NEW YORK – A team led by researchers at Northwestern University has developed a charge detection mass spectrometry-based method for top-down single-cell proteomics.
Detailed in a BioRxiv preprint published this month, the approach could enable scientists to detect intact proteins in single cells with a throughput of 1,000 or more cells per day.
While development of the technique is still in the early stages, Neil Kelleher, director of the Chemistry of Life Processes Institute at Northwestern University and senior author on the preprint, said he and his colleagues hope to within several years achieve parity with bulk top-down proteomics analyses.
The method employs the proteoform imaging mass spectrometry (PiMS) approach that Kelleher and coauthors described in a paper published last year in Science Advances. In that paper, they used a charge detection mass spec-based technique they termed individual ion MS (I2MS), which allows researchers to analyze ions individually, reducing the complexity of the signals generated.
The researchers took advantage of this technique to perform top-down mass spec imaging. Traditionally, such imaging experiments have been limited to relatively small proteoforms, but the ability of I2MS to resolve the complicated mix of ions produced in mass spec imaging allowed Kelleher and his colleagues to significantly extend the proteoform mass range they were able to address. Combining I2MS with ambient nanospray desorption electrospray ionization (nano-DESI), they imaged human kidney tissue, identifying 169 proteoforms with masses up to 70 kDa.
In their recent preprint, Kelleher and his colleagues applied the technique to high-throughput analysis of single cells. The researchers took cells from rat hippocampus and drop cast them on a glass slide. Then, as they wrote, they rastered "a liquid droplet across the glass slide" while continuously sampling the droplet using a nano-DESI source. When the droplet hit one of the cells on the slide, the source ionized the contents of the cell, producing proteoform ions that were introduced into the mass spectrometer and analyzed via I2MS.
Using this approach, the researchers were able to identify 169 proteoforms in the rat brain cells sampled. They also showed they could classify a portion of the analyzed cells as either neurons, astrocytes, or microglia based on their proteoform content. Additionally, because the approach does not use liquid chromatography to separate samples prior to introduction into the mass spec, it has very high throughput. The workflow used in the preprint was able to analyze around 80 cells per hour. The researchers analyzed a total of 10,836 single cells over the course of 10 days.
The research presented in the preprint is largely a proof of principle, with considerable development work still needed, but Kelleher suggested it points a way toward the collection of top-down proteomics data in single cells at very high-throughput.
Top-down proteomics is appealing in that by analyzing intact proteins, as opposed to proteins digested in peptides as is done in bottom-up proteomics, researchers are better able to observe features like splice variants, truncations, and post-translational modifications that play key roles in biological processes.
Throughput is a significant concern in proteomics generally, but it is especially so for single-cell proteomics where data from large numbers of cells is needed for robust analyses and where each cell constitutes its own mass spec experiment. The ability to analyze 1,000 or more cells per day would be a major boon in this regard.
"It is very bold and exciting," Nikolai Slavov, associate professor of bioengineering at Northeastern University and a pioneer of single-cell proteomics, said of the Kelleher lab's top-down single-cell work. "The number of single cells analyzed per unit time is remarkably high."
He also noted that the ability to collect proteoform-level information is significant.
"Clearly, we all want to know the proteoforms, and when we do a bottom-up analysis our knowledge of proteoforms is indirect," he said. "The intact protein analysis gives us a direct measurement of what we would really like to know."
Slavov, who was not involved in the preprint work, noted, however, that the technique is not yet capable of generating much quantitative information, a shortcoming that Kelleher acknowledged.
"Quantitation is hard," Kelleher said, observing that one of the main difficulties in this regard is introducing enough ions into the mass spectrometer to do good quantitation.
When only a few ions for a given proteoform make it into the mass spectrometer, background noise can make it difficult to accurately quantify them, Slavov said.
Addressing this "mostly has to do with improving the efficiency of ionization and the delivery of proteoforms into the [mass spec]," he said. "I think that is a very important future direction for this to go into."
"We know we're not getting a good [ion] extraction. It's a low percentage of proteoforms that are coming out of the cell. We get about 10,000 ions [per cell on average]," Kelleher said, adding that his team's initial focus has been demonstrating the technique's throughput.
He said he believes that significantly improved ion extraction and, consequently, quantitation, is possible, citing the example of single-cell RNA sequencing.
"The early days of single-cell early RNA-seq were 2012, 2014, and then by 2016, you could get 2,000 to 3,000 transcripts in a single cell," he said. "And we're going to go through the same maturation curve. My hope is to get to the kind of depth that we can in bulk [top-down analysis] in three to four years' time."
Recent bulk top-down studies have identified in the range of 2,000 to 3,000 proteins and 20,000 to 30,000 proteoforms with around 1,000 to 2,000 of those proteoforms being reproducibly quantifiable.
Bottom-up single-cell proteomics workflows continue to advance, as well. At this month's American Society for Mass Spectrometry annual conference in Houston, Bruker highlighted the capacity of its new timsTOF Ultra instrument for single-cell proteomics, noting in press materials that the instrument can identify roughly 5,000 proteins at the single-cell level and quantify more than 4,800 proteins at coefficients of variation of under 20 percent.
Additionally, Karl Mechtler, proteomics head at Vienna's Research Institute for Molecular Pathology, presented data at Bruker's user meeting this week that his lab had generated using the Ultra instrument. Using the system to look at single HeLa and K562 cells, his lab identified, respectively, 3,803 and 3,221 proteins.
On the throughput front, Slavov and his colleagues have continued development of their plexDIA method, which combines non-isobaric mass tagging with data-independent acquisition mass spec to enable multiplexed single-cell mass spec experiments run on short LC gradients. To date, the researchers have published a version of the method that combines three-plex labeling with five-minute LC gradients, but Slavov said he believes the multiplexing level can be pushed to as much as 20-plex while the LC run time can also be significantly reduced, perhaps down to the one-minute range.