Engineering a Novel Probiotic Toolkit in Escherichia coli Nissle 1917 for Sensing and Mitigating Gut Inflammatory Diseases

Abstract

Inflammatory bowel disease (IBD) is characterized by chronic intestinal inflammation with no cure and limited treatment options that often have systemic side effects. In this study, we developed a target-specific system to potentially treat IBD by engineering the probiotic bacterium Escherichia coli Nissle 1917 (EcN). Our modular system comprises three components: a transcription factor-based sensor (NorR) capable of detecting the inflammation biomarker nitric oxide (NO), a type 1 hemolysin secretion system, and a therapeutic cargo consisting of a library of humanized anti-TNFα nanobodies. Despite a reduction in sensitivity, our system demonstrated a concentration-dependent response to NO, successfully secreting functional nanobodies with binding affinities comparable to the commonly used drug Adalimumab, as confirmed by enzyme-linked immunosorbent assay and in vitro assays. This newly validated nanobody library expands EcN therapeutic capabilities. The adopted secretion system, also characterized for the first time in EcN, can be further adapted as a platform for screening and purifying proteins of interest. Additionally, we provided a mathematical framework to assess critical parameters in engineering probiotic systems, including the production and diffusion of relevant molecules, bacterial colonization rates, and particle interactions. This integrated approach expands the synthetic biology toolbox for EcN-based therapies, providing novel parts, circuits, and a model for tunable responses at inflammatory hotspots.

Keywords: E. coli Nissle 1917 (EcN); IBD; TNFα; engineered probiotic; inflammation; nanobodies; nitric oxide.

Figure 1

Design and characterization of the NO detection module. (a) NO-dependent activation from NorR. The NorR transcription factor (represented by the purple chevron) binds its cognate binding site at the promoter pNorVβ (black arrow). When not bound to NO (yellow circles), NorR acts as a competitive inhibitor of its NO-bound form and represses pNorVβ. However, at high NO concentrations, the NO-bound form of NorR is predominant and acts as a positive inducer of pNorVβ. The presence of norR in the inducible operon generates a positive feedback mechanism. Ribosomes are represented in red and the sfGFP gene in green. (b) Construct variants characterized. Our original construct β-1 consisted of sfGFP and norR, preceded by one RBS each, and placed under the control of the optimized promoter pNorVβ. To avoid read-through, we placed a double-terminator at the end of the operon. We normalized the responses of β-1, β-2, and β-3 to a negative control (Neg) and compared to a positive control (WT). Neg consisted of sfGFP and norR genes, preceded by one RBS each, and did not contain any promoter, accounting for the intrinsic leakiness of our module. WT consisted of sfGFP and norR genes, preceded by one RBS each, and placed under the control of the wild-type promoter pNorV.) (c) Time-lapse fluorescence assay for construct characterization. We have grown each construct for 16 h (x-axis) on a microplate reader where green fluorescence (arbitrary units) and measured the culture’s OD₆₀₀ every 15 min. The y-axis represents normalized fluorescence values (sfGFP/OD₆₀₀). Each panel grid represents a different concentration of DETA/NO used to test individual constructs. The DETA/NO gradients we used were 0, 8, 31, 125, 500, and 2000 μM. Each line color represents a construct. Line shadings represent the standard deviation of our biological replicates (n = 3). We performed all measurements with both biological and technical triplicates. Notice that measurements are on log₂ scale to facilitate data visualization. (d) Fold of change for each construct. Each curve represents the fold of change for each construct at T = 8 h along a gradient of NO concentrations. Line shadings represent the standard deviation of our biological replicates (n = 3). Notice that measurements are on the log₂ scale to facilitate data visualization. (e) Rate of change for each construct. The bar plots represent the rate of change for each construct for each DETA/NO change of concentration at T = 8 h. We calculated rates of change as the relative increase in fluorescence (reported as percentages, y-axis) from an initial NO concentration to the next incremental one. We have performed such calculations for each consecutive pair of concentrations (x-axis). Error bars represent the standard deviation of our biological replicates (n = 3).

Figure 2

Design and characterization of the purified anti-TNFα nanobodies. (a) Design of monovalent and bivalent anti-TNFα nanobodies. We linked bivalent nanobody constructs via a short peptide linker (EPKTPKPQPAAA). To characterize the nanobodies, we added a myc-tag and a his-tag to their C-terminal sites. Their expression was under the control of the inducible pBad system, which relies on the addition of l-arabinose. We induced the expression of nanobodies with the pBad inducible system. We harvested monovalent nanobodies via periplasmic extraction and bivalent nanobodies through whole-cell lysis. We purified all nanobodies by immobilized metal anion chromatography. (b) Testing binding capability of purified nanobodies with ELISA. We tested TNFα-binding using an ELISA by capturing the purified nanobodies via their myc-tag. Then, we visualized the binding of nanobodies to biotinylated TNFα with the streptavidin-peroxidase. We measured the absorbance of each well with a plate reader and analyzed the fold change with R studio. (c) Overview of the cell assay used to determine anti-inflammatory properties of purified anti-TNFα nanobodies. We incubated Human THP-1 monocytes with rTNFα and different purified anti-TNFα nanobodies. We assessed the immune response of the monocytic cell line to rTNFα by quantitatively determining the IL1B expression levels with the use of RT-qPCR. The binding of the nanobodies to rTNFα is supposed to inhibit the inflammatory effect observed in untreated but stimulated THP-1 cells. (d) IL1B expression compared to GAPDH in human THP-1 monocytic cell line. Quantitative analysis of the inflammatory IL1B expression levels revealed a decreased immune response of rTNFα-stimulated cells when purified nanobodies were added, compared to untreated cells (labeled as “TNF”, pink line). Adalimumab is an anti-TNFα monoclonal antibody frequently used in the clinic to treat IBD patients and served in this experiment as a positive control.

Figure 3

Design and characterization of arabinose- and NO-induced anti-TNFα nanobodies secretion in E. coliNissle 1917 and E. coliMC1061. (a) Principle of NO-induced nanobody secretion with the hemolysin A secretion system. NO is a small organic molecule able to surpass the double membrane of E. coli. NO binding to the PnorV-β promoter induces the expression of the monovalent nanobody candidate Nb1, which is tagged with a myc- and HlyA-tag. NorR expressions result in a positive feedback loop, enhancing the nanobody expression further. Thanks to the HlyA-tag, the produced nanobodies are secreted by the hemolysin A secretion system in a one-step manner into the extracellular space. (b) Arabinose-induced secretion of monovalent and bivalent nanobodies with E. coliMC1061. Western blot and ELISA analysis revealed successful secretion of functional monovalent and bivalent nanobodies upon overnight arabinose induction in E. coliMC1061. (c) Arabinose-induced secretion of monovalent and bivalent anti-TNFα nanobodies in EcN and MC1061. ELISA analysis shows a successful secretion of functional monovalent Nb1 and bivalent Nb8 nanobodies upon overnight arabinose induction, retaining their TNFα-binding capabilities regardless of the HlyA-tag. (d) NO-induced secretion of monovalent anti-TNFα nanobodies with a single-RBS system in E. coliMC1061. The NO-induced monovalent nanobody secretion was achieved using the single-RBS system (β-1). This yielded a more dynamic response to NO than the previous two-RBS system (β-2) (Figure S16) and a lower baseline expression of monovalent nanobody candidate Nb1 in E. coliMC1061. The absence of the two secretion system components (HlyB and HlyD) resulted, as expected, in no secretion of nanobodies. With increasing NO levels, higher nanobody expression can be observed. A baseline expression in the absence of NO is still present yet weaker than in the β-2 system (Figure S16).

Figure 4

Reaction-diffusion model was evaluated on key parameters. The model’s purpose was to explore which parameters could be essential for the efficacy of our system. We set parameters that were not varied to their default values, except for the sensing threshold, which we decreased by a factor of 10 during simulations as done in a recent study, for visibility reasons. We simulated each parameter configuration 10 times. Line shadings represent the standard deviation. (a) Illustration of the components of the reaction-diffusion model. (b) Relationship between bacterial density and TNFα concentrations. (n = 240). (c) Relationship between sensing threshold and TNFα concentrations. (n = 280). (d) Relationship between nanobody production and bacterial density. (n = 300). (e) Relationship between nanobody production and TNFα concentrations. (n = 380).