Novel Fluorescence Sensing Methods for High Throughput Screening

Abstract

We describe two new methods of fluorescence sensing for use in high throughput screening (HTS). Modulation sensing transforms analyte-dependent intensity changes into a change in the low-frequency modulation signal. Polarization sensing transforms an intensity change into a change in polarization. Both methods are internally calibrated by using a reference film immediately adjacent to the sample, which can be readily located on the HTS plate or on a nearby optical component and provides an intensity or polarization reference. Modulation sensing and polarization sensing were both shown useful for measurements of fluorophore concentrations, pH, or calcium concentrations in the wells of HTS plates. Sensing with a reference film provides the opportunity to internally reference HTS measurements without the need for additions to the sample. This approach can provide standardization for assays performed at different times.

INTRODUCTION

The technology for high throughput screening (HTS) is rapidly evolving.^1–3 This development is being driven by the large number of bioactive compounds that need to be screened for a variety of biological activities, and the numerous substances obtained from combinatorial synthesis.^4,5 HTS is presently accomplished with highly automated instruments for sample manipulation and measurements. At present, fluorescence methods are the method of choice for HTS and ultra-HTS.⁶ The measurements are typically performed using fluorescence intensity, but there is a trend toward wavelength ratios, polarization, lifetimes, and correlation methods as the observable quantities. Measurements of these latter values have the advantage of being mostly independent of the total intensity, making the measurements less sensitive to changes in light intensity, concentrations, or photobleaching.

Numerous fluorescent probes are available that are specific for a variety of analytes.⁷ However, most sensing fluorophores display changes in intensity in response to the analytes, and relatively few wavelength-ratiometric probes are available. Useful wavelength-ratiometric probes are available for pH, Ca²⁺, and Mg²⁺, but the probes for Na⁺ and K⁺ display small spectral shifts that are often inadequate for quantitative measurements. Collisional quenching by species such as O₂ and Cl-occur with a change in intensity and lifetime but without an emission spectral shift. Because of the large number of sensing fluorophores that display intensity changes, it is valuable to have methods that convert these changes into ratiometric observables. Additionally, such methods may find use in HTS to provide assays that can be performed with simple instrumentation in a high throughput format.

In the past several years, a variety of new approaches to sensing have appeared. These methods typically include a fluorescent reference placed immediately adjacent to the sample, with absorption properties such that it can be excited with the same light source as used for the sample. The reference is typically a fluorophore embedded in an organic matrix, but internal references can also be used. In this method the reference film contains a luminescent metal-ligand complex (MLC) that displays a decay time on the microsecond time scale.^8–10 When the reference film is excited with amplitude-modulated light with a frequency near 1–5 MHz, its emission is usually completely demodulated as a result of the long decay time. In contrast, the modulation of a typical nanosecond-timescale fluorophore is near unity.^9,10 The modulation of the combined emission from the sample and reference is then equivalent to the fractional intensity of the nanosecond sensing fluorophore, and can be used to determine the analyte concentration.^9,10 An advantage of this approach is that the measurement is ratiometric relative to a highly stable reference signal.

Another approach to sensing is polarization sensing.^11–14 In this case the reference signal is obtained from a stretched film that contains an oriented fluorophore. The emission from such films is highly polarized even when the excitation is not polarized. This film is placed adjacent to the sample, which displays a change in intensity in response to the analyte. The sensing fluorophore only needs to display a change in intensity, not a change in the polarization of its emission. The polarization of the combined emission depends on the relative intensities of the polarized reference film and the sample intensities, and can thus be used to quantify the analyte. In the present report, we show how modulation sensing and polarization sensing can be adapted to a HTS format.

MATERIALS AND METHODS

Modulation and polarization sensing in microwell plates was accomplished as shown in Figure 1. In each case the light source was a blue light-emitting diode (LED) from Nichia Chem. Ind. (Tokushima, Japan). For modulation sensing, the LED output was amplitude modulated as described previously.^15,16 For polarization sensing, the LED output was constant. In both cases the excitation light was isolated using a 7–59 Coming® filter (Coming Incorporated, Acton, MA), which transmitted wavelengths above 400 nm with a band pass of 80 nm.

FIG. 1.

Experimental configuration for modulation (top) and polarization sensing (bottom) in high throughput screening. Modulation sensing is performed with an amplitude-modulated LED, an excitation filter (F₁), and a long-lifetime reference film (LLR). Polarization sensing is accomplished with a constant LED source, a polarized reference film (PR), and an an analyzer polarizer (P). F₂ is the emission filter.

For modulation sensing, we placed a long-lifetime reference film immediately beneath the sample well. The emission was isolated with a 3–68 Coming filter and taken to the detector of a frequency-domain fluorometer.¹⁷ For polarization sensing, a stretched oriented polarized reference film was positioned beneath the sample well. The emission was detected through an emission filter and a polarizer (Fig. 2). The theory for polarization and modulation sensing is discussed in more detail in the Appendix to this paper.

FIG. 2.

Schematic of optical system for polarization-based sensing in high throughput screening. (Left) LED = light-emitting diode; F_ex = excitation filter; PR =reference film; S = sample well; F_em = emission filter; P = polarizer. (Right) I_‖^R and ${\text{I}}_{\perp }^{\text{R}}$ = vertical and horizontal fluorescence intensity components emitted by the reference film PR; I_‖^S and ${\text{I}}_{\perp }^{\text{S}}$ = vertical and horizontal fluorescence intensity components emitted by the sample; I_‖^R = I_‖^S and ${\text{I}}_{\perp }^{\text{R}}+{\text{I}}_{\perp }^{\text{S}}$ = cumulative (total) fluorescence intensity components.

Fluorescein and 6-carboxyfluorescein (6-CF) were obtained from Aldrich (Milwaukee, WI) and fluo-3 was obtained from Molecular Probes, Inc. (Eugene, OR). Ru(bpy)₃Cl₂, Ru(bpy)₂dppz(PF₆)₂, and Ru(bpy)₂(dpb)(PF₆)₂ were synthesized in this laboratory.^* Reference films containing the MLCs were prepared as described previously.^9,10

For polarization sensing, the reference film contained pyridine 2 from Exciton, Inc. (Dayton, OH). The pyridine 2 molecules were oriented by heating the film and mechanical stretching.¹⁸ The stretch ratios R_s, is given by R_s, = N^3/2 where N is the physical fold of the stretch.

RESULTS

Modulation sensing in HTS

Modulation sensing is a relatively recent development, and it is useful to describe the principles. In frequency-domain fluorometry, the intensity decay of the sample is determined from the frequency response, which is the phase and modulation values observed over a range of light modulation frequencies.¹⁷ Figure 3 shows the simulated frequency responses for a signal produced by both a long-lifetime reference (τ_R = 1,500 ns) and a short-lifetime sensing fluorophore (τ_s = 4 ns). The frequency responses show a biphasic response that is due to the two widely different lifetimes. The decrease in modulation and modest maximum in the phase angle near 0.1 MHz is due to the long decay time. At frequencies above 0.5 MHz, this component is completely demodulated, so that it no longer contributes to the phase angle. The contribution of the reference film to the modulation is revealed by the decrease in modulation below the low-frequency starting value of 1.0. The modulation is constant until about 10 MHz, where the short lifetime now contributes to the phase and modulation values.

FIG. 3.

Simulated frequency responses for a short-lifetime sensing fluorophore (τ_S = 4 ns) and a long-lifetime reference (τ_R = 1,500 ns). The dashed vertical lines indicate the range of frequencies suitable for modulation sensing.

The frequency range of interest for modulation sensing is the constant region from 1 to 10 MHz, where the modulation is mostly independent of the frequency. As shown in the Appendix, the observed modulation (m_obs) in this frequency range is essentially equivalent to the fractional intensity of the short-lifetime component (f_s): m_obs = f_s. The short-lifetime component displays a modulation from near 1.0 up to about 10 MHz when the 4-ns decay time begins to result in demodulation. As the fractional intensity of the nanosecond decay time decreases, so does the modulation between 1 and 10 MHz. The boron idea of modulation sensing is to use a short-lifetime sensing fluorophore that changes intensity in response to the analyte. The change in intensity is detected as a change in modulation at 1–10 MHz. The analyte concentration is calculated in terms of these modulation values.

Long-lifetime reference films for modulation sensing

Modulation sensing requires a long-lifetime reference. We examined several ruthenium MLCs in polyvinyl alcohol (PVA) films. The frequency responses for three such films are shown in Figure 4. All of these films display long mean decay times near 1,600 ns. Note that the modulation of these long-lifetime films is near zero above 1 MHz. The intensity decays of [Ru(bpy)₃]²⁺ and [Ru(bpy)₂dppz]²⁺ were not quite single exponentials, as can be seen from the lack of overlap of the phase and modulation values with the best single-decay time fits (dashed lines in Fig. 4). Of the three MLCs, only [Ru(bpy)₂(dpb)]²⁺ displays a single exponential decay (lower panel). While a single decay time is not essential for modulation sensing, we feel it is preferable because a short decay time component can result in modulations greater than zero over a wide range of frequencies, possibly extending into the 1- to 10-MHz region. In principle, this is not a problem and is accounted for during the calibration. For simplicity, we selected the nearly monoexponential [Ru(bpy)₂(dpb)]²⁺ for our reference fluorophore.

FIG. 4.

Frequency-domain intensity decay of Ru-metal-ligand complexes in PVA films. The solid lines are the best two-decay time fits, and the dashed lines are the best single-decay time fits.

LEDs are now more widely used as light sources in fluorescence sensing and spectroscopy. Hence it is of interest to determine the stability of their optical output. Figure 5 shows the output intensity from the MLC reference film for excitation with our blue LED over 400 min. There was no detectable change in the intensity of the reference film. In addition to demonstrating the stability of the LED output, this result also demonstrates the stability and lack of photobleaching of the MLC reference film. The MLCs are known to be chemically and photochemically stable. We have kept solutions of some MLCs in room light for years without any significant changes in intensity or decay time. We feel it should be possible to use the same film as an intensity reference for reading large numbers of plates, or the film could be incorporated into the design of the microwell plate as an intensity reference.

FIG. 5.

Stability test for the long-lifetime reference film containing [Ru(bpy)₂dpb]²⁺ with LED excitation.

Examples of modulation sensing.

As an initial example of modulation sensing, we performed a simple concentration assay using fluorescein. The emission spectra (Fig. 6, inset) reveal the contribution of fluorescein and the MLC film to the total emission seen through the long-pass filter (dashed line). Increasing concentrations of fluorescein result in increasing modulation values at 2 MHz. The increase in modulation is hyperbolic and not linear because of the constraint on f_S + f_R = 1.0. The useful range of concentrations spans an approximate 50-fold range. The concentration range could be extended by using MLC films of various intensities, or by choosing the observation wavelength to relatively select more or less of the reference emission. For instance, the modulation volume could be made more sensitive to the fluorescence by using a 520-nm observation wavelength, and less sensitive by using a larger observation wavelength.

FIG. 6.

Dependence of the low-frequency modulation at 2 MHz on the fluorescein concentration. The inset shows the emission spectra from the fluorescein and the reference film. The dashed line shows the transmission of the emission filter.

Measurements of pH are among the most common analytical procedures. The usefulness of modulation sensing for PH was demonstrated using 6-CF. The intensity of 6-CF increases dramatically as one of the carboxyls is deprotonated near pH 6.5.¹⁹ Figure 7 shows the pH-dependent modulation using 6-CF. The modulation displays the usual sigmoidal dependence with a midpoint near pH 6. For a typical accuracy in the measured modulation near ± 0.007, one can expect a pH accuracy, at the center of the titration curve, near ± 0.05. This accuracy is inferior to that obtained with a glass electrode, and the pH range is limited. However, one can imagine situations where this accuracy is adequate and an optical measurement is preferred. In particular, biological assays are typically performed using a narrow range of pH values, and fluorescence detection is preferred in HTS. Also, one could use pH fluorophores such as SNAFL® or SNARFL® (Molecular Probes, Inc., Eugene, OR), which displays spectral shift in response to pH.²⁰ In this case, changes in the observation wavelength may allow the measurable range of pH to be increased.

FIG. 7.

Modulation sensing of pH using 6-carboxyfluorescein.

As a final example of modulation sensing, we developed an assay for calcium based on fluo-3. This fluorophore is essentially nonfluorescent in the absence of Ca²⁺; its intensity increases about 100-fold on binding calcium²⁰ (Fig. 8, inset). This large change in intensity results in a dramatic change in the 2-MHz modulation, from 0.07 to 0.85 (Fig. 8). One can readily imagine such assays being used to study signaling and activation in cell cultures.

FIG. 8.

Calcium sensing using fluo-3 and the Ru-MLC reference film. The inset shows the emission spectra of the fluo-3 and reference film.

Polarization sensing in HTS

Polarization sensing is a recent development in fluorescence sensing technology. The theory for polarization sensing is relatively simple, and has been described in detail.^11–14 However, because this is a new concept that is not widely understood, we present a more detailed theory description in the Appendix. The basic idea is to mix highly polarized fluorescence of the reference with the nonpolarized or orthogonally polarized sample fluorescence. The resulting polarization of combined reference and sample emission is directly proportional to the sample intensity and can be used to quantify the analyte concentration. As shown in the Appendix, there are two principal factors affecting the resolution of polarization sensing experiments. First, the available range of polarization change depends on the polarization from the reference emission. Hence, it is important to use a highly polarized reference such as highly oriented polymer film.¹⁸ Second, one must choose the initial conditions for the ratio of sample to reference fluorescence intensities.

Highly polarized reference film for polarization sensing.

For randomly oriented fluorophores with linearly polarized excitation, it is well known that the maximum polarization is 0.5²¹ This limit is due to the orientation photoselection that occurs in optical absorption. Less well known is the high polarization values that can be achieved with oriented samples.¹⁸ Such values are shown in Figure 9 for a PVA film containing pyridine 2. This elongated fluorophore becomes oriented along the PVA stretch axis, resulting in a dramatic increase in polarization to about 0.85. Importantly, these high polarization values can be obtained even with an unpolarized excitation source, in this case the blue LED, without an excitation polarizer. This occurs because the electronic transition moments of pyridine 2 molecules are aligned, and the emission is due to this oriented population. This property of the stretched reference film makes it useful even with an unpolarized excitation source, simplifying the instrumentation. Additionally, it is known that such films are stable and retain their orientation for extended periods of time.

FIG. 9.

Effect of the stretching ratio (R_S) on the emission polarization of pyridine 2 in a PVA film.

Examples of polarization sensing.

Polarization sensing can be readily adapted for measuring the concentration or intensities of a fluorophore. This use is shown in Figure 10 for increasing concentrations of fluorescein. At low fluorescein concentrations, the polarization is high because most of the emission originates with the oriented film. As the fluorescein concentration increases, the polarization decreases. In this case the fluorescein is free in solution and its polarization is near zero. One can readily imagine the use of such an assay for immunoassays in which the presence of antigen results in an increased intensity from the sample well. In such assays the fluorophore is typically bound to a macromolecule, resulting in polarized emission. It is important to stress that, for unpolarized (isotropic) excitation, the polarization of the sample observed along the excitation beam propagation is zero independently of the sample polarization properties observed with right-angle geometry.

FIG. 10.

Effect of fluorescein concentration in a polarization sensor.

pH measurements are often used in HTS assays. A polarization pH assay is shown in Figure 11. As the pH increases, the fluorescein intensity increases and the anisotropy decreases. Similarly, calcium can be measured in the microwell plates using fluo-3 (Figure 12), which again shows decreasing polarization with increasing calcium concentrations.

FIG. 11.

Polarization sensing of pH using 6-carboxyfluorescein.

FIG. 12.

Polarization sensing of calcium using fluo-3.

DISCUSSION

When first introduced, most fluorescence plate readers provided only intensity measurements. However, it is now possible to perform the measurements typical of advanced fluorescence spectroscopy in the HTS format. Wavelength-ratiometric and polarization methods are becoming routine, and it seems likely that nanosecond time-resolved measurements will soon be practiced. A wide variety of probes are now known that display changes in lifetime in response to electrolytes or biochemical interactions.^22–26 Such measurements previously required sophisticated instrumentation, but can now be accomplished with solid-state light sources and simple electronics.^{15,16,26–29} Because of these advances in opto-electronic technology, the full power of time-resolved fluorescence and novel fluorescence sensing methods will be utilized in a HTS format for both basic research and drug discovery. However, there are still many circumstances when accurate intensity measurements are the basis for quantization. The methods of modulation sensing and polarization sensing may provide simple and accurate intensity-based assays that are transferrable between microwell plates and/or between HTS instruments based on the stability of MLC-labeled and stretched reference films. One can also imagine modulation sensing in which the reference is provided by labeled beads dropped into each well in the plates.

APPENDIX

Theoretical description of modulation sensing

The principle of modulation sensing is easily derived from the theory for frequency-domain decay time measurements.^29,30 In this method, the sample is excited with light that is amplitude modulated at a circular frequency ω in radians/s. The emission from the sample is then delayed in phase by an angle ϕ and demodulated by a factor m according to

$$\text{tan}\varphi =\omega \tau$$ (1)

$$m={\left(1+{\omega }^2{\tau }^2\right)}^{-1/2}$$ (2)

where τ is the decay time of the sample. These expressions are strictly valid for single exponential decays. More complex expressions are needed for multiexponential decays, but these expressions are not needed for the present analysis.

In modulation sensing, one typically uses a sensing fluorophore with a nanosecond decay time (τ_S) and a reference fluorophore with a microsecond decay time (τ_R). The observed modulation of the emission from the sensing and reference fluorophores is given by

$$m_{obs}\text{= }{f}_S{m}_S+f_R{m}_R$$ (3)

where f_i is the fractional intensity of each fluorophore to the combined emission from the sample and reference, with f_S + f_R = 1.0. For decay times near 1 ns and 1 μs, and modulation frequencies near 1 MHz, m_S ≃ 1.0 and m_R ≃ 0.0. Under these conditions the observed modulation becomes equal to the fractional intensity of the nanosecond-decay-time fluorophore

$$m_{obs}\text{= }{f}_S$$ (4)

The modulation measurement is ratiometric relative to the long-lifetime reference film.

To obtain the maximum change in modulation, it is important to select the initial conditions. The primary factor is the relative intensity change of the sensing fluorophore in response to the analyte, and the initial fractional intensities of the reference (f_R⁰) and sensing (f_S⁰) fluorophores. Suppose the intensity of the sensing fluorophore decreases n-fold when saturated with analyte. The fractional intensity of this fluorophore then decreases from f_S⁰ to

$$f_{\text{S}}=\frac{f_S^0/n}{f_S^0/n+f_R^0}=\frac{f_S^0}{f_S^0+n{f}_R^0}$$ (5)

The change in modulation Δf_S is given by

$$\Delta f_S=f_S^0-f_{\text{S}}$$ (6)

Simulated values of Δf_S are shown in Figure 13 for various values of $f_{\text{S}}^0$ and n. If the sensing fluorophore displays only a modest change in intensity (n ≤ 2), then it is preferable to start with roughly equal initial intensities for the sample and reference. If the sensing fluorophore displays a large decrease in intensity (n = 25–100), then it is preferable to begin the assay with a large fractional intensity of the sensing fluorophore. Such considerations can be used to maximize the dynamic range of a modulation assay.

FIG. 13.

Change in fractional intensity (Δf_S) expected for different intensity ratios (n) of the sensing fluorophore in the absence and presence of analyte.

Theoretical description of polarization sensing

Polarization sensing is based on the additivity property of parallel and orthogonal fluorescence intensity components, I_‖ and I_⊥ The optical geometry used for HTS polarization sensing is shown in Figure 2. The fluorescence polarization is defined by

$$P=\frac{I_{\Vert }^T-I_{\perp }^T}{I_{\Vert }^T+I_{\perp }^T}$$ (7)

where the superscript T indicate total sum of intensities from the sample and reference for parallel (‖) and orthogonal (⊥) components, respectively. The total parallel intensity is given by

$$I_{\Vert }^T=I_{\Vert }^R+I_{\Vert }^S$$ (8)

and the total orthogonal intensity is given by

$$I_{\perp }^T=I_{\perp }^R+I_{\perp }^S$$ (9)

The polarization is given by

$$P=\frac{I_{\Vert }^R+I_{\Vert }^S-I_{\perp }^R-I_{\perp }^S}{I_{\Vert }^R+I_{\Vert }^S+I_{\perp }^R+I_{\perp }^S}$$ (10)

When the emission intensity from the sample is zero, the polarization is determined by the reference film,

$$P_0=\frac{I_{\Vert }^R-I_{\perp }^R}{I_{\Vert }^R+I_{\perp }^R}$$ (11)

If the emission intensity from the sample dominates, the polarization tends toward that from the sample, which is near zero. The polarization of the reference is typically high, above 0.8, because of the strongly oriented fluorophores in a stretched film independent of light polarization. This allows convenient use of an unpolarized excitation beam. The use of an in-line geometry and unpolarized excitation makes measurements independent from internal change of sample polarization. Suppose the sensing fluorophore is only weakly fluorescent in the absence of analyte, and its intensity increases in the presence of analyte. In the absence of analyte, the sensing fluorophore is only weakly fluorescent, and the polarization of the combined emission is near that of the reference film. As the analyte concentration increases, the intensity of the sensing fluorophore increases and the polarization of the combined emission decreases.

In order to obtain the maximal change in polarization, one should consider the initial conditions for the assay. For unpolarized excitation and unpolarized emission from the sensing fluorophore, $I_{\Vert }{}^S=I_{\perp }^S=I^S$. The polarization of the emission is thus given by

$$P=\frac{I_{\Vert }{}^R-I_{\perp }^R}{I_{\perp }^R+{\text{I}}_{\Vert }{}^R+2{\text{I}}^{\mathrm{S}}}$$ (12)

For the limit I^S → 0, then P → P₀, which is the polarization of the reference film itself. For a high intensity from the sensing fluorophore, I^S → ∞ and P → 0.

For purposes of simulations, we define

$$k_0=\frac{I_{\perp }^S}{I_{\perp }^R}=\frac{I^S}{I_{\perp }^R}$$ (13)

as the initial values in the absence of analyte. Suppose the intensity of the sensing fluorophore decreases n times in response to analyte. Then the change in polarization is given by

$$\Delta P=P_0-P$$ (14)

where

$$P_0=\frac{D_0-1}{D_0+1+2k_0}$$ (15)

$$P=\frac{D_0-1}{D_0+1+2k_0/n}$$ (16)

and

$$D_0=I_{\Vert }{}^{\text{R}}/I_{\perp }^R$$ (17)

is the parallel and orthogonal ratio of intensities for the oriented film only.

Simulations of the expected change in polarization ΔP are shown in Figure 14. Overall, the maximal changes in polarization depend on the ratio of initial intensities of the sample $(I_{\perp }^S)$ and reference $(I_{\perp }^R)$, k₀. Suppose this initial ratio is 1.0. Then a decrease in the intensity of the sensing fluorophore results in an increase in the polarization (ΔP < 0). Conversely, an increase in the sensing fluorophore intensity results in a decrease in the polarization (ΔP > 0). One can use Figure 14 to select the initial conditions for a polarization assay so as to obtain the maximum changes in the polarization. If the intensity of the sensing fluorophore decreases (n = 10 or 20), then it is preferable to start the assay with higher intensities from the sensing fluorophore (k₀ ≃ 4). If the sensing fluorophore increases in intensity (n = 0.1), then the assay should be started with a smaller intensity from the sensing fluorophore (k₀ ≃ 0.2).

FIG. 14.

Change in polarization expected for different values of the initial intensity ratio $k_0=I^S/I_{\perp }^R$ is dependent on the intensity ratio prior to addition of the analyte and the intensity ratio of the sensing fluorophore in the absence and presence of analyte (n). That is, the analyte causes an n-fold decrease in intensity of the sensing fluorophore.