Molecular epidemiology of vancomycin-resistant Enterococcus faecium: a prospective, multicenter study in South American hospitals

JOURNAL OF CLINICAL MICROBIOLOGY, May 2010, p. 1562–1569 Vol. 48, No. 50095-1137/10/$12.00 doi:10.1128/JCM.02526-09Copyright © 2010, American Society for Microbiology. All Rights Reserved.

Molecular Epidemiology of Vancomycin-Resistant Enterococcus faecium: aProspective, Multicenter Study in South American Hospitals�

Diana Panesso,1,2† Jinnethe Reyes,1† Sandra Rincon,1 Lorena Díaz,1 Jessica Galloway-Pena,2,7

Jeannete Zurita,3 Carlos Carrillo,4‡ Altagracia Merentes,5 Manuel Guzman,5Javier A. Adachi,6 Barbara E. Murray,2,7 and Cesar A. Arias1,2*

Molecular Genetics and Antimicrobial Resistance Unit, Universidad El Bosque, Bogota, Colombia1; Center for the Study of Emerging andRe-Emerging Pathogens, Division of Infectious Diseases, Department of Internal Medicine, University of Texas Medical School at

Houston, Houston, Texas2; Hospital Vozandes, Quito, Ecuador3; Laboratorio Clínico Carlos Carrillo, Lima, Peru4;Centro Medico Caracas, Caracas, Venezuela5; University of Texas, M. D. Anderson Cancer Center, Houston, Texas6; andDepartment Microbiology and Molecular Genetics, University of Texas Medical School at Houston, Houston, Texas7

Received 29 December 2009/Returned for modification 8 February 2010/Accepted 25 February 2010

Enterococcus faecium has emerged as an important nosocomial pathogen worldwide, and this trend hasbeen associated with the dissemination of a genetic lineage designated clonal cluster 17 (CC17). Entero-coccal isolates were collected prospectively (2006 to 2008) from 32 hospitals in Colombia, Ecuador, Peru,and Venezuela and subjected to antimicrobial susceptibility testing. Genotyping was performed with allvancomycin-resistant E. faecium (VREfm) isolates by pulsed-field gel electrophoresis (PFGE) and mul-tilocus sequence typing. All VREfm isolates were evaluated for the presence of 16 putative virulence genes(14 fms genes, the esp gene of E. faecium [espEfm], and the hyl gene of E. faecium [hylEfm]) and plasmidscarrying the fms20-fms21 (pilA), hylEfm, and vanA genes. Of 723 enterococcal isolates recovered, E. faecaliswas the most common (78%). Vancomycin resistance was detected in 6% of the isolates (74% of which wereE. faecium). Eleven distinct PFGE types were found among the VREfm isolates, with most belonging tosequence types 412 and 18. The ebpAEfm-ebpBEfm-ebpCEfm (pilB) and fms11-fms19-fms16 clusters weredetected in all VREfm isolates from the region, whereas espEfm and hylEfm were detected in 69% and 23%of the isolates, respectively. The fms20-fms21 (pilA) cluster, which encodes a putative pilus-like protein,was found on plasmids from almost all VREfm isolates and was sometimes found to coexist with hylEfm andthe vanA gene cluster. The population genetics of VREfm in South America appear to resemble those ofsuch strains in the United States in the early years of the CC17 epidemic. The overwhelming presence ofplasmids encoding putative virulence factors and vanA genes suggests that E. faecium from the CC17genogroup may disseminate in the region in the coming years.

Enterococci are now recognized as important nosocomialpathogens worldwide and in the United States are ranked asthe second most common cause of nosocomial infections,after staphylococci (14). The two most common enterococ-cal species isolated from clinical samples are Enterococcusfaecalis and E. faecium; however, the proportions of isolatesof these two species have dramatically changed in the lastdecade. Whereas up until the early to mid-1990s E. faecaliswas the overwhelmingly predominant species isolated inU.S. hospitals (37), by 2008, the proportion of nosocomial E.faecalis/E. faecium strains was ca. 1.5:1, and there was animportant increase in the incidence of E. faecium nosoco-mial infections (14). Moreover, more than 80% of E. fae-cium isolates currently recovered from U.S. hospitals areresistant to vancomycin, and virtually all of them (�90%)exhibit ampicillin resistance (14). On the contrary, the prev-alence of vancomycin resistance in E. faecalis remains low

(�7% of isolates), and ampicillin resistance continues to beextremely rare. This change in the epidemiology of entero-coccal infections has been attributed to the increased abilityof a genogroup of E. faecium (designated clonal cluster 17[CC17]) to colonize the gastrointestinal tract of humans,cause disease (37), and exhibit high levels of resistance tomost antienterococcal antibiotics. Several virulence and col-onizing factors have been postulated to explain this in-creased virulence (4, 25, 28) and include the following: (i)the presence of an intact acm gene, which encodes a colla-gen adhesin and which has been associated with the patho-genesis of endocarditis in members of CC17 (25); (ii) the espgene of E. faecium (espEfm), which codes for an enterococcalsurface protein, which has been shown to play a role inbiofilm formation (12), and which transiently aggravatesexperimental urinary tract infection (18); (iii) the fms (E.faecium surface protein-encoding) genes, which encode cellwall-anchored proteins, including subunits of the enterococ-cal pili (13, 31); and (iv) the hyl gene of E. faecium (hylEfm;which encodes a putative glycosyl hydrolase), which is car-ried by transferable plasmids that have been shown to in-crease the ability of a laboratory strain of E. faecium tocolonize the gastrointestinal tracts of mice and also enhancethe virulence of a commensal strain of E. faecium in exper-imental peritonitis (4, 28).

* Corresponding author. Mailing address: Division of InfectiousDiseases, University of Texas Medical School at Houston, 6431 FanninSt., MSB 2.112, Houston, TX 77030. Phone: (713) 500-6765. Fax: (713)500-5495. E-mail: [email protected].

† Diana Panesso and Jinnethe Reyes contributed equally to thisstudy.

‡ Deceased.� Published ahead of print on 10 March 2010.

1562

In South America (Brazil and Argentina), vancomycin-resis-tant (VR) enterococcal infections have been described since1998 (9, 20). In a prospective multicenter surveillance studyconducted in 2003, the prevalence of VR among enterococci inColombia was found to be lower (9.7%) than that in theUnited States (1), and the isolation of similar percentages ofVR enterococci has been described more recently (21). Al-though the emergence of E. faecium CC17 has been docu-mented in Brazil, Chile, and Paraguay (16, 19, 39), prospectivestudies have not been performed and limited data regardingthe presence of the potential virulence determinants of CC17E. faecium (including hylEfm-containing plasmids) in SouthAmerica are available. Thus, we performed a multinational,multicenter prospective surveillance study with the aim ofcharacterizing the population genetics of enterococci circulat-ing in the northern region of South America. Clinical isolates(excluding colonizing isolates) were collected prospectivelyfrom 32 hospitals in four countries (Colombia, Ecuador, Peru,and Venezuela) and were further characterized at the molec-ular level.

(This study was presented in part at the 48th Annual Inter-science Conference on Antimicrobial Agents and Chemother-apy/Infectious Diseases Society of America 46th Annual Meet-ing, Washington, DC, 25 to 28 October 2008, abstr. C2-1998.)

MATERIALS AND METHODS

Specimen collection. A multicenter study was performed from February 2006through February 2008 to evaluate the molecular epidemiology of enterococci infour South American countries. Each participating tertiary-care hospital (a totalof 32 hospitals, from the Latin American Network of Antimicrobial Resistance)in Colombia (22 hospitals in six cities), Ecuador (5 hospitals in one city), Peru (3hospitals in one city), and Venezuela (2 hospitals in one city) collected consec-utive enterococcal isolates (duplicate organisms from the same patient wereexcluded). The samples were collected by the clinical laboratory from eachparticipating hospital and corresponded to samples from hospitalized patients inColombia, Peru, and Venezuela. In Ecuador, the clinical laboratory also col-lected samples from ambulatory services and outpatient clinics. The isolatesoriginated from the following clinical specimens: blood, urine, secretions fromsurgical wounds, peritoneal fluid, abdominal abscesses, joint aspirates, osteomy-elitis aspirates, pleural fluid, pericardial effusion, cerebral abscesses, and cere-brospinal fluid (CSF). In an attempt to avoid isolates that likely representedcolonization, enterococci recovered from sputum, rectal swabs, catheters, or skin(unless they originated in an infected surgical wound) were excluded. Eachhospital identified the microorganisms by using either automated methods (per-formed with the Vitek or MicroScan system) or manual methods, and once thecorresponding isolate was included in the study, it was sent to the referencelaboratory (located in Bogota, Colombia) via courier in transport medium(Amies; BBL). Upon arrival, the reference laboratory confirmed the purity of theisolate and confirmed the identification by molecular methods using multiplexPCR for enterococci (see below).

Susceptibility testing and molecular methods for species-specific identifica-tion and van genotype determination. Susceptibility tests for enterococci wereperformed by an agar dilution method with ampicillin, ciprofloxacin, chloram-phenicol, linezolid, vancomycin, and teicoplanin. The isolates were also evalu-ated for high-level resistance (HLR) to streptomycin (2,000 �g/ml) and genta-micin (500 �g/ml). All isolates from blood and CSF identified as E. faecalis werescreened for the presence of the �-lactamase enzyme by use of a nitrocefin test,and all susceptibility tests were conducted by the methodology suggested by theCLSI (8). MIC determinations were carried out with the inclusion of a controlreference strain (E. faecalis ATCC 29212). Species-specific identification of theenterococci and determination of the van genotype of the vancomycin-resistantisolates were performed by PCR, as described previously (2, 10) (primers for thedetection of the recently described vanL genotype [6] were not included). E.faecium BM4147 (vanA), E. faecalis V583 (vanB), and E. gallinarum BM4174(vanC1) were used as control strains for MIC determinations and PCR assays.

Molecular typing. Pulsed-field gel electrophoresis (PFGE) of the vancomycin-resistant E. faecium (VREfm) isolates was performed by use of some modifica-

tions of a previously described method (22), and the banding patterns wereinitially interpreted by visual inspection, according to the criteria specified byTenover et al. (35). Subsequently, cluster analysis was performed by the methodof Dice and the unweighted-pair group method using average linkages(UPGMA). The band tolerance was set at 1.5%, and the threshold cutoff valuewas set at 85%. Representative isolates of each PFGE subtype of VREfm werefurther genotyped by multilocus sequence typing (MLST) by a standard protocoldescribed previously (15, 29). Fragments of seven housekeeping genes (atpA, ddl,gdh, purK, gyd, pstS, and adk) were sequenced, the allelic profiles were obtained,and the sequence type (ST) for each unique allelic profile was designated on thebasis of the information at the MLST website (http://efaecium.mlst.net).

Detection of plasmids and putative virulence genes. The preparation of colonylysates on nylon membranes and hybridization under high-stringency conditionswere performed as described previously (33). DNA probes for the hylEfm, espEfm,and 14 fms genes were obtained by using previously published primers (11, 27,32); the probes were radiolabeled by use of the RadPrime DNA labeling system(Invitrogen, Carlsbad, CA). Plasmid detection in representative isolates ofVREfm belonging to CC17 was performed by the use of S1 nuclease and PFGE,according to a previously described protocol (4, 5). This methodology allows thedetection and estimation of the size of large bacterial plasmids in the presence ofgenomic DNA by PFGE (5). Plasmid bands were subsequently hybridized withprobes targeting the hylEfm, fms20, fms21 (pilA), and vanA genes to determine ifthese genes were colocated in the same plasmid, as was reported previously (17).E. faecium TX0016 (strain DO) (23) and ERV-99 (4) were used as controls forcolony hybridizations and the PFGE-S1 nuclease experiments.

RESULTS

Phenotypic characteristics of enterococcal isolates from thenorthern region of South America. A total of 760 consecutiveEnterococcus sp. isolates was collected in the four countries.Thirty-seven isolates were not included in the study due toprotocol violations (contamination, isolates were from thesame patient, the source was not included in the protocol, ormisidentification). Among the isolates included (a total of723), Colombian hospitals contributed 309 (43%) of the iso-lates, Peruvian hospitals contributed 164 (23%), Ecuadorianhospitals contributed 148 (20%), and Venezuelan hospitalscontributed 102 (14%). The majority (78%) were E. faecalisisolates mostly recovered from urine (54%) and blood (14%)(no �-lactamase-producing isolate was detected). E. faeciumcomprised 15% (n � 111) of the enterococcal isolates (Table1). The most common clinical sources of E. faecium includedurine (38%), blood (21%), and surgical wound infections(13%). Other enterococcal species also identified by PCR in-cluded E. avium, E. hirae, E. casseliflavus, E. gallinarum, and E.durans. We were unable to determine the enterococcal speciesfor 5% (n � 35) of the isolates.

The resistance rates of the enterococci from the Andeanregion are shown in Table 1. All E. faecalis isolates were sus-ceptible to ampicillin and linezolid, whereas the rates of HLRto gentamicin and streptomycin were 28% and 30%, respec-tively; ciprofloxacin resistance was found in 29% of the E.faecalis isolates (Table 1). In contrast, the rates of resistance toampicillin and vancomycin and HLR to gentamicin and strep-tomycin in E. faecium were much higher (76%, 31%, 36%, and44%, respectively). The highest rates of resistance to ampicillinand vancomycin and HLR to gentamicin and streptomycin inE. faecium isolates were found in Peru (90%, 48%, 55%, and65%, respectively), and the lowest were found in Venezuela(50%, 25%, 8%, and 21%, respectively). Among the ampicil-lin-resistant E. faecium isolates from the Andean region, atotal of 51 isolates (46%) had MICs of �64 �g/ml. Linezolidwas active against all the E. faecium isolates tested, and only

VOL. 48, 2010 ENTEROCOCCI IN LATIN AMERICAN HOSPITALS 1563

4% of the E. faecium isolates tested were resistant to chlor-amphenicol.

E. faecium CC17 is also predominant in the northern regionof South America. Thirty-five VREfm isolates were found, andall of them exhibited the VanA phenotype (high levels ofresistance to both vancomycin and teicoplanin) and harboredthe vanA gene cluster. Typing by PFGE revealed 11 differentbanding patterns (patterns A to K) (Fig. 1). A single PFGEbanding pattern was predominant in each country (referred toas type A in Venezuela, type B in Peru, type C in Colombia,and type D in Ecuador) (Fig. 1), although isolates from differ-ent countries (e.g., types A and E) could be found to haverelated PFGE types. MLST analysis of representative isolatesof each PFGE type indicated that the most frequent ST wasST412; other STs included ST17, ST18, ST125, ST203, ST280,and ST282 (all of which were associated with the CC17 geneticlineage). A representative of PFGE pattern J from a Peruvianhospital was found to belong to a novel ST (ST494) which hadnot been reported previously.

Only three vancomycin-resistant E. faecalis were isolated inthe 2 years of the study; all three isolates were recovered inColombia, harbored the vanB gene cluster, and were previ-ously characterized as belonging to the same unique clonalcluster that has been circulating in that country since 2001(ST2 by trilocus sequence typing and MLST) (3, 7).

CC17 vancomycin-resistant E. faecium isolates from SouthAmerica are enriched in the fms genes. It had previously beenshown that the members of CC17 of E. faecium harbor anumber of genes encoding sortase-anchoring motif (LPXTG)-

containing cell wall proteins (13, 31); at least 15 of these genesencode predicted microbial surface components recognizingadhesive matrix molecule (MSCRAMMs) proteins in a U.S.strain (strain TX0016 [strain DO]) that was sequenced (32). Ithas been proposed that these genes function as importantvirulence determinants mediating adhesion to mammalian tis-sues, and several are enriched in the CC17 genogroup (11, 13,31). Thus, we set out to investigate if these genes were alsofrequently found in CC17 isolates of E. faecium from LatinAmerica. Figure 2 shows the results of colony hybridizations ofisolates with several putative virulence genes (including 14 ofthe fms genes; acm, which may be present as an intact gene orpseudogene, was not tested since it has been detected in almostall isolates studied previously [26]). Indeed, the four clusters ofgenes encoding MSCRAMM proteins (31) were highly repre-sented among the South American E. faecium isolates tested.The ebpAEfm-ebpBEfm-ebpCEfm (pilB) and fms11-fms19-fms16clusters, which harbor genes encoding putative proteins withconserved pilin motifs and E-boxes of pilus proteins of Gram-positive bacteria, were present in all our South American VRisolates (Fig. 2). Interestingly, the scm gene, which has beenshown to encode a collagen adhesin (the second collagen ad-hesin from E. faecium), was present in more than 90% of theisolates. The other fms genes were also frequently present atrates ranging from 67% to 97% in the VREfm isolates. TheespEfm and hylEfm genes, which have previously been associatedwith CC17 (28, 38), were detected in South American VR E.faecium isolates (69% and 23% of isolates, respectively).

TABLE 1. Resistance profiles of enterococci from the northern region of South America

Organism(no. of isolates)

% (no.) of isolates resistant to the following antimicrobial agentsa:

VAN TEC CIP CHL LZD AMP HLR-GEN HLR-STR

EnterococciColombia (309) 5 (16) 3 (10) 30 (93) 12 (38) 0 (0) 10 (31) 19 (60) 29 (89)Perú (164) 9 (15) 8 (14) 55 (91) 17 (28) 0 (0) 16 (27) 53 (87) 47 (77)Ecuador (148) 5 (7) 3 (5) 25 (37) 17 (26) 0 (0) 10 (15) 28 (42) 28 (42)Venezuela (102) 9 (9) 6 (6) 28 (29) 9 (9) 0 (0) 12 (12) 15 (15) 17 (17)Total (723) 6 (47) 5 (35) 35 (250) 14 (101) 0 (0) 12 (85) 28 (204) 31 (225)

Enterococcus faecalisColombia (251) 1 (3) 0 (0) 26 (65) 13 (33) 0 (0) 0 (0) 17 (44) 27 (67)Perú (128) 0 (0) 0 (0) 48 (61) 20 (26) 0 (0) 0 (0) 54 (69) 44 (56)Ecuador (122) 0 (0) 0 (0) 22 (27) 20 (25) 0 (0) 0 (0) 26 (32) 27 (33)Venezuela (59) 0 (0) 0 (0) 19 (11) 13 (8) 0 (0) 0 (0) 22 (13) 17 (11)Total (560) 0.5 (3) 0 (0) 29 (164) 16 (92) 0 (0) 0 (0) 28 (158) 30 (167)

Enterococcus faeciumColombia (41) 24 (10) 24 (10) 61 (25) 7 (3) 0 (0) 76 (31) 34 (14) 44 (18)Perú (29) 48 (14) 48 (14) 93 (27) 3 (1) 0 (0) 90 (26) 55 (16) 65 (19)Ecuador (17) 29 (5) 29 (5) 59 (10) 0 (0) 0 (0) 88 (15) 47 (8) 41 (7)Venezuela (24) 25 (6) 25 (6) 58 (14) 4 (1) 0 (0) 50 (12) 8 (2) 21 (5)Total (111) 31 (35) 31 (35) 68 (76) 4 (5) 0 (0) 76 (84) 36 (40) 44 (49)

Other enterococcal speciesColombia (17) 18 (3) 0 (0) 18 (3) 12 (2) 0 (0) 0 (0) 12 (2) 23 (4)Perú (7) 14 (1) 0 (0) 43 (3) 14 (1) 0 (0) 14 (1) 28 (2) 28 (2)Ecuador (9) 22 (2) 0 (0) 0 (0) 11 (1) 0 (0) 0 (0) 22 (2) 22 (2)Venezuela (19) 16 (3) 0 (0) 21 (4) 0 (0) 0 (0) 0 (0) 0 (0) 5 (1)Total (52) 17 (9) 0 (0) 19 (10) 8 (4) 0 (0) 2 (1) 11 (6) 17 (9)

a VAN, vancomycin; TEC, teicoplanin; CIP, ciprofloxacin; CHL, chloramphenicol; LZD, linezolid; AMP, ampicillin; HLR-GEN, high-level resistance to gentamicin;HLR-STR, high-level resistance to streptomycin.

1564 PANESSO ET AL. J. CLIN. MICROBIOL.

The fms20-fms21 (pilA) cluster and hylEfm are carried onplasmids associated with the vanA gene in South AmericanVREfm. It was previously shown that hospital-associated E.faecium isolates from diverse geographical locations carrylarge, transferable plasmids containing the hylEfm gene (whichencodes a putative family 84 glycosyl hydrolase enzyme) that,additionally, are associated with vancomycin and aminoglyco-side resistance genes (4). These hylEfm-containing plasmidshave emerged as important virulence and colonization deter-minants of CC17 of E. faecium (4, 28). More recently, one ofthe fms clusters (fms20-fms21 [pilA]) encoding pilus-like pro-teins was also found to be located in a large hylEfm-carryingplasmid in U.S. strain TX0016 (strain DO) (17). Table 2 andFig. 3 show the results of S1 nuclease digestion; PFGE; and thehybridization of representatives of South American VR E.faecium isolates with probes targeting the fms20-fms21 (pilA)cluster, hylEfm, and vanA. When it was present, the fms20-fms21 (pilA) cluster was always encoded on plasmids rangingfrom ca. 60 kb to 242 kb. Of note, in some strains, fms21-containing plasmids were found not to harbor the other mem-ber of the cluster (fms20), and in three strains (all from Peru)

an additional copy of fms21 was detected in a different plasmid(strains P575, P1139, and P1986; Table 2 and Fig. 3). Further-more, the fms20-fms21 cluster was found to be in the sameplasmid as hylEfm in three strains (estimated plasmid sizes, 190and 230 kb) and on the same vanA-containing plasmid in threestrains (Table 2). No plasmid was found to carry the fms20-fms21 cluster, hylEfm, and vanA together.

DISCUSSION

In the work described here, we performed the first prospec-tive, multinational study evaluating the molecular epidemiol-ogy of VR enterococci in the northern region of South Amer-ica. Our results indicate that the prevalence of vancomycinresistance is relatively low in the participating countries (only6% of isolates included in this study) compared to that in theUnited States. In our previous multicenter study performedwith isolates from 15 hospitals in Colombia in 2001 and 2002,the rate of vancomycin resistance among enterococcal isolateswas 9.7%. Also, unlike in the United States, E. faecalis contin-ues to be, by far, the most frequent enterococcal species iso-

FIG. 1. Molecular typing of VR E. faecium from the Andean region of South America. The phylogenetic tree was constructed by use of theDice coefficient and UPGMA clustering; the band tolerance was set at 1.5%, and the threshold cutoff value was set at 85%. ST, sequence type;CC17, clonal cluster 17.

VOL. 48, 2010 ENTEROCOCCI IN LATIN AMERICAN HOSPITALS 1565

lated in this region (ratio of E. faecalis to E. faecium, 5:1).Vancomycin resistance in E. faecalis was found sporadically inonly three isolates from one city of the participating countries,and ampicillin was active against all E. faecalis isolates, under-scoring the fact that �-lactams and glycopeptides continue tobe useful for the treatment of enterococcal infections in thenorthern region of South America. On the other hand, the

rates of HLR to aminoglycosides in E. faecalis were ca. 30%,and there were important regional variations; for example, inthe participating hospitals from Peru, more than 50% of the E.faecalis isolates exhibited HLR to gentamicin and 44% exhib-ited HLR to streptomycin, indicating that the treatment ofendovascular infections with the combination of a cell wallagent and an aminoglycoside in some areas of South America

FIG. 2. Frequency of detection of putative virulence genes in 35 VR E. faecium isolates from the Andean region of South America. Thenumbers on the left-hand side indicate the four clusters of genes encoding characterized putative proteins of the enterococcal pili (30).

TABLE 2. Plasmid or chromosomal location of fms20, fms21 (pilA), hylEfm, and vanA in vancomycin-resistant Enterococcus faecium isolatesfrom the Andean region of South America

Country Code Source PFGEtype ST

Plasmid or chromosomal locationa

Estimated plasmid size(s)fms20 fms21(pilA) hylEfm vanA

Perú P575 Urine G 280 � (P) � (P) � (P) � (P) hylEfm and fms20-fms21, �190 kb;fms21, �70 kb; vanA, �60 kb;

P1123 Blood B 412 � (P) � (P) � � (P) fms20-fms21, �70 kb; vanA, �48 kbP1137 Wound infection E 18 � � � � (P) vanA, �60 kbP1139 Urine F 280 � (P) � (P) � � (P) fms20-fms21, �190 kb; fms21, �70

kb; vanA, �48 kbP1140 Wound infection A 412 � (P) � (P) � � (P) fms20-fms21 and vanA, �60 kbP1190 Blood E 125 � (P) � (P) � (P) � (P) hylEfm and fms20-fms21, �230 kb;

vanA, �60 kbP1985 Urine B 412 � (P) � (P) � � (P) fms20-fms21, �200 kb; vanA, �48 kbP1986 Blood J 494 � (P) � (P) � � (C) fms20-fms21, �80 kb; fms21, �60 kbP2022 Abdominal abscess H 282 � (P) � (P) � � (P) fms20-fms21, �75 kb; vanA, �60 kbP2074 Blood F NDb � (P) � (P) � (P) � (P) hylEfm and fms20-fms21, �220 kb;

vanA, �70 kb

Colombia C497 Blood C 18 � � � (P) � (P) hylEfm and vanA, �160 kbC621 Peritoneal fluid A 412 � � (P) � � (P) fms21 and vanA, �60 kbC1688 Blood C ND � (P) � (P) � � (C) fms20-fms21, �180 kbC1904 Blood E ND � � � (P) � (P) hylEfm and vanA, �160 kbC1910 Urine C 18 � (P) � (P) � � (P) fms20-fms21, �180 kb; vanA, �48 kb

Ecuador E417 Blood D 17 � (P) � (P) � � (P) fms20-fms21, �180 kb; vanA, �48kb;E422 Urine I 203 � (P) � (P) � � (C) fms20-fms21, �70 kb

Venezuela V689 Wound infection A 412 � (P) � (P) � � (P) fms20-fms21 and vanA, �60 kbV2216 Abdominal abscess A ND � (P) � (P) � � (P) fms20-fms21 and vanA, �70 kb

a P, plasmid; C, chromosomal.b ND, not done; MLST was not performed because the isolate had a PFGE pattern closely related to that of another isolate whose MLST profile was determined.

1566 PANESSO ET AL. J. CLIN. MICROBIOL.

may be challenging and alternative bactericidal therapiesshould be sought. Also, we did not find important differencesin phenotypic or genotypic characteristics between VREfmisolates from urine and blood (except that the blood isolateshad higher rates of resistance to chloramphenicol [20%] thanthose VREfm isolates that originated in urine [0%]).

Although our findings indicate that the proportion of E.faecalis/E. faecium is considerably higher in South Americanhospitals than in U.S. hospitals (14), the South AmericanVREfm isolates have genotypic and phenotypic characteristicssimilar to those of their U.S. counterparts: (i) South American

VREfm isolates mostly belong to ST412 and ST18 (the CC17lineage), indicating that a hospital-associated genetic lineage ispresent in the region and supporting previous findings of stud-ies performed in other countries of South America (16, 19, 39);(ii) the South American CC17 VREfm isolates also harbor animportant number of MSCRAMM genes, including the fourgene clusters encoding predicted components of the E. faeciumpili (particularly the ebpEfm and the fms11-fms19-fms16 clus-ters) (13, 31, 32), a finding that is consistent with the fact thatthe presence of these genes appears to be more reliably asso-ciated with members of the CC17 genetic lineage (11) than the

FIG. 3. Plasmids containing the fms20 gene or the fms21 gene (pilA), or both genes, in representative E. faecium isolates from the Andeanregion of South America. (A) Genetic organization of fms20-fms21 (pilA) in E. faecium TX0016 (strain DO). This locus also contains a predictedclass C sortase-encoding gene (srtC4) and a class A sortase-encoding gene (srtA) (30). (B) S1 nuclease digestion of total DNA of E. faecium strains,followed by PFGE (left panel) and hybridization with an fms20 probe (right panel). Lane 1, bacteriophage lambda ladder (molecular sizes [inkilobases] are shown to the left); lanes 2 and 14, TX0016 (strain DO); lanes 3, ERV-99 (4); lanes 4, P1985 (Peru); lanes 5, P1123 (Peru); lanes6, P2022 (Peru); lanes 7, E417 strain (Ecuador); lanes 8, P1139 (Peru); lanes 9, P575 (Peru); lanes 10, P2074 (Peru); lanes 11, E422 (Ecuador);lanes 12, P1986 (Peru); lanes 13, C1688 (Colombia); lanes 15, V2216 (Venezuela). Plasmid bands are shown as linearized fragments on the gel;the white arrows indicate the plasmid bands hybridizing with the fms20 probe. (C) S1 nuclease digestion of total DNA of E. faecium strains,followed by PFGE (left panel) and hybridization with an fms21 (pilA) probe (right panel). Lane 1, bacteriophage lambda ladder (molecular sizes[in kilobases] are shown to the left); lanes 2 and 14, TX0016 (strain DO); lanes 3, ERV-99 (4); lanes 4, P1985 (Peru); lanes 5, P1123 (Peru); lanes6, P2022 (Peru); lanes 7, E417 (Ecuador); lanes 8, P1139 (Peru); lanes 9, P575 (Peru); lanes 10, P2074 (Peru); lanes 11, E422 (Ecuador); lanes 12,P1986 (Peru); lanes 13, C1688 (Colombia); lanes 15, V2216 (Venezuela). The white arrows indicate the plasmid bands hybridizing with the fms21(pilA) probe.

VOL. 48, 2010 ENTEROCOCCI IN LATIN AMERICAN HOSPITALS 1567

presence of espEfm and hylEfm; and (iii) high-level of resistanceto ampicillin is commonly found in E. faecium isolates from theregion, with the majority of isolates exhibiting MICs of �64�g/ml.

Recently, a genotypic analysis of early clinical isolates of E.faecium in the United States (11) showed that CC17 E. faeciumisolates have been circulating in the United States since at least1982, and it was postulated that an ancestral genogroup of E.faecium enriched with the fms genes was able to subsequentlyacquire additional virulence and antibiotic resistance determi-nants (such as espEfm, hylEfm, HLR to ampicillin, and the vangene clusters) and establish itself as an epidemic hospital-associated pathogen in the United States. Our results suggestthat the population genetics of VREfm in the northern regionof South America resembles that of isolates responsible forearly outbreaks in the United States (in about 1982) (11), andit is tempting to speculate that an increase in the frequency ofE. faecium infections in South America may be expected in thefollowing decade as members of the hospital-associated lin-eage disseminate and establish themselves as nosocomialpathogens in the region. Nonetheless, it is important to pointout that the number of hospitals participating in this studydiffered in each country; while 22 hospitals from six differentcities were represented in Colombia, a lower number of hos-pitals in the other three countries participated in the study.Thus, it is difficult to make countrywide generalizations of themolecular epidemiology of enterococcal isolates in Peru, Ven-ezuela, and Ecuador. However, our data provide importantinformation on the type of organisms currently circulating inthe region.

An important goal of our study was to determine if SouthAmerican VREfm isolates also contain genes encoding puta-tive pilus proteins that are present on plasmids coexisting withhylEfm or vancomycin resistance gene clusters, as previouslyshown for other strains (4, 17). Enterococcal pili are importantcell surface structures that have been implicated in the patho-genesis of experimental endocarditis and urinary tract infec-tions (24, 34). These trimeric units are composed of a majorbackbone subunit and two minor subunits (36) with an N-terminal signal sequence and a C-terminal cell wall-sortingsignal (30) encoded by different genes. Indeed, we found thatone of these clusters (fms20-fms21 [pilA]) is carried by largeplasmids in the majority of isolates studied. Moreover, in somestrains, the fms20-fms21 (pilA) cluster coexists on the sameplasmid with the hylEfm gene or with vanA, providing furtherevidence that the plasmid dissemination of virulence, resis-tance, and/or colonization determinants may be a strategyadopted by hospital-associated lineages of E. faecium world-wide. This finding also supports the hypothesis that the An-dean region of South America may be in the early stages of anE. faecium epidemic and that as the fms20-fms21/hylEfm-carry-ing plasmids disseminate, the ability of the organisms to colo-nize the gastrointestinal tract of patients and cause disease mayincrease.

In summary, this is the first multicenter, multinational pro-spective study of the molecular epidemiology of VR entero-cocci in South America. Our results indicate that the preva-lence of E. faecium is low among clinical isolates, but genotypicanalysis supports the fact that the population genetics of E.faecium is similar to that seen in the United States more than

25 years ago, suggesting that the dissemination of these highlyadapted E. faecium organisms in the hospital environment mayoccur in South America in the coming years through the clonalexpansion of hospital-adapted isolates and with the horizontaltransfer of virulence and colonization determinants.

ACKNOWLEDGMENTS

This report is dedicated to the memory of Carlos Carrillo.We thank the Universidad El Bosque for financial support and are

indebted to Karen Jimenez and German Contreras for technical as-sistance. We also thank Maria Virginia Villegas for facilitating thecoordination of the participating centers in Colombia. We are gratefulto Kavindra S. Singh and Jaime Moreno for providing technical ex-pertise and Jouko Sillanpaa for critical review of the manuscript.

The following personnel and hospitals participated in the collectionof isolates: in Bogota, Colombia, Claudia Londono and Martha Her-rera (Fundacion Salud Bosque); Constanza Correa (Hospital SimonBolívar); Norma Montoya (Clínica de Occidente); Wilson Daza andMartha Uzeta (Clínica del Nino); Narda Olarte and Martha Garzon(Hospital El Tunal); Gloria Gallo (Hospital Santa Clara); FernandoPenaloza and Nubia Escobar (Hospital Occidente de Kennedy); Mar-tha Ruiz (Clínica San Pedro Claver); Carlos Alvarez, Nidia Torres, andZiomara Gonzalez (Hospital San Ignacio); Clara Luz Rico (FundacionSanta Fe de Bogota); Giovanni Rodríguez and Deise Rojas (ClínicaInfantil Colsubsidio); Juan Benavides, Maritza Perez, and EsperanzaGuevara (Clínica Saludcoop Jorge Pineros Corpas); and Patricia Ar-royo (Instituto Nacional de Cancerología); in Cali, Colombia, MaríaVirginia Villegas and Beatriz Vanegas (Centro Medico Imbanaco);María del Socorro Rojas (Clínica Saludcoop Occidente Cali); andErnesto Martínez and Nancy Villamarín (Hospital Universitario delValle); in Medellín, Colombia, Sergio Jaramillo and Jaime Lopez(Hospital Pablo Tobon Uribe) and Magda Cardenas (Clínica Salud-coop Juan Luis Londono); in Bucaramanga, Colombia, Adriana Pinto(Clínica La Foscal and Fundacion Cardiovascular); in Neiva, Colom-bia, Marino Cabrera and Luz Eneyda Quintero (Hospital Universita-rio Hernando Moncaleano Perdomo) and Carmen Elisa Llanos (Hos-pital Universitario San Jorge); in Ecuador, Hospital Vozandes,Hospital Eugenio Espejo, Hospital Baca Ortiz, Hospital Carlos AndresAndrade Marín, and Hospital General de las Fuerzas Armadas; inPeru, Gene Martínez Medina and Susana Kuwae de Okuhama (Labo-ratorio Clínico Carlos Carrillo); Federico Yanez Rojas and LilianaAlvarado (Hospital Nacional Sergio Bernales); Greenlandia FerreyrosBrandon and María Silva (Instituto Nacional de Enfermedades Neo-plasicas); and Rosa Avurio Usca and Gladys Patino Soto (HospitalNacional Hipolito Unanue); and in Venezuela, Adele Rizzi (CentroMedico de Caracas and Hospital Vargas de Caracas).

This work was funded in part by an independent research grant fromPfizer, SA. C.A.A. and B.E.M. are supported by a K99/R00 Pathway toIndependence Award (award 4R00 AI72961) and R01 grant AI067861from the National Institute of Allergy and Infectious Diseases, respec-tively. D.P. was partially funded by a graduate scholarship from theInstituto Colombiano para el Desarrollo de la Ciencia y Tecnología,Francisco Jose de Caldas, COLCIENCIAS; and S.R. was supported byan ASM-PAHO Infectious Disease Epidemiology and SurveillanceFellowship.

C.A.A. has received lecture fees from Pfizer, Novartis, and Merckand grant support from Pfizer. B.E.M. has had grant support fromJohnson & Johnson, Astellas, Palumed, and Intercell and has served asconsultant for Astellas Pharma US Inc., Theravance Inc., Cubist, Tar-ganta Therapeutics Corporation, Johnson & Johnson, Pfizer, Astra-Zeneca, and Wyeth-Ayerst. J.Z. reports the receipt of research grantsfrom Pfizer and Wyeth. M.G. has served as consultant for Pfizer,Merck and Co., Wyeth, and Becton and Dickinson.

REFERENCES

1. Arias, C. A., J. Reyes, M. Zuniga, L. Cortes, C. Cruz, L. Rico, and D.Panesso. 2003. Multicentre surveillance of antimicrobial resistance in en-terococci and staphylococci from Colombian hospitals, 2001–2002. J. Anti-microb. Chemother. 51:59–68.

2. Arias, C. A., B. Robredo, K. V. Singh, C. Torres, D. Panesso, and B. E.Murray. 2006. Rapid identification of Enterococcus hirae and Enterococcus

1568 PANESSO ET AL. J. CLIN. MICROBIOL.

durans by PCR and detection of a homologue of the E. hirae mur-2 gene inE. durans. J. Clin. Microbiol. 44:1567–1570.

3. Arias, C. A., S. Rincon, S. A. Chowdhury, E. Martínez, W. Coronell, J. Reyes,S. R. Nallapareddy, and B. E. Murray. 2008. MRSA USA300 clone andVREF—a U.S.-Colombian connection? N. Engl. J. Med. 359:2177–2179.

4. Arias, C. A., D. Panesso, K. V. Singh, L. B. Rice, and B. E. Murray. 2009.Cotransfer of antibiotic resistance genes and a hylEfm-containing virulenceplasmid in Enterococcus faecium. Antimicrob. Agents Chemother. 53:4240–4246.

5. Barton, B. M., G. P. Harding, and A. J. Zuccarelli. 1995. A general methodfor detecting and sizing large plasmids. Anal. Biochem. 226:235–240.

6. Boyd, D. A., B. M. Willey, D. Fawcett, N. Gillani, and M. R. Mulvey. 2008.Molecular characterization of Enterococcus faecalis N06-0364 with low-levelvancomycin resistance harboring a novel D-Ala–D-Ser gene cluster, vanL.Antimicrob. Agents Chemother. 52:2667–2672.

7. Chowdhury, S. A., C. A. Arias, S. R. Nallapareddy, J. Reyes, R. J. Willems,and B. E. Murray. 2009. A trilocus sequence typing scheme for hospitalepidemiology and subspecies differentiation of an important nosocomialpathogen, Enterococcus faecalis. J. Clin. Microbiol. 47:2713–2719.

8. Clinical and Laboratory Standards Institute. 2009 Performance standardsfor antimicrobial susceptibility testing; 19th informational supplementM100-S19. Clinical and Laboratory Standards Institute, Wayne, PA.

9. Dalla Costa, L. M., D. C. Souza, L. T. Martins, R.C. Zanella, M. C. Bran-dilone, S. Bokermann, H. S. Sader, and H. A. Souza. 1998. Vancomycin-resistant Enterococcus faecium: first case in Brazil. Braz. J. Infect. Dis. 2:160–163.

10. Dutka-Malen, S., S. Evers, and P. Courvalin. 1995. Detection of glycopep-tide resistance genotypes and identification to the species level of clinicallyrelevant enterococci by PCR. J. Clin. Microbiol. 33:24–27. (Erratum, 33:1434.)

11. Galloway-Pena, J. R., S. R. Nallapareddy, C. A Arias, G. M. Eliopoulos, andB. E. Murray. 2009. Analysis of clonality and antibiotic resistance amongearly clinical isolates of Enterococcus faecium in the United States. J. Infect.Dis. 200:1566–1573.

12. Heikens, E., M. J. Bonten, and R. J. Willems. 2007. Enterococcal surfaceprotein Esp is important for biofilm formation of Enterococcus faeciumE1162. J. Bacteriol. 189:8233–8240.

13. Hendrickx, A. P., M. J. Bonten, M. van Luit-Asbroek, C. M. Schapendonk,A. H. Kragten, and R. J. Willems. 2008. Expression of two distinct types ofpili by a hospital-acquired Enterococcus faecium isolate. Microbiology 154:3212–3223.

14. Hidron, A. I., J. R. Edwards, J. Patel T. C. Horan, D. M. Sievert, D. A.Pollock, and S. K. Fridkin. 2008. Antimicrobial-resistant pathogens associ-ated with healthcare-associated infections: annual summary of data reportedto the national healthcare safety network at the Centers for Disease Controland Prevention, 2006–2007. Infect. Control. Hosp. Epidemiol. 29:996–1011.(Erratum, 30:107, 2009.)

15. Homan, W. L., D. Tribe, S. Poznanski, M. Li, G. Hogg, E. Spalburg, J. D.Van Embden, and R. J. Willems. 2002. Multilocus sequence typing schemefor Enterococcus faecium. J. Clin. Microbiol. 40:1963–1971.

16. Khan, M. A., J. B. Northwood, R. G. J. Loor, A. T. R Tholen, E. Riera, M.Falcon, A. van Westreenen, and J. P. Hays. 2009. High prevalence of ST-78infection-associated vancomycin-resistant Enterococcus faecium from hospi-tals in Asuncion, Paraguay. Clin. Microbiol. Infect. doi:10.1111/j.1469-0691.2009.02898x.

17. Kim, D. S., K. V. Singh, S. R. Nallapareddy, X. Qin, D. Panesso, C. A. Arias,and B. E. Murray. 2010. The fms21 (pilA)-fms20 locus encoding one of fourdistinct pili of Enterococcus faecium is harbored on a large transferableplasmid associated with gut colonization and virulence. J. Med. Microbiol.59:505–507.

18. Leendertse, M., E. Heikens, and L. M. Wijnands. 2009. Enterococcal surfaceprotein transiently aggravates Enterococcus faecium-induced urinary tractinfection in mice. J. Infect. Dis. 200:1162–1165.

19. Lopez, M., J. C. Hormazabal, A. Maldonado, G. Saavedra, F. Baquero, J.Silva, C. Torres, and R. del Campo. 2009. Clonal dissemination of Entero-coccus faecalis ST201 and Enterococcus faecium CC17-ST64 containingTn5382-vanB2 among 16 hospitals in Chile. Clin. Microbiol. Infect. 15:586–588.

20. Marín, M. E., J. R. Mera, R. C. Arduino, A. P. Correa, T. M. Coque, D.

Stamboulian, and B. E. Murray. 1998. First report of vancomycin-resistantEnterococcus faecium isolated in Argentina. Clin. Infect. Dis. 26:235–236.

21. Moet, G. J., R. N. Jones, D. J. Biedenbach, M. G. Stilwell, and T. R. Fritsche.2007. Contemporary causes of skin and soft tissue infections in North Amer-ica, Latin America, and Europe: report from the SENTRY AntimicrobialSurveillance Program (1998–2004). Diagn. Microbiol. Infect. Dis. 57:7–13.

22. Murray, B. E., K. V. Singh, J. D. Heath, B. R. Sharma, and G. M. Weinstock.1990. Comparison of genomic DNAs of different enterococcal isolates usingrestriction endonucleases with infrequent recognition sites. J. Clin. Micro-biol. 28:2059–2063.

23. Nallapareddy, S. R., G. M. Weinstock, and B. E. Murray. 2003. Clinicalisolates of Enterococcus faecium exhibit strain-specific collagen binding me-diated by Acm, a new member of the MSCRAMM family. Mol. Microbiol.47:1733–1747.

24. Nallapareddy, S. R., K. V. Singh, J. Sillanpaa, D. A. Garsin, M. Hook, S. L.Erlandsen, and B. E. Murray. 2006. Endocarditis and biofilm-associated piliof Enterococcus faecalis. J. Clin. Invest. 116:2799–2807.

25. Nallapareddy, S. R., K. V. Singh, and B. E. Murray. 2008. Contribution ofthe collagen adhesin Acm to pathogenesis of Enterococcus faecium in exper-imental endocarditis. Infect. Immun. 76:4120–4128.

26. Nallapareddy, S. R., K. V. Singh, P. C. Okhuysen, and B. E. Murray. 2008.A functional collagen adhesin gene, acm, in clinical isolates of Enterococcusfaecium correlates with the recent success of this emerging nosocomialpathogen. Infect. Immun. 76:4110–4119.

27. Rice, L. B., L. Carias, S. Rudin, C. Vael, H. Goossens, C. Konstabel, I. KlareS.R. Nallapareddy, W. Huang, and B. E. Murray. 2003. A potential virulencegene, hylEfm, predominates in Enterococcus faecium of clinical origin. J. In-fect. Dis. 187:508–512.

28. Rice, L. B., V. Lakticov, L. Carias, S. Rudin, R. Hutton, and S. H. Marshall.2009. Transferable capacity for gastrointestinal colonization in Enterococcusfaecium in a mouse model. J. Infect. Dis. 199:342–349.

29. Ruiz-Garbajosa, P., M. J. Bonten, D. A. Robinson, J. Top, S. R. Nallapa-reddy, C. Torres, T. M. Coque, R. Canton, F. Baquero, B. E. Murray, R. delCampo, and R. J. Willems. 2006. Multilocus sequence typing scheme forEnterococcus faecalis reveals hospital-adapted genetic complexes in a back-ground of high rates of recombination. J. Clin. Microbiol. 44:2220–2228.

30. Schneewind, O., P. Model, and V. A. Fischetti. 1992. Sorting of protein A tothe staphylococcal cell wall. Cell 70:267–281.

31. Sillanpaa, J., S. R. Nallapareddy, V. P. Prakash, X. Qin, M. Hook, G. M.Weinstock, and B. E. Murray. 2008. Identification and phenotypic charac-terization of a second collagen adhesin, Scm, and genome-based identifica-tion and analysis of 13 other predicted MSCRAMMs, including four distinctpilus loci, in Enterococcus faecium. Microbiology 154:3199–3211.

32. Sillanpaa, J., V. P. Prakash, S. R. Nallapareddy, and B. E. Murray. 2009.Distribution of genes encoding MSCRAMMs and pili in clinical and naturalpopulations of Enterococcus faecium. J. Clin. Microbiol. 47:896–901.

33. Singh, K. V., T. M. Coque, G. M. Weinstock, and B. E. Murray. 1998. In vivotesting of an Enterococcus faecalis efaA mutant and use of efaA homologs forspecies identification. FEMS Immunol. Med. Microbiol. 21:323–331.

34. Singh, K. V., R. J. Lewis, and B. E. Murray. 2009. Importance of the epalocus of Enterococcus faecalis OG1RF in a mouse model of ascending uri-nary tract infection. J. Infect. Dis. 200:417–420.

35. Tenover, F. C., R. D. Arbeit, R. V. Goering, P. A. Mickelsen, B. E. Murray,D. H. Persing, and B. Swaminathan. 1995. Interpreting chromosomal DNArestriction patterns produced by pulsed-field gel electrophoresis: criteria forbacterial strain typing. J. Clin. Microbiol. 33:2233–2239.

36. Ton-That, H., and O. Schneewind. 2003. Assembly of pili on the surface ofCorynebacterium diphtheriae. Mol. Microbiol. 50:1429–1438.

37. Treitman, A. N., P. R. Yarnold, J. Warren, and G. A. Noskin. 2005. Emergingincidence of Enterococcus faecium among hospital isolates (1993 to 2002).J. Clin. Microbiol. 43:462–463.

38. Willems, R. J., W. Homan, J. Top, M. van Santen-Verheuvel, D. Tribe, X.Manzioros, C. Gaillard, C. M. Vandenbroucke-Grauls, E. M. Mascini, E.van Kregten, J. D. van Embden, and M. J. Bonten. 2001. Variant esp gene asa marker of a distinct genetic lineage of vancomycin-resistant Enterococcusfaecium spreading in hospitals. Lancet 357:853–855.

39. Willems, R. J., J. Top, M. van Santen, D. A. Robinson, T. M. Coque, F.Baquero, H. Grundmann, and J. M. Bonten. 2005. Global spread of vanco-mycin resistant Enterococcus faecium from distinct nosocomial genetic com-plex. Emerg. Infect. Dis. 11:821–828.

VOL. 48, 2010 ENTEROCOCCI IN LATIN AMERICAN HOSPITALS 1569