Multidrug-Resistant Methicillin-Resistant Staphylococcus aureus Associated with Bacteremia and Monocyte Evasion, Rio de Janeiro, Brazil

We typed 600 methicillin-resistant Staphylococcus aureus (MRSA) isolates collected in 51 hospitals in the Rio de Janeiro, Brazil, metropolitan area during 2014–2017. We found that multiple new clonal complex (CC) 5 sequence types had replaced previously dominant MRSA lineages in hospitals. Whole-genome analysis of 208 isolates revealed an emerging sublineage of multidrug-resistant MRSA, sequence type 105, staphylococcal cassette chromosome mec II, spa t002, which we designated the Rio de Janeiro (RdJ) clone. Using molecular clock analysis, we hypothesized that this lineage began to expand in the Rio de Janeiro metropolitan area in 2009. Multivariate analysis supported an association between bloodstream infections and the CC5 lineage that includes the RdJ clone. Compared with other closely related isolates, representative isolates of the RdJ clone more effectively evaded immune function related to monocytic cells, as evidenced by decreased phagocytosis rate and increased numbers of viable unphagocytosed (free) bacteria after in vitro exposure to monocytes.

We typed 600 methicillin-resistant Staphylococcus aureus (MRSA) isolates collected in 51 hospitals in the Rio de Janeiro, Brazil, metropolitan area during 2014-2017. We found that multiple new clonal complex (CC) 5 sequence types had replaced previously dominant MRSA lineages in hospitals. Whole-genome analysis of 208 isolates revealed an emerging sublineage of multidrug-resistant MRSA, sequence type 105, staphylococcal cassette chromosome mec II, spa t002, which we designated the Rio de Janeiro (RdJ) clone. Using molecular clock analysis, we hypothesized that this lineage began to expand in the Rio de Janeiro metropolitan area in 2009. Multivariate analysis supported an association between bloodstream infections and the CC5 lineage that includes the RdJ clone. Compared with other closely related isolates, representative isolates of the RdJ clone more eff ectively evaded immune function related to monocytic cells, as evidenced by decreased phagocytosis rate and increased numbers of viable unphagocytosed (free) bacteria after in vitro exposure to monocytes.
The sample comprised roughly equal numbers of isolates from blood samples from BSI patients, nonblood samples from patients with infections at another body site, and nasal swab samples; samples were collected during 2014-2017, most in 2015 and 2016. Patient age was available for 450 patients ( Table 2). The research protocols were submitted to the Human Research Ethics Committee (CAAE submission no. 41614914. 4.00005257) of the Hospital Universitário Clementino Fraga Filho, Universidade Federal do Rio de Janeiro (Rio de Janeiro, Brazil); the study was considered non-human subject research.

Molecular Typing and Susceptibility Testing
We used restriction-modification (RM) tests to determine CC (10) and multiplex PCR to type SCCmec (11).
We used PCR to screen for the lukSF-PV, agrII, SCC-mecIII, and seh genes as previously described (12). We conducted antibiogram and susceptibility tests for glycopeptide drugs as recommended by Clinical and Laboratory Standards Institute guidelines (13).
To estimate when the ST105-SCCmecII-t002 lineage emerged in Rio de Janeiro, we used a Bayesian phylogenetic framework to analyze 73 genomes that passed our Mash Screen (17) quality cutoffs. We selected MRSA strain FCFHV36, the closest complete reference genome available in GenBank, using the WhatsGNU topgenome (-t) option (18). We used the SNP alignment to infer an initial phylogenetic tree in RAxML version 8.2.4 before using ClonalFrameML (19) to detect and mask areas of recombination. We used the SNP recombination-masked alignment to estimate divergence times in BEAST version 2.6.2 (20). We found a positive correlation between genetic divergence and isolation time using TempEst version 1.5.3 (21). We plotted the chronograms based on the maximum clade credibility tree using the TreeAnnotator program and visualized in FigTree version 1.4.3 (Appendix 3).

Genomic Island Characterization
We used Geneious Prime version 2020.1.2 to manually inspect the ΦSA3, vSa-α, vSa-β, vSa-β, and SaPI-1 genomic islands (22,23) and Swiss-Prot (Uniprot Consortium, https://www.uniprot.org) to annotate paralogues. To map the genetic context of genomic islands, we randomly selected representative genome sequences from different phylogenetic locations of the tree showing the most common CC5 lineages in the Rio de Janeiro metropolitan area ( Figure 1). We determined gene presence or absence using BLAST analysis (https://blast.ncbi. nlm.nih.gov).

Phagocytosis Assays
We subjected the selected isolates to phagocytosis (Appendix 3 Table 1). In this assay, we considered the entire process of phagocytosis (i.e., binding and uptake) by detecting all cell-associated bacteria, whether internalized or externally attached, after washing. We cultured bacteria at 37°C for 18 h at 250 rpm in brainheart infusion broth (Becton Dickinson, https:// www.bd.com) before treating with 25 nmol SYTO 9 stain (Thermo Fisher Scientific, https://www.thermofisher.com) for 15 min and washing in phosphatebuffered saline (1× phosphate-buffered saline, pH 7.2). We incubated bacterial cells at 37°C for 30 min in 5% carbon dioxide with THP-1 monocytes in Roswell Park Memorial Institute 1640 medium for a multiplicity of infection of 10 (24). We did not use antimicrobial drugs at any time during these assays. We washed the infected monocytes with PBS once and then centrifuged them at 200 × g for 5 min. We resuspended THP-1 cells in PBS and analyzed them by flow cytometry (FACSCalibur; Becton Dickinson). We acquired 10,000 live THP-1 cells (as calculated by forward scatter and side scatter gating) and analyzed data using FlowJo10 software (https://www.flowjo. com). We calculated the number of bacteria-associated THP-1 cells as the frequency of fluorescent (i.e., SYTO 9-positive) THP-1 cells compared with total live THP-1 cells. In addition, we counted and compared the number of viable unphagocytosed bacterial cells in the culture supernatant of each assay at 0 and 30 min after incubation.

Statistical Analyses
We analyzed molecular typing, antimicrobial testing, and epidemiologic data using Pearson χ 2 tests.
To assess the association of the CC5-SCCmecII group and the ST105-SCCmecII-t002 sublineage with BSI, we used Stata 16.0 (https://www.stata.com) to conduct a Mantel-Haenszel test stratified on a composite variable informed by participant age (>60 years vs. <60 years), year of specimen collection (2014, 2015, or 2016-2017), and hospital type (public vs. private). For the analysis of year of specimen collection, we combined data from 2016 and 2017 because few isolates were collected during 2017. We analyzed phagocytosis assays using a 1-way analysis of variance and Tukey multiple comparison test in GraphPad Prism 6 (GraphPad Software, Inc., https://www.graphpad.com).

Distribution of Genotypes (CC-SCCmec) and Clinical Data
In the univariate analysis, we found that the distribution of genotypes was associated with MRSA infection site (Figure 2,  In addition to being the most frequent MRSA clone, RdJ might be responsible for the higher frequency of CC5-SCCmecII isolates from blood samples. ST105-SCCmecII-t002 isolates were more common among

Whole-Genome Phylogenetic Analysis of MRSA CC5 Isolates
The whole-genome phylogenetic analysis grouped CC5 isolates from this study into 3 of the 4 major phylogenetic groups corresponding mostly to the ST105(CC5)-SCCmecII-t002, ST5-SCCmecII-t539, and ST5-SCCmecIV-t002 genotypes and distributed widely throughout the CC5 tree ( Figure 1). All SC-CmecIV isolates clustered in the CC5-Basal clade. Isolates with the multidrug-resistant ST5(CC5)-SCCmecII-t539 genotype clustered with members of the paraphyletic group CC5-IIA described by Challagundla et al. (8). Most other CC5 isolates, including isolates of genotype ST105(CC5)-SCCmecII-t002, were grouped in clade CC5-IIB. Although most of these isolates form the RdJ clade, which is found mostly in Rio de Janeiro, nearby outgroups to this clade are composed of previously sequenced isolates from São Paulo and Porto Alegre (25) and North America, as well as a few isolates from this study In comparison with other CC5 genomes, the clade that includes the ST105 genomes lacked key virulence genes. In addition to the apparent loss of the enterotoxin P gene (sep) noted by Challagundla et al. (8), isolates from this clade uniformly lacked the splD gene encoding serine protease D (Figure 1).

Discussion
Using molecular typing and phylogenetic analysis, we identified a third epidemic lineage of MRSA in Rio de Janeiro. CC5, and to a lesser extent CC30, have become the most prevalent MRSA lineages in Rio de Janeiro hospitals, replacing the previously dominant ST1(CC1)-SCCmecIV lineage, which had replaced the BEC lineage ST239(CC8)-SCCmecIII during 2004-2008 (4). At the time when the ST1(CC1)-SCCmecIV lineage replaced BEC, CC5 comprised only 10% of isolates (4); CC5 now constitutes >60% of isolates. Previously dominant clones, especially BEC, carried resistance to many non-β-lactam antimicrobial drugs, antiseptics, and heavy metals whereas the currently dominant strains are more susceptible (1).
Although CC5-SCCmecII was the predominant genotype in our sample, the proportions of the second and third most frequent genotypes, CC5-SCC-mecIV and lukSF-PV-positive CC30-SCCmecIV, also had increased from prior studies (4). CC5-SCCmecIV (related to USA800), which was first isolated in children at a hospital in Portugal in 1992 (26), was overrepresented among patients <5 years of age in our sample. Some studies have suggested that this strain is more common among children (27), although the nature of this association remains unclear. The lukSF-PV-positive CC30-SCCmecIV genotype is related to the USA1100/Oceania South West Pacific clone (1), and is a distant relative of the historically epidemic and especially virulent phage type 80/81 lineage (28). We previously showed that, in contrast to the 80/81 lineage, ST30(CC30)-SCCmecIV MRSA from Brazil displays a natural attenuation of the Agr and SaeRS virulence regulators (29), which might explain why this lineage was responsible for only 9.6% of BSIs in this study.
The large number of MRSA isolates genotyped in this study enabled us to assess the distribution of MRSA genotypes by patient age and sites of infection or colonization. We identified associations between the CC5-SCCmecII genotype, BSIs, and older age, possibly because of the increased virulence or invasiveness of this genotype. The CC5-SCCmecII genotype also is found in the USA100 lineage ST5(CC5)-SCCmecII that was dominant among hospitals in the United States during the late 1990s (30), before the emergence of the USA300 clone (31). USA100 is still found in hospitals in the United States (32) and around the world (1).
In our sample, most (75/114; 65.8%) CC5-SC-CmecII isolates belonged to ST105 and shared spatype t002, suggesting the emergence of a new clone. ST105(CC5)-SCCmecII strains have previously infected humans and domestic animals (33), and 4 isolates from this lineage were reported in a hospital in São Paulo (7). Reports from other countries have occasionally shown a substantial prevalence of this lineage, including a study that showed colonization among 22.4% of patients admitted to a hospital in Pennsylvania, USA (34). Another study showed that ST105(CC5)-SCCmecII was the predominant lineage among patients who had MRSA BSI in Switzerland (35). In Portugal, ST105(CC5)-SCCmecII has been reported as the most abundant MRSA colonizing patients >60 years of age (33); a multicenter study identified this lineage as the second most common clone among patients who had BSIs (36). The first vancomycin-resistant S. aureus isolate in Portugal belonged to this lineage (33), a troubling finding because most vancomycin-resistant S. aureus isolates have belonged to the CC5 lineage (25).
Few studies exist on the molecular epidemiology of MRSA in Brazil and in other countries from South America; existing studies are based on a limited number of samples (5)(6)(7)9). As a result, the full extent of the dissemination of the ST105-SCCmecII-t002 genotype in Latin America is unknown. Since the late 2000s, ST105-SCCmecII-t002 has been reported as the second or third most frequent MRSA lineage in hospitals in the United States and some countries in Europe (33)(34)(35). For example, researchers documented an outbreak of ST105-SCCmecII-t002 MRSA among 18 neonates at Mount Sinai Hospital (New York, NY, USA) during 2014-15 (38). In addition, ST105 isolates comprised 87.5% of delafloxacin-resistant MRSA strains collected in 7 hospitals in New York (39). Altogether, these data show that ST105 is a major MRSA lineage not only in Rio de Janeiro but also in other countries. ST105-SCCmecII-t002 also might have spread in other regions of Brazil; therefore, more studies are needed to better track and investigate this lineage.
We used Bayesian molecular clock analysis to estimate the expansion of the RdJ clade in Rio de Janeiro  in 2009 (95% highest posterior density 2007-2010), which is consistent with previous estimates that date the origin of the ST105 lineage to the mid-1990s (8). The ST105 clade is characterized by a lack of virulence genes that are common among other CC5 strains. All ST105(CC5)-SCCmecII isolates lacked the sep gene encoding enterotoxin P, as noted by Challagundla et al. (8). In addition to its emetic properties, enterotoxin P is a superantigen that induces T-cell proliferation and production of proinflammatory cytokines (40). ST105(CC5)-SCCmecII-t002 strains showed resistance to fluoroquinolones, macrolides, and lincosamides. ST105(CC5)-SCCmecII isolates also lacked the serine protease encoding gene splD, despite the presence of the splABCF genes of the spl operon. Although the specific role of SplD in S. aureus pathogenesis is not known, some researchers have proposed that Spl serine proteases might use proteolysis to modulate host proteins critical to bacterial pathogenesis (41). Future work should address implications of the absence of SplD in the ST105(CC5)-SCCmecII lineage.
Compared with representatives of the ST5-SC-CmecII-t539 and ST5-SCCmecIV-t002 lineages, representative isolates of the RdJ clade showed increased evasion of phagocytosis mechanisms upon exposure to monocytic cells (i.e., THP-1). Multiple factors, including phagocytosis rate and the activity of toxic compounds released by monocytes, might affect the number of viable unphagocytosed bacterial cells (42). Moreover, we observed an increased number of viable RdJ free cells. The basis of this phenotype is unclear and deserves further study. Le Pabic et al. (43) implicated the small noncoding RNA, SprC, and its effect on regulation of the major autolysin Atl in S. aureus evasion of phagocytosis by human monocytes and macrophages. However, we did not find any differences in the sprC gene of the 6 representative strains tested, suggesting that the observed evasion might be multifactorial, probably linked to the production of several bacterial molecules (42).
One limitation of this study is the lack of more extensive clinical data such as the presence of indwelling catheters or lines and underlying conditions that might have affected our estimates. The association between ST105-SCCmecII-t002 and BSIs was attenuated when accounting for other variables such as hospital type and year of isolation, but still might be of clinical relevance. Access to more extensive clinical data would enable further exploration of this relationship. In addition, our reliance on samples from Rio de Janeiro might have affected our phylogenetic analysis; focused sampling in other geographic locations might show a more widespread epidemic.
In summary, we uncovered a new MRSA clone in hospitals in the Rio de Janeiro metropolitan area. Our findings emphasize the dynamic nature of the local rise and decline of various MRSA clones. In addition, these data indicate that MRSA clonal dynamics also might be associated with different manifestations of disease and host factors, such as age. This analysis revealed the emergence of a novel multidrug-resistant MRSA clone associated with BSIs. This association might be critical for assessing the clinical and epidemiologic risks associated with the spread of this clone and the biologic basis for its putative enhanced invasiveness.

About the Author
Ms. Viana is a doctoral candidate at Federal University of Rio de Janeiro, Rio de Janeiro, Brazil. Her main research interests are genomics and phylogenetic evolution of methicillin-resistant Staphylococcus aureus.