Molecular recognition between ligands and nucleic acids: DNA

John R. Kiser, Richard W. Monk, Rondey L. Smalls, and Jeffrey T. Petty ... Alexander L. Satz, Christine M. White, Terry A. Beerman, and Thomas C. Brui...
1 downloads 0 Views 1MB Size
Chem. Res. Tonicol. 1991,4, 661-669

66 1

Molecular Recognition between Ligands and Nucleic Acids: DNA Binding Characteristics of Analogues of Hoechst 33258 Designed To Exhibit Altered Base and Sequence Recognition K. Ekambareswara Rao and J. William Lown* Department of Chemistry, University of Alberta, Edmonton, Alberta, Canada T6G 2G2 Received May 29, 1991 The DNA binding characteristics of new analogues (2-8) of Hoechst 33258 (l),containing pyridine and benzoxazole units and designed for altered base specificity, were evaluated using UV, fluorescence, and circular dichroism studies. Like Hoechst 33258 the new analogues also bind through the minor groove of B-DNA in a nonintercalative fashion. The interaction of the compounds with poly(dA-dT) is salt independent. The studies with poly(dA-dT), ct DNA, and poly(dG-dC) indicated a decrease in the relative binding strength of the new analogues to DNAs compared with the parent molecule, Hoechst 33258. Compounds 5 and 7 showed acceptance of GC bases adjacent to AT base pairs. None of the compounds studied exhibited affinity for A-DNA, double-stranded RNA, or Z-DNA. Structure-DNA binding relationships of the new analogues compared with their parent molecule, Hoechst 33258, are discussed.

Introduction In recent years considerable interest has been evoked in examining the molecular recognition of DNA by small molecules (I). Understanding of molecular mechanisms of interactions of peptides and small molecules with DNA should assist in the development of more potent targetoriented (sequence-specific) compounds (2). There is growing evidence that the potency of several antitumor agents, which exert their action by interacting with cellular DNA, is related to their specific interaction at selected sequences (3). Rational design, synthesis, and systematic study of several analogues related to distamycin and netropsin antibiotics by biophysical, footprinting, and NMR techniques (2, 4) revealed that (i) molecular and global electrostatics (5),(ii) van der Waals interactions (6), (iii) hydrogen bonds (7), and (iv) "phasing" (8, 9) play important roles in molecular recognition of DNA by these agents. Encouraged by these results, we embarked on a systematic study of alternative model systems for sequence-selective DNA binding. One of such models we are currently examining is Hoechst 33258. Hoechst 33258 is a synthetic bis-benzimidazole derivative that binds to B-DNA double helix preferentially at 5 contiguous AT base pairs in a nonintercalative fashion, resulting in substantial enhancement of its fluorescence, and which is consequently widely used in cytological and histological studies of DNA (10, 11). The interaction of Hoechst 33258 with DNA has been extensively studied by biophysical methods (12-14), footprinting (15,16),NMR (17), and X-ray (18-20) techniques. Although the molecular constitution of the crystals studied by two groups is similar, there exists some controversy over binding contacts between the dye and DNA (a shift of one base pair in binding site is proposed) (18, 19). We are interested and involved in understanding the molecular recognition process between groove binding ligands and DNA (2,4). There are two possible approaches to investigation of the molecular contacts between such ligands and DNA (i) keep the receptor (DNA) structure unchanged and alter the ligand structure systematically or (ii) retain the ligand structure unchanged and alter the receptor structure (21, 22). Although both approaches

* Author to whom correspondence should be addressed.

could, in principle, lead to better understanding of specific interactions between the ligand and receptor, there is one main disadvantage with the latter. Any change in the DNA base sequence or structure would be expected to perturb DNA conformation, which would therefore influence the binding of the ligand to the receptor. Such a study, without establishing the DNA structure itself, could lead to misinterpretation of the components of molecular recognition. For this reason we have adopted the first approach. Recently we reported the design, synthesis, footprinting, and preliminary binding studies on some new analogues of Hoechst 33258 (23). The present report deals with characterization of their DNA binding properties in detail, comparing the results with their parent molecule. The study focuses on favorable and unfavorable changes in the structure of the ligands in terms of binding properties and redesigning new molecules. Though all the compounds were studied to establish binding characteristics, compound 7 was examined in the most detail in comparison to the parent compound, Hoechst 33258.

Materlais and Methods Hoechst 33258 was obtained from Aldrich Chemical Co. The polynucleotides poly(dA-dT),' poly(dG-dC),and poly(dG-msdC) were from Pharmacia P-L Biochemicals. They were used without further purification. The calf thymus (ct)and T4 coliphage DNAs were from Sigma. Calf thymus (ct) DNA was deproteinated by the phenol extraction method and extensively dialyzed against 20 mM NaCl solution, pH 7.1, before use. All other reagents were of analytical grade. The synthesis and characterization of the new analogues have been reported (23). Stock solutions of the ligands were prepared in 20 mM NaCl solution, pH 7.1. All the experiments were carried out in 20 mM NaCl solution, pH 7.1, and at a room temperature of 20 i 1 O C unless otherwise specified. The concentrations of the ligands and DNAs are expressed in moles per liter. They were determined by using the molar extinction coefficient values (mM-l cm-') of the polynucleotides and DNAs (24) and c3% (compound 1, Hoechst 33258) = 41.0, tW (compound 2) = 30.4, (compound 3) = 32.8, t W (compound

'

Abbreviations: poly(dA-dT), alternating heteropolymer poly(dAdT)-poly(dA-dT); poly(dG-dC), alternating heteropolymer poly(dGdC).poly(dG-dC);?oly(dG-mSdC), alternating heteropolymer poly(dGmsdC).poly(dG-m dC); ct, calf thymus.

0893-228~/91/2704-0661$02.50/0 0 1991 American Chemical Society

Rao and Lown

662 Chem. Res. Toxicol., Vol. 4, No. 6, 1991

4) = 29.3, em (compound 5) = 33.7, em (compound 6) = 30.0, em (compound 7) = 33.3, and 9%(compound 8) = 25.0. Precipitation of ct DNA on addition of ligands above 0.15 ratio (r? is observed as reported earlier (25). However, no precipitation is encountered with the polynucleotides poly(dA-dT)and poly(dG-dC)up to an r'of 0.4. The W absorption spectra were recorded on a Hewlett-Packard 8450A diode array spectrophotometer. The fluorescence measurements were made on a Turner 430 spectrofluorimeter. The circular dichroism (CD) spectra were recorded on a Jasco ORD/UV 5 spectropolarimeter. The path length of the CD cell was 0.5 cm. The CD values are expressed as molar ellipticity ( 6 ) in deg cm2 dmol-'. The symbol r' is the input ratio of the concentrations of the ligand and DNA. The association constants (K,) for compounds 1, 2, 5 , and 7 were determined from the Scatchard equation (26). It is assumed that an independent noncooperative type of binding takes place between the ligands and DNAs: r/cf = KO(rb- r)

(where r = cb/cp; cb and cp are the concentrations of the bound ligand and nucleic acid, respectively, cf is the concentration of the free ligand, KOis the intrinsic binding constant, and rb is the binding stoichiometry). A plot of r/cf versus r gives a straight line with intercept K , on the r/cf axis and rb on the r axis. K , can also be determined from the known values of KOand rb (K, = K&). The experhenkd values were fitted by the method of least squares to obtain a straight line. c b was determined from AA/Ac, where AA is the decrease in absorbance of the ligand at a wavelength due to binding of it to DNA and Ac is the difference in the extinction coefficients of free and bound ligands. The extinction coefficient of the bound ligand is measured in the presence of 40-fold excess of DNA. The ethidium bromide (EtdBr) assay was carried out as reported earlier (27). The other CD experiments for evaluation of helix and groove specificity of the new compounds were carried out as reported by Rao et al. (24).

OH

H OH

6

',HC

P

N

!.

I

Results The structures of the new analogues and Hoechst 33258 are shown in Figure 1. The UV absorption spectra of two representative compounds 1 and 7 are shown in Figure 2. The spectra of all compounds are characterized by a peak and t above 300 nm (see Materials and Methods for A,, values) and a peak or shoulder below 300 nm. The higher wavelength peak could be due to the presence of benzimidazole, benzoxazole, and/or azabenzimidazole nuclei as in the case of Hoechst 33258 (2.5). The lower wavelength peak or shoulder could arise from the presence of the phenolic function at the terminus. Binding of Ligands to DNA. The binding of the ligands to nucleic acids, poly(dA-dT), ct DNA, and poly(dG-dC) was studied by a change in their absorption spectra in the presence of different DNAs (representative spectra are shown in Figure 2). In general, a red shift of the higher wavelength peak associated with a hypochromic effect results from the ligand-DNA interaction. The extent of the red shift is different for different DNAs. The red shift values range from 11to 26 nm with poly(dA-dT). The red shift in the UV absorption peak of the compounds with poly(dA-dT) is in the order 1, 8 > 2 > 5 > 7 > 3 > 6. All the spectra are characterized by isosbestic points at wavelengths that are different for different DNAs. The UV spectral characteristics of the interaction of the new compounds with DNAs are consistent with the reported results for compound 1, Hoechst 33258, and its interaction with DNA (14). The red shift in the longwavelength peak of the ligands upon binding to nucleic acids could be due to perturbation in the coupled chromophoric system of the ligands. This perturbation might occur as a sequel to an enhanced degree (compared with

H

H

'i

1

Figure 1. Structures of Hoechst 33258 (1) and the new analogues (2-8).

the free ligands) of conjugation of electrons via the benzimidazole (or benzoxazole or azabenzimidazole)and phenol rings of the ligands in the complexes with DNAs as in the case of distamycin or its synthetic analogues (24, 28). The red shift associated with the hypochromic effect in the long-wavelengthpeaks of the absorption spectra of the ligands upon interaction with DNAs indicates the binding of the ligands to DNA. The extent of red shift (23) follows the order poly(dA-dT) > ct DNA > poly(dG-dC) in the case of compounds 1 and 8. However, for the other compounds (2-7) the extent of the red shift is the same with all the DNAs. Similar changes in the UV spectrum of Hoechst 33258 (14) or distamycin (29) in the presence of DNAs were reported, which results from the binding of compounds to DNAs. The presence of an induced band in the CD spectra of DNA-ligand complexes in the 300-400-nm region c o n f i i s the complex formation between the new analogues and DNA. Representative CD spectra for compounds 1 and 7 with poly(dA-dT), ct DNA, and poly(dG-dC) are shown in Figure 3. The ligands alone do not exhibit any CD absorption. Similar induced bands characterize the complex formation between Hoechst 33258 or distamycin (or

Chem. Res. Toxicol., Vol. 4, No. 6,1991 663

DNA Binding of Hoechst 33258 Analogues

I1

I 0.60,

,

,

,

I

,

,

,

A

Figure 2. UV absorption spectra of compound 1 (panel I) and compound 7 (panel 11) alone (-) and in the presence of DNAs (- - -): (A) poly(dA-dT),(B) ct DNA, (C) poly(dG-dC). The numbers on the curves indicate P/D ratios corresponding to the spectrum. The concentrations of each compound: (in panel I) (A) 13.2 pM,(B)14.1 p M , and (C)12.0 pM;(in panel 11) (A) 11.7 pM,(B)12.7 rM, and (C) 9.1 pM.

its synthetic analogues) and DNA (13, 30, 31). A comparative study of the CD spectra obtained with different ligands in the presence of DNAs revealed the following characteristic features. Salient Features of the CD Spectra. (1) With Poly(dA-dT). (i) All the ligands give an induced Cotton effect due to complexation with poly(dA-dT) at different wavelengths in the 360-390-nm region, and the shapes of the induced bands are different for different ligands. (ii) All the spectra of ligand-polynucleotide complexes are characterized by isosbestic points. The crossover points for the ligand-polynucleotide complexes remain the same as in the case of polynucleotide alone. (iii) In the case of compounds 1, 5, and 7 with increasing saturation of the polynucleotide upon addition of ligand, a shoulder (with compounds 5 and 7) or a well-defined additional band (with compound 1) appears in the CD spectrum of ligand-polynucleotide complex around 340 nm, increasing with increased r’ (above 0.2) (28). This could be an indication of a nonspecific binding mode of the ligand to DNA. (iv) The interaction of the ligands does not alter

the conservative CD spectrum of the polynucleotide. (2) With ct DNA. (i) The induced bands due to complexation of the ligands with ct DNA are centered around 370-385 nm. (ii) All the spectra are characterized by isosbestic points different for different ligands up to r’of 0.15, and the crossover points for the ligand-DNA complexes remain the same as in the case of free DNA up to r’ = 0.15. However, it is difficult to comment on the preservation of the conservative spectrum of DNA upon interaction with ligands at or above r’ = 0.15. (iii) Compounds 1,2, and 5 give a shoulder around 340 nm in their CD spectra of complexes with DNA. (iv) Compound 2, unlike the other compounds, gives a broad band in the 320-400-nm region on interaction with ct DNA (figure not shown). This broad band resolves into two shoulders at 340 and 375 nm at a drug-to-DNA concentration ratio (r9 of about 0.08. (3) With Poly(dG-dC). (i) All the compounds including Hoechst 33258 give induced CD signals on interaction with poly(dG-dC) in the 330-390-nm region. The strength and shapes of the induced bands are different for different

Rao and Lown

664 Chem. Res. Toxicol., Vol. 4, No. 6, 1991

c

8 -X 0

I

-5

-1c

M,

1

I1

!a 15

10

5

3

3

X

-x

I

L

z

F!

o

0

-5 -1 0 -1 5

-15

i

-20 2

Figure 3. (I) CD spectra of the DNAs in the absence (-1 and presence (-- -1 of compound 1: (A) poly(dA-dT)(170pM), (B)ct DNA (280 rM),(C) poly(dG-dC)(170pM).(II) CD spectra of the DNAs in the absence (-) and presence (---) of compound 7: (A) poly(dA-dT) (190pM), (B)ct DNA (180 pM),(C) poly(dG-dC) (160 rM). The numbers on curves indicate the r'corresponding to the spectrum.

ligands. (ii) In general, the strength of the induced band due to complex formation between ligands and poly(dGdC) is weak compared to the induced band due to ligand-poly(dA-dT) complexes. (iii) A shift in crossover point is noticed with the complexes of all the compounds with poly(dG-dC). (iv) There is no pronounced effect on the conservative nature of the B-DNA spectrum of poly(dGdC) due to interaction with the ligands. (v) Hoechst 33258 (compound 1) gives a positive induced band at 335 nm and a small negative band at 380 nm on interaction with poly(dG-dC). This band at 335 nm appears to be similar to the band at 340 nm observed in the CD spectrum of poly(dA-dT) (or ct DNA) on complexation with compound 1 (or 5 or 7) at higher r', which arises mainly due to nonspecific interactions. This band almost disappears at high salt concentrations (Figure 4). (vi) A notable feature with compound 2 is that its complex with poly(dG-dC) has a negative band at 330 nm and a positive band at 385 nm (data not shown), which resembles the mirror image of the CD spectrum of compound 1 and poly(dG-dC) complex (Figure 3). Similar negative induced CD bands are observed with the NSC 101327-poly(dG-dC) complex (33) and the mPD derivative-poly(dG-dC) complex under 70%

s

I , ,,

00

0.2

,

0.4

,

,

0.6

,

,

0.8

, \ 1.0

Molarity of NaCl

Figure 4. Stability of DNA-ligand complexes with increasing salt concentration measured at saturation level with poly(dA-dT) (open symbols) and poly(dG-dC) (closed symbols). Compound 1 ( O ) , compound 2 (A),compound 5 (IJ), and compound 7 (v).

v/v ethanol conditions (34). Plausibly this could be due to external binding of the ligand to the DNA.

Chem. Res. Toxicol., Vol. 4, No. 6, 1991 665

DNA Binding of Hoechst 33258 Analogues

0

f

l

"

"

'

"

'

1

I

I

I

01

0.2

03

I

0.4

r

3 2 t 24

16

0

0

0

0.1

02

0.3

0.4

Figure 5. Variation of the molar ellipticity of the induced bands of the various DNAs as a function of r'for the following: (A) compound (170 pM),-ct DNA (A)(280 pM), and -poly(dG-dC) (0)(170 pM) complexes; (B)compound 2-poly(dA-dT) (0) l-poly(dA-dT) (0) (200 pM),-ct DNA (A)(180 pM), and -poly(dG-dC) (0)(170 pM) complexes; (C)compound 5-poly(dA-dT) (0) (190 pM),-ct DNA (A)(220 pM),and -poly(dC-dC) (0)(170 pM) complexes; (D)compound 7-poly(dA-dT) (0) (190 pM),-ct DNA (A)(180 pM),and -poly(dG-dC) (0)(160 pM)complexes.

Ionic Strength Dependence of the Binding. Figure 4 shows the ionic strength dependence of the interaction of the new ligands with the three DNAs. No significant change in the strength of the induced band of the complexes of the ligands (1-8) with poly(dA-dT) or ct DNA is observed with variation of NaCl strength from 20 mM to 1M. In contrast, the results reveal that the interaction of the ligands and poly(dG-dC) is significantly salt dependent and the induced CD signal due to binding of ligands to polymer diminishes with high salt concentration. From the salt titration experiment it can be understood that (i) the [e] values of the poly(dA-dT)-ligand complexes are insensitive to ionic strength of the medium, (ii) at any given ionic strength the [e] value of the compound 1poly(dA-dT) complex is much higher than the [e] value of any of the new analogue-poly(dA-dT) complexes, and (iii) the ligand-poly(dG-dC) interaction is salt dependent. Base Preference. General base preference of the ligands could be ascertained from the relative affinity of the ligands for DNAs with different base compositions. The affinity of the ligands for base pairs was evaluated from (i) measurements of the variation in the molar ellipticity (at peak wavelength) of the induced band as a function of the r', and (ii) the affinity constants of the compounds for various DNAs. Panels A, B, C, and D (for compounds 1, 2, 5, and 7, respectively) of Figure 5 show the change

in molar ellipticity of the induced bands due to the interaction of the ligands with DNAs as a function of the input ratio of ligand to DNA, r'. It is apparent from Figure 5 that the slope for poly(dA-dT)-ligand complexes is higher for all the ligands, and this indicates the general preference of the ligands for AT bases like the parent compound 1. Figure 4 shows that the complex of compound 7 and poly(dG-dC) is relatively stable at high ionic strength conditions compared with the other ligand-poly(dG-dC) complexes. A notable feature in Figure 5D is that the curves for complexes of poly(dAdT) and ct DNA have similar slopes. This might be an indication that compound 7 tolerates GC bases adjacent to AT base pairs. However, there is no indication of a specific binding of the new analogues to long GC stretches. Binding Constants (K,) of the Ligands. The binding constants of compounds 1,2,5, and 7 for poly(dA-dT) and ct DNA were determined by Scatchard plots (26). A representative plot for compounds 1 and 7 with poly(dAdT) is shown in Figure 6. The K, values determined by this method are given in Table I. The Scatchard method has been applied assuming that the interaction of the ligands with nucleic acids is of an independent noncooperative type. The linearity of the plots supports this assumption under the experimental conditions employed. The binding stoichiometry values (rb)of the ligands are

666

Rao and Lown

Chem. Res. Toxicol., Vol. 4 , No. 6, 1991

Table 11. Apparent Affinity Constants Determined by EtdBr Assay K,, M-I compd poly(dA-dT) ctDNA polv(dG-dC) 1 6.33 x 107 1.85 x 107 1.1 x 108 7.6 X lo6 2 2.64 X lo7 1.72 X 10' 5.10 X loe 4.5 X lo6 3 9.50 X lo6 4.72 X lo6 3.5 X lob 4 5.34 X lo6 1.43 X lo7 8.3 X lo6 5 2.38 X lo7 9.43 x i o 6 5.0 x 106 6 1.19 x 107 7 3.96 X lo7 3.85 X lo7 1.1 X lo6 4.7 x 106 8 1.90 x 107 1.22 x 107

10 0.8

0.6 0.4

0.2

n 0

0.02

0.04

0.06

0.08

0.10

0.12

r

Figure 6. Scatchard plots of r/cf against r for compound 1 (0) and compound 7 (A)with poly(dA-dT). Determination of binding constants, K,, and binding stoichiometry, rb, is described under Materials and Methods. Table I. Association Constants and Binding Site Size

Ka

poly(dA-dT), compd 1

2 5

7 a

lo6 M-I 58 f 0.6

(62)' 0.43 f 0.2 0.26 f 0.08 0.92 f 0.33

CtDNA, 103 M-1 30 f 0.1 (960)O 3.8 f 0.06 0.99 f 0.12 7.9 f 0.1

rb

poly(dA-dT) 0.095 (0.ll)O 0.11 0.10 0.105

&DNA 0.08 (0.048)" 0.09 0.08 0.095

From ref 32.

also given in Table I. For comparison, the K , values reported for Hoechst 33258 by Mikhailov et al. (32)are also given. The K , values obtained by this method for the parent compound are in agreement with the reported values of Mikhailov et al. (32). Recently, Loontiens et al. (21,22) have reported K , values for compound 1 in the order lo8 M-' for both poly(dA-dT) and ct DNA by fluorimetric analysis. However, the K, values clearly show that the affinity of the new analogues is somewhat weaker than the parent compound. Both parent and new analogues have similar binding site sizes. Scatchard analysis has not been performed for all the compounds because the four compounds studied showed very much the same trend as seen in the EtdBr displacement assay. The EtdBr assay data can be used to study the relative rather than their absolute affinity constants toward DNAs ( 2 3 ,while the detailed Scatchard analysis gives more accurate K, values. The new ligands including parent molecule are more effective in displacing EtdBr from its complexation site on poly(dA-dT) (35, 36) than in the poly(dG-dC) complex. The amount of the compound in the study required to quench fluorescence by 50% is measured in each case. Using these values, the relative binding constants were determined, and the data are presented in Table 11. These data also indicate that the compounds have more affinity for poly(dA-dT) than poly(dG-dC). It is apparent that compounds 5 and 7 are the better GC acceptors in the new series of compounds. DNA Helix Specificity. Hoechst 33258, compound 1, does not interact with double-stranded RNA or A-DNA and other non-B-DNA-type structures (12,37). The specificity of the new ligands for B-type structures has been evaluated using compound 7 as a representative model compound since it showed better interaction capacity with DNAs among the new ligands. The results of the CD studies for the evaluation of the DNA backbone specificity of compound 7 may be sum-

OK, were measured from Cw values (27). Buffer was 20 m M NaCl, pH 7.1 at room temperature. Excitation and emission wavelengths were 525 and 600 nm, respectively.

marized as follows: (i) Compound 7 shows no induced band in the CD spectrum of the A-conformation of ct DNA in the presence of 78% v/v alcohol (38) up to r' = 0.4 (figure not shown). However, 7 interacts with the B-form of ct DNA (in 20 mM-1 M NaC1) and gives strong induced bands (Figure 3). (ii) A similar study with poly(A).poly(U), a double-stranded RNA (with A-form structure), showed no induced band with the compound up to r' = 0.3 (figure not shown). These two experiments demonstrate that the new ligands do not recognize the A-DNA structure. (iii) The inability of compound 7 to interact with poly(dGm5dC) in the presence of 7 mM magnesium chloride (under which conditions the polymer exists in a Z-like structure) (39) is seen in the CD spectrum of the polymer. These experiments lend support for the strict preference of the new ligands for B-DNA structures. Binding via the Minor Groove of DNA. The X-ray crystallographic and physicochemical studies demonstrate that Hoechst 33258 binds in the minor groove of B-DNA (18,19,28,36,40).Binding of the new analogues to DNA via the minor groove is evident from the following experiments. (i) Interaction of the compounds with T4 phage DNA is demonstrated by UV and thermal melting studies (23). The observation of a red shift with hypochromic effect in the long-wavelength peak of the UV spectrum of compound 7 on the addition of T4 DNA confirms the binding of compound 7 to T4 DNA (figure not shown). A red shift of 6 nm is noted in the higher wavelength peak at saturation level. This suggests that the presence of glucose residues in the major groove does not prevent the interaction of the ligands with DNA and confirms the interaction site for these compounds is the minor groove. (ii) The displacement of ethidium bromide (EtdBr) from its intercalation site on nucleic acids provides additional evidence for the binding of these compounds in the minor groove of the DNA. (iii) The minor groove binding nature of the new ligands is further confirmed by a competitive binding experiment with distamycin, a well-recognized minor groove binder. Distamycin, by virtue of its strong affinity for poly(dA-dT), displaces compound 7 from its interaction site in the minor groove of poly(dA-dT) when added to an equilibrium mixture of compound 7 and poly(dA-dT). From Figure 7 it is evident that distamycin progressively displaces compound 7 and forms a complex with an induced band at 330 nm that is characteristic of distamycin-poly(dA-dT) interaction. This further supports the minor groove binding nature of the new Hoechst analogues.

Discussion The X-ray crystallographic studies of Hoechst 33258 with oligonucleotides indicated different binding sites from absolute A A T T (19)to ATTC with the piperazine ring at

DNA Binding of Hoechst 33258 Analogues /

I

i\

I

I

I

I

Chem. Res. Toxicol., Vol. 4, No. 6, 1991 667 I

I

1

10

-0 C

m

5

l-

-10

-20 230

270

310

350

340

430

480

A (nm) F i g u r e 7. CD spectra of poly(dA-dT) (210 WM)(- -) alone and (-) in the presence of compound 7 (curve 1) at r'= 0.12 and under the condition when distamycin was added t o the equilibrium mixture of compound 7 and poly(dA-dT) a t increasing concentrations of r' = 0.024, 0.12, and 0.222 (curves 2, 3, and 4, respectively.

-

the GC base pair (18). However, the footprinting studies revealed a site of at least four AT base pairs (16,41) with ready acceptance of a G at the end of the binding site (15, 16, 23). All these studies indicate that the parent compound, Hoechst 33258, is an AT sequence preferred ligand. Hoechst 33258 binds to the DNA in at least two modes. The primary binding is through specific interactions via hydrogen bonding and van der Waals contacts at AT sequences (18,191. The specific binding of Hoechst 33258 results in high fluorescence (11,13,21,22,42). The second mode of binding occurs at high concentrations of the dye molecule, is salt dependent, and quenches the fluorescence. The UV and CD studies of the present compounds indicated a single isosbestic point up to a ligand-to-DNAbase pair ratio of about 0.15, and above this ratio, a second isosbestic point is observed which could be due to nonspecific interactions such as electrostatic attractions. The CD study also indicates that the second mode of interaction occurs at higher concentration and is salt dependent (Figure 4). Specifically, with ct DNA above an r'of 0.15 it is difficult to comment on preservation of conservative B-DNA-like spectrum, indicating several modes of binding. The UV and CD spectra characteristics of the new ligands are comparable with those of the parent compound. The red shift of the UV absorption peak of the ligands in the presence of DNAs confirms the binding of the new compounds to DNAs. However, the UV study did not indicate any definite base preference of the compounds. CD titration of the ligand-DNA complexes with increasing salt concentrations indicated that specific hydrogen-bonding interactions take place between this class

of new ligands and poly(dA-dT) which are independent of the ionic strength of the medium. With poly(dG-dC) only nonspecific charge neutralization interactions are predominant as in the case of netropsin and distamycin (29). However, the interaction of compound 7 with poly(dG-dC) is relatively strong compared with the interactions of the other compounds, and also the CD titrations of compound 7 with ct DNA indicate that this compound can tolerate GC bases adjacent to AT sequences like imidazole-containing lexitropsins (31). The increased GC tolerance of the new compounds, especially compound 7, could be due to involvement of the extra nitrogen introduced into the ring at position 16 (see Figure 1, structure 1, for numbering) with 2-NH2of guanine base. However, compounds 5 and 6 which also possess this type of nitrogen atom show weak affinity for ct DNA compared with compound 7. Hoechst 33258 and the oligopeptide antibiotics distamycin and netropsin do not bind to non-B-DNA structures (29). The current results of compound 7 with poly(dGm5dC) in the presence of 5 mM magnesium chloride and poly(A).poly(U) suggest that this compound does not interact with these polymers, which exist in Z- and A-like structures, respectively. The new compound did not show binding to the A-form of ct DNA under high salt and ethanol conditions. However, the binding of the same compound to B-form of ct DNA is evident from Figure 3. This confirms the absolute specificity of the new compounds to B-DNA helix like the parent compound. Like the parent compound the new analogues bind in the minor groove of B-DNA. Relationship between the Structure of the Ligands and Their DNA Binding Characteristics. In general, modification in the structure of the parent molecule for altered specificity leads to a decrease in binding affinity to DNA as in the case of lexitropsins or distamycin and netropsin analogues (7, 31). The comparison of the structureDNA binding properties for the ligands revealed that the N(2) and N(3) (see structure 1 of Figure 1 for numbering) are necessary for effective DNA binding. Compounds 3 and 4, where N(2) and N(3) are replaced by an oxygen, respectively, showed little or no binding to DNAs. The introduction of an oxygen in place of N(1) (compound 2) results in alteration toward GC acceptance though not to a marked extent as seen in Figure 5B. Replacement of C(16) in compound 2 by a nitrogen as in the case of compound 5 indicated better alteration in specificity toward GC bases and also improved binding strength to DNAs compared with compound 2. However, the change in the position of the oxygen in molecule 5 as in the case of molecule 6 decreased the binding strength to a greater extent, indicating the requirement of the presence of N(2) for better binding. The best structure in terms of acceptance of GC base and overall binding strength for DNAs in the new series of molecules is that of compound 7,wherein C(16) is replaced by a nitrogen. This compound shows acceptance of GC bases and better interaction tendency with DNAs compared with the other new structures in the series. The highly modified structure 8 with greater curvature than the parent molecule does not show any alteration in specificity from Hoechst 33258 except in forming weaker complexes with DNAs. It is noteworthy that the introduction of one more piperazine ring does not result in acceptance of GC bases in any way. The physicochemical and footprinting studies (23) lead to the conclusion that the introduction of two azabenzimidazole rings in the Hoechst 33258 molecule in place of

668 Chem. Res. Toxicol., Vol. 4, No. 6, 1991

the two benzimidazole rings (compound 7 has one azabenzimidazole ring) might alter the specificity toward GC with a comparable binding strength to that of Hoechst 33258. Synthesis of such a molecule is in progress in this laboratory. The new analogues of the Hoechst 33258 exhibited significant topoisomerase inhibitory properties. Compound 7 is a better inhibitor of the topoisomerases I and I1 both in whole cells as well as in vitro than the parent compound (43). Currently, the new compounds are under investigation for their therapeutic potential.

Acknowledgment. This research was supported by a grant (to J.W.L.) from the Natural Sciences and Engineering Research Council of Canada.

References (1) Dervan, P. B. (1986) Design of sequence specific DNA binding molecules. Science 232, 464-471. (2) Lown, J. W. (1988) Lexitropsins: Rational design of DNA sequence reading agents as novel anticancer agents and potential cellular probes. Anti-Cancer Drug Des. 3, 25-40. (3) Pullman, B., and Jortner, J., Eds. (1990) Molecular basis of specificity in nucleic acid-drug interaction, Kleuwer Academic Publishers, London. (4) Lown, J. W. (1990) Molecular mechanisms of DNA sequence recognition by groove binding ligands: Biochemical and biological consequences. In Molecular Basis of Specificity in Nucleic Acid-Drug Interaction (Pullman, B., and Jortner, J., Eds.) pp 103-122, Kluwer Academic Publishers, London. (5) Kissinger, K. L., Krowicki, K., Dabrowiak, J. C., and Lown, J. W. (1987) Molecular recognition between oligopeptides and nucleic acids. Monocationic imidazole lexitropsins that display enhanced GC sequence dependent DNA binding. Biochemistry 26, 5590-5595. (6) Lee, M., Krowicki, K., Hartley, J. A., Pon, R. T., and Lown, J. W. (1988) Molecular recognition between oligopeptides and nucleic acids: Influence of van der Waals contacts in determining the 3'-terminus of DNA sequences read by monocationic lexitropsins. J. Am. Chem. SOC.110, 3641-3649. (7) Lown, J. W., Krowicki, K., Bhat, U. G., Skorobogaty, A., Ward, B., and Dabrowiak, J. C. (1986) Molecular recognition between oligopeptides and nucleic acids. Novel imidazole-containing oligopeptides related to netropsin that exhibit altered DNA sequence specificity. Biochemistry 25, 7408-7416. (8) Kissinger, K. L., Dabrowiak, J. C., and Lown, J. W. (1990) Molecular recognition between oligopeptides and nucleic acids. DNA binding specificity of a series of bis-netropsin analogues deduced from footprinting analysis. Chem. Res. Toxicol. 3, 162-168. (9) Rao, K. E., Zimmermann, J., and Lown, J. W. (1991) Sequence-selective DNA binding by linked bis-N-methylpyrrole dipeptides: An analysis by MPE footprinting and force field calculations. J. Org. Chem. 56, 786-797. (10) Latt, S. A., and Wohlleb, J. C. (1975) Optical studies of the interaction of 33258 Hoechst with DNA, chromatin, and metaphase chromosomes. Chromosoma 52, 297-316. (11) Stokke, T., and Steen, H. B. (1985) Multiple binding modes for Hoechst 33258 to DNA. J. Histochem. Cytochem. 33, 333-338. (12) Muller, W., and Gautier, F. (1975) Interaction of heterochromatic compounds with nucleic acids. A-T specific non-intercalative ligands. Eur. J. Biochem. 54, 385-394. (13) Bontemps, J., Houssier, C., and Fredericq, E. (1975) Physicochemical study of the complexes of '33258 Hoechst" with DNA nucleohistones. Nucleic Acids Res. 2, 971-984. (14) Steiner, R. F., and Sternberg, H. (1979) The interaction of Hoechst 33258 with natural and biosynthetic nucleic acids. Arch. Biochem. Biophys. 197, 580-588. (15) Harshman, K. D., and Dervan, P. B. (1985) Molecular recognition of B-DNA by Hoechst 33258. Nucleic Acids Res. 13, 4825-4835. (16) Murray, V., and Martin, R. F. (1988) Sequence specificity of lBI-labeled Hoechst 33258 damage in six closely related DNA sequences. J. Mol. Biol. 201, 437-442. (17) Parkinson, J. A., Barber, J., Doubls, K. T., Rosamond, J., and Sharples, D. (1990) Minor groove recognition of the self complementary duplex (dCGCGAATTCGCG)* by Hoechst 33258: A high field NMR study. Biochemistry 29, 10181-10190. (18) Pjura, P. E., Grzeskowiak, K., and Dickerson, R. E. (1987)

Rao and Lown Binding of Hoechst 33258 to the minor groove of B-DNA. J. Mol. Bioi. 197, 257-271. (19) Teng, M.-K., Usman, N., Frederick, C. A., and Wang, A. H. J. (1988) The molecular structure of the complex of Hoechst 33258 and the DNA dcdecamer d(CGCGATATCTGCG). Nucleic Acids Res. 16, 2671-2690. (20) Carrando, M. A. A. F. C. T., Col., M., Aymami, J., Wang, A. H. J., Mavel, G. A,, Von Boom, J. H., and Rich, A. (1989) Binding of a Hoechst dye to d(CGCGATATCGCG) and its influence on the conformation of the DNA fragment. Biochemistry 28,7849-7859. (21) Loontiens, F. G., McLaughlin, L. W., Dickmann, S., and Clegg, R. M. (1991) Binding of Hoechst 33258 and 4',6-diamino-2phenylindole to self-complementary decadeoxynucleotides with modified exocyclic base substituents. Biochemistry 30, 182-189. (22) Loontiens, F. G., Regenfuss, P., Zechel, A,, Dumortier, L., and Clegg, R. M. (1990) Binding characteristics of Hoechst 33258 with ct DNA, poly(dA-dT), and d(CGCGAATTCGCG): Multiple stoichiometries and determination of tight binding with a wide spectrum of site affinities. Biochemistry 29, 9029-9039. (23) Bathini, Y., Rao, K. E., Shea, R. G., and Lown, J. W. (1990) Molecular recognition between ligands and nucleic acids: Novel pyridine- and benzoxazole-containing agents related to Hoechst 33258 that exhibit altered DNA sequence specificity deduced from footprinting and spectroscopic studies. Chem. Res. Toxicol. 3, 268-280. (24) Rao, K. E., Dasgupta, D., and Sasisekharan, V. (1988) Interaction of synthetiac analogues of distamycin and netropsin with nucleic acids. Does curvature of ligand play a role in distamycin-DNA interactions? Biochemistry 27, 3018-3024. (25) Comings, D. E. (1975) Mechanisms of chromosome binding. VII. Hoechst 33258-DNA interaction. Chromosoma 52,229-243. (26) Scatchard, G. (1949) The attractions of proteins for small molecules and ions. Ann. N.Y. Acad. Sci. 51, 660-663. (27) Rao, K. E., Shea, R. G., Bathini, Y., and Lown, J. W. (1990) Molecular recognition between oligopeptides and nucleic acids: DNA sequence specificity and binding properties of thiazole-lexitropsins incorporating the concepts of base site acceptance and avoidance. Anti-Cancer Drug Des. 5, 3-10. (28) Zimmer; C., Reinert, K. E., Luck, G., Wahnert, U., Lober, G., and Thrum, H. (1971) Interaction of the oligopeptide antibiotics netropsin and distamycin A with nucleic acids. J. Mol. Biol. 58, 329-348. (29) Zimmer, C., and Wahnert, U. (1986) Non-intercalating DNAbinding ligands: Specificity of the interaction and their use as tools in biophysical, biochemical and biological investigations of the genetic material. Prog. Biophys. Mol. Biol. 47, 31-112. (30) Zimmer, C. (1975) Effects of antibiotics netropsin and distamycin A on the structure and function of nucleic acids. Bog. Nucleic Acid Res. Mol. Biol. 15, 285-318. (31) Zimmer, C., Luck, G., Burckhardt, G., Krowicki, K., and Lown, J. W. (1988) Effect of ligand structure on the dA-dT specific DNA groove binding of non-intercalating drugs. In Biomolecular Structure and Expression (Sarma, R. H., Ed.) Vol. 2, pp 301-303, Adenine Press, New York. (32) Mikhailov, M. V., Zasedatelev, A. S., and Gurskii, G. V. (1981) Mechanism of the recognition of AT pairs in DNA by molecules of the dye Hoechst 33258. Mol. Biol. (Moscow) 15,541-554 and 690-705. (33) Luck, G., Zimmer, C., and Baguley, B. C. (1984) Interaction of nucleic acids with a non-intercalative antileukemic compound containing bisquaternary heterocycles. Biochim. Biophys. Acta 782, 41-48. (34) Rao, K. E., Ramesh, N., Choudhury, D., Brahmachari, S. K., and Sasisekharan, V. (1989) Role of the environment in the interaction of nonintercalators with Z-DNA. J. Biomol. struct. Dyn. 7, 335-345. (35) Baguley, B. C. (1982) Nonintercalative DNA-binding antitumor compounds. Mol. Cell Biochem. 43, 167-181. (36) Langlois, R. G., and Jensen, R. H. (1975) Interaction between pairs of DNA-specific fluorescent stains bound to mammalian cells. J. Histochem. Cytochem. 27, 72-79. (37) Zesedatelev, A. V., Mikhailov, M. V., Krylov, A. S., and Gurskii, G. V. (1980) Mechanism of recognition of A-T pairs in DNA by Hoechst 33258. Dokl. Akad. Nauk. SSSR 255, 756-760. (38) Ivanov, V. I., Minchenkova, L. E., Minyat, E. E., Frank-Kamenetskii, M. D., and Schyolkina, A. (1974) The B to A transition of DNA in solution. J. Mol. Biol. 87, 817-833. (39) Behe, M., and Felsenfeld, G. (1981) Effects of methylation on a synthetic polynucleotide: The B-Z transition in poly(dGm6C).poIy(dG-m5C), h o c . Natl. Acad. Sci. U.S.A. 78,1619-1623.

Chem. Res. Toxicol. 1991,4, 669-677 (40) Sahar, E., and Latt, S. A. (1980)Energy transfer and binding competition between dyes used to enhance staining differentiation in metaphase chromosomes. Chromosoma 79,1-28. (41) Portugal, J., and Waring, M. J. (1988)Assignment of DNA binding sites for DAPI (4',6'-diamidine-2-phenylindone)and Hoechst 33258 (bis-benzimide). A comparative footprinting study.

669

Biochim. Biophys. Acta 949,158-168. (42) Jorgenson, K.F., Varshney, V., and van de Sande, J. H. (1988) Interaction of Hoechst 33258 with repeating synthetic DNA polymers and natural DNA. J. Biomol. Struct. Dyn.5,1005-1023. (43) Beerman, T.A., Sigmund, R., McHugh, M., L o w , J. W., Rao, K. E., and Bathini, Y. (1991)Biochim. Biophys. Acta (in press).

Formation of Glutathionyl-Spironolactone Disulfide by Rat Liver Cytochromes P450 or Hog Liver Flavin-Containing Monooxygenases: A Functional Probe of Two-Electron Oxidations of the Thiosteroid? Caroline J. Decker,+ John R. Cashman,t Katsumi Sugiyama,? David Maltby,t and Maria Almira Correia*pf Departments of Pharmacology and Pharmaceutical Chemistry and Liver Center, University of California, Sun Francisco, California 94143 Received July 15, 1991 We have previously reported that the diuretic thiosteroid spironolactone (SPL) inactivates rat liver microsomal cytochromes P450 [P450 (P450 3A and P450 2C11)] in a mechanism-based fashion, and we have identified two polar SPL metabolites (SPL-sulfinic acid and -sulfonic acid), formed in a partition ratio of =201 in such rat liver microsomal'incubations [Decker et al. (1989) Biochemistry 28,5128-51361, We proposed a t the time that these metabolites were most likely derived from further enzymatic (or nonenzymatic) oxidations of the one-electron oxidation product [SPL-thiyl radical (SPL-S')] and/or the two-electron-oxidized species [SPL-sulfenic acid (SPL-SOH)]. In those studies, glutathione (GSH) was found to attenuate both SPL-mediated P450 loss as well as polar metabolite formation by ~ 4 0 % .We have now reexamined this in greater detail and report that it is due to GSH trapping of an electrophilic oxidized SPL species to form an adduct that we have isolated and unambiguously characterized by mass spectral analyses as the glutathionyl-SPL adduct (SPL-SSG). Moreover, we have found not only that rat liver microsomal formation of this adduct is enhanced a t pH 9.0, the pH optimum for flavin-containing monooxygenase (FMO), but also that such adduct formation was indeed efficiently catalyzed by purified hog liver FMO. Because FMO oxidations of thiols are thought to entail a two-electron process to form the corresponding sulfenic acids, we infer that such a SPL-SSG adduct most likely reflects FMO-catalyzed oxidation of SPL to SPL-SOH, which on leaving the FMO active site is then trapped by GSH. Moreover, if not intercepted by external nucleophiles, FMO-generated SPL-SOH may attack rat liver microsomal P45Os, thus accounting for the GSH-inhibitable component of SPL-mediated P450 loss. This possibility is strengthened by the finding that GSH-mediated attenuation of such a SPL-mediated P450 loss was diminished by thermal inactivation of rat liver microsomal FMO. The isolation and characterization of the SPL-SSG adduct, we believe, not only rationalize the previously reported GSH-mediated attenuation of P450 loss but also provide the first direct evidence for the intermediacy of GSH conjugates in reduction of FMO-generated sulfenic acids. Furthermore, because we detected no polar oxidized SPL metabolites in purified FMO-catalyzed systems even in the absence of GSH, it appears that these metabolites must be derived from the one-electron-oxidized SPL-S' species.

Introduction The antiminer~ocorticoidspironolactone (spL)lor its deacetylated (SPL-SH) is known to inactivate the adrenal and testicular cytochromes P450 (P450s),in various species (1-5). It is now well established that SPL deacetylation is critical to its P450 inactivation, since tissues deficient in microsomal deacetylases or blockade

* T~whom correspondence should be addressed at the Department of Pharmacology, Box 0450, University of California, San Francisco, San Francisco, CA 94143. 'Department of Pharmacology and Liver Center. t Department of Pharmaceutical Chemistry and Liver Center. 0893-228x/91/2704-0669$02.50/0

of deacetylation by specific inhibitors of microsomal esterases result in little or no SPL-mediated P450 inactivation (1, 3, 4). We have shown that in vivo administration of Abbreviations: CID, collision-induced dissociation; MS/MS, mass spectrometry; P450, cytochrome P450; SPL-SH, deacetylated spironolactone; DEX, dexamethasone; DETAPAC, diethylenetriaminepentaacetic acid; DTT, dithiothreitol; ESR, electron spin resonance; FMO, flavin-containing monooxygenase; GSH, glutathione; GSSG,oxidized glutathione; GS', glutathionyl radical; HPLC, high-performance liquid chromatography; +LSIMS, liquid secondary ion mass spectral analyses in the positive mode; NADPH, nicotinamide adenine dinucleotide phosphate, reduced form; RSOH, organic sulfenic acid; RSSC, organic (aryl alkyl) glutathionyl disulfide; SPL, spironolactone; SPLSSG, S P L GSdadduct; SPL-S', SPL-thiyl radical; SPL-SOH, SPL-sulfenic acid.

0 1991 American Chemical Society