HENRY M. SOBELL

IMPLICATIONS IN MOLECULAR BIOLOGY

Soliton concepts provide a strong rationale for expecting coherent nonlinear excitations to extend over multiple base pairs to provide either transient or permanent conformational changes in specific DNA regions. These will always be present to a certain extent at normal thermal energies; however, their concentrations will depend on temperature, pH, ionic strength, hydration, extent of superhelicity and other thermodynamic factors.

The stability of a premelton is further expected to reflect the collective properties of nucleotide base sequences in extended DNA regions. Since the ease with which beta-structural elements form is correlated with the magnitude of localized base-stacking energies, base sequences with minimal base overlap (i.e., as occur, for example, in alternating purine-pyrimidine sequences) are favored, along with sequences that contain high A-T/G-C base ratios. The energetics in the kink and antikink regions are another important factor. The tendency of a premelton to localize within a given DNA region depends on the depth of the energy minimum in the central core region, coupled with the height and separation of the energy domain walls (i.e., that are associated with the kink and the antikink structures) on either side.

The presence of premeltons undergoing low amplitude breather motions within promoter regions can serve the important function of providing nucleation centers for site-specific DNA melting by the RNA polymerase enzyme, and perhaps, by other proteins necessary for gene activation.

Key evidence for the existence of premeltons in eukaryotic promoter regions is provided by the discovery that the intercalator, 1, 10- phenanthroline-copper (I), cleaves DNA at hypersensitive sites in naked relaxed circular DNA molecules containing the histone gene cluster of D. melanogaster, many of these located at the 5' ends of genes. Remarkably, this small molecule mimics the larger micrococcal nuclease in recognizing these sites, cleaving them with equal frequencies. Pancreatic DNase (I) cleaves these same sites (with similar frequencies), as well as other unrelated sites.

The addition of the single-strand specific DNA binding protein to these same DNA molecules made superhelical selectively melts DNA at many of these sites, as revealed by S1 nuclease studies in combination with electron microscopy. Important additional information is provided by studies of this same gene cluster in active chromatin, where the micrococcal nuclease cleaves hypersensitive sites at the 5' ends of genes, while the pancreatic DNase (I) cleaves hypersensitive sites at both the 3' and 5' ends of genes.

What determines the degree to which a given DNA site is hypersensitive to nuclease action?

I propose the central beta-structural element within premeltons to be the substrate recognized and cleaved by these agents. As already discussed, premeltons are expected to arise with varying probabilities and to have different lifetimes in different DNA regions. Premeltons arising with higher probability and having longer lifetimes will therefore be cleaved more frequently than those arising elsewhere.

Low amplitude breather motions within the premelton cause the central beta-structural element to alternate between its two different energy states. I propose the highest energy state (i.e., the high-energy beta-structural element) to be the substrate cleaved by either the micrococcal nuclease or 1, 10-phenanthroline copper(I), and the lowest energy state (i.e., the low-energy beta-structural element) to be the substrate cleaved by the pancreatic DNase (I).

Premeltons containing lower energies undergo smaller breather motions. These premeltons may not have enough energy to open the central beta-structural element wide enough to act as a substrate for 1,10- phenanthroline copper (I) or the micrococcal nuclease. Premeltons such as these, however, can remain as substrates for the pancreatic DNase (I). This would explain the wider range of hypersensitive sites recognized and cleaved by this enzyme, when compared with 1, 10- phenanthroline copper (I) and micrococcal nuclease.

As described previously, premeltons at the 5' ends of genes serve to initiate site-specific DNA melting by the RNA polymerase to begin transcription, while premeltons at the 3' ends of genes facilitate the detachment of the RNA polymerase to terminate transcription.

It is well known that genes undergoing active DNA transcription are particularly sensitive to pancreatic DNase, the level of transcriptional activity being correlated with the degree of nuclease sensitivity. This sensitivity strongly suggests the presence of beta-DNA regions in actively transcribed chromatin.

Actinomycin binds to beta-DNA found within the boundaries connecting double-stranded B-DNA with single-stranded DNA in the transcription complex. This immobilizes (i.e., "pins") the complex, interfering with the elongation of growing RNA chains. In extremely active genes, RNA polymerases lie in a close-packed arrangement along DNA. Interference with the movement of one polymerase by actinomycin is expected to affect the movement of other polymerases. This explains why nucleolar RNA synthesis is so sensitive to the presence of actinomycin D.

Micrococcal nuclease and pancreatic DNase (I) have been used to probe the structure of DNA in inactive chromatin. Micrococcal nuclease (as well as 1, 10- phenanthroline copper (I)) has been observed to cut DNA every 200 base pairs to generate the basic subunit structure of inactive chromatin, the nucleosome; while pancreatic DNase gives a more complex cleavage pattern, maximal cleavage occurring every 10, 11; 21, 22, etc. base pairs in DNA regions immediately proximal to the 2-fold axis of the nucleosome. As one gets further away from this 2-fold axis, however, this periodicity becomes more exactly 10-fold.

These data are consistent with the presence of (static) high-energy beta-structural elements lying between nucleosomes, unwinding DNA to form the higher order solenoid structure in chromatin (i.e., the 300 Angstrom fiber), and the presence of lower-energy beta-structural elements in nucleosomal DNA that appear (i.e., either transiently or permanently) to relieve the bending strain energy in its left-handed toroidal super-helical structure.

Finally, it is instructive to explore the topological relationship between a left-handed toroidal helix and a right-handed interwound helix using a length of rubber tubing to simulate DNA (i.e., this is best achieved, using a 3/8" diameter tubing that is approximately 40" long). One begins by forming several turns (say, five) of a left-handed toroidal helix, connecting its ends (to establish topological linking), and then releasing the bending strain energy present in the left-handed toroidal helix to observe the spontaneous rearrangement of DNA that leads to the right-handed interwound superhelical form. The latter form contains (negatively superhelical) torsional strain energy that can be used to form premeltons and other DNA premelting manifestations in DNA. This topological inter-conversion may play a key role in activating eukaryotic gene expression.