Two “what if” experiments

Three perspectives are better than one. And it has been over 15 years since the two papers under discussion were published--a long time to remember who was thinking what. So the three of us (Roger, Jun, and I) will have three separate goes at reconstructing the events. For over 10 years, and through the early 1980s, our lab had concentrated on gene regulation in phage lambda. By 1985 we had clear ideas for how lambda regulatory proteins bind DNA and turn transcription on and off. Would these ideas help us to understand how eukaryotic transcriptional regulatory proteins work? Unlike in the lambda case, eukaryotic genes are wrapped around histones to form nucleosomes; the DNA is sequestered in a nucleus; and, in higher eukaryotes, transcriptional activators can work "at a distance," i.e., when bound to sequences many hundreds, or even thousands, of base pairs from the target gene. It could have turned out, I guess, that the principles of gene regulation in higher organisms would turn out to be fundamentally different from those found in lambda. Perhaps a new mode of binding to DNA was required, for example; perhaps activation at a distance requires special functions not represented among the lambda regulators; and so on. I found these ideas unpleasant. And in any case, isn't it better to see how far what we know will take us than to assume new complexities? We picked the simplest eukaryotic example we could find: the Gal (galactose utilization) genes of yeast, the organism introduced a few years earlier to the lab by Lenny Guarente, Sandy Johnson, Bob West, and Roger Yocum. Addition of galactose to growing cells dramatically induces transcription of the Gal genes, an effect mediated by a single transcriptional activator, Gal4. Gal4, we learned, binds to DNA some 250 base pairs away from its target genes (Giniger et al., 1985; West et al., 1984; Guarente et al., 1981). Did Gal4 recognize DNA in some new way, i.e., in some way not illustrated by the lambda regulatory proteins? How did activation at a distance, in this case 250 base pairs, work? To understand our frame of mind (mine anyway), here is a brief summary of four key concepts extracted from the lambda studies as of 1986 (Ptashne, 1986).