The StcE metalloprotease of enterohaemorrhagic Escherichia coli reduces the inner mucus layer and promotes adherence to human colonic epithelium ex vivo

Abstract Enterohaemorrhagic Escherichia coli (EHEC) is a major foodborne pathogen and tightly adheres to human colonic epithelium by forming attaching/effacing lesions. To reach the epithelial surface, EHEC must penetrate the thick mucus layer protecting the colonic epithelium. In this study, we investigated how EHEC interacts with the intestinal mucus layer using mucin‐producing LS174T colon carcinoma cells and human colonic mucosal biopsies. The level of EHEC binding and attaching/effacing lesion formation in LS174T cells was higher compared to mucin‐deficient colon carcinoma cell lines, and initial adherence was independent of the presence of flagellin, Escherichia coli common pilus, or long polar fimbriae. Although EHEC infection did not affect gene expression of secreted mucins, it resulted in reduced MUC2 glycoprotein levels. This effect was dependent on the catalytic activity of the secreted metalloprotease StcE, which reduced the inner mucus layer and thereby promoted EHEC access and binding to the epithelium in vitro and ex vivo. Given the lack of efficient therapies against EHEC infection, StcE may represent a suitable target for future treatment and prevention strategies.

Before adhering to the intestinal epithelium, EHEC must penetrate the mucus layer, which acts as a physicochemical barrier and protects the underlying epithelium from pathogens and foreign antigens. In the colon, the mucus layer is around 400 μm thick and formed of two layers (Johansson et al., 2014;McGuckin, Lindén, Sutton, & Florin 2011).
Whereas the inner layer is dense, firmly attached to the epithelium, and virtually free of bacteria, the outer layer is loose, easily penetrable, and densely colonised by the gut microbiota (Johansson et al., 2008). The mucus layer is comprised of mucin glycoproteins secreted by epithelial goblet cells (McGuckin et al., 2011). Around 20 mucins have been identified so far with the majority bound to the cell surface forming the glycocalyx. In contrast, gel-forming mucin glycoproteins are secreted from goblet cell granules and oligomerize into complex macromolecular structures incorporating water and thereby forming the inner and outer mucus layer (Juge, 2012;McGuckin et al., 2011). In the human intestine, MUC2 is the major secreted mucin of the mucus layer (Johansson et al., 2008).
Although the interaction of EHEC with the intestinal epithelium has been intensely studied, its relationship with the mucus layer remains largely unknown. In this study, we have investigated EHEC binding and its effect on the mucus layer in mucus-producing human intestinal epithelial cell lines and mucosal biopsy samples. We next examined which adhesins were involved in binding to mucus-deficient Caco-2 and mucus-producing LS174T cells. Infections were performed with isogenic EHEC deletion mutants in intimin (Δeae), E. coli common pilus (Δecp), long polar fimbriae (ΔlpfA1), and flagellin (ΔfliC). As no significant differences in binding were detected after 1 hr of infection ( Figure S1), the incubation period was prolonged to 3 hr to allow sufficient time for expression of adherence factors.
Whereas none of the adhesins tested significantly affected EHEC binding to Caco-2 cells, increased binding to LS174T cells was observed in the absence of flagellin (Figure 2a). In addition, the intimin-negative mutant showed reduced binding to LS174T cells, although this did not reach significance. Subsequent immunofluorescence staining to detect actin pedestals demonstrated EHEC A/E lesion formation in LS174T but not in Caco-2 cells after 3 hr of infection. As expected, EHEC Δeae did not form A/E lesions in either cell line (Figure 2b). No pedestal formation was observed in LS174T or Caco-2 cells after 1 hr (data not shown), and wild-type EHEC demonstrated actin recruitment in Caco-2 cells after 6 hr of infection ( Figure 2b).

| Influence of EHEC infection on mucus production
In order to determine the effect of EHEC infection on mucus expression, LS174T cells were incubated with EHEC for up to 6 hr, and gene expression of the two major secreted mucins, MUC2 and MUC5AC, was examined by quantitative reverse transcription polymerase chain reaction (qPCR). No significant changes were observed in EHECinfected versus non-infected cells (Figure 3a). In contrast, immunofluorescence staining demonstrated significantly decreased MUC2 protein levels after EHEC infection (Figure 3b,c). Counterstaining of cell nuclei with 4',6-diamidino-2-phenylindole (DAPI) confirmed that this was not due to cell detachment ( Figure 3b). In addition, MUC5AC was only produced by a small subset of cells, and staining intensity was not influenced by EHEC infection (Figure 3d,e).

| StcE decreases MUC2 protein levels and enhances EHEC binding to colonic epithelium
Previous studies have identified a secreted EHEC metalloprotease (StcE) with mucinase activity for human saliva (Grys, Siegel, Lathem, & Welch, 2005). In order to investigate whether StcE affected intestinal MUC2 levels during EHEC infection, we generated an isogenic stcE FIGURE 1 EHEC binding to mucus-producing LS174T and mucusdeficient HT-29 and Caco-2 cells. (a) Immunofluorescence staining for MUC2 (green) and cell nuclei (blue). Bar =10 μm. (b) Adherence of EHEC TUV 93-0, 85-170 and Sakai after 1 hr of infection. Adhesion was determined by counting colony-forming units and is expressed as percentage of cell-bound bacteria relative to the inoculum. ***p < 0.001 versus Caco-2 and HT-29 cells deletion mutant in strain TUV 93-0 by Lambda Red recombination. As shown in Figure 4a and b, deletion of stcE significantly impaired reduction of MUC2 levels in EHEC-infected LS174T cells. This was restored to wild-type levels after complementation with StcE (Figure 4a,b). In contrast, complementation with catalytically inactive StcE (E447D) did not exhibit any effect, and MUC2 levels were comparable to those of the deletion mutant (Figure 4a,b). In addition to immunofluorescence staining, StcE-dependent MUC2 reduction was confirmed by FIGURE 2 Involvement of EHEC adhesins in binding to Caco-2 and LS174T cells. (a) Adherence of wild-type (wt) and adhesindeficient EHEC strains after 3 hr of infection. Adhesion was determined by colony-forming unit counting and is expressed as percentage of cell-bound bacteria relative to the inoculum. ***p < 0.001 versus wt. (b) Fluorescent actin staining to identify A/E lesion formation. Caco-2 and LS174T cells were infected with EHEC 85-170 wt or Δeae for 3 or 6 hr (Caco-2 for 6 hr) and stained for actin (green) and E. coli (red). Inserts in top right corner show enlarged image areas containing EHEC bacteria with and without actin pedestals (LS174T wt and Δeae, respectively). Bar = 5 μm To confirm the effect of StcE on mucus production and EHEC adherence in a biologically relevant model, we performed in vitro organ culture (IVOC) of human colonic biopsy samples (Lewis et al., 2015).
Whereas the epithelium was covered by a firmly adherent inner mucus layer in non-infected tissue samples, diminished MUC2 levels and exposure of underlying crypt-associated goblet cells was evident after 8 hr of EHEC infection (Figure 5a,b). In contrast, incubation with EHEC ΔstcE resulted only in marginal disruption of the mucus layer leading to the appearance of "dark holes" devoid of MUC2 staining. Attenuated mucus reduction by EHEC ΔstcE was restored to wild-type levels after complementation with wild-type but not catalytically inactive StcE ( Figure 5a,b). To further determine the impact of decreased mucus levels on EHEC adherence to the biopsy epithelium, we removed the mucus layer after the IVOC and quantified adherent EHEC by immunofluorescence staining. As shown in Figure 5c, EHEC bacteria were predominantly attached to the intercrypt surface epithelium, and no adhering bacteria were detected in non-infected control samples.
Quantification of adherent bacteria revealed a significant decrease in epithelial binding of EHEC ΔstcE compared to the wild-type and complemented strain (Figure 5d).

| DISCUSSION
EHEC is a colonic pathogen which binds tightly to human intestinal epithelium by forming A/E lesions (Lewis et al., 2015). To do so, the bacteria must first interact with and penetrate the thick mucus layer overlying the colonic epithelium (Johansson et al., 2014;McGuckin et al., 2011). How this happens is still largely unknown due to a paucity of suitable animal models (Ritchie, 2014). In addition, most human cell line models used to study EHEC infection (e.g. T84, Caco-2, and HT-29 colon carcinoma cells) are devoid of a MUC2-containing mucus layer  EHEC StcE reduces mucin levels and promotes A/E lesion formation on LS174T cells. Cells were infected with wild-type TUV 93-0 (wt), an isogenic stcE mutant (ΔstcE), ΔstcE complemented with wild-type (comp), or catalytically inactive StcE (inact), or left non-infected (NI) for (a-d) 6 hr or (e) 3 hr. (a) MUC2 expression was determined by immunofluorescence staining, bar = 5 μm or (c) agarose gel electrophoresis and Western blotting and quantified by integrated density measurement (b and d, respectively). Means with different letters are significantly different (p < 0.001 for b, p < 0.01 for d). n = 2 in duplicate for c and d. (e) A/E lesion formation was assessed by immunofluorescence staining and expressed as percentage of bacteria associated with actin pedestals relative to the total number of adherent bacteria, **p < 0.01, *p < 0.05 produce MUC2 glycoproteins (van Klinken et al., 1996), we found that EHEC O157:H7 prototype strains adhered better to this cell line compared to mucus-deficient Caco-2 and HT-29 cells. A similar phenotype has recently been reported for the related A/E pathogen enteropathogenic E. coli, which demonstrated increased binding to LS174T versus HT-29 cells (Walsham et al., 2016). In addition, adhesion of Salmonella enterica to mucus-producing HT-29 MTX cells was significantly increased compared to mucus-deficient HT-29 and Caco-2 cells (Gagnon, Zihler Berner, Chervet, Chassard, & Lacroix, 2013). Although where deletion of flagellin resulted in induction of the type III secretion system (Soscia, Hachani, Bernadac, Filloux, & Bleves, 2007). Furthermore, deletion of long polar fimbriae enhanced EHEC adherence to human intestinal explants suggesting cross talk between bacterial adhesion systems (Fitzhenry et al., 2006). Alternatively, the lack of motility associated with these mutants may lead to enhanced binding or retention in the mucus layer (Rossez, Wolfson, Holmes, Gally, & Holden, 2015). It was further noted that EHEC A/E lesion formation in LS174T cells occurred earlier than in Caco-2 cells, thereby explaining the observed dependency on intimin for adherence to this cell line.

Whether enhanced EHEC binding and A/E lesion formation in
LS174T cells is ultimately linked to mucus production or associated with expression of other cell surface receptors remains to be established. In addition, the role of other EHEC adhesins in binding to human intestinal mucus should be investigated.
In the second part of this study, we investigated the influence of EHEC infection on mucus production. Previous reports on the murine A/E pathogen Citrobacter rodentium demonstrated dynamic changes in mucus thickness during infection (Gustafsson et al., 2013). During the acute phase, mucin expression and goblet cell numbers were reduced showing EHEC bacteria (red) and cell nuclei (blue). Shown are merged colour images of both fluorescence channels. c = crypt; bars = 20 μm. (d) Quantification of adherent bacteria per mm 2 surface area. ***p < 0.001, *p < 0.05. n = 4 in triplicate in the distal colon, and this was dependent on the host adaptive immune response (Bergstrom et al., 2008;Lindén, Florin, & McGuckin, 2008). In contrast, increased mucus secretion at later stages of infection promoted expulsion of bacteria into the gut lumen and clearance of infection (Bergstrom et al., 2010;Gustafsson et al., 2013). The protective role of the mucus layer during Citrobacter infection was further demonstrated by studies in Muc2-deficient mice, which exhibited increased weight loss and mortality due to elevated pathogen burden at the mucosal surface (Bergstrom et al., 2010). Similar to Citrobacter, infection with atypical but not typical enteropathogenic E. coli increased mucus production in rabbit ileal loops and mucin-producing HT-29 MTX cells, which promoted bacterial growth in the absence of luminal flow (Vieira et al., 2010).
Apart from MUC2, LS174T cells also secrete MUC5AC, which is usually expressed in the stomach but also associated with colon cancer (Bartman et al., 1999;van Klinken et al., 1996). In our studies, no changes in MUC2 and MUC5AC gene expression were detected in EHEC-infected LS174T cells. This is in contrast to previous studies demonstrating increased MUC2 gene expression in HT-29 cells infected with EHEC (Xue et al., 2014). This discrepancy is likely related to the difference in cell line models, and as HT-29 cells do not express MUC2 glycoproteins (Huet et al., 1995;Walsham et al., 2016), the relevance of this finding remains questionable.
In contrast to the gene expression results, MUC2 protein levels were significantly decreased in EHEC-infected LS174T cells. This is in agreement with a recent study reporting reduction of the MUC2-containing inner mucus layer in EHEC-infected colonic organoids . One likely explanation for this phenotype is the production of mucus-degrading enzymes, which has been reported in enteropathogens (McGuckin et al., 2011). Shigella flexneri and enteroaggregative E. coli secrete the mucin-degrading serine protease Pic (Henderson, Czeczulin, Eslava, Noriega, & Nataro, 1999), and

Vibrio cholerae produces the mucinolytic metalloproteases Hap and
TagA (Silva, Pham, & Benitez, 2003;Szabady, Yanta, Halladin, Schofield, & Welch, 2011). Interestingly, the latter has high sequence homology to the StcE zinc metalloprotease from EHEC, which is located on the pO157 virulence plasmid and secreted by a type II secretion system (Lathem et al., 2002). StcE has been shown to modulate the host immune response by cleaving C1 esterase inhibitor and CD43 and CD45 neutrophil glycoproteins (Lathem et al., 2002;Szabady, Lokuta, Walters, Huttenlocher, & Welch, 2009) and also reduces the viscosity of human saliva by degrading mucin 7 and glycoprotein 340 (Grys et al., 2005). However, recombinant StcE did not show any activity against purified human MUC2 (Grys, Walters, & Welch, 2006), and infection of human colonic organoids with an stcE mutant only resulted in a slight decrease in mucus reduction compared to the wild-type strain . Here, we found that deletion of stcE significantly diminished reduction of mucin levels during EHEC infection of LS174T cells. As catalytic inactivation of StcE had the same effect, it is likely that StcE lowers mucin levels by MUC2 degradation. However, other mechanisms related to mucin glycoprotein synthesis and secretion cannot be excluded at this stage. Diminished mucus reduction in ΔstcE-infected cells was accompanied by a decrease in A/E lesion formation despite adhesion levels comparable to the wild-type strain. Our findings confirm previous results in HEp-2 cells demonstrating reduced actin pedestal formation but not overall adherence by ΔstcE versus wild-type EHEC (Grys et al., 2005). HEp-2 cells do not secrete a mucus layer, and the authors suggested that StcE promotes intimate adherence by cleaving cell surface mucins and thereby exposing host cell receptors. To evaluate our findings in a physiologically relevant setting, we used IVOC of human colonic biopsies to investigate the impact of StcE on mucus production and EHEC adherence ex vivo. Previous studies from our laboratory have shown that EHEC binds to human colonic biopsy epithelium by forming A/E lesions (Lewis et al., 2015). Similar to our results in LS174T cells, reduction of the inner MUC2-containing mucus layer in IVOC was dependent on the presence of catalytically active StcE. However, lack of StcE caused a more pronounced effect on EHEC binding than in the cell line model with an almost complete loss of adherence to the colonic epithelium. It is likely that infection kinetics, environmental cues or host response may fine-tune expression and/ or activity of StcE and other EHEC mucinases, and this may explain the differences in StcE-mediated mucin reduction observed in IVOC and colonic organoids .
Notably, some decrease in mucus levels independent of StcE was evident in LS174T cells and colonic biopsies, suggesting the activity of other mucinolytic enzymes during EHEC infection. Apart from StcE, another type II-secreted zinc metalloprotease has recently been identified in intestinal and extraintestinal E. coli strains (Nesta et al., 2014).

Although SslE facilitates mucus penetration of extraintestinal E. coli
in HT-29 MTX cells and enhances bacterial adhesion to the epithelial surface (Valeri et al., 2015), this protein is rarely detected in EHEC and therefore unlikely to contribute to its pathogenesis (Feng et al., 2007;Nesta et al., 2014

| Quantification of adherence
Infected cells were lysed with 1% Triton X-100 in PBS for 15 min. Cell lysates were serially diluted in PBS, and appropriate dilutions were plated on LB agar. Plates were incubated overnight, and colonyforming units were counted the next day.

| IVOC of endoscopic biopsies
This study was performed with approval from the University of East

| Bacterial mutagenesis and complementation
An stcE mutant in strain TUV 93-0 was generated by Lambda Red recombination (Datsenko & Wanner, 2000). Briefly, the helper plasmid pKD46 carrying the Lambda Red recombinase was electroporated into TUV 93-0, and recombinant bacteria were selected on LB agar containing 100 μg/ml ampicillin. Subsequently, the kanamycin resistance cassette from plasmid pKD4 was polymerase chain reaction (

| Quantitative reverse transcription polymerase chain reaction
Total cellular RNA was extracted using the Promega SV RNA extraction kit according to the manufacturer's instructions. RNA concentration and purity was determined using a NanoDrop ND-1000 spectrophotometer (Fisher Scientific), and RNA integrity was confirmed by agarose gel electrophoresis. For cDNA synthesis, 1 μg RNA was reverse-transcribed using qScript cDNA supermix (Quanta BioSciences). Quantitative PCR was carried out using SYBR Green JumpStart Taq ReadyMix (Sigma) and an ABI7500 Taqman lightcycler (Applied BioSciences). Primers were purchased from Sigma-Genosys, and the following sequences were used for amplification of mucin

| Statistics
All data are shown as means ± standard error of the mean of three independent experiments performed in duplicate unless indicated otherwise. Statistical analysis was performed using GraphPad Prism software (version 5). Student's t test or one-way analysis of variance with Tukey's multiple comparisons test was used to determine differences between two or multiple groups, respectively. For biopsy data, the nonparametric Kruskal-Wallis test with Dunn's post-test was used to analyse multiple groups. A p value of <0.05 was considered significant.

SUPPORTING INFORMATION
Additional Supporting Information may be found online in the supporting information tab for this article.