Discrimination between Bacterial Spore Types Using Time-of-Flight

Apr 12, 2001 - Discrimination between Bacterial Spore Types Using Time-of-Flight Mass Spectrometry and Matrix-Free Infrared Laser Desorption and ...
0 downloads 0 Views 103KB Size
Download PDF
Anal. Chem. 2001, 73, 2331-2337

Discrimination between Bacterial Spore Types Using Time-of-Flight Mass Spectrometry and Matrix-Free Infrared Laser Desorption and Ionization J. N. Ullom,*,† M. Frank,† E. E. Gard,† J. M. Horn,† S. E. Labov,† K. Langry,† F. Magnotta,† K. A. Stanion,† C. A. Hack,‡ and W. H. Benner‡

Lawrence Livermore National Laboratory, Livermore, California 94550, and Lawrence Berkeley National Laboratory, Berkeley, California 94720

We demonstrate that molecular ions with mass-to-charge ratios (m/z) ranging from a few hundred to 19 050 can be desorbed from whole bacterial spores using infrared laser desorption and no chemical matrix. We have measured the mass of these ions using time-of-flight mass spectrometry and we observe that different ions are desorbed from spores of Bacillus cereus, Bacillus thuringiensis, Bacillus subtilis, and Bacillus niger. Our results raise the possibility of identifying microorganisms using mass spectrometry without conventional sample preparation techniques such as the addition of a matrix. We have measured the dependence of the ion yield from B. subtilis on desorption wavelength over the range 3.05-3.8 µm, and we observe the best results at 3.05 µm. We have also generated mass spectra from whole spores using 337-nm ultraviolet laser desorption, and we find that these spectra are inferior to spectra generated with infrared desorption. Since aerosol analysis is a natural application for matrix-free desorption, we have measured mass spectra from materials such as ragweed pollen and road dust that are likely to form a background to microbial aerosols. We find that these materials are readily differentiated from bacterial spores. Techniques for rapidly and accurately identifying microorganisms are desirable for a wide range of environmental and medical applications. Mass spectrometry has been used to identify microorganisms for over 25 years1 and in recent years matrixassisted laser desorption and ionization time-of-flight mass spectrometry (MALDI-TOF-MS) has emerged as a particularly promising technique for this purpose. Identification can be performed from intact cells,2-10 and as a measure of sensitivity of the technique, MALDI-TOF has been used to distinguish between †

Lawrence Livermore National Laboratory. Lawrence Berkeley National Laboratory. (1) Anhalt, J. P.; Fenselau C. Anal. Chem. 1975, 47, 219. (2) Holland, R. D.; Wilkes, J. G.; Rafii, F.; Sutherland, J. B.; Persons, C. C.; Voorhees, K. J.; Lay, J. O. Rapid Commun. Mass Spectrom. 1996, 10, 1227. (3) Krishnamurthy, T.; Ross, P. L. Rapid Commun. Mass Spectrom. 1996, 10, 1992. (4) Claydon, M. A.; Davey, S. N.; Edwards-Jones, V.; Gordon, D. B. Nat. Biotechnol. 1996, 14, 1584. ‡

10.1021/ac001551a CCC: $20.00 Published on Web 04/12/2001

© 2001 American Chemical Society

strains of the same bacterial species.4,5,7-9 However, MALDI-TOF requires some preparation of the analyte material, namely, the addition of a chemical matrix to enhance desorption and ionization. In addition, the analyte is sometimes preprocessed, for instance with a corona plasma discharge, to disrupt the organisms and increase the accessibility of signature molecules or biomarkers.11 This sample preparation is an impediment to real-time and in situ analysis. As a step toward real-time identification of microorganisms, we used both a 337-nm nitrogen laser and an infrared laser tunable between 2.9 and 3.8 µm to desorb ions for TOF-MS from whole bacterial spores with no matrix or preprocessing. The use of these types of laser for matrix-free desorption from microorganisms is novel. Previously, fast atom bombardment, carbon dioxide lasers, and plasma desorption were explored.12,13 We selected the nitrogen laser because of the widespread availability of these instruments and because their output wavelength is well-matched to the tryptophan absorption feature. We selected the infrared laser because such sources are newly available and because their output is well-matched to an absorption feature near 3 µm caused by OH, NH, CH2, and CH3 stretching modes. This absorption feature is common both to organic molecules, such as proteins, and to water. We find that the infrared laser desorbs both more types of ions and heavier ions than the nitrogen laser and the desorption mechanisms studied previously. (5) Haag, A. M.; Taylor, S. N.; Johnston, K. H.; Cole, R. B. J. Mass Spectrom. 1998, 33, 750. (6) Welham, K. J.; Domin, M. A.; Scannell, D. E.; Cohen, E.; Ashton, D. S. Rapid Commun. Mass Spectrom. 1998, 12, 1998. (7) Arnold, R. J.; Reilly, J. P. Rapid Commun. Mass Spectrom. 1998, 12, 630. (8) Hathout, Y.; Demirev, P. A.; Ho, Y. P.; Bundy, J. L.; Ryzhov, V.; Sapp, L.; Stutler, J.; Jackman, J.; Fenselau, C. Appl. Environ. Microbiol. 1999, 65, 4313. (9) Leenders, F.; Stein, T. H.; Kablitz, B.; Franke, P.; Vater, J. Rapid Commun. Mass Spectrom. 1999, 13, 943. (10) Saenz, A. J.; Petersen, C. E.; Valentine, N. B.; Gantt, S. L.; Jarman, K. H.; Kingsley, M. T.; Wahl, K. L. Rapid Commun. Mass Spectrom. 1999, 13, 1580. (11) Birmingham, J.; Demirev, P.; Ho, Y. P.; Thomas, J.; Bryden, W.; Fenselau, C. Rapid Commun. Mass Spectrom. 1999, 13, 604. (12) Heller, D. N.; Fenselau, C.; Cotter, R. J.; Demirev, P.; Olthoff, J. K.; Honovich, J.; Uy, M.; Tanaka, T.; Kishimoto, Y. Biochem. Biophys. Res. Commun. 1987, 142, 194. (13) Heller, D. N.; Cotter, R. J.; Fenselau, C.; Uy, O. M. Anal. Chem. 1987, 59, 2806.

Analytical Chemistry, Vol. 73, No. 10, May 15, 2001 2331

We studied desorption from whole spores of Bacillus cereus, Bacillus thuringiensis, Bacillus subtilis, and Bacillus niger. Laser targets were prepared by evaporating an aqueous spore suspension on the sample holder. We find both differences and similarities among the biomarkers desorbed from the four spore types. The biomarkers desorbed from B. subtilis and B. niger are entirely distinct from those desorbed from the distantly related B. cereus and B. thuringiensis. However, the biomarkers desorbed from the closely related B. subtilis and B. niger differ only slightly. Likewise, there are substantial similarities between the biomarkers desorbed from the closely related B. cereus and B. thuringiensis. In general, the number of biomarkers obtained using matrix-free desorption is less than is typically observed with MALDI. For one of the species, B. subtilis, we measured the dependence of the ion yield on the desorption wavelength over the range 3.05-3.8 µm and we achieved the best results at 3.05 µm. We also conducted initial measurements of the sensitivity of matrix-free infrared desorption. Our results are particularly significant in light of recently developed techniques for the real-time characterization of individual aerosol particles by time-of-flight mass spectrometry.14-17 Ion generation by laser interrogation of aerosol particles drawn through a pumped inlet is ideal for real-time analysis but is not presently amenable to the addition of a matrix or significant preprocessing of the aerosol. However, since we have shown that biomarkers can be generated from microorganisms without a matrix, a real-time monitor of microbial aerosols merits consideration. Such an instrument must be able to distinguish between background aerosols and pathogenic microbes. Therefore, we have also measured mass spectra from materials that might be present in common background aerosols and we find them to be significantly different from spore spectra. EXPERIMENTAL APPROACH Four types of bacterial cells were obtained from the American Type Culture Collection (ATCC): B. subtilis (ATCC 6051), B. niger (ATCC 9372), B. cereus (ATCC 14579), and B. thuringiensis (ATCC 10792). All four organisms are biosafety level 1. The four cell types comprise two pairs of closely related organisms. The ATCC describes B. niger as a variant of B. subtilis and B. thuringiensis as a subspecies of cereus. Analysis of the DNA in seven genes indicates that B. thuringiensis and B. cereus diverge from B. subtilis by 32% but from each other by only 2.5%.18 (It should be noted that different strains of B. subtilis and B. thuringiensis were examined in ref 18 than were used here.) Detection and identification of the cell types B. cereus and B. thuringiensis are of particular interest because of their close relation to the pathogenic Bacillus anthracis. After inoculation (1:50) with stationary-phase cultures, bacterial cells were grown in 1/4X TY sporulation media with agitation for at least 3 days at 30 °C.19 Spores were then prepared by the (14) For a tutorial, see: Johnston, M. V. J. Mass Spectrom. 2000, 35, 585. (15) Hinz, K. P.; Kaufmann, R.; Spengler, B. Aerosol Sci. Technol. 1996, 24, 233. (16) Gieray, R. A.; Reilly, P. T. A.; Yang, M.; Whitten, W. B.; Ramsey, J. M. J. Microbiol. Methods 1997, 29, 191. (17) Gard, E. E.; Kleeman, M. J.; Gross, D. S.; Hughes, L. S.; Allen, J. O.; Morrical, B. D.; Fergenson, D. P.; Dienes, T.; Galli, M. E.; Johnson, R. J.; Cass, G. R.; Prather, K. A. Science 1998, 279, 1184. (18) Helgason, E.; Okstad, O. A.; Caugant, D. A.; Johansen, H. A.; Fouet, A.; Mock, M.; Hegna, I.; Kolsto, A. B. Appl. Environ. Microbiol. 2000, 66, 2627. (19) Lazzarini, R. A.; Santangelo, E. J. Bacteriol. 1967, 94, 125.

2332

Analytical Chemistry, Vol. 73, No. 10, May 15, 2001

method of Bourne et al.20 Sporulated cells were harvested by centrifugation, washed twice in 1 M NaCl, and then washed twice more with deionized, sterile water. Washed sporulated cells were then resuspended in 50mM TRIS, pH 7.0, and treated with 250 µg/mL lysozyme for an incubation period of 1 h at 37 °C to lyse vegetative cells. Lysozyme was removed from treated spores by centrifugation and subsequent washing of the spore pellet in 50 mM TRIS, pH 7.0, followed by deionized water washes. Prepared spores were stored in deionized water at 5 °C. Samples were prepared by pipetting several microliters of solution to the sample probe, which was then dried for several minutes under flowing nitrogen gas before being inserted into the mass spectrometer. Sample material was desorbed and ionized by the idler beam from a custom-built infrared optical parametric oscillator (OPO). The OPO used an unseeded linear Fabry-Perot cavity design with bulk lithium niobate as the nonlinear medium. The OPO was endpumped at 1.06 µm by a Big Sky CFR 200-10 Nd:YAG laser and operated singly resonant on the signal wavelength. The duration of the YAG pulses was less than 13 ns. The crystal and optics were coated for operation over an idler range of 3.2-3.8 µm. In practice, the idler was angle-tilt tunable from 2.9 to 4.2 µm with a line width of ∼2 cm-1. The delivered energy decreased sharply at both wavelength extremes. The maximum total conversion efficiency of the signal and idler was ∼25%. Pulse energies were measured using a calibrated pick-off and a Molectron J4-09 energy meter. We typically used pulse energies between 1 and 4 mJ. We profiled the beam with a slit and power meter and determined that the focused beam spot was elliptical and that its intensity fell to half-maximum over an area of 0.9 mm2. Hence, typical intensities were 1-4 × 107 W/cm2. Mass spectra obtained using IR desorption were generated in a custom-built linear time-of-flight mass spectrometer operated in positive ion mode. The acceleration voltage was 23 kV with a 5-kV extraction pulse applied after a delay of 1.86 µs to optimize the mass resolution near m/z ) 1000. A pulsed electrostatic gate was sometimes used to deflect an initial surge of light ions away from the detector. The gate was set to suppress ions with m/z < 380. The ion gate was only useful at high laser energies; for instance, at 4 mJ/pulse. We did not observe biomarkers suitable for differentiating spore types among these initial light ions which, presumably, consisted of low-mass molecules or fragments of heavier molecules. A microchannel plate was used to measure ion flight times. The output of the microchannel plate was amplified and then recorded on a LeCroy 7200 digital oscilloscope set to digitize at 2.5 ns/point. Data traces shown in this report were smoothed for clarity. A 5-point Gaussian smooth was used in Figures 1 and 4-6. In Figure 2a and b, a 78-point Gaussian smooth was used for m/z < 8000 and a 408-point Gaussian smooth for m/z > 18 000. Mass spectra obtained using UV desorption were generated in a Perceptive Biosystems Voyager DE-STR mass spectrometer operated in linear mode. The UV source was a 337-nm N2 laser with the pulse energy set to 25 µJ. The beam spot size was measured to be 0.03 mm2 from the dimensions of the ablation region in targets of organic matrix. The pulse duration was 4 ns, yielding a beam intensity of 2 × 107 W/cm2. Hence, the intensities of the UV and IR laser sources used in this study were similar, (20) Bourne, N.; Fitzjames, P. C.; Aronson, A. I. J. Bacteriol. 1991, 173, 6618.

but the beam spot of the IR laser was 30 times larger than the UV. We found that pulses from the infrared laser generated lowmass ions even when no target material was present on the stainless steel sample probe. We found that metal foils of platinum and aluminum yielded fewer ions than stainless steel and therefore added a layer of foil to the sample tip. This background was further reduced by irradiating the foil with the IR laser for 5 min before adding sample material. Thick samples also helped to suppress background ions. When dilute samples were analyzed in order to determine the analytical sensitivity limits, significant numbers of background ions were present at values of m/z as high as 1000. The source of the background ions is not known, but they may be residual organics, possibly pump oil, adsorbed on the metal surface. The flight tube was evacuated with two turbo pumps backed by an oiled rotary vane pump, and the vacuum interlock was evacuated with an oiled rotary vane pump. Mass calibrations were generated using flight-time measurements of Cs(CsI)1-4, the polypeptide substance P, the dimer of substance P, doubly charged insulin, insulin, and myoglobin. Chemicals were used as provided from Sigma (St. Louis, MO). For a mass range of interest, reference peaks that spanned the range were fit to the functional form mass ) R(time - β)2, where R and β are fitting parameters. For the comparison of B. subtilis and pure surfactin, substance P was added to both targets to provide an internal calibration. Measurement of wavelength-dependent effects was complicated by the spatial motion of the OPO beam when the lithium niobate crystal was tilted to generate different wavelengths. A steering mirror was used to compensate for this motion and keep the position of the beam spot constant. In addition, flat acceleration plates were used in the mass spectrometer to reduce the dependence of the ion yield on beam spot position. We verified experimentally that the variation of the beam spot size with wavelength was negligible. We deduced the infrared attenuation length in spores of B. subtilis from transmission measurements taken with a Bomen MD100 FT-IR spectrometer. Given the ratio, T, of the transmission through KCl pellets containing and lacking spores, the attenuation length is given by dfv/ln(T-1), where d is the pellet thickness and fv is the volume fraction occupied by spores. It should be noted that since both scattering and absorption contribute to attenuation, the attenuation length is only a lower bound on the absorption length. Sensitivity measurements were performed by drying increasingly diluted spore solutions on the target. Spore concentrations were measured using a Coulter Multisizer II particle counter. To calculate the number of spores interrogated by a laser pulse, we used the measured area of the beam spot and assumed a uniform distribution of spores on the target. It was important that spore solutions dried homogeneously on the sample probe since it was larger than the beam spot. To improve homogeneity, target foils were roughened with a fine file before solution was applied. Without this roughening, the final distribution of spores from dilute solutions was heterogeneous to the eye. RESULTS AND DISCUSSION Biomarkers. Mass spectra from four Bacillus spore types obtained using matrix-free infrared desorption and ionization at λ

Figure 1. Mass spectra from whole spores of B. subtilis, B. niger, B. cereus, and B. thuringiensis over the range m/z ) 400-1500. Spectra are averages of 25-50 shots from two or three samples. The energy per shot was 3.4-4.25 mJ.

) 3.34 µm are shown in Figure 1. The spectra of B. subtilis and B. niger are very similar in this mass range but differ markedly from the spectra of B. cereus and B. thuringiensis. The spectra of B. cereus and B. thuringiensis share sets of features near m/z ) 905 and 1330. The spectrum of B. thuringiensis includes features below m/z ) 720 that are weakly present, if present at all, in B. cereus. However, comparison of spectra on the basis of peak heights is notoriously unreliable and therefore we do not claim to be able to discriminate between B. cereus and B. thuringiensis on the basis of data from this mass range. Differences of 1-2 in m/z, for example between peaks from B. subtilis and B. niger, are within the measurement uncertainty. The closely related spore types B. subtilis and B. niger can be differentiated by biomarkers shown in Figure 2a near m/z ) 2900. For B. subtilis, the laser fluence required to produce these features reliably lies between 1.5 and 2.7 mJ at 3.43 µm, which is higher than the fluence needed to produce the features near m/z ) 1050. The laser fluence required to generate the biomarkers near m/z ) 1050 lies between 0.8 and 1.0 mJ at 3.43 µm. Even at the highest laser power accessible, we do not observe any biomarkers from B. subtilis and B. niger heavier than those in Figure 2a. As shown in Figure 2b, the spectra of B. cereus and B. thuringiensis contain biomarkers of significantly greater mass than those in Figure 1. These biomarkers provide a means of differentiating the two spore types; in particular, B. cereus displays a doublet structure near m/z ) 7300 which is not present in the spectra from B. thuringiensis. It can be seen to the right of the scale break at m/z ) 8000 that a biomarker is present from both spores near m/z ) 19 050. This is the heaviest peak Analytical Chemistry, Vol. 73, No. 10, May 15, 2001

2333

Figure 3. Mass spectra of whole spores of B. subtilis, B. niger, B. cereus, and B. thuringiensis obtained using matrix-free 337-nm desorption. A spectrum from the bare gold substrate is also shown and can be seen to contribute the strong feature at m/z ) 591 due to 3Au+. The spectra are 50 shot averages.

Figure 2. (a) Mass spectra from whole spores of B. subtilis and B. niger from m/z ) 2800 to 3100. (b) Mass spectra from whole spores of B. thuringiensis and B. cereus from m/z ) 2000 to 8000 and from m/z ) 18 000 to 20 000. The vertical scale is reduced by a factor of 4 above m/z ) 18 000. The desorption wavelength was 3.34 µm, and the energy per shot was 3.4-4.25 mJ. Mass labels are derived from local Gaussian fits. Below m/z ) 8000, the Gaussian centroids depend on the choice of end points by ∼3 Da or less. The full width half-maxima of the peaks near m/z ) 19 050 are roughly 400 Da.

observed. The large width of this peak suggests both that many ions of interest reach the detector and that substantial fragmentation of these ions occurs. The peaks in Figure 2b are observed on top of a large continuum background, which has been removed in the plot. Not shown in Figure 2b are a pair of minor peaks near m/z ) 14 630 and 15 720 from both B. cereus and B. thuringiensis. Mass spectra obtained from whole spores using matrix-free desorption and 337-nm ultraviolet laser pulses are shown in Figure 3. The laser intensity, 2 × 107 W/cm2, was chosen to yield the highest quality spectra. Lower laser intensities created no ions; higher intensities created a large, unresolvable background at low mass and no high-mass ions. It can be seen that the ultraviolet laser is able to desorb the same complex of ions near m/z ) 1075 from B. subtilis and B. niger as the infrared laser. However, the ultraviolet laser does not generate the features from B. thuringiensis and B. cereus observed using infrared desorption. In addition, the ultraviolet laser does not desorb any biomarkers above the range shown in Figure 3. Hence, we conclude that infrared radiation near 3 µm desorbs both more types of ions and heavier ions than 337-nm ultraviolet radiation. The ability of the infrared 2334 Analytical Chemistry, Vol. 73, No. 10, May 15, 2001

laser to desorb heavy ions is important for bacterial identification since heavy, complex ions are more likely to be species-specific. It is possible, but not proven, that small molecules activated by the infrared laser act as de facto matrix material for the desorption and ionization of heavier, identifying molecules. We have tentatively identified some of the ions in Figures 1-3. The cluster of ions near m/z ) 1075 from B. subtilis and B. niger is believed to be the lipoheptapeptide surfactin based on previous MALDI-TOF mass spectrometry of whole spores of B. subtilis8 and vegetative cells.9 Surfactin, a natural antibiotic, is described in ref 21. Surfactin can contribute multiple peaks to the mass spectrum near m/z ) 1075 because of natural variation in the length of its fatty acid chain and because of variation in the cation. The mass spectrum of pure surfactin (Sigma, Catalog No. S3523) obtained using matrix-free infrared desorption is shown in Figure 4 along with that of B. subtilis. It can be seen that the spore features are a subset of the pure surfactin spectrum. The cluster of ions near m/z ) 907 in B. thuringiensis has also been observed using MALDI.22 On the basis of Figure 1, we believe that a closely related or identical compound is present in B. cereus. An unresolved but important question is the origin of the ions observed in Figures 1-3. In a separate series of experiments, spores of B. subtilis were collected by 2 min of centrifugation at 16000g. The supernatant was then drawn off, dried, and used as a target in the mass spectrometer. The dried supernatant yielded (21) Zuber, P.; Nakano, M. M.; Marahiel, M. A. In Bacillus subtilis and Other Gram-Positive Bacteria; Sonenshein, A. L., Hoch, J. A., Losick R., Eds.; American Society for Microbiology: Washington, DC, 1993; Chapter 61. (22) Hathout, Y.; Ho, Y. P.; Ryzho, V.; Demirev, P.; Fenselau, C. Mass spectrometry techniques for structural characterization of a new class of cyclic lipopeptides. 48th American Society for Mass Spectrometry Conference, Long Beach CA, June 11-15, 2000; oral presentation.

Figure 4. Mass spectra of pure surfactin and spores of B. subtilis generated with a desorption wavelength of 3.05 µm. A fluence of 1.5 mJ/pulse was used for both targets. The microchannel plate signal in the surfactin spectrum has been divided by 2 for easier comparison.

the same biomarkers as the spore solution, suggesting that some of the observed molecules leave the spores in solution. It is possible that these detached molecules are desorbed and ionized preferentially from the dried spore solution. Proteins have been detected previously in the acqueous supernatant of bacterial cells.5 We continue to observe the same biomarkers from B. subtilis spores that have been pelleted, separated from the supernatant, resuspended in deionized water, and immediately dried and used as targets. We also observe the same biomarkers when spores are pelleted, resuspended in acetone, allowed to soak for 1 h with periodic agitation, pelleted again, resuspended in deionized water, and immediately dried and used as targets. These two observations suggest that the biomarker molecules are present both in solution and on the spores and can be desorbed in both conditions. Wavelength Dependence. The use of a tunable IR laser allowed us to probe the wavelength dependence of ion generation. Mass spectra of B. subtilis obtained with different desorption wavelengths but constant desorption power are shown in Figure 5a. The energy per laser shot was 1.5 mJ. The shortest wavelength in Figure 5a, 3.05 µm, is the shortest wavelength at which our OPO can achieve this fluence. When mass spectra are compared, large sample-to-sample variations in peak heights must be taken into account. The traces shown in Figure 5a are averages of 30 shot spectra from six to nine different samples. Peak heights at m/z ) 1075 are shown in Figure 5b. Error bars show the standard deviation among the samples averaged at each wavelength. It can be seen that the standard deviations are comparable to the means. Trunks on the error bars show the standard deviation divided by the square root of the number of samples, which gives the error in the mean (∆m) if the peak heights obey a Gaussian distribution. It can be seen that there is a statistically significant reduction in ion intensity at wavelengths above 3.4 µm. Also shown in Figure 5b is the inverse of the infrared attenuation length in B. subtilis. The reduction in ion intensity above 3.4 µm correlates with a reduction in absorption by the spores. In addition, the mean ion yield at 3.2 µm is 2∆m lower than the yield at 3.4 µm even though the spore absorption is similar at the two wavelengths. We speculate that wavelengths near the center of the 3.4-µm absorption feature desorb ions more efficiently than wavelengths on the red shoulder of the 3-µm absorption feature. While the precise mechanisms of desorption and ionization in matrix-free prepara-

Figure 5. (a) Mass spectra obtained using desorption wavelengths spanning the range 3.05-3.8 µm. The energy per shot was 1.5 mJ for all wavelengths. The displayed spectra are the average of six to nine 30-shot spectra from different samples. (b) Black squares show the signal intensity at m/z ) 1075 plotted versus desorption wavelength. The continuous curve shows the inverse of the attenuation length of infrared light in spores of B. subtilis. The attenuation length has a mininimum of 4-5 µm for 3-µm light.

tions are unknown, from the data in Figure 5, we can make the modest statement that absorption of laser light in the molecular modes of the target must play a role. Sensitivity. Most of the spectra shown in this report are averages of many laser shots on roughly 107 spores. However, we also characterized the sensitivity of matrix-free desorption and ionization by measuring mass spectra from increasingly dilute samples. Because the real-time identification of bioaerosols is of particular interest to us, sensitivity measurements were performed using single laser shots on fresh targets. This approaches the conditions employed in existing aerosol analyzers that interrogate particles with either a single laser pulse or a pair of pulses in a two-step volatilization-ionization sequence.14 We observed the biomarker near m/z ) 2900 from B. subtilis in roughly half the single-shot spectra from targets of 2 × 104 spores, as estimated from the spore concentration and laser beam spot size. The desorption wavelength was 3.43 µm, and the energy per pulse was 4.4 mJ. At this level of dilution, none of the structure in Figure Analytical Chemistry, Vol. 73, No. 10, May 15, 2001

2335

2a was apparent: the biomarker appeared as a broad peak spanning the range m/z ) 2800-3000. Sensitivity measurements were complicated by ions generated directly from the substrate. The signal from substrate ions grew more prominent as targets grew more dilute and made it impossible to accurately characterize the detection threshold of the biomarker from B. subtilis near m/z ) 1075. We note that ions from the substrate will be absent in a true aerosol analyzer. In similar measurements carried out on B. thuringiensis, the cluster of biomarkers near m/z ) 1350 was present in roughly half the single-shot spectra from targets of 1.5 × 105 spores. The desorption wavelength was 3.43 µm, and the energy per pulse was 3.4 mJ. These were the most readily observed biomarkers from B. thuringiensis. If a spore diameter is taken to be 1 µm, then an aerosol particle with 10-µm diameter contains at most 1000 spores, so it is clear that somewhat higher sensitivity is necessary for accurate identification of spore-containing aerosol particles. Additional, secondary ionization is one possible avenue to achieve this goal. Further optimization of the desorption wavelength and power is also possible. Background Signal Sources. Microbial aerosols will be encountered amidst a background of other organic and inorganic aerosols.23 Hence, it is important to establish that bacterial spores can be differentiated from likely background particles using matrix-free IR desorption and ionization. While an exhaustive catalog is beyond the scope of this report, mass spectra from a limited number of background materials are compared to B. subtilis in Figure 6. The background materials we examined do not show spectral features at values of m/z as high as the spores and can clearly be distinguished from them. The spectra of giant and short ragweed pollen share features between m/z ) 450 and 650. The spectra of Arizona road dust and kaolin, an industrially important aluminosilicate clay, share features below m/z ) 350. A spectrum from the cleaned platinum foil used as a target substrate is also included. CONCLUSIONS We have compared mass spectra generated from whole bacterial spores by matrix-free ultraviolet and infrared laser desorption. We find that a 3-µm infrared laser desorbs both more types of ions and heavier ions than a 337-nm ultraviolet laser. The 3-µm laser also desorbs significantly heavier ions than other matrix-free techniques used previously on bacteria.12,13 The highest mass-to-charge ratio observed with the infrared laser was 19 050. For matrix-free desorption, infrared radiation at 3 µm may provide a good compromise between the gentle desorption and low ionization of a 10.6-µm carbon dioxide laser and the high ionization but fragmentation-prone desorption of an ultraviolet laser. Mass spectra generated using matrix-free infrared desorption from the spore types B. subtilis and B. niger can clearly be distinguished from those of B. cereus and B. thuringiensis. The distinction is made easier by species-specific lipopeptides with masses near 1000 Da which are readily volatilized and yield strong mass spectral signatures. Discrimination between pairs of closely related spore types, i.e., B. subtilis and B. niger, and B. cereus and (23) Airborne Particles; Subcommittee on Airborne Particles, Committee on Medical and Biologic Effects of Environmental Pollutants, Division of Medical Sciences, Assembly of Life Sciences, National Research Council, University Park Press: Baltimore, MD, 1979.

2336 Analytical Chemistry, Vol. 73, No. 10, May 15, 2001

Figure 6. Mass spectra from Arizona road dust, kaolin, short ragweed pollen, giant ragweed pollen, spores of B. subtilis, and bare platinum. All spectra were generated with a desorption wavelength of 3.2 µm. The pulse energy for each target was chosen to yield the clearest peak definition. The pulse energy for road dust was 0.7 mJ, for kaolin, 2.5 mJ, for short ragweed, 2.3 mJ, for greater ragweed, 3.0 mJ, for B. subtilis, 1.5 mJ, and for platinum, 3.5 mJ. Ion gates were not used for any of these measurements. It can be seen from the trace for B. subtilis that there is no need to deflect light ions with the gates at this laser power level. The lowest calibration mass was 393 Da so absolute masses below this value should be used with caution.

B. thuringiensis, is possible but more difficult. These spectra are differentiated by ions near m/z ) 2900 and above whose chemical nature is not known but whose masses approach and enter the regime of coat proteins. The ability of the 3-µm laser to desorb heavy ions is therefore crucial for discrimination between spore species. We have tuned the desorption wavelength through the natural absorption feature of the spores near 3 µm caused by OH, NH, CH2, and CH3 stretching frequencies. We observe strong ion desorption when the laser is tuned to this resonance and weak desorption at longer wavelengths. These results suggest that matrix-free desorption using lasers tuned to the amide I absorption feature near 6 µm merits study. We have shown that the biomarkers desorbed from spores are different from and heavier than ions desorbed from a small number of potential background materials such as road dust and pollen. We have also shown that biomarkers can be observed from single-shot spectra of as few as 2 × 104 spores. Future work will focus on improving this sensitivity. In general, we hope this exploration of matrix-free infrared laser desorption for TOF-MS advances efforts directed toward the identification of microorganisms without reagents or preprocess-

ing. One application of matrix-free infrared desorption that we are currently exploring is the real-time analysis of microbial aerosols. Because of its simplicity and speed, matrix-free infrared desorption may also find applications beyond microorganism identification.

National Laboratory under Contract W-7405-ENG-48 and supported by the Director, Office of Energy Research, Office of Health and Environmental Research, Human Genome Project, U.S. Department of Energy under Contract DE-AC03-76SF00098.

ACKNOWLEDGMENT The assistance of Sue Martin with the particle counter is gratefully acknowledged. This work was performed under the auspices of the U.S. Department of Energy by Lawrence Livermore

Received for review December 29, 2000. Accepted March 7, 2001. AC001551A

Analytical Chemistry, Vol. 73, No. 10, May 15, 2001

2337