Name:
Barry Dickson
Position:
Scientific director/senior scientist, Research Institute of Molecular Pathology
Background:
• Group leader, Research Institute of Molecular Pathology — 1998-2002
• Junior group leader, University of Zurich Institute of Zoology — 1996-1998
• Postdoc, molecular/cell biology, University of California, Berkeley — 1994-1995
• PhD, developmental genetics, University of Zurich — 1992
• BSc, genetics, University of Melbourne — 1987
At the Research Institute of Molecular Pathology in Vienna, Barry Dickson investigates the function of neural circuits in Drosophila, but his work led him to spearhead the creation of a genome-wide transgenic RNAi library for conditionally inactivating genes in the model organism.
More recently, he helped develop a second transgenic RNAi library, which will be officially unveiled at the Genetics Society of America's annual Drosophila Research Conference next month, designed to overcome some of the potency and specificity limitations of the first.
This week, RNAi News spoke with Dickson about the new library.
Let's begin with an overview of your lab's research.
Our primary interest is in Drosophila behavior. The interest in transgenic RNAi came out of asking, "How can we more efficiently look for genes that are involved in behavior?"
With classical mutagenesis screens, where you make an animal entirely homozygous for a mutation, you very often end up with a dead fly. We said, "Wouldn't it be great if we could target gene disruption specifically to the nervous system?"
This was the thinking about six or seven years ago, and around that time transgenic RNAi was shown to work quite robustly in flies. We decided to make a library so we could start screening it to knock down gene function specifically in the nervous system.
That's what led us to create our first library of 22,000 lines, which was published in 2007.
Can you give some details on that first library?
It is based on the GAL4/UAS system, using GAL4 to drive expression of long double-stranded RNAs in specific cells. There are large numbers of GAL4 lines available — basically any fly researcher interested in a certain tissue has a set of GAL4 lines for that tissue. This first library relied on P-element transformation, which until a few years ago was all we really had for generating transgenics in flies.
[Although] this first library was published in the summer of 2007, it was available already in the spring through the Vienna Drosophila RNAi Center, which was set up to distribute these lines.
The main stock collection for Drosophila is [at the Bloomington Drosophila Stock Center at Indiana University]. We had asked them, when we started out making this library, if they would be able to handle it. They said, "No, it's more lines than we have already ourselves," but we were then able to raise some funds from the city of Vienna and the Austrian Federal Ministry to set up a stock center here.
What led to the effort to put out a second library?
One reason is that, generally, you'd like to have a second, independent RNAi line for every gene. So in part it's just to have a second set of lines. But we also realized some of the limitations with the first library, which were all really due to the method of transgenesis.
The main limitation is that the expression levels can be quite variable because the P element inserts at random; in some places it's very well expressed, in some places it's poorly expressed. And so you can get false negatives because the RNAi transgene just doesn't get expressed. Also, because these are inserted at random, you can misregulate flanking genes and that can cause phenotypes. We don't think this happens too much, but we have seen a few cases.
So there are false negatives and false positives due to the fact that we used P element transgenesis. Then, around 2004, Michelle Calos at Stanford [University] developed the phiC31 system for site-specific transgenesis, which is just beautiful. [With it], you can target all the transgenes to one particular site in the genome.
We invested about a year or two finding a site that would give low basal levels of expression but could be induced very strongly and didn't disrupt any endogenous genes. Then, using the phiC31 [system], we inserted the transgenes at that site.
The hairpins themselves have been designed independently, so the second library contains RNAi constructs different from the first library. The design has largely been done by Michael Boutros at [the German Cancer Research Center] in Heidelberg, using better bioinformatics to make them more potent and more specific.
And the second library has 7,500 lines?
7,707 so far, we're aiming to reach about 10,000 by the end of the year.
The second library offers advantages over the first, but you envision that they will both continue to be used, correct?
Yeah, I expect so. For a researcher who is interested in gene X, they will want to get as many RNAi lines as they can. … If you want to do a screen, [however], you'll probably want to use a library that has the most effective [lines], and for that they would use our second library, because we know that the false-negative and false-positive rates are much lower. Also, because we now only need one insertion of each transgene, there are also fewer lines to screen.
Do you plan to publish details about this library, as well?
Yeah, at some point we will, although there are no immediate plans. The focus for now has been on generating the lines and getting them out to the community.
With the bulk of this second library in place, is there additional work that needs to be done? Are you thinking about a third library?
No, we don't see much point. There's also a P-element library … in Japan, so I'm not sure there is much point in now making a fourth. Our new library looks really good, and I don't think we can make the libraries much more potent or more specific.
There are always new methods coming out — there was a recent one from Mike Levine [at the University of California, Berkeley], that uses microRNAs to trigger RNAi. To us, this doesn't look like it's going to be more effective, mainly because of the difficulty of designing a single potent siRNA, but you never know.