IKKα Contributes to Canonical NF-κB Activation Downstream of Nod1-Mediated Peptidoglycan Recognition

Background During pathogen infection, innate immunity is initiated via the recognition of microbial products by pattern recognition receptors and the subsequent activation of transcription factors that upregulate proinflammatory genes. By controlling the expression of cytokines, chemokines, anti-bacterial peptides and adhesion molecules, the transcription factor nuclear factor-kappa B (NF-κB) has a central function in this process. In a typical model of NF-κB activation, the recognition of pathogen associated molecules triggers the canonical NF-κB pathway that depends on the phosphorylation of Inhibitor of NF-κB (IκB) by the catalytic subunit IκB kinase β (IKKβ), its degradation and the nuclear translocation of NF-κB dimers. Methodology Here, we performed an RNA interference (RNAi) screen on Shigella flexneri-induced NF-κB activation to identify new factors involved in the regulation of NF-κB following infection of epithelial cells by invasive bacteria. By targeting a subset of the human signaling proteome, we found that the catalytic subunit IKKα is also required for complete NF-κB activation during infection. Depletion of IKKα by RNAi strongly reduces the nuclear translocation of NF-κB p65 during S. flexneri infection as well as the expression of the proinflammatory chemokine interleukin-8. Similar to IKKβ, IKKα contributes to the phosphorylation of IκBα on serines 32 and 36, and to its degradation. Experiments performed with the synthetic Nod1 ligand L-Ala-D-γ-Glu-meso-diaminopimelic acid confirmed that IKKα is involved in NF-κB activation triggered downstream of Nod1-mediated peptidoglycan recognition. Conclusions Taken together, these results demonstrate the unexpected role of IKKα in the canonical NF-κB pathway triggered by peptidoglycan recognition during bacterial infection. In addition, they suggest that IKKα may be an important drug target for the development of treatments that aim at limiting inflammation in bacterial infection.


Introduction
During pathogen infection, structurally conserved microbial molecules are recognized by germline-encoded pathogen recognition receptors (PRRs) that function as sensors for non-self detection and initiate innate immunity [1,2]. PRRs include transmembrane proteins such as Toll-like receptors and C-type lectin receptors, as well as cytoplasmic proteins such as retinoic acid-inducible gene (RIG)-I-like receptors and NOD-like receptors [3,4,5]. They are expressed in macrophages and dendritic cells but also in various non-professional immune cells including epithelial and endothelial cells. PRRs recognize a large variety of pathogen associated molecular patterns (PAMPs) from both extracellular and intracellular pathogens including lipopolysaccharide, peptidoglycan, lipoproteins, dsRNA, ssRNA, CpG-DNA and flagellin [6]. Signaling pathways of PAMP recognition converge into the activation of the pleiotropic transcription factor nuclear factor-kappa B (NF-kB) that, in the context of innate immunity, regulates the expression of proinflammatory genes encoding cytokines, chemokines, anti-bacterial peptides and adhesion molecules [7]. The mammalian NF-kB family consists of the members RelA/p65, RelB, c-Rel, p50 (NF-kB1) and p52 (NF-kB2) [8]. All five proteins share a Rel homology domain and form homo-and heterodimers that regulate transcription by binding to kB sites in promoters or enhancers of target genes. In unstimulated cells, most of the NF-kB dimers are sequestrated in the cytoplasm by the proteins of the Inhibitor of NF-kB (IkB) family whose prototype is IkBa. In the canonical pathway of NF-kB activation triggered by most stimuli including bacterial and viral infection, cytokines and stress-induced responses, phosphorylation of IkBa on Serine 32 and Serine 36 residues by the IkB kinase (IKK) complex is a decisive regulatory step [9]. The IKK complex is comprised of three subunits: two catalytic subunits, IKKa and IKKb, and the regulatory scaffold component NF-kB essential modulator (NEMO). The respective contribution of IKKa and IKKb in the phosphorylation of IkBa is unclear. Although it is generally accepted that IKKb is critical for IkBa phosphorylation through the canonical pathway, two recent reports demonstrate the equal importance of IKKa for the activation of NF-kB by the inflammatory cytokines interleukin-1 (IL-1) in mouse embryonic fibroblasts and tumor necrosis factor alpha (TNFa) in HeLa cells [10,11]. The phosphorylation of IkBa is followed by its rapid polyubiquitination and subsequent degradation by the 26S proteasome complex [12]. The release of NF-kB with unmasked nuclear localization sequence leads then to the translocation of the transcription factor to the nucleus where it regulates gene expression [13].
Although the role of NF-kB is central to many pathways triggered by pathogen recognition, the molecular processes that govern its activation are only partially elucidated. In particular, the mechanisms triggered by the detection of invasive bacteria such as the pathogen Shigella flexneri remain largely uncharacterized. S. flexneri makes use of a type III secretion (T3S) apparatus to locally rearrange the host actin cytoskeleton and penetrate into intestinal epithelial cells [14]. Once internalized, bacteria multiply in the host cytoplasm and use actinbased motility to spread to adjacent epithelial cells. During infection, massive inflammation is observed in colonic mucosal tissues [15]. In infected epithelial cells, intracellular bacteria release peptidoglycanderived peptides that are specifically recognized by Nod1 [16]. Upon ligand binding, Nod1 homo-dimerizes and recruits the downstream kinase RICK/RIPK2 through heterologous caspase-recruitment domain interactions [17]. This converges to the sequential recruitment and activation of the TAK1/TAB1/TAB2 and IKKa/IKKb/IKKc complexes, the nuclear translocation of NF-kB and the upregulation of proinflammatory genes encoding for cytokines and chemokines, including interleukin-8 (IL-8) and TNFa [18]. The chemokine IL-8 recruits polymorphonuclear cells to the site of infection and therefore contributes to contain the dissemination of bacteria within the intestinal tissue. Interestingly, S. flexneri uses the T3S apparatus to secrete several effectors that alter multiple signaling pathways in infected cells and reduce the expression of proinflammatory genes [19]. Among others, the effector OspF suppresses the expression of IL-8 by dephosphorylating the MAP kinases p38 and ERK in the nucleus of infected cells [20,21], thereby impairing the phosphorylation of Histone H3, a process that regulates the access of chromatin to transcription factors.
Here, we performed an RNA interference (RNAi) screen on S. flexneri-induced NF-kB activation to identify new factors involved in the regulation of NF-kB following infection of epithelial cells by invasive bacteria. By targeting a subset of the human signaling proteome, we identified IKKa as a protein required for S. flexneriinduced NF-kB nuclear translocation and IL-8 secretion in HeLa cells. This result was unexpected because, except for IL-1 and TNFa [10,11], it is generally accepted that IKKb is the component of the IKK complex involved in the canonical pathway of NF-kB activation. Depletion of IKKa or IKKb indicated that S. flexneriinduced NF-kB activation in HeLa cells requires indeed both catalytic subunits. We further characterized the role of IKKa and found that, during S. flexneri infection, IKKa was required for the phosphorylation of IkBa on serines 32 and 36, and for its degradation. Experiments performed with the synthetic Nod1 ligand L-Ala-D-c-Glu-meso-diaminopimelic acid (Tri-DAP) indicated that IKKa was involved in Nod1-mediated signaling pathway of NF-kB activation. Taken together, these results show that, although Nod1 signaling triggers the canonical pathway of NF-kB activation, both IKKa and IKKb are required for full NF-kB activation.

Antibodies and reagents
Antibodies against NF-kB p65, IkBa and IKKa were obtained from Santa Cruz Biotechnology (Santa Cruz, USA) whereas the anti-actin was from Chemicon (Billerica, USA) and the anti-phospho-IkBa was from Cell signaling technology (Beverly, USA). The anti-mouse IgG-Cy5 was obtained from Zymed (San Francisco, USA) and the anti-rabbit IgG-HRP and anti-mouse IgG-HRP from GE Healthcare (Pittsburgh, USA). Hoechst and FITC-phalloidin were from Invitrogen (Carlsbad, USA), TNFa from R & D systems (Minneapolis, USA).

In vitro diced siRNA library
An in vitro diced siRNA library targeting 132 genes coding for a subset of the signaling proteome was generated as previously described [22,23,24]. Briefly, for each gene, a 600 base pair cDNA was generated by PCR from a total cDNA library. An additional set of nested primers was used to add T7 promoters at both ends of the final cDNA fragment. Nested PCR products were subject to in vitro transcription, dicing, and purification to produce gene specific siRNA pools. Dicing was performed with the turbo dicer siRNA generation kit from Genlantis (San Diego, USA). The concentration of all siRNA pools was normalized.

Bacterial strains
The S. flexneri strains M90T wild-type and the icsA (virG) deletion mutant (DvirG) were generously provided by Dr. P. Sansonetti (Institut Pasteur, Paris, France). All strains were transformed with the pMW211 plasmid to express the DsRed protein under control of a constitutive promoter. The pMW211 plasmid was a generous gift from Dr. D. Bumann (Biozentrum, University of Basel, Switzerland). The DospF deletion mutant used in IL-8 expression experiments, was generated from the DvirG mutant by allelic exchange using a modification of the lambda red-mediated gene deletion [25]. Briefly, the genes for lambda red recombination were expressed from the pKM208 plasmid [26]. The chloramphenicol resistance cassette (cat) of the pKD3 plasmid was amplified using the primers listed in Table 1. After DpnI digestion, the PCR product was electroporated into the DvirG mutant. Recombinants were selected on TSB plates containing 5 or 10 mg ml 21 chloramphenicol. The cat cassette was removed by transformation of pCP20 and incubation at 30uC on TSB plates containing 100 mg ml 21 ampicillin [25]. Single colonies were screened by PCR.

Infection assay
Bacteria were routinely grown in tryptic soy broth (TSB) medium, used in exponential growth phase, and treated with poly-L-lysine prior infection. HeLa cells, seeded in 96-well plates, were serum starved for 30 min and infected with S. flexneri at a multiplicity of infection (MOI) of 10. Immediately after adding bacteria, the plates were centrifuged for 5 min at 2000 rpm and placed at 37uC for 30 min. Extracellular bacteria were killed by addition of gentamycin (50 mg/ml).

Immunofluorescence
Cells were fixed with 4% PFA for 6 min and permeabilized in 0.5% Triton X-100 for 10 min. They were, then, incubated with a mouse monoclonal anti-p65 antibody (1 mg/ml) overnight at 4uC and stained with a Cy5-conjugated secondary antibody and Hoechst (10 mg/ml) for 40 min at room temperature.
siRNA screen of S. flexneri-induced p65 nuclear translocation The in vitro diced siRNA library was screened on S. flexneriinduced p65 translocation assay in a 96-well format. The firefly luciferase (GL3) siRNA was used as a non-silencing negative control as described previously [22,23,24]. siRNA pools against Nod1, RIPK2 and Src were used as positive controls. The screen was performed three times in duplicate as follows. Three thousand HeLa cells per well were transfected by reverse transfection with the individual 132 siRNA pools in 96-well plates. After 48 hours, cells were infected with DsRed S. flexneri at MOI of 10 for 90 min and then fixed, permeabilized, and stained for p65, F-actin, and DNA. Images were acquired at 12 random sites of each well using the automated ImageXpress microscope (Molecular devices, Sunnyvale, USA). At each site, images at 360 nm, 480 nm, 594 nm, 640 nm were acquired to visualize Hoechst, Phalloidin, DsRed S. flexneri and p65, respectively. The nuclear localization of p65 was automatically quantified by using the Enhanced-Translocation module of MetaXpress (Molecular devices, Sunnyvale, USA). Briefly, the Hoechst staining was used as a mask to automatically identify nuclei in the p65 staining image. The cytoplasmic area was defined by a ring around each nucleus. For each cell, the ratio of p65 intensity in the nucleus and in the cytoplasmic ring defined as the Nuc/Cyt p65 ratio was calculated and averaged over several thousands of cells per well. The results of the screen were expressed as individual scores. The score of a particular gene represents the fold standard deviation from the mean of the GL3 control wells. A negative or a positive sign was assigned to the score when the Nuc/Cyt p65 ratio was lower or higher than the GL3 control ratio, respectively.

Enzyme-linked Immunosorbent Assay (ELISA)
IL-8 secretion was measured by ELISA in the supernatant of HeLa cells 6 hours post infection. Cell-free supernatants from triplicate wells were analyzed for their IL-8 content using a commercial ELISA kit (BD Pharmingen, San Jose, USA).

Western blot analysis
HeLa cells were transfected with siRNAs in a 6-well plate. 72 hours post transfection, cells were lysed in Phosphosafe Extraction Buffer (Novagen, Darmstadt, Germany) supplemented with 1x protease inhibitor cocktail (Calbiochem, Darmstadt, Germany). Protein concentration was measured using the bicinchonic acid (BCA) kit (Pierce, Rockford, USA). Equal amounts of proteins were resolved by SDS-PAGE and transferred to Hybond C-Extra membrane (Amersham Bioscience, Pittsburgh, USA) for immunoblotting with individual antibodies. Primary antibodies were detected using horseradish peroxidase-conjugated anti-rabbit or anti-mouse IgG antibodies, and visualized with the ECL system (Pierce). Quantification of the blots was performed using the densitometry feature of Photoshop.

Tri-DAP treatment
Cells were serum starved 30 min before treatment with L-Ala-c-D-Glu-mesoDAP (Tri-DAP). Tri-DAP treatment was performed by calcium phosphate transfection using 0.4 mg/ml Tri-DAP final concentration for 90 min. Then cells were fixed with 4% PFA and analyzed by immunofluorescence as described above.

Statistical analysis
Data are expressed as mean 6 standard deviation calculated from the number of replicates specified in figure legends. p values were calculated with a two-tailed two-sample equal variance t-test.

Results
RNAi screen identifies the role of IKKa in the nuclear translocation of NF-kB p65 during infection of epithelial cells by S. flexneri Over the past 10 years, RNAi has been an essential tool to study gene function in mammalian cells [24,27,28]. Using S. flexneriinduced NF-kB activation in HeLa cells as a model system, we performed an image-based RNAi screen to identify new proteins involved in the molecular mechanisms that control NF-kB activation following pathogen recognition. The nuclear localization of NF-kB p65 visualized by anti-p65 immunofluorescence microscopy was used as readout for NF-kB activation. In uninfected cells, p65 was mainly localized in the cytoplasm ( Figure 1A, left panel). In contrast, following infection with dsRedexpressing S. flexneri, a strong nuclear translocation of p65 was observed ( Figure 1A, right panel). To validate the RNAi approach, silencing of the intracellular pattern recognition receptor Nod1 involved in S. flexneri recognition was tested. HeLa cells were transfected with pools of in vitro diced small interference RNA (siRNA) targeting Nod1 or the firefly Luciferase used as control (GL3). After 48 hours, cells were infected with S. flexneri, stained for p65 and DNA with an anti-p65 antibody and Hoechst, respectively. Visual inspection of images showed that S. flexneriinduced p65 nuclear translocation was suppressed in Nod1depleted cells ( Figure 1B). This observation was quantified by measuring for each cell the ratio of p65 intensity in the nucleus and in the cytoplasm by automated image processing ( Figure 1C). The Hoechst staining was used to automatically identify nuclei whereas the cytoplasmic area was defined by a ring around each nucleus. Quantification of the Nucleus/Cytoplasm p65 intensity ratio (Nuc/Cyt p65 ratio) confirmed that the depletion of Nod1 inhibited the nuclear translocation of p65 induced during infection ( Figure 1D), and therefore, that the RNAi approach was suitable to identify new proteins involved in the activation of NF-kB during S. flexneri infection of epithelial cells. An in vitro diced siRNA library targeting 132 genes from the human signaling proteome was screened on S. flexneri-induced p65 nuclear translocation as described in Materials and Methods. For each gene, a score representing the fold standard deviation from the mean of GL3 control wells was calculated (Figures 1E and  Table S1). A negative or a positive sign was attributed to the score when the Nuc/Cyt p65 ratio was lower or higher than the GL3 ratio, respectively. As expected, Nod1 and RIPK2, two key proteins involved in S. flexneri-induced NF-kB activation [16,29], as well as Src, a tyrosine kinase required for bacterial entry into cells [30], obtained strong negative scores (Figures 1E and Table S1). Using an arbitrarily determined cut off score value of +/2 3.5, the proteins CHUK/IKKa, FGD3, Arf1, GIT1, Rab23, ARHN, RabGGTA, HIPK2 and CLTC were classified as hits ( Figures 1E and Table S1). The identification of IKKa, also known as CHUK, was unexpected as peptidoglycan recognition via Nod1 triggers the canonical NF-kB pathway, and was, therefore, thought to be exclusively dependent on IKKb and NEMO. The contribution of IKKa in the mechanisms that control NF-kB activation following S. flexneri infection was then further explored. Depletion of IKKa inhibits S. flexneri-induced p65 nuclear translocation and IL-8 expression without affecting bacterial invasion First, to validate the role of IKKa in S. flexneri-induced p65 nuclear translocation, we tested a pool of four synthetic IKKa siRNAs that had no overlapping sequences with the in vitro diced siRNAs used in the screen. In conditions where around 90% of endogenous protein was depleted (Figure 2A), a massive reduction of p65 nuclear translocation was observed (Figures 2B and 2C). This result was not explained by a reduction of bacterial uptake as the number of internalized bacteria was similar in control and in IKKa-depleted cells ( Figure 2B). Taken together, these results indicated that IKKa is required for the activation of NF-kB during infection of epithelial cells by S. flexneri.
During infection, NF-kB positively regulates the expression of multiple proinflammatory genes [31]. In particular, it induces the expression of the chemokine IL-8, which recruits PMNs on site of infection, and thereby limits the spread of bacterial invasion within the intestinal tissue [18]. To test whether IKKa contributed to the upregulation of IL-8 expression during infection, we measured by ELISA the secretion of IL-8 in the supernatant of control and IKKa-depleted HeLa cells six hours post infection. To increase the amount of IL-8 produced in response to infection, cells were infected with a mutant of S. flexneri deleted for the type III effector OspF (DospF) that dampens inflammation signaling by dephosphorylating p38 in the nucleus of infected cells [20]. Consistent with NF-kB data, a reduction of IL-8 secretion was observed in response to S. flexneri DospF infection when cells were depleted for IKKa ( Figure 2D), showing that IKKa is involved in the signaling pathways that control the expression of a critical inflammatory chemokine during bacterial infection.

Both IKKa and IKKb contribute to the phosphorylation and the degradation of IkBa during S. flexneri infection
It is generally believed that IKKb and NEMO are the two subunits of the IKK complex involved in the canonical pathway of NF-kB activation. To test the contribution of all IKK subunits in the activation of NF-kB during infection of epithelial cells by S. flexneri, we monitored the localization of p65 in cells depleted for IKKa, IKKb or NEMO. As shown in Figure 3A, the depletion of either of these proteins reduced the nuclear translocation of p65, indicating that both catalytic subunits IKKa and IKKb, as well as the scaffolding function of NEMO were required to fully activate the NF-kB pathway during infection. IKKb regulates the canonical pathway of NF-kB activation by phosphorylating IkBa at positions serine 32 and 36, thereby inducing its polyubiquitination and subsequent degradation [9]. To test whether IKKa regulated NF-kB via a similar mechanism, we monitored the phosphorylation of IkBa at serines 32 and 36 (pIkBa) and its degradation following infection of HeLa cells by S. flexneri. Whereas massive phosphorylation and degradation of IkBa were observed in control cells following infection, these two processes were strongly reduced after IKKa knockdown ( Figure 3B), indicating that IKKa largely contributes to the phosphorylation and degradation of IkBa during infection by S. flexneri.

IKKa is involved in Nod1-mediated peptidoglycan recognition
During S. flexneri infection, the activation of NF-kB is initiated by the recognition of peptidoglycan fragments via the intracellular receptor Nod1 [16]. To specifically examine the implication of IKKa in Nod1-mediated signaling, the effect of IKKa depletion was directly tested in cells exposed to the Nod1 ligand Tri-DAP. Whereas Tri-DAP treatment induced a clear nuclear translocation of p65 in control cells, this translocation was severely impaired in IKKa-depleted cells ( Figures 4A and 4B), demonstrating that IKKa is required for the activation of NF-kB following Nod1mediated peptidoglycan recognition.

Discussion
Although NF-kB is a transcription factor that has been the subject of intensive research by academic laboratories and the pharmaceutical industry, the complex molecular mechanisms controlling its activation are only partially elucidated. Here we performed an RNAi screen to identify new proteins involved in the mechanisms that control NF-kB activation following pathogen recognition using infection of epithelial cells by the invasive bacterium S. flexneri as a model system. The presence of S. flexneri in the cytoplasm is recognized via the detection of peptidoglycan by the receptor Nod1 [16]. This recognition leads to NF-kB activation and the upregulation of proinflammatory genes that orchestrate the host inflammatory response [31]. The protein IKKa was identified in the screen as a protein required for the activation of NF-kB in response to S. flexneri infection. Since bacterial uptake was not affected by the depletion of IKKa, this result indicated that IKKa is involved in the activation of NF-kB in response to infection. The depletion of IKKb or NEMO also induced a reduction of S. flexneri-induced-p65 nuclear translocation, confirming that all subunits were required for the full activation of NF-kB during infection. As the translocation of p65 depends on the degradation of IkBa primed by its phosphorylation on serines 32 and 36, the role of IKKa in these two processes was analyzed. A reduction of IkBa phosphorylation and delayed degradation were observed in cells depleted for IKKa, indicating that similar to IKKb, IKKa triggers NF-kB activation by inducing the phosphorylation and the degradation of IkBa. Taken together, these results provided new evidence for a role of IKKa in the canonical pathway of NF-kB activation, and were in line with few reports indicating that the dominant model of canonical NF-kB activation based on IKKb and NEMO is incomplete. In particular, it has been reported that IKKa is the key subunit responsible for the Receptor activator of NF-kB (RANK)-induced classical NF-kB activation in mammary epithelial cells [32]. In addition, Solt et al. showed that IL-1-induced NF-kB activation requires the interaction of IKKa with NEMO and occurs in the absence of IKKb [11]. Finally, it has been shown recently that both IKKa and IKKb contribute to IkBa phosphorylation and NF-kB activation in response to TNFa stimulation in HeLa cells [10]. For the first time, our results demonstrate that IKKa is also implicated in the canonical pathway of NF-kB activation triggered by bacterial infection. To further demonstrate that pathogen recognition induced NF-kB activation in an IKKa-dependent manner during S. flexneri infection, we directly tested whether IKKa was involved in the activation of NF-kB induced by Nod1-mediated peptidoglycan recognition. For this purpose, controls or IKKa-depleted cells were directly stimulated with the purified Nod1 ligand Tri-DAP. The analysis of the p65 nuclear translocation showed that IKKa was required for the full activation of NF-kB in response to Tri-DAP treatment, indicating that IKKa is involved in the molecular mechanism signaling the recognition of peptidoglycan-derived peptides by Nod1.
Following infection of epithelial cells by S. flexneri, the activation of NF-kB leads to the upregulation of genes encoding for inflammatory cytokines including IL-8 and TNFa [31]. To confirm the implication of IKKa in the expression of genes induced by the canonical NF-kB pathway, the deletion of IKKa was tested on S. flexneri-induced IL-8 expression. Consistent with NF-kB data, the secretion of IL-8 was reduced when IKKa was depleted, showing that the contribution of IKKa to NF-kB activation has a functional impact on the amplitude of the inflammatory response mounted in response to bacterial infection. These results suggest that inhibition of IKKa activity may be critical to control inflammation upon bacterial infection. Inhibitors of IKKb have demonstrated therapeutic benefits in various animal models of inflammatory diseases and are currently in early clinical trials [33,34,35]. The data presented in this manuscript suggest that IKKa inhibitors should also be developed and used in combination with IKKb inhibitors to limit inflammation during bacterial infection or in inflammatory disorders that may involve Nod1 signaling, including asthma, eczema and inflammatory bowel diseases [36,37,38].

Supporting Information
Table S1 Results of the screen on S. flexneri-induced p65 nuclear translocation. Scores are fold standard deviations from the mean of GL3 control p65 ratios. Gene names are based on the NCBI nomenclature. (PDF)