| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
* National Institute on Aging, National Institutes of Health, Baltimore, Maryland; and Department of Pathology, Interdisciplinary Graduate School of Medicine and Engineering, University of Yamanashi, Yamanashi, Japan
Correspondence: Address reprint requests to Nikolai M. Soldatov, Tel.: 410-558-8343; E-mail: soldatovn{at}grc.nia.nih.gov.
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONSUPPLEMENTARY MATERIALACKNOWLEDGEMENTSREFERENCES |
|---|
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONSUPPLEMENTARY MATERIALACKNOWLEDGEMENTSREFERENCES |
|---|
; Vermeer et al., 2004). Owing to advances in live cell imaging, it has become possible to monitor the activity of a variety of signaling molecules in live cells. Fluorescence resonance energy transfer (FRET) microscopy is a popular tool for revealing the interaction of signaling molecules, highlighting nanometer-scale associations in cellular milieu (Wallrabe et al., 2003; Kobrinsky et al., 2004; Sharma et al., 2004). FRET essentially involves the transfer of energy from an excited donor fluorophore to an acceptor fluorophore, providing that the two fluorophores are within a certain proximity (1–10 nm). Thus, FRET microscopy is often regarded as a molecular ruler, which forms the rationale for its use in imaging molecular interactions (Kenworthy, 2001). Additionally, FRET microscopy can be used for the detection of conformational changes in reporter molecules, where fluorescent groups are attached to the same molecule. Rearrangements of the fluorophores, induced by activation of opposing processes (e.g., phosphorylation/dephosphorylation), are then detected by FRET microscopy and thereby represent the dynamic activity of a functional signaling domain. Analysis of signaling in the plasma membrane presents certain challenges reflecting a dynamic process taking place in a confined space. Numerous studies aimed at the plasma membrane have provided indirect evidence for the existence of distinct signaling domains. However, despite an increasing interest in plasma membrane signaling, the image analysis of the formation and function of signaling domains is not yet completely developed. In this work, we propose a wavelet-based image analysis to detect and quantify dynamic changes in functional signaling domains with an example of the protein kinase C (PKC) activity in plasma membranes of live cells. The PKC signaling cascade is important for various cell functions (Shirai and Saito, 2002), and phosphorylation of plasma membrane proteins represents a key step in cell signaling. We expressed the C kinase activity reporter (CKAR) of PKC phosphorylation activity (Violin et al., 2003) confined to the plasma membrane, offering a method for detecting phosphorylation/dephosphorylation activity of protein kinase C and phosphatase in such an environment. We utilized this system as a model for developing a wavelet-based approach for FRET microscopy, allowing for the detection of macro- and microfunctional signaling domains.
Our overall approach is illustrated in Fig. 1, which begins with a narrowing-down of manually outlined regions of interest (ROIm) defined as the plasma membrane region with obvious signals of fluorescence and little or no interference from fluorescence of other cellular compartments (Fig. 1 A, left panel). Our aim is to narrow-down ROIm using statistical comparisons of FRET matrices and to identify domains of signaling within the statistically redefined ROIs (Fig. 1 A, right panel). FRET values within the redefined ROIs are linearized over the plasma membrane perimeter transforming two-dimensional images of the plasma membrane into one-dimensional signals (Fig. 1 B). These signals are subjected to a continuous wavelet transform and temporal differences between wavelet coefficient matrices are used to identify heterogeneous domains. FRET intensity in identified domains can then be compared over time to ascertain their dynamics and potential for representing functional signaling domains. The spatial resolution of this method is limited by the diffraction limit of 250 nm (the pixel size in our experiments). This method provides a means of elucidating the spatial heterogeneity in the dynamics of phosphorylation/dephosphorylation of PKC substrates in the plasma membrane and may be extended to a wide variety of recently developed FRET-based indicators (Zaccolo, 2004).
|
|
MATERIALS AND METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONSUPPLEMENTARY MATERIALACKNOWLEDGEMENTSREFERENCES |
|---|
Cell culture and transfection
COS1 cells were grown on poly-D-lysine-coated coverslips (MatTek, Ashland, MA) in DMEM supplemented with 10% fetal calf serum. COS1 cells were transfected with CKAR-pcDNA3 using the Effectene kit (Qiagen, Valencia, CA).
FRET imaging and image analysis
Images were recorded in live transfected COS1 cells with a Hamamatsu digital camera C4742-95 mounted on the Nikon epifluorescent microscope TE200 (60 x 1.2 N.A. objective) equipped with multiple filter sets (Chroma Technology, Rockingham, VT). Excitation light was provided by a 175 W Xenon lamp. C-Imaging (Compix, Cranberry Township, PA) and MetaMorph (Universal Imaging) software were used to obtain and analyze FRET images. FRET was quantified with the following three filter sets: for the acceptor monomeric yellow fluorescent protein (YFP) cube, excitation filter 500/20 nm, dichroic beam splitter 515 nm, emission filter 535/30 nm; for the donor monomeric cyan fluorescent protein (CFP) cube, excitation filer 436/20 nm, dichroic beam splitter 455 nm, emission filter 480/30 nm; and for the FRET (CFP/YFP) cube, excitation 436/20 nm, dichroic beam splitter 455 nm, emission filter 535/40 nm. Excitation filter sets were changed by a high-speed filter wheel system (Lambda 10-2, Sutter Instruments, Novato, CA). The simultaneous acquisition of two fluorescence images (donor and FRET) was achieved with the Dual-View system (Optical Insights, LLC, Santa Fe, NM), which was attached to the microscope. The time of acquisition varied from 50 to 300 ms, but was held constant during a given experiment. ROIm was selected from the plasma membrane of a control cell, where there were obvious signals of fluorescence and little or no interference from fluorescence from the intracellular compartment. ROIm was selected using the C-Imaging software program, where intensity (I) from three filter sets was determined after background subtraction. FRET values were calculated in each pixel and averaged across the ROIm (with one ROIm per cell). The full scale of digitized resolution ranged from 0 to 255. Corrected intensity of FRET (FRETc) was calculated as
 | (1) |
. This calibration is based on the measurements of fluorescence of the CFP- and YFP-labeled constructs expressed in COS1 cells in all three filter sets to determine bleedthrough in CFP, YFP, and FRET filters (for additional details, see Kobrinsky et al., 2003). Corrected FRET values were normalized by donor and acceptor levels, 
according to Xia and Liu (2001). Experiments resulting in images with pixel shifts were discarded and only experiments with no pixel shifts were analyzed.
Redefining the plasma membrane region
For pixel-by-pixel calculations, three consecutive images of one and the same cell before ACh or PDBu exposure (control images), along with three consecutive images after prolonged PKC activation when stabilized effect (SE) was reached (at 5 min for ACh and 15 min for PDBu), were converted, by importation to Excel (MicroSoft, Redmond, WA), into a spreadsheet view in a matrix format. ROIm of the control cell image (see above) was superimposed onto all FRETc matrices, and fluorescence values outside the ROIm were zeroed. Then the paired t-test was applied to each pixel comparing mean values between the control and SE images. Timepoints to be compared with control images (5 min for ACh and 15 min for PDBu) were selected based on stabilization of changes in FRET signals. For example, average FRET over ROI was found to be decreased to approximately the same extent as FRET in images sampled at 14- and 16-min (see Fig. 2 B, inset). This corresponds to the time required for stabilization of the PDBu application (10–15 min) reported by Violin et al. (2003). Statistical difference between the means of each pixel was determined, and the ROI were narrowed down to only those pixels exhibiting a statistically significant difference (p < 0.05). Any gaps were filled by including statistically nonsignificant pixels to connect these regions (up to 3 pixels thick), redefining a continuous plasma membrane region. For subsequent analyses, final matrices were linearized or transformed into one-dimensional signals using a simple algorithm (see Supplementary Material) in MatLab (The MathWorks, Natick, MA).
|
; 2003). The values of angular second momentum (ASM), sum of variance (SVar), and difference variance (DVar) were calculated from normalized gray-tone (8-bit) spatial-dependence matrices (co-occurrence matrices per Haralick et al., 1973), reflecting spatial homogeneity, spatial heterogeneity, and contrast, respectively, according to
 | (2) |
 | (3) |
 | (4) |
)
 | (5) |
 | (6) |
Wavelet analysis
The continuous wavelet transform (CWT) of a linearized spatial signal f(x) (amplitude over distance) is defined as
 | (7) |
represents the wavelet basis function. Calculated coefficients (Cab) of the wavelet transform reflect an amplitude or the correlation between the original signal and the basis function at different scales and positions in space. The scale parameter is inversely related to frequency. Thus large-scale values allow for the global analysis of the low-frequency components of the spatial signal, whereas small scales provide a means for local analysis of the high-frequency components. For more details, Aristizabal and Glavinovic (2003) provide an excellent introduction to the fundamentals of wavelet analysis.
There are several families of wavelets and the choice of a wavelet basis function is often made arbitrarily or via a trial-and-error check. Unlike some wavelets that are infinite and defined by analytic formulas, Daubechies (db) wavelets are the limits of an iterative process and have the value zero everywhere outside a specific interval or support. Thus, in addition to other mathematical properties, db wavelets have compact support and can be used in local analysis (Misiti et al., 2000). For comparison purposes, a CWT with db4, db9, and the Haar wavelet basis functions was applied to each signal of linearized FRET values in the statistically defined ROI. The numbers associated with the db wavelets correspond to the order of the wavelet and the Haar wavelet is the simplest basis function, resembling a step function. The number of examined scales ranged from 32 to 96, depending on the length of a given signal, but was held constant for each cell. Temporal differences (D) in wavelet coefficient matrices (C) were assessed by computing the squares of the differences 
where the subscripts C and t represent control and time. Difference coefficient matrices were normalized by the maximal square difference and rendered in two-dimensional pseudo-color maps as a function of scale and space. Domains were identified initially by visual inspection of areas of "hot" color, which indicate a change in signal heterogeneity. Each initial domain was then scrutinized using a one-way ANOVA with Dunnett's multiple comparisons test, where the mean and standard deviation of FRETc values within a domain were compared at the zero, 1-min, and SE timepoints. Only those domains that showed statistically significant temporal changes in FRETc values were finally selected as signaling domains of interest. Standard deviations of wavelet coefficients were calculated also at each scale, providing an objective measure of coefficient dispersion that could be monitored over time. All calculations associated with wavelet analysis were performed using MatLab.
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONSUPPLEMENTARY MATERIALACKNOWLEDGEMENTSREFERENCES |
|---|
). A decrease in FRET reflects an increase of PKC-mediated phosphorylation, whereas an increase in FRET corresponds to an increase in phosphatase activity. CFP and YFP fluorophores contained mutation A206K to reduce homoaffinity of all GFPs and preclude nonspecific clustering and intermolecular FRET by CFP-YFP dimerization (Zacharias et al., 2002).
Exposure to the PKC-activator phorbol dibutyrate (PDBu) produced changes in FRET signals, allowing for the real-time imaging of PKC-activation. An example of the PDBu effect on CKAR phosphorylation expressed in COS1 cells is shown in Fig. 2 A. It was essential to define the plasma membrane region that was the source of the FRET signal. To this end, we redefined the selected ROIm (middle panel) based on a statistical analysis (lower panel). The number of the statistically significant pixels was 
40% of the pixels in the ROIm. Corrected FRET values in the redefined plasma membrane ROIs were greater than those in the ROIm, but followed a similar trend (Fig.2 B, inset). These plasma membrane regions were subsequently linearized for texture and wavelet analysis (Fig. 2 B). Fig. 2 C (left panel) shows the same image of a CKAR-expressing COS1 cell obtained with CFP filter before application of PDBu. The CFP (donor) intensity profile along the plasma membrane (Fig. 2 C, right panel) represents the level of CKAR expression. One can see that the CKAR expression is not homogeneous along the plasma membrane. FRET values normalized by the level of expression (FRETN) are also plotted in Fig. 2 C (right panel). These values represent a balance between phosphorylation/dephosphorylation of CKAR and were not always correlated with the level of CKAR expression (compare FRETN and CFP intensity in Fig. 2 C, right panel). A high level of CKAR expression is observed in the left portion of Fig. 2 C, right panel, representing 0–100 pixels of the linearized plasma membrane. However, it corresponds to a low level of FRET signal in the respective region in Fig. 2 B, and FRETN in Fig. 2 C, right panel, indicating a low level of phosphatase activity or a high level of PKC activity in this part of the plasma membrane. The heterogeneity in the distribution of FRET levels along the plasma membrane under control conditions was produced by the heterogeneity of both CKAR expression and the balance of PKC-mediated phosphorylation/phosphatase activity. Our goal was to identify functional signaling domains based on temporal changes in the FRET signal along the plasma membrane that are not affected by heterogeneity of CKAR expression. The difference in expression levels is most likely produced by much slower processes than the time frame of changes we observed in the CKAR signal induced by PKC activation (Violin et al, 2003). Therefore, the identification of functional signaling domains was based on fast temporal changes (seconds and minutes) in FRET signaling, where the FRET-based CKAR reporter expression in each particular pixel of the obtained images is likely to be similar or relatively constant.
We applied a continuous wavelet transform (CWT) to the linearized membrane corrected FRET values at selected timepoints of PDBu application using, for comparison, the db4, db9, and Haar wavelet basis functions (Fig. 2 D). COS1 cells showed episodes of transient heterogeneity in the distribution of corrected FRET at high and low scale frequencies. After prolonged PDBu exposure, the distribution of the corrected FRET and wavelet coefficients along the plasma membrane became more homogeneous (Fig. 2 D). Texture analysis (Murata et al., 2001; 2003) revealed that the homogeneity of the corrected FRET values over the plasma membrane increased twofold after 15 min of PDBu application. Comparing the texture analysis parameters for the control and 15-min PDBu application data, we found that the ASM parameter (homogeneity) increased from 0.0011 to 0.0025, SVar (heterogeneity) decreased from 24,600 to 13,000, and DVar (contrast) decreased from 1330 to 625. The temporal heterogeneity of the corrected FRET distribution was observed in plots of the squares of differences in wavelet coefficients in the space-frequency domain (Fig. 3). These wavelet coefficient plots provided means of identifying domains at different spatial frequencies (see brackets in Fig. 2, B–D). A one-way ANOVA with Dunnett's multiple comparisons test was applied for 0, 1, and 15-min timepoints and statistically significant domains were selected. One of the important findings was that FRET changes in identified microdomains did not always correspond to the overall mean changes along the plasma membrane (Table 1). The size of the signaling domains was variable (see regions marked by brackets 1–3 in Fig. 2, B–D). Although obvious polarized changes in FRET in one part of the plasma membrane can define a macrodomain (Fig. 2 B), the sizes of microdomains typically did not exceed a few pixels, ranging from 1 to 5 µm. Similar domains of interest were identified with all three tested wavelet basis functions (Fig. 2 D). No significant changes in the structure of the domains were found for cells incubated with control solution at 5- and 15-min intervals (data not shown). Identification of functional signaling microdomains based on temporal changes of FRET values along the plasma membrane was not dependent on the difference in the level of CKAR expression, as can be seen from Fig. 2 B (FRETc) and Fig. 2 C (CKAR expression level), where signaling microdomains (marked by brackets 1–3) are localized in a plasma membrane region with a relatively high and stable level of CKAR expression.
|
|
|
|
). We expressed the human M1 muscarinic type 1 receptor along with the CKAR construct in COS1 cells. Application of acetylcholine (ACh; 10 µM) induced a transient increase in FRET signal followed by a continuous decline (Fig. 5 A, inset). Similar to the PDBu treatment, wavelet analysis of the linearized FRETc image (Fig. 5 A) revealed membrane microdomains (marked by brackets 1–4) that responded to 1-min ACh application differently from the overall plasma membrane FRETc plotted in the inset (Fig. 5 A). For example, the overall FRET signal at the 1-min timepoint was slightly increased (Fig. 5 A), as was FRETc in microdomain 1, whereas microdomains 3 and 4 showed a decrease in FRET signal (Table 1). This may suggest that the phosphorylation/dephosphorylation status of the CKAR reporter was not uniform over the plasma membrane. With the exception of ASM, parameters of the texture analysis of the ACh-stimulated PKC phosphorylation followed a similar trend as the PDBu exposure, where ASM decreased from 0.0043 to 0.0009, SVar decreased from 54,100 to 10,500, and DVar decreased from 5570 to 1320. Overall, there is general agreement between the wavelet and texture analyses, whereby both methods suggest a decrease in the heterogeneity of corrected FRET values after extended (5 min) PKC activation by ACh. A three-dimensional representation of FRETc and identified signaling domains in ROIs for both PDBu- and ACh-treated cells (Fig. 6; note that regions outside the ROIs were zeroed as described previously) revealed the spatial localization and temporal characteristics of identified domains. For example, the FRETc intensity in domain 2 for the PDBu-treated cell (previously identified in Fig. 2) is shown to increase at 1 min and decrease after 15 min of PDBu application (Fig. 6 A). This is in contrast to domains 1 and 3 that show decreasing FRETc at both the 1- and 15-min timepoints. A similar heterogeneity (Fig. 6 B) was revealed for the COS1 cell exposed to ACh (Fig. 5), where domain 1 shows a significant increase in FRET at 1 min, while domain 4 exhibits a significant decrease (Fig. 6 B). On the other hand, domains 2–4 all reveal significantly decreased FRETc after 5 min of ACh exposure.
|
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONSUPPLEMENTARY MATERIALACKNOWLEDGEMENTSREFERENCES |
|---|
We have applied the CWT to FRET signals in linearized ROI located within plasma membranes. This method of analysis is typically applied to time-series data. Unlike the traditional fast Fourier transform, the CWT co-localizes in both frequency (scale) and space (or time) domains. Although space-frequency localization can be achieved with a windowed Fourier transform, the efficient multiresolution representation of signals that contain transient properties, such as those observed in this study, is often better realized using wavelet analysis (Laine, 2000). For example, with a windowed Fourier transform, the width of the temporal or spatial window is prespecified and remains constant over the duration of the signal. However, a short window would be more desirable for assessing the high-frequency components of a signal, whereas a longer window or region would be needed for determining low-frequency information. Wavelet analysis represents a logical extension of the windowed Fourier technique, providing variable-sized windowing based on the scale of the wavelet basis function. Thus, wavelet analysis captures signal properties that may be missed by traditional techniques, such as fractal scaling (Arneodo et al., 1995). The wavelet decomposition consists of calculating a resemblance index (coefficients) between a signal and a wavelet basis function. Wavelet coefficient plots provide visualization of the changes in the heterogeneity of FRET signals at different timepoints and localize these changes within the plasma membrane ROI. For these signals and the wavelets analyzed in this study (db4, db9, and Haar), our approach appears to be independent of the choice of the wavelet basis function (Fig. 2 D).
Image correlation microscopy/spectroscopy (ICM) is another image analysis technique that has shown considerable utility in characterizing the distributional heterogeneity in static images (Petersen et al., 1993). This analysis involves the calculation of an autocorrelation function, G(
), such that
 | (8) |
represents a spatial shift. In other words, G() is the correlation between a function, f(x), and shifted versions of itself, as opposed to wavelet analysis, where a function is correlated with shifted and scaled versions of a wavelet basis function (Eq. 7). The autocorrelation function can be obtained by computing the inverse Fourier transform of the product of the Fourier transform of the signal and its complex conjugate (Press et al., 1992). A Gaussian function is then fitted to the normalized correlation data to estimate parameters that are used to calculate the density of fluorescent entities. Recently, ICM has been extended to include two dimensions and time-lapse sequences to map the molecular density, dynamics, and interactions of 
5-integrin in migrating cells (Wiseman et al., 2004). It would be of interest to compare the performance of ICM and wavelet-based analysis of FRET microscopy in future studies. Some immediate benefits of wavelet analysis include its multiresolution properties, the availability of commercially supported software, and its utilization in developing robust indices of spatial heterogeneity (Sharifi-Salamatian et al., 2004).
Both our wavelet-based approach and ICM are sensitive to the signal/noise ratio. Petersen et al. (1993) demonstrated that the density of fluorescent entities in an image calculated via ICM can result in erroneously high values in the presence of a low signal/noise ratio, and cautioned that careful controls are necessary in the presence of significant background fluorescence. To evaluate this issue with our wavelet-based method, five one-dimensional signals were generated that were 400 elements in length, where each value was randomly chosen between 0 and 1. We then added 0, 0.25, 0.5, 1, or 2 units to elements 197–203 in each signal, which were selected to represent a domain of interest. Wavelet analysis was conducted over 32 scales using the db4 wavelet basis function, and the difference matrices were computed with the zero-added signal serving as a control (see Wavelet Analysis in Materials and Methods). These matrices show that the domain was clearly identified for the 0.5-, 1-, and 2-unit added signals, but not the 0.25-unit added signal (see Fig. S1 in Supplementary Material). The 0.25-unit added signal was the only signal where the mean value of the domain (elements 197–203) did not significantly differ from that of the entire signal (Student's t-test; data not shown). Furthermore, as with the ICM method, a number of false-positives become evident as the signal/noise ratio decreases. However, our approach of conducting a statistical analysis of identified domains over time would serve to exclude the false-positives and retain domains with significantly changing corrected FRET intensities. The caveat of Petersen et al. (1993) is thus echoed here in that replications and/or measures over time with careful controls are required to ascertain the effects of noise.
In this study, the application of the CWT revealed a new aspect of PKC activation in real biological milieu. The transient increase in the FRET signal observed after 1 min of PDBu or ACh exposure (Figs. 2 B and 5 A) may be associated with a PKC-dependent activation of phosphatase, as it was hypothesized by Braz et al. (2004). In our control experiments, a pre-incubation of cells with a specific PKC inhibitor (Gö6983) or a phosphatase inhibitor (calyculin A) both completely eliminated this transient activation of phosphatase activity (Fig. 4 A). The lack of transient phosphatase activation has shifted the apparent balance in favor of PKC-induced phosphorylation and resulted in decrease of heterogeneity in plasma membrane domains. Note that overall, changes in FRETc over the plasma membrane were not significant, and only one domain showed a statistically significant decrease in FRETc values in the presence of calyculin A (Fig. 4 C, left panel), representing an increase in PKC-induced phosphorylation. The subsequent continuous stable decrease in FRETc reflects stimulation of PKC-dependent phosphorylation in membrane domains, which is sensitive to the PKC inhibitor Gö6983 (Fig. 4 A and Fig. 4 B, right panel) and highlights the complexity of the PKC signaling cascade.
In conclusion, we present a new strategy for the identification of functional signaling micro- and macrodomains of the plasma membrane by coupling FRET microscopy with wavelet analysis. This strategy includes: 1), redefining the plasma membrane into ROIs based on statistical analysis; 2), applying the CWT to FRET values contained within the linearized ROIs; and 3), identifying the heterogeneous regions from wavelet coefficient plots with consequent statistical analysis of corresponding regions in the plasma membrane. The implementation of this strategy facilitated the identification of plasma membrane domains of PKC signaling based on statistical analysis. These functional signaling domains putatively represent a dynamic balance of PKC phosphorylation and phosphatase activity. Furthermore, the balance in these domains is not necessarily represented by the average balance on the macrodomain scale, thereby offering a method to explore the importance of local heterogeneity in cell signaling in different experimental systems employing a variety of cell imaging techniques such as FRET, confocal, and fluorescence microscopy.
|
SUPPLEMENTARY MATERIAL |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONSUPPLEMENTARY MATERIALACKNOWLEDGEMENTSREFERENCES |
|---|
|
ACKNOWLEDGEMENTS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONSUPPLEMENTARY MATERIALACKNOWLEDGEMENTSREFERENCES |
|---|
This study was supported by the Intramural Research Program of the National Institute on Aging, National Institutes of Health. None of the authors declare any competing financial interests that could be perceived as influencing this research.
|
FOOTNOTES |
|---|
Donald E. Mager's present address is Dept. of Pharmaceutical Sciences, University at Buffalo, State University of New York, Buffalo, NY.
Submitted on October 5, 2004; accepted for publication February 11, 2005.
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONSUPPLEMENTARY MATERIALACKNOWLEDGEMENTSREFERENCES |
|---|
Arneodo, A., E. Bacry, P. V. Graves, and J. F. Muzy. 1995. Characterizing long-range correlations in DNA sequences from wavelet analysis. Phys. Rev. Lett. 74:3293–3296.[CrossRef][Medline]
Braz, J. C., K. Gregory, A. Pathak, W. Zhao, B. Sahin, R. Klevitsky, T. F. Kimball, J. N. Lorenz, A. C. Nairn, S. B. Liggett, I. Bodi, S. Wang, A. Schwartz, E. G. Lakatta, A. A. DePaoli-Roach, J. Robbins, T. E. Hewett, J. A. Bibb, M. V. Westfall, E. G. Kranias, and J. D. Molkentin. 2004. PKC- regulates cardiac contractility and propensity toward heart failure. Nat. Med. 10:248–254.[CrossRef][Medline]
Haralick, R. M., K. Shanmugam, and I. Dinstein. 1973. Textural features for image classification. IEEE Trans. 3:610–621.
Kenworthy, A. K. 2001. Imaging protein-protein interactions using fluorescence resonance energy transfer microscopy. Methods. 24:289–296.[CrossRef][Medline]
Kobrinsky, E., E. Schwartz, D. R. Abernethy, and N. M. Soldatov. 2003. Voltage-gated mobility of the Ca2+ channel cytoplasmic tails and its regulatory role. J. Biol. Chem. 278:5021–5028.
Kobrinsky, E., K. J. F. Kepplinger, A. Yu, J. B. Harry, H. Kahr, C. Romanin, D. R. Abernethy, and N. M. Soldatov. 2004. Voltage-gated rearrangements associated with differential ß-subunit modulation of the L-type Ca2+ channel inactivation. Biophys. J. 87:844–857.
Laine, A. F. 2000. Wavelets in temporal and spatial processing of biomedical images. Annu. Rev. Biomed. Eng. 02:511–550.[CrossRef]
Maxfield, F. R. 2002. Plasma membrane microdomains. Curr. Opin. Cell Biol. 14:483–487.[CrossRef][Medline]
Misiti, M., Y. Misiti, G. Oppenheim, and J. M. Poggi. 2000. Wavelet Toolbox User's Guide. The MathWorks, Natick, MA.
Murata, S., P. Herman, and J. R. Lakowicz. 2001. Texture analysis of fluorescence lifetime images of nuclear DNA with effect of fluorescence resonance energy transfer. Cytometry. 43:94–100.[CrossRef][Medline]
Murata, S., P. Herman, K. Mochizuki, T. Nakazawa, T. Kondo, N. Nakamura, J. R. Lakowicz, and R. Katoh. 2003. Spatial distribution analysis of AT- and GC-rich regions in nuclei using corrected fluorescence resonance energy transfer. J. Histochem. Cytochem. 51:951–958.
Petersen, N. O., P. L. Hoddelius, P. W. Wiseman, O. Seger, and K. E. Magnusson. 1993. Quantitation of membrane receptor distributions by image correlation spectroscopy: concept and application. Biophys. J. 65:1135–1146.
Press, W. H., S. A. Teukolsky, W. T. Vetterling, and B. P. Flannery. 1992. Numerical Recipes in FORTRAN, 2nd Ed. Cambridge University Press, New York.
Sharifi-Salamatian, V., B. Pesquet-Popescu, J. Simony-Lafontaine, and J. P. Rigaut. 2004. Index for spatial heterogeneity in breast cancer. J. Microsc. 216:110–122.[Medline]
Sharma, P., R. Varma, R. C. Sarasij, I. K. Gousset, G. Krishnamoorthy, M. Rao, and S. Mayor. 2004. Nanoscale organization of multiple GPI-anchored proteins in living cell membranes. Cell. 116:577–589.[CrossRef][Medline]
Shirai, Y., and N. Saito. 2002. Activation mechanisms of protein kinase C: maturation, catalytic activation, and targeting. J. Biochem.(Tokyo). 132:663–668.
Vermeer, J. E., E. B. Van Munster, N. O. Vischer, and T. W. Jr. Gadella. 2004. Probing plasma membrane microdomains in cowpea protoplasts using lipidated GFP-fusion proteins and multimode FRET microscopy. J. Microsc. 214:190–200.[Medline]
Violin, J. D., J. Zhang, R. Y. Tsien, and A. C. Newton. 2003. A genetically encoded fluorescent reporter reveals oscillatory phosphorylation by protein kinase C. J. Cell Biol. 161:899–909.
Wallrabe, H., M. Elangovan, A. Burchard, A. Periasamy, and M. Barroso. 2003. Confocal FRET microscopy to measure clustering of ligand-receptor complexes in endocytic membranes. Biophys. J. 85:559–571.
Wiseman, P. W., C. M. Brown, D. J. Webb, B. Hebert, N. L. Johnson, J. A. Squier, M. H. Ellisman, and A. F. Horwitz. 2004. Spatial mapping of integrin interactions and dynamics during cell migration by image correlation microscopy. J. Cell Sci. 117:5521–5534.
Xia, Z., and Y. Liu. 2001. Reliable and global measurement of fluorescence resonance energy transfer using fluorescent microscopes. Biophys. J. 81:2395–2402.
Zaccolo, M. 2004. Use of chimeric fluorescent proteins and fluorescence resonance energy transfer to monitor cellular responses. Circ. Res. 94:866–873.
Zacharias, D. A., J. D. Violin, A. C. Newton, and R. Y. Tsien. 2002. Partitioning of lipid-modified monomeric GFPs into membrane microdomains of live cells. Science. 296:913–916.
This article has been cited by other articles:
 |
|
 |
|
  D. E. Mager, E. Kobrinsky, A. Masoudieh, A. Maltsev, D. R. Abernethy, and N. M. Soldatov Analysis of Functional Signaling Domains from Fluorescence Imaging and the Two-Dimensional Continuous Wavelet Transform Biophys. J., October 15, 2007; 93(8): 2900 - 2910. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
