RNAi: sense and antisense

Gene silencing by small interfering RNAs (siRNAs) has emerged as an extremely useful technology for discovery biology, but off-target activity has proved a stumbling block. Detailed mechanistic insights and chemical modification of siRNAs can now improve the robustness of RNAi experiments, as Kate McDonald reports.

The use of RNAi technology to explore gene function and interactions has literally exploded over the last five years or so, both as a research tool and in potential therapeutics, and at the forefront of that explosion has been the use of synthetic siRNAs. Last year, US company Dharmacon became the first in the world to produce a human genome-wide siRNA library, which is allowing researchers to go where no one has gone before.

Clinical trials using siRNA technology as a treatment for diseases such as macular degeneration are moving ahead at pace in the US and there is a great deal of excitement about its potential. However, one of the "hot-button issues" in siRNA technology, as Dharmacon's executive vice president, Dr William Marshall, terms it, is the problem of specificity and resultant off-target events, in which knocking down unintended targets can cause toxicity.

In several recent papers in the journals Nature Methods and RNA, Dharmacon's researchers have uncovered the source of these events and discovered a way to improve the outcomes.

"An off-target event is mediated by an siRNA and isn't secondary to knocking down the gene that you are trying to target," Marshall says. "If you get multiple siRNAs that target the same gene, you'll get consistent profiles which are secondary to knocking down the gene. Additional down regulation events that are specific for each individual siRNA also occur (off-targets), but we couldn't find a consistent pattern that was indicative of off-target events mediated by siRNAs in general."

What Dharmacon has discovered is that siRNA can act like a microRNA (miRNA), the cell's native method of using the RNA-induced silencing complex (RISC) to co-ordinately regulate families of genes.

"It does this by targeting 3 prime (3') untranslated regions with short areas of complementarity at the 5' end of the guide strand," Marshall says. "To induce the siRNA mechanism you require 18-19 nucleotides of exact complementarity. These short areas of miRNA, known as seed regions, are six to seven nucleotides long and bind in the untranslated region of several messages. They then co-ordinately regulate the expression of the transcripts by causing translational repression or sequestering them in the P-bodies."

(P-bodies are intracellular organelles in which mRNA degradation takes place through nonsense mediated decay and have recently been shown to act in the microRNA pathway.)

"What we discovered and reported in this Nature Methods paper is the reason for RISC-mediated off-target events: it is that the siRNA acts like an miRNA," Marshall says. "What we did next was to add chemical modifications to the siRNA molecule, including some that were originally designed to stabilise siRNA molecules in blood.

"Part of the problem of using medicinal chemistry on siRNAs is that you introduce a modification and guess what? The siRNA doesn't work anymore. It happens all the time. But instead of throwing them away we decided to make lemonade out of lemons by employing the modifications to alter function and specificity.

"Through a systematic series of experiments we were able to identify a novel modification pattern that we employ in the development of our ON-TARGETplus siRNA. Essentially, we modify the seed regions of the strands with 2'O-methyl residues in a particular pattern and we showed that this will disrupt the microRNA like binding but maintain the siRNA binding.

"With this in hand we now have a new generation of siRNA molecules. It does a lot to reduce off-targets - we have shown that in 90 per cent of the cases we can eliminate off-target events. It's very exciting."

Global RNA initiative

Marshall was in Australia recently as a guest of local distributors Millennium Science to present a series of talks on his company's modified siRNA molecules. He also took time out to talk to various institutions on something close to Dharmacon's heart, the RNAi Global Initiative.

This initiative was established by Dharmacon to allow the non-profit research sector to get in on the revolution in discovery biology made possible by genome-wide siRNA screening, through a collaborative endeavour with a lower entry price than for-profit entities.

"One of the things about having the first genome-wide collection [of siRNAs] was that the pharmaceutical industry was an area that could afford to mobilise financial resources for the collection. When we would talk to non-profit institutes they were extremely interested, but it was too [expensive]," he says.

Dharmacon decided to encourage universities and public research institutes to come together to share new breakthroughs in RNAi science and to develop shared protocols for research purposes, all at a reduced cost. The initiative, which had its first meeting at the end of 2005, now involves 17 non-profit institutes, including Cancer Research UK, the German Cancer Centre DKFZ, Harvard Medical School, Stanford University, the University of Texas Southwestern Medical Center, the University of Texas MD Anderson Cancer Centre, GTI in Scotland and the Weizmann Institute in Israel, amongst others. The full list is on the RNAi Global Initiative website at www.rnaiglobal.org

"What we are able to do is transfer the collection to these groups at an affordable price," Marshall says. "In return for this we created a global collaboration to try to work together to overcome all the inherent challenges of doing screens of this magnitude.

"We have meetings twice a year where everyone gets together to exchange information about their research interests, latest findings, best practices, and generally share information related to accelerating the scientific discoveries made possible by genome-wide RNAi screening. We have a password-protected website where we keep the protocols and we have a chat room that can be accessed by any member. We also have monthly teleconferences where we talk about the latest results."

Marshall believes there is huge value in a global effort to exploit the potential of RNAi. "The sharing it allows is incredible," he says. "The goal is to advance science and medicine and to do it as quickly as possible.

"One of the really interesting aspects is the very early access to information that allows for cross-referencing the effects of inhibiting a gene target in various cellular processes, thereby enhancing the probability of identifying a good target for a drug intervention.

"For example, after a genome-wide cancer cell killing screen you identify that inhibition of certain genes can potentiate the activity of a chemotherapy agent. These become interesting targets for drug development. Now, with good cross-lab comparability, one can use the results generated by a different group that might be studying neurobiology using a genome-wide screen and they have identified hits that cause neuropathy.

"If you cross-reference this very early on and you happen to see that your top two hits for cancer intervention also happen to cause neuropathy, you may want to move down the cancer hit list to identify better targets with a lower probability of causing side effects. There is incredible power in this approach to significantly accelerate basic biology and medical discovery."

One of the RNAi Global Initiative's first tasks was to develop a standard called MIARE, standing for minimal information about RNAi experiments. It is similar to MIAME, a microarray protocol standard, except that MIARE is based on a prospective, rather than a retrospective, analysis.

"What we are trying to do is prospectively develop standards so that each of the labs' results can be highly comparable," Marshall says. "We did an identical experiment with 10 labs around the world. Dharmacon received all of the data back from those assays - this was done with a collection of siRNAs targeting over 900 genes including all the kinases, some phosphatases and cell cycle genes. What this does is give the members a teaching set that allows the prospective generation of a set of guiding principles for doing experiments.

"We explored all of the metrics that would apply in high-throughput screening, identified the optimal metrics that should be applied in general, normalisation factors as well so that you can normalise the results from different labs and really get good comparability, and then proposed MIARE as a global standard.

"This has all moved very quickly - we first came together in October last year and we have already submitted our first group publication."

Recent developments

Marshall is executive vice president of Dharmacon, now a part of the Fisher Biosciences group, of which he is also vice president of technology and business development.

He has a bachelor's degree in biochemistry from the University of Wisconsin and a PhD in chemistry from the University of Colorado, where he worked with the 'father of nucleic acid synthesis', Professor Marvin Caruthers.

"[Caruthers] is the one who discovered and developed the original DNA synthesis methods that are most widely used in the world today," Marshall says. "Some of the discoveries that were made in his lab at the University of Colorado have been commercialised by Dharmacon. The focus had moved from the synthesis of DNA, which he had done in the past, to the synthesis of RNA. And while it sounds like a trivial difference - there is only a 2'-hydroxyl separating them - it turns out that there are some significant challenges in these molecules."

Also at Colorado was Steven Scaringe, who established Dharmacon in 1995 to develop RNA synthesis technology and new RNA oligo-dependent applications. So when RNAi was creating shockwaves in the late '90s, Dharmacon began to specialise in developing the potential of siRNA.

Marshall was enticed to join Dharmacon after 10 years at Amgen, where he was, amongst other things, associate director of research and head of the nucleic acid and peptide technology department.

"I began looking at nucleic acid therapeutics, antisense therapeutics, and also built a high-throughput nucleic acid synthesis facility to support the genome sequencing project," he says.

From there he moved into the inflammation therapeutic area, where he investigated different targets for intervention by either small molecules or antibody-based drugs, before coming full circle and leading the nucleic acid and peptide technology group. The first peptibody his group worked on at Amgen is now in clinical trials for increasing platelet production.

His interest in siRNAs was sparked while still at Amgen with the publication of a paper on how siRNAs could mediate gene down-regulation in Drosophila. His group conducted a study designed to compare their current antisense platform for functional genomics with RNAi in terms of both gene knockdown potency and potential toxicities.

"Being somewhat dogmatic after building an antisense platform, I was resistant to trying RNAi, but I had some clever, persistent people in my group who finally pushed me enough and we gave it a shot. We found that the RNAi uniformly knocked the genes down more potently with no toxicity. From that point on it became the functional genomics tool of choice."

Marshall's Amgen group had been using Dharmacon's RNA synthesis methods, partly due to Marshall's association with Scaringe in Caruthers' lab.

"Then a venture capital company recruited me to move from Amgen to Dharmacon. We took the company from being mainly a synthesis house to having an integrated solutions provider with a bioinformatics group, a biology research and development group and a chemistry R&D group. We employed a lot of automation and novel systems and databases to ramp up production to fulfil the demand. The market was really exploding."

Massive market

And explode the market has. The landmark paper on RNAi in C. elegans by Andrew Fire and Craig Mello that won them the Nobel Prize this year was only published in 1998, and since then the understanding of the mechanism and its potential has increased beyond expectation.

At Dharmacon, the company set out to think very systematically about exactly what and how researchers would need to do an effective RNAi experiment, Marshall says.

"We thought about a very step-wise, systematic process where you have to be able to make the molecules. With our chemistry in hand we had that under control. The next thing was, how do we deliver these molecules into the cell, so we have worked extensively in exploring novel delivery systems that would be designed to deliver siRNAs. Everything out there was designed to deliver plasmids.

"An siRNA is much easier to deliver but it was like using a sledgehammer to pound a nail into the wall. The same delivery process, when used with siRNAs, caused toxicity issues. We identified novel lipids that would act more like a ball-peen hammer, a small hammer to knock a small nail into the wall. With that we were able to create our DharmaFECT siRNA transfection reagents, a series of four lipid solutions that all have slightly different properties that allow you to get siRNA into a very broad range of cells with minimal toxicity.

"What we can do now is explore a much broader range of cells, get into cells that no one thought possible, and when we couldn't do it with chemical delivery methods we have partnered with companies that can do electroporation (electric-pulse mediated transfer).

"Using one of those two methods gives you the ability to get into almost any cell."

The next step was to be able to design an active siRNA molecule of the literally thousands of siRNAs possible that target a specific gene. In 2004, Dharmacon published a paper in Nature Biotechnology describing the very first report of rational design of siRNA molecules.

"In that report we essentially created a set of siRNAs at random, tested their activity at knocking down the gene and then we used that as a teaching set to test hypotheses about what might be important aspects of highly functional siRNA. Based on these studies, we developed a weighted algorithm allowing us to predict the active siRNAs that will hit that gene. All we need is the target gene sequence information.

"We then moved from a teaching set and exploded the number of molecules that we looked at, moving from what was an eight-component algorithm to a 66-component algorithm. We call it our SMARTselection siRNA design algorithm - and all of our products are derived using the algorithm utilise this technology.

"Essentially what we do is employ the algorithm to in silico predict the active molecules, and with that in hand we can make large collections of these siRNAs targeting the entire human genome, the entire mouse genome and the rat genome."

Collaborative agreements

In addition to bringing out the first human genome-wide siRNA library last year, Dharmacon recently signed a multi-year collaborative agreement with Abbott Laboratories to develop siRNA in a variety of therapeutic areas. The initial focus, Marshall says, will be on cancer.

"We will focus on novel targets that we can inhibit with siRNA that wouldn't be considered 'druggable', or not amenable to making a traditional organic small molecule, but that show profound activity in the disease. RNAi allows you a means to really inhibit these targets. We will then focus on how to stabilise the siRNA in biological systems.

"Our next major focus will be on understanding how to effectively deliver the siRNA molecules. We potentially need to get them into solid tumours, into various tissues in order to treat diseases."

Dharmacon has also entered into a multi-year agreement with Alcon, based in Dallas-Fort Worth, the US's largest manufacturer of eye care products. This collaboration involves the development of therapeutic siRNAs to treat a wide range of eye diseases.

A very recent collaboration has been with Lentigen, a company that has developed a very high-titre lentivirus system, to develop highly active titred lentiviral particles to target any gene.

"Our goal there is that there are some cells that we just can't get into today but we are going to be able to. That's what we ultimately want to do - reach every cell possible and do the kind of experiments where you simply can't use siRNAs. siRNAs give you this fantastic opportunity to do almost everything and when you can't cover the ground, we are now going to have another option."