Staphylococcus epidermidis Phages Transduce Antimicrobial Resistance Plasmids and Mobilize Chromosomal Islands

Multidrug-resistant strains of S. epidermidis emerge in both nosocomial and livestock environments as the most important pathogens among coagulase-negative staphylococcal species. The study of transduction by phages is essential to understanding how virulence and antimicrobial resistance genes spread in originally commensal bacterial populations.

mass spectrometry (LC-MS/MS) analysis of phage 48 virions identified 12 phageencoded proteins (Table S1). In addition, a capsid morphogenesis protein encoded by the cpm gene (gp14; GenBank accession no. QRX38697) of the newly discovered chromosomal island of S. epidermidis strain SE48 was detected. This points to its effect on capsid remodeling associated with island mobilization, which was first observed and described in S. epidermidis here. The chromosomal island, designated SeCI SE48 , exhibited similarities in architecture and the integrase gene with S. epidermidis fusidic acid resistance islands (14) (Fig. 2).
The genomes of all the phages studied were assembled using short Illumina reads. They were arranged in functional modules, and their size was slightly above 40 kb (Fig. 3A). As no defined termini of genomic DNA were obvious from the Illumina read alignment, phages most likely use a headful packaging mechanism and have circularly permuted genomes with redundant ends (41). The analyzed S. epidermidis phages exhibited high average nucleotide identity (94 to 98%) to each other and 67 to 70% identity to S. aureus phietavirus 80a, suggesting that Phietavirus is a common phage genus in both staphylococcal species (Fig. 4). Comparison of the genomes of phages 27, 456, and 459 revealed high similarity in the virion structure module. Moreover,  48 and phage 48-derived particle with altered capsid architecture due to interaction with SeCI SE48 -encoded proteins. (G) Detail of the baseplate of phage E72. In a specific baseplate conformation, a ring of density is present at the end of the tail (cyan arrow). In total, 13 baseplates in this conformation were observed. (H) Detail of the baseplate of phage 456. An extra ring of density is located above the one present in phage E72 (orange arrow). In total, 15 baseplates in this conformation were observed. Bars, 100 nm; black for images A to F (all at the same scale) and white for images G and H. phages 48 and 27 shared almost identical lysogeny and DNA metabolism modules, while the other modules differed ( Fig. 3A; also, see Table S2 in the supplemental material).
When particular functional genome modules in their respective order were focused on, the following differences were observed. A comparison of the amino acid sequences of the terminase small subunit (TerS), which tends to be one of the least conserved phage proteins (42), divides known S. epidermidis phietaviruses into two subgroups, where phages 27, 48, and E72 represent one and phages 456 and 459 represent the FIG 2 Comparison of phage-inducible chromosomal island SeCI SE48 with selected chromosomal islands. (A) Islands identified in staphylococcal genomes available in GenBank. SeCI SE48 genes are color coded according to their product function, and the corresponding gene product number is given above them. Chromosomal islands obtained from complete genomes of staphylococcal strains as 20-kb sequences starting with the integrase gene are indicated with asterisks. (B) Detailed comparison of SeCI SE48 with SePI fusB-857 and SaPI N315 . Genes are color coded according to their product function. Identity based on tblastx comparison is represented by shaded boxes with a cutoff of 70%. Fišarová et al. other (Fig. 5A). The genes encoding the capsid scaffolding protein (Csp) and the major capsid protein (Mcp) exhibited notable differences when phage 48 is compared with other phages analyzed in this study. Based on the Csp sequence analysis, all of the studied phages except phage 48 grouped into one cluster (Fig. 5B), while they formed two distinct clusters based on Mcp analysis (Fig. 5C). Phylogenetic trees based on alignments of TerS, Csp, and Mcp show reticulate relationships and mosaicism among phietaviruses and the relatedness of S. epidermidis phages to those of S. aureus, especially 71 and X2 (43) and ETA (44), while S. aureus phages NM4 (45) and 80a (46) are more distantly related (Fig. 5).
Phage Transduction in Staphylococcus epidermidis associated with host range, the cell wall hydrolase (Hyd) encoded by the hyd gene, hypothesized to be associated with the baseplate (21), varies slightly across all S. epidermidis phages (Table S2). Besides this, a noncoding intron was predicted in the hyd gene of phage E72 (Fig. 3A). Genes for upper tail fiber (fibU) and a hypothetical protein following the hyd gene are replaced by three diverse genes in phage 456. This is manifested as an extra ring of density above the baseplate ( Fig. 1G and H).
The lysis module comprising genes for holin (hol) and amidase (ami) in phage 48 differs significantly from the other S. epidermidis phages ( Fig. 3A; Table S2). Its amidase gene has only about 50% identity to other S. epidermidis phages but exhibits 78% identity to S. aureus phage B236 (23).
In the S. epidermidis phages studied here, no known genes associated with lysogenic conversion were identified in the region following the lysis genes (Fig. 3A). Phages 27, 456, and 459 harbor the same putative endonuclease gene in this region. In E72, a gene encoding a different putative endonuclease (gp30) is located here and is followed by a gene encoding a hypothetical protein and a gene for a putative membrane-associated protein (gp32). In this region, the phage 48 genome contains three genes, which encode short hypothetical proteins (Table S2).
The S. epidermidis phages studied here have almost identical integrases of the serine-type family (Table S2). An integrase gene with 95% identity was found in a prophage sequence of S. epidermidis NCTC 13924 (GenBank accession no. NZ_LR134536). In the laboratory-lysogenized strains prepared in this study, the genomes of all the phages integrated into the same site in the gene encoding a FAD-dependent oxidoreductase (GenBank accession no. WP_080035152 in strain 1457), which in S. aureus was annotated as a probable pyridine nucleotide-disulfide oxidoreductase family protein, resulting in the split of this gene into two open reading frames (ORFs). The sequence ATATTAAT of the assumed overlap region, where the crossover between attB and attP occurs, is flanked by imperfect inverted repeats (Fig. 3B).
Phage growth characteristics. The host range determined on 35 S. epidermidis and two S. aureus strains (Table 1) and 35 field isolates of seven CoNS species showed that the phages studied here and the previously sequenced phage PH15 (36) were species specific for S. epidermidis alone. Most of the susceptible strains were isolated in the 1970s (39, 47), while recent S. epidermidis isolates were predominantly resistant (17, 48-52) ( Table 1). The broadest host range on S. epidermidis strains (n = 35) was that of phage E72 (34%), followed by 456 (31%) and 27 (26%), and the narrowest were the host ranges of phages 48 and 459 (both 17%). Phage susceptibility testing was conducted by the plate method; however, the ability of phages to effectively propagate on hosts may also depend on the type of culture, i.e., liquid or solid. Strain 1457 did not exhibit sensitivity to phages 27, 48, and 456 (Table 1), but when these phages were incubated with strain 1457 in liquid medium (meat-peptone broth [MPB] or tryptone soy broth [TSB]), they were able to propagate, which enabled their use in transduction experiments. No correlation was found between lytic ability and adsorption kinetics. After 15 min, a portion of the studied S. epidermidis phages remained unadsorbed for both phage-resistant and -susceptible strains.
Mobilization and transfer of a chromosomal island. While the sequencing data of phage 48 propagated on S. epidermidis strain SE48 were being analyzed, the sequence of a new chromosomal island SeCI SE48 was identified. The distribution of its coverage by short Illumina sequencing reads showed that the island was packaged in the phage virions in linear form from the pac site by a headful mechanism. The pac site of SeCI SE48 was predicted at the position between gp12, which encodes an unknown protein, and ptiB, which encodes a phage transcription inhibitor (Fig. 2B). Using the sequencing data of phage 48, the ratio of mean coverage of SeCI SE48 and phage 48 DNA sequences was 1.5 Â 10 22 , which indicates that SeCI SE48 is packaged at high frequency. In the host chromosome, SeCI SE48 is integrated into attB at the 39 end of the groEL gene (Fig. 2B). As no antimicrobial resistance factors encoded by SeCI SE48 were predicted, the transduction of this island by phage 48 into the recipient strain 1457 was performed without antibiotic selection (Table 2), and the transductants were selected by pulsed-field gel electrophoresis (PFGE) analysis (Fig. 6A). Strain 1457 does not contain any chromosomal island adjacent to the groEL gene. After the transfer to strain 1457, the SeCI SE48 integrated at the 39 end of groEL gene, which was not altered due to its integration, as the last 18 bp of the gene overlap the SeCI SE48 att site. PICIs with the same type of integrase and excisionase are localized downstream of the groEL gene in several staphylococcal species, including S. aureus, Staphylococcus lugdunensis, Staphylococcus pasteuri, Staphylococcus warneri, and Staphylococcus haemolyticus ( Fig. 2A). Of the chromosomal islands described above, SeCI SE48 exhibited the highest similarity with SePI fusB-857 , which encodes fusidic acid resistance (12), and a different integrase (Fig. 2B).
The burst size of phage 48 on the transductant strain 1457(SeCI SE48 ) was significantly decreased (4.3-fold) compared to that on wild-type strain 1457. SeCI SE48 integration also led to a loss of sensitivity of strain 1457(SeCI SE48 ) to phages 459 and E72 (Table 1).
Plasmid transductions. All bacterial sequences were obtained using a hybrid assembly of Illumina reads and long Oxford Nanopore reads. All the propagating strains differing in their sequence types (STs) used here carry one or more plasmids, some of which determine antibiotic resistance (Table 3). Phages 27, 48, and 459 transferred naturally occurring plasmids (Table 2; Fig. 6B). Then, a chloramphenicol resistance- Lysis results for phage c : . c The result of each of the three replicates is given. C, confluent lysis; S, lysis; P, individual plaques; L, lysis from without (early bacterial lysis induced by high-multiplicity virion adsorption without phage propagation or abortive infection, manifested by a turbid zone); 2, no lysis. For the C, S, P pattern, the formation of single plaques at higher phage dilution was confirmed for all sensitive strains. encoding plasmid, pBTn (53), electroporated into S. epidermidis strain 1457 was transferred by all the phages except phage 27 into S. epidermidis strains 1457 and A6C ( Table 2; Fig. 6B). The transduction efficiency differed according to the phage and recipient strain used (Table 2). Interspecies plasmid transfer from strains SE27, SE48, SE456, and SE459 by phages able to propagate on them (listed in Table 1) into other staphylococcal species, including three strains of Staphylococcus sciuri, two strains of Staphylococcus chromogenes, two strains of Staphylococcus xylosus, one strain each of Staphylococcus petrasii and Staphylococcus simulans, and seven S. aureus strains, was not successful (data not shown).
Application of transducing phages for genetic studies. Our results showed that the transfer of pBTn via electroporation to the 37 clinical S. epidermidis strains from University Hospital Tübingen tested here failed in 26 cases (70.3%) (Table S3). Therefore, effectively transducing S. epidermidis phages could be considered another laboratory tool for the genetic manipulation of S. epidermidis. Phage E72 exhibited high transducing efficiency (up to 10 24 ) by transferring pBTn from strain 1457 (pBTn) to 21 of 37 (56.8%) S. epidermidis strains, including 10 of 19 (52.6%) strains that could not receive pBTn using electroporation or transduction by phage 187 (18) (Table S3). Moreover, we observed that phage E72 could be propagated to a high titer of ;10 11 PFU/ml, which is an advantage for successful genetic manipulation using transduction (54).

DISCUSSION
A detailed genomic and phenotypic analysis of five S. epidermidis phietaviruses allowed us to assess their transduction potential and putative impact on the evolution of S. epidermidis. Published phietavirus genomes are exclusive to S. aureus and S. epidermidis phages (searched using NCBI Virus). Before this study, phietaviruses amounted to nine of the 13 S. epidermidis siphoviral genomes available in NCBI Virus, which together with the phages described here makes this genus an abundant group of temperate phages in S. epidermidis.
Our results confirmed that generalized transduction in S. epidermidis is mediated by temperate phages using the headful mechanism for DNA packaging (55). Genomes of the studied phages had no obvious pac site, thus resembling S. aureus phage 11 (56). During the packaging initiation, the DNA is probably cut not at a precise location but at scattered locations within a large region of up to several kilobases (57). This relaxed specificity enables the packaging of heterogeneous DNA with pseudo-pac sites into the virion capsids. Differences in pseudo-pac site homology may therefore lead to an altered frequency of transduction. A phage-encoded small terminase subunit (TerS) is essential for phage genome and plasmid packaging into the capsid, but it is not required for the packaging of phage-inducible islands that encode distinct TerS recognizing their own specific pac site (58). We hypothesize that the high efficiency of plasmid transduction by phages 27, 48, and E72 is determined by the same TerS type that is distinct from the other studied phages.
The phage susceptibility of the tested bacterial strains does not always correlate with the adsorption rate of the phage. Therefore, it may depend on many postadsorption factors, as was recently reviewed for S. aureus (59). For generalized transduction, the permissivity of a recipient strain to productive phage infection is not required for receiving foreign DNA; however, cell wall penetration is a necessary prerequisite (29). Differences in baseplate structural proteins may relate to the different ability to infect the host (60). Phage 456 differed from other S. epidermidis phages except for phage CNPx (61) in several baseplate component-encoding genes, including fibU for upper tail fiber protein, which is not essential for either infectivity or assembly of the baseplate in phage 80a (21). We assume it may play a supporting role in host recognition. Phages 456 and E72, which have similar RBP, exhibited the broadest but not identical host ranges. In addition, phage E72 was able to lyse some of the strains resistant to all other tested phages. This could be connected with a small transmembrane domain protein, encoded by gp24 localized in the tail structure region, which could play a role in the host cell envelope penetration via holin regulation.
Horizontal gene transfer in S. aureus is common within clonal lineages, while transfer between them is rare due to specific defense mechanisms, such as restriction-modification systems (62). However, we were able to show that phages 27 and 48 transferred plasmids from S. epidermidis CC5 to CC14. Based on our previous findings on  Table 2. The transferred plasmids in their supercoiled and relaxed forms are indicated with arrows. Fišarová et al. transduction to S. aureus strains insensitive to the lytic action of a transducing phage (29), we analyzed interspecies transfer; however, we found no evidence for such an ability of any of the phages. Nevertheless, the plasmids that were transferred between S. epidermidis clonal complexes show high overall identity to plasmids of Staphylococcus hominis, S. warneri, S. aureus and Macrococcus caseolyticus (Table 3). Only a few previous studies demonstrated plasmid transfer between different staphylococcal species using electroporation (63) or by transduction (64).
To compare the transduction ability of the tested phages, the transfer frequency of the 11.2-kb chloramphenicol resistance plasmid pBTn was assessed. The transduction rates of the tested phages differed on strain 1457 but not on strain A6C, where all the transducing phages transferred pBTn with a higher rate. Only phage 27 was incapable of transduction into strains 1457 and A6C, which may be connected with host surface structures, as was demonstrated in streptococci (65). In naturally occurring plasmids, more efficient transduction of smaller ones was observed, as was described before (66).
The transfer of PICIs has already been well described for S. aureus (SaPIs), but this study provides the first evidence of such a transfer in S. epidermidis. Recently, PICIs have also been predicted in genomes of other Gram-positive cocci (67). They are only distantly related to each other and to S. aureus SaPIs, but they share a genome organization and content. Therefore, they represent convergent evolution that suggests their high selective value. In the previously described staphylococcal PICIs, the accessory genes encode toxins and/or antibiotic resistance determinants (12,13,68), but SeCI SE48 harbors genes for putative membrane proteins. SeCI SE48 is integrated at the 39 end of the host groEL gene, similar to S. aureus and S. epidermidis PICIs with the same type of integrase (12)(13)(14).
The role of the PICI-encoded proteins PtiA, PtiM, and PtiB in the inhibition of phage late transcription through the interaction with phage transcriptional regulator RinA was demonstrated previously (69). The same mechanism was possibly responsible for the decreased phage susceptibility in strain 1457 after SeCI SE48 was integrated into its chromosome. An alteration of the phage life cycle by PICI, leading to the formation of small-headed virions that are unable to carry the entire phage genome, was described (25). In staphylococcal phage 80a, the capsid size change is caused by SaPI1 proteins redirecting the phage capsid architecture from T=7 to T=3 (70). The observed  (6), gene encoding streptomycin aminoglycoside 6-adenyltransferase; fmhA, gene encoding aminoacyltransferase possibly involved in methicillin and lysostaphin resistance; ant(4), gene encoding aminoglycoside nucleotidyltransferase; abc, gene encoding ABC transporter (may be involved in drug resistance; multiple abc genes are present in appropriate plasmids). c If more than one bacterial species is present, one plasmid is given for each of them; plasmids with overall nucleotide sequence identity less than 75% are noted as novel. NA, not performed. d Nearest ST; no match with the arcC gene was found.
formation of small-headed virions in phage 48 suggests a similar architecture shift in S. epidermidis. The primary sequence and predicted secondary structure of the SeCI SE48encoded protein, involved in the capsid assembly of small-headed virions, differs from those in SaPI1 (70). This suggests that SeCI SE48 uses a different mechanism of phage capsid resizing.
The homologues of phage 80a-encoded PICI derepressors, such as dUTPase, Sri, and ORF15, are encoded by genes in the DNA metabolism module and play an essential role in chromosomal island mobilization (27,71). In S. epidermidis phages 27 and 48, homologues of these PICI mobilization determinants are identical (Fig. 3), but these phages differ significantly in their capsid proteins. While the minor capsid protein has been shown to not affect plasmid or PICI transduction (72), the role of other capsid proteins has been only partially described (70). In phage 48, the capsid scaffolding protein was distinct from all the tested phages. High-frequency transducing phages 48 and E72 encode almost identical major capsid proteins that are very different from all other phages. Thus, the capsid proteins seem to play an important role not only in the mobilization of islands but also in the transfer of plasmids, where the successful transduction is dependent on a complex of genetic determinants of both the mobile element and the phage.
The transfer of plasmids to bacterial cells is an important technique used in research in S. epidermidis. However, far fewer clinical S. epidermidis than S. aureus isolates can be transformed by electroporation, and the efficiency is orders of magnitude lower, despite attempts at optimization (73,74). While there are many studies of phage-mediated plasmid transduction in S. aureus, there is a lack of similar studies in S. epidermidis. Recently, Winstel et al. (18) established a plasmid transfer method using phage 187, which is capable of efficient transduction to CoNS strains and specific S. aureus strains with CoNS-type wall teichoic acid. The phages used in our study also exhibited a high efficiency of plasmid transduction in strains where phage 187 failed (Table S3). They even transferred relatively large plasmids that generally cannot be easily transferred via electroporation or by phage 187. Therefore, the phages described in this study, especially E72 and 48, could be considered suitable laboratory tools for the transduction of plasmids and chromosomal islands to S. epidermidis strains, which could promote future CoNS research.
Hájek (Palacký University, Olomouc, Czech Republic). Thirty-seven clinical S. epidermidis isolates were obtained from the Medical Microbiology department of the University of Tübingen (Table S3). Phage E72 was induced from S. epidermidis strain E72, a clinical isolate from infected teeth of a patient in University Hospital Tübingen, and propagated on S. epidermidis strain 1457 (49). S. aureus phages 11 and 80a (75) were used in the adsorption assay. Phage 187 (18) was used for the comparison of pBTn transduction efficiency with phage E72.
For the determination of growth properties of tested phages, a collection of S. epidermidis strains was used (Table 1). S. aureus strains 8325-4 (17) and RN4220 (51) and 35 field isolates of 7 CoNS, including S. chromogenes, S. hominis, S. haemolyticus, S. petrasii, S. sciuri, S. simulans and S. xylosus, were used in adsorption and/or host range assays. In addition to natural plasmids, the transposon plasmid pBTn, encoding chloramphenicol resistance (53), was used in a phage transduction assay. All the assays with pBTn were performed at 30°C to avoid loss of the plasmid. The pBTn was transferred to strain 1457 via electroporation, and this strain then served as the donor.
Growth medium and phage propagation. S. epidermidis phages were propagated on their propagation strains at 37°C on 1.5% meat peptone agar (MPA), prepared from 13 g nutrient broth CM0001 (Oxoid), 3 g yeast extract LP0021 (Oxoid), 5 g peptone LP0037 (Oxoid), 15 g agar LP0013 (Oxoid), and distilled water to a final volume of 1 liter (pH 7.4), overlaid with a top soft layer of MPA (0.7% agar) containing 2 mM CaCl 2 . The lysates were centrifuged at 5,000 Â g for 15 min and filtered using a polyethersulfone membrane filter with a pore size of 0.45 mm (TPP, Switzerland).
Morphological and proteomic characterization. For electron microscopy, particles of phages E72, 27, 456, and 459 were purified in a CsCl density gradient (16). The transducing lysate of phage 48 was purified by fast protein liquid chromatography (FPLC) and ultrafiltration as described previously (76) with minor modifications. For the FPLC purification, a CIMmultus QA 8-ml monolithic column (Bia Separations, Slovenia) and NGC chromatography system (Bio-Rad, USA) were used. Phage lysate was mixed with 40 mM Tris buffer (pH 7.5) in a 1:1 ratio at a final NaCl concentration of 0.35 M. For Fišarová et al. ultrafiltration, Pellicon XL50 cassettes with a Biomax 300-kDa membrane (Millipore, USA) were used. Retentate and fractions containing phages were desalted and concentrated using Amicon Ultra 0.5-ml centrifugal filters (Millipore, USA) with a 100-kDa cutoff. Phages were diluted to an A 280 of 0.5 to 1.0, and negatively stained samples were prepared by applying 4 ml of the diluted sample onto copper grids coated with a 12-nm carbon layer, stained with 2% uranyl acetate. Samples were observed using a Tecnai F20 electron microscope (Thermo Fisher Scientific, USA) operated at 200 kV with Â50,000 magnification.
Structural proteins of phage 48 and phage 48-derived small-headed particles purified in a CsCl gradient were determined by LC-MS/MS analysis performed using the Ultimate 3000 RSLCnano system (Thermo Fisher Scientific, USA) connected to an Impact II Qq time-of-flight mass spectrometer (Bruker, Bremen, Germany) as described previously (64). An in-house database of proteins encoded by strain SE48 and phage 48 was used for the final search. Only proteins that were identified based on at least two peptides were reported. Detected bacterial metabolic proteins were filtered out from the final report.
Plasmid DNA was isolated from the transductants with a NucleoSpin plasmid kit (Macherey-Nagel, Düren, Germany) according to the manufacturer's protocol using prolonged lysis with lysostaphin as described before (78). Plasmid DNA was analyzed by agarose gel electrophoresis in 1.2% agarose gel (Serva) in 1Â Tris-acetate-EDTA (TAE) buffer. SmaI (New England BioLabs, Ipswich, MA, USA) macrorestriction analysis by pulsed-field gel electrophoresis (PFGE) was performed using a CHEF Mapper XA system (Bio-Rad) to confirm the transfer and integration of SeCI SE48 .
Electroporation. Electroporation was conducted according to the transformation method for S. epidermidis described by Monk et al. (79). Briefly, overnight S. epidermidis cultures with an optical density at 600 nm (OD 600 ) of 0.5 were chilled on ice, washed in 10% glycerol, and 50 ml was aliquoted to each tube. Before electroporation, competent cells were resuspended in 50 ml of 10% glycerol and 500 mM sucrose. A volume of 5 ml plasmid pBTn DNA (500 ng/ml) was added to the bacterial cells, transferred to a 1-mm electroporation cuvette (Bio-Rad), and then pulsed at 21 kV/cm, 100 X, and 25 mF. TSB supplemented with 500 mM sucrose was added to the electroporated bacteria, incubated at 30°C for 1 h before plating on TSA plates with chloramphenicol (10 mg/liter), and incubated at 30°C for 48 h.
Determination of host range, phage adsorption, and one-step growth curve. Phage lysates were spotted onto the examined strains at 10Â, 100Â, and 1,000Â routine test dilutions (RTD) (80). Productive phage infection was confirmed by the formation of single plaques. Adsorption curves of the phages were determined as described previously (81) using an input ratio (IR) of 10. The burst size of phage 48 on bacterial strains 1457 and 1457(SeCI SE48 ) was determined from the one-step growth curve as described previously (82) with minor modifications. Briefly, overnight bacterial culture was cultivated to an OD 600 of 0.5 (10 8 CFU/ml). Bacterial cultures were synchronized at 10°C for 30 min. The phage lysate was mixed with the bacterial culture at an IR of 0.01, and the mixture was incubated at 37°C. The titer of free phage particles was determined after 0 to 90 min at 5-min intervals.
Statistical analysis. Plasmid transduction efficiency and phage 48 burst size comparisons were evaluated by statistical analysis using the t test. A P value of 0.05 was used as the threshold for statistical significance.
Genome analysis. For whole-genome sequencing on the Illumina platform, the phage DNA was isolated using a phenol-chloroform method (83), and bacterial DNA was isolated from cultures cultivated in MPB using a High Pure PCR template preparation kit (Roche Diagnostics, Mannheim, Germany) according to the manufacturer's instructions with the modification of adding lysostaphin (40 mg/ml, Sigma-Aldrich) for cell lysis. The 500-bp sequencing library was prepared with a NEBNext Ultra II DNA library prep kit for Illumina (New England BioLabs). The samples were sequenced using a MID output cartridge in a 150-bp paired-end mode on an Illumina NextSeq sequencing platform (Illumina, San Diego, CA, USA). The quality of sequencing reads was analyzed with FastQC v0.11.8 (84). Bases of lower quality and adapters were trimmed using the sliding window model in Trimmomatic Galaxy v0. 36.5 (85).

SUPPLEMENTAL MATERIAL
Supplemental material is available online only.