NEW YORK (GenomeWeb) – A team led by researchers at Toronto's Lunenfeld-Tanenbaum Research Institute has completed an analysis of tumor necrosis factor-related apoptosis-inducing ligand (TRAIL) signaling that identified potential cancer drug targets that could be used in combination therapy.
Detailed in a paper published this week in Science Signaling, the study used a combination of RNA interference and cDNA overexpression screens along with immunoprecipitation mass spectrometry and phosphoproteomics to map protein signaling linked to TRAIL-induced apoptosis.
Via this analysis, the Lunenfeld-Tanenbaum researchers identified 169 protein kinases that appeared play a role in TRAIL-induced apoptosis. Using an information flow analysis to map links between these kinases and substrates undergoing significant changes in phosphorylation in response to TRAIL stimulation, they identified 10 kinases with particularly strong links to TRAIL-induced apoptosis.
As the Science Signaling authors noted, TRAIL has been shown to preferentially induce apoptosis in tumor cells, making it a potentially promising therapeutic agent. However, early clinical trials have found that many tumors are resistant to TRAIL or quickly develop resistance to it.
With this in mind, the researchers set out to profile signaling pathways involved in TRAIL-induced apoptosis in hopes of better understanding the underlying mechanisms and identifying proteins that could be targeted as part of combination therapies designed to counter the development of resistance, Karen Colwill, a Lunenfeld-Tanenbaum researcher and senior author on the paper, told GenomeWeb.
More generally, Colwill said, she and her colleagues were interested in investigating workflows for integrating phenotypic data generated through approaches like RNAi loss-of-function screens with signaling network maps built via methods like IP-MS and mass spec-based phosphoproteomics.
"We wanted to do this in a system that we thought would have some medical relevance," she said. "And certainly the findings that TRAIL is able to preferentially induce apoptosis in tumor cells was quite attractive, and knowing that in clinical trials right now they are seeing a lot of resistance [to TRAIL therapy], it just made sense to see if we could look for a candidate that could be used in combination with TRAIL."
The researchers first used an RNAi screen targeting 715 transcripts to protein kinases and other kinase types to identify kinases involved in TRAIL-induced caspase activation in the colon cancer cell line DLD-1. They combined this with a cDNA overexpression screen of 107 protein kinases, identifying between the two screens 221 proteins and 169 protein kinases – 137 (109 protein kinases) of which functioned to increase resistance to TRAIL, 80 (56 protein kinases) that were sensitizers, and four (all protein kinases) that either increased resistance or sensitized depending on the screen.
Colwill and her colleagues then performed IP-MS using as baits 104 of the 107 kinases investigated in the overexpression screen. Running the analysis on an ABI QSTAR Elite QTOF, they mapped the protein interactions of these kinases, building a network consisting of 1,646 interacting protein pairs, with the majority of kinases having between two and 10 interactors, with a median of nine interactors per protein.
They followed this IP-MS analysis with a phosphoproteomic analysis looking at changes in protein phosphorylation levels following TRAIL stimulation. This, the authors noted, helped build out their network map by identifying proteins affected by TRAIL stimulation but which might not have been detected in the IP-MS analysis due to the transient nature of their interactions with the bait kinases.
Using a Thermo Fisher Scientific LTQ Orbitrap, the team investigated protein phosphorylation levels in SILAC labeled DLD-1 cells stimulated with TRAIL, measuring phosphorylation at 0, 5, and 60 minutes after stimulation. They identified a total of 4,701 unique phosphopeptides, 353 of which were significantly modulated by TRAIL.
Colwill said that while researchers have previously done large-scale RNAi screens to identify proteins involved in TRAIL-induced apoptosis, the Science Signaling study was the first effort she was aware of to integrate this sort of information with protein interaction and phosphoproteomic analyses.
"We did [the] screens ... to identify the kinases that have some influence on TRAIL signaling," she said. "And then we wanted to [investigate] how they might be manifesting this influence by looking at the physical network and seeing if we could map things out. So that is really where the [interaction] network came in and then also the phosphoproteomics."
To further winnow down their list of interesting kinases for follow-up, the researchers used an information flow analysis in which they tried computationally to sort out how signals traveled between kinases, and between kinases and substrates affected by TRAIL stimulation. In this way they were able to identify the apoptosis-modifying kinases most closely linked to substrates modified by TRAIL stimulation, the idea being that these kinases played particularly important roles in TRAIL-induced apoptosis.
In total, the researchers identified 10 protein kinases that they recommended "be further investigated as possible candidates for combination therapy."
They followed up with additional research into two of these 10, AAK1 and PXK, identified as contributing to TRAIL resistance. Measuring the uptake of TRAIL in DLD-1 cells overexpressing either of the two proteins, the researchers found that they promoted TRAIL receptor endocytosis, which, they noted, could account for their involvement in resistance to TRAIL-induced apoptosis.
These were particularly interesting targets, Colwill said, given that they had not previously been tied to apoptosis.
"There are actually a fair amount of studies that have been done on kinases linked to apoptosis, but AAK1 and PXK really hadn't been identified before," she noted. "So they were novel."
Colwill said the paper was intended in part as a resource the research community could use to identify possible targets worth following up on.
"We were really just presenting our screens and the possibilities," she said. "One thing to do further now would be to study some of the proteins we had hits from in the screens to see if we could validate them further."