Amiloride intercalates into DNA and inhibits DNA topoisomerase II.

Amiloride is capable of inhibiting DNA synthesis in mammalian cells in culture. Recent evidence indicates that the enzyme, DNA topoisomerase II, is probably required for DNA synthesis to occur in situ. In experiments to determine the mechanism of inhibition of DNA synthesis by amiloride, we observed that amiloride inhibited both the catalytic activity of purified DNA topoisomerase II in vitro and DNA topoisomerase II-dependent cell functions in vivo. Many compounds capable of inhibiting DNA topoisomerase II are DNA intercalators. Thus, we performed studies to determine if and how amiloride bound to DNA. Results indicated that amiloride 1) shifted the thermal denaturation profile of DNA, 2) increased the viscosity of linear DNA, and 3) unwound circular DNA, all behavior consistent with a DNA intercalation mechanism. Furthermore, quantitative and qualitative measurements of amiloride fluorescence indicated that amiloride (a) bound reversibly to purified DNA under conditions of physiologic ionic strength, and (b) bound to purified nuclei in a highly cooperative manner. Lastly, amiloride did not promote the cleavage of DNA in the presence of DNA topoisomerase II, indicating that the mechanism by which amiloride inhibited DNA topoisomerase II was not through the stabilization of a "cleavable complex" formed between DNA topoisomerase II, DNA, and amiloride. The ability of amiloride to intercalate with DNA and inhibit topoisomerase II is consistent with the proposed planar, hydrogen-bonded, tricyclic nature of amiloride's most stable conformation. Thus, DNA and DNA topoisomerase II must be considered as new cellular targets of amiloride action.

somerase I1 activity is probably required for DNA synthesis by mammalian cells in culture (6)(7)(8). Here we present evidence that amiloride inhibits the catalytic activity of purified DNA topoisomerase I1 in vitro and in uiuo, that this inhibition occurs because amiloride intercalates into DNA, but that the intercalation into DNA by amiloride does not result in topoisomerase I1mediated cleavage of DNA (9)(10)(11).

MATERIALS AND METHODS
Enzymes, Nucleic Acids, and Antitumor Drugs-Calf thymus DNA topoisomerase I1 was purified to greater than 95% homogeneity according to the procedure of Halligan et al. (12). Pd-knotted DNA was purified from P4 tailless capsids as previously described (13). Plasmid pBR322 was purified using a discontinuous, two-step CsClethidium bromide gradient technique (14). 4'-(9-Acridinylamino)methane sulfon-m-anisidide (m-AMSA,' NSC 249992) and its isomer o-AMSA (NSC 156306) were obtained from the Drug Synthesis and Chemistry Branch, Division of Cancer Treatment, National Cancer Institute. All drugs were dissolved in dimethyl sulfoxide (Me,SO) at 1 mg/ml. Amiloride HCl was a gift from Merck Sharp and Dohme.
Assays for DNA Topoisomerase II Function-End-labeled pBR322 DNA was prepared as published (15). The cleavable-complex assay, the topoisomerase I1 unknotting-inhibition assay, and the topoisomerase I1 relaxation-inhibition assay were performed as described (9,16,17). The final concentration of MezSO in all assays did not exceed 1% (v/v). This was especially important for the cleavable-complex assay in which MezSO at concentrations greater than 1% stimulates topoisomerase 11-mediated DNA cleavage.
DNA-Calf thymus DNA (Worthington) was dissolved in 0.15 M sodium chloride, 0.015 M sodium citrate, pH 7.0, to a final concentration of 3 mg/ml, and sonicated as described by Davidson et al. (18). DNA concentrations were determined using an extinction coefficient of 6600 M at 260 nm and are expressed in terms of nucleotide equivalents per liter. The purified sonicated DNA samples displayed an A2M1/A280 ratio of between 1.85 and 1.9 and a total hyperchromicity at 260 nm of 30%. These spectral properties are consistent with published values (19). The DNA contained less than 0.5% residual RNA as determined by the method of Savitsky and Stand (20).
Thermal Denaturation-Thermal denaturation experiments were performed by the method of Cory et al. (21,22) on a Varian CARY 2290-UV-VIS recording spectrophotometer with a five cell temperature-controlled turret. The average value of the T, for calf thymus DNA was 58.1 "C ( n = 40, S.D. = 1.5 "C).
Unwinding Angle Determination-The unwinding angle determination was done using the viscometric titration technique of Revet et al. (25). The high copy number 6.9-kilobase plasmid pOPlA6 (26) was isolated by the method of Garger et al. (14) and shown to be supercoiled DNA by gel electrophoresis.
Cell Culture and Synchronization-The human promyelocytic cell line, HL-60, was grown in spinner culture in RPMI 1640 medium containing 10% (v/v) fetal bovine serum. BALB/c 3T3 fibroblasts (clone A31) were grown in DMEM supplemented with 7.5% fetal bovine serum. HL-60 cells in log phase growth were synchronized by growth for 24 h in the presence of 0.15 pg/ml aphidicolin.
Assay for Adhesion of HL-60 Cells-Synchronized HL-60 cells were washed once in RPMI 1640 medium, 10% (v/v) fetal bovine serum, resuspended in the same at 1.5-2.0 X lo6 cells/ml, and incubated at 37 "C in the presence of 200 nM phorbol dibutyrate and other additions as specified. After 4 h, the number of adherent cells was determined as described previously (27).
Assay for DNA Synthesis in Aphidicolin-synchronized HL-60 Cells-Aphidicolin-synchronized HL-60 cells were released from the block and allowed to traverse S phase by washing the cells once with RPMI 1640 medium, 10% fetal bovine serum and resuspension at 1 X 10' cells/ml of the same in the absence or presence of amiloride or DNA topoisomerase I1 inhibitors, as indicated. After 1 h of incubation at 37 "C, ['Hlthymidine was added to a final concentration of 1 pCi/ ml and incubation continued at 37 ' C for 4 h. Incorporation of ['HI thymidine was halted and measured by precipitation with trichloroacetic acid, passing the sample over a GF/C filter, washing the filter with methanol, and counting the filter using Aquasol-2.
Fibroblost Mitogenesis Assay-Monolayers of BALB/c 3T3 fibroblasts were rendered quiescent by culture for 24 h in DMEM, 0.5% fetal bovine serum. The cells were placed under these "serum-starved" conditions to ensure arrest in the Go/GI stage of the cell cycle. Reentry into the cell cycle was initiated at time to by addition of 10% fetal bovine serum. Stimulation by serum was limited to a 7-h pulse, for at time t7 the monolayers were washed and reincubated in DMEM, 0.5% fetal bovine serum. Cells were exposed to amiloride or other DNA topoisomerase inhibitors for one 7-h interval, either to to t7 or t7 to tlr. Exposure to inhibitors was terminated by washing the monolayers and reincubation in fresh DMEM, 0.5% fetal bovine serum. All monolayers received the same number of washes and medium changes. At t23 ['Hlthymidine (1 pCi/ml) was added and incorporation into DNA allowed to proceed for 1 h and determined as described previously (4).
Determination of Amiloride Binding to Calf Thymus DNA and to Purified Nuclei Using the Fluorescent Properties of Amiloride-DNA prepared as described above was dialyzed into 25 mM Hepes, 10 p~ EDTA, pH 7.3. Assays were performed in 0.3 ml of this same buffer in the presence of DNA (13 X M of DNA phosphate), 10 p~ amiloride, and 0-100 mM NaCl as indicated. Concentrations of amiloride substantially greater than 10 p~ could not be used because of inner filter effects. Concentrations of DNA less than 2 5 X M of DNA phosphate could not be easily studied because of diminution in the "quenched signal/noise ratio." Fluorescence excitation spectra were determined in a Perkin-Elmer 512 Fluorescence Spectrometer with the emission wavelength set at 420 nm.
Nuclei were isolated from HL-60 cells as described by Kreutter et al. (28). Purified nuclei were suspended in buffer consisting of 25 mM Hepes, pH 7.3, 5 mM MgC12, 250 mM sucrose at a final concentration of 60 X lo6 nuclei/ml, using aliquots of 0.3 ml/assay. Amiloride (3 pl in Me2SO) was added from 100 X concentrated stocks to final desired concentrations and the nuclei incubated for 30 min at 37 "C. To conclude incubation and separate nuclear-bound amiloride from free amiloride, samples were diluted with ice-cold buffer, spun in a Microcentrifuge for 30 s, the nuclear pellets washed once with ice-cold buffer, and resuspended in 0.35 ml of buffer. The content of nuclearbound amiloride was measured by excitation at 340 nm and quantitation of fluorescent emission at 420 nm.

RESULTS
Amiloride Inhibits the Catalytic Activities of Purified Mammalian DNA Topoisomerase 1 1 in Vitro-Two different measures of catalytic activity of purified calf thymus DNA topoisomerase I1 were assayed in vitro. The ability to strand-pass was monitored using the bacteriophage P4 unknotting assay and the ability to relax supercoiled DNA was measured using pBR322 plasmid DNA. Amiloride, like the DNA topoisomerase I1 inhibitor novobiocin, inhibited in a dose-dependent manner the topoisomerase 11-mediated conversion of P, DNA from the knotted to the unknotted form ( Fig. 1). Inhibition of DNA topoisomerase I1 catalysis by amiloride occurred with an apparent Ki of -1.25 mM. As well, amiloride inhibited the ability of topoisomerase I1 to relax supercoiled DNA, with an apparent Ki of -0.5 m M (Fig. 2). Although amiloride is a weak inhibitor by comparison to most other known DNA topoisomerase I1 inhibitors (e.g. m-AMSA, Ki = 25 p~, Ref. 9; ellipticine, Ki -4 p~, Refs. 10 and 11; epipodophyllotoxin VM-26, Ki -5 p~, Ref. 16), amiloride is commonly used in experiments with cultured cells a t concentrations of 1 mM or greater. Thus, it was necessary to determine whether amiloride inhibited functions associated with DNA topoisomerase I1 in intact cells.
Amiloride Inhibits Topoisomerase 11-dependent Cell Functions in Vivo-Although mammalian DNA topoisomerase I1 has been suggested to play a role in numerous cellular (nuclear) processes (i.e. DNA replication, chromosome sorting, and transcription; for review see Ref. 29), there does not exist an assay of intact mammalian cell function as a standard barometer of DNA topoisomerase I1 activity in vivo. However, using three cellular assay systems in which DNA topoisomerase I1 activity appears to be required, we determined the likelihood that amiloride inhibited DNA topoisomerase I1 in intact cells.
Amiloride Inhibits Phorbol Dibutyrate-induced Adhesion of HL-60 Cells-HL-60 cells, in human promyelocytic cell line, differentiates in response to tumor-promoting phorbol diesters, such as phorbol dibutyrate (30, 31). Protein kinase C, a calcium-and phospholipid-dependent protein kinase, is activated by phorbol dibutyrate (reviewed in Ref. 32). Previous work from our laboratory indicated that protein kinase C phosphorylated purified DNA topoisomerase I1 in vitro, resulting in the activation of the topoisomerase (33). These findings indicated the possibility that DNA topoisomerase I1 played a role in phorbol diester induced-differentiation of HL-60 cells. Consistent with this hypothesis, we have reported previously that the phorbol diester-induced differentiation of HL-60 cells (as assessed by induction of cell adhesion) was blocked by the topoisomerase I1 inhibitors novobiocin and m-AMSA, but not by the inactive o-AMSA (33). Using an improved version of this assay, Fig. 3 indicates that amiloride, like novobiocin and m-AMSA, inhibited phorbol diester-induced adhesion of synchronized HL-60 cells in a dose-dependent manner, with an apparent Ki of 200 p~. These results The strand-passing activity of calf thymus DNA topoisomerase I1 was monitored by the P,-unknotting assay (13).   suggest that amiloride may have the ability to inhibit topoisomerase I1 in intact HL-60 cells. However, we have previously shown that amiloride, unlike novobiocin and m-AMSA, can inhibit protein kinase C in vitro and in vivo with a Ki of -1 mM (27). Thus, additional assays of topoisomerase I1 function in intact cells were explored.
Amiloride Inhibits DNA Synthesis in Synchronized HL-60 Cells-DNA topoisomerase I1 may be required for DNA synthesis (6,7), and the enzyme has been shown to be associated with newly replicated DNA (8). Aphidicolin is a reversible, potent, and specific inhibitor of DNA polymerase a and is often used to synchronize mammalian cells at G , / S (34).   (4,35). The ability of amiloride to inhibit mitogenesis was ascribed initially to the ability of amiloride to inhibit growth factor induced Na'/H' exchange (35, 36). However, work by ourselves (4,27) and others (37, 38) indicated that amiloride can inhibit many cellular processes including RNA synthesis, protein synthesis, and protein kinase activity. In addition, we observed that inhibition by amiloride required only that the drug be present in the hours immediately preceding S phase, a time period incompatible with a requirement for Na'/H' exchange (4).
However, as shown in Table I, the DNA topoisomerase I1 inhibitors novobiocin and m-AMSA, but not o-AMSA, blocked DNA synthesis in mitogen-(10% serum) stimulated fibroblasts with the same time dependence as that of amiloride. Taken together, these observations suggest that DNA topoisomerase I1 activity is required for DNA synthesis and that inhibition of DNA synthesis by amiloride may, at least in part, result from inhibition of topoisomerase 11.
Elucidating the Mechanism by Which Amilori.de Inhibits DNA Topoisomerase II-To determine the mechanism by which amiloride inhibits the catalytic activity of DNA topoisomerase 11 in vitro and in vivo, studies were performed to assess if and how amiloride might interact directly with DNA.
Thermal Denaturation of DNA-The ability of an agent to alter the thermal denaturation profile of DNA (in excess of ionic strength effects alone) is used as an indication of the ability of the compound to bind to DNA (22). Table I1 shows the increase in thermal denaturation temperature of amiloride-DNA complexes (AT,,,), measured at fixed DNA concentration and varying amiloride concentration. Comparison with the standard, ethidium bromide, indicates that amiloride interacts with DNA with a lower affinity. Control experiments indicated that the increased ionic strength due to the addition of the amiloride HC1 salt could not account for the AT,,, generated by amiloride. These results indicate that amiloride binds to DNA, although these observations alone cannot distinguish between intercalation into DNA and other modes of DNA binding.
Viscometric Titrations of Linear DNA-One characteristic of intercalative binding to DNA is the increase in DNA length that results when a ligand stacks between the base pairs. Sonicated DNA in solution exhibits hydrodynamic behavior similar to that of a rigid rod. The helical extension afforded to rod-like DNA by intercalators can be measured by viscometric titration, and the relative increase in DNA contour length (L/L,J uersu the ligand/nucleotide ratio can be determined. Fig. 4 shows a comparison of the viscometric titrations  with amiloride and ethidium bromide at low ionic strength, 1 mM ammonium fluoride buffer. The ethidium bromide data are consistent with the neighbor exclusion principle in that saturation occurs at 0.25 DNA phosphates, indicating binding of one ethidium molecule per two base pairs. While the initial slope of the amiloride titration is the same as that for ethidium bromide, the titration saturated at a drug/phosphate ratio of approximately 0.1 to 0.15, indicating binding of one amiloride molecule per four base pairs at saturation. In general, high concentrations of salt weaken the binding of compounds to DNA. In analogy to Fig. 4, viscometric titration of DNA with amiloride and ethidium bromide were performed in high ionic strength buffer, 500 mM ammonium fluoride. The tightly binding intercalator, ethidium bromide, again showed neighbor-excluded intercalation (saturation at 0.25 DNA phosphates per drug molecule), with a viscosity slope only slightly lower than that in low salt. The weaker binding amiloride, however, showed no intercalation (zero slope) in the presence of 500 mM ammonium fluoride (data not shown).
Viscometric Titrations of Circular DNA: Determining the Unwinding Angle-Although the increase of viscosity of linear DNA which occurs in the presence of amiloride is consistent with intercalative binding, binding to the DNA surface could stiffen the helix and result in an increase in DNA viscosity. However, a property exclusive to intercalative DNA binding is the resulting left-handed unwinding of the DNA helix. This unwinding is necessary for accommodation of the intercalating chromophore without breakage of the sugar-phosphate backbone (39). To determine if amiloride could unwind DNA, a series of viscometric titrations were performed with varying concentrations of closed, circular, supercoiled plasmid DNA and amiloride or ethidium bromide, and the data expressed as a Vinograd plot (25). By comparing the results obtained with amiloride with those obtained with ethidium bromide (a standard of known unwinding angle, Ref. 50), we calculated that amiloride generated a DNA-unwinding angle of 5.9" (data not shown). Thus, amiloride binds to DNA via intercalation.
Amiloride Fluorescence Reflects Reversible Binding to DNA-To demonstrate directly the ability of amiloride to bind to DNA under conditions of physiological ionic strength, the fluorescent properties of amiloride were used. The excitation spectra of amiloride alone, DNA alone, and amiloride and DNA together in 25 mM Hepes, pH 7.3,5 mM MgCl, are shown in Fig. 5. Amiloride alone fluoresced with an intensity centered at approximately 350 nm. No appreciable fluorescence was observed for DNA alone (Fig. 5A). Addition of DNA to amiloride resulted in a decrease (quench) of the fluorescence intensity of amiloride. Furthermore, the shape of the spectra changed from a poorly resolved doublet to a single broad peak centered at approximately 370 nm. The difference spectrum (amiloride alone minus amiloride in the presence of DNA) is shown in Fig. 5B. The magnitude of the fluorescence decrease was dependent on the ionic strength. As the ionic strength was increased (by adding increasing amounts of NaCl), the magnitude of the fluorescence quenching was decreased, suggesting that increasing ionic strength promoted dissociation of bound amiloride from DNA. Ionic strength did not significantly alter the excitation spectrum of amiloride alone. At physiologic ionic strength (in the presence of 100 mM NaCl), significant binding of amiloride to DNA was still observed as evidenced by fluorescence quenching.
Amiloride Binds to Purified Nuclei in a Highly Cooperative Manner-To determine if amiloride could bind to native DNA (chromatin), amiloride fluorescence was used to measure directly binding to purified, intact nuclei isolated from HL-60 cells. Mixtures of amiloride and nuclei were incubated for 30 min at 37 "C followed by very rapid separation of nuclearbound amiloride from free amiloride. Fig. 6 indicates that amiloride bound to nuclei in a specific and saturable manner with an apparent I ( d of 1.3 mM. Transformation of the binding isotherm data by Hill plot (Fig. 7) reveals that the binding of amiloride to nuclei was highly cooperative, exhibiting a Hill coefficient of 3.8. Such highly cooperative binding is characteristic of the interaction of some DNA intercalators with DNA (40-42) and with chromatin (43).  Fig. 6. The linear region of the plot shows a slope (Hill coefficient) of -3.8, indicative of a highly cooperative interaction between amiloride and the nuclei. The x-intercept of the plot gave an estimate of the K d of 1.3 mM under these conditions. the action of DNA topoisomerase 11 (9,11). Amiloride has been shown to inhibit numerous protein kinases (27, 38), and for protein kinases the apparent mechanism of inhibition is competition with ATP (27, 38). Since DNA topoisomerase I1 also utilizes ATP, we used a classic enzyme kinetic approach to determine whether amiloride inhibited topoisomerase I1 by competing with ATP. Measuring initial rates of reaction, and at fixed concentrations of amiloride (0.3, 1.0, or 3.0 mM), neither varying the concentration of DNA over an 8-fold range (from approximately one-half of the K , to -4 times the K,) nor varying the concentration of ATP (0.3 mM to 3.0 mM, K , was 1 mM) appeared to affect either the Vmax of DNA topoisomerase I1 or the K , for ATP (data not shown). In contrast, the topoisomerase I1 inhibitor novobiocin showed marked competition with ATP (data not shown), as has been reported previously (44). Thus, amiloride does not appear to inhibit topoisomerase I1 by competing with either DNA or ATP.

Kinetics
Amiloride Does Not Generate the Cleavable Complex-Many DNA intercalators have been shown to induce protein-linked DNA breaks in cultured cells and in isolated nuclei, the protein being linked to the 5' ends of the broken DNA strands (9,10,15,(45)(46)(47). The DNA breakage induced by the intercalative drugs has been demonstrated to be mediated by DNA topoisomerase I1 (9,10). The intercalators interfere with the breakage reunion reaction of the DNA topoisomerase I1 by stabilizing a drug-topoisomerase 11-DNA ternary intermediate, giving rise to a "cleavable complex." Treatment of this cleavable complex with protein denaturants results in both single and double strand DNA breaks and the covalent linking of one topoisomerase I1 subunit to each 5' phosphate end of the broken DNA. Since many DNA intercalators stimulate the formation of the cleavable complex using purified DNA and purified mammalian DNA topoisomerase I1 (9-11), we have used this assay system to determine if the mechanism of inhibition of DNA topoisomerase I1 by amiloride is via stabilization of the cleavable complex. In the presence of purified DNA topoisomerase 11, amiloride failed to stimulate the generation of cleaved DNA fragments, whereas m-AMSA, the established intercalative topoisomerase inhibitor, strongly promoted topoisomerase 11-mediated DNA cleavage (data not shown). Amiloride, however, is not the only example of a DNA intercalator which does not generate the cleavable complex. Ethidium bromide, a far more potent DNA intercalator than amiloride, does not cause formation of the cleavable complex (Ref. 11 and data not shown). Furthermore, novobiocin, a nonintercalating antibiotic which inhibits topoisomerase I1 via competition with ATP (Ref. 44 and data not shown) does not promote topoisomerase 11-mediated DNA cleavage. Clearly, only some topoisomerase I1 inhibitors generate the cleavable complex. Amiloride, ethidium bromide, and novobiocin do not.

DISCUSSION
Previous work by our laboratory and others had indicated that amiloride was most effective in inhibiting DNA synthesis in cultured cells when given during the prereplicative period (4,37,49). While examining the role of DNA topoisomerase I1 in DNA synthesis we noticed that DNA topoisomerase I1 inhibitors also were most effective in inhibiting DNA synthesis if given during the prereplicative period. These observations led us to determine whether amiloride was also a DNA topoisomerase I1 inhibitor. The data presented here clearly indicate that amiloride can bind to purified nuclei, intercalate into DNA, and inhibit the catalytic activity of topoisomerase I1 but not via stabilization of a ternary DNA-topoisomeraseamiloride intermediate, the cleavable complex.
The viscometric titrations with linear calf thymus DNA indicate clearly that amiloride binds to DNA by a mechanism which serves to modify the hydrodynamic behavior of the DNA. This has been interpreted to be a lengthening or stiffening of the DNA helix. In conjunction with the unwinding angle determination, the linear DNA viscosity titrations indicate that amiloride intercalates with the DNA and probably saturates the DNA-binding sites at a ratio of one amiloride molecule per four base pairs. The unwinding angle determined for amiloride (5.9") is low compared to most intercalating compounds. However, given that amiloride binds only weakly to DNA it is possible that all of the amiloride was not bound during the determination of the unwinding angle. Unbound drug would result in a lower estimate of the unwinding angle value.
At present we do not understand the mechanism by which amiloride inhibits the catalytic activity of DNA topoisomerase 11. The data indicate that the mechanism of inhibition does not involve competition with ATP or with DNA, or occur via generation of the cleavable complex. Possession of such properties by a DNA intercalator is not without precedent, however. Like amiloride, ethidium bromide intercalates into DNA, inhibits the catalytic activity of topoisomerase 11, but does not generate the cleavable complex (11). Clearly, definition of the mechanism of inhibition of topoisomerase I1 by amiloride awaits further work.
Using three model assay systems, we have demonstrated that amiloride inhibits intact cellular functions sensitive to inhibition by classic topoisomerase I1 inhibitors. These findings could be used to suggest that inhibition of topoisomerase I1 is the primary mechanism by which amiloride inhibits cellular replication. However, because amiloride is known to be capable of inhibiting so many processes we can only add topoisomerase I1 to the list of potential cellular targets of amiloride action.
Although amiloride is a DNA intercalator, its affinity for DNA is substantially less than other DNA intercalators (9)(10)(11). Amiloride consists of a substituted pyrazine ring with an acylguanidino group attached to the ring at position 2, and this structure is usually depicted as shown in Fig. 8A. At first glance the monocyclic structure of amiloride would not lead one to expect that it might intercalate with DNA. However, studies by Smith et al. (3) using natural abundance 'H, 13C, and 15N NMR in conjunction with quantum mechanics computations suggest that the ground-state tautomer of amiloride hydrochloride is the planar, hydrogen-bonded, tricyclic acylamino tautomer (Fig. 8B). This conformation of amiloride has been observed in crystal structures (3) and could easily intercalate intooDNA: it provides a planar ring system approximately 10 A across with a set of hydrogen bond-donating groups on either end of the system. These hydrogen bond groups, the 5-amino and the terminal-nitrogen atoms of the guanidino moiety, could interact with the phosphodiester backbone of B-DNA since the backbone P atoms are approx-