Name: Frank Buchholz
Position: Group leader, Max Planck Institute of Molecular Cell Biology and Genetics
Background: Postdoc, University of California, San Francisco — 1998-2001
Postdoc, European Molecular Biology Laboratory — 1997-1998
PhD, Biology, EMBL and University of Heidelberg — 1997
Researchers from the Max Planck Institute of Molecular Cell Biology and Genetics this week published data from experiments in which they generated an endoribonuclease-prepared siRNA library for the human genome using and compared it with standard synthetic siRNAs for silencing efficacy and specificity.
Results of the work, which was done in collaboration with Cenix BioScience and Rosetta Inpharmactics, and appears in this month’s Nature Methods, indicate that these esiRNAs are comparable to their siRNA counterparts.
According to the paper’s authors, the esiRNAs also offer “an efficient, cost-effective, and specific alternative to presently available mammalian resources.” In fact, one of the paper’s co-authors hopes to find a market for the technology.
Christophe Echeverri, CEO and CSO of Cenix, this week said his company holds an exclusive, worldwide license to the UCSF intellectual property covering esiRNAs, adding that Cenix has been working with Frank Buchholz, a group leader at MPI-CBS and the senior author on the paper, on “exploring and developing the commercial potential of this technology.”
In an e-mail to RNAi News, Echeverri wrote that Cenix has an ongoing licensing program to provide commercial entities with non-exclusive rights to use the esiRNA technology for internal research and development efforts. He added that an undisclosed US-based large pharma has already taken such a license.
However, since Cenix is not a reagent provider, the company is “open to out-licensing [the esiRNA technology] to any commercial partner” capable of marketing it globally, Echeverri wrote.
This week, RNAi News spoke with Buchholz to get more details on the technology and the Nature Methods paper.
Let’s start with these endoribonuclease-prepared siRNAs. Can you give an overview of what they are and how they’re produced?
They’re basically the equivalent of what scientists use in Drosophila and C. elegans where RNAi is triggered using long double-stranded RNA. But of course in many mammalian cells these long double-stranded RNAs would trigger an interferon response.
What we do in vitro is what in Drosophila and C. elegans is done in vivo by Dicer. We start off with a long double-stranded RNA and digest it in vitro with RNase III or recombinant Dicer. Hence, we generate a pool of many different siRNAs, so what you end up with in the test tube is basically a mixture of many different siRNA sequences that all target the same transcript.
How long ago was this developed?
The first publication [on esiRNAs] was in 2002 in PNAS. They were discovered when I was at UCSF by a postdoc in the lab.
How did the idea come about to try using these for large-scale RNAi screens?
Originally, I didn’t pay attention to RNAi. I’m a mouse geneticist by training. I did hear about RNAi in C. elegans, of course, and in Drosophila, but knowing the long double-stranded RNAs wouldn’t work in mammalian cells, I didn’t pay too much attention.
One of the postdocs in the lab I was in at UCSF came from [a] Drosophila research [background] doing RNAi, and he was actually one of the first who discovered and published that long double-stranded RNA that people were using to do RNAi in Drosophila was actually cleaved in vivo by an RNase III enzyme. In a group meeting, he described that and had thought that maybe if one takes this step out and does this in vitro, maybe this would work in mammalian cells. And he showed that.
He took the gene-specific PCR product, transcribed that into long double-stranded RNA, and digested it to silence the gene. At the time, I was doing microarray experiments [wherein] we printed our own chips and where we used a PCR product to print on a glass slide. I immediately thought [the postdoc’s findings were] fantastic and we can just combine these two things — we can use the PCR products we have for microarray experiments and make esiRNAs to knock down each individual gene. This is one of the ideas that I took back to Germany when I started my own lab.
[Here at the Max Planck Institute] we then made further improvements … to really be able to do high-throughput production of the esiRNAs. This was a technical challenge we solved, and now we can make up to 2,000 esiRNAs a day if we want to.
Can you talk a bit about the paper and how you tested the esiRNAs?
We realized that we needed some bioinformatic help, so we teamed up with a bioinformatics group here and developed the software program DEQOR. Like some other algorithms that help you to design good siRNAs, [DEQOR] allows you to design good esiRNAs.
We then compared the silencing efficiencies of these [bioinformatically] improved esiRNAs with the latest algorithm-designed siRNAs, which we ordered from [Ambion]. The outcome of this was basically that the silencing efficacy was very comparable between siRNAs and esiRNAs.
There was a side aspect, which actually was quite interesting in that there was a very strong correlation between how well the silencing worked with the chemically synthesized siRNAs and esiRNAs, and that gave us the idea that maybe the efficiency of how well one can silence a transcript may be gene-dependent more than sequence-dependent. But that was only briefly mentioned in the manuscript.
Of course, we were aware of the off-target effects that have been observed with siRNA. The people that first published this were from Rosetta [Inpharmactics], so we called them up and said, “[Your data] is very interesting and we’d be interested to find out how esiRNAs compare in specificity to siRNAs.”
So we sent them esiRNAs to some genes and they performed microarray experiments. What they saw was that esiRNAs had a very big reduction in off-target effects so that silencing was a lot more specific with esiRNAs than with the siRNAs. Based on the information that we have, the most specific silencing they get is with esiRNAs.
Are there any concerns that one should keep in mind with these as far as false phenotypes and that sort of thing?
We’ve done screens now with these esiRNA libraries and we’re quite happy with the results. When we do validation and so on, we also lose some of the phenotypes that we see with the esiRNAs although the validation rate is very high. What we haven’t done is a real comparison between an siRNA screen and an esiRNA screen with the same assay to really demonstrate that you get fewer false positives in a screen.
What we know is we observe fewer off-target effects when you do a microarray experiment. How much that translates into fewer false positives in a screen we do not know. We’ve done a smaller subset of a gene family and we’ve seen there that we also get fewer false phenotypes with esiRNAs than with siRNAs, but we need to do a really large study. [We need to have] the same assay, the same conditions for transfection, and so on to then look at how well the esiRNA library performs compared to an siRNA library in a genome-wide screen. That is really required to say, “The esiRNA is the better resource.”
Is that the next step for you?
We haven’t done this yet, but in the long run that is something someone should do. If nobody else is interested, we should do this.
You mentioned Rosetta. You also worked with Cenix BioScience in doing this. What are the plans for the techniques that come out of this work? Is this all freely available or will it be commercialized?
The esiRNA generation was patented by UCSF when I was still there, and Cenix holds that patent. They have the exclusive license as far as I know. But for the academic world it should be freely available, and we’re currently talking with some distributors to see how we can distribute it. We already have some of the resources available through the [German Resource Center for Genome Research, or] RZPD, which is a genomics [non-profit] company here in Germany.