The interaction between ethidium bromide and nucleic acids shows a pronounced metachromatic effect which has been used to obtain quantitative data on the.
J. Mol. Biol. (1965) 13, 269-282
Complex Formation between Ethidium Bromide and Nucleic Acids M.J.
WARING
Sub-department of Chemical Microbiology, Department of Biochemistry University of Cambridge, Cambridge, England (Received 23 March 1965, and in revised form 12 May 1965) The interaction between ethidium bromide and nucleic acids shows a pronounced metachromatic effect which has been used to obtain quantitative data on the process of complex formation. Ethidium binds strongly to both DNA and RNA at sites which appear to be saturated when one drug molecule is bound for every 4 or 5 nucleotides. After the primary sites have been filled, a secondary binding process can occur, leading to the precipitation of a complex containing one drug molecule bound per nucleotide. The strong primary binding to DNA is not influenced by the base-composition or by denaturation ofthe DNA, but is sensitive to changes in the salt concentration, particularly when magnesium ions are present. Addition of magnesium chloride causes a marked reduction in the strength of the interaction without significantly affecting the number of sites available to bind the drug. The process of complex formation is shown to be reversible in solution by demonstrating an exchange reaction between free and bound ethidium. Complexes between ethidium and DNA can be dissociated by using a cation-exchange resin. The binding of ethidium to DNA follows a similar pattern to that found with proflavine, suggesting that the forces involved in the binding ofthe two drugs may be similar.
1. Introduction Ethidium bromide is a trypanocidal drug which has been found to inhibit nucleic acid synthesis in a variety of organisms (Newton, 1957; Kerridge, 1958; Kandaswamy & Henderson, 1962; Tomchick & Mandel, 1964). Evidence concerning its probable mode of action in vivo has been reviewed by Newton (1963), leading to the conclusion that the action of the drug on living organisms is correlated with its ability to interfere with nucleic acid synthesis. In cell-free systems the drug has been shown to inhibit the DNAdependent DNA polymerase and RNA polymerase of Escherichia coli (Elliott, 1963; Waring, 1964). Moreover, Elliott (1963) described the formation of a complex between ethidium and DNA, which raises the possibility that the primary action of the drug might be to bind to the DNA in living cells and thus interfere with the replication and transcription of the genome. A similar suggestion was made by Seaman & Woodbine (1953) to account for the antibacterial action of dimidium bromide, the lO-methyl homologue of ethidium bromide. The experiments described in this report were undertaken with the objects of characterizing the ethidium-DNA complex and obtaining quantitative data on its formation. The interaction was studied by the spectrophotometric method employed 269
270
M. J. WARING
by Peacocke & Skerrett (1956) to investigate the interaction between proflavine and nucleic acids. This method was particularly suitable for studies on the ethidium-DNA complex because of the large metachromatic effect on the spectrum ofthe drug (Elliott, 1963). Measurements made under conditions similar to those used in the assay of DNA-dependent RNA polymerase (see below) have been used to interpret the inhibition of this enzyme system by ethidium bromide in terms of the binding of the drug to the DNA primer (Waring, 1965). A linear relationship was found between inhibition of RNA polymerase activity and the amount of ethidium complexed to the primer, in agreement with the hypothesis that the action of the drug on isolated enzyme systems, as well as its growth-inhibitory activity in vivo, is due to its capacity to form complexes with nucleic acids.
Ethidium bromide: 2: 7 diamino-u-phenyl-Iu-ethyl phenanthridinium bromide.
2. Materials and Methods (a) Nucleic acid preparations
Micrococcus Iqeodeikticus DNA was prepared from spray-dried cells (California Corporation for Biochemical Research) according to the method of Marmur (1961). After 3 extractions with chloroform-octanol and treatment with RNase, the material was further purified by shaking it at room temperature for 30 min with an equal volume of water-saturated redistilled phenol. The emulsion was broken by centrifuging at 12,000 rev.fmin for 10 min, the upper layer removed and precipitated with 2 vol. of ice-cold 95% ethanol and the fibrous DNA collected on a glass rod. The product was dissolved in one-tenth strength saline-citrate (Marmur, 1961) and extracted with ether 5 times to remove traces of phenol, freed from ether under a current of air at 37°C, and stored at 0 to 4°C in the presence of a few drops of chloroform. DNA from bacteriophage T2 was isolated as already described (Waring, 1965). Calf -thymus DNA was a product of the Sigma Chemical Company. E. coli DNA, a gift from Dr K. McQuillen, was extracted with phenol and reprecipitated with ethanol as described above before use. Solutions of these materials were also stored at 0 to 4°C in the presence of chloroform. Rat liver ribosomal RNA prepared according to Munro & Korner (1964) was kindly provided by Dr A. J. Munro. Highly polymerized yeast RNA was obtained from British Drug Houses, Ltd. These preparations were dissolved at 2 mgfml. in 0·01 M·tris-HCI (pH 7'4) containing 0·01 M-NaCI and were frozen until required.
ETHIDIUM-NUCLEIC ACID COMPLEXES
271
(b) Ethidium bromide
Generous supplies of ethidium bromide were given by Dr G. Woolfe, of Boots Pure Drug Co. Ltd., Nottingham, England. Dr Woolfe also provided 14C·labelled ethidium bromide of specific activity 0·17 p.c/mg. Solutions were prepared in water at 1 mg/mI., and stored at 0 to 4°C in the dark. (c) Buffer
All solutions used in the course of this work were prepared with doubly distilled water. Unless otherwise noted, experiments were performed in the presence of 0·04 M·tris-HCI buffer (pH 7'9) containing recrystallized primary standard grade tris (Sigma Chemical Co.). (d) Spectrophotometricmeasurements
Readings were made in a Zeiss PMQ II spectrophotometer. Solutions of ethidium bromo ide in the standard tris-HCI buffer were found to follow Beer's law over the whole range of concentrations tested (up to 2·5x 10- 4 M); and there was no evidence at any stage of adsorption of the drug or its complexes with nucleic acids on to the surfaces of glass spectrophotometer cuvettes, At 480 mp' the molar extinction coefficient of the drug was found to be 5600. Measurements were made at room temperature and no significant variation was noted in values obtained between 18 and 25°C. Readings on solutions containing nucleic acids were made with reference to a blank containing the same concentration of nucleic acid in buffer; this procedure corrected for any slight absorption of these materials in the range 400 to 600 mp.. The rate of formation of complex appeared to be very rapid. Readings obtained a few minutes after mixing solutions of the drug and DNA were found to be unchanged when repeated at frequent intervals over the following 24 hr. In practice, solutions were allowed to stand at room temperature for at least 10 min before measurements were made.
3. Results (a) Treatment of results When the sites on nucleic acids which bind a drug are of a single type and behave independently of each other, the binding process can be described in simple massaction terms, giving: c(n - r) (1) k= r or r r n -=--(~) c k k where k is the dissociation constant of the complex, c is the molar concentration of free drug, r is the number of drug molecules bound per nucleotide and n is the number of binding sites per nucleotide. The functions rand n are expressed in terms of the nucleotide as the unit of nucleic acid, because of the uncertainty about precise molecular weights of DNA and RNA. According to this treatment, a plot of ric versus r would yield a straight line of gradient - 11k, with the intercept on the r-axis equal to n. If, however, the binding process involves more than a single type of site, or binding at one site affects the interaction at neighbouring sites, then a more detailed treatment must be considered (Peacocke & Skerrett, 1956; Steiner & Beers, 1961; Cavalieri & Nemchin, 1964). Such modes of interaction are indicated by curvature in plots of ric versus r, In the present experiments, values of rand c were calculated from spectrophotometric measurements of the fraction of ethidium in a complexed form in solutions containing known amounts of drug and nucleic acid. Concentrations of nucleic acid are expressed in terms of molarity with respect to nucleotide phosphorus.
272
M. J. WARING
(b) Measurements of the metachromatic effect
The spectral shift associated with the binding of ethidium to nucleic acids is illustrated in Fig. l(a). Both DNA and RNA are able to cause a shift in the visible region spectrum of the drug, moving the absorption maximum from 479 mfl- to 516 to 518 mfl-. This effect is evident as a change in colour from yellow-orange to bright pink. Figure l(a) shows the results when the nucleic acids are present in excess, that is to say, further additions produce no significant change in the spectrum ofthe drug. Minor differences are evident between the shifts produced by the various polynucleotides, but the general effect is the same.
450
550
600 400
450
500
550
600
Wavelength (m,u) FIG. 1. Effect of nucleic acids on the absorption spectrum of ethidium bromide. Solutions contained 1·25 X 10- 4 M-ethidium bromide and the absorbancy was measured using a I-em light path. In panel (a) the concentration of each nucleic acid was 1·5 X 10- 3 M. In panel (b) T2 DNA was present at the following concentrations: curve A, zero; B, 1·5 X 10- 4 M; C, 3 X 10- 4 M; D, 5 X 10- 4 M; E, 1·2 X 10- 3 M.
Figure l(b) shows the effect of increasing concentrations ofT2 DNA on the absorption spectrum of ethidium bromide. The peak can be seen to shift progressively towards a limit (curve E) which represents the spectrum of the drug in a fully complexed form. All the curves pass through an isosbestic point at 510 tsu», indicating that they result from the contributions of two forms of ethidium, free and bound, each form having a characteristic absorption spectrum. The proportions of free and bound drug can be estimated most accurately from measurements of the absorbancy in the region where the difference between curves A and E is greatest. Suitable wavelengths for this purpose are 460, 470 and 480 mfl-. At these wavelengths the absorbancy in curve C corresponds to that of a mixture in which 54'7% of the drug is complexed: when the predicted spectrum of such a mixture was interpolated from curves A and E, it agreed with curve C to within 0·007 optical density unit over the whole range between 400 and 600 tup:
ETHIDIUM-NUCLEIC ACID COMPLEXES
273
The technique used in binding studies was that of Peacocke & Skerrett (1956). A solution of ethidium bromide containing a high concentration of nucleic acid was prepared and diluted with a solution of the drug alone at the same concentration as in the mixture; in this way measurements at the chosen wavelengths were made on solutions containing a fixed drug concentration and decreasing nucleic acid concentrations. Results from an experiment with T2 DNA are given in Fig. 2, showing the
o
1
2
3
DNA concentration (MxI03)
FIG. 2. Absorbancy of ethidium bromide in the presence of T2 DNA. Measurements were made with a. Lcm light path on solutions containing 1·25x 10- 4 M-ethidium bromide.
constant absorbancy at 510 mfL and the tendency of the absorbancy at 460 and 480 mfL to attain a constant value in the presence of high DNA concentrations. At 540 tsu», however, the absorbancy continues to rise slowly as the DNA concentration increases. Thus while the absorbancy of complexed ethidium at 460 and 480 mfL appears to be independent of the total amount of bound drug, the absorbancy at 540 mfL is affected by the level of binding. Peacocke & Skerrett (1956) noted a similar effect in their studies on the interaction between proflavine and nucleic acids. It is possible that this dependence of the absorbancy at 540 mfL on r reflects the existence of nearestneighbour interactions between bound drug molecules. Since estimates of r from spectrophotometric measurements are only possible when the absorbancy of the bound drug does not vary with r, calculations were confined to results obtained at 460,470 and 480 mfL and the average value taken. Individual estimates ofthe fraction of bound drug rarely differed from the average by more than 0·5 % of the total drug concentration.
(c) Binding of ethidium to DNA and RNA A binding curve calculated from the results in Fig. 2 is shown in Fig. 3(a). The diagram includes results from similar measurements with RNA. It can be seen that the binding to both types of nucleic acid rose rapidly to an r value of about 0'18, and only increased much more slowly after this binding ratio was reached. However, at values of r below 0,18, the concentration of free drug, c, was so low that accurate
274
1'r1. J. WARING
0·3.-------.-----,-----,.--=.e--.,
0'2
0'1 (0)
oL-.-----L-----"--------'------' 8 2 4 6 Free ethidium bromide (M X 10')
0·2..-----...,.------,----".-----,
0"
o
2
4
6
Free ethidium bromide (M X 106)
FIG. 3. Bindingofethidium to T2 DNA (-0-0-) and rat liver ribosomal RNA (-e-e-). In panel (a), the total drug concentration was 1·25 X 10 - 4 M, and measurements of absorbency were made using a I-em light path. In panel (b) the drug concentration was 2·5 X 10- 5 M and a 4-cm light path was used.
estimation was impossible. In order to measure c in this region of binding, the spectrophotometric method was adapted to cover a lower range of c values with increased sensitivity to small changes in c. This was achieved by lowering the total drug concentration by a factor of 5 and making measurements with a 4-cm light path. Results obtained under these conditions (Fig. 3(b)) show that the level of binding to DNA in the presence of a given concentration offree drug is higher than that reached by RNA. However, Fig. 3(a) shows that this situation is reversed at higher values of c. Thus the binding curves for DNA and RNA must cross at some point in the region of c equal to 5 to 10 X 10 - 6 M. Plots of ric uersu« r for T2 DNA and ribosomal RNA are included in Figs 4 and 5. It is clear that these plots are not linear, although a straight line could be drawn through most of the points obtained with T2 DNA. This straight line yielded the values k = 5XlO- 7 1\I and n = 0·19. The plot for RNA (Fig. 5) shows considerable curvature, so that the difference between the binding of ethidium to DNA and to l~NA seen in Fig. 3 could not be ascribed to a major difference in one of the parameters k or n. The significance of the curvature in the plot for RNA is discussed later.
ETHIDIUM-NUCLEIC ACID COMPLEXES
3
I
I
(a)
\. .\
2
-
\ \
-
,,
>
