Revealing Protein Flexibility: The Griffin Lab Advances HDX Technology

By Eric Sauter

For Pat Griffin’s lab, it has been what he calls a “typical year”—including several successful collaborations and an important new study in the journal Structure. Years of steady progress have added up to notable advances in hydrogen-deuterium exchange (HDX), a technology that Griffin and his colleagues have helped to pioneer.

“We set the HDX technology bar very high with the ability to do high-precision analysis with in-depth statistical analysis,” said Griffin, chair of The Scripps Research Institute (TSRI) Department of Molecular Therapeutics. “We now have collaborations with many of the leading structural biologists because of the success of our HDX platform.”

So far, Griffin said, they have approximately 40 collaborations going on, including work with Emory University, University of California at San Francisco (UCSF), University of Alabama, Harvard, Baylor, Eli Lilly & Company and Pfizer. Lilly was a believer early on, having worked with Griffin for nearly a decade and recently adopting the technology in-house.

HDX, which is coupled with mass spectrometry, is an approach to obtaining information on changes in the flexibility of proteins upon interaction with other molecules such as hormones or other proteins. HDX is versatile and sensitive and can provide precise information on protein structure, dynamics and function.

The process involves exchanging a hydrogen atom with its heavy isotope deuterium—or the other way around. The rates of these exchange reactions vary, depending on the protein’s flexibility, but if you measure the results at various time intervals you start to paint a picture of the protein in action.

HDX, which was first discovered in the 1950s but didn’t really come into its own until the last decade or so, is being used to characterize protein-ligand and protein-protein interactions that are critical for signal transduction in cells—a prime target for drug development.

The technology can be applied with imaging technologies such as nuclear magnetic resonance (NMR), but when coupled with mass spectrometry researchers can start to analyze very large protein complexes and proteins in membranes. Typically, HDX with mass spectrometry is performed in combination with other structural technologies such as crystallography, cryo-electron microscopy and NMR.

Griffin first came across the technology while a scientist at Merck Research Laboratories.

“One of my early projects at Merck was a nuclear receptor,” Griffin said. “Our understanding was that the ligand changed the conformation of the receptor in a highly dynamic fashion, so we needed an approach to get a fingerprint of that interaction. Crystallography provided information on how the compound was bound to the receptor, but there was no information on the subtle differences between ligands that afford distinct biological response. So we set out to use HDX to map the changes.” But the technology had a few problems—it was a manual process, massively time-consuming and, as Griffin put it, not terribly robust, which limited its potential in the industry.

In 2002, Griffin left Merck to head a company called ExSAR that focused on the application of HDX technology to probe the mechanism of action of ligands on the same receptor he had worked on at Merck. However, this time ExSAR had been focusing on automating the platform because, without it, HDX would never be able to impact drug discovery.

In 2004 Griffin arrived in Jupiter, and he began efforts to replicate and advance the technology at TSRI. The acceleration of HDX software began in earnest soon after.

“Bruce Pascal [bioinformatics project manager at Scripps Florida Informatics Core] came on board and helped improve the software from day one—he’s a fantastic software engineer,” Griffin said. “People think we’re Microsoft when they ask about software, but basically it’s just one guy—Bruce.”

The Engineer

Pascal’s first upgrade was called HD Desktop—it is now HDX Workbench—that resulted in a study that made the cover of the Journal of the American Society of Mass Spectrometry.

“Pat likes to look at proteins with and without ligands,” Pascal said. “Our HDX platform tells him where binding likely occurs and where exactly the rate of exchange differs for all the other ligands he’s interested in—and we have a tool that can pull up all of those results side-by-side so you can see, for example, these 10 ligands are the same, but this one has a different rate of exchange in this region of the protein.”

As for speed, the software delivers same-day results.

“You can get the data in seven hours after loading your samples, and you can analyze it in an hour,” Pascal said. “In eight hours you can have an entire data set ready to publish.”

The ability to analyze data so rapidly is basically a complete reinvention of the HDX platform.

“The thing that we brought to the HDX space, which hadn’t been done before, was rapid statistical analysis on the samples,” Griffin said. “Instead of just doing one experiment and saying, ‘Look, these things are different,’ we could perform statistical analysis, point out the significant changes that take place and mathematically link the data so you could quantify and start building models.”

Enter the Data Generator

Anybody with a mass spectrometer can do this, Griffin said, partially because he has made the HDX software available to all academics and also plans an open source version for industry.

But while other software platforms can expedite the HDX process, Griffin and his colleagues still consider theirs to be the best—because of its comprehensive nature and integrated statistical analysis.

Devrishi Goswami (aka “Dev”) is one of the Griffin lab’s “Data Machines,” and he plays a central role in many HDX collaborations, most of which are funded by the National Institutes of Health through subcontracts with Scripps.  

“Dev is very good at looking at the data and saying what it means to our collaborators,” Griffin said. “He can write and he communicates well, so he’s very popular with them.”

Goswami earned his PhD at Germany’s Max Planck Institute of Biophysics in Frankfurt—where he was serious about protein structural biology. Griffin’s lab suited his interests perfectly.

“We’re using HDX on larger and more complicated proteins, and now we’re getting into things like multiprotein complexes,” Goswami said. For him, the best part of the process is applying the automated system to these ever-increasing levels of complexity. Even he is a bit awed by the speed, pointing out that what used to take a month can now be done in a single day.

The Complexity of Complex Proteins

Still, there are many challenges. “Some proteins are difficult to analyze,” Goswami said. “For example, membrane proteins present a problem due to their lipid environment, and highly dynamic proteins present an analytical challenge.” But these issues ultimately make it more likely that HDX will be the one technique that will be used to analyze them.

“The platform in our lab is a kind of frontrunner,” Goswami said. “With larger protein complexes, you might have 10 different proteins, but only want to analyze one of them—that’s a challenge because removing that one protein out of the complex mixture is a pain and has to be done in minutes. We’re working on a way to automate that process, to improve the way we isolate specific protein data. It all goes together; the software and the lab techniques have to march hand-in-hand.”

A Cell study, published last June in collaboration with investigators at UCSF, is a prime example of the power of the HDX platform. “In this study we were able to accurately detect the conformational changes of a 30,000 Dalton molecular weight protein that was present in a protein complex mixture with a mass of 195,000 Daltons of unique sequence. We are now in the process of analyzing the same protein in a larger complex of around 350,000 Daltons”, said Goswami.

The power of the HDX technology and robustness of the automated platform in Griffin lab is evident from the number of collaborations they undertake. “I am part of at least 10 active collaborations with different academic and industry labs,” said Goswami.

In past three years, Goswami has authored/coauthored 11 papers in journals including Cell (2014 Jun 19;157(7):1685-97), Structure (2014 Jul 8;22(7):961-73; 2013 Nov 5;21(11):1942-53), Angewandte Chemie (2014 Jan 3;53(1):132-5), PNAS (2013 Jul 9;110(28):11475-80), Diabetes (2014 Apr;63(4):1394-409), JBC (2014 May 23;289(21):14941-54; 2013 Oct 18;288(42):30285-99; 2012 Dec 28;287(53):44546-60) and JASMS (2013 Oct;24(10):1584-92).

“These are all very important and significant contributions to each of the fields that our collaborators are focused on,” said Griffin. “These manuscripts not only illustrate the high productivity of Dev, but they speak to the efficiency of our platform.”

Send comments to: press[at]scripps.edu