Microscopes: why seeing smaller is not always better

How do fluorescence microscopes work?

Early in the 20th century, the phenomenon of fluorescence was applied to microscopy. Fluorescence is a luminescence phenomenon. Usually we see objects when light is reflected from them - the colour of an object depends on what wavelengths it reflects. With fluorescence, a photon (a light 'particle') of certain wavelength is absorbed by a molecule, and then re-emitted at a longer wavelength.

Fluorescence is a very commonly used technique in biological imaging. Biological materials usually scatter a lot of light making it difficult to see beyond the surface of the cell. With fluorescence, the emitted light is always longer wavelength than the excitation light, so the light scattered from the cell surface can be separated from the emitted fluorescent light using dichroic mirrors that reflect the excitation light into the sample but let through the fluorescence light, making it possible to see structures inside the cell.

Some biological materials are naturally fluorescent, but there are also many fluorescent dyes and proteins available that can be used to highlight specific parts of a cell, for example the nucleus, or they can be attached to specific proteins in cells so that it is possible to follow their movement inside the cell.

Recently discovered photoswitchable fluorescent dyes and proteins have many applications in fluorescence imaging. These molecules can exist in two states: a bright, fluorescent state, and a dark, nonfluorescent state. The switching between these states is done by irradiating the molecules with two different wavelengths of light.

One application of photoswitchable molecules is protein tracking. If the fluorescent molecules are attached to a specific protein, and a small part of them are activated, it is easier to follow where the proteins move than having all the proteins in the cell emitting light. Also, the exact moment of the activation can be controlled.

How can we produce a high resolution microscope which won't kill our biological samples?

Since the resolution of a light microscope depends on the wavelength, an obvious way to improve resolution is to decrease the wavelength. However, as we move from the visible spectrum towards the ultraviolet (UV) spectrum, the light becomes toxic to living materials. Even the least harmful UVA radiation has the ability to break bonds in DNA causing mutations and stopping the cell to function in a normal way.

Resolution improvement in the z-direction (depth, essentially) has been achieved by the use of two opposing objective lenses. Because of the increased angle from which light is collected, resolution of about 100 nm is achievable. However, resolution is improved only in one direction, and this technique suffers from technical difficulties, such as keeping the objectives accurately aligned.

Photoswitchable molecules could make fluorescence imaging possible in nanometre scale with living samples. If the molecules are switched on in a small spot on the sample, and another doughnut-shaped beam is used around it to switch off the molecules, the effective spot where from where fluorescence is emitted becomes much smaller. In fact, resolution on the scale of tens of nanometers has been achieved by Stimulated Emission Depletion Microscopy (STED), which is based on similar principles, but so far this has not shown to be generally compatible with live cell imaging. If photoswitchable proteins or dyes are used, high intensities are no longer needed.

Alternatively to scanning a spot across the sample, a grating pattern can be projected onto the object to squeeze the fluorescence into thin lines. The grating pattern can then be scanned across the sample. Although several images are required to construct the final high-resolution image, this approach still makes the data acquisition quicker than the process of scanning a spot across the object.

Yet another approach for nanoscale imaging with photoswitchable dyes and proteins is to first switch off all the molecules in the sample, then adjust the activation intensity so that only few molecules are switched on. Depending on the brightness of the molecules, the centroid position of the molecules can calculated with the accuracy of few tens of nanometers. After imaged, the molecules can be switched off, and the process can be repeated by switching on different molecules. The final image can be reconstructed by combining a stack of these images. The drawback of this approach is that by imaging only a few molecules per image, thousands of images are required for the final high resolution image, making this technique presently too slow for imaging living samples.

Although most of these superresolution techniques are not yet commercially available, this could change quickly in a matter of years.

For more information

Invitrogen.com - Fluorescence Tutorial
http://probes.invitrogen.com

NobelPrize.org - Fluorescence Microscope
http://nobelprize.org