NEW YORK (GenomeWeb) – A team led by researchers at the Austrian Academy of Sciences has used a thermal-stability profiling-based approach for proteome-wide drug screening.
The approach, which was detailed in a paper published last month in Nature Methods, offers a potential alternative to traditional chemical proteomic methods, Kilian Huber, a researcher at the Structural Genomics Consortium and first author on the study, told GenomeWeb.
Thermal-stability profiling works by comparing the thermal stability of proteins in a sample before and after treatment with a compound of interest. Because binding of a ligand typically enhances a protein's thermal stability, changes in this measure can be used to detect and study protein targets of a particular agent.
Applying a version of the methodology presented last year by researchers at chemical proteomics outfit Cellzome, a part of GlaxoSmithKline's R&D division, Huber and his colleagues coupled thermal-stability profiling to quantitative mass spectrometry, allowing them to looking at ligand binding across the entire proteome.
Like chemical proteomic approaches, the method allows researchers to screen drugs or other agents against wide swaths of proteins to measure binding activity. However, according to Huber, thermal-stability profiling also offers several advantages over conventional chemical proteomic methods.
For one, he noted, chemical proteomics typically uses arrays containing immobilized versions of the agent under investigation. Samples are run against these arrays and the proteins that bind to the immobilized agents can then be measured.
The immobilization process, however, requires modification of the agents being studied, which could, in theory, alter their behavior.
"In most cases we test to see if the modification affects the activity of the chemical compound to the cognate target," Huber said. But, he noted, nonetheless, "you are still changing the molecule."
Unlike conventional chemical proteomic approaches, however, thermal-stability profiling works in intact cells and doesn't require researchers to modify the agents they are investigating. Rather, an agent can be introduced into the cells of interest, which can then be subjected to increased heat sufficient to denature proteins that have not bound to the agent being studied. These denatured proteins will clump together and consequently will not be introduced into the mass spectrometer, allowing researchers to identify those proteins that did bind to the agent.
The ability to work with intact cells also potentially offers a more realistic picture of in vivo uptake and binding of a drug, Huber noted. When used with cell lysates, as is typical of chemical proteomics workflows, researchers may expose their compound of interest to portions of the cell that they would never contact in vivo.
"You have to break open the cell, and so you lose all the [cellular] compartments," he said. "So you may pick up interactions between your ligand of interest and proteins that, actually, your compound may never reach in an intact cell. For instance, you may see some mitochondrial proteins as targets, but in an intact cell the agent would never actually make it to the mitochondria, so they are not real targets."
Indeed, Huber said, he and his colleagues used the technique to compare results from cell lysates versus intact cells and found presumably false positive hits in the cell lysate data that they did not pick up when using intact cells.
In the Nature Methods study, the researchers used thermal-stability profiling combined with mass spec analysis on a Thermo Fisher Scientific Q Exactive to look at the targets of the drugs methotrexate (in K562 leukemia cells) and (S)-crizotinib (in SW480 colon carcinoma cells) as well as the metabolite 2'3'-cGAMP (in RAW macrophages).
In each case, Huber said, they identified the drug's intended target as one of the top hits. They also made identifications of other targets that, he noted, suggested potential additional uses for the technique.
For instance, some of the additional targets identified might represent not direct binding of the drug but, rather, the effect of downstream signaling processes, he said, adding that, while it was difficult at present to distinguish between such events, it could be possible with more experiments and more data to use thermal-stability profiling as a tool for studying not just drug-protein binding, but the signaling networks activated by these initial binding events.
"You are looking at intact protein complexes and intact compartments, so it may actually shed some light on a few signaling pathways that are difficult to tackle if you can't work with intact cells," Huber said.
"We need to do quite a bit of experimenting to develop a better understanding of the basic melting behaviors of the cellular proteins," he added. "But it gets easier the more experience you have."
Also interesting, Huber said, was the fact that in the case of methotrexate the researchers were able to identify not only the drug's target but also proteins known to be targeted by metabolites of the drug.
"If you apply a drug to intact cells, the cells modify the drug and these metabolites [stemming from this modification] may exert a different pharmacology compared to the original compound," he said. "So if you use this methodology, you are able to track these off-target metabolites, which we managed to do in this case."
Huber said he and his colleagues have several research aims for the technique moving forward. For instance, he suggested it might be useful for stratifying patients by their response to given drugs by looking at their target expression levels. Additionally, he noted, the ability to look at off-target effects could make it a useful approach for studies into repurposing existing drugs.
They also hope to use the method to explore the underlying mechanisms of several drugs whose modes of action remain poorly understood, he said, adding that while some groups have attempted to study these drugs using traditional chemoproteomics efforts, he believed thermal-stability profiling might prove more successful due to its ability to work in intact cells.
"The reason no one has found targets may be because you have had to use lysates," he said. "Here you can use an intact cell, so that is definitely something we want to try."
Huber added that the ability of the method to screen metabolites against intact cells could also prove useful in drug development given that many small molecule drugs mimic cellular metabolites.
Specifically, he and his colleagues are now looking to see if they can identify new interactors of 2'3'-cGAMP, which is involved in stimulating production of pro-inflammatory cytokines. Such interactors could be "very interesting targets for immun-oncology and inflammatory diseases," Huber said.