KINGSTON, R.I. – March 10, 2021 – Researchers from the University of Rhode Island and Rensselaer Polytechnic Institute have teamed up to perfect a fast, accurate test combining the use of a solid-state nanopore and machine learning to verify the identity and purity of synthetic heparan sulfate The work has implications for better characterizing viruses such as COVID-19 and ensuring the purity of the clinical anticoagulant heparin.
Discovered more than a century ago, naturally sourced sulfated polysaccharide heparin has grown into a major clinical anticoagulant with more than 500 million doses administered annually to treat clotting disorders and prevent clotting in patients undergoing kidney dialysis and surgery. Sourced from pig intestines, heparin supplies have experienced batch inconsistencies and contamination, causing sickness and death.
The paper detailing the work, “Synthetic heparan sulfate standards and machine learning facilitate the development of solid-state nanopore analysis,” was published this week by the journal Proceedings of the National Academy of Sciences.
The collaboration brought together the labs of RPI Professor Robert Linhardt, an internationally recognized leader in glycoscience and synthetic heparin research, and URI Professor Jason Dwyer, recipient of an Innovation Award from the Federation of Analytical Chemistry and Spectroscopy Societies for his advances in nanopore sensing and whose earlier research developed a simpler, faster way to analyze naturally sourced heparin for impurities. This collaboration focused on a class of molecules known as heparan sulfates.
“Heparan sulfates are a type of long-chain linked sugar molecules called polysaccharides that are as important to our biology as DNA,” said Linhardt and Ke Xia, a research scientist at RPI and lead author of the paper, in an email. “This paper demonstrates it is possible to reduce the time required for sequencing heparan sulfate chains from years to minutes. This accomplishment will improve our understanding of polysaccharides and the glycome as much as the use of high-speed sequencers has improved the understanding of DNA and the genome.”
At RPI, Linhardt, in collaboration with the University of North Carolina, has developed a synthetic heparin offering greater production control than naturally sourced heparin.
“RPI is extraordinarily good at making and understanding this type of molecule, and we wanted to work with them for that reason,” said Dwyer. “What we’re able to do is look at their samples with high throughput one molecule at a time, which is a tremendous advance over conventional studies that get a sort of average picture by looking at many, many molecules all at once. And when you look at something, a single molecule at a time, you can learn new and important things.”
Nanopore sensors – a nanofluidic hole, which is about one hundred thousandth times smaller than a human hair, located in an impermeable, silicon-based membrane – have been a promising tool in single-molecule research for more than 20 years. Using an ionic current, the sensor analyzes molecules as they pass through the pore. Dwyer’s lab specializes in using special surface coatings to tailor the nanopore to work with the particularly challenging polysaccharides.
While nanopores have been used in DNA sequencing for more than 15 years, analyzing molecules like polysaccharides (or glycans) is far more complex.
“The human genome project sequenced DNA and that was a billion-dollar project. You had to figure out at each position which of four bases you had. It’s basically a lottery with four numbers at each position of the DNA,” Dwyer said. “With polysaccharide sequencing, you have 120 bases to choose from – and you have to get them right at every single position. So, the lottery odds are a lot worse. Basically, there’s much more variety in these types of molecules – both chemical and structural.”
“We feel nanopores are well-suited to analyze polysaccharides, which can be even a more useful bio marker than DNA,” added James Hagan, a third-year doctoral student who oversaw the experiment in Dwyer’s lab. “These results build on earlier work in our laboratory that revealed key operating principles and showed us that nanopores had tremendous promise to characterize a wide range of unique polysaccharides.”
Rensselaer provided Dwyer’s lab with four high-purity synthetic heparan sulfate samples, each with four different types of sugar units that could make different sequences. The combination of the solid-state nanopore, high-purity samples and machine learning allowed for a clear classification of the four molecules using only a minimal amount of sample.
“Fast nanopore analysis – 10 minutes for each analysis – gave clear differences between the signal generated by our four heparan sulfates providing us sequence information,” said Linhardt and Xia. “Using machine learning we only required the results of 500 to 2,000 chains of each heparan sulfate to obtain a nearly 97% accurate prediction of the sequences. We have trained the machine to quickly read very long sequences comprised of four letters (sugar units). We are working to extend this technology to read more complex sequences made up of more sugar units in real time.”
“This is really further proof of the importance of nanopore sensors for polysaccharide sensing,” Hagan said. “Together with the machine learning, the nanopore measurements were able to get down to the specific sequence of each of the molecules. With this combined approach we were able to differentiate each of the monosaccharides in the polysaccharide chain, which is really the most exciting outcome of this collaboration.”
The potential of this work goes beyond the quality assurance of synthetic heparin. The polysaccharides used in the paper, glycosaminoglycans, are present in all biological lifeforms and serve different and vital physiological roles.
“To make our study more relevant today, COVID-19 has its surface decorated with glycans,” Hagan said. “By building on this work, you could sequence those markers and pry yet more secrets from this virus, even to the point of better therapeutic interventions.”
This material is based upon work supported by the National Science Foundation under Grant No. CHE-1808344 to Prof. Dwyer, and the National Institute of Health under Grants No. CA231074, DK111958, and HL125371 to Prof. Linhardt. Any opinions, findings, and conclusions or recommendations expressed in this material are those of the author(s) and do not necessarily reflect the views of the NSF or NIH.