Identification of non fermenting Gram negative bacilli isolated in cystic fibrosis by

Transcription

1 JCM Accepts, published online ahead of print on 6 August 2008 J. Clin. Microbiol. doi: /jcm Copyright 2008, American Society for Microbiology and/or the Listed Authors/Institutions. All Rights Reserved. 1 2 Identification of non fermenting Gram negative bacilli isolated in cystic fibrosis by Matrix assisted laser desorption ionization time-of-flight mass spectrometry Nicolas Degand 1,2, Etienne Carbonnelle 1, 2, Brunhilde Dauphin 2, Jean-Luc Beretti 1, Muriel Le Bourgeois 3, Isabelle Sermet-Gaudelus 4, Christine Segonds 5, Patrick Berche 1,2, Xavier Nassif 1,2, Agnès Ferroni 1*. 1 Assistance Publique-Hôpitaux de Paris, Laboratoire de Microbiologie, Hôpital Necker- Enfants Malades, Paris, France, 2 Université Paris Descartes, Faculté de médecine, Paris, France, 3 Service de Pneumologie Pédiatrique, 4 Service de Pédiatrie Générale, Hôpital Necker-Enfants Malades, Paris, France, 5 Observatoire Cepacia, Hôpital Purpan, Toulouse, France. Short title: Cystic fibrosis bacterial identification Key words : MALDI-TOF-MS, cystic fibrosis, bacterial identification, Gram negative bacilli * Corresponding author Agnès Ferroni Laboratoire de Microbiologie Hôpital Necker-Enfants Malades 149 rue de Sèvres Paris, France Tel: Fax: agnes.ferroni@nck.aphp.fr 1

2 Abstract The identification of nonfermenting Gram-negative bacilli isolated from cystic fibrosis (CF) patients is usually achieved using phenotypic based techniques and eventually molecular tools. These techniques remain time consuming, expensive, and technically demanding. We used a method based on Matrix Assisted Laser Desorption Ionization Time-Of-Flight Mass spectrometry (MALDI-TOF-MS) for the identification of these bacteria. A set of reference strains belonging to 58 species of clinically relevant nonfermenting Gramnegative bacilli was used. To identify peaks discriminating between these various species, the profile of 10 isolated colonies obtained from 10 different passages was analyzed for each referenced strain. Conserved peaks with a relative intensity above 0.1 were retained. The spectra of 559 clinical isolates were then compared with that of each of the 58 reference strains : 400 Pseudomonas aeruginosa, 54 Achromobacter xylosoxydans, 32 Stenotrophomonas maltophilia, 52 Burkholderia cepacia complex (Bcc), 1 Burkholderia gladioli, 14 Ralstonia mannitolilytica, 2 Ralstonia picketti, 1 Bordetella hinzii, 1 Inquilinus limosus, 1 Cupriavidus respiraculi, 1 Burkholderia thailandensis. Using this database, 549 strains were correctly identified. Nine Bcc and 1 R. mannnitolilytica strains were identified as belonging to the appropriate genus but not the correct species. We subsequently engineered Bcc and Ralstonia specific databases using additional reference strains: using these databases, correct identification for these species increased from 83 to 98% and 94 to 100 % of cases, respectively. Altogether, these data demonstrates that, in CF patients, MALDI-TOF-MS is a powerful tool for rapid identification of nonfermenting Gram-negative bacilli

3 Introduction Pseudomonas aeruginosa is the main bacterial pathogen isolated from the sputum of patients suffering from Cystic Fibrosis (CF). Other non fermenting Gram negative bacilli of clinical importance that can be isolated from this location are Achromobacter xylosoxydans, Stenotrophomonas maltophilia or species belonging to the Burkholderia cepacia complex (Bcc). Other non fermenting bacteria belonging to the genus Ralstonia or Burkholderia are less likely to be isolated. Recently, novel bacterial species have been isolated from CF patients as Ralstonia mannitolilytica (6,10), Pandoraea apista (4,22), Inquilinus limosus (5,25). Bacteria isolated during CF do not have the same virulence, thus pointing out the importance of bacterial identification by routine clinical microbiology laboratories for the management of CF patients. Moreover, the emergence of new bacterial species requires accurate identification tools in order to understand their clinical relevance and distribution in CF. Conventional phenotypic methods and commercial kits are sometimes not suitable for strains isolated from CF patients. These pathogens often lack key phenotypic characters required for their identification (1,10,20,23,24,26). In addition, in some circumstances, misidentification is due to the fact that the species are not in the database of commercial kits (10,23). Molecular tools such as 16S rrna gene sequencing provide reliable results (10,16,26). Other techniques such as FISH (27) and amplified ribosomal DNA restriction assays are available (22). Despite their good accuracy, these molecular techniques cannot be used routinely as they are expensive, time consuming and technically demanding. Several studies have reported the use of Matrix Assisted Laser Desorption Ionization Time- Of-Flight Mass Spectrometry (MALDI-TOF-MS) for bacterial identification (8,9,13-15,17). MALDI-TOF-MS can examine the profile of proteins detected directly from intact bacteria. This technique, based on relative molecular masses, is a soft ionization method 3

4 allowing desorption of peptides and proteins from whole different cultured microorganisms. Ions are separated and detected according to their molecular mass and charge. For a given bacterial strain, this approach yields reproducible spectrum within minute, consisting of a series of peaks corresponding to m/z ratios of ions released from bacterial proteins during laser desorption. Recently, we engineered a strategy to identify bacteria belonging to the Micrococaceae family (2). The aim of this present work is to extend this strategy to non fermenting bacilli recovered in CF patients, thus opening the path towards rapid, accurate and cheap means of bacterial identification in routine laboratories. Our first step was to build a complete database for all species belonging to the group of non fermenting Gram negative bacilli recovered in human, including those isolated in CF patients. We then validated this database using the identification by MALDI-TOF MS of clinical strains recovered from CF patients. Material and methods Bacterial strains. The reference strains used to engineer the MALDI-TOF MS database belonged to 58 species of non fermenting Gram negative bacilli that can possibly be recovered from CF patients. Three databases were engineered, and the 87 strains used to complete these databases are described in tables 1, 2 and 3. The tested strains used to validate the databases were : (i) 512 clinical isolates of non fermenting Gram negative bacilli recovered from the sputum of CF children attending the pediatric department of the Necker-Enfants malades hospital (Paris, France) between 01/01/06 and 31/12/06 : 400 P. aeruginosa strains (101 patients), 54 Achromobacter xylosoxydans (12 patients), 32 S. maltophilia (12 patients), 9 R. 4

5 mannitolilytica (1 patient), 14 Bcc (2 patients ), 1 Burkholderia gladioli, 1 Bordetella hinzii and 1 I. limosus. These strains were identified by phenotypical tests or molecular method as previously described (10). Briefly, isolates displaying Green or Yellow-Green pigmentation, positive oxidase test, growth at 42 C, growth on cetrimide agar and susceptibility to colimycin (disk diffusion method) were identified as Pseudomonas aeruginosa. Isolates that did not express those criteria were identified using the API 20NE system (biomerieux, Marcy-l'Etoile, France). The results of the API 20NE tests were interpreted using the APILAB PLUS software package (biomerieux). When the API 20NE system did not identify a P. aeruginosa or a A. xylosoxydans or a S. maltophilia, the identification was further pursued by sequencing an internal fragment of the 16Sr RNA gene as previously described (10). Sixteen of the 400 P. aeruginosa strains and 26 of the 112 non P. aeruginosa strains required molecular methods for identification. B. cepacia strains were identified by 16SrDNA sequencing. Identification was confirmed by the Observatoire National des Cepacia using ARDRA (amplified rdna restriction analysis) (21,22). Species-specific reca PCR and/or Burkholderia cepacia complex-reca restriction analysis was used to circumvent the limitations of ARDRA within the Burkholderia cepacia complex (19). (ii) 47 clinical strains obtained from the Observatoire National des Cepacia (Toulouse, France): 38 Bcc strains (11 Burkholderia vietnamiensis, 10 Burkholderia multivorans, 8 B. cenocepacia, 5 Burkholderia stabilis, 1 Burkholderia dolosa, 1 Burkholderia pyrrocinia, 2 B. cepacia), 1 Burkholderia thailandensis, 5 R. mannitolilytica, 2 Ralstonia picketti, 1 Cupriavidus respiraculi. All bacterial strains used in the study were stored at -80 C in trypticase soy broth supplemented with 15% glycerol. MALDI-TOF-MS. The strains were grown on Mueller-Hinton agar and incubated for 24h at 37 C. Most of the isolates grew after 24h but some strains that didn t grow after 24h were 5

6 further incubated for 48h or 72h. An isolated colony was harvested in 20 µl of sterile water. One µl of this mixture was deposited on a target plate (Bruker Daltonics, Bremen, Germany) in two replicates, and allowed to dry at room temperature. One microliter of absolute ethanol was then added in each well, and the mixture allowed to dry. One µl of matrix solution DHB (2,5-dihydroxybenzoic acid, 50 mg/ml, 30% acetonitrile, 0.1% trifluoroacetic acid) was then added and allowed to co-cristallize with the sample. Samples were processed in the MALDI- TOF-MS spectrometer (Autoflex, Bruker Daltonics) with the flex control software (Bruker Daltonics). Positive ions were extracted with an accelerating voltage of 20 kv in linear mode. Each spectrum was the sum of the ions obtained from 200 laser shots performed in five different regions of the same well. The spectra have been analyzed in an m/z range of 2,000 to 20,000. The analysis was performed with the flex analysis software and calibrated with protein calibration standard I (Bruker Daltonics). The data obtained with the two replicates were added to minimize random effect. The presence and absence of peaks were considered as fingerprints for a particular isolate. The profiles were analyzed and compared using the newly developed software BGP database available on the website Numerical data obtained from the spectrometer acquisition software (peak value and relative intensity for each peak) are sent to the BGP software. This software identifies the number of common peaks between the spectra of the tested strain and each set of peaks specific of a reference strain contained in the database (i.e non fermenting Gram negative bacilli database). The software determines a percentage for each reference strain (100 x number of common peaks between the tested strain and the peaks specific of one reference strain/total number of peaks specific of one reference strain). The identification of the tested strain correspond to the species of the reference strain having the best match in the database. The greater the difference between the first and second match, the 6

7 better is the discrimination between species. A difference of at least 10% is required to obtain a good identification Results Engineering of the non fermenting Gram negative bacilli database. The database was engineered using a previously described strategy (2). A set of reference strains has been selected as belonging to clinically relevant nonfermenting Gram-negative bacilli, including those representative of the species routinely recovered from CF patients (table 1). Figure 1 shows the spectrum obtained with 6 species of non fermenting Gram negative bacilli. Ten isolates of each of these selected strains grown on Mueller Hinton 24H were analyzed by MALDI-TOF-MS as described in the material and methods section. For each strain, we retained only those peaks with a relative intensity above 0.1 that were constantly present in all 10 sets of data obtained for a given strain. The standard deviation for each conserved peak did not exceed 6 m/z value. Table 4 shows the m/z values of the selected peaks that have been retained for 8 reference strains. The set of peaks was specific of each selected strain. We next aimed at determining whether the above database could be used for the identification of non fermenting Gram negative bacilli, thus demonstrating that the set of peaks of each selected strain is, at least partially, conserved among isolates of the same species. The database was tested using the set of strains described in the material and methods section. For each isolate, all peaks with intensity 0.02 were retained and were compared with that of the specific peaks of each reference strain included in the database using the BGP-database software, taking into account a possible error of +/- 10 m/z value. Figure 2 shows the example of one P. aeruginosa strain. We next analyzed for all tested strains the percentage of common 7

8 peaks obtained with each of the reference strains. Only the first and second best matches were retained (Table 5). We considered a difference 10% as the minimum required to give a correct identification. A correct identification was obtained for all strains except those belonging to the Bcc and the Ralstonia genus (Table 5). The latter were identified as belonging to the Bcc or the Ralstonia genus but the species identification was not correct. Engineering of specific databases. In order to improve the identification of strains belonging to the Bcc and the Ralstonia genus, we engineered Bcc and Ralstonia specific databases. We hypothesized that including several reference strains for one given genospecies would improve identification of the tested strains. The Bcc specific database encompassed the 9 Bcc reference strains used to engineer the non fermenting Gram negative bacilli database and additional 21 Bcc reference strains provided by the LMG bacterial collection (table2), so that each species was represented by several reference strains. For each of these reference strains, only peaks with a relative intensity above 0.1 that were constantly present in all 10 sets of data were retained. However, in order to increase the ability of this database to discriminate between the species, only those peaks that were constantly conserved in all reference strains of the same species were retained in the database. This Bcc specific database was then tested using all clinical strains belonging to the Bcc. This approach improved the differentiation between Bcc species: only one B. dolosa strain was falsely identified as B. multivorans. Six B. cenopacia strains could not be differentiated from B. cepacia. However, we noticed that a peak (m/z 7546) was constantly present in B. cepacia strains and constantly absent in B. cenopacia. On the basis of this peak, a correct identification was obtained for all B. cenopacia strains. 8

9 The same strategy as that used for Bcc database was thus applied to engineer a specific Ralstonia database, using supplementary reference strains provided by the LMG bacterial collection (table 3). This new database allowed the correct identification of all strains. Discussion The bacterial species responsible for infection in cystic fibrosis patients do not display the same pathogenicity and thus do not require the same clinical management, thus pointing out the need for a correct and rapid identification of the bacteria isolated. This study demonstrates that the strategy described above is suitable for accurate species identification of non fermenting Gram negative bacilli isolated in cystic fibrosis. Indeed, all CF P. aeruginosa strains were accurately identified. Prior to this study, we have tested 12 strains belonging to 4 species (P. aeruginosa, A. xylosoxidans, B. cenocepacia, S. maltophilia) using 3 different media (MH, chocolate agar and blood agar) and 2 times of culture (18 h and 24h). Identifications were the same regardless of the media used to grow the bacteria (data not shown). Furthermore, the spectra were similar after 18 or 24 H of growth. In addition, identification was not affected if the strain grew only after 48H of incubation. The mucoid character of the strains recovered from patients with chronic colonization is frequently an obstacle to the accurate identification using biochemical tests. Nevertheless, none of the mucoid strains in our study was misidentified. Only one mucoïd strain had a low percentage of matches with the P. aeruginosa reference strain (38%), but the MALDI-TOF-MS identification was correct as the profile gave 22% of matches with the second choice. All S. maltophilia and A. xylosoxydans were correctly identified. It should be pointed out that frequent misidentification occurs between these species and Bcc strains (20). Infection by Bcc species in CF patients has been shown to increase morbidity and premature 9

10 mortality (7,18), and may represent a contraindication to lung transplantation (12). Furthermore, the risk associated with dissemination of B. cenocepacia strains requires specific means in order to prevent the bacterial spread (11). The rapid identification of B. cenocepacia using the Bcc database allows to immediately implement a specific clinical management before the results of molecular identification. Despite the good results achieved for Bcc strains, the wrong identification obtained for one strain points out the need of a larger set of clinical strains to improve the identification of Bcc species. R. mannitolilytica, I. limosus, B. hinzii and B. gladioli were correctly identified by MALDI- TOF-MS. Of note, the strain identified as B. gladioli matched also with the reference strain of B. cocovenenans (data not shown), a bacterial species that has been shown to be a junior synonym of B. gladioli (3). Conventional identification is not reliable for this species as well as for other species, such as R. mannitolilytica or I. limosus isolated during our study. API 20NE gave a very good identification of Inquilinus limosus as Sphingomonas paucimobilis, as previously described, which can thus lead to an underestimation of this emerging pathogen (25). The bacteria that are rarely isolated in CF patients can then be accurately identified using our database, as we make sure that the database is as complete as possible using many reference strains, even strains belonging to rarely isolated species CF patients. In our experience, all strains included in the database would have allowed identification of 100% of the non fermenting Gram negative bacilli isolated in the CF patients in our hospital. Emerging new bacterial species will give a spectrum that does not match with any of the reference spectra contained in the database. However, tested strains for which identification is not obtained by MALDI-TOF can be identified using a molecular biology approach, thus improving rapidly the database with new species. Such a rapid improvement is usually not 10

11 achievable using biochemical kits. Moreover, when 2 bacterial cultures are mixed, the global spectrum results is the sum of the 2 spectra, with specific peaks of both. Actually, MALDI-TOF MS bacterial identification is a phenotypic method but analysis of the origin of the ions detected in the spectra show that the majority of the peaks correspond to ribosomal proteins (8), which are proteins that represent a great proportion of the whole bacterial proteome and that are constantly expressed and conserved in bacteria. Metabolic characters lack specificity since the result can either be positive or negative and many biochemical tests correspond to universal metabolic pathways that can be common to several bacteria. A laboratory technician, without spectrometric background, can easily use this method. After depositing the sample and matrix as described in the material and methods section, the spectrometer can be programmed so that the laser impact automatically over the entire surface of the matrix, and there is no human intervention on the software for the species identification. The calibration with the standard, and the treatment of data (selection of peaks with relative intensity 0.02 and calculation of peaks ratios between database and tested strains by the BGP software) are automatically processed. Therefore, this is a very basic software which does not require any expertise nor subjective process and can be easily performed even with a non experimented operator. The time required to run 50 samples is about 1 hour Altogether, these data show that bacterial identification by MALDI-TOF-MS may improve the clinical management of CF patients. As this is a very low-cost technique (about 0.9 euro /50 samples), it is likely that considering the speed with which reliable identification can be obtained, this technique ultimately replaces the actual routine phenotypic assays

12 Acknowledgment The authors thank Gilles Quesnes and Eric Frapy for their beneficial technical assistance Potential conflicts of interest. All authors: no conflicts. 12

13 References 1. Brisse, S., S. Stefani, J. Verhoef, A. Van Belkum, P. Vandamme, and W. Goessens Comparative evaluation of the BD Phoenix and VITEK 2 automated instruments for identification of isolates of the Burkholderia cepacia complex. J. Clin. Microbiol. 40: Carbonnelle, E., J. L. Beretti, S. Cottyn, G. Quesne, P. Berche, X. Nassif, and A. Ferroni Rapid identification of Staphylococci isolated in clinical microbiology laboratories by matrix-assisted laser desorption ionization-time of flight mass spectrometry. J. Clin. Microbiol. 45: Coenye, T., B. Holmes, K. Kersters, J. R. Govan, and P. Vandamme Burkholderia cocovenenans (van Damme et al. 1960) Gillis et al and Burkholderia vandii Urakami et al are junior synonyms of Burkholderia gladioli (Severini 1913) Yabuuchi et al and Burkholderia plantarii (Azegami et al. 1987) Urakami et al. 1994, respectively. Int. J. Syst. Bacteriol. 49: Coenye, T., L. Liu, P. Vandamme, and J. J. LiPuma Identification of Pandoraea species by 16S ribosomal DNA-based PCR assays. J. Clin. Microbiol. 39: Coenye, T., J. Goris, T. Spilker, P. Vandamme, and J. J. LiPuma Characterization of unusual bacteria isolated from respiratory secretions of cystic fibrosis patients and description of Inquilinus limosus gen. nov., sp. nov. J. Clin. Microbiol. 40: Coenye, T., P. Vandamme, and J. J. LiPuma Infection by Ralstonia species in cystic fibrosis patients: identification of R. pickettii and R. mannitolilytica by polymerase chain reaction. Emerg. Infect. Dis. 8: Corey, M., and V. Farewell Determinants of mortality from cystic fibrosis in Canada, Am. J. Epidemiol. 143: Demirev, P. A., Y. P. Ho, V. Ryzhov, and C. Fenselau Microorganism identification by mass spectrometry and protein database searches. Anal. Chem. 71: Fenselau, C., and P. A. Demirev Characterization of intact microorganisms by MALDI mass spectrometry. Mass. Spectrom. Rev. 20:

14 Ferroni, A., I. Sermet-Gaudelus, E. Abachin, G. Quesne, G. Lenoir, P. Berche, and J. L. Gaillard Use of 16S rrna gene sequencing for identification of nonfermenting gram-negative bacilli recovered from patients attending a single cystic fibrosis center. J. Clin. Microbiol. 40: Govan, J. R., P. H. Brown, J. Maddison, C. J. Doherty, J. W. Nelson, M. Dodd, A. P. Greening, and A. K. Webb Evidence for transmission of Pseudomonas cepacia by social contact in cystic fibrosis. Lancet 342: Hadjiliadis, D Special considerations for patients with cystic fibrosis undergoing lung transplantation. Chest 131: Holland, R. D., C. R. Duffy, F. Rafii, J. B. Sutherland, T. M. Heinze, C. L. Holder, K. J. Voorhees, and J. O. Lay, Jr Identification of bacterial proteins observed in MALDI TOF mass spectra from whole cells. Anal. Chem. 71: Holland, R. D., J. G. Wilkes, F. Rafii, J. B. Sutherland, C. C. Persons, K. J. Voorhees, and J. O. Lay, Jr Rapid identification of intact whole bacteria based on spectral patterns using matrix-assisted laser desorption/ionization with time-of-flight mass spectrometry. Rapid Commun Mass Spectrom 10: Krishnamurthy, T., and P. L. Ross Rapid identification of bacteria by direct matrix-assisted laser desorption/ionization mass spectrometric analysis of whole cells. Rapid Commun Mass Spectrom 10: Lambiase, A., V. Raia, M. Del Pezzo, A. Sepe, V. Carnovale, and F. Rossano Microbiology of airway disease in a cohort of patients with cystic fibrosis. BMC Infect. Dis. 6: Lay, J. O., Jr MALDI-TOF mass spectrometry of bacteria. Mass Spectrom Rev 20: Lipuma, J. J Update on the Burkholderia cepacia complex. Curr. Opin. Pulm. Med. 11: Mahenthiralingam, E., J. Bischof, S. Byrne, C. Radomski, J. Davies, Y. Av-Gay, and P. Vandamme DNA-Based diagnostic approaches for identification of Burkholderia cepacia complex, Burkholderia vietnamiensis, Burkholderia multivorans, Burkholderia stabilis, and Burkholderia cepacia genomovars I and III. J Clin Microbiol 38:

15 McMenamin, J. D., T. M. Zaccone, T. Coenye, P. Vandamme, and J. J. LiPuma Misidentification of Burkholderia cepacia in US cystic fibrosis treatment centers: an analysis of 1,051 recent sputum isolates. Chest 117: Segonds, C., T. Heulin, N. Marty, and G. Chabanon Differentiation of Burkholderia species by PCR-restriction fragment length polymorphism analysis of the 16S rrna gene and application to cystic fibrosis isolates. J Clin Microbiol 37: Segonds, C., S. Paute, and G. Chabanon Use of amplified ribosomal DNA restriction analysis for identification of Ralstonia and Pandoraea species: interest in determination of the respiratory bacterial flora in patients with cystic fibrosis. J. Clin. Microbiol. 41: Shelly, D. B., T. Spilker, E. J. Gracely, T. Coenye, P. Vandamme, and J. J. LiPuma Utility of commercial systems for identification of Burkholderia cepacia complex from cystic fibrosis sputum culture. J. Clin. Microbiol. 38: van Pelt, C., C. M. Verduin, W. H. Goessens, M. C. Vos, B. Tummler, C. Segonds, F. Reubsaet, H. Verbrugh, and A. van Belkum Identification of Burkholderia spp. in the clinical microbiology laboratory: comparison of conventional and molecular methods. J. Clin. Microbiol. 37: Wellinghausen, N., A. Essig, and O. Sommerburg Inquilinus limosus in patients with cystic fibrosis, Germany. Emerg. Infect. Dis. 11: Wellinghausen, N., J. Kothe, B. Wirths, A. Sigge, and S. Poppert Superiority of molecular techniques for identification of gram-negative, oxidase-positive rods, including morphologically nontypical Pseudomonas aeruginosa, from patients with cystic fibrosis. J. Clin. Microbiol. 43: Wellinghausen, N., B. Wirths, and S. Poppert Fluorescence in situ hybridization for rapid identification of Achromobacter xylosoxydans and Alcaligenes faecalis recovered from cystic fibrosis patients. J. Clin. Microbiol. 44:

16 355 Figures legends : 356 Figure 1 : MALDI-TOF-MS spectrum of 6 reference strains Figure 2 : Identification of a P. aeruginosa strain using the BGP software. The best match was obtained with the P.aeruginosa reference strain. The second best match was with F. oryzihabitans 16

17 Table 1 : reference strains used to establish the MALDI-TOF-MS non fermenting Gram negative bacilli database Species Reference strain Species Reference strain Pseudomonas aeruginosa CIP Ralstonia mannitolilytica CIP T Pseudomonas fluorescens CIP T Ralstonia pickettii CIP T Pseudomonas mosselii CIP Cupriavidus gilardii CIP T Pseudomonas putida CIP T Cupriavidus pauculus CIP T Pseudomonas stutzeri CIP T Cupriavidus respiraculi LMG Pseudomonas mendocina CIP T Bordetella avium CIP T Pseudomonas alcaligenes CIP T Bordetella bronchiseptica CIP T Pseudomonas pseudoalcaligenes CIP T Bordetella hinzii CIP T Pseudomonas oryzihabitans CIP T Alcaligenes faecalis subsp. faecalis CIP T Pseudomonas luteola CIP T Aeromonas sobria CIP T Stenotrophomonas maltophilia CIP T Aeromonas hydrophila subsp. hydrophila CIP T Achromobacter xylosoxydans subsp. xylosoxydans CIP T Aeromonas veronii CIP T Achromobacter xylosoxydans subsp. denitrificans CIP T Aeromonas caviae CIP T Achromobacter piechaudii CIP Delftia acidovorans CIP T Burkholderia cepacia complex Shewanella putrefaciens CIP T. Burkholderia cepacia CIP T Plesiomonas shigelloides CIP 63.5 T. Burkholderia multivorans CIP T Chryseobacterium indologenes CIP T. Burkholderia cenocepacia CIP T Elizabethkingia meningoseptica CIP T. Burkholderia stabilis CIP T Sphingobacterium multivorum CIP T. Burkholderia vietnamiensis CIP T Sphingobacterium spiritivorum CIP T. Burkholderia dolosa CIP Brevundimonas diminuta CIP T. Burkholderia ambifaria CIP T Brevundimonas vesicularis CIP T. Burkholderia anthina CIP T Sphingomonas paucimobilis CIP T. Burkholderia pyrrocinia CIP T Inquilinus limosus CIP T Burkholderia gladioli CIP T Pandoraea apista CIP T Burkholderia cocovenenans ATCC Pandoraea norimbergensis LMG Burkholderia glumae NCPPB 2391 Pandoraea promenusa LMG Burkholderia plantarii ATCC Pandoraea pulmonicola LMG Burkholderia thailandenis LMG Pandoraea sputorum LMG Burkholderia glathei CIP T Burkholderia andropogonis ATCC CIP: Collection de l Institut Pasteur (Paris, France) NCPPB : National Collection of Plant Pathogenic Bacteria (York, U.K). LMG : Laboratorium voor Microbiologie Bacteriënverzameling (Ghent, Belgium) T: Type strain of the species.

18 Table 2 : reference strains used to establish the MALDI-TOF-MS Bcc database Species Reference strain Burkholderia cepacia CIP T Burkholderia cepacia LMG 6889 Burkholderia cepacia LMG 2161 Burkholderia multivorans CIP T Burkholderia multivorans LMG Burkholderia multivorans LMG Burkholderia cenocepacia CIP T Burkholderia cenocepacia LMG Burkholderia cenocepacia LMG Burkholderia cenocepacia LMG Burkholderia cenocepacia LMG Burkholderia cenocepacia LMG Burkholderia cenocepacia LMG Burkholderia cenocepacia LMG Burkholderia stabilis CIP T Burkholderia stabilis LMG 6997 Burkholderia stabilis LMG 7000 Burkholderia vietnamiensis CIP T Burkholderia vietnamiensis LMG 6998 Burkholderia vietnamiensis LMG 6999 Burkholderia dolosa CIP Burkholderia dolosa LMG Burkholderia ambifaria CIP T Burkholderia ambifaria LMG Burkholderia ambifaria LMG Burkholderia anthina CIP T Burkholderia anthina LMG Burkholderia pyrrocinia CIP T Burkholderia pyrrocinia LMG Burkholderia pyrrocinia LMG CIP : Collection de l Institut Pasteur (Paris, France) LMG : Laboratorium voor Microbiologie Bacteriënverzameling (Ghent, Belgium) T: Type strain of the species.

19 Table 3 : reference strains used to establish the MALDI-TOF-MS Ralstonia database Species Reference strain Cupriavidus gilardii CIP T Cupriavidus gilardii LMG 3399 Cupriavidus gilardii LMG 3400 Cupriavidus pauculus CIP T Cupriavidus pauculus LMG 3245 Cupriavidus pauculus LMG 3317 Cupriavidus respiraculi LMG Cupriavidus respiraculi LMG Ralstonia mannitolilytica CIP T Ralstonia mannitolilytica LMG Ralstonia mannitolilytica LMG Ralstonia pickettii CIP T Ralstonia pickettii LMG 5942 Ralstonia pickettii LMG 7001 CIP: Collection de l Institut Pasteur (Paris, France) LMG : Laboratorium voor Microbiologie Bacteriënverzameling (Ghent, Belgium) T: Type strain of the species.

20 Table 4 : m/z of the selected peaks of 8 species of non fermenting Gram negative bacilli. Achromobacter xyloxosidans subsp xylosoxydans CIP 71.32T / / / / / / / / / / / / / / / / / / / / / /-2 Pseudomonas aeruginosa CIP / / / / / / / / / / / / /-5 Pseudomonas fluorescens CIP 69.13T / / / / / / / / /-1 Pseudomonas putida CIP T / / / / / / / / / / / / /-3 Burkholderia cenocepacia CIP T / / / / / / / / / / / / / /-4 Burkholderia gladioli CIP T / / / / / / / / / / / / / / / / / / /-4 Stenotrophomonas maltophilia CIP 60.77T / / / / / / / /-2 Ralstonia mannitolilytica CIP T / / / / / / / / / /-4

21 Table 5 : identification of non fermenting Gram negative bacilli by MALDI-TOF-MS average % of common Number of average % of common number of strains for Number of strains for Number peaks between the tested strains for which peaks between the tested which the difference of which the of species strain and the best match the best match is strain and the second best common peaks between identification is tested of the general database the correct match of the general the best and second best correct using the Bcc strains (extreme) identification database (extreme) match is less than 10% or Ralstonia database P. aeruginosa (ext : ) (ext : 18-83) 1 4 A. xylosoxydans 2 4 (2 nd choice : A (ext : ) (ext : 50-88) piechaudii) S. maltophilia (ext : ) (ext : 29-67) 0 B. cenocepacia (ext : ) (ext : 56-86) 7 22 B. vietnamiensis (ext : ) (ext : 67-81) 4 11 B. multivorans (ext : ) (ext : 74-87) 0 9 B. stabilis 5 72 (ext : 67-80) (ext : 56-69) 2 5 B. cepacia 2 88 (ext : 81-94) 2 62 (ext : 59-64) 0 2 B. pyrrocina B. dolosa Ralstonia (ext : 70-80) (ext : 56-70) 4 mannitolilytica Ralstonia picketti 2 73 (ext : 70-80) 2 58 (ext : 42-75) 0 2 Cupriavidus respiraculi multivorans 5 The misidentified strain was identified as B. B. gladioli B. hinzii I. limosus B. thailandensis The 4 misidentified strains were identified as 3 B. cepacia and 1 B. cocovenenans 2 The misidentified strain was identified as B. cepacia 3 The misidentified strain was identified as R. picketti 4 These strains were retested, and the second set of data provided the same identification as the initial one.

22 P. aeruginosa CIP T S. maltophilia CIP T A. xylosoxydans subsp.xylosoxydans CIP T B. cenocepacia CIP T R. mannitolilytica CIP T I. limosus CIP T

23 Profile Name : P. aeruginosa Match : 13/13 1 Isolate database N m/z values P. aeruginosa m/z values Profile Name : F. oryzihabitans Match: 7/12 2 Isolate database N m/z values F. orzyhabitans m/z values : the 13 peaks obtained for tested isolate matched all 13 peaks of P. aeruginosa database 2 : seven peaks obtained for tested isolate matched 7/12 peaks of F. orizyhabitans database