According to the World Health Organization, influenza infects nearly five million people every year, killing 250,000 to 500,000. Antiviral drugs are available to treat the virus, but some strains have developed resistance to these treatments, creating a need to identify new drug targets. One approach researchers are taking in the hunt for new targets is to perform genome-wide RNA interference (RNAi) screens to identify previously unrecognized host factors that are required for viral replication.

Although such RNAi screens have now been preformed for several viruses including HIV, dengue, and West Nile Virus (1–3), RNAi screens using influenza with human cells have remained a challenge.

Finding a simple, strong reporter

The major challenge in the development of high-throughput RNAi screens with influenza has been the advancement of an easy readout with a good signal-to-noise ratio. Normally, researchers insert a reporter gene into the virus genome—such as green fluorescent protein (GFP) or luciferase—to determine whether the viral genome has been taken up by host cells during the assay. But this approach has proven problematic when using the influenza genome.

“[Inserting a reporter gene] has been difficult to do for flu because it has a segmented genome,” said Megan Shaw, assistant professor of microbiology at Mount Sinai School of Medicine. Shaw is co-author on a recently published article in the journal Nature describing a set of human host factors necessary for influenza virus replication (4). “It’s very difficult to insert an extra open reading frame. That was definitely holding us back. We had to come up with a unique strategy to do that with our particular screen.”

An alternative strategy to tagging GFP is immunostaining with an antibody that targets the virus, a successful approach reported by researchers from the Max Planck Institute in another recent Nature paper (5). But because this technique requires a significant amount of work and time, Shaw and her colleagues decided to try a new approach.

They replaced the coding region of the influenza haemagglutinin (HA) protein with a luciferase reporter. This reporter enabled easy assessment of viral activity in infected cells using a luminometer.

“Luciferase is a fantastic reporter for high-throughput systems,” said co-author Sumit Chanda, associate professor at the Infectious and Inflammatory Disease Center at the Burnham Institute for Medical Research. “You get an amplification in signal, and you get really high sensitivity. The downside is that you have to take out a viral gene to put in the luciferase, which interferes with the replication of the virus.”

Replacing the HA coding region, however, turned out to be a very delicate task because the packaging signals have to be maintained.

“People have tried to do this in the past and have failed because they disrupted the packaging signals,” said Shaw. “If you disrupt the packaging signals, then you don’t incorporate that segment. That’s the reason that the influenza field was held back.”

Still, the eventual replacement of the HA coding segment created a new problem for the team: without HA, the modified virus would not infect host cells. HA is an essential envelope protein in the influenza replication process that helps bind influenza virus to the target cell and fuse the host and viral membranes. Without the ability to produce HA, the modified virus would not work in their RNAi screen.

So Shaw, Chanda, and their colleagues needed to develop a complementing system that would produce HA for the virus. After researching HA-expressing cell lines, the team amplified the virus in Madin-Darby canine kidney (MDCK) cells that were transfected with a plasmid to express HA. This technique provided the virus with the necessary HA to infect lung cells in the RNAi screen, and also enabled the researchers to use luciferase as their reporter with strong signal-to-noise ratio.

Reducing false positives: Toxicity and interferons

Even when an RNAi screen reveals what appears to be the identity of a host factor necessary for viral replication—and thereby a potential target for future drugs—researchers need to proceed with caution because false positives pose another challenge in RNAi screens.

One source of these false positives is toxicity. “Toxicity is the major issue,” said Shaw. “You have to realize that you’re never going to find a required host factor if it also kills the cell. If you knock it down, the cell dies, and, for sure, your virus is not going to grow either.”

Researchers are still experimenting with techniques to overcome the impact of toxicity in their RNAi screens. For Shaw’s team, titrating siRNA into their assay to produce different levels of host-cell toxicity enabled the team to measure its effects on viral replication. The team developed an algorithm to determine whether the toxic effect was solely responsible for the decrease in replication, or if the silenced gene also impacted replication.

“That is an important aspect we’re still working out,” said Shaw.

Interferons—another source of error—impede viral replication in infected host cells and increase infection resistance in healthy cells. siRNA transfection sometimes accidentally triggers the production of interferons in host cells, reducing virus growth and producing a false positive result. The solution to this is to monitor interferons, said Shaw. “We…made sure that none of our siRNAs are inducing interferons that would explain the [reduced viral] activity,” she said.

Assimilating the results: The next big challenge

Recently, three papers reported on the use of RNAi screens to identify human host factors necessary for influenza infection. The RNAi approach used by Shaw and her colleagues identified 295 genes that significantly reduced viral replication in the cells (4), while the researchers from the Max Planck Institute reported finding 287 human host cell genes influencing influenza virus replication (5). Furthermore, in a paper that appeared in the journal Cell (6), a team from the Broad Institute in Cambridge, Massachusetts validated 616 genes that significantly impacted at least one of three tested virus phenotypes. Though the results appear prolific, the amount of similar data across all three published gene lists is low, according to Shaw, which highlights a growing concern among researchers using RNAi screens.

“When you look at the gene level, there isn’t a very high level of overlap,” said Shaw. “That’s obviously a concern for people in the field. They don’t know how to interpret it.” To better understand these results, Shaw and Chanda believe that scientists need to take a step back and focus not on the genes, but rather on the pathways in which those genes are involved. ”It might be that my particular screen is picking up protein A in the pathway, and someone else’s screen is picking up protein C in the pathway,” said Shaw. “So it comes out being a different gene in the gene list, but it’s the same pathway. And that’s really the important information—that the flu requires that particular pathway.”

According to Chanda, this is a natural limitation of RNAi: because the technique is a genetic tool, it can only provide a list of genes that interact with the virus replication cycle, either directly or indirectly. And the majority of the RNAi screen results probably fall into the indirect category.

“If you take those indirect effects and compile them into pathways and genome phylogeny and groups of biochemical complexes, you start to see a bigger picture: that we’re really pulling out different pieces of the same puzzle,” said Chanda.

And this represents the next big challenge. Gene identification is step one, but successfully assimilating and interpreting data from various RNAi screens is a critical next step. Shaw and Chanda believe that further analysis of these data will lead researchers to a better understanding of how viruses interact with specific cellular pathways. From this understanding, new therapeutic approaches could be developed to treat infections from viruses such as influenza.

The data from these RNAi screens is just one piece to a much larger puzzle of how these viruses interact with the host cells during replication. Chanda envisions not only compiling and comparing data from RNAi screens, but also examining data from other approaches such as protein-protein interaction data, gene expression, and genetics data.

“Right now we have a parts list but no instructions on how these piece fit together,” said Chanda. “Bringing together the different approaches to interrogate phenotypes at the level of the genome will help support all of the different approaches. We’ll get a clear picture of what’s going on at the gene level.”

But before these analyses can be accomplished, RNAi approaches will need to be refined over the next few years. The standardization of screening protocols, reporters, and data submission will facilitate a greater understanding of how gene function contributes to the different pathways of cellular processes.

“This is what happens with a new field; [it’s] just like microarrays,” said Shaw. “As more and more people start doing it, you start to realize what type of analysis needs to be done to interpret it. It’s definitely a moving field.”

References

1. Brass, A.L., D.M. Dykxhoorn, Y. Benita, N. Yan, A. Engelman, R.J. Xavier, J. Lieberman, and S.J. Elledge. 2008. Identification of host proteins required for HIV infection through a functional genomic screen. Science 319:921–926.
2. Krishnan, M.N., A. Ng, B. Sukumaran , F.D. Gilfoy, P.D. Uchil, H. Sultana, A.L. Brass, R. Adametz, et al. 2008. RNA interference screen for human genes associated with West Nile virus infection. Nature. 455:242–245.
3. Sessions, O.M., N.J. Barrows, J.A. Souza-Neto, T.J. Robinson, C.L. Hershey, M.A. Rodgers, J.L. Ramirez, G. Dimopoulos, et al. 2009. Discovery of insect and human degue virus host factors. Nature 458:1047-1050.
4. Karlas, A., N. Machuy, Y. Shin, K.P. Pleissner, A. Artarini, D. Heuer, D. Becker, H. Khalil, et al. 2010. Genome-wide RNAi screen identifies human host factors crucial for influenza virus replication. Nature. 463:818-822.
5. König, R., S. Stertz, Y. Zhou, A. Inoue, H.H. Hoffmann , S. Bhattacharyya, J.G. Alamares, D.M. Tscherne, et al. Human host factors required for influenza virus replication. Nature. 463:813–817.
6. Shapira S.D., I. Gat-Viks, B.O. Shum, A. Dricot, M.M. de Grace, L. Wu, P.B. Gupta, T. Hao, et al. 2009. A physical and regulatory map of host-influenza interactions reveals pathways in H1N1 infection. Cell. 139:1255–1267.