RNAi is hot these days, with basic science and business both moving at a brisk clip. To pick just one example of each, in the June issue of Annals of Surgery, now available online, researchers reported on RNAi preventing colorectal cancer from metastasizing to the liver. And earlier this week, San Francisco-based Sirna Therapeutics Inc. announced a $45 million financing. (See BioWorld Today, May 24, 2006.)

But as technologies mature, some of the bloom inevitably comes off the rose, and the hard work of transforming a basic discovery into something useful includes poking around and discovering drawbacks. RNAi is no exception, with two recent publications in Nature and Nature Methods focused on preventing toxicity of RNAi approaches rather than bringing its promise ever closer to the clinic in new indications.

The Nature paper appeared in the May 25, 2006, issue and was co-authored by scientists from Stanford University and Hepadnavirus Testing Inc., a Mountain View, Calif.-based company. It describes high toxicity of vector-delivered DNA coding sequences for inhibitory RNA molecules known as short-hairpin RNA, or shRNA. In the cell, enzymes cut precursor shRNA into silencing RNA, or siRNA, which then is assembled into a silencing complex and prevents protein expression.

In the Nature paper, first author Dirk Grimm, senior author Mark Kay and their colleagues showed that if there is too much virally delivered therapeutic shRNA around, processing of the cell's own microRNA, or miRNA (yet another form of RNAi), can suffer to the point of causing severe cellular toxicity. It's not clear which specific processing step was affected, but of about 50 shRNAs the scientists tried, three-quarters were severely toxic and half killed their test mice within two months.

The scientists did find a shRNA that knocked down hepatitis B virus for more than a year without killing the murine subjects, showing that the technique does not necessarily co-opt cellular machinery to a toxic degree. Kay told BioWorld Today that the rules governing toxicity were unclear so far, with seemingly similar shRNAs showing strikingly different degrees of toxicity. One important rule that emerged, though, is that with shRNA, shooting for maximal gene expression is more likely to lead to problems than to success. The usual approach with gene therapy is to try and maximize gene expression by any means necessary, but for shRNA, using weaker promoters appears to be a more promising approach.

William Marshall is executive vice president of research and development for Lafayette, Colo.-based Dharmacon Research Inc., which applies RNAi technologies to functional genomics. Dharmacon announced this week that the RNAi Global Initiative, which consists of Dharmacon and 14 research centers, has made progress in standardizing research protocols so that results will be comparable across laboratories - another decidedly glory-free task that is a sure sign of a maturing technology.

Marshall noted that the Nature paper "really has to do with expressed RNAs," or RNAs delivered via a gene vector, a technique would be used only for particularly recalcitrant cell types.

"The biggest issue with siRNA is getting them into the cell," Marshall said. In most cell types, those problems can be overcome, so "the viral-based approach is useful for very difficult-to-transfect cells, or for long-term expression," Marshall said.

Most RNAi, including the RNAi currently in clinical trials, is delivered directly as silencing RNA, or siRNA, which does not need the cell's processing machinery. So "from the viewpoint of the chemically synthesized short siRNAs that are in the clinic now, [the findings] are less directly relevant." (Senior author Kay himself was less certain about that sanguine assessment, saying that the paper "suggests, but does not prove," that the competition for cellular resources occurs upstream of siRNAs.)

If It's Not One Thing, It's Another

Not that siRNA doesn't have its own toxicity issues, because it does. It's just that the mechanism is different. In the March 2006 issue of Nature Methods, Dharmacon's Marshall and his colleagues delved into the details of that toxicity.

While the main problem with shRNA is the hijacking of the cell's miRNA machinery, siRNA can lead to so-called off-target events; while going after one gene, "a variety of other genes can also be misregulated," Marshall said.

The Nature Methods paper, co-authored by scientists from Dharmacon and Agilent Technologies, showed that contrary to popular opinion, the overall sequence of siRNA makes very little contribution to its specificity, or lack thereof. Instead, off-targeting occurred when the so-called "seed region" of a siRNA - a series of five to six nucleotides near the beginning of the 3-prime strand - matched with an off-target gene.

The scientists cautioned in their paper that "because the sheer number of genes that contain matches with any given siRNA seed region is very large in comparison to the number of actual off-targets for that siRNA, the value of the identified parameter (by itself) is limited."

Nevertheless, Marshall pointed out that the paper implies some ways to limit off-target toxicity for siRNAs, among them the paradoxical fact that a mixture of siRNAs may be a more specific way to target a gene, because the overall effect of the different siRNA sequences is pooled, while the off-targeting effect of the different seed regions is diluted.

Kay said that overall, his findings "shouldn't dampen the enthusiasm for shRNA therapies." He added that his lab is basically interested in hepatitis and hemophilia, and for those particular diseases, he feels shRNA, warts and all, is the most promising approach.

"I'm not married to any one technique," he said, adding that his lab has in fact switched gene delivery techniques several times. Kay is more skeptical of the siRNA approach than other practitioners, saying that "it has yet to be proven, except for a very few examples, that one can successfully treat a disease in an animal model with siRNA therapy."

What his Nature paper showed, he said, is that when using shRNA, "you need to do things in balance. And if you do things in balance, there is a high likelihood of success."