Skip to main content
Premium Trial:

Request an Annual Quote

Q&A: MSKCC's Nikolaus Schultz on siRNA Screens and Off-Target Effects

Premium

schultz.jpgName:
Nikolaus Schultz

Position:
Manager, biocomputing/cancer genomics, computational biology center, Memorial Sloan-Kettering Cancer Center

Background:
• Postdoc, computational biology center, Memorial Sloan-Kettering Cancer — 2004-2009
• PhD, biochemistry, Freie Universitat Berlin — 2004
• Teaching faculty, physiology summer course, Marine Biological Laboratory — 2001-2003

Researchers from Memorial Sloan-Kettering Cancer Center this month published the results of an RNAi screen designed to identify novel components of the transforming growth factor-beta pathway, which is implicated in a number of cancers, in a human cell line.

Although they did not find any new players in this pathway, their data indicates that multiple microRNAs are acting on the type II TGF-beta receptor, while an unusually high number of off-target effects suggested that “misleading results in siRNA screens using large libraries with single-assay readout is substantial.”

Gene Silencing News spoke with one of the paper's co-authors, Nikolaus Schultz, about the findings.

You work in Chris Sander's lab. What's the focus of the research there?

It's mostly a focus on the different pathways that are dysregulated in cancer. He's always had a strong link in the field of microRNAs, so doing an siRNA screen to look for components in a cancer system was an ideal combination of all the different skills of the lab. The real goal [of the work that led to the paper] was to identify new, unknown members of the TGF-beta pathway.

Can you give an overview of the experiments conducted?

We had in-house an siRNA library and we were looking for a way to utilize it to find new pathway genes. It is quite an old library now — it was six years ago when we started doing the screen. It was synthesized and HCl purified here, but it was never completed to cover the whole genome. It's a 6,000-gene library with three to five siRNAs per gene — 20,000 individual siRNAs, roughly.

So we developed an assay for the [transforming growth factor-beta] pathway, [which is involved in a number of cellular processes, including proliferation, apoptosis, and differentiation], knowing that one of the most reliable downstream effectors of the pathway is the [transcription factor] SMAD2, and when that goes into the nucleus you know it's phosphorylated. So we tagged SMAD2 with [green fluorescent protein] and found that we can reliably stimulate cells with TGF-beta and within minutes see nuclear translocation of the reporter. The model ... was an epithelial cell line — a good model of cells that are typically deregulated in cancer, and it was easy to transfect with siRNAs and easy to grow.

We knocked down genes one by one, and were looking for a knockdown of a gene that would give us the phenotype of a blocked nuclear translocation. So in the positive control, if you knock down one of the two TGF-beta receptors, you block phosphorylation and thereby block translocation of the GFP SMAD2 into the nucleus, and you'd see that as a hit. When we ran this screen against the 6,000 genes, we noticed that 1 percent of the library showed a decrease in nuclear translocation, suggesting that we had hit pathway members that play a role in phosphorylation, or at least nuclear translocation.

We had 200 candidates if you focus on the top hits, but what struck us right away was that there was no redundancy in the hits, meaning that even though we had, per gene, three to five siRNAs in the library, not a single gene showed up with two or more siRNAs in the hit list. [This suggested] that something wasn't right, but given that [RNAi] was a pretty new field and the library design rules weren't perfect, we thought that there could still be something in there and we'd try to find the new gene. We looked and, synthesizing additional siRNAs targeting the same genes, we could never repeat the phenotypes. That's when we started getting suspicious and started wondering about off-target effects. We asked, “What are the known players in the pathway, and could those be affected by the off-target effects?” We found that the TGF-beta receptors, in particular the Type II TGF-beta receptor, seemed to be the ones that were always knocked down when we had this phenotype.

Ironically, we had one gene, PRKACA, for which we purchased an additional siRNA from Dharmacon, which gave us the same phenotype. So we thought, “Now we have two independent siRNAs that give us the same phenotype.” Later we found out that, even though we had two, both of those … had off-target effects. That little finding didn't make it into the paper, but it's a nice example of how, even when you think you have two separate siRNAs showing the same phenotype, you may still be dealing with an off-target effect. It's not likely, but it's still possible.

So you realized that you'd have to do something to tease the off-targets apart from what you were looking for.

Exactly. We had a big problem with off-target effects, but thought, “All we need is one siRNA that's not off target, and decided to test the top 200 hits, one by one, to see whether they affected the TGF-beta receptors.

Instead of doing this tediously by q-PCR, we decided to use a branched DNA assay — the Panomics QuantiGene assay — which allowed us to scan this much more quickly. So we repeated the transfection and used the assay to measure mRNA levels of the two receptors. We found that virtually all of [the siRNAs] did knock down TGF-beta receptors 1 and 2. The few that didn't turned out to be false positives — we couldn't repeat the effect on SMAD2.

By the end of it, we had not a single new member in the TGF-beta pathway, which was disappointing. [Laughs] But then we realized that what we actually had is something very interesting. So we turned the whole thing around and said, “We found that these really important tumor-suppressor genes, especially the type II TGF-beta receptor, might be super-sensitive to regulation by small RNAs, be they siRNAs in our system or microRNAs in vivo — and there was already one link between the type II TGF-beta receptor to a specific microRNA, microRNA-20a, and a specific cancer type. We also tested a few additional microRNAs and showed that they titrate the levels of the type II TGF-beta receptor, even in the cells we looked at. Of course, the other thing we had was a new resource to study sequence motifs that are involved in what we quickly realized were microRNA-like off-target effects.

[In the end], we don't call into question the results of carefully exercised screens [performed by other groups], especially if they find multiple hits for the same gene and can verify them. … But it [raises a red flag over] the everyday use of siRNAs. For instance, a lot of people still use very few controls and are not aware of off-target effects. But here we show that 1 percent of the library [has measurable off-target effects on either Type I or II TGF-beta receptor].

We were surprised by the extent of the off-target effects. … Every siRNA has off-target effects, but you wouldn't expect by chance that 1 percent hits your gene of interest. There have been reports of off-target effects in screens, but never to this extent.

The Scan

Missed Early Cases

A retrospective analysis of blood samples suggests early SARS-CoV-2 infections may have been missed in the US, the New York Times reports.

Limited Journal Editor Diversity

A survey finds low diversity among scientific and medical journal editors, according to The Scientist.

How Much of a Threat?

Science writes that need for a provision aimed at shoring up genomic data security within a new US bill is being questioned.

PNAS Papers on Historic Helicobacter Spread, Brain Development, C. difficile RNAs

In PNAS this week: Helicobacter genetic diversity gives insight into human migrations, gene expression patterns of brain development, and more.