Epithelial micro-invasion by Streptococcus pneumoniae induces epithelial-derived innate immunity 1 during colonisation at the human mucosal surface

Epithelial micro-invasion by Streptococcus pneumoniae induces epithelial-derived innate immunity 1 during colonisation at the human mucosal surface 2 3 4 5 Caroline M Weight*1, Cristina Venturini1, Sherin Pojar2, Simon P. Jochems2, Jesús Reiné2, Elissavet 6 Nikolaou2, Carla Solórzano2, Mahdad Noursadeghi1, Jeremy S Brown3, Daniela M. Ferreira2, Robert 7 S Heyderman1 8 9 1Division of Infection and Immunity, University College London, London, United Kingdom, 10 2Department of Clinical Sciences, Liverpool School of Tropical Medicine, Liverpool, United Kingdom, 11 3Department of Respiratory Medicine, University College London, London, United Kingdom 12 13 14 15 16


INTRODUCTION 46 47
Colonisation of upper respiratory tract (URT) mucosa by a range of pathogenic bacteria is a 48 necessary precursor to both disease and onward transmission. Although Streptococcus pneumoniae 49 is a common coloniser of the human nasopharynx, it is estimated to be responsible for >500,000 50 deaths due to pneumonia, meningitis and sepsis in children under five years of age worldwide 1 . 51 52 In Europe and North America, there has been a dramatic effect of pneumococcal conjugate vaccine 53 (PCV) on vaccine serotype (VT) invasive disease, carriage and transmission 2 . However, the 54 emergence of non-VT pneumococcal disease worldwide and the more modest impact of PCV on 55 colonisation in high transmission settings, threaten this success 3 . Control of pneumococcal URT 56 colonisation in humans is not fully understood 4 , and so defining the mechanistic basis for host control 57 of pneumococcal colonisation at the mucosal surface is therefore crucial for the further optimisation 58 of therapeutic interventions which target carriage and transmission. The process of transmission is 59 not fully understood, but mucosal inflammation, potentially enhanced by co-infection with viruses 60 such as Influenza A, has been proposed to mediate bacterial shedding from the nasopharynx 5 . 61 62 Naturally acquired immunity to S. pneumoniae proteins are primarily mediated by mucosal T cells 63 and anti-protein antibodies, controlled by Treg 6-8 . The role of anti-capsule polysaccharide antibodies 64 in naturally acquired immune control remains unresolved 9 . URT epithelium is central to this 65 immunity 10 , binding and transporting antibodies, sensing bacteria via a range of surface and 66 intracellular pathogen-associated molecular patterns (PAMPs) receptors, and rapidly transducing 67 signals to recruit innate and inflammatory immune mechanisms 11,12 . 68 69 Murine models suggest that adherence of S. pneumoniae to the mucosal epithelium may be followed 70 by paracellular transmigration and tight junction modulation 13,14 . In contrast, studies of immortalised 71 epithelial cell monolayers implicate endocytosis of S. pneumoniae, mediated though pneumococcal 72 protein C-polymeric immunoglobulin receptor interactions 15,16 . The relative importance of epithelial 73 endocytosis and paracellular migration in colonisation and invasion remains uncertain 16 , but could 74 influence epithelial sensing of this otherwise extracellular pathogen 14 . For example this may include 75 Nod1 signalling via peptidoglycan 17 and TLR4 signalling via pneumolysin, a pore-forming toxin that 76 induces inflammation and mediates both clearance and transmission in an infant mouse model [18][19][20] . We have used human epithelial cell lines and different strains of S. pneumoniae to further probe 88 these mechanisms and derive an epithelial transcriptome module to interrogate the host response 89 in the EHPC model. We show that pneumococcal colonisation in humans is characterised by micro-90 colony formation, junctional protein association and migration across the epithelial barrier without 91 disease, which we have termed micro-invasion. This pattern of bacterial-host cell association shapes 92 epithelial sensing of S. pneumoniae, which is partially pneumolysin dependent. Together, our data 93 suggest that pneumococcal engagement with the epithelium in the early phases of colonisation may 94 occur without eliciting a marked host response, but as colonisation becomes more established, 95 epithelial sensing of S. pneumoniae enhances innate immunity/ inflammation which we propose 96 promotes clearance.  Figure 1a). Colonisation was frequently characterised 129 by co-association between S. pneumoniae and JAM-A (Figure 1f). This intimate association between 130 inoculated pneumococci and the mucosa during asymptomatic pneumococcal carriage in humans is 131 suggestive of an active engagement process which may influence the outcome of colonisation. 132 133 Further visualization of this bacteria-host interaction by transmission electron microscopy has proven 134 problematic, but we have been able to demonstrate the integrity of the nasal epithelial curette biopsy 135 samples ( Supplementary Figures 1bi and 1bii). We provide evidence of diplococci on the epithelial 136 cell surface and chains or micro-colonies of diplococci that may have been dislodged from the 137 epithelial surface ( Supplementary Figures 1biii and 1biv).   Transmigration across the epithelium (endocytosis or paracellular movement), as measured by the 181 culture of pneumococci from the basal chamber of transwell inserts, was most marked with 23F 182 (Figure 3c, five-fold higher compared to TIGR4, after three hours infection). A similar pattern of 183 epithelial interaction was also seen at one hour suggesting these observations are not simply 184 explained by differential growth.

Endocytosis-related micro-invasion and transmigration 204
To confirm that the pneumococcal endocytosis leads to transmigration across the epithelium, 205 endocytosis was inhibited with dynasore and nystatin 16 . After one hour of infection, cellular uptake 206 of 23F was inhibited by 82% ( Figure 4a) and transmigration across the cell monolayer by 85% ( Figure  207 4b). This inhibition of endocytosis and transmigration was also seen with TIGR4 (data not shown). 208 In agreement with previous studies using A549 and Calu3 cells 30 , we found that viable pneumococci  indicating that an NFkB activation pathway was the dominant driver for responses to these strains. 266 In contrast, cellular responses to TIGR4 and 23F strains revealed enriched binding sites for more 267 diverse transcription factors, suggesting broader molecular pathways by which these strains may 268 influence cellular function. This analysis revealed particular enrichment of binding sites for beta-beta-269 alpha zinc finger superfamily of transcription factors, including KLF4, suggesting that these strains 270 upregulate mitogen activated protein kinase pathways upstream of these transcription factors 32 . 271

272
The sets of genes upregulated by each strain individually were subjected to core and interactome 273 Next, we sought to test whether these epithelial transcriptomic responses to S. pneumoniae were 287 also evident in the 6B EHPC model. We compared nasal transcriptomes sampled at two or nine days 288 after inoculation to pre-inoculation samples. These revealed enrichment of 162 transcripts among 289 individuals who became colonised (differential gene expression in Supplementary Excel file 4). 290 These genes included Claudin 5 and Claudin 17, Defensin β 103A/B, Cadherin 16, Desmocollin 1 291 and Gap junction protein α1, which suggests cytoskeletal re-organisation two days post-inoculation 292 in carriage positive individuals. By day 9 post-inoculation, molecules such as CCR3, matrix 293 metalloproteinase 12 and MHC II molecules were enriched, suggesting activation of the nasal 294 mucosa 34 . However, these genes were not significantly enriched for any immune response pathway 295 annotation perhaps because of the statistical stringency necessary for genome-wide multiple testing 296 and were not epithelial-specific. 297 298 Gene expression modules can successfully detect specific transcriptional programs in bulk tissue 35 . 299 Therefore, as an alternative approach, we sought to target specific epithelial responses in vivo using 300 the in vitro interactome. A direct comparison of transcripts for 6B revealed FOSL-1, an AP-1 301 transcription factor subunit, to be enriched with carriage, which has previously been reported to 302 become activated following pneumococcal challenge in BEAS-2B and HEK293 cells 36   It is widely accepted that colonisation of the URT mucosa by many potentially pathogenic bacterial 318 species, involves transient association with the overlying mucus layer but that adherence to the 319 epithelial surface avoids mucus entrapment and muco-ciliary clearance 14,37 . While evading immunity, 320 bacterial replication then occurs prior to onward transmission to a new host. Even for S. pneumoniae, 321 the mechanisms underlying this transition from acquisition to established colonisation, and then 322 transmission and/or disease, are not well understood. By combining an EHPC model and in vitro 323 human cell culture systems, we show that pneumococcal colonisation leads to epithelial adherence, 324 micro-colony formation and migration across the epithelial barrier without disease, which we have 325 epithelium. This process appears to be association with the formation of epithelial folds different from 338 the membrane ruffles previously reported with conventionally intracellular pathogens such as 339 Salmonella typhimurium 42 . Our study emphasises the active nature of these processes without 340 marked cellular damage or loss of barrier function. Micro-invasion of the epithelium may overcome 341 the inaccessibility of PAMP receptors located either at the epithelial basolateral surface or 342 intracellularly 43,44 . We have found that at early time points, micro-invasion in vitro is also 343 characterised by pneumococcal co-association with JAM-A and β catenin without evidence of barrier 344 dysfunction. In murine models at later time points, TLR-dependent Claudin 7 and Claudin 10 down-345 regulation enhanced pneumococcal translocation across the epithelium has been observed 13 . While 346 this may occur as inflammation becomes more prominent, our data suggest that micro-invasion can 347 happen without barrier disruption and that the epithelium plays an active role in the regulation of 348 these complex host-pathogen interactions 12 . How micro-invasion relates to the risk of invasive 349 pneumococcal disease remains to be determined but as demonstrated by the EHPC model, micro-350 invasion may occur in healthy individuals without symptoms. 351

352
The differential impact of micro-invasion on the epithelial transcriptomic response was striking with 353 considerable enrichment of multiple diverse transcription factors and signalling pathways related to 354 innate immunity with the most invasive pneumococcal strains. The interactome module highlights 355 that although enrichment of gene activation in response to all pneumococcal strains is apparent (the 356 core module), a more diverse range of epithelial gene activation occurs between strains, reflecting 357 specific host-pathogen interactions. Nasal colonisation in murine models is proinflammatory 45,46 , but 358 our in vitro and EHPC experiments with serotype 6B suggests this is not always the case, whereby 359 pneumococcal-host cell adherence, growth, endocytosis and paracellular migration can be 360 established without a marked host response. In the EHPC, clearance occurred around the time the 361 epithelial transcriptomic response was most prominent. Weiser and colleagues argue that the 362 colonising pneumococcus induces a host inflammatory response that mediates clearance, but also 363 promotes nutrient acquisition, mucus production and onward transmission 14,18,47 , which is supported 364 by murine models 48 and epidemiological studies of viral-coinfection 49 . It has been suggested that 365 pneumolysin induces neutrophil influx and degranulation, leading to increased secretions from the 366 nasopharynx, so promoting transmission 5,18 . We therefore suggest that the epithelial innate response 367 resulting from pneumococcal colonisation could promote both clearance and transmission. 368

369
The enrichment of binding sites for the transcription factor KLF4 transcription factors seen in our in 370 vitro data suggests a counter-regulatory process aimed at minimising inflammation. KLF4 plays a 371 role in barrier function, suppression of NFκB-dependent IL-8 secretion, regulation of IL-10 expression 372 via TLR9-MyD88 and Yes-1, in response to S. pneumoniae, which is partially dependent on autolysin 373 LytA 32,50,51 . Thus, by exploiting micro-invasion, the pneumococcus may carefully calibrate the host 374 innate immune/ inflammatory response to promote survival through transmission. 375 376 A range of pneumococcal PAMPs including pneumolysin, may trigger this epithelial sensing 377 process 18,52 . We found pneumolysin to be a prominent inducer of epithelial surface molecule 378 upregulation, cytokine production and transcriptomic inflammatory response in vitro. Following 379 internalisation of the pneumococcus by neutrophils, pneumolysin has been shown to induce ROS 380 following bacterial autolysis, which leads to cellular activation 53 . We speculate that pneumolysin 381 In conclusion, our data implicates the novel finding of micro-invasion during pneumococcal 397 colonisation of otherwise healthy humans, promoting epithelial-derived innate immunity/ 398 inflammation and ultimately clearance. The pathways critical for onward pneumococcal transmission 399 remain to be determined in humans but based on murine models, we propose that this epithelial 400 response may also facilitate transmission. The balance between innate immunity/ inflammation-401 driven transmission and clearance may be further modulated by the frequency of carriage events, 402 pneumococcal strain co-colonisation, viral co-infections and other environmental pressures 1,22,48 . 403 Our approach combining in vitro with in vivo human systems offers a potentially tractable way to 404 further interrogate these processes.     Pneumococcal adhesion and micro-colony formation on the epithelial surface may lead to micro-929 invasion; internalisation of the bacteria and/ or transmigration across the epithelial barrier (micro-930 invasion). The epithelial-derived response is dependent on the subsequent pattern of interactions. 931 Micro-invasion amplifies epithelial sensing and inflammation/ innate immunity, which we postulate 932 leads to immune cell engagement. This process of epithelial sensing inflammation/ innate immunity 933 may enhance both clearance and transmission. Co-association with junctional proteins may facilitate 934 migration across the barrier. 935 n/a n/a n/a n/a n/a 6 Culture (CFU/ml) 0 1.9x10 4 5.7x10 3 3. Culture (CFU/ml) 0 0 0 0 0 Microscopy (counts) 0 0 0 0 0 0 0 0 LytA PCR n/a n/a n/a n/a n/a