In 2006, researchers from Stanford University reported that there can be too much of a good thing as far as the delivery of short interfering RNAs is concerned. Of about 50 RNAs the scientists gave to mice, three-quarters were severely toxic and half killed their test subjects within two months. (See BioWorld Today, May 26, 2006.)

The 2006 paper was "very provocative," David Bumcrot told BioWorld Today, noting that "Fatality" was the first word in the title: "When you see a paper like that, you definitely take notice."

Gene vectors are a delivery method that probably would not be used for very many cell types.

But the 2006 paper nevertheless "created a concern . . . that maybe this is a property of RNA interference in general," Bumcrot said.

In a new paper in the Sept 27, 2007 issue of Nature, senior author Bumcrot, and his colleagues at Alnylam Pharmaceuticals Inc., the Massachusetts Institute of Technology, the University of Texas and the Swiss Federal Institute of Technology addressed that concern, and showed that delivering siRNAs directly to the cytoplasm appears to avoid the toxicity problems that were reported in the earlier work.

The authors of the 2006 paper delivered short hairpin RNAs or shRNAs via gene therapy. Such shRNAs are expressed in the cell nucleus, processed into siRNAs by the cell's own machinery, and exported into the cytoplasm.

And that appears to be the key problem. Cells also are producing their own microRNAs or miRNAs, which are critical for cell function and also need to be ferried from the nucleus to the cytoplasm. SiRNAs and miRNAs compete with each other for spots on the ferry.

"The choke point was the export step," said Bumcrot, who is director of research at Cambridge, Mass.-based Alnylam. That competition, in turn, led to reduced levels of several key miRNAs in the cytoplasm, leaving the cells unable to function normally.

In the new paper, the researchers used three siRNAs targeting three different liver-expressed genes, as well as a control siRNA.

They synthesized the siRNAs chemically and delivered them directly to the cytoplasm. That method "bypasses the nuclear export step that is critical for shRNA," Bumcrot explained.

The scientists administered the siRNAs to mice and hamsters, and first tested whether the levels of endogenous miRNAs were adversely affected. Looking at the same miRNAs as the 2006 paper, Bumcrot and his team found no decrease in their cytoplasmic levels.

The finding that miRNA levels themselves were unaffected "might have been expected," Bumcrot said, given that siRNAs do not need to be exported from the nucleus. So the team went on to test whether the function of the miRNAs was somehow affected.

"We wanted to rule out that there could have been competition at other levels" downstream from the nuclear export, Bumcrot said.

On the functional level, too, miRNAs appeared to be unaffected by siRNAs. Bumcrot said that at levels of siRNA that reduced target gene expression by 80 percent, there was no change in the mRNA or protein levels of the genes targeted by miR-122, a key microRNA that was affected by shRNA in the 2006 paper.