Rewiring of cell morphology by Small protein B enhances stress tolerance and colonization in Aeromonas veronii under adverse conditions

Abstract

Pathogens often display morphological plasticity to withstand environmental stress and thrive in complex host immune environments. In this study, we demonstrated the morphological adaptation of Aeromonas veronii, a severe pathogen with a wide environmental distribution. Our results establish the critical role of Small protein B (SmpB) in morphological adaptation and reveal a conserved dual-safety regulatory mechanism mediated by ArgR. A. veronii exhibited morphological changes and gained enhanced stress resistance in response to environmental cues. We identified the trans-translation component SmpB as critical for this morphological adaptation, independent of its canonical role in trans-translation. Furthermore, SmpB transcriptionally up-regulates peptidoglycan biosynthesis genes. A convolutional neural network model predicted ArgR as a transcriptional regulator of smpB. Subsequent biochemical assays confirmed that ArgR directly bound to the smpB promoter and repressed its transcription by sequestering RNA polymerase. Moreover, the interaction between ArgR and SmpB promoted the affinity of ArgR for the smpB promoter. SmpB-mediated morphological rewiring enhanced A. veronii’s intestinal colonization and virulence in a mouse infection model. Collectively, our study reveals a novel mechanism in which SmpB, operating in a negative feedback loop with ArgR, modulates cell wall synthesis and enhances bacterial ecological fitness. These insights into host–pathogen interactions identify promising targets for innovative antimicrobial therapies and diagnostic strategies.

Graphical Abstract

Graphical Abstract.

Introduction

Bacteria exhibit remarkable morphological plasticity in response to environmental stress, enabling survival under adverse conditions. For instance, cold and starvation conditions induce Mycobacterium tuberculosis, Vibrio alginolyticus, and Vibrio cholerae to transition into a viable but non-culturable state characterized by altered morphology that enhances bacterial resistance to stress and contributes to long-term survival [1–3]. Similarly, bacteria alter their morphology in response to external compounds. The absence of phosphate, cysteine, or glutathione causes Streptomyces israelensis to elongate into branched or filamentous forms, while the addition of these compounds restores a regular rod-shaped morphology [4]. Biotin depletion in Corynebacterium diphtheriae results in the formation of irregularly sized, branching rod-shaped cells [5], while nutrient deprivation induces an elongated morphology in Pseudomonas aeruginosa, Pseudomonas putida, and Pseudomonas fluorescens in contrast to the short, rod-shaped cells observed on nutrient-rich media [6,7].

These morphological changes are not passive but rather are tightly regulated by bacterial genetic programs that allow dynamic physiological responses to adapt to environmental stresses, including nutrient limitation [8], DNA damage [9], and antibiotic exposure [10]. This plasticity is often reversible, enabling bacteria to sense and adapt to fluctuations in both external and internal environments. For example, uropathogenic Escherichia coli (UPEC) adopts a filamentous form to facilitate biofilm detachment [11], while Caulobacter crescentus adopts a filamentous morphology to escape biofilms and acquire available nutrients [12]. Such transformations hinge on the synthesis of peptidoglycan (PG), a key component of the bacterial cell wall that maintains cell shape, size, osmotic stability, and viability [13–15]. In Vibrio species, the transport of C55-P peptidoglycan precursors is essential for growth, maintaining cell shape in the intestine, and adapting to alkaline environments [16]. The assembly of PG in E. coli occurs in four distinct phases, involving the synthesis of lipid II and its integration into the existing PG layer through complex enzymatic reactions [17]. This complex process is frequently targeted by antibiotics. Bacteria, in turn, often bolster cell wall integrity as a survival strategy.

A key regulator of bacterial stress response is ArgR, a transcriptional repressor of arginine metabolism. ArgR senses environmental levels of arginine and other amino acids, influencing anaerobic adaptation, acid resistance, biofilm formation, and growth under unfavorable conditions [18–20]. Its broad role in nutritional and environmental adaptation highlights its regulatory importance. However, its connection to Small protein B (SmpB), implicated in bacterial stress response, remains unresolved. SmpB, an essential component of the trans-translation system, partners with transfer messenger RNA (tmRNA) to rescue stalled ribosomes, supporting bacterial growth and stress resistance [21]. Conventionally, SmpB’s contribution of morphological adaptation is attributed to this trans-translation function. Yet, whether SmpB operates through additional mechanisms and which aspects of morphological remodeling it affects remain largely unknown.

Bacterial morphological plasticity frequently contributes to enhanced survival under host immune pressure, environmental stress, and antibiotic exposure. To dissect these adaptive responses, we explored Aeromonas veronii, a zoonotic pathogen known for its morphological plasticity, stress resistance, and multi-drug resistance [22,23]. A. veronii can infect hosts through type III and type VI secretion systems and expression of diverse virulence factors, causing gastroenteritis, septicemia, and wound infections, posing a significant threat to public health [24–27]. Our results demonstrate that under stress conditions, A. veronii undergoes pronounced morphological shifts, including reduced growth rates, altered cell shapes, and enhanced resistance to various stressors. Crucially, we identified that SmpB controls cell morphology by regulating PG synthesis, thereby affecting intestinal colonization. We further uncovered that ArgR functions as an inhibitor of SmpB, fine-tuning its expression through a novel dual-level regulatory mechanism. Mouse infection experiments further confirmed the importance of SmpB in driving morphological adaptation, which is pivotal for colonization and virulence. By elucidating the trans-translation-independent function of SmpB in morphological adaptation and its regulation by ArgR, this study unveils a mechanism by which pathogens sense environmental cues, remodel their morphology, and promote infection. These findings reveal novel host–pathogen interactions, with bacterial morphological remodeling serving as a biomarker for infection dynamics and drug resistance, and its regulatory mechanisms offering promising targets for new antimicrobial therapies and diagnostics.

Materials and methods

Bacteria strains and plasmids

The strains, plasmids, and primers used in this study are listed in Supplementary Tables S3, S4, and S5, respectively. The plasmids were constructed using conventional molecular biology techniques, including PCR, DNA cleavage, ligation, and transformation, followed by positive colony selection [28]. Endonucleases and DNA polymerase were purchased from New England Biolabs, and the DNA purification kit, oligonucleotide synthesis, and DNA sequencing services were provided by Sangon Biotech, China. To create a series of site-directed mutants in pBT-SmpB and pTRG-ArgR, mutagenesis was performed with the primers listed in Supplementary Table S5 according to the manufacturer's instructions for the QuickChange Kit (Qiagen, Shenzhen, China). Mutants of A. veronii C4 were generated using allelic exchange with pRE112 [29]. Briefly, the flanking regions of each target gene were amplified using primer pairs and cloned into pRE112. The resulting plasmids were transformed into E. coli WM3064, and positive colonies were selected on LB agar containing chloramphenicol and diaminopimelic acid. After E. coli WM3064 clones harboring pRE112 derivatives were conjugated with A. veronii C4, the target DNA fragments were excised, and the knockout strains were screened on LB plates supplemented with 6% sucrose.

Medium and growth conditions

Bacteria were cultured in LB broth or on LB agar plates at 37°C. A final concentration of 0.1 mM IPTG (isopropyl-β-d-1-thiogalactopyranoside) was added to induce Lac or T7 promoters. Antibiotics were used at the following concentrations: 25 μg/ml chloramphenicol, 100 μg/ml ampicillin, 25 μg/ml kanamycin, 20 μg/ml tetracycline, and 20 μg/ml terramycin, as indicated in specific experiments. The culture was incubated at 37°C with agitation overnight and then inoculated into fresh medium until the exponential or stationary phase.

Growth measurement and fluorescent analysis

Bacterial growth was monitored by measuring the optical density (OD) at 600 nm of the culture using a spectrophotometer (Shanghai Spectrum Instruments Co., Ltd, Shanghai, China) at the indicated time points. The experimental strains carried a green fluorescent protein (GFP) tag via the plasmid pBBR-PsmpB-GFP. The total fluorescence was measured by analyzing the aliquots with a fluorescent plate reader (Infinite® 200 PRO, Tecan, Shanghai, China), with excitation and emission wavelengths set at 485 nm and 525 nm, respectively. Each treatment was performed in triplicate using separate cultures, and the fluorescence values were normalized to the OD.

Bacterial stress resistance assay

Bacterial strains were cultured in M9 medium to the stationary phase, adjusted to an OD₆₀₀ of 1.0, centrifuged to remove the medium, and resuspended in phosphate-buffered saline (PBS). To evaluate stress resistance, cells were exposed to sodium dodecylsulfate (SDS), lysozyme, or bile salts for 1 h, or to 2 M NaCl to induce osmotic stress for 2 h. After treatment, cells were centrifuged to remove the stress agents and resuspended in 300 μl of PBS. The resulting bacterial suspensions were serially diluted 10-fold for six steps, and 2 μl of each dilution was spotted onto agar plates to quantify viable cells. Untreated cultures served as controls. In parallel, both treated and control cells were stained with 10 μg/ml propidium iodide (PI) for 1 h and analyzed by flow cytometry to determine the proportion of dead cells. Bacterial stress resistance was assessed by combining spot dilution assay results with flow cytometry-based dead cell quantification.

Mouse acute virulence and colonization assay

Four-week-old KM mice were purchased from Hunan SJA Laboratory Animal Co., Ltd (Changsha, Hunan, China). Each mouse was orally administered with 200 μl of mixed antibiotics (ampicillin 1 mg/ml, kanamycin 0.5 mg/ml, neomycin 1 mg/ml, metronidazole 1 mg/ml, and gentamicin 1 mg/ml) once a day without food for 2 days before the bacterial infection. After 2 days, the bacterial suspension was administered orally to each mouse at a dosage of 2.5 × 10⁸ colony-forming units (CFU)/g body weight in 200 μl of PBS, and the mice were fed. After 48 h, the mice were euthanized, and the heart, liver, spleen, lung, kidney, small intestine, cecum, and colon were collected. One part was fixed in fixative solution and sliced, while the other was weighed and homogenized for plate counting. For the acute toxicity assay, mice were intraperitoneally injected with 10⁵ CFU/g body weight. After 12 h, the lungs, liver, kidneys, and spleen of the mice were collected for HE staining (hematoxylin–eosin staining), and the number of colonizing bacterial colonies was quantified. All the animal experiments were approved and conducted in accordance with the guidelines and recommendations from the ethical committee of Yunnan Agricultural University (HNUAUCC-2025-00 020).

Microscopy

A single bacterial colony was inoculated into M9 medium and cultured for 12 h to the stationary phase. A 1 ml aliquot of bacterial suspension was collected by centrifugation and washed twice with PBS. The bacterial cells were then treated with 500 μl of 2.5% glutaraldehyde solution at 4°C for 1 h. The cells were washed with PBS to remove residual glutaraldehyde and dehydrated with a gradient of ethanol concentrations (30, 50, 70, 80, 90, and 100%). The dehydrated cells were dropped onto a silicon wafer, dried, and sputter-coated with gold before imaging with a scanning electron microscope. Transmission electron microscopy (TEM) was performed essentially as described earlier [30]. For the Nile red staining experiment, bacteria were collected by centrifugation and fixed with 2.5% glutaraldehyde. The fixed cells were resuspended in 1 ml of PBS, and 10 μl of 0.1 mg/ml Nile red (YeaSen Biotechnology Co., Ltd) was added to stain for 10 min. Excess dye was removed by centrifugation, followed by PBS wash and resuspension. Fluorescence microscopy was performed using a Nikon ECLIPSE Ni-U microscope. Image overlay and analysis were performed using ImageJ software.

Flow cytometry assay

Single colonies of A. veronii C4 were inoculated overnight in M9 medium. Cultures were diluted to 10⁸ CFU/ml in PBS and exposed to 2 M NaCl, 0.2% SDS, 2% bile salts, and 0.5 mg/ml lysozyme, fosfomycin, or vancomycin. Treatment with 75% ethanol for 30 min served as the positive control, while untreated samples were the negative control. Cells were next stained with 10 μg/ml PI for 1 h, and analyzed using a CytoFLEX flow cytometer (Beckman Coulter), recording 100,000 events per sample. Gating strategies and parameters were established using positive and negative control groups, followed by analysis of the experimental groups. Dead cells were identified by a PE-A+ signal peak. The corresponding bar graph showed the quantification of PE-A+ dead cells. Data are shown as the mean ± standard deviation (SD) from three independent experiments. Statistical significance was determined by one-way analysis of variance (ANOVA) (P < 0.05). Data were acquired with the CytoFLEX software (Beckman Coulter) and processed with FlowJo software to quantify the proportion of dead cells under various stress conditions.

RNA-seq

Total RNA was extracted using a conventional method [31], followed by removal of rRNA using specific and biotinylated probes. The RNA was then purified and fragmented. Next, a cDNA library was constructed using the Illumina TruSeq Stranded Kit (Illumina, USA), incorporating both first and second strand synthesis steps. The cDNA was then adenylated, ligated to sequencing adaptors, and amplified to create the cDNA library. Libraries were prepared using the Illumina TruSeq Stranded mRNA Library Prep Kit without TEX nuclease treatment. Finally, the cDNA library was sequenced using the HiSeq platform (Illumina, USA). Sequencing was performed with a read length of 150 bp, generating an average of 2.31 Gb of data per sample. The average alignment rate to the reference genome was 89.25%, with clean reads showing average Q20 and Q30 values of 98.07% and 94.27%, respectively. High-throughput sequencing enabled detection of potential novel transcripts. For each sample, candidate regions were selected based on a length >100 bp, an average coverage depth >8×, and a minimum distance of 60 bp from adjacent annotated genes.

The raw sequences were filtered and cleaned with quality control pipelines from BGI Genomics to eliminate reads with low quality scores, adaptor contamination, and high ratios of unknown nucleotides (N). The resulting cleaned reads were aligned to the reference genomes using HISAT (version: v2.0.1-beta hisat2) and applied to new transcript prediction, assessment of single-nucleotide polymorphisms (SNPs) and insertions/deletions (INDELs), and operon annotation. The RNA transcripts were classified as either coding or non-coding based on their predicted encoding status. Raw read counts were normalized using DESeq, and statistical significance was assessed based on a negative binomial distribution model. P-values were calculated and adjusted for multiple testing using the Benjamini–Hochberg method to obtain false discovery rates (FDRs). Differentially expressed genes (DEGs) were identified by log2(fold change) >1 and padj < 0.05. The coding RNA was documented to the reference gene list using Bowtie2 to calculate gene expression levels, followed by hierarchical clustering (R software hclust function) and functional enrichment analysis (OmicShare tools).

Peptidoglycan purification and analysis

PG samples were prepared and analyzed in triplicate as described [32]. Bacteria were grown to the exponential phase in M9 medium and then harvested, followed by heat treatment in 10% SDS solution for 30 min. Sacculi (the PG-containing shell remaining after bacterial cell lysis) were repeatedly washed with MilliQ water by centrifugation (12 000 rpm, 20 min, 20°C) until total removal of SDS, followed by digestion with protease E (100 mg/ml) for 1 h at 60°C. Finally, samples were treated with lysozyme (100 mg/ml) for 16 h at 37°C. Coagulated proteins were removed by centrifugation (10 min, 12 000 rpm). Sodium borohydride was added for reduction, and the pH of the samples was adjusted to pH 3–5 with orthophosphoric acid. Muropeptides were analyzed using an LCMS-IT-TOF (ion trap time-of-flight mass spectrometry–liquid chromatography) system (Shimadzu, Japan). Elution of muropeptides was detected at 205 nm. Muropeptides were separated at 45°C using a linear gradient from buffer A [0.1% (v/v) formic acid] to buffer B [0.1% (v/v) formic acid, 40% (v/v) acetonitrile], with a flow rate of 0.175 ml/min. The eluent was analyzed by LCMS-IT-TOF (Shimadzu, Japan). Muropeptides were detected at 205 nm and via MS. The total ion chromatogram (TIC) of the extracted PG samples was obtained using LCMS-IT-TOF. Relative total PG amounts were determined by comparing the total chromatogram intensities (total peak areas), normalized to equivalent cell numbers, and extracted using identical volumes. Peaks were identified based on mass-to-charge ratios (m/z) of PG components and referenced structural data. All the experiments were performed in triplicate, and statistical significance was evaluated using one-way ANOVA with a significance threshold of P < 0.05.

Convolutional neural network method in machine learning

A total of 6242 experimentally validated interactions between transcription factors (TFs) and DNA in bacteria were extracted from the CollecTF database [33]. From this, a random set of 5500 interactions was selected as the training set using 10-fold cross-validation, while 742 interactions were reserved as an independent test set 1. This process was repeated eight times to minimize bias. In addition, the TF ArgR and its 62 binding sequences were designated as an independent test set 2 [34]. Recently, the application of convolutional neural networks (CNNs) has gained popularity for analyzing DNA sequences due to their effectiveness in capturing local sequence patterns [35]. In this study, a CNN was employed to construct a multi-label classification model, with the aim of predicting the binding of TFs to specific DNA sequences and their corresponding binding probabilities. The CNN model consisted of two hidden layers, each with 256 neurons, and was trained over 1000 epochs. To prevent overfitting, dropout was applied. The input was a matrix with a length of L representing the sequence. Each sequence was composed of 16 combinations of two adjacent bases, such as AA, AT, AG, and AC, as well as four possibilities for the remaining single base, i.e., A, T, G, or C. Each combination of two adjacent bases or the remaining single base was represented as one of the 20 one-hot vectors. For example, for a sequence ATCCA, AT = [0,1,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0, 0,0] T, CC = [0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,1,0,0,0,0]T, and A = [0,0,0,0,0,0,0, 0,0,0,0,0,0,0, 0,0, 1,0,0,0] T. The prediction tool and analysis scripts developed in this study have been made open source and are available on Zenodo.

Quantitative real-time PCR analysis (qRT-PCR)

Bacterial cultures were grown in M9 or LB medium to the stationary phase before harvesting for subsequent experiments. For experiments investigating the effect of arginine on the expression levels of argR and smpB, cultures were grown in M9 minimal medium until the logarithmic phase. Then, 10 mM arginine was added, and samples were collected at different time points for analysis. Total RNA was extracted using the TRIzol reagent according to the manufacturer's instructions. After removing genomic DNA using DNase I treatment (Wazyme, Nanjing, China), RNA quality was assessed by agarose gel electrophoresis. A total of 1000 ng of RNA was reverse transcribed into cDNA in a 10 μl reaction system using the ReverTra Ace qPCR RT Kit (Toyobo, Shanghai, China), following the manufacturer's instructions. The resulting cDNA was appropriately diluted and used for RT-qPCR analysis.The qPCR was performed on an ABIPrism® 7300 instrument (ABI, New York, NY, USA) for fluorescent detection using SYBRR® Green Realtime PCR Master Mix (Toyonbo, Shanghai, China). The qPCR primers are listed in Supplementary Table S4. The relative amounts of RNAs were calculated using the comparative Ct method with gyrB RNA as the reference gene [36].

Bacterial one-hybrid and two-hybrid assays

The E. coli XL1-Blue MRF' strain was chosen for the routine propagation of all recombinant plasmids. For bacterial one-hybrid assay, a pair of pBXcmT and pTRG derivatives was employed [37]. Similarly, for the two-hybrid assay, the conjugated plasmids, pBT and pTRG derivatives were used. First, a suitable concentration of 3-amino-1,2,4-triazole (3-AT) that does not cause self-activation was determined using positive and negative control strains. The experimental strains were then cultured to the stationary phase, and equal amounts of bacterial suspensions were serially diluted and spotted onto plates containing 3-AT. Plates without 3-AT were used as negative controls. All plates were incubated in the dark at 30°C for 2 days. Growth on 3-AT-containing plates indicates a positive interaction between the tested elements.

Electrophoretic mobility shift assays (EMSAs)

Probes for each sequence were generated using Cy5-labeled primers and A. veronii C4 genomic DNA as the template [38]. For the EMSA, 40 ng or a gradient concentration of purified ArgR proteins were incubated with probes in a 20 μl reaction system containing binding buffer [PBS, 0.1 M KCl, 2.5 mM MgCl2, 0.2 mM dithiothreitol (DTT), 10% glycerol, and 10 mM arginine] at 37°C for 1 h. The samples were then separated by electrophoresis [100 V, 1.5 h, 4.5% native polyacrylamide gel electrophoresis (PAGE)] and imaged with a multifunctional Biomolecule imager (Cytiva, USA).

Microscale thermophoresis (MST)

The MST assay was performed using purified enhanced GFP (eGFP)-tagged ArgR protein as the fluorescence molecule and PCR-amplified smpB promoter as the ligand in the MST reaction buffer (2 mM KH₂PO₄, 8 mM Na₂HPO₄, 350 mM NaCl, 0.05% Tween-20, and 10 mM arginine, pH 7.4). Approximately 10 μl of continuously diluted ligands were then incubated with a volume of 10 nM fluorescence molecules at 30°C for 1 h. The reaction was loaded onto MONT.115 capillary tube plates (Nanotemper Technologies, Germany), which were measured with 30% LED excitation and medium MST power. The binding affinity (dissociation factor, K_d) was calculated with the MO. Affinity Analysis software (Nanotemper Technologies, Germany).

GST pull-down and western blotting

The whole-cell lysate of E. coli BL21 (DE3), co-transformed with pET-SmpB and pGEX-3X-ArgR, was loaded onto 2 ml of settled glutathione S-transferase (GST)–Sepharose resin (Sangon Biotech, China) and incubated for 1 h at 4°C. After washing the beads three times with 10 ml of wash buffer (50 mM Tris–HCl, pH 8.0), 1 ml of elution buffer (10 mM reduced l-glutathione, 50 mM Tris–HCl, pH 8.0) was loaded to elute the samples. In the meantime, the co-transformant containing both pET-SmpB and pGEX-3X was selected as the negative control. Eluates of 10 μl were subjected to 15% denaturing SDS–PAGE for western blotting. The gel was then transferred onto a polyvinylidene difluoride (PVDF) membrane with a pore size of 0.45 μm using the Trans-Blot Turbo Transfer System (BioRad, USA) and processed according to the standard protocols of Molecular Cloning [39]. The primary antibody, anti-His Tag antibody, and the secondary antibody, goat anti-mouse IgG and horseradish peroxidase (HRP) conjugate, were purchased from Boster Biological Technology, China. TMB Chromogen Solution for blotting was provided by Beyotime Biotechnology, China.

Confocal microscopy and FRET analyses

The microscopic and fluorescence resonance energy transfer (FRET) analyses were performed in accordance with the protocols [40]. To measure cyan fluorescent protein (CFP) and FRET, cells were excited with a 435 nm laser, and fluorescence was collected in the CFP channel with a standard 480/10 filter, while the FRET signal was measured with a 525/20 filter. To measure yellow fluorescent protein (YFP), cells were excited with a 505 nm laser, and the emission was measured with a 525/20 filter. FRET measurements were carried out in late exponential phase growing E. coli cells after one washing step with PBS buffer. A 200 μl aliquot of cell suspension was loaded into a 96-well microtiter plate and subsequently measured in a SYNERGY microplate reader. The cells were excited at 435 nm, and spectra were recorded from 465 nm to 600 nm. Emission peaks of cyan fluorescence were observed at 480 nm, while peaks of yellow fluorescence were detected at 525 nm. FRET values (RCY) were calculated as ratios from the emission maxima of CFP (480 nm) and YFP (525 nm).

Preparation of RNA polymerase (RNAP)

The purification protocol was adapted from Tang et al. [41]. The core RNAP contained α, β, and β′ subunits. E. coil BL21(DE3) strains, transformed with pET-28a-α or pET-28a-σ³², were cultured in LB broth supplemented with kanamycin at 37°C until they reached an OD₆₀₀ of 0.6. Induction was initiated by the addition of IPTG to a final concentration of 0.1 mM, followed by further incubation at 30°C for 5 h with agitation. Cells were harvested and resuspended in buffer A (20 mM Tris pH 7.9), and lysed by sonication. Crude His-tagged α and σ32 were absorbed on Ni²⁺-NTA agarose (Wazyme, Nanjing, China), and eluted in buffer A containing 350 mM imidazole. Inclusion bodies containing crude β and β′ were isolated and washed from induced E. coil BL21(DE3) strains of transformed plasmid pET-28a-β and pET-28a-β′, respectively. The inclusion bodies were dissolved in solubilization buffer (Wanleibio, Shenyang, China). The core RNAP contained 40 μg (∼1 nM) of hexahistidine-tagged α, 150 μg (∼1 nM) of crude β, and 300 μg (∼2 nM) of crude β′ in buffer (20 mM Tris–HCl pH 7.9, 100 mM KCl, 1 mM EDTA). Following dialysis, the post-reconstitution purification of RNAP was performed using metal ion affinity chromatography. The separately purified core enzyme complex will be mixed with σ32 to obtain RNAP holoenzyme.

Bioinformatics analysis

Alignment of amino acid sequences of ArgR and SmpB among different strains was carried out using ESPript 3.0 (http://espript.ibcp.fr/ESPript/cgi-bin/ESPript.cgi) [42]. Conserved sites of ArgR and SmpB were analyzed using WebLogo3 [43]. The promoter sequence alignment between different strains was performed using T-COFFEE [44].

Statistical analysis

The presented data were obtained from a minimum of three independent experiments and were expressed as means ± SD. Statistical significance was determined with the Student's t-test, where P < 0.05 and P < 0.01 were represented as significant and extremely significant differences, respectively.

Results

SmpB is required for morphological plasticity and conditional fitness in A. veronii

A. veronii demonstrates remarkable adaptability, enabling it to compete for ecological niches and better withstand external challenges. To investigate the general mechanism underlying morphological plasticity, we established a model system using A. veronii. We observed that A. veronii exhibited distinct morphologies depending on nutrient availability (Fig. 1A, B). In nutrient-rich LB medium, A. veronii displayed a short rod shape (1–1.5 μm) and grew rapidly. In contrast, in nutrient-poor M9 medium, they exhibited a slower growth rate (Supplementary Fig. S1A), increased cell length, and enhanced resistance to high osmotic pressure (NaCl), SDS, bile salts, and lysozyme (Fig. 1A–C). These findings suggest that nutrient availability significantly influences bacterial morphology.

Figure 1.

SmpB is required for morphological alterations and enhances environmental stress resistance in A. veronii under low-nutrient conditions. (A) Cell morphology of A. veronii under the indicated culture conditions was observed using fluorescence microscopy after Nile red staining. Bacterial cultures were grown in nutrient-rich LB medium and nutrient-limited M9 medium until reaching the stationary phase, with an OD₆₀₀ between 2.0 and 2.2. Scale bars represent 5 μm. (B) Scanning electron microscopy (SEM) images depicting the morphologies of A. veronii cultured in LB or M9. Scale bars represent 3 μm. Bacterial lengths in LB and M9 medium were quantified using Image J. n = 200. (C) Stress resistance assays were performed with serially diluted A. veronii pre-cultured in M9 medium. The selected stress conditions included 2 M NaCl (high osmotic pressure), 2% bile salt, 0.2% SDS, and 0.5 mg/ml lysozyme. (D) Cell morphology of A. veronii derivatives cultured in M9 medium was observed using fluorescence microscopy after Nile red staining. Scale bars represent 5 μm. (E) SEM images illustrate the morphologies of A. veronii derivatives with the indicated genotypes grown in M9 medium. Scale bars represent 1 μm. Bacterial lengths were measured using ImageJ. n = 200. (F–  I) Death rates of the indicated strains pre-cultured in M9 medium were assessed under the same stress conditions as described in (C), as determined by PI staining followed by flow cytometry analysis. The mean fluorescence intensity (MFI) was calculated after gating based on the positive control. The data are presented as means, and P-values were determined using an unpaired two-tailed t-test, with statistical significance denoted as ***P < 0.001, **P < 0.01 in (B), (E), and (F–I).

Moreover, deletion of smpB in A. veronii impaired this morphological adaptation (Fig. 1D, E). The ΔsmpB strain exhibited reduced growth in M9 medium and failed to elongate (Fig. 1D, E), while its morphology and size in LB medium were comparable with those of the wild type (WT) (Supplementary Fig. S1B, C), indicating that smpB is specifically required for morphological adaptability. Consistently, the ΔsmpB mutant also revealed increased sensitivity to high osmotic pressure, SDS, bile salts, and lysozyme (Fig. 1F–I; Supplementary Fig. S1D–F). Complementation of smpB in the ΔsmpB mutant restored the defects in M9-induced morphological elongation and the associated adaptive resistance to stressors such as SDS, bile salt, NaCl, and lysozyme. Together, these results demonstrate that SmpB plays a crucial role in environment-induced morphological adaptation and stress resistance of A. veronii.

SmpB regulates bacterial peptidoglycan synthesis independently of its trans-translation rescue

SmpB is traditionally recognized for its canonical role in trans-translation, where it rescues stalled ribosomes to preserve translation fidelity. To uncover whether this function contributes to the morphological changes observed in A. veronii under nutrient deprivation in M9 medium, we examined tmRNA, another indispensable component in the trans-translation complex. Unexpectedly, the ΔtmRNA mutant of A. veronii retained the capacity for adaptive elongation in M9 medium (Fig. 2A, B; Supplementary Fig. S2). This suggests that SmpB mediates morphological adaptation independently of its role in trans-translation, implying the involvement of an alternative regulatory mechanism.

Figure 2.

The regulatory effects of SmpB on PG synthesis are independent of trans-translation. (A and B) Morphological comparison between the WT and ΔtmRNA pre-cultured in M9 medium, observed via fluorescence microscopy (A) and TEM (left, B). Quantification of lengths between WT and ΔtmRNA strains grown in M9 medium using Image J (right, B). n = 200. (C) Venn diagram illustrating the number of DEGs identified in RNA-seq analysis comparing the WT versus ΔsmpB and ΔtmRNA. (D) Dot plot showing the log2(fold change) values of DEGs in comparisons with both the WT versus ΔsmpB and the WT versus ΔtmRNA. The P-value cut-off was set to 0.05. (E) Kyoto Encyclopedia of Genes and Genomes (KEGG) enrichment analysis of the DEGs specifically changed in the comparison between the WT and ΔsmpB (red dots in D). (F) Heat map displaying normalized count numbers of PG synthesis genes based on RNA-seq analysis. (G) Depiction of the PG synthesis pathway. Genes changed in the WT versus ΔsmpB are highlighted in yellow; green triangles represent significant down-regulation in ΔsmpB; red filled pentagons indicate significant up-regulation in ΔtmRNA, and open pentagons denote genes up-regulated but not reaching statistical significance in ΔtmRNA. (H) Quantification of purified bacterial PG using the LC-MS chromatogram. The bars on the right panel represent total labeled peaks (left panel) of the indicated strains. Analysis was performed with three biological replicates, and P-values were determined using an unpaired two-tailed t-test, where *P < 0.05 indicates statistical significance. (I) Assessment of the death rates of the indicated strains pre-cultured in M9 medium under the specified antibiotics, determined by PI staining followed by flow cytometry analysis. The mean fluorescence intensity (MFI) was calculated after gating based on the positive control. The data are presented as means, and P-values were determined using an unpaired two-tailed t-test, with statistical significance denoted as ***P < 0.001, **P < 0.01. (J) Quantification of cell wall thickness of the indicated strains through ultrathin sectioning using TEM. The yellow line represents the thickness of the PG layer.

To achieve unbiased insights into SmpB-dependent but tmRNA-independent transcriptional changes, we performed RNA-seq analysis on ΔsmpB, ΔtmRNA, and WT A. veronii cultured in M9 medium. A total of 338 DEGs were identified between the WT and ΔsmpB, and 434 DEGs between the WT and ΔtmRNA. Among these, 119 DEGs showed a similar trend of alteration in both the WT versus ΔsmpB and the WT versus ΔtmRNA comparisons, suggesting that these genes were transcriptionally regulated by trans-translation. In contrast, 219 DEGs were exclusively present in the comparison between the WT and ΔsmpB, with either unaltered or opposite changes in the ΔtmRNA strain (Fig. 2C, D; Supplementary Fig. S3). Notably, several genes involved in PG synthesis, such as murCDEFG, mraY, ftsW, and ftsI, were significantly down-regulated in ΔsmpB but not in ΔtmRNA mutants (Fig. 2E–G). RT-qPCR analysis also confirmed that deletion of smpB led to significant down-regulation of the aforementioned PG biosynthesis genes (Supplementary Fig. S3D). This led us to hypothesize that the morphological defects observed in ΔsmpB were attributed to attenuation of PG synthesis. Consistent with this, purified cell walls from ΔsmpB exhibited a marked decrease in PG abundance (Fig. 2H; Supplementary Fig. S4; and Supplementary Tables S1, S2). Given that proteins encoded by these PG synthesis genes are common targets of antibiotics, we interrogated whether loss of smpB compromised antibiotic resistance, especially those targeting cell wall components. Indeed, ΔsmpB displayed elevated sensitivity to vancomycin, cefotaxime, and d-cycloserine (Fig. 2I; Supplementary Fig. S5). Consistently, the ΔsmpB mutant showed a significant reduction in PG thickness, as observed via TEM (Fig. 2J). Importantly, the defects in antibiotic resistance, cell wall thickening, and PG synthesis were restored by complementation of smpB in ΔsmpB (Fig. 2H–J). These data suggest that smpB plays a crucial role in regulating PG synthesis independent of its trans-translation function, underpinning its essential involvement in morphological adaptation and antibiotic resistance.

Identification of regulatory factors binding the smpB promoter using a convolutional neural network approach

CNNs are computational learning architectures that implement linear convolutions followed by non-linearities to achieve advanced classifications and regressions in high-dimensional problems [45]. They have been extensively applied to predict DNA–protein binding, RNA–protein interaction, signal peptides, and other protein sorting signals [46]. In this study, we developed a CNN model to search for TFs binding to the smpB promoter (Fig. 3A). The performance of the model was evaluated using an independent test set 1, yielding an area under the curve (AUC) of 0.908, indicating high predictive accuracy (Fig. 3B). Furthermore, the model was applied to predict TF binding for 62 DNA sequences in independent test set 2. Of these, 61 sequences were correctly predicted to bind the ArgR TF, while only one sequence was incorrectly predicted to bind the OxyR TF, demonstrating the robustness of the model. The CNN model further identified the cis-acting sequence within the smpB promoter region with high binding probabilities for each TF. Among the candidates, ArgR exhibited the strongest binding affinity for the smpB promoter (Fig. 3C). To validate the in silico predictions, a 3-AT resistance-based one-hybrid assay was performed. The plasmids containing the smpB promoter were co-transformed with plasmids encoding individual TFs, including ArgR, OmpR, OxyR, and IscR. The results confirmed that ArgR exerted robust binding activity on the smpB promoter compared with OxyR and OmpR, in agreement with the predictions (Fig. 3D; Supplementary Fig. S6A).

Figure 3.

ArgR functions as a negative regulator of smpB transcription. (A) Schematic representation of the basic architecture of the CNN. (B) Evaluation of the performance of the CNN on the independent test set 1. (C) Probability assessment of each predicated TF interacting with smpB 5′ cis-acting sequences. (D) Experimental evaluation of the interaction between each predicated TF and the smpB promoter using bacterial one-hybrid assay, conducted in the absence or presence of 3-AT and streptomycin. Results of the bacterial one-hybrid assay are summarized. Further details are provided in the Materials and Methods and in Supplementary Fig. S6. (E) Determination of the interaction between ArgR and the smpB promoter using EMSA. The smpB promoter was Cy5-labeled, and purified ArgR protein was titrated in at the indicated concentrations. (F) Assessment of the dissociation factor (K_d) between ArgR and the smpB promoter via MST assay. Dose–response fits are shown on the graph, with the x-axis representing the concentration of the smpB promoter following serial dilution and the y-axis depicting the average Fnorm. n = 3 biological replicates. Data are shown as the mean ± SD. (G) Sequences of the smpB promoter region. ArgR-binding sites are highlighted in yellow, RNAP-binding sites are represented in red, regions -35 and -10 are underlined, and RBS sites are in blue. (H) AlphaFold3 prediction indicates that ArgR has a strong binding affinity for the smpB promoter. The characters marked in red represent the predicted interaction sites. (I) Transcriptional activity of the smpB promoter was negatively regulated by ArgR in an E. coli reporter strain, where eGFP expression was controlled by the smpB promoter. Further details are provided in the Materials and methods. (J) Quantification of smpB transcription in cultures grown in M9 medium using RT-qPCR. *P < 0.05; n.s, P > 0.05, indicating no statistical significance, as determined using a Student's t-test. (K) Evaluation of the competition between ArgR and RNAP in binding the smpB promoter using the EMSA method.

ArgR acts as a transcriptional repressor of the smpB gene

To confirm the direct binding of ArgR to the smpB promoter, we assessed interaction in vitro between the smpB promoter and the purified ArgR using an EMSA. The migration of the smpB promoter probes slowed progressively with increasing concentrations of purified ArgR, indicating a direct interaction between ArgR and the smpB promoter (Fig. 3E). Additionally, the binding affinity between ArgR and the smpB promoter was quantified using the MST assay, yielding a K_d of 257.3 ± 5.96 nM (Fig. 3F). By searching the consensus sequence for ArgR recognition, we identified two conserved motifs within the promoter region. Mutations in these motifs significantly reduced ArgR binding activity (Supplementary Fig. S6B, C). Promoter truncation of smpB revealed that the core region required for ArgR recruitment was located between −89 and −10 bp relative to the transcription start site (Fig. 3G). AlphaFold3 predictions further indicated that ArgR strongly interacted with the smpB promoter, with the ArgR N-terminal domain targeting the σ32-binding site (Fig. 3H). Mutation of this σ32 site dramatically impaired ArgR binding, as confirmed by MST assay (Supplementary Fig. S6C). These results suggested that ArgR directly binds to the smpB promoter and probably interferes with its transcription.

To evaluate the functional impact of ArgR on smpB expression, we developed a reporter system by fusing the smpB promoter to an eGFP gene, resulting in a transcriptional fusion plasmid PsmpB-eGFP. Co-expression of PsmpB-eGFP with ArgR resulted in significantly dampened fluorescence compared with the control strain with an empty vector, confirming that ArgR suppressed smpB expression (Fig. 3I). To delineate functional regions within ArgR that mediated smpB suppression, we constructed truncations of its N- and C-terminal domains: ArgRΔC81, with 81 amino acids deleted at the C-terminus; and ArgRΔN81, with 81 amino acids trimmed at the N-terminus. Both truncated forms of ArgR displayed suppression of PsmpB-eGFP transcription, but the C-terminal truncation had a more pronounced effect than the N-terminal truncation (Fig. 3I). This suggests that intact ArgR is required for transcriptional repression of smpB, and the C-terminus of ArgR seems to be more dispensable. To validate the role of ArgR in smpB regulation in vivo, we knocked out the argR gene in A. veronii via the allelic exchange method. In M9 medium, the relative mRNA abundance of smpB was significantly elevated in the ΔargR strain compared with the WT, whereas no difference was observed under nutrient-rich LB conditions (Fig. 3J; Supplementary Fig. S7). This discrepancy presumably implies the nutrient-dependent regulatory role of ArgR in smpB repression.

The complete RNAP is composed of the RNAP core enzyme and the σ factor that recognizes DNA-binding sequences [47]. We found a σ32-binding site on the smpB promoter. Interestingly, the ArgR-binding site overlapped with the RNAP-binding site within the smpB promoter (Fig. 3G, H). This implied that ArgR probably obstructed the accessibility of RNAP to the smpB promoter, consequently suppressing its expression. To test this possibility, we purified the core enzyme of RNAP and σ32, and subsequently examined their competition with ArgR for smpB promoter binding in vitro. First, RNAP complexed with σ32 bound to PsmpB. Upon titration of ArgR, the binding complex of RNAP-PsmpB disappeared. Instead, the band corresponding to ArgR-PsmpB emerged in EMSA, confirming the hindrance and replacement model in ArgR-mediated transcription down-regulation (Fig. 3K). This suggests that ArgR negatively regulates smpB transcription by impeding RNAP access to its binding site. Collectively, these results establish ArgR as a transcriptional repressor of smpB, acting by directly binding to the promoter and interfering with RNAP recruitment, particularly under nutrient-limited conditions.

Direct interaction between ArgR and SmpB enhances ArgR binding to the smpB promoter

The direct interaction between SmpB and ArgR was validated using a bacterial two-hybrid system in the E. coli XL1-Blue MRF′ reporter strain (Fig. 4A). This interaction was further corroborated in E. coli BL21 (DE3) cells expressing CFP-tagged SmpB and YFP-tagged ArgR, which co-localized with one another and exhibited a FRET signal, indicating their proximity and potential complex formation (Fig. 4B). Moreover, GST-tagged ArgR sufficiently pulled down SmpB in a GST pull-down assay, providing further biochemical evidence for the direct interaction between these two proteins (Fig. 4C). To delineate the key regions of SmpB necessary for its interaction with ArgR, we performed bacterial two-hybrid assays using full-length and truncated variants of smpB. Full-length SmpB efficiently interacted with full-length ArgR, while truncation mutants lacking either 34 amino acids at the N-terminus (SmpB-ΔN34) or 31 amino acids at the C-terminus (SmpB-ΔC31) failed to interact with the intact ArgR (Table 1; Supplementary Fig. S8A). Subsequent mutational analysis of conserved sites within these regions revealed that both G₁₃₃K and D₁₃₈KR mutations within SmpB were indispensable for its interaction with full-length ArgR (Table 1; Supplementary Fig. S8B, C), which was further supported by molecular docking analysis (Supplementary Fig. S8D). Likewise, the key regions and residues within ArgR critical for SmpB binding were similarly identified through bacterial two-hybrid assays. Mutations in the S₅₄R, D₁₂₁, and G₁₃₄ residues of ArgR disrupted the SmpB–ArgR interaction (Table 1; Supplementary Fig. S9A–C), as confirmed through docking simulations (Supplementary Fig. S9D). Consistent with these results, GST pull-down assays further validated the pivotal role of D₁₃₈KR residues of SmpB and the S₅₄R residue of ArgR in mediating this interaction (Fig. 4D). Subsequently, an interaction model integrating these data was constructed to illustrate the SmpB–ArgR binding mechanism (Fig. 4E).

Figure 4.

Identification of key residues of SmpB and ArgR protein interaction. (A) Detection of interaction between SmpB and ArgR at the protein level using an E. coli-based two-hybrid assay. The combinations of plasmids expressing ArgR or SmpB are shown on the right. (B) Validation of SmpB and ArgR interaction using the FRET method; see the Materials and Methods for details. (C and D) Validation of key amino acids required for ArgR–SmpB complex formation via GST pull-down assay. The GST pull-down and input samples were blotted using antibodies against 6×His tag or GST. The annotated amino acid point mutation was validated for binding activity with alanine. (E) Molecular docking model illustrating the interaction between ArgR and SmpB. Yellow markings indicate the binding sites of ArgR, while red markings indicate the binding sites of SmpB. (F) Enhancement of ArgR recruitment onto the smpB promoter by titration of purified SmpB protein, as demonstrated by EMSA.

Table 1.

Validation of ArgR–SmpB interaction sites utilizing the bacterial two-hybrid experiment

Plasmid	Plasmid	5 mM 3-AT	12 mM 3-AT	12 mM 3-AT + streptomycin
pTRG-ArgR	pBT-SmpB-ΔC30	+	–	–
	pBT-SmpB-ΔN34	+	–	–
	pBT-SmpB(GS11AA)	+	+	+
	pBT-SmpB(TI14AA)	+	+	+
	pBT-SmpB(FI26AA)	+	+	+
	pBT-SmpB(EAG32AAA)	+	+	+
	pBT-SmpB(GK133AA)	+	–	–
	pBT-SmpB(DKR138AAA)	+	–	–
	pBT-SmpB(KP152AA)	+	+	+
pBT-SmpB	pTRG-ArgR-ΔC81	+	–	–
	pTRG-ArgR-ΔN81	+	–	–
	pTRG-ArgR(Q50A)	+	+	+
	pTRG-ArgR(SR54AA)	+	–	–
	pTRG-ArgR(D121A)	+	–	–
	pTRG-ArgR(G134A)	+	–	–

Notably, CFP–SmpB and ArgR–YFP appeared to localize near the DNA of E. coli cells in the FRET assay (Fig. 4B), suggesting a possible role in transcriptional regulation. To assess the functional impact of SmpB on ArgR binding to the smpB promoter, we performed EMSA with purified SmpB and ArgR proteins. While SmpB alone showed no affinity for the smpB promoter, it enhanced the ArgR binding to the smpB promoter in a dose-dependent manner, suggesting a cooperative interaction (Fig. 4F). This observation implies a possible negative feedback loop, where SmpB enhances ArgR-mediated repression of smpB transcription, thereby enabling dynamic regulation of smpB expression while ArgR levels remain stable. In conclusion, these results provide compelling evidence for the direct interaction between SmpB and ArgR proteins, revealing that SmpB facilitates ArgR recruitment to its own promoter, potentially as part of a negative feedback regulatory mechanism.

ArgR plays a minor role in bacterial morphology and stress resistance

Since ArgR was identified as the regulator of smpB, we hypothesized that ArgR influenced bacterial morphology and stress through SmpB. Deletion of smpB resulted in altered cell morphology (Fig. 5A). The overexpression of argR in the ΔargR mutant led to noticeable morphological alterations, whereas such changes were not observed in ΔargR (Fig. 5A). Stress resistance experiments revealed that deletion of smpB significantly decreased the ability of the strain to withstand stress, whereas deletion or overexpression of argR had a minor impact on stress tolerance (Fig. 5B). Given that ArgR regulates arginine biosynthesis in an arginine concentration-dependent manner, we investigated the relationship between amino acid sensing and argR transcription. The supplementation of arginine in M9 medium, unlike other amino acids, modestly increased argR transcription while concurrently repressing smpB transcription (Fig. 5C). This effect mirrored the smpB suppression observed under LB medium conditions (Supplementary Fig. S10B). However, arginine supplementation in M9 medium did not alter bacterial morphology (Supplementary Fig. S11), despite suppressed smpB transcription. The extent of arginine-induced gene expression changes under these conditions may be too small to cause detectable morphological changes. Given these weak data, although arginine treatment may mimic smpB suppression observed in M9 medium, ArgR’s role in this pathway appears limited, suggesting that other nutrient-sensing mechanisms contribute to smpB inhibition. Although our results demonstrate that ArgR negatively regulates smpB, the regulatory relationship does not appear to be mediated through a simple linear signaling pathway. Instead, additional upstream factors likely act in concert with ArgR to regulate smpB expression. Identifying these regulatory components will be a key focus of future research.

Figure 5.

The role of ArgR in arginine sensing and regulation of smpB-mediated morphological adaptation. (A) Morphology analysis of the strains cultivated in M9 medium, observed via fluorescence microscopy. Scale bars represent 2.5 μm. (B) Stress resistance assays using serially diluted A. veronii with the indicated genotypes, pre-cultured in M9 medium. Selected stress conditions included 2 M NaCl, 0.5 mg/ml lysozyme, 0.2% SDS, cefotaxime, and d-cycloserine. (C) Transcription levels of argR and smpB in the WT strain were measured by RT-qPCR after titration of the indicated amino acids into M9 bacterial culture for 4, 8, and 12 h. P-values were determined using an unpaired two-tailed t-test, with statistical significance denoted as ***P < 0.001; **P < 0.01; ns, P > 0.05 in (C).

SmpB is essential for intestinal colonization and virulence of A. veronii

Morphological plasticity often influences virulence and transmission efficiency in pathogens. For example, in Campylobacter jejuni, the rod-shaped mutant strain exhibits a significant reduction in intestinal colonization in chickens, along with defects in motility and biofilm formation [48]. As A. veronii is an intestinal pathogen, we examined whether SmpB regulates its virulence and colonization capacity. As anticipated, loss of smpB (ΔsmpB) resulted in a severe growth defect and a markedly reduced intestinal colonization rate. Complementation of smpB in the ΔsmpB strain fully restored both the growth rate and colonization efficiency (Fig. 6A). To further validate the role of smpB in virulence, a mouse acute toxicity model was employed. Following infection with the WT strain, histological analysis via H&E staining revealed clear pathological damage in the liver and kidney, including red blood cell extravasation and infiltration of inflammatory cells, indicative of organ damage. In contrast, infection with the ΔsmpB strain resulted in minimal pathological changes, demonstrating a significant attenuation of virulence (Fig. 6B). We further analyzed the systemic colonization of A. veronii in various organs. Deletion of smpB significantly impaired colonization of the liver, kidney, and spleen (Fig. 6C). Both deletion and overexpression of argR caused mild inflammatory cell infiltration, and argR deletion only slightly reduced colonization in the liver and kidney (Fig. 6B, C).

Figure 6.

SmpB regulates the intestinal colonization and virulence of A. veronii. (A) Colonization rates of A. veronii with the indicated genotypes in the cecum, colon, and small intestine of KM mice. The data are presented as means. n = 4 or 5 for biological replicates as indicated in the dot plot. P-values were determined using an unpaired two-tailed t-test, with statistical significance denoted as *P < 0.05; ns, P > 0.05. (B and C) H&E-stained tissue sections and bacterial colonization counts of various organs following intraperitoneal challenge in mice. Red arrows indicate red blood cell extravasation, blue arrows denote inflammatory cell infiltration, and yellow boxes highlight disrupted boundaries between the red and white pulp in the spleen. P-values were determined using an unpaired two-tailed t-test.

Discussion

Dynamic alterations in cell morphology represent a prevalent strategy employed by bacteria to adapt to diverse environmental challenges. Many bacterial species modify their morphology to survive under stress or during host infection. For example, Vibrio parahaemolyticus transitions from rod-shaped to spherical forms during starvation [49], and elongates its cell length under alkaline conditions to enhance resistance [50]. Similarly, Listeria monocytogenes undergoes morphological alteration in response to alkaline stress to improve its resilience [51], while C. crescentus adopts a filamentous form to optimize nutrient uptake and adapt to a freshwater environment [12]. These examples, along with others, underscore the importance of morphological plasticity as a survival mechanism that enables bacteria to thrive in a fluctuating environment. Actinomyces israelii develops a branched or filamentous rod shape when subjected to limitations in phosphate, cysteine, or glutathione [4]. Similarly, P. aeruginosa, P. putida, and P. fluorescens elongate into slim, rod-like cells under nutrient-poor conditions [7,52]. Thus, it is reasonable that M9 medium induces A. veronii to adopt an elongated morphology, enhancing its adaptability to environmental stress. However, the morphological plasticity of bacteria poses a major obstacle to the effective prevention and control of infections in clinical settings. Changes in cellular shape have been shown to alter drug resistance [53], stress tolerance [54], and immunogenicity [11], ultimately complicating pathogen management and treatment. Consequently, elucidating the molecular mechanisms underlying bacterial morphological transformation is of great significance for the development of new strategies to control pathogens and identify novel antibiotic targets. In this study, we found that A. veronii exhibits increased cell length when cultured in M9 medium, a phenotype that depends on smpB. Consistent with previous reports, we further observed that the changes in cell morphology were accompanied by alterations in drug resistance, stress tolerance, and virulence. Therefore, A. veronii employs a morphological adaptation mechanism to enhance its survival under varying environmental conditions, with smpB acting as a key molecular regulator in this process.

SmpB is primarily known as the dominant ribosome rescue element that maintains protein translation homeostasis by complexing with tmRNA to release stalled ribosomes [55]. Absence of SmpB impairs bacterial growth, intestinal colonization, virulence, and antibiotic persistence of A. veronii [56,57]. However, the extent to which the canonical ribosome rescue mechanism underlies these physiological processes remains elusive. We differentiated the role of smpB beyond its traditional trans-translation function by knocking out tmRNA. It appears that smpB-mediated PG synthesis probably operates independently of the trans-translation. Instead, SmpB acts as a TF, orchestrating the expression of a specific subset of target genes involved in cell wall biosynthesis and adaptation to stress conditions, such as nutritional deprivation. As a TF, SmpB promotes cell elongation in A. veronii by up-regulating enzymes catalyzing PG biosynthesis, particularly under nutrient-restricting environments. This mechanism enables A. veronii to thicken its cell wall and adopt an elongated morphology, facilitating survival under stress conditions. Furthermore, the role of SmpB in regulating cell wall integrity correlates strongly with increased resistance to antibiotics targeting the bacterial cell wall and enhanced tolerance to other environmental stresses, highlighting the physiological significance of SmpB in modulating cell wall-associated fitness of A. veronii. Ribosome stalling can also be induced by nutrient starvation, i.e., amino acid deficiency. How SmpB bifurcated its dual roles in ribosome rescue during translation stalling and transcriptional regulation of PG synthesis remains an intriguing question. Investigating the evolutionary mechanisms underlying this functional divergence will be a key focus in future studies.

Many environmental factors induce bacterial morphological changes, including pH shift [50], osmotic pressure fluctuations [58], nutrient deficiencies [59], and amino acid supplementation [38]. We reveal that the arginine sensor ArgR is a key upstream regulator of smpB, linking its activity to environmental stimuli. First, ArgR directly binds to the smpB promoter, repressing its transcription. Second, ArgR interacts with SmpB at the protein level, potentially sequestering SmpB and impeding its access to downstream targets, such as genes involved in PG biosynthesis. Considering that ArgR serves as the primary sensor and regulator of arginine metabolism, it is plausible that SmpB might sense the availability of arginine through ArgR, thereby modulating cell morphology. However, phenotypic analysis of the ΔargR strain yields inconsistent results, suggesting the involvement of additional, context-specific mechanisms. Overexpression of argR impairs the resistance of A. veronii to multiple stresses and induces a shortened morphology similar to the ΔsmpB strain, further confirming the negative regulation of smpB by ArgR. Notably, smpB was found to enhance ArgR’s transcriptional repression of the smpB promoter, forming a negative feedback loop. This mechanism is likely to prevent excessive PG synthesis, ensuring precise regulation of cell wall integrity. Importantly, argR expression remains stable as a global transcriptional regulator; elevated SmpB levels can bind to ArgR, enhancing ArgR’s inhibitory effect on the smpB promoter without disrupting other ArgR-regulated pathways. This regulatory strategy is both economical and highly specific. In A. veronii, the ArgR-binding motif (RDTTATGCAB) is distinct from those reported in other bacterial species. For instance, in E. coli, ArgR binds to a canonical 18 bp palindromic sequence (TGTGA-N⁶-TCACA) [34]; in P. aeruginosa and Bacillus licheniformis, the binding motifs are partially palindromic sequences of 18–20 bp, such as TGTCGCNNNNNNGNAA [60] and TTTATCATAATTATTCATT [61], respectively. In M. tuberculosis, ArgR recognizes a shorter 9–10 bp motif, TTGTTATTTTT [62]. These findings indicate that ArgR regulates its downstream targets via species-specific DNA motifs, suggesting a functional diversity of ArgR among different bacterial taxa. Indeed, this SmpB-mediated morphological plasticity has critical physiological significance to A. veronii, as SmpB is required for optimal survival, colonization, and virulence in vivo. Similar findings have been reported in V. cholerae, where deletion of DUF368, a gene responsible for morphological changes during alkaline adaptation, impairs colonization in the host [13]. Interestingly, while argR overexpression led to a ΔsmpB-like morphological defect, significant differences were not observed in virulence and intestinal colonization. This discrepancy may arise from ArgR’s broader regulatory functions, as the ΔargR strain also exhibited a significant decrease in colonization efficiency, albeit to a lesser extent. RT-qPCR was performed to compare the transcriptional levels of smpB and argR in cells cultured in LB and M9 media (Supplementary Fig. S10). During the logarithmic growth phase, argR expression was down-regulated while smpB expression was up-regulated, consistent with their proposed negative regulatory relationship. However, in the stationary phase, a nutrient-limited condition, smpB expression remained unchanged, whereas argR expression was significantly up-regulated, contrary to our expectations. We hypothesize that this discrepancy may have contributed to increased expression of σ32. Specifically, σ32 may compete with ArgR for binding to the smpB promoter region (Fig. 3), thereby interfering with ArgR-mediated repression. In E. coli K-12, the transition from the late exponential phase to the stationary phase is accompanied by increased σ32 levels [63]. To test this hypothesis, we quantified the expression of rpoH (which encodes σ32) under both LB and M9 growth conditions. As anticipated, rpoH expression was significantly elevated during the stationary phase in M9 medium, approximately 15-fold higher than in the logarithmic phase (Supplementary Fig. S10C).

Sequence alignment of ArgR and smpB across five bacterial species revealed high conservation of both proteins, including critical interaction residues (S54 in ArgR and D138 in SmpB), suggesting a shared regulatory mechanism across bacteria (Fig. 4D; Supplementary Fig. S12). The D138 residue was also essential for trans-translation, indicating a coordinated role for smpB in bacterial morphology and stress adaptation. Additionally, we identified a conserved DNA motif in the smpB promoter that bound with ArgR, further supporting the conservation of the ArgR–smpB regulatory axis across species (Supplementary Fig. S12B–D). In sum, this study uncovers an unconventional function of SmpB in regulating PG synthesis and morphological adaptation, independent of its canonical trans-translation function. Several pioneering studies have reported diverse effects of ArgR on pathogenic microorganisms [64,65]. As an arginine sensor, we initially hypothesized that ArgR could link nutrient sensing, particularly arginine metabolism, to bacterial cell morphology. However, our data from argR knockout and overexpression experiments do not fully support this hypothesis. Further studies are needed to fully understand the intricate interplay between arginine sensing, ArgR-mediated SmpB inhibition, and the regulation of bacterial morphology.

The discovery of the SmpB regulatory axis reveals a new mechanism linking bacterial metabolism to cell wall synthesis, enabling adaptation to environmental stress. Unlike conventional resistance pathways, its reliance on negative feedback regulation between ArgR and SmpB offers novel targets for antimicrobial therapy. These adaptive changes serve as biomarkers for infection dynamics and drug resistance, exposing vulnerabilities that could be exploited to hinder bacterial survival. Understanding these processes enhances our knowledge of host–pathogen interactions, paving the way for innovative treatments and improved diagnostics. Together, these findings open up new possibilities for combating bacterial infections and developing next-generation therapeutics.

Supplementary data

Supplementary data is available at NAR online.

Conflict of interest

All authors declare no conflict of interest.

Funding

The National Natural Science Foundation of China [32160038, 32260020, 32260028, and 32360047] and Yunnan Fundamental Research Projects [202501AS070036].

Data availability

The experimental and computational data that support the findings of this research are available in this article and in Additional file 1, or upon request from the corresponding author. RNA-seq data can be found in the NCBI: GSE120603 and PRJNA1202361. The tool and analysis script developed in this study for predicting upstream transcription factors based on promoters are open source and can be found on Zenodo (DOI:10.5281/zenodo.15582349).