Genomic characterization of Staphylococcus aureus isolated from patients admitted to intensive care units of a tertiary care hospital: epidemiological risk of nasal carriage of virulent clone during admission

ABSTRACT We conducted a molecular epidemiological study of Staphylococcus aureus using whole-genome sequence data and clinical data of isolates from nasal swabs of patients admitted to the intensive care unit (ICU) of Hiroshima University hospital. The relationship between isolate genotypes and virulence factors, particularly for isolates that caused infectious diseases during ICU admission was compared with those that did not. The nasal carriage rates of methicillin-resistant S. aureus (MRSA) and methicillin-susceptible S. aureus (MSSA) in patients admitted to the ICU were 7.0% and 20.1%, respectively. The carriage rate of community-acquired (CA)-MRSA was 2.3%, accounting for 32.8% of all MRSA isolates. Whole-genome sequencing analysis of the MRSA isolates indicated that most, including CA-MRSA and healthcare-associated (HA)-MRSA, belonged to clonal complex (CC) 8 [sequence type (ST) 8] and SCCmec type IV. Furthermore, results for three disease foci (pneumonia, skin and soft tissue infection, and deep abscess) and the assessment of virulence factor genes associated with disease conditions [bacteremia, acute respiratory distress syndrome (ARDS), disseminated intravascular coagulopathy (DIC), and septic shock] suggested that nasal colonization of S. aureus clones could represent a risk for patients within the ICU. Particularly, MRSA/J and MSSA/J may be more likely to cause deep abscess infection; ST764 may cause ventilation-associated pneumonia, hospital-acquired pneumonia and subsequent bacteremia, and ARDS, and tst-1-positive isolates may cause DIC onset. IMPORTANCE Nasal colonization of MRSA in patients admitted to the intensive care unit (ICU) may predict the development of MRSA infections. However, no bacteriological data are available to perform risk assessments for Staphylococcus aureus infection onset. In this single-center 2-year genomic surveillance study, we analyzed all S. aureus isolates from nasal swabs of patients admitted to the ICU and those from the blood or lesions of in-patients who developed infectious diseases in the ICU. Furthermore, we identified the virulent clones responsible for causing infectious diseases in the ICU. Herein, we report several virulent clones present in the nares that are predictive of invasive infections. This information may facilitate the design of preemptive strategies to identify and eradicate virulent MRSA strains, reducing nosocomial infections within the ICU.

trauma patients admitted to emergency rooms (ER) or intensive care units (ICU) (1).In Japan, the mortality rates of sepsis caused by S. aureus and MRSA were 21% and 25% in 2017, respectively, with death tolls of 17,157 and 4,224, respectively (2).Furthermore, 18%-31% of VAP and 10%-25% of CRBSI cases in Japan are caused by S. aureus (1).Hence, the prevalence of MRSA nasal colonization in patients admitted to the ICU may predict MRSA infections in the ICU (3,4).Indeed, active MRSA surveillance increases the chances of contact precautions and decreases empirical treatment (4,5).Moreover, early intervention for MRSA colonization using mupirocin may reduce the burden of severe MRSA infection; however, this may also induce mupirocin resistance in colonized S. aureus (6)(7)(8).
General intervention strategies are associated with significant economic costs to hospitals and patients.Therefore, screening for highly virulent clones and selective decolonization may contribute to the control of severe infection onset and diminish the risk of emerging antimicrobial resistance.
S. aureus displays diverse genotypes, and the genotypes of major clones differ between countries (9)(10)(11)(12).Previous studies have indicated that regional prevalence of clones is changing (9,13,14); thus, molecular epidemiology studies may be useful for risk assessment, treatment selection, and information update.Few large-scale epidemio logical studies on S. aureus infection in ICU are available, and no such study has been conducted using whole-genome sequencing data (10)(11)(12)(13).
We conducted a molecular epidemiological study using the whole-genome sequences of isolates collected via nasal screening patients admitted to the ICU in a teaching hospital.To highlight high-risk clones in the ICU, we compared the relationship between genotypes and virulence determinants of isolates causing infectious diseases and those from in-patients without infectious diseases.

Isolation of S. aureus
A total of 2,456 patients were admitted to ICU of Hiroshima University Hospital between 1 June 2017 and 31 May 2019 (Fig. 1).Of these patients, 505 (20.6%) were excluded from the study, 161 (6.6%) were readmitted, and 344 (14.0%) did not undergo nasal swabbing due to discharge or death within 48 h of admission.
In addition, 92/2,456 patients (3.8%) tested positive for S. aureus infection upon or after admission, and 98 isolates from lesions or blood cultures were identified as S. aureus.Among the 2,362 patients who tested negative for S. aureus infection, 486 (20.6%) were positive for S. aureus using nasal swabs and were defined as having non-infectious S. aureus.

Molecular characterization of S. aureus isolated from different sources
Whole-genome sequencing analysis was conducted for 583 S. aureus isolates; 467 were obtained from initial nasal swabs upon admission (122 MRSA and 345 MSSA).Of the remaining 116 specimens, 85 were collected from infection sites (34 MRSA and 51 MSSA), of which 37 were ICU-acquired MRSA and 6 were ICU-acquired MRSA from the infection site.Of the initial nasal specimens, 431 were collected from patients who did not present with S. aureus infection (asymptomatic) during their ICU stay (107 MRSA and 324 MSSA).Multilocus sequence typing (MLST) and SCCmec typing were performed using the corresponding sequencing data.
MLST of MSSA isolates from the initial screening of nasal swabs and patients who did not experience S. aureus infection during their ICU stay revealed various sequence types, including CC8, CC45, CC188, CC15, and CC30 (Fig. 2B).In contrast, the MLST patterns of MSSA isolates from the lesions or blood of patients with infectious MSSA included CC8  ).An increased ratio of CC8 (infection 27.5%, initial nasal swab 17.7%) and CC5 (infection 11.8%, initial nasal swab 5.5%) was observed.Overall, ST380, ST764, and ST1 were unique to MRSA, while ST508, ST45, ST188, and ST15 were unique to MSSA (Fig. 2A and B).

Prevalence of VFGs and ARGs in S. aureus isolated from nasal swabs and infection sites
A whole-genome-based phylogeny of 583 S. aureus draft genomes was reconstructed using kSNP3.0 to characterize the population structure of ICU isolates with the metadata for infection foci, ST, CC, virulence factor genes (VFGs), and antimicrobial-resistant genes (ARGs) (Fig. 5).The prevalence of VFGs and ARGs in major STs isolated from nasal swabs and infection sites was also explored (Fig. 6).
The common gene sets of ST5 MRSA from the initial nasal swab and the infection site were sec, tst-1, seg, sei, sel, sem, sen, seo, and sep (Fig. 6).The positivity of the gene set was 50%-90% in ST5 MRSA from the initial nasal swab and 100% in ST5 MRSA from the infection site, except sel (Fig. 6).A similar gene set was present in ST5 MSSA from the initial nasal swab and infection site; however, the positive ratio of tst-1 was notably lower than that of ST5 MRSA (5% vs. 92% in the initial nasal swab and 17% vs. 100% in the infection site).New York/Japan clone (ST5-SCCmecII) accounted for only 6.6% (8/122) of initial MRSA nasal carriage (Table S3), but it accounted for 66.7% of ST5 (8/12) (Fig. 5).The ST764 isolates were all MRSA, and most possessed a common set of virulence genes, including psm-mec, seb, seg, sei, sem, sen, and seo (Fig. 5 and 6).All ST764 and ST5-SCCmecII MRSA (New York/Japan clone) isolates were positive for ant(9)-Ia, erm(A), and tet(M).Some ST764 isolates also possessed fosD and qacB.The common gene sets of ST30 MRSA and ST30 MSSA from the initial nasal swab and infection site were seg, sei, sem, sen, seo, and seu (Fig. 5 and 6).Most of the ST30 MSSA isolates from the infection site possessed tst-1, and 100% of the ST30 MRSA isolates from the infection site possessed pvl although the total number was small (Fig. 6).MRSA isolates of major STs, except ST30, possessed a significantly higher ratio of aminoglycoside resistance genes, erm(A) and tet(M), suggesting multiple resistance (Fig. 5 and 6).A higher ratio of qacA positivity was also observed in ST380 MRSA and ST8 SCCmecI MRSA (Fig. 5).

DISCUSSION
This large-scale epidemiology study investigated the presence of S. aureus in the nasal cavity of more than 2,000 patients admitted to the ICU.The overall carriage rates of MRSA and MSSA were 7.0% and 20.1%, respectively.CA-MRSA accounted for 32.8% (45/137) of all MRSA isolates.MLST revealed that most MRSA isolates belonged to CC8, followed by CC5, suggesting a transition of the major CC in MRSA.Several small-scale domestic studies have reported an increase in CC8 (ST8) levels in MRSA (16)(17)(18)(19)(20).In parallel with the CC transition, the dominant SCCmec type changed from SCCmecII to SCCmecIV.Although the most predominant MRSA clone was previously CC5-ST5-SCCmecII (New York/Japan clone) (9), this clone accounted for only 6.6% of the MRSA in the current study (Table S3).Surveillance of a Japanese tertiary hospital in 2015 demonstrated that SCCmecII decreased from 90.0% to 74.3% (21).Meanwhile, in 2017, the proportions of SCCmecIV and SCCmecII in a tertiary care hospital were 52.3% and 28.5%, respectively (22), indicating an increase in SCCmecIV prevalence.The predomi nance of SCCmecIV was also observed in CA-MRSA, HA-MRSA, and MRSA isolated from nasal swabs of patients during their ICU stay (Fig. 3 and 4).
The Japan Nosocomial Infection Surveillance data indicate that the isolation rate of MRSA (computed as the number of patients with positive MRSA culture over the total number of patients from whom specimens were collected) was 6.4%-6.6%(2015-2019) (23), which was comparable to the rate obtained in this study.In comparison, Panayotis et al. (3) reported that the carriage rates of MRSA in ICU patients in South Europe, North/ Central Europe, North America, Asia, and South America were 3.5%, 4.4%, 8.8%, 12%, and 13.1%, respectively, suggesting that the carriage rate of MRSA in this study is closer to that in North America or Europe than in Asia.Additionally, a systematic review and meta-analysis performed by Wong et al. (24) reported that the ratio of CA-MRSA to total MRSA in a hospital setting was 0.7%-10.4%;the ratio of CA-MRSA in the current study was within this range.Previously, the classification of MRSA into CA-MRSA and HA-MRSA using clinical standards supported the prediction of the SCCmec type and characteristics of the isolate, such as virulence factor repertoire and antimicrobial resistance (25).However, the results of the present study indicated that CC8 was the predominant group in CA-MRSA, HA-MRSA, and ICU-acquired MRSA, with predominant SCCmec type IV (Fig. 4), suggesting that such a clinical classification will no longer provide further insights into the virulence of isolates based on clonal differences between CA-MRSA and HA-MRSA.
ST8 was the most prevalent ST among the isolates from the infection sites, particularly from deep abscesses (Fig. 7A).Of these isolates from the infection site, 28.6% was MRSA/J, an endemic clone that causes severe invasive infectious diseases (Table 1) (15,26,27).Meanwhile, MSSA/J accounted for approximately 33.3% of the ST8-MSSA isolates from the infection site.Hence, MRSA/J and/or MSSA/J were implicated in deep abscesses (Fig. 7B).Furthermore, these strains were isolated from nasal swabs of patients in the ICU, suggesting that nasal carriage of these clones could be a potential risk factor for severe invasive skin infections.Most CC5 isolates belonged to ST5 or ST764 (Fig. 5).CC5 isolates carried the egc cluster (seg, sei, sem, sen, and seo) and were frequently isolated from pneumonia and SSTI foci.Among the ST5 clones isolated from the infection site, only two isolates belonged to the authentic New York/Japan clone (ST5-SCCmecII carrying tst-1).However, these two strains caused DIC, suggesting that the New York/Japan clone maintains high virulence.The ST764 was first reported as a variant of ST5 in a 2001-2005 study in Japan (28), and several subsequent reports have been published (16,29,30).In our study, the primary focus of the ST764 isolates was pneumonia (Fig. 7A).Similarly, in 2017, Kaseida et al. (15) reported that ST764 is associated with pneumonia.Moreover, the assessment of virulence factor genes with disease conditions in our study suggested that bacteremia was associated with seb (Table 2; Table S2).The major ST carrying seb identified in this study was ST764.In contrast, ARDS is associated with mecA and psm-mec.psm-mec was present only in the ST764 and New York/Japan clones (Fig. 5).These results suggest that nasal colonization by ST764 may be a significant threat to HAP/VAP and the subsequent cause of bacteremia and ARDS in the ICU.CC30 (ST30) commonly carried the egc cluster, and most ST30-MRSA possessed tst-1 (Fig. 5).Although only a few were detected, all ST30 MRSA isolates from the infection site possessed pvl.PVL is a cytolytic toxin composed of a protein complex of the lukS-PV and lukF-PV gene products and is as an etiological agent of necrotizing pneumonia (31).USA300 is a well-known CA-MRSA of ST8 carrying pvl (15,30); however, in Japan, the MRSA-positive detection rate for pvl is only 0.7% (25).In this study, only three isolates were positive for pvl and all belonged to ST30; however, all were isolated from the infection site.ST380 is classified as CC8 and approximately 80% of ST380 cases are MRSA carrying SCCmecIV.ST380 MRSA was the second most common isolate detected during our initial screening of nasal swabs (Fig. 2).ST380 MRSA has only been reported in Japan, suggesting that it is an endemic clone (19,32).ST380 MRSA carries qacA, a resistance gene for organic cationic compounds, such as benzalkonium chloride and chlorhexidine, which may pose a risk of spreading antiseptic resistance (Fig. 4) (33,34).Although the frequency of isolation from the infection site was low, the high rate of nasal carriage raised concerns about environmental contamination and clonal spread (Fig. 2 and 7A).This study had several limitations.First, this was a single-center study involving the ICU department and did not, therefore, reflect nationwide epidemiology.Second, 16.5% of enrollments dropped out, and only nasal swabs were used to screen for S. aureus.Third, we did not examine all factors associated with patient risk.Nevertheless, this is the first large-scale epidemiological study of S. aureus infection in an ICU in Japan.While previous molecular epidemiological studies have focused on isolates from the infection foci, we attempted to include all isolates detected via carriage screening upon patient admission in our molecular epidemiological studies, better reflecting the true epidemiology of S. aureus infection in the ICU department.

Sample collection
S. aureus was isolated from clinical specimens collected from patients admitted to the ICU at Hiroshima University Hospital from 1 June 2017 to 31 May 2019.Nasal swabs were collected from hospitalized patients within 24 h of, and every week during, admission.For patients admitted several times, the isolates from the first visit were used.To characterize S. aureus-caused infectious diseases, S. aureus was isolated from blood cultures or lesions at the infection site of patients admitted to ICU.When several isolates were obtained from the same lesion or multiple blood cultures of the same patient, the first isolate was selected for analysis.S. aureus isolates from patients diagnosed with sepsis/septic shock, DIC, and ARDS were compared with those from other studies (35)(36)(37).For clinical classification, HA-MRSA was defined as MRSA isolated within 48 h of admission and fulfilling at least one of the following criteria: patient with previous MRSA-positive infection, admitted within the past year, stayed in a long-term care facility, and underwent surgery, hemodialysis, or indwelling catheter.Isolates from patients who tested negative for nasal swab upon admission but tested positive after 24 h were classified as ICU-acquired MRSA, while other isolates were defined as CA-MRSA (38,39).S. aureus was identified following the standard protocol of the VITEK2 GP Identification Card (bioMerieux Japan, Tokyo).

Construction of phylogenetic tree
kSNP3.021 (41) was employed to identify the pan-genome SNPs of the 583 S. aureus isolates and estimate the parsimony tree based on these SNPs.kSNP3.021algorithm identifies SNPs based on unique stretches of nucleotides present in all genomes having SNPs in their middle position.The optimal size of the nucleotide regions flanking the SNPs (k-mer) was identified using the program Kchooser available with the package.A k-mer size of 19 was applied to identify SNPs.The 249,203 SNP sites were extracted from the pan-genome alignment.The SNP matrix file (SNPs_all_matrix.fasta,DOI: 10.6084/ m9.figshare.25526326)generated by this tool was used to produce a parsimony tree.The consensus tree was estimated from equally parsimony trees with 100 boostraps and visualized by FigTree v.1.4.4 (https://github.com/rambaut/figtree/).

Statistical analysis
Data analysis was performed using JMP Pro (version 16.0.0).Pearson's χ 2 test or Fisher's exact test were used to compare the genotypes and virulence factors of infectious and non-infectious S. aureus, as well as identify factors involved in sepsis, ARDS, DIC, and bacteremia.Multivariate analysis was performed to identify factors associated with bacteremia, DIC, ARDS, and septic shock.A P value < 0.05 was considered statistically significant.

ETHICS
This study was approved by the Ethics Committee of Hiroshima University (E-1784).

ADDITIONAL FILES
The following material is available online.

FIG 1
FIG 1 Schematic diagram of S. aureus isolation.

FIG 3
FIG 3 Proportion of SCCmec types in each category.

FIG 4
FIG 4 Proportion of (A) STs and (B) SCCmec types in HA-MRSA and CA-MRSA.

FIG 5
FIG 5 Pan-genome single nucleotide polymorphism (SNP)-based phylogenetic analysis of 583 S. aureus isolates in this study.A parsimony tree was constructed based on the pan-genome 249,203 SNP sites with branch lengths expressed in terms of changes per number of SNPs using kSNP3.021.Heatmap: column 1, presence or absence of infection; column 2, disease focus such as pneumoniae, SSTI, and deep abscess; column 3, each color represents an ST; column 4, each color represents a CC; column 5, methicillin-resistant S. aureus (MRSA; red); column 6, SCCmec types of MRSA; columns 7 and 8, the presence/absence of VFGs and ARGs, with black representing the presence of genes.MRSA/J is an MRSA clone belonging to CC8 (ST8), and SCCmec type is IVl and possesses tst-1.MSSA/J is an MSSA clone belonging to CC8 (ST8) and possesses tst-1.New York/Japan is an MRSA clone belonging to CC5 (ST5), and SCCmec type is II and possesses tst-1.

FIG 6
FIG 6 Characteristics of VFG or ARG patterns of representative STs of S. aureus isolates on nasal survey and infection groups.The proportion of several VFGs and ARGs of each ST is presented using heatmaps.The numbers in the heatmap columns are shown as percentages.R, MRSA; S, MSSA.

FIG 7
FIG 7 Characteristics of (A) ST and (B) VFG patterns of S. aureus isolates on pneumoniae, SSTI, and deep abscess.Arrows at both ends of the outer ring on (A) pie chart indicate CCs.(B) The numbers in the VFGs heatmap are shown as percentages.

TABLE 1
Proportion of MRSA/J or MSSA/J in ST8 each category a a Numbers in brackets indicate percentages.b Not detected.

TABLE 2
Prevalence of MRSA and VFGs in bacteremia, ARDS, DIC, and septic shock a