Immunity, Vol. 23, 551–555, December, 2005, Copyright ª2005 by Elsevier Inc. DOI 10.1016/j.immuni.2005.11.007

Previews

Phosphorylation of CARMA1:The Link(er) to NF-kB Activation

Antigen receptor-induced NF-kB activation depends

on receptor-proximal and -distal signaling events.Two papers in this issue of Immunity demonstrate that

PKC-dependent phosphorylation of CARMA1 is thecritical molecular link that controls the activation of

the IKK signalosome and NF-kB.

The adaptive immune response is initiated by the anti-gen receptor-dependent activation of signaling path-ways that control the rapid proliferation and differentia-tion of lymphocytes to become effector cells. One of theearliest events is the activation of tyrosine kinases,which in turn regulate a restricted set of signaling com-ponents that control inducible gene transcription.Among these are the B cell-specific classical PKC familymember PKCb and its functional T cell homolog, thenovel PKC family member PKCq. Targeted deletion ofthese molecules in mice leads to defects in B and T cellactivation, respectively, that correlate with impairedactivation of the transcription factor nuclear factor-kB(NF-kB) (Sun et al., 2000; Pfeifhofer et al., 2003; Saijoet al., 2002; Su et al., 2002). In quiescent cells, NF-kBis retained in the cytoplasm by binding to its inhibitor,IkB. Upon lymphocyte activation, this inhibition is re-leased by activation of the IkB kinase complex, IKK,which induces phosphorylation and subsequent degra-dation of IkB.

A major issue in the field has been the identification ofthe downstream targets of PKCb and PKCq that makethe link to the activation of the IKK complex. The pro-teins CARMA1, BCL10, and MALT1/paracaspase havebeen recently identified as signaling components thatact downstream of these PKC family members(Figure 1A; Thome, 2004), but the molecular mechanisminvolved has remained a mystery. CARMA1 is a caspaserecruitment domain (CARD)-containing member of themembrane-associated guanylate kinase (MAGUK) fam-ily that binds the adaptor protein BCL10 via its N-termi-nal CARD. The N-terminal region of CARMA1 also con-tains a coiled coil motif that contributes to MALT1binding and that is separated from the MAGUK-typicalC-terminal domains by an extended linker region withno obvious homology to other known domains.

Now, two studies in this issue of Immunity reveal animportant functional role of the CARMA1 linker andshow that phosphorylation of this region by PKCq (inT cells) or PKCb (in B cells) relays antigen receptor-induced PKC activation to the downstream signalingevents controlling IKK activation (Figure 1B; Matsumotoet al., 2005; Sommer et al., 2005).

Matsumoto, Wang, and colleagues (2005) had ob-served that BCL10 associated with a phospho-protein,subsequently identified as CARMA1, upon T cell activa-tion. Previous work from the same laboratory had re-vealed that PKCq could bind to the CARMA1 linker

region in vitro (Wang et al., 2004). Focusing on theoret-ically predicted conserved PKC phosphorylation sites,the authors identified Ser552 as the major PKCq targetin vitro, and additional Ser residues 555 and 645 as sitesrelevant for NF-kB signaling. In vivo peptide mappingexperiments, on the other hand, identified Ser552 andSer565 (a site that does not match the PKC substrateconsensus) as sites that are phosphorylated uponPMA/CD28 stimulation and that are required to func-tionally reconstitute CARMA1-deficient cells.

The study of Sommer et al. (2005) addresses the sameissue in both B and T cell models. The observation thatPKCb and CARMA1 are coordinately recruited to lipidrafts upon BCR triggering and that CARMA1 binds bothPKCb isoforms via its linker region led these authors toask whether PKCb acts as an upstream kinase regulat-ing CARMA1 function in B cells. In vitro phosphoryla-tion studies identified three individual Ser residueswithin the CARMA1 linker region as the targets of PKC-mediated phosphorylation. Mutation of two of thesesites, Ser564 of murine CARMA1 (identical to Ser552of human CARMA1) and Ser657 (corresponding to hu-man Ser645) led to complete and partial reduction ofCARMA1-inducible NF-kB activity, respectively.

What are the molecular and functional consequencesof the PKC-dependent phosphorylation of the CARMA1linker region? The study of Matsumoto, Wang, and col-leagues shows that translocation into lipid rafts and im-mune synapse enrichment of the Ser552 and Ser555point mutants are not impaired, but recruitment of thedownstream signaling components BCL10 and IKKg/NEMO is affected in CARMA1-deficient cells reconsti-tuted with these mutants. Based on these findings, theauthors suggest that phosphorylation of the CARMA1linker induces a conformational change that leads toits dissociation from an inhibitor and/or exposes itsCARD motif for BCL10 recruitment.

The study of Sommer et al. provides elegant experi-mental proof for the latter model by showing thatPKCq-dependent phosphorylation of a CARMA1 linkerconstruct reduced its capacity to bind to the CARD,an effect that was not observed by a correspondinglinker construct mutated at Ser564 and Ser657. More-over, CARMA1 constructs deleted in all or part of thecritical linker region behaved like constitutively activemutants with respect to their membrane localization,lipid raft enrichment, and capacity to recruit BCL10and IKK and to induce NF-kB activation.

Collectively, the results of the two studies suggestthat CARMA1 resides in B and T cells in an inactive con-formation in which the linker region binds to and blocksthe accessibility of the CARD motif. Antigen receptor-triggered PKC-dependent linker phosphorylation is re-quired to release this inhibition, thereby allowing forBCL10 recruitment and signal propagation to the IKKcomplex. In perfect agreement with this model, Shino-hara and colleagues just reported a PKCb-dependentphosphorylation of CARMA1. This phosphorylation cor-related with CARMA1 binding to BCL10 and MALT1 andalso with recruitment of TGFb-activated kinase-1 (TAK1)

Immunity552

Figure 1. Schematic Representation of the PKC- and CARMA1-Dependent Signaling Pathway that Links Antigen Receptor Engagement to

NF-kB Activation

(A) Antigen receptor engagement leads to NF-kB activation via the activation of PKCb/PKCq, phosphorylation of CARMA1, and signal transmis-

sion via BCL10 and MALT1 to the NF-kB-regulating IKK complex.

(B) CARMA1 contains a CARD and a coiled coil motif that are separated from the MAGUK-typical PDZ, SH3, and GUK domains by a linker region.

This linker is proposed to contain a hinge region and a CARD binding domain (Sommer et al., 2005) and is the target of PKCb- or PKCq-dependent

phosphorylation (P) (Matsumoto et al., 2005; Sommer et al., 2005). The table summarizes the individual linker Ser residues identified by Matsu-

moto, Wang, et al. and Sommer et al. in human and murine CARMA1, respectively, and their confirmed (+) or excluded (2) involvement in the

indicated issues addressed by the two studies (ND, not determined).

and subsequent TAK1-dependent IKK activation (Shi-nohara et al., 2005).

Several interesting perspectives for future workarise from these findings. The studies of Xin Lin’sand David Rawling’s laboratories provide strong supportfor PKCb/q-dependent phosphorylation of hSer552/mSer564. However, the role of the functionally defectivemutant hSer555/mSer567 is less clear. It is likely that thisresidue is not a target of phosphorylation, but insteadplays a structural role, such as facilitating the conforma-tional change predicted to occur in the linker upon phos-phorylation (Matsumoto et al., 2005; Sommer et al.,2005).

Another open question that remains is the nature ofthe kinase targeting hSer565/mSer577. This residue isphosphorylated in PMA/CD28-stimulated Jurkat cells(Matsumoto et al., 2005), but not by PKCq in vitro (Som-mer et al., 2005), and does not match the consensus se-quence for a PKC-dependent phosphorylation site.However, the corresponding CARMA1 point mutant isimpaired in its capacity to activate NF-kB (Matsumotoet al., 2005; Sommer et al., 2005). It seems thus likelythat CARMA1 is phosphorylated on this site by (an)otherkinase(s) activated upon antigen receptor triggering.This is interesting in view of a recent study showing thatPI3K-dependent kinase-1 (PDK1) acts upstream ofCARMA1 in the NF-kB pathway (Lee et al., 2005).PDK1 associates with both PKCq and CARMA1 andcontrols their lipid raft recruitment in CD3/CD28-stimu-lated Jurkat T cells (Lee et al., 2005). Therefore, the pos-sibility of a direct PDK1-dependent phosphorylation ofCARMA1 should be considered.

In addition to the effects on the NF-kB pathway dis-cussed here, it will be highly interesting to address therole of CARMA1 phosphorylation in the activation of

the JNK and p38 MAPK pathways, but also in theCD40- and TLR4-dependent signaling events that havebeen reported to be impaired in CARMA1-deficient ormutant cells (Thome, 2004).

Based on the described CARMA1 linker mutants, thegeneration of knockin mice expressing these mutantsand the generation of phosphorylation site-specificantibodies should be useful in future studies exploringthe biological functions of CARMA1. Finally, as our un-derstanding of the individual kinases and their targetsites increases, the specific pharmacological inhibitionof the phosphorylation of individual CARMA1 linker res-idues may become a successful way to control diseasescaused by altered immune function.

Daniel Rueda1 and Margot Thome1

1Department of BiochemistryUniversity of LausanneBIL Biomedical Research CenterChemin des Boveresses 155CH-1066 EpalingesSwitzerland

Selected Reading

Lee, K.Y., D’Acquisto, F., Hayden, M.S., Shim, J.H. and Ghosh, S.

(2005). Science 308, 114–118.

Matsumoto, R., Wang, D., Blonska, M., Li, H., Kobayashi, M., Pappu,

B., Chen, Y., Wang, D., and Lin, X. (2005). Immunity 23, this issue,

575–585.

Pfeifhofer, C., Kofler, K., Gruber, T., Tabrizi, N.G., Lutz, C., Maly, K.,

Leitges, M. and Baier, G. (2003). J. Exp. Med. 197, 1525–1535.

Saijo, K., Mecklenbrauker, I., Santana, A., Leitger, M., Schmedt, C.

and Tarakhovsky, A. (2002). J. Exp. Med. 195, 1647–1652.

Shinohara, H., Yasuda, T., Aiba, Y., Sanjo, H., Hamadate, M., Watarai,

H., Sakurai, H. and Kurosaki, T. (2005). J. Exp. Med. 202, 1423–1431.

Previews553

Sommer, K., Guo, B., Pomerantz, J.L., Bandaranayake, A.D.,Moreno-

Garcı́a, M.E., Ovechkina, Y.L. and Rawlings, D.J. (2005). Immunity 23,

this issue, 561–574.

Su, T.T., Guo, B., Kawakami, Y., Sommer, K., Chae, K., Humphries,

L.A., Kato, R.M., Kang, S., Patrone, L. Wall, R., et al. (2002). Nat. Im-

munol. 3, 780–786.

Immunity, Vol. 23, December, 2005, Copyright ª2005 by Elsevier Inc. DOI 10

Poxviruses Aren’t StuPYD

Pathogens utilize many strategies to dampen the hostinflammatory response. In this issue of Immunity, a re-

port by Johnston and colleagues reveals a poxvirusstrategy that inhibits the inflammasome, arresting se-

cretion of interleukin-1-related cytokines, thus silenc-ing key alarms that mobilize host defenses.

The host-pathogen relationship is complex, and in mostcases a balance exists between host control and path-ogen escape from immune defenses. The poxviruses inparticular have evolved numerous strategies to inhibitor dampen the host immune response to infection (Seetet al., 2003). In their work, Johnston and colleagues de-scribe the M13L protein that is encoded by a rabbit pox-virus (myxoma virus) and that blocks the production ofinterleukin (IL)-1 family cytokines from infected cells(Johnston et al., 2005). M13L contains a pyrin domain(PYD), a member of the ‘‘death-fold’’ family that is com-posed of six tightly packed a helices to generate aGreek key fold, which can interact specifically with a cel-lular PYD-containing protein called ASC-1 (apoptosis-associated speck-like protein containing a caspase re-cruitment domain) (Masumoto et al., 1999). ASC-1 is acomponent of the ‘‘inflammasome,’’ a multiprotein com-plex located in the cytosol that is responsible for acti-vation of proinflammatory caspases (cysteine-basedaspartic acid-specific proteinases, caspase-1, -4, and -5in humans). Active caspase-1 and/or caspase-5 pro-teolytically process proIL-1b (and IL-18) to its activeform in the cytosol (Martinon and Tschopp, 2004). John-ston et al. show that M13L-PYD inhibits the activation ofprocaspase-1 and subsequent secretion of IL-1b andIL-18 from myxoma-infected cells (Johnston et al.,2005). The authors establish the importance of M13L-PYD in the course of myxoma infection by using a mu-tant virus deleted in the PYD domain of the M13L gene.This mutant virus caused a limited infection that failed tospread from the initial site of inoculation, demonstratingthat the M13L protein contributes to the virulence in thishost and underscoring the importance of the IL-1 familyin mounting effective defenses.

Tight regulation of the production of proinflammatorycytokines is required to maintain the homeostasis ofhost tissues. Insufficient control of IL-1 productionthrough deregulation of the inflammasome can lead to

Sun, Z., Arendt, C.W., Ellmeier, W., Schaeffer, E.M., Sunshine, M.J.,

Gandhi, L., Annes, J., Petrzilka, D., Kupfer, A., Schwartzberg, P.L.

and Littman, D.R. (2000). Nature 404, 402–407.

Thome, M. (2004). Nat. Rev. Immunol. 4, 348–359.

Wang, D., Matsumoto, R., You, Y., Che, T., Lin, X.Y., Gaffen, S.L. and

Lin, X. (2004). Mol. Cell. Biol. 24, 164–171.

.1016/j.immuni.2005.11.008

inflamed tissues and organ damage. There are severaltypes of these autoinflammatory diseases in humans(e.g., Muckle-Wells syndrome, familial cold urticaria,and familial Mediterranean fever) where mutations inkey components of the inflammasome are thought tocontribute to disease (Martinon and Tschopp, 2004).NALP3 promotes inflammasome assembly and is fre-quently mutated in its NACHT/NOD domain, leadingto increased ‘‘signal-independent’’ activation of IL-1b

processing (Agostini et al., 2004). Most components ofthe inflammasome (e.g., NALP, ASC-1, caspases, andothers) encode multiple protein-interaction domains(e.g., PYD, CARD, and NOD/NACT) that promote homo-typic interactions between molecules that contain simi-lar domains, forming the scaffold of the inflammasome.For instance, ASC-1 contains both a PYD and a CARDand can function as a molecular bridge through its inter-action with the PYD of various NALP and the CARD ofcaspase-1 (see Figure 1 inset).

Unlike most inflammasome components, M13L-PYDcontains a single PYD domain at its N terminus andcan interact with ASC-1 through a PYD-PYD association(Johnston et al., 2005). This association of M13L-PYDmay disrupt the ability of ASC-1 to form the bridgebetween caspases and various NALP, resulting in theinhibition of caspase-1 activation. A cellular PYD-only-protein (POP1) that interacts with ASC-1 has been de-scribed, but POP1 does not appear to arrest IL-1b pro-duction (Stehlik et al., 2003), suggesting that M13L-PYDutilizes a distinct mechanism of action from POP1. Onepossible model is shown in Figure 1, wherein M13L-PYDdisrupts the formation of a NALP3-containing inflam-masome through its interaction with ASC-1. However,several molecularly distinct inflammasomes exist, eachcomposed of a particular combination of death-fold-containing adaptor proteins (Martinon and Tschopp,2004). The exact nature of the inflammasome that is in-duced by myxoma infection, and consequently inhibitedby M13L-PYD, is currently unknown, but its future iden-tification will certainly add mechanistic insight to howM13L-PYD functions to inhibit inflammation.

How does myxoma virus infection activate the inflam-masome? Although numerous studies have dissectedthe molecular aspects of protein-protein interactionsthat regulate inflammasome assembly, to date the num-ber of known physiological ligands/stimuli that catalyzethe assembly are limited. Bacterial muropeptide compo-nents of peptidoglycans bind directly to the leucine-rich