Continuous-Flow, On-Line Monitoring of ... - ACS Publications

Continuous-Flow, On-Line Monitoring of Biospecific. Interactions Using Electrospray Mass. Spectrometry. A. C. Hogenboom,* A. R. de Boer, R. J. E. Derk...
0 downloads 0 Views 95KB Size
Download PDF
Anal. Chem. 2001, 73, 3816-3823

Continuous-Flow, On-Line Monitoring of Biospecific Interactions Using Electrospray Mass Spectrometry A. C. Hogenboom,* A. R. de Boer, R. J. E. Derks, and H. Irth

Faculty of Sciences, Division of Chemistry, Department of Analytical Chemistry and Applied Spectroscopy, Vrije Universiteit Amsterdam, De Boelelaan 1083, 1081 HV Amsterdam, The Netherlands

A continuous-flow analytical screening system is presented using electrospray mass spectrometry to measure the interaction of biologically active compounds with soluble affinity proteins. The biochemical detection system is based on a solution-phase, homogeneous assay. In a first step, compounds to be screened (e.g., biotinylated compounds, concentration range 10-1000 nmol/L) are injected into a continuous-flow reaction system and allowed to react with the affinity protein (e.g., streptavidin, concentration range 3-48 nmol/L). Subsequently, a reporter ligand (fluorescein-labeled biotin 96 nmol/L) is added to saturate the remaining free binding sites of the affinity protein and the concentration of unbound reporter ligand is measured using electrospray MS in the selectedion monitoring mode. The presence of active compounds in the sample results in an increase of the concentration of unbound reporter ligands. The feasibility of a homogeneous MS-based biochemical assay is demonstrated using streptavidin/biotin and anti-digoxigenin/digoxin as model systems. Compared to radioactive or fluorescence-based biochemical assays, the present assay format does not require the synthesis and purification of labels. Various analytical conditions were investigated to determine the ability of MS to measure the biochemical interactions. The availability of a single ligand that can be detected at 1050 nmol/L concentrations by electrospray MS is sufficient to set up the biochemical assay. For the biospecific interactions studies, detection limits of 10-100 nmol/L were obtained. The measurement of biochemical interactions has gained enormous importance in various areas of analytical chemistry. With the progress made in recombination production, more and more biological relevant proteins such as receptors, enzymes, or antibodies become available to scientists. The measurement of the interaction of small or large molecules with important biomolecular targets provides information on the potential biological activity of these compounds. Such measurements are highly relevant for example in drug discovery, but also in areas such as environmental toxicology and monitoring. Several analytical approaches have been chosen to characterize the interaction of ligands (analytes) with biomolecular targets. Affinity chromatography has been used in various forms for the 3816 Analytical Chemistry, Vol. 73, No. 16, August 15, 2001

selective preconcentration and separation of analytes based on the interaction with immobilized antibodies or receptors.1-6 Affinity SPE columns are often used in conjunction with liquid chromatography-mass spectrometry (LC-MS), allowing the determination of the molecular mass of active compounds. Immobilization of proteins, however, may lead to inactivation of the binding sites and, consequently, to a loss of binding affinity. Since the detection properties of the analyte(s) are not altered during affinity chromatography or affinity SPE, it might still be difficult to detect trace concentrations of biologically active compounds. To improve both the selectivity and sensitivity, the implementation of biochemical assays is an alternative approach to the analytical measurement of biochemical interactions. In biochemical assays, the presence of biologically active substances is detected indirectly via reporter molecules such as fluorescent, radioactive, or enzyme-labeled ligands. Since the reporter ligand can be detected at high sensitivity, detection limits of biochemical assays are in the picomole to nanomole per liter range. Biochemical assays are typically performed in batch, often in microtiter plates; however, in recent years, several concepts for the implementation in analytical techniques have been presented. Most notably, affinity electrophoresis7-13 and on-line biochemical assays14-20 have been developed to integrate biochemical interac(1) Bean, K. A.; Henion, J. D. J. Chromatogr., A 1997, 791, 119. (2) Nedved, M. L.; HabibiGoudarzi, S.; Ganem, B.; Henion, J. D. Anal. Chem. 1996, 68, 4228. (3) Riggin, A.; Sportsman, J. R.; Regnier, F. E. J. Chromatogr. 1993, 632, 37. (4) Farjam, A.; Brugman, A. E.; Soldaat, A.; Timmerman, P.; Lingeman, H.; de Jong, G. J.; Frei, R. W.; Th. Brinkman, U. A. Chromatographia 1991, 31, 469. (5) Pichon, V.; Chen, L.; Hennion, M.-C.; Daniel, R.; Martel, A.; Le Goffic, F.; Abian, J.; Barcelo´, D. Anal. Chem. 1995, 67, 2451. (6) Hage, D. J. Chromatogr., A 1998, 795, 185. (7) Schmalzing, D.; Nashabeh, W.; Yao, X. W.; Mhatre, R.; Regnier, F. E.; Afeyan, N. B.; Fuchs, M. Anal. Chem. 1995, 67, 606. (8) Mito, E.; Zhang, Y.; Esquivel, S.; Gomez, F. A. Anal. Biochem. 2000, 280, 209. (9) Zhang, Y.; Gomez, F. A. J. Chromatogr., A 2000, 897, 339. (10) Zhao, Y.; Becu, C.; Borremans, F.; Sandra, P. Electrophoresis 1999, 20, 2462. (11) Heintz, J.; Hernandez, M.; Gomez, F. A. J. Chromatogr., A 1999, 840, 261. (12) Heegaard, N. H. H.; Nilsson, S.; Guzman, N. A. J. Chromatogr., B 1998, 715, 29. (13) Chu, Y. H.; Cheng, C. C. Cell. Mol. Life Sci. 1998, 54, 663. (14) Irth, H.; Oosterkamp, A. J.; Tjaden, U. R.; van der Greef, J. Trends Anal. Chem. 1995, 14, 355. (15) Irth, H.; Oosterkamp, A. J.; van der Welle, W.; Tjaden, U. R.; van der Greef, J. J. Chromatogr. 1993, 633, 65. (16) van Bommel, M. R.; de Jong, A. P. J. M.; Tjaden, U. R.; Irth, H.; van der Greef, J. J. Chromatogr., A 2000, 886, 19. 10.1021/ac010026o CCC: $20.00

© 2001 American Chemical Society Published on Web 07/11/2001

tions into analytical technologies using high-sensitivity labels as reporter ligands. Next to fluorescence and radioacitivity detection, mass spectrometry has become an important detection technique in the measurement of biochemical interactions.21-34 The most widely used ionization technique has been electrospray ionization (ESI) due to the relatively soft nature of this electrostatic process enabling weak noncovalently bound complex protein interactions to be directly detected. ESI-MS observations for these weakly bound systems reflect to some extent the nature of the interaction found in the condensed phase.25 The stoichiometry of the complex is obtained from the resulting mass spectrum because the molecular weight of the complex is directly measured.26,27 MS is a powerful technique due to the advantages afforded in terms of speed of analysis and sensitivity. The observation of ESI-MS applicability in this field was first described by Katta and Chait.28 in their studies on globin-heme interaction of myoglobin and the receptor-ligand complex by Ganem et al.29 Following these initial reports, several other types of protein-ligand binding were studied including enzyme-substrate pairings,30 protein-cofactor,31 and protein-DNA complexes.32 Careful experimental consideration should be taken when these studies are attempted. In other words, ESI interface conditions typically need to be as gentle as possible to maintain the intact complex.26 Alternatively to measuring the intact protein-ligand complexes, analytical techniques have been reported that are based on the separation of protein-ligand complexes from unbound, nonactive compounds by size exclusion chromatography and related techniques, the subsequent dissociation of the complexes, and the measurement of the dissociated active ligands by LCMS.35-40 These techniques are mostly applied in drug discovery, for example, in the screening of combinatorial libraries. (17) Emneus, J.; Marko-Varga, G. J. Chromatogr., A 1995, 703, 191. (18) Onnerfjord, P.; Eremin, S. A.; Emneus, J.; Marko-Varga, G. J. Chromatogr., A 1998, 800, 219. (19) Miller, K. J.; Herman, A. C. Anal. Chem. 1996, 68, 3077. (20) Graefe, K. A.; Tang, Z.; Karnes, H. T. J. Chromatogr., B 2000, 745, 305. (21) Natsume, T.; Nakayama, H.; Jansson, O.; Isobe, T.; Takio, K.; Mikoshiba, K. Anal. Chem. 2000, 72, 4193. (22) Sannes-Lowery, K. A.; Griffey, R. H.; Hofstadler, S. A. Anal. Biochem. 2000, 280, 264. (23) Veenstra, T. D. Biophys. Chem. 1999, 79, 63. (24) Pramanik, B. N.; Bartner, P. L.; Mirza, U. A.; Liu, Y. H.; Ganguly, A. K. J. Mass Spectrom. 1998, 33, 911. (25) Loo, J. A. Mass Spectrom. Rev. 1997, 16, 1. (26) Smith, R. D.; Light-Wahl, K. J. Biol. Mass Spectrom. 1993, 22, 493. (27) Light-Wahl, K. J.; Schwartz, B. L.; Smith, R. D. J. Am. Chem. Soc. 1994, 116, 5271. (28) Katta, V.; Chait, B. T. J. Am. Chem. Soc. 1993, 115, 6317. (29) Ganem, B.; Li, Y.-T.; Henion, J. J. Am. Chem. Soc. 1991, 113, 6294. (30) Ganem, B.; Li, Y.-T.; Henion, J. J. Am. Chem. Soc. 1991, 113, 7818. (31) Drummond, J. T.; Ogorzalek Loo, R. R.; Matthews, R. G. Biochemistry 1993, 32, 9282. (32) Greig, M. J.; Gaus, H.; Cummins, L. L.; Sasmor, H.; Griffey, R. H. J. Am. Chem. Soc. 1995, 117, 10765. (33) Schwartz, B. L.; Gale, D. C.; Smith, R. D.; Chilkoti, A.; Stayton, P. S. J. Mass Spectrom. 1995, 30, 1102. (34) Eckart, K.; Spies, J. J. Am. Soc. Mass Spectrom. 1995, 6, 912. (35) Davis, R. G.; Anderegg, R. J.; Blanchard, S. G. Tetrahedron 1999, 55, 11653. (36) Blom, K. F.; Larsen, B. S.; McEwen, C. N. J. Comb. Chem. 1999, 1, 82. (37) Huang, P. Y.; Carbonell, R. G. Biotechnol. Bioeng. 1999, 63, 633. (38) Kaur, S.; McGuire, L.; Tang, D. Z.; Dollinger, G.; Huebner, V. J. Protein Chem. 1997, 16, 505. (39) Kelly, M. A.; Liang, H. B.; Sytwu, I. I.; Vlattas, I.;-A Lyons, N. L.; Bowen, B. R.; Wennogle, L. P. Biochemistry 1996, 35, 11747. (40) van Elswijk, D.; van der Greef, J.; Irth, H. J. Mass Spectrom., submitted for publication.

In the present paper, we are demonstrating the use of ESIMS as a detection method in homogeneous biochemical assays. Rather than measuring protein-ligand complexes or dissociated ligands directly, a reporter ligand is employed to indirectly measure the interaction of active analytes with biomolecular targets. This approach is similar to fluorescence- or radioactivitybased biochemical assays but uses ESI-MS to detect the reporter ligand. The reporter ligand selected for the biochemical assays preferably has a high affinity for the biomolecular target to be screened and can be detected with high sensitivity using ESIMS. In contrast to biochemical assays based on fluorescent- or radioactive-labeled ligands, there is no need to synthesize labels since known, active ligands can be employed in the biochemical assay, significantly reducing the assay development time. This is particularly advantageous for those affinity proteins, where the chemical modification of known active ligands, e.g., the attachment of a fluorophor, leads to a dramatic increase of binding affinity. The concept proposed is similar to the biochemical assays based on fluorescence detection.14 The assay is carried out in a continuous-flow postcolumn reaction detection system where the affinity protein and reporter ligand are sequentially added to the carrier solution of a flow injection system or the eluate of an LC separation column. The reaction mixture is directed toward an ESI-MS which monitors the concentration of the MS reporter ligand. The presence of active compounds in the sample will result in an increase of the free, unbound concentration of the reporter ligand measured by MS in the selected-ion monitoring (SIM) mode. In parallel, the molecular mass of the active ligand can be established by interpreting the full-scan trace of the chromatogram. By employing MS as detection technique for biochemical assays, the analytical methods developed simultaneously provide biochemical (binding affinity) and chemical information (molecular mass) and may substantially reduce the time to discover and characterize biologically active compounds in areas such as combinatorial chemistry and natural product screening. EXPERIMENTAL SECTION Materials. Avidin, streptavidin, biotin, fluorescein-labeled biotin (fluorescein-biotin), digoxigenin, and digoxin were purchased from Sigma (St. Louis, MO). N-Biotinyl-6-aminocaproic hydrazide, N-biotin hydrazide, N-biotinyl-L-lysine, and biotin-Nsuccinimidyl ester were from Fluka Chemie (Buchs, Germany). Anti-digoxigenin (FAB fragments) was purchased from Roche Diagnostics GmbH (Mannheim, Germany). Sodium phosphate, ammonium acetate, ammonium formate, HEPES, Gly-Gly, triethanolamine, acetic acid, and formic acid were from Sigma. Acetonitrile and methanol were obtained at Baker (Deventer, The Netherlands) and were of HPLC-grade. Binding buffer consisted of either sodium phosphate, ammonium acetate, or ammonium formate (10 mmol/L, pH 7.5). The carrier solution is binding buffer/organic modifier (90:10%, v/v). All the stock solutions of biotin, fluorescein-biotin, digoxin, and digoxigenin were prepared in methanol. Dilutions were made in the carrier solution. Stock solutions and dilutions of avidin, streptavidin, and anti-digoxigenin were prepared in binding buffer. Apparatus. The flow injection (FI) system (Figure 1) consisted of three Shimadzu (Kyoto, Japan) LC-10Ai pumps used to deliver the carrier solution/eluent, the reporter ligand solution, and the affinity protein solution. Sample injection was performed with a Analytical Chemistry, Vol. 73, No. 16, August 15, 2001

3817

Figure 1. Scheme of the on-line continuous-flow biochemical detection system: 1, carrier solution; 2, reagent pump with protein or antibody in carrier solution; 3, reaction coil; 4, reagent pump with reporter ligand in carrier solution; 5, triple-quadrupole MS or Q-TOF MS-MS.

Gilson (Villiers-le-Bel, France) 234 autosampler equipped with a Rheodyne six-port injection valve (injection loop, 20 µL). A PerkinElmer (Applera, Norwalk, CT) LS-4 fluorescence spectrometer and a Micromass (Wythenshawe, Manchester, U.K.) Quattro II triplequadrupole mass spectrometer (MS-MS) equipped with an ESI source were used for detection. Masslynx software (version 2.22) running under Windows NT was used for control of the system and data acquisition. Full-scan sensitivity may be a limiting factor for future development; therefore, preliminary studies into the utility of a novel orthogonal-acceleration time-of-flight mass spectrometer were investigated. A Micromass Q-TOF MS-MS instrument equipped with a Micromass Z-spray electrospray source was used in the anti-digoxigenin/digoxin experiments. Instrument control, data acquisition, and data processing were done using Micromass Masslynx software (version 3.4) running under Windows NT. The FI carrier solution consisted of binding buffer/organic modifier (90:10%, v/v) and was pumped at a flow rate of 50 µL/ min. The different avidin/streptavidin concentrations (in nmol/ L) and fluorescein-biotin solutions were prepared in binding buffer/organic modifier and added to the carrier solution via an inverted Y-type mixing union. The avidin/streptavidin and fluorescein-biotin solution were pumped at a flow rate of 50 and 100 µL/min, respectively. Knitted 300-µm-i.d. poly(tetrafluoroethylene) reaction coils with internal volumes of 17 and 33 µL for reactions I and II, respectively, were used. In the anti-digoxigenin/digoxin experiments, reaction coils with internal volume of 65 and 33 µL for reactions I and II, respectively, were used. The reactions were performed at room temperature (20 °C). When MS and fluorescence detection were used in parallel, another Y union was inserted after the second reaction coil to split the flow rate to 180 µL/min entering the fluorescence detector and 20 µL/min entering the ESI source. Fluorescence detection was performed at an excitation wavelength of 486 nm and an emission wavelength of 520 nm. Mass spectrometry was operated in SIM in the positive-ion (PI) mode on [M + H]+ and on one or more intense fragment ions (Quattro II). The ESI source conditions (on Quattro II) for the streptavidin/biotin and the anti-digoxigenin/digoxin system were as follows: source temperature, 80 °C; capillary voltage, 3.2 3818

Analytical Chemistry, Vol. 73, No. 16, August 15, 2001

kV. The sample cone voltage was 30 and 60 V for biotin and fluorescein-biotin, respectively. For both digoxigenin and digoxin, the sample cone was 20 V. The ESI source conditions (on the Q-TOF) for the anti-digoxigenin/digoxin system were as follows: source temperature, 80 °C; desolvation temperature, 120 °C; capillary voltage, 3.0 kV. The sampling cone and extraction cone voltage were 15 and 0 V, respectively. In Q-TOF MS experiments, mass spectrometry was operated at a 20-kHz frequency with a spectrum integration time of 5 s in “full-scan” MS in the PI mode in the range m/z 350-850 (“interscan” time 0.1 s). Nitrogen (99.999% purity; Praxair, Oevel, Belgium) was used with flow rates of 20 (for nebulization) and 350 L/h for (for drying/ desolvation). Argon (99.9995% purity; Praxair) was used for MSMS experiments. Batch Experiments. The performance of the biochemical assay was tested under equilibrium conditions and in the FI mode with fluorescence detection.41 To model physiological systems, phosphate buffer is generally used as binding buffer. However, phosphate buffers are not amenable to MS interfaces. To select a carrier solution composition that would be compatible with MS, two aqueous buffer solutions were selected, i.e., ammonium acetate and ammonium formate, to test how the biospecific interaction will behave compared to conventionally used phosphate buffer. In all batch experiments, carrier solution compositions of buffer (10 mmol/L; pH 7.5)/methanol (90:10%, v/v) were used and FI (20-µL loop) were performed with fluorescence detection. In the first batch experiment, fluorescein-biotin (36 nmol/L) is injected into the FI-MS system. In the second experiment, avidin (6 nmol/L) was incubated for 15 min with fluorescein-biotin (36 nmol/L). In the third batch, avidin (6 nmol/L) was first incubated with fluorescein-biotin (36 nmol/L) and after 15 min biotin (600 nmol/L) was added. In the last experiment, first avidin (6 nmol/ L) was first incubated with biotin (600 nmol/L) and after 15 min fluorescein-biotin (36 nmol/L) was added. Similar batch experiments were performed for the antidigoxigenin/digoxin system. In the first batch experiment, digoxin (concentration range: 0, 50, 100, 200, 300 nmol/L) was injected into the FI-MS system to obtain calibration curves (n ) 3). In a second experiment, the various concentration levels of digoxin were incubated with 200 nmol/L anti-digoxigenin and injected to assess the anti-digoxigenin/digoxin interaction. In the third experiment, anti-digoxigenin (200 nmol/L) was first incubated with a large excess (2000 nmol/L) of digoxigenin and later digoxin (in same concentration range as mentioned above) was added and injected into FI-MS system The above-mentioned three experiments were also performed for digoxigenin in various concentrations. RESULTS AND DISCUSSION Assay Setup. A homogeneous, on-line continuous-flow biochemical system based on fluorescence detection was reported by Oosterkamp et al.41 In the present paper, a similar continuousflow analytical screening system is described using MS to measure the interaction of the analyte(s) with an affinity protein such as an antibody, receptor, or enzyme. Figure 1 shows the FI system used for this purpose. In a first step, the sample is injected into a continuous-flow reaction system and allowed to react with the (41) Oosterkamp, A. J.; Villaverde Herraiz, M. T.; Irth, H.; Tjaden, U. R.; van der Greef, J. Anal.Chem. 1996, 68, 1201.

affinity protein (reaction 1) for 10-20 s. In the second step, a reporter ligand is added to saturate the remaining free binding sites of the affinity protein (reaction 2). The reaction time is 1020 s and depends mainly on the binding constant of reporter ligand-affinity protein complex (RP). The reaction time is chosen in a way that the association of free affinity protein molecules with the reporter ligand is favored whereas the dissociation of the analyte affinity protein complex (AP) is negligible. Finally, the concentration of free reporter ligand is detected by using ESIMS in the SIM mode. Generally, in biochemical analysis, a phosphate buffer is used to mimic physiological conditions (pH of ∼7.5). The percentage of organic modifier is usually kept as low as possible to prevent denaturation of the proteins. In addition, a blocking reagent, such as Tween-20 is added to prevent nonspecific binding of the protein (and protein-ligand complex) to the surface of reaction capillaries. However, nonvolatile additives in the eluent, such as phosphate buffer and blocking reagent, are not compatible with MS detection. Various reaction conditions were monitored using a series of MScompatible solvents and compared with the responses observed in the fluorescence detection. Optimization of MS Conditions. MS Response and MSCompatible Buffers. Different organic and inorganic buffers, such as ammonium acetate, ammonium formate, HEPES, Gly-Gly, and triethanolamine, were selected to study the response of biotin and fluorescein-biotin in MS and compared to phosphate buffer. Biotin and fluorescein-biotin were dissolved in the carrier solution compositions of buffer (10 mmol/L; pH 7.5)/methanol (50:50%, v/v) at concentrations of 10 ng/µL. Both infusion and 20-µL loop injection experiments were performed with detection in MS in full-scan and SIM mode. Main optimization criteria are the maximum response of biotin and fluorescein-biotin with lowest interference of the carrier solution. HEPES, Gly-Gly, and triethanolamine give very high background response, which significantly hampers the detection of biotin and fluorescein-biotin. Phosphate buffer and ammonium acetate/ammonium formate give a factor 10 and 100 less background response, respectively. As regards sensitivity, ammonium acetate and ammonium formate gave the highest response for biotin and fluorescein-biotin. All stock solutions were prepared in methanol (biotin/fluorescein-biotin) or binding buffer (protein). MS Response and Organic Modifier. To select a carrier solution composition that would provide an overall maximum response for MS detection, two modifiers were selected, acetonitrile and methanol, and two buffers, i.e. ammonium acetate (10 mmol/L; pH 7.5) and ammonium formate (10 mmol/L; pH 7.5). Biotin and fluorescein-biotin were dissolved in various binding buffer/organic solvent mixtures ranging from 90:10 (v/v) to 50:50% (v/v) at two concentration levels (0.01 and 1 ng/µL), and 20 µL was injected and analyzed by MS in the full-scan and SIM mode. The maximum response was found with 50% methanol, which was a factor ∼2 higher than for 10% methanol. Since the proteins can denaturate or protein-ligand complexes can dissociate at relatively low percentages of organic modifier, in further experiments only 10% methanol is used in the carrier solution. Behavior of Biochemical Assay Using Different Buffers. For studying the behavior of the biochemical interaction, two MS buffers were selected, ammonium acetate and ammonium formate

Figure 2. On-line continuous-flow monitoring of biochemical interaction with (A) fluorescence and (B) MS SIM (m/z 390) detection. Fluorescein-biotin (96 nmol/L), streptavidin (32 nmol/L), 20-µL loop injections of 1000 nmol/L biotin (n ) 3).

(10 mmol/L; pH 7.5), which gave the highest response in MS experiments. The FI biochemical fluorescence-based detection system was used to test the behavior of the biochemical assay under different buffer conditions. The results were compared with phosphate buffer. Four batch experiments (n ) 3) were performed using each of the three buffers (for details, see Experimental Section). First, the response of fluorescein-biotin incubated with avidin (second batch experiment) was compared with standard injections of fluorescein-biotin. The response of free fluoresceinbiotin decreases by a factor of ∼2 due to complexation with avidin. In the third batch experiment, avidin was first incubated with fluorescein-biotin and later a large amount of biotin was added. No significant exchange of biotin for fluorescein-biotin is observed; hence, a similar response as for the second batch experiment was observed. In the last experiment, biotin was incubated with avidin and later fluorescein-biotin was added. The increase in response due to increase in concentration of free fluorescein-biotin was almost equal to injection of fluorescein-biotin. More importantly, these experiments clearly show that when compatible MS buffers are used in the biochemical assay with fluorescence detection, no marked change in the response based on buffers used is observed. For the detection of biospecific interaction with MS, either ammonium acetate or ammonium formate can be used. It should also be noted that no blocking reagents were used in the experiments. One has to keep in mind that this is not a generic solution/method. Every biochemical assay (ligand-protein/ antibody-antigen interaction) can behave differently. On-Line Continuous-Flow Biochemical Interaction. Figure 2 illustrates the basic performance of the on-line MS assay. For comparison, a homogeneous fluorescence assay has been set up in parallel. For this purpose, the carrier flow was split after reaction 2, 90% of the total flow being directed to a fluorescence detector (Figure 2A) and 10% to the MS (Figure 2B). The affinity interaction between streptavidin and biotin was chosen to study the characteristics of an on-line MS biochemical assay. Fluorescein-biotin Analytical Chemistry, Vol. 73, No. 16, August 15, 2001

3819

Figure 3. Curve of ratio of released reporter ligand and total affinity protein concentration against concentration analyte using (A) fluorescence and (B) MS detection.

was used as reporter ligand for both fluorescence and MS in the SIM mode (m/z 390) detection. In the fluorescence mode, the homogeneous biochemical assay is based on the quenching of the fluorescein-biotin fluorescence upon binding to streptavidin. At point 1 in Figure 2, solely carrier solution is pumped by all pumps (carrier pump, affinity protein pump, reporter ligand pump) resulting in stable baseline in both detectors. At point 2, fluoresceinbiotin is added to the reporter ligand pump, leading to an increase of the background signal in both detectors. After stabilization of the system, streptavidin is added to the affinity protein pump (point 3). The reaction of streptavidin and fluorescein-biotin leads to an almost complete disappearance of free fluorescein-biotin and, consequently, to a reduction of the baseline to the original level. When injecting active analytes such as biotin (4), the concentration of free, unbound streptavidin is reduced in reaction 1 leading to an increase of the free fluorescein-biotin concentration after reaction 2 and a positive signal in both the MS and fluorescence detector. MS is shown to mimic the response patterns in the continuous-flow experiment similar to those observed with fluorescence detection. The decrease of the unbound fluoresceinbiotin concentration upon addition of streptavidin at point 3 indicates that complex formation occurs and that the fluorescein biotin-streptavidin complex does not dissociate during the ionization phase. Complete protein-ligand complexes have been reported to stay intact in the ESI-MS process; however, gentle experimental conditions should be applied.33,34 Furthermore, when 96 nM fluorescein-biotin and 32 nM streptavidin is used, an injection of 1 µmol/L biotin results in an almost complete blocking of streptavidin, and consequently, the maximum peak height possible under the current conditions is ∼95% of the highest point (3), indicating that apparent binding of biotin to streptavidin is on the order of >95%. In Figure 3, the ratio of released reporter ligand and total affinity protein concentration is plotted against the concentration of analyte injected. The concentration of released reporter ligand, i.e., fluorescein-biotin, was measured by both fluorescence and electrospray MS detection and determined by calibration curves that were recorded under assay conditions. Both curves are almost identical, indicating that the electrospray ionization process does 3820

Analytical Chemistry, Vol. 73, No. 16, August 15, 2001

not lead to a significant dissociation of the reporter ligand-affinity protein complex. It should be emphasized, however, that the dissociation behavior has to be studied for each affinity protein individually. Optimization of Continuous-Flow MS-Based Assay. With the above-mentioned continuous-flow setup with MS, analytical performance of the assay is optimized. Oosterkamp et al.41 reported that reaction time and affinity protein concentration are most relevant parameters to be optimized. Because the interaction between biotin and streptavidin is strong (Ka ) 0.6 × 1015 L/mol) with a relatively fast association rate (k+1 ) 2.4 × 107 L mol-1 s-1) and slow dissociation rate (k-1 ) 0.4 × 10-7 s-1), the reaction times are fast, i.e., 10-20 s. Hence, the reaction coil volumes were kept as small as possible to reduce band broadening, i.e., 17 and 33 µL, for coils I and II, respectively. Influence of Organic Modifier. One of the goals in mind is the use of LC combined with biochemical MS-based detection to monitor the presence of certain active compounds. In such a setup, gradient elution with high percentages of organic modifier, e.g., methanol or acetonitrile, to elute active compounds from the LC column are used. First, the maximum tolerated percentage of organic modifier in the eluent is determined for the performance of the biochemical assay. Different buffer/organic modifier carrier solution mixtures ranging from 95:5 (v/v) to 20:80% (v/v) are used in the biochemical MS-based method. Biotin is diluted in the same buffer/organic modifier carrier solution mixtures (at two concentration levels) and 20 µL is injected in the continuous-flow biochemical MS-based system. The presence of high percentages of organic modifier, e.g., 40 and 80% for acetonitrile and methanol, respectively, does not seem to hamper the interaction between biotin and streptavidin. However, with higher organic modifier percentages, a decrease of the MS response was observed. The overall response seems to be the same when using acetonitrile or methanol in the eluent. However, the responses for both biotin and fluorescein-biotin decrease more rapidly when acetonitrile is used compared to methanol. For example, for biotin, the MS response decreases only slightly when up to 60-80% methanol is used whereas the MS response decreases by a factor of 2 when the assay is performed in 0-30% acetonitrile. For fluoresceinbiotin, these effects are more pronounced. When methanol is used. a loss of response of a factor of 2 is observed when going from 0 to 80%. However, for acetonitrile the loss in response was a factor of 5 (from 0 to 30%). This behavior is consistent with results reported in the literature.15,18 Analytical Performance. One important parameter in the biochemical assay is the concentration of streptavidin. Increasing the streptavidin concentration should result in higher concentrations of both streptavidin/biotin and streptavidin/fluoresceinbiotin. The behavior of the MS response of fluorescein-biotin is observed at different streptavidin and biotin concentrations. Figure 4 presents experimental calibration curves (n ) 3) for biotin shown at four (3, 8, 13, and 20 nmol/L) streptavidin concentration levels. As expected, the response for fluorescein-biotin increased with increasing streptavidin concentration until it reached a saturation level. For example, for biospecific interaction studies using 20 nmol/L streptavidin, the fluorescein-biotin response increases linearly between 50 and 400 nmol/L biotin. When 20 nmol/L

Figure 4. MS SIM response of fluorescein-biotin vs biotin concentration at different streptavidin concentrations: (A) 3, (B) 8, (C) 13, and (D) 20 nmol/L. Fluorescein-biotin concentration, 96 nmol/L. For all other conditions, see Experimental Section.

streptavidin is used, biotin detection limits are on the order of 50 nmol/L. FI-MS of different concentration levels (0, 10, 50, 100, 200, 300, 400, 500, 750, and 1000 nmol/L) of biotin showed a linear response of fluorescein-biotin concentration (50-1000 nmol) versus MS response (in SIM mode, n ) 3). Detection limits were on the order of 10 nmol/L. For the total biochemical system, detection limits were ∼50 nmol/L for biotin. Monitoring Bioactive Compounds. The biochemical MS assay performance was studied for various biotin derivatives, such as biotin (m/z 245), N-biotinyl-6-aminocaproic hydrazide (m/z 372), biotin hydrazide (m/z 259), N-biotinyl-L-lysine (m/z 373), and biotin-N-succinimidyl ester (m/z 342). These five different bioactive compounds were consecutively injected into the biochemical MS assay. Figure 5 shows triplicate injections in the biochemical MS-based system of the different active compounds. Each compound binds to streptavidin; hence, the MS response of peaks of the reporter ligand (fluorescein-biotin, m/z 390) is similar. The use of SIM allows specific components to be selected and monitored, e.g., protonated molecule of the biotin derivatives. In this, case no peaks were observed for biotin-N-succinimidyl ester (m/z 342), because under the applied conditions fragmentation occurred to m/z 245. In combination with full-scan MS measurements, the molecular mass of active compounds can be determined simultaneously to the biochemical measurement. Antibody-Antigen Interactions. To assess the applicability of MS to study antibody-antigen interactions, we used a model system comprising FAB fragments of anti-digoxigenin antibodies. Digoxin and digoxigenin are ligands having approximately the same affinity for the anti-digoxigenin antibodies. Both compounds were used as analytes and reporter ligands. The same MS-based biochemical assay setup was used as for the streptavidin/biotin system. Because the interaction between anti-digoxigenin and digoxin is weaker (Ka ) ∼109 L/mol) with a relatively slower association rate and dissociation rate than streptavidin/biotin, a longer reaction time is preferred. Therefore, a reaction coil volume

of 65 µL was chosen for reaction I, resulting in a reaction time of 39 s. In a series of experiments, the binding of digoxin or digoxigenin to the anti-digoxigenin antibodies was studied. All incubations were performed in batch, and injections were carried out in the carrier solution with detection in MS (FI-MS). In a first experiment, digoxin and digoxigenin are measured at various concentration levels (0, 50, 100, 200, and 300 nmol/L) in the absence of antibodies to observe their response in MS using ammonium formate/methanol (90:10%, v/v) as carrier solution. Figure 6A represents a calibration curve (n ) 3) for digoxigenin. Detection limits of 10 nmol/L were obtained for both digoxin and digoxigenin. The interaction of digoxigenin with the anti-digoxigenin antibodies was measured in a second experiment. For this purpose, various concentrations of digoxigenin were incubated with 200 nmol/L anti-digoxigenin and injected into the FI-MS system (Figure 6B). The interaction was monitored by observing the response of digoxigenin in MS at m/z 408.4. In comparison with the calibration line obtained by injection of digoxigenin in the absence of antibodies, a significant decrease of the digoxigenin response was observed for all digoxigenin concentrations injected. To demonstrate that the decrease of the free digoxigenin concentration upon incubation with anti-digoxigenin antibodies is based on specific interactions, the same experiment was repeated, but anti-digoxigenin was first incubated with a large excess (2000 nmol/L) of digoxin, i.e., a competing ligand. The resulting curve (Figure 6C) is almost identical with the calibration for digoxigenin measured in the absence of antibodies, indicating that digoxigenin is prevented from binding to the antibody due to an excess of competing ligand. A similar behavior was observed when digoxin instead of digoxigenin was used as reporter ligand. Figure 7 demonstrates that both ligands can be employed as reporter ligand in an on-line continuous-flow assay setup (see Figure 1). In Figure 7A, digoxigenin is used as reporter ligand to measure the interaction of digoxin with anti-digoxigenin antibodAnalytical Chemistry, Vol. 73, No. 16, August 15, 2001

3821

Figure 5. On-line continuous-flow monitoring of bioactive compounds using fluorescein-biotin/streptavidin assay. Monitoring fluorescein-biotin (m/z 390) and triplicate injections (in order of elution) of biotin (m/z 245), N-biotinyl-6-aminocaproic hydrazide (m/z 372), biotin hydrazide (m/z 259), N-biotinyl-L-lysine (m/z 373), and biotin-N-succinimidyl ester (m/z 342). Fluorescein-biotin 96 nmol/L, streptavidin 32 nmol/L, and all injections are 1000 nmol/L. For conditions, see Experimental Section.

Figure 6. MS response of flow injection of (A) digoxigenin, (B) digoxigenin incubated with 200 nmol/L anti-digoxigenin, and (C) 200 nmol/L anti-digoxigenin incubated with 2000 nmol/L digoxin and digoxigenin added. Concentration levels digoxigenin: 0, 50, 100, 200, and 300 nmol/L. For conditions, see Experimental Section.

ies. Injections of digoxin (1) result in a release of digoxigenin which is measured at m/z 408.4. The detection for digoxin in this format is ∼50 nmol/L. In Figure 7B, digoxin is used as reporter ligand; injections of digoxigenin (2) lead to an increase of the free digoxin concentration measured at m/z 798.5. The corresponding detection limit for digoxigenin is ∼50 nmol/L. 3822 Analytical Chemistry, Vol. 73, No. 16, August 15, 2001

Figure 7. On-line continuous-flow monitoring of biochemical interaction using MS. (A) (200 nmol/L) Anti-digoxigenin/(100 nmol/L) digoxigenin interaction (m/z 408.4) and triplicate injections (1) of (1000 nmol/L) digoxin. (B) (200 nmol/L) anti-digoxigenin/(100 nmol/L) digoxin interaction (m/z 798.5) and triplicate injections (2) of (1000 nmol/L) digoxigenin.

These experiments clearly demonstrate that ESI-MS is suitable for monitoring antibody-antigen interactions by selectively detecting free ligand molecules in the presence of antibody-ligand complexes. Moreover, the development of MS-based biochemical assays is rather straightforward since any detectable analyte can principally be used as reporter ligand. The sensitivity of the biochemical assay depends mainly on the detection sensitivity of the reporter ligand and its binding affinity for the affinity protein.

Since digoxin and digoxigenin have similar binding affinities for the anti-digoxigenin antibodies, similar assay sensitivities are obtained when both compounds are used as reporter ligands. CONCLUSIONS The applicability of a homogeneous continuous-flow MS-based biochemical assay was demonstrated for the interaction between streptavidin and biotin and anti-digoxigenin and digoxin interactions. One of the major advantages of using MS versus fluorescence detection is clearly that the choice of reporter ligands is facilitated. Principally, for the development of a new assay, known high-affinity ligands for the affinity protein to be screened are analyzed by LC-MS, and the ligand with the highest sensitivity is used as reporter ligand. With this approach, assay development times can be substantially reduced; moreover, the development of biochemical assays might become possible for those receptors, where labeling of known ligand with a fluorophor or an enzyme results in a dramatic decrease of binding affinity and, consequently, of detection sensitivity. An important requirement for the assay format presented is that the biochemical interactions between affinity protein and ligands proceed under MS-compatible conditions. Compared to the conditions typically used for biochemical assays, volatile buffers such as ammonium acetate or formate buffer have to be used rather than phosphate buffer. Moreover, salts such as sodium chloride, which are often added to adjust the background ionic strength of the assay, have to be avoided. Detection limits of the assays described are in the order of 10-50 nmol/L for high-affinity ligands. In comparison to fluorescence-based assays for the same compounds, the detection limits of the MS-based assays are ∼10-fold higher. The assay detection limits are mainly dependent on the detection limits obtained for the reporter ligand used. Screening for the best reporter ligand for MS (the maximum response) and optimizing the MS conditions for this reporter molecule will improve detection limits.

Further improvements can be made by optimum carrier solution conditions. Furthermore, the use of nanoflow ESI-MS with its higher sensitivity will be an option. In future investigations, the range of affinities that can be measured with the present assay technology has to be determined using appropriate receptors. Similar to other biochemical assays based on MS, the methodology presented allows the simultaneous measurement of affinity and molecular mass, allowing the rapid characterization of active compounds based on a library search. In comparison to the parallel detection approach published earlier,42 the peaks of the active ligand and the reporter ligand have inherently identical retention times, facilitating the interpretation of MS and MSMS spectra. The assay format presented here has the potential of multiplexing, i.e., of performing several assays in parallel. Principally, it is necessary to pump mixtures of affinity proteins and the corresponding reporter ligands, respectively, rather than single species, and monitor several m/z traces at the same time. Clearly, this approach will only be feasible for those assays where no crossreactivity exists between affinity proteins. The most important application area of MS-based assays is in drug discovery, particularly in the screening of combinatorial chemistry libraries and natural products. As presented in earlier papers,14,41 the present MS-based assays can be coupled on-line to LC, allowing the efficient screening of mixtures. The possibility to simultaneously measure biochemical affinity and the molecular mass of an active compound might significantly reduce the time to discover and characterize novel biologically active compounds. Received for review January 8, 2001. Accepted May 10, 2001. AC010026O (42) Ingkaninan, K.; de Best, C. M.; van der Heijden, R.; Hofte, A. J. P.; Karabatak, B.; Irth, H.; Tjaden, U. R.; van der Greef, J.; Verpoorte, R. J. Chromatogr., A 2000, 872, 61.

Analytical Chemistry, Vol. 73, No. 16, August 15, 2001

3823