Bench philosophy: Light-sheet microscopy

Sliced by Light
by Steven Buckingham, Labtimes 01/2015

Illuminating your specimen with a light-sheet means you can get a big increase in image resolution, and get deeper penetration, quicker scans and lower toxicity with it. All with open source hardware and software.

There is a theoretical limit to the resolution of light microscopy. Way back in 1873, Ernst Abbe broke the sad news that the best resolution you can possibly get from any microscope is determined by the diffraction properties of the light you use. To the first approximation, the minimum distance you can resolve is more or less half the wavelength of the light. One of Nature’s non-negotiables, it seems.

Of course that’s why we had to invent electron microscopy. If wavelength is what’s causing the problem, then let’s switch to “light” with a really small wavelength. A beam of electrons has a wavelength about 100,000th that of visible light and that adds up to a big leap in resolution.

Ernst Stelzer (l.) and Jan Huisken pioneered the SPIM technology for life science applications. Photo: EMBL Heidelberg

The only problem is that using an electron beam means you can only see things that are, or have been made, electron dense. Useless for all those amazing applications with fluorescent tags. And as for imaging living cells, just try putting your tissue into an electron microscope and see what happens.

But it turns out you can get around Abbe’s limit with a little bit of light trickery and some clever computing. Selective plane illumination microscopy (SPIM) is one such hack that allows you to ramp up the resolution of your images without running the expense of a confocal or two-photon setup. And because of the way SPIM works, several extras come included as standard: you can image large tissues (even whole living animals) and you can image over a long period. What is more, you can even convert your old fluorescent scope into a SPIM setup using materials found in your kitchen (provided you keep the right sort of things in your kitchen).

Fluorescence microscopy’s downsides

So how does SPIM work? Let’s start from the beginning. The basic problem with standard fluorescence microscopy is that you illuminate the whole specimen, even though you are only focussing on one tiny part. Doing that introduces all sorts of problems. For one, you get a lot of background from the out-of-focus regions above and below your plane of focus. Second, irradiating a tissue with high energy electromagnetic radiation does a good job of slowly micro-waving the specimen, placing a severe limit on how long you can image before the tissue hits medium-rare. That is a big problem when your signal is weak and you need a long time to overcome a poor signal-to-noise ratio.

Sure, you can get around a lot of these problems with confocal imaging, where a beam of light converges on a point in the specimen. But this comes at the expense of a poor axial (z-plane) resolution, not to mention great expense.

SPIM solves these problems by the simple expedient of shining an ultra-thin sheet of light through the specimen. The sheet lies at a 90° angle to the observing objective, so light dispersed by stained objects is detected by the objective. The sheet of light itself is created from a dispersed laser beam that is focused through a cylindrical lens. Alternatively, the equivalent of a sheet can be achieved by scanning a circular beam. Effectively, the specimen is optically sliced.

So what does this simple approach buy us? First, you get better contrast images. Gone is the background that mars conventional epifluorescence images. And given the clarity of each slice, moving the sheet through the sample, along the z-axis results in a better-quality, reconstructed 3D image.

Then there is the speed factor. The lack of background means you can get a good quality image from a very rapid scan, so you can image things that traditionally move too quickly. Last year Jan Huisken at the Max Planck Institute of Molecular Cell Biology and Genetics in Dresden used SPIM to capture the first high-resolution images of a beating heart of the zebrafish (Mickoleit et al. Nature Methods doi:10.1038/nmeth.3037). This was made possible not only because of the technique’s speed but also because of its ability to penetrate thicker (>1 cm) tissues.

Image stacks

There are other advantages, too. With SPIM you can easily rotate the sample, keeping the imaging hardware fixed. The significance of this is that you can build z-stacks from different angles, which in turn (with a bit of computer trickery called “multi-view fusion”) gives much better-resolved reconstructions.

But does SPIM really give you super-resolution? Well, no at least not on its own. The lateral resolution of a SPIM setup is still limited by the diffraction limit and the numerical aperture of the lens.

But some recent reports have changed all that: it seems you can get genuine super-resolution microscopy if you fiddle about with the fine details of the beam. And these developments are causing quite a stir in the field. Why? Because along with the super-resolution you get super speed thrown in for free.

Ultra-thin light-sheet

In 2011 Eric Betzig’s group at the HHMI Janelia Research Campus, USA showed that by making your light-sheet using a specially constructed beam (a “Bessel” beam), you can get notably higher resolution. Last year Betzig received the Nobel Prize in Chemistry together with Stefan Hell and William Moerner for “the development of super-resolved fluorescence microscopy”, which brings “optical microscopy into the ­nanodimension”. And even while Betzig was waiting to hear about his prize, his group was busy publishing a paper in Science showing that tweaking the light-sheet further − using a non-diffracting “lattice beam” − could push speed and resolution even further up and toxicity even further down (Bi-Chang Chen et al., Science 346, 6208: 1257998).

With all these high-tech optics, it may come as a surprise that the idea behind SPIM is not at all new. In fact, the basic idea was described as far back as 1903 by Henry Siedentopf and Richard Zsigmondy, and was given new impetus by Voie and co-authors in 1993, who used the idea of a light-sheet to optically section an entire cochlea for the first time (Voie et al., J. Microsc. 170, 229-36). A period of exploration followed, as experimenters toyed with different configurations of lens and illumination, until Ernst Stelzer and Jan Huisken, then at the EMBL in Heidelberg, developed the first SPIM microscope suitable for life science applications in 2004.

Whole bunch of applications

So how is SPIM being applied? It is one of those foundational techniques that gets the imaginative juices flowing in all sorts of directions. For one thing, not frying your specimen every time you image means you can image for longer − you can track developing embryos at the cellular or even subcellular level (remember what we said about tissue penetration?).

Then again, SPIM’s ability to penetrate tissues is especially interesting for 3D culture. There is an increasing awareness of the importance of structure for cultured cells. Where once we were happy growing cells on the flat surface of a Petri dish, consensus is emerging that cells don’t grow right unless they grow in 3D. Fortunately, many cells are happy to grow in agarose, which is also optically clear. That, coupled with the enhanced penetration offered by SPIM, opens up a host of possibilities.

But it is not just about anatomy −physiology is getting a boost from SPIM, too. In 2013, Georges Debrégas’ group at the University of Paris used zebrafish, genetically modified to express a calcium indicator (GCaMP3), to allow the activity of up to 5,000 neurons to be recorded simultaneously at 20 Hz. The resolution offered by the SPIM approach allowed single-cell (or at least near single-cell) activity to be monitored (Panier et al., Frontiers in Neural Circuits 7:65).

Eric Betzig (r.) and his group took SPIM to the next level.

The excitement about the neuro-physiological potential of SPIM is so intense that Phillip Keller and Hans-Ulrich Dodt, from the HHMI Janelia Research Campus, USA and the Vienna University of Technology, respectively, boldly declared two years ago that “Even samples up to the size of entire mammalian brains can be efficiently recorded and quantitatively analyzed” (Keller and Dodt, Curr Opinion Neurobiology 22, 138-43). Optimistic, perhaps, but last summer Keller’s lab reported that they had imaged all the neurons in the brain of zebrafish swimming in a virtual reality setup, using the SPIM light-sheet to prevent illumination of the retina (Vladimirov et al., 2014, Nature Methods 11, 883-84).

As with any technique, SPIM is not perfect. Slicing with a light-sheet causes problems when the beam hits an optically dense target, causing a stripey pattern in the downstream illuminated areas, although there are techniques to overcome this. Then again, things go wrong when a structure has a significantly higher refractive index, resulting in an unwanted focussing of the beam around them. And having to access the specimen from two angles (the light-sheet and the objective at 90°) is a bit of an imposition if you are using very large tissues.

Thinking of giving SPIM a spin but don’t have a major research grant to spend on a new setup? No worries − you can build one yourself. At least that is the belief of the people behind the OpenSPIM project (, who think that “every lab should have one”. The site offers the plans and even the complete assembly instructions for building your own SPIM system that will occupy only 300 x 450 x 150 mm of space. They also offer a free and open source programme to control the system through an Arduino (an open-access hardware controller).

DIY-SPIM project

As has been the case with open source software, it seems that open source microscopy is spawning a trend for do-it-yourself “hacking” (in the good sense of the word). For instance, we have mentioned how Betzig used a Bessel beam to get a thinner light-sheet but Bessel (and its alternative “Airy”) beams need some pretty expensive components to make. To get around that, Kishan Dholakia’s lab at the University of St. Andrews showed a slight adjustment to the OpenSPIM setup − by deliberately (but judiciously) misaligning the cylindrical lens! (Yang et al. 2014 Biomed Opt Express 5, 3434-42). And if you want to see how to set up a SPIM system and use it to image the nematode, Caenorhabditis elegans, take a look at where Claire Chardès from the Institut de Biologie du Développement de Marseille, has posted a series of, how-to, videos.

Whether or not the open source hardware and software approach will bring SPIM to the masses, SPIM and other light-sheet applications continue pushing the boundaries of live-cell imaging to a new ­level.

Last Changed: 04.02.2015