At A Glance
Name: Fabio Piano
Position: Assistant professor, biology, New York University
Background: Postdoc, Cornell University — 1996-2001; PhD, biology, NYU — 1995; BS, biology/computer science, NYU — 1988
After becoming involved with RNA interference while at Cornell, Fabio Piano — a true New Yorker at heart — returned to his academic roots to pursue the study of embryonic development at NYU.
Recently, Piano spoke with RNAi News about his work.
How did you first get involved with RNA interference?
My initial involvement with it was going to Cornell. I was interested in studying how genes evolve, how their functions evolve, and there was this technique that had just been devised whereby introducing in vitro-transcribed RNA elicited a phenocopy of the null allele. I just thought that would be a great tool [to use] to start analyzing gene function across species. So, I decided to try to test this idea, and started to use this technology to make in vitro-transcribed RNA and test it across different species and see how that would work.
In doing this, I quickly realized that we could do a more forward genetic type of analysis of other species that were not genetic model systems. In order to be able to really do this well, I thought we would have to develop technologies to do high-throughput analysis of these kinds of injections and treatments. So, I developed some techniques to do that and wanted to test [them] in C. elegans as a means to get the background. That, of course, became the beginning of what then started to be a big genome project, because we could then have that background knowledge to do the work on the other species. In a sense, I sometimes feel I'm still doing the control experiment that I wanted to do for a long time.
Can you talk about the genome work?
Basically the idea was, being that we can knock down genes using this technique, could we just go ahead and do the whole genome [of C. elegans] and then find out what the genes do in the early embryo? That was the idea of trying to basically make a whole map of the phenome that would give us the information of what a gene does in the early embryo.
I chose the early embryo because there was a lot of information already known about [it], as well as the fact that it is an amazing place to look at the effect of removing genes in terms of very basic cell biological functions [and] some beginning of developmental biological functions. We can easily see chromosomes dividing, we can easily see nuclei move, we can see meiosis, mitosis, cytokinesis, and all these processes just by looking at the embryo. So, I thought this is a very good system to do this knock down approach and just filter the whole genome through this system to identify all the genes required for these basic processes.
So how far along is this project?
I'm not the only who's been doing this. The thing I think is really wonderful is that as a community, we've done the whole genome more than once on this idea, although not the analysis of the early embryo.
Let me step back for a moment and tell you that basically the project began to take a big picture and we realized that instead of just going around choosing genes sort of randomly, we wanted to try to rank them so that we'd test the ones we thought were more interesting first. Because we were looking at the early embryo, we decided to rank them according to whether or not they are expressed in the ovary, because the ovary is the place where all the genes are expressed to make early embryogenesis happen. So, we basically have been focusing on the ovary-expressed genes. Valerie Reinke [of Yale University], who did the first microarray experiment in C. elegans, had identified early on the set of genes that we enriched in the ovary using microarrays, and we just started by knocking down all of those. We also made a cDNA library from knocking down all of those.
Finally, I can tell you that now, there are about 3,000 genes that are known to be ovary-enriched and we've analyzed all of them.
Now that you've analyzed them, what's next?
There are several things. This is an important point because what happens once you actually have the information? Our information, at the moment, is a time-lapse recording of early embryogenesis, which is very rich in information, but in the end it's just a bunch of movies. So, what we've had to do is think about how to analyze these data in a way [where] we can turn these kind of analog data into digital data [on which] we can use mining tools that allow us to analyze this. For example, you can think of the same thing as happening between a Northern blot and a microarray — it becomes numbers. We want to turn movies into numbers that we can then use the computers to analyze.
So, we developed a system that we have published where we code each defect that we see into a particular number. Then, each gene gets a signature according to the series of defects that we see. With this signature, we can do a lot of different things: We can cluster genes according to similar phenotypes. We published something we call PhenoBlast [with which] we can take a particular phenotype and search all the other phenotypes to rank them according to how similar they are to this phenotype. These kinds of tools allow us to make predictions on the gene functions ... we don't know anything about. What happens is we get a cluster, for example, of genes that are components of a polarity pathway and then we might get something new in there and we go ahead and test that.
It turns out that this is a very predictive type of analysis because phenotypes have been used for a long time. When you group genes together according to similarity in phenotype, they very often are working on the same pathway or even sometimes completely directly with each other. So, that turns out to be very good.
We're continuing to explore the kinds of information we have — for example, the next step in this line would be to try to ask: Are there any superstructures in terms of the data? You would have clusters of proteins that affect the cell cycle. Then you see clusters of proteins that we know affect chromosome segregation. Those two clusters seem to be more closely related to each other than either of them is to, say, polarity. So that allows us to start seeing the hierarchical way in which some of these subnetworks are organized within the whole network.
That's one thing. The other thing that's very important is that we're integrating this information with all the other functional genomic information.
They talk about systems biology.
Right. Systems biology is possibly many different things but the integration is certainly a component that is very important.
Even when I did my PhD, which wasn't very long ago, a typical molecular biology thesis would be you get a set of mutants, you know that they are working in a similar process, you go ahead and map them, and then you find a gene, clone it, and identify what it is. Then you make antibodies, you find out where it is. What I'm trying to tell you is, it's one gene but many techniques. With genomics, what's been going on is it's been mostly one technique but all the genes, or many of the genes.
What's next is basically taking these techniques that have gone horizontally and trying to combine them. All of a sudden, you know what a gene does when you knock it out, where it's localized, what kind of mutations you might get in that particular gene, and what structure that might have if there is a structure — things like that.
So, that's the next step in terms of putting these data together. The phenome, in a sense, is very difficult because it's all phenomena that you're seeing. You might consider the fact that, for example, you might see that a bunch of genes are always expressed together — they go up, they go down. You might even have a protein-protein interaction map that tells you that these components are hooked up with each other. But all this information cannot tell you, for example, that the protein you are interested in is important for chromosome separation. It's kind of the difference between biochemical function and cellular function.
Are you collaborating with anybody?
We are. We have a strong collaboration with Marc Vidal at the Dana-Farber [Cancer Institute], who is spearheading a protein-protein interaction map in C. elegans, as well as the ORF-eome project.
Basically, in C. elegans, because it's been the first animal for which we've had the whole genome sequence, it's pretty advanced as far as the knowledge we have about the genome, as well as the tools that we have at our disposal to analyze the genome. One of those things is the ORF-eome project, which Marc Vidal has been doing, which puts all the ORFs in a vector [in which] they can be shuttled around.
Also we've just been collaborating now with a company called Cenix [BioScience] in Germany. I've known [Cenix CEO/CSO] Chris Echeverri for a long time and since he started doing C. elegans work in Cenix, we are now collaborating with them and helping analyze their data.
Not that what you're doing isn't enough, but are there any other sorts of areas you envision getting into using RNAi?
There're a couple of things I think are very important. For one, just in the genomics part of RNAi, what's really important is that in an academic setting, you also have to put the resources [into] [generating] the databases that have to be put together in order for people to navigate these data. That's been a major focus, in a sense, because without [well-curated and easy-to-navigate databases], you end up losing a lot of information. I think it's important that … biology labs have to, when they start doing a large-scale project, develop a bioinformatics team to put that together. With that, we're beginning to [ask] questions like: Can we do a good job at trying to represent the data so that we not only are presented the results from all the different studies, but we also represent … how it's possible how one experiment, for example, targets more than one gene?
These are the kinds of questions that we're beginning to develop ideas on.
Also, the other things that we [want to be] doing in the future is to look at the mechanism of … gene action across different species — which is the original reason for getting involved [with RNAi]. So, we are really interested in using RNAi across species and finding out what the genes do in terms of network evolution.
The last thing that we are also getting involved more and more in is this connection between RNAi and microRNAs, and the possibility that they have [a role] in regulating a lot of the processes we are studying. There is really no transcription going on in the early embryo so we are interested in asking the question: [Do] microRNAs have some interesting roles in early embryogenesis?