Vol. 15, No. 18, pp. 2321-2342, September 15, 2001

REVIEW
NF-B signaling pathways in mammalian and insect innate immunity

Department of Molecular and Cellular Biology, Harvard University, Cambridge, Massachusetts 02138, USA

	Innate immunity

Top Innate immunity Innate immune signaling... Mechanisms of NF-B activation IB kinases Proteolytic processing of NF-... Rel phosphorylation Toll-like receptors Receptor proximal signaling... Conclusions References

Innate immunity is the first line of defense against infectious microorganisms. The innate immune system relies on germ line-encoded pattern recognition receptors (PRRs) to recognize pathogen-derived substances (Janeway 1989). Activation of the innate immune system through these receptors leads to the expression of a vast array of antimicrobial effector molecules that attack microorganisms at many different levels. The innate immune system appeared early in evolution, and the basic mechanisms of pathogen recognition and activation of the response are conserved throughout much of the animal kingdom (Hoffmann et al. 1999).

In contrast to innate immunity, the adaptive immune system generates antigen-specific receptors, antibodies, and T-cell receptors by somatic cell DNA rearrangement. These receptors, found only in higher eukaryotes, recognize specific pathogen-encoded proteins. Mammals have a complex immune response, which relies on communication between the innate and adaptive arms of the immune system. The innate immune response generates a costimulatory signal, which is required in combination with antigen-specific recognition to activate T-cells and the adaptive immune system. Antigen-specific recognition in the absence of costimulation can lead to anergy rather than activation (Janeway 1989). Thus, the activation of an antigen-specific response is coupled to infection through the innate immune system.

Insects have a very potent innate immune response that effectively combats a broad spectrum of pathogens. For example, Drosophila can withstand, and clear, bacterial burdens that, relative to their size, would be lethal to mammals (Hoffmann and Reichhart 1997). Induction of innate immunity in both mammals and insects leads to the activation of similar effector mechanisms, such as stimulation of cell-based phagocytic activity and expression of antimicrobial peptides (Hoffmann et al. 1999). For example, Drosophila produce a wide range of potent antimicrobial peptides in response to infection by fungi or bacteria (Hoffmann and Reichhart 1997). Induction of the antimicrobial peptides is regulated at the level of transcription, and they are expressed primarily in the fat body, the insect liver analog.

Recent studies have revealed striking similarities in the signaling pathways used by humans and flies to activate their innate immune responses. In both cases, infection leads to the activation of Toll-like receptors (TLRs), which in turn initiate intracellular signaling cascades that culminate in the activation of NF-B/Rel family transcription factors. In this review, we discuss recent advances in understanding the signaling pathways in mammalian and Drosophila innate immunity, with emphasis on the mechanisms by which NF-B/Rel family proteins are activated.

	Innate immune signaling pathways

Top Innate immunity Innate immune signaling... Mechanisms of NF-B activation IB kinases Proteolytic processing of NF-... Rel phosphorylation Toll-like receptors Receptor proximal signaling... Conclusions References

Drosophila

Drosophila has two independent immune signaling pathways, both of which lead to the activation of NF-B transcription factors. One pathway responds primarily to fungal and gram positive bacterial infection (Fig. 1), while the other responds to lipopolysaccharide (LPS) treatment or infection by gram-negative bacteria (Fig. 2) (Lemaitre et al. 1996, 1997). The antifungal pathway requires components of the Toll signaling pathway, most of which are also required during dorsoventral patterning of the embryo (Belvin and Anderson 1996). Induction of the Toll/antifungal pathway leads to the activation of two Drosophila NF-B homologs, called Dorsal and Dif, resulting in the production of antifungal peptides such as Drosomycin. Dorsal is also required in early embryogenesis for the Toll-dependent patterning of the dorsoventral axis. Dif is required for antifungal immunity in the adult fly, whereas in larvae either Dif or Dorsal is sufficient for the immune response (Manfruelli et al. 1999; Meng et al. 1999; Rutschmann et al. 2000a).

View larger version (79K): [in this window] [in a new window]   Figure 1.   The Drosophila Toll/antifungal signaling pathway. This model highlights current understanding of the Toll signaling pathway as it functions during the immune response. The pattern recognition receptors that recognize fungal pathogens are unknown, but they are believed to activate a serine protease cascade, culminating in the cleavage of the Toll ligand Spätzle. Ligand binding to Toll leads to the recruitment of two proteins, the adaptor Tube and the kinase Pelle. Recruitment of Pelle is thought to cause its activation and disassociation from Toll. Activated Pelle may then activate, directly or indirectly, a Cactus kinase that is responsible for signaling the proteasome-mediated degradation of Cactus. Currently, the biochemical steps between Pelle and Cactus degradation remain undetermined, and the Cactus kinase has not yet been identified.

View larger version (81K): [in this window] [in a new window]   Figure 2.   The Drosophila antibacterial signaling pathway. In this model the signaling pathway is activated by LPS through unidentified receptor(s) and leads to Relish cleavage. Downstream of the receptors, this signaling pathway bifurcates. One part leads to activation of the Drosophila IKK complex, which then phosphorylates Relish. The other part functions through the caspase Dredd and leads to the cleavage of phosphorylated Relish. At present it is not known whether Dredd acts directly or indirectly to cleave Relish. The IMD protein may function in one or both of these pathways. (N) Amino-terminal domain; (C) carboxy-terminal domain.

By contrast, the LPS-inducible antibacterial pathway requires the third Drosophila NF-B homolog, Relish, which activates the expression of genes encoding antibacterial peptides such as Diptericin (Hedengren et al. 1999; Rutschmann et al. 2000b). Relish is synthesized as an NF-B precursor protein that is cleaved and activated in response to bacterial infection or LPS treatment (Stöven et al. 2000). Interestingly, some antimicrobialpeptides, such as Cecropin, can be activated by either pathway and are thought to have both antibacterial and antifungal activities (Lemaitre et al. 1996; Ekengren and Hultmark 1999). In any case, one mechanism used to direct the transcription apparatus to distinct sets of antimicrobial genes in response to different pathogens is the activation of different members of the NF-B family.

In Drosophila, infection also leads to the activation of the JNK (Sluss et al. 1996) and JAK/STAT signaling pathways (Lagueux et al. 2000). The function of the JNK pathway in immunity has not been established whereas the JAK/SAT pathway is necessary for the induction of a number of complement-like proteins that have recently been shown to function by opsonizing gram-negative bacteria and promoting their phagocytosis (Lagueux et al. 2000; Levashina et al. 2001). Very little is known about the response of the insect immune system to other classes of pathogens, for example, virus, spirochetes or plasmodia.

Mammals

The mammalian innate immune system responds to a plethora of microbial-derived substances including microbial cell wall components such as LPS, peptidylglycans, and lipoproteins (Krutzik et al. 2001). The innate immune system can also be activated by bacterial DNA (Krieg 1996) or double-stranded RNA, the latter of which is common to many viruses (Mogensen and Paludan 2001). The TLRs play a central role in the recognition of many of these immunostimulatory molecules and are probably responsible for the recognition of most types of pathogens. In response to these signals, the TLRs activate signaling pathways that culminate in the expression of antimicrobial molecules (proteins, peptides, and reactive oxygen and nitrogen intermediates), cytokines, and costimulatory molecules (Fig. 3) (Medzhitov et al. 1997; Thoma-Uszynski et al. 2001; Zhang and Ghosh 2001). Thus, activation of the innate immune system immediately slows infection and activates other aspects of the immune system, primarily T-cells. The TLRs activate a number of signaling pathways including the JNK/AP-1 pathway, proapoptotic caspase cascades, and NF-B inducing pathways (Medzhitov et al. 1997; Muzio et al. 1998; Aliprantis et al. 1999, 2000). These pathways are responsible for activating the appropriate effector mechanisms and signaling molecules.

View larger version (68K): [in this window] [in a new window]   Figure 3.   The LPS signaling pathway in mammals. In this model LPS is recognized by a complex of three proteins; CD14, MD-2, and TLR4. TLR4 activates the intracellular signaling cascade by recruiting MyD88 and IRAK to the membrane. IRAK associates with the receptor complex transiently; once released IRAK can associate with and activate TRAF6. The TRAF6 RING finger, in combination with Ubc13 and Uev1A, mediates the K63-extended polyubiquitination of TRAF6 itself. The TAK1/TAB1/TAB2 complex is activated by its association with ubiquitinated TRAF6. Interestingly, the TAK1-associated protein TAB2 translocates from the membrane fraction to the cytoplasmic fraction upon treatment with IL-1. Once activated, the TAK1 complex phosphorylates and activates the IKK complex. The activated IKK complex then phosphorylates IB, leading to its ubiquitination and degradation by the proteasome.

The innate immune system plays a critical role in regulating the decision between the two types of mammalian immune responses, referred to as Type 1 and Type 2 responses. The Type 1 response functions to combat small intracellular pathogens such as bacteria whereas the Type 2 response combats larger extracellular pathogens such as helminths. The particular array of signaling molecules expressed by antigen presenting cells (APCs) determines whether a Type 1 or Type 2 response is activated. The expression of these signaling molecules, cytokines and costimulatory molecules, is controlled by the types of pathogens sensed by APCs through PRR, such as the TLRs. Thus, in addition to its role in activating the adaptive immune system, the innate immune response plays a central role in coordinating the particular type of adaptive immune response so that it will be most effective in combating the pathogens presented (Pulendran et al. 2001).

These two types of responses in mammals bear some similarity to the two Drosophila immunity pathways. In Drosophila, the two immune responses are activated by two different types of pathogens and are specifically suited to combat the pathogen presented. The insect antibacterial pathway is activated by pathogens that would also activate a Type 1 response in mammals. On the other hand, the insect Toll/antifungal pathway responds best to infection by fungal pathogens, which form large hyphal structures. Similar stimuli, such as helminths and certain fungi, activate a Type 2 response in mammals (Hoffmann et al. 1999; Pulendran et al. 2001). Perhaps the similarity between the two types of immune responses, found in both insects and mammals, is due to conservation of the mechanisms used to distinguish between broad classes of pathogens. Although some of the effector mechanisms are quite different, the recognition and subsequent signaling events required for activating these pathways may be very similar throughout the animal kingdom. The identification and characterization of the receptors and signaling pathways necessary for activation of innate immunity in response to a variety of pathogens, in both flies and humans, will clarify these issues.

	Mechanisms of NF-B activation

Top Innate immunity Innate immune signaling... Mechanisms of NF-B activation IB kinases Proteolytic processing of NF-... Rel phosphorylation Toll-like receptors Receptor proximal signaling... Conclusions References

Much has been learned recently about innate immunity in both mammals and flies, and knowledge from these two systems has been highly complementary (see Table 1). The most significant common feature of innate immunity throughout the animal kingdom is the central role of the NF-B/Rel family of transcriptional activator proteins (Karin and Ben-Neriah 2000). In unstimulated cells NF-B/Rel family proteins exist as hetero- or homodimeric proteins that are sequestered in the cytoplasm by virtue of their association with a member of the IB family of inhibitor proteins. An astonishing number of extracellular signals can trigger distinct signal transduction pathways, each of which culminates in the destruction of IB proteins. These signal transduction pathways lead to the activation of the IB kinase, or IKK (for review, see Karin and Ben-Neriah 2000), and the subsequent phosphorylation of serine residues within the N-terminal destruction box of IB proteins (i.e., S32 and S36 of IB) (Ghosh et al. 1998). Phospho-IB is then recognized by the -TrCP-containing SCF ubiquitin ligase complex, leading to its ubiquitination and degradation by the proteasome (Karin and Ben-Neriah 2000). Interestingly, the active -TrCP-SCF complex that functions in NF-B signaling is modified by the ubiquitin-like protein Nedd8 (Read et al. 2000). The degradation of IB unmasks the nuclear localization signal of the NF-B/Rel family protein, leading to its nuclear translocation and binding to enhancers or promoters of target genes.

The fly IB protein is known as Cactus, and by analogy with IB regulation, Cactus degradation is thought to be controlled by signal-dependent serine phosphorylation. However, the exact sequences required for Cactus degradation have not been definitively established (Bergmann et al. 1996; Reach et al. 1996). Moreover, the kinase responsible for signal-dependent Cactus phosphorylation has not been identified. Two IKK-related kinases are encoded in the Drosophila genome, although neither has been shown to be required for the Toll-induced Cactus phosphorylation and degradation (see below for more details) (Fig. 4). Like IB, the ubiquitin/proteasome pathway is required for signal-dependent Cactus degradation. Mutants in slimb, the Drosophila -TrCP homolog, exhibit defects in dorsoventral patterning (Spencer et al. 1999), and Cactus degradation can be blocked by proteasome inhibitors (N. Pandey, N. Silverman and T. Maniatis, unpubl.). Thus, it appears that the mechanisms involved in the activation of the Drosophila Dorsal and Dif proteins during early development and antifungal immunity are highly similar to those required for the activation of NF-B in mammals.

View larger version (13K): [in this window] [in a new window]   Figure 4.   Phylogenetic comparison of IKK-related kinases. Phylogram of a branch-and-bound search using bootstrap analysis of aligned IKK sequences. Gaps were removed with Lineup; data were processed using PAUPDisplay and visualized as a midpoint rooted phylogram. Numbers indicate occurrence of nodes (out of 100 reiterations) during bootstrap analysis. Two subgroups of IKK are clearly distinguishable. One contains the IKK and IKK genes (and a related protein from the oyster, Crassostrea gigas); the other clade contains the IKK-related genes. The Drosophila IKKs include DmIKK, clearly a member of the IKK subgroup, and DmIKK which is not truly in either subgroup, although within the kinase domain it is most similar to IKK. (m) Mouse; (r) rat; (h) human; (o) oyster; (Dm) Drosophila melanogaster.

NF-B/Rel family proteins can also be sequestered in the cytoplasm as large precursors (the mammalian p105 and p100 proteins and the Drosophila Relish protein). The N termini of these proteins contain the DNA binding and dimerization domains (the Rel homology domain), whereas the C termini contain a series of ankyrin repeat sequences similar to those present in IB proteins. Thus, these precursor proteins are sequestered in the cytoplasm by virtue of their covalent attachment to an IB-like inhibitory domain. The C-terminal domain can be removed by proteolysis, either constitutively or, at least in the case of p100 and Relish, in response to signals (see discussion below).

	IB kinases

Top Innate immunity Innate immune signaling... Mechanisms of NF-B activation IB kinases Proteolytic processing of NF-... Rel phosphorylation Toll-like receptors Receptor proximal signaling... Conclusions References

Key components of the NF-B signaling pathways are the IB kinases (IKKs) (Karin and Ben-Neriah 2000). An IB kinase capable of specifically phosphorylating serines 32 and 36 of IB was originally identified as a high-molecular-weight complex (~700kD) (Chen et al. 1996; Lee et al. 1998). Subsequently, two catalytic subunits (IKK/1and IKK/2) and a structural subunit of this complex (IKK/NEMO/IKKAP) were identified and cloned (Karin and Ben-Neriah 2000). More recently, two related kinases known as IKK/IKKi and TBK/NAK/T2K were characterized and found to be in complexes distinct from that of IKK// (Fig. 4) (Peters and Maniatis 2001).

The IKK// complex can be activated by a variety of stimuli, including inducers of the innate immune response, such as infection by virus or treatment with LPS (O'Connell et al. 1998; Chu et al. 1999; Fischer et al. 1999; Hawiger et al. 1999). Activation of the complex involves the phosphorylation of two serine residues located in the "activation loop" within the kinase domain of IKK or IKK. Certain MAP3 kinases (MEKK1, MEKK2, MEKK3, and NIK) are capable of phosphorylating these serines in vitro, and activating NF-B in transfection experiments (Karin and Ben-Neriah 2000). In addition, dominant negative mutants of either MEKK1 or NIK kinase can inhibit NF-B activation in response to certain inducers. However, it has been difficult to establish definitively a role in NF-B activation for either of these kinases under physiological conditions. For example, IKK activation and IB degradation occur normally in embryonic fibroblasts from NIK knockout mice (Yin et al. 2001). Curiously, although the NIK^/ cells display normal NF-B DNA-binding activity in response to numerous stimuli, they exhibit weak activation of NF-B-dependent genes specifically in response to lymphotoxin- receptor (LTR) signaling. This can now be explained by the observation that NIK is an IKK kinase required specifically for signal-dependent p100 processing (see below for more details) (Matsushima et al. 2001; Xiao et al. 2001). Thus, the phosphorylation of IKK by NIK observed in vitro does not appear to be physiologically significant.

Similarly, although MEKK1 can phosphorylate both IKK and IKK in vitro, mekk1^/ mouse embryonic fibroblasts (MEFs) display normal NF-B activation in response to TNF (Yujiri et al. 2000). By contrast, MEKK3, which can also phosphorylate IKK in vitro, is required for TNF-induced NF-B activation. mekk3^/ MEFs exhibit a greatly decreased level of IKK activation, IB degradation, and NF-B activation in response to TNF (Yang et al. 2001).

Alternatively, in some cases the IKK complex may be activated simply by virtue of its recruitment to the receptor complex at the cell membrane (Inohara et al. 2000). This idea is based in part on the observation that IKK, which is essential for IKK activation, specifically interacts with the TNF receptor-associated protein RIP (Devin et al. 2000; Poyet et al. 2000; Zhang et al. 2000), and that the IKK complex is recruited to the activated receptor via the IKK-RIP interaction. In addition, IKK and IKK are both capable of autophosphorylation and cross-phosphorylation of their activation domains (Delhase et al. 1999; O'Mahony et al. 2000). Finally, forced multimerization of the IKK// complex can lead to its activation (Poyet et al. 2000). Thus, the recruitment of the kinase complex to the intracellular domains of various receptors, leading to its increased local concentration, may, under some circumstances, be sufficient to activate the kinase.

Whatever the mechanisms of activation, it is clear that the IKK// complex is required for NF-B activation in response to most NF-B inducers. In fact, ikk^/ mice display fetal liver apoptosis, much like the NF-B p65 subunit knockout mice (Q. Li et al. 1999; Z. Li et al. 1999; Tanaka et al. 1999). Moreover, similar to the p65 knockout mice (Rosenfeld et al. 2000; Alcamo et al. 2001), when the ikk^/ mice are crossed into a tnfr1^/ mouse the liver apoptosis is suppressed. Thus, liver apoptosis in both cases is TNF-dependent. Finally, ikk^/ MEFs do not degrade IB or activate NF-B in response to various stimuli including TNF and LPS. Interestingly IKK also plays a role in NF-B activation, as ikk^/ cells still have residual IKK activity and NF-B transcriptional response. However, IKK/ double knockout cells have no NF-B response (Li et al. 2000). Based on the observation that IKK can activate IKK, it was proposed that IKK actually functions as an IKK-kinase, phosphorylating IKK (O'Mahony et al. 2000). However, this proposal is based primarily on overexpression experiments and is not supported by any of the phenotypes observed in IKK^/ mice.

Remarkably, IKK, but not IKK, appears to play a role in keratinocyte differentiation and proliferation (Hu et al. 1999; Takeda et al. 1999). ikk^/ mice have a skin abnormality caused by the continued proliferation of stem cells and the lack of keratinocyte differentiation. Although the target of IKK in keratinocyte differentiation is unknown, it does not appear to be in the NF-B pathway, as none of the NF-B knockout mice display a similar skin phenotype. In addition, neither the IKK kinase activity nor the NF-B pathway are required for the normal keratinocyte differentiation (Hu et al. 2001). Rather, it appears that IKK controls production of a soluble factor that induces keratinocyte differentiation.

Another difference between IKK and IKK, which is discussed in more detail below, is that IKK, but not IKK, is required for the phosphorylation-dependent proteolytic processing of the p100 precursor of p52, which plays a critical role in B cell maturation and formation of secondary lymphoid organs (Senftleben et al. 2001).

The phenotype of IKK-deficient mice is consistent with the phenotype of IKK-deficient cells (Yamaoka et al. 1998). These mice display fetal liver apoptosis similar to that observed in IKK and p65 knockout mice. Furthermore, ikk^/ MEFs are sensitive to TNF and cannot activate NF-B in response to TNF or LPS (Rudolph et al. 2000). Interestingly, the human disease incontinentia pigmenti (IP) appears to be caused by IKK mutations (Smahi et al. 2000). This is a dominant X-linked disease with perinatal lethality in males. In heterozygous females, cells expressing only the mutant IKK gene (because of X chromosome inactivation) die shortly after birth. Cells from patients, or mice, with homozygous IKK mutations show no detectable IKK protein, they do not activate NF-B in response to TNF, and they are more sensitive to TNF-induced cell death (Makris et al. 2000; Schmidt-Supprian et al. 2000; Smahi et al. 2000).

The role of the other two IKK-related kinases, IKK/i and TBK1/T2K/NAK, is less clear. Transfection experiments with wild type and dominant negative mutants of recombinant TBK1 suggested that TBK1 functions in the NF-B pathway through TANK and TRAF2. However, the signaling pathways that rely on the TBK1-TANK-TRAF complex remain to be identified. The same kinase, termed NAK (NF-B Activating Kinase), was proposed to function upstream of the IKK complex and to activate IKK by direct phosphorylation of the activation loop serine residues, during PMA and growth factor-mediated signaling (for review, see Peters and Maniatis 2001).

On the other hand, deletion of this gene in mice, referred to as T2K, suggests a role in TNF signaling. The phenotype of t2k^/ mice is very similar to that of the IKK and p65 knockouts, as they display severe TNF-induced embryonic liver degeneration and apoptosis. Thus, T2K appears to be a key component in the TNF signaling pathway in liver cells. t2k^/ MEFs exhibit decreased activation of some, but not all, NF-B responsive genes in response to TNF or IL-1. However, t2k^/ MEFs display normal induction of IKK activity, IB degradation, and NF-B DNA binding activity (Peters and Maniatis 2001). These observations suggest that T2K functions after IB degradation. For example, T2K could function to directly or indirectly activate the transcriptional activity of the p65 subunit of NF-B. Other possible explanations are that T2K is required for a separate TNF-inducible pathway, such as the JNK pathway, or that T2K is required for the activation of a certain subset of NF-B dimers. Those genes whose transcription requires this particular dimer would thus be most affected by the T2K deletion.

In vitro, both TBK1/T2K/NAK and IKK can phosphorylate the N-terminal regulatory domain of IB. This phosphorylation occurs only on serine 36, even though residues 32 and 36 must be modified to induce IB degradation. However, in response to PMA stimulation, IKK associates with an unidentified kinase activity that phosphorylates serines 32 and 36 of IB. Although the identity of the IKK-associated IKK is unknown, it was demonstrated that it is neither IKK nor IKK. In Jurkat cells expression of dominant negative IKK blocks PMA- and TCR-mediated, but not TNF-induced, NF-B activation. These data argue that in mammals an alternate IKK complex exists that is responsible for IB phosphorylation and degradation in response to a certain subset of stimuli, for example, T-cell activation (Peters and Maniatis 2001).

As mentioned above, Drosophila has two IKK related genes. One, known as DmIKK, is a member of the IKK/TBK subfamily of IKKs and its function remains unknown. The other Drosophila IKK relative is known as DmIKK (or DLAK) and is required for the antibacterial immune signaling pathway (Kim et al. 2000a; Silverman et al. 2000). Although DmIKK is most similar to hIKK (and thus its name), it is in a subfamily of its own, as it has little homology with either IKK or IKK in its C-terminal half (Fig. 4). DmIKK was shown to be part of a high-molecular-weight Drosophila IKK complex that also contains DmIKK, a homolog to hIKK/NEMO/IKKAP (Silverman et al. 2000). The DmIKK complex is activated by LPS treatment, directly phosphorylates Relish, and is essential for Relish activation (cleavage) and the induction of antibacterial peptide gene transcription. These observations led to the proposal that once activated by LPS, the DmIKK complex phosphorylates Relish, which is then cleaved by an unidentified protease (Fig. 2) (Silverman et al. 2000).

DmIKK has also been shown to phosphorylate Cactus (Kim et al. 2000a) and the sites of phosphorylation have been mapped to the N-terminal regulatory domain of Cactus (Silverman et al. 2000). However, the significance of this modification is not clear. Mutations in the DmIKK or DmIKK genes, ird5 or kenny, respectively, cause defects in antibacterial immunity, but have relatively little effect on antifungal immunity and dorsoventral patterning (Rutschmann et al. 2000b; Lu et al. 2001). Curiously, Lu et al. report that the ird5 mutants have a slight dorsoventral phenotype showing dorsalization in 0.5% of embryos laid by ird5 mothers. Thus, it is possible that DmIKK plays a role in the Toll pathway in a redundant manner with other unidentified kinases.

	Proteolytic processing of NF-B/Rel precursor proteins

Top Innate immunity Innate immune signaling... Mechanisms of NF-B activation IB kinases Proteolytic processing of NF-... Rel phosphorylation Toll-like receptors Receptor proximal signaling... Conclusions References

Mammalian p100 and p105 proteins

The p50 and p52 subunits of NF-B are generated by proteolytic processing of p105 and p100 precursors, respectively. In both cases the rel homology domain is located at the N terminus of the precursor, whereas the C-terminal IB-like domain functions as a covalently attached IB inhibitor protein. Both precursors are processed in vivo and in vitro by a mechanism requiring ubiquitination and partial degradation by the 26S proteasome (Chen and Maniatis 1998; Karin and Ben-Neriah 2000). The regulation of p105 and p100 processing appears to play important biological roles, as mice lacking the precursor protein but able to express the mature protein display specific defects in the immune and inflammatory pathways (Ishikawa et al. 1997, 1998).

Studies of the sequence requirements for p105 processing revealed that p105 molecules containing ~110 amino acids beyond the C terminus of p50 are accurately processed, and a glycine-rich region (GRR), located just upstream from the p50 C terminus, is essential for cleavage (Lin and Ghosh 1996). Although the GRR was initially thought to be both necessary and sufficient for p50 generation (Lin and Ghosh 1996), subsequent studies revealed that additional sequences containing ubiquitination sites downstream from the GRR are also necessary for p105 processing (Fig. 5) (Orian et al. 1999).

View larger version (20K): [in this window] [in a new window]   Figure 5.   NF-B precursor proteins from mammals and Drosophila. The mammalian precursor proteins, p100 and p105, are processed in a ubiquitin-proteasome-dependent manner. Approximate cleavage sites are marked with orange arrows. p105 processing (and/or degradation) is controlled by two regions, the constitutive central ubiquitin region and the inducible C-terminal IKK/-TrCP-dependent phosphorylation and ubiquitination element. p100 processing is stimulated by NIK in an IKK-dependent mechanism. Response to NIK requires the NRS, a probable IKK phosphorylation site. Constitutive processing of p100 is inhibited by the processing inhibitory domain (PID), which colocalizes with a predicted death domain (DD). Relish cleavage is stimulated by LPS treatment and requires caspase proteases, not the proteasome. A potential caspase cleavage site (CCS) is found in the linker domain of Relish.

A unique feature of p105 processing is the limited degradation by the proteasome. It is possible that this is accomplished by a mechanism in which the C terminus of the precursor protein enters the proteasome and is degraded processively. In this model the GRR functions as a stop signal and protects p50 from degradation (Orian et al. 1999). This model is further supported by the recent observation that ATP-dependent proteases, including the proteasome, degrade substrates processively from a degradation signal (Lee et al. 2001). However, more studies are required to prove this model. The detailed mechanism by which NF-B precursors are processed by the proteasome is controversial. A number of studies have clearly demonstrated a precursor/product relationship for both proteins (Chen and Maniatis 1998; Karin and Ben-Neriah 2000). However, it has also been suggested that p105 and p100 are not the precursors of p50 and p52, but function primarily as IBs in the cytoplasm. Instead, p50 and p52 were proposed to be generated by a cotranslational processing mechanism (Lin et al. 1998; Heusch et al. 1999; Lin et al. 2000).

Regardless of the processing mechanism, the most important question is whether it is regulated. Regulated processing is difficult to demonstrate for p50 because expression of the p105 gene is itself regulated by NF-B. Thus, increases in p50 could be due to either increased processing or increased p105 expression. An important insight into the mechanism of p105 processing/degradation was provided by the observation that the C terminus of p105 is phosphorylated in response to signals that activate NF-B (Heissmeyer et al. 1999; Orian et al. 2000). Several different kinases have been shown to interact with and phosphorylate p105, including TPL-2 (Belich et al. 1999) and the IKKs (Heissmeyer et al. 1999; Orian et al. 2000). However, recent experiments with TPL-2 knockout mice have shown that TPL-2 is not required for LPS-dependent activation of NF-B (Dumitru et al. 2000). Thus, the connection between TPL-2 and p105 remains in question.

IKK and IKK interact with and directly phosphorylate the C terminus of p105 (Heissmeyer et al. 1999, 2001). The phosphorylated serine residues are found in a conserved DSXXXDS destruction box motif, which is recognized by the same SCF-TrCP ubiquitin ligase complex that recognizes the destruction box in IB. Similar conclusions were reached in an independent study in which the phosphorylation and ubiquitination of p105 were stimulated by cotransfection p105 and constitutively active IKK (Orian et al. 2000). However, the consequence of this phosphorylation is controversial. Heissmeyer et al. (1999) concluded that the signal-dependent phosphorylation of p105 results in its degradation rather than processing. By contrast, Orian et al. (2000) argue that the phosphorylation of p105 by IKK can lead to processing (Ciechanover et al. 2001). Thus, in spite of considerable effort, there is as yet no definitive demonstration of a signal-dependent induction of p105 processing.

A clearer picture has emerged for the regulated processing of p100 (Xiao et al. 2001). This study also sheds light on the role of the MAP3 kinase NIK in the NF-B signaling pathway. Relatively little p52 is produced in most cell types, even though p100 is present, so it appears that p100 processing is tightly regulated. The p100 (nfkb2) knockout results in defects in B cell function and abnormalities in peripheral lymphoid organs. Remarkably, a similar phenotype is also observed in alymphoplasia (aly) mice, which carry a mutation in the gene encoding NIK (Shinkura et al. 1999). Because of this similarity, Sun and coworkers systemically analyzed the role of NIK in p100 activation (Xiao et al. 2001). They found that processing of transfected p100 increases dramatically when active NIK is cotransfected and that this increase is enhanced by cotransfection of the LTR. Moreover, this processing was not observed in splenocytes from the aly mouse. In contrast to earlier studies of constitutively processed p100 (Heusch et al. 1999), a clear p100 precursor/p52 product relationship was observed in pulse chase experiments. Thus, at least in the regulated processing of p100, p52 is generated by a NIK-dependent post-translational processing mechanism.

Although NIK was proposed to be the direct upstream kinase for p100 (Xiao et al. 2001), more recent studies have shown that NIK functions in p100 processing by activating IKK. Specifically, NIK-induced p100 processing was shown to require IKK, and recombinant IKK was found to be a more efficient p100 kinase than NIK (Senftleben et al. 2001). Xiao et al. (2001) also demonstrated that phosphorylation of the C terminus of p100 leads to its polyubiquitination. Thus, it seems likely that LTR signaling activates NIK, which in turn activates IKK, which then phosphorylates p100, leading to its ubiquitination and processing by the proteasome. This conclusion is consistent with the observation that nik^/ and ikk^/ cells do not respond to LTR activation (Matsushima et al. 2001; Yin et al. 2001). Thus, p100 provides a clear example of signal-dependent processing of NF-B precursors.

Drosophila Relish protein

Like p105 and p100, Relish is a bipartite protein with an N-terminal NF-B-like Rel homology domain and a C-terminal IB-like ankyrin repeat domain (Dushay et al. 1996). In unstimulated cells the Relish C-terminal IB module sequesters its own N-terminal NF-B module in the cytoplasm. Upon activation of the antibacterial signaling pathway, Relish is proteolytically cleaved and the N-terminal NF-B module translocates into the nucleus, while the stable C-terminus remains in the cytoplasm. Thus, the regulation of Relish is unique among the NF-B precursor proteins. Whereas processing of p100 and p105 is a proteasome-dependent event that does not leave a stable C-terminal domain, Relish cleavage is not mediated by the proteasome and results in a stable C terminus. In fact, Relish processing may be controlled by a caspase protease and is stimulated by the Drosophila IKK complex (see below). It is interesting to speculate that p105 and p100 may be processed by a similar mechanism in response to inducers that have yet to be identified.

	Rel phosphorylation

Top Innate immunity Innate immune signaling... Mechanisms of NF-B activation IB kinases Proteolytic processing of NF-... Rel phosphorylation Toll-like receptors Receptor proximal signaling... Conclusions References

Another level of regulation of NF-B/Rel activity in both mammals and Drosophila is the phosphorylation of Rel proteins. A number of studies have shown that the signal-dependent activation of NF-B requires both IB degradation and Rel protein phosphorylation, and these two steps can be uncoupled (Schmitz et al. 2001). For example, in the absence of Cactus, the nuclear translocation of Dorsal remains signal dependent (Bergmann et al. 1996; Drier et al. 2000), and this correlates with the signal-dependent phosphorylation of Dorsal (Drier et al. 1999). Although multiple serine residues are phosphorylated, when serine 317 is substituted by alanine a significant embryonic phenotype is observed, and Dorsal does not translocate to the nucleus.

In mammals the NF-B p65 protein has a protein kinase A phosphorylation site on serine 276, and phosphorylation of this residue is required for efficient binding to the transcriptional activator protein CBP. According to a model proposed by Ghosh and colleagues, PKA is associated with the IB/NF-B complex in the cytoplasm in an inactive form, and signal-dependent degradation of IB allows PKA to phosphorylate serine 276 (Zhong et al. 1997). Phosphorylation of S276 induces a conformational change in p65, exposing a phosphorylation-independent CBP interaction site and creating a phosphorylation-dependent interaction with the Kix domain of CBP (Zhong et al. 1998).

The TNF-dependent phosphorylation of serine 529 has also been shown to increase the transcriptional activity of p65, and this stimulation is not at the level of nuclear translocation or DNA binding. Recently, casein kinase II was implicated in the TNF-dependent phosphorylation of serine 529 (Wang and Baldwin 1998; Wang et al. 2000). Similar to the situation with PKA, IB protects p65 from phosphorylation by constitutively active CKII, but signal-dependent degradation of IB exposes the p65 phosphorylation site to CKII activity. CKII was shown to associate with p65 in vivo, and this association decreased upon TNF induction. Moreover, CKII phosphorylates p65 at serine 529 in vitro, and inhibitors of CKII block phosphorylation and transcriptional activation in vivo (Wang et al. 2000). Thus, once released from IB, at least two kinases, PKA and CKII, phosphorylate p65, at different serine residues, to increase its transcriptional activity.

Other kinases have also been shown to act on p65. For example, IL-1 treatment has been shown to induce the phosphorylation of p65, and this phosphorylation requires phosphatidylinositol-3 kinase and Akt (Sizemore et al. 1999). Also, Akt has been implicated in Ras-induced NF-B activation (Madrid et al. 2000). In these studies, Akt signaling required IKK and serines 529 and 536 of p65 (Madrid et al. 2001). Similarly, overexpression of IKK led to the phosphorylation of p65 at serine 536 (Sakurai et al. 1999). Serine residues 529 and 536 were shown to be required for the activation of the p65 fusion proteins by activated Akt or treatment with IL-1 (Madrid et al. 2001). Thus, activated PI3K and Akt appear to induce NF-B-dependent transcription by activating p65 rather than by promoting the degradation of IB and nuclear translocation of NF-B.

As mentioned above, in mouse embryonic fibroblasts lacking functional T2K, IB phosphorylation and degradation occur in response to TNF, but NF-B reporter genes are not activated, suggesting the possibility that phosphorylation of Rel proteins by T2K is required for transcriptional activity. In addition, inactivation of the mouse gene encoding the glycogen synthase kinase-3 (GSK-3) has no effect on the degradation of IB and nuclear translocation of p65 in response to TNF, yet causes decrease in NF-B DNA-binding activity and a defect in NF-B transcriptional activity (Hoeflich et al. 2000). Thus, GSK-3 and T2K appear to act downstream of IB degradation, possibly at the level of Rel protein activation. Another possible explanation for the NF-B phenotype observed in the GSK and T2K knockout MEFs is that these kinases may play a role in the activation of only a certain subset of Rel dimers. For example, they could promote the processing of Rel protein precursors (p100 and p105), similar to the function of NIK in p100 processing (Xiao et al. 2001; Yin et al. 2001). In any case, there is ample evidence to support the idea that p65 phosphorylation is critical for full NF-B activation, and this activation can be uncoupled functionally from IB phosphorylation and degradation.

Inducible phosphorylation of Rel proteins appears to function at many different levels, including inducing conformational changes in the activation domain, increasing DNA binding activity, and promoting association with transcriptional coactivator proteins such as CBP/p300 (Schmitz et al. 2001). The overall function appears to be in the integration of signaling pathways to activate distinct Rel proteins in response to different signals.

	Toll-like receptors

Top Innate immunity Innate immune signaling... Mechanisms of NF-B activation IB kinases Proteolytic processing of NF-... Rel phosphorylation Toll-like receptors Receptor proximal signaling... Conclusions References

The TLRs are characterized by an extracellular domain containing a number of leucine-rich repeats (LRRs), a single pass transmembrane domain, and an intracellular signaling domain, referred to as a TIR (Toll/IL-1R/Resistance) domain (Wilson et al. 1997; Zhang and Ghosh 2001). The TIR domain is common to the TLRs, the IL-1R family, and a number of plant genes that are required for host defense signaling. The IL-1R family is distinct from the TLR family because of a dissimilar extracellular domain, which consists of Ig domains. In all these receptors, the TIR domain mediates activation of intracellular signaling pathways. At least 10 TLRs are found in the mammalian genome (Aravind et al. 2001; Zhang and Ghosh 2001), whereas Drosophila encodes 9 (Tauszig et al. 2000).

In mammals, TLR4 is required for LPS-mediated signaling and is believed to directly recognize LPS. LPS is a potent activator of innate immunity, and excessive exposure to LPS, or endotoxin, causes serious pathological effects (Beutler 2000). The mechanism of LPS signaling remained a mystery until the discovery of the human TLRs and the finding that hTLR4 is necessary for LPS signaling. The clearest demonstration of the role of hTLR4 in LPS signaling was provided by the positional cloning of lps, a mutation that causes hyporesponsiveness to LPS in mice. lps mutants carry loss-of-function mutations in the mTLR4 gene (Poltorak et al. 1998). Furthermore, TLR4 appears to directly bind to LPS. Lipid A is the active component of LPS and deacetylated lipid A is immunostimulatory in murine but not human systems. Two groups recently demonstrated that expression of mTLR4, but not hTLR4, enabled recognition and signaling in response to deacetylated lipid A (Lien et al. 2000; Poltorak et al. 2000). In addition to TLR4, two other extracellular proteins are involved in LPS recognition. CD14, a GPI anchored protein, has long been known to bind to LPS and is required for LPS signaling (Moore et al. 2000). MD-2 interacts with extracellular domain of TLR4 and is necessary for LPS signaling (Shimazu et al. 1999; Schromm et al. 2001). Also, Ulevitch and colleagues recently demonstrated that LPS is found in close proximity to TLR4, CD14, and MD-2 (da Silva Correia et al. 2001). This argues that TLR4, CD14, and MD-2 bind to LPS, possibly in a tetrameric complex. Interestingly, CD14 and TLR4 have also been implicated in the immune recognition of proteins encoded by viral pathogens such as RSV (Kurt-Jones et al. 2000).

Drosophila Toll, the founding member of the TLR family, is required for both development and immunity. During early embryonic patterning, Toll is required for establishing the dorsoventral axis. Toll is activated by the spatially and temporally restricted activation of its putative ligand, Spätzle (Belvin and Anderson 1996). This is believed to be accomplished by the correspondingly restricted activation of a serine protease cascade consisting of the products of the nudel, gastrulation defective, snake, and easter genes (LeMosy et al. 1999, 2001; Dissing et al. 2001). Activation of this protease cascade is regulated by the products of the windbutel and pipe genes. Interestingly, pipe is the only one of the 12 dorsal group genes expressed in a spatially restricted manner, on the ventral side of the embryo. The spatially restricted expression of pipe is therefore thought to be the key signal activating the entire cascade in the correct location to establish ventral cell fates (Sen et al. 1998). The pipe gene product is a heparin-sulfotransferase, and it has been proposed to create a heparin-modified proteoglycan that is required for local activation of the protease cascade. Activation of Toll, by Spätzle, stimulates a signaling pathway culminating in the activation of the Drosophila Dorsal protein and the expression of ventral specific genes such as twist and snail (Belvin and Anderson 1996).

As mentioned earlier the Toll signaling pathway, including Spätzle, is also responsible for the antifungal immunity in insects (Lemaitre et al. 1996). However, none of the genes that function upstream of Spätzle during development are necessary for the antifungal immune response. A mutation in a Drosophila serpin gene (serine protease inhibitor), known as necrotic, has been shown to cause constitutive activation of the antifungal Toll pathway (Levashina et al. 1999). Thus, it appears that a different serine protease cascade is activated by fungal infection and leads to the processing of Spätzle. The identity of these proteases and the mechanism of their activation remain unknown. Thus, in Drosophila, Toll is indirectly required for immune recognition of pathogens.

Analysis of the complete Drosophila genome sequence identified nine different Toll-related receptors: Toll, 18wheeler, and dTLR3-9 (Tauszig et al. 2000). It is possible that one or more of these Drosophila TLRs function as an LPS receptor, analogous to the role of TLR4 in mammals. In fact, mutants in the TLR receptor 18wheeler have some defects in antibacterial immune signaling. Specifically, the antimicrobial peptide gene attacin is not fully activated in response to Escherichia coli infection. However, induction of the other antibacterial peptides, such as diptericin, occurs at near wild-type levels in 18wheeler mutants. Therefore, 18w is not strictly required for the antibacterial immune response and cannot be the sole LPS receptor (Williams et al. 1997). The possible immune function of eight Drosophila TLRs was investigated recently (Tauszig et al. 2000). Remarkably, potentially dominant activated versions of these TLRs did not mimic antibacterial signaling, suggesting that none of the dTLRs are the LPS receptor. One possible explanation for this result is that two dTLRs must heterodimerize to create the fully functional LPS receptor, similar to TLR2 and TL6 in the recognition of peptidoglycan (Ozinsky et al. 2000). Studies with dominant negative receptors, and ultimately genetic experiments with TLR mutants, are necessary to definitively determine the role of the Drosophila TLRs in the insect antibacterial immune response. One extracellular protein involved in LPS signaling in Drosophila is DGNBP1 (Drosophila gram-negative binding protein), which was shown to bind to LPS. DGNBP1 does not have transmembrane or intracellular domain and, like CD14, is held at the membrane by a GPI anchor. Overexpression of DGNBP1 potentiates LPS signaling while antibody interference with DGNBP1 inhibits LPS signaling. This argues that DGNBP1 plays an important role in the antibacterial immune response; however it cannot be responsible for activating intracellular signaling pathways (Kim et al. 2000b).

In mammals TLR2 is thought to be directly involved in recognizing cell wall components from gram positive bacteria and spirochetes, peptidoglycans, and lipoproteins (Krutzik et al. 2001; Zhang and Ghosh 2001). Also, TLR9 is required for the recognition of bacterial DNA (Hemmi et al. 2000). Bacterial DNA is a potent immune stimulatory molecule and is specifically recognized because of its unmethylated CpG dinucleotides (Krieg 1996). Recently, TLR5 has been shown to mediate the induction of the immune response by the bacterial flagellin protein (Hayashi et al. 2001). However, the mechanisms ^<