Cryo-electron microscopy: A prime example of a tool-driven scientific revolution

Last week I had the immense pleasure again of having lunch with Freeman Dyson in Princeton. One of the myriad topics on the platter of intellectual treats on the table was the idea of science as a tool-driven rather than as an idea-driven scientific revolution. The framework was fleshed out in detail by Harvard historian of science Peter Galison in his highly readable book "Image and Logic" and was popularized by Dyson in his own book and article. I wrote a post on that particular paradigm last year.

Since physics had profited immensely from idea-driven revolutions in the 20th century (most notably relativity and quantum theory) that were enshrined by Thomas Kuhn in his idea of paradigm shifts, it took physicists some time to appreciate how tools like the cyclotron, the cloud chamber, the CCD and the laser have played an equal part in their revolutionary history. But as I told Dyson, chemists on the other hand have absolutely no problem accepting the idea of tool-driven revolutions. Chemistry more than physics is an experimental science where first principles theories are often too complicated to put into practice. Chemists have thus benefited much more from experimental toys rather than fancy theorizing, and in no other case has the ascendancy of such toys been more prominent than in the case of x-ray crystallography and NMR spectroscopy. It's hard to overstate how much these two techniques have revolutionized not just our understanding of the world of molecules but of other domains, like biology and engineering. Last year's Nobel Prize for microscopy was likewise a fitting tribute to the supremacy of tools in chemical and biological research.

Now a new technique joins the arsenal of structural weapons, and I have little doubt that it too is going to be part of a revolution - cryo-electron microscopy. ACS has a nice article on how much the technique has advanced in the last decade and how prominently it is poised to be applied to structural problems that have been recalcitrant to the old approaches. During the last few years use of the technique has skyrocketed: as this Nature article compellingly describes, cryo-EM can acquire structures of ribosomes in weeks or months that took Nobel Prize-winning scientists years to solve. And as the article says, even this revolution has benefited from a crucial tool-within-a-tool.

Over the years, gradual progress in computational power and microscope quality has yielded higher and higher resolution structures. Up until the past few years, most cryo-EM structures clocked in at well above 10-Å resolution, about the size of an amino acid. Between 2002 and 2012, only 14 structures determined by EM crossed the 4-Å threshold, dipping a toe in high-resolution territory. But a true breakthrough came in 2012 when a new toy—the direct electron detector—opened the gates, allowing for a flood of high-resolution cryo-EM structures. In 2014 alone, 27 structures have reached sub-4-Å resolution, and scientists keep pushing the boundaries. “The direct electron detector has been the biggest game changer for the electron microscopy field,” says Melanie D. Ohi of Vanderbilt University.

The direct electron detector joins a long list of specialized instruments like the Bunsen burner, the Kirchhoff spectroscope, the scintillation counter and the Geiger counter, all of which proved to be key appendages of the larger technologies which they were enabling. A good counterpart to the direct electron detector would be the CCD which revolutionized tools like cameras and telescopes and which was awarded a Nobel Prize a few years ago.

Cryo-EM will almost certainly make a big splash in the world of drug discovery in the upcoming decades. However, better experimental tools alone won't suffice for this revolution. It's sometimes underappreciated how important software and hardware were in enabling the routine application of NMR and crystallography to tough biological problems in drug design. The advent of cryo-EM similarly opens up attractive opportunities for the development of specialized software and hardware that can handle the often fuzzy, low-resolution images coming out of cryo-EM. This will especially be important for multiprotein assemblies like modular enzymes and ribosomes where multiple solutions exist for a given dataset and where computational model building will be paramount. 

As the technique proliferates, so will the data that it unearths. Someone will have to then make sense of this data, and scientific and financial rewards will await those who have the courage and foresight to found companies making specialized software for analyzing cryo-EM images. The founding of these companies with their custom hardware and software will itself be a paean to the tool-driven revolution in science, in this case one led by the computer. One tool both piggybacking on and enabling another tool, that's how science progresses.

Added: Here's a nice application of cryo-EM in resolving crystals of the protein alpha-synuclein that are essentially 'invisible'.

Image source