Results

RPS3 phosphorylation in response to NF-κB activation
To test whether RPS3 is phosphorylated during NF-κB activation, we performed 32P-labeling experiments in tumor necrosis factor (TNF)-stimulated HEK 293T cells. While RPS3 was scarcely phosphorylated in unstimulated cells, we observed a marked increase in 32P-incorporation after TNF stimulation despite no increase in RPS3 protein (Fig. 1a). To determine which RPS3 residues were phosphorylated, we immunoprecipitated RPS3 from either resting or stimulated cells and performed immunoblotting with phosphorylation-specific antibodies. Both TNF and phorbol myristate acetate/ionomycin (PMA+I) stimulated rapid phosphorylation and degradaion of IκBα within 5 min which was accompanied by RPS3 phosphorylation on serine residues (Fig. 1b and data not shown), similar to the in vivo labeling. We did not detect tyrosine- or threonine-phosphorylation of RPS3 (Fig. 1b).

RPS3 and IKKβ interaction
The activation of the inhibitor of κB kinase (IKK), consisting of a regulatory subunit IKKγ and two catalytic subunits, IKKα and IKKβ, is critical for the phosphorylation and dispatch of the inhibitory IκBs and the liberation of NF-κB20-22. Given that RPS3 can be found in the cytoplasmic p65-p50-IκBα inhibitory complex in resting cells6, we hypothesized that activated IKKβ might also bind to and phosphorylate RPS3. First, we found that ectopically expressed IKKβ and RPS3 interacted (Fig. 1c). We next examined resting Jurkat cells and detected a modest endogenous IKKβ-RPS3 interaction (Fig. 1d), potentially accounting for the basal NF-κB transcription required for cell proliferation and survival. RPS3-IKKβ association was clearly augmented upon TNF stimulation, peaking at 10 min. (Fig. 1d), following similar kinetics to RPS3 serine phosphorylation (Fig. 1b). By contrast, there was no detectable interaction between RPS3 and IKKα (Fig. 1d).

IKKβ is required for RPS3 nuclear translocation
To examine whether the RPS3-IKKβ interaction is required for RPS3 nuclear translocation, we knocked down IKKα or IKKβ expression with siRNAs (Supplementary Fig. 1) and then observed stimulation-induced RPS3 nuclear migration by confocal microscopy. Both TNF and PMA+I triggered RPS3 nuclear translocation in Jurkat cells transfected with a scrambled nonspecific (NS) siRNA (Fig. 2a)6. RPS3 nuclear translocation was only slightly, if at all, impaired by IKKα-silencing. Conversely, knockdown of IKKβ attenuated 60-70% of RPS3 nuclear accumulation following stimulation (Fig. 2a). Immunoblotting of nuclear fractions confirmed that full expression of IKKβ, but not IKKα, was necessary for activation-induced RPS3 nuclear translocation (Fig. 2b). Control immunoblots revealed that p65 nuclear translocation was blocked under the same conditions (Fig. 2b).
We next examined the nuclear translocation of RPS3 in cells ectopically expressing either kinase-dead (SSAA) or constitutively-active (SSEE) mutant IKKβ proteins. As expected, the SSEE, but not SSAA, mutant of IKKβ induced NF-κB-dependent luciferase reporter activity (Fig. 2c, left). Whereas RPS3 remained cytosolic in IKKβ (SSAA)-expressing cells (Fig. 2c, right), a substantial proportion of RPS3 translocated to the nucleus in cells expressing IKKβ (SSEE) (Fig. 2c, right). The percentage of cells containing detectable nuclear RPS3 increased 5-fold in IKKβ (SSEE)-expressing cells, but not in IKKβ (SSAA)-expressing ones (Fig. 2d and Supplementary Fig. 2). Thus, IKKβ activity is necessary and sufficient for RPS3 nuclear translocation in response to NF-κB activating stimuli.

IκBα degradation and RPS3 nuclear translocation
Importin-α regulates the nuclear import of NF-κB Rel subunits23, 24. RPS3 harbors a nuclear localization signal (NLS) sequence and its nuclear translocation occurs in parallel to, but independently of, p65 translocation6. We envisioned that RPS3 could also utilize the importin-α/β pathway. Consistent with this notion, RPS3 association with importin-α, but not importin-β, was enhanced in TNF–stimulated cells (Fig. 3a). Therefore, we examined whether RPS3 binding to importin-α is essential for nuclear translocation during NF-κB activation.
Since IκBα degradation is a prerequisite to unmask the NLS of p65, and both RPS3 and IκBα bind to p65 in the cytoplasmic inhibitory complex, we tested whether IκBα degradation is required for the liberation of RPS3. We measured the association of RPS3 with importin-α in 293T cells overexpressing wild-type IκBα or an IκBα mutant (SSAA) resistant to IKKβ-induced phosphorylation and degradation. In cells transfected with wild-type IκBα, TNF stimulation augmented the interaction of RPS3 and importin-α to a similar degree as in non-transfected cells. By contrast, we observed that the RPS3-importin-α association was abolished by the presence of non-degradable IκBα (Fig. 3b).
To examine whether IκBα is the only cytoplasmic barrier precluding RPS3 nuclear translocation, we measured both RPS3-importin-α association and nuclear RPS3 after reducing IκBα expression. Compared with nonspecific siRNA, siRNA targeting of IκBα completely depleted IκBα in Jurkat cells (Fig. 3c, input). Nevertheless, the RPS3-importin-α association was not augmented (Fig. 3c), nor was significant nuclear RPS3 detected (Fig. 3d). Moreover, cells treated with sodium pervanadate (Pv) to induce IκBα degradation through an IKK-independent mechanism25-27 did not show increased association between RPS3 and importin-α (Fig. 3e and Supplementary Fig. 3b) or nuclear accumulation of RPS3, despite complete IκBα degradation (Supplementary Fig. 3c). We further examined whether a subsequent NF-κB activation signal independently promotes the importin-α association and nuclear transport of RPS3 after IκBα degradation. We found that TNF stimulation following Pv treatment was required for the RPS3-importin-α association, comparable to TNF stimulation alone (Fig. 3e). Thus, IκBα phosphorylation and degradation itself is required but not sufficient to cause RPS3 association with importin-α followed by nuclear translocation. Rather, an additional signal, potentially IKKβ phosphorylation of RPS3, is required.

IKKβ phosphorylates RPS3 at serine 209
Although originally defined as the kinase that phosphorylates IκB19, IKKβ also phosphorylates unrelated substrates including 14-3-3β and Bcl10, which lack the IKK consensus motif (DpSGYXpS/T)28. We therefore hypothesized that IKKβ could directly phosphorylate RPS3. By in vitro kinase assays using recombinant IKK and RPS3 proteins, we observed strong incorporation of 32P in autophosphorylatd IKKα and IKKβ (Fig. 4a, lanes 2-7) as well as phosporylated GST-IκBα (1-54) (Supplementary Fig. 4), but not the GST protein alone (Fig. 4a, lanes 3 and 6), when either IKKα or IKKβ was used. We discovered that GST-RPS3 could be phosphorylated by IKKβ, but not IKKα, in vitro (Fig. 4a, compare lanes 4 and 7).
To identify the RPS3 amino acid residue(s) phosphorylated by IKKβ, we performed liquid chromatography-tandem mass spectrometry analyses using in vitro phosphorylated RPS3. The results indicated that IKKβ phosphorylated S209, located in the RPS3 C-terminus (Fig. 4b). RPS3 amino acid sequence alignment revealed that S209 is conserved in many species throughout phylogeny with the exception of Caenorhabditis elegans and Schizosaccharomyces pombe, two organisms that do not possess the NF-κB signal pathway (Supplementary Fig. 5).
To verify biochemically that S209 is an IKKβ substrate, we performed 32P-labeling in vitro kinase assays with recombinant wild-type or S209A mutant RPS3 proteins. Compared with the wild-type protein, the S209A mutation reduced IKKβ–mediated RPS3 phosphorylation (Fig. 4c). There might be alternative phosphorylation site(s) under these conditions given modest residual phosphorylated RPS3 (Fig. 4c). RPS3 S209 does not fall within a conventional IKK recognition motif, but rather resides in a sequence motif (XXXpS/TXXE), potentially recognized by casein kinase II (CK2). Although IKKβ kinase can display a CK2-like phosphorylation specificity29, no CK2 protein was detectable in our recombinant IKK proteins (Supplementary Fig. 6). Thus, RPS3 S209 phosphorylation was due to the alternate specificity of the IKKβ kinase rather than any trace amount of CK2 bound to IKKs. To determine whether S209 is the critical site at which IKKβ phosphorylates RPS3 in living cells, we transfected the wild-type or S209A mutant Flag-RPS3 alone, or together with IKKβ into cells. Indeed, we observed that overexpressing IKKβ enhanced Flag-RPS3 phosphorylation, but phosphorylation was effectively eliminated by alanine substitution indicating that S209 is the predominant target site for IKKβ phosphorylation (Fig. 4d). We next generated a phospho-S209 RPS3 antibody and confirmed that endogenous RPS3 was phosphorylated at S209 in a time-dependent manner upon TNF stimulation (Fig. 4e). Thus, the RPS3 C-terminal tail potentially contains an important regulatory site.

Phosphorylation of RPS3 and its NF-κB function
We next examined whether S209 phosphorylation plays a role in the nuclear translocation of RPS3 during NF-κB activation. Subcellular fractions from either wild-type or S209A mutant RPS3-transfected cells were prepared and blotted for heat-shock protein 90 (hsp90), a cytoplasmic protein, and poly (ADP-ribose) polymerase (PARP), a nuclear protein, confirming a clean separation (Fig. 5a). As expected, PMA+I stimulation triggered wild-type Flag-RPS3 nuclear translocation (Fig. 5a). However, RPS3 (S209A) nuclear translocation was attenuated (Fig. 5a). We also tested the impact of activating NF-κB by overexpressing IKKβ on RPS3 nuclear translocation. IKKβ overexpression activated NF-κB measured by luciferase assays (Supplementary Fig. 7), and also induced the nuclear translocation of wild-type, but not S209A, RPS3 (Fig. 5b). These data suggest that S209 phosphorylation is critical for the NF-κB activation-induced RPS3 nuclear translocation.
To examine the role of S209 phosphorylation of RPS3 to its NF-κB function6, 7, 30, we silenced endogenous RPS3 expression using an siRNA that targets the 3′ untranslated region (3′ UTR) of RPS3 mRNA, followed by complementation with either wild-type or S209A mutant RPS3 via transfection. As expected, RPS3 siRNA severely reduced endogenous RPS3 abundance compared to NS siRNA, but did not affect the robust expression of Flag-tagged RPS3 from a transfected construct lacking the 3′ UTR (Fig. 5c). We also found that RPS3 knockdown reduced TNF-induced expression of an Ig κB-driven luciferase construct6 (Fig. 5d). The impaired luciferase signal caused by RPS3 deficiency was completely restored by transfecting wild-type, but not by S209A RPS3 (Fig. 5d), despite equivalent expression (Fig. 5c). Moreover, the failure of S209A RPS3 to restore luciferase activity did not result from defective translation because the transient overexpression of green fluorescent protein (GFP) was comparable in cells complemented with wild-type or S029A RPS3 (Supplementary Fig. 8). Taken together, these data suggest that RPS3 S209 phosphorylation is critical for NF-κB activity involving the canonical Ig κB site.
We next used chromatin immunoprecipitation to determine whether S209 phosphorylation affects RPS3 and p65 recruitment to specific κB sites in intact chromatin during NF-κB activation. In RPS3 knockdown cells, PMA+I stimulated the recruitment of ectopically expressed, Flag-tagged wild-type, but not S209A RPS3 to the κB sites of the NFKBIA and IL8 promoters (Fig. 5e). While expressing RPS3 S209A had no impact on p65 nuclear translocation, it substantially attenuated p65 recruitment (Fig. 5e). Additional experiments revealed that p65 attraction to RPS3-independent NF-κB target gene promoters such as CD25 was increased (Supplementary Fig. 9), consistent with our previous observations6. There was no significant Flag-RPS3 or p65 recruitment to ACTB promoter lacking κB sites (Fig. 5e), suggesting the recruitment was κB site-specific. Thus, the recruitment of RPS3 as well as the contingent recruitment of p65 to key promoters depended on S209.
Interleukin 8 (IL-8) secretion induced by either T cell receptor (TCR) agonist stimulation or PMA+I was decreased as a consequence of reduced RPS3/p65 recruitment to the IL8 κB sites in the presence of S209A mutant compared to wild-type RPS3 (Supplementary Fig. 10). However, cell surface CD25 expression was comparable between the wild-type and S209A RPS3 transfected cells (Supplementary Fig. 11). Therefore, RPS3 S209 phosphorylation by IKKβ is apparently required for RPS3 in directing NF-κB to a specific subset of target genes.

NleH1 inhibits RPS3 phosphorylation in vitro
EHEC pathogens are important causative agents of both foodborne disease and pediatric renal failure31. EHEC utilize T3SS to inject effector proteins directly into intestinal epithelial cells32, a subset of which inhibit NF-κB-dependent innate responses9, 33-38. The E. coli O157:H7 EDL933 effector protein NleH1 binds to and attenuates RPS3 nuclear translocation, thus impairing RPS3-dependent NF-κB signaling9. We therefore hypothesized that NleH1 may function by inhibiting RPS3 S209 phosphorylation. As expected, transfecting increasing amounts of NleH1-HA plasmid blocked TNFα-induced NF-κB activation in a dose-dependent manner (Fig. 6a-b)9. Remarkably, NleH1 reduced both TNF-induced, as well as basal RPS3 phosphorylation to roughly 20% of vehicle control (Fig. 6c). Expressing NleH1 does not interfere with either TNF-induced IKK activation or IκBα degradation, consistent with the lack of NleH1 impact on p65 nuclear translocation9 (Fig. 6c).
To determine if NleH1 inhibits RPS3 phosphorylation, we infected HeLa cells with E. coli O157:H7 strains possessing or lacking either nleH1 (ΔnleH1) or with a strain lacking a functional T3SS unable to inject NleH1 into mammalian cells (ΔescN). In uninfected cells, TNF-treatment stimulated a ∼7-fold increase in RPS3 S209 phosphorylation, peaking at 30 minutes (Fig. 6d). By contrast, RPS3 S209 phosphorylation was substantially impaired in cells infected with wild-type E. coli O157:H7 (Fig. 6d). However, TNF-induced RPS3 S209 phosphorylation was unimpaired in cells infected with either ΔnleH1 or ΔescN (Fig. 6d). We showed previously that wild-type, but not ΔnleH1 or ΔescN E. coli O157:H7 significantly attenuated TNF-induced RPS3 nuclear translocation9. The parallel between RPS3 phosphorylation and its nuclear translocation during E. coli infection provides evidence in the context of an NF-κB-dependent disease process that RPS3 S209 phosphorylation is important for nuclear translocation.
Our discovery that NleH1 inhibits RPS3 S209 phosphorylation suggested that it should also blocks RPS3-dependent NF-κB target gene transcription (e.g. IL8, NFKBIA, and TNFAIP3). Indeed, these genes were only modestly upregulated in cells infected with wild-type E. coli O157:H7, but significantly induced in cells infected with either ΔnleH1 or ΔescN strains (Fig. 6e). In contrast, deleting nleH1 had no impact on the expression of RPS3-independent genes, including CD25 and TNFSF13B (Supplementary Fig. 12). Together these results demonstrate that NleH1 specifically inhibits the protective immune response by directly blocking RPS3 S209 phosphorylation and thereby impairing critical RPS3-dependent NF-κB target genes.

NleH1 inhibits RPS3 S209 phosphorylation in vivo
We previously utilized a gnotobiotic piglet infection model to determine that piglets infected with ΔnleH1 mutant died more rapidly than those infected with wild-type E. coli O157:H7. Piglets infected with ΔnleH1 displayed clinical disease consistent with a robust inflammatory response, but with reduced bacterial colonization and little diarrhea9. While seemingly paradoxical, based on our cell culture data (Fig. 6d), we hypothesized that NleH1 blocked RPS3 S209 phosphorylation in vivo, thereby preventing RPS3 nuclear translocation in infected piglets. We isolated piglet colons at necropsy, and subjected them to immunohistochemistry using our phospho-RPS3 antibody. Consistent with in vitro data, piglets infected with wild-type E. coli O157:H7 exhibited diffuse and low intensity phospho-RPS3 staining, whereas in piglets infected with ΔnleH1 mutant, phospho-RPS3 expression was florid and intense (Fig. 6f). These data demonstrate that NleH1 inhibits RPS3 S209 phosphorylation both in vitro and in vivo, which might benefit the bacterium in colonization and transmission.

NleH1 steers the IKKβ substrate specificities
NleH1 is an autophosphorylated serine-threonine kinase, which depends on the lysine 159 (K159)9. To explore the mechanism by which NleH1 inhibits RPS3 S209 phosphorylation, we first performed an in vitro kinase assay with purified wild-type His-NleH1 protein and a mutant His-NleH1 (K159A) protein, confirming that NleH1 is autophosphorylated and the K159A is an NleH1 kinase-dead mutant (Fig. 7a). To examine whether the kinase activity is required for NleH1 to inhibit IKKβ phosphorlyation of RPS3 on S209, we ectopically expressing either wild-type or K159A NleH1 in 293T cells. Wild-type NleH1 expression significantly reduced TNF-induced RPS3 S209 phosphorylation, whereas the K159A mutant failed to do so (Fig. 7b). Thus NleH1 kinase activity is required to protect RPS3 from IKKβ-mediated phosphorylation.
Citrobacter rodentium is a mouse pathogen that shares pathogenic strategies with E. coli 39, most notably for our investigation, C. rodentium NleH inhibited RPS3 nuclear translocation and RPS3-dependent NF-κB luciferase activity to an extent equivalent to E. coli NleH1 (ref. 9). We assayed RPS3 S209 phosphorylation in HeLa cells infected with different C. rodentium strains. In uninfected cells, TNF-treatment stimulated a ∼3.5-fold increase in RPS3 S209 phosphorylation (Fig. 7c). Such augmentation of RPS3 phosphorylation was reduced by about 60% by wild-type C. rodentium infection (Fig. 7c). However, RPS3 phosphorylation was enhanced when cells were infected with a C. rodentium strain lacking NleH (Fig. 7c, ΔnleH). We further examined the role of NleH1 kinase activity using the C. rodentium ΔnleH strain as a background on which to express either wild-type or K159A E. coli NleH1. Complementing ΔnleH mutant with wild-type NleH1 almost abolished TNF-induced RPS3 S209 phosphorylation whereas complementing with K159A failed to inhibit RPS3 phosphorylation (Fig. 7c). Collectively, these results demonstrate that NleH1 kinase activity is required to block RPS3 S209 phosphorylation.
We next examined whether the inhibitory activity of NleH1 is sufficiently robust to impair the strong nuclear translocation of RPS3 trigged by the constitutively-active IKKβ (IKKβ [SSEE]) (Fig. 2d). We found that ectopically expressing either wild-type or SSEE IKKβ proteins, triggered more RPS3 nuclear translocation than the kinase-dead IKKβ (SSAA) protein (Fig. 7d). RPS3 nuclear accumulation was substantially retarded by infecting cells with wild-type E. coli O157:H7 (Fig. 7d). In contrast, infecting with either ΔnleH1 or ΔescN strains only slightly impaired RPS3 nuclear translocation in either IKKβ- or IKKβ (SSEE)-expressing cells (Fig. 7d). As expected, E. coli infections did not affect the RPS3 nuclear translocation in IKKβ (SSAA)-expressing cells, where NF-κB signaling was low (Fig. 7d). Thus, during infection NleH1 is sufficiently potent to inhibit RPS3 nuclear translocation even in cells expressing constitutively-activated IKKβ.
We examined whether NleH1 could directly phosphorylate IKKβ thus inhibiting IKKβ-mediated RPS3 S209 phosphorylation. We performed in vitro kinase assays using immunoprecipitated Flag-IKKβ (K44A) as substrate and recombinant His-NleH1 as kinase, so that IKKβ autophosphorylation would not obscure NleH1-induced phosphorylation. However, we did not observe any detectable 32P incorporation in IKKβ (Supplementary Fig. 13), thus ruling out this possibility.
We then tested the hypothesis that NleH1 could alter the IKKβ substrate specificities. To this end, we performed in vitro kinase assays using both CK2 and IKK substrates for IKKβ. As expected, IKKβ phosphorylated RPS3 (Fig. 7e, lane 7) and GST-IκBα (1-54) protein (Fig. 7e, lane 9), demonstrating it harbors either CK2 or IKK substrate specificity. Preincubation of IKKβ with NleH1 reduced IKKβ-mediated RPS3 phosphorylation, i.e. the CK2 kinase specificity, but not IKKβ-mediated GST-IκBα phosphorylation, i.e. the IKK kinase specificity (Fig. 7e). Control experiments revealed no NleH1-mediated phosphorylation or autophosphorylation of RPS3 or GST-IκBα (Fig. 7e). Taken together, NleH1 blocks the CK2 substrate specificity of IKKβ thus inhibiting the IKKβ-mediated RPS3 S209 phosphorylation thus representing a novel strategy by E. coli O157:H7 to alter the host innate immune response.