Introduction

Colorectal cancer (CRC) remains one of the most prevalent and deadly malignancies worldwide, with its immunosuppressive and hypoxic tumor microenvironment (TME) being a major barrier to effective treatment. A key feature of this hostile TME is the dominance of tumor-associated macrophages (TAMs) polarized to the M2 phenotype, which suppress immune responses and promote tumor progression, rather than the tumor-fighting M1 phenotype¹.

Our group, SUSTech-Med, addresses this critical challenge by pioneering a novel bacterial immunotherapy. We leverage the excellent safety profile of the probiotic bacterium Escherichia coli Nissle 1917 (EcN) ² and the unique ability of cyclic diguanylate monophosphate (c-di-GMP) to assist selective colonization of EcN in hypoxic tumor regions. We engineered EcN to become a living therapeutic platform that locally produces and delivers c-di-GMP, a potent bacterial second messenger known to activate the innate immune STING pathway in host cells³. By reactivating the immune system from within the tumor, we aim to transform immunosuppressive tumors into immunoreactive ones³, offering a new strategic avenue for CRC therapy.

Figure 1: Our Novel Bacterial Immunotherapy for CRC

Objectives

Design and Construct

To genetically engineer EcN with two novel plasmid systems:

• pUCP20-GmR-P_fnrS-yedQ: A hypoxic-response plasmid containing the dgcQ (yedQ) gene under the control of the P_fnrS promoter for tumor-specific production of c-di-GMP.
• pTBR2iB: A "suicide" safety plasmid based on a redesigned LuxI-LuxR quorum sensing circuit, where the AHL synthase LuxI is replaced by the phospholipase A (PhlA) gene. This allows for externally inducible bacterial lysis upon addition of AHL.

In Vitro Validation

To functionally validate our engineered bacteria by demonstrating:

• Hypoxia-induced production of c-di-GMP.
• Assessment of Biofilm Formation in Engineered Escherichia coli
• AHL-induced bacterial lysis and subsequent c-di-GMP release.
• Successful polarization of macrophages from the M2 to the M1 phenotype in co-culture assays via the STING pathway.

Modeling and Prediction

To develop quantitative mathematical models that simulate the spatial diffusion of the inducer (AHL) and its relationship to c-di-GMP release and macrophage polarization, aiming to predict the optimal therapeutic dosage.

Hardware Development

To design a microfluidic gut-on-a-chip device to mimic the complex human intestinal TME for more physiologically relevant future testing, incorporating fluid flow and peristalsis.

Engineering

Our therapeutic strategy hinges on the targeted delivery of the potent immune stimulant c-di-GMP directly within the immunosuppressive TME of CRC. This was achieved through a multi-layered engineering strategy: first, we genomically engineered the EcN chassis to introduce essential nutrient auxotrophies, creating a robust "tumor-restricted survival" mechanism; second, we functionally validated a key hypoxia-responsive promoter to ensure precise spatial control of therapeutic production; and finally, we introduced two novel plasmid systems into this engineered chassis:

• A c-di-GMP production plasmid for tumor-specific synthesis of the immunostimulant.
• A "suicide" safety plasmid for externally inducible bacterial lysis and payload release.

This dual-system approach ensures precise spatial control of immune activation within the hypoxic tumor and provides a crucial biocontainment mechanism.

Genomic Engineering of the EcN Chassis for Tumor-Restricted Survival

The core objective of this genomic modification is to create a biocontainment strategy that restricts bacterial proliferation exclusively to the nutrient-rich TME.

• Method: λ-RED homologous recombination.
• Target Genes: The thyA gene, encoding thymidylate synthase, and the dapA gene, encoding dihydrodipicolinate synthase, were sequentially knocked out.
• Resulting Phenotype: The resulting ΔthyAΔdapA strain is auxotrophic for thymidine and diaminopimelic acid (DAP), which are essential for DNA synthesis and cell wall integrity, respectively.
• Expected Function: The TME is characterized by an abundance of metabolic byproducts, including high concentrations of thymidine and DAP, while normal tissues lack sufficient levels of these nutrients. This metabolic containment ensures that the engineered EcN can only replicate within the TME, preventing uncontrolled proliferation in normal tissues and providing a fundamental safety layer for our live biotherapeutic platform.

Functional Validation of the P_fnrS Hypoxia-Responsive Promoter

The core objective of this experiment was to empirically confirm the anaerobic inducibility of the P_fnrS promoter prior to its integration into the therapeutic plasmid.

• Reporter System: A pGL3-Basic vector encoding firefly luciferase.
• Assembly: The P_fnrS promoter was cloned upstream of the luciferase gene using Gibson Assembly to create the pGL3-P_fnrS reporter plasmid.
• Validation Assay: The constructed plasmid was transformed into E. coli, and luciferase activity was measured under both aerobic and anaerobic conditions.
• Result: Normalized luminescence revealed a 2.78-fold increase in promoter activity under anaerobic compared to aerobic conditions, a statistically significant (p < 0.01) induction.
• Conclusion: The data functionally validates P_fnrS as a reliable hypoxia-responsive element, confirming its suitability for driving tumor-specific expression of our therapeutic payload.

Construction of the c-di-GMP Production Plasmid

The core objective of this plasmid is to enable high-level production of c-di-GMP exclusively in the hypoxic conditions of the tumor.

• Backbone: pUCP20.
• Hypoxia-Responsive Promoter: The P_fnrS promoter was chosen to drive expression. This native E. coli promoter is strongly activated under low-oxygen conditions by the FNR (fumarate and nitrate reduction) protein, ensuring tumor-specificity.
• Effector Gene: The dgcQ (yedQ) gene, encoding diguanylate cyclase (DgcQ), was inserted downstream of P_fnrS. This enzyme catalyzes the synthesis of c-di-GMP from two molecules of GTP.
• Assembly: The P_fnrS promoter was precisely placed upstream of the dgcQ gene using Gibson Assembly, replacing the original inducible promoter (e.g., P_lac) used in initial cloning stages.
• Expected Function: Upon oral administration, engineered EcN bacteria colonize the CRC TME. The hypoxic conditions trigger FNR-mediated activation of the P_fnrS promoter, leading to robust expression of DgcQ enzyme and consequent intracellular accumulation of c-di-GMP.

Figure 2: Plasmid Profile of pUCP20-GmR-P_fnrS-yedQ

Construction of Plasmids Inducing Bacterial Lysis and Vector Release

To control the release of c-di-GMP and add a crucial layer of safety, we implemented an inducible lysis system based on a redesigned quorum-sensing circuit.

• Backbone: pUC18.
• Engineered LuxI-LuxR System: The native luxI gene (AHL synthase) was removed to eliminate background AHL production. The circuit retains:
• The luxR gene, constitutively expressed, encoding the AHL receptor protein.
• The P_luxI promoter, which is activated by the LuxR-AHL complex.
• Lysis Gene: The gene encoding Phospholipase A (PhlA) was inserted downstream of the P_luxI promoter. PhlA hydrolyzes phospholipids in the bacterial cell membrane, leading to rapid lysis and content release.
• Inducer: Acyl-Homoserine Lactone (AHL) is supplied exogenously. It diffuses into the bacteria, binds to LuxR, and the resulting complex activates the P_luxI promoter, driving PhlA expression.
• Expected Function: After allowing time for bacterial colonization and c-di-GMP production, a predefined dose of AHL is administered. AHL enters the bacteria, binds LuxR, and triggers the expression of PhlA, causing controlled lysis of the engineered EcN and the release of stored c-di-GMP into the TME⁴.

Figure 3: Plasmid Profile of pTBR2iB

Chassis Transformation and Combined System Workflow

The two plasmids were co-transformed into the EcN chassis, selected for its excellent safety profile, probiotic nature, and innate ability to colonize hypoxic niches like tumors.

The overview of the therapeutic workflow is as follows:

• Administration & Colonization: Engineered EcN is administered orally and colonizes the hypoxic regions of the colorectal tumor.
• Production: Hypoxia triggers the P_fnrS promoter on the production plasmid, leading to intracellular c-di-GMP accumulation. c-di-GMP also enhances biofilm formation, localizing the bacteria to the tumor site⁵.
• Lysis and Release: Exogenously added AHL (e.g., via local injection) activates pTBR2iB / pUC18-luxR-PhlA, causing bacterial lysis and the release of c-di-GMP (and other PAMPs like LPS).
• Immune Activation: Released c-di-GMP diffuses into tumor-associated macrophages (TAMs), activating the STING pathway, promoting repolarization from M2 to M1 phenotype, and initiating a robust anti-tumor immune response.
• Clearance: The "suicide" system ensures eventual clearance of the bacterial population, addressing safety concerns.

Modeling

The primary objective of the dry lab component is to establish a mathematical model that quantitatively describes the dynamic relationship between the concentration of the externally added inducer AHL and the polarization ratio of macrophages in an in vitro experimental system. The model aims to predict the optimal AHL addition strategy to maximize immune activation (M1 polarization) while ensuring the complete clearance of engineered bacteria, providing theoretical guidance and optimization direction for subsequent experiments.

To achieve this goal, the model is divided into two core parts:

• Spatial Diffusion Model: Simulates the three-dimensional spatial diffusion process of AHL in an in vitro culture system (e.g., a 96-well plate).
• Kinetic Model: Describes the kinetics of AHL activating the "suicide" circuit, engineered bacteria lysis and release of c-di-GMP, and c-di-GMP activating the STING pathway and driving macrophage polarization.

Spatial Diffusion Model of the Inducer

We assume that initially, the engineered bacteria are primarily sedimented at the bottom of the culture plate, while macrophages are uniformly suspended in the medium. Exogenously added AHL begins diffusion from the surface of the medium. This process follows Fick's Second Law of Diffusion⁶.

Figure 4: Spatial Diffusion Model

Kinetic Model of Pathway Activation and Polarization

This part of the model describes the kinetics of several key biochemical events:

• Degradation of AHL: We assume AHL undergoes natural degradation, following first-order kinetics;
• AHL-Induced Bacterial Lysis and c-di-GMP Release: AHL binds to and activates the LuxR protein on the suicide plasmid, initiating the expression of phospholipase A (PhlA), leading to bacterial lysis and the release of c-di-GMP. We use the Hill Function to describe this activation process and the subsequent c-di-GMP release rate.
• c-di-GMP-Mediated Macrophage Polarization: Released c-di-GMP enters the macrophage cytoplasm, activates the STING pathway, and promotes the polarization of M2 macrophages to the M1 phenotype. The Hill function is also used to describe the polarization ratio.

Figure 5: Kinetic Model of Pathway Activation and Polarization

Hardware

A significant challenge in bacterial therapy research is the inability of traditional 2D cell cultures and even animal models to fully recapitulate the complex, multi-layered physiological environment of the human intestine and its TME. To bridge this gap and enhance the translational relevance of our findings, we designed and fabricated a sophisticated three-layer microfluidic gut-on-a-chip device.

The primary purposes of this hardware are:

• To simulate the complex physiological environment and multi-layered fluidic dynamics of the human intestine.
• To provide a more physiologically relevant in vitro platform for testing our EcN, observing its interaction with immune cells (Macrophages), and evaluating the efficacy of the AHL-induced lysis strategy.
• To mimic intestinal peristalsis, a critical mechanical factor influencing bacterial distribution and tissue interaction⁷.

Device Design and Architecture

Our chip features a multi-layered structure designed to mimic key aspects of the intestinal TME:

• Material: The main material of the chip is PDMS (Polydimethylsiloxane), chosen for its excellent optical clarity, gas permeability, and biocompatibility.
• Layer 1 - Vascular Simulation Layer (Top Layer):

(1) Function: Simulates the blood flow environment.

(2) Operation: Culture medium containing various substances (e.g., nutrients, the inducer AHL) is perfused through this channel to simulate intravenous drug delivery or systemic distribution.

• Layer 2 - Liquid Extraction & Monitoring Layer (Middle Layer):

(1) Function: Serves as an interstitial space and sampling port.

(2) Design: A layer of endothelial cells is cultured between Layer 1 and Layer 2 to simulate the vascular wall and the biological barrier for substance exchange.

(3) Key Feature: This layer is equipped with a dedicated liquid extraction channel. This allows for easy and direct sampling of the medium for real-time monitoring of molecular concentrations (e.g., AHL, c-di-GMP, cytokines) without disturbing the cultured cells and organoids below.

• Layer 3 - Tumor Microenvironment Simulation Layer (Bottom Layer):

(1) Function: Hosts the core biological components to model the CRC TME.

(2) Operation: This layer is designed to culture intestinal organoids or tumor organoids derived from patient-specific cells. A co-culture medium containing our engineered EcN bacteria and macrophages is perfused into this chamber, enabling the study of bacterial targeting, immune cell interaction, and therapeutic effect in a tissue-like context.

(3) Barrier between Layers: A semi-permeable membrane is placed between Layer 2 and Layer 3. This membrane allows for the free passage of small molecules (like nutrients, AHL, c-di-GMP) while physically separating the flowing liquid and cells, facilitating the solid-liquid separation and enabling the sampling function in Layer 2.

Simulation of Intestinal Peristalsis

A unique and critical feature of our chip is its ability to simulate the mechanical forces of the human intestine.

• Mechanism: Vacuum pump channels are built on both sides of the chip, adjacent to Layers 2 and 3.
• Function: By applying cyclic suction (negative pressure) to these channels, the PDMS walls of the middle and bottom layers rhythmically stretch and relax.
• Outcome: This dynamic mechanical deformation successfully mimics the peristaltic motions of the colon, exposing the cultured cells and bacteria to physiologically relevant mechanical stresses that significantly influence cell behavior, bacterial distribution, and biofilm formation.

Application for Our Project

This gut-on-a-chip serves as a powerful testbed for our dry and wet lab components:

• Testing Inducer Diffusion: The chip can be used to validate our spatial diffusion model by visually observing and quantitatively measuring the distribution of AHL from the "vascular" layer (L1) to the "tumor" layer (L3).
• Evaluating Bacterial Function: We can directly observe the colonization behavior of engineered EcN on tumor organoids, its hypoxia-induced c-di-GMP production, and most importantly, the effect of AHL-induced lysis on the bacterial population and the subsequent immune response from the co-cultured macrophages.
• Safety Assessment: The chip provides a platform to monitor potential risks, such as the integrity of the endothelial and organoid barriers after bacterial lysis, giving insights into the safety profile of our therapy.

This custom-designed gut-on-a-chip is not merely a container for cells but an advanced in vitro model that integrates biological complexity and physiological mechanics. It represents a crucial hardware component of our project, enabling high-fidelity testing of our engineered bacterial system in a controlled yet realistic human-relevant environment.

Figure 6: (A) Traditional organoid chips; (B) Our innovative intestinal organoids

Key Features

Our project incorporates several innovative and safety-conscious features:

• Tumor-Targeted Delivery: The use of EcN and a hypoxia-inducible promoter ensures specific activation of our therapy within the tumor, minimizing off-target effects.
• Potent Innate Immune Activation: We directly deliver c-di-GMP, a highly potent STING agonist, to overcome the immunosuppressive TME.
• Dual-Controlled Safety System: Our design includes two layers of control:

(1) Spatio-Temporal Control: c-di-GMP production is restricted to hypoxic environments (tumors).

(2) Kill Switch: The inducible "suicide" system allows for the precise termination of the bacterial population via external AHL application, ensuring complete clearance after therapy and addressing biocontainment concerns.

• Biofilm Localization: The inherent property of c-di-GMP to enhance biofilm formation and reduce bacterial motility serves as a natural safety mechanism, potentially trapping the engineered bacteria within the tumor site.
• Interdisciplinary Approach: We combine sophisticated genetic engineering, quantitative mathematical modeling, and custom hardware design to thoroughly design, predict, and test our system.

Impact

Our project has the potential to make a significant impact:

• Therapeutic Impact: We propose a novel, targeted immunotherapy strategy for colorectal cancer, a disease in dire need of new treatment options. By reactivating the local immune response, our approach could be synergistic with existing therapies like checkpoint inhibitors⁸.
• Safety Impact: The dual-control system (biofilm localization + inducible lysis) sets a new standard for safety in engineered bacterial therapies, addressing a major public and regulatory concern.
• Technical Impact: We provide a modular platform based on EcN. The "suicide" circuit and the immunostimulant production module can be adapted to deliver other therapeutic agents for different cancers or diseases.
• Social Impact: Through our Human Practices (HP) work, we have engaged with healthcare professionals, the public, and specifically the elderly population (who are most affected by CRC) to understand their needs and educate them about synthetic biology and its ethical implications.

Reference

1 Li Y, Chen Z, Han J, Ma X, Zheng X, Chen J. Functional and Therapeutic Significance of Tumor-Associated Macrophages in Colorectal Cancer. Front Oncol. 2022;12:781233.

2 Chen H, Lei P, Ji H, Yang Q, Peng B, Ma J, et al. Advances in Escherichia coli Nissle 1917 as a customizable drug delivery system for disease treatment and diagnosis strategies. Mater Today Bio. 2023;18:100543.

3 Ouyang S, Song X, Wang Y, Ru H, Shaw N, Jiang Y, et al. Structural Analysis of the STING Adaptor Protein Reveals a Hydrophobic Dimer Interface and Mode of Cyclic di-GMP Binding. Immunity. 2012;36(6):1073-86.

4 Ge C, Yu Z, Sheng H, Shen X, Sun X, Zhang Y, et al. Redesigning regulatory components of quorum-sensing system for diverse metabolic control. Nature Communications. 2022;13(1):2182.

5 Zhu Q, Zheng Y, Zhou X, Wang D, Yuan M, Qian D, et al. c-di-GMP and AHL signals-triggered chemical communication under electrical signaling disruption restores Geobacter sulfurreducens biofilm formation. ISME Commun. 2024;4(1):ycae096.

6 Calleja D, Fernández-Castañé A, Pasini M, de Mas C, López-Santín J. Quantitative modeling of inducer transport in fed-batch cultures of Escherichia coli. Biochemical Engineering Journal. 2014;91:210-9.

7 Taavitsainen S, Juuti-Uusitalo K, Kurppa K, Lindfors K, Kallio P, Kellomäki M. Gut-on-chip devices as intestinal inflammation models and their future for studying multifactorial diseases. Frontiers in Lab on a Chip Technologies. 2024;Volume 2 - 2023.

8 Canale FP, Basso C, Antonini G, Perotti M, Li N, Sokolovska A, et al. Metabolic modulation of tumours with engineered bacteria for immunotherapy. Nature. 2021;598(7882):662-6.