Evolution and mutations predisposing to daptomycin resistance in vancomycin-resistant Enterococcus faecium ST736 strains

We recently identified a novel vancomycin-resistant Enterococcus faecium (VREfm) clone ST736 with reduced daptomycin susceptibility. The objectives of this study were to assess the population dynamics of local VREfm strains and genetic alterations predisposing to daptomycin resistance in VREfm ST736 strains. Multilocus sequence typing and single nucleotide variant data were derived from whole-genome sequencing of 250 E. faecium isolates from 1994–1995 (n = 43), 2009–2012 (n = 115) and 2013 (n = 92). A remarkable change was noticed in the clonality and antimicrobial resistance profiles of E. faecium strains between 1994–1995 and 2013. VREfm sequence type 17 (ST17), the prototype strain of clade A1, was the dominant clone (76.7%) recognized in 1994–1995. By contrast, clone ST736 accounted for 46.7% of VREfm isolates, followed by ST18 (26.1%) and ST412 (20.7%) in 2013. Bayesian evolutionary analysis suggested that clone ST736 emerged between 1996 and 2009. Co-mutations (liaR.W73C and liaS.T120A) of the liaFSR system were identified in all ST736 isolates (n = 111, 100%) examined. Thirty-eight (34.2%) ST736 isolates exhibited daptomycin-resistant phenotype, of which 13 isolates had mutations in both the liaFSR and cardiolipin synthase (cls) genes and showed high level of resistance with a daptomycin MIC50 of 32 μg/mL. The emergence of ST736 strains with mutations predisposing to daptomycin resistance and subsequent clonal spread among inpatients contributed to the observed high occurrence of daptomycin resistance in VREfm at our institution. The expanding geographic distribution of ST736 strains in other states and countries raises concerns about its global dissemination.

Introduction laundry facility of some US teaching hospitals [31,35]. The objectives of this study were to assess the evolution in clonality and antimicrobial susceptibility profiles of local VREfm populations, to explore the potential role of genomic mutations and nosocomial transmission in the emergence and spread of daptomycin-nonsusceptible ST736 strains.

Enterococcus faecium clinical isolates
A total of 250 E. faecium clinical isolates, including 239 VREfm and 11 vancomycin-susceptible E. faecium (VSEfm) isolates, were included in this study. All isolates were recovered from patients with an infection, with the exception of one from an environmental sample, in a tertiary-care medical center of suburban New York City. Isolates were collected from three different study periods: 1) 1994-1995 (n = 43): Forty-nine VRE isolates were randomly selected from a collection of saved isolates during an outbreak investigation from July 1994 through July 1995. Of these, 43 E. faecium-VRE isolates were included, while 6 E. faecalis-VRE isolates were excluded in the analysis; 2) 2013 (n = 92): this included all non-duplicate, consecutive VREfm isolates recovered from January through October 2013; and 3) 2009-2012 (n = 115): which comprised all confirmed daptomycin-nonsusceptible E. faecium (DNSEfm) isolates and representing daptomycin-susceptible isolates spanning different months of each year. One E. faecium isolate per patient was chosen unless there were two isolates from the same patient with different sequence types (ST) or with different susceptibility (susceptible vs. nonsusceptible) to daptomycin. All enterococci isolates were identified by conventional biochemical tests and confirmed using the MALDI Biotyper CA system (Bruker, Billerica, MA) and/or DNA sequencing analysis of 16S rRNA gene. Antimicrobial susceptibilities of E. faecium isolates were measured by broth microdilution with the MicroScan 96Plus test system and daptomycin-nonsusceptibility (MIC > 4 μg/mL) was confirmed by E-test as described previously [27].

Bayesian phylogenetic analyses
To investigate the phylogenetic relationship and the divergence time of E. faecium circulating at local patient population, Bayesian Evolutionary Analysis by Sampling Trees 2 (BEAST 2) v2.4.7 [40] was used to jointly estimate molecular clock phylogeny, rate of evolution, divergence times and other evolutionary parameters. By using Snippy, a reference-based whole-genome alignment was generated to include all 250 E. faecium samples. To reduce excessive computational load, three isolates with low depth of genomic coverage were excluded, which resulted a final of 247 samples and 487,932 sites (479,196 invariant sites) for the BEAST analysis.
Bayesian Markov Chain Monte Carlo (MCMC) analysis was performed using the Hasegawa-Kishino-Yano (HKY) [41] nucleotide substitution model, along with a coalescent constant population tree prior and a discrete gamma-distributed among-site-rate-variation model with four categories [42]. A strict molecular clock model was employed, and tip dates were set from the sample collection dates. The MCMC chain was run for 1.5 billion states and sampled every 100,000 states. 10% sampled states were discarded as burn-in. Convergence was assessed using Tracer v1.7 [43], and effective sample size (ESS) values above 200 were accepted. A maximum clade credibility (MCC) tree was summarized with TreeAnnotator (available in the BEAST 2 v2.4.7 package) and visualized in Figtree v1.4.4 (http://tree.bio.ed.ac.uk/software/figtree/).

Mutation analysis of liaFSR and cls genes
The assembled genome of each isolate with reference to the E. faecium strain E39 (ST736, accession numbers CP011281-CP011285, CP015123) was blasted against the liaS (1,069 bp) and the liaR (634 bp) gene sequences of strain DO (ST18, accession no. CP003583), and the cls (1,452 bp) gene sequence of strain UW7606x64/3 TC1 (ST192, accession no. CP013009). Corresponding gene sequences for each isolate were extracted. SNPs of liaFSR and cls genes in all E. faecium isolates and genetic alterations between isolates of each isogenic pair were called out using SNP-sites (http://dx.doi.org/10.1099/mgen.0.000056), in association with daptomycin resistance from the CARD database (https://card.mcmaster.ca/).

Statistical analysis
The Fisher's exact test from the GraphPad Prism software (version 7.0) was used to determine the statistical significance of association between the different STs of E. faecium with distinct clinical and microbiological characteristics.

Accession numbers
The complete genomes of four representative ST736 strains (E39, E232, E243, and E240) were deposited to GenBank with accession numbers of CP011281-CP011285, CP015123, and CP017787 to CP017801. The untrimmed Illumina raw sequencing reads for all 250 E. faecium isolates were uploaded to GenBank under BioProject PRJNA386994.

Clonality of VREfm population from 1994-1995 vs. 2013
To determine the population dynamics and evolution of VREfm clinical isolates and correlation between distinct clones and daptomycin susceptibility, we determined the sequence types (ST) of 92 non-duplicated, consecutive VREfm isolates from 86 inpatients hospitalized during January through October 2013. Two of these patients carried VREfm isolates with different STs (ST412 and ST736), while four patients had VREfm isolates with one isolate susceptible and another nonsusceptible to daptomycin. Eight distinct STs with three dominant clones (ST736, ST18 and ST412) were recognized among the 2013 VREfm clinical isolates. Of these, 43 (46.7%) isolates belonged to ST736, followed by ST18 (n = 24, 26.1%), ST412 (n = 19, 20.7%) and five other STs each with one to two isolates (Table A in S1 File).
In addition, the clonality of 115 E. faecium isolates from 2009 to 2012 were examined. ST736 strains were detected in all years between 2009 and 2012. The number of E. faecium isolates from different study years and distribution in clonality are summarized in Table A in S1 File.

Evolutional analysis on the emergence of VREfm ST736 clone
To explore the evolution of VREfm and estimate the possible emergence time of VREfm ST736 at our institution, we selected E. faecium isolates with a minimum breadth coverage of

Association between distinct VREfm clones and daptomycin susceptibility
To further our understanding [27] on the association between different VREfm clones and daptomycin nonsusceptibility, we analyzed 92 VREfm isolates representing an unbiased collection of all hospitalized patients with VREfm infections from January through October 2013. Interestingly, clone ST736 strains accounted for only 46.7% (43 of 92) of all VREfm isolates examined in 2013 but 76.9% (20 of 26) of DNSEfm from this study period (p = 0.0002). The prevalence of ST736 as DNSEfm (46.5%, 20 of 43) was significantly higher than that of non-ST736 VREfm isolates from the same study period (6 of 49, 12.2%, p < 0.001). The distribution of daptomycin MICs of different E. faecium clones were shown in Fig 3 and Tables B and C in S1 File. The daptomycin MIC 90 of ST736 strains was 32 μg/mL, which was significantly higher than those from other non-ST736 strains (MIC 90 : 8 μg/mL, p <0.01).

Mutations in liaFSR and cls and daptomycin resistance
The mutations and frequency of liaFSR and cls reported to be associated with daptomycin resistance in E. faecium [45] were analyzed and compared between ST736 and non-ST736 isolates ( Table 2). Two mutations in liaFSR (W73C and T120A) and several mutations in cls (N13I, N13T, A20D, H215R, R218Q) were detected, while the remaining mutations described by Arias et al. [14] and listed in the CARD database were not seen in our isolates. Strikingly, all ST736 isolates (n = 111, 100%) examined had co-mutations (W73C and T120A) in the liaFSR, which was significantly higher than that detected in non-ST736 isolates (17 of 139, 12.2%, p < 0.0001).
Thirty-eight of 111 (34.2%) ST736 isolates with liaFSR mutations were resistant to daptomycin, while all 13 ST736 isolates with both liaFSR and cls mutations exhibited a daptomycin- resistant phenotype. Moreover, ST736 isolates with cls mutations (n = 13) appeared to confer high resistance to daptomycin (MIC 50 of 32 μg/mL), compared to those without cls mutations (MIC 50 : 8 μg/mL) ( Table 3 and Table B in S1 File). Thirty of 111 (27.0%) VREfm ST736 isolates had prior exposure to daptomycin within 12 months before their recovery from patients. VREfm ST736 isolates from patients with prior daptomycin exposure were more likely to be resistant to daptomycin (20/30, 66.7%) than those without prior exposure (18 of 81, 22.2%, p < 0.0001) with a relative risk of 3.0 (95% confident interval: 1.85-4.84). Moreover, 10 of 13 ST736 isolates with cls mutations had prior use of daptomycin. As shown in Table 4, four of six patients with isogenic pair of ST736 isolates developed resistance during daptomycin therapy by inducing cls mutations.   Evolution and mutations predisposing to daptomycin resistance in E. faecium ST736 strains on the chromosome ( Table 5). In addition, mutations in different insertion sequence (IS) family transposases were noticed in the plasmids of both isolates.

Nosocomial transmission of VREfm ST736
To assess potential nosocomial transmission in spreading DNSEfm, a phylogenetic tree with ST736 isolates from 2013 was constructed (Fig 4A). Notably, 31 of 37 ST736 VREfm isolates Evolution and mutations predisposing to daptomycin resistance in E. faecium ST736 strains analyzed fell into a closely related cluster with less than 50 SNVs among different isolates. Of these, 8 VREfm isolates from 6 patients revealed a distinct subcluster with less than 10 SNVs (0-9 SNVs). As shown in Fig 4, both epidemiological links (e.g., patients staying in the same hospital bed/room, visiting the same clinic on the same day, or hospitalized during the same period with at least two weeks of overlap) and SNVs-based genomic evidence were identified among the 6 patients in this subcluster, supporting the occurrence of nosocomial transmission [46].

Discussion
In this study, we sought to determine if ST736 had long been in existence or only recently emerged at our institution. Comparative genomic analysis of clinical isolates revealed a dramatic change in the clonality and antimicrobial resistance profiles of local E. faecium population from 1994-1995 to 2013. The dominant clone of VREfm isolates in 1994-1995 was ST17 (77%), a prototype strain of CC17 or clade A1, when VREfm started emerging in the mid-1990 in the US [47]. No ST736 clones were identified among VREfm isolates from 1994-1995. By contrast, ST736 strains became predominant, accounting for 46.7% of local VREfm isolates in 2013. E. faecium ST18 appeared to be only clone continuously detectable in both study periods (1994-1995 and 2013 <0.0001). The probability of an ST736 isolate to be daptomycin-nonsusceptible is significantly higher than that of a non-ST736 isolate (p = 0.0002) with an estimated relative risk of 3.9 (95% confidence interval: 1.7 to 8.9). Our data highlight the risk of Evolution and mutations predisposing to daptomycin resistance in E. faecium ST736 strains ST736 clone in dissemination of daptomycin resistance and challenge in management of patients infected with ST736 VREfm strains. The mechanisms of resistance to daptomycin appear to be diverse for VREfm strains with different genetic backgrounds [10,45,48]. Both de novo and developed resistance during daptomycin therapy have been reported [10]. Among several genomic pathways proposed, a hierarchical adaption via sequential mutations in the liaFSR signaling system and cls appears to be the most common mechanism associated with daptomycin resistance identified in enterococci [15,16,19,[49][50][51]. Since most of published data are based on analysis of a limited number of clinical isolates and/or laboratory-derived mutants, our current report represents the largest collection of E. faecium clinical isolates that can serve as an independent validation on the . Chromosomal SNVs-based phylogenetic tree of representing VREfm ST736 isolates from 2013 (n = 37). Tree was constructed using the neighbor-joining method. ( � ): VREfm E39 was used as reference isolate. The two non-ST736 isolates were used as out of group control. (B) Possible transmission route for a selected group of 6 patients with 8 closely related E. faecium isolates. Solid arrow: transmission with epidemiological evidence [i.e., stayed in the emergency room (PT161 and PT182), on the same bed (PT161 and PT127), in the same medical intensive care unit (PT182 and PT184), or hospitalized during the same time period with overlap for at least 2 weeks (PT127 and PT191, PT161 and PT177)]; dashed arrow: patients had no direct epidemiological links. All transmissions with the exception between patients PT161 and 127 were suggested by SNVs-based minimum spanning tree [46]; Next to each arrow between two samples is the number of SNV differences between the samples. (C) SNVs identified based on whole-genome sequencing with clinical data of the patients. Sequences identical to those from isolate E232 were shown in dot (.); SNVs were listed and shadowed in gray.
https://doi.org/10.1371/journal.pone.0209785.g004 correlation between some previously described mutations and daptomycin resistance. Strikingly, we found that all ST736 VREfm isolates (n = 111) carried the two common mutations (W73C and T120A) in the liaFSR system. Such a high mutation rate in the liaFSR system among ST736 strains has not yet been documented in any E. faecium with other genetic background. Mutations in liaFSR have been associated with high daptomycin MICs [16,20] and failure in treatment of bacteremia caused by daptomycin-susceptible VREfm [52,53]. Our findings and data from other studies may serve as the basis of a potential diagnostic tool to screen and identify isolates carrying mutations in the liaFSR system, and/or isolates with a specific genotype (i.e., ST736) that may predispose to subsequent development of in vivo daptomycin resistance, thus to optimize the use of daptomycin against enterococci in clinical settings. Also, it would be worthy to investigate if relatively poor response to daptomycin therapy in patients infected with ST736 strains, as compared to those infected with non-ST736 VREfm strains.
The estimated emergence of ST736 strains in the mid-2000s at our institution seems coincident with the initial clinical use of daptomycin after FDA's approval in 2003. It is unclear if the co-mutation (W73C and T120A) of the liaFSR in ST736 strains resulted from a serial of evolutional events and/or positive selections by daptomycin, and if this genetic alteration indeed contributed to its rapid expansion and dissemination at our institution with a relatively high usage of daptomycin.
Mutations in either liaFSR or cls alone may not be sufficient in conferring phenotypic resistance to daptomycin in enterococci [21,45]. In our study, only about one-third (38/111, 34.2%) of ST736 strains carrying the two mutations in liaFSR exhibited daptomycin-resistant phenotype (MIC > 4 μg/mL). Also, one VREfm isolate (non-ST736) harboring cls mutations was susceptible to daptomycin. Nevertheless, in an E. faecium clone like ST736 with all strains harboring a predisposing genetic alteration in the liaFSR, subsequent mutation in the cls gene seems to be sufficient to confer daptomycin resistance. This is based on our observation that all 13 VREfm ST736 isolates with mutations in both liaFSR and cls exhibit daptomycin-resistant phenotype. It is noteworthy that the most common cls mutation among ST736 strains is an asparagine to threonine (N13T) substitution, differing from those (N13I and N13S) reported in enterococci with other genetic background [14,21,22,45]. The N13T substitution of cls was previously described as N12T in one California E. faecium isolate with high-level daptomycin resistance (MIC of >256 μg/mL) [54]. Moreover, the thirteen ST736 isolates carrying mutations in both liaFSR and cls displayed a daptomycin MIC 50 of 32 μg/mL (ranging from 8 to >256 μg/mL), which is much higher than that of daptomycin-resistant ST736 isolates without cls mutations (MIC 50 of 8 μg/mL), indicating that co-mutations in both liaFSR and cls might have predisposed to high daptomycin MICs in ST736 strains. On the other hand, 25 of 38 (65.8%) ST736 DNSEfm strains did not carry cls mutations, suggesting diverse mechanisms of resistance to daptomycin in E. faecium [45] and a necessity of further exploring other alternative genomic pathways associated with daptomycin resistance. Notably, VREfm isolate E253 had a daptomycin MIC of >256 μg/mL. In addition to a mutation in the cls (A20D), two new genetic alterations, including mutation in a tyrosine kinase (P90H) and a frameshift in a gene coding the TlyA family rRNA (cytidine-2'-O)-methyltransferase, were also identified on the chromosome of this isolate, as well as mutations in insertion sequence (IS) family transposases on the plasmid. Previously, mutation in histidine kinase has been reported in a daptomycin-resistant E. faecium isolate [19]. Given the important role of protein kinases in regulating bacterial physiology and stress response [55], it would be interesting to determine if bacterial kinases, including tyrosine kinase described in this study, are associated with daptomycin resistance in enterococci. There was no significant genetic alteration in an isogenic pair of VREfm isolates with different daptomycin MICs (4 μg/mL for E232 and >256 μg/mL for E243), highlighting the complexity and other potential mechanisms such as small RNA [48] and/or differential gene expression that may have involved in daptomycin resistance.
The observed high occurrence of DNSEfm at our institution likely resulted from the unique genetic characteristics of ST736 strains that predispose to daptomycin resistance and nosocomial dissemination of DNSEfm. Previously, we reported that 81.7% (17 of 21) DNSEfm from 2009 to 2012 had a prior exposure to daptomycin [27]. In this study, we noticed that the majority (76.9%, 10 of 13) of DNSEfm isolates with cls muations were from patients with prior exposure to daptomycin while hospitalized during 2009 to 2012. The high prevalence of ST736 strains harboring a predisposing mutation in the liaFRS system seems to increase the likelihood of developing resistance by either introducing or selecting mutation in the cls during daptomycin therapy. Only 6 of 20 (30%) patients with DNSEfm in 2013 had prior exposure to daptomycin. Nosocomial transmission might have occurred and attributed to the observed high prevalence of DNSEfm in 2013 in spite of the implementation of hospital-wide enhanced infection control measures, including the use of ultraviolet environmental disinfection in patient rooms [56].
The limitations of this study include: 1) all VREfm isolates examined were from a single institution; 2) the lack of VREfm isolates between 1996 and 2008 for a more precisely evolutional analysis; and 3) we only examined mutations in liaFSR and cls that are known to be associated with daptomycin resistance in E. faecium. Other mutations and additional daptomycin resistance mechanisms are under further investigation.
In conclusion, our study demonstrates an evolutionary change in clonality and antimicrobial susceptibility of E. faecium population over the past 20 years and a recent emergence of VREfm clone ST736 associated with daptomycin nonsusceptibility at our institution. For clone ST736 strains, resistance to daptomycin likely resulted from predisposing genetic alterations in the liaFSR, totaling by mutations in the cls and possibly other genes and pathways, through evolution and/or sequential adaption. The high occurrence of ST736 strains with increased risk of developing daptomycin resistance during therapy and nosocomial dissemination of VREfm [57] might have contributed to the observed high prevalence of DNSEfm at our institution.