ESF Workshop ‘Proteomics: Focus on Protein Interactions’

Approximately 70 scientists met, from May 11–13 2001, in Villa Mondragone in the hills south of Rome to discuss strategies to describe the complete protein interaction network inside a cell. Very few of the participants still needed to be convinced that this is an essential step if we want to try to interpret the functional information contained in genomic databases. This was well accepted before the workshop started. The discussion revolved around how this can be achieved most effectively and which methods we should focus on if we want to get reliable biological information. Most of the high throughput methods that are currently used in large genomic protein interaction projects were represented by at least one of the 17 invited speakers. For the sake of simplicity the workshop presentations were divided into four sessions encompassing genetic methods, protein and peptide arrays, mass spectrometry and bioinformatic methods. Pierre Legrain, whose presentation is reported in more detail on page 301, focussed on the critical comparison of the different approaches that have been utilized in recent large scale 2-hybrid interaction screenings. Particularly surprising, and perhaps disappointing, is the finding that two large projects that aimed at deciphering the complete protein interaction map in S. cerevisiae show only a 15% overlap and recapitulate no more than 13% of the published interactions detected by the community of yeast biologists. Pierre Legrain suggested that an approach based on the expression of protein fragments, instead of full-length proteins, might contribute to decreasing the number of false negatives, as demonstrated in the Helicobacter pylori protein interaction project. Andreas Pluckthun and Brian Kay described two alternative methods, ribosome display and phage-display, that have a genomic potential. Ribosome display, although still in a development phase, holds great promises since it offers the potential to screen a number of partners that is by three to four logs larger than conventional display methods. Panning of peptide repertoires of random sequence displayed on filamentous phage capsids, on the other hand, not only permits one to infer the identity of natural protein partners but also allows precise mapping of the interaction sites (reviewed on page 304). Furthermore, this approach provides leads to develop molecules that, by binding at high affinity to either partner, disrupt the formation of a protein complex in a cell. Genetic methods are selective, although a large-scale screen in an array format has been described. By this approach each single interaction is tested independently and problems due to selective growth disadvantage of specific clones may be overcome. The array approach can be better implemented when proteins or peptides are orderly spotted or synthesized on solid supports, for instance a cellulose membrane, or a glass slide, as in DNA array. The technology of protein chips is far from being as accessible to the non-specialised laboratory as DNA chip technology. The problems to be overcome range from the difficulties experienced in Comparative and Functional Genomics Comp Funct Genom 2001; 2: 296–297. DOI: 10.1002 / cfg.98