Anal. Chem. 2007, 79, 6312-6318
Tandem Dye Acceptor Used To Enhance Upconversion Fluorescence Resonance Energy Transfer in Homogeneous Assays Terhi Rantanen,* Henna Pa 1 kkila 1 , Laura Ja 1 msen, Katri Kuningas, Telle Ukonaho, Timo Lo 1 vgren, and Tero Soukka
Department of Biotechnology, University of Turku, Tykisto¨katu 6A, FIN-20520 Turku, Finland
In fluorescence resonance energy transfer (FRET)-based assays, spectral separation of acceptor emission from donor emission is a common problem affecting the assay sensitivity. The challenge derives from small Stokes shifts characteristic to conventional fluorescent dyes resulting in leakage of donor emission to the measurement window intended only to collect the acceptor emission. We have studied a FRET-based homogeneous bioaffinity assay utilizing a tandem dye acceptor with a large pseudo-Stokes shift (139 nm). The tandem dye was constructed using B-phycoerythrin as an absorber and multiple Alexa Fluor 680 dyes as emitters. As a donor, we employed upconverting phosphor particles, which uniquely emit at visible wavelengths under low-energy infrared excitation enabling the fluorescence measurements free from autofluorescence even without time-resolved detection. With the tandem dye, it was possible to achieve four times higher signal from a single binding event compared to the conventional Alexa Fluor 680 dye alone. Tandem dyes are widely used in cytometry and other multiplex purposes, but their applications can be expanded to fluorescencebased homogeneous assays. Both the optimal excitation and emission wavelengths of tandem dye can be tuned to a desired region by choosing appropriate fluorophores enabling specifically designed acceptor dyes with large Stokes shift. Fluorescence-based detection systems have established their position in diagnostics within last decades due to their versatility and convenience of use.1,2 However, there are some fundamental problems related to fluorescence measurements. Biological matrixes and other sample materials generate autofluorescence under excitation and thereby cause an elevated background signal impairing assay sensitivity. Long-lifetime fluorescent labels (e.g., lanthanide chelates and cryptates) with time-resolved detection technology have overcome this problem by eliminating shortlifetime autofluorescence.3,4 Colorful samples can yet interfere with * Corresponding author. Telephone: +358-2-333-8095. Fax: +358-2-333-8050. E-mail:
[email protected]. (1) Soini, E.; Hemmila¨, I. Clin. Chem. 1979, 25, 353-361. (2) Hemmila¨, I. Clin. Chem. 1985, 31, 359-370. (3) Siitari, H.; Hemmila¨, I.; Soini, E.; Lo¨vgren, T.; Koistinen, V. Nature 1983, 301, 258-260.
6312 Analytical Chemistry, Vol. 79, No. 16, August 15, 2007
measurements in separation-free homogeneous assays5,6 by absorbing the excitation and emission light. This problem can be avoided by utilizing infrared excited upconverting phosphor (UCP) particles7 consisting of host material doped with lanthanide ions8 to exploit such excitation and anti-Stokes emission wavelengths that are compatible with most colorful matrixes, even with commonly used whole blood.9 This finding can enable the construction of simpler and more rapid diagnostic systems, as there is no longer a need for pretreatment of the whole blood sample (e.g., removal of cells) before fluorescence detection. Fluorescence-based homogeneous assay methods usually exploit fluorescence resonance energy transfer (FRET), where energy is transferred nonradiatively from one fluorescent molecule (donor) to a second fluorophore (acceptor). According to Fo¨rster, the excitation spectrum of the acceptor must overlap with the emission spectrum of the donor.10,11 The relative orientation of the transition dipoles of the participants also has an influence on the efficiency of energy transfer, and the FRET process is strongly dependent on the distance between the participants. Therefore, the closer proximity of acceptors bound to a donor compared to the unbound acceptors is utilized in homogeneous assay principles. Ideally, in order to avoid interfering background, the emission peak of an acceptor in FRET-based homogeneous assay should be located at a region, where no emission from a donor is detected. The choice of measurement window is especially important with lanthanide-based probes due to the sharp and intense emission peaks of lanthanide ions along with almost nonexistent background at the wavelengths few dozens of nanometers aside from the peaks.12 In contrast, conventional fluorescent donors may have an emission peak with a red tail extending to the higher wavelengths, complicating the discrimination of acceptor emission. (4) Hemmila¨, I.; Dakubu, S.; Mukkala, V. M.; Siitari, H.; Lo¨vgren, T. Anal. Biochem. 1984, 137, 335-343. (5) Ullman, E. F. J. Clin. Ligand Assay 1999, 22, 221-227. (6) Ullman, E. F. J. Chem. Educ. 1999, 76, 781-788. (7) Corstjens, P. L.; Li, S.; Zuiderwijk, M.; Kardos, K.; Abrams, W. R.; Niedbala, R. S.; Tanke, H. J. IEE Proc.: Nanobiotechnol. 2005, 152, 64-72. (8) Soukka, T.; Kuningas, K.; Rantanen, T.; Haaslahti, V.; Lo ¨vgren, T. J. Fluoresc. 2005, 15, 513-528. (9) Kuningas, K.; Pa¨kkila¨, H.; Ukonaho, T.; Rantanen, T.; Lo ¨vgren, T.; Soukka, T. Clin. Chem. 2007, 53, 145-146. (10) Fo ¨rster, T. Ann. Phys. 1948, 2, 55-75. (11) Stryer, L. Annu. Rev. Biochem. 1978, 47, 819-846. (12) Bunzli, J. C. G.; Piguet, C. Chem. Soc. Rev. 2005, 34, 1048-1077. 10.1021/ac070376w CCC: $37.00
© 2007 American Chemical Society Published on Web 07/12/2007
On the other hand, the spectral overlap between the acceptor excitation and the emission of the donor should be as complete as possible to achieve efficient energy transfer. Often one has to compromise either on the emission or on the excitation requirements due to the characteristically small Stokes shifts (the difference between the wavelengths of excitation and emission maximum13) of acceptor dye molecules. Typically, the Stokes shift of a bright organic fluorescent probe is ∼20 nm, and therefore, even extremely high quality band-pass filters do not ensure the favorable spectral separation. Quantum dots14-18 with their size-tunable emission maximum and Stokes shift over 200 nm can offer a solution to the spectral separation in homogeneous FRET-based assays. However, complications with their surface functionalization and synthesis still limit the use of quantum dots.14 Another solution is the polystyrene nanospheres loaded with two or more fluorescent dyes.19 These spheres contain dye, which has an excitation peak overlapping maximally with the excitation source, and another dye or set of dyes, that are capable of efficiently transferring the energy to the longest-wavelength dye in the sphere. Hence, a greater Stokes shift (over 100 nm) is achieved. The dye content in the sphere should be carefully optimized for each component in order to gain maximal loading, but still avoid self-quenching. However, both the distribution and the distance of the dye molecules in a sphere are random, complicating the repeatability of the loading. Besides those two particulate probes mentioned above, it is possible to achieve pseudo-Stokes shift over 100 nm with tandem dyes.20,21 Tandem dyes are also called bichromophoric dyes and energy-transfer cassettes. As the name tandem suggests, there are two covalently conjugated dye molecules instead of one. The first dye (absorber) is able to transfer energy to the second dye (emitter) in proximity. In order to enable the intramolecular energy transfer in a tandem dye, the excitation spectrum of the emitter must overlap with the emission spectrum of the absorber and the distance should be less than 10 nm. There are several published applications of tandem dyes. These dyes are used as laser dyes to improve the efficiency of lasers and to reduce resonator loss.22,23 Multiplex assays benefit most from the tandem dyes, and there are numerous publications especially on the field of DNA detection24-26 and cytometry.27-29 (13) Lakowicz, J. R. Principles of fluorescence spectroscopy; Plenum press: New York, 1983. (14) Medintz, I. L.; Uyeda, H. T.; Goldman, E. R.; Mattoussi, H. Nat. Mater. 2005, 4, 435-446. (15) Liu, T. C.; Liu, B. S.; Zhang, H. L.; Wang, Y. J. Fluoresc. 2005, 15, 729733. (16) Charbonniere, L. J.; Hildebrandt, N.; Ziessel, R. F.; Loehmannsroeben, H. G. J. Am. Chem. Soc. 2006, 128, 12800-12809. (17) Willard, D. M.; Carillo, L. L.; Jung, J.; Van Orden, A. Nano Lett. 2001, 1, 469-474. (18) Clapp, A. R.; Medintz, I. L.; Mauro, J. M.; Fisher, B. R.; Bawendi, M. G.; Mattoussi, H. J. Am. Chem. Soc. 2004, 126, 301-310. (19) Bhalgat, M. K.; Haugland, R. P.; Pollack, J. S.; Swan, S. J. Immunol. Methods 1998, 219, 57-68. (20) Burghart, A.; Thoresen, L. H.; Chen, J.; Burgess, K.; Bergstrom, F.; Johansson, L. B. A. Chem. Commun. 2000, 2203-2204. (21) Wan, C. W.; Burghart, A.; Chen, J.; Bergstrom, F.; Johansson, L. B. A.; Wolford, M. F.; Kim, T. G.; Topp, M. R.; Hochstrasser, R. M.; Burgess, K. Chem.-Eur. J. 2003, 9, 4430-4441. (22) Arden, J.; Deltau, G.; Huth, V.; Kringel, U.; Peros, D.; Drexhage, K. H. J. Lumin. 1991, 48-9, 352-358. (23) Su, J. H.; Tian, H.; Chen, K. C. Dyes Pigm. 1996, 31, 69-77. (24) Berti, L.; Medintz, I. L.; Tom, J.; Mathies, R. A. Bioconjugate Chem. 2001, 12, 493-500.
Tissue sections have been analyzed with a confocal laser scanning microscope by utilizing tandem dyes along with conventional dyes.30 Homogeneous hydrolytic enzyme activity measurement has also been reported.31 In this application, the tandem dye becomes functional after an enzymatic cleavage of the protective group. Tong et al. developed a trifluorophore and used it for a genetic mutation analysis.32 Another inventive approach was to conjugate a prompt emitter dye to a long-lifetime absorber resulting in a tandem dye with a long emission decay time.33 The aim of this study was to utilize the large pseudo-Stokes shift of the tandem dye and the far-red region of the spectrum in a homogeneous assay even though the donor particulate emits strongest at the green wavelengths. In this research, we have studied a tandem dye as an acceptor molecule in a previously described homogeneous upconversion FRET-based assay utilizing a UCP particle as a donor.34 EXPERIMENTAL SECTION Reagents. Fluorescent phycobiliprotein, B-phycoerythrin (BPE), was purchased from Cyanotech Corp. (Kailua-Kona, HI), and small molecular weight fluorophore, Alexa Fluor 680 (AF680) succinimidyl ester, was from Molecular Probes Invitrogen (Paisley, UK). D-Biotin was from Sigma-Aldrich (St. Louis, MO) and bovine serum albumin fraction V (BSA) from Bioreba (Nyon, Switzerland). Infrared to visible upconverting anti-Stokes phosphor PTIR550/F consisting of NaYF4 host lattice doped with Yb3+ and Er3+ ions was purchased from Phosphor Technology Ltd. (Stevenage, UK). An aqueous solution of an ammonium salt of a poly(acrylic acid) Additol XW330 (MW 30 000-50 000) was obtained from Surface Specialties Austria GmbH (Werndorf, Austria). Detergents Tween 20 and Tween 85 were purchased from E. Merck (Darmstadt, Germany), and streptavidin was from SpaBioSpa (Milan, Italy). N-(3-Dimethylaminopropyl)-N′-ethylcarbodiimide hydrochloride (EDC) and N-hydroxysulfosuccinimide sodium salt (sulfo-NHS) were from Fluka (Buchs, Switzerland). Assay buffer (50 mM Tris-HCl, pH 7.8, containing 9 g/L NaCl, 0.5 g/L NaN3, 5 g/L BSA, 0.1 g/L Tween 40, 0.5 g/L bovine γ-globulin, and 20 µM DTPA) was from Innotrac Diagnostics (Turku, Finland). Conjugation and Biotinylation of Acceptor Dyes. Tandem dye was constructed from BPE conjugated to several AF680 dyes and biotin molecules. Carrier protein dye was a mimic of tandem (25) Jiao, G. S.; Thoresen, L. H.; Kim, T. G.; Haaland, W. C.; Gao, F.; Topp, M. R.; Hochstrasser, R. M.; Metzker, M. L.; Burgess, K. Chemistry 2006, 12, 7616-7626. (26) Tyagi, S.; Marras, S. A. E.; Kramer, F. R. Nat. Biotechnol. 2000, 18, 11911196. (27) Gruber, R.; Reiter, C.; Riethmuller, G. J. Immunol. Methods 1993, 163, 173179. (28) Roederer, M.; Kantor, A. B.; Parks, D. R.; Herzenberg, L. A. Cytometry 1996, 24, 191-197. (29) Tjioe, I.; Legerton, T.; Wegstein, J.; Herzenberg, L. A.; Roederer, M. Cytometry 2001, 44, 24-29. (30) Uchihara, T.; Kondo, H.; Akiyama, H.; Ikeda, K. J. Histochem. Cytochem. 1995, 43, 103-106. (31) Takakusa, H.; Kikuchi, K.; Urano, Y.; Kojima, H.; Nagano, T. Chem.-Eur. J. 2003, 9, 1479-1485. (32) Tong, A. K.; Li, Z. M.; Jones, G. S.; Russo, J. J.; Ju, J. Y. Nat. Biotechnol. 2001, 19, 756-759. (33) Maliwal, B. P.; Gryczynski, Z.; Lakowicz, J. R. Anal. Chem. 2001, 73, 42774285. (34) Kuningas, K.; Rantanen, T.; Ukonaho, T.; Lo ¨vgren, T.; Soukka, T. Anal. Chem. 2005, 77, 7348-7355.
Analytical Chemistry, Vol. 79, No. 16, August 15, 2007
6313
dye, but the absorber dye (BPE) was replaced by BSA, which does not absorb at the same wavelength range as fluorescent BPE. Conventional biotinylated BPE alone and biotinylated AF680 alone were also utilized as acceptor dyes. See Supporting Information for conjugation of tandem dye (biotinylated BPE-AF680) and carrier protein dye (biotinylated BSA-AF680) and biotinylation of BPE and AF680. Preparation of Submicrometer-Sized UCP particles. The original size of the commercially available PTIR550/F UCP particles was 2.3-6.0 µm according to the manufacturer. These particles were ground in a planetary ball mill (Planetary Mono Mill pulverisette 6, Fritsch GmbH, Idar-Oberstein, Germany) for 90 min with 600 rpm, reversing the direction of rotation after every 5 min to gain colloidal particles with a diameter under 400 nm. A zirconia bowl (Fritsch GmbH) with a volume of 12 mL and smooth zirconia balls (Retch GmbH, Haan, Germany) with a diameter of 2 mm were used in milling due to their resistance to erosion. The masses of UCP particles and grinding balls in milling were 765 mg and 2 g, respectively. Milling was done in aqueous suspension containing 2 mL of 10 mM borate buffer, pH 9.0, with 0.1% (w/v) Additol XW330. Grinding balls were first separated from milled UCP particles by filtering the slurry through a nylon mesh with a pore size of 300 µm (Spectrum Laboratories Inc., Rancho Dominquez, CA). The volume of remaining particle suspension was adjusted to 10 mL with dimethyl sulfoxide (DMSO) containing 0.1% (w/v) Additol XW330 and bath sonicated (Finnsonic m03, Finnsonic Oy, Lahti, Finland) for 4.5 min to disperse the aggregates. Particles were next allowed to settle in a 15-mL test tube for 1 h. After the sedimentation period, the upper, colloidal portion was recovered and the particles in this portion were additionally washed twice with 1.5 mL of DMSO using centrifugation at 10000g. Finally, colloidal UCP particles were suspended in 1.5 mL of DMSO and stored in slow rotation at room temperature. Characterizations of the milled UCP particles were done as previously described.8 The concentration of colloidal material was determined by weighting the dried solids from a known volume of particle suspension. The size distribution of milled particles diluted in 10 mM borate buffer, pH 8.5, containing 1 g/L Tween 20 was estimated by Coulter N4plus submicrometer particle size analyzer (Beckman Coulter, Fullerton, CA). Streptavidin Coating of Submicrometer-Sized UCP Particles. The coating protocol for UCP particles was previously illustrated,34 but some modifications were introduced. The overnight treatment of ball-milled UCP particles with a poly(acrylic acid) was conducted at 35 °C with a smaller concentration of Additol XW330 (0.1% w/v) in an aqueous solution. The duration of the carboxylic group activation step with EDC and sulfo-NHS was shortened to last only 45 min, and the amount of added streptavidin was increased to 0.75 mg/mL instead of the previous 0.5 mg/mL. The completed streptavidin-coated UCP particles were stored at room temperature in slow rotation (4 rpm) in a 5 mM borate buffer, pH 8.5, containing 2 g/L Tween 85, 5 g/L BSA, and 0.5 g/L NaN3. Characterizations of coated UCP particles were performed as previously described.35 The concentration of particles was quanti(35) Kuningas, K.; Rantanen, T.; Karhunen, U.; Lo ¨vgren, T.; Soukka, T. Anal. Chem. 2005, 77, 2826-2834.
6314
Analytical Chemistry, Vol. 79, No. 16, August 15, 2007
fied, the total amount and the fraction of free streptavidin were measured, and the size distribution of the coated UCP particles was determined. Spectral Characterizations. The fluorescence emission spectrum of each acceptor containing AF680 was measured with a Varian Cary Eclipse fluorescence spectrophotometer (Varian Scientific Instruments, Mulgrave, Australia). The biotinylated acceptors were diluted in 0.05 M Tris-HCl, pH 7.75, 0.9% NaCl, 0.05% NaN3 (TSA) so that the concentration of AF680 (37.5 nM) was equal in every sample according to the absorbance reading at 679 nm. The spectra were measured with 0.75-nm resolution utilizing the fluorescence mode of the fluorescence spectrophotometer. Excitation wavelength was 680 nm, excitation slit 10 nm, and emission slit 5 nm. The fluorescence properties of the tandem dye were investigated in more detail. The emission spectrum was remeasured utilizing the same slits as above, but using another excitation wavelength, 560 nm. An excitation spectrum was measured reading the emission at 702 nm with 10-nm emission slit and 5-nm excitation slit. The anti-Stokes photoluminescence emission spectrum of the ball-milled, streptavidin-coated UCP particles was measured as reported before.8 Particles were diluted to concentration of 0.015 g/L in assay buffer. Optimization of Acceptor Concentration in Homogeneous Assay. To determine the optimal amount of biotinylated acceptor, different concentrations were tested in a previously described FRET-based homogeneous model assay.34 All four biotinylated acceptors were used (tandem dye, AF680, carrier protein dye, BPE). The concentration of an acceptor was calculated according to the amount of acceptor conjugates and not according to the concentration of AF680. Measurements were done from black, half-area microtitration wells (Costar Corning, NY) with modified fluorometer, Plate Chameleon (Hidex Oy, Turku, Finland). The fluorometer was equipped with a 200-mW infrared laser module (Roithener Lasertechnik, Vienna, Austria), RG-850 long-pass excitation filter (Andover Corp., Salem, NH), and Hamamatsu R4632 photomultiplier tube (Hamamatsu Photonics, Shizuoka, Japan).8 Four replicas of each acceptor concentration were used, and the total reaction volume in the well was 80 µL containing 0.015 mg/mL streptavidincoated UCP particles, free D-biotin (0 or 20 µM), and biotinylated acceptor molecule (0-12 nM). Free D-biotin and UCP particles were first added to a well in 32 µL and in 24 µL of assay buffer, respectively. After 15-min incubation at room temperature with 900 rpm shaking, the biotinylated acceptor molecules were added to the reaction wells in 24 µL of assay buffer. Incubation proceeded, protected from light, for 15 min to allow the binding of the acceptors to the unoccupied streptavidins. The anti-Stokes photoluminescence of UCP particles was measured using band-pass emission filter of 665/10 nm (center wavelength 665 nm, half-width 10 nm, peak transmittance g80%; Spectra-Physics, Mountain View, CA) combined with an absorptive neutral density filter (optical density 1.0, i.e., average T ) 10%; Thorlabs, Newton, NI) collecting the emission light for 2000 ms under continuous laser excitation at 980 nm. The FRET emission of the acceptors was measured, respectively, but using band-pass emission filter of 740/40 nm (peak T ) 85%; Chroma Technology
Figure 1. Principle of upconversion FRET-based homogeneous bioaffinity assay for biotin. The UCP particles coated with streptavidin (donors) produce emission at 540 nm and at 653 nm under infrared excitation (980 nm). The sensitized emission of biotinylated tandem dyes (acceptors) is detected at 575 nm and at 704 nm, but only the emission at red region is measured. UCP, upconverting phosphor; UC-FRET, upconversion FRET; AF680, Alexa Fluor 680; BPE, B-phycoerythrin.
Corp., Rockingham, VT) combined with a wide band-pass filter glass 800/200 nm (peak T g 80%; Chroma Technology Corp.), and these signals were corrected for variation in laser intensity by using the donor emission signals at 665 nm as described previously.34 A standard emission band-pass filter with a central wavelength at 704 nm could not be chosen for measurement of sensitized acceptor emission because of inadequate stop-band blocking at 653 nm (donor emission). The best available choise was the filter we used (740/40 nm), which still had 50% transmittance at 720 nm. Competitive Ligand-Binding Assay for Biotin. After comparing all four acceptors in an acceptor optimization assay, only tandem dye and biotinylated AF680 were compared in a homogeneous competitive biotin assay. The principle of this assay is similar to the protocol described above. This time the reaction mixture contained 0.015 mg/mL streptavidin-coated UCP particles, various amounts of free D-biotin (0-100 and 10 000 nM), and an optimal amount of biotinylated acceptor (1.6 or 4.4 nM) in 80 µL of assay buffer. The highest concentration of free D-biotin was used to verify the background level of the assay. RESULTS AND DISCUSSION Principle of the Competitive Ligand-Binding Assay for Biotin. The principle of a homogeneous ligand-binding assay based on upconversion FRET between the streptavidin-coated UCP particles and biotinylated acceptor molecules is described in Figure 1. The emission of the acceptor dye is modulated by a binding event; energy transfer from UCP particle through FRET process is possible only to the acceptors in proximity. Unbound acceptors in the solution are too far to accept energy nonradiatively. This competitive assay is considered to be a back-titration
assay36 as the tracer (labeled biotin) is added only after the binder (streptavidin-coated UCP particle) and the analyte (D-biotin) are first incubated together to allow better sensitivity and to facilitate the manual pipetting. There are various alternatives within fluorescent acceptors suitable for this application. Proper overlap in the fluorescence spectrum with the donor and the possibility for biotinylation are the main requirements for the acceptor. D-Biotin binds to the streptavidin on the surface of UCP particles, and biotinylated acceptor molecules occupy the remaining binding sites. The UCP particles are excited with infrared light (980 nm), and the sensitized emission of the bound acceptor molecules is measured above 700 nm. In Figure 1, tandem dye also emits at green wavelengths (575 nm), but the red emission is only measured. Characterization of Biotinylated Acceptor Dyes. In the tandem dye, B-phycobiliprotein functions as an absorber and AF680 as an emitter offering the fluorescent elements whereas biotin provides the specific binding to streptavidin. The optimal tandem dye conjugate was found to consist of five AF680 dye molecules on average per one fluorescent BPE accompanied by at least one biotin. Other ratios in the conjugation were also tested, but the amount of Alexa Fluor was critical (Figure S-1 in Supporting Information). Too many dye molecules introduced in one BPE molecule lead to a higher overall signal, but the increase was not linear due to the self-quenching. Low labeling degrees lead to an inefficient internal FRET between the absorber and the emitter. These findings are consistent with the results reported by Blomberg et al.37 They found that in time-resolved FRET application optimal labeling degree of IgG was 2-5 Alexa546 molecules/1 immunoglobulin. BPE is a 1.5 times larger molecule than IgG, suggesting the optimal labeling degree to be 3-7. In Supporting Information, it is also shown that the emitter dye AF680 can be replaced with another emitter dye (Alexa Fluor 700, AF700; Figure S-2) if the conjugation ratio is optimized for the new emitter. The number of biotin molecules was not so crucial considering our application. The desired biotinylation degree was around one to two. The more biotin is in one molecule, the faster is the kinetics in assay, but the amount of biotin does not affect the result at the equilibrium. In carrier protein dye, the fluorescent BPE was substituted with BSA, which is practically nonfluorescent at visible wavelengths. This protein has a 3.6 times smaller molecular weight than BPE, and therefore, the labeling degree should be kept under that of the tandem dye avoiding the self-quenching of AF680 fluorescence. Two dye molecules per BSA was found to be the optimal amount. If this labeling degree is again compared to the published findings by Blomberg et al.,37 the congruence is evident considering the smaller size of BSA. Most of the BSA contained at least one biotin. There was no optimization step in the biotinylation of BPE or AF680 alone. BPE is quite a large protein (molecular weight 240 000), and at least two biotins were coupled to most of the BPE molecules. Instead, small AF680 has only one reactive NHS group leading automatically to a biotinylation degree of one. (36) Rodbard, D.; Ruder, H. J.; Vaitukai, J.; Jacobs, H. S. J. Clin. Endocrinol. Metab. 1971, 33, 343-355. (37) Blomberg, K.; Hurskainen, P.; Hemmila¨, I. Clin. Chem. 1999, 45, 855861.
Analytical Chemistry, Vol. 79, No. 16, August 15, 2007
6315
Unbiotinylated acceptors can increase the background signal, but they do not have binding capability and therefore no functionality in assays. The emission spectra of the biotinylated acceptors (excluding BPE) were measured from samples containing the same amount of AF680 adjusted by an absorbance reading. The shape of the spectrum at the region of the emission maximum of AF680 (∼700 nm) was the same when excited at 680 nm (data not shown). However, the intensity of emission decreased if more than one dye molecule was introduced into one conjugate. Compared to the AF680 alone, the intensity was only 65 and 63% for the tandem dye and the carrier protein dye, respectively. This phenomenon refers more likely to a relevant environmental change of the AF680 in protein conjugate than merely to the self-quenching of AF680 emission when these dye molecules are brought too close to each other, because the emission from similar tandem dyes with lower AF680 content did not show any higher AF680 intensity per one dye. Roederer et al.28 stated that the increase in the amount of organic fluorophore in phycobiliprotein-based tandem conjugate increases the potential for self-quenching and insolubility. Considering the molecular weight of the protein in the dye conjugates (BPE or BSA), the carrier protein dye has the most dense dye content. If the above-described decline of AF680 emission intensity in protein conjugates were purely caused by self-quenching, the difference between the tandem dye and the carrier protein dye would be expected to be more apparent. There was no evidence of insolubility with any of the conjugates. Despite the lower signal from individual AF680 dyes, the advantage of tandem dye is clearly seen if the excitation and emission spectra of acceptor (tandem dye) and the emission spectrum of the donor (UCP particle) are illustrated in the same figure (Figure 2). The anti-Stokes emission of UCP is generated when it absorbs two infrared (980 nm) photons sequentially and combines the energy before emission. The tandem dye we constructed matches perfectly with the UCP donor. The acceptor molecule is capable of exploiting both of the main emission peaks of the donor (540 and 653 nm) due to the total spectral overlap. In our application, homogeneous biotin assay, the red emission of the tandem dye is measured at a wavelength region where the emission of the donor is minimal, resulting in a low background signal. BPE alone does not emit at this region (Figure S-3 in Supporting Information) leaving only the emission of AF680 detectable. An interesting observation in Figure 2 is that maximal excitation is achieved at the wavelength characteristic of BPE (absorber; 565 nm) and the emission maximum is still at the wavelength specific to AF680 (emitter; 704 nm) illustrating the internal FRET between the absorber and the emitter in the tandem dye. However, not all of the BPE emission (575 nm) is attenuated due to the energy transferred. This results probably from the heterogeneity of the tandem dye conjugates having random distribution of AF680 molecules; the distance of emitters may vary and a small fraction of the BPE molecules may even lack AF680 molecules. The emitter and absorber must be close enough (conjugated) or no energy transfer occurs (Figure S-4 in Supporting Information). The spectra of biotinylated BPE and AF680 alone (Figure S-3 in Supporting Information) did not differ significantly from those parts of the tandem dye spectra having excitation maximums at 6316 Analytical Chemistry, Vol. 79, No. 16, August 15, 2007
Figure 2. Excitation (dashed line) and emission spectra (thin solid line) of biotinylated tandem dye and the emission spectrum of streptavidin-coated UCP particle (thick solid line). The excitation maximum of tandem dye appears at the region of BPE excitation (565 nm), but the maximum emission is still observed at the region of AF680 emission (704 nm) indicating the internal FRET between these participants in tandem dye. The overlap of the excitation spectrum of tandem dye and the emission spectrum of UCP particle is almost complete, enabling efficient FRET process from the UCP to the tandem dye. The center wavelength of the emission filter utilized in upconversion FRET-based homogeneous bioaffinity assay is 740 nm, but the pass-band is wide enough (40 nm half-width at half-height) to collect the emission from the main peak of tandem dye. Intensities are normalized to the same scale. a.u., arbitrary unit; UCP, upconverting phosphor.
566 and 678 nm and emission maximums at 575 and 699 nm, respectively. The Stokes shift for BPE was 9 nm and for AF680 21 nm, which are considerably smaller compared with the tandem dye having a pseudo-Stokes shift of 139 nm enabling good spectral separation. The overlap of acceptor excitation and donor emission was not so total with all of the acceptors. Carrier protein dye and AF680 alone can only exploit the smaller red emission peak of the donor (653 nm). BPE alone utilizes merely the green main emission peak of the donor (540 nm) in the FRET process, but the signal is almost undetectable at wavelengths above 700 nm due to the absence of BPE emission at this region. Characterization of UCP Particles. After processing the originally micrometer-sized UCP particles in a planetary ball mill, the average particle size in the colloidal proportion was 340 nm. The size profile was similar with that obtained with the miniature bead mill procedure our group used earlier.8 The average size of colloidal UCP used in homogeneous assays has previously varied from 270 to 390 nm.9,34,38 However, the old protocol took 2 days and nights, whereas the grinding with the planetary ball mill is already finished in 2 h. Nanosized particles with a diameter below 100 nm would have an advantage in FRET-based homogeneous assays because the surface layer capable of donating the energy is relatively larger compared to the core volume. However, it is difficult to obtain such small particles in high yield by grinding micrometer-sized UCP material,8 and therefore, the straight synthesize of nanoparticles would be preferable.39-43 (38) Kuningas, K.; Ukonaho, T.; Pa¨kkila¨, H.; Rantanen, T.; Rosenberg, J.; Lo ¨vgren, T.; Soukka, T. Anal. Chem. 2006, 78, 4690-4696.
Figure 3. Titration of biotinylated acceptor molecules: tandem dye (BPE-AF680; filled circle), AF680 (open triangle), carrier protein dye (BSA-AF680; filled diamond), and BPE (open square). Constant amounts of streptavidin-coated UCP particles (donors) were mixed with various amounts of biotinylated acceptor molecules in the absence and in the excess of free D-biotin (analyte). The represented upconversion FRET signals were calculated by subtracting the background signal of the assay (excess of D-biotin) from the maximum signal (absence of D-biotin). cts, counts.
The luminescence of the ground UCP was still bright, and the particles were further modified by coating them with streptavidin. Total concentration for streptavidin was 18.1 µg of streptavidin/ mg of particles. The amount of free streptavidin, unconjugated or detached, was only 1.4%. This indicates the good quality of streptavidin-coated UCP particles. Free streptavidin in an assay is a serious problem, as it binds analyte molecules without any effect on the measured signal, leading to an overall diminished signal level. Optimization of Acceptor Concentration in Homogeneous Assay. The optimal concentration of biotinylated acceptor molecule in the biotin assay was explored by changing the amount of the acceptor while maintaining the amount of donor constant. The concentration of donor particles was selected based on adequate red emission intensity. Background signal of the assay was determined by blocking the binding sites on the surface of the donor particle with an excess of free D-biotin prior to the addition of biotinylated acceptor dyes. Maximum signal was obtained without blocking D-biotin. The data are presented in the form of specific upconversion FRET signal calculated by subtracting the background signal of the assay from the maximum signal (Figure 3). The hooklike curves indicate that when the amount of biotinylated acceptor is added in the reaction well, the signal rises until it reaches a pinnacle. After this point, the solid surface is overpopulated with acceptor molecules leading to a self-quenching (39) Hirai, T.; Orikoshi, T.; Komasawa, I. Chem. Mater. 2002, 14, 3576-3583. (40) Heer, S.; Lehmann, O.; Haase, M.; Gudel, H. U. Angew. Chem., Int. Ed. 2003, 42, 3179-3182. (41) Heer, S.; Kompe, K.; Gudel, H. U.; Haase, M. Adv. Mater. 2004, 16, 21022105. (42) Vetrone, F.; Boyer, J. C.; Capobianco, J. A.; Speghini, A.; Bettinelli, M. J. Phys. Chem. B 2003, 107, 1107-1112. (43) Yi, G. S.; Sun, B. Q.; Yang, F. Z.; Chen, D. P.; Zhou, Y. X.; Cheng, J. Chem. Mater. 2002, 14, 2910-2914.
and to a decrease in a signal. The optimal concentration of acceptor is near to the pinnacle. The signal of BPE was not significantly distinguished from the background, and therefore, it was meaningless to analyze the data of this acceptor. For the tandem dye and the carrier protein dye, the optimal acceptor concentration was 1.6 nM and the maximum signals differed from the background signals by a factors of 10 and 14, respectively. The optimum point for AF680 was somewhere between 4.4 and 6.5 nM. Due to the steep drop after the latter concentration, we considered the concentration of 4.4 nM to be preferable. The ratio between the maximum signal and background signal was up to 28. AF680 alone reached the saturation level of the signal with higher acceptor concentration than the other two acceptors due to the smaller size of the conjugate. The large size of the proteincontaining acceptors obscures some of the binding sites on the solid surface leading to earlier saturation. The highest specific signal from one binding event was obtained with the tandem dye. The carrier protein dye and AF680 alone had a four times lower signal level compared to the tandem dye when the acceptor concentration was 1.6 nM. If the maximum signals achieved with optimal concentrations for each acceptor are compared, the difference between the signal level of the tandem dye and the carrier protein dye or AF680 alone is 4- or 2-fold, respectively. However, conjugation of AF680 dye molecules to a protein (e.g., in tandem dye and in carrier protein dye) causes a few drawbacks. First, the distance between the donor particle and the acceptor dye is increased due to the protein component affecting negatively the FRET process. Second, the different environment close to the protein alters the fluorescence intensity of the AF680 as noticed in the context of the acceptor characterizations. Self-quenching may further decrease the signal even more if the dye molecules are conjugated too densely. These are probably the reasons why the carrier protein dye containing approximately two dye molecules did not bring significant improvements compared to AF680 alone. In addition, as consisting of even larger carrier protein (BPE) for AF680 dyes, the farthermost dye molecules in the tandem dye may be completely incapable of participating in the FRET process due to the substantial distance. The predominance of the tandem dye must be explained by other factors. First, the better overlap of the donor and acceptor fluorescence spectra offers efficient excitation of the tandem dye, which is a prerequisite for bright emission. Second, the internal FRET in the tandem dye also directs the emission of the absorber to the red wavelength region. In addition, the molar extinction coefficient of BPE (absorber) is exceptionally high, enabling the efficient utilization of the green main emission peak of the donor. As a result, the advantages of the tandem dye are more significant than the disadvantages related to the large carrier protein. Utilization of small molecular weight tandem dyes instead of conjugates containing large fluorescent protein may provide further improvement to this technology. When only one emitter is coupled with a small molecular weight absorber, the selfquenching inside the conjugate could be avoided. In addition, the competition between the biotinylated acceptor and the small analyte would be more proper if the size difference was less remarkable. Analytical Chemistry, Vol. 79, No. 16, August 15, 2007
6317
Figure 4. Standard curves of upconversion FRET-based biotin assay after 15-min incubation utilizing biotinylated tandem dye (1.6 nM, open circle) and biotinylated AF680 (4.4 nM, filled triangle) as an acceptor molecule. IC50 values were 9.6 nM and 12.7 nM, respectively.
Competitive Ligand-Binding Assay for Biotin. The homogeneous model assay for biotin was the goal where all the previous optimizations were aiming. A sigmoidal-shape standard curve should result when the amount of biotin is varied in the assay. In the case of AF680 alone, there was a small elevation in the signal before the curve began to fall (Figure 4). This results from the four binding sites of the strepavidin molecule enabling several biotinylated AF680 conjugates to bind closely leading to the selfquenching of the emission signal with small amounts of analyte. When enough analyte molecules are in the reaction to displace some of the densely bound acceptors, the signal raises slightly before the analyte concentration reaches the point where signal begins to fall due to increased competition. Instead, tandem dye is a substantially larger conjugate and does not fit so densely on the solid surface. The signal starts to drop earlier than expected because these large acceptors do not reach every free binding site as the adjacent acceptor molecule may obscure some of the sites. The IC50 values (analyte concentrations that inhibited 50% of the maximum signal) did not differ significantly despite the acceptor molecule. They were 9.6 and 12.7 nM for assays utilizing the tandem dye and AF680, respectively. These values are consistent with results of Kuningas et al.34 from similar assays. The similarity of the IC50 values was not a surprise as this feature is affected by the binding capacity of the donor particles rather than the nature of the acceptor dye. By reducing the amount of analyte binding sites (streptavidin) in the reaction, it is possible to adjust the IC50 value of the assay to lower analyte concentrations. This action requires that signal level still remains at a measurable level.
6318 Analytical Chemistry, Vol. 79, No. 16, August 15, 2007
A difference in the signal levels of the two acceptors was evident. With the tandem dye it was possible to achieve 1.7 times greater signal than with the conventional AF680 dye. The background signal in the assay utilizing tandem dye rose along with the overall signal level, and there was no improvement in signal-to-background ratio. The ratio was still over 8-fold and, therefore, completely adequate for bioaffinity assays. Elevated background can be caused by extensive dimensions of the BPE molecule. When the specific signal from the bound tandem dye is generated via the FRET process, only those emitters of the tandem dye close to the donor particle participate, but in the case of nonspecific signal generation via radiative excitation of unbound tandem dye, all emitters are equal and produce background signal. Nonspecific binding of the BPE protein to the donor particle may be another reason for background signal. The difference between the maximum signal of tandem dye and AF680, however, was significantly larger than the difference between assay backgrounds, indicating that the higher maximum signal of tandem dye is due to improved spectral overlap and enhanced energy transfer. CONCLUSIONS We have demonstrated enhancement of the sensitized acceptor emission from one binding event by utilizing a tandem dye rather than a conventional acceptor dye molecule along with an upconverting phosphor as an energy donor in a homogeneous bioaffinity assay. The use of tandem dyes might provide advantages in many FRET-based homogeneous assays as both the excitation and emission wavelengths of the dye can be tuned to a desired wavelength region. ACKNOWLEDGMENT This study was supported by the Finnish Funding Agency for Technology and Innovation (Tekes) and the Academy of Finland (Grant 209417). The authors are thankful for technological support from Hidex Oy in anti-Stokes photoluminescence measurement. SUPPORTING INFORMATION AVAILABLE Conjugation and biotinylation protocols for tandem dye, carrier protein dye, BPE, and AF680. The fluorescence spectra of following biotinylated dyes: BPE alone, AF680 alone, BPE and AF680 together but not conjugated, successful BPE-AF700 tandem dye, two unsuccessful BPE-AF680 tandem dyes. This material is available free of charge via the Internet at http://pubs.acs.org.
Received for review February 23, 2007. Accepted June 18, 2007. AC070376W