J. Mol. Bid. (1975) 92,449-466
Circular Dichroism of Flow-oriented
Nucleic Acids
I. Experimental Results Su-YUN
&iVNO
AND
G. HOLZWARTHt
Department of Biophysics University of Chicago Chicago, Ill. 60637, U.S.A. (Received 20 August 1974) Specific ultraviolet circular dichroism spectra with light propagating parallel to the helix axis (Ed - ~a),, and perpendicular to the helix axis ( l n - +)A have been measured for DNA in buffer, DNA in high salt, DNA in ethylene glycol, and double-stranded viral RNA in buffer. The data are obtained by measuring both the CDS and the absorption spectra of nucleic acid solutions oriented by shear gradient in a multi-capillary flow cell with the incident light beam parallel to the direction of flow. It is shown that the conservative isotropic CD spectrum of DNA in buffer originates almost entirely from (en - en),, ; the ( l n - l n)1 component is nearly zero. The predo minance of the conservative ( l n - en),, spectrum over the (Q - ~a)~ spectrum indicates that the base planes of DNA in buffer are essentially perpendicular to the helix axis as in the B-form. The unusual ss well as in 95% profile of the isotropic CD spectrum of DNA in 6 M-Lick, ethylene glycol, is a result of extensive cancellation between the positive conservative ( l n - en),, component and the negative non-conservative (Q - •n)~ component. For double-stranded RNA, the specific CD pattern observed is very is positive nondifferent from that of B-form DNA. The ( l n - en),, component conservative and (Q - QR)~ is positive conservative. The nature of these specific CD spectra implies that the base planes are tilted with respect to the helix axis. The CD measurements on flow-oriented solutions presented here reveal the directional nature of the interaction between light and the nucleic acid molecules ; this directional nature is obliterated in conventional CD measurements of isotropic solution.
1. Introduction The conformation of nucleic acids in solution is known to vary with pH, temperature, and solvent conditions. Much of our insight into nucleic acid conformations in solution comes from ultraviolet circular dichroism and optical rotatory dispersion studies (for reviews see Yang & Samejima, 1969; Brahms & Brahms, 1970; Bush & Brahms, 1973). Conventionally, all CD$ measurements for nucleic acids are made on isotropic media, in which the nucleic acid molecules orient randomly in space. However, the circular dichroism of an individual molecule is dependent on the orientation of the molecule in the light beam (Tinoco & Hammerle, 1956; Stephen, 1967; Tinoco, t To whom reprint requests should be addressed. Present Laboratories, P.O. Box 46, Linden, N.J. 07036, U.S.A. $ Abbreviation used: CD, circular dicbroism. 449
address:
Exxon
Corporate
Research
S. CHUNG
450
AND
G. HOLZWARTH
1962). Therefore, one would expect to gain more structural information if one could measure the CD of an oriented solution. We report such measurements in the present paper. The CD measured for randomly oriented polynucleotides, (en - ~a), is an average of the specific CD with light propagating parallel to the helix axis, (Q, - E&, and the specific CD with light propagating in the two directions perpendicular to the helix axis, (en - E& and (en - &a. Thus, (EL - 4
= [(EL - 411 + (CL - 422 + (EL - %)X4/3.
Because each specific CD curve can be positive or negative, the spatial averaging which occurs in isotropic solution often involves extensive cancellation among these specific CD curves. In order to orient the molecules,
we have constructed
an optical
flow cell composed
of a parallel array of small-bore capillary tubes. When a nucleic acid solution is pumped through the capillaries, the molecules become oriented. Both CD and absorption spectra can be measured on such oriented solutions with the light propagating parallel to the direction of flow. The absorption curves are used to estimate the degree of orientation; this allows us to evaluate (en - &a and the average of (en - E& and (en - ea)ss. The experiments therefore yield significant data not obtainable in the isotropic CD measurement and reveal the directional nature of the interaction between light and the polynucleotide. In this paper we present the CD and absorption spectra for various nucleic acid solutions partially oriented by flow. The solutions studied are DNA in buffer, DNA in concentrated salt or ethylene glycol solution, and double-stranded viral RNA in buffer. In a study to be published elsewhere (Chung & Holzwarth, unpublished data) we attempt to construct a link between our observations and molecular conformation by calculating the CD curves of oriented polynucleotides in A, B and C-DNA forms, as well as RNA-11-a (Fuller et al., 1965; Langridge et al., 1960; Marvin et aE., 1961; Arnott, 1970). We use Johnson & Tinoco’s (1969) theory for this calculation.
2. Materials and Methods (a) Materials Calf thymus DNA (M, w 6 x 10s) was purchased from Worthington Biochemical Corporation. The double-stranded RNA (M, 1 to 2 x 106), from silkworm cytoplasmic polyhedrosis virus, was a generous gift of Dr K. I. Miura (National Institute of Genetics, Misima, Japan). The isolation and characterization of this RNA have been reported (Miura et al., 1968). Ethylene glycol and LiCl were obtained from J. T. Baker Chemical Co. and from Mallinkrodt, respectively. The LiCl was purified with activated charcoal and filtered through a 0.65 pm Millipore filter. Solutions were prepared by first dissolving the nucleic acid in a small amount of phosphate or Tris*HCl buffer (pH 7-O). This stock solution was then diluted to the desired fhral concentration. LiCl with Tris*HCl buffer or ethylene glycol with appropriate electrolyte were added dropwise to the stock solution at 0°C. The resultant 4 final solutions had the following compositions: (1) 0.01 M-NaF, 0.01 M-sodium phosphate buffer (pH 7*0), termed “bu.tYer” for DNA; (2) 6 M.-LiCl, 0.01 M-TriseHCl (pH 7-O), low4 M-EDTA, termed “high salt” ; (3) 95% ethylene glycol, 0.06 M-KF, low4 M-EDTA, termed “ethylene glycol” ; (4) 0.1 aa-NaF, 0.01 ~-sodium phosphate buffer (pH 7*0), termed “buffer” for RNA. Polymer concentrations were determined speotrophotometrioally, using czzao= 6600
CIRCULAR cm-l
M-I(P)
for DNA
and RNA
DICHROISM in buffer,
t Hearst, 1968), and czeo = 6790 cm-l
~-l
OF NUCLEIC
461
ACIDS
ss well as for DNA in high salt (Tunis-Schneider f or DNA in ethylene glycol (Green & Mahler,
1968). (b) Flow system The apparatus used to orient the polymers includes a peristaltic pump (Harvard Apparatus no. 1215), a glass flow meter (Gilmont Instruments), an air-bubble trap, and a flow cell. The inlets and outlets of these elements were interconnected in series by Tygon or Teflon tubing to form a closed circulating loop. FluId outlet
FluId Inlet
Light
A’
f
Window
Light
\
Body (plastIcI
Window
FIG. 1. Sohematio cross-seational view of the flow oell used to measure absorption spectra and CD curves of oriented solutions. Fluid enters the tube at the left, then flowe to the oapillary array via a dead-space between window and the open ends of the tubes. After passing down the length of the capillaries the fluid leaves via the outlet at right. The dead-space is much smaller in practice than shown in the diagram. In addition, a device which guarantees radially symmetrio flow into the dead-space has been deleted for alar&y. Light passes through the cell parallel to the oapillary axes.
The flow cell was designed to obtain a high degree of orientation and at the same time, to allow a maximum amount of light to pass through the cell. The cell, shown schematically in Fig. 1, is constructed of a capillary centerpiece, two fused silica windows and a structural body. The centerpiece consists of 125 round capillary tubes glued together in hexagonal close-packing array with axes precisely parallel to one another. The borosilicate glass capillaries, manufactured by Friedrich and Dimmock, Inc., have O-5 mm inner diameter, 0.1 mm wall thickness, and 28 mm length. The space between the inner surfaces of the two windows is 30 mm, leaving a 1 mm gap at each end between the open ends of the capillaries and the window for fluid entering and leaving the capillaries. The body is specially constructed to bring fluid into the centerpiece symmetrically from all radial directions (not shown in Fig. 1).
462
S. CHUNG
AND
G. HOLZWARTH
An essential feature of this cell is that the mean orientation of macromolecules has cylindrical symmetry and the propagation direction of light is parallel to the unique direction of orientation (cylinder axis). This has two advantages over the conventional methods for measuring flow dichroism (Wada & Kozawa, 1964) in whieh light propagates perpendicular to the direction of orientation. First of all, molecules such as nucleic acids, with anisotropic orientation of chromophores, possess strong linear dichroism for light propagating in a direction perpendicular to their helices. If light propagates perpendicular to the direction of orientation, the presence of linear dichroism, which may be 1000 times greater than the CD, causes an artifact in the oriented CD measurement by several identified mechanisms (Disch & Sverdlik, 1969; Tunis-Schneider & Maestre, 1970), and perhaps also by non-ideality of the Pock&’ cell (Velluz et al., 1965). In our apparatus, by contrast, a light beam propagating down the capillary axis sees very little linear dichroism. The second advantage of our system is t,hat it allows linear dichroism of long-chain molecules to be studied with an unpolarized light beam in an ordinary spectra . photometer. Four hydrodynamic factors also entered into the design of the cell and the flow rates used with it (Chung, 1974). (1) The average shear gradient (G} should be as large as possible in order to maximize the degree of orientation of the polymer. This means that the tube radius r0 should be small, because for a long round capillary (G) is equal to 8Q/3r03, where Q is the average volume flow rate. (2) In order to maintain laminar flow, the Reynolds number should not exceed 2000. (3) The tubes must be substantially longer than the transition length required to establish the classical parabolic velocity distribution in the capillary (Langhaar, 1942). (4) The time required to orient the DNA molecule in the tube must be short compared to the residence time of the molecule in the tube. Under the conditions of our experiments, the Reynolds number is 500 or less, the transition length is 1 cm or less, and the residence time is about 3 times the relaxation time (Callis & Davidson, 1969b). The degree of orientation obtained ranged from 50/:, t.o 3O”/b depending on the nature of the solution.
(c) Optical mea8urement.s All ultraviolet absorption spectra were measured with a Cary model 15 spectrophotometer. Absorption spectra of oriented solutions were measured by placing the flow cell (Fig. 1) in the Cary 16 sample compartment. The cell was aligned with capillary axes parallel to the propagation direction of the unpolarized spectrometer beam. For transitions polarized perpendicular to the helix axis, the absorption is expected to increase when the molecules are oriented by flow; for transitions polarized parallel to the helix axis, the absorbance will decrease with flow. In our experimental conditions, the absorbance responds almost instantaneously as the flow is turned on and off. The flow-induced absorbance change, AAf, which is equal to the absorbance with flow, A,, minus the absorbance without flow, A, is constant for a given flow rate and is reversible. This is sufficient indication that the polymer is not degraded by shear. Moreover, the intrinsic viscosity of the macromolecules measured after the flow experiment was not significantly AAf/A increases with increasing different from that measured before. The flow dichroism (G) and approaches a limiting value. This limiting value is determined by the molecular weight of DNA (Callis & Davidson, 1969a) and by the hydrodynamic properties of the flow cell. Most of the measurements reported in this paper were made at (G> = lo4 s-l. The CD measurements were made on a Cary 60 recording spectropolarimeter with a model 6002 CD aooessory. The flow cell was located in the sample compartment of the Gary 60 on a special cell holder which allowed precise alignment with respect to the incoming beam. The alignment of the flow cell is critical because any residual linear dichroism appears as an artifact in the measurement. The cell was aligned by minimizing its linear dichroism measured in the Cary 60 using a method described previously (Mandel & Holzwarth, 1970). The error introduced into the oriented CD measurement by the residual linear dichroism artifact is less than 5%. We were also concerned that polarized light propagating down the very narrow capillaries might be depolarized by reflection from the capillary walls. The depolarization of the beam by the flow cell was tested by measuring the CD of a d-lo-camphor sulfonic acid
CIRCULAR
DICHROISM
OF NUCLEIC
ACIDS
463
solution after the beam passed through the flow cell which was filled with dilute nucleic acid solution (with or without flow). The depolarization effect was found negligible. 3. Results (a) DNA in buffer solution The flow dichroism and flow circular dichroism of DNA in buffer (O-01 M-NaF, 0.01 M-sodium phosphate, pH 7.0) was measured at DNA concentration 5 x 10m5 r~(P). The absorption spectra with and without flow between 220 and 320 nm are shown in Figure 2. The absorbance increases as the DNA helices are lined up along the direction of light propagation; this is in accord with the expectations if the transition dipole moments are polarized in the plane of the bases and the bases are perpendicualr to the helix axis. The flow dichroism AA,/A is constant between 280 nm and 240 nm in the central portion of the 260 nm absorption band. This observation is in agreement with previous linear dichroism studies using a different type of apparatus, where the light travels perpendicular to the direction of orientation (Wada, 1964; Gray & Rubenstein, 1968). On both the high and low energy sides of this band, the flow dichroism decreases. On. account of the uncertainty in the measurement of small absorbance change, the drop of AA,/A value above 280 nm is not established with certainty. However, the decrease in AA,/A below 240 suggests the existence of a substantial transition, presumably n + n*, polarized parallel to the helix axis (Callis $ Davidson, 19690; Gray & Rubenstein, 1968).
-4
a
Wavelength
P
(nm)
FIQ. 2. Absorption and flow diahroism of calf-thymus DNA in 0.01 M-N@ and 0.01 x-phosphate buffer, pH 7.0. The average shetlr gradient
is about 104s-‘*dA,/A = (An,l - &,rl,,)/ A no Ilor~
The flow CD, measured under conditions identical to those of the flow dichroism, is shown in Figure 3. The CD without flow is the familiar conservative spectrum with a peak at 275 nm, a trough at 245 nm, and a crossover point at the maximum of the absorption band, 258 nm (Brahms & Mommaerts, 1964). When the solute molecules are oriented by flow, both the peak and the trough increase in amplitude without a significant shift in their position or in the crossover point. Thus, the flowinduced CD change A( cL - E& which is equal to the CD with flow (cL - C& minus
S. CHUNG
464
AND
G. HOLZWARTH
..‘. ; ‘: :.,:
30
240
200
No flow ,
280 Wavelength
320
FIQ. 3. Circular dichroism of flow-oriented calf thymus DNA in 0.01 M-NaF and 0.01 M-phosphate buffer, pH 7.0. DNA concentration is 5 x 10e5 M(P) and the average shear gradient
Under the oonditions
-
4
=
of this experiment,
(EL
-
d&/A
&ow
-
(CL
-
&II
ilOW
= 0.09 at 260 mn.
the CD without flow (Ed - eR), has the same shape as the istropic CD spectrum. However, the magnitude of A(cL - E& is determined by the degree of orientation. Very similar data were obtained previously for calf thymus DNA (Mason & McCaffery, 1964) and for salmon sperm DNA in buffer (Wooley & Holzwarth, 1971). The characteristic flow-induced absorption and CD changes were not observed for DNA denatured by heat. It is important to note that the flow-induced CD changes varied with shear gradient in a manner exactly parallel to the absorbance changes. This proves that the changes in CD spectrum with flow result from orientation, not from oonformational change. (b) DNA in high salt and ethylene glycol The flow dichroism absorption spectra of DNA in high salt (6 M-Lick, 0.01 &r-T&. HCl and 10e4 M-EDTA) and in ethylene glycol (0.05 M-KF, 10v4 M-EDTA, and 95% ethylene glycol) are both very similar to the spectra in buffer. The flow dichroism AA,/A at a fixed flow rate is constant in the central region of the absorption band between 240 and 280 nm. No absorbance or absorbance change is detected in the non-absorbing spectral region (X > 310 run). The CD spectra with and without flow for DNA in 6 M-LiCl are shown in Figure 4. The unoriented solution CD has been reported before (Nelson & Johnson, 1970;
CIRCULAR
DICHROISM
I 10
I
L
OF NUCLEIC
I
240 Wavelength
I 280
I
ACIDS
466
I 320
(nm)
FIG. 4. Circular diohroism of flow-oriented calf thymus DNA in 6 M-LiCl, 0.01 M-TrisaHCl and lo-* M-EDTA. DNA conoentration is 6 x 10e6 M(P) and the units of circular dichroism are liters per mol cm on a nucleotide basis.
4% - %)I = (EL- 4lOW - (CL- l a)nonon. Under the conditions
of this experiment
AA,/A = 0.07 at 260 nm.
Tunis-Schneider & Maestre, 1970). It consists of a prominent negative band at 245 nm, which has about the same peak position, shape and magnitude as the negative lobe of the conservative spectrum of DNA in buffer, and several less distinct weak bands between 265 and 310 run. Astonishingly, with flow, the CD spectrum changes shape remarkably; the CD increases greatly above 260 nm, peaking at 275 nm, whereas the CD change below 260 nm is relatively much smaller. As a result, the difference curve d(cL - E& is semiconservative. Unoriented DNA in ethylene glycol (Fig. 5) has a non-conservative CD spectrum similar to that of DNA in LiCl (Green & Mahler, 1971). The CD spectral change which accompanies orientation in ethylene glycol is shown in Figure 5. The observed curve is almost purely positive non-conservative and resembles that observed for DNA in 6 M-LiCl. The changes in absorbance and CD again vary with flow rate in a parallel manner for DNA in both LiCl and ethylene glycol; this excludes the possibility that the flow-induced CD change in either case is due to flow-induced conformational change. (c) Double-stranded RNA in buffer solution The flow dichroism of double-stranded cytoplasmio polyhedrosis virus RNA in buffer (0.1 x:-NaF, 0.01 M-phosphate buffer, pH 7.0) was measured at RNA concentration 3.5 x 10es M(P). The flow dichroism is similar to that of DNA in buffer
466
S. CHUNG
AND
G. HOLZWARTH
(Fig. 2) ; &&/A remains constant in the central region 240 nm to 280 nm of the absorption band due to the w --f r * transition. As in the case of DNA, the dichroism L&/A decreases below 240 nm. This implies the existence of a transition parallel to the helix axis in this region, possibly an n -+ ZT* transition. Our observed flow dichroism is consistent with the previous result obtained by a different apparatus, where the
I
I
I
240
'0
I
I
200
320
Wavelength
FIU. 6. Cirouhr diohroism of flow-oriented odf thymus DNA in 95% ethylene glyool, O-05 M-KF, and lo-4 M-EDTA. DNA oonoentration is 6.6 x 10m6 M(P) and the units of oircular diohroism are liters per mol cm on a nuoleotide basis. 4% Under the conditions
-
4
= (EL -
of this experiment
L&/A
&ow
-
( EL -
4nomv.
= 0.14 at 260 nm.
\
a0 200
I 240
I
Wavelength
I 280
I
1 320
(nm)
FICA 6. Circular diohroism of flow-oriented, double-stranded, oytoplasmio-polyhedrosis-virus RNA in O-l M-N&F and O-01 M-phosphate buffer (pH 7.0). RNA concentration is 3.5 x 1Om6 M(P) and average shear gradient is 10” s- I. The units of oiroular diohroism are liters per mol cm on a nucleotide basis. Under the conditions of this experiment AA,/A = 0.06 at 260 nm.
CIRCULAR
DICHROISM
OF NUCLEIC
ACIDS
457
light propagates perpendicular to the direction of orientation (Wada et al., 1971). Moreover, no absorbance or absorbance change was observed extending beyond the absorptive spectral region (h > 310 nm). The flow CD spectra measured under identical conditions as the flow dichroism are presented in Figure 6. The CD spectrum for the unoriented solution displays the typical non-conservative profile of double-stranded RNA (Miura et al., 1968; Samejima et al., 1968). It consists of an intense positive peak at 260 nm with two minor negative peaks on each side (295 and 235 nm), and an equally intense negative peak located at 215 nm. At steady-state flow, the positive peak increases in direct proportion to the flow dichroism AA,/A. However, no significant change for X > 280 nm was observed.
4. Resolution of Spectra The CD of a flow-oriented solution is determined by the specific CD of the constituent molecules and by the orientations of the molecules with respect to the incident light. Molecular quantities may be extracted from the CD measurement of floworiented nucleic acid solutions (Pigs 3 to 6) by using the absorption data to estimate the degree of orientation, as follows. Assume that the nucleic acid molecules are rigid rods with cylindrical symmetry. A set of orthogonal principal axes (1, 2, 3) is chosen as the molecule-fixed co-ordinate system with S-axis coincident with the helix axis. Similarly, a second set of orthogonal axes (l’, 2’, 3’) is chosen as the space-fixed co-ordinate system with the 3’-axis coincident with the direction of light propagation. (Capillary ttt Flow
axis)
3’
FIG. 7. Co-ordinate system for partially oriented mecromolecule. The helix axis is designated as 3-axis. The space-fixed co-ordinates [l’, 2’, 3’1 are m-ranged with 3’-axis parallel to the capillary axis. The angles 0 and 4 define the orientation of the helix in the [l’, 2’, 3’1 co-ordinate frame.
As shown in Figure 7, the orientation of the helix axis of a molecule in space is then specified by two angles ti and 4 (Fig. 2). For a partially oriented system, the orientational distribution functionf(0,$)d8d$ (Scheraga et al., 1951) is then used to describe the fraction of molecules with helix axis oriented between 0 --f 8 + d8 and Q, -+ # + d#. In the special case when f is independent of +, which is true for our flow cell on the average, the absorption and CD of a partially oriented solution, as measured with light
S. CHUNG
458
AND
G. HOLZWARTH
propagating in the 3’-direction, can be expressed (Wada & Kozawa, 1964; Go, 1967; Chung, 1974) as A, = (1 - b)A + b/l, (1) (EL,- %)t = (1 - b) (EL - 4
+ b(% - %)33 >
(2)
where A, and A are the absorbance with and without flow, (fL - E& and (Ed - cR) are the CD with and without flow. The molecular quantity A, is the absorbance measured with light polarized perpendicular to the helix axis and (cL - c&s is the CD measured with light propagating parallel to the helix axis. The parameter b is related to the orientational distribution function f (0,$) by b = (VW J-x” j; [(3 co&’ - l)/Z]f(e,$) The orientational
distribution
sinedBd+ .
(3)
function f (e,#) is normalized:
The form of equations (1) and (2) deserves comment. The equations imply that the partially oriented system can be viewed as a combination of two well-defined fractions of molecules. One fraction of molecules, b, is perfectly oriented parallel to the light propagation direction, whereas the remaining fraction of molecules, (1 - b), is totally random. Thus, b serves as a parameter indicating the degree of orientation. In the special case of an isotropic solution, we have b = 0 and equations (1) and (2) are then reduced to the expressions A = (A,, + 2A,)/3
(EL- %J = [(CL- 411 + (EL- %)a2 + (EL- 4331/3 9
(4)
(5)
where A,, is the absorbance measured with light polarized parallel to the helix axis and (en - eahl and (eL - ~s)ss are the CD measured with light propagating in the 1 and 2 directions, which are normal to the helix axis. In order to apply equations (2) and (5) to analyze the observed isotropic and tloworiented CD spectra (Figs 3 to 6) in terms of intrinsic molecular quantities (Ed - ~s)~~, h - d2, and k L - ~a)ss, we need to know the degree of orientation b, which can be determined from the absorption measurement with and without flow. Since the transition electric dipole moments for the 260 nm rr -+ V* transitions are polarized parallel to the base planes (Fucaloro & Forster, 1971), the specitic absorbances A,, and A, depend in a simple way upon the isotropic absorbance A and the inclination angle CCbetween the normal of the base plane and the helix axis: A, = (3/2)A cosaa
(6)
A,, = 3A sirPa
(7)
dA,/A
= b(l - 3 sin2a)/2 .
(8)
In equation (El), we can use the measured value of AA,/A to evaluate b, provided we have a reasonable estimate of a. In practice, we do not need to know a with high precision, since a is small for nucleic acid structures and thus the term 3 sirPa makes only a small contribution to the estimated b value. For instance, 3 &Pa varies from 0 to 0.1 as a increases from 0” to 10”.
CIRCULAR
DICHROISM
OF NUCLEIC
ACIDS
459
Let us first resolve the spectra of DNA in buffer (Figs 2 and 3). Under buffer conditions employed, the DNA conformation is presumably B-form (Bram & Beeman, 1971). According to the X-ray diffraction data on the molecular model of B-DNA, the normals of the base planes make an angle of 4 to 5” with respect to the helix axis. Substituting this value and flow dichroism value AA,/A = 0.09 in equation (8) the degree of orientation b is determined as O.lSjO.02. It should be pointed out that this b value is estimated on the assumption that the 260 nm transitions are polarized 100% in the base planes.
-1 I/ I 1 I i
8 4 n [Q 2 -4 -8
!O u‘0
240 Wavelength
(nm)
FIG. 8. Observed speoific 8nd isotropic oircular dichroism spectra, of DNA in 0.01 M-N&F and 0.01 an-phosphate buffer (pH 7.0). --O--O--, Smn of the specific circular diohroism spectra perpendicular to the helix axis (Q - cB)ll + (Ed - +&; (* . . . . . . * . . ) specifk ciroular dichroism spectrum parallel to the helix axis (cL - s&; and ( ) isotropic ciroular diahroism ( cL - +J. The units of circular dichroism are liters per mol cm on a nucleotide basis.
Based on 6 = 0.18 and equations (2) and (5), the resolved specific CD spectra parallel to the helix axis, (cL - ~&a, and perpendicular to the helix axis, (eL - E& + (9, - 42% are given in Figure 8. The latter quantity is given as (cL - E&~ + (Q, - +& because for a system with cylindrical symmetry, (E= - Q& and (EL - GJ22 are indistinguishable experimentally. In practice,
which is just twice the CD one would observe for light propagating perpendicular to the helix axis. The accuracy of band amplitudes are determined to about &lo%. These uncertainties are mostly from the initial errors in the measurement of (Ed - E& and less in the estimation of parameter b. The most important feature in these curves is the relative magnitude of the different specific CD spectra. The peak intensity of (Ed - E& is at least fifteen times larger than the (Ed - l a)* component. As a result, the isotropic CD, (Ed - ~a), arises mostly from (Ed - l &a. The (Ed - ~&a spectrum, peaking at 275 and 245 nm, has all the characteristics of the conservative spectrum of (Q, - +J. 30
460
S. CHUNG
AND
G. HOLZWARTH
In order to evaluate the degree of orientation of the flowing DNA in high salt and ethylene glycol, we need to know the base inclination angle a with respect to the helix axis. The precise conformation of DNA in these solvents is not known, but we tentatively assume that the DNA is a C-form-like structure as suggested by Nelson & Johnson (1970). This gives a = 8”. The measured values of flow dichroism AA,/il were 0.14 and 0.07 for DNA in high salt (Fig. 4) and in ethylene glycol (Fig. 5) ; the parameter 6 is thus found to be 0.29f0.02 for DNA in high salt and 0*14&0.02 for DNA in ethylene glycol. It is worthwhile to point out that the calculated value of b is rather insensitive to a small variation in the base orientational angle a, when a is small. For instance, b varies from 0.28 to 0.33 for a = 3” to a = 13”. Thus, as long as the base orientation is not far from 8”, our estimated b values should be reasonable.
I 200
I 240
I
Wavelength
I 200
320
(nm)
FIG. 9. Observed specificand isotropic circulrtr dichroism spectra, of DNAin 6a6-LiCl. --O--O--, Sum of the specifio circular dichroism speotraperpendicula to the helix axis (q, - +JI1 + ( Q~-
Based on the above estimated b values, the resolved specific CD spectra along and perpendicular to the helix axis for DNA in LiCl and DNA in ethylene glycol are given in Figures 9 and 10. The general patterns of these resolved spectra are extremely similar for DNA in 6 M-LiCI and in ethylene glycol. The (Ed - +&s curve is semiconservative with a positive peak at 274 & 2 nm and a smaller peak at 245 f 2 nm. The sum of the (Ed - E& and (Ed - E& is comparable in size with the (Ed - E& component but opposite in sign. Accordingly, the unusual pattern of the isotropic spectrum results from the large cancellation between these two components. The oriented CD data thus provide further evidence that the conformation of DNA in LiCl or ethylene glycol differs from that of DNA in buffer. If the conformation in buffer is precisely B-form, then the conformation in LiCl is not B-form. However, the
CIRCULAR
DICHROISM
1
OF NUCLEIC
ACIDS
461
200
Wavelength
(nm)
10. Observed specific and isotropic circular dichroism spectra, of DNA in ethylene glyool. 0, -O--, Sum of the specific circular diohroism spectra perpendicular to the helix axis (EL - E&1 + (EL - E&z; (-. . . . . . . . .) specific oiroular diohroism spectrum parallel to the helix ) isotropic circular diohroism (c,, - cR). The units of oircular axis (Ed - c&; and ( diohroism are liters per mol cm on a nuoleotide basis. Fm.
conformation in LiCl then also may differ from the C-form found for Li-DNA fibers by X-ray analysis. We will explore this point, as well as contrary X-ray studies, in the Discussion. Next we will consider the resolution of the specific CD spectra of double-stranded RNA. A preliminary X-ray diffraction analysis (Miura et aZ., 1968) reports that CPV RNA has a structure very similar to that of rice dwarf virus RNA and other double-stranded RNAs (Sat0 et al., 1966; Langridge et al., 1964). Therefore, we have used the refined molecular model of double-stranded RNA-11-a form (Arnott et d.. 1969), in which the base orientation angle cc with respect to the helix axis is 14”. Substituting cc = 14” and the flow dichroism AA,/A = 0.06 into equation (8), the that the degree of orientation parameter b is obtained as 0.14 f 0.02. It is noteworthy degree of orientation obtained for double-stranded RNA is very close to that of the DKA. This is a fair indication that the RNA molecules are quite rigid under the current conditions. Based on b = O-14, the flow CD spectra are analyzed and resolved into two independent specific CD components. The CD spectra with light propagating parallel to the helix, (en - ~&,a, and perpendicular to the helix axis, (en - E& + (ei, - eR)a2, with a single are shown in Figure 11. It is seen that (en - ~&a is non-conservative intense positive peak at 255 nm, whereas the sum (c~ - E& + (Ed - l &2 is conservative in profile with the positive and negative bands situated at 275 nm and 245 nm, respectively. These results are in strong contrast to the specific CD spectra of DNA in which (Ed - &a is conservative whereas the sum of (Ed - cR)ll and is non-conservative and small. For RNA, the maximum intensity of (EL - %J22
402
S. CHUNG
AND
I 200
G. HOLZWARTH
I 240 Wavelength
I 280
1 320
(nm)
FIG. 11. Observed specific and ieotropio ciroular dichroism spectra of double-stranded RNA in 0.1 M-N~.F and 0.01 aa-phosphate buffer (pH 7.0). --O--O--, Sum of the specific ciroular diohroism spectra perpendiculez to the helix axis (q, - E& + (eL - r&2; (. . . . . . . . . .) speci& ) isotropic oimular dichroism spectrum parallel to the helix axis ( cL - +Ja3; and ( airoular diahroism (cL - c~). The units of ciroular dichroism are liters per mol cm on a nucleotide
(eL - ~a)~~+ (Ed - l a)ss is about l/3 to l/4 that of (Q, - e&s. As in the circular dichroism spectra of isotropic solution, there is a minor negative peak at 295 nm for (Ed - E& + (en - E&~. The origin of this negative band around 295 nm is not clear; it has been tentatively attributed to an n -+ rr* transition (Sarkar et al., 1967).
5. Discussion The only previous experimental CD study of oriented polynucleotides is that of Mason & McCaffery (1964), who studied DNA in buffer. Our results are in qualitative agreement with theirs. However, the earlier workers did not know the degree of orientation of their sample. A quantitative comparison is therefore precluded. In order to facilitate a structural interpretation of CD spectra, it is useful to compare the measured results to CD curves calculated for the geometries deduced from analysis of X-ray difFraction by fibers. Johnson t Tinoco (1969) have developed a theory which predicts with substantial success the isotropic CD spectra of polynucleotides. Their theory considers only interactions among the TI + VT*transitions of the bases. We have extended their theory to allow prediction of oriented CD spectra in a subsequent paper (Chung & Holzwarth, unpublished data). For DNA in buffer solution, we observed that the specific CD parallel to the helix is strictly conservative and is much larger than the specific CD atis kL - d3 perpendicular to the helix axis (cL - •a)~ which is near zero. This is in excellent accord with the theoretical prediction for the CD spectra of the molecular model B-form (Chung & Holzwarth, unpublished data). This adds independent evidence that DNA in buffer exists in B-form. Moreover, the surprisingly good agreement between the experiment and theory for B-DNA suggests that the theory has predictive value.
CIRCULAR
DICHROISM
OF
NUCLEIC
ACIDS
463
The structure of DNA in high salt and in ethylene glycol has been controversial. The specific CD spectra in these solvents are readily distinguished from those obtained on DNA in low-salt buffer. There are three possible explanations for these differences: (1) solvent-induced change in secondary structure; (2) differential scattering of the left and right-circularly polarized light from compact or aggregated DNA molecules; (3) ordered tertiary structure of the DNA molecules. The light scattering effects on absorption and CD curves are now partially understood (Gordon, 1971; Schneider, 1973). The most obvious evidence for light scattering effects is a substantial long tail in the non-absorbing region (X > 320 run) in both absorption and CD spectra. Such an effect has been observed for the absorption and CD spectra of T2 phage (Dorman & Maestre, 1973). However, we do not see such anomalous spectral tails in both absorption or CD measurements with or without flow. The absence of such light scrtttering effect argues against (2) and (3) above. Moreover, we verified that the measured CD expressed as molar extinction coef%cient differences (eL - Ed) is independent of DNA concentration over the range 5 x 1O-5 M(P) to 1.5 x 10m3 M(P). The DNA concentration used in the flow experiment is extremely low, about 5 x low5 M(P). This is not a favorable condition for intermolecular aggregation. In addition the hydrodynamic behavior of DNA in the presence of univalent cations does not support an aggregation model (Ross & Schruggs, 1968). Finally, the systems which are most susceptible to intermolecular aggregation exhibit enormously enhanced CD spectra, sometimes orders of magnitude stronger than the conventional CD spectra. For instance, the DNA-poly-L-lysine complex (Shapiro et al., 1969; Haynes et al., 1970) and DNA in polyethylene oxide (Jordan et al., 1972) which have been proved to exist in large aggregated form, give rise to CD about ten times larger than the normal CD of nucleic acids without grossly disturbing the secondary structure of DNA. The CD of DNA in ethylene glycol and high salt definitely do not belong to this category at least in terms of magnitude. Based on the above erguments, we conclude that the DNA molecules in high salt or ethylene glycol possess a novel secondary structure which is different from that of DNA in buffer solution. If DNA in LiCl or ethylene glycol is not in the standard B-form, what is its conformation under these solvent conditions? We know that the DNA does not exist as a denatured random coil because it remains hypochromic relative to thermally denatured DNA (Green & Mahler, 1968,1971). On the other hand, if electrolyte is not provided in the ethylene glycol solvent, the DNA does become denatured. Tunis-Schneider & Maestre (1970) have shown that the CD spectra of unoriented LfDNA %lms, at 75% humidity, are very similar to the CD spectra in ethylene glycol and in high salt. Since Li-DNA fibers at 75% humidity exist in C-form by X-ray analysis (Marvin et al., 1961), several groups (TunisSchneider & Maestre, 1970; Nelson t Johnson, 1970) have proposed that DNA in high salt is intermediate between B and C-forms whereas the DNA structure in ethylene glycol is a C-like conformation. Studies of the conformational changes of DNA monitored by CD, with increasing ethylene glycol or LiCl concentration, show that the transition is non-co-operative (Ivanov et d., 1973; Chung, 1974). These results suggest that the DNA conformation is gradually perturbed from the regular B-form as the LiCl or ethylene glycol concentration increases. However, at 6 M-LiCI or 95% ethylene glycol, the canonical C-form is not necessarily the end point of this non-co-operative transition.
464
S. CHUNG
AND
G. HOLZWARTH
As an aid to the interpretation of the experimental CD data for DNA in 6 M-LicI, we have calculated the specific CD curves of C-DNA. The results are presented in detail elsewhere (Chung & Holzwarth, unpublished data). We 6nd that the profile of the specific CD spectra of DNA in ethylene glycol and 6 M-Lick are in fairly good agreement with those of calculated C-form, although the isotropic CD curves are not in good accord. Both theory and experiment show a large cancellation between Moreover, the observed differences in kL - 43 and (6~ - QA~ + kL - d2. specific CD curves between DNA in buffer and DNA in LiCl are consistent with the calculated differences between B and C-form. These results suggest that the base conformation of DNA in LiCl or ethylene glycol is similar to the base conformation of the molecular model of C-DNA. Different conclusions have been reached from X-ray scatteriug. Luzzati and co-workers (Luzzati et al., 1964) found some years ago that the small-angle X-ray scattering curve of DNA in ethylene glycol (0.1 M-NaCl) is in B-form. More recent wide-angle X-ray scattering studies of dense gels of DNA in 6 M-Lick also indicate that DNA under these conditions is in B-form (Maniatis, 1971; Carlson, 1973; Maniatis et al., 1974). The origin of the discrepancy between X-ray scattering and CD conclusions is not clear; it is possible that a role is played by the 104-fold difference in DNA concentration necessary for the two techniques. Alternatively, the two techniques may have unequal sensitivity to speci6c structural features such as the base conformation or the structure of the sugar-phosphate backbone. In the case of RNA, the strictly non-conservative nature of the observed (E= - E&, and the conservative nature of the observed (E= - E& + (E= - E&~ are in marked contrast to those for DNA in buffer. These data support earlier conclusions from isotropic CD studies that the RNA has a very different conformation from DNA. X-ray analysis shows that in RNA fibers the base planes tilt at an angle of 14” toward the helix axis ; moreover, the base pairs move a distance of 5 A along the positive dyad axis relative to the base-pair positions in B-DNA (Arnott, 1970). The specific CD data presented here, when coupled with calculations of the specific CD spectra to be presented elsewhere (Chung & Holzwarth, unpublished data), suggest that similar structutal differences between RNA and B-DNA also occur in solution.
6. Conclusions The important features of the experimentally measured specific circular dichroism spectra parallel to the helix axis, (cL - Q&, and perpendicular to the helix axis, (Ed - E& + (Ed - &c, are summarised as follows. (1) For DNA in low salt buffer (Ed - E& is much larger than (Ed - E& + (Ed - E&,~, and is strictly conservative. The isotropic circular dichroism (Ed - zR) is mostly contributed from the (EL - E& component. The nature of these spectra suggest that DNA exists in the B-form, in which the base planes are perpendicular to the helix axis. (2) For DNA in high salt or in ethylene glycol, the specific circular dichroism kL - d3 and (Q, - E& + (Ed - E&,~ are comparable in magnitude but opposite in sign. (Ed - l &,a is positive and semiconservative whereas (bL - E& + (Ed - +JZ2 is strictly negative non-conservative. The peculiar non-conservative shape of the isotropic circular dichroism spectrum is a result of cancellation between the two
CIRCULAR
DICHROISM
OF NUCLEIC
465
ACIDS
intense specific spectra. It is not surprising then that it is more diEcult to draw qualitative conclusions concerning the base conformation from the isotropic spectrum than from the specific spectra. By comparing the experimental and calculated data, we suggest that the base orientation and position with respect to the helix axis are similar to those of C-DNA. (3) The profiles as well as the relative magnitudes of the specific circular dichroism of double-stranded RNA are very different from those of B-DNA. In contrast to the conservative (cL - E& spectrum of B-DNA, the (en - E&, spectrum of double-stranded RNA is positive non-conservative. Moreover, the (en - ~~~~~+ (cL - E&~ spectrum of double-stranded RNA is conservative and its magnitude is by no means negligibly small as in the case of B-DNA. The isotropic circular dichroism spectrum of double-stranded RNA is non-conservative and is twice as intense as that of B-DNA. The relative intensities of the specific circular dichroism spectra, and the non-conservative nature of (en - &e spectrum are in agreement with expectations if the base configuration of double-stranded RNA and of B-DNA in fibers are maintained in solution. The results have demonstrated how the subtle combination of the two specific circular dichroism spectra gives rise to the conventional isotropic spectrum. In many cases, there is a large cancellation between the two specific spectra in making up the isotropic curve. It has also been shown that the relative magnitudes and profiles of the specific circular dichroism spectra are very sensitive to the nucleic acid conformations. We believe that this method can be profitably extended to the study of other nucleic acid conformations and to the elucidation of the interaction of nucleic acids with proteins and smaller molecules of biological importance.
The authors for his skilful
are grateful
to K. Miura
for a generous
gift of RNA
and to John
Hanacek
construction of the flow cell. This research was funded by grant NS-07286 from the U.S. National Institute of Neurological Diseases and Stroke. One of us (S. C.) was supported by National Institutes of Health training grant GM780, while the other author
(G. H.) wss the recipient
of Research
Career
Development
award
GM15050.
REFERENCES Arnott,
S. (1970).
Prog.
Biophys.
Mol.
BioE. 21, 265-319.
Amott, S., Dover, S. D. &zWonacott, A. J. (1969). Acta
Cryetallogr. sect. B, 25, 2192-2206. Brahms, J. & Brahms, S. (1970). In Fine Structure of Proteins and Nucleic Acids (Fssman, G. D. & Timasheff, S. N., eds), vol. 4, chap. 3, Marcel Dekker, Inc., New York. Brahms, J. & Mommaerts, W. F. H. M. (1964). J. Mol. Biol. 10, 73-88. Bram, S. & Beeman, W. W. (1971). J. Mol. Biol. 55, 311-324. Bush, C. A. BE Brahms, J. (1973). In Physical Chemical Properties of Nucleic Acids (Duchesne, J., ed.), vol. 2, pp. 147-186, Academic Press, New York. Callis, P. R. & Davidson, N. (1969a). Biopolymers, 7, 335352. Callis, P. R. & Davidson, N. (19695). Biopolymers, 8, 379-390. Carlson, R. D. (1973). Ph.D. dissertation, University of Wisconsin, Madison, Wisconsin. Chung, S. Y. (1974). Ph.D. dissertation, University of Chicago, Chicago, Illinois. Disch, R. L. & Sverdlik, D. I. (1969). Anal. Chem. 41, 82-86. Dorman, B. P. & Msestre, M. F. (1973). Proc. Nat. Acad. Sci., U.S.A. 70, 255-259. Fucaloro, A. F. & Forster, L. S. (1971). J. Amer. Chem. Sot. 93, 6443-6448. Fuller, W., Wilkins, M. H. F., Wilson, H. R. & Hamilton, L. D. (1965). J. Mol. BioZ. 12, 60-80.
466
S. CHUNG
AND
G. HOLZWARTH
Go, N. (1967). J. Phys. Sot. (Japan), 23, 88-97. Gordon, D. J. (1972). Biochemistry, 11, 413-418. Gray, D. M. & Rubenstein, I. (1968). Biopolymers, 6, 16051631. Green, G. & Mahler, H. R. (1968). Biopolymers, 6, X09-1514. 10, 2200-2216. Green, G. & Mahler, H. R. (1971). Biochemistry,
Haynes, M., Garret, R. A. & Gratzer, W. B. (1970). Biochemistry, 9, 4410-4416. Ivanov, V. I., Minchenkova, L. E., Schyolkina, A. K. & Poletayev, A. I. (1973). Biopolymere, 12, 89-110. Johnson, W. C., Jr & Tinoco, I., Jr (1969). Biopolymers, 7, 727-749. Jordan, C. F., Lerxnan, L. S. & Venable, J. H., Jr (1972). Nature New BioZ. 236, 67-70. Langridge, R., Marvin, D. A., Seeds, W. E., Wilson, H. R., Hooper, C. W., Wilkins, M. H. F. & Hamilton, L. D. (1960). J. MoZ. BioZ. 2, 38-64. Langridge, R., Billeter, A., Borst, P., Burdon, R. H. & Weissmann, C. (1964). Proc. Nat. Acad. Sci., U.S.A. 54, 114119. Langhaar, H. L. (1942). J. AppZ. Me&an. 9, 55-58. Luzzati, V., Mathis, A., Masson, F. & Witz, J. (1964). J. Mol. BioZ. 10, 28-41. Mandel, R. & Holzwarth, G. (1970). Rev. Sci. In&rum. 41, 755-758. Maniatis, T. P. (1971). Ph.D. dissertation, Vanderbilt University, Nashville, Tennessee. Maniatis, T. P., Venable, J. H., Jr I% Lerman, L. S. (1974). J. Mol. BioZ. 84, 37-64. Marvin, D. A., Spencer, M., Wilkins, M. H. F. & Hamilton, L. D. (1961). J. Mol. BioZ. 3, 547-565. Mason, S. F. & McCaffery, A. J. (1964). Nature (London), 204, 468-470. Miura, K., Fujii, I., Sakaki, T., Fuke, M. & Kawase, S. (1968). J. ViroZ. 2, 1211-1222. Nelson, R. G. & Johnson, W. C., Jr (1970). Biochem. Biophys. Res. Commun. 41, 211-216. Ross, P. D. & Schruggs, R. L. (1968). Biopolymers, 6, 1005-1018. Samejima, T., Hashizume, H., Imakor, K., Fujii, I. & Miura, K. (1968). J. Mol. BioZ. 34, 39-48.
Sarkar, P. K., Wells, B. & Yang, J. T. (1967). J. Mol. BioZ. 25, 563-566. Sato, T., Kyogoku, Y., Higuchi, S., Mitsui, Y., Iitaka, Y., Tsuboi, M. & Miura, K. (1966). J. Mol. BioZ. 16, 180-190. Scheraga, H. N., Edsall, J. T. & Gadd, J. 0. (1951). J. Chem. Phys. 19, 1101-1108. Schneider, A. S. (1973). Methods Enzymol. 27D, 751-767. 8, 3219-3232. Shapiro, J. T., Leng, M. & Felsenfeld, G. (1969). Biochemistry, Stephen, M. J. (1957). Proc. Camb. Philoe. Sot. 54, 81-88. Tinoco, I., Jr (1962). Adv. Chem. Phys. 4, 113-160. Tinoco, I., Jr & Hammerle, W. G. (1956). J. Phya. Chem. 60, 1619-1623. Tunis-Schneider, M. J. B. & Hearst, J. E. (1968). Biopolymers, 6, 1218-1223. Tunis-Schneider, M. J. B. & Maestre, M. F. (1970). J. Mol. BioZ. 52, 521-541. Velluz, L., Legrand, M. & Grosjean, M. (1965). Optical Circular Dichroism, Academic Press, Inc., New York and London. Wada, A. (1964). Biopolymers, 2, 361-380. Wada, A. & Kozawa, S. (1964). J. Polymer Sci. A2, 853-864. Wada, A., Kawata, I. & Miura, K. I. (1971). BiopoZymers, 10, 1153-1157. Wooley, S. Y. & Holzwarth, G. (1971). J. Amer. Chem. Sot. 93, 4066-4068. Yang, J. T. & Samejima, T. (1969). Prog. Nucleic Acid Res. 9, 223-300.