The strains and plasmids used in this work are presented in Supplementary Table S1. Paenibacillus polymyxa SC2-M1 and its mutants, Escherichia coli DH5α, E. coli BL21, and Trans 110 were cultured in Luria-Bertani (LB) broth at 37°C. When plasmids were presented in cells, antibiotics were added to LB medium as follows: chloramphenicol (6 μg/ml for strain SC2-M1 and its mutants), kanamycin (50 μg/ml), tetracycline (20 μg/ml), ampicillin (100 μg/ml), and erythromycin (12.5 μg/ml). To measure carbon source utilization by the strains, basal medium (MgSO₄ 0.02%, NaCl 0.02%, (NH₄)₂SO₄ 0.8%, K₂HPO₄ 0.368%, KH₂PO₄ 0.132%) with a 1% sole carbon source was used.

Competent cells of E. coli were prepared using the transformation and storage solution (TSS) method and transformed by the heat pulse method (Chung et al., 1989). Competent cells were prepared and P. polymyxa SC2-M1 were electrotransformed as previously described (Hou et al., 2016).

To construct knockout SC2-M1 strain mutants, 650–1,000 bp of homologous arm-flanking target genes were cloned from genomic DNA using PCR. The cat gene, encoding the chloramphenicol resistance cassette, was cloned from pDG1661 (Guérout-Fleury et al., 1996) using PCR to replace the target knockout genes. Two homologous arms and cat cassettes were fused by fusion PCR (Cha-Aim et al., 2012), and the fusion DNA fragments were ligated into the thermo-sensitive elimination plasmid pRN5101 to construct knockout vectors. The plasmid was demethylated by being further transformed into Trans 110. The demethylated knockout vectors were transferred into the SC2-M1 strain. Double-crossover recombinants were detected as previously reported (Hou et al., 2016), and the knockout strains were confirmed by PCR using the corresponding primers (Supplementary Table S2). For in situ complementation of the knockout genes, 650–1,000 bp of homologous arms fused with the corresponding complete target genes were cloned by PCR from genomic DNA (not involved in the resistance cassette), and ligated into pRN5101 to construct the complement vectors. Demethylated complementary vectors were introduced into the mutant strains. Double-crossover recombinants were detected as described above, and the complemented strains were confirmed by PCR using the corresponding primers (Supplementary Table S2). For ChIP-seq analysis, we obtained a Flag-tagged strain using homologous recombination. The Flag-tag was added in the front of termination codon of msmR1 and a chloramphenicol resistance cassette was also added after the Flag tag using overlapping PCR. The homologous arms (650–1,000 bp) were cloned by PCR amplification, and ligated to pRN5101 to construct the plasmid pRN5101-msmR1-Flag. Demethylated pRN5101-msmR1-Flag were introduced into SC2-M1 and double-crossover recombinants were detected as described above, designated as strain SC2-M1-Flag. For overexpression of target genes, fragments carrying complete target genes were cloned from genomic DNA, fused with the promoter fragments, and ligated into pHY300PLK between BglII and SalI sites by ClonExpress MultiS One Step Cloning Kit (Vazyme Biotech, Nanjing, China) for overexpression in SC2-M1. To evaluate the promoter activity of target DNA fragments, the fragments were cloned from genomic DNA, and the promoterless green fluorescent protein (gfp) fragment from pGFP4412 was cloned (Li et al., 2019). The two fragments were fused using PCR and ligated into pHY300PLK between BglII and SalI sites as described above, thus generating the pHY300PLK-promoter-gfp plasmid.

The SC2-M1 and SC2-M1-Flag strains were cultured in 50 ml of LB medium at 37°C with shaking at 180 rpm for 10 h, inoculated with an initial OD₆₀₀ of 0.8 in 50 ml of LB broth by 2% (v/v), and cultured at 37°C with shaking at 180 rpm for 7 h. The two strains were then fixed with 1% formaldehyde for 10 min at 28°C and the reaction was stopped with 125 mM glycine for 5 min. The cells of the two strains were washed thrice with phosphate buffer solution (PBS) at 8000 rpm for 5 min at 4°C and stored at-80°C. ChIP was performed by Wuhan IGENEBOOK Biotechnology Co. Ltd., China. The manufacturer’s instructions of Illumina TruSeq ChIP Sample Prep Set A were followed to construct sequencing libraries using chromatin immunoprecipitated DNA which was then sequenced on an Illumina Xten using the PE 150 method (Chen et al., 2021). The obtained reads were filtered by Trimmomatic (v 0.30). Clean reads were then mapped to the P. polymyxa SC2 genome using Bwa (v 0.7.15), which allowed for up to two mismatches (Pjanic, 2017). MsmR1 peaks were produced and visualized using MACS software and Integrative Genomics Viewer (IGV, v 2.3.91), respectively (Zhang et al., 2008; Thorvaldsdóttir et al., 2013). To understand the potential roles of genes, Kyoto Encyclopedia of Genes and Genomes (KEGG) and multiple EM for motif elicitation (MEME) were used for pathway enrichment analysis and motif analysis, respectively (Chen et al., 2021; Wang et al., 2022).

SteadyPure Universal RNA Extraction Kit (Accurate Biology, Hunan, China) was used to purify the total RNA from the target bacteria. cDNA was obtained from 1 μg of total RNA using the Evo M-MLV RT Kit with gDNA clean for qPCR II (Accurate Biology, Hunan, China). qRT-PCR analysis was performed on a Roter-Gene Q PCR system using the SYBR Green Pro Taq HS qPCR Kit (Accurate Biology, Hunan, China). The relative expression of target genes was calculated using the 2^−ΔΔCt method (Liu et al., 2021). The housekeeping gene glyceraldehyde-3-phosphate dehydrogenase (gapdh) was used as an internal control.

An 828 bp fragment (275 amino acids) of msmR1 was cloned from the genomic DNA of strain SC2-M1 using primers MsmR1F and MsmR1R (Supplementary Table S2). The fragments of msmR1 and plasmid pET-28b(+) were simultaneously cleaved with XhoI and SalI, purified, ligated using T4 DNA ligase, and transferred into E. coli DH5α to obtain pET28b(+)-msmR1.

To overexpress the MsmR1 protein, the plasmid pET28b(+)-msmR1 was transferred into E. coli BL21. The transformants were cultured in kanamycin-resistant LB broth at 37°C with shaking at 180 rpm for 10 h. Next, the cells were inoculated into 50 ml of kanamycin-resistant LB broth at 2% (v/v) and cultured at 37°C with shaking at 180 rpm for approximately 1.2 h until the OD₆₀₀ reached 0.5–0.6. Then the cells were induced by 0.4 mM IPTG at 16°C for 20 h, harvested, washed, resuspended in PBS, and sonicated on ice. MsmR1-His6 protein was purified using BeyoGold™ His-tag Purification Resin (Beyotime, China) following the manufacturer’s instructions. The purified protein was added to a final concentration of 10% (v/v) glycerol and stored at-80°C. EMSA was performed using a chemiluminescent EMSA kit (Beyotime, China; Zhang et al., 2021). The concentration of labeled probes and competitive probes were about 0.4 and 4 μM, respectively. The gradient concentration of protein were set up for the EMSA experiment.

Promoter reporter plasmids were transformed into E. coli DH5α cells. The strains were inoculated in LB plate containing 20 μg/ml tetracycline and fluorescence was observed with an OLYMPUS fluorescence microscope at 1000× (Li et al., 2019). The fluorescence intensity of the promoters was detected at different time points as previously reported (Li et al., 2019).

An Oxford cup assay was used to evaluate polymyxin production by the SC2-M1 strain, as previously described (Liu et al., 2021). Bacteria were inoculated in liquid LB medium and cultured at 37°C with shaking at 180 rpm for 10 h. Bacteria were then inoculated with an initial OD₆₀₀ of 0.8 in 50 ml fermentation medium (sucrose 40 g/l, (NH₄)₂SO₄ 8 g/l, CaCO₃ 5 g/l, KH₂PO₄ 0.2 g/l, MgSO₄ 0.2 g/l, NaCl 0.2 g/l) at 2% (v/v) and cultured at 37°C with shaking at 180 rpm for 48 h.

A motility assay was performed as previously reported (Guo et al., 2019). Bacteria were cultured on LB medium for approximately 12 h, after which the OD₆₀₀ was adjusted to 0.6. Diluted cell solutions (2 μl) were added to the center of the plate containing 0.5% glucose, 0.22% NaCl, 0.13% yeast extract, 0.45% tryptone, and 0.5% agar.

Bacteria were cultured overnight in liquid LB medium and adjusted to an OD₆₀₀ of 0.8. Diluted strain solutions were inoculated at 2% in 50 ml of fresh LB medium. The bacteria were cultured at 37°C with shaking at 180 rpm. The OD₆₀₀ and number of live bacterial cells were measured every 3 h. A total of 100 μl of the strain solution was serially diluted with 900 μl of sterile water; 100 μl of the dilution was then spread on an LB plate. The plates were subsequently incubated for 24 h at 37°C.

The growth curves of bacteria on the carbon sources were determined using a basal medium. Bacteria were cultured overnight in LB medium. The cultures were centrifuged, washed thrice with basal medium, and resuspended to inoculate 50 ml of fresh basal medium with 1% sole carbon source; the original OD₆₀₀ value was approximately 0.1. Cultures were incubated at 37°C with shaking at 180 rpm, and growth curves were determined according to OD₆₀₀ every 3 h.

For data analysis, the analysis of variance (ANOVA) and Least Significant Difference (LSD) test (p ≤ 0.05) were performed using statistical software SPSS (v 19.0, SPSS Inc., Chicago, IL, USA; Wang et al., 2020).

The MsmR1 of strain SC2, which is classified as an AraC/XylS family transcriptional regulator, contains two domains: an N-terminal heterologous dimerization domain and a C-terminal DNA-binding domain (Figure 1A). The C-terminus of MsmR1 shows high similarities to that of AraC family members: RhaR, AraC, UreR, MelR, MsmR, XylS, and Toxt. Seven α-helix were identified in the C-terminal domain of MsmR1, which combined to form two helix-turn-helix (HTH) motifs (Figure 1A).

AraC family TFs are often auto-regulated (Sun et al., 2016). Hence, the promoter activity of msmR1 was evaluated using the pHY300PLK-msmR1-gfp plasmid. This plasmid was found to possess promoter activity in E. coli DH5α cells. In contrast, the control (pHY300PLK-gfp) showed no fluorescence (Supplementary Figure S2A). The pHY300PLK-msmR1-gfp plasmid was introduced into the strains SC2-M1 and SC2-M1-ΔmsmR1 which was a msmR1 knock out strain obtained via homologous recombination (Supplementary Figure S2B). The fluorescence intensity of the strain SC2-M1-ΔmsmR1 was significantly higher than that of the SC2-M1 strain (Figure 1B), revealing that MsmR1 could auto-regulate its transcription. EMSA was further performed to determine whether MsmR1 is capable of direct auto-regulation. MsmR1 was found to specifically bind to its own promoter resulting in a retarded band (Figure 1C), indicating that MsmR1 auto-regulates its expression.

To investigate the function of MsmR1, msmR1 knock out strain SC2-M1-ΔmsmR1 and overexpression strain SC2-M1-pmsmR1 formed by ligation into pHY300PLK plasmid (Supplementary Figure S2C), were used. Meanwhile, SC2-M1-p was set as the control. Notably, the amount of polymyxin in the mutant SC2-M1-ΔmsmR1 was significantly less than that in the SC2-M1 strain; the diameter of the inhibitory zone was decreased by 9.0% (Figure 1D; Supplementary Figure S2D). The qRT-PCR results revealed that the polymyxin synthetase genes were downregulated in the knockout strain compared with SC2-M1 (Figure 1E). In SC2-M1-pmsmR1, the inhibitory zone diameter was increased by 10.03% compared with SC2-M1-p (Figure 1D; Supplementary Figure S2D). The results were consistent with the qRT-PCR results which presented a significant increase in expression of polymyxin synthetase genes in SC2-M1-pmsmR1 when compared with that in SC2-M1-p (Figure 1F).

Given that AraC family members are involved in carbohydrate metabolism (Cortés-Avalos et al., 2021), the growth curves of the SC2-M1 and SC2-M1-ΔmsmR1 strains were determined using LB and basal media with sole carbon sources. Although phenotypic differences were noted between the two strains using glucose, xylose, galactose, fructose, and lactose, the differences were not significant (Supplementary Figure S2E).

To determine the target genes directly regulated by MsmR1 in P. polymyxa SC2, ChIP-seq experiments were performed using C-terminal FLAG-tagged msmR1 in SC2-M1. The qRT-PCR results demonstrated that the expression of msmR1 was high during the early stage and low during the decline phase (Supplementary Figure S3A). Based on the qRT-PCR results and the growth curve, a culture time of 7 h (mid-logarithmic phase) was selected for subsequent ChIP-seq analysis (Supplementary Figure S3).

Approximately 3.83–4.86 million reads in the ChIP-seq results were derived from the six samples, which were mapped to the P. polymyxa SC2 genome (Supplementary Table S3). Overall, 649 peaks were identified between the two strains, SC2-M1 and SC2-M1-Flag. Among these, 96.15% were located in the coding sequence (CDS) and 3.85% were located in the intergenic regions. To categorize the functions of MsmR1 target genes, KEGG enrichment analysis was performed. The target genes were enriched in various functions, including cellular processes (1.56%), environmental information processing (6.25%), genetic information processing (10.94%), and metabolism (81.25%; Supplementary Table S4). In the metabolism category, carbohydrate metabolism was the most enriched pathway, while amino acid metabolism, metabolism of cofactors and vitamins, lipid metabolism, energy metabolism, and nucleotide metabolism were also significantly enriched (Figure 2). Hence, MsmR1 primarily regulates basal metabolism in SC2 cells. According to the predicted results of MEME, we selected the fragments related to primary metabolism for further verification (Supplementary Table S5).

EMSA was performed to further validate the potential direct binding fragments with MsmR1. As shown in Figures 3A–K, MsmR1 could bind to the upstream fragments of oppC3, sucA, sdr3, yycN, pepF, pppL, ydfp, PPSC2_23180, gabD, araA, and bglX3. The qRT-PCR results revealed that compared with strain SC2-M1, the relative expression of genes oppC3, sucA, sdr3, yycN, pepF, pppL, ydfp, and PPSC2_23180 were significantly decreased in msmR1 mutant (p < 0.05, Figure 3L), which suggested that MsmR1 could positively regulate these genes. While no significant difference was observed in the expression of gabD, araA, and bglX3 between SC2-M1 and SC2-M1-ΔmsmR1 (p < 0.05, Figure 3L). These results indicated that MsmR1 might not regulate the expression of gabD, araA, and bglX3, even though MsmR1 could bind to them. Additionally, we also determined two MsmR1-binding motif by MEME (Figure 3M).

As described previously, carbohydrate and amino acid could affect the biosynthesis of polymyxin (Yu et al., 2018). The oppC3 gene encodes a peptide ABC transporter together with the OppB protein forming a transmembrane channel that can transport and release peptides and nickel into cells (Ge et al., 2018; Slamti and Lereclus, 2019). The gene sdr3 whose predicted product belongs to the short-chain dehydrogenase/reductase (SDR) family is important for carbohydrate, amino acid, lipid metabolism, and redox sensor mechanisms (Kavanagh et al., 2008). oppC3 and sdr3 affected the growth state of strain SC2-M1 in LB medium (Figures 4A,B). Therefore, we determined the expression levels of primary metabolism-related genes, such as those involved in carbohydrate, amino acid, and fatty acid metabolism (Supplementary Table S2), and all of these genes were downregulated in the mutant strains compared with SC2-M1 (Figures 4C,D). The growth state of SC2-M1-ΔoppC3 and SC2-M1-Δsdr3 in basal medium with the sole carbon source supported their roles in carbohydrate metabolism (Supplementary Figures S4, S5). In the present study, polymyxin synthesis was significantly reduced in the knockout strains SC2-M1-ΔoppC3 and SC2-M1-Δsdr3 compared with SC2-M1, and was recovered in the complement/overexpression strain (Figures 4E,F). Moreover, the qRT-PCR results confirmed that polymyxin synthetase genes were downregulated in the knockout strain compared with SC2-M1 (Figures 4C,D). The oppC3 and sdr3 genes indirectly influence polymyxin synthesis by reducing the rate of primary metabolism and nutrient uptake, and further reduce precursors and energy for polymyxin synthesis.

The sucA gene encodes 2-oxoglutarate dehydrogenase (E1), part of the 2-oxoglutarate dehydrogenase complex (Gayán et al., 2019). In the TCA cycle, α-ketoglutarate is converted to succinate, which requires the catalytic action of the 2-oxoglutarate dehydrogenase complex (sucAB) and succinate dehydrogenase (sucCD; Walshaw et al., 1997). To assess the function of sucA in SC2, a knockout mutant strain SC2-M1-ΔsucA was generated. The colony morphology on LB plates appeared flat and transparent following sucA knockout, whereas the original morphology was recovered in the complemented strain SC2-M1-ΔsucA/sucA (Figure 5A).

The growth curve in LB medium was determined to evaluate the growth status, and showed that growth of the mutant strain SC2-M1-ΔsucA was markedly reduced compared to SC2-M1 and SC2-M1-ΔsucA/sucA (Figure 5B). Simultaneously, live bacterial counts were determined every 3 h; those of SC2-M1-ΔsucA were greatly reduced compared with those of SC2-M1 and SC2-M1-ΔsucA/sucA, which was consistent with the OD₆₀₀ results (Figure 5C). Moreover, expression of the related gene sucB was markedly downregulated, while other genes involved in the TCA cycle, such as sucC, sucD, citC, lpdV, and pdhC (Supplementary Table S2) were markedly upregulated (Figure 5D). Due to disruption of the TCA cycle, bacteria may metabolize carbohydrate sources through the pentose phosphate pathway and glyoxylate cycle. Therefore, the energy decreased and the bacterial mass was further reduced.

During carbon source (xylose, arabinose, glucose, galactose, fructose, mannose, sucrose, maltose, trehalose, lactose and raffinose) utilization, the SC2-M1-ΔsucA strain presented a longer lag phase compared with the SC2-M1 strain. For the utilization of xylose, galactose, fructose, lactose, and raffinose, SC2-M1-ΔsucA presented a markedly lower OD₆₀₀ value during the stationary phase compared with SC2-M1 (Supplementary Figure S6). Hence, sucA have essential roles in carbon source utilization.

The pepF gene encodes an oligoendopeptidase F. In Bacillus subtilis, the pepF homolog phrA can regulate spore formation (Kanamaru et al., 2002), and in Aeromonas hydrophila, pepF can regulate biofilm formation (Du et al., 2016). qRT-PCR results revealed that pepF regulated spore and biofilm formation. In other words, the relative expressions of genes related to biofilm and spore formation were downregulated (Figure 6A). The relative expressions of these genes were also downregulated in strain SC2-M1-ΔmsmR1 compared with SC2-M1 (Supplementary Figure S7). Overall, MsmR1 could positively regulate spore and biofilm formation.

Chemotaxis enables microbes to sense environmental signals and move to more favorable environments, which is important for bacterial colonization (Ud-Din and Roujeinikova, 2017). We found that motility was lost following sdr3 deletion and was increased in the SC2-M1-psdr3 strain compared with that in the SC2-M1-p strain (Figure 6B). Moreover, the relative expression levels of genes encoding chemotaxis and flagellum (Supplementary Table S2) were downregulated in the SC2-M1-Δsdr3 strain compared with those in the SC2-M1 strain (Figure 6C). Taken together, these results indicate that MsmR1 may positively regulate cell motility by directly binding to the sdr3 promoter.

The yycN gene encodes GNAT family acetyltransferase. The ydfp gene acts as a DoxX family protein, combined with SodA and SseA, forming a membrane-associated oxidoreductase complex (Nambi et al., 2015). The pppL gene encodes a PPM family protein phosphatase, which participates in environmental stresses (Shi et al., 1998). Hence, MsmR1 in P. polymyxa SC2 may also be involved in other biological processes.