Researchers from the Swiss Federal Institute of Technology (ETH Zurich) and Alnylam Pharmaceuticals this week published data shedding new light on the process by which systemically administered, cholesterol-conjugated siRNAs are taken up into mammalian cells.
Among the key findings are that high-density and low-density lipoproteins, as well as the mammalian homolog of a gene known to influence RNA transport in C. elegans, are critical for delivering the siRNAs into tissues. The findings, while preliminary, could point the way toward overcoming the delivery challenges that have plagued the RNAi therapeutics field, according to the research team.
The research, which appears online in Nature Biotechnology, builds on earlier work by Alnylam in which company researchers found that conjugation of siRNAs to cholesterol improved the RNAi agent’s in vivo pharmacokinetic properties without affecting gene-silencing activity (see RNAi News, 11/12/2004).
“It was known that when you conjugate siRNAs with cholesterol … [the RNAi molecules] are being taken up in tissues,” Markus Stoffel, a professor at the ETH Zurich’s Institute of Molecular Systems Biology and a senior author on the Nature Biotechnology paper, told RNAi News this week.
“But there was nothing known about how these compounds were being taken up,” he said. “And if you want to improve specificity and efficiency of uptake, you need to know something about [the process].”
In their previous experiments, Alnylam investigators had used cholesterol-conjugated siRNAs to knock down apolipoprotein B, a protein essential for the formation of low-density lipoproteins and associated with cholesterol metabolism, in mice.
In looking to understand more about how this silencing occurred, Stoffel and his colleagues again used siRNAs targeting apoB, but radiolabeled the oligos in order to identify the serum components to which they would bind.
Additionally, they bound the siRNAs to a number of other lipophilic conjugates besides cholesterol “to determine whether different lipophilic modifications facilitate uptake of chemically modified siRNAs and silencing of gene expression in vivo,” according to the paper.
The researchers found that siRNAs conjugated to cholesterol, stearoyl, docosanyl, and lithocholic-oleyl associated with HDL particles, while siRNAs bound to short- and medium-chain fatty acids, such as lauroyl, myristoyl, and palmitoyl linked to either serum albumin or remained unbound.
Since mice exhibit very low LDL levels in plasma, cholesterol-conjugated siRNAs were also evaluated in human and hamster plasma, where they were found to associate with LDL, HDL, and albumin, the paper’s authors wrote. Binding affinity was highest with the lipoproteins and relatively weak with albumin.
By comparing livers selectively transfused with either lipoprotein-linked siRNAs, albumin-linked siRNAs, or unbound siRNAs, the researchers revealed that lipoprotein binding was required to achieve cellular uptake in vivo.
To further validate this finding, cholesterol-conjugated siRNAs were administered to mutant mice lacking the plasma membrane receptors to which the lipoprotein particles dock. The researchers found that siRNA uptake did not occur under these conditions, Stoffel said.
“At first we thought that maybe the whole particle, with the siRNA, is being taken up” by the cell, he said. “But we showed that this is not the case … because the clearance of the siRNA is much faster than the lipoprotein particle.”
This differential clearance “suggests that that lipoprotein-bound cholesterol-siRNA is carried by LDL and HDL particles but taken up by cells through a mechanism independent of endocytosis of whole lipoprotein particles,” according to the Nature Biotechnology paper.
After this discovery, “we were stuck for several months and didn’t know what to do,” Stoffel said. Then, a search of the literature turned up work by Harvard University’s Craig Hunter, who had discovered years before a set of genes called SID, or systemic interference defective, that plays a role in RNA transport in C. elegans (see RNAi News, 7/9/2004).
One of those genes, Sid-1, has a mammalian homolog that Stoffel homed in on.
“When we knocked [the homolog] out, we prevented delivery of these siRNAs” to hepatocytes, he said.
In the course of their research, Stoffel and his colleagues also found that HDL and LDL particles mediate delivery of cholesterol-conjugates siRNAs to different tissues.
“Whereas LDL-bound cholesterol-siRNAs are mainly taken up by the liver, HDL-bound cholesterol-siRNAs are taken up by various tissues, including adrenal, ovary, kidney and liver,” they wrote.
Furthermore, they discovered that the silencing effect of the conjugated siRNAs could be significantly enhanced if they are bound to lipoproteins before administration.
“It was known that when you conjugate siRNAs with cholesterol … [the RNAi molecules] are being taken up in tissues. But there was nothing known about how these compounds were being taken up.”
Chance plays a big role in determining whether the cholesterol-conjugated siRNAs associate with HDL, LDL, albumin, or remain unbound and are quickly excreted, Stoffel said. In order to ensure that the siRNAs bound to certain lipoproteins, HDL for example, the researchers isolated these particles outside of the animal, “pre-loaded” them with the RNAi molecules, and re-injected them.
“When we do this, we increase the efficiency of delivery and siRNA activity 8- to 15-fold,” he said. “We don’t achieve this [kind of knockdown] when we just inject the siRNA into the blood because [lipoprotein association depends on] chance — it’s a concentration-dependent effect.”
Though much work remains to be done to better understand the role Sid-1 plays in lipid-conjugated siRNA uptake in vivo, Stoffel said he is optimistic that the data published in Nature Biotechnology will go far in addressing the delivery issue that continues to hang over the RNAi therapeutics field.
“There is an opportunity to exploit [these findings] by using … HDL-like particles that can be loaded” with siRNAs as drug-delivery agents, he said.
Stoffel said that his lab is currently exploring the use of artificial lipoprotein particles for this purpose since the use of native particles strikes him as too time-consuming for clinical applications.
“It could be done, but it’s impractical [to] isolate the particles, load them, and then give them back” to a patient, he said. “It would be much easier to formulate” them synthetically.
Additionally, the use of artificial lipoprotein particles would allow for the integration of targeting ligands that could improve delivery specificity.
“Basically, [you could] make a particle that doesn’t recognize the LDL receptor, [for instance], but maybe some other receptor on a specific cell type,” he said. In this way, one “might be able to achieve very specific delivery. There is a good chance this will work in the future.”
Officials from Alnylam were unavailable for comment on whether the company is incorporating the findings in its drug-development efforts. Stoffel is a member of Alnylam’s scientific advisory board.