Quorum-sensing gene regulates hormetic effects induced by sulfonamides in Comamonadaceae

ABSTRACT

Hormesis is a toxicological phenomenon whereby exposure to low-dose stress results in the stimulation of various biological endpoints. Among these, the induction of cell proliferation by antibiotics is critical, but the underlying molecular mechanisms remain poorly understood. Here, we showed that sulfonyl-containing chemicals (e.g., sulfonamides) can induce cell proliferation hormesis in Comamonas testosteroni. An investigation of the hormesis mechanism revealed that low-dose sulfonamides potentially interact with the LuxR-type quorum-sensing protein LuxR solo, thereby triggering the transcription of 3-ketoacyl-CoA thiolase, a key enzyme of fatty acid β-oxidation. This provides additional ATP, NADPH, and acetyl-CoA for macromolecule biosynthesis, allowing cells to synthesize sufficient nucleotides to support rapid cell growth. Our work reports on a previously unknown mechanism for the hormetic effect and highlights its generality in the Comamonadaceae family.

IMPORTANCE

Antibiotics can induce dose-dependent hormetic effects on bacterial cell proliferation, i.e., low-dose stimulation and high-dose inhibition. However, the underlying molecular basis has yet to be clarified. Here, we showed that sulfonamides play dual roles as a weapon and signal against Comamonas testosteroni that can modulate cell physiology and phenotype. Subsequently, through investigating the hormesis mechanism, we proposed a comprehensive regulatory pathway for the hormetic effects of Comamonas testosteroni low-level sulfonamides and determined the generality of the observed regulatory model in the Comamonadaceae family. Considering the prevalence of Comamonadaceae in human guts and environmental ecosystems, we provide critical insights into the health and ecological effects of antibiotics.

INTRODUCTION

Hormesis commonly refers to the adaptive response of organisms (e.g., microbial, plant, and animal species) to low-dose stressors (e.g., antibiotics, oxygen, ionic liquids, metals, organic pollutants, macromolecules, and nanomaterials) (1–5). Such defensive responses against environmental deterioration can improve the functional ability of cells and thus are crucial to the survival of biological species (3, 6–8). Currently, increasing attention has been paid to the hormetic responses of bacteria exposed to antibiotics, given the close relationship to human and ecosystem health. For instance, antibiotics can dramatically alter the physiological functions of bacterial cells, even at concentrations far below lethal doses, which may contribute to human gut disorders and ecosystem dysfunction (9, 10).

At high concentrations, antibiotics act as bacterial killers, but at low concentrations, they can induce hormesis, leading to various favorable outcomes for susceptible bacteria (11–13). To date, a broad range of response endpoints for hormesis have been studied, including cell growth, secondary metabolic processes, and cellular functions, such as bioluminescence and biofilm formation (14–17). Some specific hormetic dose responses have been mechanistically explained, usually based on receptors and signaling pathways (16, 18). For example, low doses of piperacillin can trigger secondary metabolite biogenesis via the OxyR and SoxR regulons in Burkholderia thailandensis (19), neomycin and erythromycin can enhance bacterial bioluminescence in Vibrio fischeri via the LuxR quorum-sensing (QS) system (20, 21), and tetracycline can up-regulate type III secretion system expression and consequently biofilm formation in Pseudomonas aeruginosa (22). Notably, although cell growth is one of the most common endpoints of hormesis, existing hormetic studies have mainly focused on microbial growth kinetics at hormetic concentrations of antibiotics (23–29). However, the underlying molecular basis has yet to be clarified, probably because cell proliferation involves a complex array of responses, including both intrinsic and extrinsic cellular metabolic reactions.

Herein, we address this issue using Comamonas testosteroni bacteria, which are frequently present in diverse habitats, including activated sludge, marshes, marine habitats, plant and animal tissues, and the human gut (30–32). Their diversified niches reflect effective adaptation to various physiochemical conditions and are thus a good model for exploring the molecular mechanism underlying hormesis. Sulfonamides (SAs) are the most extensively used antibiotics for bacterial treatment (33). Importantly, no SA-related resistance genes (e.g., sul1, sul2, and drfA) were found in the complete genome of C. testosteroni (30). We focused on the induction of cell proliferation by sulfamethoxazole (SMX) and sulfadiazine (SD) as a model for exploring how SA stress induces hormetic effects. Following the experimental procedure in Fig. S1, we provided a comprehensive understanding of the hormetic molecular mechanisms by which the LuxR solo transcription factor (TF) up-regulates fatty acid β-oxidation (FAO) through 3-ketoacyl-CoA thiolase (EC 2.3.1.16) under low-dose SA stress (<250 µg/L). Moreover, rapid cell metabolism accelerated the respiratory dissimilatory nitrate reduction to ammonium (DNRA) pathway in response to a reduced cellular redox state. Our results highlight the significant role of LuxR solo in regulating hormesis on cell growth as well as its generality in the Comamonadaceae family.

RESULTS

Hormetic effect of SAs on C. testosteroni cell growth and morphology

The growth of C. testosteroni cells was examined upon the addition of 12 gradient concentrations (0–100 mg/L) of SAs (SMX, SD, and their mixture at 1:1 concentration). When cells were in the logarithmic growth phase (approximately 24 h), low-dose SAs (5–250 μg/L) significantly accelerated cell proliferation [reflected by optical density (OD)₆₀₀], whereas SAs at concentrations above 1 mg/L inhibited cell growth (Fig. 1a). These results demonstrated that SA stress (SMX, SD, or their mixture) induced a hormetic effect on C. testosteroni cell proliferation. Notably, the hormetic effect elicited by the SA mixture showed no significant difference from half the summation of the hormetic effects induced by separate SMX and SD exposure at all administered doses, except for 100 mg/L, suggesting that the functional groups in SMX and SD that induce these hormetic effects are identical (Table S1). Using flow cytometry to measure active biomass, we further confirmed the occurrence of the hormetic response of C. testosteroni under SA stress, which started at approximately 12 h (1.09-fold, Wilcox test, P < 0.05) and peaked at 24 h (1.24-fold, Wilcox test, P < 0.05) (Fig. S2).

Fig 1.

Hormetic effects of SAs on C. testosteroni cell growth and morphology. We first investigated the effects of SMX, SD, and their mixture (1:1 in concentration) on cell growth (a), represented as OD₆₀₀ measured after 24 h of cell growth in a 96-well microplate containing MSM; mean and standard deviation of each result were calculated from six replicates. The dashed line represents the average cell density without SA treatment. The two-sample Wilcox test was used to evaluate significant differences in cell density between cultures with/without SA treatment; *P < 0.05. We used inverted microscopy images to capture the morphology variation of C. testosteroni after 24 h of cell growth inside the ONIX microfluidic platform containing MSM under different concentrations of SA mixture (b). Brightfield images were acquired using a 100× oil immersion microscope objective. Scale bar, 2 µm.

Consistently, cell phenotypes changed along with the hormetic effect on cell growth. Under low-dose SA stimulation (5–250 μg/L, including individual and mixed SAs), cells experienced a rod-to-ellipsoid transition and formed diplococci or short chains (Fig. 1b; Fig. S3). Under SA mixture exposure, the mean cell length fluctuated slightly (0.54 ± 0.14 µm; n = 100) compared to the untreated cells (1.64 ± 0.15 µm) (Table S2). In contrast, under the inhibitory effect of high-dose SAs (>500 µg/L), cells transitioned to very long filaments with normal cell widths (up to 10-fold the mean length of cells at the same growth stage without SA treatment) (Fig. 1b; Fig. S3). Approximately 90% of cells were longer than 2.5 µm, more than 50% of which were >5 µm (n = 100) (Table S2). The filamentary formation was due to SA-induced DNA damage and replication perturbation, a common phenomenon observed with SA antibiotic treatment (34, 35). However, the rod-to-ellipsoid shape alteration has not been reported previously. During growth, cells change their overall shape and size in rapid response to changing conditions by adjusting the synthesis of the cell envelope and cell division, which are tightly coordinated with DNA replication and protein synthesis through central metabolism (36).

These observations collectively give rise to two key questions: (i) which metabolic responses are elicited in C. testosteroni cells during the hormetic effect triggered by SA and (ii) what mechanisms underlie the induction of metabolic changes by SAs?

SAs trigger metabolic adaptation responses in C. testosteroni

We next conducted transcriptomics analysis based on RNA sequencing (RNA-seq) of C. testosteroni log-phase cells (24 h) under SA mixture exposure to explore metabolic responses. A comparison between cells grown with/without 50 µg/L SA mixture revealed a total of 486 significant differentially expressed genes (DEGs) (11.38% of all genes in the C. testosteroni genome), including 422 up-regulated DEGs. A comparison between cells grown with/without 1 mg/L SA mixture revealed 673 significant DEGs (18.18% of all genes), including 378 up-regulated DEGs. Based on gene set enrichment analysis (GSEA) of differentially expressed proteins, cellular metabolism– and signal transduction–related biological processes were significantly up-regulated (Fig. 2a through c; Table S3).

Fig 2.

Global metabolic adaptation responses of C. testosteroni during growth under SA mixture treatment. Enriched Kyoto Encyclopedia of Genes and Genomes (KEGG) pathways among (a) up-regulated genes and (b) down-regulated genes under SA mixture treatment (50 µg/L and 1 mg/L, at a 1:1 concentration of sulfamethoxazole and sulfadiazine). Closed circles represent pathways, and the size of each circle corresponds to the number of genes associated with that pathway. A false discovery rate (FDR) q-value <0.05 was considered the cutoff for each enriched pathway. In subfigure (c), the schematic shows a metabolic regulatory network in C. testosteroni cells determined by transcriptomic analysis. The targeted LC-MS/MS-based metabolomics approach was further used to profile fatty acid (d) and central carbon (e) metabolites in C. testosteroni cells grown with/without 50 µg/L SA mixture treatment. Significance (*P < 0.05) was determined using one-way analysis of variance followed by Tukey HSD post hoc tests. (d) Intracellular pools (ng/mL) of saturated fatty acids (SFAs), monosaturated fatty acids (MUFAs), and polyunsaturated fatty acids (PUFAs). Data are expressed as mean ± cumulative standard deviation of six biological replicates. (e) Heatmap shows fold change (log₂FC) in an intracellular pool of central carbon, divided into TCA cycle-related organic acids, phosphorylated intermediates, and nucleotides. Relative metabolite concentrations were normalized to a mean equal to 0 and standard deviation equal to 1. Metabolite abbreviations for panels c and e are as follows: glucose-6-phosphate, G6P; fructose-6-phosphate, F6P; fructose-1,6-bisphosphate, FBP; dihydroxyacetone phosphate, DHAP; glyceraldehyde-3-phosphate, GAP; 6-phosphogluconate, 6 PG; ribulose 5-phosphate, Ru5P; xylulose-5-phosphate, Xu5P; ribose-5-phosphate, R5P; sedoheptulose-7-phosphate, S7P; erythrose 4-phosphate, E4P; 1,3-biphosphoglycerate, 1,3BPG; 3-phosphoglycerate, 3 PG; 2-phosphoglycerate, 2 PG; phosphoenolpyruvate, PEP; oxaloacetate, OAA; α-ketoglutarate, αKG.

Among the catabolism categories, “fatty acid degradation” was the most significantly enriched biological process for both SA-tested concentrations [Fig. 2a; Table S3, n = 14, (false discovery rate) q-value <0.05], in which the activities of aldehyde dehydrogenase, acyl-CoA dehydrogenase, 3-hydroxy acyl-CoA dehydrogenase, and 3-ketoacyl-CoA thiolase were significantly up-regulated for both SA-tested concentrations (Fig. S4a, |log₂FC| > 1.5, FDR q-value <0.01). These enzymes are actively involved in FAO to generate NADH, flavin adenine dinucleotide (FADH₂), acetyl-CoA, and NADPH (37). In the next few steps, the generated NADH and FADH₂ enable the production of more ATP (Fig. S5a), while acetyl-CoA enters the TCA cycle to produce citrate and other TCA cycle intermediates (Fig. S5b). Furthermore, NADPH can serve as a coenzyme for anabolic building blocks required for cell proliferation, such as lipid and nucleic acid synthesis (Fig. S5c). Increased carbon generated in the TCA cycle is further converted into phosphoenolpyruvate (PEP), which subsequently enters the gluconeogenesis and pentose phosphate (PP) pathways to provide additional substrates and energy for the biosynthesis of amino acids, lipids, and nucleotides. The enzymes involved in these processes were significantly up-regulated (Fig. S4b through d), especially the rate-limiting cataplerotic enzyme of the gluconeogenesis pathway, PEP carboxykinase (Pck) (38), which showed a 2.43-fold and 2.21-fold increase under 50 µg/L and 1 mg/L SA mixture exposure, respectively (Fig. S4d). Compared with other nutrients, nucleotide production is of particular importance for proliferating cells, as it is needed to synthesize ribosomal RNA, duplicate the genome, and maintain the transcriptome (39). Thus, we speculated that SAs may directly facilitate the oxidative decomposition of fatty acids to generate ATP, NADPH, and acetyl-CoA for further purine and pyrimidine biosynthesis, thus allowing C. testosteroni cells to synthesize sufficient nucleotides to support rapid cell growth. In contrast, cells exposed to the 1 mg/L SA mixture showed a significant decrease in folate biosynthesis (Fig. 2b; Fig. S4f and Table S3, n = 10, FDR q-value <0.05). This was expected because the bacteriostatic mechanisms of SAs can block the production of folic acid and therefore inhibit nucleotide synthesis (40, 41) (e.g., purine metabolism and pyrimidine metabolism), thus resulting in the growth inhibition effect observed at 1 mg/L SAs.

We next used a targeted metabolomics approach based on ultra-performance liquid chromatography–tandem mass spectrometry (UPLC-MS/MS) to validate the RNA-seq results. We profiled fatty acids and central carbon metabolites in C. testosteroni log-phase cells with/without 50 µg/L SA mixture treatment. Targeted fatty acid quantification showed that SA mixture exposure significantly decreased almost all oxidizable long-chain fatty acid contents (Fig. 2d), including saturated fatty acids (C16-22:0), monosaturated fatty acids (C15-24:1), and polyunsaturated fatty acids [C18:2(n-6), C18:2(n-6) T, C18:3(n-3), and C20:3(n-3)]. Furthermore, compared to growth without SA treatment, the relative abundance of TCA cycle-related organic acids and phosphorylated intermediates was significantly increased under SA treatment (Fig. 2e, 5.5-fold, P < 0.05). The cellular metabolite pools suggested that enhanced FAO decreased fatty acid content, thereby promoting carbon availability in the TCA cycle and accumulation of carbon in the upper gluconeogenesis pathway in SA-exposed cells, consistent with transcriptional pathway analysis.

Among the catabolism categories, the “nitrogen metabolism” biological process also exhibited significant enrichment (Fig. 2a; Table S3, n = 18, FDR q-value <0.05). Most notably, the nitrate and nitrite reductases Nap/Nrf, which encode the respiratory dissimilatory nitrate reduction to ammonium pathway, were highly expressed under 50 µg/L and 1 mg/L SA mixture exposure, respectively (Fig. S4e). It included the catalytic subunit NapA (2.24-fold and 2.19-fold), non-heme iron–sulfur cluster protein NapG (3.44-fold and 1.25-fold), pathway-specific chaperone NapD (2.48-fold and 1.12-fold), and pentaheme cytochrome c nitrite reductase NrfA (2.14-fold and 1.29-fold). The DNRA pathway produces energy (ATP) through oxidative phosphorylation via an electron transport chain (42), which can be used to maintain cellular activities and promote cell growth. Furthermore, this periplasmic pathway is also considered an effective electron sink to consume excess reduction forces, such as NADH and FADH₂ (43, 44). Hence, the increased expression of respiratory DNRA genes in C. testosteroni cells growing with the SA mixture may maintain redox in rapidly proliferating cells.

To exclude the potential effects of RNA degradation, we next investigated the expression levels of three key 3′−5′ exoribonucleases, namely, RNase II, RNase R, and PNPase, which are known to play major roles in RNA degradative activity (45, 46). Our results showed that the expression levels of these exoribonucleases did not show significant up-regulation, indicating that the overall RNA degradation activity remained relatively stable throughout the duration of our experiment (Fig. S6a). Additionally, considering the potential oxidative stress response in aerobic metabolism, we assessed the expression differences of several key regulators involved in oxidative stress, including OxyR, SoxRS, and RpoS, under SA treatment. However, no significant changes were observed in their expression levels (Fig. S6b). These findings not only confirmed RNA stability but also indicated the absence of stress-induced responses in our experimental conditions. Furthermore, the findings obtained from quantitative reverse transcriptase polymerase chain reaction (qRT-PCR) were consistent with the observations from RNA-seq and metabolomic analyses, providing additional support for the significant enrichment in fatty acid metabolism, biosynthesis, and DNRA pathways. Particularly, specific genes involved in fatty acid oxidation exhibited a 1.3–2.5-fold increase, PEP carboxykinase (Pck) displayed a 1.4–2.0-fold increase, and the Nap/Nrf systems showed a 1.3–2.2-fold increase (Fig. S7).

SAs require transcriptional regulator LuxR solo to induce metabolic responses

Given the crucial role of signaling processes in reshaping metabolic activities to sustain cell growth (47), we directed our attention to the signal transduction-related category. Here, we found that “quorum sensing” was the highly enriched biological process for both SA concentrations tested (Fig. 2a; Table S3, n = 21, FDR q-value <0.05). Notably, prior studies exploring the transcriptional modulation of bacterial gene expression using subinhibitory concentrations of antibiotics have also demonstrated the activation of quorum sensing under similar conditions (48). Hence, we hypothesized that low-dose SAs could act as signaling molecules to activate QS, thereby influencing cellular metabolic responses.

Among QS genes, the LuxR family transcriptional regulator (3.8-fold and 2.9-fold, FDR q-value <0.01) and ABC transporter (1.8-fold and 1.6-fold, FDR q-value <0.01) were highly expressed under 50 µg/L and 1 mg/L SA mixture exposure, respectively (Fig. 3a). The qRT-PCR results further validated that the LuxR family transcriptional regulator was significantly induced (2.66–2.89-fold) under SA treatment (Fig. S7). Sequence alignment and domain analysis of the identified LuxR transcriptional regulator (Table S4, accession no: WP_003076066.1) showed that the identified LuxR family protein was comprised of two complete functional domains, similar to typical QS proteins (Table S5), including a C-terminal region containing a predicted helix-turn-helix (HTH) motif implicated in DNA binding (e-value = 1.90e−11) and an N-terminal autoinducer-binding domain (e-value = 3.60e−12). Generally, an archetypical QS system in gram-negative bacteria is mediated by the LuxI and LuxR protein families (49). LuxI-type proteins are N-acylhomoserine lactone (AHL) synthases that synthesize AHL signals, and LuxR family protein is a regulator that directly binds to cognate AHL (49). These protein–AHL complexes then bind to the specific gene promoter sequence to regulate the expression of the QS target gene (49). Thus, we used BLASTP to identify proteins that may encode LuxI homologs. However, no unpaired or extra genes coding for LuxI homologs were present in the remainder of the genome. LuxR proteins with the same modular structure as QS LuxRs but devoid of a cognate LuxI AHL synthase are called solos (50). Current evidence suggests that LuxR solo may exhibit a high degree of variability in both the type of ligands to which they respond and the mechanism by which they regulate target genes (51, 52). For example, the subfamily of LuxR solos in Xanthomonas no longer respond to endogenously produced AHLs but do respond to plant signals to participate in interkingdom communication between the host and pathogen (53). Considering their potential ability to respond to various exogenous signals, we hypothesize that LuxR solo may provide the major pathway for SAs to induce metabolic responses. Thus, we next generated a C. testosteroni mutant based on LuxR solo gene knockdown to investigate the possible involvement of LuxR solo in the metabolic response (Fig. S8a and S8b). In the absence of SAs, the deletion of LuxR solo gene had no significant effect on cell growth (Fig. S17). We then tested the ability of the deletion mutant ΔLuxR to trigger hormetic effects on cell growth under SA mixture treatment. Notably, the ΔLuxR strains completely abrogated the SA-mediated growth induction effect and exhibited reduced cell density (up to 64%) despite low doses (<250 µg/L). These effects were rescued by plasmid-encoded expression of LuxR in the ΔLuxR strain (Fig. 3B; Fig. S8c and d). These findings highlighted the essential role of LuxR solo in the induction of metabolic responses by SAs. At low concentrations, we hypothesized that SAs may bind to the membrane protein LuxR solo as an AHL analog and subsequently activate two key processes: (i) FAO, providing supplementary nutrients for biosynthesis, and/or (ii) DNRA, which modulates ATP formation and redox balance (Fig. 3C).

Fig 3.

Sulfonamides require transcriptional regulator LuxR solo to induce metabolic responses. (a) The heatmap shows fold change in gene expression (log₂FC) in QS pathway under SA mixture treatment [50 µg/L and 1 mg/L, 1:1 concentration (mg/L) of sulfamethoxazole and sulfadiazine]; *P < 0.05. Red represents screened potential SA signal transduction gene LuxR solo (|log₂FC| > 1.5, FDR q-value <0.01). (b) LuxR solo is required for SA-mediated metabolic responses, as determined using C. testosteroni mutants ΔLuxR and ΔLuxR-pBBRLuxR to trigger hormesis effect on cell growth (OD₆₀₀) under SAs. The dashed line represents the average cell density without SA treatment. The Wilcox test was used to evaluate significant differences in cell density (OD₆₀₀) between cultures with/without SA treatment. Significance corresponds to adjusted Wilcox test P-values (*P < 0.05). (c) Combining subfigures (a and b), we show a proposed schematic overview of metabolism regulation by the LuxR family transcriptional regulator. Subfigures (d–f) verify the interaction of LuxR solo protein and SAs based on solubility measurements of overexpressed LuxR solo protein and docking studies. (d) Western blotting was used to show LuxR solo in soluble (S) and pellet (P) fractions of cell lysates from recombinant Escherichia coli strain BL21(DE3) supplied with 75 µM SMX or SD. M denotes marker (representative bands are labeled). Docking view of SMX (e) and SD (f) at LuxR solo binding site. Green dashed lines indicate hydrogen bonds between ligand and amino acid residues.

Potential interaction of SAs with LuxR solo via sulfonyl groups

To test the first mechanistic hypothesis, we investigated the binding of SA molecules to LuxR solo proteins. Generally, QS-LuxR proteins are unfolded or proteolyzed or form inclusion bodies in the absence of a cognate molecule, whereas in the presence of and when bound to cognate molecules, they become folded and soluble (54–56). Therefore, we used this biochemical “folding switch” feature to explore the potential interaction between LuxR solo and SAs. We cultured E. coli BL21(DE3) overexpressing His-LuxR in the absence and presence of SMX and SD and identified the presence of soluble LuxR solo using western blot analysis. When LuxR solo was overexpressed in the absence of SAs, almost all the LuxR solo protein was found in the particulate fraction, indicating that it accumulated as insoluble inclusion bodies. In contrast, the addition of SMX or SD increased LuxR solo solubility, with approximately half of the total LuxR solo protein found in the soluble fraction (Fig. 3d). Furthermore, we examined LuxR solo solubility across a gradient of low SA concentrations (ranging from 0 to 30 µg/L) at 16°C and 20°C, and LuxR solo consistently remained in the particulate fraction under these conditions (Fig. S9a and S9b). Importantly, the growth of E. coli BL21(DE3) carrying the expression vector used in this study remained largely unaffected by SA interference, which did not impact LuxR solo expression (Fig. S9c). Additionally, we utilized UPLC-MS/MS to monitor the concentration of SMX and SD in the culture medium. These measurements revealed a rapid decrease of approximately 78.7% ± 2.85% in the levels of SMX and SD within the initial 24 h, with this concentration remaining stable over the subsequent 72-h incubation period (Fig. S9d). These findings provide further support for the hypothesis that LuxR solo proteins may indeed interact with SAs.

Molecular docking was then performed to visualize the binding process. First, the SWISS-MODEL server was used to successfully generate a 3D structure for the LuxR solo protease using the crystal structure of SdiA in complex with 3-oxo-C6-homoserine lactone (PDB ID: 4Y15) as the template. Both SMX and SD were then docked onto the template. Results showed that SMX and SD were stable in the binding pocket of LuxR solo protease. The minimum binding free energy of SMX with the LuxR solo protein was −7.89 kcal mol^–1, stabilized in the binding pocket by four hydrogen bonds, i.e., between the sulfonyl group and residues Ser-43 and Tyr-63 and between the imino/amino group and residues Asp-80 and Val-68. In addition, a π–π stacking interaction existed between the benzene ring of SMX and the alkyl atoms of Tyr-71 (Fig. 3e; Fig. S10). Similarly, the minimum binding free energy with SD was −7.40 kcal mol^–1, with Asp-80, Cys-45, Tyr-63, and Ser-43 serving as anchoring points for the ligands (Fig. 3f; Fig. S11). Tyr-63 and Ser-43 formed hydrogen bonds with sulfonyl groups, Asp-80 and Cys-45 formed hydrogen bonds with the imino/amino groups, and Tyr-71 formed a C-H bond with the benzene ring. As sulfonyl groups were crucial in binding LuxR solo with SMX or SD, we docked 14 other sulfonyl-containing chemicals with the LuxR solo template using the same docking process. Among the compounds that induced hormetic effects after 24 h of growth, all showed binding free energies below −5.14 kcal mol^–1 (Fig. S12). These results suggest that C. testosteroni hormesis is strongly influenced by the interaction between LuxR solo and sulfonyl functional groups.

LuxR solo regulates 3-ketoacyl-CoA thiolase to activate fatty acid β-oxidation

Due to the presence of the LuxR-type HTH motif in the C-terminal domain and in agreement with the HTH position/function relationship postulated by Pérez-Rueda et al. (57), we postulated that the LuxR solo–SA complex may bind to the target gene promoter region. Therefore, we next identified the direct functional target of LuxR solo–SA in C. testosteroni. We performed RNA-seq of the C. testosteroni mutant ΔLuxR treated with 50 µg/L SA mixture and determined that LuxR solo regulated the fatty acid oxidation and/or nitrogen metabolism pathways, although other activation pathways could not be ruled out. We defined the target gene of LuxR solo based on the following criteria: (i) the gene significantly expressed in wild-type (WT) C. testosteroni cells was no longer changed after deleting LuxR solo gene upon low-dose SA treatment, and (ii) its upstream promoter region contained the lux-box bound by LuxR solo. Nine clusters associated with FAO and nine clusters associated with DNRA were investigated. We observed that DNRA genes in the deletion mutant ΔLuxR were up-regulated, a transcriptional pattern similar to that observed in SA-exposed WT cells (Fig. S13). In contrast, the FAO gene clusters that showed significant changes in WT cells disappeared in the ΔLuxR mutants (Fig. 4a). Analysis based on qRT-PCR further validated the RNA-seq results, showing that only significant changes in the fatty acid oxidation pathway were lost in mutant cells (Fig. S14). Thus, these results suggest that LuxR solo may potentially regulate the FAO pathway but not the DNRA pathway.

Fig 4.

LuxR solo directly up-regulates fatty acid β-oxidation. Subfigure (a) summarizes the fatty acid β-oxidation pathway (top), and the heatmap (below) shows related fold change in gene expression (log₂FC) in wild-type and mutant ΔLuxR upon low-dose SA mixture treatment (50 µg/L, 1:1 concentration). P-values are compared to strains without SA treatment (*P < 0.05). Three binding sites (lux-like box) were identified in the 3-ketoacyl-CoA thiolase promoter region (b). Sequences of putative LuxR binding sites (left) are shown in the 5′–3′ direction (gray); red represents conserved binding sites. EMSA (right) shows that LuxR solo binds to the 3-ketoacyl-CoA thiolase promoter region. Conditions are indicated below each lane.

To further identify the direct binding site of LuxR solo, we focused on three operons involved in FAO, aldehyde dehydrogenase (WP_012836683.1), 3-hydroxyacyl-CoA dehydrogenase (WP_012839896.1), and 3-ketoacyl-CoA thiolase (WP_012836887.1, WP_012838403.1, WP_012839895.1, and WP_003072098.1). The LuxR family regulator is reported to control target genes by binding to a conserved 20 bp lux-box CTG-(N10)-CAG sequence in the gene promoter, in which five of the six bases are essential for LuxR binding (58). Therefore, we retrieved the conserved lux-box in the promoter region of six FAO genes using MEME tools (59). Results showed that putative lux-like box motifs were only identified in the promoter regions of 3-ketoacyl-CoA thiolase (WP_012838403.1, WP_012839895.1, and WP_003072098.1), located −61.5, –60.5, and −55.5 bp upstream of the transcription start sites (TSSs) (Fig. 4b, left), respectively. As noted previously, LuxR in V. fischeri is a transcriptional activator rather than a repressor when the position of the lux-box is located −60 bp from the center to the TSS. This suggests that the LuxR solo protein identified here is also a transcriptional activator that interacts with the C-terminal domain of the alpha subunit of RNA polymerase (NAP; subunit composition α2ββ′σ) (60).

To test whether LuxR solo can directly bind to the 3-ketoacyl-CoA thiolase promoter region, we performed an electrophoretic mobility shift assay (EMSA). Using the 670, 570, and 600 bp promoter sequences of 3-ketoacyl-CoA thiolase (WP_012838403.1, WP_012839895.1, and WP_003072098.1), we observed a LuxR solo-dose-dependent mobility shift, consistent with the formation of a DNA–protein complex (Fig. 4b, right). The LuxR solo protein exhibited robust binding to the putative promoter region even at low concentrations (Fig. S15a). However, this binding dramatically diminished when we mutated the identified lux-box within the promoter region of 3-ketoacyl-CoA thiolase (Fig. S15a and b). This suggested that the QS regulator LuxR solo can directly activate the expression of 3-ketoacyl-CoA thiolase.

DISCUSSION

From an eco-evolutionary perspective, the ability to adapt to antibiotics is a trait that has evolved in single-celled organisms living in natural environments (61). While antibiotics have traditionally been perceived as weapons that can inhibit bacterial growth, recent work also emphasizes their role as concentration-dependent signaling molecules (62–64). In the current study, we showed that SAs function as signaling molecules that can modulate cell physiology and phenotype of bacteria at lower concentrations.

In response to SA stress, C. testosteroni cells undergo adaptive metabolic reprogramming characterized by the broad activation of FAO (Fig. 2c). Cell metabolism can be perceived as a complex network of pathways with plasticity, feedback loops, and crosstalk to ensure cell fitness (65). Plasticity is crucial and can be partially provided by the FAO process, which generates additional ATP and NADPH, eliminates potentially toxic lipids, and provides metabolic intermediates for cell growth. Notably, NADPH serves as a coenzyme for anabolic enzymes, significantly impacting the cell’s ability to grow and survive (66). Therefore, metabolic reprogramming must meet the requirement of producing reduced NADP⁺ (67). Generally, this is achieved through three primary enzymatic reactions, i.e., oxidation of glucose via the PP pathway, metabolism of malate to pyruvate, and oxidation of isocitrate to α-ketoglutarate. However, C. testosteroni strains lack the genes associated with G6P dehydrogenase in the oxidative PP pathway (68, 69). Evolutionarily, the absence of the oxidative PP pathway is a more recent metabolic adaptation observed in various thermophilic organisms, archaea, and aerobic bacteria (70). This absence results in NADPH production solely from the TCA cycle, which is insufficient to support anabolism in rapidly proliferating cells (71). Previous studies have proposed that C. testosteroni may leverage transhydrogenase conversion of NADH to NADPH to meet the required NADPH flux (71). Our findings illustrated that, under SA stress, the increased FAO-dependent NADPH production could also serve as compensation for the deficit in NADPH production within the oxidative PP pathway, fulfilling the biosynthetic needs of proliferating cells exposed to low-dose SAs.

Accordingly, the enhanced FAO process decreases the content of intracellular fatty acids (Fig. 2d). Accumulating evidence emphasizes the pivotal role of fatty acids in molding cell size, mainly through their involvement in phospholipid synthesis (72, 73). For instance, reducing fatty acid availability has been shown to decrease the size of various microorganisms, including gram-negative E. coli, gram-positive Bacillus subtilis, and unicellular Saccharomyces cerevisiae (66, 72, 74). Our findings support the idea that cell size hinges on the cell envelope’s capacity, which is tightly linked to fatty acid availability. As fatty acid availability decreases under low-dose SA stress, cells adopt an ovoid shape with the lowest surface-to-volume ratio. Simultaneously, cell size is coordinated with other biosynthetic processes to preserve cell envelope integrity.

Mechanically, the induction of FAO relied on the QS transcriptional regulator LuxR solo, whose activation was triggered by AHL-like signals such as SAs. Our analysis of deletion and complementation mutants strongly supported the essential role of LuxR solo in enabling cells exposed to SA to detect stressful environments and respond by initiating a switch to fatty acid metabolism (Fig. 3). SAs might influence LuxR solo, possibly via the sulfonyl group, hinting at a potential association between them. Following this, the transcription of degradative thiolase (3-ketoacyl-CoA thiolase, EC2.3.1.16) activated, which catalyzed the final step of the FAO cycle, leading to the regeneration of acyl-CoA for continued β-oxidation and the release of acetyl-CoA for the citric acid cycle (75). This activation of FAO promoted cell proliferation. However, the heightened aerobic metabolism also generated reactive oxygen species from NADH, which can disrupt cellular redox homeostasis. To counterbalance this, the cytoplasmic pathway for nitrate (NO₃⁻) and nitrite (NO₂⁻) reduction, involving periplasmic reductases Nap and Nrf, is concurrently enhanced in proliferating cells. This periplasmic pathway served as an effective electron sink to maintain the redox state by facilitating electron transfer from the quinone pool to NapABC or cytochrome bc₁ (76). In summary, low-dose SA stress triggers intrinsic and extrinsic molecular mechanisms that modify core cellular metabolic processes to support the three basic needs of proliferating cells, i.e., rapid ATP generation to maintain energy status, increased biosynthesis of macromolecules, and tightened maintenance of appropriate cellular redox status. Conversely, high-dose SA stress exhibits antimetabolite effects, causing cell death by inhibiting dihydrofolate reductase (35). The hormetic response, therefore, is governed by partial versus complete inhibition of the target and associated consequences. Partial inhibition allows bacteria to reorganize metabolic pathways (i.e., enhanced FAO pathway) to mount a counterattack. In contrast, complete or near-complete inhibition drains resources, leading to cell death (Fig. 5).

Fig 5.

Proposed regulatory pathway of C. testosteroni hormesis under SAs. At low exposure, SAs potentially bind to LuxR solo and cause accumulation of LuxR solo–SA complexes. Accumulated LuxR solo–SA complexes then activate fatty acid β-oxidation in low dose–exposed bacteria, thus exhibiting hormesis of cell growth. As doses increase at high exposure, more SAs bind to dihydropteroate synthase and inhibit folate biosynthesis. When inhibition surpasses stimulation, toxicity is observed.

Furthermore, this regulatory pathway demonstrates a high degree of conservation across organisms within the Comamonadaceae family, which includes notable denitrifiers like Comamonas, Acidovorax, and Variovorax (77). Importantly, a substantial portion of species within the Comamonadaceae family, as listed in the NCBI taxonomy database, possesses both the LuxR solo transcription factor and the 3-ketoacyl-CoA thiolase receptor identified in our study (Fig. 6a; Fig. S16). To confirm the universality of this regulatory model, we randomly selected four strains from the latter two genera, namely, Acidovorax avenae, Acidovorax delafieldii, Variovorax paradoxus, and Variovorax soli, and cultured them under SAs-induced stress conditions. All four strains exhibited significant hormetic effects on cell growth when exposed to SAs (Fig. 6b), confirming the potential wide applicability of this regulatory model within the Comamonadaceae family. Overall, our findings provide novel mechanistic insight into the genes and enzymes involved in the hormetic effect of antibiotics in bacteria. Importantly, the inherent ability of Comamonas species to survive in ecological niches and the human gut makes them formidable candidates to cause mild but persistent infections (78), especially in individuals with predisposing conditions. Hence, comprehending the intricacies of these adaptive mechanisms is essential for the development of effective strategies to both prevent and manage such infections.

Fig 6.

LuxR solo is conserved in the Comamonadaceae family, including the genera Comamonas, Acidovorax, and Variovorax (a). Numbers represent species counts in each genus that contain the LuxR solo protein. (b) Hormetic effects on cell growth in Variovorax paradoxus, Variovorax soli, Acidovorax avenae, and Acidovorax delafieldii. OD₆₀₀ was measured after 24 h of cell growth in a 96-well microplate containing MSM, stains, and different concentrations of sulfamethoxazole and sulfadiazine mixtures (1:1 in concentration); mean and standard deviation of each result were calculated from 12 replicates. The dashed line represents the average cell density without SA treatment. The Wilcox test was used to evaluate significant differences in cell density between cultures with/without SA treatment. Significance (*P < 0.05) was determined using the Wilcox test.

Limitations of the study

Although we identified LuxR solo induction of FAO as a major mechanism underlying Comamonas hormesis, other pathways may also exist given that LuxR homologs can crosstalk with other cytoprotective TFs to mediate group behavior (79). Therefore, further work is needed to explore the potential link between LuxR solo and other TFs in Comamonas hormesis.

MATERIALS AND METHODS

Bacterial strains and culture

Bacterial strain growth experiments were conducted using microplate reader kinetic assays. For C. testosteroni CNB-2, overnight cultures were inoculated in 50 mL of nutrient broth (NB) medium at 30°C/200 rpm. These cultures (OD₆₀₀ ≈ 0.1) were then washed twice with phosphate-buffered saline (PBS) buffer and inoculated at a 3% inoculation proportion (vol:vol) into 200 µL of mineral salt medium (MSM) (Table S6) containing potential hormesis inducers in 96-well plates. The potential inducers included 14 sulfonyl-containing compounds (structures and properties are listed in Table S7). The plates were sealed, and optical density was monitored using a Spark 10 M microplate reader (Tecan, Switzerland) at 600 nm (OD₆₀₀) with fast continuous shaking. Escherichia coli strains DH5αλpir, β2155, and WM3064 were used to construct the C. testosteroni CNB-2 mutants, and strain BL21(DE3) was used for protein expression. The E. coli cells were grown in sterile NB broth at 37°C/200 rpm using 25 µg/mL gentamycin and 50 µg/mL ampicillin for the corresponding mutants. All bacterial strains, gene deletion mutants, and plasmids used in this study are listed in Table 1.

TABLE 1.

Strains and plasmids used in this study

Strain or plasmid	Relevant characteristics	Source
Comamonas testosteroni
CNB-2	Wild type
CNB-2/ΔLuxR	LuxR disrupted in CNB-2, GmR	This paper
CNB-2/ΔLuxR-pBBRLuxR	LuxR complemented in CNB-2/ΔLuxR, KmR	This paper
Variovorax paradoxus	Wild type
Variovorax soli	Wild type
Acidovorax avenae	Wild type
Acidovorax delafieldii	Wild type
Escherichia coli
DH5αλpir	Cloning vehicle: F- supE44 ΔlacU169 (φ80 lacZΔM15) hsdR17 recA1 endA1 gyrA96 thi-1 relA1	Sangon Biotech
β2155	Transconjugation donor: F‘ strA hsdS Δ(lacZ)M15 ΔdapA::erm pir::RP4(::kan from SM10)	Sangon Biotech
WM3064	Auxiliary bacteria for parental junction: thrB1004 pro thi rpsL hsdS lacZΔM15 RP4-1360 Δ(araBAD)567 ΔdapA1341::[erm pir]	Sangon Biotech
BL21(DE3)	Expression host: F- ompT r-B m-B; DE3 is a λ derivative carrying lacI and T7 RNA polymerase genes under placUV5 control	Sangon Biotech
Plasmids
pJQ200SK	Helper plasmid, GmR	Sangon Biotech
pCVD442	Suicide vector, GmR	Sangon Biotech
pCVD442-ΔLuxR::Gm	Vector pCVD442 containing Gm gene cassette-truncated LuxR gene with flanking sequences for generating mutant ΔLuxR	This paper
pBBR1MCS2	Broad host range of cloning vector, KmR	Sangon Biotech
pBBR1MCS2-LuxR	Complementation plasmid	This paper

The CellASIC ONIX Microfluidic Platform (Merck Millipore, Germany) was used to maintain cells growing in a monolayer to monitor growth behavior under SA stress. A cell inoculum (50 µL; OD₆₀₀ ≈ 0.1) was loaded onto a microfluidic plate B04A-03 at 4 psi for 15 s. PBS was then added at 1 psi for 30 s to remove non-trapped cells. Subsequently, 250 µL of MSM containing different concentrations of SAs was pipetted into the wells and incubated under a total pressure of 2 psi (flow rate of 10 µL/h) for 72 h. In all cases, images were taken using a MshOt microscope equipped with an MD3 digital camera (MshOt Microscopy Imaging Expert; Guangzhou, China) and analyzed using an MshOt Digital Microscope Imaging System v1.0. Time-lapse images were captured at 1 h intervals for 48 h. A bright-field image was acquired at each time point to evaluate growth behavior.

Construction of C. testosteroni CNB-2 mutants lacking LuxR

All primer pairs used for gene knockout are listed in Table 2. Upstream and downstream genes flanking LuxR were amplified by PCR using two primer sets, i.e., LuxR-5F/LuxR-5R and LuxR-3F/LuxR-3R. The gentamicin resistance gene (Gm), which confers resistance to gentamycin, was amplified from the plasmid pJQ200SK by PCR using the primer set Gm-F and Gm-R. The upstream fragment, Gm, and downstream fragment junction was amplified via fusion PCR and cloned into pCVD442, a suicide plasmid containing the ampicillin resistance gene, to obtain recombinant plasmids. Then, pCVD442 containing the LuxR gene was introduced into the E. coli strain β2155 via electroporation to obtain donor plasmids and introduced into the recipient CNB-2 strain via conjugation. Transconjugants were selected on NB plates supplemented with ampicillin (50 µg/mL) and gentamycin (25 µg/mL). The C. testosteroni CNB-2/ΔluxR mutants were confirmed by PCR and sequencing using two primer sets, i.e., LuxR-outF/LuxR-outR and LuxR-inF/LuxR-inR.

TABLE 2.

Primer pairs used in this study

Primers	Description	Source
Used in construction of mutants lacking LuxR
LuxR-5F	CTACTCGCTCAGCCAGTTCACGTC, 5′ LuxR homologous arm	This paper
LuxR-5R	GATTGCGATGCTCATCTCAGGCC	This paper
LuxR-3F	GAACGCACTGTGGAAAACCACCTG, 3′ LuxR homologous arm	This paper
LuxR-3R	CAAGATTCCGCACAGCCTGTTTGC	This paper
LuxR-GmF	GGCCTGAGATGAGCATCGCAATCagaaatgcctcgacttcgc, primer for GmR gene amplification	This paper
LuxR-GmF	CAGGTGGTTTTCCACAGTGCGTTC ttaggtggcggtacttggg	This paper
LuxR-outF	GCTGAACCATCGACTCCCGACAAGCAAC, external primer for mutant confirmation	This paper
LuxR-outR	CTCGGCGTTACTCGCCAGCCTCTAC	This paper
LuxR-inF	CTTCCAGCGCTCGAGACGATGCG, internal primer for mutant confirmation	This paper
LuxR-inR	CGAATACAGGCACTTGCAGACCGTGG	This paper
Used in construction of complementation mutants carrying LuxR
pBBR1-F	GTGAGTTAGCTCACTCATTAGGCAC, universal primer for positive clone identification	This paper
pBBR1-R	CACTCATCGCAGTCGGCCTATTG	This paper
LuxR-comF	ATCGATAAGCTTGATcaatgcgcttggcaagctcgcgg, primer for LuxR gene amplification	This paper
LuxR-comR	CTGCAGGAATTCGAT gatagtcaggcgacgatctctccgttgcgg	This paper
Used for qRT-PCR analysis
NapA-a	GAGCACCGCTCGTATGAGTT	This paper
NapA-b	ATTGCAGTGCTTGGAGACGA	This paper
NapD-a	AGGCGCCCAAATACATGGAA	This paper
NapD-b	CGTTGATTACGCCATCGAGC	This paper
NrfA-a	GGCTATGCGTGTTGATGTGC	This paper
NrfA-b	AGACCACGATGCTGAGTGTG	This paper
GluD-a	AGCATTTCACGGGCGAAGTA	This paper
GluD-b	ATGGCCGGCATGTACAAGAA	This paper
Fadi-a	GCAGAAGTTGGTGGACCTGT	This paper
Fadi-b	AAGACAGCGTGTATGCCTCC	This paper
Fadj-a	TCTCCAACCTGCCTTTCACC	This paper
Fadj-b	AGTCAGTGGATTCCAGCAGC	This paper
Fade-a	CTGCCAGGCGAATCATGTTG	This paper
Fade-b	TGATCAACTGCGCCGGTATT	This paper
Fadl-a	ATGCCGTACTTGAGCTGCTT	This paper
Fadl-b	TCCCTGGCTGTCATCAACAC	This paper
Pck-a	GATCTGGTCTTACGGCTCGG	This paper
Pck-b	CCACGTGGTACTTCTTGCCT	This paper
LuxR-a	CATGCCTTCGGTCATCGACA	This paper
LuxR-b	TCATCCTCGATCAGAACGGC	This paper
NirK-a	CACGCACGCTGAAATAAGGC	This paper
NirK-b	GCTGCACAACTGGACGATTG	This paper
Used for EMSA
M13F	TGTAAAACGACGGCCAGT	This paper
M13R	CAGGAAACAGCTATGACC	This paper

Complementation mutants carrying LuxR

All primer pairs used for gene complementation are listed in Table 2. To construct the mutant complemented with C. testosteroni LuxR, we amplified the C. testosteroni LuxR gene by PCR using LuxR-comF and LuxR-comR primers. The In-Fusion technique was used to clone the DNA fragment into the EcoRV site of the kanamycin-resistant plasmid vector pBBR1MCS2 to yield pBBR1MCS2-LuxR. The plasmid was then transformed into the E. coli strain WM3064 via electroporation and conjugation with the ΔluxR mutant. Transconjugants were selected on NB plates containing kanamycin (50µg/mL). The C. testosteroni CNB-2/ΔluxR-pBBRLuxR mutant was confirmed by PCR using the pBBR1-F and pBBR1-R primers.

RNA-seq analysis

RNA-seq was used to quantify transcriptional abundance of the C. testosteroni CNB-2 wild-type, ΔLuxR, and ΔluxR-pBBRLuxR strains under different SA stresses: (i) WT cells were cultured for 24 h at 30°C/200 rpm with/without SA treatment (50 µg/L and 1 mg/L, mixture of SMX and SD), three groups; (ii) ΔLuxR cells were cultured for 24 h at 30°C/200 rpm with/without SA treatment (50 µg/L, mixture of SMX and SD); and (ⅲ) ΔLuxR-pBBRLuxR cells were cultured for 24 h at 30°C/200 rpm without SA treatment (50 µg/L, mixture of SMX and SD). All cells were harvested during the logarithmic growth phase at 24 h, ensuring identical growth stages (Fig. S2 and S17). All strains were cultivated in suspension cultures overnight and conducted in triplicate. Total RNA was extracted using a Qiagen mini-RNA prep kit (Qiagen, Germany) following the manufacturer’s instructions. Extracted RNA was kept at −80°C before cDNA library construction. Total RNA concentration, RNA integrity number (RIN), and RNA quality number (RQN) were evaluated using an Agilent 2100 Bioanalyzer (Santa Clara, USA). Samples with an RIN/RQN value above 8.0 were collected for sequencing. Paired-end sequencing was performed on the Illumina HiSeq 2500 platform with a read length of 150 nucleotides (San Diego, CA, USA). The raw RNA-seq data were deposited in the NCBI GEO Short Read Archive (SRA) under accession number PRJNA933284. Sequencing reads were assembled and analyzed using the NCBI Prokaryotic Genome Annotation Pipeline with the reference CNB-2 strain genome (GenBank: CP001220.2). Data were normalized by calculating fragments per kilobase per million mapped fragments (FPKM). Significant changes in gene expression were defined based on ≥1-fold change in FPKM and FDR q-value <0.01.

Gene set enrichment analysis

Gene set enrichment analysis v4.0.1 (JAVA version) was obtained from the Gene Set Enrichment Analysis website (http://software.broadinstitute.org/gsea/downloads.jsp). Reference gene sets were acquired from curated Kyoto Encyclopedia of Genes and Genomes (KEGG) (c2) and gene ontology biological process (c5) data sets within the Molecular Signature Database (MSigDBv2.5, http://software.broadinstitute.org/gsea/msigdb/collections.jsp). The expression data set, phenotype class, and reference gene sets were loaded into the GSEA software. The analyses were performed on the SA-treated group (50 µg/L and 1 mg/L) versus the untreated group, with 1,000 permutations, using the default weighted enrichment statistical method. Gene sets with FDR q-values <0.05 were considered significantly enriched.

Quantitative real-time polymerase chain reaction

Under treatment with SMX (50 µg/L), SD (50 µg/L), and their mixture (50 µg/L, 1:1 concentration), the differential expression levels of nitrogen metabolism genes (NapA, NapD, NirD, NrfA, and GluD), fatty acid β-oxidation genes (Fadi, Fadj, Fade, and Fadl), biosynthesis gene Pck, and QS gene LuxR solo were validated using qRT-PCR. To perform reverse transcription analysis, RNA samples were used as templates to synthesize cDNA using a cDNA synthesis kit (Qiagen, Germany). The resultant cDNA was then used for qRT-PCR using the LightCycler 96 Real-Time PCR system (Roche Diagnostics, Switzerland). The relative expression level (copy number of mRNA transcript) of each target gene was normalized to the cDNA concentration and compared with control samples (no SA treatment). All primer pairs used for qRT-PCR analysis are listed in Table 2.

Targeted metabolomics analysis with LC-MS/MS

Fatty acids

The C. testosteroni WT and ΔluxR mutant cells (~10⁸) cultured with/without SA mixture at 24 h were homogenized with 300 µL of isopropanol/acetonitrile (1:1) containing mixed internal standards and centrifuged at 4°C/12,000 rpm for 10 min. The supernatant was then injected into the LC-MS/MS system for analysis. Ultra-high-performance liquid chromatography–tandem mass spectrometry (UHPLC-MS/MS) (ExionLC AD UHPLC-QTRAP 6500+, AB SCIEX Corp., Boston, MA, USA) was used to quantify fatty acids at Novogene Co., Ltd. (Beijing, China). Separation was performed on a Waters Acquity UPLC BEH C18 column (2.1 × 100 mm, 1.7 µm) maintained at 50°C. The mobile phase, consisting of 0.05% formic acid in water and isopropanol/acetonitrile (1:1), was delivered at a flow rate of 0.30 mL/min. The ratio of the concentration of the standard to the internal standard was used as the abscissa, and the ratio of the peak area of the bar to the internal standard was used as the ordinate to investigate standard solution linearity. The limit of quantification was determined by the signal-to-noise ratio (S/N), which compares the signal measured by the standard solution concentration with the blank matrix. All fatty acid substances were tested, and their categories are listed in Table S8.

Central carbon

The C. testosteroni WT and ΔluxR mutant cells (~10⁸) cultured with/without SA mixture at 24 h were individually ground with liquid nitrogen. The homogenate was resuspended in 500 µL of prechilled 80% methanol and 0.1% formic acid in a vortexing well. The samples were incubated on ice for 5 min, then centrifuged at 15,000 rpm and 4°C for 10 min. Aliquots of the supernatant were diluted to a final solution containing 53% methanol using LC-MS grade water. The samples were transferred to a fresh Eppendorf tube and centrifuged at 15,000 × g and 4°C for 20 min. Finally, the filtrate was injected into the UHPLC-MS/MS system for analysis. Each experimental sample was taken in equal volume and blended as quality control samples. The blank sample was a 60% methanol aqueous solution containing 0.1% formic acid instead of the experimental sample, and the pretreatment process was the same as the experimental sample.

For UHPLC-MS/MS analysis, a QTRAP 6500+ mass spectrometer was operated in the positive polarity mode with a curtain gas of 35 psi, collision gas of medium, ion spray voltage of 4 500 V, temperature of 550°C, ion source gas of 1:60, and ion source gas of 2:60. A negative ion mode QTRAP 6500+ mass spectrometer was operated in the negative polarity mode with a curtain gas of 35 psi, collision gas of medium, ion spray voltage of −4,500 V, temperature of 550°C, ion source gas of 1:60, and ion source gas of 2:60.

Based on the Novogene database, samples were detected using multiple reaction monitoring. The data files generated by UPLC-MS/MS were processed using SCIEX OS v1.4 to integrate and correct the peak. The main parameters were set as follows: minimum peak height, 500; signal/noise ratio, 5; and Gaussian smooth width, 1. The area of each peak represents the relative content of the corresponding substance. All central carbon substances were tested, and their categories are listed in Table S9.

Molecular docking and molecular dynamics simulation

The protease sequence of LuxR solo was downloaded from GenBank (accession no: WP_003076066.1) in FASTA format. To build the 3D model of the LuxR solo protease, the target sequence information was submitted to the SWISS-MODEL server (80) (http://swissmodel.expasy.org). Templates showing the highest quality were selected for model building. The predicted model output generated as a PDB file was downloaded for further analysis and visualized using SPDBV v4.10 (81). The structural coordinates of the potential protease activators were separated from the crystal structure of the SdiA protease in a complex with 3-oxo-C6-homoserine lactone (PDB ID: 482), available from the Protein Data Bank. Molecular docking simulations were used to explore the binding mode of the sulfonyl-containing compounds (Table S7) onto the 3D model of the LuxR solo protease using AUTODOCK tools v1.5.6 (82). Before docking, polar-H atoms were added to the LuxR solo model, and the macromolecule file was then saved in pdbqt format to be used for docking. The AutoGrid program generated ligand-centered maps with a grid dimension of 40 × 40 × 40. The Gridbox center was set to the x, y, and z coordinates −4.297, 17.693, and −27.541, respectively. Polar-H charges of the Gasteiger type were assigned, non-polar-H atoms were merged with the carbons, and internal degrees of freedom and torsions were set. Default settings were used for all other parameters. The PyMol package (83) was used to visualize the binding interactions between these ligands and the 3D LuxR solo protease.

Western blotting to assess LuxR solo solubility

LuxR solo gene was PCR-amplified and digested with Xbal/Xhol, with the fragment then introduced into Xbal/Xhol-treated pET-22b(+) to generate pET-22b(+)-LuxR. The sequence-validated plasmid was used to transform E. coli BL21(DE3) cells for expression. A sterile culture tube containing NB broth supplemented with ampicillin was inoculated with a single colony of E. coli BL21(DE3) cells carrying pET-22b(+)-LuxR. Overnight cultures were inoculated into 100 mL of fresh medium and grown at 37°C/200 rpm to an OD₆₀₀ of ∼0.6. Isopropyl-β-D-thiogalactoside (0.5 mM) was added to the cultures, which were subsequently divided into four 10 mL aliquots. Two aliquots received 75 µM SMX or SD, and two received an equivalent volume of dimethyl sulfoxide (0.1% vol/vol). After shaking at 20°C/180 rpm for 20 h, the cells were harvested by centrifugation (4,000 × g, 10 min, 4°C), resuspended in 2 mL of PBS buffer, and disrupted using an ultrasonic cell crusher (JY88-II). For protein solubility assessment, total cell lysates were separated into soluble and pellet fractions by ultra-centrifugation at 4°C/45,000 × g for 30 min. Protein samples from the soluble and pellet fractions were fractionated by sodium dodecyl sulfate–polyacrylamide gel electrophoresis (SDS-PAGE) and western blotting.

Identification of lux-box

To determine the presence of lux-box in specific promoters of FAO genes, upstream sequences were retrieved using RSAT tools (84), and promoter regions were identified using BPROM (85). Twenty base pairs of palindromic sequences in the promoters were then identified using the motif discovery tool of MEME (59). Identified sequences were then aligned with known lux-box sequences.

Purification of LuxR solo protein

The purification of LuxR solo was carried out at 4°C using AKTA pure 25 M1 (Cytiva). The SMX-induced E. coli BL21(DE3) pET-22b(+)-LuxR cell pellet was resuspended in lysis buffer (50 mM Tris, 300 mM NaCl, 0.1% Triton X-100, 0.2 mM phenylmethylsulfonyl fluoride, pH 8.0). Subsequently, the suspension was stirred for 30 min and sonicated on ice for 2 min in 15-s on/15-s off cycles at 30% power. The sonication cycle was repeated after the suspension rested on ice for 5 min, after which cell debris was pelleted via centrifugation (8,000 × g, 20 min, 4°C). The crude extract was then loaded onto a nickel metal affinity column (5 mL) equilibrated with 5 CV binding buffer (50 mM Tris, 300 mM NaCl, pH 8.0) and incubated for 1 h at room temperature. Finally, the protein was eluted with elution buffer (50 mM Tris, 300 mM NaCl, 200 mM imidazole, pH 8.0), and the effluent fraction was collected. His-LuxR was dialyzed into protein preservation buffer (50 mM Tris, 300 mM NaCl, 0.1% sarkosyl, 2 mM DTT, pH 8.0), concentrated after dialysis with PEG 20000, and filtered through a 0.45-µm membrane. The desired protein fractions were pooled, analyzed by SDS-PAGE and UV-vis spectroscopy, dispensed in 1 mL/tube, and stored at −80°C.

Electrophoretic mobility shift assay

For the preparation of fluorescent fluorescein amidite (FAM)-labeled probes, the promoter region of the pccdA-GFP plasmid was PCR-amplified with Dpx DNA polymerase from the plasmid 3-ketoacyl-CoA thiolase using the M13F/M13R primers listed in Table 2. The FAM-labeled probes were purified using the Wizard SV Gel and PCR Clean-Up System (Promega, USA) and quantified with a NanoDrop 2000C (Thermo, USA). EMSA was performed in a 20 µL reaction volume containing 50 ng of probe and various proteins in a reaction buffer of 50 mM Tris-HCl (pH 8.0), 100 mM KCl, 2.5 mM MgCl₂, 0.2 mM DTT, 2 µg of polydIdC, and 10% glycerol. After incubation for 30 min at room temperature, the reaction system was loaded into a 6% PAGE gel buffered with 0.5× tri-boric acid. Gels were scanned using an ImageQuant LAS 4000 mini (GE Healthcare).

Quantification and statistical analysis

All statistical analyses were performed in R. Analysis details can be found in figure legends and Results. Figures were prepared using the basic R package and ggplot2. Significance (*P < 0.05) was determined using the Wilcox test.

DATA AVAILABILITY

The raw RNA-seq data were deposited in the NCBI GEO Short Read Archive (SRA) under accession number PRJNA933284. Any additional information required to reanalyze the data reported in this paper is available from the lead contact upon request.

SUPPLEMENTAL MATERIAL

The following material is available online at https://doi.org/10.1128/aem.01662-23.

Supplemental figures and tables. aem.01662-23-s0001.docx.

Text S1, Fig. S1 to S17, and Tables S1 to S9.

ASM does not own the copyrights to Supplemental Material that may be linked to, or accessed through, an article. The authors have granted ASM a non-exclusive, world-wide license to publish the Supplemental Material files. Please contact the corresponding author directly for reuse.