Craig Powell, M.D., Ph.D., is an assistant professor of neurology and psychiatry at the University of Texas Southwestern. Dr. Powell developed the first-ever animal model of autism, and he is the mentor of ASF 2011 post-doctoral grantee Haley Speed. ASF intern Max Rolison interviewed Dr. Powell about his current research and the use of mouse models in the study of autism.
Max Rolison: How did you get interested and involved in autism?
Craig Powell: I got involved in autism research when I was interested in molecular basis of cognitive behaviors. I learned that some of the trans-synaptic cell adhesion molecules that Dr. Sudhof, my colleague was studying, were implicated genetically in autism. And at the same time another colleague Luis Parada was studying mice with a conditional knockout of the gene PTEN and as I was helping him to characterize the behavior and cognitive function of the animals, a paper came out suggesting that the PTEN mutation was linked to autism, and so suddenly I was studying three or four different mouse models of genetic causes of autism. And that’s basically how I got my start.
MR: What is a knockout and knock-in mouse model?
CP: A knockout mouse model is one in which you remove the gene of interest, and a knock-in mouse model in which you place something into the mouse genome instead of normal gene. For example, putting an autism-causing mutation into a mouse gene.
MR: How do you determine which genes to isolate?
CP: It is determined by studies conducted by human clinical geneticists. There are several large groups of labs working in human clinical genetics. Those labs identify genes that are associated with autism. Once those genes are identified, the gene that is mutated, we can then model that mutation in a mouse and study the effects of that mutation in greater depth. There are very similar genes in humans and in mice, so transitioning from a human to a mouse works. In a mouse model we are able to see what the functional relevance of a gene is by knocking in or knocking out that gene and observing the alterations that occur. To simplify, we model a genetic mutation in a mouse model in some way and then see how the mutation affects brain function and how the brain function is abnormal compared to a mouse without that mutation.
MR: What symptoms of autism are you focusing on with the mice?
CP: We are not focusing on any particular behavior, by design. We have to rely on the particular mouse model and what it is telling us. Some of our models present with social interaction deficits, cognitive problems, and intellectual disabilities. In some mice we see only repetitive behaviors. For example, mice spending twice as much time in self-grooming behaviors compared to normal mice. Additionally, we see mice with an insistence on sameness. They don’t switch as well when they are taught to do a task one way and then have to change to do it a new way, which is characteristic of sameness and repetitive behavior. Our goal is to try to use mouse models to see what behavior abnormalities arise, and if we can, fix any behavior abnormalities. We focus in on parts of brain that are responsible for those abnormalities, and go from there.
MR: What is the basis of your selection of genes? Do you create models based off of tissue and genetic work in humans, or do you create models based on phenotypic presentations?
CP: My laboratory prefers to look at genes that are involved in the function of the synapse, the connections between neurons in the brain. We look at cell adhesion molecules, including neuroligins and neurexins. Shank3, a synaptic protein that binds to other proteins in the post synaptic neuron, is a protein that we have studied a good deal. My lab is interested in synaptic function and its impact on subsequent behavior.
Other labs study human genes and report their findings, which help us to determine which genes to study. But even when you isolate a gene, there is no obvious way to treat that genetic defect in humans. We only know that genetic defect causes autism symptoms in humans. We then use the mutant gene in a mouse model and create the same mutation in a mouse. With the mouse, we can then look at behavioral abnormalities, such as repetitive behaviors, social deficits, communication deficits, impaired motor function, heightened anxiety, seizures, learning and memory problems, and intellectual disabilities in mouse models. We look to see if the mutation in a mouse results in symptoms that can be associated with autism, as seen in humans.
At that point, we only know that the mutated gene causes abnormal behaviors. Then my lab tries to figure out what are the brain function abnormalities based on the mutation. We take the brain out of the mouse and do molecular studies to see the changes. We isolate synaptic proteins. We also take slices of the brains and record neural function in live tissue of brain, which involves poking electrodes into tissue and attaching electrodes to cells and recording the effects. We then take what we learned about the brain, and see what we can fix in the mouse to make its behavior closer to normal. For example, if there is an over-activation of kinase or an enzyme, we can give an inhibitor to make the activity closer to normal. Then, we look to see how the treatment affects behavior.
We are focused on finding novel therapeutic targets. All central nervous system drugs work by binding to molecules in the brain. So, if we can find out what’s wrong on a molecular level, we can find molecular targets for drugs. At this point, we are using some drugs that already exist to treat problems that we find.
Because we are looking at one specific genetic cause at a time, a treatment identified may only treat one specific type of autism. But we hope that multiple genetic causes will converge on the same pathways for treatment. There are many different types of autism, so they won’t all converge, but we hope that several do so they can be treated in the same manner.
MR: How do you develop mice to have the specific genes that you want?
CP: It involves making a DNA construct, transfecting mouse embryonic stem cells with the gene, and then generating the embryonic stem cell into the blastocyst stage, and putting the blastocyst into a pseudo-pregnant female. The female then gives birth to mice with some cells with the normal gene and some cells with the mutated gene. This mouse with multiple types of cells is known as a chimera. There are some cells with the mutation and some without the mutation. Our hope is that some cells in the testicles or ovaries will have mutant cells, so that the chimeric mouse will produce gametes with the mutant gene when it is bred and transmit a mutation in the germ line. The offspring with that mutation is then our mouse model.
MR: How does your work with mouse models compare to post-mortem human brain tissue research?
CP: Post-mortem tissue work is incredibly important because it is one of the few ways we get any clue about what happens in human brains. The problem is that there are very few brains available to study. It is a limited resource and it is used up very quickly. Also, the brain tissue is no longer living. In mouse models, you are able to study living tissue and observe electrical impulses in and between neurons. As a result, in post-mortem brain tissue, you can measure the levels of proteins, the structures of certain neurons, and the brain as a whole, generally, but you can’t study function. Studying mouse models is very useful to correlate what is found in human brains with what is found in animals. We can’t study function in the post-mortem brain, so we can only see a trace of what might have been happening.
MR: What are you currently researching in your lab?
CP: In particular, we are studying several Shank3 gene mutations, neuroligin3 mutations, neuroligin1 mutations, and neurexin mutations. We take recordings in the brain from the hippocampus, cortex, and dorsal striatum to see the electrophysiological differences stemming from these mutations. We are starting with the study of synaptic genes and going to structural, functional, and eventually behavioral outcomes. Our ultimate goal is to use a drug to reverse the brain function problem and then see if that drug reverses or treats the behavioral differences in the genetic mouse model.
MR: Can you explain the role of neuroligins in autism?
CP: We know that there are a handful of families that have affected individuals with mutations in neuroligin 3 or neuroligin 4. We know that mutations in neuroligin 3, neuroligin 4, and neuroligin 1 are associated with a very small percentage of autism cases in humans. What we believe is that when you perturb these neuroligins, you alter the connections between neurons and the brain, which then alters the function of synapses, and eventually alters higher order cognitive function. We understand a lot more about what a neuroligin binds to at a synapse and what happens when you delete or mutate a neuroligin in mouse models, but we haven’t yet pinned down how the abnormality leads to autism in humans.
MR: What value does this research have for families?
CP: In the case of a few genetic causes of autism, such as Fragile X and tuberous sclerosis, drugs that are used in mouse models have been used in the treatment of human patients. We can identify potential novel drug targets in mouse models, and then use drugs that affect that target in a clinical setting once we know what the drug does in the model. That method has been used before, and we are continuing on that path.
Without the use of mouse models, the only alternative is to pick a drug that you think might help with autism and see if it works in treatment of patients. There are almost an infinite number of drugs and compounds that can potentially be given to children with autism. Through the use of mouse models, we can make much better educated guesses as to which drugs will be effective in treating children with autism.
We can’t study the brain of a living child in as much detail as we can the brain of a mouse. Studying mouse models is the only way we can learn about what is happening in enough detail at this point.
MR: Have any treatments been developed that address your findings?
CP: Not yet. One of our mouse models with Dr. Luis Parada’s lab looked at a PTEN conditional mouse model. Among autism patients with excessively large heads, about 17 percent have been found to have overactive mTOR kinase protein. The drug Rapamycin decreases overactive mTOR. Rapamycin reverses molecular abnormalities and social deficits in the PTEN mouse model. What we would like to do next is to give Rapamycin or a similar drug to patients with a form of autism that is known to be associated with a PTEN gene mutation. The issue with this, though, is that we need to find patients because a PTEN gene mutation is not something that is commonly tested. We only want to try Rapamycin on patients with autism with that specific mutation. Once we do this, it might lead to other pilot studies in other types of autism, but that is too far off and too speculative.
Currently, Abilify and Risperidone are the only FDA approved drugs for the treatment of autism. At this point, no mouse models have led to novel treatment for autism; however, studies in humans haven’t identified truly novel drug treatments either. The drugs that are currently used to treat autism were not developed based on something found in human autism studies. But researchers are on the cusp of developing treatments for fragile x and tuberous sclerosis, based on mouse models. Fragile x is the most prevalent, known genetic cause of autism. If we can begin to slice up the autism pie into different genetic and environmental causes, we hope to treat each sliver in a customized manner. Really where we are headed is to identify custom treatment based on a mutation. Then, we can see if there are common themes among different mutations.