|Home | About | Journals | Submit | Contact Us | Français|
The sea anemone Nematostella vectensis is the leading developmental and genomic model for the phylum Cnidaria, which includes anemones, hydras, jellyfish, and corals. In insects and vertebrates, the NF-κB pathway is required for cellular and organismal responses to various stresses, including pathogens and chemicals, as well as for several developmental processes. Herein, we have characterized proteins that comprise the core NF-κB pathway in Nematostella, including homologs of NF-κB, IκB, Bcl-3, and IκB kinase (IKK). We show that N. vectensis NF-κB (Nv-NF-κB) can bind to κB sites and activate transcription of reporter genes containing multimeric κB sites or the Nv-IκB promoter. Both Nv-IκB and Nv-Bcl-3 interact with Nv-NF-κB and block its ability to activate reporter gene expression. Nv-IKK is most similar to human IKK/TBK kinases and, in vitro, can phosphorylate Ser47 of Nv-IκB. Nv-NF-κB is expressed in a subset of ectodermal cells in juvenile and adult Nematostella anemones. A bioinformatic analysis suggests that homologs of many mammalian NF-κB target genes are targets for Nv-NF-κB, including genes involved in apoptosis and responses to organic compounds and endogenous stimuli. These results indicate that NF-κB pathway proteins in Nematostella are similar to their vertebrate homologs, and these results also provide a framework for understanding the evolutionary origins of NF-κB signaling.
Under many circumstances, the NF-κB transcription factor signaling pathway is activated to enable animals to respond to environmental stresses, such as pathogens, chemicals, and UV light (14). Once activated, NF-κB transcription factors alter the expression of target genes to counteract these stresses. Examples of genes regulated by NF-κB include those encoding innate immune response factors (e.g., cytokines and antimicrobial peptides), antioxidizing enzymes (e.g., superoxide dismutase and nitric oxide synthase), and antiapoptosis molecules (e.g., Bcl-2, tumor necrosis factor receptor-associated factors [TRAFs], and inhibitor of apoptosis proteins [IAPs]) (14). NF-κB family members have additional roles in development. For example, the Dorsal protein controls the establishment of dorsal-ventral polarity in the Drosophila melanogaster embryo, and mammalian NF-κB proteins control various aspects of immune and liver cell growth and survival (23, 36).
In Drosophila, there are three NF-κB proteins, while in humans and other mammals, there are five NF-κB proteins. All NF-κB family transcription factors have a conserved DNA-binding/dimerization domain called the Rel homology domain (RHD), and all NF-κB family proteins bind specific DNA sequences (κB sites) as homodimers or heterodimers. NF-κB proteins can be divided into two classes based on sequences C-terminal to the RHD. One class (the “NF-κB proteins”) includes Drosophila Relish and the mammalian p50/p105 and p52/p100 proteins. Proteins of this class contain C-terminal ankyrin repeat sequences, which restrict the protein to the cytosol and inhibit the DNA-binding activity of the RHD. The C-terminal regions of the Relish, p105, and p100 NF-κB proteins must be proteolytically cleaved for the proteins to become active. In contrast, members of the “Rel class” of proteins (Dorsal and Dif in Drosophila and c-Rel, RelA, and RelB in mammals) contain C-terminal transactivation domains that are not removed.
The overall regulation of NF-κB activity in mammals has been extensively characterized (14). In its inactive state, NF-κB is bound to its inhibitor, IκB, in the cytoplasm. Upon activation of the pathway by an upstream signal, IκB is phosphorylated by an IκB kinase (IKK), which targets IκB for ubiquitination and degradation. Free NF-κB then translocates to the nucleus and binds to κB sites in the promoters of specific target genes to alter transcription. One well-characterized transcriptional target of NF-κB is the IκBα gene (15). Thus, NF-κB regulates its own activity by promoting the transcription of its primary inhibitor in a negative feedback loop.
In mammals, there are several IκB proteins that have overlapping but nevertheless distinct functions (these proteins are IκBα, IκBβ, IκB, and Bcl-3). All IκB proteins contain multiple copies of ankyrin repeat domain sequences, which mediate binding to the RHD sequences. The C-terminal regions of the mammalian NF-κB proteins also contain ankyrin repeat domains (Fig. (Fig.1A)1A) and consequently function as IκBs to inhibit the nuclear translocation of NF-κB. These C-terminal regions are phosphorylated and degraded in response to certain NF-κB-activating signals.
Nematostella vectensis is a small, burrowing sea anemone native to estuaries on the Atlantic coast of North America (29). Nematostella is a member of the phylum Cnidaria, which includes sea anemones, hydras, jellyfish, and corals. Like all anemones, it has two tissue layers, the endoderm and the ectoderm, and lacks the mesodermal layer found in triploblastic metazoans such as nematodes, insects, and vertebrates. The most recent common ancestor of Nematostella and triploblasts is thought to have lived approximately 600 million years ago (26, 27).
Nematostella is emerging as the leading model system for cnidarians, in part because its complete genome sequence is known (28). Mining of genomic and expressed sequence tag (EST) databases revealed a single NF-κB gene and two IκB-like genes in Nematostella (33). The RHD of the Nematostella vectensis NF-κB protein (Nv-NF-κB) is ~50% identical to the RHDs of human p50 and p52 but is only ~35% identical to the Rel proteins (RelA, c-Rel, and RelB). Moreover, the RHD is immediately followed by a glycine-rich region, which is also present in mammalian p50 and p52. Surprisingly, the Nv-NF-κB protein lacks the C-terminal ankyrin repeat domain found in all other NF-κB proteins (Fig. (Fig.1A).1A). The ankyrin repeats of the Nv-IκB proteins, Nv-IκB and Nv-Bcl-3, are 42% and 48% identical to the ankyrin repeat domains of human p105 and Bcl-3, respectively. Additionally, Nematostella has two naturally occurring Nv-NF-κB alleles whose encoded proteins differ at 10 amino acid residues and have distinct DNA-binding and transactivating abilities (35).
NF-κB homologs have been identified in a few other basal marine organisms. The coral Acropora millepora, an anthozoan cnidarian like Nematostella, has ESTs corresponding to NF-κB (GenBank accession number EZ002200.1) (21) and IκB (GenBank accession number DY582971) (22). However, the genome of the cnidarian Hydra magnipapillata apparently lacks an NF-κB homolog (6). A single NF-κB homolog exists in the demisponge Amphimedon queenslandica of the phylum Porifera (13). This sponge NF-κB protein contains both the RHD and the C-terminal ankyrin repeat domain, suggesting that the independent NF-κB and IκB genes in Nematostella are the result of a gene splitting event. The biological function of the NF-κB signaling pathway in these simple marine organisms is not known; however, NF-κB has been proposed to play a role in mediating host-symbiont interactions in Cnidaria (37).
In an effort to understand the evolutionary origins of NF-κB signaling, we have cloned cDNAs for Nv-NF-κB, Nv-IκB, Nv-Bcl-3, and Nv-IKK and expressed these proteins in a variety of cell-based assays. We demonstrated that the activities of these Nematostella proteins are conserved with respect to their vertebrate homologs. In addition, we showed that Nv-NF-κB is expressed primarily in a subset of ectodermal cells in both juvenile and adult Nematostella anemones.
The cDNA for the Nv-NF-κB Ser allele protein has been described previously (35). Genomic draft assemblies and EST databases from www.StellaBase.org (34) and the Joint Genome Institute (28) were used to design primers to amplify full-length cDNAs for Nv-IκB and Nv-Bcl-3. A 3′ rapid amplification of cDNA ends (RACE) PCR product generated with the SMART RACE cDNA amplification kit (Clontech) was sequenced to complete a partial 5′ predicted coding sequence for Nv-IKK. To isolate cDNAs encoding Nv-IκB, Nv-Bcl-3, and Nv-IKK, total RNA was extracted from animals by use of TRIzol (Invitrogen), and cDNA was created using Moloney murine leukemia virus (MMLV) reverse transcriptase (Promega). Gene-specific primers were then used to amplify each of the three full-length cDNAs. These cDNAs were then subcloned into a pcDNA3.1(+) (Invitrogen), pcDNA-FLAG, or pcDNA-Myc expression vector. We previously identified two Nv-NF-κB alleles whose proteins are in part defined by a Cys/Ser polymorphism at residue 67 (35); in this study, we used the Nv-NF-κB Ser allele in all in vitro experiments. A complete list of primers and details about plasmid constructions are included in the supplemental material.
The yeast pGBT9 expression vector has been described previously (19). Three different segments of the Nv-NF-κB cDNA encoding amino acids (aa) 3 to 440, 3 to 393, and 388 to 440 were amplified by PCR and subcloned into pGBT9 to create GAL4-Nv-NF-κB expression vectors. Additionally, a PCR fragment encoding aa 3 to 393 of the Nv-NF-κB protein was subcloned into the pcDNA-Myc expression vector to create Myc-Nv-NF-κB-ΔC.
The glutathione S-transferase (GST) bacterial expression plasmid pGEX-KG has been described previously (12). GST-Nv-IκB aa 20 to 67 was created by subcloning nucleotides corresponding to codons 20 through 67 of the Nv-IκB cDNA into pGEX-KG. Site-directed mutagenesis was used to make point mutations in GST-Nv-IκB aa 20 to 67 to create S42A, S47A, and S42A/S47A (S42,47A) mutants.
To create the Nv-ikb promoter luciferase reporter plasmid, primers were used to amplify 221 nucleotides of genomic DNA upstream of the Nv-ikb start codon. This sequence was cloned into the pGL3 luciferase reporter plasmid (Promega). Site-directed mutations of the two κB sites in the Nv-ikb promoter were performed by overlapping PCR-based mutagenesis.
A293 human kidney carcinoma cells and DF-1 chicken fibroblasts were grown in Dulbecco's modified Eagle's medium (DMEM) (Invitrogen) supplemented with 10% fetal bovine serum (Biologos), 50 U/ml penicillin, and 50 μg/ml streptomycin. Transfection of human A293 cells with expression plasmids was performed using polyethylenimine (Polysciences, Inc.) essentially as described previously (35). Briefly, on the day of transfection, cells were incubated with plasmid DNA at a 1:6 ratio of DNA to polyethylenimine. Media were changed 24 h posttransfection, and whole-cell lysates were prepared 24 h later.
Western blots were performed on whole-cell lysates prepared in AT buffer (20 mM HEPES, pH 7.9, 150 mM NaCl, 1 mM EDTA, 1 mM EGTA, 20% wt/vol glycerol, 1% [wt/vol] Triton X-100, 20 mM NaF, 1 mM Na4P2O7·10H2O, 1 mM dithiothreitol, 1 mM Na3VO4, 1 mM phenylmethylsulfonyl fluoride, 1 μg/ml leupeptin, 1 μg/ml pepstatin A, 10 μg/ml aprotinin) as described previously (35). Antisera against FLAG (catalog no. 2368; Cell Signaling) and Myc (SC-40; Santa Cruz Biotechnology) epitopes were used at dilutions of 1:1,000 (for FLAG) and 1:100 (for Myc). Nitrocellulose membranes were incubated with either an anti-rabbit (FLAG) or anti-mouse (Myc) horseradish peroxidase-linked secondary antiserum, and complexes were detected with SuperSignal West Dura extended-duration substrate (Pierce).
To create Nematostella whole-cell lysates, animals were washed three times with distilled water (dH2O), pulverized with a pestle in AT lysis buffer, and then sonicated two times for 10 s each in a 550 Sonic Dismembrator (Fisher Scientific). NaCl was added to the lysates to a final concentration of 150 mM and samples were centrifuged at ~16,000 × g in a microcentrifuge for 20 min at 4°C. The supernatant was discarded, the pellet was resuspended in AT buffer, and this extract was used in Western blotting or electrophoretic mobility shift assays (EMSAs).
An affinity-purified rabbit polyclonal antiserum was prepared (Open Biosystems) against a C-terminal peptide (PADFLQQGVSTQNPSNM) of the Cys allele Nv-NF-κB protein (35). When the anti-Nv-NF-κB antiserum was used, filters were blocked overnight at 4°C in phosphate-buffered saline (PBS) containing 8% milk, 5% normal goat serum (Gibco), and 0.05% Tween 20. Nitrocellulose membranes were then incubated with Nv-NF-κB antiserum (1:2,000 dilution in blocking buffer) for 2 h at room temperature, and immunoreactive bands were detected with an anti-rabbit horseradish peroxidase-conjugated secondary antibody and substrate as described above.
Coimmunoprecipitations were performed essentially as described previously (11). A293 cells were transfected with 5 μg of pcDNA3.1(+) or expression plasmids for FLAG-Nv-NF-κB, Myc-Nv-IκB, Myc-Nv-Bcl-3, or both FLAG-Nv-NF-κB and Myc-Nv-IκB or Myc-Nv-Bcl-3. Whole-cell lysates were harvested in AT buffer 48 h posttransfection. Protein concentrations were determined with the Bio-Rad protein assay reagent, and 1% of total protein was removed as an input control. The remaining lysate was incubated with anti-Myc antiserum. Input controls and immunoprecipitates were analyzed for both FLAG- and Myc-tagged proteins by Western blotting.
EMSAs were performed essentially as described previously (35). T4 polynucleotide kinase (New England BioLabs) was used to end label oligonucleotide probes with 32P. The probes were the major histocompatibility complex (MHC) probe (5′-TCGAGAGGTTGGGGATTCCCCACCCG-3′), the κB1 probe (5′-TCGAGAGGTCGGGGAATCCCCCCCCG-3′), and the κB2 probe (5′-TCGAGAGGTCGGGAAATTCCCCCCCG-3′) (κB consensus sites are underlined). The MHC probe contains the same κB site sequence (from the human MHC gene) that is contained in the 3×-κB luciferase reporter plasmid (see below). Whole-cell extracts were prepared in AT buffer as previously described (19, 35). Lysates from either A293 cells or Nematostella containing 50 μg of protein were incubated with the radiolabeled probe (100,000 cpm) in a 50-μl reaction volume for 30 min at room temperature. Samples were electrophoresed on 5% nondenaturing polyacrylamide gels, and protein-DNA complexes were detected by autoradiography. For supershifts, 1.5 μl of anti-FLAG or anti-Nv-NF-κB antiserum was added to the cell lysates and the samples were incubated at 4°C for 12 h before the addition of the radiolabeled probe.
Luciferase reporter assays were performed in A293 cells essentially as described previously (11). Three reporter plasmids containing κB site-containing promoter sequences upstream of the luciferase gene were used: (i) plasmid 3×-κB, with three κB sites from the human MHC-1 promoter (24); (ii) pNV-IκB-WT-pro-κB3, with the wild-type (WT) Nv-ikb promoter cloned from Nematostella; and (iii) pNV-IκB-dmut-pro-κB3, which has the Nv-ikb promoter with mutations in both predicted κB sites. For transfections, 0.5 μg of the luciferase reporter and 0.5 μg of a pGK-βgal normalization plasmid were used, and the total amount of DNA was kept equal by the addition of the pcDNA empty vector. Two days after transfection, lysates were prepared and luciferase activity was measured with the Luciferase Assay System (Promega) according to the manufacturer's instructions. Values were normalized to β-galactosidase activity in all assays as described previously (19). GAL4 site LacZ reporter gene assays were performed in Saccharomyces cerevisiae as described previously (19).
Chicken DF-1 fibroblast cells were used for indirect immunofluorescence because they have a flat morphology that makes them useful for subcellular localization experiments (32). DF-1 cells were transfected with 3 μg of empty vector or a FLAG or Myc expression vector. Two days after transfection, cells were transferred onto glass coverslips. The next day, indirect immunofluorescence was performed with rabbit anti-Nv-NF-κB (1:60) or mouse anti-Myc (1:60) primary antiserum. Primary antibodies were then detected with either fluorescein isothiocyanate (FITC)-conjugated anti-rabbit IgG (1:60; Sigma) or Texas Red-conjugated anti-mouse IgG (1:80; Vector Labs) antiserum as described previously (32). Coverslips were mounted in a glycerol-based mounting medium containing DAPI (4′,6-diamidino-2-phenylindole), and samples were visualized on an Olympus 1X70 microscope.
Nv-IKK immune complex kinase assays were performed essentially as described previously (12). Human A293 cells were transfected with pcDNA-FLAG-Nv-IKK, and the kinase was immunoprecipitated with anti-FLAG beads (Sigma). Immunoprecipitates were incubated with 4 μg of GST or the GST-Nv-IκB aa 20 to 67 fusion proteins (WT and S42A, S47A, and S42,47A mutants) and 5 μCi [γ-32P]ATP in kinase reaction buffer for 30 min at 30°C. After denaturation, samples were electrophoresed on an SDS-polyacrylamide gel, and 32P-labeled GST-Nv-IκB proteins were detected by phosphorimaging. In parallel, 4-μg samples of GST and the GST-Nv-IκB proteins were electrophoresed on a 10% SDS-polyacrylamide gel, and proteins were detected by staining with Coomassie blue (Bio-Rad).
Anemones were fixed in 4% formaldehyde, washed consecutively in 5, 10, 15, 20, 25, and 30% sucrose, mounted in OCT compound (Tissue Tek), sectioned into 12-μm transverse sections, and mounted onto slides. Prior to immunohistochemistry, slides were heated at 60°C for 30 min, washed twice in xylene, three times in 100% ethanol, once in 70% ethanol, and twice in dH2O, then heated in a 98°C bath of citrate acid buffer (10 mM citric acid monohydrate, 10 mM sodium citrate tribasic dehydrate, 0.05% Tween 20, pH 6.0) for 30 min, and washed once in dH2O. Slides were incubated in blocking buffer (PBS, pH 7.4, with 5% normal goat serum [Gibco], 1% bovine serum albumin [Sigma], 0.05% Tween 20) for 2 h. Slides were then incubated with anti-Nv-NF-κB antiserum (1:100 dilution in blocking buffer) overnight at 4°C. The next day, slides were washed four times in PBS-T (1× PBS with 0.05% Tween 20) and incubated for 2 h with FITC-conjugated anti-rabbit secondary antibody (Sigma) (1:160 dilution in blocking buffer). Slides were washed three times in PBS-T, and coverslips were mounted with Vectashield HardSet mounting medium containing DAPI (Vector Labs). Slides were visualized by fluorescence microscopy.
For whole-mount immunofluorescence, 4-week-old anemone polyps were fixed in 4% formaldehyde overnight, transferred to glass tubes, heated in citrate acid buffer, washed in dH2O, and blocked as described above. Fixed anemones were then incubated with primary and secondary antisera and visualized by fluorescence microscopy.
A BLASTp search was performed with the kinase domains of human IKKα, IKKβ, IKK, and TBK to identify putative IKK family members from a phylogenetically broad range of taxa. When a single taxon was found to possess multiple proteins that had already been characterized as IKKs, they were labeled sequentially (e.g., Strongylocentrotus purpuratus has two IKKs, which were designated SpIKK1 and SpIKK2). Previously uncharacterized proteins were simply designated kinases (e.g., Branchiostoma floridae has BfKIN1 and BfKIN2). The MEK kinases have approximately 28% sequence identity to the kinase domain of Nv-IKK and were used as an outgroup. Accession numbers for all sequences used can be found in the supplemental material (Table S1).
Thirty-eight kinase sequences were aligned using ClustalW2 (http://www.ebi.ac.uk/Tools/clustalw2/index.html). All positions harboring alignment gaps were deleted to produce a gap-free alignment consisting of 209 amino acid residues from the kinase domains of these 38 proteins. The phylogenetic relationships among kinase sequences were inferred from the gap-free alignment using maximum likelihood analysis performed by using RAxML (v.7.0.3) (31) as implemented in the CIPRES portal (v.2.0; http://www.phylo.org/portal2/home.action). One-hundred replicates of the bootstrap were performed to evaluate the support for specific clades. Trees were then visualized and illustrated with FigTree v1.1.2 (http://tree.bio.ed.ac.uk/software/figtree/).
Pooled adult Nematostella cDNA was sequenced de novo by using a Genome Analyzer (Illumina). A total of 37,468,448 single reads were assembled with Velvet and Oases (40) into a preliminary draft of 96,756 contigs (unpublished). Contigs were compared to Nematostella clustered ESTs (28), and 30,813 unique ESTs were added to the contig data sequences to create an inclusive collection of sequenced cDNA.
The proteins encoded by 416 validated human NF-κB target genes (from www.nf-kb.org) were used in a tBLASTn (1) search against Nematostella contigs/ESTs. By use of an E value cutoff of less than 10−10, a total of 7,058 Nematostella-homologous contigs/ESTs corresponding to 224 human proteins were identified. The directionality of the contigs/ESTs was determined by alignment to the human proteins. The contigs/ESTs were then used in a nucleotide BLAST search to map their location on the Nematostella genome (28). For these loci, the location of 2,000 bp upstream was determined.
An NF-κB-binding motif (MA0105.1) from the JASPAR transcription factor database (5) was used to identify potential NF-κB-binding sites in the Nematostella genome. The motif was entered into FIMO (3) and used to identify 57,324 κB sites with a P value cutoff of 10−4. The locations of these κB sites were compared to the 2,000-bp regions upstream of the NF-κB target gene contigs/ESTs. Nematostella homologs to human proteins were determined by the lowest E value at given loci with a minimum cutoff of 10−20. This list was further curated to remove redundant genomic loci. From that list, we identified 134 different Nematostella genes with κB sites within 2,000 bp of the start of the contig/EST. These 134 different genes were homologous to 74 different human proteins (see Table S2 in the supplemental material). Gene Ontology (GO) terms for these human genes/proteins were determined with DAVID (8, 17).
The sequences for the Nematostella cDNAs amplified in this study are available in the GenBank database (http://www.ncbi.nlm.nih.gov) under the following accession numbers: EU092640.1 (Nv-NF-κB Ser allele), HM754642 (Nv-NF-κB Cys allele), EU092641.1 (Nv-IκB), HM754643 (Nv-Bcl-3), and HM754644 (Nv-IKK).
Putative Nv-nfkb, Nv-ikb, Nv-bcl3, and Nv-ikk genes and ESTs were previously identified through bioinformatic searches of N. vectensis genomic and EST databases (22, 33, 35). Based on those and new bioinformatics analyses of predicted gene sequences and ESTs, primers were designed, and the four full-length cDNAs were amplified from adult N. vectensis mixed-population cDNA. These cDNAs were then cloned into a pcDNA-based FLAG expression vector and were subjected to DNA sequencing to predict the protein sequences and domains (as shown in Fig. Fig.1A).1A). Empty-vector and FLAG expression plasmids for the four Nv proteins were transfected into human A293 cells, and each expressed an appropriately sized protein, as determined by anti-FLAG Western blotting (Fig. (Fig.1B1B).
Nv-NF-κB was previously reported to bind DNA in an EMSA (35). To demonstrate the specificity of this binding, cell lysates from A293 cells transfected with empty vector or FLAG-Nv-NF-κB were used in an EMSA using a probe containing a consensus κB site (κB1) that is located upstream of the Nv-ikb gene (Fig. (Fig.2A;2A; see also Fig. 8A). With extracts from FLAG-Nv-NF-κB-transfected cells (Fig. (Fig.2A,2A, lane 2), we detected a DNA-protein complex that was not present with extracts from empty-vector-transfected cells (lane 1). This new DNA-protein complex was supershifted with both anti-FLAG (Fig. (Fig.2A,2A, lane 3) and anti-Nv-NF-κB (lane 4) antisera. A DNA-protein complex of similar size was detected using extracts from adult Nematostella (Fig. (Fig.2A,2A, lane 5), and this complex was also supershifted by anti-Nv-NF-κB antiserum (lane 6).
As shown in Fig. Fig.2B,2B, cotransfection of A293 cells with the multimeric 3×-κB luciferase reporter plasmid and increasing amounts of the FLAG-Nv-NF-κB expression plasmid resulted in a dose-dependent increase in expression from the reporter plasmid. At the optimal ratio of expression plasmid to reporter plasmid, there was an approximately 19-fold increase in reporter gene expression compared to the level seen in cells transfected with the empty-vector control. Taken together, these results show that Nv-NF-κB can specifically bind to a consensus κB site and activate expression of a κB site-containing reporter plasmid.
Because mammalian NF-κB proteins p50 and p52 are not usually activators of transcription by themselves (15), we next sought to characterize further the sequence requirements for transactivation by Nv-NF-κB. First, in κB site reporter gene assays in A293 cells, we found that deletion of residues (aa 394 to 440) C-terminal to the RHD substantially reduced the ability of Nv-NF-κB to activate transcription (Fig. (Fig.2C).2C). Second, to determine whether Nv-NF-κB sequences have an inherent ability to activate transcription (i.e., when not bound directly to DNA), we measured their transactivation abilities as GAL4 fusion proteins in yeast (which does not contain any NF-κB proteins). As shown in Fig. Fig.2D,2D, GAL4-Nv-NF-κB (aa 3 to 440) activated transcription approximately 45-fold above the level induced by GAL4 (aa 1 to 147) alone. Deletion of sequences C-terminal to the RHD (GAL4-NF-κB-ΔC) essentially abolished the ability of GAL4-Nv-NF-κB to activate transcription. However, the Nv-NF-κB C-terminal sequences could not by themselves activate transcription. Thus, Nv-NF-κB appears to have an intrinsic transactivating ability that requires sequences C-terminal to the RHD for optimal activity in both yeast and mammalian cells.
In insects and vertebrates, IκB proteins act as direct inhibitors of NF-κB complexes (14). Our previous bioinformatic and phylogenetic analyses suggested that the protein we named Nv-IκB is an IκB ortholog (35); however, the ankyrin repeat is a common protein motif, and there are over 146 ankyrin proteins in Nematostella. Therefore, it was important to demonstrate that Nv-IκB could interact with Nv-NF-κB and inhibit its activity.
To determine whether Nv-NF-κB and Nv-IκB can interact, A293 cells were cotransfected with expression plasmids for FLAG-Nv-NF-κB and Myc-Nv-IκB, whole-cell lysates were then immunoprecipitated with an anti-Myc antibody, and samples were analyzed by Western blotting with either anti-FLAG or anti-Myc antiserum. FLAG-Nv-NF-κB was detected in anti-Myc immunoprecipitates only when FLAG-Nv-NF-κB and Myc-Nv-IκB were coexpressed (Fig. (Fig.3A,3A, upper left panel). Western blotting of the whole-cell lysates used in these immunoprecipitations confirmed that FLAG-Nv-NF-κB and Myc-Nv-IκB were expressed in the appropriate cell extracts (Fig. (Fig.3A,3A, right two panels). Moreover, transfection of increasing amounts of an Nv-IκB expression plasmid inhibited the ability of FLAG-Nv-NF-κB to activate expression of the 3×-κB reporter plasmid in A293 cells (Fig. (Fig.3B).3B). These results indicate that Nv-IκB can bind to and inhibit the activity of Nv-NF-κB.
As determined by indirect immunofluorescence, Nv-NF-κB, when expressed alone, localized exclusively in the nuclei of DF-1 chicken fibroblast cells (Fig. (Fig.3C).3C). On the other hand, Myc-Nv-IκB was distributed approximately equally throughout the cytoplasm and nucleus (Fig. (Fig.3D),3D), similar to what is seen when mammalian IκBα is overexpressed in tissue culture cells (30). Coexpression of Myc-Nv-IκB and Nv-NF-κB resulted in an entirely cytoplasmic localization for both proteins (Fig. (Fig.3E);3E); that is, Nv-NF-κB was detected exclusively in the cytoplasm of 98% of the cells that expressed both Nv-NF-κB and Myc-Nv-IκB (Fig. (Fig.3F3F).
In mammals, Bcl-3 is an ankyrin repeat-containing protein that can bind to NF-κB complexes, such as p50-p50 or p52-p52, but unlike other IκB proteins, Bcl-3 does not always inhibit their DNA-binding or nuclear translocation ability (41). The six ankyrin repeats of Nv-Bcl-3 share 48% sequence identity to six of the seven C-terminal ankyrin repeats of human Bcl-3 (see Fig. S1 in the supplemental material). In coimmunoprecipitation experiments in A293 cells, FLAG-Nv-NF-κB and Myc-Nv-Bcl-3 interacted (Fig. (Fig.4A),4A), and in a reporter assay, Nv-Bcl-3 inhibited the ability of Nv-NF-κB to activate expression of the 3×-κB reporter plasmid (Fig. (Fig.4B).4B). By indirect immunofluorescence, when Nv-NF-κB and Myc-Nv-Bcl-3 were coexpressed in chicken fibroblasts, Nv-NF-κB and Nv-Bcl-3 showed variable, but always overlapping, subcellular localization (Fig. (Fig.4C).4C). Among cells that expressed both Nv-NF-κB and Myc-Nv-Bcl-3, expression was nuclear in 16%, cytoplasmic/nuclear in 56%, and exclusively cytoplasmic in 27% of cells (Fig. (Fig.4D4D).
A comprehensive search of Nematostella genomic and EST databases identified partial sequences for a single IKK-like protein, which we have termed Nv-IKK. Using 3′ RACE and gene-specific PCR, we amplified, sequenced, and cloned a full-length cDNA for Nv-IKK. The Nv-IKK kinase domain is ~59% identical to the kinase domains of human IKK/TBK but is only ~27% identical to that of human IKKα/IKKβ (see Fig. S2 in the supplemental material). According to a phylogenetic analysis using maximum likelihood, the vertebrate IKKs resolve into distinct and well-supported IKKα, IKKβ, IKK, and TBK clades (Fig. (Fig.5A).5A). There is also strong support for sister group relationships between the vertebrate IKKα and IKKβ clades and between the vertebrate IKK and TBK clades. Along with other invertebrate IKKs, Nv-IKK groups with the vertebrate IKK/TBK clade with robust bootstrap support (91%). The MEK kinases, including a single Nematostella sequence, form a well-resolved group distant from either of the two major IKK clades, consistent with the hypothesis that the MEK kinases comprise an outgroup relative to the IKKs.
To establish that the Nv-IKK protein has catalytic activity, we performed an immune complex kinase assay using immunoprecipitated FLAG-Nv-IKK and a bacterially expressed fusion protein containing aa 20 to 67 of Nv-IκB. This region of Nv-IκB contains three serines, two of which (Ser42 and Ser47) are separated by four amino acids, in a region that shares some sequence similarity to the IKK phosphorylation sites (Ser32 and Ser36) in human IκBα (Fig. (Fig.5B).5B). We also created mutants with single Ser-to-Ala changes in Nv-IκB at residues 42 and 47 (S42A and S47A), as well as the S42,47A double mutation. The wild-type (WT) and S42A GST-Nv-IκB peptides were both phosphorylated by FLAG-Nv-IKK (Fig. (Fig.5B).5B). In contrast, the S47A and S42,47A mutated peptides were not detectably phosphorylated by Nv-IKK. These results indicate that S47A is the primary site of Nv-IKK-mediated phosphorylation within the aa 20 to 67 Nv-IκB peptide. Nv-IKK was unable to phosphorylate any serines in GST-Nv-IκB containing aa 68 to 105 (see Fig. S4 in the supplemental material), further indicating that phosphorylation of S47 is specific. Of note, Nv-IKK did not detectably phosphorylate a peptide (aa 1 to 55) from human IκBα (Fig. S5), which contains the Ser residues (aa 32 and 36) phosphorylated by human IKKβ.
To identify cells in Nematostella that express Nv-NF-κB, anti-Nv-NF-κB antiserum was used in immunohistochemistry on sections and whole mounts from 4-week-old (juvenile) polyps (Fig. (Fig.6A)6A) and adult animals (Fig. (Fig.6D).6D). In whole-mount staining of 4-week-old polyps (Fig. (Fig.6B),6B), no signal was detected when preimmune serum was used, whereas numerous, scattered immunopositive cells were detected on the outside of the animal with anti-Nv-NF-κB antiserum. In transverse sections through the bodies of both polyps (Fig. (Fig.6C)6C) and adults (Fig. (Fig.6E),6E), anti-NF-κB immunoreactivity was detected in scattered ectodermal cells. A longitudinal section of an adult animal showed ectodermal staining of NF-κB-positive cells along the entire animal (Fig. (Fig.6F).6F). At high magnification, Nv-NF-κB was detected in the cytoplasm of the ectodermal cells as judged by exclusion from the DAPI-stained nuclei (insets in Fig. 6C and E). Anti-Nv-NF-κB Western blotting of extracts from Nematostella adults primarily detected a band of approximately 50 kDa, which comigrated with Nv-NF-κB overexpressed in A293 cells (Fig. (Fig.6G).6G). Taken together, these results indicate that the ectodermal anti-Nv-NF-κB cytoplasmic staining seen in sections of Nematostella corresponds to the Nv-NF-κB protein.
A literature-based compilation of over 400 mammalian NF-κB target genes has been assembled at www.nf-kb.org. We used 416 genes from that list to predict computationally the number of human NF-κB target genes that are also Nv-NF-κB targets based on gene/protein homology and the presence of at least one consensus κB site within 2,000 bp upstream of the predicted Nematostella gene homologue. Based on this analysis, we found that 224 of the 416 human targets have one or more homologues in Nematostella. It was predicted that 35% (78/224) of those Nematostella homologues of human NF-κB target genes are also targets of Nv-NF-κB. Sorting of these predicted Nv-NF-κB target genes by Gene Ontology-based functions indicates the four most significant biological process categories include genes involved in the regulation of apoptosis (22 genes), the response to organic substances (19 genes), the response to endogenous stimuli (14 genes), and phosphate metabolism (20 genes) (Fig. (Fig.7).7). In these four GO categories, approximately 35 to 65% of the human NF-κB target genes are also predicted to be Nv-NF-κB target genes (Fig. (Fig.7,7, top). The predicted Nv-NF-κB target gene homologs in each of these four high-scoring GO categories are listed in Fig. Fig.77 (bottom).
In vertebrates, one of the best-characterized target genes of NF-κB is the IκBα gene, which contains multiple upstream κB sites, and upregulation of IκB expression by activated NF-κB functions as part of negative regulatory loop (14). Moreover, the Nv-IκB gene was among the Nv-NF-κB targets identified in our computational analysis of targets (Nematostella homolog 2666209_2; see Table S2 in the supplemental material). Analysis of the N. vectensis genomic database (at www.Stellabase.org) revealed two κB sites within the first 337 nucleotides upstream of the start codon of the Nv-ikb gene (Fig. (Fig.8A,8A, upper panel); these sites were termed κB1 (from nucleotide −337 to −327) and κB2 (from −177 to −167). In an EMSA using whole-cell lysates from A293 cells transfected with either empty vector or Nv-NF-κB, we found that Nv-NF-κB bound the κB1 site as efficiently as a κB site from the human MHC promoter (Fig. (Fig.8A,8A, lower panel). In contrast, Nv-NF-κB weakly bound the κB2 site. The strong binding to the κB1 site was competed by the wild-type κB1 probe but not by a mutant κB site probe and was supershifted by anti-Nv-NF-κB antiserum but not by preimmune serum (see Fig. S6 in the supplemental material).
We next determined whether Nv-NF-κB could activate the Nv-ikb promoter. To test this, we subcloned 221 bp upstream of the Nv-ikb gene that contained the two κB sites into a luciferase reporter plasmid and then performed reporter gene assays with A293 cells (Fig. (Fig.8B).8B). Nv-NF-κB increased expression from the Nv-ikb promoter reporter by 2.2-fold compared to empty-vector-transfected cells. The relatively weak activation of the Nv-ikb reporter plasmid by Nv-NF-κB may, in part, be due to the high basal activity of this promoter in A293 cells. Mutation of both κB1 and κB2 sites reduced the ability of Nv-NF-κB to activate the Nv-ikb reporter plasmid from 2.2-fold to 1.3-fold. These results show that Nv-NF-κB can activate a bona fide Nematostella gene promoter in a κB site-dependent manner.
In this work, we have characterized the function and interaction of four proteins that appear to comprise the core elements of the Nematostella vectensis NF-κB transcription factor pathway. Currently, Nematostella is the only diploblastic animal in which the activity of NF-κB pathway proteins has been studied, and it may serve as a model for NF-κB signaling in other members of the phylum Cnidaria.
Nv-NF-κB is most similar to the processed versions of the NF-κB proteins p50 and p52 among human proteins (33). We demonstrated that lysates from both A293 cells transfected with Nv-NF-κB and adult Nematostella animals possess κB site DNA-binding activity that can be supershifted with anti-Nv-NF-κB antiserum. Additionally, we showed that Nv-NF-κB can activate expression of reporter genes in both yeast and human cells and that this transactivating ability requires sequences C-terminal to the RHD. Although the human NF-κB proteins p50 and p52 are not generally activators of transcription by themselves, the processed form of the Drosophila NF-κB protein Relish can activate transcription (38, 39), and this activity also requires sequences C-terminal to the RHD (9).
The role of IκB as a negative regulator of NF-κB appears to be conserved in Nematostella. As in the insect and vertebrate NF-κB pathways, Nv-IκB interacts directly with Nv-NF-κB, blocks its ability to activate transcription, and can sequester Nv-NF-κB in the cytoplasm. Additionally, the Nv-ikb gene has two predicted κB sites upstream of its start codon. Nv-NF-κB binds strongly to one of these sites (κB1), and the two κB sites are necessary for Nv-NF-κB-dependent transactivation of an Nv-ikb promoter reporter locus. These results suggest that Nv-IκB has a function and regulation similar to those of vertebrate IκBs. In both polyp and adult Nematostella anemones, Nv-NF-κB is primarily detected in the cytoplasm of a subset of cells in the ectoderm. Taken together, these results suggest that Nv-NF-κB is retained in the cytoplasm of these ectodermal cells by interaction with Nv-IκB. Nevertheless, we can detect some constitutive Nv-NF-κB DNA-binding activity in whole animal extracts of “uninduced” Nematostella (Fig. (Fig.2A2A).
Nv-Bcl-3 also interacted directly with Nv-NF-κB and blocked its ability to activate transcription of a reporter locus in A293 cells. Human Bcl-3 can also interact directly with p50 and p52 homodimers and, under different circumstances, can be an activator or repressor of NF-κB-dependent transcription (25). Given that our experiments were performed in human A293 cells, it is unclear whether Nv-Bcl-3 acts as a repressor, activator, or both for Nv-NF-κB in Nematostella. Nv-Bcl-3 lacks an obvious phosphodegron motif, suggesting that it is not a target of Nv-IKK phosphorylation. Unlike Nv-IκB, Nv-Bcl-3 did not block nuclear localization of Nv-NF-κB in all cells, even though Nv-NF-κB and Nv-Bcl-3 invariably colocalized, whether in the nucleus, the cytoplasm, or both. One explanation for the cell-to-cell variability in subcellular localization of Nv-Bcl-3/Nv-NF-κB complexes in chicken fibroblasts is that their localization is cell cycle dependent. Of note, the Nv-bcl3 gene has four predicted upstream κB sites (see Table S2 in the supplemental material), suggesting that it could be a target for Nv-NF-κB.
The human IκB kinase family contains four members, and these can be subdivided into two subfamilies based on sequence similarity within their kinase domains. IKKα and IKKβ are approximately 52% identical to each other but only about 27% identical to IKK and TBK, which are approximately 61% identical to each other (16). Maximum likelihood analysis indicates that Nv-IKK is more closely related to the IKK/TBK subfamily than to the canonical IκBα kinases IKKα/IKKβ. The only other cnidarian with a sequenced genome is H. magnipapillata, which has a single IKK-like protein that emerges at the base of the IKK/TBK family on the maximum likelihood tree. In a search of ESTs from the coral A. millepora, we identified a partial transcript that predicts a protein with 75% amino acid identity to a portion of the Nv-IKK kinase domain (see Fig. S3 in the supplemental material); however, this coral sequence was too short to be included in our phylogenetic analysis. Nevertheless, these results suggest that IKK/TBK orthologs are the only IKK-like kinases present in Cnidaria. Thus, an IKK/TBK protein may represent the ancestral form of IKK, or alternatively, IKKα/IKKβ genes have been lost in cnidarians. C-terminal to the kinase domain, Nv-IKK shares little amino acid similarity to human IKK/TBK and lacks residues similar to characterized leucine zipper and helix-loop-helix domains found in all four human IKKs (16).
The primary signal-induced phosphorylation sites for human IKKβ in IκBα are two serines, Ser32 and Ser36, within the sequence DSGLDS (7). Although human TBK can phosphorylate one of these IκBα sites (Ser36) in kinase assays in vitro, the ability of TBK to function as an in vivo kinase of IκBα has been questioned (7, 16). Additionally, IKK only weakly phosphorylates Ser36 of IκBα in an in vitro kinase assay (18). We have found that Nv-IKK can specifically phosphorylate Nv-IκB at Ser47 within the sequence GFGGSL. Of note, Ser47 of Nv-IκB is in a sequence context (SL) that is identical to the optimal consensus target phosphorylation sequence for human IKK (18). Our results establish that Nv-IKK has catalytic activity and that it can phosphorylate a Ser residue in sequences N-terminal to the ankyrin repeat domain of Nv-IκB; however, it will be essential to show that Nv-IKK can phosphorylate Nv-IκB in Nematostella.
Nv-NF-κB is expressed in the cytoplasm of a subset of scattered ectodermal cells of adult and juvenile anemones. The location of the Nv-NF-κB-positive cells is similar to that of other Nematostella cells, including mucus gland cells (10) and cnidocytes (20), identified in histological studies. Interestingly, both mucus secretion and cnidocyte activation are integral parts of the cnidarian response to many environmental stressors (2, 4). Of note, three of the four top-scoring gene categories of biological processes that might be controlled by Nv-NF-κB (responses to organic compounds and endogenous stimuli and the control of apoptosis) could easily be interpreted to be part of an environmental sensing system. Alternatively, the Nv-NF-κB-positive ectodermal cells might be primitive immune cell-like cells, which mediate innate immune responses through activation of NF-κB. Nematostella possesses many genes homologous to those of the insect and vertebrate innate immune responses, including Toll-like receptors, MyD88, and TRAFs (22, 33). Our results do not address whether Nv-NF-κB has a role in early development in Nematostella, similar to that of Dorsal in Drosophila. Future priorities will be to identify the function of Nv-NF-κB-positive cells in Nematostella and the types of signals that can induce nuclear translocation of Nv-NF-κB.
The results shown in this paper indicate that the molecular functions of some members of the Nv-NF-κB pathway have been conserved in both cnidarians and bilaterians over the last 600 million years of evolutionary history. Further studies of the NF-κB pathway in Nematostella will lead to a better understanding of the origins of this key signaling pathway, which is involved in the control of a variety of stress and developmental processes in insects and humans.
This work was supported by grant MCB-0920461 from the National Science Foundation (to J.R.F. and T.D.G.) and Public Health Service grant CA047763 from the National Cancer Institute (to T.D.G.). F.S.W. was supported by a predoctoral grant of the Superfund Basic Research Program at Boston University (5 P42 ES07381) and a Warren-McLeod Fellowship, and F.S.W. and M.R.G. were supported by ARRA supplement CA047763-21S3 (to T.D.G.). N.T.-K. was supported by funding from Conservation International, and N.T.-K. and D.J.S. were supported by Warren-McLeod Fellowships. E.D. was supported by a summer fellowship from New England Biolabs, and H.G. and E.D. were supported by funds from the Boston University Undergraduate Research Opportunities Program.
We thank Ryan Thompson and Ning Jiang for helpful discussions, Jan Blom, Jeannette Connerney, Todd Blute, and Tina Matos for help with immunohistochemistry and microscopy, Brian Granger for help with the bioinformatics analysis, and the BB522 class (Boston University, 2010) for help with the reporter gene assay used for Fig. Fig.2D2D.
Published ahead of print on 28 December 2010.
‡Supplemental material for this article may be found at http://mcb.asm.org/.