RNA silencing and epigenetics: Rome, 15.11.2012 – Forum


David Baulcombe

2012 Balzan Prize for Epigenetics

For his fundamental contribution to the understanding of epigenetics and its role in cell and tissue development under normal and stressful conditions.

First, I should explain what I mean by “RNA silencing” and “epigenetics” because they are rather specialised terms and they may seem rather obscure even to the learned and well read attendees at an Interdisciplinary Forum of the Balzan Foundation. However, before I start, I should make a confession. It is that I did not start my research career with a grand vision. I knew that I wanted to work on genes and the regulation of their expression because I thought it was one of the most important topics in biology but I did not have a clear view as to how to proceed. I am here because of chance, lucky breaks and because I have a slightly belligerent streak that leads me away from the mainstream. You can already see that I am not a dedicated follower of sartorial fashion and the same applies with science fashion. I hope you find some interest in this description of my scientific ramblings in which I try to persuade you that our findings with plants raise important questions about the blurred distinction between nature and nurture in general biology.

RNA silencing

I will start with “RNA silencing” because it was my way into epigenetics. RNA silencing inactivates gene expression through processes acting at the level of RNA rather than DNA. My work on RNA silencing began when I had the good fortune in the 1980s to work with Mike Bevan who is one of the pioneers of transgenic plants. At that time my interests were in plant viruses and, for various reasons, transgenic plants expressing viral genes were a useful. Some of our experiments were designed to generate virus-resistant transgenic plants using a type of genetic immunisation strategy. The aim was to find out about the mechanism of virus replication and to produce genetically modified crops that would be protected against virus disease.

Like many early users of transgene technology in plants we observed that some transgenic lines expressed the transgene and others did not. This variation was expected. However there was a rather puzzling anomaly in that the virus-resistant lines were those in which the transgene was silent: if the transgene was expressed at a high level the lines were fully susceptible. This finding was unlike more straightforward situations in which the phenotype of a gene would be associated with its expression. To reconcile our observations with common sense we proposed that the virus resistance could be due to the same process that silenced the transgene.

The virus used in these experiments had an RNA genome. It replicated as RNA and it did not use DNA at all in its life cycle. Therefore, if the virus and transgene were suppressed by the same mechanism it followed that the silencing mechanism did not require a DNA target. It would operate at the RNA level. I described these data in a seminar in Basel in the mid 1990s when I referred to the effect as “gene silencing”. Ingo Potrykus, who later hit the headlines with his work on golden rice, pointed out that “gene silencing” was an inappropriate term. He suggested that “RNA silencing” would be a better. I have followed his suggestion ever since, as have others.

Other reports, appeared at about the same time, or shortly afterwards, indicating that RNA silencing also occurs in fungi and in animals including worms, flies and mammals. Most notably, Fire and Mello made the seminal discovery that RNA silencing could be initiated by double stranded RNA. They referred to their version of RNA silencing as RNA interference.

In my laboratory, we were interested in the specificity determinant rather than the initiator because the transgenic virus resistance due to RNA silencing was always virus-specific. A transgene based on potato virus X, for example, could only confer resistance against potato virus X or indeed some strains of potato virus X whereas a transgene based on potato virus Y was similarly specific to that virus. It seemed as if the transgenes conferred resistance in a manner that was specific to their nucleotide sequence – the order of the A, C G and T residues.

Andrew Hamilton, I and others I hypothesised that, to explain this sequence specificity, there had to be an RNA that would bind by Watson Crick base pairing to the target RNA of silencing. This antisense RNA would be produced either directly or indirectly from the transgene. Andrew’s first attempts to identify this molecule were not successful. However we realised eventually that the hypothetical RNA is much smaller than we had expected and that his methods were not suitable.

Andrew then adapted his assay methods and, in several different experimental systems, he identified antisense RNAs corresponding to the target of RNA silencing that were about 25 nucleotides long. He later further resolved these RNAs into species of 21, 22 and 24nt in length. Similar RNAs were later implicated in RNA silencing in animals and fungi. The common presence of these short RNAs in different systems confirmed that RNA interference of Fire and Mello, our work with plants and the experiments with other systems were all addressing a common process. The small RNAs were later named as small interfering (si)RNAs.

The details of RNA silencing emerged soon afterwards in what was one of the most rewarding periods of my career. My laboratory, using genetic screens in plants, provided information that was relevant to researchers in animals and vice versa. In just a few years the global silencing community established a core mechanism of RNA silencing in plants and animals that is the basis of our current work.

Two of the most important discoveries, from the Hannon group, were the Dicer and Slicer proteins. Dicer was able to generate siRNAs from a long double stranded RNA precursor and the Slicer protein is the effector of silencing that binds to the siRNA. The siRNA binds to target RNAs by Watson Crick base pairing and Slicer cleaves them so that, if they are a messenger RNA they cannot be translated into protein.

The evolutionary origins of RNA silencing.

Many types of eukaryote including animals, plants, fungi and protozoans have versions of RNA silencing. It is likely therefore that RNA silencing was active in a primitive unicellular eukaryotic cell. It may even have evolved earlier during a phase of evolution known as the RNA world that existed before DNA acquired its key role in the genetic code and inheritance. In that scenario the RNA silencing mechanism would have been lost in the evolutionary lineage leading to modern bacteria because they do not use RNA silencing or have the main proteins in the core silencing pathway.

Primitive cells in which RNA silencing first evolved would have been good hosts for viruses, selfish DNA and they would have needed defense systems. RNA silencing is an attractive candidate for such a primitive defense system because it can target the parasite without affecting the host. This property is achieved because the siRNA would be derived from the virus or selfish DNA and because the genome sequence of these parasites is distinct from the sequence of the host cell.

Modern plants and some animals including worms and insects have retained this defense role of RNA silencing. Plants that are mutant for Dicer or Slicer, for example, are more sensitive than the wild type plants to virus disease. They are also less able than wild type plants to protect themselves against a type of selfish DNA that is able to transpose from one site to another in the genome. However, RNA silencing has diversified in the different plant and animal lineages since the evolutionary divergence of animals and plants and it now has additional roles in different organisms. It regulates gene expression, for example, in many cells by targeting RNA degradation or by preventing its translation into protein. In some organisms it produces an RNA that moves between cells or even, in plants, from the shoot to the root. It may also influence recombination between parental genomes during the production of eggs and sperm and it has particular roles in genome regulation that are specific to some protozoans.


To illustrate the diversification of RNA silencing, and to introduce epigenetics, I would like to describe an experiment in which we modified viruses to carry parts of host genes. The aim of these experiments was to develop a system for analysis of gene function. The idea was based on our understanding, described above, that RNA silencing is part of a virus defense system. With an unmodified virus the siRNAs correspond to the viral genome and RNA silencing targets the virus rather than the host. However, if host sequence is introduced into the viral genome, the siRNAs now correspond to the virus and that host element. There is silencing of host RNA if the viral insert corresponds to a gene and the symptoms on the infected plant resemble the appearance of a mutant in that gene. This system and its variations have been useful for cataloguing the function of many genes in plants and animals because it is relatively easy to carry out a survey of many thousands of genes.

In some experiments the virus-induced silencing of the host element persisted only for as long as the virus in the infected plant. However, in other instances, the silencing effect persisted for several generations. The virus was not transmitted between generations and the DNA sequence of the target gene was unaltered. These characteristics – persistence of effect through cell division and without genetic change – conform precisely to the definition of an epigenetic effect.

Other well known examples of epigenetic effects include vernalisation in plants and X chromosome inactivation in female mammals. Vernalisation occurs when young winter annual plants experience cold. The cold treatment results in epigenetic silencing of a floral repressor gene that normally prevents flowering. However, once that gene is epigenetically silenced, the plants are competent to respond to light and they flower when the day length has lengthened in the following spring or summer.

X chromosome inactivation in mammals silences one of the two X chromosomes in female mammals. Once silenced this chromosome remains inactive for the lifetime of the plant although its DNA sequence does not change and it can be reactivated in the next generation progeny. X chromosome inactivation is important as a means of dosage compensation. It ensures that the X chromosome produces the same amount of protein in females with two copies and males with only one.

A hallmark of epigenetics is that there are separate establishment and maintenance phases. Establishment in the three examples given above is due, respectively, to the viral RNA, an effect of cold and an ill defined trigger associated with early development. Maintenance is similarly diverse and it can involve chemical modification of the DNA or of proteins associated with DNA. The RNA-directed silencing described above, for example, is dependent on methylation of DNA at the silenced locus and our findings follow from the seminal discovery of Michael Wassenneger that RNA can direct DNA methylation in plants. This separation of establishment and maintenance is important because it ensures that continued presence of the initiator is not required for persistence of the epigenetic effect.

DNA methylation is effective as a maintenance mechanism of RNA silencing for two reasons. First it affects the proteins bound to the DNA and prevents the DNA from being transcribed into RNA. The second reason is because the pattern of DNA replication is stable from one generation to the next in the absence of the initiator RNA. The enzyme that adds methyl groups to the C residues of DNA can replicate the pattern of DNA methylation from the parental strands of the DNA onto the daughters.

Much of our work in recent years has been aimed at unravelling the mechanism of RNA-directed epigenetic modification in plants and some of the progress involves details that will not be of interest in a general forum. However, I would like to refer briefly to four features of this molecular stamp collecting because they are relevant to later points.

First we have discovered that two atypical forms of the RNA polymerase are involved in epigenetic RNA silencing. One of these is thought to produce the precursor of siRNAs. The other, from findings of Pikaard and colleagues, produces scaffold RNAs that are attached to their DNA template. These scaffold RNAs can be targeted directly by the siRNAs that are bound to Slicer proteins. The Slicer protein is then able to recruit the enzymes that introduce DNA methylation marks associated with epigenetic changes to the target region of the chromosome. This aspect of the mechanism is important because it explains how RNA can find a target in double stranded DNA.

A third discovery, by us and others, is of endogenous siRNAs with the potential to influence epigenetic features of the genome. Epigenetic RNA silencing, therefore, has effects that go beyond virus resistance. It is a constant and extensive background to the behaviour of the genome. A surprising fourth element of the silencing machinery is mobility between cells. The epigenetic siRNAs spread through the plant and mediate epigenetic modification of the genome in the recipient cells.

The understanding of RNA silencing mechanisms allows us to investigate the biological role of the RNA-directed epigenetic changes in plants. We are particularly interested, for example, to find out whether viruses and stress can induce siRNAs that then lead to epigenetic changes. Such a process could explain how plants can become acclimatised to different environments and some of our current projects are addressing that possibility.

In our recent work we have become particularly interested in epigenetics and the effects of hybridisation between different species or varieties of plant. I hypothesised that siRNA from the genome of one parent of the hybrid could find novel targets in the second parental genome and induce epigenetic marks that are absent from the parents. These novel epigenetic marks could then be inherited between generations. These hybrid-specific marks could affect gene expression if they are in or adjacent to genes and they would cause the heritable phenotype of hybrids to be outside the range of the parents. Such phenotypes are well known to plant breeders and they are referred to as transgressive. Our recent work in tomato is consistent with the proposed involvement of siRNA in silencing and it is taking us into new areas that may be relevant to evolution and to methods for crop plant improvement.

The findings are relevant to evolution because they imply a new consequence of hybridisation beyond the creation of gene combinations that do not exist in the parents. It would also be a process that generates new heritable variation. Following from this point, because modern organisms are the progeny of past hybridisation events, it follows that some of the heritable variation between varieties or species will be epigenetic rather than genetic.

There is, therefore, the potential for organisms to use hybrid-induced epigenetic variation to experiment with the effects of silencing a gene or set of genes. An epigenetic mutation conferring a fitness benefit would be selected for and eventually stabilized by genetic mutation. Conversely it would be selected against if it has a deleterious effect or it would revert to the original unmodified epigenetic condition.

My laboratory is testing the importance of hybrid-induced and other forms of epigenetic variation in evolution. To initiate this study we will first explore the extent to which heritable phenotypic variation is due to epigenetics. We are using seed plants including Arabidopsis and tomato. A green alga – Chlamydomonas is also a useful system because it has a rapid life cycle in liquid culture. We are planning to isolate mutants in the RNA silencing and epigenetic pathways so that we can explore their potential of these mechanisms to influence adaptation to modified environments. Plant breeders may also be interested in heritable epigenetic variation affecting key traits. They would have the opportunity to produce new varieties by selection for epigenetic as well as genetic markers.

Another practical application of RNA silencing and epigenetics is based on our discovery that epigenetic siRNAs are mobile in the plant. We are using this discovery to develop a novel strategy for crop improvement based on epigenetic rather than genetic modification.

The strategy requires first that a target gene is identified that has an adverse effect on a crop. It could be a gene affecting, for example, storage or processing quality of the crop. The next step is to produce a plant with epigenetic siRNA corresponding to this gene. The plant could be infected with a suitably modified virus or it could be a GM variety and it would be grafted as a stock onto the unmodified crop as a scion using methods that are standard horticultural practise.

Once the graft union is established the epigenetic siRNA should spread across the graft union and mediate epigenetic silencing of the target gene in the crop. The silencing would be transient if the mobile siRNA were a standard genetic regulator. However, as this siRNA mediates epigenetic changes in the recipient tissue we predict that the silencing effect would persist in these cuttings and their progeny.

These plants would be epigenetically rather than genetically modified and I am interested to find out how well such plants perform in the field and whether they would require regulatory approval before they could be used as crops. If successful this strategy could provide a useful additional tool for plant breeders and biotechnologists. It would have an advantage over genetic modification or conventional breeding in that the improvement could be introduced into multiple varieties with relative ease. It would be possible therefore to retain or even increase the diversity of varieties used in agriculture.

Epigenetics and animals including humans

I am a Professor of Botany and so most of the examples given above are with plants. However there are indications that some of these plant findings are relevant to animals including humans. There are, for example, reports of environmentally induced and heritable epigenetic effects in humans. Study of a population in the north of Sweden indicates that body mass index or mortality risk ratio correlates with smoking or diet in the grandparents. There are also indications in animals that RNA silencing and epigenetics are connected in germline and other cells through the action of small RNAs known as piRNAs.

I do not expect that there will be an exact parallel of epigenetic phenomena in plants and animals because the underlying molecular mechanisms are likely to differ to some extent. However it may well be that the human condition could be improved if we had answers to the following questions that are similar to those that we are addressing in our plant work:
· to what extent does the environment cause the epigenetic marks to change during animal development. Do stresses or chemicals induce epigenetic changes in adult or somatic stem cells that could have long lasting effects in behaviour or disease, including cancer. Such somatic epigenetic changes would not be transmitted to the next generation but they nevertheless could have a profound effect on the health or well being of the organism.
 · to what extent is heritable phenotypic variation between individuals determined by genetic and epigenetic factors? It has been difficult to link many characteristics including behavioural differences to genetic factors and so heritable epigenetic variation should be assessed in large scale studies of that are at present used to assess the association of phenotype with genetic marks.

Both questions address the relationship between the nurture of an organism and, through stable epigenetic marks, its nature. In some instances this could be an epigenetic nature that is inherited from one generation to the next. In other examples the epigenetic effect, like vernalisation and X chromosome inactivation, could be reset during gamete formation of each generation. I hope that our work with plants might provide a guide at least to the research that will address these important questions.

Insert your email address here to sign up for our newsletter with all the latest updates on the International Balzan Prize Foundation .

By subscribing, I confirm that I have read and understand the General Terms and Conditions of EU Regulation 2016/679, in particular with regard to my rights, and hereby consent to the processing of my personal in the manner and for the purposes indicated therein.
Fondazione Internazionale Premio Balzan