The type IV secretion system (T4SS) of Helicobacter pylori triggers massive inflammatory responses during gastric infection by mechanisms that are poorly understood. Here we provide evidence for a novel pathway by which the T4SS structural component, CagL, induces secretion of interleukin-8 (IL-8) independently of CagA translocation and peptidoglycan-sensing nucleotide-binding oligomerization domain 1 (NOD1) signalling. Recombinant CagL was sufficient to trigger IL-8 secretion, requiring activation of α5β1 integrin and the arginine–glycine–aspartate (RGD) motif in CagL. Mutation of the encoded RGD motif to arginine-glycine-alanine (RGA) in the cagL gene of H. pylori abrogated its ability to induce IL-8. Comparison of IL-8 induction between H. pylori ΔvirD4 strains bearing wild-type or mutant cagL indicates that CagL-dependent IL-8 induction can occur independently of CagA translocation. In line with this notion, exogenous CagL complemented H. pylori ΔcagL mutant in activating NF-κB and inducing IL-8 without restoring CagA translocation. The CagA translocation-independent, CagL-dependent IL-8induction involved host signalling via integrin α5β1, Src kinase, the mitogen-activated protein kinase (MAPK) pathway and NF-κB but was independent of NOD1. Our findings reveal a novel pathway whereby CagL, via interaction with host integrins, can trigger pro-inflammatory responses independently of CagA translocation or NOD1 signalling.
Helicobacter pylori is a Gram-negative bacterium that persistently infects the human stomach and is associated with chronic gastritis, peptic ulcer and gastric cancer. Marked IL-8 induction during gastric infection by H. pylori plays an important role in pathogenesis (Crabtree, 1996; Naumann and Crabtree, 2004). Neutrophils and macrophages are recruited to infected gastric tissues upon prolonged secretion of IL-8 during H. pylori infection (Crabtree, 1996), augmenting release of cytokines, growth factors and reactive oxygen species. This aggravated microenvironment in the gastric mucosa is highly conducive to chronic inflammation and tumorigenesis.
Helicobacter pylori strains that express a type IV secretion system (T4SS) trigger significantly more potent induction of IL-8 compared with T4SS-deficient strains (Censini et al., 1996; Audibert et al., 2001). T4SSs are macromolecular transporters found in many bacterial pathogens. The virulence-associated T4SS of H. pylori is encoded by 28–30 genes in the cag pathogenicity island (PAI) (Censini et al., 1996). H. pylori strains carrying the cag PAI genetic locus induce elevated pro-inflammatory responses and are associated with a higher cancer risk compared with cag PAI-negative strains (Crabtree et al., 1994; Blaser et al., 1995; Peek et al., 1995). The H. pylori cag PAI-encoded T4SS, like the prototypical T4SS of the plant pathogen Agrobacterium tumafeciens, consists of extracellular pilus subunits, membrane-spanning channel-forming subunits, energetic subunits and a coupling protein (Alvarez-Martinez and Christie, 2009). Only one protein substrate of the H. pylori T4SS has so far been identified, which is the 125 kDa protein, CagA (Hatakeyama, 2008; Backert et al., 2010). Upon translocation into the host cell, CagA can be phosphorylated by Src tyrosine kinase. Translocated CagA perturbs cell–cell junctions, enhances cell proliferation and alters pro-inflammatory responses (Hatakeyama, 2008; Backert et al., 2010).
Helicobacter pylori T4SS interacts with human gastric epithelial cells through binding to the transmembrane receptor α5β1 integrin (Kwok et al., 2007; Jimenez-Soto et al., 2009). This interaction is mediated by the surface-exposed T4SS component, CagL, via its arginine–glycine–aspartate (RGD) motif and is critical for both CagA translocation and CagA phosphorylation (Kwok et al., 2007). CagL also activates metalloprotease ADAM17 causing downregulation of gastric H,K-ATPase α subunit (Saha et al., 2010) and triggers cell spreading (Tegtmeyer et al., 2010). Sequence polymorphisms in cagL have been shown to be associated with gastric cancer risk (Yeh et al., 2011; Rizzato et al., 2012).
The precise mechanism by which the H. pylori T4SS apparatus is specifically recognized by the host immune system and triggers marked induction of IL-8 remains unknown. One notion, here termed the ‘substrate theory’, suggests that H. pylori T4SS substrates are responsible for activating the pro-inflammatory response. Favouring this theory is the observation that the cell wall component, peptidoglycan, can be delivered by H. pylori into infected gastric adenocarcinoma cells, AGS, in a T4SS-dependent manner, thereby activating the intracellular pathogen recognition receptor, peptidoglycan-sensing nucleotide-binding oligomerization domain 1 (NOD1), and activating NF-κB and IL-8 induction (Viala et al., 2004; Allison et al., 2009). Also, the T4SS substrate CagA enhances IL-8 induction after 8–12 h of infection of AGS (Brandt et al., 2005). Complementing the substrate theory is an alternative view, here referred to as the ‘direct activation theory’. Several lines of indirect evidence support this notion, which suggests that the T4SS apparatus per se elicits host pro-inflammatory responses independently of its substrates. First, H. pylori mutants deficient in the cag PAI-encoded genes cagβ (also known as virD4), cagG or cagI, although defective in T4SS translocation function, are still capable of stimulating a significant IL-8 response (Fischer et al., 2001; Selbach et al., 2002). Second, deletion of cagA has little effect on IL-8 induction until around 9 h post infection (hpi) (Censini et al., 1996; Fischer et al., 2001; Selbach et al., 2002). Third, pro-inflammatory chemokine secretion by NOD1-knockout murine gastric epithelial cells upon H. pylori infection was only moderately abrogated compared with wild-type cells (Viala et al., 2004). These observations all suggest that apart from CagA and peptidoglycan, additional T4SS-associated factors could contribute to IL-8 induction.
In this study, we have characterized a novel pathway by which the H. pylori T4SS triggers IL-8 induction. Our findings suggest that apart from IL-8 induction via translocated CagA and peptidoglycan, the H. pylori T4SS machinery can induce IL-8 via interaction of its surface-exposed constituent CagL with the host receptor integrin β1 and the subsequent activation of MAPKs and NF-κB.
The RGD motif of CagL and β1 integrin play significant roles in triggering IL-8 secretion
CagL is a T4SS-associated surface protein that binds integrin α5β1 via an arginine–glycine–aspartate (RGD) motif (Kwok et al., 2007). To test the hypothesis that CagL may induce IL-8 secretion upon binding to integrin α5β1, we compared the ability of CagL RGD motif variant strains to induce IL-8. The CagL variant strains were constructed by ‘knock-in’ chromosomal complementation, allowing the variant genes to be expressed under the control of the resident native promoter (Fig. S1). In contrast to P12ΔcagA and P12ΔvirD4 deletion mutants that lack CagA translocation but still induce T4SS-dependent IL-8 secretion, deletion of cagL completely abolished CagA translocation (Fig. S2A) and reduced IL-8 induction to levels obtained using P12ΔcagPAI (Fig. 1; P < 0.0001 versus P12, two-way anova Bonferroni post-test). Knock-in of P12ΔcagL with wild-type cagL (P12cagLWT) fully restored IL-8 secretion and CagA translocation, while substitution of the RGD motif with arginine-glycine-alanine (RGA) failed to restore CagA translocation (Fig. S2A) or IL-8 induction (Fig. 1; P < 0.0001 versus P12cagLWT, two-way anova Bonferroni post-test). In contrast, a more conserved substitution of the RGD motif with arginine–alanine–aspartate (RAD) restored CagA translocation (Fig. S2A) and partially restored IL-8 induction (Fig. 1; P < 0.0001 versus P12cagLWT and P12ΔcagL, two-way anova, Bonferroni post-test). These phenotypes of the P12cagLRGA and P12cagLRAD mutants were confirmed with multiple independent clones (Table S3).
Furthermore, the IL-8 level induced by P12cagLRGA was a minor fraction of that secreted in response to P12ΔcagA or P12ΔvirD4 (Fig. 1; both P < 0.0001 versus P12cagLRGA, two-way anova, Bonferroni post-test), suggesting that the RGD motif of CagL plays a more crucial role in T4SS-mediated IL-8 induction than simply facilitating CagA translocation. To further examine the role of CagL in IL-8 induction independently of its function in mediating T4SS effector translocation, we constructed the double mutant P12cagLRGAΔvirD4. As expected, P12cagLRGAΔvirD4, like P12cagLRGA and P12ΔvirD4, showed no detectable CagA translocation (Fig. S2B). The ability of P12cagLRGAΔvirD4 to induce IL-8 was, however, markedly reduced compared with that of P12cagLWTΔvirD4 or P12ΔvirD4 (Fig. 1). This significant difference in pro-inflammatory response to P12cagLWTΔvirD4 or P12ΔvirD4 versus P12cagLRGAΔvirD4 suggests that CagL and its RGD motif can contribute to IL-8 induction independently of their functions in mediating T4SS substrate translocation. Furthermore, the results indicate that while CagL was required for approximately 90% of the total IL-8 level induced by wild-type P12, the contribution of CagL to IL-8 induction in the absence of T4SS substrate translocation accounted for approximately 25% of the total IL-8 level induced by wild-type P12 under the conditions used. Also of interest is the observation that the IL-8 response to P12cagLRGA was slightly stronger than that in response to P12cagLRGAΔvirD4 (Fig. 1), suggesting that the former possesses some T4SS translocation activity, albeit below immunoblot detection, and that this residual level of T4SS activity was entirely abolished in the absence of VirD4.
Having ascertained that the RGD motif of CagL played an important role in IL-8 induction, we next examined the contribution of β1 integrin activation to H. pylori IL-8 induction. In agreement with previous findings (Hutton et al., 2010), β1 integrin activation-blocking antibody, AIIB2, significantly attenuated IL-8 secretion in response to H. pylori (Fig. S3). Importantly, our data indicate for the first time that β1 integrin contributes to both CagA-dependent and CagA-independent pro-inflammatory response because AIIB2 treatment significantly reduced IL-8 secretion in response to both P12 wt and P12ΔcagA strains. AIIB2 exerted similar effects on H. pylori P1 strains (Fig. S3), indicating that the contribution of β1 integrin to CagA-dependent and -independent IL-8 induction is not strain-specific.
Taken together, these findings suggest that CagL and β1 integrin-dependent signal transduction contribute majorly to H. pylori-induced IL-8 secretion in both CagA-dependent and CagA-independent manners.
Purified recombinant CagL induces IL-8 secretion in an α5β1 integrin-dependent manner
We then further examined the hypothesis that CagL can directly stimulate IL-8 secretion by AGS cells. Incubation of AGS cells with purified native CagL, but not heat-inactivated CagL (HI CagL), stimulated significant IL-8 secretion (Fig. 2A) in a dose-dependent manner (Fig. S4). In contrast, the CagLRGA mutant protein was significantly less potent compared with wild-type CagL (Fig. 2A), indicating that efficient IL-8 induction by recombinant CagL alone also requires the RGD motif. Furthermore, integrin α5- and integrin β1-function blocking antibodies, BIIG2 and AIIB2, respectively, abrogated IL-8 induction by wild-type CagL (Fig. 2B) and CagLRGA (Fig. S5). The latter is consistent with the recent observation that a helper sequence outside the RGD motif contributes to binding of CagL to integrin α5β1 (Conradi et al., 2012). The integrin αvβ5-function blocking antibody P1F6, which is widely used to specifically block integrin αvβ5-mediated functions (Weinacker et al., 1994; Beauvais et al., 2009), caused a slight but reproducible inhibition of IL-8 release by CagL (Fig. 2B), suggesting that integrin αvβ5 might also play a minor role in IL-8 induction by CagL. Additionally, manganese chloride, which stabilizes the active and ligand-bound state of integrin heterodimers (Mould et al., 1998), caused a 15% increase in the level of IL-8 induced by CagL (Fig. 2B). These results collectively suggest that CagL can directly induce IL-8 secretion through interaction with integrin α5β1, and possibly also integrin αvβ5. Furthermore, the possibility that the observed IL-8 secretion is due to Escherichia coli lipopolysaccharide (LPS) contamination can be ruled out because AGS cells are entirely non-responsive to E. coli LPS (Backhed et al., 2003; Su et al., 2003) due to its lack of expression of the essential TLR-4 cofactor MD2 (Smith et al., 2003). In agreement, colistin, a LPS antagonist, had no effect at concentrations up to 50 μg ml−1 on the induction of IL-8 by CagL. Given that LPS and bacterial lipopeptides are relatively heat stable, the observation that only native and not HI CagL elicited IL-8 induction (Fig. 2A) further argues against a contribution by contaminating LPS or lipopeptide. The latter is also supported by evidence that lipopeptide responsive Toll-like receptor (TLR)-2 is not expressed in AGS (Smith et al., 2003), and by the non-responsiveness of AGS cells used in this study to TLR-2 agonist Pam3Cys (not shown).
CagL-induced IL-8 secretion occurs via signal transduction involving Src, Ras, ERK and NF-κB, but independently of NOD1 signalling
Activation of integrin α5β1 can result in downstream activation of the tyrosine kinase Src, the small GTP-binding protein Ras, and the extracellular signal-regulated kinase (ERK) mitogen-activated protein kinase (MAPK) pathway. Indeed, the well-characterized and specific inhibitors, Ras inhibitor farnesyl thiosalicylic acid (FTS) (Marciano et al., 1995; Marom et al., 1995), Src kinase inhibitor Src I-1 (Tian et al., 2001; Bain et al., 2007), MEK1/2 inhibitor U0126 (Bain et al., 2007) and the c-Jun amino-terminal kinase (JNK) inhibitor SP600125 (Bain et al., 2007), all significantly inhibited IL-8 secretion by CagL in a dose-dependent manner (Figs 2C and S6). In contrast, p38 inhibitor SB203580 (Bain et al., 2007) had no significant effect on the activation of IL-8 induction by CagL (Fig. 2C) at concentrations up to 10 μM. The specificities of SrcI-1, U0126, SP600125 and SB203580 were confirmed by Western blot (Fig. S7). In addition, CagL activated ERK, JNK and Src kinase in a RGD motif-dependent manner (Fig. S8). These findings suggest that activation of Src, the ERK MAPK pathway and JNK, but not p38, play an important role in CagL-mediated IL-8 induction. The > 30% reduction in IL-8 level seen with the cholesterol-depleting agent methyl-β-cyclodextrin (MβCD) (Fig. 2C) suggests that cholesterol-rich lipid domains also play a role in CagL-mediated IL-8 induction. The proteasome-specific inhibitor MG-132 (Hellerbrand et al., 1998) and the highly specific IκB kinase inhibitor, BMS-345541 (Burke et al., 2003), both significantly inhibited IL-8 induction by CagL in AGS cells in a dose-dependent manner (Figs 2C and S6). Trypan blue exclusion and MTT proliferation assays confirmed that the inhibitor concentrations used had no detectable effect on the viability of AGS cells (data not shown). Taken together, these results indicate that IL-8 induction by CagL involves not only cholesterol-rich lipid domains and activation of the Ras-Raf-ERK MAPK pathway, but also activation of JNK and NF-κB.
We also stimulated NOD1 knock-down cells with CagL or the NOD1 agonist, Tri-DAP. Although NOD1 has been shown to play a role in IL-8 induction by H. pylori (Viala et al., 2004; Allison et al., 2009), CagL-mediated IL-8 secretion was unaffected by NOD1 knock-down (Fig. 2D). Conversely, IL-8 induction in response to transfected Tri-DAP was significantly diminished in NOD1 knock-down cells compared with control cells. These results indicate clearly that CagL-mediated IL-8 stimulation does not require NOD1.
The metalloprotease ADAM17 has been implicated to play a role in CagL-mediated repression of H, K-adenosine triphosphatase α-subunit (Saha et al., 2010). We therefore tested whether ADAM17 might contribute to CagL-mediated IL-8 induction. Knock-down of ADAM17 mRNA level by 80% had no detectable effect on the level of IL-8 induced by recombinant CagL (Fig. S9), suggesting that ADAM17 does not play a role in CagL-mediated IL-8 induction.
Recombinant CagL restores IL-8 induction by H. pyloricagL-deletion mutant
We next examined whether purified recombinant CagL could functionally complement a cagL-deletion mutant in triggering IL-8 secretion during H. pylori infection. AGS cells were infected with P12cagLWT or P12ΔcagL, in the absence and presence of recombinant CagL. Remarkably, native CagL protein, but not heat-inactivated CagL, restored stimulation of IL-8 secretion by P12ΔcagL to > 40% of that induced by wild-type P12 (Fig. 3A) and in a dose-dependent manner (Fig. S10). These observations suggest that recombinant CagL could to a large extent mimic endogenous CagL in triggering IL-8 secretion during infection of AGS. In addition, recombinant CagL enhanced IL-8 secretion by AGS in response to P12cagLWT and P12cagLRGA (both P < 0.001) by a comparatively moderate level (Fig. 3A). This difference is most likely due to the presence of endogenous CagL in these strains.
Importantly, CagL-mediated rescue of IL-8 induction in response to P12ΔcagL occurred independently of CagA-mediated signal transduction because complementation with exogenous CagL did not restore translocation and phosphorylation of CagA (Fig. 3B). In contrast to genetic complementation (Figs 1 and 3B), this approach allows the roles of CagL in CagA translocation and IL-8 induction to be uncoupled. Thus, these findings together with the results of the mutagenesis study (Fig. 1) reveal for the first time that CagL is capable of signalling and inducing IL-8 during H. pylori infection via pathway(s) independent of CagA translocation.
Such CagA translocation-independent, CagL-dependent IL-8 induction was significantly enhanced by the RGD motif of CagL with evidence as follows. First, CagLWT and not CagLRGA measurably enhanced P12ΔcagL-mediated IL-8 secretion at 7 hpi (10-fold that of P12ΔcagL alone; P < 0.001) (Fig. S11A). Second, CagLRGA compared with CagLWT facilitated a significantly less substantial increase in P12ΔcagL-mediated IL-8 secretion at 24 hpi (5.5-fold instead of 7.5-fold increase; P < 0.0001) (Fig. S11C).
In addition, exogenous CagLWT restored IL-8 induction by P12ΔcagL to levels similar to those induced by P12ΔcagA or P12ΔvirD4 (Fig. S11B), both of which are defective for CagA translocation but capable of T4SS-dependent IL-8 induction. In contrast, exogenous CagLWT only minimally enhanced IL-8 secretion in response to P12ΔcagPAI (Fig. S11A and B). To examine this effect further, we measured binding of exogenous CagL to H. pylori P12 using an attachment assay in which H. pylori P12ΔcagL or P12ΔcagPAI were cultured with immobilized recombinant CagLWT, CagLRGA, bovine serum albumin (BSA), H. pylori-specific polyclonal antibody (positive control) or pre-immune antibody (negative control). Using this approach, we found that P12ΔcagL bound efficiently to recombinant CagL in a RGD-independent manner, but P12ΔcagPAI binding to CagL was not different from background BSA or pre-immune antibody binding (Fig. S12A). We also used super-resolution microscopy to confirm that recombinant CagLWT efficiently coated P12ΔcagL but not P12ΔcagPAI (Fig. S12B).
Altogether, these results indicate that exogenous CagL binds to cagPAI-encoded T4SS surface components and triggers a substantial IL-8 response during H. pylori infection in a CagA-independent but partially RGD-dependent manner.
Activation of β1 integrin and JNK, but not NOD1, is critical for CagL-dependent induction of IL-8 secretion by H. pylori
As described earlier, recombinant CagL alone utilized both ERK and JNK signalling via β1 integrin and Src kinase activation to stimulate IL-8 release by AGS (Fig. 2). We also observed that IL-8 induction by AGS in response to P12ΔvirD4 was significantly attenuated upon inhibition of Src, ERK and JNK by SrcI-1, U0126 and SP600125, respectively (Fig. S13), suggesting that Src, ERK and JNK-mediated signalling might contribute to T4SS-substrate-independent but CagL-dependent IL-8 induction during H. pylori infection. To further test this hypothesis, we incubated AGS with P12ΔcagL and exogenous CagL in the presence or absence of various specific inhibitors. AIIB2 and SP600125 significantly inhibited the ability of exogenous CagL to restore IL-8 induction by P12ΔcagL without affecting IL-8 induction by P12ΔcagL alone (Fig. 4A). The Src inhibitor Src I-1 or MEK1/2 inhibitor U0126 also significantly attenuated the ability of exogenous CagL to restore IL-8 secretion by H. pylori P12ΔcagL, while causing a slight inhibition of IL-8 secretion by P12ΔcagL alone. These data suggest that the activation of β1 integrin and JNK is critical for CagA translocation-independent, CagL-dependent IL-8 induction during H. pylori infection whereas activation of Src and ERK contributes to both CagL-dependent and CagL-independent signal transduction.
Because β1 integrin has been implicated in peptidoglycan delivery and NOD1-mediated signal transduction by H. pylori (Hutton et al., 2010), we examined the role of NOD1 in CagL-dependent IL-8 induction. NOD1 knock-down cells secreted reduced IL-8 compared with the control cells upon infection by P12 wild-type, but not T4SS-defective strains P12ΔcagL and P12ΔcagPAI (Fig. 4B), confirming that NOD1 contributes to IL-8 induction by H. pylori in a T4SS-dependent manner (Viala et al., 2004; Allison et al., 2009). A NOD1-dependent effect on IL-8 secretion was also not observed when P12ΔcagL-inoculated cells were incubated with CagL (Fig. 4B), indicating that CagA translocation-independent, CagL-dependent IL-8 secretion did not require NOD1 signalling. In addition, NOD1 knock-down cells infected with P12ΔcagL alone secreted around 60% less IL-8 than NOD1 knock-down cells infected with P12ΔvirD4 (Fig. S14). Altogether, these results clearly indicate that there is a NOD1-independent and T4SS substrate-independent IL-8 response to H. pylori P12 and that this response depends on CagL.
CagL stimulates nuclear translocation of NF-κB p65 subunit in a T4SS-dependent but CagA-independent manner
To examine whether CagA translocation-independent, CagL-dependent IL-8 induction during H. pylori infection also involved NF-κB activation, we tracked nuclear translocation of endogenous NF-κB subunit p65 by immunofluorescent labelling. Nuclear p65 preceded IL-8 secretion and was evident at all time points in P12-inoculated cells, peaking at 2 hpi and dropping to a comparatively lower proportion of activated cells at 7 hpi that was then sustained through to 24 hpi (Fig. 5A and B). This reduction agrees with published data showing asynchronized oscillation of H. pylori-induced NF-κB p65 nuclear translocation following the highly synchronized onset of activation seen immediately upon infection (Bartfeld et al., 2010). In contrast, nuclear p65 was absent from P12ΔcagL-inoculated cells until 3 hpi. However the proportion of CagLWT-treated P12ΔcagL-inoculated cells positive for nuclear p65 by 7 hpi was significantly greater than cells stimulated with P12ΔcagL alone (P < 0.001, two-way anova). This increase coincided with the initiation of IL-8 secretion by P12ΔcagL-inoculated CagLWT-treated cells. The proportion of CagL-dependent NF-κB activated cells was similar to that of P12-inoculated cells sampled at 7 hpi and was also sustained through to 24 hpi (Fig. 5C), coinciding with continued accumulation of secreted IL-8 (Fig. 5A). No such significant increase in nuclear p65 occurred in P12ΔcagPAI or BHI inoculated cells at any time point, regardless of whether treated with CagLWT or not, although nuclear p65 was detected in a very small proportion of cells inoculated with P12ΔcagPAI by 24 hpi in the presence and absence of CagLWT. These results were all in agreement with the IL-8 secretion data (Fig. 5A). Thus in addition to CagA, CagL is another H. pylori T4SS-encoded virulence factor that promotes NF-κB activation. Moreover, CagA translocation-independent, CagL-dependent nuclear translocation of NF-κB p65 correlates with CagA translocation-independent, CagL-dependent IL-8 induction, and these responses to CagL are enhanced in the presence of cag PAI.
Pronounced IL-8 secretion in the gastric mucosa is a hallmark of H. pylori-associated chronic gastritis. Although the predominant role of the cag PAI-encoded T4SS in H. pylori-mediated IL-8 induction has been well established (Crabtree et al., 1994; Audibert et al., 2001; Fischer et al., 2001; Viala et al., 2004; Brandt et al., 2005; Allison et al., 2009), the precise mechanisms by which the T4SS apparatus stimulates IL-8 induction has been a longstanding controversy. Upon delivery into host cells by the T4SS, H. pylori protein CagA and the cell wall component peptidoglycan stimulate secretion of IL-8 (Viala et al., 2004; Brandt et al., 2005). However, the H. pylori ΔvirD4 mutant that lacks the T4SS coupling protein and hence is defective in T4SS substrate translocation is still able to trigger IL-8 induction, suggesting that IL-8 induction can occur in the absence of substrate translocation (Fischer et al., 2001; Selbach et al., 2002). Here, we have shown that CagL alone can stimulate IL-8 induction in gastric epithelial cells independently of any additional H. pylori factors. This response required integrin α5β1 as well as the native conformation and RGD motif of CagL. CagL-mediated pro-inflammatory signal transduction also involved Src kinase, ERK, JNK and NF-κB activation, corroborating previous speculations that interaction of CagL with integrin α5β1 could activate MAPK pathway and innate immune responses (Snider et al., 2008; Hutton et al., 2010).
Identification of specific T4SS components that directly induce IL-8 has been difficult because the cag PAI genes necessary for IL-8 induction are also required for translocation of CagA, which itself exacerbates IL-8 induction. In particular, knocking out cagL concomitantly abrogates CagA translocation (Fischer et al., 2001; Kwok et al., 2007). Here we used a model system based on functional complementation of H. pylori ΔcagL mutant with exogenous CagL to dissect the mechanistic role of CagL in IL-8 induction. In contrast to genetic complementation of H. pylori ΔcagL, exogenous CagL protein restored IL-8 induction, but not CagA translocation, by H. pylori ΔcagL. This novel approach has allowed us to disengage the contribution of CagL to IL-8 induction from its role in CagA translocation for the first time. Using NOD1 knock-down cells, we have further demonstrated that CagL-dependent IL-8 induction by H. pylori was also independent of NOD1 signalling. We did not investigate the role of the related peptidoglycan-sensing molecule, NOD2, in these IL-8 responses because it was previously shown that H. pylori bacteria in exponential phase contain more of the peptidoglycan breakdown products that preferentially activate NOD1 and not NOD2 signalling (Chaput et al., 2006).
We also observed that IL-8 secretion induced upon complementation of H. pylori P12ΔcagL by recombinant CagL was approximately 20-fold or up to 9-fold greater than that induced by recombinant CagL or P12ΔcagL alone respectively. This interesting observation indicates that while CagL alone is capable of inducing a moderate level of IL-8 through direct interaction with host integrin, maximal IL-8 induction to the level seen in H. pylori infection requires synergy between CagL and the rest of the T4SS apparatus. Indeed we also observed that this synergy required components of the T4SS other than CagL, and that recombinant CagL could attach to P12ΔcagL but not P12ΔcagPAI. This is consistent with the recent findings suggesting that CagL interacts with the T4SS components CagI and CagH, and that such interaction influences type IV secretion of CagA and IL-8 induction (Shaffer et al., 2011). Moreover, recent data suggest that CagL expression may influence maximal CagI expression (Pham et al., 2012). Whether this influence occurs at the translational and/or transcriptional level remains to be elucidated. Here we hypothesize that the cooperative interaction of CagL with other T4SS components and/or host cell surface integrin receptors could trigger rearrangement of the T4SS apparatus and/or integrin-associated protein complexes, and hence heighten activation of downstream signalling and subsequent IL-8 response. Our findings support the following model (Fig. 6): binding of CagL to integrin α5β1 can directly trigger activation of Src kinase, ERK, JNK and NF-κB, and subsequently IL-8 induction; meanwhile, CagL can also synergize with the rest of T4SS to elicit an enhanced response, which like the response triggered by CagL alone, is also dependent on signalling via integrin α5β1, Src kinase, ERK, JNK and NF-κB activation but independent of CagA translocation or NOD1 signalling. The precise molecular mechanism underlying the synergy between CagL and the rest of the T4SS in IL-8 induction is currently being investigated in our laboratories.
This study also reveals that the role of β1 integrin in H. pylori pathogenesis is even more important than previously envisaged: β1 integrin is a critical receptor for H. pylori-mediated signal transduction that functions by facilitating translocation of the signalling effectors CagA and peptidoglycan into host cells (i.e. substrate theory), as well as by interacting with CagL (i.e. direct activation theory). In order to further ascertain the role of CagL-β1 integrin interactions in IL-8 induction, we examined the contribution of the RGD motif of CagL to H. pylori-mediated induction of IL-8 secretion using amino acid substitution mutants of H. pylori as the precise structural configuration of the RGD motif is known to be critical for their recognition by integrins (Ruoslahti, 1996; Arnaout et al., 2002; Takagi, 2004). The aspartate residue in the RGD motif is crucial for determining integrin-binding affinity as it not only forms a ternary complex with a divalent ion in the binding site but also makes significant contact with multiple residues in the βA domain of integrin αvβ3, and possibly also integrin α5β1 (Xiong et al., 2002; Takagi et al., 2003), whereas the glycine residue of the RGD motif interacts less extensively with the integrin binding pocket and substitutions at that position are less likely to block integrin binding (Ruoslahti, 1996; Xiong et al., 2002). Our findings are consistent with these notions. Compared with the substitution of the RGD motif to RAD in CagL, we observed that substitution of RGD to RGA had a greater impact, ablating CagA translocation and severely attenuating IL-8 secretion. Nevertheless, recombinant CagLRGA was still able to enhance IL-8 secretion in response to P12ΔcagL, albeit to a significantly lesser extent than CagLWT. Incomplete inhibition of IL-8 induction upon mutation of the RGD motif to RGA was also observed with recombinant CagLRGA alone or by the H. pylori mutant P12ΔcagLRGA. This moderate IL-8 promoting effect of CagLRGA is consistent with the previous finding that CagLRGA is still able to bind β1 integrin (Wiedemann et al., 2012), albeit at a fourfold reduction (Kwok et al., 2007). Thus, CagL-dependent signalling leading to IL-8 secretion is at least in part dependent on its RGD motif for maximal efficacy, and these results confirm those in our previous study (Kwok et al., 2007) despite some controversy in the literature (Jimenez-Soto et al., 2009; Schuelein et al., 2011). The controversy could be partly due to variable expression of the cagL mutants in these studies as different constitutive promoters were used in the various CagL mutant-expressing constructs (Kwok et al., 2007; Jimenez-Soto et al., 2009). In order to robustly and independently test the functions of the CagL-RGA and CagL-RAD mutants, we constructed the amino acid substitution strains using an approach independent of those used in previous studies, whereby the cagL gene in the CagLRAD and CagLRGA mutants was placed under the control of its native promoter to preserve the regulatory control and stoichiometry of the cagG-cagL operon [the latter was suggested by a recent study to be important for T4SS functions (Shaffer et al., 2011)]. In addition, the mutant phenotypes have been thoroughly verified using multiple independent clones. Together with the fact that ‘knocking in’ of CagLRGD restored the ability of P12ΔcagL to induce IL-8 and translocate CagA, these results confirmed that the observed effects of the RAD and RGA mutations were neither coincidental nor polar.
Helicobacter pylori stimulates an array of signal transduction cascades within the host cell leading to profound effects including changes in cell morphology and motility via cytoskeletal rearrangement (Selbach et al., 2002; Snider et al., 2008), and altered expression of host transcriptional factors (Backert et al., 2005). Extensive cross-talk and potential redundancies arise as a result of H. pylori triggering multiple cascades via a single bacterial factor [e.g. CagA initiating both morphology-related and pro-inflammatory cascades (Backert et al., 2010)], while simultaneously effecting similar outcomes via disparate bacterial factors [e.g. cell scattering via CagA-independent, T4SS-dependent factors (Snider et al., 2008) and pro-inflammatory cascades via peptidoglycan (Viala et al., 2004; Allison et al., 2009)]. Like CagA, CagL is emerging as a multifaceted virulence factor capable of triggering numerous host cell signalling cascades, a notion that is elaborated in further detail in this study. It has been shown previously that in addition to being critical for CagA translocation (Fischer et al., 2001; Selbach et al., 2002), CagL mediates Src and focal adhesion kinase (FAK) activation (Kwok et al., 2007), epidermal growth factor receptor activation leading to cell spreading (Tegtmeyer et al., 2010), interacts with integrin αvβ5 to promote gastrin expression (Wiedemann et al., 2012), and suppresses gastric acid secretion via metalloprotease ADAM17 activation (Saha et al., 2010). Here we showed that CagL triggered additional signalling pathways leading to pro-inflammatory responses. Using a pharmacological approach, we examined the signalling factors contributing to CagL-mediated IL-8 induction. First, we found that IL-8 induction arising from the binding of CagL via its RGD motif to α5β1 integrin required cholesterol-rich lipid raft domains. This is in agreement with a requirement of lipid raft integrity for several T4SS-dependent consequences of H. pylori infection such as β1 integrin-mediated pro-inflammatory responses (Hutton et al., 2010), CagA translocation (Lai et al., 2008), and the accumulation of activated IκB kinase (IKK) complex component IKKβ in lipid raft domains (Rieke et al., 2011). Second, we found that inhibition of Src, Ras or ERK attenuated CagL-mediated IL-8 induction. These data suggest that CagL, by binding to α5β1 integrin, activates the well-characterized integrin signalling pathway of α5β1 integrin-Src-Ras-Raf-ERK (Schlaepfer and Hunter, 1998). Together with the previous findings of CagL-dependent FAK activation (Kwok et al., 2007), our data imply that CagL may stimulate integrin-mediated MAPK activation in a FAK-, Src- and Ras-dependent manner. One pathway by which integrin can stimulate signalling to ERK is via recruitment of the Grb2 adaptor protein whereby FAK-bound Grb2 recruits Ras to the FAK-Grb2-integrin signalling, resulting in activation of the Raf-MEK-ERK cascade (Schlaepfer et al., 1994; Zhu and Assoian, 1995; Schlaepfer and Hunter, 1998). Src also plays an important role in integrin-dependent activation of ERK activation partly because Grb2 binding to FAK is significantly promoted by the phosphorylation of FAK by Src (Schlaepfer and Hunter, 1996). It would be of interest to examine whether CagL-mediated ERK activation also requires Grb2 binding to FAK. In examining other MAPK pathways, we found that inhibition of JNK, but not p38, also attenuated IL-8 induction by CagL. CagA-independent JNK activation via α5β1 integrin by H. pylori has been reported and was postulated to be CagL-mediated (Snider et al., 2008). Our findings support this hypothesis.
Third, we used inhibition of proteasomal IκB degradation together with direct tracking of NF-κB subunit p65 to provide evidence that CagA translocation-independent, CagL-dependent NF-κB activation contributes to H. pylori IL-8 induction. Activation of NF-κB and AP-1 transcriptional factors by H. pylori and hence induction of IL-8 expression has been described (Glocker et al., 1998; Naumann et al., 1999; Meyer-ter-Vehn et al., 2000; Allison et al., 2009). However while the induction of MAPK pathways leading to AP-1 activation is well documented (Karin et al., 1997), the precise mechanisms of H. pylori-induced IκB kinase (IKK) activation and subsequent NF-κB-mediated IL-8 induction, and why this often coincides with MAPK activation, remains to be clarified. One of the mechanisms by which the Ras-Raf-MEK-ERK pathway can activate NF-κB is via the serine-threonine mitogen-activated 90 kDa ribosomal S6 kinase (pp90rsk1 or RSK1). Upon stimulation by mitogen, RSK1 is phosphorylated and activated by ERK (Anjum and Blenis, 2008). Active RSK1 can phosphorylate IκBα on serine 32, thereby triggering ubiquitin-dependent IκBα degradation in the 26S proteasome (Ghoda et al., 1997; Schouten et al., 1997), an important step leading to NF-κB activation. Alternatively another substrate of ERK, the nuclear kinase mitogen- and stress-activated protein kinase-1 (MSK1), can directly phosphorylate the NF-κB p65 subunit leading to its transcriptional activation (Vermeulen et al., 2003). Whether RSK1 and/or MSK1 play(s) a role in CagL-integrin-dependent activation of NF-κB remains to be investigated. In addition, recent data indicating a direct role for Src in phosphorylation of IKKβ (Rieke et al., 2011) suggests the possibility that NF-κB activation may result from enhanced IκB degradation mediated by CagL-activated Src.
In contrast to ERK, JNK typically has an antagonistic relationship with NF-κB (Papa et al., 2004) and so is unlikely to directly contribute to NF-κB activation in this context. However, activated JNK directs phosphorylation of the critical AP-1 subunit c-Jun. JNK activation by H. pylori has been shown to be T4SS-dependent but peptidoglycan-independent (Snider et al., 2008; Allison et al., 2009), which is in agreement with our findings. Moreover, H. pylori-mediated AP-1 activation depends on T4SS-dependent ERK activation (Naumann et al., 1999; Meyer-ter-Vehn et al., 2000), possibly via upregulation of the expression of the AP-1 subunit c-Fos through activation of the ERK substrate Elk1 (Mitsuno et al., 2002). These data, together with our finding of a contribution of JNK and ERK to CagA translocation-independent, CagL-dependent IL-8 induction, suggest that investigation of the precise contribution of CagL to AP-1-mediated pro-inflammatory responses is warranted.
This study documents a novel pathway of IL-8 induction triggered by the H. pylori T4SS. Building on previous studies (Kwok et al., 2007; Shaffer et al., 2011), our findings pinpoint CagL, a T4SS surface component, as a central contributor to T4SS pro-inflammatory and translocation functions. This notion flags CagL and its receptor, integrin, as potential drug targets for alleviating H. pylori-induced inflammation. Our findings also highlight the novel concept that apart from translocation substrates, surface component(s) of a bacterial T4SS apparatus can also be recognized by the host innate immune system in the induction of potent pro-inflammatory responses.
Cell and bacterial culture
AGS cells were routinely cultured in RPMI (Invitrogen, Carlsbad, CA, USA) supplemented with 10% (v/v) fetal bovine serum (FBS, Invitrogen) in a humidified incubator at 37°C with 5% CO2. AGS cells stably expressing NOD1-specific shRNA (AGS-siNOD1) and control cells expressing enhanced green fluorescent protein (EGFP)-specific shRNA (AGS-siEGFP), were maintained as for AGS cells with occasional culture in RPMI supplemented with G148 (400 μg ml−1) (Allison et al., 2009). The significant downregulation of NOD1 expression (> 75%) in AGS-siNOD1, compared with AGS-siEGFP control, has been described previously (Allison et al., 2009). H. pylori strains (listed in Table S1) were routinely cultured on GC agar (Oxoid, Basingstoke, UK) supplemented with 10% (v/v) horse serum (Invitrogen), vitamin mix, vancomycin and nystatin as described previously (Brandt et al., 2005). For cell-culture inoculum, H. pylori strains were cultured in brain heart infusion (BHI) or heart infusion broth (Oxoid) supplemented with 10% (v/v) FBS, vitamin mix and vancomycin (Sigma, St Louis, MO, USA). H. pylori growth medium was further supplemented with chloramphenicol (Cm; 10 μg ml−1 for plate culture, 4 μg ml−1 for broth culture) or kanamycin sulfate (Km; 15 μg ml−1) as required. All H. pylori culture was performed at 37°C under microaerobic conditions generated using the CampyGen system (Oxoid); broth cultures were shaken at 120 r.p.m. E. coli strain DH5α was propagated as described previously (Gorrell et al., 2005).
Chemicals, antibodies and siRNAs
Antagonists, agonists, siRNAs, transfection reagents and antibodies used in this study were obtained from the following suppliers: U0216, Cell Signaling Technology (Beverly, MA, USA); MG-132, BMS-345541 and SB203580, Merck KGaA (Darmstadt, Germany); Src Inhibitor-1, SP600126, methyl-β-cyclodextrin (MβCD), manganese chloride and colistin, Sigma; Farnesyl thiosalicylic acid (FTS), Cayman Chemicals (Ann Arbor, MI, USA); Phorbol-12-myristate-13-acetate (PMA), Merck KGaA (Darmstadt, Germany); NOD1-targeting siRNAs (ID 20324 and 20322), Life Technologies (Carlsbad, CA, USA); negative control siRNA (Cat. No. 1022076), QIAGEN (Hilden, Germany); pooled ADAM17-targeting siRNAs (J-003453-05) and negative control siRNA (D-001810-03), Dharmacon RNA Technology (Lafayette, CO, USA); NOD1 ligand l-Ala-γ-d-glutamyl-meso-diaminopimelic acid (Tri-DAP), EMC microcollections GmbH (Tuebingen, Germany); Lipofectamine 2000, Life Technologies; FuGene HD, Roche (Basel, Switzerland); integrin αvβ5-specific mouse monoclonal antibodies (mAb) P1F6, Millipore (Billerica, MA, USA); rat mAb AIIB2 (specific for integrin β1) (Hall et al., 1990) and BIIG2 (specific for integrin α5), developed by C.H. Damsky were obtained as conditioned culture media ultrafiltration concentrates from the Developmental Studies Hybridoma Bank developed under the auspices of the NICHD and maintained by The University of Iowa, Department of Biology, Iowa City, IA 52242. AIIB2 and BIIG2 concentrates used with H. pylori were dialysed overnight at 4°C against 1000 volumes PBS using 3 kDa cut-off Slide-A-lyser dialysis cups (Pierce, Rockford, IL, USA) to ensure the removal of any residual gentamicin remaining from hybridoma culture. Phospho-ERK (Thr202/Tyr204)-specific rabbit mAb (clone D13.14.4E), phospho-JNK (Thr183/Tyr185)-specific rabbit mAb (clone 81E11), phospho-p38 (Thr180/Tyr182)-specific rabbit mAb (clone 3D7) and glyceraldehyde-3-phosphate dehydrogenase (GAPDH)-specific rabbit mAb (clone 14C10) were purchased from Cell Signaling Technology (Boston, MA, USA); PY99 phosphotyrosine-specific mouse mAb and anti-CagA rabbit polyclonal antibody, Santa Cruz Biotechnology (Santa Cruz, CA, USA). Actin-specific mouse mAb (clone ACTN05; catalogue No. Ab3280), Abcam (Cambridge, MA, USA).
Recombinant protein production
Recombinant CagL was purified as described previously (Kwok et al., 2007) with the following exceptions. Induction of expression was carried out at 16°C for 20 h. CagL was purified under native condition by affinity chromatography through HisTrap HP (GE Healthcare, UK) and then gel filtration through Superdex75 (GE Healthcare) according to procedures described elsewhere (Kwok et al., 2007). The apparent size of CagL eluted was approximately 30 kDa. Circular dichroism analysis based on the data sets of 33 reference proteins (Toumadje et al., 1992) was used to confirm the folded conformation of purified CagL and CagLRGA (Kwok et al., 2007). Heat-inactivated CagL (HI CagL) was prepared by heating CagL in 50 mM Tris-HCl (pH 8), 100 mM NaCl and 25 mM KCl at 95°C for 30 min.
Stimulation of AGS cells with recombinant CagL protein
All experiments were performed in duplicate. AGS were grown in 24-well plates (3 × 104 cells per well) for 20 h at 37°C in 5% CO2. AGS were washed three times with warm PBS and then treated in duplicate with 200 μl of antibody or inhibitor (AIIB2, BIIG2, colistin, manganese chloride, Src-I, U0126, SP600125 or SB203580) in RPMI supplemented with 100 units ml−1 of penicillin and 100 μg ml−1 streptomycin (P/S) for 30 min at 37°C. Then 50 μl of CagL (or RPMI as negative control) was added to AGS to a final concentration of 8 μM. After 1 h incubation at 37°C, 250 μl of warm RPMI (P/S) was added and the incubation continued. Treatment with farnesyl thiosalicylic acid (FTS) was conducted using RPMI (P/S) supplemented with 5% FBS. For treatment with methyl-β-cyclodextrin (MβCD), BMS-345541 or MG-132, inhibitors were removed after 30 min, 1 h or 2 h incubation, respectively, by washing cells three times with warm PBS prior to incubation with CagL (4 or 0 μM) in RPMI (P/S) at 37°C. For all treatments, spent culture medium was collected at 24 h post stimulation with CagL and centrifuged at 4000 g at 4°C for 5 min to remove cell debris. The amount of IL-8 secreted was assayed in duplicate using the human IL-8 ELISA set (BD Biosciences, USA) according to manufacturer's instructions. Statistical analysis of P12ΔcagL supplemented with CagLWT was performed on values subtracted for P12ΔcagL alone. Viability of inhibitor-treated AGS cells was verified by trypan blue exclusion assay (Sigma) and 1-(4,5-Dimethylthiazol-2-yl)-3,5-diphenylformazan (MTT) assay (Sigma) according to manufacturer's instruction.
H. pylori mutagenesis
Recombinant DNA techniques, preparation of Km (aphA3) and Cm (cat) resistance cassettes and transformation of H. pylori were performed as described previously (Gorrell et al., 2005). All primers are listed in Table S2. Construction of donor DNA species for generation of CagL deletion and amino acid substitution mutants was performed as described in supplementary Fig. S1. For transformation of H. pylori, donor DNA was PCR amplified from the resulting constructs by Vent DNA polymerase (New England Biolabs, USA) using primer pair cagNF/cagIR. The phenotypic and genetic integrity of the derived mutants was confirmed by infection assays and sequence analysis of HpcagNF/HpcagLR amplicons, respectively, of at least three independent clones. P12ΔvirD4 partial deletion mutant was created by insertion of cat into BclI sites at bases 557452 and 557626 of the P12 genome. Clones were confirmed phenotypically by infection assays and genetically by sequence analysis of primer pair virD4F2/virD4R2 amplicons. P12cagLWTΔvirD4 and P12cagLRGAΔvirD4 double mutants were created by insertion of cat into BclI sites at bases 557452 and 557626 of the genome of P12cagLWT and P12cagLRGA respectively. Multiple independent clones were confirmed phenotypically by infection assays and genetically by sequence analysis using the primer pairs virD4F2/virD4R2 and cagNF/cagIR.
H. pylori stimulation of AGS cells
AGS cells seeded in 24-well plates (5 × 104 cells per well) were used 48 h after seeding. For experiments using AGS-siNOD1 and AGS-siEGFP cells, maintenance media was changed to serum-free RPMI 24 h after seeding. All AGS cell incubations were performed at 37°C with 5% CO2. Plate-grown H. pylori was used to inoculate BHI broth to 0.1–0.2 OD550 and grown overnight to 0.6–0.8 OD550. For comparison of H. pylori isogenic mutants, AGS cell maintenance media was directly inoculated with H. pylori at a multiplicity of infection (moi) of 100 cfu cell−1, as determined by optical density against a reference standard curve. For CagL stimulation prior to H. pylori inoculation, cells were washed three times with PBS before the addition of 250 μl of RPMI supplemented with 8 μM CagL, or RPMI alone. After 2 h incubation, media was supplemented with 250 μl of RPMI containing H. pylori strains (final moi of 30) or equivalent volume of sterile BHI broth as appropriate per well. For inhibitor studies, cells were washed three times with PBS before pre-treatment with 200 μl of RPMI containing inhibitor or integrin-blocking antibody. After 30 min pre-treatment, wells were supplemented with 50 μl of RPMI containing 39 μM CagL (final 8 μM) or RPMI alone, followed after 2 h by an additional 250 μl of RPMI containing H. pylori strains (final moi of 30) or sterile BHI broth. The amount of IL-8 secreted by the AGS cells was assayed using the human IL-8 ELISA set (BD Biosciences) according to manufacturer's instructions. Experiments were performed in duplicate.
Statistical analyses were performed using Prism 5.0 (GraphPad software, San Diego, CA, USA). Significance was determined using Student's t-test, one-way anova or two-way anova with post-tests as appropriate.
Supplementary information materials and methods has details on immunoblot analysis, transfections, qRT-PCR and p65 immunofluorescence.
We are grateful to Mr S. Lim and Ms V. Zhang for technical assistance; Ms P. Everingham for help with data analysis using the software Linreg; the Monash Protein Purification Unit and Dr R. Law for the purification of CagL and optimization of the purification protocol respectively; Prof. David Jans and Dr Kylie Wagstaff for reagents and advice for DTAF protein labelling; Dr Judy Callaghan (Monash Micro Imaging) and Dr Marko Lampe (Leica Microsystems) for assistance with GSDIM microscopy and invaluable advice; Prof. S. Backert for providing the H. pylori strains P12, P12ΔcagA, P12ΔcagPAI and P1. The work was supported by NHMRC project Grants APP545983 and APP1006010, and the Victorian Government's Operational Infrastructure Support Program.