https://orcid.org/0000-0002-8523-103X
Abstract
Corynebacterium glutamicum is a key industrial workhorse for producing amino acids and high-value chemicals. Balancing metabolic flow between cell growth and product synthesis is crucial for enhancing production efficiency. Developing dynamic, broadly applicable, and minimally toxic gene regulation tools for C. glutamicum remains challenging, as optogenetic tools ideal for dynamic regulatory strategies have not yet been developed. This study introduces an advanced light-controlled gene expression system using light-controlled RNA-binding proteins (RBP), a first for Corynebacterium glutamicum. We established a gene expression regulation system, ‘LightOnC.glu’, utilizing the light-controlled RBP to construct light-controlled transcription factors in C. glutamicum. Simultaneously, we developed a high-performance light-controlled gene interference system using CRISPR/Cpf1 tools. The metabolic flow in the synthesis network was designed to enable the production of chitin oligosaccharides (CHOSs) and chondroitin sulphate oligosaccharides A (CSA) for the first time in C. glutamicum. Additionally, a light-controlled bioreactor was constructed, achieving a CHOSs production concentration of 6.2 g/L, the highest titer recorded for CHOSs biosynthesis to date. Herein, we have established a programmable light-responsive genetic circuit in C. glutamicum, advancing the theory of dynamic regulation based on light signaling. This breakthrough has potential applications in optimizing metabolic modules in other chassis cells and synthesizing other compounds.
Introduction
As an industrial model microorganism, Corynebacterium glutamicum is widely used in fields such as the food and medical industries, showing remarkable potential for synthesizing various novel chemicals (1,2). However, target product synthesis in C. glutamicum often encounters difficulties, including organic acid by-product accumulation and noticeable bacterial growth inhibition (3). The metabolic and regulatory networks in the host cell tend to prioritize resources for cell growth and reproduction, making it challenging to balance cell growth with target product synthesis through static regulation alone (4). Therefore, achieving an optimal microbial cell factory requires directional and synergistic dynamic regulation of multiple key genes in the metabolic pathway (5–7). However, at present, effective dynamic control tools remain limited in C. glutamicum. Therefore, developing a more universal and refined dynamic control technology for C. glutamicum is of particular significance (8).
In recent years, researchers have developed precise and effective tools for regulating gene expression within living cells, focussing on both temporal and spatial control (9,10). Chemically induced gene expression systems are well established, offering timely gene regulation with high induction levels and minimal background expression (11,12). However, these chemical molecular inducers face challenges such as physiological toxicity, pleiotropy, poor universality and uncontrollable diffusion (13), which restrict their removal during fermentation and hinder the system's ability to toggle between active and inactive states. Additionally, although some regulatory elements can activate or repress gene transcription in response to specific metabolites, not all intracellular metabolites have natural biosensors. Current dynamic regulatory strategies mainly focus on identifying and using transcription factor-promoters responsive to metabolites. This approach is not limited in scope but also affects the expression strength of the original promoter. Temperature-controlled gene circuits, which rely on temperature-sensitive transcriptional regulators and their corresponding promoters, often experience regulatory delays that can impact bacterial growth and essential enzyme activities (14).
Conversely, light-induced switches offer several advantages, including high resolution, rapid transmission, good reversibility, controllable intensity and minimal toxicity (15,16). These features have made light-induced switches increasingly popular for developing precise spatiotemporal control technologies in cellular metabolism (17,18). However, current light control systems that utilize photosensitive transcription factors typically require specific promoter DNA-binding motifs (10). This necessitates fusing the operon sequence for transcription factor recognition with regulatory gene promoters, a process that can diminish promoter activity and universality across various chassis cells and varying gene expression scenarios. Furthermore, there are inherent delays in gene expression, transitioning from DNA to mRNA and subsequently from mRNA to protein. Starting regulation at the post‐RNA transcription stage of metabolism results in shorter delays than those of the previously studied gene expression systems.
Therefore, the technology of dynamic gene regulation based on RNA level is highly expected. RNA-binding proteins (RBPs) are crucial in regulating cellular RNA functions, recognizing and binding to specific RNA sequences or structures (19). By integrating an RBP with a photosensitive structural domain responsive to blue light, the resulting fusion protein can bind specific RNA sequences upon exposure. This enables the photoregulation of RNA functions and metabolism within cells. When combined with various RNA effectors, these fusion proteins can facilitate optogenetic control over RNA localization, splicing, translation and stability (20). This system holds promise for controlling gene circuits, enhancing artificial metabolic pathway adaptability in host cells and improving biomanufacturing efficiency through precise regulation of RNA metabolism.
Therefore, this study aimed to develop an efficient and controllable dynamic gene regulation system by constructing a high-performance, advanced light-controlled gene expression system, ‘LightOnC.glu’ using a light-controlled RBP in C. glutamicum. This light control system does not require specific promoter DNA-binding motifs and offers good generalizability. Initially, we constructed a synthetic light-switchable activating protein, LicV, which is a fusion protein consisting of the transcription anti-terminator protein LicT and the VVD photosensitive domain. In darkness, LicV remains monomeric and does not bind to mRNA. Concurrently, the ribonucleic acid antiterminator (RAT) adopts a stem-loop structure that inhibits transcription. Upon light exposure, LicV dimerizes and binds to the RNA, altering the RAT structure and allowing transcription to proceed, thereby initiating gene expression. We then developed a mutant library of the linker region between LicV recombinant proteins and the LicV gene RBS sequence, creating multiple light control systems with finely tunable induction characteristics, such as background noise, maximum activation level, activation kinetics and photosensitivity. Further studies indicated that this system could integrate with the CRISPR-dCpf1 system to construct a ‘NOT’ Boolean logic gate and develop a high-performance light-controlled gene interference system. Subsequently, we used the chitin oligosaccharides (CHOSs) and chondroitin sulphate oligosaccharides A (CSA) synthesis pathways as a proof of concept. By strategically designingthe metabolic flows of these synthesis networks through light-controlled gene expression regulation, we induced dynamic gene interference within competing pathways, improving the target product titer. A 5-L light-controlled bioreactor was designed for light culture to validate the adaptability of the light-controlled gene expression system at the microbial reactor scale. Finally, the titer of CHOSs increased to 6.2 g/L, marking the highest microbial synthesis titer of CHOSs to date. Our discoveries offer crucial theoretical and technological guidance for enhancing the use of light-controlled bioreactors in industrial applications.
This work provides a novel and effective toolkit for dynamic regulation in C. glutamicum, expanding the scope of control genetics from photosensitive and growth-regulating proteins to dynamically managing metabolic fluxes. By integrating control genetics into dynamic metabolic pathway regulation, we enriched the theory and methodology of light signaling‐based dynamic regulation. The framework developed here for designing and optimizing light-responsive gene circuits based on RBPs may be useful for engineering other microorganisms.
Material and methods
Bacterial strains, plasmids, culture media and materials
Genetic design and sequence reading were carried out by SnapGene 6.0.2. Primers designed and used for genetic modification in this study are listed in Supplementary Table S1. Supplementary Table S2 depicts all strains and plasmids used in this study. Escherichia coli JM109, which was used for vector construction, was cultured in Luria–Bertani medium (10 g/L of tryptone, 5 g/L of yeast extract and 5 g/L of NaCl) with 50 μg/mL of kanamycin for selection. C. glutamicum S9114 and its derivatives were cultivated aerobically at 30°C in LBB medium (LB broth with 18.5 g/L brain-heart infusion) with Kanamycin (25 μg/mL) as required. C. glutamicum competent cells were prepared and electrotransformed following a previously described protocol (21,22).The fermentation media for the CHOSs-producing strains contain the following components with the unit of g/L: glucose, 100.0; N-acetylglucosamine (GlcNAc), 20.0; corn syrup, 10.0; KH2PO4, 1.0; (NH4)2SO4, 20.0; MgSO4, 0.5; calcium carbonate, 20.0; and FeSO4, 0.18. The fermentation media for the CSA-producing strains contain the following components with the unit of g/L: glucose, 100.0; corn syrup, 10.0; KH2PO4, 1.0; (NH4)2SO4, 20.0; MgSO4, 0.5; calcium carbonate, 20.0; p-Nitrophenyl sulfate, 2.0; 3′-Phosphoadenosine-5′-phosphate, 0.43; and FeSO4, 0.18.
All chemicals were purchased from Sangon Biotech (Shanghai, China). The DNA gel purification kit, plasmid extraction kit, restriction enzymes, and T4 DNA ligase were purchased from Thermo Fisher Scientific (Waltham, MA, USA). The PrimeSTAR HS DNA polymerase used for fragment amplification was purchased from Takara Biomedical Technology (Beijing, China). Taq DNA polymerase employed for colony polymerase chain reaction (PCR) was purchased from BIO SCI Technology (Hangzhou, China). A seamless cloning kit and annealing buffer for DNA oligonucleotides were purchased from Beyotime Biotechnology (Shanghai, China). Oligonucleotides were synthesized by GENEWIZ (Suzhou, China). The Modified Bradford Protein Assay Kit was purchased from Sangon Biotech Co. Ltd. (Shanghai, China).
DNA cloning and light irradiation
We screened the transcriptional anti-termination protein-photosensitive protein complex to select photoreceptive RBPs suitable for C. glutamicum. The RNA-binding structural domain LicTCAT from the transcriptional anti-termination protein LicT derived from Bacillus subtilis, E. coli, Clostridium botulinum, Serratia marcescens, and Enterococcus faecalis (abbreviated as BsLicT, EcLicT, CbLicT, SmLicT and EfLicT, respectively), were fused with EL222 from Erythrobacter litoralis, Nc light–oxy-gen–voltage (LOV) domain (Vivid, VVD domain) from Neurospora crassa, VfLOV from Vaucheria frigida, AsLOV2-SsrA from Avena sativa and E. coli / SspB from E. coli, and Bphp1/Q-PAS1 gene sequences derived from Rhodopseudomonas palustris, respectively. The amino acid sequence of the linker was (GGGGS)3, and its expression was initiated by the constitutive promoter PJ23119. The DNA sequences of fusion proteins are listed in Supplementary Note. These 25 fusion protein-coding genes were inserted into the pJYW-4 plasmid at multiple cloning sites for Ptac promoter-regulated expression. Using RAT-F/RAT-R and mcherry-F/mcherry-R as primers, we inserted genes comprising RAT antiterminator and mCherry fluorescent protein reporter coding genes from E. coli DH5-α into the upstream position of transcriptional anti-termination protein-photosensitive protein fusion proteins. This established a reporter system to detect the anti-termination effects of different transcriptional anti-termination protein-photosensitive protein fusion proteins. The resulting recombinant plasmid was named pJYW-4-RAT-cherry-LicV1-25 (Supplementary Table S2). Transforming this recombinant plasmid, pJYW-4-RAT-cherry-LicV1-25, into C. glutamicum S9114 yielded the recombinant strain, CG-RClicV1-25.
Recombinant strains containing four blue light-sensing proteins, EL222, NcLOV, VfLOV and AsLOV2-SsrA/SspB, were incubated in the dark for 10 h to detect light-regulated gene expression. The experimental group was then irradiated with 2.5 mW/cm2 of blue light (460 nm peak) from an LED lamp, with light intensity regulated using neutral-density filters. A neutral-density filter adjusted light intensity. The recombinant strain containing the near-infrared light-sensitive protein Bphp1/Q-PAS1 was subjected to a 10 h incubation to detect light-regulated gene expression. Subsequently, the experimental group was irradiated with 2.5 mW/cm2 red light (760 nm peak) emitted from an LED lamp, with light intensity regulated by another neutral‐density filter. The light intensity was measured using a photometer (Sanwa LX-2). Shaker flasks were wrapped in tin foil and incubated under light protection for 24 h for dark manipulation. mCherry fluorescence and the optical density of cell at 600nm (OD600 nm) were measured using Synergy 2 multi-mode zymography (BioTek) (excitation, 580 nm; emission, 610 nm). Fluorescence values were normalized to the OD600 nm of each sample. Quantification of mCherry expression in the plate cells were conducted using the ImageJ image processing program.
The primers linker5-F/R, linker10-F/R, and linker15-F/R were phosphorylated at the 5′ end using a phosphorylation kit. The recombinant plasmid pJYW-4-RAT-cherry-LicV0 served as a template to construct the linker region between LicTCAT-VVD utilizing the 5′ phosphorylated linker5-F/R, linker10-F/R, and linker15-F/R primers, respectively. The resulting mutant library of LicTCAT-VVD was then transformed into C. glutamicum S9114. The recombinant LicTCAT-VVD protein in the obtained linker-5, linker-10 and linker-15 mutant libraries carried 5, 10 and 15 amino acid mutation sites, respectively, with all amino acids randomly mutated into any combination of amino acids. The mCherry fluorescent protein-coding gene was used as a reporter gene to validate the effect of binding the LicTCAT-VVD fusion protein with the RNA anti-terminator sequence. This binding inhibited transcription termination and enhanced protein expression under blue and dark illumination conditions.
Recombinant strains from the transformed plate medium were selected, inoculated into 96-well plates, and cultured in the dark until the time of the log phase of cell growth (10 h). Subsequently, they were incubated at 30°C under 2.5 mW/cm2 blue light irradiation for 24 h. The control group was cultured in the dark throughout the culture cycle. After centrifugation at 10 000 × g for 10 min, the cells were resuspended in PBS until the optical density (OD600 nm) at 600 nm was 0.1. Cell analysis and sorting were conducted using a BD FACSIOR flow cytometer with a 561 nm laser, based on mCherry fluorescence intensity, to determine recombinant strain expressing the LicTCAT-VVD fusion protein with higher expression levels. Following this, the sorted cells were cultured in the dark at 30°C for 18 h to determine recombinant strains with low mCherry fluorescence intensity. This ‘ blue light-dark’ screening cycle was carried out three times to enrich the recombinant strains of C. glutamicum expressing LicTCAT-VVD fusion protein, to screen out the recombinant strains with high expression activation level of mCherry reporter gene after blue light irradiation but low background activation level under dark conditions.
To detect LicTm-mediated resistance to transcriptional termination, the gene was synthesized from the full-length sequence of the transcription-resistant termination protein LicT from B. subtilis, incorporating two amino acid mutations, H207D and H269D, to obtain the mutant LicTm. The primers LicTm-F/LicTm-R were utilized to integrate the LicTm gene fragment into plasmid pJYW-4-RAT-mcherry-LicV2, resulting in the recombinant plasmid pJYW-4-RAT-mcherry-LicTm. This plasmid was subsequently transformed into C. glutamicum strain S9114 (23). A mutant library of the licT-VVD RBS sequence was constructed using phosphorylated primers RBS1-F/RBS1-R and RBS2-F/RBS2-R. The recombinant plasmid of lict-VVD with different RBS sequences was then transformed into C. glutamicum S9114 cells and the grown recombinants were cultured and screened in a 96-well plate. The fluorescence content of recombinant strains with different Lict-VVD expression levels was validated using mCherry fluorescent protein as a reporter gene under blue light and dark irradiation.
Light irradiation
For the recombinant strains containing four blue light-sensing proteins, EL222, NcLOV, VfLOV and AsLOV2-SsrA/SspB, we detected the light-regulated gene expression by incubating the strains in the dark for 10 h. The experimental group was then irradiated with blue light emitted from LED lamps at a light intensity of 2.5 mW/cm2 (460 nm peak), and the light intensity was adjusted using a neutral density filter. For the recombinant strains containing the near-infrared light-sensing protein Bphp1/Q-PAS1, to detect light-regulated gene expression. The recombinant strains were incubated in the dark for 10 h, followed by irradiation of the experimental group with red light (760 nm peak) emitted from an LED lamp at a light intensity of 2.5 mW/cm2, also adjusted using a neutral-density filter. A Sanwa photometer (Sanwa, LX-2) measured light intensity. For dark manipulation, shaker flasks were wrapped in tin foil and incubated for up to 24 h under light protection. Fluorescence imaging was conducted using a Leica SP8 confocal laser scanning microscope equipped with an HCPLAPOCS2 63.0 × 1.40 oil objective lens. Sorting was conducted with a BD FACSJESTTM flow cytometer with a 561-nm laser.
Light-inducible transcription by the CRISPRi system
Primers dcpf1-F/dcpf1-R were used to amplify the linearized plasmid using plasmid pJYS3_ΔcrtYF as a template (plasmid sequence reference: http://www.addgene.org/search/catalog/plasmids/?q=pJYS3_%CE%94crtYF) (24). The linearized plasmid was then transformed into E.coli JM109, and transformants were selected for sequencing. The recombinant plasmid containing the mutation site D917A was named pJYS3_ΔcrtYF-dcpf1. The recombinant strain C. glutamicum S9114-dcpf1 was obtained by integrating dcpf1 into the genome of C. glutamicum S9114. The crRNA used in this study was designed using the CRISPR‐DT tool (25). Sequences of all crRNA are shown in Supplementary Table S5.
The recombinant plasmid pJYS3_ΔcrtYF-crmcherry, with crRNA corresponding to the mchery gene, was amplified using primer mcherry-crRNA-F/mcherry-crRNA-R and pJYS3_ΔcrtYF as a template. The corresponding crRNA fragment of the mcherry gene was incorporated into pJYW-4-RAT-cherry-LicV27 using primer crm-F/crm-R, and Ptac-M promoter was added to the mcherry gene with primers pm-F/pm-R, resulting in the recombinant plasmid pJYW-4-RAT-crmcherry-mcherry-LicV. Transformation of plasmid pJYW-4-RAT-crmcherry-mcherry-LicV into the C. gltamicum S9114-dcpf1 strain yielded the recombinant strain CG-mCherryi. The relative fluorescence intensity of this recombinant strain was assessed through shaking flask fermentation, induced by blue light irradiation, or incubated in the dark for 24 h.
RT–qPCR
Gene expression was quantitatively analyzed using fluorescence to detect the transcription level of the gene. During the fermentation process, bacterial samples were collected and centrifuged to obtain the bacteria cells, which were then ground with liquid nitrogen for total RNA extraction. Total RNA was extracted using the method described in the RNA extraction kit (TAKARA), and cDNA was obtained by reverse transcription with the Prime Script RT Reagent Kit Perfect Real Time (TAKARA). Using cDNA as the template and 16s rRNA as the reference gene, a reaction system was prepared with the kit (SYBR® Premix Ex Taq™, TAKARA). qRT-PCR detection and data analysis were conducted in a light cycler real-time PCR amplifier.
Construction of CHOSs and CSA biosynthetic strains
Using primers scgna1-F and scgna1-R, the coding gene GNA1 for glucose-6-phosphate acetyltransferase (ScGna1) was amplified from the genome of Saccharomyces cerevisiae S288C. Using primers ecyqab-F and ecyqab-R, the coding gene yqaB for phosphogluconate phosphatase from E. coli K12 was amplified from the genome of E. coli K12. Using primers mlnodC-F and mlnodC-R, the N-acetylglucosamine transferase coding gene nodC from Mesorhizobium loti was amplified from the genome of M. loti. Plasmid pJYW-4 served as the expression vector for GNA1, yqaB and nodC. The ClonExpress II One-Step Cloning Kit (Novozen Biotechnology Co., Ltd.) was used to connect the expression vector to the gene fragment. The linearized vector obtained by PCR and the target gene fragment, which had homologous ends of the vector, were recovered and mixed at a molar ratio of 3:1. Then, 5 × Ce II Buffer 4 L and Exnase II 2 L were added, followed by ddH2O to bring the total volume of the connection system reach 20 μL. The reaction was conducted at 37°C for 30 min, then cooled to 4°C for heat preservation. The 10 μL ligation system was used to transform competent cells of E. coli JM109 (refer to the Takara E. coli competent kit instructions for cell preparation). Transformants with correct colony PCR results were selected and sent to Suzhou Genewiz Biotechnology Co., Ltd. for sequencing verification, resulting in the recombinant plasmid pJYW-4-GNA1-yqaB-nodC. The plasmid was then transformed into C. glutamicum using electric shock transformation, yielding the recombinant strain CG-GNA1-yqaB-nodC for CHOSs synthesis.
Using primers eckfoC-F/eckfoC-R and eckfoA-F/eckfoA-R, the genes eckfoC and eckfoA from E. coli K4 were amplified from its genome. Primers scc4st-F and scc4st-R were used to amplify the sulfotransferase-coding gene C4ST from the genome of Saccharomyces cerevisiae S288c. Plasmid pJYW-4 served as the expression vector for kfoC, kfoA and C4ST genes. The ClonExpress II One-Step Cloning Kit (Novozen Biotechnology Co., Ltd.) was used to connect the expression vector to the gene fragment. The linearized vector obtained by PCR and the target gene fragment with homologous ends of the vector were recovered and mixed at a molar ratio of 3:1. To make the total volume of the connection system mixture 20 μL, 5 × Ce II Buffer 4 L, Exnase II 2 L and ddH2O were added. The reaction was carried out at 37°C for 30 min, then reduced to 4°C for heat preservation. A 10 μL ligation system was used to transform competent E. coli JM109 cells (refer to the Takara E. coli competent kit instructions for the preparation method). Transformants with correct colony PCR results were selected and sent to Suzhou Genewiz Biotechnology Co., Ltd. for sequence verification, yielding the recombinant plasmid pJYW-4-kfoC-kfoA-C4ST. Plasmid pJYW-4-kfoC-kfoA-C4ST was then transformed into C. glutamicum S9114 by electric shock transformation to obtain the recombinant strain CG-kfoC-kfoA-C4ST for CSA synthesis.
The primers zwf-crRNA-F/zwf-crRNA-R, pfk-crRNA-F/pfk-crRNA-R and murAA-crRNA-F/murAA-crRNA-R were employed to amplify the recombinant plasmid pJYS3_ΔcrtYF containing crRNA corresponding to the zwf, pfk and murAA genes, using pJYS3_ΔcrtYF as the template. The resulting recombinant plasmid pJYS3_ΔcrtYF-crzwf/pfk/murAA was amplified using pJYS3_ΔcrtYF as the template, and the recombinant plasmid pJYS3_ΔcrtYF-crzwf/pfk/murAA was amplified using the primers cr-F/cr-R, which integrated the array fragment of the corresponding crRNAs of the zwf/pfk/murAA genes into the pJYW-4-GNA1-yqaB-nodC, to obtain the recombinant plasmid pJYW-4-CHOS- crzwf/pfk/murAA. Similarly, primers cr-F/cr-R were used to integrate the crRNA array fragment for zwf/pfk/murAA genes into pJYW-4-kfoC-kfoA-C4ST resulting in the recombinant plasmid pJYW-4-CSA-crzwf/pfk/murAA.
Using pJYS3_ΔcrtYF-crzwf/pfk/murAA as a template, primers rat-cr-F/rat-cr-R were used to integrate the array fragment of crRNA corresponding to the zwf/pfk/murAA genes into pJYW-4-RAT-mCherry-LicV27, resulting in the recombinant plasmid pJYW-4- RAT-crzwf/pfk/murAA -LicV. The RAT-crzwf/pfk/murAA-LicV fragment was then cloned into the plasmid pJYW-4-GNA1-yqaB-nodC using the primers RL-F/RL-R to obtain the recombinant plasmid pJYW-4-CHOS-RAT-crzwf/pfk/murAA-LicV. Similarly, the RAT-crzwf/pfk/murAA-LicV fragment was cloned into the plasmid pJYW-4-kfoC-kfoA-C4ST using the primers RL-F/RL-R to obtain the recombinant plasmid pJYW-4-CSA-RAT-crzwf/pfk/murAA-LicV.
The recombinant plasmids pJYW-4-CHOS-crzwf/pfk/murAA and pJYW-4-CHOS-RAT-crzwf/pfk/murAA-LicV were transformed into C. glutamicum S9114-dcpf1 to obtain the recombinant strains CG-CHOS-zwf/pfk/murAAi and CG-CHOS-zwf/pfk/murAAi-R. Similarly, recombinant plasmids pJYW-4-CSA-crzwf/pfk/murAA and pJYW-4-CSA-RAT-crzwf/pfk/murAA-LicV were transformed into C. glutamicum S9114-dcpf1 strain to obtain the recombinant strains CG-CSA-zwf/pfk/murAAi and CG-CSA-zwf/pfk/murAAi-R.
Recombinant strains CG-GNA1-yqaB-nodC, CG-CHOS-zwf/pfk/murAAi, CG-CHOS-zwf/pfk/murAAi-R, CG-kfoC-kfoA-C4ST, CG-CSA-zwf/pfk/murAAi, and CG-CSA-zwf/pfk/murAAi-R were fermented in shake flasks. Strains CG-CHOS-zwf/pfk/murAAi-R and CG-CSA-zwf/pfk/murAAi-R were exposed to blue light (460 nm peak) irradiation conditions after 8 h of inoculation. After 75 h, the titers of CHOSs and CSA in the fermentation broth were measured.
Fluorescence assays and quantification of CHOSs/CSA
The background fluorescence of the strain without fluorescent protein expression (FPbg) and the background OD (ODbg) of the medium were corrected, and the relative fluorescence intensities were calculated using Equation (1) as follows (26).
CHOSs (CHOS2, CHOS3, CHOS4, and CHOS5) were detected using high-performance liquid chromatography with a Waters e2695 high-performance liquid chromatograph and a 2998 photodiode array detector. The column temperature was 30°C, and the mobile phase consisted of a 70% acetonitrile aqueous solution with a flow rate of 10 μL. CHOSs were identified using a Xevo G2-S Q-TOF-MS spectrometer (Waters, Milford, MA, USA). Mass spectrometry was set to the positive ion electrospray ionization detection mode, with a measurement scanning range of 100–1600 m/z. For MS, the low collision energy was set to 6 Ev, and the high collision energy was set to 20 eV for MS/MS (27).
The CS titer assay utilized the sulfuric acid-carbazole method. Anhydrous sodium tetraborate (4.77 g) was dissolved in 500 mL of concentrated sulfuric acid to create a borax sulfate solution. A carbazole solution (25 g/L) was prepared by dissolving carbazole in anhydrous ethanol and storing it in a brown bottle at 4°C. Glucuronic acid concentrations used were 10, 20, 30, 40, 50, and 100 mg/L. The absorbance was measured at 530 nm after the color reaction with borax sulfuric acid-carbazole. A standard curve correlating glucuronic acid concentration with A530 was plotted. The extracted and purified CS samples were diluted by appropriate multiplicity and they were detected using the sulfuric acid-carbazole method. The concentration of the corresponding glucuronic acid was calculated according to the standard curve, and the content of CS was calculated using the following formula:
The sulfonation level of CSA was defined as the ratio of the disaccharide (Di-4S) content in CSA to the total Di-4S and chondroitin disaccharide (Di-0S) amounts. Ultra-performance liquid chromatography-mass spectrometry (UPLC-MS) was employed for separation and detection, with peak area ratios of Di-4S and Di-0S calculated for quantitative analysis. The sample was treated with 10 U/mL chondroitin sulfate lyase ABC incubated at 37°C for 12 h for complete disaccharide decomposition. Subsequently, boiling in a water bath for 10 min deactivated proteins, followed by centrifugation at a high speed of 10 000 × g for 20 min to remove impurities through the MW3000 concentration tube. The purified disaccharide was detected via UPLC-MS (28).
Kinetic model simulation
To simulate the relative concentrations of specific metabolites or relative fluxes in specific pathways, a linear pathway kinetic model based on the Michaelis-Menten equation was used (29). The kinetic model construction and dynamic simulations were executed in MATLAB R2023a. In this model, x(1)–x(11) represent the relative concentration of the metabolite of interest or the relative flux in the pathway of interest, respectively, while v(1)–v(11) represent the reaction rates at each enzymatic step reaction. To simulate pathway overexpression and weakening, the maximum reaction rate (vmax) was adjusted as indicated in the code. The Km parameter was randomly sampled between 0.5 and 1.5 for all enzymes to produce an average Km of 1. Michaelis–Menten kinetics characterized the reaction kinetics (30).
Mass balances for all metabolites resulted in the following differential equations:
Fed-batch culture in 5L bioreactor
The seed medium used for the fed-batch culture contained the following components (g/L): glucose (25.0), corn syrup (20.0; KH2PO4, 1.0; (NH4)2SO4 (0.5), and urea (1.25). The fermentation medium used for the fed-batch culture contained the following components (g/L): glucose, 50.0; corn syrup, 10.0; KH2PO4, 1.0; (NH4)2SO4, 20.0; MgSO4, 0.5; CaCO3, 20.0; and Fe2(SO4)3, 0.18. The final pH of the seed and fermentation media was adjusted to 7.0. Kanamycin (25 mg/L) was added to all media to detect transformants or recombinant cultures (8).
The seed culture was fermented in a 500 ml shake flask containing 50 ml of seed culture medium. The fermentation temperature of the seeds was 30°C, and the seeds were cultured for 16 h at the speed of 220 rpm. The seed fermentation broth was then inoculated into a 5-L fermenter (Shanghai Bailun Biotechnology Co., Ltd.) to ensure that the initial OD600 nm of the fermentation medium was 1.6. The 5‐L bioreactor was equipped with a light source installed on the tank wall to enable irradiation. By measuring the light intensity at different positions in the light-control bioreactor under different lighting conditions, the spatial distribution of light intensity within the bioreactor was obtained. The average light intensity in the tank was maintained at 2.5 mW/cm2 of blue light (460 nm peak). The fermentation medium began with an initial volume of 2 L. To maintain a pH of 7.0, 29% NH3 was added while the temperature was kept at 30°C. The air volume and initial speed were set at 0.8 Nm3/h and 25 Hz, respectively. To keep the glucose and GlcNAc concentration within the range of 5 g/L, 500 g/L of glucose and 200 g/L of GlcNAc were added, respectively, and the corresponding addition rate was adjusted according to the change of glucose and GlcNAc concentration in the fermentation medium.
Statistical analysis
In this study, the experiments were performed in independent triplicates for each strain. The data are expressed as the mean ± SD. Statistical data analysis was carried out using t-tests in the GraphPad Prism 9.0.0 software. Statistical significances are indicated as * and ** for P-values of p < 0.05 and < 0.01, respectively.
Results
Screening of light-controlled RNA binding proteins in C. glutamicum
In this study, we aimed to identify effective light-controlled RBPs in C. glutamicum. We fused the transcription anti-termination protein LicT with a light-sensitive domain. The conformation of the light-sensitive domain changed under light stimulation, leading to a change in the conformation of fusion protein LicV. Consequently, the LicV dimer stabilized and specifically bound to the RNA sequence of the RNA anti-terminator (RAT), preventing the formation of an RNA terminator stem-loop. This light-dependent binding between LicV and the target RNA allowed us to regulate it by controlling the light source. To do this, we searched sequence databases (UniProt and InterPro) and synthesized transcription anti-terminal proteins BsLicT, EcLicT, CbLicT, SmLicT, and EfLicT from B. subtilis, E. coli, C. botulinum, S. marcescens, and E. faecalis, respectively. Simultaneously, we synthesized EL222 from E. litoralis, NcLOV (VVD) from N. crassa, VfLOV from V. frigida, AsLOV2-SsrA/ from A. sativa and E. coli/SspB from E. coli, and Bphp1/Q-PAS1 from R. palustris. Many photosensitive domains found in nature undergo light-dependent transformations in their oligomeric state, enabling them to optically control signal transduction and transcription events. Among them, EL222, NcLOV, VfLOV, and AsLOV2-SsrA/SspB are sensitive to blue light signals within the 450–470 nm range, leading to dimerization. In contrast, Bphp1/Q-PAS1 responds to near-infrared signals at 750–780 nm, resulting in dimerization (Figure 1A, B). We fused LicT proteins from different sources with photosensitive proteins to identify light-controlled RNA binding protein fusions that can respond to corresponding light signal wavelengths.
Construction strategy of optogenetic circuits based on light-controlled RBPs. (A) Light-induced homodimerization changes in light-sensitive transcription factors, observed in EL222 and LOV. (B) Variations in optical heterodimerization states in response to different wavelengths of light, demonstrated in CRY2/CIB1 and Bphp/PpsR2. Specifically, CRY2/CIB1 is responsive to blue light at 460 nm, while Bphp/PpsR2 is activated by red light at 760 nm. (C) A light-dependent gene expression system utilizing fusion proteins of RBPs and light-sensitive proteins. mCherry fluorescence was measured in strains containing various fusion light-sensitive proteins after 24 hours of exposure under both dark and light-induced conditions. Strains with EL222, NcLOV, AuLOV and CRY2/CIB1 were exposed to LED light at a peak wavelength of 460 nm, whereas strains with Bphp/PpsR2 were exposed to LED light at a peak wavelength of 760 nm. (D) The structure of BsLicT-NcLOV (BsLicV) was constructed in AlphaFold and the docking of BsLicT-NcLOV with RAT-terminator was performed using HDOCK server. (E) PyMOL (version 4.3.0) software was used to analyze the forces between proteins and nucleic acids in three dimensions.
The results (Figure 1C) showed that the recombinant strain CG-RClicV2 containing pJYW-4-RAT-mcherry-LicV2 exhibited the highest activation rate (1.41-fold). Furthermore, a photocontrol system utilizing the NcLOV photocontrol protein showed light induction, with its cofactor being flavin mononucleotide (FMN), a compound abundant in bacterial cells. In contrast, the light control system utilizing the Bphp1/Q-PAS1 NIR photoreceptor protein showed minimal light-induced effect, which may be because the Bphp1/Q-PAS1 system depends on the cofactor biliverdin, which is also present in the intracellular synthesis pathway of C. glutamicum. However, the limited synthesis of biliverdin restricts the regulatory ability of the Bphp1/Q-PAS1 system, necessitating further investigation into the cofactor in subsequent studies. Further studies should explore the relationship between the cofactor biliverdin concentration and the regulatory strength of the light-controlled gene expression system. The BsLicTCAT-NcLOV fusion protein dimer was docked onto the nucleic acid structure using HDOCK (Figure 1D). Three-dimensional forces between proteins and nucleic acids were analyzed using Pymol software (Version 4.3.0) (Figure 1E). In this study, we selected a light-controlled gene expression system based on the BsLicTCAT-NcLOV fusion protein for further experiments, aiming to achieve a light-controlled gene expression system with enhanced activation efficiency.
Fine-tuning the regulatory dynamics of the ‘LightOnC.glu’ system
To enhance the dynamic regulatory range of the light-controlled gene expression system ‘LightOnC.glu’ (Figure 2A), we constructed LicTCAT-VVD fusion protein granule libraries using different linkers. These included Linker-5, linker-10, and linker-15 mutant libraries, each containing mutation sites with 5, 10, and 15 amino acids between the LicTCAT-VVD recombinant proteins within the recombinant strains. All amino acids were randomly mutated to form different amino acid combinations. By employing fluorescence-activated cell sorting (FACS) screening, we identified LicTCAT-VVD fusion protein gene regulation systems with different linkers, each exhibiting varied dynamic ranges of fluorescence response (Figure 2B). From the mutant library, we obtained mutants (B.s.linker 2- B.s.linker 8) exhibiting a greater fluorescence dynamic range than that of the original LicTCAT-VVD combination (B.s.linker1). Sequencing identified several mutants with enhanced FPbg and photodynamic range. The LicTCAT-VVD fusion protein with linker8 (PLAPDCIFSE) showed the highest dynamic change (7.8 times), designated as LicV26, and the corresponding recombinant strain was CG-RClicV26 (Figure 2C). The amino acid sequences of the linker between LicT and VVD are shown in Supplementary Table S3.
Creation and screening of a light-activated switch based on BsLicT-NcLOV. (A) Schematic diagram illustrating the transcriptional regulation by the light-activated switch BsLicT-NcLOV. Blue light irradiation induces the dimerization of the light-switchable RBP, enhancing its binding to the RAT. This interaction prevents the formation of the RNA terminator stem-loop structure, thereby facilitating the transcription of the reporter gene. (B) Strategy for screening the light-activated switch BsLicT-NcLOV. (C) Validation of BsLicT-NcLOV fusion proteins with various linkers, assessed by measuring mCherry expression under light and dark conditions. LicT: plasmid expressing the LicTm protein.
We proceeded by assessing the effect of LicTCAT-VVD expression on the functionality of the light control system, achieved by randomly altering the RBS sequence upstream of the LicV1 coding gene. This involved constructing an RBS sequence mutant library upstream of the LicV1 coding gene, leading to recombinant strains with different LicV1 expression levels. Consequently, we constructed libraries of light-controlled components with different dynamic regulation ranges suitable for specific experimental conditions. First, 192 colonies were randomly selected and inoculated into a 96-well plate under light-induced culture, with subsequent measurement of fluorescence intensity. From these, 42 strains exhibiting the highest fluorescence intensity were cultured in the dark, and the fluorescence intensity was determined (Figure 3A). Mutants (RBS2-RBS13) with expanded dynamic fluorescence ranges were then selected for sequencing, and the RBS sequences are shown in Supplementary Table S4. Among them, the recombinant strain LicTCAT-VVD with the RBS8 sequence (TATAACTTCAGGCAGAGATC) showed the highest dynamic change (14.6 times) and it was labeled as LicV27; its corresponding recombinant strain was CG-RClicV27 (Figure 3B). The expression intensity of different RBS sequences was verified by fluorescence intensity characterization (Supplementary Figure S1). The results show that high LicV expression appears to lead to the formation of light-independent homodimers, which generates high background noise under light conditions, resulting in a reduction of regulatory folds. In contrast, low LicV expression could not form a sufficient amount of homodimers to initiate the transcription process, resulting in limited expression of the target gene. Overall, the regulatory window of the system is highly tunable through alterations in linker sequence and LicV protein expression levels. LicV27, characterized by low leakage expression and a high dynamic range, was selected for further studies.
Fine-tuning the regulatory range of the light-controlled system. (A) A total of 192 colonies were randomly selected and inoculated into a 96-well plate. After 24 h of incubation under blue light induction, fluorescence was measured. The 42 strains exhibiting the highest fluorescence were transferred to a fresh 96-well plate and incubated under dark conditions for 24 h before a second fluorescence measurement. (B) Fine-tuning the regulatory window of the light-controlled system involves modifying the RBS sequence upstream of LicV to adjust the concentration of LicV in the light-controlled system. The mCherry expression from light-controlled systems with varying RBS sequences was measured under both light and dark conditions to calculate the ON/OFF ratio.
Construction of genetic logic gates using the light-controlled gene expression system (development of CRISPRi system to control the central metabolic flux)
Additionally, engineering design concepts were employed to combine different dynamic regulatory elements with gene control elements, leading to the development of multilevel and multifunctional regulatory elements. Our objective was to reverse the activation response of the circuit by creating a ‘NOT’ Boolean logic gate. We utilized the C. glutamicum CRISPR interference (CRISPRi) system, as constructed in a previous study (22), where the inactive Cpf1 (dCpf1) protein bound to and suppressed the transcription of target genes. For the design of this ‘NOT’ gate, we coupled a temperature-sensitive biosensor with a CRISPRi-based logic ‘NOT’ gate to construct a gene circuit capable of repressing metabolic modules.
In this gene circuit, crRNA expression was driven by the constitutive promoter Pj23119, while mCherry was controlled by the strong constitutive promoter of Ptac-M as a gene reporter to test repressive activity (Figure 4A). Figure 4B depicts the mCherry-expressing strains with dCpf1 and corresponding crRNA that were validated by fermentation to exhibit lower mCherry fluorescence intensity than that of the control strains from 8 to 24 h after activating the blue light signal at 8 h. The fluorescence intensity was only 0.19-fold lower than that of the control under total darkness treatment at 24 h, indicating effective suppression of mCherry gene expression (Figure 4B, C). Combining intricate gene circuits with dynamic regulatory elements based on synthetic biology concepts such as ‘non-gates’ and ‘logic gates’ expands the utility of dynamic regulatory tools. This approach aims to achieve simultaneous up- and downregulation of multiple genes, each with different expression times and levels. This simultaneous up and downregulation across multiple genes at different expression times and levels is expected to offer valuable scientific and effective guidance for developing dynamic regulatory systems.
Optogenetic gene interference system in C. glutamicum based on CRISPR-dCpf1. (A) Schematic diagram of the optogenetic gene interference system mediated by CRISPR-FnCpf1. (B) Evaluation of the inhibitory effect of the optogenetic gene interference system using mCherry as the reporter protein. (C) Fluorescence imaging of mCherry at specified time points in C. glutamicum using the optogenetic gene interference system (light on), with strains cultured in the dark as a control (ck). The scale bar represents 10 μm. (D) Schematic diagram illustrating the metabolites (x1–x10) and reactions (v1–v10) used for reprogramming and dynamic model simulation. (E) Dynamic model simulation of UDP-GlcNAc synthesis levels before and after inhibition of the HMP, EMP, and PSP pathways. Random values between 0% and 97%, 0% and 95% or 0% and 77% were generated as the inhibition strengths for zwf, pfkA or murAA, respectively. The intercellular levels of GlcN-6-P were studied under these three conditions (weakening zwf, pfkA, or murAA) through 1000 simulations (lines represent the mean values and shaded areas represent the standard deviations).
To better understand the effect of gene expression intensity on the metabolic network, a linear pathway kinetics model was constructed. This model integrated the UDP-GlcNAc synthesis pathway, characterized by the Michaelis–Menten equation, to simulate UDP-GlcNAc synthesis – a key precursor for multiple functional sugars. Kinetic modeling simulations were conducted to inhibit three pathways (HMP, EMP and PSP) (Figure 4D). In these experiments, random values between 0 and 97%, 0–95% or 0–77% were generated to represent the inhibition strength of zwf, pfk or murAA, respectively. The intracellular levels of UDP-GlcNAc were then investigated through 1000 simulations for each case (weakened zwf, pfk or murAA). Figure 4D illustrates that inhibiting EMP or PSP pathways resulted in a more significant enhancement of UDP-GlcNAc levels than that of the HMP pathway. Additionally, three randomly generated deterrent intensities were simultaneously integrated into the model, resulting in a greater accumulation of UDP-GlcNAc (Figure 4E). The predicted UDP-GlcNAc concentration positively correlated with the deterrent strength generated by zwf, pfk and murAA. Significantly, a synergistic effect was observed when simultaneously inhibiting HMP, EMP and PSP pathways (Figure 4E). Building on these simulation results, we combined these three pathways to further increase the UDP-GlcNAc levels.
Dynamic regulation improves CHOSs and CSA production
To validate whether light-signal-responsive genetic circuits could control cellular metabolism, the de novo biosynthesis pathway of CHOSs and CSA was introduced into C. glutamicum. The biosynthesis pathway for functional sugars such as CHOSs and CSA involves a variety of key node compounds in central carbon and nitrogen metabolism and the PSP. Conventional metabolic engineering strategies struggle to achieve dynamic and global regulation of metabolic flux and can interfere with bacterial cell wall synthesis, affecting cell growth and metabolism.
In order to construct the synthesis pathway of chitosan oligosaccharide, it is necessary to introduce the key enzyme coding genes (glucose-6-phosphate acetyltransferase-encoding gene GNA1/glucose-6-phosphate phosphatase-encoding gene yqaB/N-acetylglucosamine transferase-encoding gene nodC) into C. glutamicum. The key enzyme coding genes (chondroitin synthase-encoding gene KfoC/UDP-glucosamine isomerase-encoding gene KfoA/sulfotransferase-encoding gene C4ST) were introduced into C. glutamicum, which improved the synthesis pathway of CSA. Although CHOSs and CSA have different genes related to the synthesis pathway, they have the same competitive pathways related to cell growth in the process of biosynthesis. By dynamically regulating the pathways related to cell growth, the cells can increase the metabolic flux required for the synthesis of target products under the condition of satisfying growth.
Therefore, we employed a light-controlled CRISPRi gene expression regulation tool to dynamically regulate the expression of genes zwf (introducing metabolic flow into the pentose phosphate pathway (PPP)), pfk (introducing metabolic flow into the glycolysis pathway) and murAA (introducing metabolic flow into the PSP), which are related to key competing pathways (Figure 5A). The light-control gene-disrupting bacteria CGCH3-CHOS-zwf/pfk/murAAi-R and CGCS3-CSA-zwf/pfk/murAAi-R were constructed, and the light-control gene-disrupting bacteria CGCH3-CHOS-zwf/pfk/murAAi and CGCS3-CSA-zwf/pfk/murAAi were constructed with the strains CGCH3-CHOS-zwf/pfk/murAAi and CGCS3-CSA-zwf/pfk/murAAi. The strains CGCH3- GNA1-yqaB-nodC and CGCS3-kfoC-kfoA-C4ST were used as controls. The CHOSs titer of the recombinant strain CGCH3-CHOS-zwf/pfk/murAAi-R reached 1288.4 mg/L, while that of CGCH3-CHOS-zwf/pfk/murAAi was 542.7 mg/L and that of CGCH3-GNA1-yqaB-nodC was 251.8 mg/L. The CSA titer of CGCS3-CSA-zwf/pfk/murAAi-R reached 291 mg/L, while that of CGCS3-CSA-zwf/pfk/murAAi was 87.3 mg/L and that of CGCS3-kfoC-kfoA-C4ST was 49.7 mg/L (Figures 5 and 6).
Dynamic regulation of CHOSs biosynthesis. (A) Schematic representation of the de novo biosynthetic metabolic network for CHOSs, highlighting the PPP, EMP and PSP modules as endogenous pathways that compete with the CHOSs synthesis pathway. A light-responsive genetic circuit was developed using a CRISPRi-based NOT gate as a light-signal-responsive biosensor. (B) Mass spectra of GlcNAc, CHOS2, CHOS3, CHOS4 and CHOS5 in the samples. (C) Optical density at 600 nm (OD600 nm) and CHOS synthesis efficacy of the CHOSs-producing strains CG-GNA1-yqaB-nodC, CG-CHOS-zwf/pfk/murAAi, and CG-CHOS-zwf/pfk/murAAi-R under different fermentation processes. Data are presented as the mean ± standard deviation (s.d.) of three replicates.
Dynamic regulation of CSA biosynthesis. (A) Schematic representation of the de novo biosynthetic metabolic network for CSA, highlighting the PPP, EMP and PSP modules as endogenous pathways that compete with the CHOSs synthesis pathway. A light-responsive genetic circuit was constructed using a CRISPRi-based NOT gate as a light-signal-responsive biosensor. (B) Mass spectra of CS and CSA in the samples. (C) Optical density at 600 nm (OD600 nm) and CSA synthesis efficacy of the CSA-producing strains CG-kfoC-kfoA-C4ST, CG-CSA-zwf/pfk/murAAi and CG-CSA-zwf/pfk/murAAi-R under different fermentation processes. Data are presented as the mean ± standard deviation (s.d.) of three replicates.
The PPP, the glycolysis pathway/ the Embden–Meyerhof pathway (EMP) and the peptidoglycan synthesis pathway (PSP) are the three natural competing pathways for the biosynthesis of CHOSs, CSA and most functional carbohydrate chemicals in C. glutamicum. All three pathways are essential for cell growth and, therefore, cannot be knocked out. Therefore, we created inhibition cascades composed of crRNA that are not controlled by the light-controlled RBPs, targeting the key genes zwf of PPP, pfk of the glycolytic pathway, and murAA of the peptidoglycan pathway. To fine-tune repressive activity, three crRNAs were designed to target the initial sites of the zwf, pfk and murAA coding regions. The fermentation results showed that the recombinant strains CG-CHOS-zwf/pfk/murAAi and CG-CSA-zwf/pfk/murAAi, which circumvented the light control system of RBPs, enhanced the titer of the target product of the original strain. However, the growth biomass of the strain exhibited a significant decrease in biomass and hindered growth, unlike the wild-type strain. In contrast, recombinant strains CG-CHOS-zwf/pfk/murAAi-R and CG-CSA-zwf/pfk/murAAi-R, regulated by the light-controlled RBPs, showed improved titer of the target product as well as better strain growth than that of the non-light-controlled strains (Figure 5 and 6). The delayed activation of the inhibitory circuit might minimize disruptions to normal cell growth, leading to enhanced target product synthesis.
Production of CHOSs in a 5-liter light-controlled bioreactor
We conducted a scale-up test of a CHOSs production strain utilizing a light-controlled system in a 5-liter fermentor to advance the transition of light-regulated gene expression from laboratory settings to industrial use. In the turbulent stirring bioreactor, we designed an external annular illumination device (LED strip) outside a 5-L glass bioreactor to provide the required light source, adjusting illumination intensity by controlling the number or power of annular LED tubes. By assessing light intensity at different positions within the fermenter across different lighting conditions, the spatial distribution of the light intensity in the photoreactor was obtained (Figure 7A, B). We maintained an average light intensity of 2.5 mW/cm2 blue light (peak at 460 nm) in the tank. Two strains, CG-CHOS-zwf/pfk/murAAi, and CG-CHOS-zwf/pfk/murAAi-R, were employed in the scaled-up assay. During fed-batch fermentation, a glucose solution (500 g/L) was added to the fermentation process to maintain glucose concentration in the fermentation broth at approximately 5 g/L. The maximum CHOSs titers in CG-CHOS-zwf/pfk/murAAi and CG-CHOS-zwf/pfk/murAAi-R reached 2.5 and 6.2 g/L, respectively (Figure 7C). RT-PCR was used to measure the relative transcription levels of genes regulated by CG-CHOS-zwf/pfk/murAAi-R at different fermentation stages. After 20 h of growth,the mRNA levels of zwf, pfk, and murAA decreased to 63%, 41.3% and 78.7% of those at 8 hours under dark conditions (Figure 7D). A kinetic model simulation (Figure 7E) was conducted to inhibit the three pathways (HMP, EMP, and PSP). Based on the experimental results, repression intensities of zwf, pfk, or murAA were randomly generated between 0 and 97%, 0 and 95%, and 0 and 77%, respectively. The CHOSs levels were then studied across 1000 simulations for these three cases (weakened zwf, pfk, or murAA). Figure 7F illustrates that integrating three randomly generated repression intensities simultaneously into the model led to an increased accumulation of CHOSs than that of inhibiting HMP, EMP, or PSP alone. These results highlighted the importance of timely regulating metabolic flux in key metabolic modules in cell growth, which plays a vital role in the biosynthesis of functional sugars, such as CHOSs.
Production of CHOSs in a 5-L fermenter. (A) Configuration of stirred-tank reactor light sources during cultivation (left, figure created by FigDraw) and spatial distribution of light field inside the bioreactor under illumination with different power light sources (right). (B) CHOS production by strains CG-CHOS-zwf/pfk/murAAi and CG-CHOS-zwf/pfk/murAAi-R after 75 h of cultivation in a 5-L fermenter. (C) CHOSs synthesis curve and cell growth curve of the recombinant engineering strain CG-CHOS-zwf/pfk/murAAi-R in a 5-L fermenter. (D) Relative normalized expression of zwf, pfk, and murAA in strain CG-CHOS-zwf/pfk/murAAi-R before and after light signal control (samples taken at 8 and 20 h, with blue light activation at 10 h). Data are presented as the mean ± standard deviation (s.d.) of three replicates. (E) Schematic diagram illustrating the metabolites (x1–x11) and reactions (v1–v11) used for reprogramming and dynamic model simulation. (F) Dynamic model simulation of CHOSs synthesis levels before and after inhibition of the HMP, EMP, and PSP pathways. Random values between 0% and 97%, 0% and 95% or 0% and 77% were generated as the inhibition strengths for zwf, pfk and murAA, respectively. Intracellular levels of GlcN-6-P were analyzed under these conditions (weakening zwf, pfk or murAA) through 1000 simulations (lines represent the mean values; shaded areas represent the standard deviations).
Discussion
Precise regulation of gene expression is crucial for understanding the life activities of organisms. Traditionally, gene regulation methods have relied on chemical molecular inducers. However, this approach faces considerable challenges, including the toxicity of chemical molecules to cells, non-specific gene regulation and the complexities of downstream separation and purification (10,31). These limitations have restricted their industrial applicability. Optogenetics, a promising synthetic biology tool, has emerged as a solution by enabling control over metabolic flux distribution (17,32). It addresses the drawbacks of conventional dynamic regulation tools, such as poor generalizability, limited regulation range, slow response speed, and difficulty in separation and purification (33). Optogenetics extends the scope of dynamic regulation methods, making it an attractive option in industrial applications. Current methods of light-based gene regulation typically rely on the recognition of target DNA sequences (usually located in the promoter region) using light-controlled transcription factors (34,35). However, the limited options for designing DNA target sequences limit the widespread use of optogenetic tools depending on light-controlled factors.
In this study, we introduced ‘LightOnC.glu,’ the first optogenetic tool for C. glutamicum, to the best of our knowledge. This is the inaugural report of using RNA-based binding proteins in conjunction with photosensitive proteins to regulate gene expression in C. glutamicum. By adjusting the parameters of the expression components, we constructed a series of highly adjustable light-control tools characterized by their regulatory window, activation kinetics and photosensitivity, providing flexibility under various experimental conditions. The gene expression regulation system using light-controlled transcription factors and RBPs exhibits clear advantages over existing light-activated systems. Firstly, it does not rely on specific DNA-binding motifs of promoters and avoids fusing transcription factor operon sequences with regulatory gene promoters. Thus, eliminating interference with promoter activity and enhancing versatility across different chassis cells and gene expression conditions. Secondly, regulating metabolism at the RNA transcription level proves more efficient than at the DNA to mRNA or mRNA to protein stages. Thirdly, the system's design is simple and compact, consisting only of the photosensor component, which simplifies genetic engineering operations. Additionally, its highly tunable induction characteristics—such as window adjustment, activation kinetics and photosensitivity—offer adaptability for various experimental setups. Finally, integrating a light-controlled gene expression system into the fermentation process at the bioreactor scale enhances the design and operation of bioreactors. During this period, the light signal can be applied or removed immediately to control batch or fed-batch fermentation processes.
Based on this, we combined the light-controlled gene expression regulation system with the CRISPR/Cpf1 system to develop a high-performance light-controlled gene interference system. This system downregulates metabolic flux in competitive pathways, redirecting it towards the synthetic pathway of the target metabolite. Employing the biosynthesis of CHOSs and CSA as examples, we rationally designed the metabolic flow using a gene expression regulation system based on light-controlled RBP, enabling efficient synthesis of these compounds in C. glutamicum. This study not only fills the gap in light-controlled gene regulation tools for C. glutamicum, but it also shows that the light-controlled RBP gene expression system effectively enhances the synthesis of target chemicals, offering significant practical value in metabolic engineering.
However, some work requires further research. In light-controlled bioreactors, the penetration depth of light signals is usually limited. In order to solve this problem, it is necessary to improve the mass transfer and regulation performance from the aspects of optical control system expression elements and bioreactor design. First, for photosensitive proteins, it is necessary to analyze and remodel the protein structure in detail, and develop photosensitive protein variants with higher light sensitivity, so that less light intensity in the bioreactor can achieve stronger regulation effect. Additionally, additional research is needed to clarify how the gene circuit responds to the complex interplay of light patterns and to develop a statistical model that links the light-controlled RBP expression system with target gene expression levels. This model will be instrumental in optimising light-controlled fermentation processes. Moreover, exploring other RNA-binding structural domains and photosensitive proteins that respond to different wavelengths could extend the application spectrum of RBP-based photoregulatory systems in dynamic regulation. Expanding the use of photoconvertible RBPs in biological research would enable the simultaneous regulation of multiple RNAs, enhancing our understanding of complex biological processes. In addition, high-performance computer clusters are necessary for simulating computational fluid dynamics to account for different parameters that react to temporal and spatial dynamic changes in the light and flow fields of light-controlled cells. Implementing an artificial neural network to analyze this dataset could develop a black-box model capable of predicting flow field characteristics under any set of operating conditions within the normal operating range. This model could support real-time process control and fermentation process analysis, guiding the rational design and optimization of light patterns and reactor structures in light-controlled bioreactors, tailored to the metabolic performance of the light-controlled cells.
Conclusion
In conclusion, we have developed ‘LightOnC.glu’, the inaugural light-controlled gene expression system for C. glutamicum, characterized by its single-component structure, direct activation, highly tunable induction properties, precise spatiotemporal resolution, and strong adaptability. This system precisely regulates RNA metabolism to control gene circuits, enhancing biomanufacturing efficiency by optimizing artificial metabolic pathways in host cells. By rationally designing light-controlled bioreactors and optimising the fermentation process, we have successfully scaled light-regulatory gene expression systems from laboratory settings to industrial applications.
Data availability
The MATLAB code used in this study can be found on GitHub (https://github.com/Arturia123/Dynamic-model-of-metabolic-pathways-related-to-UDP-GlcNAc-and-CHOSs) and Zenodo (https://doi.org/10.5281/zenodo.14015764). The DNA sequences of plasmid with the necessary annotations used in this study are shown in Supplementary Note.
Supplementary data
Supplementary Data are available at NAR Online.
Acknowledgements
Author contributions: Chen Deng: Conceptualization, formal analysis, methodology, validation and writing—original draft. Ruijie Xin: Conceptualization, formal analysis and visualization. Xingjian Li: Formal analysis. Jie Zhang: Formal analysis and methodology. Liqiang Fan and Yongjun Qiu: Supervised the project. Liming Zhao: Supervision, validation and writing – review & editing.
Funding
This work was financially supported by the National Natural Science Foundation of China [32301214, 32327801]; Chenguang Program of Shanghai Education Development Foundation and Shanghai Municipal Education Commission.
Conflict of interest statement. The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.
Comments