Natural and synthetic double-stranded DNA binding ... - Springer Link

JBIC (1999) 4 : 431–440

Q SBIC 1999

ORIGINAL ARTICLE

Emiko Kikuta 7 Naomi Katsube 7 Eiichi Kimura

Natural and synthetic double-stranded DNA binding studies of macrocyclic tetraamine zinc(II) complexes appended with polyaromatic groups

Received: 15 February 1999 / Accepted: 22 April 1999

Abstract The characteristic binding mode of zinc(II) complexes of macrocyclic tetraamines (1,4,7,10-tetraazacyclododecane, cyclen) appended with one or two arylmethyl group(s) [(4-quinolyl)methyl-, 1,7-bis(4-quinolyl)methyl-, (1-naphthyl)methyl-, 1,7-bis(1-naphthyl)methyl-, and (9-acridinyl)methyl-cyclen] to doublestranded calf thymus DNA and synthetic DNAs [poly(dA)7poly(dT), poly(dA-dT)2, poly(dI)7poly(dC), poly(dI-dC)2, poly(dG)7poly(dC), and poly(dG-dC)2] has been examined by spectrophotometric methods, Tm measurement, and inhibition of these DNA-directed transcriptions in vitro. Various hypochromic and bathochromic effects on the pendant aromatic absorption spectra of the complexes were observed in titration with the native and synthetic DNA. The binding constants Kapp (p[bound cyclen derivatives]/[unbound cyclen derivatives][DNA phosphates] M –1), at 25 7C in 10 mM EPPS (pH 8.0) containing 0.1 M Na c, were determined and compared with those of the corresponding Zn 2c-free ligands. The results showed that the Zn 2c-cyclen complexes interact with the DNA more strongly than the corresponding diprotonated ligands, leading to a stronger stacking of the pendant aromatic rings. The binding of Zn 2c–(9-acridinyl)methyl-cyclen to calf thymus DNA was competed by an AT-selective, minor groove binder, distamycin, but not by a major groove binder, methyl green. In an unusual interaction of excess Zn 2c–(9-acridinyl)methyl-cyclen with poly(dA)7poly(dT), the Zn 2c-cyclen moiety went into the minor groove to make coordination bonds with the deprotonated imides of the thymines, resulting in disruption of the poly(dA)7poly(dT) duplex. Thymine-containing DNA-directed transcription with Escherichia coli RNA polymerase in vitro was inhibited by the

E. Kikuta 7 N. Katsube 7 E. Kimura (Y) Department of Medicinal Chemistry, Faculty of Medicine Hiroshima University, Kasumi 1-2-3, Minami-ku, Hiroshima 734-8551, Japan e-mail: ekimura6ipc.hiroshima-u.ac.jp Fax: c81-82-2575324

Zn 2c–(9-acridinyl)methyl-cyclen. The 50% inhibition concentrations of the transcription (IC50) were 22–45 mM with poly(dA)7poly(dT) or poly(dA-dT)2 as templates, while with poly(dG-dC)2 as a template the IC50 value was 110 mM. Key words Zn 2c-cyclen complex 7 Double-stranded DNA interaction 7 Transcription inhibition 7 AT-selective binding

Introduction A zinc(II) complex 1a of the macrocyclic tetraamine 1,4,7,10-tetraazacyclododecane (cyclen) has a unique propensity to bind with deprotonated imides, whose derivatives include thymine (T), uracil (U), or barbitals.

Scheme 1

432

Thus, 1a selectively recognizes thymine (as T –) or uracil (as U –) among DNA (or RNA) bases by reversibly forming a stable complex 2 in physiological pH aqueous solution (Scheme 1) [1]. Furthermore, with the cyclen appended with aromatic rings such as acridine (1b) [2–4], the resulting 1 : 1 complex 3 became ca. 20 times more stable at pH 8.0, i.e. Kappp3.2!10 3 M –1 (for 2) versus Kappp6.3!10 4 M –1 (for 3). Very recently, we have found that Zn 2c-cyclen complexes 1a and 1b interact with U in single-stranded poly(U), unperturbed by the presence of the anionic phosphodiester backbone [5, 6]. The Zn 2c-cyclen complexes, moreover, seemed to disrupt the A-U hydrogen bonds in poly(A)7poly(U), as demonstrated by lowering of the RNA duplex melting temperature (Tm). The present study was designed to deepen our earlier knowledge about the Zn 2c-cyclen derivatives as selective T (or U)-recognizing small molecules in doublestranded DNA and develop a new type of DNA-interacting probe. We now have added newly synthesized cyclen derivatives appended with one (4) or two quinolines (5), and one (6) or two naphthalenes (7). For double-stranded DNAs, naturally occurring calf thymus DNA and synthesized poly(dA)7poly(dT), poly(dAdT)2, poly(dI)7poly(dC), poly(dI-dC)2, poly(dG)7poly(dC), and poly(dG-dC)2 have been studied.

Materials and methods Reagents Distamycin A (Sigma), methyl green (Sigma), and ethidium bromide (Merck) were dissolved in water. All the cyclen derivatives except for 7 and its Zn 2c-free ligand were dissolved in 10 mM EPPS (pH 8.0). Compound 7 and its Zn 2c-free ligand were dissolved in 10 mM EPPS (pH 8.0) containing 50% dimethyl sulfoxide. Concentrations were determined spectrophotometrically: ε303p34,000 for distamycin [7], ε638p85,300 for methyl green [7], ε480p5850 for ethidium bromide [8], ε361p10,030 for 1b, ε361p9980 for Zn 2c-free 1b, ε317p3900 for 4, ε316p3100 for Zn 2c-free 4, ε317p7400 for 5, ε316p6200 for Zn 2c-free 5, ε283p7000 for 6, ε283p6800 for Zn 2c-free 6, ε283p15,300 for 7, ε283p12,300 for Zn 2c-free 7. DNAs Calf thymus DNA (Sigma) was dissolved in water, sonicated, and filtered. Poly(dA)7poly(dT), poly(dA-dT)2, poly(dC)7poly(dG), poly(dC-dG)2, poly(dI)7poly(dC), poly(dI-dC)2, poly(dA), and poly(dT) were purchased from Amersham Pharmacia and dissolved in water. Their concentrations per phosphates were determined spectrophotometrically: ε253p6600 for calf thymus DNA [9], ε260p6000 for poly(dA)7poly(dT) [9], ε262p6600 for poly(dA-dT)2 [10], ε253p7400 for poly(dG)7poly(dC) [9], ε254p8400 for poly(dG-dC)2 [9], ε245p5300 for poly(dI)7poly(dC) [11], ε251p6900 for poly(dI-dC)2 [12], ε257p8600 for poly(dA) [13], and ε264p8520 for poly(dT) [14]. They were stored at –20 7C in a final concentration of 1 mM per phosphates in 10 mM EPPS (pH 8.0) containing 0.1 M NaNO3.

UV and visible absorption studies Spectrophotometric titration experiments were performed with a Hitachi U-3500 spectrophotometer at 25 7C. A 1 cm quartz cuvette was used, with continuous stirring throughout the course of the titration. A stock solution of DNAs [1.0 mM per phosphates in 10 mM EPPS (pH 8.0) containing 0.1 M NaNO3] was added in increasing amounts to a 2 ml solution of the cyclen complexes in the same buffer. The decrease in the absorption of the Zn 2ccyclen complexes or their Zn 2c-free ligands at their absorption maxima were measured. The apparent binding constants Kapp were determined from the plots of D/Dεapp versus D, where D is the concentration of DNA in phosphates, Dεapppεf –εapp and Dεpεf –εb [15], where εb and εf correspond to the extinction coefficient of the DNA-bound form of the cyclen derivatives and the extinction coefficient of the DNA-unbound cyclen derivatives, respectively. The apparent extinction coefficient, εapp, was obtained by calculating Aobsd/[cyclen derivatives], where Aobsd corresponds to the observed absorbance at absorption maxima. The data were fitted to Eq. 1, wherein a slope equal to 1/Dε and a y-intercept equal to Kapp/Dε were obtained. εb was determined from Dε, and Kapp from the ratio of the slope to the y-intercept: D/DεapppD/DεcKapp/Dε

(1)

The concentration of the DNA-bound cyclen derivatives (Cb) was determined as follows [16]: Cbp(εfCtotal–Aobs)/(εf–εb)

(2)

The concentration of the DNA-unbound cyclen derivatives (Cf) was determined by: CfpCtotal–Cb

(3)

DNA melting studies Thermal melting curves of native and synthetic DNAs [50 mM in 10 mM EPPS buffer (pH 8.0 at 25 7C) containing 15 mM NaNO3] were followed on a Hitachi U-3500 spectrophotometer equipped with a thermoelectric cell temperature controller (B0.5 7C) and a stirrer unit. A 1 cm quartz cuvette was used. The temperature was raised at the rate of 0.5 7C/min. Tm values were determined by differentiation of the melting curves. The DTm value for each compound was calculated as the temperature difference between the compound’s Tm and the DNA’s Tm. DNA transcription inhibition assay The inhbitory activities of Zn 2c-cyclen complexes in transcription of DNA in vitro were examined using calf thyums DNA or syn-

433 thetic DNA polymers as a template and Escherichia coli RNA polymerase (from Sigma) [17]. Reaction mixtures (total volume 50 ml) contained each 200 mM of GTP, CTP, ATP, and UTP except where [ 3H]-UTP (50 mCi/mmol, 200 mM), [ 3H]-ATP (50 mCi/mmol, 200 mM), or [a- 32P]-CTP (6 Ci/nmol, 200 mM) were substituted for non-radioactive substrate, 10 mM MgCl2, and 0.2 mM dithiothreitol, template DNAs, and cyclen derivatives in 50 mM Tris-HCl (pH 8.0). The phosphate concentrations of the template DNAs were as follows: calf thymus DNA, 250 mM; poly(dA)7poly(dT), 20 mM; poly(dA-dT)2, 20 mM; poly(dG-dC)2, 20 mM. The reactions were started by adding 5 ml of RNA polymerase (0.22 mg protein/ml). All the reactions were terminated after 60 min at 37 7C by pouring the 50 ml of the reaction solution on to a DE81 filter (DEAE cellulose ion exchanger from Whatman). Filters were dried for 10 min and washed four times for 10 min with 5% Na2HPO4 solution. The incorporation of the labeled substrates into the synthesized RNA was counted with a liquid scintillation counter.

Results UV and visible absorption spectral changes of the aromatic pendants in interaction with calf thymus DNA The UV and visible absorption spectra of the aromatic pendants of 1b, 4, and 5 changed significantly as a result of their binding to calf thymus DNA. Monitoring such changes would be helpful in estimating the binding constants. In titration of the Zn 2c-cyclen derivatives (20–40 mM) with DNA (0–333 mM) in EPPS buffer (pH 8.0) containing 0.1 M NaNO3 at 25 7C, the absorption maxima of 1b (lmaxp361 nm, εp10,030), 4 (lmaxp317 nm, εp3900), and 5 (lmaxp317 nm, εp7400) decreased with increasing concentration of DNA (hypochromicity) (Fig. 1a–c). Isosbestic points were observed near 310 nm for 1b, 322 nm for 4, and 320 nm for 5 binding to DNA, suggesting that for each Zn 2c-cyclen complex there is a single mode of binding to DNA. Among these, 1b and 5, moreover, showed significant red shifts (bathochromism) as [DNA] increased. The bathochromism of 4 was limited in comparison with that of 5, indicating that the stacking mode of the quinolines was not identical. This fact may result from difference in the single stacking by 4 and double stacking by 5 [in view of significantly higher complexation constants with dT – (or U –) and the remarkably different chemical shifts in 1H NMR data (unpublished), the aromatic rings in 5 are proposed to act as bisintercalators to sandwich thymines]. For comparison, the Zn 2c-free ligands of 1b, 4, and 5 were similarly studied; they showed lesser bathochromisities and smaller hypochromisities than the corresponding Zn 2c-cyclen complexes (data not shown). The changes in the maximum absorbances allowed us to construct the half-reciprocal plots for [DNA] with matching [Zn 2c-cyclen derivatives] according to Eq. 1 (see Materials and methods). From the linear half-reciprocal plots for 1b, 4, and 5 (Fig. 1a’–cb), apparent binding constants (Kapp) were obtained from the ratio of the slope to the y-intercept (at [DNA] 1 100 mM,

where the plots gave straight lines; the binding was quantitative [18]), which were 3.0!10 4 M –1 for 1b, 1.5!10 3 M –1 for 4, and 3.3!10 4 M –1 for 5 in 10 mM EPPS buffer (pH 8.0) containing 0.1 M NaNO3 at 25 7C. Note that these values were obtained at [DNA] more than five times [Zn 2c-cyclen derivatives] (i.e. rp[1b]/ [DNA]~0.2), where the interaction was derived mainly from the p-p stacking in the intact double strands (supplemented by electrostatic forces), but not significantly from the Zn 2c–(dT –) complexation. Similar spectral titrations were conducted for the corresponding Zn 2cfree ligands, which gave a smaller Kapp value of 1.4!10 4 M –1 for the Zn 2c-free ligand of 1b. The Kapp values of the Zn 2c-free ligands of 4 and 5 could not be determined since their hypochromisities were too small to construct the half-reciprocal plot. The Kapp values for 6 and 7 could not be determined because their absorption maxima (lmaxp283 nm, ep7000 for 6 and lmaxp283 nm, ep15,300 for 7) overlapped with that of calf thymus DNA. UV and visible absorption spectral changes of 1b in interaction with synthetic DNAs To study further how the Zn 2c-cyclen derivatives recognized the specific DNA sequences, the UV and visible absorption spectral changes of 1b were investigated in the presence of synthetic double-stranded DNAs. The spectral changes of 20 mM 1b with poly(dA-dT)2 (0–200 mM) in 10 mM EPPS buffer (pH 8.0) containing 0.1 M NaNO3 at 25 7C is illustrated in Fig. 2a, which shows one isosbestic point at 310 nm and gives almost a linear half-reciprocal plot (Fig. 2ab). A similar spectral change with one isosbestic point at 310 nm and linear half-reciprocal plots were observed for 1b binding to homo- and hetero-GC and -IC double-stranded DNAs (data not shown). From these linear half-reciprocal plots, the Kapp values were calculated (see Table 1). In the titration of 1b with poly(dA)7poly(dT), mixed equilibria occurred apparently with two different isosbestic points at 340 nm for lower concentration of DNA (~47 mM) and 290 nm for higher concentration of DNA ( 1 47 mM) (Fig. 2b), giving very complex, nonlinear half-reciprocal plots (Fig. 2bb). However, if the titration was started with very high [DNA] ( 1 80 mM) with respect to [1b] (~10 mM), the spectral change showed one isosbestic point at 310 nm (Fig. 2c) and gave a linear half-reciprocal plot (Fig. 2cb) to allow determination of the Kapp value of 1.1!10 5 M –1. Without Zn 2c, the hypochromism and bathochromism were very limited for any DNA (Table 2). Typical titrations of 20 mM of the Zn 2c-free 1b (lmaxp361 nm, ε361p9980) with poly(dA-dT)2 (0–214 mM) and poly(dA)7poly(dT) (0–214 mM) are shown in Fig. 2d and e. Similar spectral changes were observed in interaction of the Zn 2c-free 1b with other DNAs (data not shown). From the half-reciprocal plots, we have estimated all the Kapp values for the Zn 2c-free 1b, which

434 Fig. 1 Spectrophotometric titrations of 1b (20 mM) (a), 4 (40 mM) (b), and 5 (25 mM) (c) with calf thymus DNA. The concentrations of DNA phosphates (mM) (rp[Zn 2ccyclen complex]/[DNA phosphates]) are: (a) 0, 29 (0.64), 57 (0.32), 83 (0.21), 107 (0.16), 130 (0.13), 145 (0.11), 167 (0.09), and 194 (0.08) from the top to bottom curve at 361 nm; (b) 0, 49 (1.59), 95 (0.82), 140 (0.53), 180 (0.40), 222 (0.32), 281 (0.25), 298 (0.23), and 333 (0.2) at 317 nm; (c) 0, 29 (0.87), 58 (0.43), 86 (0.28), 113 (0.21), 140 (0.17), 165 (0.14), and 190 (0.12) at 317 nm. Arrows indicate isosbestic points. The half-reciprocal plots ([DNA] 1 100 mM) are shown for for 1b (a’), 4 (b’), and 5 (c’)

are all about one order of magnitude smaller than those for the corresponding Zn 2c-cyclen complexes (see Table 1). In order to know more about the DNA interaction mode of 1b, groove binders and an intercalator displacement experiments were performed (Fig. 3a). Distamycin is known to bind to the minor groove at the AT base pairs by hydrogen bonds with adenine N3 and thymine O2 [19, 20]. The absorbance at 361 nm of 1b decreased upon addition of calf thymus DNA in 10 mM EPPS (pH 8.0) containing 0.1 M NaNO3 at 25 7C. When the mixing ratio of [1b]/[DNA phosphates] was [10.5 mM]/[96.0 mM], the ratio of the DNA-bound and DNA-unbound 1b ([1bbound]/[1bfree]) was calculated to be [8.4 mM]/[2.1 mM] using Eqs. 2 and 3 (see Materials and methods). Upon addition of distamycin, a decrease was observed in the ratio of [1bbound]/[ 1btotal]. The 50% dissociation of the DNA-bound 1b occurred with 6 mM distamycin. Likewise tested was the displacement by methyl green, which is known to interact with the hydrophobic surface in the major groove [21] and has AT specificity [22]. However, in comparison to distamycin,

Table 1 Apparent binding constants (Kapp M –1) of 1b and its Zn 2c-free ligand with native and synthetic DNAs, as determined by spectrophotometric titrations in 10 mM EPPS (pH 8.0) containing 0.1 M NaNO3 at 25 7C DNA

1b

Zn 2c-free 1b

Ratio

Calf thymus DNA poly(dA)7poly(dT) poly(dA-dT)2 poly(dG) poly(dC) poly(dG-dC)2 poly(dI)7poly(dC) poly(dI-dC)2

3.0!10 4a 1.1!10 5b 5.0!10 4a 5.6!10 4a 1.7!10 4a 4.7!10 4a 2.5!10 4a

1.4!10 4a 3.5!10 3a 2.9!10 3a 6.4!10 3a 3.8!10 3a 3.3!10 3a 9.3!10 3a

2.1 31.4 17.2 8.8 4.5 14.4 2.7

a

By the titration of 1b (20 mM) with DNAs (0–220 mM). The half-reciprocal plots were made at [DNA] 1 100 mM b By the titration of 1b (10 mM) with poly(dA)7poly(dT) (80–180 mM). The half-reciprocal plot was made at [DNA] 1 80 mM

methyl green did not so significantly promote dissociation of the DNA-bound 1b (50% dissociation occurred with 30 mM methyl green). Ethidium bromide was also tested as a non-base-selective intercalator. Ethidium

435 Fig. 2 Spectrophotometrictitration of 1b (20 mM) with a poly(dA-dT)2 or b poly(dA)7poly(dT); c 1b (10 mM) with poly(dA)7poly(dT); the Zn 2c-free ligand of 1b (20 mM) with d poly(dA-dT)2 or e poly(dA)7poly(dT). The concentration of DNA phosphates (mM) (r value) are: a 0, 29 (0.67), 57 (0.33), 83 (0.22), 107 (0.17), 130 (0.13), 145 (0.12), 167 (0.10), 187 (0.09), 200 (0.08) from the top to bottom curves at 361 nm; b 0, 9 (2.15), 19 (1.01), 29 (0.65), 38 (0.49), 47 (0.40), 74 (0.24), 99 (0.18), 122 (0.14), 145 (0.12), 167 (0.10), and 187 (0.08) at 361 nm; c 0, 80.1 (0.11), 101 (0.085), 128 (0.065), 153 (0.052), and 176 (0.044) at 361 nm; d 0, 58 (0.30), 113 (0.15), 165 (0.10), and 214 (0.07) at 361 nm; e 0, 58 (0.30), 113 (0.15), 165 (0.10), and 214 (0.07) at 361 nm. The half-reciprocal plots are shown for for 1b (20 mM) with poly(dA-dT)2 (a’), 1b (20 mM) with poly(dA)7poly(dT) (b’), and 1b (10 mM) with poly(dA)7poly(dT) (c’)

bromide also decreased [1bbound] with an IC50 value of 20 mM. Since distamycin, methyl green, and ethidium bromide were reported to possess nearly the same binding affinities to calf thymus DNA (Kappp4!10 5 M –1 for distamycin, 3!10 5 M –1 for methyl green, and 5!10 5 M –1 for ethidium bromide at pH 7.4 and [Na c]p0.1 M) [7, 23], the outstanding effect by distamycin further supports the AT-sequence preference and minor groove specificity of the Zn 2ccyclen derivatives when they interact with calf thymus DNA.

The interaction of the Zn 2c-free ligand of 1b with DNA was also blocked, but almost indiscriminately by distamycin and ethidium bromide (Fig. 3b). When the mixing ratio of [Zn 2c-free 1b]/[DNA phosphates] was [10.1 mM]/[96.0 mM], where the ratio of the [Zn 2c-free 1bbound]/[Zn 2c-free 1bfree] was calculated to be [6.4 mM]/[3.7 mM], the 50% dissociation of the Zn 2cfree 1b occurred with the same 1.7 mM distamycin and ethidium bromide.

436 Table 2 Extent of hypochromism (at 361 nm) and red shift (bathochromism)

Hypochromism (%) a

Calf thymus DNA poly(dA)7poly(dT) poly(dA-dT)2 poly(dG)7poly(dC) poly(dG-dC)2 poly(dI)7poly(dC) poly(dI-dC)2

Red shift (nm) b

1b

Zn 2c-free 1b

1b

Zn 2c-free 1b

54 54 45 61 63 54 45

11 16 5 14 21 13 10

5 5 5 5 5 5 5

ND c ND ND ND ND ND ND

a

Calculated from Dε (Eq. 1 in Materials and methods) Calculated from the differences between observed lmax of 1b or Zn 2c-free 1b in the absence of DNAs and in the presence of excessive DNAs c Red shift was not detected b

DNA melting studies Additional information on the DNA binding properties of the Zn 2c-cyclen derivatives were obtained from the DNA melting studies. The DNA [all 50 mM in 10 mM EPPS (pH 8.0 at 25 7C) containing 15 mM NaNO3] melting profiles in the presence of increasing concentration of 1b and its Zn 2c-free ligand are shown in Fig. 4. The melting temperatures (Tm) of poly(dA)7poinitially increased ly(dT) and poly(dA-dT)2 for poly(dA)7poly(dT) and {DTmpc4.7 7C DTmpc3.4 7C for poly(dA-dT)2 at rp[1b]/[DNA phosphates]p0.1 in Fig. 4b and c}. However, later, Tm began to decrease [e.g. DTmp–3.3 7C for poly(dA)7poly(dT) at rp0.2 and DTmp–1.3 7C for poly(dA-dT)2 at rp0.3]. The melting profile of calf thymus DNA (AT contentp58%) was somewhat similar to that of these AT polymers (Fig. 4a). On the other hand, the double strands poly(dI)7poly(dC) and poly(dI-dC)2 were steadily stabilized by 1b (Fig. 4d and e). Likewise, the Zn 2c-free ligand of 1b steadily stabilized the double strand of calf thymus DNA and all the synthesized DNAs (Fig. 4a’–eb). Inhibition of DNA-directed in vitro transcription To test if the interaction of the Zn 2c-cyclen complexes with DNA affects some biochemical processes, Zn 2c-

Fig. 3 Competition profiles of groove binders and an intercalator in DNA binding of a 1b and b the Zn 2c-free 1b. The cross point of the middle dashed line indicates 50% dissociation concentration (IC50)

cyclen complexes 1a, 1b, 4, 5, 6, 7, and their Zn 2c-free ligands were examined for their ability to inhibit in vitro transcription from calf thymus DNA (250 mM) as a template by E. coli RNA polymerase. The calf thymus DNA-directed transcription was assayed by measuring the incorporation of [ 3H]-UTP using the same method as applied to the actinomycin D transcriptional inhibition test [17]. Reactants also contained all other nucleotide substrates (each 200 mM) needed for RNA synthesis. Indeed, the inhibition of the transcription took

437

place with the Zn 2c-cyclen complexes 1b, 5, and 7, but not with the Zn 2c-free ligands. The 50% inhibition concentrations (IC50) of the Zn 2c-cyclen complexes are summarized in Table 3. The most effective complex was the bis-naphthyl-pendant Zn 2c-cyclen 7. Then, we tested 1b to inhibit the transcription from the synthetic DNAs, i.e. with poly(dA-dT)2, poly(dA)7poly(dT), and poly(dG-dC)2 (all 20 mM) as templates. The inhibition profiles are shown in Fig. 5. For the poly(dA-dT)2 template, incorporation of either [ 3H]-ATP or [ 3H]-UTP substrate was inhibited to the same degree with IC50p36 mM and 33 mM, respectively (Fig. 5a). On the other hand, for the poly(dA)7po-

Fig. 4 Native and synthetic DNAs [all 50 mM in 10 mM EPPS (pH 8.0 at 25 7C) containing 15 mM NaNO3] melting profiles in the presence of a-e 1b or a’-e’ Zn 2c-free 1b. a, a’ Calf thymus DNA, b, b’ poly(dA)7poly(dT), c, c’ poly(dA-dT)2,d, d’ poly(dI)7poly(dC), and e, e’ poly(dI-dC)2

Fig. 5 Inhibition profiles of in vitro transcription using E. coli RNA polymerase: a [ 3H]-ATP or [ 3H]-UTP incorporation directed by poly(dA-dT)2, b [ 3H]-ATP or [ 3H]-UTP incorporation directed by poly(dA)7poly(dT), and c [a- 32P]-CTP incorporation directed by poly(dG-dC)2

438 Table 3 The 50% inhibition (IC50) of calf thymus DNA (250 mM)-directed transcription by cyclen derivatives with or without Zn 2c Cyclen derivatives

1a 1b 4 5 6 7

IC50 (mM) Zn 2c complex

Zn 2c-free ligand

1 200 130 1 200 95 1 200 55

1 200 1 200 1 200 1 200 1 200 1 200

ly(dT) template the incorporation of [ 3H]-ATP (IC50p22 mM) was not the same as of [ 3H]-UTP (IC50p45 mM), the former being more effectively blocked (Fig. 5b). This fact is compatible with the prediction that 1b strongly binds to the poly(dT) strand. For the poly(dG-dC)2 template, the incorporation of [a- 32P]-CTP was weakly inhibited (IC50p110 mM), implying that in the transcription this template was not as effectively blocked as the AT polymers by 1b (Fig. 5c).

Discussion The UV and visible spectra of the cyclen derivatives in the presence of calf thymus DNA were measured for acridine (1b) and quinoline pendants (4 and 5), whose absorption spectra did not overlap with those of the DNA bases. The observed higher hypochromisms for those of the Zn 2c-cyclen complexes over the Zn 2cfree ligands suggest stronger electronic interactions between the intercalating chromophores and the DNA bases [15, 24]. Since the strength of this electronic interaction is expected to decrease as the cube of the distance of separation between the chromophore and the DNA bases, the larger hypochromism means closer proximity of the chromophores in the Zn 2c-cyclen complexes to the DNA bases. In addition to the greater decrease in the intensity, the higher bathochromisms are also consistent with stronger interactions of the acridine and quinoline with the DNA base stack. From the smooth UV absorption changes of 1b (Fig. 1a), 4 (Fig. 1b), and 5 (Fig. 1c) with calf thymus DNA, and of 1b with poly(dA-dT)2 (Fig. 2a), poly(dA)7poly(dT) (Fig. 2c), homo- and hetero-GC polymers, and homo- and hetero-IC polymers, the binding constants Kapp were determined from the half-reciprocal plots at [DNA] 1 [1b] (i.e. r~0.11–0.2), which are shown to be all larger than those for the Zn 2c-free ligands (Table 1). The results suggest that the Zn 2c-cyclen moiety may interact with the nucleophilic atoms such as adenine N3, thymine O2, inosine N3, and cytosine O2 in the minor groove, and guanuine N7 and inosine N7 in the major groove (Scheme 2). The stronger electrostatic interactions by the Zn 2c-cyclen than the

Scheme 2

diprotonated cyclen in the DNA grooves would make the intercalation of the aromatic pendants more favorable. It is to be emphasized that, under the given conditions, the DNAs were seen to remain double-stranded to permit regular intercalations with the polyaromatic pendants. Of special comment are the binding constants Kapp of 1.1!10 5 M –1 for 1b with poly(dA)7poly(dT) and 5.0!10 4 M –1 for 1b with poly(dA-dT)2, which are 31.4 and 17.2 times higher than those for the Zn 2c-free 1b, respectively. The Kapp ratios with other DNAs, in particular poly(dG-dC)2 and poly(dI-dC)2, are significantly smaller. This fact indicates that the Zn 2c-cyclen moiety prefers the AT sequence over the GC or IC sequence, owing to some contribution of the Zn 2c–(dT –) bonding. When poly(dA)7poly(dT) (0–200 mM) was gradually added to 1b (20 mM), emergence of two isosbestic points suggests that the binding modes differed at lower and higher concentration of DNA (Fig. 2b). For comparison, we have measured spectral changes of 1b with each component DNA, i.e. single-stranded poly(dA) and poly(dT). The single-stranded poly(dA) (Fig. 6a) showed small hypochromicity with one isosbestic point at 295 nm, and a linear half-reciprocal plot. On the other hand, poly(dT) (Fig. 6b) showed stronger hypochromism and more remarkable bathochromism, which is somewhat analogous to the initial titration spectra in Fig. 2b. However, the half-reciprocal plots revealed a totally non-linear relation. In this titration of 1b (20 mM) with poly(dA)7poly(dT), [1b] is initially in large excess, permitting the strong interaction of Zn 2ccyclen moiety with dT to come into play, leading to some disruption of the A-T duplex. However, the spec-

439

Fig. 6 Spectrophotometric titration of 1b with a poly(dA) and b poly(dT). Initial concentrations of 1b are 20 mM. The concentrations of DNA phosphates (mM) (r) are: a 0, 10 (1.99), 24 (0.81), 41 (0.46), and 58 (0.31) from the top to bottom curves at 361 nm; b 0, 2.0 (10.2), 3.8 (5.24), 5.7 (3.42), 7.0 (2.74), 8.3 (2.29), 9.5 (1.97), 11.5 (1.59), 13.8 (1.29), 16.0 (1.09), and 18.7 (0.90) at 361 nm

tral titration of 1b (10 mM) with a large excess poly(dA)7poly(dT) (80–180 mM) (i.e. r~0.11) (Fig. 2c) gave a linear half-reciprocal plot (Fig. 2cb), suggesting regular interaction with the intact double strand predominantly by p-p stacking under these conditions. On the other hand, the hetero AT polymer, poly(dA-dT)2, showed only one equilibrium (Fig. 2a), probably because single-strand poly(dA-dT) could maintain the double helical structure by itself, even if the Zn 2c-cyclen disrupts the double strand at [1b] 1 1 [DNA]. The DNA melting behaviors of poly(dA)7poly(dT) and poly(dA-dT)2 [all 50 mM in 10 mM EPPS (pH 8.0 at 25 7C) containing 15 mM NaNO3] in the presence of 1b support this conclusion based on the spectrophotometric titrations. The Tm increased at the lower concentration of 1b (i.e. at r~0.1–0.2), implying that the intact double strand was initially stabilized predominantly by the intercalation effect by the pendant acridine (Fig. 4b and c). A similar Tm increase was seen with the Zn 2c-free 1b (Fig. 4b’ and cb). However, at r 1 0.2 or 0.3, the effect of 1b reversed and Tm started to decrease, indicating that the effect of the interaction of the Zn 2c-cyclen moiety with thymine surpassed the mere p-p stacking effect, destabilizing the doublestranded structures. Such an abnormal Tm behavior was not observed with homo- and hetero-IC polymers (Fig. 4d and e), as anticipated from the smooth spectral

titration data. In our recent study [5, 6], 1b simply stabilized poly(dG)7poly(dC). In the biochemical test, Zn 2c-cyclen derivatives 1b, 5, and 7 inhibited the calf thymus DNA (250 mM)-templated transcription more strongly than the other Zn 2c-cyclen derivatives. The IC50 values were found to correlate with the DNA binding properties of the complexes. Among these, 1b Zn 2c-cyclen (Kappp3.0!10 4 M –1) and 5 (Kappp3.3!10 4 M –1), which showed higher affinity to calf thymus DNA as concluded from the spectrophotometric titration, inhibited the transcription more strongly, while 4 (Kappp1.5!10 3 M –1) and 6 inhibited transcription much more weakly. In the absence of Zn 2c, all these ligands failed to inhibit the transcription, which again proves the essential role of the Zn 2c-cyclen moiety. When the template DNA consisted of A and T, the incorporation of A from [ 3H]-ATP or U from [ 3H]UTP into RNA was examined. The ratio of [ 3H]-ATP incorporation to [ 3H]-UTP incorporation was identical when the reaction was directed by poly(dA-dT)2 (Fig. 5a). The Zn 2c-cyclen complex 1b might bind to substrate (each [NTP subthe [ 3H]-UTP strate]p200 mM) to inhibit the transcription under this condition. However, this was disproved by the equal uptake of [ 3H]-ATP and [ 3H]-UTP. Therefore, it is concluded that 1b essentially bound to T bases in the template DNA, but not to the monomeric UTP substrate under the given conditions. On the other hand, when poly(dA)7poly(dT) was used as a template DNA, the [ 3H]-ATP incorporation was about twice lower than the [ 3H]-UTP incorporation into RNA (Fig. 5b). This result is consistent with the stronger binding of 1b to poly(dT) in the template DNA to make its transcription unfavorable.

Conclusion Zn 2c-macrocyclic tetraamine complexes appended with polyaromatic pendants 1b, 4, and 5 bind to doublestranded calf thymus DNA and synthetic DNAs, as studied by UV and visible spectrophotometric titrations. At relatively dilute concentration, the acridinependant Zn 2c-cyclen 1b showed 31.4 and 17.2 times higher affinity for intact double-stranded homo- and hetero-AT polymers than its Zn 2c-free ligand, which is to be compared with 8.8 and 4.5 times higher affinity for GC polymers. This fact suggests that the Zn 2c-cyclen moiety determines the preference to the AT sequence over the GC sequence. Stronger hypochromism and bathochromism of the absorption spectra of Zn 2ccyclen complexes with DNA titration support the idea that the electrostatic interaction of the Zn 2c-cyclen moiety with DNA grooves brings about a closer proximity of the aromatic ring to the DNA bases. Binding of 1b to calf thymus DNA was blocked by an AT-selective, minor groove binder (distamycin), but less significantly by a major groove binder (methyl green) or a

440

intercalating agent (ethidium bromide). Complex spectral changes in the titration of poly(dA)7poly(dT) by 1b at high concentration illustrates mixed equilibria, including dissociation of the poly(dT) strand. From the DNA melting experiments it is concluded that, at lower concentration, 1b stabilized the thymine-containing DNAs predominantly by the intercalation effect of the acridine, but at higher concentration, 1b destabilized the duplexes by the Zn 2c–(dT –) interaction to intervene into the AT hydrogen bonds. The DNA binding properties of the Zn 2c-cyclen complexes brought about inhibition of these DNA-directed transcriptions in vitro. The Zn 2c-cyclen complexes, having high binding affinities to calf thymus DNA, showed the higher inhibition. Among those tested, the bis-naphthalenependant Zn 2c-cyclen 7 showed the highest activity. The Zn 2c-cyclen complex 1b inhibited AT polymer-directed transcription to a greater degree than GC polymer-directed transcription, suggesting again the stronger perturbation effect on the AT duplex. The stronger uptake inhibition of ATP over UTP when the transcription was directed by the homo-AT polymer supports the stronger binding of 1b to the poly(dT) strand than to poly(dA). Acknowledgements This research was supported by a Grand-inAid for Scientific Research on Priority Areas “Biometallics” (No. 08249103) for E.K.

References 1. Shionoya M, Kimura E, Shiro M (1993) J Am Chem Soc 115 : 6730–6737 2. Shionoya M, Ikeda T, Kimura E, Shiro M (1994) J Am Chem Soc 116 : 3848–3859 3. Tucker JHR, Shionoya M, Koike T, Kimura E (1995) Bull Chem Soc Jpn 68 : 2465–2469 4. Koike T, Gotoh T, Aoki S, Kimura E, Shiro M (1998) Inorg Chim Acta 270 : 424–432 5. Kimura E, Ikeda T, Aoki S, Shionoya M (1998) JBIC 3 : 259–267 6. Kimura E, Koike T, Shionoya M (1997) Pure Appl Chem 69 : 2187–2195 7. Tuite E, Sehlstedt U, Hagmer P, Norden B, Takahashi M (1997) Eur J Biochem 243 : 482–492 8. Breslauer KJ, Remeta DP, Chou W-Y, Ferrante R, Curry J, Zaunczkowski D, Synder JD, Marky LA (1987) Proc Natl Acad Sci USA 84 : 8922–8926 9. Wells RD, Larson JE, Grant RC, Shortle BE, Cantor CR (1970) J Mol Biol 54 : 465–497 10. Inman RB, Baldwin RL (1962) J Mol Biol 5 : 172–184 11. Inman RB, Baldwin RL (1964) J Mol Biol 8 : 452–469 12. Grant RC, Harwood SJ, Wells RD (1968) J Am Chem Soc 90 : 4474–4476 13. Chamberlin MJ (1965) Fed Proc 24 : 1446–1457 14. Ts’o PO, Rapaport SA, Bollum FJ (1966) Biochemistry 5 : 4153–4170 15. Kumar CV, Asuncion EH (1993) J Am Chem Soc 115 : 8547–8553 16. Chaires JB, Dattagupta N, Crothers DM (1982) Biochemistry 21 : 3933–3940 17. Chu W, Shinomiya M, Kamitori KY, Kamitori S, Carlson RG, Weaver RF, Takusagawa F (1994) J Am Chem Soc 116 : 7971–7982 18. Zimmermann HW (1986) Angew Chem Int Ed Engl 25 : 115–130 19. Coll M, Frederick CA, Wang AH-J, Rich A (1987) Proc Natl Acad Sci USA 84 : 8385–8389 20. Van Dyke MW, Hertzberg RP, Dervan PB (1982) Proc Natl Acad Sci USA 79 : 5470–5474 21. Kim SK, Nordén B (1993) FEBS Lett 315 : 61–64 22. Fox KR, Higson SL, Scott JE (1992) Eur J Histochem 36 : 263–270 23. Kumar KA, Muniyappa K (1992) J Biol Chem 267 : 24824–24832 24. Dougherty G, Prigram WJ (1982) CRC Crit Rev Biochem 12 : 103–132