Comparison of Genotypic and Phenotypic Methods for Species-Level Identification of Clinical Isolates of Coagulase-Negative Staphylococci
http://www.100md.com
微生物临床杂志 2005年第5期
Eijkman-Winkler Institute, University Medical Center Utrecht, Utrecht, The Netherlands
ABSTRACT
To compare commonly used phenotypic methods with genotypic identification methods 47 clinical isolates of coagulase-negative staphylococci (CONS), 10 CONS ATCC strains, and a Staphylococcus aureus clinical isolate were identified using the API Staph ID test, BD Phoenix Automated Microbiology System, and 16S rRNA gene and tuf gene sequencing. When necessary part of the sodA gene was sequenced for definitive identification. The results show that tuf gene sequencing is the best method for identification of CONS, but the API Staph ID test is a reasonably reliable phenotypic alternative. The performance of the BD Phoenix Automated Microbiology System for identification of CONS is poor. The present study also showed that although genotypic methods are clearly superior to phenotypic identifications, a drawback of sequence-based genotypic methods may be a lack of quality of deposited sequences in data banks. In particular, 16S rRNA gene sequencing suffers from the lack of high quality among sequences deposited in GenBank. Furthermore, genotypic identification based on 16S rRNA sequences has limited discriminating power for closely related Staphylococcus species. We propose partial sequencing of the tuf gene as a reliable and reproducible method for identification of CONS species.
INTRODUCTION
The human skin and mucous membranes represent a diverse environment of bacteria, the normal microfloras (5). Probably the most important bacteria of this microflora are members of the genus Staphylococcus. The genus Staphylococcus is currently divided in 38 species and 17 subspecies, half of which are indigenous to humans (13). Staphylococci generally have a benign or symbiotic relationship with their host. However, they may develop into a pathogen if they gain entry into the host tissue through trauma of the cutaneous barrier, inoculation by needles, or implantation of medical devices.
In last two decades, coagulase-negative staphylococci (CONS) have also emerged as significant pathogens, especially in immunocompromised patients, premature newborns, and patients with implanted biomaterials. The most frequently encountered CONS species associated with human infections is Staphylococcus epidermidis, in particular in association with intravascular catheters. In addition, S. epidermidis is the predominant agent of nosocomial bacteremia, prosthetic-valve endocarditis, surgical wounds, central nervous system shunt infections, intravascular catheter-related infections, peritoneal dialysis-related infections, and infections of prosthetic joints (5, 10, 19). The second most frequently encountered CONS species is S. haemolyticus. S. haemolyticus has been implicated in native-valve endocarditis, septicemia, peritonitis, and wound, bone, and joint infections (5, 10, 19). Other CONS species are involved in a variety of infections. For example, S. saprophyticus is an important pathogen in human urinary tract infections, especially in young, sexually active females, and S. lugdunensis has been implicated in arthritis, catheter infections, and prosthetic joint infections (5, 10, 19).
Because of the increasing clinical significance of CONS, accurate species identification of CONS is highly desirable to permit a more precise determination of host-pathogen relationship of CONS. Phenotypic identification of CONS appears to be unsatisfactory, unreliable, and irreproducible. Commercial identification systems and automated systems are not able to make a reliable distinction between the different species of CONS because of the variable expression of the phenotypic characters (1, 2, 3, 7, 11, 18). Other tests, such as enzyme electrophoresis or analysis of cellular fatty acid composition, have also failed to make a reliable identification (20). Therefore, methods based on sequencing of PCR amplicons were evaluated and compared to phenotypic identifications of CONS. A number of PCR amplicon-sequencing-based methods for identification of CONS have been reported, i.e., targeting the 16S rRNA, sodA, and tuf genes (8, 15, 17). Multiple copies of the 16S rRNA gene are present on the chromosome of most bacteria. The sequence variation between the different copies of the 16S rRNA gene seems to be relatively low (4, 16). The sodA gene encodes the manganese-dependent superoxide dismutase. This metalloprotein inactivates harmful superoxide radicals (12, 22). The tuf gene, which encodes the elongation factor Tu, is involved in peptide chain formation and is part of the ribosome (9). These genes are essential constituents of the bacterial genome and are therefore preferred for diagnostic purposes.
The aim of the present study was to compare commonly used phenotypic methods with genotypic identification methods for clinical CONS isolates. For that purpose, 57 CONS isolates were phenotypically identified by means of the API Staph ID test and the BD Phoenix Automated Microbiology System. Furthermore, the same 57 CONS isolates were genotypically identified by sequencing of PCR amplicons based on sequences of Staphylococcus 16S rRNA, tuf, and sodA genes.
MATERIALS AND METHODS
Bacterial isolates. The bacterial isolates used in this study consisted of 10 ATCC reference strains of various CONS species, 47 clinical CONS isolates, and an S. aureus clinical isolate. The reference strains are listed in Table 1 and were obtained from the American Type Culture Collection (ATCC). The clinical isolates were a random selection from a collection of blood isolates from the neonatal intensive care unit of the University Medical Center of Utrecht and have been described previously (14). The organisms were identified by Gram staining, catalase, coagulase, and DNase tests. All strains were grown overnight on blood agar plates at 37°C.
Phenotypic identification. Phenotypic identification of bacterial isolates was achieved by using the API Staph ID test (BioMerieux) and the BD Phoenix Automated Microbiology System (Becton Dickinson, software version 4.01). For the API Staph ID test, results were interpreted using the API LAB ID computer software. Both biochemical tests were performed according to the manufacturers' instructions.
Extraction of genomic DNA. After overnight growth on blood agar plates at 37°C, genomic DNA of pure cultures was extracted using a Dneasy Tissue Kit (QIAGEN) according to the manufacturer's instructions. Lysates were frozen at –70°C until they were used for PCR.
PCR amplification. The sequences of the primers used in this study are shown in Table 2. All primers were purchased from Invitrogen. The PCR mixtures contained 10 μl PCR buffer (10x, as supplied with Taq polymerase), 8 μl deoxynucleoside triphosphates (200 μM concentrations of each), 2 μl of genomic DNA (120 ng/ml), 0.2 μl Taq DNA polymerase (0.2 U) and 2.5 μl of each of the primers (20 μM) in a final volume made up to 100 μl with water. The thermal cycling conditions were 5 min at 95°C for 1 cycle, followed by 30 cycles of 1 min at 95°C, 1 min at 55°C, and 1 min at 72°C. The last cycle was performed at 10 min at 72°C. Amplification of a 1,500-bp-long fragment of the 16S rRNA gene was achieved by PCR from genomic DNA with the forward primer Epsilon F and the reverse primer 1510R. These primers have been described previously (6). For tuf and sodA, new primers were designed. tuf and sodA sequences were obtained from GenBank. Multiple sequence alignments were carried out by using the program Align Plus of the Scientific & Educational software (version 3). Based on these sequence alignments, regions of the tuf and sodA genes highly conserved among CONS were chosen. PCR primers were designed from these regions. Primer pair tuf-F and tuf-R was designed to amplify a 412-bp-long fragment of the tuf gene. Primer pair sodA-F and sodA-R was designed to amplify a 236-bp-long fragment of the sodA gene. The PCR conditions were the same as described before except for the amount of the primers (4 μl of the primers tuf-F and tuf-R [10 μM] and 8 μl of the primers sodA-F and sodA-R [10 μM]) and the annealing temperature (47°C for sodA). After PCR amplification 10 μl of each reaction mixture was analyzed on a 1.2% agarose gel. The DNA fragments were visualized and photographed under UV light after ethidium bromide staining. The PCR products were purified using the QIAquick PCR Purification Kit protocol (QIAGEN) according to the manufacturer's instructions.
DNA sequencing. Forward and reverse sequencing of a 320-bp-long part of the 16S rRNA amplicon and forward and reverse sequencing of a 412-bp-long fragment of the tuf gene and a 236-bp-long fragment of the sodA gene were obtained by using the forward primer Epsilon F, the reverse primer 320R, the forward primer tuf-F, the reverse primer tuf-R, the forward primer sodA-F, and the reverse primer sodA-R, respectively. The reverse primer 320R has been described previously (6). Sequencing reactions were carried out with the reagents of a BigDye Terminator v1.1 cycle sequencing ready reaction kit (QIAGEN). All PCR sequence reaction mixtures contained in the final concentration 4 μl Ready Reaction mix (2.5x), 2 μl BigDye sequencing buffer (5x), 2 μl template (purified PCR product), and 4 μl primer (10 ng/μl), in a final volume made up to 20 μl with water. The thermal cycling conditions were 1 min at 96°C for 1 cycle followed by 25 cycles of 10 s at 96°C, 5 s at 50°C, and 4 min at 60°C. The PCR sequence products were purified using the BigDye Terminator PCR Purification Kit protocol (QIAGEN) according to the manufacturer's instructions. The PCR sequence products were separated by electrophoresis on a 0.2-mm 4% polyacrylamide denaturing gel and recorded with an ABI 377 DNA sequencer (Applied Biosystems). The results were processed into sequence data with the sequence ABI Prism 377 collection software. To exclude sequencing mistakes, the forward nucleotide sequences were aligned with the reverse nucleotide sequences by using the program Align Plus. For identification the 16S rRNA, tuf, and sodA partial sequences were compared with sequences from GenBank. The partial tuf gene sequence for S. sciuri and the four S. xylosus isolates are deposited in GenBank (accession numbers AY763434 to AY763438, respectively).
Phylogenetic comparisons. Phylogenetic trees based on the 16S rRNApart and tufpart sequences were generated using the program Data Analysis in Molecular Biology and Evolution. The results were processed into a tree with the neighbor-joining method. All trees were resampled with 1,000 bootstrap replications to test the robustness of the data. Pairwise comparisons of the 16S rRNApart and tufpart sequences for homology were obtained using the program Align Plus.
RESULTS
Phenotypic and genotypic identification of CONS. A total of 10 CONS ATCC strains and 47 clinical CONS isolates and an S. aureus clinical isolate were identified using the API Staph ID test, the BD Phoenix Automated Microbiology System, and 16S rRNA gene and tuf gene sequencing. When no discrimination could be made based on 16S rRNA and tuf sequences, part of the sodA gene was sequenced for definitive identification. All PCR products generated for sequence analysis were of the expected size, i.e., 1,500 bp, 412 bp, and 236 bp for 16S rRNA, tuf, and sodA, respectively (data not shown). Partial sequencing of the 16S rRNA amplicons and sequencing of the tuf and sodA amplicons resulted in diagrams devoid of overlapping peaks, which confirmed that these strains contain a single type of 16S rRNA, tuf, and sodA gene. For identification the 16S rRNA, tuf, and sodA sequences were compared with sequences deposited in GenBank. Homology values above 97% were considered reliable. However, in some cases more than one species fulfilled this criterion. Definitive species identification was achieved on the basis of the combined sequence data for the 16S rRNA, tuf, and sodA genes (Table 3). The ATCC strains were identified as one S. capitis, one S. sciuri, three S. epidermidis, and four S. xylosus strains and one S. saprophyticus strain, in agreement with the naming by the ATCC. The clinical isolates consisted of one S. aureus, 31 S. epidermidis, six S. haemolyticus, five S. capitis, two S. warneri, and two S. hominis isolates and one S. cohnii isolate. The API Staph ID test misidentified seven clinical isolates. All S. capitis isolates were identified as S. epidermidis. Both S. hominis isolates were identified as S. lugdunensis. The Phoenix system misidentified 17 clinical isolates and was not able to identify 2 clinical isolates. All misidentified clinical isolates were S. epidermidis isolates, which were identified as members of six different species (S. warneri, S. capitis, S. hominis, S. simulans, S. saprophyticus, and S. chromogenes). Three ATCC strains were misidentified by the Phoenix system. One S. xylosus ATCC strain was named as S. simulans, and two S. epidermidis ATCC strains were named as S. schleiferi or S. saprophyticus.
16S rRNA gene sequencing showed correct identification except for the six S. capitis isolates, which were named as either S. epidermidis or S. caprae. Both S. hominis isolates and the S. cohnii isolate were identified as either S. hominis or S. xylosus and as either S. cohnii or S. saprophyticus, respectively. Similar problems occurred for two ATCC strains. One S. xylosus ATCC strain was identified as either S. xylosus or S. saprophyticus, and the S. saprophyticus ATCC strain was identified as S. xylosus. tuf gene sequencing showed correct identification except for five ATCC strains. The S. xylosus ATCC strains were named as S. saprophyticus and the S. sciuri ATCC strain was named as S. capitis. In these cases the homology values were below 97% and therefore considered not reliable. sodA gene sequencing agreed with our definitive identification. DNA amplification for the sodA gene of the S. sciuri ATCC strain failed. Therefore, genotypic identification was not possible. An explanation may be that the designed sodA reverse primer has five mismatches within the sodA sequence from the S. sciuri ATCC strain. The sodA reverse primer contains no more than one mismatch within the sodA sequences from the other isolates.
Phylogenetic comparisons. Separate phylogenetic trees, inferred from the neighbor-joining method, were constructed with the sequence data for 16S rRNA and tuf gene (Fig. 1 and 2). Although tuf is well conserved in Staphylococcus, variability was observed in the nucleotide sequences. Four Staphylococcus species (S. saprophyticus, S. xylosus, S. sciuri, and S. cohnii) harbor an insertion of three nucleotides. The inserted amino acid is leucine for S. sciuri and alanine for S. saprophyticus, S. xylosus, and S. cohnii. The insertion of three nucleotides was considered a single event rather than three distinct events. We replaced the insertion with a single mutation of one nucleotide (T for A) to avoid unreliable differences in phylogenetic relationships. On the basis of 16S rRNA sequences, S. epidermidis and S. capitis constitute a group of related species. On the other hand, from the tuf sequence analysis, association of S. warneri with the S. epidermidis group was inferred. In both 16S rRNA and tuf sequence analysis, S. saprophyticus clearly forms a distinct group which includes S. xylosus and S. cohnii with high bootstrap values of 100% (tuf) and 86% (16S rRNA).
Intraspecies sequence divergence was observed in our collection. On the basis of tuf sequences, the intraspecies sequence divergence was 0.5% among 34 S. epidermidis isolates, 0.3% among 6 S. haemolyticus isolates, and 0.8 to 1.1% among 4 S. xylosus isolates. On the basis of 16S rRNA sequences, only intraspecies sequence divergence (0.3 to 1%) among four S. xylosus isolates was observed. Pairwise comparison of the 16S rRNA sequences (data not shown) showed that the most closely related Staphylococcus species were S. saprophyticus and S. xylosus with 99.7% identity. On the other hand, the most divergent species pair was S. aureus and S. cohnii with 92.2% identity. Using tuf sequences the most closely related Staphylococcus species were S. warneri and S. capitis with 97% identity. The most divergent species pair was S. xylosus and S. sciuri with 85.7% identity. Furthermore, pairwise comparison of the tuf sequences revealed that their mean identity (92.6%) is lower than the mean identity (95.9%) of 16S rRNA sequences.
DISCUSSION
Clinically significant CONS should be identified at the species level by a reliable and reproducible method to provide a better understanding of pathogenic potential of various CONS species. Phenotypic methods for identification of CONS appear to be unreliable. The purpose of the present study was to compare commonly used phenotypic methods, the API Staph ID test and the BD Phoenix Automated Microbiology System, with PCR amplicon-sequencing based methods targeting the 16S rRNA, tuf, and sodA genes.
The results of the phenotypic identification methods indicate that the API Staph ID is a much more reliable phenotypic method than the BD Phoenix Automated Microbiology System. The results obtained with the genotypic methods indicate that the PCR amplicon-sequencing-based techniques are specific and sensitive for identification of Staphylococcus species. 16S rRNA gene sequencing misidentified eight isolates (six clinical isolates and two ATCC strains). An explanation for these misidentifications may be that the deposited sequences in GenBank are incorrectly assigned to the various Staphylococcus species.
tuf gene sequencing misidentified five ATCC strains. These misidentifications occurred because tuf gene sequences of the species S. xylosus and S. sciuri have never been deposited in the GenBank database. The S. xylosus ATCC strains and S. sciuri ATCC strain were therefore named after closely related species, S. saprophyticus and S. capitis, respectively. sodA gene sequencing agreed with our definitive identification. No misidentifications were observed. A drawback of sodA gene-based genotypic identification may be the high interspecies sequence divergence, which makes it difficult to find conserved regions to design primers.
The 16S rRNA and tuf sequence-based relationships obtained were in accordance with phylogenetic trees published previously (15, 21). Both the 16S rRNA and tuf sequence-derived trees indicate that S. sciuri is the most distant Staphylococcus species. This result is in agreement with phylogenetic studies performed earlier with 16S rRNA and tuf sequence analysis (15, 21). Furthermore, the tuf sequence-derived tree indicates that S. warneri is associated with the S. epidermidis group, which includes S. capitis. This is in agreement with results earlier described in PCR-sequencing assays targeting the sodA gene (17). Tuf-derived data more often showed intraspecies sequence divergence than the 16S rRNA-derived data. Apparently, the 16S rRNA gene is more highly conserved than the tuf gene. Pairwise comparison of the tuf sequences revealed that their mean identity (92.6%) is lower than the mean identity (95.9%) of 16S rRNA sequences. These results indicate that tuf constitutes a more discriminatory target gene than the 16S rRNA gene to differentiate closely related Staphylococcus species.
In conclusion, the present study demonstrates the superiority of genotypic methods over phenotypic assays for identification of CONS species. tuf gene sequencing is the best method for the identification of CONS, but the API Staph ID test is a reasonably reliable phenotypic alternative. The performance of the BD Phoenix Automated Microbiology System using software version 4.01 for identification of CONS is poor. The present study also shows that although genotypic methods are clearly superior to phenotypic identifications, a drawback of sequence-based genotypic methods may be a lack of quality of deposited sequences in data banks. In particular, 16S rRNA gene sequencing suffers from the lack of high quality among deposited sequences in GenBank. Furthermore, genotypic identification based on 16S rRNA sequences has limited discriminating power for closely related Staphylococcus species. We propose PCR and sequencing of the tuf gene as a reliable and valuable approach for the genotypic identification of CONS species. Furthermore, the tuf sequence polymorphisms have potential for the development of biprobe identification assays based on real-time PCR (6). This method might be a useful diagnostic tool for a reliable and simple identification of clinically relevant CONS species.
ACKNOWLEDGMENTS
We thank D. de Jong for helpful advice on generating phylogenetic trees.
REFERENCES
Bannerman, T. L., K. T. Kleeman, and W. E. Kloos. 1993. Evaluation of the Vitek Systems Gram-Positive Identification card for species identification of coagulase-negative staphylococci. J. Clin. Microbiol. 31:1322-1325.
Birnbaum, D., M. Kelly, and A. W. Chow. 1991. Epidemiologic typing systems for coagulase-negative staphylococci. Infect. Control Hosp. Epidemiol. 12:319-326.
Calvo, J., J. L. Hernandez, M. C. Farinas, D. Garcia-Palomo, and J. Aguero. 2000. Osteomyelitis caused by Staphylococcus schleiferi and evidence for misidentification of this Staphylococcus species by an automated identification system. J. Clin. Microbiol. 38:3887-3889.
Cilia, V., B. Lafay, and R. Christen. 1996. Sequence heterogeneities among 16S ribosomal RNA sequences, and their effect on phylogenetic analyses at the species level. Mol. Biol. Evol. 13:451-461.
Crossley, K. B., and G. L. Archer. 1997. The staphylococci in human disease. Churchill Livingstone, Inc., New York, N.Y.
Edwards, K. J., M. E. Kaufman., and N. A. Saunders. 2001. Rapid and accurate identification of coagulase-negative staphylococci by real-time PCR. J. Clin. Microbiol. 39:3047-3051.
Grant, C. E., D. L. Sewell, M. Pfaller, R. V. Bumgardner, and J. A. Williams. 1994. Evaluation of two commercial systems for identification of coagulase-negative staphylococci to species level. Diagn. Microbiol. Infect. Dis. 18:1-5.
Gribaldo, S., B. Cookson, N. Saunders, R. Marples, and J. Stanley. 1997. Rapid identification by specific PCR of coagulase-negative staphylococcal species important in hospital infection. J. Med. Microbiol. 46:45-53.
Grunberg-Manango, M. 1996. Regulation of the expression of aminoacyl-tRNA synthetases and translation factors, p. 1432-1457. In F. C. Neidhardt and R. Curtiss III, J. L. Ingraham, Edmund C. C. Lin, K. Brooks Low, B. Magasanik, W. S. Reznikoff, M. Riley, M. Schaechter, and H. E. Umbarger (ed.), Escherichia coli and Salmonella: cellular and molecular biology, 2nd ed., vol. 2. ASM Press, Washington, D.C.
Huebner, J., and D. A. Goldmann. 1999. Coagulase-negative staphylococci: role as pathogens. Annu. Rev. Med. 50:223-236.
Ieven, M. J., J. Verhoeven, S. R. Pattyn, and H. Goossens. 1995. Rapid and economical method for species identification of clinically significant coagulase-negative staphylococci. J. Clin. Microbiol. 33:1060-1063.
Karavolos, M. H., M. J. Horsburgh., E. Ingham, and S. J. Foster. 2003. Role and regulation of the superoxide dismutases of Staphylococcus aureus. Microbiology 149:2749-2758.
Kloos, W. E., and T. L. Bannerman. 1994. Update on clinical significance of coagulase-negative staphylococci. Clin. Microbiol. Rev. 7:117-140.
Krediet, T. G., E. M. Mascini, E. van Rooij, J. Vlooswijk, A. Paauw, and A. Fleer. 2004. Molecular epidemiology of coagulase-negative staphylococci causing sepsis in a neonatal intensive care unit over an 11-year period. J. Clin. Microbiol. 42:992-995.
Martineau, F., F. J. Picard, D. Ke, S. Paradis, P. H. Roy, M. Ouellette, and M. G. Bergeron. 2001. Development of a PCR assay for identification of staphylococci at genus and species level. J. Clin. Microbiol. 39:2541-2547.
Ohta, T. 1991. Multigene families and the evolution of complexity. J. Mol. Evol. 33:34-41.
Poyart, C., G. Quesne, C. Boumaila, and P. Trieu-Cuot. 2001. Rapid and accurate species-level identification of coagulase-negative staphylococci by using the sodA gene as a target. J. Clin. Microbiol. 39:4296-4301.
Renneberg, J., J. K. Rieneck, and E. Gutschik. 1995. Evaluation of Staph ID 32 system and Staph-Zym system for identification of coagulase-negative staphylococci. J. Clin. Microbiol. 33:1150-1153.
Rupp, M. E., and G. L. Archer. 1994. Coagulase-negative staphylococci: pathogens associated with medical progress. Clin. Infect. Dis. 19:231-245.
Stoakes, L., M. A. John, R. Lannigan, B. C. Schieven, M. Ramos, D. Harley, and Z. Hussain. 1994. Gas-liquid chromatography of cellular fatty acids for identification of staphylococci. J. Clin. Microbiol. 32:1908-1910.
Takahashi, T., I. Satoh, and N. Kikuchi. 1999. Phylogenetic relationships of 38 taxa of the genus Staphylococcus based on 16S rRNA gene sequence analysis. Int. J. Syst. Bacteriol. 49:725-728.
Valderas, M. W., J. W. Gatson, N. Wreyford, and M. E. Hart. 2002. The superoxide dismutase gene sodM is unique to Staphylococcus aureus: absence of sodM in coagulase-negative staphylococci. J. Bacteriol. 184:2465-2472.(E. Heikens, A. Fleer, A. )
ABSTRACT
To compare commonly used phenotypic methods with genotypic identification methods 47 clinical isolates of coagulase-negative staphylococci (CONS), 10 CONS ATCC strains, and a Staphylococcus aureus clinical isolate were identified using the API Staph ID test, BD Phoenix Automated Microbiology System, and 16S rRNA gene and tuf gene sequencing. When necessary part of the sodA gene was sequenced for definitive identification. The results show that tuf gene sequencing is the best method for identification of CONS, but the API Staph ID test is a reasonably reliable phenotypic alternative. The performance of the BD Phoenix Automated Microbiology System for identification of CONS is poor. The present study also showed that although genotypic methods are clearly superior to phenotypic identifications, a drawback of sequence-based genotypic methods may be a lack of quality of deposited sequences in data banks. In particular, 16S rRNA gene sequencing suffers from the lack of high quality among sequences deposited in GenBank. Furthermore, genotypic identification based on 16S rRNA sequences has limited discriminating power for closely related Staphylococcus species. We propose partial sequencing of the tuf gene as a reliable and reproducible method for identification of CONS species.
INTRODUCTION
The human skin and mucous membranes represent a diverse environment of bacteria, the normal microfloras (5). Probably the most important bacteria of this microflora are members of the genus Staphylococcus. The genus Staphylococcus is currently divided in 38 species and 17 subspecies, half of which are indigenous to humans (13). Staphylococci generally have a benign or symbiotic relationship with their host. However, they may develop into a pathogen if they gain entry into the host tissue through trauma of the cutaneous barrier, inoculation by needles, or implantation of medical devices.
In last two decades, coagulase-negative staphylococci (CONS) have also emerged as significant pathogens, especially in immunocompromised patients, premature newborns, and patients with implanted biomaterials. The most frequently encountered CONS species associated with human infections is Staphylococcus epidermidis, in particular in association with intravascular catheters. In addition, S. epidermidis is the predominant agent of nosocomial bacteremia, prosthetic-valve endocarditis, surgical wounds, central nervous system shunt infections, intravascular catheter-related infections, peritoneal dialysis-related infections, and infections of prosthetic joints (5, 10, 19). The second most frequently encountered CONS species is S. haemolyticus. S. haemolyticus has been implicated in native-valve endocarditis, septicemia, peritonitis, and wound, bone, and joint infections (5, 10, 19). Other CONS species are involved in a variety of infections. For example, S. saprophyticus is an important pathogen in human urinary tract infections, especially in young, sexually active females, and S. lugdunensis has been implicated in arthritis, catheter infections, and prosthetic joint infections (5, 10, 19).
Because of the increasing clinical significance of CONS, accurate species identification of CONS is highly desirable to permit a more precise determination of host-pathogen relationship of CONS. Phenotypic identification of CONS appears to be unsatisfactory, unreliable, and irreproducible. Commercial identification systems and automated systems are not able to make a reliable distinction between the different species of CONS because of the variable expression of the phenotypic characters (1, 2, 3, 7, 11, 18). Other tests, such as enzyme electrophoresis or analysis of cellular fatty acid composition, have also failed to make a reliable identification (20). Therefore, methods based on sequencing of PCR amplicons were evaluated and compared to phenotypic identifications of CONS. A number of PCR amplicon-sequencing-based methods for identification of CONS have been reported, i.e., targeting the 16S rRNA, sodA, and tuf genes (8, 15, 17). Multiple copies of the 16S rRNA gene are present on the chromosome of most bacteria. The sequence variation between the different copies of the 16S rRNA gene seems to be relatively low (4, 16). The sodA gene encodes the manganese-dependent superoxide dismutase. This metalloprotein inactivates harmful superoxide radicals (12, 22). The tuf gene, which encodes the elongation factor Tu, is involved in peptide chain formation and is part of the ribosome (9). These genes are essential constituents of the bacterial genome and are therefore preferred for diagnostic purposes.
The aim of the present study was to compare commonly used phenotypic methods with genotypic identification methods for clinical CONS isolates. For that purpose, 57 CONS isolates were phenotypically identified by means of the API Staph ID test and the BD Phoenix Automated Microbiology System. Furthermore, the same 57 CONS isolates were genotypically identified by sequencing of PCR amplicons based on sequences of Staphylococcus 16S rRNA, tuf, and sodA genes.
MATERIALS AND METHODS
Bacterial isolates. The bacterial isolates used in this study consisted of 10 ATCC reference strains of various CONS species, 47 clinical CONS isolates, and an S. aureus clinical isolate. The reference strains are listed in Table 1 and were obtained from the American Type Culture Collection (ATCC). The clinical isolates were a random selection from a collection of blood isolates from the neonatal intensive care unit of the University Medical Center of Utrecht and have been described previously (14). The organisms were identified by Gram staining, catalase, coagulase, and DNase tests. All strains were grown overnight on blood agar plates at 37°C.
Phenotypic identification. Phenotypic identification of bacterial isolates was achieved by using the API Staph ID test (BioMerieux) and the BD Phoenix Automated Microbiology System (Becton Dickinson, software version 4.01). For the API Staph ID test, results were interpreted using the API LAB ID computer software. Both biochemical tests were performed according to the manufacturers' instructions.
Extraction of genomic DNA. After overnight growth on blood agar plates at 37°C, genomic DNA of pure cultures was extracted using a Dneasy Tissue Kit (QIAGEN) according to the manufacturer's instructions. Lysates were frozen at –70°C until they were used for PCR.
PCR amplification. The sequences of the primers used in this study are shown in Table 2. All primers were purchased from Invitrogen. The PCR mixtures contained 10 μl PCR buffer (10x, as supplied with Taq polymerase), 8 μl deoxynucleoside triphosphates (200 μM concentrations of each), 2 μl of genomic DNA (120 ng/ml), 0.2 μl Taq DNA polymerase (0.2 U) and 2.5 μl of each of the primers (20 μM) in a final volume made up to 100 μl with water. The thermal cycling conditions were 5 min at 95°C for 1 cycle, followed by 30 cycles of 1 min at 95°C, 1 min at 55°C, and 1 min at 72°C. The last cycle was performed at 10 min at 72°C. Amplification of a 1,500-bp-long fragment of the 16S rRNA gene was achieved by PCR from genomic DNA with the forward primer Epsilon F and the reverse primer 1510R. These primers have been described previously (6). For tuf and sodA, new primers were designed. tuf and sodA sequences were obtained from GenBank. Multiple sequence alignments were carried out by using the program Align Plus of the Scientific & Educational software (version 3). Based on these sequence alignments, regions of the tuf and sodA genes highly conserved among CONS were chosen. PCR primers were designed from these regions. Primer pair tuf-F and tuf-R was designed to amplify a 412-bp-long fragment of the tuf gene. Primer pair sodA-F and sodA-R was designed to amplify a 236-bp-long fragment of the sodA gene. The PCR conditions were the same as described before except for the amount of the primers (4 μl of the primers tuf-F and tuf-R [10 μM] and 8 μl of the primers sodA-F and sodA-R [10 μM]) and the annealing temperature (47°C for sodA). After PCR amplification 10 μl of each reaction mixture was analyzed on a 1.2% agarose gel. The DNA fragments were visualized and photographed under UV light after ethidium bromide staining. The PCR products were purified using the QIAquick PCR Purification Kit protocol (QIAGEN) according to the manufacturer's instructions.
DNA sequencing. Forward and reverse sequencing of a 320-bp-long part of the 16S rRNA amplicon and forward and reverse sequencing of a 412-bp-long fragment of the tuf gene and a 236-bp-long fragment of the sodA gene were obtained by using the forward primer Epsilon F, the reverse primer 320R, the forward primer tuf-F, the reverse primer tuf-R, the forward primer sodA-F, and the reverse primer sodA-R, respectively. The reverse primer 320R has been described previously (6). Sequencing reactions were carried out with the reagents of a BigDye Terminator v1.1 cycle sequencing ready reaction kit (QIAGEN). All PCR sequence reaction mixtures contained in the final concentration 4 μl Ready Reaction mix (2.5x), 2 μl BigDye sequencing buffer (5x), 2 μl template (purified PCR product), and 4 μl primer (10 ng/μl), in a final volume made up to 20 μl with water. The thermal cycling conditions were 1 min at 96°C for 1 cycle followed by 25 cycles of 10 s at 96°C, 5 s at 50°C, and 4 min at 60°C. The PCR sequence products were purified using the BigDye Terminator PCR Purification Kit protocol (QIAGEN) according to the manufacturer's instructions. The PCR sequence products were separated by electrophoresis on a 0.2-mm 4% polyacrylamide denaturing gel and recorded with an ABI 377 DNA sequencer (Applied Biosystems). The results were processed into sequence data with the sequence ABI Prism 377 collection software. To exclude sequencing mistakes, the forward nucleotide sequences were aligned with the reverse nucleotide sequences by using the program Align Plus. For identification the 16S rRNA, tuf, and sodA partial sequences were compared with sequences from GenBank. The partial tuf gene sequence for S. sciuri and the four S. xylosus isolates are deposited in GenBank (accession numbers AY763434 to AY763438, respectively).
Phylogenetic comparisons. Phylogenetic trees based on the 16S rRNApart and tufpart sequences were generated using the program Data Analysis in Molecular Biology and Evolution. The results were processed into a tree with the neighbor-joining method. All trees were resampled with 1,000 bootstrap replications to test the robustness of the data. Pairwise comparisons of the 16S rRNApart and tufpart sequences for homology were obtained using the program Align Plus.
RESULTS
Phenotypic and genotypic identification of CONS. A total of 10 CONS ATCC strains and 47 clinical CONS isolates and an S. aureus clinical isolate were identified using the API Staph ID test, the BD Phoenix Automated Microbiology System, and 16S rRNA gene and tuf gene sequencing. When no discrimination could be made based on 16S rRNA and tuf sequences, part of the sodA gene was sequenced for definitive identification. All PCR products generated for sequence analysis were of the expected size, i.e., 1,500 bp, 412 bp, and 236 bp for 16S rRNA, tuf, and sodA, respectively (data not shown). Partial sequencing of the 16S rRNA amplicons and sequencing of the tuf and sodA amplicons resulted in diagrams devoid of overlapping peaks, which confirmed that these strains contain a single type of 16S rRNA, tuf, and sodA gene. For identification the 16S rRNA, tuf, and sodA sequences were compared with sequences deposited in GenBank. Homology values above 97% were considered reliable. However, in some cases more than one species fulfilled this criterion. Definitive species identification was achieved on the basis of the combined sequence data for the 16S rRNA, tuf, and sodA genes (Table 3). The ATCC strains were identified as one S. capitis, one S. sciuri, three S. epidermidis, and four S. xylosus strains and one S. saprophyticus strain, in agreement with the naming by the ATCC. The clinical isolates consisted of one S. aureus, 31 S. epidermidis, six S. haemolyticus, five S. capitis, two S. warneri, and two S. hominis isolates and one S. cohnii isolate. The API Staph ID test misidentified seven clinical isolates. All S. capitis isolates were identified as S. epidermidis. Both S. hominis isolates were identified as S. lugdunensis. The Phoenix system misidentified 17 clinical isolates and was not able to identify 2 clinical isolates. All misidentified clinical isolates were S. epidermidis isolates, which were identified as members of six different species (S. warneri, S. capitis, S. hominis, S. simulans, S. saprophyticus, and S. chromogenes). Three ATCC strains were misidentified by the Phoenix system. One S. xylosus ATCC strain was named as S. simulans, and two S. epidermidis ATCC strains were named as S. schleiferi or S. saprophyticus.
16S rRNA gene sequencing showed correct identification except for the six S. capitis isolates, which were named as either S. epidermidis or S. caprae. Both S. hominis isolates and the S. cohnii isolate were identified as either S. hominis or S. xylosus and as either S. cohnii or S. saprophyticus, respectively. Similar problems occurred for two ATCC strains. One S. xylosus ATCC strain was identified as either S. xylosus or S. saprophyticus, and the S. saprophyticus ATCC strain was identified as S. xylosus. tuf gene sequencing showed correct identification except for five ATCC strains. The S. xylosus ATCC strains were named as S. saprophyticus and the S. sciuri ATCC strain was named as S. capitis. In these cases the homology values were below 97% and therefore considered not reliable. sodA gene sequencing agreed with our definitive identification. DNA amplification for the sodA gene of the S. sciuri ATCC strain failed. Therefore, genotypic identification was not possible. An explanation may be that the designed sodA reverse primer has five mismatches within the sodA sequence from the S. sciuri ATCC strain. The sodA reverse primer contains no more than one mismatch within the sodA sequences from the other isolates.
Phylogenetic comparisons. Separate phylogenetic trees, inferred from the neighbor-joining method, were constructed with the sequence data for 16S rRNA and tuf gene (Fig. 1 and 2). Although tuf is well conserved in Staphylococcus, variability was observed in the nucleotide sequences. Four Staphylococcus species (S. saprophyticus, S. xylosus, S. sciuri, and S. cohnii) harbor an insertion of three nucleotides. The inserted amino acid is leucine for S. sciuri and alanine for S. saprophyticus, S. xylosus, and S. cohnii. The insertion of three nucleotides was considered a single event rather than three distinct events. We replaced the insertion with a single mutation of one nucleotide (T for A) to avoid unreliable differences in phylogenetic relationships. On the basis of 16S rRNA sequences, S. epidermidis and S. capitis constitute a group of related species. On the other hand, from the tuf sequence analysis, association of S. warneri with the S. epidermidis group was inferred. In both 16S rRNA and tuf sequence analysis, S. saprophyticus clearly forms a distinct group which includes S. xylosus and S. cohnii with high bootstrap values of 100% (tuf) and 86% (16S rRNA).
Intraspecies sequence divergence was observed in our collection. On the basis of tuf sequences, the intraspecies sequence divergence was 0.5% among 34 S. epidermidis isolates, 0.3% among 6 S. haemolyticus isolates, and 0.8 to 1.1% among 4 S. xylosus isolates. On the basis of 16S rRNA sequences, only intraspecies sequence divergence (0.3 to 1%) among four S. xylosus isolates was observed. Pairwise comparison of the 16S rRNA sequences (data not shown) showed that the most closely related Staphylococcus species were S. saprophyticus and S. xylosus with 99.7% identity. On the other hand, the most divergent species pair was S. aureus and S. cohnii with 92.2% identity. Using tuf sequences the most closely related Staphylococcus species were S. warneri and S. capitis with 97% identity. The most divergent species pair was S. xylosus and S. sciuri with 85.7% identity. Furthermore, pairwise comparison of the tuf sequences revealed that their mean identity (92.6%) is lower than the mean identity (95.9%) of 16S rRNA sequences.
DISCUSSION
Clinically significant CONS should be identified at the species level by a reliable and reproducible method to provide a better understanding of pathogenic potential of various CONS species. Phenotypic methods for identification of CONS appear to be unreliable. The purpose of the present study was to compare commonly used phenotypic methods, the API Staph ID test and the BD Phoenix Automated Microbiology System, with PCR amplicon-sequencing based methods targeting the 16S rRNA, tuf, and sodA genes.
The results of the phenotypic identification methods indicate that the API Staph ID is a much more reliable phenotypic method than the BD Phoenix Automated Microbiology System. The results obtained with the genotypic methods indicate that the PCR amplicon-sequencing-based techniques are specific and sensitive for identification of Staphylococcus species. 16S rRNA gene sequencing misidentified eight isolates (six clinical isolates and two ATCC strains). An explanation for these misidentifications may be that the deposited sequences in GenBank are incorrectly assigned to the various Staphylococcus species.
tuf gene sequencing misidentified five ATCC strains. These misidentifications occurred because tuf gene sequences of the species S. xylosus and S. sciuri have never been deposited in the GenBank database. The S. xylosus ATCC strains and S. sciuri ATCC strain were therefore named after closely related species, S. saprophyticus and S. capitis, respectively. sodA gene sequencing agreed with our definitive identification. No misidentifications were observed. A drawback of sodA gene-based genotypic identification may be the high interspecies sequence divergence, which makes it difficult to find conserved regions to design primers.
The 16S rRNA and tuf sequence-based relationships obtained were in accordance with phylogenetic trees published previously (15, 21). Both the 16S rRNA and tuf sequence-derived trees indicate that S. sciuri is the most distant Staphylococcus species. This result is in agreement with phylogenetic studies performed earlier with 16S rRNA and tuf sequence analysis (15, 21). Furthermore, the tuf sequence-derived tree indicates that S. warneri is associated with the S. epidermidis group, which includes S. capitis. This is in agreement with results earlier described in PCR-sequencing assays targeting the sodA gene (17). Tuf-derived data more often showed intraspecies sequence divergence than the 16S rRNA-derived data. Apparently, the 16S rRNA gene is more highly conserved than the tuf gene. Pairwise comparison of the tuf sequences revealed that their mean identity (92.6%) is lower than the mean identity (95.9%) of 16S rRNA sequences. These results indicate that tuf constitutes a more discriminatory target gene than the 16S rRNA gene to differentiate closely related Staphylococcus species.
In conclusion, the present study demonstrates the superiority of genotypic methods over phenotypic assays for identification of CONS species. tuf gene sequencing is the best method for the identification of CONS, but the API Staph ID test is a reasonably reliable phenotypic alternative. The performance of the BD Phoenix Automated Microbiology System using software version 4.01 for identification of CONS is poor. The present study also shows that although genotypic methods are clearly superior to phenotypic identifications, a drawback of sequence-based genotypic methods may be a lack of quality of deposited sequences in data banks. In particular, 16S rRNA gene sequencing suffers from the lack of high quality among deposited sequences in GenBank. Furthermore, genotypic identification based on 16S rRNA sequences has limited discriminating power for closely related Staphylococcus species. We propose PCR and sequencing of the tuf gene as a reliable and valuable approach for the genotypic identification of CONS species. Furthermore, the tuf sequence polymorphisms have potential for the development of biprobe identification assays based on real-time PCR (6). This method might be a useful diagnostic tool for a reliable and simple identification of clinically relevant CONS species.
ACKNOWLEDGMENTS
We thank D. de Jong for helpful advice on generating phylogenetic trees.
REFERENCES
Bannerman, T. L., K. T. Kleeman, and W. E. Kloos. 1993. Evaluation of the Vitek Systems Gram-Positive Identification card for species identification of coagulase-negative staphylococci. J. Clin. Microbiol. 31:1322-1325.
Birnbaum, D., M. Kelly, and A. W. Chow. 1991. Epidemiologic typing systems for coagulase-negative staphylococci. Infect. Control Hosp. Epidemiol. 12:319-326.
Calvo, J., J. L. Hernandez, M. C. Farinas, D. Garcia-Palomo, and J. Aguero. 2000. Osteomyelitis caused by Staphylococcus schleiferi and evidence for misidentification of this Staphylococcus species by an automated identification system. J. Clin. Microbiol. 38:3887-3889.
Cilia, V., B. Lafay, and R. Christen. 1996. Sequence heterogeneities among 16S ribosomal RNA sequences, and their effect on phylogenetic analyses at the species level. Mol. Biol. Evol. 13:451-461.
Crossley, K. B., and G. L. Archer. 1997. The staphylococci in human disease. Churchill Livingstone, Inc., New York, N.Y.
Edwards, K. J., M. E. Kaufman., and N. A. Saunders. 2001. Rapid and accurate identification of coagulase-negative staphylococci by real-time PCR. J. Clin. Microbiol. 39:3047-3051.
Grant, C. E., D. L. Sewell, M. Pfaller, R. V. Bumgardner, and J. A. Williams. 1994. Evaluation of two commercial systems for identification of coagulase-negative staphylococci to species level. Diagn. Microbiol. Infect. Dis. 18:1-5.
Gribaldo, S., B. Cookson, N. Saunders, R. Marples, and J. Stanley. 1997. Rapid identification by specific PCR of coagulase-negative staphylococcal species important in hospital infection. J. Med. Microbiol. 46:45-53.
Grunberg-Manango, M. 1996. Regulation of the expression of aminoacyl-tRNA synthetases and translation factors, p. 1432-1457. In F. C. Neidhardt and R. Curtiss III, J. L. Ingraham, Edmund C. C. Lin, K. Brooks Low, B. Magasanik, W. S. Reznikoff, M. Riley, M. Schaechter, and H. E. Umbarger (ed.), Escherichia coli and Salmonella: cellular and molecular biology, 2nd ed., vol. 2. ASM Press, Washington, D.C.
Huebner, J., and D. A. Goldmann. 1999. Coagulase-negative staphylococci: role as pathogens. Annu. Rev. Med. 50:223-236.
Ieven, M. J., J. Verhoeven, S. R. Pattyn, and H. Goossens. 1995. Rapid and economical method for species identification of clinically significant coagulase-negative staphylococci. J. Clin. Microbiol. 33:1060-1063.
Karavolos, M. H., M. J. Horsburgh., E. Ingham, and S. J. Foster. 2003. Role and regulation of the superoxide dismutases of Staphylococcus aureus. Microbiology 149:2749-2758.
Kloos, W. E., and T. L. Bannerman. 1994. Update on clinical significance of coagulase-negative staphylococci. Clin. Microbiol. Rev. 7:117-140.
Krediet, T. G., E. M. Mascini, E. van Rooij, J. Vlooswijk, A. Paauw, and A. Fleer. 2004. Molecular epidemiology of coagulase-negative staphylococci causing sepsis in a neonatal intensive care unit over an 11-year period. J. Clin. Microbiol. 42:992-995.
Martineau, F., F. J. Picard, D. Ke, S. Paradis, P. H. Roy, M. Ouellette, and M. G. Bergeron. 2001. Development of a PCR assay for identification of staphylococci at genus and species level. J. Clin. Microbiol. 39:2541-2547.
Ohta, T. 1991. Multigene families and the evolution of complexity. J. Mol. Evol. 33:34-41.
Poyart, C., G. Quesne, C. Boumaila, and P. Trieu-Cuot. 2001. Rapid and accurate species-level identification of coagulase-negative staphylococci by using the sodA gene as a target. J. Clin. Microbiol. 39:4296-4301.
Renneberg, J., J. K. Rieneck, and E. Gutschik. 1995. Evaluation of Staph ID 32 system and Staph-Zym system for identification of coagulase-negative staphylococci. J. Clin. Microbiol. 33:1150-1153.
Rupp, M. E., and G. L. Archer. 1994. Coagulase-negative staphylococci: pathogens associated with medical progress. Clin. Infect. Dis. 19:231-245.
Stoakes, L., M. A. John, R. Lannigan, B. C. Schieven, M. Ramos, D. Harley, and Z. Hussain. 1994. Gas-liquid chromatography of cellular fatty acids for identification of staphylococci. J. Clin. Microbiol. 32:1908-1910.
Takahashi, T., I. Satoh, and N. Kikuchi. 1999. Phylogenetic relationships of 38 taxa of the genus Staphylococcus based on 16S rRNA gene sequence analysis. Int. J. Syst. Bacteriol. 49:725-728.
Valderas, M. W., J. W. Gatson, N. Wreyford, and M. E. Hart. 2002. The superoxide dismutase gene sodM is unique to Staphylococcus aureus: absence of sodM in coagulase-negative staphylococci. J. Bacteriol. 184:2465-2472.(E. Heikens, A. Fleer, A. )