also note that the surface area so estimated is closer to that from high-temperature gas-solid interaction measurements’ and thus helps close the gap between these two methods of measurement. Acknowledgment. We gratefully acknowledge support of this project through a grant, No. DMR 8111111, from
the Division of Materials Research, Metallurgy Polymers and Ceramic section of the National Science Foundation. We are extremely indebted to Professor Pierotti for allowing us to use his data prior to publication. We also thank Professor S. Fain for pointing out several errors. Registry No. K r , 7439-90-9.
Picosecond Fluorescence Anisotropy Decay in the Ethldlum/DNA Complex Douglas Magde,” Marlna Zappala, Wayne H. Knox, and Thomas M. Nordlund Department of Physics and Astronomy, Institute of Optics, and Department of Radietion Biology and Blophysics, University of Rochester, River Station, Rochester, New York 14627 (Received: November 15, 1982)
Ethidium bromide intercalated into calf thymus or salmon sperm DNA exhibits fluorescence anisotropy relaxation due to torsional motion (and, at long times, bending) of the DNA which is well described by the elastic continuum model for times greater than 0.5 ns. A fast component of amplitude 0.025 f 0.01 and characteristic time 100 (-50, +loo) ps is present at short times and is affected very weakly or not at all by solution viscosity. Such fast relaxation may be attributed to “wobbling”of the dye within a binding site, although it is not completely certain that all other possibilities have been rigorously excluded.
Introduction Deoxyribonucleic acid (DNA) is a double-stranded linear polymer which codes genetic information in higher organisms. Both replication and transcritpion of DNA imply subtle control of motion of the polymer. Ethidium bromide (EB) is a fluorescent dye probe which intercalates between adjacent base pairs of the DNA helix. The time dependence of fluorescence depolarization of EB can be monitored in order to determine the local torsional flexibility of DNA. We report here the extension of such anisotropy decay measurements down to the true picosecond regime. Our time resolution is improved by a factor of ten or more over earlier studies.’-5 Wahl and his group were the first to observe this behavior,’ although their data had only limited time resolution and their original interpretive model has been superceded. Barkley and Zimm (BZ) proposed a different theoretical treatment.6 Schurr and co-workers introduced a CW mode-locked laser with single-photon correlation to improve the measurement2 and proposed their own theoretical analysis,’ which was independent of but largely equivalent to that of BZ. At almost the same time, Zewail and his group3 also studied anisotropy decay over times from below 1 to about 100 ns, using the BZ formulation to analyze their results. They have since refined their analyses and extended somewhat their measurement^.*#^ Their last paper5 (MRZ) will constitute the takeoff point for our analyses, since it describes very succinctly the issues left unresolved by the nanosecond studies. In particular, all workers agree that an extrapolation of their nanosecond data back to an initial anisotropy implies small but significant “missing amplitude”. Instead of the theoretically expected value of 0.4, MRZ can account for only about 0.36; other workers found somewhat larger discrepancies, which could, so far as one can tell from published reports, be partly instrumental. More recently, Schurr and co-
* Author to whom correspondence should be addressed at the Department of Chemistry, University of California, L a Jolla, CA 92093
worker^*^^ have extended the theory further. In particular, they consider large angles of reorientation, rather than the small angles assumed by BZ. This matters for long time behavior, including even that of MRZ. However, it is irrelevant for the issue treated here. Experimental Section Single-photon correlation offers excellent linearity and sensitivity with good, but not picosecond, time resolution. To reach shorter times, we have combined a very stable, mode-locked and Q-switched, Nd3+:YAG laser with a highly sensitive jitter-free streak camera to form a system capable of extensive signal averaging. This instrumentation has been thoroughly described previously10and has been utilized in applications which challenge its time resolution and its sensitivity for weak emission. The present application tests its ability to distinguish a very small, possibly very fast component of a multiple decay process. This is a difficult task and only the outstanding performance of the jitter-free streak system tempted us to undertake the effort. In order to facilitate comparison with earlier work, many of our measurements were made on calf-thymus DNA, chopped to molecular weight 100000. In order to have some evidence for the generality of the results, measurements were also carried out on salmon sperm DNA. Both (1)D.Genest and Ph. Wahl, Biochim. Biophys. Acta, 521,502(1978). (2)J. C. Thomas, S. A. Allison, C. J. Appellof, and J. M. Schurr, Biophys. Chem., 12,177 (1980). (3)D. P. Millar, R. J. Robbins, and A. H. Zewail, h o c . Natl. Acad. Sci. U.S.A., 77,5593 (1980). (4)D.P. Millar, R. J. Robbins, and A. H. Zewail, J. Chem. Phys., 74, 4200 (1981). (5)D.P.Millar, R. J. Robbins, and A. H. Zewail, J. Chem. Phys., 76, 2080 (1981). ( 6 ) M. D. Barkley and B. H. Zimm, J. Chem. Phys., 70,2991 (1979). (7)S. A. Allison and J. M. Schurr, Chem. Phys., 41,35 (1979). (8)S.A. Allison, J. H. Shibata, J. Wilcoxon, and J. M. Schurr, Biopolymers, 21,729 (1982). (9)J. M. Schurr, Chem. Phys., 65, 417 (1982). (10)W. Knox and G. Mourou, Opt. Commun., 37, 203 (1981).
@ 1983 American Chemical Society
The Joufnal of Physical Chemlsfty, Vol. 87, No. 17, 1983 3287
Picosecond Fluorescence in the Ethldlum/DNA Complex 0.4 r
o.6
0.3 0.2
t
; I =:
f Y
0.I
0
100
200
300
400
r
/
0*41 t / 0.2
I
0
500
species of DNA and Ethidium bromide were obtained from Sigma Chemical Co. With the exception of experiments on EB-glycerol solutions, all measurements on both free ethidium and DNA-EB complexes were made with a buffer of 0.1 M Tris-HC1 (pH 7.7), 0.15 M NaCl, again chosen to mimic previous studies. For some measuremenb sucrose was added to adjust the solution viscosity. Concentrations were determined gravimetrically for sucrose and photometrically for other compounds, assuming decadic molar absorptivities: 4260) = 6412 cm-' M-' of phosphate for calf thymus and 6300 cm-' M-' for salmon sperm DNA" and ~(480) = 5860 cm-' M-' for free dye.12 All measurements were made at ambient temperature, T = 23 f 2 "C. Luminescence was excited with about 20 WJof light at 530 nm in a 30-ps pulse. Either 50 or 100 laser shots were averaged for each experimental run. The instrument response was measured by observing scattering from synthetic coffee creamer. After carefully verifying that the response of the detection system was identical for both polarizations, data were collected by using vertically polarized excitation, defined by a crystal polarizer, and detection at right angles. Vertical and horizontal emission polarizations were selected with Polaroid sheet polarizers. Wavelength discrimination was accomplished with a KV550 and an OG570 Schott glass filters. Time calibration used multiple reflections from a coated etalon having 125-ps delay between successive reflections. A key advantage of the high sensitivity of this sytem with extensive averaging is the ability to use a very small solid angle of detection, namely, 0.001 sr. Use of too large a solid angle is a contributing factor in frequent failures to observe "limiting anisotropy" of 0.4.
Results For our geometry, the anisotropy r is calculated in terms of the parallel and perpendicular intensities I,,,L(t)as (1) r ( t ) = (111- IJ/(I,l + 21,) Figure 1 shows typical data together with an instrument response curve. The response illustrated is not the shortest possible. It is broadened by optical path differences of up to 4 mm for different portions of the sample. Because of the desire to keep the DNA (phosphate) to dye ratio (P/D) high, without increasing P too far, and to collect as much emission as possible without saturating the absorption, which distorts r,13we illuminated a horizontal sheet of
I
I
4
5
Figure 2. Dependence of the correlation time for rotational diffusion T In free EB as a function of macroscopic solution vlscoslty. Theoretical fit to a straight line agrees with the Stokes-Elnstein relation.
O.I
rr I \ 100
0
200
300
400
500
TIME (pr) Figure 3. Time dependence of anisotropy decay r ( f )for EB in solution with DNA present, all other conditions being the same as In Figure 1. One typical experimental run is shown. Theoretical fit Is the convolution for the curve known to hold at nanosecond tlmes, as detailed In the text, with the Instrument functlon shown at reduced amplitude. Note the fast lnitlal decay not accounted for by that theoretical curve alone.
solution using a cylindrical lens. The instrument response is highly reproducible, fully justifying deconvolution procedure~.'~ Free EB anisotropy relaxes by rotational diffusion. Although multiexponential in principle, the decay is adequately fit by a single exponential convolved with the instrument response, as illustrated in Figure 1. The rotational correlation time was measured in plain buffer as well as in 20% and 35% sucrose solutions. The correlation times and viscosities were 120 (1cP), 230 (1.9 cP), and 525 ps (4.3 cP). The three data points are well fit by a linear dependence on viscosity as shown in Figure 2. If one assumes the Stokes-Einstein relation 7 = 1/6D = TV/kT (2) one may deduce for an assumed spherical volume an "effective radius" of 0.49 nm. In glycerol, we find that EB shows constant r for all t , in accord with previous work. MRZ find r = 0.39 f 0.01 in glycerol. Our data are consistent with this, but the single-photon detection is capable of higher precision; hence we take their value as a standard in calibrating our own data. The "missing" 0.01, if real, is presumably an effect of vibronic relaxation, fast even on our time scale, so that the absorption and emission dipoles are not exactly parallel. Figure 3 shows a typical r(t) curve for EB bound to calf thymus DNA in 20% sucrose/buffer. The DNA concentration was 2 mM in phosphate and the EB concentration ~~
(11)G.Felsenfeld and S. Z. Hirschman, J.Mol. Biol., 13,407 (1965). (12)C.G.Reinhardt and T. R. Krugh, Biochemistry, 17,4845(1978).
I
3
VISCOSITY (cP)
TIME (pa) Figure 1. The time dependence of anisotropy decay r ( f )for EB In solution without DNA polymer. One typical experimentalrun Is shown. Theoretical fit Is the convolution of a single exponentlal with the instrument functlon shown at reduced amplitude.
I
2
~~~
(13)D.Magde, J. Chem. Phys., 68, 3717 (1978). (14)Ph.Wahl, Biophys. Chem., 10,91 (1979).
3288
The Journal of Physical Chemistry, Vol. 87, No. 17, 1983
was 25 pM, giving a P / D ratio of 80. In separate absorption experiments, we confirmed that the equilibrium constant for binding is 2.5 x lo5 L mol-’ for calf thymus DNA in accord with previous so that essentially all dye is bound. For salmon sperm DNA, the equilibrium constant is 8 X lo4 L mol-’. The theoretical curve, appropriate only for short times, drawn in Figure 3 is that obtained by using the MRZ parameterization of BZ: r ( t ) = 0.39(0.105+ 0.585 exp[-0.088 - At’/2] + 0.310 exp[(-0.088 - At’/2)/4]) (3) The t l l z dependence is characteristic of torsional motion of the elastic continuum, which is the intermediate zone of Schurr’s ana lyse^.^^^^ The numerical coefficients depend on the orientation of the EB with respect to the polymer and are obtained from other data. The value 0.088 in the exponentials accounts for the “missing amplitude“ which reduces r(0) below 0.39. The parameter A = 5.1 X 10-37-’/2/ps1/2CP-I/~ and the value 0.39 are fits to their data. We observe that our data reveal the actual time course of the fast component in r ( t ) ,with about the expected amplitude. We repeated the measurement at the same P / D for solutions with 20,35,45, and 60% sucrose, corresponding to viscosities of 1.9, 4.3, 9.4, and 50 cP. In our fits, the theoretical parameters A was scaled as the square root of viscosity in accord with theory. Unfortunately this dependence has not been tested, but the variation is not important in isolating the fast decay. We repeated the same viscosity series with salmon sperm DNA, and we also explored the effect of changing P / D from 250 to 25 for each DNA type. All resulting r ( t ) look much the same. This is in contrast to the result of MRZ who found a several nanosecond “fast decay” for 60% sucrose DNA/EB solution but did not examine any intermediate case. We cannot extend our time range to see if there is yet another component on that time scale, but there would not seem to be amplitude available for it. Considering all our data, we assign an amplitude of 0.025 f 0.01 and a characteristic relaxation time of 100 (-50, +loo) ps to the fast component. We do not have the precision to say whether or not the fast decay follows a single exponential.
Discussion There have been a number of suggestions made for a fast initial decay to account for the “missing amplitude”. In their first paper Millar et al.3 postulated a decay less than 10 ps without invoking a particular mechanism. We do not see anything that fast. Thomas et al.2considered the possibility of weak joints in DNA so that a short segment might twist independent of its neighbors and contribute a fast relaxation mode with an exponential relaxation. Their experiments set an upper limit of 20 base pairs for the possible length of such a segment based upon an observation that the exponential zone, if present, could not extend to times longer than 4 (15) J. B. LePecq and C. Paoletti, J. Mol. Biol., 27, 87 (1967).
Magde et al.
ns. The present results reduce this time by a factor of 20 or more. Because the transition from the exponential to the intermediate zones scales with the square of the rod length, the possible length of a single rod or segment is reduced to below 3 base pairs for the plain buffer solution. More importantly, both the correlation time characterizing the initial exponential zone and the extent of depolarization occurring in that zone should be viscosity dependent, which we do not observe. We conclude that some depolarization due to this mechanism might occur, but it does not appear to be sufficient to account for all our observations without an additional relaxation mechanism. We rule out depolarization due to features of the electronic spectrum of EB (except for the possible difference between 0.40 and 0.39 f 0.01) on two grounds: first, such polarization should be faster than the approximately 100 ps we observe and second, if present, it should apply also to glycerol-EB, which does have the same absorption spectrum as DNA-EB. Intermolecular energy transfer is ruled out on the basis of independence of r ( t ) on P/D, unless there might be cooperativity in binding favoring binding of a second dye at fortuitously the correct distance from the first. Even then the anisotropy decay curves should still change with P/D, but we would not have the precision and sensitivity to resolve small changes over a large range of P / D ratios. Energy transfer in EB-DNA has been treated by Genest et al.16J7 MRZ reject the possibility that relaxation of a weakly or outside-bound, nonintercalated dye (which they believe is present) is responsible for the bulk of the fast component. We concur with their argument. They consider it possible that there might be rapid motion of the EB-base pair group. This sounds the same as the suggestion of Thomas et a1.2 discussed above. Finally there is the possibility of “wobbling” of EB “within” a binding site. This mechanism is discussed explicitly by BZ and also by MRZ. If the wobble is confined to torsional motion, as has been assumed implicitly, then an rms anglular spread of about 18’ is needed to account for the loss of anisotropy from 0.39 to 0.365. If angular displacement is also possible in polar angles, then one may estimate’* that an rms angular spread less than 10’ suffices. Whether that is plausible is unclear, but our results at present are best ascribed to such motion, characterized by a correlation time of about 100 ps, unaffected by macroscopic solution viscosity, in ”typical” if not necessarily all DNA.
Acknowledgment. We thank Mr. B. Wittmershaus, Dr. L. Rosenberg, Dr. G. Mourou, and Prof. R. Knox for their essential contributions. The work was supported in part by National Science Foundation Grants PCM-80-11819 and PCM-80-18488and by the sponsors of the Laboratory for Laser Energetics. (16) D. Genest, Ph. Wahl, and J. C. Auchet, Biophys. Chem., 1, 266 (1974). (17) D. Genest and Ph. Wahl, Biophys. Chem., 7, 317 (1978). (18) Pointed out by an anonymous referee.
