Anal. Chem. 1996, 68, 2270-2276
Reduction of Luminescent Background in Ultrasensitive Fluorescence Detection by Photobleaching Rhett L. Affleck,†,§ W. Patrick Ambrose,† James N. Demas,‡ Peter M. Goodwin,† Jay A. Schecker,†,⊥ Ming Wu,†,| and Richard A. Keller*,†
Chemical Science and Technology Division, Los Alamos National Laboratory, Los Alamos, New Mexico 87545, and Department of Chemistry, University of Virginia, Charlottesville, Virginia 22901
Chemical Science and Technology Division, Los Alamos National Laboratory. University of Virginia. § Present address: Pharmacopeia Inc., 101 College Rd. East, Princeton, NJ 08540. ⊥ Present address: Los Alamos Science, Los Alamos National Laboratory. | Present address: Department of Advanced Technology, Brookhaven National Laboratory, Upton, NY 11973. (1) Shera, E. B.; Seitzinger, N. K.; Davis, L. N.; Keller, R. A.; Soper, S. A. Chem. Phys. Lett. 1990, 174, 553-557. (2) Soper, S. A.; Shera, E. B.; Martin, J. C.; Jett, J. H.; Hahn. J. H.; Nutter, H. L.; Keller, R. A. Anal. Chem. 1991, 63, 432-437. (3) Soper, S. A.; Davis, L. M.; Shera, E. B. J. Opt. Soc. Am. B 1992, 9, 17611769. (4) Whitten, W. B.; Ramsey, J. M.; Arnold, S.; Bronk, B. V. Anal. Chem. 1991, 63, 1027-1031. (5) Soper, S. A.; Mattingly, Q. L.; Vegunta, P. Anal. Chem. 1993, 65, 740747. (6) Wilkerson, C. W.; Goodwin, P. M.; Ambrose, W. P.; Martin, J. C.; Keller, R. A. Appl. Phys. Lett. 1993, 62, 2030-2032. (7) Lee, Y.-H.; Maus, R. G.; Smith, B. W.; Winefordner, J. D. Anal. Chem. 1994, 66, 4142-4149. (8) Rigler, R.; Widengren, J.; Mets, U. In Fluorescence Spectroscopy; Wolfbeis, O. S., Ed.; Springer-Verlag: Berlin, 1993; pp 13-24. (9) Mets, U.; Rigler, R. J. Fluor. 1994, 4, 259-264. (10) Nie, S.; Chiu, D. T.; Zare, R. N. Science 1994, 266, 1018-1021. (11) Nie, S.; Chiu, D. T.; Zare, R. N. Anal. Chem. 1995, 67, 2849-2857. (12) Li, L.-Q.; Davis, L. M. Appl. Opt. 1995, 34, 3208-3217. (13) Shapiro, H. M. Practical Flow Cytometry; Wiley-Liss: New York, 1995; pp 133-143.
natural combination of single-molecule detection (SMD) and flow cytometry has created an exciting arena of analytical techniques that have just begun to be explored. To detect single molecules, it is important to minimize background from scattering and from fluorescent impurities in the solvent. Approaches to limiting the background include extremely high rejection, complementary excitation and emission filters, time gating to eliminate Rayleigh and Raman scattering, and small probe volumes. In spite of all of these improvements, the detection limits and the ability to discriminate between different fluorophores are still limited by background luminescence. The signal-to-noise ratio for single-molecule detection is inversely related to the probe volume. Impressive signal-to-noise ratios have been demonstrated for probe volumes in the femtoliter range for static SMD in liquids, where single molecules diffuse in and out of the probe volume.8-11 While it is important to use very small probe volumes, in order to process samples efficiently and to ensure detection of most of the sample, the detection volume must be large enough that the sample stream flows through the central portion of the detection volume. The detection volume is defined by the intersection of the focused excitation laser and the image of a spatial filter in the flow cell. To incorporate the usefulness of flow-based SMD techniques, we find it convenient to use a detection volume on the order of a picoliter, and this increases the solvent background. Therefore, every possible action must be used to reduce the level of luminescent impurities in the solvent. A fluorescent molecule emits many photons while passing through the excitation laser beam, resulting in a burst of photons distinguishable from the background. If the sample is appropriately dilute, most bursts represent the passage of a single fluorescent molecule. Background interference can arise from two sources. There may be fluorescent impurities with high luminescence quantum yields and extinction coefficients that will produce photon bursts similar to those of the analyte. This socalled bursting background can be indistinguishable from the analyte and, thus, represents a particularly serious interference. In addition to the bursting background, there is a quasi-continuum of photons that arises from higher concentrations of less luminescent materials. This nonbursting background makes it more difficult to detect single-molecule bursts, especially when the number of photons detected per burst is small. Thus, in SMD,
2270 Analytical Chemistry, Vol. 68, No. 13, July 1, 1996
S0003-2700(95)01251-0 CCC: $12.00
In luminescence-based ultrasensitive analysis, such as single-molecule detection by flow cytometry, the luminescence background from impurities present in the solvent or reagents can ultimately determine the detection limits. A simple, versatile method for reducing luminescence background is described. The method is based on photobleaching the reagent stream immediately before it enters the detection flow cell. Dramatic reduction (an order of magnitude or more) of both low-level continuous background and single-molecule fluorescence bursts is demonstrated. Application and enhancements of the technique are discussed. Several groups have developed the capability to detect and identify single fluorescent molecules as they flow or diffuse through a focused laser beam.1-12 Our approach is based upon flow cytometry, which is commonly used to analyze cells or particles.13 In flow cytometry, a hydrodynamically focused analyte stream flows past the fluorescence detector at a concentration such that only one analyte particle is in the observation region at any given time. Due to limitations in detection sensitivity, flow cytometry had been restricted to studies of particles such as cells or chromosomes containing large numbers of fluorophores. The † ‡
© 1996 American Chemical Society
both nonbursting and bursting backgrounds must be minimized. Experience has shown that, no matter how carefully the reagents and solvents are purified, the background due to fluorescent impurities sets detection limits. The extent of the background problem can be appreciated by recognizing that, in our SMD experiments, sample concentrations are typically 10-13 M. Thus, highly fluorescent, strongly absorbing impurities need to be reduced below 10-15 M to keep them from contributing more than 1% of the burst signal. Of course, less efficient or more weakly absorbing impurities at higher concentrations can contribute significant nonbursting background. Even in the most highly purified water, both bursting and nonbursting backgrounds limit detection. The problem is further exacerbated in samples that contain other components. Bioanalytical samples, in particular, include buffers, enzymes, or biological extracts. Further purification, especially in biological systems, may be either futile or undesirable, resulting in either destruction of sensitive biological molecules or the addition of new luminescent impurities. We present here a simple, versatile method for reducing both bursting and nonbursting backgrounds. The approach is based on photobleaching the impurities on-line prior to detection. It is particularly well suited for flow cytometry configurations where a small sample is introduced into a much larger flowing sheath stream or for a stationary reaction phase where the reagent is introduced into a flowing stream bathing the sample. In a sheath flow cell used for ultrasensitive fluorescence detection in capillary electrophoresis14 or SMD in a flow cytometer, most of the background can arise from the sheath stream. Thus, cleaning up the sheath by photobleaching before combining it with the sample can greatly reduce the background. The methodology is universal and works regardless of the nature of the fluorescent impurities. Emissive molecules generally have some photochemical sensitivity, and this photochemistry, even if very slight, can be used to destroy the interferent. Without prior knowledge of the nature or source of the luminescent impurities, photobleaching can decrease the luminescent background without adversely affecting nonabsorbing components of the solution. We describe below a specific implementation and possible enhancements and variations of the technique. PHOTOLYSIS CONSIDERATIONS Simple Beer’s law considerations for optically dilute samples lead to the equation
[C]out ) [C]in e-2303 QpdIotR
(1)
[C]out is the concentration of molecules that survive photobleaching; [C]in is the concentration of interfering, fluorescent impurities that enter the photolysis chamber; is the molar extinction coefficient (L mol-1 cm-1); Qpd is the quantum yield of photodestruction of C (mol ein-1); Io is the incident light irradiance (einstein cm-2 s-1), assumed to be constant across the photolysis chamber; and tR is the residence time of a molecule in the photolysis chamber. Plug flow is assumed throughout the photolysis chamber. Qpd and are intrinsic properties of the unknown impurity, while Io and tR are subject to experimental control and their product should be maximized in order to (14) Cheng, Y. F.; Wu, S.; Chen, D. Y.; Dovichi, N. J. Anal. Chem. 1990, 62, 496-503.
minimize interferents. These considerations led to the design of the on-line, axially irradiated photolysis chamber shown in Figure 1. Because of the exponential dependence on the survival probability with IotR (eq 1), even relatively small increases in IotR can dramatically enhance the photolysis efficiency. Therefore, design considerations should be directed toward maximizing this quantity. Clearly, for a given molecule with a specific Qpd, IotR can be increased with longer path lengths and higher laser power. Our current long path length design destroys effectively even relatively photostable molecules using modest photolysis powers. A transverse photolysis configuration is also possible. Generally, this gives a shorter path length, and although this allows for a more compact apparatus, a higher average photolysis power would be required. A pulsed laser was used for these experiments. In some cases, high peak powers introduce the possibility of excited state absorptions that can initiate photochemistry not accessible to single-photon absorptions. Therefore, high peak power pulsed lasers might prove more efficient at eliminating some luminescent impurities than continuous wave photolysis. However, at our peak power of 200 W, there is no evidence for multiphoton photolysis; the observed Qpd values are the same as those obtained at low powers. In our current design, we used the same laser for photolysis and detection. This may prove optimum in many cases, since it guarantees that we are decomposing fluorescent impurities that are excited by our analysis beam. However, there are two reasons why using shorter photolysis wavelengths might be desirable. First, many molecules have stronger absorptions at shorter wavelengths. Thus, using a shorter wavelength photolysis beam could enhance the photolysis efficiency. This would also be true if one were exciting on the long wavelength edge of the absorption spectrum of the interferent; then a relatively small blue shift in the photolysis wavelength would move up the absorption band to the peak of the lowest energy absorption, dramatically increasing . Second, it is common for photochemistry to increase in efficiency (Qpd) as the photon energy increases. Additionally, a much more powerful photolysis laser may be available than the one used to excite the sample. In some cases, combining several laser wavelengths may prove useful for eliminating multiple impurities with different action spectra. An important issue concerning the photolysis wavelength is the possibility of the photodestruction of a necessary reagent in the sample flow (e.g., exonuclease in our approach to DNA sequencing15-17 ). Most proteins do not absorb enough energy for photodestruction unless the excitation wavelength goes into the UV (
