Liquid Chromatography Coupled On-Line to Flow ... - ACS Publications

Jul 15, 2003 - (0.5 nmol/L), digoxin (0.1 nmol/L), and gitoxigenin (50 nmol/L). The applicability of LC coupled on-line to flow cytometry was demonstr...
1 downloads 0 Views 118KB Size
Download PDF
Anal. Chem. 2003, 75, 4272-4278

Liquid Chromatography Coupled On-Line to Flow Cytometry for Postcolumn Homogeneous Biochemical Detection T. Schenk,*,† A. Molendijk,† H. Irth,‡ U. R. Tjaden,§ and J. van der Greef§

Kiadis B.V., Niels Bohrweg 11-13, 2333 CA, Leiden, The Netherlands, Faculty of Sciences, Division of Chemistry, Department of Analytical Chemistry and Applied Spectroscopy, Vrije Universiteit Amsterdam, De Boelelaan 1083, 1081 HV Amsterdam, The Netherlands, and Leiden/Amsterdam Center for Drug Research, Division of Analytical Biosciences, Leiden University, P.O. Box 9502, 2300 RA, Leiden, The Netherlands

The feasibility of flow cytometry as read-out principle for homogeneous cell- or bead-based assays coupled on-line to LC is demonstrated using digoxin-coated beads (DigBeads) and fluorescent-labeled anti-digoxin (AD-FITC) as model system. The assay is carried out in a postcolumn continuous-flow reaction detection system where the ADFITC and Dig-Beads are simultaneously added to the eluate of an LC separation column. Binding of AD-FITC to Dig-Beads results in a constant amount of fluorescence associated with the beads, which is detected by the flow cytometer. The presence of active compounds, such as digoxin and its analogues, in the sample will results in a decrease of the AD-FITC-Dig-Bead complex and, consequently, in the bead-associated fluorescence. Hence, the bead-associated fluorescence detected is inversely related to the digoxin concentration. A data-handling algorithm was developed in-house for adequate analysis of raw data output from the flow cytometer. Various conditions that influence the performance of this novel LC-biochemical detection (LC-BCD) system were investigated to determine the optimal settings of the bead-based biochemical interaction. The optimized flow injection bead-based assay was capable of detecting very low concentrations of digoxigenin (0.5 nmol/L), digoxin (0.1 nmol/L), and gitoxigenin (50 nmol/L). The applicability of LC coupled on-line to flow cytometry was demonstrated by the individual detection of digoxin, digoxigenin, and gitoxigenin in a single LC analysis. The successful coupling of LC on-line to flow cytometry principally enables the use of a wide range of new homogeneous assay formats in LC-BCD, such as membrane-bound receptor assays, cell-binding assays, and functional cell-based assays. Next to the ability to use insoluble targets, and also multiplexing assays, i.e., performing a number of assays simultaneously, using color- or size-coded beads becomes at hand in LCBCD. * Corresponding author. Tel: 0031 (0)71 5810000. Fax: 0031 (0)71 5810001. E-mail: [email protected]. † Kiadis B.V. ‡ Vrije Universiteit Amsterdam. § Leiden University.

4272 Analytical Chemistry, Vol. 75, No. 16, August 15, 2003

Detectors traditionally used in LC respond only to the physical and chemical properties of analytes while, especially in bioanalysis and drug discovery, often the functional aspects of a compound are much more of interest. Hence, the detection of biological activity directly in the effluent of an LC column using a continuousflow biochemical assay is a more effective approach to obtain information about the biological activity of the individual compounds present in a mixture. The coupling of liquid chromatography with biochemical detection (LC-BCD) has proven to be a very powerful tool in the fields of bioanalysis1,2 and drug discovery.3 The main advantage of LC-BCD over batch biochemical assays is that (cross-reactive) compounds are separated prior to the biochemical assay. In addition, parallel mass spectrometry allows identification of the active compounds in the complex mixture in one single run, hereby overcoming the tedious fraction collection and manual operations required to prepare the fractions for batch biochemical assays. To allow the on-line coupling of LC with a biochemical assay, the assay has to be performed in a continuous-flow mode. In the past decade, many different continuous-flow assay formats have been introduced. These assays can be classified as being either homogeneous or heterogeneous. A homogeneous assay does not require separation of the different reactants prior to the readout. An example of such a homogeneous assay in LC-BCD is the estrogen receptor assay developed by Oosterkamp et al.4 However, most LC-BCD systems developed so far require a separation step in the assay and are therefore classified as heterogeneous assays. Different separation devices (affinity columns,5 restricted access columns,1 cross-flow membranes,6 and a free-flow electrophoresis chip7) have been introduced to obtain the required separation of (1) Oosterkamp, A. J.; Irth, H.; Marko-Varga, G.; Heintz, L.; Kjellstrom, S.; Alkner, U. J. Clin. Ligand Assay 1997, 20 (1), 40-48. (2) Miller, K. J.; Herman, A. C. Anal. Chem. 1996, 68, 3077-3082. (3) Schobel, U.; Frenay, M.; van Elswijk, D. A.; McAndrews, J. M.; Long, K. R.; Olson, L. M.; Bobzin, S. C.; Irth, H. J. Biomol. Screening 2001, 6 (5), 291303. (4) Oosterkamp, A. J.; Villaverde Herraiz, M. T.; Irth, H.; Tjaden, U. R.; van der Greef, J. Anal. Chem. 1996, 68, 1201-1206. (5) Schenk, T.; Irth, H.; Marko-Varga, G.; Edholm, L.-E.; Tjaden, U. R.; van der Greef, J. J. Pharm. Biomed. Anal. 2001, 26, 975-985. (6) Lutz, E. S.; Irth, H.; Tjaden, U. R.; van der Greef, J. J. Chromatogr., A 1996, 755, 179-187. (7) Mazereeuw, M.; de Best, C. M.; Tjaden, U. R.; Irth, H.; van der Greef, J. Anal. Chem. 2000, 72, 3881-3886. 10.1021/ac0341822 CCC: $25.00

© 2003 American Chemical Society Published on Web 07/15/2003

assay reactants. A disadvantage of the introduction of separation devices in the postcolumn detection system is the demand to regenerate the device after a fixed number of analyses. Second, additional band-broadening is introduced, which reduces the resolution obtained in the LC separation. Finally, nonspecific binding of target proteins and sample components will occur, which often leads to clogging of the complete detection system. The continuous-flow biochemical assays developed so far are all based on soluble target molecules. So far, no membrane-bound receptor assay, cell-binding assay, or functional cell-based assay has been reported in LC-BCD. This greatly limits the use of LCBCD in drug discovery, because membrane-bound receptors or enzymes and ion channels constitute an important class of biological targets for future drugs.8 Because of these drawbacks homogeneous assays are preferred above heterogeneous assays. In the past decade, different readout techniques, such as UV absorbance9 and fluorescence detection,5 have been used in LCBCD. Fluorescence detection is most frequently used because it is more sensitive and selective compared to UV absorbance. However, when LC-BCD is used in drug discovery for the detection of active compounds in natural extracts, even fluorescence detection is not selective enough and gives rise to many problems concerning native fluorescent compounds present in the natural extract. Recently, Hogenboom et al.10 reported an elegant way to overcome both drawbacks concerning heterogeneous assays and native fluorescence by using mass spectrometry as readout for continuous-flow assays. This approach also offers the opportunity to perform multiple assays simultaneously (multiplexing).11 Unfortunately, MS is not compatible with buffer salts used (and often essential) in biochemical reactions. Hence, when MS is used as readout in biochemical assays, one is limited to the use of volatile buffer in the reaction media. In this paper, we present the use of flow cytometry as readout for LC-BCD. Flow cytometry is a readout technique that has been used for the analysis of cells for many years.12 In flow cytometry, cells under investigation are fed into a coaxial liquid stream called the sheath fluid.13 The sheath fluid carries the cells through a laser beam, hereby illuminating the cells one by one. The optical system of the flow cytometer is designed to measure both the scattered light and the fluorescent responses from the individual cells. The light scattering provides information about the size and shape of the cells. Simultaneously, the fluorescent signals provide information about fluorescent-labeled ligands bound to the cell surface or fluorescent probes present in the cells. Because the laser of a flow cytometer only illuminates a volume as small as the cell itself, the cell-associated fluorescence can be distinguished from the fluorescence in the solution surrounding the cell.14 The physical basis for discrimination of bound and free label is (8) Hopkins, A. L.; Groom, C. R. Nat. Rev. Drug Discovery 2002, 1, 727-730. (9) Ingkaninan, K.; de Best, C. M.; van der Heijden, R.; Hofte, A. J.; Karabatak, B.; Irth, H.; Tjaden, U. R.; van der Greef, J.; Verpoorte, R. J. Chromatogr., A 2000, 872 (1-2), 61-73. (10) Hogenboom, A. C.; de Boer, A. R.; Derks, R. J.; Irth, H. Anal. Chem. 2001, 73 (16), 3816-3823. (11) Derks, R. J.; Hogenboom, A. C.; van der Zwan, G.; Irth, H. Anal. Chem., in press. (12) Macey, M. G., Ed. Flow Cytometry Clinical Applications; Blackwell Scientific Publications: Oxford, U. K., 1994. (13) McCoy, J. P. Hematol.-Oncol. Clin. N. Am. 2002, 16 (2), 229-243. (14) Shapiro, H. M. Practical Flow Cytometry, 3rd ed.; Wiley-Liss: New York, 1995; p 295.

discussed by Murphy.15 These characteristics of flow cytometry have been used successfully to perform homogeneous cell- or bead-based assays in batch for many years. In cell-based assays, a flow cytometer can detect the binding of fluorescent ligands to membrane-bound receptors present at the cell surface.16,17 Loading cells with the appropriate probe also allows the detection of cellular responses (such as intracellular calcium release,18,19 changes in membrane permeability,20 changes in membrane potential,21 or changes of the intracellular pH22) caused by binding of ligands or other stimuli. Flow cytometry has also been applied to the analysis of receptor-G protein interactions in cell membranes.23 The development of several mixing devices to facilitate on-line reagent addition to a cell suspension and to reduce the transfer time to the detection point have enabled the monitoring of fast cellular responses and subsecond kinetic measurements.24-28 In the past decade, the throughput of flow cytometry has been drastically increased by the introduction of flow injection cytometry29 and plug flow cytometry,30 permitting aspiration of samples from microplate wells and delivery to the flow cytometer for analysis at rates approaching 100 samples/min.31 This high throughput in combination with the ability of multiplexing, in which many assays are performed simultaneously,32,33 allows the fast screening of large combinatorial or natural product libraries against multiple targets. However, despite all recent advances in the area of flow cytometry, distinction between cross-reactive compounds present in a complex mixture remains unfeasible using the current approaches. The on-line coupling of LC with flow cytometry would result in a new extremely powerful analytical tool. The use of flow cytometry in LC-BCD would overcome the drawbacks and limitations currently encountered in LC-BCD, such as regeneration of a separation device, band-broadening, clogging of the continuous-flow assay system, interference from native fluorescence of natural extracts, and inability to use insoluble targets such as membrane proteins or cells. (15) Murphy, R. F. Ligand Binding, Endocytosis, and Processing. In Flow Cytometry and Sorting, 2nd ed.; Melamed, M. R., Lindmo, T., Mendelsohn, M. L., Eds.; Wiley-Liss: New York, 1991; pp 355-366. (16) Sklar, L. A.; Edwards, B. S.; Graves, S. W.; Nolan, J. P.; Prossnitz, E. R. Annu. Rev. Biophys. Biomol. Struct. 2002, 31, 97-119. (17) Waller, A.; Pipkorn, D.; Sutton, K. L.; Lindermann J. J.; Omann G. M. Cytometry 2001, 45 (2), 102-114. (18) Brewis, I. A.; Morton, I. E.; Mohammad, S. N.; Browes, C. E.; Moore H. D. J. Androl. 2000, 21 (2), 238-249. (19) Burchiel, S. W.; Edwards, B. S.; Kuckuck, F. W.; Lauer, F. T.; Prossnitz, E. R.; Ransom, J. T.; Sklar, L. A. Methods 2000, 21 (3), 221-230. (20) Reynolds, J. E.; Li, J.; Eastman, A. Cytometry 1996, 25 (4), 349-357. (21) Rottenberg, H.; Wu, S. Biochim. Biophys. Acta 1998, 1404 (3), 393-404. (22) Tarnok, A.; Dorger, M.; Berg, I.; Gercken, G.; Schluter, T. Cytometry 2001, 43 (3), 204-210. (23) Sarvazyan, N. A.; Lim, W. K.; Neubig, R. R. Biochemistry 2002, 41 (42), 12858-12867. (24) Kachel, V.; Glossner, E.; Schneider, H. Cytometry 1982, 3 (3), 202-212. (25) Omann, G. M.; Coppersmith, W.; Finney, D. A.; Sklar, L. A. Cytometry 1985, 6, 69-73. (26) Pennings, A.; Speth, P.; Wessels, H.; Haanen, C. Cytometry 1987, 8, 335338. (27) Kelley, K. A. Cytometry 1989, 10, 796-800. (28) Graves, S. W.; Nolan, J. P.; Jett, J. H.; Martin, J. C.; Sklar, L. A. Cytometry 2002, 47 (2), 127-137. (29) Lindberg, W.; Ruzicka, J.; Christian, G. D. Cytometry 1993, 14, 230-236. (30) Edwards, B. S.; Kuckuck, F.; Sklar, L. A. Cytometry 1999, 37 (2), 156-159. (31) Kuckuck, F.; Edwards, B. S.; Sklar, L. A. Cytometry 2001, 44 (1), 83-90. (32) Kellar, K. L.; Iannone, M. A. Exp. Hematol. 2002, 30 (11), 1227-1237. (33) Carson, R. T.; Vignali, D. A. J. Immunol. Methods 1999, 227 (1-2), 41-52.

Analytical Chemistry, Vol. 75, No. 16, August 15, 2003

4273

In this article, the coupling of LC on-line with flow cytometry is described using a bead-based assay as a model for cell-based assays. This model assay is based on the binding of AD-FITC to Dig-Beads resulting in a constant amount of fluorescence associated with the beads, which is detected by a flow cytometer. A flow injection homogeneous assay is developed to optimize the different conditions influencing the performance of the assay, such as reaction temperature, reaction time, and concentration of the assay reagents. During the optimization, special attention is paid to the compatibility of the bead-based assay with the eluent of a reversed-phase separation system containing organic modifiers. After the successful coupling of LC with the optimized bead-based assay, the applicability of on-line LC-flow cytometry is demonstrated by the simultaneous detection of digoxin, digoxigenin, and gitoxigenin. EXPERIMENTAL SECTION Materials. Anti-digoxigenin-fluorescein Fab fragments (ADFITC) and “blocking reagent for ELISA” were purchased from Roche (Mannheim, Germany). Digoxigenin, digoxin, and gitoxigenin were obtained from Sigma (St. Louis, MO). Hydrazine beads (2.96 µm) were purchased from Bangs Laboratories (Fisher), methanol and DMSO were obtained from J. T. Baker (Deventer, The Netherlands). Phosphate-buffered saline solution (PBS) was prepared by dissolving one PBS tablet from ICN (Aurora, OH) in 100 mL of water. The addition of 0.5% (w/w) Tween 20 to the PBS solution resulted in a buffer referred to as PBST. The addition of 0.1% blocking reagent for ELISA to the PBST solution resulted in a buffer referred to as PBSTBR. The sheet liquid solution used for the flow cytometer consisted of one PBS tablet per 20 L of Milli-Q water. All other chemicals were purchased from Merck (Darmstadt, Germany) and were analytical grade. Instrumentation. All experiments were carried out using a Becton Dickinson (San Jose, CA) FACScan flow cytometer equipped with a FACSflow supply system. To allow a continuous input of beads from the flow assay into the flow cytometer, some minor hardware adjustments were required. First, the outside tube from the sample inlet system was removed. This allowed the connection of the reaction coil outlet to the inlet tube of the flow cytometer consisting of standard 1/16-in. stainless steel tubing. Second, the sample pressure tubing of the flow cytometer was blocked to mimic the presence of a sample tube and to allow sample pressure build-up required for Becton Dickinson’s Cellquest software to record data. Continuous-flow experiments were carried out using a Gilson 234 autoinjector (Villiers-le-Bel, France) and Valco VICI six-port injection valves (Schenkon, Switzerland) (injection loop 5 µL). Agilent (Waldbronn, Germany) IsoPump G1310 A 1100 series were used to displace solvents. Solvents were degassed using an Agilent in-line four-channel degasser. Two Pharmacia (Uppsala, Sweden) 10-mL Superloops were used for addition of the AD-FITC solution and the bead suspension using water as displacement solution. These Superloops were placed in a Shimadzu (Kyoto, Japan) TO-10ACvp column oven to allow temperature control at 18 °C. An additional Shimadzu TO-10ACvp column oven was used to control the reaction temperature and the temperature of the isocratic LC separation. All connecting tubing consisted of 0.13-mm-i.d. PEEK. Between the LC system and the postcolumn biochemical detection system, a custom-made “flow-mix-splitter” was used (Accurate; LC-Packings, Amsterdam, 4274 Analytical Chemistry, Vol. 75, No. 16, August 15, 2003

Figure 1. Scheme of the postcolumn continuous-flow homogeneous bead-based assay for digoxin. (1) Effluent of the LC column; (2) ADFITC solution is added to the LC effluent; (3) Dig-Beads solution is added to the LC effluent. (A) In the reaction coil, the binding of ADFITC to the Dig-Beads takes place resulting in highly fluorescent beads; (B) when digoxin or analogues elute from the LC column, the binding of these compounds reduces the amount of AD-FITC/DigBeads complex formed, resulting in a reduction of the bead-associated fluorescence. For detailed conditions, see Experimental Section.

The Netherlands). All solutions in the biochemical detection system were added together using a so-called low-dead-volume Valco VICI mixing cross (Schenkon, Switzerland). Synthesis of the Digoxin-Bound Support. Digoxin (13.2 mg) was suspended in 2 mL of ethanol, and 2 mL of a 0.1 M sodium metaperiodate was added dropwise in the absence of light under stirring. After 30 min, the excess of metaperiodate was titrated with an aqueous solution of sodium bisulfite (1 M) until the brown color had disappeared. After addition of 5 mL of saturated ammonium sulfate solution, the ethanol phase was recovered and used without further treatment for the synthesis of the digoxinbound support. Hydrazine Beads (50 mg) were washed three times with 0.1 M phosphate buffer, pH 6.15, and subsequently suspended in 1 mL of the same buffer. The ethanol phase, containing the digoxin, was added and stirred for 72 h at room temperature in the dark. Finally, the digoxin-coated beads (DigBeads) were washed with and stored in PBS containing 0.01% sodium azide at 4 °C. Batch Experiments. To verify whether the immobilization of digoxin to the support was successful and to determine the saturation binding curve, increasing concentrations of AD-FITC were incubated with Dig-Beads. After 2 h of incubation, the fluorescence associated with the beads was measured using the flow cytometer. Additional experiments were performed to study whether the binding of AD-FITC to the Dig-Beads was a specific interaction of the antibody fragments to its immobilized antigens or was caused by nonspecific binding. For this purpose, AD-FITC was incubated with increasing concentrations of digoxin for 2 h. Subsequently, Dig-Beads were added and incubated for 2 h. All batch experiment samples were prepared in PBSTBR. Flow Injection Bead-Based Assay. A scheme of the flow injection bead-based assay is displayed in Figure 1. The carrier flow consisted of PBST. The Dig-Beads were dissolved in carrier solution. AD-FITC was dissolved in PBSTBR to reduce nonspecific binding of the antibody fragments to the inner surface of the system. Different reaction coil volumes were used to optimize the reaction time, all consisting of knitted 0.2-mm i.d. × 0.4 mm o.d. PTFE tubing. The concentrations of AD-FITC and Dig-Beads were also varied during optimization of the assay conditions. All solutions were pumped at a flow rate of 25 µL/min. The solutions were added together using a so-called low-dead-volume mixing cross (Valco VICI; Schenkon, Switzerland). During optimization, 12 nmol/L digoxin dissolved in carrier solution (5 µL) was used to obtain information about the assay response.

Figure 2. Flow injection analysis of (1) 120, (2) 12, and (3) 1.2 nmol/L digoxin using the continuous-flow bead-based assay. (A) A typical scatterplot presenting the forward scatter and side scatter of every single bead detected by the flow cytometer; (B) histogram presenting the amount of Dig-Beads per second detected by the flow cytometer; (C) a dot plot presenting the bead-associated fluorescence in time; (D) the bead-associated fluorescence in time after transforming the data with the in-house-developed data-handling algorithm. For detailed conditions, see Experimental Section.

On-Line Coupling of LC to the Continuous-Flow BeadBased Assay. A Vydac (Hesperia) C4 reversed-phase, 250 × 1.0 mm i.d. packed with 5-µm particles (300 Å) LC column was used to perform isocratic separations of digoxin analogues. Elution was performed at 50 °C using methanol/water (40/60, v/v) at a flow rate of 100 µL/min. All samples (5 µL) were dissolved in eluent, unless stated otherwise. When coupling the continuous-flow beadbased assay on-line to LC, the eluent was first diluted down to 10% methanol by postcolumn addition of 300 µL/min water using the Accurate “flow-mix-splitter”. With the latter device, the flow of 400 µL/min was split to a flow of 25 µL/min entering the biochemical detection system. Data Analysis. All data were recorded using Becton Dickinson’s Cellquest software. To allow processing of the continuousflow data, a data analysis algorithm was developed in-house. This algorithm consisted of the following steps: (i) export of the raw data in ASCII format using FSC assistant software (Becton Dickinson); (ii) taking the median of the fluorescence/beads per second; (iii) applying the Savitzky-Golay algorithm.34 RESULTS AND DISCUSSION Concept of the Postcolumn Continuous-Flow Bead-Based Assay. The objective of the presented work was to study the applicability of flow cytometry as readout principle in LC-BCD to overcome the problems and limitations currently encountered in LC-BCD. Hereto, a detection system for digoxin based on beads was chosen to serve as a model system for cell-based assays. The affinity interaction between AD-FITC and digoxin-coated beads was chosen to study the characteristics of the on-line flow cytometry biochemical assay. Figure 1 shows a schematic representation of the postcolumn biochemical detection system. The affinity protein (AD-FITC) and the digoxin-coated beads are allowed to react (reaction A) in the reaction coil. The complexation results in a continuous flow of highly fluorescent-associated beads into the flow cytometer creating a stable baseline. When active compounds are injected, i.e., digoxin or its analogues, these compounds occupy the binding site of the AD-FITC (reaction B). (34) Savitzky, G.; Golay, M. J. E. Anal. Chem. 1964, 36, 1627-1638.

This will lead to a decrease in the affinity protein/beads complex leading to a temporary reduction of the baseline (“assay response”) in the flow cytometer. The amplitude of the assay response is a measure for the amount of active compound in the sample. Batch Experiments. After immobilization of digoxin on the beads, batch experiments were performed to demonstrate the fundamental aspects of this assay. The AD-FITC was able to bind to the digoxin-coated beads, and this binding could be prevented by preincubation of the antibody fragments with free digoxin. As expected, these experiments demonstrated that the flow cytometer was capable of selectively detecting only the fluorescence associated to the beads; i.e., the AD-FITC in the liquid surrounding the beads did not interfere with the detection (data available upon request). As discussed before, this is a characteristic advantage of the use of a flow cytometer as detection device.14 In batch biochemical assays, incubation times commonly used are in the order of several hours. When incubation times of 2 h were used, a detection limit for digoxin was obtained of 60 pmol/L. To predict the influence of shortening the reaction time to several minutes, which is common in postcolumn continuous-flow assays, batch experiments were performed with incubation times of 2 min. The reduction of the reaction time from 2 h to 2 min resulted in an increase of the detection limit for digoxin to 0.5 nmol/L. These results indicated that the development of a continuous-flow beadbased assay for digoxin was feasible. Continuous-Flow Assay. To investigate the performance of the assay in a continuous-flow format, experiments were performed using the setup presented in Figure 1. This stand-alone continuous-flow detection system is referred to as a “flow injection detection system” in which an autoinjector, equipped with an injection valve, is used for introduction of the (digoxin) sample in the carrier flow. For these experiments, the AD-FITC was added at a concentration of 200 nmol/L, the Dig-Beads were added at a concentration of 2.5 µg/mL, and a reaction coil of 150 µL (2 min reaction time) was used. Panels A-C of Figure 2 show the raw data from the flow cytometer after flow injection analysis of (1) 120, (2) 12, and (3) 1.2 nmol/L digoxin. Every dot in Figure 2A Analytical Chemistry, Vol. 75, No. 16, August 15, 2003

4275

Figure 3. Influence of the temperature on the assay response of the continuous-flow homogeneous bead-based assay.

represents a bead illuminated by the laser of the flow cytometer with the forward light scatterplotted on the horizontal axis and the side light scatterplotted on the vertical axis. As can be observed in Figure 2B, a stable bead flow of ∼60 beads/s is detected, indicating that we were able to create a constant flow of Dig-Beads into the flow assay without problems caused by precipitation of the beads in the Superloop or clogging of the system. In Figure 2C, the bead-associated fluorescence is plotted as a function of time. The beads, coming out of the continuousflow assay, are highly fluorescent, proving that the AD-FITC is capable of binding to the Dig-Beads in the relatively short reaction time of 2 min. A strong temporary reduction in bead-associated fluorescence is the result of the injection of (1) 120, (2) 12, and (3) 1.2 nmol/L digoxin, respectively, indicating that the assay response is concentration dependent. These results prove the concept of a continuous-flow homogeneous bead-based assay technology. The injection of 120 nmol/L digoxin caused a maximum reduction of the bead-associated fluorescence. All ADFITC formed complexes with the injected digoxin in the sample and no antibodies fragments associated with the beads. Figure 2C also shows a wide distribution of the amount of bead-associated fluorescence. This limits the minimum detectable assay response. Analysis of the same data using an in-house-developed datahandling algorithm resulted in a data output presented in Figure 2D. A comparison of Figure 2C and D shows the importance of data handling for the use of signal identification and sensitivity.

Influence of the Temperature. In continuous-flow postcolumn biochemical assays, incubation times are limited to a few minutes. The use of longer incubation times would require the use of larger reaction coils resulting in unacceptable bandbroadening, which diminishes the resolution obtained in the LC separation. One way to obtain a substantial response from a biochemical reaction in the relatively short reaction times is increasing the reaction speed by increasing the reaction temperature. To optimize the reaction temperature for this biochemical assay, the flow injection detection system was operated at different temperatures using 100 nmol/L AD-FITC, 2.5 µg/mL Dig-Beads, and a reaction coil of 150 µL (2-min reaction time). Figure 3 shows the influence of the temperature on the assay response. This figure clearly demonstrates that increasing the temperature results in an increase of the assay response with a maximum at 50 °C due to increased reaction kinetics. When the temperature is elevated higher that 50 °C, the assay response decreases again, until finally at 80 °C no assay response is observed. This decrease in assay response is probably caused by degradation of the AD-FITC proteins at high temperatures. In further optimization experiments, the optimum temperature of 50 °C will be used. Influence of the Dig-Beads Concentration. The influence of the Dig-Beads’ concentration on the performance of the continuous-flow bead-based assay was studied using 100 nmol/L AD-FITC and a reaction coil of 150 µL (2-min reaction time). Figure 4A shows that an increase of the Dig-Beads concentration in the assay results in a decrease of the assay response. This phenomenon is a consequence of a competition between the free digoxin and digoxin beads for the AD-FITC. The more Dig-Beads are available, the more AD-FITC will bind to the beads. Consequently, less AD-FITC will be available to bind to the free digoxin. The relatively strong increase of the noise shown in Figure 4A, when the bead concentration is reduced below 7 µg/mL, is a consequence of the lack of data points available per second to obtain a representative measurement. Above 7 µg/mL, the noise increases slightly with increasing bead concentration. This is caused by the reduction of AD-FITC binding to the Dig-Beads resulting in a reduction of the bead-associated fluorescence, as explained earlier. The reduced fluorescence per bead causes an increase of the noise. In Figure 4B, the signal-to-noise ratio (S/N) is plotted against the Dig-Bead concentration. A clear optimum is observed at a Dig-Bead concentration of ∼7 µg/mL.

Figure 4. Influence of the Dig-Beads concentrations on the performance of the continuous-flow homogeneous bead-based assay. 4276

Analytical Chemistry, Vol. 75, No. 16, August 15, 2003

Figure 5. Influence of the AD-FITC concentration on the signalto-noise ratio using different reaction coils on the continuous-flow homogeneous bead-based assay.

The relatively strong increase of the S/N when the bead concentration is increased up to 7 µg/mL (Figure 4B) is caused by decrease of the noise (see Figure 4A). The decrease of the S/N upon further increasing the bead concentration (Figure 4B) is caused by a combination of a decreasing signal and an increasing noise (see Figure 4A). For further optimization experiments, the optimum temperature of 50 °C and the optimum DigBead concentration of 7 µg/mL will be used. Influence of the AD-FITC Concentration and Reaction Time. Two important parameters that can significantly influence the performance of the assay are the AD-FITC concentration and the reaction time. In biochemical assays, the optimum concentration of the reagents often depends strongly upon the reaction time of the assay. For this reason, the AD-FITC concentration and the reaction time were optimized simultaneously. Figure 5 presents the influence of the AD-FITC concentration on the signal-to-noise ratio using different reaction coils in the continuous-flow homogeneous bead-based assay. Irrespective of the reaction time, the signal-to-noise ratio increases with increasing AD-FITC concentration. At reaction times of 2 or 3 min, a maximum assay response is reached with about 200-300 nmol/L AD-FITC. However, when a reaction time of 0.5 or 1 min is used, much higher AD-FITC concentrations are required to reach a maximum assay response. Figure 5 also shows that the assay response at a reaction time of 2 min is significantly higher compared to a reaction time of 0.5 or 1 min. The increase of the reaction time from 2 to 3 min results in a minimal increase of the assay response. One important consequence of increasing the reaction time that is not observed in Figure 5 is band-broadening of the analytes introduced in the continuous-flow assay. This band-broadening diminishes the resolution obtained in the LC separation when the assay is coupled to LC. For this reason, the reaction time for future experiments was held at 2 min. With a reaction time of 2 min, an insignificant gain in assay response is observed when the AD-FITC concentration is increased above 200 nmol/L. Hence, future experiments are performed using 200 nmol/L AD-FITC. Concentration Response Curves. The continuous-flow homogeneous bead-based assay for digoxin and analogues is now optimized under physiological conditions, resulting in the following optimal conditions: a reaction temperature of 50 °C, 200 nmol/L AD-FITC, 7 µg/mL Dig-Beads, and 2-min reaction time. Using these optimum conditions, concentration response curves were

Figure 6. Concentration response curves for digoxin, digoxigenin, and gitoxigenin using the continuous-flow homogeneous bead-based assay at the optimum conditions: 200 nmol/L AD-FITC, 7 µg/mL DigBeads, 2-min reaction time, and reaction temperature of 50 °C.

determined (n ) 3) for digoxigenin, digoxin, and gitoxigenin. Figure 6 shows the typical sigmoidal curves obtained after injection of increasing concentrations of the active compounds. The limit of detection (S/N ) 3) and IC50 values were 0.2 and 8 nmol/L for digoxigenin, 0.1 and 3 nmol/L for digoxin, and 50 nmol/L and 5 µmol/L for gitoxigenin. The high IC50 value of gitoxigenin, compared to the other two compounds, is caused by the weak affinity of this compound for AD-FITC. Consequently, this weak affinity results in a high detection limit for gitoxigenin. The dynamic range for all compounds was at least 2 orders of magnitude. When the detection limit for digoxin after batch incubation and in the continuous-flow assay is compared using a reaction time of 2 min, a decrease can be observed from 0.5 to 0.1 nmol/L in favor of the continuous-flow assay. This can be explained by the poor reproducibility of the manually controlled reaction time of only 2 min in batch, leading to increased standard deviations in the results and hence increased detection limits over continuous-flow experiments. On-Line Coupling of LC to the Bead-Based Assay. Traditionally, biochemical assays are mainly performed using batch incubations in microtiter plates. Transformation of the batch biochemical assays to a continuous-flow mode is performed for only one reason, i.e., the use of LC combined with the biochemical assay to detect the presence of active compounds in a single analysis. Because reversed-phase LC (RP-LC) is the most widely applied separation technique in both bioanalysis and drug discovery, we decided to couple the newly developed continuousflow bead-based assay to a reversed-phase chromatographic system. Since RP-LC uses organic modifiers to separate the compounds in a mixture, the influence of organic modifiers on the performance of the continuous-flow bead-based assay was studied. Figure 7 presents the influence of methanol on the performance of the bead-based assay using different percentages of methanol in the carrier solution of the flow injection system. As expected, the presence of methanol in the carrier solution reduces the assay response. A loss in response of only a factor of 2 is observed when the methanol content is increased from 0 to 40% in the carrier solution. Although the assay tolerates a high percentage of methanol, it was decided to allow only 10% methanol Analytical Chemistry, Vol. 75, No. 16, August 15, 2003

4277

Figure 7. Influence of methanol on the assay response of the continuous-flow homogeneous bead-based assay.

in the bead-based assay after coupling to LC. This was achieved by diluting the 100 µL/min 40% methanol eluting from the C4 RP-LC column with water. The LC was operated at 50 °C to elute digoxin and its analogues from the LC column using only 40% methanol. Figure 8 shows a typical LC-flow cytometry chromatogram obtained after LC bead-based assay of a mixture of (1) 92 nmol/L digoxigenin, (2) 8 µmol/L gitoxigenin, and (3) 13 nmol/L digoxin. With the powerful combination of LC and flow cytometry, the biologically active compounds present in a mixture can be detected within a single analysis. CONCLUSIONS A continuous-flow homogeneous bead-based assay using flow cytometry coupled to LC was developed. Hereto, the association of AD-FITC with Dig-Beads was used as model assay for cellbased assays. The reaction kinetics of the association reaction between the affinity protein and the antigen-coated beads is fast enough to be used in a continuous-flow postcolumn reaction detection system. The reaction times in postcolumn biochemical assays are generally in the order of minutes, because longer reaction times would cause unacceptable band-broadening, diminishing the resolution obtained in the LC. With the optimized bead-based assay format, detection limits were in the order of 0.1(35) Schenk, T.; Molendijk, A.; Irth, H.; Tjaden, U. R.; van der Greef, J., in preparation.

4278 Analytical Chemistry, Vol. 75, No. 16, August 15, 2003

Figure 8. Chromatogram obtained using LC-flow cytometry of a mixture containing (1) 92 nmol/L digoxigenin, (2) 8 µmol/L gitoxigenin, and (3) 13 nmol/L digoxin.

10 nmol/L for high and weak affinity ligands in flow injection analysis. The applicability of LC coupled to flow cytometry was demonstrated by the separation and biochemical detection of a mixture of digoxin and its analogues in a single run. The successful application of beads in continuous-flow assays opens the way to development of different bead-based assays. Moreover, the new methodology allows the implementation of other insoluble targets such as membrane-bound receptors, cell fragments, or complete living cells. In addition, multiplexed biochemical assays, as commonly performed in flow cytometry, may become possible for postcolumn biochemical detection. The combination of LC and flow cytometry-based biochemical detection can be a powerful tool for the discovery of active compounds in natural product screening. The applicability of LC-flow cytometry for the discovery of active compounds in natural products is presented elsewhere.35 Future research will be focused on the use of LC-flow cytometry to perform cell-based assays and to enable multiplexing.

Received for review February 24, 2003. Accepted May 6, 2003. AC0341822