Refinements in design have simplified confocal microscopy to the extent that it has become a standard research tool in cell biology. However, as confocal microscopes have become more powerful, they have also become more demanding of their optical components. In fact, optical aberrations that cause subtle defects in image quality in widefield microscopy can have devastating effects in confocal microscopy. Unfortunately, the exacting optical requirements of confocal microscopy are often hidden by the optical system that guarantees a sharp image, even when the microscope is performing poorly. Optics manufacturers provide a wide range of microscope objectives, each designed for specific applications. This report demonstrates how the trade-offs involved in objective design can affect confocal microscopy.
Over the past ten years, confocal microscopy has developed from a technique limited to specialists in microscopy into a standard research tool. Its proliferation of applications results as much from the rapid technological developments in confocal microscopy as from the maturing user interface of commercial confocal microscope systems. The latest systems are nearly turnkey systems with which even novice microscopists can rapidly collect high-quality images. Ironically, the same technical developments that have stimulated the spread of confocal microscopy in experimental biology have also pushed the limits of the optics of confocal microscopes in a way that makes understanding the optical properties of a confocal microscope more important than ever.
The most common application of confocal microscopy is to compare the distributions or behaviors of multiple probes in the same cells. Such studies have been made possible by the development of confocal microscopes that are capable of efficiently collecting multiple colors of fluorescence and developing new dyes that have extended the useful spectrum of fluorescence microscopy. Depending on the configuration of the microscope, such studies may require optics that use wavelengths of light ranging from the ultraviolet through the infrared. The requirements for accurate color imaging have been further increased by the development of colorimetric methods of quantitative microscopy, such as fluorescence ratio measurements of ion concentrations.
Freedom from chromatic aberration is only part of the equation of optical design, which also considers monochromatic aberrations and parameters such as high photon efficiency, field size, field flatness, working distance, and the ability to image deep into aqueous biological tissues. Since the design of microscope optics reflects a compromise of these various parameters, manufacturers typically design a variety of different microscope objectives, with each representing a particular set of design trade-offs and each suited to a particular application. The studies discussed here demonstrate that the choice of microscope objective can have profound effects on the results of confocal microscopy experiments. They emphasize the importance of carefully choosing a microscope objective that is appropriate for the experimental application.
An ideal lens would focus all colors of light to the same point. In reality, all lenses have chromatic aberration, a property in which different colors of light are focused to different points. When observing a sample through the microscope eyepiece, this defect makes objects appear to have colored edges. When imaging a sample in color confocal microscopy, this defect results in different colors of excitation illumination being focused to different points in the sample and different colors of emission being collected from different points in the sample. Horizontal displacements in the image plane, termed lateral chromatic aberration, result in different magnification of different colors. This problem can be minimized by restricting the analysis to the center of the microscope field. However, vertical color displacements along the focal axis, referred to as axial chromatic aberration, are present throughout the microscope field. This is clearly a problem for any researcher trying to use color confocal microscopy to determine the relative distribution of multiple probes. The images presented in Figures 1 through 3 show how the results of color confocal imaging critically depend on the nature of the microscope objective.
The figures provide a comparison of the performance of a plan fluor 40x objective, which is designed for maximum UV light transmission, with that of a plan apochromat 100x objective, designed with minimal chromatic aberration. For one aspect of the comparison, a vertical series of images of the reflection from the face of a glass coverslip was collected, using either 488 or 647-nanometer light. As the different colors of light are reflected by a single surface, a lens free of chromatic aberration would focus the different colors to the same focal plane. When reproduced as a vertical cross-section, images collected with an ideal lens would show a single horizontal line in which the two colors overlap completely. The images in the top half of Figure 1(a) illustrate xz-sections of the glass reflection, with the focal (z) axis oriented vertically, and demonstrate that the plan apochromat 100x objective does a credible job of approaching this ideal, resolving the two colors of light to a depth within 0.1 micrometer of one another. In contrast, the minimally corrected 40x plan fluor objective detects the reflection of 647-nanometer light approximately 1.2 micrometers above the 488-nanometer light. In the images of the glass reflection, 647-nanometer light is shown in red, and 488-nanometer light is shown in blue. The scale bar indicates a distance of 1 micrometer.
The effect of this vertical discrepancy is apparent in fluorescence. The bottom half of this panel (Figure 1(a)) shows vertical cross-sections of image volumes of beads labeled with three fluorophores. When imaged with the 100x plan apochromat objective, the three colors are coincident, resulting in a white, reasonably round image. In contrast, the 40x objective generates an image in which the far-red fluorescence (680-nanometer emissions, depicted in blue) is distinctly vertically displaced from that of either the red (600 nanometer) or the green (520 nanometer) fluorescence. In addition, a slight horizontal displacement of the far-red fluorescence reflects lateral chromatic aberration.
To provide a biological test specimen, advantage was taken of the fact that endosomes can be labeled by incubating cells with fluorescently tagged endocytic ligands. The large number of molecules included in each endosome ensures that each will contain the same ratio of fluorescent probes. Thus, endosomes provide an excellent test of color confocal imaging as they present subresolution objects in biological samples with color that will reflect the relative fluorescent contributions of each fluorophore. Figure 1(b) shows a color image of cells in which the endosomes have been labeled with both fluoresceintransferrin (F-Tf, which fluoresces green) and Cy5-transferrin (Cy5-Tf, which fluoresces far-red), collected using the 100x plan apochromat objective. The co-localization of the two probes is apparent in the constant yellow-orange colors of the individual endosomes and is more apparent in comparisons of the high magnification images of F-Tf (Figure 1(c)) and Cy5-Tf (Figure 1(d)). In these images (Figure 1(b) through Figure 1(d)), the scale bar represents a length of 10 micrometers.
When switching to the 40x plan fluor objective, however, slight differences in the focal plane are apparent even between the red and green fluorescence, as shown in cells labeled with Tf conjugated to both fluorescein and rhodamine (Figure 2(a)). The differences in focal planes between far-red and green fluorescence result in startling differences in the apparent distributions of F-Tf and Cy5-Tf, which now appear to be completely distinct (Figure 2(b)).
For a cell biologist who is assessing the relative distributions of various probes, these displacements would have disastrous consequences. The discrepancy in focal plane can be circumvented by summing the entire vertical series of the image volume into a single projection, which removes the effect of differences in the focal plane of the two colors (Figure 2(c)). The consistent yellow color of every endosome in the projection demonstrates the constant ratio of the two probes in each endosome. However, as this procedure discards all vertical information, it is seldom an appropriate way of presenting confocal images. The effects of axial chromatic aberration can also be minimized by measuring the axial offset of different colors and combining the images collected from the focal planes appropriate for each color. An example of this procedure is shown in Figure 2(d), which shows the combined green and far-red images collected 1.2 micrometers apart with the 40x plan fluor objective. The scale bar in the images of Figure 2 represents a 10-micrometer length. The correction achieved by combining images from different focal planes is apparent when comparing the images of the individual probes, shown in Figure 3. Whereas Figure 3(a) and Figure 3(b) show poor agreement in distribution between F-Tf and Cy5-Tf in images collected at the same focal plane, a comparison of Figure 3(b) and Figure 3(c) shows that the far-red image can be superimposed with the green fluorescence image collected 1.2 micrometers deeper. In these images (Figure 3(a) through Figure 3(c)), the scale bar corresponds to a length of 5 micrometers.
While chromatic aberration results in different colors of light being focused to different points in the image volume, spherical aberration can profoundly decrease the signal in a confocal microscope. A spherically aberrated lens focuses axial and peripheral rays to different points, thus blurring the image of a point source of light. In the same way that the confocal pinhole so effectively improves image contrast by rejecting out-of-focus light, it effectively eliminates much of the fluorescence of an object imaged with spherical aberration.
For many samples, the primary source of spherical aberration derives from the difference between the refractive index of the immersion medium and the mounting medium. Until recently, the highest-resolution, best-corrected microscope objectives were designed for use with oil as an immersion fluid. For these objectives, spherical aberration is minimized only when the entire light path has the refractive index of immersion oil (which is the same as that of glass) and accumulates with distance into a medium with a different refractive index. Since most samples — particularly living samples — are mounted in media with a refractive index that is significantly lower than immersion oil, spherical aberration has thus limited the depth of image volumes using oil immersion objectives.
Figure 4 shows the effects of spherical aberration in confocal microscopy, where the oil immersion 100x plan apochromat objective was used to collect an image of cells labeled with F-Tf and mounted at a depth of either 0 micrometers (the surface of the coverslip) (Figure 4(a)) or 35 micrometers into an aqueous medium (Figure 4(b)). In both cases, the endosomes appear sharply defined, but the accumulated spherical aberration of a light path 35 micrometers into an aqueous medium has profoundly compromised the fluorescent signal in Figure 4(b). The scale bar represents a length of 10 micrometers in each of the images presented in Figure 4.
Recently, optical manufacturers have addressed this problem by designing objectives that use water as the objective immersion medium. For aqueous samples, the matched refractive index of immersion and sample media makes spherical aberration independent of imaging depth. This enables these objectives to collect images as deeply as the working distance permits, frequently hundreds of micrometers into a sample. The success of this design is shown in Figure 4 ((c) and (d)) , which shows images of cells labeled with F-Tf, collected either zero (Figure 4(c)) or 66 micrometers into an aqueous medium (Figure 4(d)) using a water immersion 60x plan apochromat objective. Here, the fluorescent signal is unaffected by the light path through the aqueous sample medium. This new generation of high numerical aperture, plan apochromat water immersion objectives has significantly helped confocal microscopy realize its potential in three-dimensional biological imaging.
The color performance of this water immersion objective falls between that of the oil immersion plan fluor and the plan apochromat objectives discussed previously. The cross-sectional images of the reflection of a glass surface in the top half of Figure 5(a) show that 647-nanometer light is focused approximately 0.6 micrometer above 488-nanometer light. Similar patterns for volumes collected either 0 or 63 micrometers into an aqueous medium show that this discrepancy is independent of imaging depth.
The bottom-left image of those in Figure 5(a) shows that this color discrepancy results in the far-red image of a triple-labeled bead being displaced from that of either the red or green fluorescence images. In the fluorescent bead images, 520-nanometer emissions are shown in green, 600-nanometer emissions in red, and 680-nanometer (far-red) emissions in blue. The bottom-right portion of this panel demonstrates the critical importance of coverslip thickness correction in minimizing spherical aberration in this kind of objective. Since the correction for spherical aberration depends on the length of the optical path through the coverslip, it is adjustable by a collar that is set according to the thickness of the coverslip. While the bead cross-section on the left was collected with the collar set to the measured coverslip thickness (174 micrometers), the image on the right was collected with the collar misadjusted to 150 micrometers. The spherical aberration that results from an improper collar setting severely attenuates the fluorescence signal and compromises vertical resolution. Although this misadjustment was chosen to dramatize the point, we emphasize that such an error is easily encountered in practice because the actual thickness of coverslips can vary by more than 40 micrometers around their nominal value. Proper collar setting can only be ensured through measurements of individual coverslips. In all images in Figure 5(a), the focal axis is oriented vertically, and the scale bar indicates a distance of 1 micrometer.
The images shown in Figure 5(b) and Figure 5(c) are of a field of cells labeled with both F-Tf and Cy5-Tf, and were collected at a depth of 63 micrometers into an aqueous buffer. The scale bar represents 10 micrometers in these two images. Although the images appear sharp, the axial chromatic aberration of this objective has the effect of making the distributions of F-Tf and Cy5-Tf appear discrete (Figure 5(b)). Nonetheless, the projection of the vertical series of images of these cells (Figure 5(c)) shows that the two probes equally label all endosomes. In agreement with the reflection images shown in Figure 5(a), axial chromatic aberration is similarly apparent in images of endosomes collected at a depth of zero micrometers (at the coverslip surface). Little difference is observed in the focal plane of red and green fluorescence in images of endosomes labeled with Tf conjugated to both fluorescein and rhodamine (Figure 6(a)), but F-Tf and Cy5-Tf again appear to label discrete populations of endosomes (Figure 6(b)).
The distribution of the two probes can be better compared when allowing for the difference in focal plane and combining the Cy5-Tf image with a fluorescein image collected 0.6 micrometer deeper. The co-localization of the two probes is now apparent in the constant yellow color of the endosomes shown in Figure 6(c). Apparent differences in the distribution of the two probes shown in images collected at a single focal plane (Figure 6(d) and Figure 6(e)) disappear when the Cy5-Tf image is compared with the F-Tf image collected 0.6 micrometer deeper (Figure 6(f)). The scale bar represents 10 micrometers length in Figure 6(a) through Figure 6(c), and corresponds to 5 micrometers in Figure 6(d) through Figure 6(f).
The effects of chromatic aberration can also be quantified by measuring fluorescence ratios in images of endosomes labeled with multiple probes. Semi-logarithmic plots of ratio histograms are presented in Figure 7(a) for cells labeled with F-R-Tf, with images collected using the 100x plan apochromat (red curve), the 40x plan fluor (blue curve), and the 60x plan apochromat water immersion objective (green curve). Figure 7(b) displays ratio histograms of cells labeled with F-Tf and Cy5-Tf using the same three objectives. Figure 7(a) shows that the ratio of rhodamine to fluorescein (red to green) emissions is reasonably constant for all three objectives. In contrast, Figure 7(b) shows that while the 100x plan apochromat still shows minimal variation in the ratio of Cy5 to fluorescein (far-red to green), the water immersion 60x plan apochromat shows more variation, and the 40x plan fluor shows even more. The effects of chromatic aberration in the 60x plan apochromat and 40x plan fluor are more apparent when the distributions for a single focal plane are compared to those measured in the projections of the vertical image series for each objective (Figure 7(c) and Figure 7(d), respectively). In each case, the narrow distribution of fluorescence ratios in the projected images reports the nearly constant ratio of the two probes in the endosomes, which is misrepresented in any single focal plane image.
The studies presented here demonstrate vividly how chromatic and spherical aberrations compromise the performance of confocal imaging. At the same time, they emphasize the critical importance of objective choice in confocal microscopy. Poor color correction will lead to erroneous interpretations of the relative distributions of multiple probes, one of the major applications of confocal microscopy. The ratio quantifications also demonstrate how chromatic aberrations compromise microscope quantitation. Spherical aberration resulting from mismatched immersion and sample media worsens vertical resolution and can obliterate fluorescence detection altogether. Although particularly apparent in the high-contrast images provided by confocal microscopy, these errors also compromise images collected by conventional epifluorescence and transillumination techniques.
Chromatic aberration has been characterized primarily with respect to UV fluorophores. Other results demonstrate significant axial chromatic aberration in the far-red as well, a range increasingly used by microscopists with the development of new far-red fluorescent probes and lasers capable of exciting them. Although the studies presented here involve characterization of point sources, these samples readily display problems that would be manifest (but not necessarily obvious) with more extensive probes. Such probes include the cytosolic dyes that are frequently used in fluorescence ratio measurements of ionic concentrations.
For example, fluorescence of the cytosolic pH indicator SNARF-1 is excited by 488-nanometer light, and pH is measured from the ratio of fluorescence at 580 nanometers to that at 640 nanometers. For measurements of the cytosolic pH of thin cells, it is easy to imagine how the fluorescence ratio can be influenced by the position of the cell relative to the separate focal planes of the green excitation light, the red emissions and the far-red emissions. Although these errors will frequently simply add variation to measurements (doubling or tripling standard deviations in the presented quantifications), they can also systematically affect ratio measurements, for example, when comparing ratios in different parts of cells of varying thickness.
A final consequence of chromatic aberration, although not discussed here, is how it affects fluorescence detection. In a confocal system with chromatic aberration, the difference between the excitation and emission wavelengths results in the excitation of one volume and imaging of a different volume, attenuating the fluorescence signal. For beam-scanning systems (most confocal microscopes), this loss of signal worsens with distance off axis as the imaged spot is focused farther and farther from the confocal pinhole.
It must be emphasized that these observations are not unique to the particular objectives used in the data presented, or to a particular manufacturer. The characterizations of the color correction were identically reproduced in three examples of the 60x water immersion objective and two examples of the 100x oil immersion objective on both Quantum and Diaphot microscope stands. Significant axial chromatic aberration has been identified in objectives from every major manufacturer. Indeed, the ubiquity of chromatic aberration has led to the development of alternative microscope designs that address chromatic aberration either by imposing special auxiliary correcting lenses or avoiding refraction altogether and turning to reflecting objectives. Nonetheless, microscope optics is likely to continue to evolve, and preliminary experience with the new generation of CFI60 optics from Nikon, Inc. indicates that axial chromatic correction is much improved.
Superficially, it may appear that some objectives are simply better than others. However, as discussed previously, objective design represents a compromise of design parameters. As such, each objective reflects a different set of design compromises according to its intended application. Nonetheless, it is possible that researchers will require more than what is currently available in optical design, forcing them to design experiments that do not stretch the limits of objective design.
If an oil immersion objective must be used to image a sample in an aqueous medium — for example, in studies of living cells — spherical aberration can be minimized either by minimizing the optical path through the aqueous medium or by using an oil with a refractive index tailored to the spherical aberration induced by the aqueous optical path.
The problem of chromatic aberration can be minimized in several ways. If a particular experiment requires the use of an objective with significant chromatic aberration, the most obvious solution is to simply avoid the use of far-red fluorescing dyes such as Cy5 altogether. The results discussed in this review demonstrate satisfactory agreement between green and red fluorescence with all the tested objectives. In general, it is best to use dyes with maximum excitation and emission near the specific wavelengths for which a particular objective has been corrected. A second solution, discussed above, is to collect images of each fluorophore in a vertical series of focal planes and combine the different color images according to the difference in focal plane between the two. This is a simple solution, but one that will not be appropriate when rapid image acquisition is necessary, as with live cells. Analyses not shown here also indicate that this solution does not have sufficient accuracy to correct images intended for ratio quantification. A more expensive (or less widely accessible) solution is to use two-photon microscopy, which is inherently unaffected by chromatic aberration as long as no detector pinhole is used. However, its multicolor capability is only now being explored.
Kenneth W. Dunn and Exing Wang - Department of Medicine, Indiana University, School of Medicine, 1120 South Drive, FH115, Indianapolis, Indiana 46202-5116.
This article originally appeared in BioTechniques: Vol. 28, 542 (March 2000).