Genotypic Diversity of Coagulase-Negative Staphylococci Causing Endocarditis: a Global Perspective

Coagulase-negative staphylococci (CNS) are important causesof infective endocarditis (IE), but their microbiological profilesare poorly described. We performed DNA target sequencing andsusceptibility testing for 91 patients with definite CNS IEwho were identified from the International Collaboration onEndocarditis—Microbiology, a large, multicenter, multinationalconsortium. A hierarchy of gene sequences demonstrated greatgenetic diversity within CNS from patients with definite endocarditisthat represented diverse geographic regions. In particular,rpoB sequence data demonstrated unique genetic signatures withthe potential to serve as an important tool for global surveillance.

INTRODUCTION

Top

INTRODUCTION

Coagulase-negative staphylococci (CNS) are increasingly importantcauses of community- and health care-associated infective endocarditis(7, 12, 17). Although more than 40 species make up this heterogeneousgroup of microorganisms, identification of CNS to the specieslevel often is not performed because of laboratory uncertaintyabout its clinical relevance or the absence of reliable identificationsystems (10). There is, however, growing evidence that identificationof CNS to the species level may alter diagnostic and therapeuticclinical decision making where specific species have uniquevirulence factors (e.g., Staphylococcus lugdunensis) (2) orunusual antibacterial resistance patterns (e.g., glycopeptideresistance with S. epidermidis and S. haemolyticus) (4-6). Inrecent years, partial 16S rRNA gene sequencing has emerged asan accurate and reliable method to identify CNS, but this moleculartarget is limited by having less than 1% sequence divergenceamong some CNS species. Alternative gene targets such as tuf(elongation factor Tu) (10) and rpoB (RNA polymerase βsubunit) (9, 14) have been evaluated, but to our knowledge,no studies have applied gene sequencing of these targets frompatients with definite CNS endocarditis.

Given the limited frequency of CNS endocarditis in a singleinstitution, the International Collaboration on Endocarditis—Microbiology(ICE-Micro), a large, multicenter, multinational consortium,provided a unique opportunity to improve our understanding ofthe spectrum of CNS microorganisms implicated in prostheticand native valve endocarditis. We performed gene sequencingwith multiple DNA targets to identify CNS from patients withdefinite endocarditis to the species level. We also evaluatedthe potential for a hierarchy of sequence data to provide greaterspecificity for species identification, serve as an epidemiologictool to assess clonality, and predict antimicrobial resistance.

MATERIALS AND METHODS

Top

MATERIALS AND METHODS

CNS isolates from patients with definite endocarditis were submittedby ICE-Micro investigators representing a collection from 18medical centers in 12 countries. Conventional identificationand susceptibility testing were performed at a central laboratorywith a commercially available panel processed on the MicroscanWalkaway instrument (PC-21; Dade Behring, Deerfield, IL) bya standard laboratory protocol. Rifampin susceptibility testingwas performed by E-test (AB Biodisk, Solna, Sweden). Interpretationof antimicrobial susceptibility results was based on Clinicaland Laboratory Standards Institute guidelines (8). TemplateDNA preparation and amplification were performed directly onfrozen stocks as previously described (18). Amplification of16S rRNA, tuf, and rpoB genes was achieved with the followingprimer pairs: 16S rRNA 5F (5'-TTGGAGAGTTTGATCCTGGCTC-3') and1194R (5'-ACGTCATCCCCACCTTCCTC-3'), tuf Tseq271 (5'-AAYATGATIACIGGIGCIGCICARATGGA-3')and Tseq1138 (5'-CCIACIGTICKICCRCCYTCRCG-3') (13), and rpoB2491F (5'-AACCAATTCCGTATIGGTTT-3') and 3241R (5'-GCIACITGITCCATACCTGT-3')(9).

PCR products were bidirectionally sequenced with original amplificationprimers, and sequences were compared to related sequences inSmartGene IDNS-Bacteria (SmartGene Inc., Raleigh, NC) by a standardlaboratory protocol (18). Nucleotide and amino acid sequencealignments and phylogenetic trees were constructed by the neighbor-joiningmethod with Kimura's two-parameter distance correction modeland 1,000 bootstrap replications in the MEGA version 3.1 softwarepackage (11). Rifampin resistance was evaluated by using therifampin resistance-determining region of the rpoB gene as describedby Murphy et al. (15) for all strains of CNS.

RESULTS

Top

RESULTS

Ninety-one isolates represented 91 patients with definite endocarditis(prosthetic [n = 36], native [n = 34], and other [n = 21]) withmedian and mean ages of 68 and 63 years, respectively. The numbersof infections classified as health care or community associatedwere similar. With phylogenetic analysis as the "gold standard"for final identification, S. epidermidis (n = 65) was the mostcommon CNS identified, followed by S. lugdunensis (n = 8), S.hominis (n = 6), S. capitis (n = 5), S. haemolyticus (n = 3),and other (n = 3). Identification by conventional methods agreedwith gene sequencing results for 79 (87%) of 91 isolates (Table1). Phenotypic methods misidentified six S. epidermidis isolatesas S. hominis subsp. hominis (three), S. capitis subsp. urealyticus(one), S. hyicus (one), and S. warneri (one). Additionally,phenotypic methods misidentified one isolate each of S. epidermidisas S. aureus and S. pasteuri as S. warneri. The characterizationof 91 CNS isolates by three genetic targets is summarized inFig. 1. Overall, diversity between species in decreasing orderwas found with rpoB, tuf, and 16S sequences. Neighbor-joiningdendrograms for all gene targets clearly showed distinct clustersfor all of the species, with the tuf and rpoB genes having moreintraspecies variability. No groups with five or more isolateshaving identical rpoB sequences clustered within specific institutionsor geographic regions including the distinct group of S. epidermidissequences. No association was observed between the type of valveand the distinct S. epidermidis clusters (data not shown). Thedistribution of antimicrobial susceptibilities for each speciesof CNS is described in Table 2. Notably, 42% of the S. epidermidisand 100% of the S. lugdunensis isolates tested were susceptibleto oxacillin. Susceptibility patterns did not appear to varywith geographic distribution for S. epidermidis (data not shown).When rpoB amino acid sequences were analyzed for markers ofrifampin resistance in CNS isolates, we identified three distinctsequences resulting from four amino acid alterations among alleight rifampin-resistant isolates (Table 3). All rifampin-resistantisolates were S. epidermidis (MIC, >32 µg/ml), eachhaving two unique amino acid substitutions within the rpoB genethat are known to confer rifampin resistance on S. aureus. Sixof these eight isolates had D471E and I527M, one had H481N andI527M, and one had D471Y and H481N amino acid substitutions.No unique amino acid substitutions were observed for non-S.epidermidis CNS species.

View this table:
[in this window]
[in a new window]

TABLE 1. Identification by conventional method and gene sequencing

View larger version (41K):
[in this window]
[in a new window]

FIG. 1. Neighbor-joining radial dendrograms of the 16S rRNA (A), tuf (B), and rpoB (C) genes of CNS isolates from patients with endocarditis. Each entry represents a unique sequence among the study isolates. The number of isolates sharing 100% identity with the representative sequence is noted. Intraspecies variability is recorded as the percent difference between isolates of the same species, as well as the number of base pair positions in the interrogated sequence with instability. To the side of each tree, the relative distance from each species to its nearest neighbor is recorded in terms of percent distance. A superscript letter a indicates two isolates identified by Microscan as S. hominis subsp. novobiosepticus that had unique tuf and rpoB sequences which differed by at least 4 bp (0.6%) and 2 bp (0.3%) from the other S. hominis sequences, respectively. A superscript letter b indicates rifampin-resistant isolates.

View this table:
[in this window]
[in a new window]

TABLE 2. Distribution of antimicrobial susceptibilities of CNS isolates

View this table:
[in this window]
[in a new window]

TABLE 3. Variability of the rpoB amino acid sequence for S. epidermidis isolates^a

DISCUSSION

Top

DISCUSSION

The specimen repository of the ICE-Micro multinational consortiumprovided us with a unique opportunity to examine CNS isolatesassociated with invasive disease. To our knowledge, this studyis the first to fully characterize the species distributionand susceptibility patterns of CNS isolates from patients withdefinite endocarditis. Sequencing of the rpoB gene served asa robust target for identification to the species level, suggestedan absence of clonality in strains causing S. epidermidis endocarditis,and identified high-level rifampin resistance in S. epidermidisisolates.

Historically, investigators have relied on conventional methodsfor the identification and susceptibility testing of CNS; thismay have served as a barrier to the full appreciation of theepidemiology of CNS disease. In fact, previous reviews of CNSendocarditis have discussed the spectrum of disease from theperspective of phenotypic identifications with limited attentionto specific species (17, 22). We found S. epidermidis as themost common CNS species, an observation corroborated by previousreports on non-endocarditis patients with invasive CNS disease(1, 3, 10, 16, 19). The rank order of non-S. epidermidis infectionsdiffers in our report from that in others. We report S. lugdunensisas the second most common pathogen, whereas non-endocarditisstudies have observed S. hominis, S. haemolyticus, and S. capitisas the next most commonly encountered CNS pathogens (1, 19,21).

The clinical implications of more accurate identification ofCNS by gene sequencing are not fully known. In our study, weobserved that the rpoB gene serves as a reliable indicator ofgenetic diversity, which may be helpful as an epidemiologicaltool to distinguish multiple CNS strains. Also, fast identificationof two distinct CNS strains may prove useful when interpretingthe clinical significance of blood cultures in patients withintracardiac devices. With our limited data set, we did notobserve distinct regional variations among clusters of CNS isolates;however, the ability to rapidly identify isolates to the speciesand subspecies levels may prove valuable for monitoring thedissemination of unusual strains between and within institutions.Only by gene sequencing were we able to identify and now reportthe first case of S. pasteuri endocarditis. Additionally, rpoBgene sequencing affords greater specificity by providing a uniquegenetic signature for CNS species that may have implicationsfor global surveillance. While multilocus sequence typing schemeshave emerged as important tools to assess clonal complexes forS. epidermidis, the optimal discriminatory loci to serve asthe gold standard have not been firmly established (20). Theuse of the rpoB gene shows promise as a marker for unique clones,but its use as a sole epidemiological target warrants furtherstudy by multilocus sequence typing.

Accurate isolate identification to the species level is alsoimportant for examining susceptibility patterns and alertingclinicians to those species with known increased virulence.Although we did not observe reduced susceptibilities to glycopeptidesin S. epidermidis and S. haemolyticus, these species reportedlyexhibit heterogeneous susceptibility to glycopeptides (4-6)and their identification may heighten clinician awareness oftherapeutic failures of vancomycin therapy. Similarly, identificationof S. lugdunensis may alter the diagnostic approach becausethis species is considered to be more susceptible to beta-lactamagents, more virulent, and associated with a higher mortalitycompared with other CNS species (2). Finally, we report thefirst use of the rifampin resistance-determining region of therpoB gene to predict rifampin resistance in isolates identifiedas S. epidermidis. Given the need for rifampin in patients withCNS prosthetic valve endocarditis, a reliable method to identifylocations known to confer rifampin resistance by single-stepmutations may be clinically important.

In conclusion, the ICE-Micro consortium enabled us to evaluatea large number of CNS isolates from patients with definite endocarditisfrom diverse geographic regions with a hierarchy of gene sequencedata. From this extensive global repository, we provide noveland valuable information about the genetic diversity of CNSspecies that cause endocarditis, suggesting that strains causingCNS endocarditis have unique genetic signatures which are foundacross vast geographic distances. The clinical significanceof more accurate identification to the species level remainsto be defined, but rpoB sequence analysis may serve as a usefultool for surveillance and may improve our understanding of thehost-pathogen relationships in native and prosthetic valve endocarditis.

ACKNOWLEDGMENTS

There are no conflicts to declare.

ICE-Micro receives support from Cubist Pharmaceuticals (C.W.W.).

FOOTNOTES

* Corresponding author. Mailing address: University of Utah School of Medicine, ARUP Laboratories, 500 Chipeta Way, Salt Lake City, UT 84108. Phone: (801) 583-2787. Fax: (801) 584-5207. E-mail: cathy.petti{at}aruplab.com

Published ahead of print on 26 March 2008.

REFERENCES

Top