Gene-Environment Interaction in the Etiology of Autism Spectrum Disorder
"We literally went through a period of time where it was either genetics or the environment....
[W]hat's been missing in these studies is really an understanding of how the genetics have worked with environmental exposure."

Interview with Alycia Halladay, PhD, senior director of environmental and clinical sciences at Autism Speaks, the world's leading autism research and advocacy organization. She received her Ph.D. in psychology from Rutgers University, where she worked on animal studies of chemical exposures and their effects on brain development. After a postdoctoral fellowship, Halladay became an Assistant Research Professor at Rutgers and furthered her study of environmental exposures as they relate to autistic regression. In 2005, she joined the National Alliance for Autism Research as the associate director of research for environmental sciences, before it merged with Autism Speaks the following year.
In her current role, Halladay has a hand in all projects, research grants, and initiatives involving gene-environment interactions at Autism Speaks. Founded in 2005, the organization funds a number of studies that explore causes of autism, as well as prevention and treatment options. It also attempts to raise awareness of autism spectrum disorders, and advocate for those with autism and their families.
Interviewed by Meeri Kim, PhD, March 2014
Interviewed by Meeri Kim, PhD, March 2014
You have been involved in autism research for over a decade. How has it evolved, particularly in the light of gene-environment interactions?
I have to tell you, I look back at when I first started in the field 13 years ago: the understanding of what an animal model of autism was was very rudimentary, and what exposures were possibly involved in autism was pretty much an open book. There was so much we just didn't know, and now I can look back and realize how far we've come. We literally went through a period of time where it was either genetics or the environment. I remember going to scientific meetings, and scientists would argue about whether it was genes or the environment. It would get contentious at times. But within the last 3 to 5 years, I've noticed a real trend where researchers who 5, 10 years ago said it had nothing to do with the environment now say that autism is probably a complex interaction of genetics and the environment. So that is movement.
I have to tell you, I look back at when I first started in the field 13 years ago: the understanding of what an animal model of autism was was very rudimentary, and what exposures were possibly involved in autism was pretty much an open book. There was so much we just didn't know, and now I can look back and realize how far we've come. We literally went through a period of time where it was either genetics or the environment. I remember going to scientific meetings, and scientists would argue about whether it was genes or the environment. It would get contentious at times. But within the last 3 to 5 years, I've noticed a real trend where researchers who 5, 10 years ago said it had nothing to do with the environment now say that autism is probably a complex interaction of genetics and the environment. So that is movement.

What has contributed to this shift in thinking, and why does some resistance still exist?
New paradigms are always going to take some time to be accepted. If you think about the percentage of cases of autism that stem from just genetic causes, combined with the ones that stem from just environmental causes — it's not very many. We can identify maybe 10-15% of cases of autism that we can narrow down to a known genetic cause, and probably even less can be attributed to a single environmental exposure. Think about that: it means anywhere between 70 to 80% of the causes of autism are somewhere in between, where there's a mix. Now of course those numbers could change a little the more we know about genetics and the more we understand about multiple environmental exposures. The shift in thinking, I believe, has to do with the increased knowledge of both genetic, epigenetic, and environmental factors. The more research is done, the more it is clear that in most cases, there are multiple causes at play. There is an increase in the prevalence of autism, and a lot of it has to do with increased recognition, increased awareness, and better diagnosis. But there's a lot that we don't know. It could be that autism is on the rise, and if that's the case, why is that? This has investigators thinking.
New paradigms are always going to take some time to be accepted. If you think about the percentage of cases of autism that stem from just genetic causes, combined with the ones that stem from just environmental causes — it's not very many. We can identify maybe 10-15% of cases of autism that we can narrow down to a known genetic cause, and probably even less can be attributed to a single environmental exposure. Think about that: it means anywhere between 70 to 80% of the causes of autism are somewhere in between, where there's a mix. Now of course those numbers could change a little the more we know about genetics and the more we understand about multiple environmental exposures. The shift in thinking, I believe, has to do with the increased knowledge of both genetic, epigenetic, and environmental factors. The more research is done, the more it is clear that in most cases, there are multiple causes at play. There is an increase in the prevalence of autism, and a lot of it has to do with increased recognition, increased awareness, and better diagnosis. But there's a lot that we don't know. It could be that autism is on the rise, and if that's the case, why is that? This has investigators thinking.
How do our genes interact with the environment? Have there been studies that back up this interplay?
When people think about genetics, they think about their structural DNA — the double helix. But there's a number of other mechanisms, called epigenetics, that affect how those genes are being expressed that has nothing to do with the actual coding of the genes. There are different ways that epigenetics can affect gene expression. For example, methylation is one type of epigenetic mechanism: there are regions on your genome where methyl group molecules attach and as a result, that gene is no longer expressed. If there's no methyl group attached, it expresses normally. It's not just DNA which is the major factor. Environmental exposure affects the epigenome and methylation. It doesn't have to be chemicals — there are a number of studies that have found maternal grooming behavior and stress can also have an epigenetic effect. The structure of the genome has received the most attention in autism research, but recently it has been recognized things may influence the epigenome and, therefore, gene expression.
When people think about genetics, they think about their structural DNA — the double helix. But there's a number of other mechanisms, called epigenetics, that affect how those genes are being expressed that has nothing to do with the actual coding of the genes. There are different ways that epigenetics can affect gene expression. For example, methylation is one type of epigenetic mechanism: there are regions on your genome where methyl group molecules attach and as a result, that gene is no longer expressed. If there's no methyl group attached, it expresses normally. It's not just DNA which is the major factor. Environmental exposure affects the epigenome and methylation. It doesn't have to be chemicals — there are a number of studies that have found maternal grooming behavior and stress can also have an epigenetic effect. The structure of the genome has received the most attention in autism research, but recently it has been recognized things may influence the epigenome and, therefore, gene expression.

A number of medical health disorders probably underestimate the value of epigenetic gene expression, with autism being one of them.
If you think about a way you could study epigenetic effects, probably the best way would be to look at twins — identical twins, where one of them has autism and the other doesn't. What's going on there? These people have supposedly the same DNA code. There was a major research publication this past year authored by Wong et. al. They showed that in the twin with autism, there were regions of the DNA that were differentially methylated, and a lot of them were on genes that have been implicated through independent genetic studies to be associated with autism. I think that's huge — definitely an important milestone.
How can an environmental exposure translate into changes in the epigenome?
There's so much science doesn't know, but I'll put out there what people are suspecting. The first is something called a susceptibility gene: sometimes mutations in our DNA could lead to an increased risk of having a physical, emotional, or mental disability. The gene can function and encode a protein normally, but then an environmental exposure completely blocks the effect of that protein. You have two things that by themselves wouldn't have been all that harmful, but given the wrong exposure and the wrong person, it's producing something. There's the possibility of this happening in the individual with autism, but there can also be genetic susceptibility in the mother. For example, some people have hypothesized about antioxidant chemicals that eliminate toxins in the body. Under normal circumstances, a woman's children may be completely healthy; but when they're exposed to some sort of toxic environmental factor, that process is really disabled to the point where they're not able to eliminate toxins in their body, and therefore her fetus is affected.
In epigenetics, which I mentioned earlier, environmental exposures could "turn on" or "turn off" genes by decreasing or increasing methylation. Methylation is just one type of epigenetic mechanism, there are a few others. One thing that really hasn't received a whole lot of attention in epidemiology, but has in the animal model realm, is the idea of a preconceptional exposure, or an exposure that happens before you're pregnant or reproduce. In animal studies, you can take an animal and expose it to a certain toxin. For five generations down, those offspring are still showing either pathophysiological changes or behavioral changes. It's not the chemical itself because it's not in the equation anymore — it's the way that chemical has influenced gene expression in generation after generation.
What are some examples of environmental exposures that could have harmful effects on the epigenome and possibly increase the risk of autism for future children?
The term “environmental exposures” tend to get a bad rap. When people visualize that term, they think of a chemical plant that's emitting all of these toxic fumes, or chemicals that they put on couches to prevent fires. But in reality, it could be a pharmaceutical exposure, a chemical exposure, stress — anything that affects the body's physiology that is not genetic. There's also been a long history of the word "environmental" being tied to maternal neglect in autism, which is awful. They say, oh there must have been something I could have done differently. That's absolutely not the case. You're exposed to things every day that you have no control over whatsoever. I think the word environmental exposure infers that there's something parents are doing, or not doing that they could be doing better, and that's not what it is. One of the goals of better awareness and education is to get people to think of the term differently.
But certainly there are some environmental exposures that have been shown to be associated with autism. One of them is prenatal valproic acid, which is a drug that was used to control seizures. In women who took it while pregnant, their offspring has certainly shown an increased risk of autism. There's been a couple of studies showing that women who are exposed to higher levels of air pollution have an increased risk of having a child with autism. That risk is even higher if their child has a specific genetic mutation. It's not just an either/or thing — it's a combination of two things. There has been a couple studies that show pesticide exposure is associated with an increased risk.
Are there some cutoffs in terms of how much of a certain exposure one should avoid? Have we learned lessons that could help use prevent autism in future generations?
With valproic acid, there is a certain window of susceptibility, which tends to be at the end of the second trimester. Also, since its harmful effects were illustrated, there's been improvements in drug development that have allowed for other classes of drugs that don't have this effect, but at the same time can control seizures. In the case of air pollution, the good news is that clear air standards have been improving air quality for the last 20 years. If we keep on with those efforts, we can get even better and hopefully have an effect on autism rates down the road. In other good news, we also know that folic acid supplementation during pregnancy reduces the risk of autism. On an individual level, there are many things that women can be doing to reduce their exposures. We don't know if these actions will prevent autism, but they are actionable and reasonable. (we have a link on our website about this) The field of epigenetics is just starting to be funded in a way that's beginning to produce results. These results now need to be turned into action via regulatory agencies or professional societies.
Autism is four times more prevalent in males than females. Do you think epigenetic factors might help explain this?
Definitely. Researchers are looking at genes on the sex chromosomes and how they are regulated. For some disorders that resemble autism (like Rett Syndrome), the answer is found though epigenetic regulation of the X chromosome. Some researchers are questioning the existence of a true difference in prevalence between males and females, and consider that females just present the disorder differently — missing full “diagnostic classification.” Also some recent studies show that there might be something going on that protects females against some of the risk factors. We need to better understand the sex differences since that's kind of a huge area about where genes and environment can interact. Here you have this whole extra chromosome that's different, and there's such a disparity in the prevalence of autism, with the disease affecting 4 boys for every girl.
Do you have any recommendations for ways to improve on studies of germline-environment interaction in the etiology of autism and other neurodevelopmental disorders?
None of these things by themselves cause autism, and what's been missing in these studies is really an understanding of how the genetics have worked with environmental exposure. Most of them don't collect a whole lot of environmental information — that's the first problem. The second problem is that most of the information they do collect is at the time the child was diagnosed. There are just a few studies (like MARBLES and EARLI, and possibly the NCS) that are collecting prenatal biosamples longitudinally and also measuring autism as an outcome. Science really needs to make sure we're looking at prenatal exposures during critical time periods. Autism could also be taking advantage of epidemiological projects that are studying other outcomes and then adding on ASD. But at the same time, there needs to be more animal work that studies the multigenerational and transgenerational effects of different exposures. Again, there are only a few research groups doing this. There should be more.
Autism Speaks is funding a project to create a standardized, gestational exposure instrument that could be used across both genetic and clinical studies, for researchers to collect environmental information in a very standardized way. Everybody would collect the same information, ask the same questions, write them down the same way — so that there's consistency. Also, we need to be monitoring exposures and collecting biosamples throughout the lifespan so we can track outcomes later on. So data on my kids, my kids' kids, etc. and be able to track them back. That's partially a possibility in some of the studies from Scandinavian countries because they have these huge medical registries that includes hospital and pharmacy records through multiple generations. For example, gather data from all women who were given a certain drug during pregnancy; you can follow them down: how many of them have children or grandchildren affected by autism?
We need to think about the vulnerability of the germline. We need to think about exposures that can affect neurodevelopment.
How can an environmental exposure translate into changes in the epigenome?
There's so much science doesn't know, but I'll put out there what people are suspecting. The first is something called a susceptibility gene: sometimes mutations in our DNA could lead to an increased risk of having a physical, emotional, or mental disability. The gene can function and encode a protein normally, but then an environmental exposure completely blocks the effect of that protein. You have two things that by themselves wouldn't have been all that harmful, but given the wrong exposure and the wrong person, it's producing something. There's the possibility of this happening in the individual with autism, but there can also be genetic susceptibility in the mother. For example, some people have hypothesized about antioxidant chemicals that eliminate toxins in the body. Under normal circumstances, a woman's children may be completely healthy; but when they're exposed to some sort of toxic environmental factor, that process is really disabled to the point where they're not able to eliminate toxins in their body, and therefore her fetus is affected.
In epigenetics, which I mentioned earlier, environmental exposures could "turn on" or "turn off" genes by decreasing or increasing methylation. Methylation is just one type of epigenetic mechanism, there are a few others. One thing that really hasn't received a whole lot of attention in epidemiology, but has in the animal model realm, is the idea of a preconceptional exposure, or an exposure that happens before you're pregnant or reproduce. In animal studies, you can take an animal and expose it to a certain toxin. For five generations down, those offspring are still showing either pathophysiological changes or behavioral changes. It's not the chemical itself because it's not in the equation anymore — it's the way that chemical has influenced gene expression in generation after generation.
What are some examples of environmental exposures that could have harmful effects on the epigenome and possibly increase the risk of autism for future children?
The term “environmental exposures” tend to get a bad rap. When people visualize that term, they think of a chemical plant that's emitting all of these toxic fumes, or chemicals that they put on couches to prevent fires. But in reality, it could be a pharmaceutical exposure, a chemical exposure, stress — anything that affects the body's physiology that is not genetic. There's also been a long history of the word "environmental" being tied to maternal neglect in autism, which is awful. They say, oh there must have been something I could have done differently. That's absolutely not the case. You're exposed to things every day that you have no control over whatsoever. I think the word environmental exposure infers that there's something parents are doing, or not doing that they could be doing better, and that's not what it is. One of the goals of better awareness and education is to get people to think of the term differently.
But certainly there are some environmental exposures that have been shown to be associated with autism. One of them is prenatal valproic acid, which is a drug that was used to control seizures. In women who took it while pregnant, their offspring has certainly shown an increased risk of autism. There's been a couple of studies showing that women who are exposed to higher levels of air pollution have an increased risk of having a child with autism. That risk is even higher if their child has a specific genetic mutation. It's not just an either/or thing — it's a combination of two things. There has been a couple studies that show pesticide exposure is associated with an increased risk.
Are there some cutoffs in terms of how much of a certain exposure one should avoid? Have we learned lessons that could help use prevent autism in future generations?
With valproic acid, there is a certain window of susceptibility, which tends to be at the end of the second trimester. Also, since its harmful effects were illustrated, there's been improvements in drug development that have allowed for other classes of drugs that don't have this effect, but at the same time can control seizures. In the case of air pollution, the good news is that clear air standards have been improving air quality for the last 20 years. If we keep on with those efforts, we can get even better and hopefully have an effect on autism rates down the road. In other good news, we also know that folic acid supplementation during pregnancy reduces the risk of autism. On an individual level, there are many things that women can be doing to reduce their exposures. We don't know if these actions will prevent autism, but they are actionable and reasonable. (we have a link on our website about this) The field of epigenetics is just starting to be funded in a way that's beginning to produce results. These results now need to be turned into action via regulatory agencies or professional societies.
Autism is four times more prevalent in males than females. Do you think epigenetic factors might help explain this?
Definitely. Researchers are looking at genes on the sex chromosomes and how they are regulated. For some disorders that resemble autism (like Rett Syndrome), the answer is found though epigenetic regulation of the X chromosome. Some researchers are questioning the existence of a true difference in prevalence between males and females, and consider that females just present the disorder differently — missing full “diagnostic classification.” Also some recent studies show that there might be something going on that protects females against some of the risk factors. We need to better understand the sex differences since that's kind of a huge area about where genes and environment can interact. Here you have this whole extra chromosome that's different, and there's such a disparity in the prevalence of autism, with the disease affecting 4 boys for every girl.
Do you have any recommendations for ways to improve on studies of germline-environment interaction in the etiology of autism and other neurodevelopmental disorders?
None of these things by themselves cause autism, and what's been missing in these studies is really an understanding of how the genetics have worked with environmental exposure. Most of them don't collect a whole lot of environmental information — that's the first problem. The second problem is that most of the information they do collect is at the time the child was diagnosed. There are just a few studies (like MARBLES and EARLI, and possibly the NCS) that are collecting prenatal biosamples longitudinally and also measuring autism as an outcome. Science really needs to make sure we're looking at prenatal exposures during critical time periods. Autism could also be taking advantage of epidemiological projects that are studying other outcomes and then adding on ASD. But at the same time, there needs to be more animal work that studies the multigenerational and transgenerational effects of different exposures. Again, there are only a few research groups doing this. There should be more.
Autism Speaks is funding a project to create a standardized, gestational exposure instrument that could be used across both genetic and clinical studies, for researchers to collect environmental information in a very standardized way. Everybody would collect the same information, ask the same questions, write them down the same way — so that there's consistency. Also, we need to be monitoring exposures and collecting biosamples throughout the lifespan so we can track outcomes later on. So data on my kids, my kids' kids, etc. and be able to track them back. That's partially a possibility in some of the studies from Scandinavian countries because they have these huge medical registries that includes hospital and pharmacy records through multiple generations. For example, gather data from all women who were given a certain drug during pregnancy; you can follow them down: how many of them have children or grandchildren affected by autism?
We need to think about the vulnerability of the germline. We need to think about exposures that can affect neurodevelopment.