By Doug Macron
Researchers from the University of Kentucky this month published data in Molecular Therapy showing that 21-nucleotide-long siRNAs cause retinal toxicity even if they are formulated for delivery into the cell, potentially providing insight into the abandonment of several siRNA drugs that had been in human testing as treatments for wet age-related macular degeneration.
The work also follows previous studies from the same group demonstrating that these canonical siRNAs suppress angiogenesis, regardless of their sequences or targets, by triggering an immune response rather than through the RNAi pathway
However, the findings should not be taken to mean that therapeutics based on the gene-silencing technology cannot be developed, only that they will need to be designed in ways to avoid undesirable effects, Jayakrishna Ambati, a University of Kentucky researcher and senior author of the latest paper, told Gene Silencing News this week.
“The field is well-aware that there are several challenges to translating siRNAs to the clinic,” he said. First among them is the issue of delivery, and a number of modalities are under development to address this. “The second is off-target ... and immune activation effects. I'm sure different strategies can be employed to minimize or evade these effects.”
In the end, studies that identify the various pitfalls facing RNAi drugs will serve to increase the overall understanding of the biology of RNAi, and “better understanding of the biology will make for better chemistry and better drugs,” Ambati said.
In early 2008, Ambati and colleagues reported in Nature that 21-nucleotide-long siRNAs, including ones that were at the time in clinical trials as AMD therapies, suppressed angiogenesis in the retinas of mouse models (GSN 3/27/2008). These included Opko Health's VEGF-targeting drug bevasiranib and Merck's drug AGN211745, which targeted VEGF receptor-1 and was licensed to Allergan.
Notably, the investigators found that they could achieve the same anti-angiogenic effect with siRNAs against any target. “No matter what siRNA we used, as far as blood vessel growth in the retina was concerned, the result was always the same: that is, angiogenesis was suppressed” due to the activation of toll-like receptor 3, Ambati said at the time.
About a year later, the University of Kentucky investigators added to their findings with experiments that showed that a non-targeted 21-nucleotide-long siRNA would inhibit blood and lymphatic vessel growth in vivo as effectively as an siRNA against VEGF (GSN 4/9/2009).
According to Ambati, his team started the experiment that led to the newest paper after both Opko and Allergan pulled the plug on clinical trials of their respective AMD drugs (GSN 3/12/2009 & 5/28/2009)
“We showed that blood vessel growth suppression was a generic effect, but that was actually the effect that was desired in the eye in clinical trials,” he said. “It seemed to us that if [the trials] had been stopped, there must have been some reason why the endpoints were not being met. We wondered if there might be some toxicity accompanying the beneficial effect of blood vessel growth inhibition.
“That's what prompted [the latest work] and, indeed, we found that the beneficial effect of blood vessel growth suppression was accompanied by this undesirable effect” of retinal toxicity, he added.
In Molecular Therapy, Ambati tested a variety of 21-mers, both naked and cholesterol-conjugated to enable their cellular entry, against targets including VEGF and luciferase in the eyes of mice, and found that all of the siRNAs induced retinal degeneration by activating surface TLR3 on retinal pigmented epithelial cells.
Ambati noted that TLR3 was activating its downstream transcription factor, interferon regulatory factor-3, which caused retinal death. “We also showed, for the first time, that 21-nucleotide siRNAs can physically bind TLR3 and cause its phosphorylation,” he said.
Consistent with their previous data, Ambati's group also found that retinal degeneration did not occur with siRNAs shorter than 21 nucleotides, “opening the possibility that you can get siRNAs that are safe for retinal use and still achieve” target gene silencing, he said.
Ambati noted that 21 nucleotides is the minimum length required to “span two particular domains of two TLR3 molecules to enable them to dimerize … which is essential for TLR3 activation. It's that structural requirement that is being met by the 21-nucleotide sequence.”
Ambati said that chemical modifications could theoretically be used to avoid TLR3 activation, but that doing so depends on the modification being used.
“People have looked at 2'-O-methylated sequences, which seem to reduce TLR7 activation, but we've shown in our earlier paper that that particular modification does not prevent TLR3 from being activated,” he said. “There are other modifications that we and others have described, like the 2-thiouridine modification, that can block TLR3 from being activated … but really, the simplest way to overcome this problem is just to make the oligo a little bit shorter” than 21 nucleotides.
Have topics you'd like to see covered in Gene Silencing News? Contact the editor
at dmacron [at] genomeweb [.] com