Using RNAi-based screens may help better identify drug targets, especially for cancer indications, but researchers must be aware of certain pitfalls associated with the technology that could affect outcomes, according to presentations made at this year’s American Association for Cancer Research annual meeting.
It is widely known that cancer therapeutics are only effective in a relatively small portion of patients, a dynamic that creates the need for better drug target-identification techniques.
While RNAi-based screens may help to address this issue, researchers must develop strategies to make the most of the gene-silencing technology.
During a session on siRNA screens at AACR, officials from two companies with experience with RNAi screens — Merck subsidiary Rosetta Inpharmatics and Cenix BioScience — offered a glimpse into how they approach this kind of project.
Merck at Work
“For marketed [cancer] drugs — I’m talking about the Iressas to the Avastins to the other … billion-dollar drugs that are out there — they’re working in about a quarter of the patients,” Stephen Friend, executive vice president and franchise head of oncology at Merck, said during the session, which was webcast this week.
“If there is no other reason to look for good targets, it’s that the … drugs that we have today are really ones that are not bringing forward solutions that are meeting patient needs,” he added. “Is there a way to better validate those early targets?”
For Merck, a company that has embraced RNAi for both research and therapeutic applications, that “better way” includes using genome-wide siRNA screens, according to Peter Linsley, vice president of research at Rosetta Inpharmactics.
“The discovery of [RNAi] in mammalian cells … has revolutionized the study of functional genomics in human cells,” he said during a presentation at AACR. “We’re now about to do genetic screens in human cells, which was a dream of many people for a long time.”
Operating under the umbrella of a big pharma, Rosetta favors a “parallel approach” for RNAi screening in which researchers treat each gene with an siRNA in individual wells and compare the whole genome side by side in one experiment, Linsley said.
Though resource-intensive, this approach “is one pharma favors because the human genome is relatively small compared to the size of chemical screens pharma is used to doing.”
While a chemical screen “might be a million or more compounds, the human genome is only 20,000 or so genes, which translates to 200 or so 96-well plates,” Linsley noted. “So for pharma, this approach is well within [their capabilities] and the way [they] like to do things.”
Working in conjunction with Merck’s automated biotechnology facility in North Wales, Penn., Seattle-based Rosetta is able to complete a genome-scale screen in cultured human cells in around one or two months, Linsley said.
Yet, a genome-wide screen may not always be the answer — at least not at first, he added. So Merck has adopted “a rapid mini-screen approach to optimize [those] screens that we will later push on to genome-scale screens.”
In 2006, Rosetta ran more than 100 mini-screens of 300 to 3,600 genes, Linsley said.
“Some of [the screens] we kill because we can’t make them work, [in] some of them we find all we need to know [without taking them further], and a few of them we advance to genome-scale screens,” he said.
Overall, “we found that this combination of mini-screens and a few genome-scale screens done at different centers is actually a very good strategy.”
Of particular use have been so-called “drug-enhancer/suppressor screens,” which look for genes whose expression can enhance or suppress the potency of specific cancer drugs, Linsley noted.
“In mammalian cells … [these screens] are really powerful tools to identify pathways, which lead us to new targets,” he said. “And understanding the pathways leads us to develop hypotheses for how patients can be treated with drugs.”
Cenix, meanwhile, incorporates a multipass approach to conducting RNAi screens, according to CEO and CSO Christophe Echeverri.
In the first pass, “you’re usually dealing with the scale of thousands of genes, [a technique that] is usually devoted strictly to detection,” he said.
“The discovery of [RNAi] in mammalian cells … has revolutionized the study of functional genomics in human cells. We’re now about to do genetic screens in human cells, which was a dream of many people for a long time.”
“We try to make sure we make it as inclusive as possible … so we use the siRNAs at quite high concentrations, we use them individually for each gene, and we really try to make sure we have the most sensitive possible detection,” Echeverri explained. “We know that under these conditions we’re going to generate a lot of false positives, but that’s fine because the point of this is to reduce the scale of the problem to something that is more manageable.”
The second screening stage focuses on the positive results from the first stage. “This is where you start cleaning out issues of specificity [and] making sure things are reproducible,” Echeverri said. In this step, researchers “typically” repeat the experiments with a new batch of siRNAs to [guard] against non-specific effects, he said. The company then tests each positive gene — even genes that were positive with only one of the three siRNAs used, he added.
“We usually expect a very high attrition rate” during this stage, he said.
Finally, Cenix conducts a “final confirmation where we make sure to correlate the phenotype severity with the target silencing level that gives us our ultimate confirmation that what we’re looking at really is related to a loss-of-function phenotype,” Echeverri said.
Despite the seemingly straightforwardness of the screening process, Echeverri warned of a number of small issues that, if not addressed, can throw a wrench into the gears of a well-designed screening experiment.
Aside from the use of positive and negative siRNA controls, “you really want to monitor all sources of variation — well positions within each plate, inter-plate variation, inter-experiment variation, inter-operate variations are [all] good to keep in mind,” he said. “This is the kind of thing you want to do to convince yourself that the kind of data you’re generating really is authentic and clean.”
Although the kind of large-scale automation employed by Merck in its genome-wide screens is useful, Echeverri stressed that systems for handling the data coming out of the research are just as important.
“People usually think a lot about lab automation, liquid-handling robots, that sort of thing — and that’s important,” he said. “However, the other half of the equation is the computing power … and computing infrastructure. You do need to think about putting in place a good laboratory information-management system that’s well-adapted for the kind of experimental workflows that you’re going to be working with.
“You want to have a serious relational database system and you want good, serious image-analysis systems to be able to convert the images into … numerical data … if you’re using microscopy-based read-outs,” he said.
Additionally, “in all these studies, you should not consider the kinds of tests you need to do for establishing RNAi specificity as something you’ll do as a follow-up to a screen,” Echeverri added. “This is something that is part of the screen and something that should be considered always as such.”