Unlocking and opening a DNA gate

The DNA topoisomerases are enzymes that solve various entanglement problems of intracellular DNA. In their presence, DNA strands or double helices can pass through one another as if there were no physical boundaries in between. Manipulations of DNA by a DNA topoisomerase often require displacements between different parts of the enzyme–DNA complex, over distances of tens of angstroms. Such a requirement is particularly evident in reactions catalyzed by the type II DNA topoisomerases. A type II DNA topoisomerase catalyzes the ATP-dependent transport of one DNA double helix through another (1, 2). Each enzyme is made of two identical halves and possesses two protein gates, an ATP-operated entrance gate that admits the DNA segment to be transported (the T segment), and a second gate for the exit of the admitted T segment after its passage through an enzyme-bound DNA segment termed the G segment (reviewed in refs. 3 and 4; see figure 4B in ref. 4 for a sketch of the type II DNA topoisomerase catalyzed reaction). Slicing a DNA double helix through the entire interface between the two enzyme halves and the enzyme-bound DNA double helix clearly requires large movements within the enzyme–DNA complex. However, there have been few direct studies of such movements before the elegant experiments described in this issue of PNAS by Smiley et al. (5), in which single-molecule fluorescence resonance energy transfer (FRET) is used to monitor the opening and closing of the DNA gate by Drosophila DNA topoisomerase II.

Opening a DNA gate by a type II DNA topoisomerase involves unlocking the gate by covalent catalysis that transiently breaks the G segment and widening the gate by moving the two G segment halves away from each other. The unlocking reaction utilizes transesterification chemistry, in which a pair of symmetry-related enzyme tyrosyl groups attack a pair of phosphoryl groups in the two strands of the G segment to sever the DNA backbone bonds and form, at the same time, two DNA–protein covalent bonds. For Drosophila DNA topoisomerase II, a member of the A-subfamily of the type II enzymes, the two positions of DNA cleavage are 4 bp apart. Locking the DNA gate utilizes the same transesterification chemistry, and breakage of the DNA–protein phosphotyrosine links is accompanied by reformation of the DNA backbone bonds.

Monitoring a DNA Gate by FRET

Smiley et al. (5) used several well designed double-stranded DNA oligomers in their work. The nucleotide sequence is based on one specifically recognized by the Drosophila enzyme, with a pseudodyad that coincides with that of the enzyme in the enzyme–DNA complex. In the substrate used in most of their experiments, the donor (D) and acceptor (A) fluorophores are conjugated to two bases, one on each complementary DNA strand and separated by 14 bp. Thus, in a typical DNA B-helical geometry, D and A are expected to be ≈54 Å apart on the same side of the double helix, and changes in this spacing are in a range well suited for study by FRET. In both bulk and single-molecule measurements, the authors observed efficient FRET between D and A in the enzyme-bound DNA, and there is a large decrease in FRET in the presence of both ATP and Mg²⁺, but not if either ATP or Mg²⁺ is omitted or if the temperature is much lower than the normal reaction temperature for the Drosophila enzyme. It was also shown that omission of ATP leads to no significant changes in the rotational mobility of the fluorophores, and thus all of these results suggest that the observed FRET decrease under normal enzyme reaction conditions is caused by an enzyme-modulated change in the distance between a D–A pair (5).

Putative C and O states are nearly equally populated in a steady-state population.

Single-molecule measurements, in which the DNA is bound to an Avidin-coated glass slide at one biotin-tagged end, show that under normal enzyme reaction conditions, an enzyme-bound DNA fluctuates between a high and a low FRET state, corresponding to a distance change of ≈20 Å between D and A. The authors suggest that the high and low FRET states correspond to the closed (C) and open (O) states of the DNA gate, respectively. Reaction lifetime measurements of these states for several hundred molecules are consistent with a transition between the C and O states; under steady–state ATP hydrolysis, both the forward and reverse rate constants for this transition are ≈1 s⁻¹, indicating equal populations of molecules in the C and O states under steady–state ATP hydrolysis (5).

Two Puzzles

Both the bulk and single-molecule FRET results provide much-needed information on the dynamics of opening a DNA gate by a type IIA DNA topoisomerase. However, the results are puzzling in two respects. First, it is surprising that the putative C and O states are nearly equally populated in a steady- state population. Since the discovery of the first DNA topoisomerase and the proposal of the transesterification mechanism over three decades ago, numerous experiments to trap the postulated covalent complexes had been carried out in which alkali or detergent was added to a reaction mixture. Covalent complexes were indeed found in such experiments, and there is strong evidence that the trapped complexes do indeed represent intermediates of the normal reactions (see, for example, ref. 6). In all of these experiments, only a few percent of the DNA-bound enzyme can be trapped as the covalent complex. Thus, it has been assumed that, in a steady-state population, only a small fraction of any topoisomerase–DNA complex is in the O state; the O state is supposedly favored only in the presence of particular classes of compounds, many of which are antibiotics and anticancer agents in clinical use (7, 8).

Furthermore, based on a crystal structure of a large yeast DNA topoisomerase II fragment and molecular modeling, a G segment-bound type IIA enzyme with a widely open DNA gate is believed to possess a ≈50-Å hole bounded by the severed G segment on one side and the exit gate on the other side (9). The presence of a large hole in a type IIA DNA topoisomerase like yeast or Drosophila DNA topoisomerase II is consistent with an earlier finding thata linear DNA, but not any ring-shaped DNA, can thread through the yeast enzyme in a “closed-clamp” conformation, in which the entrance gate of the enzyme is locked by adenosine 5′-[β,γ-imido]triphosphate (AMPPNP) binding (10). All attempts to thread a linear DNA through a G segment-bound enzyme in a closed-clamp state were unsuccessful, however, and this failure was attributed to the DNA gate in a G segment-bound enzyme being mostly in the C state, with a much smaller hole relative to that of the enzyme–DNA complex in its O state (3, 9).

The second puzzle in the single molecule FRET data is related to a plausible role of the T segment in the opening of the DNA gate. In the measurements of Smiley et al. (5), the DNA oligomers to which the enzyme molecules are bound are affixed at one of their ends to a coated glass slide. Therefore, there are no diffusible DNA molecules in those experiments to act as a T segment in the topoisomerase-catalyzed reactions. If the observed high and low FRET states represent the closed and widely open states of the DNA gate, respectively, then an inescapable conclusion would be that the ATP-dependent opening and closing of the DNA gate by a type II DNA topoisomerase does not require the participation of a T segment. Why is such a conclusion puzzling?

In the opening of the DNA gate by a type II DNA topoisomerase, it has been thought that opening an unlocked gate probably involves a T segment (3). This idea stems from two findings. First, if the active-site tyrosine of a type II enzyme is mutated to a phenylalanine to prevent unlocking of the DNA gate, binding of AMPPNP can still close the entrance gate of a DNA-bound enzyme, but this closure can no longer capture a T segment. Thus, in agreement with the available x-ray structures (4), it appears that the channel between the ATPase domains in the closed-clamp form of the enzyme is not sufficiently large to accommodate a T segment, which in turn hints that the capture of a T segment might help the unlocking and widening of the DNA gate. Second, there is the issue regarding the exit of the T segment after its passage through the DNA gate. As mentioned earlier, when the DNA gate is widely open, the cavity between the G segment and the exit gate of the protein is much larger than the diameter of a DNA double helix. So, after a T segment passes through the opened DNA gate, why should it rush out of the protein exit gate rather than stay put in the hole? The large hole is, after all, lined with positive charges for the comfort of a negatively charged DNA (9). This question led to the idea that the closed state of the DNA gate, whether unlocked or locked, is the stable state; the spontaneous closing of the DNA gate is believed to drive the expulsion of the T segment (3, 9). In this scenario, closing the entrance gate of the enzyme by ATP binding traps the T segment in between the ATPase domains and forces the opening of the normally closed DNA gate; once the T segment passes through the DNA gate into the large hole, the DNA gate immediately returns to its more stable closed state, reducing the size of the hole containing the T segment and forcing the opening of the exit gate for the expulsion of the T segment (3, 9). Clearly, such a view is incongruous with the idea suggested by the single molecule FRET data that the T segment is not involved in the opening and closing of the DNA gate (5).

Closing of the DNA gate is believed to drive the expulsion of the T segment.

Plausible Explanations

So, how might these puzzles be resolved? On one hand, each of the earlier findings has its own limitations. For example, as pointed out by Smiley et al. (5), trapping of the covalent intermediates by alkali or detergents could have perturbed the DNA breakage-rejoining equilibrium in favor of the rejoined state; the low-FRET state probably represents a heterogeneous population of molecules, among which perhaps only a small fraction could be trapped by the addition of protein denaturants (5). The failure in threading a linear DNA through the central hole of a DNA-bound type II DNA topoisomerase is indicative of a large contraction of the hole upon binding a G segment to an enzyme in its closed-clamp conformation, but it does not preclude the possibility of equally populated C and O states under normal reaction conditions: The DNA threading experiment requires the presence of AMPPNP to keep the entrance gate locked (10), and the FRET data suggest that, in a DNA–enzyme complex, AMPPNP might be keeping the DNA gate in a partially rather than fully opened state (5); thus, the size of the central annulus when the DNA gate is not fully open might not be sufficiently large for a DNA double helix to thread through. Finally, in view of the intricate coupling of the C and O states to ATP binding and hydrolysis and the complex interplay between various conformation changes within an enzyme–DNA complex (3), explanations other than the specific ones discussed above are clearly plausible for the T segment exit problem.

On the other hand, it is also plausible that the two observed FRET states do not correspond to the C and O states of the DNA gate. Conceptually, it would seem more reasonable that unlocking the DNA gate precedes a large increase in distance between the pair of fluorophores straddling the positions of DNA cleavage. However, one could also argue for the contrary: namely, the unlocking of the DNA gate might be preceded by an enzyme-mediated distortion of the DNA G segment (bending, twisting, and/or stretching). In the latter scenario, efficient breakage of the DNA backbone bonds and/or widening an unlocked DNA gate in the contorted DNA might be triggered by an additional event, such as the capture of a T segment. Indeed, if the strands of an enzyme-bound G segment are normally intact, then distorting the G segment may offer a way to efficiently cleave the strands.

Closing Remarks

The new FRET results, especially the single-molecule data, are sure to stimulate further experiments. Although three-bpdimensional structures of several type II DNA topoisomerase fragments have already provided invaluable insight into the transport of one DNA double helix through another (see, for example, ref. 11), additional data, especially three-dimensional structures of DNA–enzyme complexes, are needed for the elucidation of the intricate reaction steps. The power of single-molecule FRET measurements, nicely illustrated by Smiley et al. (5), will surely be applied to pairs of fluorophores placed at different locations in the DNA and/or enzyme, and such data should provide a wealth of information for elucidating the molecular movements during the ATP-dependent transport of a DNA double helix by a type II DNA topoisomerase.