LOADING
To effectively develop long-acting drugs based on human-derived GLP-1, we plan to use the intestinal probiotic Escherichia coli Nissle 1917 (EcN) as the chassis, and induce its overexpression via daily beverages to produce sequence-optimized GLP-1. This approach aims to address the short in vivo circulation half-life of the protein from both "qualitative" and "quantitative" perspectives. To fully verify the feasibility of the designed pathway and overcome the challenges of oral bacterial delivery, we have simultaneously established a characterization system for in vitro intestinal environment simulation as turbidostat and hydrogel delivery system. For the precise regulation of EcN's survival status and the guarantee of drug safety, we have also designed essential biosafety modules to both actively and passively regulate the suicide mechanism of E. coli. At this point, four modules: sensing circuits, GLP-1 protein engineering, biosafety and hardware characterization system, have been formally established.
The accumulation of GLP-1 is primarily achieved through sensing circuit modules. We aim to regulate the synthesis and expression of GLP-1 via a dual-signal "AND-gate" pathway(BBa_2581ND93 and BBa_25MR74ON) responsive to protocatechuic acid (PCA)[1] (BBa_25N5FWAK) and bile salts (BS)(BBa_K1962010), along with signal Amplification and Feedback pathways, referred to as the AAF module, enables probiotics colonized in the intestine to sense ingested green tea metabolites and efficiently synthesize GLP-1.
Circuit 1: And-gate Sensing
We use PCA and BS as the "inducer" and "localizer" respectively, and construct an AND-gate logic via the PCA-inducible operon, BS-activated operon elements, and based on the principle of intein trans-splicing[2].
Circuit 2: Amplification & Feedback
Building on the upstream AND-gate signal, we incorporate the Amp30E signal amplification system[3] and glucose uptake negative-response promoter element[4] to regulate the final synthesis of GLP-1 based on feeding and fasting states.
Native human GLP-1 is susceptible to degradation by enzymes such as DPP-IV, resulting in a relatively short in vivo circulation half-life[5]. Therefore, we plan to modify amino acid sequences and introduce non-canonical amino acid (ncAA) substitutions at the cleavage sites, aiming to further explore the enhancement of GLP-1's activity on the basis of improved stability.
To prevent engineered EcN from horizontal gene transfer (HGT) and gene leakage in the intestine or in vitro environments, we have designed an active and passive dual-input OR-gate regulated biosafety Circuit[6] (BBa_25NKC5AV). This allows for flexible regulation of the strain's survival and death through the active administration of substances such as arabinose, or via passive response to in vitro low temperature.
Oral bacterial agents must resist the effects of gastric acid and digestive enzymes to reach the intestine, where they are released and colonize. To this end, based on the pH gradients in the gastrointestinal tract, we have selected hydrogels with good biocompatibility as delivery carriers and conducted relevant tests[7]. Furthermore, we have developed an in vitro turbidostat reactor to determine the activation time of gene circuits[8], and to predict the actual administration needs of agents like green tea for inducing GLP-1 production.
Figure 1. Design and schematic diagram of GlucoXpert
During the implementation of Human Practices, several instructors pointed out that Escherichia coli Nissle 1917 (EcN) is not suitable for long-term intestinal colonization in humans and exhibits low abundance in the gut microbiota. They thus recommended using Lactobacillus, Bacteroides, and other strains as chassis for subsequent engineering. Given the limited timeframe for our wet experiments, and considering that EcN has been widely used in multiple studies for the treatment of gastrointestinal diseases with its safety sufficiently validated, we decided to focus our subsequent experiments primarily on E. coli.
For the specific application of strain selection: First, experiments related to genetic circuits and biosafety require frequent plasmid construction and circuit function verification, with no high demands on protein expression levels. Therefore, we mainly used commercial competent cells such as TOP10 and DH5α as chassis strains for characterization. Second, experiments in the protein module require secretory expression and purification of proteins, and it was necessary to preliminarily verify whether EcN has the ability to synthesize and secrete GLP-1. We thus first used BL21(DE3) as the plasmid expression host, and after verifying the plasmid function, we introduced the plasmid into EcN via electroporation, accordingly characterized EcN's ability to synthesize and secrete modified GLP-1.
In summary, we will complete the characterization of different E. coli strains and preliminarily verify the expression of GLP-1 in EcN. If subsequent experiments proceed smoothly, we will continue to refine the experimental design in EcN and attempt to further advance wet lab experiments using other strains that are more conducive to intestinal colonization and optimize secretory expression.
The AAF Sensing Circuits consist of an AND-gate Sensing and a signal Amplification & Feedback genetic circuit for intelligent and precise regulation of the GLP-1 synthesis process.
To achieve spatiotemporal regulation of GLP-1 synthesis and secretion in engineered strains, we designed and constructed a dual-signal "AND-gate" sensing module, which corely integrates three functional elements: a protocatechuic acid (PCA)-responsive operon, a bile salt (BS)-inducible sensor and a coupled orthogonal split intein-split transcription factor system (the SspGyrB and ECF16 coupled system developed by Professor Baojun Wang's research group) (BBa_25YZWH1A and BBa_259H9ZPUM). The engineered EcN strain with this module can respond to PCA generated via intestinal flora metabolism following preprandial green tea consumption in the intestinal tract and pre-synthesize GLP-1. It then reduces sugar intake by inhibiting appetite and delaying gastric emptying, while also regulating the expression of hormones in the human body, such as insulin and glucagon.
The regulation of the PCA-responsive sensor relies on the inhibitory effect of the repressor protein PcaV (BBa_25OBUNJP) on its cognate promoter PLV (BBa_25Y8JP8F): After phenolic acids in green tea are metabolized by intestinal flora to produce PCA, PCA specifically binds to PcaV, relieving PcaV's inhibition on PLV and thus activating the transcriptional expression of downstream genes[1].
The Bile Salt sensing device is derived from the standardized part submitted by iGEM Dundee 2016, and consists of the bile salt-responsive promoter acrRA (BBa_K318514) and the cognate transcriptional activator gene ramA (BBa_K1962009). It can sense bile salts and activate the expression of downstream genes. Since bile salts in the intestine exhibit the feature of "stable baseline concentration and transient postprandial elevation", this module functions as an ideal signal for confirming the strain's intestinal localization.
The implementation of the AND-gate relies on the trans-splicing mechanism of split inteins. The N-terminal and C-terminal halves of a split intein can be fused and expressed with the halves of the target protein, respectively. Only when both chimeric halves are present simultaneously, the split intein excises itself through self-catalyzed trans-splicing, achieving precise ligation of the extein halves at both ends and forming a fully functional protein[2]. We placed the two halves of the aforementioned orthogonal chimeric system under the regulation of the PCA-responsive promoter and the BS-inducible promoter, respectively. When there is a synergistic response to both PCA and BS signals, the two chimeric SspGyrB halves can undergo trans-self-splicing to reconstitute into the functional transcription factor ECF16. This transcription factor specifically activates its cognate promoter P16 (BBa_25RT9PC8), ultimately driving the expression of downstream GLP-1.
Figure 2. Schematic diagram of the AND-gate Sensing genetic circuit
For the verification of this genetic circuit, we plan to replace the downstream GLP-1 with a GFP reporter. First, we will complete the cloning, construction, gene synthesis, and functional characterization of the two inducible promoters on the pSB3K3 plasmid (provided by Prof. Baojun Wang). We replace the two halves of the downstream SspGyrB-ECF16 chimeric system with GFP reporter gene, while optimizing the expression levels of the regulatory repressor PcaV and transcription factor RamA, ultimately verifying and determining the optimal induction concentrations and timings of PCA and BS for target genes. Subsequently, we will construct an expression cassette consisting of the P16 promoter and GFP on another vector, the pSB4A3 plasmid (provided by Prof. Baojun Wang as well), then co-transform this plasmid together with optimized "AND-gate" logic plasmid into E. coli TOP10. By detecting fluorescence levels, we can verify whether the split intein SspGyrB can achieve efficient splicing, thereby reconstruct the complete transcription factor and ultimately implementing the "AND-gate" logic function.
Considering the low abundance of E. coli in the intestine and the loss of GLP-1 it secretes due to intestinal enzymatic hydrolysis and transmembrane processes, we plan to introduce a signal amplification module to enhance the synthetic expression level of GLP-1. Additionally, we will dynamically regulate the amplification intensity based on feeding and fasting states to reduce the metabolic burden of the strain during its resting phase.
The signal amplification module is constructed using standardized Amp30E elements, with its core components including transcription factors HrpS (BBa_25L4IUPG), HrpR (BBa_K4907021) and the cognate promoter PhrpL (BBa_K4907019). Among these, HrpR and HrpS can form a ultrasensitive high-order co-complex, which acts on the PhrpL promoter containing the σ54 consensus binding sequence, significantly amplifying the expression level of downstream genes[3].
Feedback regulation is achieved through the interaction between the inhibitor HrpV and HrpS. The hrpV gene is placed under the control of the glucose uptake-repressed promoter GURB3-2 (BBa_25TRS3OS), whose activation depends on its Crp-binding site. When glucose concentration in the intestine increases, the concentration of cyclic adenosine monophosphate (cAMP) will decrease, which in turn inhibits the expression of genes downstream of the GURB3-2 promoter[4].
Figure 3. Schematic diagram of the signal Amplification and Feedback genetic circuit
After integrating the above modules, the operating mechanism of this gene circuit is as follows: Upon sensing signals, the upstream AND-gate sensing module activates the expression of HrpS and HrpR proteins. These proteins amplify the signal via the PhrpL promoter, driving the massive synthesis and secretion of GLP-1 to lower blood glucose. During the fasting state, the GURB3-2 promoter is activated, enabling the strain to synthesize HrpV which inhibits the excessive expression of GLP-1 and reduces metabolic stress. After feeding, the increased glucose concentration in the intestine represses the GURB3-2 promoter, leading to decreased HrpV expression. The amplification system remains continuously activated, promoting the large-scale synthesis of GLP-1 and thereby exerting hormone-regulated functions such as blood glucose reduction and weight loss.
Glucagon-like peptide-1 (GLP-1) is a core bioactive peptide that regulates glucose homeostasis and appetite, with significant clinical value in the treatment of diabetes. However, native human-derived GLP-1 faces the critical limitation of rapid enzymatic degradation in vivo, primarily by dipeptidyl peptidase IV (DPP-IV)[9] and neprilysin (NEP-24.11), with a plasma half-life of only about 2 minutes. Such rapid degradation severely limits the therapeutic efficacy of native GLP-1-based drugs, requiring frequent administration to maintain effective plasma concentrations. This not only markedly reduces patient compliance but also restricts the depth and breadth of its clinical applications.
To address this core challenge, we adopt a design strategy inspired by semaglutide, employing site-specific incorporation of non-canonical amino acids (ncAAs) to modify GLP-1. NcAA modification offers remarkable advantages in enhancing resistance to proteolysis, achieved through multiple synergistic mechanisms: (i) introducing ncAAs near susceptible enzymatic cleavage sites to generate steric hindrance; (ii) exploiting the unique chemical properties of ncAAs, which are poorly recognized and degraded by conventional enzymatic systems[10], thereby effectively extending the circulation half-life of GLP-1; (iii) leveraging ncAAs with their specific chemical modification sites (e.g., 5-hydroxytryptophan (5-HTP) can undergo rapid chemoselective azo-coupling reaction (CRACR) to generate multiple functional protein[11]), thereby further enhancing the plasma protein binding capacity of GLP-1.
Based on literature review and expert consultation, we initially select two ncAAs that can be biosynthetically incorporated in vivo: sulfotyrosine (sTyr) [12] and 5-HTP for protein engineering, and design a two-stage workflow consisting of in silico prediction followed by in vitro validation.
In terms of dry lab, we will first perform molecular docking to analyze the interactions between native GLP-1 and NEP-24.11, constructing a 3D model of the binding complex to identify key contact sites. We then screen potential mutation sites and evaluate the impact of substitutions on binding free energy, predicting positions that significantly reduce GLP-1 & NEP-24.11 affinity. In parallel, to ensure that engineered GLP-1 maintains or enhances its binding capacity to the GLP-1 receptor (GLP-1R), we assess the receptor-binding potential of all GLP-1 residues using molecular dynamics simulations and Rosetta-based free energy calculations on the GLP-1 & GLP-1R complex. By integrating both NEP-resistance and GLP-1R affinity analyses, we identify optimal mutation sites and ncAA combinations.
Figure 4. Molecular docking of GLP-1 with NEP-24.11 and GLP-1R, respectively
Building on these computational predictions, we proceed to wet lab experimental validation of ncAA incorporation. First, we select a GLP-1 variant with preliminary amino acid substitutions reported in patents as the starting scaffold[13], and express both native and engineered GLP-1. Second, we employ sfGFP reporter protein harboring amber codon substitutions (BBa_25R2L2YD) to verify the feasibility of site-specific ncAA incorporation in E. coli, while optimizing tRNA sequence features to enhance incorporation efficiency. Third, we purify GLP-1 proteins and evaluate them using GLP-1&GLP-1R His pulldown assays and plasma half-life measurements as the primary criteria for selection. Ultimately, depending on the timeframe, we plan to establish a biosynthetic system for selected ncAA. Through metabolic engineering, we optimize host cell metabolic pathways by enhancing ncAA biosynthesis (push strategy), improving its incorporation efficiency into GLP-1 (pull strategy), and blocking competing metabolic and degradative pathways (block strategy), enabling efficient and specific incorporation of ncAAs into GLP-1 for stable expression.
Figure 5. Site-specific incorporation of non-canonical amino acids via Genetic Codon Expansion
As a live bacterial therapeutic, the biological safety of exogenous engineered bacteria must be a core consideration when they enter the human intestinal tract. To prevent HGT of the strain if it escapes into the external environment, while maintaining the flexibility of the therapeutic regimen, we designed an "OR-gate dual-input responsive" kill-switch. This switch regulates the survival status of the engineered bacteria using two signals including "active arabinose administration" and "passive temperature sensing".
The core of the temperature-responsive module is a sequence-optimized transcriptional regulatory protein TlpA* (BBa_25XF4HZ5) derived from Salmonella. Under low-temperature conditions (<35°C), TlpA* can form a homodimer, which in turn inhibits the activity of the PtlpA promoter (BBa_25RFWS4Y); when the temperature increases, the homodimer dissociates, the inhibitory effect on PtlpA will be relieved, and the promoter regains its transcriptional function. By placing the tlpA gene under the control of the PtlpA promoter, the PtlpA-tlpA* self-repressive regulatory system (BBa_2575N2V6) can be constructed[6].
Based on the aforementioned system, we place the araC gene which encodes the repressor protein of the arabinose operon under the regulation of the PtlpA-tlpA* self-repressive system. Meanwhile, the ccdB gene (BBa_K3512001) which encodes the DNA gyrase toxic protein will be placed under the control of the arabinose-responsive PBAD promoter (BBa_25NOQJ8F). When the strain escapes to a low-temperature environment outside our body, the PtlpA-tlpA* self-repressive system is activated, inhibiting the expression of araC. This reduces the synthesis of the repressor protein AraC, thereby relieving the inhibition of the PBAD promoter. When arabinose is actively administered, arabinose binds to AraC and induces a conformational change of the protein, which relieves the inhibition of PBAD as well. Both conditions can activate PBAD, driving the expression of the downstream ccdB and ultimately inducing the strain to undergo suicide.
To prevent the strain's premature suicide due to leaky expression of CcdB during intestinal colonization, we introduced CcdA (BBa_K3512002), a specific antagonist of CcdB. (The coding genes ccdA and ccdB were both derived from a plasmid of Salmonella provided by Professor Baojun Wang's research group.) By adjusting the expression level of ccdA, the normal function of the strain in the intestine is ensured. Based on the above design, we can construct the following genetic circuit.
Figure 6. Schematic diagram of the biosafety module's OR-gate genetic circuit
The specific experimental design for this circuit proceeds as follows: Firstly, using green fluorescent protein (GFP) as the reporter gene, we systematically characterize the regulatory effect of the PtlpA-tlpA* self-repressive system and optimize culture condition parameters. Subsequently, we construct the complete OR-gate regulatory circuit and verify the validity of the OR-gate logic by detecting fluorescent signal intensity. Then, we amplify the ccdB gene, construct an independent expression vector, and verify the toxicity of CcdB to E. coli. Finally, we assemble the complete kill-switch circuit containing ccdB, characterize the bactericidal efficiency of the arabinose-temperature dual-input system, further optimize the expression levels of the CcdA/CcdB antagonistic system, and ultimately construct a plasmid meeting biosafety requirements.
Engineered bacteria must first withstand the challenge of gastric acid and digestive enzymes before they can achieve precise colonization in the intestinal tract. To this end, we have adopted a hydrogel-based probiotic delivery system that combines biodegradability, excellent biocompatibility, and economic practicality, thereby enabling the release of live biotherapeutics with gastrointestinal pH-responsive properties. Moreover, to further address the limitations of in vivo experiments, we have designed an in vitro turbidostat, which is used to simulate the dynamic balance of intestinal flora and build a dedicated platform for gene circuit testing. For details, please refer to our Hardware page.
Through literature research, we initially identified two cross-linked hydrogel systems: carboxymethyl starch (CMS)–calcium chloride (CaCl2)[7] and sodium alginate (SA)–CaCl2[14]. Subsequently, we prepared simulated gastrointestinal fluids (including simulated gastric fluid, SGF, and simulated intestinal fluid, SIF) and conducted in vitro digestion experiments to determine the gastric acid resistance and intestinal release characteristics of the hydrogels. Building on this foundation, we sought to enhance the hydrogels' ability to support intestinal colonization and adhesion of probiotics.
Given that probiotic adhesion primarily relies on non-specific surface interactions and specific interactions between the mucus layer and receptors, and that thiol-containing compounds can form strong disulfide bonds with mucus glycoproteins, thereby enhancing the intestinal adhesion of probiotics. Thus, we will choose thiol-modified sodium carboxymethyl cellulose (CMC-SH) as a modified cross-linking agent. This agent will enable us to construct microcapsules capable of prolonging the adhesion time of EcN in the intestine, facilitating the development of truly long-acting therapeutics[15].
Figure 7. Schematic diagram of constructing the hydrogel delivery carrier
The turbidostat is designed to simulate the dynamic balance of bacterial density in the intestinal tract, thereby providing a relatively constant initial bacterial load for gene circuit testing and facilitating the characterization of parameters such as "time required for induced expression" and "inducer dosage"[8].
We will construct a simplified in vitro turbidostat and maintained the concentration of engineered bacteria in the reactor at a stable state with the OD value fluctuating slightly within 3 hours, by adjusting devices including the feeding device, reactor, peristaltic pump, and thermostatic magnetic stirrer. Subsequently, we can add a certain amount of inducer using a syringe, collect samples at regular intervals, and evaluate the expression efficiency using GFP fluorescence levels as an indicator, thereby further optimizing the relevant conditions.
Figure 8. Schematic diagram of the in vitro turbidostat
[1] Jiang, S., Chen, H., Chen, S., Chen, N., Yang, H., Duan, Y., Ao, S., Wang, R., Wang, X., Zhang, Y., & Yuan, J. (2025). Genetically Encoded Biosensors for Constrained Biological Functions in Probiotic Escherichia coli Nissle. ACS synthetic biology, 14(1), 296–303.
[2] Pinto, F., Thornton, E. L., & Wang, B. (2020). An expanded library of orthogonal split inteins enables modular multi-peptide assemblies. Nature communications, 11(1), 1529.
[3] Wang, B., Barahona, M., & Buck, M. (2014). Engineering modular and tunable genetic amplifiers for scaling transcriptional signals in cascaded gene networks. Nucleic acids research, 42(14), 9484–9492.
[4] Ding, D., Zhu, Y., Bai, D., et al. (2025). Monitoring and dynamically controlling glucose uptake rate and central metabolism. Nature Chemical Engineering, 2, 50–62.
[5] Yang, D., de Graaf, C., Yang, L., Song, G., Dai, A., Cai, X., Feng, Y., Reedtz-Runge, S., Hanson, M. A., Yang, H., Jiang, H., Stevens, R. C., & Wang, M. W. (2016). Structural Determinants of Binding the Seven-transmembrane Domain of the Glucagon-like Peptide-1 Receptor (GLP-1R). The Journal of biological chemistry, 291(25), 12991–13004.
[6] Rottinghaus, A. G., Ferreiro, A., Fishbein, S. R. S., Dantas, G., & Moon, T. S. (2022). Genetically stable CRISPR-based kill switches for engineered microbes. Nature communications, 13(1), 672.
[7] Feng, P., Bai, X., Ma, X., Kong, H., & Yang, R. (2024). Interfacial-engineered living drugs with "ON/OFF" switching for oral delivery. Nanoscale, 16(28), 13399–13406.
[8] Takahashi, C. N., Miller, A. W., Ekness, F., Dunham, M. J., & Klavins, E. (2015). A low cost, customizable turbidostat for use in synthetic circuit characterization. ACS synthetic biology, 4(1), 32–38.
[9] Gare, C. L., White, A. M., & Malins, L. R. (2025). From lead to market: chemical approaches to transform peptides into therapeutics. Trends in biochemical sciences, 50(6), 467–480.
[10] Zhang, W. H., Otting, G., & Jackson, C. J. (2013). Protein engineering with unnatural amino acids. Current opinion in structural biology, 23(4), 581–587.
[11] Chen, Y., Tang, J., Wang, L., Tian, Z., Cardenas, A., Fang, X., Chatterjee, A., & Xiao, H. (2020). Creation of Bacterial cells with 5-Hydroxytryptophan as a 21st Amino Acid Building Block. Chem, 6(10), 2717–2727.
[12] Chen, Y., Jin, S., Zhang, M., Hu, Y., Wu, K. L., Chung, A., Wang, S., Tian, Z., Wang, Y., Wolynes, P. G., & Xiao, H. (2022). Unleashing the potential of noncanonical amino acid biosynthesis to create cells with precision tyrosine sulfation. Nature communications, 13(1), 5434.
[13] Ma, Y. P., Gu, J. W., & Ma, J. (2019). A GLP-1 mutant and its preparation method and use [Chinese Patent Publication No. CN 110256553 A]. National Intellectual Property Administration of the People's Republic of China (CNIPA).
[14] Chu, L., Deng, Y., Zhang, M., Chen, J., Lian, Y., Chen, B., Xie, L., & Jiang, Y. (2024). The characteristics of sodium alginate-tremella polysaccharide assembled hydrogel induced by calcium ion and its protective effect on Lactobacillus rhamnosus. Food Hydrocolloids, 160, 110732.
[15] Zhang, H., Liu, Z., Fang, H., Chang, S., Ren, G., Cheng, X., Pan, Y., Wu, R., Liu, H., & Wu, J. (2023). Construction of Probiotic Double-Layered Multinucleated Microcapsules Based on Sulfhydryl-Modified Carboxymethyl Cellulose Sodium for Increased Intestinal Adhesion of Probiotics and Therapy for Intestinal Inflammation Induced by Escherichia coli O157:H7. ACS applied materials & interfaces, 15(15), 18569–18589.