Detection and Identification of Microorganisms by Gene Amplification and Sequencing

Transcription

1 MEDICAL MICROBIOLOGY L. Barth Reller and Melvin P. Weinstein, Section Editors INVITED ARTICLE Detection and Identification of Microorganisms by Gene Amplification and Sequencing Cathy A. Petti Departments of Medicine and Pathology, University of Utah School of Medicine, and Associated Regional University Pathologists Laboratories, Salt Lake City, Utah Gene amplification and sequencing have led to the discovery of new pathogens as agents of disease and have enabled us to better classify microorganisms from culture. Sequence-based identification of bacteria and fungi using culture is more objective and accurate than conventional methods, especially for classifying unusual microorganisms that are emerging pathogens in immunocompromised hosts. Although a powerful tool, the interpretation of sequence-based classification can be challenging as microbial taxonomy grows more complex, without known clinical correlatives. Additionally, broad-range gene polymerase chain reaction and sequencing have emerged as alternative, culture-independent methods for detecting pathogens from clinical material. The promise of this technique has remained strong, limited mainly by contamination and inadequate sensitivity issues. This review explains sequence-based microbial classification, with emphasis on relating the complex world of microbial taxonomy to a clinical context. Additionally, this review discusses a rational approach to broad-range bacterial polymerase chain reaction and gene sequencing when applied directly to clinical samples. Gene amplification and sequencing of broad-range gene targets for bacteria and fungi have emerged as important tools to diagnose infections. During the past decade, clinical laboratories have applied PCR amplification and gene sequencing to characterize microorganisms from culture, and occasionally, to directly detect pathogens from patient samples. Gene sequencing is a more accurate and reproducible method to identify microorganisms and has increased our ability to capture the diversity of microbial taxa [1]. This new technology has resulted in the identification of unusual microorganisms [2] and the detection of novel, difficult-to-cultivate microorganisms, such as Tropheryma whipplei [3, 4]. However, clinicians are now faced with interpreting microbiological reports that include taxa that are unfamiliar and cannot be assimilated in a meaningful clinical setting. Also, application of broad-range bacterial and fungal PCR directly from clinical material is now more widely available, shifting from research to the clinical setting. This review describes sequence-based identification of bacteria and fungi, with particular emphasis on improving our understand- Received 3 January 2007; accepted 5 January 2007; electronically published 2 March Reprints or correspondence: Dr. Cathy A. Petti, University of Utah School of Medicine, 30 N East, Salt Lake City, UT (cathy.petti@aruplab.com). Clinical Infectious Diseases 2007; 44: by the Infectious Diseases Society of America. All rights reserved /2007/ $15.00 DOI: / ing of the increasingly complex world of microbial taxonomy as it relates to the clinical context. Additionally, this review discusses a rational approach to broad-range bacterial PCR and gene sequencing when applied directly to clinical samples. SELECTION OF GENE TARGETS For bacteria, mycobacteria, and fungi, many gene targets have been recognized as useful tools for identification. Broad-range gene targets for viruses have not been developed, owing to their greater genetic diversity and our inability to find a common genetic link. The gene targets that are selected must be functionally constant, serving as molecular clocks of microbial evolutionary change (phylogeny). The gene should have a conserved segment that is common to all bacteria (or fungi) and that is flanked by variable or highly variable regions. Conserved regions are responsible for the universality of the gene target where PCR and DNA sequencing primers anneal. During cycle sequencing, variable or highly variable regions generate unique nucleotide base fragments or sequences that serve as signatures for different species. The sequence is then compared with reference sequences that are deposited in public or private sequence libraries to determine relatedness. The acceptable degree of difference between the 2 sequences for classification to genus or species is variable, depending on the gene target and microorganism, and is the subject of ongoing debate CID 2007:44 (15 April) MEDICAL MICROBIOLOGY

2 Bacteria. Classic methods to identify bacteria are based on characteristics observed in known strains with predictable biochemical and physical properties under optimal growth conditions. When common microorganisms present with uncommon phenotypes, reliance on phenotype can compromise accurate identification [5]. Identification by gene sequencing is more objective, does not require optimal growth or even a viable microorganism, and has the added capability of defining taxonomical relationships among bacteria. The gene target that is most commonly used for bacterial identification is 16S rrna (or 16S rdna), an 1500 base pair gene that codes for a portion of the 30S ribosome (figure 1). Partial (500 base pair) 16S rrna gene sequencing has emerged as a more accurate and faster method to identify a wide variety of aerobic and anaerobic bacteria and has been successfully implemented in clinical laboratories [6 10]. Bacteria may have only a single copy or multiple copies of this gene, and this can make interpretation difficult when base pair changes exist among copies (copy variants). Conversely, multiple copies can be helpful to improve the sensitivity of PCR amplification, especially when applying this technology directly from clinical samples. A major limitation of the 16S rdna sequence is its inability to discriminate among all bacterial taxa. For example, Bacillus cereus and Bacillus anthracis have identical 16S rdna sequences and, in fact, cannot be separated reliably at the molecular level by their genomic DNA, because their differences primarily lie in the acquisition of virulence plasmids. For other microorganisms that share complete sequence identity, alternative gene targets can provide better separation of closely related species. The rpob gene, a gene that encodes the b-subunit of bacterial RNA polymerase, is a valuable gene target when partial 16S rrna gene sequencing results cannot discriminate among taxa [11]. Additionally, rpob sequence data are more straightforward, without copy variants, because bacteria only have 1 copy of this gene. The rpob gene has been particularly useful for identification of rapidly-growing mycobacteria, such as Mycobacterium chelonae and Mycobacterium abscessus. These 2 species have indistinguishable 16S rdna sequences but exhibit 13% sequence divergence with rpob [12]. Other bacterial gene targets that may provide better separation of certain species include tuf (elongation factor Tu), gyra or gyrb (gyrase A or B), soda (manganese-dependent superoxide dismutase), and heat shock proteins. Fungi. Similar to bacteria, classic identification of fungi requires biochemical testing and expertise in recognizing the morphology of fungal reproductive structures macroscopically and microscopically. Identification by gene sequencing does not require viable organism or sporulation (molds), enabling a more rapid diagnosis. Gene targets for identification of medically important yeast and fungi are not as well codified as those for bacteria. The optimal targets appear to be internal transcribed spacer regions ITS1 and ITS2, which are variable regions located between conserved genes encoding for 18S, 5.8S, and 28S rrna (figure 2). The ITS region has proven useful for identification of yeasts, such as Candida, Cryptococcus, Trichosporon [13, 14], and Aspergillus species; zygomycetes; dematiaceous molds; and other medically relevant fungi [15, 16]. The ITS regions have limitations, and alternative gene targets are often necessary to identify certain genera, such as D1 and D2 domains of the 28S rrna subunit [17], elongation factor a (e.g., Fusarium species), and b-tabulin (e.g., Phaeoacremonium species).. INTERPRETATION OF GENE SEQUENCES FROM PURE CULTURE Accurate classification of gene sequences to a particular genus or species requires analysis with a high-quality, comprehensive reference library. MicroSeq, GenBank, Ribosomal Database Project, Ribosomal Differentiation of Microorganisms, and SmartGene are all useful databases, but each has unique strengths and limitations. For example, GenBank is a large, public database with 1200,000 named 16S rdna sequences, with the limitation of less oversight resulting in reference sequences of poor quality. MicroSeq (Applied Biosystems) is an established, commercially available reference sequence library with S rdna sequences derived from type strains of microorganisms, which may yield higher-quality sequence data Figure 1. A schematic for 16S rrna (an 1500 base pair gene), located on the small ribosomal subunit (30S). Circles represent conserved regions that serve as gene targets for PCR amplification and DNA sequencing of bacteria. MEDICAL MICROBIOLOGY CID 2007:44 (15 April) 1109

3 Figure 2. Fungal ITS regions between the genes for the small and large rdna subunits. Circles represent conserved regions (outside ITS1 and ITS2 variable regions) that serve as gene targets for PCR amplification and DNA sequencing of fungi. but should be used cautiously, because use of a single type strain to represent entire taxa is often inadequate. Reference databases are largest for 16S rdna (bacteria) and ITS1/2 (fungi) sequences, but sequences from other gene targets are increasing rapidly. Nucleotide sequences are generally reported in terms of percent identity. This term refers to the number of identical nucleotide bases shared by the query and reference sequences divided by the number of nucleotide bases sequenced. Figure 3 is an example of a report generated for the microorganism Acinetobacter baumannii. No consensus guideline exists for establishing criteria on the basis of percent identity scores to classify microorganisms to their respective genera by gene sequencing, especially for fungi. For bacteria identified by the 16S rrna gene, most taxonomists accept a percent identity score of 97% and 99% to classify a microorganism to genus and species, respectively. As mentioned previously, this classification schema most likely will continue to evolve as we learn more about heterogeneity within certain taxa. For example, sequence-based classification differs for Mycobacteria species, owing in large part to their homology within the 16S rrna region, and a sequence must share at least 99.6% identity with the reference strain for species identification. Percent identity scores for genus and species classification also vary with different gene targets. Mycobacterium species sequenced from reca and rpob can be classified to distinct species sharing only 97% 98% identity. Challenges arise when microorganisms cannot be discriminated by 16S rrna gene sequencing. For example, members of the Streptococcus mitis group, which include Streptococcus pneumoniae, have indistinguishable 16S rrna gene sequences, sharing between 99% 100% identities when they are aligned (figure 4A). Sequence similarity can be assessed further by constructing a relatedness diagram (phylogenetic tree) that estimates the evolutionary distances among sequences. Figures 4B and 4C show phylogenetic trees for the Streptocococcus mitis group, created with both 16S rrna and tuf gene sequences. For this group of microorganisms, the tuf gene has greater sequence divergence and, thus, provides a better separation of species for a more accurate sequence-based classification. Alternative gene targets are also helpful for the identification of Bordetella pertussis, Bordetella parapertussis, and Bordetella bronchiseptica; Neisseria meningitidis and Neisseria cinerea; and Escherichia coli and Shigella species in which each group has indistinguishable 16S rrna gene sequences. Gene sequencing has enabled us to identify new and less common microorganisms that are agents of human disease. It is unclear whether these unusual microorganisms have emerged as new human pathogens or whether they have been agents of human disease that were previously misclassified by conven- Figure 3. Representation of identity scores from the Ribosomal Differentiation of Microorganisms database. The query isolate shares 100% identity with Acinetobacter baumannii. The Smith-Waterman score assesses relatedness on the basis of the number of aligned bases and percent identity CID 2007:44 (15 April) MEDICAL MICROBIOLOGY

4 Figure 4. A, Each nucleotide base of the query sequence compared with the reference sequence by aligning the 2 sequences. With partial 16S rrna gene sequencing, Streptococcus mitis and Streptococcus pneumoniae differ by only 2 base pairs, whereas with tuf gene sequencing, these organisms differ by 11 base pairs. B and C, Distance matrices from phylogenetic trees calculated by various methods. Generally, the association between 2 sequences can be calculated by adding their horizontal distances. For S. mitis and S. pneumoniae, the horizontal distance calculated on 16S rrna tree is small (0.005), compared with tuf gene tree, which demonstrates greater evolutionary distance (10.01). tional methods. A case report of bacteremia from Devosia species, a gram-negative bacillus, may in fact be the first description of this microorganism causing human disease, but without 16S rrna gene sequencing, this bacteria previously may have been classified by conventional methods as an environmental gramnegative rod or Pseudomonas species. Interpreting microbiological reports with unfamiliar taxa is difficult, especially when we are unable to associate these microorganisms with particular clinical syndromes. For example, Helcococcus species identified in joint fluid using sequencing may be classified as a viridans group Streptococcus species using conventional methods. Clinicians may interpret a report of an unfamiliar genus, such as Helcococcus species, as more significant than a report of viridans group streptococci, for which clinicians have associative clinical experience. Similarly, clinical laboratories should apply gene sequencing for microbial identification cautiously and judiciously. A single blood culture with colonies suggestive of diphtheroids should not be sequenced routinely for identification, because clinicians may infer from the sequencing result, Leifsonia aquaticum (formerly part of Corynebacterium genus), that this microorganism is a new pathogen with a high likelihood for causing disease, which is a misleading inference. When unusual taxa are reported by gene sequencing, we propose that clinical laboratories provide additional information to aid clinicians in placing new species in a more familiar microbial framework (figure 5). Overall, sequence-based identification from culture is an extremely valuable tool, especially for classifying unusual microorganisms that are emerging pathogens in immunocompromised hosts. With 16S rrna gene sequencing, we have been able to reliably identify Inquinilus limosus from respiratory samples, a known pathogen in patients with cystic fibrosis. For fungi, better characterization of zygomycetes by ITS sequencing can be extremely useful for better predicting antifungal activity, which varies among genera. Finally, microbial identification by gene sequencing has improved our ability to recognize new, emerging pathogens, and better defines taxonomical relationships and may further our understanding of microbial pathogenesis. Figure 5. Reporting of novel species using phylogenetic trees. An unusual species, Tannerella forsythensis, can be reported in a framework that demonstrates its relationship to more commonly known anaerobes. MEDICAL MICROBIOLOGY CID 2007:44 (15 April) 1111

5 APPLICATIONS FOR CLINICAL SAMPLES Broad-range gene PCR and sequencing from clinical material provide an alternative, culture-independent method for detecting pathogens. The promise of this technique has remained strong, especially for the bacterial domain, but its application has not been reliably reproducible, owing, in large part, to issues with contamination and inadequate sensitivity [18]. The 16S rrna gene has been the most widely used gene for this application, because it is present in all bacteria and has conserved regions for designing broad-range primers, and many bacteria contain multiple copies of this gene in their genomes. The most successful application for broad-range PCR directly from clinical material has been pathogen discovery, especially for acute illnesses or chronic inflammatory syndromes with histopathological evidence suggesting an infectious etiology. In 1990, the agent of bacillary angiomatosis, subsequently identified as Bartonella henselae, was discovered using 16S rdna amplification and sequencing from tissue sections obtained from patients with HIV infection [19]. Tropheryma whipplei, a previously uncultured bacterium, was first detected with 16S rdna primers in a small bowel biopsy sample obtained from a patient with Whipple disease [3, 4]. In both cases, the pathogen was previously detected by special staining methods, such as Warthin-Starry (B. henselae) or periodic acid-schiff (T. whipplei), with an inability to cultivate the microorganism. Causation can be more difficult to establish when there is no definitive histopathogical correlative of an infectious process. In addition to bacterial DNA contamination of PCR reagents and other materials, some investigators have suggested that bacterial DNA can be detected in tissues that were previously considered to be sterile [20, 21], but this claim needs to be confirmed with more definitive studies. In general, the finding of 16S rdna alone is not sufficient to infer causation for an acute or chronic inflammatory syndrome. The application of broad-range bacterial PCR for less defined clinical syndromes has contributed modestly to our understanding of these illnesses. Among 46 patients with unexplained life-threatening illnesses that were suggestive of an infectious etiology [22], 16S rdna PCR and sequencing detected bacteria in 8 specimens, including 2 cases involving N. meningitidis (in a CSF specimen), 3 cases involving S. pneumoniae (in a CSF specimen in 1 case and in a pleural fluid specimen in 2 cases), and 1 case involving Stenotrophomonas maltophilia (in a bone marrow aspirate specimen). Two positive PCR specimens from blood culture material detected pathogens with uncertain clinical significance, suggesting possible contamination, a known limitation of broad-range PCR. More importantly, of the 6 PCR-positive, culture-negative samples that were clinically relevant, 5 patients had received antibacterial therapy for at least 24 h prior to specimen collection. Broad-range PCR has considerable advantages to culture, but it has not drastically contributed to identifying the etiologic agents of unexplained, culture-negative illnesses. It is important to note that culture-based methods have exceedingly high sensitivity as a result of decades of efforts to improve its capabilities to recover 1 viable microorganism from a sample. Similarly, culture has the added advantage of culturing large volumes of clinical material (in milliliters), whereas PCR is limited to much smaller volumes (in microliters). Although PCR detects 1 gene copy theoretically, it often does not approach this sensitivity because of the presence of PCR inhibitors and background material from human DNA. Perhaps our greatest lesson is to understand the unique capabilities of broad-range PCR, which will enable us to optimize its diagnostic value. These targeted applications will be explored through selected publications that have used broad-range bacterial PCR for specific clinical syndromes. Bacteremia and endocarditis. The incidence of culturenegative infective endocarditis remains high and, in hopes of better defining the etiologic agents responsible for this disease, broad-range bacterial PCR has been applied to blood sample, blood culture, and cardiac valve specimens from patients with this disease [23 26]. Although the use of broad-range PCR for patients with endocarditis can be helpful, false-negative PCR results occur even when blood or valve culture results are positive. In a study of patients undergoing heart valve resection for culture-positive and -negative endocarditis [23], 16S rrna PCR failed to detect 2 cases of viridans group streptococcal endocarditis, whereas culture failed to recover T. whipplei. In an additional case involving negative blood and valve culture results, PCR was positive for viridans group streptococci, but gram-positive cocci were found on Gram stain from the resected valve. A similar study found broad-range PCR additive for patients with Bartonella quintana endocarditis and those receiving prolonged courses of antimicrobial therapy prior to valvular excision [24]. Reports of PCR contamination and its specificity are variable, but clinicians should interpret positive PCR results with caution, especially in the absence of histological evidence for endocarditis. Overall, broad-range bacterial PCR methods have proven to be useful for diagnosis of endocarditis when applied to blood samples, blood culture bottles, or valvular material for patients already receiving antimicrobial therapy, with positive histology, or for disease caused by fastidious pathogens (e.g., Bartonella species and Coxiella burnetii). Meningitis and brain abscess. Broad-range bacterial PCR has been applied to CSF samples, with similar results to those of blood samples and with bacterial DNA contamination and sensitivity remaining problematic [27, 28]. In a prospective study of 227 patients with meningitis in their differential diagnosis, 16S rdna PCR had a sensitivity of 86%, specificity of 97%, positive predictive value of 80%, and negative predictive 1112 CID 2007:44 (15 April) MEDICAL MICROBIOLOGY

6 value of 98% [27]. For 5 patients with clinically significant positive culture results (3 with N. meningitidis, 1 with Listeria monocytogenes, and 1 with Pantoae species), the PCR results were negative. Of 6 patients with negative culture and positive PCR results, 5 had received antibacterial therapy prior to sample collection. These findings again reflect the advantages of molecular methods when patients have received prior antibacterial therapy. There have been no large studies that have applied broad-range bacterial PCR to samples of fluid or tissue from brain abscesses. We have experience with 2 cases in which the Gram stain from the abscess demonstrated gram-positive cocci with negative culture results. 16S rrna amplification and sequencing detected Streptococcus anginosus group in samples obtained from each patient. Again, broad-range bacterial PCR should be used as an adjunctive method to culture for the diagnosis of meningitis and brain abscesses when the clinical presentation and laboratory findings (e.g., CSF parameters, Gram stain, and histological stains) suggest a bacterial etiology and may prove to be useful, particularly for patients who have received antibacterial therapy at the time of sample collection, when culture results remain negative. Bone and joint infections. The etiologic agents of osteomyelitis, septic arthritis, and prosthetic joint infections often have been difficult to cultivate, with a significant proportion of culture-negative infections. The value of broad-range PCR has been studied in this area extensively [29, 30], and unlike obtainment of blood or CSF specimens, extraction of bacterial DNA from bone and joint tissue presents many challenges. Disruption of the biofilm, which develops in patients with orthopaedic implants, can be difficult, but it is essential to release bacterial DNA to improve the sensitivity of broad-range bacterial PCR. In a recent study of osteoarticular infections, 16S rdna methods detected pathogens in 16 cases in which culture results were negative, with 7 patients having received prior antibacterial therapy [29]. Conversely, in 4 cases of culturepositive S. aureus infection, broad-range bacterial PCR results were negative. False-positive results occurred for both culture and broad-range PCR, with Staphylococcus epidermidis, Propionibacterium acnes, and Pseudomonas aeruginosa reported as the most common PCR contaminants. For patients with septic arthritis, broad-range bacterial PCR also has not been proven to be superior to conventional methods [30]. PCR may prove to be useful in select cases of bone and soft-tissue infections in patients who have received prior antibacterial therapy, when fastidious pathogens are suspected (e.g., Streptobacillus moniloformis and Kingella kingae in children) or when the pathogen is uncultivable (e.g., Borrelia burgdorferi). CONCLUSIONS Gene amplification and sequencing have led to the discovery of new pathogens as agents of disease and have enabled us to better classify microorganisms from culture. The emergence of more unusual microorganisms from culture may represent new human pathogens or reflect our evolving expertise of microbial taxonomy of species, which were not previously appreciated with the use of conventional methods of identification. The direct application of broad-range bacterial PCR to clinical samples is an important, adjunctive tool to culture and has proven to be valuable in certain clinical contexts. When culture results remain negative and the clinical suspicion for infection remains high, broad-range PCR is extremely useful, especially when histological stains yield positive results, patients have received antimicrobial therapy prior to sample collection, or the likely agents are fastidious or uncultivable pathogens. Overall, gene amplification and sequencing from culture and clinical material may improve our understanding of microbial pathogenesis and better predict responses to therapy and outcome. Acknowledgments I gratefully acknowledge Dr. Kenneth H. Wilson for his insightful discussions and careful review of this manuscript. I also thank Keith Simmon for his technical assistance and figure illustrations. Potential conflicts of interest. C.A.P. has consulted for Diagnostic Hybrids and has served on the speaker s bureau for Gilead. References 1. Clarridge JE. Impact of 16S rrna gene sequence analysis for identification of bacteria on clinical microbiology and infectious diseases. Clin Microbiol Rev 2004; 17: Drancourt D, Berger P, Raoult D. Systematic 16S rrna gene sequencing of atypical clinical isolates identified 27 new bacterial species associated with humans. J Clin Microbiol 2004; 42: Wilson KH, Blitchington R. Frothingham R, et al. Phylogeny of the Whipple s disease-associated bacterium. Lancet 1991; 338: Relman DA, Schmidt TM, MacDermott RP, et al. Identification of the uncultured bacillus of Whipple s disease. N Engl J Med 1992; 327: Petti CA, Polage CR, Schreckenberger P. Misidentification of microorganisms by conventional methods: role of 16S rrna gene sequencing. J Clin Microbiol 2005; 43: Tang YW, Ellis NM, Hopkins MK, et al. Comparison of phenotypic and genotypic techniques for identification of unusual aerobic pathogenic gram-negative bacilli. J Clin Microbiol 1998; 36: Bosshard PP, Abels S, Zbinden R, et al. Ribosomal DNA sequencing for identification of aerobic gram-positive rods in the clinical laboratory (an 18-month evaluation). J Clin Microbiol 2003; 41: Simmon KE, Croft AC, Petti CA. Application of SmartGene IDNS software to partial 16S rrna gene sequences for a diverse group of bacteria in a clinical laboratory. J Clin Microbiol 2006; 44: Hall L, Doerr KA, Wohlfiel SL. Evaluation of the MicroSeq system for identification of mycobacteria by 16S ribosomal DNA sequencing and its integration into a routine clinical mycobacteriology laboratory. J Clin Microbiol 2003; 41: Patel JB, Wallace RJ, Brown-Elliott BA, et al. Sequence-based identification of aerobic actinomycetes. J Clin Microbiol 2004; 42: La Scola B, Bundi Gundi VAKB, Khamis A, et al. Sequencing of the rpob gene and flanking spacers for molecular identification of Acinetobacter species. J Clin Microbiol 2006; 44: Adekambi T, Drancourt M. Dissection of phylogenetic relationships among 19 rapidly growing Mycobacterium species by 16S rrna, hsp65, MEDICAL MICROBIOLOGY CID 2007:44 (15 April) 1113

7 soda, reca and rpob gene sequencing. Int J Syst Evol Microbiol 2004;54: Chen YC, Eisner JD, Kattar MM, et al. Identification of medically important yeasts using PCR-based detection of DNA sequence polymorphisms in the internal transcribed spacer 2 region of the rrna genes. J Clin Microbiol 2000; 38: Ciardo DE, Schar G, Bottger EC, et al. Internal transcribed spacer sequencing versus biochemical profiling for identification of medically important yeasts. J Clin Microbiol 2006; 44: Iwen PC, Hinrichs SH, Rupp ME. Utlization of the internal transcribed spacer regions as molecular targets to detect and identify human fungal pathogens. Med Mycol 2002; 40: Schwarz P, Bretagne S, Gantier JC, et al. Molecular identification of zygomycetes from culture and experimentally infected tissues. J Clin Microbiol 2006; 44: Hall H, Wohlfiel S, Roberts GD. Experience with the MicroSeq D2 large-subunit ribosomal DNA sequencing kit for identification of filamentous fungi encountered in the clinical laboratory. J Clin Microbiol 2004; 42: Corless CE, Guiver M, Borrow R, et al. Contamination and sensitivity issues with a real-time universal 16S rrna PCR. J Clin Microbiol 2000; 38: Relman DA, Loutit JS, Schmidt TM, et al. The agent of bacillary angiomatosis: an approach to the identification of uncultured pathogens. N Engl J Med 1990; 323: Nikkari S, McLaughlin IJ, Bi W, et al. Does blood of healthy subjects contain bacterial ribosomal DNA? J Clin Microbiol 2001; 39: Dagan R, Shriker O, Hazan I, et al. Prospective study to determine clinical relevance of detection of pneumococcal DNA in sera of children by PCR. J Clin Microbiol 1998; 36: Nikkari S, Lopez FA, Lepp PW, et al. Broad-range bacterial detection and the analysis of unexplained death and critical illness. Emerg Infect Dis 2002; 8: Goldenberger D, Kunzli A, Vogt P, et al. Molecular diagnosis of bacterial endocarditis by broad-range PCR amplification and direct sequencing. J Clin Microbiol 1997; 35: Breitkopf C. Hammel D, Scheld HH, et al. Impact of a molecular approach to improve the microbiological diagnosis of infective heart valve endocarditis. Circulation 2005; 111: Millar BC, Moore JE. Current trends in the molecular diagnosis of infective endocarditis. Eur J Clin Microbiol Infect Dis 2004; 23: Bosshard PP, Kronenberg A, Zbinden R, et al. Etiologic diagnosis of infective endocarditis by broad-range polymerase chain reaction: a 3- year experience. Clin Infect Dis 2003; 37: Schuurman T, de Boer RF, Kooistra-Smid AMD, et al. Prospective study of use of PCR amplification and sequencing of 16S ribosomal DNA from cerebrospinal fluid for diagnosis of bacterial meningitis in a clinical setting. J Clin Microbiol 2004; 42: Deutch S. Pedersen LN, Podenphant L, et al. Broad-range real time PCR and DNA sequencing for the diagnosis of bacterial meningitis. Scand J Infect Dis 2006; 38: Fenollar F, Roux V, Stein A, et al. Analysis of 525 samples to determine the usefulness of PCR amplification and sequencing of the 16S rrna gene for diagnosis of bone and joint infections. J Clin Microbiol 2006; 44: Jalava J, Skurnik M, Toivanen P, et al. Bacterial PCR in the diagnosis of joint infection. Ann Rheum Dis 2001; 60: CID 2007:44 (15 April) MEDICAL MICROBIOLOGY