A biological screen is only as good as the validation experiments that follow. Long gone are the days when a microarray data dump (or data vomit, as I like to refer to it) made for a good publication, not to mention interesting reading. I still shudder from the memory of reading (or more accurately, staring blankly at) pages and pages of up or down-regulated genes with no context and no explanation to their interaction, involvement, or function in the problem at hand. It’s not enough to pull out novel players in an siRNA screen if the experiments to confirm and extend those results are not done. Screens and other techniques with the potential of providing large data sets are finally coming into their own (or perhaps they had reached “their own” long ago and I am just now catching on).

One of the biggest science stories of the past few years was the generation of iPS (induced pluripotent stem cells) from fully differentiated adult cells, providing an alternative to deriving embryonic (ES) stem cells from embryonic tissues. All it took was the expression of a handful of key genes encoding transcription factors important for the maintenance of cells in an undifferentiated, pluripotent, stem cell state. Since that key study, and the many that followed, the discovery and characterization of novel regulatory pathways has “bootstrapped” or built upon/extended the known network of factors, limiting both the search and the results.

Stephen Elledge, a professor in the Department of Genetics at Harvard Medical School, and colleagues, took a different approach to identify new pathways important for the maintenance of cells in an undifferentiated, pluripotent state, capable of self-renewal – the hallmarks of ES cells. Elledge’s group used an siRNA screen to identify genes that when knocked down, lead to ES cell differentiation and loss of pluripotency. The investigators successfully navigated the murky waters of this genome-wide screen by performing a barrage of follow-up validation experiments which demonstrated that the genes they pulled out were in fact involved in and important for the maintenance of ES cells in an undifferentiated form. The authors narrowed down the 148 candidate genes to 104 by ensuring that multiple siRNAs led to the same de-differentiation phenotype and that the targeted genes are in fact expressed in embryonic tissues and stem cell lines. Of those, the authors selected 8 genes to pursue further – I would have loved to be a fly on the wall during those discussions. How do you choose 8 out of 100? Really wish they had shown their reasoning in the paper.

Four divergent assays demonstrated that the 8 chosen genes were important for maintenance of ES cells, including changes in ES cell morphology and marker expression following knockdown of the chosen ones. Based on the strength of the phenotypes observed in these validation experiments, the authors narrowed their work further to two genes encoding the transcriptional regulators Cnot3 (I pronounce this as ‘snot’. I don’t know why) and Trim28. Both of these transcription factors were expressed at high levels in embryonic tissues and ES cell lines; expression levels decreased upon differentiation. The authors then identified 1669 binding sites for Cnot3 and Trim28 in the mouse ES cell genome, including the promoter regions of multiple pluripotency genes. Cnot3 and Trim28 were found to form a transcription network, distinct from those regulated by previously characterized pluripotency regulators regulating gene expression of a cluster of 326 genes.

Talk about follow through. This group covered their bases and made a screen, which at its core is nothing but data vomit, into a building block for a new avenue in iPS and ES biology. They defined one piece of their data, leaving the rest for others to pursue. Sometimes anal retentiveness (is that a word) is really not a bad thing.

–

Hu, G., Kim, J., Xu, Q., Leng, Y., Orkin, S., & Elledge, S. (2009). A genome-wide RNAi screen identifies a new transcriptional module required for self-renewal Genes & Development, 23 (7), 837-848 DOI: 10.1101/gad.1769609