Unbiased discovery of interactions at a control locus driving expression of the cancer-specific therapeutic and diagnostic target, mesothelin.

Although significant effort is expended on identifying transcripts/proteins that are up-regulated in cancer, there are few reports on systematic elucidation of transcriptional mechanisms underlying such druggable cancer-specific targets. The mesothelin (MSLN) gene offers a promising subject, being expressed in a restricted pattern normally, yet highly overexpressed in almost one-third of human malignancies and a target of cancer immunotherapeutic trials. CanScript, a cis promoter element, appears to control MSLN cancer-specific expression; its related genomic sequences may up-regulate other cancer markers. CanScript is a 20-nt bipartite element consisting of an SP1-like motif and a consensus MCAT sequence. The latter recruits TEAD (TEA domain) family members, which are universally expressed. Exploration of the active CanScript element, especially the proteins binding to the SP1-like motif, thus could reveal cancer-specific features having diagnostic or therapeutic interest. The efficient identification of sequence-specific DNA-binding proteins at a given locus, however, has lagged in biomarker explorations. We used two orthogonal proteomics approaches--unbiased SILAC (stable isotope labeling by amino acids in cell culture)/DNA affinity-capture/mass spectrometry survey (SD-MS) and a large transcription factor protein microarray (TFM)--and functional validation to explore systematically the CanScript interactome. SD-MS produced nine candidates, and TFM, 18. The screens agreed in confirming binding by TEAD proteins and by newly identified NAB1 and NFATc. Among other identified candidates, we found functional roles for ZNF24, NAB1 and RFX1 in MSLN expression by cancer cells. Combined interactome screens yield an efficient, reproducible, sensitive, and unbiased approach to identify sequence-specific DNA-binding proteins and other participants in disease-specific DNA elements.