Molecular Characterization by Whole-Genome Sequencing of Clinical and Environmental Serratia marcescens Strains Isolated during an Outbreak in a Neonatal Intensive Care Unit (NICU)

The whole-genome sequencing (WGS) of eighteen S. marcescens clinical strains isolated from 18 newborns hospitalized in the Neonatal Intensive Care Unit (NICU) at Pescara Public Hospital, Italy, was compared with that of S. marcescens isolated from cradles surfaces in the same ward. The identical antibiotic resistance genes (ARGs) and virulence factors were found in both clinical and environmental S. marcescens strains. The aac(6′)-Ic, tetA(41), blaSRT-3, adeFGH, rsmA, and PBP3 (D350N) genes were identified in all strains. The SRT-3 enzyme, which exhibited 10 amino acid substitutions with respect to SST-1, the constitutive AmpC β-lactamase in S. marcescens, was partially purified and tested against some β-lactams. It showed a good activity against cefazolin. Both clinical and environmental S. marcescens strains exhibited susceptibility to all antibiotics tested, with the exception of amoxicillin/clavulanate.


Introduction
Serratia marcescens is a Gram-negative bacterium that has been considered, for a long time, to be an environmental organism because of its ability to survive in surfaces, soil, and aquatic reservoirs [1]. Chromosomal-encoded genetic factors such as efflux pump systems, porins, virulence factors, and the capacity to acquire antibiotic resistance genes (ARGs) by horizontal transfer facilitate the survival of S. marcescens in different environments, causing nosocomial infections [2][3][4][5]. Additionally, S. marcescens has been found as bacterial contaminant of the blood donors, after venipuncture, or of blood platelet concentrates [6]. Over the last decades it has emerged as clinical pathogen in different hospital settings, in particular in intensive care units (ICUs) and neonatal intensive care units (NICUs) [7][8][9][10][11][12]. S. marcescens is responsible for a wide range of asymptomatic and severe clinical manifestations such as respiratory, urinary, and bloodstream infections, meningitis, sepsis, keratitis, conjunctivitis, and surgical wound infections [13,14]. S. marcescens outbreaks in NICUs are responsible for newborn morbidity and mortality [15]. The last European Centre for Disease Prevention and Control (ECDC) report indicates that in 2017, Serratia spp. was in 6th and 9th place in European ICU-acquired pneumonia episodes, ICU-acquired bloodstream infection, and ICU-acquired urinary tract infection episodes, respectively [16]. Most clinical S. marcescens showed resistance to different classes of antibiotics and, especially, to many extended-spectrum β-lactams [17,18]. In S. marcescens multidrug resistance could be intrinsic or plasmid acquired, and the most common mechanism of resistance was that mediated by serine-β-lactamases, such as AmpC, TEM-type, KPC-type, GES-type, and metallo-β-lactamases [18][19][20][21][22][23][24]. Carbapenemases-producing S. marcescens strains are more frequently found [25], and they represent a serious risk because this organism is intrinsically resistant to colistin, which is considered the last drug option for carbapenem resistant infections [26]. Simultaneous outbreaks of S. marcescens and ESBLs-producing Enterobacterales (i.e., K. pneumoniae) have also been described as a situation of greater risk [27,28].
In the present study we have correlate the genetic background of clinical and environmental S. marcescens strains during an outbreak occurred in the period February-July 2021 at the NICU of the Pescara Public Hospital (central Italy). The ARGs and virulence factors were examined by whole-genome sequencing. The isolation, partial purification, and kinetic characterization of the SRT-3, AmpC β-lactamases, was also performed.

Clinical Strains
From February to July 2021, eighteen S. marcescens clinical strains were isolated from ear, pharyngeal, and rectal swabs of eighteen newborns recovered in NICU of Pescara Public Hospital, central Italy. The NICU is equipped with eight bed stations for the management of the critically ill newborns. The swabs were inoculated in Brain Heart Infusion broth medium (BHI) (Liofilchem, Roseto degli Abruzzi, Italy) and incubated overnight at 37 • C. 100 µL of the overnight cultures were plated on MacConkey agar (Liofilchem s.r.l, Roseto degli Abruzzi, Italy) and incubated at 37 • C for 18 h. One single and pure colony, for each sample, was used to identify S. marcescens by Matrix-Assisted Laser Desorption Ionization Time-Of-Flight Mass Spectrometry (MALDI-TOF, MS) (Bruker Daltonics, Billerica, MA, USA).

Environmental Sampling
Environmental samples were collected from numerous surfaces within NICU: doors and door handles, walls, floors, cradles, shelves, benches, hoods, sinks, ventilators, milk collecting devices, medical records, trolleys, stethoscopes, and other personal medical devices. Samples from all environmental surfaces were obtained with a cotton swab wet with a sterile saline solution. The swabs were then inoculated in BHI and then incubated overnight at 37 • C. The procedure was repeated every week for five months. The bacteria identification was performed following the same procedure used for clinical strains, as described in the previous paragraph.

Testing for Antimicrobial Susceptibility
The antimicrobial activity of S. marcescens isolates was tested against the routinely clinical antibiotics such as amikacin, amoxicillin/clavulanate, piperacillin/tazobactam, cefepime, cefotaxime, ceftazidime, ciprofloxacin, ertapenem, meropenem, fosfomycin, gentamicin, and colistin using the Phoenix Automated Microbiology System (BD Diagnostic Systems, Sparks, MD, USA). For antimicrobial susceptibility, each bacterial culture was adjusted to a 0.5-0.6 McFarland standard in order to obtain a final inoculum of 5 × 10 5 CFU/mL. Results for antimicrobial susceptibility were interpreted in accordance with EUCAST guidelines [29].

SRT-3 Extraction and β-Lactamase Activity
An overnight culture of S. marcescens SM_PE/1 in 1 L of BHI broth was harvested by centrifugation at 2500× g (4 • C) for 15 min, then washed twice with 50 mM sodium phosphate buffer pH 6.5. The crude extract was obtained by sonication using a probe of 3 mm in diameter (5 times for 1 min with 2 min off, on ice, at 60 Watt). The cell debris was removed by high-speed centrifugation at 105,000× g for 40 min. SRT-3 enzyme was purified in FPLC system using cation exchange Sepharose S FF equilibrated in 50 mM sodium phosphate buffer pH 6.5. The fractions containing SRT-3 were eluted in a linear gradient of NaCl 1 M in the same buffer. The β-lactamase activity was performed following the hydrolysis of some β-lactams (cefazolin, cefotaxime, ceftazidime, cefepime, meropenem) at 25 • C in 25 mM sodium phosphate buffer (pH 7.0). Data were collected with a Perkin-Elmer Lambda 25 spectrophotometer (Perkin-Elmer Italia, Monza, Italy). K m and V max were determined under initial-rate conditions using the Hanes-linearization method and the Origin Pro 8.5.1 program to generate Michaelis-Menten curves [35,36].

Epidemiological Outbreak
S. marcescens clinical strains were isolated from 18 newborns (eight boys and ten girls): 15 born preterm (gestational age from 25 to 36 weeks) and 3 born at term. All newborns were hospitalized in the NICU for an average time of 41.4 days. Namely, at birth, all newborns underwent ear, pharyngeal, and rectal swabs, which were found to be positive for S. marcescens. Two newborns (gestational age: 25 weeks) were symptomatic and were treated with antibiotics until a negative swab was obtained, with good clinical conditions at the end of the treatment. The antibiotic therapy was meropenem 20 mg/Kg every 12 h for infants younger than 8 days and every 8 h hours for infants older than 8 days. The remaining 16 newborns were asymptomatic, and the swabs were repeated approximately once a week until a negative swab was obtained.
From fifty environmental samples analyzed, one swab, collected from cradle surfaces, was positive for S. marcescens. All other medical devices and surfaces analyzed were found to be negative. Immediately after environmental samples microbiological analysis, a sanitation procedure was started, reinforcing all recommended measures such as hand washing with alcohol-based solutions, glove use for assisting and caring newborns, and routine renewal of water and air filters [37]. The last environmental sampling carried out at the beginning of July 2021 yielded negative results.

Antimicrobial Susceptibility
All S. marcescens strains, 18 clinical and 1 environmental isolate, were analyzed against a large panel of antibiotics routinely used in the Clinical Microbiology and Virology Unit of Pescara Public Hospital. The S. marcescens analyzed showed the same susceptibility profile (Table 1) exhibiting susceptibility to all antibiotics tested, with the exception of amoxicillin/clavulanate association (MIC > 16 mg/L) and colistin (MIC > 8 mg/L).

WGS Analysis
Thus, S. marcescens clinical strains of eighteen newborns and one S. marcescens isolated from cradle surface were included in the study in order to characterize their WGS. On the basis of bioinformatic analysis, the nineteenth S. marcescens showed the genome size ranging from 4.864.007-4.876.696 bp ( Table 2). A minor part of the genome was, undoubtedly, loss during library preparation. In all S. marcescens, including clinical and environmental strains, no plasmids were identified. However, both clinical and environmental strains showed the same ARGs and virulence factors profile. The aac(6 )-Ic, tetA (41), and bla SRT-3 genes were identified in all strains. Regarding virulence factors, the clinical and environmental S. marcescens strains possessed adeFGH, rsmA, CRP elements, and penicillin binding protein-3 gene (PBP3) with D350N amino acid substitution.

Discussion
S. marcescens infections represent a serious problem in hospital settings, in particular in NICUs and ICUs, which are considered the high-risk wards. Our study described an outbreak occurred at NICUs of Pescara Public Hospital (central Italy) involving 18 newborns, two of which were symptomatic. The early etiological diagnosis allowed neonatologists to treat the two symptomatic newborns with a specific antibiotic therapy. Luckily, the outbreak was limited to a small number of cases and none of newborns died. The average time during which the 18 infants were admitted to the NICU was 41.4 days and this probably contributed to an increase of the risk factor for the acquisition of S. marcescens infection. It is well known that bacterial colonization in preterm newborns is widespread because their intestinal microbiota has not yet been established [38,39] and their immune system is immature [40,41]. Other risk factors for S. marcescens outbreaks are represented by length of stay of newborns in NICU and the broad spectrum administered empirically [42,43].
The extensive collection of environmental samples showed the presence of S. marcescens strain only in one cradle, while the other investigated samples yielded negative results. In early 2011, in the same hospital, an outbreak of S. marcescens which involved six neonates and the environmental reservoir was represented by two soap dispensers [44]. This is a confirmation that S. marcescens outbreak, in the same ward, is a recurring problem. Similar studies have been conducted in other Italian regions [45][46][47][48]. Casolari et al. described two consecutive S. marcescens outbreaks in a NICU of a North Italy hospital in a period of 10 years with an overall mortality of 7% [47].
The 19 S. marcescens analyzed in the present study showed the typical resistance to colistin and amoxicillin-clavulanate association, but they are susceptible to meropenem which was chosen as preferred antibiotic for therapy. The molecular characterization of both clinical and environmental S. marcescens was performed using a next-generation sequencing platform in order to obtain the WGS of each strain. All strains sequenced exhibited the same ARGs and virulence factors profile. The virulence factors ade-FGH, rsmA, In order to verify the kinetic behavior of the SRT-3, the enzyme was extracted by S. marcescens SM_PE/1 and purified by one chromatographic step. The molecular mass and isoelectric point of 41 kDa and 9.2, respectively, were determined for SRT-3. The hydrolysis of SRT-3 was monitored spectrophotometrically against some cephalosporins and one carbapenem. In detail, we have tested cefazolin (2nd generation cephalosporin), cefotaxime and ceftazidime (3rd generation cephalosporins), cefepime (4th generation cephalosporin), and meropenem (carbapenem). Cefazolin was well hydrolyzed by SRT-3 with V max = 10.5 µmol/s/mg and K m = 170 µM. On the contrary, cefotaxime, ceftazidime cefepime, and meropenem were not hydrolyzed by SRT-3 enzyme.

Discussion
S. marcescens infections represent a serious problem in hospital settings, in particular in NICUs and ICUs, which are considered the high-risk wards. Our study described an outbreak occurred at NICUs of Pescara Public Hospital (central Italy) involving 18 newborns, two of which were symptomatic. The early etiological diagnosis allowed neonatologists to treat the two symptomatic newborns with a specific antibiotic therapy. Luckily, the outbreak was limited to a small number of cases and none of newborns died. The average time during which the 18 infants were admitted to the NICU was 41.4 days and this probably contributed to an increase of the risk factor for the acquisition of S. marcescens infection. It is well known that bacterial colonization in preterm newborns is widespread because their intestinal microbiota has not yet been established [38,39] and their immune system is immature [40,41]. Other risk factors for S. marcescens outbreaks are represented by length of stay of newborns in NICU and the broad spectrum administered empirically [42,43].
The extensive collection of environmental samples showed the presence of S. marcescens strain only in one cradle, while the other investigated samples yielded negative results. In early 2011, in the same hospital, an outbreak of S. marcescens which involved six neonates and the environmental reservoir was represented by two soap dispensers [44]. This is a confirmation that S. marcescens outbreak, in the same ward, is a recurring problem. Similar studies have been conducted in other Italian regions [45][46][47][48]. Casolari et al. described two consecutive S. marcescens outbreaks in a NICU of a North Italy hospital in a period of 10 years with an overall mortality of 7% [47].
The 19 S. marcescens analyzed in the present study showed the typical resistance to colistin and amoxicillin-clavulanate association, but they are susceptible to meropenem which was chosen as preferred antibiotic for therapy. The molecular characterization of both clinical and environmental S. marcescens was performed using a next-generation sequencing platform in order to obtain the WGS of each strain. All strains sequenced exhibited the same ARGs and virulence factors profile. The virulence factors ade-FGH, rsmA, CRP, and PBP3, found in our strains, are likely responsible for resistance to various antimicrobials. In particular, the adeFGH is one of the three parts which compose the RND efflux pumps (i.e., AdeABC, AdeFGH, and AdeIJK) that contribute to antibiotic resistance in several pathogens [49,50]. The rsmA and CRP belong to the RND antibiotic efflux pump and it contributed to antibiotic resistance in particular to fluoroquinolone and macrolide resistance [51]. The PBP3 is an essential protein involved in many interactions within the di- visome of the bacteria [52] and, as well as other penicillin-binding proteins, represents major targets for β-lactam antibiotics [53]. The substitution D350N in PBP3 has also been found in other bacterial species (i.e., Haemophilus influenzae) [54]. Concerning the other resistance genes found in this study, the aac(6 )-Ic was identified in 1992 in S. marcescens strains and it represents an intrinsic acetyltransferase, which may play a role in primary metabolism of bacterium. The aac(6 )-Ic is a chromosomal gene which confers resistance to aminoglycosides (i.e., gentamycin, tobramycin, netilmicin, and amikacin) [55,56]. The tetA(41) is a tetracycline efflux pumps protein found in environmental and clinical S. marcescens [57,58]. It is most closely related to tet(39) protein of Acinetobacter [59]. Some ARGs and virulence genes identified in our strains can be commonly found in S. marcescens [6].
The SRT-3 enzyme belongs to the SRT AmpC family, which is characteristic of S. marcescens clinical isolates [60]. β-lactamases are the commonest cause and the most efficient mechanism of bacterial resistance to β-lactam antimicrobial agents. As confirmed in the last standard numbering scheme [61], three conserved elements have been identified in class C β-lactamases: the first element is the active site with the catalytic serine at position 64 and downstream one helix-turn is located a lysine residue (K67) whose side-chain also points into the active site; the second element 150YSN152 is located on a short loop in the all α domains where it forms one side of the catalytic cavity. The side-chains of the first and third residues point to the active-site cleft; the third element, 315KTG317, is on the innermost strand of the β-sheet (α/β domain) and forms the opposite wall of the catalytic cavity [62]. In the present study, we have characterized the SRT-3 enzyme which showed a narrow spectrum of substrate. Indeed, it was able to hydrolyze only cefazolin, a 2nd generation of cephalosporins. Based on our knowledge, in literature, there are no kinetic data related to SRT-3 enzyme. Similarly, to SST-1, which showed more than 97% of amino acid homology, the SRT-3 is an AmpC enzyme able to hydrolyze only old generation cephalosporins. The S. marcescens strains, isolated during the outbreak at NICU of Pescara Public Hospital, did not harbor plasmids or genes encoding for extended-spectrum β-lactamasases or carbapenemases. Thus, all ARGs found in the genome of S. marcescens strains were constitutive and not acquired by gene horizontal transfer. We suppose this is due to the fact that, during the same period, no other antibiotic resistant bacteria were identified in NICU.

Conclusions
In the present study we have described the second S. marcescens outbreak that occurred at NICU of Pescara General Hospital in the period 2011-2021. Fortunately, in the recent outbreak only two newborns were symptomatic, and the total number of cases were small. However, ongoing surveillance and infection control programs are necessary to identify the possible infection reservoir in order to adopt, as quickly as possible, the appropriate containment measures.