Matrix-assisted laser desorption/ionization time of flight mass spectrometry (MALDI-TOF MS) and molecular techniques (16S rRNA gene sequencing) permit the identification of rare bacterial pathogens in routine diagnostics. The goal of this protocol lies in the combination of both techniques which leads to more accurate and reliable data.
There are a number of rare and, therefore, insufficiently described bacterial pathogens which are reported to cause severe infections especially in immunocompromised patients. In most cases only few data, mostly published as case reports, are available which investigate the role of such pathogens as an infectious agent. Therefore, in order to clarify the pathogenic character of such microorganisms, it is necessary to conduct epidemiologic studies which include large numbers of these bacteria. The methods used in such a surveillance study have to meet the following criteria: the identification of the strains has to be accurate according to the valid nomenclature, they should be easy to handle (robustness), economical in routine diagnostics and they have to generate comparable results among different laboratories. Generally, there are three strategies for identifying bacterial strains in a routine setting: 1) phenotypic identification characterizing the biochemical and metabolic properties of the bacteria, 2) molecular techniques such as 16S rRNA gene sequencing and 3) mass spectrometry as a novel proteome based approach. Since mass spectrometry and molecular approaches are the most promising tools for identifying a large variety of bacterial species, these two methods are described. Advances, limitations and potential problems when using these techniques are discussed.
Secure identification of rare pathogens in routine diagnostics is hampered by the fact that classical cultural and biochemical methods are cumbersome and sometimes questionable. Furthermore, a diagnostic microbiology laboratory has to process a large number of pathogens, ranging from a few hundred to several thousands, daily, which requires the use of automated systems. In addition to the management of a high daily throughput, the precise identification of bacterial species is needed. This is warranted since they differ in their antimicrobial susceptibility pattern and therefore correct identification provides the clinician with essential information to choose appropriate antibiotics (e.g., Enterococcus spp., Acinetobacter spp.) 12,43.
Automated microbial identification systems (aMIS) apply standardized sets of enzymatic reactions to characterize the metabolic properties of bacterial isolates 13,15,16,26,27. Although the cartridges used in these systems utilize a large number of different biochemical reactions, e.g., 47 in the GN card of the aMIS used in this study 52, this strategy permits secure identification only for a limited set of bacteria. Furthermore, the database, an advanced expert system, is clearly focused on detection of relevant and highly relevant bacteria of medical importance 13,15,16,36. Two further systems, widely used in laboratories, also apply this biochemical approach for bacterial identification. Recent studies demonstrate a comparable identification accuracy between the aMIS used in this study and one of the competitors (93.7% and 93.0% respectively), while the 3rd aMIS has an identification accuracy of only 82.4% on species level 35. Such discrepancies may be explained by the quality of the underlying identification data references, the versions of kits and software, differences in metabolism and proficiency of technical personnel 35,36.
Two automated MALDI-TOF MS systems (MALDI-TOF microbial identification system, mMIS) are mainly used. These systems allow for detection of a large number of bacterial species based on their protein fingerprint mass spectra. For instance, the database of the mMIS used contains 6,000 reference spectra. Identification systems based on mass spectrometry offer fast and reliable detection of a great variety of microorganisms including rare pathogens 11,48,51. To date only a few direct comparisons are available between the mMIS used in this study and its competitor 19,33. According to Daek et al. both systems provide a similar high rate of identification accuracy, but the mMIS used in this study seems to be more reliable in species identification 19.
Similarly, molecular techniques addressing well conserved but also distinct genes (e.g., 16S rDNA or rpoB) permit a clear species identification 3,22,61. Among these, the 16S rDNA is the most widely used housekeeping gene because of its presence in all bacteria 34. Its function remains unchanged and finally, with roughly 1,500 bp, it is long enough to be suitable for bio-informatics 14,34. Many researchers regard 16S rRNA gene analysis as the "gold-standard" for bacterial identification 21. This is due to the fact that few laboratories use DNA-DNA hybridization techniques to date for identification of rare or new bacteria 14,34. Additionally, more and more databases are available which can be used for 16S rRNA gene analysis 50. However, it has to be taken into account that 16S rDNA based detection systems have a limited sensitivity compared to standard PCR protocols. Moreover, the molecular approach is sophisticated, time consuming and requires highly trained personnel as well as dedicated laboratory facilities and is, therefore, not easily implemented into routine diagnostics 55. Furthermore, it has been shown that the combination of at least two different methods of bacterial identification leads to highly accurate strain identification. The combination of MALDI-TOF MS and 16S rDNA sequencing permits the identification of large numbers of different bacterial species with high accuracy. Recently the combination of MALDI-TOF MS and 16S rRNA gene analysis was presented for bacterial identification studying epidemiological questions and rare pathogens 56.
1. Extraction of Bacterial DNA
2. 16S rDNA PCR
3. DNA Agarose Gel Electrophoresis
4. Sanger Sequencing
5. MALDI-TOF MS
NOTE: The mass spectrometer uses a 337 nm N2 laser and is operated by a specific control software (Materials Table). Spectra are recorded in linear mode and a mass range between 2,000 to 20,000 Da is covered. Data interpretation and allocation of scores to the samples is carried out in real time by an analysis software (Materials Table) (Figure 1d).
MALDI-TOF MS is a novel, fast and inexpensive method for microbiological routine diagnostics. Bacterial species identification by MALDI-TOF MS produces spectra mainly composed of ribosomal proteins but also other "very conserved proteins with house-keeping functions affected to a minimal extent by environmental conditions" 17.The database of this mMIS contains a large set of reference spectra and even bacteria which are rarely found in clinical isolates can be securely identified 7,56,57. Score values show the reliability of the identified species. Scores above 2.300 represent a highly probable species identification, a score between 2.000 and 2.300 indicates a secure species identification, a score between 1.700 and 2.000 stands for a probable species identification and a score below 1.700 is non-reliable. In the case of more than one species being identified with a score above 2.000, additional tests have to be applied and this is the reason why we have combined different methods such as 16S rDNA sequencing, API and techniques addressing the bacterial morpho- and/or phenotype such as Gram stain, motility etc. In the case where the score is below 1.700 the above mentioned tests have to be used to achieve reliable species identification. A combination of MALDI-TOF MS and 16S rDNA sequencing has been demonstrated recently 56-58. It should be pointed out that clear guidelines for species determination based on 16S rDNA sequence homologies are still lacking 22,61. For practical interpretation, homologies of ≥97%, as proposed by Stackebrandt and Göbel 61, are used 56. Advances in DNA sequencing allow the determination of whole bacterial genomes by next generation techniques at relatively low cost and, therefore, in principle identification of bacteria based on their genome sequence. This technique has already been used in genome based outbreak management or in epidemiology 9,29. However, the strongest limitation today is the lack of easy to use software packages for the end user 65.
Seven examples of rare occurring bacteria, isolated from patient samples during routine diagnostics and analyzed by both MALDI-TOF MS and 16s rRNA gene sequencing are listed in Table 1. MALDI-TOF spectra of strain #1 to #6 are shown in Figure 3 and a spectrum of strain #7, Sphingobacterium spiritivorum, is shown in Figure 1e. All isolates show high scores in MALDI-TOF MS analysis and 99 to 100% 16S rDNA sequence homologies. Thus, both techniques lead to secure genus and species identification. However, as pointed out in a recent publication on comparing identification methods of Myroides sp., the combination of MALDI-TOF MS and 16S rRNA gene sequencing led to more reliable results 56. These data illustrate that a single method used may not always be sufficient to achieve a reliable identification result. The combination of two independent methods leads to a higher accuracy and expands the commercial MALDI-TOF MS database with in-house entries resulted in an unequivocal species identification 56.
Strain #1 was identified as Chryseobacterium gleum (MALDI-TOF MS score 2.490, sequence homology 99%). Chryseobacteria are Gram-negative, non-fermenting rods and related to Myroides spp. Chryseobacterium spp. are regarded as emerging pathogens, associated with septicemia, pneumonia and urinary tract infections. The genus Chryseobacterium comprises a large number of species and the best studied species is Chryseobacterium indologenes. Similar to infections caused by Myroides sp., mostly immunocompromised patients are affected 6,10,60. Strain #2 was identified as Myroides odoratimimus (MALDI-TOF MS score 2.436, sequence homology 99%) and strain #3 as Myroides odoratus (MALDI-TOF MS score 2.237, sequence homology 99%) Myroides sp. are Gram-negative, nonfermenting rods which are associated with severe diseases such as sepsis, cellulitis or pneumonia. Mostly immunocompromised patients with underlying hematologic or oncologic diseases are affected 8,18,42. Strain #4 was identified as Sphingobacterium multivorum32,71 (MALDI-TOF MS score 2.092, sequence homology 99%).
The bacteria are Gram-negative non-fermenting rods, which may cause septicemia in immunocompromised patients 5,24,44,53. In addition, a case of fatal meningoencephalitis has been reported 70. The stains #5 (MALDI-TOF MS score 2.282, sequence homology 100%) and #6 (MALDI-TOF MS score 2.289, sequence homology 100%) were identified as Wohlfahrtiimonas chitiniclastica. These bacteria are short, non-motile, Gram-negative rods which were first isolated from larvae of the parasitic fly Wohlfahrtia magnifica and described in 2008 66. These zoonotic bacteria are regarded as emerging pathogens and causative agents for bacteremia, septicemia or soft tissue infections 4,40,64. Interestingly, most patients reported were either homeless and/or suffered from alcoholism and, therefore, the occurrence of these rare pathogens may be explained by the higher rate of ectoparasites in the homeless population 4,54. Strain #7, Sphingobacterium spiritivorum (MALDI-TOF MS score 2.269, sequence homology 100%), is a rare pathogen. It may cause infections in immunocompromised patients and first human isolates were described in 1982 31,71. A recent report identifies Sphingobacterium spiritivorum as causative agent for a fatal case of bacteremia and sepsis in a patient with acute myeloid leukemia 38.
Figure 1: Workflow for nucleic acid and/or mass spectrometry based detection of clinically relevant bacteria in the medical microbiology laboratory. Starting from a pure culture (a), target DNA is amplified in a thermocycler (b). PCR amplicons are sequenced in a four capillary sequencer using Sanger technology (c). Sequencing results are determined as percent homologies of query sequences to database entries by computer assisted comparison. A second, proteomics based method, uses mass spectrometry (d) and in the center a mass spectroscopic result of strain #1 in Table 1, Sphingobacterium spiritivorum, including its MALDI-TOF MS score, is given (e). A combination of mass spectrometry and 16S rRNA gene sequencing may be necessary, in order to achieve a more reliable and accurate species identification. Please click here to view a larger version of this figure.
Figure 2: Analysis procedure using MALDI-TOF MS for identification of clinically relevant bacteria. An operator takes whole cell bacterial material (a) with a toothpick from a pure culture (b) and places bacterial smears on a steel target (c). The steel target is inserted into the loading port of the mass spectrometer. When the proper vacuum of 5 x 10-6 mbar is reached, the target will be moved to the ionization chamber. Using an N2-laser, ions are created by soft desorption which are then accelerated in an electrostatic field (d) and separated in the flight tube (e). The time of flight (TOF) needed for the ions to reach the detector of the flight tube (f) is directly related to their mass and forms the basis of subsequent calculations of mass peaks. Specific identification software (Materials Table) then assigns score values to the mass spectrum fingerprint of ionized bacterial proteins. Please click here to view a larger version of this figure.
Figure 3: MALDI-TOF spectra and assigned scores of rare pathogens. In this figure, representative MALDI-TOF spectra of the first six rare pathogens listed in Table 1 are shown. Species name and corresponding MALDI-TOF MS score are noted in each spectrum. Please click here to view a larger version of this figure.
Strain | MALDI identification | MALDI score | 16s rDNA result | BLAST homology |
#1 | Chryseobacterium gleum | 2.211 | Chryseobacterium gleum | 99% identity |
#2 | Myroides odoratimimus | 2.397 | Myroides odoratimimus | 99% identity |
#3 | Myroides odoratus | 2.237 | Myroides odoratus | 99% identity |
#4 | Sphingobacterium multivorum | 2.093 | Sphingobacterium multivorum | 99% identity |
#5 | Wohlfahrtiimonas chitiniclastica | 2.282 | Wohlfahrtiimonas chitiniclastica | 100% identity |
#6 | Wohlfahrtiimonas chitiniclastica | 2.289 | Wohlfahrtiimonas chitiniclastica | 100% identity |
#7 | Sphingobacterium spiritivorum | 2.269 | Sphingobacterium spiritivorum | 100% identity |
Table 1: Representative MALDI-TOFs results of rare occurring bacteria in comparison to 16s rRNA gene sequencing.
Both MALDI-TOF MS and 16S rRNA gene sequencing offer the possibility to identify large numbers of different bacteria. MALDI-TOF MS is a fast and inexpensive method, which is easy to handle and large databases of bacterial mass spectra are available. For this reason, MALDI-TOF MS is a rapid, cost effective and reliable method to conduct screening studies focused on rare bacterial pathogens 17,20,39,51. In a prospective study comparing MALDI-TOF MS with other phenotypic identification methods, Seng et al. demonstrated cost effectiveness and speed of MALDI-TOF MS 59 and Tan et al. reported a reduction of reagent and labor costs for bacterial identification in their laboratory setting by 56.9% annually 62. Such cost reductions are supported by a note of Gaillot et al. 28
On the other hand, 16S rDNA sequencing is more time consuming, laborious and warrants specialized personnel to perform the analysis 34. Nevertheless, both approaches are suitable for routine diagnostics and it could be demonstrated that the combination of these two methods leads to a higher reliability and more secure identification accuracy, which is especially beneficial in case of doubtful results 56. Therefore, we propose the combination of both methods as the best approach to verify the identification of rare occurring bacteria in routine diagnostics. For reliable species identification, it is absolutely imperative to use pure bacterial cultures because mixed cultures prevent species determination. As a plausibility check, species identification obtained with either MALDI TOF MS or 16S rDNA sequencing has to be in accordance with the results from Gram-staining, biochemical characterization and the clinical presentation 49.
Mass spectrometry as an analytical tool for bacterial identification was first proposed in 1975 51. However, it took until the late 1980s to establish a practicable approach for protein analysis. The "soft desorption ionization" technology was introduced by Koichi Tanaka in 1985 63, who received the Nobel Prize for chemistry in 2002, allowing analysis of intact proteins. At the same time the matrix-assisted ionization time of flight mass spectrometry was introduced by Hillenkamp and Karas as they were the first to use an organic acid to analyze biomolecules 37 and this is the method which is still used today. Although the MALDI technology has been used sporadically 23,30,41, it took until 2004 when Bruker Daltonics introduced its microflex MALDI Biotyper system (Pittcon Conference & Expo, 2004, press release), that MALDI-TOF MS based microbial identification became a routine diagnostic technique 46. Recent approaches in MALDI-TOF MS techniques gave the opportunity to identify yeasts, detect multiresistant bacteria and perform antimicrobial susceptibility testing 17,51. Moreover, certain procedures offer the possibility to directly analyze bacteria from primary samples, such as urine or blood cultures 17,51.
Although the use of this system is principally easy, there are some pitfalls which may influence the results. The age of the microorganisms for instance affects the bacterial protein expression and therefore the reproducibility of the results. First, growth conditions may be a problem. As an example, enterobacteria such as Escherichia coli grow faster than non-fermenting bacteria (e.g., Pseudomonas aeruginosa) and consequently have to be analyzed earlier 17. Second, the matrix used for MALDI-TOF MS consists of small organic acid molecules that have a strong laser optical absorption for the wavelength of the laser used. Prior to analysis the matrix is added to the sample and both components undergo a crystallization process forming a solid solution.
This explains why changes in the matrix may affect the accuracy of the bacterial identification 17. Therefore, freshly prepared matrix solutions, not older than 7 days, should be used. Third, the medium from which bacteria are picked may influence the identification results 17,68. For instance, crystal violet, which is a component of MacConkey agar, interferes with mass spectra 17. Additionally, too many bacteria applied on the steel target will lead to lower scores and therefore affect the identification accuracy. Therefore, it is advisable to define clear criteria as to how an analysis is carried out. However, misleading results performed by MALDI-TOF MS are mostly caused by insufficient reference spectra contained in the database. (Currently the data base contains >6,000 entries.) This is especially still the case for the identification of anaerobic bacteria 17. However, additional spectra can be added by the user. MALDI-TOF MS library for instance contains 98 additional reference spectra of isolates which are not adequately addressed by the original database, such as Mycoplasma sp., Myroides sp., Legionella sp., Roseomonas sp., Comamonas sp. and Chryseobacterium sp. Consequently, additional reference spectra will lead to a higher identification accuracy 59,69. Finally, in some cases the spectra of different species are very similar. This can lead to misidentification in cases such as the discrimination of Escherichia coli and Shigella spp. or Streptococcus pneumoniae and Streptococcus mitis 17,51.
Sequencing of 16S rDNA and additional genes such as rpoB have simplified the molecular identification of rare or unknown bacteria. These genes are common in most bacteria and their individual function is identical. However, since they possess enough genetic variability to produce results which permit a differentiation on genus and species level, they can be used for bacterial identification 34. According to Stackebrandt and Göbel, homologies <97% represent different species 61. However, homologies >97% do not necessarily lead to a secure species identification 34,47. These uncertainties have several reasons.
The quality of the databases which are used to calculate the homologies is sometimes questionable 34. In cases where high homologies at the 16S rDNA level exist, uncertainties may result in species identification. A combination of different genetic regions may therefore lead to more secure results. Moreover, there are no general guidelines for the interpretation of 16S rDNA sequencing data 22,34,61. However, increasing the reliability of the existing databases will lead to a higher accuracy for bacterial identification using this molecular approach 34. Regarding the sequencing reaction we need to mention that the accuracy of the sequencing results mostly relies on the quality of the underlying PCR. Furthermore, since the results are being gained by comparing them to the entries listed in public databases, quality of sequence entries and maintenance of the data is of crucial importance.
In general, although both approaches, MALDI-TOF MS and 16S rRNA gene sequencing, have advantages they have also weaknesses. The combination of both methods, however, leads to a high accuracy for bacterial identification. For instance, in a previous study we could demonstrate that MALDI-TOF MS is able to distinguish between Myroides odoratimimus and Myroides odoratus 56. In a few cases the MALDI score suggested unreliable species identification, while the results obtained by 16S rDNA sequencing could confirm the species identity. Mass spectral fingerprints of these isolates were then created and introduced into our MALDI reference spectra database. Future research focusing on other rare bacteria may demonstrate the applicability of the proposed strategy.
The authors have nothing to disclose.
The authors would like to thank Prof. Enno Jacobs for his continuing support.
CHROMASOLV, HPLC grade water, 1 L | Sigma-Aldrich Chemie, München, Germany | 270733 | |
Tissue Lyser LT | Qiagen, Hilden, Germany | 85600 | Oscillating Homogenizer |
Glass-beads 1,0mm | VWR International, Darmstadt, Germany | 412-2917 | |
Thermomixer 5436 | Eppendorf, Hamburg, Germany | 2050-100-05 | |
QIAamp DNA Mini Kit (250) | Qiagen, Hilden, Germany | 51306 | |
Taq PCR Core Kit (1000 U) | Qiagen, Hilden, Germany | 201225 | |
Forward Primer TPU1 (5´-AGA GTT TGA TCM TGG CTC AG-3’) | biomers.net GmbH, Ulm, Germany | – | |
Reverse Primer RTU4 (5´-TAC CAG GGT ATC TAA TCC TGT T-3´) | biomers.net GmbH, Ulm, Germany | – | |
Mastercycler | Eppendorf, Hamburg, Germany | - | Thermocylcer |
Reaction tube 1.5 mL | SARSTEDT, Nümbrecht, Germany | 72,692 | |
Reaction tube 2 mL | SARSTEDT, Nümbrecht, Germany | 72,693,005 | |
PCR 8er-CapStrips | Biozym Scientific, Hessisch Oldendorf, Germany | 711040X | |
PCR 8er-SoftStrips | Biozym Scientific, Hessisch Oldendorf, Germany | 711030X | |
Sharp R-ZV11 | Sharp Electronics, Hamburg, Germany | – | Microwave |
Titriplex III (EDTA Na2-salt dehydrate; 1 kg) | Merck, Darmstadt, Germany | 1084211000 | |
SeaKem LE Agarose | Biozym Scientific, Hessisch Oldendorf, Germany | 849006 | |
(2 x 500 g) | |||
SmartLadder SF – 100 to 1000 bp | Eurogentec, Lüttich, Belgium | MW-1800-04 | |
Bromphenol blue (25 g) | Sigma-Aldrich Chemie, München, Germany | B0126 | |
Xylene cyanol FF (10 g) | Sigma-Aldrich Chemie, München, Germany | X4126 | |
ComPhor L Maxi | Biozym, Hessisch Oldendorf, Germany | – | |
Ethidium bromide solution 1 %(10 mL) | Carl Roth, Karlsruhe, Germany | 2218.1 | |
Gel Doc 2000 | Bio-Rad Laboratories, München, Germany | – | Gel-documentation system |
ExoSAP-IT (500 reactions) | Affymetrix UK, Wooburn Green, High Wycombe, United Kingdom | 78201 | |
Buffer (10 x) with EDTA | Life Technologies, Darmstadt, Germany | 402824 | |
BigDye Terminator Kit v1.1 | Life Technologies, Darmstadt, Germany | 4337450 | |
Hi-Di formamide (25 mL) | Life Technologies, Darmstadt, Germany | 4311320 | |
DyeEx 2.0 Spin Kit (250) | Qiagen, Hilden, Germany | 63206 | |
3130 Genetic Analyzer | Life Technologies, Darmstadt, Germany | – | Sequenzer |
MicroAmp optical 96-well reaction plate with barcode | Life Technologies, Darmstadt, Germany | 4306737 | |
3130 Genetic Analyzer, plate base 96-well | Life Technologies, Darmstadt, Germany | 4317237 | |
3130 Genetic Analyzer, plate retainer 96-well | Life Technologies, Darmstadt, Germany | 4317241 | |
3130 Genetic Analyzer, well plate septa | Life Technologies, Darmstadt, Germany | 4315933 | |
3130 Genetic Analyzer, POP-7 Polymer, 7 mL | Life Technologies, Darmstadt, Germany | 4352759 | |
3130 Genetic Analyzer, 4-Capillary Array, 50 cm | Life Technologies, Darmstadt, Germany | 4333466 | |
Sequencing Analysis Software 5.4 | Life Technologies, Darmstadt, Germany | – | |
microflex (the MALDI TOF MS maschine) | Bruker Daltonik, Bremen, Germany | – | |
MALDI Biotyper (the MALDI TOF MS system) | Bruker Daltonik, Bremen, Germany | – | our mMIS |
VITEK MS | bioMérieux, Nürtingen, Germany | 2nd mMis | |
flexControl 3.4 (control software) | Bruker Daltonik, Bremen, Germany | – | |
Biotyper Realtime Classification 3.1 (RTC), (analysis software) | Bruker Daltonik, Bremen, Germany | – | |
α-cyano-4-hydroxycinnamic acid, HCCA, 1 g | Bruker Daltonik, Bremen, Germany | 201344 | |
Peptide Calibration Standard II | Bruker Daltonik, Bremen, Germany | 222570 | |
MSP 96 target polished steel | Bruker Daltonik, Bremen, Germany | 8224989 | |
peqgreen | peqlab | 37-5010 | |
MALDI Biotyper Galaxy | Bruker Daltonik, Bremen, Germany | Part No. 1836007 | |
Vitek 2 | bioMérieux, Nürtingen, Germany | our aMis | |
MicroScan | Beckman Coulter | 2nd aMis | |
BD Phoenix™ Automated Microbiology System | BD | 3rd aMis | |
Staphylococcus aureus subsp. aureus Rosenbach (ATCC® 25923™) | ATCC | postive control for PCR |