Bacterial strains V. splendidus AJ01 used in this study were previously isolated from SUS diseased sea cucumber and characterized in our laboratory (Zhang et al., 2016). Bacterial were cultured at 28°C in 2216E medium [5 g/L tryptone, 1 g/L yeast extract, and 0.01 g/L FePO₄ in filtered natural seawater]. The overnight bacteria culture was diluted 1:100 into fresh 2216E medium and cultured until it reached at different growth stages (OD₆₀₀ = 0.6, 1.2, and 2.0. These three OD’s are represented log phase, early stationary phase, and late stationary phase, respectively). Tetracycline-induced persister cells were selected as described by our previous work (Li et al., 2021). The samples without tetracycline treatment at OD₆₀₀ of 0.6, 1.2, and 2.0 were served as the control. RNA isolation, cDNA library preparation, Illumina sequencing, and mapping of reads were conducted on the Illumina HiSeq/MiSeq sequencing platform at Novogene Biotech (Beijing, China). The trimming and quality control of raw Illumina reads were conducted to obtain clean reads. The clean reads were mapped to the genome of Vibrio atlanticus strain LGP32 (FM954973.2) using Bowtie2 (Langmead and Salzberg, 2012). Cluster Profiler software was used to analyze KEGG pathway enrichment of differential gene sets. p_adj < 0.05 was considered as the threshold of significant enrichment. p_adj was the corrected value of p, and a small p_adj value of differentially expressed gene indicated high significance of the differential expression (Jin and Hammell, 2018). p_adj = p^* m/n, such that p is the original value of p, m is the maximum number of tests that minimizes Padj, and n is the total number of tests (Hu et al., 2018). Interactive Pathways (iPath) analysis using iPath3.0.¹

Metabolome samples preparation were the same as the transcriptome as mentioned above. The overnight bacteria culture was diluted 1:100 into fresh 2216E medium and cultured until it reached at OD₆₀₀ = 1.2. The sample numbers of VS and VST were labeled. Each sample had six biological replicates. The resulting samples were used for non-target metabolomics liquid chromatography-mass spectroscopy analysis and conducted at Shanghai Luming Biological Technology Co. Ltd (Shanghai, China). Multivariate statistical analyses were applied for metabolites. Normalization and analysis of metabolome data were completed using Microsoft Excel. Principal-component analysis (PCA) was used to reduce the high dimension of the data set and analyze the covariance variation and emphasize the outlier in clustering. The significantly altered abundance of metabolites was analyzed and outlined in KEGG (https://www.kegg.jp/).

Five hundred μM 6-MP (Krishnan et al., 2009) and 50 μM mycophenolic acid (MPA) (Zhou et al., 2020) were used to inhibit purine metabolism to ATP and GTP, respectively. For dose-dependent inhibition assay, 1 mL overnight cultured V. splendidus AJ01 was treated with three concentrations of 6-MP (50 μM, 250 μM, and 500 μM) and four concentrations of MPA (5 μM, 10 μM, 20 μM, and 50 μM) for 3 h before adding 250 μg/mL tetracycline for 4 h at 28°C, respectively. Then the cells were collected by centrifugation at 5,000 g for 5 min. The cells were resuspended and washed with phosphate buffered saline (PBS) [137 mM NaCl, 2.7 mM KCl, 8 mM Na₂HPO₄, and 1.5 mM KH₂PO₄] three times to isolate unlysed persister cells and total cells. One milliliter of the cell culture was serially diluted by PBS, and 10 μL samples were spotted onto 2216E agar plates. Colonies were counted after overnight incubation at 28°C, and those dilutions yielding 20–200 colonies were enumerated to calculate colony-forming units (CFU). For membrane potential inhibition, 10 μM cyanide m-chlorophenyl hydrazone (CCCP) was added to the bacteria for 30 min before tetracycline treatment. Inhibition of efflux pump was treated with 100 μM NMP was added to the bacterial culture medium at the same time points with tetracycline.

Intracellular ATP levels of the bacteria cultures were measured using BacTiter-Glo Microbial Cell Viability Assay (Promega, G8231) following the manufacturer’s instructions. Cells were suspended to an OD₆₀₀ of 0.6 (around 10⁵ bacterial cells) in 2216E medium incubated with the addition of various concentration gradient of chemicals before ATP measurement under a fluorescence microplate reader (ThermoFisher). Intracellular ATP concentration was determined through normalizing ATP levels by cell number and single cell volume.

One milliliter overnight cultured V. splendidus AJ01 was treated with 6-MP or MPA for 3 h before adding 250 μg/mL tetracycline for 4 h at 28°C, respectively. And 1 mL overnight cultured cells were treated with exogenous ATP and 250 μg/mL tetracycline for 4 h at 28°C. For cells staining, cells were collected by centrifugation at 5,000 g for 5 min, resuspend and washed with PBS three times to isolate unlysed persister cells and total cells. Then cells were incubated with 1 μM fluorescein isothiocyanate (FITC) (Sigma, aldrich) for 20 min (Polak et al., 2019). Three washing steps with 1 mL PBS were carried out to remove the free fluorescent dye (Polak et al., 2019). For protein aggresomes visualization, cellular “dark foci” (brightfield) and “green fluorescent foci” (green fluorescent staining) as a marker for the formation of protein aggresomes (Pu et al., 2018). Intracellular green fluorescent foci were imaged using FITC filter sets. Image analysis software and counted dark foci with the ImageJ (Le et al., 2021).

The method for insoluble protein aggresomes isolation was described by Pu et al. (2018). Two milliliters of bacterial cultures at OD₆₀₀ = 1 were rapidly cooled on ice for 10 min and centrifugated at 5,000 g for 10 min at 4°C to collect bacteria. Pellets were suspended in 100 μL buffer A [10 mM potassium phosphate buffer (PPB), pH 6.5, 1 mM EDTA, 20% (w/v) sucrose, 1 mg/mL lysozyme] and incubated for 30 min on ice. Cell lysate was added with 500 μL of buffer B (10 mM PPB, pH 6.5, 1 mM EDTA), then sonication in ice bath. Intact cells were removed by centrifugation at 2,000 g for 10 min at 4°C. The pellet fractions were resuspended in 500 μL of buffer C (buffer B with 2% NP40) to dissolve membrane proteins and the aggregated proteins were isolated by centrifugation at 2,000 g for 10 min at 4°C. Then the membrane and aggregated proteins were centrifugation by 15,000 g for 30 min at 4°C. The pellet fractions were resuspended in 400 μL of buffer B by brief sonification and centrifugation by 15,000 g for 30 min at 4°C. NP40-insoluble pellets were washed twice with 400 μL of buffer B and re-suspended in 50 μL of buffer B. Aggregated proteins of each sample tetracycline-induced persister cells and active cells were cut out of the gel and characterized on the LC–MS platform at a biotechnology company (Shanghai Sangon Biological Engineering Technology & Services Co. Ltd., China). VENNY 2.1² analyzed the common of protein aggresomes. KOBAS 3.0³ analyzed the enrichment of KEGG.

Intracellular tetracycline concentrations of the cultures were measured using Tetracycline enzyme-linked immunosorbent assay (ELISA) diagnostic kit (Lunchangshuo, SU-B99855) following the manufacturer’s instructions. The cells were collected by centrifugation at 5,000 g for 5 min and washed three times with PBS. An aliquot of 5 mL was sonicated (30% output, 2 s ultrasound, and 8 s interval time for a total of 2 min) in an ice bath. The resulting supernatant was collected for the detection of tetracycline following the instructions of the tetracycline ELISA diagnostic kit. Intracellular tetracycline concentrations were measured under a microplate reader (Molecular Devices, SpectraMax 190) at 450 nm.

Membrane potential of cells was measured using BacLight™ Bacterial Membrane Potential Kit (Life Technologies, B34950) following the manufacturer’s instructions. One milliliter of 10⁶ CFU of bacteria were used and stained with 10 μL of 3 mM 3, 3-diethyloxa-carbocyanine iodide [DiOC₂(3)], followed by incubation for 30 min. Cultures with the addition of various concentration gradient of CCCP (0 μM, 10 μM, 25 μM, and 75 μM) to the depolarized sample and mix for 30 min. Samples were analyzed using a FACScan flow cytometer (Becton Dickinson, San Jose, CA) using the following settings: fluorescein isothiocyanate (FITC) voltage, 250 V; mCherry voltage, 650 V. The red/green (mCherry/FITC) values for each cell were determined, normalized, and then compared between samples. The relative membrane potential of test samples was determined.

To observe the single cell was performed on an agarose gel pads as described previously (Zhang et al., 2019). Low melting agarose (1.5%) were prepared using the sandwich method (Kim and Gadd, 2019). Briefly, 1.5% of low-melting agarose was added to 2216E liquid medium and melted by microwaving. DiOC₂(3) exhibits green fluorescence in all bacterial cells, but the fluorescence shifts toward red emission as the dye molecules self-associate at the higher cytosolic concentrations caused by larger membrane potentials. When we observed single cell membrane potential changes using fluorescence microscopy, in cells with higher cytoplasmic concentrations, self-associating red fluorescence still appears, however only for 1–2 s. Subsequently, the red fluorescence of bacteria at high membrane potential disappeared, and the green fluorescence of bacteria with high membrane potential is stronger than that of low membrane potential. Therefore, the DiOC₂(3) fluorescence value was measured under a fluorescence microscope (ZEISS, Axiovert A1) maintained at 28°C with the excitation wavelength at 488 nm and detected with emission filters suitable for FITC (green).

To make fluorescence measurements, cells were treated with 0.1 μM Hoechst 33342 for 60 min at 28°C in the dark. For the efflux pump inhibition experiment, bacteria cultures were added N-Methylpyrrolidone (NMP) incubated at 28°C for 4 h, which can effectively block efflux pumps through competitive substrate export (Donner et al., 2017; Camp et al., 2021). The cells then were loaded onto 1.5% agarose pads, which made with the same 2216E growth medium. Cells were imaged with a fluorescence microscope. Intracellular Hoechst 33342 were imaged using 4′,6-diamidino-2-phenylindole (DAPI) filter sets. Image Analysis Software and analyzed with the ImageJ.

Intracellular pH measurement was conducted as previously described (Song et al., 2021). Briefly, the tested bacteria were washed with PBS for three times and were then treated with 10 μM BCECF-AM (Beyotime, S1006) for 30 min at 28°C. The fluorescence intensity value was measured under a fluorescence microscope with the excitation wavelength at 488 nm and emission wavelength at 535 nm.

Quantitative real-time PCR (qRT-PCR) was carried out as described previously (Li et al., 2021). Total RNA was isolated from cells using a Bacterial RNA Isolation kit (Omega, R6950-01). The purity and quantity of the total RNA were determined using a NanoDropTM 2000 spectrophotometer (Thermo Fisher Scientific). By using a Primescript™ 1st cDNA Synthesis kit (TaKaRa, RR036A) reverse transcription-PCR was carried out on 1 μg of total RNA according to the manufacturer’s instructions. qRT-PCR was performed in 96-well plates, amplifications were performed in a 20 μL of reaction volume containing 8 μL of diluted cDNA (1:100 dilution of cDNA with sterile water), 0.8 μL of each primer, and 10.4 μL of SYBR Green Mix (Takara, DRR420A). The reaction mixtures were carried out in an ABI 7500 RT-PCR detection system (Applied Biosystems). The 2^-△△CT method was used to analyze the expression level of each gene. Data are shown as the relative mRNA expression compared with control with the endogenous reference 16S rRNA gene. Primers used for qRT-PCR are listed in the Supplementary Table S1.

In our previous study, we defined the live AJ01 by 10 × MIC tetracycline treatment for 4 h as tetracycline-induced AJ01 persister cells (Li et al., 2021), which was consistent with Keren ‘s experimental method (Keren et al., 2011). To investigate which pathways are associated with tetracycline-induced persister cell formation, integrated transcriptomes and metabolomes were employed to identify the differentially expressed genes (DEGs) and metabolic products of tetracycline-induced persister cells (VST) compared to active cells (VS). The DEGs of VST vs. VS at three stages were determined (Supplementary Figure S1A). A total of 1,166 DEGs containing 654 up-regulated and 512 down-regulated genes were obtained between VST and VS at OD₆₀₀ = 0.6. When the OD₆₀₀ increased to 1.2, 1,018 DEGs with 565 up-regulated and 453 down-regulated genes were detected. At the condition of OD₆₀₀ = 2.0, there were 638 up-regulated and 472 down-regulated DEGs (Supplementary Figure S1A; Supplementary Tables S1–S4). KEGG analysis revealed a total of 66 pathways were significantly enriched in the three stages (Supplementary Figure S1B; Supplementary Tables S5–S7), the common pathways including 7 pathways for ribosome synthesis, purine metabolism, biofilm formation, homologous recombination, histidine metabolism, tyrosine metabolism, and amino acid synthesis (Supplementary Figure S1B). Interactive Pathways Explorer (iPath3.0) analysis revealed that the VST DEGs were enriched in most metabolic pathways, in which carbohydrate metabolism and nucleotide metabolism were the most significant ones (Supplementary Figure S1C). Notably, the purine metabolism genes of purM, purB, purH, and gmk were significantly down-regulated (green numbers, p < 0.05) in tetracycline-induced persister cells (Supplementary Figures S1D,E). The expression of these genes were further validated by qRT-PCR, and found that they showed similar depressed expression profiles to transcriptome analysis between tetracycline-induced persister cells and active cells (Supplementary Figure S2).

Building on the transcriptome results, liquid chromatography-mass spectroscopy-based metabolome analysis was conducted for VST and VS at OD₆₀₀ = 1.2. A total of 237 metabolites were identified as differential produts (Figure 1A), in which 19.0, 30.8, 17.7, and 10.1% abundance were categorized as carbohydrates, amino acids, lipids, and nucleotides, respectively (Figures 1B). The two groups of VST and VS were individually clustered together by PCA analysis, demonstrating that tetracycline-induced persister cells and active cells had a distinct metabolic signature (Figure 1C). To explore the metabolic pathways that distinguish VST and VS, KEGG was performed to analyze the metabolites of differential abundance. Fourteen metabolic pathways were enriched as shown in Figure 1D and Supplementary Table S8. The six common pathways were purine metabolism (p = 2.061 × 10⁻⁶), ABC transporters (p = 0.0037), arginine biosynthesis (p = 0.0058), amino sugar and nucleotide sugar metabolism (p = 0.0078), phosphotransferase system (PTS) (p = 0.0085), and aminoacyl-tRNA biosynthesis (p = 0.0098) (Figure 1D). The purine metabolism pathway was the most enriched. By comparing the metabolites of purine metabolism pathway of VST and VS, 40 metabolites had significant differential abundance, the identified metabolites were categorized as purine metabolism (25%), purine nucleosides (20%), purines and purine derivatives (55%) (Figure 1E). The purine metabolites of tetracycline-induced persister cells decreased by 57.5% compared with active cells, there are 22 metabolites, including 2-Methyladenosine, Xanthosine, N6-Methyladenosine, 2-Phenylaminoadenosine, Guansine monopjpsphate, and Dexyadenosine monophosphate, were decreased (Figure 1E and Supplementary Table S9). Meanwhile, the intracellular ATP level of tetracycline-induced persister cells was significantly lower than that of active cells (Supplementary Figure S3).

To further verify that the down-regulated purine metabolism was associated with persister cells formation, the change of persister cell formation was investigated using different purine metabolism inhibitors (Figure 2). We found that ATP levels were significantly reduced after 6-mercaptopurine (6-MP, inhibits ATP production) treatment (Figures 2A,B), and that a higher ratio of persister cells was also detected from 100 μM 6-MP to 500 μM 6-MP treatments (Figure 2C). On the contrary, inhibiting GTP synthesis via MPA treatment significantly increased intracellular ATP levels (Figures 2D,E), and reduced persister cell formation in a dose-dependent manner (Figure 2F). Similarly, the number of persister cells were also decreased after supplying exogenous ATP (Figure 2G), but there were no significant change in persister cell number with exogenous GTP (Figure 2H). All these results support that the depressed purine metabolism, i.e., the ATP synthesis pathway, promote the formation of persister cell in AJ01.

In our previous work, we found that a novel phenotypic feature-dark foci was detected in tetracycline-induced persister cells but was not in active cells (Li et al., 2021). To further confirm reducing purine metabolism promoted AJ01 persistence that was related to the protein aggresome formation, we counted the proportion of bacteria that formed protein aggresomes by normalized to cell numbers by microscopy observation. Accordingly, the fraction of cells with protein aggresomes, as assessed by the presence of dark foci or fluorescent foci. The percentages of cells with dark foci after treatment with different concentrations of tetracycline were 11.2, 42.3, 79.1, and 95.3%, respectively (Figure 3A). Within the decreased ATP levels by 6-MP treatment group, the proportion of cells containing protein aggresomes, increased with 6-MP concentration, such as 11.5, 13.9, 46.2, and 90.1%, respectively (Figure 3B). In contrast, increased ATP levels by supplying exogenous MPA significantly reduced the percentages of cells with protein aggresome (Figure 3C). Similarly, the direct addition of 20 mM exogenous ATP also reduced the proportion of cells containing protein aggresomes (Supplementary Figure S4).

Subsequently, we extracted the insoluble protein aggresomes, a total of 490 and 348 insoluble proteins were obtained from tetracycline-induced persister cells and active cells, respectively (Figure 3D and Supplementary Table S10). Among them, analysis of common proteins using Venn, tetracycline-induced persister cells and active cells identified 273 proteins (Figure 3E). KEGG pathway analysis revealed that most proteins aggregated are ribosome proteins and proteins associated with metabolic pathways (Figure 3F). Strikingly, six proteins of PurD, PurH, PurA, GuaB, GuaA, and AdK associated with the purine metabolism were detected in active cells (Figure 3G), and seven proteins of PurD, PurC, PurH, PurA, GuaB, Ndk, and Adk were also detected in tetracycline-induced persister cells (Figure 3G), and the abundance of these proteins were shown in Supplementary Table S11. Protein NdK and PurC were aggregated only in tetracycline-induced persister cells, and its protein abundance was 0.73 and 0.44, respectively (Supplementary Table S11). The main function of NdK and PurC were to maintain the balance of ATP and NTP concentration in cells through phosphate group transfer reaction and to participate in the UTP and GTP biosynthesis process at the same time (Francois-Moutal et al., 2014). These results confirm that tetracycline-induced persister cells with protein aggresome formation that is associated with purine metabolism inhibition and ATP levels reduction.

Reduction in the intracellular antibiotic accumulation is another factor to promote the formation of persister cells (Roy et al., 2021). To confirm the connection between purine metabolism and intracellular tetracycline accumulation, we investigated whether exogenous 6-MP addition decreased the intracellular concentration of tetracycline. Cells treated with 6-MP showed a significant, though no dramatic, decreased in tetracycline that was consistent with our hypothesis (Figure 4A, p = 0.045). Firstly, DiOC₂(3) was used to measure transmembrane potential of bacteria. DiOC₂(3) exhibits green fluorescence in all bacterial cells, as the dye molecules self-associate at the higher cytosolic concentrations caused by larger membrane potentials, the green fluorescence is brighter than other low membrane potential after the red fluorescence disappears (Jiang et al., 2020). Consistently, higher membrane potential was found upon 6-MP addition as shown by higher DiOC₂(3) fluorescence intensity (Figures 4B,C), low membrane potential was found upon ATP addition as shown by low DiOC₂(3) fluorescence dim (Figures 4B,C), membrane protential was increased though marginally by 2.1%. The change showed a trend in 6-MP treated cells; lower intracellular tetracycline concentration and increased membrane potential (Supplementary Figure S5). Meanwhile, the increase of bacteria stained by PI after the addition of ATP indicates that the cells died (Figure 4C). In contrast, CCCP, the inhibitor of membrane potential treatment increased the concentration of tetracycline (Figures 4D, 5B–D), and also decreased the percentage of persister cells (Figure 4E). Together, these results support that the inhibition of purine metabolism is associated with intracellular tetracycline accumulation, which is achieved by increasing membrane potential.

To further address the potential mechanism of membrane potential-mediated intracellular tetracycline concentrations, then we examined how membrane potential affects substrate transport by using a fluorescent dye, Hoechst 33342 (Le et al., 2021). First, dead cells were excluded by PI staining. For active cells, we found that DiOC₂(3)-bright cells were Hoechst 33342-dim, whereas DiOC₂(3)-dim cells were Hoechst 33342-bright (Figure 5A). This observation suggested that the coexistence of cells with two distinct Hoechst 33342 intensities was caused by the heterogeneous on membrane potential, suggesting that membrane potential might regulate the efflux of intracellular Hoechst 33342. In contrast, membrane potential inhibitor CCCP treatment decreased DiOC₂(3) fluorescence intensity in a dose-dependent manner (Figures 5B–E), and also increased Hoechst 33342 fluorescence intensity (Figures 5F–I). Inhibition of efflux pumps by NMP could increase accumulation of Hoechst 33342 fluorescence intensity (Figures 5J–M). These results suggest that the inverse correlation of fluorescence intensity observed between DiOC₂(3) and Hoechst 33342, thus, the efflux of Hoechst 33342 is driven by membrane potential. After tetracycline treatment, persister cells with bright DiOC₂(3) fluorescence intensity of also had dim Hoechst 33342 fluorescence intensity (Supplementary Figure S6). NMP could increase accumulation of Hoechst 33342 fluorescence intensity in 6-MP treatment group, which was consistent with NMP treatment group (Supplementary Figure S6D). Similarly, the concentration of intracellular tetracycline after NMP treatment was also increased by 6-MP treatment (Figures 4F,G). Together, these results reveal that membrane potential mediates tetracycline accumulation by effecting efflux.

The membrane potential and transmembrane proton pH gradient are related (Song et al., 2021). A pH shift in the intracellular microenvironment leads to a decrease in the transmembrane proton pH gradient (Lewis, 2007). We assayed the intracellular pH by using the pH indicator BCECF-AM and found the BCECF-AM fluorescence signal significantly decreased with an increase in tetracycline or 6-MP treatments (Figures 6A–C, and Supplementary Table S12). Similarly, 6-MP treatment also led to a significant increase in DiOC₂(3) fluorescence in the presence or absence of tetracycline, indicating that the membrane potential was increased (Figure 5B). To further validate the change of membrane potential depending on transmembrane proton pH gradient, the pH of culture was adjusted by adding 40 mM potassium benzoate and 40 mM methylamine hydrochloride (Supplementary Figure S7A). When the pH of the growth medium had been adjusted to be acidic (pH = 6.30), 11% increase of DiOC₂(3) fluorescence intensity of membrane potential was detected in 6-MP treated bacteria (Figure 6D). In contrast, when the pH was adjusted to alkaline (pH = 7.95), membrane potential was decreased by 38% (Figure 6D). These results support that the inhibition of purine metabolism increases membrane potential depending on dissipation of transmembrane proton pH gradient.