At A Glance
Name: Paul Riley
Position: Senior lecturer, molecular medicine unit, Institute of Child Health, University College London
Background: BSc, zoology, University of Leeds — 1990; PhD, molecular biology/endocrinology, University of London — 1995; Postdoc, Samuel Lunenfeld Research Institute, Mount Sinai Hospital, Toronto — 1996-1999; Wellcome Trust Fellow, Institute of Molecular Medicine, Oxford University — 1999
At the Institute of Child Health in London, Paul Riley investigates the genes and mechanisms that control cardiac development. While his work does not directly relate to RNAi, the gene-silencing technology has found a significant place in his lab.
Recently, he and his colleagues published a paper in Biological Procedures Online detailing an assay system for the quantification of siRNA efficacy and specificity that they developed to deal with RNAi sequence specificity issues they encountered in their research.
He recently spoke with RNAi News about his work.
Could you give an overview of what you do at the ICH?
We’re interested in molecular processes that underlie normal heart development with a view toward finding out what goes awry in the disease situation. We’re taking an animal-model system approach, mainly working with mice — knocking out genes, with a focus on cardiac transcription factors and their target genes, and trying to get an idea of the underlying events essential for vertebrate cardiac morphogenesis.
Our focus is primarily on the basic helix-loop-helix transcription factor called Hand1, previously known as Hxt, eHand or Thing1, which I’ve been working on for a number of years dating back to my time in Toronto [at the Samuel Lunenfeld Research Institute.]
Can you explain what Hand1 is involved in?
The gene was actually knocked out whilst I was a postdoc in Jay Cross’ lab in Toronto. We were coming at it from an angle of looking at its role in placentation since it is expressed at quite high levels in the trophoblast of the developing placenta. It turned out that, indeed, there was a placentation defect in Hand1-null embryos with a failure in differentiation of so-called trophoblast giant cells. These cells are responsible for creating an appropriate implantation chamber for the developing mouse embryo and establishing the maternal-fetal interface.
But the gene was also expressed at high levels in the heart during development, so we employed a technique called tetraploid rescue to rescue the placentation defect and look at the role of the gene in the embryo proper. What we found was that the embryos were rescued partially but they still died from heart failure. Specifically, the rescued embryos had a failure in cardiac looping morphogenesis, an essential and highly conserved process in vertebrates responsible for appropriate orientation of the chambers of the heart and their alignment with the great vessels. Basically, in the absence of Hand1 we observed defects in the looping process, leading to embryonic lethality at around mid-gestation.
The question still remained however, as to what is the precise function of Hand1 in the developing heart? This is the major research question we’re trying to address now. The simple fact that Hand1 mutant embryos died of a failure in looping morphogenesis doesn’t really give us a true insight into a specific molecular or cellular role for the gene in the developing heart per se.
What sort of gene knockout technologies are you using now?
One way to get at what Hand1 is doing during cardiac morphogenesis is to identify and functionally characterise Hand1 target genes. To initially screen for Hand1 downstream targets I carried out an RDA or representational difference analysis screen. For want of a better description it’s a poor man’s microarray, established sometime before microarray technology was widely available.
RDA is essentially a PCR-based enrichment of differentially expressed transcripts between two pools of RNA. I carried out the RDA on Hand1-null embryoid bodies versus wild-type embryoid bodies. Embryoid bodies are derived from in vitro differentiated embryonic stem cells, which in culture can form essentially all of the cell types of the developing embryo. What we were able to do was separately in vitro differentiate wild type and Hand1-null ES cells, extract RNA from each genotype pool, make cDNA and following successive rounds of subtraction and PCR amplification enrich for genes present in wild type but absent or present at low level in the mutant population. Thus we were able to enrich for genes down-regulated in the absence of Hand1. By reversing the RDA we were also able to enrich for genes up-regulated in the absence of Hand1 and consequently sequence and identify potential targets for Hand1.
We’ve subsequently been focusing on a number of target genes that came out of the RDA screen, following their verification by standard techniques such as real-time PCR and northern blot analysis.
When did RNAi start coming into play?
Interestingly enough, one of the targets that came out of the screen was a gene called Thymosin beta4, and we’ve published on this in a Mechanisms of Development paper. Essentially, Thymosin beta4 is involved in actin-based cell motility, regulating actin polymerization, and lamellipodia formation, which are essential processes for cell movement and directional motility. We’ve had a long standing hypothesis that Hand1 might be involved in establishing morphogenetic cues via regulating cell movement in the developing heart. So any genes that are downstream of Hand1 putatively involved in a cell movement pathway are of interest.
Thymosin beta4, we figured, would be quite a difficult gene to target by conventional homologous recombination in ES cells for a couple of reasons. Firstly, it maps to the X chromosome, and since people generally use male-derived, XY embryonic stem cells for gene targeting, to ensure a higher rate of germline transmission, mapping to the X means there is only a single copy of the gene so any targeting event would automatically create a functional null. Secondly, from some in vitro studies we had reason to believe that Thymosin beta4 might be essential for cell division or cytokinesis and consequently cell survival. Therefore, in attempting to create a null ES cell line we would be unable to appropriately selective the targeted cells since they would be unlikely to survive.
At around this time we had learned about a very interesting study carried out by some of my old colleagues at the Samuel Lunenfeld in Toronto, namely Tilo Kunath and Gerry Gish in Janet Rossant and Tony Pawson’s groups, respectively. They had established a means of in vivo gene targeting or knock down using RNAi. They knocked down RasGAP, a cell signaling molecule for which they already knew the conventional knockout phenotype, taking an in vivo RNAi approach. Rather than going down a germ-line transmission route, they made chimeras with RNAi, targeted ES cells, and then compared the embryonic phenotype following knock down with that of the published RasGAP knockout. Not only did they successfully phenocopy the conventional knock out but they also discovered that different copy number of the RNAi transgene could lead to different levels of knock-down resulting in a hypomorphic series of phenotypes; examination of which is a great way to tease out more subtle gene function. Following the publication of this study a couple of years ago in Nature Cell Biology, I contacted the authors and they kindly supplied us with the RNAi vectors to enable us to adopt an analogous in vivo approach with RNAi for Thymosin beta4. We have further extended the RNAi design, making it conditional so that we can restrict knock down of Thymosin b4 to the developing heart.
Is that ongoing now, or do you have some indication already about its role?
Establishing the in vivo model is ongoing. We have carried out in vitro work to assess the efficacy and specificity of shRNAs directed against Thymosin beta4 and these studies led to the development of the reporter assay detailed in Biological Procedures Online earlier this year.
From the outset there were issues with specificity of RNAi here because Thymosin beta4 has a very close homolog called Thymosin beta10. These b-thymosins are very small genes with significant sequence overlap, so it was quite difficult to work out if we were knocking down one versus the other with conventional techniques such as RT-PCR, Northerns, and Westerns due to substantial cross reactivity. Therefore, we couldn’t really work out just how specific the shRNAs we had selected were, and consequently had to develop an alternative assay to address the specificity question.
Can you give an overview of the assay system?
Essentially, what we decided to do was take the target sequence within Thymosin beta4 against which we had designed the shRNA and transfer that target sequence into a reporter system and assay how efficiently the shRNA could knock down reporter activity. We were able to insert target sequences for either Thymosin beta4 or Thymosin beta10 at the 3’ end of the luciferase gene between the polyA and the actual luciferase cassette itself.
Transfection into cells of the target-sequence reporter alone results in high luciferase activity due to the strong upstream SV40 promoter in the vector. Co-transfection of the corresponding shRNA to the target sequence, should result in knockdown of luciferase activity. Different shRNAs can then be tested against their corresponding target sequence for optimal efficacy of knockdown and also for specificity as to whether say a Thymosin beta4 shRNA can non-specifically knock down a Thymosin beta10 target reporter or vice versa.
As I’m sure you’re aware, there have been many reports which comment on the specificity of RNAi in a number of different systems and claim a single nucleotide change as sufficient to disrupt RNAi specificity. We actually found this not to be the case in our system. In fact we observed significant knock down of a Thymosin beta10 target reporter with one particular Thymosin beta4 shRNA where the number of mismatches was 12 out of 23. This highlighted the need to consider so-called off-target effects as a very obvious caveat to RNAi experimental design.
We thought it was worthwhile publishing this RNAi reporter assay system in some form to bring to people’s attention that specificity at the single-nucleotide level is not always the case and that consequently use of an assay system to check RNAi sequences should be seen as an essential pre-requisite for planned RNAi experiments.
We had a few problems convincing people that this was a significant enough issue to publish. We received mixed comments at review such as ‘this may not be of sufficient widespread interest’ and ‘that having to clone in each target sequence into the reporter vectors could be time consuming’. I don’t really feel a need to respond to the first comment but on the cloning front; since annealed oligos can be inserted into a single directional restriction site the cloning is actually quite trivial and rapid. Clearly we acknowledge the assay is limited to focus on a small number of genes in any one study and has limited value if you are taking a high throughput knockdown of multiple genes.
We always felt RNAi specificity is a big issue. The viewpoint that specificity down to the single nucleotide level applies across the board is clearly not the case. All we’re really saying is that one should take the time to check specificity first, and it’s relatively straightforward to do using an adaptation of a conventional reporter assay system. It was quite an unexpected find that such significant mismatch could actually lead to knockdown.