Engineered E. coli Nissle 1917 for the delivery of matrix-tethered therapeutic domains to the gut

Abstract

Mucosal healing plays a critical role in combatting the effects of inflammatory bowel disease, fistulae and ulcers. While most treatments for such diseases focus on systemically delivered anti-inflammatory drugs, often leading to detrimental side effects, mucosal healing agents that target the gut epithelium are underexplored. We genetically engineer Escherichia coli Nissle 1917 (EcN) to create fibrous matrices that promote gut epithelial integrity in situ. These matrices consist of curli nanofibers displaying trefoil factors (TFFs), known to promote intestinal barrier function and epithelial restitution. We confirm that engineered EcN can secrete the curli-fused TFFs in vitro and in vivo, and is non-pathogenic. We observe enhanced protective effects of engineered EcN against dextran sodium sulfate-induced colitis in mice, associated with mucosal healing and immunomodulation. This work lays a foundation for the development of a platform in which the in situ production of therapeutic protein matrices from beneficial bacteria can be exploited.

Fig. 1. Probiotic-associated therapeutic curli hybrids (PATCH).

a Schematic overview of engineered curli production. Genetically engineered E. coli Nissle 1917 (EcN) with csg (curli) operon deletion (PBP8 strain) containing plasmids encoding a synthetic curli operon capable of producing chimeric CsgA proteins (yellow chevrons with appended bright green domains), which are secreted and self-assembled extracellularly into therapeutic curli hybrid fibers. b CsgA (yellow), the main proteinaceous component of the E. coli biofilm matrix, was genetically fused to a therapeutic domain—in this case, TFF3 (PDB ID: 19ET, bright green), which is a cytokine secreted by mucus-producing cells. The flexible linker (black) includes a 6xHis tag for detection purposes. c Engineered bacteria are produced in bulk before delivery to the subject via oral or rectal routes. A site of colonic inflammation is highlighted in red. d Interaction of PATCH and the colonic mucosa. Inflammatory lesions in IBD result in loss of colonic crypt structure, damage to epithelial tissue, and compromised barrier integrity (left panel, (−) PATCH). The resulting invasion of luminal contents and recruitment of immune cells to the site exacerbates the local inflammation. The application of PATCH (right panel, (+) PATCH) reinforces barrier function, promotes epithelial restitution, and dampens inflammatory signaling to ameliorate IBD activity.

Fig. 2. Production of curli fiber variants from engineered EcN.

a Schematic of quantitative Congo Red (CR)-binding assay (Yellow ovals = E. coli, Orange and green lines = engineered curli fibers, Red triangles = Congo Red) (left). Normalized amyloid production of each EcN variant, as measured by CR-binding assay (right), after induction with arabinose (Ara) at 37 °C in LB media. b Schematic of whole-cell filtration ELISA for the monitoring of modified curli production from bacterial culture (Purple Y shapes with blue dots = Anti-HIS-HRP, Pink stars = HRP Substrates, Yellow stars = HRP products) (left). Relative CsgA production between EcN variants, derived from anti-6xHis antibody-based detection. c–g Scanning electron micrographs of EcN transformed with plasmids encoding various proteins: c GFP d wt-CsgA e CsgA-TFF1 f CsgA-TFF2 g CsgA-TFF3 (scale bar = 1 μm). h Relative TFF3 production (using anti-TFF3 antibody) of induced and non-induced PBP8 library. Data are represented as mean ± standard error of the mean (SEM). (ns) p > 0.05, *p ≤ 0.05, **p ≤ 0.01, ***p ≤ 0.001, ****p ≤ 0.0001, one-way ANOVA followed by Dunnett’s test. Source data are provided as a Source Data file.

Fig. 3. Effects of curli fiber expression on EcN pathogenicity.

a Percent of bacteria that invaded a monolayer of Caco-2 after 2 h of co-incubation with EcN, PBP8 variants, and S. typhimurium. b Bacterial translocation to the basolateral compartment of polarized Caco-2 cells exposed to bacterial library for 5 h. c Epithelial permeability of polarized Caco-2 24 h post-infection, quantified via FITC-dextran (MW 3000–5000) translocation. d IL-8 secretion from the basolateral compartment of polarized Caco-2 cells 24 h post-infection. Data are represented as mean ± SEM. (ns) p > 0.05, *p ≤ 0.05, **p ≤ 0.01, ***p ≤ 0.001, ****p ≤ 0.0001, one-way ANOVA followed by Tukeyʼs test. Source data are provided as a Source Data file.

Fig. 4. Residence time of engineered EcN in the mouse gut and in vivo curli expression.

a In vivo residence time for EcN variants, as measured by CFU counting from fecal samples (Blue = EcN GFP, Red = EcN wt-CsgA, Green = EcN CsgA-TFF1, Pink = EcN CsgA-TFF2, Orange = EcN CsgA-TFF3, Black = PBS Control). b IVIS images of mice that received either PBS or luminescent EcN variants at various time points post-inoculation. c Relative CsgA production from fecal sample analysis of mice 5 days post-inoculation, as measured by ELISA. Data are represented as mean ± SEM. (ns) p > 0.05, *p ≤ 0.05, **p ≤ 0.01, ***p ≤ 0.001, ****p ≤ 0.0001, one-way ANOVA followed by Fischerʼs Least Significant Difference (LSD) multiple comparison. Source data are provided as a Source Data file.

Fig. 5. Immunohistological visualization of engineered EcN strains in tissue sections.

These sections are taken from proximal colons of mice receiving different bacteria. Sectioning protocol was designed to preserve mucus and luminal content. Sections were stained with fluorescently labeled antibodies: anti-E-cadherin (green), anti-Muc2 (red), and anti-LPS (blue). The first column shows bright-field images of the sections. The last column shows an overlay of all stains. The white dotted lines represent the boundary of the epithelium and mucus layers. The leftmost parts represent the epithelium, the center parts represent the mucus layers and the rightmost parts represent the lumen (scale bar = 100 μm).

Fig. 6. Therapeutic efficacy of engineered EcN against a mouse model of DSS-induced colitis.

a Schematic of administration schedule. PBP8 variants (10⁸ CFU) were administered rectally daily. Antibiotics and inducers were administered continuously via drinking water. Weight change (b, N = 9) and disease activity index (c, N = 5–7) were monitored over time, averaged across two independent experiments. Activity index criteria are described in Table 1 (Blue = PBS (DSS−), Red = PBS (DSS+), Green = PBP8 (DSS+), Pink = PBP8 wt-CsgA (DSS+), Orange = PBP8 CsgA-TFF3 (DSS+)). d Colon length at endpoint from two independent experiments (N = 5–7). e Combined DSS colitis histopathology score reflecting severity of inflammation (see Table 2 for details), at endpoint from two independent experiments (N = 5–10). f–j Representative histology of distal colon sections stained with haematoxylin and eosin from each experimental group: f PBS (DSS−) – healthy control, g PBS (DSS+) – colitic, h PBP8 (DSS+), i PBP8 wt-CsgA (DSS+), j PBP8 CsgA-TFF3 (DSS+). Image markers indicate complete loss of crypt and goblet cell depletion (X), immune cell infiltration (I), tissue edema (E), partial loss of the crypts (L), and recovery of crypts (R) (scale bar = 500 μm) k–l IL-6 and IL-17A protein levels from tissue homogenates, as determined by multiplex ELISA. Data are presented as protein concentration per 100 mg of tissue from two independent experiments (N = 6–9). Data are represented as mean ± SEM. (ns) p > 0.05, *p ≤ 0.05, **p ≤ 0.01, ***p ≤ 0.001, ****p ≤ 0.0001. The time course experiments (b and c) were analyzed using two-way ANOVA following by Dunnett’s multiple comparison. One-way ANOVA followed by Fischerʼs LSD multiple comparison was used for d, e, k, and l. Source data are provided as a Source Data file.