The bacterial strains and plasmids used in this study are listed in Table 1 . The S. epidermidis wild-type (WT) strain and its isogenic mutants were routinely grown in Tryptic soy broth (TSB) or Brain Heart Infusion (BHI) with a shaking speed of 220 rpm at 37°C, and E. coli strains were cultured in Lysogeny Broth (LB). When necessary, antibiotics were added to the above media to the following final concentrations: 50 μg·ml^-1 carbenicillin and 10 μg·ml^-1 chloramphenicol.

The primers used in this study are listed in Supplementary Table 1 . For construction of the sarZ knockout mutant, upstream and downstream DNA fragments flanking of the sarZ gene were PCR amplified from the genomic DNA of S. epidermidis strain RP62A.The two DNA fragments were further combined by fusion PCR and then inserted into the pUCm-T vector through TA cloning. The resulting vector was taken as the template to obtain the two homology arms with restriction endonuclease sites Apa I and Nco I by PCR amplification. The homologous arms were then cloned into E. coli-staphylococcal shuttle vector pKOR1 by using conventional molecular cloning technique, yielding the recombinant plasmid pKOR1-ΔsarZ. The plasmid was transformed into E. coli strain DC10B for restriction modification and subsequently electroporated into S. epidermidis strain RP62A. For construction of the sarZ psmβ double mutant, the two homology arms flanking the entire psmβ operon were obtained by adopting the cloning strategy mentioned above. Instead, the recombinant plasmid pKOR1-Δpsmβ were generated through In-Fusion cloning according to the manual of ClonExpress Ultra One Step Cloning Kit (Vazyme, Nanjing, China). The remaining steps for allelic replacement were performed as described previously (Zhu et al., 2017). The desired mutant strains were screened by PCR with a pair of primers complementary to regions outside the homology arms and eventually confirmed by DNA sequencing.

For complementation of the sarZ mutant strain, a DNA fragment encompassing the sarZ gene and its promoter region was amplified, and then ligated into pCNcat to create pCNcat-sarZ through In-Fusion cloning. For complementation of the double mutant strain, a DNA fragment containing the entire psmβ operon and its promoter sequences was cloned into pCNcat as described above. The promoter sequences were predicted by using BDGP Neural Network Promoter Prediction software (http://www.fruitfly.org/seq tools/).

To detect the hemolytic activity of S. epidermidis wild-type and its isogenic mutants, all the strains were cultivated to exponential phase in BHI medium. Then, a 2.5 μl aliquot of each culture was withdrawn and spotted onto Columbia blood agar plate. The plates were incubated at 37°C for approximately 24 hours and the hemolytic zones were examined and photographed.

To investigate the effect of sarZ mutation on S. epidermidis biofilm formation, a semiquantitative microplate assay involving crystal violet staining was performed as described previously (Zhu et al., 2017).

Differential biofilm formation was further observed using scanning electron microscopy (SEM). Briefly, S. epidermidis cells were seeded into a 6-well tissue culture plate containing segments of central venous catheter, and grown in TSB medium at 37°C for 24 hours. After that, catheter segments were washed three times with 1×PBS to remove planktonic bacteria and fixed with 2.5% glutaric glutaraldehyde at 4°C for 6 hours. After progressive alcohol dehydration, catheter segments were air-dried, mounted onto the SEM holder with black glue and gold-sputtered. SEM micrographs were taken at ×3000 and ×10000 magnification.

Crude PSMs were extracted from S. epidermidis according to the protocol reported by Joo et al. (Joo and Otto, 2014) with some modifications. Briefly, S. epidermidis cells were cultured overnight in BHI medium, and subsequently centrifuged to collect the supernatant. The supernatant was clarified by filtration, and then added with 1/3 of 100% 1-butanol to make 25% of 1-butanol. Then, the mixture was shaken vigorously (260 rpm) at 37°C for 2 hours. After brief centrifugation, the upper phase (1-butanol phase) was harvested and then evaporated using conventional rotary evaporation at 60°C. To accelerate the evaporation process, the distilled water was added into the 1-butanol phase (1-butanol phase: distilled water=1:2 (v/v)). Finally, the precipitates were redissolved in distilled water to acquire the crude PSMs and then visualized by 12% SDS-PAGE.

Hemolytic activity of the crude PSMs was determined with sheep red blood cells. One hundred microliters of crude PSMs extracts were mixed with 900 μl of 1×PBS buffer containing 3% sheep red blood cells. The mixtures were then incubated at 37°C for 2 hours. After centrifugation, the absorbance of the supernatant was measured at 540 nm to quantify the released hemoglobin in order to evaluate the extent of erythrocyte lysis.

After reduced by 10 mM DTT at 56°C for 1 hours and alkylated by 50 mM IAM at room temperature in dark for 40 min, the crude PSMs were lyophilized to near dryness, and then resuspended in 0.1% formic acid. HPLC-MS/MS experiments were carried out on an Easy-nLC 1000 system connected to an Orbitrap Exploris™ 240 Mass Spectrometer (Thermo Fisher Scientific, USA) equipped with an ESI nanospray source. Chromatographic separation was performed on an in-house made NanoColumn (15cm, ID 150 μm, 3μm, C18) with a flow rate of 600 nL/min. The LC linear gradient was used from 6% to 9% B for 5 min, from 9% to 14% B for 15 min, from 14% to 30% B for 30 min, from 30% to 40% B for 8 min and from 40% to 95% B for 2 min. Solvent A was 0.1% formic acid in water, and solvent B was 0.1% formic acid in acetonitrile/water (80:20, v/v). For ionization, a spray voltage of 2.2 kV and a 320°C capillary temperature was used. Peptides were analyzed in positive mode from 350 to 1,500 m/z, followed by data-dependent higher-energy collision dissociation (HCD) MS/MS scans using a normalized collision energy of 30%.

The raw MS files were analyzed and searched against the UniProt Staphylococcus protein databases using Byonic. The parameters were set as follows: the protein modifications were carbamidomethylation (C) (fixed), oxidation (M) (variable), Acetyl (Peptide N-term) (variable), Formyl (Peptide N-term) (+27.99) (variable); the enzyme specificity was set to trypsin; the maximum missed cleavages were set to 3; the precursor ion mass tolerance was set to 20 ppm, and MS/MS tolerance was 0.02 Da. Only high confident identified peptides were chosen for downstream analysis. The relative quantification of PSMs peptides were performed by integration of extracted ion chromatograms of formylated and deformylated forms.

S. epidermidis cells were grown in six milliliters of BHI medium at a flask-to-media volume ratio of 5:1 to the mid-logarithmic phase (4 hours) and stationary phase (10 hours), respectively. Total RNA was prepared using the RNeasy-mini kit (Qiagen) according to the manufacturer’s instructions with some modifications as described before (Zhu et al., 2022). Afterward, 0.5 μg of total RNA was taken to synthesize cDNA using the Go Script™ Reverse Transcription Mix (Promega). The qPCR reactions were run on a LightCycler^® 96 instrument. Fast qPCR Master Mix (2×) (Roche) was used along with specific primers as listed in Supplementary Table 1 . Each reaction was carried out in triplicate, and the gyrB gene was employed as reference gene for normalization. The relative quantity of target genes was calculated using the 2^-ΔΔct method.

The sarZ gene was amplified with the primers in Supplementary Table 1 and inserted into the vector pET28a to construct a recombinant expression plasmid pET28a-sarZ. The resulting plasmid was then transferred into E. coli BL21 star (DE3). The expression strain was grown in LB medium with 50 μg·ml^-1 kanamycin at 37°C, shaking at 220 rpm until OD₆₀₀ was 0.6, and then added with a final concentration of 1 mM isopropyl β-D-thiogalactopyranoside (IPTG) at 37°C for additional 4 hours’ cultivation. The culture was harvested by centrifugation, and the pellet was resuspended in lysis buffer (20 mM Na₂HPO₄, 500 mM NaCl, 20 mM imidazole, pH 7.4) and then homogenized on ice by sonication. After centrifugation, the supernatant was collected and purified through Ni-IDA sefinose Resin (Shanghai Sangon Biotech). The His-tagged SarZ protein was eluted with 20, 100, 150, 300 mM imidazole. The purity of the recombinant protein was checked by 12% SDS-PAGE. The protein concentration was determined using a Bradford protein assay kit.

For EMSA assay, the 5’-biotin-labeled DNA fragments containing the promoter regions of the psms and ica operon were amplified from the S. epidermidis genomic DNA using the primers listed in Supplementary Table 1 . Then, the biotin-labeled DNA fragments were incubated with increasing concentrations of purified SarZ protein at room temperature for 20 min in binding buffer (Shanghai Beyotime Biotech). For competitive EMSA, a 100-fold and 200-fold molar excess of unlabeled fragment was respectively added to the reaction mixture for 20 min prior to addition of a constant amount of labeled fragment. After incubation, the mixtures were electrophoresed on a 5% native polyacrylamide gel in 0.5× TBE buffer and then the gel was blotted onto a positively charged nylon membrane. The shifted bands were detected and analyzed using the LightShift Chemiluminescent EMSA kit (Shanghai Beyotime Biotech). The rpsJ (encoding 30S ribosomal protein S10) gene was designated as negative control for SarZ-DNA binding.

A 330-bp fragment (from 735785 bp to 736114 bp), covering the 276-bp of psmβ promoter region in the EMSA assay, was synthesized and then ligated into pUC57 vector through in-fusion cloning. The promoter fragment was fluorescently labeled by PCR amplification with the primers listed in Supplementary Table 1 . Footprinting assays were performed according to the method described previously (Queck et al., 2008), with some modifications. Briefly, binding reactions were set up in 40 μl volumes, which contained 1×binding buffer (Shanghai Sangon Biotech), 50 ng/μl salmon sperm DNA, 500ng FAM-labeled DNA fragment and various amounts of purified SarZ. After incubation at room temperature for 30 min, 5 microliters of enzyme mixture comprising 1× RQI buffer (Shanghai Sangon Biotech), 10 mM CaCl₂, 0.1U/μl DNase I (Thermo Fisher) was added. The DNase I digestion was carried out for about 55 s and terminated by adding 10 μl of 0.5 M EDTA. Final DNA fragments were extracted using DiaSpin PCR Product Purification Kit (Shanghai Sangon Biotech) and detected with an Applied Biosystems 3730XL DNA analyzer. Electropherograms were analyzed and aligned using the GeneMapper software (Applied Biosystems). The assay was repeated at least three times with similar results.

Experimental data obtained were analyzed using GraphPad Prism8 and compared by the independent-sample t-test or one-way analysis of variance (ANOVA). P< 0.05 were considered statistically significant.

To confirm the regulatory role of SarZ on virulence factors in S. epidermidis, the sarZ deletion mutant strain was successfully constructed via homologous recombination ( Supplementary Figures 1 – 4 ). It was found that the sarZ mutant strain displayed a significantly larger zone of hemolysis than the WT strain when aliquots of both strains were spotted onto Columbia blood agar plate and then incubated at 37 °C for 24 hours. Complementation of the sarZ mutant strain by pCNcat-sarZ restored the hemolytic zone at size close to the WT strain, whereas empty vector had no effect on the hemolysis ( Figure 1 ). These results indicated that SarZ inhibits the hemolytic activity of S. epidermidis.

To investigate whether SarZ influences the biofilm formation of S. epidermidis, a semiquantitative biofilm assay was carried out. As shown in Figures 2A, B , the sarZ mutant strain formed significantly reduced biofilm compared to the WT strain. The amount of biofilm was restored to the wild-type level after complementation of the sarZ mutant strain. Furthermore, the SEM images showed that the WT strain generated a dense biofilm comprising multiple layers of bacterial cells on the catheter surface, whereas the sarZ mutant strain just formed a few single-layer bacterial clusters ( Figure 2C ). Altogether, these findings confirmed that SarZ promotes biofilm formation in S. epidermidis.

Since PSMs in Staphylococcus were reported to have cytolytic capacity toward erythrocytes and neutrophils during the past decade, it was speculated that SarZ might regulate the hemolytic activity by inhibiting the production of PSMs. To test the hypothesis, crude PSMs were isolated from the spent medium of the sarZ mutant and its parent strain by 1-butanol extraction, and their hemolytic activities were further compared. As expected, butanol extract of the sarZ mutant strain exhibited significantly stronger cytolytic activity against sheep erythrocytes than the counterpart of the WT strain ( Figures 3A, B ). Furthermore, the crude PSMs were resolved by SDS-PAGE. Compared with the WT strain, the protein band corresponding to PSMs were obviously more intensive in the sarZ mutant strain ( Figure 3C ), indicating that PSMs production was significantly higher in the sarZ mutant strain than in the WT strain.

Subsequently, in order to identify the PSMs member whose production is primarily affected by SarZ, HPLC-MS/MS was performed to distinguish the differences in PSMs content between the sarZ mutant strain and its parent strain. A total of six PSMs that were previously found to be produced by S. epidermidis, including PSMα, PSMβ1, PSMβ2, PSMγ, PSMδ and PSMε, were detected. Among them, PSMβ1 and PSMα were only detected in the sarZ mutant strain, whereas no or little difference in PSMγ and PSMδ production was observed, suggesting that amounts of PSMβ1 and PSMα in the mutant strain were considerably higher than that in the parent strain ( Figure 3D ). In addition, PSMβ2 production was also obviously higher in the mutant strain. It is noteworthy that in both strains, the amount of β-type PSMs was more abundant than that of PSMα ( Figure 3D ), which is in accordance with previous study. These above results indicated that SarZ may inhibit the hemolytic activity of S. epidermidis by downregulating the production of the PSMs family, particularly the β-type PSMs.

To confirm the substantial differences in the pattern of PSMs production between the sarZ mutant strain and the WT strain, transcript levels of all psm genes in both strains were measured using qRT-PCR. As a result, gene expression of psmα , psmβ1, psmβ2, psmβ3, psmδ and psmε was indeed upregulated obviously in the sarZ mutant strain ( Figures 4A–F ), which coincides with the MS data. In particular, the transcript levels of β-type psm were increased significantly, with a 10-, 8-, 11-fold increase in the logarithmic phase and 7-, 13-,8-fold increase in the early stationary phase for psmβ1, β2 and β3, respectively ( Figures 4B–D ). Second, the transcript levels of psmα were also elevated markedly, with a respective 8- and 7-fold change in the logarithmic and early stationary phase ( Figure 4A ). As anticipated, the transcript levels of psmγ, also known as hld, did not differ between the wild type and sarZ mutant strain ( Figure 4G ). Consistent with the remarkably decreased biofilm formation, transcript level of icaA gene was notably downregulated in the sarZ mutant strain ( Figure 4H ). However, no difference was observed in the icaR transcript levels ( Figure 4I ). Taken together, these results demonstrated that SarZ inhibits transcription of psm genes, especially psmβ operon. Meanwhile, SarZ contributes to biofilm formation by activating transcription of ica operon independently of icaR.

Obviously increased expression of β-type psm revealed by both the HPLC-MS/MS and qRT-PCR in the sarZ mutant led us to speculate that SarZ may regulate the hemolytic activity of S. epidermidis mainly by controlling the expression of β-type psm. In order to verify our speculation, the sarZ psmβ double mutant strain was then constructed successfully by allelic replacement ( Supplementary Figures 5 – 7 ; Figure 5A ). No significant differences in the growth curves were monitored between the WT, sarZ mutant and sarZ psmβ double mutant strains ( Figure 5B ). As expected, the double mutant strain displayed a significant decreased hemolytic activity compared with the sarZ single mutant strain, reaching the level equivalent to the WT strain. The hemolytic zone was restored to the size comparable to the sarZ single mutant strain after complementation of the double mutant strain with the intact psmβ operon ( Figure 5C ). Taken together, these data demonstrated that the regulation of hemolytic activity by SarZ is dependent on the presence of psmβ operon in S. epidermidis.

As a member of the MarR-family proteins (Dyzenhaus et al., 2023), SarZ contains one predicted helix-turn-helix (HTH domain) DNA-binding domains. Since all psm genes transcription is altered in the sarZ mutant, the ability of SarZ binding directly to their promoter regions was investigated. The recombinant His-tagged SarZ was used for EMSA with biotin labeled DNA fragments containing the respective promoter regions of psm genes (262-bp psmα, 276-bp psmβ, 230-bp psmε and 183-bp psmγ). Despite repeated attempts, the promoter region of psmδ was always unable to be fluorescently labeled to conduct the assay. As shown in Figure 6 , the promoter fragments of psmα, psmβ and psmε formed retarded DNA-protein complex with SarZ in a dose-dependent manner, respectively ( Figures 6A–C , lane 2 to lane 4). The addition of a 200-fold excess of unlabeled identical DNA fragments as a specific competitor completely blocked SarZ-biotin-DNA complex formation ( Figures 6A–C , lane 6). As expected, psmγ promoter fragment did not form a shifted complex with SarZ ( Figure 6D ). We also examined the ability of SarZ binding to the 165-bp labeled DNA fragment containing the ica operon promoter. As illustrated in Figure 6E , with the increase of SarZ dosage, the shifted band of ica operon promoter emerged and was strengthened. As a negative control, a 119-bp DNA fragment of rpsJ gene did not form a shifted complex with SarZ under the same condition ( Figure 6F ). Collectively, these data indicated that SarZ can specifically bind to the promoter region of the psm genes and ica operon, excluding the psmγ.

Since SarZ could directly bind to the promoter region of the psmβ operon, we were interested in the SarZ recognition sites on the psmβ operon promoter. We performed DNase I footprinting analysis with the 330-bp psmβ operon promoter fragment labeled with 6-carboxyfluorescein (6-FAM) ( Figure 7 ). Without addition of the SarZ protein, the 6-FAM labeled DNA fragment was uniformly digested, as reflected by the uniform distribution of the 6-FAM signals ( Figure 7A ). The two regions from -138 to -90 bp and -34 to -5 bp, relative to the putative transcription start site of the psmβ operon, were protected, as indicated by the disappearing nucleotide peaks in Figure 7A , compared with Figure 7B . These data indicated that the SarZ recognition site may lie in the 48-bp region (tcctatatgtttatatcaataaaatagagtgcaatacagttgtgcatg) and the 41-bp region (atgaaaaatgcaacaaattgagtcaaattaactttatagta) of the psmβ promoter.