One cannot measure an information content from a single sequence. Dyad
symmetries in DNA (palindromes) are an exception because both the sequence of
the palindrome and its complement are available. This allows us to estimate
how much information appears in the *lac* operator
(Beckwith, 1978; Goeddel *et al*., 1978;
Sadler *et al*., 1983a). Gilbert and Maxam (1973) found that the
tetrameric *lac* repressor protein protects 24 base pairs
from DNase digestion.
This is a region from -13 to +10, where the zero is the central base. More
recently, exonuclease III digestion gave the range -14 to +16
(Shalloway *et al*., 1980).
To analyze the site
we extended the range -16 to +16 by 5 bases on both
sides (Fig. 5).
This range includes the
"extended operator" (Dickson *et al*., 1975;
Heyneker *et al*., 1976). As with
other operators, the sequence was compared to its complement using the program
Rseq. The central position was included,
giving
*R*_{sequence} = 19.2 bits per
site. Because there are only two examples, there is a large sampling error.
If there is only one functional *lac* repressor
binding site in the *E. coli*
genome, then
*R*_{frequency} = 21.9 bits per site.
"Pseudo"-operator sequences
exist for which there is no known function
(Reznikoff *et al*., 1974; Winter and
von Hippel, 1981). If we include the strong secondary "pseudo"-operator,
*R*_{sequence} = 16.22.6 and
*R*_{frequency} = 20.9 bits.