Impact Factor 4.259 | CiteScore 4.30
More on impact ›

Original Research ARTICLE

Front. Microbiol., 19 May 2015 |

Unsuitability of MALDI-TOF MS to discriminate Acinetobacter baumannii clones under routine experimental conditions

Clara Sousa1, João Botelho2, Filipa Grosso2, Liliana Silva2, João Lopes3 and Luísa Peixe2*
  • 1Centro de Engenharia Biológica, Universidade do Minho, Braga, Portugal
  • 2UCIBIO/REQUIMTE, Laboratório de Microbiologia, Departamento de Ciências Biológicas, Faculdade de Farmácia, Universidade do Porto, Porto, Portugal
  • 3Departamento de Farmácia Galénica e Tecnologia Farmacêutica, Faculdade de Farmácia, Universidade de Lisboa, Lisboa, Portugal

MALDI-TOF MS (matrix-assisted laser desorption/ionization time-of-flight mass spectrometry) is now in the forefront for routine bacterial species identification methodologies, being its value for clonality assessment controversial. In this work we evaluated the potential of MALDI-TOF MS for assisting infection control by depicting Acinetobacter baumannii clones. Mass spectra of 58 A. baumannii clinical isolates belonging to the worldwide spread lineages (ST98, ST103, ST208, and ST218) isolated in our country, were obtained and analyzed with several chemometric tools (pseudo gel views, peakfind function, and partial least squares discriminant analysis). The clonal lineages were obtained using the “Oxford” scheme, belonging ST98, ST208, and ST218 to the international clone II and ST103 to an epidemic clonal lineage (SG5). Additionally, mass spectra of a highly diverse international collection of 38 isolates belonging to 22 sequence types (STs) were obtained for further comparisons. Pseudo gel views and direct peak pattern analysis did not allow the discrimination of A. baumannii isolates belonging to ST98, ST103, ST208, or ST218. Moreover, a partial least square discriminant analysis of the mass spectra considering two spectral ranges (2–20 kDa and 4–10 kDa) revealed a poor degree of discrimination with only 64.6 and 65.8% of correct ST assignments, respectively. Also, mass spectra of the international isolates (n = 38, 22STs) revealed a very congruent peak pattern among them as well as among the four lineages included in this work. Despite the increasing interest of MALDI-TOF MS for bacterial typing at different taxonomical levels, we demonstrated, using routine experimental conditions, the unsuitability of this methodology for A. baumannii clonal discrimination.


During the last decade, the rate of nosocomial infections caused by multidrug-resistant Acinetobacter baumannii (MDRAB) has increased worldwide. In particular, the growing number of carbapenem-resistant A. baumannii isolates, mainly due to the production of carbapenem-hydrolyzing class D β-lactamases (CHDLs) jeopardizes the treatment of infections caused by this agent (Higgins et al., 2010). The quick and reliable clonality assessment is crucial to rapidly trace its dissemination, assist antibiotherapy, and implement measures to constrain its dissemination.

Pulsed-field gel electrophoresis (PFGE), amplified fragment length polymorphism (AFLP) analysis, multilocus sequence typing (MLST), DiversilabTM typing, other PCR fingerprinting methods, and multiple-locus variable tandem repeat number analysis (MVLA) are among the current methods used for genotyping bacteria (Grosso et al., 2011b; Karah et al., 2012; Zander et al., 2012; Zarrilli et al., 2013). PCR and sequence-based methods which describe isolates numerically are easy to use in international networks. Specifically for A. baumannii, Sequence Groups (SGs) identification or Trilocus (Turton et al., 2007) sequence-based typing (3LST – selectively amplifies alleles of ompA, csuE, and blaOXA-51 genes), blaOXA-51-like genotyping and two MLST schemes1 (“Oxford” and Institute Pasteur) are available. The 3LST, blaOXA-51-like genotyping, and DiversilabTM typing (which consists in a repetitive extragenic palindromic PCR – rep-PCR) allow the rapid identification of the main multidrug-resistant A. baumannii lineages, being already established the concordance between some of these methods (Zarrilli et al., 2013). The MLST schemes share three loci but reveal different discriminatory abilities, with our data pointing for a higher resolution, congruent to PFGE analysis and CHDL content, of the “Oxford” scheme (Grosso et al., 2011b). MLST analysis in several collections resulted in the recognition of major clonal complexes (CCs) and sequence types (STs) responsible for antimicrobial resistance dissemination, such as CC92 (ST92, ST208, ST218), CC109 (ST109), CC103 (ST103, ST133), and CC113 (ST113) according with the Oxford scheme (Zarrilli et al., 2013).

However, most of these methods are too expensive and/or time consuming. Spectroscopic techniques might constitute reliable alternatives for bacterial typing at different taxonomic levels, with variable degrees of success reported from their application to several microorganisms (Mencacci et al., 2013; Šedo et al., 2013; Vaz et al., 2013; Branquinho et al., 2014a,b; Novais et al., 2014; Sousa et al., 2014a). Recently, we developed and validated a mathematical model for typing A. baumannii clones, most of them included in this study, based on spectra obtained by a competitive spectroscopic technique, Fourier transform infrared spectroscopy with attenuated total reflectance (FTIR-ATR), that could be routinely used (Sousa et al., 2014b). Moreover, using matrix assisted laser desorption/ionization time-of-flight mass spectrometry (MALDI-TOF MS) and chemometrics we were able to circumvent difficulties associated with the discrimination by MALDI-TOF MS of Acinetobacter species within the A. baumannii-calcoaceticus complex (Böhme et al., 2010; Espinal et al., 2011; Sousa et al., 2014a). In which concerns the suitability of MALDI-TOF MS for bacterial typing at subspecies level, controversial data is available (Barbuddhe et al., 2008; Dieckmann and Malorny, 2011; Gekenidis et al., 2014; Lasch et al., 2014; Novais et al., 2014; Wolters et al., 2011).

In this work, we evaluated the ability of MALDI-TOF MS combined with several chemometric tools for assisting infection control by depicting A. baumannii clones.

Materials and Methods

Rational of the Study

To evaluate the potential of MALDI-TOF MS to discriminate A. baumannii clones, mass spectra of 58 Portuguese clinical isolates belonging to ST98, ST103, ST208, and ST218 according with the “Oxford” scheme were obtained and analyzed with several chemometric tools. Mass profiles analysis was performed by different approaches: (i) spectral overview on the basis of pseudo gel views; (ii) analysis of the mass peak pattern according the number of peaks found with the Matlab function peakfind; and (iii) partial least squares discriminant analysis (PLSDA). The last two approaches also allowed the estimative of correct predictions for each ST and the PLSDA model was performed considering the entire (2–20 kDa) and a selected (4–10 kDa) spectral range where the majority of the mass peaks (about 90%) were located. Additionally, mass spectra of 38 isolates from different countries and belonging to 22 distinct STs were obtained and the respective pseudo gel views compared among each other and with the ones of ST98, ST103, ST208, and ST218 isolates. We further explored a possible correlation between the mass profiles of the isolates tested and clustering/classification obtained with other methods for major lineages A. baumannii identification, as AFLP, SG typing, and the MLST scheme of Institute Pasteur.

Bacterial Isolates

Fifty-eight A. baumannii clinical isolates were selected among a previously published collection of 318 CHDL-producing Acinetobacter spp. recovered from six geographically distant Portuguese hospitals (2001–2012). The selected isolates are representatives of the main lineages disseminated in Portugal before 2012: ST98 (n = 10); ST103 (n = 9); ST208 (n = 17; corresponding to a reassignment of ST92 isolates), and ST218 (n = 22; Grosso et al., 2011a,b; Sousa et al., 2014b). Additionally, 38 isolates belonging to 22 different STs were used for further comparisons. The main characteristics of the isolates used in this study, including the relationship between CHDL content, PFGE types, SGs, and the STs obtained according with Institute Pasteur and “Oxford” scheme1 among others were presented in Table 1.


TABLE 1. Epidemiological details of the Acinetobater baumannii isolates included in this study.

MALDI-TOF MS Experiments

Mass spectra were obtained from cell extracts prepared according the manufacturer instructions. Briefly, overnight cultures in Muller-Hinton agar were suspended in HPLC water and treated with ethanol (75%). After centrifugation and removal of the supernatant, cells were extracted with 25 μL of 70% formic acid followed by addition of 25 μL of acetonitrile and vortexing at 2000 rpm for 1 min. Samples were spotted onto MALDI ground steel target (AnchorChip) followed by drying and the addition of 1 μL of the chemical matrix (saturated solution of α-cyano-4-hydroxycinnamic acid in 50% of acetonitrile and 2.5% of trifluoroacetic acid). Spectra were randomly obtained from blind samples in the linear positive mode at a laser (nitrogen) frequency of 20 Hz in the range of 2–20 kDa with a Microflex III instrument (Bruker Daltonic, Bremen, Germany). Each recorded spectrum is the result of six series of 40 single laser shots in different locations. The experiments were performed in quadruplicate using four distinct spots of the MALDI target (instrumental replicates) at least in two different days (biological replicates) with two different bacterial cultures or extracts. External calibration of the mass spectra was performed using Escherichia coli DH5 alpha standard peaks (BTS).

Data Analysis

Due to the large amount of data generated by MALDI experiments, mean spectrum for each isolate was generated from the instrumental and biological replicates and considered for further analysis. Zero-line and low S/N ratio mass spectra were not considered to the average. The pseudo gel views were generated by the dedicated Matlab-based software MicrobeMS2 that provides direct access to the spectra of Bruker’s proprietary file format. In these gel views the intensities are gray-scaled (log scale) being the mass/charge ratios (m/z) the abscissa and spectral indices the ordinate values. In these bar code spectra only the information of peak presence or absense is employed, while the peak intensity is neglected.

Mass spectra were also analyzed with the peakfind function of the PLS Toolbox for Matlab (arguments: 9- number of points in Savitzky–Golay filter, 6- tolerance on the estimated residuals; peaks heights are estimated to be >toleranceresiduals) and 19- window width for determining local maxima to evaluate the intra and inter-ST variability among the four distinct STs. The method starts by estimating the peak mass-to-charge ratios of each ST. For this task, spectra of all isolates belonging to one ST were averaged, the result was submitted to the peak identification method and the peak locations were stored in a vector generating a “peak prototype” for each ST. This method was repeated for all STs. Then, the same peak identification method was run for each isolate individually. Peak positions of each “ST prototype” were compared with peak positions of each isolate. When a “ST prototype” peak location matched a peak location of an isolate, a value 1 was assigned for that peak; otherwise a value 0 was assigned. This procedure creates a vector of 0s and 1s for each pair “ST prototype”/isolate. Note that peak locations were considered to match if they were located within a mass-to-charge ratio difference lower than 7 m/z units [if for a certain peak location n, | m/z(prototype)n–m/z(sample)n| < 7, that peak is considered to match]. For each isolate, a percentage of matching peaks was estimated for each ST. Isolates were considered to belong to the ST yielding the highest percentage of peak matches.

For clustering purposes spectra were analyzed by PLSDA after the pre-processing mean-centring (Savitzky and Golay, 1964; Geladi and Kowalsky, 1986; Barker and Rayens, 2003). This pre-processing method allows removing the influence of different sample amounts and/or equipment variations in the peaks intensity. In PLSDA, to each known sample (xi) is assigned a vector of 0s with the value 1 at the position corresponding to its ST (yi). The structure of the PLSDA model is described by Eqs 1 and 2. Model loadings (P and Q) and corresponding scores (T and U) are obtained by sequentially extracting the components or latent variables (LVs) from matrices X (the spectra) and Y (the matrix codifying the STs).


The algorithm correlates the scores of each block (T and U), yielding an internal regression matrix. This internal regression can be transformed on a regression matrix (B). In this case, the regression matrix is composed by three vectors: one regression vector corresponding to each ST. E and F are the residual matrices and depend on the number of LV selected. Predictions for new samples are obtained by multiplying a new spectrum (xnew) by the regression matrix (B).


The prediction (ynew = [ynew,1,ynew,2,....,ynew,n]) is then converted in a class assignment. In PLSDA a probability value for each assignment is estimated for each sample. The model number of LV was optimized using the leave-one-sample-out cross-validation procedure in order to prevent overfitting. All chemometric models were performed in Matlab version 7.4 Release 2007a (MathWorks, Natick, MA, USA) and PLS Toolbox version 4.2.1 for Matlab (Eigenvector Research, Manson, WA, USA).


Spectral Overview

Mean mass profiles of the isolates belonging to the four studied STs are presented in Figure 1. A very similar peak pattern can be found among the isolates of a single ST (data not shown) and among the four STs with almost no differences between them. The common peaks found among the four STs are summarized in the figure. Figure 2 exhibits the pseudo gel views generated with MicrobeMS software considering all the isolates included in this study (ST98, S103, ST208, ST218 plus the 22 STs of the international collection). A high degree of consistency in the peak pattern can be found among the isolates of a single ST but also among the four STs (Figure 2A). It was impossible to obtain any degree of isolate discrimination according the ST solely based in the presence and/or absence of specific mass peak profiles even considering isolates epidemiologically unrelated and belonging to diverse STs (Figure 2B).


FIGURE 1. Mass profiles of ST98, ST103, ST208, and ST218 isolates in the region of 2–13 kDa [obtained from the mean spectra of each sequence types (ST)].


FIGURE 2. Pseudo gel view representation obtained from the average mass spectra of each Acinetobacter baumannii isolate of ST98, ST103, ST208, and ST218 (A) and of A. baumannii isolates of the international collection (B).

Peakfind Function of Matlab

Supplementary Figure S1 summarizes the peak positions founded with the peakfind function in the mass spectra from all isolates. No peaks were found above 17 kDa. Similarly to the observations in the pseudo gel view analysis, a very consistent peak pattern was observed among all the isolates with a low inter and intra-ST spectral variability. The comparison of each isolate peak profile with the four mean-ST peak profiles (see Materials and Methods) was used for estimating the ST of each isolate, Figure 3. Nevertheless, it was only possible to correctly predict the ST for 58.6% of the isolates. ST103 isolates were always correctly predicted; however, 19/58 isolates were erroneously predicted as ST103, meaning a high sensitivity (9/9 = 100%) and low specificity (30/51 = 58.8%) for the ST103 prediction. Moreover, it was possible to correctly predict 70% of the ST98 isolates; 41.2% of the ST208 isolates and 50% of the ST218 isolates.


FIGURE 3. Sequence type assignments for all isolates considering the spectral matching method. Colors identify the real ST of each isolate and bars correspond to the method’s ST predictions (58.6% of correct predictions). Legend: ST98, ST103, ST208, and ST218.

Partial Least Squares Discriminant Analysis

The PLSDA models developed considering the entire (2–20 kDa) and a selected (4–10 kDa) spectral range are presented in Figures 4A,B, respectively. Considering the entire spectral range (Figure 4A) it was not possible to clearly discriminate the four STs as no individualized clusters could be found in the score map of the model. The PLSDA model was able to correctly predict the STs of 64.6% of the isolates (Table 2) being ST103 the one with a larger percentage of correct predictions (75.7%). The worst cases were observed for ST98 and ST208 for which plus than 40% of the isolates were erroneously predicted. Similarly, the PLSDA model obtained with the selected spectral range (4–10 kDa) did not allow the discrimination of the four STs (Figure 4B) still being ST103 the best predicted (73.7%), Table 2. Moreover, the total percentage of correct predictions was slightly higher (65.8%) for each of the three remaining STs considering the 4–10 kDa range.


FIGURE 4. Score plot corresponding to the first three latent variables (LVs) of the partial least squares discriminant analysis (PLSDA) regression model using the entire, 2–20 kDa, (A) and a selected, 4–10 kDa, (B) spectral range. Legend: ST98, ST103, ST208, and ST218.


TABLE 2. Confusion matrix for both of the A. baumannii PLSDA discrimination models, considering 20 LV (values are in %).

Partial least squares discriminant analysis models were also developed (whenever possible and according to the available information, please see Table 1) to correlate the mass profiles of the isolates with the results obtained from less discriminatory typing methods as AFLP, SG, and MLST scheme of Institute Pasteur. However, the clustering analysis was not congruent with any of these methods (data not shown).


In the last years an impressive growing number of studies unveiling the MALDI-TOF MS potential for routine bacterial species identification has been published (Böhme et al., 2010; Carbonnelle et al., 2010; Branquinho et al., 2014a,b; Sousa et al., 2014a). However, the suitability of this mass spectrometry technique for bacterial discrimination at the subspecies level has been barely explored, with contradictory outcomes for particular species (Barbuddhe et al., 2008; Dieckmann and Malorny, 2011; Wolters et al., 2011; Gekenidis et al., 2014; Lasch et al., 2014; Novais et al., 2014). The goal of this work was to assess the ability of MALDI-TOF MS to depict A. baumannii clones, with particular interest for the worldwide spread lineages (ST98, ST103, ST208, and ST218), contributing to a better understanding of the capabilities and limitations of this technique in bacterial typing. Analysis of the mass spectra of the four STs was attempted using three approaches and revealed some mass-to-charge ratios already identified as Acinetobacter genus and A. baumannii species-specific (Böhme et al., 2010; Espinal et al., 2011; Šedo et al., 2013; Sousa et al., 2014a). However, the high similarity among mass profiles of the A. baumannii lineages analyzed prevented the STs discrimination either by peak pattern direct analysis (Figure 1) or based on the presence/absence of specific peaks depicted in the pseudo gel views (Figure 2A). Moreover, attempting to assign a ST based on the comparison of each isolate’s mass profile with the mean-ST profile also resulted in a low percentage of correct identifications (58%, Figure 3B). Previous studies, including from our group (Novais et al., 2014; Sousa et al., 2014a), have demonstrate that the use of specific and optimized chemometric tools in MALDI-TOF MS data analysis improves the bacterial discrimination derived from this spectroscopic methodology. In this context, we attempted to discriminate the isolates with a PLSDA analysis considering two distinct mass ranges (Figures 4A,B). Although the degree of discrimination slightly improved, only 64.6 and 65.8% of correct STs predictions were obtained for the two considered mass ranges, demonstrating the current inadequacy of MALDI-TOF MS for discrimination of major A. baumannii STs. The difficulty to differentiate these STs based on MALDI-TOF MS analysis could possibly be associated with the relatedness of their allelic profiles. In fact, ST98 is a double locus variant of ST208 and ST218 in gyrB and gpi and ST208 a single locus variant of ST218 in the gpi allele. Despite the low ability to discriminate the four STs, A. baumannii ST103 isolates always presented the higher rate of correct ST predictions whether considering the comparison of the ST-mean mass profiles or the chemometric approach. It is of note that the allelic profile of ST103 isolates is the most dissimilar one, presenting only one common allele with ST98, the gpi one3. This fact could contribute to a more dissimilar ribosomal protein/peptide profile of ST103 isolates and its subsequent higher rate of correct identifications. The difficulty to differentiate these four STs based on their mass spectra suggests that these clones possess a very similar profile in what concerns to the molecules routinely observed in these MALDI experiments.

As a high throughput technique, MALDI-TOF MS competes with other spectroscopic techniques as Raman (Maquelin et al., 2006), and Fourier Transform Infrared Spectroscopy (FTIR) for bacterial typing at different taxonomic levels. It is of note, the suitability of FTIR to discriminate A. baumannii lineages, including the STs included in this study (Sousa et al., 2014b), which is also a sensitive, quick and low cost technique. Nevertheless, with the recognition of the interest in the microbiological diagnostic of MALDI-TOF MS, associated with the increasing availability of MALDI-TOF MS equipment in routine laboratories, there is a particular interest on methodology developments assisting bacterial epidemiology. In this way, further assays testing the ability of MALDI-TOF MS to discriminate A. baumannii lineages with different sample preparation conditions and matrix solutions should be conducted. It also should be noted that, despite our MALDI data had been compared with four distinct classification methods (two different MLST schemes, AFLP and SG), presenting different typing resolutions, it was not congruent with the grouping obtained with any of these approaches. In this way, it does not offer, in these experimental conditions, an advantage over other rapid methods such as DiversiLab rep-PCR-based typing, trilocus sequence-based typing, or single-locus-sequence-based typing of blaOXA-51-like genes. However, we do not exclude the possibility that other classification method could somehow fit MALDI-TOF MS data.


In this work we evaluate the ability of MALDI-TOF MS to discriminate A. baumannii clones. This mass spectroscopic technique, which revealed in previous studies a high discrimination power for species identification within the A. calcoaceticus–A. baumannii complex, demonstrates an insufficient result when used for discrimination at the clonal level. These findings suggest that the detected molecules, mainly ribosomal peptides and/or proteins, remain unchanged during clonal diversification in this species. Further studies, namely using different sample preparation conditions, are needed to provide further insights on the suitability of MALDI-TOF MS for typing A. baumannii at a subspecies level.

Conflict of Interest Statement

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.


This work was funded by FEDER funds through the Operational Programme for Competitiveness Factors – COMPETE and by National Funds through FCT – Foundation for Science and Technology under the project UID/MULTI/04378/2013. LS was supported by doctoral grant (SFRH/BD/88028/2012) and FG by a post doc grant (SFRH/BPD/95556/2013) from Fundação para a Ciência e a Tecnologia. CS was funded by a CIENCIA2008 contract from FCT.

Supplementary Material

The Supplementary Material for this article can be found online at:


  1. ^
  2. ^
  3. ^


Barbuddhe, S. B., Maier, T., Schwarz, G., Kostrzewa, M., Hof, H., Domann. E., et al. (2008). Rapid identification and typing of listeria species by matrix-assisted laser desorption ionization-time of flight mass spectrometry. Appl. Environ. Microbiol. 74, 5402–5407. doi: 10.1128/AEM.02689-07

PubMed Abstract | CrossRef Full Text | Google Scholar

Barker, M., and Rayens, W. (2003). Partial least squares for discrimination. J. Chemometr. 17, 166–173. doi: 10.1002/cem.785

CrossRef Full Text | Google Scholar

Böhme, K., Fernández-No, I. C., Barros-Velázquez, J., Gallardo, J. M., Calo-Mata, P., and Cañas, B. (2010). Species differentiation of seafood spoilage and pathogenic Gram-negative bacteria by MALDI-TOF mass fingerprinting. J. Proteome Res. 9, 3169–3183. doi: 10.1021/pr100047q

PubMed Abstract | CrossRef Full Text | Google Scholar

Branquinho, R., Sousa, C., Lopes, J., Pintado, M. E., Peixe, L. V., and Osório, H. (2014a). Differentiation of Bacillus pumilus and Bacillus safensis using MALDI-TOF/MS. PLoS ONE 9:e110127. doi: 10.1371/journal.pone.0110127

PubMed Abstract | CrossRef Full Text | Google Scholar

Branquinho, R., Sousa, C., Osório, H., Meirinhos-Soares, L., Lopes, J., Carriço, J. A., et al. (2014b). Bacillus invictae sp. nov., isolated from a health product. Int. J. Syst. Evol. Microbiol. 64, 3867–3876. doi: 10.1099/ijs.0.067850-0

PubMed Abstract | CrossRef Full Text | Google Scholar

Carbonnelle, E., Mesquita, C., Bille, E., Day, N., Dauphin, B., Beretti, J. L., et al. (2010). MALDI-TOF mass spectrometry tools for bacterial identification in clinical microbiology laboratory. Clin. Biochem. 44, 104–109. doi: 10.1016/j.clinbiochem.2010.06.017

PubMed Abstract | CrossRef Full Text | Google Scholar

D’Arezzo, S., Capone, A., Petrosillo, N., and Visca, P. (2009). Epidemic multidrug-resistant Acinetobacter baumannii related to European clonal types I and II in Rome (Italy). Clin. Microbiol. Infect. 15, 347–357. doi: 10.1111/j.1469-0691.2009.02668.x

PubMed Abstract | CrossRef Full Text | Google Scholar

D’Arezzo, S., Principe, L., Capone, A., Petrosillo, N., Petrucca, A., and Visca, P. (2011). Changing carbapenemase gene pattern in an epidemic multidrug-resistant Acinetobacter baumannii lineage causing multiple outbreaks in central Italy. J. Antimicrob. Chemother. 66, 54–61. doi: 10.1093/jac/dkq407

PubMed Abstract | CrossRef Full Text | Google Scholar

Dieckmann, R., and Malorny, B. (2011). Rapid screening of epidemiologically important Salmonella enterica subsp. enterica serovars by whole-cell matrix-assisted laser desorption ionization-time of flight mass spectrometry. Appl. Environ. Microbiol. 77, 4136–4146. doi: 10.1128/AEM.02418-10

PubMed Abstract | CrossRef Full Text

Espinal, P., Seifert, H., Dijkshoorn, L., Vila, J., and Roca, I. (2011). Rapid and accurate identification of genomic species from the Acinetobacter baumannii (Ab) group by MALDI-TOF MS. Clin. Microbiol. Infect. 18, 1097–1103. doi: 10.1111/j.1469-0691.2011.03696.x

PubMed Abstract | CrossRef Full Text | Google Scholar

Gekenidis, M. T., Studer, P., Wüthrich, S., Brunisholz, R., and Drissner, D. (2014). Beyond the matrix-assisted laser desorption ionization (MALDI) biotyping workflow: in search of microorganism-specific tryptic peptides enabling discrimination of subspecies. Appl. Environ. Microbiol. 80, 4234–4241. doi: 10.1128/AEM.00740-14

PubMed Abstract | CrossRef Full Text | Google Scholar

Geladi, P., and Kowalsky, B. R. (1986). Partial least-squares regression: a tutorial. Anal. Chim. Acta. 185, 1–17. doi: 10.1016/0003-2670(86)80028-9

CrossRef Full Text | Google Scholar

Goic-Barisic, I., Towner, K. J., Kovacic, A., Sisko-Kraljevic, K., Tonkic, M., Novak, A., et al. (2011). Outbreak in Croatia caused by a new carbapenem-resistant clone of Acinetobacter baumannii producing OXA-72 carbapenemase. J. Hosp. Infect. 77, 368–369. doi: 10.1016/j.jhin.2010.12.003

PubMed Abstract | CrossRef Full Text | Google Scholar

Grosso, F., Carvalho, K. R., Quinteira, S., Ramos, A., Carvalho-Assef, A. P., Asensi, M. D., et al. (2011a). OXA-23-producing Acinetobacter baumannii: a new hotspot of diversity in Rio de Janeiro? J. Antimicrob. Chemother. 66, 62–65. doi: 10.1093/jac/dkq406

PubMed Abstract | CrossRef Full Text | Google Scholar

Grosso, F., Quinteira, S., and Peixe, L. (2011b). Understanding the dynamics of imipenem-resistant Acinetobacter baumannii lineages within Portugal. Clin. Microbiol. Infect. 17, 1275–1279. doi: 10.1111/j.1469-0691.2011.03469.x

PubMed Abstract | CrossRef Full Text | Google Scholar

Higgins, P. G., Dammhayn, C., Hackel, M., and Seifert, H. (2010). Global spread of carbapenem-resistant Acinetobacter baumannii. J. Antimicrob. Chemother. 65, 233–238. doi: 10.1093/jac/dkp428

PubMed Abstract | CrossRef Full Text | Google Scholar

Karah, N., Sundsfjord, A., Towner, K., and Samuelsen, Ø. (2012). Insights into the global molecular epidemiology of carbapenem non-susceptible clones of Acinetobacter baumannii. Drug Resist. Updat. 15, 237–247. doi: 10.1016/j.drup.2012.06.001

PubMed Abstract | CrossRef Full Text | Google Scholar

Lasch, P., Fleige, C., Stämmler, M., Layer, F., Nübel, U., Witte, W., et al. (2014). Insufficient discriminatory power of MALDI-TOF mass spectrometry for typing of Enterococcus faecium and Staphylococcus aureus isolates. J. Microbiol. Methods 100, 58–69. doi: 10.1016/j.mimet.2014.02.015

PubMed Abstract | CrossRef Full Text | Google Scholar

Maquelin, K., Dijkshoorn, L., van der Reijden, T. J., and Puppels, G. J. (2006). Rapid epidemiological analysis of Acinetobacter strains by Raman spectroscopy. J. Microbiol. Methods 64, 126–131. doi: 10.1016/j.mimet.2005.04.028

PubMed Abstract | CrossRef Full Text | Google Scholar

Mencacci, A., Monari, C., Leli, C., Merlini, L., De Carolis, E., Vella, A., et al. (2013). Typing of nosocomial outbreaks of Acinetobacter baumannii by use of matrix-assisted laser desorption ionization-time of flight mass spectrometry. J. Clin. Microbiol. 51, 603–606. doi: 10.1128/JCM.01811-12

PubMed Abstract | CrossRef Full Text | Google Scholar

Novais, Â., Sousa, C., de Dios Caballero, J., Fernandez-Olmos, A., Lopes, J., Ramos, H., et al. (2014). Discrimination of high-risk Escherichia coli clones from phylogenetic groups B2 (ST131) and D (ST69, ST405, ST393) by MALDI-TOF mass spectrometry. Eur. J. Clin. Microbiol. Infect. Dis. 33, 1391–1399. doi: 10.1007/s10096-014-2071-5

PubMed Abstract | CrossRef Full Text

Savitzky, A., and Golay, M. J. E. (1964). Smoothing and differentiation of data by simplified least squares procedures. Anal. Chem. 36, 1627–1639. doi: 10.1021/ac60214a047

PubMed Abstract | CrossRef Full Text | Google Scholar

Šedo, O., Nemec, A., Křížová, L., Kačalová, M., and Zdráhal, Z. (2013). Improvement of MALDI-TOF MS profiling for the differentiation of species within the Acinetobacter calcoaceticus- Acinetobacter baumannii complex. Syst. Appl. Microbiol. 36, 572–578. doi: 10.1016/j.syapm.2013.08.001

PubMed Abstract | CrossRef Full Text | Google Scholar

Seifert, H., Dolzani, L., Bressan, R., van der Reijden, T., van Strijen, B., Stefanik, D., et al. (2005). Standardization and interlaboratory gel electrophoresis-generated fingerprints reproducibility assessment of pulsed-field of Acinetobacter baumannii. J. Clin. Microbiol. 43, 4328–4335. doi: 10.1128/JCM.43.9.4328-4335.2005

PubMed Abstract | CrossRef Full Text

Sousa, C., Botelho, J., Silva, L., Grosso, F., Nemec, A., Lopes, J., et al. (2014a). MALDI-TOF MS and chemometric based identification of Acinetobacter calcoaceticus-Acinetobacter baumannii complex species. Int. J. Med. Microbiol. 304, 669–677. doi: 10.1016/j.ijmm.2014.04.014

PubMed Abstract | CrossRef Full Text | Google Scholar

Sousa, C., Silva, L., Grosso, F., Lopes, J., and Peixe, L. (2014b). Development of a FTIR-ATR based model for typing clinically relevant Acinetobacter baumannii clones belonging to ST98, ST103, ST208 and ST218. J. Photochem. Photobiol. B 133, 108–114. doi: 10.1016/j.jphotobiol.2014.02.015

PubMed Abstract | CrossRef Full Text | Google Scholar

Turton, J., Gabriel, S., Valderrey, C., Kaufmann, M., and Pitt, T. (2007). Use of sequence-based typing and multiplex PCR to identify clonal lineages of outbreak strains of Acinetobacter baumannii. Clin. Microbiol. Infect. 13, 807–815. doi: 10.1111/j.1469-0691.2007.01759.x

PubMed Abstract | CrossRef Full Text | Google Scholar

Vaz, M., Meirinhos-Soares, L., Sousa, C. C., Ramirez, M., Melo-Cristino, J., and Lopes, J. A. (2013). Serotypes discrimination of encapsulated Streptococcus pneumoniae strains by Fourier-transform infrared spectroscopy. J. Microbiol. Methods 93, 102–107. doi: 10.1016/j.mimet.2013.02.008

PubMed Abstract | CrossRef Full Text | Google Scholar

Wolters, M., Rohde, H., Maier, T., Belmar-Campos, C., Franke, G., Scherpe, S., et al. (2011). MALDI-TOF MS fingerprinting allows for discrimination of major methicillin-resistant Staphylococcus aureus lineages. Int. J. Med. Microbiol. 301, 64–68. doi: 10.1016/j.ijmm.2010.06.002

PubMed Abstract | CrossRef Full Text | Google Scholar

Zander, E., Nemec, A., Seifert, H., and Higgins, P. (2012). Association between β-lactamase-encoding blaOXA-51 variants and DiversiLab rep-PCR-based typing of Acinetobacter baumannii isolates. J. Clin. Microbiol. 50, 1900–1904. doi: 10.1128/JCM.06462-11

PubMed Abstract | CrossRef Full Text | Google Scholar

Zarrilli, R., Pournaras, S., Giannouli, M., and Tsakris, A. (2013). Global evolution of multidrug-resistant Acinetobacter baumannii clonal lineages. Int. J. Antimicrob. Agents 41, 11–19. doi: 10.1016/j.ijantimicag.2012.09.008

PubMed Abstract | CrossRef Full Text | Google Scholar

Keywords: Acinetobacter baumannii, typing, MALDI-TOF MS, subspecies, chemometrics, sequence type

Citation: Sousa C, Botelho J, Grosso F, Silva L, Lopes J and Peixe L (2015) Unsuitability of MALDI-TOF MS to discriminate Acinetobacter baumannii clones under routine experimental conditions. Front. Microbiol. 6:481. doi: 10.3389/fmicb.2015.00481

Received: 19 February 2015; Accepted: 01 May 2015;
Published online: 19 May 2015

Edited by:

Antonio C. M. Correia, Universidade de Aveiro, Portugal

Reviewed by:

Raffaele Zarrilli, University of Naples Federico II, Italy
Kjell Sergeant, Luxembourg Institute of Science and Technology, Luxembourg

Copyright © 2015 Sousa, Botelho, Grosso, Silva, Lopes and Peixe. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) or licensor are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

*Correspondence: Luísa Peixe, UCIBIO/REQUIMTE, Laboratório de Microbiologia, Departamento de Ciências Biológicas, Faculdade de Farmácia da Universidade do Porto, R. Jorge Viterbo Ferreira, 228, 4050-313 Porto, Portugal,