At A Glance
Name: Tom Tuschl
Associate Professor and Head of laboratory for RNA molecular biology at Rockefeller University
Background: The Max Planck Institute for Biophysical Chemistry, Department of Cellular Biochemistry — 1999-2002
Whitehead Institute for Biomedical Research; MIT, Department of Biology — 1995-1999
PhD, Chemistry, University of Regensburg — 1995
In 2001, Tom Tuschl established himself as an RNAi pioneer with his publication of a paper in Nature — written with MIT colleagues Phil Zamore, Phil Sharp, and Dave Bartel — describing how 21-nucleotide siRNA duplexes suppress gene expression in mammalian cells, including human embryonic kidney and HeLa cells [Nature 2001 May 24;411(6836):494-8].
In 2003, he joined the faculty of Rockefeller University in New York, where he is associate professor and head of the Laboratory for RNA Molecular Biology. RNAi News recently visited Tuschl at his lab to discuss his work and his vision for RNAi in the future.
People talk about “Tuschl’s rules” on siRNA selection …
The Tuschl rules are really no rules. The only rule that there is: that you need double-stranded RNA and you may need an overhang if you use a certain minimum size, that’s 21. If you make double-stranded RNA longer than 21 nucleotides, then they become so long that they can’t be processed by Dicer. This is why the hairpin siRNAs work fine; they are between 20 to 30 base pairs in length.
And the overhangs: the reason they were made [in the sequence] TT was really to facilitate the chemical synthesis in the lab. The TT synthesis reagent was a lot less expensive than the RNA uridine reagent, and very early on we tried to cut down on the cost of expensive RNA synthesis. TT is also the one that doesn’t have a lot of protection groups and is compatible with all sorts of RNA chemistry; that was another reason.
I don’t know if you noticed, but TT is also my initials. I actually didn’t even notice it, but eventually somebody pointed that out. But the fact is, TTs really do not contribute much to specificity or target recognition so it doesn’t really matter what your target sequence is.
So there are no rules. These were initial rules just for cost saving. Basically everything works. There are some variations in the activity, and then my TT rule will remedy these variations in activity.
This is a complex combination of the sequence you’re targeting, there’s the accessibility of the target RNA, the way things enter this RISC complex. There is some sequence specificity that affects the ability of the siRNA duplex to mediate its gene silencing. Because every sequence is unique, or you try to select unique sequences to avoid targeting of another gene, you basically start over with the same problem every time you make a new sequence, and it’s hard to come up with very general rules. Basically, you have to make a couple of siRNAs, some will be good, some will be very good, and some will not be so good, but the machine has been made such that you can deal with basically any sequence. Otherwise viruses or transposons would evolve such that they would become resistant to gene silencing.
What do you see as the biggest challenge for RNAi researchers?
It depends what you want to try to do. If you really want to deliver siRNAs into animals to turn off a gene in a specific tissue to save you time in making knockout animals, then the big challenge is really delivery, formulating the nucleic acid such that it gets into that tissue of interest, or it gets into any tissue. I think that’s the current status. It’s very hard to deliver to almost everything, with the exception of the liver, where you can use the high-pressure tail-vein injection method.
If you really want to enhance delivery, you have to form particles, you have to change the charge distribution of the nucleic acid or the hydrophobicity, or you have to conjugate the nucleic acid to a molecule that specifically interacts with a cell-surface receptor to enhance its pharmacologic properties. So that’s the big challenge at the moment. And that’s, of course, true also for therapeutic development.
In terms of the biology, nobody really understands what’s going on. I mean, you have all the proteins, everything is in your body, and what is it waiting for? There’s the hypothesis that it’s there as a biodefense system, there’s a hypothesis there that its regulating transposons, but if that’s the case, you should find the small RNAs from those sources in a cell. The presence of siRNAs or small RNAs would tell you whether there is such a process going on at the time you’re harvesting the tissue. And that, nobody has really looked at carefully.
What do you think is going on?
Well, I’m very intrigued at the moment because when we do these experiments, what we mostly find are microRNAs.
So this is probably why this whole RNAi machinery is there, because you express these microRNA genes and you have to use these microRNAs to regulate other genes. And these microRNAs are very well regulated; they are tissue-specific; they get turned on during development at specific times. At the moment, we don’t understand what those microRNAs are doing — if they do RNAi, if they do translational regulation like the few microRNAs that have been characterized in nematodes. We’re just trying to find out what the networks of gene regulation are.
In the cell, there are typically about 20 or some microRNA genes expressed; in humans you may have 200 to 300 microRNA genes, and I think about 200 have been identified at the moment. The question is: What do these 200 genes regulate, what is their target, and how is their mechanism different from RNA interference? What are the proteins that associate with the microRNA, and when we study microRNA can we learn something about how to design siRNA better? Is there any link to disease from these microRNA genes? These are the current questions, and I suspect that there may be certain tissues where there are not only microRNAs but a lot of the siRNAs that regulate transposon mobility or that regulate chromatin structures.
There was a number of recent papers that suggest small RNA can also induce changes in heterochromatin structures, and that’s been shown in plants, in tetrahymena, but there’s evidence lacking for this in human systems. This may be lacking because people have looked in the wrong cells, and it may be mostly in germ lines. Stem cell-like tissues have a completely different chromatin structure than differentiated cells, where a lot of genes have been silenced. I think there’s a possibility that in stem cells or developing germ line cells there could be a lot of small RNA that have more to do with transcriptional regulation, [and] heterochromatin, than with the conventional RNAi.
It’s just biology, basically. Finding protein, protein-interacting partners, and the associated RNA, and learn[ing] about the specificity of these interactions and the regulation of these proteins — that’s our big task at the moment. And then [to] see if there are any cases of dysregulation out there that could be linked to disease. There is evidence there are links to disease and there are a number of papers that get often overlooked where they say [that] polymorphisms in certain populations in the untranslated regions of mRNA can be linked to disease. It’s never been clear why modifications in the untrans- lated regions could have such an impact on disease, and I think one possibility is that these polymorphisms are found in microRNA recognition sites. So, microRNAs recognize not the coding region, but they are thought to recognize mostly the untranslated region. When they recognize the untranslated region, they prevent translation. So, if you had polymorphisms that affected microRNA recognition patterns, you would lose the regulation of that gene by a microRNA.
There’s a lot of effort at the moment to predict targets of microRNAs. For example, you use the human genome, cDNA sequences, and then when you find certain target sites predicted, you go into another sequenced organism, like the mouse, and double-check if you find, with similar frequency in the same related mRNA sequence, also microRNA binding sites. Then you can go in an even further species, and then if you find binding sites for microRNA [are] conserved in more than three species or more than two species, you can be 100 percent sure almost that this is the gene that’s going to be regulated or affected when a microRNA is being expressed.
I think the reason these microRNAs are so conserved is that they have more than one target and then evolution becomes very difficult. As the microRNA sequence changes, five other target genes have to co-evolve at the same time, and that’s basically impossible. And so, you find microRNAs frozen in evolution — that 21-nucleotide sequence[s] that are conserved from C. elegans, Drosophila, to mouse, to human, to fish, that’s quite unusual, and the reason for that is they have multiple targeting sites.
This may also be the main reason why RNAi is in every step. Because you at least need to express the microRNAs in every step.
Where do you see RNAi in the long-term? Do you see drugs on the market in, say, 10 years?
Yeah. I think so. The development is fairly rapid at the moment. The only hurdle we have is the delivery to get it work as a drug. Since that’s the only bottleneck, there will be a lot of investment from NIH, I’m sure, to enhance development of delivery techniques. I’ve talked about nanoparticles, liposome formulations, variations of siRNA, conjugated siRNAs that have different biodistribution, different uptake.
I think there will be lots of funding and there will be lots of creative ideas, because that’s the only problem. There are minor problems like specificity, off-target activities, interference of an siRNA with the natural regulatory mechanisms, that may require a little bit of sequence optimization, but these are do-able things that we understand fairly well at this point.
And the other applications of RNAi for screening purposes are just basic cell biology. In 10 years, I think every institution will have a set of validated siRNAs in the freezer. Then all the cell biologists have to do is to develop a reporter system that faithfully mirrors the process [they’re] looking at, then [they’re] going to knockout every gene in the human genome and see which gene affects the process that [they’re] studying.
So the problem with off-target effects, you think people have a handle on it?
Well, I think you have to be aware of it. It’s a very potent response, RNAi: that’s why it works so well. So, obviously the side effects with a very potent drug could also be potent. There’s no question about it.
But, you have computer predictions. You know the genome, you have it sequenced. So, you can predict by sequence homology the off-target activities, sooner or later, and that, I think, is a very important development that has to be made. But that is not a complicated problem; it just needs a bit of computer power and there are software tools that already use the genome-wide analysis to prevent off-target activity.
If you cannot completely avoid off-target activity, because there is always some similarity to another sequence of an mRNA, it’s also not going to be a problem for downstream analysis. You can make three or more siRNAs against different regions of a gene, so the off-target effects for everything will be different. Then you can subtract these differences from the common responses. You still can perfectly understand what your gene is doing downstream, what it’s regulating. That is what people are doing.
It’s still a reliable method to learn about gene function.