Affinity Assays Using Fluorescence Anisotropy with Capillary

A novel approach to detecting affinity interactions that combines fluorescence anisotropy with capillary electro- phoresis (FACE) was developed. In th...
0 downloads 0 Views 194KB Size
Anal. Chem. 2004, 76, 7380-7386

Affinity Assays Using Fluorescence Anisotropy with Capillary Electrophoresis Separation Rebecca J. Whelan,† Roger K. Sunahara,‡ Richard R. Neubig,‡ and Robert T. Kennedy*,†,‡

Department of Chemistry and Department of Pharmacology, University of Michigan, Ann Arbor, Michigan 48109

A novel approach to detecting affinity interactions that combines fluorescence anisotropy with capillary electrophoresis (FACE) was developed. In the method, sample is injected into a capillary filled with buffer that contains a fluorescent probe that possesses low fluorescence anisotropy. If proteins or other large molecules in the sample bind the fluorescent probe, their migration through the capillary can be detected as a positive anisotropy shift. Thus, the method provides both separation and confirmation of binding to the probe. Calculations based on combining the Perrin equation and dissociation constant were used to predict the effect of conditions on aniostropy detection. These calculations predict that low probe concentrations yield the best sensitivity while higher concentrations increase the dynamic range for detection of binding partner. The assay was applied to detection of G proteins using BODIPY FL GTPγS as the fluorescent probe. Experimental measurements exhibited trends in anisotropy with varying probe and protein concentrations that were consistent with the calculations. The limit of detection for Gri1 was 3 nM when the electrophoresis buffer contained 250 nM BODIPY FL GTPγS. FACE affinity assay is envisioned as a method that can quantify selected binding partners and screen complex samples for compounds that possess affinity for a particular small molecule that is used as a probe.

of analytical techniques including affinity CE (ACE),4,5 affinity probe CE (APCE),6-8 immunoassay,9,10 and nonequilibrium CE of equilibrium mixtures.11,12 Such assays have the potential to be used for high-throughput assays in clinical, environmental, food, and pharmaceutical applications, drug discovery, and physicochemical characterization of noncovalent interactions. In addition, since cellular signal transduction is controlled by affinity interactions, analytical tools that incorporate or measure such interactions may be useful in understanding their normal function and their derangement in various disease states. Complex formation within a CE experiment is usually detected by the mobility shift of one or both binding molecules. In ACE, the shift in migration time of injected binding partner is measured in sequential assays as the concentration of ligand in the electrophoresis buffer is increased. In APCE, a fluorophore (affinity probe) is premixed with sample and complexes with the binding partner are detected as new peaks with shifted migration times relative to free probe. (Some authors have used the term APCE for an ACE experiment in which the injected binding partner is fluorescently labeled.13 In this paper, we reserve the APCE term for separation of premixed solutions described above.) More recently, laser-induced fluorescence (LIF) anisotropy detection has also been used with CE to confirm the presence of a complex.11,14-17 LIF anisotropy is useful as a detector of noncovalent affinity interactions because of its sensitivity to the local environment of the fluorophore.18 Upon excitation of fluorescence with polarized light, a population of excited-state molecules oriented parallel with

Analytical methods that combine separations with affinity interactions, such as Western blotting, gel mobility-shift assays, and affinity chromatography, have become essential tools for determining trace compounds in complex mixtures and for studying affinity interactions between biomolecules. Considerable research interest has focused on incorporating affinity interactions into capillary electrophoresis (CE) to gain advantages of improved mass sensitivity, automation, speed, and reduced reagent usage associated with CE.1-3 Biomolecular affinity interactions have been incorporated into CE assays in a variety of ways resulting in a set

(4) Chu, Y. H.; Avila, L. Z.; Gao, J. M.; Whitesides, G. M. Acc. Chem. Res. 1995, 28, 461-468. (5) Colton, I. J.; Carbeck, J. D.; Rao, J.; Whitesides, G. M. Electrophoresis 1998, 19, 367-382. (6) Shimura, K.; Karger, B. L. Anal. Chem. 1994, 66, 9-15. (7) Jameson, E. E.; Cunliffe, J. M.; Neubig, R. R.; Sunahara, R. K.; Kennedy, R. T. Anal. Chem. 2003, 75, 4297-4304. (8) German, I.; Buchanan, D. D.; Kennedy, R. T. Anal. Chem. 1998, 70, 45404545. (9) Schultz, N. M.; Kennedy, R. T. Anal. Chem. 1993, 65, 3161-3165. (10) Schultz, N. M.; Huang, L.; Kennedy, R. T. Anal. Chem. 1995, 67, 924929. (11) Berezovski, M.; Krylov, S. N. J. Am. Chem. Soc. 2002, 124, 13674-13675. (12) Berezovski, M.; Nutiu, R.; Li, Y. F.; Krylov, S. N. Anal. Chem. 2003, 75, 1382-1386. (13) Shimura, K.; Kasai, K. Anal. Biochem. 1995, 227, 186-194. (14) Ye, L. W.; Le, X. C.; Xing, J. Z.; Ma, M. S.; Yatscoff, R. J. Chromatogr., B 1998, 714, 59-67. (15) Wan, Q. H.; Le, X. C. Anal. Chem. 1999, 71, 4183-4189. (16) Wan, Q. H.; Le, X. C. Anal. Chem. 2000, 72, 5583-5589. (17) Le, X. C.; Wan, Q. H.; Lam, M. T. Electrophoresis 2002, 23, 903-908. (18) Lakowicz, J. R. Principles of Fluorescence Spectroscopy; Kluwer Academic/ Plenum: New York, 1999.

* To whom correspondence should be addressed. E-mail: [email protected]. Fax: 734-615-6462. Phone: 734-615-4363. † Department of Chemistry. ‡ Department of Pharmacology. (1) Heegaard, N. H. H. J. Mol. Recognit. 1998, 11, 141-148. (2) Heegaard, N. H. H.; Kennedy, R. T. Electrophoresis 1999, 20, 3122-3133. (3) He, X. Y.; Ding, Y. S.; Li, D. Z.; Lin, B. C. Electrophoresis 2004, 25, 697711.

7380 Analytical Chemistry, Vol. 76, No. 24, December 15, 2004

10.1021/ac0489566 CCC: $27.50

© 2004 American Chemical Society Published on Web 11/17/2004

Figure 1. Principle of FACE affinity assays. Electrophoresis buffer contains labeled probe that undergoes rapid rotation because of its small size. Sample containing unlabeled protein(s) is injected and separated by electrophoresis. As protein migrates through the capillary, probe-protein complexes continually form and dissociate during the separation. Unbound probe has low anisotropy. Complex has high anisotropy, resulting in an anisotropy shift as the protein migrates through the detection point.

the excitation source is produced. Rapid molecular rotation scrambles the orientation of the excited population, resulting in fluorescence emission that is depolarized (isotropic) relative to the excitation source. If the same fluorophore interacts with another molecule to form an affinity complex, the complex will rotate more slowly than fluorophore alone, and its fluorescence will retain a larger degree of polarization (anisotropy). The combination of CE with anisotropy detection has proven extremely useful for confirming the presence of complexes in various CE experiments.11,14-17 CE with anisotropy detection has previously been performed by injecting a fluorescent probe into a capillary and monitoring the anisotropy of the fluorophore zone as it migrates through the detector. The unlabeled binding partner may either be premixed with the fluorescent probe in the sample11,14-16 or added to the buffer background.15,16 We have investigated affinity assays that use fluorescence anisotropy with capillary electrophoresis (FACE), in which the fluorescent probe is present in the electrophoresis buffer and unlabeled binding partners are detected as they migrate through the detection window as increases in anisotropy of the fluorescent probe (Figure 1). The presence of a probe in the electrophoresis buffer means that the separation is performed under equilibrium conditions and that dissociation of low-affinity complexes during separation is not detrimental to complex detection. In addition, FACE affinity assays do not require that analytes be fluorescently labeled; therefore, multiple unlabeled analyte molecules may be simultaneously analyzed for their interaction with the probe. This is in contrast to previously reported uses of anisotropy detection for CE that require that each analyte be fluorescently labeled.11,14-17 We characterize this method and demonstrate its application to detection of G proteins based on the use of BODIPY-GTPγS (BGTPγS) as the affinity-anisotropy probe. EXPERIMENTAL METHODS Chemicals. Unless otherwise stated, all chemicals were from Sigma Chemical Co. (St. Louis, MO) and of the highest purity

Figure 2. Schematic drawing of the FACE instrument. Solid lines show the path of light radiation; dashed lines show electrical connections. Excitation and emission paths are perpendicular to each other.

available. Tris-glycine buffer (10×) was purchased from BioRad Laboratories (Hercules, CA). All solutions were prepared with 18.1 MΩ‚cm deionized water from an E-Pure water purification system (Barnstead International Co., Dubuque, IA). BGTPγS was from Molecular Probes (Eugene, OR). Expression and purification of r-myristoylated GRi1 (GRi1) and His6-GRo (GRo) were performed as previously described,19,20 and the proteins were stored at -80 °C. Prior to analysis, proteins were rapidly thawed and diluted to the desired concentration in buffer. GRi1 samples were prepared in TGEMD buffer (25 mM Tris, 192 mM glycine, 1 mM EDTA, 10 mM MgCl2, 1 mM dithiothreitol, pH 8.5). Mixtures of GRi1 and GRo were prepared in Tris-glycine buffer containing 5 mM MgCl2 (TGM buffer). The concentration of GRi1 in stock solution was estimated by determining the amount of GRi1 required to saturate a standard of BGTPγS (measured by anisotropy on a plate-reading fluorometer) and fitting to a binding isotherm with Kd ) 150 nM.21 Samples contained rhodamine 110 (R110) as an internal standard. FACE Instrumentation. Figure 2 shows a schematic diagram of the components of the FACE instrument. Separations were performed in 50-cm lengths of unmodified fused-silica capillary with an inner diameter of 50 µm and an outer diameter of 360 µm (Polymicro Technologies, Phoenix, AZ). A detection window was made 30 cm from the inlet by using an electrical arc to remove a small amount of polyimide coating from the capillary. Separation voltage was provided by a Spellman CZE 1000 R power supply (Plainview, NY). Fluorescence detection was performed oncolumn. Vertically polarized light from a Kr+ laser (Omnichrome model 643-R-NIKN-A01, Melles Griot Laser Group, Carlsbad, CA) (19) Mumby, S. M.; Linder, M. E. Methods Enzymol. 1994, 237, 254-268. (20) Lee, E.; Linder, M. E.; Gilman, A. G. Methods Enzymol. 1994, 237, 146164. (21) McEwen, D. P.; Gee, K. R.; Kang, H. C.; Neubig, R. R. Anal. Biochem. 2001, 291, 109-117.

Analytical Chemistry, Vol. 76, No. 24, December 15, 2004

7381

was spectrally filtered to select the 488-nm component (Corion P10488-F, Spectra Physics, Mountain View, CA) that was then passed though a polarizer (10LP-vis, Newport Corp., Irvine, CA) to further isolate the vertically polarized component. Excitation was focused into the capillary lumen using a biconvex lens (Newport). Fluorescence emission was collected at 90° to the excitation using a 32× microscope objective (LD Achrostigmat, 0.40 NA Ph1, Carl Zeiss, Thornwood, NY). Emission was spatially filtered by a 0.8-mm slit placed in the microscope’s back focal plane. Emission was recollimated by a plano-convex lens (Newport) and sent through a polarizing beam splitter cube (Newport) to separate parallel and perpendicular emissions. Each polarization channel was spectrally filtered by a 520 ( 10-nm band-pass filter (Corion S10-520-F, Spectra Physics) and detected by a side-on photomultiplier tube (PMT; R1477, Hamamatsu USA, Bridgewater, NJ) enclosed in a shuttered light-tight box (Products for Research, Danvers, MA). Current from the PMT was converted to voltage and filtered by a low-noise current preamplifier (SR570, Stanford Research Systems, Sunnyvale, CA). Output voltage was collected by a data acquisition card (AT-MIO-16XE-50, National Instruments, Austin, TX) and computer using LabView (National Instruments) software written in-house. Fluorescence anisotropy (r) was calculated as follows:

r ) (I| - GI⊥)/(I| + 2GI⊥)

(1)

value of the probe is known; for BGTPγS, the factory default value and the manually optimized values were not significantly different. Before each assay, the PMT voltage was optimized by an autocalibration routine on the most fluorescent well. RESULTS AND DISCUSSION Calculation of Trends in Affinity Assays Using FACE. To guide the development of conditions for affinity assays using FACE, we performed calculations to predict fluorescent probe anisotropy as a function of binding partner in solution. Anisotropy for a mixture of fluorescing species was calculated using the property that anisotropy is linearly additive, weighted by the fractional contribution of individual species (bound and free fluorophore in this case).18 The concentrations of bound and free fluorophore were calculated from Kd and initial concentrations. The anisotropy of bound and free fluorophore was calculated using the Perrin equation:22

r)

7382

Analytical Chemistry, Vol. 76, No. 24, December 15, 2004

1 + (τ/θ)

(2)

where ro is the fundamental anisotropy, τ is the fluorescence lifetime, and θ is the rotational correlation time. Rotational correlation times were calculated using18

θ) where I| is the intensity of emitted light parallel to excitation, I⊥ is the intensity of emitted light perpendicular to the excitation, and G is an experimentally determined factor that corrects for any bias in sensitivity of the I| and I⊥ detection channels. G was determined as previously described.18 To determine G, it was necessary to use horizontally polarized light,18 which was produced by placing a depolarizing optic (DPU-15-vis, Optics for Research, Caldwell, NJ) before the polarizer and rotating the polarizer 90°. Assay Conditions. Injections were performed using hydrodynamic flow by placing the capillary inlet in a sample vial elevated 10 cm relative to the outlet for between 5 and 60 s. For GRi1 assays, 20-nL samples of protein in TGEMD ([GRi1] from 66 to 200 nM) were injected. For the separation of GRi1 and GRo, 2-nL samples of G proteins in TGM (1 µM each) were injected. All samples contained 14 nM R110 as an internal standard. Electrophoresis buffer was 25 mM Tris, 192 mM glycine, pH 8.5 (Tris-glycine) containing BGTPγS as the probe. BGTPγS concentration varied from 250 to 1000 nM. The applied separation voltage was 20 kV (electric field, 400 V/cm). At the start of each day, or before changing BGTPγS buffer, the capillary was conditioned by rinsing for 5 min each with 50 mM NaOH, water, Tris-glycine, and Trisglycine with BGTPγS. Anisotropy values determined on the CE instrument were compared to values measured on a commercial plate-reading fluorometer (Fusion Universal Microplate Analyzer, Packard BioSciences, Meriden, CT). For the plate reader fluorometer assays, samples containing GRi1 and BGTPγS in the required concentrations were mixed, dispensed (50 µL/well), and analyzed immediately using an excitation wavelength of 485 nm and emission wavelength of 535 nm. The factory default G value (0.9) was used for all assays. The G value used by the plate-reading fluorometer can be optimized experimentally if the anisotropy

ro

ηV ηM ) (νj + h) RT RT

(3)

where η is the viscosity, V is the molar volume, R is the gas constant (8.31 × 107 erg/mol‚K), T is the temperature, M is the molecular weight, νj is the specific volume of the protein (in mL/ g), and h is the hydration factor (g of H2O/g of protein).18 Use of eqs 2 and 3 assumes that the molecules are spherical, the dye moiety does not undergo free rotation relative to the rest of the molecule or complex, and effects of binding other than changes in molecular volume, such as dequenching of fluorescence,21 do not affect anisotropy. To calculate θ for BGTPγS using eq 3, V was estimated by comparison with space-filling atomic models of other dye molecules.23 To calculate θ for the complex, the expression on the right side of eq 3 was used with values for νj (0.735 mL/g) and h (0.20) that have been reported as typical for globular proteins.18 At η ) 0.01 P and T ) 293 K, the θ for free BGTPγS was calculated to be 0.30 ns and θ for BGTPγS-GRi1 complex was 17.6 ns. These θ values yield a predicted r ) 0.019 for BGTPγS (0% bound) and r ) 0.30 for BGTPγS-GRi1 (100% bound), using ro ) 0.4 and τ ) 6 ns for BODIPY dye (www.probes.com). Using the predicted r values for bound and free BGTPγS and Kd ) 150 nM,21 the total anisotropy was calculated for a range of BGTPγS and GRi1 concentrations with results shown in Figure 3. The plot shows that, as expected, at each probe concentration the anisotropy increases with protein concentration because a greater fraction of probe is bound, and the total anisotropy becomes weighted toward the high anisotropy value of the complex. The rate of change with protein is greatest at low probe (22) Perrin, F. J. Phys. Radium 1926, 7, 390-401. (23) Dutt, G. B.; Doraiswamy, S.; Periasamy, N.; Venkataraman, B. J. Chem. Phys. 1990, 93, 8498-8513.

Figure 3. Surface generated for calculated anisotropy of a mixture of bound and free fluorescent probe as a function of probe and protein concentration assuming that the probe is BGTPγS, the protein is GRi1, and Kd ) 150 nM.

concentrations, yielding the greatest sensitivity and best detection limit. For example, if the noise in the anisotropy measurement is 0.5 anisotropy milliunits, the protein detection limit would be 1 nM for 5 nM probe and 10 nM for 2000 nM probe. On the other hand, low probe concentrations limit the dynamic range, with saturation occurring at a lower protein concentration. These trends result from two opposing phenomena: more complexes form at higher probe concentration, which serves to increase anisotropy, but greater probe concentration means that the average anisotropy value receives a large contribution from free, low anisotropy species. Calculations indicate that, for a system with lower Kd (higher affinity), the slope of a calibration curve would be steeper with increasing protein, because a greater fraction of probe is bound at a given protein concentration. High Kd (low affinity) has the same qualitative effect as high background probe concentration, increasing the dynamic range at the expense of sensitivity. These calculations offer predictive insight on appropriate conditions for FACE affinity assays. If the function of the assay is qualitative screening for protein molecules able to bind probe, the largest anisotropy signal will result from maximizing protein concentration and minimizing probe concentration. For quantification purposes requiring a large dynamic range, it may be best to use higher concentrations of probe. To obtain equilibrium binding constant information, it would be necessary to maintain [probe] > [protein] so that complex formation does not result in probe (ligand) depletion within the capillary. Characterization of the FACE Instrument. The ability of the FACE instrument to measure anisotropy accurately was investigated by analyzing identical samples on the CE instrument and a commercial plate-reading fluorometer. Samples containing 500 nM BGTPγS and GRi1 from 0 to 130 nM were prepared and analyzed using both the plate reader and the FACE instrument. To avoid possible complications resulting from dilution of narrow

Figure 4. (A) Electropherograms from a FACE affinity assay with 250 nM BGTPγS in the electrophoresis buffer and 200 nM GRi1 injected. Dashed line is fluorescence parallel to the excitation, and solid line is fluorescence perpendicular to the excitation. (B) Anisotropy electropherograms with 250 nM BGTPγS in the electrophoresis buffer: (a) 200 nM GRi1 injected (calculated from traces in Figure 4A); (b) sample buffer injected. Trace a has been vertically offset by 0.06 anisotropy unit.

injected peaks during a CE assay, premixed solutions were continually pushed through the capillary and the anisotropy was measured for at least 30 s. Across all values of [GRi1], the anisotropy values measured on the plate-reading fluorometer and on the FACE instrument were within 8% of each other, indicating that the FACE instrument measures anisotropy accurately. The limit of detection of the FACE instrument for premixed solutions of BGTPγS/GRi1 was calculated to be 2 nM. FACE Affinity Assay for G proteins. FACE was evaluated for samples containing 200 nM GRi1 and 14 nM R110 as internal standard with 250 nM BGTPγS in the electrophoresis buffer. Figure 4 illustrates electropherograms recorded on the parallel and perpendicular channels (Figure 4A) and the calculated anisotropy trace (Figure 4B, trace a). The internal standard migrated at ∼100 s, and although it resulted in a positive fluorescence feature, the parallel and perpendicular components of the emission were equal, meaning that it did not exhibit a significant change in anisotropy relative to the background. Given the comparable size of R110 and BGTPγS, the lack of an anisotropy change for R110 is expected. The peak at 125 s had positive anisotropy as indicated by both trace a in Figure 4B and the greater fluorescence on the parallel channel in Figure 4A. This Analytical Chemistry, Vol. 76, No. 24, December 15, 2004

7383

peak increased linearly with GRi1 concentration and was absent with injections of buffer containing no protein. It was concluded that the anisotropy peak at 125 s resulted from the detection of BGTPγS/GRi1 complexes. We observed that the complex yielded not only an anisotropy increase but an increase in fluorescence as well (Figure 4A). While the increase in fluorescence associated with the protein peak was partially due to dequenching of fluorescence for BGTPγS bound to the protein.;21 preliminary investigations (data not shown) with other binding systems reveal that positive fluorescence peaks are detected even for binding pairs with no dequencing. This effect may be due to analyte molecules concentrating fluorescent ligand in the analyte zone; however, the mechanism is still under investigation. Given these observations, it is possible to use direct fluorescence to detect the proteins in these assays; however, without the anisotropy trace it would not be possible to confirm which peaks in a given trace represent complex. The significance of this additional information is apparent from a comparison of panels A and B of Figure 4, which shows that the anisotropy trace is much simplified compared to the fluorescence traces by minimizing signals unrelated to binding. Two other features consistently appeared in the electropherograms, a positive fluorescence feature at ∼75 s and a sharp rise and dip below baseline fluorescence levels at ∼175 s, followed by a gradual return to baseline by 200 s. The “negative fluorescence” region of the electropherogram (between 175 and 200 s) exhibited positive anisotropy. These peaks, apparent even without protein or R110 in the sample, may indicate a fluorescent product in the sample buffer (Figure 4B, trace b). In addition, it is anticipated that, with fluorophore in the electrophoresis buffer, baseline disturbances may result from displacement (as occurs with indirect detection) or stacking and depletion of fluorophore caused by differences in conductivity and mobility of the separation and sample buffers. The positive anisotropy at 175 s is not likely due to an actual anisotropy change but instead an artifact of the low fluorescence in this region. Thus, while the use of anisotropy measurements enables distinction of complex from some components (such as the sample buffer and internal standard peak), some artifacts may still occur such as the peak at 175 s. Choosing appropriate buffers can eliminate some of these peaks, as the following discussion of Figure 5 illustrates. To illustrate the potential of the method for detecting mixtures of binding partners, samples containing GRi1 and GRo were analyzed. Electrophoresis of these mixtures resulted in two features with greater fluorescence in the parallel channel than the perpendicular channel, as shown in Figure 5A. The calculated anisotropy electropherogram (Figure 5B) confirms that these features result from the detection of complexes of BGTPγS and GR subunits that are formed dynamically during the separation. When TGM is used as sample buffer (Figure 5) rather than TGEMD (Figure 4), the baseline artifacts are reduced in size, further suggesting that these features result from sample buffer additives. Figure 5 demonstrates the ability of FACE to detect multiple unlabeled protein analytes as a result of their independent interaction with a small affinity and anisotropy probe. Affinity Assays by FACE Assays Compared with Calculations. We next evaluated how well the calculations could predict the effect of assay conditions on FACE measurements. To perform 7384 Analytical Chemistry, Vol. 76, No. 24, December 15, 2004

Figure 5. (A) Electropherograms from a FACE affinity assay with 250 nM BGTPγS in the electrophoresis buffer and a mixture containing GRi1 and GRo (1 µM each) injected. Dashed line is fluorescence parallel to the excitation; solid line is fluorescence perpendicular to the excitation. (B) Anisotropy electropherogram calculated from traces in (A).

this comparison, FACE assays were performed with electrophoresis buffer containing BGTPγS at 250, 500, 750, and 1000 nM and with samples containing GRi1 at 0, 66, 133, and 200 nM. Assays were performed for each of the 16 possible combinations in triplicate. Peak anisotropy values from the assays as a function of protein concentration are compared to calculated values for the four probe concentrations in Figure 6. (The solid lines on the plots correspond to slices taken from the surface plot in Figure 3.) The lower absolute values of the calculated curves may be due to underestimation of free BGTPγS anisotropy in the calculations. Predicted anisotropy for free BGTPγS was 0.019 while the measured anisotropy was 0.035. Using the measured anisotropy value for free probe in the calculations results in a much better fit to the experimental data, as shown by the dashed lines in Figure 6, illustrating that most of the error is associated with calculation of the small fluorophore anisotropy. The calculations are also limited by not accounting for dilution of protein during separation; i.e., the calculations used the injected protein concentration, not protein concentration at the detector. Despite these limitations, the experimental and calculated results agree well with respect to the two important trends we noted when evaluating the calculated surface (Figure 3)snamely, the increase of anisotropy

Figure 6. Comparison of predicted and measured anisotropy values. Anisotropy values at a range of probe and protein concentrations were predicted by assuming BGTPγS probe, GRi1 protein, and Kd ) 150 nM as in Figure 3. Solid line shows the results of predictions using theoretical anisotropy values for free and bound probe. Dashed line was calculated using an experimentally determined value for free probe anisotropy. Open circles are anisotropy peak heights measured from affinity FACE assays. Error bars are (1 standard deviation (n ) 3). (A) [BGTPγS] ) 250 nM. (B) [BGTPγS] ) 500 nΜ. (C) [BGTPγS] ) 750 nM. (D) [BGTPγS] ) 1000 nM.

with protein concentration and the decrease of anisotropy with probe concentration. In addition, the improvement of sensitivity with lower probe concentration is also apparent by comparing the slopes of the data. (The slope increases from 2.2 × 10-4 to 5.3 × 10-4 aniostropy units/nM for a decrease in probe concentration from 1000 to 250 nM.) These results confirm that the calculations used here are useful for establishing initial conditions for affinity FACE assays. The data in Figure 6 also illustrate that the reproducibility and linearity of the responses over different conditions is sufficient for accurate analysis. Using the data for the lowest concentration of BGTPγS used in the electrophoresis buffer (250 nM), the detection limit for unlabeled GRi1 was 3 nM. FACE Compared with Other Affinity CE Methods. As several affinity CE techniques have been developed, it is of interest to consider how this technique offers new capabilities. Like the technique described here, APCE (including variants with anisotropy detection) offers the ability to detect multiple unlabeled analytes; however, APCE separations are performed under nonequilibrium conditions, limiting that technique to detecting fairly long-lived affinity complexes. For APCE to be successfully applied, the separation time should be shorter than the half-life of complex (t1/2) given by

t1/2 ) ln 2/koff

(4)

where koff is the dissociation rate constant. For example, for an association rate constant (kon) of 3 ×104 M-1 s-1 and Kd of 100 nM, the koff ) 3 × 10-3 s-1 and the separation should be faster than 230 s (3.8 min). Detection of a complex with similar on rates but Kd ) 1 µM, however, would require a 23-s separation. While such fast separations are possible when using short capillaries and high electric fields,24,25,26 they are not routinely achieved. In comparison, if we consider using FACE with 100 nM fluorescent probe in the electrophoresis buffer, a protein complex with Kd ) 1 µM is predicted to result in a detection limit (S/N ) 3) of 5 nM protein. (This calculation uses the anisotropy values for BGTPγS and G protein complex.) Even a very weak affinity interaction, Kd ) 1 mM, is predicted to give a detectable anisotropy feature upon injection of a moderate concentration of protein (4.4 µM). FACE is therefore potentially applicable to the analysis of a wider range of affinities than is APCE-LIF. Weaker affinity systems are typically analyzed by ACE. In conventional ACE, the binding of injected receptor to ligand in the electrophoresis buffer must result in a shift in the receptor’s electrophoretic mobility in order to be detected;4 this becomes (24) Kennedy, R. T.; German, I.; Thompson, J. E.; Witowski, S. R. Chem. Rev. 1999, 99, 3081-3132. (25) Kennedy, R. T. Anal. Chim. Acta 1999, 400, 163-180. (26) Buchanan, D. D.; Jameson, E. E.; Perlette, J.; Malik, A.; Kennedy, R. T. Electrophoresis 2003, 24, 1375-1382.

Analytical Chemistry, Vol. 76, No. 24, December 15, 2004

7385

less likely as smaller or less charged ligands are used. In FACE, however, small fluorescent ligands (probes) are likely to undergo sizable changes in anisotropy as the result of binding to injected protein. In addition, conventional ACE requires multiple analyses and the comparison of protein mobility with and without ligand in the electrophoresis buffer, whereas in an affinity FACE assay, evidence of complex formation can be acquired from a single electropherogram. At the same time, the technique shares with ACE the advantage of being applicable over a wide range of affinities. A previous combination of ACE and anisotropy detection has been reported in which a fluorescent analyte was injected into buffer containing a large molecule binding partner.17 This approach does allow analysis of weaker affinity systems and it can

allows separation and detection of multiple unlabeled binding partners. Thus, FACE expands the utility of affinity-based CE methods by being applicable to a wide range of affinities and by providing sensitive detection of multiple unlabeled binding partners in a single assay. These properties allow the assay to be used for analyzing or screening complex mixtures for binding partners potentially enabling several important applications including discovering drugs or investigating cellular signal transduction mechanisms. ACKNOWLEDGMENT This work was supported by funding from an NSF grant (CHE 0232440).

detect binding with a single assay; however, it requires that each

Received for review July 16, 2004. Accepted October 6, 2004.

injected analyte be fluorescent whereas the method described here

AC0489566

7386

Analytical Chemistry, Vol. 76, No. 24, December 15, 2004