The value of the twist energy parameter (E(T)) of pBR322 is determined near zero superhelix density from topoisomer distributions created under various conditions. The resulting value, E(T) = 1155 +/- 65, at 37 degrees C is essentially unaffected by adding 10 mM Mg2+, or by changing the kind of Topo I from chicken-red-cell to calf-thymus. This value significantly exceeds that (E(T) = 950 +/- 80) measured for p30 delta DNA under indentical conditions by the same method in the preceding paper. Decreasing the temperature from 37 to 21 degrees C yields a slightly larger value, E(T) = 1340 +/- 130, but the statistical significance of the increase is marginal. Attempts to determine reliable E(T) values for pBR322 at higher superhelix densities by ethidium binding were frustrated by the fact that good fits of the equilibrium dialysis results could not be achieved using a single value of E(T). Moreover, the curves of apparent E(T) versus binding ratio r vary considerably from one preparation to another, and for a given preparation vary with time after cell lysis up to about seven weeks, after which they settle in to nearly reproducible behavior. The apparent E(T) values obtained from competitive dialysis experiments are typically rather low (E(T) similar to 700) for small r and nearly native superhelix density, and rise up to 1300 to 1500 with increasing binding ratio (up to r = 0.055) and decreasing negative superhelix density. The observed trend of apparent E(T) values extrapolates to a still larger value at r approximate to 0.07, sigma = 0, which significantly exceeds the value obtained from topoisomer distributions near r = 0, sigma = 0. These findings cast considerable doubt on the reliability of apparent E(T) values obtained for pBR322 by ethidium binding studies. The differences between pBR322 and p30 delta are attributed to one or more different secondary conformations that prevails over part of the pBR322 sequence, and to the metastable prevalence of one or more of those as dye is added. In the case of pBR322 DNA, the torsion constant, molar ellipticity at 250 nm, and plateau diffusion coefficient from dynamic light scattering at K = 4.5 x 10(5) cm(-1) all vary with superhelix density in a similar way, showing dips at sigma = -0.010 and jumps at sigma = -0.037. This behavior, which is similar to that observed previously for pUC8 dimer DNA, suggests that two successive transitions in secondary structure are induced in pBR322 by increasing negative superhelix density. The possible role of such structural transitions in the contrasting behaviors of pBR322 and p30 delta is discussed.
