NEW YORK (GenomeWeb) – A team led by researchers at the Karolinska Institute and Singapore's Agency for Science, Technology and Research have developed a method for studying protein complexes based on their thermal stability.
Detailed in a paper published this week in Science, the method, called thermal proximity coaggregation (TPCA), allows researchers to monitor protein complexes in vivo in essentially any cell or tissue type and could aid investigations into how different drugs and disease states modify these structures, said Pär Nordlund, a professor at the Karolinska Institute and senior author on the study.
The technique, the authors wrote, is based on the idea that "interacting proteins coaggregate upon heat denaturation leading to similar solubility across different temperatures." Different proteins denature to different extents at different temperatures, and by profiling the levels of a protein's denaturation across different temperature points, researchers can generate a melting curve for that protein.
The extent of a protein's denaturation can be determined by measuring its solubility, with denatured proteins becoming insoluble. Different proteins looked at in isolation will have different melting curves, and therefore different levels of solubility at different temperatures. However, interacting proteins will exhibit similar levels of solubility, leading to their melting curves shifting to become more similar. The TPCA method determines proteins that are likely interactors by looking for those shifts towards increased melting curve similarity.
The approach is based on earlier work by Nordlund and his colleagues in which they developed the underlying cellular thermal shift assay (CETSA). Much of that research was done in collaboration with Cellzome, a part of GlaxoSmithKline's R&D division, where researchers have explored the technique for studying the effect of drugs on their targets and downstream signaling.
In a 2014 Science paper, Nordlund and his co-authors used 10-plex TMT isobaric labeling mass spec experiments to establish thermal profiles for more than 7,000 human proteins. The 10-plex reagents allowed the researchers to profile the proteins of interest at 10 different temperatures, which let them gather data at enough points to put together high-quality denaturation curves.
That work focused on analyzing drug-protein interactions, which they did by looking for proteins who exhibited higher thermal stability upon addition of the drug of interest, the idea being that proteins tend to have higher thermal stability when bound to a ligand like a drug.
In the case of this week's Science paper, Nordlund and his colleagues hypothesized that a denaturing protein will pull its interacting proteins out of solution with it, allowing them to analyze the components of the larger complex.
"Truthfully, we don't quite understand how large protein complexes unfold," he said, "but what we think happens is that when you have a larger protein complex, some subunits start to unfold when you heat the cell, so the complex is still intact but some subunits have unfolded, and that makes the [entire] complex precipitate."
The researchers used the approach for a variety of analyses looking at protein complexes in K562 cells. In one, they looked at protein complexes modulated during the S-phase of the cell cycle, identifying 18 such complexes, 15 of which are known to be involved in the S-phase.
They also analyzed TCPA data across five different cell lines, finding both a core set of common complexes across the different lines as well as complexes unique to each line. In one instance, they identified BRAF and RAF1 proteins as likely interactors in the colon cancer cell line HCT116, which, they noted, fits with the "expected dimerization of BRAF and RAF1" driven by the KRAS G13D mutant known to be active in HCT116 cells. The analysis also indicated that BRAF/CRAF interaction with MEK is heightened in the HCT116 cells but dampened in the melanoma cell line A375, which they also analyzed — a finding that likewise fits with previous observations and, the authors wrote, indicates that "TPCA could potentially capture cell-specific interactions and pathways."
Nordlund suggested that while the technique has the potential to predict new complexes, it will likely prove most useful not as a standalone discovery tool, but as a method used in combination with existing protein complex databases and resources to understand how complexes might vary across cells or due to different disease states or drug treatments or other perturbations.
He said that the key advantage of the technique compared to existing approaches for studying protein complexes like yeast two-hybrid and affinity-purification mass spec experiments is that it can be applied in vivo.
For instance, he said, in yeast two-hybrid experiments, problems such a protein misfolding lead to a high false-positive rate. Experiments like AP-MS, which are done in cell lysate, suffer from the facts that "in lysate, some complexes are thought to dissociate" and "that you can get false aggregation," he said.
"The key thing [with TCPA] is that it is a direct method for looking in any cell line, in any tissue, at the status of the protein complexes and particularly for comparing them across different cellular states," he said.
Nordlund said he was most interested currently in using the technique to study questions of drug resistance in cancer, and added that he had recently moved his group at Karolinska to a clinical cancer department to get better access to clinical samples.
"There are very few biomarkers for cancer drug resistance," he said. "You have genomics, and there are a few mutations that are nearly always there, but then there are these hundreds of other mutations where we don't have a clue if they do something or not. Then you have transcriptomics where you get 20 percent of the transcriptome changing in these experiments, and you don't know what is what. Then you move to proteomics, and that starts to give you a little more sharpness, but there you start to run into problems with getting enough cells to do these experiments, and so on.
"So, there is a huge need for a novel way to look at these events in the cell," he said.