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

Abstract

Coagulase-negative staphylococci (CNS) are important causes of infective endocarditis (IE), but their microbiological profiles are poorly described. We performed DNA target sequencing and susceptibility testing for 91 patients with definite CNS IE who were identified from the International Collaboration on Endocarditis—Microbiology, a large, multicenter, multinational consortium. A hierarchy of gene sequences demonstrated great genetic diversity within CNS from patients with definite endocarditis that represented diverse geographic regions. In particular, rpoB sequence data demonstrated unique genetic signatures with the potential to serve as an important tool for global surveillance.

Coagulase-negative staphylococci (CNS) are increasingly important causes of community- and health care-associated infective endocarditis (7, 12, 17). Although more than 40 species make up this heterogeneous group of microorganisms, identification of CNS to the species level often is not performed because of laboratory uncertainty about its clinical relevance or the absence of reliable identification systems (10). There is, however, growing evidence that identification of CNS to the species level may alter diagnostic and therapeutic clinical decision making where specific species have unique virulence factors (e.g., Staphylococcus lugdunensis) (2) or unusual antibacterial resistance patterns (e.g., glycopeptide resistance with S. epidermidis and S. haemolyticus) (4-6). In recent years, partial 16S rRNA gene sequencing has emerged as an accurate and reliable method to identify CNS, but this molecular target is limited by having less than 1% sequence divergence among some CNS species. Alternative gene targets such as tuf (elongation factor Tu) (10) and rpoB (RNA polymerase β subunit) (9, 14) have been evaluated, but to our knowledge, no studies have applied gene sequencing of these targets from patients with definite CNS endocarditis.

Given the limited frequency of CNS endocarditis in a single institution, the International Collaboration on Endocarditis—Microbiology (ICE-Micro), a large, multicenter, multinational consortium, provided a unique opportunity to improve our understanding of the spectrum of CNS microorganisms implicated in prosthetic and native valve endocarditis. We performed gene sequencing with multiple DNA targets to identify CNS from patients with definite endocarditis to the species level. We also evaluated the potential for a hierarchy of sequence data to provide greater specificity for species identification, serve as an epidemiologic tool to assess clonality, and predict antimicrobial resistance.

MATERIALS AND METHODS

CNS isolates from patients with definite endocarditis were submitted by ICE-Micro investigators representing a collection from 18 medical centers in 12 countries. Conventional identification and susceptibility testing were performed at a central laboratory with a commercially available panel processed on the Microscan Walkaway instrument (PC-21; Dade Behring, Deerfield, IL) by a standard laboratory protocol. Rifampin susceptibility testing was performed by E-test (AB Biodisk, Solna, Sweden). Interpretation of antimicrobial susceptibility results was based on Clinical and Laboratory Standards Institute guidelines (8). Template DNA preparation and amplification were performed directly on frozen stocks as previously described (18). Amplification of 16S rRNA, tuf, and rpoB genes was achieved with the following primer pairs: 16S rRNA 5F (5′-TTGGAGAGTTTGATCCTGGCTC-3′) and 1194R (5′-ACGTCATCCCCACCTTCCTC-3′), tuf Tseq271 (5′-AAYATGATIACIGGIGCIGCICARATGGA-3′) and Tseq1138 (5′-CCIACIGTICKICCRCCYTCRCG-3′) (13), and rpoB 2491F (5′-AACCAATTCCGTATIGGTTT-3′) and 3241R (5′-GCIACITGITCCATACCTGT-3′) (9).

PCR products were bidirectionally sequenced with original amplification primers, and sequences were compared to related sequences in SmartGene IDNS-Bacteria (SmartGene Inc., Raleigh, NC) by a standard laboratory protocol (18). Nucleotide and amino acid sequence alignments and phylogenetic trees were constructed by the neighbor-joining method with Kimura's two-parameter distance correction model and 1,000 bootstrap replications in the MEGA version 3.1 software package (11). Rifampin resistance was evaluated by using the rifampin resistance-determining region of the rpoB gene as described by Murphy et al. (15) for all strains of CNS.

RESULTS

Ninety-one isolates represented 91 patients with definite endocarditis (prosthetic [n = 36], native [n = 34], and other [n = 21]) with median and mean ages of 68 and 63 years, respectively. The numbers of infections classified as health care or community associated were similar. With phylogenetic analysis as the “gold standard” for final identification, S. epidermidis (n = 65) was the most common 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 agreed with gene sequencing results for 79 (87%) of 91 isolates (Table 1). Phenotypic methods misidentified six S. epidermidis isolates as 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. epidermidis as S. aureus and S. pasteuri as S. warneri. The characterization of 91 CNS isolates by three genetic targets is summarized in Fig. 1. Overall, diversity between species in decreasing order was found with rpoB, tuf, and 16S sequences. Neighbor-joining dendrograms for all gene targets clearly showed distinct clusters for all of the species, with the tuf and rpoB genes having more intraspecies variability. No groups with five or more isolates having identical rpoB sequences clustered within specific institutions or geographic regions including the distinct group of S. epidermidis sequences. No association was observed between the type of valve and the distinct S. epidermidis clusters (data not shown). The distribution of antimicrobial susceptibilities for each species of CNS is described in Table 2. Notably, 42% of the S. epidermidis and 100% of the S. lugdunensis isolates tested were susceptible to oxacillin. Susceptibility patterns did not appear to vary with geographic distribution for S. epidermidis (data not shown). When rpoB amino acid sequences were analyzed for markers of rifampin resistance in CNS isolates, we identified three distinct sequences resulting from four amino acid alterations among all eight rifampin-resistant isolates (Table 3). All rifampin-resistant isolates were S. epidermidis (MIC, >32 μg/ml), each having two unique amino acid substitutions within the rpoB gene that are known to confer rifampin resistance on S. aureus. Six of these eight isolates had D471E and I527M, one had H481N and I527M, and one had D471Y and H481N amino acid substitutions. No unique amino acid substitutions were observed for non-S. epidermidis CNS species.

TABLE 1.

Identification by conventional method and gene sequencing

Final identification by phylogenetic resolution (no. of isolates)	Identity according to following test method:
	Conventionala	16S rRNA	tuf	rpoB
S. epidermidis (59)	S. epidermidis	S. epidermidis	S. epidermidis	S. epidermidis
S. epidermidis (3)	S. hominissubsp.hominis	S. epidermidis	S. epidermidis	S. epidermidis
S. epidermidis (1)	S. capitissubsp.urealyticus	S. epidermidis	S. epidermidis	S. epidermidis
S. epidermidis (1)	S. hyicus	S. epidermidis	S. epidermidis	S. epidermidis
S. epidermidis (1)	S. warneri	S. epidermidis	S. epidermidis	S. epidermidis
S. lugdunensis (8)	S. lugdunensis	S. lugdunensis	S. lugdunensis	S. lugdunensis
S. hominis (6)	S. hominis	S. hominis/xylosus	S. hominis	S. hominis
S. haemolyticus (3)	S. haemolyticus	S. haemolyticus	S. haemolyticus	S. haemolyticus
S. capitis (5)	S. capitis	S. capitis/caprae/arlettae/epidermidis	S. capitis	S. capitis
S. schleiferi (1)	S. schleiferi	S. schleiferi	Staphylococcus sp.	S. schleiferi
S. aureus (1)	S. epidermidis	S. aureus	S. aureus	S. aureus
S. pasteuri (1)	S. warneri	S. pasteuri/aureus	Staphylococcus sp.	S. pasteuri
S. warneri (1)	S. warneri	S. warneri/pasteuri	S. warneri	S. warneri

^aBoldface type indicates misidentification by conventional methods.

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.

TABLE 2.

Distribution of antimicrobial susceptibilities of CNS isolates

Antibacterial agent and MIC (μg/ml)a	No. of isolates identified by rpoB sequence as:
	S. epidermidis (n = 65)	S. lugdunensis (n = 8)	S. hominis (n = 6)	S. capitis (n = 5)	S. haemolyticus (n = 3)	S. pasteuri (n = 1)	S. schleiferi (n = 1)	S. warneri (n = 1)
Oxacillin
≤0.25	26	5	2	3	1	0	1	0
0.5b	2	3	1	0	0	0	0	0
1	1	0	0	0	0	0	0	0
≥2	36	0	3	2	2	1	0	1
Vancomycin
≤1	27	8	4	4	2	1	0	0
2	38	0	2	1	1	0	1	1
Linezolid
≤0.5	46	7	0	3	1	0	0	0
1	17	1	6	1	2	1	1	1
2	2	0	0	1	0	0	0	0
Daptomycin, ≤1	65	8	6	5	3	1	1	1
Rifampin
≤1	57	8	6	5	3	1	1	1
2	0	0	0	0	0	0	0	0
≥4	8	0	0	0	0	0	0	0

^aBreakpoints for susceptibility are those defined by the Clinical and Laboratory Standards Institute (8). An MIC in boldface type represents the breakpoint for resistance.

^bThe breakpoint for S. lugdunensis oxacillin resistance is ≥4 μg/ml.

TABLE 3.

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

No. of isolates	Nucleotide change(s) in rpoB	Amino acid alteration	Rifampin resistance (MIC [μg/ml])
55	None	None	Susceptible (<1)
1	GCT → GAT	A534 → D	Susceptible (<1)
1	CGT → CAT	R503 → H	Susceptible (<1)
1	GAT → TAC	D471 → Yb	Resistant (>32)
	CAC → AAC	H481 → Nb
1	CAC → AAC	H481 → Nb	Resistant (>32)
	ATA → ATG	I527 → Mb
6	GAT → GAA	D471 → Eb	Resistant (>32)
	ATA → ATG	I527 → Mb

^aAmino acid sequences of the rpoB gene were compared to GenBank sequence NP_763861 (ATCC 12228).

^bRegions known to be involved with resistance to rifampin (15).

DISCUSSION

The specimen repository of the ICE-Micro multinational consortium provided us with a unique opportunity to examine CNS isolates associated with invasive disease. To our knowledge, this study is the first to fully characterize the species distribution and susceptibility patterns of CNS isolates from patients with definite endocarditis. Sequencing of the rpoB gene served as a robust target for identification to the species level, suggested an absence of clonality in strains causing S. epidermidis endocarditis, and identified high-level rifampin resistance in S. epidermidis isolates.

Historically, investigators have relied on conventional methods for the identification and susceptibility testing of CNS; this may have served as a barrier to the full appreciation of the epidemiology of CNS disease. In fact, previous reviews of CNS endocarditis have discussed the spectrum of disease from the perspective of phenotypic identifications with limited attention to specific species (17, 22). We found S. epidermidis as the most common CNS species, an observation corroborated by previous reports on non-endocarditis patients with invasive CNS disease (1, 3, 10, 16, 19). The rank order of non-S. epidermidis infections differs in our report from that in others. We report S. lugdunensis as the second most common pathogen, whereas non-endocarditis studies have observed S. hominis, S. haemolyticus, and S. capitis as the next most commonly encountered CNS pathogens (1, 19, 21).

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

Accurate isolate identification to the species level is also important for examining susceptibility patterns and alerting clinicians to those species with known increased virulence. Although we did not observe reduced susceptibilities to glycopeptides in S. epidermidis and S. haemolyticus, these species reportedly exhibit heterogeneous susceptibility to glycopeptides (4-6) and their identification may heighten clinician awareness of therapeutic failures of vancomycin therapy. Similarly, identification of S. lugdunensis may alter the diagnostic approach because this species is considered to be more susceptible to beta-lactam agents, more virulent, and associated with a higher mortality compared with other CNS species (2). Finally, we report the first use of the rifampin resistance-determining region of the rpoB gene to predict rifampin resistance in isolates identified as S. epidermidis. Given the need for rifampin in patients with CNS prosthetic valve endocarditis, a reliable method to identify locations known to confer rifampin resistance by single-step mutations may be clinically important.

In conclusion, the ICE-Micro consortium enabled us to evaluate a large number of CNS isolates from patients with definite endocarditis from diverse geographic regions with a hierarchy of gene sequence data. From this extensive global repository, we provide novel and valuable information about the genetic diversity of CNS species that cause endocarditis, suggesting that strains causing CNS endocarditis have unique genetic signatures which are found across vast geographic distances. The clinical significance of more accurate identification to the species level remains to be defined, but rpoB sequence analysis may serve as a useful tool for surveillance and may improve our understanding of the host-pathogen relationships in native and prosthetic valve endocarditis.