To study the mechanisms that control epithelial commitment and differentiation we have used undifferentiated HT-29 colon cancer cells and a subpopulation of mucus secreting cells obtained by selection of HT-29 cells in 10−6 M methotrexate (M6 cells) as experimental models. We isolated cDNAs encoding transcripts overexpressed in early confluent M6 cells regarding steady-state levels in HT-29 cells by subtractive hybridisation. Fifty-one cDNA clones, corresponding to 34 independent transcripts, were isolated, partially sequenced by their 5′ end, and classified into four groups according to their identity: transcripts that included a repeated sequence of the Alu family (10 clones, among them those encoding ribonucleoprotein RNP-L and E-cadherin), transcripts encoded by the mitochondrial genome (nine clones), transcripts encoding components of the protein synthesis machinery (23 clones, including the human ribosomal protein L38 not previously cloned in humans) and nine additional cDNAs that could not be classified in the previous groups. These last included ferritin, cytokeratin 18, translationally controlled human tumour protein (TCHTP), mt-aldehyde dehydrogenase, as well as unknown transcripts (three clones), and the human homologues of the molecular motor kinesin KIF3B and of the ser/thr protein kinase EMK1. Spot dot and Northern blot analyses showed that ser/thr protein kinase EMK1 was differentially expressed in M6 cells when compared with parental HT-29 cells. Steady-state levels of EMK1 were higher in proliferating, preconfluent, M6 and HT-29 cells than in 2 days post confluence (dpc) and 8dpc M6 and HT-29 cells. Transcripts that included an Alu repeat were also shown to be differentially expressed and accumulated in differentiating M6 cells when analysed by Northern blot. The significance of the transcripts cloned is discussed in the context of the commitment and differentiation of the M6 cells to the mucus secreting lineage of epithelial cells.