A Conversation with SFARI Investigators Peter Kind and Loren Frank

Peter Kind is Director of the Simons Initiative for the Developing Brain (SIDB) and professor at the University of Edinburgh. Loren Frank is a professor at the University of California, San Francisco and a Howard Hughes Medical Institute Investigator.

The Simons Foundation Autism Research Initiative (SFARI) recently announced the development of eight new genetic rat models for studies of autism spectrum disorder. The rat models, maintained in the outbred Long-Evans background strain, are available for use from the Medical College of Wisconsin. I recently spoke with SFARI Investigators Peter Kind and Loren Frank on their work with these rat models and their views on how these models will aid autism research.

Peter Kind is Director of the Simons Initiative for the Developing Brain (SIDB) and SFARI’s partner in the behavioral phenotyping of SFARI’s autism rat models. Kind is also a professor at the University of Edinburgh, where his laboratory focuses on understanding the roles that synaptic function and dysfunction play in the manifestation of neurodevelopmental conditions.
Loren Frank is a professor at the University of California, San Francisco, a Howard Hughes Medical Institute Investigator and a SFARI Investigator. Frank’s laboratory has an expertise in understanding how the hippocampus supports memory storage, memory retrieval and memory-guided decision-making. He was previously awarded a SFARI grant to examine learning and memory in Fmr1 rats, a model of fragile X syndrome.
The interview has been edited for clarity and brevity.

Peter, what rat models are you using in your lab, and what do you see as the benefits and limitations of rat models of neurodevelopmental conditions compared with other animal models?

Peter: Between the SIDB Edinburgh site and the Bangalore site, we have somewhere between 15 and 20 different rat models, all to monogenic forms of autism spectrum disorder, typically the high-confidence genes on the SFARI Gene list.

We started with the fragile X and the SynGAP rats because Fmr1 and Syngap are two genes we were particularly interested in. We moved to the rat model primarily because of the complex social behaviors that rats have that mice do not. Rats evolved to live in large groups of up to 150 individuals, and they have numerous cooperative behaviors, such as hunting and playing. Mice are much more loner animals and have evolved far fewer social interactions.

But I would never say one model is better than another model. In fact, I’m a firm believer that we need more models, not fewer. We’ve learned a huge amount from flies, from fish[…]. It really comes down to the hypothesis you’re testing and the experiments you need to do to test that hypothesis. Different models are better than others to address particular questions. We thought the complex social repertoire of behaviors in the rat would be a particular advantage for studying autism.

There is also a lot of autism data coming out of mice1, and we thought that cross-species validation would be very important to understand any translatability across mammalian species2. I know people look at rats and just see big mice, but there’s actually about 12 million years of evolution between them. To put that into perspective, there’s six million years of evolution between a chimpanzee and a human. So rats and mice are very distinct species, but offer many of the same advantages in terms of large litter size and shortened developmental periods. Rat neurons are also much more robust in culture than mouse neurons, and that’s turned into a huge advantage for many of our colleagues. Also, from a lab-specific point of view, we are very interested in using longitudinal functional magnetic resonance imaging, and because of the rat’s bigger brain, we can get much more resolution using rats.

Fmr1 knock-out (KO) mice and rats show species-specific differences in spatial navigation. Spatial learning and memory in the water maze are not affected by the loss of the fragile X mental retardation protein (FMRP) in rats. Fmr1 KO rats exhibited learning that was similar to wild-type rats, as measured as decreased path length to reach a submerged escape platform (A) and increased crossing of the platform (B) over trials. In contrast, prior work has shown that Fmr1 KO mice exhibit learning deficits in this same spatial task when compared to wild-type mice (data not shown; e.g., Baker K.B. et al.1). Image from Till S.M. et al.2

Loren, with your lab’s focus on the fundamental science of how hippocampal circuits support memory and decision-making, what were some unexpected lessons that came out of your lab’s early use of various rat models?

Loren: We have always been a rat lab for some of the reasons that Peter mentioned, including how social rats are. Also, just cognitive capacity-wise, we can get our rats to do more complicated behaviors that we cannot get a mouse to do.

When we started to look at autism rat models, we started with an inbred Sprague Dawley strain. From a genetic homogeneity perspective, inbred is the way to go. But it’s not obvious from a ‘modeling human conditions’ perspective. In fact, we have found that the wild-type inbred Sprague Dawley strain performs at the same cognitive level as hippocampally lesioned Long-Evans animals. Sprague Dawley rats are also closer to blind, so you need really bright lights for behavioral tasks with them. Since we are interested in hippocampal-dependent cognition and behavior in our lab, we always work with an outbred Long-Evans strain.

Peter: I completely agree with what Loren said. Both the Sprague Dawley and Long-Evans models we use are all outbred strains. And the cognitive capacity of the rats is a big plus. We wanted to use some of the complex behavioral tasks developed in Richard Morris’ and Emma Wood’s laboratories, here at the University of Edinburgh, with our autism models, and we can’t get mice to do those tasks.

“…we don't ask if our rats look like a human with the corresponding condition […]. Instead, we ask how a given mutation affects the natural behaviors of a rat.”

- Peter Kind

It sounds like autism rat models have a lot to offer to the field but that the models need to be carefully characterized. The SIDB is taking the lead on behavioral phenotyping the SFARI autism rat models. Peter, can you tell us about this pipeline and the design aspects added to ensure reproducibility and replicability?

Peter: When we set up the pipeline, we wanted our tasks to be as unbiased and as comprehensive as possible. In that context, it’s important to remember that the behavioral repertoire of a rat is very distinct from the behavioral repertoire of a human. So, we don’t ask if our rats look like a human with the corresponding condition, something often referred to as ‘face validity.’ Instead, we ask how a given mutation affects the natural behaviors of a rat, including a lot of spontaneous behaviors that rely on a wide range of brain areas.

To keep the pipeline standardized and to make sure that we can reproduce behaviors across sites, we also set it up at two different locations, running many of the tasks in both Bangalore, India, and here in Edinburgh, Scotland. All of the people across those sites are trained to do the behavioral assays using identical, and very detailed, protocols that take into account not just what the tasks are but how the tasks are run and the context in which they are run. For example, tasks are run for every rat model in the exact same order and at the same point in the animal’s light-dark cycle. Everybody running the tasks and the Principal Investigators involved also meet once a week to make sure things remain standardized. And at those sessions, the discussions can get very lively and everyone contributes to how the data should be analyzed and interpreted. With all of this, we’re trying to build what we loosely call an ‘ethogram’ for each animal model.

“We have to really understand the whole process of what makes these animals different, behaviorally, so that we know how to relate these differences at the neural level.”

- Loren Frank

Finally, with Oliver Hardt, now at McGill University, we’ve also built something that we call the ‘Habitat,’ where 50 to 60 animals can be put in at the same time. We are beginning to characterize how the animals behave in this much more naturalistic environment and to cross-compare that to our more standardized tasks. And that’s getting really exciting. We’re actually revealing phenotypes in some animals that didn’t really show much in our standard pipeline, but when you put them together in these big cohorts, these phenotypes start to emerge.

Loren: It is deeply heartening, Peter, to hear what you are doing. It’s really important to understand the complex behaviors of these animals as a starting point. The brain has changed in a whole bunch of complicated ways as a result of a genetic mutation. We have to really understand the whole process of what makes these animals different, behaviorally, so that we know how to relate these differences at the neural level. It’s an incredibly important and an incredibly hard problem.

Phenotyping rats. Rat models are phenotyped according to a rigorous pipeline that assesses behaviors relevant to autism spectrum disorder, such as social and motor skills, learning and sensory processing. Different cohorts of rats are run through pipeline 1 and pipeline 2 to overcome potential order effects on behaviors and to limit the number of tasks each animal is put through. For both pipeline 1 and 2, some groups live in the Habitat prior to and during participation in the pipelines. (*MoSeq was originally developed by SFARI Investigator Sandeep Robert Datta to study mouse behaviors3.

Loren, your lab only recently became interested in autism research. What prompted this, and what have you found in your studies of the Fmr1 knockout rats?

Loren: Our lab has always been interested in fundamental cognitive abilities, like creating memories through the events of daily life, and using those to make decisions. In early conversations with SFARI, there was a recognition that we might understand enough about these processes to be able to apply what we know to disease models. So we picked Fmr1 because it was well studied on the mouse model side, so we knew we would have something to compare to.

David Kastner, a postdoc in my lab who’s also an M.D. psychiatrist, asked what behavioral assays we could use. One assay used a lot in studies of psychiatric disease in humans, as well as mice and rat models, is something called ‘prepulse inhibition.’ It’s a really simple paradigm. You basically give a first sound and then a loud second sound follows, and then you ask how much the animal jumps. This seems sort of silly in a way, but this behavior is altered in schizophrenia, and it can be different between males and females. At the time, there were various reports in literature that Fmr1 mice were either the same as humans, the opposite, or showed no effect with prepulse inhibition.

And what we found was that we had to reengineer the whole paradigm. The way it had been done before was with an arbitrarily loud first sound, an arbitrary delay and an arbitrarily loud second sound, and then a small number of trials to measure how much that affected the animal. So we came up with a new, consistent set of paradigms and a new mathematical model for understanding prepulse inhibition, and we found no consistent differences in prepulse inhibition across Fmr1 mutant and non-mutant rats. But we did find consistent differences between males and females, so we know that it’s not that we’re just ignoring things4.

It was a real lesson for us. Behavior has to be done really carefully and thoughtfully. Behavior is the output of an incredibly complicated neural process. So all the things Peter was talking about needing, including replicating things across cohorts, are important to really be confident about behavioral results.

Peter: I’ll just add something to that, following on beautifully from what Loren just said. We’ve just characterized another rat autism model that lacks Nlgn3 and found that the animals freeze less in a standard, cued fear-conditioning task5. This would have been typically interpreted as the rats having a problem with fear learning. But by running numerous different tasks, we found that the animals were actually learning, but using a different expression of fear. Whenever you’re faced with fear, you either do fight, flight or freeze. Normally, rats freeze, but our animals were responding with flight. As Loren said, behavior is incredibly complicated, and you need really good behaviorists, and you need to be very careful with your interpretations on what are believed to be standard tests. The tests are standard, but the interpretations can be often very difficult.

Systematic examination of prepulse inhibition. When an animal or person responds less to a startlingly loud noise that has been preceded by a quieter sound, this is referred to as prepulse inhibition (PPI) of the startle reflex. Changes in PPI have been linked to a variety of neuropsychiatric conditions. Using a new method for assessing PPI, Frank’s laboratory recently showed that Fmr1 KO rats exhibited the same levels of PPI response as wild-type rats, but that there were differences between wild-type males and females. Image from Miller E.A. et al.4

The missions of the SIDB are to both uncover biological mechanisms of autism and to develop therapeutic interventions for autism. Peter, in your view, how might research using rat models help in the development of treatments for autism?

 Peter: When people think of treatments, they typically think of therapeutic testing in animal models. Pharmaceutical companies have used rats for decades because they are particularly well suited, as the metabolism of the rat is closer than mice’s to that of a human. However, there is also still much to be learned from fundamental science that has important implications for neurodevelopmental conditions. For example, Adrian Bird’s work on Rett Syndrome showed that if you re-express the gene even late in an animal’s life, the animal recovers amazing function6. That challenged the whole idea that there are critical periods during development when treatments would be more effective. But we have no idea how generalizable this is. So a focal point for SIDB is to determine whether the monogenic forms of autism are issues of neural development or neural maintenance. I’d like to see the field start classifying, gene by gene, which conditions have a critical period for intervention — those that are truly neural developmental — and which are associated with neural maintenance and able to be corrected even later in life. That stratification has huge implications for treatment strategies in humans.

In terms of testing pharmaceuticals, we also need to remember that it’s not just about bringing things into clinical trials. Every time you give a drug, that’s a valid fundamental experiment on cellular biochemistry and physiology as well as the circuits involved.

Loren: I can just add one other perspective on this. People are starting to think about therapies that might live at a circuit level, as opposed to a pharmacological level. These animal models have real power to help us figure out the systems and to know how they respond.

“…the addition of all of these animal models […] is a huge service to the field. It allows you to ask questions about whether there are common phenotypes and common underlying mechanisms.”

- Loren Frank

SFARI has provided the field with a number of resources, including human genetics data with their autism cohorts and a variety of animal models, and they continue to fund innovative autism research. But what in particular would you like to see SFARI doing in the future to help facilitate autism research?

Loren: The question for me is: What are the things that SFARI can do that nobody else is going to do? Their amazing human genetic work, in terms of putting together the Simons Simplex Collection, Simons Searchlight and SPARK cohorts, has really been fairly unique. And the addition of all of these animal models, doing all the work of creating them and making them easily available, is a huge service to the field. It allows you to ask questions about whether there are common phenotypes and common underlying mechanisms.

I’d also like to see them continue the support for fundamental biology that helps everybody in the community. And I’d like them to think hard about other long-term initiatives, like they did with the Circuit Dynamics request for applications. Things that we could really push and that allow people to try out some slightly crazy things that could move the needle and really change things. For example, our lab is thinking about whether we can teach animals to amplify patterns of brain activity — you know, neurofeedback. And if so, can we then translate that to disease models where we’re using in-built learning mechanisms to push the circuits to a state that’s back toward the standard behavior.

Peter: SFARI has always appreciated the importance of fundamental biology and exploratory science and that you often don’t know where the next big discovery is going to come from. We can learn amazing things about fundamental and autism biology. I think SFARI has done an incredible job at casting that wide net. The SFARI meetings are definitely amazing, bringing everyone together and fostering collaborative work.

SFARI could also play a big part in guiding clinical trials. If a scientific standard was set by SFARI, the premier foundation for autism research, it would change how we think about translating something through to a clinical trial.


  1. Baker K.B. et al. Genes Brain. Behav. 9, 562-574 (2010) PubMed
  2. Till S.M. et al. Hum. Mol. Genet. 24, 5977-5984 (2015) PubMed
  3. Wiltschko A.B. et al. Neuron 88, 1121-1135 (2015) PubMed
  4. Miller E.A. et al. Mol. Psych. Epub ahead of print (2020) PubMed
  5. Anstey N. J. et al. bioRxiv (2020) Preprint
  6. Guy J. et al. Science 315, 1143-1147 (2007) PubMed
Recent SFARI Conversations