A cross-species inducible system for enhanced protein expression and multiplexed metabolic pathway fine-tuning in bacteria

Abstract

Inducible systems are crucial to metabolic engineering and synthetic biology, enabling organisms that function as biosensors and produce valuable compounds. However, almost all inducible systems are strain-specific, limiting comparative analyses and applications across strains rapidly. This study designed and presented a robust workflow for developing the cross-species inducible system. By applying this approach, two reconstructed inducible systems (a 2,4-diacetylphloroglucinol-inducible system P_phlF3R1 and an anhydrotetracycline-inducible system P_tet2R2*) were successfully developed and demonstrated to function in three model microorganisms, including Escherichia coli, Bacillus subtilis and Corynebacterium glutamicum. To enhance their practicality, both inducible systems were subsequently placed on the plasmid and genome for detailed characterization to determine the optimal expression conditions. Furthermore, the more efficient inducible system P_tet2R2* was employed to express various reporter proteins and gene clusters in these three strains. Moreover, the aTc-inducible system P_tet2R2*, combined with T7 RNA polymerase and dCas12a, was utilized to develop a single-input genetic circuit that enables the simultaneous activation and repression of gene expression. Overall, the cross-species inducible system serves as a stringent, controllable and effective tool for protein expression and metabolic pathway control in different bacteria.

Graphical Abstract

Graphical Abstract.

Introduction

Inducible systems enable precise control over gene activation and repression through the addition of an inducer, serving as fundamental components of metabolic engineering and synthetic biology. Compared to constitutive expression systems, inducible systems facilitate dynamic gene regulation (1), reducing the metabolic burden on host cells and improving the production of various industrial products, including recombinant proteins (2), platform chemicals (3) and biopolymers (4). Moreover, inducible systems are instrumental in designing complex genetic circuits, such as toggle switches (5), oscillators (6–8), autoregulatory latches (9) and logic gates (10), facilitating precise control of living cells by applying a programming mindset to biological systems (11). Recently, inducible systems have been widely applied in model organisms such as Escherichia coli, Bacillus subtilis and Saccharomyces cerevisiae (12–14). For example, 12 highly optimized small-molecule inducible systems were developed in E. coli using a directed evolution strategy, successfully achieving lower background levels, a high dynamic range, increased sensitivity and low crosstalk (15). These systems were then utilized to construct ‘Marionette’ strains and optimize a biosynthetic pathway to synthesize lycopene (15). Additionally, a maltose-inducible system for B. subtilis was constructed by combining promoter-based mutagenesis with host-specific metabolic engineering of transactivation components, resulting in stringent, robust and homogeneous gene regulation upon maltose induction (16). These inducible systems are developed through a comprehensive workflow that combines rational design and random mutagenesis, minimizing effort, accelerating development and enhancing theoretical research and applications in model microorganisms (17,18).

With the rapid development in genomics and synthetic biology, microbial research has increasingly multiple species (19,20). Comparative genomic analyses across species have revealed gene variations, functional alterations and their associations with phenotypic traits (21). However, the use of different expression systems across species often results in complex experimental designs, labor-intensive construction processes and reduced comparability of results. For instance, in developing Mismatch-Clustered Regularly Interspaced Short Palindromic Repeats interference (Mismatch-CRISPRi) system to investigate co-variation expression-fitness relationships of essential genes, Hawkins et al. had to construct an isopropyl β-D-1-thiogalactopyranoside (IPTG)-inducible system for expression of dCas9 in E. coli and a xylose-inducible system for the same purpose in B. subtilis (22). Moreover, when expressing many novel enzyme-encoding genes (23), biosynthetic pathways or uncharacterized DNA fragments (24), variations in expression levels and intracellular environments across strains can significantly affect protein functionality. Although tools such as CRISPR systems (e.g. Cas9 and Cas12a) and T7 expression systems have been widely applied across various strains, their implementation usually necessitates independent designs for each strain, adding complexity to the experimental workflow (4,25–27). Despite the construction of numerous inducible systems in various strains in recent years, most of these systems remain strain-specific (15,16). Few cross-species inducible systems have been reported, and these are primarily designed for specific metabolites or constructed and optimized mainly in one strain, with functionality verified in other strains (28). Due to their poor performance, these cross-species inducible systems cannot facilitate comparative analyses or applications across different strains.

In this study, a robust workflow for creating cross-species inducible system was developed and successfully implemented in representative organisms, including E. coli, B. subtilis and C. glutamicum, resulting in the creation of 2,4-diacetylphloroglucinol (DAPG) and anhydrotetracycline (aTc)-inducible systems. These systems were meticulously optimized and characterized to determine induction concentration across the three organisms. The aTc-inducible system P_tet2R2* demonstrated superior performance in all three strains, including low leakage, broad dynamic range, sufficient expression intensity and appropriate sensitivity. Then, the more efficient inducible system P_tet2R2*was employed to effectively regulate the expression of various reporter proteins (sfGFP, mCherry and mScarlet3) and gene clusters (crtEIB, crtEIBY and vioABCDE) in these three strains. Additionally, a single-input genetic circuit was developed based on T7 RNA polymerase (T7 RNAP) and dCas12a for simultaneous activation and repression of gene expression. Overall, the cross-species inducible system developed in this study demonstrated low leakage, high dynamic range and robust expression in both E. coli, B. subtilis and C. glutamicum, offering highly efficient tools for applications in synthetic biology and metabolic engineering.

Materials and methods

Chemicals and reagents

All chemicals used in this study were obtained from Sangon Biotech (Shanghai, China) unless otherwise specified. DAPG (S22165) and aTc (S25906) were obtained from Yuanye Bio-Technology (Shanghai, China). Plasmids extraction kit, DNA gel purification kit and restriction enzymes were obtained from Thermo Fisher Scientific (Waltham, USA). PrimeSTAR HS DNA polymerase was obtained from Takara Biomedical Technology (Beijing, China). Hieff Canace® Gold High-Fidelity DNA Polymerase (Cat No.10148) was obtained from Yeasen Biotechnology (Shanghai) Co., Ltd. T4 DNA ligase was obtained from BiOligo Biotechnology Shanghai. StarLighter HotStart Taq Pro Polymerase Chain Reaction (PCR) Mix was purchased from Beijng Foreverstar Biotech. Seamless Cloning Kit was obtained from Beyotime Biotechnology (Shanghai, China). Oligonucleotides were synthesized by GENEWIZ (Suzhou, China). Luria–Bertani medium (10 g·L^–1 tryptone, 5 g·L^–1 yeast extract and 5 g·L^–1 NaCl) was used for E. coli and B. subtilis cultivation. LB broth supplemented with 18.5 g·L⁻¹ of brain heart infusion (LBB) was used for C. glutamicum cultivation. Ampicillin (100 μg·ml^–1), spectinomycin (100 μg·ml^–1) and kanamycin (50 μg·ml^–1) were used for selection in E. coli; Kanamycin (25 μg·ml^–1) was used for selection in C. glutamicum; Kanamycin (50 μg·ml^–1) and chloromycetin (5 μg·ml^–1) were used for selection in B. subtilis.

Strains construction

Table 1 lists all strains used in this study. E. coli JM109 was utilized for plasmid construction. Plasmids were constructed using the T4 DNA ligase or the Seamless Cloning Kit according to the manufacturer’s instructions. E. coli K12 and C. glutamicum ATCC 13032 were used as expression hosts. B. subtilis G600 (B. subtilis 168 Δepr::XylR-P_xylA-comK-comS, ΔtrpC2::trpC0, ΔgudB::gudB⁺, aprE⁰, nprE⁰, bpr⁰, mpr⁰, nprB⁰), constructed in our previous study (29), was also used as the expression host. The expression strains used for inducible systems characterization were constructed by transforming the expression plasmids to E. coli K12, G600 and C. glutamicum ATCC 13032. All recombinant C. glutamicum strains were transformed via electroporation (30). All recombinant B. subtilis were transformed by a previously reported method (31). The construction of strains based on the CRISPR/Cas12a gene editing system was conducted according to previously reported methods (26,27,32), which included the techniques for creating a CRISPR RNA array and editing the genome with the assistance of CRISPR/Cas12a.

Table 1.

Strains used in this study

Strain	Characteristic	Source
Strains
E. coli JM109	recA1, endA1, thi, gyrA96, supE44, hsdR17Δ (lac-proAB)/F′[traD36, proAB+, laclq, lacZΔ M15]	Lab stock
E. coli K12	F-, λ-, ilvG-, rfb-50, rph-1	Lab stock
K12-PphlF3R1-sfGFP	E. coli K12 derivate, pHT-PphlF3R1-sfGFP	This study
K12-Ptet2R2*-sfGFP	E. coli K12 derivate, pHT-Ptet2R2*-sfGFP	This study
C. glutamicum ATCC 13 032	Wild-type (WT) C. glutamicum strain	Lab stock
13032-PphlF3R1-sfGFP	C. glutamicum ATCC 13032 derivate, pJYW-PphlF3R1-sfGFP	This study
13032-Ptet2R2*-sfGFP	C. glutamicum ATCC 13032 derivate, pJYW-Ptet2R2*-sfGFP	This study
G600	B. subtilis 168 derivate, Δepr::XylR-PxylA-comKS, ΔtrpC2::trpC0, ΔgudB::gudB+, aprE0, nprE0, bpr0, mpr0, nprB0	(29)
G600-PphlF3R1-sfGFP	G600 derivate, pHT-PphlF3R1-sfGFP	This study
G600-Ptet2R2*-sfGFP	G600 derivate, pHT-Ptet2R2*-sfGFP	This study
K12T7T	E. coli K12 derivate, leuO::Ptet2R2*-T7 RNAP	This study
BST7T	G600 derivate, ydjC::Ptet2R2*-T7 RNAP	This study
K12T7Tt	K12T7T derivate, leuO::Ptet2R2*-T7 RNAP, gsk::Ptet2R2*-dCas12a	This study
BST7Tt	BST7T derivate, ydjC::Ptet2R2*-T7 RNAP, lacA::Ptet2R2*-dCas12a	This study
BST7Tg	BST7T derivate, ydjC::Ptet2R2*-T7 RNAP, lacA::Pgrac1R-dCas12a	This study
K12T7Tt-msc	K12T7Tt derivate, leuO::Ptet2R2*-T7 RNAP, gsk::Ptet2R2*-dCas12a, plfA::Pveg-mScarlet3	This study
BST7Tt-msc	BST7Tt derivate, ydjC::Ptet2R2*-T7 RNAP, lacA::Ptet2R2*-dCas12a, ytoA::Pveg-mScarlet3	This study
BST7Tg-msc	BST7Tt derivate, ydjC::Ptet2R2*-T7 RNAP, lacA:: Pgrac1R-dCas12a, ytoA::Pveg-mScarlet3	This study
R1	BST7Tt derivate, ydjC::Ptet2R2*-T7 RNAP, lacA::Ptet2R2*-dCas12a, PT7tet-ribDEAHT	This study
R2	R2 derivate, ydjC::Ptet2R2*-T7 RNAP, lacA::Ptet2R2*-dCas12a, PT7tet-ribDEAHT, PT7tet-purE	This study
R11	R1 derivate, ydjC::Ptet2R2*-T7 RNAP, lacA::Ptet2R2*-dCas12a, PT7tet-ribDEAHT, pHT-PT7tet-gdh	This study
R12	R1 derivate, ydjC::Ptet2R2*-T7 RNAP, lacA::Ptet2R2*-dCas12a, PT7tet-ribDEAHT, pHT-Ptet2R2*-CrribC	This study
R13	R1 derivate, ydjC::Ptet2R2*-T7 RNAP, lacA::Ptet2R2*-dCas12a, PT7tet-ribDEAHT, pHT-Ptet2R2*-CrribC-PT7tet-gdh	This study
R21	R2 derivate, ydjC::Ptet2R2*-T7 RNAP, lacA::Ptet2R2*-dCas12a, PT7tet-ribDEAHT, pHT-PT7tet-gdh	This study
R22	R2 derivate, ydjC::Ptet2R2*-T7 RNAP, lacA::Ptet2R2*-dCas12a, PT7tet-ribDEAHT, pHT-Ptet2R2*-CrribC-PT7tet-gdh	This study
R23	R2 derivate, ydjC::Ptet2R2*-T7 RNAP, lacA::Ptet2R2*-dCas12a, PT7tet-ribDEAHT, pHT-Ptet2R2*-CrribC-PT7tet-gdh	This study
L1	BST7Tt derivate, ydjC::Ptet2R2*-T7 RNAP, lacA::Ptet2R2*-dCas12a, PT7tet-ispDF, pHT-PT7tet-crtEIB	This study
L2	BST7Tt derivate, ydjC::Ptet2R2*-T7 RNAP, lacA::Ptet2R2*-dCas12a, PT7tet-fni, pHT-PT7tet-crtEIB	This study
L3	BST7Tt derivate, ydjC::Ptet2R2*-T7 RNAP, lacA::Ptet2R2*-dCas12a, PT7tet-dxS, pHT-PT7tet-crtEIB	This study
L4	BST7Tt derivate, ydjC::Ptet2R2*-T7 RNAP, lacA::Ptet2R2*-dCas12a, PT7tet-dxR, pHT-PT7tet-crtEIB	This study
L5	L1 derivate, ydjC::Ptet2R2*-T7 RNAP, lacA::Ptet2R2*-dCas12a, PT7tet-ispDF, PT7tet-fni, pHT-PT7tet-crtEIB	This study
L6	L1 derivate, ydjC::Ptet2R2*-T7 RNAP, lacA::Ptet2R2*-dCas12a, PT7tet-ispDF, PT7tet-dxS, pHT-PT7tet-crtEIB	This study
L7	L1 derivate, ydjC::Ptet2R2*-T7 RNAP, lacA::Ptet2R2*-dCas12a, PT7tet-ispDF, PT7tet-dxR, pHT-PT7tet-crtEIB	This study
L8	L5 derivate, ydjC::Ptet2R2*-T7 RNAP, lacA::Ptet2R2*-dCas12a, PT7tet-ispDF, PT7tet-fni, PT7tet-dxS, pHT-PT7tet-crtEIB	This study
L9	L5 derivate, ydjC::Ptet2R2*-T7 RNAP, lacA::Ptet2R2*-dCas12a, PT7tet-ispDF, PT7tet-fni, PT7tet-dxR, pHT-PT7tet-crtEIB	This study
L10	L8 derivate, ydjC::Ptet2R2*-T7 RNAP, lacA::Ptet2R2*-dCas12a, PT7tet-ispDF, PT7tet-fni, PT7tet-dxS, PT7tet-dxR, pHT-PT7tet-crtEIB	This study
L11	L10 derivate, ydjC::Ptet2R2*-T7 RNAP, lacA::Ptet2R2*-dCas12a, PT7tet-ispDF, PT7tet-fni, PT7tet-dxS, PT7tet-dxR, pHT-Ptet2R2*-CrhepS-PT7tet-crtEIB	This study

Plasmids construction

Supplementary Table S1 lists all sequences of promoters, and Supplementary Table S2 lists all plasmids used in this study. The expression plasmids discussed in the study primarily belong to the pHT and pJYW series, used for E. coli, B. subtilis and C. glutamicum expression, respectively. The subsequent sections would focus on the construction of pHT, as the procedures for pJYW were similar and would not be detailed further. Nine inducible expression cassettes were amplified from the genome of Marionette-Wild and BST7L/R by PCR (4,15). Followed by Gibson assembly (33), the nine linearized fragments were cloned on vector pHT-GP0s and pJYW4, which resulted in the original plasmids (pHT-P_grac-sfGFP, pHT-P_xylA-sfGFP, pHT-P_tet-sfGFP, pHT-P_phlF-sfGFP, pHT-P_cym-sfGFP, pHT-P_van-sfGFP, pHT-P_betl-sfGFP, pHT-P_ttg-sfGFP and pHT-P_pca-sfGFP). pHT-P_grac-sfGFP and pHT-P_xylA-sfGFP were reconstructed to pHT-P_grac-sfGFP-(a-i) and pHT-P_xylA-sfGFP-(a-i) by circular plasmid PCR. The remaining seven plasmids were first subjected to circular plasmid PCR to remove the repressor cassette to obtain pHT-P_inducer-sfGFP0, and then reconstructed using circular PCR on this basis to obtain the promoters (P_phlF3, P_cym1, P_van2, P_van3, P_betl2, P_ttg1, P_pca2, P_tet2) that can be expressed in B. subtilis and C. glutamicum. Three different constitutive promoters (P_SPL-L2, P_spovG and P_xylR) and seven repressor proteins were combined and ligated to the corresponding eight promoters to obtain the expression plasmids (e.g. pHT-P_phlF3R1-sfGFP) carrying a complete inducible system. To mutate and screen the aTc-inducible system, the ORI-cgl replication origin was amplified and ligated into pHT-P_tet2R2-sfGFP, creating the shuttle plasmid pHTcgl-P_tet2R2-sfGFP. After the screening, the mutant pHTcgl-P_tet2R2*-sfGFP was obtained. For plasmid stability, the P_tet2R2*-sfGFP cassette was re-cloned into pHT and pJYW for further modification and characterization. The linearized fragments of mcherry, mScarlet3, vioABCDE and crtEIBY were amplified and cloned on vector pHT-P_tet2R2*-sfGFP, pHT-T7Ts and pHT-T7Ds.

Analysis of fluorescence intensity

Recombinant E. coli and B. subtilis strains with fluorescent proteins were precultured in Luria-Bertani medium for 10 h and further inoculated into 200 μl of Luria–Bertani medium with 1% proportion in 96-well plates (Corning, 3603). The 96-well plates were subsequently incubated on a Titramax-1000 multi-plate shaker at 750 rpm at 37°C (Heidolph instrument, Schwabach, German) (34). The culture medium used for cultivating C. glutamicum was LBB medium, with an incubation temperature of 30°C. Then, sfGFP fluorescence (excitation, 480 nm; emission, 516 nm) and OD₆₀₀ (the value of absorbance at 600 nm) were measured using a microplate multimode reader (BioTek, Cytation 3) directly (35). To calculate the relative fluorescence intensities, background OD of the medium (OD_bg) and background fluorescence of the strain without fluorescent protein expression (FP_bg) was excluded, and equation (1) was applied:

(1)

Dose–response curves to inducible and T7-inducible systems were fitted to the equation (2):

(2)

where is the inducer concentration, is the relative expression activity of the promoter ( and are the maximum/minimum activities), is the threshold, and is the cooperativity.

Construction of a mutated repressor expression library

Error-prone PCR was used to introduce random mutations into the coding sequence of the repressor, while, at the same time, a degenerate primer was employed to introduce an RBS library (TAACTACATAAATMAGGMRGWATCAC), thereby expanding the diversity of the library. Subsequently, the resulting mixed library was ligated into the backbone of shuttle plasmid pHTcgl-P_tet2R2-sfGFP using the Gibson assembly, and the mixed plasmids were then transformed into E. coli. After transformation, the individual colonies were cultured in a 96-well plate for 24 h. Next, colonies that exhibited low leaky expression without aTc but strong expression with aTc were chosen. Plasmids from these selected colonies were transformed into B. subtilis and C. glutamicum, where the expression levels before and after induction were similarly assessed. Finally, mutants showing low leakage and high induced expression in both E. coli, B. subtilis and C. glutamicum were sequenced.

Batch fermentation in shake flasks

For protein fermentation, the single colony of strains with the protein (sfGFP, mcherry and mScarlet3) was inoculated into the Luria–Bertani medium. After 12 h of incubation at 37°C and 220 rpm, seed cultures were obtained. Then, 2 ml of seed cultures were added to 38 ml of Terrific-broth medium (24 g·L^–1 yeast extract, 12 g·L^–1 tryptone, 4 ml·L^–1 glycerin, 16.43 g·L^–1 K₂HPO₄·3H₂O and 2.31 g·L^–1 KH₂PO₄) and continued to grow for an additional 48 h. The appropriate antibiotics were added to the culture medium to maintain plasmid replication.

For lycopene, β-carotene, violacein and riboflavin fermentation, seed cultures were prepared in the same manner as protein fermentation. The fermentation medium (85 g·L^–1 glucose, 6 g·L^–1 urea, 15 g·L^–1 yeast extract, 15 g·L^–1 tryptone, 12.5 g·L^–1 K₂HPO₄·3H₂O, 2.5 g·L^–1 KH₂PO₄ and 3 g·L^–1 MgSO₄) was used for shake flask fermentation. Glucose and MgSO₄ were sterilized separately and added to the shake flask before inoculation. Moreover, appropriate inducer DAPG and aTc were supplied to open the inducible systems.

Quantitative reverse-transcription PCR

A RNA preparation kit was used to extract RNA from each strain, and a reverse-transcription RT-PCR kit was used to generate complementary DNA as a qPCR template. The cell was cultured in a 10 ml shake tube containing 2 ml of LB medium at 30°C or 37°C with shaking at 220 rpm for 24 h. Lysozyme was used for the disruption of bacterial cell walls. Briefly, the high concentration of RNA was then obtained by buffer YCA purification and ethanol precipitation. The OD₂₆₀/OD₂₈₀ ratio of RNA was ensured to be 2.0, and the RNA concentration of each sample was diluted in the same range (315 ng/μl). TB Green qPCR was used for reverse transcription, and 1 μg of RNA was added to each reaction. The total reaction volume of quantitative reverse-transcription polymerase chain reaction (RT-qPCR) was 20 μl. The RpoB gene of E. coli, the rpsJ gene of B. subtilis and the gap gene of C. glutamicum were selected as the internal reference genes for quantitative fluorescence PCR. Reactions were performed in a 96-well plate using the SYBR Premix Ex Taq. The qPCR reaction was performed using the LightCycler 480 II Real-Time PCR instrument (Roche Applied Science). The RNA extraction kit was purchased from Tiangen DP430, and the reverse transcription and fluorescence quantitative PCR kit [HiScript III RT SuperMix for qPCR (+gDNA wiper)] was purchased from Vazyme Biotech Co., Ltd.

Protein analysis by SDS-PAGE

One milliliter of culture sample was collected at 24 h of fermentation and centrifuged (12 000 × g for 10 min at 4°C). The supernatant was removed to obtain cell precipitation and suspended in ddH₂O. These steps were repeated twice; then an ultrasonic disruptor was used to break cells and obtain cell lysate. The cell lysate was treated with 4 × SDS-PAGE sample buffer and then boiled for 10 min to get samples. Prepared samples and markers were separated via SDS-PAGE using a 10% Bolt Bis-TrisPlus gel in MES SDS running buffer (Thermo Scientific, Waltham, USA). After that, Coomassie Brilliant Blue R250 staining was used to reveal proteins of different molecular weights.

Measurement of lycopene and riboflavin

Lycopene and β-carotene were extracted from the cell pellet using acetone as previously described (36). The sample was analyzed using an HPLC system (Agilent Technologies 1260) equipped with a C18 column (4.6 mm × 250 mm; 5 μm; Waters, Ireland) and a variable wavelength scanning UV detector (VWD) set to 474 nm. The mobile phase consisted of 50% (v/v) acetonitrile, 30% (v/v) methanol and 20% (v/v) isopropanol at a flow rate of 1 ml·min⁻¹ and a temperature of 40°C.

For riboflavin measurements, 200 μl of the fermentation culture was diluted with 800 μl of 0.05 M NaOH and centrifuged at 10 000 × g for 2 min to remove cells. The supernatant was then diluted with a 0.1 M acetate-sodium acetate buffer solution (pH 4.42) to bring it within the linear range of the spectrophotometer and the absorbance at 444 nm (OD₄₄₄) was measured (37). The riboflavin concentration was calculated using a validated standard equation: Y = (OD₄₄₄ - 0.0203) × DF/0.0163 (R² = 0.9997; Y, the riboflavin concentration of sample (mg·l⁻¹); OD₄₄₄, the value of absorbance at 444 nm; DF, dilution fold; OD₄₄₄ was controlled within the range of 0.1–0.8 by dilution).

Statistical analysis

The data were presented as mean ± standard deviation, and each experiment was independently repeated at least three times unless stated otherwise. Two-tailed Student’s t-test was used for two-group comparisons, and a one-way analysis of variance followed by post-hoc Dunnett’s test was used for multiple-group comparisons. P-values were reported as n.s. (not significant) when P> 0.05, * for P < 0.05, ** for P < 0.01, *** for P < 0.001 and **** for P < 0.0001 to indicate statistical significance.

Results and discussion

Design and construction of cross-species inducible systems in bacteria

The cross-species inducible system enabled comparative analysis across strains and supported the development of universally applicable systems with complex functions (28). As illustrated in Figure 1A, the method for construction of a cross-species inducible system in bacteria involves four key steps: (i) characterization of reported inducible systems in different strains to assess their expression levels; (2) reconstruction and modification through rational design and directed evolution to fine-tune the balance between P_inducer and P_repressor expression; (3) further optimization of the system by placing it on the genome and plasmid to determine optimal expression conditions, and (4) application of the inducible system for protein expression and chemical production. To validate the effectiveness of the approach, representative Gram-negative and Gram-positive model bacteria, including E. coli, B. subtilis and C. glutamicum, were selected to develop a universal inducible system for these three strains.

Figure 1.

The framework for constructing a cross-species inducible system in bacteria. (A) Schematic diagram of the method for constructing a cross-species inducible system in bacteria: characterization, reconstruction and modification, optimization and application. (B) The characteristics of the nine reported inducible systems in E. coli, B. subtilis and C. glutamicum were evaluated. The relative fluorescence expression intensity after 24 h, with and without the addition of the corresponding inducer, was used as the indicator for the inducible systems. The data represent the mean ± SD (standard deviation) of three experiments each in triplicate. (C) The 14 modified promoters (P_tet1-2, P_phlF1-3, P_cym1, P_van1-3, P_betl1-2, P_ttg1, P_cau1-2), along with the original seven promoters, were used to express in B. subtilis and C. glutamicum. The data represent the mean ± SD (standard deviation) of three experiments each in triplicate.

Nine reported inducible systems were selected, including the IPTG-inducible system P_grac and the xylose-inducible system P_xylA from B. subtilis, as well as seven optimized systems from E. coli that were induced by different inducers (P_tet, P_phlF, P_cym, P_van, P_betl, P_ttg, P_cau). Then, their relative fluorescence intensity at 24 h was used as the expression intensity, which was assessed with and without the inducer in E. coli, B. subtilis and C. glutamicum (Figure 1B). Unfortunately, the inducible system derived from B. subtilis (P_grac) can be used in C. glutamicum but failed to express and induce in E. coli (Figure 1B). Meanwhile, P_xylA exhibited a significantly lower expression level in C. glutamicum compared to B. subtilis. Notably, P_xylA achieved high expression levels in E. coli under both induced and uninduced conditions (Figure 1B). This discrepancy may arise from a mismatch between the expression strength of P_inducer and P_repressor.

In addition, the other seven inducible systems from E. coli showed no expression in B. subtilis and C. glutamicum (Figure 1B), indicating that the P_inducer of these systems was unable to function. The performance of the nine inducible expression systems in these three strains indicated that constructing a cross-species inducible system required not only that P_inducer and P_repressor functioned effectively across multiple strains but also that their expression compatibility was achieved to facilitate inducible expression. Directly transferring existing strain-specific inducible systems was not feasible. Therefore, these inducible systems were reconstructed to get the cross-species inducible systems. First, we want to modify the inducible systems from B. subtilis (P_grac and P_xylA) to function in E. coli. The repressor of P_grac and P_xylA was deleted to assess their maximum expression capacity, and several strategies were employed, including introducing an additional copy of lac/xyl operator (lacO/xylO), adding the repressor with a constitutive promoter, and substituting the original P_inducer with constitutive promoters P₁₃ and P₂₁ along with lacO/xylO. However, these reconstructed inducible systems still failed to achieve induction in E. coli (Supplementary Figures S1 and S2).

Seven inducible systems from E. coli were then modified through two approaches to construct inducible systems in B. subtilis and C. glutamicum. One approach was to modify the P_inducer of these systems (P_tet, P_phlF, P_cym, P_van, P_betl, P_ttg, P_cau) by combining the corresponding operators with the constitutive promoters P_SPL-L6 in B. subtilis and C. glutamicum (Supplementary Figure S3A). The other approach involved replacing the non-conserved sequences of the original P_inducer with conserved sequences that are recognizable by B. subtilis and C. glutamicum (Supplementary Figure S3B). As a result, a total of 14 modified promoters (P_tet1-2-representing both P_tet1 and P_tet2, P_phlF1-3, P_cym1, P_van1-3, P_betl1-2, P_ttg1, P_cau1-2), along with the original seven promoters (Supplementary Table S1), were used to express in B. subtilis and C. glutamicum. The results showed that eight promoters (P_phlF3, P_cym1, P_van2, P_van3, P_betl2, P_ttg1, P_cau2, P_tet2) were successfully achieving expression in B. subtilis and C. glutamicum (Figure 1C).

To obtain functional repressor expression cassettes in three hosts, three constitutive promoters (P_SPL-L2, P_spovG and P_xylR) from E. coli and B. subtilis with different expression intensities were used to drive the expression of repressors (Supplementary Figure S4). As a result, 24 different inducible systems were assembled, and they were named P_inducerR1-3. For instance, promoter P_phlF3 with P_SPL-L2-phlF was named P_phlF3R1, promoter P_phlF3 with P_spovG was named P_phlF3R2, and promoter P_phlF3 with P_xylR was named P_phlF3R3. The determination of repressor expression strength for these 24 combinations in B. subtilis indicated that six inducible systems (P_phlF3R1, P_van2R3, P_tet2R1, P_tet2R2, P_cau2R3 and P_betl2R1) displayed significant variations, with expression levels at 0.069, 0.460, 0.719, 0.286, 0.786 and 0.578 times that of the corresponding promoter without repressor under non-inducing conditions (Figure 2A). The results indicated that repressor expression did not consistently correlate with promoter strength, highlighting the importance of compatibility between the repressor and P_repressor. Six inducible systems (P_phlF3R1, P_van2R3, P_tet2R1, P_tet2R2, P_cau2R3 and P_betl2R1), which exhibited relatively more suitable repressor protein expression levels compared to the other combinations, were characterized for their inducive performance and leakage expression in E. coli, B. subtilis and C. glutamicum.

Figure 2.

Reconstruction and modification of inducible systems in E. coli, B. subtilis and C. glutamicum. (A) Characteristics of the eight modified promoters combined with repressors expressed by different promoters (P_SPL-L2, P_spovG and P_xylR) in B. subtilis. The data represent the mean ± SD (standard deviation) of three experiments each in triplicate. P-values were reported as n.s. (not significant)when P>0.05, * for P<0.05, ** for P<0.01, *** for P<0.001, and **** for P<0.0001 to indicate statistical significance. (B) Characteristics of the modified inducible systems P_phlF3R1, P_tet2R1, P_tet2R2, P_van2R3, P_cau2R3 and P_betl2R1 in E. coli, B. subtilis and C. glutamicum. The IR is defined as the ratio of expression levels with inducer to those without. The data represent the mean ± SD (standard deviation) of three experiments each in triplicate.

The induction ratio (IR), defined as the ratio of expression levels with the inducer to those without, was employed as a preliminary measure of system performance (38). To better evaluate the changes in the inducible system after adding the inducer, to better evaluate the changes in the inducible system after adding the inducer, we defined no response as an IR of 1.15 or less, and an IR of 2 or greater as a significant difference. Among them, the P_phlF3R1, P_tet2R1 and P_tet2R2 exhibited significant differences in expression levels with and without the inducer in E. coli, B. subtilis and C. glutamicum (Figure 2B), indicating that they can be served as initially available inducible systems across all three strains. In contrast, P_van2R3 showed inducible expression in E. coli and B. subtilis but exhibited almost no response to the inducer in C. glutamicum. Similarly, P_cau2R3 induced expression only in E. coli and showed no response to the inducer in B. subtilis and C. glutamicum. Additionally, P_betl2R1 exhibited a peculiar phenomenon of inhibition upon inducer addition in all three strains (Figure 2B). In summary, through rational modification of P_inducer and the repressor expression cassette, the DAPG-inducible system P_phlF3R1, the aTc-inducible system P_tet2R1 and P_tet2R2 capable of functioning in E. coli, B. subtilis and C. glutamicum were obtained. However, since P_tet2R1 and P_tet2R2 still exhibit leak expression in E. coli, B. subtilis and C. glutamicum, the next step will be to optimize these systems for enhanced performance.

Optimization of the cross-species inducible systems

To enhance the performance of P_tet2R1 and P_tet2R2 in E. coli, B. subtilis and C. glutamicum, the components of both inducible systems were optimized, and a mutated repressor expression library was constructed (Figure 3A). Compared to P_tet2R1, the P_repressor of P_tet2R2 (P_SPL-L2) was replaced with promoter P_spovG, increasing the IR from 33.6 to 45.8 in E. coli, from 2.38 to 4.38 in B. subtilis and from 1.89 to 6.69 in C. glutamicum (Figure 3B and C). Then another tet operator (tetO) was introduced into P_tet2R getting inducible system P_tet2R2O to diminish leakage expression and enhance the IR. Compared to P_tet2R2, the leakage expression of P_tet2R2O decreased by 13.8%, 94.4% and 18.9% in E. coli, B. subtilis and C. glutamicum, respectively. However, the maximum expression levels were also reduced by 23.7%, 78.1% and 71.7% in three strains, respectively (Figure 3B and C). This phenomenon may be due to the increased binding strength between the repressor and the promoter upon adding the tetO (Figure 3B). In conclusion, modifying the promoter alone cannot perfectly achieve low leakage and high expression levels.

Figure 3.

Characteristics and modification of the aTc-inducible system in E. coli, B. subtilis and C. glutamicum. (A) Schematic diagram of the optimization of the aTc-inducible system using error-prone PCR to further reduce leakage expression. (B) Diagram of the structural components of seven different aTc-inducible systems, highlighting key regulatory elements and their configurations. The symbol ‘*’ in 3A refers to the name of the tetR mutant, tetR* and the name of inducible system with tetR*. (C) Comparison of the inducible performance of the seven aTc-inducible systems in E. coli, B. subtilis and C. glutamicum, showing fluorescence intensity without and with induction, as well as the IR, defined as the ratio of expression levels with inducer to those without. The data represent the mean ± SD (standard deviation) of three experiments each in triplicate. (D) Confocal images of E. coli, B. subtilis and C. glutamicum strains from the green channel (515–564 nm) under aTc-induced conditions, each carrying one of seven different aTc-inducible systems. Image analysis was performed with the Leica TCS SP8 software package and ImageJ. Scale bar: 10 μm. Each experiment was independently repeated three times with similar results.

The mutated repressor expression library was constructed to further reduce leakage expression (Figure 3A). Error-prone PCR was used to introduce random mutations into the repressor, while a degenerate primer generated an RBS library to increase diversity. The library was then ligated into pHTcgl-P_tet2R2-sfGFP and transformed into E. coli. Following transformation, colonies exhibiting low leaky expression without aTc and strong expression with aTc were selected and then transformed into B. subtilis and C. glutamicum for further evaluation. Ultimately, mutants demonstrating low leakage and high expression across all strains were sequenced. The process resulted in one well-performing mutant, designated P_tet2R2*. In E. coli, B. subtilis and C. glutamicum, the leakage of P_tet2R2* was reduced by 52.9%, 97.7% and 80.8%, respectively, with an increase in IR from 45.8 to 90.0 in E. coli, from 4.83 to 174 in B. subtilis and from 6.69 to 36.5 in C. glutamicum (Figure 3B). However, the further addition of tetO into P_tet2R2*(P_tet2R2*O) showed similar results as P_tet2R2, significantly reducing the maximum expression intensity (Figure 3B). As a result, P_tet2R2* was considered the optimal cross-species inducible system. Sequencing results showed that the RBS sequence of P_tet2R2* was RBS0, and the repressor tetR exhibited a mutation from Phe to Tyr at position 119 (tetR*). RBS0 refers to the RBS obtained after introducing mutations, representing the specific RBS variant used in the system optimization. However, the exact reason for the excellent performance of P_tet2R2* was still unclear, whether it stems from the enhanced translation strength of RBS0 or the increased affinity of tetR*.

To assess each factor’s impact, the repressor tetR* and RBS0 of P_tet2R2* were reverted to the original tetR and RBS, creating P_tet2R2*1 and P_tet2R2*2, respectively. Compared to P_tet2R2, RBS0 reduced leakage by 24.0% in E. coli, 97.0% in B. subtilis and 74.7% in C. glutamicum, while increasing IR by 0.29-, 28.19- and 2.94-fold, respectively, in these strains. The F119Y mutation of tetR* reduced leakage by 14.4% in E. coli, 22.2% in B. subtilis and 26.2% in C. glutamicum, while increasing IR by 0.24-, 0.33- and 0.43-fold, respectively, in these strains (Figure 3B). These results suggested that RBS0 was more effective in reducing leakage, especially in B. subtilis and C. glutamicum. Meanwhile, predictions of translation strength for RBS0 revealed that it increased by 45-, 7.8- and 62-fold in E. coli, B. subtilis and C. glutamicum, respectively, compared to the original RBS (Supplementary Figure S5). These results suggested that RBS0 was more effective in reducing leakage and expression intensity by increasing the expression level of the repressor protein, particularly in B. subtilis and C. glutamicum. Additionally, tetR* may contribute to leakage reduction by enhancing the affinity of the repressor protein for tetO to a certain extent. Thus, the combined effect of the two mutations is superior to that of a mutation alone, facilitating the construction of an aTc-inducible system with low leakage and high expression intensity in the three strains. To better compare the performance of different aTc-inducible systems across the three bacterial strains, confocal images were used to visualize their differences, as shown in Figure 3D and Supplementary Figure S6. The confocal images provided a clearer representation of the distinctions among the inducible systems under inducer-added conditions, which were consistent with the fluorescence intensity characterization results.

Characterization of the cross-species inducible systems

Since DAPG and aTc impact the growth of the three strains, the addition concentration needs to be optimized. As shown in Supplementary Figure S7A, three strains exhibited marked differences in resistance to DAPG: E. coli tolerated up to 1000 μM, B. subtilis up to 100 μM and C. glutamicum up to 10 μM, with the least impact on growth observed when DAPG was added at 6 h. In contrast, all three strains exhibited similar resistance to aTc, with the addition of 10 μM at 6 h not affecting growth (Supplementary Figure S7B). Therefore, a gradient of DAPG (0–1000 μM) and aTc (0–10 μM) was added at 6 h.

Based on the above conditions, the DAPG-inducible system P_phlF3R1 and aTc-inducible system P_tet2R2* were placed on the plasmid and genome of E. coli, B. subtilis and C. glutamicum to test, respectively (Figure 4A and Supplementary Figure S8A). The commonly used plasmid pHT was employed as the expression vector for E. coli (high copy number, 300–500 copies per cell) and B. subtilis (low copy number, 10–15 copies per cell), while the plasmid pJYW was used as the expression vector for C. glutamicum (high copy number, 300–500 copies per cell). For further precise evaluation, some key features including leakage expression (l), dynamic range (μ), maximum expression intensity (m) and sensitivity (K_a) were measured (Figure 4B). Previous reports on inducible systems indicate that their leakage expression typically ranges from 1000 to 2000 a.u.; however, there are few accurate descriptions regarding other parameters (4). For the DAPG-inducible system P_phlF3R1, Figure 4C and Supplementary Figure S8A present the induction curves and four key parameters obtained from plasmid-based expression in three different strains, respectively. The leakage expression of P_phlF3R1 in E. coli, B. subtilis and C. glutamicum in plasmid was measured at 264.8, 4945.7 and 1950.2 a.u., respectively, with maximum expression levels of 85 169.7, 28 717.5 and 4474.9 a.u. The dynamic ranges were 320.63, 4.80 and 1.29, while the sensitivities were 257.6, 25.43 and 0.79 μM, respectively (Figure 4C). These parameters suggested that P_phlF3R1 exhibited optimal performance in E. coli, characterized by low leakage and high maximum expression intensity, which led to an extensive dynamic range. In contrast, C. glutamicum showed the lowest protein expression levels (Figure 4C). The maximum expression intensity of the inducible system P_phlF3R1 on the genome significantly decreased in all three strains, which was attributed to the reduced plasmid copy number (Supplementary Figure S8A). In C. glutamicum, leakage expression also decreased accordingly. However, in E. coli and B. subtilis, leakage expression increased, likely due to differences in the expression levels of the repressor protein between the genome and the plasmid. The induction curves suggested that higher DAPG concentrations could boost expression levels, though this effect is constrained by DAPG’s antibiotic properties. For P_phlF3R1, despite its high leakage, which may make the system unsuitable for expressing toxic proteins or genes requiring tight regulation, the exogenous synthesis of DAPG presents the potential for developing a quorum-sensing system that enables self-induced expression.

Figure 4.

The induction curves of DAPG- and aTc-inducible systems in E. coli, B. subtilis and C. glutamicum. (A) Schematic diagram of the characterization of DAPG- and aTc-inducible systems on the plasmid. (B) Key features for an effectively inducible system. The key parameters of the induction curves are presented in dashed lines. The numbers in the box, from left to right and top to bottom, represent leakage expression (l), maximum expression intensity (m), dynamic range (μ), and sensitivity (K_a), respectively. (C) Induction curves and key parameters of the DAPG-inducible system P_phlF3R1 on the plasmid in E. coli, B. subtilis and C. glutamicum. The data represent the mean ± SD (standard deviation) of three experiments each in triplicate. (D) Induction curves and key parameters of the aTc-inducible system P_tet2R2* on the plasmid in E. coli, B. subtilis and C. glutamicum. The data represent the mean ± SD (standard deviation) of three experiments each in triplicate.

For the aTc-inducible system P_tet2R2*, leakage expression levels on the plasmid in E. coli, B. subtilis and C. glutamicum were measured at 539.3, 113.2 and 273.5 a.u., respectively, corresponding to 2.036, 0.022 and 0.140 times that of the DAPG-inducible system P_phlF3R1 (Figure 4D). The maximum expression intensity reached 70 336.6, 24 719.7 and 8404.5 a.u., representing 0.825, 0.861 and 1.878 times that of the DAPG-inducible system P_phlF3R1 (Figure 4D). The dynamic ranges were 129.28, 217.37 and 29.73, while the sensitivities were 0.97, 1.43 and 0.16 μM, respectively (Figure 4D). The maximum expression intensity and leakage expression intensity of the inducible system P_tet2R2* on the genome significantly decreased in all three strains, which was attributed to the reduced plasmid copy number (Supplementary Figure S8B). Compared to P_phlF3R1, P_tet2R2* demonstrated significantly lower leakage expression in plasmid across the three strains, resulting in a high dynamic range and characteristics approaching those of an ideal inducible expression system.

To further investigate the effect of the inducible systems P_tet2R2* and P_phlF3R1 on GFP transcription levels with and without inducer, we conducted qPCR analysis (Supplementary Figure S9). The results showed that, in all three strains, transcription levels were very low without inducer, indicating strong repression of GFP transcription by the repressor proteins. Transcription levels were significantly higher under the condition of with inducer, which was consistent with the results obtained from fluorescence intensity measurements. We also compared these two inducible systems with commonly used inducible expression systems in three different strains (Supplementary Table S3). In E. coli, both P_tet2R2* and P_phlF3R1 exhibited lower leakage and maximum expression intensity compared to the T7 system (BL21(DE3)-T7lac), likely due to the excessive strength of the T7 expression system. In B. subtilis, both P_tet2R2* and P_phlF3R1 displayed higher maximum expression intensities compared to the commonly used IPTG and xylose-inducible systems, with P_tet2R2* showing low leakage expression. In C. glutamicum, P_tet2R2* and P_phlF3R1 displayed lower leakage compared to the P_trc. Overall, P_tet2R2*showed superior performance across all three strains, while P_phlF3R1 showed higher leakage expression in B. subtilis and C. glutamicum.

In conclusion, two systems capable of inducible expression in E. coli, B. subtilis and C. glutamicum were obtained through rational and non-rational approaches. In particular, the resulting aTc-inducible system P_tet2R2* demonstrated low leakage, high maximum expression levels, wide dynamic range and high sensitivity, which had potential applications in protein expression and biosynthetic gene regulation.

Regulation of proteins and metabolites production via aTc-inducible system P_tet2R2*

To verify the versatility of the modified aTc-inducible system, P_tet2R2* was used for multiple proteins and gene clusters. First, three reporter proteins (sfGFP, mCherry and mScarlet3) were expressed in E. coli, B. subtilis and C. glutamicum, respectively. The strains expressing three distinct reporter proteins under the control of the P_tet2R2* were cultured for 24 h under both aTc-free and aTc-induced conditions. In the absence of aTc, none of the proteins were expressed in any of the strains. Conversely, in the presence of aTc, all three proteins exhibited high expression levels in three strains, with fluorescence visible (Figure 5A). SDS-PAGE analysis of the lysed cells confirmed these results and demonstrated that the protein expression levels among the three strains are highest in E. coli, followed by B. subtilis, and then C. glutamicum (Figure 5B). These results indicated that, despite the use of different expressed proteins, P_tet2R2*consistently exhibited low leakage expression and high maximum expression strength across the three strains. These findings align with previous characterizations, further demonstrating the stability of this inducible expression system and its potential applications in metabolic engineering.

Figure 5.

Application of the aTc-inducible system P_tet2R2* for E. coli, B. subtilis and C. glutamicum in protein expression and biosynthetic gene cluster regulation. (A) Comparison of bacterial colors with and without aTc using P_tet2R2* to express three fluorescent proteins (sfGFP, mCherry and mScarlet3) in E. coli, B. subtilis and C. glutamicum. (B) The results of SDS-PAGE for three fluorescent proteins (sfGFP, mCherry and mScarlet3) expressed using P_tet2R2* with and without aTc. (C) The synthetic pathway and shake flask fermentation of lycopene and β-carotene in E. coli, B. subtilis and C. glutamicum. The figure shows major intermediate products with solid lines for single-step conversions and dashed lines for multi-step reactions, which are omitted for clarity. The synthetic gene cluster crtEBI and crtEBIY, responsible for lycopene and β-carotene biosynthesis, was regulated by P_tet2R2* within the plasmids. The titers of lycopene and β-carotene were measured both in the presence and absence of aTc. (D) The synthetic pathway and shake flask fermentation of violacein in E. coli, B. subtilis and C. glutamicum. The synthetic gene cluster vioA/B/C/D/E, responsible for violacein biosynthesis, was regulated by P_tet2R2* within the plasmids.

The capability of P_tet2R2* to drive the expression of a complex biosynthetic pathway in E. coli, B. subtilis and C. glutamicum was then examined. Three gene clusters for the biosynthesis of lycopene, β-carotene and violacein were driven by P_tet2R2*. Lycopene and β-carotene are synthesized from the precursors isopentenyl pyrophosphate (IPP) and dimethylallyl pyrophosphate (DMAPP), which is produced via the 2-C-methyl-D-erythritol-4-phosphate (MEP) pathway. The synthesis of lycopene and β-carotene requires the exogenous gene clusters crtEBI and crtEBIY to be introduced into E. coli, B. subtilis and C. glutamicum (Figure 5C). The strains expressing gene clusters crtEBI and crtEBIY under the control of the P_tet2R2* were cultured for 24 h under both aTc-free and aTc-induced conditions. As shown in Figure 5C, the lycopene and β-carotene successfully induced expression in all three strains. In the absence of aTc, only trace amounts of lycopene and β-carotene were detected. Upon aTc induction, lycopene titers in E. coli, B. subtilis and C. glutamicum reached 7.8, 98.2 and 4.2 mg·L⁻¹, respectively, while β-carotene titers were 4.5, 18.5 and 3.5 mg·L⁻¹ (Figure 5C). B. subtilis demonstrated the highest production of both lycopene and carotene, with increases of 982- and 185-fold, respectively, compared to non-induced conditions (Figure 5C). These titers represent the highest reported for B. subtilis to date. Additionally, C. glutamicum was reported for the first time as a host for β-carotene production. Interestingly, E. coli, typically employed for terpene biosynthesis (39), exhibited much lower lycopene and carotene titers compared to B. subtilis (Figure 5C), indicating that further optimization of production conditions may be necessary.

In addition, the biosynthetic pathway for violacein uses L-tryptophan as the precursor, producing a navy color through the catalysis of five heterologous enzymes (VioA/B/C/D/E) from Chromobacterium violaceum. The strains expressing gene cluster vioABCDE under the control of the P_tet2R2* were cultured for 24 h under both aTc-free and aTc-induced conditions. The results showed leaky expression in E. coli, with only a faint color change observed upon induction (Figure 5D). In contrast, minimal leaky expression was detected in B. subtilis and C. glutamicum, both of which exhibited a pronounced color change in response to aTc induction (Figure 5D). Overall, P_tet2R2* successfully induced the expression of three reporter proteins and three gene clusters in E. coli, B. subtilis and C. glutamicum, demonstrating low leakage and high expression intensity in all strains. The three gene clusters, with lengths of 3.4, 4.6 and 7.3 kb, were all expressed effectively, highlighting the P_tet2R2*’s significant advantage for producing complex metabolic pathways. In conclusion, this system simplifies heterologous expression by reducing the need to construct extensive strain-specific expression cassettes.

To enhance the functionality of the inducible expression system, an aTc-T7 system was constructed and evaluated in E. coli and B. subtilis. The P_tet2R2*-T7RNAP cassette was integrated into E. coli K12 and B. subtilis G600, generating the K12T7T and BST7T strains, respectively (Supplementary Figure S10A). Expression of the gene of interest (GOI) was regulated by the hybrid promoter P_T7tet, which combines T7 and tetO elements. To enable the function of the promoter P_T7tet in strain K12T7T, we tested three commonly used plasmid vectors—pCYC, pBBR and pRSF (Supplementary Figure S10B). Among these, only pBBR exhibited superior performance and was selected for subsequent experiments. For strain BST7T, we had previously screened expression vectors in our study (4). In that study, the pHT vector demonstrated the best performance and was therefore chosen for further experiments. Accordingly, pBBR-P_T7tet-sfGFP and pHT-P_T7tet-sfGFP were constructed for characterization in strains K12T7T and BST7T, respectively. While the aTc-T7 system demonstrated robust performance in B. subtilis, it did not significantly mitigate expression leakage or enhance maximum expression levels in E. coli (Supplementary Figure S10C). Therefore, future studies should focus on optimizing P_tet2R2*, potentially by fine-tuning RBS strength, to achieve improved expression dynamics.

Construction of a single-input genetic circuit based on the aTc-inducible system P_tet2R2*

Although the constructed inducible systems effectively control gene expression due to their low leakage and high expression levels, their functionality is limited in activating gene expression. Meanwhile, repression of competing and branching pathways is essential to prevent undesired metabolic flux. To achieve both gene activation and repression, a single-input genetic circuit was developed by combining CRISPR interference (CRISPRi) with the T7 expression system, consisting of an activation module and a repression module (Figure 6A). Notably, the activation module featured a T7 RNAP expression cassette integrated into the genome, regulated by the aTc-inducible system P_tet2R2*. This module drove the T7 promoter to express the GOI either in the genome or plasmid. The repression module comprised a dCas12a expression cassette, regulated by P_tet2R2* and an expression cassette for the crRNA array (Figure 6A). Due to the low leakage of P_tet2R2*, the circuit remained silent in the absence of aTc, ensuring no impact on cellular growth and metabolism. Following the addition of aTc, the circuit quickly expressed the GOI and repressed gene of uninterest (GOU), thereby maximizing the output of the desired product.

Figure 6.

Construction, characteristics, and application of single-input genetic circuit in E. coli and B. subtilis. (A) Schematic diagram of the single-input genetic circuit, consisting of an activation module and a repression module. (B) Characteristics of the single-input genetic circuit in E. coli, where the crRNA targeting the gene of mScarlet3 is controlled by the inducible promoter P_tet2 and the constitutive promoter P_veg. The expression levels of sfGFP and mScarlet3 were evaluated in the presence and absence of aTc. (C) Characteristics of the single-input genetic circuit in B. subtilis, where the crRNA targeting the gene of mScarlet3 is controlled by the inducible promoter P_tet2 and the constitutive promoter P_veg The expression levels of sfGFP and mScarlet3 were evaluated in the presence and absence of aTc. (D) Shake flask fermentation of riboflavin in various engineered strains, which included: replacing the promoters of the ribDEAHT gene cluster, the purEKBCSQLFMNHD gene cluster, and the gdh gene with P_T7tet, and using repression module to repress ribC. (E) Shake flask fermentation of lycopene in various engineered strains, which included: replacing the promoters of the ispDF, fni, dxS and dxR gene with P_T7tet, and using repression module to repress hepS.

The single-input genetic circuit was constructed by integrating the P_tet2R2*-dCas12a cassette into E. coli K12T7T and B. subtilis BST7T to create the K12T7Tt and BST7Tt strains, respectively. To verify the circuit’s functionality, the GOI and GOU were set to genes of sfGFP and mScarlet3. The gene of mScarlet3 was expressed under the control of the constitute promoter P_veg and integrated into the genome, resulting in the strain K12T7Tt-msc and BST7Tt-msc. The crRNA (mc1/mc2/mc3) targeting the gene of mScarlet3 was driven by aTc-inducible promoter P_tet2. Both crRNA and the P_T7tet-sfGFP expression cassettes were placed on the plasmid. To better highlight the changes, the data were normalized by defining the fluorescence expression intensities of sfGFP after activation and mScarlet3 when not repressed as 100%, with no fluorescence defined as 0. As shown in Figure 6B, the expression of mScarlet3 in E. coli decreased by 81.3%, 78.1% and 69.2% without aTc. With the addition of aTc, the expression of sfGFP was activated, while mScarlet3 expression was repressed by 87.6%, 87.4% and 91.8%. The expression of mScarlet3 was repressed in the absence of aTc, likely due to leakage from the P_tet2R2* in E. coli. To investigate this possibility, the promoter of crRNA was replaced with the constitutive promoter P_veg. The expression of mScarlet3 was almost completely repressed (91–97%) regardless of aTc presence (Figure 6B).

In B. subtilis, the expression of mScarlet3 decreased by 44.4%, 14.1% and 15.6% without aTc, likely due to the more stringent behavior of the P_tet2R2* in this organism (Figure 6C). Upon aTc addition, sfGFP expression was induced, and mScarlet3 was repressed by 96.0%, 89.1% and 98.5%. Similar to E. coli, replacing the promoter of crRNA with P_veg resulted in decreasing 92–98% repression of mScarlet3 expression in both the presence and absence of aTc (Figure 6C). These observations demonstrated that the single-input genetic circuit successfully activated expression of sfGFP and repressed expression of mScarlet3 in both E. coli and B. subtilis in the presence of aTc. In contrast, when aTc was absent, the use of P_tet2R2* and P_veg to drive crRNA array expression resulted in partial and almost complete repression, with the degree of repression differing between E. coli and B. subtilis. This distinction was attributed to leakage expression of dCas12a and crRNA. The higher leakage of P_tet2R2* in E. coli compared to B. subtilis aligned with the greater level of repression observed in E. coli in the absence of aTc. These results suggested that the expression level of crRNA was the limiting factor in CRISPRi system. When both crRNA and dCas12a were initially expressed at low levels, enhancing expression of crRNA significantly improved the repression effect. This indicated that dCas12a could still function effectively in CRISPRi system even at low expression levels. To verify this hypothesis, we constructed a dual-input genetic circuit (Supplementary Figure S11A), using the low-leakage IPTG-inducible system P_grac1R to express dCas12a, and P_veg and P_tet2R2* to express crRNA. When crRNA was expressed by P_veg (high expression level), CRISPRi system achieved nearly 100% repression, both with and without IPTG induction (dCas12a: high and low expression levels) (Supplementary Figure S11B). However, when crRNA was expressed by P_tet2R2*, repression was minimal without inducers. Upon adding aTc, which increased expression of crRNA, repression reached nearly 100%. Adding IPTG, which increased dCas12a expression, resulted in only 60% repression. When both aTc and IPTG were added, complete repression was achieved. These results confirmed our hypothesis that the expression level of crRNA was the limiting factor for the functionality of the CRISPRi system. Overall, these results underscored the importance of an inducible system with minimal leakage in constructing an efficient repression module.

Due to the suboptimal performance of gene circuits in E. coli compared to the ideal performance of single-input genetic circuits in B. subtilis, further exploration was made to enhance chemical production expression. Riboflavin (vitamin B2) is an essential micronutrient as it is a direct precursor of the coenzymes flavin adenine (FAD) and flavin mononucleotide (FMN), which are crucial for cellular processes (40). In B. subtilis, the synthesis of riboflavin is subjected to negative feedback regulation with GTP and FMN by the guanine and FMN riboswitches, respectively (Supplementary Figure S12A). In addition, riboflavin kinase, encoded by the essential gene ribC, can convert riboflavin to FMN, which not only consumes riboflavin but also further strengthens the feedback inhibition of the FMN riboswitch. Moreover, the synthesis of riboflavin from GTP and ribulose-5-phosphate depends on the availability of ribulose-5-phosphate, which can be adjusted by overexpressing glucose dehydrogenase (gdh) to modify carbon flow in the gluconate bypass.

To enhance riboflavin production, the ribDEAHT gene cluster, purEKBCSQLFMNHD gene cluster and gdh were selected for overexpression using the P_T7tet, while ribC was chosen as a target gene for repression module. The results of the activation and repression combinations of different genes showed that (Figure 6E), relative to the wild-type strain, the yield of riboflavin significantly increased when the feedback inhibition of FMN was relieved. However, further alleviation of GTP feedback inhibition (5.0-fold that of the wild-type strain) or overexpression of gdh (8.1-fold that of the wild-type strain) did not lead to a noticeable enhancement in riboflavin production. The most significant increases in yield were observed with the relief of FMN feedback inhibition and the repression of ribC (50.9-fold that of the wild-type strain), while additional modifications actually resulted in a decrease in yield. These results indicated that relieving FMN feedback inhibition and repressing the degradation pathway of riboflavin had the greatest impact on riboflavin production. These findings also indicated that this single-input genetic circuit can serve as a foundational platform, enabling the maximization of target product expression through the screening and combination of genes for activation and repression.

Next, the single-genetic circuit was employed to enhance lycopene production. The promoter P_T7tet was used to activate the expression of key genes, increasing precursors IPP and DMAPP, while CRISPRi was applied to repress the competing metabolic flux from FPP to vitamin K2. Target genes reported to enhance lycopene production were selected, including the genes of 1-deoxy-D-xylulose-5-phosphate synthase (DXS), 1-deoxy-D-xylulose-5-phosphate reductoisomerase (DXR), C-methyl-D-erythritol-4-phosphocytidine transferase (IspD), 2-C-methyl-D-erythritol-2,4-cyclic pyrophosphate synthase (IspF) and Type 2 isopentenyl diphosphate isomerase (Fni) for overexpression, as well as hepS for repression module (Supplementary Figure S12B). Starting with the BST7Tt strain, the native promoters of four selected genes (ispDF, dxS, dxR and fni) were replaced with P_T7tet. After iterative replacements, the plasmid pHT-P_T7tet-crtEBI, carrying the exogenous gene cluster crtEBI, was introduced, resulting in lycopene-producing strains L1–L10 and the control strain WT (BST7Tt with pHT-P_T7tet-crtEBI). Following fermentation, the strain with the highest lycopene production was further modified by introducing a crRNA expression cassette P_tet2-Cr_hepS to repress the expression of hepS getting strain L11. As shown in Figure 6E and Supplementary Figure S12C, in the strains L1–L10 equipped with only the activation module, activation of all four target genes resulted in the highest lycopene production, which was 3.29 times that of the control strain, reaching 212.37 mg·L⁻¹. With the addition of the repression module, lycopene production further increased to 232.1 mg·L⁻¹, which was 3.61 times that of the control strain. This represents the highest lycopene production reported in B. subtilis and provides a foundation for the industrial production of lycopene.

Conclusions

This study successfully developed and implemented a robust workflow for constructing cross-species inducible system. The DAPG- and aTc-inducible systems were constructed in E. coli, B. subtilis and C. glutamicum through the workflow, which were then meticulously characterized and used to express various reporter proteins and gene clusters. The results demonstrated that aTc-inducible system P_tet2R2* was an ideal system with low leakage, broad dynamic range, sufficient expression intensity, and appropriate sensitivity. In addition, the aTc-inducible system was used to construct T7 expression system and single-input genetic circuits for protein and chemical production. Together, our results suggest this workflow facilitates the development of cross-species inducible systems and provides highly efficient tools for applications in synthetic biology and metabolic engineering.

Data availability

The sequences of plasmid with the necessary annotations used in this study are deposited in Zenodo (10.5281/zenodo.8416166).

Supplementary data

Supplementary Data are available at NAR Online.

Funding

National Natural Science Foundation of China [32270096]; the Postgraduate Research & Practice Innovation Program of Jiangsu Province [1012050205238420]; Jiangsu Basic Research Center for Synthetic Biology [BK20233003]; Natural Science Foundation of Jiangsu Province [BK20240202]; Fundamental Research Funds for the Central Universities [JUSRP202404017, JUSRP622004]. Funding for open access charge: National Natural Science Foundation of China.

Conflict of interest statement. None declared.