The Structure of Duplex DNA

clive delmonte clived at
Mon Sep 14 12:54:57 EST 1998


You are right, and it is timely, to raise the topological papers by Crick,
Wang & Bauer (J Mol Biol 129 (1979) 449-461), and Bauer, Crick & White
(Scientific American 243 (1980) 100-113).  Their treatments did not apply to
every side-by-side model, however :

"..This proves that the linking number is not equal to zero (at least for
the great majority of those, i.e., side-by-side, molecules.),

and also only to such side-by-side models

" which the two ...chains do not coil around each other ...but instead
lie side by side over most of their length, having only a few helical

Their treatment only applies to side-by-side models, like that of Rodley et
al., for which there is a net winding because an equal number of right and
left turns do not get you back to an equivalent repeat point in space
because of the D chirality of asymmetric carbon in all the pentoses.

The model I have come upon has exactly and identically zero plectonaemic
turns in its relaxed state over any length of duplex and is therefore not
covered and constrained by the two papers just cited.

Thermodynamics has always been a delight to me, though I find it difficult!
Chemical energy is dissipated in all directions; that is, it is not a
vector, but torque is a vector.  Therefore chemical energy would seem to be
unable to either wind or unwind two plectonaemic strands where no breakage
of covalent bonds takes place.  My solution to this conundrum is to say that
neither winding nor unwinding does actually take place !

But tell me - let me tease you a little - what do you make of the results of
Kabata et al. (SCIENCE 262 (1993) 1561-1563) who found that RNA polymerase
slides along its DNA duplex substrate and does not follow a helical path ?

An absolutely crucial test of the new model offers itself, though we need
someone to carry it out.  Drew & Travers (J Mol Biol 186 (1985) 773-790)
have constructed a 169 bp circular, covalently closed length of duplex DNA.
With this length of DNA there can only be a very limited number of
topoisomers present if the ring closure, of itself, could introduce only a
very few superhelical turns.  That is, all the 169 bp molecules would share
a range of superhelical turns between about -1 and +1.  Therefore in
suitably mild denaturing conditions, in the absence of any strand breakage,
the 169 bp sequence would be largely separable into complementary covalently
closed single-stranded DNA circles interlinked at very few places according
to the new model. Thus the most direct route to distinguishing between the
DNA double helix and the new model is based on topology.  The complementary
strands in that 169 bp sequence may actually be linked by 0, or perhaps ± 1
superhelical turns due to the algebra of the enzymic closure reaction used
on the 169 bp linear duplex, and the denatured covalently closed circular
strands could be decorated with a suitable protein and then be examined
under AFM to determine their linking number.

Therefore, if the 169 bp circular DNA is denatured under mild conditions,
perhaps using aqueous urea, it could then be decorated with the
single-strand DNA binding protein, the product of gene 32 from phage T4
(Scanning Microscopy 9 (1995) 705-727), so as to be rendered easily visible
under AFM, and the actual numbers of crossovers then counted.

According to the double helix model, LK will lie in the range 16 to 18, or
so, while the new model predicts LK is -1, 0 or +1.

A very similar experiment could be carried out with the small circular DNAs
described by Bates & Maxwell (EMBO J 8 (1989) 1861-1866), where, for the
smallest, the double helix model predicts LK = 9, and the new model predicts
LK = -2.

Is there any research group with the interest and ability to conduct this
experiment ?

clive delmonte
clived at

More information about the Xtal-log mailing list