Skip to main content

Using RNAi: Hairpins versus siRNAs


By Meredith W. Salisbury


It’s no secret that RNAi, which has enjoyed a wildfire-like spread through the scientific community, is evolving just as rapidly as it’s taking hold. Even experts in the field are hard-pressed to point to one use or technique over others and say, “This is the way it ought to be done.”

The same holds for methods of inducing RNAi. When the field began, chemically synthesized siRNAs were used essentially across the board. But by 2002, Greg Hannon’s Cold Spring Harbor Laboratory team had published its work using hairpin RNA, showing that it could be used with vectors or plasmids and seemed to cause more stable gene silencing.

Even now, no one can say with certainty whether one is better than the other — in all likelihood, one of the techniques will be more suitable for certain kinds of experiments, while the other will be optimal for different ones. “Time and experience will really help identify what methods will provide the most value,” says Stephen Scaringe, CSO of Dharmacon, a vendor of both short interfering RNAs and short hairpin RNAs.

Today, vendors agree that chemically synthesized siRNAs are the most highly sought means of using RNAi. They’re “by far the revenue leader,” according to David Dorris, business development manager for Ambion, another vendor selling both tools. But Scaringe points out that viral vectors, used with hairpins, are also popular.

To help navigate the issue, Genome Technology chatted with experts in the field to figure out pros, cons, and in betweens of each method. What follows is a look at how time, type of study, ease of use, reliability, and cost could factor into your decision to use synthetic or hairpin RNAs.

Time investment

Everyone seems to agree that in a head-to-head competition, chemically synthesized siRNAs beat out the vector-based alternatives when it comes to how much time goes into preparing for the experiment.

Tom Tuschl explains the process when he wants to knock out a gene: “Within two hours I have my siRNA design. I can synthesize four RNAs in three hours, then I deprotect them. The next day at noon I have my RNAs hybridized and ready to go in tissue culture. After two or three days I check the phenotyping. In four days or five days I have completed from the design to the experiment — you can’t really do it that fast with vectors.”

Most people aren’t like Tuschl, and don’t design or synthesize their own siRNAs. For those people, argues Scaringe, the process is even less time-consuming. As CSO of Dharmacon, one of the largest siRNA and shRNA providers, he says, “All one has to do is identify the gene, perhaps provide the sequence of the gene.” Dharmacon handles the design and testing, and “four days later, the compound’s on their desk,” Scaringe says.

Vectors, on the other hand, require much more attention. “With vectors, they’re going to spend the next four or five days building it,” Scaringe adds.

Tuschl points out that the common-sense step of verifying your hairpin sequences would take an additional two or three days. Scientists could easily omit that step, Tuschl says, but that leaves them susceptible to acquiring results that involve a cloning mutation or other problem, rendering the experiment useless. Also, Tuschl notes, after spending more time on developing the vectors, “then all you really have is just a plasmid that you can transfect,” he says. It’ll take “another three weeks to establish a cell line that responds and that is homogeneous in its knockdown properties.”

Nature of study

Much of the debate between using siRNAs and hairpins relates to the exact type and goal of the study in question. Choosing siRNAs for most experiments is the easier way to go, notes Cold Spring Harbor’s Hannon, but for certain cases, the hairpin’s unique advantages make it the better option. “With the hairpin you can induce silencing that’s permanent,” he says. For research examining silencing in the long run, siRNAs won’t do the trick — their effect wears off after a while. Tuschl, meanwhile, argues that the hairpin effect isn’t as permanent as one might think: “When your cells die, you don’t have any advantage for expression over synthetic RNAs,” he says.

For many ex-periments, it’s enough to see the silencing take place in two to three days, which is the strength of the siRNA. Tuschl says using those in certain slow-division cell types such as hepatocytes or non-dividing primary microphages means that silencing effects can last as long as two weeks.

Hairpins, on the other hand, are like the Energizer bunny — they keep going and going. Ambion Senior Scientist David Brown says the beauty of the vector-based system is that it “allows you to generate stable cell lines. … You have these on a vector that’s integrated and they can express forever.”

Another particular use for the hairpin, Scaringe predicts, will be as a therapeutic. “It may be that making one compound, a single-strand 45- or 50-mer, is easier in the [FDA] regulatory process than two strands that have to be brought together to make a third product,” he says. “[Hairpins] may provide an advantage in the development of a therapeutic RNA.”

Delivery is another issue that can make the decision between siRNAs and shRNAs an easy one. For your run-of-the-mill experiment using cultured cells, Hannon says, siRNAs are the way to go. But for in vivo research, disease modeling, or certain cells that don’t react well to transfection, hairpins are likely the answer. Mike Hemann, a postdoc fellow at Cold Spring Harbor Lab, has been working on modeling lymphomagenesis in mice. “For us, siRNA strategies wouldn’t work,” he says. “We really needed to be able to stably modify our target cells and put the lymphoma back into a mouse and have the repression last for a number of weeks.” The particular cells in question were also problematic. “We needed to be able to use retroviruses to encode these hairpin constructs because we can’t get good transfection,” Hemann says. “We could only target [the cells] by viral integration.”

It’s not an uncommon problem, according to Tuschl. Primary muscle cells, such as those in cardiac muscles, don’t react well to the often-toxic transfected siRNA. For those cells, you have to “turn to the viral vector” by using hairpins, he says.

For most standard experiments, though, “you would pick a cell line that’s robust,” Tuschl says. “In this case you would use synthetic RNA.”

Ambion’s Brown points out that hairpins can also be used with transfection. “You can go ahead and generate a hairpin RNA outside of cells and transfect [it] into cells just like you would with an siRNA,” he explains. But they don’t seem to work as well that way: “We’ve done direct comparisons,” Brown says, “and we find that the siRNAs are significantly more active” than hairpins introduced in this fashion. “In our hands, that’s the least effective method of knocking down target gene expression,” he adds.

Ease of use

Even Greg Hannon, king of the hairpin realm, says that for basic experiments with cultured cells, “the easiest thing to do is order an siRNA.” Hairpins are widely acknowledged to need more attention and to be more complex than their siRNA cousins.

But that’s not to say hairpins are impossible. Once an shRNA is designed, “it’s just a plasma transfection protocol, or a retroviral infection protocol,” Hannon says. “These are very common techniques.”

So common, actually, that Dharmacon’s Scaringe believes scientists’ comfort level with using vector-based techniques has greatly contributed to the popularity of the hairpin. “Most of the users are cell biologists. They’re familiar with vectors, they may have access to make their own vectors and their own hairpins. But most labs don’t have their own synthesizers,” he notes. Despite the popularity of synthetic siRNAs, he says, many scientists still prefer to use vectors with hairpins because they’re more familiar with them.

But the design of the sequence for the vector may cause some problems. “There are more constraints,” Scaringe says. “It’s best to start with a G. Some sequences transcribe better than others.” Though his company is working on algorithms that would improve selection of hairpins, it’s still not as easy as plugging in a sequence and popping out a synthesized siRNA.

There are other design challenges, too. Hemann at Cold Spring Harbor has worked closely with Hannon’s lab to develop a hairpin strategy that would work for his research. Factors included choosing the best retrovirus, best placement of the hairpin within the virus, and length of the hairpin sequence. Hannon’s team has gone with 29-base hairpins, unlike many other groups, which more often use 19- or 20-base shRNAs. “One of the nicer things about having a longer hairpin is actually when the hairpin is recognized by [the Dicer enzyme] and cleaved, it generates a number of siRNAs, which allows us to have a potentially greater range of effectiveness against a target gene,” Hemann says.

Tuschl says interferon problems cause trouble for both techniques. With a high enough concentration of hairpins, the sense strand from one hairpin could fold onto the antisense strand of another, he says. “That could dimerize into the interferon,” he says. But the main disadvantage of the siRNA — having to use a transfecting reagent — “can also lead to interferon toxin,” a cell response to invaders that renders RNAi useless.


Debates over which method is more reliable are largely speculative in the nascent RNAi world. Still, leaders in the field see certain factors pushing for one technique or the other.

Reproducibility is one such issue, according to Tuschl. Because vector-based hairpins start out within the nucleus, they have to be expressed at a high enough level to work their way out into the cytoplasm to have a silencing effect, he says. “There are no well-defined rules for how you do this in a reproducible manner,” he says. Meanwhile, because siRNAs start farther out, eliminating the first steps necessary to turn on a hairpin, being able to redo an experiment with the same results is much more straightforward, he contends.

That links to what Scaringe considers the confidence factor. The chemical synthesis method “is more developed,” he says. “Confidence in the specificity and activity is relatively much higher” than in the hairpin systems.

While proponents of hairpins find them veryeffective, Tuschl says finding the right hairpin takes a lot more fishing than finding the right siRNA. “People have to screen 10 hairpins to find one that’s really good. With synthetic RNA it’s more like one in three, two in five,” he says. But he acknowledges that this isn’t a huge drawback for the hairpin method: “If you make 10 hairpins or two hairpins, it’s basically the same amount of time.”


Of course, no scientist can easily ignore cost differential when deciding which RNAi method to use. Figures from vendors such as Dharmacon and Ambion vary depending on whether scientists ask for plain old siRNAs or ones chosen with the special design algorithms that are supposed to increase the effectiveness of the strand, but in general, the average cost runs around $200 to $250 for a single siRNA duplex, according to Dharmacon’s Scaringe and Ambion’s David Dorris.

Scaringe says that vector-based hairpins cost anywhere from $50 to $150 for materials, and then each scientist has to factor in his own expenses for labor and time.

Most seem to agree that synthesis is the cheapest way to go. “There’s nothing that beats the synthetic RNAs in terms of cost,” according to Tuschl.

But Larry Wang, a scientist in business development at RNAi supply vendor GenScript, argues that hairpins can wind up being less expensive because “this is a one-time deal,” he says. Once you order your plasmid, “you can grow as much as you like in your lab. For synthetic, you have to keep on ordering.”


No matter how you look at it, either way of using RNAi has had — and will continue to have — a dramatic impact on the field. “The technology is fantastic. It’s completely changing the kinds of experiments that we can do and that we’re proposing,” says Cold Spring Harbor’s Hemann.

And while no one’s decisively picking one method over another, a number of people see the beginnings of a trend toward researchers using the techniques in conjunction with one another. Naturally, that’s a path that could lead to more sales for vendors, who are encouraging it.

“Do I use vectors or do I use chemical synthesis?” posits Scaringe. “We think the better question is, ‘How do I use them together?’”

David Brown at Ambion has seen the same trend. “A lot of people have siRNAs and vectors that target the same sequences and they use them for what [works best for each study].”

Head to head, how RNAs stack up
  siRNA Hairpin

About $250 per duplex

$50-150 for materials, plus labor and time

Up to two weeks

Permanent, if integrated well
Cell types

Robust cell lines

Primary cells where siRNA is too toxic
Time needed

Four or five days from design to experiment

About four or five days to create plasmid; two or three more for QC; three weeks for cell line development, if needed
Nature of study Most experiments

In vivo, especially

The Scan

Pfizer-BioNTech Seek Full Vaccine Approval

According to the New York Times, Pfizer and BioNTech are seeking full US Food and Drug Administration approval for their SARS-CoV-2 vaccine.

Viral Integration Study Critiqued

Science writes that a paper reporting that SARS-CoV-2 can occasionally integrate into the host genome is drawing criticism.

Giraffe Species Debate

The Scientist reports that a new analysis aiming to end the discussion of how many giraffe species there are has only continued it.

Science Papers Examine Factors Shaping SARS-CoV-2 Spread, Give Insight Into Bacterial Evolution

In Science this week: genomic analysis points to role of human behavior in SARS-CoV-2 spread, and more.