CRP-BINDING SITES - EVIDENCE FOR 2 STRUCTURAL CLASSES WITH 6-BP AND 8-BP SPACERS

While classifying protein binding DNA sequences of the type GTGN(x)CAC , based on the size of N(x) [Shumilov, Mol. Biologya (Engl. Transl.) 2 1 (1987) 168-187], we had previously found that the cyclic AMP recepto r protein (CRP)-binding sites found in the Escherichia coli genome are of at least two classes: (i) those with a conventional 6-bp spacer (N 6) and (ii) those with a potential 8-bp spacer (N8) [Barber and Zhurki n, J. Biomol. Struct. Dyn. 8 (1990) 213-232]. In this paper, we presen t the first experimental evidence that CRP binds to DNA with an N8 spa cer with relatively high affinity, as measured by gel electrophoresis of CRP-DNA complexes. We have tested two types of N8 spacers: A+T-rich and G+C-rich. Compared with the affinity of CRP for a reference site with an N6 spacer, the binding strength of CRP toward an A+T-rich N8 s equence is lower and that toward a G + C-rich N8 site is comparable. J ust like DNA sites with N6 spacers, those with N8 spacers utilize both halves of the symmetrical protein recognition sequences, TGTGA and TC ACA. Because of the increased number of nucleotides in the N8 spacer, the two recognition sequences in DNA will have an increased distance a nd a helical twist between them. These would cause displacement of the two recognition sequences with respect to the two symmetrically locat ed alpha-helices of the CRP dimer, if there is no change in the DNA co nformation. To explain the proper alignment of the recognition element s in DNA and CRP, we propose that the orientation of the two recogniti on elements in DNA is restored to the original orientation as in the N 6 spacer, and the physical distance in DNA between the recognition seq uences is decreased. The mechanism of such changes conceivably depends on the nature of the N8 sequence. A+T-rich N, sequences supposedly un wind in the center to realign the angular orientation of the two recog nition sequences and bend into the minor groove facing the protein to reduce the distance, whereas for the G+C-rich N8 sequences, a transiti on from B- to A-DNA would bring about the required unwinding and compr ession.