Total internal reflection fluorescence microscopy (TIRFM) is a technique designed to probe the surface of fluorescently labeled living cells with an evanescent wave generated by a light beam traveling between two media of differing refractive indices. In practice, an incident laser beam is reflected at a critical angle (total internal reflection) when it encounters the interface between a microscope glass coverslip and the aqueous medium containing the cells. Fluorophores within a few 10s of nanometers of the surface (between 10 and 200 nanometers) are excited by the evanescent wave, while those farther away are largely unaffected, as the energy of the evanescent wave declines exponentially with distance from the coverslip. Therefore, TIRFM results in a high signal level arising from fluorophores residing close to the coverslip, superimposed on a very dark background, providing the best possible signal-to-noise ratio. The extreme limitation on excitation depth is ideal for studying single molecules or membrane and organelle components near the surface of the coverslip in adherent cells (see Figure 8(e)). As excitation is limited to the thin region adjacent to the coverslip, photobleaching and phototoxicity are also limited to these areas, rendering TIRFM one of the most useful methodologies for long-term observations. The technique has become a fundamental tool to investigate a wide spectrum of phenomena in cell and molecular biology.

Deconvolution analysis is a technique that applies algorithms to a through-focus stack of images acquired along the optical (z) axis to enhance photon signals specific for a given image plane or multiple focal planes in an image stack. The microscope must be equipped with a high-precision motorized focus drive in order to guarantee image acquisition at precisely defined intervals between focal planes in the specimen. In a typical application (see Figure 8(f)), deconvolution analysis is utilized to deblur and remove out-of-focus light from a particular focal plane of interest using widefield fluorescence excitation and emission (although the technique is useful for other illumination modes as well). The most sophisticated applications apply deconvolution analysis to an entire image stack to produce projection views or three-dimensional models. The stack of widefield images used for deconvolution analysis captures the theoretical maximum number of photons emitted by the specimen. The process of deconvolution reassigns the "blur" intensity arising from photons emitted above and beneath the focal plane to the plane of origin. Therefore, deconvolution uses practically all of the available emission intensity and offers the best possible light budget, thus making this technique the method of choice for extremely photosensitive specimens.

An adaptation of the resonance energy transfer phenomenon to fluorescence microscopy, fluorescence or F?rster resonance energy transfer (FRET) is used to obtain quantitative temporal and spatial information about the binding and interaction of proteins, lipids, enzymes, and nucleic acids in living cells. FRET microscopy is performed using either steady state or time-resolved techniques, but time-resolved FRET imaging has the advantage of more accurately mapping the donor-acceptor distance. A standard widefield fluorescence microscope equipped with the proper excitation and emission filters and a sensitive video camera can be utilized to perform FRET imaging. Biosensors that sandwich an environmentally sensitive protein or peptide between two FRET-capable fluorescent proteins are currently enjoying widespread use in cell biology. These probes are readily imaged in widefield fluorescence microscopy using sensitized emission FRET techniques coupled to ratiometric analysis. In addition, spectral imaging and linear unmixing with laser scanning confocal microscopes is useful for monitoring FRET in biosensors and other fluorescent protein applications.

Fluorescence lifetime imaging microscopy (FLIM) is a sophisticated technique that enables simultaneous recording of both the fluorescence lifetime and the spatial location of fluorophores throughout every location in the image. The methodology provides a mechanism to investigate environmental parameters such as pH, ion concentration, solvent polarity, non-covalent interactions, viscosity, and oxygen tension in single living cells, presenting the data in a spatial and temporal array. FLIM measurements of the nanosecond excited state lifetime are independent of localized fluorophore concentration, photobleaching artifacts, and path length (specimen thickness), but are sensitive to excited state reactions such as resonance energy transfer. In fact, combining FLIM with FRET by monitoring the change in lifetime of the fluorescent donor before and after being involved in resonance energy transfer is considered to be one of the best approaches for examining this phenomena.

Translational mobility (lateral diffusion coefficients) of fluorescently labeled macromolecules and small fluorophores can be determined by fluorescence recovery after photobleaching (FRAP) techniques. In FRAP, a very small, selected region (several micrometers in diameter) is subjected to intense illumination, usually with a laser, to produce complete photobleaching of fluorophores in the region. The result is a dramatic reduction or annihilation of fluorescence. After the photobleaching pulse, the rate and extent of fluorescence intensity recovery in the bleached region is monitored as a function of time at lower excitation intensity to generate information about repopulation by fluorophores and the kinetics of recovery (Figure 9). FRAP is generally conducted using EGFP or other fluorescent proteins. Related photoactivation techniques are based on specialized synthetic caged fluorophores or similarly endowed fluorescent proteins that can be activated by a brief pulse of ultraviolet or violet light. Photoactivation and FRAP can be used as complementary techniques to determine mobility parameters.

In a technique related to FRAP (termed fluorescence loss in photobleaching; FLIP), a defined region of fluorescence within a living cell is subjected to repeated photobleaching by illumination with intense irradiation. Over a measured time period, this action will result in complete loss of fluorescence signal throughout the cell if all of the fluorophores are able to diffuse into the region that is being photobleached. By calculating the rate at which fluorescence is extinguished from the entire cell, the diffusional mobility of the target fluorophore can be determined. Furthermore, FLIP will readily identify the location and nature of any diffusional barriers between the individual compartments of a cell, such as the barrier between the soma and axon of a neuron.

The increasing use of multiple fluorescent proteins with highly overlapping emission spectra in live-cell imaging often requires unique solutions to provide adequate separation of signals from different targets. Spectral imaging relies on specialized hardware (primarily coupled to confocal microscopes) to separate the emission light into its spectral components. Linear unmixing is a computational process related to deconvolution that uses the unique spectral profile of each fluorophore in the specimen to reassign signal to the appropriate pixels in the final image. Although together these analytical tools can be employed to discriminate between distinct fluorophores having highly overlapping spectra, they operate at the cost of requiring that significantly more photons be detected from each pixel, which can be problematic for live-cell imaging of specimens labeled with dim fluorophores or low abundance targets.

Used primarily with laser scanning confocal or multiphoton microscopy, fluorescence correlation spectroscopy (FCS) is a technique designed to determine molecular dynamics in volumes containing only one or a few molecules, yielding information about chemical reaction rates, diffusion coefficients, molecular weights, flow rates, and aggregation. In FCS, a small volume (approximately one femtoliter; the diffraction limited focal spot of a laser beam) is irradiated with a focused laser beam to record spontaneous fluorescence intensity fluctuations arising from the dynamics of fluorescent molecules occupying the volume as a function of time (see Figure 10). Relatively small fluorophores diffuse rapidly through the illuminated volume to generate short, randomized bursts of intensity. In contrast, larger complexes (fluorophores bound to macromolecules) move more slowly and produce a longer, more sustained time-dependent fluorescence intensity pattern.

The inherent dynamics and spatial distribution of fluorescently labeled structures can be difficult to analyze when these entities are densely packed and overlapping within specific regions of living cells. Fluorescent speckle microscopy (FSM) is a technique compatible with almost all imaging modalities that takes advantage of a very low concentration of fluorescently labeled subunits to reduce out-of-focus fluorescence and improve visibility of labeled structures and their dynamics in thick regions. FSM is accomplished by labeling only a fraction of the entire structure of interest. In that sense, it is similar to performing FCS over an entire field of view, albeit with more emphasis on spatial patterns than on quantitative temporal analysis. Speckle microscopy (as it is often called) has been especially useful in defining the mobility and polymerization of cytoskeletal elements, such as actin and microtubules, in highly motile cells.

The major benefit of coherent anti-stokes Raman scattering microscopy (CARS) is that it enables investigations of biomolecules without the addition of synthetic fluorescent labels or endogenous coupling to fluorescent proteins. Instead, the technique is based on the vibrational properties of the target molecule and does not require the species to be electronically excited by ultraviolet or visible light. In practice, fast (picosecond or lower) laser pulses in the near-infrared region from two sources are focused onto the specimen with a microscope objective and raster scanned in the lateral and axial planes. The pulses are separated in frequency by a selected molecular vibrational mode and generate a new beam, which has a wavelength shorter than the incident beams, at the objective focal point. The secondary beam produces a concentration profile of the target species and enables the construction of a three-dimensional image of the specimen. Because most of the biomolecules found in living cells have similar building blocks, it is difficult to achieve real molecular specificity in CARS experiments. Furthermore, CARS is based on resonance, meaning its sensitivity is restricted by the fact that thousands of molecules must be present in the focal volume for a significant signal level to be produced. Therefore, CARS (in its present form) must target abundant molecules and may be restricted to relying on external labeling (such as the use of deuterated compounds) to provide real molecular specificity.

Harmonic generation microscopy is an emerging technique that may ultimately see widespread use in live-cell imaging. Harmonic generation occurs when an optical excitation event involving two or more photons at a particular frequency results in cooperative emission at multiple harmonics (primarily, the second and third) without absorption of the photons. Generation of the harmonic frequencies is essentially a non-linear scattering process yielding an emitted photon wavelength that is twice the frequency or half the wavelength (for second harmonic generation) of the incident illumination. In optical microscopy, transparent specimens that lack symmetry (such as living cells) are ideal candidates for imaging with harmonic generation techniques. Unlike the situation with typical probes and traditional fluorescence microscopy illumination techniques, changing the excitation illumination wavelength produces a corresponding change in the emission wavelength. In addition, the emitted light is coherent and retains phase information about the specimen.

For ultra-high optical resolution, near-field scanning optical microscopy (NSOM) is currently the photonic technique of choice. Near-field imaging occurs when a sub-micron optical probe is positioned a very short distance from the sample and light is transmitted through a small aperture at the tip of this probe. The near-field is defined as the region above a surface with dimensions less than a single wavelength of the light incident on the surface. Within the near-field region evanescent light is not diffraction limited and nanometer spatial resolution is possible. This phenomenon enables non-diffraction limited imaging and spectroscopy of a sample that is simply not possible with conventional optical imaging methodology. Although currently seldom applied to live-cell imaging, related techniques such as atomic force microscopy are being explored to examine the surface of living cells.

Exhibiting spatial resolution well beyond the diffraction limit, stimulated emission depletion microscopy (STED) is an emerging superresolution technique that uses a donut-shaped depletion beam surrounding a smaller excitation beam to achieve an axial resolution lower than 50 nanometers. The technique relies on inhibiting fluorescence of excited molecules at the periphery of a laser scanning focal spot using synchronized laser pulses for excitation of fluorophores and spatially coordinated circular STED pulses to deplete emission. Resulting fluorescence is inhibited at the periphery of the spot, but not in the center, thus dramatically reducing the fluorescence spot size with a concomitant increase in resolution. STED has been demonstrated to be a useful tool for examining living cells at high resolution. Other emerging superresolution techniques, such as photoactivated localization microscopy (PALM) and structured illumination microscopy (SIM) will probably become essential tools in live-cell imaging in the near future.

Conclusions

The increasing use of genetically-encoded fluorescent proteins and advanced synthetic fluorophores for live-cell imaging has opened the door to a wide spectrum of new optical modalities that are useful for monitoring temporal dynamics and spatial relationships. The microscopist now has a full complement of tools to view and record image data of cellular processes that occur over a large range of timescales and at multiple resolutions. Slower events are readily observed and recorded using laser scanning confocal microscopy, while the more rapid dynamics are accessible through the use of spinning disk techniques. Additionally, multiphoton microscopy enables imaging deep within thick tissues and total internal reflection techniques are able to probe the membrane surface with confocal precision. Advanced fluorescence methodology, such as FRET, FLIM, FRAP, FCS, FSM, SIM, PALM, and STED, can be used to monitor protein-protein interactions, often at resolutions greater than those allowed by the diffraction barrier. As fluorophore, microscope, and detector technology becomes increasingly more advanced, a still wider range of phenomena will be put "under the microscope".