• We are sorry, but NCBI web applications do not support your browser and may not function properly. More information
Logo of aemPermissionsJournals.ASM.orgJournalAEM ArticleJournal InfoAuthorsReviewers
Appl Environ Microbiol. Dec 2008; 74(24): 7767–7778.
Published online Oct 24, 2008. doi:  10.1128/AEM.01402-08
PMCID: PMC2607147

Rapid Classification and Identification of Salmonellae at the Species and Subspecies Levels by Whole-Cell Matrix-Assisted Laser Desorption Ionization-Time of Flight Mass Spectrometry[down-pointing small open triangle]

Abstract

Variations in the mass spectral profiles of multiple housekeeping proteins of 126 strains representing Salmonella enterica subsp. enterica (subspecies I), S. enterica subsp. salamae (subspecies II), S. enterica subsp. arizonae (subspecies IIIa), S. enterica subsp. diarizonae (subspecies IIIb), S. enterica subsp. houtenae (subspecies IV), and S. enterica subsp. indica (subspecies VI), and Salmonella bongori were analyzed to obtain a phylogenetic classification of salmonellae based on whole-cell matrix-assisted laser desorption ionization-time of flight mass spectrometric bacterial typing. Sinapinic acid produced highly informative spectra containing a large number of biomarkers and covering a wide molecular mass range (2,000 to 40,000 Da). Genus-, species-, and subspecies-identifying biomarker ions were assigned on the basis of available genome sequence data for Salmonella, and more than 200 biomarker peaks, which corresponded mainly to abundant and highly basic ribosomal or nucleic acid binding proteins, were selected. A detailed comparative analysis of the biomarker profiles of Salmonella strains revealed sequence variations corresponding to single or multiple amino acid changes in multiple housekeeping proteins. The resulting mass spectrometry-based bacterial classification was very comparable to the results of DNA sequence-based methods. A rapid protocol that allowed identification of Salmonella subspecies in minutes was established.

The genus Salmonella comprises two species, Salmonella enterica and Salmonella bongori (29). S. enterica is further divided into six subspecies, S. enterica subsp. enterica (subspecies I), S. enterica subsp. salamae (subspecies II), S. enterica subsp. arizonae (subspecies IIIa), S. enterica subsp. diarizonae (subspecies IIIb), S. enterica subsp. houtenae (subspecies IV), and S. enterica subsp. indica (subspecies VI), which have been recognized on the basis of variation in biochemical characteristics and DNA-DNA hybridization results (9, 14, 19, 20, 21, 38). The division of Salmonella into these seven groups has been confirmed by sequence analysis of housekeeping genes and invasion-associated protein genes (4). Based on these results, an evolutionary model has been proposed, which separates predominantly diphasic flagellar expression serotypes from monophasic serotypes (3, 35). Other investigations based on 23S rRNA sequence analyses (7) and amplified fragment length polymorphism (34) supported this model.

Salmonellae are important food-borne pathogens that are responsible for serious cases of food-borne illness. In subspecies I, which is responsible for 99% of the cases of human salmonellosis (41), there are over 2,500 known serovars that differ in prevalence and the diseases that they cause in different hosts. S. bongori and S. enterica subsp. arizonae are typically associated with cold-blooded hosts, whereas the other subspecies are associated with warm-blooded hosts or both types of hosts.

Rapid and reliable identification of pathogenic microorganisms, including Salmonella, is important for surveillance, prevention, and control of food-borne diseases. The established methods for bacterial identification in clinical microbiology are often time-consuming and laborious. Identification of Salmonella subspecies by biochemical typing procedures requires long incubation times, and therefore there is a delay in final identification. These procedures require experience in interpretation and can be limited by subjectivity and low specificity. There is an increasing need for alternative procedures that allow rapid and reliable identification of microorganisms. In recent years, several reports have shown the feasibility of using matrix-assisted laser desorption ionization (MALDI)—time of flight (TOF) mass spectrometry (MS) to identify microorganisms (2, 8, 11, 12, 15, 16, 17, 18). In whole-cell MALDI-TOF MS, characteristic “fingerprint” spectra are obtained from whole (“intact”) cells without biomarker prefractionation, digestion, separation, or cleanup. The procedure is very fast, requires minimal amounts of biological material (subcolony amounts), is suitable for high-throughput routine analysis, and therefore has great potential for applications in clinical microbiology or environmental monitoring. The observed protein biomarkers are typically highly expressed proteins with housekeeping functions, such as ribosomal or nucleic acid-binding proteins (28, 30), which are highly conserved in bacteria; therefore, the method can be universally applied. New applications have included detection of plasmid insertion in Escherichia coli (33), differentiation between isogenic teicoplanin-susceptible and teicoplanin-resistant strains of methicillin-resistant Staphylococcus aureus (25), and discrimination between wild-type and ampicillin-resistant E. coli strains (5). While several studies demonstrated the applicability of this technique for bacterial species identification, few studies have examined its potential for discrimination at levels below the species level (1, 15, 22, 31, 32, 42). The aim of the present investigation was to study the suitability of using whole-cell MALDI-TOF MS for differentiation of salmonellae at the species and subspecies levels. As a prerequisite, reproducible and highly informative MALDI-TOF-MS spectra were acquired under defined conditions for a collection of well-characterized strains belonging to different subspecies and serotypes isolated from diverse animals or food samples over several years. A MALDI-TOF MS protocol was developed which allows subspecies identification of Salmonella isolates in a few minutes using subcolony amounts of bacterial biomass grown on agar plates. Using sinapinic acid, high-quality mass spectra for a molecular mass range from 2,000 to 40,000 Da were obtained. Phylogenetic classification was based on MS bacterial typing using variations in mass data for multiple housekeeping proteins.

MATERIALS AND METHODS

Bacterial strains.

A total of 126 strains representing all known S. enterica subspecies and S. bongori were selected (Table (Table1).1). A detailed strain list is shown in Table S1 in the supplemental material. Epidemiologically unlinked strains belonging to various serotypes originating from different regions and herds in Germany were carefully selected from the strain collection of the National Salmonella Reference Laboratory. These strains were isolated between 2000 and 2008 from diverse animal and food samples. All of the strains were identified by biochemical assays (19-21) and serotyping using the Kauffmann-White scheme (14).

TABLE 1.
Salmonella strains used in this studya

Bacterial isolation and culture conditions.

Standardized conditions were used for bacterial growth. Liquid cultures containing Luria-Bertani broth (Merck, Darmstadt, Germany) were inoculated and incubated overnight at 37°C. For routine measurements, Salmonella strains were streaked onto Mueller-Hinton agar plates (Oxoid, Greve, Denmark) and incubated at 37°C for 24 ± 1 h, and single colonies were selected. In order to compare the effects of different growth conditions on the expression of protein biomarker ions, selected strains were also grown on Mueller-Hinton blood agar (Oxoid), Columbia agar (Merck), Columbia blood agar (Merck), Gassner agar (Merck), plate count agar (Merck), and sheep blood agar (Merck).

Preparation of samples for MALDI-TOF MS of whole bacterial cells.

Individual colonies were removed from plates using a sterile pipette tip and applied directly as a thin film onto a 384-position MALDI sample target (Bruker Daltonics, Bremen, Germany). The samples were immediately mixed with 1 μl of sinapinic acid obtained from Bruker Daltonics (Bremen, Germany) (25 mg/ml in 50% acetonitrile [Sigma-Aldrich]) supplemented with 0.6% trifluoroacetic acid (Roth, Germany). The matrix sample spots were crystallized by air drying. The other matrices tested included 2,5-dihydroxybenzoate and α-cyano-4-hydroxycinnamic acid (Bruker Daltonics, Bremen, Germany).

MALDI-TOF MS parameters.

All mass spectra were acquired with an Ultraflex II MALDI-TOF/TOF mass spectrometer (Bruker Daltonics, Bremen, Germany) equipped with an all-solid-state smartbeam Nd:YAG laser and operated at 100 Hz in the positive linear mode (delay, 100 ns; voltage, 25 kV; molecular mass range, 2.2 to 40 kDa) under control of Flexcontrol software (version 3.0; Bruker Daltonics). Each spectrum was obtained by averaging up to 10,000 laser shots acquired at the minimum laser power necessary for ionization of the samples. The spectra were externally calibrated by using a standard calibration mixture (protein calibration standard I supplied by Bruker Daltonics).

Data evaluation.

Reference spectra for a set of Salmonella strains belonging to the different S. enterica subspecies and S. bongori were analyzed at least in duplicate. Mass data files were transferred to the Flexanalysis software (version 3.0; Bruker Daltonics) and processed with baseline correction, Gaussian smoothing, and peak finding. Average mass values were determined. Spectra were internally calibrated using a set of ribosomal biomarker proteins common to all Salmonella spp. For the initial data evaluation, spectra were imported into the BioNumerics 5.1 software (Applied Maths, Belgium) for visualization in gel view representation and calculation of dendrograms. For phylogenetic classification the profiles of biomarkers (with a binary table extracted from Table S2 in the supplemental material) were processed with BioNumerics 5.1 using the simple matching similarity coefficient, and dendrograms were constructed using complete linkage. Peak lists were imported into the SARAMIS software (Spectral Archiving and Microbial Identification System, release 3.36; Anagnostec GmbH, Germany). In the first step, consensus spectra, called superspectra, were calculated for each of the six S. enterica subspecies and S. bongori using multiple measurements for strains belonging to the different taxa. These spectra consisted of sets of biomarker ions that were specific at the different taxonomic levels and were present in >95% of the strains. In the second step, the superspectra were compared to identify peaks that were present in all superspectra (category I), were present in only S. bongori superspectra or in all S. enterica subspecies superspectra (category II), or were present in the superspectra for only one of the subspecies (category III). The remaining peaks were designated category IV mass peaks. Using the SARAMIS software, a point system for each superspectrum was created based on peak lists with mass signals weighted according to their specificity. Using this approach, highly specific biomarkers were upweighted in the identification routine, while nonspecific, variable, and low-intensity peaks were downweighted or ignored in the identification process. Automated computer-aided identification was performed by comparing peak lists for individual samples, including samples of Salmonella subspecies whose genomes have not been sequenced, with the established reference database of superspectra, generating a ranked list of matching spectra.

Identification of biomarker proteins based on database searches.

The m/z peaks obtained were subjected to an online TagIdent protein database search (13). An unrestricted search with pI values was performed. Theoretical masses of protein sequences were calculated using the PeptideMass tool (13). The BLAST servers at www.sanger.ac.uk and www.expasy.ch were used for protein-versus-translated DNA BLAST searches for strains of S. bongori, S. enterica subsp. enterica, and S. enterica subsp. arizonae whose genomes have been sequenced.

RESULTS

Optimization of experimental parameters.

Species can be readily identified using whole-cell MALDI-TOF MS by detecting a limited number of specific biomarker peaks. Typically, 5 to 10 peaks in the molecular mass range from 2,000 to 11,000 Da are sufficient to discriminate bacteria at the species level. Different protocols are available for sample preparation, the matrix used, and measurement parameters. Our initial experiments showed that for identification at levels below the species level the requirements related to information content, reproducibility, mass accuracy, and quality of spectra are significantly greater than the requirements for routine species identification by whole-cell MALDI-TOF MS. In order to establish a standardized analytical protocol, several experimental, sample preparation, and MS parameters that can affect the reproducibility and accuracy of data were evaluated. These parameters included the type and concentration of matrix, the sample preparation procedure, the matrix solvent mixture, the concentration of acid added to the matrix, and the growth medium, as well as measurement parameters such as the laser energy and the number of shots summarized. Our aim was to determine the simplest procedure that has the potential for automation and that results in MS data with high information content (large number of peaks) and a balanced number of peaks in the low-molecular-mass and especially high-molecular-mass (>13,000 Da) regions. Previous suspension of cells in water or solvent mixtures did not result in an improvement compared to the direct application of cells to the MALDI target. Different matrices were tested, including 2,5-dihydroxybenzoic acid, α-cyano-4-hydroxycinnamic acid, and sinapinic acid. The use of 2,5-dihydroxybenzoic acid resulted in many fewer peaks, especially in the higher-molecular-mass range, and was less suitable for automated measurements due to the heterogeneous crystallization of this compound. α-Cyano-4-hydroxycinnamic acid produced very homogeneous crystal layers, but the spectra were much less informative, especially in the higher-molecular-mass range, than the spectra obtained with sinapinic acid, which was chosen for further experiments. Various concentrations of sinapinic acid were tested. As a general rule, higher concentrations favored the detection of higher-molecular-mass protein peaks but resulted in decreased peak intensities for low-molecular-mass peaks (data not shown). As a result of the optimization process, the overall number of biomarker peaks (>300) and especially the information for the higher-molecular-mass range (>10,000 to 40,000 Da), which includes important discriminative information due to the higher probability of mutations occurring in larger proteins, were significantly increased, thereby increasing the probability of detecting discriminative biomarkers for differentiation of closely related bacteria, such as Salmonella subspecies (see the sample spectrum in Fig. S3 in the supplemental material).

The influence of growth conditions on the MALDI-TOF MS patterns of different Salmonella subspecies was analyzed by subculturing strains on several different media. Very similar patterns were generated, but slight variations in the expression of proteins were observed, which resulted in medium-dependent clustering of MALDI-TOF data from multiple analyses of strains (Fig. (Fig.1).1). However, the expression of genus-, species-, and subspecies-identifying biomarker ions was found to be largely stable under different conditions. The overall quality of spectra produced after cultivation on Gassner agar was reduced, and the peaks were less intense. Blood-containing media have the disadvantage that the profiles can be contaminated by blood-related proteins (e.g., m/z 15,048 and m/z 16,075 together with their doubly charged variants). Therefore, Mueller-Hinton agar was chosen as the standard medium for routine analysis.

FIG. 1.
MALDI-TOF MS profiles and dendrogram of S. enterica subsp. arizonae (Arizonae) and S. enterica subsp. houtenae (Houtenae) grown on different agar media. Duplicate spectra for representative samples are shown, as visualized using a gel view representation. ...

Data evaluation and assignment of biomarker peaks to known proteins.

A total of 254 spectra for 126 strains of S. enterica and S. bongori and one E. coli strain were analyzed. Figure Figure22 shows an overlay of mass spectra for S. bongori and S. enterica subspecies in the molecular mass range from 13,230 to 15,270 Da. The spectra display high overall levels of similarity, but slight mass shifts of peaks in the protein profiles of different Salmonella species and subspecies were detected. The SARAMIS software was used to identify biomarker peaks and to assign them to one of the following categories: biomarkers that were specific for the genus (or a higher-level taxon) (category I), biomarkers that were species specific for either S. enterica or S. bongori (category II), biomarkers that were subspecies specific (category III), or biomarkers that were not genus specific or were specific for more than one species or subspecies but were reproducibly present in all strains of the species or subspecies analyzed (category IV). All other peaks that were variable within a taxonomic group (e.g., peaks that were potentially strain specific or serovar specific or displayed variable expression) were eliminated from the analysis at this point. A discussion of the discriminative potential of MALDI-TOF MS at levels below the subspecies level will be presented elsewhere. Using the Tagident tool (www.expasy.ch) and translated DNA BLAST searches of genome sequence information available for S. bongori, S. enterica subsp. arizonae, and several S. enterica subsp. enterica serovars, many of the observed biomarker ions could be tentatively assigned. Mass values for well-resolved peaks observed in spectra of different strains were matched to the same protein species within a mass tolerance window of ±1 Da in the molecular mass range from 2,000 to 20,000 Da. Larger and low-intensity protein peaks were tentatively assigned by using a larger mass tolerance window (±5 Da). Table Table22 summarizes the assignments for selected biomarker peaks, together with calculated masses and posttranslational modifications indicating peak presence or absence. For E. coli only the peaks that were also found in Salmonella are indicated. Many proteins produced, in addition to a singly protonated protein signal [(M+H)+], the corresponding doubly protonated protein signal [(M+2H)2+]. A complete list of the peaks, including doubly charged and unidentified biomarker ions, is shown in Table S2 in the supplemental material. Most of the peaks detected corresponded to ribosomal proteins of the large (505) and small (305) subunits. Information for methionine loss or posttranslational modifications like methylation, acetylation, and β-methylthiolation that are known to occur in E. coli was accounted for in the peak assignments (2, 43). Most known ribosomal proteins of the large and small subunits in the molecular mass range from 2,000 to 20,000 could be tentatively assigned; the exceptions were ribosomal proteins L6 and L9. Protein peaks of ribosomal proteins L1 to L5 and S2 to S4 with molecular masses greater than 20,000 Da were present at lower intensities and were not used during the routine identification process. In addition, the identities of several proteins other than ribosomal proteins were proposed on the basis of the observed MALDI-TOF masses; these proteins included ribosome-associated inhibitor A, an RNA chaperone, a ribosome modulation factor, a carbon storage regulator, the Gns protein, several cold shock proteins, translation initiation factor IF-1, DNA-binding proteins HU-alpha and -beta and H-NS, a probable σ54 modulation protein, the 10-kDa chaperonin CH10, integration host factor subunit beta, glutaredoxin-1, inorganic pyrophosphatase, and several uncharacterized proteins. Several low-intensity peaks and some peaks with high intensities could not be assigned. This could have been due to unknown posttranslational modifications or proteolytic cleavage of signal peptides.

FIG. 2.
Overlay of MALDI mass spectra for strains of S. bongori (group V) (red), S. enterica subsp. enterica (subspecies I) (gray), S. enterica subsp. salamae (subspecies II) (green), S. enterica subsp. arizonae (subspecies IIIa) (dark blue), S. enterica subsp. ...
TABLE 2.
Selected genus-, species-, and subspecies-specific biomarker peaks obtained by whole-cell MALDI-TOF MS of salmonellae and tentative assignment of proteinsa

Designation of GIBIs based on MALDI-TOF MS analysis.

Ions that were reproducibly detected in both Salmonella spp. (S. enterica and S. bongori) were designated genus-identifying biomarker ions (GIBIs). Table S2 lists 57 peaks (including doubly charged proteins) that were present in all Salmonella spectra (category I) and were selected as biomarkers to unequivocally discriminate Salmonella spp. from other bacterial genera (see the supplemental material). Forty-five of these peaks were related to ribosomal proteins from the large and small subunits. Other GIBIs corresponded to a carbon storage regulator (CsrA; 6,857 Da), a cold shock protein (CspC; 7,272 Da), translation initiation factor IF-1 (8,119 Da), a 10-kDa chaperonin (CH10; 10,188 Da), DNA-binding protein H-NS (15,412 Da), and other unidentified proteins. A subset of genus-specific ribosomal proteins was used for internal calibration. All Salmonella strains had 29 peaks in common with E. coli, which originated from 12 identical ribosomal proteins, a carbon storage regulator, a cold shock protein, DNA-binding protein H-NS, RNA-binding translation initiation factor IF1, and two unidentified proteins.

Discrimination of S. bongori and S. enterica and designation of SIBIs based on MALDI-TOF MS analysis.

MS analyses of multiple S. enterica and S. bongori strains revealed a high overall level of similarity of their whole-cell protein profiles, but a limited number of species-identifying biomarker ions (SIBIs) were reproducibly detected, resulting in clear discrimination of the two species based on comparison of their fingerprint patterns (Fig. (Fig.3).3). The designations of SIBIs are shown in Table S2 in the supplemental material (category II). Twenty-seven S. bongori-specific peaks that were present in the spectra of all S. bongori strains but were not present in the spectrum of any of the S. enterica strains analyzed were detected, while 14 peaks were present in the spectra of all S. enterica strains but not in S. bongori spectra. The Salmonella species comparative sequencing BLAST server (Sanger Institute, United Kingdom) was used to compare protein sequences of the S. bongori and S. enterica subsp. enterica strains whose genomes have been sequenced and to identify point mutations. Observed mass differences corresponded to slight sequence variations in ribosomal proteins, other proteins that were tentatively assigned to protein Gns, to cold shock proteins, or to ribosome-associated inhibitor A, or protein peaks that could not be assigned. Amino acid differences between S. bongori and S. enterica subsp. enterica are indicated in Table S2 in the supplemental material. SIBIs for S. bongori, for example, were identified at m/z 9,207, corresponding to a V→A change in the ribosomal protein S16, at m/z 12,549, corresponding to an S→N change in ribosome-associated inhibitor A, at m/z 14,726, corresponding to a G→S change in ribosomal protein S9, and at m/z 15,976, corresponding to an R→L change in ribosomal protein L13. The SIBIs highly characteristic of S. enterica included peaks at m/z 8,920 (L28), m/z 12,522 (ribosome-associated inhibitor A), m/z 14,696 (S9), and m/z 16,019 (L13).

FIG. 3.
Salmonella species discrimination: WARD-generated dendrogram obtained using Pearson correlation (BioNumerics 5.1) and gel view of mass data obtained from whole-cell MALDI-TOF MS analyses of S. bongori and S. enterica strains.

Discrimination of S. enterica subspecies and designation of SSIBIs based on MALDI-TOF MS analysis.

Sequence variations corresponding to single or multiple amino acid changes in proteins detected in MALDI-TOF analyses of different S. enterica subspecies also allowed MS-based subtyping of S. enterica strains into subgroups corresponding to the different S. enterica subspecies. Subspecies-identifying biomarker ions (SSIBIs) are shown in Table Table22 (category III). While most of the GIBIs and SIBIs were related to ribosomal proteins, most of the SSIBIs were tentatively assigned to other proteins or were not identified, indicating that a classification based solely on the ribosomal protein subset is not sufficient for differentiation of very closely related microorganisms. Protein assignment was assisted by cross-comparison of protein sequences identified by TagIdent mass searches using protein-versus-translated DNA BLAST searches in genome databases for S. enterica subsp. arizonae and S. enterica subsp. enterica strains. Comprehensive annotation of the identified biomarkers should be possible after completion of genome projects for the other S. enterica subspecies and/or structural analysis of posttranslational modifications possibly occurring in the unidentified proteins. The peaks specific for S. enterica subsp. enterica included, in addition to several unidentified protein peaks, peaks for ribosomal proteins L25 (m/z 10,542), L7/L12 (m/z 12,183), and L17 (m/z 14,395), a ribosome modulation factor (m/z 6,572), cold shock-like protein CspH (m/z 7,662), glutaredoxin-1 (m/z 9,924), and the putative uncharacterized protein YigF (m/z 13,445). S. enterica subsp. arizonae could be unequivocally identified by the presence of a peak at m/z 13,352 corresponding to ribosomal protein RL20, which has a T→S amino acid change compared to all other S. enterica subspecies (m/z 13,366). Other tentatively assigned S. enterica subsp. arizonae-specific biomarkers were assigned to protein Gns (m/z 6,479), the RNA chaperone CspE (m/z 7,333), integration host factor subunit beta (m/z 10,608), a probable σ54 modulation protein (m/z 10,884), and ribosomal proteins L25 (m/z 5,270), L7/L12 (m/z 12,283), S13 (m/z 13,045), L20 (m/z 13,352), putative modified S11 (m/z 13,729), and S8 (m/z 13,897).

Phylogenetic classification of Salmonella spp.

Dendrograms were constructed based on MALDI-TOF MS biomarker profiles (with a binary table extracted from Table S2 in the supplemental material) using the simple matching similarity coefficient and complete linkage (Fig. (Fig.4).4). Category IV biomarker ions, which were present in more than one subspecies, were also included in this analysis. The topology obtained using MALDI-TOF MS profiling strongly resembled the topologies of dendrograms constructed using combined coding sequences of several housekeeping genes and invasion genes (3, 4, 23) and therefore may indicate the evolutionary relationships of the subspecies of S. enterica. In agreement with evidence obtained from genomic DNA hybridization (19, 21) and multiple gene sequence analyses, MALDI-TOF MS protein profiling indicated that group V (S. bongori) is the most divergent taxon in the salmonellae. S. enterica is subdivided into six subspecies. Subspecies I (S. enterica subsp. enterica), II (S. enterica subsp. salamae), IIIb (S. enterica subsp. diarizonae), and VI (S. enterica subsp. indica), which are predominantly diphasic in terms of flagellar expression, cluster apart from the monophasic salmonellae (S. enterica subsp. arizonae [subspecies IIIa], S. enterica subsp. houtenae [subspecies IV], and S. bongori [V]), supporting the evolutionary model of Selander et al. (35, 40).

FIG. 4.
Phylogenetic classification of Salmonella species and subspecies based on variations in whole-cell MALDI-TOF mass data for multiple protein biomarkers. The dendrogram was calculated based on the binary table extracted from Table S2 in the supplemental ...

DISCUSSION

Microbial species identification by whole-cell MALDI-TOF MS is generally achieved by comparison of experimental mass data for a set of intact protein ions desorbed from whole bacterial cells with a database of reference spectra (fingerprint-based approach). In mass spectra obtained using whole cells, typically up to 30 constant peaks are detected, predominantly in the molecular mass range from 4,000 to 13,000 Da (11, 18, 42); this number of peaks has been shown to be sufficient for bacterial identification at the genus and species levels for many food-borne or clinically relevant bacterial pathogens, such as E. coli, Campylobacter, Salmonella, Pseudomonas, Yersinia, and Listeria (6, 23, 27, 28, 44), as well as environmental isolates (10, 31). Fingerprint-based approaches for subtyping bacteria at levels below the species level tend to be less useful than approaches used for species identification, primarily because of the high overall similarity of MS fingerprints within species and the difficulty of reproducibly detecting sufficient numbers of biomarkers with specificities below species-level specificity (22, 32, 42). In order to differentiate bacteria at levels below the species level, spectra with a high number of reproducible protein peaks were required. The information content of the mass spectra obtained in this study was substantially increased, especially for molecular masses greater than 13,000 Da, compared to the information obtained in previous MALDI-TOF MS studies performed for species identification (26, 27, 31, 42). Typically, more than 300 peaks, mainly between 2,000 and 25,000 Da, were detected, and even at molecular masses of greater than 35 kDa, subspecies-specific protein peaks were detectable (see Fig. S3 in the supplemental material). Due to the clonality of the genus Salmonella and the intrasubspecies variability of the protein profiles, simple clustering of mass data from bacterial fingerprints initially did not result in clear discrimination of the strains at the subspecies level. Therefore, a bioinformatics-based approach that was recently proposed by Teramoto et al. (39) was used. MALDI-TOF MS profiles of whole bacterial cells have been proposed to be generally dominated by abundant, cytosolic proteins that are highly basic, a feature that is known to result in efficient ionization in the MALDI process (28, 30). Proteins that fulfill these criteria in particular include ribosomal proteins, proteins involved in DNA or RNA binding in general, and other abundant proteins, most of which have high isoelectric points. Indeed, most of the pI values calculated for the tentatively assigned proteins detected in this study were greater than 9. This was true not only for the subset of ribosomal proteins but also for many other observed proteins, like cold shock-like protein CspH, translation initiation factor IF-1 (pI 9.23), DNA-binding proteins HU-alpha and -beta (pI 9.69 and 9.57), the ribosome modulation factor (pI 10.56), and integration host factors A and B (both pI 9.34). In many cases detected proteins with lower pI values were known to be very abundant proteins; these proteins included the nucleoid-associated protein H-NS (pI 5.32), the RNA chaperone CspE (pI 8.08), glutaredoxin-1 (pI 5.63), and the phosphocarrier protein HPr (pI 5.6). The bacterial ribosome consists of more than 50 ribosomal “housekeeping” proteins, and sequence variations that occur at the subspecies level should enable phylogenetic classification based on mass data for multiple protein biomarkers. The key to Salmonella subtyping was establishment of an optimized sample preparation and MALDI measurement procedure, with which almost all of the expected mass peaks for ribosomal subunit proteins could be detected together with a high number of other peaks that could be tentatively assigned to known proteins or left unidentified, probably due to unknown posttranslational modifications. The assignment procedure was complicated by various posttranslational modifications occurring in ribosomal proteins which have to be accounted for, including N-terminal methionine loss, methylation, β-methylthiolation, oxidation, or acetylation. However, since many of these modifications appeared to be conserved posttranslational modifications in Salmonella and E. coli, information on posttranslational modifications could be extracted from previous reports on the detection of ribosomal proteins in E. coli (2, 43). By identifying and selecting biomarker subsets specific at different taxonomic levels, a sufficient number of SSIBIs were identified so that salmonellae could be subtyped at a level below the species level. This bioinformatics-based approach for phylogenetic classification of bacterial strains uses selected biomarkers that can be identified by comparing experimental mass data to translated gene databases mainly for microorganisms whose genomes have been sequenced (28, 39). The principle is similar to the multilocus enzyme electrophoresis (4) principle and the multilocus sequence typing (24, 37) principle, which is based on a combination of several sequence types for multiple housekeeping genes. Whereas fingerprint-based approaches require collection of MS fingerprint data, bioinformatics-based approaches benefit from the exponentially increasing number of publicly available sequenced microbial genomes. Therefore, they are less dependent on experimental and biological factors that influence the spectra, and results obtained in different laboratories can be compared more easily. Moreover, compared to DNA sequence-based approaches that require gel electrophoresis or DNA sequencing, the sample preparation and measurement procedures are much simpler and faster. Automated measurement combined with computer-aided evaluation of the data allowed very rapid identification with high throughput capabilities. Unlike fingerprint-based approaches, whose results generally do not reflect phylogenetic relationships, MS typing of multiple housekeeping proteins can also be used to probe the evolutionary history of related bacteria, since the proteomics-based method is based on detection of genetic variation, which means accumulating nonsynonymous point mutations at multiple housekeeping loci. The genus Salmonella forms a single DNA homology group comprising seven subgroups and more than 2,500 serovars. The data clearly separated the S. bongori and S. enterica subgroups (Fig. (Fig.2)2) and supported evidence from other studies that strongly differentiated the group V strains from all other salmonellae (4, 29, 35, 36). The next two most divergent subgroups were S. enterica subsp. arizonae and S. enterica subsp. houtenae, which is in agreement with the comparative analysis of the combined coding sequences of five housekeeping genes and seven invasion genes (4), the sequence comparison of 23S rRNA (7), and amplified fragment length polymorphism analyses (34, 40). Diphasic subspecies I, II, IIIb, and VI were placed in a discrete branch of the dendrogram and were separated from the predominantly monophasic (sub)species.

The major advantages of MALDI-TOF MS-based bacterial typing compared to other typing methods are the ease and speed of the procedure and the possibility of automation and high-throughput analysis. This study used a true “whole-cell” procedure with minimal sample preparation that consisted of transferring subcolony amounts of bacterial biomass grown on agar plates directly to a MALDI sample plate, followed by on-target extraction of proteins. An automated measurement method was used, which allowed analysis of one sample in less than 2 min on a 384-well sample plate. Using the SARAMIS software, automated computer-aided identification of salmonellae was achieved by comparing mass spectra for individual samples, including samples of strains whose genomes have not been sequenced, with a reference database of superspectra containing peak lists weighted according to their specificities at the different taxonomic levels. Alternatively, artificial superspectra composed of mass lists calculated from sequences of protein biomarkers can be used as reference spectra for microorganisms whose genomes have been sequenced. It is expected that the bioinformatics-approach will be especially advantageous compared to fingerprint-based approaches for discrimination and identification of very closely related microorganisms as bacterial subspecies, strains, or serovars. The cost of consumables is minimal, and the whole process takes less than 5 min.

Supplementary Material

[Supplemental material]

Acknowledgments

This work was supported by a grant from the German Federal Ministry of Economics and Technology (AiF/ProInno II grant KF 0350101 MD6).

We are grateful to Gabor Balizs for fruitful discussions and support.

Footnotes

[down-pointing small open triangle]Published ahead of print on 24 October 2008.

Supplemental material for this article may be found at http://aem.asm.org/.

REFERENCES

1. Arnold, R. J., and J. P. Reilly. 1998. Fingerprint matching of E. coli strains with matrix-assisted laser desorption/ionization time-of-flight mass spectrometry of whole cells using a modified correlation approach. Rapid Commun. Mass Spectrom. 12:630-636. [PubMed]
2. Arnold, R. J., and J. P. Reilly. 1999. Observation of Escherichia coli ribosomal proteins and their posttranslational modifications by mass spectrometry. Anal. Biochem. 269:105-112. [PubMed]
3. Boyd, E. F., J. Li, H. Ochman, and R. K. Selander. 1997. Comparative genetics of the inv-spa invasion gene complex of Salmonella enterica. J. Bacteriol. 179:1985-1991. [PMC free article] [PubMed]
4. Boyd, E. F., F. S. Wang, T. S. Whittam, and R. K. Selander. 1996. Molecular genetic relationships of the salmonellae. Appl. Environ. Microbiol. 62:804-808. [PMC free article] [PubMed]
5. Camara, J. E., and F. A. Hays. 2007. Discrimination between wild-type and ampicillin-resistant Escherichia coli by matrix-assisted laser desorption/ionization time-of-flight mass spectrometry. Anal. Bioanal. Chem. 389:1633-1638. [PubMed]
6. Chen, P., Y. Lu, and P. B. Harrington. 2008. Biomarker profiling and reproducibility study of MALDI-MS measurements of Escherichia coli by analysis of variance-principal component analysis. Anal. Chem. 80:1474-1481. [PubMed]
7. Christensen, H., S. Nordentoft, and J. E. Olsen. 1998. Phylogenetic relationships of Salmonella based on rRNA sequences. Int. J. Syst. Bacteriol. 48:605-610. [PubMed]
8. Claydon, M. A., S. N. Davey, V. Edwards-Jones, and D. B. Gordon. 1996. The rapid identification of intact microorganisms using mass spectrometry. Nat. Biotechnol. 14:1584-1586. [PubMed]
9. Crosa, J. H., D. J. Brenner, W. H. Ewing, and S. Falkow. 1973. Molecular relationships among the salmonellae. J. Bacteriol. 115:307-315. [PMC free article] [PubMed]
10. Dieckmann, R., I. Graeber, I. Kaesler, U. Szewzyk, and H. von Döhren. 2005. Rapid screening and dereplication of bacterial isolates from marine sponges of the Sula Ridge by intact-cell-MALDI-TOF mass spectrometry (ICM-MS). Appl. Microbiol. Biotechnol. 67:539-548. [PubMed]
11. Fenselau, C., and P. A. Demirev. 2001. Characterization of intact microorganisms by MALDI mass spectrometry. Mass Spectrom. Rev. 20:157-171. [PubMed]
12. Friedrichs, C., A. C. Rodloff, G. S. Chhatwal, W. Schellenberger, and K. Eschrich. 2007. Rapid identification of viridans streptococci by mass spectrometric discrimination. J. Clin. Microbiol. 45:2392-2397. [PMC free article] [PubMed]
13. Gasteiger, E., C. Hoogland, A. Gattiker, S. Duvaud, M. R. Wilkins, R. D. Appel, and A. Bairoch. 2005. Protein identification and analysis tools on the ExPASy server, p. 571-607. In J. M. Walker (ed.), The proteomics protocol handbook. Humana Press, Totowa, NJ.
14. Grimont, P. A. D., and F.-X. Weill. 2007. Antigenic formulae of the Salmonella serovars, 9th ed. WHO Collaborating Centre for Reference and Research on Salmonella. Institut Pasteur, Paris, France.
15. Hettick, J. M., M. L. Kashon, J. E. Slaven, Y. Ma, J. P. Simpson, P. D. Siegel, G. N. Mazurek, and D. N. Weissman. 2006. Discrimination of intact mycobacteria at the strain level: a combined MALDI-TOF MS and biostatistical analysis. Proteomics 6:6416-6425. [PubMed]
16. Keys, C. J., D. J. Dare, H. Sutton, I. Wells, M. Lunt, T. McKenna, M. McDowall, and H. N. Shah. 2004. Compilation of a MALDI-TOF mass spectral database for the rapid screening and characterization of bacteria implicated in human infectious diseases. Infect. Genet. Evol. 4:221-242. [PubMed]
17. Krishnamurthy, T., and P. L. Ross. 1996. Rapid identification of bacteria by direct matrix-assisted laser desorption/ionization mass spectrometric analysis of whole cells. Rapid Commun. Mass Spectrom. 10:1992-1996. [PubMed]
18. Lay, J. O. 2001. MALDI-TOF mass spectrometry of bacteria. Mass Spectrom. Rev. 20:172-194. [PubMed]
19. LeMinor, L., M. Veron, and M. Popoff. 1982. Taxonomie des Salmonella. Ann. Inst. Pasteur Microbiol. 133B:224-243.
20. LeMinor, L., M. Veron, and M. Popoff. 1982. Proposition pour une nomenclature des Salmonella. Ann. Inst. Pasteur Microbiol. 133B:245-254. [PubMed]
21. LeMinor, L., M. Y. Popoff, B. Laurent, and D. Hermant. 1986. Characterization of a 7th subspecies of Salmonella: S. choleraesuis subsp. indica subsp. nov. Ann. Inst. Pasteur Microbiol. 137B:211-217. (In French.) [PubMed]
22. Leuschner, R. G. K., N. Beresford-Jones, and C. Robinson. 2004. Difference and consensus of whole cell Salmonella enterica subsp. enterica serovars matrix-assisted laser desorption/ionization time-of-flight mass spectrometry spectra. Lett. Appl. Microbiol. 38:24-31. [PubMed]
23. Li, J., H. Ochman, E. A. Groisman, E. F. Boyd, F. Solomon, K. Nelson, and S. K. Selander. 1995. Relationship between evolutionary rate and cellular location among the Inv/Spa invasion proteins of Salmonella enterica. Proc. Natl. Acad. Sci. USA 92:7252-7256. [PMC free article] [PubMed]
24. Maiden, M. C. J., J. A. Bygraves, E. Feil, G. Morelli, J. E. Russell, R. Urwin, Q. Zhang, J. Zhou, K. Zurth, D. A. Caugant, I. M. Feavers, M. Achtman, and B. G. Spratt. 1998. Multilocus sequence typing: a portable approach to the identification of clones within populations of pathogenic microorganisms. Proc. Natl. Acad. Sci. USA 95:3140-3145. [PMC free article] [PubMed]
25. Majcherczyk, P. A., T. McKenna, P. Moreillon, and P. Vaudaux. 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. [PubMed]
26. Mandrell, R. E., L. A. Harden, A. Bates, W. G. Miller, W. F. Haddon, and C. K. Fagerquist. 2005. Speciation of Campylobacter coli, C. jejuni, C. helveticus, C. lari, C. sputorum, and C. upsaliensis by matrix-assisted laser desorption ionization-time of flight mass spectrometry. Appl. Environ. Microbiol. 71:6292-6307. [PMC free article] [PubMed]
27. Mazzeo, M. F., A. Sorrentino, M. Gaita, G. Cacace, M. Di Stasio, A. Facchiano, G. Comi, A. Malorni, and R. A. Siciliano. 2006. Matrix-assisted laser desorption ionization-time of flight mass spectrometry for the discrimination of food-borne microorganisms. Appl. Environ. Microbiol. 72:1180-1189. [PMC free article] [PubMed]
28. Pineda, F. J., M. D. Antoine, P. A. Demirev, A. B. Feldman, J. Jackman, M. Longenecker, and J. S. Lin. 2003. Microorganism identification by matrix-assisted laser/desorption ionization mass spectrometry and model-derived ribosomal protein biomarkers. Anal. Chem. 75:3817-3822. [PubMed]
29. Reeves, M. W., I. M. Evins, A. A. Heiba, B. D. Plikaytis, and J. J. Farmer III. 1989. Clonal nature of Salmonella typhi and its genetic relatedness to other salmonellae as shown by multilocus enzyme electrophoresis, and proposal of Salmonella bongori comb. nov. J. Clin. Microbiol. 27:313-320. [PMC free article] [PubMed]
30. Rhyzov, V., and C. Fenselau. 2001. Characterization of the protein subset desorbed by MALDI from whole bacterial cells. Anal. Chem. 73:746-750. [PubMed]
31. Ruelle, V., B. El Moualij, W. Zorzi, P. Ledent, and E. De Pauw. 2004. Rapid identification of bacterial strains by matrix-assisted laser desorption/ionization time-of-flight mass spectrometry. Rapid Commun. Mass. Spectrom. 18:2013-2019. [PubMed]
32. Rupf, S., K. Breitung, W. Schellenberger, K. Merte, S. Kneist, and K. Eschrich. 2005. Differentiation of mutans streptococci by intact cell matrix-assisted laser desorption/ionization time-of-flight mass spectrometry. Oral Microbiol. Immunol. 20:267-273. [PubMed]
33. Russell, S. C., N. Edwards, and C. Fenselau. 2007. Detection of plasmid insertion in Escherichia coli by MALDI-TOF mass spectrometry. Anal. Chem. 79:5399-5406. [PubMed]
34. Scott, F., J. Threlfall, and C. Arnold. 2002. Genetic structure of Salmonella revealed by fragment analysis. Int. J. Syst. Evol. Microbiol. 52:1701-1713. [PubMed]
35. Selander, R. K., J. Li, and K. Nelson. 1996. Evolutionary genetics of Salmonella enterica, p. 2691-2707. In F. C. Neidhardt, R. Curtiss III, J. L. Ingraham, E. C. C. Lin, K. B. Low, B. Magasanik, W. S. Reznikoff, M. Riley, M. Schaechter, and H. E. Umbarger (ed.), Escherichia coli and Salmonella: cellular and molecular biology, 2nd ed., vol. 2. American Society for Microbiology, Washington, DC.
36. Selander, R. K., J. Li, E. F. Boyd, F.-S. Wang, and K. Nelson. 1994. DNA sequence analysis of the genetic structure of populations of Salmonella enterica and Escherichia coli, p. 17-49. In F. G. Priest, A. Ramos-Cormenzana, and B. J. Tindall (ed.), Bacterial diversity and systematics. Plenum Press, New York, NY.
37. Spratt, B. G. 1999. Multilocus sequence typing: molecular typing of bacterial pathogens in an era of rapid DNA sequencing and the Internet. Curr. Opin. Microbiol. 2:312-316. [PubMed]
38. Stoleru, I. H., L. Le Minor, and A. M. Lheritier. 1976. Polynucleotide sequence divergence among strains of Salmonella subgenus IV and closely related organisms. Ann. Inst. Pasteur Microbiol. 127A:477-486. [PubMed]
39. Teramoto, K., H. Sato, L. Sun, M. Torimura, H. Tao, H. Yoshikawa, Y. Hotta, A. Hosoda, and H. Tamura. 2007. Phylogenetic classification of Pseudomonas putida strains by MALDI-MS using ribosomal subunit proteins as biomarkers. Anal. Chem. 79:8712-8719. [PubMed]
40. Torpdahl, M., and P. Ahrens. 2004. Population structure of Salmonella investigated by amplified fragment length polymorphism. J. Appl. Microbiol. 97:566-573. [PubMed]
41. Uzzau, S., D. J. Brown, T. Wallis, S. Rubino, I. Leori, S. Bernard, J. Casadesus, D. J. Platt, and J. E. Olsen. 2000. Host adapted serotypes of Salmonella enterica. Epidemiol. Infect. 125:229-255. [PMC free article] [PubMed]
42. Vargha, M., Z. Takats, A. Konopka, and C. H. Nakatsu. 2006. Optimization of MALDI-TOF MS for strain level differentiation of Arthrobacter isolates. J. Mol. Methods 66:399-409. [PubMed]
43. Wilcox, S. K., G. S. Cavey, and J. D. Pearson. 2001. Single ribosomal protein mutations in antibiotic-resistant bacteria analyzed by mass spectrometry. Antimicrob. Agents Chemother. 45:3046-3055. [PMC free article] [PubMed]
44. Williams, T. L., D. Andrzejewski, J. O. Lay, Jr., and S. M. Musser. 2003. Experimental factors affecting the quality and reproducibility of MALDI TOF mass spectra obtained from whole bacteria cells. J. Am. Soc. Mass Spectrom. 14:342-351. [PubMed]

Articles from Applied and Environmental Microbiology are provided here courtesy of American Society for Microbiology (ASM)
PubReader format: click here to try

Formats:

Related citations in PubMed

See reviews...See all...

Cited by other articles in PMC

See all...

Links

Recent Activity

Your browsing activity is empty.

Activity recording is turned off.

Turn recording back on

See more...