Vouume 29, Nusoen 11 PHYSICAL REVIEW LETTERS 11 Seprenen 1972 *Work partially supported by the U. S, Atomic Energy ‘Commission under Contract No. AT(1-1)~3237. 'N, M. Kroll, A. Ron, and N. Rostoker, Phys, Rev. Lett: 18, 88 (1964) Pv. Teytovich, Non-Linear Effects én Plasmas (Plex tum, New York, 1970), Chap. 2. %p, Wolff, In Proceedings of the Second International Conference on Light Scattering in Solids (Flammarion, Paris, 197%), p. 180. 43. M, Daweon, Phys. Fluids 5, 445 (1962). 'B.1. Cohen, A. N. Kaufmann, and K. M. Watson, Phys. Rev. Lett. 29, 581 (1972). Thermodynamic Fluctuat ns in a Reacting System—Measurement by Fluorescence Correlation Spectroscopy Douglas Magde,* Elliot Elson, and W. W. Webbt Comell University, Maca, New York 14850 ‘Received 10 July 1972) ‘The temporal correlations of thermodynamic concentration fluctuations have been mea- sured in a chemically reactive system at equilibrium by observing fluctuations of the flu ‘orescence of a reaction product. The experiment yields the chemical rate constants and iffuston coefficients and shows the coupling among them, Data are reported for binding of ethidium bromide to DNA, ‘The time correlations of thermodynamte con centration fluctuations in reactive multicompo- nent systems at equilibrium are determined by the kinetics of chemical reactions and diffusion processes. Purely diffusive fluctuations have been measured with great success by quasielastic light scattering’ and extension to reaction kinetics has prompted several experiments? and attracted considerable theoretical attention.’ However, it now appears that homogeneous chemical kinetics are not amenable to scattering studies because the dielectrie-constant changes that reveal the fluctuations are usually too small. In contrast, optical absorbance coefficients and fluorescent quantum yields frequently display large changes. Hence we chose to observe intrinsic concentra- ton fluctuations using 2 fluorescent indicator. We report here direct observations of fluctua tions about thermodynamic equilibrium in a re~ active multicomponent system of biophysical in- terest. We have studied the reversible binding to DNA of ethidium bromide (BtBr), a dye that in- hibits nucleic acid synthesis.‘ The complex of dye and DNA is strongly fluorescent; thus flucta- ations in the number of molecules of the complex. ina small volume defined by a beam of exciting radiation are indicated by fluctuations of the total fluorescent power.* ‘The principal chemical reaction between the DNA (A) and the EtBr (8) to form the fluorescent complex (C) is supposed to be a single-step bi- molecular process with rate constants k, and hy (in fact the system is more complex, as will be seen): Avec, CO) The equilibrium constant is defined in terms of the equilibrium concentrations C4, Cy, and To 85 K=h,/ky=C/C,Cp. Thermodynamic concen tration fluctuations decay via the chemical reac- tion as exp(-R¢), where the inverse relaxation time R is? R=b(CyATy) thy ® Since diffusion provides alternative relaxation paths, the correct description of the fluctuation spectrum requires a set of coupled differential equations which for ideal solutions assume the form 2.8C,/ot= D,V*6C,+DaT py 8C yy ®) where 6C,=6C,(, ) are the local concentration fluctuations of the three reactants of Eq, (1); the D, are corresponding diffusion coefficients, ‘and the 7, are the elements of the matrix of linear chemical interaction coefficients implied by Bq, (1). In our experiment, fluctuations of the concen- tration C of the fluorescent complex cause the ‘Muctuations in a photodetector current, The autocorrelation function of the photocurrent flue~ tuation bi()= i!) ~ (i), with I7| written 7, is G\(1)= (Gill 7) (0). Neglecting shot noise for simplicity, the photocurrent due to fluorescence Induced by the exciting radiation 1(@) is i= 2eQfIMCE, Nar, ® 705Vouuste 29, Nunn 11 PHYSICAL REVIEW LETTERS 11 Serrensen 1972 where sf accounts for quantum efficiency, photo- multiplier gain, and geometrical and filtering losses. @ and ¢ are the fluorescence quantum efficiency and extinction coefficient for the EtBr- DNA complex. In contrast with quasielastic scattering experi: lr ‘ments, there is no spatial coherence; the coher- ence time of the broad emission spectrum is ~10-" sec and a random fluorescent emission delay of ~10°* see uncouples the fluorescence from the coherent exciting radiation, Thus fluo rescent emission from different molecules is uncorrelated so that intensities are added, Thus (ge Q)* [IFMF VOCE, NOC LE, 147) dv dv, (3) G,(7) is evaluated in terms of solutions of Ea. (3) and statistical weights determined by the equiparti- tion theorem," We will report details elsewhere, ‘The convolution of the spatial concentration correlation function with the Gaussian profile of the cylindrical laser beam leads to characteristic diffusion times of the form 1,=w7/4D,, where j=A, B; wv ts the beam radius at 1(0)/Iya, =e ". Hence the solutions of Eq. (3) may be reduced to convenient form using the expansion parameter (D, ~_D,)Ru* <1. Noting that D,~De «Dy and specializing to an = LEREI ago (da.40h where P is the total power in the incident light beam, volume, and Aglt)= KE g( 47/7)" ay 14, WTeKC, Tear, 5gR TFKC, Tor/e 4,0) 1 ie exe, tore (t with characteristic times 1,=74(1+KC,), 7. =tg(L+KC,)/KC,. ‘The time dependence of A,(r) is determined en- tirely by the slower diffusing species through 14 A, reflects diffusion of the small molecule 25 slowed by chemical interaction with the slower diffusers; A. is dominated by the exponential factor due to the chemical relaxation of Eq, (2). ‘The convolution with /(*) in Eq. (4) introduces factors of the form (1*7/r,)" replacing the more ‘usual exponential form. ‘The mean photocurrent is (i(/))=geQlPC, and the relative root-mean-square fluctuating photo- current is At) Bigs GON «6 are ye, GS aqgy HE wy In our experiments ;w*t~ 10"* em? and T,~10°7 -M or ~10* molecules /em® s0 that 54,.,/4él) 10", ‘To measure G(r) we excited the EtBr-DNA fluorescent complex with the 514.5-nm line of a stabilized argon laser, usually at 1.5 mW power to minimize photodecomposition with the beam focused to w=5,5 jm in a cell of thickness 1= 150 hm, ‘The orange fluorescence was collected by 106 ) ~CoC,-K*, we obtain the form of G,(2) useful for our experiments: 6) 1 is the length of the illuminated (cylindrical) a parabolic reflector, passed through a saturated K,Cr,0, solution filter to reject the laser Light, and collected on an $-20 photocathode, ‘The photo current fluctuations were analyzed, after filter- {ng out the steady component, with two 100-chan~ nel SAICOR autocorrelators. From 10° to 10° intensity samples were recorded for each time delay, The response of the entire system was proven with a stable white-light source that could be weakly modulated at ~0.01% rms. ‘The optical properties of the DNA-EtBr complex are ¢=3.8 (aM em)" at 514.5 nm," Q= 14%." In the free dye, @ is only ~0.%." Our calf thymus DNA was phenol extracted and sonicated to a molecular ‘weight of 2x10", ‘The experiments reported here were all made at 22°C in a suitable buffer (10°* ‘M Na-ethylene-diamine-tetra-acetic acid, 10* M tris-HCl, pH 8, 10°' AL NaCl. Measurements of G(s) for two “pure” dyes, rhodamine 6G and EtBr, demonstrated the method for nonreactive species. Accord with each factor of Eq, (7) was verified. ‘The observed 6,(7) «(1+7/%9)"!, where T)=10"/4D, is determined by the diffusion coefficients D,, of the pure dyes. We obtained D, &1.5%10"* em"/sec for both dyes.Vouuste 29, Numer 11 PHYSICAL REVIEW LETTERS 11 Seeresenee 1972 © Tq 6x 10° Molar p= 2.2 x10 Molor B= 7.9 x 15® Molar 4 7 3 Lo = i Bs 2 x10 Motor bac Bq 49x10 Noor he & + 55 x10 Motr gh 2s (seconds) FIG, 1, Typical data points for G,(r) at two values of jp fitted by Ea. (6). Although 7) =86+7 msec, the uncertainties of w? and thus Dp may be ~50%. These values of Dy are reasonable." ‘To derive the kinetic parameters, G,(7) was measured as a function of C4 from 2x10"7 to 8 x10"" IM. Over this range the relative amplitudes of A,(0) and the 4 (0) change considerably. At the higher @,, the term A (x) dominates, as {illustrated by the fit to measured points in Fig. (a), so that a set of values of 7, may easily be determined as a function of C, a5 shown in Fig. 2, yielding a value for k,/ky. At the low C,, the major term 4.(r) is domin- ated by the exponential factor due to reaction, Kinetics. This range yields an estimate of f, the limiting value of R as C,+C,=0. Given trial estimates from the limiting cases, intermediate concentration data as in Fig. 1(b) can be analyzed reiteratively to improve the estimates of K and R, The derived dependence of , and R on Cy is, exhibited in Fig. 2, The characteristic time 7. is not reliably obtained as an independent experi- ‘mental result, The dependence of 7, and 7. on 1w* in Eq, (6) was confirmed in the limiting cases. Equation (1) oversimplifies the mechanism of inding. Surface sites for EtBr in DNA* and puzzling kinetic complexities" have been ob- served. We attribute to these complications the sox 3 40 i g : po = ao ox,ten won | 8 3 6 R Left scale 3 : 3 8 7 R Right scale 20 © 2 12 dio ® ° LS Ta concentration (10° Mover) _FIG. 2, Magnitude and variations of 7, and R™! with Tq, which yleld the chemical parameters of the reac~ ton; see text. observation that the measured value of [6,(0)}# has approximately $ of the calculated magnitude while (j(0)) remains rather consistent with the pure-dye experiments and calculations, The re duced amplitude of G (0) can be interpreted as a shift of some of the fluctuation spectrum out of the range 50 usec <7 <2 sec. With Eq. (6) and the typical data set shown in Fig. 2, the dependence of R on C, yields &,=20 #7 soc" and k,= (1,840.8) x10" see"™ M“, whence K=hy/ky gives 4x10 =K=20%10* M+, The de- pendence of 7, on C4 independently determines K with greater precision as K=(5.41)%10° a", By conventional fluorescent titration in the same apparatus we obtained K=(6+1)%10° MI, Using K=5,4x10" IM“ and the most precise datum on Rat the largest C,, the best values of the kinetic parameters are obtained as #,=27 sec“ and k, 1,5x10" sec" M7, Independent values of 402 10 sec" and &,= 210" sec"? a” have been determined by conventional temperature perturba- tion methods at 92°C." A reasonable correction for tho small temperature dependence leads to agreement well within the combined uncertainties. We conclude that our method of fluorescence correlation spectroscopy has successfully re - ‘vealed the ubiquitous fluctuations around equilib- rium in a reactive system with sufficient preci- sion to identify and measure chemical reaction Kinetics as well as diffusion and to demonstrate 707