation transcripts transcripts −/− −/− relative to to relative BMDCs BMDCs BMDCs BMDCs 30 , κ 3 α 1 B ------ .

© 2012 Nature America, Inc. All rights reserved. r R Published online at The authors declare no competing interests.financial with contributions from all authors. work and J.P. the bioinformatic analysis. V.F.-S.S. and A.H. wrote the manuscript from E.I.Z. and A.H. J.D.-T. and J.D.K. outcarried the computational modeling workexperimental with assistance from J.Q.H., T.Y., R.F. and M.A., and guidance V.F.-S.S. and A.H. the designed study. V.F.-S.S. and M.M. outcarried all (A.H.), AI090935 (A.H.), GM085325 (J.P.) and AI081923 (E.I.Z.). for modeling advice, and C. Brown and M. Karin (University of SanCalifornia Diego) and recombinant proteins, S. A. Basak, Wu, P. Loriaux, R. Tsui for computational We thank Z. Tao and G. Ghosh (University of SanCalifornia Diego) for plasmids Note: Supplementary information is available in the codes. Accession the in v available are references associated any and Methods M of development the for therapeutics. also DC-mediated but studies mechanistic for only not hypotheses generate to serve may systems signaling of analysis tive quantita a and modeling kinetic such, As TLRs. by DC-specific activation as RelB such properties, type–specific cell qualitative ingly control of the state steady of seem may system a determine signaling quantitative molecular type–specific cell that indicate type–specific results The mechanism. cell other any or specificities interaction protein type–specific cell invoke to need no was There control. RelB DC-like to MEFs produce engineering by genetically experimentally NF- the of lation the to shift sufficient constants ( rate kinetic two are differences found biochemical and key the DCs, that and MEFs types, cell two in mechanisms control RelB- containing dimers. We contrasted the steady-state and NF- dynamic multiple of regulation and tion genera the for accounts that model kinetic first the knowledge our I NF- the of control homeostatic and dynamic NF- the of models mathematical refined Iteratively networks. signaling of regulation the studying for tool a as itself lends equations, rate kinetic of terms in reactions cal PAMP exposure. transient to response in program activation and in DC biology ensures irreversible execution of a terminal maturation I DNA by the I for by I inhibited is RelB-p50 Although c-Rel. or RelA of that from distinct is RelB of specificity. cRel-p50 versus control dynamic stimulus-responsive is the that RelB-p50 possibility A second of question the address to c-Rel but required are tools other thus, expression; programs, RelB control to out turned gene RelB-dependent and c-Rel–dependent between overlap suggests profiling transcriptomic Our RelA. or Rel by c- controlled those from are distinct genes target RelB-p50 that is s e l c i rt A CO AUTH Ac e κ Relb e

eprints and permissions information is available online at at online available is information permissions and eprints p ethods r B proteins in fibroblasts in proteins B Mathematical Mathematical modeling, which we used here to biochemi describe r Traf3 k MPETI s i n i nowl κ o t mRNA synthesis and NIK half-life); a threefold increase was was increase threefold a half-life); NIK and synthesis mRNA s B feedback control than RelA, which is efficiently stripped off off stripped efficiently is which RelA, than control feedback B O n / i −/− n R R

o d CON f e NG embryos. This study was supported by GM085763 (A.H.), GM071573

x t ed . h h FI e t m g κ κ TRIBUTI

p N me B l B . a h A α α p κ t GEO: microarray data, data, microarray GEO: NC t n in resting cells, it may make for a poorer substrate substrate for a it poorer may make cells, resting in 3 e p B signaling system. We confirmed this prediction prediction this We confirmed system. signaling B 6 r : ts / . We speculate that the involvement of RelB-p50 RelB-p50 of involvement the that We . speculate . / IA w in in silico ON w L w I . N S n a TERESTS 21 t u model from MEF-like to DC-like regu to DC-like MEF-like from model , r 22 e . c , 37– o m κ B–I 4 / d 1 o . In this study, we developed to to developed we study, this In . i κ f κ i n B dimers, namely RelA- and and RelA- namely dimers, B B system have addressed the the addressed have system B d e G r o / n 1 S l 0 i E n . κ 1 e 3 0

B RelA-p50 dimer by dimer RelA-p50 B v 4 3 e 9 8 r h 9 s / t i n 0 t o p i n . . 2 :

/ o 4 / w f 4

t w 6 h . w e

. p n a

a p t u e o r r . e n . c l i o n m e - - - - - /

31. 30. 29. 28. 27. 26. 25. 22. 12. 11. 10. 9. 8. 7. 6. 5. 4. 3. 2. 1. 24. 23. 21. 20. 19. 18. 17. 16. 15. 14. 13.

orasi P, yek RP & rv, . ifrnil neatos f Rel-NF- of interactions Differential R. Bravo, & R.P. Ryseck, P., Dobrzanski, the to contribute mechanisms distinct TwoT. Wirth, & B. Kistler,T., Lernbecher, G. NF- Bonizzi, the via regulation Transcriptional G. Ghosh, & G. Natoli, A., Hoffmann, J.Q. He, Waterfield, Jin, M., Reiley,W., I S.C. Sun, & M. Zhang, W., NF single A A. Hoffmann, & A. Caldwell, V.F., R., Tsui,Shih, hh V.F. Shih, Pomerantz, J.L. & Baltimore, D. Two pathways to NF- to pathways Two D. Baltimore, & J.L. Pomerantz, Yang,H. Hofmann, J., Mair, F., Greter, M., Schmidt-Supprian, M. & Becher, B. NIK signaling T. Kobayashi, L. Wu, Li, M. in relB of role The M. Gerloni, & S. Schoenberger, P., Castiglioni, M., Zanetti, Carrasco, D., Ryseck, R.P. & Bravo, R. Expression of relB transcripts during lymphoid Steinman, R.M. & Banchereau, J. Taking dendritic cells into medicine. TNF/ E.G. Pamer, & W.A. Kuziel, plasmacytoid C.A., T.P.,Salazar-Mather,N.V.,Biron, Serbina, and cells interferon-producing 1 type professional IPC: Y.J. immunity.Liu, of control the and cells Dendritic R.M. Steinman, & J. Banchereau, Wuerzberger-Davis, S.M. T. Li, S. Basak, dimeric multiple of V.F.activation Shih, and S., Generation Basak, A. Hoffmann, & Gasparini, C., Foxwell, B.M. & Feldmann, M. RelB/p50 regulates CCL19 production, of expression Differential S.W. Krause, & R. Andreesen, K., Mondal, C., Ammon, NF- of Modulation G. Natoli, & S. Pantano, S., Saccani, antigen- cell dendritic conditions ligation CD40 R. Thomas, & B.J. O’Sullivan, E.F.Lind, E. Derudder, NF- in Crosstalk S. Ghosh, & M.S. Hayden, A., Oeckinghaus, EMBO J. EMBO I with complexes cells. lymphoid in (1994). RelB of activation constitutive (2004). specific of module. signaling 203 pathway. signaling (2004). Tpl2 the of component signaling. non-canonical and canonical (2011). (2002). Res. FoxP3 and Th2 a to Th17/Th1 helper T a from by transfer of RelB-silenced bone marrow dentritic cells is associated with a change effector responses. of (2011). immune development the cell-mediated for and required is cells cells T T in not but cells dendritic in development. (1998). 839–847 aha b rgltn epeso o te iae IKK formation. tissue lymphoid secondary and cell B proper for kinase the of expression differentiation. regulating by pathway USA Sci. Acad. specificity. signaling cytokine-receptor and pathogen- their determines (2007). 369–381 NF- of activation sustained through function presenting antigens. soluble of p100. for roles critical Chem. activation: Biol. J. receptor lymphotoxin-beta and factor-alpha Immunol. Nat. edii cls u nt f ypodrltd CD8 lymphoid-related of not but cells dendritic interference. RNA through cells (2003). response. immune adaptive the regulating cells. antigen-presenting in dendritic expression Development specific development: organ (2007). 419–426 infection. bacterial against defense immune innate mediate cells dendritic iNOS-producing precursors. cell dendritic Nature rncito fcos ihn h NF- (2008). 3139–3150 the within factors transcription maturation. (2009). DC human promote to fails but (2000). to macrophages and dendritic cells. the transcription factor NF- dimers. (2002). 5491–5498 , 2413–2418 (2006). 2413–2418 ,

59 et al.

t al. et 392 , 197–205 (2010). 197–205 , Mol. Cell Mol. t al. et et al. et

et al. et aDV 13 et al. et Immunity et al. et t al. et Immune modulation and tolerance induction by RelB-silenced dendritic , 245–252 (1998). 245–252 , κ , 4608–4616 (1994). 4608–4616 , irRA mdlt te ocnncl rncito fco NF- factor transcription noncanonical the modulate MicroRNAs

t al. et et al. et bnig ie b Rl:5 dimers. RelB:p52 by sites binding B 118 Rescue of TRAF3-null mice by p100 NF- p100 by mice TRAF3-null of Rescue eB s seta fr h dvlpet f yli-eae CD8 myeloid-related of development the for essential is RelB Suppression of ongoing experimental autoimmune myasthenia gravis myasthenia autoimmune experimental ongoing of Suppression Immunity

t al. et A 278 A fourth I fourth A 12 Nat. Immunol. Nat.

Dendritic cells require the NF- the require cells Dendritic 106 NCE ONLINE PUBLIC ONLINE NCE iei cnrl f eaie edak euaos f NF- of regulators feedback negative of control Kinetic κ

, 1221–1231 (1993). 1221–1231 , B 11 , 695–708 (2011). 695–708 , Oncogene RelB/p50 dimers are differentially regulated by tumor necrosis tumor by regulated differentially are dimers RelB/p50 ciain f IKK of Activation , 23278–23284 (2003). 23278–23284 , α

RF i a rtcl atr o dnrtc el auain and maturation cell dendritic for factor critical a is TRAF6 19 J. Immunol. J. , 9619–9624 (2009). 9619–9624 , , 1563–1574 (2003). 1563–1574 , determine pools of constitutive and inducible NF- inducible and constitutive of pools determine , 59–70 (2003). 59–70 ,

19 et al. κ Annu. Rev. Immunol. Rev. Annu. κ B protein within the NF- the within protein B , 353–363 (2003). 353–363 , B during human mononuclear phagocyte differentiation

25 Nuclear export of the NF-

11 , 6706–6716 (2006). 6706–6716 ,

J. Immunol. J. 181 , 799–805 (2010). 799–805 , Biochem. Biophys. Res. Commun. , 354–363 (2008). 354–363 , κ sgaig system. signaling B α Cell Res. Cell agt ee dpns n recognition on depends genes target A TION

n. Y cd Sci. Acad. NY Ann. 178 u. . Immunol. J. Eur.

κ o. el Biol. Cell. Mol. + 23 regulatory T-cellprofile. regulatory B2 pathway for cross-presentation for pathway B2

. x. Med. Exp. J.

, 5480–5487 (2007). 5480–5487 , α 21 , 275–306 (2005). 275–306 , + κ edii cells. dendritic nature immunology nature B signaling module. signaling B , 86–102 (2011). 86–102 , MO J. EMBO κ κ MO J. EMBO κ B. B inhibitor I κ B deficiency.B B activity by exchange of exchange by activity B κ Mol. Cell Mol. Immunity κ α B kinase is an essential an is kinase B κ B. uig macrophage during B signaling pathways. signaling B o. el Biol. Cell. Mol.

κ . Immunol. J.

208 sse fr both for system B

24 13 23 39

987 κ

6040–6048 , 4060–4069 ,

4202–4210 , 1917–1929 , 2215–2223 ,

B 10 34 268 Immunity J. Exp. Med. Exp. J. α Nature 249–257 , κ , 693–695 , is required Proc. Natl. Proc. , 188–200 , B activity.B Cell , 99–105 Inflamm. κ B/RelA

128 168 449 28

κ κ κ α 9 B B B - , , , , ,

Bergqvist, S. promoter Csf2 the to RelB of Recruitment D.L. Longo, & P. Ghosh, C.Y., Sasaki, responses immune in defects Multifocal R. Bravo, & H. Yang, G., F.,Warr,Weih, a of structure X-ray G. Ghosh, & D. V.Y., Vu, Wang, D.B., Huang, A.K., Moorthy, A.J. Fusco, factor. enhances RelA-mediated transcription of granulocyte-macrophage colony-stimulating mice. RelB-deficient in sites. NF- modes. distinct two with Natl. Acad. Sci. USA Sci. Acad. Natl. κ B p50/RelB/DNA complex reveals assembly of multiple dimers on tandem on dimers multiple of assembly reveals complex p50/RelB/DNA B J. Mol. Biol. Mol. J. J. Biol. Chem. Biol. J. et al. et et al. NF- Kinetic enhancement of NF-

373

κ 106 286 B p52:RelB heterodimer recognizes two classes of classes two recognizes heterodimer p52:RelB B , 723–734 (2007). 723–734 , J. Immunol. J. EMBO Rep. EMBO

, 19328–19333 (2009). 19328–19333 , , 1093–1102 (2011). 1093–1102 , aDV A NCE ONLINE PUBLIC ONLINE NCE

158

10 , 152–159 (2009). 152–159 , , 5211–5218 (1997). 5211–5218 , κ BxDNA dissociation by I A TION

κ B κ α B sites B . Proc. κ B 41. 40. 39. 38. 37.

Hoffmann, A., Levchenko, A., Scott, M.L. & Baltimore, D. The I The D. Baltimore, & M.L. Scott, A., Levchenko, A., Hoffmann, I A. Hoffmann, & C.S. Huang, S.L., Werner, S., Basak, J.D., Kearns, E.L. O’Dea, inducer than rather amplifier an as UV A. Hoffmann, & J.D. Kearns, E.L., O’Dea, Werner, S.L., Barken, D. & Hoffmann, A. Stimulus specificity of gene expression programs NF- NF- of module: temporal control and selective gene activation. gene selective and control temporal module: expression. gene inflammatory NF- control to feedback negative provides determinedtemporalby control IKKofactivity. (2002). κ B activity. B κ B activity. B et al. et Mol. Syst. Biol. Syst. Mol. A homeostatic model of I of model homeostatic A Mol. Cell Mol.

30 J. Cell Biol. Cell J. , 632–641 (2008). 632–641 ,

3 , 111 (2007). 111 , κ

B oscillations, signaling dynamics, and dynamics, signaling oscillations, B 173 κ Science B metabolism to control constitutive control to metabolism B , 659–664 (2006). 659–664 ,

309 Science , 1857–1861, (2005). ticles e l c i rt A κ

B-NF- 298 , 1241–1245 , κ B signaling B κ Bepsilon © 2012 Nature America, Inc. All rights reserved. MgCl mM NaCl,1.5 mM 4207.9, pHHEPES mM EDTA;20 mMC:Buffer 0.1and extraction buffer salt (Buffer high A: 10 mM HEPES by pH prepared 7.9, 10 mM were KCl, 0.1 mM BMDCs EGDA and BMDMs from extracts nuclear and malized for total protein or cell numbers beforeBiochemical immunoblot analyses. analysis. Cytoplasmic (Treestar). software FlowJo using analyzed were Data LSRII). ined for CFSE dilution and production of TNF and IL-2 by flow cytometry (BD later with 5 described as ratios cell DC:T indicated the CD4 CFSE-labeled with and cultured washed ester (CFSE) succinimidyl DCs were (Sigma). with labeled carboxyfluorescein for 323-339–specific ovalbumin transgenic ( of B6.Cg-Tg from spleens Technologies) Stem Cell T cells (5 × 10 for alone 2 or h medium at 37 Naive °C. (Anaspec) CD4 (OT-II) 339 peptide 200 with pulsed were row DCs in cocultures. cell Antigen presentation DC–T software. FlowJo with performed was analysis data and were in cells a acquired either or FACSCalibur an C6 (BD Accuri Biosciences) anti-IA and (GL-1) anti-CD86 (16-10A1), anti-CD80 (3/23), anti-CD40 (HL3), anti-CD11 Pharmingen: BD from purchased were dead cells and indicated antibodies for DC maturation analyses. All antibodies for FCS to exclude 5% D (7-AAD) 7-amino-actinomycin with containing stained were Cells min. 10 PBS in CD16/CD32 anti-mouse with blocked and Antibody staining and flow cytometry. protocol. manufacturer’s to according Biotec) (Miltenyi microbeads CD11c for were CD11c enriched tion with 10 mM EDTA pH 8.0 for 5 min. Single-cell suspension of splenocytes by incuba for at 30 min 37 °C followed Roche) D (2 mg/ml, collagenase with purification. DC Splenic studies. maturation DC spontaneous for BMDCs 9–11 day and maturation, DC TLR-induced investigate to used were BMDCs 6–7 described previously as analyses experimental to subjected and on days 3, were was 6 replaced collected cells medium and 8, and floating BMDC BMDCs. derive to IL-4 ng/ml 10 and GM-CSF ng/ml 20 with d 6–11 for them cultured or BMDMs derive to DMEM L929-conditioned with week 10 × 2 seeded prepared from were mouse femurs. Wesuspensions made from bone-marrow Primary MEFs were generated from E12.5–14.5 embryos. BMDMs and BMDCs Rel mice were generated by cross-breeding ing Rel Diego. San California, of University the of Committee Use San Diego. All procedures were approved by the Institutional Animal Care and maintained in pathogen–freespecific condition at the University of California, culture. cell and Animals conjugated anti-rabbit secondary antibody (18-8816) were from eBioscience. antibody (4994) was from Cell Signaling. Immunoprecipitation beads and HRP- Cancer Institute, Biological Resources Branch, Frederick, Maryland, USA. NIK Nationalterminuswere US from C p100 toantibody and p100/p52 p105/p50, Biotechnology. Cruz Santa from were (sc-18867) CD16/CD32 and (sc-1615) (sc-861), USF-2 (sc-6933), TRAF3 (sc-7606), I (sc-70), c-Rel (sc-226), IKK2 inhibitor (PS-1145) were from Sigma. Antibodies to RelA RelB(sc-372), 0.5 and (Invivogen), 500 ng/ml Pam3CSK4 (Invivogen), 100 ng/ml LPS (Sigma, B5:055) Reagents. METHODS ONLINE nature immunology nature −/− −/− Nfkbia 2 and Relb , 0.2 mM EDTA and 25% glycerol). Immunoprecipitation-immunoblotting µ g/ml LTßR agonist (Biogen) to stimulate cells. Cycloheximide and Cycloheximide cells. stimulate to (Biogen) LTßRagonist g/ml GM-CSF and IL-4 were from Peprotech. We used 0.1 0.1 used We Peprotech. from were IL-4 and GM-CSF −/− −/− Nfkbia µ M M OT-II for peptide 5 h in the presence of brefeldin A and exam mice were generated by cross-breeding by cross-breeding were generated mice Tnf 4 6 cells/well) were obtained by negative enrichment (>90% purity; bone marrow cells on 10-cm plate and cultured them for one one for them cultured and plate 10-cm on cells marrow bone −/− −/− with with Tnf Whole cell extracts were prepared in RIPA buffer and nor −/− κ Rel + Spleens were cut into small fragments and digested and digested were fragments cut into Spleens small cells by immunomagnetic cell sorting using MACS cell sorting by cells immunomagnetic B Wild-type and gene-deficient C57BL/6 mice were were mice C57BL/6 gene-deficient and Wild-type

Rel α −/− (sc-371), I (sc-371), −/− and µ g whole ovalbumin (Sigma, 5 5 (Sigma, ovalbumin whole g Relb Relb −/− Nfkbib Single-cell suspensions were collected −/− mice were generated by were mice cross-breed generated κ B mice. mice. α β 4 3 (sc-945), I (sc-945), β −/− . T cells were restimulated 72 h 72 restimulated were cells T . + TCR (Jackson Laboratory) and TCR Laboratory) (Jackson T cells (5 × 10 T cells , α GM-CSF–derived bone mar GM-CSF–derived Nfkbie Nfkbib -tubulin (sc-5286), (sc-5286), -tubulin TcraTcrb Rel b (AF6-120.1). Stained Stained (AF6-120.1). −/− −/− κ −/− B and Nfkbie ε and (sc-7155), IKK1 IKK1 (sc-7155), 4 )425 Cbn/J mice mice Cbn/J )425 2 4 Nfkbia . Typically, day day Typically, . T cells/well) at T cells/well) Nfkb2 µ Relb M OVAM 323- −/− µ −/− Nfkb2 −/− −/− CpG M β Tnf mice. mice. -actin -actin mice. −/− −/− + - - - - -

as indicated. withPrism software (GraphPad). Error bars were shown aseither s.d. or s.e.m. Statistics. ode15s solver. Detailed descriptions are included in the sented in this paper. Computational DC-specificMEF-andparameterizationsimulations experimentalonwerebased predata were performed in Matlabconstraints-based usingparameterization. the Refinementmodel(versionthe ofand 5.1) 3.1) (version model published previously a on based developed was equations kinetic action mass involving Computationalmodeling. experiments. independent four of representative experi independent (mean three ments of representative are shown data precipitation immuno chromatin and qRT-PCR levels. basal over induction fold derive with gene-specific and system ( realplex primers Mastercycler Eppendorf (Stratagene), reagent Mix Master PCR Green SYBR and using (Invitrogen), RT II oligo(dT) with SuperScript synthesis cDNA first-strand after performed was PCR FI as defined was phenotype RelB tiple expression viewer) expression tiple SAM paired by two-class analyzed and WT method by mloess normalized and was preprocessed were data chip raw The facility. expression core Biogem gene UCSD v.2 by performed Illumina the to hybridization and Labeling 0.1 with stimulated BMDCs (WT), wild-type using h) 24 and h 8 h, (1 experi ments time-course of set one from collected was RNA (Qiagen). Kit Mini analysis. expression Gene were acquired with a Zeiss Axio Z1 microscope. Transducedcells were selected withpuromycin hydrochloride(Sigma). Images reagent (Invitrogen) for 48 h. Supernatant wastransfection filtered 2000 and used toLipofectamine infect MEFs. with cells 293T into pCL.Eco with together transfected and methods standard by generated were constructs pBabe-puro transduction. gene Retrovirus-mediated ties of antibodies were confirmed ( (for RelB, p50 and p52) was performed as previously RelAdescribed DNA binding activity was the focus ( Similarly, nuclear extracts were preincubated with ingRelBthem with RelA and andc-Rel antibodies c-Rel( antibodies when were ablated of RelA and c-Rel–containing DNA binding describedactivities by preincubat analysis,EMSA, chromatin immunoprecipitation were performed previouslyas 42. FI example, for FI genotypes, different for calculated was (FI) tion ( analyses ing ( distance clustered Euclidian and with clustering induction fold hierarchical by maximum its to normalized were gene each in maturation listed DC are significant. TLR-elicited deemed during were induction 5% twofold least below at with rate Genes discovery false the at identified WT and points between 43. Rel

uz M.B. Lutz, Boonstra, A. Boonstra, ihy ue edii cls rm os bn marrow. bone mouse from cells dendritic pure highly dose and differential toll-like receptor ligation. ligation. receptor (2003). toll-like differential and dose antigen on dependency development: cell 2 and 1 type helper T directing in cells (1999). 77–92 −/− β -actin as normalization control to relate signals to those in MEFs or or MEFs in those to signals relate to control normalization as -actin Relb 21 2 Rel Supplementary Table 3 Table Supplementary Statistical significance was calculated by two-tailed Student’stwo-tailed by calculated was significance Statistical 6 −/− , n log in 4 4 −/− . In EMSAs focusing on RelB-DNA binding activity, nuclear extracts . The RelB phenotype was defined as FI as defined was phenotype RelB The . Supplementary Table 1 Table Supplementary Fig. 5 Fig. t al. et ± et et al. Relb s.d.). Quantification of mRNA and protein abundance are are abundance protein and mRNA of Quantification s.d.). 2 n dacd utr mto fr eeaig ag qatte of quantities large generating for method culture advanced An Flexibility of mouse classical and plasmacytoid-derived dendritic dendritic and plasmacytoid-derived of classical mouse Flexibility cl ars ete tmcus (p ad Pam and (CpG timecourse either across scale −/− f and and BMDCs during TLR stimulation time courses were were courses time stimulation TLR during BMDCs 4 Rel 7 Supplementary Table 2 Table Supplementary . Class pairing was defined by corresponding time time by corresponding defined was pairing . Class The RelA–RelB mathematical model (version 5.0) (version5.0) RelA–RelB The model mathematical −/− RNA extraction was performed with RNAeasy RNAeasy with performed was extraction RNA µ Relb M CpG (Invivogen) or 500 ng/ml Pam ng/ml 500 or (Invivogen) CpG M 2 2 Supplementary Fig. 2e n eprmna data experimental and 4 −/− ). Data analysis used the the used analysis Data ). 4 6 WT 5 implemented in the MeV implemented (mul program . Genes differentially regulated between between regulated differentially . Genes BMDCs. Differentially expressed genes expressed Differentially BMDCs. – FI – . In heat maps, expression values of of values expression maps, heat In . Fig. 3 Fig.3 Rel eB o Rl-F expressing RelB-GFP or RelB- b ). Antibody-shift ablation analysis −/− b and Relb . x. Med. Exp. J. ), the average fold induc fold average the ), Relb −/− . muo. Methods Immunol. J. SupplementaryFig. 3b Supplementary Note Fig. 5 Fig. WT . Quantitative (q)RT- Quantitative . ). −/− –FI 2 doi:10.1038/ni.2446 and and 0 2 e WT ht loe for allowed that 2 ). For phenotyp For ). Rel and the specifici ∆

FI , ( −/− 197 ∆ Rel Ct) method method Ct) , the c-Rel- the , Rel−/− , 101–109 101–109 , −/− 3 Relb 3 CSK CSK , and and ,

t -test 223 . −/− 4 4 ). ------) , .

ai, . Wek CH & obi, . irary rts n consequences. and truths Microarray J. Corbeil, & C.H. Woelk, R., Sasik, C.S. Cheng, Endocrinol. NF- (2011). by elements response interferon of repression

33 t al. et , 1–9 (2004). 1–9 , h seiiiy f nae mue epne i efre by enforced is responses immune innate of specificity The κ B p50. p50. B c. Signal. Sci.

4 . Mol. J. , ra11 ra11 , 47. 46.

Saeed, A.I. Saeed, applied microarrays of analysis Significance G. Chu, & Tusher,V.G.,R. Tibshirani, (2006). (2001). o h inzn rdain response. radiation ionizing the to et al. et TM4 microarray software suite. software microarray TM4 rc Nt. cd Si USA Sci. Acad. Natl. Proc. Methods Enzymol. Methods nature immunology nature

411 5116–5121 , , 134–193 ,