Tuning gene expression through antisense transcript dynamics

How is gene expression controlled? This most fundamental of questions in molecular biology has been intensively studied over many decades. The overwhelming paradigm has been that expression is regulated by transcription factors that bind to regulatory DNA to switch on or off associated genes. The realisation about 10 years ago that such processes can be significantly influenced by noise has augmented but not overturned this framework. However, recent experiments from the genomic era of massive sequencing have begun to reveal gaps in this paradigm. In particular these experiments have revealed that most genomes are pervasively transcribed, so that not only protein-coding genes but also many other regions of the genome are transcribed to produce so called non-coding RNA. As a result, DNA is often transcribed not only in the direction needed to make a protein (the sense direction) but also in the opposite (antisense) direction. The question then arises as to what all this extra transcription is doing: is it an accident, the inevitable by-product of the noisy cellular environment, or does it perform some regulatory function? Increasingly, the conclusion that it does have an important regulatory role is becoming accepted. However, the actual mechanistic role played by non-coding, often antisense, RNA is very unclear. The majority of previous studies have probed these questions at the level of a whole genome from which it is very difficult to draw conclusions about the regulation of specific genes. In this project, we propose to take a tightly focused look at antisense RNA regulation in the context of a plant flowering gene called FLC.

FLC is a repressor of flowering, and is a gene whose quantitative level of transcription is vital in ensuring that the plant flowers at an optimal time for reproductive success. Under normal conditions the precise level of transcription is believed to be controlled by two antisense non-coding RNAs. Here, we are seeking to understand how the differential production of these two antisense RNAs is able to tune the expression of the FLC gene. One possibility that we will attempt to prove (or disprove) is that each individual FLC gene makes only one of the two antisense transcripts for extended periods of time and switches randomly back and forth between one state and the other over time. The two different states are believed to have very different effects on the expression state of the FLC gene. By controlling the length of time the system spends in one or other of the (bistable) states, the sense FLC expression level can then be precisely tuned. We will test this hypothesis (and others) to reveal in unprecedented detail how non-coding RNA works as a quantitative regulator of expression. If we can get to the heart of this mechanism we will have moved closer to answering our starting question: how is gene expression controlled?

The role played by non-coding antisense RNAs (asRNAs) in controlling gene expression is poorly understood. Although much is known at a genomic level about non-coding RNA, mechanistic understanding of its role in the control of specific genes is still lacking. Our objective is therefore to dissect how non-coding asRNA controls quantitative expression of a specific gene, the floral repressor FLC in Arabidopsis.

In this system, quantitative control is believed to depend on the differential production of two alternatively polyadenylated antisense transcripts. We will therefore investigate how the choice of polyadenylation site can affect sense transcription. This will be achieved by an iterative combination of mathematical modelling and targeted experiments. Our working hypothesis which we will seek to confirm (or refute) is that the antisense transcripts act in opposing positive feedback loops to generate bistability at an individual FLC locus. Regulatory control is exerted by the fraction of time that a given locus spends in one of the two transcriptional states, with only relatively rare stochastic flips between them. Such positive feedback will depend critically on interactions between antisense transcripts and histone modifying enzymes, whose action may alter the local chromatin landscape, thereby reinforcing one or both of the antisense transcriptional states. This model, which will be simulated numerically to generate rigorous predictions, will be tested by genetic perturbation (e.g. deletion of key methyltransferases), measurement of histone modification/polymerase levels by Chromatin ImmunoPrecipitation, and also by monitoring transcript levels inside individual cells. The latter experiment will directly test the bistability hypothesis.

The end result of our work will be a focused, quantitative understanding of asRNA-mediated gene regulation that will serve as a paradigm for how antisense transcripts can mediate quantitative variation in gene expression.

The impact of this research will primarily be felt within the academic community, as detailed above. However, the fundamental nature of the problem being addressed, namely transcriptional regulation, ensures that the results of our work should have very wide impact across all of biology. This work will be in the vanguard of a new quantitative understanding of transcription, supplementing the traditional view of transcription factors with a much more sophisticated picture encompassing the role of non-coding, often antisense RNA. Moreover, the programme of work described here will study these questions in a multicellular organism rather than the standard yeast or bacterial unicellular model systems.

Eventually an improved understanding of transcriptional regulation can hardly fail to generate significant benefits for human health and biotechnology. However, we acknowledge that the time horizon for such enhanced understanding to percolate into new, economically important areas will be rather long.