A Parallel Multiharmonic Frequency-Domain Fluorometer for

Under our experimental conditions, the frequency-domain profile for a train of .... Two systems were investigated for this express purpose: (1) a mixt...
0 downloads 0 Views 169KB Size
Anal. Chem. 1998, 70, 3384-3396

A Parallel Multiharmonic Frequency-Domain Fluorometer for Measuring Excited-State Decay Kinetics Following One-, Two-, or Three-Photon Excitation A. Neal Watkins, Christine M. Ingersoll, Gary A. Baker, and Frank V. Bright*

Department of Chemistry, Natural Sciences Complex, State University of New York at Buffalo, Buffalo, New York 14260-3000

We report on the performance of a new, multiharmonic frequency-domain instrument that uses the high harmonic content of a passively mode-locked, pulse-picked femtosecond Ti-sapphire laser as the excitation source for the determination of one-, two-, or three-photon excited timeresolved fluorescence anisotropy and intensity decay kinetics. In operation, the new instrument can provide a complete frequency-domain data set at 100 modulation frequencies in less than 1 min. The new instrument exhibits 5-10-ps measurement precision and it can rapidly and accurately recover complex excited-state fluorescence anisotropy and intensity decay kinetics under one-, two-, or three-photon excitation for dilute or optically dense samples that exhibit single or multiexponential decay kinetics. This latter aspect of the instrument is demonstrated by successfully determining the excitedstate intensity decay kinetics for a dilute aqueous solution of rhodamine 6G dissolved in a high concentration of bromocresol green. This approach is extended by determining the excited-state fluorescence intensity decay kinetics of dilute fluorescein directly in undiluted, whole blood as a function of pH under two-photon excitation conditions. The high-speed capabilities of the new instrument are exploited by performing two-photon excited fluorescence anisotropy decay experiments on the fly for site-selectively labeled bovine serum albumin as it undergoes enzymatic digestion by trypsin. Tunable, high repetition rate mode-locked Ti-sapphire lasers that produce ultrashort (fs) pulses have made possible numerous advances in the chemical sciences.1-3 In addition to the obvious improvement in time resolution afforded by these lasers, their high peak power (∼100 kW) makes them attractive sources for multiphoton excitation (MPE) schemes wherein the analyte of interest absorbs simultaneously two or more longer wavelength photons to produce an excited electronic state.4 Although a bit more technically challenging, MPE experiments offer substantial (1) Spence, D. E.; Kean, P. N.; Sibbert, W. Opt. Lett. 1991, 16, 42-4. (2) Cerullo, G.; DeSilvestri, S.; Magni., V. Opt. Lett. 1994, 19, 1040-2. (3) Ultrafast Processes in Spectroscopy; Svelto, O., DeSilvestri, S., Denardo, G., Eds.; Plenum Press: New York, 1996. (4) Callis P. R. Annu. Rev. Phys. Chem. 1997, 48, 271-99.

3384 Analytical Chemistry, Vol. 70, No. 16, August 15, 1998

advantages compared to the more conventional one-photon excitation schemes. Multiphoton excitation was first proposed in 1931 by Go¨ppertMeyer.5 Thirty years latter, Kaiser and Garrett first demonstrated two-photon excited fluorescence from Eu2+ doped within CaF2.6 Three-photon excited fluorescence was subsequently demonstrated by Singh and Bradley in 1964 on naphthalene crystals.7 Following from these seminal efforts, multiphoton excited fluorescence has been used to perform quantitative measurements in strongly absorbing media,8,9 in tissue,10 and in whole blood,11 it has been used as a detection method in liquid chromatography12,13 and capillary separations,14 it has allowed remarkable advances in high-resolution fluorescence microscopy,15-26 and it is able to quantifying fluorescent species in dilute solutions27-35 (5) Go ¨ppert-Mayer, M. Ann. Phys. 1931, 9, 273-95. (6) Kaiser, W.; Garrett, C. G. B. Phys. Rev. Lett. 1961, 7, 229-31. (7) Singh, S.; Bradley, L. T. Phys. Rev. Lett. 1964, 12, 612-4. (8) Wirth, M. J.; Lytle, F. E. Anal. Chem. 1977, 49, 2054-7. (9) Steffen, R. L.; Lytle, F. E. Anal. Chim. Acta 1988, 215, 203-10. (10) Guo, Y.; Wang, Q. Z.; Zhadin, N.; Liu, F.; Demos, S.; Calistru, D.; Tirksliunas, A.; Katz, A.; Budansky, Y.; Ho, P. P.; Alfano, R. R. Appl. Opt. 1997, 36, 968-70. (11) Burke, T. G.; Malak, H.; Gryczynski, I.; Mi, Z.; Lakowicz, J. R. Anal. Biochem. 1996, 242, 266-70. (12) Sepaniak, M. J.; Yeung, E. S. Anal. Chem. 1977, 49, 1554-6. (13) Sepaniak, M. J.; Yeung, E. S. J. Chromatogr. 1981, 211, 95-102. (14) Gostkowski, M. L.; McDoniel, J. B.; Wei, J.; Curey, T. E.; Shear, J. B. J. Am. Chem. Soc. 1998, 120, 18-22. (15) Denk, W.; Strickler, J. H.; Webb, W. W. Science 1990, 248, 73-6. (16) Ha¨nninen, P. E.; Hell, S. W. Bioimaging 1994, 2, 117-22. (17) Soeller, C.; Cannell, M. B. Plugers Arch.: Eur. J. Physiol. 1996, 432, 55561. (18) Gu, M. Opt. Lett. 1996, 21, 988-90. (19) Yu, W. M.; So, P. T. C.; French, T.; Gratton, E. Biophys. J. 1996, 70, 62636. (20) Bhawalker, J. D.; Swiatkiewicz, J.; Pan, S. J.; Samarabandu, J. K.; Liou, W. S.; He, G. S.; Berezeny, R.; Cheng, P. C.; Prasad, P. N. Scanning 1996, 18, 562-6. (21) Robertson, G.; Armstrong, D.; Dymott, M. J. P.; Ferguson, A. I.; Hogg, G. L. Appl. Opt. 1997, 36, 2481-3. (22) Schrader, M.; Bahlmann, K.; Hell, S. W. Optik 1997, 104, 116-24. (23) Robinson, M. K. Biophotonic Int. 1997, (Sept/Oct), 38-45. (24) Masters, B. R.; So, P. T. C.; Gratton, E. Biophys. J. 1997, 72, 2405-12. (25) Parasassi, T.; Gratton, E.; Yu, W. M.; Wilson, P.; Levi, M. Biophys. J. 1997, 72, 2413-29. (26) Xu, C.; Webb, W. W. In Topics in Fluorescence Spectroscopy. Volume 5: Nonlinear and Two-Photon Induced Fluorescence; Lakowicz, J. R., Ed.; Plenum Press: New York, 1997; Chapter 11. (27) Pfeffer, W. D.; Yeung, E. S. Anal. Chem. 1986, 58, 2103-5. S0003-2700(98)00348-5 CCC: $15.00

© 1998 American Chemical Society Published on Web 07/03/1998

even at the single-molecule level.36 Multiphoton excited fluorescence has also been used to determine multiphoton absorption cross sections37-39 and multiphoton excited excitation spectra40 and perform steady-state fluorescence measurements on complex biological samples.41-48 More recently, multiphoton excited fluorescence has been expanded with time-resolved spectroscopy for imaging49 and excited-state fluorescence anisotropy and intensity decay measurements.50-56 The latter area is of particular interest here. Several groups have demonstrated time-resolved multiphoton excited fluorescence (MPEF) in the time and frequency domains as a tool for determining excited-state anisotropy and intensity decay kinetics.50-56 The primary advantages associated with the frequency-domain methodology are its high accuracy and the speed with which one can measure the phase angle and demodulation factor. Unfortunately, if the actual fluorescence anisotropy or intensity decay kinetics are (as are most) more complex than a single-exponential decay, one must acquire frequency-domain data at a number of modulation frequencies over a large frequency range to recover the true fluorescence anisotropy or intensity decay kinetics.57-62 In all previous frequency-domain MPEF instruments, one acquired the requisite frequency-domain data (28) Freeman, R. G.; Gilliland, D. L.; Lytle, F. E. Anal. Chem. 1990, 62, 22169. (29) Wirth, M. J.; Fatunmbi, H. O. Anal. Chem. 1990, 62, 973-6. (30) Van de Nesse, R. J.; Mank, A. J. G.; Hoornweg, G. P.; Gooijer, C.; Brinkman, U. A. T.; Velthorst, N. H. Anal. Chem. 1991, 63, 2685-8. (31) Fisher, W. G.; Lytle, F. E. Anal. Chem. 1993, 65, 631-5. (32) Lytle, F. E.; Dinkel, D. M.; Fisher, W. G. Appl. Spectrosc. 1993, 47, 20027. (33) Mertz, J.; Xu, C.; Webb, W. W. Opt. Lett. 1995, 2532-4. (34) Fisher, W. G.; Wachter, E. A.; Armas, M.; Seaton, C. Appl. Spectrosc. 1997, 51, 218-26. (35) Overway, K. S.; Lytle, F. E. Appl. Spectrosc. 1998, 52, 298-302. (36) Walser, D.; Plakhotnik, T.; Renn, A.; Wild, U. P. Chem. Phys. Lett. 1997, 270, 16-22. (37) (a) Xu, C.; Zipfel, W.; Shear, J. B.; Williams, R. M.; Webb, W. W. Proc. Natl. Acad. Sci. U.S.A. 1996, 93, 10763-8. (b) Xu, C., Ph.D. Thesis, Cornell University, 1996. (38) Xu, C.; Webb, W. W. J. Opt. Soc. Am. B 1996, 13, 481-91. (39) Xu, C.; Guild, J.; Webb, W. W.; Denk, W. Opt. Lett. 1995, 20, 2372-4. (40) Fisher, W. G.; Wachter, E. A.; Lytle, F. E.; Armas, M. E.; Seaton, C. Appl. Spectrosc. 1998, 52, 536-45. (41) Birge, R. R.; Bennett, J. A.; Pierce, B. M.; Thomas, T. M. J. Am. Chem. Soc. 1978, 100, 1533-9. (42) Birge, R. R. Acc. Chem. Res. 1986, 19, 138-46. (43) Rehms, A. A.; Callis, P. R. Chem. Phys. Lett. 1987, 140, 83-9. (44) Jiang, S. P. Prog. React. Kinet. 1989, 15, 77-92. (45) Kawski, A.; Gryczynski, Z.; Gryczynski, I.; Lakowicz, J. R.; Piszczek, G. Z. Naturforsch. 1996, 51A, 1037-41. (46) Shear, J. B.; Brown, E. B.; Webb, W. W. Anal. Chem. 1996, 68, 1778-83. (47) Chen, Z.; Kaplan, D. L.; Yang, K.; Kumar, J.; Marx, K. A.; Tripathy, S. K. Appl. Opt. 1997, 36, 1655-9. (48) Gryczynski, I.; Malak, H.; Lakowicz, J. R. Biospectroscopy 1997, 3, 97-101. (49) French, T.; So, P. T. C.; Weaver, D. J.; Coelhosampaio, Gratton, E.; Voss, E. W.; Carrero, J. J. Microsc. 1997, 185, 339-53. (50) Lakowicz, J. R.; Gryczynski, I.; Gryczynski, Z.; Danielsen, E.; Wirth, M. J. J. Phys. Chem. 1992, 96, 3000-6. (51) Lakowicz, J. R.; Grycznski, I. Biophys. Chem. 1992, 45, 1-6. (52) Lakowicz, J. R.; Gryczynski, I.; Danielsen, E.; Frisoli, J. Chem. Phys. Lett. 1992, 194, 282-7. (53) Gryczynski, I.; Malak, H.; Lakowicz, J. R. Chem. Phys. Lett. 1995, 245, 305. (54) Kierdaszuk, B.; Malak, H.; Gryczynski, I.; Callis, P.; Lakowicz, J. R. Biophys. Chem. 1996, 62, 1-13. (55) Lakowicz, J. R.; Gryczynski, I.; Malak, H.; Gryczynski, Z. J. Phys. Chem. 1996, 100, 19406-11. (56) Lakowicz, J. R.; Gryczynski, I. In Topics in Fluorescence Spectroscopy. Volume 5: Nonlinear and Two-Photon Induced Fluorescence; Lakowicz, J. R., Ed.; Plenum Press: New York, 1997; Chapter 5.

in a serial or stepwise fashion over the frequency range of interest. While this methodology works, it can easily take an hour or more to acquire a single 15-20 frequency data set on a reasonably fluorescent sample. Moreover, the total data acquisition time with a serial frequency-domain instrument, assuming the source depth of modulation is constant across the frequency profile, is linearly dependent on the actual number of modulation frequencies within a particular data set. For example, if one acquired a data set with a serial phase-modulation instrument at 20 frequencies and the total acquisition time was 1 h, a data set with 100 discrete frequencies would take at least 5 h to acquire serially. In this paper we report on a new frequency-domain instrument that uses the harmonic content of a passively mode-locked, pulsepicked femtosecond Ti-sapphire laser as the excitation source and digital parallel acquistion57,58 to acquire one-, two-, or threephoton excited frequency-domain fluorescence data simultaneously at 100 modulation frequencies (200 data points) in less than 1 min. We illustrate the potential of this new instrument by recovering the fluorescence excited-state intensity and anisotropy decay kinetics for a series of common fluorophores, fluorophore mixtures, and biologically relevant fluorophores under one-, two-, or three-photon excitation. We then use the instrument to perform direct excited-state fluorescence lifetime measurements in optically dense media such as a concentrated solution of bromocresol green and whole, untreated human blood. Finally, we use the highspeed capabilities of the new instrument to acquire two-photon excited fluorescence anisotropy decay data “on the fly” at 0.5min time intervals for site-selectively labeled bovine serum albumin as it undergoes enzymatic digestion by trypsin. THEORY SECTION Excited-State Intensity Decay Measurements in the Frequency Domain. For any time-resolved intensity decay, I(t), one can write n

I(t) )

∑ Re

-t/τi

i

(1)

i)1

where n is the number of discrete emissive components, τi is the excited-state fluorescence lifetime for component i and Ri is the preexponential factor associated with τi. In the frequency domain, the sample under study is excited with sinusoidally modulated light and the experimentally measured parameters are the frequency-dependent phase angle (θ) and demodulation factor (M). From the phase-modulation data, one can recover all the terms that describe I(t).59-62 Rotational Reorientation Dynamics in the Frequency Domain. For a simple isotropic rotor, the time-resolved fluorescence anisotropy decay, rx(t), is described by a single rotational (57) Feddersen, B. A.; Piston, D. W.; Gratton, E. Rev. Sci. Instrum. 1989, 60, 2929-36. (58) Mitchell, G. W. Picosecond Multiharmonic Fourier Fluorometer U.S. Patent 4,937,457, 1990. (59) Jameson, D. M.; Gratton, E.; Hall, R. D. Appl. Spectrosc. Rev. 1984, 20, 55-106. (60) Lakowicz, J. R.; Laczko, G.; Gryczynski, I.; Szmacinski, H.; Wiczk, W. J. Photochem. Photobiol., B: Biol. 1988, 2, 295-311. (61) Bright, F. V.; Betts, T. A.; Litwiler, K. S. CRC Crit. Rev. Anal. Chem. 1990, 21, 389-405. (62) Bright, F. V. Appl. Spectrosc. 1995, 49, 14A-9A.

Analytical Chemistry, Vol. 70, No. 16, August 15, 1998

3385

reorientation time, φ:59-62

rx(t) ) ro,x exp(-t/φ)

(2)

In this expression, ro,x is the limiting anisotropy at a particular set of excitation conditions. Under one-, two-, or three-photon excitation, the values for ro,x (x is the number of photons used to produce the excited state) range from -0.20 to +0.40, -0.29 to +0.57, or -0.33 to +0.67, respectively.63-67 For more complicated systems, rx(t) can be written as n

rx(t) ) ro,x

∑ β exp(-t/φ ) i

i

(3)

i)1

where n is the number of discrete rotational events that describe rx(t), φi is rotational reorientation time i, and βi is the fractional contribution of the total fluorescence anisotropy decay that arises from φi. The recovered φi terms are nonlinear functions of principle rotational diffusion coefficients that describe the rotational motion around the three principal axes of the rotating unit. In the frequency domain, rx(t) is determined from frequencydependent measurements of the differential polarized phase angle, ∆ () θ⊥ - θ|) and the polarized modulation ratio, Λ () m|/m⊥).59-62 Figure 1 presents a simulated series of ∆ and Λ traces for a hypothetical fluorophore that has an excited-state fluorescence lifetime of 4.00 ns and a rotational reorientation time of 1.00 ns when it is excited under one-, two-, or three-photon excitation conditions. The limiting anisotropies for the individual simulations were set at 0.40, 0.57, or 0.67, representing the limiting positive values when a fluorophore is excited under one-, two- or threephoton excitation conditions, respectively. The most obvious feature seen in these simulations is that ∆ and Λ increase significantly for the two- and three-photon excitation cases relative to the one-photon excitation case. This arises because of the increased photoselection associated with MPE.63-67 Statistics. All results reported in this paper are the average of at least triplicate runs on discrete samples. The uncertainties in all the reported values are given as the mean ( one standard deviation. The actual imprecision in each measured datum was used when fitting all the frequency-domain data. EXPERIMENTAL SECTION Reagents and Materials. The following chemicals and materials were used in these experiments: Bio-Gel P-2 Gel (BioRad Laboratories); acetone, dimethylformamide (DMF), N,N′-bis(2,5-di-tert-butylphenyl)-3,4,9,10-perylenedicarboximide (BTBP), propylene glycol, perylene, rubrene, 2,5-diphenyl-1,3,4-oxadiazole (PPD), and 2,5-diphenyloxazole (PPO) (Aldrich Chemical Co.); (63) Wirth, M. J.; Koskelo, A. C.; Mohler, C. E.; Lentz, B. L. Anal. Chem. 1981, 53, 2045-8. (64) Mohler, C. E.; Wirth, M. J. J. Chem. Phys. 1988, 88, 7369-75. (65) Wirth, M. J.; Koskelo, A.; Sanders, M. J. Appl. Spectrosc. 1980, 35, 14-8. (66) Callis, P. R. In Topics in Fluorescence Spectroscopy. Volume 5: Nonlinear and Two-Photon Induced Fluorescence; Lakowicz, J. R., Ed.; Plenum Press: New York, 1997; Chapter 1. (67) Johnson, C. K.; Wan, C. In Topics in Fluorescence Spectroscopy. Volume 5: Nonlinear and Two-Photon Induced Fluorescence; Lakowicz, J. R., Ed.; Plenum Press: New York, 1997; Chapter 2.

3386 Analytical Chemistry, Vol. 70, No. 16, August 15, 1998

Figure 1. Differential polarized phase angle (upper panel) and polarized modulation ratio (lower panel) simulations for a hypothetical fluorophore that has an excited-state fluorescence lifetime of 4.00 ns and a rotational reorientation time of 1.00 ns under conditions where the limiting anisotropy is maximal: 0.40 (one photon), 0.57 (two photon), and 0.67 (three photon).

4-(dicyanomethylene)-2-methyl-6-(p-dimethylaminostyryl)-4H-pyran (DCM, Acros); rhodamine 6G (R6G, Exciton); fluorescein, fluorescein-5-maleimide, BODIPY C1-IA, BODIPY FL IA, and BODIPY 530/550 IA (Molecular Probes); bilirubin, flavin mononucleotide (FMN), flavine adenine dinucleotide (FAD), poly(ethylene glycol) (PEG) oligomers with an average molecular weight of 200, 300 and 400, trypsin-type III from bovine pancreas (10 600 units/mg of solid), 12 000 mw cutoff dialysis tubing (cellulose membrane), and essentially fatty acid-free bovine serum albumin (BSA) (Sigma); Na2HPO4, NaH2PO4‚2H2O, methanol, benzene, KOH, and hydrochloric acid (Fisher Scientific Co.); and ethanol (200 proof, Quantum Chemical Corp.). Unless noted to the contrary, all reagents were of the highest purity available and they were used as received without further purification. Aqueous solutions were prepared in doubly distilleddeionized water or phosphate buffer. All stock solutions were refrigerated in the dark at 4 °C until they were used. Multifrequency fluorescence experiments were generally performed at 21.0 ( 0.1 °C unless otherwise noted. FAD Purification. Commercial FAD is generally contaminated with a small amount of FMN. All our FAD samples were purified by passing an aqueous solution through a Bio-Gel P-2 loaded chromatographic column, using 0.05 M pH 7.5 phosphate buffer as the eluent. Pure FAD was found in those yellow column fractions that first eluted.

Figure 2. Simplified schematic of the new frequency-domain instrument that uses the harmonic content of a mode-locked, pulse-picked femtosecond Ti-sapphire laser as the excitation source and digital parallel acquisition to acquire one-, two-, or three-photon excited frequencydomain fluorescence data. Dashed lines denote optical paths. Solid lines are electrical connections. See text for further details and abbreviations.

Preparation of Fluorescein-Labeled Bovine Serum Albumin (BSA-FL). BSA contains a free thiol residue at position 34.68 We labeled BSA at position 34 with the thiol-reactive fluorescein derivative by dissolving BSA in a pH 7.0 phosphate buffer solution and then adding a 5-fold molar excess of fluorescein-5-maleimide in DMF. The reaction was carried out in a polypropylene centrifuge tube (Cole-Parmer). The tube was sealed, and the reaction was allowed to continue for 2 h in the dark at ambient temperature. The reaction mixture was transferred to a cellulose dialysis tube, and the preparation was dialyzed exhaustively against the labeling buffer solution until no fluorescence from the free probe was detected in the dialysate. All dialysis steps were carried out in the dark at 4 °C and were normally complete after 5 days if the dialysate was changed every 12 h. After the dialysis was complete, the solution containing the fluorescein-labeled BSA was removed from the dialysis tube, placed in a polypropylene centrifuge tube, sealed, and kept in the dark at 4 °C until use. Similar protocols have been used successfully to label thiol residues.69 The molar ratio of fluorescein to BSA in the final BSAFL was 0.95 ( 0.05. Trypsin Digestion of BSA-FL. A stock solution of 15 µM BSA-FL was prepared in phosphate buffer (0.1 M, pH 7.6). A stock trypsin solution was prepared by dissolving 150 mg of trypsin in 5 mL of phosphate buffer (0.1 M, pH 7.6). To initiate the digestion reaction, we added 500 µL of the trypsin stock solution directly to 3.000 mL of the stock BSA-FL solution; we then mixed the solution by hand by repeatedly inverting the cuvette for 3-5 s and then recorded the fluorescence information as a function of (68) Brown, J. R.; Shockley, P. Serum Albumins: Structure and Characterization of its Lipid Binding Sites. In Lipid Protein Interactions; Jost, P. C., Griffith, O. H., Eds., John Wiley & Son, New York, 1982. (69) Yao, Y.; Scho¨neich, C.; Squier, T. C. Biochemistry 1994, 33, 7797-810.

time. All trypsin digestion experiments were performed at 37 ( 0.1 °C. Fresh trypsin preparations were always used because trypsin will undergo autodigestion. Instrumentation. Figure 2 presents a simplified schematic of the new parallel multiharmonic frequency-domain fluorometer that has been constructed in our laboratory. The excitation source consists of a passively mode-locked, femtosecond Ti-sapphire laser (Coherent, model Mira 900F) that is pumped with 14 W (all lines) from a cw argon ion laser (Coherent, model Innova 400). The optimized operating parameters for the Ti-sapphire system with the long-wavelength optics (tunable continuously from 900 to 1000 nm) installed and the system tuned to ∼900 nm are as follows: repetition rate, ∼76 MHz; pulse duration, 150-170 femtosecond; and average mode-locked power, 1.1 W. The infrared output from the mode-locked Ti-sapphire laser is split by fused-silica beam splitters BS1 and BS2. The split beam produced by BS1 (∼4% of the primary power) is directed to a photodiode (Thor Labs, model DET200) that is used to track the Ti-sapphire repetition rate and synchronize the pulse picker driver electronics to the Ti-sapphire laser. The split beam produced at BS2 is directed into an optical spectrum analyzer (Imaging and Sensing Technology, model E201 LSA). The spectrum analyzer monitors the nominal Ti-sapphire wavelength output, the laser optical bandwidth/spectral profile, and serves as a simple mode-locking diagnostic. After passing through BS1 and BS2, the infrared pulse train is directed into a pulse picker (Coherent, Model 9200) where a TeO2 acousto-optically driven Bragg cell selectively extracts pulses from the 76-MHz infrared pulse train. In its normal configuration, the pulse picker produces variable repetition rate pulses by integer division of the fundamental pulse repetition rate between 9.5 MHz Analytical Chemistry, Vol. 70, No. 16, August 15, 1998

3387

(divide by 8) and 18.6 kHz (divide by 4096). The femtosecond train of picked pulses that exit from the Bragg cell serve as the modulated source for all subsequent frequency-domain measurements. Under our experimental conditions, the frequency-domain profile for a train of 150-170-fs-long pulses picked at, for example, 4 MHz is a continuous series of sinusoidally modulated wave forms at 4 MHz, 8 MHz, 12 MHz, etc., that persist up to the laser bandwidth limit. For the current Ti-sapphire system, this upper frequency limit is between 5 and 7 THz. Of course, the practical upper frequency limit for the instrument is set by the detector rise times and the bandwidth of the detection/cross-correlation electronics. The pulse picker “pick” frequency (f1) is derived and controlled from a high-precision, master ((1 µHz) synthesized function generator (FS1; Stanford Research, model DS345-02) whose TTL output is connected to the model 9200 controller. In order to provide phase coherence between the excitation source and the detection electronics, we phase lock the high-stability crystal clock within FS1 to a second synthesized function generator (FS2; Stanford Research model DS345) and then use FS2 to generate the cross-correlation heterodyne frequencies within the photomultiplier tube detectors (vide infra). On leaving the pulse picker, the laser beam takes two possible pathways to reach the sample. In the first path, moveable mirror MM1 is translated out of the infrared beam path and the beam is directed into a second and/or third harmonic generation system (CSK, model Super Doubler/Tripler). The second (Vis) or third (UV) harmonics are then directed with moveable mirrors MM2 or MM3 to mirror M2 (moveable mirror MM4 is translated out of the UV/Vis beam pathway). In a second pathway, mirror MM1 intercepts the pulse-picked infrared laser beam and directs it to MM4, which in turn directs the beam to M2. The exact polarization of the IR/Vis/UV beams are controlled by a series of beam translators and/or two half-wave plates. The full set of control optics that is used to ensure the correct polarization for the UV, Vis, or IR laser beams is omitted from Figure 2 for clarity. After the excitation beam polarization is selected, the laser beam is directed through beam splitter BS3. The split beam is directed to and reflected off the hypotenuse of a triangular quartz cuvette and passed through an appropriate neutral density filter (ND1), and the beam impinges onto the reference photomultiplier tube photocathode (DET1; Hamamatsu, model R928). The remainder of the excitation beam passes through a variable neutral density filter (ND2) and a fused silica Glan-Thompson “cleanup” polarizer (P1) that is oriented so that all samples are excited with vertically polarized electromagnetic radiation. The laser beam is finally focused into the sample by using a fused silica lens (L1). For all work reported here, standard 1-cm quartz cuvettes were used. The sample fluorescence is observed at 90° with respect to the excitation beam trajectory. The sample fluorescence is collected by a lens (L2) and collimated, passed through appropriate optical filters (F1) and a fused silica Glan-Thompson polarization analyzer (P3) that is oriented at the magic angle (54.7°), 0°, or 90°, and ultimately focused with a lens (L3) onto the photocathode of a second photomultiplier tube detector (DET2; Hamamatsu, model R928). In certain experiments, the optical filter is removed and a scanning emission monochromator is 3388 Analytical Chemistry, Vol. 70, No. 16, August 15, 1998

placed between L3 and DET2. The methodology used to acquire the frequency-domain information involves simultaneous cross-correlation within DET1 and DET2 at all modulation frequencies.57,58 Toward this end, the output from FS2 is typically adjusted to a frequency (f2) that is 23.0 Hz greater than the frequency selected for the master oscillator (FS1) that drives the pulse picker (i.e., f2 ) f1 + 23.0 Hz). The output from FS2 is then connected to a power amplifier (A1; SLM-Aminco/Spectronic Instruments), directed through a power splitter (PS; SLM-Aminco/Spectronic Instruments), and sent to a pair of matched harmonic comb generators (HCG1 and HCG2; SLM-Aminco/Spectronic Instruments) that are each connected directly to the second dynode on the reference and sample photomultiplier tube sockets (SLM-Aminco/Spectronic Instruments). On passing through the harmonic comb generators, the harmonics of f2 (i.e., f2, 2f2, 3f2, ...) are produced and all are directed simultaneously into DET1 and DET2 where simultaneous crosscorrelation is effected with all the harmonics of f1.57,58 The photocurrent outputs from DET1 and DET2 are directed to a data acquisition/control module (SLM-Aminco/Spectronic Instruments, model 4850) which is operated from a PC using software available from SLM-Aminco/Spectronic Instruments. The detector signals are essentially digitized with a high-speed analog-to-digital converter and the time-domain data are Fourier transformed into the frequency domain. The frequency-domain data of interest are encoded at the individual cross-correlation frequency harmonics. That is, if we drive the pulse picker at 4 MHz and use a 23.0-Hz cross-correlation frequency, the information (i.e., θ, M or ∆, Λ) associated with the 4-MHz data is encoded at 23.0 Hz, the 8-MHz data are found at 46.0 Hz, the 12-MHz data are associated with the 69.0-Hz signal, information at 16 MHz is found at 92.0 Hz, etc. The pulse picker was normally operated at 4.75 MHz, and the cross-correlation frequency was set at 23.0 Hz unless otherwise noted. Other primary/cross-correlation frequency combinations have been demonstrated with the current instrument. As a general rule it is critical that one choose ∆f to be a prime number. RESULTS AND DISCUSSION One-Photon Excitation. Intensity Decay Kinetics. Figure 3 presents the phase angle (solid symbols) and demodulation (open symbols) data for 0.10 µM R6G dissolved in ethanol that was acquired with the new instrument using one-photon excitation at 450 nm. These data represent two separate frequency-domain experiments that were acquired by using 4.00- and 4.75-MHz base frequencies. The solid traces represent the best fit to the experimental data by a single-exponential rate law (τ ) 3.900 ( 0.002 ns; χ2 ) 1.04), and the lower panel illustrates the residual plots for this particular data set. On the basis of the χ2 and the residual plots, one can clearly see that the intensity decay kinetics are well described by a single-exponential rate law. We compare the excited-state fluorescence lifetime for R6G recovered using the new frequency-domain fluorometer to values determined in our laboratory with another phase-modulation instrument and laser excitation at 514.5 nm (τ ) 3.85 ( 0.01 ns). These values are equivalent to one another at greater than the 95% confidence level. To further test the new instrument, we performed a simple set of quenching experiments on a series of 1.0 µM fluorescein

Figure 3. One-photon excited phase-modulation data, fits, and residual plot for 0.10 µM R6G dissolved in ethanol acquired with the new instrument. Two data sets that were acquired at f1 ) 4.00 and 4.75 MHz are overlayed here. λex ) 450 nm.

samples dissolved in 0.1 M aqueous KOH as a function of added KI. The Stern-Volmer plots of τo/τ vs [I-] (not shown) between 0.0 and 0.2 M KI was linear (r2 ) 0.993) and superimposable on a plot of Io/I vs [I-]. This behavior is indicative of an exclusively dynamic quenching process.70 The Stern-Volmer constant and bimolecular quenching rate (9.0 ( 0.2 M-1 and 2.25 ( 0.05 × 109 M-1 s-1, respectively) were within 5% of the values that we determined in our laboratory on an IBH 5000-W SAFE timecorrelated single-photon-counting fluorescence lifetime instrument. As an additional test of the new instruments accuracy, we measured the excited-state intensity decay kinetics for several dilute fluorophore solutions and compared the recovered excitedstate lifetimes to values published in the open literature or to values determined in our laboratory on the exact same samples using another phase-modulation instrument. The results of these comparative experiments are shown in Table 1. The overall agreement between the excited-state lifetimes that we determined using the new instrument is excellent on samples that exhibit excited-state fluorescence lifetimes between 0.2 and 11 ns. Moreover, in those cases where actually uncertainties are available, the results that were obtained with the new instrument are statistically equivalent to the benchmarks to at least the 95% confidence level. Careful inspection of Table 1 also shows that the new instrument is capable of performing very well on systems that (70) Eftink, M. R. In Topics in Fluorescence Spectroscopy. Volume 2: Principles; Lakowicz, J. R. Ed.; Plenum Press: New York, 1991; Chapter 2.

exhibit multiexponential intensity decays. Two systems were investigated for this express purpose: (1) a mixture of fluorescein and bilirubin in a BSA solution at pH 7.00 and (2) a dilute solution of purified FAD in pH 7.50 buffer. In both samples, a singleexponential rate law does not describe the experimental data very well (χ2 > 1.8; nonrandom residuals) and an adequate fit is achieved only when the phase-modulation data are fit to a doubleexponential rate law. The kinetic parameters recovered for these two systems agree very closely with values recovered on a different instrument or with values published in the literature from another laboratory. The results for FAD are especially encouraging (the double-exponential decay is a result of intramolecular quenching of the flavin chromophore by the adenine residue)75 because we were successful at accurately recovering a fractional intensity (do not confuse fri and Ri; fri ) Riτi/∑Rjτj) term associated with the shorter-lived component (∼0.3-ns species) that is less than 10% of the total fluorescence intensity. Anisotropy Decay Kinetics. Figure 4 presents the differential polarized phase angle and polarized modulation ratio data (left half of the figure) for 1.0 µM R6G dissolved in 200 MW PEG acquired with the new instrument under one-photon excitation at 450 nm. The corresponding residual plots are shown in the righthand panel sets. The solid traces through the experimental data represent the best fit to the experimental data by a singleexponential rate law (ro ) 0.38 ( 0.01; φ ) 6.34 ( 0.1 ns; χ2 ) 1.07). These data are clearly well modeled by an isotropic rotor model, and these results agree well with values determined on another instrument (vide infra). Perylene is a well-known anisotropic rotor that exhibits two discrete rotational reorientation times.76,77 Figure 5 presents the one-photon excited differential polarized phase angle and polarized modulation ratio data (left half of the plot) for 1.0 µM perylene dissolved in 400 MW PEG acquired using the new instrument with excitation at 450 nm. The basic presentation format is identical to Figure 4. The solid curves represent the best fit between the experimental data and an isotropic rotor model. The χ2 for this particular fit is 19.4, the residuals are clearly systematic, and the fit is obviously inadequate. The dashed curves represent the best fit of the same data to an anisotropic rotor model with two discrete rotational reorientation times (ro ) 0.31 ( 0.02; φ1 ) 2.09 ( 0.2 ns; β1 ) 0.58 ( 0.06; φ2 ) 0.48 ( 0.11 ns; β2 ) 0.42 ( 0.05; χ2 ) 1.29). Table 2 compares the rotational reorientation dynamics that are recovered with the new instrument to values determined on a different instrument or published in the open literature. The general agreement between the two columns of results is very good, and the values are statistically equivalent to one another at the 90% confidence level. These data also show that the new frequency-domain instrument is capable of acquiring accurate rotational reorientation times and resolving anisotropic rotational (71) Bourson, J.; Doizi, D.; Lambert, D.; Sacaze, T.; Valeur, B. Opt. Commun. 1989, 72, 367-70. (72) Drake, J. M.; Lesiecki, M. L.; Camaioni, D. M. Chem. Phys. Lett. 1985, 113, 530-4. (73) Meyer, M.; Mialocq, J. C.; Rouge´e, M. Chem. Phys. Lett. 1988, 150, 48490. (74) Heitz, M. P.; Bright, F. V. J. Phys. Chem. 1996, 100, 6889-97. (75) Visser, A. J. W. G. Photochem. Photobiol. 1984, 40, 703-6. (76) Lakowicz, J. R.; Cherek, H.; Maliwal, B. P. Biochemistry 1985, 24, 376-83. (77) Barkley, M. D.; Kowalczyk, A. A.; Brand, L. J. Chem. Phys. 1981, 75, 358193.

Analytical Chemistry, Vol. 70, No. 16, August 15, 1998

3389

Table 1. Recovered One-Photon Excited Fluorescence Lifetimes (ns) Measured with the New MHF System Compared to Values Recovered Using Other Instrumentation system fluorescein in KOH R6G in EtOH DCM in MeOH

DCM in DMSO

rubrene in benzene BTBP in acetone FMN in pH 7.5 buffer bilirubin + BSA in pH 7.00 buffer fluorescein + bilirubin + BSA in pH 7.00 buffer FAD in pH 7.5 buffer

reference values 4.00 ( 0.02b 3.85 ( 0.01b 1.28c 1.31d 1.38e 1.31 ( 0.008b 2.15c 2.18d 2.24e 2.20 ( 0.01b 11.0 ( 0.024b 3.68 ( 0.01f 4.70g 4.65 ( 0.03b 0.194 ( 0.007b 3.88(0.61):0.21(0.39)b,h 2.82(0.72):0.31(0.28)g,h 2.75(0.70):0.28(0.30)b,h

new MHF systema 3.98 ( 0.004 3.90 ( 0.002 1.31 ( 0.005

2.15 ( 0.006 10.6 ( 0.08 3.72 ( 0.002 4.55 ( 0.01 0.209 ( 0.009 3.74(0.54):0.29(0.46)h 2.71(0.66):0.35(0.34)h

a Measured in this laboratory using the new instrument. λ ) 450 nm. T ) 21.0 ( 0.1 °C b Measured in this laboratory on a commercial SLM ex 48000 MHF, with a Pockels cell modulator, and a CW argon ion laser as the excitation source. c From ref 71 at 25 °C. d From ref 72 at 25 °C.e From f g ref 73 at 20 °C. From ref 74 at 20 °C. From ref 75 at 20 °C. h τ1(R1):τ2(R2).

Figure 4. One-photon excited differential polarized phase angle and polarized modulation ratio data, fits and residual plots for 1.0 µM R6G dissolved in 200 MW PEG acquired with the new instrument. λex ) 450 nm.

Figure 5. One-photon excited differential polarized phase and polarized modulation data, fits and residual plots for 1.0 µM perylene dissolved in 400 MW PEG acquired with the new instrument. λex ) 450 nm.

reorientation dynamics. The biggest exception is the perylene/ 400 MW PEG data where the agreement is only fair. We speculate that these results arise for two possible reasons. First, under the particular experimental conditions that were used to acquire these data, the perylene dynamics are relatively fast and one of the motions is subnanosecond. Thus, even though we can clearly discriminate between isotropic and anisotropic models (Figure 5), these dynamics remain a difficult resolution problem. Second, PEG solutions are extremely hygroscopic and we suspect that some of the differences between entries in Table 2 result from slightly different water contents between the two PEG samples.

Two-Photon Excitation. Steady-State Emission Spectra. Prior to performing any frequency-domain two-photon excited emission experiments, we investigated the steady-state emission spectra and the effects of incident laser power on the static fluorescence from 1.0 µM R6G dissolved in ethanol when excited at 450 (onephoton) and 900 nm (two-photon). The emission spectra (not shown) exhibit identical spectral profiles and the peak maximums were always within 2 nm of one another. A plot of the fluorescence intensity vs incident laser power (not shown) demonstrates that the R6G emission observed when the sample is excited at 900 nm results from a two-photon excitation process (measured

3390 Analytical Chemistry, Vol. 70, No. 16, August 15, 1998

Table 2. Recovered One-Photon Excited Rotational Reorientation Times (ns) Measured with the New MHF System Compared to Values Recovered Using Other Instrumentation system

reference values

R6G in water at 21 °C R6G in PEG 200 at 21 °C R6G in PEG 400 at 21 °C fluorescein in propylene glycol at 25 °C

0.180 ( 0.025b 6.58 ( 0.11b 9.11 ( 0.03b 5.8c 5.91 ( 0.05b 12.6 (0.62): 2.03 (0.38)c,d 2.56 (0.57): 0.28 (0.43)b,d

perylene in propylene glycol at -9 °C perylene in PEG 400 at 21 °C

new MHF systema 0.177 ( 0.021 6.34 ( 0.10 8.72 ( 0.06 5.91 ( 0.07 12.8 (0.56): 1.94 (0.44)d 2.09 (0.58): 0.48 (0.42)d

a Measured in this laboratory using the new instrument. λ ) 450 nm. b Measured in this laboratory on a commercial SLM 48000 MHF, with ex a Pockels cell modulator, and a CW argon ion laser as the excitation source. R6G excited at 514.5 nm. Fluorescein excited at 488.0 nm. Perylene excited at 351.1 nm. c From ref 76. Fluorescein excited at 488 nm. Perylene excited at 351.1 nm. d φ1(β1):φ2(β2).

Figure 6. One- and two-photon excited phase-modulation data sets and fit to a single-exponential rate law for 1.0 µM R6G dissolved in ethanol acquired with the new instrument. One-photon excitation at 450 nm. Two-photon excitation at 900 nm.

slope, 1.9 ( 0.03; predicted slope for a biphotonic process, 2.0). Similar experiments were carried out on all the samples described in this section prior to making any frequency-domain measurements, and the measured slope was always between 1.8 and 2.1. Intensity Decay Kinetics. Figure 6 presents the phase angle and demodulation data for 1.0 µM R6G dissolved in ethanol that we acquired with the new instrument under one- and two-photon excitation at 450 and 900 nm, respectively. The total acquisition time for each data set was only 1 min. The points represent the experimental data, and the solid curve denotes the best fit between the data and a single-exponential rate law (〈τ〉 ) 3.92 ( 0.011 ns; 〈χ2〉 ) 1.12). These results show that the overlap between the one- and two-photon excited phase-modulation data is extremely high and the recovered excited-state lifetimes for R6G in ethanol under one- or two-photon excitation are statistically equivalent. Figure 7 presents phase-modulation data for 5.0 µM BODIPY FL1A dissolved in 0.10 M phosphate buffer (pH 7.00) acquired using the new instrument with two-photon excitation at 900 nm. The solid curves illustrate the best fit between the experimental data and a double-exponential intensity decay model (τ1 ) 5.58 ns, fr1 ) 0.98; τ2 ) 0.09 ns; fr2 ) 0.02; χ2 ) 1.03). The best fit to a single-exponential rate law under one- or two-photon excitation (not shown) for all the free BODIPY dyes generally exhibits statistically greater χ2 values (1.4 e χ2 < 2.1) relative to the doubleexponential fit. However, in the double-exponential decay fits, one of the recovered lifetimes is generally subnanosecond and contributes less than 5% to the total emission. For this reason,

Figure 7. Two-photon excited phase-modulation data, fit and residual plot for 5.0 µM BODIPY FL IA dissolved in buffer acquired with the new instrument. The points are the experimental data, and the solid curves are the best fit to a double-exponential rate law. λex ) 900 nm.

we compare here only the average BODIPY excited-state fluorescence lifetimes under one- and two-photon excitation. We feel it is entirely possible that these particular BODIPY dyes are intrinsically multiexponential and/or different photoselection rules might allow one to access different excited-state processes under one- and two-photon excitation. These issues are currently under investigation in our laboratory. Table 3 compares the average excited-state fluorescence lifetimes for R6G dissolved in ethanol or in the various molecular weight PEG samples along with the three BODIPY dyes dissolved in buffer under one- and two-photon excitation. Inspection of these results shows that, within our existing measurement precision and subject to the constraints placed on the BODIPY samples (vide supra), the excited-state fluorescence lifetimes for each of these Analytical Chemistry, Vol. 70, No. 16, August 15, 1998

3391

Table 3. Comparison between the Average Recovered One- and Two-Photon Excited Fluorescence Lifetimes (ns) Measured with the New MHF System at 21.0 ( 0.1 °C sample

one-photon excitation at 450 nm

two-photon excitation at 900 nm

R6G in ethanol R6G in 200 MW PEG R6G in 300 MW PEG R6G in 400 MW PEG BODIPY C1-IA BODIPY FL IA BODIPY 530/550 IA

3.90 ( 0.002 4.22 ( 0.014 4.77 ( 0.012 4.78 ( 0.013 7.50 ( 0.12 5.61 ( 0.04 6.73 ( 0.09

3.92 ( 0.011 4.28 ( 0.017 4.71 ( 0.036 4.77 ( 0.025 7.44 ( 0.17 5.50 ( 0.09 6.50 ( 0.15

Figure 8. One- and two-photon excited dynamical measurements in an optically dense aqueous sample that contains 10 µM aqueous R6G and excess bromocresol green. (Upper panel) One- (open symbols) and two-photon excited (solid symbols) phase-modulation data sets acquired with the new instrument and fits to singleexponential decay models. (Lower panel) Absorbance spectrum for the optically dense solutions. One-photon excitation at 450 nm. Twophoton excitation at 900 nm.

species are statistically equivalent and thus independent of the excitation mode (one vs two photon). One of the primary attractions of MPE as an analytical tool is the ability to excite fluorescent species in optically dense samples.8,9 In Figure 8 we present phase angle-modulation data (upper panel) under one- and two-photon excitation for a 10 µM aqueous R6G solution that contains a high concentration of bromocresol green. The absorbance spectrum for this solution in a standard 1-cm quartz cuvette is presented in the lower panel of Figure 8. This sample exhibits an absorbance in excess of four over the wavelength region where one typically excites R6G under onephoton conditions. In the upper panel, the one-photon excited phase-modulation data for this sample are represented by the open 3392 Analytical Chemistry, Vol. 70, No. 16, August 15, 1998

symbols and the corresponding solid curves that pass nearest these points denote the best fit between the data and a singleexponential decay model. These one-photon excited phasemodulation data are highly distorted, the R6G fluorescence is so weak that the emission detector was operated at the highest recommended voltage and gain, unrealistic phase angles (>90°) are observed, and the fit between the data and a single-exponential rate law is poor (〈τ〉 ) 7.1 ns; χ2 ) 971). The filled symbols in Figure 8 (upper panel) represent the phase-modulation data obtained for the exact same sample when excited by two 900-nm photons. These data do not exhibit any of the distortions seen under one-photon excitation. The solid curves passing through these data points correspond to the best fit between the data and a single-exponential rate law (τ ) 3.88 ( 0.07 ns; χ2 ) 1.14). The recovered excited-state fluorescence lifetime under two-photon excitation is statistically equivalent to the R6G excited-state lifetime measured in the absence of bromocresol green (Table 1). These results demonstrate the potential of the new frequency-domain instrument for making accurate and precise excited-state lifetime measurements directly in optically dense media. To better assess the potential of the new instrument for making reliable excited-state fluorescence lifetime measurements in optically dense media, we repeated the fluorescein/I- experiments that we described above, but we added enough bromocresol green to produce an absorbance in excess of four at 450 nm. The phasemodulation traces that we acquired for these samples were all similar in basic character to those shown in Figure 8. Under twophoton excitation conditions, the phase-modulation data (not shown) are well described by a single-exponential decay law (χ2av ) 1.27) and the recovered Stern-Volmer constant and bimolecular quenching rate for the quenching process in these optically dense samples were 9.5 ( 0.6 M-1 and 2.07 ( 0.17 × 109 M-1 s-1, respectively. These values agree remarkably well with the results in the absence of any bromocresol green (vide supra) and argue convincingly that the new instrument is capable of rapidly making accurate excited-state fluorescence lifetime measurements in optically dense samples. To more fully investigate the potential of the new instrument, we performed a series of two-photon excited fluorescence experiments on 15 µM fluorescein dissolved in undiluted, whole human blood at several pH values. At pH 5.0 in neat buffer, fluorescein exists exclusively in the anion form (carboxylate or phenolate), exhibits a quantum yield of 0.37, and has an excited-state fluorescence lifetime of 3.0 ns.78 As one raises the buffer pH to (78) Sjo ¨back, R.; Nygren, J.; Kubista, M. Spectrochim. Acta A 1995, 51, L7L21.

Table 4. Recovered Excited-State Fluorescence Lifetimes (ns) and Fractional Contributions, from Global Analysis,79 for 15 µM Fluorescein in Undiluted, Whole Blood as a Function of pH under Two-Photon Excitation at 900 nm pH

τ1

τ2

fr1

5.1 6.8 7.5

2.88 ( 0.21 2.88 ( 0.21 2.88 ( 0.21

3.80 ( 0.11 3.80 ( 0.11 3.80 ( 0.11

0.92 ( 0.08 0.31 ( 0.06 0.02 ( 0.02

7.5, fluorescein exists predominately as a dianion that exibits an excited-state fluorescence lifetime of 4.1 ns and a quantum yield of 0.93.78 In neat buffers, we performed a set of phase-modulation experiments on fluorescein excited under one- and two-photon conditions at pH 5.0 and 7.5. The phase-modulation data were single exponential at these pH values regardless of the excitation mode, and we recovered excited-state fluorescence lifetimes of 2.97 ( 0.05 and 4.02 ( 0.01 ns at pH 5.0 and 7.5, respectively. We next performed a set of similar experiments on fluorescein dissolved in whole blood at pH 5.1, 6.8, and 7.5 under one- and two-photon excitation conditions at 450 and 900 nm, respectively. In these particular experiments, we repositioned the sample cuvette so that the excitation and emission only passed through ∼1 mm of the blood sample. The one-photon excited phasemodulation data could not be measured for the blood samples at all under any conditions we attempted. By using two-photon excitation at 900 nm, we obtained phase-modulation data (not shown) on our initial attempt. The two-photon excited phasemodulation data for fluorescein in blood at pH 5.1 and 7.5 were best described by a single-exponential rate law with excited-state fluorescence lifetimes of 2.85 ( 0.3 and 3.84 ( 0.02 ns. Interestingly, the corresponding data at pH 6.8 could only be well fit to a double-exponential decay law. Simultaneous global analysis79 of all the pH-dependent two-photon excited phase-modulation data sets on the fluorescein/blood system yielded the results shown in Table 4. These results are very much in line with the known protolytic equilibria for fluorescein78 and demonstrate several key aspects of the new instrument. First, the new instrument can directly recover accurate excited-state intensity decay information from dilute fluorophores dissolved in whole blood. Second, multiexponential intensity decay kinetics can be resolved in whole blood. Finally, we can resolve excited-state fluorescence lifetimes in an optically dense media like whole blood that differ by only 1 ns. Together these results argue for the utility of the new instrument for analyzing samples of clinical interest and the like. Anisotropy Decay Kinetics. Figure 9 presents the differential polarized phase angle and polarized modulation ratio data (left half of the figure) for 10 µM R6G dissolved in 200 MW PEG acquired using the new instrument under one- and two-photon excitation. The corresponding residual plots for the two-photon data are shown in the right-hand panel sets. The solid traces through the experimental data represent the best fit between the experimental data and a single-exponential rate law. For the onephoton excitation at 450 nm the recovered parameters are ro ) (79) Beechem, J. M.; Gratton, E.; Ameloot, M.; Knutson, J. R.; Brand, L. In Topics in Fluorescence Spectroscopy. Volume 2: Principles; Lakowicz, J. R. Ed.; Plenum Press: New York, 1991; Chapter 5.

Figure 9. One- and two-photon excited differential polarized phase angle and polarized modulation ratio data sets, fits, and residual plots for 10 µM R6G dissolved in PEG MW 200 acquired with the new instrument. The solid squares are the experimental data under onephoton excitation (450 nm), and the solid curves are the best fit to a single-exponential rate law. The solid circles are the experimental data under two-photon excitation (900 nm), and the solid curves are the best fit to a single-exponential rate law. Residuals are shown only for the two-photon excited data.

0.38 ( 0.01 and φ ) 6.34 ( 0.1 ns (χ2 ) 1.06). Under two-photon excitation at 900 nm the recovered parameters are ro ) 0.49 ( 0.02 and φ ) 6.58 ( 0.83 ns (χ2 ) 1.24). These data are clearly well described by an isotropic rotor model, and these results agree well with values determined using another phase-modulation instrument under one-photon excitation (vide infra). The recovered rotational reorientation times are statistically equivalent to one another and the limiting anisotropy under two-photon excitation is clearly, as expected,63-67 much greater than the one-photon anisotropy. We performed a simultaneous global analysis79 on the two data sets shown in Figure 9 where we linked φ between the two data sets and allowed only the limiting anisotropies to float. The fit (not shown) using this protocol was very good (χ2global ) 1.13), the residuals were random, and the recovered rotational reorientation time was 6.41 ( 0.29 ns. The limiting anisotropy terms remained statistically equivalent to the values recovered in the discrete analyses. A second series of experiments (not shown) on R6G dissolved in 400 MW PEG yielded rotation reorientation times and limiting anisotropies under one- and two-photon excitation of 8.67 ( 0.24 ns and 0.38 ( 0.01, 8.72 ( 0.65 ns and 0.49 ( 0.02, respectively. Global analysis of these data sets resulted in a recovered rotational reorientation time of 8.89 ( 0.21 ns. Ingersoll et al.80 recently illustrated the potential of frequencydomain fluorescence spectroscopy with one-photon excitation as (80) Ingersoll, C. M.; Watkins, A. N.; Baker, G. A.; Bright, F. V. Appl. Spectrosc., in press.

Analytical Chemistry, Vol. 70, No. 16, August 15, 1998

3393

a tool for acquiring nanosecond and subnanosecond dynamical data on protein systems on the fly as they undergo reaction. To further explore the potential of the new instrument for on-the-fly measurements in a multiphoton excitation mode, we investigated bovine serum albumin that we labeled site selectively at cysteine34 with the fluorescent probe fluorescein (BSA-FL) as the protein reacted with the enzyme trypsin. Cys-34 is located within loop 1 of domain I in the BSA protein, and fluorescently labeled BSA has been studied extensively in our laboratories.80-84 Previous work has demonstrated that the fluorescence anisotropy decay for fluorescently labeled BSA is best described by an anisotropic rotor model with two discrete rotational reorientation times. The biexponential decay of fluorescence anisotropy results primarily from simultaneous global rotational reorientation of the entire BSA protein and local precessional motion of the fluorescent reporter group that is attached at Cys-34. Trypsin cleaves peptides at the carboxy side of lysine and arginine.85,86 In BSA-FL, the key lysine residues are located at positions 20 and 41.68 Thus, reaction of BSA-FL with trypsin results in the excision of a small peptide segment consisting of a 21-amino acid peptide that hosts the FL reporter group attached at Cys-34. Under our experimental conditions, this reaction occurs in ∼15 min. Following the protocol developed by Ingersoll et al.,80 we recorded the differential phase and polarized modulation ratio data on BSA-FL under two-photon excitation conditions at 900 nm as the protein was enzymatically digested by trypsin. We acquired complete differential phase and polarized modulation data sets every 0.5 min for a period of 1 h and then analyzed the available data using global analysis.79 We anticipated the BSA-FL/trypsin system to behave as follows. Based on our previous work,80-84 the native system before addition of trypsin should be described by a two-term rate law (i.e., global protein motion and FL local motion).80-84 As we add trypsin, the enzyme should begin to digest the BSA and excise a small 21-amino acid-long peptide fragment that hosts the FL residue. Thus, one might anticipate the system rotational reorientation dynamics to be described by a triple-exponential decay model (local FL motion within the BSA, BSA global motion, and motion of the FL residue attached to the excised peptide unit) during the digestion process. Global analysis of all the available on-the-fly, two-photon excited BSA-FL data files revealed that the system is described best by an anisotropic rotor model with two discrete motions at all times (χ2global ) 1.26, random residuals). Tests of other models80 with greater or fewer terms and/or other global linking schemes did not yield any improvement in the fit compared to the two-term rate law. The best-fit results are summarized in Figure 10, and they show several interesting trends. First, the longer of the two rotational reorientation times can be linked at all reaction times without any statistical decrease in fit quality or increase in residuals. The recovered rotational reorientation time under these (81) Wang, R.; Sun, S.; Bekos, E. J.; Bright, F. V. Anal. Chem. 1995, 67, 14959. (82) Lundgren, J. S.; Heitz, M. P.; Bright, F. V. Anal. Chem. 1995, 67, 377581. (83) Ingersoll, C. M.; Jordan, J. D.; Bright, F. V. Anal. Chem. 1996, 68, 31948. (84) Jordan, J. D.; Dunbar, R. A.; Bright, F. V. Anal. Chem. 1995, 67, 2436-43. (85) Stryer, L. Biochemistry, 3rd ed.; W. H. Freeman and Co.: New York, 1988; Chapter 3, p 56, Chapter 8, p 178. (86) Wilkinson, J. M. In Practical Protein Chemistry: A Handbook; Darbre, A., Ed.; Wiley: Chichester, 1988; Chapter 3, p 132.

3394 Analytical Chemistry, Vol. 70, No. 16, August 15, 1998

Figure 10. Summary of the on-the-fly fluorescence anisotropy decay kinetics of BSA-FL reacting with trypsin. (Top panel) Fractional contributions to rotational reorientation times 2 (b) and 1 (O) as a function of reaction time. (Lower panel) Rotational reorientation times for φ1 as a function of reaction time. φ2 is independent of reaction time, 35 ( 0.5 ns.

conditions is 35 ( 0.5 ns. On the basis of our previous work,80-84 we assign this slower motion to the global rotational reorientation of the intact BSA-FL protein. Second, the faster rotational reorientation time begins at 0.61 ( 0.04 ns and it decreases to ∼0.15 ns within 10-12 min. This rotational reorientation time most likely reflects the local FL dynamics within the intact BSAFL molecule convolved with the FL dynamics that are associated with the 21-amino acid-long peptide fragment that is excised by trypsin from the intact BSA-FL molecule. We are currently unable to resolve these two subnanosecond dynamical processes, but it is satisfying to see that the rotational reorientation time for free FL in buffer under the same experimental conditions is about 150170 ps. Third, we see that the fractional contribution associated with the faster rotational reorientation time, β1, increases from 0.25 to 0.97 as the reaction proceeds. As a consequence, the fractional contribution associated with the global BSA rotational reorientation, β2, decreases from 0.75 to only 0.03 as the BSA protein is essentially digested completely into smaller peptide segments. These results highlight the speed with which the new instrument can acquire on-the-fly dynamical information in a multiphoton excitation mode. Three-Photon Excitation. Steady-State Emission Spectra. Prior to performing any three-photon experiments, we investigated the steady-state emission spectra and the effects of incident laser power on PPD dissolved in ethanol when excited at 257.2 (one photon) and 900 nm (three photon). The emission spectra (not

Figure 11. One- and three-photon excited phase-modulation data sets and fits for PPD dissolved in ethanol. One-photon excitation at 257.2 nm. Three-photon excitation at 900 nm acquired with the new instrument.

shown) exhibit reasonably similar spectral profiles regardless of the excitation mode. The PPD fluorescence intensity vs incident laser power plot (not shown) demonstrated that the PPD emission seen when exciting at 900 nm results predominately from a threephoton excitation process (measured slope, 3.02 ( 0.19; predicted slope for an exclusively three-photon event, 3.0). Intensity Decay Kinetics. Figure 11 presents phase angle and demodulation data for PPD dissolved in ethanol acquired under one- and three-photon excitation at 257.2 (1 µM PPD) and 900 nm (10 mM PPD), respectively. Very similar data were also obtained by using the third harmonic on the new instrument to excited PPD via a one-photon process (not shown). The 257.2nm data were obtained by using an SLM 48000 MHF in concert with a Lexel model 95 SHG intracavity doubled argon ion laser as the excitation source. The open points illustrate the experimental data obtained under one-photon excitation at 257.2 nm, and the dashed curve through these points denotes the best fit between the experimental data and a single-exponential rate law (τ ) 1.30 ( 0.02 ns; χ2 ) 1.14). The solid symbols illustrate the experimental phase-modulation data that were acquired with threephoton excitation at 900 nm by using the new instrument. The solid curves passing through these points represent the best fit between the experimental data and a single-exponential rate law (τ ) 1.43 ( 0.14 ns; χ2 ) 1.27). As a comparison, Lakowicz and co-workers reported the excited-state lifetime for PPD in ethanol to be 1.20 ( 0.2 ns.87 Anisotropy Decay Kinetics. Figure 12 presents the differential polarized phase angle and polarized modulation ratio data (left half of the figure) for PPD dissolved in propylene glycol under one- and three-photon excitation. The corresponding residual plots for the three-photon data are shown in the right-hand panel sets. The solid traces passing through the data points represent the best fit between the experimental data and an isotropic rotor model. For the one-photon excitation at 257.2 nm, the recovered parameters are ro ) 0.25 ( 0.05 and φ ) 5.82 ( 0.3 ns (χ2 ) 1.21). For three-photon excitation at 900 nm, the recovered dynamical parameters are ro ) 0.58 ( 0.07 and φ ) 6.86 ( 1.15 ns (χ2 ) 1.36). These data are reasonably well described by an (87) Lakowicz, J. R.; Cherek, H.; Balter, A. J. Biochem. Biophys. Methods 1981, 5, 131-46.

Figure 12. One- and three-photon excited differential polarized phase angle and polarized modulation ratio data sets, fits, and residual plots for PPD dissolved in propylene glycol acquired with the new instrument. The open squares are the experimental data acquired under one-photon excitation (257.2 nm), and the solid curves are the best fit to a single-exponential rate law. The solid circles are the experimental data under three-photon excitation (900 nm), and the solid curves are the best fit to a single-exponential rate law. Residuals are shown only for the three-photon excited data.

isotropic rotor model, and attempts to fit these data to other models failed to yield any perceptible improvement in the fit quality and/or residuals. Interestingly, although the one- and three-photon limiting anisotropies generally behave as anticipated, the recovered rotational reorientation times for PPD under oneand three-photon excitation are not statistically equivalent. At this time we are not sure of the origin behind such a difference. We speculate that it may simply arise from slight differences between the samples used for the one- and three-photon experiments (e.g., because of our current optical configuration, the samples for the three-photon experiments are much more concentrated compared to the one-photon excited samples) and/or local sample heating at 257.2-nm excitation. CONCLUSIONS We have reported the construction and analytical performance of a new frequency-domain fluorometer capable of accurate and precise time-resolved fluorescence anisotropy and intensity decay measurements under one-, two-, or three-photon excitation conditions. The new instrument, based on a mode-locked, pulse-picked femtosecond Ti-sapphire laser, uses a parallel data acquisition strategy which allows one to acquire complete frequency-domain data sets in as little as 0.5 min. We demonstrated the potential of this new instrument by measuring the excited-state fluorescence lifetimes and rotational reorientation times for fluorophores and fluorophore mixtures that exhibit well-known nanosecond and subnanosecond single-exponential and multiexponential decay kinetics under one-, two-, or three-photon excitation. The instruAnalytical Chemistry, Vol. 70, No. 16, August 15, 1998

3395

ment was able to accurately recover the fluorescence decay kinetics and offered a measurement precision on the order of 5-10 ps. We tested the instrument’s capabilities further by performing time-resolved intensity decay measurements on dilute fluorophore solutions in optically dense samples such as bromocresol green and undiluted whole blood under two-photon excitation conditions. The results of these experiments showed that the new instrument can accurately recover the true fluorophore decay kinetics in these optically dense samples. We also demonstrated the speed of the new instrument by recording the time-resolved anisotropy decay kinetics associated with fluorescein-labeled bovine serum albumin under two-photon excitation conditions on the fly as it was digested by the enzyme trypsin. Complete, 100 frequency data sets were acquired on the fly every 0.5 min. The results provided insights into the complex events that occur as BSA-FL is digested by trypsin. The time resolution of the new instrument is currently limited by the photomultiplier tube detectors which set the practical upper modulation frequency limit to between 250 and 300 MHz. We are currently exploring interfacing the instrument

3396 Analytical Chemistry, Vol. 70, No. 16, August 15, 1998

with a microchannel plate photomultiplier tube as a means to increase the upper frequency limit and hence the time resolution. Dedication. This paper is dedicated to Professor Gary M. Hieftje on the occasion of his winning the 1998 American Chemical Society Award for Excellence in Teaching. ACKNOWLEDGMENT This work was supported in part by the National Science Foundation, the Office of Naval Research, and the Division of Chemical Sciences, Office of Basic Energy Sciences, Office of Energy Research, United States Department of Energy. We also acknowledge Dr. Robert D. Fugate for his helpful discussions and insights. Lexel Lasers is acknowledged for providing the model 95 SHG intracavity double argon ion laser that was used in part of this work. Finally, we thank one of the Reviewers whose comments prompted us to attempt measurements in whole blood. Received for review March 27, 1998. Accepted June 1, 1998. AC9803481