Global bottleneck holding up discovery using cryo-EM is solved by MRC Laboratory of Molecular Biology in Cambridge
Put together by hand using tweezers, the process of creating each grid for cryo-electron microscopy is a laborious one. But the world’s laboratories need a million a year, leading to a huge backlog. Step forward Cambridge’s
MRC Laboratory of Molecular Biology where the whole process has been reimagined from the ground up.
The use of cryo-electron microscopy has grown rapidly in recent years, helping us to make key discoveries about human biology and accelerate the development of new drugs.
Despite the substantial price tag of the machines, which can run into the millions of pounds, and the significant running costs, their use in determining three-dimensional structures of proteins has led to many more of these microscopes being adopted by laboratories around the world.
But scientists in these labs are being frustrated by global supply issues surrounding a small, but essential, consumable at the heart of this technique.
Currently, it can take half a year or more for an order for grids – used to hold your specimen during flash-freezing – to arrive. That is because they are made by hand, and a person armed with tweezers must handle each one several times before it is ready.
At the MRC Laboratory of Molecular Biology in Cambridge, which has pioneered the use and development of cryo-EM, Chris Russo’s group in the LMB’s Structural Studies Division is focused on improving the technology around it.
For their latest work, Chris and Katerina Naydenova, a PhD student in his group, have developed a manufacturing process that will solve this bottleneck for good.
“Cryo-EM has gone from this niche technique that only a few people were doing, to being properly mainstream biology,” Chris tells the Cambridge Independent.
“This is the primary method of determining structures now. This is a big change and our lab has been at the forefront of doing this for many years now.
“We have projects on various aspects, such as trying to reduce the cost of the microscopes for the process and that is well under way and has been quite successful already.”
Indeed, the lab has already built its first prototypes of affordable cryo-EM microscopes.
“Other companies are doing this as well, with our encouragement and cajoling, and they are starting to sell them,” Chris continues.
“We’ve got to a state where the number of cryo-EM microscopes in the world has increased probably by a factor of 10 in the last five years.
“The throughput of each of these microscopes has also gone up by an order of magnitude I would say. They have automated systems in them. You can load grids in and out of them quickly, so we are on an exponential trajectory for cryo-EM.
“So far there is no sign of it tailing off, although eventually it must.
“We’ve reached the point where there are a lot of microscopes distributed throughout the world and there is a lot of need for determining structures and this is supplanting the current technology, which is crystallography.
“But a key part of this process is the grids that you put the specimens on. Until now, they have been made by hand one at a time. There is a person in a factory with a pair of tweezers, picking up a grid and putting it down. They have to manually touch it four or five times just to create each little device for this structure determination work and they cannot keep up.
“They keep trying to hire more people and tripled the size of the factory but it just doesn’t scale fast enough to meet demand.
“So the current state is that everyone is waiting from six months to a year to get their grids. It’s a dire state.”
The LMB is among those waiting on large orders of grids.
“We knew this would keep happening and we warned the companies,” explains Chris.
“We realised there was no way they would do it, so Katerina and I decided that we need to sort this ourselves.”
Chris’s group had already developed a technique for improving the grids themselves, which solved the issue of specimens moving slightly when electron beams are fired at them to take the 2D images, from which a 3D structure is built up.
In 2020, in a paper published in Science, they described how to make improved all-gold grids called HexAuFoil, which completely eliminate the movement of the specimen during imaging.
“We had developed new grid technology which we knew everyone would want. It gives you better images for everything. The demand for this is going to be enormous – even more than the original gold grids that we developed six or seven years ago now.
“So we said let’s rethink the whole thing from the ground up. How do we start with raw materials and end up with a scalable process that can make hundreds of thousands of grids?
“Sitting in the canteen two years ago, we asked how many does the world need?”
Currently, about a million grids a year are required. By 2025, as the technique continues to be adopted and the costs come down, that is expected to rise to two to three million a year.
“We needed a process that enables a company to be able to make about 10,000 a day,” said Chris. “You just can’t do that with people handling grids one at a time with tweezers.
“We know how to do semiconductor processing, so we thought we would work it out.
“The surprise was that when we looked in the literature, the process for making the grids themselves was commercialised so long ago that no one knew how to do it except certain companies, working in areas like photographic film, and they were nothing to do with the cryo-EM world.
“So we reverse engineered the whole thing from scratch using basic semiconductor processing, MEMS technology, microlithography and set up a small clean room here in our biology lab to do all this processing and we worked out all the details.
“The process is complicated, but we sorted it all out, so now we can make these things by the thousands.
“Katerina can make more in a couple of days than a company can with a staff of 12.
“It’s going to change the whole dynamic and hopefully eliminate this bottleneck as well as providing this new grid design for the world.”
The technique has been licensed to Quantifoil in Germany, a sister company of Melbourn-based SPT Labtech.
“We are thrilled because we get to work with a local company and they have four engineers here in Cambridge who are taking this process and commercialising it, with all the quality checks.
“They’ve been coming to the lab and reproduced the whole thing in-house with Katerina watching over their shoulder. Now they’ve ordered the equipment and are in the process of rolling it out.
“We hope that within six months they will be manufacturing these things in Cambridge and, as they scale, will use the factory in Germany they bought as well to ramp up production.”
When the Cambridge Independent caught up with her, Katerina was awaiting her flight home from the US, where she had been presenting the technique.
“The semiconductor processing is super simple, but the challenge was to define the pattern of the small holes that are required for the grid. This uses a different method of lithography developed by some people in Switzerland. We got in touch to get some starting materials.
“Something nobody had figured out was how to replicate the pattern of small holes using a metal because the metals are crystalline. If you try to outline very small nanometre-scale circles using a crystalline material it doesn’t work – it doesn’t form a circle, so we developed a different process for depositing this metal, such that the size of the crystals is kept very small compared to the size of the holes. That was one step of the process that was new.”
Solving this was key, because the grids require holes to be exceptionally small and round.
“Our previous work found that there was a simple rule, based on physics experiments and calculations, that to have a stable specimen you need the diameter of the hole to be roughly 10 times the thickness of the specimen.
“Usually the thickness of the sample is 30 nanometres, so the hole needs to be 300 nanometres, which is smaller than the wavelength of visible light. That’s the scale we’re working at. For the roundness, that relates to the fact that microscopes nowadays use automation. The automation software relies on being able to see all the holes for data collection,” explains Katerina.
“We require accurate targeting on this scale to acquire images. Because the holes are so small, we can pack them very closely together, which also works for the automation and fast data collection. You have many more holes for a grid.
“Currently, what happens is people might have a two- or three-day microscope session. They put a grid into the microscope and start collecting the data but halfway through the session, they run out of areas to shoot. They might not have another grid. But using grids with many small holes, you have so many areas, this situation should never happen.”
The LMB turned to Moorfield Nanotechnology in Manchester for a custom order of equipment required to make the gold films, depositing the metal in a modified way. The company is now selling the equipment.
“We love it when this happens, because we’re after the science and the devices,” said Chris. “We know they will be useful because we’re our own customers. We know what we need. If we can bring the commercial partners in and get them to develop the product, they’ll have a new product that they would never even realise had potential.”
Quantifoil has the rights to the patent for commercial use.
“But for academic use, not commercial purposes, you are free to reproduce it or modify it,” explains Katerina, who adds that she expects others to use the technique for other grid types too.
Quantifoil and SPT Labtech expect to have the grids available for sale during 2022.
And the LMB will be among thousands of labs desperate to get their hands on these commercially-produced HexAuFoil grids, the cost of which is nominal compared to the running costs of cryo-EM.
“Around Christmas last year was the first time I made nice, usable grids with this process,” recalled Katerina. “Now I’ve had enough of making them myself!”
Making the grids
The process starts from a plain silicon wafer, patterning it with an array of small holes. This is covered with a thin metal foil before outlining each grid (a 3mm circle) and growing gold grid bars on top by electroplating, before finally releasing the individual grids from the wafer.
The method can make 600 grids at a time from a single four-inch wafer.
Crucially, it does not require the handling of the individual devices at any stage of the process.
To create the holes with spaces of just 300nm, a technique called phase-interference lithography was deployed, while to keep the holes round, the gold film was evaporated while cooling the template with liquid nitrogen in a method they named ‘cryoEvap’.
The HexAuFoil and manufacturing method have been licensed to Quantifoil Micro Tools in Jena, Germany, part of the SPT Life Sciences group based on Melbourn Science Park.
The work was funded by UKRI MRC, the Wellcome Trust, a Vice-Chancellor’s Award (Cambridge Commonwealth, European and International Trust) and a Sir John Bradfield Scholarship from Trinity College, Cambridge.
How cryo-EM works
Cryo-EM relies on flash-freezing a microscopic protein sample in vitreous ice that is just a single molecule thick. This is done at a speed fast enough to prevent the water from forming ice crystals – instead, it is glass-like.
Electron beams are then fired at this sample – held in place by a grid as a support structure – and hundreds of thousands of 2D images of the individual molecules within the sample are captured, from all angles. Computational construction enables the 3D model to be built up, revealing the protein structure at atomic resolution, or close to it.
The models even show how structures within the molecules move while performing their functions.
The data helps scientists understand human biology, explore diseases and design drugs that can interact with these structures.
Its development led to the LMB’s Richard Henderson being jointly awarded the Nobel Prize in Chemistry 2017.