AudioHelicase podcast: Whitehead's Mary Gehring discusses how gene regulation can be passed from one generation to the next

In this episode of AudioHelicase, Whitehead Institute Member Mary Gehring discusses how her research on the plant Arabidopsis thaliana reveals how gene regulation can be passed from one generation to the next.

Subscribe to AudioHelicase on iTunes or SoundCloud and never miss an episode!


Lisa Girard: Welcome to AudioHelicase, the podcast of Whitehead Institute, unwinding the science and the people behind some of the Institute’s most exciting discoveries. I’m Lisa Girard, Director of Communications at Whitehead Institute. In this episode, I’m talking with Mary Gehring, Member of Whitehead Institute and an associate professor of biology at MIT.  Mary’s work is focused on understanding how epigenetics controls gene expression in the model plant Arabidopsis. While many of us are familiar with the idea of the sequence of our DNA itself dictating when and how much our genes are expressed, it’s been discovered that there is actually another way that some genes can be heritably regulated that’s not due to our DNA sequence, but instead for example to chemical tags that may be present on different genes, which impact their expression. Epigenetic dysregulation has been associated with a host of diseases including fragile X syndrome and Rett syndrome, and uncovering its mysteries will contribute to a more holistic understanding of gene regulation.

Girard: So Mary, I’ve heard various definitions of epigenetics, but how would you define  it?

Mary Gehring: The question of what is epigenetics is obviously, you won’t get the same answer from two biologists, necessarily. For some people, epigenetics just means gene expression and how gene expression is regulated. For me that’s a little too broad. When I think about epigenetics, I think about heritability also being a component—heritable changes in phenotype that are not dependent on the DNA sequence. And phenotype could be a developmental phenotype, or it could be something like a gene expression state. Even when we say epigenetics is something that happens that is not depending on the DNA sequence, in reality, it’s really hard in many cases to prove that DNA sequence isn’t important. So I think there are some things that maybe we think now are epigenetic, but in the future we might find a different explanation.

Girard: If epigenetics is the heritable changes in phenotype that aren’t dependent on the plant’s DNA sequence, what’s the epigenome?

Gehring: So when we’re talking about characterizing the epigenome, we’re usually referring to looking at what is actually there on the DNA, what are the modifications that are present. So, in my lab, we work a lot on DNA methylation, which is a modification to cytosine. We’re also talking about when we’re talking about epigenome even perhaps things that are associated with the genome, like non-coding transcripts or scaffold transcripts that are potentially interacting with the DNA. All of these modifications—DNA methylation, histone modifications, histone variants—all of these are found across eukaryotes. Now there’s different lineages that may have more or fewer of these.  Some species have gained or lost some of these mechanisms, but they are very, very common across eukaryotic life.

Girard: “Eukaryote” describes a diverse group, including plants, animals, fungi, and other organisms.  Why have you focused on epigenetics specifically in plants?

Gehring: I’ve always thought plants are fascinating, because they are so plastic, they’re so responsive to the environment and all of these external signals. I think these questions of how do you actually regulate those things at the level of the genome, at the level of the DNA, is really interesting.

Girard: Which aspects of epigenetics in plants is your lab currently working on?

Gehring: We’re broadly interested in understanding epigenetic’s dynamics during plant growth and development. A couple of key areas that we’re interested in is how is epigenetic information inherited from one generation to the next. We know in plants that a lot of epigenetic information appears to be inherited between generations, and so what factors influence that? Whether that’s specific features of the genome or the epigenome or specific genes and proteins that are important. That’s one broad set of questions.  

Another broad area is really understanding what’s happening to the epigenome during reproductive development in plants, because we know that it’s really at the reproductive stage of the life cycle that we see remodeling of the epigenome in this extra-embryonic lineage called the endosperm—this tissue that’s also formed by fertilization, like the embryo, that’s in seeds, and it supports the embryo’s development during seed development. Endosperm is not only a good model, but really the only tissue where imprinting takes place in plants, so that’s why we’re looking at it.

Girard:  So you mentioned imprinting. Can you tell a little bit more about what imprinting is?  

Gehring: Imprinting is when alleles or copies of genes are expressed differently depending on whether they are inherited through the male parent, through the dad, or through the female parent, the mom. There is a set of genes in plants that are imprinted. There’s somewhere around 100 or 200 genes that are imprinted. And about 50 of those are primarily expressed from the copy that’s inherited from the father and the rest are primarily expressed from the copy that’s inherited from the mother.  We’re really interested in imprinting, because it is an example of epigenetic regulation. For an imprinted gene, the two copies may have very similar sequence or even identical sequence for some of the things we look at, yet, they’re expressed differently. Even though they’re in the same cells, they’re in the same nucleus, the copy on the maternally inherited chromosome is not expressed in the same way as the copy on the paternally inherited chromosome.

Girard: In the plant that you’re studying, Arabidopsis, which genes coming from the male parent are imprinted?  Are these different from the genes coming from the female that are imprinted?

Gehring: What we’ve found in Arabidopsis, and this seems to be true in some other plant species as well, is that many of the genes that are paternally expressed—imprinted genes, so they are only expressed in the allele from dad—they’re enriched for transcription factors and for chromatin-modifying enzymes. And so we’re really excited about those genes, and we’re working some more on some of those genes to understand what are the functions of these chromatin modifiers that are imprinted.  Are they themselves important in regulating imprint expression?

For the genes that are maternally expressed—imprinted genes—really it’s a broader class of genes.  There’s a lot of interesting things in there. There’s genes involved in cell wall biosynthesis, there’s genes involved in hormone biosynthesis, transcription factors, and things like that. But there’s not one type of gene is imprinted.

Girard: What role does methylation—or the addition of methyl groups to certain parts of the genome—play in imprinting?

Gehring: Well, we’ve worked a lot on DNA methylation, in the context of imprinting and understanding whether or not differential DNA methylation between the maternal and paternal alleles is important for imprint expression. And it clearly is. So we know that a signature of imprinted genes, particularly the paternally expressed imprinted genes, is that the maternally inherited copy is less methylated than the paternally inherited copy. So we have a difference in gene expression and a difference in methylation. And that difference in methylation often is not happening right in the coding sequence of the gene, it’s happening upstream or downstream in 5’ or 3’ regulatory regions.

Girard: Do those differences in methylation actually affect whether or not a gene is imprinted?

Gehring: We’ve found a lot of correlations with methylation differences in imprinting, and now for a couple of genes, we’ve been actually able to show that this difference is causal for a difference in imprinting by changing the methylation state, by putting methylation someplace where it shouldn’t be and making a gene imprinted that is normally not imprinted, or vice versa.

Girard: Methylation isn’t the only way that gene expression can be controlled. Have you come across other mechanisms that affect the expression of imprinted genes?

Gehring: One thing that we’ve gotten interested in more recently is understanding how small non-coding RNAs might be involved in regulating imprinting expression. So we have evidence that this small RNA pathway, which is under paternal control, is actually being used to dampen expression of the maternally inherited genomes. And that’s something we’re quite excited about, trying to figure out exactly what the molecular mechanism of that is.

Girard: Can you tell us about an interesting example of an imprinted gene you’ve come across so far?

Gehring: So, in terms of thinking about how methylation patterns are inherited between generations, one thing we’ve found recently is that regulation of this DNA demethylase gene, ROS1, is crucial for maintaining methylation patterns across generations—so for maintaining fidelity and epigenetic inheritance. So we showed a few years ago that ROS1, which is an enzyme that removes methylcytosine from the genome, that it is itself regulated it’s expression is regulated by DNA methylation, so it has a methylated region 5’ of the gene, and that methylation actually promotes expression of ROS1, and demethylation reduces its expression. That’s the opposite of what you typically associate with methylation, although actually that’s also what we find for paternally expressed imprinted genes in the endosperm. I think the notion that methylation is only repressive is not correct, although it’s certainly true that if you have methylation right over the transcription start site, that can be repressive, but I think when there’s a little distance between the transcription start site and the methylation, we’ve certainly found both for ROS1 and for the paternally expressed imprinted genes that that’s associated with promoting transcription. Now we don’t know how, we’re trying to figure that out.

It does look like you really need ROS1 to be under tight control that’s tied to the methylation status of the genome in order to maintain the proper balance of methylation and demethylation activities. And if ROS1, its demethylase is just expressed, regardless of what’s going on with methylation at its 5’ region, and thus elsewhere in the genome, what you find is this epigenome that gets progressively worse across generations, at least in more gene-rich regions of the genome. We actually found that the genome is actually really plastic, and we found in heterochromatic regions, that whereas if you initially disrupt this ROS1 regulatory mechanism, you lose methylation at those regions, they can actually bounce back in later generations. There’s something else when you disrupt this mechanism at ROS1 over generational time, there’s something else that kicks in to try and restore the proper methylation patterning.


Girard: That was Mary Gehring, a Member of Whitehead Institute and an associate professor of biology at MIT. You can learn more about Whitehead science on our website at And you can listen to other AudioHelicase episodes on SoundCloud and iTunes. For Whitehead Institute, I’m Lisa Girard. Thanks for listening.


Produced by Nicole Giese Rura

Original music by Chocolat Billy. CC BY-NC-ND 4.0



Communications and Public Affairs
Phone: 617-452-4630