PA54 is a short-chain polyamide that can be entirely derived from renewable resources by the polymerization of the bio-based monomers cadaverine and succinate. At present, the production of bio-based PA54 relies on two distinct bacterial strains to independently synthesize cadaverine and succinate. Two separate microbial processes introduces challenges including process complexity, high consumption of acid-base buffers, and substantial generation of waste salts. To address these limitations, we aimed to construct an intelligent cell factory capable of concurrently and efficiently producing both cadaverine and succinate.

To achieve this goal, four core engineering modules were designed: developing genetic switch , engineering key enzyme , cofactor regulation, and finally constructing an intelligent cell. First, a switchable system was screened and optimized to enable the transition from aerobic lysine accumulation to anaerobic simultaneous synthesis of the two target monomers. Second, guided by computational simulations and machine learning, lysine decarboxylase (CadA) was subjected to AI-based design and rational design to enhance both catalytic efficiency and enzyme stability. Then, cofactor regeneration and conversion were carried out to optimize the intracellular NADPH/NADH. Finally, an intelligent cell factory was constructed by integrating the three modules together that can simultaneously synthesize cadaverine and succinate from glucose.

Application of synthetic biology strategies allowed the engineered system to achieve simultaneous cadaverine and succinate production within a single bacterial cell. This configuration facilitates acid-base self-neutralization, thereby reducing buffer consumption and waste salt formation, while also enabling in situ utilization of CO₂ released from decarboxylation for succinate synthesis. In summary, these modifications will enhance carbon atom economy and offer a novel strategy for sustainable and green biomanufacturing.

We selected Escherichia coli NT1003 as the chassis organism for this project, primarily due to its superior performance in lysine biosynthesis and its high amenability to genetic engineering. This metabolically engineered strain is a high-yield lysine producer, capable of efficiently accumulating lysine as the direct precursor to cadaverine. This provides a robust metabolic foundation for subsequent cadaverine synthesis via lysine decarboxylation.

Additionally, E. coli NT1003 exhibits facultative anaerobic metabolism, enabling efficient growth and lysine accumulation under aerobic conditions, while anaerobic conditions favor succinate production. This physiological property aligns well with our two-stage fermentation strategy of “aerobic growth—anaerobic production,” rendering NT1003 an ideal platform for the simultaneous synthesis of cadaverine and succinate within a single cell.

In summary, the chassis strain E. coli NT1003 not only satisfies the project’s metabolic requirements but also provides a favorable platform for constructing an intelligent, efficient, and environmentally sustainable single-cell factory, ultimately facilitating the green production of fully bio-based PA54.

To achieve precise temporal separation between lysine accumulation and cadaverine production, we first designed and implemented two genetic switch systems. The primary objective was to ensure lysine accumulation during the aerobic growth phase through strict repression of lysine decarboxylase (CadA) expression, while enabling CadA activation exclusively during the anaerobic phase to catalyze lysine decarboxylation.

CadA expression was regulated by modulating culture temperature, which was a strategy easily scalable for industrial fermentation. TlpA39 (BBa_K2572000) and its cognate promoter PtlpA (BBa_K2500003) were obtained from the part pages. TlpA39 functions as a temperature-sensitive recombinant transcriptional repressor: at temperatures below 39 °C, TlpA39 forms a dimer that binds PtlpA, thereby repressing downstream gene expression; at temperatures above 39 °C, TlpA39 dissociates into monomers, losing its promoter-binding capability and allowing activation of downstream genes. This temperature range is compatible with E. coli growth. A genetic circuit of PJ23100-TlpA39-PtlpA-CadA was designed to dynamically regulate CadA expression through controlled temperature shifts(Figure1).

A-IDP (Artificial Intrinsically Disordered Protein) is a rationally designed protein intended to emulate the characteristics of naturally occurring intrinsically disordered proteins. Its highly flexible disordered regions enable it to function as a “molecular net,” dynamically interacting with and binding target biomolecules.

In this study, A-IDP was employed to modulate CadA activity. Under aerobic conditions, A-IDP encapsulated CadA, inhibiting its enzymatic activity and preventing lysine consumption . Under anaerobic conditions, specific proteases (TEV/TVMV proteases) could recognize and cleave their corresponding recognition sites (TEV-site/TVMV-site), releasing A-IDP from CadA. This removal alleviates the inhibitory interaction, enabling CadA to catalyze the conversion of lysine to cadaverine(Figure2).

In subsequent wet-lab experiments, the temperature-sensitive switch-based dynamic regulatory system with the better functionality was selected for all subsequent experiments.

To enhance the stringency and precision of CadA regulation, promoters of varying strengths were systematically evaluated (Figure3). Building upon this, a high-efficiency RNA thermometer was incorporated to further refine translational control. At low temperatures, the RNA thermometer induces the formation of a stem-loop structure at the ribosome binding site (RBS), preventing ribosome binding and thereby inhibiting translation of downstream genes (Figure4). At elevated temperatures, the stem-loop unfolds, permitting normal translation.

Through combined transcriptional and translational optimizations of the switch genetic circuit, the system achieves a finely tuned temperature-response gradient, enabling more precise and robust control of CadA expression.

Lysine decarboxylase (CadA) catalyzes the decarboxylation of lysine to produce cadaverine and CO₂, as a key enzyme in our engineered cell factory. To enhance its catalytic performance, we sought to engineer CadA through AI-based design and rational design, focusing on two key aspects: catalytic efficiency and stability.

To enhance catalytic efficiency, a computationally guided approach was employed. Using the crystal structure of CadA (PDB No. 6YN6) as a starting point, essential structural components were stabilized, and substrate-binding modes were analyzed through molecular simulations. Systematic evaluation of hydrogen bonds, salt bridges, and hydrophobic interactions between the enzyme and substrate enabled identification of residues critical for substrate binding and maintenance of the catalytic microenvironment.

Based on this interaction analysis, key residues located near the active site and directly or indirectly involved in substrate binding or cofactor coordination were selected, including I182, H245, K246, W333, E526, K527, and Y652 (Figure 5). These candidate sites were then input into a site-directed modification prediction model to perform virtual amino acid substitutions and predict corresponding catalytic efficiencies (K_cat). This process identified several mutant combinations with high predicted enzymatic activity: H245Q, W333E, E526N, K527S, G655F; H245Q, W333H, E526N, K527S, G655F; H245Q, W333H, E526N, K527S, G655K; H245Q, W333E, E526N, K527S, G655K; H245F, W333E, E526N, K527S, G655F (Table 1).

To assess the functional contribution of key residues, single-site mutations were introduced at I182, H245, K246, W333, E526, K527, and Y652, and their effects on enzymatic activity were evaluated. Given that conventional saturation mutagenesis entails considerable redundancy and labor, we employed a systematic approach: initial screening was conducted via alanine scanning, followed by further substitutions at positions with amino acids of diverse physicochemical properties (E, L, S, F, V) to explore the functional roles.

A complementary strategy was subsequently implemented to enhance protein stability. CadA functions as a decamer (Figure6), formed by the assembly of five dimers, and its decarboxylation activity is strongly dependent on the integrity of this quaternary structure. Accordingly, stabilizing the decameric assembly was prioritized. Computational simulations guided the design of several stabilizing mutations, including F14C/K44C, V12C/K41C, F14C/L45C, L93C/E445C, Y13C/P36C, and E104K. In addition, previously reported stabilizing mutations were incorporated, such as V12C/K44C, F14C/D41C, and L89C/E445C, to further enhance structural stability.

For the co-produced succinate, metabolic flux analysis performed by the in silico team identified glyceraldehyde-3-phosphate dehydrogenase (GAP) as a key modification target. To enhance the availability of reducing power required for succinate synthesis, GAP was overexpressed. Succinate is a reductive product whose biosynthesis consumes substantial reducing power.

Four NADP⁺-dependent GAPs originated from different sources were selected, including Corynebacterium glutamicum ATCC 13032 (CggapC), Bacillus subtilis (BsrocG), Clostridium saccharobutylicum DSM 13864 (CsgapC), and Streptococcus mutans (SmgapN). These enzymes catalyze the conversion of D-glyceraldehyde 3-phosphate to 3-phospho-D-glycerate and simultaneously generate NADPH, thereby increasing intracellular reducing power.

Additionally, to mitigate the effects of excess reducing power on cellular metabolism and lysine accumulation during the aerobic growth phase, the expression of glyceraldehyde-3-phosphate dehydrogenase was further optimized using plasmids of varying copy numbers.

To further enhance succinate production, cofactor distribution and flow were rationally engineered. For the NADH required in the succinate synthesis, the soluble pyridine nucleotide transhydrogenase (sthA) from E. coli MG1655 catalyzes the conversion of NADPH to NADH, which was selected to balance different types of cofactor. Co-expression of sthA with CggapC enabled optimization of the intracellular cofactor pool, thereby improving reducing power allocation and enhancing succinate biosynthesis.

Based on preliminary experiments, the temperature-responsive switch regulating CadA expression and the engineered cofactor system were optimized and integrated into the lysine-producing strain E. coli NT1003. A single-cell factory was constructed to simultaneously produce cadaverine and succinate, the two monomers of PA54.