The E. coli, Staphylococcus aureus, and Serratia sp. ATCC 39006 strains and plasmids used in this study are listed in Supplementary Table 1. The primers and oligonucleotides are listed in Supplementary Table 2. All the strains and plasmids constructed in this study have been verified by sequencing. The E. coli and S. aureus strains were grown in Luria–Bertani broth (LB: yeast extract 5 g/L, tryptone 10 g/L, and NaCl 10 g/L) at 37°C. Wild-type Serratia sp. ATCC 39006 and the derived strains were grown at 30°C in LBG medium (10 g/L tryptone, 10 g/L NaCl, 5 g/L yeast extract, and 20 g/L glucose) or M9 medium [2% (w/v) glucose or another carbon source, 2 mM MgSO₄, 0.1 mM CaCl₂, 17.1 g/L Na₂HPO₄^.12H₂O, 3.0 g/L KH₂PO₄, 1.0 g/L NH₄Cl, and 0.5 g/L NaCl]. The Serratia sp. ATCC 39006 strains were cultured in 25 mL of liquid medium in 100-mL-volume shake flasks at 200 rpm or on LBG or M9 solid medium supplemented with 1.8% (w/v) agar. Appropriate antibiotics were added to the medium when necessary. The final concentrations of antibiotics are listed as follows: kanamycin (Km), 50 μg/mL; gentamicin (Gm), 15 μg/mL.

To acquire a reporter strain that can be used to detect the pig promoter-binding protein, the lacZ gene was cloned into the broad-host vector, pBBR1MCS5, resulting in pBBR1MCS5-lacZ. The pig promoter was inserted upstream of the lacZ open reading frame (ORF) to yield the reporter vector pBBR1MCS5-P_pig-lacZ. The reporter vector was then transformed into the Serratia sp. ATCC 39006 pigA-deletion mutant ΔpigA (the construction process of ΔpigA is shown below) to generate the reporter strain, ΔpigA/pBBR1MCS5-P_pig-lacZ. Subsequently, transposon mutagenesis was performed as described by Liu et al. (2017), with a slight modification. The mini-Tn5-Km-carrying suicide vector, pRL27, was conjugated to the ΔpigA/pBBR1MCS5-P_pig-lacZ strain, and the resulting transposon insertion cells were plated on an M9 plate containing 50 μg/mL kanamycin and 15 μg/mL gentamicin and incubated at 30°C. The colonies were inoculated on LB solid medium containing 15 μg/mL gentamicin and X-gal. After incubating at 30°C for 2 days, colonies showing differences in color (more blue or more white) compared to the control strain (ΔpigA/pBBR1MCS5-P_pig-lacZ) were selected to identify the transposon insertion site. The transposon insertion sites were mapped as described by Liu et al. (2017).

pMMB1, a suicide plasmid for Serratia sp. ATCC 39006 gene manipulation, was constructed by replacing the DNA replication origin of pK19mobsacB with a π protein-dependent oriR6K from pKNG101. To construct an in-frame fnr deletion mutant (Δfnr), a 951-bp upstream homologous fragment (−789 to +162 bp relative to the fnr start codon) and a 1,032-bp downstream homologous fragment (+700 to +1,731 bp relative to the fnr start codon) were amplified using primer pairs WL1003/WL1027 and WL1028/WL1006 separately. The two fragments were digested with BamHI/EcoRI and EcoRI/PstI, respectively, and ligated to BamHI/PstI-digested pMMB1 to generate an fnr in-frame deletion vector, pMMB1-Dfnr. pMMB1-Dfnr was verified by sequencing and then transferred from E. coli S17-1(λpir) to Serratia sp. ATCC 39006 by conjugation as per the following procedure: Overnight cultured S17-1(λpir)/pMMB1-Dfnr and Serratia sp. ATCC 39006 were harvested by centrifugation. The cells were washed twice with sterile LB medium, mixed and dropped onto LB plate, and cultured at 30°C for 12 h. Then, the cells from the lawn were harvested by scraping, washed twice with 10 mM MgSO₄ buffer, serially diluted, and plated on M9 solid medium containing 50 μg/ml kanamycin. After culturing at 30°C for 2–3 days, the transformants were picked and inoculated into LC medium (yeast extract, 5 g/L; tryptone, 10 g/L), cultured at 30°C and 200 rpm for 12 h, and then serially diluted and plated onto an LCS plate (LC medium supplemented with 200 g/L sucrose). The right Δfnr deletion mutant was verified by colony PCR using primer pairs WL1007 and WL1008, WL1009 and WL1010, and WL1011 and WL1012. A prodigiosin-disrupted strain, ΔpigA, was constructed by the same method using primer pairs WL1644 and WL1645, WL1646 and WL1647, WL1648 and WL1649, WL1650 and WL1651, and WL1652 and WL1653.

For complementation of fnr, primers WL1301 and WL1302 were used to amplify the DNA fragment containing the ribosome binding site, and the complete ORF of the fnr gene. Then, the fragment was digested with BamHI/EcoRI and ligated into the BamHI and EcoRI sites of pBBR1MCS2-P_aacC1 (complementary vector containing a constitutive aacC1 promoter from pUC-Gm, constructed by our laboratory), resulting in the fnr-complemented vector pBBR1MCS2-Cfnr, which was then transformed into Δfnr via conjugation to obtain the complemented strain Cfnr.

To construct the wild-type (WT)/fnr-FLAG transformant encoding a C-terminal 3 × FLAG-tagged Fnr, the upstream/downstream homologous arm flanking the 3′-terminus region of the fnr gene was amplified and cloned into pMMB1. Next, the resulting plasmid was amplified with the primers WL1966 and WL1967 to acquire fragment A, and the oligonucleotide (shown in Supplementary Table 2) encoding 3 × FLAG (DYKDDDDKGDYKDDDDKIDYKDDDDK) was used as a template and amplified by primers WL1592 and WL1963 to acquire fragment B. Finally, the resulting fragments in step two were digested with restriction enzymes, purified, and ligated to obtain the fnr C-terminal 3 × FLAG-tag knock-in plasmid, pMMB1-fnr-FlagKin. The plasmid was conjugated into WT to obtain the C-terminal 3 × FLAG-tagged fnr transformant WT/fnr-FLAG. The C-terminal 3 × FLAG-tagged pigA transformants were constructed in the same way, resulting in WT/pigA-FLAG and Δfnr/pigA-FLAG.

To substitute the constitutive promoter P_aacC1 for the native pig operon promoter P_pig, P_pig and its flanking homologous arms were first amplified using the primers WL1979 and WL1980 and subcloned into the pMMB1 vector to generate an intermediate plasmid. Reverse PCR was performed using the intermediate plasmid as a template with the primers WL1981 and WL1982 to acquire the P_pig-free part. Second, the aacC1 promoter region (P_aacC1) from the pUC-Gm plasmid was ligated to the P_pig-free part, resulting in the P_pig-replacement plasmid, pMMB1-P_aacC1-pig. The plasmid was conjugated into WT and Δfnr to replace the pig promoter with a constitutive expressing promoter, P_aacC1, resulting in strains WT/P_aacC1-pig and Δfnr/P_aacC1-pig, respectively.

Prodigiosin production and yield analysis was performed according to the procedure described by Slater et al. (2003) with slight modifications. To perform fermentation, Serratia sp. ATCC 39006 strains stored at −80°C were inoculated into LBG medium and cultured at 30°C, 200 rpm, overnight to obtain activated seed medium. Subsequently, 250 μL seed medium was inoculated into 25 mL fermentation broth (LBG) in 100 mL volume shake flasks and cultured at 30°C, 200 rpm, for 12 h. For prodigiosin yield quantification, 3 mL of fermentation broth was centrifuged at 13,000 g for 5 min to harvest cells. The cells were resuspended in 1 mL HCl-acidified methanol (pH 3.0) and placed in a dark room for 30 min to extract prodigiosin. Next, a second centrifugation was performed, and the pellet was discarded; the absorbance of the supernatant at 534 nm (A₅₃₄) was measured in a cuvette. The anaerobic fermentation procedure was modified from that of Liu et al. (2020). Briefly, Serratia sp. ATCC 39006 that had been cultured overnight was inoculated into 50 mL of LBG medium in 60-mL serum vials. After purging with nitrogen gas for 10 min, the vials were sealed to ensure anoxic conditions. All serum vials were placed at 30°C, 200 rpm, for 12 h before the analysis of growth and prodigiosin production.

Serratia sp. ATCC 39006 cells grown in fermentation medium for 12 h were collected for RNA extraction, and total RNA was extracted using the TRNzol universal total RNA extraction agent (Tiangen, China) following the protocol of the manufacturer. Reverse transcription PCR was performed to acquire cDNA, as described by Liu et al. (2017). qRT-PCR assays were performed using SYBR Green qPCR Mix (Biosharp, China), and the results were analyzed using a CFX96 Touch System (Bio-Rad, United States). The qRT-PCR condition was described by Liu et al. (2017), the primers used are listed in Supplementary Table 2, and the 16S rRNA gene was taken as an internal control. All experiments were performed at least three times.

The Serratia sp. ATCC 39006 fnr encoding sequence was amplified by the primers WL1228 and WL1229. The resulting DNA fragments were inserted into the protein expression plasmid pET-28a(+) to obtain pET28-39006-Fnr. After being screened by DNA sequencing, the vector was transformed into E. coli BL21 (DE3) to generate the BL21/pET28-39006-Fnr strain, which was used to heterologously express N-terminal His₆-tagged Fnr protein. For protein expression and purification, the heterologous protein expression strains were induced by 0.4 mM IPTG and cultured at 16°C, 180 rpm, for 12 h. The broth was centrifuged at 4°C, 11,000 g, for 10 min to harvest induced cells, and the cells were washed and resuspended in ice-cold phosphate-buffered saline (PBS) (pH 7.6) buffer containing 10 mM DTT and protease inhibitor (CoWin Biosciences). The cells were then broken in a sonicator (JY88-IIN, Ningbo Scientz Biotechnology Co., Ltd.) on ice, and low-temperature centrifugation was applied to remove the pellet. The supernatant-containing soluble His₆-39006 Fnr protein was purified by affinity chromatography on Ni-agarose resin (CoWin Biosciences). After washing with washing buffer (ice-cold PBS buffer containing 50 mM imidazole and 10 mM DTT), the target protein was eluted from the resin with elution buffer (ice-cold PBS buffer containing 250 mM imidazole and 10 mM DTT). After verification by SDS-PAGE, the protein was dialyzed in dialysis buffer [20 mM HEPES, pH 7.6, 1 mM EDTA, 10 mM (NH₄)₂SO₄, 10 mM DTT, 30 mM KCl, 0.2% (w/v) Tween-20, and 20% (v/v) glycerol] at 4°C. The concentration of the dialyzed protein was measured using Bradford assay, and the protein was subpackaged and preserved at −80°C.

Electrophoretic mobility shift assays (EMSAs) were performed as described by Sun et al. (2016). The primers used to amplify the EMSA probes are listed in Supplementary Table 2. The mutated pig promoter probe was amplified using primers WL1265 and 1266 (listed in Supplementary Table 2), and the plasmid pMV-P2M was used as template.

5′-Rapid amplification of cDNA ends (5′-RACE) assays was carried out as described by Sun et al. (2016). The primers used for reverse-transcribed target genes and to identify the transcriptional start site of the target genes are listed in Supplementary Table 2.

To perform the chromatin immunoprecipitation assay, Serratia sp. ATCC 39006 cells were obtained after fermentation for 12 h, and the following steps were performed as described by Zhu et al. (2016), with minor modifications. In brief, Serratia sp. ATCC 39006 cells were cross-linked by 1% formaldehyde at 30°C, 150 rpm, for 25 min, and the cross-linking was terminated by the addition of 0.1 M glycine. After washing twice with PBS, the cells were suspended in lysis buffer (50 mM HEPES-Na, pH 8.0, 137 mM NaCl, 1 mM EDTA, and 10 mM Tris–HCl, pH 8.0) and sonicated to shear the genomic DNA to 250–500-bp fragments. The sonicated mixture was centrifuged at 4°C, 13,000 g, for 10 min to collect the supernatant. The protein concentration of the supernatant was measured, and lysis buffer was added to 4 mg total protein to a final sample volume of 1 ml. Next, 10 μL of the sample was frozen at −80°C as the input. The rest of the sample was added to 40 μL anti-FLAG affinity gel (Bimake), which was pre-balanced with bovine serum albumin and salmon sperm DNA, and incubated at 4°C for 4 h, with gentle rotation. The gel was then washed, and the protein–nucleic acid complex was eluted. The input and eluate were reverse cross-linked, and the DNA fragments were extracted. DNA from the inputs and samples was dissolved in ddH₂O for the qPCR analysis. The primers used for the qPCR analysis of the immunoprecipitated DNA are listed in Supplementary Table 2.

Serratia sp. ATCC 39006 cells grown in liquid medium were harvested and disrupted in a sonicator. After centrifuging at 4°C, 13,000 g, for 10 min, the supernatant was collected to measure the protein concentration. The total protein of each sample was adjusted to equal amounts to perform western blotting assays. The western blotting assay was performed as described by Yan et al. (2020), using anti-FLAG-tag mouse monoclonal antibody (CoWin Biosciences) as the primary antibody and HRP-conjugated goat anti-mouse IgG (CoWin Biosciences) as the secondary antibody. An eECL western blot kit (CoWin Biosciences) was used for chemiluminescence detection of the target band.

Motility bioassays were performed as described by Fineran et al. (2005b) and Pan et al. (2021), with slight modifications. Overnight cultured Serratia sp. ATCC 39006 strains were adjusted to an OD₆₀₀ of 1.0, and 3 μL of medium was spotted onto an LBG plate containing 0.3% agar (for swimming test) or 0.6% agar (for swarming test). The plates were cultured at 30°C for 24 h before examining the halo sizes. An Oxford cup assay was used to determine the bacteriostatic activity, as described by Slater et al. (2003), with slight modification. Carbapenem-sensitive S. aureus cells were overlayed on LB medium containing 0.6% agar. Overnight cultured Serratia sp. ATCC 39006 medium was adjusted to an OD₆₀₀ of 1.0 and centrifuged. Next, 100 μL filtered supernatant was injected into the Oxford cup, and the plates were incubated at 30°C for 24 h. The diameter of the inhibition zone around the Oxford cup was measured to indicate the anti-bacterial activity. Secreted cellulase and pectinase activities were assessed as described by Liu et al. (2019), with a minor modification. Serratia sp. ATCC 39006 strains, adjusted to the same OD₆₀₀ value, were dropped onto M9 plates supplemented with different carbon sources and placed at 30°C for 48 h before the activity test. For the cellulase activity test, 0.5% (w/v) carboxymethylcellulose sodium was supplied. The plate was stained for 30 min using 0.3% Congo red solution, followed by destaining for 30 min with 0.9% NaCl solution. For the pectinase activity test, 0.5% (w/v) pectin was added to the medium. The plate was stained for 5 min with iodine solution containing 3.3 g/L iodine and 6.6 g/L KI, and the plate was washed with water for 1 min. The exoenzyme activity was assessed by measuring the diameter of the clear zone.

We performed transposon mutagenesis analysis to identify the direct transcriptional regulator of the pig operon. First, a pigA in-frame deletion mutant harboring a pig promoter-lacZ reporter vector (pBBR1MCS5-P_pig-lacZ) was constructed to eliminate the interference of the red prodigiosin. Second, transposon insertion mutation was performed, and the transformants were screened on X-gal-containing plates for significantly altered X-gal hydrolysis ability by observing colony color changes. Approximately 5,000 mutants were screened, seven of which had deeper blue colonies and were selected for mapping the transposon insertion site. Two of the insertion mutants with higher X-gal hydrolysis ability (bluer colony) were inserted into two loci in the open reading frame of the ser39006_013370 gene (Supplementary Figure 1A). This gene was annotated as the Crp/Fnr family transcriptional regulator Fnr, and the predicted amino acid sequence showed a high similarity to E. coli Fnr (identity = 94.40%) (Supplementary Figure 1B).

The fnr in-frame deletion mutant Δfnr and the complementation strain Cfnr were constructed to identify whether Fnr affected prodigiosin biosynthesis. The plate assay showed that the Δfnr colony was redder than the colony of the WT strain (Figure 1A). Moreover, a higher production of prodigiosin of Δfnr was observed in the shake-flask fermentation assay, in which the prodigiosin yield of Δfnr was 46.2% higher than that of WT (Figure 1B). For both plate assay and the liquid fermentation assays, prodigiosin production was restored in the complementary strain Cfnr (Figures 1A,B). Although the cell growth in Δfnr was slightly enhanced compared to that in WT (Figure 1C), the prodigiosin yield per cell unit (A₅₃₄/OD₆₀₀) for Δfnr was significantly higher than that for WT (Figure 1D), indicating that the elevation of prodigiosin yield in Δfnr was not due to the change in biomass. Thus, Fnr negatively affects prodigiosin synthesis in Serratia sp. ATCC 39006.

Fnr is well recognized as a global transcriptional regulator that is activated under anaerobic conditions, and prodigiosin itself is not made anaerobically (Heinemann et al., 1970). Therefore, we considered whether Fnr is the repressor responsible for inhibiting prodigiosin synthesis under anaerobic conditions. Anaerobic fermentation was performed in the serum vial purged with nitrogen, and all of the examined strains were found to be able to grow under anaerobic conditions (Figure 1C). Similar to WT, no prodigiosin production was detected in Δfnr (Figures 1A,B,D), indicating that prodigiosin synthesis in an anaerobic environment was not suppressed by Fnr. Thus, this report is the first to demonstrate that the deficiency of prodigiosin under anaerobic conditions in Serratia sp. ATCC 39006 was not due to Fnr repression and that Fnr was involved in the regulation of prodigiosin synthesis under aerobic conditions.

As Fnr is a transcriptional regulator, it is possible that Fnr represses prodigiosin production by regulating the transcription of the prodigiosin biosynthetic gene cluster. The total RNA of WT, Δfnr, and Cfnr cells grown in the aerobic fermentation broth was isolated for qRT-PCR, and the transcription levels of 15 pig genes arranged in the prodigiosin synthesis operon were measured. The transcription levels of pig genes in Δfnr were higher than those in WT, and the transcription was restored in the Cfnr strain (Figure 2). Meanwhile, the transcription level of the housekeeping gene, recA, showed no significant difference in WT, Δfnr, and Cfnr (Supplementary Figure 2), indicating that the effect of Fnr on the pig operon was specific. These results show that the transcription profiles of pig genes in WT, Δfnr, and Cfnr strains were consistent with the prodigiosin yield (Figure 1C), indicating that Fnr represses prodigiosin production via negatively regulating the transcription of prodigiosin biosynthetic genes.

To confirm this result, a 3 × FLAG tag was knocked into the C-terminus of PigA in WT and Δfnr, resulting in WT/pigA-FLAG and Δfnr/pigA-FLAG, respectively. Western blotting assay was performed to compare the expression of labeled PigA in these two strains. The results showed that the level of PigA in Δfnr/pigA-FLAG was higher than that in WT/pigA-FLAG (Figure 3A), which is coincident with the transcriptional analysis (Figure 2), indicating that Fnr regulates pig gene expression at the transcription level. We also found that, when the native pig promoter was replaced by a constitutive promoter P_aacC1 in WT and Δfnr, the prodigiosin yield and the transcription of pigA in the resulting strains, WT/P_aacC1-pig and Δfnr/P_aacC1-pig, were shown to be not significantly different (Figures 3B,C), suggesting that Fnr regulates pig gene expression by binding to the pig promoter in vivo. To verify this hypothesis, a 3 × FLAG tag was knocked into the C-terminus of Fnr, resulting in the strain WT/fnr-FLAG, and ChIP-qPCR was performed. Growth and prodigiosin production in WT/fnr-FLAG were not significantly different compared to WT, indicating that the knock-in tag does not affect the physiological function of Fnr (Figure 3D). An anti-FLAG monoclonal antibody was used to monitor the binding of FLAG-labeled Fnr to the pigA promoter region, and the WT strain was used as the negative control. The enrichment level of Fnr on the pigA promoter in WT/fnr-FLAG was 87-fold higher than that in the negative control (Figure 3E), indicating that Fnr directly binds to the pig promoter in vivo. For further confirmation of the binding of Fnr to the pigA promoter, His₆-tagged Fnr was heterologously expressed and purified for an in vitro gel shift assay. EMSA experiments found that the recombinant His₆-Fnr bound to the labeled promoter region of pigA (Probe pigA) specifically in vitro, indicating that Fnr represses pig gene transcription directly (Figure 3F). According to the abovementioned results, Fnr represses prodigiosin biosynthesis by directly binding to the promoter region of the pig operon and negatively regulates the transcription of pig genes.

To clarify the underlying regulatory mechanism of Fnr in prodigiosin synthesis, 5′-RACE was performed to identify the transcription start site (TSS) of pigA. The TSS was located at a guanosine 44 nt upstream of the translation start codon, which is coincident with a former study (Slater et al., 2003; Figure 4A). The −10 region and −35 region were also predicted based on the TSS. As the amino acid sequence of Serratia sp. ATCC 39006 Fnr is highly similar to that of E. coli Fnr (identity = 94.40%) (Supplementary Figure 1B), the conservative binding sequence of E. coli Fnr (TTGAT-N₄-ATCAA) was found in the spacer between the −10 region and −35 region (Figure 4A; Kovacs et al., 2005). To verify whether this sequence is the Fnr binding site, the predicted binding site was mutated (Figure 4B), and the binding activity of Fnr to the probe containing the mutated sequence (Probe pigA-M) was evaluated by EMSA. The result shows that Fnr was unable to bind to the mutant probe (Figure 4B), and the Fnr binding site in the pigA promoter was identified as the region from 17 to 30 nt upstream of the pigA TSS (Figure 4A), indicating that Fnr represses pig gene transcription by binding to the spacer between the −10 region and −35 region.

In the abovementioned experiments, we have revealed that Fnr represses prodigiosin production by reducing the transcription of pig genes. As the Fnr homolog in other bacteria has been reported to be a global regulator modulating various physiological processes, we investigated whether Fnr affects other metabolisms beyond prodigiosin synthesis in Serratia sp. ATCC 39006 by plate assay. The swimming ability was reduced in Δfnr but restored in Cfnr (Figure 5A), indicating that Fnr promotes Serratia sp. ATCC 39006 swimming. Given that FlhD (SER39006_RS11350) and FlhC (SER39006_RS11355) were necessary for the motility of Serratia sp. ATCC 39006 (Hampton et al., 2016), the transcription levels of flhD and flhC in WT, Δfnr, and Cfnr were analyzed. The results showed that the transcription levels of both genes were decreased in Δfnr (Figure 5A), suggesting that Fnr may promote Serratia sp. ATCC 39006 swimming ability by increasing the transcription of flhD and flhC. Besides this, Fnr negatively controlled the bacteriostatic activity of Serratia sp. ATCC 39006 (Figure 5B). Transcription analysis showed that carR (ser39006_RS009915), which encodes the pathway-specific activator of the bacteriostatic carbapenem (McGowan et al., 1995), was elevated in Δfnr (Figure 5B), suggesting that Fnr negatively controls bacteriostatic activity through carR. The results of ChIP-qPCR showed that the enrichment of Fnr on the flhDC promoter or carR promoter was higher (25- and 7.6-fold, respectively) than that in the negative control (Figure 5C), indicating that Fnr regulates the motility and synthesis of carbapenem at the transcriptional level by directly binding to the promoter regions of pathway-specific regulators. Therefore, Fnr acts as a global regulator to control multiple physiological processes in Serratia sp. ATCC 39006.