Modelling and identification of transcription-factor binding motifs in human chondrogenesis

Chondrogenesis is a dynamical process of cartilage formation involving many transcription factors and transcription-factor binding motifs (TFBMs). Using a series of microarray-derived mRNA profiles for human chondrogenesis, a computational model was built to identify critical TFBMs and model-driven prediction was evaluated experimentally using a promoter competition assay. First, the model was used to approximate the observed mRNA levels using frequencies of potential TFBMs which appeared in the regulatory DNA sequences. Potential TFBMs were chosen from a random population of 5-bp DNA fragments, and the selected genes included chondrocyte specific genes such as type II collagen. Using the Akaike information criterion and the genetic algorithm, six sets of putative TFBMs was chosen from the 5'-end flanking region encompassing 125-, 250-, 500-, and 1000-bp sequences. The minimum model error was obtained in the model with the 500-bp regulatory region. Second, the promoter competition assay was conducted for the selective collagen genes. The assay supported the stimulatory role of a novel TFBM (5'-AGGGG-3') in induction of type II and type III collagens.