Long-term RNAi can be deadly

Long-term RNAi can be deadly
25th May 2006
By Charles W. Choi
The Scientist
Long-term use of RNA interference (RNAi) can be fatal in mice, scientists report in this week's Nature. However, some short hairpin RNAs (shRNAs) suppressed viral infections without killing the mice, suggesting that the technology may still be useful -- if used carefully.
This shows RNAi therapies "can be very effective, but [scientists] will have to really test the sequences and doses [they] want in clinical trials," John Rossi of the Beckman Research Institute of the City of Hope in Duarte, Calif., who did not participate in this study, told The Scientist.
Study author Mark Kay at Stanford University in California and his colleagues, led by postdoctoral researcher Dirk Grimm, began investigating the long-term effects of RNAi after a 2003 study of shRNAs against hepatitis B yielded some disappointing results. Although the initial findings were promising, "then we saw some dying two or three weeks later," Kay told The Scientist.
During the current study, the researchers investigated the effects of 49 distinct shRNAs, unique in length and sequence. The shRNAs were directed against six targets, using a virus vector directed toward the adult mouse liver. They found 36 shRNAs caused dose-dependent liver injury. Within two months, 23 shRNAs had killed their hosts.
Experiments with the vector alone revealed the virus was not at fault. Enzyme-linked immunosorbent assay (ELISA)-based screening and Western blot analysis found no evidence of RNAi activation of interferon responses. Another theory was that the shRNAs were toxic because they acted on non-target RNAs, "but we had similar lethal results with six different targets and 23 different shRNAs, making off-targeting unlikely as the cause," Kay said.
High shRNA levels were often linked with toxicity, suggesting that toxic shRNAs oversaturated cellular shRNA processing machinery, preventing vital microRNAs (miRNAs) from accessing those proteins. The researchers performed in vitro fluorescence studies with human liver cells that abundantly expressed the most common liver miRNA, miR-122. They found that highly expressed, toxic shRNAs inhibited miR-122 processing, while shRNAs that did not kill their host appeared to have little or no effect on miR-122 processing.
Some shRNAs were effective, the researchers found. Specifically, in mice given human hepatitis B virus and low doses of weakly expressed shRNAs, Kay and his team reported a more than 32-fold and 150-fold decrease in serum viral DNA within two weeks and four months, respectively -- all without signs of toxicity.
Overexpressing the nuclear export protein exportin-5 in mice enhanced the RNA silencing effects of both shRNAs and miRNAs, suggesting this protein may be one of the important resources over which shRNAs compete with miRNAs. "Exportin-5 is in pretty low abundance in most cells, and may be an easy thing to saturate," Rossi said.
The findings are "provocative and striking," Timothy Nilsen at Case Western Reserve University in Cleveland, who did not participate in the research, told The Scientist. The correlations between expression and toxicity aren't perfect, so "it will be interesting to see if toxic or lethal hairpins expressed with a weaker promoter were both efficacious and didn't have a toxic effect. That would clearly indicate that it's levels of expression that are important."
Irvin Chen at the University of California at Los Angeles, also not a co-author, noted he also has found that some shRNAs directed against CCR5 for HIV-1 therapy proved toxic with long-term use in primary human T-cells and chimeric mice, while other shRNAs were not toxic in the same setting. "The more data we accumulate about what shRNAs and siRNAs are and are not toxic, the better we can get at the mechanisms and in the future hopefully be able to predict toxicity," he said.
Comments: 0
Votes:28