Toll like Receptor signalling by Prevotella histicola activates alternative NF-κB signalling in Cystic Fibrosis bronchial epithelial cells compared to P.aeruginosa

Cystic Fibrosis (CF), caused by mutations affecting the CFTR gene, is characterised by viscid secretions in multiple organ systems. CF airways contain thick mucus, creating a gradient of hypoxia, which promotes the establishment of polymicrobial infection. Such inflammation predisposes to further infection, a self-perpetuating cycle in mediated by NF-κB. Anaerobic Gram-negative Prevotella spp. are found in sputum from healthy volunteers and CF patients and in CF lungs correlate with reduced levels of inflammation. Prevotella histicola (P.histicola) can suppress murine lung inflammation, however, no studies have examined the role of P.histicola in modulating infection and inflammation in the CF airways. We investigated innate immune signalling and NF-kB activation in CF epithelial cells CFBE41o-in response to clinical stains of P.histicola and Pseudomonas aeruginosa (P.aeruginosa). Toll-Like Receptor (TLR) expressing HEK-293 cells and siRNA assays for TLRs and IKKa were used to confirm signalling pathways. We show that P.histicola infection activated the alternative NF-kB signalling pathway in CF bronchial epithelial cells inducing HIF-1α protein. TLR5 signalling was responsible for the induction of the alternative NF-kB pathway through phosphorylation of IKKα. The induction of transcription factor HIF-1α was inversely associated with the induction of the alternative NF-kB pathway and knockdown of IKKα partially restored canonical NF-kB activation in response to P.histicola. This study demonstrates that different bacterial species in the respiratory microbiome can contribute differently to inflammation, either by activating inflammatory cascades (P.aeruginosa) or by muting the inflammatory response by modulating similar or related pathways (P.histicola). Further work is required to assess the complex interactions of the lung microbiome in response to mixed bacterial infections and their effects in people with CF.


Introduction 51
Cystic Fibrosis (CF), is an autosomal recessive life-limiting disease, characterised by viscid 52 secretions in multiple organ systems due to mutations affecting the CFTR gene, which codes 53 for a cAMP-regulated chloride channel, found on epithelial surfaces, including the airways, showed that a strain of P.nigrescens isolated from a person with CF induced a lower pro-70 inflammatory cytokine expression in CF bronchial epithelial cells than P.aeruginosa,

71
suggesting that the presence of certain Prevotella species in CF lungs may lower the 72 inflammatory response and could therefore be beneficial to the host (6).

78
NF-κB activation occurs via two major signalling pathways, the canonical and the non-79 canonical NF-κB signalling pathways.

129
The minimum amount of bacteria required to provoke a significant response from CFBE41o-130 cells (0-4h, anaerobic conditions) was determined by screening of 3 different P.aeruginosa 131 isolates as described (6). Growth curves of P. histicola and P.aeruginosa under anaerobic 132 conditions revealed no differences in the growth rates between the two species (supplement 133 S1). (37°C, 5% CO 2 , 95% mixed gas). All tissue culture flasks and plates were pre-coated with a 140 1 % PurCol type 1 collagen solution (Nutacon) and passaged as described previously (18).  Student's t-test were employed to assess the null hypothesis. Where more than two groups 196 were compared either one-or two-way ANOVA was utilised to assess differences between  Fig 1E).

219
As we did not observe nuclear p65-DNA binding in response to infection with P.histicola, we time point (Fig 1G and Fig 1H). A significant increase in IL-6 gene expression was observed 225 in CFBE41o-cells at 2 hours in response to infection with P.histicola (p<0.001, Fig 1F).

256
To confirm that TLR2 and TLR5 signalling following P.histicola infection would lead to NF-B 257 activation, CFBE41o-cells were transiently transfected with siRNA targeted against TLR2 or 258 TLR5 and infected with P.histicola. In response to infection we observed reduced levels of

260
but not when cells were transfected with siRNA targeted against TLR2 (Fig 3D). We also did 261 not find any significant reduction in RelB DNA-binding in response to flagellin stimulation in 262 these cells. However, flagellin alone significantly induces p65-DNA binding in CFBE41 cells

263
(p<0.001, Fig 2E), but has no effect on RelB-DNA binding ( Fig 2D). As a control, P.aeruginosa 264 infection in TLR2 and TLR5 transfected cells did not lead to any significant changes in nuclear 265 p65-DNA binding (Fig 2E).

266
Overall, these results show that P.histicola engages with CF epithelial cells via TLR5 and does 267 not engage TLR2 signalling.   Fig 3A), while infection with P.aeruginosa resulted in phosphorylation of

290
IKKα at 60 minutes post infection only (p<0.001, Fig 3A). siRNA knockdown of IKKα resulted 291 in a significant reduction in the DNA-binding abilities of nuclear RelB in response to infection 292 with P.histicola (p<0.001, Fig 3B). As expected, P.aeruginosa infection significantly induced 293 p65-DNA binding (p<0.001, Fig 3C), but P.histicola infection did not (scrambled IKK siRNA 294 and P.histicola, Fig 3C). siRNA IKK in P.histicola infected CFBE41o-had no inducing effect 295 on p65-DNA binding in CFBE41o-, and p65-DNA levels remained significantly lower compared 296 to those in P.aeruginosa infected cells (p<0.001, Fig 3C) in growth rates between the two species (supplement S1), suggesting that any differences

405
Furthermore, we used flagellin derived from Salmonella, which was highly pro-inflammatory,

406
but we cannot exclude that other flagellins (e.g. from P. aeruginosa) may provoke a response 407 of different intensity.

408
As we observed a strong phosphorylation of IKKα induced by P.histicola, we further sought to 409 establish if such IKKα activation was involved in the activation of the alternative NF-B 410 signalling pathway and if this would have any regulatory effect on the canonical pathway.

411
Knockdown of IKKα significantly reduced P.histicola induced activation of the alternative NF-

412
B signalling (RelB-DNA binding), but we saw an induction in p65 (see 'scrambled and 413 P.histicola' in Fig 3C) indicating a role for IKKα in the regulation and suppression of p65.

415
(PHDs) and subsequent polyubiquitination promotes HIF degradation (47 show a significant upregulation of IL-6 expression, a cytokine known for both pro-inflammatory 431 and anti-inflammatory actions (50).