Biocurious is a weblog about biology, quantified.

Fluorescence Nanoscopy Just Keeps Getting Better

by Andre on 2 April 2008

If you’ve been following the development of optical microscopy at all, you’re familiar with the recent advances that have been made in sub-diffraction limited optical microscopy. The new results keep coming fast and furious, pushing the field forward at an amazing rate. It’s interesting that the important advances are obvious even if their method of implementation is not. What I mean is that the goals of this research are clear: increase sensitivity, spatial resolution in 2D and 3D, temporal resolution, and the range of dye colours that can be simultaneously visualized. Of course, anyone can give you an important goal (fusion anyone? How about a time machine?) but having a practical approach is the challenge.

In sub-diffraction limited optical microscopy, there are currently two approaches that are driving things forward on all of these fronts.

One is the PALM/STORM* approach of Eric Betzig and Xiaowei Zhuang. PALM and STORM both take advantage of the fact that single fluorophores can be localized with nanometer accuracy when they are well separated, even if they can’t be resolved into spots when they are closer than the standard diffraction limit. This idea is turned into an imaging technique using photoactivatable dyes that can be turned on and off within the sample so that the dyes that are “on” at any given point are far enough apart to be localized before the next set of dyes are activated and in turn localized. By doing this procedure many times, the dye locations can be combined into a high resolution image:


The other approach is called Stimulated Emission Depletion or (STED) microscopy and has been pioneered by Stefan Hell and his group. It also takes advantage of photo-activatable fluorescence to beat the diffraction limit, but in a very different way. They use one beam to make a doughnut-shaped “depletion” patch and another to make an activating spot in the middle (we Canadians might say a Timbit excitation?). By increasing the intensity of the depletion beam, they can effectively squeeze the region where excitation is likely down to a spot smaller than the diffraction limit. This spot can then be scanned around a sample as in a confocal microscope to produce a nanoscale resolution image.

The image below compares the results of these two approaches to images taken using their most closely related diffraction-limited cousins.

STED/PALM comparisonThe image is taken from Hell’s excellent review of these techniques that appeared last year in Science (2007) 315:1153

Because PALM and STORM require the collection of a lot of speckled images to reconstruct the full distribution of dyes, it was clear that STED would have the advantage in time resolution.*** Now, in an article published online in Science [abstract, subscription required for full article], they’ve really driven that message home with a report of fluorescence imaging with 63 nm resolution at 28 frames per second.

“Video-rate imaging was accomplished by scanning the excitation and depletion beam pair in the focal plane by means of a 16 kHz resonant mirror in one direction; the second direction was scanned by moving the sample with a piezo actuator.” They also tightened their doughnut down to the point where the number of photons emitted from the hot spot was just big enough to provide the intensity needed to distinguish the signal from the background.

They used this method to image vesicles moving in neurons that are normally hard to see directly because they move in a nerve terminal that is only about 1 micron in diameter. As soon as you have a few vesicles passing each other in a space that size, a standard microscope (even a really good one) won’t be able to resolve them. Experts in the field can tell us if they learned anything interesting about vesicle traffic, but to me that’s not really the point. The paper is really about showing what can be done and getting cell biologists excited about the possibility of using STED for their favourite system.

It’s only a matter of time (I would say a few years at most) before this technology becomes available on a commercial instrument. When that happens and more biologists start to get their hands on one, we’ll really start to see its impact. In a huge understatement, Volker Westphal et al. start their paper with this:

Many questions in the life sciences could be answered if lens-based optical microscopy featured the resolution of electron microscopy, or if the electron microscope operated under physiological conditions.

*PALM stands for photo-activated localization microscopy. STORM stands for stochastic optical reconstruction microscopy.

**The statistics of my schematic are way off. In my animated gif it is, conveniently, exactly the needed remaining pixels that are filled in. This is especially important in the last couple of frames. In a real reconstruction, because it is based on stochastic activation of dyes, really filling out the image at the end takes a very long time. This is one of the things that limits the time resolution of the technique. I think the first demonstrations of PALM and STORM used acquisition times of over an hour on fixed samples to get enough data for their images. These times have now been shortened, but video rate seems like a long shot.

***But people are clever and there could always be surprises.

  1. sam    3760 days ago    #

    Thanks for this post. Cool animated GIF! :)

    I have a few review posts about breaking the optical diffraction limit:

    And a couple with updates:

  2. Mike Beversluis    3760 days ago    #

    These are neat techniques, but keep in mind that the sample prep is pretty similar to what you need to do for an environmental SEM, which sort of begs the question.

    The last talk I went to on STED was neat because they were using a CW laser, but it took ~200 mW (!) of visible light focused down to a ~300nm spot to get the maximum resolution. And that was with fluorospheres, which are really bright. I suspect bleaching and cell destruction can be an issue at those wavelengths (needed for the depletion beam) and powers.

    I keep wondering when another group will adopt the STED approach, but getting all of the beams to overlap in a 4pi microscope seems really technically challenging.

  3. Andre    3760 days ago    #

    Thanks Sam. You inspired me!

    I don’t know much about environmental SEM, but I assume you’re getting images only of the sample surface which is a different story (still potentially interesting, of course). More importantly, this most recent example was done on living cells.

  4. sam    3760 days ago    #

    Mike Beversluis, your point about the sample prep (basically killing the cells and slicing them) is an important one. however, i know labs that are currently doing PALM/STORM on living cells!

    and STED is really hard. there are ways around photobleaching from the high powers (i.e. modulating the lasers with an AOM, for instance, AKA “T-REX”). and the Hell lab has already done some live-cell STED (in an FCS mode) on membrane dynamics.

    none of these techniques is a golden bullet, but they’re great to add to our bag of tricks!

  5. Andre    3760 days ago    #

    Wait, a “golden bullet?” Isn’t it a silver bullet? agrees with me. :)

  Textile help