Denise Barlow: A career in epigenetics

Could you characterize your career in a single sentence?

The drive to see a biological problem with my own eyes. Sometimes there is a general view about a biological problem and people approach it from a general viewpoint. But they don't look at the problem itself. They look at what people think. I think you get much further if you look at the problem with your own eyes.

You stuck to this rule several times in your career and it paid off, didn't it?

In the imprinting field, any time I did something, it didn't agree with what people expected. For example, when we were looking at how DNA methylation modifies an imprinted gene, to the shocking surprise of many, we found that DNA methylation is not on the silent allele but on the active allele. And it isn't sitting in the promoter but it's sitting in the middle of the gene, in an intron. And this was completely contrary to expectations. I would present the data at meetings and people would come to me, red-faced, and say: "How can you say that methylation is on the active allele?!”

About Dr. Barlow Denise Barlow was born in Yorkshire, UK in 1950. At first she was trained to become a nurse but then she entered science with a degree in Zoology with Physiology & Biochemistry from Reading University (1978) and PhD from Warwick Univeristy (1981). She performed her post-doctoral research with Brigid Hogan at ICRF Mill Hill Laboratories, London (1981-5), and with Hans Lehrach at EMBL Heidelberg, Germany (1985-8). She started her own lab as a Junior Group Leader at the newly opened Institute of Molecular Pathology (IMP) in Vienna, Austria (1988-95). After a brief interlude at The Netherlands Cancer Institute, Amsterdam (1996-2000), she returned to Austria. Denise initially joined the Institute of Molecular Biology of the Austrian Academy of Sciences in Salzburg, and upon its closure in 2003 she joined the Center for Molecular Medicine of the Austrian Academy of Sciences in Vienna, until her retirement in February 2015.She is a recipient of numerous awards and honours including honorary professorship of genetics at University of Vienna (2003), EMBL Alumni Association Austrian Chapter Achievement Award Medal (2014) and Erwin Schrödinger Award (2014). She has been an EMBO member since 1995.Career HighlightsAll mammals inherit a complete set of genes from both the maternal and paternal parent. However, each parent can leave a mark or ‘imprint’ on about 200 of our 23,000 genes, which tells that gene to be silent just on one of the parental chromosomes. These genes are called imprinted genes and although it is still largely a mystery why they exist only in mammals and flowering plants, they have provided a large amount of important information about how genes are silenced by epigenetic mechanisms.Denise Barlow has made major discoveries in the imprinting field, including:•The identification and characterization of the first imprinted gene Igf2r (Insulin-like growth factor type 2 receptor), which is a ‘scavenger’ receptor for the Igf2 growth hormone and an embryonic growth suppressor that opposes the effect of Igf2r.^1-3•The first identification of a gametic DNA methylation imprint combined with the suggestion that it acts by “repressing a repressor.”⁴•The discovery of a lncRNA (Airn—Antisense Igf2r RNA non-coding) together with the first identification of an imprint control element that controls Airn expression.^5-7•The demonstration that Airn represses Igf2r by transcriptional interference and that it is able to regulate an entire cluster of imprinted genes.^8,9•The first proposal that genomic imprinting could originate from a methylation-based host defense response present in one gamete only.¹⁰

More on this later. Let's start at the beginning. Can you recall your student years? What or who was it that made you go into science and molecular biology?

I had a 2-step career - my school did not prepare its students for University. So, I did a pre-nursing course aged 16-18 y at a local Technical college and then from 18 - 22 y of age, I trained and worked as a State Registered Nurse (SRN) in the UK. It was during my nursing training that I started to ask questions about human physiology and disease that could not be answered by my nurse tutors and I felt that I had to know more. It was this feeling "that I had to know more" that continually drove me in my life. I thought if I went to University this would give me the ability and freedom to do anything.

However, I did not have the qualifications to enter University so, I got a grant to go to another Technical college to take A-levels (from 22-24 y of age). Then I went to Reading University UK for 3 y for a BSc Honors course (aged 25-28). I did not find my University course (Zoology with Physiology & Biochemistry) very interesting until the 3rd year when I started to read research papers. I remember being most excited by developmental biology questions. It was only late in my 3^rd year that I considered staying on in research. I found a PhD position at Warwick University supervised by Derek Burke one of the pioneers of the interferon field (my project was to investigate when mouse embryo developed an interferon response to virus infections and I could show it happens rather late in development).

I was still excited by developmental biology particularly in mammals and was lucky enough to be invited to join the lab of Brigid Hogan who was one of the top developmental biologists in the UK at the ICRF in London. We worked on isolating genes expressed in the very early stages of embryonic development -notably this was a time before the mouse genome was sequenced, before PCR, and before many modern molecular biology techniques - but we succeeded in the end to isolate these genes. However, I was frustrated by the then inability to identify what would be the most important genes for embryonic development. For this reason, I was interested in mouse mutants that had been characterized as having a defect in development that mapped to a single gene. However, the problem was there were currently no techniques to isolate the defective gene. Until I heard about Hans Lehrach.

Later you spent some time in his lab amidst the rise of genomics. Can you recount those days?

I met Hans Lehrach at a CSH meeting where he explained that he was developing many wonderful molecular genetics techniques that would allow you to quickly isolate any gene in the mouse genome. These were very exciting days. Many techniques were being developed in Hans's lab. My particular favorite was 'pulsed field gene electrophoresis' that allowed you to separate very, very, large chromosome fragments on a gel - and this is what I did so that I could isolate a mutant gene that had been mapped to a specific chromosome region in the mouse. However, it turned out that chromosomes were much larger than we expected and the mutant genes we were seeking were not so easily isolated. Han's lab was at EMBL Heidelberg, which was a fantastic environment for any young scientist - there were amazing seminars from guest speakers and a very interactive atmosphere inside and outside the lab. It was also my first time working outside the UK and I liked it very much - so that eventually I decided not to return to the UK.

You were searching for genes that were behind certain diseases. How was this done in a pre-genomic era?

Mapping, mapping and more mapping. Basically we had to find a piece of DNA that was close to our mutant gene. Before the mouse genome was sequenced this was done using genetic maps that measured recombination frequencies between 2 other genes. This required a lot of mouse breeding and careful analysis by mouse geneticists - none of this was done by me. Luckily, the gene I was interested in was shown by other scientists to lie in a small genomic deletion - and this was the best system to have.  This gene was known as a maternal effect gene (short form is Tme) and it was assumed to be an imprinted gene - i.e., expressed only from a maternally-inherited chromosome. However, no imprinted genes had yet been isolated so we were not sure how to look for it. When I first starting gene hunting for Tme - I decided to use my 'pulsed field gene electrophoresis' approach to map the distance between genes that were shown by others to lie outside this small genomic deletion. This as I would know how much DNA I had to search for my gene of interest.

However, this was tedious and I was not making good progress. Then I happened to be at EMBL, teaching on a course, and went to the library (much more complete than ours at the relatively new IMP) and I looked up the latest issue of a journal reporting on human chromosome maps. I knew that human chromosome 6 looked very similar to the mouse chromosome that contained the Tme gene that I was interested in. And I found that 4 new genes had been mapped to this human chromosome and I also found by reading the journals that other scientists had isolated the mouse equivalent gene. I had to wait until I got back to Vienna, then I wrote a letter (yes a letter!) to ask these scientists if they would send me a fragment of their mouse genes. All of them were very nice and agreed to do this. Then I had to test if any of these 4 new genes were inside the small genomic deletion containing my deleted gene. Surprisingly and very, very luckily for me - it turned out that 3 of these genes were, and, one of them was the gene I had sought for (called Igf2r).

How did you find Igf2r is imprinted?

In 1990 I was doing Northern blots using Tme crosses. I had embryos, in which only the maternal chromosome was intact and the paternal one had a deletion in this region and it would be the other way around in other embryos. I did Northern blots with RNA isolated from these embryos and I would ask "is it expressed in one embryo and absent in the other?" I had a collection of genes in that locus and each time I would probe for them, I would find they were expressed from both chromosomes at half the level of a diploid wild-type. I stripped the Northern blot 6 times and the seventh time I used it, I probed for Igf2r and I could not see any signal from the paternal chromosome. But the blot was very faint, everyone at the institute was skeptical, and I had to expose it for 10 days, until it was clearly visible (Fig. 1).

Figure 1.

The original Northern blot showing imprinted expression of Igf2r. The samples were prepared from mice with Igf2r locus deleted on the maternal chromosome (lanes 1, 2) and their wildtype littermates (lanes 3, 4), or from mice with Igf2r locus deleted on the paternal chromosome (lane 5) and their wildtype littermates (lane 6). The results of this experiment were published in ref. 1.

Your lab later discovered the way Igf2r is regulated by lncRNA Airn. How did you hit upon the idea Igf2r might be regulated by transcription on the opposite DNA strand?

Once we knew that Igf2r is an imprinted gene and only expressed from the maternal chromosome - we knew we must compare the maternal and paternal copy of this gene. A key feature of imprinted genes is that they have the same DNA sequence on both parental chromosomes, so we knew we had to look for extra information added on top of the DNA sequence. This type of information is known as 'epigenetic'. In 1991, just after we identified Igf2r, the only type of epigenetic modification was DNA methylation. So a PhD student (Reinhard Stöger) looked for any methylation marks that were different on the maternal and paternal chromosome. We found 2 regions that had a differential mark. However, only one had all the expected features of something that would control maternal-specific expression of Igf2r. Unexpectedly, this mark was not on the silent paternal copy of Igf2r, but was on the active maternal copy. In addition, this key methylation mark was siting inside the Igf2r gene in intron 2. This made us think that perhaps DNA methylation was being used to express the maternal copy of Igf2r. This interpretation went against all the then current thinking about how DNA methylation should regulate genes - but it was supported by experiments from other scientists who inactivated the mouse methylation system and also by other scientists who examined other imprinted genes.

With this discovery of a key methylation mark (now known as gametic differential methylated regions or gDMRs), we were a bit stuck. We could show it was important by deleting it from a mouse - but we did not understand how it could work. Then we had another very lucky accidental finding. An Austrian Diploma student (Walter Lerchner) in the lab did an experiment that I initially had told him not to do - and found a strange RNA being made in intron 2 of Igf2r. Another PhD student (Anton Wutz) was able to show that this RNA was only made from the paternal chromosome. However, it took a postdoc (Robert Lyle) a few years before we could understand that this was not an artifact but a strange, and very long non-coding RNA. This lncRNA is now named Airn (Antisense Igf2r RNA Non-coding). We were very surprised that it was so long - up until the time of its discovery very few non-coding RNAs were known and none were very long. However, now we know of several examples of lncRNA that are very long.

Airn is 118 kb and yet your work has shown that its function lies with transcriptional interference at the Igf2r promoter. Why does nature go to such length as to transcribe huge chunks of genome that apparently go to waste?

This is going to be one of the puzzles for the next generation of scientists. Work from a PhD student (Paulina Latos) and postdoc (Florian Pauler) has shown that Airn needs to be only 32 kb long to transcriptionally interfere and silence the Igf2r promoter. So why is it 118 kb long? The answer may lie in the fact that the Airn lncRNA can silence some genes that lay a few hundred kilobasepairs away, but only in a tissue-specific manner. A postdoc in the lab (Quanah Hudson) is currently testing if the extra length of Airn is needed to silence these distant genes. However, it may turn out that it is less cost to transcribe a lncRNA than it is to specify a particular length. We do not know yet.

What's your favourite hypothesis about the evolutionary origins of imprinting?

Ah! Difficult. And I do not want to give one because I think all the data is not yet in. We think that imprinted expression only occurs in placental and marsupial mammals and is even absent from egg laying mammals. This seems acceptable, although not much work has been done analyzing imprinted expression in egg laying mammals. I often read reviews claiming that imprinted expression in mammals is important for the embryo or for the placenta. I also read some reviews claiming it is important for maternal brain function. However, there are 2 problems that confound this viewpoint. Firstly, we did not initially appreciate that imprinted expression can arise in a tissue-specific manner. In the beginning scientists only looked at mid-gestation embryos and sometimes at the placenta. But no one has yet systematically looked for the presence of imprinted gene expression in all adult tissues or in different developmental stages. So we do not know the full list of imprinted genes and we also do not know which tissues show most imprinted expression. Secondly, we did not appreciate that the beginning, that most imprinted genes occur in small clusters containing from 2–10 genes. We do not know yet, if all of the genes in one cluster were equally selected to show imprinted expression or if some can be described as "innocent bystanders." So I will reserve my opinion on the evolutionary purpose of imprinted gene expression until we have more information on these very interesting questions.

Can you guess what the primary signal for parental imprinting could be?

My guess is that imprinted genes arose from an existing host defense response present in oocytes and sperm that is directed toward invading viruses, which today comprise about 50% of the human genome. Today we know that a short DNA sequence, known as the imprint control element, controls imprinted expression. The idea is that a long time ago at the start of mammalian evolution, a viral like element jumped close to a gene and it could exert a silencing effect in cis (i.e., on the same chromosome only). However, thanks to the existing host defense response, this viral element was methylated in one of the gametes. Thus it was only active as a silencer in the other gamete. There is evidence to support this hypothesis (that I first proposed in 1993) that comes form the demonstration that one of the enzymes that methylate an imprint control element in one gamete - acts to methylated retroviruses in the other gamete. This mechanism whereby the host defense response is hijacked to induce imprinted gene expression is analogous to the behavior of plant transposons as described by Barbara McClintock. She could show that transposons silenced a gene they landed close to, however, if the transposon was silenced by DNA methylation - it lost the ability to silence in cis.

How would you explain epigenetics to a lay person?

I once explained it as “Epigenetics has always been all the weird and wonderful things that can't be explained by genetics.” Because this is how the field was in its early days - many incredible biological phenomena that appeared to be inherited in a non-genetic manner.

How do you envision the future of the imprinting field?

I think the imprinting field has been a bit diverted in recent years concentrating on chromatin modifications such as histone marks, which are not epigenetic as they are not inherited through cell division and, although some are associated with silent genes, none have been shown to cause gene silencing. The role of imprinted lncRNAs in inducing imprinted gene silencing is, in my opinion, the future of the field. This together with the discovery that lncRNAs are widespread in the genome, most interestingly, shows that epigenetic silencing of imprinted genes may also be relevant for non-imprinted genes.

You are known as a pioneer of gender equality in science. How has the position of women changed over the years?

It has changed a little but not much. I think that now one can identify slightly more women at the excellence level in science. But we will not see significant change until there are more women in decision roles in recruitment, in organizing meetings, in directing the future of science.

How much can one plan a career, how much is chance and predictable?

I think you have to plan - you need this basic idea and question to drive you forward. But, you have to have luck. First and foremost you need luck to have good students and Postdocs in your lab and good luck to be helped and mentored and supported at critical times in your career by more senior PIs. Then you need luck not only in fortuitous discoveries (Louis Pasteur said “Chance favors only the prepared mind”), but also in opportunities being there when you are ready for them. Most of the time in your career, you get the job you want because you are in the right place at the right time.

If you had a chance to rewind the tape of your career, would you change anything?

No. There are many things I could change, including some things that could have made my scientific life easier. But if I could change these things I would not be able to ask questions in the same rigorous way and I would not be the person that I am today.

Disclosure of Potential Conflicts of Interest

No potential conflicts of interest were disclosed.

Funding

Recent work in the Barlow group was supported by the Austrian Science Fund [FWF F43-B09, FWF W1207-B09].