Using yeast protein arrays made by Invitrogen, scientists at Yale have mapped over 4,000 different phosphorylation events involving 82 yeast kinases.
The study, published in last week's Nature, is the first one to use protein chips to identify phosphorylation events in a high-throughput, large-scale manner. Prior to this study, only about 160 interactions between yeast kinases and their targets had been identified, according to the scientists.
The new phosphorylation map could be scrutinized by drug makers who are looking at human protein kinases as drug targets, said Jason Ptacek, the first author of the study and a graduate student in Michael Snyder's lab. Because protein kinases can turn pathways on and off through phosphorylation events, they are a popular class of drug targets. Drugs that target protein kinases include the leukemia drug Gleevac and the breast cancer therapeutic Herceptin.
"Two-thirds of proteins in yeast are conserved in man, so many kinases have a comparative protein in humans. I'm sure people will look up their [homologous] kinase in yeast and then investigate if that kinase also phosphorylates that protein in humans."
"I'm sure somebody who's studying a human kinase will take a look at this work that's been done one yeast," said Ptacek. "Two-thirds of proteins in yeast are conserved in man, so many kinases have a comparative protein in humans. I'm sure people will look up their [homologous] kinase in yeast and then investigate if that kinase also phosphorylates that protein in humans."
In traditional kinase studies, researchers localize target substrates of interest within a cell and run assays to see if the kinase is able to phosphorylate them when added to the outside of the cell, Ptacek said.
Ptacek's large-scale study came about after a study by Stuart Schreiber and Gavin MacBeath, published in Science in 2000, showed that a kinase is able to phosphorylate a protein on a glass surface.
"That was the proof-of-principle study, but it only involved three proteins," said Ptacek.
In the current study, Ptacek and his colleagues applied each of the 82 unique kinases to two identical yeast proteome chips containing 4,400 proteins each. Each kinase was applied to two proteome chips because the researchers wanted to make sure that the phosphorylation events were replicable.
After about two and a half years of these experiments, results yielded 4,192 interactions between the yeast kinases and around 1,300 different protein targets. The interactions were compiled into a searchable map, which is available online at http://networks.gersteinlab.org/phosphorylome/.
"One of the most challenging parts of the study was the data analysis," said Ptacek. "We had to write new software algorithms to identify hits."
Another challenge was to get the phosphorylation assay working with as many kinases as possible, Ptacek added.
"It took a while just to get the assay working to make it such that we could do it with different kinases," he said.
To validate that the phosphorylation being seen on the chips was actually happening in vivo, the researchers removed certain kinases of interest from cells and showed that the substrate target that had been identified was no longer phosphorylated.
"We did that for about a dozen substrates," said Ptacek.
In analyzing their phosphorylation map, Ptacek and his colleagues noticed several patterns in protein phosphorylation. One pattern was that in many cases, both the kinase and substrate physically interacted with another protein. This pattern is compatible with the theory that an adapter protein mediates the phosphorylation event by bringing the substrate and the kinase together.
Another pattern that the researchers noticed was that the same transcription factor tended to turn on both the kinase and the target.
The number of substrates that a kinase phosphorylated ranged from three to 270. On average, each kinase phosphorylated 50 substrates.
In several cases, the researchers noticed that kinases that appeared to be very similar had very different substrates. In fact, three-quarters of the substrates identified were recognized by only one or two kinases.
"That was something we didn't know," said Ptacek. "We thought perhaps the kinases would have a much higher degree of overlap between what substrates they recognize."
In following up on the yeast phosphorylation mapping study, the researchers would like to identify sites of phosphorylation, and to understand the functional significance of the phosphorylation events.
"If we can mutate kinases so that they no longer phosphorylate their target, we might be able to see a phenotypic change in cells that would help us understand the functional significance of the kinase," said Ptacek. "For most of the phosphorylation events that we identified, the functional significance is not known."
The researchers would also like to do a similar type of mapping study using human kinases and human protein chips.
"That job will be much larger," said Ptacek. "Humans have 158 kinases. Yeast only have 120. Also, the number of proteins in humans is significantly larger."
Ptacek estimated that there are about one million different proteins in humans. That number accounts for different splice variants and protein modifications.
A human phosphorylation mapping study would utilize Invitrogen's ProtoArray Human Protein Microarray. Launched about a year ago, the array contains more than 1,800 human proteins from the kinase, membrane-associated, cell-signaling and metabolic protein classes.
— Tien Shun Lee ([email protected])