- Email: [email protected]
A data treatment method for detecting fluorescence anisotropy peaks in capillary electropherograms
Analytica Chimica Acta 739 (2012) 99–103
Contents lists available at SciVerse ScienceDirect
Analytica Chimica Acta journal homepage: www.elsevier.com/locate/aca
A data treatment method for detecting fluorescence anisotropy peaks in capillary electropherograms Ryan A. Picou a , Indu Kheterpal b , S. Douglass Gilman a,∗ a b
Department of Chemistry, Louisiana State University, Baton Rouge, LA 70803, USA Pennington Biomedical Research Center, LSU System, Baton Rouge, LA 70808, USA
h i g h l i g h t s
g r a p h i c a l
a b s t r a c t
We explain the absence of fluorescence anisotropy peaks for capillary separations. We developed a method to visualize fluorescence anisotropy peaks for separations. This method was applied to a separation of amyloid beta peptide aggregates.
a r t i c l e
i n f o
Article history: Received 18 April 2012 Accepted 12 June 2012 Available online 19 June 2012 Keywords: Fluorescence anisotropy Data treatment Capillary electrophoresis Amyloid beta peptide
a b s t r a c t A data treatment method is presented to detect fluorescence anisotropy (FA) peaks in capillary electrophoresis electropherograms. The data treatment method converts plots of fluorescence anisotropy vs. time that contain no peaks that are distinguishable from the noise of the anisotropy background into plots that show distinct fluorescence anisotropy peaks. The method was demonstrated using laserinduced fluorescence anisotropy data from individual A (1–42) aggregates separated using capillary electrophoresis. Applying this data treatment method enabled the detection of anisotropy peaks for individual A aggregate fluorescence peaks that were not observed prior to the data treatment method. The data treatment method is not specifically designed for A aggregate analysis or capillary electrophoresis, and it should be applicable to other applications and other separation methods with FA detection. © 2012 Elsevier B.V. All rights reserved.
1. Introduction Fluorescence anisotropy (FA) is a measure of the degree of emission depolarization of a fluorescent molecule after being excited with a polarized excitation source [1]. A common cause of emission depolarization is molecular rotational diffusion during the excited state lifetime of the fluorophore [1]. In general, molecules with smaller rotational diffusion rates, i.e. slower rotation in solution, exhibit larger FA. Similarly, molecules that have large rotational diffusion rates exhibit smaller FA.
∗ Corresponding author. E-mail address: [email protected] (S.D. Gilman). 0003-2670/$ – see front matter © 2012 Elsevier B.V. All rights reserved. http://dx.doi.org/10.1016/j.aca.2012.06.023
Laser-induced fluorescence anisotropy (LIFA) has been demonstrated as an on-column detection technique for capillary electrophoresis (CE) for studies of biomolecular interactions, e.g. protein–protein or oligonucleotide–protein interactions [2–7]. Picou et al. recently studied the separation and detection of individual A amyloid aggregates by CE-LIF using ThT in the running buffer as the fluorescent probe [8]. It was hypothesized that differences in fluorescence anisotropy between aggregates could indicate relative size differences between separated aggregates. Plots of FA vs. time showed that FA due to aggregates and the ThT background [9] were indistinguishable; i.e. no fluorescence anisotropy peaks were observed although there were obvious peaks in plots of fluorescence vs. time [8]. In order to evaluate if CE-LIFA can be used to differentiate individually detected A
100
R.A. Picou et al. / Analytica Chimica Acta 739 (2012) 99–103
aggregates, it is important that the FA of the aggregates be easily distinguishable from the background anisotropy. Most CE-LIFA studies in the literature did not include plots of fluorescence anisotropy or polarization vs. time [2,3,6,7,10,11]. This suggests that other studies using FA for separations detection may have faced the same issues that we encountered when plotting FA vs. time, i.e. high FA noise and no observed FA peaks despite low background fluorescence and peaks with high S/N in the corresponding plots of fluorescence vs. time. In order to understand why our anisotropy plots did not result in distinct peaks as demonstrated in the literature [4,5] and to better visualize changes in FA vs. time that are evident in the raw fluorescence data [8], we have developed a simple data treatment method. This method enables differentiation of anisotropies due to fluorescent peaks (peptide aggregates here) and background signals. This data treatment method effectively extracts anisotropy peaks from FA electropherograms that were originally embedded in the background FA noise. This method allows simple visualization of anisotropy peaks vs. time.
Fig. 1. Bottom two plots are electropherograms with LIF detection showing an individual A (1–42) aggregate peak at 120.65 s (left y-axis). The dashed plot and dark solid plot are the parallel and perpendicular fluorescence emissions, respectively. The uppermost plot (lighter solid) is the fluorescence anisotropy, r (right y-axis), calculated from the parallel and perpendicular emissions using Eq. (1).
2. Experimental 2.1. Chemicals and peptide samples The chemicals, solutions and sample preparations used in this study have been described in detail previously [8]. Tris (hydroxymethyl) aminomethane (Tris) was purchased from Fisher Scientific (Fair Lawn, NJ). Thioflavin T (ThT) was purchased from Sigma (St. Louis, MO). The electropherograms presented in this work were performed using 10.00 mM Tris buffer at pH 7.79. Thioflavin T (15 M) was included in the running buffer to detect A (1–42) aggregates [8,12]. The A (1–42) peptide samples were prepared as previously described [13,14]. The monomer-equivalent concentration of the A (1–42) was 5 M as determined by HPLC-UV [15]. Analysis of samples by TEM, CE-UV and CE-LIF confirmed that they contained A (1–42) aggregates [8].
(PMT) (H9306-04, Hamamatsu; Bridgewater, NJ). The outputs of the PMTs were filtered by 500 Hz low pass RC filters and collected by a National Instrument PCI-6024E DAQ board at a scan rate of 1000 Hz. The CE-LIFA data collection was controlled by a program written in LabView 5.0, and the data were analyzed using Origin Pro 7.5 and Microsoft Excel 2007. Eq. (1) was used to calculate the fluorescence anisotropy, r, where I|| , and I⊥ , are the measured parallel and perpendicular fluorescence intensities (relative to the polarization of the excitation source). The factor G corrects the anisotropy for the differences in sensitivities of the two PMTs. For this work, G was empirically adjusted to 1 by optimizing the potentials applied to both PMTs so that the responses and sensitivities were equal for a small fluorescent molecule, Lucifer Yellow (Molecular Probes, Eugene, OR) over a range of concentrations (data not shown). The applied potentials for PMT|| and PMT⊥ were 1000 and 935 V, respectively.
2.2. Capillary electrophoresis r= Capillary electrophoresis was performed in a fused-silica capillary (ID = 50 m, OD 366 m; Polymicro Technologies, Phoenix, AZ) cut to 60.0 cm total length with a window created at 23.0 cm from the inlet. The electrophoretic potential was applied across the capillary using a Spellman CZE1000R high-voltage power supply (Hauppauge, NY) at 25.0 kV (417 V/cm), which produced an electrophoretic current of 5–6 A. Sample injections were performed for 5.0 s at 25.0 kV. Prior to each injection, the sample was vortexed to suspend any aggregates that might have settled to the bottom of the sample vial. 2.3. Laser-induced fluorescence anisotropy detection system The FA detection system used here has been described in detail previously [8]. This detector is based on the design reported by Whelan et al. [4] and has been modified for the detection of narrow peaks (3–15 ms) due to individual aggregates. A schematic of the detection system is shown in Supplementary material (Fig. S1). The beamsplitter cube (10FC16PB.3, Newport Corporation; Irvine, CA) separates the fluorescence emission into parallel and perpendicular components relative to the polarization of the 445 nm diode laser excitation source (Model no. LDCU12/7532, Power Technology, Inc; Alexander, AR). Both emission components were spectrally filtered by a 490 nm bandpass filter and 440 nm notch filter. Then the two emission components were detected simultaneously and equidistant from the capillary by two identical photomultiplier tubes
I|| − GI⊥ I|| + 2GI⊥
(1)
3. Results and discussion The aggregation of A peptides has been linked to Alzheimer’s disease. The data presented here are from a capillary electrophoretic separation of a mixture of A (1–42) peptide aggregates. The peptides were non-covalently labeled with the amyloid dye, thioflavin T (ThT). The peaks in the electropherograms presented here are due to individual A (1–42) aggregates [8]. Fig. 1 presents a representative section from electropherograms for samples containing A (1–42) aggregates. The individual parallel and perpendicular fluorescence signals (I|| , and I⊥ ) are plotted in the lower part of the figure (left y-axis), and the FA calculated from these signals (Eq. (1)) is plotted in the top part of the figure (right y-axis). The fluorescence peak in Fig. 1 is due to an individual A (1–42) aggregate with ThT noncovalently bound to it [8]. The binding of ThT to A causes excitation and emission spectral shifts and enhancement at the wavelengths being used for excitation and detection (440 and 490 nm) [16,17]. Fig. 1 shows only 1 of the 178 aggregate peaks that were detected for this injection of A (1–42) aggregates. Fig. 1 shows that the fluorescence anisotropies due to A aggregates and due to background fluorescence of ThT [9] are indistinguishable, i.e. no FA peaks are apparent in the plot of FA vs. time although fluorescence peaks for I|| , and I⊥ and their different intensities are obvious. The slight difference in the
R.A. Picou et al. / Analytica Chimica Acta 739 (2012) 99–103
101
Fig. 2. (A) Electropherograms with LIF detection showing peaks due to two individual A (1–42) aggregates at 121.84 s and 122.02 s. The dashed plot and solid plot are the parallel and perpendicular fluorescence emissions, respectively. The bottom two plots are without any addition of z, and the top two plots include the additions of z = 0.5. (B) Plots of FA vs. time. The lighter solid plot (lower) was calculated from the parallel and perpendicular LIF emissions without addition of z (Eq. (1)). The darker solid plot (upper) was calculated with the addition of z = 0.5 (Eq. (3)).
background fluorescence intensities (I|| , and I⊥ for unbound ThT) is also apparent in Fig. 1. Based on reports in the literature, we expected that fluorescence peaks resulting from the A (1–42) aggregates would produce distinguishable peaks in a plot of FA vs. time [4]. Whelan et al. demonstrated the use of CE-LIFA for affinity assays to monitor the fluorescence anisotropy of G␣i1 protein complex formation with BODIPY-GTP␥S (BGTP␥S) [4]. In their study, the fluorescent probe, BODIPY, was included in the electrophoresis buffer, similar to ThT in this paper [4]. As shown in Fig. 4 of Whelan et al. [4], the parallel and perpendicular fluorescence signals of the G␣i1 /GTP␥S complex were approximately 40 and 25 RFU above background signals of 35 RFU for both the parallel and perpendicular channels [4]. A fluorescence anisotropy peak was clearly observed in plots of FA vs. time for the G␣i1 /GTP␥S complex with an intensity of approximately 0.08 above a fluorescence anisotropy baseline noise of less than 0.005 [4]. An important difference between the data reported by Whelan et al. and that presented here is the heights of the analyte peaks relative to the fluorescence background signal. For example, in Fig. 1 the ratios of the A aggregate peak heights to the intensity of the background signals for the parallel and perpendicular channels are 11.2
Fig. 3. (A) Electropherogram with LIF detection showing a peak for an individual A (1–42) aggregate at 148.35 s. The dashed plot and solid plot are the parallel and perpendicular fluorescence emissions, respectively. (B) Plots of FA vs. time calculated with different levels of z (0, 0.5, 1.5, 3.0 and 5.0) added. Note that all of the y axes in (B) are the same except in the plot where z = 0.
and 10.3, respectively. The equivalent ratios observed by Whelan et al. for the G␣i1 /GTP␥S complex were about 1.1 and 0.7 [4]. It was hypothesized that uniformly adding an artificial background signal to the data for both fluorescence channels, I|| , and I⊥ , prior to calculating the anisotropy might enable differentiation of the anisotropies of the aggregate peaks and background signal, resulting in the visualization of anisotropy peaks. This would greatly facilitate identification and interpretation of differences between FA of A aggregates and that of the background. Adding an artificial signal to all fluorescence data points would have the same effect on the data as if ThT were producing a high fluorescence background signal similar to the BODIPY background signal observed in the work of Whelan et al. [4].
102
R.A. Picou et al. / Analytica Chimica Acta 739 (2012) 99–103
Fig. 2A shows plots of the original fluorescence data (bottom two traces) and the fluorescence data after adding an artificial signal, z, of 0.5 to the original data (top two traces). In Fig. 2, the ratio of the fluorescence intensity of the A aggregate peak at 121.8 s to that of the background for the parallel and perpendicular channels are 7.0 and 8.4, respectively. The added value of 0.5 RFU produced ratios of A peak height:background signal of 1.1 and 0.7, which are the same ratios of analyte signal:background signal observed by Whelan et al. [4]. Fig. 2B shows the fluorescence anisotropy before and after the addition of the 0.5 RFU background signal, z. In Fig. 2B, the upper trace is the anisotropy calculated from the original fluorescence data, and the lower trace is the anisotropy calculated from the modified fluorescence data. As hypothesized, the addition of the artificial background signal results in a relative enhancement of the anisotropy of the aggregate peaks relative to the background anisotropy. It is clear that the fluorescence anisotropy peaks correspond in time to the fluorescence peaks in Fig. 2A. This result can be explained mathematically starting with the Eq. (1), where r is the original anisotropy, i.e. the anisotropy before the fluorescence signals are artificially adjusted. Eq. (1) was used to calculate the anisotropy data plotted in the upper trace of Fig. 2B. Recalling that this method adds a constant artificial signal, z, to all the fluorescence data, the modified anisotropy, rm , can be described by Eq. (2), which can be reduced to Eq. (3). Eq. (3) can be used to calculate the modified anisotropy, which is plotted vs. time in the lower plot in Fig. 2B. Subtracting Eq. (3) from Eq. (1) gives Eq. (4), which is useful for considering the differences observed between the two plots in Fig. 2B. rm =
(I|| + z) − (I⊥ + z) (I|| + z) + 2(I⊥ + z)
(2)
rm =
I|| − I⊥ I|| + 2I⊥ + 3z
(3)
r − rm =
I|| − I⊥ I|| + 2I⊥
−
I|| − I⊥ I|| + 2I⊥ + 3z
(4)
Only the second term on the right side of Eq. (4) is dependent on z. For lower fluorescence intensities (lower I|| , and I⊥ ), the z term is more important, and greater differences will result between the original and modified anisotropies. Conversely, adding z to higher fluorescence intensities, e.g. like those produced by aggregates, has less of an impact on the second term of Eq. (4), which translates to smaller differences between the original and modified anisotropies. In general, by adding z the anisotropy of the background fluorescence is reduced to a greater extent than the anisotropy of the A aggregate peaks. After addition of an artificial background signal, our data resemble the data obtained by Whelan et al., and peaks become apparent in a plot of anisotropy vs. time as shown in Fig. 2B. The natural background fluorescence in our data is due to the fluorescence of unbound ThT, and this fluorescence has an FA value of about 0.2 at the ThT concentrations used in this work [8,9]. If the background ThT fluorescence did not exhibit FA, the plot of FA vs. time would be similar to the upper plot in Fig. 2B, except it would be centered at 0.0 instead of 0.2. The high noise in the plot of FA vs. time would still largely mask the FA peaks without data treatment. The FA values calculated with Eq. (3) are not identical to the true (original) anisotropy values. The data treatment method presented here enables visualization and identification of FA peaks in a plot of FA vs. time. Fluorescence peaks that do not exhibit FA will not produce peaks in a plot of FA vs. time generated using Eq. (3). Once the FA peaks have been identified with the help of this method, the
Fig. 4. S/N of four FA peaks for individual A aggregates plotted as a function of z. The diamonds, triangles, circles and squares are S/N values for FA from fluorescence peaks whose parallel fluorescence S/N values were calculated to be 127, 65, 30 and 15, respectively. The S/N values plotted here are tabulated in Supplementary material.
true anisotropy values can be easily calculated using Eq. (1) for data points at the migration times of interest. The impact of the value of z added on the plots of FA vs. time produced using Eq. (3) was investigated. Fig. 3A shows another time segment of parallel and perpendicular polarized fluorescence data from the same electropherogram shown in Figs. 1 and 2. Fig. 3B shows five plots of FA vs. time, which were calculated from the fluorescence data in Fig. 3A using several values of z (z = 0, 0.5, 1.5, 3.0 and 5.0 RFU). Consistent with Eq. (3) and the discussion above, larger z values result in lower FA values, and FA peaks become apparent in the electropherograms. The S/N of some of the FA peaks were analyzed as a function of the magnitude of z added. Fig. 4 shows a plot of the S/N of FA vs. z for selected peaks from the same electropherogram. These S/N values are tabulated in Supplementary material (Table S1). The peaks in Fig. 4 were selected based on the S/N of their corresponding fluorescence peaks prior to any data treatment (S/N = 15, 30, 65, and 127). In general, the S/N values of the anisotropy peaks initially increase as z is increased before approaching a limiting value. For the peaks examined here, the S/N values of the anisotropy peaks plateau at a maximum S/N when z is approximately 10 RFU. The plots in Fig. 4 show that the maximum S/N for FA was greater if the corresponding fluorescence peak had a greater S/N, and the S/N for FA reaches a maximum value at a lower value of z if the corresponding fluorescence peak had a lower S/N. For z = 20 RFU, the anisotropy S/N values are 5.9, 11, 29, 49 and 143 for fluorescence peaks that had S/N values of 15, 30, 65, 127 and 323, respectively. 4. Conclusions A simple and effective data treatment method was described for identifying and visualizing fluorescence anisotropy peaks in plots of FA vs. time for capillary electrophoresis separations of A (1–42) peptide aggregates. The data treatment method revealed fluorescence anisotropy peaks that were masked by the high anisotropy noise of the background signal prior to the data treatment. Although this method did alter the FA values in the peaks, true FA values could be easily calculated once peaks of interest had been identified. While this data treatment method was applied to CE separations with laser-induced fluorescence anisotropy detection, it is equally applicable to other separation techniques using fluorescence anisotropy detection and for other analytes.
R.A. Picou et al. / Analytica Chimica Acta 739 (2012) 99–103
Acknowledgments This work was supported by Louisiana State University, the National Science Foundation CHE-0505972 (SDG), the Rosalinde and Arthur Gilbert Foundation/AFAR Award (IK) and Louisiana Board of Regents (IK). Ryan Picou was supported by a Louisiana Board of Regents Fellowship. Rebekah Cerqua was supported by NSF Research Experiences for Undergraduates (REU) CHE-0648841. The authors would like to thank the LSU School of Veterinary Medicine Microscopy Center in the Department of Comparative Biomedical Sciences for TEM work. Appendix A. Supplementary data Supplementary data associated with this article can be found, in the online version, at http://dx.doi.org/10.1016/j.aca.2012.06.023. References [1] J.R. Lakowicz, Principles of Fluorescence Spectroscopy, 3rd ed., Springer, Singapore, 2006.
103
[2] X.C. Le, Q.-H. Wan, M.T. Lam, Electrophoresis 23 (2002) 903. [3] Q.-H. Wan, X.C. Le, Anal. Chem. 72 (2000) 5583. [4] R.J. Whelan, R.K. Sunahara, R.R. Neubig, R.T. Kennedy, Anal. Chem. 76 (2004) 7380. [5] P. Yang, R.J. Whelan, E.E. Jameson, J.H. Kurzer, L.S. Argetsinger, C. Carter-Su, A. Kabir, A. Malik, R.T. Kennedy, Anal. Chem. 77 (2005) 2482. [6] P. Yang, R.J. Whelan, Y. Mao, A.W.-M. Lee, C. Carter-Su, R.T. Kennedy, Anal. Chem. 79 (2007) 1690. [7] L. Ye, X.C. Le, J.Z. Xing, M. Ma, R. Yatscoff, J. Chromatogr. B 714 (1998) 59. [8] R.A. Picou, D.P. Schrum, G. Ku, R.A. Cerqua, I. Kheterpal, S.D. Gilman, Anal. Biochem 425 (2012) 104. [9] R. Khurana, C. Coleman, C. Ionescu-Zanetti, S.A. Carter, V. Krishna, R.K. Grover, R. Roy, S. Singh, J. Struct. Biol. 151 (2005) 229. [10] M. Song, Y. Zhang, T. Li, Z. Wang, J. Yin, H. Wang, J. Chromatogr. A 1216 (2009) 873. [11] X. Wang, Y. Song, M. Song, Z. Wang, T. Li, H. Wang, Anal. Chem. 81 (2009) 7885. [12] M. Kato, H. Kinoshita, M. Enokita, Y. Hori, T. Hashimoto, T. Iwatsubo, T. Toyo’oka, Anal. Chem. 79 (2007) 4887. [13] R. Picou, J.P. Moses, A.D. Wellman, I. Kheterpal, S.D. Gilman, Analyst 135 (2010) 1631. [14] R.A. Picou, I. Kheterpal, A.D. Wellman, M. Minnamreddy, G. Ku, S.D. Gilman, J. Chromatogr. B 879 (2011) 627. [15] B. O’Nuallain, A.K. Thakur, A.D. Williams, A.M. Bhattacharyya, S. Chen, G. Thiagarajan, R. Wetzel, Methods Enzymol. 413 (2006) 34. [16] M. Biancalana, S. Koide, Biochim. Biophys. Acta 1804 (2010) 1405. [17] H. Levine III, Amyloid 12 (2005) 5.
Contents lists available at SciVerse ScienceDirect
Analytica Chimica Acta journal homepage: www.elsevier.com/locate/aca
A data treatment method for detecting fluorescence anisotropy peaks in capillary electropherograms Ryan A. Picou a , Indu Kheterpal b , S. Douglass Gilman a,∗ a b
Department of Chemistry, Louisiana State University, Baton Rouge, LA 70803, USA Pennington Biomedical Research Center, LSU System, Baton Rouge, LA 70808, USA
h i g h l i g h t s
g r a p h i c a l
a b s t r a c t
We explain the absence of fluorescence anisotropy peaks for capillary separations. We developed a method to visualize fluorescence anisotropy peaks for separations. This method was applied to a separation of amyloid beta peptide aggregates.
a r t i c l e
i n f o
Article history: Received 18 April 2012 Accepted 12 June 2012 Available online 19 June 2012 Keywords: Fluorescence anisotropy Data treatment Capillary electrophoresis Amyloid beta peptide
a b s t r a c t A data treatment method is presented to detect fluorescence anisotropy (FA) peaks in capillary electrophoresis electropherograms. The data treatment method converts plots of fluorescence anisotropy vs. time that contain no peaks that are distinguishable from the noise of the anisotropy background into plots that show distinct fluorescence anisotropy peaks. The method was demonstrated using laserinduced fluorescence anisotropy data from individual A (1–42) aggregates separated using capillary electrophoresis. Applying this data treatment method enabled the detection of anisotropy peaks for individual A aggregate fluorescence peaks that were not observed prior to the data treatment method. The data treatment method is not specifically designed for A aggregate analysis or capillary electrophoresis, and it should be applicable to other applications and other separation methods with FA detection. © 2012 Elsevier B.V. All rights reserved.
1. Introduction Fluorescence anisotropy (FA) is a measure of the degree of emission depolarization of a fluorescent molecule after being excited with a polarized excitation source [1]. A common cause of emission depolarization is molecular rotational diffusion during the excited state lifetime of the fluorophore [1]. In general, molecules with smaller rotational diffusion rates, i.e. slower rotation in solution, exhibit larger FA. Similarly, molecules that have large rotational diffusion rates exhibit smaller FA.
∗ Corresponding author. E-mail address: [email protected] (S.D. Gilman). 0003-2670/$ – see front matter © 2012 Elsevier B.V. All rights reserved. http://dx.doi.org/10.1016/j.aca.2012.06.023
Laser-induced fluorescence anisotropy (LIFA) has been demonstrated as an on-column detection technique for capillary electrophoresis (CE) for studies of biomolecular interactions, e.g. protein–protein or oligonucleotide–protein interactions [2–7]. Picou et al. recently studied the separation and detection of individual A amyloid aggregates by CE-LIF using ThT in the running buffer as the fluorescent probe [8]. It was hypothesized that differences in fluorescence anisotropy between aggregates could indicate relative size differences between separated aggregates. Plots of FA vs. time showed that FA due to aggregates and the ThT background [9] were indistinguishable; i.e. no fluorescence anisotropy peaks were observed although there were obvious peaks in plots of fluorescence vs. time [8]. In order to evaluate if CE-LIFA can be used to differentiate individually detected A
100
R.A. Picou et al. / Analytica Chimica Acta 739 (2012) 99–103
aggregates, it is important that the FA of the aggregates be easily distinguishable from the background anisotropy. Most CE-LIFA studies in the literature did not include plots of fluorescence anisotropy or polarization vs. time [2,3,6,7,10,11]. This suggests that other studies using FA for separations detection may have faced the same issues that we encountered when plotting FA vs. time, i.e. high FA noise and no observed FA peaks despite low background fluorescence and peaks with high S/N in the corresponding plots of fluorescence vs. time. In order to understand why our anisotropy plots did not result in distinct peaks as demonstrated in the literature [4,5] and to better visualize changes in FA vs. time that are evident in the raw fluorescence data [8], we have developed a simple data treatment method. This method enables differentiation of anisotropies due to fluorescent peaks (peptide aggregates here) and background signals. This data treatment method effectively extracts anisotropy peaks from FA electropherograms that were originally embedded in the background FA noise. This method allows simple visualization of anisotropy peaks vs. time.
Fig. 1. Bottom two plots are electropherograms with LIF detection showing an individual A (1–42) aggregate peak at 120.65 s (left y-axis). The dashed plot and dark solid plot are the parallel and perpendicular fluorescence emissions, respectively. The uppermost plot (lighter solid) is the fluorescence anisotropy, r (right y-axis), calculated from the parallel and perpendicular emissions using Eq. (1).
2. Experimental 2.1. Chemicals and peptide samples The chemicals, solutions and sample preparations used in this study have been described in detail previously [8]. Tris (hydroxymethyl) aminomethane (Tris) was purchased from Fisher Scientific (Fair Lawn, NJ). Thioflavin T (ThT) was purchased from Sigma (St. Louis, MO). The electropherograms presented in this work were performed using 10.00 mM Tris buffer at pH 7.79. Thioflavin T (15 M) was included in the running buffer to detect A (1–42) aggregates [8,12]. The A (1–42) peptide samples were prepared as previously described [13,14]. The monomer-equivalent concentration of the A (1–42) was 5 M as determined by HPLC-UV [15]. Analysis of samples by TEM, CE-UV and CE-LIF confirmed that they contained A (1–42) aggregates [8].
(PMT) (H9306-04, Hamamatsu; Bridgewater, NJ). The outputs of the PMTs were filtered by 500 Hz low pass RC filters and collected by a National Instrument PCI-6024E DAQ board at a scan rate of 1000 Hz. The CE-LIFA data collection was controlled by a program written in LabView 5.0, and the data were analyzed using Origin Pro 7.5 and Microsoft Excel 2007. Eq. (1) was used to calculate the fluorescence anisotropy, r, where I|| , and I⊥ , are the measured parallel and perpendicular fluorescence intensities (relative to the polarization of the excitation source). The factor G corrects the anisotropy for the differences in sensitivities of the two PMTs. For this work, G was empirically adjusted to 1 by optimizing the potentials applied to both PMTs so that the responses and sensitivities were equal for a small fluorescent molecule, Lucifer Yellow (Molecular Probes, Eugene, OR) over a range of concentrations (data not shown). The applied potentials for PMT|| and PMT⊥ were 1000 and 935 V, respectively.
2.2. Capillary electrophoresis r= Capillary electrophoresis was performed in a fused-silica capillary (ID = 50 m, OD 366 m; Polymicro Technologies, Phoenix, AZ) cut to 60.0 cm total length with a window created at 23.0 cm from the inlet. The electrophoretic potential was applied across the capillary using a Spellman CZE1000R high-voltage power supply (Hauppauge, NY) at 25.0 kV (417 V/cm), which produced an electrophoretic current of 5–6 A. Sample injections were performed for 5.0 s at 25.0 kV. Prior to each injection, the sample was vortexed to suspend any aggregates that might have settled to the bottom of the sample vial. 2.3. Laser-induced fluorescence anisotropy detection system The FA detection system used here has been described in detail previously [8]. This detector is based on the design reported by Whelan et al. [4] and has been modified for the detection of narrow peaks (3–15 ms) due to individual aggregates. A schematic of the detection system is shown in Supplementary material (Fig. S1). The beamsplitter cube (10FC16PB.3, Newport Corporation; Irvine, CA) separates the fluorescence emission into parallel and perpendicular components relative to the polarization of the 445 nm diode laser excitation source (Model no. LDCU12/7532, Power Technology, Inc; Alexander, AR). Both emission components were spectrally filtered by a 490 nm bandpass filter and 440 nm notch filter. Then the two emission components were detected simultaneously and equidistant from the capillary by two identical photomultiplier tubes
I|| − GI⊥ I|| + 2GI⊥
(1)
3. Results and discussion The aggregation of A peptides has been linked to Alzheimer’s disease. The data presented here are from a capillary electrophoretic separation of a mixture of A (1–42) peptide aggregates. The peptides were non-covalently labeled with the amyloid dye, thioflavin T (ThT). The peaks in the electropherograms presented here are due to individual A (1–42) aggregates [8]. Fig. 1 presents a representative section from electropherograms for samples containing A (1–42) aggregates. The individual parallel and perpendicular fluorescence signals (I|| , and I⊥ ) are plotted in the lower part of the figure (left y-axis), and the FA calculated from these signals (Eq. (1)) is plotted in the top part of the figure (right y-axis). The fluorescence peak in Fig. 1 is due to an individual A (1–42) aggregate with ThT noncovalently bound to it [8]. The binding of ThT to A causes excitation and emission spectral shifts and enhancement at the wavelengths being used for excitation and detection (440 and 490 nm) [16,17]. Fig. 1 shows only 1 of the 178 aggregate peaks that were detected for this injection of A (1–42) aggregates. Fig. 1 shows that the fluorescence anisotropies due to A aggregates and due to background fluorescence of ThT [9] are indistinguishable, i.e. no FA peaks are apparent in the plot of FA vs. time although fluorescence peaks for I|| , and I⊥ and their different intensities are obvious. The slight difference in the
R.A. Picou et al. / Analytica Chimica Acta 739 (2012) 99–103
101
Fig. 2. (A) Electropherograms with LIF detection showing peaks due to two individual A (1–42) aggregates at 121.84 s and 122.02 s. The dashed plot and solid plot are the parallel and perpendicular fluorescence emissions, respectively. The bottom two plots are without any addition of z, and the top two plots include the additions of z = 0.5. (B) Plots of FA vs. time. The lighter solid plot (lower) was calculated from the parallel and perpendicular LIF emissions without addition of z (Eq. (1)). The darker solid plot (upper) was calculated with the addition of z = 0.5 (Eq. (3)).
background fluorescence intensities (I|| , and I⊥ for unbound ThT) is also apparent in Fig. 1. Based on reports in the literature, we expected that fluorescence peaks resulting from the A (1–42) aggregates would produce distinguishable peaks in a plot of FA vs. time [4]. Whelan et al. demonstrated the use of CE-LIFA for affinity assays to monitor the fluorescence anisotropy of G␣i1 protein complex formation with BODIPY-GTP␥S (BGTP␥S) [4]. In their study, the fluorescent probe, BODIPY, was included in the electrophoresis buffer, similar to ThT in this paper [4]. As shown in Fig. 4 of Whelan et al. [4], the parallel and perpendicular fluorescence signals of the G␣i1 /GTP␥S complex were approximately 40 and 25 RFU above background signals of 35 RFU for both the parallel and perpendicular channels [4]. A fluorescence anisotropy peak was clearly observed in plots of FA vs. time for the G␣i1 /GTP␥S complex with an intensity of approximately 0.08 above a fluorescence anisotropy baseline noise of less than 0.005 [4]. An important difference between the data reported by Whelan et al. and that presented here is the heights of the analyte peaks relative to the fluorescence background signal. For example, in Fig. 1 the ratios of the A aggregate peak heights to the intensity of the background signals for the parallel and perpendicular channels are 11.2
Fig. 3. (A) Electropherogram with LIF detection showing a peak for an individual A (1–42) aggregate at 148.35 s. The dashed plot and solid plot are the parallel and perpendicular fluorescence emissions, respectively. (B) Plots of FA vs. time calculated with different levels of z (0, 0.5, 1.5, 3.0 and 5.0) added. Note that all of the y axes in (B) are the same except in the plot where z = 0.
and 10.3, respectively. The equivalent ratios observed by Whelan et al. for the G␣i1 /GTP␥S complex were about 1.1 and 0.7 [4]. It was hypothesized that uniformly adding an artificial background signal to the data for both fluorescence channels, I|| , and I⊥ , prior to calculating the anisotropy might enable differentiation of the anisotropies of the aggregate peaks and background signal, resulting in the visualization of anisotropy peaks. This would greatly facilitate identification and interpretation of differences between FA of A aggregates and that of the background. Adding an artificial signal to all fluorescence data points would have the same effect on the data as if ThT were producing a high fluorescence background signal similar to the BODIPY background signal observed in the work of Whelan et al. [4].
102
R.A. Picou et al. / Analytica Chimica Acta 739 (2012) 99–103
Fig. 2A shows plots of the original fluorescence data (bottom two traces) and the fluorescence data after adding an artificial signal, z, of 0.5 to the original data (top two traces). In Fig. 2, the ratio of the fluorescence intensity of the A aggregate peak at 121.8 s to that of the background for the parallel and perpendicular channels are 7.0 and 8.4, respectively. The added value of 0.5 RFU produced ratios of A peak height:background signal of 1.1 and 0.7, which are the same ratios of analyte signal:background signal observed by Whelan et al. [4]. Fig. 2B shows the fluorescence anisotropy before and after the addition of the 0.5 RFU background signal, z. In Fig. 2B, the upper trace is the anisotropy calculated from the original fluorescence data, and the lower trace is the anisotropy calculated from the modified fluorescence data. As hypothesized, the addition of the artificial background signal results in a relative enhancement of the anisotropy of the aggregate peaks relative to the background anisotropy. It is clear that the fluorescence anisotropy peaks correspond in time to the fluorescence peaks in Fig. 2A. This result can be explained mathematically starting with the Eq. (1), where r is the original anisotropy, i.e. the anisotropy before the fluorescence signals are artificially adjusted. Eq. (1) was used to calculate the anisotropy data plotted in the upper trace of Fig. 2B. Recalling that this method adds a constant artificial signal, z, to all the fluorescence data, the modified anisotropy, rm , can be described by Eq. (2), which can be reduced to Eq. (3). Eq. (3) can be used to calculate the modified anisotropy, which is plotted vs. time in the lower plot in Fig. 2B. Subtracting Eq. (3) from Eq. (1) gives Eq. (4), which is useful for considering the differences observed between the two plots in Fig. 2B. rm =
(I|| + z) − (I⊥ + z) (I|| + z) + 2(I⊥ + z)
(2)
rm =
I|| − I⊥ I|| + 2I⊥ + 3z
(3)
r − rm =
I|| − I⊥ I|| + 2I⊥
−
I|| − I⊥ I|| + 2I⊥ + 3z
(4)
Only the second term on the right side of Eq. (4) is dependent on z. For lower fluorescence intensities (lower I|| , and I⊥ ), the z term is more important, and greater differences will result between the original and modified anisotropies. Conversely, adding z to higher fluorescence intensities, e.g. like those produced by aggregates, has less of an impact on the second term of Eq. (4), which translates to smaller differences between the original and modified anisotropies. In general, by adding z the anisotropy of the background fluorescence is reduced to a greater extent than the anisotropy of the A aggregate peaks. After addition of an artificial background signal, our data resemble the data obtained by Whelan et al., and peaks become apparent in a plot of anisotropy vs. time as shown in Fig. 2B. The natural background fluorescence in our data is due to the fluorescence of unbound ThT, and this fluorescence has an FA value of about 0.2 at the ThT concentrations used in this work [8,9]. If the background ThT fluorescence did not exhibit FA, the plot of FA vs. time would be similar to the upper plot in Fig. 2B, except it would be centered at 0.0 instead of 0.2. The high noise in the plot of FA vs. time would still largely mask the FA peaks without data treatment. The FA values calculated with Eq. (3) are not identical to the true (original) anisotropy values. The data treatment method presented here enables visualization and identification of FA peaks in a plot of FA vs. time. Fluorescence peaks that do not exhibit FA will not produce peaks in a plot of FA vs. time generated using Eq. (3). Once the FA peaks have been identified with the help of this method, the
Fig. 4. S/N of four FA peaks for individual A aggregates plotted as a function of z. The diamonds, triangles, circles and squares are S/N values for FA from fluorescence peaks whose parallel fluorescence S/N values were calculated to be 127, 65, 30 and 15, respectively. The S/N values plotted here are tabulated in Supplementary material.
true anisotropy values can be easily calculated using Eq. (1) for data points at the migration times of interest. The impact of the value of z added on the plots of FA vs. time produced using Eq. (3) was investigated. Fig. 3A shows another time segment of parallel and perpendicular polarized fluorescence data from the same electropherogram shown in Figs. 1 and 2. Fig. 3B shows five plots of FA vs. time, which were calculated from the fluorescence data in Fig. 3A using several values of z (z = 0, 0.5, 1.5, 3.0 and 5.0 RFU). Consistent with Eq. (3) and the discussion above, larger z values result in lower FA values, and FA peaks become apparent in the electropherograms. The S/N of some of the FA peaks were analyzed as a function of the magnitude of z added. Fig. 4 shows a plot of the S/N of FA vs. z for selected peaks from the same electropherogram. These S/N values are tabulated in Supplementary material (Table S1). The peaks in Fig. 4 were selected based on the S/N of their corresponding fluorescence peaks prior to any data treatment (S/N = 15, 30, 65, and 127). In general, the S/N values of the anisotropy peaks initially increase as z is increased before approaching a limiting value. For the peaks examined here, the S/N values of the anisotropy peaks plateau at a maximum S/N when z is approximately 10 RFU. The plots in Fig. 4 show that the maximum S/N for FA was greater if the corresponding fluorescence peak had a greater S/N, and the S/N for FA reaches a maximum value at a lower value of z if the corresponding fluorescence peak had a lower S/N. For z = 20 RFU, the anisotropy S/N values are 5.9, 11, 29, 49 and 143 for fluorescence peaks that had S/N values of 15, 30, 65, 127 and 323, respectively. 4. Conclusions A simple and effective data treatment method was described for identifying and visualizing fluorescence anisotropy peaks in plots of FA vs. time for capillary electrophoresis separations of A (1–42) peptide aggregates. The data treatment method revealed fluorescence anisotropy peaks that were masked by the high anisotropy noise of the background signal prior to the data treatment. Although this method did alter the FA values in the peaks, true FA values could be easily calculated once peaks of interest had been identified. While this data treatment method was applied to CE separations with laser-induced fluorescence anisotropy detection, it is equally applicable to other separation techniques using fluorescence anisotropy detection and for other analytes.
R.A. Picou et al. / Analytica Chimica Acta 739 (2012) 99–103
Acknowledgments This work was supported by Louisiana State University, the National Science Foundation CHE-0505972 (SDG), the Rosalinde and Arthur Gilbert Foundation/AFAR Award (IK) and Louisiana Board of Regents (IK). Ryan Picou was supported by a Louisiana Board of Regents Fellowship. Rebekah Cerqua was supported by NSF Research Experiences for Undergraduates (REU) CHE-0648841. The authors would like to thank the LSU School of Veterinary Medicine Microscopy Center in the Department of Comparative Biomedical Sciences for TEM work. Appendix A. Supplementary data Supplementary data associated with this article can be found, in the online version, at http://dx.doi.org/10.1016/j.aca.2012.06.023. References [1] J.R. Lakowicz, Principles of Fluorescence Spectroscopy, 3rd ed., Springer, Singapore, 2006.
103
[2] X.C. Le, Q.-H. Wan, M.T. Lam, Electrophoresis 23 (2002) 903. [3] Q.-H. Wan, X.C. Le, Anal. Chem. 72 (2000) 5583. [4] R.J. Whelan, R.K. Sunahara, R.R. Neubig, R.T. Kennedy, Anal. Chem. 76 (2004) 7380. [5] P. Yang, R.J. Whelan, E.E. Jameson, J.H. Kurzer, L.S. Argetsinger, C. Carter-Su, A. Kabir, A. Malik, R.T. Kennedy, Anal. Chem. 77 (2005) 2482. [6] P. Yang, R.J. Whelan, Y. Mao, A.W.-M. Lee, C. Carter-Su, R.T. Kennedy, Anal. Chem. 79 (2007) 1690. [7] L. Ye, X.C. Le, J.Z. Xing, M. Ma, R. Yatscoff, J. Chromatogr. B 714 (1998) 59. [8] R.A. Picou, D.P. Schrum, G. Ku, R.A. Cerqua, I. Kheterpal, S.D. Gilman, Anal. Biochem 425 (2012) 104. [9] R. Khurana, C. Coleman, C. Ionescu-Zanetti, S.A. Carter, V. Krishna, R.K. Grover, R. Roy, S. Singh, J. Struct. Biol. 151 (2005) 229. [10] M. Song, Y. Zhang, T. Li, Z. Wang, J. Yin, H. Wang, J. Chromatogr. A 1216 (2009) 873. [11] X. Wang, Y. Song, M. Song, Z. Wang, T. Li, H. Wang, Anal. Chem. 81 (2009) 7885. [12] M. Kato, H. Kinoshita, M. Enokita, Y. Hori, T. Hashimoto, T. Iwatsubo, T. Toyo’oka, Anal. Chem. 79 (2007) 4887. [13] R. Picou, J.P. Moses, A.D. Wellman, I. Kheterpal, S.D. Gilman, Analyst 135 (2010) 1631. [14] R.A. Picou, I. Kheterpal, A.D. Wellman, M. Minnamreddy, G. Ku, S.D. Gilman, J. Chromatogr. B 879 (2011) 627. [15] B. O’Nuallain, A.K. Thakur, A.D. Williams, A.M. Bhattacharyya, S. Chen, G. Thiagarajan, R. Wetzel, Methods Enzymol. 413 (2006) 34. [16] M. Biancalana, S. Koide, Biochim. Biophys. Acta 1804 (2010) 1405. [17] H. Levine III, Amyloid 12 (2005) 5.