Coordination of cyclic di-GMP and 4-hydroxybenzoic acid in regulating antifungal antibiotic biosynthesis in Lysobacter enzymogenes

Small molecules are able to regulate numerous cellular processes through binding to various bacterial receptor proteins, but the mechanisms and functions by which these chemicals coordinate and execute remain poorly understood. 4-hydroxybenzoic acid (4-HBA) and cyclic di-GMP (c-di-GMP) are two such molecules with distinct structures that are produced in Lysobacter enzymogenes to synergistically affect the secretion of an antifungal antibiotic, known as heat-stable antifungal factor (HSAF). In our earlier studies, we showed that CdgL, a YajQ-like protein without DNA-binding domain, was able to physically interact with LysR, a transcription factor, to enhance its binding affinity toward the upstream region of the HSAF biosynthesis operon promoter, hence increasing the HSAF biosynthesis. Interestingly, 4-HBA or c-di-GMP can bind to its cognate receptor of LysR or CdgL, respectively, to regulate the HSAF biosynthesis. Further, c-di-GMP acts by binding to CdgL to induce the dissociation of the CdgL-LysR complex, leading to decreased downstream expression. We now showed that CdgL controlled the transcription of lenB2, which encodes an oxygenase to convert chorismate to 4-HBA. Notably, overexpression of cdgL was found to stimulate lenB2 transcription, which likely increased the intracellular 4-HBA content. Also, 4-HBA could bind to LysR to interrupt the LysR-CdgL complex formation and release of CdgL, which caused a lower affinity of LysR toward DNA and hence decreased HSAF operon expression. These findings, along with our earlier report, allow us to propose a coordination mechanism demonstrating how the HSAF biosynthesis is co-regulated by 4-HBA and c-di-GMP through interactions with their cognate receptors. This new mechanism shall shed light on improving the HSAF yield for practical usage.

Bacteria of the genus Lysobacter is gradually recognized as a rich source of crop protecting agents, due to their great ability to produce abundant extracellular lytic enzymes and antimicrobial secondary metabolites (Folman et al. 2003;Kobayashi et al. 2005;Qian et al. 2009;Xie et al. 2012). L. enzymogene OH11 is one of such plant-associated soil proteobacteria. It can secrete a diffusible antifungal antibiotic, known as heat-stable antifungal factor (HSAF), into the surrounding environment to kill neighbor fungal pathogens through targeting their sphingolipids biosynthesis pathway (Yu et al. 2007;Li et al. 2008;Zhao et al. 2017Zhao et al. , 2019. The biosynthesis of HSAF in L. enzymogenes is controlled by a gene cluster known as HSAF biosynthesis gene operon (Lou et al. 2011;Wang et al. 2017), whose transcription is influenced by multiple factors, including small molecules and transcription factors (Qian et al. 2013; Wang et al. 2014;Chen et al. 2017;Su et al. 2017Su et al. , 2018. 4-hydroxybenzoic acid (4-HBA) is one of such small molecules that is generated from a conserved shikimate pathway in Xanthomonas campestris and L. enzymogenes (Zhou et al. 2013;Su et al. 2017). Previously, we showed that the L. enzymogenes LenB2, a petridinedependent dioxygenase-like protein, can convert chorismate, the end-product of the shikimate pathway, to 4-HBA . Mutation of lenB2 lowers the production of 4-HBA, leading to blockage of HSAF production . Exogenous supplement of the physiological amount of 4-HBA into the lenB2 mutant culture can rescue the HSAF production back to the wild type level . The regulatory pathway of 4-HBA involves LysR, a transcription factor that directly binds to the upstream region of the HSAF biosynthesis operon promoter (pHSAF), to activate the operon transcription . Indeed, the direct 4-HBA-LysR binding (K d , 5 μM) was confirmed by an in vitro assay . To date, however, it remains unclear whether the binding of 4-HBA with LysR could strengthen the transcription of the HSAF biosynthesis operon under the in vivo condition or not.
Besides 4-HBA, we found cyclic di-GMP (c-di-GMP), another small molecule, is also engaged in regulating the HSAF biosynthesis Xu et al. 2018;Qian et al. 2020). Indeed, c-di-GMP is a ubiquitous second messenger in bacteria that functions by binding with protein receptors or RNA riboswitches (Römling et al. 2005(Römling et al. , 2013. We have shown that elevated intracellular c-di-GMP levels inhibit HSAF biosynthesis, and further identified CdgL as a c-di-GMP-binding protein that is required for HSAF production (Han et al. 2020). CdgL is a protein homologous to YajQ from plant pathogenic X. campestris, in which it directly interacts with c-di-GMP to contribute to bacterial virulence ). In our earlier study, we found that at low intracellular c-di-GMP level, CdgL forms a stable protein complex with LysR to strengthen the binding affinity of LysR with pHSAF, the promoter region of the HSAF biosynthesis operon, thereby enhancing HSAF operon expression (Han et al. 2020). At high intracellular c-di-GMP level, c-di-GMP binding to CdgL can induce the disassembly of the CdgL-LysR complex to release CdgL from the LysR-DNA complex, leading to decreased operon expression (Han et al. 2020). These earlier findings collectively reveal that HSAF biosynthesis in L. enzymogenes can be fine-tuned by designing an interconnected cellular network constituted by LenB2, 4-HBA, LysR, c-di-GMP, and CdgL. However, the mechanism of how these components are linked remains unclear.
In the present study, we showed that inactivation of the c-di-GMP receptor CdgL remarkably impaired the transcription of lenB2, leading to lower amounts of both 4-HBA and HSAF. Overexpression of cdgL, in contrast, could promote the lenB2 transcription and inhibit the HSAF yield. We further found that higher amount of lenB2 transcription is correlated with increasing 4-HBA levels, which can disassociate the LysR-CdgL complex, leading to blocked HSAF generation. These observations, along with our earlier finding of c-di-GMP-induced disassembly of the CdgL-LysR complex (Han et al. 2020), led us to propose a model addressing how the regulatory effects of two small molecules (4-HBA and c-di-GMP) in HSAF biosynthesis are linked through targeting the same protein complex (CdgL-LysR) formed by their respective receptors. The potential significance for this newly-discovered mechanism in facilitating ecological adaption of L. enzymogenes in natural niches is also discussed.

CdgL stimulated the transcription of lenB2
We have carried out a CdgL-mediated transcriptomics study and discovered that the transcription of the HSAF biosynthesis operon was positively controlled by CdgL (Han et al. 2020). A closer analysis of the transcriptomics data led us to observe that lenB2 formed another CdgL regulon, with the inactivation of CdgL blocking the lenB2 transcription (Han et al. 2020). To confirm this result, we measured the mRNA levels of lenB2 in the cdgL mutant and found that the lenB2 level was reduced by approximately 10 folds to that of wild-type (Fig. 1a). As expected, the plasmid-borne cdgL gene restored the lenB2 transcript abundance back to normal (Fig. 1a). These results suggest that intracellular CdgL stimulated, directly or indirectly, the expression of lenB2. To validate this finding, we further investigated the effect of a plasmid-boren cdgL on the growth of wild-type OH11 for checking the lenB2 transcript level. Introduction of the plasmid-borne cdgL (Additional file 1: Figure  S1) into the wild-type OH11 decreased its growth at the stationary phase, but not at the logarithmic phase (Fig.  1b). Based on this finding, we subsequently collected bacterial samples at OD 600 of 1.0 ( Fig. 1b) for qRT-PCR, and the result of Fig. 1c showed that introducing the plasmid-borne cdgL significantly increased the lenB2 transcript abundance by approximately 15 folds compared to that of wild type carrying an empty vector.
CdgL regulated production of 4-HBA and 3-HBA Findings of CdgL in stimulating lenB2 transcription revealed a potential role of CdgL in 4-HBA production, as we already knew that LenB2 is required for the generation of 4-HBA . By HPLC analysis, we found that the 4-HBA level was indeed significantly reduced in the cdgL mutant, while the plasmid-borne cdgL complementation restored the amount of 4-HBA to the wild-type level (Fig. 2a). Since we have previously shown that LenB2 is essential for the production of both 4-HBA and 3-HBA in vivo, with 4-HBA regulating the HSAF biosynthesis and 3-HBA involved in a yellowpigment development , we thus further quantified the amounts of secreted 3-hydroxybenzoic acid (3-HBA) in the test strains (Fig. 2a). As expected, mutation of cdgL also remarkably impaired the 3-HBA yield and yellow-pigment amounts, which were in agreement with those of the control strain, the lenB2 mutant ( Fig. 2a-d). Exogenous supplement of 3-HBA at physiological range or introducing the plasmid-borne cdgL into the cdgL mutant both rescued the observed deficiencies ( Fig. 2a-d). These results revealed that CdgL could control 4-HBA and 3-HBA synthesis through altering the transcription of lenB2.

CdgL accumulation impaired HSAF biosynthesis
Previously, we found that CdgL is required for the c-di-GMP-mediated HSAF generation (Han et al. 2020). This, along with the involvement of CdgL in 4-HBA synthesis, indicated that CdgL might also participate in the 4-HBA-mediated pathway to control HSAF production. Considering that overexpression of cdgL could stimulate the transcription of lenB2 (Fig. 1c), we thus assumed that wild-type OH11 containing a plasmid-borne cdgL may produce a higher amount of HSAF. Unexpectedly, the data shown in Fig. 3a did not seem to support this hypothesis, since we found that wild-type OH11 with a plasmid-borne cdgL, like the cdgL mutant (Han et al. 2020), also displayed a significant reduction in the amount of HSAF. This phenotype was further confirmed by the reduced transcription abundance of lafB in the cdgL overexpression strain compared to that of the vector control (Fig. 3b). Considering that overexpression of cdgL promoted the expression of lenB2, we thus tested to see whether the lower amount of HSAF in the cdgL overexpression strain is associated with the higher lenB2 expression. For this purpose, we overexpressed lenB2 in wild-type OH11, and indeed found that it reduced the HSAF yield compared to that of the vector control ( Fig.  3c). However, we still could not conclude that reduction in the amount of HSAF in the lenB2 overexpression strain is either due to the increased levels of 4-HBA, or to the accumulated levels of 3-HBA since we previously showed that LenB2 could convert chorismate to both 4-HBA and 3-HBA . To address this issue, we overexpressed ubiC in the wild-type OH11, which is proven to explicitly convert chorismate to 4-HBA, as described previously (Siebert et al. 1994). We subsequently found that the wild-type OH11 carrying an overexpressed ubiC indeed produced lower amounts of HSAF compared to the vector control (Fig. 3d). Taken together, these results reveal that the cdgL overexpression could stimulate lenB2 expression, the translation of which seems to generate higher amounts of 4-HBA to block HSAF biosynthesis.

4-HBA disrupted the LysR-CdgL association
To test the above hypothesis, we further investigated the potential effect of 4-HBA on the LysR-CdgL association, as disruption of this binary complex has been shown to block HSAF production, according to our earlier report (Han et al. 2020). For this purpose, we carried out a series of microscale thermophoresis (MST) assays to a qRT-PCR analyses of lenB2 mRNA abundance in the wild-type OH11 and its derivatives. b The growth of the wild-type OH11 carrying a plasmid-borne cdgL in the HSAF-producing medium, 1/10 TSB. Cells at OD 600 of 1.0 (indicated by dashed lines) were collected for qRT-PCR assay. c Effect of cdgL expression in the wild-type OH11 on the mRNA abundance of lenB2. ΔcdgL, the cdgL deletion mutant; ΔcdgL (pBBR) and ΔcdgL (cdgL), the mutant strain carrying an empty vector or a plasmid-borne cdgL; OH11(cdgL), the wild-type strain carrying a plasmid-borne cdgL. In all assays, average data from three experiments are shown, ± SD. **P < 0.01 quantify the CdgL-LysR binding affinity in the presence or absence of 4-HBA. We found that 4-HBA indeed strongly inhibited the CdgL-LysR binding. Without 4-HBA, LysR-His 6 could bind to GST-CdgL with a high affinity (K d , 0.56 μM, Fig. 4a), which is consistent with our recent report (Han et al. 2020). Yet, in the presence of 4-HBA at concentrations ranging from 0.5 to 5 μM, the binding affinity of LysR-His 6 with GST-CdgL was gradually decreased (Fig. 4b-d), with the binding completely abolished when the level of 4-HBA reached 5 μM (Fig.  4d). As a control, 3-HBA, the structural analog of 4-HBA, exhibited only a small inhibition effect on the LysR-CdgL binding at a concentration of 5 μM (Fig. 4e). These results suggested that 4-HBA binds specifically to LysR  to impair the LysR-CdgL formation, resulting in a lower HSAF production. To further support this hypothesis, we carried out a series of EMSA assay. Results of Fig. 4f showed that the CdgL-LysR-DNA ternary complex formation was still very stable in the presence of moderately lower 4-HBA concentrations of 5 and 10 μM. However, 5 μM of 4-HBA could efficiently disassociate the CdgL-LysR complex as determined by the MST method (Fig. 4d). The precise reason for this inconsistent phenomenon is unclear, but the MST technique may be more sensitive in detecting the 4-HBA-induced disassembly of the CdgL-LysR complex than that of EMSA. In agreement, when a higher concentration (20 μM) of 4-HBA ) was applied, the amount of the CdgL-LysR-DNA complex decreased significantly while the LysR-DNA binary complex reappeared (Fig. 4f). Notably, 4-HBA did not interact with the test DNA probe, according to our earlier work  or CdgL (Additional file 1: Figure  S2). Therefore, these data suggest that 4-HBA very likely Fig. 2 Effect of CdgL on the production of 4-HBA, 3-HBA, and yellow-pigment in L. enzymogenes. a Quantification of 3-HBA and 4-HBA levels produced in the wild-type OH11 and its derivatives as measured by HPLC. b Effect of exogenous addition of 3-HBA or 4-HBA on the production of yellow pigment of the ΔcdgL or ΔlenB2 strain. c Effect of CdgL on the production of yellow pigment in L. enzymogenes. d Representative results showing the yellow or white phenotype from the test strains. OH11, the wild-type strain; ΔcdgL, the cdgL deletion mutant; ΔcdgL (pBBR) and ΔcdgL (cdgL), the mutant strain carrying an empty vector or a plasmid-borne cdgL; ΔlenB2, the lenB2 deletion mutant. Average data from three experiments are shown, ± SD. **P < 0.01 binds with LysR  to change the CdgL-LysR complex conformation, leading to CdgL release from the LysR-CdgL-DNA ternary complex. Without CdgL, the LysR-DNA binding was impaired, which decreased the extent of HSAF generation.

Discussion
It seems that L. enzymogenes has evolved a strategy to secrete HSAF as an antifungal weapon to kill surrounding fungi as foods in nutrient-poor condition and in the presence of fungi (Qian et al. 2009), while HSAF production would be needless to prevent energy loss in a nutrient-rich environment or in the absence of fungi (Li et al. 2006;Yu et al. 2007). However, the underlying mechanism remains poorly understood. Our current work, along with those previously published Han et al. 2020), allow us to propose a rational mechanism (Fig. 5) regarding how L. enzymogens finetunes the HSAF biosynthesis. We argued that overexpression of cdgL, stimulated by an unknown environmental signal (likely the presence of fungi), could induce the accumulation of CdgL, which boosted the transcription of lenB2 (via an unidentified mechanism) to increase intracellular 4-HBA levels. Next, 4-HBA binding to LysR ) could impair the LysR-CdgL complex formation to release CdgL from the LysR-CdgL-DNA ternary complex. This would weaken the binding affinity of LysR to the upstream region of the HSAF operon promoter, resulting in lower HSAF operon expression and production. This strategy could enable L. enzymogenes to gain a better ecological fitness to turn on HSAF production to kill nearby fungi for food or to turn off this event to avoid energy waste in a relatively nutrient-rich environment or in the absence of fungi.
Notably, our findings presented in this work have also provided some clues to connect the pathways of 4-HBA and c-di-GMP, two structurally unrelated small molecules yet acting synergistically to regulate HSAF biosynthesis. Previously we showed that c-di-GMP binding to its receptor CdgL can weaken the CdgL-LysR binding and lead to impaired HSAF biosynthesis (Han et al. 2020). Our present study further showed that 4-HBA also used a similar strategy by binding to a different receptor protein of LysR ) to impair the CdgL-LysR complex formation. Thus, either 4-HBA or c-di-GMP could regulate the HSAF biosynthesis through binding and targeting to its cognate receptor in the same protein complex in a synergistic way. To the best of our knowledge, such a mechanism linking the interplay of Fig. 3 CdgL and LenB2 accumulations blocked HSAF biosynthesis. a Quantification of HSAF levels measured by HPLC in the wild-type OH11 carrying a plasmid-borne cdgL. Relative amounts of HSAF (y-axis) are expressed as peak intensities from the HPLC chromatogram per unit of bacterial optical density, OD 600 (Qian et al. 2013). b qRT-PCR analyses of lafB mRNA abundance in the wild-type OH11 carrying a plasmid-borne cdgL. c, d Quantification of HSAF levels measured by HPLC in the wild-type OH11 carrying a plasmid-borne lenB2 (c) or ubiC (d). OH11(pBBR), the wild-type strain carrying an empty vector; OH11(cdgL), the wild-type OH11 possessing a plasmid-borne cdgL; OH11(lenB2), the wild-type strain carrying a plasmid-borne lenB2; OH11(ubiC), the wild-type strain carrying a plasmid-borne ubiC. Average data from three experiments are shown, ± SD. **P < 0.01 two small molecules has never been documented for bacteria to date. According to an earlier study (Wang et al. 2018), 4-HBA can act as a versatile intermediate for numerous value-added bio-products, such as resveratrol and ubiquinone. Unlike these cases, we demonstrated that 4-HBA could play a regulatory/signaling role in L. enzymogenes, since it could bind to and disrupt the LysR-CdgL complex formation to inhibit HSAF biosynthesis. Such a function of 4-HBA is also specific since its structural analog 3-HBA failed to do the job. Thus, our findings also shed light on the functional and mechanistic diversity of 4-HBA in bacterial cells.
In our earlier work , we showed that exogenous addition of 0.5-μM 4-HBA that is 10-fold lower than the affinity of 4-HBA binding with LysR (K d , 5 μM) is already effective in restoring HSAF production in the lenB2 mutant to the wild-type level. Thus, 4-HBA in L. enzymogenes might have other unidentified, Fig. 4 Effect of 4-HBA on the CdgL-LysR binding and the CdgL-LysR-DNA ternary complex formation. a Microscale thermophoresis (MST) showing that GST-CdgL interacted with LysR-His6 with a K d of 0.56 μM. FNorm was plotted on a linear y-axis in per mil (‰) against the total concentration of thetitrated partner on a log 10 x-axis as described previously (Seidel et al. 2013). b, c, d Characterization of the LysR-CdgL binding affinity via MST in the presence of various concentrations of 4-HBA: 0.5 μM (b),1 μM (c), or 5 μM (d). e Effect of 3-HBA on the interaction between CdgL and LysR. f Effect of 4-HBA on the LysR-CdgL-DNA ternary complex formation determind by electrophoretic mobility shift assay membrane-bound, or cytoplasmic sensors or receptors other than LysR. In addition, several other interesting questions remain: For example, how CdgL, a protein without any known DNA binding domain, affects the transcription of lenB2? Does it work with a transcription factor? Also, does 4-HBA affect the synthesis of c-di-GMP or vice versa? In-depth investigations are required for these issues.

Conclusions
At present, it is clear that a single bacterial cell could produce a variety of small molecules with diversified functions. How these chemicals establish cellular communication is not fully understood. The biocontrol bacterium, L. enzymogenes, produces two small molecules, 4-HBA and c-di-GMP, both of which could control the antifungal HSAF biosynthesis. In this work, we show that the c-di-GMP-binding receptor CdgL promoted the transcription of lenB2 to increase the intracellular levels of 4-HBA. As a potential feedback, 4-HBA binds with the transcription factor LysR to disrupt the CdgL-LysR complex, resulting in low HSAF production. Therefore, L. enzymogenes seems to be able to adopt a LysR-CdgL In the presence of nearby fungi, L. enzymogenes appears to be able to stimulate the HSAF (red asterisk) production and secretion to kill them for foods. In this case, the CdgL level was low, unable to significantly activate the transcription of lenB2 (thin black arrow), leading to low 4-HBA production (blue pentagram). It is important to note that the potential activation mechanism remains unknown (question mark). Similarly, the intracellular c-di-GMP (black circle) levels are also likely low, according to our earlier work (Han et al. 2020). In the absence of c-di-GMP or 4-HBA, the CdgL-LysR complex was stable enough to strengthen the binding of LysR to the HSAF operon promoter region, leading to high operon expression (thick red arrow). b In the absence of surrounding fungi or presence of a nutrient-rich signal (lightning symbol), L. enzymogenes could turn off the HSAF biosynthesis to avoid energy waste. At this moment, CdgL likely accumulated via an unknown manner to stimulate the transcription (thick black arrow) of lenB2 to produce higher intracellular levels of 4-HBA. Subsequently, 4-HBA binding to LysR disrupted the LysR-CdgL complex to release CdgL from the LysR-CdgL-DNA ternary complex, resulting in a weak LysR binding to DNA and leading thus to a low HSAF operon expression (thin red arrow). According to our earlier data (Han et al. 2020), the level of c-di-GMP at this moment was high enough to bind with CdgL and to release CdgL from the LysR-CdgL-DNA ternary complex, leading to further decrease the HSAF operon expression complex to interconnect the 4-HBA and c-di-GMP pathways to control the HSAF biosynthesis synergistically in a rather elegant way. Thus, our findings reveal a previously uncharacterized mechanism by which two small molecules are functionally linked. Engineering this novel mechanism may be helpful for enhancing the HSAF yield to serve as an applicable bio-fungicide.

Bacterial strains, plasmids and growth conditions
The bacterial strains and plasmids used in this study are listed in the Additional file 1: Table S1. Escherichia coli strains were grown in LB medium with appropriate antibiotics at 37°C. L. enzymogenes strains were cultivated in LB medium or 1/10 Tryptic Soy Broth (TSB) at 28°C. When required, appropriate antibiotics (kanamycin, Km, and gentamicin, Gm) were added into media.

Complementation and overexpression
The coding region of the CdgL-Flag fusion was PCRamplified by primers (Additional file 1: Table S2). The purified product was cloned into the broad-host vector pBBR1-MCS5 to generate a final construct (Additional file 1: Table S1). This construct was transformed into the wild-type OH11 or mutants by electroporation as described previously (Qian et al. 2012), with the transformants selected on the LB agar containing Km and Gm.

Quantitative RT-PCR, qRT-PCR
The wild-type OH11 and its derivatives were cultivated in 1/10 TSB medium and collected at OD 600 of 1.0. Bacterial RNA Kit (No. R6950-01, OMEGA, China) was used to extract total RNA from collected cells. To remove genomic DNA, 400 ng RNA samples were treated with RNase inhibitors and DNase I (No.E1091, OMEGA, China). Next, 2 μg RNA sample was used to synthesize cDNA using PrimerScript™ RT reagent Kit with gDNA Eraser (No. RR047A, TaKaRa, Japan). qRT-PCR was carried out by an Applied Biosystems 7500 system using the primers listed in the Additional file 1: Table S2 with the following program: 95°C for 30 s, followed by 40 cycles of 95°C for 5 s, 60°C for 34 s, before fluorescence detection followed by a melting curve determined with 1 cycle of 95°C for 15 s, 60°C for 60 s and 95°C for 30 s. The 16S rDNA was used as an internal control as described previously (Qian et al. 2013(Qian et al. , 2014. Data were analyzed using SPSS 14.0 (SPSS Inc., Chicago, IL, USA). The hypothesis test of percentages (t-test, P = 0.05 or 0.01) was used to determine the significant differences in gene expression.

Quantification of yellow pigments
Extraction of yellow pigment was carried out as described previously (Zhou et al. 2013). In brief, cells from 8 mL of the culture of the wild-type OH11 and its derivatives were harvested at the stationary phase (OD 600 , 3.0). The collected cells were re-suspended in 750 μL methanol, followed by addion of the same volume of chloroform via thoroughly mixing, which were further subjected to centrifugation at 13,000 rpm for 10 min for collecting the supernatant. The amount of pigment was expressed as the absorbance of the crude pigment extract (absorbance at 445 nm) (Poplawsky and Chun 1997;Zhou et al. 2013) HSAF extraction and quantification HSAF was extracted from 50 mL of L. enzymogenes cultures grown in 1/10 TSB for 24 h at 28°C with shaking at 200 rpm. A total of 4 mL of culture was mixed with an equal volume of ethyl acetate with 12 μL hydrochloric acid. After shaking, the ethyl acetate phase was collected and evaporated to dryness. The final HSAF-containing residue was dissolved in 200 μL of methanol. HSAF was then determined by HPLC and quantified as per unit of OD 600 as described earlier (Yu et al. 2007;Qian et al. 2013). All tested strains were carried out in triplicate, with each analysed in three technical replicates. Data were analyzed using SPSS 14.0 (SPSS Inc., Chicago, IL, USA). The hypothesis test of percentages (t-test, P = 0.05 or 0.01) was used to determine significant differences in HSAF level.
Extraction and quantification of 3-HBA and 4-HBA 3-HBA and 4-HBA were extracted from 15 mL of L. enzymogenes cultures grown in 1/10 TSB for 2 days at 28°C with shaking at 200 rpm. Cells were removed by centrifugation (13,000 rpm for 3 min). The harvested supernatant was mixed with an equal volume of ethyl acetate, followed by shaking at 28°C for 30 min. The ethyl acetate phase was collected by centrifugation (13, 000 rpm for 10 min) and further evaporated to dryness. Both 3-HBA and 4-HBA were detected and quantified using HPLC, as described previously .

Microscale thermophoresis, MST
The affinity of protein-protein interaction was determined by MST using Monolith NT.115 (NanoTemper Technologies, Germany) as described previously Xu et al. 2018;Han et al. 2020). For the LysR-CdgL binding assay, the purified LysR-His was labeled with the fluorescent dye NT-647-NHS (Nano Temper Technologies) via amine conjugation. A constant concentration (40 μM) of the labeled protein in the MST buffer (50 mM Tris, pH 7.5, 150 mM NaCl,10 mM MgCl 2 , 0.05% Tween 20) was titrated against GST-CdgL dissolved in MST buffer at concentrations ranging from 440 μM to 13.4 nM. The MST premium-coated capillaries (Monolith NT.115 MO-K005, Germany) were used to load the samples into the MST instrument at 25°C using 80% MST and 20% LED power. FNorm was plotted on a linear y-axis in per mil (‰) against the total concentration of the titrated partner on a log 10 x-axis, as reported previously (Seidel et al. 2013). The remaining tests were run similarly to the LysR-CdgL binding assay. All experiments were performed in triplicate. Data were analyzed using Nanotemper Analysis software 2.2.4.4577 (NanoTemper Technologies).

Electrophoretic mobility shift assay, EMSA
EMSA assay was carried out as follows: Biotin-labeled fragment containing the promoter region of lafB (pHSAF) was amplified using the 5′-end biotinylated primers listed in the Additional file 1: Table S2. Purified proteins and DNA probe was mixed and incubated in the test system for 30 min at 28°C according to the specifications of LightShift® Chemiluminescent EMSA Kit (ThermoFisher, Waltham, USA). The reaction system was loaded onto the polyacrylamide gel, electrophoresed, transferred to a nylon membrane and crosslinked as described in the manufacturer protocol. The biotinylated DNA fragments were detected by chemiluminescence using Versa Doc imaging system (Bio-Rad, Philadelphia, USA).
Additional file 1: Table S1. Strains and plasmids used in this study. Table S2. Primers used in this study. Figutre S1. Western blot analyses of the CdgL-Flag fusion expression in Lysobacter enzymogenes. Figure  S2. MST showing no detectable GST-CdgL binding to 4-HBA.