NEW YORK – While technical hurdles remain, nanopore-based protein analysis is approaching feasibility as scientists have recently demonstrated potential methods for tackling the challenge.
Both academic and industry-affiliated research groups have published on techniques for moving full-length proteins through nanopores in a controlled manner and making identifications based on the resulting signal, and at least one company, Portal Biotech, has begun to place nanopore-based protein analyzers in the hands of early adopters.
Nanopore biosensing typically uses the changes in electric current produced as a molecule translocates through the pore to determine its identity. To date, this approach has mainly been used for sequencing nucleic acids — most prominently by Oxford, UK-based Oxford Nanopore Technologies — but researchers have also been pursuing it for protein analysis.
Compared to nucleic acid sequencing, nanopore-based protein analysis is significantly more difficult. Researchers must contend with 20 amino acids when analyzing proteins versus just four bases in DNA, as well as a multitude of post-translational modifications. They also face the challenge of having to unfold proteins prior to passing them through the pores with enough control and consistency to generate reproducible signals for protein identification.
The upside of nanopore-based protein measurements could be substantial, as it could enable detailed analyses of full-length proteins with single-molecule sensitivity, something not currently possible using established proteomic technologies like mass spectrometry or immunoassays.
Perhaps most immediate on the horizon is nanopore-based protein fingerprinting, in which nanopores are used not to read an amino acid sequence but to generate a "fingerprint" signal that is characteristic of a protein. Researchers can use this fingerprint to identify the protein as well as use deviations from the expected signal to identify potential variations in its sequence or posttranslational modifications.
True de novo sequencing, in which researchers read each amino acid in a protein as it passes through a nanopore, is likely a longer way off, with some in the field still questioning whether this will ultimately be possible. Some also debate whether it will prove necessary or if fingerprinting approaches will be sufficient.
The translocation challenge
For either application, an effective strategy for moving target proteins through a pore is key. Broadly speaking, such strategies fall into two categories — enzymatic, in which enzymes like DNA helicases or the protein unfoldase ClpX are used to translocate proteins, and non-enzymatic, in which physical forces such as electrophoresis or electro-osmotic flow are used to translocate target proteins.
In a study published last year in Nature Nanotechnology, a team led by Hagan Bayley, professor of chemical biology at the University of Oxford and one of the founders of Oxford Nanopore, described the use of electro-osmosis for nanopore protein analysis. The process uses nanopores selective for particular ions in a liquid. When a potential is applied to the nanopore, it pulls those ions toward it, which carry water molecules with them. This flow of liquid can be used to unfold proteins and move them through the nanopore.
Bayley said that he and his colleagues view electro-osmosis as "the most promising method” for protein translocation, noting that the enzymatic tools used for nanopore-based DNA sequencing are not necessarily well suited to protein analysis. A non-enzymatic approach could also be simpler than an enzyme-based one, as it would require less manipulation of the target proteins.
Aleksei Aksimentiev, a physicist at the University of Illinois at Urbana-Champaign whose work includes research into nanopore-based biosensing, highlighted demonstrations by Bayley's lab and others of the use of electro-osmotic force as an important recent advance for the field. Prior to this work, he said, it was unclear whether this force would be sufficient to move proteins through nanopores. "It looks like it is, and that is a really great development," he said.
Aksimentiev noted that there are difficulties to the electro-osmotic approach, particularly around controlling the consistency of protein movement through the pore, which may vary across the length of the pore and depend on the charge of the protein sequence passing through the pore at a given time. Enzymatic approaches could provide more consistent translocation, he said, though electro-osmotic force alone could prove sufficient for fingerprinting applications.
On the enzymatic side, Oxford Nanopore and a team including Aksimentiev and Delft University of Technology researcher Cees Dekker have separately explored strategies that link DNA to target proteins and then use DNA helicases to pull the protein-DNA complex through a nanopore. One potential limitation to this approach, however, is read length. Xiuqi Chen, a postdoctoral researcher in Dekker's lab, said that he and his colleagues are currently able to look at peptides in the range of 20 to 30 amino acids using the technique.
Additionally, several researchers are pursuing methods that combine enzymatic and non-enzymatic approaches. This month, a team of researchers led by Jeff Nivala, a molecular engineering professor at the University of Washington, published a study in Nature in which they used electrophoresis to pull proteins through a nanopore and then ClpX to pull it back out of the pore in a slower, more reproducible manner, generating better signals for identifying the protein and detecting amino acid variants and PTMs.
Nivala said that while researchers are still exploring various methods for translocating proteins through nanopores, he believes some sort of enzymatic motor will be key. "I just don't think [you have] the reproducibility, the control of how the strand moves if you don't have a motor," he said. "It's just too variable, in my opinion."
Nanopore protein analysis firm Portal Biotech is also taking a combined approach, pairing electro-osmotic force with a nanopore engineered to include a proteosome to unfold and translocate target proteins. In 2021, Giovanni Maglia, the company's cofounder and CSO as well as a professor of chemical biology at Groningen University, published a paper in Nature Chemistry on a nanopore-proteosome complex his lab had engineered and used for protein measurements. In 2023, he and his colleagues published a study in Nature Biotechnology on the use of electro-osmotic flow for unfolding and translocating proteins.
The combination of these forces allows the company to pass proteins through its nanopores with "almost metronomic precision," said Andrew Heron, Portal Biotech cofounder and CEO. Heron was previously senior director of advanced research at Oxford Nanopore.
Near-term potential
With these advances in controlling unfolding and translocation, the field has drawn closer to being able to make meaningful protein measurements. Currently, these are fingerprinting measurements and are typically made in a targeted fashion on small sets of proteins or peptides, often either purified or synthetically generated.
Nanopore-based protein fingerprinting can potentially provide information that is difficult to generate using approaches like mass spec, said Chen. He cited as an example more in-depth analysis of known proteins to allow for the identification and localization of multiple PTMs or other alterations.
With mass spec, "when you have multiple sites of modification, it can be a problem to identify which amino acids are being modified, but with [nanopores] we know the peptide is going through the pore in a certain way, and the signal is different at certain steps, so we can pretty reliably say which amino acids are modified."
In a recent study published in the Journal of the American Chemical Society, Bayley and colleagues used chemical binders (phos-tags) specific to protein phosphorylation to further distinguish the signal produced by these modifications, allowing them to detect phosphorylation along a polypeptide chain of more than 700 amino acids.
Bayley noted, however, that "technical improvements" are still needed. "If a 1,000 amino acid polypeptide goes through the pore and is phosphorylated at, say, seven positions, do we see all seven?" he said. "Right now, the answer is 'not always.' And so that has to be improved."
Nivala said that based on recently and soon-to-be published studies, nanopore-based protein fingerprinting "is probably here" for applications in which researchers are looking at a well-defined set of protein targets.
He offered as an example an experiment in which, "I'm looking at this particular peptide or this particular protein, and I want to know, does it have this particular modification at this site?"
Nivala said he expects this will be expanded to broader fingerprinting efforts in the near term, in which experimentally generated nanopore data can be compared to an empirically built database of nanopore data for specific proteins. The next step, he said, could be to generate databases of simulated nanopore data, in which the signal produced by a given protein could be predicted by its sequence and identifications could be made by matching experimentally generated nanopore signal to the predicted signals, much as is done in discovery mass spec.
Nivala suggested that, initially, methods for proteome-scale fingerprinting will likely be built on such an approach. He noted that in their recent Nature paper, he and his colleagues demonstrated that they could predict nanopore signals based on an amino acid sequence that matched a protein's actual experimental signal.
Portal Biotech is working along similar lines. Heron said the company "routinely measures proteins that are more than 100 kilodaltons in size." He noted that one advantage of measuring such large proteins, as opposed to shorter peptides, is that their signature is highly unique, making for easier identification.
"In the mass spec space, even though it is very accurate, if you have a peptide, it is still quite ambiguous what molecule it came from. But if you feed a 50- or 100-kilodalton protein [through a nanopore], it becomes totally unambiguous what the molecule is because you've got so much data," Heron said.
Portal Biotech, which was founded in 2021, has been using its experimental nanopore data to train machine-learning algorithms to predict nanopore signals from protein sequences with the goal of producing databases of simulated signals that can be used to make identifications. Heron said the company is now at the point where it is capable of producing such simulated databases and identifying individual proteins out of thousands of potential candidates with accuracies above 90 percent, though it has not yet published peer-reviewed research detailing this work.
The road to de novo sequencing
True de novo sequencing of proteins is a harder problem. Identifying each of 20 amino acids as they pass through a nanopore along with PTMs of interest is a daunting challenge. Adding to the challenge is the fact that as a protein translocates, multiple amino acids are interacting with the nanopore at any moment in time, creating highly complex signals that are difficult to deconvolute.
"You have 20 amino acids, and then we suspect that around eight amino acids typically contribute to a given signal," Chen said. "You're in the billions of parameters … so the complexity could be very formidable." He added that it was unclear to him if a supervised machine-learning algorithm could ever deal with that level of complexity.
Chen suggested that one possible way to address this issue could be to reduce the complexity of potential amino acid combinations by, for instance, using features like the different charge states of different amino acids to narrow down the possible amino acids present, thus reducing the size of the combinatorial space.
Nivala said that while he believes de novo sequencing will ultimately prove possible, innovation on the chemistry side, and in nanopore design in particular, will be necessary. He suggested that shallower pores that contain fewer amino acids at a time could be key to simplifying the signal enough to enable sequencing-based approaches.
Aksimentiev likewise said that nanopores designed to confine a smaller number of amino acids — perhaps around two or three — would be key to enabling de novo sequencing. Another possible option, he said, would be to sever amino acids one at a time from target proteins and feed them singly through the nanopore. "The problem there is that you have to capture all of them and capture them in sequence," he said. "That is a technical challenge."
Heron expressed optimism that the field would ultimately achieve de novo sequencing, saying that improvements in nanopore technology will be needed but that "the majority of the work will be on training and building better and better machine learning."
He cited the example of nanopore-based DNA sequencing, which he noted has seen dramatic advances in capabilities since the field's initial efforts. "We think we'll walk a similar pathway with proteins," he said. "It's very hard to say how long that pathway is because it just requires time, people, effort, and money."
Commercial prospects
Nanopore-based protein analysis has to date remained largely an academic enterprise, with commercial entrants having thus far failed to bring products to market.
French firm DreamPore launched in 2017 with the goal of developing a nanopore protein analysis system, though the company now appears to be inactive and did not return requests for comment.
In 2018, Dekker and his Delft colleague Chirlmin Joo launched a company called Bluemics to commercialize nanopore-based protein detection technology developed in their labs, but they shut down the company after struggling to find investors.
Portal Biotech appears to be off to a more promising start. According to Heron, the company, which raised $11 million in venture funding in 2022, has begun placing instruments with early adopters ranging from academic labs to small and large biopharma companies, though he declined to name any specific firms. The company is also analyzing samples sent by outside customers, Heron said. "It's still at the stage where we continue to refine the software and capabilities, but the most important thing is for us to put it in the hands of our customers, for them to tell us how we can improve it further for their particular applications."
Oxford Nanopore also continues to work on applying its technologies to protein analysis. In an email, Lakmal Jayasinghe, senior VP of R&D biologics at the company, said it has "made considerable progress toward this goal," though he added that it is still working "to establish methods to prepare full-length proteins that can be threaded through the nanopore in a uniform fashion" and that "amino [acid] calling is particularly challenging." He added that while the company's ultimate aim is de novo sequencing, it also intends "to develop application-specific products and methods to enable protein fingerprinting."
Some of Nivala's work, including that presented in the recent Nature study, is partially funded by Oxford Nanopore. In the study, he and his colleagues used the company's MinIon platform for their measurements.
As with other entrants in the single-molecule protein analysis space, like Quantum-Si and Nautilus Biotechnology, nanopores will likely first be used for narrowly targeted applications.
Initially, "we will be going into the targeted proteomic space," Heron said. "A protein or a limited library of proteins [of interest] is very accessible for us." He said the company plans at first to focus on the drug development space, facilitating, for instance, in-depth characterization of drug biomarkers or targets or, in the case of biologics, the drugs themselves.
Longer term, though, Heron said he expects the technology will be useful for "deep discovery" proteomics. His cofounder Maglia echoed this sentiment. "There are complexities of course, but I think there is no intrinsic limitation," he said. "With time, you can tackle these problems."