The topological state of cellular DNA is intimately connected with its function. Without topoisomerases, cells cannot replicate or package their DNA, or express their genes—and they die. Inhibitors of topoisomerases have therefore become important pharmaceutical agents, targeted at infectious organisms and malignant cells.

Two classes of bacterial topoisomerase inhibitors have been developed as antibiotics. The coumarins, including novobiocin and coumermycin A1, are natural products derived from Streptomyces species. They inhibit the ATP binding of the bacterial type II topoisomerases, DNA gyrase and topoisomerase IV. These antibiotics are not used to treat infections in humans, but research continues to identify clinically effective variants.

The quinolone antibiotics, also inhibitors of bacterial DNA gyrase and topoisomerase IV, first appeared in 1962 with the introduction of nalidixic acid (Figure 1). This compound had limited effectiveness and is no longer used clinically in the United States, but the continued development of this class of drugs eventually led to the introduction of the fluoroquinolones, exemplified by ciprofloxacin (Cipro). The quinolones act by blocking the last step of the topoisomerase reaction in bacteria, the resealing of the DNA strand breaks. Ciprofloxacin is a broad-spectrum antibiotic that works on a wide range of disease-causing bacteria. It is one of the few antibiotics reliably effective in treating anthrax infections and is considered a valuable agent in protection against possible bioterrorism. Quinolones are selective for the bacterial topoisomerases, inhibiting the eukaryotic enzymes only at concentrations several orders of magnitude greater than the therapeutic doses.

Some of the most important chemotherapeutic agents used in cancer treatment are inhibitors of human topoisomerases. Tumor cells generally contain elevated levels of topoisomerases, and agents targeted to these enzymes are much more toxic to the tumors than to most other tissue types. Inhibitors of both type I and type II topoisomerases have been developed as anticancer drugs.

Camptothecin, isolated from a Chinese ornamental tree and first tested clinically in the 1970s, is an inhibitor of eukaryotic type I topoisomerases. Clinical trials indicated limited effectiveness in treating cancer, however, despite its early promise in preclinical work on mice. Two effective derivatives were developed in the 1990s: irinotecan (Campto) and topotecan (Hycamtin), used to treat colorectal cancer and ovarian cancer, respectively (Figure 2). Additional derivatives are likely to be approved for clinical use in the coming years. All of these drugs act by trapping the topoisomerase-DNA complex in which the DNA is cleaved, inhibiting religation.

The human type II topoisomerases are targeted by a variety of antitumor drugs, including doxorubicin (Adriamycin), etoposide (Etopophos), and ellipticine (Figure 3). Doxorubicin, effective against several kinds of human tumors, is in clinical use. Most of these drugs stabilize the covalent topoisomerase–cleaved DNA complex.

All of these anticancer agents generally increase the levels of DNA damage in targeted, rapidly growing tumor cells, but noncancerous tissues can also be affected, leading to a more general toxicity and unpleasant side effects that must be managed during therapy. As cancer therapies become more effective and survival statistics for cancer patients improve, the independent appearance of new tumors is becoming a greater problem. In the continuing search for new cancer therapies, topoisomerases are likely to remain prominent targets.

In large chromosomes, with their DNA highly complexed with proteins and thereby constrained, topological challenges accompany every process in DNA metabolism. The simple movement of a DNA polymerase through the DNA during replication (defining a structure called a replication fork) leads to overwinding ahead of the fork and underwinding behind it. Some of the challenges are extraordinary, such as when two replication forks converge in a eukaryotic chromosome, or when intertwined chromosomal DNAs must be separated at cell division. A complete understanding of DNA metabolism in cells will require more detailed information on how every system interfaces with the proteins that solve topological problems.

321

Lebowitz, J. 1990. Through the looking glass: The discovery of supercoiled DNA. Trends Biochem. Sci. 15:202–207.

Vinograd, J., J. Lebowitz, R. Radloff, R. Watson, and P. Laipis. 1965. The twisted circular form of polyoma viral DNA. Proc. Natl. Acad. Sci. USA 53:1104–1111.

By 1962, the double-helical structure of DNA was established, but little was known about the detailed structure of chromosomes. Researchers had been surprised by reports that the E. coli chromosome was a continuous circle. Were circular chromosomes unusual, or were they widespread? Two research groups, led by Renato Dulbecco and Jerome Vinograd, both at the California Institute of Technology, took up the problem of DNA structure, using the mammalian polyoma virus. The experiments with polyoma DNA led the Vinograd group to the concept of supercoiling.

Analytical ultracentrifugation measures the migration of molecules through a density gradient when ultracentrifugal force is applied (see Figure 16-2); molecules that migrate farther through the gradient have a larger sedimentation coefficient (S). Using this and other methods, researchers found three forms of polyoma DNA in the isolated preparations, which migrated with sedimentation coefficients of 20S, 16S, and 14S. Because the 20S and 16S forms (forms I and II, respectively) were most abundant, the researchers focused on those two. The two forms had the same molecular weight, so the different sedimentation velocities had to involve different conformations. Form I could be isolated in almost pure preparations, if care was taken while preparing the DNA.

Both the Dulbecco and the Vinograd groups published reports indicating that form I could be converted to form II by the addition of reagents that promote DNA strand cleavage. The researchers developed a model in which they assumed form I to be circular and form II to be linear. The observed kinetics suggested that the conversion occurred in a single step. This challenged the model, because two strands would have to be cleaved to generate the linear molecule, and both would have to be broken at the same position.

The Vinograd group decided to examine all three forms of polyoma DNA (20S, 16S, and 14S) by electron microscopy. Philip Laipis carried out the first experiments. His examination of forms I and II yielded the unexpected result that both were circular (Figure 1). Laipis was a relatively inexperienced undergraduate, and some of the other researchers in the lab initially assumed he had made a mistake in preparing the samples. However, several repetitions yielded the same result. Only form III (14S) was linear. Careful controls eliminated any possibility that the result reflected a selective elimination of linear forms during preparation of the DNA for examination; when the researchers premixed measured amounts of forms III and II and then examined them, the linear and circular DNAs were always there in the expected proportions. Additional kinetic studies showed that only one strand break was needed to convert form I to form II. Cleaving form I with endonuclease I (which cleaves both strands) produced only the form III (14S) DNA, with no form II. Forms I and II also had identical buoyant densities, making it unlikely that some non-DNA mass was removed in converting form I to form II.

The important clue lay in the electron micrographs. The form I molecules appeared much more twisted on themselves than the form II molecules. Denaturation experiments showed that the strands of form II could be separated, but those of form I could not. The researchers gradually established that form II was a nicked DNA circle. Understanding form I required a little more work. A comment from colleague Robert Sinsheimer led Vinograd to focus on the twisted nature of the form I DNA. His subsequent modeling with phone cords was documented in 1990 in a delightful retrospective article by Jacob Lebowitz, one of the authors on the 1965 paper. Although the term “supercoiling” was not yet in common use, the discovery of the supercoiled nature of polyoma DNA opened an entirely new field of investigation.

Wang, J.C. 1971. Interaction between DNA and an Escherichia coli protein ω. J. Mol. Biol. 55:523–533.

Over the course of the twentieth century, life scientists were getting used to a fundamental idea: if a change occurs in any cellular structure, there is at least one enzyme that catalyzes it. The first discovery of an enzyme involved in DNA supercoiling was made by James Wang and reported in 1971.

Working at the University of California, Berkeley, Wang had initiated studies of superhelicity in small DNAs that could be isolated from bacteria. Focusing on the DNA from bacteriophage λ, he noticed that E. coli extracts contained a macromolecule that converted the superhelical form of the circular DNA to a relaxed form. He did what any good molecular biologist would do: he purified the macromolecule. The result was a protein that he dubbed the ω (omega) protein. Later, it became known as DNA topoisomerase I.

Initially, the purification was incomplete. However, Wang could establish that the macromolecule was a protein, because its activity was not lost after extended dialysis (using a membrane that allowed small molecules to escape but retained larger proteins); activity was lost when the preparation was heated to 50°C (a temperature that denatures most proteins). Wang demonstrated that the conversion of a DNA circle with 150 superhelical turns to a relaxed circle did not happen in one step. Using sedimentation velocity studies and electron microscopy, he showed that the change was progressive, with one or a few superhelical turns lost in each catalytic step. The activity affected only negative, not positive, superhelicity.

Wang concluded that the enzyme had two activities: a nicking activity and a strand-joining activity. He proposed that the ω protein acted by transiently introducing a break into one strand, creating a swivel in the unbroken strand that would allow the removal of superhelical turns. The reaction required no enzymatic cofactors, suggesting that the reaction pathway featured a transient covalent intermediate. His speculation, shown in Figure 2, proved to be largely correct. Overall, the study produced a remarkable set of insights that thoroughly framed the subsequent study of this and related topoisomerases.

Brown, P.O., and N.R. Cozzarelli. 1979. A sign inversion mechanism for enzymatic supercoiling of DNA. Science 206: 1081–1083.

In 1976, Martin Gellert and colleagues reported the discovery of a second topoisomerase in E. coli. The enzyme, DNA gyrase, had the novel property that it could introduce negative supercoils into DNA, hydrolyzing ATP in the process. DNA gyrase was quickly shown to be critical to DNA replication and other processes, and there was great interest in determining how it worked. Many researchers expected that DNA gyrase generated a net negative superhelicity by relaxing positive supercoils, using a mechanism much like that exhibited by the ω protein, with the creation of a break in one strand and rotation of that strand about the other.

Nicholas Cozzarelli and colleagues, at the University of Chicago, began to focus on experimental observations that did not fit this scheme. First, when active DNA gyrase was acting on a DNA, and the gyrase-DNA combination was treated with a protein denaturant, double-strand breaks were introduced into the DNA. Gyrase molecules were covalently linked to the 5′-phosphoryl groups in the DNA on both ends of the break. This implied that the normal mechanism of gyrase action involves an intermediate in which both strands, not just one, are cleaved. The research group also noticed that gyrase has the unusual capacity to catenate (interlink) two DNA circles. Such a reaction would require the formation of at least a transient double-strand break in one of the DNAs.

Pooling this and other information, Cozzarelli proposed a very different mechanism for gyrase action, one he dubbed “sign inversion” (Figure 3). He imagined that in a circular DNA, gyrase would bind to two segments of DNA that crossed over each other, thereby stabilizing a positive crossover, or node. The creation of such a node would necessarily create a compensating negative node elsewhere in the DNA molecule. Gyrase would then invert the sign of the bound node by breaking both DNA strands, passing the unbroken DNA segment through the break, and resealing the break on the other side. This would change the sign of the node to negative and effectively fix two negative supercoils in the DNA.

The sign inversion model made a novel and unique prediction. DNA gyrase would do something very different from the ω protein: it would change superhelicity in increments of 2 rather than increments of 1. This prediction was not trivial to test. Supercoiled circular DNA (such as plasmid DNA) is isolated from cells as a heterogeneous mixture of topoisomers with a roughly Gaussian distribution of linking numbers. Gyrase could shift the center of that distribution, but highlighting individual reaction steps to observe the predicted Lk increments of 2 would be difficult. Cozzarelli and his student, Patrick O. Brown, found a way to overcome the problem.

325

They initially focused on a particularly small circular DNA, a plasmid of about 2,400 bp called p15. Such a small DNA limited the total number of topoisomers in the Gaussian distribution and facilitated the separation of one topoisomer from another on an agarose gel. Using the ω protein, Brown and Cozzarelli took a naturally supercoiled preparation of p15 DNA and relaxed it completely. They then ran the DNA on an agarose gel under conditions in which the topoisomers of p15 were well separated. They cut the most abundant topoisomer out of the gel and extracted it, effectively isolating a DNA preparation with one topoisomer only.

With a topologically pure DNA in hand, the key experiment could be done. The researchers added enough DNA gyrase (about two heterotetramers per DNA molecule) to ensure that essentially every DNA circle had a bound gyrase. After incubating the DNA and gyrase for 3 minutes, they added ATP, but only enough (30 μM, or about one-tenth of the K_m) to support a slow reaction. The results, shown in Figure 4, are most striking at the 5 second time point. The major product is a species with a change in linking number (ΔLk) of 22. A small amount of DNA with a ΔLk of 24 is also evident. Markers showing topoisomers differing by a ΔLk of 1 are shown in the lanes marked MW. At later time points, the DNA becomes more supercoiled, but topoisomers with an odd-numbered ΔLk do not appear.

Brown and Cozzarelli carried out one additional test. After 5 minutes, the p15 DNA was highly supercoiled. They then added novobiocin, an antibiotic that inhibits supercoiling but not relaxation by gyrase (see Highlight 9-2). After another 30 minutes of incubation, much of the DNA had been substantially relaxed (see Figure 4). The topoisomers present included species with superhelicity changes of 0, –2, and –4 relative to the starting material. This demonstrated that gyrase promoted both supercoiling and relaxation of DNA in increments of 2. The result fulfilled a key expectation of any enzymatic reaction—that the reaction pathway is the same in the forward and reverse directions. Overall, the experiments constituted a compelling case for the sign inversion model and provided the impetus to eventually define two separate classes of topoisomerases.

These advances helped explain the mechanism of action of a range of important antibiotics and antitumor drugs (see Highlight 9-2). They were among a string of important contributions from the Cozzarelli lab, first at Chicago and later at the University of California, Berkeley. With an ebullient personality and a creative intellect, Cozzarelli inspired a generation of scientists, both as a mentor and a colleague. Cozzarelli died due to complications of a treatment for Burkitt’s lymphoma in 2006, at the age of 67. But his lab motto “Blast Ahead” lives on.