NEW YORK – Roswell Biotechnologies presented the first peer-reviewed data on its molecular electronics technology last week in a publication in the Proceedings of the National Academy of Sciences.
In collaboration with George Church, of Harvard Medical School and the Wyss Institute, and Jim Tour of Rice University, as well as researchers from the University of California, San Diego, Roswell presented data from its ME 1947 molecular electronics chip. The paper largely recapitulated the company's presentation at its product unveiling in November, suggesting that the technology could be used to analyze binding dynamics of DNA polymerases, small molecules, antibodies, aptamers, oligonucleotides, CRISPR-Cas enzymes, and more.
Both Tour and Church are scientific advisers to the company. Tour is a self-described minor shareholder, and Church said in an email that he has received stock options and consulting fees. However, the paper did not disclose these relationships in the competing interest statement, which only stated that "[a]ll authors having the Roswell affiliation … are employed by Roswell Biotechnologies."
The paper adds in-depth data on the fundamental workings of the sensor, Roswell cofounder and CSO Barry Merriman said in an email. "We show the majority of the signal pulse is resulting from a field effect, where charges moving on the probe-target complex exert an electric field that is gating the current flow through the molecular wire." It also considers the detection limits of the technology and suggests ways in which the company could further improve it, he added.
"It's a nice proof of concept for being able to make measurements of multiple types of molecules using an integrated complementary metal-oxide semiconductor (CMOS) device," said David Walt, a professor at the Wyss Institute and an inventor and entrepreneur in biological detection. He is a cofounder of companies including Illumina, Quanterix, and Vizgen. But Roswell will have to improve its detection limit, he suggested, to get to where the state of the art is headed.
"There's a difference between measuring many single molecules and measuring single molecules at concentrations that are relevant for early diagnosis of cancer or neurodegenerative diseases," he said. "They’re not measuring in concentration ranges that people would particularly care about."
For example, the paper presents data from DNA hybridization experiments showing detection at 20 nanomolar concentrations. "That's a millionfold lower than what PCR and other techniques can measure," he said.
Frances Ligler, a biosensor researcher and professor of biomedical engineering at Texas A&M University, said in an email that the company's results "look quite impressive," adding that how the technology handles complex samples will be important.
Initially a DNA sequencing startup, San Diego-based Roswell has expanded its field of view to a variety of applications that could be informed by biomolecule binding dynamics, including infectious disease diagnostics, drug discovery assays, and methylation sequencing.
At present, its chips feature about 16,000 features, each of which consist of a circuit that can be connected by a peptide bridge attached to a molecule of interest. Each circuit registers electric current levels, and secondary patterns of binding and unbinding can be used to fingerprint molecules.
"The chip digitizes the resulting picoamp-scale current-versus-time readout from each sensor element of the array at a rate of 1,000 frames per second," the authors wrote. "This provides detailed electrical signatures of the single-molecule interactions between the probe and targets present in a solution-phase test sample. This platform is used to measure the interaction kinetics of single molecules, without the use of labels, in a massively parallel fashion."
Both the authors and Walt placed the Roswell chip into a long history of research. The authors said the chip realizes a 50-year-old idea to merge biological molecules with CMOS chips and an opportunity to "[move] away from photon-based detection" of biomolecules. Thermo Fisher Scientific's IonTorrent sequencer, acquired through Life Technologies in 2013, also uses CMOS chips and did away with light-based detection. Merriman and Paul Mola, Roswell's CEO and cofounder, both worked at Life Technologies until 2014.
"What's impressive, in terms of engineering design, is to make this something that could conceivably be scaled," Walt said.
The authors suggested that the technology could follow projected trends in CMOS fabrication and eventually reach a density of 100 million sensors per square millimeter. Roswell has projected that its next generation of chips, expected by the end of 2023, could have up to 1 million features, and the generation after that as many as 20 million.
More features on a chip would help with limits of detection, Walt said. "But more sensors could also lead to more nonspecific binding. [I] will be interested to follow the progress to see how things scale."
Merriman said that Roswell has unpublished experimental data detecting molecules "down to sub-picomolar concentration," or more than 20,000 times below the published results. Moreover, the firm has the option to apply electrical forces to draw target molecules closer to the sensors. This alone could provide a millionfold boost to the limit of detection, he suggested, and longer observation times or observations across multiple sensors could also increase sensitivity.
Even when presented with target molecules at higher concentrations, as shown in the paper, the chip has utility, Merriman said. "We can measure standard quantities 'better': the concentration, down to low limits; the standard binding kinetics of the target molecule; on rate, off rate, dwell time, binding affinity (Kd)," he said. In addition, it can help make observations not previously possible, for example distinguish between misformed and well-formed antibodies, he suggested. Ligler and Walt suggested that Roswell still had work to do when it comes to complex samples that it would encounter in diagnostics applications.
"I always look at how well a detection method works in complex samples," Ligler said. "The amount of data on this topic is limited to spiking a few samples into contrived saliva — whatever that is. But the authors do realize that nonspecific finding is an issue and understand that they need to address it."
Merriman responded that for DNA binding, the data in the paper show "a similar dose-response of this sensor to target concentration," whether the sample is oligonucleotides in solution, part of a PCR product, or spiked with salmon sperm or human saliva. "This shows the sensor retains specificity against all these forms of complex background that are relevant to assays. … We will be trying other bodily fluids such as blood as well as develop relevant assays, but based on these results, we consider this low risk," he said.
Potential applications outlined by the authors include "next-generation" microarrays that are faster and could require less sample preparation, label-free protein identification, CRISPR-based diagnostics, and Cas enzyme evolution.
"Clearly, these results are exciting and the progress of this technology bears watching," Ligler said.