1310
OPTICS LETTERS / Vol. 31, No. 9 / May 1, 2006
Prism-based multicolor fluorescence correlation spectrometer Ling Chin Hwang Department of Chemistry, National University of Singapore, 3 Science Drive, Singapore 117543, Singapore, and Ecole Polytechnique Fédérale de Lausanne, Laboratoire d’Optique Biomédicale, CH-1015 Lausanne, Switzerland
Marcel Leutenegger, Michael Gösch, and Theo Lasser Ecole Polytechnique Fédérale de Lausanne, Laboratoire d’Optique Biomédicale, CH-1015 Lausanne, Switzerland
Per Rigler and Wolfgang Meier Physikalische Chemie, Universität Basel, Klingelbergstrasse 80, 4056 Basel, Switzerland
Thorsten Wohland Department of Chemistry, National University of Singapore, 3 Science Drive, Singapore 117543, Singapore Received November 29, 2005; accepted January 18, 2006; posted January 30, 2006 (Doc. ID 66314) We report the design and application of a prism-based detection system for fluorescence (cross) correlation spectroscopy. The system utilizes a single laser wavelength for the simultaneous excitation of several dyes of different emission spectra. Fluorescence light is spectrally separated with a prismatic setup, and wavelengths are selected by scanning a fiber-coupled avalanche photodiode across the image spots. Multicolor autocorrelations are demonstrated with standard and tandem dyes, and fluorescence cross-correlation measurements of biotinylated nanocontainers and streptavidin are presented. This spectrometer offers high optical stability and no focal volume mismatch for the multicolor detection of molecular dynamics and interactions, with single-molecule sensitivity. © 2006 Optical Society of America OCIS codes: 170.0170, 300.6280, 230.5480.
Fluorescence correlation and fluorescence crosscorrelation spectroscopy (FCS and FCCS, respectively) are widely used tools for the study of molecular dynamics and interactions. These techniques use the fluctuations in fluorescence signals caused by random diffusion of fluorescent particles through a laser-illuminated confocal volume to measure the occurrences and characteristic times of molecular processes.1 Because of their single-molecular sensitivity FCS and FCCS have found various applications in chemical and biological studies.2 Multicolor fluorescence detection has been demonstrated in confocal microscopy and extended to FCS and FCCS measurements to study the mutual interactions of biomolecules.3,4 The excitation of several fluorophores in a multicolor setup for FCS and FCCS in general requires several laser beams matched to the same focal spot, or the combination of a high-frequency pulsed IR laser and dyes with similar two-photon excitation spectra. This complexity has been addressed by using fluorophores with overlapping one-photon excitation spectra but different emission wavelengths. Using only a single laser wavelength for excitation, autocorrelation and cross-correlation measurements have been performed with as many as three differently labeled molecules to study their mutual interactions.5–7 With more excitation lines applied for investigation of complex biomolecular systems, individual specified sets of dichroic mirrors and bandpass filters have to 0146-9592/06/091310-3/$15.00
be aligned for each range of wavelengths detected. This not only complicates the optical setup but also contributes to higher transmission losses through each optical component. In this work we simplify the detection path of the FCS instrument by using a dispersive prism to spectrally separate the emission light. The dispersion causes a wavelength-dependent deflection angle such that the FCS signal can be focused in well-separated spots for the spectral ranges of interest. Scanning an optical fiber through these foci selects different spectral ranges for detection. To prove this concept we show that FCS and FCCS can be performed with a single laser wavelength and prism-based detection. A schematic diagram of the prism-based FCS setup is shown in Fig. 1. An argon-ion laser 共 = 488 nm兲 is used for the excitation of several fluorescent dyes. The laser beam diameter is expanded and coupled into the back aperture of the objective (Olympus 40⫻, NA 1.15). The fluorescence emission from the sample is collected by the objective and separated from the excitation and backscattered light with a dichroic mirror. The fluorescence light is focused by the microscope tube lens into a 50 m pinhole. An achromat 共f = 100 mm兲 collimates the emission light, which then passes a 30° isosceles prism (N-BK7, index of refraction nd = 1.5168, Abbe number d = 64.1673; for Sellmeier coefficients see Ref. 8) dispersing the fluorescence light. For a reference wavelength of 580 nm the prism is tilted about the optical axis by 8.1° with © 2006 Optical Society of America
May 1, 2006 / Vol. 31, No. 9 / OPTICS LETTERS
Fig. 1. (Color online) Optical setup of prism-based fluorescence correlation spectrometer. A single laser wavelength excites differently emitting fluorophores. The emitted light is collimated and chromatically dispersed by an isosceles prism and focused onto an array of optical fibers that is coupled to avalanche photodiodes. The inset shows (a) the normalized emission spectra of Rhodamine green (RhG), R-phycoerythin (RPE) and Alexa 647-R-phycoerythrinstreptavidin (AXSA); (b)–(d) the colored autocorrelation data curves recorded by scanning a 100 m optical fiber along the imaged spots from green, yellow, to red wavelengths, respectively. The black curves show the experimental fits. F, excitation filter; Obj, microscope objective, L1–L5, lenses; DM, dichroic mirror; PH, pinhole; P, dispersive prism; OF, optical fibers; APD1–3, avalanche photodiodes.
the incident beam striking at 23.1° to the prism surface and exiting symmetrically at a minimum deviation angle of 16.2° for optimum transmission. The focusing lens (achromat f = 120 mm) brings the wavelengths into focus at different positions in the focal plane. A 1.2 times magnified image of the pinhole is formed for each wavelength and distributed along the image plane. The desired wavelength range to be detected is defined by the core diameter of the optical fiber positioned at the image plane. An increase in focal length of the focusing lens narrows the width of the detected wavelength ranges and increases the spectral channel separation. To first establish that we can perform FCS with a prism as a dispersive element, we scanned a single 100 m multimode optical fiber along the image plane. The beam diameter and focal length of the L5 lens resulted in a NA of 0.028, much smaller than the fiber NA of 0.22, which assured optimal coupling efficiency of the fiber. Because of the nonlinear dispersion of the prism the lateral spectral shifts are not linear. Figure 2(a) illustrates the lateral displacement y共兲 of different wavelengths with respect to the reference wavelength of 580 nm. We used an avalanche photodiode for detecting single photons. The detector signal was autocorrelated by a hardware correlator. The system was calibrated with a standard fluorophore, Rhodamine Green (RhG) and a tandem dye Alexa-647 R-phycoerythrin-streptavidin (AXSA) having two emission peaks at 575 and 667 nm (the emission spectra shown in the inset in Fig. 1). Using optimum laser powers of 600 W for the RhG and 50 W for the AXSA, we align the fiber for the best count rates in the green, yellow, and red channels, respectively. From the fits of the autocorre-
1311
lation curves, the ratio of axial to radial dimensions of the confocal volume K was between 2 and 4. We measured a diffusion time of 134.5 s for the RhG. Using the diffusion coefficient of Rhodamine of 2.8 ⫻ 10−6 cm2 / s, the beam cross-section radius and the effective confocal volume were determined to be 390 nm and 0.89 fl, respectively. The count rate per molecule (cpm) calculated from the autocorrelation curves, 26.5 kHz (RhG), 9.5 kHz (the AXSA in the yellow channel), and 47.2 kHz (the AXSA in the red channel), was corrected for laser background in the blue region and Raman scattering of water in the yellow region. The cpm for the RhG was reduced by approximately 60% as compared with that for typical FCS. This is due to the narrower spectral bandwidth of the green channel (19 nm as calculated below) compared with an emission bandpass filter with a spectral range of 40 nm. An optical fiber array was constructed consisting of an optic fiber array holder with grooves to fix 3 m ⫻ 105 m core diameter bare fibers with 250 m cladding and buffer diameter. The centers of the cores were separated by 250 m when clamped next to each other. The detection efficiency of a fiber for monochromatic light with wavelength is given by the overlap integral of the wavelength-dependent image with the fiber core. To calculate the transmission T共兲 of a fiber over the full spectral range this integral is computed for all . The prism dispersion leads to a lateral shift y共兲 of the image with the wavelength. A lateral displacement of the fiber allows the selection of different wavelength ranges, whereas the bandwidth is proportional to the core diameter. Figure 2(b) shows the calculated spectral bandwidths 19, 29, and 45 nm at FWHM from the blue, yellow, and red wavelength regions, respectively. The steepness of the spectral filtering of our spectrometer is measured between the 10% and 90% transmission values and was 4 – 11 nm for the green to red bandwidths, as compared with commercial bandpass filters and dichroic mirrors of 5 and 15– 25 nm, respectively. Here, to show the simultaneous autocorrelations and cross correlations of two binding components, we clamped
Fig. 2. (Color online) (a) Plot of lateral displacement y共兲 against wavelength caused by prism dispersion. The nonlinear dispersion depends on the glass properties of the prism. (b) Plot of transmission T共兲 versus wavelength shows the laser line 488 nm and the spectral bandwidths calculated for optic fibers with core diameters of 105 m with cladding and buffer diameters of 250 m.
1312
OPTICS LETTERS / Vol. 31, No. 9 / May 1, 2006
Fig. 3. (Color online) Cross-correlation data (gray curves) and their fitting curves (black curves) between green and red channels and their intensity traces. (a) Crosscorrelation curves show binding of biotinylated RhG nanocontainers and AXSA. Inset, schematic drawing of the dual-color complex. Red intensity spikes show oligomerization of AXSA to the nanocontainers. (b) Negative control shows neither cross correlation nor intensity spikes from binding. Correlations were measured at 60 s each for 20⫻ with a laser line of 488 nm at power 50 W.
the fibers at alternate grooves and mounted the holder on an x, y, z translation stage that moves in micrometer steps. Calibration measurements were again repeated with the same dyes. The green channel was aligned to an optimum cpm for the RhG of 21 kHz. The red channel had a cpm of 9 kHz for the AXSA. For cross-correlation measurements, RhG-filled, 5% biotinylated nanocontainers9 were measured with the AXSA to detect biotin–streptavidin complexes. The nanocontainers were prepared by dispersing amphiphilic triblock copolymers in an aqueous solution. An amphiphilic triblock copolymer (PMOXA-PDMSPMOXA, JW 05), having an average molecular weight of approximately 8000 g / mol, was synthesized by an established procedure.10 As a negative control, all the AXSA biotin-binding sites were first saturated with excess unlabeled biotin before adding the RhG nanocontainers. For both experiments a sample droplet of RhG nanocontainers was first measured to give an autocorrelation curve in the green channel. Since there was negligible cross talk from the RhG to the red channel, there neither was an autocorrelation found in the red channel nor were cross correlations observed between the channels. After adding the AXSA to the sample droplet, oligomers started forming between the nanocontainers and the AXSA, as was evident from the intensity spikes and the rising cross-correlation amplitude (Fig. 3). The negative control did not show any cross correlations or intensity spikes despite autocorrelation curves in both the green and red channels. We have successfully shown a prism-based fluorescence correlation spectrometer by using a single laser wavelength for the simultaneous excitation of three
dyes. The spectrometer is capable of selecting emission wavelengths without the use of dichroic mirrors and emission filters. By aligning a fiber optic array along the focal plane, we have simultaneously performed dual-color autocorrelations and cross correlations of two binding components. To allow the optimization of detection ranges for each fiber individually, the fibers could be fixed separately and aligned independently. An alternative is to use a Silicon photodiode array of up to 76 elements (Hamamatsu, Japan) to detect the whole spectrum. A grating-based setup has been shown to work with FCS; however, detection efficiency decreases mainly because of losses to higher diffraction orders.11 The prism-based spectrometer presented here achieved a cpm more than double that attained for fluorescent dyes and quantum dots measured on the grating setup. In addition, we present here for what we believe to be the first time experiments showing the possibility of performing FCCS with a setup that uses a dispersive element in the detection pathway. With an increasing number of fluorophores, namely, tandem dyes and quantum dots that can be excited with a single laser wavelength and emit at separate wavelengths, the prismbased fluorescence correlation spectrometer is a promising tool to investigate molecular dynamics and binding processes in multicolor systems. The authors thank Antonio Lopez for his assistance in constructing the optic fiber array holder. This research was funded by the Swiss National Science Foundation, the National Center for Competence in Nanoscale Science, Ecole Polytechnique Fédérale de Lausanne, and the National University of Singapore. T. Wohland’s e-mail address is
[email protected]. References 1. D. Magde, E. L. Elson, and W. W. Webb, Phys. Rev. Lett. 29, 705 (1972). 2. N. L. Thompson, A. M. Lieto, and N. W. Allen, Curr. Opin. Struct. Biol. 12, 634 (2002). 3. T. D. Lacoste, X. Michalet, F. Pinaud, D. S. Chemla, A. P. Alivisatos, and S. Weiss, Proc. Natl. Acad. Sci. U.S.A. 97, 9461 (2000). 4. P. Schwille, F. Meyer-Almes, and R. Rigler, Biophys. J. 72, 1878 (1997). 5. L. C. Hwang and T. Wohland, ChemPhysChem 5, 549 (2004). 6. L. C. Hwang and T. Wohland, J. Chem. Phys. 122, 114708 (2005). 7. L. C. Hwang, M. Gösch, T. Lasser, and T. Wohland, have submitted a paper called “Simultaneous multicolor fluorescence cross-correlation spectroscopy to detect, higher order interactions using single wavelength laser excitation,” to Biophys. J. 8. SCHOTT, http://www.schott.com/optics_devices/ english/products/flash/abbediagramm_flash.htm. 9. P. Rigler and M. Meier, J. Am. Chem. Soc. 128, 367 (2006). 10. C. Nardin, T. Hirt, J. Leukel, and W. Meier, Langmuir 16, 1035 (2000). 11. M. Burkhardt, K. G. Heinze, and P. Schwille, Opt. Lett. 30, 2266 (2005).
