A Genomic Snapshot of Antibiotic-ResistantEnterococcus faecalis within Public Hospital Environments in South Africa

Enterococci are among the most common opportunistic hospital pathogens. This study used whole-genome sequencing (WGS) and bioinformatics to determine the antibiotic resistome, mobile genetic elements, clone and phylogenetic relationship of Enterococcus faecalis isolated from hospital environments in South Africa. This study was carried out from September to November 2017. Isolates were recovered from 11 frequently touched sites by patients and healthcare workers in different wards at 4 levels of healthcare (A, B, C, and D) in Durban, South Africa. Out of the 245 identified E. faecalis isolates, 38 isolates underwent whole-genome sequencing (WGS) on the Illumina MiSeq platform, following microbial identification and antibiotic susceptibility tests. The tet(M) (31/38, 82%) and erm(C) (16/38, 42%) genes were the most common antibiotic-resistant genes found in isolates originating from different hospital environments which corroborated with their antibiotic resistance phenotypes. The isolates harboured mobile genetic elements consisting of plasmids (n = 11) and prophages (n = 14) that were mostly clone-specific. Of note, a large number of insertion sequence (IS) families were found on the IS3 (55%), IS5 (42%), IS1595 (40%), and Tn3 transposons the most predominant. Microbial typing using WGS data revealed 15 clones with 6 major sequence types (ST) belonging to ST16 (n = 7), ST40 (n = 6), ST21 (n = 5), ST126 (n = 3), ST23 (n = 3), and ST386 (n = 3). Phylogenomic analysis showed that the major clones were mostly conserved within specific hospital environments. However, further metadata insights revealed the complex intraclonal spread of these E. faecalis major clones between the sampling sites within each specific hospital setting. The results of these genomic analyses will offer insights into antibiotic-resistantE. faecalis in hospital environments relevant to the design of optimal infection prevention strategies in hospital settings.


Introduction
Te surveillance of hospital environments can be a useful tool to better understand the opportunistic microbial communities within the hospital (Comar et al., 2019), to identify the source of an outbreak [1], and to evaluate the efcacy of environmental disinfection or other infection prevention and control measures [2]. Inadequate control practices have played a signifcant role in the dissemination, persistence, and intra and interhospital spread of drug-resistant organisms.
Regrettably, good clinical trials comparing diferent approaches to and the impact of infection prevention and control interventions on the control of drug-resistant bacteria in hospitals and other healthcare facilities are minimal [3,4]. Accurate identifcation of resistant bacterial reservoirs and modes of transmission helps to inform such interventions.
Te latest successes in tracing worldwide epidemics [5] and nosocomial outbreaks [6] have been attributed to wholegenome sequencing (WGS). Genomic comparison has aided our understanding of the evolution and spread of infectious agents. Comparative genomic analyses have been made possible through the use of WGS, showing the extent of genomic variation, which may result in varied phenotypes, thus expanding our understanding of diverse genomic determinants such as antibiotic resistance genes and their mobile genetic elements in bacterial species [6,7].
Enterococcus faecalis (E. faecalis) is a good indicator bacterium in hospital environment monitoring, and being Grampositive cocci, these opportunistic pathogens not only form noxious bioflms on implanted medical devices and catheters but also they cause abdominal infections, urinary tract infections, surgical site wound infections, bacteremia, endocarditis, and burn wound infection [8]. Antibiotic resistance is either intrinsic or through sporadic mutation or through the acquisition of foreign genetic material, by horizontal gene exchange occurring with the aid of mobile genetic elements plasmids, prophages, and insertion sequences [9,10]. Difculties in treating E. faecalis and E. faecium (the most prevalent species in human) have emerged due to acquired resistance, predominantly multidrug resistance to universally used drugs as well as vancomycin [11]. A number of previous surveillance studies involving E. faecalis in Africa have focused either on wastewater treatment plants (WWTPs) and hospital efuent but not on the internal hospital environment [12,13]. Moreover, in South Africa, studies on the contamination of E. faecalis, using high discrimination resolution typing, are scarce. Tis study, therefore, uses WGS in delineating the resistome, mobile genetic elements, clones and phylogenomic relationship of E. faecalis isolated from the hospital environment in places frequently touched by patients and healthcare workers at four diferent levels of healthcare in the metropolitan city of Durban, South Africa.

Enterococcus Collection.
Te isolates described here at a genomic level were part of a previously reported study and appeared in [14]. Briefy, four public hospitals situated in the eTekwini region in Durban, South Africa were sampled and referred to as A, B, C, and D, representing central, tertiary, regional, and district facilities, respectively. Each hospital is classifed according to the Department of Health and depends on the level of service it can provide to the community. Te sample sites viz. ward telephones, ventilators, blood pressure apparatus, patient fles, drip stands, sinks, occupied beds, unoccupied beds, nurses' tables, mops, and the door handle of the linen room were situated in the general ward, intensive care unit, and paediatrics. Suspected colonies were phenotypically identifed using API 20 Strep kits (bioMerieux SA, Marcy I 'Etoile, France). Staphylococcus aureus ATCC 29213 and E. faecalis ATCC 29212 served as controls. Molecular confrmation of isolates was also previously described [15]. To summarise, of the 620 samples taken, 295 Enterococcus spp. were obtained, of which 245 were confrmed as Enterococcus faecalis via phenotypic and molecular assays. A subsample of 38 E. faecalis isolates was selected for genotypic characterization by WGS and bioinformatics analysis (Table 1).

DNA Isolation, Genome Sequencing, Assembly, and
Annotation. Genomic DNA (gDNA) was extracted using GenElute bacterial genomic DNA kit (Sigma-Aldrich, St. Louis, Missouri, United States) according to the manufacturer's instructions. Te quantifcation of extracted gDNA was determined on a Nanodrop ND1000 spectrophotometer (Termo Scientifc, Waltham, USA) and Qubit 2.0 fuorometer (Invitrogen, Oregon, USA) and verifed on an agarose gel electrophoresis. Multiplexed paired-end libraries (2 × 300 bp) were prepared using the Nextera DNA Flex sample preparation kit (Illumina, San Diego, California, United States) and sequences determined on an Illumina MiSeq platform with 100x coverage at the National Institute of Communicable Diseases Sequencing Core Facility, South Africa. Te resulting raw reads were checked for quality, trimmed, and de novo assembled into contigs using the CLC Genomics Workbench version 10.1 (CLC, Bio-QIAGEN, Aarhus, Denmark). Default parameters were used for all software unless otherwise specifed. Te CheckM tool version 0.9.7 [18] was used to verify that the sequence reads were not from mixed species using lineage-specifc marker sets from other genetically well-characterised closely related E. faecalis isolates. Te de-novo-assembled reads were uploaded in GenBank and annotated using National Centre   Global Health, Epidemiology and Genomics for Biotechnology Information (NCBI) prokaryotic genome annotation pipeline and Rapid Annotations using Subsystems Technology (RAST) 2.0 server [19].

WGS-Based Molecular Typing of E. faecalis Isolates.
Multilocus sequence typing (MLST) was performed in silico using the WGS data online platform tool MLST 1.8 [20] which also predicted the allelic profles of the seven housekeeping genes, aroE, gdh, gki, gyd, psts, xpt, and yqil of E. faecalis as described previously [21].

Phylogenomic Analysis of Enterococcus faecalis Isolates.
Te de-novo-assembled contigs were uploaded, and the analysis was submitted to CSI (called SNPs & Infer) Phylogeny-1.4 (https://cge.cbs.dtu.dk/services/CSIPhylogeny-1.2), an online service which identifes single-nucleotide polymorphism (SNPs) from WGS data, flters and validates the SNP positions, and then infers phylogeny based on concatenated SNP profles [22]. Te pipeline was run with default parameters: a minimal depth at SNP positions of 10 reads, a minimal relative depth at SNP positions of 10%, a minimal distance between SNPs of 10 bp, a minimal Z-score of 1.96, a minimal SNP quality of 30, and a minimal read mapping quality of 25. A bootstrapped with 100 replicate indicators was applied to identify recombined regions and provide phylogenetic accuracy in groups with little homoplasy. Te fgtree (https://tree.bio.ed.ac.uk/software/fgtree/) was used to edit and visualise the phylogenetic tree. Te phylogeny was visualised alongside metadata for isolate demographics (including hospital, source, ward), sequence type, and antibiotic resistome using Phandango [23] to provide a comprehensive analysis to understand the complex intraclonal spread within each specifc hospital setting using the generated phylogenomic tree.

Data
Availability. Te raw read sequences and the assembled whole-genome contigs have been deposited in GenBank. Te data are available under project number PRJNA 23601.
For the tertiary hospital, it was the ward phones (14.4%) and mops (10.8%). Te central hospital showed a similar outcome as the regional hospital with the highest contaminated sites as occupied beds (22.2%), nurses' tables, and mop (20.3% each). Looking at the highest contaminated sites, the risk of mops as vehicles of contamination and probable cross-contamination within the hospital environment is highlighted.

Vancomycin
Susceptibility. Te authors of [11] described the antibiotic susceptibility testing, and these revealed that none of the 245 identifed E. faecalis isolates was vancomycin-resistant (VRE). However, a total of thirty-eight (38) were of intermediate susceptibility to vancomycin and were selected for genotypic characterization by WGS and bioinformatics analysis (Table 1). Tese 38 vancomycin-intermediate isolates were predominantly resistant to tetracycline (n � 31, 81%) followed by resistance to erythromycin (n � 18, 47%). A small number of isolates showed aminoglycoside resistance (gentamicin (n � 4) and streptomycin (n � 6)). Te majority of the isolates were susceptible to ampicillin, penicillin, teicoplanin, and levofoxacin while all isolates were susceptible to nitrofurantoin (Table S2)

Antibiotic Resistance Profles and Resistance Genes of E. faecalis Isolates.
Te phenotypic resistance profles displayed by the isolates are shown in Table S1. In total, 14 antibiotic resistance genes (ARGs) and variants were detected (Table 1). Tere were no specifc diferences in the resistome with regard to their hospital levels and wards. Te frequency of ARGs ranged between 2-13 genes, with ffteen isolates carrying 3 resistance genes. Acquired ARGs confer resistance to tetracycline (tet(M) and tet(L)), macrolidelincosamide-streptogramin B (MLS B ) (erm(B) and mphD), aminoglycosides (sat4A, aph3-lll, ant6-la, aac6-aph2), trimethoprim-sulfamethoxazole (dfrG and dfrK), and phenicols (catA and optrA) were found in the isolates as shown in Table 1. Te tet(M) and erm(B) genes were found in 82% (31/38) and 42% (16/38) of the isolates, respectively. Te dfrG gene predominately caused resistance to trimethoprim-sulfamethoxazole (Tables 1 and S2). In some cases, the phenotypes were not corroborated by ARGs, as some isolates expressed resistance to antibiotics in the absence of the associated resistance determinants (Tables 1 and S1).
A myriad of FAMILIES was found in the isolates with no association with respect to the hospital and ward. Te 5 major FAMILIES were IS3 (predicted to be linked with Enterococcus faecium/Streptococcus agalactiae source), IS5 (predicted to be associated with Cyanotheca sp. source), IS1595 (predicted to be linked with Bacillus subtilis), ISL3 (predicted to be linked with Streptococcus mutans/thermophilus), and IS607 (predicted to be linked with both Campylobacter sp. and Virus NY2A) (Table S3). Te transposase (Tn3) linked to Bacillus thuringiensis was found in 7 of the isolates identifed from diferent sources (Table S3). All isolates lacked integrons and their associated gene cassettes.

Phylogenomic and Metadata Analysis.
A phylogenetic tree reconstructed to analyse the genetic relationships between the isolates revealed a high divergence of isolates according to the diferent levels of care (Figure 1). For instance, each hospital was generally associated with specifc dominant clones (i.e., ST40 and ST498 were mostly found in hospital A; ST16, ST126, and ST386 were found in hospital B; and ST21 was predominately found in hospital C (Table 1 and Figure 1)).
Te phylogenomic tree coupled with metadata visualization analysis provided a more in-depth insight into the characteristics and distinctions between isolates and revealed the intraclonal spread of E. faecalis strains between diferent sources within the same hospital ( Figure 1). Specifcally, ST21 was found on the drip stand, patient fle, sink, and nurse table in both ICU and paediatric wards of hospital C. Similarly, ST40 was found on the phone, patient fles, mops, occupied bed, and nurse tables of the paediatric ward hospital A. Te ST16 clone was isolated from the mop (paediatric ward), phone, and BP apparatus (ICU) of hospital B. More so, ST386 was linked with the phone, BP apparatus, and unoccupied bed in the paediatric ward of hospital B, while ST126 was found with the occupied bed and nurses table in the ICU of the same hospital. Te prophage Entero_vB_IME197 was mostly found among all ST21 strains and one ST41 strain (2CPF3) that derived from the regional hospitals; however, another ST21 strain (2UIK3) from a tertiary hospital also carried this prophage. Te same prophage has been reported in an E. feacium strain belonging to ST 179 sourced from a wastewater plant (Mbanga et al., 2021). It has been reported that for Enterococcus species, it is more likely that isolates with common phagerelated genes share common environments or a specifc niche (Bonacina et al., 2016)

Discussion
In line with the global trend, reports on bacterial contamination in hospital environments are increasing in Africa across all sectors [31], and E. faecalis is one of the most common enterococcal species isolated from the hospital environment. Tis is evident from our results where E. faecalis (n � 245) was the most prevalent organism compared to E. faecium (n � 50). E. faecalis is recognised as an important hospital-associated pathogen responsible for approximately 80-90% of cases reported in hospital settings, Global Health, Epidemiology and Genomics  6 Global Health, Epidemiology and Genomics followed by 5-10% E. faecium [32] and hence has been placed in the category of pathogens posing a major threat to healthcare systems [33]. Furthermore, E. faecalis represents a major infection prevention attributed to their ability to persist for long periods on hands and remain viable on environmental surfaces (inanimate surfaces) due to their microbial structure thus, which can serve as a reservoir for ongoing transmission in the absence of regular decontamination [34]. Additionally, E. faecalis possesses the ability to acquire additional resistance through the transfer of mobile genetic elements such as plasmids, prophages, and insertion sequences [35,36]. Te acquisition of resistance and genetic elements poses a therapeutic challenge.
Te WGS results showed that none of the E. faecalis harboured vancomycin-resistant genes, which corroborated with the phenotypes [11]. Tis afrms the view of Ellington et al. on the role of WGS in antimicrobial susceptibility testing of bacteria for the explanation of phenotypic results for samples [37] and further confrms WGS as a more discriminatory tool to infer antibiotic susceptibility compared to relying fully on phenotypic testing alone. Additionally, the discordance between the phenotype and the genotype in Enterococcus spp. could be due to impaired penetration of the antibiotic, hetero-resistance, or limitations of the current traditional phenotypic (antimicrobial susceptibility testing) methods [38][39][40]. More so, while the presence of specifc resistance genes suggests a potential mechanism for antibiotic resistance, it is important to note that other factors, such as gene expression, regulatory elements, and efux pumps, can infuence the observed phenotypic resistance [38,40]. Te majority of the isolates were susceptible to ampicillin, penicillin, teicoplanin, levofoxacin, and nitrofurantoin confrming their use as treatment options in South Africa, particularly ampicillin (the drug of choice for E. faecalis infections) [41].
In our result, tetracycline demonstrated reduced susceptibility to E. faecalis mediated mostly by the ribosomal protection protein, tet(M) [10,42]. Tis was consistent with previous studies that found tet(M) as the dominant gene causing tetracycline resistance in E. faecalis isolates across all one-health sectors (human-animal-environment interface) [31]. For instance, in a 2014 hospital-based study in China by Jia et al. [43], tet(M) was found to cause tetracycline-resistantE. faecal isolates. Similarly, Said et al. [44] also detected tet(M) as 96.1% of all tetracycline-resistantEnterococcus isolates in Egypt. However, the tetracycline resistance exhibited by the 2SIL2 isolates was mediated by both the ribosomal-protection gene [tet(M)] and the active-efux gene [tet(L)]. Tis indicates the signifcant role played by efux pumps in mediating antibiotic resistance [45]. Te low prevalence of the tet(L) was not unusual and pointed to the fact that ribosomal protection protein is the main mechanism of tetracycline-resistantE. faecalis isolates. Our result showed that a moderate level of erythromycin resistance was mediated by erm(B) genes which are the most common mechanism of resistance reported for the macrolide class of antibiotics in Africa [31] and globally [10,46] for Enterococcus.
High levels of resistance to aminoglycosides have developed since its clinical utility for achieving bactericidal synergism in combination with cell wall-active agents, which is important in the treatment of enterococcal endocarditis [47]. Tis level of resistance is owing to the appearance of aminoglycoside-modifying enzymes and contradicts the synergistic advantage of the combinations in the clinical setting. Te gene encoding the most known enzyme conferring resistance to aminoglycosides excluding streptomycin is aac-6′-Ie-aph-2″, typically discovered inside Tn4001 in staphylococci and other variants in enterococci [48]. From our result, only a small number of isolates showed aminoglycoside resistance across the diferent levels of care, which corresponded to the aminoglycosidemodifying enzymes found. However, these isolates exhibited high-level resistance encoding a set of enzymes (sat4A, aph3-lll, ant6-la, aac6-aph2) although this was not unusual  Figure 1: Te whole-genome MLST phylogenomic branch and metadata of isolate information (including isolate identity, hospital, source, and ward) and WGS in-silico typing (sequence type and antibiotic resistome) coupled using Phandango (https://jameshadfeld. github.io/phandango/) E. faecalis isolates at diferent level of care in Durban, South Africa. Te linking lines in the phylogenetic tree diferentiate between the diferent clades. Metadata annotations show that there were generally distinct major sequence types between the 4 hospital environments; however, within each hospital, there was the spread of these major clones between diferent sources in the wards.
as some Enterococcus spp. are known to produce low-level resistance to aminoglycosides by limiting drug uptake, which is associated with proteins involved in electron transport [49]. Furthermore, although the OptrA gene associated with linezolid resistance was detected in only one isolate (2SIL2); however, it was unexpressed as the isolate was susceptible to linezolid (Tables 1 and S2). Tis fnding was similiar to a study conducted in China by Li et al. (2020), where the OptrA gene was identifed in eighteen linezolid-sensitive enterococci [50]. A noticeable polyclonal nature was observed in the E. faecalis isolates with 15 distinct STs, including one novel STs, highlighting the diverse nature of the strains in the province. Te major STs found such as ST16, ST40, and ST21 were previously reported in Saudi Arabia, China, Tunisia, France, and Spain from human subjects, hospitalised patients, animals, and wastewater (Farman et al., 2019;[51-54]. Similarly, other studies have also reported ST126, ST23, and ST386 in diferent settings (human, animal, and environment) and hence do not suggest any kind of host specifcity in these major STs reported in this study [55]. However, unlike other countries, the population structure E. faecalis from diferent settings in South Africa is minimally monitored, if at all, making it difcult to correlate our results with studies in South Africa. In a clonal analysis report of E. faecalis sourced from patients in a Chinese tertiary hospital, ST16 was predominantly present in urinary tract infections. In a Saudi-Arabian study, it accounted for the second most strain types and was also related to urinary tract infections. Tis calls for the need for E. faecalis to be included in surveillance schemes to enable the monitoring of the molecular epidemiology of isolates collected over larger tempo-spatial scales using high throughput technologies such as WGS [56]. Such surveillance would help microbiologists and public health practitioners to gain better insights into the evolution and dissemination of E. faecalis.
Characterizing the mobile genetic elements of the isolates indicated that the majority of E. faecalis in diferent hospitals are likely reservoirs for diverse mobile genetic elements and associated ARGs (especially for tetracycline and erythromycin). Tere was a higher plasmid prevalence rate (seven rep families) and the detection of two or more distinct replicons in one strain. Accordingly, this fnding agrees with the fact that numerous types of plasmids are often present in enterococci in a clinical setting [57][58][59]. More so, other studies have shown that single isolates of E. faecalis may harbour multiple plasmids [57,60].
Tere was no specifc pattern between the acquisition of insertion sequence families or transposable elements with respect to the word and level of care; however, the presence of major FAMILIES in E. faecalis clones implies that these elements are spread by horizontal gene transfer (HGT) [35]. Moreover, the acquisition of these elements can lead to transposition in the genome aid in the transfer of resistance genes, enabling it to adapt to new environmental challenges and colonise new niches [61]. For instance, the IS3 family upstream of the EmrB gene has been reported for enhanced erythromycin resistance [61]. Te ability of these clones to acquire novel genetic features may contribute to their increased persistence and highlights their potential public health threat.
Comparative phylogenomics using WGS SNPs analysis revealed a higher genetic diversity between the strains with respect to each specifc hospital. Tis implied that the major clones were mostly hospital-specifc, which was in concordance with the in-silico MLST typing scheme ( Figure 1). Interestingly, a study by Kawalec et al. [62] also found a higher diversity in the clonal structure of E. faecalis strains among hospitals in Poland. Visualizing the phylogenomic tree with metadata revealed the major clones in various hospitals. Tis further depicted the intraclonal spread of E. faecalis strains between diferent sources within the same hospital, reiterating the need for phylo/meta-analysis to increase confdence in molecular epidemiological studies. For instance, at the paediatric ward of hospital A, the ST40 clone was isolated from a phone, nurse table, patient fles, mop, and occupied bed, which may be due to hand contamination by patients and/or healthcare workers (nurses, janitor staf, etc.) (Figure 1). A similar scenario occurred in hospital B, where ST386 was found in the paediatric ward (on the phone, BP apparatus and unoccupied bed) while ST126 was isolated in the ICU (on the nurses' table and occupied bed). Reports on enterococci transient carriage on the hands of healthcare workers and patients as well as their presence on medical equipment or environmental surfaces have been documented in several studies [63][64][65][66]. Other studies have reported the movement of colonised patients among diferent settings in the hospital as responsible for these patterns of transmission [64,67]. Moreover, hospitals B and C observed intraward spreads (both ICU and paediatric ward) of ST16 and 21, respectively, from diferent sites in each hospital. Te transmission of enterococcal strains has been documented within medical units, giving credence to the study fndings [68,69].
Frequent contact with healthcare providers and movement of colonised patients among diferent healthcare settings are probable means for these patterns of transmission in hospitals A, B, and C. However, there were limited isolates from the district hospital (Hospital D) due to the number of isolates obtained for any detailed comparative analysis. Even though the fndings of our study may not be generalised to the overall situation in the country, this study improves the understanding of the prevalence, genetic content, and relatedness of E. faecalis contamination in hospital environments. It is thus recommended that scheduled periodic identifcation of transmitting sources in hospitals' inanimate environment, strict enforcement and adhesion of IPC practices amongst health workers, and isolation of colonised patients should be imposed to reduce the incidence and transmission of E. faecalis in hospital environments. More so, the study was limited by the number of isolates selected for sequencing, and hence, there is a need for large-scale genomic epidemiology to elucidate the population structure in various hospital environments in South Africa.

Conclusion
Tis genomic analysis provided a snapshot of the hospital inanimate environment as a reservoir of resistant E. faecalis, its associated mobilome (plasmids, prophages, insertion sequences, and transposons), and revealed a complex intraclonal spread of E. faecalis major clones between the sites within each specifc hospital setting. Tis study enhances our understanding of how E. faecalis spreads in hospital environments and investigates the role of the hospital environment in acquiring resistant genes, which poses a threat to the efective treatment of common infectious diseases. Tis will aid in the design of optimal infection prevention and control strategies in clinical settings. Te fndings may not represent the entire population of E. faecalis strains within the hospital settings; however, the observed trends in the major clones provide valuable preliminary information about their distribution and transmission patterns. Tese results can serve as a basis for further investigations and hypothesis generation in future studies.

Data Availability
Te generated data used to support the fndings of this study are included in the article.

Disclosure
Any opinion, fndings, and conclusions or recommendations expressed in this material is those of the authors, and neither the NRF nor the other funding bodies accept any liability in this regard.

Conflicts of Interest
Te authors declare that they have no conficts of interest.