A Novel Role for IκB Kinase (IKK) α and IKKβ in ERK-Dependent Up-Regulation of MUC5AC Mucin Transcription by Streptococcus pneumoniae¹

In the innate immune system, epithelial cells represent the first line of host defense against invading microbes at mammalian mucosal surfaces (1). Epithelial cells are not passive barriers, rather, they elaborate diverse molecules involved in efficient pathogen clearance. Mucins, the major constituent of mucus secretions, facilitate the mucociliary clearance by physically trapping inhaled microbial pathogens in the upper respiratory tract (2, 3). Thus, up-regulation of mucins in infectious disease represents an important host innate defense response against invading microbes (3). Excess mucin production, however, overwhelms the mucociliary escalator, contributing to airway obstruction. Similarly, prolonged elevation of mucus production in the middle ear restricts movement of the ossicles, contributing to hearing impairment during otitis media (4, 5, 6, 7). Indeed, mucin concentration in middle ear effusions correlates positively with hearing loss (8). Tight regulation of mucin production is thus critical for host mucosal defense, and yet how mucin is up-regulated in upper respiratory infectious diseases, such as those involving Streptococcus pneumoniae, remain largely unknown.

S. pneumoniae, a Gram-positive bacterium, is an important human pathogen that colonizes the upper respiratory tract (9). It causes potentially life-threatening diseases such as pneumonia, septicemia, and meningitis as well as otitis media, the most common childhood bacterial infection and the leading cause of conductive hearing loss (10). Using animal models of infection, numerous S. pneumoniae virulence factors have been identified (11, 12). Among them, a 53-kDa cytoplasmic protein, pneumolysin, is a key virulence factor produced by virtually all clinical isolates and released during respiratory infections in humans by bacterial autolysis (13, 14, 15). Pneumolysin is classically defined as a cytolytic toxin (16). At low sublytic concentrations, however, pneumolysin also serves as a potent inducer of host defense responses by, for example, activation of host complement and induction of immune cell-mediated inflammation (17, 18). The mechanisms by which S. pneumoniae induces these responses and, in particular, the receptor-mediated signaling pathways involved in mucin up-regulation, are unknown.

TLRs are integral to mediating effective host defense responses against invading pathogens (19, 20). To date, 11 members of the human TLR family have been identified. TLR-mediated MyD88-dependent signaling pathways activate NF-κB and MAPK pathways (20). A canonical NF-κB pathway is commonly activated by proinflammatory cytokines as well as by pathogen-associated molecular patterns (21). These inducers interact with the TNFR and TLR-IL-1R superfamilies, causing activation of the IκB kinase (IKK)⁴ complex. The most common form of this complex is comprised of IKKα and IKKβ catalytic subunits plus the regulatory subunit IKKγ (also known as NF-κB essential modulator). Following IKKγ dissociation, IKKβ drives the phosphorylation, polyubiquitination, and subsequent degradation of IκBα. Released NF-κB dimers (most commonly the p50–p65 heterodimer) translocate to the nucleus, bind DNA, and activate gene transcription. A noncanonical NF-κB pathway was recently discovered (21, 22), typically induced by lymphotoxin β receptor, B cell activating factor of the TNF family, and CD40L. This alternative pathway is strictly dependent on the formation of IKKα homodimers that in turn induce posttranslational processing of p100 to the DNA-binding subunit p52, ultimately leading to nuclear translocation of p52-RelB heterodimers and activating gene transcription.

In contrast to NF-κB, MAPK pathways are quite diverse. MAPKs are a superfamily of serine/threonine protein kinases widely conserved among eukaryotes. They transduce a variety of external signals, leading to an array of cellular responses that include growth, differentiation, apoptosis, and host defense response. To date, three major MAPK pathways have been identified in mammals: ERK, stress-activated protein kinase/JNK, and p38. Growth factor-induced ERK activation is relatively well-elucidated, but the signaling mechanisms underlying TLR-mediated activation of ERK remain to be determined.

In this study, we investigate the molecular mechanisms underlying S. pneumoniae-induced up-regulation of MUC5AC, a prominent mucin in respiratory secretions. We find that pneumolysin plays an important role in MUC5AC induction via TLR4-MyD88-IL-1R-associated kinase 1 (IRAK1)-TNFR-associated factor 6 (TRAF6)-dependent activation of ERK in vitro and in vivo. Furthermore, delayed ERK activation appears to be crucial in mediating MUC5AC induction. We also provide evidence for novel roles for IKKα and IKKβ in mediating MUC5AC induction via an ERK1-dependent, but IκBα-p65- and p100-p52-independent, mechanism. Our data thus reveal a novel IKK-mediated signaling pathway through which S. pneumoniae stimulates production of MUC5AC.

PD98059, SB203580, MG-132, and PMA were purchased from Calbiochem. Recombinant human epidermal growth factor (EGF) was purchased from R&D Systems. Polymyxin B was purchased from Sigma-Aldrich.

Clinical isolates of S. pneumoniae wild-type (WT) strains D39, R6, 6B, 3, 19F, and 23F were used in this study (23, 24). D39 isogenic pneumolysin-deficient mutant (Ply mt) was developed through insertion-duplication mutagenesis as described (13). S. pneumoniae was grown on chocolate agar plates and in Todd-Hewitt broth supplemented with 0.5% yeast extract (THY) at 37°C in a humidified 5% CO₂ water-jacketed incubator. The whole bacterial cells cultured in THY (Autolysate) were harvested at 10,000 × g for 20 min at 4°C to obtain supernatant and pellet after growth to early stationary phase. The bacterial culture supernatant was filtered with 0.2-μm pore size membrane to completely remove bacteria. The bacterial pellet was suspended in PBS for the preparation of live bacteria. The bacterial cell suspension was sonicated on ice three times at 150 W for 3 min at 5-min intervals. Residual intact cells were removed by centrifugation at 12,000 × g for 20 min at 4°C. The bacterial lysate was stored at −80°C and 5 μg/ml lysate was used in all experiments.

The 6 × His-tag fused pneumolysin was expressed in and purified from an Escherichia coli strain and residual LPS was removed by passage over End-X resin as previously described (25). The cytotoxicity of purified pneumolysin was quantified by lactate dehydrogenase (LDH) release assay using the CytoTox 96 Nonradioactive Cytotoxicity Assay kit (Promega).

All media described below except medium for primary human bronchial epithelial (NHBE) were supplemented with 10% FBS (Invitrogen Life Technologies), penicillin (100 U/ml), and streptomycin (0.1 mg/ml). HM3 (human colon epithelial) cells were maintained in DME H-21 (University of California Cell Culture Facility, San Francisco, CA), and mouse embryonic fibroblast (MEF) cells were maintained in DMEM (Invitrogen Life Technologies). WT, IKKα^−/−, and IKKβ^−/− MEFs were provided by Dr. I. M. Verma (The Salk Institute, La Jolla, CA) (26, 27). p65^−/− MEF was provided by Dr. C.-Y. Wang (University of Michigan, Ann Arbor, MI) (28). Tlr4^Lps-d and ERK1^−/− MEFs were isolated from Tlr4^Lps-d mutant and ERK1^−/− mice, respectively. HeLa (human cervix epithelial) cells were maintained in MEM (American Type Culture Collection). HMEEC-1 (human middle ear epithelial) cells were previously described (29) and maintained in bronchial epithelial basal medium (Cambrex). NHBE cells were purchased from Cambrex and maintained in bronchial epithelial growth medium (Cambrex) supplemented with defined growth factors and retinoic acid as per Clonetics’ instructions. Only NHBE cells at passages 4 were used for experiments. All cells were cultured at 37°C in a humidified 5% CO₂ water-jacketed incubator. For air-liquid interface culture, NHBE cells were cultured as previously described (30). Briefly, NHBE cells were seeded at 2 × 10⁴ cells/cm² onto 24-mm diameter, 0.4-μm pore size, semipermeable membrane inserts (Transwell Permeable Supports; Corning) in bronchial epithelial growth medium. The cultures were grown submerged for the first 7 days; an air-liquid interface was then created by removing media from the apical compartment of the cultures. Culture media were changed every other day until creation of the air-liquid interface, then changed daily by replacing fresh media only to the basal compartment. NHBE cells were grown in air-liquid interface for 2 wk before use in experiments.

The expression plasmids TLR2 dominant negative (DN), TLR4 DN, MyD88 DN, IRAK1 DN, TRAF6 DN, IκBα DN, ERK2 DN, IKKα DN, WT IKKα, IKKβ DN, and WT IKKβ have been previously described (31, 32, 33, 34, 35, 36, 37). The expression plasmid ERK1 DN was provided by Dr. Cobb (University of Texas Southwestern Medical Center, Dallas, TX). The construct containing the 5′-flanking region of human MUC5AC gene was previously described (38). A deletion mutant of the promoter region was constructed by restriction enzyme digestion and PCR amplification. Generated DNA fragments were ligated into a luciferase reporter gene (pGL3-Basic vector; Promega). A MUC5AC thymidine kinase (TK) luciferase construct (300TK) was obtained by subcloning the MUC5AC promoter (base pair −3752/−3452) to upstream of the TK-32 promoter. An NF-κB mutant TK construct was created by replacing the NF-κB site (−3612/−3600) with an EcoRI site. All constructs were confirmed by DNA sequencing. Transient transfections were conducted using TransIT-LT1 reagent (Mirus) following the manufacturer’s instructions. In cotransfections, empty vector was used as a control. Transfected cells were pretreated with chemical inhibitors for 1 h, followed by S. pneumoniae or pneumolysin treatment for 7 h before lysis for luciferase assay. The luciferase assay was conducted in triplicate and luciferase activity was normalized with respect to β-galactosidase activity.

RNA-mediated interference for down-regulating ERK1 and IKKα expressions was done by the transfection of siRNA-ERK1 and siRNA-IKKα as previously described (39). siRNA-ERK1 (human MAPK3), siRNA-IKKα (human CHUK), and an siCONTROL Nontargeting siRNA Pool were purchased from Dharmacon. Cells, 40∼50% confluent HeLa, were transfected with a final concentration of 100 nM siRNA using Lipofectamine 2000 (Invitrogen Life Technologies). Forty hours after the start of transfection, cells were treated with S. pneumoniae for the indicated time before being lysed for Western blot analysis and real-time quantitative PCR.

Abs were used to analyze total cell lysates as per the manufacturers’ instructions. Phospho-ERK (Thr²⁰²/Tyr²⁰⁴), ERK, phospho-Ets-like transcription factor (ELK) 1 (Ser³⁸³), phospho-IKKα (Ser¹⁸⁰)/IKKβ (Ser¹⁸¹), IKKα, IKKβ, and GST Abs were purchased from Cell Signaling. Monoclonal anti-β-actin was purchased from Sigma-Aldrich. p100/p52 (N-terminal) Ab was provided by Dr. L.-F. Chen (University of Illinois at Urbana-Champaign, Urbana, IL).

MUC5AC protein released from cells was measured with slight modifications to the ELISA described previously (40). Briefly, serial dilutions of each sample were plated in duplicate on 96-well microtiter plates and dried overnight at 40°C. After washing, plates were blocked with 2% BSA overnight at 4°C. To detect MUC5AC, Ag-coated wells were incubated with a 1/100 dilution of anti-mucin 5AC primary Ab (45M1; Neo Markers) for 1 h at 37°C. After washing, the plates were incubated with a 1/10,000 dilution of HRP-conjugated anti-mouse IgG for 1 h. Plates were incubated with tetramethylbenzidine peroxidase substrate (Bio-Rad), and the reaction was stopped with 1 N H₂SO₄. Absorbance was measured at 450 nm. Bovine submaxillary gland mucin (type I; Sigma-Aldrich) was used as a standard. The data was expressed as a fold induction on the same experimental day due to various mucin productions with cell passage in HM3 cells.

Total RNA was isolated using TRIzol reagent following the instruction of Invitrogen Life Technologies. Predeveloped TaqMan Assay Reagents (Applied Biosystems) were used for Q-PCR. Synthesis of cDNA from total RNA was performed with MultiScribe Reverse Transcriptase. Primers and probes for human and mouse MUC5AC have been previously described (37) and synthesized by the Applied Biosystems Customer Oligo Synthesis Service (Applied Biosystems). Reactions were amplified and quantified using a 7500 Real-Time PCR System and the manufacturer’s software (Applied Biosystems). Relative quantities of MUC5AC mRNA were calculated using the comparative cycle threshold method and normalized by cyclophilin (human) and GAPDH (mouse) for the amount of RNA used in each reaction (Applied Biosystems).

Cells were harvested in cell lysis buffer following p44/42 MAPK Assay kit (Biolabs). Cell lysates were immunoprecipitated with immobilized phospho-ERK (Thr²⁰²/Tyr²⁰⁴) mAb (Cell Signaling), and total cell extracts were prepared with SDS-PAGE sample buffer. ERK activity was quantified in each immunoprecipitate using the Assay kit.

ERK1^−/− mice were previously described (41). Tlr4^Lps-d mice (homozygous mice for the point mutation (Pro⁷¹² to His) in the Tlr4 gene), BALB/c and C57BL/6 mice were purchased from The Jackson Laboratory. Sex- and age-matched background mice were used as WT controls. All animal experiments were approved by the Institutional Animal Care and Use Committee at the House Ear Institute and the University of Rochester.

In our mouse model of upper respiratory (middle ear) infections, the middle ear of anesthetized mice (WT, ERK1^−/−, and Tlr4^Lps-d) were inoculated transtympanically with live S. pneumoniae (6.25 × 10⁵ CFU), lysate (equivalent of 6.25 × 10⁵ CFU), or pneumolysin (200 ng). Saline was injected as a control. Bullae were dissected from mice 9 h after treatment, then subjected to MUC5AC mRNA expression analysis by Q-PCR. To examine the effect of systemic ERK inhibition on S. pneumoniae-induced MUC5AC mRNA expression, WT mice were pretreated with ERK inhibitor PD98059 (1 mg/kg, i.p.) or DMSO as vehicle control for 2 h before S. pneumoniae inoculation. Total RNA was extracted from whole murine bullae with TRIzol and Q-PCR was performed as described above.

Up-regulated mucin production represents an important innate defense response against infectious agents such as S. pneumoniae. Excess mucin production, however, may lead to detrimental airway obstruction or to conductive hearing loss in the middle ear. The molecular mechanism by which mucin is induced during S. pneumoniae infection remains unknown. Because MUC5AC has been identified as a prominent mucin in respiratory secretions and in middle ear effusions, we focused on elucidating MUC5AC up-regulation. To determine whether S. pneumoniae induces MUC5AC transcription, we first used Q-PCR to quantify MUC5AC mRNA expression in human epithelial cells following incubation with S. pneumoniae strains. As shown in Fig. 1,A, all clinical isolates tested were capable of inducing MUC5AC expression. Transcriptional regulation of MUC5AC was evaluated by transfecting epithelial cells with a MUC5AC promoter-driven luciferase reporter construct before treatment with S. pneumoniae. MUC5AC transcriptional activity was indeed increased upon exposure to all clinically isolated S. pneumoniae strains tested, indicating that S. pneumoniae induces MUC5AC expression at transcriptional levels (Fig. 1,B). These results suggest that the ability to induce mucin is well-conserved among clinical isolates. Consistent with MUC5AC mRNA induction, ELISA revealed increased MUC5AC protein production in response to S. pneumoniae (Fig. 1 C).

To evaluate the generalizability of these data, we assayed MUC5AC expression in a variety of epithelial cell lines including HeLa, HM3, and HMEEC-1 as well as primary bronchial epithelial NHBE cells. As shown in Fig. 1,D, S. pneumoniae induced MUC5AC mRNA expression in all human epithelial cells tested, indicating that MUC5AC induction by S. pneumoniae may be generalizable to most human epithelial cells. To confirm whether MUC5AC is also induced in vivo, we next determined the effects of S. pneumoniae on Muc5ac expression in mice. As shown in Fig. 1,E, clinical isolates of S. pneumoniae (strains D39 and 6B) strongly induced Muc5ac mRNA expression in murine upper respiratory tracts. We confirmed again these results in two mouse strains, BALB/c and C57BL/6 (Fig. 1 F). Collectively, these data demonstrate that S. pneumoniae induces mucin MUC5AC transcription.

To identify the bacterial factors responsible for MUC5AC induction, we first compared the mucin-inducing activity of live bacteria, bacterial culture supernatant, autolysate, and bacterial lysate containing soluble cytoplasmic components. As shown in Fig. 2,A, MUC5AC transcription was increased the most in response to S. pneumoniae lysate vs live bacteria and bacterial culture supernatant. Similar results were also observed in mice in vivo (Fig. 2,B). Interestingly, autolysate induced MUC5AC expression to a degree relatively comparable with lysate (Fig. 2 A). Therefore, we conclude that S. pneumoniae cytoplasmic components are likely the primary mediators of MUC5AC induction.

S. pneumoniae produces numerous virulence factors contributing to bacterial pathogenesis during infection (12). Of these, pneumolysin is a major cytoplasmic protein in inducing host immune responses (18, 42). To determine whether pneumolysin is responsible for MUC5AC induction, WT S. pneumoniae strain D39 (D39 wt) and its isogenic Ply mt were compared in their MUC5AC-inducing activity. As shown in Fig. 2,C, D39 wt potently induced MUC5AC transcription whereas Ply mt did not, indicating that pneumolysin is important in MUC5AC induction. Indeed, purified pneumolysin up-regulated MUC5AC transcription in a dose-dependent manner (Fig. 2,D). Because a 500 ng/ml dose of pneumolysin induced MUC5AC transcription less strongly than 100∼200 ng/ml, we examined pneumolysin cytotoxicity with an LDH release assay in HM3, HeLa, and NHBE epithelial cells. As shown in Fig. 2,E, pneumolysin cytotoxicity was dose-dependent and peaked at 500 ng/ml, which may account for the decrease in MUC5AC expression at this dose. To eliminate the possibility that the MUC5AC transcription was induced by LPS contaminated during purification from E. coli, we pretreated pneumolysin with polymyxin B, a well-characterized LPS inhibitor (26, 27, 28). As shown in Fig. 2,F, polymyxin B pretreatment did not significantly reduce pneumolysin-induced MUC5AC transcription, indicating that potential LPS contamination was not involved. Finally, we compared MUC5AC expression at mRNA levels induced by S. pneumonia and pneumolysin. As shown in Fig. 2,G, D39 wt and purified pneumolysin induced MUC5AC mRNA to similar levels in epithelial cells. Similar results were observed in mice in vivo (Fig. 2 H). Taken together, these data demonstrate that S. pneumoniae pneumolysin plays a major role in inducing mucin MUC5AC.

Because of the important roles for TLR2 and TLR4 in cellular responses to pneumococcal components (43, 44, 45), we investigated whether TLR2 or TLR4 is involved in up-regulation of MUC5AC by S. pneumoniae. As shown in Fig. 3,A, overexpressing DN TLR4, but not TLR2, inhibited MUC5AC mRNA expression, suggesting TLR4-dependent MUC5AC induction by S. pneumoniae. The known importance of MyD88-IRAK1-TRAF6 signaling downstream of TLR4 led us to examine the involvement of this pathway in MUC5AC induction. As shown in Fig. 3,B, overexpressing DN MyD88, IRAK1, or TRAF6 inhibited MUC5AC expression at mRNA levels. Similarly, pneumolysin-induced MUC5AC transcription was also abolished by overexpressing DN TLR4, MyD88, IRAK1, or TRAF6 (Fig. 3,C). These data implicate involvement of the TLR4-MyD88-IRAK1-TRAF6 signaling pathway in S. pneumoniae-induced MUC5AC up-regulation. To further confirm the requirement for TLR4, we next evaluated the effect of S. pneumoniae on MUC5AC transcription in WT and Tlr4^Lps-d MEF cells. Fig. 3,D shows that S. pneumoniae greatly up-regulated MUC5AC transcription in WT MEFs whereas MUC5AC induction was abolished in Tlr4^Lps-d MEFs, indicating that TLR4 is indeed required for MUC5AC induction by S. pneumoniae. Similarly, induction of Muc5ac mRNA by both D39 wt and purified pneumolysin was also abolished in Tlr4^Lps-d mutant mice (Fig. 3 E). We conclude from these data that TLR4-dependent signaling is required for MUC5AC induction by S. pneumoniae in vitro and in vivo.

Having demonstrated that the TLR4-dependent signaling is required in MUC5AC induction, still unclear is which downstream signaling pathway transduces signals to the nucleus to up-regulate MUC5AC transcription. Among the numerous host signaling pathways downstream of TLR4, MAPKs have been shown to play important roles in mediating a variety of cellular responses (46). We have previously shown that p38 signaling is required for nontypeable Haemophilus influenzae (NTHi)-induced MUC5AC transcription (47). Thus, we initially investigated whether S. pneumoniae-induced MUC5AC transcription is also mediated by MAPK signaling. As shown in Fig. 4,A, pretreatment with SB203580, a specific chemical inhibitor for p38, did not significantly reduce MUC5AC transcription whereas pretreatment with PD98059, a specific chemical inhibitor for ERK, did. Similar results were obtained using pneumolysin (Fig. 4,B). Importantly, in air-liquid interface cultures of NHBE cells, PD98059 inhibited pneumolysin-induced MUC5AC expression at mRNA levels (Fig. 4,C). In addition, D39 wt and purified pneumolysin potently induced phosphorylation of ERK as shown by Western blot analysis (Fig. 4 D). Together, these data suggest that, in contrast to Gram-negative NTHi, Gram-positive S. pneumoniae uses ERK but not p38 to induce MUC5AC transcription.

To confirm the requirement of ERK in S. pneumoniae-induced MUC5AC transcription in vitro, we used more specific approaches to inhibit ERK signaling. As shown in Fig. 4, E and F, overexpressing DN ERK1 abolished MUC5AC expression at transcriptional and mRNA levels. In comparison, overexpressing DN ERK2 reduced MUC5AC expression to a lesser extent, suggesting that ERK1 may be more important in mediating MUC5AC expression by S. pneumoniae. Indeed, ERK1 knockdown using siRNA-ERK1 also abolished MUC5AC expression (Fig. 4,G, upper panel). The efficiency of siRNA-ERK1 in reducing endogenous ERK1 protein was confirmed by Western blot analysis (Fig. 4,G, lower panel). We then compared MUC5AC expression in WT and ERK1^−/− MEFs. As shown in Fig. 4 H, S. pneumoniae markedly induced MUC5AC transcription in WT but not in ERK1^−/− MEFs.

The in vivo relevance of these data was confirmed by using PD98059 in mice. As shown in Fig. 4,I, Muc5ac expression was abolished by pretreatment with PD98059 in two WT mouse strains (BALB/c and C57BL/6). Moreover, D39 wt and purified pneumolysin greatly induced Muc5ac expression at mRNA levels in BALB/c mice but not in ERK1^−/− mice (Fig. 4,J). Finally, we confirmed that ERK signaling is downstream of pneumolysin-activated TLR4 by comparing WT and Tlr4^Lps-d MEFs by performing Western blot analysis. As shown in Fig. 4 K, phosphorylation of ERK by pneumolysin was reduced in Tlr4^Lps-d MEF cells compared with WT MEF cells. Taken together, our data indicate that TLR4-dependent ERK activation is required for MUC5AC induction by S. pneumoniae in vitro and in vivo.

Based on previous studies showing that the duration of ERK activation can be relevant in directing the outcome of ERK activation by various inducers (48), we evaluated the role of ERK activation in MUC5AC induction by S. pneumoniae. As shown in Fig. 5,A, phosphorylation of ERK increased significantly by 10-min posttreatment with S. pneumoniae, declined at 30 min, increased again at 60 min, and declined thereafter, outlining a bimodal pattern of ERK activation. Immunoprecipitation with anti-phospho-ERK Abs and in vitro kinase assays using ELK1-GST as ERK substrate revealed the same bimodal ERK activation. We next sought to determine which phase of ERK activation is responsible for maximal MUC5AC induction. Because EGF induces transient activation of ERK (49), we confirmed transient activation by Western blot analysis using various concentrations of EGF (Fig. 5,B). Then, we compared MUC5AC induction by S. pneumoniae with that induced by EGF at various concentrations (Fig. 5 C). Interestingly, S. pneumoniae potently induced MUC5AC transcription by ∼6-fold over the control, while 2.5 ng/ml EGF yielded only ∼2-fold induction. Higher concentrations of EGF (10∼50 ng/ml) induced even lower MUC5AC transcription compared with low-dose EGF (2.5 ng/ml). Thus, the “second wave” of ERK phosphorylation may play an important role in S. pneumoniae-induced MUC5AC expression.

Having shown that the TLR4-dependent ERK activation plays a critical role in MUC5AC induction by S. pneumoniae, how ERK is activated remains unclear. We previously reported that the NF-κB signaling cascade is involved in mucin induction by NTHi (36, 37). In addition, the NF-κB pathway represents one of the major signaling pathways downstream of TLR4 and involved in innate and acquired immune responses (50). We therefore investigated whether the NF-κB signaling cascade is involved in S. pneumoniae-induced MUC5AC transcription. First, we assessed the effects on MUC5AC induction of inhibiting IKKα and IKKβ, essential kinases upstream of NF-κB. As shown in Fig. 6A, overexpressing DN IKKα reduced MUC5AC induction by S. pneumoniae whereas, surprisingly, overexpressing DN IKKβ enhanced it. IKKα therefore appears to be a positive regulator of MUC5AC induction, while IKKβ acts as a negative regulator. Such distinct roles for IKKα and IKKβ were confirmed at mRNA levels as shown in Fig. 6,B. Consistent with these results, S. pneumoniae induced the phosphorylation of both IKKα and IKKβ in a time-dependent manner (Fig. 6 C).

We next confirmed whether MUC5AC induction requires IKKα using a more specific siRNA approach. As shown in Fig. 6,D (upper panel), IKKα knockdown using siRNA-IKKα greatly reduced S. pneumoniae-induced MUC5AC mRNA expression. The efficiency of siRNA-IKKα in reducing endogenous IKKα protein was confirmed by Western blot analysis (Fig. 6,D, lower panel). As further confirmation of these results, we assessed MUC5AC up-regulation in WT, IKKα^−/−, and IKKβ^−/− MEFs (Fig. 6,E, upper panel). The absence of IKKα and IKKβ proteins was again confirmed by Western blot analysis (Fig. 6 E, lower panel). In support of our findings to this point, MUC5AC induction was abolished in IKKα^−/− MEFs but markedly enhanced in IKKβ^−/− MEFs. Responsiveness to S. pneumoniae was restored by transfection of WT IKKα and IKKβ, respectively. Taken together, our data suggest that IKKα and IKKβ play distinct roles in mediating MUC5AC induction by S. pneumoniae.

We next sought to determine whether IKKα and IKKβ act downstream of pneumolysin-activated TLR4. As shown in Fig. 6 F, phosphorylation of IKKα and IKKβ by pneumolysin was reduced in Tlr4^Lps-d MEF cells compared with WT MEF cells. Thus, it is evident that IKKα and IKKβ activation are TLR4-dependent and play distinct roles in mediating S. pneumoniae-induced MUC5AC transcription.

Because IKKα and IKKβ are known to be distinctly involved in MUC5AC induction, we next investigated whether the NF-κB signaling cascade is involved in MUC5AC induction by S. pneumoniae. We first investigated whether an NF-κB binding site in the MUC5AC promoter is required for induction. Analysis of MUC5AC promoter-driven luciferase activities revealed a S. pneumoniae response element between base pairs −3752 and −3452 of the MUC5AC promoter (Fig. 7,A). This 300-bp region was subcloned upstream of a TK-32 promoter (300TK); similar results were observed with the full-length 3.7-kb and 300TK promoter constructs. Sequence analysis revealed an NF-κB binding site between base pairs −3612 and −3600. Selective mutagenesis was performed to replace the NF-κB binding site with an EcoRI site. Surprisingly, as shown in Fig. 7 A, 300TK (NF-κB mt) retained the ability to up-regulate MUC5AC promoter activation in response to S. pneumoniae treatment. This result indicates that NF-κB transcription factor may not be critically involved in mediating S. pneumoniae-induced MUC5AC expression.

IKKα- and IKKβ-dependent activation of NF-κB is mediated by two distinct, canonical and noncanonical signaling pathways. In the canonical pathway, IKKβ is both necessary and sufficient for phosphorylation of IκBα, while the role of IKKα is unclear (22). In contrast, the noncanonical pathway depends only on the IKKα homodimer, which is essential in phosphorylating p100 before ubiquitination and subsequent processing to p52 (22). To determine whether the IKKβ-mediated canonical NF-κB pathway is involved in MUC5AC induction, we compared S. pneumoniae-induced MUC5AC transcription in WT and p65^−/− MEFs; p65 (RelA) is a key subunit of NF-κB complex in the canonical pathway. As shown in Fig. 7,B (upper panel), S. pneumoniae-induced MUC5AC transcription was almost equal in both WT and p65^−/− MEFs, indicating that p65 is not required. The absence of p65 protein was confirmed by Western blot analysis (Fig. 7,B, lower panel). Consistent with this result, ERK phosphorylation is also roughly equal in WT and p65^−/− MEFs (Fig. 7,C). Moreover, S. pneumoniae-induced ERK phosphorylation was not affected by pretreatment with MG-132, a chemical inhibitor known to prevent IκBα degradation and subsequent inhibition of NF-κB activation (51) (Fig. 7,D). Finally, overexpressing DN IκBα did not alter MUC5AC induction (Fig. 7 E). Together, our data indicate that the IκBα-p65-dependent NF-κB pathway is not required for S. pneumoniae-induced MUC5AC transcription.

We next evaluated whether the noncanonical NF-κB pathway is involved in S. pneumoniae-induced MUC5AC transcription. As shown in Fig. 7 F, S. pneumoniae did not induce posttranslational processing of p100 to the DNA-binding subunit p52, a mandatory event in noncanonical NF-κB signaling. Thus, S. pneumoniae- induced MUC5AC transcription appears to be independent of both the IκBα-p65-dependent canonical and p100-p52-dependent noncanonical NF-κB pathways. IKKs therefore appear to play a novel role in mediating induction of MUC5AC expression by S. pneumoniae.

We have demonstrated that delayed ERK activation and IKKα signaling are positively involved in MUC5AC induction by S. pneumoniae in an IκBα-p65- and p100-p52-independent manner. We further hypothesized that IKKα acts upstream of ERK in mediating MUC5AC induction by S. pneumoniae. To test our hypothesis, we assessed the effect of IKKα knockdown using IKKα-siRNA on S. pneumoniae-induced ERK activation. Efficient reduction of endogenous IKKα protein was confirmed by Western blot analysis (Fig. 8,A). RNA-mediated interference for down-regulating IKKα greatly reduced the second wave of ERK activation, especially ERK1, at 60 min, leaving transient activation of ERK at 10 min relatively unchanged, indicating that IKKα is involved in delayed ERK activation by S. pneumoniae. We confirmed the involvement of IKKα as well as IKKβ in ERK activation by comparing ERK phosphorylation in WT, IKKα^−/−, and IKKβ^−/− MEFs. As shown in Fig. 8,B, phosphorylation of ERK was markedly reduced at 60 min in IKKα^−/− MEFs as compared with that at 10 min, whereas ERK activation was slightly enhanced in IKKβ^−/− MEFs at both 10 and 60 min compared with that in WT MEFs, confirming that IKKα is indeed an upstream activator for delayed ERK activation; in contrast, IKKβ may be somehow negatively involved in mediating ERK activation. Furthermore, immunoprecipitation with anti-phospho-ERK Abs and in vitro kinase assay using ELK1-GST as the ERK substrate revealed a significant reduction in ERK kinase activity at 60 min in IKKα^−/− MEFs and increased kinase activity of ERK in IKKβ^−/− MEFs (Fig. 8,C). Using PMA, a specific inducer of MUC5AC via ERK in human epithelial cells, we next examined whether IKKα mediates delayed ERK activation generally for other ERK inducers or specifically for S. pneumoniae. As shown in Fig. 8 D, PMA-induced ERK activation remained unaffected at both 10 and 60 min in WT and IKKα^−/− MEFs. We therefore conclude that S. pneumoniae induces delayed ERK activity via a unique pathway positively regulated by IKKα and negatively regulated by IKKβ.

We finally confirmed our conclusions by assessing ERK-mediated MUC5AC induction in the setting either of IKKβ inhibition or IKKα activation in ERK1^−/− and WT MEFs treated with the ERK inhibitor PD98059. If ERK1 signaling is already abrogated either by using PD98059 or using ERK1-deficient MEFs, altering IKKβ or IKKα signaling should not further enhance MUC5AC induction by S. pneumoniae. Indeed, as shown in Fig. 8 E, overexpressing DN IKKβ or WT IKKα did not enhance S. pneumoniae-induced MUC5AC transcription in WT MEFs pretreated with PD98059 compared with WT MEFs not pretreated with PD98059. Likewise, overexpressing DN IKKβ or WT IKKα did not further enhance S. pneumoniae-induced MUC5AC transcription in ERK1^−/− MEFs. Taken together, our data demonstrate distinct roles for IKKα and IKKβ in mediating S. pneumoniae-induced MUC5AC expression by acting upstream of ERK1, thus revealing novel roles for IKKs in mediating up-regulation of MUC5AC mucin via an IκBα-p65- and p100-p52-independent mechanism.

Here, we show that pneumolysin, a key cytoplasmic virulence protein well-conserved among all clinical isolates of S. pneumoniae, is crucial for induction of MUC5AC, a prominent mucin in respiratory secretions. We further found that S. pneumoniae induces MUC5AC via TLR4-dependent activation of ERK both in vitro and in vivo. Interestingly, delayed ERK activation appears to play an important role in mediating MUC5AC induction by S. pneumoniae. Our studies also provide evidence of distinct roles for IKKα and IKKβ via an ERK-dependent but IκBα-p65- and p100-p52-independent mechanism, thus revealing novel roles for IKKs in mediating up-regulation of MUC5AC mucin by S. pneumoniae (Fig. 9).

The finding that pneumolysin induces MUC5AC via TLR4 is consistent with a previous report that TLR4 recognizes pneumolysin to stimulate TNF-α and IL-6 production (43). However, recent studies by van Rossum et al. (44) suggest that TLR2 is essential for clearance of S. pneumoniae. It is likely that TLR2 is the main pattern recognition receptor for live S. pneumoniae, while pneumolysin is predominantly recognized by TLR4. This model is supported by the recent work of Schmeck et al. (45) who observed that live S. pneumoniae-induced NF-κB activity is exclusively dependent on TLR2 whereas pneumolysin-induced NF-κB activity is exclusively dependent on TLR4. Their results are congruent with our finding that pneumolysin-induced activation of IKK-ERK signaling is mainly TLR4-dependent. In contrast, van Rossum et al. (44) showed that pneumolysin promotes clearance of colonized S. pneumoniae in a TLR4-independent manner. The pore-forming effect of pneumolysin activates p38, contributing to the production of proinflammatory cytokines independent of both TLR2 and TLR4 (52). It is logical that pneumolysin-activated p38 leads to increased bacterial clearance by inducing proinflammatory responses independent of TLR.

A novel finding in the present study is the distinct roles of IKKα and IKKβ in mediating MUC5AC induction by S. pneumoniae. As described, these IKKs appear to act upstream of ERK, independent of both the IκBα-p65-mediated canonical and the p100-p52-mediated noncanonical NF-κB pathways. This unexpected finding seems to reveal novel functions for IKKα and IKKβ in regulating ERK-mediated biological responses during pneumococcal infection. It should be noted that no direct physical interaction was observed between ERK and IKKα or IKKβ when we performed coimmunoprecipitation experiments (data not shown), implicating the involvement of an as-yet unidentified signaling intermediate. Recently, two separate studies have concluded that IKKβ mediates LPS- or TNF-α-induced ERK activation via tumor progression locus 2 (Tpl2) (53, 54). However, by using Tpl2 siRNA knockdown, we found that Tpl2 is not involved in mediating MUC5AC induction or ERK activation by S. pneumoniae (data not shown). Moreover, their studies described IKKβ as a positive regulator of ERK via Tpl2 whereas we found that IKKβ acts as a negative regulator for ERK activation by S. pneumoniae. Thus, our results suggest novel roles for IKKα and IKKβ in mediating ERK-dependent MUC5AC mucin up-regulation by S. pneumoniae.

Differences in the duration and timing of ERK activation may account for the very different outcomes observed with various ERK inducers (55). EGF transiently activates ERK (56, 57), but induces MUC5AC transcription to a lesser extent compared with S. pneumoniae. To our surprise, S. pneumoniae induced bimodal activation of ERK. Delayed ERK activation is evidently important in mediating MUC5AC induction, and IKKα mediates this second wave induced by S. pneumoniae. Why does treatment with S. pneumoniae and EGF lead to two different patterns of ERK activation? The direct means by which IKKα induces ERK activation also remain to be determined. In addition, we cannot rule out the possibility that other pathways essential for MUC5AC expression, in addition to the ERK pathway described here, are activated by pneumolysin and not by EGF.

Pneumolysin is a multifunctional protein possessing both pore-forming cytolytic and immunomodulatory properties (58). Pneumolysin is thought to be present at relatively low concentrations early in infection, when it mediates both immunomodulatory effects and low sublytic activities (17, 18). Later in infection, pneumolysin rises to lytic concentrations at which it mainly forms membrane pores that cause direct cellular and tissue damage, promoting widespread pneumococcal dissemination (16). In this study, we found that 100∼200 ng/ml doses of pneumolysin induce high MUC5AC expression via TLR4-dependent activation of ERK, though this concentration causes ∼5∼25% cytotoxicity as measured by LDH release assay. In contrast, 500 ng/ml pneumolysin induces pronounced cytotoxicity (60% by LDH assay) but significantly lower MUC5AC expression. Recently, Ratner et al. (52) demonstrated that pneumolysin increases cell membrane permeability by forming membrane pores, and the induced osmotic stress activates p38 independent of both TLR2 and TLR4. Implied therein is that activation of p38 is cytotoxicity-mediated and independent of TLR signaling. We also observed p38 activation by 200 ng/ml pneumolysin (data not shown), but p38 is apparently not involved in MUC5AC induction as shown in Fig. 4 B. MAPKs are important to both cell survival and apoptosis. Although p38 activation is proapoptotic, activation of ERK can protect the cell. The dynamic balance between these opposing pathways is crucial to determining whether a cell will survive or undergo apoptosis (59). Therefore, it is likely that low sublytic concentrations of pneumolysin are sufficient to activate ERK, which predominates at these concentrations for survival and MUC5AC induction via TLR4. At higher lytic concentrations, p38 activation predominates in proapoptotic responses, and interference with ERK activation by cell damage may account for lower MUC5AC induction.

S. pneumoniae is capable of adhering to mucosal cells in the upper respiratory tract, but pneumococcal carriage is initially asymptomatic. During the symptom-free period of colonization, pneumococci interact with host epithelial cells and activate intracellular signaling pathways. In response, epithelial cells produce mucus, which facilitates bacterial clearance from airways via the mucociliary escalator. However, S. pneumoniae also develops disease by provoking inflammatory responses, invading adjacent tissues and sometimes the bloodstream. Proinflammatory cytokine release increases rapidly in response to bacterial cell wall components released by pneumococcal autolysis (60), which may also release relatively low concentrations of pneumolysin early in infection. Released pneumolysin could interact with TLR4 to activate ERK signaling pathways and rapidly increase MUC5AC expression as shown in this study, both in human epithelial cells in vitro and in mice in vivo. Increased mucus production can lead to airway obstruction which, in the setting of chronic inflammation, incapacitates the mucociliary escalator and impairs bacterial clearance. When bacterial load rises later in infection, increased bacterial autolysis releases lytic concentrations of pneumolysin, directly and significantly damaging the respiratory epithelium and facilitating dissemination of pneumococci from alveoli into the bloodstream (61, 62, 63).

In the present study, pneumolysin was identified as a key virulence factor in S. pneumoniae-induced MUC5AC expression via TLR4-dependent activation of ERK. These findings were consistent in our in vitro and in vivo models, suggesting that the same mechanisms likely act in vitro and in vivo. Furthermore, IKKα and IKKβ appear crucial in mediating MUC5AC induction via delayed ERK activation. In the context of previous reports, our findings outline a complex host signaling network responding to S. pneumoniae infection. The signaling molecules that directly mediate IKKα and IKKβ regulation of ERK activation are yet to be determined. Additional studies of how these pathways interact in vivo are also necessary to better characterize the integrated host response to S. pneumoniae infection.

We are grateful to Drs. D. Briles, J. R. Zhang, C. J. Kirschning, M. H. Cobb, C. Y. Wang, L. F. Chen, and I. M. Verma for kindly providing various reagents. We also appreciate the effort in professional editing of this manuscript by Erika Szymanski.

The authors have no financial conflict of interest.