The interaction of estradiol-receptor protein with the genome: an argument for the existence of undetected specific sites.

In extracts from rat and calf uterus, the steroid hormone 17 beta-estradiol stimulates the binding of its specific receptor protein to DNA. This interaction appears to be of low affinity (half of the estradiol-activated, 5S receptor bound at 300-400 mug/ml DNA) and nonspecific with respect to DNA base sequence. No binding to double-stranded RNA is observed. These findings are consistent with several in vivo observations. In particular, when the cytoplasmic receptor protein binds hormone, it migrates to the cell nucleus to an extent consistent with its affinity for DNA in vitro, and this in vivo nuclear binding is uniform and nonsaturable in the testable range (to greater than 3 times 10-4 sites per cell). The level of biological response appears to parallel the hormone dose up to these high levels of receptor binding. How are these observations to be reconciled with the prevalent view of steroid receptors as gene control proteins regulating transcription at specific loci on the genome? Our model is based on an analogy with the DNA binding properties of the E. coli lac repressor protein. We believe that the estradiol receptor exerts its effect by binding to a small number of high affinity sites on the genome, while also having a finite low affinity for nonspecific DNA sequences. These nonspecific loci, because of their vast number, completely mask the presence of the high affinity sites. We estimate that up to 10-3 specific sites, with affinities in the range 10- minus 8 minus 10- minus 10 M, could exist without being detected by bulk binding assays currently in use. However, alternative approaches should allow detection of these sites, and some of these are suggested.