Sentinel Surveillance reveals phylogenetic diversity and detection of linear plasmids harboring vanA and optrA among enterococci collected in the United States

ABSTRACT

Enterococcus faecalis and Enterococcus faecium are frequent causes of healthcare-associated infections. Antimicrobial-resistant enterococci pose a serious public health threat, particularly vancomycin-resistant enterococci (VRE), for which treatment options are limited. The Centers for Disease Control and Prevention’s Division of Healthcare Quality Promotion Sentinel Surveillance system conducted surveillance from 2018 to 2019 to evaluate antimicrobial susceptibility profiles and molecular epidemiology of 205 E. faecalis and 180 E. faecium clinical isolates collected from nine geographically diverse sites in the United States. Whole genome sequencing revealed diverse genetic lineages, with no single sequence type accounting for more than 15% of E. faecalis or E. faecium. Phylogenetic analysis distinguished E. faecium from 19 E. lactis (previously known as E. faecium clade B). Resistance to vancomycin was 78.3% among E. faecium, 7.8% among E. faecalis, and did not occur among E. lactis isolates. Resistance to daptomycin and linezolid was rare: E. faecium (5.6%, 0.6%, respectively), E. faecalis (2%, 2%), and E. lactis (5.3%, 0%). All VRE harbored the vanA gene. Three of the seven isolates that were not susceptible to linezolid harbored optrA, one chromosomally located and two on linear plasmids that shared a conserved backbone with other multidrug-resistant conjugative linear plasmids. One of these isolates contained optrA and vanA co-localized on the linear plasmid. By screening all enterococci, 20% of E. faecium were predicted to harbor linear plasmids, whereas none were predicted among E. faecalis or E. lactis. Continued surveillance is needed to assess the future emergence and spread of antimicrobial resistance by linear plasmids and other mechanisms.

IMPORTANCE

This work confirms prior reports of E. faecium showing higher levels of resistance to more antibiotics than E. faecalis and identifies that diverse sequence types are contributing to enterococcal infections in the United States. All VRE harbored the vanA gene. We present the first report of the linezolid resistance gene optrA on linear plasmids in the United States, one of which co-carried a vanA cassette. Additional studies integrating epidemiological, antimicrobial susceptibility, and genomic methods to characterize mechanisms of resistance, including the role of linear plasmids, will be critical to understanding the changing landscape of enterococci in the United States.

INTRODUCTION

Enterococci, including Enterococcus faecalis and Enterococcus faecium, are frequently reported pathogens responsible for healthcare-associated infections (HAIs). The Centers for Disease Control and Prevention (CDC) deemed vancomycin-resistant enterococci (VRE), which account for ~30% of healthcare-associated enterococcal infections, a serious public health threat, particularly because treatment options are limited, especially for vancomycin-resistant E. faecium (1, 2). There were an estimated 54,457 VRE infections in hospitalized patients in 2017 (1, 3). Although VRE infections decreased from 84,800 to 54,500 between 2012 and 2017, the rate of VRE infections increased 16% from 2019 to 2020 during the COVID-19 pandemic (1, 3, 4). Increased vigilance and further action are needed to reverse this increase.

Linezolid and daptomycin are antibiotics used for the treatment of infections caused by VRE. Although rates of resistance to linezolid and daptomycin among US enterococci are low (5), resistance to these antibiotics has emerged (5, 6). Resistance to linezolid is associated with the presence of mutations in the V domain of the 23S rRNA gene or the acquisition of genes encoding resistance to oxazolidinones, including cfr, cfr(B), cfr(D), optrA, and poxtA (7–9). Mutations in the liaFSR operon that encode a 3-component system that regulates cell envelope integrity, and mutations in cls that encodes cardiolipin synthase, are associated with daptomycin resistance (10). However, resistance to daptomycin is multifactorial, and the presence of mutations in liaFSR and cls alone may not always result in elevated MICs and additional mechanisms exist (11–15).

Surveillance data are crucial in our efforts to combat resistance; strengthening surveillance is one of the US Government’s goals in the National Strategy for Combating Antibiotic-Resistant Bacteria (Federal Task Force on Combating Antibiotic-Resistant Bacteria) (16). Antimicrobial susceptibility testing (AST) using reference methods generates helpful and necessary phenotypic surveillance data (17–19); however, molecular surveillance is also needed to detect and characterize emerging resistance mechanisms and identify dominant strains in the healthcare setting. For example, the spread of vancomycin-resistant E. faecium isolates with pstS-null sequence types has been reported internationally, and the ST1478 pstS-null sequence type associated with increased daptomycin nonsusceptibility was reported to have rapidly disseminated across acute care hospitals in Canada between 2013 and 2018 (20–23). However, there is a lack of data available to determine if pstS-null sequence types are circulating in the United States.

The U.S. CDC Division of Healthcare Quality Promotion (DHQP) Sentinel Surveillance system studies the antimicrobial resistance profiles and resistance mechanisms of a convenience sample of bacteria collected annually from hospitalized patients from geographically diverse facilities in the United States. Sentinel Surveillance isolates are selected without regard to antimicrobial susceptibility profile (i.e., both resistant and susceptible isolates). The CDC performed surveillance for E. faecalis and E. faecium using the DHQP Sentinel Surveillance system between 2018 and 2019. Isolates were characterized using both reference AST and whole genome sequencing (WGS) to provide insights into the in vitro activity of various antibiotics used to treat enterococci, describe the molecular epidemiology of both vancomycin-susceptible and -resistant isolates at nine sites in the United States, and describe the molecular mechanisms of resistance observed.

RESULTS

Bacterial isolate collection and identification

During the 2018–2019 collection period, 404 enterococci were included in this study from nine different facilities across the United States (Table 1). Of the 205 isolates submitted as E. faecalis, all were identified as E. faecalis by matrix-assisted laser desorption/ionization-time of flight mass spectrometry (MALDI-TOF MS) (Table S1). Of the 199 isolates submitted as E. faecium, 190/199 isolates were identified as E. faecium by MALDI-TOF MS. There were nine of 199 isolates that were submitted as E. faecium that were identified as E. lactis by MALDI-TOF MS. Whole genome sequencing was performed on all isolates. Average nucleotide identity (ANI) was performed, and four of 199 isolates submitted as E. faecium were identified as E. lactis. Ultimately, phylogenetics was used to distinguish between closely related E. faecium and E. lactis, discussed in greater detail below. A total of 180 isolates were identified as E. faecium clade A, and 19 isolates were identified as E. lactis (previously known as E. faecium clade B) by phylogenetic separation (Fig. S1). Enterococci were isolated from urine (n = 168, 41.6%), blood (n = 130, 32.2%), and other sources (n = 106, 26.2%) including wounds, tissues, and body fluids (Table S1).

TABLE 1.

Number of isolates and demographics^a

	E. faecalis (n = 205)	E. faecium (n = 180)	E. lactis (n = 19)
Source
Blood	63 (30.7%)	60 (33.3%)	7 (36.8%)
Urine	91 (44.4%)	71 (39.4%)	6 (31.6%)
Other	51 (24.9%)	49 (27.2%)	6 (31.6%)
Site
CA	21 (10.2%)	17 (9.4%)	2 (10.5%)
IA	26 (12.7%)	26 (14.4%)	-
MD	21 (10.2%)	16 (8.9%)	2 (10.5%)
MN	14 (6.8%)	12 (6.7%)	2 (10.5%)
NM	20 (9.8%)	18 (10.0%)	1 (5.3%)
NY	24 (11.7%)	24 (13.3%)	4 (21.1%)
NC	27 (13.2%)	16 (8.9%)	2 (10.5%)
PA	27 (13.2%)	18 (10.0%)	2 (10.5%)
WA	25 (12.2%)	33 (18.3%)	4 (21.1%)
Multi-locus sequence types (STs)	ST179 (n = 26); ST16 (n = 20); ST40 (n = 20); ST6 (n = 19); ST21 (n = 7); ST64 (n = 7); ST109 (n = 6); ST103 (n = 5); ST778 (n = 5); others (n = 90)	ST80 (n = 26); ST17 (n = 23); ST18 (n = 22); ST412 (n = 21); ST117 (n = 18); ST736 (n = 9); ST78 (n = 6); ST1478 (n = 5); others (n = 50)	ST240 (n = 4); ST94 (n = 3); ST800 (n = 2); others (n = 10)
Patient age (median)	64	61	57

^a CA, California; IA, Iowa; MD, Maryland; MN, Minnesota; NM, New Mexico; NY, New York; NC, North Carolina; PA, Pennsylvania; WA, Washington. - indicates no E. lactis isolates were received from that site.

Antimicrobial susceptibility testing

A summary of the susceptibility profiles of E. faecalis, E. faecium, and E. lactis isolates is presented (Table 2). E. faecium MIC susceptibility test interpretive criteria (breakpoints) were used for E. lactis isolates for the purposes of categorical interpretation due to the genomic similarity between E. faecium and E. lactis (24). All E. faecalis and E. lactis were susceptible to ampicillin and penicillin, but most E. faecium isolates were resistant to both (n = 167, 92.8%). Resistance to vancomycin was uncommon among E. faecalis isolates (n = 16, 7.8%) but common among E. faecium (n = 141, 78.3%). No E. lactis isolates displayed vancomycin resistance. Resistance to daptomycin was rare, with 2.0% (4/205) of E. faecalis, 5.6% (10/180) of E. faecium, and 5.3% (1/19) of E. lactis isolates displaying resistance. Resistance to linezolid was also rare with 2.0% (4/205) of E. faecalis, 0.6% (1/180) of E. faecium, and no E. lactis isolates displaying resistance. Three vancomycin-resistant E. faecalis displayed resistance to linezolid and intermediate susceptibility to daptomycin. Two vancomycin-resistant E. faecium displayed resistance or intermediate susceptibility to linezolid; both were categorized as susceptible-dose dependent (SDD) to daptomycin. It should be noted that the Clinical and Laboratory Standards Institute (CLSI)-adopted SDD breakpoint for E. faecium is based on a dosage of 8–12 mg/kg, which is higher than the Food and Drug Administration (FDA)-approved dose (25).

TABLE 2.

Antimicrobial susceptibility testing results for E. faecalis, E. faecium, and E. lactis, including vancomycin-resistant E. faecalis and E. faecium, 2018–2019 DHQP Sentinel Surveillance system isolates^a

	% Susceptible	% Intermediate	% Resistant	MIC range (µg/mL)	MIC50 (µg/mL)	MIC90 (µg/mL)
E. faecalis (n = 205)
Ampicillin	100.0	N/A	0.0	≤0.25 to 2	1	1
Penicillin	100.0	N/A	0.0	≤0.25 to 8	2	4
Vancomycin	92.2	0.0	7.8	0.5 to > 64	1	2
Teicoplanin	92.2	0.0	7.8	≤0.12 to > 64	0.5	1
Daptomycin	80.0	18.0	2.0	≤0.25 to 16	2	4
Linezolid	98.0	0.0	2.0	≤0.5 to 16	2	2
Tedizolidb	98.0	N/A	2.0	≤0.12 to 2	0.5	0.5
Quinupristin/dalfopristinc	N/A	N/A	N/A	0.5 to > 8	8	8
Levofloxacin	77.1	0.0	22.9	0.5 to > 8	1	>8
Doxycycline	45.9	47.3	6.8	≤2 to 32	8	8
Tigecyclined	100.0	N/A	N/A	0.06 to 0.25	0.12	0.12
Eravacyclineb,e	95.1	N/A	4.9	≤0.016 to 0.12	0.06	0.06
Chloramphenicol	89.3	1.5	9.3	≤2 to > 32	4	16
Gentamicinf	77.6	N/A	22.4	-	-	-
Streptomycinf	83.4	N/A	16.6	-	-	-
VR E. faecalis (n = 16)
Ampicillin	100.0	N/A	0.0	0.5 to 2	1	1
Penicillin	100.0	N/A	0.0	2 to 8	4	4
Vancomycin	0.0	0.0	100.0	>64	>64	>64
Teicoplanin	0.0	0.0	100.0	32 to > 64	>64	>64
Daptomycin	75.0	18.8	6.3	0.5 to 8	1	4
Linezolid	81.3	0.0	18.8	≤0.5 to 16	2	16
Tedizolidb	81.3	N/A	18.8	≤0.12 to 2	0.25	2
Quinupristin/dalfopristinc	N/A	N/A	N/A	0.5 to > 8	8	8
Levofloxacin	0.0	0.0	100.0	8 to > 8	>8	>8
Doxycycline	75.0	25.0	0.0	≤2 to 8	4	8
Tigecyclined	N/A	N/A	N/A	0.06 to 0.25	0.12	0.12
Eravacyclineb,e	93.8	N/A	6.3	≤0.016 to 0.12	0.06	0.06
Chloramphenicol	62.5	18.8	18.8	8 to > 32	8	>32
Gentamicinf	18.8	N/A	81.3	-	-	-
Streptomycinf	68.8	N/A	31.3	-	-	-
E. faecium (n = 180)
Ampicillin	7.2	N/A	92.8	≤0.25 to > 32	>32	>32
Penicillin	6.7	N/A	93.3	≤0.25 to > 32	>32	>32
Vancomycin	21.1	0.6	78.3	0.5 to > 64	>64	>64
Teicoplanin	21.1	0.6	78.3	0.5 to > 64	>64	>64
Daptomycing	94.4	N/A	5.6	≤0.25 to 16	4	4
Linezolid	98.3	1.1	0.6	≤0.5 to 8	2	2
Tedizolid	N/A	N/A	N/A	≤0.12 to 1	0.5	0.5
Quinupristin/dalfopristinh	N/A	N/A	N/A	0.25 to 8	1	2
Levofloxacin	5.0	2.8	92.2	1 to > 8	>8	>8
Doxycycline	50.6	13.9	35.6	≤2 to 32	4	16
Tigecyclined	N/A	N/A	N/A	≤0.03 to 0.5	0.06	0.12
Eravacyclineb,e	88.3	N/A	11.7	≤0.016 to > 0.25	0.06	0.12
Chloramphenicol	87.2	11.1	1.7	4 to > 32	8	16
Gentamicinf	88.9	N/A	11.1	-	-	-
Streptomycinf	60.6	N/A	39.4	-	-	-
VR E. faecium (n = 141)
Ampicillin	0.0	N/A	100.0	32 to > 32	>32	>32
Penicillin	0.0	N/A	100.0	>32	>32	>32
Vancomycin	0.0	0.0	100.0	>64	>64	>64
Teicoplanin	0.0	0.7	99.3	16 to > 64	>64	>64
Daptomycing	95.7	N/A	4.3	≤0.25 to 16	4	4
Linezolid	98.6	0.7	0.7	≤0.5 to 8	2	2
Tedizolid	N/A	N/A	N/A	≤0.12 to 1	0.5	0.5
Quinupristin/dalfopristinh	89.4	3.5	7.1	0.25 to 8	1	2
Levofloxacin	0.0	0.0	100.0	>8	>8	>8
Doxycycline	52.5	13.5	34.0	≤2 to 32	4	16
Tigecyclined	N/A	N/A	N/A	≤0.03 to 0.5	0.06	0.12
Eravacyclineb, e	88.7	N/A	11.3	≤0.016 to > 0.25	0.06	0.12
Chloramphenicol	87.2	11.3	1.4	4 to > 32	8	16
Gentamicinf	90.8	N/A	9.2	-	-	-
Streptomycinf	58.9	N/A	41.1	-	-	-
E. lactis (n = 19)
Ampicillin	100.0	N/A	0.0	≤0.25 to 2	1	2
Penicillin	100.0	N/A	0.0	≤0.25 to 8	4	8
Vancomycin	100.0	0.0	0.0	0.5 to 1	0.5	1
Teicoplanin	100.0	0.0	0.0	0.5 to 2	1	1
Daptomycing	94.7	N/A	5.3	0.5 to 8	4	4
Linezolid	100.0	0.0	0.0	1 to 2	2	2
Tedizolid	N/A	N/A	N/A	0.25 to 0.5	0.5	0.5
Quinupristin/dalfopristinh	N/A	N/A	N/A	1 to 4	2	2
Levofloxacin	89.5	5.3	5.3	0.5 to > 8	1	4
Doxycycline	94.7	0.0	5.3	≤2 to 16	≤2	≤2
Tigecyclined	N/A	N/A	N/A	0.06	0.06	0.06
Eravacyclineb, e	100.0	N/A	0.0	0.03 to 0.06	0.06	0.06
Chloramphenicol	94.7	5.3	0.0	4 to 16	4	8
Gentamicinf	100.0	N/A	0.0	-	-	-
Streptomycinf	94.7	N/A	5.3	-	-	-

^a N/A, isolates were tested, but no breakpoints were available; VR, vancomycin resistant.

^b For tedizolid and eravacycline, % nonsusceptible is reported as there is no resistant category.

^c E. faecalis is intrinsically resistant to quinupristin/dalfopristin.

^d U.S. FDA MIC susceptibility test interpretive criteria (breakpoint) for tigecycline (susceptible ≤0.25 μg/mL) is only available for vancomycin-susceptible E. faecalis. The FDA breakpoint for tigecycline was only applied to vancomycin-susceptible E. faecalis isolates (n=189). CLSI breakpoints are not established.

^e U.S. FDA MIC susceptibility test interpretive criteria (breakpoint) for eravacycline (susceptible ≤0.06 μg/mL) was used as CLSI breakpoints are not established.

^f E. faecalis and E. faecium isolates are intrinsically resistant to aminoglycosides. High-level gentamicin resistance and high-level streptomycin resistance were tested using 500 µg/mL and 1000 µg/mL of the antibiotic, respectively. Susceptibility implies synergy between the aminoglycoside and the cell wall active agent and resistance implies no synergy between the aminoglycoside and the cell wall active agent; therefore, MIC range, MIC₅₀, and MIC₉₀ are indicated as (-).

^g For daptomycin and E. faecium/E.lactis, there is no susceptible breakpoint so susceptible-dose dependent (SDD) is reported. Given the close genetic relationship between E. faecium and E. lactis, E. faecium daptomycin breakpoints were applied to the E. lactis isolates.

^h For quinupristin/dalfopristin, CLSI guidance is to report only on vancomycin-resistant E. faecium.

Whole genome sequencing, phylogenetics and MLST distribution

All isolates underwent WGS, and to better understand the diversity within each species and the relationship between genetic relatedness, resistance, and demographics, we performed multi-locus sequence typing (MLST) and core genome phylogenetic reconstruction and compared with AST. A summary of sequence types (STs) is provided in Table 1. Overall, STs for E. faecalis, E. faecium, and E. lactis were highly diverse: for E. faecalis, the Simpson’s Diversity Index was 0.95; for E. faecium, it was 0.92; and for E. lactis, it was 0.89.

E. faecalis did not cluster phylogenetically within collection sites, and 73 STs were represented (Fig. 1). Vancomycin resistance was identified in multiple sites within a few E. faecalis STs: ST109 (n = 5), ST778 (n = 4), ST6 (n = 4), and ST1011 (n = 3). Daptomycin resistance was identified infrequently among ST16 (n = 2), ST109 (n = 1), and ST40 (n = 1), whereas daptomycin-intermediate E. faecalis were spread across many different STs. Linezolid resistance was identified in ST6 (n = 3) and ST480 (n = 1) isolates (Table S1).

Fig 1.

E. faecalis maximum likelihood phylogeny based on core gene alignments generated by Roary and RaxML. Annotation rings (from inner to outer) show isolate ID, state of origin, multi-locus sequence type (ST), minimum inhibitory concentration (MIC) interpretations for vancomycin, daptomycin, and linezolid. ST obtained using E. faecalis scheme from PubMLST, MIC interpretations represent susceptible in white; intermediate in gray; and resistant in black. Black circles on branches represent 100% support for the branch out of 100 bootstraps, and the tree is rooted at the midpoint. Abbreviations: LZD, linezolid; DAP, daptomycin; VAN, vancomycin; ST; multi-locus sequence type, CA, California; IA, Iowa; MD, Maryland; MN, Minnesota; NM, New Mexico; NY, New York; NC, North Carolina; PA, Pennsylvania; and WA, Washington.

Performing a phylogenetic analysis of isolates submitted as E. faecium, we observed two clades, A (Fig. 2) with n = 180 isolates with 44 STs and B (Fig. 3) with n = 19 isolates with 13 STs, identified by previously established reference isolates Aus0004 (26) and COM15 (27), respectively (28). Vancomycin resistance was identified throughout the E. faecium tree across 24 STs with the most in the most prevalent STs, for example, ST17 (n = 22), ST18 (n = 21), and ST80 (n = 21) (Fig. 2). Vancomycin resistance was not present in any E. lactis isolate (Fig. 3). Daptomycin resistance was also distributed across several STs throughout E. faecium (Fig. 2), with the following STs represented ST80 (n = 4), ST18 (n = 1), ST203 (n = 1), ST736 (n = 1), ST789 (n = 1), ST1471 (n = 1), and ST2181 (n = 1). A single E. lactis isolate, ST2186, displayed daptomycin resistance. Linezolid resistance in E. faecium was found only in ST80 (n = 1), and linezolid-intermediate isolates belonged to ST78 (n = 1) and ST117 (n = 1). One linezolid-resistant (ST80) isolate and one linezolid-intermediate (ST78) isolate were also resistant to vancomycin; both were categorized as SDD to daptomycin.

Fig 2.

E. faecium maximum likelihood phylogeny based on core gene alignments generated by Roary and RaxML. Annotation rings (from inner to outer) show isolate ID, state of origin, multi-locus sequence type (ST), minimum inhibitory concentration (MIC) interpretations for vancomycin, daptomycin, and linezolid. ST obtained using E. faecium scheme from PubMLST, MIC interpretations following E. faecium guidance represents susceptible – white; susceptible dose-dependent – gray; and resistant – black. Black circles on branches represent 100% support for the branch out of 100 bootstraps. Abbreviations: LZD, linezolid; DAP, daptomycin; VAN, vancomycin; ST, multi-locus sequence type; CA, California; IA, Iowa; MD, Maryland; MN, Minnesota; NM, New Mexico; NY, New York; NC, North Carolina; PA, Pennsylvania; and WA, Washington.

Fig 3.

E. lactis maximum likelihood phylogeny based on core gene alignments generated by Roary and RaxML. Annotation (from left to right) shows isolate ID, state of origin, multi-locus sequence type (ST), MALDI-TOF MS identified species, ANI identified species, minimum inhibitory concentration (MIC) interpretations for vancomycin, daptomycin, and linezolid. ST obtained using E. faecium scheme from PubMLST, MIC interpretations following E. faecium guidance represents susceptible in white; susceptible dose-dependent in gray; and resistant in black. Black circles on branches represent 100% support for the branch out of 100 bootstraps. Abbreviations: LZD, linezolid; DAP, daptomycin; VAN, vancomycin; ST, multi-locus sequence type; CA, California; IA, Iowa; MD, Maryland; MN, Minnesota; NM, New Mexico; NY, New York; NC, North Carolina; PA, Pennsylvania; and WA, Washington.

We assessed the distribution of the rapidly spreading and global pstS-null STs (20) (strains that are nontypable due to the absence of the pstS locus) and found that 6/141 (4.3%) vancomycin-resistant E. faecium contained the pstS-null ST, including one ST1421 (WA; SDD to daptomycin MIC 2 µg/mL) and 5 ST1478 (PA, IA, and MN; all daptomycin SDD with MICs of 2 or 4 µg/mL). When compared phylogenetically to global isolates using a core gene analysis, the ST1421 isolates clustered closely with isolates from Australia (29). However, the 5 US ST1478 were more similar in general to each other [40–115 core gene single nucleotide polymorphisms (SNPs)] than the rest of the Canadian isolates (23) (150–1897 core SNPs) except for one Canadian isolate (ID:01A18435), which ranged between 74 and 108 core SNPs to the 5 US ST1478 (Fig. S2).

Antimicrobial resistance gene and mutation analyses

Sixteen (7.8%) of the 205 E. faecalis were resistant to vancomycin; all 16 harbored the vanA resistance gene. The remaining 189 vancomycin-susceptible isolates did not contain known mechanisms associated with vancomycin resistance. Among the E. faecium, 141 (78.3%) of 180 were resistant to vancomycin and all harbored vanA. One vancomycin-intermediate (MIC 8 µg/mL) E. faecium harbored vanA. Thirty-eight (21.1%) of 180 E. faecium isolates were vancomycin susceptible and did not harbor a known vancomycin resistance gene.

Of the four E. faecalis isolates showing resistance to daptomycin (MIC ≥8 µg/mL), only one isolate (sample ID: 2006286, MIC 16 µg/mL) contained the LiaF amino acid insertion insI177 previously associated with resistance (Table S1) (30). No other genes containing mutations previously associated with daptomycin resistance (e.g., cls, gdpD, gshF, liaFSR, and yybT) (15, 30), were identified among the remaining 201 E. faecalis isolates. Of the 10 E. faecium isolates showing resistance to daptomycin (MIC ≥8 µg/mL), five contained mutations previously associated with resistance, including isolates 2006249 (Cls R218Q), 2005969 (Cls A20P, LiaR W73C, LiaS T120A), and 2001563, 2008334, 2001367 (LiaR W73C, LiaS T120A), and five did not have any genes containing mutations previously associated with daptomycin resistance [e.g., cfa, cls, divIVA, dltABCD, gdpD, gshF, liaFSR, mprF, rpoB, and yycFG (5, 10–12, 15)]. One E. lactis isolate showed resistance to daptomycin but did not harbor genes containing mutations previously associated with daptomycin resistance. Among the 170 daptomycin SDD E. faecium isolates (MIC ≤4 µg/mL), 32 (17.8%) isolates (MIC range: 2–4 µg/mL) harbored mutations previously associated with resistance, including 2005978 with Cls R218Q, LiaR W73C, LiaS T120A, 2000749 with LiaR W73C, and 30 samples with LiaR W73C, LiaS T120A (12, 15). The remaining 138 SDD isolates did not contain mutations previously associated with daptomycin resistance (5, 10–12, 15).

Of the four linezolid-resistant E. faecalis, three ST6 isolates contained the 23S rRNA G2576T mutation and one ST480 isolate contained optrA. One E. faecium isolate was linezolid-resistant and two were intermediate (MIC 4 µg/mL). One of the linezolid-intermediate isolates neither harbored a 23S rRNA mutation nor did it carry optrA, cfr, cfr(B), cfr(D), or poxtA. The remaining linezolid-resistant and linezolid-intermediate isolates contained optrA. Of the 177 linezolid-susceptible E. faecium isolates, three carried cfr(B).

optrA locations and linear plasmids

To analyze the genomic context of the three optrA-positive enterococci, the isolates were sequenced using both short- and long-read sequencing to create annotated hybrid assemblies. In E. faecalis isolate 2001206, optrA was chromosomally encoded on a Tn6674-like transposon. The 60 kb region in 2001206 surrounding optrA was similar to several publicly available genomes from the National Center for Biotechnology Information (NCBI) with high identity (>99.2%) and perfect coverage including EFS17 (Accession CP085289), JF31-223 (Accession CP102065), JARB-HU0796 (Accession AP027136), and AT40a (Accession CP097069). The Tn6674-like region in 2001206 had 99.8% identity in the overlapping regions of Tn6674 [Accession MK737778 (31)] with an ~900 bp internal deletion that did not appear to interrupt any coding genes. OptrA is located on this transposon in addition to resistance genes spc, erm(A), and fexA. In the two E. faecium isolates, 2001564 and 2008300, we identified optrA on plasmids that did not circularize with sizes of 114,744 bp and 106,518 bp, respectively.

S1 nuclease pulsed-field gel electrophoresis (PFGE) was used to assess plasmid topology. S1 nuclease, which converts supercoiled plasmid DNA into linear molecules, was used to compare digested total DNA to undigested total DNA for all three optrA-positive isolates. Linear plasmids were not identified in 2001206; however, linear fragments of sizes ~114 kb and ~106 kb were observed in S1 nuclease-digested and undigested samples for isolates 2001564 and 2008300, respectively, which is characteristic of a plasmid with linear topology (32) (Fig. 4). The linear plasmids correspond approximately in size to the assembled contigs carrying optrA as identified with WGS analysis. Following the nomenclature pELF, enterococcal linear form plasmid (33), the plasmids were designated pELF_2001564 and pELF_2008300.

Fig 4.

Confirmation of the presence of linear plasmids by S1 nuclease pulsed-field gel electrophoresis (PFGE) of optrA-positive enterococci. Lane 1, PFG MidRange Marker; lane 2, 2001206 (undigested); lane 3, 2001206 (digested); lane 4, 2001564 (undigested, ~114 kb linear plasmid present); lane 5, 2001564 (digested, ~114 kb linear plasmid present); lane 6, 2008300 (undigested, ~106 kb plasmid present); and lane 7, 2008300 (digested, ~106 kb plasmid present).

We compared the two plasmids with one another and other known enterococcal linear form plasmids (Fig. 5). The linear plasmids pELF_2008300 (106,518 bp) and pELF_2001564 (114,744 bp) share 96,190 bp at 99.99% identity. The unique 10,326 bp region in pELF_2008300 is flanked by IS1216 elements and carries several recombinases, transposases, and protein domains related to cell wall degradation. The main unique 17,708 bp region in pELF_2001564 carried a transposon with the vanA operon (Fig. 5). Upon querying NCBI for other highly similar sequences to pELF_2008300 and pELF_2001564, both scored the highest blastn identity, 100% for both, and highest coverage, 89% and 98%, respectively, with the linear plasmid pBA17124_P1, which carries optrA and was found in a clinical E. faecium isolated in 2019 from a patient in India (Fig. 5) (34). We then compared the plasmids in this study with the first described enterococcal linear form plasmids, pELF1, and pELF2 (33, 35) (Table S2). All four plasmids shared at least 75,950 bp, >97% identity, with one another including genes associated with DNA partitioning, transfer, and replication ftsK, soj, and repB, respectively. Although pELF_2008300 and pELF_2001564 uniquely carry optrA in a region that is inserted into radC, pELF2 has a copy of radC interrupted by a different antimicrobial resistance gene region. All three carry the antimicrobial resistance gene erm(A) in this region as well, whereas pELF1 does not carry radC or erm(A).

Fig 5.

Comparison of this study’s linear plasmids with optrA, the highly similar plasmid pBA17124_P1(34), and the first documented Enterococcal Linear Form plasmids, pELF1 and pELF2 (33, 35). Accession and plasmid length are shown under the plasmid name. Regions of similarity between sequences are connected by vertical blocks when there is a minimum of 500 bp and 80% BLASTN identity. Coding sequences are shown with arrows, and gene functions are color-coded as follows: vancomycin-related resistance genes are in orange, other resistance genes in red, radC in purple, DNA replication in brown, plasmid partitioning machinery in blue, parM in light blue, and IS elements or transposases in yellow.

To predict the distribution of pELF1-like plasmids in this data set, we queried short-read assemblies with in silico PCR using primers developed previously to identify the conserved terminal end sequences of pELF1-like plasmids (36). Among E. faecium isolates, 36/180 (20.0%) were predicted to harbor pELF1-like plasmid products, whereas no E. faecalis or E. lactis isolates contained predicted products. The isolates predicted to harbor pELF1-like plasmid products had 15 different STs within E. faecium and were found across seven different sites. Of all VRE isolates, 31/157 (19.7%) were predicted to harbor pELF1-like plasmid products, although this may be an underestimate due to primer-template variation and the potential for short-read assemblies to miss the end regions of these plasmids.

DISCUSSION

We characterized 404 enterococci from CDC’s 2018–2019 DHQP Sentinel Surveillance collection. Although all study isolates were submitted as E. faecalis or E. faecium as identified by clinical laboratories, the use of phylogenetic analysis revealed that 9.5% of submitted E. faecium isolates were E. lactis (formerly clade B E. faecium). MALDI-TOF MS and ANI analysis were not sufficient to differentiate E. faecium from E. lactis isolates. To date, custom in-house databases may be necessary to distinguish E. faecium from E. lactis isolates using MALDI-TOF MS (37). Recently, PCR assays have been designed to differentiate between E. faecium and E. lactis isolates (38). It is unknown how important differentiating these two species is for patient care.

Overall, resistance to vancomycin was more common among E. faecium isolates (78.3% VRE) than E. faecalis (7.8% VRE) or E. lactis (0% VRE), as previously demonstrated by others (18). The percent of E. faecium isolates showing resistance to vancomycin in this study was consistent with recent surveillance data examining AST profiles from bloodstream infections in the United States (2010–2019) (39) that found 72.5% of E. faecium were vancomycin-resistant. For all E. faecium isolates in our study, vanA was identified as the mechanism of vancomycin resistance, whereas Carvalhaes et al. identified vanA in 96.3% and vanB in 3.7% among vancomycin-resistant E. faecium isolates collected in the United States (39). The percent of vancomycin-resistant E. faecalis isolates was 7.8%, which is higher than the study by Carvalhaes et al. (39) that found that 3.6% of E. faecalis were resistant. All E. faecalis isolates in this study harbored vanA, compared with Carvalhaes et al. in which 85.5% of isolates contained vanA and 14.5% vanB (39). The elevated percent of vancomycin-resistant E. faecalis and lack of vanB seen in our surveillance may be explained by the variation in isolate source, as the Carvalhaes et al. study (39) focused on only isolates from bloodstream infections (39) and Sentinel Surveillance included isolates from bloodstream infections and from urine and other sources. In addition, the observation that vanA was the sole contributor to vancomycin resistance is not unexpected, as vanA has historically been reported as the predominant genotype in Canada, Europe, and the United States (40, 41).

Resistance to linezolid was rare in the 2018–2019 Sentinel Surveillance set, with 2% of E. faecalis isolates and 0.6% of E. faecium isolates showing resistance. This may represent a small increase in linezolid resistance among E. faecalis as previous DHQP Sentinel Surveillance from 2013 to 2016 showed 0.3% of E. faecalis and 0.6% of E. faecium isolates were linezolid-resistant, although the 2013–2016 surveillance period differed by two sites (5). Other US surveillance systems have reported E. faecalis and E. faecium linezolid resistance rates of 0.1% and 0.5%, respectively, among bloodstream infections in the United States from 2010 to 2019 (39). The frequency of daptomycin-resistant E. faecalis was 2% and E. faecium was 5.6% among the 2018–2019 Sentinel Surveillance set, compared with 0.5% and 5%, respectively, in the 2013–2016 DHQP Sentinel Surveillance set. This may represent a small increase in daptomycin resistance among E. faecalis. Other US surveillance systems have reported E. faecalis and E. faecium daptomycin resistance rates of 0% and 0.8%, respectively, among bloodstream infections in the United States from 2010 to 2019 (39). Continued surveillance is necessary to monitor resistance development to linezolid and daptomycin.

Overall, the diversity of genetic lineages in E. faecalis, E. faecium, and E. lactis was high, similar to previous surveillance of bloodstream infections caused by Enterococcus spp. (28). Vancomycin-resistant isolates had diverse genetic lineages even at single collection sites. This may indicate that we need to better understand the contribution of horizontal gene transfer to the spread of antimicrobial resistance among Enterococcus spp., as seen in several hospital-specific cases (35, 42).

Recently, strains of E. faecium that are nontypable due to the absence of the pstS MLST locus, referred to as pstS-null sequence types (e.g., ST1421, ST1424, and ST1478), have been reported to be rapidly spreading in Australia, the United Kingdom, Denmark, and Canada (20–23, 29). In addition, an increased proportion of daptomycin nonsusceptibility has been reported for ST1478 isolates (20). These pstS-null isolates were not a predominant ST among this set; however, they were identified in geographically diverse locations, and continued surveillance is necessary to monitor their distribution in the United States.

The mechanisms contributing to daptomycin resistance in enterococci have not been fully elucidated and resistance is associated with multiple pathways (11–13). Mutations in the LiaFSR system and the phospholipid metabolism enzyme, Cls, previously associated with daptomycin resistance (11, 12, 15, 30), were identified for only a subset of daptomycin-resistant E. faecalis and E. faecium isolates. These mutations were also detected among 18.8% of daptomycin-SDD E. faecium isolates, indicating the contribution of a variety of different resistance mechanisms.

Among the three E. faecium isolates that were not susceptible to linezolid, two isolates harbored optrA that were encoded on linear plasmids. This is the first description of the optrA gene on a linear plasmid in the United States. The optrA-encoding linear plasmids (designated pELF_2008300 and pELF_2001564) shared 99.9% sequence identity and were identified in two different E. faecium STs (ST78 and ST80), both of which are considered hospital-adapted sequence types associated with nosocomial infections worldwide (43). Both plasmids further shared 100% sequence identity and 89%–98% sequence coverage with a plasmid found in a clinical E. faecium isolated in 2019 from India (34), demonstrating the potential global spread of optrA on a linear plasmid (44).

Notably, one of the optrA-carrying linear plasmids also harbored a vanA cassette, leading to vancomycin resistance. The convergence of these resistance genes on the same genetic element/linear plasmid is concerning given that plasmids belonging to the pELF family have been shown to spread between Enterococcus spp. during healthcare-associated outbreaks (35, 42). In addition, previous laboratory work has demonstrated self-transmissibility and a low fitness cost of these plasmids, which could further facilitate the spread of these elements along with the associated resistance genes (33, 36, 45). Although studies indicate that linear enterococcal plasmids harboring vancomycin resistance genes may have only recently emerged, reports of linear plasmids harboring vanA, alone or in combination with other AR genes, have been reported elsewhere (33, 34, 36, 46).

Although bacterial linear plasmids were first described in the late 70s (47), the first report of a linear enterococcal plasmid was not until 2019 (33). Since then, linear plasmids in enterococci have been reported from multiple countries (33, 34, 36, 46); surveillance of VRE in Japan identified 3.4% of VRE that harbored pELF1-like plasmids, detected exclusively in E. faecium (36). In the present study, 19.7% of VRE were predicted to contain pELF1-like plasmid sequences similarly detected exclusively in E. faecium, indicating a relatively broad dissemination of pELF1-like plasmids among US E. faecium clinical isolates. Increased surveillance is necessary to better understand the distribution of these linear plasmids, their antimicrobial resistance gene content, and their clinical relevance/how they affect patient outcomes.

This work has limitations. First, the study design is based on a convenience sample rather than a random sample with limited metadata collected and no clinical outcome data. Thus, we cannot account for unknown biases that may have occurred. In addition, isolates were collected from nine sites and may not be representative of the true diversity at that site or in the United States. However, this study provides both reference AST and WGS data for more than 400 isolates collected from diverse sources.

E. faecium isolates showed higher levels of resistance to more antibiotics than E. faecalis, and resistance to linezolid and daptomycin was rare among the DHQP’s 2018–2019 Sentinel Surveillance collection. Whole genome sequencing revealed diverse STs among E. faecalis and E. faecium and the presence of optrA, associated with acquired linezolid resistance, encoded on a chromosomal transposon and on linear plasmids containing multiple antimicrobial resistance genes, including the vanA gene mediating vancomycin resistance. These data provide a snapshot of the AST profiles and molecular epidemiology of enterococci, important healthcare-associated pathogens. In the US, ongoing surveillance is necessary to detect possible future changes in AST patterns and/or monitor for increases in the mechanisms and emergence of new mechanisms of resistance.

MATERIALS AND METHODS

Bacterial isolates and identification

DHQP Sentinel Surveillance isolates are clinical isolates representing a convenience sample of bacteria collected annually from hospitalized patients from nine facilities in geographically distinct locations across the United States, including sites in California, Iowa, Maryland, Minnesota, New Mexico, New York, North Carolina, Pennsylvania, and Washington. Isolates are collected systematically at each site by choosing the first consecutive isolates of each organism. Isolates were cultured on BD BBL Trypticase soy agar II with 5% sheep blood (SBA) (Thermo Fisher Scientific, Waltham, MA, United States). Species identification was confirmed using MALDI-TOF mass spectrometry using a MALDI Biotyper, version 3.1 (Bruker Daltronics, Bremen, Germany).

Antimicrobial susceptibility testing

Broth microdilution (BMD) AST testing was performed on all isolates following the CLSI guidelines (48). BMD panels were prepared in-house consisting of 14 antibiotics at varying 2-fold serial dilutions in sterile 96-well microtiter plates (Caplugs, Rancho Domingues, CA, United States) and stored at −70°C until use. Each well contained 100 µL cation-adjusted Mueller-Hinton Broth (BD Difco, Sparks, MD, United States), and for daptomycin, there was a supplement of 50 µg/mL calcium chloride. See Table 2 for a complete list of antibiotics tested. Enterococcus faecalis ATCC 29212 and Staphylococcus aureus ATCC 29213 were used as quality control strains. Samples were streaked to purity on SBA. Colonies were resuspended in 5 mL REMEL 0.85% sterile saline (Thermo Fisher Scientific, Waltham, MA, United States) to a turbidity equivalent to a 0.5 MacFarland standard. Panels were inoculated with a 95-pin sterile inoculator (10 µL pickup; Caplugs, Buffalo, NY, United States), incubated at 35°C, then read at 16–20 hours for all antibiotics; vancomycin and gentamicin were also read at 24 hours; and streptomycin was also read at 24 and 48 hours. Results were interpreted according to CLSI guidelines (49).

Whole genome sequencing and analysis

All isolates underwent whole genome short-read sequencing (Illumina MiSeq or NovaSeq, San Diego, CA). Genomic DNA was extracted from colonies cultured overnight on SBA using the Promega Maxwell 16 Cell Low Elution Volume DNA Purification Kit and Maxwell 16 MDx Instrument (Madison, WI, United States). Sequencing using the MiSeq was performed as previously described (5). For NovaSeq WGS, genomic DNA was sheared to a mean size of 600 bp using a Covaris LE220 focused ultrasonicator (Covaris Inc., Woburn, MA). DNA fragments were AMPure XP bead cleaned (Beckman Coulter Inc., Indianapolis, IN) and used to prepare dual-indexed sequencing libraries using NEBNext Ultra library prep reagents (New England Biolabs Inc., Ipswich, MA) and barcoding indices synthesized in the CDC Biotechnology Core Facility. Libraries were analyzed for size and concentration, normalized, pooled, and denatured for loading onto flow cells for cluster generation. Sequencing was performed using the MiSeq Reagent Kit v2 (500 cycles) and Illumina MiSeq System (San Diego, CA) or on a NovaSeq using the NovaSeq 6000 Reagent Kit (500 cycles). On completion, sequence reads were filtered for read quality, base called, and demultiplexed using bcl2fastq (v2.19). Short-read sequences (MiSeq and NovaSeq) were analyzed using the QuAISAR-H pipeline; a description of publicly available tools and versions can be found on the QuAISAR-H repository (https://github.com/DHQP/QuAISAR_singularity). Specifically, raw short-read sequences were processed with BBDuk v38.90 (https://sourceforge.net/projects/bbmap/) and fastp v.0.20.1 (50) to remove adapters and PhiX then quality trim sequences. ANI was computed using pyANI v 0.2.10 (51) against NCBI’s RefSeq database. Trimmed sequences were used for high-quality single-nucleotide variant (hqSNV) analyses and hybrid genome assembly.

Antimicrobial resistance genes and point mutations were identified using GAMMA (52) against a manually curated non-redundant database comprising antimicrobial resistance genes from ResFinder (53), ARG-ANNOT (54), and NCBI’s AMRFinderPlus (55) (all accessed 06/06/2024); minimum of 98% sequence identity and 90% sequence coverage was used. Mutations associated with linezolid resistance within the 23S rRNA genes from E. faecium and E. faecalis were determined by mapping short-read sequences to a linezolid-susceptible reference using Bowtie2 (56), processed using SamTools (57), then hqSNV-called using VarScan v2.4.1 (58) using mpileup2cns with the parameters: --min-coverage 8 --min-reads2 2 --min-var-freq 0.01. Potential pELF1-like linear plasmids were identified in the short-read assembled genomes using ipcress with default settings for in silico PCR with previously published primers targeting the ends of pELF1(36).Table S1

Phylogenetic and MLST analyses

Short-read assembled genomes were annotated with Prokka v1.14.3 (59), and a core genome alignment was created using Roary v3.12.0 and MAFFT v7.407 with default parameters. A maximum likelihood tree was created on the core genome alignment using RAxML v8.2.12 with model GTRCAT and 100 bootstraps. Aus0004 and 1–231-501, were used as references to identify E. faecium, COM15, was used as the reference for E. lactis and ATCC19433 as a reference for E. faecalis. The diversity of E. faecalis, E. faecium, and E. lactis MLSTs was assessed using the Simpson’s Diversity Index with R’s vegan package. Global sequence data for the pstS-null analysis were obtained from NCBI’s SRA and GenBank databases pulled from the following BioProjects: PRJEB13012, PRJNA816868, and PRJEB25215. Assemblies were pulled from NCBI unless they were not available, in which case sequence data were pulled from NCBI and assembled as described before. Then, the MLST was identified to separate based on ST and core-gene based maximum likelihood phylogenetic trees for ST1421 and ST1478 were generated as described before. Core SNPs were identified using snp-dists (https://github.com/tseemann/snp-dists) on the core gene alignment generated by Roary.

Identification of linear plasmids by S1 nuclease pulsed-field gel electrophoresis

Isolates identified as non-susceptible (intermediate or resistant) to linezolid (MIC >4 µg/mL) were first analyzed for the presence of optrA using PCR following the previously described procedure (5). PCR samples that were optrA-positive were further analyzed by S1 nuclease pulsed-field gel electrophoresis (PFGE) as described previously (5). Briefly, samples were cultured to purity on SBA. Cells were embedded in 1.5% SeaKem Gold agarose (Thermo Fisher Scientific, Waltham, MA, United States) and lysed in Gram-positive lysis buffer (0.05% Tris, 0.1% EDTA, 0.01% N-lauroylsarcosine sodium salt). Agarose plug slices were cut and digested using 20U S1 nuclease (Takara Bio, Mountain View, CA, United States). Samples were then subjected to PFGE at 6 V/cm, 14°C for 20 hr with switch times of 1 and 25 s.

Long-read analyses

Isolates identified as carrying optrA were selected for long-read sequencing [Oxford Nanopore Technologies (ONT) MinION, Oxford, UK]. High molecular weight (HMW) genomic DNA was extracted using the Qiagen QIAamp DNA Blood Mini Kit (Germantown, MD), quantified using the Qubit dsDNA Broad Range Assay Kit (ThermoFisher Scientific, Waltham, MA), and purity was confirmed (ThermoFisher Scientific Nanodrop 2000 Spectrophotometer). The genomic library was prepared using 400 ng of HMW genomic DNA and the ONT Rapid Barcoding Sequencing Kit (SQK-RBK004), followed by sequencing on the R9.4.1 flow cell; the optional AMPure XP bead purification step was included during library prep. Guppy v.3.0.7 was used to live basecall long-read sequences (ONT MinION) using the fast basecalling model, excluding poor-quality reads (quality score <Q7). Porechop v.0.2.3 (https://github.com/rrwick/Porechop) was used to demultiplex and trim adapters, then Filtlong v0.2.0 (https://github.com/rrwick/Filtlong) was used to additionally filter out poor quality reads with parameters --min_length 1000 --keep_percent 90 --target_bases 5000000000 --trim --split 500. Using quality-controlled short- and long-read sequences, hybrid genome assemblies were generated for each sample using Unicycler v.0.4.8 (60) with default parameters. ContigExtender v0.1 (61) with default parameters was used to extend the linear plasmid pELF_2001564. Known linear form enterococcal plasmids were used for comparison (Table S2). To compare plasmids, assemblies were annotated with the Prokaryotic Genome Annotation Pipeline (62), and then visualized with BLASTN and Easyfig (63).

DATA AVAILABILITY

Raw sequencing reads and assembled genomes were deposited in NCBI GenBank under PRJNA573902; accessions are listed in Table S1.

SUPPLEMENTAL MATERIAL

The following material is available online at https://doi.org/10.1128/aac.00591-24.

Supplemental material. aac.00591-24-s0001.docx.

Figures S1 and S2; Table S2.

Table S1. aac.00591-24-s0002.xlsx.

Characteristics of isolates, including public accessions, taxonomic identification, site, source, identified resistance mechanisms, and antimicrobial susceptibility testing.

ASM does not own the copyrights to Supplemental Material that may be linked to, or accessed through, an article. The authors have granted ASM a non-exclusive, world-wide license to publish the Supplemental Material files. Please contact the corresponding author directly for reuse.