Linking Genes to Pathways
Hyperactivity of the Wnt/beta-catenin signaling pathway has been associated with several forms of cancer. Furthermore, diseases such as osteoporosis and Alzheimer disease have been linked to decreased beta-catenin activity. A research team at Merck & Co. led by Michele Cleary, Ph.D., senior director of automated biotechnology, designed a lentivirus-based short hairpin (sh)RNA screen to identify genes involved in the regulation of Wnt/beta-catenin signaling.
The large-scale screen tested a 14,000-vector shRNA library against 5,000 human genes. The assay was based on a luciferase reporter of beta-catenin driven transcription that was developed in the laboratory of Randy Moon at the University of Washington.
The Merck team’s screening strategy involved pretreating the cells with an inhibitor of a naturally occurring protein complex containing GSK3β (a serine/threonine kinase) that acts to downregulate beta-catenin activity.
“We treated the cells with a dose of the GSK3β inhibitor that was just below the level needed to detect a change in reporter activity. By taking this approach, we were able to screen for genes whose RNAi-mediated loss synergizes with this low dose of GSK3β inhibitor to potently activate the reporter,” explains Dr. Cleary.
She describes two main advantages of using lentivirus delivery of shRNA versus transfection of siRNA: improved delivery efficiency in cell types that are difficult to transfect, and the ability of lentiviruses to integrate into the host genome and provide long-term gene silencing in dividing and nondividing cells. The Merck group developed protocols for preparing thousands of lentivirus constructs in parallel in microtiter plates at a reasonable cost.
From the screen they identified “new modulators of the beta-catenin signaling pathway, most notably dihydrofolate reductase (DHFR).” Follow-up studies showed that DHFR inhibition decreases GSK3β-mediated phosphorylation of beta-catenin, leading to accumulation of beta-catenin in cells.
Dr. Cleary points out that methotrexate, a drug used to treat inflammation, targets DHRF; GSK3β inhibition appears to diminish the inflammatory response; and the actions of methotrexate and the GSK3β inhibitor are synergistic. These observations, combined with evidence that patients with rheumatoid arthritis who are treated with methotrexate have an increased risk of certain cancers, highlight the potential for certain types of anti-inflammatory agents to activate a potent oncogenic pathway.
The key challenges in applying RNAi technology “lie in applying the appropriate resources to the sometimes extensive follow-up needed to understand how a particular gene feeds into the relevant pathways and how this information can help advance a drug discovery program,” says Dr. Cleary.
Particularly if it is not possible to screen in replicates—for example, “when working with large RNAi libraries or expensive assay readouts.” Follow-up screens to confirm hits and ensure on-target silencing are critical for a successful screening campaign, she adds.
The Translational Genomics Research Institute (TGen) is applying high-throughput, high-content RNAi to advance a systems-biology approach aimed at mapping perturbations across the transcriptome to molecular networks and using this information to support iterative testing and refinement of models of gene activity related to molecular and cellular processes.
The goal of this effort, says Spyro Mousses, Ph.D., director of the Center of BioIntelligence of TGen, is to develop “a more functional way of interrogating the human genome and assigning a trait to a gene.” The advent of siRNA libraries offered a means to look across the entire genome and identify groups of genes that converge on a particular phenotypic profile.
Key technology improvements will include continued miniaturization of assays beyond the current 384-well format, which will help further reduce the cost of screening, notes Dr. Mousses. He describes the ability to marry high-content and high-throughput RNAi screening as one of the most significant innovations enabling the simultaneous collection and assessment of multiple response parameters.
Advances in informatics, with the availability of more powerful mapping and knowledge mining tools, are facilitating data interpretation and helping researchers identify multiple genes linked to a phenotype and group them into pathways. This information has a range of applications, including target discovery, pharmacogenomics, probing cellular response to a drug, and biomarker discovery.
Dr. Mousses emphasizes the importance of “understanding the limitations of what inferences you can make from the data,” learning how to minimize off-target effects, artifacts, and false-positive or -negative results, and recognizing the risks of pooling siRNAs. “These limitations do not decrease the value of the technology,” he adds, and there are many strategic approaches for dealing with them.