S. aureus strain Newman, an agr group I prototype used as the backbone strain in this study, was originally isolated from a secondarily infected tubercular osteomyelitis lesion (Duthie and Lorenz, 1952). N315, an agrII prototype MRSA with defective agr system, was also used as control background strain (Tsompanidou et al., 2011). MW2 is a prototypical agr group III strain originally isolated from a child with fatal septicemia and septic arthritis (Mei et al., 2011). XQ is a clinical community-associated MRSA belongs to group IV prototype strain isolated from an adolescent patient with staphylococcal scalded skin syndrome (Rao et al., 2015). Escherichia coli strain DH5α was used for plasmid construction and genetic manipulation. All strains were stored in 10% glycerol at −80°C (Supplementary Table 1). E. coli strains were grown in Luria–Bertani broth, whereas S. aureus strains were cultivated in trypticase soy broth (TSB; Sigma) or brain–heart infusion medium (Sigma). When necessary, ampicillin (100 μg/mL) and chloramphenicol (20 μg/mL) were added to the medium. The temperature-sensitive S. aureus plasmid pBT2 and E. coli–S. aureus shuttle plasmid pLI50 kindly provided by Prof. Baolin Sun (University of Science and Technology of China, China) were used for gene mutation and complementation assay, respectively. Transformations of S. aureus strains were performed by electroporation (Bio-Rad Gene Pulser). The primers used in this study were presented in Supplementary Table 2.

The NewmanΔagrBDC and N315ΔagrBDCA markerless deletion mutants were constructed with pBT2 plasmid using homologous recombinant strategy as described previously (Yuan et al., 2013). Take NewmanΔagrBDC construction as an example, a 803-bp DNA fragment upstream of agrBDC locus was amplified from Newman genome DNA using primer pairs, ΔagrBDC up-for and ΔagrBDC up-rev (Supplementary Table 2). The polymerase chain reaction (PCR) product was digested with HindIII and SalI and subcloned into the same site of pBT2 to obtain plasmid pagrBDCF. Then, a 957-bp DNA fragment downstream of agrBDC locus was amplified and subcloned into the BamHI and EcoRI site of pagrBDCF, yielding the ΔagrBDC knockout plasmid (pagrBDC). The pagrBDC plasmid was identified by restriction enzyme digestion and DNA sequencing and then electrotransformed into S. aureus RN4220 and Newman to construct NewmanΔagrBDC markerless deletion mutant by homologous recombination. The mutants were screened and confirmed by PCR amplification and DNA sequencing. The N315ΔagrBDCA markerless deletion mutant was constructed with similar strategy.

The agrBDC fragments of agrII, agrIII, and agrIV were separately obtained from N315, MW2, and XQ by PCR amplification and were, respectively, inserted into the SalI and BamHI sites of pBT2 plasmid to yield pagrBDC-II, pagrBDC-III, and pagrBDC-IV knock-in vectors. Then, these pagrBDC vectors were sequentially introduced into S. aureus RN4220 and NewmanΔagrBDC deletion mutant strain to construct allelic replacement strains via homologous recombination. The substitution of agrBDC fragments of allelic congenic strains was also verified by PCR amplification and DNA sequencing.

Full gene DNA of types I through IV agr clusters and their promoters were separately amplified from Newman, N315, MW2, and XQ using the primers described in Supplementary Table 2 and subcloned into the pLI50 plasmid. All positive recombinant plasmids were confirmed by DNA sequencing. Then, these plasmids were, respectively, transformed into E. coli DH5α, S. aureus RN4220, and finally into NewmanΔagrBDC or N315ΔagrBDCA mutant to construct types I–IV agrBDCA genes complemented strains.

The hemolytic activity, pigment formation, exoprotein production, and gene transcription of agr allelic congenic strains were analyzed to assess the influence of agr system replacement on S. aureus biological characteristics.

For hemolytic activity assessment, overnight culture of each single S. aureus colony was diluted into the same colony-forming units (CFUs) and plated on rabbit’s blood agar plates, followed by overnight growth at 37°C for hemolysis analysis. Furthermore, the hemolytic activity was also evaluated according to the hemolysis of rabbit erythrocytes. As the method described previously (Pader et al., 2014), 100 μL overnight culture supernatant was mixed with 6% rabbit blood in phosphate-buffered saline and then incubated at 37°C for 20 min. The unlysed blood cells were removed by centrifugation, and the erythrocyte lysis was determined with the OD543 values of supernatant.

To evaluate staphyloxanthin production, the diluted overnight inoculation of S. aureus strains was also cultured on TSB plate to assess the effect of agr replacement on pigment formation. Moreover, staphyloxanthin productions were also quantitatively analyzed and adapted from a previously published method (Liu et al., 2018). Cells in 1-mL overnight culture were collected and washed thrice with sterilized water and then resuspended with 200 μL methanol and heated at 55°C for 3 min. After the cells were centrifuged at 10,000 × g for 1 min, the OD462 value of supernatant was detected.

The overnight culture supernatants of S. aureus were collected by centrifugation. Proteins in supernatants were precipitated with trichloroacetic acid and analyzed by sodium dodecyl sulfate–polyacrylamide gel electrophoresis (SDS-PAGE) in a 12% polyacrylamide gel.

The change in gene transcription was appraised with quantitative real-time PCR (qRT-PCR). Cells were harvested at 6 h after inoculation, which represent the mid-log phase of the S. aureus growth. The total bacterial RNA was isolated using SV Total RNA Isolation System kit (Z3100; Promega, United States). The contaminated genomic DNA in RNA was degraded with DNaseI. The cDNA was prepared using PrimeScript RT Reagent Kit (RR047A; Takara, Japan) and used for qRT-PCR using GoTaq ^® qPCR Master Mix (A6001; Promega, United States) on an ABI SimpliAmp PCR detection system (United States). Specific primers for qRT-PCR (Supplementary Table 3) were designed according to the target gene sequences. All PCR reactions were performed in triplicate, with 16S rDNA as internal control. The relative expression of gene products was normalized to the housekeeping gene 16S and calculated using the 2^–ΔΔCT method.

The agr systems were divided into four agr groups named agrI, agrII, agrIII, and agrIV in S. aureus (Wang and Muir, 2016). AgrA is highly conserved, whereas AgrB, AgrD, and AgrC are variable among the four agr groups (Supplementary Figure 1). In this study, S. aureus Newman and N315 were selected for agr congenic strains construction to assess the effects of divergent agr alleles. The whole RNAII transcripts (involved agrB, agrD, agrC, and agrA genes) were allelic substituted to construct congenic strains in N315, whereas only agrB, agrD, and agrC genes were replaced in Newman congenic strains. The congenic strains were constructed via homologous recombination and confirmed by PCR amplification (Figure 1 and Supplementary Table 2) and DNA sequencing.

S. aureus is able to secrete a variety of toxins, such as α-hemolysin, bicomponent leukocidins, γ-hemolysin, Panton–Valentine leukocidin (PVL), β-hemolysin, δ-hemolysin, PSMs, and so on (Cheung and Otto, 2012). These exotoxins have leukotoxic and hemolytic activities and are widely associated with the pathogenicity of S. aureus (Powers et al., 2015; Seilie and Bubeck Wardenburg, 2017). α-Hemolysin, γ-hemolysin, PVL, δ-hemolysin, and PSM belong to pore-forming toxins capable of forming transmembrane aqueous channel and leading to host cell lysis (Reyes-Robles and Torres, 2017). β-Hemolysin is a neutral sphingomyelinase capable of digesting sphingomyelin into ceramide and phosphorylcholine (Vandenesch et al., 2012). The production of hemolysins in S. aureus is tightly regulated, and the agr system plays an important role in this process (Johansson et al., 2019). It was reported that the agr system controls the expression of α-, β-, and γ-hemolysin and PVL by RNAIII and regulates the transcription of δ-hemolysin and PSMs through AgrA (Arya and Princy, 2013; Singh and Ray, 2014). The hemolytic capabilities of S. aureus vary with their agr types. As shown in Figure 2A, the Newman (agrI) and XQ (agrIV) strains have strong hemolytic activities, the MW2 (agrIII) strain showed weak hemolysis, and the N315 (agrII) strain presented the lowest hemolytic toxicity (Figure 2A).

To assess the effects of different agr alleles on S. aureus hemolytic activity, the agrBDC genes in Newman strain (agrBDC-I) were deleted and in situ substituted with other three agrBDC alleles (agrBDC-II, agrBDC-III, and agrBDC-IV). As expected, the hemolytic activity of the NewmanΔagrBDC strain was significantly lower than that of the wild strain (Figure 2B). The knock-in of type I agrBDC genes back to the genome of NewmanΔagrBDC mutant recovered the hemolytic activity of revertant strain (Figure 2B). Contrary to our expectations, no visible hemolysis and very low hemolytic activity were discovered on three congenic replacement strains harboring heterologous agr systems (Figure 2B). The hemolytic activity controlled by agr system seemed to be severely suppressed when the three heterologous agr alleles were in situ recombined into the genome of NewmanΔagrBDC mutant.

To investigate this further, four recombinant pLI50 plasmids containing types I to IV agrBDCA genes and their promoter sequences were, respectively, transformed into the ΔagrBDC mutant strains of Newman or N315. As shown in Figures 2C,D, the four agr complemented strains showed similar hemolytic activities to each other. The differences in hemolysis across agr groups were significantly weakened in the same Newman or N315 background when compared with the standard agr allelic strains.

Staphyloxanthin, a carotenoid pigment produced by S. aureus, protects bacteria from neutrophil oxidants (Xue et al., 2019). When the agrBDC sequence was deleted, NewmanΔagrBDC mutant strain lost staphyloxanthin production and formed white colonies (Figures 3B,C). All congenic strains of Newman constructed by genomic replacement or plasmid complemented did not bring back the yellow color of wild Newman strain and showed white colonies (Figures 3B,C). The deletion of genes agrBDCA of N315 did not lead to visible color change (Figure 3D). However, four agr alleles complemented mutants in N315 presented white colonies in contrast to the golden colonies from the wild N315 and N315ΔagrBDCA mutant strains (Figure 3D).

Five genes, crtOPQMN, organized in an operon are responsible for the biosynthesis of staphyloxanthin (Xue et al., 2019). Two colorless farnesyl diphosphate successively catalyzed by five enzymes (CrtM, CrtN, CrtP, CrtQ, CrtO) to yield orange staphyloxanthin. Dehydrosqualene desaturase, CrtN, catalyzes the formation of the first deep yellow-colored carotenoid intermediate product, 4,4′-diaponeurosporene (Wieland et al., 1994). The sigma factor B (SigB) plays an essential role in regulating staphyloxanthin biosynthesis by binding to the promoter that laid in upstream of crtO (Kullik et al., 1998; Pelz et al., 2005). In the present study, the transcription levels of genes crtN and sigB were analyzed and found to be down-regulated in congenic strains when compared with the wild strains, Newman or N315 (Figure 4). It can be inferred that the down-regulated transcriptions of crt operon, especially that of crtN, are responsible for the weaken pigmentations of congenic strains, and the decreased SigB expression may play an important role in these processes.

To explore the influences of agr alleles on S. aureus exoprotein expressions, overnight culture supernatants were collected and analyzed by SDS-PAGE. As shown in Figure 5A, each agr type strain has its distinctive exoprotein profiles. Deletion of agrBDC genes led to significant changes in exoprotein expressions of Newman (Figure 5A). The knock-in of agrII, agrIII, or agrIV agrBDC genes did not change the exoprotein expressions of NewmanΔagrBDC; the three allelic strains exhibited similar exoprotein profiles as that of NewmanΔagrBDC mutant (Figure 5A). However, when the four allelic agrBDCA genes were introduced by the pLI50 plasmid, the exoprotein patterns of four agr allelic complemented strains changed, which were highly similar to each other (Figure 5B) but different from those of wild agr allelic strains and the genomic replacement strains (Figures 5A,B). The agr allele–dependent difference in exoprotein expression was diminished when in the same Newman background. The similar results were also observed in N315 (Figure 5C). These complement strains exhibited significantly higher exoprotein expression than wild strains (Figures 5B,C). The two most abundant bands at approximately 25 and 35 kDa were identified as bicomponent hemolysin, α-hemolysin, or serine protease with mass spectrometry (Supplementary Tables 4, 5), which have been demonstrated to be regulated by agr system.

The ability of S. aureus to invade host tissues and cause infections depends on the production of a number of virulence factors. Many of them are regulated by the agr system such as α-toxin, β-toxin, δ-toxin, serine protease, fibrinolysin, PSM, and enterotoxin B (Seilie and Bubeck Wardenburg, 2017). In this study, we sought to examine the effects of different agr alleles on the transcription of virulence factors as hla, hlb, hld, hlg, psmα, psmβ, pvl, and aur (Figure 6 and Supplementary Figure 4).

We found that the transcription of most toxins that are known to be activated by agr system (Bronesky et al., 2016) was now down-regulated in the NewmanΔagrBDC mutant strain; the transcriptional factors and some surface proteins (Foster et al., 2014) that are usually inhibited by agr were now up-regulated. When the agrBDC genes of Newman strain (agrI) were genomically replaced by agrBDC genes from the other three agr alleles, the resulting mutants showed very low agr expression (data not shown). This is consistent with the finding of hemolytic assay, staphyloxanthin formation test, and exoprotein profile analysis. As the agrBDC deletion was complemented by plasmid pLI50, the transcription of agrA and RNAIII of NewmanΔagrBDC returned to the original level of wild Newman (Supplementary Figures 3B,E). While the relative mRNA level of agrA increased 30–50 times and the RNAIII increased about thousand folds compared with that of wild N315 (Supplementary Figures 3C,F), the virulence factor transcription of these allelic complemented strains constructed in the same background varied in the same trend, showing an elimination of agr-dependent difference in identical background (Figure 6 and Supplementary Figure 4). For example, the transcription levels of hla gene of four agr congenic Newman strains increased 15–30 times compared with the wild Newman (Figure 6B), whereas they significantly increased 150–300 times when the four agr alleles were, respectively, complemented in agr defective N315 by plasmid pLI50 (Figure 6C). The hlb mRNA level showed 260- to 990-fold increase in N315 background, whereas only 30- to 72-fold in Newman background (Figures 6E,F). Similar changes were also observed in transcriptions of proteases such as aureolysin-encoding gene (aur, Figures 6H,I) and PVL-encoding gene (pvl, Figures 6K,L). Interestingly, the levels of psmα and psmβ in all the four complemented strains were significantly decreased in the Newman background (Supplementary Figures 4H,K) compared with the wild Newman strain, but increased hundreds or thousands of folds in the N315 background (Supplementary Figures 4I,L).