SEARCH

SEARCH BY CITATION

Keywords:

  • Discrimination;
  • MALDI-TOF mass spectrometry;
  • S. mitis ;
  • S. pneumoniae ;
  • VGS ;
  • viridans group streptococci

Abstract

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. Transparency Declaration
  9. References

Accurate species-level identification of alpha-hemolytic (viridans) streptococci (VGS) is very important for understanding their pathogenicity and virulence. However, an extremely high level of similarity between VGS within the mitis group (S. pneumoniae, S. mitis, S. oralis and S. pseudopneumoniae) often results in misidentification of these organisms. Earlier, matrix-assisted laser desorption ionization–time of flight mass spectrometry (MALDI-TOF MS) has been suggested as a tool for the rapid identification of S. pneumoniae. However, by using Biotyper 3.0 (Bruker) or Vitek MS (bioMérieux) databases, Streptococcus mitis/oralis species can be erroneously identified as S. pneumoniae. ClinProTools 2.1 software was used for the discrimination of MALDI-TOF mass spectra of 25 S. pneumoniae isolates, 34 S. mitis and three S. oralis. Phenotypical tests and multilocus gene typing schemes for the S. pneumoniae (http://spneumoniae.mlst.net/) and viridans streptococci (http://viridans.emlsa.net/) were used for the identification of isolates included in the study. The classifying model was generated based on different algorithms (Genetic Algorithm, Supervised Neural Network and QuickClassifier). In all cases, values of sensitivity and specificity were found to be equal or close to 100%, allowing discrimination of mass spectra of different species. Three peaks (6949, 9876 and 9975 m/z) were determined conferring the maximal statistical weight onto each model built. We find this approach to be promising for viridans streptococci discrimination.


Introduction

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. Transparency Declaration
  9. References

Streptococcus pneumoniae, Streptococcus mitis, Streptococcus pseudopneumoniae and Streptococcus oralis are closely related species of viridans group streptococci (VGS) colonizing the human oral cavity; however, their pathogenic properties differ significantly. While S. pneumoniae is a major human pathogen associated with community-acquired pneumonia, meningitis and otitis media, other representatives of this group are commensals and can cause infections only when they gain access to the blood stream or in an immunocompromized host. Clinical laboratories must be able to accurately differentiate S. pneumoniae from other VGS commonly found in clinical samples to facilitate appropriate antimicrobial therapy. However, conventional phenotypic methods such as colony morphology, bile solubility and optochin susceptibility testing, as well as commercial systems (API 20 Strep and Vitek 2; bioMe′rieux, Marcy l'Etoile, France), do not always provide accurate identification [1-4].

A variety of genes have been used as targets for a PCR-based discrimination of VGS: pneumolysin (ply) [5], autolysin (lytA) [6], pneumococcal surface antigen A (psaA) [7], and the DNA fragment of unknown function Spn9802 [8]. However, the application of this strategy is complicated by the reports that Streptococcus mitis and Streptococcus oralis harbour the genes encoding autolysin and pneumolysin [9-11]. Sequence analysis of 16S rRNA genes also cannot be applied for discrimination of VGS due to 99% similarity in nucleotide composition of 16S rRNA genes in these bacteria [12, 13]. Sequence analysis of rnpB (RNA subunit of endonuclease P), sodA (manganese-dependent superoxide dismutase), tuf (elongation factor Tu), groESL (heat shock proteins) and rpoB (b-subunit of bacterial RNA polymerase) genes is more promising but the data are insufficient for final conclusions.

At present the most reliable identification of VGS can be achieved by using multilocus sequence analysis (MLSA) [14]. This approach is based on the construction of a phylogenetic tree on the concatenated sequences of seven housekeeping gene fragments and displaying the clonal relationship between unknown strains under study and strains stored in the public databases. http://viridans.emlsa.net/ allows streptococcal strains to be assigned to species within the VGS. Correct S. pneumoniae identification can also be achieved using the mlst.net database, which was designed mainly for intraspecies typing of pneumococci [15]. Unfortunately, MLSA is relatively expensive and time-consuming.

Recently direct bacterial profiling by means of matrix-assisted laser desorption ionization–time of flight mass spectrometry (MALDI-TOF MS) has been suggested as a tool for the rapid identification of different bacteria. Unfortunately, using the Biotyper 3.0 (Bruker Daltonics, Bremen, Germany) database Streptococcus mitis/oralis species can be erroneously identified as S. pneumoniae, due to the exceptional similarity of their mass spectra [16-19].

In this study, a number of mathematical classifying algorithms were applied to sort mass spectra of VGS by the selection of a set of mass peaks discriminating phenotypically and genetically characterized isolates of different VGS species. From these classes, a number of classifying models were generated and compared by the parameters of sensitivity and specificity. Finally, successful models were blind tested on the randomly selected S. pneumoniae and S. mitis strains.

Methods

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. Transparency Declaration
  9. References

Strains

A total of 62 VGS were included in the study. Thirty-four of them were identified in our previous study as S. mitis and three as S. oralis [20]. Twenty-five isolates were identified as S. pneumoniae based on bile solubility, optochin susceptibility and positive results of the ‘Slidex® pneumo-kit’ (bioMerieux®, Marcy-l'Etoile, France) test. For S. pneumoniae isolates, serotypes were determined using antisera obtained from the Staten Serum Institute (Copenhagen, Denmark) according to the manufacturer's recommendations. All VGS isolates were stored at −80°C in CRYOBANK (Copan, Italy) vials.

Before mass spectrometry analysis and genetic studies, isolates were subcultured on Columbia agar (Oxoid Ltd, Basingstoke, UK) with addition of 5% sheep blood (overnight incubation at 35°C in air with 5% CO2), and phenotypic tests (bile solubility, optochin susceptibility and the ‘Slidex® pneumo-kit’ test) were repeated. An optochin susceptibility test was carried out using the standard diagnostic optochin discs (Research Centre on Pharmacotherapy, St Petersburg, Russia), in air with 5% CO2.

Gene analysis: MLST and MLSA schemes

For the genetic manipulations, streptococcal DNA was extracted using the ‘DNA-express’ kit (‘Lytech’ Ltd, Moscow, Russia), in accordance with the manufacturer's instructions. MLST and MLSA were performed as described by Enright and Spratt [15] and by Bishop et al. [14], respectively. Results were analysed using the MLST (http://www.mlst.net) and MLSA (http://viridans.emlsa.net/) databases.

MALDI TOF data acquisition

For MALDI-TOF mass spectrometry analysis, two to three isolated colonies of fresh bacterial cultures (18 h) were picked up with a sterile 1.0 μl plastic loop (FL Medical, Torreglia, Italy) and suspended in 300 μL of pure water (Fluka, St. Louis, MO, USA). After precipitation with ethanol (900 μl), the pellet was treated with a formic acid/acetonitrile mixture. Bacterial extracts (1 μl) were spotted onto a MALDI sample target, overlaid with 1 μL of the matrix (saturated solution of alpha-cyano-4-hydroxy cinnamic acid (CHCA) in 50% acetonitrile/2.5% TFA), and dried in the air. Mass spectra were recorded on a Microflex MALDI-TOF mass spectrometer (Bruker Daltonics) equipped with a N2 337 nm laser. Not less than four mass spectra were obtained from each isolate, with a total of 250 shots per spectrum (50 shots at each of five different spot positions). For visual spectra inspection the FlexAnalysis 2.4 software (Bruker Daltonics) was used. BioTyper 3.0 software (Bruker Daltonics) was used for the comparison and identification of spectra.

Data processing with ClinProTools software

ClinProTools 2.1 software (Bruker Daltonics) was used for the recognition of peptide patterns. This software was originally provided as a part of the ClinProt system solution for searching for biomarkers in human body fluids (serum, plasma, urine, saliva, cerebral spinal fluid, etc.) associated with the different diseases. It allows a measurement and visualization of peptide and protein differences in mass spectra of different samples [21]. Spectra pretreatment, peak picking and peak calculation operations were performed using the standard mode. Classification models were generated using Genetic Algorithm (GA), Supervised Neural Network (SNN) and QuickClassifier (QC) algorithms.

For each model, recognition capability (RC), relative number of correctly classified spectra for the given model under the constraint that all tested data are previously used for the determination of the model, and cross-validation (CV), a quantitative measure for the reliability of a model that can be used to predict how a model will behave in the future, were calculated.

External validation was carried out on the blindly selected isolates; the values of sensitivity and specificity were determined.

Results

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. Transparency Declaration
  9. References

MLST and MLSA analysis

MLST was performed on 25 isolates of S. pneumoniae, 34 of S. mitis and three of S. oralis. All S. pneumoniae isolates were susceptible to optochin; nine of them belong to serotype 23F, five to 6B, five to 19F, two to 18, and serotypes 14, 19A, 9L and 35F were presented by one isolate each.

The dendrogram depicting clonal relationships between isolates is presented in Fig. 1(a). For two S. oralis isolates, it failed to amplify the spi gene locus, so these strains were discarded.

image

Figure 1. (a) Phylogenetic trees constructed on the concatenated sequences of six housekeeping gene fragments (aroE, gdh, gki, recP and spi и xpt) and displaying the clonal relationship between the strains under study and 6497 S. pneumoniae strains stored in the S. pneumoniae/MLST database (http://spneumoniae.mlst.net/). A dendrogram was constructed by the neighbour-joining algorithm; the evolutionary distances were computed using the p-distance method. (b) Phylogenetic trees constructed on the concatenated sequences of seven housekeeping gene fragments (map, pfl, ppaC, pyk, rpoB, sodA and tuf) and displaying the clonal relationship between the strains under study and 244 strains of the mitis group (S. mitis, S. pneumoniae, S. pseudopneumoniae and S. oralis) stored in the eMLSA database (http://viridans.emlsa.net/). A dendrogram was constructed by a neighbour-joining algorithm; the evolutionary distances were computed using the p-distance method. In both figures, optochin-susceptible strains are marked by red filled circles, while optochin-resistant strains are marked by green filled circles. Red unfilled circles indicate the strains that were primarily characterized as optochin susceptible, but optochin susceptibility has not been confirmed for repeated tests (see Ikryannikova et al., [20]). Asterisks indicate two S. oralis strains for which we were unable to amplify the spi locus in the MLST scheme.

Download figure to PowerPoint

Most isolates of the ‘pneumococci’ group (24 of 25) fell into the pneumococcal cluster (Fig. 1a, on top of the picture), while mitis and oralis isolates formed the separate (‘non-pneumococcal’) cluster. For isolates of the ‘non-pneumococcal’ cluster, alleles of all gene fragments were found to be at least one or more per cent different from any known for pneumococcus allele. Only one strain (43 741) of the ‘pneumococci’ group unexpectedly fell into the ‘non-pneumococcal’ cluster. This strain was stably optochin susceptible and serotyped as 19F. Then we built the phylogenetic tree using ClonalFrame v. 1.1 algorithms, which allow us to take into account possible recombinations in bacterial populations; the distribution of strains appeared to be the same (data not shown).

Randomly selected isolates from pneumococcal (n = 13) and from non-pneumococcal (n = 22) clusters, including two S. oralis isolates negative on the spi locus, were additionally analysed by the MLSA scheme.

A dendrogram based on concatenated sequences of seven housekeeping genes from isolates included in the study and 244 strains of the mitis group (S. mitis, S. pneumoniae, S. pseudopneumoniae and Soralis) stored in the http://viridans.emlsa.net/ database was created (Fig. 1b). All isolates from the ‘pneumococcal’ cluster as defined by MLST analysis fell into the S. pneumoniae cluster on MLSA analysis, while 21 of 22 isolates from the ‘non-pneumococcal’ cluster, including isolate 43 741, fell into the S. mitis group. It was one isolate from ‘non-pneumococcal’ cluster, which grouped to the S. oralis branch, along with two S. oralis isolates for which spi locus was failed to amplify in MLST scheme (these are marked by asterisks in Fig 1b).

Part of the isolates identified by MLSA as S. pneumoniae (n = 13) or Smitis (n = 21) formed the main groups for the generating of classifying models. Three strains attributed to S. oralis formed the ‘oralis’ group. Other isolates included in the study (11 S. pneumoniae and 14 S. mitis) formed the auxiliary groups for the validation of models.

Acquisition of mass spectrometry data and generating of the classifying model

About 120 peaks with signal-to-noise ratios greater than 5 were detected between m/z 2000 and 20 000 in the mass spectra of 62 isolates under study.

All isolates of the ‘pneumococcal’ cluster (see Fig. 1) were identified by BioTyper sofware as S. pneumoniae with a spectral score of ≥2.3. Most of the isolates of the ‘non-pneumococcal’ cluster were identified as S. pneumoniae with a score of ≥2.0, and some isolates were identified as S. pneumoniae or S. oralis, or S. pseudopneumoniae, or S. cristatus, with scores not less than 2.0.

Three isolates of the S. oralis cluster (Fig. 2) were also attributed to S. pneumoniae/S. oralis/S. pseudopneumoniae/S. cristatus species.

image

Figure 2. The mass peak of 6949 m/z is presented in S. mitis isolates (green line) and is not presented in S. pneumoniae isolates (a). 2D distribution diagram for two of the three best peaks (6949 and 9975 m/z): ellipses correspond to 95% confidence interval, red and green spots denote S. pneumoniae and S. mitis strains, respectively (b).

Download figure to PowerPoint

Further, mass spectra obtained from the strains of the main group, 13 of S. pneumoniae and 21 of S. mitis, were compared using ClinProTools 2.1 software. A number of different classification models were generated using GA, SNN and QC algorithms. RC, CV, sensitivity and specificity of the models are presented in Table 1.

Table 1. Parameters of the classifying models generated on the basis of different algorithms
ModelRecognition capabilityCross-validationSensitivity,%Specificity,%
  1. a

    A model generated by three peaks (6949, 9876 and 9975 m/z) conferring the significant weight onto the classification.

GA100.00100.00100.098.6
SNN100.0098.65100.0100.0
QC100.0099.46100.0100.0
QC-3a98.4098.4497.9100.0

In the GA model, the number of peaks can be defined by user. Varying this parameter, we created a model in which 17 peaks gave 100% values of RC and CV (Table 1). The number of peaks in both the other models, SNN and QC, was detected automatically. Six peaks were determined in the SNN model and five in the QC model. A CV value of 100% was reached for the GA model only, while in other models it was slightly less. Note that cross-validation is a variant of an automatic validation during model generation: a small part of all spectra is left out in model generation and cluster analysis, then these spectra are classified, and the number of correct and wrong class predictions is determined. This procedure is repeated several times, and the correct and wrong class predictions are accumulated for each class.

There were three peaks (6949, 9876 and 9975 m/z) conferring the maximal statistical weights, which were found in each model generated. Fig. 2(a) shows an example of the peak of m/z 6949, which contributes significantly to the species discrimination model: one can see that this minor peak arises in almost all S. mitis isolates mass spectra, and is absent in the S. pneumoniae mass spectra. Two-dimensional distribution for the two peaks (6949 and 9975 m/z) is presented in Fig. 2(b). In accordance with this diagram, even these two peaks were enough for the convincing discrimination of streptococcal species. Actually, forcing only these three peaks (6949, 9876 and 9975 m/z) into a model (designated as QC-3 in Table 1) was followed by only a minor decrease of RC and CV values (98.40% and 98.44%, respectively).

Each of the models was evaluated blindly using 25 isolates of the auxiliary group (11 of pneumococci and 14 of S. mitis). All three models showed almost 100% values of sensitivity and specificity (Table 1).

The number of S. oralis isolates (n = 3) was insufficient to create a reliable model. However, brief computations showed that all classifiers were able to discriminate S. pneumoniae from S. oralis isolates, but not S. mitis from S. oralis isolates.

Discussion

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. Transparency Declaration
  9. References

Accurate species-level identification of VGS isolates is important both in terms of practice and for understanding the pathogenic mechanisms of the particular species. For mitis group streptococci, species identification is especially difficult. Numerous approaches utilizing the different phenotypic species traits do not give a good sensitivity and cannot be considered reliable.

Introduction of MALDI-TOF-based bacterial protein profiling significantly improved the process of identifying microorganisms in routine practice. This approach is based on the investigation of complex mass spectra of peptides and small proteins, which contain unique m/z ‘signatures’ for different microorganisms due to the inherent variations in their masses, and allowed successful identification and discrimination of a broad spectrum of organisms. First attempts to discriminate VGS using the MALDI-TOF MS were promising [16, 19, 22], but later it become clear that Biotyper 3.0 (Bruker) is unable to solve the problem [17, 18, 23, 24]. Very recently, a novel Vitek MS (bioMérieux) system was introduced and showed relatively good results [25].

We hypothesized that application of alternative algorithms to the analysis of mass spectra may be more fruitful and tried to summarize the differences in mass spectra, instead of looking for the similarities between them. This task can be carried out with ClinProTools software. Previously, few attempts at intraspecies differentiation of some microorganisms using a similar approach were made. Thus, specific mass peaks were identified as ‘clonal-identifying’ markers for two main phylogenetic lineages, ST-1 and ST-17, of Group B streptococcus, which are associated with major causes of meningitis and late-onset diseases in neonates [26]. However, no ‘clonal’ biomarker ions were identified for the other three major STs. Mass spectrometry profiles followed by ClinProTools software analysis were exploited to find a reproducible model able to identify the Pantone-Valentine leukocidin (PVL) in Staphylococcus aureus strains [27]. The peak at 4448 m/z was found to be the most relevant peak for differentiating between PVL-producing and non-PVL-producing strains; however, subsequent works have shown that this peak seems to be independent of the presence of PVL [28, 29]. A good example of differentiation of cfiA-positive (B. fragilis division I) and cfiA-negative (B. fragilis division II) strains using ClinProTools software was demonstrated in the work of Nagy et al. [30].

Our data confirm that, in spite of the extraordinary similarity of their mass spectra, there are a number of minor mass peaks allowing us to discriminate S. pneumoniae from S. mitis. Defining an appropriate set of these peaks, we can differentiate these species with a sensitivity and specificity equal or close to 100%. Three mass peaks (6949, 9876 and 9975 m/z) seem to be of interest as contributing most to the models. Note that the similar peak, 6954 m/z, was revealed as specific for the non-pneumococcal mitis group species in the very recently published work of Werno et al. [31] regarding the manual inspection of VGS mass spectra. Unfortunately, we still cannot identify proteins attributed to these peaks, using the available methods, but it may be a challenge for future investigations. Also, it would be extremely interesting to build the discriminating models for all members of the mitis group (i.e. S. pneumoniae, S. mitis, S. oralis and S. pseudopneumoniae).

Acknowledgements

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. Transparency Declaration
  9. References

This work was supported by the BRUKER Daltonic company (Germany) and was partially supported by RFBR (research project No. 08-04-01597-a). We would also like to thank V. A. Karpov for the oligonucleotide primer synthesis and M. M. Chukin and T. A. Akopian for their assistance in the sequence instrumentation. This publication made use of the http://viridans.emlsa.net/ database (website http://viridans.emlsa.net/) and the S. pneumoniae MLST database (website http://spneumoniae.mlst.net/) held at the Department of Infectious Disease Epidemiology, Imperial College, London.

References

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. Transparency Declaration
  9. References
  • 1
    Teles C, Smith A, Ramage G, Lang S. Identification of clinically relevant viridans group streptococci by phenotypic and genotypic analysis. Eur J Clin Microbiol Infect Dis 2011; 30: 243250.
  • 2
    Simoes A, Valente C, de Lencastre H, Sa-Leao R. Rapid identification of noncapsulated Streptococcus pneumoniae in nasopharyngeal samples allowing detection of co-colonization and reevaluation of prevalence. Diagn Microbiol Infect Dis 2011; 71: 208216.
  • 3
    Song J-H, Baek JY. Incorrect identification of Streptococcus pneumoniae and its effect on antimicrobial resistance rates. Int J Antimicrob Agents 2009; 33: 9394.
  • 4
    Richter SS, Heilmann KP, Dohrn CL, Riahi F, Beekmann SE, Doern GV. Accuracy of phenotypic methods for identification of Streptococcus pneumoniae isolates included in surveillance programs. J Clin Microbiol 2008; 46: 21842188.
  • 5
    Toikka P, Nikkari S, Ruuskanen O, Leinonen M, Mertsola J. Pneumolysin PCR-based diagnosis of invasive pneumococcal infection in children. J Clin Microbiol 1999; 37: 633637.
  • 6
    McAvin JC, Reilly PA, Roudabush RM et al. Sensitive and specific method for rapid identification of Streptococcus pneumoniae using real-time fluorescence PCR. J Clin Microbiol 2001; 39: 34463451.
  • 7
    Morrison KE, Lake D, Crook J et al. Confirmation of psaA in all 90 serotypes of Streptococcus pneumoniae by PCR and potential of this assay for identification and diagnosis. J Clin Microbiol 2000; 38: 434437.
  • 8
    Suzuki N, Yuyama M, Maeda S, Ogawa H, Mashiko K, Kiyoura Y. Genotypic identification of presumptive Streptococcus pneumoniae by PCR using four genes highly specific for S. pneumoniae. J Med Microbiol 2006; 55: 709714.
  • 9
    Seki M, Yamashita Y, Torigoe H, Tsuda H, Sato S, Maeno M. Loop-mediated isothermal amplification method targeting the lytA gene for detection of Streptococcus pneumoniae. J Clin Microbiol 2005; 43: 15811586.
  • 10
    Verhelst R, Kaijalainen T, De Baere T et al. Comparison of Five Genotypic techniques for identification of optochin-resistant pneumococcus-like isolates. J Clin Microbiol 2005; 41: 35213525.
  • 11
    Whatmore AM, Efstratiou A, Pickerell AP et al. Genetic relationships between clinical isolates of Streptococcus pneumoniae, Streptococcus oralis, and Streptococcus mitis: characterization of “atypical” pneumococci and organisms allied to S. mitis harboring S. pneumoniae virulence factor-encoding genes. Infect Immun 2000; 68: 13741382.
  • 12
    Haanpera M, Jalava J, Huovinen P, Meurman O, Rantakokko-Jalava K. Identification of alpha-hemolytic streptococci by pyrosequencing the 16S rRNA gene and by use of VITEK 2. J Clin Microbiol 2007; 45: 762770.
  • 13
    Arbique JC, Poyart C, Trieu-Cuot P et al. Accuracy of phenotypic and genotypic testing for identification of Streptococcus pneumoniae and description of Streptococcus pseudopneumoniae sp. J Clin Microbiol 2004; 42: 46864696.
  • 14
    Bishop CJ, Aanensen DM, Jordan GE, Kilian M, Hanage WP, Spratt BG. Assigning strains to bacterial species via the internet. BMC Biol 2009; 7: 3.
  • 15
    Enright MC, Spratt BG. A multilocus sequence typing scheme for Streptococcus pneumoniae: identification of clones associated with serious invasive disease. Microbiology 1998; 144: 30493060.
  • 16
    Friedrichs C, Rodloff AC, Chhatwal GS, Schellenberger W, Eschrich K. Rapid identification of viridans streptococci by mass spectrometric discrimination. J Clin Microbiol 2007; 45: 23922397.
  • 17
    La Scola B, Raoult D. Direct identification of bacteria in positive blood culture bottles by Matrix-Assisted Laser Desorption Ionisation Time-of-Flight mass spectrometry. PLoS ONE 2009; 4: e8041.
  • 18
    Stevenson LG, Drake SK, Murray PR. Rapid identification of bacteria in positive blood culture broths by Matrix-Assisted Laser Desorption Ionization–Time of Flight mass spectrometry. J Clin Microbiol 2010; 48: 444447.
  • 19
    Ferroni A, Suarez S, Beretti J-L et al. Real-time identification of bacteria and Candida species in positive blood culture broths by Matrix-Assisted Laser Desorption Ionization–Time of Flight mass spectrometry. J Clin Microbiol 2010; 48: 15421548.
  • 20
    Ikryannikova LN, Lapin KN, Malakhova MV et al. Misidentification of alpha-hemolytic streptococci by routine tests in clinical practice. Infect Genet Evol 2011; 11: 17091715.
  • 21
    Ketterlinus R, Hsieh S-Y, Teng Sh-H, Lee H, Pusch W. Fishing for biomarkers: analyzing mass spectrometry data with the new ClinProTools software. Biotechniques 2005; 38: S37S40.
  • 22
    Rupf S, Breitung K, Schellenberger W, Merte K, Kneist S, Eschrich K. Differentiation of mutans streptococci by intact cell matrixassisted laser desorption/ionization time-of-flight mass spectrometry. Oral Microbiol Immunol 2005; 20: 267273.
  • 23
    Risch M, Radjenovic D, Han JN, Wydler M, Nydegger U, Risch L. Comparison of MALDI TOF with conventional identification of clinically relevant bacteria. Swiss Med Wkly 2010; 140: w13095.
  • 24
    Welker M, Moore ERB. Applications of whole-cell matrix-assisted laser-desorption/ionization time-of-flight mass spectrometry in systematic microbiology. Syst Appl Microbiol 2011; 34: 211.
  • 25
    Martiny D, Busson L, Wybo I, Haj RAE, Dediste A, Vandenberga O. Comparison of the Microflex LT and Vitek MS systems for routine identification of bacteria by Matrix-Assisted Laser Desorption Ionization–Time of Flight mass spectrometry J. J Clin Microbiol 2012; 50: 1313.
  • 26
    Lartigue M-F, Kostrzewa M, Salloum M et al. Rapid detection of “highly virulent” Group B Streptococcus ST-17 and emerging ST-1 clones by MALDI-TOF mass spectrometry. J Microbiol Methods 2011; 86: 262265.
  • 27
    Bittar F, Ouchenane Z, Smati F, Raoult D, Rolain J-M. MALDI-TOF-MS for rapid detection of staphylococcal Panton–Valentine leukocidin. Int J Antimicrob Agents 2009; 34: 467470.
  • 28
    Szabados F, Becker K, von Eiff C, Kaase M, Gatermann S. The matrix-assisted laser desorption/ionisation time-of-flight mass spectrometry (MALDI-TOF MS)-based protein peaks of 4448 and 5302 Da are not associated with the presence of Panton-Valentine leukocidin. Int J Med Microbiol 2011; 301: 5863.
  • 29
    Carbonnelle E, Benito Y, Lina G, Nassif X, Vandenesch F, Laurent F. Detection of Panton–Valentine toxin in Staphylococcus aureus by mass spectrometry directly from colony: time has not yet come. Int J Antimicrob Agents 2010; 36: 187194.
  • 30
    Nagy E, Becker S, Soki J, Urban E, Kostrzewa M. Differentiation of division I (cfiA-negative) and division II (cfiA-positive) Bacteroides fragilis strains by matrix-assisted laser desorption/ionization time-of-flight mass spectrometry. J Med Microbiol 2011; 60: 15841590.
  • 31
    Werno AM, Christner M, Anderson TP, Murdoch DR. Differentiation of Streptococcus pneumoniae from non-pneumococcal streptococci of the Streptococcus mitis group by MALDI-TOF MS. J Clin Microbiol 2012; 50: 28632867.