Anal. Chem. 2005, 77, 1488-1495
Identification of Microbial Mixtures by Capillary Electrophoresis/Selective Tandem Mass Spectrometry Anren Hu,†,‡ Pei-Jen Tsai,‡ and Yen-Peng Ho*,†
Department of Chemistry, National Dong Hwa University, Hualien, Taiwan, and Department of Laboratory Medicine & Biotechnology, Tzu Chi University, Hualien, Taiwan
In this paper, we propose a new strategy for identifying specific bacteria in bacterial mixtures by using CEselective MS/MS of peptide marker ions associated with the bacteria of interest. We searched the CE-MS/MS spectra acquired from the proteolytic digests of pure bacterial cell extracts against protein databases. The identified peptides that match the protein associated with the corresponding species were selected as marker ions for bacterial identification. Specific peptide marker ions were obtained for each of the following three pathogens: Pseudomonas aeruginasa, Staphylococcus aureus, and Staphylococcus epidermidis. To identify a bacterial species in a sample, we performed CE-MS/MS analysis of the selected marker ions in the proteolytic digest of the cell extract and then performed protein database searches. The selected peptides that we identified correctly from Xcorr values ranking at the top of the search results allowed us to identify the corresponding bacterial species present in the sample. We have applied this method successfully to the identification of various mixtures of the three pathogens. Even minor bacterial species present at a concentration of 1% can be identified with great confidence. This method for CE-MS/MS analysis of bacteriaspecific marker peptides provides excellent selectivity and high accuracy when identifying bacterial species in complex systems. In addition, we have used this approach to identify P. aeruginasa in a saliva sample spiked with E. coli and P. aeruginasa. The detection and identification of microorganisms is critical when diagnosing infectious diseases and detecting biohazards in the environment.1 For example, the use of protein biomarkers to identify microorganisms has been very successful. Both matrixassisted laser desorption/ionization (MALDI) and electrospray ionization (ESI) mass spectrometry (MS) can be used to analyze protein and, therefore, to identify bacteria.2-4 Currently, MALDI* Corresponding author. Tel: 886-3-8633591. E-mail:
[email protected]. † National Dong Hwa University. ‡ Tzu Chi University. (1) Fenselau, C., Ed. Mass spectrometry for the characterization of microorganisms; American Chemical Society: Washington, DC, 1994. (2) van Baar, B. L. M. FEMS Microbiol. Rev. 2000, 24, 193-219. (3) Fenselau, C.; Demirev, P. A. Mass Spectrom. Rev. 2001, 20, 157-171.
1488 Analytical Chemistry, Vol. 77, No. 5, March 1, 2005
MS is the technique used most often to obtain mass spectra of proteins for bacterial identification.3,4 Protein-based approaches include profiling bacterial proteins and mass matching the proteins/peptides obtained from bacterial samples to those present in proteome databases. The successful identification of bacteria by protein profiling relies on a high level of mass spectral reproducibility or a complex search algorithm to a created fingerprint library.5 To overcome the problem of poor spectral reproducibility, an approach has been described for microorganism identification by a combination of mass spectrometry and protein database searches6 in which protein masses in mass spectra are compared with those in the proteome databases. Probable organism sources were assigned to each mass, and the best match (i.e., the one having the most matched proteins) should, in theory, lead to the identification of the correct microorganism. Spectral reproducibility is not critical when this approach is used as long as the observed masses match protein masses that are available in the database. False identification can occur, however, because of overlap of the masses of some proteins in different microorganisms or because of poor mass accuracy. If the reliability of database search results is in doubt, a significance level testing7,8 may be performed. Another way to avoid limited mass accuracy in protein mass measurements is to apply tandem mass spectrometry to obtain a partial protein sequence. Thus, proteins extracted from viral or bacterial samples can be digested by trypsin, and the sequence information of the peptides obtained by tandem mass spectrometry can be used to identify the microorganism’s source by performing proteome database searches. Yao et al. have demonstrated a method for virus identification based on the MS/MS analysis of tryptic peptides and the construction of databases of virus-specific tryptic peptide masses.9 Warscheid et al. have proposed a strategy to identify Bacillus (4) Lay, J. O., Jr. Mass Spectrom. Rev. 2001, 20, 172-194. (5) Wahl, K. L.; Wunschel, S. C.; Jarman, K. H.; Valentine, N. B.; Petersen, C. E.; Kingsley, M. T.; Zartolas, K. A.; Saenz, A. J. Anal. Chem. 2002, 74, 6191-6199. (6) Demirev, P. A.; Ho, Y. P.; Ryzhov, V.; Fenselau, C. Anal. Chem. 1999, 71, 2732-2738. (7) Pineda, F. J.; Lin, J. S.; Fenselau, C.; Demirev, P. A. Anal. Chem. 2000, 72, 3739-3744. (8) Demirev, P. A.; Lin, J. S.; Pineda, F. J.; Fenselau, C. Anal. Chem. 2001, 73, 4566-4573. (9) Yao, Z. P.; Demirev, P. A.; Fenselau, C. Anal. Chem. 2002, 74, 2529-2534. 10.1021/ac0484427 CCC: $30.25
© 2005 American Chemical Society Published on Web 01/28/2005
spores through a combination of peptide sequencing of tryptic digests of specific spore proteins and database searching.10,11 English et al. have employed a constructed database of tryptic peptides generated from specific biomarkers to analyze Bacillus spores.12 Most of these database searching approaches have employed MALDI-MS to analyze the samples. Direct identification of bacteria through the use of MALDIMS provides a number of advantages, such as speed, high sensitivity, simplified mass spectra of dominantly singly charge ions, and tolerance to contaminants. When a large set of digested peptides obtained from a complex microorganism is analyzed, however, MALDI-MS provides spectra that are too complicated for interpretation and a low sensitivity for peptide sequencing. Moreover, MALDI-MS is relatively difficult to couple on-line with sample pretreatment and separation techniques and is not readily amenable to automation. ESI-MS, on the other hand, allows on-line detection to be combined with sample purification, concentration, and separation techniques such as microdialysis, solid-phase extraction, liquid chromatography, and capillary electrophoresis.13-17 Thus, ESI-MS is a powerful tool for the analysis of complex systems. Although ESI-MS has been used less frequently for microbial identification, Xiang et al. have characterized microorganisms by performing global ESI-MS/MS analyses of cell lysates.18 Goodacre et al. have reported the direct ESI-MS analysis of whole bacterial cells without prior separation.19,20 Several laboratories, including those of Krishnamurthy, Lubman, and Williams, have used LC-ESI-MS to obtain reproducible protein profiles from bacterial samples.21-24 Ho et al. have demonstrated the application of protein profiling by LC-ESI-MS with database searches to effect bacterial identification.25 Recently, Dworzanski et al. proposed a method for bacterial identification using LC-MS/MS data that are searched against a bacterial proteome database generated by translating all putative open reading frames into protein sequences.26 (10) Warscheid, B.; Fenselau, C. Anal. Chem. 2003, 75, 5618-5627. (11) Warscheid, B.; Jackson, K.; Sutton, C.; Fenselau, C. Anal. Chem. 2003, 75, 5608-5617. (12) English, R. D.; Warscheid, B.; Fenselau, C.; Cotter, R. J. Anal. Chem. 2003, 75, 6886-6893. (13) Liu, C. L.; Hofstadler, S. A.; Bresson, J. A.; Udseth, H. R.; Tsukuda, T.; Smith, R. D.; Snyder, A. P. Anal. Chem. 1998, 70, 1797-1801. (14) Estrela, R. D. C. E.; Salvadori, M. C.; Suarez-Kurtz, G. Rapid Commun. Mass Spectrom. 2004, 18, 1147-1155. (15) Pruvost, A.; Ragueneau, I.; Ferry, A.; Jaillon, P.; Grognet, J. M.; Benech, H. J. Mass Spectrom. 2000, 35, 625-633. (16) Zamfir, A.; Seidler, D. G.; Schonherr, E.; Kresse, H.; Peter-Katalinic, J. Electrophoresis 2004, 25, 2010-2016. (17) Janini, G. M.; Conrads, T. P.; Wilkens, K. L.; Issaq, H. J.; Veenstra, T. D. Anal. Chem. 2003, 75, 1615-1619. (18) Xiang, F.; Anderson, G. A.; Veenstra, T. D.; Lipton, M. S.; Smith, R. D. Anal. Chem. 2000, 72, 2475-2481. (19) Vaidyanathan, S.; Rowland, J. J.; Kell, D. B.; Goodacre, R. Anal. Chem. 2001, 73, 4134-4144. (20) Vaidyanathan, S.; Kell, D. B.; Goodacre, R. J. Am. Soc. Mass Spectrom. 2002, 13, 118-128. (21) Krishnamurthy, T.; Davis, M. T.; Stahl, D. C.; Lee, T. D. Rapid Commun. Mass Spectrom. 1999, 13, 39-49. (22) Williams, T. L.; Leopold, P.; Musser, S. Anal. Chem. 2002, 74, 5807-5813. (23) Zheng, S.; Schneider, K. A.; Barder, T. J.; Lubman, D. M. BioTechniques 2003, 35, 1202-1212. (24) Chong, B. E.; Kim, J.; Lubman, D. M.; Tiedje, J. M.; Kathariou, S. J. Chromatogr., B 2000, 748, 167-177. (25) Ho, Y. P.; Hsu, P. H. J. Chromatogr., A 2002, 976, 103-111. (26) Dworzanski, J. P.; Snyder, A. P.; Chen, R.; Zhang, H. Y.; Wishart, D.; Li, L. Anal. Chem. 2004, 76, 2355-2366.
In the paper, we describe a method that uses CE-MS/MS of selected peptide ions to identify specific bacteria from bacterial mixtures. This approach is similar to that of selected reaction monitoring (SRM)27-31 with the exception that the allowable maximum mass range covering all product ions is monitored. The characterization of microbial particles by CE alone has been reported.32-34 A limited number of papers have reported efforts to identify microorganisms using CE-MS.35 CE allows the rapid and efficient separation of biological molecules with minimum sample preparation and sample consumption.36-38 Selected MS/MS monitoring coupled with separation methods such as CE provides a highly selective and sensitive analytical method for characterizing the analytes present in a complex mixture.39 In this study, we subjected the proteolytic digests of cell extracts obtained from pure pathogens of interest to CE-MS/MS analysis and then performed database searches. We selected the identified peptides that match the protein associated with the corresponding bacterial species as marker ions for bacterial identification of the mixtures. To identify a bacterial species in a sample, we performed CEMS analysis on the proteolytic digest of the cell extract and monitored only the selected marker peptide masses using MS/ MS. Thus, the corresponding bacterial species could be identified if the selected peptides were identified correctly from the database searches. It would be rare for two isobaric peptides to coelute during CE separation of the complex proteolytic digests. Therefore, the peptide chosen for MS/MS analysis is less likely to be a mixture, which reduces the possibility of database search failures that are due to complicated MS/MS data. Because the same proteins may be shared by different bacteria or the same peptide may be a portion of proteins having different sequences, database searches against the MS/MS data of the corresponding peptides can lead to multiple or false identifications of microorganisms. Preliminary analysis of pure pathogens through database searching can determine whether the peptides lead to multiple or false microbial identification; these peptides can then be excluded for selective MS/MS. We have also applied this powerful method of microbial identification to identify pathogens present in a saliva sample spiked with bacterial mixtures. EXPERIMENTAL SECTION Materials. Methyl alcohol was obtained from J. T. Baker (Phillipsburg, NJ). Acetic acid, ammonium acetate, and ammonia (27) King, R. C.; Gundersdorf, R.; Fernandez-Metzler, C. L. Rapid Commun. Mass Spectrom. 2003, 17, 2413-2422. (28) Barbarin, N.; Mawhinney, D. B.; Black, R.; Henion, J. J. Chromatogr., B 2003, 783, 73-83. (29) Zweigenbaum, J.; Henion, J. Anal. Chem. 2000, 72, 2446-2454. (30) Onorato, J. M.; Henion, J. D.; Lefebvre, P. M.; Kiplinger, J. P. Anal. Chem. 2001, 73, 119-125. (31) Zhang, H. W.; Henion, J. J. Chromatogr., B 2001, 757, 151-159. (32) Kourkine, I. V.; Ristic-Petrovic, M.; Davis, E.; Ruffolo, C. G.; Kapsalis, A.; Barron, A. E. Electrophoresis 2003, 24, 655-661. (33) Rodriguez, M. A.; Armstrong, D. W. J. Chromatogr., B 2004, 800, 7-25. (34) Desai, M. J.; Armstrong, D. W. Microbiol. Mol. Biol. Rev. 2003, 67, 38-51, table of contents. (35) Chong, B. E.; Kim, J.; Lubman, D. M.; Tiedje, J. M.; Kathariou, S. J. Chromatogr., B 2000, 748, 167-177. (36) Hernandez-Borges, J.; Neususs, C.; Cifuentes, A.; Pelzing, M. Electrophoresis 2004, 25, 2257-2281. (37) Janini, G. M.; Zhou, M.; Yu, L. R.; Blonder, J.; Gignac, M.; Conrads, T. P.; Issaq, H. J.; Veenstra, T. D. Anal. Chem. 2003, 75, 5984-5993. (38) Figeys, D.; Aebersold, R. Electrophoresis 1998, 19, 885-892. (39) Heinig, K.; Henion, J. J. Chromatogr., B 1999, 735, 171-188.
Analytical Chemistry, Vol. 77, No. 5, March 1, 2005
1489
solution were purchased from Reidel-de Hae¨n (Seelze, Germany). Ammonium bicarbonate and LB broth were purchased from the Sigma Chemical Co. (St. Louis, MO). Porcine trypsin (sequence grade) was obtained from Promega (Madison, WI). Water was purified using a Milli-Q system (Millipore, Bedford, MA). Bacterial Sample Preparations. We investigated three pathogens: Pseudomonas aeruginasa (Gram negative), Staphylococcus aureus (Gram positive), and Staphylococcus epidermidis (Gram positive). Bacterial cells were grown in LB broth at 37 °C for 24 h. The cells were harvested, washed three times with water, boiled in water for 10 min, lyophilized to dryness, and then stored at -20 °C. Water was used for protein extraction from the cells. The lyophilized bacteria were suspended in water at a concentration of 5 mg/mL. Most of the bacterial mixtures were prepared by mixing different dry weights of bacteria, ranging from 0.5 to 5 mg, in 1 mL of water. The 1:50, 1:100, and 1:10:100 mixtures were prepared in water (10 mL) by mixing 50 mg of the major component with 0.5-5 mg of the minor component. The cell suspension was centrifuged, and 1 mL of the supernatant was collected, filtered through a 4.5-µm PVDF membrane, and dried on a SpeedVac vacuum concentrator. For proteolytic digestion, the protein extract was redissolved in ammonium bicarbonate buffer (55 mM, pH 8.5; 5 µL) and treated with trypsin solution (0.1 mM in 55 mM ammonium bicarbonate, pH 8.5; 0.5 µL) for 6 h. The peptide digest was dried on a SpeedVac vacuum concentrator and redissolved in water (1 µL). To prepare the saliva samples, adult male saliva (5 mL) was spiked with a single colony of Escherichia coli and P. aeruginasa species. The sample was cultivated at 37 °C for 12 h and amplified in LB broth at 37 °C for 20 h. The bacterial cells were harvested as described in the previous paragraph. CE System. The experiments were performed on a homebuilt CE system equipped with a model PS/EH30R03.0 power supply (Glassman High Voltage, Inc., Whitehouse Station, NJ). A poly(vinyl alcohol)-coated capillary (100 cm × 75 µm i.d. × 380 µm o.d.) was obtained from Agilent Technologies (Palo Alto, CA). Pressure injection at 100 mbar was applied for 5 s. The temperature was maintained at 25 °C, and the applied potential was 20 kV. The running buffer was a mixture of 10 mM ammonium acetate (70%) and 10 mM acetic acid (30%) at pH 4.8. The capillary was conditioned with water and running buffer for 30 min each before the first run and for 5 min between runs. MS Analysis. An ion trap mass spectrometer (LCQ Duo, Finnigan, San Jose, CA) equipped with an electrospray ionization source was used for all MS analyses. The coupling to the CE apparatus was effected through a Finnigan coaxial sheath-flow interface. The spray voltage was maintained at 4 kV. The capillary voltage and temperature were maintained at 27 V and 200 °C, respectively. The sheath gas flow was set at 20 (data-dependent experiments) or 40 (selected MS/MS experiments) arbitrary units. The sheath liquid was 50% MeOH containing 0.5% acetic acid at a flow rate of 4 µL/min. The mass spectrometer was operated under the control of an Xcaliber program. Spectra were collected in the positive-ion mode. The autogain control was maintained at 5 × 108, the microscan count was 3, and the maximum injection time was 200 ms. When the spectra were acquired in a datadependent mode, a full MS scan was acquired from m/z 400 to 2000, followed by three MS/MS scans between m/z 400 and 2000 1490 Analytical Chemistry, Vol. 77, No. 5, March 1, 2005
Figure 1. Data-dependent total ion chromatogram obtained from the CE-MS/MS analysis of S. aureus.
of the most, second-most, and third-most intense ions of the full MS scan. The wideband excitation for the MS/MS scan was set at 42% of the normalized collision energy for 30 ms. In the selective MS/MS experiments, the spectra were acquired in the SRM mode while the mass range of the monitored product ion was set to the allowable maximum value to cover the dissociation fragments. The experiment differs from that of normal SRM, which only monitors the abundant and structurally unique transitions (from molecular ion to specific product ions). Database Searches. The MS/MS spectra were searched against the NCBInr database using SEQUEST (ThermoFinnigan Inc.). The protein database comprising all kingdoms was downloaded from the web site of the National Center for Biotechnology Information (NCBI; ftp://www.ncbi.nlm.nih.gov/blast/db). In the data-dependent experiments, the mass tolerances of the precursor and fragment ions were set to 2.5. Only those peptides that gave cross-correlation scores (Xcorr) of >1.8 for singly charged peptides, >2.2 for doubly charged peptides, and >3.3 for triply charged peptides [each with delta-correlation scores (DelCn) > 0.1] were considered as legitimate identifications. Each identified peptide leads to a matched protein and organism source associated with the protein. Because the nature of the sample is known, only the bacterial sources were assigned for each identified peptide. For the selective MS/MS experiments of bacterial mixtures, the mass tolerances of the precursor and fragment ions were set to 0.5. Peptides that gave cross-correlation scores (Xcorr) of >1.5 for singly charged peptides and >1.9 for doubly charged peptides [each with delta-correlation scores (DelCn) of >0.1] were considered as legitimate identifications. RESULTS AND DISCUSSION We have investigated the identification of bacteria through selective MS/MS analysis and database searches. Before the selective MS/MS analysis of bacteria could be performed, it was necessary to acquire the MS/MS spectra of proteolytic digests of cell extracts obtained from bacterial samples in a data-dependent mode. The peptides were separated by CE and subjected to MS/ MS analyses. Figure 1 displays the data-dependent total ion chromatogram for S. aureus. The ion chromatogram consists of many full scans of mass spectra of the separated peptides, with
Table 1. Database Search Results for the Top 10 Identified Peptides Generated from the CE-MS/MS Analysis of the Proteolytic Digest of the S. aureus Cell Extract Xcorr
peptide mass (charge)
retention time (min)
3.77
1892.9 (+2)
47.9
-.SLEEGQAVEFEVVEGDR.-
2.83
1844.8 (+2)
28.8
-.VYNDYSIEEHNGNYK.-
2.75
1697.8 (+2)
43.5
-.FIEDIDHVNPDEVR.-
17-30
2.69
809.4 (+1)
31.4
-.ATDFIDK.-
51-57
2.66
1601.8 (+1)
39.0
-.TDLINAVAEQADLTK.-
4-18
2.65
1076.5 (+1)
30.3
-.ETVGNVTDNK.-
6-25
2.56
1262.6 (+1)
45.6
-.SGEESEVLVADK.-
2.36
2918.4 (+2)
33.9
-.GFGFIEVEGENDVFVHFSAINQDGYK.-
2.35
1397.6 (+1)
35.5
-.INEYTGSNNEEK.-
2.33
977.5 (+1)
23.9
-.EFNPDIDK.-
2.26
1175.6 (+1)
34.2
-.NLENEGKEDK.-
sequence
position
protein
MW
29-45
cold-shock protein hypothetical protein hypothetical protein hypothetical protein DNA-binding protein II hypothetical protein hypothetical protein cold-shock protein hypothetical protein lmo1241
5030
156-170
92-103 3-28 29-40 242-249 26-35
22806
microbial source S. aureus S. intermedius S. aureus
6283
S. aureus
6996
S. aureus
9602
S. aureus
6996
S. aureus
21243
S. aureus
5030 5634 51957
hypothetical protein
6996
S. aureus S. intermedius S. aureus Listeria monocytogenes S. aureus
Table 2. Identities of the Selected Peptide Markers of Three Pathogens: S. aureus, P. aeruginasa, and S. epidermidis bacterial species S. aureus P. aeruginasa
S. epidermidis
isolated m/z (charge)
sequence
peptide position
1076.5 (+1) 1262.6 (+1) 1638.8 (+1) 1036.0 (+2)
-.ETVGNVTDNK.-.SGEESEVLVADK.-.IEDTDFAAETANLTK.-.TVIHTDNAPAAIGTYSQAIK.-
6-25 92-103 307-321 4-23
1040.5 (+2)
-.YGPVDGDPITSTEEIPFDK.-
1086-1104
950.0 (+2)
-.AHLVDLAQHNPEELNAK.-
37-53
every full scan followed by three MS/MS scans of the most-intense precursor ions. The serrated appearance of the ion chromatogram arose because the data acquisition alternated between MS and MS/MS modes. To resolve as many peptides as possible, we lowered the migration time by using a low sheath gas flow.40,41 All of the tandem mass spectra of the peptides were searched against the NCBInr protein database by using the SEQUEST application. Table 1 lists the top 10 identified peptides that gave cross-correlation (Xcorr) scores and delCN values higher than the set values described in the Experimental Section as well as the bacterial sources of the matched proteins. According to the results we obtained from the data-dependent analysis, we imposed several criteria to select peptide markers for bacterial identification: (1) Peptides that lead to the correct bacterial source were chosen as markers. (2) The peptide signal in the full scan spectrum must have been high (ion count of >1 × 106 in the centroid mode). (3) The peptides that correspond to several microbial sources were excluded; some proteins may be shared in common by different bacteria or the same peptide (40) Jauregui, O.; Moyano, E.; Galceran, M. T. J. Chromatogr., A 2000, 896, 125-133. (41) McCourt, J.; Bordin, G.; Rodriguez, A. R. J. Chromatogr., A 2003, 990, 259269.
protein
MW
hypothetical protein hypothetical protein flagellin putative translation initiation inhibitor, yjgF family accumulation-associated protein immunodominant antigen A
6 996 21 243 34 341 13 583 15 7025 24 514
sequence may be associated with several different proteins, so database searches against the MS/MS data of the corresponding peptides will lead to multiple identifications of microorganisms. For instance, the first peptide listed in Table 1 matches a protein that belongs to two different species of bacteria, and thus, we excluded it from our selective MS/MS experiments. We chose the ions at m/z 1076.5 and 1262.6 because they have higher ion abundances. Table 2 presents the selected peptides of the three bacteria that we chose from the data-dependent experiments; these selected peptides were used for subsequent bacterial identification. Panels a and b in Figure 2 display two selected CE-MS/MS ion chromatograms for the tryptic digest of the cell extract obtained from P. aeruginasa. The two ions at m/z 1036.0 (+2) and 1638.8 (+1) that are listed in Table 2 were monitored in the SRM mode, while the product ion mass window was set to the allowable maximum mass range to cover the dissociation fragments. The separation efficiency was not ideal: it was compromised by reducing the separation time. Product ion mass spectra of the two selected precursor ions were searched against the protein database by using SEQUEST. The identified peptides were arranged in the order of their Xcorr values. If the identified peptides have the same sequence, then only the one having the Analytical Chemistry, Vol. 77, No. 5, March 1, 2005
1491
Table 3. Top Four Ranked SEQUEST Search Results According to the Xcorr Values Generated from the MS/MS Data of Two Selected Ions for Each Bacterial Sample bacterial species
peptide mass (charge)
peptide sequence
Xcorr
protein
MW
P. aeruginasa
2071.1 (+2) 1638.8 (+1) 1638.9 (+1)
-.TVIHTDNAPAAIGTYSQAIK.-.IEDTDFAAETANLTK.-.VLSGLHMPSSGSIIIK.-
3.36 3.17 1.95
13 583 34 341 13 419
2071.0 (+2)
-.ENTKGHLTYYSVSDPVFN.-
1.69
1076.5 (+1) 1262.6 (+1) 1076.6 (+1)
-.ETVGNVTDNK.-.SGEESEVLVADK.-.SMEFIGKHK.-
2.25 2.00 1.39
1262.6 (+1)
-.ERTDEDARDR.-
1.19
1899.0 (+2)
-.AHLVDLAQHNPEELNAK.-
2.91
2080.0 (+2)
-.YGPVDGDPITSTEEIPFDK.-
2.86
2080.0 (+2)
-.ADVIKVSDNELEFLTGSDK.-
1.22
hypothetical protein flagellin ABC-type sugar transport system, ATPase component hypothetical protein STY2363 hypothetical protein hypothetical protein transcription antiterminator DNA-directed RNA polymerase specialized σ subunit immunodominant antigen A accumulation-associated protein putative fructokinase
2079.2 (+2)
-.VLLALKAFSMFSLAPLVMK.-
0.99
S. aureus
S. epidermidis
highest Xcorr value was listed and it is considered as one entry of search results. Table 3 lists the first four search results that have the highest Xcorr values generated from the MS/MS data without considering the threshold Xcorr value. The two selected peptide ions associated with P. aeruginasa were correctly identified as having the highest Xcorr scores. For instance, the product ions observed in the tandem mass spectrum of the doubly charged ions at m/z 1036.0 (Figure 2c) match the b- and y-type ions and identify the peptide sequence TVIHTDNAPAAIGTYSQAIK. A database search led to the identification of a hypothetical protein associated with the corresponding bacteria. Indeed, the matched protein leads to the correct microorganism source (P. aeruginasa). The third search result identified a peptide having a mass of 1638.9 (singly charged) that is associated with a protein from Haemophilus somnus. The fourth search result gave a peptide with a mass of 2071.0 (doubly charged) that is associated with Salmonella enterica. The Xcorr values for the third and fourth matches are 1.95 and 1.69, respectively, which are much lower than the first two matches (3.36 and 3.17). Furthermore, when targeting the specific peptides corresponding to P. aeruginasa, the two selected peptides that we identified from the Xcorr values ranked as the top two. It is clear that we had identified the corresponding bacterium. Because we subjected only the MS/MS spectra of selected ions to SEQUEST searching, we greatly reduced the time spent performing database searches to minutes from the hours required for normal data-dependent searches (Table 1). The total analysis time from sample injection to data analysis is ∼20 min. The analysis time is longer than that for direct MALDI analysis because of the coupling to CE but is comparable with or better than that for other MS methods coupled to separation tools. The major timeconsuming step in the present analysis is proteolytic digestion (6 h) although the digestion time can be improved by either fast9,10 or microwave-assisted proteolysis.42 We applied our selective MS/ MS method to the identification of another two bacteria: S. aureus and S. epidermidis. For their CE-MS/MS analysis, we selected 1492
Analytical Chemistry, Vol. 77, No. 5, March 1, 2005
diaminopimelate decarboxylase
7 432 6996 21 243 21 577
microbial source P. aeruginasa P. aeruginasa Haemophilus somnus Salmonella enterica S. aureus S. aureus Bacillus subtilis
25 094
Desulfitobacterium hafniense
24 514
S. epidermidis
157 025
S. epidermidis
34 963
Petunia integrifolia Microbulbifer degradans
43 251
Figure 2. Selected CE-MS/MS ion chromatograms at m/z (a) 1036.0 (+2) and (b) 1638.8 (+1) for the tryptic digest of the cell extract obtained from P. aeruginasa. (c) Product ions mass spectrum of the doubly charged ions at m/z 1036.0.
two ions for each bacterial sample. Table 3 lists the first four search results that gave the highest Xcorr values from the MS/ MS data for each sample. The search results for the S. aureus and S. epidermidis samples reveal that the two selected peptides for each bacterium corresponded correctly with the top Xcorr scores. The results demonstrate that careful selection of the species-specific markers provides excellent selectivity for bacterial identification. Next, we investigated the effect that the number of peptide ions selected for MS/MS has on the identification process. The maximum number of selected ions allowed by the data acquisition system running in SRM mode is 12. We performed these experiments by choosing one, two, three, and eight peptide ions associated with P. aeruginasa. Selection of one, two, and three ions for MS/MS analysis gave top Xcorr scores in the search results and led to the correct identification of P. aeruginasa (data not shown). When we selected eight ions for MS/MS analysis, the MS/MS spectra of five of the ions were detected and the search results provided an unambiguous identification of the bacterium, P. aeruginasa. The missing three ions may be attributable to the long acquisition cycle time required for sequential MS/ MS of eight ions. The MS/MS of each ion takes ∼1.8 s, and these three ions could be lost during the acquisition of the other five ions under the present experimental conditions. Although the run acquisition cycle time can be shortened, it will sacrifice the spectral quality and decrease the confidence of the search results. Basically, the selected MS/MS of more ions should favor a more positive identification of the bacterium if all of these selected ions identify the same bacterial source. In the present experimental setup, we took into account the number of bacterial species to be identified and the number of good candidate ions available when deciding how many ions for each species to monitor. One advantage of selected MS/MS is its high selectivity when complex systems are analyzed. To demonstrate this advantage, we analyzed bacterial mixtures using the selective MS/MS method. Three bacteria were mixed in different ratios, and one selected peptide ion for each bacterium was monitored and subjected to MS/MS analysis and database searches. Panels a-c in Figure 3 display the selective CE-MS/MS ion chromatograms for a mixture of S. aureus, P. aeruginasa, and S. epidermidis at a dry weight ratio of 1:1:1. As mentioned above, the separation of these three ions was compromised by the short elution time. The purpose of using CE was to provide a simple method to reduce sample complexity. We subjected all of the spectra acquired in the three selected ion chromatograms to database searches. Figure 3d displays one of the MS/MS spectra of the doubly charged ions at m/z 1040.5 corresponding to the selected MS/ MS ion chromatogram for S. epidermidis. Our peak assignments are based on the search result that gave the highest Xcorr score among the output data associated with the ion at m/z 1040.5. Table 4 lists the Xcorr scores for the three selected peptides; the rankings of the Xcorr values among all the SEQUEST search results are also indicated. The peptide (m/z 1040.5) corresponding to S. epidermidis has the highest Xcorr value; the peptide (m/z 1638.8) corresponding to P. aeruginasa and the peptide (m/z 1076.5) associated with S. aureus have the second- and third(42) Wu, C.-H.; Lin, S.-S.; Lee, M.-J.; Sun, C.-M.; Ho, Y.-P. Proceedings of the 52nd ASMS Conference on Mass Spectrometry and Allied Topics, Nashville, TN, 2004.
Figure 3. Selective CE-MS/MS ion chromatograms of the three ions at m/z (a) 1076.5, (b) 1638.8, and (c) 1040.5 for the bacterial mixture of S. aureus, P. aeruginasa, and S. epidermidis (1:1:1). (d) An MS/MS spectrum of the doubly charged ions at m/z 1040.5 corresponding to the selected MS/MS ion chromatogram for S. epidermidis. Table 4. Xcorr Values and Their Rankings in Database Search Results for Each Species-Specific Peptide Ion Obtained from Various Bacterial Mixtures Xcorr (ranking) mixing ratio (P.a:S.a:S.e)
P. aeruginosa m/z 1638.8a
S. aureus m/z 1076.5a
S. epidermidis m/z 1040.5a
1:1:1 1:0:1 1:0:4 0:10:1 1:0:10 10:0:1 1:10:10 1:50:0 1:0:50 0:100:1 10:100:1
2.33 (2) 3.07 (2) 2.03 (2)
2.04 (3)
3.09 (1) 3.25 (1) 3.30 (1) 2.83 (1) 3.29 (1) 3.13 (1) 2.90 (1)
a
2.06 (2) 2.19 (2) 3.06 (2) 1.87 (3) 1.55 (2) 1.83 (2) 2.09 (1)
1.88 (2) 2.15 (1) 2.34 (1) 1.98 (2)
3.69 (1) 1.22 (4) 1.74 (3)
Selected ion.
highest Xcorr values, respectively. Our results indicate that all three of the peptides associated with their specific bacteria could be identified correctly, which allows the positive identification of their bacterial sources. Analytical Chemistry, Vol. 77, No. 5, March 1, 2005
1493
Figure 4. Selective CE-MS/MS ion chromatograms of the three ions at m/z (a) 1076.5, (b) 1638.8, and (c) 1040.5 for the bacterial mixture of S. aureus, P. aeruginasa, and S. epidermidis (10:10:1).
Next, we analyzed various bacterial mixtures that differ in their mixing ratios and compositions; Table 4 also summarizes the Xcorr scores for the selected peptides and their rankings. To tighten the criteria for positive identification of the bacteria, we restricted the correct identification to having the Xcorr values among the top rankings of the total search results. Our identifications were very successful. Ten out of 11 total analyses indicated that the selected peptides were all among the top two or three candidates, depending on the number of bacterial species mixed together. In general, we identified the minor components with high confidence. The most intriguing results listed in Table 4 are those provided by the mixtures that contain their minor component at 1% abundance relative to the major component. Even when the minor component was in such a low relative concentration, our method successfully identified the correct bacterial component. To the best of our knowledge, this report is the first that demonstrates the detection of a minor bacterial component present in such a low relative concentration. Figure 4 illustrates selective CE-MS/MS ion chromatograms recorded from the mixture of S. aureus, P. aeruginasa, and S. epidermidis (10:10:1). We note that the detection of the minor component (∼1%) is pretty much the limit under the present experimental conditions. As indicated in Figure 4c, the selected MS/MS ion chromatogram for S. epidermidis is noisy and only three MS/MS spectra correctly identify the selected peptide. The only search result that provided a false identification of the minor component was that of the 100:1 mixture of S. aureus and S. epidermidis. The Xcorr value for the doubly charged peptide ion at m/z 1040.5, which corresponds to S. epidermidis (the minor component), is 1.22 and ranked fourth. In addition to approaching the detection limit, the poor match of this spectrum may arise because of the reduced duty cycle. When two selected ions were acquired alternately, the duty cycle for each ion was reduced by half relative to the duty cycle obtained for a single ion acquisition. This effect should be more significant when dealing with the minor component. To eliminate this sampling problem, we may perform selective MS/MS of a single ion associated with the minor component. Figure 5a displays a selective MS/MS ion chromatogram of a peptide (m/z 1040.5) corresponding to S. epidermidis 1494 Analytical Chemistry, Vol. 77, No. 5, March 1, 2005
Figure 5. (a) MS/MS ion chromatogram of a single selected peptide (m/z 1040.5) corresponding to S. epidermidis from a 100:1 mixture of S. aureus and S. epidermidis. (b) An MS/MS spectrum of the ions at m/z 1040.5 corresponding to the selected MS/MS ion chromatogram.
obtained from a 100:1 mixture of S. aureus and S. epidermidis; a tandem mass spectrum of the selected ion is displayed in Figure 5b. The product ion assignments correspond to the correct sequence of the selected peptide. The Xcorr score was improved from 1.22 to 1.96 and ranked first. To simulate a real bacteria-containing sample, we spiked a saliva sample with a single colony of the bacteria P. aeruginasa and E. coli. We cultured the sample and collected the bacterial cells as described in the Experimental Section. We analyzed the tryptic digest of the protein extract obtained from the bacterial cells by using both direct data-dependent and selective MS/MS approaches. The search results for the data-dependent analysis reveal 10 matched peptides associated with E. coli and only 1 matched peptide associated with P. aeruginasa among the top 20 identified peptides (data not shown). It is reasonable to expect that E. coli outgrew P. aeruginasa during cell culturing under the growth condition we used, which would cause more proteins from E. coli to be present in the cell extract. The search result definitely identifies the bacterium E. coli, but it leaves some uncertainty in the identification of P. aeruginasa. When the tryptic digest was analyzed using the selective MS/MS analysis of peptides targeted for P. aeruginasa (Figure 6a), we correctly assigned the chosen peptide at m/z 1638.8 eluted at 10.22 min (Figure 6b) with its Xcorr value (2.01) ranking first. Another peak centered at 8.26 min in the chromatogram corresponds to an MS/MS spectrum of a peptide that we matched with a partial sequence of a protein belonging to Vibrio parahaemolyticus. The Xcorr value (1.71) ranked fourth, and the presence of this microorganism is not realized. It would not be a surprise if there were other bacterial species present in the sample, considering the complexity of the saliva sample. It might be possible that the MS/MS of the selected ion at certain values of m/z identifies a bacterial species that gives a higher Xcorr score than that for the bacteria investigated. When such uncertainty appears, we may perform the selective MS/MS of more than one species-specific peptide. This process should increase the confidence level for the positive (or negative) identification of the bacteria under investigation.
peptide ion that are specific to the bacterial species of interest and exclude the peptides that lead to multiple or false microbial identifications. Consequently, the selective CE-MS/MS analyses of bacteria-specific marker peptides provide the advantages of excellent selectivity and high accuracy in the identification of bacterial species present in complex systems. This selective MS/ MS approach should also be suitable for coupling to other separation methods such as liquid chromatography. We have applied the selective CE-MS/MS method to the successful identification of bacterial mixtures. The minor bacterial species present at even 1% relative abundance can be identified with high confidence. In addition, we have used this method also to demonstrate the identification of P. aeruginasa in a saliva sample spiked with both E. coli and P. aeruginasa. Figure 6. (a) Selective MS/MS ion chromatogram of a single peptide at m/z 1638.8 targeted for P. aeruginasa. (b) A tandem mass spectrum of the correctly identified peptide at m/z 1638.8.
CONCLUSIONS Our proposed selective CE-MS/MS method identifies specific bacteria in bacterial mixtures. Our preliminary analyses of pure pathogens through database searches selected abundant marker
ACKNOWLEDGMENT We thank the National Science Council of the Republic of China for supporting this research financially.
Received for review December 9, 2004.
October
21,
2004.
Accepted
AC0484427
Analytical Chemistry, Vol. 77, No. 5, March 1, 2005
1495
