SXT-induced stable phenotype of S. aureus SCVs

Among the 30 clinical wild-type isolates, 3 strains (2, 15 and 29) exhibited the SCVs phenotype after 11, 15 and 5 days of subculturing with SXT, respectively (Fig. 1). The induced small colonies were less than one-tenth the size of the colonies related to their parental strains and exhibited slower growth. Even after 10 passages of subculture on antibiotic-free plates, the SCVs maintained the small colony size, indicating their stability as a distinct cell type.

Fig. 1

Phenotype of wild-type and small-colony variant (SCV) originated from representative strain 15 of S. aureus. A SCV phenotype observed after 15 days of induction with 12/238 µg/ml of SXT, cultured for 24 h. B SCV cultured for 24 h after screening through ≥ 10 passages of subculture. C SCV cultured for 48 h after screening through ≥ 10 passages of subculture. D Wild-type parental strain cultured for 24 h. E Wild-type parental strain cultured for 48 h

Identification and genotyping of SCVs

The 3 selected SCVs were confirmed to be S. aureus through Vitek MS detection and 16S rRNA sequencing. PFGE analysis revealed that the induced SCVs exhibited the same pattern profiles as their parental strains (Fig. 2), indicating clonality between the SCVs and the wild-type strains [11]. Additionally, agr typing showed that the induced SCV and its parental strain belonged to the same type (strains 2 and 29: agr type II; strain 15: agr type I). This information confirmed that the SCVs originated from their respective parental strains.

Fig. 2

Pulsed-field gel electrophoresis (PFGE) patterns (SmaI digest) of three pairs of S. aureus isolates. P represents the parent normal S. aureus, S represents the small-colony variant (SCV), and M represents the molecular marker

Thymidine, menadione, and hemin compensation in induced SCVs

Since the formation of S. aureus SCVs are usually linked with deficiencies in the utilization of menadione, hemin or thymidine by bacteria, we conducted the auxotrophism test on the induced SCVs to find the possible mechanisms [12]. The results showed that the colony sizes of the 3 induced SCVs strains on MH plates supplemented with each compound separately were not significantly different from those on the plates without these agents, as shown in Fig. S1. This indicated that our SCVs were not dependent on thymidine, menadione, or hemin for their growth.

Growth curves of SCVs

The growth kinetics of SCVs in TSB were examined, as depicted in Fig. S2. Compared to the control parental strains, the SCVs exhibited a noticeably reduced in vitro growth over the monitoring period. Particularly for the SCVs derived from strains 15 and 29, their OD600 values at 16 h were only 55.3% and 42.7% of those observed in their respective parental strains.

Antimicrobial susceptibility

Commonly used antibiotics MICs of the isolates as determined by E-test are shown in Table S3. The results showed that the MICs of penicillin to SCV 29, linezolid to SCV 2, oxacillin, ciprofloxacin, vancomycin and clindamycin to SCV 15, gentamicin and amikacin to SCVs 15 and 29, tigecycline to SCVs 2 and 29, and rifampicin and SXT to three SCVs were elevated when compared with corresponding parental strains. There was no change in the susceptibility of levofloxacin and erythromycin for all SCVs.

Analysis of genome sequence and single nucleotide polymorphisms

Based on the kinetics curves of bacterial growth and the timing of SCV formation, strain 15 (day 1) and its corresponding SCV (day 15) were chosen for whole-genome sequencing analysis. The genome features of both strains are summarized in Table 1. A phylogenetic tree reveals a close relationship between strain 15 and its SCV form with strains of S. aureus USA300 FPR3757, COL, and NCTC 8325 (Fig. S3). Mauve analysis shows a few observed genomic rearrangements in the SCV compared to the parental strain, potentially resulting from genome shuffling (Fig. S4).

Table 1 The comparative genomic features between the SCV and its parental S. aureus strain (strain 15)

Identification of SNPs, insertions, and deletions revealed 29 mutation events in the SCV compared to its parental strain, with half of them being synonymous mutations. These missense mutations primarily affected 13 genes, including spa, clfB, deoB, mgtE, asp2, and those located on phage genome (holin, SAOUHSC_01556, 02022, 02025, 02029, 02074, 02076, and 02216). The mutation events are listed in Table 2.

Table 2 Single nucleotide polymorphisms (SNPs) in SCV compared with its parental strain (strain15) by whole-genome sequencingIdentification and DEGs

A total of 54,727,711 raw reads were obtained through RNA sequencing. After quality checking, 39,294,597 reads (19,342,594 for the parental group and 19,952,003 for the SCV group) were generated (Table 3). Subsequent analysis of gene expression profiles identified 2,418 expressed genes, among which 277 (11.5%) were differentially regulated, including 150 significantly increased and 127 prominently decreased DEGs (Table S4). Figure 3 displays a visualization of all the DEGs between the SCV and its parental strain. The heat map (Fig. 3B) clearly demonstrates several genes that were markedly down-regulated in the SCV, such as NWMN_2268 (encoding lactate permease) and NWMN_0176 (encoding L-lactate dehydrogenase), both of which play crucial roles in metabolic pathways. The representative DEGs are listed in Table S5.

Table 3 The data of RNA sequencing after quality control for samples of S. aureus strain and its SCVFig. 3

RNA Sequencing analysis of DEGs between parental S. aureus strain 15 and its SCV. A Scatter plot illustrating the RNA sequencing analysis of DEGs between the parental S. aureus strain 15 and its induced SCV. The DESeq2 v 1.10.1 package was employed to identify the DEGs, considering a |log2(fold change)|≥ 1 and a false discovery rate (FDR) ≤ 0.05. The x- and y-axes represent the normalized signal values of samples in the two groups. Genes up-regulated by the SCV are denoted by green dots, while genes down-regulated by the SCV are represented by red dots. Blue dots indicate genes that were not differentially expressed. B Heat map displaying the DEGs between the SCV and its parental strain. A1-A3 correspond to parental strain samples, while B1-B3 represent SCV samples. DEGs stands for differentially expressed genes. Each box in the heat map represents a gene, with the color indicating the level of gene expression. Each column represents the expression of each gene in different samples, and each row represents the expression of all genes in each sample

To validate the transcriptome results, the expression levels of 49 selected genes (23 up-regulated and 26 down-regulated genes) were reanalyzed using qRT-PCR. As shown in Fig. 4, the transcription levels of 48 out of 49 DRGs (98.0%) validated by qRT-PCR were consistent with the findings from RNA sequencing. Contradictory data (RNA sequencing: up-regulated; qRT-PCR: down-regulated) were observed for the gene NWMN_2331 (Fig. 4E). Although there were isolated discrepancies between the two technologies, overall, the transcriptome data were reliable.

Fig. 4

qRT-PCR analysis for confirmation of transcription levels of selected 48 DEGs between SCV and its parental strain. A qRT-PCR analysis revealed that 23 DEGs were up-regulated in the SCV. B 25 DEGs were found to be down-regulated in the SCV. The black bar represents SCVs, while the gray bar represents the parental strain. Significance levels: *, P < 0.05, **, P < 0.01, ***, P < 0.001, ****, P < 0.0001. The conformance of the data between RNA sequencing and qRT-PCR, analyzed with log2FC (y-axis), is depicted in panels C, D, and E. C, D The verification of 23 up-regulated and 25 down-regulated determinants in the SCV showed matching results between the two detection technologies. E The expression of NWMN_2331, as assayed by qRT-PCR, was confirmed to be opposite to that observed in RNA sequencing. The black bar represents qRT-PCR, while the gray bar represents RNA sequencing. qRT-PCR stands for quantitative real-time polymerase chain reaction, while DEGs refers to differentially expressed genes. FC represents fold change

Bioinformatics analysisGO enrichment analysis of DEGs

The GO enrichment analysis revealed that the DEGs were enriched in three GO domains: biological process (63 terms), cellular component (109 terms), and molecular function (55 terms) (Table S6). Further exploration showed significant enrichment in certain GO items. For example, the biological process domain had a notable enrichment of genes involved in metabolic process (53 genes), cellular process (51 genes), and single-organism process (38 genes). In the cellular component domain, enrichment was observed in cell and cell part (102 genes each), as well as membrane (53 genes). The molecular function domain showed enrichment in catalytic activity (42 genes) and binding (20 genes) (Fig. 5, Table S6). Most of the GO items for biological functions were enriched with down-regulated genes in SCV, and there were almost twice as many the down-regulated genes as the up-regulated genes (Table S6). Besides, from GO analysis, we could find that down-regulated genes were significantly enriched in “response to stimulus”, “growth” and “membrane part” categories, which may imply that dysfunction in these functions occurred in SCV.

Fig. 5

GO enrichment analysis of DEGs between SCV and its parental strain. GO databases were utilized to perform GO functional annotation on the DEGs using the R package GOseq v 1.18. The annotated determinants (y-axis) are categorized into three GO domains: biological process (17 terms), cellular component (6 terms), and molecular function (7 terms). The x-axis represents the number of DEGs. The blue histogram corresponds to the up-regulated genes, while the red histogram represents the down-regulated genes. GO stands for Gene Ontology

KEGG pathway enrichment analysis

Mapping the DEGs to KEGG pathways identified 68 different pathways associated with 94 DEGs (Table S7). As shown in Fig. 6A, the most prevalent pathways among the DEGs were metabolic pathways (58 genes), followed by biosynthesis of secondary metabolites (25 genes), microbial metabolism in diverse environments (25 genes), and biosynthesis of antibiotics (22 genes). Additionally, pathways related to carbon metabolism (14 genes), ABC transporters (14 genes), biosynthesis of amino acids (13 genes), purine metabolism (11 genes), glycolysis/gluconeogenesis (10 genes), pyruvate metabolism (10 genes), and glycine, serine, and threonine metabolism (9 genes) were significantly enriched (Fig. 6A). Furthermore, a few DEGs were associated with arginine biosynthesis, pyrimidine metabolism, quorum sensing (6 genes each), methane metabolism, cysteine and methionine metabolism, citrate cycle, propanoate metabolism, glyoxylate and dicarboxylate metabolism, and glycerolipid metabolism (5 genes each). The bubble chart in Fig. 6B illustrates 20 significantly enriched pathways based on three dimensions: rich factor, p-value, and gene number.

Fig. 6

Enrichment analysis of KEGG pathways for DEGs between SCV and its parental strain. The enrichment analysis of KEGG pathways was conducted using the KEGG database. A The top 20 enriched KEGG pathways of the DEGs are presented. These pathways are arranged in order of the number of DEGs, with metabolic pathways being the most enriched, followed by biosynthesis of secondary metabolites, microbial metabolism in diverse environments, and biosynthesis of antibiotics. B Scatterplots depict the 20 most significant KEGG enrichment pathways of the DEGs. The selection of these pathways was based on three factors: the rich factor corresponding to each pathway (horizontal axis), the p-adjust value (color of the dots), and the number of DEGs enriched in each pathway (size of the dots). The vertical axis represents the names of the pathways. KEGG refers to the Kyoto Encyclopedia of Genes and Genomes

Analysis of differential metabolites

After processing the raw mass spectrometry data using XCMS, a package based on the R language, we utilized metaX, a metabolomics data analysis software also based on the R language, along with KEGG to obtain the first-level identification results of metabolites. Out of the 27,176 identified metabolites, 2,992 differential ions (11%) were characterized, including 1,345 up-regulated ions (4.9%) and 1,647 down-regulated ions (6.1%) in the SCV (Table S8). It is worth noting that the ions with the highest fold of up-regulation in the SCV were all associated with carboxylic acids and derivatives (Table S9). The KEGG pathway analysis revealed that the differentially regulated ions were primarily enriched in pathways related to metabolism, with only a small fraction associated with cellular processes, environmental information processing, genetic information processing, and human disease (Fig. 7). Among the metabolic pathways, the three most enriched pathways in terms of ions were metabolic pathways (34 differentially regulated ions), biosynthesis of secondary metabolites (16 differentially regulated ions), and biosynthesis of amino acids (13 differentially regulated ions). Additionally, ABC transporters from environmental information processing and aminoacyl-tRNA biosynthesis from genetic information processing each enriched 11 differentially regulated ions.

Fig. 7

KEGG analysis of differentially regulated ions between SCV and its parental strain. In the bubble chart, the 60 most significantly enriched pathways are displayed out of a total of 73 enriched pathways. The selection of these 60 pathways was based on the number of metabolites, p-value, and enrichment factor. The enrichment factor represents the ratio of differential metabolites located in a specific KEGG pathway to the total number of metabolites in that pathway. A smaller p-value indicates a higher level of KEGG enrichment

Analysis of transcriptomic and metabolomic data

The conjoint analysis of transcriptomic and metabolomic data utilized DEGs and significantly different metabolites (SDMs) that were enriched in KEGG pathways. A heatmap, generated using the Spearman method, was used to visualize the correlation between DEGs and SDMs (Fig. S5). By examining the overlapping pathways, a total of thirty-five pathways were identified, with the majority being involved in metabolic pathways (Fig. 8). Notably, the highlighted common pathways included purine metabolism, pyruvate metabolism, ABC transporters, and arginine metabolism. These pathways had an impact on bacteria in terms of nucleic acid synthesis and energy metabolism. For instance, out of the 14 DEGs from ABC transporters, 10 were connected to the iron complex transport system permease protein. Furthermore, 11 SDMs, which included several amino acids, are important components of ABC transporters.

Fig. 8

Bubble chart of overlapped pathways in KEGG analysis. This figure displays a bubble chart representing the 35 pathways that overlap between the metabolomic and transcriptomic data in the KEGG analysis. In the chart, DEGs are marked with circles, while differentially regulated metabolites are marked with triangles. The enrichment factor is a measure indicating the ratio of differential genes/metabolites located within a specific KEGG pathway to the total number of genes/metabolites in that pathway. A smaller p-value corresponds to a higher level of KEGG enrichment