Characterization of KPC-Producing Serratia marcescens in an Intensive Care Unit of a Brazilian Tertiary Hospital

Serratia marcescens has emerged as an important opportunistic pathogen responsible for nosocomial and severe infections. Here, we determined phenotypic and molecular characteristics of 54 S. marcescens isolates obtained from patient samples from intensive-care-unit (ICU) and neonatal intensive-care-unit (NIUC) of a Brazilian tertiary hospital. All isolates were resistant to beta-lactam group antibiotics, and 92.6% (50/54) were not susceptible to tigecycline. Furthermore, 96.3% showed intrinsic resistance to polymyxin E (colistin), a last-resort antibiotic for the treatment of infections caused by MDR (multidrug-resistant) Gram-negative bacteria. In contrast, high susceptibility to other antibiotics such as fluoroquinolones (81.5%), and to aminoglycosides (as gentamicin 81.5%, and amikacin 85.2%) was found. Of all isolates, 24.1% were classified as MDR. The presence of resistance and virulence genes were examined by PCR and sequencing. All isolates carried KPC-carbapenemase (blaKPC) and extended spectrum beta-lactamase blaTEM genes, 14.8% carried blaOXA–1, and 16.7% carried blaCTX–M–1group genes, suggesting that bacterial resistance to β-lactam antibiotics found may be associated with these genes. The genes SdeB/HasF and SdeY/HasF that are associated with efflux pump mediated drug extrusion to fluoroquinolones and tigecycline, respectively, were found in 88.9%. The aac(6′)-Ib-cr variant gene that can simultaneously induce resistance to aminoglycoside and fluoroquinolone was present in 24.1% of the isolates. Notably, the virulence genes to (i) pore-forming toxin (ShlA); (ii) phospholipase with hemolytic and cytolytic activities (PhlA); (iii) flagellar transcriptional regulator (FlhD); and (iv) positive regulator of prodigiosin and serratamolide production (PigP) were present in 98.2%. The genetic relationship among the isolates determined by ERIC-PCR demonstrated that the vast majority of isolates were grouped in a single cluster with 86.4% genetic similarity. In addition, many isolates showed 100% genetic similarity to each other, suggesting that the S. marcescens that circulate in this ICU are closely related. Our results suggest that the antimicrobial resistance to many drugs currently used to treat ICU and NIUC patients, associated with the high frequency of resistance and virulence genes is a worrisome phenomenon. Our findings emphasize the importance of active surveillance plans for infection control and to prevent dissemination of these strains.

Serratia marcescens has emerged as an important opportunistic pathogen responsible for nosocomial and severe infections. Here, we determined phenotypic and molecular characteristics of 54 S. marcescens isolates obtained from patient samples from intensive-care-unit (ICU) and neonatal intensive-care-unit (NIUC) of a Brazilian tertiary hospital. All isolates were resistant to beta-lactam group antibiotics, and 92.6% (50/54) were not susceptible to tigecycline. Furthermore, 96.3% showed intrinsic resistance to polymyxin E (colistin), a last-resort antibiotic for the treatment of infections caused by MDR (multidrug-resistant) Gram-negative bacteria. In contrast, high susceptibility to other antibiotics such as fluoroquinolones (81.5%), and to aminoglycosides (as gentamicin 81.5%, and amikacin 85.2%) was found. Of all isolates, 24.1% were classified as MDR. The presence of resistance and virulence genes were examined by PCR and sequencing. All isolates carried KPC-carbapenemase (bla KPC ) and extended spectrum beta-lactamase bla TEM genes, 14.8% carried bla OXA−1 , and 16.7% carried bla CTX−M−1group genes, suggesting that bacterial resistance to β-lactam antibiotics found may be associated with these genes. The genes SdeB/HasF and SdeY/HasF that are associated with efflux pump mediated drug extrusion to fluoroquinolones and tigecycline, respectively, were found in 88.9%. The aac(6 )-Ib-cr variant gene that can simultaneously induce resistance to aminoglycoside and fluoroquinolone was present in 24.1% of the isolates. Notably, the virulence genes to (i) pore-forming toxin (ShlA); (ii) phospholipase with hemolytic and cytolytic activities (PhlA); (iii) flagellar transcriptional regulator (FlhD); and (iv) positive regulator of prodigiosin and serratamolide production (PigP) were present in 98.2%. The genetic relationship among the isolates determined by ERIC-PCR demonstrated that the vast majority of isolates were grouped in a single cluster with 86.4% genetic similarity. In addition, many isolates showed 100% genetic similarity to each other, suggesting that the S. marcescens that circulate in this ICU are closely related. Our results suggest that the antimicrobial resistance to many drugs INTRODUCTION Serratia marcescens is a Gram-negative bacillus that naturally resides in the soil and water and produces a red pigment at room temperature. Although previously considered non-pathogenic, this species has emerged as a prominent opportunistic pathogen found in nosocomial outbreaks in neonatal intensive care Units (NICUs), intensive care units (ICUs) and other hospital units over the last few decades (Enciso-Moreno et al., 2004;Moradigaravand et al., 2016;Ghaith et al., 2018).
The true occurrence of S. marcescens is still underestimated (Zingg et al., 2017). In NICUs, studies have showen that infected newborns are a potential source of S. marcescens (Cristina et al., 2019), although there is a constant increase of S. marcescens bacteremia across all age groups (Vetter et al., 2016;Phan et al., 2018). S. marcescens increasingly adapts to hospital environments (Yoon et al., 2005;Gastmeier, 2014). It accounts for 15% of all isolates from nosocomial infections (Raymond and Aujard, 2000). Although it is difficult to identify the source of S. marcescens during outbreaks, it is the third most frequent pathogen identified (Gastmeier et al., 2007), and more than one clone can be usually identified (David et al., 2006;Montagnani et al., 2015;Dawczynski et al., 2016).
Serratia marcescens associated with hospital outbreaks or epidemic events are commonly resistant to several antibiotics (Moradigaravand et al., 2016;Cristina et al., 2019). In fact, one important feature of S. marcescens is its resistance to narrow-spectrum penicillins and cephalosporins; nitrofurantoin; tetracycline; macrolides; cefuroxime; cephamycins; fluoroquinolone, and colistin (Stock et al., 2003;Liou et al., 2014;Moradigaravand et al., 2016;Sandner-Miranda et al., 2018). The resistance to some of these molecules may be intrinsic to this specie and is explained by either the presence of resistance genes on the chromosome or by the acquisition of such genes via horizontal transfer. It is noteworthy that the latter mechanism is considered the most important event that leads to multiple antibiotic resistance (von Wintersdorff et al., 2016;Sandner-Miranda et al., 2018).
Extended-spectrum β-lactamases (ESBLs) are a group of bacterial enzymes that can be rapidly transferred via plasmid exchange (Rawat and Nair, 2010) causing resistance to a broad range of β-lactams (Naas et al., 2008). Carbapenemases are the most versatile family of β-lactamases able to hydrolyze carbapenems and many other β-lactams  including penicillins, cephalosporins, and monobactams (Anderson et al., 2007;Marschall et al., 2009). In general, bacteria carrying the bla KPC and/or ESBLs genes usually harbor other resistance genes associated with several classes of antimicrobials (Tzouvelekis et al., 2012;Cao et al., 2014;Ribeiro et al., 2016).
Since S. marcescens has been acquiring a range of ESBLs and commonly exhibit co-resistance to many other classes of antibiotics, the infections caused by these multidrug-resistant (MDR) isolates impair therapy and limit treatment options (Yu et al., 1998;Mostatabi et al., 2013;Herra and Falkiner, 2018).
In this study, we investigated the phenotypic characteristics regarding antimicrobial resistance and the genotypic traits of S. marcescens isolated from a tertiary care hospital's ICUs including the search for resistance and virulence genes as well as the genetic relationship among the isolates. Our report describes MDR profile and KPC-producing S. marcescens isolates and highlight the importance of monitoring S. marcescens infection and the need of constant surveillance to support continuous and effective measures to prevent the spread of these strains.

Study Design and Bacterial Isolates
From February 2014 to June 2015, a total of 54 S. marcescens were isolated of clinical specimens collected from 45 patients admitted to intensive-care-unit (ICU) and neonatal intensive care unit (NICU) of a tertiary care government hospital in Palmas, Tocantins, Brazil. Since 2013, there has been an increase in detection of S. marcescens isolates from hospital inpatients, and in 2015, the hospital reported an apparent S. marcescens outbreak that occurred from July to August 2015. Appropriate intervention measures were established, such as reviewing the infection control policies, hand antisepsis practices and determination of trends of isolation of S. marcescens over time. To trace the source of the infection, bacteria were isolated from various samples obtained from clinical indications of infections during the patients' ICU stay.
As part of the control measures, surveillance cultures were obtained from tracheal aspirate and rectal swabs from all ICU patients, on admission (within the first 24 h) and during the stay (once a week). Blood, wound, catheter tip, drain, sputum, urine, rectal swab, and tracheal aspirate samples were primarily sent to the hospital's laboratory, processed and cultured by standard microbiological techniques. The blood samples were inoculated first in blood culture bottles (Hemoprov-NewProv, Brazil). All clinical samples, including blood culture bottles giving positive signals were cultured onto MacConkey agar (Probac, Brazil), blood agar (Probac, Brazil), and chocolate blood agar (Probac, Brazil). Plates were incubated at 37 • C for up to 48 h. S. marcescens were identified by Gram staining, cultural characteristics in MacConkey agar (Probac, Brazil), blood agar (Probac, Brazil), and biochemical tests (Bactray I, II,III; Laborclin, Brazil). The antimicrobial susceptibility profile was determined by Kirby-Bauer disk diffusion method. All S. marcescens isolates and the microbiological reports prepared at the hospital were sent to the Central Laboratory of Public Health of Tocantins (LACEN-TO) for further phenotypic validations.

Genomic DNA Extraction
Isolates of S. marcescens were subcultured on Brain Heart Infusion (BHI) agar (Oxoid, United Kingdom) and incubated for 24 h at 37 • C. All samples were submitted to genomic DNA extraction using the Wizard Genomic DNA Purification Kit (Promega, Madison, WI, United States), according to manufacture's instructions.

Virulence Gene Detection
The presence of four virulence genes were assessed by PCR: genes PigP, a positive regulator of prodigiosin and serratamolide production; FlhD, a flagellar transcriptional regulator; ShlA, a pore-forming toxin with hemolytic activity; PhlA, a phospholipase A with hemolytic activity. The primers sequences amplicon sizes and annealing temperatures are listed in Table 1. Amplicons were analyzed by gel electrophoresis in 1.0% agarose and visualized under ultraviolet (UV) light.

Sequence Analysis of Antibiotic-Resistance Markers and Virulence Genes
One amplicon of each studied gene was randomly selected for confirmation of identity by DNA sequencing using an automated sequencer (ABI 3500xL Genetic Analyzer; Applied Biosystems, Foster City, CA, United States). After amplification, we extracted the PCR products from agarose gels using the Illustra GFX PCR DNA (GE Healthcare), which were purified using the Gel Band Purification Kit (GE Healthcare), both according to manufacturer's instructions. Obtained sequences were edited with Bioedit v7.0.5 (Hall, 1999), compared with the nr database using the Blastn tool 1 and submitted to the GenBank database. Genes and their respective accession numbers: bla CTX -MK576103; bla KPC -MK576104; bla OXA -MK576105; bla TEM -MK576106; SdeB -MN583232; SdeY -MN583233; HasF -MN583234; aac(6 )-Ib-cr -MN583235; FlhD -MN583236; PigP -MN583237; ShlA -MN583238; PhlA -MN583239). Access to genetic heritage was approved by the National System for the Management of Genetic Heritage (SisGen n • AFF27ED).

Enterobacterial Repetitive Intergenic Consensus Polymerase Chain Reaction
Enterobacterial repetitive intergenic consensus PCR (ERIC-PCR) analysis was performed to evaluate the genetic similarity among the 54 S. marcescens isolates using the primers and conditions previously described by Versalovic et al. (1994). PCR reactions were performed using the enzyme TaKaRa Ex Taq DNA Polymerase (Takara Bio, Kusatsu, Japan). The BioNumerics program version 5.1 (AppliedMaths, Keistraat, Belgium) was used to construct the unweighted pair group mean method (UPGMA) similarity dendrogram with Dice's similarity coefficient, following .

Statistical Analyses
In the analysis of contingency tables, we used Fisher's exact test and/or Barnard's exact test. Maximum likelihood did not present superior efficiency in relation to the previous methods (data not show). It was used logistic regression model with two predictor variables x 1 and x 2 . Statistical software R was used in all data analysis.

Ethical Considerations
In our study, we did not use/collect human genetic material and biological samples. Strains were part of the collection of the Central Laboratory of Public Health, (LACEN-TO), a health-care facility that is a reference in diagnosis in the state of Tocantins, Brazil. It was a retrospective study, and epidemiological data were obtained from a database or similar, which will be kept confidential in accordance with the with the terms of Resolution 466/12 of the National Health Council. These epidemiological data were also provided by LACEN-TO.

Resistance Phenotype-Genotype Correlation and Genetic Markers for Virulence Factors
The correlation between the results of phenotypic and genotypic detection and the presence of virulence genes is shown in Figure 3.

ERIC-PCR
The ERIC-PCR results indicated that the majority of the isolates presented a rate of genetic similarity above 85% (Figure 4). Almost all strains (96.3%) were grouped into a large cluster named B cluster, sharing 86.4% of genetic similarity. In addition, the B cluster was separated into two sub-clusters named B1, with 21 isolates, and B2, with 31 isolates, sharing a genetic similarity of 96.1% and 100%, respectively. Although the cluster B1 presented two subgroups with 4 and 17 isolates, they showed 100% genetic similarity in each one. Interestingly, two strains (Sm38 and Sm40) were grouped separately within the A cluster and presented 71.4% of genetic similarity (Figure 4).

DISCUSSION
Serratia marcescens is a prominent opportunistic pathogen that frequently causes infections in intensive care, surgical and dialysis units (Krishnan et al., 1991;Martineau et al., 2018). In Brazil, there are only few studies on S. marcescens (Ribeiro et al., 2013;Silva et al., 2015). Therefore, we here describe the presence of MDR S. marcescens isolates producing KPCcarbapenemase (bla KPC ) and extended spectrum beta-lactamase (bla TEM , bla CTX−M−1group e bla OXA−1 , 4and 30 ) in the state of Tocantins, Brazil. Tocantins, located southeast of the Northern Region, is the newest state of Brazil and shares borders with six states presenting intensive migration flow.
Serratia marcescens were isolated mainly from male patients with 60 or more years of age, similarly to previous studies that demonstrated advanced age male patients as presenting a higher risk of contracting S. marcescens infections (Ulu-Kilic et al., 2013;Kim et al., 2015;O'Horo et al., 2017). Samples with higher amounts of S. marcescens were those from tracheal aspirate, followed by blood, rectal swab, and wounds. Our findings corroborate studies by Kim et al. (2015) and Liou et al. (2014) that reported the respiratory tract as the main route of infection for S. marcescens. Other studies have also reported S. marcescens in other sites as bloodstream (Seeyave et al., 2006) and wounds (Us et al., 2017), demonstrating the versatility of these strains in colonizing the host and affecting a wide variety of physiological system.
In addition to the intrinsic resistance to the antibiotics AMP, SAM, CXM, FOX, and CST, we found multidrug-resistant (MDR) S. marcescens isolates to beta-lactam, glycylcycline, and/or aminoglycoside and quinolone group antibiotics. This is in line with other studies that have also reported MDR S. marcescens mainly to beta-lactam, aminoglycoside, and quinolone antibiotics groups (Stock et al., 2003), in hospital environment (Merkier et al., 2013), and particularly in critically ill patients and neonatal intensive care units (Maragakis et al., 2008). We also observed a significant resistance to colistin and tigecycline in several colonization sites, as shown by a previous study (Silva et al., 2017).
All the S. marcescens strains tested here were resistant to β-lactams, including carbapenems antibiotics. Resistance to carbapenems used to be uncommon among Serratia species (Stock et al., 2003), but many resistant strains have now emerged throughout the world (Lin et al., 2016). Although few studies have related resistance to carbapenems in S. marcescens in Brazil (Milisavljevic et al., 2004;da Costa Guimarães et al., 2013;Ribeiro et al., 2013), Silva et al. (2015) obtained similar results. They have isolated S. marcescens resistant to imipenem, meropenem and ertapenem in samples from different infection sites of ICU patients in another Brazilian locality. Both results are troubling. Infections caused by carbapenem-resistant bacteria often do not respond to conventional treatment (Okoche et al., 2015), as produced carbapenemases hydrolyze not only carbapenems but also penicillins, cephalosporins and monobactams (Queenan and Bush, 2007). The most common carbapenem resistance of S. marcescens in Brazil is due to the production of carbapenamases, especially the KPC-2 type (da Costa Guimarães et al., 2013;Ribeiro et al., 2013), that is encoded by the gene bla KPC−2 .
Enterobacteriaceae, such as S. marcescens, have the genes bla TEM−1 and bla SHV−1 ; these genes express classical class A betalactamases, encoded by plasmid that hydrolyze first generation penicillins and cephaloporins (Bush, 2010). We found gene bla TEM in all isolates while gene bla SHVvariants was not detected. It is noteworthy that even though S. marcescens also carries the gene bla CTX−M (Yu et al., 2003;Kim et al., 2005;Tenover et al., 2013), few of our strains had the gene. Some strains also carried the genes bla OXA−1 , 4 and 30 , that have been reported in few studies in Brazil or in other countries, either alone or associated with extended spectrum beta-lactamases genes (ESBL) (bla TEM , bla SHV e bla CTX−M ) in S. marcescens strains. Although there are discrepancies in frequency rate and in genotyping of ESBL-producing S. marcescens (Cheng et al., 2006), the observed beta-lactam antibiotic resistance may have also been caused by the genes bla TEM , bla CTX−M , and bla OXA , since the production of broad-spectrum beta-lactamases enzymes (TEM-1, TEM-2, SHV-1, OXA-1) generate resistant to ampicillin, ticarcillin, piperacillin, piperacillin/tazobactam and cephalosporin antibiotics, and the enzymes CTX-M have hydrolytic activity against cefotaxime (Levy and Marshall, 2004;Sugumar et al., 2014). In our study, most strains of S. marcescens were only sensitive to aminoglycosides (gentamicin and amikacin) and fluoroquinolone (ciprofloxacin). Aminoglycosides are the oldest antibiotics that have been used less frequently in the last years, thus possibly preserving activity against some resistant bacteria that cause difficult to cure infections (Falagas et al., 2008;Gad et al., 2011). The observed low resistance to ciprofloxacin (18.18%) is in agreement with the results obtained by Sheng et al. (2002) who observed 20-30% resistance to quinolone in S. marcescens isolates. However, it is important to consider that S. marcescens is highly adaptable, so rates of resistance to fluoroquinolones diverge considerably among institutions (Young et al., 1980;Mahlen, 2011;Sader et al., 2014), including within Brazilian ones.
Resistance to fluoroquinolones may be caused by alterations in the target enzymes DNA gyrase and topoisomerase IV, and by acquisition of the transferable plasmid-mediated quinolone resistance (PMQR) determinants qnr, qepA, aac(6')-Ib-cr, and oqxAB (Veldman et al., 2011;Poirel et al., 2012;Moradigaravand et al., 2016). The gene aac(6 )-Ib-cr, a variant gene of the aminoglycoside acetyltransferase, was also present in most of strains that presented fluoroquinolone and/or aminoglycoside resistance. This finding is consistent with others studies that have shown that aac(6 )-Ib-cr may be associated with antibacterial resistance against fluoroquinolone and aminoglycoside (Kim et al., 2009(Kim et al., , 2011 antibiotics. Three RND-type efflux have been reported in S. marcescens, namely SdeAB (Kumar and Worobec, 2005a), SdeCDE (Kumar and Worobec, 2005a;Begic and Worobec, 2008), and SdeXY (Chen et al., 2003). SdeAB and SdeXY interact with HasF (an outer membrane component, TolC homolog gene) contributing to resistance against a wide variety of antimicrobial agents FIGURE 4 | Dendrogram representing the genetic relationship between 54 strains of Serratia marcescens associated with patient localization, isolation site, antibiotic resistance, and virulence genes. ICU: Intensive Care Unit. NICU: Neonatal Intensive Care Unit. *Patients presenting S. marcescens in more than one infection site. # multidrug resistance (MDR) pattern. (Begic and Worobec, 2008;Hornsey et al., 2010). Although we did not analyze the genes SdeA e SdeX, the genes SdeB and SdeX were present in all strains and the gene HasF was also found in most strains. Drug extrusion by efflux pumps as SdeAB-HasF comprises one of the main mechanisms for fluoroquinolones antibiotic resistance (Dalvi and Worobec, 2012). Additionally, Hornsey et al. (2010) has shown the intrinsic activity of the SdeXY-HasF efflux pump is responsible for the lower susceptibility to ciprofloxacin. Thus, in addition to the presence of the genes aac(6 )-Ib-cr [associated with plasmid-mediated quinolone resistance (PMQR)], the genes SdeB, SdeY, and HasF that encode RND-type efflux pump may have contributed to the observed ciprofloxacin resistance.
Our data shows high prevalence of tigecycline-resistant S. marcescens strains and SdeY and HasF genes. The reduced sensitivity of S. marcescens to tigecycline may be related to the up-regulation of the SdeXY-HasF efflux pump (Hornsey et al., 2010). Although the gene SdeX was not analyzed, our findings strongly suggest that these genes may be responsible for the high tigecycline resistance.
Many bacteria produce virulence factors as hydrolytic enzymes and toxins that enable host invasion, bacterial proliferation and inhibit host defense mechanisms, sometimes resulting in host death (Aggarwal et al., 2017). S. marcescens strains were also evaluated for the presence of toxin genes ShlA and phlA, and all but the Sm40 strain carried these genes. Our findings are in agreement with other studies that reported the presence these genes in S. marcescens (strain SEN) (Aggarwal et al., 2017). FlhDC has been proposed as a regulator controlling flagellum biogenesis, biofilm formation, cell septation and expression of virulence factors during swarming (Givskov et al., 1995;Fraser and Hughes, 1999;Chilcott and Hughes, 2000;Lin et al., 2010). In our study, the presence of FlhC was not analyzed but the gene FlhD was present in almost all isolates. S. marcescens produces biosurfactant serrawettin and red pigment prodigiosin used in surfaces colonization (Hejazi and Falkiner, 1997). PigP is a positive regulator of prodigiosin production that regulates swarming and hemolysis through serratamolide production (Fineran et al., 2005;Shanks et al., 2013). In our study, PigP gene was found in almost all isolates (98.15%). Overall, our results suggest that the combination of these virulence genes could have contributed to the pathogenicity of S. marcescens strains.
The dendrogram based on ERIC-PCR fingerprint analysis demonstrated that the vast majority of isolates are closely related, sharing a genetic similarity of 86.4% (except for two strains). In addition, many isolates showed 100% similarity to each other. A study conducted by Ferreira-Firmo et al. (2019) evaluated nine S. marcescens from different clinical sources and three hospitals in Northeast Brazil showed a greater genetic diversity among the studied strains. Lin et al. (2016) studied 83 carbapenem-resistant S. marcescens isolates recovered from Zhejiang Provincial 501 People's Hospital, China, from which they found 63 blaKPC-2 positive strains sharing nine different profiles. Our results demonstrate the predominance of few genetic profiles grouped together, with similarity above 85%, indicating that, although bacteria have been isolated from different patients and devices, the circulating S. marcescens in this hospital is highly genetically related.
Our study has some limitations worth noting. There was a high number of S. marcescens isolated from rectal swabs and tracheal aspirate, and both cultures are recommended for surveillance in ICU and NICU patients. However, microbiological reports sent to LACEN were not clear regarding how many of these samples were analyzed for both clinical and surveillance purposes. This study was further limited by the duration of the research, which was relatively short (2014)(2015), and by conventional phenotypic and genotypic techniques, that have their own particular strengths and limitations in detecting MDR strains. Nonetheless, we intend to extend the analysis period of S. marcescens isolates, and whole genome sequencing to type relevant MDR strains must be performed. We expect that this study broadens our understanding on epidemiology, antibiotic resistance, and putative virulence factors of these strains, and provides relevant information for the prevention and management of S. marcescens infections.
In conclusion, S. marcescens represents a problem for public health (Kurz et al., 2003) and the resistance pattern exhibited by clinical isolates along with the transmission to other clones show the importance of researching factors associated with the increase in frequency and/or emergence of infections caused by S. marcescens in Brazilian hospitals. The occurrence of antibiotic-resistant bacteria considerably varies according to country, region and susceptible population, and the mitigation of this problem in ICUs is especially associated with actions to control the spreading of such drugresistant bacteria (Ko et al., 2002;Hou et al., 2015). Thus, it is crucial to eliminate sources of resistance development and associated reservoirs as well as to overtake standardized sanitation and enforce mandatory notification to gather important data for continuous risk assessment evaluation and effective decision making to control these species in hospital environments.

ETHICS STATEMENT
The studies involving human participants were reviewed and approved by the Committee of Ethics in Human Research of the Federal University of São Carlos (no. 1.088.936). In this work, all Serratia marcescens and the anonymous archival data related patient age, gender, and sample type were obtained from the Central Laboratory of Public Health of Tocantins (LACEN/TO, data's owner). Permission to conduct the present study was obtained from the Health Department of the State of Tocantins (Secretaria da Saúde do Estado do Tocantins -SESAU) and LACEN/TO. Patient consent was not required, since the data presented in this study do not relate to any specific person or persons. Written informed consent from the participants' legal guardian/next of kin was not required to participate in this study in accordance with the national legislation and the institutional requirements.

AUTHOR CONTRIBUTIONS
RF, GR, MD, MO-S, MB, and M-CP performed the experiments. FG and EL aided with statistical analysis. EC aided with the sequencing analysis and the sequence submission to the NCBI platform. M-CP conceived and supervised the study. M-CP, AC, IM, and AP-S wrote the manuscript and analyzed the data. MO-S performed ERIC-PCR.

FUNDING
This work was supported by the Fundação de Amparo a Pesquisa do Estado de São Paulo (FAPESP grants 2020/05699-6 to M-CP and FAPESP grant 2013/22581-5 to AP-S) and Conselho Nacional de Desenvolvimento Científico e Tecnológico (CNPq, grant 2013/485873 to M-CP). This study was partially financed by the Fundação de Amparo á Pesquisa do Estado de São Paulo-Brazil (FAPESP) as a fellowship to MD (FAPESP fellowship 2018/24213-7), and by the Coordenação de Aperfeiçoamento de Pessoal de Nível Superior -Brazil (CAPES) -Finance Code 001 as a fellowship to GR, FG, and EC.

ACKNOWLEDGMENTS
The authors thank the Laboratório Central de Saúde Pública do Tocantins, Palmas, TO, Brazil (LACEN-TO) who kindly provided the S. marcescens strains, and to Secretaria de Saúde do Estado do Tocantins (SESAU-TO) for facilitating the development of project.