Prediction of active nodes in the transcriptional network of neural tube patterning

Abstract

A transcriptional network governs patterning in the developing spinal cord. As the developmental program runs, the levels of sequence-specific DNA-binding transcription factors (SSTFs) in each progenitor cell type change to ultimately define a set of postmitotic populations with combinatorial codes of expressed SSTFs. A network description of the neural tube (NT) transcriptional patterning process will require definition of nodes (SSTFs and target enhancers) and edges (interactions between nodes). There are 1,600 SSTF nodes in a given mammalian genome. To limit the complexity of a network description, it will be useful to discriminate between active and passive SSTF nodes. We define active SSTF nodes as those that are differentially expressed within the system. Our system, the developing NT, was partitioned into two pools of genetically defined populations by using flow sorting. Microarray comparisons across the partition led to an estimate of 500-700 active SSTF nodes in the transcriptional network of the developing NT. These included most of the 66 known SSTFs assembled from review articles and recent reports on NT patterning. Empirical cutoffs based on the performance of knowns were used to identify 188 further active SSTFs nodes that performed similarly. The general utility and limitations of the population-partitioning paradigm are discussed.