Proteins that rise and fall together
Quantitative mass spectrometry and bioinformatics pave the way for identifying new drug targets
by Parizad Bilimoria
Too much junk building up in your Inbox? Just like you might go through your thousands of messages and flag the ones to delete, your cells have a system for sorting through their thousands of proteins and marking the ones to trash.
This system is important in all cells, and may be especially important in the brain, where buildup of unwanted proteins could impair how we think or behave. Cells have a huge molecular ensemble devoted to this housekeeping task, including key players known as ubiquitin ligases—enzymes that target specific proteins that need to go. Working in concert with other enzymes, they attach flags to these proteins, consisting of the small protein ubiquitin. The flags invite the attentions of the cell’s protein degradation machine.
Genetic studies have revealed that specific ubiquitin ligases are mutated in many neurologic conditions, ranging from degenerative diseases like Parkinson’s to developmental disorders like Angelman syndrome or autism spectrum disorders. By discovering what proteins these enzymes normally flag for destruction, researchers hope to reveal the cellular pathways disrupted in these conditions and uncover new drug targets.
But finding these targets is easier said than done. That’s why Judith Steen, PhD, assistant professor of Neurobiology at the F.M. Kirby Neurobiology Center at Children’s, is developing new methods in proteomics and bioinformatics to identify these proteins, especially those active in neurons.
Ubiquitin ligases are enzymes that tag proteins with ubiquitin—a molecular flag that promotes the proteins’ degradation. Ubiquitin ligases are mutated in a number of neurologic diseases.
Judith Steen, PhD, is developing new methods in proteomics and bioinformatics to identify the proteins normally targeted by ubiquitin ligases.
A tough quest
Classic methods of finding targets of ubiquitin ligases are cumbersome and not at all comprehensive. Sometimes, a particular sequence in a protein’s amino acid code suggests its susceptibility to a particular ubiquitin ligase. But these sequences aren’t always known, and aren’t reliable enough indicators on their own.
Instead, researchers must perform labor-intensive biochemical assays. Imagine you’re looking for proteins targeted by a ubiquitin ligase known as the anaphase promoting complex (APC), famous for its control of cell division. You’d start by making all your candidate proteins and tagging them with radioactivity. Next, you’d gather extracts from cells in different stages of the cell-division cycle (duplication of DNA, lining up of chromosomes, actual cell separation, etc.), throughout which APC activity is changing. Then you’d mix together the proteins and the cell extracts in different combinations, looking to see which radioactive proteins are diminished when APC activity is high—those would likely be its targets—and vice versa.
All of this would take months, and only at the final step of the experiment would your main work begin: You’d have to go through the complex mix of molecules one-by-one to look for changes in the abundance of each test protein.
To speed up this type of analysis, the joint laboratory of Judith Steen and Hanno Steen, PhD, director of the Proteomics Center at Children's Hospital Boston, has been developing a completely different strategy. Judith Steen recently applied it to studying Angelman syndrome, a rare developmental disorder causing intellectual disability, speech impairment, seizures and motor difficulties—best known for its feature of frequent smiling and laughter.
In Angelman syndrome, and in some cases of autism spectrum disorder, the ubiquitin ligase Ube3a is mutated or lost. But why and how does this cause disease? Collaborating with Michael Greenberg, PhD, now chair of Neurobiology at Harvard Medical School, Steen began looking for answers.
Hunting in the brain
It was already known that neurons without Ube3a have defective synapse plasticity, meaning that their synapses—the points of contact with other neurons—have trouble tuning their properties in response to the inputs they receive. This is the process thought to underlie learning and memory, and problems at synapses may underlie Angelman syndrome and autism.
But how exactly does Ube3a control synapse plasticity? Genetic studies have shown that Ube3a’s job as a ubiquitin ligase is clearly relevant to its role in disease, but finding the synapse proteins that Ube3a might degrade has been no easy task. Although Ube3a was identified as the causative gene for Angelman syndrome in 1997, a decade later its synaptic targets remained unknown.
Studying mice, the Judith Steen and Greenberg labs put quantitative mass spectrometry to work. Mass spectrometry is a way of identifying the proteins in a complex mixture, in this case, Ube3a targets—ubiquitin-tagged proteins that were abundant in the control mice but missing in mice lacking Ube3a.
In mass spectrometry, proteins are first digested into little pieces using enzymes that act as scissors, cutting at predictable sites. These protein bits, or peptides, are introduced into the mass spectrometer in electrically charged forms (the charge is crucial for what comes next).
Once in the mass spectrometer, these peptides are collided with molecules of inert gases, like helium. The collisions fragment them further, and the precise mass of each fragment is then measured. The mass, combined with knowledge of the electric charge, tells researchers which amino acids make up the peptide fragment. When pieced together, this information ultimately reveals protein identity.
Quantitative mass spectrometry goes a step further, revealing how much of a protein is present in a sample. When looking for the proteins tagged by ubiquitin ligases, quantification is invaluable, because changes in a protein’s levels are often the main clue that it’s being degraded.
The final step is bioinformatics—strategic analysis of the quantitative mass spectrometry data. A particularly useful tool is “co-regulation” analysis, which combs the data for proteins whose levels go up or down together—such as the targets of a specific ubiquitin ligase.
To identify potential targets of Ube3a, the Steen and Greenberg teams performed quantitative mass spectrometry, looking for proteins that were tagged with ubiquitin (blue “Ub”) in mice with Ube3a, but not in those without Ube3a. These proteins, “co-regulated” by Ube3a, are denoted by the yellow circles and pink ellipses. Image courtesy of Sasha Singh.
Through these experiments and analyses, conducted by lab manager Zachary Waldon, the Steen-Greenberg team found sacsin. This protein is known to be mutated in Charelvoix-Saguenay spastic ataxia, a neurologic disorder with intriguing similarities to Angelman syndrome. But because of its extremely large size, sacsin proved hard to study. So, adding a new twist to their strategy, the team compared sacsin’s amino acid sequence with that of a previously-known target of Ube3a and defined a trademark sequence that the two shared, and that may be common in many Ube3a targets.
Another brain protein called Arc, thought to be essential for synapse plasticity, carries a similar sequence. Biochemical experiments confirmed it as a bona fide target of Ube3a in the brain.
Publishing in Cell in 2010, the researchers, led by Paul Greer, PhD and Rikinari Hanayama, PhD of Harvard suggest that Ube3 controls synapse plasticity by controlling the number of neurotransmitter receptors in the synapse—at least in part by tagging Arc for degradation. Arc promotes the traffic of receptor molecules out of the synapse, so having too much Arc—as happens in mice without Ube3a—results in too few neurotransmitter receptors in the synapse. This, in turn, results in decreased synaptic transmission. These findings could help explain the cognitive dysfunction seen in Angelman syndrome and suggest new avenues for drug development.
In the pipeline
Steen is now setting her sights on other ubiquitin ligases active in neurons—including Parkin, which is mutated in some familial forms of Parkinson disease. “I’d like to apply this approach to several other ubiquitin ligases,” she says, “Now that we have the pipeline to this kind of work.”
Postdoctoral fellow Marc Kirchner, PhD, in the Steen & Steen Laboratory is developing a new strategy for co-regulation analysis that will be invaluable in finding ubiquitin ligase targets. His method, described in the journal Bioinformatics, will allow researchers to better identify the co-regulated proteins in quantitative mass spectrometry experiments that employ a system called isobaric mass tagging. This tagging can be used to follow protein dynamics in different biological situations, such as different parts of the cell cycle.
“We define a reference behavior, and then we try to find proteins that fall into that reference behavior,” explains Kirchner. “Then we rank them based on how similarly to the reference they behave, and that gives the biologist the short list.”
In the case of ubiquitin ligases mutated in neurologic disease, this short list of targets might be a precious set of clues into the signaling pathways that the disease disrupts. Kirchner likens the process to finding needles in a haystack. The mathematical advances of his co-regulation analysis get rid of a lot of the hay, and make it easier to apply knowledge of the needles we already know about to the search for unknown ones.
“We tailored the statistics to the structure of the data,” Kirchner says. “And this increased the power of our statistical tests. And because we tailored the statistics, the question of how similarly two proteins behave now has a more accurate answer.”
Together, the Ube3a tale and improved bioinformatics techniques set the stage for bigger and bolder hunts for enzyme targets. Just like a suspect’s trash gives detectives clues about his behavior, the proteins that a ubiquitin ligase marks for disposal will likely give insight into its function in the brain, and provide new drug targets in the battle against neurological disorders.