The high speed of neuronal signaling and the complex three-dimensional (3D) structure of individual neurons and their networks complicate the study of in vivo neuronal functionality. Conventional laser scanning microscopy is too slow to image signaling in a 3D volume. Although clever schemes have been designed to greatly increase this speed, most of these methods rely on physical movement of some combination of beam-scanning mirrors, microscope objective or biological sample.

There are practical limits on the speed with which physical objects can be moved, which limit the imaging speeds of these physical inertia-based systems to below 100 Hz. Although such speeds allow many valuable biological experiments to be performed, there is a strong desire for speeds in the kilohertz range that would allow effectively simultaneous measurements at multiple points.

Recent years have witnessed the introduction of inertia-free acousto-optic deflectors (AODs). These devices use sound waves of a fixed frequency to control the deflection of the excitation laser in a laser-scanning microscope. Inertia-based systems are limited to steering a beam along a continuous path through the sample. But AOD-based scanning is not restricted to continuous path scans. Because sound frequency is used to control beam deflection, the beam can be moved almost instantly by a desired distance. The user can choose to either emulate a continuous path scan or operate in a 'random access' mode by jumping to specific points and staying at each one as long as desired to take a measurement before moving to the next point.

Peter Saggau's lab at Baylor College of Medicine is one of the pioneers in the use of AODs for microscopy. Most previous work has been limited to using a pair of AODs to position a beam in a lateral plane, but in 2005 Saggau demonstrated that AODs were not limited to lateral scanning (Reddy & Saggau, 2005). By applying counter-propagating sound waves that constantly varied in frequency to a pair of AODs, they effectively created a cylindrical lens of variable focal length and thus could control the axial focal position of the scanning beam. It turns out that it is possible to use two such pairs of AODs laterally and axially without interference between the two.

To demonstrate the power of this scanning method they built a random-access multiphoton microscope with a four-AOD 3D scanning system and used it for high-speed 3D imaging of dye-filled pyramidal neurons in hippocampal brain slices (Reddy et al., 2008). They compared the imaging performance of AOD-based lateral scanning coupled with axial movement of the objective to fully AOD-based 3D scanning and showed there was a minimal performance difference between them.

Of course, you do not need high speeds to image static neurons in three dimensions. The value of the method is in its ability to sample multiple user-defined points on a neuron fast enough such that the sampling is effectively simultaneous. To demonstrate this capability, the researchers filled neurons with a calcium dye by patch pipette and acquired a full 3D image. This allowed them to select points of interest along neuronal dendrites. They then induced a short train of action potentials in the cell soma and measured the calcium response at the selected sites at speeds of up to 10 kHz. This showed clear reductions in the calcium response at points farther from the cell soma (Fig. 1).

Figure 1: Fast 3D imaging of dendritic calcium signals by random-access multiphoton microscopy.
figure 1

Image of a pyramidal neuron visualized in three dimensions with superimposed calcium transient graphs color-coded to indicate the axial depth of the recording site over a 30-μm range. Reprinted from Nature Neuroscience.

Full size image

Presently the effective axial range of the method is limited to 50 μm when using a high-magnification objective, so the system is not exploiting the full capabilities of multiphoton lasers. According to Saggau, some design changes already underway should substantially increase this range. While this system is already useful, when this limitation is overcome, the resulting system will offer a powerful tool for examining neuronal signaling in the intact brain.