| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
PERSPECTIVES IN CELL PHYSIOLOGY
1Department of Cellular and Physiological Sciences, University of British Columbia, Life Sciences Institute, Vancouver, British Columbia, Canada; and 2Department of Physiology, University of Arizona, Arizona Health Sciences Center, Tucson, Arizona
Submitted 29 February 2008 ; accepted in final form 17 March 2008
ABSTRACT
Colocalization, in which images of two or more fluorescent markers are overlaid, and coincidence between the probes is measured or displayed, is a common analytical tool in cell biology. Interpreting the images and the meaning of this identified coincidence is difficult in the absence of basic information about the acquisition parameters. In this commentary, we highlight important factors in the acquisition of images used to demonstrate colocalization, and we discuss the minimum information that authors should include in a manuscript so that a reader can interpret both the fluorescent images and any observed colocalization.
The Nyquist Criteria
When the term colocalization is used in optical microscopy, the objects that are being compared are usually pixels or voxels in digital images (A pixel is the smallest element in a two-dimensional digital image and a voxel is its three-dimensional counterpart). When we ask if two molecules are colocalized, we are, in effect, asking "Is the voxel illuminated by a fluorophore attached to (or part of) molecule A the same voxel as that illuminated by another fluorophore attached to molecule B?" Thus the size of the voxels is important for both planning and interpreting the experiments. When the image is digitized (voxelized), it has the same effect as placing a grid over the image and replacing the objects at each grid point (voxel) with the total intensity of all of the objects contained in it. As voxel size increases, resolution decreases so that separate structures become lumped together in the same voxel. At some point, the voxelized image will no longer accurately represent the original object and artifacts may appear. The voxel size at which an object is accurately represented by an image is given by the Nyquist criteria, which are a function of several variables: the numerical aperture of the objective, the emission wavelength of the fluorophore, the type of microscope (wide field or confocal) and, for depth, the refractive index of the medium. Further details can be found in Castleman (2) and van der Voort and Strasters (13). The formulas and a summary of appropriate values for a conventional microscope, using oil immersion lenses of various numerical apertures and illuminating wavelengths, are given in Table 1. Confocal microscopes, using laser light and a typical pinhole size of 
250 nm, have Nyquist values that are similar to those listed (13). Ideally, imaging parameters should be optimized to reach these values to obtain the best resolution for evaluating colocalization. In real systems, noise will increase the theoretical values listed in the table, and values within 10% of those tabulated will still give an acceptable result.
|
|
Diffuse Labeling
In some cases, one of the fluorophores is present in almost every voxel. This can occur when labeling a cytosolic protein or a widely distributed protein like actin. Under these conditions, colocalization by chance has almost 100% probability, making any observed or measured colocalization, not significant. Experimental approaches other than colocalization by optical microscopy are required to resolve this problem.
Deconvolution
Wide-field images contain much out-of-focus light, which will confound colocalization analysis. For this reason, it is mandatory that wide-field images be deconvolved before colocalization analysis is attempted. Authors should state which algorithm or software package was used. It should be noted that the measurement of colocalization in confocal images is also improved by deconvolution (7, 11).
Bleed Through
A fundamental assumption made when measuring colocalization is that the signal at each wavelength is independent of the other. Bleed through of signal from one channel to another can result in an apparent colocalization when none, in fact, exists. For example, even with modern filters, there is significant bleed through from FITC to rhodamine, or from cyan fluorescent protein to yellow fluorescent protein. Bleed through should be estimated by creating a control with labeling in one channel and none in the other and then recording the signal coming from the unlabeled wavelength(s). The resultant image should be random, having no discernable pattern, and its intensity should be less than the chosen threshold. If, however, there is significant bleed through, the control images can be used in concert with spectral linear unmixing (8, 14) to separate the two signals. Once corrected, the images can then be used for analysis. Some newer instruments have spectral unmixing built into their software and correct for bleed through at the time of acquisition, although it is important to check how well the signals are separated before accepting the results (15).
Colocalization in X-Y versus Z
In all optical microscopes, except the 4Pi, the resolving power in Z is only one-third to one-quarter of that in X and Y (9). The consequence is that colocalization is most powerful when performed on structures that run along the XY plane, parallel to the coverslip, whereas it is least accurate for objects that are running parallel to the Z-axis of the microscope. The consequences of this are best seen in a structure that runs both parallel and perpendicular to the XY plane. This is diagrammed in Fig. 2A, which shows a cross-section through a cell in which fluorophores of the same color are separated by 200 nm, and the two different fluorophores are interspersed at 100-nm intervals. Figure 2B shows the result of superimposing these planes–the pixels that lie in the transverse plane are not colocalized, but those at the edges, lying along the Z-axis appear to be. This effect can be observed in real cells. Figure 2C shows a rat ventricular myocyte labeled with antibodies specific for vinculin (green) and caveolin-3 (red). A 1-µm thick section from the cell's center is displayed in Fig. 2Ci and demonstrates almost complete colocalization at the cell surface running parallel to the Z-axis (arrowheads). Figure 2Cii is an image of the top surface of the same cell running parallel to the XY axis, which shows <5% colocalization (boxed regions). A stereo pair of the complete data stacks (Fig. 2Ciii) shows that colocalization is confined to those regions of the sarcolemma that are oriented largely parallel to the microscope's Z-axis, but these molecules are, in fact, not colocalized.
|
Sampling the Cell
The great advantage of using optical microscopy to assess colocalization, as opposed to other techniques including electron microscopy, is that images of entire cells or their substructures can be readily collected and analyzed, and regional variations can be visualized. When focusing an analysis on specific regions of a cell, care needs to be taken when deciding which part of the cell is sampled, because this can dramatically affect the results. Specifically, two recent papers (3, 10) have shown that regional variations in colocalization can occur in the absence of any obvious structural differences.
Guidelines for Publication
To enable readers to interpret images and to evaluate the meaning of colocalization from a given experiment, manuscripts should include the following information:
In addition, the following conditions should be satisfied.
Conclusion
Colocalization can be a powerful method for determining whether molecules have the potential to associate or interact, but it cannot confirm that they do. The inherent limitations of the technique mean that confirmation may require electron microscopy (1, 5) or other microscopic techniques such as fluorescence resonance energy transfer (6, 12).
GRANTS
This work was funded by the American Heart Association GIA 93014700 and National Heart, Lung, and Blood Institute Grant HL-66044 (to R. M. Lynch), a Grant-in-Aid from the Heart and Stroke Foundation of British Columbia and the Yukon, and by grants from the National Sciences and Engineering Research Council and the Canadian Institutes of Health Research (to E. D. W. Moore).
FOOTNOTES
Address for reprint requests and other correspondence: E. D. W. Moore, Dept. of Cellular and Physiological Sciences, Univ. of British Columbia, Life Sciences Institute, 2350 Health Sciences Mall, Vancouver, British Columbia V6T 1Z3, Canada (e-mail: edmoore{at}interchange.ubc.ca)
The costs of publication of this article were defrayed in part by the payment of page charges. The article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
REFERENCES
1. Anderson JB, Carol AA, Brown VK, Anderson LE. A quantitative method for assessing co-localization in immunolabeled thin section electron micrographs. J Struct Biol 143: 95–106, 2003.[CrossRef][Web of Science][Medline]
2. Castleman KR. Resolution and sampling requirements for digital image processing, analysis, and display. In: Electronic Light Microscopy, edited by Shotton D. New York: Wiley-Liss, 1993, p. 71–93.
3. Dan P, Lin E, Huang J, Biln P, Tibbits GF. Three-dimensional distribution of cardiac Na+-Ca2+ exchanger and ryanodine receptor during development. Biophys J 93: 2504–2518, 2007.[CrossRef][Web of Science][Medline]
4. Gonzalez RC, Woods RE. Digital Image Processing. New York: Addison-Wesley, 1992.
5. Griffiths G, Parton RG, Lucocq J, van Deurs B, Brown D, Slot JW, Geuze HJ. The immunofluorescent era of membrane traffic. Trends Cell Biol 3: 214–219, 1993.[CrossRef][Medline]
6. Jares-Erijman EA, Jovin TM. FRET imaging. Nat Biotechnol 21: 1387–1395, 2003.[CrossRef][Web of Science][Medline]
7. Landmann L. Deconvolution improves colocalization analysis of multiple fluorochromes in 3D confocal data sets more than filtering techniques. J Microsc 208: 134–147, 2002.[Web of Science][Medline]
8. Nadrigny F, Rivals I, Hirrlinger PG, Koulakoff A, Personnaz L, Vernet M, Allioux M, Chaumeil M, Ropert N, Giaume C, Kirchhoff F, Oheim M. Detecting fluorescent protein expression and co-localisation on single secretory vesicles with linear spectral unmixing. Eur Biophys J 35: 533–547, 2006.[CrossRef][Web of Science][Medline]
9. Schrader M, Bahlmann K, Giese G, Hell SW. 4Pi-confocal imaging in fixed biological specimens. Biophys J 75: 1659–1668, 1998.[Medline]
10. Scriven DR, Klimek A, Asghari P, Bellve K, Moore ED. Caveolin-3 is adjacent to a group of extradyadic ryanodine receptors. Biophys J 89: 1893–1901, 2005.[CrossRef][Web of Science][Medline]
11. Sedarat F, Lin E, Moore ED, Tibbits GF. Deconvolution of Confocal Images of Dihydropyridine and Ryanodine Receptors in Developing Cardiomyocytes. J Appl Physiol 97: 1098–1103, 2004.
12. Sekar RB, Periasamy A. Fluorescence resonance energy transfer (FRET) microscopy imaging of live cell protein localizations. J Cell Biol 160: 629–633, 2003.
13. van der Voort HTM, Strasters KC. Restoration of confocal images for quantitative image analysis. J Microsc 178: 165–181, 1995.[Web of Science]
14. Zimmermann T, Rietdorf J, Pepperkok R. Spectral imaging and its applications in live cell microscopy. FEBS Lett 546: 87–92, 2003.[CrossRef][Web of Science][Medline]
15. Zucker RM, Lerner JM. Wavelength and alignment tests for confocal spectral imaging systems. Microsc Res Tech 68: 307–319, 2005.[CrossRef][Web of Science][Medline]
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| Visit Other APS Journals Online |
