Occult Vancomycin-Resistant Enterococcus faecium ST117 Displaying a Highly Mutated vanB2 Operon

Rare information is available on clinical Enterococcus faecium encountered in Sardinia, Italy. This study investigated the antimicrobial susceptibility profiles and genotypic characteristics of E. faecium isolated at the University Hospital of Sassari, Italy, using the Vitek2 system and PCR, MLST, or WGS. Vitek2 revealed two VanB-type vancomycin-resistant Enterococcus faecium (VREfm) isolates (MICs mg/L = 8 and ≥32) but failed to detect vancomycin resistance in one isolate (MIC mg/L ≤ 1) despite positive genotypic confirmation of vanB gene, which proved to be vancomycin resistant by additional phenotypic methods (MICs mg/L = 8). This vanB isolate was able to increase its vancomycin MIC after exposure to vancomycin, unlike the “classic” occult vanB-carrying E. faecium, becoming detectable by Vitek 2 (MICs mg/L ≥ 32). All three E. faecium had highly mutated vanB2 operons, as part of a chromosomally integrated Tn1549 transposon, with common missense mutations in VanH and VanB2 resistance proteins and specific missense mutations in the VanW accessory protein. There were additional missense mutations in VanS, VanH, and VanB proteins in the vanB2-carrying VREfm isolates compared to Vitek2. The molecular typing revealed a polyclonal hospital-associated E. faecium population from Clade A1, and that vanB2-VREfm, and nearly half of vancomycin-susceptible E. faecium (VSEfm) analyzed, belonged to ST117. Based on core genome-MLST, ST117 strains had different clonal types (CT), excluding nosocomial transmission of specific CT. Detecting vanB2-carrying VREfm isolates by Vitek2 may be problematic, and alternative methods are needed to prevent therapeutic failure and spread.


Introduction
Nosocomial Enterococcus faecium (E. faecium) infections are caused by a specific hospitalassociated (HA) E. faecium subpopulation, capable of prolonged survival outside the human body, contributing to cross-contamination through direct contact between patients and healthcare workers. This leads to local outbreaks, as well as increased length of hospitalization, mortality rate, and healthcare costs [1,2]. HA-E. faecium infections are characterized by ampicillin resistance (AREfm) and often by multidrug resistance (MDR) [1].
In their pathway towards MDR, HA-AREfm can acquire vancomycin resistance determinants under antibiotic pressure, becoming the worrisome vancomycin-resistant E. faecium (VREfm), which is responsible for more severe infections in high-risk patients, especially from intensive care unit (ICU) and surgery wards, with limited therapeutic options [3].
The most common vancomycin resistance determinant in clinical settings is vanA, but in the last few years, vanB clones have been emerging, overtaking vanA in some European countries [4,5]. Both vanA and vanB determinants are associated with mobile genetic elements (MGEs)-the vanA gene cluster as a part of Tn3 family transposon Tn1546, and vanB as part of Tn1547 or Tn1549/Tn5382-like conjugative transposons, which play a crucial role in disseminating vancomycin resistance intra-and interspecies [6,7].
The VanA-and VanB-type glycopeptide resistance expressions are regulated by a two-component regulatory system, VanR A S A and VanR B S B , respectively, composed of a membrane-associated sensor kinase (VanS A and VanS B ) and a cytoplasmic response regulator (VanR A and VanR B ) [8]. The cytoplasmic response regulator VanR acts as a transcriptional activator controlling the transcription of vanR A S A regulatory and vanHAXYZ resistance genes by PR and PH promoters' activation in the vanA operon, and vanR B S B and vanY B WH B BX B genes by activating the PR B and PY B promoters in the vanB operon [8]. Unlike vanA, which confers inducible resistance to high levels of both glycopeptides vancomycin and teicoplanin, the vanB operon is not induced by teicoplanin, and it encodes for a variable level of inducible vancomycin resistance, in some cases below the European Committee on Antimicrobial Susceptibility Testing (EUCAST) breakpoint of 4 mg/L. Thus, vanB-containing isolates can go undetected by phenotypic methods, as recently described for the Vitek2 instrument [9].
Genomic surveillance through whole-genome sequencing (WGS) has proved its usefulness for accurate and effective infection control practices, for detecting variants of VREfm that are missed by conventional screening techniques, and for molecular epidemiology studies also tracing intra-hospitals dissemination of nosocomial clones [16].
In this study, we aimed to characterize, through phenotypic and molecular methods, E. faecium isolated from 2013-2018 in the University Hospital of Sassari, Italy. WGS was used to characterize the vanB determinants of the first described vanB-VREfm isolates in our hospital.

Phenotypic and Genotypic (PCR) Characterization of E. faecium Isolates
Overall, 39 E. faecium isolates collected from patients from ICU (23%), Surgery (54%), and Medicine (23%) wards of the University Hospital of Sassari were analyzed. A description of the characteristics of the isolates is given in Table 1. The antibiotic resistance pattern was similar among all isolates regardless of isolation date, source, and ward. The majority (92%) were ampicillin resistant (AMP MIC ≥ 32 mg/L), as well as resistant to high-level streptomycin (92%), high-level gentamicin (80%), imipenem (92%), and ciprofloxacin and erythromycin (100%). All isolates were susceptible to both linezolid and tigecycline. Two isolates (SSM5811 and SSM5812) showed resistance to vancomycin by the Vitek2 system exhibiting VanB phenotypes with VAN MIC = 8 mg/L and ≥32 mg/L, respectively, and TEC MIC ≤ 0.5 mg/L ( Table 1). The two isolates were collected from the same patient (from a wound swab and a central vascular catheter 24hr later) who was admitted at the ICU in May 2018.  The presence of the vanB gene in the two VREfm isolates SSM5811 and SSM5812 was confirmed by PCR. A third isolate (SSM5777), isolated in 2017 from ICU, also carried the vanB gene, even displaying susceptibility to glycopeptides (VAN MIC ≤ 1 mg/L and TEC MIC ≤ 0.5 mg/L) by Vitek2 using the gram-positive susceptibility test cards P592 and P658.
The three vanB E. faecium isolates were then tested by MicroScan using gram-positive card AB33 (Table 2) and by disk diffusion; both methods were also able to detect the vancomycin resistance in the occult SSM5777 isolate ( Table 2). As a result of disk diffusion for vancomycin susceptibility in the SSM5777 isolate, a fuzzy inhibition zone edge was observed in the SSM5777 isolate. Despite the zone diameter of 12 mm, colonies were found within the inhibition zone, so the isolate was reported as resistant according to the notes of the EUCAST clinical breakpoint.
A further investigation of the three vanB-VREfm isolates was conducted by placing the isolates on BHI broth with increasing concentrations of vancomycin (4, 10, 25, 50, 100 mg/L). All three isolates were able to grow at all tested concentrations at 37 • C after 24 h, including at 100 mg/L after 48 h of incubation. Following exposure to 100 mg/L vancomycin, glycopeptide susceptibility was retested, growing the isolates in the absence of vancomycin. This enabled the Vitek2 system to detect the vancomycin resistance in SSM5777 as well. Glycopeptides MICs were determined by both the Vitek2 system and MicroScan before exposure to vancomycin, and by Vitek2 following exposure to vancomycin, as shown in Table 2.

WGS In Silico Analysis
Ten genomes of E. faecium isolates were selected for WGS according to hospital associated STs. Five ST117-including the vanB-VREfm, two ST916, and one isolate from ST80, ST78, and ST780-were sequenced by an Illumina platform and assembled to the scaffold level. The accession numbers, type of data, and status of the sequences are listed in Supplementary Table S1a. The presence of the vanB 2 gene was confirmed in VREfm isolates, as part of a chromosomally integrated Tn1549 transposon. The Tn1549 assembled sequence resulted in identical SSM5811 and SSM5812 isolates sharing 99.97% of sequence identity, with the Tn1549 sequence of E. faecium isolate E7654 from Utrecht, the Netherlands (Accession no. LR135324.1), and E. faecium AUS0004 (Accession no. CP003351.1). Compared to the vanB 2 gene cluster of AUS0004 (VAN MIC ≥ 16 mg/L and TEC MIC = 2 mg/L) used as a reference strain, our VREfm displayed single nucleotide mutations leading to amino acid substitutions in VanS (D105N), VanW (S204F), VanH (G55S/A231G), and VanB (S23A_I151M_M308L_M321V) proteins ( Figure 1). The assembled Tn1549 of the SSM5777 isolate showed 99.98% sequence identity with those of isolates SSM5811 and SSM5812, while the vanB 2 operon showed a slight genetic variation (99.89%), having a wild type vanS gene and different point mutations leading to amino acid substitutions in VanW (T61I_S94P) (Figure 1).
The SSM5777 was re-sequenced and analyzed after exposure to 100 mg/L of vancomycin, showing the same operon mutations observed before vancomycin exposure.
The SSM5777 was re-sequenced and analyzed after exposure to 100 mg/L of vancomycin, showing the same operon mutations observed before vancomycin exposure.  Table S1b), while fluoroquinolones resistance was associated with mutations in the quinolone resistance-determining region of topoisomerase genes leading to amino acid substitution in gyrA (Ser83-Tyr) and parC (Ser80-Ile). For these ten isolates, the phenotypic antibiotic susceptibility patterns aligned with the resistant genes identified (Supplementary Table S1b).

Discussion
Hospital settings provide a reservoir for MDR pathogens; thus, understanding molecular resistance mechanisms and epidemiology of circulating strains is crucial for preventing infection, including the emergence and dissemination of VREfm, which is of particular clinical concern [17].
In this study, we performed phenotypic and genotypic characterization of E. faecium isolates recovered during 2013-2018 from patients recovering in the main wards of the

Discussion
Hospital settings provide a reservoir for MDR pathogens; thus, understanding molecular resistance mechanisms and epidemiology of circulating strains is crucial for preventing Antibiotics 2023, 12, 476 6 of 10 infection, including the emergence and dissemination of VREfm, which is of particular clinical concern [17].
In this study, we performed phenotypic and genotypic characterization of E. faecium isolates recovered during 2013-2018 from patients recovering in the main wards of the University Hospital of Sassari. Our findings evidenced three isolates characterized by harboring highly mutated vanB 2 operons that were associated with variable vancomycin MICs representing, to our knowledge, the first description of vanB-VREfm in our hospital and in Sardinia.
These types of E. faecium going undetected by routine phenotypic testing are at risk of silent spread and are currently considered a developing threat because they can revert to a vancomycin-resistant phenotype upon exposure to vancomycin, causing treatment failures. This ability is typical of vanA E. faecium [12,14,18]; however, Hashimoto et al. described that Japan HA-E. faecium isolates harbor stealthy vanB 2 operons producing mutated VanW and VanB 2 proteins, which regained a resistant phenotype on exposure to vancomycin thanks to a mutation in the VanS sensor [15]. Our occult vanB 2 isolate showed three mutations in the VanB 2 protein that were the same as in the other two vanB 2 VREfm. Two of these mutations, V321M and M308L, were described to be involved in the decrease of vancomycin MIC in vanB 2 -VREfm isolates from Japan and Taiwan, respectively [15,19]. The presence of a VanS mutation (D105N) located in the same periplasmic domain [20] in reverted Japanese isolates [15] may be responsible for the higher vancomycin MIC in our VREfm compared to the occult VREfm, lacking this mutation.
Moreover, the analysis of the vanB 2 operon highlighted additional mutations in VanW and VanH proteins, including a A231G missense mutation in VanH, which occurs in an amino acidic residue essential for its dehydrogenase activity [21]. Considering this, we hypothesized that our vanB 2 -VREfm may represent reverted-stealthy vanB 2 E. faecium, and that vancomycin exposure may lead to similar mutations in the occult vanB 2 -VREfm, as previously demonstrated in Japanese isolates [15]. However, re-sequencing of the vanB 2 operon after vancomycin exposure did not detect additional mutations, suggesting an alternative mechanism for the increased vancomycin MIC. Further studies are needed to confirm this hypothesis and to clarify the role of single mutations in the modulation of the vanB 2 operon. The Tn1549 transposons detected in different years showed high nucleotide identity with common and univocal mutations, suggesting a common ancestral source.
Moreover, most E. faecium isolates, including the VSEfm, showed a consistently multidrug-resistant phenotype and genotype over the years. In sequenced isolates, an analysis of acquired antimicrobial resistance genes revealed that aminoglycosides and macrolide resistances were associated with the carriage of the aph(3 )-IIIa and/or aac(6 )-Ii, and/or aac(6 )-Ie-aph(2 )-Ia genes and the msr(C) and/or ermB genes, respectively.
Furthermore, despite the low number of isolates analyzed in this study, the polyclonal nature of nosocomial E. faecium STs is clearly shown.
Our results highlighted that vanB 2 -VREfm and nearly half of VSEfm analyzed were from the hospital-associated (HA) ST117/CC17 or CladeA1 [22]. These STs are frequently associated with nosocomial outbreaks in Europe [23][24][25] and have also been reported in Italian hospitals [26], along with the other HA-STs that were detected here (ST78, ST916, ST80, and ST780).
Finally, cgMLST subdivided ST117 into different CTs excluding the nosocomial transmission of a specific CT. However, the presence of ST117 over the years and its association to vancomycin resistance determinants compared to the other STs also capable of escaping phenotypic detection makes ST117 even more dangerous, confirming its high potential for diffusion.

E. faecium Isolates, Antimicrobial Susceptibility
Thirty-nine E. faecium strains isolated from patients admitted at the University Hospital of Sassari, Italy during 2013-2018, all part of the strains collection of the Microbiology Unit of the Biomedical Sciences Department of the University of Sassari, were included in the study. The isolation was performed at the Division of Microbiology and Virology of the University Hospital of Sassari, Italy using Slanetz Bartley or Blood Agar media (Microbiol UTA, Cagliari, Italy). Species identification was determined by Maldi-TOF Mass Spectrometry (Bruker, Billerica, MA, USA). After identification, antimicrobial susceptibility testing was conducted by Vitek2 using the AST cards P592 and P658 (bioMerieux, Marcy-l'Étoile, France), which are specifically for gram-positive Staphylococcus spp., Enterococcus spp., and Streptococcus agalactiae. For Enterococcus spp., among the antibiotics tested, we report results for ampicillin (AMP), high-level gentamicin (GEN), high-level streptomycin (STR), ciprofloxacin (CIP), imipenem (IMI), erythromycin (ERY), linezolid (LNZ), teicoplanin (TEC), vancomycin (VAN), and tigecycline (TGC) following the European Committee on Antimicrobial Susceptibility Testing (EUCAST) guidelines [27].
Antimicrobial susceptibility testing was also performed by MicroScan autoSCAN-4 using the AB33 Cards (Beckman Coulter Inc., CA, USA), following the EUCAST guidelines [27]. Glycopeptide susceptibility was additionally tested by the disk diffusion method [27] with E. faecalis ATCC 29212 as the control. . The presence of vanA, vanB, vanC1, and vanC2/C3 genes was found by Multiplex-PCR as previously described [28]. MLST was carried out on all isolates with the standard primers included in the E. faecium MLST scheme [29]. In addition, purified amplicons with DNA clean and concentrator TM -5 columns (Zymo Research, Irvine, CA, USA) were sequenced at the LMU Sequencing Service (Munich, Germany). Obtained sequences were analyzed using Geneious Pro 4.8.4 (http://www.geneious.com/ accessed on 1 March 2017). Allelic profiles and STs were assigned according to the E. faecium MLST database (http://pubmlst.org/efaecium/ accessed on 16 December 2018) and included in CCs based on eBURSTv3 analysis (http:// eburst.mlst.net/ accessed on 16 December 2018) used to assign isolates to genetic complexes and to evaluate the genetic relationships of the STs.
Generated sequences were assembled de novo into contigs using the software Velvet version 1.2.10 [30]. Contigs were reordered against the reference genome of E. faecium AUS0004 (Accession no. CP003351.1) using Mauve. Whole-genome alignment was done using the Artemis Comparative Tool (ACT) and MUMmer. The NUCmer tool of the MUMmer software was used to search individual gene sequences within the genomes. All ten genomes were submitted to the RAST platform for annotation (https://rast.nmpdr. org/ accessed on on 15 May 2021. The assembled genomes were subjected to the online search tools MLST 2.0 (Achtman MLST scheme) and ResFinder 4.0 available at the Center for Genomic Epidemiology (CGE) (http://www.genomicepidemiology.org/ accessed on 30 June 2021. The vanB 2 operon was resequenced using specific primers designed on the AUS0004 reference strain (CP003351) and shown in Supplementary Table S1c. Amplicons were purified, sequenced, and analyzed as described above.

Phylogenetic Analysis by cgMLST
The Ridom SeqSphere+ version 6.0.2 [31] whole-genome typing platform was used to infer phylogenetic relatedness of the isolates based on 2554 loci of the E. faecium cgMLST v1.1 and E. faecium MLSTv1.0 schemes, with cgMLST complex type/cluster-alert distance set as 20.

Conclusions
In this work, we highlighted a highly mutated vanB 2 -VREf ST117 occult isolate by Vitek2 with the ability to increase its vancomycin MIC after exposure to vancomycin. This isolate was found to have similarities to both vanA-VVEfm and stealthy vanB 2 -E. faecium.
The risk of non-detection of vanB 2 -VREfm by the Vitek2 system is of great concern, and E. faecium should always be processed using an alternative method to Vitek2 to avoid therapeutic failure and vanB 2 -VREfm spread.
Stricter preventive measures, such as isolation of infected/colonized patients and more accurate disinfection procedures, can reduce the risk of dissemination of these coverts and all HA-VREfm/VSEfm clones.