A host anti-amyloidogenic stomach-specific protein inhibits colonization and biofilm formation by adherent invasive Escherichia coli in the colon

TL;DR

A protein from the stomach helps prevent harmful bacteria from forming biofilms in the gut, which is linked to inflammatory bowel disease.

Contribution

The study shows that Gastrokine-1 inhibits biofilm formation and amyloid production by adherent invasive Escherichia coli in the gut.

Findings

Gastrokine-1 does not affect initial colonization by AIEC but is needed for its clearance from the distal gut.

Gastrokine-1 inhibits biofilm formation and amyloid fiber production by AIEC and curli proteins.

AIEC biofilms were found in the distal gut of Gastrokine-1-deficient mice.

Abstract

Gastrokine-1 (Gkn1) is an anti-amyloidogenic host protein secreted into the gut lumen by the stomach. Gut bacteria make functional amyloids to facilitate biofilm formation and biofilms in the gastrointestinal tract are associated with a variety of disorders, including inflammatory bowel disease. Adherent invasive Escherichia coli (AIEC) is a pathobiont that produces amyloids, forms biofilms, and is associated with inflammatory bowel disease. We therefore investigated whether Gkn1 is required to clear AIEC from the gastrointestinal tract by comparing the course of infection in wild-type and Gkn1-deficient (Gkn1−/−) mice. Our findings reveal that Gkn1 does not impact initial colonization by AIEC, but is required for effective clearance of AIEC from the distal GI tract. We also find that Gkn1 inhibits biofilm formation by AIEC and that Gkn1 inhibits the formation of amyloid fibers by the functional E. coli amyloid curli. Furthermore, biofilms of AIEC were evident in the distal gut of Gkn1−/− mice. Together these results indicate that Gkn1 inhibits bacterial amyloid fiber formation, bacterial biofilms, and facilitates clearance of a biofilm forming, IBD-associated, pathobiont from the distal gut. In addition, as the stomach is the sole source of Gkn1, these results implicate the stomach as a source of protection from intestinal biofilms.

Figures

A](#F1)). Both WT and Gkn1^−/−^ mice initially exhibited high levels of fecal LF82 that were not statistically different 2 days and 4 days after the LF82 challenge ([Fig. 1B](#F1)). The LF82 levels in stool began to decrease by day 6 post infection, but there were more LF82 bacteria present in the stool of Gkn1^−/−^ mice compared to WT mice ([Fig. 1B](#F1)). On day 7, the contents from the intestine – ileum, cecum, proximal colon, and distal colon – were collected and assessed for LF82 colonization. There were significantly more LF82 present in the terminal ileum, cecum, proximal colon and dis

A](#F2)). Both WT and Gkn1^−/−^ mice initially exhibited high levels of fecal NRG857c on day 2, 4, and 7 after the NRG857c challenge and these levels were not significantly different between WT and Gkn1^−/−^ mice ([Fig. 2B](#F2)). The NRG857c levels in stool began to decrease on day14 post infection in WT mice, with significantly more AIEC present in the stool of Gkn1^−/−^ mice compared to WT mice on days 14, 21 and 28 post-infection ([Fig. 2B](#F2)). Most WT mice had cleared NRG857c from the feces by 28 days post infection but all the Gkn1^−/−^ mice had persistent NRG857c at this time point.

A](#F3)). Viability of AIEC was not decreased by Gkn1 ([Fig. 3A](#F3)). Visualization of biofilm formation using fluorescent staining confirmed that Gkn1 inhibits biofilm formation by both LF82 and NRG57c ([Fig. 3B](#F3), [C](#F3)). Thus, Gkn1 is sufficient to inhibit biofilm formation by AIEC.

A](#F4)). This fibrillation of CsgA was delayed by substoichiometric levels of Gkn1 in a dose dependent manner (1, 2 and 5uM Gkn1) and was entirely prevented by equimolar (10uM) and higher (20uM) concentrations of Gkn1. Thus, Gkn1 delays, and can entirely prevent, amyloid fiber formation by *E. coli* CsgA peptide.

Full text

INTRODUCTION

Adherent-invasive Eschericia coli (AIEC) is a pathobiont associated with Crohn’s disease (CD)^1,2^. Strains of AIEC isolated from CD patients, including LF82 and NRG857c, can form biofilms as part of their virulence mechanism^3–5^. Formation of biofilms by AIEC requires type IV secretion systems (T4SS) and the presence of extracellular curli fibers^6^. Curli fibers are amyloid fibers formed by the secreted E. coli peptide CsgA^7^. The ability of AIEC to form extracellular amyloids and form biofilms is one reason this pathobiont can persist in the gut and potentially drive inflammation in CD^8^. In addition, biofilm formation can provide resistance to antibiotics, allow intracellular survival of AIEC in macrophages, and allow prolonged survival in the gut lumen^1,9–11^. Thus, understanding host mechanisms capable of preventing AIEC biofilms in vivo may provide insight into targeting this pathobiont for treatment of CD.

Gastrokine-1 (Gkn1) is a protein made abundantly and constitutively in the stomach by differentiated gastric epithelial cells^12^. The stomach lining is the only site of Gkn1 production, where it is secreted along with gastric mucus into the gut lumen^12,13^. Gkn1 is a very stable protein that can be found intact in the lumen of the distal GI tract where it can promote gut health^14^. Mice lacking Gkn1 are otherwise healthy but are susceptible to chemically induced colitis^14^. The function of Gkn1 is not fully understood but Gkn1 can inhibit amyloid fiber formation by the model amyloid Aβ^15^. Given that Gkn1 is found in the lumen of the distal gut, protects against colitis, and inhibits amyloid fiber formation, we investigated whether Gkn1 is important for clearance of the amyloid- and biofilm-forming CD-associated pathobiont AIEC from the distal gut.

METHODS

Animals.

All animal protocols were approved by the University of Notre Dame Institutional Animal Care and Use Committee (protocol #23-12-8242). All experiments were carried out in accordance with the ARRIVE (Animal Research: Reporting of In Vivo experiments) guidelines and in accordance with the NIH Guidelines for the Care and Use of Laboratory Animals. GKN1^+/−^ mice on a C57Bl/6 genetic background were generated by the trans-NIH Knockout-Out Mouse Project (KOMP) and bred and maintained in our facility on that background. Mice were provided food and water ad libitum and kept on a 12/12-h light/dark cycle. Mice were housed in a specific pathogen-free barrier facility under BSL 2 conditions. All cages contained at least one enrichment device in the form of a cardboard tube or “house”. Male and female mice were used as no gender specific effects were observed. WT mice were either Gkn1^+/+^ or Gkn1^+/−^ as no heterozygous phenotype was observed.

Bacterial Strains & Culture

The AIEC LF82 strain was isolated from the ileal mucosa of a CD patient^16^. AIEC NRG857c strain was also isolated from the ileum of a CD patient^17^ and was a gift from Brain Coombes (McMaster University, Ontario, Canada). The E. coli CsgA mutant was produced by the Chapman lab^18^.

AIEC Inoculation of mice and enumeration of bacteria fecal or gut contents

WT or Gkn1^−/−^ mice (8 to 12-weeks of age) were pretreated with 0.1g/ml of streptomycin (200ul/mouse) by gavage 24hr prior to AIEC exposure and then inoculated by gavage with E. coli strain LF82 or NRG857c (1 × 10^9^ CFU) (200ul/mouse). For LF82, fecal samples were collected on days 2, 4 and 6 post-inoculation and the contents of the ileum, cecum, proximal colon and distal colon were collected on day 7. For NRG857c, fecal samples were collected on days 2, 4, 7, 14, 21 and 28 post-inoculation and contents of the ileum, cecum, proximal colon and distal colon were collected on day 28. Fecal samples or gut contents were collected into ice cold PBS or LB and dissociated with a mixer mill, serially diluted and plated onto LB-AMP agar (100mg/ml ampicillin). Counts of CFUs were normalized to the dry weight of the fecal or gut contents.

Recombinant Gastrokine-1 (mGkn1) Production

Mouse Gkn1 was produced using the yeast Kluyveromyces lactis (K. lactis) protein expression system according to manufacturer’s instructions (NEB). Briefly, mouse Gkn1 cDNA without the endogenous secretion signal and an added c-terminal HA epitope tag was amplified by PCR and cloned into the pKLAC2 DNA vector containing the K. lactis protein secretion signal. The linearized plasmid was transformed into competent K. lactis strain GG799 and K. lactis isolates were screened on YCB agar (30mM Tris (pH7.5), 11.7g YCB powder, 20g Bacto agar /L) acetamide plates and clones selected based on secretion of HA-Gkn1 protein. To produce HA-Gkn1 protein, K. lactis was cultured in YPDGal (10g Yeast extract, 20g Bacto Peptone, 2% galactose /L) for 6 days and yeast was removed by centrifugation (10,000g, 20min) and filtration (0.2uM). The supernatant was mixed with agarose beads conjugated to anti-HA antibody (Pierce) overnight (4C). Agarose beads were washed and HA-Gkn1 was eluted using 3M sodium isothiocyanate followed by buffer exchange and concentration using 3KDa spin columns. Purified Gkn1 protein concentration was assessed by A_300_ and confirmed for purity and concentration by SDS-PAGE and staining with colloidal Coomassie.

In vitro biofilm assays:

Biofilm formation was quantified using standard conditions for crystal violet staining^18^. Briefly, AIEC (LF82 or NRG857c) were cultured overnight in LB and diluted (1:200) into M63 media in a 96 well plate and incubated (24hr, 30C) in statice culture with or without Gkn1. The microbial density was measured (A_600_), and the plate was then washed repeatedly to remove planktonic bacteria and air dried. Wells were then stained with 0.1% crystal violet, washed, solubilized in 30% acetic acid and biofilms quantified by A_590_. Biofilms were quantified as the ration of A_590_/A_600_. To visualize biofilms, bacteria were cultured as described above, with or without Gkn1, on glass coverslips (24hr, 30C), washed and stained with Film Tracer SYPRO Ruby biofilm matrix stain (Invitrogen) and visualized by brightfield fluorescence (Leica DM5500).

Measurement of amyloid fiber formation

E. coli curli fiber formation was assessed as described^18^. Briefly, recombinant CsgA monomers (10uM) were incubated (24hr, RT) in excess Thioflavin T (ThT) and formation of amyloid fibers was assessed by fluorescence (excitation 438nm, emission 495nm) at 20-minute intervals.

Assessment of Gkn1 binding to E. coli:

AIEC LF82 (CsgA +ve) or CsgA-ve E. coli were cultured overnight and diluted (1:200) in M63 media and kept in static culture (24hr, 30C). Bacterial cultures were then incubated (1hr, 4C) with Alexa Fluor 488-conjugated Gkn1 and covered with 1% agarose and visualized for binding of Gkn1 to the surface of live bacteria using a Zeiss LSM 980 confocal microscope.

In vivo biofilm studies

WT and Gkn1^−/−^ mice were inoculated by gavage with AIEC NRG857c, as described above. On day 7 after inoculation gut tissues were collected and fixed in Methyl Carnoys (60% methanol, 30% chloroform, 10%acetic acid), processed and embedded in paraffin, and sectioned at 4uM for immunostaining. Tissue sections were deparaffinized, hydrated and incubated (1hr, RT) in blocking buffer (2.5% donkey serum, 1% BSA in TBS-T (Tris buffered saline with 0.1%Tween)). Tissues were then incubated with antibody to NRG857c (E. coli O83 antibody) in blocking buffer overnight (4C), washed and incubated (1hr, RT) in Alexa Flur 488-conjugated anti-Rabbit antibody (Jackson Immunoresearch), washed and mounted in ProLong Gold media with DAPI (Invitrogen). Images were collected with both brightfield fluorescence (Leica DM5500) and confocal microscopy Zeiss LSM 980).

Statistics

For enumeration of bacteria CFU in fecal or gut contents data is presented as mean with SEM and analyzed by Student’s T test (Mann-Whitney) or ANOVA with post hoc Tukey’s test. For in vitro biofilm assays data was analyzed by ANOVA with Dunnett testing. Significance was inferred by *p < 0.05, **p < 0.01, ***p < 0.005, ****p < 0.001. Graphpad prism 10 was used to graph data and calculate statistics.

RESULTS

Gkn1 is required to effectively clear Adherent Invasive E. coli from the distal gut.

Gkn1^−/−^ mice are more susceptible to chemically induced colitis than their WT littermates^14^. This suggests that Gkn1, a stomach specific protein secreted into the lumen of the gut, can protect against inflammation in the distal intestine. The pathobiont AIEC has been implicated in IBD and can cause colitis in mouse models of diseases^24,25^. We therefore tested whether Gkn1 limits distal gut infection and aids in the clearance of AIEC strain LF82 by comparing the course of infection in Gkn1^−/−^ vs. WT mice.

WT & Gkn1^−/−^ mice were pretreated with streptomycin 24 hours prior to infection to facilitate LF82 colonization, as described^19–21^. Mice were subsequently infected by oral gavage with LF82 (10^9^ Colony Forming Units (CFU) and monitored for LF82 levels in their stool (Fig. 1A). Both WT and Gkn1^−/−^ mice initially exhibited high levels of fecal LF82 that were not statistically different 2 days and 4 days after the LF82 challenge (Fig. 1B). The LF82 levels in stool began to decrease by day 6 post infection, but there were more LF82 bacteria present in the stool of Gkn1^−/−^ mice compared to WT mice (Fig. 1B). On day 7, the contents from the intestine – ileum, cecum, proximal colon, and distal colon – were collected and assessed for LF82 colonization. There were significantly more LF82 present in the terminal ileum, cecum, proximal colon and distal colons of Gkn1^−/−^ mice, compared to WT mice (Fig. 1C). In addition, WT mice frequently cleared AIEC from the cecum, proximal colon and distal colon, whereas Gkn1^−/−^ mice did not.

This indicates that WT and Gkn1^−/−^ mice are equally susceptible to initial infection by LF82, but that Gkn1^−/−^ mice are less able to clear LF82 from the distal gut following infection.

LF82 and NRG857c are both AIEC strains isolated from patients with Crohn’s disease^4^. While both strains are closely related and exhibit similar traits, NRG857c exhibits greater within-host fitness and a more robust and persistent infection capacity, compared to strain LF82^22–24^. We therefore investigated whether the requirement for Gkn1 in clearance of distal gut AIEC extends to the more virulent AIEC strain NRG857c. Mice were infected by oral gavage with NRG857c (10^9^ CFU) following the same protocol used for LF82 and monitored for 28 days post infection (Fig. 2A). Both WT and Gkn1^−/−^ mice initially exhibited high levels of fecal NRG857c on day 2, 4, and 7 after the NRG857c challenge and these levels were not significantly different between WT and Gkn1^−/−^ mice (Fig. 2B). The NRG857c levels in stool began to decrease on day14 post infection in WT mice, with significantly more AIEC present in the stool of Gkn1^−/−^ mice compared to WT mice on days 14, 21 and 28 post-infection (Fig. 2B). Most WT mice had cleared NRG857c from the feces by 28 days post infection but all the Gkn1^−/−^ mice had persistent NRG857c at this time point. On day28 the contents from the intestine – ileum, cecum, proximal colon, and distal colon were collected and assessed for NRG857c colonization. There were significantly more AIEC bacteria present in the cecum, proximal colon and distal colons of Gkn1^−/−^ mice, compared to WT mice (Fig. 2C). In addition, WT mice frequently cleared AIEC from the different regions of the colon whereas Gkn1^−/−^ mice did not (Fig. 2C). This indicates Gkn1 does not impact initial infection by the more virulent AIEC strain NRG857c, but that Gkn1 is required to effectively clear NRG857c from the distal gut. Together, these results indicate that a protein made exclusively in the stomach is required for the normal clearance of AIEC from the distal gut.

Gkn1 inhibits biofilm formation by AIEC in vitro

Gkn1 has a BRICHOS domain and, like other BRICHOS domain-containing proteins, Gkn1 inhibits amyloid fiber formation by amyloid-beta^11,14,15^. Some strains of E. coli, including AIEC, can generate functional amyloids through the secretion of curli peptide subunit CsgA, which forms amyloids and facilitates biofilm formation^19^. The ability to form biofilms is a virulence mechanism of E. coli. We observed impaired clearance of AIEC from the distal gut of Gkn1^−/−^ mice. We therefore tested whether recombinant Gkn1 inhibits microbial growth or biofilm formation by AIEC in vitro.

To assess the effect of Gkn1 on AIEC biofilm formation, overnight cultures of LF82 or NRG857c were diluted in M63 media and grown at 30C for 24 hrs in the presence or absence of Gkn1 followed by identification of biofilm formation using crystal violet staining. Recombinant mGkn1 inhibited biofilm formation by both LF82 & NRG857c strains in a dose dependent manner (Fig. 3A). Viability of AIEC was not decreased by Gkn1 (Fig. 3A). Visualization of biofilm formation using fluorescent staining confirmed that Gkn1 inhibits biofilm formation by both LF82 and NRG57c (Fig. 3B, C). Thus, Gkn1 is sufficient to inhibit biofilm formation by AIEC.

Gkn1 binds to CsgA-expressing E. coli and inhibits curli amyloid fiber formation

Our finding that Gkn1 inhibits biofilm formation by AIEC in vitro led us to investigate whether Gkn1 inhibits E. coli amyloid formation. CsgA is the major curli peptide subunit of E. coli and structurally constitutes the amyloid fibers which facilitate biofilm formation^7,18^. Gkn1 inhibits amyloid formation by the model human amyloid peptide Aβ, suggesting that Gkn1 might inhibit E. Coli biofilm formation by inhibiting bacterial amyloid formation. We therefore investigated whether Gkn1 interacts with E.coli in a CsgA dependent manner and whether Gkn1 inhibits curli amyloid formation in vitro.

To assess the ability of Gkn1 to inhibit E. coli curli amyloid fiber formation in vitro, purified CsgA peptide amyloid fiber formation in the presence or absence of mGkn1 was assessed using the ThT fluorescence assay. Incubation of CsgA peptide (10uM) alone resulted in elevated ThT fluorescence indicative of amyloid fiber formation beginning at 3 hours and reaching maximal fluorescence at 8 hours of incubation (Fig. 4A). This fibrillation of CsgA was delayed by substoichiometric levels of Gkn1 in a dose dependent manner (1, 2 and 5uM Gkn1) and was entirely prevented by equimolar (10uM) and higher (20uM) concentrations of Gkn1. Thus, Gkn1 delays, and can entirely prevent, amyloid fiber formation by E. coli CsgA peptide.

To visualize the interaction of E.coli CsgA with Gkn1 in vitro, overnight cultures of curli -ve (curli KO) and curli +ve (LF82) E. coli were incubated (1hr; 4C) with fluorescently tagged recombinant mGkn1 (Alexa Fluor 488) and visualized by confocal microscopy under 1% agarose. Gkn1 was observed decorating the outer surface of curli +ve E. coli, but not curli -ve E. coli, indicating that Gkn1 decorates the surface of E. coli in a CsgA dependent manner (Fig. 4B). Together, these results are consistent with the ability of Gkn1 to bind E. coli and prevent biofilm formation.

Gkn1 inhibits AIEC biofilm formation in vivo

The persistence of AIEC in the distal gut of Gkn1^−/−^ mice and inhibition of E. coli amyloid fiber formation and inhibition of AIEC biofilms by Gkn1 led us to hypothesize that Gkn1 might be important for preventing AIEC biofilms in the distal gut. To test this, WT and Gkn1^−/−^ mice were gavaged with AIEC strain NRG857c (as described) and tissues were collected and fixed in Methyl-Carnoys to preserve mucus structure followed by visualization of AIEC with an NRG857c - specific antibody. In the colon of Gkn1^−/−^ mice, layers of AIEC consistent with biofilms were evident and these were reduced or absent from the distal gut of WT mice (Fig. 5). Confocal images confirmed the biofilm like structure of these AIEC groups in Gkn1^−/−^ mice. These observations suggest that Gkn1 is required to prevent AIEC biofilm formation in the distal gut and that the presence of these biofilms may contribute to the persistence of AIEC in the distal gut of Gkn1^−/−^ mice.

DISCUSSION

The results indicate that Gkn1 is required for clearance of AIEC from the distal gut and for prevention of biofilms by AIEC. One potential mechanism for Gkn1 control of AIEC colonization is inhibition of amyloid formation. Amyloids are key components of E. coli biofilms, and the present work is the first to demonstrate that Gkn1 inhibits bacterial amyloid fiber formation. Prior studies have shown that Gkn1 also inhibits fiber formation by the model human amyloid Aβ^15^. Thus, inhibition of amyloid fiber formation may be a general property of Gkn1, suggesting that Gkn1 may inhibit amyloids and amyloid based biofilms of other microbes, such as Salmonella, Streptococcus, Bacillus, Pseudomonas and others. Consistent with this, antibodies to human amyloid Aβ, that also bind CsgA, inhibit biofilm formation by S. Typhimurium in vitro and in vivo^25^. Inhibition of amyloid with antibody also increased the antibiotic sensitivity of catheter-associated S. Typhimurium, suggesting that Gkn1 might also increase antibiotic sensitivity of biofilm forming pathobionts. In addition to facilitating biofilms, microbial amyloids can activate inflammatory signals by binding toll-like receptors (TLRs) or facilitating PAMP binding to TLRs^26,27^. We did not observe differences in inflammation between WT and Gkn1^−/−^ mice colonized with AIEC, suggesting that Gkn1 binding to AIEC amyloids does not inhibit inflammation in this model. Whether Gkn1 inhibits other bacterial biofilms, amyloid activation of TLRs, or colonization by other gut pathogens remains to be determined.

Gkn1 is constitutively and abundantly made in the stomach where it is secreted into the gut contents^12,28^. Thus, the present results implicate the stomach as an important component in defense against persistent AIEC colonization. A role for the stomach in defense against AIEC has not been described. However, increased abundance of E. coli in the gut microbiome occurs following bariatric surgery^29–31^. It will be interesting to determine whether bariatric surgery decreases Gkn1 protein levels in the distal gut and whether this is associated with colonization by AIEC. The constitutive expression of Gkn1 is inhibited by NSAIDs^14,32,33^. NSAIDs increase colonization and inflammation caused by AIEC^34^. One possibility to be explored is whether NSAIDs promote AIEC colonization by inhibiting Gkn1 expression. Human patients lacking Gkn1 have not been identified and polymorphisms in Gkn1 are not evident in the human genome. However, whether patients have autoantibodies against Gkn1 that block its function, similar to antibodies that block intrinsic factor, has not been explored^35^. It is not clear why production of a protein in the stomach is required for clearance AIEC from the distal gut. It is possible that Gkn1 expression in the stomach is important for protection from other amyloid forming pathobionts that colonize the proximal gut, or that early events in AIEC pathogenesis such as induction of CsgA, begin in the proximal gut and are counteracted by Gkn1.

AIEC is associated with CD, but biofilms are a feature of other gut disorders, including ulcerative colitis, irritable bowel syndrome, peptic ulcers and other GI diseases^36,37^. Gkn1^−/−^ mice are more susceptible to chemically induced ulcerative colitis and exogenous Gkn1 promotes intestinal barrier function, but a role for Gkn1 in other GI disorders has not been investigated^14^. Gkn1^−/−^ mice are resistant to diet induced obesity and have reduced abundance of small bowel Firmicutes, especially Erysipelotrichia, suggesting that Gkn1 may play other roles in the health of the proximal gut by regulating the abundance of commensals^28^. Although biofilms are virulence features of many organisms, commensal organisms also form biofilms in the healthy gut where Gkn1 is present^36,37^. This may reflect the fact that not all biofilms require microbial functional amyloids for formation, such that Gkn1 inhibits amyloid based biofilms but not biofilms that form independent of amyloids.

Gkn1 is highly resistant to degradation and digestion in the stomach and GI tract. Gkn1 is stable, and highly conserved across mammals^13,38^. This suggests that Gkn1 protein might be suitable as a food or supplement to inhibit amyloid based biofilms and colonization of the distal gut by AIEC, or to increase antibiotic sensitivity of AIEC to promote intestinal health in CD and other GI disorders. Lastly, the lack of animal models of in vivo gastrointestinal biofilms has been a barrier to progress in understanding biofilms roles in pathogenesis and in advancing new therapeutic strategies to target biofilms^37^. Gkn1^−/−^ mice may therefore provide an important platform for advancing our understanding of gastrointestinal biofilms.