NEW YORK – Recently developed methods for automating isobaric labeling workflows aim to streamline multiplexing of proteomics experiments.
The approaches, developed separately by life sciences firm Calico and proteomics sample prep company PreOmics, could help drive uptake of isobaric labeling and Thermo Fisher Scientific's TMTpro label reagents in particular, which allow multiplexing of up to 16 samples.
As data-independent acquisition (DIA) mass spec workflows have advanced technically and grown in popularity, some researchers have suggested that isobaric tagging approaches like TMT labeling may see reduced uptake.
However, recent advances in mass spec instrumentation and new TMT reagents and methods have boosted the technique's utility. The ability to automate the workflow could further enhance its appeal.
Isobaric labeling (of which TMT is the most popular commercial version) uses stable isotope tags attached to peptides of interest to enable relative or absolute quantitation of proteins via tandem mass spectrometry. Digested peptides are labeled with tags that fragment during MS2 to produce signals corresponding to the amount of peptide present in a sample. The approach is commonly used to multiplex samples, which improves throughput and reduces variation.
Fiona McAllister, principal investigator, proteomics at Alphabet-subsidiary Calico, said that her lab considered using DIA for its large-scale studies but determined that "for very large cohort studies in the thousands [of samples], the time savings with TMTpro is unbeatable."
She noted that run times for a single-shot TMT experiment is longer than for a DIA experiment, but the ability to multiplex up to 16 samples in a run substantially outweighs that factor. She and her colleagues calculated that analyzing 1,000 samples would take around three months of non-stop mass spec run time using DIA methods compared to around 11 days using TMTpro.
McAllister added that the RTS-MS3 real-time search method developed by Harvard University professor Steven Gygi (in whose lab McAllister was a post-doc) and using Thermo Fisher's FAIMS ion mobility device had allowed her team to "dig much deeper" into samples without needing fractionation. She added that she considered another advantage of TMT the fact that, if fractionation is needed, "TMT samples can be easily fractionated without a huge time penalty compared to DIA."
Sample prep, however, remained a major challenge, McAllister and her colleagues noted in a January Journal of Proteome Research study detailing their development of an automated workflow for plasma proteomics using the TMTpro reagents (which Thermo Fisher licenses from UK-based proteomic firm Proteome Sciences.)
"The standard TMT sample preparation protocol is complicated, with more steps than label-free workflows," they wrote, noting that "the main barrier to performing large-scale studies using TMT is sample preparation since the labeling chemistry is very time-consuming to perform manually."
They added that, "while there have been multiple automated sample preparation platforms for label-free proteomics reported," only one such platform existed for isobaric labeling: PreOmics' PreOn system.
In fact, McAllister said that PreOmics released the PreOn system while her team was in the middle of developing its automated TMT method, meaning there were no automation options when they began their work. To address the problem, they devised a system they called the AutoMP3 platform, which uses a Hamilton Vantage liquid handler to process samples in a 96-well plate format. Using the AutoMP3 system to prep TMTpro experiments, the Calico researchers found they could prepare 96 samples in two days, five-fold faster than the 10 days they said manual preparation would have taken.
One of the biggest challenges to automating the TMT workflow, McAllister said, was the large volume of reagents involved in running a 16-plex experiment and the fact that such volumes are not compatible with 96-well plate formats.
"This was a huge barrier," she said, noting that, ultimately, Aleksandr Gaun, a senior research associate at South San Francisco-based Calico and first author on the JPR paper, devised a method in which samples were split between multiple wells to allow lower volumes per well.
"We were initially concerned that extra variability and loss could be an issue, but after extensive optimization, this performs extremely well in our hands," McAllister said.
Another challenge to automating TMT workflows is the fact that these labeling reagents "are very volatile," said Russell Golson, chief commercial officer at Martinsried, Germany-based PreOmics, which launched its PreOn proteomics sample prep platform in May 2019. Last year, the company updated the system to work with the TMTpro reagents, which upped TMT multiplexing from an 11-plex to 16-plex.
The volatility of the reagents means that the labels can't be added at the beginning of the sample prep process along with the rest of the reagents, Golson said. "So we have basically a pause step so that when the robot gets to the 'add TMT stage,' it pauses and alerts you and then you add the TMT."
The Calico researchers likewise highlighted this step as difficult to automate, and Harvard's Gygi, who was not involved in either project said that the volatility of the TMT reagents meant that "you are likely limited to semi-automated workflows where you stop the work and add the TMT reagents at that stage — usually manually."
Golson said the effort to automate the process emerged from discussions with pharmaceutical and other industry customers whose scientists were asking for increased access to mass spectrometry.
"Basically what people were saying to us more and more was that they were being pushed by biologists who wanted more access to mass spectrometry, but their cores couldn't prepare all of these samples," he said. "They wanted to be able to have a small benchtop [instrument] that they or a technician could put their samples on and run through a dedicated [sample prep] menu and press go, and then the mass spec specialist could come pick these up off the [sample prep] robot and know they were in great condition to inject into the mass spectrometer."
Golson said that a major focus of PreOmics' automation effort was paring down the number of steps involved by improving the compatibility of the different stages to eliminate as much clean-up as possible.
He noted that many proteomics workflows use "homebrew" methods developed by individual labs.
"One of the main problems with homebrew is the lack of chemical compatibility between one step and another, and that means a lot of transfer steps and clean-up steps between phases," he said. "That brings a longer cycle time, and it also means that if you have small, precious samples, you have four or five chances to lose that precious sample along the way."
He said that by reducing these gaps in compatibility, PreOmics had moved from an eight-step protocol to a three-step protocol. The PreOn system can prepare a 16-plex TMT sample in about two hours.
Golson said the company's customers for the system are largely biopharma and pharma firms doing large-scale mass spec experiments. He said that the company has also received interest from academic core facilities.
Brett Phinney, director of the proteomics core facility at the University of California, Davis, said in an email that his lab "would love to try the PreOmics kit," though he added that the cost made him reluctant.
Golson said that the PreOn platform costs around $95,000. Researchers have to use the company's reagent kits on the system, which cost around $20 per sample, not including the cost of the TMT reagents themselves.
TMT labeling has also become a key technology for single-cell proteomics, where it enables experiments using a carrier sample to boost the signal of peptides present in extremely small target samples. Neither the AutoMP3 nor the PreOn approaches are aimed at the single-cell proteomics space, but in January researchers at Brigham Young University published a study in Analytical Chemistry detailing an automated version of their previously developed NanoPOTS (nanodroplet processing in one pot for trace samples) sample prep technique.
The NanoPOTS (nanodroplet processing in one pot for trace samples) platform shrinks sample processing volumes down to less than 200 nanoliters to limit sample loss and speed trypsin digestion, which has slow kinetics at small sample volumes. The automated version, called autoPOTS, uses a robotic pipetting platform and autosampler to automate the NanoPOTS process.
Gygi said that automation becomes more important as researchers begin to move to experiments running hundreds of samples or more. He suggested that labs have not previously implemented automation for TMT sample prep in large part because they were not running such large-scale experiments.
"It seems that for most of the work we do, we are dealing with less than 16 samples — certainly less than 100," he said. "We can usually pack all of the replicates, time-series, dose-response, et cetera types of experiments into a single 16-plex. You don’t need a plate-based method for this. This even holds true for up to around 100 samples. The mass spectrometer is the bottleneck and not the prep. However, this changes as you move beyond 100 samples."
"Certainly, for large-scale studies, the automation strategies could be really important," he added.
That suggests automation could become more relevant as proteomics continues to move in the direction of larger experimental cohorts. That is particularly true of the single-cell space given the large number of cells required to collect good data on questions of interest.
Gygi also said that the development of paramagnetic bead-based single-pot solid-phase-enhanced sample preparation (SP3) methods had also allowed for improved automation of steps like protein precipitation.
The SP3 approach was first published on by researchers at the European Molecular Biology Laboratory in 2014. Last year, PreOmics entered into a licensing agreement with EMBL for the technology.
"We didn't have a good way to automate until fairly recently," he said.