A team led by scientists at Baylor College of Medicine has elucidated a protein interaction network that indicates that common protein pathways underlie various distinct subtypes of autism spectrum disorders.
The interactome could prove useful in classifying autism patients according to functional pathways and protein interactions, said Huda Zoghbi, a Baylor professor and leader of the study. Ultimately, she added, it could enable more effective development and targeting of therapies for the disorder.
Broadly, autism can be divided into two main classes: idiopathic, which consists of classic autism symptoms like impaired social skills and delayed language development; and syndromic, in which these classic symptoms are part of a broader clinical phenotype, such as fragile X, tuberous sclerosis, and Phelan-McDermid syndrome. Both types, Zoghbi told ProteoMonitor, are highly heterogeneous, with dozens or perhaps hundreds of different genes involved, making study of the disorder a difficult endeavor.
Investigating it "one gene at a time to come up with therapeutics is going to be a big challenge," she said. "So we're trying to identify which subgroups of genes might interact directly, might affect a certain pathway, which will help classify different autism subtypes."
The goal, Zoghbi said, is that by identifying interacting genes and their corresponding proteins, researchers could ID protein pathways that could serve as therapeutic targets for broad "subgroups of autism cases, rather than just one specific subtype."
She compared the work to similar research into stratifying cancer patients according to the pathways active in their tumors. "There are many genetic ways to get cancer, but at the end of the day many times there are certain pathways affected that affect cell growth and division," she said. "So if you intervene at that final point, you can treat many kinds of cancers with one particular class of drug."
"The hope is that maybe the same thing could happen in autism," she added. "There are many molecular perturbations that end up giving you [the autism] phenotype, and putting together the protein [interaction] network is one of the experiments we do to enhance the knowledge about these intermediate steps."
The network could also be helpful in making sense of the sometimes unclear genetic data that's emerged from microarray- and sequencing-based investigations into the disease, Zoghbi said.
"There are lots of sequencing efforts for autism samples, and often you find a lot of sequence variants and can't tell which are important," she said. "Having this network, if you find some of these genes that alter a protein sequence and you find that protein actually existing in the network and interacting with a known autism coding gene, you can pursue that further. It can help you discover new autism genes and be more certain about them as you go back and forth between the genetic and the functional studies."
The study, which was published this week in Science Translational Medicine, set out to identify protein partners of 26 known autism-associated proteins. Using a yeast two-hybrid screen of a human cDNA library, the researchers found 539 protein partners having a total of 848 interactions, only 32 of which had been previously reported.
Among the more notable discoveries was the connectivity between two syndromic proteins – SHANK3 and TSC1 – linked to two different forms of the disorder. These proteins had at least 21 partners in common, suggesting that the distinct pathways they are involved in exist together in a larger protein complex, meaning that the two forms of the disease could share a common therapeutic target.
The study is the second large protein interaction project to emerge from Baylor in recent weeks. In May, a research team led by Baylor College of Medicine professor Bert O'Malley published a paper in Cell detailing the results of an almost decade-long study of endogenous coregulator protein networks that implicated roughly 11,000 proteins in the regulation of gene expression (PM 06/03/2011).
That study used immunoprecipitation followed by mass spectrometry to identify protein interactions, an approach Zoghbi said she plans to use in the next stage of her research.
While yeast two-hybrid systems like that used in the autism work tend to identify direct, transient protein-protein interactions, immunoprecipitation approaches are generally better for isolating more stable interactions like those involved in protein complexes, making them complementary approaches, Zoghbi said.
"One could start from either [approach]," she said. "We were interested in direct partners of autism proteins and now that we have this network where we have the key players and we know what directly physically interacts with what, we are now going to go back and create [antibody] reagents that will allow us to take the same proteomic approach that [O'Malley] took."
"[O'Malley] has a big collection of proteins, and now he needs to figure out what is directly interacting with what," Zoghbi added. "We know what is interacting with what, and now we need to figure out what [protein] subgroups are in stable complexes [together]."
The decision to take the yeast two-hybrid approach also stemmed from a lack of good reagents for immunoprecipitation, she said, noting that the scientists didn't have "outstanding antibodies that allowed you to be sure that you were hitting the right protein and finding its partner."
To get around this lack of antibodies, Zoghbi plans to tag target proteins in vivo in mouse models, an effort she said her lab will begin within the next several months.
"It's going to take us I would imagine months to years to really get that data because we'll have to create special tools" for the tagging, she said. "We're not going to do it for all of the proteins. We're just going to select a few. It will be a long-term experiment."
Have topics you'd like to see covered in ProteoMonitor? Contact the editor at abonislawski [at] genomeweb [.] com.