A team led by researchers at the University of Chicago has completed a protein interaction study that indicates the protein HER3 may play a larger role in cancer than previously thought.
The study, detailed in a paper published last week in PLoS One, measured binding between HER family receptor phosphosites and Src homology 2, or SH2, protein domains traditionally associated with cancer. Using fluorescence polarization to detect low-affinity interactions, the researchers found that HER3 was far more effective than assumed at recruiting SH2 domain-containing proteins.
The results, said Richard Jones, a researcher at the University of Chicago and author on the paper, bolster the case that HER3 could be an important driver of cancers linked to the HER family of receptors.
HER1 and HER2 have long been implicated in cancers like small cell lung and breast cancer, with drugs including AstraZeneca's Iressa and Roche/Genentech's Herceptin targeting these proteins. HER3, however, has not traditionally been considered a good target for treatment, in part because it was perceived to have less ability to recruit the SH2 domain-containing proteins involved in cancer-related processes like mitogenesis, cell survival, and cell motility.
Past studies of this SH2 recruitment, however, have largely relied on protein array-based techniques biased in favor of high-affinity binding, Jones told ProteoMonitor. This, he said, has led to measurements that potentially undercounted more ephemeral binding events.
Using fluorescence polarization, Jones said, he and his colleagues were able to detect significantly more interactions between HER phosphosites and SH2 domains, including many more lower-affinity events. Screening 89 HER tyrosine sites against 93 SH2 domains and 2 phosphotyrosine binding domains, the researchers identified a total of 1,405 unique interactions, 1,169 of them novel. Additionally, the authors found that HER3 was equally effective as HER1 and HER2 at recruiting SH2 domains.
Fluorescence polarization proved more sensitive than protein array-based techniques because it enabled use of higher protein concentrations, Jones said. Both methods use fluorescently labeled HER phosphopeptides, but in the case of the protein arrays, these peptides begin to collect around their fluorescent rhodamine labels when added to the SH2 protein arrays in the high concentrations needed to observe low-affinity interactions.
"They simply start to aggregate and precipitate, and so you have this artificial cutoff with all your interactions based on this problem," he said.
Fluorescence polarization, on the other hand, uses a solution-phase assay where the rhodamine-labeled phosphopeptides are held at a low concentration and the SH2 proteins are added at increasing concentrations, which avoids the peptide aggregation problem.
"We're simply flipping around what we're looking at," Jones said. "We're actually keeping the peptide at a constant nanomolar range and then inferring the interaction affinity based on titrating the proteins in. And because the proteins don't have any hydrophobic moiety attached to them, you can actually go up to quite high concentrations and then you can end up characterizing very weak binding interactions."
After gathering the binding data, the researchers then validated a number of the observed interactions using competition assays, a technique, Jones said, that could provide a more sensitive validation than conventional methods based on affinity strength.
"Typically, in many cases in biochemistry affinities are used as a gauge for whether or not something is real," he said. "So most of the interactions that have been characterized for the particular [SH2] domain families are fairly tight, with a few exceptions that in many cases people have argued maybe aren't real because they aren't high affinity."
"What we're trying to suggest is that affinity isn't relevant for human biology," Jones said. "And in this case it appears that there have been many, many interactions that have been underappreciated, and if you perform real competition experiments it will validate that these really are true interactions between peptides that are going into the binding pockets of proteins. They are simply very low affinity, and very transient."
The findings, Jones noted, lend support to a growing number of studies indicating a potential role for HER3 in cancer.
"HER3 has kind of been ignored in the field because it was thought not to have any tyrosine kinase activity, and therefore it was thought not to have a role in cancer," he said. However, Jones said, in recent years researchers have detected low amounts of tyrosine kinase activity with HER3 as well as a link between increased expression of the protein and poor prognosis in breast cancer patients.
Drug companies have also taken notice, as pharmas including Daiichi Sankyo, Roche, and Aveo Pharmaceuticals have agents targeting HER3 in clinical trials.
"There's an increasing notion that HER3 seems to be playing an important role in [breast] cancer, but it's always been a bit perplexing," Jones said. "How could HER3 play any role? It has barely enough kinase activity to measure. So how could this actually occur?"
"So what our paradigm would suggest," he said, "is that there are actually two parts to the story: One in which the receptor kinase needs to get activated potentially to start off a chain of events, and then the second [part] is [it needs] to actually recruit all the downstream players to make something actually happen. And it appears that in that second step HER3 is better than [HER1 and HER2]."
Jones cautioned, however, that the interactions he and his colleagues observed have yet to be validated in vivo. With that in mind the researchers are currently working to develop an assay for such in vivo validation, he said, noting that conventional immunoprecipitation techniques are ill-suited to such work due to the transient nature of the interactions they hope to validate.
"What we really need is a next generation of in-cell interaction approaches that really dynamically follow and track the interactions of proteins in real time," Jones said, noting that the researchers are working with several other scientists on a biomolecular complementation luciferase assay that could prove useful in their validation efforts.
"The notion is that you simply tag many proteins with half reporter constructs such as fluorescent proteins or luciferase," he said. "And then when two proteins are interacting, even when it is a transient kind of case, they will generate light, and then you can follow it in real time. Then you can ask the questions: Do I see those interactions under specific conditions? And can I make certain point mutations in parts of the proteins that would be predicted to interact and disrupt the observed interaction?"
Jones noted, though, that "to tag so many proteins in cells in specific ways and then make point mutations in specific ways" is a significant undertaking. The researchers, he said, are currently "planning a representative [validation] subset of some of the most interesting interactions that were observed in this particular study."