Lectin-Based Affinity Capture for MALDI-MS Analysis of Bacteria

Samples in phosphate buffer or urine were applied to the capture surface and allowed to interact. The capture surface was then washed to remove salts ...
6 downloads 0 Views 46KB Size
Anal. Chem. 1999, 71, 1460-1463

Technical Notes

Lectin-Based Affinity Capture for MALDI-MS Analysis of Bacteria Jonathan Bundy and Catherine Fenselau*

Department of Chemistry and Biochemistry, University of Maryland, College Park College Park, Maryland 20742

Immobilized lectins have now been incorporated into affinity surfaces that can be used to isolate broad classes of samples for mass spectrometric analysis. A carbohydrate and a bacterial species that displays the carbohydrate binding motif were isolated and concentrated out of solutions containing salt, urea, buffers, and other contaminants that are deleterious to MALDI mass spectrometry. Concanavalin A was immobilized to a gold foil via a self-assembled monolayer. Samples in phosphate buffer or urine were applied to the capture surface and allowed to interact. The capture surface was then washed to remove salts and other unbound components and subjected to matrix-assisted laser desorption/ionization on a time-of-flight mass spectrometer. The lectin-derivatized surface allowed samples to be concentrated and readily characterized at relatively low levels. The ability to rapidly identify and characterize microorganisms is important in many practical situations. In actual clinical or environmental samples, the microorganisms will likely be present in low levels in a complex matrix that may contain salts, other proteins, and environmental debris, which may make direct mass spectrometric identification difficult or impossible due to signal suppression caused by these contaminating components. Obviously, a simple way to selectively capture and concentrate microorganisms before mass spectrometric analysis would be attractive. In recent years, there has been considerable interest in the coupling of affinity chromatographic techniques with MALDI mass spectrometry as a means to clean up samples. These techniques make an ideal partner for MALDI because the affinity ligand may be immobilized to a solid support, which can also serve as the mass spectrometer sample surface. The most common interaction employed with mass spectrometry has been antibody binding to protein antigens,1-4 although other protein-ligand interactions

(1) Papac, D. I.; Hoyes, J.; Tomer, K. B. Anal. Chem. 1994, 66, 2609-2613. (2) Brockman, A. H.; Orlando, R. Anal. Chem. 1995, 67, 4581-4585. (3) Nelson, R. W.; Krone, J. R.; Bieber, A. L.; Williams, P. Anal. Chem. 1995, 67, 1153-1158. (4) Liang, X.; Lubman, D.; Rossi, D. T.; Nordbloom, G. D.; Barksdale, C. M. Anal. Chem. 1998, 70, 498-503.

1460 Analytical Chemistry, Vol. 71, No. 7, April 1, 1999

such as nucleic acid binding5 and metal ion binding1 have also been employed. Several different coupling chemistries have been used to immobilize proteins to a gold surface, followed by direct MALDI/TOF analysis of captured molecules2.6 Another approach has employed immobilization of the ligand of interest with a linker which may be cleaved via the application of laser light during a LD or MALDI experiment.7,8 Also, a thin nitrocellulose film has been used to immobilize antibodies for use as a capture surface for MALDI.4 Antibodies are also useful for the capture of bacteria. In a clinical situation, bacteria would be present in plasma, urine, or mucus which would have to be extensively cleaned up, e.g., by centrifugal filtration, dialysis, or chromatography, before MALDIMS could be performed. When antibody-based affinity capture is coupled with MALDI, it has previously been shown that protein or peptide analytes present in whole blood,3 blood plasma,4 human tears,2 and ascitic fluid1 can be isolated and readily analyzed. However, the high specificity of antibodies may be a disadvantage when the objective is to isolate a large class or group of microorganisms from a sample. Another class of binding proteins that have been employed to isolate bacteria is the lectins, defined as carbohydrate-binding proteins of nonimmune origin.9 Lectins may interact with bacteria by binding to the myriad of carbohydrate structures present at the cell surface, e.g., techoic acids, lipopolysaccharides, and peptidoglycans. It has been reported that the lectin concanavalin A is reactive with a variety of Gram-negative bacterial species, presumably by binding to lipopolysaccharides at or near the cell surface.10 Since lectins are generally specific for a particular carbohydrate structural motif, many bacterial species may bind to a single lectin and a single species may bind a variety of lectins with different carbohydrate specificities. This latter characteristic has been used to differentiate species of Bacillus (5) Hutchens, T. W.; Yip, T.-T. Rapid Commun. Mass Spectrom. 1993, 7, 576580. (6) Brockman, A. H.; Orlando, R. Rapid Commun. Mass Spectrom. 1996, 10, 1688-1692. (7) Ching, J.; Voivodov, K. I.; Hutchens, T. W. Bioconjugate Chem. 1996, 7, 525-528. (8) Ching, J.; Voivodov, K. I.; Hutchens, T. W., Presented at the 44th ASMS Conference on Mass Spectrometry and Allied Topics, Portland, OR, 1996. (9) Lis, H.; Sharon, N. Chem. Rev. 1998, 98, 637-674. (10) Pistole, T. G. Annu. Rev. Microbiol. 1981, 35, 85-112. 10.1021/ac981119h CCC: $18.00

© 1999 American Chemical Society Published on Web 02/20/1999

anthracis11 and Bacillus thuringiensis.12 Lectins have also been proposed to determine the presence of Neisseria gonorrheae in clinical isolates.13 A limitation of these early applications was the reliance on traditional detection systems such as agglutination or fluorescence, both of which may produce false positive results due to nonspecific agglutination or high background fluorescence. In this work, we combine lectin binding with MALDI mass spectrometry. The use of lectins allows the isolation of microorganisms from sample matrixes which may be prohibitive for direct MALDI analysis of phospholipid or other biomarkers.14,15 Lectins are first immobilized to gold foils, which serve as the sample support for MALDI. Bacterial samples are then applied to the probes and allowed to incubate. The probe may then be washed to remove interfering sample components. A MALDI matrix solution is applied, and MALDI/TOF analysis is performed. Samples present in physiological buffer and urine may be affinity captured directly on the MALDI probe and readily analyzed with minimal sample preparation. We also demonstrate the ability of the lectin probes to capture high-mannose complex carbohydrate samples from buffers. EXPERIMENTAL SECTION Preparation of Lectin Affinity Probes. The affinity probes were produced using a previously reported procedure2.16 Briefly, gold foil (99.9% pure, 0.1-mm thickness, Aldrich Chemical Co., Milwaukee, WI) was cut to approximately the same size as the sample well of our MALDI mass spectrometer. The foil was extensively washed with 2-propanol (Fisher Scientific, Pittsburgh, PA) which had been dried over molecular sieves before use. The probes were then immersed in a saturated solution of dithiobis(succinimidylpropionate) (Pierce, Rockford, IL) in dry 2-propanol for 20 min with gentle rocking. The probes were then repeatedly washed with dry 2-propanol, followed by dry ethanol. The probes were then immersed into a 20 µg/mL solution of the lectin concanavalin A (ConA) in “binding buffer” (100 mM potassium phosphate, pH 7.4, containing ∼100 µM Ca2+ (CaCl2)) for 20 min with gentle rocking. An excess of lectin was used in order to ensure maximal binding to the activated surface. Following the coupling reaction, the probes were extensively washed with binding buffer, followed by washes with polished water. The probes were then stored at 5 °C. Bacteria Growth and Sample Preparation. The bacteria used in this study, Escherichia coli (ATCC 11775), was grown by conventional methods in nutrient broth with aeration for 24 h at 37 °C. Cultures were harvested by chilling in an ice bath followed by centrifugation. The supernatant liquid was discarded and the (11) Cole, H. B.; Ezzell, J. W., Jr.; Keller, K. F.; Doyle, R. J. J. Clin. Microbiol. 1984, 19, 48-53. (12) DeLucca, A. J. Can. J. Microbiol. 1984, 30, 1100-1104. (13) Schaffer, R. L.; Keller, K. F.; Doyle, R. J. J. Clin. Microbiol. 1979, 10, 669672. (14) Heller, D. N.; Fenselau, C.; Cotter, R. J.; Demirev, P.; Oltoff, J. K.; Honovich, J.; Uy, O. M.; Tanaka, T.; Kishimoto, Y. Biochem. Biophys. Res. Commun. 1987, 142, 194-199. (15) Fenselau, C.; Fabris, D.; Hathout, Y.; Ho, Y.-P.; Thomas, J.; Cotter, R.; Bryden, W. In Proceedings of the First Joint Services Workshop On Biological Mass Spectrometry; U.S. Army Research and Development Center, Edgewood, MD.; U.S. Army: Edgewood, MD, 1997; pp 25-29. (16) Dogurel, D.; Williams, P.; Nelson, R. W. Anal. Chem. 1995, 67, 43434348.

pellet washed three times with polished water, followed by lyophilization. Oligomannose 9, a high-mannose complex carbohydrate (Oxford Glycosystems, Abingdon, UK) was initially used as a test substrate for the concanavalin A affinity probes. The sample was dissolved to a concentration of 200 µg/mL in the binding buffer solution. Several dilutions of this solution were made in the same buffer for sensitivity studies. Samples of E. coli were spiked into human urine to a concentration of 5 mg of dry cells/mL, corresponding to a cell count of ∼109 cells/mL, as estimated by the use of a PetroffHauser counting chamber (Fisher Scientific). For sensitivity studies, dilutions were also made of this stock suspension with unspiked urine. All samples were vortexed thoroughly before dilution, to ensure a homogeneous suspension of bacteria. Affinity Capture and Mass Spectrometry. The solutions were then applied to the lectin affinity probes, which had been affixed to the MALDI sample slide with double-sided tape. A 0.5µL aliquot of solution, plus an equal volume of binding buffer, was applied to the probe (to ensure optimal binding pH). The MALDI slide was then placed in a humid enclosure and allowed to incubate at room temperature for 2 h. The long incubation time was used to ensure maximal binding of the bacteria to the lectin and was used for all experiments reported here. After incubation, the probe surface was washed extensively with polished water and 0.5 µL of MALDI matrix solution (2,5-dihydroxybenzoic acid, 50 mM in 70:30 (v/v) acetonitrile/water and 0.1% trifluoroacetic acid) was applied to the probe. Appropriate controls were run using unmodified gold as the capture surface. MALDI time-of-flight mass spectra were run on a Kratos (Manchester, UK) Kompact MALDI 4 TOF mass spectrometer equipped with pulsed ion extraction. A nitrogen laser operating at 337 nm was used to achieve MALDI. Spectra were run in positive ion mode with a linear flight path and an extraction voltage of -20 kV. Spectra were typically the average of 100 laser shots taken across the probe. The instrument was calibrated with signals of the positive MH+ ion of substance P (mass 1348.6 Da) or melittin (mass 2847.5 Da) and the sodium ion (mass 22.98 Da). For accurate calibration, the standards were deposited on unmodified pieces of gold foil. Electron Microscopy. Electron microscopy was performed using an Amray model 1820D scanning electron microscope. Samples of E.coli spiked into urine were subjected to affinity capture on both lectin-derivatized and plain gold probe surfaces. The captured samples were washed by the identical procedure used for the mass spectrometric studies, with the exception that MALDI matrix was not applied. The samples were introduced into the electron microscope with no further sample preparation. RESULTS AND DISCUSSION Sequentially diluted solutions of the high-mannose complex carbohydrate oligomannose 9, dissolved in 100 mM potassium phosphate buffer, were subjected to affinity capture with a concanavalin A affinity probe and washed. This compound was selected in order to examine whether the lectin probes would function properly. Since ConA is reactive toward mannose residues, this carbohydrate should bind readily to this lectin. The mass spectra depicted in Figure 1a,b show that this analyte is Analytical Chemistry, Vol. 71, No. 7, April 1, 1999

1461

Figure 1. MALDI mass spectra obtained of oligomannose 9 (M + Na 1906 Da, M + K 1923 Da) captured from phosphate buffer solution with a concanavalin A probe: (a) 58 pmol, (b) 580 fmol, and (c) control on unmodified gold (58 pmol applied).

detectible to a level of 2 µg/mL, which corresponds to a maximum 580 fmol of analyte retained on the probe after washing. Therelatively low resolution in the spectra results from the use of relatively high laser power to desorb the sample and analysis in the linear mode.17 A control experiment was carried out, in which the same amount of carbohydrate was loaded onto an unmodified gold surface and subjected to the identical capture and wash procedure (Figure 1c). No signal was detected for the carbohydrate on the washed surface. Having succeeded with this compound, we sought to extend the use of the lectin probes for the isolation of bacteria from physiologic samples. E. coli, a common urinary tract pathogen, was spiked into a sample of human urine, which was incubated with a ConA-derivatized probe and washed as described in the Experimental Section. The results obtained are depicted in Figure 2. An array of phospholipid peaks is observed corresponding to (17) Cotter, R. J. Time-of-Flight Mass Spectrometry: Instrumentation and Applications in Biological Research; American Chemical Society: Washington, DC, 1997.

1462 Analytical Chemistry, Vol. 71, No. 7, April 1, 1999

Figure 2. MALDI mass spectra of E. coli 11775 spiked into urine and captured with a concanavalin A probe: (a) 0.5 µL of stock solution (∼109 cells/mL), (b) 1:100 dilution, and (c) control on unmodified gold (0.5 µL applied). Peak assignments are given as PE(x:y), where x denotes the number of carbon atoms in the fatty acid side chains and y the number of unsaturated bonds. PE, phosphatidylethanolamine; PHG, polar headgroup.

various species of phosphatidylethanolamine, which have previously been shown to be biomarkers for Gram-negative bacteria analyzed by desorption mass spectrometry.14,15,18,19 The limit of detection was observed with a 100-fold dilution of the stock bacterial solution (Figure 2b) and corresponds to ∼5000 bacterial cells applied on the capture surface, based on the estimated cell count in the stock solution (see Experimental Section). This is comparable to detection limits observed in our laboratory with purified bacterial preparations, analyzed by “conventional” dried droplet MALDI. As with the carbohydrate samples, biomarker signals were not observed when an aliquot was applied to an unmodified gold surface (Figure 2c). (18) Ho, Y.-P.; Fenselau, C. Anal. Chem. 1998, 70, 4890-4895. (19) Fenselau, C., Ed. Mass Spectrometry for Characterization of Microorganisms; ACS Symposium Series 541; American Chemical Society: Washington, DC, 1994.

Electron microscopy was used to confirm the presence or absence of bacteria on lectin-activated biocapture surfaces. The electron micrographs obtained revealed that the lectin-modified probe surface was covered with large patches of agglutinated bacteria, while only scattered individual cells were present on the plain gold surface. This is consistent with the MALDI biomarker analyses. CONCLUSIONS The work presented here illustrates the feasibility of employing lectin-modified, gold-surfaced mass spectrometer probes to capture carbohydrate or microbial samples from complex sample matrixes prior to mass spectrometric analysis. The technique could be extended to glycoproteins, glycolipids, and other samples displaying the appropriate carbohydrate moieties. Experimental evidence indicates that the binding is specific and not the result of nonspecific interactions with the gold surface. Target analytes may be isolated, concentrated, and analyzed from solutions that

contain high concentrations of other chemicals that would degrade direct mass spectrometric analysis. Sensitivity is estimated at ∼5000 cells of E. coli. ACKNOWLEDGMENT The authors acknowledge Brandon Falk and Tim Maugel for expert technical assistance. Scanning electron microscopy was performed at the Laboratory for Biological Ultrastructure, a core facility of the College of Life Sciences at the University of Maryland. This work was supported by a contract from the Applied Physics Laboratory of Johns Hopkins University, a grant from the Joint Institute for Food Safety and Nutrition of the Food and Drug Administration/University of Maryland, and a training grant from the Department of the Army. Received for review October 12, 1998. Accepted January 11, 1999. AC981119H

Analytical Chemistry, Vol. 71, No. 7, April 1, 1999

1463