Transforming growth factor beta-activated kinase 1(TAK1)-dependent checkpoint in the survival of dendriticcells promotes immune homeostasis and functionYanyan Wanga,1, Gonghua Huanga,1, Peter Vogelb, Geoffrey Nealec, Boris Reizisd, and Hongbo Chia,2

Departments of aImmunology and bPathology, and cHartwell Center for Bioinformatics and Biotechnology, St. Jude Children’s Research Hospital, Memphis,TN 38105; and dDepartment of Microbiology and Immunology, Columbia University, New York, NY 10032

Edited by Kenneth M. Murphy, Washington University, St. Louis, MO, and accepted by the Editorial Board December 27, 2011 (received for review September22, 2011)

Homeostatic control of dendritic cell (DC) survival is crucial foradaptive immunity, but the molecular mechanism is not well de-fined. Moreover, how DCs influence immune homeostasis understeady state remains unclear. Combining DC-specific and -inducibledeletion systems, we report that transforming growth factor beta-activated kinase 1 (TAK1) is an essential regulator of DC survivaland immune system homeostasis and function. Deficiency of TAK1in CD11c+ cells induced markedly elevated apoptosis, leading tothe depletion of DC populations, especially the CD8+ and CD103+

DC subsets in lymphoid and nonlymphoid tissues, respectively.TAK1 also contributed to DC development by promoting the gen-eration of DC precursors. Prosurvival signals from Toll-like recep-tors, CD40 and receptor activator of nuclear factor-κB (RANK) areintegrated by TAK1 in DCs, which in turn mediated activation ofdownstream NF-κB and AKT-Foxo pathways and establisheda gene-expression program. TAK1 deficiency in DCs caused a mye-loid proliferative disorder characterized by expansion of neutro-phils and inflammatory monocytes, disrupted T-cell homeostasis,and prevented effective T-cell priming and generation of regulatoryT cells. Moreover, TAK1 signaling in DCs was required to preventmyeloid proliferation even in the absence of lymphocytes, indicatinga previously unappreciated regulatory mechanism of DC-mediatedcontrol of myeloid cell-dependent inflammation. Therefore, TAK1orchestrates a prosurvival checkpoint in DCs that affects the homeo-stasis and function of the immune system.

innate immunity | immune tolerance

Dendritic cells (DCs) are a heterogeneous population of im-mune cells specialized to capture, process, and present

antigens to T lymphocytes (1). DCs can be divided into threemain populations: conventional DCs (cDCs), type I IFN-secretingplasmacytoid DCs (pDCs), and migratory DCs (2–4). cDCs can befurther divided into CD8+ and CD8– DC subsets in lymphoidorgans, whereas the CD103+ DC subset in nonlymphoid organsserves as a functional equivalent of CD8+ cDCs in lymphoidorgans (2–4). Homeostasis of DCs is maintained through a dy-namic balance of three main mechanisms. First, DCs are con-tinuously replenished by blood-borne precursors that are furtherderived from macrophage and DC precursor (MDP) and com-mon DC precursor (CDP) populations in the bone marrow (BM)(2, 5). Second, ∼5% of DCs are dividing at any given time. Al-though limited, the self-renewal ability in situ contributes to themaintenance of the DC population (6–8). Finally, apoptotic celldeath of DCs is a central mechanism for down-modulating DCnumbers to ensure immune homeostasis (9). Recent studies haveprovided important insight into the regulation of DC survival byBcl-2 family members, including the antiapoptotic members Bcl-2 and Bcl-xL and the proapoptotic Bim (10–12). Expression ofthese Bcl-2 family proteins is dynamically regulated by signalstransduced from prosurvival cytokines, ligands for Toll-likereceptors (TLRs), and other extracellular stimuli, thereby dic-tating the decision between survival and death (9). However, it is

largely unexplored how the expression of Bcl-2–related proteinsand apoptosis of DCs are regulated by the intracellular signalingnetwork.Despite a well-established role for DCs to induce antigen-

specific immune activation and tolerance (13, 14), how DC sur-vival affects immune homeostasis in the steady state, in the ab-sence of encounter with foreign antigens, remains incompletelyunderstood. Enhancing DC survival through transgenic expres-sion of the baculoviral caspase inhibitor p35 or deletion of Fasleads to chronic lymphocyte activation and systemic autoimmunity(15, 16), whereas increasing DC numbers by FMS-like tyrosinekinase 3 ligand (FLT3L) administration protects mice from au-toimmunity via inducing regulatory T-cell (Treg) accumulation(17–19). Conversely, constitutive ablation of cDCs through theexpression of diphtheria toxin in CD11c+ cells has been reportedto cause fatal T-cell–mediated autoimmune disease (20) ora myeloid proliferative syndrome without obvious alterations ofT-cell homeostasis (21). Mechanistically, peripheral DCs havebeen implicated in promoting homeostatic proliferation andsurvival of T cells (22–24) as well as inducing a low-level tonicsignaling in naive T cells to program their responsiveness toforeign antigens (25). Moreover, although a direct role for DCsto promote Treg expansion and accumulation has been estab-lished by recent elegant studies (17, 26), how this effect con-tributes to immune homeostasis is not fully understood (20, 21,27, 28). Because many of the previous studies have used non-physiological approaches to modulate DC survival, how endog-enous regulators of DC survival influence immune homeostasisand function remains to be established.Transforming growth factor beta-activated kinase 1 (TAK1)

(encoded by Map3k7) is arguably the most widely used MAPkinase kinase kinase in the immune system (29). The essentialrole of TAK1 in host immune defense was first demonstrated inDrosophila (30, 31) and later confirmed in murine cells followingstimulation through TLRs and proinflammatory cytokine re-ceptors (32, 33). TAK1 also mediates the intracellular sensorpathway mediated by nucleotide-binding oligomerization domain

Author contributions: Y.W., G.H., and H.C. designed research; Y.W., G.H., P.V., and G.N.performed research; B.R. contributed new reagents/analytic tools; Y.W., G.H., P.V., G.N.,and H.C. analyzed data; and Y.W., G.H., and H.C. wrote the paper.

The authors declare no conflict of interest.

This article is a PNAS Direct Submission. K.M.M. is a guest editor invited by the EditorialBoard.

Data deposition: The microarray results reported in this paper have been deposited in theGene Expression Omnibus (GEO) database, www.ncbi.nlm.nih.gov/geo (accession no.GSE34417).1Y.W. and G.H. contributed equally to this work.2To whom correspondence should be addressed. E-mail: hongbo.chi@stjude.org.

See Author Summary on page 1834.

This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10.1073/pnas.1115635109/-/DCSupplemental.

www.pnas.org/cgi/doi/10.1073/pnas.1115635109 PNAS | February 7, 2012 | vol. 109 | no. 6 | E343–E352

IMMUNOLO

GY

PNASPL

US

1 (NOD1) and NOD2 (34, 35), but TLR8-induced activation ofNF-κB and JNK is independent of TAK1 (36). In lymphocytes,TAK1 is an essential component of antigen receptor signalingand promotes lymphocyte proliferation and survival and adaptiveimmune functions (33, 37–40). Moreover, TAK1 is critical forthe survival of hematopoietic stem cells and progenitors (41).These results indicate a cell context-dependent function forTAK1 in the immune and hematopoietic systems.Whereas a role for TAK1 in the initiation of innate immune

responses upon pathogen recognition is well established, itsrole in the homeostatic control of innate immune cells such asDCs has not been examined. To investigate the function ofTAK1 in DCs, we generated DC-specific TAK1-deficient miceand found that TAK1 was essential for the homeostasis of DCsby promoting their survival. Using an inducible deletion system,we further identified a direct role of TAK1 to actively maintainmature DCs and BM precursors. Moreover, TAK1 deficiency inDCs caused a myeloid proliferative disorder, disrupted T-cellhomeostasis under steady state, and prevented effective T-cellpriming and Treg generation. Our studies demonstrate that aTAK1-mediated checkpoint in DC survival has a key role inthe homeostasis and function of the innate and adaptiveimmune systems.

ResultsCell-Autonomous Role of TAK1 in Regulating DC Populations. To in-vestigate the function of TAK1 in DCs, we generated DC-spe-cific TAK1-deficient mice by crossing mice bearing floxed andnull alleles of the Map3k7 gene with transgenic mice expressingCre under the control of the CD11c promoter to generateMap3k7flox/null CD11c-Cre mice (called “Map3k7DC mice” here-after) (37, 42). TAK1 was efficiently deleted in splenic DCs butnot T cells from these mice (SI Appendix, Fig. S1A). Flowcytometry analysis of splenic DC populations in Map3k7DC miceshowed that the percentage and cell number of cDC (CD11c+

MHC-II+) and pDC (CD11clomPDCA-1+) populations weremarkedly reduced compared with those of wild-type (WT) mice(Fig. 1A). Within cDCs, the CD8+ subset was more profoundlyaffected than the CD8– (CD11b+) subset (Fig. 1A). Similardefects were observed in other lymphoid organs, includingmesenteric lymph nodes (MLNs) and thymus (Fig. 1B and SIAppendix, Fig. S1B). Moreover, nonlymphoid organs, such as theliver, had lower DC frequency with a preferential reduction ofthe CD103+ DC subset (Fig. 1C). This result is consistent withthe recent findings that nonlymphoid CD103+ DCs and lym-phoid organ-resident CD8+ cDCs are developmentally andfunctionally related (8, 43–45). These results reveal a key role for

Fig. 1. TAK1 has a key cell-autonomous role in regulating DC populations. (A) Flow cytometry of splenic cDC, pDC, CD8+ cDC, and CD11b+ cDC populations inWT andMap3k7DC mice. MHC-II, MHC class II. (Right) The proportion and cell numbers of the splenic DC populations in WT andMap3k7DC mice. (B and C) Flowcytometry of DC populations in the MLNs (B) and liver (C) of WT and Map3k7DC mice. (D) Flow cytometry of the contributions of spike BM-derived (CD45.1.2+)and WT or Map3k7DC donor BM-derived cells (CD45.2.2+) in the mixed chimeras after 2 mo of reconstitution. Data are representative of three to five in-dependent experiments. Data represent the mean ± SEM.

E344 | www.pnas.org/cgi/doi/10.1073/pnas.1115635109 Wang et al.

TAK1 to regulate DC populations in both lymphoid and non-lymphoid organs.To address whether the reduction of DCs in Map3k7DC mice

was an intrinsic defect, we generated mixed BM chimeras. Spe-cifically, we transferred WT or Map3k7DC BM cells (CD45.2.2+;donor) and WT BM cells (CD45.1.2+; spike) at a ratio of 1:1 intolethally irradiated WT (CD45.1.1+) recipients. After reconsti-tution, we analyzed splenic DC populations in the mixed chimeras.Compared with WT BM-derived donor cells, Map3k7DC donorcells in the chimeras contained a greatly reduced DC population(Fig. 1D). Therefore, TAK1 has a key cell-autonomous function inmaintaining the DC pool.

TAK1 Promotes DC Survival and Development but Not Proliferation.The overall size of DC populations is dependent upon the ratesof apoptosis and proliferation as well as replenishment fromDC precursors (2, 5, 9). We first measured caspase activity, ahallmark of apoptotic cell death (46). The percentage of caspase-positive DCs from Map3k7DC mice was significantly highercompared with that in WT cells (Fig. 2A). Further, terminaldeoxynucleotidyl transferase (TdT)-mediated dUTP nick endlabeling (TUNEL) assay also revealed elevated apoptosis ofMap3k7DC DCs (Fig. 2B).We next tested the possibility that the decreased DC pop-

ulation in Map3k7DC mice was partly ascribed to defective DCproliferation in situ (6–8). Notably, TAK1 is essential for mediatingcell cycle progression in lymphocytes (33, 37–40). We thereforemeasured DC proliferation in vivo using the Bromodeoxyuridine(BrdU) incorporation assay. DCs in WT and Map3k7DC mice in-corporated BrdU to a comparable degree (Fig. 2C), indicating thelack of a role for TAK1 in DC proliferation.Finally, we determined whether TAK1 regulated DC devel-

opment by acting on DC precursors. One of the earliest pre-cursor populations identified for DCs is MDPs (47–49). Because

CD11c-Cre mice do not allow deletion in MDPs, we crossedMap3k7flox/null mice with Rosa26-Cre-ERT2 mice (a Cre-ER fu-sion gene was recombined into the ubiquitously expressedRosa26 locus) to generateMap3k7flox/null Rosa26-Cre-ERT2 mice(called “Map3k7CreER mice” hereafter). Three days after treat-ment of WT and Map3k7CreER mice with tamoxifen in vivo, thepercentage of MDPs, defined by Lin–Sca-1–CSF1R(CD115)+ inMap3k7CreER mice was significantly decreased compared withthat in WT mice (Fig. 2D). Next, we cultured WT andMap3k7CreER BM cells with FLT3L in the presence of 4-hydroxytamoxifen (4-OHT). After 4 d of culture, Map3k7CreER

BM cells showed considerable defects to up-regulate CSF1R,a defining molecule for MDPs (49) (Fig. 2E). These data indicatean important role of TAK1 in the generation of MDPs, therebycontributing to DC development. Altogether, TAK1 is criticalfor the survival and development but not proliferation of DCs.

TAK1 Is Essential for Maintaining DC Survival by IntegratingProsurvival Receptor Signals. To directly investigate the effects ofTAK1 on the homeostasis of mature DCs, we analyzed DCs inWT and Map3k7CreER mice after short-term treatment with ta-moxifen. Acute systemic deletion of TAK1, although lacking celltype-specific precision, allowed us to exclude the compensatoryeffects that might have developed due to the continuous lossof TAK1 in DCs or precursors. Following in vivo tamoxifen-mediated TAK1 deletion, splenic cDCs were markedly reduced,with a more prominent reduction in the CD8+ subset. The pDCpopulation was also greatly reduced (Fig. 3A). Accordingly,caspase activation in DCs was elevated in these cells (Fig. 3B).The effect of TAK1 deletion on DC maintenance was fur-ther tested in vitro. BM cells from WT or Map3k7CreER mice(CD45.2+) were mixed with an equal number of WT (CD45.1+)spike cells, and after 7 d of culture with FLT3L to derive DCs,4-OHT was added. After culturing for 6 additional days, CD11c+

Fig. 2. Deletion of TAK1 results in excessive apoptosis of DCs and loss of MDPs. (A) Caspase activity in freshly isolated WT andMap3k7DC splenic DCs, assessedby staining with FITC-conjugated VAD-FMK. (Right) The proportion of caspase+ cells. (B) Measurement of apoptosis of WT and Map3k7DC splenic DCs byTUNEL staining. (C) Flow cytometry of BrdU incorporation into splenic DCs of WT and Map3k7DC mice after an in vivo BrdU pulse for 24 h. (Right) Thepercentage of proliferating splenic DCs in WT and Map3k7DC mice. (D) Flow cytometry of MDPs (Lin–Sca-1–CSF1R+) in the BM Lin– cells of WT and Map3k7CreER

mice after 3 d of tamoxifen treatment in vivo. (Right) The proportion of MDPs among BM Lin– cells. (E) BM cells fromWT andMap3k7CreER mice were culturedwith FLT3L in the presence of 0.5 μM 4-OHT. The expression of CSF1R and CD11c was analyzed by flow cytometry at day 4 (Left) and the percentage of CSF1R+

cells in CD11c– cells was calculated (Right). Data are representative of three independent experiments. Data represent the mean ± SEM.

Wang et al. PNAS | February 7, 2012 | vol. 109 | no. 6 | E345

IMMUNOLO

GY

PNASPL

US

cells from Map3k7CreER BM cells were selectively reduced (Fig.3C), associated with enhanced apoptosis (Fig. 3D), comparedwith WT or spike-derived cells. These results demonstrate thatone essential mechanism for the prosurvival function of TAK1 isthrough the maintenance of mature DC survival.The life span of DCs is determined by signals from pathogens

and T cells that are transduced through TLRs, CD40, and re-ceptor activator of nuclear factor-κB (RANK) (9, 10). To test theintrinsic requirement of TAK1 in mediating these prosurvivalreceptor signals, we generated FLT3L-derived DCs from WTand Map3k7DC BM cells and treated purified DCs with LPS,α-CD40, or RANKL for 2 d. In the absence of treatment, TAK1-deficient DCs had more spontaneous death compared with WTcells. LPS treatment resulted in enhanced survival of WT cells,but this effect was substantially reduced in Map3k7DC DCs.Moreover, TAK1-deficient DCs remained nearly completelyunresponsive to the survival effects induced by α-CD40 orRANKL stimulation (SI Appendix, Fig. S2). These findingsidentify TAK1 as an important sensor that links signaling activ-ities of extracellular receptors and survival of DCs.

TAK1-Dependent Pro- and Antiapoptotic Signaling Mechanisms inDCs.We dissected the biochemical and molecular mechanisms bywhich TAK1 regulates DC survival. We first measured the effectsof TAK1 deficiency on the activity of NF-κB, which is importantfor DC survival (50). Constitutive NF-κB activity, assessed byIκBα phosphorylation, was considerably diminished in freshlyisolated Map3k7DC splenic DCs (Fig. 4A), indicating a role forTAK1 to mediate NF-κB activation in DCs in vivo. Activity ofAKT (also known as protein kinase B, PKB), another anti-

apoptotic kinase in DCs (51), was also diminished in TAK1-deficient DCs. Accordingly, phosphorylation of the AKT down-stream target Foxo1 (Ser256), which exerts a potent inhibitoryeffect on Foxo1 activity, was decreased (Fig. 4A). The AKT-Foxoaxis is important for the expression of the proapoptotic moleculeBim (52). Indeed, Bim expression was up-regulated in TAK1-deficient DCs, whereas the level of the antiapoptotic Bcl2 wasnot altered (Fig. 4A). In addition, production of reactive oxygenspecies (ROS), which is under the control of TAK1 signaling inother cell types (53), was undisturbed in TAK1-deficient DCs (SIAppendix, Fig. S3). Therefore, TAK1 regulates the activities ofthe NF-κB pathway and the AKT-Foxo-Bim signaling axis, andthe interplay of these pathways likely contributes to the physio-logical function of TAK1 in mediating DC survival.To further understand the function of TAK1 in DCs, we used

a functional genomics approach to analyze TAK1-dependent genesignature. To obviate compensatory effects due to sustained loss ofTAK1, we purified splenic DCs from WT and Map3k7CreER miceafter acute tamoxifen treatment and used microarrays to comparetheir gene expression profiles. A total of 550 probe sets showedequal or greater than twofold change (with false discovery rate<0.05) between WT and TAK1-deficient DCs. TAK1-deficientDCs showed up-regulated levels of several proapoptotic factorsincluding Apaf-1, Bim (encoded by Bcl2l11), Caspase 6, andGadd45a and reduced expression of transcription factors Nfkb2and Relb. In addition, expression of several cytokines, chemo-kines, and their receptors was altered in TAK1-deficient DCs (Fig.4B). Real-time PCR analysis confirmed the altered expression ofselected genes (Fig. 4C). Interestingly, IL-10 and the IL-10 targetgene SOCS3 were found to be significantly up-regulated in DCs

Fig. 3. TAK1 is essential for the maintenance of mature DCs. (A) Flow cytometry of splenic cDC, pDC, CD8+ cDC, and CD11b+ cDC populations in WT andMap3k7CreER mice after 3 d of tamoxifen treatment. (Right) The proportion and cell numbers of the splenic DC populations. (B) Caspase activity in splenic DCsfrom tamoxifen-treated WT and Map3k7CreER mice, assessed with FITC-VAD-FMK. (Right) The proportion of apoptotic VAD-FMK+ splenic DCs. (C and D) BMcells from WT or Map3k7CreER mice (CD45.2+) were cultured with WT CD45.1+ BM cells at a 1:1 ratio in the presence of FLT3L, and 0.5 μM 4-OHT was added atday 7. The distribution (C) and apoptosis (D) of CD45.1– and CD45.1+ populations in the gated CD11c+ cells were analyzed by flow cytometry at the indicatedtime points. Data are representative of three independent experiments. Data represent the mean ± SEM.

E346 | www.pnas.org/cgi/doi/10.1073/pnas.1115635109 Wang et al.

from tamoxifen-treated Map3k7CreER mice (Fig. 4 B and C). DCsfrom Map3k7DC mice showed a similar increase in the expressionof IL-10 and SOCS3, as well as proapoptotic Bim and Gadd45a(SI Appendix, Fig. S4). Furthermore, elevated STAT3 phosphor-ylation was observed in DCs from Map3k7DC and tamoxifen-treated Map3k7CreER mice (SI Appendix, Fig. S5), thus indicatingan active autocrine IL-10 signaling in TAK1-deficient DCs.Therefore, TAK1 orchestrates a prosurvival program in DCs byregulating expression of apoptotic molecules and further linksthem with expression of immune response genes.

TAK1 Functions in DCs to Promote T-Cell Priming and Treg Generation.DCs are the most potent antigen-presenting cells (APCs) toactivate naive T cells. To investigate the role of TAK1 signalingin DCs to mediate T-cell–dependent immune responses, wetransferred antigen-specific CD8+ T cells from OT-I TCR-trans-genic mice (specific for the OVA257–264 peptide) into WT andMap3k7DC mice, followed by immunization with the cognate an-tigen. T cells isolated from Map3k7DC hosts contained fewer do-nor-derived antigen-specific T cells (Fig. 5A). We obtained similarresults for antigen-specific CD4+ T-cell responses using OT-IITCR-transgenic T-cells (specific for OVA323–339) (Fig. 5B).Moreover, OT-II T cells activated with TAK1-deficient DCs invitro were substantially impaired in their expansion (SI Appendix,Fig. S6A). These findings demonstrate that TAK1 signaling inDCs is required for the activation of antigen-specific naive T cells.In addition to T-cell priming, DCs have been recently shown

to promote Treg cell generation (17, 54–57). Treg cells eitherdevelop in the thymus, known as naturally occurring Treg (nTreg)cells, or are induced from naive T cells in the periphery undersubimmunogenic antigen stimulation, known as induced Treg(iTreg) cells (58). Whereas nTreg cells were undisturbed in thethymus of Map3k7DC mice, they were considerably reduced inperipheral lymph organs (Fig. 5C). Given the critical importance

of TAK1 signaling in Treg cell generation (37, 38), it remainedpossible that the reduction of nTreg cells could result from an off-target deletion of TAK1 in T cells in Map3k7DC mice. However,in mixed chimeras as described above (Fig. 1D), the nTreg pop-ulation derived from Map3k7DC BM cells developed normally(SI Appendix, Fig. S7), indicating a non-cell–autonomous defecttriggered by the abnormality of DCs.We next tested whether TAK1 regulated induction of iTreg cells.

We adoptively transferred naive OT-II CD4+ T cells into WT andMap3k7DC hosts, followed by s.c. antigen immunization or ad-ministration of the antigen in the drinking water. In both models,donor cells developed into a considerably smaller Foxp3+ popu-lation in Map3k7DC recipients relative to WT hosts (Fig. 5 D andE), indicating a role of TAK1 in DCs to mediate induction ofantigen-specific iTreg cells. This result was further confirmed by theimpaired ability of TAK1-deficient DCs tomediate iTreg inductionin vitro (SI Appendix, Fig. S6B). We conclude that TAK1 functionin DCs is critical for both naive T-cell priming and Treg generation.

TAK1 Deficiency in DCs Disrupts Homeostasis of T Cells and MyeloidCells. How DCs affect homeostasis of the immune systemremains incompletely understood. We therefore tested whetherdefective DC survival in Map3k7DC mice influenced T-cell ho-meostasis. Map3k7DC mice had reduced proportions of T cells,especially CD8+ T cells, in the thymus, spleen, and peripherallymph nodes (PLNs) (Fig. 6A). Moreover, splenic T cells fromMap3k7DC mice were partially impaired to produce IFN-γ andIL-2 upon acute polyclonal stimulation (Fig. 6B), indicative ofaltered homeostasis of T cells in the steady state.Despite diminished DC and T-cell populations, the spleen and

PLN sizes of Map3k7DC mice were markedly increased (SI Ap-pendix, Fig. S8A). Histological analysis showed that the structureof the secondary lymphoid organs was disrupted, associated withaltered distribution of CD3+ T cells. In addition, the spleen of

Fig. 4. TAK1 regulates the activities of NF-κB and AKT-Foxo signaling and the expression of apoptotic factors and immune response genes. (A) Western blotanalyses of p-IκB, p-AKT, p-Foxo1, Bim, and Bcl-2 expression in splenic DCs fromWT andMap3k7DC mice. Numbers below lanes indicate band intensity relativeto that of β-actin (loading control). (B) WT and Map3k7CreER mice were treated with tamoxifen for 3 d and RNA of splenic DCs was analyzed for the com-parison of expression profiles. A subset of genes differentially regulated in WT and TAK1-deficient DCs (false discovery rate < 0.05) is shown. Normalized geneexpression is displayed. (Scale bar shows SD from the mean.) (C) WT and Map3k7CreER mice were treated with tamoxifen for 3 d and RNA of splenic DCs wasanalyzed for the expression of the indicated genes. Data are representative of three independent experiments. Data represent the mean ± SEM.

Wang et al. PNAS | February 7, 2012 | vol. 109 | no. 6 | E347

IMMUNOLO

GY

PNASPL

US

Map3k7DC mice showed severe fibrosis, detected by trichromeand reticulin staining (Fig. 6C). Moreover, increased numbers ofMac2-reactive myeloid cells were evident in the mutant spleenand PLNs (Fig. 6C and SI Appendix, Fig. S8B). Flow cytometryanalysis of spleen, MLNs, and PLNs in Map3k7DC mice revealeda marked increase in Gr1+CD11b+ myeloid cells that likelycorresponded to neutrophils and Gr1+ inflammatory monocytes(Fig. 6D) (59). Analysis of additional markers in Map3k7DC

lymphoid organs confirmed the expansion of neutrophils (Ly6G+

Ly6C+) and monocytes (CD115+CD11b+), with the latter pop-ulation mainly consisting of the Ly6C+ inflammatory subsetrather than the Ly6C– subset (SI Appendix, Fig. S9) (59). Flowcytometry and blood counts identified a similar shift towardneutrophils and monocytes in peripheral blood ofMap3k7DC mice(SI Appendix, Fig. S10). Further, myeloid cell infiltration was alsoobserved in various nonlymphoid organs from Map3k7DC mice,including the liver, lung, kidney, and colon (SI Appendix, Fig. S11).In contrast, Map3k7DC BM contained largely normal populationsof neutrophils and monocytes (SI Appendix, Fig. S12). These

results collectively indicate the development of a myeloid pro-liferative syndrome in Map3k7DC mice.

Myeloid Proliferation Results from Loss of TAK1 in DCs Independentlyof Lymphocytes. We dissected the cellular mechanisms thatcaused defective homeostasis of myeloid cells inMap3k7DC mice.Such a disorder could reflect a direct role of TAK1 in myeloidcells (for example, by nonspecific deletion) or could be due todysregulated functions of T cells, especially Treg cells, or trig-gered by a mechanism that directly sensed the abnormality ofDCs in vivo. To distinguish these possibilities, we analyzed themixed BM chimeras composed of Map3k7DC donor and spikecells, as described above (Fig. 1D). No myeloid expansion wasobserved in the mixed chimeras that contained normal DC pop-ulations derived from the spike cells (Fig. 7A), suggesting that anormal DC population is important to prevent myeloid expansion.We next determinedwhethermyeloid proliferation inMap3k7DC

mice resulted from altered T-cell function or tolerance. In par-ticular, depletion of DCs has been implicated in the development

Fig. 5. TAK1 signaling in DCs is required for T-cell priming and Treg generation. (A and B) Antigen-specific OT-I CD8+ (CD45.1+) (A) or OT-II CD4+ T cells(Thy1.1+) (B) were transferred into WT and Map3k7DC mice followed by antigen immunization in the presence of IFA. At day 4 after immunization, DLN cellswere examined for the percentages of donor T cells. (C) Proportion of Foxp3+ nTreg cells among CD4+ T cells in the thymus, spleen, PLNs, and MLNs of WT andMap3k7DC mice. (Right) The proportion of Foxp3+ cells in CD4+ T cells of WT andMap3k7DC mice. (D) WT andMap3k7DC mice were transferred with naive OT-IICD4+ T cells (Thy1.1+) and immunized as described in B, and DLN cells were examined for the percentage of Foxp3+ cells among donor cells. (E) After adoptivetransfer of naive OT-II CD4+ T cells, WT and Map3k7DC mice were fed with water supplemented with OVA protein. Five days later, the spleen and PLN cellswere examined for the percentages of donor CD4+ T cells (Thy1.1+) and of Foxp3+ cells among donor cells. Data are representative of two to four independentexperiments. Data represent the mean ± SEM.

E348 | www.pnas.org/cgi/doi/10.1073/pnas.1115635109 Wang et al.

of autoimmune diseases that is dependent upon dysregulatedautoreactive T cells and/or defective Treg-mediated immune sup-pression (17, 20). However, T cells in Map3k7DC mice did notappear to be activated, indicated by the largely normal expressionof the homeostasis markers CD62L and CD44 and the absence ofelevated serum autoantibodies (SI Appendix, Fig. S13). To furtherdissect the contribution of T cells, we crossed Map3k7DC miceonto the alymphoid Rag1−/− background. Notably, Rag1 de-ficiency elevated the population of myeloid cells compared withthat in immunocompetent WT mice (Fig. 7B; compare with Fig.6D), probably triggered by the lymphopenic environment. How-ever, deletion of TAK1 in DCs on the same Rag1−/− backgroundfurther expanded the myeloid populations including neutrophilsand Ly6C+ inflammatory monocytes (Fig. 7B and SI Appendix, Fig.S14). Therefore, defective control of lymphoid and myeloid pop-ulations in Map3k7DC mice was uncoupled after depletion of lym-

phocytes, which revealed an important role for TAK1 signaling inDCs to maintain myeloid homeostasis in a manner independentlyof lymphocytes.We further explored the mechanism by which lack of DCs

triggered myeloid expansion. Depletion of DCs has recently beenshown to increase the levels of FLT3L, a key growth factor formyeloid cell generation (21, 60). Indeed, the serum concentra-tion of FLT3L was significantly elevated in Map3k7DC micecompared with the WT controls (Fig. 7C). Thus, FLT3L was up-regulated in the absence of a normal DC population that likelycontributed to the abnormal myeloid expansion.

DiscussionHomeostasis of DCs is essential for the generation of adaptiveimmunity and the maintenance of immune tolerance. Here wehave identified TAK1 as a critical regulator of DC survival and

Fig. 6. Map3k7DC mice show diminished conventional T cells but expanded myeloid cells. (A) Distribution of CD4+ and CD8+ T cells in the thymus, spleen, andPLNs of WT and Map3k7DC mice. (Right) The proportion of CD4+ and CD8+ T cells in WT and Map3k7DC mice. (B) Expression of IFN-γ and IL-2 in splenic CD4+

and CD8+ T cells was determined by intracellular cytokine staining after stimulation with PMA and ionomycin. (C) Spleen sections from WT and Map3k7DC

mice were analyzed by H&E, Masson’s trichrome, and reticulin staining for fibrosis and by anti-CD3 and anti-Mac2 immunohistochemistry. (D) Flow cytometryof myeloid cells (Gr1+CD11b+) in the spleen, PLNs, and MLNs of WT and Map3k7DC mice. Data are representative of at least three independent experiments.Data represent the mean ± SEM.

Wang et al. PNAS | February 7, 2012 | vol. 109 | no. 6 | E349

IMMUNOLO

GY

PNASPL

US

homeostasis. Using DC-specific and acute deletion systems, wehave established that TAK1 is required to maintain DC pop-ulations by promoting the survival of DCs and generation of theirprecursors. TAK1 integrates prosurvival receptor signals and inturn activates NF-κB and AKT-Foxo pathways while controllingexpression of Bim and other proapoptotic molecules. TAK1signaling in DCs facilitates T-cell priming and generation of Tregcells, indicating a role of TAK1 to mediate both the immuno-genic and tolerogenic activities of DCs. Moreover, loss of thisactive control mechanism in DCs has profound effects on thehomeostasis of T cells and myeloid populations, and additionalanalysis reveals that TAK1 acts in DCs to control myeloid cell-mediated inflammation in a lymphocyte-independent manner.Therefore, a TAK1-dependent checkpoint in DC survival orches-trates immune function and homeostasis.DC homeostasis is dependent upon the interplay of three main

processes: development from precursors, division in situ, andapoptotic cell death. Combining tissue-specific and inducibledeletion systems, we found that both acute and sustained loss ofTAK1 resulted in an extensive apoptotic death of DCs. Thisresult contrasted with the observations in neuronal cells in whichshort-term inhibition of TAK1 protects neurons from apoptosis,whereas prolonged inhibition is not protective (61). Therefore,our findings indicate a direct effect of TAK1 to actively maintainDC survival. Moreover, TAK1 contributed to DC developmentby maintaining the MDP population. However, TAK1 was dis-pensable for DC division, which differed from a critical re-quirement of TAK1 in clonal expansion of lymphocytes (33, 37–40). Therefore, TAK1 has a unique role in DCs to selectivelyregulate their survival and development.Among DC populations, the CD8+ and CD8– cDC subsets

efficiently present antigens to CD8+ and CD4+ T cells, re-spectively (62). The loss of CD8+ cDCs was more profoundrelative to the CD8– subset in Map3k7DC mice. However, theremaining CD8– cDCs probably exhibited qualitative changes,reflected by their up-regulated CD11b levels on a per cell basis(Fig. 1A). Accordingly, both CD8+ and CD4+ T-cell responseswere diminished in response to antigen challenge, in agreementwith a requirement of TAK1 in the contact hypersensitivity re-sponse (63). Deletion of TAK1 was also incompatible with thesurvival of pDCs, and a more severe reduction of pDCs uponacute deletion of TAK1 than DC-specific TAK1 deficiencyprobably reflected the significant but incomplete deletion effi-ciency in pDCs in the CD11c-Cre line (42). Finally, the CD103+

DCs in the nonlymphoid organs were as sensitive to TAK1 de-ficiency as CD8+ cDCs in the lymphoid organs, which provides

added genetic evidence supporting the notion that these twopopulations are developmentally and functionally linked (8, 43–45). Therefore, TAK1 is critically required for the survival of allmajor DC subsets.The molecular mechanisms in DC survival remain to be fully

established. Among the best understood pathways for the sur-vival of DCs are the Bcl-2 family members, as previous work hasimplicated a pivotal role for Bcl-2, Bcl-xL, and Bim in DC sur-vival (9). How these factors are regulated by the intracellularsignaling network is not well defined. Here we found an elevatedBim expression in TAK1-deficient DCs, which likely resultedfrom the altered AKT-Foxo signaling axis (52). In addition,TAK1 was required for NF-κB activity in DCs. Moreover,TAK1-deficient DCs showed an enhanced autocrine IL-10 sig-naling, which likely further contributed to the increased celldeath, given the important role of IL-10 in promoting DC apo-ptosis (64). Taken together, TAK1 affects signal transductionand gene expression of both proapoptotic and antiapoptoticpathways and further links them with expression of immune re-sponse genes, thereby orchestrating a prosurvival program inDCs. Consistent with this notion, TAK1 integrates multipleprosurvival receptor signals in DCs.As the most potent type of APCs, DCs are essential for the

initiation and propagation of adaptive immune responses, buthow DCs are involved in the maintenance of nTreg cell homeo-stasis remains under debate. Map3k7DC mice contained a greatlyreduced number of nTreg cells in the periphery under steadystate. These results differ from the earlier reports excludinga role for DCs to regulate nTreg cells (20, 21), but are consistentwith more recent studies implicating a requirement for DCs tomaintain the nTreg population (17, 27, 65). Interestingly, despitedefective iTreg generation and nTreg homeostasis, T cells inMap3k7DC mice exhibited no signs of prominent activation. In-stead, we found reduced populations of CD4+ and CD8+ T cellsand diminished production of IFN-γ and IL-2 in T cells fromMap3k7DC mice. Our results are in agreement with a recent DCablation study in lupus-prone MRL/lpr mice that reveals an im-portant role for DCs to maintain conventional T cells and theirability to produce IFN-γ (65). Thus, TAK1 acts in DCs to or-chestrate T-cell priming and homeostasis but not self-tolerance.Unexpectedly, deficiency of TAK1 in DCs results in a prom-

inent myeloid proliferative disease characterized by expansion ofneutrophils and Ly6C+ inflammatory monocytes, suggesting thatDCs provide a negative signal to restrain myeloid expansion andinflammation under physiological conditions. The myeloid defectslargely recapitulate the phenotypes of mice with constitutive loss

Fig. 7. Myeloid expansion in Map3k7DC mice results from altered DC–myeloid cell crosstalk independently of lymphocytes. (A) Flow cytometry of spike BM-derived and WT or Map3k7DC BM-derived myeloid cells in the mixed chimeras (Fig. 1D). (B) Flow cytometry of myeloid cells in the spleen, PLNs, MLNs, andthymus of Rag1−/− and Map3k7DC Rag1−/− mice. (C) Analysis of FLT3L levels in the serum from WT and Map3k7DC mice. Each symbol represents an individualmouse and short horizontal lines indicate the mean.

E350 | www.pnas.org/cgi/doi/10.1073/pnas.1115635109 Wang et al.

of cDCs (20, 21), although the precise mechanism involvedremains in debate. In particular, dysregulated autoreactive Tcells and impaired Treg cells have been suggested as importantmechanisms for excessive inflammation (17, 20). However, de-letion of TAK1 in DCs disrupted myeloid homeostasis even ona Rag1−/− background. Although our results do not exclude thecontribution of Treg or conventional T cells, they provide de-finitive evidence that DC signaling can directly impact myeloidcell homeostasis independently of lymphocytes, thus establishinga previously unappreciated mechanism of DC-mediated activecontrol of myeloid expansion and inflammation. The crosstalkbetween DCs and myeloid cells is likely mediated by FLT3L,given the key role of FLT3L in myeloid cell expansion (21, 60),although this action does not appear to affect myeloid pop-ulations in the BM. Recently, expansion of myeloid populationsand increases of serum FLT3L levels have been reported in micewith DC-selective deletion of A20, a negative regulator of NF-κBsignaling (66, 67). However, unlike DCs lacking TAK1, A20-deficient DCs show enhanced responses to CD40 and RANK-mediated survival signals and trigger profound activation of Tcells and B cells and development of autoimmunity (66, 67).Therefore, immune homeostasis under steady state is dependentupon molecular signals in DCs, especially a delicate threshold ofNF-κB signaling.The life span of DCs is important for the strength of adaptive

immunity (10, 68, 69). Our results have identified a TAK1-de-pendent checkpoint in DC survival that influences the magnitudeand quality of adaptive immune responses, as well as homeo-stasis of the immune system. In Drosophila TAK1 has a crucialfunction in mounting host defense reactions (30, 31). Not onlydoes the mammalian immune system retain this function forinnate defense responses (32, 33), but also DCs have evolved toacquire this evolutionarily conserved pathway to regulate theirlife span and further imprint innate and adaptive immunity. Thismolecular pathway in DCs may be explored for the developmentof DC-based therapeutic strategies.

Materials and MethodsMice and BM Chimeras. C57BL/6, CD45.1, Thy1.1, Rag1−/−, OT-I, and OT-II micewere purchased from the Jackson Laboratory. Foxp3gfp mice were kindlyprovided by A. Rudensky (Memorial Sloan-Kettering Cancer Center, NewYork, NY) (70). Floxed Map3k7 (Map3k7flox/null) mice were bred with CD11c-Cre or Rosa26-Cre-ERT2 mice (37, 42, 46) and have been backcrossed to theC57BL/6 background for at least eight generations. WT controls were in thesame genetic background and included Cre+ mice to account for Cre effects.For in vivo tamoxifen treatment, WT and Map3k7CreER mice were injected i.p. with 1 mg per mouse of tamoxifen (Sigma) for 3 consecutive days beforefurther analysis. For mixed BM experiments, BM cells from WT or Map3k7DC

CD45.2.2+ mice were mixed with cells from CD45.1.2+ mice at a 1:1 ratio andtransferred into lethally irradiated (11 Gy, split dose) CD45.1.1+ mice, asdescribed previously (71). Animal protocols were approved by InstitutionalAnimal Care and Use Committee of St. Jude Children’s Research Hospital.

Cell Purification and Cultures. Mouse spleens were digested with CollagenaseD (Roche), and DCs (CD11c+TCR–CD19–DX5–) were sorted on a Reflection cellsorter (i-Cyt). DCs from nonlymphoid organs were isolated as described inref. 8. Lymphocytes were sorted for naive T cells (CD4+CD62LhiCD44loCD25–).

For DC culture, BM cells were cultured in medium containing 10% FBS andmouse FLT3L (200 ng/mL); CD11c+ cells were sorted at day 7, washed ex-tensively, and treated with 10 μg/mL LPS, 10 μg/mL α-CD40, 1 μg/mL RANKL,or mock for 2 d for the analysis of cell survival. For the mixed culture, WT orMap3k7CreER BM cells were cultured with CD45.1+ BM cells at a 1:1 ratio inthe presence of FLT3L. From day 7, 0.5 μM 4-OHT (Sigma-Aldrich) was addedto each well to delete TAK1. For DC–T-cell coculture, splenic DCs and OT-IIT cells (Foxp3GFP) were mixed at a 1:10 ratio in the presence of 1 μg/mLOVA323–339 peptide and cultured for 5 d, followed by flow cytometry analysis.

Antigen Challenge. Antigen-specific T cells from OT-I (CD45.1+; 1 × 106) or OT-II TCR-transgenic mice (Thy1.1+; 2 × 106) were sorted and transferred into WTand Map3k7DC mice intravenously. Twenty-four hours later, the mice wereinjected s.c. with OVA257–264 or OVA323–339 (5 μg) in the presence of in-complete Freund’s adjuvant (IFA; Difco). At day 4 after immunization,draining lymph node (DLN) cells were isolated for further analyses (71). Inthe model of oral antigen stimulation, after adoptive transfer of naive OT-IIT cells, mice were fed with water supplemented with 20 mg/mL OVA protein(Grade VI; Sigma-Aldrich) for a total of 5 d (71).

Flow Cytometry. Flow cytometry was performed as described previously (46,71, 72). For intracellular cytokine detection, the cells were stimulated for 5 hwith phorbol 12-myristate 13-acetate (PMA) and ionomycin in the presenceof monensin before staining according to the manufacturer’s instructions(BD Biosciences). For caspase activity detection, cells were stained withCaspACE FITC-VAD-FMK in situ marker according to the manufacturer’sinstructions (Promega). For TUNEL staining, freshly isolated splenocytes wereused according to the protocol supplied (BD Biosciences). ROS were mea-sured by incubation for 30 min at 37 °C with 10 μM CM-H2DCFDA (Invi-trogen). For BrdU labeling, mice were injected i.p. with 1 mg BrdU (5-bromo-2-deoxyuridine); 24 h later, mice were killed and BrdU incorporation wasanalyzed according to the manufacturer’s instructions (BD Biosciences) (46).

RNA and Protein Analyses. Real-time PCR analysis was done with primers andprobe sets from Applied Biosystems as described in refs. 46 and 71. Micro-array analysis was performed as described previously (46, 72), and the resultshave been deposited in the GEO database (GSE34417). The heat maps weregenerated to show relative expression of genes after z-transformation,which subtracts the mean of each from each individual value for that geneand then divides by the SD. This method sets all of the genes into the samescale with a mean of 0 and SD of 1. The scale bar shows the distance fromthe mean expression as SD units. Immunoblot was performed as described inrefs. 46 and 71, using the following antibodies: p-STAT3, p-AKT (Ser473), p-Foxo1 (Ser256), and p-IκBα (all from Cell Signaling Technology); Bcl-2 (SantaCruz); Bim (Abcam); and β-actin (Sigma). The serum levels of FLT3L andautoantibodies were measured by ELISA as described in ref. 73.

Histopathology and Immunohistochemistry. The spleen and PLNs were pro-cessed as described previously (72) and stained with hematoxylin and eosin(H&E), Masson’s trichrome, and reticulin stains for the visualization of col-lagen deposition. Immunohistochemistry was performed as described pre-viously (72). Peripheral blood samples were collected for blood counts.

Statistical Analysis. P values were calculated using Student’s t test. P values<0.05 were considered significant.

ACKNOWLEDGMENTS. This work is supported by the National Institutes ofHealth (R01 NS064599 and K01 AR053573), the Cancer Research Institute, theNational Multiple Sclerosis Society (RG4180-A-1), and the Hartwell Foundation.

1. Steinman RM (2007) Lasker Basic Medical Research Award. Dendritic cells: Versatilecontrollers of the immune system. Nat Med 13:1155–1159.

2. Liu K, Nussenzweig MC (2010) Origin and development of dendritic cells. ImmunolRev 234:45–54.

3. Helft J, Ginhoux F, Bogunovic M, MeradM (2010) Origin and functional heterogeneityof non-lymphoid tissue dendritic cells in mice. Immunol Rev 234:55–75.

4. Watowich SS, Liu YJ (2010) Mechanisms regulating dendritic cell specification anddevelopment. Immunol Rev 238:76–92.

5. Liu K, et al. (2009) In vivo analysis of dendritic cell development and homeostasis.Science 324:392–397.

6. Kabashima K, et al. (2005) Intrinsic lymphotoxin-beta receptor requirement forhomeostasis of lymphoid tissue dendritic cells. Immunity 22:439–450.

7. Liu K, et al. (2007) Origin of dendritic cells in peripheral lymphoid organs of mice. NatImmunol 8:578–583.

8. Ginhoux F, et al. (2009) The origin and development of nonlymphoid tissue CD103+DCs. J Exp Med 206:3115–3130.

9. Chen M, Wang J (2010) Programmed cell death of dendritic cells in immuneregulation. Immunol Rev 236:11–27.

10. Hou WS, Van Parijs L (2004) A Bcl-2-dependent molecular timer regulates the lifespanand immunogenicity of dendritic cells. Nat Immunol 5:583–589.

11. Chen M, Huang L, Wang J (2007) Deficiency of Bim in dendritic cells contributes tooveractivation of lymphocytes and autoimmunity. Blood 109:4360–4367.

12. Chen M, Huang L, Shabier Z, Wang J (2007) Regulation of the lifespan in dendritic cellsubsets. Mol Immunol 44:2558–2565.

13. Jung S, et al. (2002) In vivo depletion of CD11c+ dendritic cells abrogates priming ofCD8+ T cells by exogenous cell-associated antigens. Immunity 17:211–220.

14. Steinman RM, Hawiger D, Nussenzweig MC (2003) Tolerogenic dendritic cells. AnnuRev Immunol 21:685–711.

Wang et al. PNAS | February 7, 2012 | vol. 109 | no. 6 | E351

IMMUNOLO

GY

PNASPL

US

15. Chen M, et al. (2006) Dendritic cell apoptosis in the maintenance of immunetolerance. Science 311:1160–1164.

16. Stranges PB, et al. (2007) Elimination of antigen-presenting cells and autoreactiveT cells by Fas contributes to prevention of autoimmunity. Immunity 26:629–641.

17. Darrasse-Jèze G, et al. (2009) Feedback control of regulatory T cell homeostasis bydendritic cells in vivo. J Exp Med 206:1853–1862.

18. Swee LK, Bosco N, Malissen B, Ceredig R, Rolink A (2009) Expansion of peripheralnaturally occurring T regulatory cells by Fms-like tyrosine kinase 3 ligand treatment.Blood 113:6277–6287.

19. O’Keeffe M, et al. (2005) Fms-like tyrosine kinase 3 ligand administration overcomesa genetically determined dendritic cell deficiency in NOD mice and protects againstdiabetes development. Int Immunol 17:307–314.

20. Ohnmacht C, et al. (2009) Constitutive ablation of dendritic cells breaks self-toleranceof CD4 T cells and results in spontaneous fatal autoimmunity. J Exp Med 206:549–559.

21. Birnberg T, et al. (2008) Lack of conventional dendritic cells is compatible with normaldevelopment and T cell homeostasis, but causes myeloid proliferative syndrome.Immunity 29:986–997.

22. Brocker T (1997) Survival of mature CD4 T lymphocytes is dependent on majorhistocompatibility complex class II-expressing dendritic cells. J Exp Med 186:1223–1232.

23. Gruber A, Brocker T (2005) MHC class I-positive dendritic cells (DC) control CD8 T cellhomeostasis in vivo: T cell lymphopenia as a prerequisite for DC-mediatedhomeostatic proliferation of naive CD8 T cells. J Immunol 175:201–206.

24. Zaft T, Sapoznikov A, Krauthgamer R, Littman DR, Jung S (2005) CD11chigh dendriticcell ablation impairs lymphopenia-driven proliferation of naive and memory CD8+T cells. J Immunol 175:6428–6435.

25. Hochweller K, et al. (2010) Dendritic cells control T cell tonic signaling required forresponsiveness to foreign antigen. Proc Natl Acad Sci USA 107:5931–5936.

26. Suffner J, et al. (2010) Dendritic cells support homeostatic expansion of Foxp3+regulatory T cells in Foxp3.LuciDTR mice. J Immunol 184:1810–1820.

27. Bar-On L, Birnberg T, Kim KW, Jung S (2011) Dendritic cell-restricted CD80/86deficiency results in peripheral regulatory T-cell reduction but is not associated withlymphocyte hyperactivation. Eur J Immunol 41:291–298.

28. Bar-On L, Jung S (2010) Defining dendritic cells by conditional and constitutive cellablation. Immunol Rev 234:76–89.

29. Huang G, Shi LZ, Chi H (2009) Regulation of JNK and p38 MAPK in the immunesystem: Signal integration, propagation and termination. Cytokine 48:161–169.

30. Vidal S, et al. (2001) Mutations in the Drosophila dTAK1 gene reveal a conservedfunction for MAPKKKs in the control of rel/NF-kappaB-dependent innate immuneresponses. Genes Dev 15:1900–1912.

31. Silverman N, et al. (2003) Immune activation of NF-kappaB and JNK requiresDrosophila TAK1. J Biol Chem 278:48928–48934.

32. Shim JH, et al. (2005) TAK1, but not TAB1 or TAB2, plays an essential role in multiplesignaling pathways in vivo. Genes Dev 19:2668–2681.

33. Sato S, et al. (2005) Essential function for the kinase TAK1 in innate and adaptiveimmune responses. Nat Immunol 6:1087–1095.

34. Hasegawa M, et al. (2008) A critical role of RICK/RIP2 polyubiquitination in Nod-induced NF-kappaB activation. EMBO J 27:373–383.

35. Kim JY, Omori E, Matsumoto K, Núñez G, Ninomiya-Tsuji J (2008) TAK1 is a centralmediator of NOD2 signaling in epidermal cells. J Biol Chem 283:137–144.

36. Qin J, et al. (2006) TLR8-mediated NF-kappaB and JNK activation are TAK1-independent and MEKK3-dependent. J Biol Chem 281:21013–21021.

37. Wan YY, Chi H, Xie M, Schneider MD, Flavell RA (2006) The kinase TAK1 integratesantigen and cytokine receptor signaling for T cell development, survival and function.Nat Immunol 7:851–858.

38. Sato S, et al. (2006) TAK1 is indispensable for development of T cells and preventionof colitis by the generation of regulatory T cells. Int Immunol 18:1405–1411.

39. Liu HH, Xie M, Schneider MD, Chen ZJ (2006) Essential role of TAK1 in thymocytedevelopment and activation. Proc Natl Acad Sci USA 103:11677–11682.

40. Schuman J, et al. (2009) A critical role of TAK1 in B-cell receptor-mediated nuclearfactor kappaB activation. Blood 113:4566–4574.

41. Tang M, et al. (2008) TAK1 is required for the survival of hematopoietic cells andhepatocytes in mice. J Exp Med 205:1611–1619.

42. Caton ML, Smith-Raska MR, Reizis B (2007) Notch-RBP-J signaling controls thehomeostasis of CD8- dendritic cells in the spleen. J Exp Med 204:1653–1664.

43. Varol C, et al. (2009) Intestinal lamina propria dendritic cell subsets have differentorigin and functions. Immunity 31:502–512.

44. Bedoui S, et al. (2009) Cross-presentation of viral and self antigens by skin-derivedCD103+ dendritic cells. Nat Immunol 10:488–495.

45. Bogunovic M, et al. (2009) Origin of the lamina propria dendritic cell network.Immunity 31:513–525.

46. Yang K, Neale G, Green DR, He W, Chi H (2011) The tumor suppressor Tsc1 enforcesquiescence of naive T cells to promote immune homeostasis and function. NatImmunol 12:888–897.

47. Fogg DK, et al. (2006) A clonogenic bone marrow progenitor specific formacrophages and dendritic cells. Science 311:83–87.

48. Varol C, et al. (2007) Monocytes give rise to mucosal, but not splenic, conventionaldendritic cells. J Exp Med 204:171–180.

49. Waskow C, et al. (2008) The receptor tyrosine kinase Flt3 is required for dendritic celldevelopment in peripheral lymphoid tissues. Nat Immunol 9:676–683.

50. Ouaaz F, Arron J, Zheng Y, Choi Y, Beg AA (2002) Dendritic cell development andsurvival require distinct NF-kappaB subunits. Immunity 16:257–270.

51. Del Prete A, et al. (2004) Defective dendritic cell migration and activation of adaptiveimmunity in PI3Kgamma-deficient mice. EMBO J 23:3505–3515.

52. Fu Z, Tindall DJ (2008) FOXOs, cancer and regulation of apoptosis. Oncogene 27:2312–2319.

53. Lam CR, et al. (2011) TAK1 regulates SCF expression to modulate PKBα activity thatprotects keratinocytes from ROS-induced apoptosis. Cell Death Differ 18:1120–1129.

54. Yamazaki S, et al. (2007) Dendritic cells are specialized accessory cells along with TGF-for the differentiation of Foxp3+ CD4+ regulatory T cells from peripheral Foxp3precursors. Blood 110:4293–4302.

55. Yamazaki S, et al. (2008) CD8+ CD205+ splenic dendritic cells are specialized to induceFoxp3+ regulatory T cells. J Immunol 181:6923–6933.

56. Coombes JL, et al. (2007) A functionally specialized population of mucosal CD103+DCs induces Foxp3+ regulatory T cells via a TGF-beta and retinoic acid-dependentmechanism. J Exp Med 204:1757–1764.

57. Sun CM, et al. (2007) Small intestine lamina propria dendritic cells promote de novogeneration of Foxp3 T reg cells via retinoic acid. J Exp Med 204:1775–1785.

58. Curotto de Lafaille MA, Lafaille JJ (2009) Natural and adaptive foxp3+ regulatoryT cells: More of the same or a division of labor? Immunity 30:626–635.

59. Auffray C, Sieweke MH, Geissmann F (2009) Blood monocytes: Development,heterogeneity, and relationship with dendritic cells. Annu Rev Immunol 27:669–692.

60. Hochweller K, et al. (2009) Homeostasis of dendritic cells in lymphoid organs iscontrolled by regulation of their precursors via a feedback loop. Blood 114:4411–4421.

61. Neubert M, Ridder DA, Bargiotas P, Akira S, Schwaninger M (2011) Acute inhibition ofTAK1 protects against neuronal death in cerebral ischemia. Cell Death Differ 18:1521–1530.

62. den Haan JM, Lehar SM, Bevan MJ (2000) CD8(+) but not CD8(-) dendritic cells cross-prime cytotoxic T cells in vivo. J Exp Med 192:1685–1696.

63. Zhao YG, Wang Y, Hao W, Wan YY (2011) An essential role for TAK1 in the contacthypersensitivity response. Cell Mol Immunol 8:315–324.

64. Chang WL, et al. (2007) Exposure of myeloid dendritic cells to exogenous orendogenous IL-10 during maturation determines their longevity. J Immunol 178:7794–7804.

65. Teichmann LL, et al. (2010) Dendritic cells in lupus are not required for activation of Tand B cells but promote their expansion, resulting in tissue damage. Immunity 33:967–978.

66. Kool M, et al. (2011) The ubiquitin-editing protein A20 prevents dendritic cellactivation, recognition of apoptotic cells, and systemic autoimmunity. Immunity 35:82–96.

67. Hammer GE, et al. (2011) Expression of A20 by dendritic cells preserves immunehomeostasis and prevents colitis and spondyloarthritis. Nat Immunol 12:1184–1193.

68. Nopora A, Brocker T (2002) Bcl-2 controls dendritic cell longevity in vivo. J Immunol169:3006–3014.

69. Josien R, et al. (2000) TRANCE, a tumor necrosis factor family member, enhances thelongevity and adjuvant properties of dendritic cells in vivo. J Exp Med 191:495–502.

70. Fontenot JD, et al. (2005) Regulatory T cell lineage specification by the forkheadtranscription factor foxp3. Immunity 22:329–341.

71. Huang G, Wang Y, Shi LZ, Kanneganti TD, Chi H (2011) Signaling by the phosphataseMKP-1 in dendritic cells imprints distinct effector and regulatory T cell fates. Immunity35:45–58.

72. Shi LZ, et al. (2011) HIF1alpha-dependent glycolytic pathway orchestrates a metaboliccheckpoint for the differentiation of TH17 and Treg cells. J Exp Med 208:1367–1376.

73. Chi H, et al. (2006) Dynamic regulation of pro- and anti-inflammatory cytokines byMAPK phosphatase 1 (MKP-1) in innate immune responses. Proc Natl Acad Sci USA103:2274–2279.

E352 | www.pnas.org/cgi/doi/10.1073/pnas.1115635109 Wang et al.