Name:
Christina Leslie
Position:
Assistant professor, computational biology, Memorial Sloan-Kettering Cancer Center
Background:
• Research scientist, Columbia University — 2004-2007
• Assistant professor, computer science, Columbia University — 2000-2003
• Postdoc, Columbia University — 1999-2000
• PhD, mathematics, University of California, Berkeley — 1998
• BMath, University of Waterloo, Canada — 1992
This week, investigators published in the online version of Nature Biotechnology data from statistical analyses demonstrating that the transfection of small RNAs into cells disrupts gene regulation by endogenous microRNAs.
According to the team, the results support the hypothesis that small RNAs such as siRNAs introduced into a cell compete with endogenous microRNAs for RISC and related intracellular machinery.
This week, RNAi News spoke with Memorial Sloan-Kettering Cancer Center researcher Christina Leslie, who is a co-senior author of the paper, about the findings.
Let's start with an overview of your lab.
I run a computational systems biology lab, so what we do is apply computational approaches, especially machine-learning methods, to look at high-throughput data. Generally, we're interested in understanding the regulation of gene expression, including transcriptional regulation, and now microRNA-mediated regulation.
How did the work looking at the effect of small RNA transfection on endogenous microRNAs come about?
When Lee Lim did the first genome-wide study of miRNA regulation in 2005 — he transfected miRNAs into HeLa cells and measured expression changes with microarrays — my collaborator, [Harvard Medical School's] Debbie Marks … looked at the data and saw that many genes were in fact up-regulated. She had a hunch that this up-regulation might be [due to] a competition effect [for the intracellular small RNA-processing machinery].
We were looking at a lot of transfection data because we've been developing a microRNA target-prediction method that is based on supervised learning — learning from the transfection data, learning from the mRNA-level response to be able to predict the efficiency of the [target] site. We heard about [Marks'] hypothesis and were in the position to try to get statistical support that it was happening.
We knew statistically how to look for evidence and found it; then we started collecting more and more data because we knew that [this effect] would be important to know about.
[ pagebreak ]
Can you give an overview of the experiments you did?
When you put an siRNA or microRNA into a cell, you expect that dozens or hundreds of genes are down-regulated, but some are up-regulated. That effect was dismissed as a secondary effect, so we wanted to see [if] targets of endogenous microRNAs are part of this up-regulation.
The hypothesis is that if you put a lot of exogenous small RNAs into the cell, then the cell's own microRNAs are competing with the transfected small RNAs for protein machinery like RISC. If that's true and they are getting out-competed, then their targets should be de-repressed.
The simplest thing to look for is [whether] the targets of endogenously expressed microRNAs go up. To do that, you need to know the endogenously expressed microRNAs … and [using existing data] you can figure out, say, the top 10 expressed microRNAs in a cell.
You also need a target-prediction method, and our approach was to use a very simple method, just to look for a conserved 7-mer seed match, because we wanted the result to be easily reproducible and we didn't want it to be contingent on what method you used. You can get the same effect with different target-prediction methods, but this is just very simple.
Then, we looked for all the genes that have at least one site for an endogenously expressed microRNA and no target sites for the transfected small RNA, and we looked at the distribution of expression changes of these genes.
Typically, to see if their transfection works, people look at the targets of the transfected microRNA and expect that the distribution of expression changes of these genes shifts downward, to the left, compared to all genes or non-target genes. What we're seeing with these endogenous targets is that they are shifting to the right — they are going up.
And this bears out the hypothesis that there is this competition going on for the small RNA-processing machinery?
This is a statistical analysis, so we can't really address mechanism. What we can say is that it is a statistically significant effect [involving] an up-regulation of targets of endogenously expressed microRNAs.
We did another statistical analysis, a regression model, in the paper trying to pin it down a little more … and we showed that you can figure out, to a large extent, what the endogenously expressed microRNAs are just from these fold changes after transfection.
The idea is that … you [can] know the microRNA you transfected, but [may not] know any of the endogenously expressed microRNAs. But [with] a target-prediction method, you predict target sites for every seed class of human microRNAs … you [can] just select sites that help you explain expression changes in the regression model. If you do that over many independent experiments and see the microRNAs that helped you explain the … up-regulation, then the microRNAs you pick are actually the ones that are endogenously expressed in the cell type.
[In light of these findings], the simplest explanation [for the up-regulation of gene expression] is this competition effect … [although] there may be other explanations that are consistent with the data.
The reaction people have had is that they believe this is plausible and that [they] would guess this is happening; there is only so much protein machinery and it can get saturated. In other systems, the idea of saturation/competition is well accepted, and no one rejects out of hand that this is happening [with endogenous miRNAs]. It's an intuitive result, but hadn't been shown before.
[ pagebreak ]
What is your take on the implication of these findings for people doing research or developing RNAi or microRNA therapeutics?
There are important ramifications. For anyone using RNAi or doing functional studies, they understand now that siRNAs have off-target effects. But this is another kind of unintended and broad consequence of putting siRNAs into the cell … and is something to be aware of. It's not that are just a few dozen off-target effects, but there is a more pervasive effect.
When it comes to RNAi therapeutics, this effect has to be quantified and understood. We didn't have very much data for dose response or dynamics, but [with the] little data we had, we showed there is a dose response; as you titrate down … you get less of [this saturation] effect, but there is also less of an effect on your target. So it doesn't seem like you can titrate this away.
The other thing is that the genes that are affected and highly regulated by endogenously expressed microRNAs … [include] cell-cycle genes and oncogenes. … So, if you are, as a consequence of putting small RNAs into a cell, up-regulating oncogenes, that's obviously a problem.
It just has to be investigated and people have to be aware that this effect can happen.
Is there any follow-up work to these findings going on in your lab?
We are trying to do some quantifications of dose response in collaboration with … a colleague in my department, [Gregoire Altan-Bonnet]. We're trying to collect some data and measure more carefully what's going on.
I don't think we're in a position to figure out the details of the mechanism; there are other labs that are much better equipped to do that.