Review of “A predictive computational framework for direct reprogramming between human cell types” from Nature Genetics by Stuart P. Atkinson

Blind alleys, one-way streets, and a lot of negative results. Reprogramming experiments come across such problems with unfortunate certainty and, while they are invaluable to the final result, this time could be much better spent. Jose M Polo and Julian Gough sought to reduce the time spent and increase the likelihood of success in direct reprogramming (or transdifferentiation) experiments by creating a predictive computational system which can indicate which reprogramming factors are likely to induce the conversion of one cell type into another.

Their results, published in Nature Genetics, could save both money and time and, in doing so, revolutionize the field of direct reprogramming and boost clinical aspirations for this process [1].

The computational framework described, known as ‘Mogrify’ to its creators, holds information regarding transcription factors (TFs) which discriminate cell lineages. This includes information on how each TF is regulated, how direct the action of a TF is, and its specificity. With all this information, Mogrify can choose which TFs are likely to combine to activate the pertinent regulatory networks.

After construction, the authors demonstrated Mogrify’s power by testing it against well-understood validated human cell conversions. Mogrify correctly identified:

• NANOG, OCT4, and SOX2 in human induced pluripotent stem cell (hiPSC) production from fibroblasts
• CEBPα and PU.1 in the conversion of B cells and fibroblasts into macrophage-like cells
• Transcription factors known to be important in the conversion of human dermal fibroblasts into cardiomyocytes and hepatocytes

The real test came when the authors attempted “new” cell conversions using TFs generated from Mogrify’s databases. However, Mogrify came through, and correctly identified:

• FOXQ1, SOX9, MAFB, CDH1, FOS, and REL in the conversion of fibroblasts into keratinocytes with the expected morphological and molecular characteristics
• SOX17, TAL1, SMAD1, IRF1, and TCF7L1 in the conversion of keratinocytes into micro¬vascular endothelial cells (iECs) with the expected morphological and molecular characteristics

The authors note that finalizing conditions for cell conversions will still need trial-and-error experimentation as the data input to Mogrify is limited by what we currently understand, their new program will give many a confident head start in reprogramming experiments. There are, however, some other important caveats that the authors identify. Firstly, Mogrify is not designed to identify factors which extinguish a given donor cells signature and, secondly, multiple other factors which can influence cell conversion (e.g. noncoding RNAs, small molecules, epigenetic factors, and signaling pathways) are not incorporated.

However, one cannot help see the usefulness of this network biology approach to transdifferentiation and, as a final bonus to the entire field, Mogrify itself is a free tool for all to use. So get over to www.mogrify.net and see what you can convert today!

References

1. Rackham OJ, Firas J, Fang H, et al. A predictive computational framework for direct reprogramming between human cell types. Nat Genet 2016;48:331-335.