The wide spectrum of fluorescent proteins and derivatives uncovered thus far are quite versatile and have been successfully employed in almost every biological discipline from microbiology to systems physiology. These unique probes have proven extremely useful as reporters for gene expression studies in both cultured cells and entire animals. In living cells, fluorescent proteins are most commonly utilized to track the localization and dynamics of proteins, organelles, and other cellular compartments, as well as a tracer of intracellular protein trafficking. Quantitative imaging of fluorescent proteins is readily accomplished with a variety of techniques, including widefield, confocal, and multiphoton microscopy, to provide a unique window for exposing the intricacies of cellular structure and function.
The complex spectral and physical properties of fluorescent proteins affect the accuracy and utility of any quantitative measurement. Many of these properties, such as the molar extinction coefficient, quantum yield, photobleaching rate, and pH dependence on spectral profiles, can be easily measured with purified proteins in vitro. However, other important and experimentally critical properties, including the time course of chromophore formation (maturation) and protein degradation rates in vivo, are more difficult to ascertain. For typical experiments, the brightest, most stable monomeric fluorescent protein derivative should be selected to reduce the necessity for applying complicated background and photobleaching corrections in less demanding applications.
In selecting fluorescent protein vectors for imaging experiments, the commercially available enhanced variants from jellyfish and reef corals, as well as their derivatives should be seriously considered. These are available with fluorescence profiles in the cyan (ECFP), green (EGFP), and yellow (EYFP) spectral regions, and new monomeric red-emitting fluorescent proteins have now been introduced. Illustrated in Figure 1 are a selection of images acquired using a variety of commercially available fluorescent protein fusion vectors targeting several different sub-cellular locations. The fluorescence emission spectral profiles for the probes depicted in Figure 1 cover a bandwidth of almost 180 nanometers, featuring maxima that range from 440 to 618 nanometers. These probes are the brightest, most stable fluorescent protein variants and permit imaging with illumination at low light levels, which minimizes problems such as photobleaching (discussed below) and phototoxicity. Many of the fluorescent proteins variants (including EGFP, ECFP, and EYFP) can form dimers at sufficiently high concentrations, an artifact that can perturb membrane structure or lead to incorrect assumptions when conducting quantitative advanced fluorescence techniques, such as resonance energy transfer (FRET).
In addition to the intrinsic brightness of genetically engineered fluorescent proteins, the expression level is another factor to consider in producing a significantly bright intracellular signal from a fusion construct. Use of vectors with strong promoters to enhance transcription levels and the application of proper codons to optimize translation are essential for increasing expression levels of the fusion protein, thereby enhancing the overall signal level. This is particularly important when cellular autofluorescence makes it difficult to distinguish fluorescent fusion protein emission from background fluorescence.
A variety of techniques can be applied to enhance the expression level and brightness of cells harboring a fluorescent protein fusion product. Addition of sodium butyrate (approximately 1 to 5 millimolar) to the culture medium will increase the overall gene expression levels in stable cells lines expressing a fusion protein. In addition, the use of transiently transfected cells is also an attractive option, because they often display much higher levels of expression than stably transfected cells. A degree of caution should be exercised with transient transfections, however, due to possible over expression artifacts, including protein aggregation or saturation of protein targeting machinery, leading to inappropriate localization. Finally, increasing the number of tandem fluorescent protein sequences (double or triple) in the cloning vector is an alternative method for increasing fusion protein brightness levels.
Critical Fluorescent Protein Properties
Most of the commercially available fluorescent protein vectors feature improved performance due to optimization of codon usage for translation in mammalian cells and site-directed mutagenesis to increase the efficiency of chromophore maturation at higher temperatures. For the more demanding imaging applications in which the expression or target abundance levels of fluorescent proteins are low, the intrinsic properties of the probes will likely be a limiting factor. It is therefore necessary to fully understand the inherent properties of fluorescent proteins and apply the necessary controls to minimize artifacts that may arise during imaging of the probes in live cell experiments.
Photon-induced destruction of the chromophore, a phenomenon known as photobleaching, in fluorescent proteins is typically reduced in speed and magnitude when compared to traditional synthetic fluorophores, such as fluorescein and rhodamine under similar conditions. This resistance to photobleaching is thought to arise from the protection of the fluorescent protein chromophore by the tightly packed surrounding beta-can protein structure. Regardless, it is still important to rigorously perform bleaching control measures in quantitative imaging experiments.
The rate of photobleaching in fluorescent proteins can be determined by acquiring a time-lapse image series on labeled control cells, followed by quantitatively measuring the gradual loss in emission intensity, which is usually manifested by an exponential decay. The photobleaching fluorescence loss curve (as illustrated in Figure 2) can then be employed to correct the imaging experiments. Although not generally a significant problem, if photobleaching becomes the limiting factor when imaging living cells, antifade reagents such as ascorbic acid, Trolox, or Oxyrase (oxygen and free radical scavengers) can be added to the culture medium.
Although photobleaching is often the ultimate limiting factor in fluorescence microscopy, the photobleaching rate of many fluorescent protein derivatives is relatively slow (see Figure 2). Another limiting factor is fluorophore saturation, which does not occur in widefield microscopy but can be a significant problem with laser scanning confocal instruments where excitation intensities high enough to reach saturation are sometimes employed. Fluorophore saturation occurs when all of the available molecules are constantly in the excited state, so that an increase in excitation light does not produce a corresponding increase in fluorescence emission. In this case, it is very difficult to calibrate fluorescence signals due to non-linearity, a problem that can only be corrected by lowering the laser input power.
Intracellular environmental conditions, especially pH and high localized concentrations of other monovalent and divalent cations, can alter the brightness level of fluorescent protein derivatives. For example, the wild-type green fluorescent protein (wtGFP) displays relatively even brightness from pH 5 to pH 10, whereas enhanced green and yellow fluorescent proteins are both quenched at acidic pH levels, reducing their effectiveness when targeted at organelles (lysosomes) or sub-cellular compartments having low pH interiors. Other derivatives, such as ECFP and DsRedFP are less sensitive to low pH. However, many fluorescent protein fusion products will be localized in regions near physiological pH, which should not seriously affect fluorescence intensity.
Among the other physical parameters inherent to fluorescent proteins that may dramatically alter quantitative imaging (but cannot be accurately measured in vitro), the chromophore folding kinetics, or turn-on time can have the most serious consequences. Derivatives of the Aequorea victoria jellyfish (and most reef coral) fluorescent proteins require oxidation for chromophore formation, but oxygen does not readily penetrate the dense protein structure surrounding the tripeptide fluorophore. Therefore, a considerable delay (often exceeding several hours) can occur between protein expression and the appearance of fluorescence. Unfortunately, it is not often easy to distinguish between delays due to fluorescent protein expression (the ideal measurement) and those due to slow chromophore formation (an artifact). The reverse process, in vivo degradation of mature fluorescent proteins, is also very difficult to monitor, but overwhelming evidence suggests that many of these probes are very stable.
Widefield, confocal, and multiphoton fluorescence microscopes, either in the upright or inverted configuration, are the ideal instruments for observing fluorescent proteins in living cells and tissues. The required illumination sources range from mercury and xenon arc-discharge lamps for widefield, to gas and semiconductor continuous wave lasers in confocal and mode-locked pulsed lasers for multiphoton microscopy (see Table 1). Green fluorescent protein and its variants are readily imaged with the bright blue spectral lines of a mercury lamp or the 488-nanometer line emitted by an argon-ion or krypton-argon gas laser. In addition, a host of new solid-state lasers that produce lines in many portions of the visible spectrum, including the violet and blue regions, are being introduced. Fluorescent protein variants having spectral profiles shifted to the blue, red, and near-infrared wavelengths are often imaged with more efficiency using xenon or metal-halide lamps, as well as lasers emitting lines closely matching the excitation maxima. Regardless of the light source, the primary issue for imaging fluorescent proteins is that enough excitation light is available to obtain reasonable signal levels.
Aside from objectives, which are discussed below, the two most critical components for imaging fluorescent proteins are the emission filter and the detector. In widefield microscopes, the excitation filter, dichromatic mirror, and emission filter are combined into an optical block with the bandwidth of the excitation filter being determined by the light source used for excitation. For example, a mercury lamp requires an excitation filter bandpass having high transmission between 450 and 490 nanometers to image green fluorescent protein. Shorter wavelength bands are necessary for cyan and blue variants, whereas longer wavelength bands are used for yellow, orange, and red derivatives. Laser lines have bandwidths only a few nanometers in size, and therefore, do not normally require the use of wavelength selection with interference filters. The dichromatic mirror cut-on wavelength should clearly separate the excitation and emission spectral profiles by efficiently reflecting the former and transmitting the latter. There is very little margin for error in selection of suitable dichromatic mirrors. Emission filters, on the other hand, can be fine-tuned to pass bandwidths varying from 20 to hundreds of nanometers, depending upon the experimental requirements. These filters are the most flexible in terms of determining the level of fluorescence emission intensity passed to the detector. Catalogued in Table 1 are a list of suggested filter combinations for quantitative imaging of fluorescent proteins. These filter recommendations, which include only bandpass emission filters (in effect, neglecting longpass filter suggestions), should be regarded merely as a starting point for designing experiments.
Performance of several popular green fluorescent protein filter sets can be judged by comparing images from the same viewfield captured with each of the individual filter combinations, as illustrated in Figure 3. The specimen is an adherent culture of embryonic rat thoracic aorta myoblasts that were transfected with a vector containing a fusion protein combining the gene for human cytoplasmic beta-actin with a red-shifted (enhanced) green fluorescent protein variant, EGFP. Expression of the fusion protein is manifested by its incorporation into growing actin filaments that enable visualization of the subcellular structures using fluorescence microscopy with the appropriate filter combinations (see Figure 3). The thoracic aorta myoblasts were also stained with MitoTracker Red CMXRos (mitochondria; red fluorescence) and Hoechst 33258 (DNA in the nucleus; blue fluorescence). Note the absence of signal from the blue and red fluorophores with the Piston and Endow bandpass filters (Figure 3(a) and 3(b)), but the significant degree of orange-red fluorescence produced by the mitochondrial probe with the Endow longpass filter (Figure 3(c)). The bandpass and longpass GFP filter combinations, having bandwidths of 30 nanometers (Piston), 50 nanometers (Endow BP), and more than 100 nanometers (Endow LP), are also useful with a wide variety of synthetic fluorophores that are excited by light in the blue portion of the visible spectrum.
Although green fluorescent protein can generally be imaged with a standard fluorescein filter combination if the intensity is high enough, the use of specialized filters (see Figure 3) will enable fine-tuning of signal collection to include or exclude other components. A dedicated filter combination for quantitative imaging of GFP should use a narrow bandpass emission filter to maximize the fluorescent protein versus background signal (usually autofluorescence). With a narrow bandpass emission filter, the sharp greenish-blue fluorescent protein signal (Figure 3(a)) is preferentially passed to the detector over the greenish-yellow (Figure 3(b)) or greenish-orange (Figure 3(c)) background of cellular autofluorescence and other fluorophores emitting longer wavelengths. Autofluorescence usually exhibits a very broad spectrum, whereas the green fluorescent protein emission spectrum is relatively narrow. In principle, the narrowest bandpass filter should be chosen to increase the discrimination of fluorescent protein emission over other background signals. However, the optimal passband of the filter will be limited by the required signal-to-noise ratio and detector characteristics, as discussed below. The optimal bandpass should be determined for each imaging situation. In addition, although enhanced green fluorescent protein is illustrated in Figure 3, the specific filter combination chosen depends not only on signal strength, but also on the fluorescence emission spectral profile of the protein variant (see Table 1) being imaged.
Blue fluorescent protein variants can be imaged in widefield fluorescence microscopy using standard DAPI filter combinations with excitation filter profiles in the 330 to 380 nanometer range, dichromatic mirrors with cut-on wavelengths of 385 to 400 nanometers, and either bandpass or longpass emission filters (420 nanometers and above). These filter combinations are not very useful for cyan fluorescent proteins, however, which produce optimal images with filters designed to image fluorophores excited with blue-violet light (400 to 440 nanometers). Dichromatic mirrors for cyan fluorescent proteins should have cut-on wavelengths between 455 and 460 nanometers, as well as emission filters that transmit light between 460 and 500 nanometers. Many of the standard blue-violet longpass and bandpass filter combinations can be employed with success to image cyan fluorescent protein derivatives (including Cerulean and AmCyan1), but the microscope and aftermarket filter manufacturers also provide custom sets designed specifically for these probes.
Although yellow fluorescent protein derivatives (including enhanced, Venus, and Citrine) usually produce acceptable images using filter combinations designed for FITC and green fluorescent proteins, better results are obtained when filter parameters are optimized for the slightly longer absorption and emission wavelength profiles displayed by these probes. The ideal combinations for yellow fluorescent proteins are excitation filters with a bandpass region spanning 490 to 510 nanometers, a dichromatic mirror with a cut-on wavelength of 515 nanometers, and an emission filter passing wavelengths between 520 and 550 nanometers. A longpass emission filter with a cut-on wavelength of 515 nanometers or slightly higher will also produce good images with yellow fluorescent proteins.
The spectral profiles of orange and red fluorescent proteins span a large wavelength region and there is, unfortunately, no single filter combination that is ideal for imaging the entire collection of these fluorophores. DsRed fluorescent protein variants can often be visualized with standard TRITC filter combinations, as well as many of the longpass and bandpass sets designed for other probes absorbing in the green region of the visible spectrum. Orange fluorescent proteins (mOrange and CoralHue Orange) should be imaged with excitation filters in the 500 to 540 nanometer region, coupled with a dichromatic mirror having a cut-on wavelength of approximately 550 nanometers and bandpass emission filters featuring wavelengths between 560 and 600 nanometers. Several of the standard green excitation filter combinations loosely fit the requirements, but specialized filter parameters must be designed for optimum imaging of specific fluorophores in this category. Longer wavelength red fluorescent proteins (mCherry, HcRed1, mRaspberry, and mPlum) can usually be imaged with yellow excitation filter combinations. For example, the Nikon Y-2E/C set, having a bandpass excitation filter of 540 to 580 nanometers, a dichromatic mirror with a cut-on wavelength of 595 nanometers, and a bandpass emission filter capturing photons between 600 and 660 nanometers, can be used for many of these red probes. Custom filter combinations are offered by the microscope and filter manufacturers.
In laser scanning confocal microscopy, the choice of excitation wavelengths is restricted to a narrow range of available laser lines for each instrument. Typical microscopes, such as the Nikon A1 HD25/A1R HD25, are equipped with lasers that span the violet (405 and 440 nanometers), blue (457, 477, and 488 nanometers), green (514 and 543 nanometers), yellow-orange (568 and 594 nanometers), and red (633 and 647 nanometers) spectral regions. These excitation wavelengths can be effectively used to excite most of the fluorescent proteins listed in Table 1, depending upon the availability of appropriate dichromatic mirrors and emission filters. Multiphoton microscopes can also be utilized to image most of the popular fluorescent protein derivatives in the restricted spatial volume that is characteristic of these instruments.
Table 1 - Fluorescent Protein Filter Combination Parameters
CWL / BW (nm)
CWL / BW (nm)
(% of EGFP)
|GFP (wt)||Argon (488)||450/50||480LP||510/50||48|
|Green Fluorescent Proteins|
|Azami Green||Argon (488)||470/40||495LP||520/30||121|
|Blue Fluorescent Proteins|
|Cyan Fluorescent Proteins|
|CoralHue Cyan||Argon (488)||450/50||480LP||505/35||73|
|Yellow Fluorescent Proteins|
|Orange and Red Fluorescent Proteins|
|CoralHue Orange||He-Ne (543)||525/40||550LP||580/40||92|
|Optical Highlighter Fluorescent Proteins; (N) = Native (P) = Photoconverted|
|PA-GFP (N)||Diode (405)||400/60||465LP||530/50||8|
|PA-GFP (P)||Argon (488)||480/40||505LP||535/40||41|
|PS-CFP (N)||Diode (405)||395/50||430LP||470/60||16|
|PS-CFP (P)||Argon (488)||470/50||500LP||530/40||15|
|PS-CFP2 (N)||Diode (405)||395/50||430LP||470/60||26|
|PS-CFP2 (P)||Argon (488)||470/50||500LP||530/40||32|
|Kaede (N)||Argon (488)||485/40||510LP||535/30||259|
|Kaede (P)||Kr-Ar (568)||540/50||570LP||590/30||59|
|mEosFP (N)||Argon (488)||490/30||510LP||535/30||128|
|mEosFP (P)||Kr-Ar (568)||540/50||570LP||600/40||68|
|Kindling (N/P)||Kr-Ar (568)||560/50||590LP||625/50||12|
|Dronpa (P)||Argon (488)||485/30||505LP||530/30||240|
Presented in Table 1 is a compilation of properties displayed by several of the most popular and potentially useful fluorescent protein variants. Along with the common name and/or acronym for each fluorescent protein, the optimum laser wavelength line, as well as suggested starting points for the excitation and emission filter bandwidths (and center wavelengths) are listed. Also included in the table is the relative brightness and recommended dichromatic mirror parameters. The computed brightness values were derived from the product of the molar extinction coefficient and quantum yield, divided by the value for EGFP. This listing was created from scientific and commercial literature resources and is not intended to be comprehensive, but instead represents fluorescent protein derivatives that have received considerable attention in the literature and may prove valuable in research efforts.
Detectors typically preferred for quantitative microscopy are the photomultiplier and cooled charge-coupled device (CCD) camera systems. Because photomultipliers are not imaging devices, raster scanning (as used in laser confocal microscopy) must be performed across the specimen to build an image point-by-point. Alternatively, CCD image sensors contain arrays of photodiodes that produce two-dimensional images limited in size only by the number of individual diode elements. Aside from signal-to-noise considerations, the most important aspect of a detector system is the linearity and offset. Both photomultipliers and CCD arrays are highly linear and higher end camera systems enable adjustment of the offset (often termed black level), as do all photomultiplier configurations. In practice, offset should be adjusted to read zero when fluorescence emission is absent from the specimen. If there is a nonzero offset in the system, it will not be possible to obtain quantitative differences between images.
In laser scanning confocal microscopes, the photomultipliers usually exhibit lower dynamic range (8 bits or 256 gray levels) and detection efficiency (15 to 30 percent) when compared to a cooled CCD camera system. Although it would be desirable to increase the detection efficiency, the 8-bit dynamic range is usually not a problem because generally fewer than 255 photons per pixel are collected for each scan (even for very well stained specimens). Still, the excellent linearity of the photomultiplier response, along with the background rejection properties coupled with the availability of appropriate lasers make the confocal microscope an excellent choice for quantitative imaging of fluorescent proteins. Multiphoton microscopy, which eliminates photodamage away from the focal plane, avoids chromatic aberrations, and provides high resolution three-dimensional measurements, is also a very useful tool for imaging fluorescent proteins.
The most important parameter for defining the utility of quantitative imaging is the signal-to-noise ratio (abbreviated S/N). In modern commercial fluorescence microscopes this value is generally derived from the shot noise, which is defined as the square root of the number of photons collected by the detector. Several systematic errors may be introduced in the detection system, but they are generally small compared to the shot noise for CCDs and photomultipliers. For example if a detector collects 100 photons per pixel, the noise would be expected to be 10 (square root of 100), or 10 percent of the signal. Likewise, for a signal of 10,000 photons per pixel, the noise (100) is only 1 percent of the signal. These examples reflect the typical signal levels in fluorescence microscopy, and errors introduced in the detection electronics are usually less and 1 percent. At first appearance, it would seem obvious that the highest signal-to-noise level is the most desirable, but in reality, factors such as photobleaching, fluorophore saturation, the number of fluorophores in a specimen, and spatial resolution limit the obtainable signal-to-noise ratio.
In most cases, an equal ratio exists between the fluorescent protein label and the cellular target protein to which it is fused. Therefore, the concentration of fluorescent molecules in a cell can be calculated from the total cellular fluorescence using known concentrations of recombinant fluorescent protein as a reference standard. Enhanced green fluorescent protein is typically employed for this purpose because of its photostability and brightness. For calculation of intracellular EGFP concentrations, known quantities of the marker can be embedded in polyacrylamide slabs and imaged alongside cells expressing fusion constructs. Reference standards have also been developed for calculating the density of membrane-bound proteins through attaching precise quantities of histidine-tagged EGFP to coated beads. As fluorescent proteins are increasingly utilized to label molecules in entire organisms using knock-in methodology, quantitative analysis of protein expression will become an increasingly useful technique.
Choosing Objectives for Fluorescent Protein Imaging
Although it may be the most important consideration when designing protocols for an imaging experiment, the selection of objectives is often given the least amount of thought. The objective parameters are paramount in determining the available spatial resolution and the amount of light gathered for the image sensor. In general, objectives are defined by three principal criteria: numerical aperture, magnification, and the degree of optical correction. The least important parameter is magnification despite widespread belief to the contrary. Magnification determines only how large an object will appear in the eyepiece or detector, but does not play a role in resolution (the smallest detail that can be clearly observed) or brightness (the amount of fluorescence signal that will be collected) of the image.
Resolution and brightness are a function of the numerical aperture, which is the most important criterion for selection of an objective. The numerical aperture defines the amount of light that will enter the objective front lens, so the highest possible numerical aperture should be chosen for quantitative imaging of fluorescent proteins. The general principle behind image collection (of any form) is simple: as the signal-to-noise ratio increases, so does the quality of the image. This concept is particularly critical to image collection with a laser scanning confocal microscopy system, where the number of photons gathered is often quite small (usually only a few per pixel). Therefore, careful selection of the proper objective is particularly useful when optimizing photon collection strategies. Note that the overall brightness of an image is proportional to the fourth power of the objective numerical aperture divided by the square of the magnification.
Illustrated in Figure 4 is an example of the principle of numerical aperture importance with regard to resolution and image brightness. The specimen is a culture of adherent human osteosarcoma epithelial cells (U2OS line) transfected with a plasmid vector encoding for enhanced green fluorescent protein fused to the mitochondrial targeting nucleotide sequence from subunit VIII of human cytochrome C oxidase. Upon transcription and translation of the plasmid in transfected mammalian hosts, the mitochondrial localization signal is responsible for transport and distribution of the fluorescent protein chimera throughout the cellular mitochondrial network. Tubular mitochondria can be subsequently visualized using fluorescence microscopy.
In Figure 4, the same viewfield was imaged using objectives having the same optical correction (plan fluorite) and numerical aperture (1.3), but with magnifications ranging from 40x to 100x. Although the number of pixels and detector conditions utilized in collecting the images for Figure 4 were identical, the mitochondria are brightest when imaged through the 40x objective (Figure 4(a)). In contrast, the higher magnification 60x and 100x objectives (Figure 4(b) and 4(c), respectively) yield progressively darker images, with the 100x image being almost indiscernible. This objective can still be used to image the specimen, but the detector gain must be significantly increased, resulting in a deterioration of the signal-to-noise ratio and a generally inferior image. Note that the resolving power of the objectives (Figure 4(d) through 4(f)) is comparable due to the identical numerical aperture values.
One of the primary concepts to be gleaned from the data in Figure 4 is to avoid excessive magnification when choosing objectives for imaging of fluorescent proteins (and other fluorophores, for that matter). Simply increasing the digital enlargement (zoom) during image collection on a confocal microscope using the 40x objective results in an image equivalent in size to the 100x lens (Figures 4(d) and 4(f)). The resolution of both objectives is the same because they have identical numerical aperture values. The argument concerning the relative unimportance of magnification should not be taken to suggest that using the 60x or 100x objectives is not beneficial. In fact, selecting a high magnification objective is often necessary when imaging very small objects, such as peroxisomes or secretory granules, using a widefield microscope. Because the image size relative to detector size plays an important role in determining spatial sampling frequency, the optimal magnification is determined by the parameters of the digital camera system (CCD pixel size and the intermediate magnification factor). Thus, the best choice of objective usually depends on the optical configuration of the instrumentation in addition to the specific requirements of a particular experiment.
The use of highly corrected objectives (apochromat) should be carefully considered and avoided if possible. The typical optical corrections included in objectives for chromatic aberration and flatness of field require additional lens elements as the degree of correction is increased (for example, from fluorite to apochromat). With each additional lens, the total transmission of light through the objective decreases, which further restricts the amount of signal able to reach the detector. Although highly corrected lenses are required for specialized applications, such as multicolor imaging, a high signal-to-noise ratio (and thus an exceptional quality image) will be more difficult to obtain, especially when imaging some of the dimmer fluorescent proteins, such as HcRed or ECFP. In summary, a 40x fluorite objective (numerical aperture of 1.3) is preferable for imaging specimens containing a single fluorescent protein, but for multicolor experiments (two or more fluorescent proteins), an apochromat objective is necessary for color correction.
Multicolor Imaging with Fluorescent Proteins
The primary reasoning behind the development of a full color palette of fluorescent proteins is to simultaneously track two or more cellular events. Given the ever-growing number of fluorescent protein variants currently available (see Table 1), the optimum pairings may not be obvious. The broad excitation and emission spectral profiles exhibited by fluorescent proteins and their color-shifted genetic variants often require specialized considerations when designing live-cell imaging experiments using two or more of these unique probes simultaneously. Of primary concern are potential bleed-through artifacts (in effect, the spillover of fluorescence from one probe into the channel configured for a second probe) resulting from a significant degree of spectral overlap usually exhibited by fluorescent protein combinations.
Bleed-through (often referred to as crossover or crosstalk) can occur during both excitation and emission of different fluorescent proteins. When examining the properties of fluorescent probes, bleed-through occurs toward the blue end of the spectrum (shorter wavelengths), whereas the artifact is most pronounced at longer wavelengths (red end) for emission. As an example, a green fluorescent protein can often be detected through red emission filters, but a red fluorescent protein is generally not imaged using a green filter. For this reason, multicolor imaging of fluorescent proteins should proceed with the longest wavelength emission probe imaged first, using excitation wavelengths that do not cross over to the shorter wavelength probe. For example to image EYFP and EGFP in the same cell using an argon-ion laser, the EYFP signal would be collected first at with the 514-nanometer spectral line, which is beyond the absorption profile of EGFP, and emission gathered with a 530-550 nanometer bandpass filter. The emission filter bandwidth is not particularly critical since only the EYFP is excited. EGFP is imaged using a second scan from the 477-nanometer argon-ion laser line together with a very narrow 490 to 500 nanometer bandpass emission filter. This filter should be correctly positioned to exclude EYFP fluorescence and allow specific capture of signal from EGFP (a somewhat tedious task). Although the peak of the 488-nanometer argon-ion laser is closer to the excitation maximum of EGFP, it is too close to the emission filter bandpass region, resulting in the possibility of interference from reflected laser light with image collection.
An important point to consider when imaging two or more fluorescent proteins is that multicolor imaging with filters requires careful attention in order to control bleed-through of fluorescence emission into unintended channels. However, it should be noted that significant restriction of filter bandwidth occurs at the expense of signal-to-noise (because the collection bandwidth is severely restricted) and temporal resolution due to the separate collection strategies required. Secondly, specialized optical designs, often particular to the experiment, are usually required to achieve optimal imaging conditions. For example, the specialized filter combinations necessary for collection of green fluorescent protein variants in the presence of yellow derivatives are not commonplace in standard confocal microscopes. As the number and spectral profile variations of fluorescent proteins continues to grow, the need for specific filter combinations will increase accordingly.
Illustrated in Figure 5 are a series of images captured in cell lines simultaneously transfected with two or more fluorescent proteins fused to sub-cellular localization targets. The HeLa cells presented in Figure 5(a) were labeled with EYFP (nucleus), ECFP (Golgi), and DsRed2FP (mitochondria), three probes that are clearly separated by the widefield fluorescence filter combinations used in capturing the image (Nikon CFP, YFP HYQ, and Cy3 HYQ). Opossum kidney epithelial cells (OK line) transfected with EGFP (tubulin), ECFP (nucleus), and DsRed2FP (mitochondria) and imaged with standard filter combinations are featured in Figure 5(b). Similarly, the rabbit kidney cells (RK-13 line) in Figure 5(c) were transfected with EYFP (endoplasmic reticulum) and ECFP (peroxisomes) and imaged with the Nikon CFP and YFP HYQ filter sets. The African water mongoose cells (APM) cells illustrated in Figure 5(d) were transfected with EGFP (tubulin) and DsRed2FP (nucleus), and the human osteosarcoma cells (U2OS line; Figure 5(e)) were labeled with the same nuclear probe along with ECFP (mitochondria). Finally, the baby hamster kidney cells (BHK cell line) presented in Figure 5(f) were transfected with EGFP (peroxisomes) and DsRed2FP (mitochondria). All of the images featured in Figure 5 were captured using a widefield fluorescence microscope and Nikon filter combinations optimized for the appropriate fluorescent proteins.
Although the brightest fluorescent protein classes are the green and yellow (see Table 1), their spectral maxima are separated by only by an average of 20 to 25 nanometers. Multicolor experiments using paired green and yellow fluorescent proteins are possible, but bleed-through between filter combinations becomes a significant problem. As a result, the green and yellow combinations are not often used together. One of the most popular dual probe combinations is cyan and yellow fluorescent proteins (ECFP and EYFP, see Figure 5). Even though ECFP is not much brighter than EBFP, its use presents two advantages in that ECFP is efficiently excited with the 457-nanometer line from argon-ion lasers, and the probe does not photobleach as readily (as illustrated in Figure 2).
Combining DsRed derivatives with either green or yellow fluorescent proteins can generate a useful pairing of probes with emission spectra that can be readily separated. However, care should be exercised in selecting DsRed variants due to differences in maturation rates. Furthermore, because DsRed and its derivatives form obligate tetramers in vivo, they may produce unintended effects on the biology of the protein system being investigated (such as aggregation or localization to organelles other than those intended). There is no effective rule of thumb available to predict whether DsRed derivatives will work as a fusion tag, and some proteins may tolerate the oligomerization whereas others will not. For triple labeling using only fluorescent proteins (Figures 5(a) and 5(b)) the combination of cyan, yellow (or green), and DsRed offers an excellent solution.
There are several compromises that must be considered when using two or more fluorescent proteins in the context of live-cell imaging. For example, the rate of data acquisition slows with the addition of each color due to the requirement for different image collection conditions. Addressing fluorescence bleed-through becomes more complex, especially because the emission spectral profiles of fluorescent proteins tend to be quite broad (and often overlapping). In addition, because the popular fluorescent proteins vary in relative brightness (Table 1), each color may require different signal integration times. Cyan and blue fluorescent proteins are quite dim and may require longer collection periods that saturate the brighter green and yellow fluorescent proteins. Therefore, it is important to balance the parameters of the experiment (in effect, fast collection versus the optimal separation) with the choice of fluorescent proteins. For example, combining EYFP and DsRedFP is a good choice for fast and efficient image capture, because these fluorescent proteins can share a single excitation setting, whereas using ECFP with EYFP can provide a higher degree of spectral separation.
Imaging Optical Highlighter Fluorescent Proteins
When designing experiments using optical highlighter fluorescent proteins, several important aspects should be considered with regards to live-cell imaging, microscope configuration, vector construction, and the choice of proteins in order to optimize image acquisition parameters. These unique probes are primarily used to monitor dynamic processes in living cells, which often requires maintaining the cell culture in a healthy condition on the microscope stage for extended periods of time. A variety of specialized heated imaging chambers fabricated as stage inserts are commercially available for long-term cell maintenance on the microscope, but the investigator must also consider several essential auxiliary requirements, such as a carbon dioxide source, light-tight microscope enclosures, and vibration isolation. In short, the experimental conditions necessary to successfully image living cells for lengthy periods should be carefully established before embarking on experiments with optical highlighter proteins.
Several of the optical highlighter fluorescent proteins can be successfully imaged using traditional widefield fluorescence microscopy techniques, but serious quantitative investigations involving photobleaching methodology and photoactivation often must be performed on a confocal or multiphoton microscope equipped with specialized laser systems. Specifically, PA-GFP, PS-CFP, Kaede, EosFP, and Dronpa all require near-ultraviolet or violet illumination for photoactivation or photoconversion, but longer wavelength sources (blue and green excitation) are necessary for imaging the proteins. Therefore, confocal microscopes should be equipped with an ultraviolet or violet laser, as well as visible light lasers emitting blue and green spectral lines.
The most popular current laser configuration for optical highlighters is a 405-nanometer diode, 488-nanometer argon-ion, and a 543-nanometer helium-neon, although ultraviolet high-energy argon, diode-pumped solid-state, and helium-cadmium lasers may serve as well for photoactivation (keeping in mind the potential cell damage artifacts possible with ultraviolet irradiation). Other useful imaging lasers include several new diode and helium-neon models and the multi-spectral krypton-argon systems. Multiphoton instruments should ideally be configured with a broadband Ti:Sapphire laser featuring tunable output in the wavelength range between 750 and 1100 nanometers.
Illustrated in Figure 6 is the photoconversion of a PS-CFP2 fluorescent protein fusion product with human beta-actin using a 405-nanometer diode laser for imaging and conversion, as well as the argon-ion 488-nanometer spectral line for imaging and tracking of the photoconverted protein. The fusion construct plasmid vector was used to transfect opossum kidney (OK line) epithelial cells in a live-cell imaging chamber for expression of labeled filamentous actin. Figure 6(a) shows a single cell before photoconversion, imaged with the diode laser. A region of interest was drawn around a large portion of the actin cytoskeletal network (Figure 6(b); red box), and then illuminated at 405-nanometers with 40-percent laser power. When imaged with both the diode and argon-ion lasers (Figure 6(b)), the photoconverted PS-CFP2 fusion protein is clearly visible in green fluorescence, contrasting the non-converted probe (blue). After 10 minutes (Figure 6(c)), the photoconverted actin has begun to be incorporated into filaments outside the region of interest, and at 30 minutes (Figure 6(d)) much of the cytoskeletal network is labeled with the optical highlighter.
The delineation of specific regions for illumination during photoconversion and photoactivation is most efficiently conducted under the control of a acousto-optic tunable filter (AOTF), which can be conveniently used to define circular, elliptical, rectangular or free-hand selections for irradiation of the specimen. The AOTF also serves to fine-tune laser power levels to modulate intensity at the specimen, and is capable of sequentially scanning specimens using a technique known as high-speed channel switching or multitracking. This approach enables the sequential excitation and collection of emission from several fluorescent probes to avoid bleed-through artifacts and more accurately separate emission spectra.
Defining regions of interest in widefield microscopy presents a greater challenge. Research-level microscopes are often equipped with a variable aperture that can be adjusted to illuminate selected regions in the specimen, although with much less accuracy than a confocal microscope AOTF system. Widefield instruments must also be configured with specialized fluorescence filter combinations in order to image several of the optical highlighter proteins. Comparatively, the confocal and multiphoton instruments are far more suitable for quantitative analysis of cellular dynamics using photoconversion and photoactivation techniques than are widefield microscopes and should therefore be the first choice for these applications.
Generation of sufficiently high fluorescence signal levels for optimal image acquisition is also an important factor to consider when using optical highlighters. In this regard, the absorption molar extinction coefficient and fluorescence quantum yield values for the native and photoconverted protein species can be employed as a gauge to determine relative brightness levels. For example, both Kaede and mEosFP have similar spectral bandwidth profiles in the green (native) and red (photoconverted) regions of the visible light spectrum. However, the native green form of Kaede has approximately twice the fluorescence emission intensity or brightness level of green mEosFP due to a higher extinction coefficient and quantum yield. Therefore, before photoconversion, the imaging signal-to-noise ratio for Kaede is much higher than that of mEosFP. After photoconversion, both proteins feature similar brightness levels (see Table 1) and produce comparable images.
Among the optical highlighter proteins currently available, Dronpa, Kaede, and mEosFP are the brightest, followed by PA-GFP, which is similar in brightness level to the photoconverted species of Kaede, but significantly lower than the native protein. The commercially available PS-CFP2 variant is about one-half as bright as Kaede, whereas the kindling and PS-CFP proteins, both in the native and photoconverted forms, are by far the dimmest optical highlighters, yielding only about 25-percent of the brightness level exhibited by photoconverted Kaede (Table 1). The latter highlighters exhibit significantly lower signal-to-noise ratios during imaging, a factor that should be considered when planning investigations of biological targets having low abundance levels or requiring high imaging speeds.
The battery of fluorescent proteins and imaging tools currently available enables the cell biologist to readily monitor protein dynamics in living cells, continuing to provide numerous new insights into the behavior of proteins, organelles, and cells. In so doing, these remarkable probes have ushered in a new era of cell biology in which both steady-state and kinetic microscopy techniques can be used to decipher pathways and mechanisms of biological processes. Coupled to the rapid advances made in developing live-cell imaging microscope systems, including low light level electron multiplying CCD cameras, spinning disk confocal instruments, and slit-scanning microscopes, the stage is set for imaging fluorescent proteins and optical highlighters over the entire visible and near-infrared spectral regions with near real-time precision.
David W. Piston - Department of Molecular Physiology and Biophysics, Vanderbilt University, Nashville, Tennessee, 37232.
George H. Patterson and Jennifer Lippincott-Schwartz - Cell Biology and Metabolism Branch, National Institute of Child Health and Human Development, National Institutes of Health, Bethesda, Maryland, 20892.
Nathan S. Claxton and Michael W. Davidson - National High Magnetic Field Laboratory, 1800 East Paul Dirac Dr., The Florida State University, Tallahassee, Florida, 32310.