Anal. Chem. 1994,66, 1500-1506
Antibody-Mediated Fluorescence Enhancement Based on Shifting the Intramolecular Dimer == Monomer Equilibrium of Fluorescent Dyest Ai-Ping Wei,* Donald K. Blumenthai,g and James N. Herron'p* Departments of Pharmaceutics, Bioengineeringl and Pharmacology and Toxicologyl University of Utah, Salt Lake Cityl Utah 84108
A novel concept is described for directly coupling fluorescence emission to protein-ligand binding. It is based on shifting the intramolecular monomer + dimer equilibrium of two fluorescent dyes linked by a short spacer. A 13-residue peptide, recognized by a monoclonal antibody against human chorionic gonadotrophin (hCG), was labeled with fluorescein (F) and tetramethylrhodamine (T) at its N- and C-terminus, respectively. Spectral evidence suggests that when the conjugate is free in solution, F and T exist as an intramolecular dimer. Fluorescencequenching of fluorescein and rhodamine is -98% and -90%, respectively,due to dimerization. When the doublelabeled peptide is bound to anti-hCG, however, the rhodamine fluorescence increases by up to 7.8-fold, depending upon the excitation wavelength. This is attributed to the dissociation of intramolecular dimers brought about by conformational changes of the conjugate upon binding. Fluorescein fluorescence, on the other hand, was still quenched because of excitedstate energy transfer and residual ground-state interactions. Antibody binding also resulted in a -3.4-fold increase in fluorescence anisotropy of the peptide. These changes in intensity and anisotropy allow direct measurement of antigenantibody binding with a fluorescence plate reader or a polarization analyzer, without the need for separation steps and labeling antibodies. Because recent advances in peptide technology have allowed rapid and economical identification of antigen-mimicking peptides, the doublelabeled peptide approach offers many opportunities for developing new diagnostic assays and screening new therapeutic drugs. It also has many potential applications to techniques involving recombinant antibodies, biosensors, cell sorting, and DNA probes. Most clinical immunoassaysare indirect and heterogeneous. They require labeled antibodies for binding detection and solid carriers (membranes, beads, plates) for separating bound from free antigens.' These assays usually consist of multiple steps and demand highly trained operators and expensive reagents, making them inherently slow and difficult to automate. t Abbreviations: F,fluoresccin;T, tetramethylrhodamine;hCG, human chorionic gonadotrophin; pep, an epitope peptide of hCG: GSGSRLFGPSDTC, where the standard one-letter code is used to represent amino acid residues; Mab, monoclonal antibody; K4 dissociation constant; FAB, fast atom bombardment; BSA, bovine serum albumin. 8 Departments of Pharmaceutics and Bioengineering. I Department of Pharmacology and Toxicology. (1) Immunoassay. A Practical Guide; Chan, D. W., Perlstein, M. T., Eds.; Academic Press: New York, 1987.
1500 Analytical Chemistty, Vol. 66, No. 9, May 1, 1994
Homogeneous assays, on the other hand, are separation-free and offer the advantage of easy automation, random sample accessibility, and simple adaptation to existing instrumentation.1.2A number of homogeneous systems have been studied in the past 20 years based on detection schemes such as fluorescence polarization,3fluorescence energy transfer? timeresolved fluore~cence,~ enzyme inhibition? substrate inhibit i ~ nand , ~ chemiluminescence.s However, these approaches were usually limited to small molecules and in many instances required chemical labeling of antibodies. In our search for new detection strategies for high molecular weight substances, we have developed a method employing fluorescently labeled synthetic peptides as tracer molecules. These tracers allow direct fluorometric measurement of binding events without separation steps or labeled antibodies. Our approach uses antigen-mimicking peptides of six to eight amino acid residues in length as tracer antigens. Their primary structure may be either a sequential epitope found in the original antigen or a nonhomologous amino acid sequence recognized by the antibody. Recent advances in building and screening complex peptide libraries have allowed the rapid identification of such peptides for almost any monoclonal a n t i b ~ d y . ~ -Because l~ of their small size and chain flexibility, antigen-mimicking peptides offer unique opportunities for homogeneous assay of macromolecules. In a separate study, l 3 we exploited the size difference between the peptide (labeled with a single fluorophore) and the antibody-peptide complex to develop a fluorescence polarization immunoassay. It exhibited 100-fold higher sensitivity than a comparable assay employing the native macromolecular antigen as tracer.14 We believe that synthetic peptides may open a new window for extending previous assay techniques suitable for small (2) Gosling, J. P. Clin. Chem. 1990, 36, 1408-1427. (3) Watson,R.A. A.;Landon, J.;Shaw,E. J.;Smith,S.D. Clin. Chim.Acra 1976, 73, 51-55. (4) Ullman, E. F.; Khanna, P. L. Methods Enzymol. 1981, 74, 28-60. (5) Barnard, G.; Kohen, F.; Mikola, H.; Lovgren, T. Clin. Chem. 1989,35, 555559. (6) Burd, J. F.; et al. Clin. Chem. 1977, 23, 1402-1408. (7) Morris, D. L.; et al. Anal. Chem. 1981, 53, 658-665. (8) Campbell, A. K.; Roberts, P. A.; Patel, A. In Alternatiue Immunoassays; Collins, W. P., Ed.; John Wiley and Sons: Chichater, U.K., 1985. (9) Geysen, H. M.;Rodda, S. J.; Mason,T. J. Mol. Immunol. 1986,23,709-715. (10) Houghten, R. A.; Pinilla, C.; Blondelle, S. E.; Appel, J. R.; Dooley, C. T.; CUCNO,J. H. Nature 1991, 354, 84-86. (11) Lam, K. S.; Salmon, S. E.; Hersh, E. M.; Hruby, V. J.; Kazmierski, W. M.; Knapp, R. J. Nature 1991,354, 82-83. (12) Scott, J. K.; Smith, G. P. Science 1990, 249, 386-390. (13) Wei, A.-P.; Herron, J. N. Anal. Chem. 1993, 65, 3372-3377. (14) Urios, P.; Cittanova, N.; Jayle, M. FEBS Lett. 1978, 94, 54-58.
0003-2700/94/036&1500$04.50/0
0 1994 American Chemical Society
molecules to high molecular weight analytesof clinical interest. In this paper, we describe a system consisting of an epitope of human chorionic gonadotrophin (hCG), labeled with fluorescein (F) and tetramethylrhodamine (T). It is wellknown that some fluorescent dyes (fluoresceins, rhodamines, cyanines) form dimers in aqueous solution when they are at high concentrationsor within close proximity to each other.15-17 Dipole-dipole interactions within the resonating dimeric structurecan result in significant fluorescencequenching.lS2O For this reason, dye dimerization has been regarded as an adverse effect in biological applications.21 This phenomenon, however, may be used to advantageto quench the fluorescence of free tracers and to enhance the fluorescence of bound tracers if the dimer -L monomer equilibrium can be modulated by antibody-antigen reactions. In order to achieve this, two dye molecules are conjugated to the ends of a small peptide. When the conjugate is free, because the peptide is small and flexible, the dyes can form an intramoleculardimer which will exhibit low fluorescence intensity. Upon binding to its antibody, however, the dye-peptide conjugate may undergo conformational changes which can result in dissociation of the intramolecular dimer and increase in fluorescence intensity. When used in a noncompetitive assay, this property allows direct homogeneous detection of antibodies. When used in a competitive assay, high molecular weight antigens can be conveniently assayed by mixing the antibody with a doublelabeled epitope peptide (as tracer antigen). In the absence of any sample, fluorescence intensities are high due to the bound tracer. When a sample is added, analyte antigens can displacethe bound tracer from the antibody into bulk solution, resulting in a decrease in fluorescence intensity. Therefore, the net fluorescence decrease is related to the analyte concentration. This concept is depicted schematically in Figure 1. The intramolecular monomer + dimer equilibrium is essentially a molecular switch that can be toggled directly by binding events- process that would otherwise be difficult to achieve with high molecular weight protein antigens. Applications of this technology may well extend beyond clinical immunoassays to other biotechnological areas such as biosensors,22 DNA probes,23 enzyme assays,Z4 recombinant antibodies,Z5 peptide libraries:-' and 2D molecular assembl y .26
EXPERIMENTAL SECTION Cbemicals. 5-Carboxylfluoresceinsuccinimidyl ester and tetramethylrhodamine-5-(and 6-) maleimide were products of Molecular Probes (Eugene, OR). Mouse immunoglobulin (15) Rohatgi, K. K.; Singhal, G. S. J. Phys. G e m . 1966, 70, 1695-1701. (16) Rohatgi, K. K.; Mukhopadhyay, A. K. Chem. Phys. Leu. 1971,12,259-260. (17) West, W.; Pearce, S. J. Phys. Chem. 1965.69, 1894-1903. (18) Arbeloa, I. L. J. Chem. Soc.. Furuduy Trans. 2 1981, 77, 1735-1742. (19)Arbtloa, I. L. J. Chem. Soc., Furuduy Trans. 2 1981, 77, 1725-1733. (20) Arbeloa, I. L.; Ojcda, P. R. Chem. Phys. Len. 1982.87, 556-560. (21) Bailey, M. P.; Rocks, B. F.; Riley, C. J . Phurm. Biomed. Anal. 1987.5.649658. (22) Barnard, S. M.; Walt, D. R. Science 1991, 251,927-929. (23) Morrison, L. E.; Halder, T. C.; Stols, L. M. Anal. Biochem. 1989,183,231244. (24) Matayoshi, E. D.; Wang, G. T.; Krafft, G. A.; Erickson, J. Science 1990,247, 954-958. (25) Lemer, R. A.; Kang, A. S.; Bain, J. D.; Burton, D. R.; Carlos F. Barbas, I. Science 1992, 258, 1313-1 3 14. (26) Charych, D. H.; Nagy, J. 0.;Spevak, W.; Bednarski, M. D.Science 1993,261, 585-588.
LI
4
O
A
4I
Double Labeled Epitope Peptide (by Fluorescein & Tetramethylrhodamine)
+
Dominant Species in Aaueous Solution
3
'2
U
Antibody
.^
/
A
L
i
e
W.L
Spectrum of the stacked peptide conjugates
a a
(HighM.W. Analyte Antigen)
Competition
Spectrum of the double labeled peptide when bound
Flguro 1. Schematic lllustratfon of the concept of using an epltope peptide labeled wtth two fluoresoent dyes as tracer antlgens for the homogeneous detectionof both antibodies and high molecular weight
analytes.
G (IgG) and bovine serum albumin (BSA) were purchased from Sigma (St. Louis, MO). Anti-hCG monoclonal antibodies (Mab) and human chorionicgonadotrophin (reference standard, 550 international units (IU)/vial) were provided by Organon Teknika, Boxtel, The nether land^.^^ Antibody concentration was determined by absorption at 278 nm, using of 14. The molar concentraan extinction coefficient e)(:t tion of hCG was converted from international units per milliliter using a specific activity of 11 200 IU/mg and a molecular mass of 38 kDa.27 All solutionswere made in 100 mM phosphate buffer (pH 7.4) unless otherwise indicated. Epitope Mapping and Peptide Synthesis. A panel of 221 overlapping octapeptides derived from the primary sequence of hCG28$29 was screened for specific binding to a monoclonal antibody against hCG. The mapping experiments were performed by Chrion Mimotopes (Clayton, Australia) using the method of Geysen et al.9 The octapeptide showing the highest affinity was synthesized in bulk quantity using the F-moc m e t h o d ~ l o g yand ~ ~purified on reversed-phase HPLC to >95% purity. Its chemical identitywas confirmedby amino acid analysis and fast atom bombardment (FAB) mass spectrometry.31 Fluorescent Labeling of Peptide. The epitope peptide was first reacted with tetramethylrhodamine-5-maleimide in 50 mM phosphate buffer (pH 6) for 48 h at 4 O C to make pepT. Purified pepT was reacted overnight with 5-carboxylfluorescein succimidyl ester in 50 mM borate buffer (pH 8.5) to make FpepT. A gradient of acetonitrile in water was used to purify these conjugates.13 FAB mass spectrometry was used to confirm the chemical identity of FpepT. The concentration (27) Van Erp, R.; Gribnau, T. C. J.; van Sommmn, A. P. G.; Bloemets, H. P. J. J. Immunol. Merhods 1991,140,235-241. (28) Carlsen, R. B.; Bahl, 0.P.; Swaminathan, N. J. Biol. Chem. 1973,248,68106827. (29) Bellisario, R.; Carlsen, R. B.; Bahl, 0. P. 1. Biof.Chem. 1973,248,67966809. (30) Solid Phase Peptide Synfiresis, 2nd ed.; Stewart, J. M., Young, J. D., Ed.; Pierce: Rockford, IL, 1984. (31) Muss Specfromerry; McCloskey, J. A., Ed.; Academic Press: New York 1990.
Ana&tkaIChemktry, Vd. 66, No. 9, May 1, 1984
1501
of this conjugate was determined by its absorbance at 560 nm using a molar extinction coefficient),,;e:( of 37 500 M-l cm-l. Spectral Analysis. The visibleabsorption spectra of FpepT in the absence and presence of anti-hCG were subjected to multiple linear regression analysisusing the model where AFpep, A F ~ =~ aTA ~ p e p+ B A p e p ~+ 6
(1)
A p e p ~and , A F ~ , , are T the absorption spectra (400-650 nm) of single-labeled peptides (Fpep and pepT) and double-labeled peptide (FpepT), respectively; a and 0 are linear coefficients to be determined; and e is the residual term. The analysis was performed on an Apple Macintosh computer using Statworks (Cricket Software Inc., Philadelphia, PA). Fluorescence Measurements. Fluorescence spectra, intensity, and anisotropy measurements were taken with an ISS PC-1 fluorometer (ISS, Champaign, IL). An excitation wavelength of 493 or 561 nm (fwhm dispersion 4 nm) was used, and fluorescence emission was measured through a 589nm interference filter (fwhm = 10 nm; Oriel, Stratford, CT). Temperature was controlled at 6 OC using a water bath. In all titration experiments, the overall titrating volume added to the sample was less than 4% of the total sample volume. Binding Experiments and Data Analysis. Three different methods were used for evaluating binding affinity. In the first of these (method I), two identical solutions of FpepT (1.3 X lo-’ M) were titrated with 5-pL aliquots of either a stock solution of anti-hCG (sample) or a 1:l mixture of BSA and mouse IgG (reference). Fluorescence intensities in sample and reference cuvettes were denoted as I, and Ir,respectively. The enhancement factor was defined as
It can be shown that E and the total antibody concentration (Po)are related as follows: KdE Po=-
Em-E
+-L$ E,
(3)
where Po,Lo,Em,and Kd are the total antibody concentration, total FpepT concentration, maximum enhancement, and dissociation constant, respectively. The E vs Podata set was fit to this equation using Kaleidagraph (Abelbeck Software). In the second method (method 11), a sample solution of antihCG (1.4 X M) was titrated with 5-pL aliquots of stock FpepT. The same amount of FpepT was also added to a reference solution of BSA and mouse IgG (1.5 X lo-’ M). and The relationship between total FpepT concentration (Lo) E is given by the equation KdEm Lo = 2P,-Em - E Em-E
(4)
where the parameters are defined the same as in eq 3. The E vs Lodata set was fit to this equation using Kaleidagraph. In the third method (method 111), which involves competitive 1502
AnalytlcalChemistry, Vol. 66,No. 9, May 1, 1994
binding, a mixture of FpepT and anti-hCG was titrated with 5-pL aliquots of stock hCG solution in the sample cuvette. The reference intensity (Ir)was obtained by carrying out the same titration in a reference cuvette without anti-hCG. Values of E were calculated according to eq 2. If K1 and K2 are defined as the disassociation constants for hCG and FpepT, respectively, we can show that the enhancement factor ( E ) and the total hCG concentration (Ly) follow the equation
where Po and Lq are total antibody concentration and total tracer concentration, respectively. The E vs Ly data set was fit to this equation using Kaleidagraph to obtain K1 and Kz. Equations 3-5 were derived from the basic mass law of binding equilibrium. Readers should refer to Herron or Pesce for general derivation procedure^.^^^^^
RESULTS AND DISCUSSION HCC Epitope and Its Conjugate with Dyes. Of the 221 octapeptides tested, 2 showed strong interaction with the antihCG antibody: SRLPGPSD and RLPGPSDT. These results indicate that this monoclonal antibody recognizes a single linear epitope whose “core” sequence is the heptapeptide RLPGPSD. In order to conjugate fluorescent dyes to this peptide without compromising its binding properties, the following sequence was synthesized: GSGSRLPGPSDTC. Except for Cys, this peptide (pep) corresponds to a naturally occurring sequence near the C-terminus of hCG 0 chain. The measured molecular mass of 1234 Da for pep by FAB mass spectrometry was in excellent agreement with the predicted value of 1233.38 Da; amino acid analysis also gave results consistent with the predicted amino acid sequence. It should be noted that for a protein antigen of unknown primary sequence or a nonprotein antigen, it is still possible to identify high-affinity peptides using the Geysen method or other more recent approaches based on peptide libraries.llJ4 The single-labeled pep-GSGSRLPGPSDTC-T (pepT)showed a peak at 1715 Da on the FAB mass spectrogram, in good agreement with the predicted value of 1714.88 Da. The double-labeled pep had the chemical structure of F-GSGSRLPGPSDTC-T (FpepT) and is illustrated in Figure 2A. Its molecular mass, however, was found to be 2091 Da, rather than the expected value of 2073.19 Da. We attribute this 18-unit difference in mass to the entrapment of a bonded water molecule in the conjugate due to the peptide hydrophilicity and two bulky fluorophores. Fluorescein and tetramethyrhodamine were used in this study because they are commonly used in fluorescent immunoassays21and constitute a well-characterized fluorescence energy-transfer pair.4 Also, the large Stoke’s shift of fluorescence emission (-90 nm) exhibited by the F-T pair helps to eliminate interference from scattered light and serum fluorescen~e.~~ (32) Herron, J. N. In Fluorescein Hapten. An Immunological Probe; Vass, E.W . , Jr., Ed.; CRC Press: Boca Raton, FL, 1983; pp 53-55. (33) Fluorescence Spectroscopy An Introduction for Biology and Medicine; Pescc, A. J., Rosen, C.-G., Pasby, T. L., Us.; M a r 4 Dckker: New York, 1971. (34) Houghten, R. A.; Appcl, J. R.; Blondelle, S. E.; Cuervo, J. H.; Dooley, C.T.; Pinilla, C. BioTechniques 1992, 13, 412-421.
:A1
111
v)
C
A
3 1owoe
I
500
550
e
600
io
Wavelength (nm)
E VJ 5s *
5000
.CI
B 4000-
I
400
450
500
550
600
650
Wavelength (nm) Figure 2. (A) Structure of the double-labeled tracer consistlng of an epltope peptlde of hCG labeled wlth fluorescein (F) and tetramethylrhodamine (T). (B) Absorption spectra of the FpepT conjugate (solid line). The dashed line shows a mumple llnear regresslon fit of the measured spectra according to eq 1. Reskiuaisbetween the measured and fitted values are shown In inset.
Dye dimerization is usually characterized by changes in absorption spectra and significant quenching of fluorescence.18-20The visible absorption spectrum of monomeric fluorescein or rhodamine dyes usually consists of an intense transition at the red edge and a shoulder located 2C-30 nm to the blue. The formation of dimers results in a "flip-flop" of these relative peak intensities so that the shorter wavelength transition becomes hyper~hromic.~~ Comparison of the absorption spectra of FpepT with those of Fpep and pepT revealed that the major absorption peak of fluorescein was blue-shifted by 2 nm, while that of rhodamine was red-shifted by 9 nm when in the FpepT conjugate. The result of spectral analysis according to eq 1 is shown in Figure 2B. If there is no ground-state interaction between F and T in the FpepT conjugate, we should expect the residual values (e) to be random with a mean value of zero. Systematic changes in residual values shown in Figure 2B are evidence of significant ground-stateinteractions between F and T. As expected, these static interactions have profound effects on fluorescence quenching. The intrinsic fluorescence of rhodamine in the FpepT conjugate was 10-fold lower (Le., 90%quenching) than in the pepT conjugate, as a consequence of dimerization with fluorescein (Figure 3A). On the other hand, the fluorescence of fluorescein in the FpepT conjugate was 64-fold lower (Le., 98%quenching) than in the Fpepconjugate (Figure 3B). This higher quenching efficiency was probably due to both dimerization (a static process) and excited-stated energy transfer to rhodamine (a dynamic process). These results (35) Wolfbeis, 0. S.;Lciner, M.Anal. Chim. Acta 1985, 167, 203-215. (36) Edmundson, E.;Ely, K.;Herron, J. Mol. Immunology 1984, 21, 561-576.
d
500
550
600
650
Wavelength (nm) Figure 3. (A) Technical fluorescence spectra of Fpep (I), pepT (2), and FpepT (3) when excited at 493 nm (lo-' M). Compared to Fpep, the fluorescein fluorescence was quenched by 98% In FpepT. (B) Technical fluorescencespectra of pepT (1) and FpepT (2) when excited at 550 nm (lo-' M). Compared to pepT, the rhodamine fluorescence was quenched by 90% in FpepT. Spectra were measured in 100 mM phosphete buffer, pH 7.4 at 6 'C.
suggest that free FpepT indeed exists as an intramolecular dimer. Binding of FpepT with Anti-hCG Msb. Figure 4A shows the fluorescence spectra of FpepT with increasing amounts of anti-hCG Mab. As anti-hCG antibody was added, the fluorescence of rhodamine (Amx = 570 nm) increased up to 5-fold as a result of diminished static interactions with fluorescein. This phenomenon was accompanied by changes in the absorption spectra of FpepT. Figure 4B illustrates that the major absorption peaks of FpepT shifted toward each other when bound to anti-hCG. Also, the longer wavelength absorption band of rhodamine (56 1 nm) increased by a factor of 1.6 because of dimer dissociation. As a result, the absorption spectrum of FpepT in the presence of antibody was better approximated by the simple sum of the spectra obtained for Fpep and pepT (Figure 4B, inset). This indicated that the absorption bands of F and T in FpepT were less perturbed when the conjugate was bound-further evidence that F and T indeed dissociated upon binding to the antibody. However, it should be noted that the measured spectrum of the bound FpepT does not completely coincide with the fitted spectrum due to residual spectral perturbations. Even though the static interactions between F and T were = 515 nm) diminished after binding, the fluorescence (A,, of fluorescein remained quenched (Figure 4A). This was attributed, in part, to excited-state energy transfer from F to T. The distance at which 50% energy-transfer efficiency Ana&ticalChemistry, Vol. 88, No. 9, May 1, 1994
1503
W
Qe+O 500
525
550
575
600
625
650
Wavelength bun)
I
20 B
le-?
38-7 Ab (M)
2e-7
4e-7
5e-7
F w e5. Fluorescenceenhancementfactor (9as a function of antbody concentration. Ex = 493 nm and Em = 589 nm. The inset shows fluorescence intensities of FpepT (1.3 X lo-' M) as a function of antUlCG (I,, filled circle), or a 1:1 mixture of BSA and mouse IgG (I,, open circle). A value of 2.2 X 10-7Mwas obtained for 6,and 4.1 for according to method I. When Ex = 561 nm, a value of 6.8 was obtained for (data not shown). Measurements were made in 100 mM phosphate buffer, pH 7.4 at 6 OC. In all titration experiments, the overall titrating volume added to the sample was less than 4 96 of the total sample volume.
e,"
500 5IO 600 650 wnv.L.rylh (nm)
400 4SO
I
4 0 0 4 5 0 5 0 0 5 5 0 6 0 0 6 5 0
Wavelength (nm) Flgure 4. (A) Technical fluorescence spectra of FpepT M) as aliquots of anti-hCG Mab were added (Ex = 493 nm). The total antihCGconcentration changed from zero to 2.6 X lO-'M at an increment of 3.7 X 10" M wlth 2 % dilution. Note that while the rhodamine fluorescence was enhanced, the fluorescein fluorescence remained constant. (B) Comparison of the absorption spectra of FpepT (6.5 X lod M) in the absence (dashed) and presence (solid) of anti-hCG antibodies (7.5 X loa M). The spectra after binding coukl be better approximated with eq 1 (Inset). Spectra were measured In 100 mM phosphate buffer, pH 7.4 at 6 OC.
occurs (R,)is 54 8, for the F-T pair.4 Even if the peptide became fully extended after binding, the end-to-end distance of 47 8L37 would still allow 70% energy-transfer efficiency. This effect, plus the residual static quenching, accounts for the low fluorescence of fluorescein. Because fluorescence intensity is a relative parameter, the constant fluorescence at 5 15nm may serve as an internal self-reference point to correct for instrument fluctuations and to establish intrinsic standard curves. For example, one could relate the intensity ratio between 570 and 515 nm with the mole fraction of bound FpepT, which can then be used to calculate the antibody or analyte concentration, according to the system composition. Figure 5 is a typical titration for the binding of FpepT to anti-hCG Mab (fluorescence vs antibody concentration). While the addition of anti-hCG resulted in a gradual increase in fluorescence intensity, the same amount of nonspecific proteins (BSA + mouse IgG) did not have any effect on the fluorescence of FpepT (Figure 5, inset), indicating that the (37) Proteins Structure and Molecular Properties; Creighton, T. E., Ed.; W.H. Freeman and Co.: N e w York, 1984.
1504 Analytical Chemistry, Vol. 66, No. 9,May 1, 1994
enhanced fluorescence was a result of specific binding. The fluorescence enhancement factor ( E ) as a function of antibody concentration (Po)was fit to eq 3. Themaximum enhancement factor at 589 nm was found to be 4.1 (Ex = 493 nm) or 6.8 (Ex = 561 nm, data not shown). According to eq 2 (Zs/Zr = 1 + E ) , these results indicate that when FpepT is completely bound, its fluorescence at 589 nm increased by 5.1- or 7.8-fold. The dependence of on excitation wavelength reflects the molecular mechanism of fluorescence enhancement and may be interpreted as follows. When the major absorption transition of rhodamine was excited (Ex = 561 nm), the 7.8-fold increase in fluorescence was brought about by diminished static interactions and an increase in absorptivity at 561 nm (Figure 4B). When the major absorption transition of fluorescein was excited (Ex = 493 nm), on the other hand, the value of E:' was determined by fluorescence energy transfer from F to T. As will be discussed later, binding of FpepT to antibody increased the angle between the absorption dipole of F and the emission dipole of T. This reduced the energy-transfer efficiencyfrom F to T and yielded a smaller value. A dissociation constant (&) of (2.2 f 0.3) X l C 7 ( N = 6) was obtained from Figure 5. In a previous article where the same peptide was studied with one label (Le., pepT), a value of 0.67 X M was determined for K d . I 3 The 3-fold decrease in the binding affinity of FpepT relative to pepT was probably due, in part, to the free energy necessary to dissociate the intramolecular dimers. In a separate binding experiment, an antibody solution of fixed concentration was titrated with aliquots of FpepT according to method I1 described earlier. Results are shown in Figure 6. The E vs Lo data set was fit with eq 4, yielding a dissociation constant of (2.1 f 0.4) X M ( N = 3). This is in excellent agreement with the result of Figure 5. Exchange of FpepT from Binding Complex by bCG. Competitive binding experiments were carried out according to method 111 to examine the specificity and reversibility of
40000
i
I
0'4
1
Q)
c 0
ln
e a
ii
-
.-m>
1
-mm a
F
O.Oe+O
I 5.0e-8
1.5e-7
1.08-7
2.0e-7 2e-7 3e-7 4e-7 5e-7 Concentration (MI Flguro 8. Fluorescence anisotropy of FpepT (1.6 X 1od M) at 589 nm. When exclted at 550 nm, the anisotropy increased with antChCG concentration (filled circles) and remained constant with BSA IgG (open circles). The error bars represent the standard devletions of three different experlments. The anisotropy of FpepT was A, = 0.1087 f 0.0014 (N = 17) when unbound and Increased m e than %fold when bound. A value of 0.3444 for bound FpepT was determined from a double-reclprocal plot of l/(A A,) vs l/Pw When exclted at 490 nm, the anlsotropy of unbound FpepT was 0.0436 f 0.0023 (N= 17). Increasing antChCG concentration (filled squares) resulted In further depolarkatlon, while BSA 4- IgG (open squares) had no effect. Condltlons are the same as In Figure 5. Oe+O
F p W (MI
Flgurs6. FluorescenceIntensitiesas8 function of FpepT concentration. Ex = 561 nm and Em = 589 nm. The sample cuvette contained 1.4 X lo-' ManU-hCG(fllledcircles, ZJ, and the referencecuvettecontained 1.5 X lo-' M BSA IgG (open circles, ZJ. A K, value of 2.1 X lo-' M was obtained by fittlng the Evs FpepT data set according to method 11. This value was in excellent agreement with the result of Figure 5. Condltlons are the same as In Figure 5.
+
I
1
w hca Conoentrallon (Y)
4
'c
0.4
0.0 0
20
40
60
80
100
hCG (nM) Flguo 7. Fluorescenceenhancementfactor (9 of FpepTas a functlon of hCG concentration. Ex = 561nm and Em = 589 nm. The dlssociation constants for hCG and FpepT was found to be 4.9 X 10-lo and 2.4 X lo-' M, respecthrely, according to method 111. A mixture of FpepT (5.5 X 10-8M)andantChCGMab(4.5X 104M)wastltrated byaliquots of hCG stock solution (Inset, filled clrcle). The reference cuvette containedonly FpepTwithout any antibody(lnset,open clrcle). Conditions are the same as In Figure 5.
anti-hCG binding to FpepT. Figure 7 shows the fluorescence intensity of FpepT in the presence of antibody as a function of hCG concentration (inset). The decrease was specific to hCG because little change was observed when BSA and mouse IgG was added. The enhancement factor ( E ) vs hCG concentration data were fit to eq 5 . The Kavalues for antibody binding to hCG and FpepT were found to be 4.9 X 1O-Io and 2.4 X lo-' M, respectively. In spite of the 490-fold lower affinity exhibited by the tracer relative to hCG, the typical detection limit of hCG was -1 X le9M. This level of sensitivity is at least 100-fold higher than comparable homogeneous assays for hCG.14 We attribute this performance to the nature of fluorescenceenhancement, which allows preferential measurement of the bound species. If the binding affinity of the tracer can be improved to the same level as the native antigen, the detection limit of hCG should fall within the range of lo-'* M.13 The limiting parameter in that case
le-7
+
-
will become the instrument sensitivity and the brightness of fluorophores. There are two ways to improve peptide affinities for anti-hCG. One is to use the existing peptide as a template to systematically substitute both D- & L-amino acids at key positions and search for better binder~.3~ The other method is to search for tight binders from a more complex combinatorial peptide library of over 200 million peptides.I1J4 Fluorescence Anisotropy. The molecular mechanism of fluorescence enhancement may be further understood by examining the fluorescence anisotropy of FpepT at 589 nm as a function of anti-hCG concentration. When excited at 550 nm, the fluorescence anisotropy of free FpepT was 0.1087 f 0.0014 ( N = 17). It increased with antibody concentration up to a maximum value of 0.3444 (Figure 8). A similar result was obtained for the fluorescence anisotropy of fluorescein (measured at 515 nm) when excited at 490 nm (data not shown). This dramatic increase in anisotropy was due to the difference in rotational mobility between FpepT (MW = -2000) and the antibody (MW = 150 000). It suggests that the rotational motion of FpepT was significantly hindered upon binding to anti-hCG.39 It is important to note that this increase in fluorescence polarization is also accompanied by an increase in fluorescence intensity. In this context, the double-labeled tracer offers significant advantages over its single-labeled counterpart because the level of total fluorescence is biased in favor of the concentration of the bound species. The effects of intensity change on fluorescence polarization immunoassays are discussed in detail by Wei and Herren." When the tracer was excited at 490 nm, however, the fluorescence anisotropy at 589 nm for free FpepT was 0.0436
-
~
(38) Geysen, H. M. U S . Patent 4,833,092, 1989. (39) Principles ofNuorescence Spcrroscopy; Lakowicz, J. R., Ed.;Plenum: New York, 1983.
Analytical Chemistry, Vol. 88, No. 9, May 1, 1994
1505
f 0.0023 ( N = 17), indicating significant fluorescence depolarization. This was probably caused by the noncollinearity between the absorption dipole of fluorescein and the emission dipole of rhodamine. We know that the intrinsic anisotropy (A,) of a fluorophore in the absence of other depolarization effects is described by ( A , = (2/5)((3 cos2 a - 1)/2), where a is the angular displacement between the absorption and emission dipoles. Typically, the largest A, values are observed for the longest wavelength absorption band because absorption and emission involving the same electronic transition normally have collinear moments.39 Under otherwise identical conditions, the difference in observed anisotropy at two excitation wavelengths reveals the difference in a. A smaller anisotropy value indicates a larger angle a, and vice versa. Interestingly, as the antibody concentration was increased, more depolarization occurred (Figure 8). This suggested that the angle between the molecular planes of fluorescein and rhodamine had in fact become larger after binding.39 Consistent with the previous discussion, these results provide additional evidence that the observed fluorescence enhancement was indeed due to conformational changes that occurred in FpepT upon binding to antibody. Fluorophore Selection. Proper labeling of the peptide with fluorophores is probably the most critical consideration in designing new tracers. Although the core peptide is usually 6 to 8 amino residues in length, a longer sequence is needed in order to reduce steric hindrance. The peptide length, in turn, determines the fluorophore selection. Assuming a typical dimerization constant (&m) of 2500 M-1 (as exhibited by rhodamines), the concentration of fluorescent dyes required to form 90% dimers is 1.8 X lo-* M, which is 1 X 1019 molecules/cm3. Under this condition, the average distance between two dyes is'60 A. Sincethe distance between adjacent a-carbons in an extended polypeptide is 3.63 A,3760 A should correspond to '16 residues. In other words, if the dye has a Kdm of 2500 M-l, the two fluorophores should be spaced by 16 residues according to this calculation. If the peptide is shorter than a 16-mer, fluorophores with smaller Kdm may be used in order to achieve effective intramolecular dimerization, and vice versa. The dimerization constants of fluorescein, eosin, rhodamine B, and rhodamine 6G are 5,110,2100, and 5600 M-l, r e s p e ~ t i v e l ywhile , ~ ~ ~that ~ ~of~cyanines ~ varies in the range of 103-106 M-*, depending upon the alkyl chain length of their str~ctures.1~In addition to the above considerations, the attachment of dyes to a spacer may alter their dimerization properties. Also, the conformational restrictions imposed on the peptide by intramolecular dimerization may change its binding properties with antibody.
CONCLUSIONS We have shown that fluorescent dye dimerization-a phenomenon that has largely been regarded as an adverse effect in biological applications because of fluorescence quenching-can be effectively used to advantage as tracer (40)Aguirresacona, I. U.; Arbeloa, F. L.; Arbeloa, I. L. J. Chem. Educ. 1989,66, 866-869. (41) Rohatgi, K. K.; Mukhopadhyay, A. K. J. Indian Chem. SOC.1972.49.131 11320. (42) Hlady, V.;van Wagenen, R. A.; Andradc,S. D. In Protein Adsorption; Andrade, J . D., Ed.; Plenum Press: New York, 1985; pp 81-119.
1508 Analytical Chemism, Voi. 66, No. 9, May 1, 1994
antigens in homogeneous immunoassays. This was achieved by coupling the antibody-antigen binding event to the dissociation of intramolecular dimers of two fluorescent dyes linked via a short antigen-mimicking peptide. When the double-labeled peptide is free, the two labels form an intramolecular dimer and their fluorescenceis quenched. When the peptide binds to an antibody, the fluorescence increases because of dimer dissociationbrought about by conformational changes in the dye-peptide conjugate. With the recent development of techniques for constructing and screening peptide libraries, antigen-mimicking peptides should become readily available at low cost. This makes the present method not only feasible but also practical. The double-labeled peptide technique is a promising assay, especially for large molecular analytes, given that fluorescent homogeneous assays are simpler, more reproducible, and as sensitiveas the cumbersome, separation-based methodologies currently in use. This method can be readily adapted to existing instrumentation such as 96-well fluorescence plate readers and fluorescence polarization analyzers. Compared to the polarization-based methods, however, this technique is faster, less expensive, and more sensitive (given similar other conditions) because there is no need for polarized light. It may also have potential impact in biosensors. For example, one major obstacle in the development of evanescent immunosensors has been parasitic fluorescence from unbound tracer antigens in bulk solution.42 However, this problem can be avoided by employing a tracer that only fluoresceswhen bound to the immobilized antibody and is nonfluorescent when unbound in bulk solution. The reduced background fluorescence should significantly improve sensor sensitivity. Because this method does not require chemical labeling of antibodies, it may be used for screening libraries of recombinant antibodies. Finally,in analogy to the antibody-antigen system, the concept of intramolecular dye dimerization might also be utilized in sequence-specific DNA assays. If a DNA probe is used to link two dyes, hybridizing with its target DNA will bring about a transition from intramolecular dimers to monomers. The target DNA can, therefore, be measured from the net fluorescence increase. ACKNOWLEDGMENT We thank AKZO Corporate Research America, Inc., for their generous financial support. Additional support to this project was also provided by NIH Grant AI 22898 awarded to J.N.H. and a predoctoral fellowship awarded to A,-P.W. from the NIH Biotechnology Training Grant GM 08393. We acknowledge our colleagues C.A. Gentry, W. Jiskoot, R. W. Schackmann, and A. H. Terry for assistingin peptide synthesis, stimulating discussions,and critical reading of the manuscript.
Recelved for revlew October 29, 1993. Accepted February 15, 1994."
Abstract published in Advance ACS Abstracts. April 1, 1994.
