Transposon insertion sequencing identifies novel genes involved in product synthesis in gas-fermenting Clostridium ljungdahlii Authors: Zhiyi Zhang, Fengjuan Yang, Yihan Liu, Wei Ye, Dechen Jia, Yi Tian, Weihong Jiang, Qiyao Wang, Yang Gu Published: 2026 ABSTRACT The efficient utilization of industrial one-carbon (C1) gases, such as carbon dioxide (CO2) and carbon monoxide (CO), has attracted widespread interest in recent years. Clostridium ljungdahlii, a key chemoautotrophic bacterium, can ferment syngas (a mixture of CO2, CO and H2) into value-added chemicals, making it a subject of extensive research. However, despite advances in genetic tool development for C. ljungdahlii, the rapid identification of genes underlying crucial cellular functions remains a bottleneck, hindering both a deeper understanding and successful genetic engineering of this bacterium. In this study, we constructed a genome-wide random mutagenesis library of C. ljungdahlii using a mariner transposon system, enabling the identification of numerous functional genes associated with cellular fitness to syngas. We overcame the library's relatively low plasmid curing efficiency by employing an IPTG-regulated conditional replicon. Subsequent high-throughput transposon insertion sequencing (Tn-seq) of this library allowed for the rapid exploration of a broad set of genes potentially involved in growth on syngas. Experimental validation of these candidate genes revealed several that affect the production of ethanol and acetate (two major metabolites) of C. ljungdahlii in gas fermentation. The data generated in this study provide a valuable resource for future research and optimization of C. ljungdahlii. Furthermore, the strategy presented here can be applied to identify genes associated with other important phenotypes in this bacterium. Introduction Gas-fermenting clostridia, belonging to acetogenic bacteria, are anaerobic microorganisms with unique abilities to assimilate C1 gases (CO2 and CO) [[1], [2], [3]]. Their Wood-Ljungdahl pathway enables efficient carbon fixation, converting CO2 and CO into acetyl-CoA [1,4], which can then be further synthesized into value-added chemicals such as ethanol, acetate, butyrate, and butanol [1,[5], [6], [7]]. The unique metabolic properties of gas-fermenting Clostridium species have not only garnered widespread attention in the field of biomanufacturing but also demonstrated potential applications in reducing industrial greenhouse gas emissions and decreasing reliance on fossil fuels [1,[8], [9], [10]]. The most significant advancement in gas fermentation using Clostridium species in recent years has been the successful large-scale production of ethanol through the fermentation of industrial carbon-rich off-gases (CO/CO2) [11]. This milestone underscores the great potential of microbial gas fermentation technology for future biomanufacturing. As anaerobic bacteria, gas-fermenting clostridia exhibit "autotrophic" physiological and metabolic traits that distinguish them from conventional industrial microbes [[12], [13], [14]]. However, progress in strain engineering and optimization has been constrained by fundamental gaps in our understanding of their gene functions and metabolic networks. To address these challenges, there is an urgent need for functional genomics studies that can systematically identify genes associated with important phenotypes. A major methodology in functional genomics is the construction of genome-wide random insertion mutant libraries based on transposons, combined with high-throughput screening. In the solventogenic and intestinal Clostridium species, several transposon systems have been developed and applied, including Tn916 [[15], [16], [17], [18], [19], [20]], Tn1545 [21,22], mariner [[23], [24], [25], [26]], Mu [27] and Tn5 [28]. In contrast, for gas-fermenting Clostridium species, which are attracting growing interest due to their ability to utilize CO2/CO as carbon sources, in-depth functional genomics studies remain limited, largely attributable to the technical challenges in their genetic manipulation. To date, the mariner transposon has been shown to function in gas-fermenting Clostridium species and has been used to achieve chromosomal integration of heterologous genes [29,30]. In a recent study, a genome-scale insertion mutant library was constructed in the gas-fermenting Clostridium autoethanogenum using a mariner-derived transposon system [31]. Screening of this library subsequently identified a set of putatively essential genes required for growth on CO as the sole carbon and energy source [31]. Clostridium ljungdahlii has gained significant attention in recent years as a gas-fermenting clostridia capable of converting CO or CO2 into valuable chemicals such as acetate, ethanol, and 2,3-butanediol [[32], [33], [34]]. Through genetic engineering, it has been further modified to produce non-native compounds including acetone, isopropanol, and 3-hydroxybutyrate [29,30,35,36]. A major ongoing research focus for this acetogen is the comprehensive and rapid identification of functional genes linked to its key phenotypic traits, including growth on CO2/CO, carbon fixation, energy metabolism, and product synthesis, which would provide critical insights for a deeper understanding and effective optimization of the strain. In this study, we applied Tn-seq to systematically identify genes associated with fitness and product synthesis of C. ljungdahlii under syngas-containing conditions. We developed an inducible mariner transposon system and constructed a genome-wide mutant library with 75% coverage. By using this library through condition-specific differential culturing followed by transposon sequencing, we identified a set of previously uncharacterized genes that play roles in product synthesis during gas fermentation. Their functional contributions and biological importance in C. ljungdahlii were further validated. Materials and methods Strains and media The bacterial strains employed in this study are listed in Table S1. Culture conditions were as following: (i) For plasmid construction, the E. coli strain TOP10 was cultured in LB medium or on LB agar plate at 37 °C supplemented with 12.5 μg/mL chloramphenicol. All the primers used in this study were synthesized by BioSune (Biosune, Shanghai, China). (ii) For library construction, the C. ljungdahlii strain DSM 13528 and its derivatives were cultured in YTF (yeast-tryptone-fructose) [37] or YT (YTF without fructose) medium and cultivated at 37 °C, supplemented with thiamphenicol (Tm, 5 μg/mL) and erythromycin (Em, 10 μg/mL) when needed. (iii) For cultivation on syngas, the engineered strains of C. ljungdahlii grew in fructose-free ATCC 1754 medium [38] supplemented with thiamphenicol (5 μg/mL) when needed. All the manipulations of C. ljungdahlii strains were conducted under anaerobic conditions. The full composition for YTF, YT, and ATCC 1754 media is as follows:(i)YTF medium (per liter, pH 6.0): yeast extract 10 g, tryptone 16 g, fructose 10 g, sodium chloride 0.2 g, l-cysteine 0.3 g, 500 × trace element solution 2 mL (per liter : nitrilotriacetic acid 10 g, MnSO4 5 g, (NH4)2SO4·FeSO4·6H2O 4 g, CoCl2·6H2O 1 g, ZnSO4·7H2O 1 g, CuCl2·2H2O 0.1 g, NiCl2·6H2O 0.1 g, Na2MoO4·2H2O 0.1 g, Na2SeO4 0.1 g, Na2WO4·2H2O 0.1 g), 1000 × vitamin solution 1 mL (per liter: vitamin B6 100 mg, thiamine 100 mg, vitamin B2 100 mg, calcium d-(+)-pantothenate 50 mg, thioctic acid 50 mg, p-aminobenzoic acid 50 mg, nicotinic acid 50 mg, vitamin B12 50 mg, biotin 50 mg, folic acid 50 mg). YTF agar was prepared by adding 15 g of agar per liter to the YTF media.(ii)YT medium (per liter, pH 6.0) was obtained by removing the fructose component from YTF media and adding 5 mL of reducing agent stock (per liter: NaOH 9 g, l-cysteine·HCl 40 g, Na2S·9H2O 40 g).(iii)Fructose-free ATCC 1754 medium (per liter, pH 6.0): yeast extract 0.5 g, 2-morpholinoethanesulfonic acid 5 g, mineral solution stock 30 mL (per liter: NaCl 80 g, NH4Cl 100 g, KCl 10 g, KH2PO4 10 g, MgSO4·7H2O 20 g, CaCl2·2H2O 4 g), 500 × trace element solution 2 mL, 1000 × vitamin solution 1 mL, 5 mL reducing agent stock (same as above). (i)YTF medium (per liter, pH 6.0): yeast extract 10 g, tryptone 16 g, fructose 10 g, sodium chloride 0.2 g, l-cysteine 0.3 g, 500 × trace element solution 2 mL (per liter : nitrilotriacetic acid 10 g, MnSO4 5 g, (NH4)2SO4·FeSO4·6H2O 4 g, CoCl2·6H2O 1 g, ZnSO4·7H2O 1 g, CuCl2·2H2O 0.1 g, NiCl2·6H2O 0.1 g, Na2MoO4·2H2O 0.1 g, Na2SeO4 0.1 g, Na2WO4·2H2O 0.1 g), 1000 × vitamin solution 1 mL (per liter: vitamin B6 100 mg, thiamine 100 mg, vitamin B2 100 mg, calcium d-(+)-pantothenate 50 mg, thioctic acid 50 mg, p-aminobenzoic acid 50 mg, nicotinic acid 50 mg, vitamin B12 50 mg, biotin 50 mg, folic acid 50 mg). YTF agar was prepared by adding 15 g of agar per liter to the YTF media. YTF medium (per liter, pH 6.0): yeast extract 10 g, tryptone 16 g, fructose 10 g, sodium chloride 0.2 g, l-cysteine 0.3 g, 500 × trace element solution 2 mL (per liter : nitrilotriacetic acid 10 g, MnSO4 5 g, (NH4)2SO4·FeSO4·6H2O 4 g, CoCl2·6H2O 1 g, ZnSO4·7H2O 1 g, CuCl2·2H2O 0.1 g, NiCl2·6H2O 0.1 g, Na2MoO4·2H2O 0.1 g, Na2SeO4 0.1 g, Na2WO4·2H2O 0.1 g), 1000 × vitamin solution 1 mL (per liter: vitamin B6 100 mg, thiamine 100 mg, vitamin B2 100 mg, calcium d-(+)-pantothenate 50 mg, thioctic acid 50 mg, p-aminobenzoic acid 50 mg, nicotinic acid 50 mg, vitamin B12 50 mg, biotin 50 mg, folic acid 50 mg). YTF agar was prepared by adding 15 g of agar per liter to the YTF media. (ii)YT medium (per liter, pH 6.0) was obtained by removing the fructose component from YTF media and adding 5 mL of reducing agent stock (per liter: NaOH 9 g, l-cysteine·HCl 40 g, Na2S·9H2O 40 g). YT medium (per liter, pH 6.0) was obtained by removing the fructose component from YTF media and adding 5 mL of reducing agent stock (per liter: NaOH 9 g, l-cysteine·HCl 40 g, Na2S·9H2O 40 g). (iii)Fructose-free ATCC 1754 medium (per liter, pH 6.0): yeast extract 0.5 g, 2-morpholinoethanesulfonic acid 5 g, mineral solution stock 30 mL (per liter: NaCl 80 g, NH4Cl 100 g, KCl 10 g, KH2PO4 10 g, MgSO4·7H2O 20 g, CaCl2·2H2O 4 g), 500 × trace element solution 2 mL, 1000 × vitamin solution 1 mL, 5 mL reducing agent stock (same as above). Fructose-free ATCC 1754 medium (per liter, pH 6.0): yeast extract 0.5 g, 2-morpholinoethanesulfonic acid 5 g, mineral solution stock 30 mL (per liter: NaCl 80 g, NH4Cl 100 g, KCl 10 g, KH2PO4 10 g, MgSO4·7H2O 20 g, CaCl2·2H2O 4 g), 500 × trace element solution 2 mL, 1000 × vitamin solution 1 mL, 5 mL reducing agent stock (same as above). For all the abovementioned media (per liter), 500 μL of 0.1% (w/v) resazurin solution was added. Plasmid construction The plasmids and primers used in this study are listed in Tables S1 and S2, respectively. All the plasmids used for C. ljungdahlii in this study were derived from the plasmid pMTL83151. The plasmid pCYG4 was constructed from the previously reported pYG3 [39]. Briefly, the ParaE promoter was PCR-amplified from C. ljungdahlii genomic DNA using primers ParaE-F/ParaE-R. This promoter was then used as a template with primers ParaE-F and CD-R to generate the ParaE-CD0164 fragment. Simultaneously, the LacI-Pfac-LacO fragment was amplified from plasmid LacI-Pfac-LacO [38] using primers ParaE-CD0164-F and LacI-Pfac-LacO-R. These two fragments, ParaE-CD0164 and LacI-Pfac-LacO, were subsequently fused by overlap extension PCR with primers ParaE-F and LacI-Pfac-LacO-R, yielding a larger fragment ParaE-CD0164-LacI-Pfac-LacO. Finally, this large fragment and the pYG3 vector were separately digested with Bpu10I and BseRI, and then ligated to yield the target plasmid, pCYG4. The construction of the plasmid for overexpressing the CLJU_c04490 gene was performed as follows. The plasmid pMTL83151-Ppta-IPA1 [36] was linearized by double digestion with KpnI and NcoI, and the CLJU_c04490 gene was PCR-amplified from C. ljungdahlii genomic DNA using primers CLJU_c04490-F and CLJU_c04490-R. The purified PCR product was subsequently assembled with the linear pMTL83151 vector using a Hieff Clone™ Plus Multi One Step Cloning Kit (Yeasen Biotech Co., China), yielding the target plasmid pMTL83151-Ppta-CLJU_c04490. The plasmids for overexpressing the other genes were constructed via the same steps. The construction of the CRISPRi (CRISPR interference) plasmid for repressing the CLJU_c09160 gene was performed as follows. The plasmid pZG-ddFnCas12a [40] was linearized by double digestion with BamHI and NcoI to provide the vector backbone. The sgRNA expression cassette was then generated via a two-step PCR process: first, a primer extension reaction was performed using primers CLJU_c09160-sg-F and CLJU_c09160-sg-R without a DNA template; second, the product of this reaction served as the template for amplification with primers Pthl-F and Fdx-NcoI-R. Finally, the resulting cassette was assembled with the linearized pZG-ddFnCas12a backbone using a Hieff Clone™ Plus Multi One Step Cloning Kit (Yeasen Biotech Co., China), yielding the plasmid pZG-ddFnCas12a-CLJU_c09160. The other plasmids for CRISPRi were constructed following the same steps. Plasmid isolation and purification of DNA fragments from agarose gels were carried out using the EasyPure® Plasmid MiniPrep Kit (TransGen Bioech, Beijing, China) and EasyPure® PCR Purification Kit (TransGen Bioech, Beijing, China), respectively. Restriction enzymes and ligase were purchased from Thermo Fisher Scientific (Thermo Fisher Scientific, Vilnius, Lithuania) and Takara (Takara, Dalian, China), respectively. Plasmids were constructed using the Hieff Clone™ Plus Multi One Step Cloning Kit (Yeasen Biotech Co., China). Plasmid transfer in C. ljungdahlii and isolation of transposon insertion mutants The plasmid pCYG4 was transferred into C. ljungdahlii via electroporation as described previously [40]. The transformed cells (containing the pCYG4 plasmid) were recovered in the liquified YTF medium supplemented with Tm for 20 h. Subsequently, the bacterial culture was spread on the YTF agar plates (containing Tm). When colonies were visible on the plates, all colonies were harvested and transferred into the liquified YTF medium (no antibiotics). Next, IPTG was added into the medium to drive plasmid elimination. After several rounds of subculturing, 200 μL of bacterial culture was spread onto the YTF agar plates (no antibiotics) again. When colonies were visible, some were picked and patch plated onto the YTF agar plates containing either Tm or Em to confirm the percentage of chromosomal transposon insertion and plasmid elimination in C. ljungdahlii cells. Inverse PCR for identification of transposon insertion sites on chromosome Inverse PCRs for identifying the insertion sites of pYG3 in C. ljungdahlii were carried out according to previously described methods [30]. Briefly, genomic DNA was isolated from C. ljungdahlii strains. A total of 10 μg of genomic DNA was digested overnight with 2 μL of HindIII restriction endonuclease (200 ng/μL). The enzyme was inactivated by incubation at 65 °C for 30 min. The digested DNA was then diluted to a concentration of 5 ng/μL, and 1 μL of T4 DNA ligase (Takara, Dalian, China) was then added to facilitate self-ligation of the DNA fragments. Ligation was carried out at 16 °C overnight. Next, the T4 DNA ligase was inactivated by heating at 65 °C for 30 min. Inverse PCR was performed in 50 μL reaction volumes using the primers CatP-INV-F1 and CatP-INV-R1. The resulting PCR products were sequenced using the CatP-INV-F1 primer. Cultivation on syngas Cultivation of the C. ljungdahlii strains on syngas was performed in 125 mL serum bottles (Sigma-Aldrich, Merck Life Science, Gillingham, UK) with a 30 mL working volume. Briefly, 200 μL of frozen stock was inoculated into the liquified YTF medium (5 mL) and incubated anaerobically at 37 °C for 24 h. When the optical density (OD600) of the grown cells reached ∼ 0.8, 5% (v/v) of the inoculum was transferred to 30 mL of YT medium (for continuous passaging) or fructose-free ATCC 1754 medium (for endpoint cultivation assays) with uncontrolled pH. The culture bottles were pressurized to 0.2 MPa with a CO/CO2/H2/N2 mixture (56%/20%/9%/15%). The gases in the headspace were replaced every 24 h. The serum bottles were placed in a shaking incubator (37 °C, 100 rpm). Analytical methods Cell growth was determined by measuring the optical density at 600 nm (OD600) using a spectrophotometer (DU730; Beckman Coulter, USA). The concentrations of ethanol and acetate were quantified using the Agilent 7890A gas chromatography system equipped with a capillary column (Alltech ECTM-WAX) and a flame ionization detector. The analytical conditions were consistent with those described in previous studies [41]. The gases in the headspace were determined by a gas chromatograph (Agilent Technologies-7890A) equipped with a thermal conductivity detector (TCD) and packed column [42]. Helium gas was used as the carrier gas at a flow rate of 2 mL/min. The initial oven temperature was set at 60 °C and held for 1 min, followed by a ramp of 20 °C/min to 80 °C and maintained for 1 min. Subsequently, the oven was heated at 40 °C/min to 140 °C and held for 4.5 min. Total carbon consumption was determined as follows. At 24, 48, and 72 h of fermentation, the headspace pressure in the serum bottle was measured using a handheld pressure gauge (AZ82100, AZ Instrument, Taiwan, China) connected to the bottle headspace. After the pressure reading stabilized, the value was recorded in megapascals (MPa). Subsequently, a gas-tight syringe was used to collect gas samples from the headspace, which were then injected into a gas chromatograph (Agilent Technologies-7890A) to determine the molar mass of carbon dioxide (CO2) and carbon monoxide (CO) in the headspace. Using the obtained data, the amounts (in moles) of CO and CO2 were calculated according to the ideal gas law (n = PV/RT), where P is the measured headspace pressure, V is the headspace volume of the serum bottle, R is the ideal gas constant, and T is the absolute temperature of the culture environment. Off-target analysis of sgRNAs was performed using the Benchling platform (https://www.benchling.com) to evaluate the off-target specificity of each sgRNA. This platform employs an off-target prediction model based on the MIT scoring algorithm [43]. The underlying principle involves aligning each sgRNA sequence against the reference genome to identify potential off-target sites containing the TTN PAM sequence, allowing for a certain number of mismatches. Subsequently, a risk score is calculated for each potential off-target site based on the position of mismatches (mismatches in the seed region proximal to the PAM are assigned higher weights) and the degree of PAM sequence matching. Finally, an overall off-target score for each sgRNA is aggregated using the formula "100/(100 + sum of risk scores of all off-target sites)". The score ranges from 0 to 100, with higher scores indicating lower off-target risk. Screening experiments The preserved library culture was inoculated into a fructose-containing medium and grown to an OD600 of approximately 1.0. It was then subcultured at a 5% inoculation ratio into YT medium and incubated under two distinct gas atmospheres (N2 as the control condition and syngas as the selective pressure condition). When the cultures entered the logarithmic growth phase, they were again subcultured at a 5% inoculation ratio into fresh YT medium. This process was repeated for a total of five consecutive passages. Transposon insertion sequencing analysis Tn-seq experimental procedures and data analysis were performed as follows [44]. The gDNA (5 μg) was diluted in 100 μL TE, and sonicated in a Bioruptor (Diagenode, Belgium) with 30 s ON/60 s OFF for 12 cycles, yielding 200 ∼ 800 bp fragments. End repair and 3′ end A-tailing was performed with 1 μg sheared DNA and the VAHTS turbo DNA library preparation kit (Vazyme, Nanjing, China), and adaptor ligation was performed with forked adaptors. Two rounds of PCR amplification were performed to enrich for sequences adjacent to transposon insertion sites and attach Illumina P5 and P7 hybridization sequences and barcodes for multiplexing. Final products (300 to 500 bp) were isolated from a 2% agarose gel and extracted with a gel purification column (Qiagen, Germany). High-throughput sequencing was performed on HiSeq 2500 platforms (Illumina, San Diego, CA, USA). The sequencing protocol was 65-cycle single end read plus 7-cycle index read. Sequencing generated 2 to 3 million reads for each library. The raw data were processed to remove adaptor sequences by CutAdapt software, and then mapped to the C. ljungdahlii DSM 13528 genome using Bowtie. After mapping, the reads per TA site were tallied and assigned to annotated genes or intergenic regions using EL-ARTIST. Essential loci were determined using the HMM module of EL-ARTIST (window size of 10 TA sites, P-value of 0.01) [45]. The total read counts of all loci within a given gene were summed to represent the read counts of that gene. Finally, the fold change (FC) for each gene was calculated as the normalized read count in the experimental group (S group) divided by that in the corresponding control group (N group). Subsequently, the differential analysis was conducted using the limma package in R, which employs an empirical Bayesian approach to estimate moderated t-statistics. To control the false discovery rate (FDR) at 5%, we applied the Benjamini-Hochberg (BH) procedure to adjust the raw P-values. Results Validation of the mariner transposon fitness in C. ljungdahlii A previously constructed mariner transposon delivery plasmid, pYG3 (Fig. S1a), had been used to develop a mutant library in the solventogenic bacterium Clostridium acetobutylicum [39]. Based on this precedent, we hypothesized that the same transposon delivery system would also be functional in the gas-fermenting species C. ljungdahlii. To test this hypothesis, pYG3 was introduced into C. ljungdahlii following the workflow outlined in Fig. S1b. Hundreds of transformants were obtained on solid agar supplemented with thiamphenicol (Tm) (Fig. S1b), and these transformants were subsequently analyzed to assess the efficiency of transposon insertion and plasmid elimination. A total of 337 colonies were selected and patch-plated onto solid agar plates containing Tm, erythromycin (Em), or no antibiotics. After overnight cultivation, all 337 colonies grew on antibiotic-free medium. Among these, 317 colonies grew on Tm-containing plates, indicating successful transposon insertion, while 226 colonies grew on Em-containing plates, which indicates retention of the original plasmid. These results indicate a plasmid elimination efficiency of only 33% ([337 − 226]/337). To further estimate transposon insertion efficiency, 31 Tm-resistant colonies on the agar plate were randomly picked out and analyzed by inverse PCR. Sequencing of the amplified DNA fragments confirmed that all 31 carried a single transposon insertion, distributed randomly across the genome (Fig. S2). These results demonstrate that the pYG3 plasmid can achieve efficient and relatively random chromosomal integration in C. ljungdahlii; however, the low plasmid elimination rate was observed in the library, which may affect genetic stability of mutants and requires further optimization. Optimization of the plasmid elimination efficiency To improve plasmid elimination efficiency, we modified pYG3 by inserting an IPTG-inducible cassette (Pptb::lacI::Pfac) upstream of the pCB102 replicon (repH), with the goal of constructing a conditionally maintained replicon (Fig. 1a). This IPTG-inducible cassette has previously been shown to interfere with the downstream replicon in the presence of IPTG, resulting in plasmid loss in Clostridium acetobutylicum and Clostridium sporogenes [46]. The resulting plasmid was designated pCYG4.Fig. 1Construction and optimization of transposon mutagenesis system. (a) A schematic diagram of the plasmid pCYG4. himar1 C9: the transposase derived from himar1; ParaE: the promoter of the araE gene in C. acetobutylicum ATCC 824; LacI: the LacI repressor protein gene; Pptb: the promoter of the ptb gene from Clostrisium beijerinckii NCIMB 8052; Pfac: the promoter of the fac gene from Clostridium pasteurianum DSM 525; LacO: the binding region of LacI; repH: the pCB102 replicon from Clostridium butyricum. (b) The workflow for assessing the efficiency of chromosomal transposon insertion and plasmid curing in C. ljungdahlii. Thiamphenicol (Tm) was used for screening clonies inserted by transposon and erythromycin (Em) was used to estimate the plasmid elimination. (c) Genetic map of mariner transposon insertion sites in the C. ljungdahlii. From inner to outer: circle 1, the density distribution of insertion sites; circle 2, the predicted genomic island loci; circle 3, GC content variation; circle 4, antisense-strand gene locations; circle 5: sense-strand gene locations.Fig. 1 Construction and optimization of transposon mutagenesis system. (a) A schematic diagram of the plasmid pCYG4. himar1 C9: the transposase derived from himar1; ParaE: the promoter of the araE gene in C. acetobutylicum ATCC 824; LacI: the LacI repressor protein gene; Pptb: the promoter of the ptb gene from Clostrisium beijerinckii NCIMB 8052; Pfac: the promoter of the fac gene from Clostridium pasteurianum DSM 525; LacO: the binding region of LacI; repH: the pCB102 replicon from Clostridium butyricum. (b) The workflow for assessing the efficiency of chromosomal transposon insertion and plasmid curing in C. ljungdahlii. Thiamphenicol (Tm) was used for screening clonies inserted by transposon and erythromycin (Em) was used to estimate the plasmid elimination. (c) Genetic map of mariner transposon insertion sites in the C. ljungdahlii. From inner to outer: circle 1, the density distribution of insertion sites; circle 2, the predicted genomic island loci; circle 3, GC content variation; circle 4, antisense-strand gene locations; circle 5: sense-strand gene locations. We then introduced pCYG4 into C. ljungdahlii to evaluate its transposon insertion efficiency into the chromosome, following the experimental procedure outlined in Fig. 1b. After electroporation and IPTG-induced plasmid curing, hundreds of transformants were obtained on antibiotic-free agar plates. Subsequently, 64 colonies were selected from agar plates and patch-plated onto solid media containing either Tm or Em. After overnight cultivation, all 64 colonies grew on Tm-containing plates, indicating successful transposon insertion in all the 64 isolates, whereas only 4 colonies (those still harboring the plasmid) grew on Em-containing plates. Accordingly, the plasmid elimination rate of pCYG4 was 93.75% (60/64). Together, these findings demonstrate that pCYG4 enables efficient chromosomal transposon insertion while overcoming the low plasmid elimination efficiency associated with pYG3. This makes pCYG4 a more suitable vector for constructing a genome-wide mutant library in C. ljungdahlii. However, it should be noted that, although IPTG-induced plasmid curing of pCYG4 in Clostridium species has been demonstrated previously [39,46], leakiness of repH expression in vivo remains a possibility, which could compromise plasmid stability and potentially affect the quality of the mutant library. Generation of a genome-wide transposon mutant library of C. ljungdahlii Since pCYG4 enabled efficient chromosomal transposon insertion and plasmid curing in C. ljungdahlii, we proceeded to construct a genome-wide transposon mutant library. This resource is designed for use with high-throughput sequencing to rapidly pinpoint critical genes involved in important phenotypes, such as growth and product formation under gas fermentation conditions. To obtain enough transformants for library construction [47,48], ten rounds of electroporation were performed (Fig. S3), yielding approximately 1.9 × 105 transformants. Based on the genome size of C. ljungdahlii (4.63 Mbp, comprising a total of 4296 coding and non-coding regions), this number of transformants represents more than 40-fold coverage of the coding and non-coding sequences. We assessed the distribution and coverage of transposon insertions by performing Transposon-Insertion Sequencing (Tn-seq) on the mutant library's genomic DNA. The results revealed insertions in 3197 genes and an equal number of non-coding regions, achieving ∼75% coverage of all genomic features (Fig. 1c & Supplementary Data1). The insertions were uniformly distributed genome-wide, showing no apparent bias or hotspots (Fig. 1c & Supplementary Data1). The 25% of genes without insertions likely include genes that are essential for cell growth under the heterotrophic (YTF medium) cultivation condition (Supplementary Data2), which typically represent 10%∼20% of a bacterial genome [48]. Given that the genes responsible for product synthesis (the primary focus of this study) are generally non-essential for growth, we consider the 75% coverage of our mutant library to be sufficient for screening these genes in C. ljungdahlii. After deduplication of the sequencing data, a total of 16,922 unique insertion sites were identified in this transposon mutant library, of which 16,192 (95.7%) were located at TA dinucleotide sites (Supplementary Data3). This high proportion indicates strong TA-site specificity of the library, consistent with the fundamental characteristics of mariner family transposons. The observed proportion is slightly lower than the theoretical optimum (typically >99%), which may be attributable to minimal carryover of vector backbone, non-specific PCR amplification products, or sequencing background noise during library construction. Such signals can be effectively removed by bioinformatics filtering, with limited impact on downstream functional analyses. Furthermore, based on the NCBI genome database, the genome of this strain contains a total of 509,018 TA sites, yielding the TA-site saturation of approximately 3.2% (Supplementary Data3). This relatively low saturation reflects a certain degree of duplication bias in transposon insertions, i.e., multiple transformants carrying identical insertion sites. This phenomenon may be related to differential clonal expansion during post-transformation cultivation, local hotspot effects for transposon insertion, or PCR amplification bias during sequencing library preparation. Despite the low TA-site saturation, the library covers nearly 1.7 × 104 unique insertion sites. Combined with an overall transformant pool size of 1.9 × 105, the library still holds potential for supporting subsequent functional screening, provided that appropriate quality control measures based on insertion diversity are implemented during data analysis. In addition, we observed a positive correlation between the number of insertion events (insertion frequency) and gene length (Fig. S4). A plausible explanation is that longer genes occupy larger physical intervals in the genome, thereby presenting a larger "target area" for transposon insertion and thus a higher likelihood of being inserted. Notably, some short genes show exceptionally high numbers of unique insertions, which may be due to the high TA content within these genes or selective pressure related to gene function. Conversely, some long genes exhibit unexpectedly low numbers of insertion events, potentially influenced by factors such as chromatin structure, expression level, and DNA accessibility. Screening of genes associated with C. ljungdahlii fitness to syngas Next, we employed the transposon mutant library to screen for genes linked to C. ljungdahlii fitness under syngas-containing conditions. The library was subjected to serial subculturing under two conditions: an experimental condition with syngas (CO/CO2/H2) and a control with nitrogen gas. This process enriches for population composition differences between the two conditions, thereby facilitating the identification of relevant genes through subsequent deep sequencing. Genes showing significant abundance differences between the syngas and control groups in the sequencing data will be identified as potential candidates associated with cellular fitness to syngas. The specific procedure is outlined in Fig. 2a: cultures at the logarithmic growth phase (OD600 ≈ 0.6) were serially subcultured at a 5% (v/v) inoculation ratio. Bacterial samples from the 1st, 3rd, and 5th generations were collected for total DNA extraction and subsequent Tn-seq analysis.Fig. 2Construction and Tn-seq analysis of the transposon mutant library of C. ljungdahlii. (a) Workflow for screening C1 gas utilization genes using Tn-seq. The mutant library was first resuscitated in YTF medium with thiamphenicol to an OD600 of ∼1.0. Cultures were then inoculated at 5% (v/v) into YT medium (without fructose) and incubated under either a nitrogen atmosphere (control) or syngas (experimental). Upon reaching an OD600 of ∼0.6, cultures were serially subcultured four more times at the same inoculation ratio. Genomic DNA was extracted from the 1st, 3rd, and 5th generation samples (six groups in total) for high-throughput sequencing to analyze changes in transposon insertion abundance. (b) Principal component analysis (PCA) of the sequenced samples. Strains S1, S3, and S5 represent the first, third, and fifth generations cultured in synthesis gas (syngas), while strains N1, N3, and N5 correspond to the first, third, and fifth generations cultivated under nitrogen (N2), respectively. Data are presented as mean ± SD (n = 5 biological replicates). (c–e) Analysis of differentially abundant genes of the 1st, 3rd and 5th generations (|log2FC| > 1, P < 0.05). Red dots: genes with significantly increased abundance; blue dots: genes with significantly decreased abundance. (f) KEGG pathway mapping and abundance heatmap of differentially abundant genes shared between the 3rd and 5th generations. The heatmap displays the log2FC values of the 41 differentially expressed genes (|log2FC| > 2, P < 0.05) that are shared between the third- and fifth-generation strains, respectively.Fig. 2 Construction and Tn-seq analysis of the transposon mutant library of C. ljungdahlii. (a) Workflow for screening C1 gas utilization genes using Tn-seq. The mutant library was first resuscitated in YTF medium with thiamphenicol to an OD600 of ∼1.0. Cultures were then inoculated at 5% (v/v) into YT medium (without fructose) and incubated under either a nitrogen atmosphere (control) or syngas (experimental). Upon reaching an OD600 of ∼0.6, cultures were serially subcultured four more times at the same inoculation ratio. Genomic DNA was extracted from the 1st, 3rd, and 5th generation samples (six groups in total) for high-throughput sequencing to analyze changes in transposon insertion abundance. (b) Principal component analysis (PCA) of the sequenced samples. Strains S1, S3, and S5 represent the first, third, and fifth generations cultured in synthesis gas (syngas), while strains N1, N3, and N5 correspond to the first, third, and fifth generations cultivated under nitrogen (N2), respectively. Data are presented as mean ± SD (n = 5 biological replicates). (c–e) Analysis of differentially abundant genes of the 1st, 3rd and 5th generations (|log2FC| > 1, P < 0.05). Red dots: genes with significantly increased abundance; blue dots: genes with significantly decreased abundance. (f) KEGG pathway mapping and abundance heatmap of differentially abundant genes shared between the 3rd and 5th generations. The heatmap displays the log2FC values of the 41 differentially expressed genes (|log2FC| > 2, P < 0.05) that are shared between the third- and fifth-generation strains, respectively. Principal Component Analysis (PCA) of the read counts revealed that all the biological replicate samples (5 biological replicates in each group) formed distinct clusters in the principal component space, demonstrating the reproducibility of the experimental system (Fig. 2b). Under the selective pressure of syngas, the experimental group exhibited a clear intergenerational evolutionary trajectory: the first (S1), third (S3), and fifth (S5) generation of samples showed a stepwise shift along the first principal component (PC1), with the divergence between successive generations becoming progressively greater (Fig. 2b). In contrast, the control group samples (N1, N3, N5), maintained under non-selective conditions, displayed a broader distribution along the second principal component (PC2) while showing less spread along PC1. This spatial separation, characterized by a directional shift of the experimental group along PC1 alongside a broader, less constrained variation of the control group along PC2, demonstrates that the gaseous environment drives rapid and directional differentiation within the mutant library through the sustained imposition of metabolic stress. As subculturing progressed across generations, bacterial strains harboring advantageous or deleterious mutations became increasingly distinguishable phenotypically under the applied selective regime. Based on the above high-throughput Tn-seq data, we performed a differential gene abundance analysis using thresholds of P < 0.05 and |log2FC| > 1. Compared to nitrogen cultivation conditions, loss-of-function mutants depleted under the syngas condition indicate genes whose presence is beneficial; thus, overexpression of these genes may contribute to cellular fitness under syngas conditions. Conversely, mutants enriched under the syngas condition may reveal genes whose reduced activity is advantageous; CRISPRi could therefore mimic this phenomenon. The results showed that the number of genes with significant abundance changes increased substantially over successive subcultures (Fig. 2c–e). In the first generation, 62 genes were significantly enriched and 127 were depleted. This number expanded to 89 enriched and 307 depleted genes by the third generation, and further to 108 enriched and 967 depleted genes by the fifth generation (Supplementary Data4). This rapid rise in the number of differentially abundant genes across generations demonstrates that the experimental procedure effectively enriches for genes associated with the fitness of C. ljungdahlii to syngas. Given the substantial number of differentially abundant genes identified in both the third- and fifth-generation samples (Fig. 2d and e), we integrated the data from these two time points and performed a KEGG pathway mapping on their overlapping gene set. Using a screening threshold of |log2FC| > 2 and P < 0.05, we identified 114 genes with significantly different abundances that were shared between the third and fifth library microbiotas from the Tn-seq datasets (Fig. S5). Subsequently, KEGG pathway mapping was performed using these 114 genes as the input set, of which 41 genes were successfully mapped to KEGG pathways, collectively involved in 44 pathways (Fig. 2f). Pathways with higher gene counts included porphyrin metabolism, folate biosynthesis (encompassing the one-carbon pool by folate), carbon metabolism, cofactor biosynthesis, secondary metabolite biosynthesis, and several carbon fixation pathways (i.e., the Wood-Ljungdahl pathway, classified in KEGG as “other carbon fixation pathways”; carbon fixation by the Calvin cycle). Notably, many of these pathways are plausibly linked to the gas-utilizing physiology of C. ljungdahlii. For instance, folate metabolism supplies the one-carbon carrier methyl-THF, which is essential for the Wood-Ljungdahl Pathway (WLP), while the biosynthesis of cofactors such as thiamine and cobalamin is critical for the activity of key enzymes like formate dehydrogenase (FDH), hydrogenase, and carbon monoxide dehydrogenase [[49], [50], [51]]. Identification of genes affecting acetate and ethanol formation of C. ljungdahlii under syngas conditions Having identified a set of genes potentially linked to C. ljungdahlii fitness under syngas conditions via Tn-seq, we next sought to investigate their specific roles in product synthesis. To this end, 25 genes were randomly selected from the top 50 genes with the most significant decrease in abundance (Table S4) for overexpression, while 20 genes were randomly selected from the top 25 genes with the most significant increase in abundance (Table S5) for transcriptional repression. As shown in Fig. 3, the overexpression of five genes significantly enhanced ethanol synthesis of C. ljungdahlii on syngas. These overexpressed genes and their encoded proteins were: CLJU_c04490 (glutamate-1-semialdehyde 2,1-aminomutase, HemL), CLJU_c25060 (molybdopterin adenylyltransferase, MoeA), CLJU_c06990 (formate dehydrogenase, FDH), CLJU_c22650 (nickel transporter system permease), and CLJU_c07720 (Crp/Fnr-family transcriptional regulator). The five engineered strains exhibited no significant differences in growth compared to the control strain (Fig. 3a). One possibility is that the inoculum used for sub-culturing was a mixed population containing a variety of mutants, which may have allowed for interactions among them. For instance, some mutants might have been promoted by others through metabolites or signaling molecules, leading to their increased abundance within the community. However, once isolated and grown individually, these strains might lose such advantages due to the absence of those community-specific promoting factors. It is therefore possible that the five engineered strains do not directly drive their own growth; rather, they may contribute to adaptive benefits through metabolic interactions or quorum sensing within a specific microbial community microenvironment.Fig. 3Characterization of the CLJU_c04490, CLJU_c25060, CLJU_c06990, CLJU_c22650, and CLJU_c07720 gene. (a–d) The growth profiles (a), total carbon consumption (b), specific productivity at 72 h of ethanol (c), and acetate (d) of five overexpression strains. The "OE" (overexpression) strains, namely OE04490, OE25060, OE06990, OE22650, and OE07720, correspond to genetically engineered derivatives overexpressing the genes CLJU_c04490, CLJU_c25060, CLJU_c06990, CLJU_c22650, and CLJU_c07720, respectively. The strain harboring the empty pMTL83151 plasmid served as the control. Data are presented as mean ± SD (n = 3 biological replicates). Statistical significance was determined by Student's t-test (∗, P < 0.05; ∗∗, P < 0.01; ns, no significance). (e) Proposed regulatory network model integrating the functions of the five overexpressed genes in the context of syngas metabolism. Key enzyme abbreviations: CODH, carbon monoxide dehydrogenase; FDH, formate dehydrogenase; HemL, glutamate-1-semialdehyde 2,1-aminomutase; MeTr, methyltransferase; MoeA, molybdopterin adenylyltransferase.Fig. 3 Characterization of the CLJU_c04490, CLJU_c25060, CLJU_c06990, CLJU_c22650, and CLJU_c07720 gene. (a–d) The growth profiles (a), total carbon consumption (b), specific productivity at 72 h of ethanol (c), and acetate (d) of five overexpression strains. The "OE" (overexpression) strains, namely OE04490, OE25060, OE06990, OE22650, and OE07720, correspond to genetically engineered derivatives overexpressing the genes CLJU_c04490, CLJU_c25060, CLJU_c06990, CLJU_c22650, and CLJU_c07720, respectively. The strain harboring the empty pMTL83151 plasmid served as the control. Data are presented as mean ± SD (n = 3 biological replicates). Statistical significance was determined by Student's t-test (∗, P < 0.05; ∗∗, P < 0.01; ns, no significance). (e) Proposed regulatory network model integrating the functions of the five overexpressed genes in the context of syngas metabolism. Key enzyme abbreviations: CODH, carbon monoxide dehydrogenase; FDH, formate dehydrogenase; HemL, glutamate-1-semialdehyde 2,1-aminomutase; MeTr, methyltransferase; MoeA, molybdopterin adenylyltransferase. Further analysis of total carbon (CO and CO2) consumption revealed that, except for strain OE07720, which showed a 24.3% increase in total carbon consumption over 72 h relative to the control, the other strains displayed no significant differences (Fig. 3b). Notably, the specific productivity of ethanol of OE04490, OE25060, OE06990, OE22650, and OE07720 was increased by 26.2%, 16.9%, 95.2%, 13.8%, and 40.9%, respectively, compared to the control (Fig. 3c); meanwhile, the specific productivity of acetate of OE04490, OE06990, and OE07720 was decreased by 5.4%, 21.4%, and 22.4%, respectively, relative to the control, whereas OE25060 and OE22650 showed no significant differences (Fig. 3d). Here, a regulatory network model integrating the functions of the five overexpressed genes in the context of syngas (CO2/CO) metabolism was proposed (Fig. 3e). Among the 20 CRISPRi-based transcriptionally repressed strains (Fig. 4a), knockdown of CLJU_c09160 (encoding a lantibiotic transporter permease) and CLJU_c23840 (encoding an ornithine/aspartate carbamoyltransferase) significantly altered ethanol synthesis (Fig. 4d). Repression of CLJU_c09160 transcription also affected cell growth and acetate synthesis (Fig. 4b & e). Therefore, the final biomass (OD600) and specific acetate productivity of TR09160 were both reduced by 35.7% compared to the control (Fig. 4b & e), while its specific ethanol productivity was increased by 57.5% (Fig. 4d). Additionally, the strain TR23840 exhibited a 15.1% increase in specific ethanol productivity relative to the control (Fig. 4d), but showed no significant differences in growth or specific acetate productivity (Fig. 4b & e). Furthermore, individual repression of transcription of either gene had no significant effect on total carbon consumption (Fig. 4c). Here, it should be noted that the CRISPRi off-target analysis in this study was performed using the Benchling platform (based on the MIT algorithm; see section 2.6). As shown in Table S3, some target genes have off-target scores that are not high (<50), suggesting a certain possibility of off-target effects, which may influence the repression efficiency of these target genes and the associated phenotypic changes. Whether and to what extent off-target effects may have impacted the repression efficiency and phenotypic changes for these target genes remain to be further investigated.Fig. 4Characterization of the CLJU_c09160 and CLJU_c23840 gene. (a) Schematic of the CRISPRi shuttle plasmid used for targeted gene repression. The promoter Pthl and P1440 were used for driving the expression of crRNA (DR and N24) and ddFnCas12a, respectively. (b–e) The growth profiles (b), total carbon consumption (c), specific productivity at 72 h of ethanol (d), and acetate (e) of two engineered strains with transcriptional repression. The control strain harbored the plasmid expressing ddFnCas12a without a targeting crRNA. The "TR" (transcriptional repression) strains, namely TR09160 and TR23840, represent genetically engineered derivatives in which transcription of the target genes CLJU_c09160 and CLJU_c23840, respectively, has been specifically repressed using the ddFnCas12a-based CRISPRi system. Data are presented as mean ± SD (n = 3 biological replicates). Statistical significance was determined by Student's t-test (∗, P < 0.05; ∗∗∗, P < 0.001; ns, no significance).Fig. 4 Characterization of the CLJU_c09160 and CLJU_c23840 gene. (a) Schematic of the CRISPRi shuttle plasmid used for targeted gene repression. The promoter Pthl and P1440 were used for driving the expression of crRNA (DR and N24) and ddFnCas12a, respectively. (b–e) The growth profiles (b), total carbon consumption (c), specific productivity at 72 h of ethanol (d), and acetate (e) of two engineered strains with transcriptional repression. The control strain harbored the plasmid expressing ddFnCas12a without a targeting crRNA. The "TR" (transcriptional repression) strains, namely TR09160 and TR23840, represent genetically engineered derivatives in which transcription of the target genes CLJU_c09160 and CLJU_c23840, respectively, has been specifically repressed using the ddFnCas12a-based CRISPRi system. Data are presented as mean ± SD (n = 3 biological replicates). Statistical significance was determined by Student's t-test (∗, P < 0.05; ∗∗∗, P < 0.001; ns, no significance). Collectively, our findings demonstrate a partial overlap between the genes associated with cellular fitness to syngas, as identified by Tn-seq, and those affecting product synthesis. This core subset of genes represents highly promising targets for both fundamental investigation and applied metabolic engineering to develop strains with enhanced production capabilities. However, one thing to note is that enrichment does not necessarily imply that the associated genes are directly involved in product synthesis in C. ljungdahlii. Inactivation of these genes may also lead to subtle growth-rate effects or bottlenecks unrelated to product formation. Discussion This study represents the first application of genome-wide Tn-seq in C. ljungdahlii. By optimizing a conditional replication system, we constructed the transposon delivery plasmid pCYG4, which enables efficient chromosomal insertion and inducible clearance, yielding a stable mutant library covering approximately 75% of genomic features. Based on this, we systematically identified gene related to cellular fitness and product synthesis under syngas conditions by gene overexpression or CRISPRi. Compared to Tn-seq studies in other gas-fermenting Clostridium species such as C. autoethanogenum [31], the gene clusters enriched in our study include many previously unreported genes (Supplementary Data2), particularly those capable of influencing the ethanol-to-acetate ratio. This work provides reliable genetic resources and functional data to support both fundamental and applied research on this bacterium. Our integrated Tn-seq analysis captured the dynamic adaptation of the transposon library population during syngas-containing cultivation. Comparative analysis identified 114 genes with consistent abundance changes shared between the 3rd and 5th generations under syngas versus control conditions (Fig. S5), implicating their potential roles in the fitness of C. ljungdahlii to syngas. Functional enrichment further indicated that these genes cluster in key metabolic categories (Fig. 2f), including folate-mediated CO/CO2 transfer, central carbon metabolism, porphyrin metabolism, and cofactor biosynthesis. Notably, most of these genes lack prior experimental characterization, despite their putative involvement in CO/CO2 metabolism. Through analysis of the genes with the most significant changes in abundance, we found that genes involved in energy/cofactor supply for carbon fixation and the maturation of key enzymes showed the significant decrease in abundance (Table S4). Among them, genes encoding Rnf complex subunits (rnfC CLJU_c11360, rnfA CLJU_c11400, rnfB CLJU_c11410) are known to be critical for ferredoxin:NAD+ oxidoreductase-driven ion gradient formation, thereby affecting intracellular ATP synthesis [52]. Their importance in gas-fermenting clostridia has also been demonstrated [53]. In addition, genes involved in cobalamin synthesis (cbiK CLJU_c31850, cbiT CLJU_c31900, cbiG CLJU_c31870, cbiE CLJU_c31910, and CLJU_c31890) and molybdenum cofactor synthesis (moeA CLJU_c07000, CLJU_c21410, CLJU_c25060, and CLJU_c33380) can affect methyl transfer and formate dehydrogenase activity [54,55], which is particularly important for the Wood-Ljungdahl carbon fixation pathway. In addition, the genes encoding the alpha subunit of formate dehydrogenase (CLJU_c06990), the CODH nickel-insertion accessory protein (CLJU_c37660), and the dihydrolipoamide dehydrogenase (CLJU_c37600) are crucial for substrate activation, maturation of key enzymes, and regeneration of reducing equivalents, respectively, within the Wood-Ljungdahl pathway [[56], [57], [58], [59]]. Hence, they may also contribute to carbon fixation in C. ljungdahlii. Notably, the abundance of several genes encoding transcriptional regulators (e.g., Crp/Fnr family CLJU_c07720, CtsR family CLJU_c41360, AraC family CLJU_c13540, and the sigma factor negative regulator RseC CLJU_c11350) were significantly decreased, suggesting that these regulatory factors play roles in modulating the growth of C. ljungdahlii on syngas. Conversely, the genes showing significantly increased abundance imply that suppressing their related metabolic pathways could provide the host with a competitive edge during gas fermentation. Our results showed a notable enrichment of genes involved in energy metabolism, carbon/nitrogen utilization, and flux distribution (Table S5). Furthermore, it remains unknown how a number of genes with significantly increased abundance are associated with the growth of C. ljungdahlii with syngas. These include the cell division protein FtsI1 gene (CLJU_c10560), the phage infection protein gene (CLJU_c32080), the ornithine/aspartate carbamoyltransferase gene (CLJU_c23840), the geranyl transferase gene (fps, CLJU_c11150), a spore germination protein KA-encoding gene (CLJU_c08490), a deacetylase gene (CLJU_c23810), a nucleotide kinase-related gene (CLJU_c17140), a hydrolase gene of unknown function (CLJU_c01110), and several hypothetical protein genes (CLJU_c01910, CLJU_RS22435). Nonetheless, these genes offer potential targets for future investigations into the physiology and metabolic mechanisms of syngas-fermenting clostridia. Based on the functional perturbation results of 45 candidate genes, we identified seven that significantly enhanced ethanol yield (Fig. 3, Fig. 4). According to their annotated functions, we propose that they may influence core metabolic processes through the following mechanisms (Fig. 3e): (i) Cofactor synthesis and precursor supply. CLJU_c04490 (HemL) catalyzes the synthesis of 5-aminolevulinate, a universal precursor for coenzyme B12 [60]. Since coenzyme B12 is a known cofactor for methyl group transfer [[61], [62], [63]], one plausible hypothesis is that enhanced HemL activity could increase coenzyme B12 availability, thereby potentially supporting CH3–COFeSP formation via methyl transfer. This might subsequently favor acetyl-CoA synthesis from CO and enhance ethanol flux; however, this proposed causal link between HemL perturbation and cobalamin-dependent methyl transfer has not been directly tested and remains to be experimentally validated. CLJU_c25060 encodes molybdopterin adenylyltransferase (MoeA), a key enzyme in the synthesis of the molybdenum cofactor (Moco). This cofactor forms the active center of various oxidoreductases, including formate dehydrogenase (FDH), and is essential for electron transfer [64,65]. We therefore speculate that its perturbation may influence Moco-dependent enzymatic steps, which could indirectly affect metabolic efficiency. (ii) Carbon fixation and substrate conversion. CLJU_c06990 encodes formate dehydrogenase (FDH), which catalyzes the reduction of CO2 to formate [66]. Overexpression of this carbon-fixing enzyme could thus potentially improve growth of C. ljungdahlii in gas fermentation, which may in turn contribute to enhanced ethanol production. CLJU_c22650 encodes a nickel permease which supplies the essential nickel cofactor to carbon monoxide dehydrogenase, a central enzyme in the carbon fixation pathway of C. ljungdahlii [67,68]. Its disruption is therefore likely to impair the function of nickel-dependent enzymes that catalyze key reactions in CO/CO2 metabolism. (iii) CLJU_c07720 encodes a Crp/Fnr family transcription factor known to regulate carbon flux in response to the intracellular ATP/ADP ratio [[69], [70], [71]]. We speculate that perturbing its expression could redirect carbon metabolism, thereby modulating the acetate/ethanol balance. The precise molecular mechanism underlying this regulation, however, warrants further investigation. The protein encoded by CLJU_c09160 belongs to the ABC transporter family, which is typically involved in the efflux of antimicrobial peptides and associated with quorum sensing and stress responses [72]. Under the reductive stress of gas fermentation, this transporter may fine-tune the intracellular redox state (e.g., the NADH/NAD+ ratio) by modulating membrane potential or proton motive force. It is well established that in acetogenic clostridia, elevated NADH levels drive a metabolic shift from acetate to ethanol synthesis, consistent with the phenotype observed in strain TR09160, i.e., a 57.5% increase in specific ethanol productivity and a 35.7% decrease in specific acetate productivity (Fig. 4d & e). Furthermore, this gene may indirectly regulate the expression of carbon metabolism-related genes through signaling networks, thereby reshaping metabolic flux [73,74]. For CLJU_c23840, its encoding product, ornithine/aspartate carbamoyltransferase, has been known to operate at the interface of nitrogen metabolism and pyrimidine synthesis [75]. Variations in its activity could affect intracellular levels of arginine and ornithine and modify the pool of pyrimidine nucleotides (e.g., UTP, CTP). These molecules not only participate in nitrogen metabolism but may also act as signaling factors influencing global gene regulation (e.g., the clostridial CodY regulon) [[76], [77], [78]]. The observed phenotype in strain TR23840, characterized by a significant increase in ethanol yield without affecting acetate production (Fig. 4d & e), suggests that this gene may trigger global transcriptional reprogramming, thereby directing carbon flux more efficiently toward ethanol synthesis. Taken together, the roles of these two genes in gas fermentation reveal the intricate metabolic regulatory network of C. ljungdahlii when utilizing syngas, reflecting the cellular strategy of achieving systemic adaptation under specific environmental pressures through the integration of resource allocation, energy balance, and stress defense. Moreover, to roughly evaluate the reliability of our mutant library screening result, we compared the essential genes reported for cell growth in gas-fermenting C. autoethanogenum with the depleted genes identified in our Tn-seq dataset [31]. This analysis revealed that our depleted gene set includes multiple homologs of previously reported essential genes, particularly those encoding ribosomal proteins, aminoacyl-tRNA synthetases, and enzymes involved in fructose metabolism. Genes encoding ribosomal proteins and aminoacyl-tRNA synthetases are not only indispensable for translation machinery assembly and tRNA aminoacylation, but have also been increasingly implicated in diverse cellular processes such as transcriptional regulation, DNA repair, stress response, and even immune modulation [79,80]. Their pronounced depletion in our Tn-seq dataset further supports their essentiality. Similarly, several genes encoding fructose-metabolizing enzymes were also found to be significantly depleted, suggesting they may be involved in fundamental metabolism of C. ljungdahlii, though the precise functions remain to be elucidated. A critical consideration in the use of a transposon mutant library for gene screening and enrichment is the potential loss of insertion mutants in genes essential for cell growth, as well as the impact of extended cultivation over many generations on library composition and distribution. These factors warrant careful consideration in the interpretation of our findings. As noted earlier, although the pCYG4 plasmid enabled genome-wide random mutagenesis in C. ljungdahlii, the transposon insertion sites covered only 75% of all genomic regions (Fig. 1c). Thus, the functions of the remaining genes not represented in the library could not be assessed. This limited coverage may be attributed to two primary factors: the potential essentiality of some genes, whose mutants would be non-viable, and the inherent insertion site preference of the mariner transposon. To overcome this limitation, complementary high-throughput technologies could be integrated. For instance, a CRISPR interference (CRISPRi) library could be designed to target the 25% of unsampled genomic regions, thereby enabling functional screening of these genes and their potential roles in key phenotypes of C. ljungdahlii. Conclusions This study established a transposon mutagenesis system for genome-wide random mutagenesis in the gas-fermenting bacterium C. ljungdahlii. By applying Tn-seq, we identified a set of genes associated with C. ljungdahlii fitness to syngas, several of which were further demonstrated to modulate ethanol and acetate synthesis. Our work provides a crucial methodology and foundational dataset for comprehensively understanding and rationally engineering C. ljungdahlii. Furthermore, the numerous uncharacterized genes identified herein represent promising targets for future research, potentially offering valuable insights into the unique physiology of this industrially relevant organism. CRediT authorship contribution statement Zhiyi Zhang: Writing – original draft, Visualization, Validation, Investigation, Conceptualization. Fengjuan Yang: Writing – original draft, Validation, Investigation, Conceptualization. Yihan Liu: Visualization, Data curation. Wei Ye: Validation. Dechen Jia: Validation, Investigation. Yi Tian: Validation, Investigation. Weihong Jiang: Supervision, Project administration. Qiyao Wang: Writing – review & editing, Supervision. Yang Gu: Writing – review & editing, Supervision, Project administration, Funding acquisition. Declaration of competing interest 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.