J. Mol. Biol. (1975) 95, 103-123

DNA-Ethidium Reaction Kinetics : Demonstration of Direct Ligand Transfer Between DNA Binding Sites JEFFREY

AND DONALD M. BROTHERS

L. BREsLoF@

Departments of Chemistry and Molecular Biophysics and Biochemistry Yale University, New Haven, Conn. 06520, U.S.A. (Received 14 October 1974) Study of the relaxation kinetics of the interaction of ethidium and DNA reveals a novel and potentially important general binding mechanism, namely direct transfer of the ligand between DNA binding sites without requiring dissociation to free ligand. The measurable relaxation spectrum shows three relaxation times, indicating that three bound dye species are present at equilibrium; about 80% of the dye is in the major intercalated form. For each relaxation the reciprocal relaxation time varies linearly with concentration up to very high DNA concentrations. The failure of the longer relaxation times to plateau at high concentration can be accounted for by including a bimolecular pathway for conversion from one complex form to another. This we envisage as direct transfer of an ethidium molecule, bound to one DNA molecule, to an empty binding site on another DNA molecule. Additional evidence for this direct transfer mechanism was obtained from an experiment showing that DNA (which binds ethidium relatively rapidly) accelerates the binding of ethidium to poly(rA)*poly(rU), presumably by first forming a DNA-ethidium complex and then transferring the ethidium to RNA. The bimolecular rate constant for transfer is found to be about four times larger than the constant for intercalating the free dye. The transfer pathway thus provides .a highly efficient means for the ligand to equilibrate over its DNA binding sites, especially at high polymer concentration. The potential importance of direct transfer for DNA-binding regulatory proteins is emphasized.

1. Introduction Proteins

specific

for binding

to a particular

RNA

or DNA.

sequence

can be expected

is documented is the bc repressor, which binds to a variety of polynucleotides with an affinity that, though much lower than for the operator sequence, would be sufficient to ensure extensive binding at the nucleic acid concentrations present in a cell (Riggs et al.,

to possess

a general

affinity

for nucleic

acids.

One example

where

1972; Lin and Riggs, be&n,

1972a).

siently

or not

this

1972). Another example is RNA polymerase (Hinkle As a consequence, many regulatory proteins may exist

at all in free

solution,

spending

most

of their

time

& Chamonly tran-

in association

with

t Present address: Division of Chemistry and Chemical Engineering, California Institute Technology,

Pasadena,

Calif. 91109, U.S.A. 103

of

104

J. L.

BRESLOFF

AND

D. M. CROTHERS

nucleic acid. In such cases a crucial question in understanding the mechanism by which the correct binding site is sought out by the ligand is, how does the protein dissociate from the incorrect or non-productive complexes in which it may be trapped by its general nucleic acid affinity? Even simple DNA ligands such as actinomycin can require hundreds of seconds for dissociation (Miiller & Crothers, 1968). If transfer to the correct binding site requires dissociation to free ligand, the dissociation process could easily become rate limiting for the final binding reaction. Studies of the rate and mechanism of ligand binding to DNA have conventionally focused on the reaction between free ligand and its binding site. This is true both for small molecule ligands such as proflavine (Li & Crothers, 1969) and actinomycin (Miiller & Crothers, 1968), and also for proteins such as the lac repressor (Riggs et al., 1970) and RNA polymerase (Hinkle & Chamberlin, 19726). In the latter case there is evidence that the binding rate is limited by the rate of dissociation of the nonproductive complexes. The point of the present paper is to document with a simple model system the existence of another kind of binding mechanism, namely the direct transfer of a ligand from one DNA molecule to another without the necessity for dissociating to free ligand. The potential importance of direct transfer is that it allows a faster path for ligands to seek out the preferred binding site than provided by a mechanism requiring dissociation. This is illustrated by comparison of proflavine binding kinetics (Li $ Crothers, 1969) with the present system, which utilizes ethidium bromide, a muchstudied DNA intercalating agent (Fuller & Waring, 1964; Waring, 1965; LePecq & Paoletti, 1967 ; Crawford & Waring, 1967 ; Bauer & Vinograd, 1968; Saucier et al., 1971; Freifelder, 1971; Kreishman et al., 1971). Even though the rate constants for binding and dissociating free ethidium are slower than for proflavine, ethidium equilibrates over its binding sites at high DNA concentration faster than proilavine. The difference is that ethidium can be transferred directly between binding sites on two DNA molecules, whereas no such step was found for proflavine. A simple physical model of the direct transfer process would seem to imply transient binding of both DNA molecules by the ligand. Hence ligands that show this property should be potentially bifunctional. In the ethidium molecule the primary binding site is clearly the phenantbridinium ring, and a possible candidate for the secondary binding site is the phenyl ring; we are presently trying to clarify this question by studying the transfer process in ethidium analogs (D. Ryan, unpublished experiments). Of more direct biological relevance is the hypothesis that potential bifunctionality may be kinetically advantageous in regulatory proteins that bind to DNA. The bc repressor may be constructed along these lines: the tetrameric structure would provide two binding sites for an operator with (partial) two-fold symmetry (Gilbert & Maxam, 1973; Sobell & Jain, 1972; Steitz et al., 1974) if the protein symmetry was rigorously maintained. Since only one of these two potential sites is occupied, the affinity of the other must be lowered by loss of symmetry. Transient binding of the weaker site to a second DNA molecule could result in formation of an intermediate with full symmetry. If the bond to the original DNA partner should break, the ligand will be transferred, with its “weak” binding site becoming the “strong” site, and vice versa. This kind of ligand transfer could circumvent kinetic problems faced by the repressor in finding the operator in the midst of a high concentration of non-specific DNA with appreciable binding afiinity and slow dissociation rate.

DNA-ETHIDIUM

REACTION

105

KINETICS

The experiments described here are relaxation kinetic measurements on the ethidium-DNA complex, using methods described previously (Li & Crothers, 1969) a!ong with equilibrium dialysis determination of binding equilibria. We find that only about 80% of the bound dye is the major intercalated species, with the remainder distributed between two minor binding modes. The direct transfer mechanism was deduced from the concentration dependence of the relaxation times, and confirmed by showing that DNA (which binds ethidium more rapidly than RNA) catalyzes the binding of ethidium to poly(rA) *poly(rU). One surprising feature of the direct transfer process is that the bimolecular rate constant for transferring an ethidium molecule is three to six times larger than for intercalating a free ethidium molecule, so that the DNA-bound ligand reacts more rapidly than a free ligand.

2. Materials and Methods (a) Materials Ethidium bromide was purchased from both Boots Pure Drug Co. Ltd, Nottingham, England (batch no. DR1731) and Cal Biochem., Los Angeles, Calif. (lot 100301 B grade). These samples were indistinguishable experimentally. Recent studies (Kreishman et aZ., 1971; Thomas & Roques, 1972) using high-resolution proton magnetic resonance speotroscopy have revealed the possibility of samples of EthBrt not completely alkylated, as well as the presence of the impurity, ethanol. 100 MHz proton magnetic resonance spectra of concentrated (0.09 M) EthBr in dimethyl sulfoxide (d6) were taken to investigate these effects. This solvent (used by Thomas & Roques, 1972) minimizes stacking interactions between EthBr molecules, making peak assignments simpler to interpret than with 2Hz0, as well as separating the free and bound methyl resonances that overlap in 2H,0. Following the analysis by Kreishman et al. (1971) the percentage alkylation, found by taking the weighted ratio of the integrals of the 5 ethyl proton resonances to the 10 aromatic proton resonances was 98&7. Hence our EthBr is totally alkylated, within experimental error. Our spectra further reveal the existence of a significant amount of ethanol as an impurity, although accurate concentration determination is difficult due to the broad resonances that arise. Elementary analysis (Alfred Bernhard& West Germany) on EthBr extensively dried under vacuum at room temperature gives results quantitatively consistent with an ethanol impurity level equal to 0.46 mol/mol EthBr. The molecular weight of the EthBr complex is then 415.5. The purity of EthBr was further monitored by paper chromatography (Schleicher & Schuell, Keene, New Hampshire; no. 2043-B) using the solvent system of CHaCH,OH/H,O/NH,OH [135/55/10, by vol.]. Application of approx. 15 ~1 of a solution of EthBr wibh an absorbance at 480 nm equivalent to 0.9 O.D. units on a quadrant of the paper resulted in a single band, detected under U.V. light, with ZZr = 0.6. A negligible amount of fluorescent material remained at the origin. However, at concentrations above the equivalent of 2 O.D. units, two diffuse bands become discernible; at a concentration equivalent to 63 O.D. units, zones with R, = 0.7 and 0.8 occur. This anomalous concentration effect was further verified by cutting out both zones, redissolving and submitting them to chromatography. Only a single zone with RF = 0.6 resulted. Calf thymus DNA (type 1) was purchased from Sigma Chemical Co., St. Louis, MO.; poly(rA) and poly(rU) were purchased from Miles Laboratories. The DNA was sonicated and purified by first dissolving in buffer I (0.006 M-Na,HPO,, 0.002 M-NaH,PC4, 0.001 M-EDTA(Na,), O*OOl M-sodium oacodylate, pH 7.2), stirring overnight at 4°C to form a gel, then sonicating with a Branson S125 ultrasonicator at ice-bath temperature for 5 min, followed by bubbling with nitrogen for 5 min, and repeating 6 times. The DNA was centrifuged and then purified by extraction with phenol, followed by extraction with ether and bubbling with nitrogen. It was then extensively dialyzed (5 times) into buffer 1. Poly(rA) .poly(rU) was formed from its constituent strands by first extensively dialyzing each component in buffer I, followed by combination in equimolar proportions. After 7 Abbreviation

used: EthBr,

3,8-diamino-6-phonyl-5-ethyl-phenanthridinium

bromide.

106

J. L.

BRESLOFF

AND

D. M. CROTHERS

~~PPI’OX. 1 day, allowed for equilibrium to be reached, these RNAs were then sonicated and treated as described above for DNA. Concentrations were determined spectraphotometrically assuming e (260 nm) = 6550 for calf thymus DNA (average of several literature values); E (260 nm) = 7140 for poly(rA)*poly(rU) (Krakauer, 1969). Molecular weights of the son&ted DNA and RNA were estimated to be roughly 105 (Schmechel & Crothers, 1971). All experiments were done in the presence of 1 M-Na+ (buffer II) prepared by addition of an aqueous solution of 4.932 M-NaNO, to buffer I in a ratio of 1: 4 (v/v), pH 6.6. The volume change on mixing these solutions has been neglected in determining concentrations, it being a 4.2 ml-decrease on addition of 200 ml NaNOs to 800 ml buffer I. Stock solutions of EthBr were prepared in buffer I to a concentration of 5 x 10-s M. This, as well as all EthBr-DNA complexes, was stored in the dark at 4°C. Care w&s taken to avoid exposing solutions containing EthBr to light for long periods, as spectral changes were observed when the free dye was left uncovered for several days in the laboratory. All stock DNA solutions were stored at 4”C, or for longer periods at -20°C. Their conoentration was determined at frequent intervals during this study. DNA concentrations are expressed in base-pair units.

(b) Methods The extinction coefficient of EthBr was determined by drying under vacuum at room temperature for 3 to 24 h, as well as under vacuum at 100°C for several hours. All procedures gave essentially identical results with the extinction coefficient at low concen(&0*3%) at 480 nm. (Neglecting the tration in buffer I at 19*O”C equal to 5850 (M cm)-l ethanol impurity gives a value within 1 o/o of that found by Waring (1965), 5600, where no correction was applied.) Deviations from Beer’s law will be described in a later paper. Because of the limited solubility of EthBr in buffer II, the corresponding extinction coefficient was determined by dilution of EthBr in buffer I by addition of salt with a correction for the volume change due to mixing. A value of 5820 at 480 nm (19.O’C) was obtained. The extinction coefficient of EthBr bound to calf thymus DNA was determined directly by equilibrium dialysis and will be described elsewhere (Bresloff & Crothers, manuscript in preparation). It was found equal to 4150 at the X,,, of the bound dye (521 nm) et 19.O”C. A description of the double beam temperature-jump apparatus used in this study has been given by Crothers (1971). Recent acquisition of a transient recorder (model 802, Biomation) has substantially increased the accuracy of analysis of relaxation spectra. Typically it was used in the pretrigger mode so that the signal level directly preceding the perturbation is recorded, along with the relaxation signel; 4 multiple exposures of the data at fractions of 1, l/2, l/5 end l/10 the sweep time are then taken with the further option of magnifying the signal amplitude available. Faster components of the spectrum were obtained by simultaneously recording this part of the signal on a storage oscilloscope. In general, 4 to 5 temperature-jumps at the wavelength of analysis were taken and the relaxation parameters averaged. A concentration dependence of the relaxation times was obtained by perturbing a series of solutions of varying DNA concentration, but holding the extent of binding (r) constant. This requires that each solution has the same monomer free dye concentration, a condition that was achieved by either preparing a concentrated DNA solution at the given monomer concentration C, and diluting this with a dye solution whose concentration equals C,, or by dialyzing a series of solutions of varying DNA concentration all to the same G, value. Both techniques were used and gave identical results. Holding constant the extent of binding is advisable as the kinetics of intercalation may vary with T. Further, for complex mechanisms such as that proposed here, one may find that the relaxation times cannot be plotted as a function of the sum of concentrations of monomer free dye and free binding sites, i.e. simultaneous variation of each of these parameters would preclude a S-dimensional plot. Holding the free dye concentration (and hence the P value) constant allows for analysis of the relaxation spectra directly in terms of the DNA ooncentration.

DNA-ETHIDITJM

REACTION

KINETICS

107

3. Results The kinetic spectrum describing complex formation between ethidium bromide and calf thymus DNA at high ionic strength (1 M-Na + ) is characterized by multiple effects associated with both the free and bound dye, which are present in equilibrium. In general, we observe instantaneous absorbance changes, arising from processes that are faster than the heating time of the instrument (about 2 ps under our experimental conditions) attributed to the free and bound dye, as well as a measurable relaxation spectrum consisting of three relaxation times. None of these measurable effects arises from the free dye alone as only rapid, unresolvable absorbance changes occur when the dye is perturbed in the absence of nucleic acid. The concentration dependenoe of these relaxation times will be a function of the rate constants describing the binding mechanism and the concentrations of reactant species, namely free dye monomers and free DNA binding sites. The relation between the concentration of free binding sites and base-pairs is therefore required to evaluate the kinetic results and to compare the binding constants so obtained with the statically determined apparent equilibrium constant. We will first describe the model that has been used for the nature of the binding site and then apply this to the relaxation data. (a) Neighbor exclusion model Use of the Scatchard analysis (Scatchard, 1949) assumes that binding sites are independent of each other and predicts that:

where C, is the concentration of free DNA binding sites, C, is the concentration of the bound dye, GE is the DNA concentration in base-pairs, B,, is the apparent number of binding sites per base-pair and r = C,/Ci. The association constant for the formation of bound dye from the free state is then given by

or

where C, is the concentration of free dye monomers. Equation (1) predicts a linear dependence of r/C, with r over the entire range of binding. A binding isotherm for the EthBr-calf thymus DNA system analyzed by this approsch is shown in Figure l(a). The value of B,, is found equal to O-360, indicating that, 36% of the base-pairs act as binding sites for EthBr. However, this interpretation cannot be representative of the actual binding process, since binding isotherms with synthetic nucleic acid homopolymers, where all base-pairs are identical, give values of B,, < I (LePecq & Paoletti, 1965; Bresloff & Crothers, manuscript in preparation). There is no physical basis for assigning some fraction of the base-pairs as binding sites, while the others are not. A model that can realistically explain the observed binding isotherm assumes that the region between any two base-pairs acts as a potential intercalation site but on binding to one such region further binding to neighboring sites is excluded, presumably through alteration of the secondary structure of the nucleic acid. This

108

J. L. BRESLOFF

AND

__I--

0 2 0 I-----0.20 00 0.15

D. M. CROTHERS (a; (a;

o ok

0.10 1

\ s

F 035 t

‘0,

0

Extent of binding(r j FIG. 1. Binding isotherm of EthBr to calf thymus DNA. (a) Scatchard analysis. The when r/C, = 0 gives B,,. The slope gives -IX&,. (b) Neighbor exclusion analysis. T C, is the monomer free dye concentration, K(0) = 1.83 x IO4 M-l is the intrinsic binding Experiments were done at 19.0%, to an isolated potential binding site and equals B,,K,,. l-0 M.

intercept = Cb/Cz, constant [Na+] =

neighbor exclusion model with an exclusion of one base-pair on each side of the intercalated site predicts a non-linear dependence of r/C, with r given by the relation (&others, 1968) r/K(O)C, = (1 - 2r)2/( 1 - r),

(2)

where K(0) is the intrinsic binding constant to an isolated potential binding site. The curve shown in Figure l(b) was calculated from the functional dependence of r/E(O) C, on r predicted by equation (2) and is in excellent agreement with the experimental data. A similar observation was also found by Bauer & Vinograd (1970) for EthBr binding to nicked circular simian virus (SV40) DNA. It is now a simple matter to determine the concentration of free DNA binding sites at any value of r. Calling P(r) the fraction of the concentration of potential binding sites at zero r that are available for binding EthBr, at a given r value, then C,(r) = F(r) Cg.

(3)

Since the binding isotherm is non-cooperative, the association may be expressed by a single equilibrium expression K(O) = c,/c,c,.

(4)

DNA-ETHIDIUM

Substituting

REACTION

KINETICS

109

equation (3) into equation (4) gives R(O) = ?$&Byr).

(5)

Comparison of equations (2) and (5) reveals that F(r) = (1 - Sr)Z/( 1 - ?“).

(6)

By using equations (3) and (6) the relation between C,(r) and C’z is obtained. Determination of the kinetic parameters using this value of Cd(r) gives directly the rate constants and equilibrium constants to an isolated potential binding site. Comparison with the equilibrium measurements can be made by noting that in the limit as r approaches zero, neighbor exclusion vanishes, so lim r/G, = K,,B,, = K(O), r-t0

as discussed by Crothers (1968), assuming all the spaces between base-pairs are potential binding sites. (b) Einetics

of the birding

reaction

The resolvable spectral change following a perturbation of the EthBr-sonicated calf thymus DNA system is completely described by three relaxation times at all DNA concentrations investigated. The two longest times (intermediate 72 with amplitude A,, and Iongest 7s with amplitude As) are found to be always closely coupled, differing only by a factor of roughly 3 to 7. The ratio of the amplitudes of these effects, As/A,, varies between about 1 and 3 over the range of DNA concentration used, with As/A, increasing with decreasing DNA concentration. The fastest time, TV, is 6 to 16 times smaller than TV, with an amplitude, A,, between 6 and 14% of the total resolvable absorbance change, A, -/- A, + A,. Plate I(a) and (lo) shows kinetic spectra obtained by analyzing for all three times. As a control, the kinetic spectrum of non-sonicated, high molecular weight calf thymus DNA was investigated and found to be indistinguishable from its sonicated counterpart. A concentration dependence of the two longest relaxation times determined at 465 mu, where A, + A, atstains a maximum, is shown in Figure 2. Both 1/T2 and l/r3 are observed to vary linearly with DNA concentration. The linear dependence of I/T~ continues, within experimental error, up to the highest concentration obtained, 81.0~ 10e4 M in base-pairs (-5 mg/ml). This dependence was found to occur also at 435,495 and 545 nm. This behavior of ~a was rather unexpected, apart from a bimolecular reaction that would give rise to a single relaxation time with this dependence, more complex systems in general predict a levelling off of the longest relaxation time with increasing DNA concentration. In order to account for the increasingly faster reaction rate with addition of DNA, it is necessary to postulate a mechanism that includes bimolecular conversion between the two bound dye species that give rise to T2 and Ts, and which is mediated by the presence of other DNA binding sites. Specifically, a mechanism which would predict the observed comentration dependence of ~a and Tgis : (ED), Lk, k-2

E + D &

(ED), + D -s-m?k-t

(ED),. k-1

(ED), + D.

Fa) G’b)

110

J. L.

BRESLOFF

AND

D. M. CROTHERS

W

5 C,”

x IO4

(M)

FIU. 2. Dependence of the two longest relaxation times, -r2 and Q, on DNA concentration at constant r. (a) l/~s VWSU~Ci; (b) l/q, verszds C,.o The insert extends the concentration dependence of 73 to 3 and 5 mg/ml of DNA. Error bars giving &lo% uncertainty are shown. T = 0.112, tf = 32*O”C.

This mechanism stipulates that free EthBr (E) can combine directly with a DNA binding site (D) in two different ways to form two complexes, (ED), and (ED),. Furthermore, conversion between the complexes can occur not only via the free state (eqn ‘7a) but can also occur through a direct transfer mode (eqn 7b), whereby EthBr bound to a DNA binding site in one way, e.g. (ED),, is able to transfer to another DNA binding site to form the other bound dye complex (ED),. (Direct transfer between identical complex forms might also occur, but will not contribute to the relaxation spectrum.) It should be stressed that the pathway represented by equation (7a) alone (or any such similar mechanism) will not predict the concentration dependence of the two longest relaxation times. However, addition of the pathway (eqn 7b) that allows conversion between (ED), and (ED), to proceed by a bimolecular mechanism will give rise to the linear dependence of l/r2 and 11~~.Equation (7b) describes the direct transfer process, which may be of general importance for transferring proteins between DNA binding sites. An additional relaxation time (TJ is observed, indicating that at least three bound

PLATE I. Relaxation spectra of the EthBr-calf thymus DNA system following a temp erature;I xnp perturbation. (a) Multiple exposures of signal taken at 5 ms/major division and e: > K1(=kl/k-l) or vice versa. Further, by requiring that K, < K(0) and that k, change monotonically with temperature, a unique set of solutions at each temperature can be determined (Bresloff, 1974). Analysis of the concentration dependence of ~a in terms of a bimoledependence cular process gives k, and k _ 2 ; the slope and intercept of the concentration of 72 yield Ic, and k-,. The two other rate constants k, and k-, are subject to the conditions that k, be a good deal larger than Ic, and similarly k, >> k-,. An estimate of K, is obtained by comparison of the kinetic data with equilibrium results, as will be presented later. The kinetic constants k,, k-,, k-, and k,, found by interpreting the concentration dependence of 72 and ~a in terms of the direct transfer mechanism, are given in Table 1. These are found to fit the data within experimental uncertainty, as is shown in Figure 4. Uncertainty limits on the kinetic constants are about 2 20%. Also in Table 1 are k, and k- ,,, the rate constants obtained from the concentration determined binding constants and independence of TV, as well as the kinetically trinsic binding constant determined from equilibrium dialysis. Arrhenius plots of the rate constants to obtain activation energies are given in Figure 5. Table 2 summarizes the thermodynamic and activation parameters of the system. (c) Additional

evidence for the direct transfer mechanism

Although the concentration dependence of the longest relaxation time requires the formulation of a bimolecular conversion pathway for equilibrium over the binding sites, additional evidence for the novel direct transfer mechanism would be certainly welcomed. This was obtained by observing the kinetics of binding of EthBr to an RNA species in the presence of a DNA species. Specifically, calf thymus DNA and poly(rA)*poly(rU) were used; the latter was chosen since, under comparable conditions, relaxation times for the EthBr-poly(rA) *poly(rU) system are roughly a factor of five to ten longer than with calf thymus DNA. Therefore, the longest relaxation time measured for a system containing both nucleic acids together monitors the binding of Et,hBr to the RNA. Qualitatively speaking, if no direct transfer can occur between the DNA and RNA so that intercalation into the RNA can proceed only via the free dye, then the terminal relaxation time measured for the DNA and RNA together will be longer than in the absence of the DNA, since now the DNA acts as a “buffer” to the dye concentration. However, if direct transfer can occur SO that

1.53x 107

1.70 x 107

2.32 x IO”

19

23

32

9.51 x 103

7.40 x 103

6.86 x lo3

6.18 x lo3

1.10 x 106

6.22 x lo5

4.81 x lo6

3.85 x 105

(M.S)-l

K(0) is the intrinsic binding constant to an isolated potential in terms of base-pairs. Solutions were in buffer II (see Materials

1.36 x lo1

15

(M.S)-l

6.36 x lo6

2.85 x 10s

2.05 x 10s

1.38 x 106

(M*S)-l

by equilibrium

230

180

160

140

2.44 x IO3

2.30 x lo3

2.23 x 10s

2.20 x 10s

in I iv-Nat

measurements.

thymes DNA interaction

1

binding site, determined and Methods).

138

54.7

35-l

22.7

Kinetic constants for the EthBr-calf

TABLE

DNA

concentrations

7.97 x 103

1.14 x 104

1.37 x 104

1.70 x 104

are expressed

9.91 x 103

1.51 x 10”

1.83 x lo*

2.24 x lo4

(M-l)

114

J. L.

BRESLOFF

AND

D.

M. CROTHERS

-(b) 60-

I

I

I

-15

0

30

45

C,” x 104(M) FIG. 4. Comparison between the experimental data (0) and the predicted curves (-) based on the proposed binding mechanism (see eqns (7a) and (7b)). The latter curves were calculated using equations (9) and (10) with the kinetic rate constants given in Table 1 and assuming that b, > k1 and k, > k-,. (a) l/~ + l/~s versus DNA concentration; (b) l/ Q. l/~s wersus DNA concentration. r = 0.112, tr = 32.O”C, 0, = 1.66 x 10-s M, c, = (0.673) c:.

intercalation into the RNA can proceed not only via the free dye, but from a DNA binding site to an RNA binding site, then as long as the rate constants for this transfer are comparable to, or larger than, the rate constants for intercalation into the RNA, the longest relaxation time with the DNA and RNA together will be faster than in the absence of the DNA. These arguments are presented in quantitative form in the Appendix. TABLE 2 Energies characterizirzg EthBr-edj’

E,+

E-,+

5.7

4.5

Ez+ 10.8

E-,+ 18.5

E,+ 15.7

thymus DNA system (kcallmol)

E.ml+

AH:

AH;

5.1

1.1

-7.8

(t = 19.OOC) (t = 19.O”C) tAS: tAS; 5.6

-2.3

E + is the activation energy far transformation from one state to another as indicated by the subscript. AH” and AS” are the thermodynamic enthalpy and entropy changes. These parameters were determined over the temperature range 16°C to 32°C in 1.0 na-Na+ (buffer II, see Materials and Methods).

DNA-ETHIDIUM

REACTION

3.2

1

32

33

3-4

KINETICS

3.3

34

I/t x 103(OK)-’

35

I/f x IO3 (OK)-’ FIG. 5. Plots of natural logarithm of the rate constants z)eew.s inverse temperature to obtain activation energies. (a) --O-O-, Ea+; -@--a--, EM,+. (b) -O-O-, Et+; -O--O-,

E-1+. (c)-@--@-,E~+;-.-.-,E-~+.

116

J. L.

BRESLOFF

AND

D. M. CROTHERS

Figure 6 describes the results of an experiment designed to detect binding to poly(rA) *poly(rU) by transfer from DNA. The concentration of poly(rA) . poly(rU) was held constant, while that of the calf thymus DNA was varied between 0.3 and 7.3 times the RNA concentration. Importantly, the extent of binding for each nucleic acid was held constant for all the solutions by dialyzing them together to the same free dye concentration. It is observed that addition of even the lowest concentration of DNA used here results in a dramatic decrease, from 45.0 to 25.4 ms, in the longest relaxation time TVover that of the RNA by itself. With increasingly higher concentrations of DNA, 71 becomes faster, and approaches the value to be expected if only the DNA were present. However, in order to analyze this experiment correctly, the amplitude dependence of T, should be determined (insert in Fig. 6). The rapid initial decrease in A, followed by its slow increase to a limiting value arises from the buffering action of the DNA: the amplitude of the process monitoring the intercalation reaction into the RNA decreases quickly to zero, so that at higher DNA concentrations the longest relaxation time that can be measured for a DNA-RNA solution reflects binding to the DNA. (The amplitude dependence observed under this condition closely resembles that found when only DNA is present.) Therefore, it is in the region of lowest DNA concentration where comparison between 71 for the DNA-RNA system and RNA alone should be made, since it is in this region where 71 for the DNA-RNA complex monitors intercalation into the RNA. The observed initial dramatic decrease in 71 with addition of calf thymus DNA confirms that transfer

40-

303 E

I -O

1

,

20

0

v

C: x 1041W

c

I

I

40 calf fhymus DNA

20

\ IO

,

0 ‘1,

I

\\ \\

0

\

‘.O t

I

RNA

I ----_ 10

---L---------I

20 C,” x 104(M)

I -------I

30 calf

40 thymus

-0

50

3

DNA

FIG. 6. Dependence of the longest relaxation time, 71 and its amplitude A I (insert) as a function of DNA concentration, for a series of solutions at constant RNA concentration. All soIutions were dialyzed to the mme free dye concentration. Calf thymus DNA and poly(rA) .poly(rU) (C’i= 7.63 x lo-* M) were used. C, = 2.8’7 x 10-s m, t, = 19.O”C.

DNA-ETHIDIUM

between ethidium

the DNA to RNA.

and RNA

REACTION

is occurring.

(d) Wavelength dependence

117

KINETICS

In effect, DNA

of the amplitude

catalyzes

changes

the binding

of

A, and A,

Valuable information on the spectral characteristics of the bound dye species (ED), and (ED), can be obtained from a wavelength dependence of the amplitudes A, and AE, are of the two longer relaxation times, 72 and ~a. These amplitudes, directly proportional to the absorbance changes associated with their corresponding relaxation times, so that the wavelength dependence reflects the difference in extinction coefficients of the free and bound dye species. The wavelength dependence of both A, and A, is found to be quite similar and follows almost exactly the form predicted by the dependence of de between free dye and total bound dye, obtained by equilibrium measurements. Figure 7 shows the absorption spectrum of free EthBr and total bound EthBr, the latter obtained directly by equilibrium dialysis with analysis by difference absorption spectroscopy. From this, de can be obtained at any wavelength. Kinetic experiments with this r = 0.12 complex as a function of wavelength enable determination of A, and A,

450 Wavelength

500

600

(nml

FIG. 7. Extinction coefficient spectrum of free EthBr (A) and bound EthBr (B). The bound dye absorption spectrum was determined directly by equilibrium dialysis of the EthBr-calf thymus DNA system: r = 0.122, CE = 29.2 x 10m4 DI. Extinction coefficients were determined a,s described in Materials and Methods, at 19.O”C in the presence of 1.0 M-NBC.

118

J. L. BRESLOFF

AND

D. M. CROTHERS

at different wavelengths, and hence a plot can be made of A, and A, versus de, as is shown in Figure 8. It is seen that both A, and A, are essentially directly proportional to Aa, so that the wavelength dependence of the magnitude of the absorbance change as well as its reversal in direction follows the dependence of AG with wavelength.

(e) “Irwtanta~eous” boutid dye effects In addition to the resolved relaxations TV, 72 and 7s) there is a thermally induced absorbance change too fast to measure with our apparatus (r < 2 ps). Figure 9 shows the wavelength dependence of the amplitude of the rapid spectral effect for a series of solutions of varying DNA concentration at a constant r value, (The oontribution from the free dye has been subtracted, its effect is quite small over most of the wavelengths studied for DNA concentrations of about 10 x IO-& M (base-pair) or higher at the low C, value used.) Results reported elsewhere (Bresloff, 1974) show that the amplitude of this effect is directly proportional to the concentration of bound dye, indicating that the effect should be assigned to a temperature-dependent extinction coefficient of the bound dye. We found at 545 nm, de/At = -5.5 M-l cm-l deg. C-l (6 decreases with increasing temperature), and at 585 nm, de/At = l-8 m-l cm-l deg. C-l.

4. Discussion (a) Nature of the reaction meclm&wn We interpret our results to show that ethidium binds to DNA in (at least) three different ways, and that the different complex forms can be interconverted by the bimolecular direct transfer pathway. To a crude approximation (neglecting kinetic coupling between the relaxations) the fastest relaxation 71 represents equilibration of the free dye over the outside-bound sites, the intermediate relaxation ra reflects equilibrium of the bound dye form 1 with form 2 via the direct transfer process, and the slowest relaxation ~a corresponds to equilibration of the free and outside-bound dye with the bound dye forms 1 and 2. Some sort of direct transfer involving outsidebound dye must occur, but we are unable to specify the mechanism in sufficient detail to determine the corresponding kinetic constants. (b) Alternatives to the proposedmec~nism Several potential objections to our proposed mechanism (eqns (7a) and (7b)) should be mentioned. It is possible that multiple relaxation times could arise from variations in the kinetic constants for different base compositions or sequences. We rejected this idea because we found virtually no variation in the relaxation spectrum for DNAs of differing base composition, and because the same multiple relaxations appeared for ethidium interacting with double helical synthetic polynucleotides (J. L. Bresloff and D. M. Crothers, paper in preparation). Furthermore, no variation with molecular weight was seen in the relaxation spectrum, so the additional forms cannot be special binding sites at helix ends. Hence we are forced to conclude that ethidium can bind in several different ways to a given region of double helix. Another alternative to equations (7a) and (7b) would be a sequential mechanism in which (ED), is an intermediate between free dye and (ED),, as was proposed for the intercalation of proflavine via the outside-bound dye form (Li & Crothers, 1969). When direct transfer occurs, the sequential and parallel reaction mechanisms are

DNA-ETHIDIUM

I

REACTION

I

119

I

I

-1500

KINETICS

0

rd, kAn -I- k-a l/T1 = i& 6, + csp + k - 6 + hifl, kicn [ (

’ )I+k-r

(A5)

When c, > 0, the quantity in parentheses in equation (A5) is always less than one. Comparison of equations (A4) and (A5) reveals that l/~~ < l/rr or T[ > T$. However, if direct transfer between the DNA and RNA can occur, then the pathway k

ED+RG

ER+D

W)

k;t

must be considered with equations (Al) and (A2). The longest relaxation time, rl, is then given by the expression:

The rate constants appearing in equation (A7) are the same as those in equa*tion (A5). This assumption is valid as long as the DNA and RNA do not interact in solution, which is not expected at such low concentrations. As long as the expression in the brackets of equation (A7) is positive, then l/~~ > l/Tr and ~~< Tr. This will be met if, for example k,, the rate constant for transfer from a DNA to an RNA binding site, is equal to or greater than k,, the rate constant for intercalation into the RNA from the free state, since then the last two terms within the brackets become fl,C,k,(lc, - k,) 2 0. Thus, if the transfer of EthBr between a DNA and RNA binding site can occur efficiently, addition of the DNA will decrease the longest relaxation time from that for the EthBr-RNA system alone. To summarize, if it is observed that addition of DNA to RNA, under the appropriate conditions, results in a decrease in the longest relaxation time when compared to the RNA alone, then this result requires a mechanism in which binding to RNA is assisted or catalyzed by DNA. Equation (A7) applies as well to transfer of a ligand between different kinds of DNA binding sites, such as productive and non-productive complexes d and r. This work was supported initially by grant no. GM12589, and subsequently grant no. CA15583, both from the National Institutes of Health. One of us (J. L. B.) was therecipient of a U.S. Public Health predoctoral fellowship (GM40024) from the National Institutes of Health. REFERENCES Bauer, W. & Vinograd, J. (1968). J. Mol. Biol. 33, 141-171. Bauer, W. & Vinograd, J. (1970). J. Mol. Biol. 47, 419-435. Bresloff, J. L. (1974). Thesis, Yale University, New Haven, Corm. Crawford, L. V. & Waring, M. J. (1967). J. Mol. Biol. 25, 23-30. Crothers, D. M. (1968). Biopolymere, 6, 575-584. Crothers, D. M. (1971). In Procedures in Nzccleic Acid Besearch (Cantoni, D. R., eds), vol. 2, pp. 369-388, Harper & Row, New York.

G. L. and Davies,

DNA-ETHIDIUM

REACTION

KINETICS

fZ4

Eigen, M. & DeMaeyer, L. (1963). In !Z’ecechn@ue of OrganicCliemti~ry (Freiss, S. L., Lewis, E. S. and Weissberger, A., eds), vol. 8(2), pp. 896-1054, Interscience, New York. Freifelder, D. (1971). J. Mol. Biol. 60, 401-403. Fuller, W. & Waring, M. J. (1964). Ber. Bunsengeselkchaft, 68, 805-808. Gilbert, W. & Maxam, A. (1973). Proo. Nat. Acad. Sci., U.S.A. 70, 3581-3584. Hinkle, D. C. & Chamberlin, M. J. (1972a). J. Mol. BioZ. 70, 157-186. Hinkle, D. C. & Chamberlin, M. J. (19725). J. Mol. BioZ. 70, 187-195. Fluorescence Spectroscopy (Chen, R. and Jovin, T. M. (1975). In TTencZs in Biochemical Edelhoch, H., eds), Marcel Dekker, in the press. Krakauer, H. (1969). Thesis, Yale University, New Haven, Conn. Kreishman, G. P., Chan, S. I. & Bauer, W. (1971). J. Mol. BioZ. 61, 45-58. LePecq, J. B. & Paoletti, C. (1965). Comp. Rend. Acad. Sci. (ser. C), 260, 7033-7036. LePecq, J. B. & Paoletti, C. (1967). J. Mol. BioZ. 27, 87-106. Li, H. J. & Crothers, D. M. (1969). J. Mol. BioZ. 39, 461-477. Lin, S. Y. & Riggs, A. D. (1972). J. Mol. BioZ. 72, 671-690. Madge, D., Elson, E. L. & Webb, W. W. (1974). Biopolymers, 13, 29-61. Miiller, W. & Crothers, D. M. (1968). J. Mol. BioZ. 35, 251-290. Pohl, F. M. & Jovin, T. M. (1972). J. NloZ. BioZ. 67, 375-396. Riggs, A. D., Bourgeois, S. & Cohn, M. (1970). J. Mol. BioZ. ,53, 401-417. Riggs, A. D., Lin, S. & Wells, R. D. (1972). Proc. Nat. Acud. Sci., U.S.A. 69, 761-764. Saucier, J. M., Fe&y, B. & LePecq, J. B. (1971). Biochimie, 53, 973-980. Scatchard, G. (1949). Ann. N.Y. Acad. Sci. 51, 660-672. Sohmechel, D. E. V. & Crothers, D. M. (1971). Biopolymers, IO, 465480. Sobell, H. M. & Jam, S. C. (1972). J. Mol. BioZ. 68, 21-34. Steitz, T. A., Richmond, T. J., Wise, D. & Engelman, D. (1974). Proc. Nat. Acad. SC&, U.S.A.

71, 593-597.

Thomas, G. & Roques, B. (1972). FEBS Lettere, 26, 169-175. Waring, M. J. (1965). J. Mol. BioZ. 13, 269-282.