by Sam Tesfai and John Jordan

(Click image for larger version.)
Figure 1. Comparison of image collection in single photon vs. multiphoton systems. Fluorescent dye is excited by single photon excitation (left image) and multiphoton excitation (localized spot indicated by white arrow on the right), illustrating that two-photon excitation is confined to the plane of focus.

Localizing the excitation fluorescence and using near-infrared wavelengths, researchers can achieve deep-tissue imaging with significantly reduced photo-damage and significantly enhanced imaging results. Even such difficult specimens as ultra-thick brain slices, eye tissue and developing embryos can be imaged.

First, many fluorescence systems rely on shorter-wavelength blue and ultraviolet (UV) light, which are absorbed by fluorophores and emitted as longer wavelengths. But light from this region of the spectrum doesn't penetrate living tissue easily. In contrast to fixed tissue, which in some cases can be imaged several hundred microns deep into the specimen, living tissue scatters so much light that much of the emitted fluorescence from the region of interest may never reach the detector. The deeper into the tissue the area of interest is, the more light scatter occurs. This often limits the collection of a full three-dimensional data set when working with live specimens. In fact, for every specimen, at some depth, there is a point where there is so much scatter that traditional fluorescence imaging techniques are no longer effective.

When living cells are far below the surface, researchers sometimes must raise power in order to excite the fluorophores. But raising the intensity of the excitation light can cause photobleaching and phototoxicity, which can be lethal to the specimen over time; the greater the light intensity, the worse the negative effects. Adding to the problem is the fact that in deep imaging, regions of the specimen above and below the focal plane that are not of interest are exposed to light, causing unwanted fluorescence. Finally, excessive scattering of the excitation light when imaging deep below the specimen's surface results in an image with poor signal-to-noise ratio and poor contrast. Where such degradation occurs, the resolution and contrast of images are lower, and the images tend to look flat and dull instead of crisp and full of contrast.

Imaging options

Figure 2. Vertical cross sections of the mouse cerebellum molecular layer (left two images) and granular layer (right two images) marked with DiI and obtained using single-photon (488 nm confocal) and two-photon (720 nm) excitation. Multiphoton excitation's superiority to confocal observation is apparent in deep observation of brain slices. Image courtesy of Yasuyuki Hayakawa, Haruo Kasai, Center for Disease Biology and Integrative Medicine, Faculty of Medicine, University of Tokyo. (Note to editor: Two versions of Figure 2 are attached here. The larger file size bitmap has the image parts only. All words --or at least "Single photon" and "Two photons" -- should be inserted below the image as they are shown in the smaller bitmap version. Please call me if you have any questions - Thanks, Ilene (914) 684-0959; ilene@edge-comm.net.)

Confocal microscopes are another solution. They are used for three-dimensional fluorescence imaging of living specimens because they are designed to optically remove out-of-focus light coming from above and below the focal plane, reducing background haze and also reducing the thickness of optical sections. Because the excitation light generates fluorescence throughout the entire depth of the sample, confocal microscopes can damage or bleach the entire volume of the specimen not only at the plane of focus, but above and below it as well. When sequentially imaging down through a three dimensional sample, the effect is sometimes to bleach and damage the lower regions before the researcher works his way down to them.

Another potential issue with confocal microscopy occurs in collecting the emitted signal of the fluorophore. The emission must travel from the sample, through the microscope optical system, and then be precisely directed through a small confocal aperture. Any deflection or scattering of the emitted fluorescence in the sample results in its rejection by the confocal aperture, and the further emission photons have to travel (from deeper within the specimen), the higher the odds that they will be scattered. In addition, a certain percentage of out of focus light that should be rejected will get scattered into the confocal aperture. This also contributes to degradation of the signal-to-noise ratio, and leads to less than optimal images. Thus, while confocal microscopy holds an extremely important place in three-dimensional imaging, there are trade-offs when attempting to view living specimens, especially when imaging very deeply and/or for long time periods. This is just as important as the excitation side of the equation in evaluating the quality of the overall imaging system.

Multiphoton advantages

Figure 3. Multiphoton image of a Living zebrafish: CFP and YFP expression during embryonic eye devolvement (25×, 1.05 NA objective). Image courtesy of Phillip Williams and Rachel Wong, University of Washington.

First theorized in the 1930s as part of a doctoral dissertation by Maria Goeppert-Mayer, one of only two women ever to receive the Nobel Prize in physics, multiphoton microscopy was not developed in a practical form until decades later. (The technology was patented by Winfried Denk, James Strickler and Watt Webb at Cornell University.) When enough photons are present at high enough densities, two or more excitation photons can be simultaneously absorbed by a single fluorophore, combining their energies to bring the fluorophore to an excited state. For instance, one can excite DAPI, a frequently used UV-excited stain, using 780-800 nm light. GFP, typically excited in fluorescence at 488 nm, can be excited using multiphoton imaging at wavelengths ranging from 850-960 nm. Whereas traditional fluorescence results in fluorophore emissions at longer wavelengths than the excitation light, the nearly simultaneous absorption of two longer-wavelength photons allows the fluorescence emission to occur at a shorter wavelength, typically in the visible light range.

Because multiphoton excitation requires the nearly simultaneous absorption of two or more photons, anything that increases photon density will increase fluorescence. This is why a burst of short pulses of highly focused light is directed at the focal point. A laser of such high peak energy is used that, during its femtosecond pulses, two photons from the same pulse excite a fluorophore essentially simultaneously, as if they were a single, short-wavelength, high-energy photon. The key is that the statistical chance of two or more photons interacting with the same fluorophore molecule occurs only at the plane of focus where there is a high density of photons. Thus, fluorescence excitation is effectively limited to the plane of focus, eliminating photobleaching and phototoxicity in the areas above and below the focal plane.

By successfully localizing the excitation fluorescence, and by using near-infrared wavelengths, researchers can achieve deep-tissue imaging with significantly reduced photodamage and significantly enhanced, higher contrast imaging results. Even such difficult specimens as ultra-thick brain slices, eye tissue and developing embryos can be imaged (Figure 3).

As with confocal microscopes, three-dimensional multiphoton imaging builds stacks of individual sections from each optical slice collected at sequential z-axis locations. But, though multiphoton microscope systems have much in common with confocal instruments, they are not identical. One key difference is that multiphoton systems do not use confocal apertures in front of the detectors. Instead, optical sectioning is accomplished as a result of the excitation process.

Benefits in emission detection

A third area of differentiation involves how deep within tissue the systems can accurately image. A good multiphoton system can image efficiently from 2 to 10 times deeper within a given specimen than a comparable confocal system. The thicker and deeper the living specimen's area of interest is, the more dramatic this contrast becomes.

Pulse width and resolution issues

Figure 4. Three-dimensionally constructed images of neurons expressing EYFP in the cerebral neocortex of a mouse under anesthesia. Cross-sectional images down to 700 microns from the surface can be observed. Image captured using the Olympus Fluoview FV1000-MPE multiphoton system with a 60x objective by Hiroaki Waki, Tomomi Nemoto, and Junichi Nabekura, National Institute for Physiological Sciences, National Institutes of Natural Sciences, Japan.

Another limiting factor is pulse width. The energy of multiphoton excitation is improved by employing more intense, shorter pulses of light. Two or more photons must be absorbed simultaneously to excite a fluorophore, and fluorescence is generated only where the laser beam is tightly focused (the plane of focus). Pulse width is a determining factor in efficiency of excitation, and most Ti:sapphire lasers have short pulse widths (such as 150 fs out of the laser). However, the pulses in most commercial multiphoton systems are inadvertently broadened by components in the microscope itself, such as lenses and power control devices. This broadening degrades the efficiency of dye excitation. Most recently, Olympus has developed a way of maintaining shorter, more tightly focused pulses with the FV1000-MPE system, optimizing the excitation efficiency of the system. By utilizing a process known as negative chirp, a specialized type of dispersion compensation, the laser's pulses are conditioned and kept short at the sample, further reducing the amount of average power necessary for imaging, which leads to less specimen damage and longer-term imaging of living tissue. The system incorporates the use of a femtosecond pulsed laser to achieve photon density at the focal plane, and is optimized for use with near-infrared wavelengths. Negative chirp ensures that the packet of light doesn't arrive already broadened. By being more efficient with the pulse, the system keeps power down, efficiently delivering a tight packet of light to the focal plane and resulting in the maximum return for the light placed into the system.

Perhaps the most often described limitation of multiphoton systems is their resolution, which generally has not been as high as with confocal systems. This assessment is based on a comparison with theoretical confocal systems that are only wavelength dependent. However, in the real world, the deeper the specimen, the less relevance the theoretical difference has. Indeed, being able to avoid confocal degradation of the image in deep scans means that investigators can image 2 to 10 times as deeply into the tissue as is possible with even the best confocal systems. There are also the advantages of the enhanced signal-to-noise ratio and enhanced image contrast. In other words, multiphoton imaging is optimized for situations where scientists need to see deeply and clearly. The Olympus FV1000-MPE system has resolved images up to 700 microns deep in brain tissue — hundreds of microns deeper than possible using confocal microscopy — and every slice has the optical resolution necessary to see the neuronal spines (Figure 4).

Multiphoton microscopy is a specialized methodology that allows imaging living tissue deeper and with less damage than ever before. Now that it has been commercialized by several companies, the technique is coming into much more widespread use, especially for repeated, long-range or time-lapse exposures. Recent advances in multiphoton technology have had the effect of encouraging more researchers to move into the world of live tissue imaging.