Various mechanisms are often employed in fluorescence microscopy applications to restrict the excitation and detection of fluorophores to a thin region of the specimen. Elimination of background fluorescence from outside the focal plane can dramatically improve the signal-to-noise ratio, and consequently, the spatial resolution of the features or events of interest. Total internal reflection fluorescence microscopy (TIRFM) exploits the unique properties of an induced evanescent wave or field in a limited specimen region immediately adjacent to the interface between two media having different refractive indices. In practice, the most commonly utilized interface in the application of TIRFM is the contact area between a specimen and a glass coverslip or tissue culture container.
The concepts underlying TIRFM are not new, and much of the recent interest in, and enthusiasm for, the technique have come about due to technological advances that facilitate its use. The availability of complete ready-to-use instrumentation systems for employment of the method, as well as developments in fluorophore technology, such as genetically encoded fluorescent species, have made it possible to investigate a number of cell membrane and other surface processes in a direct manner that was not previously possible.
Physical Basis of TIRFM
The physical phenomenon of total internal reflection (TIR) has been relied upon in such seemingly diverse applications as modern fiber optic data transmission, and in the centuries-old utilization by diamond cutters to enhance the sparkle, or "fire", of cut gemstones. In each case, refraction (or bending) of light as it encounters the interface between two media having different refractive indices (n) results in confinement of a portion or all of the light to the higher-index medium. A collimated light beam propagating through one medium and reaching such an interface is either refracted as it enters the second medium, or reflected at the interface, depending upon the incident angle and the difference in refractive indices of the two media. Total internal reflection is only possible in situations in which the propagating light encounters a boundary to a medium of lower refractive index. Its refractive behavior is governed by Snell's Law:
where n(1) is the higher refractive index and n(2) is the lower refractive index. The angle of the incident beam, with respect to the normal to the interface, is represented by θ(1), while the refracted beam angle within the lower-index medium is given by θ(2). When light strikes the interface of the two materials at a sufficiently high angle, termed the critical angle (θ(c)), its refraction direction becomes parallel to the interface (90 degrees relative to the normal), and at larger angles it is reflected entirely back into the first medium.
Although light no longer passes into the second medium when it is incident at angles greater than the critical angle, the reflected light generates a highly restricted electromagnetic field adjacent to the interface, in the lower-index medium. This evanescent field is identical in frequency to the incident light, and because it decays exponentially in intensity with distance from the interface, the field extends at most a few hundred nanometers into the specimen in the z direction (normal to the interface). In a typical experimental setup, fluorophores located in the vicinity of the glass-liquid or plastic-liquid surface can be excited by the evanescent field, provided they have potential electronic transitions at energies within or very near the wavelength bandwidth of the illuminating beam. Because of the exponential falloff of evanescent field intensity, the excitation of fluorophores is restricted to a region that is typically less than 100 nanometers in thickness. By comparison, this optical section thickness is approximately one-tenth that produced by confocal fluorescence microscopy techniques. Because excitation of fluorophores in the bulk of the specimen is avoided, confining the secondary fluorescence emission to a very thin region, a much higher signal-to-noise ratio is achieved compared to conventional widefield epifluorescence illumination. This enhanced signal level makes it possible to detect single-molecule fluorescence by the TIRFM method.
The basic concept of total internal reflection fluorescence is schematically illustrated in Figure 1, in which specimen cells incorporating fluorescent molecules (green fluorophores in the figure) are supported on a glass microscope slide. The refractive indices of the glass slide (1.518) and the aqueous specimen medium (approximately 1.35) are appropriate to support total internal reflection within the glass slide. With adjustment of the laser excitation incidence angle to a value greater than the critical angle, the illuminating beam is entirely reflected back into the microscope slide upon encountering the interface, and an evanescent field is generated in the specimen medium immediately adjacent to the interface. The fluorophores nearest the glass surface are selectively excited by interaction with the evanescent field, and secondary fluorescence from these emitters can be collected by the microscope optics.
As discussed previously, the angles taken by propagating light beams following refraction or reflection at an interface between different media depends upon the light's incidence angle at the interface, and the refractive indices of the two materials. The critical angle of incidence, beyond which total internal reflection occurs, can be calculated by manipulation of the Snell's Law expression, given above. Applying the equation to a typical biological investigation of cell membrane processes, the refractive index of the microscope slide or coverslip is represented by n(1) (approximately 1.5), while n(2) represents the refractive index of the aqueous buffer solution or cytoplasmic components (1.33 to 1.38). With n(1)greater than n(2), when θ(1) exceeds the critical angle θ(c), total internal reflection occurs within the glass medium. At the critical incidence angle, refraction occurs at 90 degrees (sin θ(2) = 1), and Snell's Law reduces to:
and therefore, the critical angle can be expressed as:
Total internal reflection does not occur suddenly as a new phenomenon at the critical angle, but a continuous transition is followed from predominant refraction with a small amount of reflection, to total reflection when the critical angle is exceeded. As the incident angle increases toward the critical angle value, the transmitted (refracted) beam diminishes in intensity while the reflected beam grows stronger. At all angles greater than the critical angle, total internal reflection is achieved, in which essentially all of the light is reflected back into the first medium. Even though the light no longer propagates into the second medium, there is a small amount of penetration of the reflected light across the interface, which then propagates parallel to the surface, creating an electromagnetic field in the second medium immediately adjacent to the interface. This field is termed the evanescent field, and within a limited region near the interface, it is capable of exciting fluorophores. The range over which excitation is possible is limited by the exponential decay of the evanescent wave energy in the z direction (perpendicular to the interface). The following equation defines this energy as a function of distance from the interface:
where E(z) is the energy at a perpendicular distance z from the interface, and E(0) is the energy at the interface. The penetration depth (d) is dependent upon the wavelength of the incident illumination (λ(i)), the angle of incidence, and the refractive indices of the media at the interface, according to the equation:
At small incidence angles, light waves propagating through the interface to the lower-refractive index medium are sinusoidal, and have a characteristic period. At increasing angle, approaching the critical value, the period of refracted rays becomes longer and the propagation direction becomes more nearly parallel to the interface. When the critical angle is achieved, the wave period becomes infinite and the refracted light wavefronts are aligned perpendicular to the interfacial surface.
To summarize, several crucial factors govern the utilization of the evanescent wave in microscopy. For total internal reflection to occur and produce an evanescent field, the refractive index of the medium of illumination incidence must be greater than that of the specimen medium (n(1) greater than n(2)), and the angle of incidence (θ(1)) must be greater than the critical angle (θ(c)). The incident illumination wavelength affects both the penetration depth of the evanescent wave and the specific fluorophores that are excited, which must have appropriate absorption characteristics in the wavelength band of the illuminant. The implication of the wavelength effects combined with the fact that the energy of the evanescent wave decreases exponentially in the z direction, is that highly specific fluorescent excitation can be induced in a very thin optical section, typically less than 100 nanometers in thickness. Although TIRFM is limited to imaging at the interface of two different media having suitable refractive indices, a great number of applications are ideally suited to the technique. One of the most active areas of research interest is in the biomedical arena, in which many compelling questions involve processes that take place at the cell surface or plasma membrane — appropriate interfaces for TIRFM investigation.
Basic Instrumental Approaches to TIRFM
There are two basic approaches to configuring an instrument for total internal reflection fluorescence microscopy: the prism method, and the objective lens method. Figure 2 illustrates these two general configurations. In the prism technique, a focused laser beam is introduced into the microscope coverslip by means of a prism attached to its surface, and the beam incidence angle is adjusted to the critical angle (see Figure 2(a)). Reliance on a prism for introduction of the illuminating beam has several limitations, primarily due to geometric constraints on specimen manipulation, and although the method has been utilized in biological applications for more than two decades, it has never become a mainstream research tool. There are many variations of the prism configuration, but most restrict access to the specimen, making it difficult to perform manipulations, to inject media into the specimen space, or to carry out physiological measurements.
Another disadvantage of the prism technique is that in most configurations based on inverted microscope designs, the illumination is introduced on the specimen side opposite the objective optics, requiring imaging of the evanescent field region through the bulk of the specimen. Placement of the prism on the objective side of the specimen to avoid this presents additional problems because of the close proximity of a short-working-distance objective to the specimen and prism location. The general complexity and precision required in configuring an imaging system to utilize total internal reflection discouraged many potential researchers before complete ("turnkey") systems became available from the microscope manufacturers. Investigators who wanted to utilize the technique were required to engineer and build their own systems, and this difficulty, combined with the necessity of setting up and maintaining an open laser on an optical bench, meant that earlier users of the prism method were more often physicists than biologists.
The objective lens technique, which is sometimes referred to as through the lens illumination, avoids many of the limitations of utilizing a prism to introduce light at the required angles (see Figure 2(b)). In this method, the objective is employed to introduce either coherent laser or non-coherent arc lamp illumination to the coverslip-specimen interface. Incidence angles greater than the critical angle are achieved by the use of objectives of high numerical aperture (ideally 1.45 or higher). Typically the numerical aperture of an objective is thought of as characterizing the light gathering capability of the lens. Conversely, the numerical aperture directly determines the range of angles at which light can exit the objective when it is utilized to deliver illumination. The relationship between numerical aperture and achievable illumination incidence angles is described by the following equation:
where NA is the numerical aperture of the objective, n represents refractive index, and θ is one-half the objective angular aperture. Combining this relationship with the condition for total internal reflection given above illustrates that living cells having a typical refractive index of 1.38 require illumination with an objective having a numerical aperture of greater than 1.38 in order to achieve total internal reflection. Light entering the objective must pass through the portion of the aperture cone corresponding to numerical aperture values larger than 1.38 in order to be totally reflected at the specimen-glass interface. If coherent laser illumination is employed, it must be focused at the periphery of the objective rear aperture to ensure that light will exit the front optical surface at an angle equal to or greater than the critical value. In the case of non-coherent illumination, such as that from an arc-discharge lamp, a mask in the form of an opaque disk must be introduced into the optical path to restrict light passing through the objective to the outer region of the rear aperture.
By confining illumination at the rear focal plane of the objective to a circular annulus region, light rays from the center of the illumination cone that would normally emerge at sub-critical angles is blocked. The resulting emission from the objective is a hollow cone of light incident upon the TIR interface at a half angle sufficient to result in total internal reflection. If significant illumination passes through the central portion of the objective rear aperture (the lower numerical aperture region), epi-illumination rather than total internal reflection is produced, lowering the signal-to-noise ratio at the image plane. In practice, the opaque light-blocking disk can be mounted on a moveable slider, facilitating rapid switching between TIRF and epi-illumination imaging modes.
Accomplishing evanescent field excitation through the use of high-aperture objectives offers greater flexibility in specimen manipulation and measurement options than does the prism-based technique, but the precise control of incident illumination angle is more difficult. When a laser source is used, the angle of incidence of light coupled into a prism can be easily varied over a wide range, allowing straightforward control of evanescent field penetration depth. With the objective system, the point of laser focus in the objective rear focal plane is off-axis (taking advantage of the outer portion of the aperture), and increasing the radial distance from the lens axis produces a corresponding increase in the angle at which light is incident on the specimen (see Figure 3).
If the numerical aperture of the objective is sufficient, the critical angle for total internal reflection can be achieved. Because the primary cellular component (cytosol) has a refractive index of approximately 1.38, an objective numerical aperture exceeding that value is required. With a 1.4 numerical aperture objective, only a few percent of the peripheral area of the lens can be utilized for total internal reflection, and the critical angle can only be marginally surpassed, making coupling of the laser into the rear aperture a very challenging procedure. Obviously, objectives of higher numerical aperture are advantageous, and provide additional working margin for fine adjustment of angles exceeding the critical angle. Once the critical angle is surpassed, further increases in the radial distance of the laser focal point from the lens axis serve to reduce the evanescent field penetration depth in a smooth and reproducible manner.
In general, total internal reflection illumination has potential benefits in any application requiring imaging of minute structures or single molecules in specimens having large numbers of fluorophores located outside of the optical plane of interest, such as molecules in solution in Brownian motion, vesicles undergoing endocytosis or exocytosis, or single protein trafficking in cells. Such specimens typically exhibit a dramatic increase in signal-to-noise ratio from restriction of the excitation region thickness. Figure 4 presents images acquired of a solution of fluorescent microspheres utilizing the TIRFM method (Figure 4(b)) and conventional epi-fluorescence illumination (Figure 4(d)). To the left of each image is its corresponding intensity histogram (Figures 4(a) and 4(c)). The improved resolution of the spheres afforded by increasing the signal-to-noise ratio (S/N) from 1.3 to 35 is apparent in the images, and in the sharp localization and higher signal intensity in the histogram corresponding to the TIRFM image (Figure 4(a)).
It has long been recognized that TIRFM could potentially become a powerful tool in answering a number of biological questions, and although utilized for over 20 years, the technique has not received a considerable amount of attention until recently. Cell-substrate contacts of human skin fibroblasts, labeled with fluorescent lipids, were investigated by TIRFM in the early 1980s. Another study carried out at approximately the same time utilized TIRFM in combination with fluorescence photobleaching recovery (FRAP) to elucidate biomolecular surface dynamics, while still another focused on energy transfer in bovine serum albumin bound to surfaces. The latter study combined TIRFM with fluorescence resonance energy transfer (FRET), another technique currently experiencing rapid growth in application.
Much of the trend toward greater utilization of TIRFM and other cutting-edge techniques is due to the increased availability of advanced modular instrumentation that makes it unnecessary to engineer and build custom systems for each particular research application. Another important factor is the development of versatile biological tools that can be applied to a wide variety of problems, the most significant of which is probably the utilization of green fluorescent protein (GFP) and its cyan, blue, yellow, and red derivatives. GFP, derived from jellyfish, does not require species-specific cofactors for expression and exhibition of fluorescence, and can be used experimentally across species. The biological fluorophore has been inserted into hundreds of proteins, through genetic recombination, and is essentially unlimited in that potential. Another promising capability results from development of GFP mutants that function as indicators of intracellular calcium in the process of neurotransmitter release. These proteins have been monitored in some studies through fluorescence resonance energy transfer.
TIRFM is an ideal tool for investigation of both the mechanisms and dynamics of many of the proteins involved in cell-cell interactions. Figure 5 presents comparative images of live cells (PtK1 kangaroo kidney epithelial cells expressing GFP-vinculin) utilizing a conventional widefield epi-fluorescence method (Figure 5(a)) and evanescent wave illumination (Figure 5(b)). The TIRFM image reveals localization of the fusion protein in cell focal adhesions at the substrate interface in dramatic contrast to the blur produced by out-of-plane fluorescence in the epi-illumination image. Live-cell imaging represents one of the most promising applications of the TIRFM technique. Protein interactions at the cell membrane surface, such as those involved in focal adhesions, have tremendous importance in cell biology. An understanding of the signals involved in normal cell growth and its attenuation resulting from cell-cell contacts (contact inhibition) may provide insight into abnormal cell growth that occurs in diseases such as cancer.
At the biomolecular level, TIRFM techniques have been utilized to image single molecules of the mutant protein GFP-Rac trafficking along thin filopodia of cells growing on a substrate (Figure 6). This protein is involved in cell motility, and knowledge of the dynamics of its interactions at the cell membrane are crucial to understanding the process. The visualization of single-molecule fluorescence with sufficient temporal resolution for dynamic studies is possible with TIRFM because of the outstanding signal-to-noise ratio afforded by the evanescent wave excitation. Figure 6 presents four sequential time lapse frames taken at 200-millisecond intervals, illustrating the movement of a GFP-Rac fusion protein molecule (arrows) through a fine filopodium of a Xenopus cell growing out on a substrate.
Although TIRFM is limited to investigation of structures and processes occurring at or near the coverslip-specimen interface, it is simultaneously fortuitous that many questions of current interest in the biological and biomedical sciences can be probed at the cell membrane. The field of neuroscience is one in which numerous fundamental questions lend themselves to study by TIRF microscopy. An ideal candidate for application of the technique is the study of neurotransmitter release and uptake at the synapse. Historically, the mechanisms of membrane trafficking and fusion, including the release (exocytosis) or uptake (endocytosis) of synaptic vesicles, have been investigated using genetic, biochemical, and electron microscopic approaches. These techniques are in some ways indirect or provide only an instantaneous representation of the processes taking place, and cannot resolve the complex dynamics of the cellular membrane activity.
Development of the patch-clamping technique has allowed capacitance measurements to be used to indicate, by extremely small electrical changes, the addition or subtraction of membrane surface area or the release of oxidative material. The shortcoming of this technique is that only fusion events are detected, and while high temporal resolution can be attained, there is very little information on the spatial location of the important events. Since all of the events are detected together, no specificity is obtained and details of other stages of vesicle trafficking, docking, and membrane fusion have typically been inferred from cell measurements combined with kinetic modeling. Although investigation by electron microscopy provides exceptional spatial resolution, live cell or dynamic studies are not possible, and correlation of the instantaneous views with other measurements is very difficult. The strength of the TIRFM method, demonstrated in the more recent investigations, is that direct visual observation of dynamic protein-vesicle interactions is possible.
Because of the ability to resolve individual vesicles optically, and to follow the dynamics of their interactions directly, TIRFM provides the capability to study the vast number of proteins involved in neurobiological processes in a manner never before possible. A recent study demonstrated the release of fluorescent lipid-containing synaptic vessels from active zones, and the subsequent transport of vesicles from a reserve pool located 20 nanometers from the plasma membrane to provide replenishment for the ones released. Another study involved direct visualization of the role of actin in the dynamic process of endocytosis in cultured mast cells. GFP-actin filaments were observed surrounding fluorescently labeled pinocytic vesicles and pulling them into the cell in a stream of actin. A time lapse TIRFM imaging sequence is presented in Figure 7 illustrating GFP-actin dynamics during endocytosis. The six sequential frames represent different time intervals over a range of 0 to 65 seconds. The white arrows in each frame indicate the GFP-actin fusion protein signal.
Multi-spectral imaging illustrating the vesicle-actin interaction during endocytosis is presented in Figure 8. In the two-channel image (Figure 8(a)), a stream of green-labeled GFP-actin is seen surrounding a vesicle containing Texas red dextran in the extracellular medium. A time lapse sequence of three two-channel images (Figure 8(b) through 8(d)) provide insight into the temporal dynamics of the actin-vesicle interaction during the pinocytosis process. A study of this type could be logically extended to label a number of synaptic proteins utilizing different GFP color variants in order to investigate their interaction and dynamics.
Although specimens in TIRFM are imaged in two dimensions, there are mechanisms by which three-dimensional information on the location of vesicles or structures in cells can be obtained, both in living cell studies and in fixed stained preparations. Figure 9 illustrates the structure in cells immunocytochemically labeled for the protein tubulin, and imaged using both widefield epifluorescence (Figure 9(a)) and evanescent wave illumination (Figure 9(b)). Structural details are revealed in the TIRFM image that could not be visualized with conventional epi-illumination. Comparison of the two image modes is emphasized by overlaying them in pseudocolor. In Figure 9(c), the epifluorescence is assigned the color green, while the TIRFM image is shown in red.
The principles of TIRFM suggest that by varying the illumination incidence angle, and consequently the penetration depth of the evanescent wave, fluorophores can be distinguished by depth on a nanometric scale. This technique of precisely controlling penetration depth is more easily accomplished in prism-type systems, and a recent technical enhancement to the method is the utilization of acousto-optical deflectors (AOD) to rapidly change the incidence angle. By rapid variation in the evanescent field depth, target vesicles or other structures can be tracked at different depths and their positions accurately determined. There are a number of potentially useful applications of AODs in TIRFM, including use as extremely fast shutters that can quickly modulate the illumination wavelength in multi-line laser equipped systems.
As discussed above, the variation of incidence angle in objective-based systems is not as easily accomplished as in those using prisms, although the newer higher numerical aperture objectives provide a considerable improvement in available range of incidence angle adjustment. In general, objective type systems are capable of detecting more emitted light, and the intensity of this signal decreases monotonically with distance. This property is an advantage in calibrating the TIRFM system so that fluorescence signal level can be related to axial position, providing another approach to three-dimensional imaging.
Prospects for Future Development
The basic theory of TIRFM is now well established, and the practical implementation of the technique has been greatly facilitated by recent technological advances. Consequently, an increasing number of biomolecular and cell biology investigations are being conducted using the technique. Configuration of TIRFM systems based on upright or inverted microscopes is relatively straightforward using a laser light source, and can be accomplished using conventional arc-lamp sources if modifications are made to block light passing through the central region of the objective. Complete modular microscope systems, configured for TIRFM in conjunction with other optical techniques, are now available, and some manufacturers provide high numerical aperture objectives designed specifically for internal reflection applications. The TIRFM technique is compatible with a wide range of illumination modes, including brightfield, darkfield, phase contrast, and differential interference contrast, as well as conventional epi-fluorescence. A particular advantage of the objective-based systems is that they can be utilized in conjunction with various mechanisms for manipulating biomolecules, such as atomic force microscopy. It is likely that TIRFM will continue to be merged with other complementary techniques.
The acquisition of image data at high temporal resolution in living cells at multiple wavelengths is an area of great promise for TIRFM, and the expanded utilization of various dye combinations is certain to reveal cellular dynamics in more detail than has previously been possible. Recently, investigators have reported dual emission-wavelength detection utilizing fluorophores that can be excited at a single wavelength, and the possibility exists to expand TIRFM capabilities further by configuring systems for multiple-wavelength excitation. Single molecule studies will be greatly enhanced with further development in dye characteristics and with continued improvement of detectors. Expansion of the TIRFM approach in cellular studies is likely to continue through the refinement of genetic and molecular manipulation techniques, combined with optical detection at the high temporal and spatial resolution afforded by evanescent wave excitation.
Because TIRFM and laser scanning confocal microscopy (LSCM) have certain common capabilities, the two techniques are naturally compared when evaluating possible approaches to research problems. Although both techniques provide optical sectioning capability, the TIRFM approach is limited to specimen regions having an appropriate refractive index interface, while confocal microscopy can selectively image virtually any specimen plane. The minimum optical section thickness produced by confocal methods, however, is approximately 600 nanometers - considerably thicker than the 100-nanometer sections typical of the TIRFM technique. In many applications, it is desirable to minimize the total illumination flux into the specimen (to reduce cell damage, for example), and since confocal instruments illuminate a relatively large specimen volume, this is more readily accomplished with TIRFM. In general, it is more economical to configure a TIRFM instrument, which does not require complex scanning systems, and can be built on nearly any modern research-level optical microscope. The complete, ready-configured systems being provided by a number of manufacturers are the most direct entry point into the TIRFM technique, and allow the combined capabilities of other powerful optical imaging modes.
Stephen T. Ross and Stanley Schwartz - Bioscience Department, Nikon Instruments, Inc., 1300 Walt Whitman Road, Melville, New York 11747.
Thomas J. Fellers and Michael W. Davidson - National High Magnetic Field Laboratory, 1800 East Paul Dirac Dr., The Florida State University, Tallahassee, Florida, 32310.