Single calponin homology domains are not actin-binding domains

In the early 1980s, the oncogene field was in transition. The biological properties of more than 20 oncogenes had been described in animal tumor models and cell transformation assays. Attention was shifting to the question of how these genes actually function. In a few cases, it was obvious. The v-Sis protein (encoded by the transforming gene of simian sarcoma virus) was shown to be related to platelet-derived growth factor and erbB (the oncogene from avian erythroblastosis virus) was found to encode a receptor for epidermal growth factor. The transforming Ras protein was known to bind GTP. This put Ras in a class with the classical signal transducing G-proteins, elongation factors and tubulin but did not clarify its mechanism of action. The biochemical analysis that was needed to answer such questions was still in its early stages when I decided to enter the field. Two main questions were being addressed: first, how do oncogenic proteins differ from their normal cellular counterparts? This question involved comparison of biochemical properties of normal and mutant proteins — a relatively logical process, and one that was made practical by the development of recombinant methods of producing these proteins. Second, how do these proteins transform cells? This was more of a guessing game. For Ras, it meant a long and relatively fruitless search for proteins that interact with Ras in its active state, and could therefore be Ras ‘effectors’. It was clear that Ras plays a major part in human cancer; by 1983, activating mutations in ras genes (single amino acid changes that seemed to make the mutant Ras proteins overactive) had been described in a variety of cell lines and tumors. (And this was before PCR had even been invented.) To me, with a degree in biochemistry and a job in a biotechnology company, it seemed an ideal moment to apply biochemistry to the question of Ras function. This was a real turning point for me. In 1984, I wound down my work on mammalian expression systems and characterization of β-interferon and geared up to enter the Ras field. The attention of my group was focussed on the biochemical properties of normal and oncogenic forms of Ras proteins produced in Escherichia coli. We were inspired and encouraged by four papers that revealed a dramatic difference between the two: the normal Ras protein could hydrolyze GTP to GDP; an oncogenic mutant (glycine to valine at codon 12), could not. This lead to a beautifully simple model. Ras proteins are simple switches. They are on when bound to GTP, and off when bound to GDP. Normal Ras turns itself off by converting GTP to GDP through its built-in GTPase activity; oncogenic Ras is stuck in the on position because its built-in GTPase is defective (see Figure 1). This brilliant model was inspiring but also provocative because it raised further questions. Most troubling to us was the fact that the mutant protein did, in fact, retain some GTPase activity: it was only about 8-fold less active than wild-type Ras. This hardly seemed sufficient to account for its awesome transforming power. Moreover, the mutant we were working on, the aspartate-12 mutant, was just as powerful at transforming cells yet it was only about 3-fold less active than wild-type Ras in hydrolyzing GTP. We therefore started to wonder whether the GTPase model needed some modification. A strong prediction of the model was that normal Ras should be in the off state (bound to GDP) in normal cells, whereas the oncogenic mutant should mostly be in the on state (bound to GTP). Using various technical tricks and an awful lot of radioactive phosphorous, we confirmed this. But in the process, we discovered that the aspartate-12 mutant was fully loaded with GTP, just like the valine-12 mutant, even though its GTPase activity as measured in the test tube was not much different Magazine R673