For those frustrated by the brute force of knockout mice when studying gene function, in vivo RNA interference (RNAi) would seem to have come to the rescue. In theory, in vivo RNAi promises a more nuanced approach that gives researchers temporal and spatial control in knocking down genes.

That doesn't mean, however, that it's easy to use. The notorious challenge with RNAi, even in cell culture, is delivery. "With in vivo, the problems are exactly the same, except ten times as hard," says Mark Behlke of Integrated DNA Technologies in Coralville, Iowa. Users struggle with getting RNA to their target organ or cell type, efficiently transfecting cells, and managing toxicity and off-target effects.

A handful of tricks can remedy these problems. For synthetic small interfering RNAs, for example, there are cationic lipid reagents to neutralize RNA's negative charge and ease it through the cell membrane; chemical modifications to protect RNA...

Tumor Knockdown

User: Matthias Ocker, professor of hepatology and oncology, University of Erlangen, Germany

The project: Knockdown of anti-apoptotic gene Bcl-2 and other genes linked to pancreatic and liver cancers

The problem: Spreading siRNA to tumors throughout the body

Approach: Ocker's lab grows tumor cells in culture and injects them as xenografts into mice. After the tumors develop, they use a low-dose (100 mg/kg bodyweight), low-volume (100 ml daily) intraperitoneal injection of unmodified siRNA in saline to systemically knock down the targeted gene. "We also observed that the vehicle you use is important for efficacy," he says. "Saline or PBS works well, but sterile water works less effectively."

• Systemic delivery makes sense in cancer, where primary tumors and metastases are dispersed

• There is no transport vehicle to target specific cell types, which would enhance efficiency.

The bottom line: "In cancers, you need a systemic approach," Ocker says. "That's the big problem today - we don't have a good carrying or target device that would facilitate uptake specifically for pancreatic cells." Solutions to that problem are slowly being found. For example, Judy Lieberman at Harvard Medical School recently showed that fusion proteins containing antibody fragments that recognize specific receptors can target breast cancer cells.

Superfine Targeting

User: Helen Lee Hellmich, assistant professor of anesthesiology, University of Texas Medical Branch, Galveston

The project: Studying the therapeutic potential of neuronal nitric oxide synthase and other genes upregulated after traumatic brain injury in rat hippocampus

The problem: How to target the right organ at the right time

Approach: Using microarrays, Hellmich identified genes upregulated for a few days (rather than hours) after injury, and thus temporally available for targeting with RNAi. Hellmich first cloned small interfering RNAs (siRNAs) into an adenoviral vector - expressed for about four weeks, it offers shorter knockdown than other viruses. For more clinically relevant results, she switched to siRNAs delivered in a cationic lipid reagent by stereotactic injection to hippocampal region CA3 - with a timescale of hours to days, small interfering RNAs provide a finer-tuned temporal delivery. Because she has little experience in RNAi tinkering, she sticks to commercial products with protocols for use. "I purchase it from the company, and I just cross my fingers and hope for the best," she says.

• siRNA gives a more predictable phenotype than a knockout mouse
• Companies provide protocols for using off-the-shelf products

• Difficult surgical targeting for delivery
• For every gene the procedure must be reoptimized

The bottom line: Intrathecal injection, the usual way of targeting the brain, wouldn't work to target a specific brain region, Hellmich notes. Her approach yields about 50%-60% knockdown in vivo. "That may be enough," she says. "If a gene is necessary for survival, you just knock down what was upregulated." Ultimately, the field will need better delivery systems, she says. The products available now "may not be optimal, for the five years I have on my grant [that] I could use them."

Viral Invaders

User: Sailen Barik, professor of microbiology and immunology, University of South Alabama, Mobile

The project: Knocking down functional subunit genes of the respiratory syncytial virus (RSV) to block infection in mice infected with the virus

The problem: Tweaking lipid-based delivery with chemical modifications and 27mer strands while keeping a protocol working overall

Approach: Barik delivers siRNAs complexed to cationic lipid transfection agents intranasally. "For respiratory viruses, this is the best route because it basically stays in the lung - it doesn't go anywhere else," he says. Naked siRNA also worked, but with low efficiency. The group later added chemical modifications, but determining where the modifications should be placed on the siRNA was a challenge, and reagents that had worked previously sometimes stopped working after modification. They also tried 27mer strands, though Barik says they did not significantly improve results.

• Lungs provide an effective delivery route

• Changing one parameter can mean reoptimizing the protocol

The bottom line: Getting different tricks to work in combination requires a lot of tinkering, and published protocols may not provide much help. "I think it is time to combine various chemical modifications with 27mers and test them in animals," says Barik. Chemical modifications are often effective, "but you have to figure out which position works for your system," he notes. What's more, combining delivery reagents with chemical modifications "throws off the whole delivery relationship," he says. "It's voodoo - half art, half science."

Vector Knockdown

User: Daniel Paskowitz, graduate student, Stanford University

The project: Knocking down expression of genes in the adult rat retina to study degenerative retinal diseases

The problem: Maintaining stability in genetic expression

Approach: Paskowitz used an adeno-associated viral (AAV) vector to deliver shRNA by injection into the retina. "DNA can be delivered to cells in a way that's much more stable" than siRNA," says Paskowitz. Although most AAV vectors take two to four weeks to turn on expression, he also used a strong promoter developed in a handful of labs that turns on expression in a few days.

• shRNA provides more flexibility with tissue targeting
• Users can switch different encoded RNAs into an optimized viral-delivery protocol
• Viral specificity for distinct cell types can be harnessed for targeting

• Variability in knockdown efficiency
• Cloning viral vectors takes longer than using straight siRNA
• Many viral vectors take time to turn on expression

The bottom line: "Not all shRNAs are alike," says Paskowitz. "Predicting which will work and which won't is still an imperfect science." Different viruses have different lengths of expression, but in retina, AAV-delivered gene expression is essentially permanent. Although the organ of interest in his case was easily accessible, he notes, viral delivery is also more flexible than most siRNA approaches. Some types of AAV specifically infect photoreceptors, while others go for pigment cells, for example, and theoretically, the approach can target distinct cell types, he says.

Full-on Knockout Replacement

User: John Stockton, manager, manipulated non-obese diabetes mouse core, Harvard Medical School

The project: Creating heritable knockdown of genes, such as CTLA4, that are involved in Type 1 diabetes

The problem: Wrestling with several insertion sites and the resulting variability in expression

Approach: "It's similar to the normal transgenic process," says Stockton, "but here, because there are several insertion sites, there's more variability in expression." Stockton and colleagues use a lentiviral vector to deliver short hairpin RNAi into mouse embryos via microinjection. They then insert the embryos into pregnancy-primed females, who deliver pups that carry partial knockdown of the targeted gene.

• Heritable and permanent gene effect
• About three times faster than making knockouts
• Applicable to species other than mice (birds, fish, rats, farm animals)

• Incomplete knockdown
• Variable expression rate
• Lentivirus requires Biosafety Level 2 precautions

The bottom line: Viral vectors encoding RNAi give a quick approximation of traditional knockouts, but so far users have succeeded in only partially blocking the function of the targeted gene. Still, the technique is not just faster but also more efficient. "On a good day, we can get 100% of the mice to be transgenic," he says. Otherwise, success rate is closer to 30%, he notes, which is still better than the rate of pronuclear injection. According to Stockton, founder mice express the construct in 10%-40% of their cells, and variability appears to stabilize over subsequent generations.

Further reading

M. Behlke, "Progress towards in vivo use of siRNAs," Mol Ther, 13:644-70, 2006. M. Amarzguioui et al., "Approaches for chemically synthesized siRNA and vector-mediated RNAi," FEBS Lett, 579:5974-81, 2005.

Interested in reading more?

Magaizne Cover

Become a Member of

Receive full access to digital editions of The Scientist, as well as TS Digest, feature stories, more than 35 years of archives, and much more!
Already a member?