The public arm of the public-private consortium responsible for developing a short hairpin RNA library targeting the entire human and mouse genomes has now published a research paper that validates the effectiveness of the library for conducting large-scale loss-of-function phenotypic screens.
The publication, which may serve as a roadmap of sorts for the application of the shRNA library in functional genomics and drug discovery, also demonstrates the power of combining such technology with high-content automated imaging.
As detailed in the March 24 issue of Cell, a group led by scientists from the Broad Institute of Harvard University, the Whitehead Institute, and MIT has developed a screen based on high-content imaging to identify genes required for mitosis in human cancer cells, and applied the screen to a set of 5,000 unique shRNA-expressing lentiviruses targeting 1,028 human genes.
The scientists behind the research are from the same team that formed The RNAi Consortium, a public-private venture established in early 2005 with the goal of generating genome-scale shRNA libraries and making them available to scientists worldwide. Members of this consortium include Harvard Medical School's Massachusetts General Hospital and Dana-Farber Cancer Institute, the Broad Institute, the Whitehead Institute for Biomedical Research, Bristol-Myers Squibb, Eli Lilly, Novartis; Sigma-Aldrich, and Academia Sinica-National Science Council in Taiwan.
"In some ways it's a manual for applying the library as well as an illustration of its effectiveness."
The set of 5,000 shRNA lentiviruses used in the research are a subset of the approximately 104,000 vectors targeting 22,000 total genes — about 12,000 in humans and 10,000 in mice — already developed by the RNAi Consortium, and available through Sigma-Aldrich and Open Biosystems.
"What's new is not explaining what we're making here, or that we're making it available through distributors," explained David Root, corresponding author on the Cell paper and director of the RNAi Consortium and the RNAi platform at the Broad Institute. "What's new is that we have characterized the performance of the library and applied it to a screen.
"It does serve to validate its effectiveness for screening, and it's also an explanation of how we have and how others can apply the library," Root added. "In some ways it's a manual for applying the library as well as an illustration of its effectiveness."
According to Root, the research is particularly notable because it is the first demonstration of using lentiviral RNAi vectors in an arrayed format for a relatively high-throughput screen of loss-of-function phenotypes in cells. Prior to this, he said, scientists have typically used basic transfection techniques or pools of viral vectors to screen for gene function in cells.
"One of the challenges to using this [shRNA] library in a powerful way — and certainly a way that people are interested in — is being able to do arrayed viral screening," Root said. "This is a lentiviral library, [which] isn't new in the sense that this library has been available for a year.
"However, being able to make virus in high throughput and screen with it shows that you can take advantage of the broad applicability of lentivirus to many cell types in an arrayed viral screening application," he added. "If you had to use it by transfection, you wouldn't reap the benefits of lentivirus in an arrayed screen."
To demonstrate the efficacy of their approach, the researchers developed a high-content imaging screen for mitotic index genes in human colon cancer cells. To do this, they added individual lentivirus vectors to HT29 cells in each well of 96-well plates, with each virus containing four or five distinct hairpin sequences targeting a specific gene.
"So you have multiple different hairpins targeting each gene, and you can easily make a virus with these four or five hairpins for each gene — in fact, we screened about 1,000 genes with 5,000 different hairpins," Root said. "Therefore you can do a viral screen and have some kind of confirmation of gene specificity by, say, looking for those genes that show phenotype with more than one hairpin for the target."
To assess the resulting phenotypes in the cancer cells, Root and colleagues fixed the cells and fluorescently stained them for phosphohistone H3, a common mitotic index marker. They then used a Cellomics high-content screening instrument and associated software to take automated images of the cells and extract mitotic index data.
"A byproduct of this was to see just how powerful it was to combine RNAi and high-content screening."
In an interesting side note, the researchers used the Whitehead Institute's CellProfiler software to extract DNA content histograms from the same primary screening images. CellProfiler is an open-access image-analysis software package (see CBA News, 07/04/05). It is expected to complement or compete with commercial image-analysis software packages, and the Cell study represents one of the more publicized examples of how it can complement existing packages.
"The high-content imaging wasn't one of the main focuses of this paper," Root said. "Our main goals were to determine how best this library could be applied, and to illustrate and determine its effectiveness in a screening situation.
"But a byproduct of this was to see just how powerful it was to combine RNAi and high-content screening," he added. "We really got a wide cross-section of phenotypes — as one often can with high-content screening — as a function of RNAi knockdown."
According to the researchers, the RNAi Consortium continues to add approximately 4,500 additional constructs per month to its current library of about 104,000 shRNA vectors. Detailed information on genes targeted in the library can be found at the RNAi Consortium shRNA library web site.
The RNAi Consortium shRNA library is distributed as bacterial glycerol stocks, plasmid DNA, or lentiviral particles by Sigma-Aldrich and as bacterial glycerol stocks by Open Biosystems, according to the RNAi Consortium website.
"Our purpose is to get this thing distributed rather than setting up a commercial enterprise with these guys," Root said. "We basically send them a copy of the library and they distribute under whatever terms they can. We want to establish good distribution channels to get it out there, and we'd like to be able to share the load."
— Ben Butkus ([email protected])