Matrix-assisted laser desorption ionization-time of flight mass spectrometry (MALDI-TOF MS) allows rapid and reliable identification of microorganisms, particularly clinically important pathogens.
We compared the identification efficiency of MALDI-TOF MS with that of Phoenix®, API® and 16S ribosomal DNA sequence analysis on 1,019 strains obtained from routine diagnostics. Further, we determined the agreement of MALDI-TOF MS identifications as compared to 16S gene sequencing for additional 545 strains belonging to species of Enterococcus, Gardnerella, Staphylococcus, and Streptococcus. For 94.7% of the isolates MALDI-TOF MS results were identical with those obtained with conventional systems. 16S sequencing confirmed MALDI-TOF MS identification in 63% of the discordant results. Agreement of identification of Gardnerella, Enterococcus, Streptococcus and Staphylococcus species between MALDI-TOF MS and traditional method was high (Crohn's kappa values: 0.9 to 0.93).
Citation: Benagli C, Rossi V, Dolina M, Tonolla M, Petrini O (2011) Matrix-Assisted Laser Desorption Ionization-Time of Flight Mass Spectrometry for the Identification of Clinically Relevant Bacteria. PLoS ONE 6(1): e16424. https://doi.org/10.1371/journal.pone.0016424
Editor: Niyaz Ahmed, University of Hyderabad, India
Received: August 4, 2010; Accepted: December 15, 2010; Published: January 25, 2011
Copyright: © 2011 Benagli et al. This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.
Funding: The authors have no support or funding to report.
Competing interests: The authors have declared that no competing interests exist.
Matrix-assisted laser desorption ionization-time of flight mass spectrometry (MALDI-TOF MS) is rapidly attracting the interest of microbiologists working in the routine labs, because of its powerful features that allow rapid and reliable identification of microorganisms.
Standardized test systems such as API® and VITEK® 2 (bioMérieux), or PHOENIX® (BD Diagnostics), complemented by traditional culture and microscopy methods, have so far been used in routine labs for the rapid identification of clinical microorganisms. With the introduction of these methods, the average time needed for a reliable and validated identification ranged from 6 h to 18 h and in the last few years, sequence analysis of small-subunit rRNAs or selected genes by PCR methods has complemented the biochemical methods, additionally decreasing throughput time and becoming in several cases the gold standard .
The recent developments of MALDI-TOF MS are rapidly changing the routine diagnostics scene. MALDI-TOF MS is a powerful method to detect and identify proteins by molecular weight determination of individual, specific fragments . The method is accurate and easy to use, allowing quick determination of molecular weights of proteins with minimal sample requirements.
MALDI-TOF MS is now widely used for the identification and characterization of clinically important microorganisms . The currently available identification databases target the identification of human pathogens  and MALDI-TOF MS represents a valid and rapid alternative to conventional methods of identification and classification of human pathogens in microbiology.
Traditionally, validation of a new identification system to be introduced in routine diagnostics consists of running parallel identifications of a large number of isolates using the new method concomitantly with set standards.
In this study we compared the identification efficiency of MALDI-TOF MS with that of Phoenix®, API® and 16S ribosomal DNA sequence analysis. In a first step we analyzed 1,019 strains obtained sequentially during three months from our routine diagnostic laboratory. In a second step, we studied in more detail 545 isolates of species belonging to the genera Enterococcus, Gardnerella, Staphylococcus, and Streptococcus and determined the agreement (and, when possible, efficiency, sensitivity and specificity) of the MALDI-TOF MS identifications as compared to 16S gene sequencing as the gold standard.
In a first step we analysed 1,019 strains obtained from the routine diagnostic lab. The results are described in Fig. 1. For 965 isolates (94.7%) the results of MALDI-TOF MS were identical with those obtained with the BD PHOENIX system and the confidence level of the MALDI-TOF MS identification was almost 100% for approximately 75% of the isolates tested. API or 16S sequencing confirmed MALDI-TOF MS identification in 63% of the discordant results. Overall, therefore, MALDI-TOF was able to identify correctly more than 98% of the isolates tested. Table 2 reports the specificity, sensitivity, PPV, NPV and efficiency values of the MALDI-TOF MS identification, compared to the classical methods, for those bacterial species for which we could analyse at least 15 isolates. With the exception of Enterobacter cloacae and Klebsiella oxytoca, for which the sensitivity values for identifications at the confidence level of at least 90% are relatively low, MALDI-TOF MS has a quite high efficiency of identification. In most cases these values are still even if the limit of acceptance of the confidence level for correct identification by MALDI-TOF MS is set at 99.9% (Table 2).
We analysed separately 76 isolates of Gardnerella, 50 of Enterococcus, and 76 of Streptococcus (mainly S. agalactiae and S. pneumoniae) by MALDI-TOF MS, API/PHOENIX and 16S sequencing. The results are described in Table 3.
For Gardnerella spp., the %AI was almost 100%. Very good concordance between 16S sequencing and MALDI-TOF MS was also observed for Enterococcus faecalis and E. faecium, for which the concordance was 100%. Results for E. gallinarum were not particularly good (44.4%), but the number of isolates investigated (9) was low. MALDI-TOF MS showed a fair performance also with Streptococcus spp., with quite high sensitivity and specificity values for S. agalactiae and S. pneumoniae (Table 4).
In a separate experiment we evaluated the efficiency of MALDI-TOF MS in the identification of Staphylococcus spp. We considered a sample of 343 staphylococci belonging to 17 species (S. aureus, S. auricularis, S. capitis, S. carnosus, S. cohnii, S. epidermidis, S. equorum, S. haemolyticus, S. hominis, S. lugdunensis, S. pasteuri, S. saprophyticus, S. schleiferi, S. sciuri, S. simulans, S. warneri, S. xylosus) and we computed sensitivity, specificity, PPV and NPV values for the four most commonly isolated taxa in our routine laboratory (S. aureus, S. epidermidis, S. hominis, and S. haemolyticus) as well as the overall %AI for all species. The outcome of the identification by MALDI-TOF MS was compared to that obtained by 16S sequencing, which for the purpose of this work was considered the gold standard, and other methods (API®, PHOENIX®) currently used in our laboratory.
Results are presented in Figure 2, Figure 3 and Figure 4. For the four species considered, MALDI-TOF MS was at least as good as the other methods in identifying the species studied, and mostly the sensitivity, specificity, PPV and NPV values were trend-wise, albeit not statistically significantly superior to those obtained for the other methods. The identification agreement between MALDI-TOF MS and the gold standard used, as represented by the Crohn's kappa values, ranged between 0.9 and 0.93, indicating an almost perfect agreement. The identification efficiency was also always over 90%. The %AI over the whole sample investigated was 88.5%.
In this study, MALDI-TOF MS has proven to be a fast, accurate and reliable technique for the identification of clinically relevant bacteria. We have observed an almost perfect agreement between identifications obtained by MALDI-TOF MS and those provided by conventional, biochemical methods. When discordant results among mass spectrometry and biochemical methods were observed, sequencing most often confirmed MALDI-TOF MS identification.
Identification of Gardnerella species by MALDI-TOF MS has proven to be reliable, needing no additional confirmations by other methods. The same applied to E. faecium and E. faecalis, both regularly identified by MALDI-TOF MS at confidence levels of almost 100% and with a %AI with 16S sequencing of 100%. We were not able to study enough samples of E. caselliflavus and E. gallinarum to report reliable values for MALDI-TOF MS identification of these two species; nevertheless, our present daily experience places MALDI-TOF MS reliability at least at the same level as PHOENIX® or API® (data not shown).
Streptococcus species are notoriously difficult to be identified and often 16S sequence data are not informative enough to distinguish species. Glazunova et al.  have produced phylogenies of this genus inferred from rpoB, sodA, gyrB and groEL sequence comparisons that were more discriminative than those derived from 16S rRNA. This would explain the rather low %AI (71.4%) observed for Streptococcus spp. other than S. agalactiae and S. pneumoniae. MALDI-TOF MS, however, provides reliable identification of species belonging to the viridans  and mutans complex ; in the latter case MALDI-TOF MS was useful also for differentiation at the subspecies level. For group A Streptococcus (S. pyogenes), MALDI-TOF MS was able to distinguish isolates from cases of necrotizing fasciitis from those associated with non-invasive infections, despite they shared the same emm type . Additional work is needed for species in this genus, however, for which a careful analysis of regional strains could also be very important. For Streptococcus and Staphylococcus species we have observed regional differences in isolates characteristics and we are currently establishing a local database for both genera, to define new SuperSpectra that should allow rapid and more reliable identification of local isolates belonging to this variable taxonomic group.
Dubois et al.  have already shown that MALDI-TOF MS is a powerful and reliable method to identify Staphylococcus species. Most Staphylococcus species are correctly identified by MALDI-TOF MS with a high level (>90%) of confidence. According to Bernardo et al.  MALDI-TOF MS can be used for rapid identification of clonal strains of S. aureus; we are currently investigating the possibility of exploiting this feature to track nosocomial or community acquired methicillin resistant S. aureus strains. We cannot exclude that in the near future MALDI-TOF MS may allow discriminating methicillin resistant from sensitive Staphylococcus isolates, as suggested by some authors , , but, probably because of the polyphyletic nature of this resistance , we have not yet achieved a robust typing at this level.
This is the first study that has tried to quantify the efficiency of MALDI-TOF MS as compared to currently used methods. We also used a mass spectrometer and a database different from that described in previous publications and we found an almost perfect concordance of our results with those so far reported in the literature. We have been able to show that MALDI-TOF MS represents a rapid, reliable and cost-effective identification technique for clinically relevant bacteria. The method, however, has some limitations. Reliable identification of organisms by this technique is possible only if the spectral database contains data from strains that have been carefully characterised by sequencing specific and informative gene regions (e.g. 16S RNA, gyrB, rpoB, or hsp60 genes for bacteria and ITS regions for fungi). Calibration and selection of internal standards is needed when the work aims at intraspecific characterization. Finally, for a conclusive analysis of region specific strains, careful internal calibration and preparation of locally adapted databases should be considered when geographic variation in the genotypic and phenotypic expression of some taxa (e.g., Streptococcus or Staphylococcus) is expected.
Overall, MALDI-TOF has found a well-deserved place in bacterial taxonomy. Problems related to the analysis of mixed cultures, presently not yet fully solved, are likely to be overcome in the near future, and attempts to identify pathogens directly in the clinical samples have already yielded promising results . Susceptibility testing of fungi and bacteria by MALDI-TOF may also be soon possible . MALDI-TOF MS may thus become the method of choice for the identification of clinically and environmentally relevant organisms.
Materials and Methods
Bacterial isolates were received during a three-month period from our bacteriological routine laboratory. Isolates were collected in sequential order and no selection criterion was applied to their choice. In a first validation step we analysed 1,019 isolates belonging to 27 bacterial genera (Table 1). Subsequently we studied 202 additional isolates of Enterococcus, Gardnerella and Streptococcus as well as 343 isolates that belonged to 17 Staphyloccoccus species.
Isolates were cultured during 24–48 h at 35°C on 5% sheep blood agar, chocolate agar, Mac Conkey agar, and Mueller Hinton agar supplemented with 5% sheep blood (bioMérieux, Lyon, France). Growing colonies were checked for mixed cultures, transferred to a new plate and the isolates growing in pure culture were subsequently conserved in 7% skimmed milk at −80°C until use.
Identification by biochemical methods and sequencing
After Gram staining and determination of catalase and oxidase activities, isolates were identified using PHOENIX® identification cards (BD Diagnostics, Sparks, MD, USA) or API® identification strips (bioMérieux, Lyon, France), both according to the manufacturer's instructions. We also used Slidex Staph Plus (bioMérieux, Lyon, France) for Staphylococcus aureus identification. Haemolytic streptococci were identified based on the combination of colony morphology, Gram staining and rapid latex agglutination test (Streptex; Remel, Lenexa, KS).
Isolates that yielded discrepant results between routine and MALDI-TOF MS identifications were subjected to partial 16S rRNA gene sequencing. Unequivocal identification was defined as the highest sequence homology (99%) with a unique species sequence in GenBank.
DNA extraction was performed with the QIAamp DNA Mini kit (QIAGEN AG, Switzerland) according to the manufacturer's instructions.
Primers used for the amplification of the partial 16S rRNA gene sequence were UNI16SRNA-L (nucleotide sequence 5′-ATTCTTAGAGTTTGATCATGGCTCA-3′) and UNI16SRNA-R (nucleotide sequence 5′-ATGGTACCGTGTGACGGGCGGTGTGTA-3′), which allowed the amplification of a 1400 bp DNA fragment , .
PCR thermal cycling conditions were 5 min at 95°C for 1 cycle, followed by 35 cycles of 30 sec at 94°C, 30 sec at 52°C, and 1 min at 72°C. The last cycle was performed at 72°C and lasted 10 min. DNA purification was performed using NucleoSpin® (Cat. No. 740609.250) according to the instructions for direct purification of PCR products. We quantified the amplified and purified DNA fragments before the sequencing reaction using a NANO DROP® ND-1000 spectrophotometer (Thermo Fisher Scientific, Houston, USA).
Sequencing reactions were carried out using Big Dye® Terminator v1.1 Cycle Sequencing Kit (Applied Biosystems, Rotkreuz, Switzerland) with a 15 µl total volume composed of 2 µl Big Dye®Terminator, 3 µl Big Dye®buffer, 3 µl primer 1 µM, 6 µl H2O and 1 µl DNA sample.
For the sequencing reaction the amplification cycle was 10 sec at 96°C, 5 sec at 50°C, and 4 min at 60°C. Sequence reactions were purified by Sephadex G-25 (Amersham Biosciences, Otelfingen, Switzerland) before sequencing on an ABI 310 Genetic Analyzer (Perkin Elmer Instrument, Applied Biosystems, Rotkreuz, Switzerland).
All samples were analyzed with a MALDI-TOF MS Axima™ Confidence spectrometer (Shimadzu-Biotech Corp., Kyoto, Japan) in positive linear mode (m/z = 2,000–20,000). A small amount of a colony of each pure culture was transferred to a FlexiMass™ target well using a disposable loop, overlaid with 0.5 µl of 2,5-dihydroxybenzoic acid matrix solution (DHB; 10 mg/ml in acetonitrile/0.1% trifluoroacetic acid 1∶1) and air-dehydrated within 1–2 min at 24–27°C.
The reference strain Escherichia coli K12 (genotype GM48) was used as a standard for calibration and as reference for quality control. Sample information such as medium and grown conditions was imported into the software Shimadzu Biotech Launchpad™, v.2.8 (Shimadzu-Biotech Corp., Kyoto, Japan). Protein mass profiles were obtained with detection in the linear positive mode at a laser frequency of 50 Hz and within a mass range from 2,000–20,000 Da. Acceleration voltage was 20 kV and extraction delay time 200 ns. A minimum of 20 laser shots per sample was used to generate each ion spectrum. For each bacterial sample, 50 protein mass fingerprints were averaged and processed. Spectra were analyzed using SARAMIS™ (Spectral Archive And Microbial Identification System, AnagnosTec GmbH, Potzdam, Germany), a software in which the identification at the species level is based on a percentage of confidence referred to reference spectra (SuperSpectra™) that contain family, genus and species specific m/z biomarkers, as described in the SARAMIS™ user manual.
Genetic data were analyzed using the software ABI Prism™ 310 Collection Genetic Analyser (Applied Biosystems, Rotkreuz, Switzerland). Alignments were performed using the BioNumerics software v.6.01 (Applied Maths, Sint-Martens-Latem, Belgium). The modular microorganism identification system AnagnosTec SARAMIS™ was used to archive and evaluate MALDI-TOF MS data. SARAMIS™ was also used to construct dendrograms to show taxonomic relationships among strains.
Agreement of identification (AI) between MALDI and classical methods (API®, PHOENIX®) was defined as the identical outcome of identification of a given isolate by all three methods. Percentage of agreement between identifications (AI%) was computed to compare the agreement of the MALDI-TOF MS identifications with those of the classical methods. 16S rRNA gene sequencing was used as a constructed gold standard only in those cases when a discordant result was observed.
When appropriate, we calculated the estimated sensitivity and specificity and the 95% confidence intervals (CI), as well as the positive predictive (PPV) and negative predictive (NPV) values compared to the constructed perfect standard, corresponding to the identification by 16S gene sequences. We calculated the estimated sensitivity and specificity by defining a positive identification by MALDI-TOF MS when the identification score was at least ≥90%. Sensitivity, specificity and efficiency were computed using DAG_Stat .
Conceived and designed the experiments: MD MT OP. Performed the experiments: CB VR. Analyzed the data: CB VR OP. Contributed reagents/materials/analysis tools: OP. Wrote the paper: CB MT OP.
- 1. Weile J, Knabbe C (2009) Current applications and future trends of molecular diagnostics in clinical bacteriology. Anal Bioanal Chem 394: 731–742.J. WeileC. Knabbe2009Current applications and future trends of molecular diagnostics in clinical bacteriology.Anal Bioanal Chem394731742
- 2. Lay JO Jr (2001) MALDI-TOF mass spectrometry of bacteria. Mass Spectrom Rev 20: 172–194.JO Lay Jr2001MALDI-TOF mass spectrometry of bacteria.Mass Spectrom Rev20172194
- 3. Cherkaoui A, Hibbs J, Emonet S, Tangomo M, Girard M, et al. (2010) Comparison of two matrix-assisted laser desorption ionization-time of flight mass spectrometry methods with conventional phenotypic identification for routine identification of bacteria to the species level. J Clin Microbiol 48: 1169–1175.A. CherkaouiJ. HibbsS. EmonetM. TangomoM. Girard2010Comparison of two matrix-assisted laser desorption ionization-time of flight mass spectrometry methods with conventional phenotypic identification for routine identification of bacteria to the species level.J Clin Microbiol4811691175
- 4. Wilkins C, Lay J (2006) Identification of microorganisms by mass spectrometry: Wiley-InterscienceC. WilkinsJ. Lay2006Identification of microorganisms by mass spectrometry: Wiley-Interscience
- 5. Glazunova OO, Raoult D, Roux V (2009) Partial sequence comparison of the rpoB, sodA, groEL and gyrB genes within the genus Streptococcus. Int J Syst Evol Microbiol 59: 2317–2322.OO GlazunovaD. RaoultV. Roux2009Partial sequence comparison of the rpoB, sodA, groEL and gyrB genes within the genus Streptococcus.Int J Syst Evol Microbiol5923172322
- 6. Friedrichs C, Rodloff AC, Chhatwal GS, Schellenberger W, Eschrich K (2007) Rapid identification of viridans streptococci by mass spectrometric discrimination. J Clin Microbiol 45: 2392–2397.C. FriedrichsAC RodloffGS ChhatwalW. SchellenbergerK. Eschrich2007Rapid identification of viridans streptococci by mass spectrometric discrimination.J Clin Microbiol4523922397
- 7. Rupf S, Breitung K, Schellenberger W, Merte K, Kneist S, et al. (2005) Differentiation of mutans streptococci by intact cell matrix-assisted laser desorption/ionization time-of-flight mass spectrometry. Oral Microbiol Immunol 20: 267–273.S. RupfK. BreitungW. SchellenbergerK. MerteS. Kneist2005Differentiation of mutans streptococci by intact cell matrix-assisted laser desorption/ionization time-of-flight mass spectrometry.Oral Microbiol Immunol20267273
- 8. Moura H, Woolfitt AR, Carvalho MG, Pavlopoulos A, Teixeira LM, et al. (2008) MALDI-TOF mass spectrometry as a tool for differentiation of invasive and noninvasive Streptococcus pyogenes isolates. FEMS Immunol Med Microbiol 53: 333–342.H. MouraAR WoolfittMG CarvalhoA. PavlopoulosLM Teixeira2008MALDI-TOF mass spectrometry as a tool for differentiation of invasive and noninvasive Streptococcus pyogenes isolates.FEMS Immunol Med Microbiol53333342
- 9. Dubois D, Leyssene D, Chacornac JP, Kostrzewa M, Schmit PO, et al. (2009) Identification of a variety of Staphylococcus species by matrix-assisted laser desorption ionization-time of flight mass spectrometry. J Clin Microbiol 48: 941–945.D. DuboisD. LeysseneJP ChacornacM. KostrzewaPO Schmit2009Identification of a variety of Staphylococcus species by matrix-assisted laser desorption ionization-time of flight mass spectrometry.J Clin Microbiol48941945
- 10. Bernardo K, Pakulat N, Macht M, Krut O, Seifert H, et al. (2002) Identification and discrimination of Staphylococcus aureus strains using matrix-assisted laser desorption/ionization-time of flight mass spectrometry. Proteomics 2: 747–753.K. BernardoN. PakulatM. MachtO. KrutH. Seifert2002Identification and discrimination of Staphylococcus aureus strains using matrix-assisted laser desorption/ionization-time of flight mass spectrometry.Proteomics2747753
- 11. Francois P, Scherl A, Hochstrasser D, Schrenzel J (2007) Proteomic approach to investigate MRSA. Methods Mol Biol 391: 179–199.P. FrancoisA. ScherlD. HochstrasserJ. Schrenzel2007Proteomic approach to investigate MRSA.Methods Mol Biol391179199
- 12. Majcherczyk PA, McKenna T, Moreillon P, Vaudaux P (2006) The discriminatory power of MALDI-TOF mass spectrometry to differentiate between isogenic teicoplanin-susceptible and teicoplanin-resistant strains of methicillin-resistant Staphylococcus aureus. FEMS Microbiol Lett 255: 233–239.PA MajcherczykT. McKennaP. MoreillonP. Vaudaux2006The discriminatory power of MALDI-TOF mass spectrometry to differentiate between isogenic teicoplanin-susceptible and teicoplanin-resistant strains of methicillin-resistant Staphylococcus aureus.FEMS Microbiol Lett255233239
- 13. Enright MC, Robinson DA, Randle G, Feil EJ, Grundmann H, et al. (2002) The evolutionary history of methicillin-resistant Staphylococcus aureus (MRSA). Proc Natl Acad Sci U S A 99: 7687–7692.MC EnrightDA RobinsonG. RandleEJ FeilH. Grundmann2002The evolutionary history of methicillin-resistant Staphylococcus aureus (MRSA).Proc Natl Acad Sci U S A9976877692
- 14. Marinach-Patrice C, Fekkar A, Atanasova R, Gomes J, Djamdjian L, et al. (2010) Rapid species diagnosis for invasive candidiasis using mass spectrometry. PLoS One 5: e8862.C. Marinach-PatriceA. FekkarR. AtanasovaJ. GomesL. Djamdjian2010Rapid species diagnosis for invasive candidiasis using mass spectrometry.PLoS One5e8862
- 15. Marinach C, Alanio A, Palous M, Kwasek S, Fekkar A, et al. (2009) MALDI-TOF MS-based drug susceptibility testing of pathogens: The example of Candida albicans and fluconazole. Proteomics 9: 4627–4631.C. MarinachA. AlanioM. PalousS. KwasekA. Fekkar2009MALDI-TOF MS-based drug susceptibility testing of pathogens: The example of Candida albicans and fluconazole.Proteomics946274631
- 16. Brosius J, Palmer ML, Kennedy PJ, Noller HF (1978) Complete nucleotide sequence of a 16S ribosomal RNA gene from Escherichia coli. Proc Natl Acad Sci U S A 75: 4801–4805.J. BrosiusML PalmerPJ KennedyHF Noller1978Complete nucleotide sequence of a 16S ribosomal RNA gene from Escherichia coli.Proc Natl Acad Sci U S A7548014805
- 17. Lane DJ (1991) 16S/23S rRNA sequencing. In: Stackebrandt E, Goodfellow M, editors. Nucleic acid techniques in bacterial systematics. Chichester: Wiley & Sons. pp. 115–175.DJ Lane199116S/23S rRNA sequencing.E. StackebrandtM. GoodfellowNucleic acid techniques in bacterial systematicsChichesterWiley & Sons115175
- 18. Mackinnon A (2000) A spreadsheet for the calculation of comprehensive statistics for the assessment of diagnostic tests and inter-rater agreement. Computers in Biology and Medicine 30: 127–134.A. Mackinnon2000A spreadsheet for the calculation of comprehensive statistics for the assessment of diagnostic tests and inter-rater agreement.Computers in Biology and Medicine30127134